[RFC PATCH] perf stat: Separate bperf from bpf_profiler

[RFC PATCH] perf stat: Separate bperf from bpf_profiler

[Dmitrii Dolgov]

It seems that perf stat -b <prog id> doesn't produce any results:

    $ perf stat -e cycles -b 4 -I 10000 -vvv
    Control descriptor is not initialized
    cycles: 0 0 0
                time        counts unit      events
	10.007641640    <not supported>      cycles

Looks like this happens because fentry/fexit progs are getting loaded, but the
corresponding perf event is not enabled and not added into the events bpf map.
I think there is some mixing up between two type of bpf support, one for bperf
and one for bpf_profiler. Both are identified via evsel__is_bpf, based on which
perf events are enabled, but for the latter (bpf_profiler) a perf event is
required. Using evsel__is_bperf to check only bperf produces expected results:

    $ perf stat -e cycles -b 4 -I 10000 -vvv
    Control descriptor is not initialized
    ------------------------------------------------------------
    perf_event_attr:
      size                             136
      sample_type                      IDENTIFIER
      read_format                      TOTAL_TIME_ENABLED|TOTAL_TIME_RUNNING
      disabled                         1
      exclude_guest                    1
    ------------------------------------------------------------
    sys_perf_event_open: pid -1  cpu 0  group_fd -1  flags 0x8 = 3
    ------------------------------------------------------------
    [...perf_event_attr for other CPUs...]
    ------------------------------------------------------------
    cycles: 309426 169009 169009
		time             counts unit events
	10.010091271             309426      cycles

The final numbers correspond (at least in the level of magnitude) to the
same metric obtained via bpftool.

Signed-off-by: Dmitrii Dolgov <9erthalion6@gmail.com>
---
 tools/perf/builtin-stat.c | 4 ++--
 tools/perf/util/evsel.h   | 5 +++++
 2 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index d3cbee7460f..23b8d684ca6 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -712,7 +712,7 @@ static int __run_perf_stat(int argc, const char **argv, int run_idx)
 		counter->reset_group = false;
 		if (bpf_counter__load(counter, &target))
 			return -1;
-		if (!evsel__is_bpf(counter))
+		if (!(evsel__is_bperf(counter)))
 			all_counters_use_bpf = false;
 	}
 
@@ -728,7 +728,7 @@ static int __run_perf_stat(int argc, const char **argv, int run_idx)
 
 		if (counter->reset_group || counter->errored)
 			continue;
-		if (evsel__is_bpf(counter))
+		if (evsel__is_bperf(counter))
 			continue;
 try_again:
 		if (create_perf_stat_counter(counter, &stat_config, &target,
diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h
index 68072ec655c..f0aa17dcf3d 100644
--- a/tools/perf/util/evsel.h
+++ b/tools/perf/util/evsel.h
@@ -269,6 +269,11 @@ static inline bool evsel__is_bpf(struct evsel *evsel)
 	return evsel->bpf_counter_ops != NULL;
 }
 
+static inline bool evsel__is_bperf(struct evsel *evsel)
+{
+	return evsel->bpf_counter_ops != NULL && list_empty(&evsel->bpf_counter_list);
+}
+
 #define EVSEL__MAX_ALIASES 8
 
 extern const char *const evsel__hw_cache[PERF_COUNT_HW_CACHE_MAX][EVSEL__MAX_ALIASES];
-- 
2.32.0

Re: [RFC PATCH] perf stat: Separate bperf from bpf_profiler

[Dmitry Dolgov]

> On Wed, Apr 12, 2023 at 08:23:16PM +0200, Dmitrii Dolgov wrote:
> It seems that perf stat -b <prog id> doesn't produce any results:
>
>     $ perf stat -e cycles -b 4 -I 10000 -vvv
>     Control descriptor is not initialized
>     cycles: 0 0 0
>                 time        counts unit      events
> 	10.007641640    <not supported>      cycles
>
> Looks like this happens because fentry/fexit progs are getting loaded, but the
> corresponding perf event is not enabled and not added into the events bpf map.
> I think there is some mixing up between two type of bpf support, one for bperf
> and one for bpf_profiler. Both are identified via evsel__is_bpf, based on which
> perf events are enabled, but for the latter (bpf_profiler) a perf event is
> required. Using evsel__is_bperf to check only bperf produces expected results:

Any thoughts on this? I would appreciate clarifications if I'm missing
something.

