1 perf.data format 1 perf.data format
2 2
3 Uptodate as of v4.7 3 Uptodate as of v4.7
4 4
5 This document describes the on-disk perf.data 5 This document describes the on-disk perf.data format, generated by perf record
6 or perf inject and consumed by the other perf 6 or perf inject and consumed by the other perf tools.
7 7
8 On a high level perf.data contains the events 8 On a high level perf.data contains the events generated by the PMUs, plus metadata.
9 9
10 All fields are in native-endian of the machine 10 All fields are in native-endian of the machine that generated the perf.data.
11 11
12 When perf is writing to a pipe it uses a speci 12 When perf is writing to a pipe it uses a special version of the file
13 format that does not rely on seeking to adjust 13 format that does not rely on seeking to adjust data offsets. This
14 format is described in "Pipe-mode data" sectio 14 format is described in "Pipe-mode data" section. The pipe data version can be
15 augmented with additional events using perf in 15 augmented with additional events using perf inject.
16 16
17 The file starts with a perf_header: 17 The file starts with a perf_header:
18 18
19 struct perf_header { 19 struct perf_header {
20 char magic[8]; /* PERFILE2 */ 20 char magic[8]; /* PERFILE2 */
21 uint64_t size; /* size of the 21 uint64_t size; /* size of the header */
22 uint64_t attr_size; /* size of an 22 uint64_t attr_size; /* size of an attribute in attrs */
23 struct perf_file_section attrs; 23 struct perf_file_section attrs;
24 struct perf_file_section data; 24 struct perf_file_section data;
25 struct perf_file_section event_types; 25 struct perf_file_section event_types;
26 uint64_t flags; 26 uint64_t flags;
27 uint64_t flags1[3]; 27 uint64_t flags1[3];
28 }; 28 };
29 29
30 The magic number identifies the perf file and 30 The magic number identifies the perf file and the version. Current perf versions
31 use PERFILE2. Old perf versions generated a ve 31 use PERFILE2. Old perf versions generated a version 1 format (PERFFILE). Version 1
32 is not described here. The magic number also i 32 is not described here. The magic number also identifies the endian. When the
33 magic value is 64bit byte swapped compared the 33 magic value is 64bit byte swapped compared the file is in non-native
34 endian. 34 endian.
35 35
36 A perf_file_section contains a pointer to anot 36 A perf_file_section contains a pointer to another section of the perf file.
37 The header contains three such pointers: for a 37 The header contains three such pointers: for attributes, data and event types.
38 38
39 struct perf_file_section { 39 struct perf_file_section {
40 uint64_t offset; /* offset from 40 uint64_t offset; /* offset from start of file */
41 uint64_t size; /* size of the 41 uint64_t size; /* size of the section */
42 }; 42 };
43 43
44 Flags section: 44 Flags section:
45 45
46 For each of the optional features a perf_file_ !! 46 The header is followed by different optional headers, described by the bits set
47 section if the feature bit is set in the perf_ !! 47 in flags. Only headers for which the bit is set are included. Each header
48 respective perf_file_section points to the dat !! 48 consists of a perf_file_section located after the initial header.
49 defines its size. !! 49 The respective perf_file_section points to the data of the additional
>> 50 header and defines its size.
50 51
51 Some headers consist of strings, which are def 52 Some headers consist of strings, which are defined like this:
52 53
53 struct perf_header_string { 54 struct perf_header_string {
54 uint32_t len; 55 uint32_t len;
55 char string[len]; /* zero terminated */ 56 char string[len]; /* zero terminated */
56 }; 57 };
57 58
58 Some headers consist of a sequence of strings, 59 Some headers consist of a sequence of strings, which start with a
59 60
60 struct perf_header_string_list { 61 struct perf_header_string_list {
61 uint32_t nr; 62 uint32_t nr;
62 struct perf_header_string strings[nr]; /* 63 struct perf_header_string strings[nr]; /* variable length records */
63 }; 64 };
64 65
65 The bits are the flags bits in a 256 bit bitma 66 The bits are the flags bits in a 256 bit bitmap starting with
66 flags. These define the valid bits: 67 flags. These define the valid bits:
67 68
68 HEADER_RESERVED = 0, /* alw 69 HEADER_RESERVED = 0, /* always cleared */
69 HEADER_FIRST_FEATURE = 1, 70 HEADER_FIRST_FEATURE = 1,
70 HEADER_TRACING_DATA = 1, 71 HEADER_TRACING_DATA = 1,
71 72
72 Describe me. 73 Describe me.
73 74
74 HEADER_BUILD_ID = 2, 75 HEADER_BUILD_ID = 2,
75 76
76 The header consists of an sequence of build_id 77 The header consists of an sequence of build_id_event. The size of each record
77 is defined by header.size (see perf_event.h). 78 is defined by header.size (see perf_event.h). Each event defines a ELF build id
78 for a executable file name for a pid. An ELF b 79 for a executable file name for a pid. An ELF build id is a unique identifier
79 assigned by the linker to an executable. 80 assigned by the linker to an executable.
