1perf-record(1) 2============== 3 4NAME 5---- 6perf-record - Run a command and record its profile into perf.data 7 8SYNOPSIS 9-------- 10[verse] 11'perf record' [-e <EVENT> | --event=EVENT] [-l] [-a] <command> 12'perf record' [-e <EVENT> | --event=EVENT] [-l] [-a] -- <command> [<options>] 13 14DESCRIPTION 15----------- 16This command runs a command and gathers a performance counter profile 17from it, into perf.data - without displaying anything. 18 19This file can then be inspected later on, using 'perf report'. 20 21 22OPTIONS 23------- 24<command>...:: 25 Any command you can specify in a shell. 26 27-e:: 28--event=:: 29 Select the PMU event. Selection can be: 30 31 - a symbolic event name (use 'perf list' to list all events) 32 33 - a raw PMU event (eventsel+umask) in the form of rNNN where NNN is a 34 hexadecimal event descriptor. 35 36 - a hardware breakpoint event in the form of '\mem:addr[:access]' 37 where addr is the address in memory you want to break in. 38 Access is the memory access type (read, write, execute) it can 39 be passed as follows: '\mem:addr[:[r][w][x]]'. 40 If you want to profile read-write accesses in 0x1000, just set 41 'mem:0x1000:rw'. 42 43--filter=<filter>:: 44 Event filter. 45 46-a:: 47--all-cpus:: 48 System-wide collection from all CPUs. 49 50-l:: 51 Scale counter values. 52 53-p:: 54--pid=:: 55 Record events on existing process ID (comma separated list). 56 57-t:: 58--tid=:: 59 Record events on existing thread ID (comma separated list). 60 61-u:: 62--uid=:: 63 Record events in threads owned by uid. Name or number. 64 65-r:: 66--realtime=:: 67 Collect data with this RT SCHED_FIFO priority. 68-D:: 69--no-delay:: 70 Collect data without buffering. 71-A:: 72--append:: 73 Append to the output file to do incremental profiling. 74 75-f:: 76--force:: 77 Overwrite existing data file. (deprecated) 78 79-c:: 80--count=:: 81 Event period to sample. 82 83-o:: 84--output=:: 85 Output file name. 86 87-i:: 88--no-inherit:: 89 Child tasks do not inherit counters. 90-F:: 91--freq=:: 92 Profile at this frequency. 93 94-m:: 95--mmap-pages=:: 96 Number of mmap data pages. Must be a power of two. 97 98-g:: 99--call-graph:: 100 Do call-graph (stack chain/backtrace) recording. 101 102-q:: 103--quiet:: 104 Don't print any message, useful for scripting. 105 106-v:: 107--verbose:: 108 Be more verbose (show counter open errors, etc). 109 110-s:: 111--stat:: 112 Per thread counts. 113 114-d:: 115--data:: 116 Sample addresses. 117 118-T:: 119--timestamp:: 120 Sample timestamps. Use it with 'perf report -D' to see the timestamps, 121 for instance. 122 123-n:: 124--no-samples:: 125 Don't sample. 126 127-R:: 128--raw-samples:: 129Collect raw sample records from all opened counters (default for tracepoint counters). 130 131-C:: 132--cpu:: 133Collect samples only on the list of CPUs provided. Multiple CPUs can be provided as a 134comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2. 135In per-thread mode with inheritance mode on (default), samples are captured only when 136the thread executes on the designated CPUs. Default is to monitor all CPUs. 137 138-N:: 139--no-buildid-cache:: 140Do not update the builid cache. This saves some overhead in situations 141where the information in the perf.data file (which includes buildids) 142is sufficient. 143 144-G name,...:: 145--cgroup name,...:: 146monitor only in the container (cgroup) called "name". This option is available only 147in per-cpu mode. The cgroup filesystem must be mounted. All threads belonging to 148container "name" are monitored when they run on the monitored CPUs. Multiple cgroups 149can be provided. Each cgroup is applied to the corresponding event, i.e., first cgroup 150to first event, second cgroup to second event and so on. It is possible to provide 151an empty cgroup (monitor all the time) using, e.g., -G foo,,bar. Cgroups must have 152corresponding events, i.e., they always refer to events defined earlier on the command 153line. 154 155-b:: 156--branch-any:: 157Enable taken branch stack sampling. Any type of taken branch may be sampled. 158This is a shortcut for --branch-filter any. See --branch-filter for more infos. 159 160-j:: 161--branch-filter:: 162Enable taken branch stack sampling. Each sample captures a series of consecutive 163taken branches. The number of branches captured with each sample depends on the 164underlying hardware, the type of branches of interest, and the executed code. 165It is possible to select the types of branches captured by enabling filters. The 166following filters are defined: 167 168 - any: any type of branches 169 - any_call: any function call or system call 170 - any_ret: any function return or system call return 171 - ind_call: any indirect branch 172 - u: only when the branch target is at the user level 173 - k: only when the branch target is in the kernel 174 - hv: only when the target is at the hypervisor level 175 176+ 177The option requires at least one branch type among any, any_call, any_ret, ind_call. 178The privilege levels may be ommitted, in which case, the privilege levels of the associated 179event are applied to the branch filter. Both kernel (k) and hypervisor (hv) privilege 180levels are subject to permissions. When sampling on multiple events, branch stack sampling 181is enabled for all the sampling events. The sampled branch type is the same for all events. 182The various filters must be specified as a comma separated list: --branch-filter any_ret,u,k 183Note that this feature may not be available on all processors. 184 185SEE ALSO 186-------- 187linkperf:perf-stat[1], linkperf:perf-list[1] 188