1 /* 2 * Copyright 2018 Advanced Micro Devices, Inc. 3 * 4 * Permission is hereby granted, free of charge, to any person obtaining a 5 * copy of this software and associated documentation files (the "Software"), 6 * to deal in the Software without restriction, including without limitation 7 * the rights to use, copy, modify, merge, publish, distribute, sublicense, 8 * and/or sell copies of the Software, and to permit persons to whom the 9 * Software is furnished to do so, subject to the following conditions: 10 * 11 * The above copyright notice and this permission notice shall be included in 12 * all copies or substantial portions of the Software. 13 * 14 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 15 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 16 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL 17 * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR 18 * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, 19 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR 20 * OTHER DEALINGS IN THE SOFTWARE. 21 * 22 * 23 */ 24 #ifndef _AMDGPU_RAS_H 25 #define _AMDGPU_RAS_H 26 27 #include <linux/debugfs.h> 28 #include <linux/list.h> 29 #include <linux/kfifo.h> 30 #include <linux/radix-tree.h> 31 #include <linux/siphash.h> 32 #include "ta_ras_if.h" 33 #include "amdgpu_ras_eeprom.h" 34 #include "amdgpu_smuio.h" 35 #include "amdgpu_aca.h" 36 37 struct amdgpu_iv_entry; 38 39 #define AMDGPU_RAS_GPU_ERR_MEM_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 0, 0) 40 #define AMDGPU_RAS_GPU_ERR_FW_LOAD(x) AMDGPU_GET_REG_FIELD(x, 1, 1) 41 #define AMDGPU_RAS_GPU_ERR_WAFL_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 2, 2) 42 #define AMDGPU_RAS_GPU_ERR_XGMI_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 3, 3) 43 #define AMDGPU_RAS_GPU_ERR_USR_CP_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 4, 4) 44 #define AMDGPU_RAS_GPU_ERR_USR_DP_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 5, 5) 45 #define AMDGPU_RAS_GPU_ERR_HBM_MEM_TEST(x) AMDGPU_GET_REG_FIELD(x, 6, 6) 46 #define AMDGPU_RAS_GPU_ERR_HBM_BIST_TEST(x) AMDGPU_GET_REG_FIELD(x, 7, 7) 47 #define AMDGPU_RAS_GPU_ERR_SOCKET_ID(x) AMDGPU_GET_REG_FIELD(x, 10, 8) 48 #define AMDGPU_RAS_GPU_ERR_AID_ID(x) AMDGPU_GET_REG_FIELD(x, 12, 11) 49 #define AMDGPU_RAS_GPU_ERR_HBM_ID(x) AMDGPU_GET_REG_FIELD(x, 14, 13) 50 51 #define AMDGPU_RAS_BOOT_STATUS_POLLING_LIMIT 100 52 #define AMDGPU_RAS_BOOT_STEADY_STATUS 0xBA 53 #define AMDGPU_RAS_BOOT_STATUS_MASK 0xFF 54 55 #define AMDGPU_RAS_FLAG_INIT_BY_VBIOS (0x1 << 0) 56 /* position of instance value in sub_block_index of 57 * ta_ras_trigger_error_input, the sub block uses lower 12 bits 58 */ 59 #define AMDGPU_RAS_INST_MASK 0xfffff000 60 #define AMDGPU_RAS_INST_SHIFT 0xc 61 62 #define AMDGPU_RAS_FEATURES_SOCKETID_SHIFT 29 63 #define AMDGPU_RAS_FEATURES_SOCKETID_MASK 0xe0000000 64 65 /* Reserve 8 physical dram row for possible retirement. 66 * In worst cases, it will lose 8 * 2MB memory in vram domain */ 67 #define AMDGPU_RAS_RESERVED_VRAM_SIZE (16ULL << 20) 68 /* The high three bits indicates socketid */ 69 #define AMDGPU_RAS_GET_FEATURES(val) ((val) & ~AMDGPU_RAS_FEATURES_SOCKETID_MASK) 70 71 #define RAS_EVENT_LOG(adev, id, fmt, ...) \ 72 amdgpu_ras_event_log_print((adev), (id), (fmt), ##__VA_ARGS__); 73 74 enum amdgpu_ras_block { 75 AMDGPU_RAS_BLOCK__UMC = 0, 76 AMDGPU_RAS_BLOCK__SDMA, 77 AMDGPU_RAS_BLOCK__GFX, 78 AMDGPU_RAS_BLOCK__MMHUB, 79 AMDGPU_RAS_BLOCK__ATHUB, 80 AMDGPU_RAS_BLOCK__PCIE_BIF, 81 AMDGPU_RAS_BLOCK__HDP, 82 AMDGPU_RAS_BLOCK__XGMI_WAFL, 83 AMDGPU_RAS_BLOCK__DF, 84 AMDGPU_RAS_BLOCK__SMN, 85 AMDGPU_RAS_BLOCK__SEM, 86 AMDGPU_RAS_BLOCK__MP0, 87 AMDGPU_RAS_BLOCK__MP1, 88 AMDGPU_RAS_BLOCK__FUSE, 89 AMDGPU_RAS_BLOCK__MCA, 90 AMDGPU_RAS_BLOCK__VCN, 91 AMDGPU_RAS_BLOCK__JPEG, 92 AMDGPU_RAS_BLOCK__IH, 93 AMDGPU_RAS_BLOCK__MPIO, 94 95 AMDGPU_RAS_BLOCK__LAST 96 }; 97 98 enum amdgpu_ras_mca_block { 99 AMDGPU_RAS_MCA_BLOCK__MP0 = 0, 100 AMDGPU_RAS_MCA_BLOCK__MP1, 101 AMDGPU_RAS_MCA_BLOCK__MPIO, 102 AMDGPU_RAS_MCA_BLOCK__IOHC, 103 104 AMDGPU_RAS_MCA_BLOCK__LAST 105 }; 106 107 #define AMDGPU_RAS_BLOCK_COUNT AMDGPU_RAS_BLOCK__LAST 108 #define AMDGPU_RAS_MCA_BLOCK_COUNT AMDGPU_RAS_MCA_BLOCK__LAST 109 #define AMDGPU_RAS_BLOCK_MASK ((1ULL << AMDGPU_RAS_BLOCK_COUNT) - 