KVM: nSVM: Check for NRIPS support before updating control field
authorBandan Das <bsd@redhat.com>
Thu, 11 Jun 2015 06:05:33 +0000 (02:05 -0400)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Fri, 10 Jul 2015 16:49:27 +0000 (09:49 -0700)
commit f104765b4f81fd74d69e0eb161e89096deade2db upstream.

If hardware doesn't support DecodeAssist - a feature that provides
more information about the intercept in the VMCB, KVM decodes the
instruction and then updates the next_rip vmcb control field.
However, NRIP support itself depends on cpuid Fn8000_000A_EDX[NRIPS].
Since skip_emulated_instruction() doesn't verify nrip support
before accepting control.next_rip as valid, avoid writing this
field if support isn't present.

Signed-off-by: Bandan Das <bsd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
arch/x86/kvm/svm.c

index 9afa233..4911bf1 100644 (file)
@@ -511,8 +511,10 @@ static void skip_emulated_instruction(struct kvm_vcpu *vcpu)
 {
        struct vcpu_svm *svm = to_svm(vcpu);
 
-       if (svm->vmcb->control.next_rip != 0)
+       if (svm->vmcb->control.next_rip != 0) {
+               WARN_ON(!static_cpu_has(X86_FEATURE_NRIPS));
                svm->next_rip = svm->vmcb->control.next_rip;
+       }
 
        if (!svm->next_rip) {
                if (emulate_instruction(vcpu, EMULTYPE_SKIP) !=
@@ -4317,7 +4319,9 @@ static int svm_check_intercept(struct kvm_vcpu *vcpu,
                break;
        }
 
-       vmcb->control.next_rip  = info->next_rip;
+       /* TODO: Advertise NRIPS to guest hypervisor unconditionally */
+       if (static_cpu_has(X86_FEATURE_NRIPS))
+               vmcb->control.next_rip  = info->next_rip;
        vmcb->control.exit_code = icpt_info.exit_code;
        vmexit = nested_svm_exit_handled(svm);