1TRACE-CMD-RECORD(1) TRACE-CMD-RECORD(1)
2
3
4
6 trace-cmd-record - record a trace from the Ftrace Linux internal tracer
7
9 trace-cmd record [OPTIONS] [command]
10
12 The trace-cmd(1) record command will set up the Ftrace Linux kernel
13 tracer to record the specified plugins or events that happen while the
14 command executes. If no command is given, then it will record until the
15 user hits Ctrl-C.
16
17 The record command of trace-cmd will set up the Ftrace tracer to start
18 tracing the various events or plugins that are given on the command
19 line. It will then create a number of tracing processes (one per CPU)
20 that will start recording from the kernel ring buffer straight into
21 temporary files. When the command is complete (or Ctrl-C is hit) all
22 the files will be combined into a trace.dat file that can later be read
23 (see trace-cmd-report(1)).
24
26 -p tracer
27 Specify a tracer. Tracers usually do more than just trace an event.
28 Common tracers are: function, function_graph, preemptirqsoff,
29 irqsoff, preemptoff and wakeup. A tracer must be supported by the
30 running kernel. To see a list of available tracers, see
31 trace-cmd-list(1).
32
33 -e event
34 Specify an event to trace. Various static trace points have been
35 added to the Linux kernel. They are grouped by subsystem where you
36 can enable all events of a given subsystem or specify specific
37 events to be enabled. The event is of the format
38 "subsystem:event-name". You can also just specify the subsystem
39 without the :event-name or the event-name without the "subsystem:".
40 Using "-e sched_switch" will enable the "sched_switch" event where
41 as, "-e sched" will enable all events under the "sched" subsystem.
42
43 The 'event' can also contain glob expressions. That is, "*stat*" will
44 select all events (or subsystems) that have the characters "stat" in their
45 names.
46
47 The keyword 'all' can be used to enable all events.
48
49 -a
50 Every event that is being recorded has its output format file saved
51 in the output file to be able to display it later. But if other
52 events are enabled in the trace without trace-cmd’s knowledge, the
53 formats of those events will not be recorded and trace-cmd report
54 will not be able to display them. If this is the case, then specify
55 the -a option and the format for all events in the system will be
56 saved.
57
58 -T
59 Enable a stacktrace on each event. For example:
60
61 <idle>-0 [003] 58549.289091: sched_switch: kworker/0:1:0 [120] R ==> trace-cmd:2603 [120]
62 <idle>-0 [003] 58549.289092: kernel_stack: <stack trace>
63 => schedule (ffffffff814b260e)
64 => cpu_idle (ffffffff8100a38c)
65 => start_secondary (ffffffff814ab828)
66
67 --func-stack
68 Enable a stack trace on all functions. Note this is only applicable
69 for the "function" plugin tracer, and will only take effect if the
70 -l option is used and succeeds in limiting functions. If the
71 function tracer is not filtered, and the stack trace is enabled,
72 you can live lock the machine.
73
74 -f filter
75 Specify a filter for the previous event. This must come after a -e.
76 This will filter what events get recorded based on the content of
77 the event. Filtering is passed to the kernel directly so what
78 filtering is allowed may depend on what version of the kernel you
79 have. Basically, it will let you use C notation to check if an
80 event should be processed or not.
81
82 ==, >=, <=, >, <, &, |, && and ||
83
84 The above are usually safe to use to compare fields.
85
86 --no-filter
87 Do not filter out the trace-cmd threads. By default, the threads
88 are filtered out to not be traced by events. This option will have
89 the trace-cmd threads also be traced.
90
91 -R trigger
92 Specify a trigger for the previous event. This must come after a
93 -e. This will add a given trigger to the given event. To only
94 enable the trigger and not the event itself, then place the event
95 after the -v option.
96
97 See Documentation/trace/events.txt in the Linux kernel source for more
98 information on triggers.
99
100 -v
101 This will cause all events specified after it on the command line
102 to not be traced. This is useful for selecting a subsystem to be
103 traced but to leave out various events. For Example: "-e sched -v
104 -e "*stat\*"" will enable all events in the sched subsystem except
105 those that have "stat" in their names.