1) 110 111 enum amdgpu_ras_gfx_subblock { 112 /* CPC */ 113 AMDGPU_RAS_BLOCK__GFX_CPC_INDEX_START = 0, 114 AMDGPU_RAS_BLOCK__GFX_CPC_SCRATCH = 115 AMDGPU_RAS_BLOCK__GFX_CPC_INDEX_START, 116 AMDGPU_RAS_BLOCK__GFX_CPC_UCODE, 117 AMDGPU_RAS_BLOCK__GFX_DC_STATE_ME1, 118 AMDGPU_RAS_BLOCK__GFX_DC_CSINVOC_ME1, 119 AMDGPU_RAS_BLOCK__GFX_DC_RESTORE_ME1, 120 AMDGPU_RAS_BLOCK__GFX_DC_STATE_ME2, 121 AMDGPU_RAS_BLOCK__GFX_DC_CSINVOC_ME2, 122 AMDGPU_RAS_BLOCK__GFX_DC_RESTORE_ME2, 123 AMDGPU_RAS_BLOCK__GFX_CPC_INDEX_END = 124 AMDGPU_RAS_BLOCK__GFX_DC_RESTORE_ME2, 125 /* CPF */ 126 AMDGPU_RAS_BLOCK__GFX_CPF_INDEX_START, 127 AMDGPU_RAS_BLOCK__GFX_CPF_ROQ_ME2 = 128 AMDGPU_RAS_BLOCK__GFX_CPF_INDEX_START, 129 AMDGPU_RAS_BLOCK__GFX_CPF_ROQ_ME1, 130 AMDGPU_RAS_BLOCK__GFX_CPF_TAG, 131 AMDGPU_RAS_BLOCK__GFX_CPF_INDEX_END = AMDGPU_RAS_BLOCK__GFX_CPF_TAG, 132 /* CPG */ 133 AMDGPU_RAS_BLOCK__GFX_CPG_INDEX_START, 134 AMDGPU_RAS_BLOCK__GFX_CPG_DMA_ROQ = 135 AMDGPU_RAS_BLOCK__GFX_CPG_INDEX_START, 136 AMDGPU_RAS_BLOCK__GFX_CPG_DMA_TAG, 137 AMDGPU_RAS_BLOCK__GFX_CPG_TAG, 138 AMDGPU_RAS_BLOCK__GFX_CPG_INDEX_END = AMDGPU_RAS_BLOCK__GFX_CPG_TAG, 139 /* GDS */ 140 AMDGPU_RAS_BLOCK__GFX_GDS_INDEX_START, 141 AMDGPU_RAS_BLOCK__GFX_GDS_MEM = AMDGPU_RAS_BLOCK__GFX_GDS_INDEX_START, 142 AMDGPU_RAS_BLOCK__GFX_GDS_INPUT_QUEUE, 143 AMDGPU_RAS_BLOCK__GFX_GDS_OA_PHY_CMD_RAM_MEM, 144 AMDGPU_RAS_BLOCK__GFX_GDS_OA_PHY_DATA_RAM_MEM, 145 AMDGPU_RAS_BLOCK__GFX_GDS_OA_PIPE_MEM, 146 AMDGPU_RAS_BLOCK__GFX_GDS_INDEX_END = 147 AMDGPU_RAS_BLOCK__GFX_GDS_OA_PIPE_MEM, 148 /* SPI */ 149 AMDGPU_RAS_BLOCK__GFX_SPI_SR_MEM, 150 /* SQ */ 151 AMDGPU_RAS_BLOCK__GFX_SQ_INDEX_START, 152 AMDGPU_RAS_BLOCK__GFX_SQ_SGPR = AMDGPU_RAS_BLOCK__GFX_SQ_INDEX_START, 153 AMDGPU_RAS_BLOCK__GFX_SQ_LDS_D, 154 AMDGPU_RAS_BLOCK__GFX_SQ_LDS_I, 155 AMDGPU_RAS_BLOCK__GFX_SQ_VGPR, 156 AMDGPU_RAS_BLOCK__GFX_SQ_INDEX_END = AMDGPU_RAS_BLOCK__GFX_SQ_VGPR, 157 /* SQC (3 ranges) */ 158 AMDGPU_RAS_BLOCK__GFX_SQC_INDEX_START, 159 /* SQC range 0 */ 160 AMDGPU_RAS_BLOCK__GFX_SQC_INDEX0_START = 161 AMDGPU_RAS_BLOCK__GFX_SQC_INDEX_START, 162 AMDGPU_RAS_BLOCK__GFX_SQC_INST_UTCL1_LFIFO = 163 AMDGPU_RAS_BLOCK__GFX_SQC_INDEX0_START, 164 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_CU0_WRITE_DATA_BUF, 165 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_CU0_UTCL1_LFIFO, 166 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_CU1_WRITE_DATA_BUF, 167 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_CU1_UTCL1_LFIFO, 168 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_CU2_WRITE_DATA_BUF, 169 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_CU2_UTCL1_LFIFO, 170 AMDGPU_RAS_BLOCK__GFX_SQC_INDEX0_END = 171 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_CU2_UTCL1_LFIFO, 172 /* SQC range 1 */ 173 AMDGPU_RAS_BLOCK__GFX_SQC_INDEX1_START, 174 AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKA_TAG_RAM = 175 AMDGPU_RAS_BLOCK__GFX_SQC_INDEX1_START, 176 AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKA_UTCL1_MISS_FIFO, 177 AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKA_MISS_FIFO, 178 AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKA_BANK_RAM, 179 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKA_TAG_RAM, 180 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKA_HIT_FIFO, 181 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKA_MISS_FIFO, 182 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKA_DIRTY_BIT_RAM, 183 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKA_BANK_RAM, 184 AMDGPU_RAS_BLOCK__GFX_SQC_INDEX1_END = 185 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKA_BANK_RAM, 186 /* SQC range 2 */ 187 AMDGPU_RAS_BLOCK__GFX_SQC_INDEX2_START, 188 AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKB_TAG_RAM = 189 AMDGPU_RAS_BLOCK__GFX_SQC_INDEX2_START, 190 AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKB_UTCL1_MISS_FIFO, 191 AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKB_MISS_FIFO, 192 AMDGPU_RAS_BLOCK__GFX_SQC_INST_BANKB_BANK_RAM, 193 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKB_TAG_RAM, 194 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKB_HIT_FIFO, 195 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKB_MISS_FIFO, 196 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKB_DIRTY_BIT_RAM, 