1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * UEFI Common Platform Error Record (CPER) support
4 *
5 * Copyright (C) 2017, The Linux Foundation. All rights reserved.
6 */
7
8 #include <linux/kernel.h>
9 #include <linux/module.h>
10 #include <linux/time.h>
11 #include <linux/cper.h>
12 #include <linux/dmi.h>
13 #include <linux/acpi.h>
14 #include <linux/pci.h>
15 #include <linux/aer.h>
16 #include <linux/printk.h>
17 #include <linux/bcd.h>
18 #include <acpi/ghes.h>
19 #include <ras/ras_event.h>
20
21 static const char * const arm_reg_ctx_strs[] = {
22 "AArch32 general purpose registers",
23 "AArch32 EL1 context registers",
24 "AArch32 EL2 context registers",
25 "AArch32 secure context registers",
26 "AArch64 general purpose registers",
27 "AArch64 EL1 context registers",
28 "AArch64 EL2 context registers",
29 "AArch64 EL3 context registers",
30 "Misc. system register structure",
31 };
32
33 static const char * const arm_err_trans_type_strs[] = {
34 "Instruction",
35 "Data Access",
36 "Generic",
37 };
38
39 static const char * const arm_bus_err_op_strs[] = {
40 "Generic error (type cannot be determined)",
41 "Generic read (type of instruction or data request cannot be determined)",
42 "Generic write (type of instruction of data request cannot be determined)",
43 "Data read",
44 "Data write",
45 "Instruction fetch",
46 "Prefetch",
47 };
48
49 static const char * const arm_cache_err_op_strs[] = {
50 "Generic error (type cannot be determined)",
51 "Generic read (type of instruction or data request cannot be determined)",
52 "Generic write (type of instruction of data request cannot be determined)",
53 "Data read",
54 "Data write",
55 "Instruction fetch",
56 "Prefetch",
57 "Eviction",
58 "Snooping (processor initiated a cache snoop that resulted in an error)",
59 "Snooped (processor raised a cache error caused by another processor or device snooping its cache)",
60 "Management",
61 };
62
63 static const char * const arm_tlb_err_op_strs[] = {
64 "Generic error (type cannot be determined)",
65 "Generic read (type of instruction or data request cannot be determined)",
66 "Generic write (type of instruction of data request cannot be determined)",
67 "Data read",
68 "Data write",
69 "Instruction fetch",
70 "Prefetch",
71 "Local management operation (processor initiated a TLB management operation that resulted in an error)",
72 "External management operation (processor raised a TLB error caused by another processor or device broadcasting TLB operations)",
73 };
74
75 static const char * const arm_bus_err_part_type_strs[] = {
76 "Local processor originated request",
77 "Local processor responded to request",
78 "Local processor observed",
79 "Generic",
80 };
81
82 static const char * const arm_bus_err_addr_space_strs[] = {
83 "External Memory Access",
84 "Internal Memory Access",
85 "Unknown",
86 "Device Memory Access",
87 };
88
cper_print_arm_err_info(const char * pfx,u32 type,u64 error_info)89 static void cper_print_arm_err_info(const char *pfx, u32 type,
90 u64 error_info)
91 {
92 u8 trans_type, op_type, level, participation_type, address_space;
93 u16 mem_attributes;
94 bool proc_context_corrupt, corrected, precise_pc, restartable_pc;
95 bool time_out, access_mode;
96
97 /* If the type is unknown, bail. */
98 if (type > CPER_ARM_MAX_TYPE)
99 return;
100
101 /*
102 * Vendor type errors have error information values that are vendor
103 * specific.
104 */
105 if (type == CPER_ARM_VENDOR_ERROR)
106 return;
107
108 if (error_info & CPER_ARM_ERR_VALID_TRANSACTION_TYPE) {
109 trans_type = ((error_info >> CPER_ARM_ERR_TRANSACTION_SHIFT)
110 & CPER_ARM_ERR_TRANSACTION_MASK);
111 if (trans_type < ARRAY_SIZE(arm_err_trans_type_strs)) {
112 printk("%stransaction type: %s\n", pfx,
113 arm_err_trans_type_strs[trans_type]);
114 }
115 }
116
117 if (error_info & CPER_ARM_ERR_VALID_OPERATION_TYPE) {
118 op_type = ((error_info >> CPER_ARM_ERR_OPERATION_SHIFT)
