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--event:: 39 List of syscalls and other perf events (tracepoints, HW cache events, 40 etc) to show. Globbing is supported, e.g.: "epoll_*", "*msg*", etc. 41 See 'perf list' for a complete list of events. 42 Prefixing with ! shows all syscalls but the ones specified. You may 43 need to escape it. 44 45-D msecs:: 46--delay msecs:: 47After starting the program, wait msecs before measuring. This is useful to 48filter out the startup phase of the program, which is often very different. 49 50-o:: 51--output=:: 52 Output file name. 53 54-p:: 55--pid=:: 56 Record events on existing process ID (comma separated list). 57 58-t:: 59--tid=:: 60 Record events on existing thread ID (comma separated list). 61 62-u:: 63--uid=:: 64 Record events in threads owned by uid. Name or number. 65 66--filter-pids=:: 67 Filter out events for these pids and for 'trace' itself (comma separated list). 68 69-v:: 70--verbose=:: 71 Verbosity level. 72 73--no-inherit:: 74 Child tasks do not inherit counters. 75 76-m:: 77--mmap-pages=:: 78 Number of mmap data pages (must be a power of two) or size 79 specification with appended unit character - B/K/M/G. The 80 size is rounded up to have nearest pages power of two value. 81 82-C:: 83--cpu:: 84Collect samples only on the list of CPUs provided. Multiple CPUs can be provided as a 85comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2. 86In per-thread mode with inheritance mode on (default), Events are captured only when 87the thread executes on the designated CPUs. Default is to monitor all CPUs. 88 89--duration:: 90 Show only events that had a duration greater than N.M ms. 91 92--sched:: 93 Accrue thread runtime and provide a summary at the end of the session. 94 95-i:: 96--input:: 97 Process events from a given perf data file. 98 99-T:: 100--time:: 101 Print full timestamp rather time relative to first sample. 102 103--comm:: 104 Show process COMM right beside its ID, on by default, disable with --no-comm. 105 106-s:: 107--summary:: 108 Show only a summary of syscalls by thread with min, max, and average times 109 (in msec) and relative stddev. 110 111-S:: 112--with-summary:: 113 Show all syscalls followed by a summary by thread with min, max, and 114 average times (in msec) and relative stddev. 115 116--tool_stats:: 117 Show tool stats such as number of times fd->pathname was discovered thru 118 hooking the open syscall return + vfs_getname or via reading /proc/pid/fd, etc. 119 120-f:: 121--force:: 122 Don't complain, do it. 123 124-F=[all|min|maj]:: 125--pf=[all|min|maj]:: 126 Trace pagefaults. Optionally, you can specify whether you want minor, 127 major or all pagefaults. Default value is maj. 128 129--syscalls:: 130 Trace system calls. This options is enabled by default, disable with 131 --no-syscalls. 132 133--call-graph [mode,type,min[,limit],order[,key][,branch]]:: 134 Setup and enable call-graph (stack chain/backtrace) recording. 135 See `--call-graph` section in perf-record and perf-report 136 man pages for details. The ones that are most useful in 'perf trace' 137 are 'dwarf' and 'lbr', where available, try: 'perf trace --call-graph dwarf'. 138 139 Using this will, for the root user, bump the value of --mmap-pages to 4 140 times the maximum for non-root users, based on the kernel.perf_event_mlock_kb 141 sysctl. This is done only if the user doesn't specify a --mmap-pages value. 142 143--kernel-syscall-graph:: 144 Show the kernel callchains on the syscall exit path. 145 146--max-stack:: 147 Set the stack depth limit when parsing the callchain, anything 148 beyond the specified depth will be ignored. Note that at this point 149 this is just about the presentation part, i.e. the kernel is still 150 not limiting, the overhead of callchains needs to be set via the 151 knobs in --call-graph dwarf. 152 153 Implies '--call-graph dwarf' when --call-graph not present on the 154 command line, on systems where DWARF unwinding was built in. 155 156 Default: /proc/sys/kernel/perf_event_max_stack when present for 157 live sessions (without --input/-i), 127 otherwise. 158 159--min-stack:: 160 Set the stack depth limit when parsing the callchain, anything 161 below the specified depth will be ignored. Disabled by default. 162 163 Implies '--call-graph dwarf' when --call-graph not present on the 164 command line, on systems where DWARF unwinding was built in. 165 166--print-sample:: 167 Print the PERF_RECORD_SAMPLE PERF_SAMPLE_ info for the 168 raw_syscalls:sys_{enter,exit} tracepoints, for debugging. 169 170--proc-map-timeout:: 171 When processing pre-existing threads /proc/XXX/mmap, it may take a long time, 172 because the file may be huge. A time out is needed in such cases. 173 This option sets the time out limit. The default value is 500 ms. 174 175PAGEFAULTS 176---------- 177 178When tracing pagefaults, the format of the trace is as follows: 179 180<min|maj>fault [<ip.symbol>+<ip.offset>] => <addr.dso@addr.offset> (<map type><addr level>). 181 182- min/maj indicates whether fault event is minor or major; 183- ip.symbol shows symbol for instruction pointer (the code that generated the 184 fault); if no debug symbols available, perf trace will print raw IP; 185- addr.dso shows DSO for the faulted address; 186- map type is either 'd' for non-executable maps or 'x' for executable maps; 187- addr level is either 'k' for kernel dso or '.' for user dso. 188 189For symbols resolution you may need to install debugging symbols. 190 191Please be aware that duration is currently always 0 and doesn't reflect actual 192time it took for fault to be handled! 193 194When --verbose specified, perf trace tries to print all available information 195for both IP and fault address in the form of dso@symbol+offset. 196 197EXAMPLES 198-------- 199 200Trace only major pagefaults: 201 202 $ perf trace --no-syscalls -F 203 204Trace syscalls, major and minor pagefaults: 205 206 $ perf trace -F all 207 208 1416.547 ( 0.000 ms): python/20235 majfault [CRYPTO_push_info_+0x0] => /lib/x86_64-linux-gnu/libcrypto.so.1.0.0@0x61be0 (x.) 209 210 As you can see, there was major pagefault in python process, from 211 CRYPTO_push_info_ routine which faulted somewhere in libcrypto.so. 212 213SEE ALSO 214-------- 215linkperf:perf-record[1], linkperf:perf-script[1] 216