1TRACE-CMD-SET(1) TRACE-CMD-SET(1)
2
6 trace-cmd-set - set a configuration parameter of the Ftrace Linux
7 internal tracer
8
10 trace-cmd set [OPTIONS] [command]
11
13 The trace-cmd(1) set command will set a configuration parameter of the
14 Ftrace Linux kernel tracer. The specified command will be run after the
15 ftrace state is set. The configured ftrace state can be restored to
16 default using the trace-cmd-reset(1) command.
17
19 -p tracer
20 Specify a tracer. Tracers usually do more than just trace an event.
21 Common tracers are: function, function_graph, preemptirqsoff,
22 irqsoff, preemptoff and wakeup. A tracer must be supported by the
23 running kernel. To see a list of available tracers, see
24 trace-cmd-list(1).
25 -e event
27 Specify an event to trace. Various static trace points have been
28 added to the Linux kernel. They are grouped by subsystem where you
29 can enable all events of a given subsystem or specify specific
30 events to be enabled. The event is of the format
31 "subsystem:event-name". You can also just specify the subsystem
32 without the :event-name or the event-name without the "subsystem:".
33 Using "-e sched_switch" will enable the "sched_switch" event where
34 as, "-e sched" will enable all events under the "sched" subsystem.
35 The 'event' can also contain glob expressions. That is, "*stat*" will
37 select all events (or subsystems) that have the characters "stat" in their
38 names.
39 The keyword 'all' can be used to enable all events.
41 -T
43 Enable a stacktrace on each event. For example:
44 <idle>-0 [003] 58549.289091: sched_switch: kworker/0:1:0 [120] R ==> trace-cmd:2603 [120]
46 <idle>-0 [003] 58549.289092: kernel_stack: <stack trace>
47 => schedule (ffffffff814b260e)
48 => cpu_idle (ffffffff8100a38c)
49 => start_secondary (ffffffff814ab828)
50 --func-stack
52 Enable a stack trace on all functions. Note this is only applicable
53 for the "function" plugin tracer, and will only take effect if the
54 -l option is used and succeeds in limiting functions. If the
55 function tracer is not filtered, and the stack trace is enabled,
56 you can live lock the machine.
57 -f filter
59 Specify a filter for the previous event. This must come after a -e.
60 This will filter what events get recorded based on the content of
61 the event. Filtering is passed to the kernel directly so what
62 filtering is allowed may depend on what version of the kernel you
63 have. Basically, it will let you use C notation to check if an
64 event should be processed or not.
65 ==, >=, <=, >, <, &, |, && and ||
67 The above are usually safe to use to compare fields.
69 -R trigger
71 Specify a trigger for the previous event. This must come after a
72 -e. This will add a given trigger to the given event. To only
73 enable the trigger and not the event itself, then place the event
74 after the -v option.
75 See Documentation/trace/events.txt in the Linux kernel source for more
77 information on triggers.
78 -v
80 This will negate options specified after it on the command line. It
81 affects:
82 -e: Causes all specified events to not be traced. This is useful for
84 selecting a subsystem to be traced but to leave out various events.
85 For example: "-e sched -v -e "*stat*"" will enable all events in
86 the sched subsystem except those that have "stat" in their names.
87 -B: Deletes the specified ftrace instance. There must be no
88 configuration options related to this instance in the command line.
89 For example: "-v -B bar -B foo" will delete instance bar and create
90 a new instance foo.
91 Note: the -v option was taken from the way grep(1) inverts the following
92 matches.
93 -P pid
95 This will filter only the specified process IDs. Using -P will let
96 you trace only events that are caused by the process.
97 -c
99 Used -P to trace the process' children too (if kernel supports it).
100 --user
102 Execute the specified command as given user.
103 -C clock
105 Set the trace clock to "clock".
106 Use trace-cmd(1) list -C to see what clocks are available.
108 -l function-name
110 This will limit the function and function_graph tracers to only
111 trace the given function name. More than one -l may be specified on
112 the command line to trace more than one function. The limited use
113 of glob expressions are also allowed. These are match* to only
114 filter functions that start with match. *match to only filter
115 functions that end with match. *match\* to only filter on
116 functions that contain match.
117 -g function-name
119 This option is for the function_graph plugin. It will graph the
120 given function. That is, it will only trace the function and all
121 functions that it calls. You can have more than one -g on the
122 command line.
123 -n function-name
125 This has the opposite effect of -l. The function given with the -n
126 option will not be traced. This takes precedence, that is, if you
127 include the same function for both -n and -l, it will not be
128 traced.
129 -d
131 Some tracer plugins enable the function tracer by default. Like the
132 latency tracers. This option prevents the function tracer from
133 being enabled at start up.
