diff options
Diffstat (limited to 'include/uapi/misc/habanalabs.h')
| -rw-r--r-- | include/uapi/misc/habanalabs.h | 1023 |
1 files changed, 896 insertions, 127 deletions
diff --git a/include/uapi/misc/habanalabs.h b/include/uapi/misc/habanalabs.h index 00b309590499..e00ebe05097d 100644 --- a/include/uapi/misc/habanalabs.h +++ b/include/uapi/misc/habanalabs.h @@ -1,6 +1,6 @@ /* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note * - * Copyright 2016-2020 HabanaLabs, Ltd. + * Copyright 2016-2022 HabanaLabs, Ltd. * All Rights Reserved. * */ @@ -30,6 +30,9 @@ */ #define GAUDI_FIRST_AVAILABLE_W_S_MONITOR 72 +/* Max number of elements in timestamps registration buffers */ +#define TS_MAX_ELEMENTS_NUM (1 << 20) /* 1MB */ + /* * Goya queue Numbering * @@ -182,6 +185,285 @@ enum gaudi_queue_id { }; /* + * In GAUDI2 we have two modes of operation in regard to queues: + * 1. Legacy mode, where each QMAN exposes 4 streams to the user + * 2. F/W mode, where we use F/W to schedule the JOBS to the different queues. + * + * When in legacy mode, the user sends the queue id per JOB according to + * enum gaudi2_queue_id below. + * + * When in F/W mode, the user sends a stream id per Command Submission. The + * stream id is a running number from 0 up to (N-1), where N is the number + * of streams the F/W exposes and is passed to the user in + * struct hl_info_hw_ip_info + */ + +enum gaudi2_queue_id { + GAUDI2_QUEUE_ID_PDMA_0_0 = 0, + GAUDI2_QUEUE_ID_PDMA_0_1 = 1, + GAUDI2_QUEUE_ID_PDMA_0_2 = 2, + GAUDI2_QUEUE_ID_PDMA_0_3 = 3, + GAUDI2_QUEUE_ID_PDMA_1_0 = 4, + GAUDI2_QUEUE_ID_PDMA_1_1 = 5, + GAUDI2_QUEUE_ID_PDMA_1_2 = 6, + GAUDI2_QUEUE_ID_PDMA_1_3 = 7, + GAUDI2_QUEUE_ID_DCORE0_EDMA_0_0 = 8, + GAUDI2_QUEUE_ID_DCORE0_EDMA_0_1 = 9, + GAUDI2_QUEUE_ID_DCORE0_EDMA_0_2 = 10, + GAUDI2_QUEUE_ID_DCORE0_EDMA_0_3 = 11, + GAUDI2_QUEUE_ID_DCORE0_EDMA_1_0 = 12, + GAUDI2_QUEUE_ID_DCORE0_EDMA_1_1 = 13, + GAUDI2_QUEUE_ID_DCORE0_EDMA_1_2 = 14, + GAUDI2_QUEUE_ID_DCORE0_EDMA_1_3 = 15, + GAUDI2_QUEUE_ID_DCORE0_MME_0_0 = 16, + GAUDI2_QUEUE_ID_DCORE0_MME_0_1 = 17, + GAUDI2_QUEUE_ID_DCORE0_MME_0_2 = 18, + GAUDI2_QUEUE_ID_DCORE0_MME_0_3 = 19, + GAUDI2_QUEUE_ID_DCORE0_TPC_0_0 = 20, + GAUDI2_QUEUE_ID_DCORE0_TPC_0_1 = 21, + GAUDI2_QUEUE_ID_DCORE0_TPC_0_2 = 22, + GAUDI2_QUEUE_ID_DCORE0_TPC_0_3 = 23, + GAUDI2_QUEUE_ID_DCORE0_TPC_1_0 = 24, + GAUDI2_QUEUE_ID_DCORE0_TPC_1_1 = 25, + GAUDI2_QUEUE_ID_DCORE0_TPC_1_2 = 26, + GAUDI2_QUEUE_ID_DCORE0_TPC_1_3 = 27, + GAUDI2_QUEUE_ID_DCORE0_TPC_2_0 = 28, + GAUDI2_QUEUE_ID_DCORE0_TPC_2_1 = 29, + GAUDI2_QUEUE_ID_DCORE0_TPC_2_2 = 30, + GAUDI2_QUEUE_ID_DCORE0_TPC_2_3 = 31, + GAUDI2_QUEUE_ID_DCORE0_TPC_3_0 = 32, + GAUDI2_QUEUE_ID_DCORE0_TPC_3_1 = 33, + GAUDI2_QUEUE_ID_DCORE0_TPC_3_2 = 34, + GAUDI2_QUEUE_ID_DCORE0_TPC_3_3 = 35, + GAUDI2_QUEUE_ID_DCORE0_TPC_4_0 = 36, + GAUDI2_QUEUE_ID_DCORE0_TPC_4_1 = 37, + GAUDI2_QUEUE_ID_DCORE0_TPC_4_2 = 38, + GAUDI2_QUEUE_ID_DCORE0_TPC_4_3 = 39, + GAUDI2_QUEUE_ID_DCORE0_TPC_5_0 = 40, + GAUDI2_QUEUE_ID_DCORE0_TPC_5_1 = 41, + GAUDI2_QUEUE_ID_DCORE0_TPC_5_2 = 42, + GAUDI2_QUEUE_ID_DCORE0_TPC_5_3 = 43, + GAUDI2_QUEUE_ID_DCORE0_TPC_6_0 = 44, + GAUDI2_QUEUE_ID_DCORE0_TPC_6_1 = 45, + GAUDI2_QUEUE_ID_DCORE0_TPC_6_2 = 46, + GAUDI2_QUEUE_ID_DCORE0_TPC_6_3 = 47, + GAUDI2_QUEUE_ID_DCORE1_EDMA_0_0 = 48, + GAUDI2_QUEUE_ID_DCORE1_EDMA_0_1 = 49, + GAUDI2_QUEUE_ID_DCORE1_EDMA_0_2 = 50, + GAUDI2_QUEUE_ID_DCORE1_EDMA_0_3 = 51, + GAUDI2_QUEUE_ID_DCORE1_EDMA_1_0 = 52, + GAUDI2_QUEUE_ID_DCORE1_EDMA_1_1 = 53, + GAUDI2_QUEUE_ID_DCORE1_EDMA_1_2 = 54, + GAUDI2_QUEUE_ID_DCORE1_EDMA_1_3 = 55, + GAUDI2_QUEUE_ID_DCORE1_MME_0_0 = 56, + GAUDI2_QUEUE_ID_DCORE1_MME_0_1 = 57, + GAUDI2_QUEUE_ID_DCORE1_MME_0_2 = 58, + GAUDI2_QUEUE_ID_DCORE1_MME_0_3 = 59, + GAUDI2_QUEUE_ID_DCORE1_TPC_0_0 = 60, + GAUDI2_QUEUE_ID_DCORE1_TPC_0_1 = 61, + GAUDI2_QUEUE_ID_DCORE1_TPC_0_2 = 62, + GAUDI2_QUEUE_ID_DCORE1_TPC_0_3 = 63, + GAUDI2_QUEUE_ID_DCORE1_TPC_1_0 = 64, + GAUDI2_QUEUE_ID_DCORE1_TPC_1_1 = 65, + GAUDI2_QUEUE_ID_DCORE1_TPC_1_2 = 66, + GAUDI2_QUEUE_ID_DCORE1_TPC_1_3 = 67, + GAUDI2_QUEUE_ID_DCORE1_TPC_2_0 = 68, + GAUDI2_QUEUE_ID_DCORE1_TPC_2_1 = 69, + GAUDI2_QUEUE_ID_DCORE1_TPC_2_2 = 70, + GAUDI2_QUEUE_ID_DCORE1_TPC_2_3 = 71, + GAUDI2_QUEUE_ID_DCORE1_TPC_3_0 = 72, + GAUDI2_QUEUE_ID_DCORE1_TPC_3_1 = 73, + GAUDI2_QUEUE_ID_DCORE1_TPC_3_2 = 74, + GAUDI2_QUEUE_ID_DCORE1_TPC_3_3 = 75, + GAUDI2_QUEUE_ID_DCORE1_TPC_4_0 = 76, + GAUDI2_QUEUE_ID_DCORE1_TPC_4_1 = 77, + GAUDI2_QUEUE_ID_DCORE1_TPC_4_2 = 78, + GAUDI2_QUEUE_ID_DCORE1_TPC_4_3 = 79, + GAUDI2_QUEUE_ID_DCORE1_TPC_5_0 = 80, + GAUDI2_QUEUE_ID_DCORE1_TPC_5_1 = 81, + GAUDI2_QUEUE_ID_DCORE1_TPC_5_2 = 82, + GAUDI2_QUEUE_ID_DCORE1_TPC_5_3 = 83, + GAUDI2_QUEUE_ID_DCORE2_EDMA_0_0 = 84, + GAUDI2_QUEUE_ID_DCORE2_EDMA_0_1 = 85, + GAUDI2_QUEUE_ID_DCORE2_EDMA_0_2 = 86, + GAUDI2_QUEUE_ID_DCORE2_EDMA_0_3 = 87, + GAUDI2_QUEUE_ID_DCORE2_EDMA_1_0 = 88, + GAUDI2_QUEUE_ID_DCORE2_EDMA_1_1 = 89, + GAUDI2_QUEUE_ID_DCORE2_EDMA_1_2 = 90, + GAUDI2_QUEUE_ID_DCORE2_EDMA_1_3 = 91, + GAUDI2_QUEUE_ID_DCORE2_MME_0_0 = 92, + GAUDI2_QUEUE_ID_DCORE2_MME_0_1 = 93, + GAUDI2_QUEUE_ID_DCORE2_MME_0_2 = 94, + GAUDI2_QUEUE_ID_DCORE2_MME_0_3 = 95, + GAUDI2_QUEUE_ID_DCORE2_TPC_0_0 = 96, + GAUDI2_QUEUE_ID_DCORE2_TPC_0_1 = 97, + GAUDI2_QUEUE_ID_DCORE2_TPC_0_2 = 98, + GAUDI2_QUEUE_ID_DCORE2_TPC_0_3 = 99, + GAUDI2_QUEUE_ID_DCORE2_TPC_1_0 = 100, + GAUDI2_QUEUE_ID_DCORE2_TPC_1_1 = 101, + GAUDI2_QUEUE_ID_DCORE2_TPC_1_2 = 102, + GAUDI2_QUEUE_ID_DCORE2_TPC_1_3 = 103, + GAUDI2_QUEUE_ID_DCORE2_TPC_2_0 = 104, + GAUDI2_QUEUE_ID_DCORE2_TPC_2_1 = 105, + GAUDI2_QUEUE_ID_DCORE2_TPC_2_2 = 106, + GAUDI2_QUEUE_ID_DCORE2_TPC_2_3 = 107, + GAUDI2_QUEUE_ID_DCORE2_TPC_3_0 = 108, + GAUDI2_QUEUE_ID_DCORE2_TPC_3_1 = 109, + GAUDI2_QUEUE_ID_DCORE2_TPC_3_2 = 110, + GAUDI2_QUEUE_ID_DCORE2_TPC_3_3 = 111, + GAUDI2_QUEUE_ID_DCORE2_TPC_4_0 = 112, + GAUDI2_QUEUE_ID_DCORE2_TPC_4_1 = 113, + GAUDI2_QUEUE_ID_DCORE2_TPC_4_2 = 114, + GAUDI2_QUEUE_ID_DCORE2_TPC_4_3 = 115, + GAUDI2_QUEUE_ID_DCORE2_TPC_5_0 = 116, + GAUDI2_QUEUE_ID_DCORE2_TPC_5_1 = 117, + GAUDI2_QUEUE_ID_DCORE2_TPC_5_2 = 118, + GAUDI2_QUEUE_ID_DCORE2_TPC_5_3 = 119, + GAUDI2_QUEUE_ID_DCORE3_EDMA_0_0 = 120, + GAUDI2_QUEUE_ID_DCORE3_EDMA_0_1 = 121, + GAUDI2_QUEUE_ID_DCORE3_EDMA_0_2 = 122, + GAUDI2_QUEUE_ID_DCORE3_EDMA_0_3 = 123, + GAUDI2_QUEUE_ID_DCORE3_EDMA_1_0 = 124, + GAUDI2_QUEUE_ID_DCORE3_EDMA_1_1 = 125, + GAUDI2_QUEUE_ID_DCORE3_EDMA_1_2 = 126, + GAUDI2_QUEUE_ID_DCORE3_EDMA_1_3 = 127, + GAUDI2_QUEUE_ID_DCORE3_MME_0_0 = 128, + GAUDI2_QUEUE_ID_DCORE3_MME_0_1 = 129, + GAUDI2_QUEUE_ID_DCORE3_MME_0_2 = 130, + GAUDI2_QUEUE_ID_DCORE3_MME_0_3 = 131, + GAUDI2_QUEUE_ID_DCORE3_TPC_0_0 = 132, + GAUDI2_QUEUE_ID_DCORE3_TPC_0_1 = 133, + GAUDI2_QUEUE_ID_DCORE3_TPC_0_2 = 134, + GAUDI2_QUEUE_ID_DCORE3_TPC_0_3 = 135, + GAUDI2_QUEUE_ID_DCORE3_TPC_1_0 = 136, + GAUDI2_QUEUE_ID_DCORE3_TPC_1_1 = 137, + GAUDI2_QUEUE_ID_DCORE3_TPC_1_2 = 138, + GAUDI2_QUEUE_ID_DCORE3_TPC_1_3 = 139, + GAUDI2_QUEUE_ID_DCORE3_TPC_2_0 = 140, + GAUDI2_QUEUE_ID_DCORE3_TPC_2_1 = 141, + GAUDI2_QUEUE_ID_DCORE3_TPC_2_2 = 142, + GAUDI2_QUEUE_ID_DCORE3_TPC_2_3 = 143, + GAUDI2_QUEUE_ID_DCORE3_TPC_3_0 = 144, + GAUDI2_QUEUE_ID_DCORE3_TPC_3_1 = 145, + GAUDI2_QUEUE_ID_DCORE3_TPC_3_2 = 146, + GAUDI2_QUEUE_ID_DCORE3_TPC_3_3 = 147, + GAUDI2_QUEUE_ID_DCORE3_TPC_4_0 = 148, + GAUDI2_QUEUE_ID_DCORE3_TPC_4_1 = 149, + GAUDI2_QUEUE_ID_DCORE3_TPC_4_2 = 150, + GAUDI2_QUEUE_ID_DCORE3_TPC_4_3 = 151, + GAUDI2_QUEUE_ID_DCORE3_TPC_5_0 = 152, + GAUDI2_QUEUE_ID_DCORE3_TPC_5_1 = 153, + GAUDI2_QUEUE_ID_DCORE3_TPC_5_2 = 154, + GAUDI2_QUEUE_ID_DCORE3_TPC_5_3 = 155, + GAUDI2_QUEUE_ID_NIC_0_0 = 156, + GAUDI2_QUEUE_ID_NIC_0_1 = 157, + GAUDI2_QUEUE_ID_NIC_0_2 = 158, + GAUDI2_QUEUE_ID_NIC_0_3 = 159, + GAUDI2_QUEUE_ID_NIC_1_0 = 160, + GAUDI2_QUEUE_ID_NIC_1_1 = 161, + GAUDI2_QUEUE_ID_NIC_1_2 = 162, + GAUDI2_QUEUE_ID_NIC_1_3 = 163, + GAUDI2_QUEUE_ID_NIC_2_0 = 164, + GAUDI2_QUEUE_ID_NIC_2_1 = 165, + GAUDI2_QUEUE_ID_NIC_2_2 = 166, + GAUDI2_QUEUE_ID_NIC_2_3 = 167, + GAUDI2_QUEUE_ID_NIC_3_0 = 168, + GAUDI2_QUEUE_ID_NIC_3_1 = 169, + GAUDI2_QUEUE_ID_NIC_3_2 = 170, + GAUDI2_QUEUE_ID_NIC_3_3 = 171, + GAUDI2_QUEUE_ID_NIC_4_0 = 172, + GAUDI2_QUEUE_ID_NIC_4_1 = 173, + GAUDI2_QUEUE_ID_NIC_4_2 = 174, + GAUDI2_QUEUE_ID_NIC_4_3 = 175, + GAUDI2_QUEUE_ID_NIC_5_0 = 176, + GAUDI2_QUEUE_ID_NIC_5_1 = 177, + GAUDI2_QUEUE_ID_NIC_5_2 = 178, + GAUDI2_QUEUE_ID_NIC_5_3 = 179, + GAUDI2_QUEUE_ID_NIC_6_0 = 180, + GAUDI2_QUEUE_ID_NIC_6_1 = 181, + GAUDI2_QUEUE_ID_NIC_6_2 = 182, + GAUDI2_QUEUE_ID_NIC_6_3 = 183, + GAUDI2_QUEUE_ID_NIC_7_0 = 184, + GAUDI2_QUEUE_ID_NIC_7_1 = 185, + GAUDI2_QUEUE_ID_NIC_7_2 = 186, + GAUDI2_QUEUE_ID_NIC_7_3 = 187, + GAUDI2_QUEUE_ID_NIC_8_0 = 188, + GAUDI2_QUEUE_ID_NIC_8_1 = 189, + GAUDI2_QUEUE_ID_NIC_8_2 = 190, + GAUDI2_QUEUE_ID_NIC_8_3 = 191, + GAUDI2_QUEUE_ID_NIC_9_0 = 192, + GAUDI2_QUEUE_ID_NIC_9_1 = 193, + GAUDI2_QUEUE_ID_NIC_9_2 = 194, + GAUDI2_QUEUE_ID_NIC_9_3 = 195, + GAUDI2_QUEUE_ID_NIC_10_0 = 196, + GAUDI2_QUEUE_ID_NIC_10_1 = 197, + GAUDI2_QUEUE_ID_NIC_10_2 = 198, + GAUDI2_QUEUE_ID_NIC_10_3 = 199, + GAUDI2_QUEUE_ID_NIC_11_0 = 200, + GAUDI2_QUEUE_ID_NIC_11_1 = 201, + GAUDI2_QUEUE_ID_NIC_11_2 = 202, + GAUDI2_QUEUE_ID_NIC_11_3 = 203, + GAUDI2_QUEUE_ID_NIC_12_0 = 204, + GAUDI2_QUEUE_ID_NIC_12_1 = 205, + GAUDI2_QUEUE_ID_NIC_12_2 = 206, + GAUDI2_QUEUE_ID_NIC_12_3 = 207, + GAUDI2_QUEUE_ID_NIC_13_0 = 208, + GAUDI2_QUEUE_ID_NIC_13_1 = 209, + GAUDI2_QUEUE_ID_NIC_13_2 = 210, + GAUDI2_QUEUE_ID_NIC_13_3 = 211, + GAUDI2_QUEUE_ID_NIC_14_0 = 212, + GAUDI2_QUEUE_ID_NIC_14_1 = 213, + GAUDI2_QUEUE_ID_NIC_14_2 = 214, + GAUDI2_QUEUE_ID_NIC_14_3 = 215, + GAUDI2_QUEUE_ID_NIC_15_0 = 216, + GAUDI2_QUEUE_ID_NIC_15_1 = 217, + GAUDI2_QUEUE_ID_NIC_15_2 = 218, + GAUDI2_QUEUE_ID_NIC_15_3 = 219, + GAUDI2_QUEUE_ID_NIC_16_0 = 220, + GAUDI2_QUEUE_ID_NIC_16_1 = 221, + GAUDI2_QUEUE_ID_NIC_16_2 = 222, + GAUDI2_QUEUE_ID_NIC_16_3 = 223, + GAUDI2_QUEUE_ID_NIC_17_0 = 224, + GAUDI2_QUEUE_ID_NIC_17_1 = 225, + GAUDI2_QUEUE_ID_NIC_17_2 = 226, + GAUDI2_QUEUE_ID_NIC_17_3 = 227, + GAUDI2_QUEUE_ID_NIC_18_0 = 228, + GAUDI2_QUEUE_ID_NIC_18_1 = 229, + GAUDI2_QUEUE_ID_NIC_18_2 = 230, + GAUDI2_QUEUE_ID_NIC_18_3 = 231, + GAUDI2_QUEUE_ID_NIC_19_0 = 232, + GAUDI2_QUEUE_ID_NIC_19_1 = 233, + GAUDI2_QUEUE_ID_NIC_19_2 = 234, + GAUDI2_QUEUE_ID_NIC_19_3 = 235, + GAUDI2_QUEUE_ID_NIC_20_0 = 236, + GAUDI2_QUEUE_ID_NIC_20_1 = 237, + GAUDI2_QUEUE_ID_NIC_20_2 = 238, + GAUDI2_QUEUE_ID_NIC_20_3 = 239, + GAUDI2_QUEUE_ID_NIC_21_0 = 240, + GAUDI2_QUEUE_ID_NIC_21_1 = 241, + GAUDI2_QUEUE_ID_NIC_21_2 = 242, + GAUDI2_QUEUE_ID_NIC_21_3 = 243, + GAUDI2_QUEUE_ID_NIC_22_0 = 244, + GAUDI2_QUEUE_ID_NIC_22_1 = 245, + GAUDI2_QUEUE_ID_NIC_22_2 = 246, + GAUDI2_QUEUE_ID_NIC_22_3 = 247, + GAUDI2_QUEUE_ID_NIC_23_0 = 248, + GAUDI2_QUEUE_ID_NIC_23_1 = 249, + GAUDI2_QUEUE_ID_NIC_23_2 = 250, + GAUDI2_QUEUE_ID_NIC_23_3 = 251, + GAUDI2_QUEUE_ID_ROT_0_0 = 252, + GAUDI2_QUEUE_ID_ROT_0_1 = 253, + GAUDI2_QUEUE_ID_ROT_0_2 = 254, + GAUDI2_QUEUE_ID_ROT_0_3 = 255, + GAUDI2_QUEUE_ID_ROT_1_0 = 256, + GAUDI2_QUEUE_ID_ROT_1_1 = 257, + GAUDI2_QUEUE_ID_ROT_1_2 = 258, + GAUDI2_QUEUE_ID_ROT_1_3 = 259, + GAUDI2_QUEUE_ID_CPU_PQ = 260, + GAUDI2_QUEUE_ID_SIZE +}; + +/* * Engine Numbering * * Used in the "busy_engines_mask" field in `struct hl_info_hw_idle' @@ -239,6 +521,85 @@ enum gaudi_engine_id { GAUDI_ENGINE_ID_SIZE }; +enum gaudi2_engine_id { + GAUDI2_DCORE0_ENGINE_ID_EDMA_0 = 