1 // SPDX-License-Identifier: GPL-2.0
3 * UEFI Common Platform Error Record (CPER) support
5 * Copyright (C) 2010, Intel Corp.
6 * Author: Huang Ying <ying.huang@intel.com>
8 * CPER is the format used to describe platform hardware error by
9 * various tables, such as ERST, BERT and HEST etc.
11 * For more information about CPER, please refer to Appendix N of UEFI
12 * Specification version 2.4.
15 #include <linux/kernel.h>
16 #include <linux/module.h>
17 #include <linux/time.h>
18 #include <linux/cper.h>
19 #include <linux/dmi.h>
20 #include <linux/acpi.h>
21 #include <linux/pci.h>
22 #include <linux/aer.h>
23 #include <linux/printk.h>
24 #include <linux/bcd.h>
25 #include <acpi/ghes.h>
26 #include <ras/ras_event.h>
29 * CPER record ID need to be unique even after reboot, because record
30 * ID is used as index for ERST storage, while CPER records from
31 * multiple boot may co-exist in ERST.
33 u64 cper_next_record_id(void)
35 static atomic64_t seq;
37 if (!atomic64_read(&seq)) {
38 time64_t time = ktime_get_real_seconds();
41 * This code is unlikely to still be needed in year 2106,
42 * but just in case, let's use a few more bits for timestamps
43 * after y2038 to be sure they keep increasing monotonically
44 * for the next few hundred years...
46 if (time < 0x80000000)
47 atomic64_set(&seq, (ktime_get_real_seconds()) << 32);
49 atomic64_set(&seq, 0x8000000000000000ull |
50 ktime_get_real_seconds() << 24);
53 return atomic64_inc_return(&seq);
55 EXPORT_SYMBOL_GPL(cper_next_record_id);
57 static const char * const severity_strs[] = {
64 const char *cper_severity_str(unsigned int severity)
66 return severity < ARRAY_SIZE(severity_strs) ?
67 severity_strs[severity] : "unknown";
69 EXPORT_SYMBOL_GPL(cper_severity_str);
72 * cper_print_bits - print strings for set bits
73 * @pfx: prefix for each line, including log level and prefix string
75 * @strs: string array, indexed by bit position
76 * @strs_size: size of the string array: @strs
78 * For each set bit in @bits, print the corresponding string in @strs.
79 * If the output length is longer than 80, multiple line will be
80 * printed, with @pfx is printed at the beginning of each line.
82 void cper_print_bits(const char *pfx, unsigned int bits,
83 const char * const strs[], unsigned int strs_size)
89 for (i = 0; i < strs_size; i++) {
90 if (!(bits & (1U << i)))
95 if (len && len + strlen(str) + 2 > 80) {
100 len = snprintf(buf, sizeof(buf), "%s%s", pfx, str);
102 len += scnprintf(buf+len, sizeof(buf)-len, ", %s", str);
108 static const char * const proc_type_strs[] = {
114 static const char * const proc_isa_strs[] = {
122 const char * const cper_proc_error_type_strs[] = {
126 "micro-architectural error",
129 static const char * const proc_op_strs[] = {
130 "unknown or generic",
133 "instruction execution",
136 static const char * const proc_flag_strs[] = {
143 static void cper_print_proc_generic(const char *pfx,
144 const struct cper_sec_proc_generic *proc)
146 if (proc->validation_bits & CPER_PROC_VALID_TYPE)
147 printk("%s""processor_type: %d, %s\n", pfx, proc->proc_type,
148 proc->proc_type < ARRAY_SIZE(proc_type_strs) ?
149 proc_type_strs[proc->proc_type] : "unknown");
150 if (proc->validation_bits & CPER_PROC_VALID_ISA)
151 printk("%s""processor_isa: %d, %s\n", pfx, proc->proc_isa,
152 proc->proc_isa < ARRAY_SIZE(proc_isa_strs) ?
153 proc_isa_strs[proc->proc_isa] : "unknown");
154 if (proc->validation_bits & CPER_PROC_VALID_ERROR_TYPE) {
155 printk("%s""error_type: 0x%02x\n", pfx, proc->proc_error_type);
156 cper_print_bits(pfx, proc->proc_error_type,
157 cper_proc_error_type_strs,
158 ARRAY_SIZE(cper_proc_error_type_strs));
160 if (proc->validation_bits & CPER_PROC_VALID_OPERATION)
161 printk("%s""operation: %d, %s\n", pfx, proc->operation,
162 proc->operation < ARRAY_SIZE(proc_op_strs) ?
