KVM: iommu: fix releasing unmapped page

There are two bugs:
- the 'error page' is forgot to be released
  [ it is unneeded after commit a2766325cf9f9, for backport, we
    still do kvm_release_pfn_clean for the error pfn ]

- guest pages are always released regardless of the unmapped page
  (e,g, caused by hwpoison)

Signed-off-by: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

virt/kvm/iommu.c

@@ -107,6 +107,7 @@ int kvm_iommu_map_pages(struct kvm *kvm, struct kvm_memory_slot *slot)
 		 */
 		pfn = kvm_pin_pages(slot, gfn, page_size);
 		if (is_error_pfn(pfn)) {
+			kvm_release_pfn_clean(pfn);
 			gfn += 1;
 			continue;
 		}
@@ -300,6 +301,12 @@ static void kvm_iommu_put_pages(struct kvm *kvm,
 
 		/* Get physical address */
 		phys = iommu_iova_to_phys(domain, gfn_to_gpa(gfn));
+
+		if (!phys) {
+			gfn++;
+			continue;
+		}
+
 		pfn  = phys >> PAGE_SHIFT;
 
 		/* Unmap address from IO address space */