Commit | Line | Data |
---|---|---|
133dc4c3 | 1 | perf-script(1) |
4778e0e8 | 2 | ============= |
0a02ad93 IM |
3 | |
4 | NAME | |
5 | ---- | |
133dc4c3 | 6 | perf-script - Read perf.data (created by perf record) and display trace output |
0a02ad93 IM |
7 | |
8 | SYNOPSIS | |
9 | -------- | |
10 | [verse] | |
133dc4c3 IM |
11 | 'perf script' [<options>] |
12 | 'perf script' [<options>] record <script> [<record-options>] <command> | |
13 | 'perf script' [<options>] report <script> [script-args] | |
14 | 'perf script' [<options>] <script> <required-script-args> [<record-options>] <command> | |
15 | 'perf script' [<options>] <top-script> [script-args] | |
0a02ad93 IM |
16 | |
17 | DESCRIPTION | |
18 | ----------- | |
19 | This command reads the input file and displays the trace recorded. | |
20 | ||
133dc4c3 | 21 | There are several variants of perf script: |
a6005123 | 22 | |
133dc4c3 | 23 | 'perf script' to see a detailed trace of the workload that was |
a6005123 TZ |
24 | recorded. |
25 | ||
cff68e58 TZ |
26 | You can also run a set of pre-canned scripts that aggregate and |
27 | summarize the raw trace data in various ways (the list of scripts is | |
133dc4c3 | 28 | available via 'perf script -l'). The following variants allow you to |
cff68e58 TZ |
29 | record and run those scripts: |
30 | ||
133dc4c3 IM |
31 | 'perf script record <script> <command>' to record the events required |
32 | for 'perf script report'. <script> is the name displayed in the | |
33 | output of 'perf script --list' i.e. the actual script name minus any | |
d3c4f798 TZ |
34 | language extension. If <command> is not specified, the events are |
35 | recorded using the -a (system-wide) 'perf record' option. | |
a6005123 | 36 | |
133dc4c3 | 37 | 'perf script report <script> [args]' to run and display the results |
d3c4f798 | 38 | of <script>. <script> is the name displayed in the output of 'perf |
5c64f99b | 39 | script --list' i.e. the actual script name minus any language |
133dc4c3 | 40 | extension. The perf.data output from a previous run of 'perf script |
a6005123 | 41 | record <script>' is used and should be present for this command to |
d3c4f798 TZ |
42 | succeed. [args] refers to the (mainly optional) args expected by |
43 | the script. | |
44 | ||
133dc4c3 | 45 | 'perf script <script> <required-script-args> <command>' to both |
d3c4f798 TZ |
46 | record the events required for <script> and to run the <script> |
47 | using 'live-mode' i.e. without writing anything to disk. <script> | |
133dc4c3 | 48 | is the name displayed in the output of 'perf script --list' i.e. the |
d3c4f798 TZ |
49 | actual script name minus any language extension. If <command> is |
50 | not specified, the events are recorded using the -a (system-wide) | |
51 | 'perf record' option. If <script> has any required args, they | |
52 | should be specified before <command>. This mode doesn't allow for | |
53 | optional script args to be specified; if optional script args are | |
133dc4c3 IM |
54 | desired, they can be specified using separate 'perf script record' |
55 | and 'perf script report' commands, with the stdout of the record step | |
d3c4f798 TZ |
56 | piped to the stdin of the report script, using the '-o -' and '-i -' |
57 | options of the corresponding commands. | |
58 | ||
133dc4c3 | 59 | 'perf script <top-script>' to both record the events required for |
d3c4f798 TZ |
60 | <top-script> and to run the <top-script> using 'live-mode' |
61 | i.e. without writing anything to disk. <top-script> is the name | |
133dc4c3 | 62 | displayed in the output of 'perf script --list' i.e. the actual |
d3c4f798 TZ |
63 | script name minus any language extension; a <top-script> is defined |
64 | as any script name ending with the string 'top'. | |
65 | ||
133dc4c3 | 66 | [<record-options>] can be passed to the record steps of 'perf script |
d3c4f798 | 67 | record' and 'live-mode' variants; this isn't possible however for |