80 81
81 struct build_id_event { 82 struct build_id_event {
82 struct perf_event_header header; 83 struct perf_event_header header;
83 pid_t pid; 84 pid_t pid;
84 uint8_t build_id[24]; 85 uint8_t build_id[24];
85 char filename[head 86 char filename[header.size - offsetof(struct build_id_event, filename)];
86 }; 87 };
87 88
88 HEADER_HOSTNAME = 3, 89 HEADER_HOSTNAME = 3,
89 90
90 A perf_header_string with the hostname where t 91 A perf_header_string with the hostname where the data was collected
91 (uname -n) 92 (uname -n)
92 93
93 HEADER_OSRELEASE = 4, 94 HEADER_OSRELEASE = 4,
94 95
95 A perf_header_string with the os release where 96 A perf_header_string with the os release where the data was collected
96 (uname -r) 97 (uname -r)
97 98
98 HEADER_VERSION = 5, 99 HEADER_VERSION = 5,
99 100
100 A perf_header_string with the perf user tool v 101 A perf_header_string with the perf user tool version where the
101 data was collected. This is the same as the ve 102 data was collected. This is the same as the version of the source tree
102 the perf tool was built from. 103 the perf tool was built from.
103 104
104 HEADER_ARCH = 6, 105 HEADER_ARCH = 6,
105 106
106 A perf_header_string with the CPU architecture 107 A perf_header_string with the CPU architecture (uname -m)
107 108
108 HEADER_NRCPUS = 7, 109 HEADER_NRCPUS = 7,
109 110
110 A structure defining the number of CPUs. 111 A structure defining the number of CPUs.
111 112
112 struct nr_cpus { 113 struct nr_cpus {
113 uint32_t nr_cpus_available; /* CPUs not 114 uint32_t nr_cpus_available; /* CPUs not yet onlined */
114 uint32_t nr_cpus_online; 115 uint32_t nr_cpus_online;
115 }; 116 };
116 117
117 HEADER_CPUDESC = 8, 118 HEADER_CPUDESC = 8,
118 119
119 A perf_header_string with description of the C 120 A perf_header_string with description of the CPU. On x86 this is the model name
120 in /proc/cpuinfo 121 in /proc/cpuinfo
121 122
122 HEADER_CPUID = 9, 123 HEADER_CPUID = 9,
123 124
124 A perf_header_string with the exact CPU type. 125 A perf_header_string with the exact CPU type. On x86 this is
125 vendor,family,model,stepping. For example: Gen 126 vendor,family,model,stepping. For example: GenuineIntel,6,69,1
126 127
127 HEADER_TOTAL_MEM = 10, 128 HEADER_TOTAL_MEM = 10,
128 129
129 An uint64_t with the total memory in kilobytes !! 130 An uint64_t with the total memory in bytes.
130 131
131 HEADER_CMDLINE = 11, 132 HEADER_CMDLINE = 11,
132 133
133 A perf_header_string_list with the perf arg-ve !! 134 A perf_header_string with the perf command line used to collect the data.
134 135
135 HEADER_EVENT_DESC = 12, 136 HEADER_EVENT_DESC = 12,
136 137
137 Another description of the perf_event_attrs, m 138 Another description of the perf_event_attrs, more detailed than header.attrs
138 including IDs and names. See perf_event.h or t 139 including IDs and names. See perf_event.h or the man page for a description
139 of a struct perf_event_attr. 140 of a struct perf_event_attr.