0, + GAUDI2_DCORE0_ENGINE_ID_EDMA_1, + GAUDI2_DCORE0_ENGINE_ID_MME, + GAUDI2_DCORE0_ENGINE_ID_TPC_0, + GAUDI2_DCORE0_ENGINE_ID_TPC_1, + GAUDI2_DCORE0_ENGINE_ID_TPC_2, + GAUDI2_DCORE0_ENGINE_ID_TPC_3, + GAUDI2_DCORE0_ENGINE_ID_TPC_4, + GAUDI2_DCORE0_ENGINE_ID_TPC_5, + GAUDI2_DCORE0_ENGINE_ID_DEC_0, + GAUDI2_DCORE0_ENGINE_ID_DEC_1, + GAUDI2_DCORE1_ENGINE_ID_EDMA_0, + GAUDI2_DCORE1_ENGINE_ID_EDMA_1, + GAUDI2_DCORE1_ENGINE_ID_MME, + GAUDI2_DCORE1_ENGINE_ID_TPC_0, + GAUDI2_DCORE1_ENGINE_ID_TPC_1, + GAUDI2_DCORE1_ENGINE_ID_TPC_2, + GAUDI2_DCORE1_ENGINE_ID_TPC_3, + GAUDI2_DCORE1_ENGINE_ID_TPC_4, + GAUDI2_DCORE1_ENGINE_ID_TPC_5, + GAUDI2_DCORE1_ENGINE_ID_DEC_0, + GAUDI2_DCORE1_ENGINE_ID_DEC_1, + GAUDI2_DCORE2_ENGINE_ID_EDMA_0, + GAUDI2_DCORE2_ENGINE_ID_EDMA_1, + GAUDI2_DCORE2_ENGINE_ID_MME, + GAUDI2_DCORE2_ENGINE_ID_TPC_0, + GAUDI2_DCORE2_ENGINE_ID_TPC_1, + GAUDI2_DCORE2_ENGINE_ID_TPC_2, + GAUDI2_DCORE2_ENGINE_ID_TPC_3, + GAUDI2_DCORE2_ENGINE_ID_TPC_4, + GAUDI2_DCORE2_ENGINE_ID_TPC_5, + GAUDI2_DCORE2_ENGINE_ID_DEC_0, + GAUDI2_DCORE2_ENGINE_ID_DEC_1, + GAUDI2_DCORE3_ENGINE_ID_EDMA_0, + GAUDI2_DCORE3_ENGINE_ID_EDMA_1, + GAUDI2_DCORE3_ENGINE_ID_MME, + GAUDI2_DCORE3_ENGINE_ID_TPC_0, + GAUDI2_DCORE3_ENGINE_ID_TPC_1, + GAUDI2_DCORE3_ENGINE_ID_TPC_2, + GAUDI2_DCORE3_ENGINE_ID_TPC_3, + GAUDI2_DCORE3_ENGINE_ID_TPC_4, + GAUDI2_DCORE3_ENGINE_ID_TPC_5, + GAUDI2_DCORE3_ENGINE_ID_DEC_0, + GAUDI2_DCORE3_ENGINE_ID_DEC_1, + GAUDI2_DCORE0_ENGINE_ID_TPC_6, + GAUDI2_ENGINE_ID_PDMA_0, + GAUDI2_ENGINE_ID_PDMA_1, + GAUDI2_ENGINE_ID_ROT_0, + GAUDI2_ENGINE_ID_ROT_1, + GAUDI2_PCIE_ENGINE_ID_DEC_0, + GAUDI2_PCIE_ENGINE_ID_DEC_1, + GAUDI2_ENGINE_ID_NIC0_0, + GAUDI2_ENGINE_ID_NIC0_1, + GAUDI2_ENGINE_ID_NIC1_0, + GAUDI2_ENGINE_ID_NIC1_1, + GAUDI2_ENGINE_ID_NIC2_0, + GAUDI2_ENGINE_ID_NIC2_1, + GAUDI2_ENGINE_ID_NIC3_0, + GAUDI2_ENGINE_ID_NIC3_1, + GAUDI2_ENGINE_ID_NIC4_0, + GAUDI2_ENGINE_ID_NIC4_1, + GAUDI2_ENGINE_ID_NIC5_0, + GAUDI2_ENGINE_ID_NIC5_1, + GAUDI2_ENGINE_ID_NIC6_0, + GAUDI2_ENGINE_ID_NIC6_1, + GAUDI2_ENGINE_ID_NIC7_0, + GAUDI2_ENGINE_ID_NIC7_1, + GAUDI2_ENGINE_ID_NIC8_0, + GAUDI2_ENGINE_ID_NIC8_1, + GAUDI2_ENGINE_ID_NIC9_0, + GAUDI2_ENGINE_ID_NIC9_1, + GAUDI2_ENGINE_ID_NIC10_0, + GAUDI2_ENGINE_ID_NIC10_1, + GAUDI2_ENGINE_ID_NIC11_0, + GAUDI2_ENGINE_ID_NIC11_1, + GAUDI2_ENGINE_ID_SIZE +}; + /* * ASIC specific PLL index * @@ -272,6 +633,49 @@ enum hl_gaudi_pll_index { HL_GAUDI_PLL_MAX }; +enum hl_gaudi2_pll_index { + HL_GAUDI2_CPU_PLL = 0, + HL_GAUDI2_PCI_PLL, + HL_GAUDI2_SRAM_PLL, + HL_GAUDI2_HBM_PLL, + HL_GAUDI2_NIC_PLL, + HL_GAUDI2_DMA_PLL, + HL_GAUDI2_MESH_PLL, + HL_GAUDI2_MME_PLL, + HL_GAUDI2_TPC_PLL, + HL_GAUDI2_IF_PLL, + HL_GAUDI2_VID_PLL, + HL_GAUDI2_MSS_PLL, + HL_GAUDI2_PLL_MAX +}; + +/** + * enum hl_goya_dma_direction - Direction of DMA operation inside a LIN_DMA packet that is + * submitted to the GOYA's DMA QMAN. This attribute is not relevant + * to the H/W but the kernel driver use it to parse the packet's + * addresses and patch/validate them. + * @HL_DMA_HOST_TO_DRAM: DMA operation from Host memory to GOYA's DDR. + * @HL_DMA_HOST_TO_SRAM: DMA operation from Host memory to GOYA's SRAM. + * @HL_DMA_DRAM_TO_SRAM: DMA operation from GOYA's DDR to GOYA's SRAM. + * @HL_DMA_SRAM_TO_DRAM: DMA operation from GOYA's SRAM to GOYA's DDR. + * @HL_DMA_SRAM_TO_HOST: DMA operation from GOYA's SRAM to Host memory. + * @HL_DMA_DRAM_TO_HOST: DMA operation from GOYA's DDR to Host memory. + * @HL_DMA_DRAM_TO_DRAM: DMA operation from GOYA's DDR to GOYA's DDR. + * @HL_DMA_SRAM_TO_SRAM: DMA operation from GOYA's SRAM to GOYA's SRAM. + * @HL_DMA_ENUM_MAX: number of values in enum + */ +enum hl_goya_dma_direction { + HL_DMA_HOST_TO_DRAM, + HL_DMA_HOST_TO_SRAM, + HL_DMA_DRAM_TO_SRAM, + HL_DMA_SRAM_TO_DRAM, + HL_DMA_SRAM_TO_HOST, + HL_DMA_DRAM_TO_HOST, + HL_DMA_DRAM_TO_DRAM, + HL_DMA_SRAM_TO_SRAM, + HL_DMA_ENUM_MAX +}; + /** * enum hl_device_status - Device status information. * @HL_DEVICE_STATUS_OPERATIONAL: Device is operational. @@ -280,6 +684,8 @@ enum hl_gaudi_pll_index { * @HL_DEVICE_STATUS_NEEDS_RESET: Device needs reset because auto reset was disabled. * @HL_DEVICE_STATUS_IN_DEVICE_CREATION: Device is operational but its creation is still in * progress. + * @HL_DEVICE_STATUS_IN_RESET_AFTER_DEVICE_RELEASE: Device is currently during reset that was + * triggered because the user released the device * @HL_DEVICE_STATUS_LAST: Last status. */ enum hl_device_status { @@ -288,7 +694,8 @@ enum hl_device_status { HL_DEVICE_STATUS_MALFUNCTION, HL_DEVICE_STATUS_NEEDS_RESET, HL_DEVICE_STATUS_IN_DEVICE_CREATION, - HL_DEVICE_STATUS_LAST = HL_DEVICE_STATUS_IN_DEVICE_CREATION + HL_DEVICE_STATUS_IN_RESET_AFTER_DEVICE_RELEASE, + HL_DEVICE_STATUS_LAST = HL_DEVICE_STATUS_IN_RESET_AFTER_DEVICE_RELEASE }; enum hl_server_type { @@ -296,9 +703,29 @@ enum hl_server_type { HL_SERVER_GAUDI_HLS1 = 1, HL_SERVER_GAUDI_HLS1H = 2, HL_SERVER_GAUDI_TYPE1 = 3, - HL_SERVER_GAUDI_TYPE2 = 4 + HL_SERVER_GAUDI_TYPE2 = 4, + HL_SERVER_GAUDI2_HLS2 = 5 }; +/* + * Notifier event values - for the notification mechanism and the HL_INFO_GET_EVENTS command + * + * HL_NOTIFIER_EVENT_TPC_ASSERT - Indicates TPC assert event + * HL_NOTIFIER_EVENT_UNDEFINED_OPCODE - Indicates undefined operation code + * HL_NOTIFIER_EVENT_DEVICE_RESET - Indicates device requires a reset + * HL_NOTIFIER_EVENT_CS_TIMEOUT - Indicates CS timeout error + * HL_NOTIFIER_EVENT_DEVICE_UNAVAILABLE - Indicates device is unavailable + * HL_NOTIFIER_EVENT_USER_ENGINE_ERR - Indicates device engine in error state + * HL_NOTIFIER_EVENT_GENERAL_HW_ERR - Indicates device HW error + */ +#define HL_NOTIFIER_EVENT_TPC_ASSERT (1ULL << 0) +#define HL_NOTIFIER_EVENT_UNDEFINED_OPCODE (1ULL << 1) +#define HL_NOTIFIER_EVENT_DEVICE_RESET (1ULL << 2) +#define HL_NOTIFIER_EVENT_CS_TIMEOUT (1ULL << 3) +#define HL_NOTIFIER_EVENT_DEVICE_UNAVAILABLE (1ULL << 4) +#define HL_NOTIFIER_EVENT_USER_ENGINE_ERR (1ULL << 5) +#define HL_NOTIFIER_EVENT_GENERAL_HW_ERR (1ULL << 6) + /* Opcode for management ioctl * * HW_IP_INFO - Receive information about different IP blocks in the @@ -333,29 +760,61 @@ enum hl_server_type { * HL_INFO_SYNC_MANAGER - Retrieve sync manager info per dcore * HL_INFO_TOTAL_ENERGY - Retrieve total energy consumption * HL_INFO_PLL_FREQUENCY - Retrieve PLL frequency + * HL_INFO_POWER - Retrieve power information * HL_INFO_OPEN_STATS - Retrieve info regarding recent device open calls + * HL_INFO_DRAM_REPLACED_ROWS - Retrieve DRAM replaced rows info + * HL_INFO_DRAM_PENDING_ROWS - Retrieve DRAM pending rows num + * HL_INFO_LAST_ERR_OPEN_DEV_TIME - Retrieve timestamp of the last time the device was opened + * and CS timeout or razwi error occurred. + * HL_INFO_CS_TIMEOUT_EVENT - Retrieve CS timeout timestamp and its related CS sequence number. + * HL_INFO_RAZWI_EVENT - Retrieve parameters of razwi: + * Timestamp of razwi. + * The address which accessing it caused the razwi. + * Razwi initiator. + * Razwi cause, was it a page fault or MMU access error. + * HL_INFO_DEV_MEM_ALLOC_PAGE_SIZES - Retrieve valid