133dc4c3 | 68 | <top-script> 'live-mode' or 'perf script report' variants. |
a6005123 | 69 | |
cff68e58 TZ |
70 | See the 'SEE ALSO' section for links to language-specific |
71 | information on how to write and run your own trace scripts. | |
72 | ||
0a02ad93 IM |
73 | OPTIONS |
74 | ------- | |
d3c4f798 TZ |
75 | <command>...:: |
76 | Any command you can specify in a shell. | |
77 | ||
0a02ad93 | 78 | -D:: |
5c64f99b | 79 | --dump-raw-trace=:: |
0a02ad93 IM |
80 | Display verbose dump of the trace data. |
81 | ||
a6005123 TZ |
82 | -L:: |
83 | --Latency=:: | |
84 | Show latency attributes (irqs/preemption disabled, etc). | |
85 | ||
86 | -l:: | |
87 | --list=:: | |
88 | Display a list of available trace scripts. | |
89 | ||
f526d68b | 90 | -s ['lang']:: |
89fbf0b8 TZ |
91 | --script=:: |
92 | Process trace data with the given script ([lang]:script[.ext]). | |
f526d68b TZ |
93 | If the string 'lang' is specified in place of a script name, a |
94 | list of supported languages will be displayed instead. | |
89fbf0b8 TZ |
95 | |
96 | -g:: | |
97 | --gen-script=:: | |
133dc4c3 | 98 | Generate perf-script.[ext] starter script for given language, |
89fbf0b8 TZ |
99 | using current perf.data. |
100 | ||
d3c4f798 TZ |
101 | -a:: |
102 | Force system-wide collection. Scripts run without a <command> | |
103 | normally use -a by default, while scripts run with a <command> | |
104 | normally don't - this option allows the latter to be run in | |
105 | system-wide mode. | |
106 | ||
646420f1 SB |
107 | -i:: |
108 | --input=:: | |
efad1415 | 109 | Input file name. (default: perf.data unless stdin is a fifo) |
646420f1 SB |
110 | |
111 | -d:: | |
112 | --debug-mode:: | |
113 | Do various checks like samples ordering and lost events. | |
d3c4f798 | 114 | |
dc323ce8 | 115 | -F:: |
176fcc5c | 116 | --fields:: |
745f43e3 | 117 | Comma separated list of fields to print. Options are: |
400ea6d3 | 118 | comm, tid, pid, time, cpu, event, trace, ip, sym, dso, addr, symoff, |
b1491ace | 119 | srcline, period, iregs, uregs, brstack, brstacksym, flags, bpf-output, brstackinsn, |
dd2e18e9 | 120 | brstackoff, callindent, insn, insnlen, synth, phys_addr, metric, misc, srccode. |
47e78084 | 121 | Field list can be prepended with the type, trace, sw or hw, |
1424dc96 | 122 | to indicate to which event type the field list applies. |
cbb0bba9 | 123 | e.g., -F sw:comm,tid,time,ip,sym and -F trace:time,cpu,trace |
c0230b2b | 124 | |
cbb0bba9 | 125 | perf script -F <fields> |
176fcc5c ACM |
126 | |
127 | is equivalent to: | |
128 | ||
cbb0bba9 | 129 | perf script -F trace:<fields> -F sw:<fields> -F hw:<fields> |
48000a1a | 130 | |
176fcc5c ACM |
131 | i.e., the specified fields apply to all event types if the type string |
132 | is not given. | |
48000a1a | 133 | |
36ce5651 AK |
134 | In addition to overriding fields, it is also possible to add or remove |
135 | fields from the defaults. For example | |
136 | ||
137 | -F -cpu,+insn | |
138 | ||
139 | removes the cpu field and adds the insn field. Adding/removing fields | |
140 | cannot be mixed with normal overriding. | |
141 | ||
176fcc5c ACM |
142 | The arguments are processed in the order received. A later usage can |
143 | reset a prior request. e.g.: | |
48000a1a | 144 | |
cbb0bba9 | 145 | -F trace: -F comm,tid,time,ip,sym |
48000a1a | 146 | |
cbb0bba9 | 147 | The first -F suppresses trace events (field list is ""), but then the |
787bef17 | 148 | second invocation sets the fields to comm,tid,time,ip,sym. In this case a |
176fcc5c | 149 | warning is given to the user: |
48000a1a | 150 | |
176fcc5c | 151 | "Overriding previous field request for all events." |
48000a1a | 152 | |
96355f2c | 153 | Alternatively, consider the order: |
48000a1a | 154 | |
cbb0bba9 | 155 | -F comm,tid,time,ip,sym -F trace: |
48000a1a | 156 | |
cbb0bba9 | 157 | The first -F sets the fields for all events and the second -F |
176fcc5c ACM |
