Hi All,
If the host system decides that it wants to push a given page out to
swap, the host will notify the host through registered mmu notifier to
inform the guest. I am wondering if there any other situations, other
than swapping, which will trigger the mmu notifier to inform host page
change to
Hi Andrea,
Few comments below.
On Thu, Jun 26, 2008 at 08:11:16PM +0200, Andrea Arcangeli wrote:
> + }
> + return need_tlb_flush;
> +}
> +
> +int kvm_unmap_hva(struct kvm *kvm, unsigned long hva)
> +{
> + int i;
> + int need_tlb_flush = 0;
> +
> + /*
> + * If mmap_sem is
Hi Andrea,
On Thu, Jul 03, 2008 at 05:17:42PM +0200, Andrea Arcangeli wrote:
> +static int kvm_unmap_rmapp(struct kvm *kvm, unsigned long *rmapp)
> +{
> + u64 *spte;
> + int need_tlb_flush = 0;
> +
> + while ((spte = rmap_next(kvm, rmapp, NULL))) {
> + BUG_ON(!(*spte & PT_
l sequence number/counter are
under the spinlock and all smb_wmb/rmb goes away).
Given mmu-notifier are in -mm and everyone seem to agrees about them
being merged during .27 merge window, I think it'd be fairly safe to
apply kvm mmu notifier patch.
I'm testing this on both mmu-notifie
On Fri, Jun 27, 2008 at 07:43:17PM -0300, Marcelo Tosatti wrote:
> These two functions share the same memslot iteration code, you could
> avoid duplication.
Ok, I'll look into cleaning this up a bit.
> > + if (unlikely(atomic_read(&vcpu->kvm->mmu_notifier_count)))
> > + return;
> > +
Hello,
this is the mmu notifier patch for kvm updated according to last
comments.
Signed-off-by: Andrea Arcangeli <[EMAIL PROTECTED]>
diff --git a/arch/x86/kvm/Kconfig b/arch/x86/kvm/Kconfig
index 8d45fab..ce3251c 100644
--- a/arch/x86/kvm/Kconfig
+++ b/arch/x86/kvm/Kconfig
@@ -21,6 +21,7 @@ con