xref: /linux/tools/perf/Documentation/perf-record.txt (revision 0883c2c06fb5bcf5b9e008270827e63c09a88c1e)
1perf-record(1)
2==============
3
4NAME
5----
6perf-record - Run a command and record its profile into perf.data
7
8SYNOPSIS
9--------
10[verse]
11'perf record' [-e <EVENT> | --event=EVENT] [-l] [-a] <command>
12'perf record' [-e <EVENT> | --event=EVENT] [-l] [-a] -- <command> [<options>]
13
14DESCRIPTION
15-----------
16This command runs a command and gathers a performance counter profile
17from it, into perf.data - without displaying anything.
18
19This file can then be inspected later on, using 'perf report'.
20
21
22OPTIONS
23-------
24<command>...::
25	Any command you can specify in a shell.
26
27-e::
28--event=::
29	Select the PMU event. Selection can be:
30
31        - a symbolic event name	(use 'perf list' to list all events)
32
33        - a raw PMU event (eventsel+umask) in the form of rNNN where NNN is a
34	  hexadecimal event descriptor.
35
36	- a symbolically formed PMU event like 'pmu/param1=0x3,param2/' where
37	  'param1', 'param2', etc are defined as formats for the PMU in
38	  /sys/bus/event_sources/devices/<pmu>/format/*.
39
40	- a symbolically formed event like 'pmu/config=M,config1=N,config3=K/'
41
42          where M, N, K are numbers (in decimal, hex, octal format). Acceptable
43          values for each of 'config', 'config1' and 'config2' are defined by
44          corresponding entries in /sys/bus/event_sources/devices/<pmu>/format/*
45          param1 and param2 are defined as formats for the PMU in:
46          /sys/bus/event_sources/devices/<pmu>/format/*
47
48	  There are also some params which are not defined in .../<pmu>/format/*.
49	  These params can be used to overload default config values per event.
50	  Here is a list of the params.
51	  - 'period': Set event sampling period
52	  - 'freq': Set event sampling frequency
53	  - 'time': Disable/enable time stamping. Acceptable values are 1 for
54		    enabling time stamping. 0 for disabling time stamping.
55		    The default is 1.
56	  - 'call-graph': Disable/enable callgraph. Acceptable str are "fp" for
57			 FP mode, "dwarf" for DWARF mode, "lbr" for LBR mode and
58			 "no" for disable callgraph.
59	  - 'stack-size': user stack size for dwarf mode
60	  Note: If user explicitly sets options which conflict with the params,
61	  the value set by the params will be overridden.
62
63        - a hardware breakpoint event in the form of '\mem:addr[/len][:access]'
64          where addr is the address in memory you want to break in.
65          Access is the memory access type (read, write, execute) it can
66          be passed as follows: '\mem:addr[:[r][w][x]]'. len is the range,
67          number of bytes from specified addr, which the breakpoint will cover.
68          If you want to profile read-write accesses in 0x1000, just set
69          'mem:0x1000:rw'.
70          If you want to profile write accesses in [0x1000~1008), just set
71          'mem:0x1000/8:w'.
72
73	- a group of events surrounded by a pair of brace ("{event1,event2,...}").
74	  Each event is separated by commas and the group should be quoted to
75	  prevent the shell interpretation.  You also need to use --group on
76	  "perf report" to view group events together.
77
78--filter=<filter>::
79        Event filter. This option should follow a event selector (-e) which
80	selects tracepoint event(s). Multiple '--filter' options are combined
81	using '&&'.
82
83--exclude-perf::
84	Don't record events issued by perf itself. This option should follow
85	a event selector (-e) which selects tracepoint event(s). It adds a
86	filter expression 'common_pid != $PERFPID' to filters. If other
87	'--filter' exists, the new filter expression will be combined with
88	them by '&&'.
89
90-a::
91--all-cpus::
92        System-wide collection from all CPUs.
93
94-p::
95--pid=::
96	Record events on existing process ID (comma separated list).
97
98-t::
99--tid=::
100        Record events on existing thread ID (comma separated list).
101        This option also disables inheritance by default.  Enable it by adding
102        --inherit.
103
104-u::
105--uid=::
106        Record events in threads owned by uid. Name or number.
107
108-r::
109--realtime=::
110	Collect data with this RT SCHED_FIFO priority.
111
112--no-buffering::
113	Collect data without buffering.
114
115-c::
116--count=::
117	Event period to sample.
118
119-o::
120--output=::
121	Output file name.
122
123-i::
124--no-inherit::
125	Child tasks do not inherit counters.
126-F::
127--freq=::
128	Profile at this frequency.
129
130-m::
131--mmap-pages=::
132	Number of mmap data pages (must be a power of two) or size
133	specification with appended unit character - B/K/M/G. The
134	size is rounded up to have nearest pages power of two value.