158 | suppresses trace events. The user is given a warning message about |
159 | the override, and the result of the above is that only S/W and H/W | |
160 | events are displayed with the given fields. | |
48000a1a | 161 | |
6ef362fd JO |
162 | It's possible tp add/remove fields only for specific event type: |
163 | ||
164 | -Fsw:-cpu,-period | |
165 | ||
166 | removes cpu and period from software events. | |
167 | ||
176fcc5c ACM |
168 | For the 'wildcard' option if a user selected field is invalid for an |
169 | event type, a message is displayed to the user that the option is | |
170 | ignored for that type. For example: | |
48000a1a | 171 | |
cbb0bba9 | 172 | $ perf script -F comm,tid,trace |
176fcc5c ACM |
173 | 'trace' not valid for hardware events. Ignoring. |
174 | 'trace' not valid for software events. Ignoring. | |
48000a1a | 175 | |
176fcc5c ACM |
176 | Alternatively, if the type is given an invalid field is specified it |
177 | is an error. For example: | |
48000a1a | 178 | |
cbb0bba9 | 179 | perf script -v -F sw:comm,tid,trace |
176fcc5c | 180 | 'trace' not valid for software events. |
48000a1a | 181 | |
176fcc5c | 182 | At this point usage is displayed, and perf-script exits. |
48000a1a | 183 | |
400ea6d3 AH |
184 | The flags field is synthesized and may have a value when Instruction |
185 | Trace decoding. The flags are "bcrosyiABEx" which stand for branch, | |
186 | call, return, conditional, system, asynchronous, interrupt, | |
187 | transaction abort, trace begin, trace end, and in transaction, | |
055cd33d AH |
188 | respectively. Known combinations of flags are printed more nicely e.g. |
189 | "call" for "bc", "return" for "br", "jcc" for "bo", "jmp" for "b", | |
190 | "int" for "bci", "iret" for "bri", "syscall" for "bcs", "sysret" for "brs", | |
191 | "async" for "by", "hw int" for "bcyi", "tx abrt" for "bA", "tr strt" for "bB", | |
192 | "tr end" for "bE". However the "x" flag will be display separately in those | |
193 | cases e.g. "jcc (x)" for a condition branch within a transaction. | |
400ea6d3 | 194 | |
e216708d AH |
195 | The callindent field is synthesized and may have a value when |
196 | Instruction Trace decoding. For calls and returns, it will display the | |
197 | name of the symbol indented with spaces to reflect the stack depth. | |
198 | ||
224e2c97 AK |
199 | When doing instruction trace decoding insn and insnlen give the |
200 | instruction bytes and the instruction length of the current | |
201 | instruction. | |
202 | ||
47e78084 AH |
203 | The synth field is used by synthesized events which may be created when |
204 | Instruction Trace decoding. | |
205 | ||
176fcc5c | 206 | Finally, a user may not set fields to none for all event types. |
cbb0bba9 | 207 | i.e., -F "" is not allowed. |
176fcc5c | 208 | |
dc323ce8 | 209 | The brstack output includes branch related information with raw addresses using the |
48d02a1d | 210 | /v/v/v/v/cycles syntax in the following order: |
dc323ce8 SE |
211 | FROM: branch source instruction |
212 | TO : branch target instruction | |
213 | M/P/-: M=branch target mispredicted or branch direction was mispredicted, P=target predicted or direction predicted, -=not supported | |
214 | X/- : X=branch inside a transactional region, -=not in transaction region or not supported | |
215 | A/- : A=TSX abort entry, -=not aborted region or not supported | |
48d02a1d | 216 | cycles |
dc323ce8 SE |
217 | |
218 | The brstacksym is identical to brstack, except that the FROM and TO addresses are printed in a symbolic form if possible. | |
219 | ||
48d02a1d AK |
220 | When brstackinsn is specified the full assembler sequences of branch sequences for each sample |
221 | is printed. This is the full execution path leading to the sample. This is only supported when the | |
222 | sample was recorded with perf record -b or -j any. | |
223 | ||
106dacd8 MS |
224 | The brstackoff field will print an offset into a specific dso/binary. |
