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, brstackinsn,
120 brstackoff, callindent, insn, insnlen, synth, phys_addr, metric, misc, srccode, ipc.
121 Field list can be prepended with the type, trace, sw or hw,
122 to indicate to which event type the field list applies.
123 e.g., -F sw:comm,tid,time,ip,sym and -F trace:time,cpu,trace
125 perf script -F <fields>
129 perf script -F trace:<fields> -F sw:<fields> -F hw:<fields>
131 i.e., the specified fields apply to all event types if the type string
134 In addition to overriding fields, it is also possible to add or remove
135 fields from the defaults. For example
139 removes the cpu field and adds the insn field. Adding/removing fields
140 cannot be mixed with normal overriding.
142 The arguments are processed in the order received. A later usage can
143 reset a prior request. e.g.:
145 -F trace: -F comm,tid,time,ip,sym
147 The first -F suppresses trace events (field list is ""), but then the
148 second invocation sets the fields to comm,tid,time,ip,sym. In this case a
149 warning is given to the user:
151 "Overriding previous field request for all events."
153 Alternatively, consider the order:
155 -F comm,tid,time,ip,sym -F trace:
157 The first -F sets the fields for all events and the second -F
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.
162 It's possible tp add/remove fields only for specific event type:
166 removes cpu and period from software events.
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:
172 $ perf script -F comm,tid,trace
173 'trace' not valid for hardware events. Ignoring.
174 'trace' not valid for software events. Ignoring.
176 Alternatively, if the type is given an invalid field is specified it
177 is an error. For example:
179 perf script -v -F sw:comm,tid,trace
180 'trace' not valid for software events.
182 At this point usage is displayed, and perf-script exits.
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,
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.
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.
199 When doing instruction trace decoding insn and insnlen give the
200 instruction bytes and the instruction length of the current
203 The synth field is used by synthesized events which may be created when
204 Instruction Trace decoding.
206 The ipc (instructions per cycle) field is synthesized and may have a value when
207 Instruction Trace decoding.
209 Finally, a user may not set fields to none for all event types.
210 i.e., -F "" is not allowed.
212 The brstack output includes branch related information with raw addresses using the
213 /v/v/v/v/cycles syntax in the following order:
214 FROM: branch source instruction
215 TO : branch target instruction
216 M/P/-: M=branch target mispredicted or branch direction was mispredicted, P=target predicted or direction predicted, -=not supported
217 X/- : X=branch inside a transactional region, -=not in transaction region or not supported
218 A/- : A=TSX abort entry, -=not aborted region or not supported
221 The brstacksym is identical to brstack, except that the FROM and TO addresses are printed in a symbolic form if possible.
223 When brstackinsn is specified the full assembler sequences of branch sequences for each sample
224 is printed. This is the full execution path leading to the sample. This is only supported when the
225 sample was recorded with perf record -b or -j any.
227 The brstackoff field will print an offset into a specific dso/binary.
229 With the metric option perf script can compute metrics for
230 sampling periods, similar to perf stat. This requires
231 specifying a group with multiple events defining metrics with the :S option
232 for perf record. perf will sample on the first event, and
233 print computed metrics for all the events in the group. Please note
234 that the metric computed is averaged over the whole sampling
235 period (since the last sample), not just for the sample point.
237 For sample events it's possible to display misc field with -F +misc option,
238 following letters are displayed for each bit:
240 PERF_RECORD_MISC_KERNEL K
241 PERF_RECORD_MISC_USER U
242 PERF_RECORD_MISC_HYPERVISOR H
243 PERF_RECORD_MISC_GUEST_KERNEL G
244 PERF_RECORD_MISC_GUEST_USER g
245 PERF_RECORD_MISC_MMAP_DATA* M
246 PERF_RECORD_MISC_COMM_EXEC E
247 PERF_RECORD_MISC_SWITCH_OUT S
248 PERF_RECORD_MISC_SWITCH_OUT_PREEMPT Sp
250 $ perf script -F +misc ...
251 sched-messaging 1414 K 28690.636582: 4590 cycles ...
252 sched-messaging 1407 U 28690.636600: 325620 cycles ...
253 sched-messaging 1414 K 28690.636608: 19473 cycles ...
254 misc field ___________/
263 --symfs=<directory>::
264 Look for files with symbols relative to this directory.
268 When printing symbols do not display call chain.
271 Stop display of callgraph at these symbols
274 --cpu:: Only report samples for the list of CPUs provided. Multiple CPUs can
275 be provided as a comma-separated list with no space: 0,1. Ranges of
276 CPUs are specified with -: 0-2. Default is to report samples on all
281 Only display events for these comms. CSV that understands
282 file://filename entries.
285 Only show events for given process ID (comma separated list).
288 Only show events for given thread ID (comma separated list).
292 Display extended information about the perf.data file. This adds
293 information which may be very large and thus may clutter the display.
