1 // SPDX-License-Identifier: GPL-2.0
3 * UEFI Common Platform Error Record (CPER) support
5 * Copyright (C) 2010, Intel Corp.
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>
28 static char rcd_decode_str[CPER_REC_LEN];
31 * CPER record ID need to be unique even after reboot, because record
32 * ID is used as index for ERST storage, while CPER records from
33 * multiple boot may co-exist in ERST.
35 u64 cper_next_record_id(void)
37 static atomic64_t seq;
39 if (!atomic64_read(&seq)) {
40 time64_t time = ktime_get_real_seconds();
43 * This code is unlikely to still be needed in year 2106,
44 * but just in case, let's use a few more bits for timestamps
45 * after y2038 to be sure they keep increasing monotonically
46 * for the next few hundred years...
48 if (time < 0x80000000)
49 atomic64_set(&seq, (ktime_get_real_seconds()) << 32);
51 atomic64_set(&seq, 0x8000000000000000ull |
52 ktime_get_real_seconds() << 24);
55 return atomic64_inc_return(&seq);
57 EXPORT_SYMBOL_GPL(cper_next_record_id);
59 static const char * const severity_strs[] = {
66 const char *cper_severity_str(unsigned int severity)
68 return severity < ARRAY_SIZE(severity_strs) ?
69 severity_strs[severity] : "unknown";
71 EXPORT_SYMBOL_GPL(cper_severity_str);
74 * cper_print_bits - print strings for set bits
75 * @pfx: prefix for each line, including log level and prefix string
77 * @strs: string array, indexed by bit position
78 * @strs_size: size of the string array: @strs
80 * For each set bit in @bits, print the corresponding string in @strs.
81 * If the output length is longer than 80, multiple line will be
82 * printed, with @pfx is printed at the beginning of each line.
84 void cper_print_bits(const char *pfx, unsigned int bits,
85 const char * const strs[], unsigned int strs_size)
91 for (i = 0; i < strs_size; i++) {
92 if (!(bits & (1U << i)))
97 if (len && len + strlen(str) + 2 > 80) {
102 len = snprintf(buf, sizeof(buf), "%s%s", pfx, str);
104 len += scnprintf(buf+len, sizeof(buf)-len, ", %s", str);
110 static const char * const proc_type_strs[] = {
116 static const char * const proc_isa_strs[] = {
124 const char * const cper_proc_error_type_strs[] = {
128 "micro-architectural error",
131 static const char * const proc_op_strs[] = {
132 "unknown or generic",
135 "instruction execution",
138 static const char * const proc_flag_strs[] = {
145 static void cper_print_proc_generic(const char *pfx,
146 const struct cper_sec_proc_generic *proc)
148 if (proc->validation_bits & CPER_PROC_VALID_TYPE)
149 printk("%s""processor_type: %d, %s\n", pfx, proc->proc_type,
150 proc->proc_type < ARRAY_SIZE(proc_type_strs) ?
151 proc_type_strs[proc->proc_type] : "unknown");
152 if (proc->validation_bits & CPER_PROC_VALID_ISA)
153 printk("%s""processor_isa: %d, %s\n", pfx, proc->proc_isa,
154 proc->proc_isa < ARRAY_SIZE(proc_isa_strs) ?
155 proc_isa_strs[proc->proc_isa] : "unknown");
156 if (proc->validation_bits & CPER_PROC_VALID_ERROR_TYPE) {
157 printk("%s""error_type: 0x%02x\n", pfx, proc->proc_error_type);
158 cper_print_bits(pfx, proc->proc_error_type,
159 cper_proc_error_type_strs,
160 ARRAY_SIZE(cper_proc_error_type_strs));
162 if (proc->validation_bits & CPER_PROC_VALID_OPERATION)
163 printk("%s""operation: %d, %s\n", pfx, proc->operation,
164 proc->operation < ARRAY_SIZE(proc_op_strs) ?
