1TRACE-CMD-RECORD(1) TRACE-CMD-RECORD(1)
2
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 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 plugin
27 Specify a trace plugin. Plugins are special Ftrace tracers that
28 usually do more than just trace an event. Common plugins are
29 function, function_graph, preemptirqsoff, irqsoff, preemptoff, and
30 wakeup. A plugin must be supported by the running kernel. To see a
31 list of available plugins, see trace-cmd-list(1).
32 -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 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 The keyword 'all' can be used to enable all events.
48 -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 -T
59 Enable a stacktrace on each event. For example:
60 <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 --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 -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 ==, >=, <=, >, <, &, |, && and ||
83 The above are usually safe to use to compare fields.
85 -R trigger
87 Specify a trigger for the previous event. This must come after a
88 -e. This will add a given trigger to the given event. To only
89 enable the trigger and not the event itself, then place the event
90 after the -v option.
91 See Documentation/trace/events.txt in the Linux kernel source for more
93 information on triggers.
94 -v
96 This will cause all events specified after it on the command line
97 to not be traced. This is useful for selecting a subsystem to be
98 traced but to leave out various events. For Example: "-e sched -v
99 -e "*stat\*"" will enable all events in the sched subsystem except
100 those that have "stat" in their names.
101 Note: the *-v* option was taken from the way grep(1) inverts the following
103 matches.
104 -F
106 This will filter only the executable that is given on the command
107 line. If no command is given, then it will filter itself (pretty
108 pointless). Using -F will let you trace only events that are caused
109 by the given command.
110 -P pid
112 Similar to -F but lets you specify a process ID to trace.
113 -c
115 Used with either -F to trace the process' children too.
116 -C clock
118 Set the trace clock to "clock".
119 Use trace-cmd(1) list -C to see what clocks are available.
121 -o output-file
123 By default, trace-cmd report will create a trace.dat file. You can
124 specify a different file to write to with the -o option.
125 -l function-name
127 This will limit the function and function_graph tracers to only
128 trace the given function name. More than one -l may be specified on
129 the command line to trace more than one function. The limited use
130 of glob expressions are also allowed. These are match* to only
131 filter functions that start with match. *match to only filter
132 functions that end with match. *match\* to only filter on
133 functions that contain match.
134 -g function-name
136 This option is for the function_graph plugin. It will graph the
137 given function. That is, it will only trace the function and all
138 functions that it calls. You can have more than one -g on the
139 command line.
140 -n function-name
142 This has the opposite effect of -l. The function given with the -n
143 option will not be traced. This takes precedence, that is, if you
144 include the same function for both -n and -l, it will not be
145 traced.
146 -d
148 Some tracer plugins enable the function tracer by default. Like the
149 latency tracers. This option prevents the function tracer from
150 being enabled at start up.
151 -D
153 The option -d will try to use the function-trace option to disable
154 the function tracer (if available), otherwise it defaults to the
155 proc file: /proc/sys/kernel/ftrace_enabled, but will not touch it
156 if the function-trace option is available. The -D option will
157 disable both the ftrace_enabled proc file as well as the
158 function-trace option if it exists.
159 Note, this disable function tracing for all users, which includes users
161 outside of ftrace tracers (stack_tracer, perf, etc).
162 -O option
164 Ftrace has various options that can be enabled or disabled. This
165 allows you to set them. Appending the text no to an option disables
166 it. For example: "-O nograph-time" will disable the "graph-time"
167 Ftrace option.
168 -s interval
170 The processes that trace-cmd creates to record from the ring buffer
171 need to wake up to do the recording. Setting the interval to zero
172 will cause the processes to wakeup every time new data is written
173 into the buffer. But since Ftrace is recording kernel activity, the
174 act of this processes going back to sleep may cause new events into
175 the ring buffer which will wake the process back up. This will
176 needlessly add extra data into the ring buffer.
177 The 'interval' metric is microseconds. The default is set to 1000 (1 ms).
179 This is the time each recording process will sleep before waking up to
180 record any new data that was written to the ring buffer.
181 -r priority
183 The priority to run the capture threads at. In a busy system the
184 trace capturing threads may be staved and events can be lost. This
185 increases the priority of those threads to the real time (FIFO)
186 priority. But use this option with care, it can also change the
187 behaviour of the system being traced.
188 -b size
190 This sets the ring buffer size to size kilobytes. Because the
191 Ftrace ring buffer is per CPU, this size is the size of each per
192 CPU ring buffer inside the kernel. Using "-b 10000" on a machine
193 with 4 CPUs will make Ftrace have a total buffer size of 40 Megs.
194 -B buffer-name
196 If the kernel supports multiple buffers, this will add a buffer
197 with the given name. If the buffer name already exists, that buffer
198 is just reset and will not be deleted at the end of record
199 execution. If the buffer is created, it will be removed at the end
200 of execution (unless the -k is set, or start command was used).
201 After a buffer name is stated, all events added after that will be
203 associated with that buffer. If no buffer is specified, or an event
204 is specified before a buffer name, it will be associated with the
205 main (toplevel) buffer.
