KVM: eventfd: Fix lock order inversion.

When registering a new irqfd, we call its ->poll method to collect any
event that might have previously been pending so that we can trigger it.
This is done under the kvm->irqfds.lock, which means the eventfd's ctx
lock is taken under it.

However, if we get a POLLHUP in irqfd_wakeup, we will be called with the
ctx lock held before getting the irqfds.lock to deactivate the irqfd,
causing lockdep to complain.

Calling the ->poll method does not really need the irqfds.lock, so let's
just move it after we've given up the irqfds.lock in kvm_irqfd_assign().

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c
index abe4d60..29c2a04 100644 (file)
--- a/virt/kvm/eventfd.c
+++ b/virt/kvm/eventfd.c
@@ -391,19 +391,19 @@ kvm_irqfd_assign(struct kvm *kvm, struct kvm_irqfd *args)
                                           lockdep_is_held(&kvm->irqfds.lock));
        irqfd_update(kvm, irqfd, irq_rt);
 
-       events = f.file->f_op->poll(f.file, &irqfd->pt);
-
        list_add_tail(&irqfd->list, &kvm->irqfds.items);
 
+       spin_unlock_irq(&kvm->irqfds.lock);
+
        /*
         * Check if there was an event already pending on the eventfd
         * before we registered, and trigger it as if we didn't miss it.
         */
+       events = f.file->f_op->poll(f.file, &irqfd->pt);
+
        if (events & POLLIN)
                schedule_work(&irqfd->inject);
 
-       spin_unlock_irq(&kvm->irqfds.lock);
-
        /*
         * do not drop the file until the irqfd is fully initialized, otherwise
         * we might race against the POLLHUP