VIVI driver issue with xawtv resize in low resolution mode

[dean <dean@sensoray.com>]

xawtv and/or libv4l2 is unmapping the V4L2 buffers before stopping the 
stream.  This causes xawtv and/or libv4l2 to retry VIDIOC_STREAMOFF 
multiple times before failing and the stream does not get restarted.

Tested on Ubuntu 8.10 with kernel upgraded to 2.6.29 and with 
v4l-dvb-b1596c6517c9.


To reproduce (does not fail if running with XV extensions):

1) load vivi driver
2) xawtv -noxv -nodga -noxv-image -c /dev/video0  (where video0 is vivi 
device, xawtv-3.95)
3) Resize the xawtv window.
4) Video output will stop and will not restart.

Should VIDIOC_STREAMOFF return -EINVAL if the stream is already off?  Is 
it ok to return 0 if the stream was turned off by an user application 
unmapping the buffers before stopping the stream?  In this example, 
videobuf_vm_close in videobuf-vmalloc.c is called before 
vidioc_streamoff in vivi.c.  videobuf_vm_close stops streaming in 
videobuf-vmalloc.c such that -EINVAL is returned on any subsequent 
VIDIOC_STREAMOFF ioctls. xawtv will not resume streaming in this case. 
This also affects other drivers using videobuf-vmalloc.  A possible 
workaround is the following(not an official patch) code below:

static int vidioc_streamoff(struct file *file, void *priv, enum 
v4l2_buf_type i)
{
	struct vivi_fh  *fh = priv;

	if (fh->type != V4L2_BUF_TYPE_VIDEO_CAPTURE)
		return -EINVAL;
	if (i != fh->type)
		return -EINVAL;
	return videobuf_streamoff(&fh->vb_vidq);
}

to:

static int vidioc_streamoff(struct file *file, void *priv, enum 
v4l2_buf_type i)
{
	struct vivi_fh  *fh = priv;

	if (fh->type != V4L2_BUF_TYPE_VIDEO_CAPTURE)
		return -EINVAL;
	if (i != fh->type)
		return -EINVAL;
	(void) videobuf_streamoff(&fh->vb_vidq);
	return 0;
}




--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list