On Fri, Feb 17, 2023 at 03:10:18PM -0800, Sean Christopherson wrote:
> Track "virtual VMSAVE/VMLOAD exposed to L1" via a governed feature flag
> instead of using a dedicated bit/flag in vcpu_svm.
>
> No functional change intended.
>
> Signed-off-by: Sean Christopherson <seanjc@xxxxxxxxxx>
> ---
> arch/x86/kvm/governed_features.h | 1 +
> arch/x86/kvm/svm/nested.c | 2 +-
> arch/x86/kvm/svm/svm.c | 5 ++---
> arch/x86/kvm/svm/svm.h | 1 -
> 4 files changed, 4 insertions(+), 5 deletions(-)
>
> diff --git a/arch/x86/kvm/governed_features.h b/arch/x86/kvm/governed_features.h
> index 0335576a80a8..b66b9d550f33 100644
> --- a/arch/x86/kvm/governed_features.h
> +++ b/arch/x86/kvm/governed_features.h
> @@ -9,6 +9,7 @@ KVM_GOVERNED_X86_FEATURE(GBPAGES)
> KVM_GOVERNED_X86_FEATURE(XSAVES)
> KVM_GOVERNED_X86_FEATURE(NRIPS)
> KVM_GOVERNED_X86_FEATURE(TSCRATEMSR)
> +KVM_GOVERNED_X86_FEATURE(V_VMSAVE_VMLOAD)
>
> #undef KVM_GOVERNED_X86_FEATURE
> #undef KVM_GOVERNED_FEATURE
> diff --git a/arch/x86/kvm/svm/nested.c b/arch/x86/kvm/svm/nested.c
> index 30e00c4e07c7..6a96058c0e48 100644
> --- a/arch/x86/kvm/svm/nested.c
> +++ b/arch/x86/kvm/svm/nested.c
> @@ -107,7 +107,7 @@ static void nested_svm_uninit_mmu_context(struct kvm_vcpu *vcpu)
>
> static bool nested_vmcb_needs_vls_intercept(struct vcpu_svm *svm)
> {
> - if (!svm->v_vmload_vmsave_enabled)
> + if (!guest_can_use(&svm->vcpu, X86_FEATURE_V_VMSAVE_VMLOAD))
> return true;
>
> if (!nested_npt_enabled(svm))
> diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
> index dd4aead5462c..b3f0271c73b9 100644
> --- a/arch/x86/kvm/svm/svm.c
> +++ b/arch/x86/kvm/svm/svm.c
> @@ -1162,8 +1162,6 @@ static inline void init_vmcb_after_set_cpuid(struct kvm_vcpu *vcpu)
>
> set_msr_interception(vcpu, svm->msrpm, MSR_IA32_SYSENTER_EIP, 0, 0);
> set_msr_interception(vcpu, svm->msrpm, MSR_IA32_SYSENTER_ESP, 0, 0);
> -
> - svm->v_vmload_vmsave_enabled = false;
> } else {
> /*
> * If hardware supports Virtual VMLOAD VMSAVE then enable it
> @@ -4153,7 +4151,8 @@ static void svm_vcpu_after_set_cpuid(struct kvm_vcpu *vcpu)
>
> svm->lbrv_enabled = lbrv && guest_cpuid_has(vcpu, X86_FEATURE_LBRV);
>
> - svm->v_vmload_vmsave_enabled = vls && guest_cpuid_has(vcpu, X86_FEATURE_V_VMSAVE_VMLOAD);
> + if (vls && !guest_cpuid_is_intel(vcpu))
> + kvm_governed_feature_check_and_set(vcpu, X86_FEATURE_V_VMSAVE_VMLOAD);
Sorry, why guest_cpuid_is_intel(vcpu)? Is it becasue that a AMD host with virtual
VMSAVE/VMLOAD capability will always expose this feature for all AMD guests?
If yes, how about adding some comments? Thanks!
B.R.
Yu