163 proc_op_strs[proc->operation] : "unknown");
164 if (proc->validation_bits & CPER_PROC_VALID_FLAGS) {
165 printk("%s""flags: 0x%02x\n", pfx, proc->flags);
166 cper_print_bits(pfx, proc->flags, proc_flag_strs,
167 ARRAY_SIZE(proc_flag_strs));
169 if (proc->validation_bits & CPER_PROC_VALID_LEVEL)
170 printk("%s""level: %d\n", pfx, proc->level);
171 if (proc->validation_bits & CPER_PROC_VALID_VERSION)
172 printk("%s""version_info: 0x%016llx\n", pfx, proc->cpu_version);
173 if (proc->validation_bits & CPER_PROC_VALID_ID)
174 printk("%s""processor_id: 0x%016llx\n", pfx, proc->proc_id);
175 if (proc->validation_bits & CPER_PROC_VALID_TARGET_ADDRESS)
176 printk("%s""target_address: 0x%016llx\n",
177 pfx, proc->target_addr);
178 if (proc->validation_bits & CPER_PROC_VALID_REQUESTOR_ID)
179 printk("%s""requestor_id: 0x%016llx\n",
180 pfx, proc->requestor_id);
181 if (proc->validation_bits & CPER_PROC_VALID_RESPONDER_ID)
182 printk("%s""responder_id: 0x%016llx\n",
183 pfx, proc->responder_id);
184 if (proc->validation_bits & CPER_PROC_VALID_IP)
185 printk("%s""IP: 0x%016llx\n", pfx, proc->ip);
188 static const char * const mem_err_type_strs[] = {
193 "single-symbol chipkill ECC",
194 "multi-symbol chipkill ECC",
202 "scrub corrected error",
203 "scrub uncorrected error",
204 "physical memory map-out event",
207 const char *cper_mem_err_type_str(unsigned int etype)
209 return etype < ARRAY_SIZE(mem_err_type_strs) ?
210 mem_err_type_strs[etype] : "unknown";
212 EXPORT_SYMBOL_GPL(cper_mem_err_type_str);
214 static int cper_mem_err_location(struct cper_mem_err_compact *mem, char *msg)
222 len = CPER_REC_LEN - 1;
223 if (mem->validation_bits & CPER_MEM_VALID_NODE)
224 n += scnprintf(msg + n, len - n, "node: %d ", mem->node);
225 if (mem->validation_bits & CPER_MEM_VALID_CARD)
226 n += scnprintf(msg + n, len - n, "card: %d ", mem->card);
227 if (mem->validation_bits & CPER_MEM_VALID_MODULE)
228 n += scnprintf(msg + n, len - n, "module: %d ", mem->module);
229 if (mem->validation_bits & CPER_MEM_VALID_RANK_NUMBER)
230 n += scnprintf(msg + n, len - n, "rank: %d ", mem->rank);
231 if (mem->validation_bits & CPER_MEM_VALID_BANK)
232 n += scnprintf(msg + n, len - n, "bank: %d ", mem->bank);
233 if (mem->validation_bits & CPER_MEM_VALID_BANK_GROUP)
234 n += scnprintf(msg + n, len - n, "bank_group: %d ",
235 mem->bank >> CPER_MEM_BANK_GROUP_SHIFT);
236 if (mem->validation_bits & CPER_MEM_VALID_BANK_ADDRESS)
237 n += scnprintf(msg + n, len - n, "bank_address: %d ",
238 mem->bank & CPER_MEM_BANK_ADDRESS_MASK);
239 if (mem->validation_bits & CPER_MEM_VALID_DEVICE)
240 n += scnprintf(msg + n, len - n, "device: %d ", mem->device);
241 if (mem->validation_bits & (CPER_MEM_VALID_ROW | CPER_MEM_VALID_ROW_EXT)) {
244 row |= cper_get_mem_extension(mem->validation_bits, mem->extended);
245 n += scnprintf(msg + n, len - n, "row: %d ", row);
247 if (mem->validation_bits & CPER_MEM_VALID_COLUMN)
248 n += scnprintf(msg + n, len - n, "column: %d ", mem->column);
249 if (mem->validation_bits & CPER_MEM_VALID_BIT_POSITION)
250 n += scnprintf(msg + n, len - n, "bit_position: %d ",
252 if (mem->validation_bits & CPER_MEM_VALID_REQUESTOR_ID)
