KVM: x86: Improve thread safety in pit

commit 2febc839133280d5a5e8e1179c94ea674489dae2 upstream.

There's a race condition in the PIT emulation code in KVM.  In
__kvm_migrate_pit_timer the pit_timer object is accessed without
synchronization.  If the race condition occurs at the wrong time this
can crash the host kernel.

This fixes CVE-2014-3611.

Signed-off-by: Andrew Honig <ahonig@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
[bwh: Backported to 3.2: adjust context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

diff --git a/arch/x86/kvm/i8254.c b/arch/x86/kvm/i8254.c
index 139415e..cced57f 100644 (file)
--- a/arch/x86/kvm/i8254.c
+++ b/arch/x86/kvm/i8254.c
@@ -264,8 +264,10 @@ void __kvm_migrate_pit_timer(struct kvm_vcpu *vcpu)
                return;
 
        timer = &pit->pit_state.pit_timer.timer;
+       mutex_lock(&pit->pit_state.lock);
        if (hrtimer_cancel(timer))
                hrtimer_start_expires(timer, HRTIMER_MODE_ABS);
+       mutex_unlock(&pit->pit_state.lock);
 }
 
 static void destroy_pit_timer(struct kvm_pit *pit)