6 perf-script - Read perf.data (created by perf record) and display trace output
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]
19 This command reads the input file and displays the trace recorded.
21 There are several variants of perf script:
23 'perf script' to see a detailed trace of the workload that was
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
28 available via 'perf script -l'). The following variants allow you to
29 record and run those scripts:
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
34 language extension. If <command> is not specified, the events are
35 recorded using the -a (system-wide) 'perf record' option.
37 'perf script report <script> [args]' to run and display the results
38 of <script>. <script> is the name displayed in the output of 'perf
39 script --list' i.e. the actual script name minus any language
40 extension. The perf.data output from a previous run of 'perf script
41 record <script>' is used and should be present for this command to
42 succeed. [args] refers to the (mainly optional) args expected by
45 'perf script <script> <required-script-args> <command>' to both
46 record the events required for <script> and to run the <script>
47 using 'live-mode' i.e. without writing anything to disk. <script>
48 is the name displayed in the output of 'perf script --list' i.e. the
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
54 desired, they can be specified using separate 'perf script record'
55 and 'perf script report' commands, with the stdout of the record step
56 piped to the stdin of the report script, using the '-o -' and '-i -'
57 options of the corresponding commands.
59 'perf script <top-script>' to both record the events required for
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
62 displayed in the output of 'perf script --list' i.e. the actual
63 script name minus any language extension; a <top-script> is defined
64 as any script name ending with the string 'top'.
66 [<record-options>] can be passed to the record steps of 'perf script
67 record' and 'live-mode' variants; this isn't possible however for
68 <top-script> 'live-mode' or 'perf script report' variants.
70 See the 'SEE ALSO' section for links to language-specific
71 information on how to write and run your own trace scripts.
76 Any command you can specify in a shell.
80 Display verbose dump of the trace data.
84 Show latency attributes (irqs/preemption disabled, etc).
88 Display a list of available trace scripts.
92 Process trace data with the given script ([lang]:script[.ext]).
93 If the string 'lang' is specified in place of a script name, a
94 list of supported languages will be displayed instead.
98 Generate perf-script.[ext] starter script for given language,
99 using current perf.data.
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
109 Input file name. (default: perf.data unless stdin is a fifo)
113 Do various checks like samples ordering and lost events.
117 Comma separated list of fields to print. Options are:
118 comm, tid, pid, time, cpu, event, trace, ip, sym, dso, addr, symoff,
119 srcline, period, iregs, uregs, brstack, brstacksym, flags, bpf-output,
120 brstackinsn, brstackoff, callindent, insn, insnlen, synth, phys_addr,
121 metric, misc, srccode, ipc, data_page_size.
122 Field list can be prepended with the type, trace, sw or hw,
123 to indicate to which event type the field list applies.
124 e.g., -F sw:comm,tid,time,ip,sym and -F trace:time,cpu,trace
126 perf script -F <fields>
130 perf script -F trace:<fields> -F sw:<fields> -F hw:<fields>
132 i.e., the specified fields apply to all event types if the type string
135 In addition to overriding fields, it is also possible to add or remove
136 fields from the defaults. For example
140 removes the cpu field and adds the insn field. Adding/removing fields
141 cannot be mixed with normal overriding.
143 The arguments are processed in the order received. A later usage can
144 reset a prior request. e.g.:
146 -F trace: -F comm,tid,time,ip,sym
148 The first -F suppresses trace events (field list is ""), but then the
149 second invocation sets the fields to comm,tid,time,ip,sym. In this case a
150 warning is given to the user:
152 "Overriding previous field request for all events."
154 Alternatively, consider the order:
156 -F comm,tid,time,ip,sym -F trace:
158 The first -F sets the fields for all events and the second -F
159 suppresses trace events. The user is given a warning message about
160 the override, and the result of the above is that only S/W and H/W
161 events are displayed with the given fields.
163 It's possible tp add/remove fields only for specific event type:
167 removes cpu and period from software events.
169 For the 'wildcard' option if a user selected field is invalid for an
170 event type, a message is displayed to the user that the option is
171 ignored for that type. For example:
173 $ perf script -F comm,tid,trace
174 'trace' not valid for hardware events. Ignoring.
175 'trace' not valid for software events. Ignoring.
177 Alternatively, if the type is given an invalid field is specified it
178 is an error. For example:
180 perf script -v -F sw:comm,tid,trace
181 'trace' not valid for software events.
