1 perf-trace(1) 1 perf-trace(1)
2 ============= 2 =============
3 3
4 NAME 4 NAME
5 ---- 5 ----
6 perf-trace - strace inspired tool 6 perf-trace - strace inspired tool
7 7
8 SYNOPSIS 8 SYNOPSIS
9 -------- 9 --------
10 [verse] 10 [verse]
11 'perf trace' 11 'perf trace'
12 'perf trace record' 12 'perf trace record'
13 13
14 DESCRIPTION 14 DESCRIPTION
15 ----------- 15 -----------
16 This command will show the events associated w 16 This command will show the events associated with the target, initially
17 syscalls, but other system events like pagefau 17 syscalls, but other system events like pagefaults, task lifetime events,
18 scheduling events, etc. 18 scheduling events, etc.
19 19
20 This is a live mode tool in addition to workin 20 This is a live mode tool in addition to working with perf.data files like
21 the other perf tools. Files can be generated u 21 the other perf tools. Files can be generated using the 'perf record' command
22 but the session needs to include the raw_sysca 22 but the session needs to include the raw_syscalls events (-e 'raw_syscalls:*').
23 Alternatively, 'perf trace record' can be used 23 Alternatively, 'perf trace record' can be used as a shortcut to
24 automatically include the raw_syscalls events 24 automatically include the raw_syscalls events when writing events to a file.
25 25
26 The following options apply to perf trace; opt 26 The following options apply to perf trace; options to perf trace record are
27 found in the perf record man page. 27 found in the perf record man page.
28 28
29 OPTIONS 29 OPTIONS
30 ------- 30 -------
31 31
32 -a:: 32 -a::
33 --all-cpus:: 33 --all-cpus::
34 System-wide collection from all CPUs. 34 System-wide collection from all CPUs.
35 35
36 -e:: 36 -e::
37 --expr:: 37 --expr::
38 --event:: !! 38 List of syscalls to show, currently only syscall names.
39 List of syscalls and other perf events <<
40 etc) to show. Globbing is supported, e <<
41 See 'perf list' for a complete list of <<
42 Prefixing with ! shows all syscalls bu 39 Prefixing with ! shows all syscalls but the ones specified. You may
43 need to escape it. 40 need to escape it.
44 41
45 --filter=<filter>:: <<
46 Event filter. This option should follo <<
47 selects tracepoint event(s). <<
48 <<
49 <<
50 -D msecs:: 42 -D msecs::
51 --delay msecs:: 43 --delay msecs::
52 After starting the program, wait msecs before 44 After starting the program, wait msecs before measuring. This is useful to
53 filter out the startup phase of the program, w 45 filter out the startup phase of the program, which is often very different.
54 46
55 -o:: 47 -o::
56 --output=:: 48 --output=::
57 Output file name. 49 Output file name.
58 50
59 -p:: 51 -p::
60 --pid=:: 52 --pid=::
61 Record events on existing process ID ( 53 Record events on existing process ID (comma separated list).
62 54
63 -t:: 55 -t::
64 --tid=:: 56 --tid=::
65 Record events on existing thread ID (c 57 Record events on existing thread ID (comma separated list).
66 58
67 -u:: 59 -u::
68 --uid=:: 60 --uid=::
69 Record events in threads owned by uid. 61 Record events in threads owned by uid. Name or number.
70 62
71 -G:: <<
72 --cgroup:: <<
73 Record events in threads in a cgroup. <<
74 <<
75 Look for cgroups to set at the /sys/fs <<
76 remove the /sys/fs/cgroup/perf_event/ <<
77 <<
78 perf trace -G A -e sched:*swit <<
79 <<
80 Will set all raw_syscalls:sys_{enter,e <<
81 _and_ sched:sched_switch to the 'A' cg <<
82 <<
83 perf trace -e sched:*switch -G <<
84 <<
85 will only set the sched:sched_switch e <<
86 other events (raw_syscalls:sys_{enter, <<
87 a cgroup (on the root cgroup, sys wide <<
88 <<
89 Multiple cgroups: <<
90 <<
91 perf trace -G A -e sched:*swit <<
92 <<
93 the syscall ones go to the 'A' cgroup, <<
94 to the 'B' cgroup. <<
95 <<
96 --filter-pids=:: 63 --filter-pids=::
97 Filter out events for these pids and f 64 Filter out events for these pids and for 'trace' itself (comma separated list).