197 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKB_BANK_RAM, 198 AMDGPU_RAS_BLOCK__GFX_SQC_INDEX2_END = 199 AMDGPU_RAS_BLOCK__GFX_SQC_DATA_BANKB_BANK_RAM, 200 AMDGPU_RAS_BLOCK__GFX_SQC_INDEX_END = 201 AMDGPU_RAS_BLOCK__GFX_SQC_INDEX2_END, 202 /* TA */ 203 AMDGPU_RAS_BLOCK__GFX_TA_INDEX_START, 204 AMDGPU_RAS_BLOCK__GFX_TA_FS_DFIFO = 205 AMDGPU_RAS_BLOCK__GFX_TA_INDEX_START, 206 AMDGPU_RAS_BLOCK__GFX_TA_FS_AFIFO, 207 AMDGPU_RAS_BLOCK__GFX_TA_FL_LFIFO, 208 AMDGPU_RAS_BLOCK__GFX_TA_FX_LFIFO, 209 AMDGPU_RAS_BLOCK__GFX_TA_FS_CFIFO, 210 AMDGPU_RAS_BLOCK__GFX_TA_INDEX_END = AMDGPU_RAS_BLOCK__GFX_TA_FS_CFIFO, 211 /* TCA */ 212 AMDGPU_RAS_BLOCK__GFX_TCA_INDEX_START, 213 AMDGPU_RAS_BLOCK__GFX_TCA_HOLE_FIFO = 214 AMDGPU_RAS_BLOCK__GFX_TCA_INDEX_START, 215 AMDGPU_RAS_BLOCK__GFX_TCA_REQ_FIFO, 216 AMDGPU_RAS_BLOCK__GFX_TCA_INDEX_END = 217 AMDGPU_RAS_BLOCK__GFX_TCA_REQ_FIFO, 218 /* TCC (5 sub-ranges) */ 219 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX_START, 220 /* TCC range 0 */ 221 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX0_START = 222 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX_START, 223 AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_DATA = 224 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX0_START, 225 AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_DATA_BANK_0_1, 226 AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_DATA_BANK_1_0, 227 AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_DATA_BANK_1_1, 228 AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_DIRTY_BANK_0, 229 AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_DIRTY_BANK_1, 230 AMDGPU_RAS_BLOCK__GFX_TCC_HIGH_RATE_TAG, 231 AMDGPU_RAS_BLOCK__GFX_TCC_LOW_RATE_TAG, 232 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX0_END = 233 AMDGPU_RAS_BLOCK__GFX_TCC_LOW_RATE_TAG, 234 /* TCC range 1 */ 235 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX1_START, 236 AMDGPU_RAS_BLOCK__GFX_TCC_IN_USE_DEC = 237 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX1_START, 238 AMDGPU_RAS_BLOCK__GFX_TCC_IN_USE_TRANSFER, 239 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX1_END = 240 AMDGPU_RAS_BLOCK__GFX_TCC_IN_USE_TRANSFER, 241 /* TCC range 2 */ 242 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX2_START, 243 AMDGPU_RAS_BLOCK__GFX_TCC_RETURN_DATA = 244 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX2_START, 245 AMDGPU_RAS_BLOCK__GFX_TCC_RETURN_CONTROL, 246 AMDGPU_RAS_BLOCK__GFX_TCC_UC_ATOMIC_FIFO, 247 AMDGPU_RAS_BLOCK__GFX_TCC_WRITE_RETURN, 248 AMDGPU_RAS_BLOCK__GFX_TCC_WRITE_CACHE_READ, 249 AMDGPU_RAS_BLOCK__GFX_TCC_SRC_FIFO, 250 AMDGPU_RAS_BLOCK__GFX_TCC_SRC_FIFO_NEXT_RAM, 251 AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_TAG_PROBE_FIFO, 252 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX2_END = 253 AMDGPU_RAS_BLOCK__GFX_TCC_CACHE_TAG_PROBE_FIFO, 254 /* TCC range 3 */ 255 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX3_START, 256 AMDGPU_RAS_BLOCK__GFX_TCC_LATENCY_FIFO = 257 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX3_START, 258 AMDGPU_RAS_BLOCK__GFX_TCC_LATENCY_FIFO_NEXT_RAM, 259 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX3_END = 260 AMDGPU_RAS_BLOCK__GFX_TCC_LATENCY_FIFO_NEXT_RAM, 261 /* TCC range 4 */ 262 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX4_START, 263 AMDGPU_RAS_BLOCK__GFX_TCC_WRRET_TAG_WRITE_RETURN = 264 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX4_START, 265 AMDGPU_RAS_BLOCK__GFX_TCC_ATOMIC_RETURN_BUFFER, 266 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX4_END = 267 AMDGPU_RAS_BLOCK__GFX_TCC_ATOMIC_RETURN_BUFFER, 268 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX_END = 269 AMDGPU_RAS_BLOCK__GFX_TCC_INDEX4_END, 270 /* TCI */ 271 AMDGPU_RAS_BLOCK__GFX_TCI_WRITE_RAM, 272 /* TCP */ 273 AMDGPU_RAS_BLOCK__GFX_TCP_INDEX_START, 274 AMDGPU_RAS_BLOCK__GFX_TCP_CACHE_RAM = 275 AMDGPU_RAS_BLOCK__GFX_TCP_INDEX_START, 276 AMDGPU_RAS_BLOCK__GFX_TCP_LFIFO_RAM, 277 AMDGPU_RAS_BLOCK__GFX_TCP_CMD_FIFO, 278 AMDGPU_RAS_BLOCK__GFX_TCP_VM_FIFO, 279 AMDGPU_RAS_BLOCK__GFX_TCP_DB_RAM, 280 AMDGPU_RAS_BLOCK__GFX_TCP_UTCL1_LFIFO0, 281 AMDGPU_RAS_BLOCK__GFX_TCP_UTCL1_LFIFO1, 282 AMDGPU_RAS_BLOCK__GFX_TCP_INDEX_END = 283 AMDGPU_RAS_BLOCK__GFX_TCP_UTCL1_LFIFO1, 