diff options
Diffstat (limited to 'drivers/gpu/drm/amd/display/dc/dcn20')
17 files changed, 1396 insertions, 208 deletions
diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/Makefile b/drivers/gpu/drm/amd/display/dc/dcn20/Makefile index e9721a906592..f57a3b281408 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/Makefile +++ b/drivers/gpu/drm/amd/display/dc/dcn20/Makefile @@ -18,6 +18,10 @@ endif CFLAGS_dcn20_resource.o := -mhard-float -msse $(cc_stack_align) +ifdef CONFIG_CC_IS_CLANG +CFLAGS_dcn20_resource.o += -msse2 +endif + AMD_DAL_DCN20 = $(addprefix $(AMDDALPATH)/dc/dcn20/,$(DCN20)) AMD_DISPLAY_FILES += $(AMD_DAL_DCN20) diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dccg.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dccg.c index 51a3dfe97f0e..31aa6ee5cd5b 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dccg.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dccg.c @@ -102,14 +102,19 @@ void dccg2_init(struct dccg *dccg) switch (dccg_dcn->base.ctx->dc->res_pool->pipe_count) { case 6: REG_UPDATE(DPPCLK_DTO_CTRL, DPPCLK_DTO_DB_EN[5], 1); + /* Fall through */ case 5: REG_UPDATE(DPPCLK_DTO_CTRL, DPPCLK_DTO_DB_EN[4], 1); + /* Fall through */ case 4: REG_UPDATE(DPPCLK_DTO_CTRL, DPPCLK_DTO_DB_EN[3], 1); + /* Fall through */ case 3: REG_UPDATE(DPPCLK_DTO_CTRL, DPPCLK_DTO_DB_EN[2], 1); + /* Fall through */ case 2: REG_UPDATE(DPPCLK_DTO_CTRL, DPPCLK_DTO_DB_EN[1], 1); + /* Fall through */ case 1: REG_UPDATE(DPPCLK_DTO_CTRL, DPPCLK_DTO_DB_EN[0], 1); break; diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dpp_cm.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dpp_cm.c index e28b8e7bedf5..2d112c316424 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dpp_cm.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dpp_cm.c @@ -52,7 +52,12 @@ static void dpp2_enable_cm_block( { struct dcn20_dpp *dpp = TO_DCN20_DPP(dpp_base); - REG_UPDATE(CM_CONTROL, CM_BYPASS, 0); + unsigned int cm_bypass_mode = 0; + //Temp, put CM in bypass mode + if (dpp_base->ctx->dc->debug.cm_in_bypass) + cm_bypass_mode = 1; + + REG_UPDATE(CM_CONTROL, CM_BYPASS, cm_bypass_mode); } diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dsc.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dsc.c index ffd0014ec3b5..e870caa8d4fa 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dsc.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dsc.c @@ -436,7 +436,7 @@ static void dsc_init_reg_values(struct dsc_reg_values *reg_vals) reg_vals->ich_reset_at_eol = 0; reg_vals->alternate_ich_encoding_en = 0; reg_vals->rc_buffer_model_size = 0; - reg_vals->disable_ich = 0; + /*reg_vals->disable_ich = 0;*/ reg_vals->dsc_dbg_en = 0; for (i = 0; i < 4; i++) @@ -518,9 +518,11 @@ static void dsc_write_to_registers(struct display_stream_compressor *dsc, const ALTERNATE_ICH_ENCODING_EN, reg_vals->alternate_ich_encoding_en, NUMBER_OF_SLICES_IN_VERTICAL_DIRECTION, reg_vals->num_slices_v - 1); - REG_SET_2(DSCC_CONFIG1, 0, + REG_SET(DSCC_CONFIG1, 0, + DSCC_RATE_CONTROL_BUFFER_MODEL_SIZE, reg_vals->rc_buffer_model_size); + /*REG_SET_2(DSCC_CONFIG1, 0, DSCC_RATE_CONTROL_BUFFER_MODEL_SIZE, reg_vals->rc_buffer_model_size, - DSCC_DISABLE_ICH, reg_vals->disable_ich); + DSCC_DISABLE_ICH, reg_vals->disable_ich);*/ REG_SET_4(DSCC_INTERRUPT_CONTROL_STATUS, 0, DSCC_RATE_CONTROL_BUFFER_MODEL0_OVERFLOW_OCCURRED_INT_EN, reg_vals->rc_buffer_model_overflow_int_en[0], diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dsc.h b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dsc.h index 168865a16288..4e2fb38390a4 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dsc.h +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dsc.h @@ -103,7 +103,7 @@ DSC_SF(DSCC0_DSCC_CONFIG0, ALTERNATE_ICH_ENCODING_EN, mask_sh), \ DSC_SF(DSCC0_DSCC_CONFIG0, NUMBER_OF_SLICES_IN_VERTICAL_DIRECTION, mask_sh), \ DSC_SF(DSCC0_DSCC_CONFIG1, DSCC_RATE_CONTROL_BUFFER_MODEL_SIZE, mask_sh), \ - DSC_SF(DSCC0_DSCC_CONFIG1, DSCC_DISABLE_ICH, mask_sh), \ + /*DSC_SF(DSCC0_DSCC_CONFIG1, DSCC_DISABLE_ICH, mask_sh),*/ \ DSC_SF(DSCC0_DSCC_STATUS, DSCC_DOUBLE_BUFFER_REG_UPDATE_PENDING, mask_sh), \ DSC_SF(DSCC0_DSCC_INTERRUPT_CONTROL_STATUS, DSCC_RATE_BUFFER0_OVERFLOW_OCCURRED, mask_sh), \ DSC_SF(DSCC0_DSCC_INTERRUPT_CONTROL_STATUS, DSCC_RATE_BUFFER1_OVERFLOW_OCCURRED, mask_sh), \ @@ -278,7 +278,7 @@ type ALTERNATE_ICH_ENCODING_EN; \ type NUMBER_OF_SLICES_IN_VERTICAL_DIRECTION; \ type DSCC_RATE_CONTROL_BUFFER_MODEL_SIZE; \ - type DSCC_DISABLE_ICH; \ + /*type DSCC_DISABLE_ICH;*/ \ type DSCC_DOUBLE_BUFFER_REG_UPDATE_PENDING; \ type DSCC_RATE_BUFFER0_OVERFLOW_OCCURRED; \ type DSCC_RATE_BUFFER1_OVERFLOW_OCCURRED; \ diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubbub.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubbub.c index ece6e136437b..31d6e79ba2b8 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubbub.c @@ -26,6 +26,7 @@ #include "dcn20_hubbub.h" #include "reg_helper.h" +#include "clk_mgr.h" #define REG(reg)\ hubbub1->regs->reg @@ -337,6 +338,7 @@ static enum dcn_hubbub_page_table_block_size page_table_block_size_to_hw(unsigne break; default: ASSERT(false); + block_size = page_table_block_size; break; } @@ -366,25 +368,24 @@ int hubbub2_init_dchub_sys_ctx(struct hubbub *hubbub, struct dcn_vmid_page_table_config phys_config; REG_SET(DCN_VM_FB_LOCATION_BASE, 0, - FB_BASE, pa_config->system_aperture.fb_base); + FB_BASE, pa_config->system_aperture.fb_base >> 24); REG_SET(DCN_VM_FB_LOCATION_TOP, 0, - FB_TOP, pa_config->system_aperture.fb_top); + FB_TOP, pa_config->system_aperture.fb_top >> 24); REG_SET(DCN_VM_FB_OFFSET, 0, - FB_OFFSET, pa_config->system_aperture.fb_offset); + FB_OFFSET, pa_config->system_aperture.fb_offset >> 24); REG_SET(DCN_VM_AGP_BOT, 0, - AGP_BOT, pa_config->system_aperture.agp_bot); + AGP_BOT, pa_config->system_aperture.agp_bot >> 24); REG_SET(DCN_VM_AGP_TOP, 0, - AGP_TOP, pa_config->system_aperture.agp_top); + AGP_TOP, pa_config->system_aperture.agp_top >> 24); REG_SET(DCN_VM_AGP_BASE, 0, - AGP_BASE, pa_config->system_aperture.agp_base); + AGP_BASE, pa_config->system_aperture.agp_base >> 24); if (pa_config->gart_config.page_table_start_addr != pa_config->gart_config.page_table_end_addr) { - phys_config.depth = 1; - phys_config.block_size = 4096; phys_config.page_table_start_addr = pa_config->gart_config.page_table_start_addr >> 12; phys_config.page_table_end_addr = pa_config->gart_config.page_table_end_addr >> 12; phys_config.page_table_base_addr = pa_config->gart_config.page_table_base_addr; - + phys_config.depth = 0; + phys_config.block_size = 0; // Init VMID 0 based on PA config dcn20_vmid_setup(&hubbub1->vmid[0], &phys_config); } @@ -553,6 +554,16 @@ static void hubbub2_program_watermarks( */ hubbub1_program_urgent_watermarks(hubbub, watermarks, refclk_mhz, safe_to_lower); hubbub1_program_stutter_watermarks(hubbub, watermarks, refclk_mhz, safe_to_lower); + + /* + * There's a special case when going from p-state support to p-state unsupported + * here we are going to LOWER watermarks to go to dummy p-state only, but this has + * to be done prepare_bandwidth, not optimize + */ + if (hubbub1->base.ctx->dc->clk_mgr->clks.prev_p_state_change_support == true && + hubbub1->base.ctx->dc->clk_mgr->clks.p_state_change_support == false) + safe_to_lower = true; + hubbub1_program_pstate_watermarks(hubbub, watermarks, refclk_mhz, safe_to_lower); REG_SET(DCHUBBUB_ARB_SAT_LEVEL, 0, diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubp.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubp.c index d3f7dd374d50..487de87b03eb 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubp.c @@ -156,7 +156,85 @@ void hubp2_program_deadline( { struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); - hubp1_program_deadline(hubp, dlg_attr, ttu_attr); + /* DLG - Per hubp */ + REG_SET_2(BLANK_OFFSET_0, 0, + REFCYC_H_BLANK_END, dlg_attr->refcyc_h_blank_end, + DLG_V_BLANK_END, dlg_attr->dlg_vblank_end); + + REG_SET(BLANK_OFFSET_1, 0, + MIN_DST_Y_NEXT_START, dlg_attr->min_dst_y_next_start); + + REG_SET(DST_DIMENSIONS, 0, + REFCYC_PER_HTOTAL, dlg_attr->refcyc_per_htotal); + + REG_SET_2(DST_AFTER_SCALER, 0, + REFCYC_X_AFTER_SCALER, dlg_attr->refcyc_x_after_scaler, + DST_Y_AFTER_SCALER, dlg_attr->dst_y_after_scaler); + + REG_SET(REF_FREQ_TO_PIX_FREQ, 0, + REF_FREQ_TO_PIX_FREQ, dlg_attr->ref_freq_to_pix_freq); + + /* DLG - Per luma/chroma */ + REG_SET(VBLANK_PARAMETERS_1, 0, + REFCYC_PER_PTE_GROUP_VBLANK_L, dlg_attr->refcyc_per_pte_group_vblank_l); + + if (REG(NOM_PARAMETERS_0)) + REG_SET(NOM_PARAMETERS_0, 0, + DST_Y_PER_PTE_ROW_NOM_L, dlg_attr->dst_y_per_pte_row_nom_l); + + if (REG(NOM_PARAMETERS_1)) + REG_SET(NOM_PARAMETERS_1, 0, + REFCYC_PER_PTE_GROUP_NOM_L, dlg_attr->refcyc_per_pte_group_nom_l); + + REG_SET(NOM_PARAMETERS_4, 0, + DST_Y_PER_META_ROW_NOM_L, dlg_attr->dst_y_per_meta_row_nom_l); + + REG_SET(NOM_PARAMETERS_5, 0, + REFCYC_PER_META_CHUNK_NOM_L, dlg_attr->refcyc_per_meta_chunk_nom_l); + + REG_SET_2(PER_LINE_DELIVERY, 0, + REFCYC_PER_LINE_DELIVERY_L, dlg_attr->refcyc_per_line_delivery_l, + REFCYC_PER_LINE_DELIVERY_C, dlg_attr->refcyc_per_line_delivery_c); + + REG_SET(VBLANK_PARAMETERS_2, 0, + REFCYC_PER_PTE_GROUP_VBLANK_C, dlg_attr->refcyc_per_pte_group_vblank_c); + + if (REG(NOM_PARAMETERS_2)) + REG_SET(NOM_PARAMETERS_2, 0, + DST_Y_PER_PTE_ROW_NOM_C, dlg_attr->dst_y_per_pte_row_nom_c); + + if (REG(NOM_PARAMETERS_3)) + REG_SET(NOM_PARAMETERS_3, 0, + REFCYC_PER_PTE_GROUP_NOM_C, dlg_attr->refcyc_per_pte_group_nom_c); + + REG_SET(NOM_PARAMETERS_6, 0, + DST_Y_PER_META_ROW_NOM_C, dlg_attr->dst_y_per_meta_row_nom_c); + + REG_SET(NOM_PARAMETERS_7, 0, + REFCYC_PER_META_CHUNK_NOM_C, dlg_attr->refcyc_per_meta_chunk_nom_c); + + /* TTU - per hubp */ + REG_SET_2(DCN_TTU_QOS_WM, 