140 141
141 struct { 142 struct {
142 uint32_t nr; /* number of events */ 143 uint32_t nr; /* number of events */
143 uint32_t attr_size; /* size of each per 144 uint32_t attr_size; /* size of each perf_event_attr */
144 struct { 145 struct {
145 struct perf_event_attr attr; /* 146 struct perf_event_attr attr; /* size of attr_size */
146 uint32_t nr_ids; 147 uint32_t nr_ids;
147 struct perf_header_string event_ 148 struct perf_header_string event_string;
148 uint64_t ids[nr_ids]; 149 uint64_t ids[nr_ids];
149 } events[nr]; /* Variable length record 150 } events[nr]; /* Variable length records */
150 }; 151 };
151 152
152 HEADER_CPU_TOPOLOGY = 13, 153 HEADER_CPU_TOPOLOGY = 13,
153 154
154 struct { !! 155 String lists defining the core and CPU threads topology.
155 /* !! 156 The string lists are followed by a variable length array
156 * First revision of HEADER_CPU_TOPOLO !! 157 which contains core_id and socket_id of each cpu.
157 * !! 158 The number of entries can be determined by the size of the
158 * See 'struct perf_header_string_list !! 159 section minus the sizes of both string lists.
159 * in this file. <<
160 */ <<
161 160
>> 161 struct {
162 struct perf_header_string_list cores; / 162 struct perf_header_string_list cores; /* Variable length */
163 struct perf_header_string_list threads; 163 struct perf_header_string_list threads; /* Variable length */
164 <<
165 /* <<
166 * Second revision of HEADER_CPU_TOPOLO <<
167 * will not consider what comes next <<
168 */ <<
169 <<
170 struct { 164 struct {
171 uint32_t core_id; 165 uint32_t core_id;
172 uint32_t socket_id; 166 uint32_t socket_id;
173 } cpus[nr]; /* Variable length records 167 } cpus[nr]; /* Variable length records */
174 /* 'nr' comes from previously processed <<
175 <<
176 /* <<
177 * Third revision of HEADER_CPU_TOPOLO <<
178 * will not consider what comes next <<
179 */ <<
180 <<
181 struct perf_header_string_list dies; / <<
182 uint32_t die_id[nr_cpus_avail]; /* fro <<
183 }; 168 };
184 169
185 Example: 170 Example:
186 sibling sockets : 0-8 !! 171 sibling cores : 0-3
187 sibling dies : 0-3 <<
188 sibling dies : 4-7 <<
189 sibling threads : 0-1 172 sibling threads : 0-1
190 sibling threads : 2-3 173 sibling threads : 2-3
191 sibling threads : 4-5 <<
192 sibling threads : 6-7 <<
193 174
194 HEADER_NUMA_TOPOLOGY = 14, 175 HEADER_NUMA_TOPOLOGY = 14,
195 176
196 A list of NUMA node descriptions 177 A list of NUMA node descriptions
197 178
198 struct { 179 struct {
199 uint32_t nr; 180 uint32_t nr;
200 struct { 181 struct {
201 uint32_t nodenr; 182 uint32_t nodenr;
202 uint64_t mem_total; 183 uint64_t mem_total;
203 uint64_t mem_free; 184 uint64_t mem_free;
204 struct perf_header_string cpus; 185 struct perf_header_string cpus;
205 } nodes[nr]; /* Variable length records 186 } nodes[nr]; /* Variable length records */
206 }; 187 };
207 188
208 HEADER_BRANCH_STACK = 15, 189 HEADER_BRANCH_STACK = 15,
209 190
210 Not implemented in perf. 191 Not implemented in perf.
211 192
212 HEADER_PMU_MAPPINGS = 16, 193 HEADER_PMU_MAPPINGS = 16,
213 194
214 A list of PMU structures, defining the 195 A list of PMU structures, defining the different PMUs supported by perf.
215 196
216 struct { 197 struct {
217 uint32_t nr; 198 uint32_t nr;
218 struct pmu { 199 struct pmu {
219 uint32_t pmu_type; 200 uint32_t pmu_type;
220 struct perf_header_string pmu_na 201 struct perf_header_string pmu_name;
221 } [nr]; /* Variable length records */ 202 } [nr]; /* Variable length records */
222 }; 203 };
223 204
224 HEADER_GROUP_DESC = 17, 205 HEADER_GROUP_DESC = 17,
225 206
226 Description of counter groups ({...} i 207 Description of counter groups ({...} in perf syntax)
227 208
228 struct { 209 struct {
229 uint32_t nr; 210 uint32_t nr;
230 struct { 211 struct {
231 struct perf_header_string stri 212 struct perf_header_string string;
232 uint32_t leader_idx; 213 uint32_t leader_idx;
233 uint32_t nr_members; 214 uint32_t nr_members;
234 } [nr]; /* Variable length records */ 215 } [nr]; /* Variable length records */
235 }; 216 };
236 217
237 HEADER_AUXTRACE = 18, 218 HEADER_AUXTRACE = 18,
238 219
239 Define additional auxtrace areas in the perf.d 220 Define additional auxtrace areas in the perf.data. auxtrace is used to store
240 undecoded hardware tracing information, such a 221 undecoded hardware tracing information, such as Intel Processor Trace data.
