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-G:: 67--cgroup:: 68 Record events in threads in a cgroup. 69 70 Look for cgroups to set at the /sys/fs/cgroup/perf_event directory, then 71 remove the /sys/fs/cgroup/perf_event/ part and try: 72 73 perf trace -G A -e sched:*switch 74 75 Will set all raw_syscalls:sys_{enter,exit}, pgfault, vfs_getname, etc 76 _and_ sched:sched_switch to the 'A' cgroup, while: 77 78 perf trace -e sched:*switch -G A 79 80 will only set the sched:sched_switch event to the 'A' cgroup, all the 81 other events (raw_syscalls:sys_{enter,exit}, etc are left "without" 82 a cgroup (on the root cgroup, sys wide, etc). 83 84 Multiple cgroups: 85 86 perf trace -G A -e sched:*switch -G B 87 88 the syscall ones go to the 'A' cgroup, the sched:sched_switch goes 89 to the 'B' cgroup. 90 91--filter-pids=:: 92 Filter out events for these pids and for 'trace' itself (comma separated list). 93 94-v:: 95--verbose=:: 96 Verbosity level. 97 98--no-inherit:: 99 Child tasks do not inherit counters. 100 101-m:: 102--mmap-pages=:: 103 Number of mmap data pages (must be a power of two) or size 104 specification with appended unit character - B/K/M/G. The 105 size is rounded up to have nearest pages power of two value. 106 107-C:: 108--cpu:: 109Collect samples only on the list of CPUs provided. Multiple CPUs can be provided as a 110comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2. 111In per-thread mode with inheritance mode on (default), Events are captured only when 112the thread executes on the designated CPUs. Default is to monitor all CPUs. 113 114--duration:: 115 Show only events that had a duration greater than N.M ms. 116 117--sched:: 118 Accrue thread runtime and provide a summary at the end of the session. 119 120--failure:: 121 Show only syscalls that failed, i.e. that returned < 0. 122 123-i:: 124--input:: 125 Process events from a given perf data file. 126 127-T:: 128--time:: 129 Print full timestamp rather time relative to first sample. 130 131--comm:: 132 Show process COMM right beside its ID, on by default, disable with --no-comm. 133 134-s:: 135--summary:: 136 Show only a summary of syscalls by thread with min, max, and average times 137 (in msec) and relative stddev. 138 139-S:: 140--with-summary:: 141 Show all syscalls followed by a summary by thread with min, max, and 142 average times (in msec) and relative stddev. 143 144--tool_stats:: 145 Show tool stats such as number of times fd->pathname was discovered thru 146 hooking the open syscall return + vfs_getname or via reading /proc/pid/fd, etc. 147 148-f:: 149--force:: 150 Don't complain, do it. 151 152-F=[all|min|maj]:: 153--pf=[all|min|maj]:: 154 Trace pagefaults. Optionally, you can specify whether you want minor, 155 major or all pagefaults. Default value is maj. 156 157--syscalls:: 158 Trace system calls. This options is enabled by default, disable with 159 --no-syscalls. 160 161--call-graph [mode,type,min[,limit],order[,key][,branch]]:: 162 Setup and enable call-graph (stack chain/backtrace) recording. 163 See `--call-graph` section in perf-record and perf-report 164 man pages for details. The ones that are most useful in 'perf trace' 165 are 'dwarf' and 'lbr', where available, try: 'perf trace --call-graph dwarf'. 