119 & CPER_ARM_ERR_OPERATION_MASK);
120 switch (type) {
121 case CPER_ARM_CACHE_ERROR:
122 if (op_type < ARRAY_SIZE(arm_cache_err_op_strs)) {
123 printk("%soperation type: %s\n", pfx,
124 arm_cache_err_op_strs[op_type]);
125 }
126 break;
127 case CPER_ARM_TLB_ERROR:
128 if (op_type < ARRAY_SIZE(arm_tlb_err_op_strs)) {
129 printk("%soperation type: %s\n", pfx,
130 arm_tlb_err_op_strs[op_type]);
131 }
132 break;
133 case CPER_ARM_BUS_ERROR:
134 if (op_type < ARRAY_SIZE(arm_bus_err_op_strs)) {
135 printk("%soperation type: %s\n", pfx,
136 arm_bus_err_op_strs[op_type]);
137 }
138 break;
139 }
140 }
141
142 if (error_info & CPER_ARM_ERR_VALID_LEVEL) {
143 level = ((error_info >> CPER_ARM_ERR_LEVEL_SHIFT)
144 & CPER_ARM_ERR_LEVEL_MASK);
145 switch (type) {
146 case CPER_ARM_CACHE_ERROR:
147 printk("%scache level: %d\n", pfx, level);
148 break;
149 case CPER_ARM_TLB_ERROR:
150 printk("%sTLB level: %d\n", pfx, level);
151 break;
152 case CPER_ARM_BUS_ERROR:
153 printk("%saffinity level at which the bus error occurred: %d\n",
154 pfx, level);
155 break;
156 }
157 }
158
159 if (error_info & CPER_ARM_ERR_VALID_PROC_CONTEXT_CORRUPT) {
160 proc_context_corrupt = ((error_info >> CPER_ARM_ERR_PC_CORRUPT_SHIFT)
161 & CPER_ARM_ERR_PC_CORRUPT_MASK);
162 if (proc_context_corrupt)
163 printk("%sprocessor context corrupted\n", pfx);
164 else
165 printk("%sprocessor context not corrupted\n", pfx);
166 }
167
168 if (error_info & CPER_ARM_ERR_VALID_CORRECTED) {
169 corrected = ((error_info >> CPER_ARM_ERR_CORRECTED_SHIFT)
170 & CPER_ARM_ERR_CORRECTED_MASK);
171 if (corrected)
172 printk("%sthe error has been corrected\n", pfx);
173 else
174 printk("%sthe error has not been corrected\n", pfx);
175 }
176
177 if (error_info & CPER_ARM_ERR_VALID_PRECISE_PC) {
178 precise_pc = ((error_info >> CPER_ARM_ERR_PRECISE_PC_SHIFT)
179 & CPER_ARM_ERR_PRECISE_PC_MASK);
180 if (precise_pc)
181 printk("%sPC is precise\n", pfx);
182 else
183 printk("%sPC is imprecise\n", pfx);
184 }
185
186 if (error_info & CPER_ARM_ERR_VALID_RESTARTABLE_PC) {
187 restartable_pc = ((error_info >> CPER_ARM_ERR_RESTARTABLE_PC_SHIFT)
188 & CPER_ARM_ERR_RESTARTABLE_PC_MASK);
189 if (restartable_pc)
190 printk("%sProgram execution can be restarted reliably at the PC associated with the error.\n", pfx);
191 }
192
193 /* The rest of the fields are specific to bus errors */
194 if (type != CPER_ARM_BUS_ERROR)
195 return;
196
197 if (error_info & CPER_ARM_ERR_VALID_PARTICIPATION_TYPE) {
198 participation_type = ((error_info >> CPER_ARM_ERR_PARTICIPATION_TYPE_SHIFT)
199 & CPER_ARM_ERR_PARTICIPATION_TYPE_MASK);
200 if (participation_type < ARRAY_SIZE(arm_bus_err_part_type_strs)) {
201 printk("%sparticipation type: %s\n", pfx,
202 arm_bus_err_part_type_strs[participation_type]);
203 }
204 }
205
206 if (error_info & CPER_ARM_ERR_VALID_TIME_OUT) {
207 time_out = ((error_info >> CPER_ARM_ERR_TIME_OUT_SHIFT)
208 & CPER_ARM_ERR_TIME_OUT_MASK);
209 if (time_out)
210 printk("%srequest timed out\n", pfx);
211 }
212
213 if (error_info & CPER_ARM_ERR_VALID_ADDRESS_SPACE) {
214 address_space = ((error_info >> CPER_ARM_ERR_ADDRESS_SPACE_SHIFT)
215 & CPER_ARM_ERR_ADDRESS_SPACE_MASK);
216 if (address_space < ARRAY_SIZE(arm_bus_err_addr_space_strs)) {
217 printk("%saddress space: %s\n", pfx,
218 arm_bus_err_addr_space_strs[address_space]);
219 }
220 }
221
222 if (error_info & CPER_ARM_ERR_VALID_MEM_ATTRIBUTES) {
223 mem_attributes = ((error_info >> CPER_ARM_ERR_MEM_ATTRIBUTES_SHIFT)
224 & CPER_ARM_ERR_MEM_ATTRIBUTES_MASK);
225 printk("%smemory access attributes:0x%x\n", pfx, mem_attributes);
226 }
227
228 if (error_info & CPER_ARM_ERR_VALID_ACCESS_MODE) {
229 access_mode = ((error_info >> CPER_ARM_ERR_ACCESS_MODE_SHIFT)
230 & CPER_ARM_ERR_ACCESS_MODE_MASK);
231 if (access_mode)
232 printk("%saccess mode: normal\n", pfx);
233 else
234 printk("%saccess mode: secure\n", pfx);
235 }
236 }
237