135	Also, by adding a comma, the number of mmap pages for AUX
136	area tracing can be specified.
137
138--group::
139	Put all events in a single event group.  This precedes the --event
140	option and remains only for backward compatibility.  See --event.
141
142-g::
143	Enables call-graph (stack chain/backtrace) recording.
144
145--call-graph::
146	Setup and enable call-graph (stack chain/backtrace) recording,
147	implies -g.  Default is "fp".
148
149	Allows specifying "fp" (frame pointer) or "dwarf"
150	(DWARF's CFI - Call Frame Information) or "lbr"
151	(Hardware Last Branch Record facility) as the method to collect
152	the information used to show the call graphs.
153
154	In some systems, where binaries are build with gcc
155	--fomit-frame-pointer, using the "fp" method will produce bogus
156	call graphs, using "dwarf", if available (perf tools linked to
157	the libunwind or libdw library) should be used instead.
158	Using the "lbr" method doesn't require any compiler options. It
159	will produce call graphs from the hardware LBR registers. The
160	main limition is that it is only available on new Intel
161	platforms, such as Haswell. It can only get user call chain. It
162	doesn't work with branch stack sampling at the same time.
163
164	When "dwarf" recording is used, perf also records (user) stack dump
165	when sampled.  Default size of the stack dump is 8192 (bytes).
166	User can change the size by passing the size after comma like
167	"--call-graph dwarf,4096".
168
169-q::
170--quiet::
171	Don't print any message, useful for scripting.
172
173-v::
174--verbose::
175	Be more verbose (show counter open errors, etc).
176
177-s::
178--stat::
179	Record per-thread event counts.  Use it with 'perf report -T' to see
180	the values.
181
182-d::
183--data::
184	Record the sample addresses.
185
186-T::
187--timestamp::
188	Record the sample timestamps. Use it with 'perf report -D' to see the
189	timestamps, for instance.
190
191-P::
192--period::
193	Record the sample period.
194
195-n::
196--no-samples::
197	Don't sample.
198
199-R::
200--raw-samples::
201Collect raw sample records from all opened counters (default for tracepoint counters).
202
203-C::
204--cpu::
205Collect samples only on the list of CPUs provided. Multiple CPUs can be provided as a
206comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2.
207In per-thread mode with inheritance mode on (default), samples are captured only when
208the thread executes on the designated CPUs. Default is to monitor all CPUs.
209
210-B::
211--no-buildid::
212Do not save the build ids of binaries in the perf.data files. This skips
213post processing after recording, which sometimes makes the final step in
214the recording process to take a long time, as it needs to process all
215events looking for mmap records. The downside is that it can misresolve
216symbols if the workload binaries used when recording get locally rebuilt
217or upgraded, because the only key available in this case is the
218pathname. You can also set the "record.build-id" config variable to
219'skip to have this behaviour permanently.
220
221-N::
222--no-buildid-cache::
223Do not update the buildid cache. This saves some overhead in situations
224where the information in the perf.data file (which includes buildids)
225is sufficient.  You can also set the "record.build-id" config variable to
226'no-cache' to have the same effect.
227
228-G name,...::
229--cgroup name,...::
230monitor only in the container (cgroup) called "name". This option is available only
231in per-cpu mode. The cgroup filesystem must be mounted. All threads belonging to
232container "name" are monitored when they run on the monitored CPUs. Multiple cgroups
233can be provided. Each cgroup is applied to the corresponding event, i.e., first cgroup
234to first event, second cgroup to second event and so on. It is possible to provide
235an empty cgroup (monitor all the time) using, e.g., -G foo,,bar. Cgroups must have
236corresponding events, i.e., they always refer to events defined earlier on the command
237line.
238
239-b::
240--branch-any::
241Enable taken branch stack sampling. Any type of taken branch may be sampled.
242This is a shortcut for --branch-filter any. See --branch-filter for more infos.
243
244-j::
245--branch-filter::
246Enable taken branch stack sampling. Each sample captures a series of consecutive
247taken branches. The number of branches captured with each sample depends on the
248underlying hardware, the type of branches of interest, and the executed code.
