Re: [PATCH 3/3] KVM: VMX: Adjust number of LBR records for PERF_CAPABILITIES at refresh

[Like Xu <like.xu.linux@gmail.com>]

On 28/7/2022 7:34 am, Sean Christopherson wrote:
> -bool intel_pmu_lbr_is_enabled(struct kvm_vcpu *vcpu)
> -{
> -	struct x86_pmu_lbr *lbr = vcpu_to_lbr_records(vcpu);
> -
> -	return lbr->nr && (vcpu_get_perf_capabilities(vcpu) & PMU_CAP_LBR_FMT);
> -}
> -
>   static bool intel_pmu_is_valid_lbr_msr(struct kvm_vcpu *vcpu, u32 index)
>   {
>   	struct x86_pmu_lbr *records = vcpu_to_lbr_records(vcpu);
> @@ -590,7 +583,9 @@ static void intel_pmu_refresh(struct kvm_vcpu *vcpu)
>   	bitmap_set(pmu->all_valid_pmc_idx,
>   		INTEL_PMC_MAX_GENERIC, pmu->nr_arch_fixed_counters);
>   
> -	if (cpuid_model_is_consistent(vcpu))
> +	perf_capabilities = vcpu_get_perf_capabilities(vcpu);
> +	if (cpuid_model_is_consistent(vcpu) &&
> +	    (perf_capabilities & PMU_CAP_LBR_FMT))
>   		x86_perf_get_lbr(&lbr_desc->records);

As one of evil source to add CPUID walk in the critical path:

The x86_perf_get_lbr() is one of the perf interfaces, KVM cannot always trust
that the number of returned lbr_desc->records.nr is always > 0,  and if not,
we have to tweak perf_capabilities inside KVM which violates user input again.

Do you have more inputs to address this issue ?

>   	else
>   		lbr_desc->records.nr = 0;