106
107 Note: the *-v* option was taken from the way grep(1) inverts the following
108 matches.
109
110 -F
111 This will filter only the executable that is given on the command
112 line. If no command is given, then it will filter itself (pretty
113 pointless). Using -F will let you trace only events that are caused
114 by the given command.
115
116 -P pid
117 Similar to -F but lets you specify a process ID to trace.
118
119 -c
120 Used with either -F (or -P if kernel supports it) to trace the
121 process' children too.
122
123 --user
124 Execute the specified command as given user.
125
126 -C clock
127 Set the trace clock to "clock".
128
129 Use trace-cmd(1) list -C to see what clocks are available.
130
131 -o output-file
132 By default, trace-cmd report will create a trace.dat file. You can
133 specify a different file to write to with the -o option.
134
135 -l function-name
136 This will limit the function and function_graph tracers to only
137 trace the given function name. More than one -l may be specified on
138 the command line to trace more than one function. The limited use
139 of glob expressions are also allowed. These are match* to only
140 filter functions that start with match. *match to only filter
141 functions that end with match. *match\* to only filter on
142 functions that contain match.
143
144 -g function-name
145 This option is for the function_graph plugin. It will graph the
146 given function. That is, it will only trace the function and all
147 functions that it calls. You can have more than one -g on the
148 command line.
149
150 -n function-name
151 This has the opposite effect of -l. The function given with the -n
152 option will not be traced. This takes precedence, that is, if you
153 include the same function for both -n and -l, it will not be
154 traced.
155
156 -d
157 Some tracer plugins enable the function tracer by default. Like the
158 latency tracers. This option prevents the function tracer from
159 being enabled at start up.
160
161 -D
162 The option -d will try to use the function-trace option to disable
163 the function tracer (if available), otherwise it defaults to the
164 proc file: /proc/sys/kernel/ftrace_enabled, but will not touch it
165 if the function-trace option is available. The -D option will
166 disable both the ftrace_enabled proc file as well as the
167 function-trace option if it exists.
168
169 Note, this disable function tracing for all users, which includes users
170 outside of ftrace tracers (stack_tracer, perf, etc).
171
172 -O option
173 Ftrace has various options that can be enabled or disabled. This
174 allows you to set them. Appending the text no to an option disables
175 it. For example: "-O nograph-time" will disable the "graph-time"
176 Ftrace option.
177
178 -s interval
179 The processes that trace-cmd creates to record from the ring buffer
180 need to wake up to do the recording. Setting the interval to zero
181 will cause the processes to wakeup every time new data is written
182 into the buffer. But since Ftrace is recording kernel activity, the
183 act of this processes going back to sleep may cause new events into
184 the ring buffer which will wake the process back up. This will
185 needlessly add extra data into the ring buffer.
186
187 The 'interval' metric is microseconds. The default is set to 1000 (1 ms).
188 This is the time each recording process will sleep before waking up to
189 record any new data that was written to the ring buffer.
190
191 -r priority
192 The priority to run the capture threads at. In a busy system the
193 trace capturing threads may be staved and events can be lost. This
194 increases the priority of those threads to the real time (FIFO)
195 priority. But use this option with care, it can also change the
196 behaviour of the system being traced.
197
198 -b size
199 This sets the ring buffer size to size kilobytes. Because the
200 Ftrace ring buffer is per CPU, this size is the size of each per
201 CPU ring buffer inside the kernel. Using "-b 10000" on a machine
202 with 4 CPUs will make Ftrace have a total buffer size of 40 Megs.
203
204 -B buffer-name
205 If the kernel supports multiple buffers, this will add a buffer
206 with the given name. If the buffer name already exists, that buffer
207 is just reset and will not be deleted at the end of record
208 execution. If the buffer is created, it will be removed at the end
209 of execution (unless the -k is set, or start command was used).
210
211 After a buffer name is stated, all events added after that will be
212 associated with that buffer. If no buffer is specified, or an event
213 is specified before a buffer name, it will be associated with the
214 main (toplevel) buffer.
215
216 trace-cmd record -e sched -B block -e block -B time -e timer sleep 1
217
218 The above is will enable all sched events in the main buffer. It will
219 then create a 'block' buffer instance and enable all block events within
220 that buffer. A 'time' buffer instance is created and all timer events
221 will be enabled for that event.