253 n += scnprintf(msg + n, len - n, "requestor_id: 0x%016llx ",
255 if (mem->validation_bits & CPER_MEM_VALID_RESPONDER_ID)
256 n += scnprintf(msg + n, len - n, "responder_id: 0x%016llx ",
258 if (mem->validation_bits & CPER_MEM_VALID_TARGET_ID)
259 scnprintf(msg + n, len - n, "target_id: 0x%016llx ",
261 if (mem->validation_bits & CPER_MEM_VALID_CHIP_ID)
262 scnprintf(msg + n, len - n, "chip_id: %d ",
263 mem->extended >> CPER_MEM_CHIP_ID_SHIFT);
269 static int cper_dimm_err_location(struct cper_mem_err_compact *mem, char *msg)
272 const char *bank = NULL, *device = NULL;
274 if (!msg || !(mem->validation_bits & CPER_MEM_VALID_MODULE_HANDLE))
278 dmi_memdev_name(mem->mem_dev_handle, &bank, &device);
280 n = snprintf(msg, len, "DIMM location: %s %s ", bank, device);
282 n = snprintf(msg, len,
283 "DIMM location: not present. DMI handle: 0x%.4x ",
284 mem->mem_dev_handle);
289 void cper_mem_err_pack(const struct cper_sec_mem_err *mem,
290 struct cper_mem_err_compact *cmem)
292 cmem->validation_bits = mem->validation_bits;
293 cmem->node = mem->node;
294 cmem->card = mem->card;
295 cmem->module = mem->module;
296 cmem->bank = mem->bank;
297 cmem->device = mem->device;
298 cmem->row = mem->row;
299 cmem->column = mem->column;
300 cmem->bit_pos = mem->bit_pos;
301 cmem->requestor_id = mem->requestor_id;
302 cmem->responder_id = mem->responder_id;
303 cmem->target_id = mem->target_id;
304 cmem->extended = mem->extended;
305 cmem->rank = mem->rank;
306 cmem->mem_array_handle = mem->mem_array_handle;
307 cmem->mem_dev_handle = mem->mem_dev_handle;
310 const char *cper_mem_err_unpack(struct trace_seq *p,
311 struct cper_mem_err_compact *cmem)
313 const char *ret = trace_seq_buffer_ptr(p);
314 char rcd_decode_str[CPER_REC_LEN];
316 if (cper_mem_err_location(cmem, rcd_decode_str))
317 trace_seq_printf(p, "%s", rcd_decode_str);
318 if (cper_dimm_err_location(cmem, rcd_decode_str))
319 trace_seq_printf(p, "%s", rcd_decode_str);
320 trace_seq_putc(p, '\0');
325 static void cper_print_mem(const char *pfx, const struct cper_sec_mem_err *mem,
328 struct cper_mem_err_compact cmem;
329 char rcd_decode_str[CPER_REC_LEN];
331 /* Don't trust UEFI 2.1/2.2 structure with bad validation bits */
332 if (len == sizeof(struct cper_sec_mem_err_old) &&
333 (mem->validation_bits & ~(CPER_MEM_VALID_RANK_NUMBER - 1))) {
334 pr_err(FW_WARN "valid bits set for fields beyond structure\n");
337 if (mem->validation_bits & CPER_MEM_VALID_ERROR_STATUS)
338 printk("%s""error_status: 0x%016llx\n", pfx, mem->error_status);
339 if (mem->validation_bits & CPER_MEM_VALID_PA)
340 printk("%s""physical_address: 0x%016llx\n",
341 pfx, mem->physical_addr);
342 if (mem->validation_bits & CPER_MEM_VALID_PA_MASK)
343 printk("%s""physical_address_mask: 0x%016llx\n",
344 pfx, mem->physical_addr_mask);
345 cper_mem_err_pack(mem, &cmem);
346 if (cper_mem_err_location(&cmem, rcd_decode_str))
347 printk("%s%s\n", pfx, rcd_decode_str);
348 if (mem->validation_bits & CPER_MEM_VALID_ERROR_TYPE) {
349 u8 etype = mem->error_type;
350 printk("%s""error_type: %d, %s\n", pfx, etype,
351 cper_mem_err_type_str(etype));
353 if (cper_dimm_err_location(&cmem, rcd_decode_str))
