Re: [PATCH v5 2/2] perf: Fujitsu: Add the Uncore PCI PMU driver

[Mark Rutland <mark.rutland@arm.com>]

On Thu, Jun 19, 2025 at 04:04:42PM +0200, Peter Zijlstra wrote:
> On Thu, Jun 19, 2025 at 02:57:55PM +0100, Mark Rutland wrote:
> > On Tue, Jun 17, 2025 at 12:36:18PM +0200, Peter Zijlstra wrote:
> > > 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.
> > 
> > These PMUs actually have an interrupt, so that might be a bit confusing.
> > That said, the counters seem to be 64-bit, so the interrupt doesn't seem
> > practically necessary today.
> 
> Yeah, I saw they had an interrupt. But them being uncore they must not
> sample, and NO_INTERRUPT means not being able to sample.
> 
> Naming urgh :-)

Yeah :)

FWIW, that came to mind due to:

  https://lore.kernel.org/lkml/20250618-perf-pmu-cap-docs-v1-1-0d34387d6e47@collabora.com/

... and if NO_INTERRUPT would be better-named as NO_SAMPLING, that might
be a good opportunity to clean that up.

> > Either way, the fujitsu_pci__event_init() function rejects sampling
> > events, so it correctly rejects sampling events.
> > 
> > IMO, it'd be a bit nicer to do the inverse, and have the perf core
> > reject sampling events by default for non-CPU PMUs.
> 
> Yeah, there's lots of cleanup to be done. Quite possibly also some
> unification between the various drivers.
> 
> Just need someone that knows what they're doing that has spare time,
> know anybody like that? :-)

I think a few folk likle that live in the nearest hall of mirrors...

If you're happy with that in concept, I can see about getting someone to
look at that -- the general idea has come up a few times.

Mark.