Re: [PATCH v2 2/2] KVM: nVMX: Check host support for MSR bitmaps

["Radim Krčmář" <rkrcmar@redhat.com>]

2017-05-25 17:04-0700, Jim Mattson:
> If the host doesn't support MSR bitmaps, but the L2 guest is
> configured to use MSR bitmaps, fail the emulated VM-entry.
> 
> Fixes: 3af18d9c5fe95 ("KVM: nVMX: Prepare for using hardware MSR bitmap")
> Signed-off-by: Jim Mattson <jmattson@google.com>
> ---
>  arch/x86/kvm/vmx.c | 4 +---
>  1 file changed, 1 insertion(+), 3 deletions(-)
> 
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index 0d72b4064a80..639188037744 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -9587,12 +9587,10 @@ static void vmx_start_preemption_timer(struct kvm_vcpu *vcpu)
>  static int nested_vmx_check_msr_bitmap_controls(struct kvm_vcpu *vcpu,
>  						struct vmcs12 *vmcs12)
>  {
> -	u64 addr;
> -
>  	if (!nested_cpu_has(vmcs12, CPU_BASED_USE_MSR_BITMAPS))
>  		return 0;
>  
> -	if (vmcs12_read_any(vcpu, MSR_BITMAP, &addr)) {
> +	if (!cpu_has_vmx_msr_bitmap()) {

VMCS bitmaps are emulated and exposed regardless of CPU features.
Doesn't this change break L1 hypervisor on CPUs that don't support MSR
bitmaps?

I think the check can just go away and can be merged into the first
patch.

Thanks.

>  		WARN_ON(1);
>  		return -EINVAL;
>  	}
> -- 
> 2.13.0.219.gdb65acc882-goog
>