On Tue, Aug 28, 2018 at 9:04 AM, Sean Christopherson
<sean.j.christopherson@xxxxxxxxx> wrote:
> VM_ENTRY_IA32E_MODE and VM_{ENTRY,EXIT}_LOAD_IA32_EFER will be
> explicitly set/cleared as needed by vmx_set_efer(), but attempt
> to get the bits set correctly when intializing the control fields.
> Setting the value correctly can avoid multiple VMWrites.
>
> Signed-off-by: Sean Christopherson <sean.j.christopherson@xxxxxxxxx>
> ---
> arch/x86/kvm/vmx.c | 44 ++++++++++++++++++++++++++++++--------------
> 1 file changed, 30 insertions(+), 14 deletions(-)
>
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index 1fcf374a1475..e58dd3a66abf 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -11896,6 +11896,17 @@ static int nested_vmx_load_cr3(struct kvm_vcpu *vcpu, unsigned long cr3, bool ne
> return 0;
> }
>
> +static u64 nested_vmx_calc_efer(struct vcpu_vmx *vmx, struct vmcs12 *vmcs12)
> +{
> + if (vmx->nested.nested_run_pending &&
> + (vmcs12->vm_entry_controls & VM_ENTRY_LOAD_IA32_EFER))
> + return vmcs12->guest_ia32_efer;
> + else if (vmcs12->vm_entry_controls & VM_ENTRY_IA32E_MODE)
> + return vmx->vcpu.arch.efer | (EFER_LMA | EFER_LME);
> + else
> + return vmx->vcpu.arch.efer & ~(EFER_LMA | EFER_LME);
> +}
This makes me a little uncomfortable from the save/restore standpoint,
though it does work given kvm's current behavior. VM-entry controls
should really only be applied when vmx->nested.nested_run_pending is
true. Can this be changed to:
if (!vmx->nested.nested_run_pending) {
return vmx->vcpu.arch.efer;
} else {
if (vmcs12->vm_entry_controls & VM_ENTRY_LOAD_IA32_EFER)
return vmcs12->guest_ia32_efer;
else if (vmcs12->vm_entry_controls & VM_ENTRY_IA32E_MODE)
return vmx->vcpu.arch.efer | (EFER_LMA | EFER_LME);
else
return vmx->vcpu.arch.efer & ~(EFER_LMA | EFER_LME);
}
Reviewed-by: Jim Mattson <jmattson@xxxxxxxxxx>