0, + QoS_LEVEL_LOW_WM, ttu_attr->qos_level_low_wm, + QoS_LEVEL_HIGH_WM, ttu_attr->qos_level_high_wm); + + /* TTU - per luma/chroma */ + /* Assumed surf0 is luma and 1 is chroma */ + + REG_SET_3(DCN_SURF0_TTU_CNTL0, 0, + REFCYC_PER_REQ_DELIVERY, ttu_attr->refcyc_per_req_delivery_l, + QoS_LEVEL_FIXED, ttu_attr->qos_level_fixed_l, + QoS_RAMP_DISABLE, ttu_attr->qos_ramp_disable_l); + + REG_SET_3(DCN_SURF1_TTU_CNTL0, 0, + REFCYC_PER_REQ_DELIVERY, ttu_attr->refcyc_per_req_delivery_c, + QoS_LEVEL_FIXED, ttu_attr->qos_level_fixed_c, + QoS_RAMP_DISABLE, ttu_attr->qos_ramp_disable_c); + + REG_SET_3(DCN_CUR0_TTU_CNTL0, 0, + REFCYC_PER_REQ_DELIVERY, ttu_attr->refcyc_per_req_delivery_cur0, + QoS_LEVEL_FIXED, ttu_attr->qos_level_fixed_cur0, + QoS_RAMP_DISABLE, ttu_attr->qos_ramp_disable_cur0); REG_SET(FLIP_PARAMETERS_1, 0, REFCYC_PER_PTE_GROUP_FLIP_L, dlg_attr->refcyc_per_pte_group_flip_l); @@ -184,6 +262,39 @@ void hubp2_vready_at_or_After_vsync(struct hubp *hubp, REG_UPDATE(DCHUBP_CNTL, HUBP_VREADY_AT_OR_AFTER_VSYNC, value); } +void hubp2_program_requestor( + struct hubp *hubp, + struct _vcs_dpi_display_rq_regs_st *rq_regs) +{ + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + + REG_UPDATE(HUBPRET_CONTROL, + DET_BUF_PLANE1_BASE_ADDRESS, rq_regs->plane1_base_address); + REG_SET_4(DCN_EXPANSION_MODE, 0, + DRQ_EXPANSION_MODE, rq_regs->drq_expansion_mode, + PRQ_EXPANSION_MODE, rq_regs->prq_expansion_mode, + MRQ_EXPANSION_MODE, rq_regs->mrq_expansion_mode, + CRQ_EXPANSION_MODE, rq_regs->crq_expansion_mode); + REG_SET_8(DCHUBP_REQ_SIZE_CONFIG, 0, + CHUNK_SIZE, rq_regs->rq_regs_l.chunk_size, + MIN_CHUNK_SIZE, rq_regs->rq_regs_l.min_chunk_size, + META_CHUNK_SIZE, rq_regs->rq_regs_l.meta_chunk_size, + MIN_META_CHUNK_SIZE, rq_regs->rq_regs_l.min_meta_chunk_size, + DPTE_GROUP_SIZE, rq_regs->rq_regs_l.dpte_group_size, + MPTE_GROUP_SIZE, rq_regs->rq_regs_l.mpte_group_size, + SWATH_HEIGHT, rq_regs->rq_regs_l.swath_height, + PTE_ROW_HEIGHT_LINEAR, rq_regs->rq_regs_l.pte_row_height_linear); + REG_SET_8(DCHUBP_REQ_SIZE_CONFIG_C, 0, + CHUNK_SIZE_C, rq_regs->rq_regs_c.chunk_size, + MIN_CHUNK_SIZE_C, rq_regs->rq_regs_c.min_chunk_size, + META_CHUNK_SIZE_C, rq_regs->rq_regs_c.meta_chunk_size, + MIN_META_CHUNK_SIZE_C, rq_regs->rq_regs_c.min_meta_chunk_size, + DPTE_GROUP_SIZE_C, rq_regs->rq_regs_c.dpte_group_size, + MPTE_GROUP_SIZE_C, rq_regs->rq_regs_c.mpte_group_size, + SWATH_HEIGHT_C, rq_regs->rq_regs_c.swath_height, + PTE_ROW_HEIGHT_LINEAR_C, rq_regs->rq_regs_c.pte_row_height_linear); +} + static void hubp2_setup( struct hubp *hubp, struct _vcs_dpi_display_dlg_regs_st *dlg_attr, @@ -196,7 +307,7 @@ static void hubp2_setup( */ hubp2_vready_at_or_After_vsync(hubp, pipe_dest); - hubp1_program_requestor(hubp, rq_regs); + hubp2_program_requestor(hubp, rq_regs); hubp2_program_deadline(hubp, dlg_attr, ttu_attr); } @@ -283,11 +394,205 @@ static void hubp2_program_tiling( PIPE_ALIGNED, 0); } +void hubp2_program_size( + struct hubp *hubp, + enum surface_pixel_format format, + const struct plane_size *plane_size, + struct dc_plane_dcc_param *dcc) +{ + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + uint32_t pitch, meta_pitch, pitch_c, meta_pitch_c; + bool use_pitch_c = false; + + /* Program data and meta surface pitch (calculation from addrlib) + * 444 or 420 luma + */ + use_pitch_c = format >= SURFACE_PIXEL_FORMAT_VIDEO_BEGIN + && format < SURFACE_PIXEL_FORMAT_SUBSAMPLE_END; + if (use_pitch_c) { + ASSERT(plane_size->chroma_pitch != 0); + /* Chroma pitch zero can cause system hang! */ + + pitch = plane_size->surface_pitch - 1; + meta_pitch = dcc->meta_pitch - 1; + pitch_c = plane_size->chroma_pitch - 1; + meta_pitch_c = dcc->meta_pitch_c - 1; + } else { + pitch = plane_size->surface_pitch - 1; + meta_pitch = dcc->meta_pitch - 1; + pitch_c = 0; + meta_pitch_c = 0; + } + + if (!dcc->enable) { + meta_pitch = 0; + meta_pitch_c = 0; + } + + REG_UPDATE_2(DCSURF_SURFACE_PITCH, + PITCH, pitch, META_PITCH, meta_pitch); + + use_pitch_c = format >= SURFACE_PIXEL_FORMAT_VIDEO_BEGIN; + if (use_pitch_c) + REG_UPDATE_2(DCSURF_SURFACE_PITCH_C, + PITCH_C, pitch_c, META_PITCH_C, meta_pitch_c); +} + +void hubp2_program_rotation( + struct hubp *hubp, + enum dc_rotation_angle rotation, + bool horizontal_mirror) +{ + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + uint32_t mirror; + + + if (horizontal_mirror) + mirror = 1; + else + mirror = 0; + + /* Program rotation angle and horz mirror - no mirror */ + if (rotation == ROTATION_ANGLE_0) + REG_UPDATE_2(DCSURF_SURFACE_CONFIG, + ROTATION_ANGLE, 0, + H_MIRROR_EN, mirror); + else if (rotation == ROTATION_ANGLE_90) + REG_UPDATE_2(DCSURF_SURFACE_CONFIG, + ROTATION_ANGLE, 1, + H_MIRROR_EN, mirror); + else if (rotation == ROTATION_ANGLE_180) + REG_UPDATE_2(DCSURF_SURFACE_CONFIG, + ROTATION_ANGLE, 2, + H_MIRROR_EN, mirror); + else if (rotation == ROTATION_ANGLE_270) + REG_UPDATE_2(DCSURF_SURFACE_CONFIG, + ROTATION_ANGLE, 3, + H_MIRROR_EN, mirror); +} + +void hubp2_dcc_control(struct hubp *hubp, bool enable, + bool independent_64b_blks) +{ + uint32_t dcc_en = enable ? 1 : 0; + uint32_t dcc_ind_64b_blk = independent_64b_blks ? 1 : 0; + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + + REG_UPDATE_4(DCSURF_SURFACE_CONTROL, + PRIMARY_SURFACE_DCC_EN, dcc_en, + PRIMARY_SURFACE_DCC_IND_64B_BLK, dcc_ind_64b_blk, + SECONDARY_SURFACE_DCC_EN, dcc_en, + SECONDARY_SURFACE_DCC_IND_64B_BLK, dcc_ind_64b_blk); +} + +void hubp2_program_pixel_format( + struct hubp *hubp, + enum surface_pixel_format format) +{ + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + uint32_t red_bar = 3; + uint32_t blue_bar = 2; + + /* swap for ABGR format */ + if (format == SURFACE_PIXEL_FORMAT_GRPH_ABGR8888 + || format == SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010 + || format == SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010_XR_BIAS + || format == SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F) { + red_bar = 2; + blue_bar = 3; + } + + REG_UPDATE_2(HUBPRET_CONTROL, + CROSSBAR_SRC_CB_B, blue_bar, + CROSSBAR_SRC_CR_R, red_bar); + + /* Mapping is same as ipp programming (cnvc) */ + + switch (format) { + case SURFACE_PIXEL_FORMAT_GRPH_ARGB1555: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 1); + break; + case SURFACE_PIXEL_FORMAT_GRPH_RGB565: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 3); + break; + case SURFACE_PIXEL_FORMAT_GRPH_ARGB8888: + case SURFACE_PIXEL_FORMAT_GRPH_ABGR8888: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 8); + break; + case SURFACE_PIXEL_FORMAT_GRPH_ARGB2101010: + case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010: + case SURFACE_PIXEL_FORMAT_GRPH_ABGR2101010_XR_BIAS: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 10); + break; + case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 22); + break; + case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F: + case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F:/*we use crossbar already*/ + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 24); + break; + + case SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 65); + break; + case SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 64); + break; + case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCbCr: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 67); + break; + case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 66); + break; + case SURFACE_PIXEL_FORMAT_VIDEO_AYCrCb8888: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 12); + break; +#if defined(CONFIG_DRM_AMD_DC_DCN2_0) + case SURFACE_PIXEL_FORMAT_GRPH_RGB111110_FIX: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 112); + break; + case SURFACE_PIXEL_FORMAT_GRPH_BGR101111_FIX: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 113); + break; + case SURFACE_PIXEL_FORMAT_VIDEO_ACrYCb2101010: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 114); + break; + case SURFACE_PIXEL_FORMAT_GRPH_RGB111110_FLOAT: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 118); + break; + case SURFACE_PIXEL_FORMAT_GRPH_BGR101111_FLOAT: + REG_UPDATE(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, 119); + break; +#endif + default: + BREAK_TO_DEBUGGER(); + break; + } + + /* don't see the need of program the xbar in DCN 1.0 */ +} + void hubp2_program_surface_config( struct hubp *hubp, enum surface_pixel_format format, union dc_tiling_info *tiling_info, - union plane_size *plane_size, + struct plane_size *plane_size, enum dc_rotation_angle rotation, struct dc_plane_dcc_param *dcc, bool horizontal_mirror, @@ -295,11 +600,11 @@ void hubp2_program_surface_config( { struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); - hubp1_dcc_control(hubp, dcc->enable, dcc->grph.independent_64b_blks); + hubp2_dcc_control(hubp, dcc->enable, dcc->independent_64b_blks); hubp2_program_tiling(hubp2, tiling_info, format); - hubp1_program_size(hubp, format, plane_size, dcc); - hubp1_program_rotation(hubp, rotation, horizontal_mirror); - hubp1_program_pixel_format(hubp, format); + hubp2_program_size(hubp, format, plane_size, dcc); + hubp2_program_rotation(hubp, rotation, horizontal_mirror); + hubp2_program_pixel_format(hubp, format); } enum cursor_lines_per_chunk hubp2_get_lines_per_chunk( @@ -652,28 +957,381 @@ void hubp2_set_flip_control_surface_gsl(struct hubp *hubp, bool enable) REG_UPDATE(DCSURF_FLIP_CONTROL2, SURFACE_GSL_ENABLE, enable ? 1 : 0); } +bool hubp2_is_flip_pending(struct hubp *hubp) +{ + uint32_t flip_pending = 0; + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + struct dc_plane_address earliest_inuse_address; + + REG_GET(DCSURF_FLIP_CONTROL, + SURFACE_FLIP_PENDING, &flip_pending); + + REG_GET(DCSURF_SURFACE_EARLIEST_INUSE, + SURFACE_EARLIEST_INUSE_ADDRESS, &earliest_inuse_address.grph.addr.low_part); + + REG_GET(DCSURF_SURFACE_EARLIEST_INUSE_HIGH, + SURFACE_EARLIEST_INUSE_ADDRESS_HIGH, &earliest_inuse_address.grph.addr.high_part); + + if (flip_pending) + return true; + + if (earliest_inuse_address.grph.addr.quad_part != hubp->request_address.grph.addr.quad_part) + return true; + + return false; +} + +void hubp2_set_blank(struct hubp *hubp, bool blank) +{ + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + uint32_t blank_en = blank ? 