354 printk("%s%s\n", pfx, rcd_decode_str);
357 static const char * const pcie_port_type_strs[] = {
359 "legacy PCI end point",
363 "upstream switch port",
364 "downstream switch port",
365 "PCIe to PCI/PCI-X bridge",
366 "PCI/PCI-X to PCIe bridge",
367 "root complex integrated endpoint device",
368 "root complex event collector",
371 static void cper_print_pcie(const char *pfx, const struct cper_sec_pcie *pcie,
372 const struct acpi_hest_generic_data *gdata)
374 if (pcie->validation_bits & CPER_PCIE_VALID_PORT_TYPE)
375 printk("%s""port_type: %d, %s\n", pfx, pcie->port_type,
376 pcie->port_type < ARRAY_SIZE(pcie_port_type_strs) ?
377 pcie_port_type_strs[pcie->port_type] : "unknown");
378 if (pcie->validation_bits & CPER_PCIE_VALID_VERSION)
379 printk("%s""version: %d.%d\n", pfx,
380 pcie->version.major, pcie->version.minor);
381 if (pcie->validation_bits & CPER_PCIE_VALID_COMMAND_STATUS)
382 printk("%s""command: 0x%04x, status: 0x%04x\n", pfx,
383 pcie->command, pcie->status);
384 if (pcie->validation_bits & CPER_PCIE_VALID_DEVICE_ID) {
386 printk("%s""device_id: %04x:%02x:%02x.%x\n", pfx,
387 pcie->device_id.segment, pcie->device_id.bus,
388 pcie->device_id.device, pcie->device_id.function);
389 printk("%s""slot: %d\n", pfx,
390 pcie->device_id.slot >> CPER_PCIE_SLOT_SHIFT);
391 printk("%s""secondary_bus: 0x%02x\n", pfx,
392 pcie->device_id.secondary_bus);
393 printk("%s""vendor_id: 0x%04x, device_id: 0x%04x\n", pfx,
394 pcie->device_id.vendor_id, pcie->device_id.device_id);
395 p = pcie->device_id.class_code;
396 printk("%s""class_code: %02x%02x%02x\n", pfx, p[2], p[1], p[0]);
398 if (pcie->validation_bits & CPER_PCIE_VALID_SERIAL_NUMBER)
399 printk("%s""serial number: 0x%04x, 0x%04x\n", pfx,
400 pcie->serial_number.lower, pcie->serial_number.upper);
401 if (pcie->validation_bits & CPER_PCIE_VALID_BRIDGE_CONTROL_STATUS)
403 "%s""bridge: secondary_status: 0x%04x, control: 0x%04x\n",
404 pfx, pcie->bridge.secondary_status, pcie->bridge.control);
406 /* Fatal errors call __ghes_panic() before AER handler prints this */
407 if ((pcie->validation_bits & CPER_PCIE_VALID_AER_INFO) &&
408 (gdata->error_severity & CPER_SEV_FATAL)) {
409 struct aer_capability_regs *aer;
411 aer = (struct aer_capability_regs *)pcie->aer_info;
412 printk("%saer_uncor_status: 0x%08x, aer_uncor_mask: 0x%08x\n",
413 pfx, aer->uncor_status, aer->uncor_mask);
414 printk("%saer_uncor_severity: 0x%08x\n",
415 pfx, aer->uncor_severity);
416 printk("%sTLP Header: %08x %08x %08x %08x\n", pfx,
417 aer->header_log.dw0, aer->header_log.dw1,
418 aer->header_log.dw2, aer->header_log.dw3);
422 static const char * const fw_err_rec_type_strs[] = {
423 "IPF SAL Error Record",
424 "SOC Firmware Error Record Type1 (Legacy CrashLog Support)",
425 "SOC Firmware Error Record Type2",
428 static void cper_print_fw_err(const char *pfx,
429 struct acpi_hest_generic_data *gdata,
430 const struct cper_sec_fw_err_rec_ref *fw_err)
432 void *buf = acpi_hest_get_payload(gdata);
433 u32 offset, length = gdata->error_data_length;
435 printk("%s""Firmware Error Record Type: %s\n", pfx,
436 fw_err->record_type < ARRAY_SIZE(fw_err_rec_type_strs) ?
