KVM: nVMX: fix lifetime issues for vmcs02
authorPaolo Bonzini <pbonzini@redhat.com>
Thu, 17 Jul 2014 10:25:16 +0000 (12:25 +0200)
committerPaolo Bonzini <pbonzini@redhat.com>
Mon, 21 Jul 2014 12:29:49 +0000 (14:29 +0200)
free_nested needs the loaded_vmcs to be valid if it is a vmcs02, in
order to detach it from the shadow vmcs.  However, this is not
available anymore after commit 26a865f4aa8e (KVM: VMX: fix use after
free of vmx->loaded_vmcs, 2014-01-03).

Revert that patch, and fix its problem by forcing a vmcs01 as the
active VMCS before freeing all the nested VMX state.

Reported-by: Wanpeng Li <wanpeng.li@linux.intel.com>
Tested-by: Wanpeng Li <wanpeng.li@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
arch/x86/kvm/vmx.c

index 7534a9f67cc866f8ab8f415a5c86c6f200af0dda..462334eaa3c008acefe7e26729e7e7f7dbe54bf6 100644 (file)
@@ -5772,22 +5772,27 @@ static void nested_free_vmcs02(struct vcpu_vmx *vmx, gpa_t vmptr)
 
 /*
  * Free all VMCSs saved for this vcpu, except the one pointed by
- * vmx->loaded_vmcs. These include the VMCSs in vmcs02_pool (except the one
- * currently used, if running L2), and vmcs01 when running L2.
+ * vmx->loaded_vmcs. We must be running L1, so vmx->loaded_vmcs
+ * must be &vmx->vmcs01.
  */
 static void nested_free_all_saved_vmcss(struct vcpu_vmx *vmx)
 {
        struct vmcs02_list *item, *n;
+
+       WARN_ON(vmx->loaded_vmcs != &vmx->vmcs01);
        list_for_each_entry_safe(item, n, &vmx->nested.vmcs02_pool, list) {
-               if (vmx->loaded_vmcs != &item->vmcs02)
-                       free_loaded_vmcs(&item->vmcs02);
+               /*
+                * Something will leak if the above WARN triggers.  Better than
+                * a use-after-free.
+                */
+               if (vmx->loaded_vmcs == &item->vmcs02)
+                       continue;
+
+               free_loaded_vmcs(&item->vmcs02);
                list_del(&item->list);
                kfree(item);
+               vmx->nested.vmcs02_num--;
        }
-       vmx->nested.vmcs02_num = 0;
-
-       if (vmx->loaded_vmcs != &vmx->vmcs01)
-               free_loaded_vmcs(&vmx->vmcs01);
 }
 
 /*
@@ -7557,13 +7562,31 @@ static void __noclone vmx_vcpu_run(struct kvm_vcpu *vcpu)
        vmx_complete_interrupts(vmx);
 }
 
+static void vmx_load_vmcs01(struct kvm_vcpu *vcpu)
+{
+       struct vcpu_vmx *vmx = to_vmx(vcpu);
+       int cpu;
+
+       if (vmx->loaded_vmcs == &vmx->vmcs01)
+               return;
+
+       cpu = get_cpu();
+       vmx->loaded_vmcs = &vmx->vmcs01;
+       vmx_vcpu_put(vcpu);
+       vmx_vcpu_load(vcpu, cpu);
+       vcpu->cpu = cpu;
+       put_cpu();
+}
+
 static void vmx_free_vcpu(struct kvm_vcpu *vcpu)
 {
        struct vcpu_vmx *vmx = to_vmx(vcpu);
 
        free_vpid(vmx);
-       free_loaded_vmcs(vmx->loaded_vmcs);
+       leave_guest_mode(vcpu);
+       vmx_load_vmcs01(vcpu);
        free_nested(vmx);
+       free_loaded_vmcs(vmx->loaded_vmcs);
        kfree(vmx->guest_msrs);
        kvm_vcpu_uninit(vcpu);
        kmem_cache_free(kvm_vcpu_cache, vmx);
@@ -8721,7 +8744,6 @@ static void nested_vmx_vmexit(struct kvm_vcpu *vcpu, u32 exit_reason,
                              unsigned long exit_qualification)
 {
        struct vcpu_vmx *vmx = to_vmx(vcpu);
-       int cpu;
        struct vmcs12 *vmcs12 = get_vmcs12(vcpu);
 
        /* trying to cancel vmlaunch/vmresume is a bug */
@@ -8746,12 +8768,7 @@ static void nested_vmx_vmexit(struct kvm_vcpu *vcpu, u32 exit_reason,
                                       vmcs12->vm_exit_intr_error_code,
                                       KVM_ISA_VMX);
 
-       cpu = get_cpu();
-       vmx->loaded_vmcs = &vmx->vmcs01;
-       vmx_vcpu_put(vcpu);
-       vmx_vcpu_load(vcpu, cpu);
-       vcpu->cpu = cpu;
-       put_cpu();
+       vmx_load_vmcs01(vcpu);
 
        vm_entry_controls_init(vmx, vmcs_read32(VM_ENTRY_CONTROLS));
        vm_exit_controls_init(vmx, vmcs_read32(VM_EXIT_CONTROLS));