1 : 0; + + REG_UPDATE_2(DCHUBP_CNTL, + HUBP_BLANK_EN, blank_en, + HUBP_TTU_DISABLE, blank_en); + + if (blank) { + uint32_t reg_val = REG_READ(DCHUBP_CNTL); + + if (reg_val) { + /* init sequence workaround: in case HUBP is + * power gated, this wait would timeout. + * + * we just wrote reg_val to non-0, if it stay 0 + * it means HUBP is gated + */ + REG_WAIT(DCHUBP_CNTL, + HUBP_NO_OUTSTANDING_REQ, 1, + 1, 200); + } + + hubp->mpcc_id = 0xf; + hubp->opp_id = OPP_ID_INVALID; + } +} + +void hubp2_cursor_set_position( + struct hubp *hubp, + const struct dc_cursor_position *pos, + const struct dc_cursor_mi_param *param) +{ + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + int src_x_offset = pos->x - pos->x_hotspot - param->viewport.x; + int src_y_offset = pos->y - pos->y_hotspot - param->viewport.y; + int x_hotspot = pos->x_hotspot; + int y_hotspot = pos->y_hotspot; + uint32_t dst_x_offset; + uint32_t cur_en = pos->enable ? 1 : 0; + + /* + * Guard aganst cursor_set_position() from being called with invalid + * attributes + * + * TODO: Look at combining cursor_set_position() and + * cursor_set_attributes() into cursor_update() + */ + if (hubp->curs_attr.address.quad_part == 0) + return; + + if (param->rotation == ROTATION_ANGLE_90 || param->rotation == ROTATION_ANGLE_270) { + src_x_offset = pos->y - pos->y_hotspot - param->viewport.x; + y_hotspot = pos->x_hotspot; + x_hotspot = pos->y_hotspot; + } + + if (param->mirror) { + x_hotspot = param->viewport.width - x_hotspot; + src_x_offset = param->viewport.x + param->viewport.width - src_x_offset; + } + + dst_x_offset = (src_x_offset >= 0) ? src_x_offset : 0; + dst_x_offset *= param->ref_clk_khz; + dst_x_offset /= param->pixel_clk_khz; + + ASSERT(param->h_scale_ratio.value); + + if (param->h_scale_ratio.value) + dst_x_offset = dc_fixpt_floor(dc_fixpt_div( + dc_fixpt_from_int(dst_x_offset), + param->h_scale_ratio)); + + if (src_x_offset >= (int)param->viewport.width) + cur_en = 0; /* not visible beyond right edge*/ + + if (src_x_offset + (int)hubp->curs_attr.width <= 0) + cur_en = 0; /* not visible beyond left edge*/ + + if (src_y_offset >= (int)param->viewport.height) + cur_en = 0; /* not visible beyond bottom edge*/ + + if (src_y_offset + (int)hubp->curs_attr.height <= 0) + cur_en = 0; /* not visible beyond top edge*/ + + if (cur_en && REG_READ(CURSOR_SURFACE_ADDRESS) == 0) + hubp->funcs->set_cursor_attributes(hubp, &hubp->curs_attr); + + REG_UPDATE(CURSOR_CONTROL, + CURSOR_ENABLE, cur_en); + + REG_SET_2(CURSOR_POSITION, 0, + CURSOR_X_POSITION, pos->x, + CURSOR_Y_POSITION, pos->y); + + REG_SET_2(CURSOR_HOT_SPOT, 0, + CURSOR_HOT_SPOT_X, x_hotspot, + CURSOR_HOT_SPOT_Y, y_hotspot); + + REG_SET(CURSOR_DST_OFFSET, 0, + CURSOR_DST_X_OFFSET, dst_x_offset); + /* TODO Handle surface pixel formats other than 4:4:4 */ +} + +void hubp2_clk_cntl(struct hubp *hubp, bool enable) +{ + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + uint32_t clk_enable = enable ? 1 : 0; + + REG_UPDATE(HUBP_CLK_CNTL, HUBP_CLOCK_ENABLE, clk_enable); +} + +void hubp2_vtg_sel(struct hubp *hubp, uint32_t otg_inst) +{ + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + + REG_UPDATE(DCHUBP_CNTL, HUBP_VTG_SEL, otg_inst); +} + +void hubp2_clear_underflow(struct hubp *hubp) +{ + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + + REG_UPDATE(DCHUBP_CNTL, HUBP_UNDERFLOW_CLEAR, 1); +} + +void hubp2_read_state_common(struct hubp *hubp) +{ + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + struct dcn_hubp_state *s = &hubp2->state; + struct _vcs_dpi_display_dlg_regs_st *dlg_attr = &s->dlg_attr; + struct _vcs_dpi_display_ttu_regs_st *ttu_attr = &s->ttu_attr; + struct _vcs_dpi_display_rq_regs_st *rq_regs = &s->rq_regs; + + /* Requester */ + REG_GET(HUBPRET_CONTROL, + DET_BUF_PLANE1_BASE_ADDRESS, &rq_regs->plane1_base_address); + REG_GET_4(DCN_EXPANSION_MODE, + DRQ_EXPANSION_MODE, &rq_regs->drq_expansion_mode, + PRQ_EXPANSION_MODE, &rq_regs->prq_expansion_mode, + MRQ_EXPANSION_MODE, &rq_regs->mrq_expansion_mode, + CRQ_EXPANSION_MODE, &rq_regs->crq_expansion_mode); + + /* DLG - Per hubp */ + REG_GET_2(BLANK_OFFSET_0, + REFCYC_H_BLANK_END, &dlg_attr->refcyc_h_blank_end, + DLG_V_BLANK_END, &dlg_attr->dlg_vblank_end); + + REG_GET(BLANK_OFFSET_1, + MIN_DST_Y_NEXT_START, &dlg_attr->min_dst_y_next_start); + + REG_GET(DST_DIMENSIONS, + REFCYC_PER_HTOTAL, &dlg_attr->refcyc_per_htotal); + + REG_GET_2(DST_AFTER_SCALER, + REFCYC_X_AFTER_SCALER, &dlg_attr->refcyc_x_after_scaler, + DST_Y_AFTER_SCALER, &dlg_attr->dst_y_after_scaler); + + if (REG(PREFETCH_SETTINS)) + REG_GET_2(PREFETCH_SETTINS, + DST_Y_PREFETCH, &dlg_attr->dst_y_prefetch, + VRATIO_PREFETCH, &dlg_attr->vratio_prefetch); + else + REG_GET_2(PREFETCH_SETTINGS, + DST_Y_PREFETCH, &dlg_attr->dst_y_prefetch, + VRATIO_PREFETCH, &dlg_attr->vratio_prefetch); + + REG_GET_2(VBLANK_PARAMETERS_0, + DST_Y_PER_VM_VBLANK, &dlg_attr->dst_y_per_vm_vblank, + DST_Y_PER_ROW_VBLANK, &dlg_attr->dst_y_per_row_vblank); + + REG_GET(REF_FREQ_TO_PIX_FREQ, + REF_FREQ_TO_PIX_FREQ, &dlg_attr->ref_freq_to_pix_freq); + + /* DLG - Per luma/chroma */ + REG_GET(VBLANK_PARAMETERS_1, + REFCYC_PER_PTE_GROUP_VBLANK_L, &dlg_attr->refcyc_per_pte_group_vblank_l); + + REG_GET(VBLANK_PARAMETERS_3, + REFCYC_PER_META_CHUNK_VBLANK_L, &dlg_attr->refcyc_per_meta_chunk_vblank_l); + + if (REG(NOM_PARAMETERS_0)) + REG_GET(NOM_PARAMETERS_0, + DST_Y_PER_PTE_ROW_NOM_L, &dlg_attr->dst_y_per_pte_row_nom_l); + + if (REG(NOM_PARAMETERS_1)) + REG_GET(NOM_PARAMETERS_1, + REFCYC_PER_PTE_GROUP_NOM_L, &dlg_attr->refcyc_per_pte_group_nom_l); + + REG_GET(NOM_PARAMETERS_4, + DST_Y_PER_META_ROW_NOM_L, &dlg_attr->dst_y_per_meta_row_nom_l); + + REG_GET(NOM_PARAMETERS_5, + REFCYC_PER_META_CHUNK_NOM_L, &dlg_attr->refcyc_per_meta_chunk_nom_l); + + REG_GET_2(PER_LINE_DELIVERY_PRE, + REFCYC_PER_LINE_DELIVERY_PRE_L, &dlg_attr->refcyc_per_line_delivery_pre_l, + REFCYC_PER_LINE_DELIVERY_PRE_C, &dlg_attr->refcyc_per_line_delivery_pre_c); + + REG_GET_2(PER_LINE_DELIVERY, + REFCYC_PER_LINE_DELIVERY_L, &dlg_attr->refcyc_per_line_delivery_l, + REFCYC_PER_LINE_DELIVERY_C, &dlg_attr->refcyc_per_line_delivery_c); + + if (REG(PREFETCH_SETTINS_C)) + REG_GET(PREFETCH_SETTINS_C, + VRATIO_PREFETCH_C, &dlg_attr->vratio_prefetch_c); + else + REG_GET(PREFETCH_SETTINGS_C, + VRATIO_PREFETCH_C, &dlg_attr->vratio_prefetch_c); + + REG_GET(VBLANK_PARAMETERS_2, + REFCYC_PER_PTE_GROUP_VBLANK_C, &dlg_attr->refcyc_per_pte_group_vblank_c); + + REG_GET(VBLANK_PARAMETERS_4, + REFCYC_PER_META_CHUNK_VBLANK_C, &dlg_attr->refcyc_per_meta_chunk_vblank_c); + + if (REG(NOM_PARAMETERS_2)) + REG_GET(NOM_PARAMETERS_2, + DST_Y_PER_PTE_ROW_NOM_C, &dlg_attr->dst_y_per_pte_row_nom_c); + + if (REG(NOM_PARAMETERS_3)) + REG_GET(NOM_PARAMETERS_3, + REFCYC_PER_PTE_GROUP_NOM_C, &dlg_attr->refcyc_per_pte_group_nom_c); + + REG_GET(NOM_PARAMETERS_6, + DST_Y_PER_META_ROW_NOM_C, &dlg_attr->dst_y_per_meta_row_nom_c); + + REG_GET(NOM_PARAMETERS_7, + REFCYC_PER_META_CHUNK_NOM_C, &dlg_attr->refcyc_per_meta_chunk_nom_c); + + /* TTU - per hubp */ + REG_GET_2(DCN_TTU_QOS_WM, + QoS_LEVEL_LOW_WM, &ttu_attr->qos_level_low_wm, + QoS_LEVEL_HIGH_WM, &ttu_attr->qos_level_high_wm); + + REG_GET_2(DCN_GLOBAL_TTU_CNTL, + MIN_TTU_VBLANK, &ttu_attr->min_ttu_vblank, + QoS_LEVEL_FLIP, &ttu_attr->qos_level_flip); + + /* TTU - per luma/chroma */ + /* Assumed surf0 is luma and 1 is chroma */ + + REG_GET_3(DCN_SURF0_TTU_CNTL0, + REFCYC_PER_REQ_DELIVERY, &ttu_attr->refcyc_per_req_delivery_l, + QoS_LEVEL_FIXED, &ttu_attr->qos_level_fixed_l, + QoS_RAMP_DISABLE, &ttu_attr->qos_ramp_disable_l); + + REG_GET(DCN_SURF0_TTU_CNTL1, + REFCYC_PER_REQ_DELIVERY_PRE, + &ttu_attr->refcyc_per_req_delivery_pre_l); + + REG_GET_3(DCN_SURF1_TTU_CNTL0, + REFCYC_PER_REQ_DELIVERY, &ttu_attr->refcyc_per_req_delivery_c, + QoS_LEVEL_FIXED, &ttu_attr->qos_level_fixed_c, + QoS_RAMP_DISABLE, &ttu_attr->qos_ramp_disable_c); + + REG_GET(DCN_SURF1_TTU_CNTL1, + REFCYC_PER_REQ_DELIVERY_PRE, + &ttu_attr->refcyc_per_req_delivery_pre_c); + + /* Rest of hubp */ + REG_GET(DCSURF_SURFACE_CONFIG, + SURFACE_PIXEL_FORMAT, &s->pixel_format); + + REG_GET(DCSURF_SURFACE_EARLIEST_INUSE_HIGH, + SURFACE_EARLIEST_INUSE_ADDRESS_HIGH, &s->inuse_addr_hi); + + REG_GET(DCSURF_SURFACE_EARLIEST_INUSE, + SURFACE_EARLIEST_INUSE_ADDRESS, &s->inuse_addr_lo); + + REG_GET_2(DCSURF_PRI_VIEWPORT_DIMENSION, + PRI_VIEWPORT_WIDTH, &s->viewport_width, + PRI_VIEWPORT_HEIGHT, &s->viewport_height); + + REG_GET_2(DCSURF_SURFACE_CONFIG, + ROTATION_ANGLE, &s->rotation_angle, + H_MIRROR_EN, &s->h_mirror_en); + + REG_GET(DCSURF_TILING_CONFIG, + SW_MODE, &s->sw_mode); + + REG_GET(DCSURF_SURFACE_CONTROL, + PRIMARY_SURFACE_DCC_EN, &s->dcc_en); + + REG_GET_3(DCHUBP_CNTL, + HUBP_BLANK_EN, &s->blank_en, + HUBP_TTU_DISABLE, &s->ttu_disable, + HUBP_UNDERFLOW_STATUS, &s->underflow_status); + + REG_GET(DCN_GLOBAL_TTU_CNTL, + MIN_TTU_VBLANK, &s->min_ttu_vblank); + + REG_GET_2(DCN_TTU_QOS_WM, + QoS_LEVEL_LOW_WM, &s->qos_level_low_wm, + QoS_LEVEL_HIGH_WM, &s->qos_level_high_wm); + +} + +void hubp2_read_state(struct hubp *hubp) +{ + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + struct dcn_hubp_state *s = &hubp2->state; + struct _vcs_dpi_display_rq_regs_st *rq_regs = &s->rq_regs; + + hubp2_read_state_common(hubp); + + REG_GET_8(DCHUBP_REQ_SIZE_CONFIG, + CHUNK_SIZE, &rq_regs->rq_regs_l.chunk_size, + MIN_CHUNK_SIZE, &rq_regs->rq_regs_l.min_chunk_size, + META_CHUNK_SIZE, &rq_regs->rq_regs_l.meta_chunk_size, + MIN_META_CHUNK_SIZE, &rq_regs->rq_regs_l.min_meta_chunk_size, + DPTE_GROUP_SIZE, &rq_regs->rq_regs_l.dpte_group_size, + MPTE_GROUP_SIZE, &rq_regs->rq_regs_l.mpte_group_size, + SWATH_HEIGHT, &rq_regs->rq_regs_l.swath_height, + PTE_ROW_HEIGHT_LINEAR, &rq_regs->rq_regs_l.pte_row_height_linear); + + REG_GET_8(DCHUBP_REQ_SIZE_CONFIG_C, + CHUNK_SIZE_C, &rq_regs->rq_regs_c.chunk_size, + MIN_CHUNK_SIZE_C, &rq_regs->rq_regs_c.min_chunk_size, + META_CHUNK_SIZE_C, &rq_regs->rq_regs_c.meta_chunk_size, + MIN_META_CHUNK_SIZE_C, &rq_regs->rq_regs_c.min_meta_chunk_size, + DPTE_GROUP_SIZE_C, &rq_regs->rq_regs_c.dpte_group_size, + MPTE_GROUP_SIZE_C, &rq_regs->rq_regs_c.mpte_group_size, + SWATH_HEIGHT_C, &rq_regs->rq_regs_c.swath_height, + PTE_ROW_HEIGHT_LINEAR_C, &rq_regs->rq_regs_c.pte_row_height_linear); + +} + static struct hubp_funcs dcn20_hubp_funcs = { .hubp_enable_tripleBuffer = hubp2_enable_triplebuffer, .hubp_is_triplebuffer_enabled = hubp2_is_triplebuffer_enabled, .hubp_program_surface_flip_and_addr = hubp2_program_surface_flip_and_addr, .hubp_program_surface_config = hubp2_program_surface_config, - .hubp_is_flip_pending = hubp1_is_flip_pending, + .hubp_is_flip_pending = hubp2_is_flip_pending, .hubp_setup = hubp2_setup, .hubp_setup_interdependent = hubp2_setup_interdependent, .hubp_set_vm_system_aperture_settings = hubp2_set_vm_system_aperture_settings, - .set_blank = hubp1_set_blank, - .dcc_control = hubp1_dcc_control, + .set_blank = hubp2_set_blank, + .dcc_control = hubp2_dcc_control, .hubp_update_dchub = hubp2_update_dchub, .mem_program_viewport = min_set_viewport, .set_cursor_attributes = hubp2_cursor_set_attributes, - .set_cursor_position = hubp1_cursor_set_position, - .hubp_clk_cntl = hubp1_clk_cntl, - .hubp_vtg_sel = hubp1_vtg_sel, + .set_cursor_position = hubp2_cursor_set_position, + .hubp_clk_cntl = hubp2_clk_cntl, + .hubp_vtg_sel = hubp2_vtg_sel, .dmdata_set_attributes = hubp2_dmdata_set_attributes, .dmdata_load = hubp2_dmdata_load, .dmdata_status_done = hubp2_dmdata_status_done, - .hubp_read_state = hubp1_read_state, - .hubp_clear_underflow = hubp1_clear_underflow, + .hubp_read_state = hubp2_read_state, + .hubp_clear_underflow = hubp2_clear_underflow, .hubp_set_flip_control_surface_gsl = hubp2_set_flip_control_surface_gsl, .hubp_init = hubp1_init, }; diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubp.h b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubp.h index d5acc348be22..1c53af4811e8 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubp.h +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubp.h @@ -72,8 +72,8 @@ SR(DCN_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB),\ SR(DCN_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB) -#define HUBP_MASK_SH_LIST_DCN2_COMMON(mask_sh)\ - HUBP_MASK_SH_LIST_DCN(mask_sh),\ +#define HUBP_MASK_SH_LIST_DCN2_SHARE_COMMON(mask_sh)\ + HUBP_MASK_SH_LIST_DCN_SHARE_COMMON(mask_sh),\ HUBP_MASK_SH_LIST_DCN_VM(mask_sh),\ HUBP_SF(HUBP0_DCSURF_SURFACE_CONFIG, ROTATION_ANGLE, mask_sh),\ HUBP_SF(HUBP0_DCSURF_SURFACE_CONFIG, H_MIRROR_EN, mask_sh),\ @@ -127,13 +127,21 @@ HUBP_SF(HUBPREQ0_DCSURF_FLIP_CONTROL2, SURFACE_TRIPLE_BUFFER_ENABLE, mask_sh),\ HUBP_SF(HUBPREQ0_VMID_SETTINGS_0, VMID, mask_sh) +/*DCN2.x and DCN1.x*/ +#define HUBP_MASK_SH_LIST_DCN2_COMMON(mask_sh)\ + HUBP_MASK_SH_LIST_DCN2_SHARE_COMMON(mask_sh),\ + HUBP_SF(HUBP0_DCSURF_TILING_CONFIG, RB_ALIGNED, mask_sh),\ + HUBP_SF(HUBP0_DCHUBP_REQ_SIZE_CONFIG, MPTE_GROUP_SIZE, mask_sh),\ + HUBP_SF(HUBP0_DCHUBP_REQ_SIZE_CONFIG_C, MPTE_GROUP_SIZE_C, mask_sh) + +/*DCN2.0 specific*/ #define HUBP_MASK_SH_LIST_DCN20(mask_sh)\ HUBP_MASK_SH_LIST_DCN2_COMMON(mask_sh),\ HUBP_SF(DCN_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB, DCN_VM_SYSTEM_APERTURE_DEFAULT_SYSTEM, mask_sh),\ HUBP_SF(DCN_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB, DCN_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB, mask_sh),\ HUBP_SF(DCN_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB, DCN_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB, mask_sh) - +/*DCN2.x */ #define DCN2_HUBP_REG_COMMON_VARIABLE_LIST \ HUBP_COMMON_REG_VARIABLE_LIST; \ uint32_t DMDATA_ADDRESS_HIGH; \ @@ -149,14 +157,11 @@ uint32_t FLIP_PARAMETERS_2;\ uint32_t DCN_CUR1_TTU_CNTL0;\ uint32_t DCN_CUR1_TTU_CNTL1;\ - uint32_t VMID_SETTINGS_0;\ - uint32_t FLIP_PARAMETERS_3;\ - uint32_t FLIP_PARAMETERS_4;\ - uint32_t VBLANK_PARAMETERS_5;\ - uint32_t VBLANK_PARAMETERS_6 + uint32_t VMID_SETTINGS_0 + #define DCN2_HUBP_REG_FIELD_VARIABLE_LIST(type) \ - DCN_HUBP_REG_FIELD_LIST(type); \ + DCN_HUBP_REG_FIELD_BASE_LIST(type); \ type DMDATA_ADDRESS_HIGH;\ type DMDATA_MODE;\ type DMDATA_UPDATED;\ @@ -262,16 +267,53 @@ bool hubp2_program_surface_flip_and_addr( const struct dc_plane_address *address, bool flip_immediate); +void hubp2_dcc_control(struct hubp *hubp, bool enable, + bool independent_64b_blks); + +void hubp2_program_size( + struct hubp *hubp, + enum surface_pixel_format format, + const struct plane_size *plane_size, + struct dc_plane_dcc_param *dcc); + +void hubp2_program_rotation( + struct hubp *hubp, + enum dc_rotation_angle rotation, + bool horizontal_mirror); + +void hubp2_program_pixel_format( + struct hubp *hubp, + enum surface_pixel_format format); + void hubp2_program_surface_config( struct hubp *hubp, enum surface_pixel_format format, union dc_tiling_info *tiling_info, - union plane_size *plane_size, + struct plane_size *plane_size, enum dc_rotation_angle rotation, struct dc_plane_dcc_param *dcc, bool horizontal_mirror, unsigned int compat_level); +bool hubp2_is_flip_pending(struct hubp *hubp); + +void hubp2_set_blank(struct hubp *hubp, bool blank); + +void hubp2_cursor_set_position( + struct hubp *hubp, + const struct dc_cursor_position *pos, + const struct dc_cursor_mi_param *param); + +void hubp2_clk_cntl(struct hubp *hubp, bool enable); + +void hubp2_vtg_sel(struct hubp *hubp, uint32_t otg_inst); + +void hubp2_clear_underflow(struct hubp *hubp); + +void hubp2_read_state_common(struct hubp *hubp); + +void hubp2_read_state(struct hubp *hubp); + #endif /* __DC_MEM_INPUT_DCN20_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c index 0b84a322b8a2..38b3c89b2a59 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c @@ -94,28 +94,34 @@ static void enable_power_gating_plane( REG_UPDATE(DOMAIN2_PG_CONFIG, DOMAIN2_POWER_FORCEON, force_on); REG_UPDATE(DOMAIN4_PG_CONFIG, DOMAIN4_POWER_FORCEON, force_on); REG_UPDATE(DOMAIN6_PG_CONFIG, DOMAIN6_POWER_FORCEON, force_on); - REG_UPDATE(DOMAIN8_PG_CONFIG, DOMAIN8_POWER_FORCEON, force_on); - /*Do not power gate DCHUB5, should be left at HW default, power on permanently*/ - /*REG_UPDATE(DOMAIN10_PG_CONFIG, DOMAIN10_POWER_FORCEON, force_on);*/ + if (REG(DOMAIN8_PG_CONFIG)) + REG_UPDATE(DOMAIN8_PG_CONFIG, DOMAIN8_POWER_FORCEON, force_on); + if (REG(DOMAIN10_PG_CONFIG)) + REG_UPDATE(DOMAIN10_PG_CONFIG, DOMAIN8_POWER_FORCEON, force_on); /* DPP0/1/2/3/4/5 */ REG_UPDATE(DOMAIN1_PG_CONFIG, DOMAIN1_POWER_FORCEON, force_on); REG_UPDATE(DOMAIN3_PG_CONFIG, DOMAIN3_POWER_FORCEON, force_on); REG_UPDATE(DOMAIN5_PG_CONFIG, DOMAIN5_POWER_FORCEON, force_on); REG_UPDATE(DOMAIN7_PG_CONFIG, DOMAIN7_POWER_FORCEON, force_on); - REG_UPDATE(DOMAIN9_PG_CONFIG, DOMAIN9_POWER_FORCEON, force_on); - /*Do not power gate DPP5, should be left at HW default, power on permanently*/ - /*REG_UPDATE(DOMAIN11_PG_CONFIG, DOMAIN11_POWER_FORCEON, force_on);*/ + if (REG(DOMAIN9_PG_CONFIG)) + REG_UPDATE(DOMAIN9_PG_CONFIG, DOMAIN9_POWER_FORCEON, force_on); + if (REG(DOMAIN11_PG_CONFIG)) + REG_UPDATE(DOMAIN11_PG_CONFIG, DOMAIN9_POWER_FORCEON, force_on); + /* DCS0/1/2/3/4/5 */ REG_UPDATE(DOMAIN16_PG_CONFIG, DOMAIN16_POWER_FORCEON, force_on); REG_UPDATE(DOMAIN17_PG_CONFIG, DOMAIN17_POWER_FORCEON, force_on); REG_UPDATE(DOMAIN18_PG_CONFIG, DOMAIN18_POWER_FORCEON, force_on); - REG_UPDATE(DOMAIN19_PG_CONFIG, DOMAIN19_POWER_FORCEON, force_on); - REG_UPDATE(DOMAIN20_PG_CONFIG, DOMAIN20_POWER_FORCEON, force_on); - REG_UPDATE(DOMAIN21_PG_CONFIG, DOMAIN21_POWER_FORCEON, force_on); + if (REG(DOMAIN19_PG_CONFIG)) + REG_UPDATE(DOMAIN19_PG_CONFIG, DOMAIN19_POWER_FORCEON, force_on); + if (REG(DOMAIN20_PG_CONFIG)) + REG_UPDATE(DOMAIN20_PG_CONFIG, DOMAIN20_POWER_FORCEON, force_on); + if (REG(DOMAIN21_PG_CONFIG)) + REG_UPDATE(DOMAIN21_PG_CONFIG, DOMAIN21_POWER_FORCEON, force_on); } -static void dcn20_dccg_init(struct dce_hwseq *hws) +void dcn20_dccg_init(struct dce_hwseq *hws) { /* * set MICROSECOND_TIME_BASE_DIV @@ -138,6 +144,45 @@ static void dcn20_dccg_init(struct dce_hwseq *hws) /* This value is dependent on the hardware pipeline delay so set once per SOC */ REG_WRITE(DISPCLK_FREQ_CHANGE_CNTL, 0x801003c); } +void dcn20_display_init(struct dc *dc) +{ + struct dce_hwseq *hws = dc->hwseq; + + /* RBBMIF + * disable RBBMIF timeout detection for all clients + * Ensure RBBMIF does not drop register accesses due to the per-client timeout + */ + REG_WRITE(RBBMIF_TIMEOUT_DIS, 0xFFFFFFFF); + REG_WRITE(RBBMIF_TIMEOUT_DIS_2, 0xFFFFFFFF); + + /* DCCG */ + dcn20_dccg_init(hws); + + /* Disable all memory low power mode. All memories are enabled. */ + REG_UPDATE(DC_MEM_GLOBAL_PWR_REQ_CNTL, DC_MEM_GLOBAL_PWR_REQ_DIS, 1); + + /* DCHUB/MMHUBBUB + * set global timer refclk divider + * 100Mhz refclk -> 2 + * 27Mhz refclk -> 1 + * 48Mhz refclk -> 1 + */ + REG_UPDATE(DCHUBBUB_GLOBAL_TIMER_CNTL, DCHUBBUB_GLOBAL_TIMER_REFDIV, 2); + REG_UPDATE(DCHUBBUB_GLOBAL_TIMER_CNTL, DCHUBBUB_GLOBAL_TIMER_ENABLE, 1); + REG_WRITE(REFCLK_CNTL, 0); + + /* OPTC + * OTG_CONTROL.OTG_DISABLE_POINT_CNTL = 0x3; will be set during optc2_enable_crtc + */ + + /* AZ + * default value is 0x64 for 100Mhz ref clock, if the ref clock is 100Mhz, no need to program this regiser, + * if