1 /* 2 * Support for generating APEI tables and recording CPER for Guests 3 * 4 * Copyright (c) 2020 HUAWEI TECHNOLOGIES CO., LTD. 5 * 6 * Author: Dongjiu Geng <gengdongjiu@huawei.com> 7 * 8 * This program is free software; you can redistribute it and/or modify 9 * it under the terms of the GNU General Public License as published by 10 * the Free Software Foundation; either version 2 of the License, or 11 * (at your option) any later version. 12 13 * This program is distributed in the hope that it will be useful, 14 * but WITHOUT ANY WARRANTY; without even the implied warranty of 15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 16 * GNU General Public License for more details. 17 18 * You should have received a copy of the GNU General Public License along 19 * with this program; if not, see <http://www.gnu.org/licenses/>. 20 */ 21 22 #include "qemu/osdep.h" 23 #include "qemu/units.h" 24 #include "hw/acpi/ghes.h" 25 #include "hw/acpi/aml-build.h" 26 #include "qemu/error-report.h" 27 #include "hw/acpi/generic_event_device.h" 28 #include "hw/nvram/fw_cfg.h" 29 #include "qemu/uuid.h" 30 31 #define ACPI_GHES_ERRORS_FW_CFG_FILE "etc/hardware_errors" 32 #define ACPI_GHES_DATA_ADDR_FW_CFG_FILE "etc/hardware_errors_addr" 33 34 /* The max size in bytes for one error block */ 35 #define ACPI_GHES_MAX_RAW_DATA_LENGTH (1 * KiB) 36 37 /* Generic Hardware Error Source version 2 */ 38 #define ACPI_GHES_SOURCE_GENERIC_ERROR_V2 10 39 40 /* Address offset in Generic Address Structure(GAS) */ 41 #define GAS_ADDR_OFFSET 4 42 43 /* 44 * The total size of Generic Error Data Entry 45 * ACPI 6.1/6.2: 18.3.2.7.1 Generic Error Data, 46 * Table 18-343 Generic Error Data Entry 47 */ 48 #define ACPI_GHES_DATA_LENGTH 72 49 50 /* The memory section CPER size, UEFI 2.6: N.2.5 Memory Error Section */ 51 #define ACPI_GHES_MEM_CPER_LENGTH 80 52 53 /* Masks for block_status flags */ 54 #define ACPI_GEBS_UNCORRECTABLE 1 55 56 /* 57 * Total size for Generic Error Status Block except Generic Error Data Entries 58 * ACPI 6.2: 18.3.2.7.1 Generic Error Data, 59 * Table 18-380 Generic Error Status Block 60 */ 61 #define ACPI_GHES_GESB_SIZE 20 62 63 /* 64 * Values for error_severity field 65 */ 66 enum AcpiGenericErrorSeverity { 67 ACPI_CPER_SEV_RECOVERABLE = 0, 68 ACPI_CPER_SEV_FATAL = 1, 69 ACPI_CPER_SEV_CORRECTED = 2, 70 ACPI_CPER_SEV_NONE = 3, 71 }; 72 73 /* 74 * Hardware Error Notification 75 * ACPI 4.0: 17.3.2.7 Hardware Error Notification 76 * Composes dummy Hardware Error Notification descriptor of specified type 77 */ 78 static void build_ghes_hw_error_notification(GArray *table, const uint8_t type) 79 { 80 /* Type */ 81 build_append_int_noprefix(table, type, 1); 82 /* 83 * Length: 84 * Total length of the structure in bytes 85 */ 86 build_append_int_noprefix(table, 28, 1); 87 /* Configuration Write Enable */ 88 build_append_int_noprefix(table, 0, 2); 89 /* Poll Interval */ 90 build_append_int_noprefix(table, 0, 4); 91 /* Vector */ 92 build_append_int_noprefix(table, 0, 4); 93 /* Switch To Polling Threshold Value */ 