225 | ||
4bd1bef8 AK |
226 | With the metric option perf script can compute metrics for |
227 | sampling periods, similar to perf stat. This requires | |
228 | specifying a group with multiple metrics with the :S option | |
229 | for perf record. perf will sample on the first event, and | |
230 | compute metrics for all the events in the group. Please note | |
231 | that the metric computed is averaged over the whole sampling | |
232 | period, not just for the sample point. | |
233 | ||
28a0b398 JO |
234 | For sample events it's possible to display misc field with -F +misc option, |
235 | following letters are displayed for each bit: | |
236 | ||
bf30cc18 AB |
237 | PERF_RECORD_MISC_KERNEL K |
238 | PERF_RECORD_MISC_USER U | |
239 | PERF_RECORD_MISC_HYPERVISOR H | |
240 | PERF_RECORD_MISC_GUEST_KERNEL G | |
241 | PERF_RECORD_MISC_GUEST_USER g | |
242 | PERF_RECORD_MISC_MMAP_DATA* M | |
243 | PERF_RECORD_MISC_COMM_EXEC E | |
244 | PERF_RECORD_MISC_SWITCH_OUT S | |
245 | PERF_RECORD_MISC_SWITCH_OUT_PREEMPT Sp | |
28a0b398 JO |
246 | |
247 | $ perf script -F +misc ... | |
248 | sched-messaging 1414 K 28690.636582: 4590 cycles ... | |
249 | sched-messaging 1407 U 28690.636600: 325620 cycles ... | |
250 | sched-messaging 1414 K 28690.636608: 19473 cycles ... | |
251 | misc field ___________/ | |
252 | ||
c0230b2b DA |
253 | -k:: |
254 | --vmlinux=<file>:: | |
255 | vmlinux pathname | |
256 | ||
257 | --kallsyms=<file>:: | |
258 | kallsyms pathname | |
259 | ||
260 | --symfs=<directory>:: | |
261 | Look for files with symbols relative to this directory. | |
262 | ||
263 | -G:: | |
264 | --hide-call-graph:: | |
265 | When printing symbols do not display call chain. | |
745f43e3 | 266 | |
64eff7d9 DA |
267 | --stop-bt:: |
268 | Stop display of callgraph at these symbols | |
269 | ||
c8e66720 | 270 | -C:: |
5d67be97 AB |
271 | --cpu:: Only report samples for the list of CPUs provided. Multiple CPUs can |
272 | be provided as a comma-separated list with no space: 0,1. Ranges of | |
273 | CPUs are specified with -: 0-2. Default is to report samples on all | |
274 | CPUs. | |
275 | ||
e7984b7b DA |
276 | -c:: |
277 | --comms=:: | |
278 | Only display events for these comms. CSV that understands | |
279 | file://filename entries. | |
280 | ||
e03eaa40 DA |
281 | --pid=:: |
282 | Only show events for given process ID (comma separated list). | |
283 | ||
284 | --tid=:: | |
285 | Only show events for given thread ID (comma separated list). | |
286 | ||
fbe96f29 SE |
287 | -I:: |
288 | --show-info:: | |
289 | Display extended information about the perf.data file. This adds | |
290 | information which may be very large and thus may clutter the display. | |
291 | It currently includes: cpu and numa topology of the host system. | |
292 | It can only be used with the perf script report mode. | |
293 | ||
0bc8d205 AN |
294 | --show-kernel-path:: |
295 | Try to resolve the path of [kernel.kallsyms] | |
296 | ||
ad7ebb9a NK |
297 | --show-task-events |
298 | Display task related events (e.g. FORK, COMM, EXIT). | |
299 | ||
ba1ddf42 NK |
300 | --show-mmap-events |
301 | Display mmap related events (e.g. MMAP, MMAP2). | |
302 | ||
96a44bbc HB |
303 | --show-namespace-events |
304 | Display namespace events i.e. events of type PERF_RECORD_NAMESPACES. | |
305 | ||
7c14898b AH |
306 | --show-switch-events |
307 | Display context switch events i.e. events of type PERF_RECORD_SWITCH or | |
308 | PERF_RECORD_SWITCH_CPU_WIDE. | |
309 | ||
3d7c27b6 JO |
310 | --show-lost-events |
311 | Display lost events i.e. events of type PERF_RECORD_LOST. | |
312 | ||
3233b37a JO |
313 | --show-round-events |
314 | Display finished round events i.e. events of type PERF_RECORD_FINISHED_ROUND. | |
315 | ||
77e0070d MD |
316 | --demangle:: |
317 | Demangle symbol names to human readable form. It's enabled by default, | |
318 | disable with --no-demangle. | |
319 | ||