249It is possible to select the types of branches captured by enabling filters. The
250following filters are defined:
251
252        - any:  any type of branches
253        - any_call: any function call or system call
254        - any_ret: any function return or system call return
255        - ind_call: any indirect branch
256        - call: direct calls, including far (to/from kernel) calls
257        - u:  only when the branch target is at the user level
258        - k: only when the branch target is in the kernel
259        - hv: only when the target is at the hypervisor level
260	- in_tx: only when the target is in a hardware transaction
261	- no_tx: only when the target is not in a hardware transaction
262	- abort_tx: only when the target is a hardware transaction abort
263	- cond: conditional branches
264
265+
266The option requires at least one branch type among any, any_call, any_ret, ind_call, cond.
267The privilege levels may be omitted, in which case, the privilege levels of the associated
268event are applied to the branch filter. Both kernel (k) and hypervisor (hv) privilege
269levels are subject to permissions.  When sampling on multiple events, branch stack sampling
270is enabled for all the sampling events. The sampled branch type is the same for all events.
271The various filters must be specified as a comma separated list: --branch-filter any_ret,u,k
272Note that this feature may not be available on all processors.
273
274--weight::
275Enable weightened sampling. An additional weight is recorded per sample and can be
276displayed with the weight and local_weight sort keys.  This currently works for TSX
277abort events and some memory events in precise mode on modern Intel CPUs.
278
279--transaction::
280Record transaction flags for transaction related events.
281
282--per-thread::
283Use per-thread mmaps.  By default per-cpu mmaps are created.  This option
284overrides that and uses per-thread mmaps.  A side-effect of that is that
285inheritance is automatically disabled.  --per-thread is ignored with a warning
286if combined with -a or -C options.
287
288-D::
289--delay=::
290After starting the program, wait msecs before measuring. This is useful to
291filter out the startup phase of the program, which is often very different.
292
293-I::
294--intr-regs::
295Capture machine state (registers) at interrupt, i.e., on counter overflows for
296each sample. List of captured registers depends on the architecture. This option
297is off by default. It is possible to select the registers to sample using their
298symbolic names, e.g. on x86, ax, si. To list the available registers use
299--intr-regs=\?. To name registers, pass a comma separated list such as
300--intr-regs=ax,bx. The list of register is architecture dependent.
301
302
303--running-time::
304Record running and enabled time for read events (:S)
305
306-k::
307--clockid::
308Sets the clock id to use for the various time fields in the perf_event_type
309records. See clock_gettime(). In particular CLOCK_MONOTONIC and
310CLOCK_MONOTONIC_RAW are supported, some events might also allow
311CLOCK_BOOTTIME, CLOCK_REALTIME and CLOCK_TAI.
312
313-S::
314--snapshot::
315Select AUX area tracing Snapshot Mode. This option is valid only with an
316AUX area tracing event. Optionally the number of bytes to capture per
317snapshot can be specified. In Snapshot Mode, trace data is captured only when
318signal SIGUSR2 is received.
319
320--proc-map-timeout::
321When processing pre-existing threads /proc/XXX/mmap, it may take a long time,
322because the file may be huge. A time out is needed in such cases.
323This option sets the time out limit. The default value is 500 ms.
324
325--switch-events::
326Record context switch events i.e. events of type PERF_RECORD_SWITCH or
327PERF_RECORD_SWITCH_CPU_WIDE.
328
329--clang-path=PATH::
330Path to clang binary to use for compiling BPF scriptlets.
331(enabled when BPF support is on)
332
333--clang-opt=OPTIONS::
334Options passed to clang when compiling BPF scriptlets.
335(enabled when BPF support is on)
336
337--vmlinux=PATH::
338Specify vmlinux path which has debuginfo.
339(enabled when BPF prologue is on)
340
341--buildid-all::
342Record build-id of all DSOs regardless whether it's actually hit or not.
343
344--all-kernel::
345Configure all used events to run in kernel space.
346
347--all-user::
348Configure all used events to run in user space.
349
350--timestamp-filename
351Append timestamp to output file name.
352
353--switch-output::
354Generate multiple perf.data files, timestamp prefixed, switching to a new one
355when receiving a SIGUSR2.
356
357A possible use case is to, given an external event, slice the perf.data file
358that gets then processed, possibly via a perf script, to decide if that
359particular perf.data snapshot should be kept or not.
360
361Implies --timestamp-filename, --no-buildid and --no-buildid-cache.
362
363SEE ALSO
364--------
365linkperf:perf-stat[1], linkperf:perf-list[1]
366