1PERF-SCRIPT(1) perf Manual PERF-SCRIPT(1)
2
6 perf-script - Read perf.data (created by perf record) and display trace
7 output
8
10 perf script [<options>]
11 perf script [<options>] record <script> [<record-options>] <command>
12 perf script [<options>] report <script> [script-args]
13 perf script [<options>] <script> <required-script-args> [<record-options>] <command>
14 perf script [<options>] <top-script> [script-args]
15
17 This command reads the input file and displays the trace recorded.
18 There are several variants of perf script:
20 'perf script' to see a detailed trace of the workload that was
22 recorded.
23 You can also run a set of pre-canned scripts that aggregate and
25 summarize the raw trace data in various ways (the list of scripts is
26 available via 'perf script -l'). The following variants allow you to
27 record and run those scripts:
28 'perf script record <script> <command>' to record the events required
30 for 'perf script report'. <script> is the name displayed in the
31 output of 'perf script --list' i.e. the actual script name minus any
32 language extension. If <command> is not specified, the events are
33 recorded using the -a (system-wide) 'perf record' option.
34 'perf script report <script> [args]' to run and display the results
36 of <script>. <script> is the name displayed in the output of 'perf
37 script --list' i.e. the actual script name minus any language
38 extension. The perf.data output from a previous run of 'perf script
39 record <script>' is used and should be present for this command to
40 succeed. [args] refers to the (mainly optional) args expected by
41 the script.
42 'perf script <script> <required-script-args> <command>' to both
44 record the events required for <script> and to run the <script>
45 using 'live-mode' i.e. without writing anything to disk. <script>
46 is the name displayed in the output of 'perf script --list' i.e. the
47 actual script name minus any language extension. If <command> is
48 not specified, the events are recorded using the -a (system-wide)
49 'perf record' option. If <script> has any required args, they
50 should be specified before <command>. This mode doesn't allow for
51 optional script args to be specified; if optional script args are
52 desired, they can be specified using separate 'perf script record'
53 and 'perf script report' commands, with the stdout of the record step
54 piped to the stdin of the report script, using the '-o -' and '-i -'
55 options of the corresponding commands.
56 'perf script <top-script>' to both record the events required for
58 <top-script> and to run the <top-script> using 'live-mode'
59 i.e. without writing anything to disk. <top-script> is the name
60 displayed in the output of 'perf script --list' i.e. the actual
61 script name minus any language extension; a <top-script> is defined
62 as any script name ending with the string 'top'.
63 [<record-options>] can be passed to the record steps of 'perf script
65 record' and 'live-mode' variants; this isn't possible however for
66 <top-script> 'live-mode' or 'perf script report' variants.
67 See the 'SEE ALSO' section for links to language-specific
69 information on how to write and run your own trace scripts.
70
72 <command>...
73 Any command you can specify in a shell.
74 -D, --dump-raw-trace=
76 Display verbose dump of the trace data.
77 -L, --Latency=
79 Show latency attributes (irqs/preemption disabled, etc).
80 -l, --list=
82 Display a list of available trace scripts.
83 -s [lang], --script=
85 Process trace data with the given script ([lang]:script[.ext]). If
86 the string lang is specified in place of a script name, a list of
87 supported languages will be displayed instead.
88 -g, --gen-script=
90 Generate perf-script.[ext] starter script for given language, using
91 current perf.data.
92 --dlfilter=<file>
94 Filter sample events using the given shared object file. Refer
95 perf-dlfilter(1)
96 --dlarg=<arg>
98 Pass arg as an argument to the dlfilter. --dlarg may be repeated to
99 add more arguments.
100 --list-dlfilters
102 Display a list of available dlfilters. Use with option -v (must
103 come before option --list-dlfilters) to show long descriptions.
104 -a
106 Force system-wide collection. Scripts run without a <command>
107 normally use -a by default, while scripts run with a <command>
108 normally don’t - this option allows the latter to be run in
109 system-wide mode.
110 -i, --input=
112 Input file name. (default: perf.data unless stdin is a fifo)
113 -d, --debug-mode
115 Do various checks like samples ordering and lost events.
116 -F, --fields
118 Comma separated list of fields to print. Options are: comm, tid,
119 pid, time, cpu, event, trace, ip, sym, dso, addr, symoff, srcline,
120 period, iregs, uregs, brstack, brstacksym, flags, bpf-output,
121 brstackinsn, brstackoff, callindent, insn, insnlen, synth,
122 phys_addr, metric, misc, srccode, ipc, data_page_size,
123 code_page_size. Field list can be prepended with the type, trace,
124 sw or hw, to indicate to which event type the field list applies.
