Re: [PATCH 3/4] KVM: x86: Remove kvm_mmu_reset_context() in kvm_set_efer()

[Marcelo Tosatti <mtosatti@redhat.com>]

On Tue, May 11, 2010 at 01:30:06PM +0800, Sheng Yang wrote:
> Modify EFER won't result in mode switch directly. After EFER.LME set, the
> following set CR0.PG would result in mode switch to IA32e. And the later
> action already covered by kvm_set_cr0().
> 
> Signed-off-by: Sheng Yang <sheng@linux.intel.com>
> ---
>  arch/x86/kvm/x86.c |    1 -
>  1 files changed, 0 insertions(+), 1 deletions(-)
> 
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 764f89b..b59fc67 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -721,7 +721,6 @@ static int set_efer(struct kvm_vcpu *vcpu, u64 efer)
>  	kvm_x86_ops->set_efer(vcpu, efer);
>  
>  	vcpu->arch.mmu.base_role.nxe = (efer & EFER_NX) && !tdp_enabled;
> -	kvm_mmu_reset_context(vcpu);

But there are different sets of shadow pagetables for NXE on/off. See
commit 9645bb56b31a1b.

Without the reset, after NXE 1->0 transition, a spte retains the NXE
validity check, and subsequent use of such gpte with bit 63 set does not
cause a fault.