On Tue, Jul 01, 2025 at 06:13:57AM +0000, Koichi Okuno (Fujitsu) wrote:
> Hi, Peter
>
> Sorry for the late reply.
>
> > On Tue, Jun 17, 2025 at 07:27:50PM +0900, Koichi Okuno wrote:
> > > + pcipmu->pmu = (struct pmu) {
> > > + .parent = dev,
> > > + .task_ctx_nr = perf_invalid_context,
> > > +
> > > + .pmu_enable = fujitsu_pci__pmu_enable,
> > > + .pmu_disable = fujitsu_pci__pmu_disable,
> > > + .event_init = fujitsu_pci__event_init,
> > > + .add = fujitsu_pci__event_add,
> > > + .del = fujitsu_pci__event_del,
> > > + .start = fujitsu_pci__event_start,
> > > + .stop = fujitsu_pci__event_stop,
> > > + .read = fujitsu_pci__event_read,
> > > +
> > > > + .attr_groups = fujitsu_pci_pmu_attr_grps,
> > > + .capabilities = PERF_PMU_CAP_NO_EXCLUDE,
> >
> > Should these drivers not also have PERF_PMU_CAP_NO_INTERRUPT ? Per them
> > being uncore they cannot generate samples.
>
> Even now, the fujitsu_pci__event_init() function rejects sampling events
> before checking PERF_PMU_CAP_NO_INTERRUPT in core.c.
> However, I think it is correct to have PERF_PMU_CAP_NO_INTERRUPT which
> means not being able to sample, so I will add NO_INTERRUPT in the next
> version.
Sorry to give contradictory feeback, but given that the majority of PMU
drivers under drivers/perf/ don't set PERF_PMU_CAP_NO_INTERRUPT, and
explicitly reject sampling events, I'd prefer to do the same for
consistency.
Peter, are you happy with that for now?
Overall it would be better for PMUs to export a positive "I support
sampling" flag, and maybe rename/replace PERF_PMU_CAP_NO_INTERRUPT
accordingly, which I'd be happy to look at in future.
Mark.
