Loading...
1// SPDX-License-Identifier: GPL-2.0
2/*
3 * UEFI Common Platform Error Record (CPER) support
4 *
5 * Copyright (C) 2010, Intel Corp.
6 * Author: Huang Ying <ying.huang@intel.com>
7 *
8 * CPER is the format used to describe platform hardware error by
9 * various tables, such as ERST, BERT and HEST etc.
10 *
11 * For more information about CPER, please refer to Appendix N of UEFI
12 * Specification version 2.4.
13 */
14
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>
27#include "cper_cxl.h"
28
29/*
30 * CPER record ID need to be unique even after reboot, because record
31 * ID is used as index for ERST storage, while CPER records from
32 * multiple boot may co-exist in ERST.
33 */
34u64 cper_next_record_id(void)
35{
36 static atomic64_t seq;
37
38 if (!atomic64_read(&seq)) {
39 time64_t time = ktime_get_real_seconds();
40
41 /*
42 * This code is unlikely to still be needed in year 2106,
43 * but just in case, let's use a few more bits for timestamps
44 * after y2038 to be sure they keep increasing monotonically
45 * for the next few hundred years...
46 */
47 if (time < 0x80000000)
48 atomic64_set(&seq, (ktime_get_real_seconds()) << 32);
49 else
50 atomic64_set(&seq, 0x8000000000000000ull |
51 ktime_get_real_seconds() << 24);
52 }
53
54 return atomic64_inc_return(&seq);
55}
56EXPORT_SYMBOL_GPL(cper_next_record_id);
57
58static const char * const severity_strs[] = {
59 "recoverable",
60 "fatal",
61 "corrected",
62 "info",
63};
64
65const char *cper_severity_str(unsigned int severity)
66{
67 return severity < ARRAY_SIZE(severity_strs) ?
68 severity_strs[severity] : "unknown";
69}
70EXPORT_SYMBOL_GPL(cper_severity_str);
71
72/*
73 * cper_print_bits - print strings for set bits
74 * @pfx: prefix for each line, including log level and prefix string
75 * @bits: bit mask
76 * @strs: string array, indexed by bit position
77 * @strs_size: size of the string array: @strs
78 *
79 * For each set bit in @bits, print the corresponding string in @strs.
80 * If the output length is longer than 80, multiple line will be
81 * printed, with @pfx is printed at the beginning of each line.
82 */
83void cper_print_bits(const char *pfx, unsigned int bits,
84 const char * const strs[], unsigned int strs_size)
85{
86 int i, len = 0;
87 const char *str;
88 char buf[84];
89
90 for (i = 0; i < strs_size; i++) {
91 if (!(bits & (1U << i)))
92 continue;
93 str = strs[i];
94 if (!str)
95 continue;
96 if (len && len + strlen(str) + 2 > 80) {
97 printk("%s\n", buf);
98 len = 0;
99 }
100 if (!len)
101 len = snprintf(buf, sizeof(buf), "%s%s", pfx, str);
102 else
103 len += scnprintf(buf+len, sizeof(buf)-len, ", %s", str);
104 }
105 if (len)
106 printk("%s\n", buf);
107}
108
109static const char * const proc_type_strs[] = {
110 "IA32/X64",
111 "IA64",
112 "ARM",
113};
114
115static const char * const proc_isa_strs[] = {
116 "IA32",
117 "IA64",
118 "X64",
119 "ARM A32/T32",
120 "ARM A64",
121};
122
123const char * const cper_proc_error_type_strs[] = {
124 "cache error",
125 "TLB error",
126 "bus error",
127 "micro-architectural error",
128};
129
130static const char * const proc_op_strs[] = {
131 "unknown or generic",
132 "data read",
133 "data write",
134 "instruction execution",
135};
136
137static const char * const proc_flag_strs[] = {
138 "restartable",
139 "precise IP",
140 "overflow",
141 "corrected",
142};
143
144static void cper_print_proc_generic(const char *pfx,
145 const struct cper_sec_proc_generic *proc)
146{
147 if (proc->validation_bits & CPER_PROC_VALID_TYPE)
148 printk("%s""processor_type: %d, %s\n", pfx, proc->proc_type,
149 proc->proc_type < ARRAY_SIZE(proc_type_strs) ?
150 proc_type_strs[proc->proc_type] : "unknown");
151 if (proc->validation_bits & CPER_PROC_VALID_ISA)
152 printk("%s""processor_isa: %d, %s\n", pfx, proc->proc_isa,
153 proc->proc_isa < ARRAY_SIZE(proc_isa_strs) ?
154 proc_isa_strs[proc->proc_isa] : "unknown");
155 if (proc->validation_bits & CPER_PROC_VALID_ERROR_TYPE) {
156 printk("%s""error_type: 0x%02x\n", pfx, proc->proc_error_type);
157 cper_print_bits(pfx, proc->proc_error_type,
158 cper_proc_error_type_strs,
159 ARRAY_SIZE(cper_proc_error_type_strs));
160 }
161 if (proc->validation_bits & CPER_PROC_VALID_OPERATION)
162 printk("%s""operation: %d, %s\n", pfx, proc->operation,
163 proc->operation < ARRAY_SIZE(proc_op_strs) ?