284 /* TD */ 285 AMDGPU_RAS_BLOCK__GFX_TD_INDEX_START, 286 AMDGPU_RAS_BLOCK__GFX_TD_SS_FIFO_LO = 287 AMDGPU_RAS_BLOCK__GFX_TD_INDEX_START, 288 AMDGPU_RAS_BLOCK__GFX_TD_SS_FIFO_HI, 289 AMDGPU_RAS_BLOCK__GFX_TD_CS_FIFO, 290 AMDGPU_RAS_BLOCK__GFX_TD_INDEX_END = AMDGPU_RAS_BLOCK__GFX_TD_CS_FIFO, 291 /* EA (3 sub-ranges) */ 292 AMDGPU_RAS_BLOCK__GFX_EA_INDEX_START, 293 /* EA range 0 */ 294 AMDGPU_RAS_BLOCK__GFX_EA_INDEX0_START = 295 AMDGPU_RAS_BLOCK__GFX_EA_INDEX_START, 296 AMDGPU_RAS_BLOCK__GFX_EA_DRAMRD_CMDMEM = 297 AMDGPU_RAS_BLOCK__GFX_EA_INDEX0_START, 298 AMDGPU_RAS_BLOCK__GFX_EA_DRAMWR_CMDMEM, 299 AMDGPU_RAS_BLOCK__GFX_EA_DRAMWR_DATAMEM, 300 AMDGPU_RAS_BLOCK__GFX_EA_RRET_TAGMEM, 301 AMDGPU_RAS_BLOCK__GFX_EA_WRET_TAGMEM, 302 AMDGPU_RAS_BLOCK__GFX_EA_GMIRD_CMDMEM, 303 AMDGPU_RAS_BLOCK__GFX_EA_GMIWR_CMDMEM, 304 AMDGPU_RAS_BLOCK__GFX_EA_GMIWR_DATAMEM, 305 AMDGPU_RAS_BLOCK__GFX_EA_INDEX0_END = 306 AMDGPU_RAS_BLOCK__GFX_EA_GMIWR_DATAMEM, 307 /* EA range 1 */ 308 AMDGPU_RAS_BLOCK__GFX_EA_INDEX1_START, 309 AMDGPU_RAS_BLOCK__GFX_EA_DRAMRD_PAGEMEM = 310 AMDGPU_RAS_BLOCK__GFX_EA_INDEX1_START, 311 AMDGPU_RAS_BLOCK__GFX_EA_DRAMWR_PAGEMEM, 312 AMDGPU_RAS_BLOCK__GFX_EA_IORD_CMDMEM, 313 AMDGPU_RAS_BLOCK__GFX_EA_IOWR_CMDMEM, 314 AMDGPU_RAS_BLOCK__GFX_EA_IOWR_DATAMEM, 315 AMDGPU_RAS_BLOCK__GFX_EA_GMIRD_PAGEMEM, 316 AMDGPU_RAS_BLOCK__GFX_EA_GMIWR_PAGEMEM, 317 AMDGPU_RAS_BLOCK__GFX_EA_INDEX1_END = 318 AMDGPU_RAS_BLOCK__GFX_EA_GMIWR_PAGEMEM, 319 /* EA range 2 */ 320 AMDGPU_RAS_BLOCK__GFX_EA_INDEX2_START, 321 AMDGPU_RAS_BLOCK__GFX_EA_MAM_D0MEM = 322 AMDGPU_RAS_BLOCK__GFX_EA_INDEX2_START, 323 AMDGPU_RAS_BLOCK__GFX_EA_MAM_D1MEM, 324 AMDGPU_RAS_BLOCK__GFX_EA_MAM_D2MEM, 325 AMDGPU_RAS_BLOCK__GFX_EA_MAM_D3MEM, 326 AMDGPU_RAS_BLOCK__GFX_EA_INDEX2_END = 327 AMDGPU_RAS_BLOCK__GFX_EA_MAM_D3MEM, 328 AMDGPU_RAS_BLOCK__GFX_EA_INDEX_END = 329 AMDGPU_RAS_BLOCK__GFX_EA_INDEX2_END, 330 /* UTC VM L2 bank */ 331 AMDGPU_RAS_BLOCK__UTC_VML2_BANK_CACHE, 332 /* UTC VM walker */ 333 AMDGPU_RAS_BLOCK__UTC_VML2_WALKER, 334 /* UTC ATC L2 2MB cache */ 335 AMDGPU_RAS_BLOCK__UTC_ATCL2_CACHE_2M_BANK, 336 /* UTC ATC L2 4KB cache */ 337 AMDGPU_RAS_BLOCK__UTC_ATCL2_CACHE_4K_BANK, 338 AMDGPU_RAS_BLOCK__GFX_MAX 339 }; 340 341 enum amdgpu_ras_error_type { 342 AMDGPU_RAS_ERROR__NONE = 0, 343 AMDGPU_RAS_ERROR__PARITY = 1, 344 AMDGPU_RAS_ERROR__SINGLE_CORRECTABLE = 2, 345 AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE = 4, 346 AMDGPU_RAS_ERROR__POISON = 8, 347 }; 348 349 enum amdgpu_ras_ret { 350 AMDGPU_RAS_SUCCESS = 0, 351 AMDGPU_RAS_FAIL, 352 AMDGPU_RAS_UE, 353 AMDGPU_RAS_CE, 354 AMDGPU_RAS_PT, 355 }; 356 357 enum amdgpu_ras_error_query_mode { 358 AMDGPU_RAS_INVALID_ERROR_QUERY = 0, 359 AMDGPU_RAS_DIRECT_ERROR_QUERY = 1, 360 AMDGPU_RAS_FIRMWARE_ERROR_QUERY = 2, 361 }; 362 363 /* ras error status reisger fields */ 364 #define ERR_STATUS_LO__ERR_STATUS_VALID_FLAG__SHIFT 0x0 365 #define ERR_STATUS_LO__ERR_STATUS_VALID_FLAG_MASK 0x00000001L 366 #define ERR_STATUS_LO__MEMORY_ID__SHIFT 0x18 367 #define ERR_STATUS_LO__MEMORY_ID_MASK 0xFF000000L 368 #define ERR_STATUS_HI__ERR_INFO_VALID_FLAG__SHIFT 0x2 369 #define ERR_STATUS_HI__ERR_INFO_VALID_FLAG_MASK 0x00000004L 370 #define ERR_STATUS__ERR_CNT__SHIFT 0x17 371 #define ERR_STATUS__ERR_CNT_MASK 0x03800000L 372 373 #define AMDGPU_RAS_REG_ENTRY(ip, inst, reg_lo, reg_hi) \ 374 ip##_HWIP, inst, reg_lo##_BASE_IDX, reg_lo, reg_hi##_BASE_IDX, reg_hi 375 376 #define AMDGPU_RAS_REG_ENTRY_OFFSET(hwip, ip_inst, segment, reg) \ 377 (adev->reg_offset[hwip][ip_inst][segment] + (reg)) 378 379 #define AMDGPU_RAS_ERR_INFO_VALID (1 << 0) 380 #define AMDGPU_RAS_ERR_STATUS_VALID (1 << 1) 381 #define AMDGPU_RAS_ERR_ADDRESS_VALID (1 << 2) 382 383 #define AMDGPU_RAS_GPU_RESET_MODE2_RESET (0x1 << 0) 384 #define AMDGPU_RAS_GPU_RESET_MODE1_RESET (0x1 << 1) 385 386 struct amdgpu_ras_err_status_reg_entry { 387 uint32_t hwip; 388 uint32_t ip_inst; 389 uint32_t seg_lo; 390 uint32_t reg_lo; 391 uint32_t seg_hi; 392 uint32_t reg_hi; 393 uint32_t reg_inst; 394 uint32_t flags; 395 const char *block_name; 396 }; 397 398 struct amdgpu_ras_memory_id_entry { 399 uint32_t memory_id; 400 const char *name; 401 }; 402 403 struct ras_common_if { 404 enum amdgpu_ras_block block; 405 enum amdgpu_ras_error_type