98 65
99 -v:: 66 -v::
100 --verbose:: !! 67 --verbose=::
101 Increase the verbosity level. !! 68 Verbosity level.
102 69
103 --no-inherit:: 70 --no-inherit::
104 Child tasks do not inherit counters. 71 Child tasks do not inherit counters.
105 72
106 -m:: 73 -m::
107 --mmap-pages=:: 74 --mmap-pages=::
108 Number of mmap data pages (must be a p 75 Number of mmap data pages (must be a power of two) or size
109 specification in bytes with appended u !! 76 specification with appended unit character - B/K/M/G. The
110 The size is rounded up to the nearest !! 77 size is rounded up to have nearest pages power of two value.
111 78
112 -C:: 79 -C::
113 --cpu:: 80 --cpu::
114 Collect samples only on the list of CPUs provi 81 Collect samples only on the list of CPUs provided. Multiple CPUs can be provided as a
115 comma-separated list with no space: 0,1. Range 82 comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2.
116 In per-thread mode with inheritance mode on (d 83 In per-thread mode with inheritance mode on (default), Events are captured only when
117 the thread executes on the designated CPUs. De 84 the thread executes on the designated CPUs. Default is to monitor all CPUs.
118 85
119 --duration:: !! 86 --duration:
120 Show only events that had a duration g 87 Show only events that had a duration greater than N.M ms.
121 88
122 --sched:: !! 89 --sched:
123 Accrue thread runtime and provide a su 90 Accrue thread runtime and provide a summary at the end of the session.
124 91
125 --failure:: !! 92 -i
126 Show only syscalls that failed, i.e. t !! 93 --input
127 <<
128 -i:: <<
129 --input:: <<
130 Process events from a given perf data 94 Process events from a given perf data file.
131 95
132 -T:: !! 96 -T
133 --time:: !! 97 --time
134 Print full timestamp rather time relat 98 Print full timestamp rather time relative to first sample.
135 99
136 --comm:: 100 --comm::
137 Show process COMM right beside its ID, 101 Show process COMM right beside its ID, on by default, disable with --no-comm.
138 102
139 -s:: 103 -s::
140 --summary:: 104 --summary::
141 Show only a summary of syscalls by thr 105 Show only a summary of syscalls by thread with min, max, and average times
142 (in msec) and relative stddev. 106 (in msec) and relative stddev.
143 107
144 -S:: 108 -S::
145 --with-summary:: 109 --with-summary::
146 Show all syscalls followed by a summar 110 Show all syscalls followed by a summary by thread with min, max, and
147 average times (in msec) and relative stdde 111 average times (in msec) and relative stddev.
148 112
149 --errno-summary:: <<
150 To be used with -s or -S, to show stat <<
151 syscalls, using only this option will <<
152 <<
153 --tool_stats:: 113 --tool_stats::
154 Show tool stats such as number of time 114 Show tool stats such as number of times fd->pathname was discovered thru
155 hooking the open syscall return + vfs_ 115 hooking the open syscall return + vfs_getname or via reading /proc/pid/fd, etc.
156 116
157 -f:: <<
158 --force:: <<
159 Don't complain, do it. <<
160 <<
161 -F=[all|min|maj]:: 117 -F=[all|min|maj]::
162 --pf=[all|min|maj]:: 118 --pf=[all|min|maj]::
163 Trace pagefaults. Optionally, you can 119 Trace pagefaults. Optionally, you can specify whether you want minor,
164 major or all pagefaults. Default value 120 major or all pagefaults. Default value is maj.
165 121
166 --syscalls:: 122 --syscalls::
167 Trace system calls. This options is en !! 123 Trace system calls. This options is enabled by default.
168 --no-syscalls. <<
169 124
170 --call-graph [mode,type,min[,limit],order[,key 125 --call-graph [mode,type,min[,limit],order[,key][,branch]]::
171 Setup and enable call-graph (stack cha 126 Setup and enable call-graph (stack chain/backtrace) recording.
172 See `--call-graph` section in perf-rec 127 See `--call-graph` section in perf-record and perf-report
173 man pages for details. The ones that a 128 man pages for details. The ones that are most useful in 'perf trace'
174 are 'dwarf' and 'lbr', where available 129 are 'dwarf' and 'lbr', where available, try: 'perf trace --call-graph dwarf'.
175 130
176 Using this will, for the root user, bu 131 Using this will, for the root user, bump the value of --mmap-pages to 4
177 times the maximum for non-root users, 132 times the maximum for non-root users, based on the kernel.perf_event_mlock_kb
178 sysctl. This is done only if the user 133 sysctl. This is done only if the user doesn't specify a --mmap-pages value.
