1 #undef TRACE_SYSTEM 2 #define TRACE_SYSTEM ras 3 #define TRACE_INCLUDE_FILE ras_event 4 5 #if !defined(_TRACE_HW_EVENT_MC_H) || defined(TRACE_HEADER_MULTI_READ) 6 #define _TRACE_HW_EVENT_MC_H 7 8 #include <linux/tracepoint.h> 9 #include <linux/edac.h> 10 #include <linux/ktime.h> 11 #include <linux/pci.h> 12 #include <linux/aer.h> 13 #include <linux/cper.h> 14 #include <linux/mm.h> 15 16 /* 17 * MCE Extended Error Log trace event 18 * 19 * These events are generated when hardware detects a corrected or 20 * uncorrected event. 21 */ 22 23 /* memory trace event */ 24 25 #if defined(CONFIG_ACPI_EXTLOG) || defined(CONFIG_ACPI_EXTLOG_MODULE) 26 TRACE_EVENT(extlog_mem_event, 27 TP_PROTO(struct cper_sec_mem_err *mem, 28 u32 err_seq, 29 const uuid_le *fru_id, 30 const char *fru_text, 31 u8 sev), 32 33 TP_ARGS(mem, err_seq, fru_id, fru_text, sev), 34 35 TP_STRUCT__entry( 36 __field(u32, err_seq) 37 __field(u8, etype) 38 __field(u8, sev) 39 __field(u64, pa) 40 __field(u8, pa_mask_lsb) 41 __field_struct(uuid_le, fru_id) 42 __string(fru_text, fru_text) 43 __field_struct(struct cper_mem_err_compact, data) 44 ), 45 46 TP_fast_assign( 47 __entry->err_seq = err_seq; 48 if (mem->validation_bits & CPER_MEM_VALID_ERROR_TYPE) 49 __entry->etype = mem->error_type; 50 else 51 __entry->etype = ~0; 52 __entry->sev = sev; 53 if (mem->validation_bits & CPER_MEM_VALID_PA) 54 __entry->pa = mem->physical_addr; 55 else 56 __entry->pa = ~0ull; 57 58 if (mem->validation_bits & CPER_MEM_VALID_PA_MASK) 59 __entry->pa_mask_lsb = (u8)__ffs64(mem->physical_addr_mask); 60 else 61 __entry->pa_mask_lsb = ~0; 62 __entry->fru_id = *fru_id; 63 __assign_str(fru_text, fru_text); 64 cper_mem_err_pack(mem, &__entry->data); 65 ), 66 67 TP_printk("{%d} %s error: %s physical addr: %016llx (mask lsb: %x) %sFRU: %pUl %.20s", 68 __entry->err_seq, 69 cper_severity_str(__entry->sev), 70 cper_mem_err_type_str(__entry->etype), 71 __entry->pa, 72 __entry->pa_mask_lsb, 73 cper_mem_err_unpack(p, &__entry->data), 74 &__entry->fru_id, 75 __get_str(fru_text)) 76 ); 77 #endif 78 79 /* 80 * Hardware Events Report 81 * 82 * Those events are generated when hardware detected a corrected or 83 * uncorrected event, and are meant to replace the current API to report 84 * errors defined on both EDAC and MCE subsystems. 85 * 86 * FIXME: Add events for handling memory errors originated from the 87 * MCE subsystem. 88 */ 89 90 /* 91 * Hardware-independent Memory Controller specific events 92 */ 93 94 /* 95 * Default error mechanisms for Memory Controller errors (CE and UE) 96 */ 97 TRACE_EVENT(mc_event, 98 99 TP_PROTO(const unsigned int err_type, 100 const char *error_msg, 101 const char *label, 102 const int error_count, 103 const u8 mc_index, 104 const s8 top_layer, 105 const s8 mid_layer, 106 const s8 low_layer, 107 unsigned long address, 108 const u8 grain_bits, 109 unsigned long syndrome, 110 const char *driver_detail), 111 112 TP_ARGS(err_type, error_msg, label, error_count, mc_index, 113 top_layer, mid_layer, low_layer, address, grain_bits, 114 syndrome, driver_detail), 115 116 TP_STRUCT__entry( 117 __field( unsigned int, error_type ) 118 __string( msg, error_msg ) 119 __string( label, label ) 120 __field( u16, error_count ) 121 __field( u8, mc_index ) 122 __field( s8, top_layer ) 123 __field( s8, middle_layer ) 124 __field( s8, lower_layer ) 125 __field( long, address ) 126 __field( u8, grain_bits ) 127 __field( long, syndrome ) 128 __string( driver_detail, driver_detail ) 129 ), 130 131 TP_fast_assign( 132 __entry->error_type = err_type; 133 __assign_str(msg, error_msg); 134 __assign_str(label, label); 135 __entry->error_count = error_count; 136 __entry->mc_index = mc_index; 137 __entry->top_layer = top_layer; 138 __entry->middle_layer = mid_layer; 139 __entry->lower_layer = low_layer; 140 __entry->address = address; 141 __entry->grain_bits = grain_bits; 142 __entry->syndrome = syndrome; 143 __assign_str(driver_detail, driver_detail); 144 ), 145 146 TP_printk("%d %s error%s:%s%s on %s (mc:%d location:%d:%d:%d address:0x%08lx grain:%d syndrome:0x%08lx%s%s)", 147 __entry->error_count, 148 mc_event_error_type(__entry->error_type), 149 __entry->error_count > 1 ? "s" : "", 150 __get_str(msg)[0] ? " " : "", 151 __get_str(msg), 152 __get_str(label), 153 __entry->mc_index, 154 __entry->top_layer, 155 __entry->middle_layer, 156 __entry->lower_layer, 157 __entry->address, 158 1 << __entry->grain_bits, 159 __entry->syndrome, 160 __get_str(driver_detail)[0] ? " " : "", 161 __get_str(driver_detail)) 162 ); 163 164 /* 165 * ARM Processor Events Report 166 * 167 * This event is generated when hardware detects an ARM processor error 168 * has occurred. UEFI 2.6 spec section N.2.4.4. 169 */ 170 TRACE_EVENT(arm_event, 171 172 TP_PROTO(const struct cper_sec_proc_arm *proc), 173 174 TP_ARGS(proc), 175 176 TP_STRUCT__entry( 177 __field(u64, mpidr) 178 __field(u64, midr) 179 __field(u32, running_state) 180 __field(u32, psci_state) 181 __field(u8, affinity) 182 ), 183 184 TP_fast_assign( 185 if (proc->validation_bits & CPER_ARM_VALID_AFFINITY_LEVEL) 186 __entry->affinity = proc->affinity_level; 187 else 188 __entry->affinity = ~0; 189 if (proc->validation_bits & CPER_ARM_VALID_MPIDR) 190 __entry->mpidr = proc->mpidr; 191 else 192 __entry->mpidr = 0ULL; 193 __entry->midr = proc->midr; 194 if (proc->validation_bits & CPER_ARM_VALID_RUNNING_STATE) { 195 __entry->running_state = proc->running_state; 196 __entry->psci_state = proc->psci_state; 197 } else { 198 __entry->running_state = ~0; 199 __entry->psci_state = ~0; 200 } 201 ), 202 203 TP_printk("affinity level: %d; MPIDR: %016llx; MIDR: %016llx; " 204 "running state: %d; PSCI state: %d", 205 __entry->affinity, __entry->mpidr, __entry->midr, 206 __entry->running_state, __entry->psci_state) 207 ); 208 209 /* 210 * Non-Standard Section Report 211 * 212 * This event is generated when hardware detected a hardware 213 * error event, which may be of non-standard section as defined 214 * in UEFI spec appendix "Common Platform Error Record", or may 215 * be of sections for which TRACE_EVENT is not defined. 