437 fw_err_rec_type_strs[fw_err->record_type] : "unknown");
438 printk("%s""Revision: %d\n", pfx, fw_err->revision);
440 /* Record Type based on UEFI 2.7 */
441 if (fw_err->revision == 0) {
442 printk("%s""Record Identifier: %08llx\n", pfx,
443 fw_err->record_identifier);
444 } else if (fw_err->revision == 2) {
445 printk("%s""Record Identifier: %pUl\n", pfx,
446 &fw_err->record_identifier_guid);
450 * The FW error record may contain trailing data beyond the
451 * structure defined by the specification. As the fields
452 * defined (and hence the offset of any trailing data) vary
453 * with the revision, set the offset to account for this
456 if (fw_err->revision == 0) {
457 /* record_identifier_guid not defined */
458 offset = offsetof(struct cper_sec_fw_err_rec_ref,
459 record_identifier_guid);
460 } else if (fw_err->revision == 1) {
461 /* record_identifier not defined */
462 offset = offsetof(struct cper_sec_fw_err_rec_ref,
465 offset = sizeof(*fw_err);
471 print_hex_dump(pfx, "", DUMP_PREFIX_OFFSET, 16, 4, buf, length, true);
474 static void cper_print_tstamp(const char *pfx,
475 struct acpi_hest_generic_data_v300 *gdata)
477 __u8 hour, min, sec, day, mon, year, century, *timestamp;
479 if (gdata->validation_bits & ACPI_HEST_GEN_VALID_TIMESTAMP) {
480 timestamp = (__u8 *)&(gdata->time_stamp);
481 sec = bcd2bin(timestamp[0]);
482 min = bcd2bin(timestamp[1]);
483 hour = bcd2bin(timestamp[2]);
484 day = bcd2bin(timestamp[4]);
485 mon = bcd2bin(timestamp[5]);
486 year = bcd2bin(timestamp[6]);
487 century = bcd2bin(timestamp[7]);
489 printk("%s%ststamp: %02d%02d-%02d-%02d %02d:%02d:%02d\n", pfx,
490 (timestamp[3] & 0x1 ? "precise " : "imprecise "),
491 century, year, mon, day, hour, min, sec);
496 cper_estatus_print_section(const char *pfx, struct acpi_hest_generic_data *gdata,
499 guid_t *sec_type = (guid_t *)gdata->section_type;
503 if (acpi_hest_get_version(gdata) >= 3)
504 cper_print_tstamp(pfx, (struct acpi_hest_generic_data_v300 *)gdata);
506 severity = gdata->error_severity;
507 printk("%s""Error %d, type: %s\n", pfx, sec_no,
508 cper_severity_str(severity));
509 if (gdata->validation_bits & CPER_SEC_VALID_FRU_ID)
510 printk("%s""fru_id: %pUl\n", pfx, gdata->fru_id);
511 if (gdata->validation_bits & CPER_SEC_VALID_FRU_TEXT)
512 printk("%s""fru_text: %.20s\n", pfx, gdata->fru_text);
514 snprintf(newpfx, sizeof(newpfx), "%s ", pfx);
515 if (guid_equal(sec_type, &CPER_SEC_PROC_GENERIC)) {
516 struct cper_sec_proc_generic *proc_err = acpi_hest_get_payload(gdata);
518 printk("%s""section_type: general processor error\n", newpfx);
519 if (gdata->error_data_length >= sizeof(*proc_err))
520 cper_print_proc_generic(newpfx, proc_err);
522 goto err_section_too_small;
523 } else if (guid_equal(sec_type, &CPER_SEC_PLATFORM_MEM)) {
524 struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
526 printk("%s""section_type: memory error\n", newpfx);
527 if (gdata->error_data_length >=
528 sizeof(struct cper_sec_mem_err_old))
529 cper_print_mem(newpfx, mem_err,
530 gdata->error_data_length);
532 goto err_section_too_small;
533 } else if (guid_equal(sec_type, &CPER_SEC_PCIE)) {
