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 Filter sample events using the given shared object file.
103 Refer linkperf:perf-dlfilter[1]
106 Pass 'arg' as an argument to the dlfilter. --dlarg may be repeated
107 to add more arguments.
110 Display a list of available dlfilters. Use with option -v (must come
111 before option --list-dlfilters) to show long descriptions.
114 Force system-wide collection. Scripts run without a <command>
115 normally use -a by default, while scripts run with a <command>
116 normally don't - this option allows the latter to be run in
121 Input file name. (default: perf.data unless stdin is a fifo)
125 Do various checks like samples ordering and lost events.
129 Comma separated list of fields to print. Options are:
130 comm, tid, pid, time, cpu, event, trace, ip, sym, dso, addr, symoff,
131 srcline, period, iregs, uregs, brstack, brstacksym, flags, bpf-output,
132 brstackinsn, brstackinsnlen, brstackoff, callindent, insn, insnlen, synth,
133 phys_addr, metric, misc, srccode, ipc, data_page_size, code_page_size, ins_lat.
134 Field list can be prepended with the type, trace, sw or hw,
135 to indicate to which event type the field list applies.
136 e.g., -F sw:comm,tid,time,ip,sym and -F trace:time,cpu,trace
138 perf script -F <fields>
142 perf script -F trace:<fields> -F sw:<fields> -F hw:<fields>
144 i.e., the specified fields apply to all event types if the type string
147 In addition to overriding fields, it is also possible to add or remove
148 fields from the defaults. For example
152 removes the cpu field and adds the insn field. Adding/removing fields
153 cannot be mixed with normal overriding.
155 The arguments are processed in the order received. A later usage can
156 reset a prior request. e.g.:
158 -F trace: -F comm,tid,time,ip,sym
160 The first -F suppresses trace events (field list is ""), but then the
161 second invocation sets the fields to comm,tid,time,ip,sym. In this case a
162 warning is given to the user:
164 "Overriding previous field request for all events."
166 Alternatively, consider the order:
168 -F comm,tid,time,ip,sym -F trace:
170 The first -F sets the fields for all events and the second -F
171 suppresses trace events. The user is given a warning message about
172 the override, and the result of the above is that only S/W and H/W
173 events are displayed with the given fields.
175 It's possible tp add/remove fields only for specific event type:
179 removes cpu and period from software events.
181 For the 'wildcard' option if a user selected field is invalid for an
182 event type, a message is displayed to the user that the option is
183 ignored for that type. For example:
185 $ perf script -F comm,tid,trace
186 'trace' not valid for hardware events. Ignoring.
187 'trace' not valid for software events. Ignoring.
189 Alternatively, if the type is given an invalid field is specified it
190 is an error. For example:
192 perf script -v -F sw:comm,tid,trace
193 'trace' not valid for software events.
195 At this point usage is displayed, and perf-script exits.
197 The flags field is synthesized and may have a value when Instruction
198 Trace decoding. The flags are "bcrosyiABExghDt" which stand for branch,
199 call, return, conditional, system, asynchronous, interrupt,
200 transaction abort, trace begin, trace end, in transaction, VM-Entry,
201 VM-Exit, interrupt disabled and interrupt disable toggle respectively.
202 Known combinations of flags are printed more nicely e.g.
203 "call" for "bc", "return" for "br", "jcc" for "bo", "jmp" for "b",
204 "int" for "bci", "iret" for "bri", "syscall" for "bcs", "sysret" for "brs",
205 "async" for "by", "hw int" for "bcyi", "tx abrt" for "bA", "tr strt" for "bB",
206 "tr end" for "bE", "vmentry" for "bcg", "vmexit" for "bch".
207 However the "x", "D" and "t" flags will be displayed separately in those
208 cases e.g. "jcc (xD)" for a condition branch within a transaction
209 with interrupts disabled. Note, interrupts becoming disabled is "t",
210 whereas interrupts becoming enabled is "Dt".
212 The callindent field is synthesized and may have a value when
213 Instruction Trace decoding. For calls and returns, it will display the
214 name of the symbol indented with spaces to reflect the stack depth.