166 167 Using this will, for the root user, bump the value of --mmap-pages to 4 168 times the maximum for non-root users, based on the kernel.perf_event_mlock_kb 169 sysctl. This is done only if the user doesn't specify a --mmap-pages value. 170 171--kernel-syscall-graph:: 172 Show the kernel callchains on the syscall exit path. 173 174--max-events=N:: 175 Stop after processing N events. Note that strace-like events are considered 176 only at exit time or when a syscall is interrupted, i.e. in those cases this 177 option is equivalent to the number of lines printed. 178 179--max-stack:: 180 Set the stack depth limit when parsing the callchain, anything 181 beyond the specified depth will be ignored. Note that at this point 182 this is just about the presentation part, i.e. the kernel is still 183 not limiting, the overhead of callchains needs to be set via the 184 knobs in --call-graph dwarf. 185 186 Implies '--call-graph dwarf' when --call-graph not present on the 187 command line, on systems where DWARF unwinding was built in. 188 189 Default: /proc/sys/kernel/perf_event_max_stack when present for 190 live sessions (without --input/-i), 127 otherwise. 191 192--min-stack:: 193 Set the stack depth limit when parsing the callchain, anything 194 below the specified depth will be ignored. Disabled by default. 195 196 Implies '--call-graph dwarf' when --call-graph not present on the 197 command line, on systems where DWARF unwinding was built in. 198 199--print-sample:: 200 Print the PERF_RECORD_SAMPLE PERF_SAMPLE_ info for the 201 raw_syscalls:sys_{enter,exit} tracepoints, for debugging. 202 203--proc-map-timeout:: 204 When processing pre-existing threads /proc/XXX/mmap, it may take a long time, 205 because the file may be huge. A time out is needed in such cases. 206 This option sets the time out limit. The default value is 500 ms. 207 208PAGEFAULTS 209---------- 210 211When tracing pagefaults, the format of the trace is as follows: 212 213<min|maj>fault [<ip.symbol>+<ip.offset>] => <addr.dso@addr.offset> (<map type><addr level>). 214 215- min/maj indicates whether fault event is minor or major; 216- ip.symbol shows symbol for instruction pointer (the code that generated the 217 fault); if no debug symbols available, perf trace will print raw IP; 218- addr.dso shows DSO for the faulted address; 219- map type is either 'd' for non-executable maps or 'x' for executable maps; 220- addr level is either 'k' for kernel dso or '.' for user dso. 221 222For symbols resolution you may need to install debugging symbols. 223 224Please be aware that duration is currently always 0 and doesn't reflect actual 225time it took for fault to be handled! 226 227When --verbose specified, perf trace tries to print all available information 228for both IP and fault address in the form of dso@symbol+offset. 229 230EXAMPLES 231-------- 232 233Trace only major pagefaults: 234 235 $ perf trace --no-syscalls -F 236 237Trace syscalls, major and minor pagefaults: 238 239 $ perf trace -F all 240 241 1416.547 ( 0.000 ms): python/20235 majfault [CRYPTO_push_info_+0x0] => /lib/x86_64-linux-gnu/libcrypto.so.1.0.0@0x61be0 (x.) 242 243 As you can see, there was major pagefault in python process, from 244 CRYPTO_push_info_ routine which faulted somewhere in libcrypto.so. 245 246Trace the first 4 open, openat or open_by_handle_at syscalls (in the future more syscalls may match here): 247 248 $ perf trace -e open* --max-events 