KVM: MMU: Fix potential memory leak with smp real-mode
authorIzik Eidus <izike@qumranet.com>
Tue, 20 Nov 2007 09:30:04 +0000 (11:30 +0200)
committerAvi Kivity <avi@qumranet.com>
Wed, 30 Jan 2008 15:53:08 +0000 (17:53 +0200)
When we map a page, we check whether some other vcpu mapped it for us and if
so, bail out.  But we should decrease the refcount on the page as we do so.

Signed-off-by: Izik Eidus <izike@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
drivers/kvm/mmu.c

index 87d8e70..8add4d5 100644 (file)
@@ -907,8 +907,10 @@ static int nonpaging_map(struct kvm_vcpu *vcpu, gva_t v, hpa_t p)
 
                        pte = table[index];
                        was_rmapped = is_rmap_pte(pte);
-                       if (is_shadow_present_pte(pte) && is_writeble_pte(pte))
+                       if (is_shadow_present_pte(pte) && is_writeble_pte(pte)) {
+                               kvm_release_page(pfn_to_page(p >> PAGE_SHIFT));
                                return 0;
+                       }
                        mark_page_dirty(vcpu->kvm, v >> PAGE_SHIFT);
                        page_header_update_slot(vcpu->kvm, table, v);
                        table[index] = p | PT_PRESENT_MASK | PT_WRITABLE_MASK |