xref: /linux/tools/perf/Documentation/perf-trace.txt (revision e0bf6c5ca2d3281f231c5f0c9bf145e9513644de)
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 events 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-v::
59--verbose=::
60        Verbosity level.
61
62-i::
63--no-inherit::
64	Child tasks do not inherit counters.
65
66-m::
67--mmap-pages=::
68	Number of mmap data pages (must be a power of two) or size
69	specification with appended unit character - B/K/M/G. The
70	size is rounded up to have nearest pages power of two value.
71
72-C::
73--cpu::
74Collect samples only on the list of CPUs provided. Multiple CPUs can be provided as a
75comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2.
76In per-thread mode with inheritance mode on (default), Events are captured only when
77the thread executes on the designated CPUs. Default is to monitor all CPUs.
78
79--duration:
80	Show only events that had a duration greater than N.M ms.
81
82--sched:
83	Accrue thread runtime and provide a summary at the end of the session.
84
85-i
86--input
87	Process events from a given perf data file.
88
89-T
90--time
91	Print full timestamp rather time relative to first sample.
92
93--comm::
94        Show process COMM right beside its ID, on by default, disable with --no-comm.
95
96-s::
97--summary::
98	Show only a summary of syscalls by thread with min, max, and average times
99    (in msec) and relative stddev.
100
101-S::
102--with-summary::
103	Show all syscalls followed by a summary by thread with min, max, and
104    average times (in msec) and relative stddev.
105
106--tool_stats::
107	Show tool stats such as number of times fd->pathname was discovered thru
108	hooking the open syscall return + vfs_getname or via reading /proc/pid/fd, etc.
109
110-F=[all|min|maj]::
111--pf=[all|min|maj]::
112	Trace pagefaults. Optionally, you can specify whether you want minor,
113	major or all pagefaults. Default value is maj.
114
115--syscalls::
116	Trace system calls. This options is enabled by default.
117
118PAGEFAULTS
119----------
120
121When tracing pagefaults, the format of the trace is as follows:
122
123<min|maj>fault [<ip.symbol>+<ip.offset>] => <addr.dso@addr.offset> (<map type><addr level>).
124
125- min/maj indicates whether fault event is minor or major;
126- ip.symbol shows symbol for instruction pointer (the code that generated the
127  fault); if no debug symbols available, perf trace will print raw IP;
128- addr.dso shows DSO for the faulted address;
129- map type is either 'd' for non-executable maps or 'x' for executable maps;
130- addr level is either 'k' for kernel dso or '.' for user dso.
131
132For symbols resolution you may need to install debugging symbols.
133
134Please be aware that duration is currently always 0 and doesn't reflect actual
135time it took for fault to be handled!
136
137When --verbose specified, perf trace tries to print all available information
138for both IP and fault address in the form of dso@symbol+offset.
139
140EXAMPLES
141--------
142
143Trace only major pagefaults:
144
145 $ perf trace --no-syscalls -F
146
147Trace syscalls, major and minor pagefaults:
148
149 $ perf trace -F all
150
151  1416.547 ( 0.000 ms): python/20235 majfault [CRYPTO_push_info_+0x0] => /lib/x86_64-linux-gnu/libcrypto.so.1.0.0@0x61be0 (x.)
152
153  As you can see, there was major pagefault in python process, from
154  CRYPTO_push_info_ routine which faulted somewhere in libcrypto.so.
155
156SEE ALSO
157--------
158linkperf:perf-record[1], linkperf:perf-script[1]
159