241 222
242 /** 223 /**
243 * struct auxtrace_index_entry - indexes a AUX 224 * struct auxtrace_index_entry - indexes a AUX area tracing event within a
244 * perf.data fil 225 * perf.data file.
245 * @file_offset: offset within the perf.data f 226 * @file_offset: offset within the perf.data file
246 * @sz: size of the event 227 * @sz: size of the event
247 */ 228 */
248 struct auxtrace_index_entry { 229 struct auxtrace_index_entry {
249 u64 file_offset; 230 u64 file_offset;
250 u64 sz; 231 u64 sz;
251 }; 232 };
252 233
253 #define PERF_AUXTRACE_INDEX_ENTRY_COUNT 256 234 #define PERF_AUXTRACE_INDEX_ENTRY_COUNT 256
254 235
255 /** 236 /**
256 * struct auxtrace_index - index of AUX area t 237 * struct auxtrace_index - index of AUX area tracing events within a perf.data
257 * file. 238 * file.
258 * @list: linking a number of arrays of entrie 239 * @list: linking a number of arrays of entries
259 * @nr: number of entries 240 * @nr: number of entries
260 * @entries: array of entries 241 * @entries: array of entries
261 */ 242 */
262 struct auxtrace_index { 243 struct auxtrace_index {
263 struct list_head list; 244 struct list_head list;
264 size_t nr; 245 size_t nr;
265 struct auxtrace_index_entry entries[PE 246 struct auxtrace_index_entry entries[PERF_AUXTRACE_INDEX_ENTRY_COUNT];
266 }; 247 };
267 248
268 HEADER_STAT = 19, 249 HEADER_STAT = 19,
269 250
270 This is merely a flag signifying that the data 251 This is merely a flag signifying that the data section contains data
271 recorded from perf stat record. 252 recorded from perf stat record.
272 253
273 HEADER_CACHE = 20, 254 HEADER_CACHE = 20,
274 255
275 Description of the cache hierarchy. Based on t 256 Description of the cache hierarchy. Based on the Linux sysfs format
276 in /sys/devices/system/cpu/cpu*/cache/ 257 in /sys/devices/system/cpu/cpu*/cache/
277 258
278 u32 version Currently always 1 259 u32 version Currently always 1
279 u32 number_of_cache_levels 260 u32 number_of_cache_levels
280 261
281 struct { 262 struct {
282 u32 level; 263 u32 level;
283 u32 line_size; 264 u32 line_size;
284 u32 sets; 265 u32 sets;
285 u32 ways; 266 u32 ways;
286 struct perf_header_string type; 267 struct perf_header_string type;
287 struct perf_header_string size; 268 struct perf_header_string size;
288 struct perf_header_string map; 269 struct perf_header_string map;
289 }[number_of_cache_levels]; 270 }[number_of_cache_levels];
290 271
291 HEADER_SAMPLE_TIME = 21, 272 HEADER_SAMPLE_TIME = 21,
292 273
293 Two uint64_t for the time of first sample and 274 Two uint64_t for the time of first sample and the time of last sample.