206 trace-cmd record -e sched -B block -e block -B time -e timer sleep 1
208 The above is will enable all sched events in the main buffer. It will
210 then create a 'block' buffer instance and enable all block events within
211 that buffer. A 'time' buffer instance is created and all timer events
212 will be enabled for that event.
213 -m size
215 The max size in kilobytes that a per cpu buffer should be. Note,
216 due to rounding to page size, the number may not be totally
217 correct. Also, this is performed by switching between two buffers
218 that are half the given size thus the output may not be of the
219 given size even if much more was written.
220 Use this to prevent running out of diskspace for long runs.
222 -M cpumask
224 Set the cpumask for to trace. It only affects the last buffer
225 instance given. If supplied before any buffer instance, then it
226 affects the main buffer. The value supplied must be a hex number.
227 trace-cmd record -p function -M c -B events13 -e all -M 5
229 If the -M is left out, then the mask stays the same. To enable all
231 CPUs, pass in a value of '-1'.
232 -k
234 By default, when trace-cmd is finished tracing, it will reset the
235 buffers and disable all the tracing that it enabled. This option
236 keeps trace-cmd from disabling the tracer and reseting the buffer.
237 This option is useful for debugging trace-cmd.
238 Note: usually trace-cmd will set the "tracing_on" file back to what it
240 was before it was called. This option will leave that file set to zero.
241 -i
243 By default, if an event is listed that trace-cmd does not find, it
244 will exit with an error. This option will just ignore events that
245 are listed on the command line but are not found on the system.
246 -N host:port
248 If another machine is running "trace-cmd listen", this option is
249 used to have the data sent to that machine with UDP packets.
250 Instead of writing to an output file, the data is sent off to a
251 remote box. This is ideal for embedded machines with little
252 storage, or having a single machine that will keep all the data in
253 a single repository.
254 Note: This option is not supported with latency tracer plugins:
256 wakeup, wakeup_rt, irqsoff, preemptoff and preemptirqsoff
257 -t
259 This option is used with -N, when there’s a need to send the live
260 data with TCP packets instead of UDP. Although TCP is not nearly as
261 fast as sending the UDP packets, but it may be needed if the
262 network is not that reliable, the amount of data is not that
263 intensive, and a guarantee is needed that all traced information is
264 transfered successfully.
265 --date
267 With the --date option, "trace-cmd" will write timestamps into the
268 trace buffer after it has finished recording. It will then map the
269 timestamp to gettimeofday which will allow wall time output from
270 the timestamps reading the created trace.dat file.
271 --max-graph-depth depth
273 Set the maximum depth the function_graph tracer will trace into a
274 function. A value of one will only show where userspace enters the
275 kernel but not any functions called in the kernel. The default is
276 zero, which means no limit.
277 --profile
279 With the --profile option, "trace-cmd" will enable tracing that can
280 be used with trace-cmd-report(1) --profile option. If a tracer -p
281 is not set, and function graph depth is supported by the kernel,
282 then the function_graph tracer will be enabled with a depth of one
283 (only show where userspace enters into the kernel). It will also
284 enable various tracepoints with stack tracing such that the report
285 can show where tasks have been blocked for the longest time.
286 See trace-cmd-profile(1) for more details and examples.
288 -H event-hooks
290 Add custom event matching to connect any two events together. When
291 not used with --profile, it will save the parameter and this will
292 be used by trace-cmd report --profile, too. That is:
293 trace-cmd record -H hrtimer_expire_entry,hrtimer/hrtimer_expire_exit,hrtimer,sp
295 trace-cmd report --profile
296 Will profile hrtimer_expire_entry and hrtimer_expire_ext times.
298 See trace-cmd-profile(1) for format.
300 --ts-offset offset
302 Add an offset for the timestamp in the trace.dat file. This will
303 add a offset option into the trace.dat file such that a trace-cmd
304 report will offset all the timestamps of the events by the given
305 offset. The offset is in raw units. That is, if the event
306 timestamps are in nanoseconds the offset will also be in
307 nanoseconds even if the displayed units are in microseconds.
308 --stderr
310 Have output go to stderr instead of stdout, but the output of the
311 command executed will not be changed. This is useful if you want to
312 monitor the output of the command being executed, but not see the
313 output from trace-cmd.