179 134
180 --kernel-syscall-graph:: 135 --kernel-syscall-graph::
181 Show the kernel callchains on the sys 136 Show the kernel callchains on the syscall exit path.
182 137
183 --max-events=N:: !! 138 --event::
184 Stop after processing N events. Note t !! 139 Trace other events, see 'perf list' for a complete list.
185 only at exit time or when a syscall is <<
186 option is equivalent to the number of <<
187 <<
188 --switch-on EVENT_NAME:: <<
189 Only consider events after this event <<
190 <<
191 --switch-off EVENT_NAME:: <<
192 Stop considering events after this eve <<
193 <<
194 --show-on-off-events:: <<
195 Show the --switch-on/off events too. <<
196 140
197 --max-stack:: 141 --max-stack::
198 Set the stack depth limit when parsing 142 Set the stack depth limit when parsing the callchain, anything
199 beyond the specified depth will be ign 143 beyond the specified depth will be ignored. Note that at this point
200 this is just about the presentation pa 144 this is just about the presentation part, i.e. the kernel is still
201 not limiting, the overhead of callchai 145 not limiting, the overhead of callchains needs to be set via the
202 knobs in --call-graph dwarf. 146 knobs in --call-graph dwarf.
203 147
204 Implies '--call-graph dwarf' when --ca 148 Implies '--call-graph dwarf' when --call-graph not present on the
205 command line, on systems where DWARF u 149 command line, on systems where DWARF unwinding was built in.
206 150
207 Default: /proc/sys/kernel/perf_event_m 151 Default: /proc/sys/kernel/perf_event_max_stack when present for
208 live sessions (without --inpu 152 live sessions (without --input/-i), 127 otherwise.
209 153
210 --min-stack:: 154 --min-stack::
211 Set the stack depth limit when parsing 155 Set the stack depth limit when parsing the callchain, anything
212 below the specified depth will be igno 156 below the specified depth will be ignored. Disabled by default.
213 157
214 Implies '--call-graph dwarf' when --ca 158 Implies '--call-graph dwarf' when --call-graph not present on the
215 command line, on systems where DWARF u 159 command line, on systems where DWARF unwinding was built in.
216 160
217 --print-sample:: <<
218 Print the PERF_RECORD_SAMPLE PERF_SAMP <<
219 raw_syscalls:sys_{enter,exit} tracepoi <<
220 <<
221 --proc-map-timeout:: 161 --proc-map-timeout::
222 When processing pre-existing threads / 162 When processing pre-existing threads /proc/XXX/mmap, it may take a long time,
223 because the file may be huge. A time o 163 because the file may be huge. A time out is needed in such cases.
224 This option sets the time out limit. T 164 This option sets the time out limit. The default value is 500 ms.
225 165
226 --sort-events:: <<
227 Do sorting on batches of events, use w <<
228 may happen, for instance, when a threa <<
229 while processing a syscall. <<
230 <<
231 --libtraceevent_print:: <<
232 Use libtraceevent to print tracepoint <<
233 the same beautifiers used in the strac <<
234 tracepoint arguments. <<
235 <<
236 --map-dump:: <<
237 Dump BPF maps setup by events passed v <<
238 living in tools/perf/examples/bpf/augm <<
239 dumps just boolean map values and inte <<
240 by default and use BTF when available, <<
241 printing using the existing 'perf trac <<
242 arguments to strings (pid to comm, sys <<
243 <<
244 <<
245 PAGEFAULTS 166 PAGEFAULTS
246 ---------- 167 ----------
247 168
248 When tracing pagefaults, the format of the tra 169 When tracing pagefaults, the format of the trace is as follows:
249 170
250 <min|maj>fault [<ip.symbol>+<ip.offset>] => <ad 171 <min|maj>fault [<ip.symbol>+<ip.offset>] => <addr.dso@addr.offset> (<map type><addr level>).
251 172
252 - min/maj indicates whether fault event is min 173 - min/maj indicates whether fault event is minor or major;
253 - ip.symbol shows symbol for instruction point 174 - ip.symbol shows symbol for instruction pointer (the code that generated the
254 fault); if no debug symbols available, perf 175 fault); if no debug symbols available, perf trace will print raw IP;
255 - addr.dso shows DSO for the faulted address; 176 - addr.dso shows DSO for the faulted address;
256 - map type is either 'd' for non-executable ma 177 - map type is either 'd' for non-executable maps or 'x' for executable maps;
257 - addr level is either 'k' for kernel dso or ' 178 - addr level is either 'k' for kernel dso or '.' for user dso.