type; 406 uint32_t sub_block_index; 407 char name[32]; 408 }; 409 410 #define MAX_UMC_CHANNEL_NUM 32 411 412 struct ecc_info_per_ch { 413 uint16_t ce_count_lo_chip; 414 uint16_t ce_count_hi_chip; 415 uint64_t mca_umc_status; 416 uint64_t mca_umc_addr; 417 uint64_t mca_ceumc_addr; 418 }; 419 420 struct umc_ecc_info { 421 struct ecc_info_per_ch ecc[MAX_UMC_CHANNEL_NUM]; 422 423 /* Determine smu ecctable whether support 424 * record correctable error address 425 */ 426 int record_ce_addr_supported; 427 }; 428 429 enum ras_event_type { 430 RAS_EVENT_TYPE_INVALID = -1, 431 RAS_EVENT_TYPE_ISR = 0, 432 RAS_EVENT_TYPE_COUNT, 433 }; 434 435 struct ras_event_manager { 436 atomic64_t seqnos[RAS_EVENT_TYPE_COUNT]; 437 }; 438 439 struct ras_query_context { 440 enum ras_event_type type; 441 u64 event_id; 442 }; 443 444 typedef int (*pasid_notify)(struct amdgpu_device *adev, 445 uint16_t pasid, void *data); 446 447 struct ras_poison_msg { 448 enum amdgpu_ras_block block; 449 uint16_t pasid; 450 uint32_t reset; 451 pasid_notify pasid_fn; 452 void *data; 453 }; 454 455 struct ras_err_pages { 456 uint32_t count; 457 uint64_t *pfn; 458 }; 459 460 struct ras_ecc_err { 461 u64 hash_index; 462 uint64_t status; 463 uint64_t ipid; 464 uint64_t addr; 465 struct ras_err_pages err_pages; 466 }; 467 468 struct ras_ecc_log_info { 469 struct mutex lock; 470 siphash_key_t ecc_key; 471 struct radix_tree_root de_page_tree; 472 bool de_updated; 473 }; 474 475 struct amdgpu_ras { 476 /* ras infrastructure */ 477 /* for ras itself. */ 478 uint32_t features; 479 uint32_t schema; 480 struct list_head head; 481 /* sysfs */ 482 struct device_attribute features_attr; 483 struct device_attribute version_attr; 484 struct device_attribute schema_attr; 485 struct bin_attribute badpages_attr; 486 struct dentry *de_ras_eeprom_table; 487 /* block array */ 488 struct ras_manager *objs; 489 490 /* gpu recovery */ 491 struct work_struct recovery_work; 492 atomic_t in_recovery; 493 struct amdgpu_device *adev; 494 /* error handler data */ 495 struct ras_err_handler_data *eh_data; 496 struct mutex recovery_lock; 497 498 uint32_t flags; 499 bool reboot; 500 struct amdgpu_ras_eeprom_control eeprom_control; 501 502 bool error_query_ready; 503 504 /* bad page count threshold */ 505 uint32_t bad_page_cnt_threshold; 506 507 /* disable ras error count harvest in recovery */ 508 bool disable_ras_err_cnt_harvest; 509 510 /* is poison mode supported */ 511 bool poison_supported; 512 513 /* RAS count errors delayed work */ 514 struct delayed_work ras_counte_delay_work; 515 atomic_t ras_ue_count; 516 atomic_t ras_ce_count; 517 518 /* record umc error info queried from smu */ 519 struct umc_ecc_info umc_ecc; 520 521 /* Indicates smu whether need update bad channel info */ 522 bool update_channel_flag; 523 /* Record status of smu mca debug mode */ 524 bool is_aca_debug_mode; 525 bool is_rma; 526 527 /* Record special requirements of gpu reset caller */ 528 uint32_t gpu_reset_flags; 529 530 struct task_struct *page_retirement_thread; 531 wait_queue_head_t page_retirement_wq; 532 struct mutex page_retirement_lock; 533 atomic_t page_retirement_req_cnt; 534 struct mutex page_rsv_lock; 535 DECLARE_KFIFO(poison_fifo, struct ras_poison_msg, 128); 536 struct ras_ecc_log_info umc_ecc_log; 537 struct delayed_work page_retirement_dwork; 538 539 /* Fatal error detected flag */ 540 atomic_t fed; 541 542 /* RAS event manager */ 543 struct ras_event_manager __event_mgr; 544 struct ras_event_manager *event_mgr; 545 546 uint64_t reserved_pages_in_bytes; 547 }; 548 549 struct ras_fs_data { 550 char sysfs_name[48]; 551 char debugfs_name[32]; 552 }; 553 554 struct ras_err_addr { 555 struct list_head node; 556 uint64_t err_status; 557 uint64_t err_ipid; 558 uint64_t err_addr; 559 }; 560 561 struct ras_err_info { 562 struct amdgpu_smuio_mcm_config_info mcm_info; 563 u64 ce_count; 564 u64 ue_count; 565 u64 de_count; 566 struct list_head err_addr_list; 567 }; 568 569 struct ras_err_node { 570 struct list_head node; 571 struct ras_err_info err_info; 572 }; 573 574 struct ras_err_data { 575 unsigned long ue_count; 576 unsigned long ce_count; 577 unsigned long de_count; 578 unsigned long err_addr_cnt; 579 struct eeprom_table_record *err_addr; 580 unsigned long err_addr_len; 