534 struct cper_sec_pcie *pcie = acpi_hest_get_payload(gdata);
536 printk("%s""section_type: PCIe error\n", newpfx);
537 if (gdata->error_data_length >= sizeof(*pcie))
538 cper_print_pcie(newpfx, pcie, gdata);
540 goto err_section_too_small;
541 #if defined(CONFIG_ARM64) || defined(CONFIG_ARM)
542 } else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
543 struct cper_sec_proc_arm *arm_err = acpi_hest_get_payload(gdata);
545 printk("%ssection_type: ARM processor error\n", newpfx);
546 if (gdata->error_data_length >= sizeof(*arm_err))
547 cper_print_proc_arm(newpfx, arm_err);
549 goto err_section_too_small;
551 #if defined(CONFIG_UEFI_CPER_X86)
552 } else if (guid_equal(sec_type, &CPER_SEC_PROC_IA)) {
553 struct cper_sec_proc_ia *ia_err = acpi_hest_get_payload(gdata);
555 printk("%ssection_type: IA32/X64 processor error\n", newpfx);
556 if (gdata->error_data_length >= sizeof(*ia_err))
557 cper_print_proc_ia(newpfx, ia_err);
559 goto err_section_too_small;
561 } else if (guid_equal(sec_type, &CPER_SEC_FW_ERR_REC_REF)) {
562 struct cper_sec_fw_err_rec_ref *fw_err = acpi_hest_get_payload(gdata);
564 printk("%ssection_type: Firmware Error Record Reference\n",
566 /* The minimal FW Error Record contains 16 bytes */
567 if (gdata->error_data_length >= SZ_16)
568 cper_print_fw_err(newpfx, gdata, fw_err);
570 goto err_section_too_small;
572 const void *err = acpi_hest_get_payload(gdata);
574 printk("%ssection type: unknown, %pUl\n", newpfx, sec_type);
575 printk("%ssection length: %#x\n", newpfx,
576 gdata->error_data_length);
577 print_hex_dump(newpfx, "", DUMP_PREFIX_OFFSET, 16, 4, err,
578 gdata->error_data_length, true);
583 err_section_too_small:
584 pr_err(FW_WARN "error section length is too small\n");
587 void cper_estatus_print(const char *pfx,
588 const struct acpi_hest_generic_status *estatus)
590 struct acpi_hest_generic_data *gdata;
595 severity = estatus->error_severity;
596 if (severity == CPER_SEV_CORRECTED)
597 printk("%s%s\n", pfx,
598 "It has been corrected by h/w "
599 "and requires no further action");
600 printk("%s""event severity: %s\n", pfx, cper_severity_str(severity));
601 snprintf(newpfx, sizeof(newpfx), "%s ", pfx);
603 apei_estatus_for_each_section(estatus, gdata) {
604 cper_estatus_print_section(newpfx, gdata, sec_no);
608 EXPORT_SYMBOL_GPL(cper_estatus_print);
610 int cper_estatus_check_header(const struct acpi_hest_generic_status *estatus)
612 if (estatus->data_length &&
613 estatus->data_length < sizeof(struct acpi_hest_generic_data))
615 if (estatus->raw_data_length &&
616 estatus->raw_data_offset < sizeof(*estatus) + estatus->data_length)
621 EXPORT_SYMBOL_GPL(cper_estatus_check_header);
623 int cper_estatus_check(const struct acpi_hest_generic_status *estatus)
625 struct acpi_hest_generic_data *gdata;
626 unsigned int data_len, record_size;
629 rc = cper_estatus_check_header(estatus);
633 data_len = estatus->data_length;
635 apei_estatus_for_each_section(estatus, gdata) {
636 if (sizeof(struct acpi_hest_generic_data) > data_len)
639 record_size = acpi_hest_get_record_size(gdata);
640 if (record_size > data_len)
643 data_len -= record_size;
650 EXPORT_SYMBOL_GPL(cper_estatus_check);