258 179
259 For symbols resolution you may need to install 180 For symbols resolution you may need to install debugging symbols.
260 181
261 Please be aware that duration is currently alw 182 Please be aware that duration is currently always 0 and doesn't reflect actual
262 time it took for fault to be handled! 183 time it took for fault to be handled!
263 184
264 When --verbose specified, perf trace tries to 185 When --verbose specified, perf trace tries to print all available information
265 for both IP and fault address in the form of d 186 for both IP and fault address in the form of dso@symbol+offset.
266 187
267 EXAMPLES 188 EXAMPLES
268 -------- 189 --------
269 190
270 Trace only major pagefaults: 191 Trace only major pagefaults:
271 192
272 $ perf trace --no-syscalls -F 193 $ perf trace --no-syscalls -F
273 194
274 Trace syscalls, major and minor pagefaults: 195 Trace syscalls, major and minor pagefaults:
275 196
276 $ perf trace -F all 197 $ perf trace -F all
277 198
278 1416.547 ( 0.000 ms): python/20235 majfault 199 1416.547 ( 0.000 ms): python/20235 majfault [CRYPTO_push_info_+0x0] => /lib/x86_64-linux-gnu/libcrypto.so.1.0.0@0x61be0 (x.)
279 200
280 As you can see, there was major pagefault in 201 As you can see, there was major pagefault in python process, from
281 CRYPTO_push_info_ routine which faulted some 202 CRYPTO_push_info_ routine which faulted somewhere in libcrypto.so.
282 <<
283 Trace the first 4 open, openat or open_by_hand <<
284 <<
285 $ perf trace -e open* --max-events 4 <<
286 [root@jouet perf]# trace -e open* --max-even <<
287 2272.992 ( 0.037 ms): gnome-shell/1370 opena <<
288 2277.481 ( 0.139 ms): gnome-shell/3039 opena <<
289 3026.398 ( 0.076 ms): gnome-shell/3039 opena <<
290 4294.665 ( 0.015 ms): sed/15879 openat(dfd: <<
291 $ <<
292 <<
293 Trace the first minor page fault when running <<
294 <<
295 # perf trace -F min --max-stack=7 --max-even <<
296 0.000 ( 0.000 ms): sleep/18006 minfault [ <<
297 __clear <<
298 load_el <<
299 search_ <<
300 __do_ex <<
301 __x64_s <<
302 do_sysc <<
303 entry_S <<
304 # <<
305 <<
306 Trace the next min page page fault to take pla <<
307 <<
308 # perf trace -F min --call-graph=dwarf --max <<
309 0.000 ( 0.000 ms): Web Content/17136 minf <<
310 js::gc: <<
311 js::gc: <<
312 js::gc: <<
313 js::gc: <<
314 js::gc: <<
315 js::gc: <<
316 js::gc: <<
317 js::All <<
318 js::All <<
319 JSThinI <<
320 Allocat <<
321 js::Con <<
322 [0x18b2 <<
323 # <<
324 <<
325 Trace the next two sched:sched_switch events, <<
326 next block:*_unplug and the next three net:*de <<
327 with a backtrace of at most 16 entries, system <<
328 <<
329 # perf trace -e sched:*switch/nr=2/,block:*_ <<
330 0.000 :0/0 sched:sched_switch:swapper/2:0 <<
331 0.015 rcu_sched/10 sched:sched_switch:rcu <<
332 254.198 irq/50-iwlwifi/680 net:net_dev_queu <<
333 __dev_q <<
334 273.977 :0/0 net:net_dev_queue:dev=wlp3s0 s <<
335 __dev_q <<
336 274.007 :0/0 net:net_dev_queue:dev=wlp3s0 s <<
337 __dev_q <<
338 2930.140 kworker/u16:58/2722 block:block_plu <<
339 2930.162 kworker/u16:58/2722 block:block_unp <<
340 4466.094 jbd2/dm-2-8/748 block:block_plug:[j <<
341 8050.123 kworker/u16:30/2694 block:block_plu <<
342 8050.271 kworker/u16:30/2694 block:block_plu <<
343 # <<
344 203
345 SEE ALSO 204 SEE ALSO
346 -------- 205 --------
347 linkperf:perf-record[1], linkperf:perf-script[ 206 linkperf:perf-record[1], linkperf:perf-script[1]
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.