94 build_append_int_noprefix(table, 0, 4); 95 /* Switch To Polling Threshold Window */ 96 build_append_int_noprefix(table, 0, 4); 97 /* Error Threshold Value */ 98 build_append_int_noprefix(table, 0, 4); 99 /* Error Threshold Window */ 100 build_append_int_noprefix(table, 0, 4); 101 } 102 103 /* 104 * Generic Error Data Entry 105 * ACPI 6.1: 18.3.2.7.1 Generic Error Data 106 */ 107 static void acpi_ghes_generic_error_data(GArray *table, 108 const uint8_t *section_type, uint32_t error_severity, 109 uint8_t validation_bits, uint8_t flags, 110 uint32_t error_data_length, QemuUUID fru_id, 111 uint64_t time_stamp) 112 { 113 const uint8_t fru_text[20] = {0}; 114 115 /* Section Type */ 116 g_array_append_vals(table, section_type, 16); 117 118 /* Error Severity */ 119 build_append_int_noprefix(table, error_severity, 4); 120 /* Revision */ 121 build_append_int_noprefix(table, 0x300, 2); 122 /* Validation Bits */ 123 build_append_int_noprefix(table, validation_bits, 1); 124 /* Flags */ 125 build_append_int_noprefix(table, flags, 1); 126 /* Error Data Length */ 127 build_append_int_noprefix(table, error_data_length, 4); 128 129 /* FRU Id */ 130 g_array_append_vals(table, fru_id.data, ARRAY_SIZE(fru_id.data)); 131 132 /* FRU Text */ 133 g_array_append_vals(table, fru_text, sizeof(fru_text)); 134 135 /* Timestamp */ 136 build_append_int_noprefix(table, time_stamp, 8); 137 } 138 139 /* 140 * Generic Error Status Block 141 * ACPI 6.1: 18.3.2.7.1 Generic Error Data 142 */ 143 static void acpi_ghes_generic_error_status(GArray *table, uint32_t block_status, 144 uint32_t raw_data_offset, uint32_t raw_data_length, 145 uint32_t data_length, uint32_t error_severity) 146 { 147 /* Block Status */ 148 build_append_int_noprefix(table, block_status, 4); 149 /* Raw Data Offset */ 150 build_append_int_noprefix(table, raw_data_offset, 4); 151 /* Raw Data Length */ 152 build_append_int_noprefix(table, raw_data_length, 4); 153 /* Data Length */ 154 build_append_int_noprefix(table, data_length, 4); 155 /* Error Severity */ 156 build_append_int_noprefix(table, error_severity, 4); 157 } 158 159 /* UEFI 2.6: N.2.5 Memory Error Section */ 160 static void acpi_ghes_build_append_mem_cper(GArray *table, 161 uint64_t error_physical_addr) 162 { 163 /* 164 * Memory Error Record 165 */ 166 167 /* Validation Bits */ 168 build_append_int_noprefix(table, 169 (1ULL << 14) | /* Type Valid */ 170 (1ULL << 1) /* Physical Address Valid */, 171 8); 172 /* Error Status */ 173 build_append_int_noprefix(table, 0, 8); 174 /* Physical Address */ 175 build_append_int_noprefix(table, error_physical_addr, 8); 176 /* Skip all the detailed information normally found in such a record */ 177 build_append_int_noprefix(table, 0, 48); 178 /* Memory Error Type */ 179 build_append_int_noprefix(table, 0 /* Unknown error */, 1); 180 /* Skip all the detailed information normally found in such a record */ 181 build_append_int_noprefix(table, 0, 7); 182 } 183 184 static int acpi_ghes_record_mem_error(uint64_t error_block_address, 185 uint64_t error_physical_addr) 