Re: [PATCH] perf stat: Support --all-kernel/--all-user

[Jin, Yao]

On 10/15/2019 11:11 PM, Arnaldo Carvalho de Melo wrote:
> Em Tue, Oct 15, 2019 at 09:57:33AM +0800, Jin, Yao escreveu:
> > On 10/15/2019 12:23 AM, Arnaldo Carvalho de Melo wrote:
> > > Em Mon, Oct 14, 2019 at 04:42:08PM +0200, Jiri Olsa escreveu:
> > > > On Fri, Oct 11, 2019 at 01:05:45PM +0800, Jin Yao wrote:
> > > > > perf record has supported --all-kernel / --all-user to configure all
> > > > > used events to run in kernel space or run in user space. But perf
> > > > > stat doesn't support these options.
> > > > >
> > > > > It would be useful to support these options in perf-stat too to keep
> > > > > the same semantics.
> > > > >
> > > > > Signed-off-by: Jin Yao <yao.jin@xxxxxxxxxxxxxxx>
> > > >
> > > > Acked-by: Jiri Olsa <jolsa@xxxxxxxxxx>
> > >
> > > What happens if one asks for both? From the patch only the --all-kernel
> > > will stick, right? Is that the behaviour for 'perf record'? Lemme see,
> > > as I recall having this discussion at that time...
> > >
> > > Yeah, same thing, and looking at that I wonder why we dong use
> > > perf_evsel__config() in 'perf stat'...
> > >
> > > Ok, I'm going to apply it.
> >
> > Hi Arnaldo,
> >
> > If we specify both, we will see the error.
>
> Cool, I looked for this code and couldn't quickly spot it nor tried
> running with both to see what happened, my bad, thanks for replying and
> making sure that that is the behaviour.
>   
> > root@kbl:~# perf stat --all-kernel --all-user -a -- sleep 1
> >   Error: option `all-user' cannot be used with all-kernel
> >
> > root@kbl:~# perf record --all-kernel --all-user -a -- sleep 1
> >   Error: option `all-user' cannot be used with all-kernel
> >
> > For why I don't set them in perf_evsel__config, because perf_evsel__config
> > is called in record but not called in stat.
>
> Ok, at some point we should try and check if we could make 'perf stat'
> use perf_evsel__config(), even if we have to separate what is shared
> into some __perf_evsel__config() that then gets called by 'perf stat'
> while the preexisting perf_evsel__config() calls it and does things
> that don't make sense for 'perf stat'.
>   

Yes, fully agree, that should be better. If we reconstruct the 
perf_evsel__config() and make it be shared by perf stat, perf record and 
others, that would be nice. :)

Thanks
Jin Yao

> > Thanks for reviewing and applying this patch.
>
> You're welcome!
>
> - Arnaldo
>   
> > Thanks
> > Jin Yao
> >
> > > > thanks,
> > > > jirka
> > > >
> > > > > ---
> > > > >    tools/perf/Documentation/perf-stat.txt |  6 ++++++
> > > > >    tools/perf/builtin-stat.c              |  6 ++++++
> > > > >    tools/perf/util/stat.c                 | 10 ++++++++++
> > > > >    tools/perf/util/stat.h                 |  2 ++
> > > > >    4 files changed, 24 insertions(+)
> > > > >
> > > > > diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt
> > > > > index 930c51c01201..a9af4e440e80 100644
> > > > > --- a/tools/perf/Documentation/perf-stat.txt
> > > > > +++ b/tools/perf/Documentation/perf-stat.txt
> > > > > @@ -323,6 +323,12 @@ The output is SMI cycles%, equals to (aperf - unhalted core cycles) / aperf
> > > > >    Users who wants to get the actual value can apply --no-metric-only.
> > > > > +--all-kernel::
> > > > > +Configure all used events to run in kernel space.
> > > > > +
> > > > > +--all-user::
> > > > > +Configure all used events to run in user space.
> > > > > +
> > > > >    EXAMPLES
> > > > >    --------
> > > > > diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
> > > > > index 468fc49420ce..c88d4e118409 100644
> > > > > --- a/tools/perf/builtin-stat.c
> > > > > +++ b/tools/perf/builtin-stat.c
> > > > > @@ -803,6 +803,12 @@ static struct option stat_options[] = {
> > > > >    	OPT_CALLBACK('M', "metrics", &evsel_list, "metric/metric group list",
> > > > >    		     "monitor specified metrics or metric groups (separated by ,)",
> > > > >    		     parse_metric_groups),
> > > > > +	OPT_BOOLEAN_FLAG(0, "all-kernel", &stat_config.all_kernel,
> > > > > +			 "Configure all used events to run in kernel space.",
> > > > > +			 PARSE_OPT_EXCLUSIVE),
> > > > > +	OPT_BOOLEAN_FLAG(0, "all-user", &stat_config.all_user,
> > > > > +			 "Configure all used events to run in user space.",
> > > > > +			 PARSE_OPT_EXCLUSIVE),
> > > > >    	OPT_END()
> > > > >    };
> > > > > diff --git a/tools/perf/util/stat.c b/tools/perf/util/stat.c
> > > > > index ebdd130557fb..6822e4ffe224 100644
> > > > > --- a/tools/perf/util/stat.c
> > > > > +++ b/tools/perf/util/stat.c
> > > > > @@ -490,6 +490,16 @@ int create_perf_stat_counter(struct evsel *evsel,
> > > > >    	if (config->identifier)
> > > > >    		attr->sample_type = PERF_SAMPLE_IDENTIFIER;
> > > > > +	if (config->all_user) {
> > > > > +		attr->exclude_kernel = 1;
> > > > > +		attr->exclude_user   = 0;
> > > > > +	}
> > > > > +
> > > > > +	if (config->all_kernel) {
> > > > > +		attr->exclude_kernel = 0;
> > > > > +		attr->exclude_user   = 1;
> > > > > +	}
> > > > > +
> > > > >    	/*
> > > > >    	 * Disabling all counters initially, they will be enabled
> > > > >    	 * either manually by us or by kernel via enable_on_exec
> > > > > diff --git a/tools/perf/util/stat.h b/tools/perf/util/stat.h
> > > > > index edbeb2f63e8d..081c4a5113c6 100644
> > > > > --- a/tools/perf/util/stat.h
> > > > > +++ b/tools/perf/util/stat.h
> > > > > @@ -106,6 +106,8 @@ struct perf_stat_config {
> > > > >    	bool			 big_num;
> > > > >    	bool			 no_merge;
> > > > >    	bool			 walltime_run_table;
> > > > > +	bool			 all_kernel;
> > > > > +	bool			 all_user;
> > > > >    	FILE			*output;
> > > > >    	unsigned int		 interval;
> > > > >    	unsigned int		 timeout;
> > > > > --
> > > > > 2.17.1