virtio_console: move early VQ enablement
authorCornelia Huck <cornelia.huck@de.ibm.com>
Mon, 20 Oct 2014 13:58:49 +0000 (15:58 +0200)
committerMichael S. Tsirkin <mst@redhat.com>
Thu, 13 Nov 2014 07:53:26 +0000 (09:53 +0200)
Commit f5866db6 (virtio_console: enable VQs early) tried to make
sure that DRIVER_OK was set when virtio_console started using its
virtqueues. Doing this in add_port(), however, means that we try
to set DRIVER_OK again when when a port is dynamically added after
the probe function is done.

Let's move virtio_device_ready() to the probe function just before
trying to use the virtqueues instead. This is fine as nothing can
fail inbetween.

Reported-by: Thomas Graf <tgraf@suug.ch>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
drivers/char/virtio_console.c

index bfa640023e64893251d2cb2cf54af5a853d89568..cf7a561fad7cd9f3cdd94fe16e5c22d65318428c 100644 (file)
@@ -1449,8 +1449,6 @@ static int add_port(struct ports_device *portdev, u32 id)
        spin_lock_init(&port->outvq_lock);
        init_waitqueue_head(&port->waitqueue);
 
-       virtio_device_ready(portdev->vdev);
-
        /* Fill the in_vq with buffers so the host can send us data. */
        nr_added_bufs = fill_queue(port->in_vq, &port->inbuf_lock);
        if (!nr_added_bufs) {
@@ -2026,6 +2024,8 @@ static int virtcons_probe(struct virtio_device *vdev)
        spin_lock_init(&portdev->ports_lock);
        INIT_LIST_HEAD(&portdev->ports);
 
+       virtio_device_ready(portdev->vdev);
+
        if (multiport) {
                unsigned int nr_added_bufs;