not, it should be programmed according to the ref clock + */ + REG_UPDATE(AZALIA_AUDIO_DTO, AZALIA_AUDIO_DTO_MODULE, 0x64); + /* Enable controller clock gating */ + REG_WRITE(AZALIA_CONTROLLER_CLOCK_GATING, 0x1); +} static void disable_vga( struct dce_hwseq *hws) @@ -163,7 +208,7 @@ void dcn20_program_tripleBuffer( } /* Blank pixel data during initialization */ -static void dcn20_init_blank( +void dcn20_init_blank( struct dc *dc, struct timing_generator *tg) { @@ -585,6 +630,12 @@ static void dcn20_init_hw(struct dc *dc) } } +#ifdef CONFIG_DRM_AMD_DC_DSC_SUPPORT + /* Power gate DSCs */ + for (i = 0; i < res_pool->res_cap->num_dsc; i++) + dcn20_dsc_pg_control(hws, res_pool->dscs[i]->inst, false); +#endif + /* Blank pixel data with OPP DPG */ for (i = 0; i < dc->res_pool->timing_generator_count; i++) { struct timing_generator *tg = dc->res_pool->timing_generators[i]; @@ -909,14 +960,14 @@ static bool dcn20_set_shaper_3dlut( result = dpp_base->funcs->dpp_program_shaper_lut(dpp_base, shaper_lut); if (plane_state->lut3d_func && - plane_state->lut3d_func->initialized == true) + plane_state->lut3d_func->state.bits.initialized == 1) result = dpp_base->funcs->dpp_program_3dlut(dpp_base, &plane_state->lut3d_func->lut_3d); else result = dpp_base->funcs->dpp_program_3dlut(dpp_base, NULL); if (plane_state->lut3d_func && - plane_state->lut3d_func->initialized == true && + plane_state->lut3d_func->state.bits.initialized == 1 && plane_state->lut3d_func->hdr_multiplier != 0) dpp_base->funcs->dpp_set_hdr_multiplier(dpp_base, plane_state->lut3d_func->hdr_multiplier); @@ -1153,8 +1204,8 @@ void dcn20_enable_plane( apt.sys_default.quad_part = 0; - apt.sys_high.quad_part = dc->vm_pa_config.system_aperture.start_addr; - apt.sys_low.quad_part = dc->vm_pa_config.system_aperture.end_addr; + apt.sys_low.quad_part = dc->vm_pa_config.system_aperture.start_addr; + apt.sys_high.quad_part = dc->vm_pa_config.system_aperture.end_addr; // Program system aperture settings pipe_ctx->plane_res.hubp->funcs->hubp_set_vm_system_aperture_settings(pipe_ctx->plane_res.hubp, &apt); @@ -1242,6 +1293,8 @@ void dcn20_pipe_control_lock_global( CRTC_STATE_VACTIVE); pipe->stream_res.tg->funcs->wait_for_state(pipe->stream_res.tg, CRTC_STATE_VBLANK); + pipe->stream_res.tg->funcs->wait_for_state(pipe->stream_res.tg, + CRTC_STATE_VACTIVE); pipe->stream_res.tg->funcs->lock_doublebuffer_disable( pipe->stream_res.tg); } @@ -1263,6 +1316,17 @@ void dcn20_pipe_control_lock( if (pipe->plane_state != NULL) flip_immediate = pipe->plane_state->flip_immediate; + if (flip_immediate && lock) { + while (pipe->plane_res.hubp->funcs->hubp_is_flip_pending(pipe->plane_res.hubp)) { + udelay(1); + } + + if (pipe->bottom_pipe != NULL) + while (pipe->bottom_pipe->plane_res.hubp->funcs->hubp_is_flip_pending(pipe->bottom_pipe->plane_res.hubp)) { + udelay(1); + } + } + /* In flip immediate and pipe splitting case, we need to use GSL * for synchronization. Only do setup on locking and on flip type change. */ @@ -1302,6 +1366,18 @@ static void dcn20_apply_ctx_for_surface( if (!top_pipe_to_program) return; + /* Carry over GSL groups in case the context is changing. */ + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; + struct pipe_ctx *old_pipe_ctx = + &dc->current_state->res_ctx.pipe_ctx[i]; + + if (pipe_ctx->stream == stream && + pipe_ctx->stream == old_pipe_ctx->stream) + pipe_ctx->stream_res.gsl_group = + old_pipe_ctx->stream_res.gsl_group; + } + tg = top_pipe_to_program->stream_res.tg; interdependent_update = top_pipe_to_program->plane_state && @@ -1387,16 +1463,16 @@ void dcn20_prepare_bandwidth( { struct hubbub *hubbub = dc->res_pool->hubbub; + dc->clk_mgr->funcs->update_clocks( + dc->clk_mgr, + context, + false); + /* program dchubbub watermarks */ hubbub->funcs->program_watermarks(hubbub, &context->bw_ctx.bw.dcn.watermarks, dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000, false); - - dc->clk_mgr->funcs->update_clocks( - dc->clk_mgr, - context, - false); } void dcn20_optimize_bandwidth( @@ -1740,8 +1816,12 @@ static void dcn20_reset_back_end_for_pipe( else if (pipe_ctx->stream_res.audio) { dc->hwss.disable_audio_stream(pipe_ctx, FREE_ACQUIRED_RESOURCE); } - } +#ifdef CONFIG_DRM_AMD_DC_DSC_SUPPORT + else if (pipe_ctx->stream_res.dsc) { + dp_set_dsc_enable(pipe_ctx, false); + } +#endif /* by upper caller loop, parent pipe: pipe0, will be reset last. * back end share by all pipes and will be disable only when disable @@ -1803,7 +1883,7 @@ static void dcn20_update_mpcc(struct dc *dc, struct pipe_ctx *pipe_ctx) { struct hubp *hubp = pipe_ctx->plane_res.hubp; struct mpcc_blnd_cfg blnd_cfg = { {0} }; - bool per_pixel_alpha = pipe_ctx->plane_state->per_pixel_alpha && pipe_ctx->bottom_pipe; + bool per_pixel_alpha = pipe_ctx->plane_state->per_pixel_alpha; int mpcc_id; struct mpcc *new_mpcc; struct mpc *mpc = dc->res_pool->mpc; @@ -1996,6 +2076,78 @@ static void dcn20_set_flip_control_gsl( } +static void dcn20_enable_stream(struct pipe_ctx *pipe_ctx) +{ + enum dc_lane_count lane_count = + pipe_ctx->stream->link->cur_link_settings.lane_count; + + struct dc_crtc_timing *timing = &pipe_ctx->stream->timing; + struct dc_link *link = pipe_ctx->stream->link; + + uint32_t active_total_with_borders; + uint32_t early_control = 0; + struct timing_generator *tg = pipe_ctx->stream_res.tg; + + /* For MST, there are multiply stream go to only one link. + * connect DIG back_end to front_end while enable_stream and + * disconnect them during disable_stream + * BY this, it is logic clean to separate stream and link + */ + link->link_enc->funcs->connect_dig_be_to_fe(link->link_enc, + pipe_ctx->stream_res.stream_enc->id, true); + + if (link->dc->hwss.program_dmdata_engine) + link->dc->hwss.program_dmdata_engine(pipe_ctx); + + link->dc->hwss.update_info_frame(pipe_ctx); + + /* enable early control to avoid corruption on DP monitor*/ + active_total_with_borders = + timing->h_addressable + + timing->h_border_left + + timing->h_border_right; + + if (lane_count != 0) + early_control = active_total_with_borders % lane_count; + + if (early_control == 0) + early_control = lane_count; + + tg->funcs->set_early_control(tg, early_control); + + /* enable audio only within mode set */ + if (pipe_ctx->stream_res.audio != NULL) { + if (dc_is_dp_signal(pipe_ctx->stream->signal)) + pipe_ctx->stream_res.stream_enc->funcs->dp_audio_enable(pipe_ctx->stream_res.stream_enc); + } +} + +static void dcn20_program_dmdata_engine(struct pipe_ctx *pipe_ctx) +{ + struct dc_stream_state *stream = pipe_ctx->stream; + struct hubp *hubp = pipe_ctx->plane_res.hubp; + bool enable = false; + struct stream_encoder *stream_enc = pipe_ctx->stream_res.stream_enc; + enum dynamic_metadata_mode mode = dc_is_dp_signal(stream->signal) + ? dmdata_dp + : dmdata_hdmi; + + /* if using dynamic meta, don't set up generic infopackets */ + if (pipe_ctx->stream->dmdata_address.quad_part != 0) { + pipe_ctx->stream_res.encoder_info_frame.hdrsmd.valid = false; + enable = true; + } + + if (!hubp) + return; + + if (!stream_enc || !stream_enc->funcs->set_dynamic_metadata) + return; + + stream_enc->funcs->set_dynamic_metadata(stream_enc, enable, + hubp->inst, mode); +} + void dcn20_hw_sequencer_construct(struct dc *dc) { dcn10_hw_sequencer_construct(dc); @@ -2020,6 +2172,8 @@ void dcn20_hw_sequencer_construct(struct dc *dc) dc->hwss.update_odm = dcn20_update_odm; dc->hwss.blank_pixel_data = dcn20_blank_pixel_data; dc->hwss.dmdata_status_done = dcn20_dmdata_status_done; + dc->hwss.program_dmdata_engine = dcn20_program_dmdata_engine; + dc->hwss.enable_stream = dcn20_enable_stream; dc->hwss.disable_stream = dcn20_disable_stream; dc->hwss.init_sys_ctx = dcn20_init_sys_ctx; dc->hwss.init_vm_ctx = dcn20_init_vm_ctx; diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.h b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.h index 2b0409454073..689c2765b071 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.h @@ -91,13 +91,9 @@ void dcn20_pipe_control_lock_global( void dcn20_setup_gsl_group_as_lock(const struct dc *dc, struct pipe_ctx *pipe_ctx, bool enable); -void dcn20_pipe_control_lock( - struct dc *dc, - struct pipe_ctx *pipe, - bool lock); -void dcn20_disable_plane(struct dc *dc, struct pipe_ctx *pipe_ctx); -void dcn20_enable_plane( - struct dc *dc, - struct pipe_ctx *pipe_ctx, - struct dc_state *context); +void dcn20_dccg_init(struct dce_hwseq *hws); +void dcn20_init_blank( + struct dc *dc, + struct timing_generator *tg); +void dcn20_display_init(struct dc *dc); #endif /* __DC_HWSS_DCN20_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_mpc.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_mpc.c index 240749e4cf83..67f0128f0b38 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_mpc.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_mpc.c @@ -368,6 +368,11 @@ void apply_DEDCN20_305_wa( { struct dcn20_mpc *mpc20 = TO_DCN20_MPC(mpc); + if (mpc->ctx->dc->debug.cm_in_bypass) { + REG_SET(MPCC_OGAM_MODE[mpcc_id], 0, MPCC_OGAM_MODE, 0); + return; + } + if (mpc->ctx->dc->work_arounds.dedcn20_305_wa == false) { /*hw fixed in new review*/ return; @@ -390,10 +395,16 @@ void mpc2_set_output_gamma( enum dc_lut_mode next_mode; struct dcn20_mpc *mpc20 = TO_DCN20_MPC(mpc); + if (mpc->ctx->dc->debug.cm_in_bypass) { + REG_SET(MPCC_OGAM_MODE[mpcc_id], 0, MPCC_OGAM_MODE, 0); + return; + } + if (params == NULL) { REG_SET(MPCC_OGAM_MODE[mpcc_id], 0, MPCC_OGAM_MODE, 0); return; } + current_mode = mpc20_get_ogam_current(mpc, mpcc_id); if (current_mode == LUT_BYPASS || current_mode == LUT_RAM_A) next_mode = LUT_RAM_B; @@ -435,23 +446,22 @@ void mpc2_assert_mpcc_idle_before_connect(struct mpc *mpc, int mpcc_id) { struct dcn20_mpc *mpc20 = TO_DCN20_MPC(mpc); unsigned int top_sel, mpc_busy, mpc_idle, mpc_disabled; - REG_GET(MPCC_STATUS[mpcc_id], MPCC_DISABLED, &mpc_disabled); - - if (mpc_disabled) { - ASSERT(0); - return; - } REG_GET(MPCC_TOP_SEL[mpcc_id], MPCC_TOP_SEL, &top_sel); - if (top_sel == 0xf) { - REG_GET_2(MPCC_STATUS[mpcc_id], - MPCC_BUSY, &mpc_busy, - MPCC_IDLE, &mpc_idle); + REG_GET_3(MPCC_STATUS[mpcc_id], + MPCC_BUSY, &mpc_busy, + MPCC_IDLE, &mpc_idle, + MPCC_DISABLED, &mpc_disabled); - ASSERT(mpc_busy == 0); - ASSERT(mpc_idle == 1); + if (top_sel == 0xf) { + ASSERT(!mpc_busy); + ASSERT(mpc_idle); + ASSERT(mpc_disabled); + } else { + ASSERT(!mpc_disabled); + ASSERT(!mpc_idle); } } diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_optc.