[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAM9d7cgEJ=0Y_Y42gN0RN95YXZPzWfXmeLKuNYqJeSWFMbQ5fA@mail.gmail.com>
Date: Mon, 19 Jun 2023 13:00:44 -0700
From: Namhyung Kim <namhyung@...nel.org>
To: Jiri Olsa <olsajiri@...il.com>
Cc: Arnaldo Carvalho de Melo <acme@...nel.org>,
Ian Rogers <irogers@...gle.com>,
Adrian Hunter <adrian.hunter@...el.com>,
Peter Zijlstra <peterz@...radead.org>,
Ingo Molnar <mingo@...nel.org>,
LKML <linux-kernel@...r.kernel.org>,
linux-perf-users@...r.kernel.org,
Kan Liang <kan.liang@...ux.intel.com>,
Andi Kleen <ak@...ux.intel.com>
Subject: Re: [PATCH 2/2] perf stat: Show average value on multiple runs
On Sat, Jun 17, 2023 at 11:45 AM Jiri Olsa <olsajiri@...il.com> wrote:
>
> On Fri, Jun 16, 2023 at 12:32:11AM -0700, Namhyung Kim wrote:
> > When -r option is used, perf stat runs the command multiple times and
> > update stats in the evsel->stats.res_stats for global aggregation. But
> > the value is never used and the value it prints at the end is just the
> > value from the last run. I think we should print the average number of
> > multiple runs.
> >
> > Add evlist__copy_res_stats() to update the aggr counter (for display)
> > using the values in the evsel->stats.res_stats.
> >
> > Signed-off-by: Namhyung Kim <namhyung@...nel.org>
>
> this is the 'real' fix right? I thought this was the way it worked before
It worked like the patch 1/2 before.
>
> anyway works nicely now, would be nice to add some tests for this,
> but not sure how bad it'd be ;-)
Maybe we can parse the raw counter output from the -v option and
calculate the average then compare.
>
> Acked/Tested-by: Jiri Olsa <jolsa@...nel.org>
Thanks,
Namhyung
>
> > ---
> > tools/perf/builtin-stat.c | 5 ++++-
> > tools/perf/util/stat.c | 22 ++++++++++++++++++++++
> > tools/perf/util/stat.h | 1 +
> > 3 files changed, 27 insertions(+), 1 deletion(-)
> >
> > diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
> > index e549862f90f0..42f84975a4d5 100644
> > --- a/tools/perf/builtin-stat.c
> > +++ b/tools/perf/builtin-stat.c
> > @@ -2829,8 +2829,11 @@ int cmd_stat(int argc, const char **argv)
> > }
> > }
> >
> > - if (!forever && status != -1 && (!interval || stat_config.summary))
> > + if (!forever && status != -1 && (!interval || stat_config.summary)) {
> > + if (stat_config.run_count > 1)
> > + evlist__copy_res_stats(&stat_config, evsel_list);
> > print_counters(NULL, argc, argv);
> > + }
> >
> > evlist__finalize_ctlfd(evsel_list);
> >
> > diff --git a/tools/perf/util/stat.c b/tools/perf/util/stat.c
> > index 0f7b8a8cdea6..967e583392c7 100644
> > --- a/tools/perf/util/stat.c
> > +++ b/tools/perf/util/stat.c
> > @@ -264,6 +264,28 @@ void evlist__copy_prev_raw_counts(struct evlist *evlist)
> > evsel__copy_prev_raw_counts(evsel);
> > }
> >
> > +static void evsel__copy_res_stats(struct evsel *evsel)
> > +{
> > + struct perf_stat_evsel *ps = evsel->stats;
> > +
> > + /*
> > + * For GLOBAL aggregation mode, it updates the counts for each run
> > + * in the evsel->stats.res_stats. See perf_stat_process_counter().
> > + */
> > + *ps->aggr[0].counts.values = avg_stats(&ps->res_stats);
> > +}
> > +
> > +void evlist__copy_res_stats(struct perf_stat_config *config, struct evlist *evlist)
> > +{
> > + struct evsel *evsel;
> > +
> > + if (config->aggr_mode != AGGR_GLOBAL)
> > + return;
> > +
> > + evlist__for_each_entry(evlist, evsel)
> > + evsel__copy_res_stats(evsel);
> > +}
> > +
> > static size_t pkg_id_hash(long __key, void *ctx __maybe_unused)
> > {
> > uint64_t *key = (uint64_t *) __key;
> > diff --git a/tools/perf/util/stat.h b/tools/perf/util/stat.h
> > index 7abff7cbb5a1..1cbc26b587ba 100644
> > --- a/tools/perf/util/stat.h
> > +++ b/tools/perf/util/stat.h
> > @@ -182,6 +182,7 @@ void evlist__save_aggr_prev_raw_counts(struct evlist *evlist);
> >
> > int evlist__alloc_aggr_stats(struct evlist *evlist, int nr_aggr);
> > void evlist__reset_aggr_stats(struct evlist *evlist);
> > +void evlist__copy_res_stats(struct perf_stat_config *config, struct evlist *evlist);
> >
> > int perf_stat_process_counter(struct perf_stat_config *config,
> > struct evsel *counter);
> > --
> > 2.41.0.162.gfafddb0af9-goog
> >
Powered by blists - more mailing lists