294 275
295 HEADER_SAMPLE_TOPOLOGY = 22, <<
296 <<
297 Physical memory map and its node assignments. <<
298 <<
299 The format of data in MEM_TOPOLOGY is as follo <<
300 <<
301 u64 version; // Currently 1 <<
302 u64 block_size_bytes; // /sys/device <<
303 u64 count; // number of n <<
304 <<
305 struct memory_node { <<
306 u64 node_id; // node index <<
307 u64 size; // size of bit <<
308 struct bitmap { <<
309 /* size of bitmap again */ <<
310 u64 bitmapsize; <<
311 /* bitmap of memory indexes th <<
312 /* /sys/devices/system/node/no <<
313 u64 entries[(bitmapsize/64)+1] <<
314 } <<
315 }[count]; <<
316 <<
317 The MEM_TOPOLOGY can be displayed with followi <<
318 <<
319 $ perf report --header-only -I <<
320 ... <<
321 # memory nodes (nr 1, block size 0x8000000): <<
322 # 0 [7G]: 0-23,32-69 <<
323 <<
324 HEADER_CLOCKID = 23, <<
325 <<
326 One uint64_t for the clockid frequency, specif <<
327 record -k' (see clock_gettime()), to enable ti <<
328 conversion into wall clock time on the reporti <<
329 <<
330 HEADER_DIR_FORMAT = 24, <<
331 <<
332 The data files layout is described by HEADER_D <<
333 holds only version number (1): <<
334 <<
335 uint64_t version; <<
336 <<
337 The current version holds only version value ( <<
338 <<
339 - Follow the 'data.*' name format. <<
340 <<
341 - Contain raw events data in standard perf for <<
342 to be sorted) <<
343 <<
344 Future versions are expected to describe diffe <<
345 to special needs. <<
346 <<
347 HEADER_BPF_PROG_INFO = 25, <<
348 <<
349 struct perf_bpil, which contains detailed info <<
350 a BPF program, including type, id, tag, jited/ <<
351 <<
352 HEADER_BPF_BTF = 26, <<
353 <<
354 Contains BPF Type Format (BTF). For more infor <<
355 refer to Documentation/bpf/btf.rst. <<
356 <<
357 struct { <<
358 u32 id; <<
359 u32 data_size; <<
360 char data[]; <<
361 }; <<
362 <<
363 HEADER_COMPRESSED = 27, <<
364 <<
365 struct { <<
366 u32 version; <<
367 u32 type; <<
368 u32 level; <<
369 u32 ratio; <<
370 u32 mmap_len; <<
371 }; <<
372 <<
373 Indicates that trace contains records of PERF_ <<
374 that have perf_events records in compressed fo <<
375 <<
376 HEADER_CPU_PMU_CAPS = 28, <<
377 <<
378 A list of cpu PMU capabilities. The fo <<
379 <<
380 struct { <<
381 u32 nr_cpu_pmu_caps; <<
382 { <<
383 char name[]; <<
384 char value[]; <<
385 } [nr_cpu_pmu_caps] <<
386 }; <<
387 <<
388 <<
389 Example: <<
390 cpu pmu capabilities: branches=32, max_precis <<
391 <<
392 HEADER_CLOCK_DATA = 29, <<
393 <<
394 Contains clock id and its reference ti <<
395 time taken at the 'same time', both va <<
396 The format of data is as below. <<
397 <<
398 struct { <<
399 u32 version; /* version = 1 */ <<
400 u32 clockid; <<
401 u64 wall_clock_ns; <<
402 u64 clockid_time_ns; <<
403 }; <<
404 <<
405 HEADER_HYBRID_TOPOLOGY = 30, <<
406 <<
407 Indicate the hybrid CPUs. The format of data i <<
408 <<
409 struct { <<
410 u32 nr; <<
411 struct { <<
412 char pmu_name[]; <<
413 char cpus[]; <<
414 } [nr]; /* Variable length records */ <<
415 }; <<
416 <<
417 Example: <<
418 hybrid cpu system: <<
419 cpu_core cpu list : 0-15 <<
420 cpu_atom cpu list : 16-23 <<
421 <<
422 HEADER_PMU_CAPS = 31, <<
423 <<
424 List of pmu capabilities (except cpu p <<
425 covered by HEADER_CPU_PMU_CAPS). Note <<
426 capabilities are also stored here. <<
427 <<
428 struct { <<
429 u32 nr_pmu; <<
430 struct { <<
431 u32 nr_caps; <<
432 { <<
433 char name[]; <<
434 char value[]; <<
435 } [nr_caps]; <<
436 char pmu_name[]; <<
437 } [nr_pmu]; <<
438 }; <<
439 <<
440 other bits are reserved and should ign 276 other bits are reserved and should ignored for now
441 HEADER_FEAT_BITS = 256, 277 HEADER_FEAT_BITS = 256,
442 278
443 Attributes 279 Attributes
444 280
445 This is an array of perf_event_attrs, each att 281 This is an array of perf_event_attrs, each attr_size bytes long, which defines
446 each event collected. See perf_event.h or the 282 each event collected. See perf_event.h or the man page for a detailed
447 description. 283 description.