134 -D
136 The option -d will try to use the function-trace option to disable
137 the function tracer (if available), otherwise it defaults to the
138 proc file: /proc/sys/kernel/ftrace_enabled, but will not touch it
139 if the function-trace option is available. The -D option will
140 disable both the ftrace_enabled proc file as well as the
141 function-trace option if it exists.
142 Note, this disable function tracing for all users, which includes users
144 outside of ftrace tracers (stack_tracer, perf, etc).
145 -O option
147 Ftrace has various options that can be enabled or disabled. This
148 allows you to set them. Appending the text no to an option disables
149 it. For example: "-O nograph-time" will disable the "graph-time"
150 Ftrace option.
151 -b size
153 This sets the ring buffer size to size kilobytes. Because the
154 Ftrace ring buffer is per CPU, this size is the size of each per
155 CPU ring buffer inside the kernel. Using "-b 10000" on a machine
156 with 4 CPUs will make Ftrace have a total buffer size of 40 Megs.
157 -B buffer-name
159 If the kernel supports multiple buffers, this will add a buffer
160 with the given name. If the buffer name already exists, that buffer
161 is just reset.
162 After a buffer name is stated, all events added after that will be
164 associated with that buffer. If no buffer is specified, or an event
165 is specified before a buffer name, it will be associated with the
166 main (toplevel) buffer.
167 trace-cmd set -e sched -B block -e block -B time -e timer sleep 1
169 The above is will enable all sched events in the main buffer. It will
171 then create a 'block' buffer instance and enable all block events within
172 that buffer. A 'time' buffer instance is created and all timer events
173 will be enabled for that event.
174 -m size
176 The max size in kilobytes that a per cpu buffer should be. Note,
177 due to rounding to page size, the number may not be totally
178 correct. Also, this is performed by switching between two buffers
179 that are half the given size thus the output may not be of the
180 given size even if much more was written.
181 Use this to prevent running out of diskspace for long runs.
183 -M cpumask
185 Set the cpumask for to trace. It only affects the last buffer
186 instance given. If supplied before any buffer instance, then it
187 affects the main buffer. The value supplied must be a hex number.
188 trace-cmd set -p function -M c -B events13 -e all -M 5
190 If the -M is left out, then the mask stays the same. To enable all
192 CPUs, pass in a value of '-1'.
193 -i
195 By default, if an event is listed that trace-cmd does not find, it
196 will exit with an error. This option will just ignore events that
197 are listed on the command line but are not found on the system.
198 -q | --quiet
200 Suppresses normal output, except for errors.
201 --max-graph-depth depth
203 Set the maximum depth the function_graph tracer will trace into a
204 function. A value of one will only show where userspace enters the
205 kernel but not any functions called in the kernel. The default is
206 zero, which means no limit.
207 --cmdlines-size size
209 Set the number of entries the kernel tracing file "saved_cmdlines"
210 can contain. This file is a circular buffer which stores the
211 mapping between cmdlines and PIDs. If full, it leads to unresolved
212 cmdlines ("<...>") within the trace. The kernel default value is
213 128.
214 --module module
216 Filter a module’s name in function tracing. It is equivalent to
217 adding :mod:module after all other functions being filtered. If no
218 other function filter is listed, then all modules functions will be
219 filtered in the filter.
220 '--module snd' is equivalent to '-l :mod:snd'
222 '--module snd -l "*jack*"' is equivalent to '-l "*jack*:mod:snd"'
224 '--module snd -n "*"' is equivalent to '-n :mod:snd'
226 --stderr
228 Have output go to stderr instead of stdout, but the output of the
229 command executed will not be changed. This is useful if you want to
230 monitor the output of the command being executed, but not see the
231 output from trace-cmd.
232 --fork
234 If a command is listed, then trace-cmd will wait for that command
235 to finish, unless the --fork option is specified. Then it will fork
236 the command and return immediately.
237
239 Enable all events for tracing:
240 # trace-cmd set -e all
242 Set the function tracer:
244 # trace-cmd set -p function
246
248 trace-cmd(1), trace-cmd-report(1), trace-cmd-start(1),
249 trace-cmd-stop(1), trace-cmd-extract(1), trace-cmd-reset(1),
250 trace-cmd-split(1), trace-cmd-list(1), trace-cmd-listen(1),
251 trace-cmd-profile(1)
252
254 Written by Tzvetomir Stoyanov (VMware) <tz.stoyanov@gmail.com[1]>
255
257 https://git.kernel.org/pub/scm/utils/trace-cmd/trace-cmd.git/
258
260 Copyright (C) 2010 Red Hat, Inc. Free use of this software is granted
261 under the terms of the GNU Public License (GPL).
262
264 1. tz.stoyanov@gmail.com
265 mailto:tz.stoyanov@gmail.com
266 07/23/2021 TRACE-CMD-SET(1)