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 69-D:: 70--no-delay:: 71 Collect data without buffering. 72 73-c:: 74--count=:: 75 Event period to sample. 76 77-o:: 78--output=:: 79 Output file name. 80 81-i:: 82--no-inherit:: 83 Child tasks do not inherit counters. 84-F:: 85--freq=:: 86 Profile at this frequency. 87 88-m:: 89--mmap-pages=:: 90 Number of mmap data pages (must be a power of two) or size 91 specification with appended unit character - B/K/M/G. The 92 size is rounded up to have nearest pages power of two value. 93 94-g:: 95 Enables call-graph (stack chain/backtrace) recording. 96 97--call-graph:: 98 Setup and enable call-graph (stack chain/backtrace) recording, 99 implies -g. 100 101 Allows specifying "fp" (frame pointer) or "dwarf" 102 (DWARF's CFI - Call Frame Information) as the method to collect 103 the information used to show the call graphs. 104 105 In some systems, where binaries are build with gcc 106 --fomit-frame-pointer, using the "fp" method will produce bogus 107 call graphs, using "dwarf", if available (perf tools linked to 108 the libunwind library) should be used instead. 109 110-q:: 111--quiet:: 112 Don't print any message, useful for scripting. 113 114-v:: 115--verbose:: 116 Be more verbose (show counter open errors, etc). 117 118-s:: 119--stat:: 120 Per thread counts. 121 122-d:: 123--data:: 124 Sample addresses. 125 126-T:: 127--timestamp:: 128 Sample timestamps. Use it with 'perf report -D' to see the timestamps, 129 for instance. 130 131-n:: 132--no-samples:: 133 Don't sample. 134 135-R:: 136--raw-samples:: 137Collect raw sample records from all opened counters (default for tracepoint counters). 138 139-C:: 140--cpu:: 141Collect samples only on the list of CPUs provided. Multiple CPUs can be provided as a 142comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2. 143In per-thread mode with inheritance mode on (default), samples are captured only when 144the thread executes on the designated CPUs. Default is to monitor all CPUs. 145 146-N:: 147--no-buildid-cache:: 148Do not update the builid cache. This saves some overhead in situations 149where the information in the perf.data file (which includes buildids) 150is sufficient. 151 152-G name,...:: 153--cgroup name,...:: 154monitor only in the container (cgroup) called "name". This option is available only 155in per-cpu mode. The cgroup filesystem must be mounted. All threads belonging to 156container "name" are monitored when they run on the monitored CPUs. Multiple cgroups 157can be provided. Each cgroup is applied to the corresponding event, i.e., first cgroup 158to first event, second cgroup to second event and so on. It is possible to provide 159an empty cgroup (monitor all the time) using, e.g., -G foo,,bar. Cgroups must have 160corresponding events, i.e., they always refer to events defined earlier on the command 161line. 162 163-b:: 164--branch-any:: 165Enable taken branch stack sampling. Any type of taken branch may be sampled. 166This is a shortcut for --branch-filter any. See --branch-filter for more infos. 167 168-j:: 169--branch-filter:: 170Enable taken branch stack sampling. Each sample captures a series of consecutive 171taken branches. The number of branches captured with each sample depends on the 172underlying hardware, the type of branches of interest, and the executed code. 173It is possible to select the types of branches captured by enabling filters. The 174following filters are defined: 175 176 - any: any type of branches 177 - any_call: any function call or system call 178 - any_ret: any function return or system call return 179 - ind_call: any indirect branch 180 - u: only when the branch target is at the user level 181 - k: only when the branch target is in the kernel 182 - hv: only when the target is at the hypervisor level 183 - in_tx: only when the target is in a hardware transaction 184 - no_tx: only when the target is not in a hardware transaction 185 - abort_tx: only when the target is a hardware transaction abort 186 187+ 188The option requires at least one branch type among any, any_call, any_ret, ind_call. 189The privilege levels may be omitted, in which case, the privilege levels of the associated 190event are applied to the branch filter. Both kernel (k) and hypervisor (hv) privilege 191levels are subject to permissions. When sampling on multiple events, branch stack sampling 192is enabled for all the sampling events. The sampled branch type is the same for all events. 193The various filters must be specified as a comma separated list: --branch-filter any_ret,u,k 194Note that this feature may not be available on all processors. 195 196--weight:: 197Enable weightened sampling. An additional weight is recorded per sample and can be 198displayed with the weight and local_weight sort keys. This currently works for TSX 199abort events and some memory events in precise mode on modern Intel CPUs. 200 201--transaction:: 202Record transaction flags for transaction related events. 203 204--force-per-cpu:: 205Force the use of per-cpu mmaps. By default, when tasks are specified (i.e. -p, 206-t or -u options) per-thread mmaps are created. This option overrides that and 207forces per-cpu mmaps. A side-effect of that is that inheritance is 208automatically enabled. Add the -i option also to disable inheritance. 209 210SEE ALSO 211-------- 212linkperf:perf-stat[1], linkperf:perf-list[1] 213