164 proc_op_strs[proc->operation] : "unknown");
165 if (proc->validation_bits & CPER_PROC_VALID_FLAGS) {
166 printk("%s""flags: 0x%02x\n", pfx, proc->flags);
167 cper_print_bits(pfx, proc->flags, proc_flag_strs,
168 ARRAY_SIZE(proc_flag_strs));
169 }
170 if (proc->validation_bits & CPER_PROC_VALID_LEVEL)
171 printk("%s""level: %d\n", pfx, proc->level);
172 if (proc->validation_bits & CPER_PROC_VALID_VERSION)
173 printk("%s""version_info: 0x%016llx\n", pfx, proc->cpu_version);
174 if (proc->validation_bits & CPER_PROC_VALID_ID)
175 printk("%s""processor_id: 0x%016llx\n", pfx, proc->proc_id);
176 if (proc->validation_bits & CPER_PROC_VALID_TARGET_ADDRESS)
177 printk("%s""target_address: 0x%016llx\n",
178 pfx, proc->target_addr);
179 if (proc->validation_bits & CPER_PROC_VALID_REQUESTOR_ID)
180 printk("%s""requestor_id: 0x%016llx\n",
181 pfx, proc->requestor_id);
182 if (proc->validation_bits & CPER_PROC_VALID_RESPONDER_ID)
183 printk("%s""responder_id: 0x%016llx\n",
184 pfx, proc->responder_id);
185 if (proc->validation_bits & CPER_PROC_VALID_IP)
186 printk("%s""IP: 0x%016llx\n", pfx, proc->ip);
187}
188
189static const char * const mem_err_type_strs[] = {
190 "unknown",
191 "no error",
192 "single-bit ECC",
193 "multi-bit ECC",
194 "single-symbol chipkill ECC",
195 "multi-symbol chipkill ECC",
196 "master abort",
197 "target abort",
198 "parity error",
199 "watchdog timeout",
200 "invalid address",
201 "mirror Broken",
202 "memory sparing",
203 "scrub corrected error",
204 "scrub uncorrected error",
205 "physical memory map-out event",
206};
207
208const char *cper_mem_err_type_str(unsigned int etype)
209{
210 return etype < ARRAY_SIZE(mem_err_type_strs) ?
211 mem_err_type_strs[etype] : "unknown";
212}
213EXPORT_SYMBOL_GPL(cper_mem_err_type_str);
214
215const char *cper_mem_err_status_str(u64 status)
216{
217 switch ((status >> 8) & 0xff) {
218 case 1: return "Error detected internal to the component";
219 case 4: return "Storage error in DRAM memory";
220 case 5: return "Storage error in TLB";
221 case 6: return "Storage error in cache";
222 case 7: return "Error in one or more functional units";
223 case 8: return "Component failed self test";
224 case 9: return "Overflow or undervalue of internal queue";
225 case 16: return "Error detected in the bus";
226 case 17: return "Virtual address not found on IO-TLB or IO-PDIR";
227 case 18: return "Improper access error";
228 case 19: return "Access to a memory address which is not mapped to any component";
229 case 20: return "Loss of Lockstep";
230 case 21: return "Response not associated with a request";
231 case 22: return "Bus parity error - must also set the A, C, or D Bits";
232 case 23: return "Detection of a protocol error";
233 case 24: return "Detection of a PATH_ERROR";
234 case 25: return "Bus operation timeout";
235 case 26: return "A read was issued to data that has been poisoned";
236 default: return "Reserved";
237 }
238}
239EXPORT_SYMBOL_GPL(cper_mem_err_status_str);
240
241int cper_mem_err_location(struct cper_mem_err_compact *mem, char *msg)
242{
243 u32 len, n;
244
245 if (!msg)
246 return 0;
247
248 n = 0;
249 len = CPER_REC_LEN;
250 if (mem->validation_bits & CPER_MEM_VALID_NODE)
251 n += scnprintf(msg + n, len - n, "node:%d ", mem->node);
252 if (mem->validation_bits & CPER_MEM_VALID_CARD)
253 n += scnprintf(msg + n, len - n, "card:%d ", mem->card);
254 if (mem->validation_bits & CPER_MEM_VALID_MODULE)
255 n += scnprintf(msg + n, len - n, "module:%d ", mem->module);
256 if (mem->validation_bits & CPER_MEM_VALID_RANK_NUMBER)
257 n += scnprintf(msg + n, len - n, "rank:%d ", mem->rank);
258 if (mem->validation_bits & CPER_MEM_VALID_BANK)
259 n += scnprintf(msg + n, len - n, "bank:%d ", mem->bank);
260 if (mem->validation_bits & CPER_MEM_VALID_BANK_GROUP)
261 n += scnprintf(msg + n, len - n, "bank_group:%d ",
262 mem->bank >> CPER_MEM_BANK_GROUP_SHIFT);
263 if (mem->validation_bits & CPER_MEM_VALID_BANK_ADDRESS)
264 n += scnprintf(msg + n, len - n, "bank_address:%d ",
265 mem->bank & CPER_MEM_BANK_ADDRESS_MASK);
266 if (mem->validation_bits & CPER_MEM_VALID_DEVICE)
267 n += scnprintf(msg + n, len - n, "device:%d ", mem->device);
268 if (mem->validation_bits & (CPER_MEM_VALID_ROW | CPER_MEM_VALID_ROW_EXT)) {
269 u32 row = mem->row;
270
271 row |= cper_get_mem_extension(mem->validation_bits, mem->extended);
272 n += scnprintf(msg + n, len - n, "row:%d ", row);
273 }
274 if (mem->validation_bits & CPER_MEM_VALID_COLUMN)
275 n += scnprintf(msg + n, len - n, "column:%d ", mem->column);
276 if (mem->validation_bits & CPER_MEM_VALID_BIT_POSITION)
