Re: [Query] About V4L2_CAP_VIDEO_CAPTURE_MPLANE device types

[Sylwester Nawrocki <snjw23@gmail.com>]

Hi Sergio,

On 05/01/2012 09:08 PM, Aguirre, Sergio wrote:
> Hi all,
> 
> I wonder if there's an example app for v4l2 devices with
> V4L2_CAP_VIDEO_CAPTURE_MPLANE capability?
> (like capture.c in V4L2 API docs)
 
There isn't at the V4L2 API docs, it probably makes sense to add one though.
I'll try to find a time to prepare a patch for that. A very simple application
with V4L2_CAP_VIDEO_CAPTURE_MPLANE support can be found here:
http://thread.gmane.org/gmane.comp.video.dri.devel/65997

It might be not that straightforward since there is the DMABUF memory type
used. Some more applications using multi-plane API are available at:
http://git.infradead.org/users/kmpark/public-apps

An MMAP memory example can be found here:
http://git.infradead.org/users/kmpark/public-apps/blob/9c057b001e8873861a70f7025214003837a0860b:/v4l2-mfc-example/mfc.c

Please see mfc_dec_setup_capture function.

> Also, does it have to be mutually exclusive with a
> V4L2_CAP_VIDEO_CAPTURE device?

No, the driver could support both. But it generally makes sense to implement
only _mplane ioctls at drivers. The multi/single-plane conversion should be 
done in user space. There were some patches for libv4l to support that
(http://www.spinics.net/lists/linux-media/msg35080.html), but such conversions
are not yet supported in the v4l libraries yet AFAIK.

I'm also not sure how is V4L2_CAP_VIDEO_CAPTURE_MPLANE handled in standard 
GStreamer plugins like v4l2src at the moment, having only multi-plane
interface at the driver might cause additional problems there.

--

Regards,
Sylwester