581 u32 err_list_count; 582 struct list_head err_node_list; 583 }; 584 585 #define for_each_ras_error(err_node, err_data) \ 586 list_for_each_entry(err_node, &(err_data)->err_node_list, node) 587 588 struct ras_err_handler_data { 589 /* point to bad page records array */ 590 struct eeprom_table_record *bps; 591 /* the count of entries */ 592 int count; 593 /* the space can place new entries */ 594 int space_left; 595 }; 596 597 typedef int (*ras_ih_cb)(struct amdgpu_device *adev, 598 void *err_data, 599 struct amdgpu_iv_entry *entry); 600 601 struct ras_ih_data { 602 /* interrupt bottom half */ 603 struct work_struct ih_work; 604 int inuse; 605 /* IP callback */ 606 ras_ih_cb cb; 607 /* full of entries */ 608 unsigned char *ring; 609 unsigned int ring_size; 610 unsigned int element_size; 611 unsigned int aligned_element_size; 612 unsigned int rptr; 613 unsigned int wptr; 614 }; 615 616 struct ras_manager { 617 struct ras_common_if head; 618 /* reference count */ 619 int use; 620 /* ras block link */ 621 struct list_head node; 622 /* the device */ 623 struct amdgpu_device *adev; 624 /* sysfs */ 625 struct device_attribute sysfs_attr; 626 int attr_inuse; 627 628 /* fs node name */ 629 struct ras_fs_data fs_data; 630 631 /* IH data */ 632 struct ras_ih_data ih_data; 633 634 struct ras_err_data err_data; 635 636 struct aca_handle aca_handle; 637 }; 638 639 struct ras_badpage { 640 unsigned int bp; 641 unsigned int size; 642 unsigned int flags; 643 }; 644 645 /* interfaces for IP */ 646 struct ras_fs_if { 647 struct ras_common_if head; 648 const char* sysfs_name; 649 char debugfs_name[32]; 650 }; 651 652 struct ras_query_if { 653 struct ras_common_if head; 654 unsigned long ue_count; 655 unsigned long ce_count; 656 unsigned long de_count; 657 }; 658 659 struct ras_inject_if { 660 struct ras_common_if head; 661 uint64_t address; 662 uint64_t value; 663 uint32_t instance_mask; 664 }; 665 666 struct ras_cure_if { 667 struct ras_common_if head; 668 uint64_t address; 669 }; 670 671 struct ras_ih_if { 672 struct ras_common_if head; 673 ras_ih_cb cb; 674 }; 675 676 struct ras_dispatch_if { 677 struct ras_common_if head; 678 struct amdgpu_iv_entry *entry; 679 }; 680 681 struct ras_debug_if { 682 union { 683 struct ras_common_if head; 684 struct ras_inject_if inject; 685 }; 686 int op; 687 }; 688 689 struct amdgpu_ras_block_object { 690 struct ras_common_if ras_comm; 691 692 int (*ras_block_match)(struct amdgpu_ras_block_object *block_obj, 693 enum amdgpu_ras_block block, uint32_t sub_block_index); 694 int (*ras_late_init)(struct amdgpu_device *adev, struct ras_common_if *ras_block); 695 void (*ras_fini)(struct amdgpu_device *adev, struct ras_common_if *ras_block); 696 ras_ih_cb ras_cb; 697 const struct amdgpu_ras_block_hw_ops *hw_ops; 698 }; 699 700 struct amdgpu_ras_block_hw_ops { 701 int (*ras_error_inject)(struct amdgpu_device *adev, 702 void *inject_if, uint32_t instance_mask); 703 void (*query_ras_error_count)(struct amdgpu_device *adev, void *ras_error_status); 704 void (*query_ras_error_status)(struct amdgpu_device *adev); 705 void (*query_ras_error_address)(struct amdgpu_device *adev, void *ras_error_status); 706 void (*reset_ras_error_count)(struct amdgpu_device *adev); 707 void (*reset_ras_error_status)(struct amdgpu_device *adev); 708 bool (*query_poison_status)(struct amdgpu_device *adev); 709 bool (*handle_poison_consumption)(struct amdgpu_device *adev); 710 }; 711 712 /* work flow 713 * vbios 714 * 1: ras feature enable (enabled by default) 715 * psp 716 * 2: ras framework init (in ip_init) 717 * IP 718 * 3: IH add 719 * 4: debugfs/sysfs create 720 * 5: query/inject 721 * 6: debugfs/sysfs remove 722 * 7: IH remove 723 * 8: feature disable 724 */ 725 726 727 int amdgpu_ras_recovery_init(struct amdgpu_device *adev); 728 729 void amdgpu_ras_resume(struct amdgpu_device *adev); 730 void amdgpu_ras_suspend(struct amdgpu_device *adev); 731 732 int amdgpu_ras_query_error_count(struct amdgpu_device *adev, 733 unsigned long *ce_count, 734 unsigned long *ue_count, 735 struct ras_query_if *query_info); 736 737 /* error handling functions */ 738 int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev, 739 struct eeprom_table_record *bps, int pages); 740 741 