216 * 217 */ 218 TRACE_EVENT(non_standard_event, 219 220 TP_PROTO(const uuid_le *sec_type, 221 const uuid_le *fru_id, 222 const char *fru_text, 223 const u8 sev, 224 const u8 *err, 225 const u32 len), 226 227 TP_ARGS(sec_type, fru_id, fru_text, sev, err, len), 228 229 TP_STRUCT__entry( 230 __array(char, sec_type, UUID_SIZE) 231 __array(char, fru_id, UUID_SIZE) 232 __string(fru_text, fru_text) 233 __field(u8, sev) 234 __field(u32, len) 235 __dynamic_array(u8, buf, len) 236 ), 237 238 TP_fast_assign( 239 memcpy(__entry->sec_type, sec_type, UUID_SIZE); 240 memcpy(__entry->fru_id, fru_id, UUID_SIZE); 241 __assign_str(fru_text, fru_text); 242 __entry->sev = sev; 243 __entry->len = len; 244 memcpy(__get_dynamic_array(buf), err, len); 245 ), 246 247 TP_printk("severity: %d; sec type:%pU; FRU: %pU %s; data len:%d; raw data:%s", 248 __entry->sev, __entry->sec_type, 249 __entry->fru_id, __get_str(fru_text), 250 __entry->len, 251 __print_hex(__get_dynamic_array(buf), __entry->len)) 252 ); 253 254 /* 255 * PCIe AER Trace event 256 * 257 * These events are generated when hardware detects a corrected or 258 * uncorrected event on a PCIe device. The event report has 259 * the following structure: 260 * 261 * char * dev_name - The name of the slot where the device resides 262 * ([domain:]bus:device.function). 263 * u32 status - Either the correctable or uncorrectable register 264 * indicating what error or errors have been seen 265 * u8 severity - error severity 0:NONFATAL 1:FATAL 2:CORRECTED 266 */ 267 268 #define aer_correctable_errors \ 269 {PCI_ERR_COR_RCVR, "Receiver Error"}, \ 270 {PCI_ERR_COR_BAD_TLP, "Bad TLP"}, \ 271 {PCI_ERR_COR_BAD_DLLP, "Bad DLLP"}, \ 272 {PCI_ERR_COR_REP_ROLL, "RELAY_NUM Rollover"}, \ 273 {PCI_ERR_COR_REP_TIMER, "Replay Timer Timeout"}, \ 274 {PCI_ERR_COR_ADV_NFAT, "Advisory Non-Fatal Error"}, \ 275 {PCI_ERR_COR_INTERNAL, "Corrected Internal Error"}, \ 276 {PCI_ERR_COR_LOG_OVER, "Header Log Overflow"} 277 278 #define aer_uncorrectable_errors \ 279 {PCI_ERR_UNC_UND, "Undefined"}, \ 280 {PCI_ERR_UNC_DLP, "Data Link Protocol Error"}, \ 281 {PCI_ERR_UNC_SURPDN, "Surprise Down Error"}, \ 282 {PCI_ERR_UNC_POISON_TLP,"Poisoned TLP"}, \ 283 {PCI_ERR_UNC_FCP, "Flow Control Protocol Error"}, \ 284 {PCI_ERR_UNC_COMP_TIME, "Completion Timeout"}, \ 285 {PCI_ERR_UNC_COMP_ABORT,"Completer Abort"}, \ 286 {PCI_ERR_UNC_UNX_COMP, "Unexpected Completion"}, \ 287 {PCI_ERR_UNC_RX_OVER, "Receiver Overflow"}, \ 288 {PCI_ERR_UNC_MALF_TLP, "Malformed TLP"}, \ 289 {PCI_ERR_UNC_ECRC, "ECRC Error"}, \ 290 {PCI_ERR_UNC_UNSUP, "Unsupported Request Error"}, \ 291 {PCI_ERR_UNC_ACSV, "ACS Violation"}, \ 292 {PCI_ERR_UNC_INTN, "Uncorrectable Internal Error"},\ 293 {PCI_ERR_UNC_MCBTLP, "MC Blocked TLP"}, \ 294 {PCI_ERR_UNC_ATOMEG, "AtomicOp Egress Blocked"}, \ 295 {PCI_ERR_UNC_TLPPRE, "TLP Prefix Blocked Error"} 296 297 TRACE_EVENT(aer_event, 298 TP_PROTO(const char *dev_name, 299 const u32 status, 300 const u8 severity), 301 302 TP_ARGS(dev_name, status, severity), 303 304 TP_STRUCT__entry( 305 __string( dev_name, dev_name ) 306 __field( u32, status ) 307 __field( u8, severity ) 308 ), 309 310 TP_fast_assign( 311 __assign_str(dev_name, dev_name); 312 __entry->status = status; 313 __entry->severity = severity; 314 ), 315 316 TP_printk("%s PCIe Bus Error: severity=%s, %s\n", 317 __get_str(dev_name), 318 __entry->severity == AER_CORRECTABLE ? "Corrected" : 319 __entry->severity == AER_FATAL ? 