186 { 187 GArray *block; 188 189 /* Memory Error Section Type */ 190 const uint8_t uefi_cper_mem_sec[] = 191 UUID_LE(0xA5BC1114, 0x6F64, 0x4EDE, 0xB8, 0x63, 0x3E, 0x83, \ 192 0xED, 0x7C, 0x83, 0xB1); 193 194 /* invalid fru id: ACPI 4.0: 17.3.2.6.1 Generic Error Data, 195 * Table 17-13 Generic Error Data Entry 196 */ 197 QemuUUID fru_id = {}; 198 uint32_t data_length; 199 200 block = g_array_new(false, true /* clear */, 1); 201 202 /* This is the length if adding a new generic error data entry*/ 203 data_length = ACPI_GHES_DATA_LENGTH + ACPI_GHES_MEM_CPER_LENGTH; 204 /* 205 * It should not run out of the preallocated memory if adding a new generic 206 * error data entry 207 */ 208 assert((data_length + ACPI_GHES_GESB_SIZE) <= 209 ACPI_GHES_MAX_RAW_DATA_LENGTH); 210 211 /* Build the new generic error status block header */ 212 acpi_ghes_generic_error_status(block, ACPI_GEBS_UNCORRECTABLE, 213 0, 0, data_length, ACPI_CPER_SEV_RECOVERABLE); 214 215 /* Build this new generic error data entry header */ 216 acpi_ghes_generic_error_data(block, uefi_cper_mem_sec, 217 ACPI_CPER_SEV_RECOVERABLE, 0, 0, 218 ACPI_GHES_MEM_CPER_LENGTH, fru_id, 0); 219 220 /* Build the memory section CPER for above new generic error data entry */ 221 acpi_ghes_build_append_mem_cper(block, error_physical_addr); 222 223 /* Write the generic error data entry into guest memory */ 224 cpu_physical_memory_write(error_block_address, block->data, block->len); 225 226 g_array_free(block, true); 227 228 return 0; 229 } 230 231 /* 232 * Build table for the hardware error fw_cfg blob. 233 * Initialize "etc/hardware_errors" and "etc/hardware_errors_addr" fw_cfg blobs. 234 * See docs/specs/acpi_hest_ghes.rst for blobs format. 235 */ 236 static void build_ghes_error_table(GArray *hardware_errors, BIOSLinker *linker) 237 { 238 int i, error_status_block_offset; 239 240 /* Build error_block_address */ 241 for (i = 0; i < ACPI_GHES_ERROR_SOURCE_COUNT; i++) { 242 build_append_int_noprefix(hardware_errors, 0, sizeof(uint64_t)); 243 } 244 245 /* Build read_ack_register */ 246 for (i = 0; i < ACPI_GHES_ERROR_SOURCE_COUNT; i++) { 247 /* 248 * Initialize the value of read_ack_register to 1, so GHES can be 249 * writable after (re)boot. 250 * ACPI 6.2: 18.3.2.8 Generic Hardware Error Source version 2 251 * (GHESv2 - Type 10) 252 */ 253 build_append_int_noprefix(hardware_errors, 1, sizeof(uint64_t)); 254 } 255 256 /* Generic Error Status Block offset in the hardware error fw_cfg blob */ 257 error_status_block_offset = hardware_errors->len; 258 259 /* Reserve space for Error Status Data Block */ 260 acpi_data_push(hardware_errors, 261 ACPI_GHES_MAX_RAW_DATA_LENGTH * ACPI_GHES_ERROR_SOURCE_COUNT); 262 263 /* Tell guest firmware to place hardware_errors blob into RAM */ 264 bios_linker_loader_alloc(linker, ACPI_GHES_ERRORS_FW_CFG_FILE, 265 hardware_errors, sizeof(uint64_t), false); 266 267 for (i = 0; i < ACPI_GHES_ERROR_SOURCE_COUNT; i++) { 268 /* 269 * Tell firmware to patch error_block_address entries to point to 270 * corresponding "Generic Error Status Block" 271 */ 272 bios_linker_loader_add_pointer(linker, 273 ACPI_GHES_ERRORS_FW_CFG_FILE, sizeof(uint64_t) * i, 274 sizeof(uint64_t), ACPI_GHES_ERRORS_FW_CFG_FILE, 275 error_status_block_offset + i * ACPI_GHES_MAX_RAW_DATA_LENGTH); 276 } 277 278 /* 279 * tell firmware to write hardware_errors GPA into 280 * hardware_errors_addr fw_cfg, once the former has been initialized. 