314
316 The basic way to trace all events:
317 # trace-cmd record -e all ls > /dev/null
319 # trace-cmd report
320 trace-cmd-13541 [003] 106260.693809: filemap_fault: address=0x128122 offset=0xce
321 trace-cmd-13543 [001] 106260.693809: kmalloc: call_site=81128dd4 ptr=0xffff88003dd83800 bytes_req=768 bytes_alloc=1024 gfp_flags=GFP_KERNEL|GFP_ZERO
322 ls-13545 [002] 106260.693809: kfree: call_site=810a7abb ptr=0x0
323 ls-13545 [002] 106260.693818: sys_exit_write: 0x1
324 To use the function tracer with sched switch tracing:
326 # trace-cmd record -p function -e sched_switch ls > /dev/null
328 # trace-cmd report
329 ls-13587 [002] 106467.860310: function: hrtick_start_fair <-- pick_next_task_fair
330 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
331 trace-cmd-13585 [001] 106467.860314: function: native_set_pte_at <-- __do_fault
332 trace-cmd-13586 [003] 106467.860314: function: up_read <-- do_page_fault
333 ls-13587 [002] 106467.860317: function: __phys_addr <-- schedule
334 trace-cmd-13585 [001] 106467.860318: function: _raw_spin_unlock <-- __do_fault
335 ls-13587 [002] 106467.860320: function: native_load_sp0 <-- __switch_to
336 trace-cmd-13586 [003] 106467.860322: function: down_read_trylock <-- do_page_fault
337 Here is a nice way to find what interrupts have the highest latency:
339 # trace-cmd record -p function_graph -e irq_handler_entry -l do_IRQ sleep 10
341 # trace-cmd report
342 <idle>-0 [000] 157412.933969: funcgraph_entry: | do_IRQ() {
343 <idle>-0 [000] 157412.933974: irq_handler_entry: irq=48 name=eth0
344 <idle>-0 [000] 157412.934004: funcgraph_exit: + 36.358 us | }
345 <idle>-0 [000] 157413.895004: funcgraph_entry: | do_IRQ() {
346 <idle>-0 [000] 157413.895011: irq_handler_entry: irq=48 name=eth0
347 <idle>-0 [000] 157413.895026: funcgraph_exit: + 24.014 us | }
348 <idle>-0 [000] 157415.891762: funcgraph_entry: | do_IRQ() {
349 <idle>-0 [000] 157415.891769: irq_handler_entry: irq=48 name=eth0
350 <idle>-0 [000] 157415.891784: funcgraph_exit: + 22.928 us | }
351 <idle>-0 [000] 157415.934869: funcgraph_entry: | do_IRQ() {
352 <idle>-0 [000] 157415.934874: irq_handler_entry: irq=48 name=eth0
353 <idle>-0 [000] 157415.934906: funcgraph_exit: + 37.512 us | }
354 <idle>-0 [000] 157417.888373: funcgraph_entry: | do_IRQ() {
355 <idle>-0 [000] 157417.888381: irq_handler_entry: irq=48 name=eth0
356 <idle>-0 [000] 157417.888398: funcgraph_exit: + 25.943 us | }
357 An example of the profile:
359 # trace-cmd record --profile sleep 1
361 # trace-cmd report --profile --comm sleep
362 task: sleep-21611
363 Event: sched_switch:R (1) Total: 99442 Avg: 99442 Max: 99442 Min:99442
364 <stack> 1 total:99442 min:99442 max:99442 avg=99442
365 => ftrace_raw_event_sched_switch (0xffffffff8105f812)
366 => __schedule (0xffffffff8150810a)
367 => preempt_schedule (0xffffffff8150842e)
368 => ___preempt_schedule (0xffffffff81273354)
369 => cpu_stop_queue_work (0xffffffff810b03c5)
370 => stop_one_cpu (0xffffffff810b063b)
371 => sched_exec (0xffffffff8106136d)
372 => do_execve_common.isra.27 (0xffffffff81148c89)
373 => do_execve (0xffffffff811490b0)
374 => SyS_execve (0xffffffff811492c4)
375 => return_to_handler (0xffffffff8150e3c8)
376 => stub_execve (0xffffffff8150c699)
377 Event: sched_switch:S (1) Total: 1000506680 Avg: 1000506680 Max: 1000506680 Min:1000506680
378 <stack> 1 total:1000506680 min:1000506680 max:1000506680 avg=1000506680
379 => ftrace_raw_event_sched_switch (0xffffffff8105f812)
380 => __schedule (0xffffffff8150810a)
381 => schedule (0xffffffff815084b8)
382 => do_nanosleep (0xffffffff8150b22c)
383 => hrtimer_nanosleep (0xffffffff8108d647)
384 => SyS_nanosleep (0xffffffff8108d72c)
385 => return_to_handler (0xffffffff8150e3c8)
386 => tracesys_phase2 (0xffffffff8150c304)
387 Event: sched_wakeup:21611 (1) Total: 30326 Avg: 30326 Max: 30326 Min:30326
388 <stack> 1 total:30326 min:30326 max:30326 avg=30326
389 => ftrace_raw_event_sched_wakeup_template (0xffffffff8105f653)
390 => ttwu_do_wakeup (0xffffffff810606eb)
391 => ttwu_do_activate.constprop.124 (0xffffffff810607c8)
392 => try_to_wake_up (0xffffffff8106340a)
393
395 trace-cmd(1), trace-cmd-report(1), trace-cmd-start(1),
396 trace-cmd-stop(1), trace-cmd-extract(1), trace-cmd-reset(1),
397 trace-cmd-split(1), trace-cmd-list(1), trace-cmd-listen(1),
398 trace-cmd-profile(1)
399
401 Written by Steven Rostedt, <rostedt@goodmis.org[1]>
402
404 git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git
405
407 Copyright (C) 2010 Red Hat, Inc. Free use of this software is granted
408 under the terms of the GNU Public License (GPL).
409
411 1. rostedt@goodmis.org
412 mailto:rostedt@goodmis.org
413 04/11/2018 TRACE-CMD-RECORD(1)