216 When doing instruction trace decoding insn and insnlen give the
217 instruction bytes and the instruction length of the current
220 The synth field is used by synthesized events which may be created when
221 Instruction Trace decoding.
223 The ipc (instructions per cycle) field is synthesized and may have a value when
224 Instruction Trace decoding.
226 Finally, a user may not set fields to none for all event types.
227 i.e., -F "" is not allowed.
229 The brstack output includes branch related information with raw addresses using the
230 /v/v/v/v/cycles syntax in the following order:
231 FROM: branch source instruction
232 TO : branch target instruction
233 M/P/-: M=branch target mispredicted or branch direction was mispredicted, P=target predicted or direction predicted, -=not supported
234 X/- : X=branch inside a transactional region, -=not in transaction region or not supported
235 A/- : A=TSX abort entry, -=not aborted region or not supported
238 The brstacksym is identical to brstack, except that the FROM and TO addresses are printed in a symbolic form if possible.
240 When brstackinsn is specified the full assembler sequences of branch sequences for each sample
241 is printed. This is the full execution path leading to the sample. This is only supported when the
242 sample was recorded with perf record -b or -j any.
244 Use brstackinsnlen to print the brstackinsn lenght. For example, you
245 can’t know the next sequential instruction after an unconditional branch unless
246 you calculate that based on its length.
248 The brstackoff field will print an offset into a specific dso/binary.
250 With the metric option perf script can compute metrics for
251 sampling periods, similar to perf stat. This requires
252 specifying a group with multiple events defining metrics with the :S option
253 for perf record. perf will sample on the first event, and
254 print computed metrics for all the events in the group. Please note
255 that the metric computed is averaged over the whole sampling
256 period (since the last sample), not just for the sample point.
258 For sample events it's possible to display misc field with -F +misc option,
259 following letters are displayed for each bit:
261 PERF_RECORD_MISC_KERNEL K
262 PERF_RECORD_MISC_USER U
263 PERF_RECORD_MISC_HYPERVISOR H
264 PERF_RECORD_MISC_GUEST_KERNEL G
265 PERF_RECORD_MISC_GUEST_USER g
266 PERF_RECORD_MISC_MMAP_DATA* M
267 PERF_RECORD_MISC_COMM_EXEC E
268 PERF_RECORD_MISC_SWITCH_OUT S
269 PERF_RECORD_MISC_SWITCH_OUT_PREEMPT Sp
271 $ perf script -F +misc ...
272 sched-messaging 1414 K 28690.636582: 4590 cycles ...
273 sched-messaging 1407 U 28690.636600: 325620 cycles ...
274 sched-messaging 1414 K 28690.636608: 19473 cycles ...
275 misc field ___________/
284 --symfs=<directory>::
285 Look for files with symbols relative to this directory.
289 When printing symbols do not display call chain.
292 Stop display of callgraph at these symbols
295 --cpu:: Only report samples for the list of CPUs provided. Multiple CPUs can
296 be provided as a comma-separated list with no space: 0,1. Ranges of
297 CPUs are specified with -: 0-2. Default is to report samples on all
302 Only display events for these comms. CSV that understands
303 file://filename entries.
306 Only show events for given process ID (comma separated list).
309 Only show events for given thread ID (comma separated list).
313 Display extended information about the perf.data file. This adds
314 information which may be very large and thus may clutter the display.
315 It currently includes: cpu and numa topology of the host system.
316 It can only be used with the perf script report mode.
319 Try to resolve the path of [kernel.kallsyms]
322 Display task related events (e.g. FORK, COMM, EXIT).
325 Display mmap related events (e.g. MMAP, MMAP2).
327 --show-namespace-events
328 Display namespace events i.e. events of type PERF_RECORD_NAMESPACES.
331 Display context switch events i.e. events of type PERF_RECORD_SWITCH or
332 PERF_RECORD_SWITCH_CPU_WIDE.
335 Display lost events i.e. events of type PERF_RECORD_LOST.
338 Display finished round events i.e. events of type PERF_RECORD_FINISHED_ROUND.
341 Display bpf events i.e. events of type PERF_RECORD_KSYMBOL and PERF_RECORD_BPF_EVENT.
344 Display cgroup events i.e. events of type PERF_RECORD_CGROUP.
346 --show-text-poke-events
347 Display text poke events i.e. events of type PERF_RECORD_TEXT_POKE and
351 Demangle symbol names to human readable form. It's enabled by default,
352 disable with --no-demangle.