cper_print_proc_arm(const char * pfx,const struct cper_sec_proc_arm * proc)238 void cper_print_proc_arm(const char *pfx,
239 const struct cper_sec_proc_arm *proc)
240 {
241 int i, len, max_ctx_type;
242 struct cper_arm_err_info *err_info;
243 struct cper_arm_ctx_info *ctx_info;
244 char newpfx[64], infopfx[64];
245
246 printk("%sMIDR: 0x%016llx\n", pfx, proc->midr);
247
248 len = proc->section_length - (sizeof(*proc) +
249 proc->err_info_num * (sizeof(*err_info)));
250 if (len < 0) {
251 printk("%ssection length: %d\n", pfx, proc->section_length);
252 printk("%ssection length is too small\n", pfx);
253 printk("%sfirmware-generated error record is incorrect\n", pfx);
254 printk("%sERR_INFO_NUM is %d\n", pfx, proc->err_info_num);
255 return;
256 }
257
258 if (proc->validation_bits & CPER_ARM_VALID_MPIDR)
259 printk("%sMultiprocessor Affinity Register (MPIDR): 0x%016llx\n",
260 pfx, proc->mpidr);
261
262 if (proc->validation_bits & CPER_ARM_VALID_AFFINITY_LEVEL)
263 printk("%serror affinity level: %d\n", pfx,
264 proc->affinity_level);
265
266 if (proc->validation_bits & CPER_ARM_VALID_RUNNING_STATE) {
267 printk("%srunning state: 0x%x\n", pfx, proc->running_state);
268 printk("%sPower State Coordination Interface state: %d\n",
269 pfx, proc->psci_state);
270 }
271
272 snprintf(newpfx, sizeof(newpfx), "%s ", pfx);
273
274 err_info = (struct cper_arm_err_info *)(proc + 1);
275 for (i = 0; i < proc->err_info_num; i++) {
276 printk("%sError info structure %d:\n", pfx, i);
277
278 printk("%snum errors: %d\n", pfx, err_info->multiple_error + 1);
279
280 if (err_info->validation_bits & CPER_ARM_INFO_VALID_FLAGS) {
281 if (err_info->flags & CPER_ARM_INFO_FLAGS_FIRST)
282 printk("%sfirst error captured\n", newpfx);
283 if (err_info->flags & CPER_ARM_INFO_FLAGS_LAST)
284 printk("%slast error captured\n", newpfx);
285 if (err_info->flags & CPER_ARM_INFO_FLAGS_PROPAGATED)
286 printk("%spropagated error captured\n",
287 newpfx);
288 if (err_info->flags & CPER_ARM_INFO_FLAGS_OVERFLOW)
289 printk("%soverflow occurred, error info is incomplete\n",
290 newpfx);
291 }
292
293 printk("%serror_type: %d, %s\n", newpfx, err_info->type,
294 err_info->type < ARRAY_SIZE(cper_proc_error_type_strs) ?
295 cper_proc_error_type_strs[err_info->type] : "unknown");
296 if (err_info->validation_bits & CPER_ARM_INFO_VALID_ERR_INFO) {
297 printk("%serror_info: 0x%016llx\n", newpfx,
298 err_info->error_info);
299 snprintf(infopfx, sizeof(infopfx), "%s ", newpfx);
300 cper_print_arm_err_info(infopfx, err_info->type,
301 err_info->error_info);
302 }
303 if (err_info->validation_bits & CPER_ARM_INFO_VALID_VIRT_ADDR)
304 printk("%svirtual fault address: 0x%016llx\n",
305 newpfx, err_info->virt_fault_addr);
306 if (err_info->validation_bits & CPER_ARM_INFO_VALID_PHYSICAL_ADDR)
307 printk("%sphysical fault address: 0x%016llx\n",
308 newpfx, err_info->physical_fault_addr);
309 err_info += 1;
310 }
311
312 ctx_info = (struct cper_arm_ctx_info *)err_info;
313 max_ctx_type = ARRAY_SIZE(arm_reg_ctx_strs) - 1;
314 for (i = 0; i < proc->context_info_num; i++) {
315 int size = sizeof(*ctx_info) + ctx_info->size;
316
317 printk("%sContext info structure %d:\n", pfx, i);
318 if (len < size) {
319 printk("%ssection length is too small\n", newpfx);
320 printk("%sfirmware-generated error record is incorrect\n", pfx);
321 return;
322 }
323 if (ctx_info->type > max_ctx_type) {
324 printk("%sInvalid context type: %d (max: %d)\n",
325 newpfx, ctx_info->type, max_ctx_type);
326 return;
327 }
328 printk("%sregister context type: %s\n", newpfx,
329 arm_reg_ctx_strs[ctx_info->type]);
330 print_hex_dump(newpfx, "", DUMP_PREFIX_OFFSET, 16, 4,
331 (ctx_info + 1), ctx_info->size, 0);
332 len -= size;
333 ctx_info = (struct cper_arm_ctx_info *)((long)ctx_info + size);
334 }
335
336 if (len > 0) {
337 printk("%sVendor specific error info has %u bytes:\n", pfx,
338 len);
339 print_hex_dump(newpfx, "", DUMP_PREFIX_OFFSET, 16, 4, ctx_info,
340 len, true);
341 }
342 }
343