448 284
449 Data 285 Data
450 286
451 This section is the bulk of the file. It consi 287 This section is the bulk of the file. It consist of a stream of perf_events
452 describing events. This matches the format gen 288 describing events. This matches the format generated by the kernel.
453 See perf_event.h or the manpage for a detailed 289 See perf_event.h or the manpage for a detailed description.
454 290
455 Some notes on parsing: 291 Some notes on parsing:
456 292
457 Ordering 293 Ordering
458 294
459 The events are not necessarily in time stamp o 295 The events are not necessarily in time stamp order, as they can be
460 collected in parallel on different CPUs. If th 296 collected in parallel on different CPUs. If the events should be
461 processed in time order they need to be sorted 297 processed in time order they need to be sorted first. It is possible
462 to only do a partial sort using the FINISHED_R 298 to only do a partial sort using the FINISHED_ROUND event header (see
463 below). perf record guarantees that there is n 299 below). perf record guarantees that there is no reordering over a
464 FINISHED_ROUND. 300 FINISHED_ROUND.
465 301
466 ID vs IDENTIFIER 302 ID vs IDENTIFIER
467 303
468 When the event stream contains multiple events 304 When the event stream contains multiple events each event is identified
469 by an ID. This can be either through the PERF_ 305 by an ID. This can be either through the PERF_SAMPLE_ID or the
470 PERF_SAMPLE_IDENTIFIER header. The PERF_SAMPLE 306 PERF_SAMPLE_IDENTIFIER header. The PERF_SAMPLE_IDENTIFIER header is
471 at a fixed offset from the event header, which 307 at a fixed offset from the event header, which allows reliable
472 parsing of the header. Relying on ID may be am 308 parsing of the header. Relying on ID may be ambiguous.
473 IDENTIFIER is only supported by newer Linux ke 309 IDENTIFIER is only supported by newer Linux kernels.
474 310
475 Perf record specific events: 311 Perf record specific events:
476 312
477 In addition to the kernel generated event type 313 In addition to the kernel generated event types perf record adds its
478 own event types (in addition it also synthesiz 314 own event types (in addition it also synthesizes some kernel events,
479 for example MMAP events) 315 for example MMAP events)
480 316
481 PERF_RECORD_USER_TYPE_START 317 PERF_RECORD_USER_TYPE_START = 64,
482 PERF_RECORD_HEADER_ATTR 318 PERF_RECORD_HEADER_ATTR = 64,
483 319
484 struct attr_event { 320 struct attr_event {
485 struct perf_event_header header; 321 struct perf_event_header header;
486 struct perf_event_attr attr; 322 struct perf_event_attr attr;
487 uint64_t id[]; 323 uint64_t id[];
488 }; 324 };
489 325
490 PERF_RECORD_HEADER_EVENT_TYPE 326 PERF_RECORD_HEADER_EVENT_TYPE = 65, /* deprecated */
491 327
492 #define MAX_EVENT_NAME 64 328 #define MAX_EVENT_NAME 64
493 329
494 struct perf_trace_event_type { 330 struct perf_trace_event_type {
495 uint64_t event_id; 331 uint64_t event_id;
496 char name[MAX_EVENT_NAME]; 332 char name[MAX_EVENT_NAME];
497 }; 333 };
498 334
499 struct event_type_event { 335 struct event_type_event {
500 struct perf_event_header header; 336 struct perf_event_header header;
501 struct perf_trace_event_type event_typ 337 struct perf_trace_event_type event_type;
502 }; 338 };
503 339
504 340
505 PERF_RECORD_HEADER_TRACING_DATA 341 PERF_RECORD_HEADER_TRACING_DATA = 66,
506 342
507 Describe me 343 Describe me
508 344
509 struct tracing_data_event { 345 struct tracing_data_event {
510 struct perf_event_header header; 346 struct perf_event_header header;
511 uint32_t size; 347 uint32_t size;
512 }; 348 };
513 349
514 PERF_RECORD_HEADER_BUILD_ID 350 PERF_RECORD_HEADER_BUILD_ID = 67,
515 351
516 Define a ELF build ID for a referenced executa 352 Define a ELF build ID for a referenced executable.
517 353
518 struct build_id_event; /* See above * 354 struct build_id_event; /* See above */