320 | --demangle-kernel:: | |
321 | Demangle kernel symbol names to human readable form (for C++ kernels). | |
322 | ||
e90debdd JO |
323 | --header |
324 | Show perf.data header. | |
325 | ||
326 | --header-only | |
327 | Show only perf.data header. | |
328 | ||
7a680eb9 AH |
329 | --itrace:: |
330 | Options for decoding instruction tracing data. The options are: | |
331 | ||
60b88d87 | 332 | include::itrace.txt[] |
7a680eb9 AH |
333 | |
334 | To disable decoding entirely, use --no-itrace. | |
335 | ||
a9710ba0 AK |
336 | --full-source-path:: |
337 | Show the full path for source files for srcline output. | |
338 | ||
6125cc8d ACM |
339 | --max-stack:: |
340 | Set the stack depth limit when parsing the callchain, anything | |
341 | beyond the specified depth will be ignored. This is a trade-off | |
342 | between information loss and faster processing especially for | |
343 | workloads that can have a very long callchain stack. | |
344 | Note that when using the --itrace option the synthesized callchain size | |
345 | will override this value if the synthesized callchain size is bigger. | |
346 | ||
fe176085 | 347 | Default: 127 |
6125cc8d | 348 | |
83e19860 AH |
349 | --ns:: |
350 | Use 9 decimal places when displaying time (i.e. show the nanoseconds) | |
351 | ||
e0be62cc JO |
352 | -f:: |
353 | --force:: | |
354 | Don't do ownership validation. | |
355 | ||
a91f4c47 DA |
356 | --time:: |
357 | Only analyze samples within given time window: <start>,<stop>. Times | |
358 | have the format seconds.microseconds. If start is not given (i.e., time | |
359 | string is ',x.y') then analysis starts at the beginning of the file. If | |
360 | stop time is not given (i.e, time string is 'x.y,') then analysis goes | |
361 | to end of file. | |
362 | ||
2ab046cd | 363 | Also support time percent with multipe time range. Time string is |
cc2ef584 | 364 | 'a%/n,b%/m,...' or 'a%-b%,c%-%d,...'. |
2ab046cd JY |
365 | |
366 | For example: | |
cc2ef584 | 367 | Select the second 10% time slice: |
2ab046cd JY |
368 | perf script --time 10%/2 |
369 | ||
cc2ef584 | 370 | Select from 0% to 10% time slice: |
2ab046cd JY |
371 | perf script --time 0%-10% |
372 | ||
cc2ef584 | 373 | Select the first and second 10% time slices: |
2ab046cd JY |
374 | perf script --time 10%/1,10%/2 |
375 | ||
cc2ef584 | 376 | Select from 0% to 10% and 30% to 40% slices: |
2ab046cd JY |
377 | perf script --time 0%-10%,30%-40% |
378 | ||
48d02a1d AK |
379 | --max-blocks:: |
380 | Set the maximum number of program blocks to print with brstackasm for | |
381 | each sample. | |
382 | ||
a14390fd ACM |
383 | --per-event-dump:: |
384 | Create per event files with a "perf.data.EVENT.dump" name instead of | |
385 | printing to stdout, useful, for instance, for generating flamegraphs. | |
386 | ||
325fbff5 NK |
387 | --inline:: |
388 | If a callgraph address belongs to an inlined function, the inline stack | |
d8a88dd2 MW |
389 | will be printed. Each entry has function name and file/line. Enabled by |
390 | default, disable with --no-inline. | |
325fbff5 | 391 | |
b585ebdb AK |
392 | --insn-trace:: |
393 | Show instruction stream for intel_pt traces. Combine with --xed to | |
394 | show disassembly. | |
395 | ||
396 | --xed:: | |
397 | Run xed disassembler on output. Requires installing the xed disassembler. | |
398 | ||
d1b1552e AK |
399 | --call-trace:: |
400 | Show call stream for intel_pt traces. The CPUs are interleaved, but | |
401 | can be filtered with -C. | |
402 | ||
403 | --call-ret-trace:: | |
404 | Show call and return stream for intel_pt traces. | |
405 | ||
99f753f0 AK |
406 | --graph-function:: |
407 | For itrace only show specified functions and their callees for | |
408 | itrace. Multiple functions can be separated by comma. | |
409 | ||
0a02ad93 IM |
410 | SEE ALSO |
411 | -------- | |
133dc4c3 IM |
412 | linkperf:perf-record[1], linkperf:perf-script-perl[1], |
413 | linkperf:perf-script-python[1] |