277 n += scnprintf(msg + n, len - n, "bit_position:%d ",
278 mem->bit_pos);
279 if (mem->validation_bits & CPER_MEM_VALID_REQUESTOR_ID)
280 n += scnprintf(msg + n, len - n, "requestor_id:0x%016llx ",
281 mem->requestor_id);
282 if (mem->validation_bits & CPER_MEM_VALID_RESPONDER_ID)
283 n += scnprintf(msg + n, len - n, "responder_id:0x%016llx ",
284 mem->responder_id);
285 if (mem->validation_bits & CPER_MEM_VALID_TARGET_ID)
286 n += scnprintf(msg + n, len - n, "target_id:0x%016llx ",
287 mem->target_id);
288 if (mem->validation_bits & CPER_MEM_VALID_CHIP_ID)
289 n += scnprintf(msg + n, len - n, "chip_id:%d ",
290 mem->extended >> CPER_MEM_CHIP_ID_SHIFT);
291
292 return n;
293}
294EXPORT_SYMBOL_GPL(cper_mem_err_location);
295
296int cper_dimm_err_location(struct cper_mem_err_compact *mem, char *msg)
297{
298 u32 len, n;
299 const char *bank = NULL, *device = NULL;
300
301 if (!msg || !(mem->validation_bits & CPER_MEM_VALID_MODULE_HANDLE))
302 return 0;
303
304 len = CPER_REC_LEN;
305 dmi_memdev_name(mem->mem_dev_handle, &bank, &device);
306 if (bank && device)
307 n = snprintf(msg, len, "DIMM location: %s %s ", bank, device);
308 else
309 n = snprintf(msg, len,
310 "DIMM location: not present. DMI handle: 0x%.4x ",
311 mem->mem_dev_handle);
312
313 return n;
314}
315EXPORT_SYMBOL_GPL(cper_dimm_err_location);
316
317void cper_mem_err_pack(const struct cper_sec_mem_err *mem,
318 struct cper_mem_err_compact *cmem)
319{
320 cmem->validation_bits = mem->validation_bits;
321 cmem->node = mem->node;
322 cmem->card = mem->card;
323 cmem->module = mem->module;
324 cmem->bank = mem->bank;
325 cmem->device = mem->device;
326 cmem->row = mem->row;
327 cmem->column = mem->column;
328 cmem->bit_pos = mem->bit_pos;
329 cmem->requestor_id = mem->requestor_id;
330 cmem->responder_id = mem->responder_id;
331 cmem->target_id = mem->target_id;
332 cmem->extended = mem->extended;
333 cmem->rank = mem->rank;
334 cmem->mem_array_handle = mem->mem_array_handle;
335 cmem->mem_dev_handle = mem->mem_dev_handle;
336}
337EXPORT_SYMBOL_GPL(cper_mem_err_pack);
338
339const char *cper_mem_err_unpack(struct trace_seq *p,
340 struct cper_mem_err_compact *cmem)
341{
342 const char *ret = trace_seq_buffer_ptr(p);
343 char rcd_decode_str[CPER_REC_LEN];
344
345 if (cper_mem_err_location(cmem, rcd_decode_str))
346 trace_seq_printf(p, "%s", rcd_decode_str);
347 if (cper_dimm_err_location(cmem, rcd_decode_str))
348 trace_seq_printf(p, "%s", rcd_decode_str);
349 trace_seq_putc(p, '\0');
350
351 return ret;
352}
353
354static void cper_print_mem(const char *pfx, const struct cper_sec_mem_err *mem,
355 int len)
356{
357 struct cper_mem_err_compact cmem;
358 char rcd_decode_str[CPER_REC_LEN];
359
360 /* Don't trust UEFI 2.1/2.2 structure with bad validation bits */
361 if (len == sizeof(struct cper_sec_mem_err_old) &&
362 (mem->validation_bits & ~(CPER_MEM_VALID_RANK_NUMBER - 1))) {
363 pr_err(FW_WARN "valid bits set for fields beyond structure\n");
364 return;
365 }
366 if (mem->validation_bits & CPER_MEM_VALID_ERROR_STATUS)
367 printk("%s error_status: %s (0x%016llx)\n",
368 pfx, cper_mem_err_status_str(mem->error_status),
369 mem->error_status);
370 if (mem->validation_bits & CPER_MEM_VALID_PA)
371 printk("%s""physical_address: 0x%016llx\n",
372 pfx, mem->physical_addr);
373 if (mem->validation_bits & CPER_MEM_VALID_PA_MASK)
374 printk("%s""physical_address_mask: 0x%016llx\n",
375 pfx, mem->physical_addr_mask);
376 cper_mem_err_pack(mem, &cmem);
377 if (cper_mem_err_location(&cmem, rcd_decode_str))
378 printk("%s%s\n", pfx, rcd_decode_str);
379 if (mem->validation_bits & CPER_MEM_VALID_ERROR_TYPE) {
380 u8 etype = mem->error_type;
381 printk("%s""error_type: %d, %s\n", pfx, etype,
382 cper_mem_err_type_str(etype));
383 }
384 if (cper_dimm_err_location(&cmem, rcd_decode_str))
385 printk("%s%s\n", pfx, rcd_decode_str);
386}
387
388static const char * const pcie_port_type_strs[] = {
389 "PCIe end point",
390 "legacy PCI end point",
391 "unknown",
392 "unknown",
393 "root port",
394 "upstream switch port",
395 "downstream switch port",
396 "PCIe to PCI/PCI-X bridge",
397 "PCI/PCI-X to PCIe bridge",
398 "root complex integrated endpoint device",
399 "root complex event collector",
400};
401
402static void cper_print_pcie(const char *pfx, const struct cper_sec_pcie *pcie,
403 const struct acpi_hest_generic_data *gdata)
404{
405 if (pcie->validation_bits & CPER_PCIE_VALID_PORT_TYPE)
406 printk("%s""port_type: %d, %s\n", pfx, pcie->port_type,
407 pcie->port_type < ARRAY_SIZE(pcie_port_type_strs) ?