355 Demangle kernel symbol names to human readable form (for C++ kernels).
358 Show perf.data header.
361 Show only perf.data header.
364 Options for decoding instruction tracing data. The options are:
366 include::itrace.txt[]
368 To disable decoding entirely, use --no-itrace.
371 Show the full path for source files for srcline output.
374 Set the stack depth limit when parsing the callchain, anything
375 beyond the specified depth will be ignored. This is a trade-off
376 between information loss and faster processing especially for
377 workloads that can have a very long callchain stack.
378 Note that when using the --itrace option the synthesized callchain size
379 will override this value if the synthesized callchain size is bigger.
384 Use 9 decimal places when displaying time (i.e. show the nanoseconds)
388 Don't do ownership validation.
391 Only analyze samples within given time window: <start>,<stop>. Times
392 have the format seconds.nanoseconds. If start is not given (i.e. time
393 string is ',x.y') then analysis starts at the beginning of the file. If
394 stop time is not given (i.e. time string is 'x.y,') then analysis goes
395 to end of file. Multiple ranges can be separated by spaces, which
396 requires the argument to be quoted e.g. --time "1234.567,1234.789 1235,"
398 Also support time percent with multiple time ranges. Time string is
399 'a%/n,b%/m,...' or 'a%-b%,c%-%d,...'.
402 Select the second 10% time slice:
403 perf script --time 10%/2
405 Select from 0% to 10% time slice:
406 perf script --time 0%-10%
408 Select the first and second 10% time slices:
409 perf script --time 10%/1,10%/2
411 Select from 0% to 10% and 30% to 40% slices:
412 perf script --time 0%-10%,30%-40%
415 Set the maximum number of program blocks to print with brstackinsn for
419 Print time stamps relative to trace start.
422 Print time stamps relative to previous event.
425 Create per event files with a "perf.data.EVENT.dump" name instead of
426 printing to stdout, useful, for instance, for generating flamegraphs.
429 If a callgraph address belongs to an inlined function, the inline stack
430 will be printed. Each entry has function name and file/line. Enabled by
431 default, disable with --no-inline.
434 Show instruction stream for intel_pt traces. Combine with --xed to
438 Run xed disassembler on output. Requires installing the xed disassembler.
441 --symbols=symbol[,symbol...]::
442 Only consider the listed symbols. Symbols are typically a name
443 but they may also be hexadecimal address.
445 The hexadecimal address may be the start address of a symbol or
446 any other address to filter the trace records
448 For example, to select the symbol noploop or the address 0x4007a0:
449 perf script --symbols=noploop,0x4007a0
451 Support filtering trace records by symbol name, start address of
452 symbol, any hexadecimal address and address range.
454 The comparison order is:
456 1. symbol name comparison
457 2. symbol start address comparison.
458 3. any hexadecimal address comparison.
459 4. address range comparison (see --addr-range).
462 Use with -S or --symbols to list traced records within address range.
464 For example, to list the traced records within the address range
465 [0x4007a0, 0x0x4007a9]:
466 perf script -S 0x4007a0 --addr-range 10
469 Only consider symbols in these DSOs.
472 Show call stream for intel_pt traces. The CPUs are interleaved, but
473 can be filtered with -C.
476 Show call and return stream for intel_pt traces.
479 For itrace only show specified functions and their callees for
480 itrace. Multiple functions can be separated by comma.
482 --switch-on EVENT_NAME::
483 Only consider events after this event is found.
485 --switch-off EVENT_NAME::
486 Stop considering events after this event is found.
488 --show-on-off-events::
489 Show the --switch-on/off events too.
492 Show callgraph with stitched LBRs, which may have more complete
493 callgraph. The perf.data file must have been obtained using
494 perf record --call-graph lbr.
495 Disabled by default. In common cases with call stack overflows,
496 it can recreate better call stacks than the default lbr call stack
497 output. But this approach is not full proof. There can be cases
498 where it creates incorrect call stacks from incorrect matches.
499 The known limitations include exception handing such as
500 setjmp/longjmp will have calls/returns not match.
504 linkperf:perf-record[1], linkperf:perf-script-perl[1],
505 linkperf:perf-script-python[1], linkperf:perf-intel-pt[1],
506 linkperf:perf-dlfilter[1]