Re: [PATCH v2 1/6] KVM: MMU: fix forgot flush vcpu tlbs

[Marcelo Tosatti <mtosatti@redhat.com>]

On Wed, Nov 17, 2010 at 06:26:51PM +0200, Avi Kivity wrote:
> On 11/17/2010 05:29 PM, Marcelo Tosatti wrote:
> >>  diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h
> >>  index ba00eef..58b4d9a 100644
> >>  --- a/arch/x86/kvm/paging_tmpl.h
> >>  +++ b/arch/x86/kvm/paging_tmpl.h
> >>  @@ -781,6 +781,7 @@ static int FNAME(sync_page)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
> >>   			else
> >>   				nonpresent = shadow_notrap_nonpresent_pte;
> >>   			drop_spte(vcpu->kvm,&sp->spt[i], nonpresent);
> >>  +			kvm_flush_remote_tlbs(vcpu->kvm);
> >>   			continue;
> >>   		}
> >
> >This is not needed. Guest is responsible for flushing on
> >present->nonpresent change.
> 
> sync_page
> drop_spte
> kvm_mmu_notifier_invalidate_page
> kvm_unmap_rmapp
> spte doesn't exist -> no flush
> page is freed
> guest can write into freed page?

Ugh right.

> I don't think we need to flush immediately; set a "tlb dirty" bit
> somewhere that is cleareded when we flush the tlb.
> kvm_mmu_notifier_invalidate_page() can consult the bit and force a
> flush if set.

Yep.