408 pcie_port_type_strs[pcie->port_type] : "unknown");
409 if (pcie->validation_bits & CPER_PCIE_VALID_VERSION)
410 printk("%s""version: %d.%d\n", pfx,
411 pcie->version.major, pcie->version.minor);
412 if (pcie->validation_bits & CPER_PCIE_VALID_COMMAND_STATUS)
413 printk("%s""command: 0x%04x, status: 0x%04x\n", pfx,
414 pcie->command, pcie->status);
415 if (pcie->validation_bits & CPER_PCIE_VALID_DEVICE_ID) {
416 const __u8 *p;
417 printk("%s""device_id: %04x:%02x:%02x.%x\n", pfx,
418 pcie->device_id.segment, pcie->device_id.bus,
419 pcie->device_id.device, pcie->device_id.function);
420 printk("%s""slot: %d\n", pfx,
421 pcie->device_id.slot >> CPER_PCIE_SLOT_SHIFT);
422 printk("%s""secondary_bus: 0x%02x\n", pfx,
423 pcie->device_id.secondary_bus);
424 printk("%s""vendor_id: 0x%04x, device_id: 0x%04x\n", pfx,
425 pcie->device_id.vendor_id, pcie->device_id.device_id);
426 p = pcie->device_id.class_code;
427 printk("%s""class_code: %02x%02x%02x\n", pfx, p[2], p[1], p[0]);
428 }
429 if (pcie->validation_bits & CPER_PCIE_VALID_SERIAL_NUMBER)
430 printk("%s""serial number: 0x%04x, 0x%04x\n", pfx,
431 pcie->serial_number.lower, pcie->serial_number.upper);
432 if (pcie->validation_bits & CPER_PCIE_VALID_BRIDGE_CONTROL_STATUS)
433 printk(
434 "%s""bridge: secondary_status: 0x%04x, control: 0x%04x\n",
435 pfx, pcie->bridge.secondary_status, pcie->bridge.control);
436
437 /* Fatal errors call __ghes_panic() before AER handler prints this */
438 if ((pcie->validation_bits & CPER_PCIE_VALID_AER_INFO) &&
439 (gdata->error_severity & CPER_SEV_FATAL)) {
440 struct aer_capability_regs *aer;
441
442 aer = (struct aer_capability_regs *)pcie->aer_info;
443 printk("%saer_uncor_status: 0x%08x, aer_uncor_mask: 0x%08x\n",
444 pfx, aer->uncor_status, aer->uncor_mask);
445 printk("%saer_uncor_severity: 0x%08x\n",
446 pfx, aer->uncor_severity);
447 printk("%sTLP Header: %08x %08x %08x %08x\n", pfx,
448 aer->header_log.dw0, aer->header_log.dw1,
449 aer->header_log.dw2, aer->header_log.dw3);
450 }
451}
452
453static const char * const fw_err_rec_type_strs[] = {
454 "IPF SAL Error Record",
455 "SOC Firmware Error Record Type1 (Legacy CrashLog Support)",
456 "SOC Firmware Error Record Type2",
457};
458
459static void cper_print_fw_err(const char *pfx,
460 struct acpi_hest_generic_data *gdata,
461 const struct cper_sec_fw_err_rec_ref *fw_err)
462{
463 void *buf = acpi_hest_get_payload(gdata);
464 u32 offset, length = gdata->error_data_length;
465
466 printk("%s""Firmware Error Record Type: %s\n", pfx,
467 fw_err->record_type < ARRAY_SIZE(fw_err_rec_type_strs) ?
468 fw_err_rec_type_strs[fw_err->record_type] : "unknown");
469 printk("%s""Revision: %d\n", pfx, fw_err->revision);
470
471 /* Record Type based on UEFI 2.7 */
472 if (fw_err->revision == 0) {
473 printk("%s""Record Identifier: %08llx\n", pfx,
474 fw_err->record_identifier);
475 } else if (fw_err->revision == 2) {
476 printk("%s""Record Identifier: %pUl\n", pfx,
477 &fw_err->record_identifier_guid);
478 }
479
480 /*
481 * The FW error record may contain trailing data beyond the
482 * structure defined by the specification. As the fields
483 * defined (and hence the offset of any trailing data) vary
484 * with the revision, set the offset to account for this
485 * variation.
486 */
487 if (fw_err->revision == 0) {
488 /* record_identifier_guid not defined */
489 offset = offsetof(struct cper_sec_fw_err_rec_ref,
490 record_identifier_guid);
491 } else if (fw_err->revision == 1) {
492 /* record_identifier not defined */
493 offset = offsetof(struct cper_sec_fw_err_rec_ref,
494 record_identifier);
495 } else {
496 offset = sizeof(*fw_err);
497 }
498
499 buf += offset;
500 length -= offset;
501
502 print_hex_dump(pfx, "", DUMP_PREFIX_OFFSET, 16, 4, buf, length, true);
503}
504
505static void cper_print_tstamp(const char *pfx,
506 struct acpi_hest_generic_data_v300 *gdata)
507{
508 __u8 hour, min, sec, day, mon, year, century, *timestamp;
509
510 if (gdata->validation_bits & ACPI_HEST_GEN_VALID_TIMESTAMP) {
511 timestamp = (__u8 *)&(gdata->time_stamp);
512 sec = bcd2bin(timestamp[0]);
513 min = bcd2bin(timestamp[1]);
514 hour = bcd2bin(timestamp[2]);
515 day = bcd2bin(timestamp[4]);
516 mon = bcd2bin(timestamp[5]);
517 year = bcd2bin(timestamp[6]);
518 century = bcd2bin(timestamp[7]);
519
520 printk("%s%ststamp: %02d%02d-%02d-%02d %02d:%02d:%02d\n", pfx,
521 (timestamp[3] & 0x1 ? "precise " : "imprecise "),
522 century, year, mon, day, hour, min, sec);
523 }
524}
525
526static void
527cper_estatus_print_section(const char *pfx, struct acpi_hest_generic_data *gdata,
528 int sec_no)
529{
530 guid_t *sec_type = (guid_t *)gdata->section_type;
531 __u16 severity;
532 char newpfx[64];
533
534 if (acpi_hest_get_version(gdata) >= 3)
535 cper_print_tstamp(pfx, (struct acpi_hest_generic_data_v300 *)gdata);
536
537 severity = gdata->error_severity;
538 printk("%s""Error %d, type: %s\n", pfx, sec_no,
539 cper_severity_str(severity));
540 if (gdata->validation_bits & CPER_SEC_VALID_FRU_ID)
541 printk("%s""fru_id: %pUl\n", pfx, gdata->fru_id);
