Loading tools/perf/Documentation/perf-periodic.txt 0 → 100644 +76 −0 Original line number Diff line number Diff line perf-periodic(1) ============ NAME ---- perf-periodic - Periodically print the performance counters at given rate SYNOPSIS -------- [verse] 'perf periodic' [-e <EVENT> | --event=EVENT] [-p pid] [<options>] DESCRIPTION ----------- This command gathers performance counter statistics and periodically prints out the counter values. OPTIONS ------- -e:: --event=:: Select the PMU event. Selection can be a symbolic event name (use 'perf list' to list all events) or a raw PMU event (eventsel+umask) in the form of rNNN where NNN is a hexadecimal event descriptor. -a:: --all-cpus:: system-wide collection from all CPUs, overrides cpumask -c cpumask:: Count only on the list of CPUs provided. Multiple CPUs can be provided as a mask: 1 = cpu0, 3 = cpu0 and cpu1 etc. Default is to count on all CPUs. -d:: Delta. Display results as a delta from the previous value. Default is to show the total acculumated value. -i iterations:: Number of times to iterate. Default 0-forever. -m operations:: Perform math operations on results. m=multiply, d=divide, a=add, s=subtract, z=zero, t=transfer, T=total. -p:: --pid=<pid>:: stat events on existing process id (comma separated list) -r fifo_name:: Specify read FIFO name for commands, such as PID to pass pid to connect to. -s delay:: Time to sleep between each sample, in milliseconds. Default 1000 (1 second). -v:: Verbose. Show more information (counter open errors etc). EXAMPLES -------- $ perf periodic -d -e r3 -e r4 -i 3 122640 2570103 108397 2159418 133372 2826949 $ perf periodic -d -e r3 -e r4 -i 3 -m td 0.032174 0.050263 0.031349 SEE ALSO -------- linkperf:perf-stat[1] tools/perf/builtin-periodic.c +3 −3 Original line number Diff line number Diff line /* * Copyright (c) 2011-2012, The Linux Foundation. All rights reserved. * Copyright (c) 2011-2013, The Linux Foundation. All rights reserved. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License version 2 and Loading Loading @@ -335,7 +335,7 @@ static void do_math_op(struct perf_setup_s *p) } } int cmd_periodic(int argc, const char **argv, const char *prefix __used) int cmd_periodic(int argc, const char **argv, const char *prefix __maybe_unused) { int status = 0; int c, i; Loading @@ -344,7 +344,7 @@ int cmd_periodic(int argc, const char **argv, const char *prefix __used) FILE *fp; int nr_counters = 0; evsel_list = perf_evlist__new(NULL, NULL); evsel_list = perf_evlist__new(); if (evsel_list == NULL) return -ENOMEM; Loading tools/perf/command-list.txt +1 −0 Original line number Diff line number Diff line Loading @@ -15,6 +15,7 @@ perf-kvm mainporcelain common perf-list mainporcelain common perf-lock mainporcelain common perf-mem mainporcelain common perf-periodic mainporcelain common perf-probe mainporcelain full perf-record mainporcelain common perf-report mainporcelain common Loading Loading
tools/perf/Documentation/perf-periodic.txt 0 → 100644 +76 −0 Original line number Diff line number Diff line perf-periodic(1) ============ NAME ---- perf-periodic - Periodically print the performance counters at given rate SYNOPSIS -------- [verse] 'perf periodic' [-e <EVENT> | --event=EVENT] [-p pid] [<options>] DESCRIPTION ----------- This command gathers performance counter statistics and periodically prints out the counter values. OPTIONS ------- -e:: --event=:: Select the PMU event. Selection can be a symbolic event name (use 'perf list' to list all events) or a raw PMU event (eventsel+umask) in the form of rNNN where NNN is a hexadecimal event descriptor. -a:: --all-cpus:: system-wide collection from all CPUs, overrides cpumask -c cpumask:: Count only on the list of CPUs provided. Multiple CPUs can be provided as a mask: 1 = cpu0, 3 = cpu0 and cpu1 etc. Default is to count on all CPUs. -d:: Delta. Display results as a delta from the previous value. Default is to show the total acculumated value. -i iterations:: Number of times to iterate. Default 0-forever. -m operations:: Perform math operations on results. m=multiply, d=divide, a=add, s=subtract, z=zero, t=transfer, T=total. -p:: --pid=<pid>:: stat events on existing process id (comma separated list) -r fifo_name:: Specify read FIFO name for commands, such as PID to pass pid to connect to. -s delay:: Time to sleep between each sample, in milliseconds. Default 1000 (1 second). -v:: Verbose. Show more information (counter open errors etc). EXAMPLES -------- $ perf periodic -d -e r3 -e r4 -i 3 122640 2570103 108397 2159418 133372 2826949 $ perf periodic -d -e r3 -e r4 -i 3 -m td 0.032174 0.050263 0.031349 SEE ALSO -------- linkperf:perf-stat[1]
tools/perf/builtin-periodic.c +3 −3 Original line number Diff line number Diff line /* * Copyright (c) 2011-2012, The Linux Foundation. All rights reserved. * Copyright (c) 2011-2013, The Linux Foundation. All rights reserved. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License version 2 and Loading Loading @@ -335,7 +335,7 @@ static void do_math_op(struct perf_setup_s *p) } } int cmd_periodic(int argc, const char **argv, const char *prefix __used) int cmd_periodic(int argc, const char **argv, const char *prefix __maybe_unused) { int status = 0; int c, i; Loading @@ -344,7 +344,7 @@ int cmd_periodic(int argc, const char **argv, const char *prefix __used) FILE *fp; int nr_counters = 0; evsel_list = perf_evlist__new(NULL, NULL); evsel_list = perf_evlist__new(); if (evsel_list == NULL) return -ENOMEM; Loading
tools/perf/command-list.txt +1 −0 Original line number Diff line number Diff line Loading @@ -15,6 +15,7 @@ perf-kvm mainporcelain common perf-list mainporcelain common perf-lock mainporcelain common perf-mem mainporcelain common perf-periodic mainporcelain common perf-probe mainporcelain full perf-record mainporcelain common perf-report mainporcelain common Loading
