KVM: MMU: fix opposite condition in mapping_level_dirty_bitmap

The condition is opposite, it always maps huge page for the dirty tracked page
Reported-by: Steve <stefan.bosak@gmail.com>
Signed-off-by: Steve <stefan.bosak@gmail.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

arch/x86/kvm/mmu.c

@@ -565,7 +565,7 @@ gfn_to_memslot_dirty_bitmap(struct kvm_vcpu *vcpu, gfn_t gfn,
 
 static bool mapping_level_dirty_bitmap(struct kvm_vcpu *vcpu, gfn_t large_gfn)
 {
-	return gfn_to_memslot_dirty_bitmap(vcpu, large_gfn, true);
+	return !gfn_to_memslot_dirty_bitmap(vcpu, large_gfn, true);
 }
 
 static int mapping_level(struct kvm_vcpu *vcpu, gfn_t large_gfn)