542 if (gdata->validation_bits & CPER_SEC_VALID_FRU_TEXT)
543 printk("%s""fru_text: %.20s\n", pfx, gdata->fru_text);
544
545 snprintf(newpfx, sizeof(newpfx), "%s ", pfx);
546 if (guid_equal(sec_type, &CPER_SEC_PROC_GENERIC)) {
547 struct cper_sec_proc_generic *proc_err = acpi_hest_get_payload(gdata);
548
549 printk("%s""section_type: general processor error\n", newpfx);
550 if (gdata->error_data_length >= sizeof(*proc_err))
551 cper_print_proc_generic(newpfx, proc_err);
552 else
553 goto err_section_too_small;
554 } else if (guid_equal(sec_type, &CPER_SEC_PLATFORM_MEM)) {
555 struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
556
557 printk("%s""section_type: memory error\n", newpfx);
558 if (gdata->error_data_length >=
559 sizeof(struct cper_sec_mem_err_old))
560 cper_print_mem(newpfx, mem_err,
561 gdata->error_data_length);
562 else
563 goto err_section_too_small;
564 } else if (guid_equal(sec_type, &CPER_SEC_PCIE)) {
565 struct cper_sec_pcie *pcie = acpi_hest_get_payload(gdata);
566
567 printk("%s""section_type: PCIe error\n", newpfx);
568 if (gdata->error_data_length >= sizeof(*pcie))
569 cper_print_pcie(newpfx, pcie, gdata);
570 else
571 goto err_section_too_small;
572#if defined(CONFIG_ARM64) || defined(CONFIG_ARM)
573 } else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
574 struct cper_sec_proc_arm *arm_err = acpi_hest_get_payload(gdata);
575
576 printk("%ssection_type: ARM processor error\n", newpfx);
577 if (gdata->error_data_length >= sizeof(*arm_err))
578 cper_print_proc_arm(newpfx, arm_err);
579 else
580 goto err_section_too_small;
581#endif
582#if defined(CONFIG_UEFI_CPER_X86)
583 } else if (guid_equal(sec_type, &CPER_SEC_PROC_IA)) {
584 struct cper_sec_proc_ia *ia_err = acpi_hest_get_payload(gdata);
585
586 printk("%ssection_type: IA32/X64 processor error\n", newpfx);
587 if (gdata->error_data_length >= sizeof(*ia_err))
588 cper_print_proc_ia(newpfx, ia_err);
589 else
590 goto err_section_too_small;
591#endif
592 } else if (guid_equal(sec_type, &CPER_SEC_FW_ERR_REC_REF)) {
593 struct cper_sec_fw_err_rec_ref *fw_err = acpi_hest_get_payload(gdata);
594
595 printk("%ssection_type: Firmware Error Record Reference\n",
596 newpfx);
597 /* The minimal FW Error Record contains 16 bytes */
598 if (gdata->error_data_length >= SZ_16)
599 cper_print_fw_err(newpfx, gdata, fw_err);
600 else
601 goto err_section_too_small;
602 } else if (guid_equal(sec_type, &CPER_SEC_CXL_PROT_ERR)) {
603 struct cper_sec_prot_err *prot_err = acpi_hest_get_payload(gdata);
604
605 printk("%ssection_type: CXL Protocol Error\n", newpfx);
606 if (gdata->error_data_length >= sizeof(*prot_err))
607 cper_print_prot_err(newpfx, prot_err);
608 else
609 goto err_section_too_small;
610 } else {
611 const void *err = acpi_hest_get_payload(gdata);
612
613 printk("%ssection type: unknown, %pUl\n", newpfx, sec_type);
614 printk("%ssection length: %#x\n", newpfx,
615 gdata->error_data_length);
616 print_hex_dump(newpfx, "", DUMP_PREFIX_OFFSET, 16, 4, err,
617 gdata->error_data_length, true);
618 }
619
620 return;
621
622err_section_too_small:
623 pr_err(FW_WARN "error section length is too small\n");
624}
625
626void cper_estatus_print(const char *pfx,
627 const struct acpi_hest_generic_status *estatus)
628{
629 struct acpi_hest_generic_data *gdata;
630 int sec_no = 0;
631 char newpfx[64];
632 __u16 severity;
633
634 severity = estatus->error_severity;
635 if (severity == CPER_SEV_CORRECTED)
636 printk("%s%s\n", pfx,
637 "It has been corrected by h/w "
638 "and requires no further action");
639 printk("%s""event severity: %s\n", pfx, cper_severity_str(severity));
640 snprintf(newpfx, sizeof(newpfx), "%s ", pfx);
641
642 apei_estatus_for_each_section(estatus, gdata) {
643 cper_estatus_print_section(newpfx, gdata, sec_no);
644 sec_no++;
645 }
646}
647EXPORT_SYMBOL_GPL(cper_estatus_print);
648
649int cper_estatus_check_header(const struct acpi_hest_generic_status *estatus)
650{
651 if (estatus->data_length &&
652 estatus->data_length < sizeof(struct acpi_hest_generic_data))
653 return -EINVAL;
654 if (estatus->raw_data_length &&
655 estatus->raw_data_offset < sizeof(*estatus) + estatus->data_length)
656 return -EINVAL;
657
658 return 0;
659}
660EXPORT_SYMBOL_GPL(cper_estatus_check_header);
661
662int cper_estatus_check(const struct acpi_hest_generic_status *estatus)
663{
664 struct acpi_hest_generic_data *gdata;
665 unsigned int data_len, record_size;
666 int rc;
667
668 rc = cper_estatus_check_header(estatus);
669 if (rc)
670 return rc;
671
672 data_len = estatus->data_length;
673
674 apei_estatus_for_each_section(estatus, gdata) {
675 if (acpi_hest_get_size(gdata) > data_len)
676 return -EINVAL;
677
678 record_size = acpi_hest_get_record_size(gdata);
679 if (record_size > data_len)
680 return -EINVAL;
681
682 data_len -= record_size;
683 }
684 if (data_len)
685 return -EINVAL;
686
687 return 0;
688}
689EXPORT_SYMBOL_GPL(cper_estatus_check);
1/*
2 * UEFI Common Platform Error Record (CPER) support
3 *
4 * Copyright (C) 2010, Intel Corp.