320 "Fatal" : "Uncorrected, non-fatal", 321 __entry->severity == AER_CORRECTABLE ? 322 __print_flags(__entry->status, "|", aer_correctable_errors) : 323 __print_flags(__entry->status, "|", aer_uncorrectable_errors)) 324 ); 325 326 /* 327 * memory-failure recovery action result event 328 * 329 * unsigned long pfn - Page Frame Number of the corrupted page 330 * int type - Page types of the corrupted page 331 * int result - Result of recovery action 332 */ 333 334 #ifdef CONFIG_MEMORY_FAILURE 335 #define MF_ACTION_RESULT \ 336 EM ( MF_IGNORED, "Ignored" ) \ 337 EM ( MF_FAILED, "Failed" ) \ 338 EM ( MF_DELAYED, "Delayed" ) \ 339 EMe ( MF_RECOVERED, "Recovered" ) 340 341 #define MF_PAGE_TYPE \ 342 EM ( MF_MSG_KERNEL, "reserved kernel page" ) \ 343 EM ( MF_MSG_KERNEL_HIGH_ORDER, "high-order kernel page" ) \ 344 EM ( MF_MSG_SLAB, "kernel slab page" ) \ 345 EM ( MF_MSG_DIFFERENT_COMPOUND, "different compound page after locking" ) \ 346 EM ( MF_MSG_POISONED_HUGE, "huge page already hardware poisoned" ) \ 347 EM ( MF_MSG_HUGE, "huge page" ) \ 348 EM ( MF_MSG_FREE_HUGE, "free huge page" ) \ 349 EM ( MF_MSG_UNMAP_FAILED, "unmapping failed page" ) \ 350 EM ( MF_MSG_DIRTY_SWAPCACHE, "dirty swapcache page" ) \ 351 EM ( MF_MSG_CLEAN_SWAPCACHE, "clean swapcache page" ) \ 352 EM ( MF_MSG_DIRTY_MLOCKED_LRU, "dirty mlocked LRU page" ) \ 353 EM ( MF_MSG_CLEAN_MLOCKED_LRU, "clean mlocked LRU page" ) \ 354 EM ( MF_MSG_DIRTY_UNEVICTABLE_LRU, "dirty unevictable LRU page" ) \ 355 EM ( MF_MSG_CLEAN_UNEVICTABLE_LRU, "clean unevictable LRU page" ) \ 356 EM ( MF_MSG_DIRTY_LRU, "dirty LRU page" ) \ 357 EM ( MF_MSG_CLEAN_LRU, "clean LRU page" ) \ 358 EM ( MF_MSG_TRUNCATED_LRU, "already truncated LRU page" ) \ 359 EM ( MF_MSG_BUDDY, "free buddy page" ) \ 360 EM ( MF_MSG_BUDDY_2ND, "free buddy page (2nd try)" ) \ 361 EMe ( MF_MSG_UNKNOWN, "unknown page" ) 362 363 /* 364 * First define the enums in MM_ACTION_RESULT to be exported to userspace 365 * via TRACE_DEFINE_ENUM(). 366 */ 367 #undef EM 368 #undef EMe 369 #define EM(a, b) TRACE_DEFINE_ENUM(a); 370 #define EMe(a, b) TRACE_DEFINE_ENUM(a); 371 372 MF_ACTION_RESULT 373 MF_PAGE_TYPE 374 375 /* 376 * Now redefine the EM() and EMe() macros to map the enums to the strings 377 * that will be printed in the output. 378 */ 379 #undef EM 380 #undef EMe 381 #define EM(a, b) { a, b }, 382 #define EMe(a, b) { a, b } 383 384 TRACE_EVENT(memory_failure_event, 385 TP_PROTO(unsigned long pfn, 386 int type, 387 int result), 388 389 TP_ARGS(pfn, type, result), 390 391 TP_STRUCT__entry( 392 __field(unsigned long, pfn) 393 __field(int, type) 394 __field(int, result) 395 ), 396 397 TP_fast_assign( 398 __entry->pfn = pfn; 399 __entry->type = type; 400 __entry->result = result; 401 ), 402 403 TP_printk("pfn %#lx: recovery action for %s: %s", 404 __entry->pfn, 405 __print_symbolic(__entry->type, MF_PAGE_TYPE), 406 __print_symbolic(__entry->result, MF_ACTION_RESULT) 407 ) 408 ); 409 #endif /* CONFIG_MEMORY_FAILURE */ 410 #endif /* _TRACE_HW_EVENT_MC_H */ 411 412 /* This part must be outside protection */ 413 #include <trace/define_trace.h> 414