222
223 -m size
224 The max size in kilobytes that a per cpu buffer should be. Note,
225 due to rounding to page size, the number may not be totally
226 correct. Also, this is performed by switching between two buffers
227 that are half the given size thus the output may not be of the
228 given size even if much more was written.
229
230 Use this to prevent running out of diskspace for long runs.
231
232 -M cpumask
233 Set the cpumask for to trace. It only affects the last buffer
234 instance given. If supplied before any buffer instance, then it
235 affects the main buffer. The value supplied must be a hex number.
236
237 trace-cmd record -p function -M c -B events13 -e all -M 5
238
239 If the -M is left out, then the mask stays the same. To enable all
240 CPUs, pass in a value of '-1'.
241
242 -k
243 By default, when trace-cmd is finished tracing, it will reset the
244 buffers and disable all the tracing that it enabled. This option
245 keeps trace-cmd from disabling the tracer and reseting the buffer.
246 This option is useful for debugging trace-cmd.
247
248 Note: usually trace-cmd will set the "tracing_on" file back to what it
249 was before it was called. This option will leave that file set to zero.
250
251 -i
252 By default, if an event is listed that trace-cmd does not find, it
253 will exit with an error. This option will just ignore events that
254 are listed on the command line but are not found on the system.
255
256 -N host:port
257 If another machine is running "trace-cmd listen", this option is
258 used to have the data sent to that machine with UDP packets.
259 Instead of writing to an output file, the data is sent off to a
260 remote box. This is ideal for embedded machines with little
261 storage, or having a single machine that will keep all the data in
262 a single repository.
263
264 Note: This option is not supported with latency tracer plugins:
265 wakeup, wakeup_rt, irqsoff, preemptoff and preemptirqsoff
266
267 -t
268 This option is used with -N, when there’s a need to send the live
269 data with TCP packets instead of UDP. Although TCP is not nearly as
270 fast as sending the UDP packets, but it may be needed if the
271 network is not that reliable, the amount of data is not that
272 intensive, and a guarantee is needed that all traced information is
273 transfered successfully.
274
275 -q | --quiet
276 For use with recording an application. Suppresses normal output
277 (except for errors) to allow only the application’s output to be
278 displayed.
279
280 --date
281 With the --date option, "trace-cmd" will write timestamps into the
282 trace buffer after it has finished recording. It will then map the
283 timestamp to gettimeofday which will allow wall time output from
284 the timestamps reading the created trace.dat file.
285
286 --max-graph-depth depth
287 Set the maximum depth the function_graph tracer will trace into a
288 function. A value of one will only show where userspace enters the
289 kernel but not any functions called in the kernel. The default is
290 zero, which means no limit.
291
292 --cmdlines-size size
293 Set the number of entries the kernel tracing file "saved_cmdlines"
294 can contain. This file is a circular buffer which stores the
295 mapping between cmdlines and PIDs. If full, it leads to unresolved
296 cmdlines ("<...>") within the trace. The kernel default value is
297 128.
298
299 --module module
300 Filter a module’s name in function tracing. It is equivalent to
301 adding :mod:module after all other functions being filtered. If no
302 other function filter is listed, then all modules functions will be
303 filtered in the filter.
304
305 '--module snd' is equivalent to '-l :mod:snd'
306
307 '--module snd -l "*jack*"' is equivalent to '-l "*jack*:mod:snd"'
308
309 '--module snd -n "*"' is equivalent to '-n :mod:snd'
310
311 --proc-map
312 Save the traced process address map into the trace.dat file. The
313 traced processes can be specified using the option -P, or as a
314 given command.
315
316 --profile
317 With the --profile option, "trace-cmd" will enable tracing that can
318 be used with trace-cmd-report(1) --profile option. If a tracer -p
319 is not set, and function graph depth is supported by the kernel,
320 then the function_graph tracer will be enabled with a depth of one
321 (only show where userspace enters into the kernel). It will also
322 enable various tracepoints with stack tracing such that the report
323 can show where tasks have been blocked for the longest time.