5 * Author: Huang Ying <ying.huang@intel.com>
6 *
7 * CPER is the format used to describe platform hardware error by
8 * various tables, such as ERST, BERT and HEST etc.
9 *
10 * For more information about CPER, please refer to Appendix N of UEFI
11 * Specification version 2.4.
12 *
13 * This program is free software; you can redistribute it and/or
14 * modify it under the terms of the GNU General Public License version
15 * 2 as published by the Free Software Foundation.
16 *
17 * This program is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20 * GNU General Public License for more details.
21 *
22 * You should have received a copy of the GNU General Public License
23 * along with this program; if not, write to the Free Software
24 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
25 */
26
27#include <linux/kernel.h>
28#include <linux/module.h>
29#include <linux/time.h>
30#include <linux/cper.h>
31#include <linux/dmi.h>
32#include <linux/acpi.h>
33#include <linux/pci.h>
34#include <linux/aer.h>
35
36#define INDENT_SP " "
37/*
38 * CPER record ID need to be unique even after reboot, because record
39 * ID is used as index for ERST storage, while CPER records from
40 * multiple boot may co-exist in ERST.
41 */
42u64 cper_next_record_id(void)
43{
44 static atomic64_t seq;
45
46 if (!atomic64_read(&seq))
47 atomic64_set(&seq, ((u64)get_seconds()) << 32);
48
49 return atomic64_inc_return(&seq);
50}
51EXPORT_SYMBOL_GPL(cper_next_record_id);
52
53static const char *cper_severity_strs[] = {
54 "recoverable",
55 "fatal",
56 "corrected",
57 "info",
58};
59
60static const char *cper_severity_str(unsigned int severity)
61{
62 return severity < ARRAY_SIZE(cper_severity_strs) ?
63 cper_severity_strs[severity] : "unknown";
64}
65
66/*
67 * cper_print_bits - print strings for set bits
68 * @pfx: prefix for each line, including log level and prefix string
69 * @bits: bit mask
70 * @strs: string array, indexed by bit position
71 * @strs_size: size of the string array: @strs
72 *
73 * For each set bit in @bits, print the corresponding string in @strs.
74 * If the output length is longer than 80, multiple line will be
75 * printed, with @pfx is printed at the beginning of each line.
76 */
77void cper_print_bits(const char *pfx, unsigned int bits,
78 const char * const strs[], unsigned int strs_size)
79{
80 int i, len = 0;
81 const char *str;
82 char buf[84];
83
84 for (i = 0; i < strs_size; i++) {
85 if (!(bits & (1U << i)))
86 continue;
87 str = strs[i];
88 if (!str)
89 continue;
90 if (len && len + strlen(str) + 2 > 80) {
91 printk("%s\n", buf);
92 len = 0;
93 }
94 if (!len)
95 len = snprintf(buf, sizeof(buf), "%s%s", pfx, str);
96 else
97 len += snprintf(buf+len, sizeof(buf)-len, ", %s", str);
98 }
99 if (len)
100 printk("%s\n", buf);
101}
102
103static const char * const cper_proc_type_strs[] = {
104 "IA32/X64",
105 "IA64",
106};
107
108static const char * const cper_proc_isa_strs[] = {
109 "IA32",
110 "IA64",
111 "X64",
112};
113
114static const char * const cper_proc_error_type_strs[] = {
115 "cache error",
116 "TLB error",
117 "bus error",
118 "micro-architectural error",
119};
120
121static const char * const cper_proc_op_strs[] = {
122 "unknown or generic",
123 "data read",
124 "data write",
125 "instruction execution",
126};
127
128static const char * const cper_proc_flag_strs[] = {
129 "restartable",
130 "precise IP",
131 "overflow",
132 "corrected",
133};
134
135static void cper_print_proc_generic(const char *pfx,
136 const struct cper_sec_proc_generic *proc)
137{
138 if (proc->validation_bits & CPER_PROC_VALID_TYPE)
139 printk("%s""processor_type: %d, %s\n", pfx, proc->proc_type,
140 proc->proc_type < ARRAY_SIZE(cper_proc_type_strs) ?
141 cper_proc_type_strs[proc->proc_type] : "unknown");
142 if (proc->validation_bits & CPER_PROC_VALID_ISA)
143 printk("%s""processor_isa: %d, %s\n", pfx, proc->proc_isa,
144 proc->proc_isa < ARRAY_SIZE(cper_proc_isa_strs) ?
145 cper_proc_isa_strs[proc->proc_isa] : "unknown");
146 if (proc->validation_bits & CPER_PROC_VALID_ERROR_TYPE) {
147 printk("%s""error_type: 0x%02x\n", pfx, proc->proc_error_type);
148 cper_print_bits(pfx, proc->proc_error_type,
149 cper_proc_error_type_strs,
150 ARRAY_SIZE(cper_proc_error_type_strs));
151 }
152 if (proc->validation_bits & CPER_PROC_VALID_OPERATION)
153 printk("%s""operation: %d, %s\n", pfx, proc->operation,
154 proc->operation < ARRAY_SIZE(cper_proc_op_strs) ?