page sizes for device memory allocation + * HL_INFO_SECURED_ATTESTATION - Retrieve attestation report of the boot. + * HL_INFO_REGISTER_EVENTFD - Register eventfd for event notifications. + * HL_INFO_UNREGISTER_EVENTFD - Unregister eventfd + * HL_INFO_GET_EVENTS - Retrieve the last occurred events + * HL_INFO_UNDEFINED_OPCODE_EVENT - Retrieve last undefined opcode error information. */ -#define HL_INFO_HW_IP_INFO 0 -#define HL_INFO_HW_EVENTS 1 -#define HL_INFO_DRAM_USAGE 2 -#define HL_INFO_HW_IDLE 3 -#define HL_INFO_DEVICE_STATUS 4 -#define HL_INFO_DEVICE_UTILIZATION 6 -#define HL_INFO_HW_EVENTS_AGGREGATE 7 -#define HL_INFO_CLK_RATE 8 -#define HL_INFO_RESET_COUNT 9 -#define HL_INFO_TIME_SYNC 10 -#define HL_INFO_CS_COUNTERS 11 -#define HL_INFO_PCI_COUNTERS 12 -#define HL_INFO_CLK_THROTTLE_REASON 13 -#define HL_INFO_SYNC_MANAGER 14 -#define HL_INFO_TOTAL_ENERGY 15 -#define HL_INFO_PLL_FREQUENCY 16 -#define HL_INFO_POWER 17 -#define HL_INFO_OPEN_STATS 18 - -#define HL_INFO_VERSION_MAX_LEN 128 -#define HL_INFO_CARD_NAME_MAX_LEN 16 +#define HL_INFO_HW_IP_INFO 0 +#define HL_INFO_HW_EVENTS 1 +#define HL_INFO_DRAM_USAGE 2 +#define HL_INFO_HW_IDLE 3 +#define HL_INFO_DEVICE_STATUS 4 +#define HL_INFO_DEVICE_UTILIZATION 6 +#define HL_INFO_HW_EVENTS_AGGREGATE 7 +#define HL_INFO_CLK_RATE 8 +#define HL_INFO_RESET_COUNT 9 +#define HL_INFO_TIME_SYNC 10 +#define HL_INFO_CS_COUNTERS 11 +#define HL_INFO_PCI_COUNTERS 12 +#define HL_INFO_CLK_THROTTLE_REASON 13 +#define HL_INFO_SYNC_MANAGER 14 +#define HL_INFO_TOTAL_ENERGY 15 +#define HL_INFO_PLL_FREQUENCY 16 +#define HL_INFO_POWER 17 +#define HL_INFO_OPEN_STATS 18 +#define HL_INFO_DRAM_REPLACED_ROWS 21 +#define HL_INFO_DRAM_PENDING_ROWS 22 +#define HL_INFO_LAST_ERR_OPEN_DEV_TIME 23 +#define HL_INFO_CS_TIMEOUT_EVENT 24 +#define HL_INFO_RAZWI_EVENT 25 +#define HL_INFO_DEV_MEM_ALLOC_PAGE_SIZES 26 +#define HL_INFO_SECURED_ATTESTATION 27 +#define HL_INFO_REGISTER_EVENTFD 28 +#define HL_INFO_UNREGISTER_EVENTFD 29 +#define HL_INFO_GET_EVENTS 30 +#define HL_INFO_UNDEFINED_OPCODE_EVENT 31 +#define HL_INFO_ENGINE_STATUS 32 + +#define HL_INFO_VERSION_MAX_LEN 128 +#define HL_INFO_CARD_NAME_MAX_LEN 16 + +/* Maximum buffer size for retrieving engines status */ +#define HL_ENGINES_DATA_MAX_SIZE SZ_1M /** * struct hl_info_hw_ip_info - hardware information on various IPs in the ASIC @@ -372,8 +831,10 @@ enum hl_server_type { * @device_id: PCI device ID of the ASIC. * @module_id: Module ID of the ASIC for mezzanine cards in servers * (From OCP spec). + * @decoder_enabled_mask: Bit-mask that represents which decoders are enabled. * @first_available_interrupt_id: The first available interrupt ID for the user * to be used when it works with user interrupts. + * Relevant for Gaudi2 and later. * @server_type: Server type that the Gaudi ASIC is currently installed in. * The value is according to enum hl_server_type * @cpld_version: CPLD version on the board. @@ -385,9 +846,19 @@ enum hl_server_type { * @tpc_enabled_mask: Bit-mask that represents which TPCs are enabled. Relevant * for Goya/Gaudi only. * @dram_enabled: Whether the DRAM is enabled. + * @security_enabled: Whether security is enabled on device. + * @mme_master_slave_mode: Indicate whether the MME is working in master/slave + * configuration. Relevant for Greco and later. * @cpucp_version: The CPUCP f/w version. * @card_name: The card name as passed by the f/w. + * @tpc_enabled_mask_ext: Bit-mask that represents which TPCs are enabled. + * Relevant for Greco and later. * @dram_page_size: The DRAM physical page size. + * @edma_enabled_mask: Bit-mask that represents which EDMAs are enabled. + * Relevant for Gaudi2 and later. + * @number_of_user_interrupts: The number of interrupts that are available to the userspace + * application to use. Relevant for Gaudi2 and later. + * @device_mem_alloc_default_page_size: default page size used in device memory allocation. */ struct hl_info_hw_ip_info { __u64 sram_base_address; @@ -397,7 +868,7 @@ struct hl_info_hw_ip_info { __u32 num_of_events; __u32 device_id; __u32 module_id; - __u32 reserved; + __u32 decoder_enabled_mask; __u16 first_available_interrupt_id; __u16 server_type; __u32 cpld_version; @@ -407,11 +878,17 @@ struct hl_info_hw_ip_info { __u32 psoc_pci_pll_div_factor; __u8 tpc_enabled_mask; __u8 dram_enabled; - __u8 pad[2]; + __u8 security_enabled; + __u8 mme_master_slave_mode; __u8 cpucp_version[HL_INFO_VERSION_MAX_LEN]; __u8 card_name[HL_INFO_CARD_NAME_MAX_LEN]; - __u64 reserved2; + __u64 tpc_enabled_mask_ext; __u64 dram_page_size; + __u32 edma_enabled_mask; + __u16 number_of_user_interrupts; + __u16 pad2; + __u64 reserved4; + __u64 device_mem_alloc_default_page_size; }; struct hl_info_dram_usage { @@ -425,13 +902,13 @@ struct hl_info_hw_idle { __u32 is_idle; /* * Bitmask of busy engines. - * Bits definition is according to `enum <chip>_enging_id'. + * Bits definition is according to `enum <chip>_engine_id'. */ __u32 busy_engines_mask; /* * Extended Bitmask of busy engines. - * Bits definition is according to `enum <chip>_enging_id'. + * Bits definition is according to `enum <chip>_engine_id'. */ __u64 busy_engines_mask_ext[HL_BUSY_ENGINES_MASK_EXT_SIZE]; }; @@ -473,15 +950,27 @@ struct hl_info_pci_counters { __u64 replay_cnt; }; -#define HL_CLK_THROTTLE_POWER 0x1 -#define HL_CLK_THROTTLE_THERMAL 0x2 +enum hl_clk_throttling_type { + HL_CLK_THROTTLE_TYPE_POWER, + HL_CLK_THROTTLE_TYPE_THERMAL, + HL_CLK_THROTTLE_TYPE_MAX +}; + +/* clk_throttling_reason masks */ +#define HL_CLK_THROTTLE_POWER (1 << HL_CLK_THROTTLE_TYPE_POWER) +#define HL_CLK_THROTTLE_THERMAL (1 << HL_CLK_THROTTLE_TYPE_THERMAL) /** * struct hl_info_clk_throttle - clock throttling reason * @clk_throttling_reason: each bit represents a clk throttling reason + * @clk_throttling_timestamp_us: represents CPU timestamp in microseconds of the start-event + * @clk_throttling_duration_ns: the clock throttle time in nanosec */ struct hl_info_clk_throttle { __u32 clk_throttling_reason; + __u32 pad; + __u64 clk_throttling_timestamp_us[HL_CLK_THROTTLE_TYPE_MAX]; + __u64 clk_throttling_duration_ns[HL_CLK_THROTTLE_TYPE_MAX]; }; /** @@ -502,10 +991,15 @@ struct hl_pll_frequency_info { * struct hl_open_stats_info - device open statistics information * @open_counter: ever growing counter, increased on each successful dev open * @last_open_period_ms: duration (ms) device was open last time + * @is_compute_ctx_active: Whether there is an active compute context executing + * @compute_ctx_in_release: true if the current compute context is being released */ struct hl_open_stats_info { __u64 open_counter; __u64 last_open_period_ms; + __u8 is_compute_ctx_active; + __u8 compute_ctx_in_release; + __u8 pad[6]; }; /** @@ -559,6 +1053,127 @@ struct hl_info_cs_counters { __u64 ctx_validation_drop_cnt; }; +/** + * struct hl_info_last_err_open_dev_time - last error boot information. + * @timestamp: timestamp of last time the device was opened and error occurred. + */ +struct hl_info_last_err_open_dev_time { + __s64 timestamp; +}; + +/** + * struct hl_info_cs_timeout_event - last CS timeout information. + * @timestamp: timestamp when last CS timeout event occurred. + * @seq: sequence number of last CS timeout event. + */ +struct hl_info_cs_timeout_event { + __s64 timestamp; + __u64 seq; +}; + +#define HL_RAZWI_PAGE_FAULT 0 +#define HL_RAZWI_MMU_ACCESS_ERROR 1 + +/** + * struct hl_info_razwi_event - razwi information. + * @timestamp: timestamp of razwi. + * @addr: address which accessing it caused razwi. + * @engine_id_1: engine id of the razwi initiator, if it was initiated by engine that does not + * have engine id it will be set to U16_MAX. + * @engine_id_2: second engine id of razwi initiator. Might happen that razwi have 2 possible + * engines which one them caused the razwi. In that case, it will contain the + * second possible engine id, otherwise it will be set to U16_MAX. + * @no_engine_id: if razwi initiator does not have engine id, this field will be set to 1, + * otherwise 0. + * @error_type: cause of razwi, page fault or access error, otherwise it will be set to U8_MAX. + * @pad: padding to 64 bit. + */ +struct hl_info_razwi_event { + __s64 timestamp; + __u64 addr; + __u16 engine_id_1; + __u16 engine_id_2; + __u8 no_engine_id; + __u8 error_type; + __u8 pad[2]; +}; + +#define MAX_QMAN_STREAMS_INFO 4 +#define OPCODE_INFO_MAX_ADDR_SIZE 8 +/** + * struct hl_info_undefined_opcode_event - info about last undefined opcode error + * @timestamp: timestamp of the undefined opcode error + * @cb_addr_streams: CB addresses (per stream) that are currently exists in the PQ + * entries. In case all streams array entries are + * filled with values, it means the execution was in Lower-CP. + * @cq_addr: the address of the current handled command buffer + * @cq_size: the size of the current handled command buffer + * @cb_addr_streams_len: num of streams - actual len of cb_addr_streams array. + * should be equal to 1 in case of undefined opcode + * in Upper-CP (specific stream) and equal to 4 incase + * of undefined opcode in Lower-CP. + * @engine_id: engine-id that the error occurred on + * @stream_id: the stream id the error occurred on. In case the stream equals to + * MAX_QMAN_STREAMS_INFO it means the error occurred on a Lower-CP. + */ +struct hl_info_undefined_opcode_event { + __s64 timestamp; + __u64 cb_addr_streams[MAX_QMAN_STREAMS_INFO][OPCODE_INFO_MAX_ADDR_SIZE]; + __u64 cq_addr; + __u32 cq_size; + __u32 cb_addr_streams_len; + __u32 engine_id; + __u32 stream_id; +}; + +/** + * struct hl_info_dev_memalloc_page_sizes - valid page sizes in device mem alloc information. + * @page_order_bitmask: bitmap in which a set bit represents the order of the supported page size + * (e.g. 0x2100000 means that 1MB and 32MB pages are supported). + */ +struct hl_info_dev_memalloc_page_sizes { + __u64 page_order_bitmask; +}; + +#define SEC_PCR_DATA_BUF_SZ 256 +#define SEC_PCR_QUOTE_BUF_SZ 510 /* (512 - 2) 2 bytes used for size */ +#define SEC_SIGNATURE_BUF_SZ 255 /* (256 - 1) 1 byte used for size */ +#define SEC_PUB_DATA_BUF_SZ 510 /* (512 - 2) 2 bytes used for size */ +#define SEC_CERTIFICATE_BUF_SZ 2046 /* (2048 - 2) 2 bytes used for size */ + +/* + * struct hl_info_sec_attest - attestation report of the boot + * @nonce: number only used once. random number provided by host. this also passed to the quote + * command as a qualifying data. + * @pcr_quote_len: length of the attestation quote data (bytes) + * @pub_data_len: length of the public data (bytes) + * @certificate_len: length of the certificate (bytes) + * @pcr_num_reg: number of PCR registers in the pcr_data array + * @pcr_reg_len: length of each PCR register in the pcr_data array (bytes) + * @quote_sig_len: length of the attestation report signature (bytes) + * @pcr_data: raw values of the PCR registers + * @pcr_quote: attestation report data structure + * @quote_sig: signature structure of the attestation report + * @public_data: public key for the signed attestation + * (outPublic + name + qualifiedName) + * @certificate: certificate for the attestation signing key + */ +struct hl_info_sec_attest { + __u32 nonce; + __u16 pcr_quote_len; + __u16 pub_data_len; + __u16 certificate_len; + __u8 pcr_num_reg; + __u8 pcr_reg_len; + __u8 quote_sig_len; + __u8 pcr_data[SEC_PCR_DATA_BUF_SZ]; + __u8 pcr_quote[SEC_PCR_QUOTE_BUF_SZ]; + __u8 quote_sig[SEC_SIGNATURE_BUF_SZ]; + __u8 public_data[SEC_PUB_DATA_BUF_SZ]; + __u8 certificate[SEC_CERTIFICATE_BUF_SZ]; + __u8 pad0[2]; +}; + enum gaudi_dcores { HL_GAUDI_WS_DCORE, HL_GAUDI_WN_DCORE, @@ -579,6 +1194,12 @@ enum gaudi_dcores { * @period_ms: Period value, in milliseconds, for utilization rate in range 100ms - 1000ms in 100 ms * resolution. Currently not in use. * @pll_index: Index as defined in hl_<asic type>_pll_index enumeration. + * @eventfd: event file descriptor for event notifications. + * @user_buffer_actual_size: Actual data size which was copied to user allocated buffer by the + * driver. It is possible for the user to allocate buffer larger than + * needed, hence updating this variable so user will know the exact amount + * of bytes copied by the kernel to the buffer. + * @sec_attest_nonce: Nonce number used for attestation report. * @pad: Padding to 64 bit. */ struct hl_info_args { @@ -591,6 +1212,9 @@ struct hl_info_args { __u32 ctx_id; __u32 period_ms; __u32 pll_index; + __u32 eventfd; + __u32 user_buffer_actual_size; + __u32 sec_attest_nonce; }; __u32 pad; @@ -607,17 +1231,22 @@ struct hl_info_args { #define HL_MAX_CB_SIZE (0x200000 - 32) /* Indicates whether the command buffer should be mapped to the device's MMU */ -#define HL_CB_FLAGS_MAP 0x1 +#define HL_CB_FLAGS_MAP 0x1 + +/* Used with HL_CB_OP_INFO opcode to get the device va address for kernel mapped CB */ +#define HL_CB_FLAGS_GET_DEVICE_VA 0x2 struct hl_cb_in { /* Handle of CB or 0 if we want to create one */ __u64 cb_handle; /* HL_CB_OP_* */ __u32 op; + /* Size of CB. Maximum size is HL_MAX_CB_SIZE. The minimum size that * will be allocated, regardless of this parameter's value, is PAGE_SIZE */ __u32 cb_size; + /* Context ID - Currently not in use */ __u32 ctx_id; /* HL_CB_FLAGS_* */ @@ -629,11 +1258,16 @@ struct hl_cb_out { /* Handle of CB */ __u64 cb_handle; - /* Information about CB */ - struct { - /* Usage count of CB */ - __u32 usage_cnt; - __u32 pad; + union { + /* Information about CB */ + struct { + /* Usage count of CB */ + __u32 usage_cnt; + __u32 pad; + }; + + /* CB mapped address to device MMU */ + __u64 device_va; }; }; }; @@ -646,16 +1280,16 @@ union hl_cb_args { /* HL_CS_CHUNK_FLAGS_ values * * HL_CS_CHUNK_FLAGS_USER_ALLOC_CB: - * Indicates if the CB was allocated and mapped by userspace. - * User allocated CB is a command buffer allocated by the user, via malloc - * (or similar). After allocating the CB, the user invokes “memory ioctl” - * to map the user memory into a device virtual address. The user provides - * this address via the cb_handle field. The interface provides the - * ability to create a large CBs, Which aren’t limited to - * “HL_MAX_CB_SIZE”. Therefore, it increases the PCI-DMA queues - * throughput. This CB allocation method also reduces the use of Linux - * DMA-able memory pool. Which are limited and used by other Linux - * sub-systems. + * Indicates if the CB was allocated and mapped by userspace + * (relevant to greco and above). User allocated CB is a command buffer, + * allocated by the user, via malloc (or similar). After allocating the + * CB, the user invokes - “memory ioctl” to map the user memory into a + * device virtual address. The user provides this address via the + * cb_handle field. The interface provides the ability to create a + * large CBs, Which aren’t limited to “HL_MAX_CB_SIZE”. Therefore, it + * increases the PCI-DMA queues throughput. This CB allocation method + * also reduces the use of Linux DMA-able memory pool. Which are limited + * and used by other Linux sub-systems. */ #define HL_CS_CHUNK_FLAGS_USER_ALLOC_CB 0x1 @@ -665,12 +1299,17 @@ union hl_cb_args { */ struct hl_cs_chunk { union { - /* For external queue, this represents a Handle of CB on the + /* Goya/Gaudi: + * For external queue, this represents a Handle of CB on the * Host. * For internal queue in Goya, this represents an SRAM or * a DRAM address of the internal CB. In Gaudi, this might also * represent a mapped host address of the CB. * + * Greco onwards: + * For H/W queue, this represents either a Handle of CB on the + * Host, or an SRAM, a DRAM, or a mapped host address of the CB. + * * A mapped host address is in the device address space, after * a host address was mapped by the device MMU. */ @@ -735,11 +1374,12 @@ struct hl_cs_chunk { __u32 pad[10]; }; -/* SIGNAL and WAIT/COLLECTIVE_WAIT flags are mutually exclusive */ +/* SIGNAL/WAIT/COLLECTIVE_WAIT flags are mutually exclusive */ #define HL_CS_FLAGS_FORCE_RESTORE 0x1 #define HL_CS_FLAGS_SIGNAL 0x2 #define HL_CS_FLAGS_WAIT 0x4 #define HL_CS_FLAGS_COLLECTIVE_WAIT 0x8 + #define HL_CS_FLAGS_TIMESTAMP 0x20 #define HL_CS_FLAGS_STAGED_SUBMISSION 0x40 #define HL_CS_FLAGS_STAGED_SUBMISSION_FIRST 0x80 @@ -769,17 +1409,47 @@ struct hl_cs_chunk { #define HL_CS_FLAGS_RESERVE_SIGNALS_ONLY 0x1000 #define HL_CS_FLAGS_UNRESERVE_SIGNALS_ONLY 0x2000 +/* + * The engine cores CS is merged into the existing CS ioctls. + * Use it to control the engine cores mode. + */ +#define HL_CS_FLAGS_ENGINE_CORE_COMMAND 0x4000 + #define HL_CS_STATUS_SUCCESS 0 #define HL_MAX_JOBS_PER_CS 512 +/* HL_ENGINE_CORE_ values + * + * HL_ENGINE_CORE_HALT: engine core halt + * HL_ENGINE_CORE_RUN: engine core run + */ +#define HL_ENGINE_CORE_HALT (1 << 0) +#define HL_ENGINE_CORE_RUN (1 << 1) + struct hl_cs_in { - /* this holds address of array of hl_cs_chunk for restore phase */ - __u64 chunks_restore; + union { + struct { + /* this holds address of array of hl_cs_chunk for restore phase */ + __u64 chunks_restore; + + /* holds address of array of hl_cs_chunk for execution phase */ + __u64 chunks_execute; + }; + + /* Valid only when HL_CS_FLAGS_ENGINE_CORE_COMMAND is set */ + struct { + /* this holds address of array of uint32 for engine_cores */ + __u64 engine_cores; + + /* number of engine cores in engine_cores array */ + __u32 num_engine_cores; - /* holds address of array of hl_cs_chunk for execution phase */ - __u64 chunks_execute; + /* the core command to be sent towards engine cores */ + __u32 core_command; + }; + }; union { /* @@ -831,6 +1501,7 @@ struct hl_cs_in { /* Context ID - Currently not in use */ __u32 ctx_id; + __u8 pad[4]; }; struct hl_cs_out { @@ -843,7 +1514,7 @@ struct hl_cs_out { /* Valid only when HL_CS_FLAGS_RESERVE_SIGNALS_ONLY is set */ struct { - /* This is the resereved signal handle id */ + /* This is the reserved signal handle id */ __u32 handle_id; /* This is the signals count */ @@ -856,9 +1527,17 @@ struct hl_cs_out { /* * SOB base address offset - * Valid only when HL_CS_FLAGS_RESERVE_SIGNALS_ONLY is set + * Valid only when HL_CS_FLAGS_RESERVE_SIGNALS_ONLY or HL_CS_FLAGS_SIGNAL is set */ __u32 sob_base_addr_offset; + + /* + * Count of completed signals in SOB before current signal submission. + * Valid only when (HL_CS_FLAGS_ENCAP_SIGNALS & HL_CS_FLAGS_STAGED_SUBMISSION) + * or HL_CS_FLAGS_SIGNAL is set + */ + __u16 sob_count_before_submission; + __u16 pad[3]; }; union hl_cs_args { @@ -866,9 +1545,13 @@ union hl_cs_args { struct hl_cs_out out; }; -#define HL_WAIT_CS_FLAGS_INTERRUPT 0x2 -#define HL_WAIT_CS_FLAGS_INTERRUPT_MASK 0xFFF00000 -#define HL_WAIT_CS_FLAGS_MULTI_CS 0x4 +#define HL_WAIT_CS_FLAGS_INTERRUPT 0x2 +#define HL_WAIT_CS_FLAGS_INTERRUPT_MASK 0xFFF00000 +#define HL_WAIT_CS_FLAGS_ANY_CQ_INTERRUPT 0xFFF00000 +#define HL_WAIT_CS_FLAGS_ANY_DEC_INTERRUPT 0xFFE00000 +#define HL_WAIT_CS_FLAGS_MULTI_CS 0x4 +#define HL_WAIT_CS_FLAGS_INTERRUPT_KERNEL_CQ 0x10 +#define HL_WAIT_CS_FLAGS_REGISTER_INTERRUPT 0x20 #define HL_WAIT_MULTI_CS_LIST_MAX_LEN 32 @@ -888,14 +1571,23 @@ struct hl_wait_cs_in { }; struct { - /* User address for completion comparison. - * upon interrupt, driver will compare the value pointed - * by this address with the supplied target value. - * in order not to perform any comparison, set address - * to all 1s. - * Relevant only when HL_WAIT_CS_FLAGS_INTERRUPT is set - */ - __u64 addr; + union { + /* User address for completion comparison. + * upon interrupt, driver will compare the value pointed + * by this address with the supplied target value. + * in order not to perform any comparison, set address + * to all 1s. + * Relevant only when HL_WAIT_CS_FLAGS_INTERRUPT is set + */ + __u64 addr; + + /* cq_counters_handle to a kernel mapped cb which contains + * cq counters. + * Relevant only when HL_WAIT_CS_FLAGS_INTERRUPT_KERNEL_CQ is set + */ + __u64 cq_counters_handle; + }; + /* Target value for completion comparison */ __u64 target; }; @@ -906,19 +1598,51 @@ struct hl_wait_cs_in { /* HL_WAIT_CS_FLAGS_* * If HL_WAIT_CS_FLAGS_INTERRUPT is set, this field should include - * interrupt id according to HL_WAIT_CS_FLAGS_INTERRUPT_MASK, in order - * not to specify an interrupt id ,set mask to all 1s. + * interrupt id according to HL_WAIT_CS_FLAGS_INTERRUPT_MASK + * + * in order to wait for any CQ interrupt, set interrupt value to + * HL_WAIT_CS_FLAGS_ANY_CQ_INTERRUPT. + * + * in order to wait for any decoder interrupt, set interrupt value to + * HL_WAIT_CS_FLAGS_ANY_DEC_INTERRUPT. */ __u32 flags; - /* Multi CS API info- valid entries in multi-CS array */ - __u8 seq_arr_len; - __u8 pad[3]; + union { + struct { + /* Multi CS API info- valid entries in multi-CS array */ + __u8 seq_arr_len; + __u8 pad[7]; + }; + + /* Absolute timeout to wait for an interrupt in microseconds. + * Relevant only when HL_WAIT_CS_FLAGS_INTERRUPT is set + */ + __u64 interrupt_timeout_us; + }; + + /* + * cq counter offset inside the counters cb pointed by cq_counters_handle above. + * upon interrupt, driver will compare the value pointed + * by this address (cq_counters_handle + cq_counters_offset) + * with the supplied target value. + * relevant only when HL_WAIT_CS_FLAGS_INTERRUPT_KERNEL_CQ is set + */ + __u64 cq_counters_offset; + + /* + * Timestamp_handle timestamps buffer handle. + * relevant only when HL_WAIT_CS_FLAGS_REGISTER_INTERRUPT is set + */ + __u64 timestamp_handle; - /* Absolute timeout to wait for an interrupt in microseconds. - * Relevant only when HL_WAIT_CS_FLAGS_INTERRUPT is set + /* + * Timestamp_offset is offset inside the timestamp buffer pointed by timestamp_handle above. + * upon interrupt, if the cq reached the target value then driver will write + * timestamp to this offset. + * relevant only when HL_WAIT_CS_FLAGS_REGISTER_INTERRUPT is set */ - __u32 interrupt_timeout_us; + __u64 timestamp_offset; }; #define HL_WAIT_CS_STATUS_COMPLETED 0 @@ -952,113 +1676,138 @@ union hl_wait_cs_args { /* Opcode to allocate device memory */ #define HL_MEM_OP_ALLOC 0 + /* Opcode to free previously allocated device memory */ #define HL_MEM_OP_FREE 1 + /* Opcode to map host and device memory */ #define HL_MEM_OP_MAP 2 + /* Opcode to unmap previously mapped host and device memory */ #define HL_MEM_OP_UNMAP 3 + /* Opcode to map a hw block */ #define HL_MEM_OP_MAP_BLOCK 4 + /* Opcode to create DMA-BUF object for an existing device memory allocation * and to export an FD of that DMA-BUF back to the caller */ #define HL_MEM_OP_EXPORT_DMABUF_FD 5 +/* Opcode to create timestamps pool for user interrupts registration support + * The memory will be allocated by the kernel driver, A timestamp buffer which the user + * will get handle to it for mmap, and another internal buffer used by the + * driver for registration management + * The memory will be freed when the user closes the file descriptor(ctx close) + */ +#define HL_MEM_OP_TS_ALLOC 6 + /* Memory flags */ #define HL_MEM_CONTIGUOUS 0x1 #define HL_MEM_SHARED 0x2 #define HL_MEM_USERPTR 0x4 #define HL_MEM_FORCE_HINT 0x8 +#define HL_MEM_PREFETCH 0x40 +/** + * structure hl_mem_in - structure that handle input args for memory IOCTL + * @union arg: union of structures to be used based on the input operation + * @op: specify the requested memory operation (one of the HL_MEM_OP_* definitions). + * @flags: flags for the memory operation (one of the HL_MEM_* definitions). + * For the HL_MEM_OP_EXPORT_DMABUF_FD opcode, this field holds the DMA-BUF file/FD flags. + * @ctx_id: context ID - currently not in use. + * @num_of_elements: number of timestamp elements used only with HL_MEM_OP_TS_ALLOC opcode. + */ struct hl_mem_in { union { - /* HL_MEM_OP_ALLOC- allocate device memory */ + /** + * structure for device memory allocation (used with the HL_MEM_OP_ALLOC op) + * @mem_size: memory size to allocate + * @page_size: page size to use on allocation. when the value is 0 the default page + * size will be taken. + */ struct { - /* Size to alloc */ __u64 mem_size; + __u64 page_size; } alloc; - /* HL_MEM_OP_FREE - free device memory */ + /** + * structure for free-ing device memory (used with the HL_MEM_OP_FREE op) + * @handle: handle returned from HL_MEM_OP_ALLOC + */ struct { - /* Handle returned from HL_MEM_OP_ALLOC */ __u64 handle; } free; - /* HL_MEM_OP_MAP - map device memory */ + /** + * structure for mapping device memory (used with the HL_MEM_OP_MAP op) + * @hint_addr: requested virtual address of mapped memory. + * the driver will try to map the requested region to this hint + * address, as long as the address is valid and not already mapped. + * the user