324
325 See trace-cmd-profile(1) for more details and examples.
326
327 -G
328 Set interrupt (soft and hard) events as global (associated to CPU
329 instead of tasks). Only works for --profile.
330
331 -H event-hooks
332 Add custom event matching to connect any two events together. When
333 not used with --profile, it will save the parameter and this will
334 be used by trace-cmd report --profile, too. That is:
335
336 trace-cmd record -H hrtimer_expire_entry,hrtimer/hrtimer_expire_exit,hrtimer,sp
337 trace-cmd report --profile
338
339 Will profile hrtimer_expire_entry and hrtimer_expire_ext times.
340
341 See trace-cmd-profile(1) for format.
342
343 -S
344 (for --profile only) Only enable the tracer or events speficied on
345 the command line. With this option, the function_graph tracer is
346 not enabled, nor are any events (like sched_switch), unless they
347 are specifically specified on the command line (i.e. -p function -e
348 sched_switch -e sched_wakeup)
349
350 --ts-offset offset
351 Add an offset for the timestamp in the trace.dat file. This will
352 add a offset option into the trace.dat file such that a trace-cmd
353 report will offset all the timestamps of the events by the given
354 offset. The offset is in raw units. That is, if the event
355 timestamps are in nanoseconds the offset will also be in
356 nanoseconds even if the displayed units are in microseconds.
357
358 --tsync-interval
359 Set the loop interval, in ms, for timestamps synchronization with
360 guests: If a negative number is specified, timestamps
361 synchronization is disabled If 0 is specified, no loop is performed
362 - timestamps offset is calculated only twice," at the beginning and
363 at the end of the trace. Timestamps synchronization with guests
364 works only if there is support for VSOCK.\n"
365
366 --stderr
367 Have output go to stderr instead of stdout, but the output of the
368 command executed will not be changed. This is useful if you want to
369 monitor the output of the command being executed, but not see the
370 output from trace-cmd.
371
373 The basic way to trace all events:
374
375 # trace-cmd record -e all ls > /dev/null
376 # trace-cmd report
377 trace-cmd-13541 [003] 106260.693809: filemap_fault: address=0x128122 offset=0xce
378 trace-cmd-13543 [001] 106260.693809: kmalloc: call_site=81128dd4 ptr=0xffff88003dd83800 bytes_req=768 bytes_alloc=1024 gfp_flags=GFP_KERNEL|GFP_ZERO
379 ls-13545 [002] 106260.693809: kfree: call_site=810a7abb ptr=0x0
380 ls-13545 [002] 106260.693818: sys_exit_write: 0x1
381
382 To use the function tracer with sched switch tracing:
383
384 # trace-cmd record -p function -e sched_switch ls > /dev/null
385 # trace-cmd report
386 ls-13587 [002] 106467.860310: function: hrtick_start_fair <-- pick_next_task_fair
387 ls-13587 [002] 106467.860313: sched_switch: prev_comm=trace-cmd prev_pid=13587 prev_prio=120 prev_state=R ==> next_comm=trace-cmd next_pid=13583 next_prio=120
388 trace-cmd-13585 [001] 106467.860314: function: native_set_pte_at <-- __do_fault
389 trace-cmd-13586 [003] 106467.860314: function: up_read <-- do_page_fault
390 ls-13587 [002] 106467.860317: function: __phys_addr <-- schedule
391 trace-cmd-13585 [001] 106467.860318: function: _raw_spin_unlock <-- __do_fault
392 ls-13587 [002] 106467.860320: function: native_load_sp0 <-- __switch_to
393 trace-cmd-13586 [003] 106467.860322: function: down_read_trylock <-- do_page_fault
394