183 At this point usage is displayed, and perf-script exits.
185 The flags field is synthesized and may have a value when Instruction
186 Trace decoding. The flags are "bcrosyiABEx" which stand for branch,
187 call, return, conditional, system, asynchronous, interrupt,
188 transaction abort, trace begin, trace end, and in transaction,
189 respectively. Known combinations of flags are printed more nicely e.g.
190 "call" for "bc", "return" for "br", "jcc" for "bo", "jmp" for "b",
191 "int" for "bci", "iret" for "bri", "syscall" for "bcs", "sysret" for "brs",
192 "async" for "by", "hw int" for "bcyi", "tx abrt" for "bA", "tr strt" for "bB",
193 "tr end" for "bE". However the "x" flag will be display separately in those
194 cases e.g. "jcc (x)" for a condition branch within a transaction.
196 The callindent field is synthesized and may have a value when
197 Instruction Trace decoding. For calls and returns, it will display the
198 name of the symbol indented with spaces to reflect the stack depth.
200 When doing instruction trace decoding insn and insnlen give the
201 instruction bytes and the instruction length of the current
204 The synth field is used by synthesized events which may be created when
205 Instruction Trace decoding.
207 The ipc (instructions per cycle) field is synthesized and may have a value when
208 Instruction Trace decoding.
210 Finally, a user may not set fields to none for all event types.
211 i.e., -F "" is not allowed.
213 The brstack output includes branch related information with raw addresses using the
214 /v/v/v/v/cycles syntax in the following order:
215 FROM: branch source instruction
216 TO : branch target instruction
217 M/P/-: M=branch target mispredicted or branch direction was mispredicted, P=target predicted or direction predicted, -=not supported
218 X/- : X=branch inside a transactional region, -=not in transaction region or not supported
219 A/- : A=TSX abort entry, -=not aborted region or not supported
222 The brstacksym is identical to brstack, except that the FROM and TO addresses are printed in a symbolic form if possible.
224 When brstackinsn is specified the full assembler sequences of branch sequences for each sample
225 is printed. This is the full execution path leading to the sample. This is only supported when the
226 sample was recorded with perf record -b or -j any.
228 The brstackoff field will print an offset into a specific dso/binary.
230 With the metric option perf script can compute metrics for
231 sampling periods, similar to perf stat. This requires
232 specifying a group with multiple events defining metrics with the :S option
233 for perf record. perf will sample on the first event, and
234 print computed metrics for all the events in the group. Please note
235 that the metric computed is averaged over the whole sampling
236 period (since the last sample), not just for the sample point.
238 For sample events it's possible to display misc field with -F +misc option,
239 following letters are displayed for each bit:
241 PERF_RECORD_MISC_KERNEL K
242 PERF_RECORD_MISC_USER U
243 PERF_RECORD_MISC_HYPERVISOR H
244 PERF_RECORD_MISC_GUEST_KERNEL G
245 PERF_RECORD_MISC_GUEST_USER g
246 PERF_RECORD_MISC_MMAP_DATA* M
247 PERF_RECORD_MISC_COMM_EXEC E
248 PERF_RECORD_MISC_SWITCH_OUT S
249 PERF_RECORD_MISC_SWITCH_OUT_PREEMPT Sp
251 $ perf script -F +misc ...
252 sched-messaging 1414 K 28690.636582: 4590 cycles ...
253 sched-messaging 1407 U 28690.636600: 325620 cycles ...
254 sched-messaging 1414 K 28690.636608: 19473 cycles ...
255 misc field ___________/
264 --symfs=<directory>::
265 Look for files with symbols relative to this directory.
269 When printing symbols do not display call chain.
272 Stop display of callgraph at these symbols
275 --cpu:: Only report samples for the list of CPUs provided. Multiple CPUs can
276 be provided as a comma-separated list with no space: 0,1. Ranges of
277 CPUs are specified with -: 0-2. Default is to report samples on all
282 Only display events for these comms. CSV that understands
283 file://filename entries.
286 Only show events for given process ID (comma separated list).
289 Only show events for given thread ID (comma separated list).
293 Display extended information about the perf.data file. This adds
294 information which may be very large and thus may clutter the display.
