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_HW_ERROR_FW_CFG_FILE "etc/hardware_errors"
32 #define ACPI_HW_ERROR_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 */
build_ghes_hw_error_notification(GArray * table,const uint8_t type)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 */
acpi_ghes_generic_error_data(GArray * table,const uint8_t * section_type,uint32_t error_severity,uint8_t validation_bits,uint8_t flags,uint32_t error_data_length,QemuUUID fru_id,uint64_t time_stamp)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 */
acpi_ghes_generic_error_status(GArray * table,uint32_t block_status,uint32_t raw_data_offset,uint32_t raw_data_length,uint32_t data_length,uint32_t error_severity)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 */
acpi_ghes_build_append_mem_cper(GArray * table,uint64_t error_physical_addr)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 void
ghes_gen_err_data_uncorrectable_recoverable(GArray * block,const uint8_t * section_type,int data_length)185 ghes_gen_err_data_uncorrectable_recoverable(GArray *block,
186 const uint8_t *section_type,
187 int data_length)
188 {
189 /* invalid fru id: ACPI 4.0: 17.3.2.6.1 Generic Error Data,
190 * Table 17-13 Generic Error Data Entry
191 */
192 QemuUUID fru_id = {};
193
194 /* Build the new generic error status block header */
195 acpi_ghes_generic_error_status(block, ACPI_GEBS_UNCORRECTABLE,
196 0, 0, data_length, ACPI_CPER_SEV_RECOVERABLE);
197
198 /* Build this new generic error data entry header */
199 acpi_ghes_generic_error_data(block, section_type,
200 ACPI_CPER_SEV_RECOVERABLE, 0, 0,
201 ACPI_GHES_MEM_CPER_LENGTH, fru_id, 0);
202 }
203
204 /*
205 * Build table for the hardware error fw_cfg blob.
206 * Initialize "etc/hardware_errors" and "etc/hardware_errors_addr" fw_cfg blobs.
207 * See docs/specs/acpi_hest_ghes.rst for blobs format.
208 */
build_ghes_error_table(GArray * hardware_errors,BIOSLinker * linker)209 static void build_ghes_error_table(GArray *hardware_errors, BIOSLinker *linker)
210 {
211 int i, error_status_block_offset;
212
213 /* Build error_block_address */
214 for (i = 0; i < ACPI_GHES_ERROR_SOURCE_COUNT; i++) {
215 build_append_int_noprefix(hardware_errors, 0, sizeof(uint64_t));
216 }
217
218 /* Build read_ack_register */
219 for (i = 0; i < ACPI_GHES_ERROR_SOURCE_COUNT; i++) {
220 /*
221 * Initialize the value of read_ack_register to 1, so GHES can be
222 * writable after (re)boot.
223 * ACPI 6.2: 18.3.2.8 Generic Hardware Error Source version 2
224 * (GHESv2 - Type 10)
225 */
226 build_append_int_noprefix(hardware_errors, 1, sizeof(uint64_t));
227 }
228
229 /* Generic Error Status Block offset in the hardware error fw_cfg blob */
230 error_status_block_offset = hardware_errors->len;
231
232 /* Reserve space for Error Status Data Block */
233 acpi_data_push(hardware_errors,
234 ACPI_GHES_MAX_RAW_DATA_LENGTH * ACPI_GHES_ERROR_SOURCE_COUNT);
235
236 /* Tell guest firmware to place hardware_errors blob into RAM */
237 bios_linker_loader_alloc(linker, ACPI_HW_ERROR_FW_CFG_FILE,
238 hardware_errors, sizeof(uint64_t), false);
239
240 for (i = 0; i < ACPI_GHES_ERROR_SOURCE_COUNT; i++) {
241 /*
242 * Tell firmware to patch error_block_address entries to point to
243 * corresponding "Generic Error Status Block"
244 */
245 bios_linker_loader_add_pointer(linker,
246 ACPI_HW_ERROR_FW_CFG_FILE,
247 sizeof(uint64_t) * i,
248 sizeof(uint64_t),
249 ACPI_HW_ERROR_FW_CFG_FILE,
250 error_status_block_offset +
251 i * ACPI_GHES_MAX_RAW_DATA_LENGTH);
252 }
253
254 /*
255 * tell firmware to write hardware_errors GPA into
256 * hardware_errors_addr fw_cfg, once the former has been initialized.