395 Here is a nice way to find what interrupts have the highest latency:
396
397 # trace-cmd record -p function_graph -e irq_handler_entry -l do_IRQ sleep 10
398 # trace-cmd report
399 <idle>-0 [000] 157412.933969: funcgraph_entry: | do_IRQ() {
400 <idle>-0 [000] 157412.933974: irq_handler_entry: irq=48 name=eth0
401 <idle>-0 [000] 157412.934004: funcgraph_exit: + 36.358 us | }
402 <idle>-0 [000] 157413.895004: funcgraph_entry: | do_IRQ() {
403 <idle>-0 [000] 157413.895011: irq_handler_entry: irq=48 name=eth0
404 <idle>-0 [000] 157413.895026: funcgraph_exit: + 24.014 us | }
405 <idle>-0 [000] 157415.891762: funcgraph_entry: | do_IRQ() {
406 <idle>-0 [000] 157415.891769: irq_handler_entry: irq=48 name=eth0
407 <idle>-0 [000] 157415.891784: funcgraph_exit: + 22.928 us | }
408 <idle>-0 [000] 157415.934869: funcgraph_entry: | do_IRQ() {
409 <idle>-0 [000] 157415.934874: irq_handler_entry: irq=48 name=eth0
410 <idle>-0 [000] 157415.934906: funcgraph_exit: + 37.512 us | }
411 <idle>-0 [000] 157417.888373: funcgraph_entry: | do_IRQ() {
412 <idle>-0 [000] 157417.888381: irq_handler_entry: irq=48 name=eth0
413 <idle>-0 [000] 157417.888398: funcgraph_exit: + 25.943 us | }
414
415 An example of the profile:
416
417 # trace-cmd record --profile sleep 1
418 # trace-cmd report --profile --comm sleep
419 task: sleep-21611
420 Event: sched_switch:R (1) Total: 99442 Avg: 99442 Max: 99442 Min:99442
421 <stack> 1 total:99442 min:99442 max:99442 avg=99442
422 => ftrace_raw_event_sched_switch (0xffffffff8105f812)
423 => __schedule (0xffffffff8150810a)
424 => preempt_schedule (0xffffffff8150842e)
425 => ___preempt_schedule (0xffffffff81273354)
426 => cpu_stop_queue_work (0xffffffff810b03c5)
427 => stop_one_cpu (0xffffffff810b063b)
428 => sched_exec (0xffffffff8106136d)
429 => do_execve_common.isra.27 (0xffffffff81148c89)
430 => do_execve (0xffffffff811490b0)
431 => SyS_execve (0xffffffff811492c4)
432 => return_to_handler (0xffffffff8150e3c8)
433 => stub_execve (0xffffffff8150c699)
434 Event: sched_switch:S (1) Total: 1000506680 Avg: 1000506680 Max: 1000506680 Min:1000506680
435 <stack> 1 total:1000506680 min:1000506680 max:1000506680 avg=1000506680
436 => ftrace_raw_event_sched_switch (0xffffffff8105f812)
437 => __schedule (0xffffffff8150810a)
438 => schedule (0xffffffff815084b8)
439 => do_nanosleep (0xffffffff8150b22c)
440 => hrtimer_nanosleep (0xffffffff8108d647)
441 => SyS_nanosleep (0xffffffff8108d72c)
442 => return_to_handler (0xffffffff8150e3c8)
443 => tracesys_phase2 (0xffffffff8150c304)
444 Event: sched_wakeup:21611 (1) Total: 30326 Avg: 30326 Max: 30326 Min:30326
445 <stack> 1 total:30326 min:30326 max:30326 avg=30326
446 => ftrace_raw_event_sched_wakeup_template (0xffffffff8105f653)
447 => ttwu_do_wakeup (0xffffffff810606eb)
448 => ttwu_do_activate.constprop.124 (0xffffffff810607c8)
449 => try_to_wake_up (0xffffffff8106340a)
450
452 trace-cmd(1), trace-cmd-report(1), trace-cmd-start(1),
453 trace-cmd-stop(1), trace-cmd-extract(1), trace-cmd-reset(1),
454 trace-cmd-split(1), trace-cmd-list(1), trace-cmd-listen(1),
455 trace-cmd-profile(1)
456
458 Written by Steven Rostedt, <rostedt@goodmis.org[1]>
459
461 https://git.kernel.org/pub/scm/utils/trace-cmd/trace-cmd.git/
462
464 Copyright (C) 2010 Red Hat, Inc. Free use of this software is granted
465 under the terms of the GNU Public License (GPL).
466
468 1. rostedt@goodmis.org
469 mailto:rostedt@goodmis.org
470
471
472
473 03/29/2021 TRACE-CMD-RECORD(1)