Commit e3f78718 authored Nov 30, 2018 by Jean-Philippe Brucker Committed by David S. Miller Dec 03, 2018

vhost: fix IOTLB locking



Commit 78139c94 ("net: vhost: lock the vqs one by one") moved the vq
lock to improve scalability, but introduced a possible deadlock in
vhost-iotlb. vhost_iotlb_notify_vq() now takes vq->mutex while holding
the device's IOTLB spinlock. And on the vhost_iotlb_miss() path, the
spinlock is taken while holding vq->mutex.

Since calling vhost_poll_queue() doesn't require any lock, avoid the
deadlock by not taking vq->mutex.

Fixes: 78139c94 ("net: vhost: lock the vqs one by one")
Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jean-Philippe Brucker <jean-philippe.brucker@arm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

parent 4763c9f9

drivers/vhost/vhost.c

+0 −3

Original line number	Diff line number	Diff line
		@@ -944,10 +944,7 @@ static void vhost_iotlb_notify_vq(struct vhost_dev *d,
		if (msg->iova <= vq_msg->iova &&
		msg->iova + msg->size - 1 >= vq_msg->iova &&
		vq_msg->type == VHOST_IOTLB_MISS) {
		mutex_lock(&node->vq->mutex);
		vhost_poll_queue(&node->vq->poll);
		mutex_unlock(&node->vq->mutex);

		list_del(&node->node);
		kfree(node);
		}

Admin message