Query the flexpriority module param instead of simply checking if
flexpriority is supported by the cpu when checking whether or not VMCS
fields need to be updated in response to the guest changing its
APIC_BASE MSR control bits. This avoids multiple unnecessary VMREADs
and a VMWRITE if flexpriority is disabled via its module param and the
cpu doesn't support X2APIC virtualization.
Signed-off-by: Sean Christopherson <sean.j.christopherson@xxxxxxxxx>
---
This is basically fixup for commit 76e97cc35522 ("KVM: VMX: check for
existence of secondary exec controls before accessing") in
kvm.git/queue. The aforementioned commit came from a series with a
prior patch that removed the module param, but the prior patch was
ultimately rejected resulting in this partially wrong code.
arch/x86/kvm/vmx.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index c7ae8ea87bc4..612fd17be635 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -10223,7 +10223,7 @@ static void vmx_set_virtual_apic_mode(struct kvm_vcpu *vcpu)
if (!lapic_in_kernel(vcpu))
return;
- if (!cpu_has_vmx_flexpriority() &&
+ if (!flexpriority_enabled &&
!cpu_has_vmx_virtualize_x2apic_mode())
return;
--
2.18.0