should check the returned address of the IOCTL to make + * sure he got the hint address. + * passing 0 here means that the driver will choose the address itself. + * @handle: handle returned from HL_MEM_OP_ALLOC. + */ struct { - /* - * Requested virtual address of mapped memory. - * The driver will try to map the requested region to - * this hint address, as long as the address is valid - * and not already mapped. The user should check the - * returned address of the IOCTL to make sure he got - * the hint address. Passing 0 here means that the - * driver will choose the address itself. - */ __u64 hint_addr; - /* Handle returned from HL_MEM_OP_ALLOC */ __u64 handle; } map_device; - /* HL_MEM_OP_MAP - map host memory */ + /** + * structure for mapping host memory (used with the HL_MEM_OP_MAP op) + * @host_virt_addr: address of allocated host memory. + * @hint_addr: requested virtual address of mapped memory. + * the driver will try to map the requested region to this hint + * address, as long as the address is valid and not already mapped. + * the user should check the returned address of the IOCTL to make + * sure he got the hint address. + * passing 0 here means that the driver will choose the address itself. + * @size: size of allocated host memory. + */ struct { - /* Address of allocated host memory */ __u64 host_virt_addr; - /* - * Requested virtual address of mapped memory. - * The driver will try to map the requested region to - * this hint address, as long as the address is valid - * and not already mapped. The user should check the - * returned address of the IOCTL to make sure he got - * the hint address. Passing 0 here means that the - * driver will choose the address itself. - */ __u64 hint_addr; - /* Size of allocated host memory */ __u64 mem_size; } map_host; - /* HL_MEM_OP_MAP_BLOCK - map a hw block */ + /** + * structure for mapping hw block (used with the HL_MEM_OP_MAP_BLOCK op) + * @block_addr:HW block address to map, a handle and size will be returned + * to the user and will be used to mmap the relevant block. + * only addresses from configuration space are allowed. + */ struct { - /* - * HW block address to map, a handle and size will be - * returned to the user and will be used to mmap the - * relevant block. Only addresses from configuration - * space are allowed. - */ __u64 block_addr; } map_block; - /* HL_MEM_OP_UNMAP - unmap host memory */ + /** + * structure for unmapping host memory (used with the HL_MEM_OP_UNMAP op) + * @device_virt_addr: virtual address returned from HL_MEM_OP_MAP + */ struct { - /* Virtual address returned from HL_MEM_OP_MAP */ __u64 device_virt_addr; } unmap; - /* HL_MEM_OP_EXPORT_DMABUF_FD */ + /** + * structure for exporting DMABUF object (used with + * the HL_MEM_OP_EXPORT_DMABUF_FD op) + * @handle: handle returned from HL_MEM_OP_ALLOC. + * in Gaudi, where we don't have MMU for the device memory, the + * driver expects a physical address (instead of a handle) in the + * device memory space. + * @mem_size: size of memory allocation. Relevant only for GAUDI + */ struct { - /* Handle returned from HL_MEM_OP_ALLOC. In Gaudi, - * where we don't have MMU for the device memory, the - * driver expects a physical address (instead of - * a handle) in the device memory space. - */ __u64 handle; - /* Size of memory allocation. Relevant only for GAUDI */ __u64 mem_size; } export_dmabuf_fd; }; - /* HL_MEM_OP_* */ __u32 op; - /* HL_MEM_* flags. - * For the HL_MEM_OP_EXPORT_DMABUF_FD opcode, this field holds the - * DMA-BUF file/FD flags. - */ __u32 flags; - /* Context ID - Currently not in use */ __u32 ctx_id; - __u32 pad; + __u32 num_of_elements; }; struct hl_mem_out { @@ -1158,7 +1907,16 @@ struct hl_debug_params_bmon { /* Trace source ID */ __u32 id; - __u32 pad; + + /* Control register */ + __u32 control; + + /* Two more address ranges that the user can request to filter */ + __u64 start_addr2; + __u64 end_addr2; + + __u64 start_addr3; + __u64 end_addr3; }; struct hl_debug_params_spmu { @@ -1167,7 +1925,11 @@ struct hl_debug_params_spmu { /* Number of event types selection */ __u32 event_types_num; - __u32 pad; + + /* TRC configuration register values */ + __u32 pmtrc_val; + __u32 trc_ctrl_host_val; + __u32 trc_en_host_val; }; /* Opcode for ETR component */ @@ -1267,16 +2029,23 @@ struct hl_debug_args { * (or if its the first CS for this context). The user can also order the * driver to run the "restore" phase explicitly * + * Goya/Gaudi: * There are two types of queues - external and internal. External queues * are DMA queues which transfer data from/to the Host. All other queues are * internal. The driver will get completion notifications from the device only * on JOBS which are enqueued in the external queues. * + * Greco onwards: + * There is a single type of queue for all types of engines, either DMA engines + * for transfers from/to the host or inside the device, or compute engines. + * The driver will get completion notifications from the device for all queues. + * * For jobs on external queues, the user needs to create command buffers * through the CB ioctl and give the CB's handle to the CS ioctl. For jobs on * internal queues, the user needs to prepare a "command buffer" with packets * on either the device SRAM/DRAM or the host, and give the device address of * that buffer to the CS ioctl. + * For jobs on H/W queues both options of command buffers are valid. * * This IOCTL is asynchronous in regard to the actual execution of the CS. This * means it returns immediately after ALL the JOBS were enqueued on their @@ -1285,7 +2054,7 @@ struct hl_debug_args { * * Upon successful enqueue, the IOCTL returns a sequence number which the user * can use with the "Wait for CS" IOCTL to check whether the handle's CS - * external JOBS have been completed. Note that if the CS has internal JOBS + * non-internal JOBS have been completed. Note that if the CS has internal JOBS * which can execute AFTER the external JOBS have finished, the driver might * report that the CS has finished executing BEFORE the internal JOBS have * actually finished executing. |