Re: [PATCH v3 5/7] KVM: x86/pmu: Defer counter emulated overflow via pmc->prev_counter

[Like Xu]

On 23/9/2022 7:59 am, Sean Christopherson wrote:
> The bigger issue is that calling __kvm_perf_overflow() here can get false negatives.
> If reprogramming fails due to contention, the reprogram_pmi bit will be left set
> and so this check in __kvm_perf_overflow() will suppress the PMI.

I understand your motivation, and before we reprocess on "left set reprogram_pmi 
bit",
we may need reproducible cases to cover "reprogramming fails".