Commit acb66dd0 authored by Izik Eidus's avatar Izik Eidus Committed by Avi Kivity

KVM: MMU: dont hold pagecount reference for mapped sptes pages

When using mmu notifiers, we are allowed to remove the page count
reference tooken by get_user_pages to a specific page that is mapped
inside the shadow page tables.

This is needed so we can balance the pagecount against mapcount
checking.

(Right now kvm increase the pagecount and does not increase the
mapcount when mapping page into shadow page table entry,
so when comparing pagecount against mapcount, you have no
reliable result.)
Signed-off-by: default avatarIzik Eidus <ieidus@redhat.com>
Signed-off-by: default avatarMarcelo Tosatti <mtosatti@redhat.com>
parent 6a544355
...@@ -634,9 +634,7 @@ static void rmap_remove(struct kvm *kvm, u64 *spte) ...@@ -634,9 +634,7 @@ static void rmap_remove(struct kvm *kvm, u64 *spte)
if (*spte & shadow_accessed_mask) if (*spte & shadow_accessed_mask)
kvm_set_pfn_accessed(pfn); kvm_set_pfn_accessed(pfn);
if (is_writeble_pte(*spte)) if (is_writeble_pte(*spte))
kvm_release_pfn_dirty(pfn); kvm_set_pfn_dirty(pfn);
else
kvm_release_pfn_clean(pfn);
rmapp = gfn_to_rmap(kvm, sp->gfns[spte - sp->spt], sp->role.level); rmapp = gfn_to_rmap(kvm, sp->gfns[spte - sp->spt], sp->role.level);
if (!*rmapp) { if (!*rmapp) {
printk(KERN_ERR "rmap_remove: %p %llx 0->BUG\n", spte, *spte); printk(KERN_ERR "rmap_remove: %p %llx 0->BUG\n", spte, *spte);
...@@ -1877,8 +1875,7 @@ static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *sptep, ...@@ -1877,8 +1875,7 @@ static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *sptep,
page_header_update_slot(vcpu->kvm, sptep, gfn); page_header_update_slot(vcpu->kvm, sptep, gfn);
if (!was_rmapped) { if (!was_rmapped) {
rmap_count = rmap_add(vcpu, sptep, gfn); rmap_count = rmap_add(vcpu, sptep, gfn);
if (!is_rmap_spte(*sptep)) kvm_release_pfn_clean(pfn);
kvm_release_pfn_clean(pfn);
if (rmap_count > RMAP_RECYCLE_THRESHOLD) if (rmap_count > RMAP_RECYCLE_THRESHOLD)
rmap_recycle(vcpu, sptep, gfn); rmap_recycle(vcpu, sptep, gfn);
} else { } else {
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment