xref: /openbmc/qemu/hw/acpi/ghes.c (revision 2e1cacfb)
1 /*
2  * Support for generating APEI tables and recording CPER for Guests
3  *
4  * Copyright (c) 2020 HUAWEI TECHNOLOGIES CO., LTD.
5  *
6  * Author: Dongjiu Geng <gengdongjiu@huawei.com>
7  *
8  * This program is free software; you can redistribute it and/or modify
9  * it under the terms of the GNU General Public License as published by
10  * the Free Software Foundation; either version 2 of the License, or
11  * (at your option) any later version.
12 
13  * This program is distributed in the hope that it will be useful,
14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
16  * GNU General Public License for more details.
17 
18  * You should have received a copy of the GNU General Public License along
19  * with this program; if not, see <http://www.gnu.org/licenses/>.
20  */
21 
22 #include "qemu/osdep.h"
23 #include "qemu/units.h"
24 #include "hw/acpi/ghes.h"
25 #include "hw/acpi/aml-build.h"
26 #include "qemu/error-report.h"
27 #include "hw/acpi/generic_event_device.h"
28 #include "hw/nvram/fw_cfg.h"
29 #include "qemu/uuid.h"
30 
31 #define ACPI_GHES_ERRORS_FW_CFG_FILE        "etc/hardware_errors"
32 #define ACPI_GHES_DATA_ADDR_FW_CFG_FILE     "etc/hardware_errors_addr"
33 
34 /* The max size in bytes for one error block */
35 #define ACPI_GHES_MAX_RAW_DATA_LENGTH   (1 * KiB)
36 
37 /* Now only support ARMv8 SEA notification type error source */
38 #define ACPI_GHES_ERROR_SOURCE_COUNT        1
39 
40 /* Generic Hardware Error Source version 2 */
41 #define ACPI_GHES_SOURCE_GENERIC_ERROR_V2   10
42 
43 /* Address offset in Generic Address Structure(GAS) */
44 #define GAS_ADDR_OFFSET 4
45 
46 /*
47  * The total size of Generic Error Data Entry
48  * ACPI 6.1/6.2: 18.3.2.7.1 Generic Error Data,
49  * Table 18-343 Generic Error Data Entry
50  */
51 #define ACPI_GHES_DATA_LENGTH               72
52 
53 /* The memory section CPER size, UEFI 2.6: N.2.5 Memory Error Section */
54 #define ACPI_GHES_MEM_CPER_LENGTH           80
55 
56 /* Masks for block_status flags */
57 #define ACPI_GEBS_UNCORRECTABLE         1
58 
59 /*
60  * Total size for Generic Error Status Block except Generic Error Data Entries
61  * ACPI 6.2: 18.3.2.7.1 Generic Error Data,
62  * Table 18-380 Generic Error Status Block
63  */
64 #define ACPI_GHES_GESB_SIZE                 20
65 
66 /*
67  * Values for error_severity field
68  */
69 enum AcpiGenericErrorSeverity {
70     ACPI_CPER_SEV_RECOVERABLE = 0,
71     ACPI_CPER_SEV_FATAL = 1,
72     ACPI_CPER_SEV_CORRECTED = 2,
73     ACPI_CPER_SEV_NONE = 3,
74 };
75 
76 /*
77  * Hardware Error Notification
78  * ACPI 4.0: 17.3.2.7 Hardware Error Notification
79  * Composes dummy Hardware Error Notification descriptor of specified type
80  */
81 static void build_ghes_hw_error_notification(GArray *table, const uint8_t type)
82 {
83     /* Type */
84     build_append_int_noprefix(table, type, 1);
85     /*
86      * Length:
87      * Total length of the structure in bytes
88      */
89     build_append_int_noprefix(table, 28, 1);
90     /* Configuration Write Enable */
91     build_append_int_noprefix(table, 0, 2);
92     /* Poll Interval */
93     build_append_int_noprefix(table, 0, 4);
94     /* Vector */
95     build_append_int_noprefix(table, 0, 4);
96     /* Switch To Polling Threshold Value */
97     build_append_int_noprefix(table, 0, 4);
98     /* Switch To Polling Threshold Window */
99     build_append_int_noprefix(table, 0, 4);
100     /* Error Threshold Value */
101     build_append_int_noprefix(table, 0, 4);
102     /* Error Threshold Window */
