KVM: MMU: Fix incorrect direct gfn for unpaged mode shadow
authorAvi Kivity <avi@redhat.com>
Tue, 28 Dec 2010 10:09:07 +0000 (12:09 +0200)
committerAvi Kivity <avi@redhat.com>
Wed, 29 Dec 2010 10:35:29 +0000 (12:35 +0200)
We use the physical address instead of the base gfn for the four
PAE page directories we use in unpaged mode.  When the guest accesses
an address above 1GB that is backed by a large host page, a BUG_ON()
in kvm_mmu_set_gfn() triggers.

Resolves: https://bugzilla.kernel.org/show_bug.cgi?id=21962
Reported-and-tested-by: Nicolas Prochazka <prochazka.nicolas@gmail.com>
KVM-Stable-Tag.
Signed-off-by: Avi Kivity <avi@redhat.com>
arch/x86/kvm/mmu.c

index fb8b376bf28cb3e04a6bb903900f32838ab02a14..fbb04aee8301efab741f80f7c590a45d7e3b298f 100644 (file)
@@ -2394,7 +2394,8 @@ static int mmu_alloc_direct_roots(struct kvm_vcpu *vcpu)
                        ASSERT(!VALID_PAGE(root));
                        spin_lock(&vcpu->kvm->mmu_lock);
                        kvm_mmu_free_some_pages(vcpu);
-                       sp = kvm_mmu_get_page(vcpu, i << 30, i << 30,
+                       sp = kvm_mmu_get_page(vcpu, i << (30 - PAGE_SHIFT),
+                                             i << 30,
                                              PT32_ROOT_LEVEL, 1, ACC_ALL,
                                              NULL);
                        root = __pa(sp->spt);