165 proc_op_strs[proc->operation] : "unknown");
166 if (proc->validation_bits & CPER_PROC_VALID_FLAGS) {
167 printk("%s""flags: 0x%02x\n", pfx, proc->flags);
168 cper_print_bits(pfx, proc->flags, proc_flag_strs,
169 ARRAY_SIZE(proc_flag_strs));
171 if (proc->validation_bits & CPER_PROC_VALID_LEVEL)
172 printk("%s""level: %d\n", pfx, proc->level);
173 if (proc->validation_bits & CPER_PROC_VALID_VERSION)
174 printk("%s""version_info: 0x%016llx\n", pfx, proc->cpu_version);
175 if (proc->validation_bits & CPER_PROC_VALID_ID)
176 printk("%s""processor_id: 0x%016llx\n", pfx, proc->proc_id);
177 if (proc->validation_bits & CPER_PROC_VALID_TARGET_ADDRESS)
178 printk("%s""target_address: 0x%016llx\n",
179 pfx, proc->target_addr);
180 if (proc->validation_bits & CPER_PROC_VALID_REQUESTOR_ID)
181 printk("%s""requestor_id: 0x%016llx\n",
182 pfx, proc->requestor_id);
183 if (proc->validation_bits & CPER_PROC_VALID_RESPONDER_ID)
184 printk("%s""responder_id: 0x%016llx\n",
185 pfx, proc->responder_id);
186 if (proc->validation_bits & CPER_PROC_VALID_IP)
187 printk("%s""IP: 0x%016llx\n", pfx, proc->ip);
190 static const char * const mem_err_type_strs[] = {
195 "single-symbol chipkill ECC",
196 "multi-symbol chipkill ECC",
204 "scrub corrected error",
205 "scrub uncorrected error",
206 "physical memory map-out event",
209 const char *cper_mem_err_type_str(unsigned int etype)
211 return etype < ARRAY_SIZE(mem_err_type_strs) ?
212 mem_err_type_strs[etype] : "unknown";
214 EXPORT_SYMBOL_GPL(cper_mem_err_type_str);
216 static int cper_mem_err_location(struct cper_mem_err_compact *mem, char *msg)
224 len = CPER_REC_LEN - 1;
225 if (mem->validation_bits & CPER_MEM_VALID_NODE)
226 n += scnprintf(msg + n, len - n, "node: %d ", mem->node);
227 if (mem->validation_bits & CPER_MEM_VALID_CARD)
228 n += scnprintf(msg + n, len - n, "card: %d ", mem->card);
229 if (mem->validation_bits & CPER_MEM_VALID_MODULE)
230 n += scnprintf(msg + n, len - n, "module: %d ", mem->module);
231 if (mem->validation_bits & CPER_MEM_VALID_RANK_NUMBER)
232 n += scnprintf(msg + n, len - n, "rank: %d ", mem->rank);
233 if (mem->validation_bits & CPER_MEM_VALID_BANK)
234 n += scnprintf(msg + n, len - n, "bank: %d ", mem->bank);
235 if (mem->validation_bits & CPER_MEM_VALID_DEVICE)
236 n += scnprintf(msg + n, len - n, "device: %d ", mem->device);
237 if (mem->validation_bits & CPER_MEM_VALID_ROW)
238 n += scnprintf(msg + n, len - n, "row: %d ", mem->row);
239 if (mem->validation_bits & CPER_MEM_VALID_COLUMN)
240 n += scnprintf(msg + n, len - n, "column: %d ", mem->column);
241 if (mem->validation_bits & CPER_MEM_VALID_BIT_POSITION)
242 n += scnprintf(msg + n, len - n, "bit_position: %d ",
244 if (mem->validation_bits & CPER_MEM_VALID_REQUESTOR_ID)
245 n += scnprintf(msg + n, len - n, "requestor_id: 0x%016llx ",
247 if (mem->validation_bits & CPER_MEM_VALID_RESPONDER_ID)
248 n += scnprintf(msg + n, len - n, "responder_id: 0x%016llx ",
250 if (mem->validation_bits & CPER_MEM_VALID_TARGET_ID)
251 scnprintf(msg + n, len - n, "target_id: 0x%016llx ",
258 static int cper_dimm_err_location(struct cper_mem_err_compact *mem, char *msg)
261 const char *bank = NULL, *device = NULL;
