On 8/2/24 13:19, Sean Christopherson wrote:
> Inject a #GP if the guest attempts to change MSR_AMD64_DE_CFG from its
> *current* value, not if the guest attempts to write a value other than
> KVM's set of supported bits. As per the comment and the changelog of the
> original code, the intent is to effectively make MSR_AMD64_DE_CFG read-
> only for the guest.
>
> Opportunistically use a more conventional equality check instead of an
> exclusive-OR check to detect attempts to change bits.
>
> Fixes: d1d93fa90f1a ("KVM: SVM: Add MSR-based feature support for serializing LFENCE")
> Cc: Tom Lendacky <thomas.lendacky@xxxxxxx>
> Signed-off-by: Sean Christopherson <seanjc@xxxxxxxxxx>
Reviewed-by: Tom Lendacky <thomas.lendacky@xxxxxxx>
(one spelling correction below)
> ---
> arch/x86/kvm/svm/svm.c | 9 +++++++--
> 1 file changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
> index c115d26844f7..550ead197543 100644
> --- a/arch/x86/kvm/svm/svm.c
> +++ b/arch/x86/kvm/svm/svm.c
> @@ -3189,8 +3189,13 @@ static int svm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr)
> if (data & ~msr_entry.data)
> return 1;
>
> - /* Don't allow the guest to change a bit, #GP */
> - if (!msr->host_initiated && (data ^ msr_entry.data))
> + /*
> + * Don't let the guest change the host-programmed value. The
> + * MSR is very model specific, i.e. contains multiple bits that
> + * are completely unknown to KVM, and the one bit known to KVM
> + * is simply a reflection of hardware capatibilies.
s/capatibilies/capabilities/
> + */
> + if (!msr->host_initiated && data != svm->msr_decfg)
> return 1;
>
> svm->msr_decfg = data;