155 cper_proc_op_strs[proc->operation] : "unknown");
156 if (proc->validation_bits & CPER_PROC_VALID_FLAGS) {
157 printk("%s""flags: 0x%02x\n", pfx, proc->flags);
158 cper_print_bits(pfx, proc->flags, cper_proc_flag_strs,
159 ARRAY_SIZE(cper_proc_flag_strs));
160 }
161 if (proc->validation_bits & CPER_PROC_VALID_LEVEL)
162 printk("%s""level: %d\n", pfx, proc->level);
163 if (proc->validation_bits & CPER_PROC_VALID_VERSION)
164 printk("%s""version_info: 0x%016llx\n", pfx, proc->cpu_version);
165 if (proc->validation_bits & CPER_PROC_VALID_ID)
166 printk("%s""processor_id: 0x%016llx\n", pfx, proc->proc_id);
167 if (proc->validation_bits & CPER_PROC_VALID_TARGET_ADDRESS)
168 printk("%s""target_address: 0x%016llx\n",
169 pfx, proc->target_addr);
170 if (proc->validation_bits & CPER_PROC_VALID_REQUESTOR_ID)
171 printk("%s""requestor_id: 0x%016llx\n",
172 pfx, proc->requestor_id);
173 if (proc->validation_bits & CPER_PROC_VALID_RESPONDER_ID)
174 printk("%s""responder_id: 0x%016llx\n",
175 pfx, proc->responder_id);
176 if (proc->validation_bits & CPER_PROC_VALID_IP)
177 printk("%s""IP: 0x%016llx\n", pfx, proc->ip);
178}
179
180static const char *cper_mem_err_type_strs[] = {
181 "unknown",
182 "no error",
183 "single-bit ECC",
184 "multi-bit ECC",
185 "single-symbol chipkill ECC",
186 "multi-symbol chipkill ECC",
187 "master abort",
188 "target abort",
189 "parity error",
190 "watchdog timeout",
191 "invalid address",
192 "mirror Broken",
193 "memory sparing",
194 "scrub corrected error",
195 "scrub uncorrected error",
196 "physical memory map-out event",
197};
198
199static void cper_print_mem(const char *pfx, const struct cper_sec_mem_err *mem)
200{
201 if (mem->validation_bits & CPER_MEM_VALID_ERROR_STATUS)
202 printk("%s""error_status: 0x%016llx\n", pfx, mem->error_status);
203 if (mem->validation_bits & CPER_MEM_VALID_PA)
204 printk("%s""physical_address: 0x%016llx\n",
205 pfx, mem->physical_addr);
206 if (mem->validation_bits & CPER_MEM_VALID_PA_MASK)
207 printk("%s""physical_address_mask: 0x%016llx\n",
208 pfx, mem->physical_addr_mask);
209 if (mem->validation_bits & CPER_MEM_VALID_NODE)
210 pr_debug("node: %d\n", mem->node);
211 if (mem->validation_bits & CPER_MEM_VALID_CARD)
212 pr_debug("card: %d\n", mem->card);
213 if (mem->validation_bits & CPER_MEM_VALID_MODULE)
214 pr_debug("module: %d\n", mem->module);
215 if (mem->validation_bits & CPER_MEM_VALID_RANK_NUMBER)
216 pr_debug("rank: %d\n", mem->rank);
217 if (mem->validation_bits & CPER_MEM_VALID_BANK)
218 pr_debug("bank: %d\n", mem->bank);
219 if (mem->validation_bits & CPER_MEM_VALID_DEVICE)
220 pr_debug("device: %d\n", mem->device);
221 if (mem->validation_bits & CPER_MEM_VALID_ROW)
222 pr_debug("row: %d\n", mem->row);
223 if (mem->validation_bits & CPER_MEM_VALID_COLUMN)
224 pr_debug("column: %d\n", mem->column);
225 if (mem->validation_bits & CPER_MEM_VALID_BIT_POSITION)
226 pr_debug("bit_position: %d\n", mem->bit_pos);
227 if (mem->validation_bits & CPER_MEM_VALID_REQUESTOR_ID)
228 pr_debug("requestor_id: 0x%016llx\n", mem->requestor_id);
229 if (mem->validation_bits & CPER_MEM_VALID_RESPONDER_ID)
230 pr_debug("responder_id: 0x%016llx\n", mem->responder_id);
231 if (mem->validation_bits & CPER_MEM_VALID_TARGET_ID)
232 pr_debug("target_id: 0x%016llx\n", mem->target_id);
233 if (mem->validation_bits & CPER_MEM_VALID_ERROR_TYPE) {
234 u8 etype = mem->error_type;
235 printk("%s""error_type: %d, %s\n", pfx, etype,
236 etype < ARRAY_SIZE(cper_mem_err_type_strs) ?
237 cper_mem_err_type_strs[etype] : "unknown");
238 }
239 if (mem->validation_bits & CPER_MEM_VALID_MODULE_HANDLE) {
240 const char *bank = NULL, *device = NULL;
241 dmi_memdev_name(mem->mem_dev_handle, &bank, &device);
242 if (bank != NULL && device != NULL)
243 printk("%s""DIMM location: %s %s", pfx, bank, device);
244 else
245 printk("%s""DIMM DMI handle: 0x%.4x",
246 pfx, mem->mem_dev_handle);
247 }
248}
249
250static const char *cper_pcie_port_type_strs[] = {
251 "PCIe end point",
252 "legacy PCI end point",
253 "unknown",
254 "unknown",
255 "root port",
256 "upstream switch port",
257 "downstream switch port",
258 "PCIe to PCI/PCI-X bridge",
259 "PCI/PCI-X to PCIe bridge",
260 "root complex integrated endpoint device",
261 "root complex event collector",
262};
263
264static void cper_print_pcie(const char *pfx, const struct cper_sec_pcie *pcie,
265 const struct acpi_generic_data *gdata)
266{
267 if (pcie->validation_bits & CPER_PCIE_VALID_PORT_TYPE)
268 printk("%s""port_type: %d, %s\n", pfx, pcie->port_type,
269 pcie->port_type < ARRAY_SIZE(cper_pcie_port_type_strs) ?