263 if (!msg || !(mem->validation_bits & CPER_MEM_VALID_MODULE_HANDLE))
267 len = CPER_REC_LEN - 1;
268 dmi_memdev_name(mem->mem_dev_handle, &bank, &device);
270 n = snprintf(msg, len, "DIMM location: %s %s ", bank, device);
272 n = snprintf(msg, len,
273 "DIMM location: not present. DMI handle: 0x%.4x ",
274 mem->mem_dev_handle);
280 void cper_mem_err_pack(const struct cper_sec_mem_err *mem,
281 struct cper_mem_err_compact *cmem)
283 cmem->validation_bits = mem->validation_bits;
284 cmem->node = mem->node;
285 cmem->card = mem->card;
286 cmem->module = mem->module;
287 cmem->bank = mem->bank;
288 cmem->device = mem->device;
289 cmem->row = mem->row;
290 cmem->column = mem->column;
291 cmem->bit_pos = mem->bit_pos;
292 cmem->requestor_id = mem->requestor_id;
293 cmem->responder_id = mem->responder_id;
294 cmem->target_id = mem->target_id;
295 cmem->rank = mem->rank;
296 cmem->mem_array_handle = mem->mem_array_handle;
297 cmem->mem_dev_handle = mem->mem_dev_handle;
300 const char *cper_mem_err_unpack(struct trace_seq *p,
301 struct cper_mem_err_compact *cmem)
303 const char *ret = trace_seq_buffer_ptr(p);
305 if (cper_mem_err_location(cmem, rcd_decode_str))
306 trace_seq_printf(p, "%s", rcd_decode_str);
307 if (cper_dimm_err_location(cmem, rcd_decode_str))
308 trace_seq_printf(p, "%s", rcd_decode_str);
309 trace_seq_putc(p, '\0');
314 static void cper_print_mem(const char *pfx, const struct cper_sec_mem_err *mem,
317 struct cper_mem_err_compact cmem;
319 /* Don't trust UEFI 2.1/2.2 structure with bad validation bits */
320 if (len == sizeof(struct cper_sec_mem_err_old) &&
321 (mem->validation_bits & ~(CPER_MEM_VALID_RANK_NUMBER - 1))) {
322 pr_err(FW_WARN "valid bits set for fields beyond structure\n");
325 if (mem->validation_bits & CPER_MEM_VALID_ERROR_STATUS)
326 printk("%s""error_status: 0x%016llx\n", pfx, mem->error_status);
327 if (mem->validation_bits & CPER_MEM_VALID_PA)
328 printk("%s""physical_address: 0x%016llx\n",
329 pfx, mem->physical_addr);
330 if (mem->validation_bits & CPER_MEM_VALID_PA_MASK)
331 printk("%s""physical_address_mask: 0x%016llx\n",
332 pfx, mem->physical_addr_mask);
333 cper_mem_err_pack(mem, &cmem);
334 if (cper_mem_err_location(&cmem, rcd_decode_str))
335 printk("%s%s\n", pfx, rcd_decode_str);
336 if (mem->validation_bits & CPER_MEM_VALID_ERROR_TYPE) {
337 u8 etype = mem->error_type;
338 printk("%s""error_type: %d, %s\n", pfx, etype,
339 cper_mem_err_type_str(etype));
341 if (cper_dimm_err_location(&cmem, rcd_decode_str))
342 printk("%s%s\n", pfx, rcd_decode_str);
345 static const char * const pcie_port_type_strs[] = {
347 "legacy PCI end point",
351 "upstream switch port",
352 "downstream switch port",
353 "PCIe to PCI/PCI-X bridge",
354 "PCI/PCI-X to PCIe bridge",
355 "root complex integrated endpoint device",
356 "root complex event collector",
359 static void cper_print_pcie(const char *pfx, const struct cper_sec_pcie *pcie,
360 const struct acpi_hest_generic_data *gdata)
362 if (pcie->validation_bits & CPER_PCIE_VALID_PORT_TYPE)
363 printk("%s""port_type: %d, %s\n", pfx, pcie->port_type,
364 pcie->port_type < ARRAY_SIZE(pcie_port_type_strs) ?