257 */
258 bios_linker_loader_write_pointer(linker, ACPI_HW_ERROR_ADDR_FW_CFG_FILE, 0,
259 sizeof(uint64_t),
260 ACPI_HW_ERROR_FW_CFG_FILE, 0);
261 }
262
263 /* Build Generic Hardware Error Source version 2 (GHESv2) */
build_ghes_v2(GArray * table_data,BIOSLinker * linker,enum AcpiGhesNotifyType notify,uint16_t source_id)264 static void build_ghes_v2(GArray *table_data,
265 BIOSLinker *linker,
266 enum AcpiGhesNotifyType notify,
267 uint16_t source_id)
268 {
269 uint64_t address_offset;
270
271 /*
272 * Type:
273 * Generic Hardware Error Source version 2(GHESv2 - Type 10)
274 */
275 build_append_int_noprefix(table_data, ACPI_GHES_SOURCE_GENERIC_ERROR_V2, 2);
276 /* Source Id */
277 build_append_int_noprefix(table_data, source_id, 2);
278 /* Related Source Id */
279 build_append_int_noprefix(table_data, 0xffff, 2);
280 /* Flags */
281 build_append_int_noprefix(table_data, 0, 1);
282 /* Enabled */
283 build_append_int_noprefix(table_data, 1, 1);
284
285 /* Number of Records To Pre-allocate */
286 build_append_int_noprefix(table_data, 1, 4);
287 /* Max Sections Per Record */
288 build_append_int_noprefix(table_data, 1, 4);
289 /* Max Raw Data Length */
290 build_append_int_noprefix(table_data, ACPI_GHES_MAX_RAW_DATA_LENGTH, 4);
291
292 address_offset = table_data->len;
293 /* Error Status Address */
294 build_append_gas(table_data, AML_AS_SYSTEM_MEMORY, 0x40, 0,
295 4 /* QWord access */, 0);
296 bios_linker_loader_add_pointer(linker, ACPI_BUILD_TABLE_FILE,
297 address_offset + GAS_ADDR_OFFSET,
298 sizeof(uint64_t),
299 ACPI_HW_ERROR_FW_CFG_FILE,
300 source_id * sizeof(uint64_t));
301
302 /* Notification Structure */
303 build_ghes_hw_error_notification(table_data, notify);
304
305 /* Error Status Block Length */
306 build_append_int_noprefix(table_data, ACPI_GHES_MAX_RAW_DATA_LENGTH, 4);
307
308 /*
309 * Read Ack Register
310 * ACPI 6.1: 18.3.2.8 Generic Hardware Error Source
311 * version 2 (GHESv2 - Type 10)
312 */
313 address_offset = table_data->len;
314 build_append_gas(table_data, AML_AS_SYSTEM_MEMORY, 0x40, 0,
315 4 /* QWord access */, 0);
316 bios_linker_loader_add_pointer(linker, ACPI_BUILD_TABLE_FILE,
317 address_offset + GAS_ADDR_OFFSET,
318 sizeof(uint64_t),
319 ACPI_HW_ERROR_FW_CFG_FILE,
320 (ACPI_GHES_ERROR_SOURCE_COUNT + source_id)
321 * sizeof(uint64_t));
322
323 /*
324 * Read Ack Preserve field
325 * We only provide the first bit in Read Ack Register to OSPM to write
326 * while the other bits are preserved.
327 */
328 build_append_int_noprefix(table_data, ~0x1ULL, 8);
329 /* Read Ack Write */
330 build_append_int_noprefix(table_data, 0x1, 8);
331 }
332
333 /* Build Hardware Error Source Table */
acpi_build_hest(GArray * table_data,GArray * hardware_errors,BIOSLinker * linker,const char * oem_id,const char * oem_table_id)334 void acpi_build_hest(GArray *table_data, GArray *hardware_errors,
335 BIOSLinker *linker,
336 const char *oem_id, const char *oem_table_id)
337 {
338 AcpiTable table = { .sig = "HEST", .rev = 1,
339 .oem_id = oem_id, .oem_table_id = oem_table_id };
340
341 build_ghes_error_table(hardware_errors, linker);
342
343 acpi_table_begin(&table, table_data);
344
345 /* Error Source Count */
346 build_append_int_noprefix(table_data, ACPI_GHES_ERROR_SOURCE_COUNT, 4);
347 build_ghes_v2(table_data, linker,
348 ACPI_GHES_NOTIFY_SEA, ACPI_HEST_SRC_ID_SEA);
349
350 acpi_table_end(linker, &table);
351 }
352
acpi_ghes_add_fw_cfg(AcpiGhesState * ags,FWCfgState * s,GArray * hardware_error)353 void acpi_ghes_add_fw_cfg(AcpiGhesState *ags, FWCfgState *s,
354 GArray *hardware_error)
355 {
356 /* Create a read-only fw_cfg file for GHES */
357 fw_cfg_add_file(s, ACPI_HW_ERROR_FW_CFG_FILE, hardware_error->data,
358 hardware_error->len);
359
360 /* Create a read-write fw_cfg file for Address */
361 fw_cfg_add_file_callback(s, ACPI_HW_ERROR_ADDR_FW_CFG_FILE, NULL, NULL,
362 NULL, &(ags->hw_error_le), sizeof(ags->hw_error_le), false);
363
364 ags->present = true;
365 }
366
get_hw_error_offsets(uint64_t ghes_addr,uint64_t * cper_addr,uint64_t * read_ack_register_addr)367 static void get_hw_error_offsets(uint64_t ghes_addr,
368 uint64_t *cper_addr,
369 uint64_t *read_ack_register_addr)
370 {
371 if (!ghes_addr) {
372 return;
373 }
374
375 /*
376 * non-HEST version supports only one source, so no need to change
377 * the start offset based on the source ID. Also, we can't validate
378 * the source ID, as it is stored inside the HEST table.