int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev, 742 unsigned long *new_cnt); 743 744 static inline enum ta_ras_block 745 amdgpu_ras_block_to_ta(enum amdgpu_ras_block block) { 746 switch (block) { 747 case AMDGPU_RAS_BLOCK__UMC: 748 return TA_RAS_BLOCK__UMC; 749 case AMDGPU_RAS_BLOCK__SDMA: 750 return TA_RAS_BLOCK__SDMA; 751 case AMDGPU_RAS_BLOCK__GFX: 752 return TA_RAS_BLOCK__GFX; 753 case AMDGPU_RAS_BLOCK__MMHUB: 754 return TA_RAS_BLOCK__MMHUB; 755 case AMDGPU_RAS_BLOCK__ATHUB: 756 return TA_RAS_BLOCK__ATHUB; 757 case AMDGPU_RAS_BLOCK__PCIE_BIF: 758 return TA_RAS_BLOCK__PCIE_BIF; 759 case AMDGPU_RAS_BLOCK__HDP: 760 return TA_RAS_BLOCK__HDP; 761 case AMDGPU_RAS_BLOCK__XGMI_WAFL: 762 return TA_RAS_BLOCK__XGMI_WAFL; 763 case AMDGPU_RAS_BLOCK__DF: 764 return TA_RAS_BLOCK__DF; 765 case AMDGPU_RAS_BLOCK__SMN: 766 return TA_RAS_BLOCK__SMN; 767 case AMDGPU_RAS_BLOCK__SEM: 768 return TA_RAS_BLOCK__SEM; 769 case AMDGPU_RAS_BLOCK__MP0: 770 return TA_RAS_BLOCK__MP0; 771 case AMDGPU_RAS_BLOCK__MP1: 772 return TA_RAS_BLOCK__MP1; 773 case AMDGPU_RAS_BLOCK__FUSE: 774 return TA_RAS_BLOCK__FUSE; 775 case AMDGPU_RAS_BLOCK__MCA: 776 return TA_RAS_BLOCK__MCA; 777 case AMDGPU_RAS_BLOCK__VCN: 778 return TA_RAS_BLOCK__VCN; 779 case AMDGPU_RAS_BLOCK__JPEG: 780 return TA_RAS_BLOCK__JPEG; 781 default: 782 WARN_ONCE(1, "RAS ERROR: unexpected block id %d\n", block); 783 return TA_RAS_BLOCK__UMC; 784 } 785 } 786 787 static inline enum ta_ras_error_type 788 amdgpu_ras_error_to_ta(enum amdgpu_ras_error_type error) { 789 switch (error) { 790 case AMDGPU_RAS_ERROR__NONE: 791 return TA_RAS_ERROR__NONE; 792 case AMDGPU_RAS_ERROR__PARITY: 793 return TA_RAS_ERROR__PARITY; 794 case AMDGPU_RAS_ERROR__SINGLE_CORRECTABLE: 795 return TA_RAS_ERROR__SINGLE_CORRECTABLE; 796 case AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE: 797 return TA_RAS_ERROR__MULTI_UNCORRECTABLE; 798 case AMDGPU_RAS_ERROR__POISON: 799 return TA_RAS_ERROR__POISON; 800 default: 801 WARN_ONCE(1, "RAS ERROR: unexpected error type %d\n", error); 802 return TA_RAS_ERROR__NONE; 803 } 804 } 805 806 /* called in ip_init and ip_fini */ 807 int amdgpu_ras_init(struct amdgpu_device *adev); 808 int amdgpu_ras_late_init(struct amdgpu_device *adev); 809 int amdgpu_ras_fini(struct amdgpu_device *adev); 810 int amdgpu_ras_pre_fini(struct amdgpu_device *adev); 811 812 int amdgpu_ras_block_late_init(struct amdgpu_device *adev, 813 struct ras_common_if *ras_block); 814 815 void amdgpu_ras_block_late_fini(struct amdgpu_device *adev, 816 struct ras_common_if *ras_block); 817 818 int amdgpu_ras_feature_enable(struct amdgpu_device *adev, 819 struct ras_common_if *head, bool enable); 820 821 int amdgpu_ras_feature_enable_on_boot(struct amdgpu_device *adev, 822 struct ras_common_if *head, bool enable); 823 824 int amdgpu_ras_sysfs_create(struct amdgpu_device *adev, 825 struct ras_common_if *head); 826 827 int amdgpu_ras_sysfs_remove(struct amdgpu_device *adev, 828 struct ras_common_if *head); 829 830 void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev); 831 832 int amdgpu_ras_query_error_status(struct amdgpu_device *adev, 833 struct ras_query_if *info); 834 835 int amdgpu_ras_reset_error_count(struct amdgpu_device *adev, 836 enum amdgpu_ras_block block); 837 int amdgpu_ras_reset_error_status(struct amdgpu_device *adev, 838 enum amdgpu_ras_block block); 839 840 int amdgpu_ras_error_inject(struct amdgpu_device *adev, 841 struct ras_inject_if *info); 842 843 int amdgpu_ras_interrupt_add_handler(struct amdgpu_device *adev, 844 struct ras_common_if *head); 845 846 int amdgpu_ras_interrupt_remove_handler(struct amdgpu_device *adev, 847 struct ras_common_if *head); 848 849 int amdgpu_ras_interrupt_dispatch(struct amdgpu_device *adev, 850 struct ras_dispatch_if *info); 851 852 struct ras_manager *amdgpu_ras_find_obj(struct amdgpu_device *adev, 853 struct ras_common_if *head); 854 855 extern atomic_t amdgpu_ras_in_intr; 856 857 static inline bool amdgpu_ras_intr_triggered(void) 858 { 859 return !!atomic_read(&amdgpu_ras_in_intr); 860 } 861 862 static inline void amdgpu_ras_intr_cleared(void) 863 { 864 atomic_set(&amdgpu_ras_in_intr, 0); 865 } 866 867 void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev); 868 869 void amdgpu_ras_set_error_query_ready(struct amdgpu_device *adev, bool ready); 870 871 bool amdgpu_ras_need_emergency_restart(struct amdgpu_device *adev); 872 873 void amdgpu_release_ras_context(struct amdgpu_device *adev); 874 875 int amdgpu_persistent_edc_harvesting_supported(struct amdgpu_device *adev); 876 877 const char *get_ras_block_str(struct ras_common_if *ras_block); 878 879 bool amdgpu_ras_is_poison_mode_supported(struct amdgpu_device *adev); 880 881 int amdgpu_ras_is_supported(struct amdgpu_device *adev, unsigned int block); 882 883 int amdgpu_ras_reset_gpu(struct amdgpu_device *adev); 884 885 struct amdgpu_ras* amdgpu_ras_get_context(struct amdgpu_device *adev); 886 887 int amdgpu_ras_set_context(struct amdgpu_device *adev, struct amdgpu_ras *ras_con); 888 889 int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable); 890 int amdgpu_ras_set_aca_debug_mode(struct amdgpu_device *adev, bool enable); 891 bool amdgpu_ras_get_aca_debug_mode(struct amdgpu_device *adev); 892 bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev, 893 unsigned int *mode); 894 895 int amdgpu_ras_register_ras_block(struct amdgpu_device *adev, 896 struct amdgpu_ras_block_object *ras_block_obj); 897 void amdgpu_ras_interrupt_fatal_error_handler(struct amdgpu_device *adev); 898 void amdgpu_ras_get_error_type_name(uint32_t err_type, char *err_type_name); 899 bool amdgpu_ras_inst_get_memory_id_field(struct amdgpu_device *adev, 900 const struct amdgpu_ras_err_status_reg_entry *reg_entry, 901 uint32_t instance, 902 uint32_t *memory_id); 903 bool amdgpu_ras_inst_get_err_cnt_field(struct amdgpu_device *adev, 904 const struct amdgpu_ras_err_status_reg_entry *reg_entry, 905 uint32_t instance, 906 unsigned long *err_cnt); 907 void amdgpu_ras_inst_query_ras_error_count(struct amdgpu_device *adev, 908 const struct amdgpu_ras_err_status_reg_entry *reg_list, 909 uint32_t reg_list_size, 910 const struct amdgpu_ras_memory_id_entry *mem_list, 911 uint32_t mem_list_size, 912 uint32_t instance, 913 uint32_t err_type, 914 unsigned long *err_count); 915 void amdgpu_ras_inst_reset_ras_error_count(struct amdgpu_device *adev, 916 const struct amdgpu_ras_err_status_reg_entry *reg_list, 917 uint32_t reg_list_size, 918 uint32_t instance); 919 920 int amdgpu_ras_error_data_init(struct ras_err_data *err_data); 921 void amdgpu_ras_error_data_fini(struct ras_err_data *err_data); 922 int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data, 923 struct amdgpu_smuio_mcm_config_info *mcm_info, 924 struct ras_err_addr *err_addr, u64 count); 925 int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data, 926 struct amdgpu_smuio_mcm_config_info *mcm_info, 927 struct ras_err_addr *err_addr, u64 count); 928 int amdgpu_ras_error_statistic_de_count(struct ras_err_data *err_data, 929 struct amdgpu_smuio_mcm_config_info *mcm_info, 930 struct ras_err_addr *err_addr, u64 count); 931 void amdgpu_ras_query_boot_status(struct amdgpu_device *adev, u32 num_instances); 932 int amdgpu_ras_bind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk, 933 const struct aca_info *aca_info, void *data); 934 int amdgpu_ras_unbind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk); 935 936 ssize_t amdgpu_ras_aca_sysfs_read(struct device *dev, struct device_attribute *attr, 937 struct aca_handle *handle, char *buf, void *data); 938 939 void amdgpu_ras_add_mca_err_addr(struct ras_err_info *err_info, 940 struct ras_err_addr *err_addr); 941 942 void amdgpu_ras_del_mca_err_addr(struct ras_err_info *err_info, 943 struct ras_err_addr *mca_err_addr); 944 945 void amdgpu_ras_set_fed(struct amdgpu_device *adev, bool status); 946 bool amdgpu_ras_get_fed_status(struct amdgpu_device *adev); 947 948 bool amdgpu_ras_event_id_is_valid(struct amdgpu_device *adev, u64 id); 949 u64 amdgpu_ras_acquire_event_id(struct amdgpu_device *adev, enum ras_event_type type); 950 951 int amdgpu_ras_reserve_page(struct amdgpu_device *adev, uint64_t pfn); 952 953 int amdgpu_ras_put_poison_req(struct amdgpu_device *adev, 954 enum amdgpu_ras_block block, uint16_t pasid, 955 pasid_notify pasid_fn, void *data, uint32_t reset); 956 957 bool amdgpu_ras_in_recovery(struct amdgpu_device *adev); 958 959 __printf(3, 4) 960 void amdgpu_ras_event_log_print(struct amdgpu_device *adev, u64 event_id, 961 const char *fmt, ...); 962 963 #endif 964