Re: [patch] perf stat: --repeat forever
From: Arnaldo Carvalho de Melo
Date: Mon Feb 18 2013 - 10:04:58 EST
Em Fri, Feb 15, 2013 at 05:28:49PM -0500, Frederik Deweerdt escreveu:
> Hi,
>
> The following patch causes 'perf stat --repeat 0' to be interpreted as
> 'forever', displaying the stats for every run.
>
> We act as if a single run was asked, and reset the stats in each
> iteration. In this mode SIGINT is passed to perf to be able to stop the
> loop with Ctrl+C.
It is not applying to my perf/core branch, please take a look at:
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux
branch perf/core
Also please consider removing the need to specify 0, i.e. I think this
is possible and shorter:
# perf stat -a --repeat sleep 1
- Arnaldo
> Regards,
> Frederik
>
>
> Signed-off-by: Frederik Deweerdt <frederik.deweerdt@xxxxxxxxx>
>
> diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt
> index cf0c310..784976a 100644
> --- a/tools/perf/Documentation/perf-stat.txt
> +++ b/tools/perf/Documentation/perf-stat.txt
> @@ -52,7 +52,7 @@ OPTIONS
>
> -r::
> --repeat=<n>::
> - repeat command and print average + stddev (max: 100)
> + repeat command and print average + stddev (max: 100). 0 means forever.
>
> -B::
> --big-num::
> diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
> index c247fac..b02b3a9 100644
> --- a/tools/perf/builtin-stat.c
> +++ b/tools/perf/builtin-stat.c
> @@ -87,6 +87,7 @@ static FILE *output = NULL;
> static const char *pre_cmd = NULL;
> static const char *post_cmd = NULL;
> static bool sync_run = false;
> +static bool forever = false;
>
> static volatile int done = 0;
>
> @@ -94,6 +95,11 @@ struct perf_stat {
> struct stats res_stats[3];
> };
>
> +static void perf_evsel__reset_stat_priv(struct perf_evsel *evsel)
> +{
> + memset(evsel->priv, 0, sizeof(struct perf_stat));
> +}
> +
> static int perf_evsel__alloc_stat_priv(struct perf_evsel *evsel)
> {
> evsel->priv = zalloc(sizeof(struct perf_stat));
> @@ -129,6 +135,22 @@ static struct stats runtime_itlb_cache_stats[MAX_NR_CPUS];
> static struct stats runtime_dtlb_cache_stats[MAX_NR_CPUS];
> static struct stats walltime_nsecs_stats;
>
> +static void reset_stats(void)
> +{
> + memset(runtime_nsecs_stats, 0, sizeof(runtime_nsecs_stats));
> + memset(runtime_cycles_stats, 0, sizeof(runtime_cycles_stats));
> + memset(runtime_stalled_cycles_front_stats, 0, sizeof(runtime_stalled_cycles_front_stats));
> + memset(runtime_stalled_cycles_back_stats, 0, sizeof(runtime_stalled_cycles_back_stats));
> + memset(runtime_branches_stats, 0, sizeof(runtime_branches_stats));
> + memset(runtime_cacherefs_stats, 0, sizeof(runtime_cacherefs_stats));
> + memset(runtime_l1_dcache_stats, 0, sizeof(runtime_l1_dcache_stats));
> + memset(runtime_l1_icache_stats, 0, sizeof(runtime_l1_icache_stats));
> + memset(runtime_ll_cache_stats, 0, sizeof(runtime_ll_cache_stats));
> + memset(runtime_itlb_cache_stats, 0, sizeof(runtime_itlb_cache_stats));
> + memset(runtime_dtlb_cache_stats, 0, sizeof(runtime_dtlb_cache_stats));
> + memset(&walltime_nsecs_stats, 0, sizeof(walltime_nsecs_stats));
> +}
> +
> static int create_perf_stat_counter(struct perf_evsel *evsel)
> {
> struct perf_event_attr *attr = &evsel->attr;
> @@ -1120,7 +1142,7 @@ int cmd_stat(int argc, const char **argv, const char *prefix __maybe_unused)
> OPT_INCR('v', "verbose", &verbose,
> "be more verbose (show counter open errors, etc)"),
> OPT_INTEGER('r', "repeat", &run_count,
> - "repeat command and print average + stddev (max: 100)"),
> + "repeat command and print average + stddev (max: 100, forever: 0)"),
> OPT_BOOLEAN('n', "null", &null_run,
> "null run - dont start any counters"),
> OPT_INCR('d', "detailed", &detailed_run,
> @@ -1220,8 +1242,12 @@ int cmd_stat(int argc, const char **argv, const char *prefix __maybe_unused)
>
> if (!argc && !perf_target__has_task(&target))
> usage_with_options(stat_usage, options);
> - if (run_count <= 0)
> + if (run_count < 0) {
> usage_with_options(stat_usage, options);
> + } else if (run_count == 0) {
> + forever = true;
> + run_count = 1;
> + }
>
> /* no_aggr, cgroup are for system-wide only */
> if ((no_aggr || nr_cgroups) && !perf_target__has_cpu(&target)) {
> @@ -1259,21 +1285,31 @@ int cmd_stat(int argc, const char **argv, const char *prefix __maybe_unused)
> * task, but being ignored by perf stat itself:
> */
> atexit(sig_atexit);
> - signal(SIGINT, skip_signal);
> + if (!forever)
> + signal(SIGINT, skip_signal);
> signal(SIGALRM, skip_signal);
> signal(SIGABRT, skip_signal);
>
> status = 0;
> - for (run_idx = 0; run_idx < run_count; run_idx++) {
> + for (run_idx = 0; forever || run_idx < run_count; run_idx++) {
> if (run_count != 1 && verbose)
> fprintf(output, "[ perf stat: executing run #%d ... ]\n",
> - run_idx + 1);
> + run_idx + 1);
>
> status = run_perf_stat(argc, argv);
> + if (forever && status != -1) {
> + print_stat(argc, argv);
> + list_for_each_entry(pos, &evsel_list->entries, node) {
> + perf_evsel__reset_stat_priv(pos);
> + perf_evsel__reset_counts(pos, perf_evsel__nr_cpus(pos));
> + }
> + reset_stats();
> + }
> }
>
> - if (status != -1)
> + if (!forever && status != -1)
> print_stat(argc, argv);
> +
> out_free_fd:
> list_for_each_entry(pos, &evsel_list->entries, node)
> perf_evsel__free_stat_priv(pos);
> diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
> index 1b16dd1..0c49e79 100644
> --- a/tools/perf/util/evsel.c
> +++ b/tools/perf/util/evsel.c
> @@ -580,6 +580,12 @@ int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads)
> return 0;
> }
>
> +void perf_evsel__reset_counts(struct perf_evsel *evsel, int ncpus)
> +{
> + memset(evsel->counts, 0, (sizeof(*evsel->counts) +
> + (ncpus * sizeof(struct perf_counts_values))));
> +}
> +
> int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus)
> {
> evsel->counts = zalloc((sizeof(*evsel->counts) +
> diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h
> index 3d2b801..364d820 100644
> --- a/tools/perf/util/evsel.h
> +++ b/tools/perf/util/evsel.h
> @@ -114,6 +114,7 @@ const char *perf_evsel__name(struct perf_evsel *evsel);
> int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads);
> int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads);
> int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus);
> +void perf_evsel__reset_counts(struct perf_evsel *evsel, int ncpus);
> void perf_evsel__free_fd(struct perf_evsel *evsel);
> void perf_evsel__free_id(struct perf_evsel *evsel);
> void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads);
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/