365 pcie_port_type_strs[pcie->port_type] : "unknown");
366 if (pcie->validation_bits & CPER_PCIE_VALID_VERSION)
367 printk("%s""version: %d.%d\n", pfx,
368 pcie->version.major, pcie->version.minor);
369 if (pcie->validation_bits & CPER_PCIE_VALID_COMMAND_STATUS)
370 printk("%s""command: 0x%04x, status: 0x%04x\n", pfx,
371 pcie->command, pcie->status);
372 if (pcie->validation_bits & CPER_PCIE_VALID_DEVICE_ID) {
374 printk("%s""device_id: %04x:%02x:%02x.%x\n", pfx,
375 pcie->device_id.segment, pcie->device_id.bus,
376 pcie->device_id.device, pcie->device_id.function);
377 printk("%s""slot: %d\n", pfx,
378 pcie->device_id.slot >> CPER_PCIE_SLOT_SHIFT);
379 printk("%s""secondary_bus: 0x%02x\n", pfx,
380 pcie->device_id.secondary_bus);
381 printk("%s""vendor_id: 0x%04x, device_id: 0x%04x\n", pfx,
382 pcie->device_id.vendor_id, pcie->device_id.device_id);
383 p = pcie->device_id.class_code;
384 printk("%s""class_code: %02x%02x%02x\n", pfx, p[2], p[1], p[0]);
386 if (pcie->validation_bits & CPER_PCIE_VALID_SERIAL_NUMBER)
387 printk("%s""serial number: 0x%04x, 0x%04x\n", pfx,
388 pcie->serial_number.lower, pcie->serial_number.upper);
389 if (pcie->validation_bits & CPER_PCIE_VALID_BRIDGE_CONTROL_STATUS)
391 "%s""bridge: secondary_status: 0x%04x, control: 0x%04x\n",
392 pfx, pcie->bridge.secondary_status, pcie->bridge.control);
394 /* Fatal errors call __ghes_panic() before AER handler prints this */
395 if ((pcie->validation_bits & CPER_PCIE_VALID_AER_INFO) &&
396 (gdata->error_severity & CPER_SEV_FATAL)) {
397 struct aer_capability_regs *aer;
399 aer = (struct aer_capability_regs *)pcie->aer_info;
400 printk("%saer_uncor_status: 0x%08x, aer_uncor_mask: 0x%08x\n",
401 pfx, aer->uncor_status, aer->uncor_mask);
402 printk("%saer_uncor_severity: 0x%08x\n",
403 pfx, aer->uncor_severity);
404 printk("%sTLP Header: %08x %08x %08x %08x\n", pfx,
405 aer->header_log.dw0, aer->header_log.dw1,
406 aer->header_log.dw2, aer->header_log.dw3);
410 static const char * const fw_err_rec_type_strs[] = {
411 "IPF SAL Error Record",
412 "SOC Firmware Error Record Type1 (Legacy CrashLog Support)",
413 "SOC Firmware Error Record Type2",
416 static void cper_print_fw_err(const char *pfx,
417 struct acpi_hest_generic_data *gdata,
418 const struct cper_sec_fw_err_rec_ref *fw_err)
420 void *buf = acpi_hest_get_payload(gdata);
421 u32 offset, length = gdata->error_data_length;
423 printk("%s""Firmware Error Record Type: %s\n", pfx,
424 fw_err->record_type < ARRAY_SIZE(fw_err_rec_type_strs) ?
