[PATCH 4.9 059/104] KVM: pci-assign: do not map smm memory slot pages in vt-d page tables

From: Greg Kroah-Hartman
Date: Fri Dec 22 2017 - 05:18:48 EST


4.9-stable review patch. If anyone has any objections, please let me know.

------------------

From: "Herongguang (Stephen)" <herongguang.he@xxxxxxxxxx>


[ Upstream commit 0292e169b2d9c8377a168778f0b16eadb1f578fd ]

or VM memory are not put thus leaked in kvm_iommu_unmap_memslots() when
destroy VM.

This is consistent with current vfio implementation.

Signed-off-by: herongguang <herongguang.he@xxxxxxxxxx>
Signed-off-by: Paolo Bonzini <pbonzini@xxxxxxxxxx>
Signed-off-by: Sasha Levin <alexander.levin@xxxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
virt/kvm/kvm_main.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -1060,7 +1060,7 @@ int __kvm_set_memory_region(struct kvm *
* changes) is disallowed above, so any other attribute changes getting
* here can be skipped.
*/
- if ((change == KVM_MR_CREATE) || (change == KVM_MR_MOVE)) {
+ if (as_id == 0 && (change == KVM_MR_CREATE || change == KVM_MR_MOVE)) {
r = kvm_iommu_map_pages(kvm, &new);
return r;
}