Em Wed, Feb 03, 2021 at 04:20:38PM -0500, Liang, Kan escreveu:
>
>
> On 2/3/2021 3:02 PM, Arnaldo Carvalho de Melo wrote:
> > Em Tue, Feb 02, 2021 at 12:09:06PM -0800,kan.li...@linux.intel.com
> > escreveu:
> > > From: Kan Liang<kan.li...@linux.intel.com>
> > > diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
> > > index c26ea822..c48f6de 100644
> > > --- a/tools/perf/util/evsel.c
> > > +++ b/tools/perf/util/evsel.c
> > > @@ -2689,6 +2689,9 @@ int evsel__open_strerror(struct evsel *evsel,
> > > struct target *target,
> > > if (perf_missing_features.aux_output)
> > > return scnprintf(msg, size, "The 'aux_output'
> > > feature is not supported, update the kernel.");
> > > break;
> > > + case ENODATA:
> > > + return scnprintf(msg, size, "Cannot collect data source with
> > > the load latency event alone. "
> > > + "Please add an auxiliary event in front of the
> > > load latency event.");
> > Are you sure this is the only case where ENODATA comes out from
> > perf_event_open()? Well, according to your comment in:
> >
> > 61b985e3e775a3a7 ("perf/x86/intel: Add perf core PMU support for
> > Sapphire Rapids")
> >
> > It should be at that point in time, so its safe to merge as-is, but then
> > I think this is fragile, what if someone else, in the future, not
> > knowing that ENODATA is supposed to be used only with that ancient CPU,
> > Sapphire Rapids, uses it?:-)
> >
> > Please consider adding a check before assuming ENODATA is for this
> > specific case.
>
> Sure, I will add a check in V2.
Do it as a separate patch, on top of what is now in tmp.perf/core.
https://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git/log/?h=tmp.perf/core
- Arnaldo