125 e.g., -F sw:comm,tid,time,ip,sym and -F trace:time,cpu,trace
126 perf script -F <fields>
128 is equivalent to:
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
134 is not given.
135 In addition to overriding fields, it is also possible to add or remove
137 fields from the defaults. For example
138 -F -cpu,+insn
140 removes the cpu field and adds the insn field. Adding/removing fields
142 cannot be mixed with normal overriding.
143 The arguments are processed in the order received. A later usage can
145 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
150 second invocation sets the fields to comm,tid,time,ip,sym. In this case a
151 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
160 suppresses trace events. The user is given a warning message about
161 the override, and the result of the above is that only S/W and H/W
162 events are displayed with the given fields.
163 It's possible tp add/remove fields only for specific event type:
165 -Fsw:-cpu,-period
167 removes cpu and period from software events.
169 For the 'wildcard' option if a user selected field is invalid for an
171 event type, a message is displayed to the user that the option is
172 ignored for that type. For example:
173 $ perf script -F comm,tid,trace
175 'trace' not valid for hardware events. Ignoring.
176 'trace' not valid for software events. Ignoring.
177 Alternatively, if the type is given an invalid field is specified it
179 is an error. For example:
180 perf script -v -F sw:comm,tid,trace
182 '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
187 Trace decoding. The flags are "bcrosyiABExgh" which stand for branch,
188 call, return, conditional, system, asynchronous, interrupt,
189 transaction abort, trace begin, trace end, in transaction, VM-Entry, and VM-Exit
190 respectively. Known combinations of flags are printed more nicely e.g.
191 "call" for "bc", "return" for "br", "jcc" for "bo", "jmp" for "b",
192 "int" for "bci", "iret" for "bri", "syscall" for "bcs", "sysret" for "brs",
193 "async" for "by", "hw int" for "bcyi", "tx abrt" for "bA", "tr strt" for "bB",
194 "tr end" for "bE", "vmentry" for "bcg", "vmexit" for "bch".
195 However the "x" flag will be displayed separately in those
196 cases e.g. "jcc (x)" for a condition branch within a transaction.
197 The callindent field is synthesized and may have a value when
199 Instruction Trace decoding. For calls and returns, it will display the
200 name of the symbol indented with spaces to reflect the stack depth.
201 When doing instruction trace decoding insn and insnlen give the
203 instruction bytes and the instruction length of the current
204 instruction.
205 The synth field is used by synthesized events which may be created when
207 Instruction Trace decoding.
208 The ipc (instructions per cycle) field is synthesized and may have a value when
210 Instruction Trace decoding.
211 Finally, a user may not set fields to none for all event types.
213 i.e., -F "" is not allowed.
214 The brstack output includes branch related information with raw addresses using the
216 /v/v/v/v/cycles syntax in the following order:
217 FROM: branch source instruction
218 TO : branch target instruction
219 M/P/-: M=branch target mispredicted or branch direction was mispredicted, P=target predicted or direction predicted, -=not supported
220 X/- : X=branch inside a transactional region, -=not in transaction region or not supported
221 A/- : A=TSX abort entry, -=not aborted region or not supported
222 cycles
223 The brstacksym is identical to brstack, except that the FROM and TO addresses are printed in a symbolic form if possible.
225 When brstackinsn is specified the full assembler sequences of branch sequences for each sample
227 is printed. This is the full execution path leading to the sample. This is only supported when the
228 sample was recorded with perf record -b or -j any.
229 The brstackoff field will print an offset into a specific dso/binary.
231 With the metric option perf script can compute metrics for
233 sampling periods, similar to perf stat. This requires
234 specifying a group with multiple events defining metrics with the :S option
235 for perf record. perf will sample on the first event, and
236 print computed metrics for all the events in the group. Please note
237 that the metric computed is averaged over the whole sampling
238 period (since the last sample), not just for the sample point.
239 For sample events it's possible to display misc field with -F +misc option,
241 following letters are displayed for each bit:
242 PERF_RECORD_MISC_KERNEL K
244 PERF_RECORD_MISC_USER U
245 PERF_RECORD_MISC_HYPERVISOR H
246 PERF_RECORD_MISC_GUEST_KERNEL G
247 PERF_RECORD_MISC_GUEST_USER g
248 PERF_RECORD_MISC_MMAP_DATA* M
249 PERF_RECORD_MISC_COMM_EXEC E
250 PERF_RECORD_MISC_SWITCH_OUT S
251 PERF_RECORD_MISC_SWITCH_OUT_PREEMPT Sp
252 $ perf script -F +misc ...
254 sched-messaging 1414 K 28690.636582: 4590 cycles ...
255 sched-messaging 1407 U 28690.636600: 325620 cycles ...
256 sched-messaging 1414 K 28690.636608: 19473 cycles ...