294 It currently includes: cpu and numa topology of the host system.
295 It can only be used with the perf script report mode.
298 Try to resolve the path of [kernel.kallsyms]
301 Display task related events (e.g. FORK, COMM, EXIT).
304 Display mmap related events (e.g. MMAP, MMAP2).
306 --show-namespace-events
307 Display namespace events i.e. events of type PERF_RECORD_NAMESPACES.
310 Display context switch events i.e. events of type PERF_RECORD_SWITCH or
311 PERF_RECORD_SWITCH_CPU_WIDE.
314 Display lost events i.e. events of type PERF_RECORD_LOST.
317 Display finished round events i.e. events of type PERF_RECORD_FINISHED_ROUND.
320 Display bpf events i.e. events of type PERF_RECORD_KSYMBOL and PERF_RECORD_BPF_EVENT.
323 Display cgroup events i.e. events of type PERF_RECORD_CGROUP.
325 --show-text-poke-events
326 Display text poke events i.e. events of type PERF_RECORD_TEXT_POKE and
330 Demangle symbol names to human readable form. It's enabled by default,
331 disable with --no-demangle.
334 Demangle kernel symbol names to human readable form (for C++ kernels).
337 Show perf.data header.
340 Show only perf.data header.
343 Options for decoding instruction tracing data. The options are:
345 include::itrace.txt[]
347 To disable decoding entirely, use --no-itrace.
350 Show the full path for source files for srcline output.
353 Set the stack depth limit when parsing the callchain, anything
354 beyond the specified depth will be ignored. This is a trade-off
355 between information loss and faster processing especially for
356 workloads that can have a very long callchain stack.
357 Note that when using the --itrace option the synthesized callchain size
358 will override this value if the synthesized callchain size is bigger.
363 Use 9 decimal places when displaying time (i.e. show the nanoseconds)
367 Don't do ownership validation.
370 Only analyze samples within given time window: <start>,<stop>. Times
371 have the format seconds.nanoseconds. If start is not given (i.e. time
372 string is ',x.y') then analysis starts at the beginning of the file. If
373 stop time is not given (i.e. time string is 'x.y,') then analysis goes
374 to end of file. Multiple ranges can be separated by spaces, which
375 requires the argument to be quoted e.g. --time "1234.567,1234.789 1235,"
377 Also support time percent with multiple time ranges. Time string is
378 'a%/n,b%/m,...' or 'a%-b%,c%-%d,...'.
381 Select the second 10% time slice:
382 perf script --time 10%/2
384 Select from 0% to 10% time slice:
385 perf script --time 0%-10%
387 Select the first and second 10% time slices:
388 perf script --time 10%/1,10%/2
390 Select from 0% to 10% and 30% to 40% slices:
391 perf script --time 0%-10%,30%-40%
394 Set the maximum number of program blocks to print with brstackinsn for
398 Print time stamps relative to trace start.
401 Print time stamps relative to previous event.
404 Create per event files with a "perf.data.EVENT.dump" name instead of
405 printing to stdout, useful, for instance, for generating flamegraphs.
408 If a callgraph address belongs to an inlined function, the inline stack
409 will be printed. Each entry has function name and file/line. Enabled by
410 default, disable with --no-inline.
413 Show instruction stream for intel_pt traces. Combine with --xed to
417 Run xed disassembler on output. Requires installing the xed disassembler.
420 --symbols=symbol[,symbol...]::
421 Only consider the listed symbols. Symbols are typically a name
422 but they may also be hexadecimal address.
424 For example, to select the symbol noploop or the address 0x4007a0:
425 perf script --symbols=noploop,0x4007a0
428 Show call stream for intel_pt traces. The CPUs are interleaved, but
429 can be filtered with -C.
432 Show call and return stream for intel_pt traces.
435 For itrace only show specified functions and their callees for
436 itrace. Multiple functions can be separated by comma.
438 --switch-on EVENT_NAME::
439 Only consider events after this event is found.
441 --switch-off EVENT_NAME::
442 Stop considering events after this event is found.
444 --show-on-off-events::
445 Show the --switch-on/off events too.
448 Show callgraph with stitched LBRs, which may have more complete
449 callgraph. The perf.data file must have been obtained using
450 perf record --call-graph lbr.
451 Disabled by default. In common cases with call stack overflows,
452 it can recreate better call stacks than the default lbr call stack
453 output. But this approach is not full proof. There can be cases
454 where it creates incorrect call stacks from incorrect matches.
455 The known limitations include exception handing such as
456 setjmp/longjmp will have calls/returns not match.
460 linkperf:perf-record[1], linkperf:perf-script-perl[1],
461 linkperf:perf-script-python[1], linkperf:perf-intel-pt[1]