Re: [PATCH v8 0/7] KVM: arm64: PMU: Use multiple host PMUs

On Tue, 07 Jul 2026 09:10:17 +0100,
Oliver Upton <[email protected]> wrote:
> 
> On Tue, Jul 07, 2026 at 08:58:50AM +0100, Marc Zyngier wrote:
> > On Mon, 06 Jul 2026 19:28:20 +0100,
> > Oliver Upton <[email protected]> wrote:
> > > 
> > > On Mon, Jul 06, 2026 at 07:03:23PM +0900, Akihiko Odaki wrote:
> > > > Akihiko Odaki (7):
> > > >       KVM: arm64: Disallow vPMU when pPMUs do not cover all CPUs
> > > >       KVM: arm64: PMU: Protect the list of PMUs with RCU
> > > >       KVM: arm64: PMU: Pass the pPMU to kvm_map_pmu_event()
> > > >       KVM: arm64: PMU: Pass the target CPU to kvm_pmu_probe_armpmu()
> > > >       KVM: arm64: PMU: Implement fixed-counters-only emulation
> > > >       KVM: arm64: PMU: Introduce FIXED_COUNTERS_ONLY
> > > >       KVM: arm64: selftests: Test PMU_V3_FIXED_COUNTERS_ONLY
> > > 
> > > Thanks for respinning the series, I left some comments. I'd really like
> > > to get this feature picked up, hope you have cycles to work on it soon.
> > > 
> > > FWIW, we're considering adding a new vCPU feature flag to deprecate all
> > > the ugliness of the old PMUv3 UAPI. I'm tempted to say that the
> > > FIXED_COUNTERS_ONLY feature is conditioned on the new feature flag...
> > 
> > It'd be good to have a strawman proposal on the list if we are going
> > to change add a new API, specially if this is likely to gate new code.
> > 
> 
> Patch 1 [*] of the series I linked does exactly this, no? Needs some

Ah, I overlooked that, apologies.

> attention like a KVM_CAP to make the bit discoverable by the VMM but
> gets the general idea across.

Not sure that's the ideal approach, specially in the light of this
"multiple PMU" series. This "strict" API mandates the selection of a
PMU (singular), while this stuff is the exact opposite (grab all the
possible PMUs).

	M.

-- 
Without deviation from the norm, progress is not possible.