257 misc field ___________/
258 -k, --vmlinux=<file>
260 vmlinux pathname
261 --kallsyms=<file>
263 kallsyms pathname
264 --symfs=<directory>
266 Look for files with symbols relative to this directory.
267 -G, --hide-call-graph
269 When printing symbols do not display call chain.
270 --stop-bt
272 Stop display of callgraph at these symbols
273 -C, --cpu
275 Only report samples for the list of CPUs provided. Multiple CPUs
276 can be provided as a comma-separated list with no space: 0,1.
277 Ranges of CPUs are specified with -: 0-2. Default is to report
278 samples on all CPUs.
279 -c, --comms=
281 Only display events for these comms. CSV that understands
282 file://filename entries.
283 --pid=
285 Only show events for given process ID (comma separated list).
286 --tid=
288 Only show events for given thread ID (comma separated list).
289 -I, --show-info
291 Display extended information about the perf.data file. This adds
292 information which may be very large and thus may clutter the
293 display. It currently includes: cpu and numa topology of the host
294 system. It can only be used with the perf script report mode.
295 --show-kernel-path
297 Try to resolve the path of [kernel.kallsyms]
298 --show-task-events Display task related events (e.g. FORK, COMM, EXIT).
300 --show-mmap-events Display mmap related events (e.g. MMAP, MMAP2).
302 --show-namespace-events Display namespace events i.e. events of type
304 PERF_RECORD_NAMESPACES.
305 --show-switch-events Display context switch events i.e. events of type
307 PERF_RECORD_SWITCH or PERF_RECORD_SWITCH_CPU_WIDE.
308 --show-lost-events Display lost events i.e. events of type
310 PERF_RECORD_LOST.
311 --show-round-events Display finished round events i.e. events of type
313 PERF_RECORD_FINISHED_ROUND.
314 --show-bpf-events Display bpf events i.e. events of type
316 PERF_RECORD_KSYMBOL and PERF_RECORD_BPF_EVENT.
317 --show-cgroup-events Display cgroup events i.e. events of type
319 PERF_RECORD_CGROUP.
320 --show-text-poke-events Display text poke events i.e. events of type
322 PERF_RECORD_TEXT_POKE and PERF_RECORD_KSYMBOL.
323 --demangle
325 Demangle symbol names to human readable form. It’s enabled by
326 default, disable with --no-demangle.
327 --demangle-kernel
329 Demangle kernel symbol names to human readable form (for C++
330 kernels).
331 --header Show perf.data header.
333 --header-only Show only perf.data header.
335 --itrace
337 Options for decoding instruction tracing data. The options are:
338 i synthesize instructions events
340 b synthesize branches events (branch misses for Arm SPE)
341 c synthesize branches events (calls only)
342 r synthesize branches events (returns only)
343 x synthesize transactions events
344 w synthesize ptwrite events
345 p synthesize power events (incl. PSB events for Intel PT)
346 o synthesize other events recorded due to the use
347 of aux-output (refer to perf record)
348 e synthesize error events
349 d create a debug log
350 f synthesize first level cache events
351 m synthesize last level cache events
352 M synthesize memory events
353 t synthesize TLB events
354 a synthesize remote access events
355 g synthesize a call chain (use with i or x)
356 G synthesize a call chain on existing event records
357 l synthesize last branch entries (use with i or x)
358 L synthesize last branch entries on existing event records
359 s skip initial number of events
360 q quicker (less detailed) decoding
361 Z prefer to ignore timestamps (so-called "timeless" decoding)
362 The default is all events i.e. the same as --itrace=ibxwpe,
364 except for perf script where it is --itrace=ce
365 In addition, the period (default 100000, except for perf script where it is 1)
367 for instructions events can be specified in units of:
368 i instructions
370 t ticks
371 ms milliseconds
372 us microseconds
373 ns nanoseconds (default)
374 Also the call chain size (default 16, max. 1024) for instructions or
376 transactions events can be specified.
377 Also the number of last branch entries (default 64, max. 1024) for
379 instructions or transactions events can be specified.
380 Similar to options g and l, size may also be specified for options G and L.
382 On x86, note that G and L work poorly when data has been recorded with
383 large PEBS. Refer linkperf:perf-intel-pt[1] man page for details.
384 It is also possible to skip events generated (instructions, branches, transactions,
386 ptwrite, power) at the beginning. This is useful to ignore initialization code.
387 --itrace=i0nss1000000
389 skips the first million instructions.
391 The 'e' option may be followed by flags which affect what errors will or
393 will not be reported. Each flag must be preceded by either '+' or '-'.
