KVM: x86: reset MMU on KVM_SET_VCPU_EVENTS

BugLink: http://bugs.launchpad.net/bugs/1655969

commit 6ef4e07e upstream.

Otherwise, mismatch between the smm bit in hflags and the MMU role
can cause a NULL pointer dereference.
Signed-off-by: Xiao Guangrong <guangrong.xiao@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

arch/x86/kvm/x86.c

@@ -2956,6 +2956,8 @@ static void kvm_vcpu_ioctl_x86_get_vcpu_events(struct kvm_vcpu *vcpu,
 	memset(&events->reserved, 0, sizeof(events->reserved));
 }
 
+static void kvm_set_hflags(struct kvm_vcpu *vcpu, unsigned emul_flags);
+
 static int kvm_vcpu_ioctl_x86_set_vcpu_events(struct kvm_vcpu *vcpu,
 					      struct kvm_vcpu_events *events)
 {
@@ -2988,10 +2990,13 @@ static int kvm_vcpu_ioctl_x86_set_vcpu_events(struct kvm_vcpu *vcpu,
 		vcpu->arch.apic->sipi_vector = events->sipi_vector;
 
 	if (events->flags & KVM_VCPUEVENT_VALID_SMM) {
+		u32 hflags = vcpu->arch.hflags;
 		if (events->smi.smm)
-			vcpu->arch.hflags |= HF_SMM_MASK;
+			hflags |= HF_SMM_MASK;
 		else
-			vcpu->arch.hflags &= ~HF_SMM_MASK;
+			hflags &= ~HF_SMM_MASK;
+		kvm_set_hflags(vcpu, hflags);
+
 		vcpu->arch.smi_pending = events->smi.pending;
 		if (events->smi.smm_inside_nmi)
 			vcpu->arch.hflags |= HF_SMM_INSIDE_NMI_MASK;