281 */ 282 bios_linker_loader_write_pointer(linker, ACPI_GHES_DATA_ADDR_FW_CFG_FILE, 283 0, sizeof(uint64_t), ACPI_GHES_ERRORS_FW_CFG_FILE, 0); 284 } 285 286 /* Build Generic Hardware Error Source version 2 (GHESv2) */ 287 static void build_ghes_v2(GArray *table_data, 288 BIOSLinker *linker, 289 enum AcpiGhesNotifyType notify, 290 uint16_t source_id) 291 { 292 uint64_t address_offset; 293 294 /* 295 * Type: 296 * Generic Hardware Error Source version 2(GHESv2 - Type 10) 297 */ 298 build_append_int_noprefix(table_data, ACPI_GHES_SOURCE_GENERIC_ERROR_V2, 2); 299 /* Source Id */ 300 build_append_int_noprefix(table_data, source_id, 2); 301 /* Related Source Id */ 302 build_append_int_noprefix(table_data, 0xffff, 2); 303 /* Flags */ 304 build_append_int_noprefix(table_data, 0, 1); 305 /* Enabled */ 306 build_append_int_noprefix(table_data, 1, 1); 307 308 /* Number of Records To Pre-allocate */ 309 build_append_int_noprefix(table_data, 1, 4); 310 /* Max Sections Per Record */ 311 build_append_int_noprefix(table_data, 1, 4); 312 /* Max Raw Data Length */ 313 build_append_int_noprefix(table_data, ACPI_GHES_MAX_RAW_DATA_LENGTH, 4); 314 315 address_offset = table_data->len; 316 /* Error Status Address */ 317 build_append_gas(table_data, AML_AS_SYSTEM_MEMORY, 0x40, 0, 318 4 /* QWord access */, 0); 319 bios_linker_loader_add_pointer(linker, ACPI_BUILD_TABLE_FILE, 320 address_offset + GAS_ADDR_OFFSET, sizeof(uint64_t), 321 ACPI_GHES_ERRORS_FW_CFG_FILE, source_id * sizeof(uint64_t)); 322 323 /* Notification Structure */ 324 build_ghes_hw_error_notification(table_data, notify); 325 326 /* Error Status Block Length */ 327 build_append_int_noprefix(table_data, ACPI_GHES_MAX_RAW_DATA_LENGTH, 4); 328 329 /* 330 * Read Ack Register 331 * ACPI 6.1: 18.3.2.8 Generic Hardware Error Source 332 * version 2 (GHESv2 - Type 10) 333 */ 334 address_offset = table_data->len; 335 build_append_gas(table_data, AML_AS_SYSTEM_MEMORY, 0x40, 0, 336 4 /* QWord access */, 0); 337 bios_linker_loader_add_pointer(linker, ACPI_BUILD_TABLE_FILE, 338 address_offset + GAS_ADDR_OFFSET, 339 sizeof(uint64_t), ACPI_GHES_ERRORS_FW_CFG_FILE, 340 (ACPI_GHES_ERROR_SOURCE_COUNT + source_id) * sizeof(uint64_t)); 341 342 /* 343 * Read Ack Preserve field 344 * We only provide the first bit in Read Ack Register to OSPM to write 345 * while the other bits are preserved. 