394 The flags are:
395 o overflow
396 l trace data lost
397 If supported, the 'd' option may be followed by flags which affect what
399 debug messages will or will not be logged. Each flag must be preceded
400 by either '+' or '-'. The flags are:
401 a all perf events
402 If supported, the 'q' option may be repeated to increase the effect.
404 To disable decoding entirely, use --no-itrace.
406 --full-source-path
408 Show the full path for source files for srcline output.
409 --max-stack
411 Set the stack depth limit when parsing the callchain, anything
412 beyond the specified depth will be ignored. This is a trade-off
413 between information loss and faster processing especially for
414 workloads that can have a very long callchain stack. Note that when
415 using the --itrace option the synthesized callchain size will
416 override this value if the synthesized callchain size is bigger.
417 Default: 127
419 --ns
421 Use 9 decimal places when displaying time (i.e. show the
422 nanoseconds)
423 -f, --force
425 Don’t do ownership validation.
426 --time
428 Only analyze samples within given time window: <start>,<stop>.
429 Times have the format seconds.nanoseconds. If start is not given
430 (i.e. time string is ,x.y) then analysis starts at the beginning of
431 the file. If stop time is not given (i.e. time string is x.y,) then
432 analysis goes to end of file. Multiple ranges can be separated by
433 spaces, which requires the argument to be quoted e.g. --time
434 "1234.567,1234.789 1235,"
435 Also support time percent with multiple time ranges. Time string is
437 'a%/n,b%/m,...' or 'a%-b%,c%-%d,...'.
438 For example:
440 Select the second 10% time slice:
441 perf script --time 10%/2
442 Select from 0% to 10% time slice:
444 perf script --time 0%-10%
445 Select the first and second 10% time slices:
447 perf script --time 10%/1,10%/2
448 Select from 0% to 10% and 30% to 40% slices:
450 perf script --time 0%-10%,30%-40%
451 --max-blocks
453 Set the maximum number of program blocks to print with brstackinsn
454 for each sample.
455 --reltime
457 Print time stamps relative to trace start.
458 --deltatime
460 Print time stamps relative to previous event.
461 --per-event-dump
463 Create per event files with a "perf.data.EVENT.dump" name instead
464 of printing to stdout, useful, for instance, for generating
465 flamegraphs.
466 --inline
468 If a callgraph address belongs to an inlined function, the inline
469 stack will be printed. Each entry has function name and file/line.
470 Enabled by default, disable with --no-inline.
471 --insn-trace
473 Show instruction stream for intel_pt traces. Combine with --xed to
474 show disassembly.
475 --xed
477 Run xed disassembler on output. Requires installing the xed
478 disassembler.
479 -S, --symbols=symbol[,symbol...]
481 Only consider the listed symbols. Symbols are typically a name but
482 they may also be hexadecimal address.
483 The hexadecimal address may be the start address of a symbol or
485 any other address to filter the trace records
486 For example, to select the symbol noploop or the address 0x4007a0:
488 perf script --symbols=noploop,0x4007a0
489 Support filtering trace records by symbol name, start address of
491 symbol, any hexadecimal address and address range.
492 The comparison order is:
494 1. symbol name comparison
496 2. symbol start address comparison.
498 3. any hexadecimal address comparison.
500 4. address range comparison (see --addr-range).
502 --addr-range
504 Use with -S or --symbols to list traced records within address
505 range.
506 For example, to list the traced records within the address range
508 [0x4007a0, 0x0x4007a9]:
509 perf script -S 0x4007a0 --addr-range 10
510 --dsos=
512 Only consider symbols in these DSOs.
513 --call-trace
515 Show call stream for intel_pt traces. The CPUs are interleaved, but
516 can be filtered with -C.
517 --call-ret-trace
519 Show call and return stream for intel_pt traces.
520 --graph-function
522 For itrace only show specified functions and their callees for
523 itrace. Multiple functions can be separated by comma.
524 --switch-on EVENT_NAME
526 Only consider events after this event is found.
527 --switch-off EVENT_NAME
529 Stop considering events after this event is found.
530 --show-on-off-events
532 Show the --switch-on/off events too.
533 --stitch-lbr
535 Show callgraph with stitched LBRs, which may have more complete
536 callgraph. The perf.data file must have been obtained using perf
537 record --call-graph lbr. Disabled by default. In common cases with
538 call stack overflows, it can recreate better call stacks than the
539 default lbr call stack output. But this approach is not full proof.
540 There can be cases where it creates incorrect call stacks from
541 incorrect matches. The known limitations include exception handing
542 such as setjmp/longjmp will have calls/returns not match.
543
545 perf-record(1), perf-script-perl(1), perf-script-python(1), perf-intel-
546 pt(1), perf-dlfilter(1)
547perf 11/22/2021 PERF-SCRIPT(1)