Re: [PATCH] KVM: x86: Avoid zapping mmio sptes twice for generation wraparound

[Gleb Natapov <gleb@redhat.com>]

On Wed, Jul 03, 2013 at 06:05:00PM +0900, Takuya Yoshikawa wrote:
> On Wed, 03 Jul 2013 10:53:51 +0200
> Paolo Bonzini <pbonzini@redhat.com> wrote:
> 
> > Il 03/07/2013 10:50, Xiao Guangrong ha scritto:
> > >> > Please wait a while. I can not understand it very clearly.
> > >> > 
> > >> > This conditional check will cause caching a overflow value into mmio spte.
> > >> > The simple case is that kvm adds new slots for many times, the mmio-gen is easily
> > >> > more than MMIO_MAX_GEN.
> > >> > 
> > > Actually, the double zapping can be avoided by moving kvm_mmu_invalidate_mmio_sptes to
> > > the end of install_new_memslots().
> > > 
> > > 
> > 
> > Yes, the actual operation would be the same as this patch.  You can
> > rename kvm_mmu_invalidate_mmio_sptes to kvm_arch_memslots_installed, or
> > something like that.  But it would have to touch all architectures.
> 
> I tried to avoid introducing x86-centric code into the generic one.
> 
> If another arch can gain something by such function, I'm willing to
> touch all arch code.
> 
Please do. X86 is the most optimized one so it does things other arches
do not yet. Slot generation update hook sounds generic enough.

--
			Gleb.