519 355
520 PERF_RECORD_FINISHED_ROUND 356 PERF_RECORD_FINISHED_ROUND = 68,
521 357
522 No event reordering over this header. No paylo 358 No event reordering over this header. No payload.
523 359
524 PERF_RECORD_ID_INDEX 360 PERF_RECORD_ID_INDEX = 69,
525 361
526 Map event ids to CPUs and TIDs. 362 Map event ids to CPUs and TIDs.
527 363
528 struct id_index_entry { 364 struct id_index_entry {
529 uint64_t id; 365 uint64_t id;
530 uint64_t idx; 366 uint64_t idx;
531 uint64_t cpu; 367 uint64_t cpu;
532 uint64_t tid; 368 uint64_t tid;
533 }; 369 };
534 370
535 struct id_index_event { 371 struct id_index_event {
536 struct perf_event_header header; 372 struct perf_event_header header;
537 uint64_t nr; 373 uint64_t nr;
538 struct id_index_entry entries[nr]; 374 struct id_index_entry entries[nr];
539 }; 375 };
540 376
541 PERF_RECORD_AUXTRACE_INFO 377 PERF_RECORD_AUXTRACE_INFO = 70,
542 378
543 Auxtrace type specific information. Describe m 379 Auxtrace type specific information. Describe me
544 380
545 struct auxtrace_info_event { 381 struct auxtrace_info_event {
546 struct perf_event_header header; 382 struct perf_event_header header;
547 uint32_t type; 383 uint32_t type;
548 uint32_t reserved__; /* For alignment 384 uint32_t reserved__; /* For alignment */
549 uint64_t priv[]; 385 uint64_t priv[];
550 }; 386 };
551 387
552 PERF_RECORD_AUXTRACE 388 PERF_RECORD_AUXTRACE = 71,
553 389
554 Defines auxtrace data. Followed by the actual 390 Defines auxtrace data. Followed by the actual data. The contents of
555 the auxtrace data is dependent on the event an 391 the auxtrace data is dependent on the event and the CPU. For example
556 for Intel Processor Trace it contains Processo 392 for Intel Processor Trace it contains Processor Trace data generated
557 by the CPU. 393 by the CPU.
558 394
559 struct auxtrace_event { 395 struct auxtrace_event {
560 struct perf_event_header header; 396 struct perf_event_header header;
561 uint64_t size; 397 uint64_t size;
562 uint64_t offset; 398 uint64_t offset;
563 uint64_t reference; 399 uint64_t reference;
564 uint32_t idx; 400 uint32_t idx;
565 uint32_t tid; 401 uint32_t tid;
566 uint32_t cpu; 402 uint32_t cpu;
567 uint32_t reserved__; /* For alignment 403 uint32_t reserved__; /* For alignment */
568 }; 404 };
569 405
570 struct aux_event { 406 struct aux_event {
571 struct perf_event_header header; 407 struct perf_event_header header;
572 uint64_t aux_offset; 408 uint64_t aux_offset;
573 uint64_t aux_size; 409 uint64_t aux_size;
574 uint64_t flags; 410 uint64_t flags;
575 }; 411 };
576 412
577 PERF_RECORD_AUXTRACE_ERROR 413 PERF_RECORD_AUXTRACE_ERROR = 72,
578 414
579 Describes an error in hardware tracing 415 Describes an error in hardware tracing
580 416
581 enum auxtrace_error_type { 417 enum auxtrace_error_type {
582 PERF_AUXTRACE_ERROR_ITRACE = 1, 418 PERF_AUXTRACE_ERROR_ITRACE = 1,
583 PERF_AUXTRACE_ERROR_MAX 419 PERF_AUXTRACE_ERROR_MAX
584 }; 420 };
585 421
586 #define MAX_AUXTRACE_ERROR_MSG 64 422 #define MAX_AUXTRACE_ERROR_MSG 64
587 423
588 struct auxtrace_error_event { 424 struct auxtrace_error_event {
589 struct perf_event_header header; 425 struct perf_event_header header;
590 uint32_t type; 426 uint32_t type;
591 uint32_t code; 427 uint32_t code;
592 uint32_t cpu; 428 uint32_t cpu;
593 uint32_t pid; 429 uint32_t pid;
594 uint32_t tid; 430 uint32_t tid;
595 uint32_t reserved__; /* For alignment 431 uint32_t reserved__; /* For alignment */
596 uint64_t ip; 432 uint64_t ip;
597 char msg[MAX_AUXTRACE_ERROR_MSG]; 433 char msg[MAX_AUXTRACE_ERROR_MSG];
598 }; 434 };
599 435
600 PERF_RECORD_HEADER_FEATURE 436 PERF_RECORD_HEADER_FEATURE = 80,
601 437
602 Describes a header feature. These are records 438 Describes a header feature. These are records used in pipe-mode that
603 contain information that otherwise would be in 439 contain information that otherwise would be in perf.data file's header.