425 fw_err_rec_type_strs[fw_err->record_type] : "unknown");
426 printk("%s""Revision: %d\n", pfx, fw_err->revision);
428 /* Record Type based on UEFI 2.7 */
429 if (fw_err->revision == 0) {
430 printk("%s""Record Identifier: %08llx\n", pfx,
431 fw_err->record_identifier);
432 } else if (fw_err->revision == 2) {
433 printk("%s""Record Identifier: %pUl\n", pfx,
434 &fw_err->record_identifier_guid);
438 * The FW error record may contain trailing data beyond the
439 * structure defined by the specification. As the fields
440 * defined (and hence the offset of any trailing data) vary
441 * with the revision, set the offset to account for this
444 if (fw_err->revision == 0) {
445 /* record_identifier_guid not defined */
446 offset = offsetof(struct cper_sec_fw_err_rec_ref,
447 record_identifier_guid);
448 } else if (fw_err->revision == 1) {
449 /* record_identifier not defined */
450 offset = offsetof(struct cper_sec_fw_err_rec_ref,
453 offset = sizeof(*fw_err);
459 print_hex_dump(pfx, "", DUMP_PREFIX_OFFSET, 16, 4, buf, length, true);
462 static void cper_print_tstamp(const char *pfx,
463 struct acpi_hest_generic_data_v300 *gdata)
465 __u8 hour, min, sec, day, mon, year, century, *timestamp;
467 if (gdata->validation_bits & ACPI_HEST_GEN_VALID_TIMESTAMP) {
468 timestamp = (__u8 *)&(gdata->time_stamp);
469 sec = bcd2bin(timestamp[0]);
470 min = bcd2bin(timestamp[1]);
471 hour = bcd2bin(timestamp[2]);
472 day = bcd2bin(timestamp[4]);
473 mon = bcd2bin(timestamp[5]);
474 year = bcd2bin(timestamp[6]);
475 century = bcd2bin(timestamp[7]);
477 printk("%s%ststamp: %02d%02d-%02d-%02d %02d:%02d:%02d\n", pfx,
478 (timestamp[3] & 0x1 ? "precise " : "imprecise "),
479 century, year, mon, day, hour, min, sec);
484 cper_estatus_print_section(const char *pfx, struct acpi_hest_generic_data *gdata,
487 guid_t *sec_type = (guid_t *)gdata->section_type;
491 if (acpi_hest_get_version(gdata) >= 3)
492 cper_print_tstamp(pfx, (struct acpi_hest_generic_data_v300 *)gdata);
494 severity = gdata->error_severity;
495 printk("%s""Error %d, type: %s\n", pfx, sec_no,
496 cper_severity_str(severity));
497 if (gdata->validation_bits & CPER_SEC_VALID_FRU_ID)
498 printk("%s""fru_id: %pUl\n", pfx, gdata->fru_id);
499 if (gdata->validation_bits & CPER_SEC_VALID_FRU_TEXT)
500 printk("%s""fru_text: %.20s\n", pfx, gdata->fru_text);
502 snprintf(newpfx, sizeof(newpfx), "%s ", pfx);
503 if (guid_equal(sec_type, &CPER_SEC_PROC_GENERIC)) {
504 struct cper_sec_proc_generic *proc_err = acpi_hest_get_payload(gdata);
506 printk("%s""section_type: general processor error\n", newpfx);
507 if (gdata->error_data_length >= sizeof(*proc_err))
508 cper_print_proc_generic(newpfx, proc_err);
510 goto err_section_too_small;
511 } else if (guid_equal(sec_type, &CPER_SEC_PLATFORM_MEM)) {
512 struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
514 printk("%s""section_type: memory error\n", newpfx);
515 if (gdata->error_data_length >=
516 sizeof(struct cper_sec_mem_err_old))
517 cper_print_mem(newpfx, mem_err,
518 gdata->error_data_length);
520 goto err_section_too_small;
521 } else if (guid_equal(sec_type, &CPER_SEC_PCIE)) {