270 cper_pcie_port_type_strs[pcie->port_type] : "unknown");
271 if (pcie->validation_bits & CPER_PCIE_VALID_VERSION)
272 printk("%s""version: %d.%d\n", pfx,
273 pcie->version.major, pcie->version.minor);
274 if (pcie->validation_bits & CPER_PCIE_VALID_COMMAND_STATUS)
275 printk("%s""command: 0x%04x, status: 0x%04x\n", pfx,
276 pcie->command, pcie->status);
277 if (pcie->validation_bits & CPER_PCIE_VALID_DEVICE_ID) {
278 const __u8 *p;
279 printk("%s""device_id: %04x:%02x:%02x.%x\n", pfx,
280 pcie->device_id.segment, pcie->device_id.bus,
281 pcie->device_id.device, pcie->device_id.function);
282 printk("%s""slot: %d\n", pfx,
283 pcie->device_id.slot >> CPER_PCIE_SLOT_SHIFT);
284 printk("%s""secondary_bus: 0x%02x\n", pfx,
285 pcie->device_id.secondary_bus);
286 printk("%s""vendor_id: 0x%04x, device_id: 0x%04x\n", pfx,
287 pcie->device_id.vendor_id, pcie->device_id.device_id);
288 p = pcie->device_id.class_code;
289 printk("%s""class_code: %02x%02x%02x\n", pfx, p[0], p[1], p[2]);
290 }
291 if (pcie->validation_bits & CPER_PCIE_VALID_SERIAL_NUMBER)
292 printk("%s""serial number: 0x%04x, 0x%04x\n", pfx,
293 pcie->serial_number.lower, pcie->serial_number.upper);
294 if (pcie->validation_bits & CPER_PCIE_VALID_BRIDGE_CONTROL_STATUS)
295 printk(
296 "%s""bridge: secondary_status: 0x%04x, control: 0x%04x\n",
297 pfx, pcie->bridge.secondary_status, pcie->bridge.control);
298}
299
300static void cper_estatus_print_section(
301 const char *pfx, const struct acpi_generic_data *gdata, int sec_no)
302{
303 uuid_le *sec_type = (uuid_le *)gdata->section_type;
304 __u16 severity;
305 char newpfx[64];
306
307 severity = gdata->error_severity;
308 printk("%s""Error %d, type: %s\n", pfx, sec_no,
309 cper_severity_str(severity));
310 if (gdata->validation_bits & CPER_SEC_VALID_FRU_ID)
311 printk("%s""fru_id: %pUl\n", pfx, (uuid_le *)gdata->fru_id);
312 if (gdata->validation_bits & CPER_SEC_VALID_FRU_TEXT)
313 printk("%s""fru_text: %.20s\n", pfx, gdata->fru_text);
314
315 snprintf(newpfx, sizeof(newpfx), "%s%s", pfx, INDENT_SP);
316 if (!uuid_le_cmp(*sec_type, CPER_SEC_PROC_GENERIC)) {
317 struct cper_sec_proc_generic *proc_err = (void *)(gdata + 1);
318 printk("%s""section_type: general processor error\n", newpfx);
319 if (gdata->error_data_length >= sizeof(*proc_err))
320 cper_print_proc_generic(newpfx, proc_err);
321 else
322 goto err_section_too_small;
323 } else if (!uuid_le_cmp(*sec_type, CPER_SEC_PLATFORM_MEM)) {
324 struct cper_sec_mem_err *mem_err = (void *)(gdata + 1);
325 printk("%s""section_type: memory error\n", newpfx);
326 if (gdata->error_data_length >= sizeof(*mem_err))
327 cper_print_mem(newpfx, mem_err);
328 else
329 goto err_section_too_small;
330 } else if (!uuid_le_cmp(*sec_type, CPER_SEC_PCIE)) {
331 struct cper_sec_pcie *pcie = (void *)(gdata + 1);
332 printk("%s""section_type: PCIe error\n", newpfx);
333 if (gdata->error_data_length >= sizeof(*pcie))
334 cper_print_pcie(newpfx, pcie, gdata);
335 else
336 goto err_section_too_small;
337 } else
338 printk("%s""section type: unknown, %pUl\n", newpfx, sec_type);
339
340 return;
341
342err_section_too_small:
343 pr_err(FW_WARN "error section length is too small\n");
344}
345
346void cper_estatus_print(const char *pfx,
347 const struct acpi_generic_status *estatus)
348{
349 struct acpi_generic_data *gdata;
350 unsigned int data_len, gedata_len;
351 int sec_no = 0;
352 char newpfx[64];
353 __u16 severity;
354
355 severity = estatus->error_severity;
356 if (severity == CPER_SEV_CORRECTED)
357 printk("%s%s\n", pfx,
358 "It has been corrected by h/w "
359 "and requires no further action");
360 printk("%s""event severity: %s\n", pfx, cper_severity_str(severity));
361 data_len = estatus->data_length;
362 gdata = (struct acpi_generic_data *)(estatus + 1);
363 snprintf(newpfx, sizeof(newpfx), "%s%s", pfx, INDENT_SP);
364 while (data_len >= sizeof(*gdata)) {
365 gedata_len = gdata->error_data_length;
366 cper_estatus_print_section(newpfx, gdata, sec_no);
367 data_len -= gedata_len + sizeof(*gdata);
368 gdata = (void *)(gdata + 1) + gedata_len;
369 sec_no++;
370 }
371}
372EXPORT_SYMBOL_GPL(cper_estatus_print);
373
374int cper_estatus_check_header(const struct acpi_generic_status *estatus)
375{
376 if (estatus->data_length &&
377 estatus->data_length < sizeof(struct acpi_generic_data))
378 return -EINVAL;
379 if (estatus->raw_data_length &&
380 estatus->raw_data_offset < sizeof(*estatus) + estatus->data_length)
381 return -EINVAL;
382
383 return 0;
384}
385EXPORT_SYMBOL_GPL(cper_estatus_check_header);
386
387int cper_estatus_check(const struct acpi_generic_status *estatus)
388{
389 struct acpi_generic_data *gdata;
390 unsigned int data_len, gedata_len;
391 int rc;
392
393 rc = cper_estatus_check_header(estatus);
394 if (rc)
395 return rc;
396 data_len = estatus->data_length;
397 gdata = (struct acpi_generic_data *)(estatus + 1);
398 while (data_len >= sizeof(*gdata)) {
399 gedata_len = gdata->error_data_length;
400 if (gedata_len > data_len - sizeof(*gdata))
401 return -EINVAL;
402 data_len -= gedata_len + sizeof(*gdata);
403 gdata = (void *)(gdata + 1) + gedata_len;
404 }
405 if (data_len)
406 return -EINVAL;
407
408 return 0;
409}
410EXPORT_SYMBOL_GPL(cper_estatus_check);