604 <<
605 PERF_RECORD_COMPRESSED <<
606 <<
607 struct compressed_event { <<
608 struct perf_event_header header <<
609 char data[] <<
610 }; <<
611 <<
612 PERF_RECORD_FINISHED_INIT <<
613 <<
614 Marks the end of records for the system, pre-e <<
615 sessions, etc. Those are the ones prefixed PER <<
616 <<
617 This is used, for instance, to 'perf inject' e <<
618 regular events, those emitted by the kernel, t <<
619 host records. <<
620 <<
621 <<
622 The header is followed by compressed data fram <<
623 into array of perf trace records. The size of <<
624 record including the header is limited by the <<
625 440
626 Event types 441 Event types
627 442
628 Define the event attributes with their IDs. 443 Define the event attributes with their IDs.
629 444
630 An array bound by the perf_file_section size. 445 An array bound by the perf_file_section size.
631 446
632 struct { 447 struct {
633 struct perf_event_attr attr; 448 struct perf_event_attr attr; /* Size defined by header.attr_size */
634 struct perf_file_section ids; 449 struct perf_file_section ids;
635 } 450 }
636 451
637 ids points to a array of uint64_t defining the 452 ids points to a array of uint64_t defining the ids for event attr attr.
638 453
639 Pipe-mode data 454 Pipe-mode data
640 455
641 Pipe-mode avoid seeks in the file by removing 456 Pipe-mode avoid seeks in the file by removing the perf_file_section and flags
642 from the struct perf_header. The trimmed heade 457 from the struct perf_header. The trimmed header is:
643 458
644 struct perf_pipe_file_header { 459 struct perf_pipe_file_header {
645 u64 magic; 460 u64 magic;
646 u64 size; 461 u64 size;
647 }; 462 };
648 463
649 The information about attrs, data, and event_t 464 The information about attrs, data, and event_types is instead in the
650 synthesized events PERF_RECORD_ATTR, PERF_RECO 465 synthesized events PERF_RECORD_ATTR, PERF_RECORD_HEADER_TRACING_DATA,
651 PERF_RECORD_HEADER_EVENT_TYPE, and PERF_RECORD 466 PERF_RECORD_HEADER_EVENT_TYPE, and PERF_RECORD_HEADER_FEATURE
652 that are generated by perf record in pipe-mode 467 that are generated by perf record in pipe-mode.
653 468
654 469
655 References: 470 References:
656 471
657 include/uapi/linux/perf_event.h 472 include/uapi/linux/perf_event.h
658 473
659 This is the canonical description of the kerne 474 This is the canonical description of the kernel generated perf_events
660 and the perf_event_attrs. 475 and the perf_event_attrs.
661 476
662 perf_events manpage 477 perf_events manpage
663 478
664 A manpage describing perf_event and perf_event 479 A manpage describing perf_event and perf_event_attr is here:
665 http://web.eece.maine.edu/~vweaver/projects/pe 480 http://web.eece.maine.edu/~vweaver/projects/perf_events/programming.html
666 This tends to be slightly behind the kernel in 481 This tends to be slightly behind the kernel include, but has better
667 descriptions. An (typically older) version of 482 descriptions. An (typically older) version of the man page may be
668 included with the standard Linux man pages, av 483 included with the standard Linux man pages, available with "man
669 perf_events" 484 perf_events"
670 485
671 pmu-tools 486 pmu-tools
672 487
673 https://github.com/andikleen/pmu-tools/tree/ma 488 https://github.com/andikleen/pmu-tools/tree/master/parser
674 489
675 A definition of the perf.data format in python 490 A definition of the perf.data format in python "construct" format is available
676 in pmu-tools parser. This allows to read perf. 491 in pmu-tools parser. This allows to read perf.data from python and dump it.
677 492
678 quipper 493 quipper
679 494
680 The quipper C++ parser is available at 495 The quipper C++ parser is available at
681 http://github.com/google/perf_data_converter/t 496 http://github.com/google/perf_data_converter/tree/master/src/quipper
682 497
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.