295 It currently includes: cpu and numa topology of the host system.
296 It can only be used with the perf script report mode.
299 Try to resolve the path of [kernel.kallsyms]
302 Display task related events (e.g. FORK, COMM, EXIT).
305 Display mmap related events (e.g. MMAP, MMAP2).
307 --show-namespace-events
308 Display namespace events i.e. events of type PERF_RECORD_NAMESPACES.
311 Display context switch events i.e. events of type PERF_RECORD_SWITCH or
312 PERF_RECORD_SWITCH_CPU_WIDE.
315 Display lost events i.e. events of type PERF_RECORD_LOST.
318 Display finished round events i.e. events of type PERF_RECORD_FINISHED_ROUND.
321 Display bpf events i.e. events of type PERF_RECORD_KSYMBOL and PERF_RECORD_BPF_EVENT.
324 Display cgroup events i.e. events of type PERF_RECORD_CGROUP.
326 --show-text-poke-events
327 Display text poke events i.e. events of type PERF_RECORD_TEXT_POKE and
331 Demangle symbol names to human readable form. It's enabled by default,
332 disable with --no-demangle.
335 Demangle kernel symbol names to human readable form (for C++ kernels).
338 Show perf.data header.
341 Show only perf.data header.
344 Options for decoding instruction tracing data. The options are:
346 include::itrace.txt[]
348 To disable decoding entirely, use --no-itrace.
351 Show the full path for source files for srcline output.
354 Set the stack depth limit when parsing the callchain, anything
355 beyond the specified depth will be ignored. This is a trade-off
356 between information loss and faster processing especially for
357 workloads that can have a very long callchain stack.
358 Note that when using the --itrace option the synthesized callchain size
359 will override this value if the synthesized callchain size is bigger.
364 Use 9 decimal places when displaying time (i.e. show the nanoseconds)
368 Don't do ownership validation.
371 Only analyze samples within given time window: <start>,<stop>. Times
372 have the format seconds.nanoseconds. If start is not given (i.e. time
373 string is ',x.y') then analysis starts at the beginning of the file. If
374 stop time is not given (i.e. time string is 'x.y,') then analysis goes
375 to end of file. Multiple ranges can be separated by spaces, which
376 requires the argument to be quoted e.g. --time "1234.567,1234.789 1235,"
378 Also support time percent with multiple time ranges. Time string is
379 'a%/n,b%/m,...' or 'a%-b%,c%-%d,...'.
382 Select the second 10% time slice:
383 perf script --time 10%/2
385 Select from 0% to 10% time slice:
386 perf script --time 0%-10%
388 Select the first and second 10% time slices:
389 perf script --time 10%/1,10%/2
391 Select from 0% to 10% and 30% to 40% slices:
392 perf script --time 0%-10%,30%-40%
395 Set the maximum number of program blocks to print with brstackinsn for
399 Print time stamps relative to trace start.
402 Print time stamps relative to previous event.
405 Create per event files with a "perf.data.EVENT.dump" name instead of
406 printing to stdout, useful, for instance, for generating flamegraphs.
409 If a callgraph address belongs to an inlined function, the inline stack
410 will be printed. Each entry has function name and file/line. Enabled by
411 default, disable with --no-inline.
414 Show instruction stream for intel_pt traces. Combine with --xed to
418 Run xed disassembler on output. Requires installing the xed disassembler.
421 --symbols=symbol[,symbol...]::
422 Only consider the listed symbols. Symbols are typically a name
423 but they may also be hexadecimal address.
425 For example, to select the symbol noploop or the address 0x4007a0:
426 perf script --symbols=noploop,0x4007a0
429 Show call stream for intel_pt traces. The CPUs are interleaved, but
430 can be filtered with -C.
433 Show call and return stream for intel_pt traces.
436 For itrace only show specified functions and their callees for
437 itrace. Multiple functions can be separated by comma.
439 --switch-on EVENT_NAME::
440 Only consider events after this event is found.
442 --switch-off EVENT_NAME::
443 Stop considering events after this event is found.
445 --show-on-off-events::
446 Show the --switch-on/off events too.
449 Show callgraph with stitched LBRs, which may have more complete
450 callgraph. The perf.data file must have been obtained using
451 perf record --call-graph lbr.
452 Disabled by default. In common cases with call stack overflows,
453 it can recreate better call stacks than the default lbr call stack
454 output. But this approach is not full proof. There can be cases
455 where it creates incorrect call stacks from incorrect matches.
456 The known limitations include exception handing such as
457 setjmp/longjmp will have calls/returns not match.
461 linkperf:perf-record[1], linkperf:perf-script-perl[1],
462 linkperf:perf-script-python[1], linkperf:perf-intel-pt[1]