Re: [PATCH v2 2/6] perf evsel: Introduce an observed performance device

[Alexander Antonov]

On 1/6/2021 11:44 AM, Namhyung Kim wrote:
> Hi,
>
> On Wed, Dec 23, 2020 at 10:03 PM Alexander Antonov
> <alexander.antonov@xxxxxxxxxxxxxxx> wrote:
> > Adding evsel::perf_device void pointer.
> >
> > For performance monitoring purposes, an evsel can have a related device.
> > These changes allow to attribute, for example, I/O performance metrics
> > to IIO stack.
> >
> > Signed-off-by: Alexander Antonov <alexander.antonov@xxxxxxxxxxxxxxx>
> > ---
> >   tools/perf/util/evsel.h | 1 +
> >   1 file changed, 1 insertion(+)
> >
> > diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h
> > index 79a860d8e3ee..c346920f477a 100644
> > --- a/tools/perf/util/evsel.h
> > +++ b/tools/perf/util/evsel.h
> > @@ -127,6 +127,7 @@ struct evsel {
> >           * See also evsel__has_callchain().
> >           */
> >          __u64                   synth_sample_type;
> > +       void                    *perf_device;
> Maybe we can use the existing 'priv' field.
>
> Thanks,
> Namhyung

Hello Namhyung,

Looks like the 'priv' field isn't used in this case. I suppose it can be
re-used in iiostat mode.

Thanks,
Alexander
> >   };
> >
> >   struct perf_missing_features {
> > --
> > 2.19.1