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_optc.c index 26a66ccf6e72..1ae973962d53 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_optc.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_optc.c @@ -535,7 +535,7 @@ void dcn20_timing_generator_init(struct optc *optc1) optc1->min_h_blank = 32; optc1->min_v_blank = 3; optc1->min_v_blank_interlace = 5; - optc1->min_h_sync_width = 8; + optc1->min_h_sync_width = 4;// Minimum HSYNC = 8 pixels asked By HW in the first place for no actual reason. Oculus Rift S will not light up with 8 as it's hsyncWidth is 6. Changing it to 4 to fix that issue. optc1->min_v_sync_width = 1; optc1->comb_opp_id = 0xf; } diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c index d200bc3cec71..e90b6bcad05b 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c @@ -82,7 +82,7 @@ #include "amdgpu_socbb.h" -#define SOC_BOUNDING_BOX_VALID false +#define SOC_BOUNDING_BOX_VALID true #define DC_LOGGER_INIT(logger) struct _vcs_dpi_ip_params_st dcn2_0_ip = { @@ -156,7 +156,117 @@ struct _vcs_dpi_ip_params_st dcn2_0_ip = { .xfc_fill_constant_bytes = 0, }; -struct _vcs_dpi_soc_bounding_box_st dcn2_0_soc = { 0 }; +struct _vcs_dpi_soc_bounding_box_st dcn2_0_soc = { + /* Defaults that get patched on driver load from firmware. */ + .clock_limits = { + { + .state = 0, + .dcfclk_mhz = 560.0, + .fabricclk_mhz = 560.0, + .dispclk_mhz = 513.0, + .dppclk_mhz = 513.0, + .phyclk_mhz = 540.0, + .socclk_mhz = 560.0, + .dscclk_mhz = 171.0, + .dram_speed_mts = 8960.0, + }, + { + .state = 1, + .dcfclk_mhz = 694.0, + .fabricclk_mhz = 694.0, + .dispclk_mhz = 642.0, + .dppclk_mhz = 642.0, + .phyclk_mhz = 600.0, + .socclk_mhz = 694.0, + .dscclk_mhz = 214.0, + .dram_speed_mts = 11104.0, + }, + { + .state = 2, + .dcfclk_mhz = 875.0, + .fabricclk_mhz = 875.0, + .dispclk_mhz = 734.0, + .dppclk_mhz = 734.0, + .phyclk_mhz = 810.0, + .socclk_mhz = 875.0, + .dscclk_mhz = 245.0, + .dram_speed_mts = 14000.0, + }, + { + .state = 3, + .dcfclk_mhz = 1000.0, + .fabricclk_mhz = 1000.0, + .dispclk_mhz = 1100.0, + .dppclk_mhz = 1100.0, + .phyclk_mhz = 810.0, + .socclk_mhz = 1000.0, + .dscclk_mhz = 367.0, + .dram_speed_mts = 16000.0, + }, + { + .state = 4, + .dcfclk_mhz = 1200.0, + .fabricclk_mhz = 1200.0, + .dispclk_mhz = 1284.0, + .dppclk_mhz = 1284.0, + .phyclk_mhz = 810.0, + .socclk_mhz = 1200.0, + .dscclk_mhz = 428.0, + .dram_speed_mts = 16000.0, + }, + /*Extra state, no dispclk ramping*/ + { + .state = 5, + .dcfclk_mhz = 1200.0, + .fabricclk_mhz = 1200.0, + .dispclk_mhz = 1284.0, + .dppclk_mhz = 1284.0, + .phyclk_mhz = 810.0, + .socclk_mhz = 1200.0, + .dscclk_mhz = 428.0, + .dram_speed_mts = 16000.0, + }, + }, + .num_states = 5, + .sr_exit_time_us = 8.6, + .sr_enter_plus_exit_time_us = 10.9, + .urgent_latency_us = 4.0, + .urgent_latency_pixel_data_only_us = 4.0, + .urgent_latency_pixel_mixed_with_vm_data_us = 4.0, + .urgent_latency_vm_data_only_us = 4.0, + .urgent_out_of_order_return_per_channel_pixel_only_bytes = 4096, + .urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = 4096, + .urgent_out_of_order_return_per_channel_vm_only_bytes = 4096, + .pct_ideal_dram_sdp_bw_after_urgent_pixel_only = 40.0, + .pct_ideal_dram_sdp_bw_after_urgent_pixel_and_vm = 40.0, + .pct_ideal_dram_sdp_bw_after_urgent_vm_only = 40.0, + .max_avg_sdp_bw_use_normal_percent = 40.0, + .max_avg_dram_bw_use_normal_percent = 40.0, + .writeback_latency_us = 12.0, + .ideal_dram_bw_after_urgent_percent = 40.0, + .max_request_size_bytes = 256, + .dram_channel_width_bytes = 2, + .fabric_datapath_to_dcn_data_return_bytes = 64, + .dcn_downspread_percent = 0.5, + .downspread_percent = 0.38, + .dram_page_open_time_ns = 50.0, + .dram_rw_turnaround_time_ns = 17.5, + .dram_return_buffer_per_channel_bytes = 8192, + .round_trip_ping_latency_dcfclk_cycles = 131, + .urgent_out_of_order_return_per_channel_bytes = 256, + .channel_interleave_bytes = 256, + .num_banks = 8, + .num_chans = 16, + .vmm_page_size_bytes = 4096, + .dram_clock_change_latency_us = 404.0, + .dummy_pstate_latency_us = 5.0, + .writeback_dram_clock_change_latency_us = 23.0, + .return_bus_width_bytes = 64, + .dispclk_dppclk_vco_speed_mhz = 3850, + .xfc_bus_transport_time_us = 20, + .xfc_xbuf_latency_tolerance_us = 4, + .use_urgent_burst_bw = 0 +}; #ifndef mmDP0_DP_DPHY_INTERNAL_CTRL @@ -695,6 +805,16 @@ static const struct dc_plane_cap plane_cap = { .fp16 = 1 } }; +static const struct resource_caps res_cap_nv14 = { + .num_timing_generator = 5, + .num_opp = 5, + .num_video_plane = 5, + .num_audio = 6, + .num_stream_encoder = 5, + .num_pll = 5, + .num_dwb = 0, + .num_ddc = 5, +}; static const struct dc_debug_options debug_defaults_drv = { .disable_dmcu = true, @@ -1812,13 +1932,13 @@ int dcn20_populate_dml_pipes_from_context( pipes[pipe_cnt].pipe.src.viewport_height = scl->viewport.height; pipes[pipe_cnt].pipe.src.viewport_height_c = scl->viewport_c.height; if (pln->format >= SURFACE_PIXEL_FORMAT_VIDEO_BEGIN) { - pipes[pipe_cnt].pipe.src.data_pitch = pln->plane_size.video.luma_pitch; - pipes[pipe_cnt].pipe.src.data_pitch_c = pln->plane_size.video.chroma_pitch; - pipes[pipe_cnt].pipe.src.meta_pitch = pln->dcc.video.meta_pitch_l; - pipes[pipe_cnt].pipe.src.meta_pitch_c = pln->dcc.video.meta_pitch_c; + pipes[pipe_cnt].pipe.src.data_pitch = pln->plane_size.surface_pitch; + pipes[pipe_cnt].pipe.src.data_pitch_c = pln->plane_size.chroma_pitch; + pipes[pipe_cnt].pipe.src.meta_pitch = pln->dcc.meta_pitch; + pipes[pipe_cnt].pipe.src.meta_pitch_c = pln->dcc.meta_pitch_c; } else { - pipes[pipe_cnt].pipe.src.data_pitch = pln->plane_size.grph.surface_pitch; - pipes[pipe_cnt].pipe.src.meta_pitch = pln->dcc.grph.meta_pitch; + pipes[pipe_cnt].pipe.src.data_pitch = pln->plane_size.surface_pitch; + pipes[pipe_cnt].pipe.src.meta_pitch = pln->dcc.meta_pitch; } pipes[pipe_cnt].pipe.src.dcc = pln->dcc.enable; pipes[pipe_cnt].pipe.dest.recout_width = scl->recout.width; @@ -2008,15 +2128,17 @@ static bool dcn20_validate_dsc(struct dc *dc, struct dc_state *new_ctx) } #endif -bool dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate) +bool dcn20_fast_validate_bw( + struct dc *dc, + struct dc_state *context, + display_e2e_pipe_params_st *pipes, + int *pipe_cnt_out, + int *pipe_split_from, + int *vlevel_out) { bool out = false; - BW_VAL_TRACE_SETUP(); - int pipe_cnt, i, pipe_idx, vlevel, vlevel_unsplit; - int pipe_split_from[MAX_PIPES]; bool odm_capable = context->bw_ctx.dml.ip.odm_capable; bool force_split = false; #ifdef CONFIG_DRM_AMD_DC_DSC_SUPPORT @@ -2024,10 +2146,7 @@ bool dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, #endif int split_threshold = dc->res_pool->pipe_count / 2; bool avoid_split = dc->debug.pipe_split_policy != MPC_SPLIT_DYNAMIC; - display_e2e_pipe_params_st *pipes = kzalloc(dc->res_pool->pipe_count * sizeof(display_e2e_pipe_params_st), GFP_KERNEL); - DC_LOGGER_INIT(dc->ctx->logger); - BW_VAL_TRACE_COUNT(); ASSERT(pipes); if (!pipes) @@ -2066,8 +2185,9 @@ bool dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, pipe_cnt = dcn20_populate_dml_pipes_from_context(dc, &context->res_ctx, pipes); + *pipe_cnt_out = pipe_cnt; + if (!pipe_cnt) { - BW_VAL_TRACE_SKIP(pass); out = true; goto validate_out; } @@ -2160,10 +2280,6 @@ bool dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, } if (force_split && context->bw_ctx.dml.vba.NoOfDPP[vlevel][context->bw_ctx.dml.vba.maxMpcComb][pipe_idx] == 1) context->bw_ctx.dml.vba.RequiredDPPCLK[vlevel][context->bw_ctx.dml.vba.maxMpcComb][pipe_idx] /= 2; - if (dc->config.forced_clocks == true) { - context->bw_ctx.dml.vba.RequiredDPPCLK[vlevel][context->bw_ctx.dml.vba.maxMpcComb][pipe_idx] = - context->bw_ctx.dml.soc.clock_limits[0].dppclk_mhz; - } if (!pipe->top_pipe && !pipe->plane_state && context->bw_ctx.dml.vba.ODMCombineEnabled[pipe_idx]) { hsplit_pipe = find_idle_secondary_pipe(&context->res_ctx, dc->res_pool, pipe); ASSERT(hsplit_pipe); @@ -2232,101 +2348,132 @@ bool dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, } #endif - BW_VAL_TRACE_END_VOLTAGE_LEVEL(); + *vlevel_out = vlevel; - if (fast_validate) { - BW_VAL_TRACE_SKIP(fast); - out = true; - goto validate_out; - } + out = true; + goto validate_out; + +validate_fail: + out = false; + +validate_out: + return out; +} + +void dcn20_calculate_wm( + struct dc *dc, struct dc_state *context, + display_e2e_pipe_params_st *pipes, + int *out_pipe_cnt, + int *pipe_split_from, + int vlevel) +{ + int pipe_cnt, i, pipe_idx; for (i = 0, pipe_idx = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) { - if (!context->res_ctx.pipe_ctx[i].stream) - continue; + if (!context->res_ctx.pipe_ctx[i].stream) + continue; - pipes[pipe_cnt].clks_cfg.refclk_mhz = dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000.0; - pipes[pipe_cnt].clks_cfg.dispclk_mhz = context->bw_ctx.dml.vba.RequiredDISPCLK[vlevel][context->bw_ctx.dml.vba.maxMpcComb]; + pipes[pipe_cnt].clks_cfg.refclk_mhz = dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000.0; + pipes[pipe_cnt].clks_cfg.dispclk_mhz = context->bw_ctx.dml.vba.RequiredDISPCLK[vlevel][context->bw_ctx.dml.vba.maxMpcComb]; - if (pipe_split_from[i] < 