522 struct cper_sec_pcie *pcie = acpi_hest_get_payload(gdata);
524 printk("%s""section_type: PCIe error\n", newpfx);
525 if (gdata->error_data_length >= sizeof(*pcie))
526 cper_print_pcie(newpfx, pcie, gdata);
528 goto err_section_too_small;
529 #if defined(CONFIG_ARM64) || defined(CONFIG_ARM)
530 } else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
531 struct cper_sec_proc_arm *arm_err = acpi_hest_get_payload(gdata);
533 printk("%ssection_type: ARM processor error\n", newpfx);
534 if (gdata->error_data_length >= sizeof(*arm_err))
535 cper_print_proc_arm(newpfx, arm_err);
537 goto err_section_too_small;
539 #if defined(CONFIG_UEFI_CPER_X86)
540 } else if (guid_equal(sec_type, &CPER_SEC_PROC_IA)) {
541 struct cper_sec_proc_ia *ia_err = acpi_hest_get_payload(gdata);
543 printk("%ssection_type: IA32/X64 processor error\n", newpfx);
544 if (gdata->error_data_length >= sizeof(*ia_err))
545 cper_print_proc_ia(newpfx, ia_err);
547 goto err_section_too_small;
549 } else if (guid_equal(sec_type, &CPER_SEC_FW_ERR_REC_REF)) {
550 struct cper_sec_fw_err_rec_ref *fw_err = acpi_hest_get_payload(gdata);
552 printk("%ssection_type: Firmware Error Record Reference\n",
554 /* The minimal FW Error Record contains 16 bytes */
555 if (gdata->error_data_length >= SZ_16)
556 cper_print_fw_err(newpfx, gdata, fw_err);
558 goto err_section_too_small;
560 const void *err = acpi_hest_get_payload(gdata);
562 printk("%ssection type: unknown, %pUl\n", newpfx, sec_type);
563 printk("%ssection length: %#x\n", newpfx,
564 gdata->error_data_length);
565 print_hex_dump(newpfx, "", DUMP_PREFIX_OFFSET, 16, 4, err,
566 gdata->error_data_length, true);
571 err_section_too_small:
572 pr_err(FW_WARN "error section length is too small\n");
575 void cper_estatus_print(const char *pfx,
576 const struct acpi_hest_generic_status *estatus)
578 struct acpi_hest_generic_data *gdata;
583 severity = estatus->error_severity;
584 if (severity == CPER_SEV_CORRECTED)
585 printk("%s%s\n", pfx,
586 "It has been corrected by h/w "
587 "and requires no further action");
588 printk("%s""event severity: %s\n", pfx, cper_severity_str(severity));
589 snprintf(newpfx, sizeof(newpfx), "%s ", pfx);
591 apei_estatus_for_each_section(estatus, gdata) {
592 cper_estatus_print_section(newpfx, gdata, sec_no);
596 EXPORT_SYMBOL_GPL(cper_estatus_print);
598 int cper_estatus_check_header(const struct acpi_hest_generic_status *estatus)
600 if (estatus->data_length &&
601 estatus->data_length < sizeof(struct acpi_hest_generic_data))
603 if (estatus->raw_data_length &&
604 estatus->raw_data_offset < sizeof(*estatus) + estatus->data_length)
609 EXPORT_SYMBOL_GPL(cper_estatus_check_header);
611 int cper_estatus_check(const struct acpi_hest_generic_status *estatus)
613 struct acpi_hest_generic_data *gdata;
614 unsigned int data_len, record_size;
617 rc = cper_estatus_check_header(estatus);
621 data_len = estatus->data_length;
623 apei_estatus_for_each_section(estatus, gdata) {
624 if (sizeof(struct acpi_hest_generic_data) > data_len)
627 record_size = acpi_hest_get_record_size(gdata);
628 if (record_size > data_len)
631 data_len -= record_size;
638 EXPORT_SYMBOL_GPL(cper_estatus_check);