xref: /linux/tools/perf/Documentation/perf-trace.txt (revision c0e297dc61f8d4453e07afbea1fa8d0e67cd4a34)
1perf-trace(1)
2=============
3
4NAME
5----
6perf-trace - strace inspired tool
7
8SYNOPSIS
9--------
10[verse]
11'perf trace'
12'perf trace record'
13
14DESCRIPTION
15-----------
16This command will show the events associated with the target, initially
17syscalls, but other system events like pagefaults, task lifetime events,
18scheduling events, etc.
19
20This is a live mode tool in addition to working with perf.data files like
21the other perf tools. Files can be generated using the 'perf record' command
22but the session needs to include the raw_syscalls events (-e 'raw_syscalls:*').
23Alternatively, 'perf trace record' can be used as a shortcut to
24automatically include the raw_syscalls events when writing events to a file.
25
26The following options apply to perf trace; options to perf trace record are
27found in the perf record man page.
28
29OPTIONS
30-------
31
32-a::
33--all-cpus::
34        System-wide collection from all CPUs.
35
36-e::
37--expr::
38	List of syscalls to show, currently only syscall names.
39	Prefixing with ! shows all syscalls but the ones specified.  You may
40	need to escape it.
41
42-o::
43--output=::
44	Output file name.
45
46-p::
47--pid=::
48	Record events on existing process ID (comma separated list).
49
50-t::
51--tid=::
52        Record events on existing thread ID (comma separated list).
53
54-u::
55--uid=::
56        Record events in threads owned by uid. Name or number.
57
58--filter-pids=::
59	Filter out events for these pids and for 'trace' itself (comma separated list).
60
61-v::
62--verbose=::
63        Verbosity level.
64
65-i::
66--no-inherit::
67	Child tasks do not inherit counters.
68
69-m::
70--mmap-pages=::
71	Number of mmap data pages (must be a power of two) or size
72	specification with appended unit character - B/K/M/G. The
73	size is rounded up to have nearest pages power of two value.
74
75-C::
76--cpu::
77Collect samples only on the list of CPUs provided. Multiple CPUs can be provided as a
78comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2.
79In per-thread mode with inheritance mode on (default), Events are captured only when
80the thread executes on the designated CPUs. Default is to monitor all CPUs.
81
82--duration:
83	Show only events that had a duration greater than N.M ms.
84
85--sched:
86	Accrue thread runtime and provide a summary at the end of the session.
87
88-i
89--input
90	Process events from a given perf data file.
91
92-T
93--time
94	Print full timestamp rather time relative to first sample.
95
96--comm::
97        Show process COMM right beside its ID, on by default, disable with --no-comm.
98
99-s::
100--summary::
101	Show only a summary of syscalls by thread with min, max, and average times
102    (in msec) and relative stddev.
103
104-S::
105--with-summary::
106	Show all syscalls followed by a summary by thread with min, max, and
107    average times (in msec) and relative stddev.
108
109--tool_stats::
110	Show tool stats such as number of times fd->pathname was discovered thru
111	hooking the open syscall return + vfs_getname or via reading /proc/pid/fd, etc.
112
113-F=[all|min|maj]::
114--pf=[all|min|maj]::
115	Trace pagefaults. Optionally, you can specify whether you want minor,
116	major or all pagefaults. Default value is maj.
117
118--syscalls::
119	Trace system calls. This options is enabled by default.
120
121--event::
122	Trace other events, see 'perf list' for a complete list.
123
124--proc-map-timeout::
125	When processing pre-existing threads /proc/XXX/mmap, it may take a long time,
126	because the file may be huge. A time out is needed in such cases.
127	This option sets the time out limit. The default value is 500 ms.
128
129PAGEFAULTS
130----------
131
132When tracing pagefaults, the format of the trace is as follows:
133
134<min|maj>fault [<ip.symbol>+<ip.offset>] => <addr.dso@addr.offset> (<map type><addr level>).
135
136- min/maj indicates whether fault event is minor or major;
137- ip.symbol shows symbol for instruction pointer (the code that generated the
138  fault); if no debug symbols available, perf trace will print raw IP;
139- addr.dso shows DSO for the faulted address;
140- map type is either 'd' for non-executable maps or 'x' for executable maps;
141- addr level is either 'k' for kernel dso or '.' for user dso.
142
143For symbols resolution you may need to install debugging symbols.
144
145Please be aware that duration is currently always 0 and doesn't reflect actual
146time it took for fault to be handled!
147
148When --verbose specified, perf trace tries to print all available information
149for both IP and fault address in the form of dso@symbol+offset.
150
151EXAMPLES
152--------
153
154Trace only major pagefaults:
155
156 $ perf trace --no-syscalls -F
157
158Trace syscalls, major and minor pagefaults:
159
160 $ perf trace -F all
161
162  1416.547 ( 0.000 ms): python/20235 majfault [CRYPTO_push_info_+0x0] => /lib/x86_64-linux-gnu/libcrypto.so.1.0.0@0x61be0 (x.)
163
164  As you can see, there was major pagefault in python process, from
165  CRYPTO_push_info_ routine which faulted somewhere in libcrypto.so.
166
167SEE ALSO
168--------
169linkperf:perf-record[1], linkperf:perf-script[1]
170