4 249 [root@jouet perf]# trace -e open* --max-events 4 250 2272.992 ( 0.037 ms): gnome-shell/1370 openat(dfd: CWD, filename: /proc/self/stat) = 31 251 2277.481 ( 0.139 ms): gnome-shell/3039 openat(dfd: CWD, filename: /proc/self/stat) = 65 252 3026.398 ( 0.076 ms): gnome-shell/3039 openat(dfd: CWD, filename: /proc/self/stat) = 65 253 4294.665 ( 0.015 ms): sed/15879 openat(dfd: CWD, filename: /etc/ld.so.cache, flags: CLOEXEC) = 3 254 $ 255 256Trace the first minor page fault when running a workload: 257 258 # perf trace -F min --max-stack=7 --max-events 1 sleep 1 259 0.000 ( 0.000 ms): sleep/18006 minfault [__clear_user+0x1a] => 0x5626efa56080 (?k) 260 __clear_user ([kernel.kallsyms]) 261 load_elf_binary ([kernel.kallsyms]) 262 search_binary_handler ([kernel.kallsyms]) 263 __do_execve_file.isra.33 ([kernel.kallsyms]) 264 __x64_sys_execve ([kernel.kallsyms]) 265 do_syscall_64 ([kernel.kallsyms]) 266 entry_SYSCALL_64 ([kernel.kallsyms]) 267 # 268 269Trace the next min page page fault to take place on the first CPU: 270 271 # perf trace -F min --call-graph=dwarf --max-events 1 --cpu 0 272 0.000 ( 0.000 ms): Web Content/17136 minfault [js::gc::Chunk::fetchNextDecommittedArena+0x4b] => 0x7fbe6181b000 (?.) 273 js::gc::FreeSpan::initAsEmpty (inlined) 274 js::gc::Arena::setAsNotAllocated (inlined) 275 js::gc::Chunk::fetchNextDecommittedArena (/usr/lib64/firefox/libxul.so) 276 js::gc::Chunk::allocateArena (/usr/lib64/firefox/libxul.so) 277 js::gc::GCRuntime::allocateArena (/usr/lib64/firefox/libxul.so) 278 js::gc::ArenaLists::allocateFromArena (/usr/lib64/firefox/libxul.so) 279 js::gc::GCRuntime::tryNewTenuredThing<JSString, (js::AllowGC)1> (inlined) 280 js::AllocateString<JSString, (js::AllowGC)1> (/usr/lib64/firefox/libxul.so) 281 js::Allocate<JSThinInlineString, (js::AllowGC)1> (inlined) 282 JSThinInlineString::new_<(js::AllowGC)1> (inlined) 283 AllocateInlineString<(js::AllowGC)1, unsigned char> (inlined) 284 js::ConcatStrings<(js::AllowGC)1> (/usr/lib64/firefox/libxul.so) 285 [0x18b26e6bc2bd] (/tmp/perf-17136.map) 286 # 287 288Trace the next two sched:sched_switch events, four block:*_plug events, the 289next block:*_unplug and the next three net:*dev_queue events, this last one 290with a backtrace of at most 16 entries, system wide: 291 292 # perf trace -e sched:*switch/nr=2/,block:*_plug/nr=4/,block:*_unplug/nr=1/,net:*dev_queue/nr=3,max-stack=16/ 293 0.000 :0/0 sched:sched_switch:swapper/2:0 [120] S ==> rcu_sched:10 [120] 294 0.015 rcu_sched/10 sched:sched_switch:rcu_sched:10 [120] R ==> swapper/2:0 [120] 295 254.198 irq/50-iwlwifi/680 net:net_dev_queue:dev=wlp3s0 skbaddr=0xffff93498051f600 len=66 296 __dev_queue_xmit ([kernel.kallsyms]) 297 273.977 :0/0 net:net_dev_queue:dev=wlp3s0 skbaddr=0xffff93498051f600 len=78 298 __dev_queue_xmit ([kernel.kallsyms]) 299 274.007 :0/0 net:net_dev_queue:dev=wlp3s0 skbaddr=0xffff93498051ff00 len=78 300 __dev_queue_xmit ([kernel.kallsyms]) 301 2930.140 kworker/u16:58/2722 block:block_plug:[kworker/u16:58] 302 2930.162 kworker/u16:58/2722 block:block_unplug:[kworker/u16:58] 1 303 4466.094 jbd2/dm-2-8/748 block:block_plug:[jbd2/dm-2-8] 304 8050.123 kworker/u16:30/2694 block:block_plug:[kworker/u16:30] 305 8050.271 kworker/u16:30/2694 block:block_plug:[kworker/u16:30] 306 # 307 308SEE ALSO 309-------- 310linkperf:perf-record[1], linkperf:perf-script[1] 311