379 */
380
381 cpu_physical_memory_read(ghes_addr, cper_addr,
382 sizeof(*cper_addr));
383
384 *cper_addr = le64_to_cpu(*cper_addr);
385
386 /*
387 * As the current version supports only one source, the ack offset is
388 * just sizeof(uint64_t).
389 */
390 *read_ack_register_addr = ghes_addr + sizeof(uint64_t);
391 }
392
ghes_record_cper_errors(const void * cper,size_t len,uint16_t source_id,Error ** errp)393 static void ghes_record_cper_errors(const void *cper, size_t len,
394 uint16_t source_id, Error **errp)
395 {
396 uint64_t cper_addr = 0, read_ack_register_addr = 0, read_ack_register;
397 AcpiGedState *acpi_ged_state;
398 AcpiGhesState *ags;
399
400 if (len > ACPI_GHES_MAX_RAW_DATA_LENGTH) {
401 error_setg(errp, "GHES CPER record is too big: %zd", len);
402 return;
403 }
404
405 acpi_ged_state = ACPI_GED(object_resolve_path_type("", TYPE_ACPI_GED,
406 NULL));
407 if (!acpi_ged_state) {
408 error_setg(errp, "Can't find ACPI_GED object");
409 return;
410 }
411 ags = &acpi_ged_state->ghes_state;
412
413 assert(ACPI_GHES_ERROR_SOURCE_COUNT == 1);
414 get_hw_error_offsets(le64_to_cpu(ags->hw_error_le),
415 &cper_addr, &read_ack_register_addr);
416
417 if (!cper_addr) {
418 error_setg(errp, "can not find Generic Error Status Block");
419 return;
420 }
421
422 cpu_physical_memory_read(read_ack_register_addr,
423 &read_ack_register, sizeof(read_ack_register));
424
425 /* zero means OSPM does not acknowledge the error */
426 if (!read_ack_register) {
427 error_setg(errp,
428 "OSPM does not acknowledge previous error,"
429 " so can not record CPER for current error anymore");
430 return;
431 }
432
433 read_ack_register = cpu_to_le64(0);
434 /*
435 * Clear the Read Ack Register, OSPM will write 1 to this register when
436 * it acknowledges the error.
437 */
438 cpu_physical_memory_write(read_ack_register_addr,
439 &read_ack_register, sizeof(uint64_t));
440
441 /* Write the generic error data entry into guest memory */
442 cpu_physical_memory_write(cper_addr, cper, len);
443 }
444
acpi_ghes_memory_errors(uint16_t source_id,uint64_t physical_address)445 int acpi_ghes_memory_errors(uint16_t source_id, uint64_t physical_address)
446 {
447 /* Memory Error Section Type */
448 const uint8_t guid[] =
449 UUID_LE(0xA5BC1114, 0x6F64, 0x4EDE, 0xB8, 0x63, 0x3E, 0x83, \
450 0xED, 0x7C, 0x83, 0xB1);
451 Error *errp = NULL;
452 int data_length;
453 GArray *block;
454
455 block = g_array_new(false, true /* clear */, 1);
456
457 data_length = ACPI_GHES_DATA_LENGTH + ACPI_GHES_MEM_CPER_LENGTH;
458 /*
459 * It should not run out of the preallocated memory if adding a new generic
460 * error data entry
461 */
462 assert((data_length + ACPI_GHES_GESB_SIZE) <=
463 ACPI_GHES_MAX_RAW_DATA_LENGTH);
464
465 ghes_gen_err_data_uncorrectable_recoverable(block, guid, data_length);
466
467 /* Build the memory section CPER for above new generic error data entry */
468 acpi_ghes_build_append_mem_cper(block, physical_address);
469
470 /* Report the error */
471 ghes_record_cper_errors(block->data, block->len, source_id, &errp);
472
473 g_array_free(block, true);
474
475 if (errp) {
476 error_report_err(errp);
477 return -1;
478 }
479
480 return 0;
481 }
482
acpi_ghes_present(void)483 bool acpi_ghes_present(void)
484 {
485 AcpiGedState *acpi_ged_state;
486 AcpiGhesState *ags;
487
488 acpi_ged_state = ACPI_GED(object_resolve_path_type("", TYPE_ACPI_GED,
489 NULL));
490
491 if (!acpi_ged_state) {
492 return false;
493 }
494 ags = &acpi_ged_state->ghes_state;
495 return ags->present;
496 }
497