vhost: fix initialization for vq->is_le
authorHalil Pasic <pasic@linux.vnet.ibm.com>
Mon, 30 Jan 2017 10:09:36 +0000 (11:09 +0100)
committerMichael S. Tsirkin <mst@redhat.com>
Fri, 3 Feb 2017 21:38:57 +0000 (23:38 +0200)
commitcda8bba0f99d25d2061c531113c14fa41effc3ae
treea8efcd70106ad7dc0fb3b56bd1a0ccb79b757d26
parent0d5415b489f68b58e1983a53793d25d53098ed4b
vhost: fix initialization for vq->is_le

Currently, under certain circumstances vhost_init_is_le does just a part
of the initialization job, and depends on vhost_reset_is_le being called
too. For this reason vhost_vq_init_access used to call vhost_reset_is_le
when vq->private_data is NULL. This is not only counter intuitive, but
also real a problem because it breaks vhost_net. The bug was introduced to
vhost_net with commit 2751c9882b94 ("vhost: cross-endian support for
legacy devices"). The symptom is corruption of the vq's used.idx field
(virtio) after VHOST_NET_SET_BACKEND was issued as a part of the vhost
shutdown on a vq with pending descriptors.

Let us make sure the outcome of vhost_init_is_le never depend on the state
it is actually supposed to initialize, and fix virtio_net by removing the
reset from vhost_vq_init_access.

With the above, there is no reason for vhost_reset_is_le to do just half
of the job. Let us make vhost_reset_is_le reinitialize is_le.

Signed-off-by: Halil Pasic <pasic@linux.vnet.ibm.com>
Reported-by: Michael A. Tebolt <miket@us.ibm.com>
Reported-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Fixes: commit 2751c9882b94 ("vhost: cross-endian support for legacy devices")
Cc: <stable@vger.kernel.org>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Tested-by: Michael A. Tebolt <miket@us.ibm.com>
drivers/vhost/vhost.c