0) { - pipes[pipe_cnt].clks_cfg.dppclk_mhz = - context->bw_ctx.dml.vba.RequiredDPPCLK[vlevel][context->bw_ctx.dml.vba.maxMpcComb][pipe_idx]; - if (context->bw_ctx.dml.vba.BlendingAndTiming[pipe_idx] == pipe_idx) - pipes[pipe_cnt].pipe.dest.odm_combine = - context->bw_ctx.dml.vba.ODMCombineEnablePerState[vlevel][pipe_idx]; - else - pipes[pipe_cnt].pipe.dest.odm_combine = 0; - pipe_idx++; - } else { - pipes[pipe_cnt].clks_cfg.dppclk_mhz = - context->bw_ctx.dml.vba.RequiredDPPCLK[vlevel][context->bw_ctx.dml.vba.maxMpcComb][pipe_split_from[i]]; - if (context->bw_ctx.dml.vba.BlendingAndTiming[pipe_split_from[i]] == pipe_split_from[i]) - pipes[pipe_cnt].pipe.dest.odm_combine = - context->bw_ctx.dml.vba.ODMCombineEnablePerState[vlevel][pipe_split_from[i]]; + if (pipe_split_from[i] < 0) { + pipes[pipe_cnt].clks_cfg.dppclk_mhz = + context->bw_ctx.dml.vba.RequiredDPPCLK[vlevel][context->bw_ctx.dml.vba.maxMpcComb][pipe_idx]; + if (context->bw_ctx.dml.vba.BlendingAndTiming[pipe_idx] == pipe_idx) + pipes[pipe_cnt].pipe.dest.odm_combine = + context->bw_ctx.dml.vba.ODMCombineEnablePerState[vlevel][pipe_idx]; + else + pipes[pipe_cnt].pipe.dest.odm_combine = 0; + pipe_idx++; + } else { + pipes[pipe_cnt].clks_cfg.dppclk_mhz = + context->bw_ctx.dml.vba.RequiredDPPCLK[vlevel][context->bw_ctx.dml.vba.maxMpcComb][pipe_split_from[i]]; + if (context->bw_ctx.dml.vba.BlendingAndTiming[pipe_split_from[i]] == pipe_split_from[i]) + pipes[pipe_cnt].pipe.dest.odm_combine = + context->bw_ctx.dml.vba.ODMCombineEnablePerState[vlevel][pipe_split_from[i]]; + else + pipes[pipe_cnt].pipe.dest.odm_combine = 0; + } + + if (dc->config.forced_clocks) { + pipes[pipe_cnt].clks_cfg.dispclk_mhz = context->bw_ctx.dml.soc.clock_limits[0].dispclk_mhz; + pipes[pipe_cnt].clks_cfg.dppclk_mhz = context->bw_ctx.dml.soc.clock_limits[0].dppclk_mhz; + } + if (dc->debug.min_disp_clk_khz > pipes[pipe_cnt].clks_cfg.dispclk_mhz * 1000) + pipes[pipe_cnt].clks_cfg.dispclk_mhz = dc->debug.min_disp_clk_khz / 1000.0; + if (dc->debug.min_dpp_clk_khz > pipes[pipe_cnt].clks_cfg.dppclk_mhz * 1000) + pipes[pipe_cnt].clks_cfg.dppclk_mhz = dc->debug.min_dpp_clk_khz / 1000.0; + + pipe_cnt++; + } + + if (pipe_cnt != pipe_idx) { + if (dc->res_pool->funcs->populate_dml_pipes) + pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, + &context->res_ctx, pipes); else - pipes[pipe_cnt].pipe.dest.odm_combine = 0; + pipe_cnt = dcn20_populate_dml_pipes_from_context(dc, + &context->res_ctx, pipes); } - if (dc->config.forced_clocks) { - pipes[pipe_cnt].clks_cfg.dispclk_mhz = context->bw_ctx.dml.soc.clock_limits[0].dispclk_mhz; - pipes[pipe_cnt].clks_cfg.dppclk_mhz = context->bw_ctx.dml.soc.clock_limits[0].dppclk_mhz; + + *out_pipe_cnt = pipe_cnt; + + pipes[0].clks_cfg.voltage = vlevel; + pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel].dcfclk_mhz; + pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel].socclk_mhz; + + /* only pipe 0 is read for voltage and dcf/soc clocks */ + if (vlevel < 1) { + pipes[0].clks_cfg.voltage = 1; + pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[1].dcfclk_mhz; + pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[1].socclk_mhz; } - pipe_cnt++; - } + context->bw_ctx.bw.dcn.watermarks.b.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.b.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + + if (vlevel < 2) { + pipes[0].clks_cfg.voltage = 2; + pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[2].dcfclk_mhz; + pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[2].socclk_mhz; + } + context->bw_ctx.bw.dcn.watermarks.c.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.c.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + + if (vlevel < 3) { + pipes[0].clks_cfg.voltage = 3; + pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[2].dcfclk_mhz; + pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[2].socclk_mhz; + } + context->bw_ctx.bw.dcn.watermarks.d.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.d.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + + pipes[0].clks_cfg.voltage = vlevel; + pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel].dcfclk_mhz; + pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel].socclk_mhz; + context->bw_ctx.bw.dcn.watermarks.a.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; + context->bw_ctx.bw.dcn.watermarks.a.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; +} + +void dcn20_calculate_dlg_params( + struct dc *dc, struct dc_state *context, + display_e2e_pipe_params_st *pipes, + int pipe_cnt, + int vlevel) +{ + int i, pipe_idx; - if (pipe_cnt != pipe_idx) { - if (dc->res_pool->funcs->populate_dml_pipes) - pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, - &context->res_ctx, pipes); - else - pipe_cnt = dcn20_populate_dml_pipes_from_context(dc, - &context->res_ctx, pipes); - } - - pipes[0].clks_cfg.voltage = vlevel; - pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel].dcfclk_mhz; - pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel].socclk_mhz; - - /* only pipe 0 is read for voltage and dcf/soc clocks */ - if (vlevel < 1) { - pipes[0].clks_cfg.voltage = 1; - pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[1].dcfclk_mhz; - pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[1].socclk_mhz; - } - context->bw_ctx.bw.dcn.watermarks.b.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.b.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.b.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - - if (vlevel < 2) { - pipes[0].clks_cfg.voltage = 2; - pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[2].dcfclk_mhz; - pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[2].socclk_mhz; - } - context->bw_ctx.bw.dcn.watermarks.c.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.c.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.c.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - - if (vlevel < 3) { - pipes[0].clks_cfg.voltage = 3; - pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[2].dcfclk_mhz; - pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[2].socclk_mhz; - } - context->bw_ctx.bw.dcn.watermarks.d.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.d.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.d.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - - pipes[0].clks_cfg.voltage = vlevel; - pipes[0].clks_cfg.dcfclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel].dcfclk_mhz; - pipes[0].clks_cfg.socclk_mhz = context->bw_ctx.dml.soc.clock_limits[vlevel].socclk_mhz; - context->bw_ctx.bw.dcn.watermarks.a.urgent_ns = get_wm_urgent(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_enter_plus_exit_ns = get_wm_stutter_enter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_exit_ns = get_wm_stutter_exit(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.pstate_change_ns = get_wm_dram_clock_change(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; - context->bw_ctx.bw.dcn.watermarks.a.pte_meta_urgent_ns = get_wm_memory_trip(&context->bw_ctx.dml, pipes, pipe_cnt) * 1000; /* Writeback MCIF_WB arbitration parameters */ dc->res_pool->funcs->set_mcif_arb_params(dc, context, pipes, pipe_cnt); @@ -2341,7 +2488,7 @@ bool dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, != dm_dram_clock_change_unsupported; context->bw_ctx.bw.dcn.clk.dppclk_khz = 0; - BW_VAL_TRACE_END_WATERMARKS(); + for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) { if (!context->res_ctx.pipe_ctx[i].stream) @@ -2361,6 +2508,11 @@ bool dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, context->res_ctx.pipe_ctx[i].pipe_dlg_param = pipes[pipe_idx].pipe.dest; pipe_idx++; } + /*save a original dppclock copy*/ + context->bw_ctx.bw.dcn.clk.bw_dppclk_khz = context->bw_ctx.bw.dcn.clk.dppclk_khz; + context->bw_ctx.bw.dcn.clk.bw_dispclk_khz = context->bw_ctx.bw.dcn.clk.dispclk_khz; + context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz = context->bw_ctx.dml.soc.clock_limits[vlevel].dppclk_mhz*1000; + context->bw_ctx.bw.dcn.clk.max_supported_dispclk_khz = context->bw_ctx.dml.soc.clock_limits[vlevel].dispclk_mhz*1000; for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) { bool cstate_en = context->bw_ctx.dml.vba.PrefetchMode[vlevel][context->bw_ctx.dml.vba.maxMpcComb] != 2; @@ -2383,8 +2535,43 @@ bool dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, pipes[pipe_idx].pipe); pipe_idx++; } +} + +static bool dcn20_validate_bandwidth_internal(struct dc *dc, struct dc_state *context, + bool fast_validate) +{ + bool out = false; + + BW_VAL_TRACE_SETUP(); + + int vlevel = 0; + int pipe_split_from[MAX_PIPES]; + int pipe_cnt = 0; + display_e2e_pipe_params_st *pipes = kzalloc(dc->res_pool->pipe_count * sizeof(display_e2e_pipe_params_st), GFP_KERNEL); + DC_LOGGER_INIT(dc->ctx->logger); + + BW_VAL_TRACE_COUNT(); + + out = dcn20_fast_validate_bw(dc, context, pipes, &pipe_cnt, pipe_split_from, &vlevel); + + if (pipe_cnt == 0) + goto validate_out; + + if (!out) + goto validate_fail; + + BW_VAL_TRACE_END_VOLTAGE_LEVEL(); + + if (fast_validate) { + BW_VAL_TRACE_SKIP(fast); + goto validate_out; + } + + dcn20_calculate_wm(dc, context, pipes, &pipe_cnt, pipe_split_from, vlevel); + dcn20_calculate_dlg_params(dc, context, pipes, pipe_cnt, vlevel); + + BW_VAL_TRACE_END_WATERMARKS(); - out = true; goto validate_out; validate_fail: @@ -2402,6 +2589,62 @@ validate_out: return out; } + +bool dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, + bool fast_validate) +{ + bool voltage_supported = false; + bool full_pstate_supported = false; + bool dummy_pstate_supported = false; + double p_state_latency_us = context->bw_ctx.dml.soc.dram_clock_change_latency_us; + + if (fast_validate) + return dcn20_validate_bandwidth_internal(dc, context, true); + + + // Best case, we support full UCLK switch