103     build_append_int_noprefix(table, 0, 4);
104 }
105 
106 /*
107  * Generic Error Data Entry
108  * ACPI 6.1: 18.3.2.7.1 Generic Error Data
109  */
110 static void acpi_ghes_generic_error_data(GArray *table,
111                 const uint8_t *section_type, uint32_t error_severity,
112                 uint8_t validation_bits, uint8_t flags,
113                 uint32_t error_data_length, QemuUUID fru_id,
114                 uint64_t time_stamp)
115 {
116     const uint8_t fru_text[20] = {0};
117 
118     /* Section Type */
119     g_array_append_vals(table, section_type, 16);
120 
121     /* Error Severity */
122     build_append_int_noprefix(table, error_severity, 4);
123     /* Revision */
124     build_append_int_noprefix(table, 0x300, 2);
125     /* Validation Bits */
126     build_append_int_noprefix(table, validation_bits, 1);
127     /* Flags */
128     build_append_int_noprefix(table, flags, 1);
129     /* Error Data Length */
130     build_append_int_noprefix(table, error_data_length, 4);
131 
132     /* FRU Id */
133     g_array_append_vals(table, fru_id.data, ARRAY_SIZE(fru_id.data));
134 
135     /* FRU Text */
136     g_array_append_vals(table, fru_text, sizeof(fru_text));
137 
138     /* Timestamp */
139     build_append_int_noprefix(table, time_stamp, 8);
140 }
141 
142 /*
143  * Generic Error Status Block
144  * ACPI 6.1: 18.3.2.7.1 Generic Error Data
145  */
146 static void acpi_ghes_generic_error_status(GArray *table, uint32_t block_status,
147                 uint32_t raw_data_offset, uint32_t raw_data_length,
148                 uint32_t data_length, uint32_t error_severity)
149 {
150     /* Block Status */
151     build_append_int_noprefix(table, block_status, 4);
152     /* Raw Data Offset */
153     build_append_int_noprefix(table, raw_data_offset, 4);
154     /* Raw Data Length */
155     build_append_int_noprefix(table, raw_data_length, 4);
156     /* Data Length */
157     build_append_int_noprefix(table, data_length, 4);
158     /* Error Severity */
159     build_append_int_noprefix(table, error_severity, 4);
160 }
161 
162 /* UEFI 2.6: N.2.5 Memory Error Section */
163 static void acpi_ghes_build_append_mem_cper(GArray *table,
164                                             uint64_t error_physical_addr)
165 {
166     /*
167      * Memory Error Record
168      */
169 
170     /* Validation Bits */
171     build_append_int_noprefix(table,
172                               (1ULL << 14) | /* Type Valid */
173                               (1ULL << 1) /* Physical Address Valid */,
174                               8);
175     /* Error Status */
176     build_append_int_noprefix(table, 0, 8);
177     /* Physical Address */
178     build_append_int_noprefix(table, error_physical_addr, 8);
179     /* Skip all the detailed information normally found in such a record */
180     build_append_int_noprefix(table, 0, 48);
181     /* Memory Error Type */
182     build_append_int_noprefix(table, 0 /* Unknown error */, 1);
183     /* Skip all the detailed information normally found in such a record */
184     build_append_int_noprefix(table, 0, 7);
185 }
186 
187 static int acpi_ghes_record_mem_error(uint64_t error_block_address,
188                                       uint64_t error_physical_addr)
189 {
190     GArray *block;
191 
192     /* Memory Error Section Type */
193     const uint8_t uefi_cper_mem_sec[] =
194           UUID_LE(0xA5BC1114, 0x6F64, 0x4EDE, 0xB8, 0x63, 0x3E, 0x83, \
195                   0xED, 0x7C, 0x83, 0xB1);
196 
197     /* invalid fru id: ACPI 4.0: 17.3.2.6.1 Generic Error Data,