Re: [RFC PATCH] perf stat: Separate bperf from bpf_profiler

[Arnaldo Carvalho de Melo]

Em Fri, Apr 21, 2023 at 10:56:10PM +0200, Dmitry Dolgov escreveu:
> > On Wed, Apr 12, 2023 at 08:23:16PM +0200, Dmitrii Dolgov wrote:
> > It seems that perf stat -b <prog id> doesn't produce any results:
> >
> >     $ perf stat -e cycles -b 4 -I 10000 -vvv
> >     Control descriptor is not initialized
> >     cycles: 0 0 0
> >                 time        counts unit      events
> > 	10.007641640    <not supported>      cycles
> >
> > Looks like this happens because fentry/fexit progs are getting loaded, but the
> > corresponding perf event is not enabled and not added into the events bpf map.
> > I think there is some mixing up between two type of bpf support, one for bperf
> > and one for bpf_profiler. Both are identified via evsel__is_bpf, based on which
> > perf events are enabled, but for the latter (bpf_profiler) a perf event is
> > required. Using evsel__is_bperf to check only bperf produces expected results:
> 
> Any thoughts on this? I would appreciate clarifications if I'm missing
> something.

Namhyung, Song, can you please take a look at this?

- Arnaldo

Re: [RFC PATCH] perf stat: Separate bperf from bpf_profiler

[Song Liu]

On Fri, Apr 28, 2023 at 6:44 PM Arnaldo Carvalho de Melo
<acme@kernel.org> wrote:
>
> Em Fri, Apr 21, 2023 at 10:56:10PM +0200, Dmitry Dolgov escreveu:
> > > On Wed, Apr 12, 2023 at 08:23:16PM +0200, Dmitrii Dolgov wrote:
> > > It seems that perf stat -b <prog id> doesn't produce any results:
> > >
> > >     $ perf stat -e cycles -b 4 -I 10000 -vvv
> > >     Control descriptor is not initialized
> > >     cycles: 0 0 0
> > >                 time        counts unit      events
> > >     10.007641640    <not supported>      cycles
> > >
> > > Looks like this happens because fentry/fexit progs are getting loaded, but the
> > > corresponding perf event is not enabled and not added into the events bpf map.
> > > I think there is some mixing up between two type of bpf support, one for bperf
> > > and one for bpf_profiler. Both are identified via evsel__is_bpf, based on which
> > > perf events are enabled, but for the latter (bpf_profiler) a perf event is
> > > required. Using evsel__is_bperf to check only bperf produces expected results:
> >
> > Any thoughts on this? I would appreciate clarifications if I'm missing
> > something.
>
> Namhyung, Song, can you please take a look at this?

Sorry for the late response. The fix looks good to me and worked well
in my test.

Reviewed-and-tested-by: Song Liu <song@kernel.org>

I guess we also need:

Fixes: 112cb56164bc2 ("perf stat: Introduce config stat.bpf-counter-events")

Thanks for the fix!
Song

Re: [RFC PATCH] perf stat: Separate bperf from bpf_profiler

[Arnaldo Carvalho de Melo]

Em Thu, May 04, 2023 at 07:01:04PM -0700, Song Liu escreveu:
> On Fri, Apr 28, 2023 at 6:44 PM Arnaldo Carvalho de Melo <acme@kernel.org> wrote:

> > Em Fri, Apr 21, 2023 at 10:56:10PM +0200, Dmitry Dolgov escreveu:
> > > > On Wed, Apr 12, 2023 at 08:23:16PM +0200, Dmitrii Dolgov wrote:
> > > > It seems that perf stat -b <prog id> doesn't produce any results:

> > > >     $ perf stat -e cycles -b 4 -I 10000 -vvv
> > > >     Control descriptor is not initialized
> > > >     cycles: 0 0 0
> > > >                 time        counts unit      events
> > > >     10.007641640    <not supported>      cycles

> > > > Looks like this happens because fentry/fexit progs are getting loaded, but the
> > > > corresponding perf event is not enabled and not added into the events bpf map.
> > > > I think there is some mixing up between two type of bpf support, one for bperf
> > > > and one for bpf_profiler. Both are identified via evsel__is_bpf, based on which
> > > > perf events are enabled, but for the latter (bpf_profiler) a perf event is
> > > > required. Using evsel__is_bperf to check only bperf produces expected results:

> > > Any thoughts on this? I would appreciate clarifications if I'm missing
> > > something.

> > Namhyung, Song, can you please take a look at this?

> Sorry for the late response. The fix looks good to me and worked well
> in my test.

> Reviewed-and-tested-by: Song Liu <song@kernel.org>

> I guess we also need:

> Fixes: 112cb56164bc2 ("perf stat: Introduce config stat.bpf-counter-events")

Thanks a lot, applied, and this is relevant in the current situation,
where we're trying to have Linux v6.4 perf tools building BPF skels by
default.

- Arnaldo