On 8/1/2025 7:39 AM, Sean Christopherson wrote:
On Wed, Jul 30, 2025, Xin Li (Intel) wrote:
Advertise support for the immediate form of MSR instructions to userspace
if the instructions are supported by the underlying CPU.
SVM needs to explicitly clear the capability so that KVM doesn't over-advertise
support if AMD ever implements X86_FEATURE_MSR_IMM.
diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
index ca550c4fa174..7e7821ee8ee1 100644
--- a/arch/x86/kvm/svm/svm.c
+++ b/arch/x86/kvm/svm/svm.c
@@ -5311,8 +5311,12 @@ static __init void svm_set_cpu_caps(void)
/* CPUID 0x8000001F (SME/SEV features) */
sev_set_cpu_caps();
- /* Don't advertise Bus Lock Detect to guest if SVM support is absent */
+ /*
+ * Clear capabilities that are automatically configured by common code,
+ * but that require explicit SVM support (that isn't yet implemented).
+ */
kvm_cpu_cap_clear(X86_FEATURE_BUS_LOCK_DETECT);
+ kvm_cpu_cap_clear(X86_FEATURE_MSR_IMM);
}
static __init int svm_hardware_setup(void)
Nice catch!
Yes, a feature needing explicit enabling effort can't be blindly
advertised until the support on all sub-arch is ready. I.e., I need to
disable it on non-Intel CPUs because it's only done for Intel.