198      * Table 17-13 Generic Error Data Entry
199      */
200     QemuUUID fru_id = {};
201     uint32_t data_length;
202 
203     block = g_array_new(false, true /* clear */, 1);
204 
205     /* This is the length if adding a new generic error data entry*/
206     data_length = ACPI_GHES_DATA_LENGTH + ACPI_GHES_MEM_CPER_LENGTH;
207     /*
208      * It should not run out of the preallocated memory if adding a new generic
209      * error data entry
210      */
211     assert((data_length + ACPI_GHES_GESB_SIZE) <=
212             ACPI_GHES_MAX_RAW_DATA_LENGTH);
213 
214     /* Build the new generic error status block header */
215     acpi_ghes_generic_error_status(block, ACPI_GEBS_UNCORRECTABLE,
216         0, 0, data_length, ACPI_CPER_SEV_RECOVERABLE);
217 
218     /* Build this new generic error data entry header */
219     acpi_ghes_generic_error_data(block, uefi_cper_mem_sec,
220         ACPI_CPER_SEV_RECOVERABLE, 0, 0,
221         ACPI_GHES_MEM_CPER_LENGTH, fru_id, 0);
222 
223     /* Build the memory section CPER for above new generic error data entry */
224     acpi_ghes_build_append_mem_cper(block, error_physical_addr);
225 
226     /* Write the generic error data entry into guest memory */
227     cpu_physical_memory_write(error_block_address, block->data, block->len);
228 
229     g_array_free(block, true);
230 
231     return 0;
232 }
233 
234 /*
235  * Build table for the hardware error fw_cfg blob.
236  * Initialize "etc/hardware_errors" and "etc/hardware_errors_addr" fw_cfg blobs.
237  * See docs/specs/acpi_hest_ghes.rst for blobs format.
238  */
239 void build_ghes_error_table(GArray *hardware_errors, BIOSLinker *linker)
240 {
241     int i, error_status_block_offset;
242 
243     /* Build error_block_address */
244     for (i = 0; i < ACPI_GHES_ERROR_SOURCE_COUNT; i++) {
245         build_append_int_noprefix(hardware_errors, 0, sizeof(uint64_t));
246     }
247 
248     /* Build read_ack_register */
249     for (i = 0; i < ACPI_GHES_ERROR_SOURCE_COUNT; i++) {
250         /*
251          * Initialize the value of read_ack_register to 1, so GHES can be
252          * writable after (re)boot.
253          * ACPI 6.2: 18.3.2.8 Generic Hardware Error Source version 2
254          * (GHESv2 - Type 10)
255          */
256         build_append_int_noprefix(hardware_errors, 1, sizeof(uint64_t));
257     }
258 
259     /* Generic Error Status Block offset in the hardware error fw_cfg blob */
260     error_status_block_offset = hardware_errors->len;
261 
262     /* Reserve space for Error Status Data Block */
263     acpi_data_push(hardware_errors,
264         ACPI_GHES_MAX_RAW_DATA_LENGTH * ACPI_GHES_ERROR_SOURCE_COUNT);
265 
266     /* Tell guest firmware to place hardware_errors blob into RAM */
267     bios_linker_loader_alloc(linker, ACPI_GHES_ERRORS_FW_CFG_FILE,
268                              hardware_errors, sizeof(uint64_t), false);
269 
270     for (i = 0; i < ACPI_GHES_ERROR_SOURCE_COUNT; i++) {
271         /*
272          * Tell firmware to patch error_block_address entries to point to
273          * corresponding "Generic Error Status Block"
274          */
275         bios_linker_loader_add_pointer(linker,
276             ACPI_GHES_ERRORS_FW_CFG_FILE, sizeof(uint64_t) * i,
277             sizeof(uint64_t), ACPI_GHES_ERRORS_FW_CFG_FILE,
278             error_status_block_offset + i * ACPI_GHES_MAX_RAW_DATA_LENGTH);
279     }
280 
281     /*
282      * tell firmware to write hardware_errors GPA into
283      * hardware_errors_addr fw_cfg, once the former has been initialized.