346 */ 347 build_append_int_noprefix(table_data, ~0x1ULL, 8); 348 /* Read Ack Write */ 349 build_append_int_noprefix(table_data, 0x1, 8); 350 } 351 352 /* Build Hardware Error Source Table */ 353 void acpi_build_hest(GArray *table_data, GArray *hardware_errors, 354 BIOSLinker *linker, 355 const char *oem_id, const char *oem_table_id) 356 { 357 AcpiTable table = { .sig = "HEST", .rev = 1, 358 .oem_id = oem_id, .oem_table_id = oem_table_id }; 359 360 build_ghes_error_table(hardware_errors, linker); 361 362 acpi_table_begin(&table, table_data); 363 364 /* Error Source Count */ 365 build_append_int_noprefix(table_data, ACPI_GHES_ERROR_SOURCE_COUNT, 4); 366 build_ghes_v2(table_data, linker, 367 ACPI_GHES_NOTIFY_SEA, ACPI_HEST_SRC_ID_SEA); 368 369 acpi_table_end(linker, &table); 370 } 371 372 void acpi_ghes_add_fw_cfg(AcpiGhesState *ags, FWCfgState *s, 373 GArray *hardware_error) 374 { 375 /* Create a read-only fw_cfg file for GHES */ 376 fw_cfg_add_file(s, ACPI_GHES_ERRORS_FW_CFG_FILE, hardware_error->data, 377 hardware_error->len); 378 379 /* Create a read-write fw_cfg file for Address */ 380 fw_cfg_add_file_callback(s, ACPI_GHES_DATA_ADDR_FW_CFG_FILE, NULL, NULL, 381 NULL, &(ags->ghes_addr_le), sizeof(ags->ghes_addr_le), false); 382 383 ags->present = true; 384 } 385 386 int acpi_ghes_record_errors(uint8_t source_id, uint64_t physical_address) 387 { 388 uint64_t error_block_addr, read_ack_register_addr, read_ack_register = 0; 389 uint64_t start_addr; 390 bool ret = -1; 391 AcpiGedState *acpi_ged_state; 392 AcpiGhesState *ags; 393 394 assert(source_id < ACPI_GHES_ERROR_SOURCE_COUNT); 395 396 acpi_ged_state = ACPI_GED(object_resolve_path_type("", TYPE_ACPI_GED, 397 NULL)); 398 g_assert(acpi_ged_state); 399 ags = &acpi_ged_state->ghes_state; 400 401 start_addr = le64_to_cpu(ags->ghes_addr_le); 402 403 if (!physical_address) { 404 return -1; 405 } 406 407 if (source_id < ACPI_GHES_ERROR_SOURCE_COUNT) { 408 start_addr += source_id * sizeof(uint64_t); 409 } 410 411 cpu_physical_memory_read(start_addr, &error_block_addr, 412 sizeof(error_block_addr)); 413 414 error_block_addr = le64_to_cpu(error_block_addr); 415 416 read_ack_register_addr = start_addr + 417 ACPI_GHES_ERROR_SOURCE_COUNT * sizeof(uint64_t); 418 419 cpu_physical_memory_read(read_ack_register_addr, 420 &read_ack_register, sizeof(read_ack_register)); 421 422 /* zero means OSPM does not acknowledge the error */ 423 if (!read_ack_register) { 424 error_report("OSPM does not acknowledge previous error," 425 " so can not record CPER for current error anymore"); 426 } else if (error_block_addr) { 427 read_ack_register = cpu_to_le64(0); 428 /* 429 * Clear the Read Ack Register, OSPM will write it to 1 when 430 * it acknowledges this error. 431 */ 432 cpu_physical_memory_write(read_ack_register_addr, 433 &read_ack_register, sizeof(uint64_t)); 434 435 ret = acpi_ghes_record_mem_error(error_block_addr, 436 physical_address); 437 } else { 438 error_report("can not find Generic Error Status Block"); 439 } 440 441 return ret; 442 } 443 444 bool acpi_ghes_present(void) 445 { 446 AcpiGedState *acpi_ged_state; 447 AcpiGhesState *ags; 448 449 acpi_ged_state = ACPI_GED(object_resolve_path_type("", TYPE_ACPI_GED, 450 NULL)); 451 452 if (!acpi_ged_state) { 453 return false; 454 } 455 ags = &acpi_ged_state->ghes_state; 456 return ags->present; 457 } 458