latency + voltage_supported = dcn20_validate_bandwidth_internal(dc, context, false); + full_pstate_supported = context->bw_ctx.bw.dcn.clk.p_state_change_support; + + if (context->bw_ctx.dml.soc.dummy_pstate_latency_us == 0 || + (voltage_supported && full_pstate_supported)) { + context->bw_ctx.bw.dcn.clk.p_state_change_support = true; + goto restore_dml_state; + } + + // Fallback #1: Try to only support G6 temperature read latency + context->bw_ctx.dml.soc.dram_clock_change_latency_us = context->bw_ctx.dml.soc.dummy_pstate_latency_us; + + voltage_supported = dcn20_validate_bandwidth_internal(dc, context, false); + dummy_pstate_supported = context->bw_ctx.bw.dcn.clk.p_state_change_support; + + if (voltage_supported && dummy_pstate_supported) { + context->bw_ctx.bw.dcn.clk.p_state_change_support = false; + goto restore_dml_state; + } + + // Fallback #2: Retry with "new" DCN20 to support G6 temperature read latency + memcpy (&context->bw_ctx.dml, &dc->work_arounds.alternate_dml, sizeof (struct display_mode_lib)); + context->bw_ctx.dml.soc.dram_clock_change_latency_us = context->bw_ctx.dml.soc.dummy_pstate_latency_us; + + voltage_supported = dcn20_validate_bandwidth_internal(dc, context, false); + dummy_pstate_supported = context->bw_ctx.bw.dcn.clk.p_state_change_support; + + if (voltage_supported && dummy_pstate_supported) { + context->bw_ctx.bw.dcn.clk.p_state_change_support = false; + goto restore_dml_state; + } + + // ERROR: fallback #2 is supposed to always work. + ASSERT(false); + +restore_dml_state: + memcpy(&context->bw_ctx.dml, &dc->dml, sizeof(struct display_mode_lib)); + context->bw_ctx.dml.soc.dram_clock_change_latency_us = p_state_latency_us; + + return voltage_supported; +} + struct pipe_ctx *dcn20_acquire_idle_pipe_for_layer( struct dc_state *state, const struct resource_pool *pool, @@ -2576,9 +2819,6 @@ static void cap_soc_clocks( && max_clocks.uClockInKhz != 0) bb->clock_limits[i].dram_speed_mts = (max_clocks.uClockInKhz / 1000) * 16; - // HACK: Force every uclk to max for now to "disable" uclk switching. - bb->clock_limits[i].dram_speed_mts = (max_clocks.uClockInKhz / 1000) * 16; - if ((bb->clock_limits[i].fabricclk_mhz > (max_clocks.fabricClockInKhz / 1000)) && max_clocks.fabricClockInKhz != 0) bb->clock_limits[i].fabricclk_mhz = (max_clocks.fabricClockInKhz / 1000); @@ -2643,6 +2883,10 @@ static void update_bounding_box(struct dc *dc, struct _vcs_dpi_soc_bounding_box_ if (dc->bb_overrides.min_dcfclk_mhz > 0) min_dcfclk = dc->bb_overrides.min_dcfclk_mhz; + else + // Accounting for SOC/DCF relationship, we can go as high as + // 506Mhz in Vmin. We need to code 507 since SMU will round down to 506. + min_dcfclk = 507; for (i = 0; i < num_states; i++) { int min_fclk_required_by_uclk; @@ -2670,6 +2914,10 @@ static void update_bounding_box(struct dc *dc, struct _vcs_dpi_soc_bounding_box_ num_calculated_states++; } + calculated_states[num_calculated_states - 1].socclk_mhz = max_clocks->socClockInKhz / 1000; + calculated_states[num_calculated_states - 1].fabricclk_mhz = max_clocks->socClockInKhz / 1000; + calculated_states[num_calculated_states - 1].dcfclk_mhz = max_clocks->dcfClockInKhz / 1000; + memcpy(bb->clock_limits, calculated_states, sizeof(bb->clock_limits)); bb->num_states = num_calculated_states; @@ -2786,8 +3034,6 @@ static bool init_soc_bounding_box(struct dc *dc, le32_to_cpu(bb->vmm_page_size_bytes); dcn2_0_soc.dram_clock_change_latency_us = fixed16_to_double_to_cpu(bb->dram_clock_change_latency_us); - // HACK!! Lower uclock latency switch time so we don't switch - dcn2_0_soc.dram_clock_change_latency_us = 10; dcn2_0_soc.writeback_dram_clock_change_latency_us = fixed16_to_double_to_cpu(bb->writeback_dram_clock_change_latency_us); dcn2_0_soc.return_bus_width_bytes = @@ -2829,7 +3075,6 @@ static bool init_soc_bounding_box(struct dc *dc, struct pp_smu_nv_clock_table max_clocks = {0}; unsigned int uclk_states[8] = {0}; unsigned int num_states = 0; - int i; enum pp_smu_status status; bool clock_limits_available = false; bool uclk_states_available = false; @@ -2851,10 +3096,6 @@ static bool init_soc_bounding_box(struct dc *dc, clock_limits_available = (status == PP_SMU_RESULT_OK); } - // HACK: Use the max uclk_states value for all elements. - for (i = 0; i < num_states; i++) - uclk_states[i] = uclk_states[num_states - 1]; - if (clock_limits_available && uclk_states_available && num_states) update_bounding_box(dc, &dcn2_0_soc, &max_clocks, uclk_states, num_states); else if (clock_limits_available) @@ -2878,17 +3119,22 @@ static bool construct( struct irq_service_init_data init_data; ctx->dc_bios->regs = &bios_regs; - - pool->base.res_cap = &res_cap_nv10; pool->base.funcs = &dcn20_res_pool_funcs; + if (ASICREV_IS_NAVI14_M(ctx->asic_id.hw_internal_rev)) { + pool->base.res_cap = &res_cap_nv14; + pool->base.pipe_count = 5; + pool->base.mpcc_count = 5; + } else { + pool->base.res_cap = &res_cap_nv10; + pool->base.pipe_count = 6; + pool->base.mpcc_count = 6; + } /************************************************* * Resource + asic cap harcoding * *************************************************/ pool->base.underlay_pipe_index = NO_UNDERLAY_PIPE; - pool->base.pipe_count = 6; - pool->base.mpcc_count = 6; dc->caps.max_downscale_ratio = 200; dc->caps.i2c_speed_in_khz = 100; dc->caps.max_cursor_size = 256; @@ -2995,6 +3241,7 @@ static bool construct( } dml_init_instance(&dc->dml, &dcn2_0_soc, &dcn2_0_ip, DML_PROJECT_NAVI10); + dml_init_instance(&dc->work_arounds.alternate_dml, &dcn2_0_soc, &dcn2_0_ip, DML_PROJECT_NAVI10v2); if (!dc->debug.disable_pplib_wm_range) { struct pp_smu_wm_range_sets ranges = {0}; diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.h b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.h index b5a75289f444..44f95aa0d61e 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.h +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.h @@ -116,6 +116,18 @@ void dcn20_set_mcif_arb_params( display_e2e_pipe_params_st *pipes, int pipe_cnt); bool dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, bool fast_validate); +bool dcn20_fast_validate_bw( + struct dc *dc, + struct dc_state *context, + display_e2e_pipe_params_st *pipes, + int *pipe_cnt_out, + int *pipe_split_from, + int *vlevel_out); +void dcn20_calculate_dlg_params( + struct dc *dc, struct dc_state *context, + display_e2e_pipe_params_st *pipes, + int pipe_cnt, + int vlevel); enum dc_status dcn20_build_mapped_resource(const struct dc *dc, struct dc_state *context, struct dc_stream_state *stream); enum dc_status dcn20_add_stream_to_ctx(struct dc *dc, struct dc_state *new_ctx, struct dc_stream_state *dc_stream); diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_stream_encoder.c index f5bcffc426b8..5c2b7b54b126 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_stream_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_stream_encoder.c @@ -373,7 +373,7 @@ static void enc2_read_state(struct stream_encoder *enc, struct enc_state *s) * * Ensure the OTG master update lock is set when changing DME configuration. */ -static void enc2_set_dynamic_metadata(struct stream_encoder *enc, +void enc2_set_dynamic_metadata(struct stream_encoder *enc, bool enable_dme, uint32_t hubp_requestor_id, enum dynamic_metadata_mode dmdata_mode) diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_stream_encoder.h b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_stream_encoder.h index 6d40e8c9b78f..3f94a9f13c4a 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_stream_encoder.h +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_stream_encoder.h @@ -104,4 +104,9 @@ void enc2_stream_encoder_dp_unblank( struct stream_encoder *enc, const struct encoder_unblank_param *param); +void enc2_set_dynamic_metadata(struct stream_encoder *enc, + bool enable_dme, + uint32_t hubp_requestor_id, + enum dynamic_metadata_mode dmdata_mode); + #endif /* __DC_STREAM_ENCODER_DCN20_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_vmid.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_vmid.c index 27679ef6ebe8..96c263223315 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_vmid.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_vmid.c @@ -23,6 +23,8 @@ * */ +#include <linux/delay.h> + #include "dcn20_vmid.h" #include "reg_helper.h" @@ -36,6 +38,38 @@ #define FN(reg_name, field_name) \ vmid->shifts->field_name, vmid->masks->field_name +static void dcn20_wait_for_vmid_ready(struct dcn20_vmid *vmid) +{ + /* According the hardware spec, we need to poll for the lowest + * bit of PAGE_TABLE_BASE_ADDR_LO32 = 1 any time a GPUVM + * context is updated. We can't use REG_WAIT here since we + * don't have a seperate field to wait on. + * + * TODO: Confirm timeout / poll interval with hardware team + */ + + int max_times = 10000; + int delay_us = 5; + int i; + + for (i = 0; i < max_times; ++i) { + uint32_t entry_lo32; + + REG_GET(PAGE_TABLE_BASE_ADDR_LO32, + VM_CONTEXT0_PAGE_DIRECTORY_ENTRY_LO32, + &entry_lo32); + + if (entry_lo32 & 0x1) + return; + + udelay(delay_us); + } + + /* VM setup timed out */ + DC_LOG_WARNING("Timeout while waiting for GPUVM context update\n"); + ASSERT(0); +} + void dcn20_vmid_setup(struct dcn20_vmid *vmid, const struct dcn_vmid_page_table_config *config) { REG_SET(PAGE_TABLE_START_ADDR_HI32, 0, @@ -54,6 +88,9 @@ void dcn20_vmid_setup(struct dcn20_vmid *vmid, const struct dcn_vmid_page_table_ REG_SET(PAGE_TABLE_BASE_ADDR_HI32, 0, VM_CONTEXT0_PAGE_DIRECTORY_ENTRY_HI32, (config->page_table_base_addr >> 32) & 0xFFFFFFFF); + /* Note: per hardware spec PAGE_TABLE_BASE_ADDR_LO32 must be programmed last in sequence */ REG_SET(PAGE_TABLE_BASE_ADDR_LO32, 0, VM_CONTEXT0_PAGE_DIRECTORY_ENTRY_LO32, config->page_table_base_addr & 0xFFFFFFFF); + + dcn20_wait_for_vmid_ready(vmid); } |