284      */
285     bios_linker_loader_write_pointer(linker, ACPI_GHES_DATA_ADDR_FW_CFG_FILE,
286         0, sizeof(uint64_t), ACPI_GHES_ERRORS_FW_CFG_FILE, 0);
287 }
288 
289 /* Build Generic Hardware Error Source version 2 (GHESv2) */
290 static void build_ghes_v2(GArray *table_data, int source_id, BIOSLinker *linker)
291 {
292     uint64_t address_offset;
293     /*
294      * Type:
295      * Generic Hardware Error Source version 2(GHESv2 - Type 10)
296      */
297     build_append_int_noprefix(table_data, ACPI_GHES_SOURCE_GENERIC_ERROR_V2, 2);
298     /* Source Id */
299     build_append_int_noprefix(table_data, source_id, 2);
300     /* Related Source Id */
301     build_append_int_noprefix(table_data, 0xffff, 2);
302     /* Flags */
303     build_append_int_noprefix(table_data, 0, 1);
304     /* Enabled */
305     build_append_int_noprefix(table_data, 1, 1);
306 
307     /* Number of Records To Pre-allocate */
308     build_append_int_noprefix(table_data, 1, 4);
309     /* Max Sections Per Record */
310     build_append_int_noprefix(table_data, 1, 4);
311     /* Max Raw Data Length */
312     build_append_int_noprefix(table_data, ACPI_GHES_MAX_RAW_DATA_LENGTH, 4);
313 
314     address_offset = table_data->len;
315     /* Error Status Address */
316     build_append_gas(table_data, AML_AS_SYSTEM_MEMORY, 0x40, 0,
317                      4 /* QWord access */, 0);
318     bios_linker_loader_add_pointer(linker, ACPI_BUILD_TABLE_FILE,
319         address_offset + GAS_ADDR_OFFSET, sizeof(uint64_t),
320         ACPI_GHES_ERRORS_FW_CFG_FILE, source_id * sizeof(uint64_t));
321 
322     switch (source_id) {
323     case ACPI_HEST_SRC_ID_SEA:
324         /*
325          * Notification Structure
326          * Now only enable ARMv8 SEA notification type
327          */
328         build_ghes_hw_error_notification(table_data, ACPI_GHES_NOTIFY_SEA);
329         break;
330     default:
331         error_report("Not support this error source");
332         abort();
333     }
334 
335     /* Error Status Block Length */
336     build_append_int_noprefix(table_data, ACPI_GHES_MAX_RAW_DATA_LENGTH, 4);
337 
338     /*
339      * Read Ack Register
340      * ACPI 6.1: 18.3.2.8 Generic Hardware Error Source
341      * version 2 (GHESv2 - Type 10)
342      */
343     address_offset = table_data->len;
344     build_append_gas(table_data, AML_AS_SYSTEM_MEMORY, 0x40, 0,
345                      4 /* QWord access */, 0);
346     bios_linker_loader_add_pointer(linker, ACPI_BUILD_TABLE_FILE,
347         address_offset + GAS_ADDR_OFFSET,
348         sizeof(uint64_t), ACPI_GHES_ERRORS_FW_CFG_FILE,
349         (ACPI_GHES_ERROR_SOURCE_COUNT + source_id) * sizeof(uint64_t));
350 
351     /*
352      * Read Ack Preserve field
353      * We only provide the first bit in Read Ack Register to OSPM to write
354      * while the other bits are preserved.
355      */
356     build_append_int_noprefix(table_data, ~0x1ULL, 8);
357     /* Read Ack Write */
358     build_append_int_noprefix(table_data, 0x1, 8);
359 }
360 
361 /* Build Hardware Error Source Table */
362 void acpi_build_hest(GArray *table_data, BIOSLinker *linker,
363                      const char *oem_id, const char *oem_table_id)
364 {
365     AcpiTable table = { .sig = "HEST", .rev = 1,
366                         .oem_id = oem_id, .oem_table_id = oem_table_id };
367 
368     acpi_table_begin(&table, table_data);
369 
370     /* Error Source Count */
371     build_append_int_noprefix(table_data, ACPI_GHES_ERROR_SOURCE_COUNT, 4);
372     build_ghes_v2(table_data, ACPI_HEST_SRC_ID_SEA, linker);
373 
374     acpi_table_end(linker, &table);
375 }
376 
377 void acpi_ghes_add_fw_cfg(AcpiGhesState *ags, FWCfgState *s,
378                           GArray *hardware_error)
379 {
380     /* Create a read-only fw_cfg file for GHES */
381     fw_cfg_add_file(s, ACPI_GHES_ERRORS_FW_CFG_FILE, hardware_error->data,
382                     hardware_error->len);
383 
384     /* Create a read-write fw_cfg file for Address */
385     fw_cfg_add_file_callback(s, ACPI_GHES_DATA_ADDR_FW_CFG_FILE, NULL, NULL,
386         NULL, &(ags->ghes_addr_le), sizeof(ags->ghes_addr_le), false);
387 
388     ags->present = true;
389 }
390 
391 int acpi_ghes_record_errors(uint8_t source_id, uint64_t physical_address)
392 {
393     uint64_t error_block_addr, read_ack_register_addr, read_ack_register = 0;
394     uint64_t start_addr;
395     bool ret = -1;
396     AcpiGedState *acpi_ged_state;
397     AcpiGhesState *ags;
398 
399     assert(source_id < ACPI_HEST_SRC_ID_RESERVED);
400 
401     acpi_ged_state = ACPI_GED(object_resolve_path_type("", TYPE_ACPI_GED,
402                                                        NULL));
403     g_assert(acpi_ged_state);
404     ags = &acpi_ged_state->ghes_state;
405 
406     start_addr = le64_to_cpu(ags->ghes_addr_le);
407 
408     if (physical_address) {
409 
410         if (source_id < ACPI_HEST_SRC_ID_RESERVED) {
411             start_addr += source_id * sizeof(uint64_t);
412         }
413 
414         cpu_physical_memory_read(start_addr, &error_block_addr,
415                                  sizeof(error_block_addr));
416 
417         error_block_addr = le64_to_cpu(error_block_addr);
418 
419         read_ack_register_addr = start_addr +
420             ACPI_GHES_ERROR_SOURCE_COUNT * sizeof(uint64_t);
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_report("OSPM does not acknowledge previous error,"
428                 " so can not record CPER for current error anymore");
429         } else if (error_block_addr) {
430             read_ack_register = cpu_to_le64(0);
431             /*
432              * Clear the Read Ack Register, OSPM will write it to 1 when
433              * it acknowledges this error.
434              */
435             cpu_physical_memory_write(read_ack_register_addr,
436                 &read_ack_register, sizeof(uint64_t));
437 
438             ret = acpi_ghes_record_mem_error(error_block_addr,
439                                              physical_address);
440         } else
441             error_report("can not find Generic Error Status Block");
442     }
443 
444     return ret;
445 }
446 
447 bool acpi_ghes_present(void)
448 {
449     AcpiGedState *acpi_ged_state;
450     AcpiGhesState *ags;
451 
452     acpi_ged_state = ACPI_GED(object_resolve_path_type("", TYPE_ACPI_GED,
453                                                        NULL));
454 
455     if (!acpi_ged_state) {
456         return false;
457     }
458     ags = &acpi_ged_state->ghes_state;
459     return ags->present;
460 }
461