Re: [PATCH v2 3/3] media: vimc: Join pipeline if one already exists

[Kaaira Gupta <kgupta@es.iitr.ac.in>]

On Tue, Jul 28, 2020 at 02:24:37PM +0200, Dafna Hirschfeld wrote:
> 
> 
> On 24.07.20 14:02, Kaaira Gupta wrote:
> > An output which is running is already part of a pipeline and trying to
> > start a new pipeline is not possible. This prevents two capture devices
> > from streaming at the same time.
> > 
> > Instead of failing to start the second capture device allow it to join
> > the already running pipeline. This allows two (or more) capture devices
> > to independently be started and stopped.
> > 
> > [Kaaira: Changed the search for an existing connected sensor, to search
> > for a non-NULL pipe instead, this helps to terminate the search at
> > output itself instead of going till the sensor, changed variable names,
> > commit message and conditions accordingly]
> > 
> > Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
> > Signed-off-by: Kaaira Gupta <kgupta@es.iitr.ac.in>
> > ---
> >   .../media/test-drivers/vimc/vimc-capture.c    | 35 ++++++++++++++++++-
> >   1 file changed, 34 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/media/test-drivers/vimc/vimc-capture.c b/drivers/media/test-drivers/vimc/vimc-capture.c
> > index c63496b17b9a..423d5e5a508d 100644
> > --- a/drivers/media/test-drivers/vimc/vimc-capture.c
> > +++ b/drivers/media/test-drivers/vimc/vimc-capture.c
> > @@ -237,16 +237,49 @@ static void vimc_cap_return_all_buffers(struct vimc_cap_device *vcap,
> >   	spin_unlock(&vcap->qlock);
> >   }
> > +static struct media_entity *vimc_cap_get_output(struct vimc_cap_device *vcap)
> > +{
> > +	struct media_entity *entity = &vcap->vdev.entity;
> > +	struct media_device *mdev = entity->graph_obj.mdev;
> > +	struct media_graph graph;
> > +
> > +	mutex_lock(&mdev->graph_mutex);
> > +	if (media_graph_walk_init(&graph, mdev)) {
> > +		mutex_unlock(&mdev->graph_mutex);
> > +		return NULL;
> > +	}
> > +
> > +	media_graph_walk_start(&graph, entity);
> > +
> > +	while ((entity = media_graph_walk_next(&graph)))
> > +		if (entity->pipe)
> > +			break;
> > +
> > +	mutex_unlock(&mdev->graph_mutex);
> > +
> > +	media_graph_walk_cleanup(&graph);
> > +
> > +	return entity;
> > +}
> > +
> >   static int vimc_cap_start_streaming(struct vb2_queue *vq, unsigned int count)
> >   {
> >   	struct vimc_cap_device *vcap = vb2_get_drv_priv(vq);
> >   	struct media_entity *entity = &vcap->vdev.entity;
> > +	struct media_pipeline *pipe = NULL;
> > +	struct media_entity *oent;
> >   	int ret;
> >   	vcap->sequence = 0;
> >   	/* Start the media pipeline */
> > -	ret = media_pipeline_start(entity, &vcap->stream.pipe);
> > +	oent = vimc_cap_get_output(vcap);
> > +	if (oent)
> > +		pipe = oent->pipe;
> > +	else
> > +		pipe = &vcap->stream.pipe;
> > +
> > +	ret = media_pipeline_start(entity, pipe);
> >   	if (ret) {
> >   		vimc_cap_return_all_buffers(vcap, VB2_BUF_STATE_QUEUED);
> >   		return ret;
> > 
> 
> I think there is actually no need to iterate the graph. If the capture is already connected to another capture
> that streams, it means that they both have the same pipe in the media core.
> So actually the code can just be:

Hi,
iterating the graph takes care of the case when output already exists.
So in case where an output is needed from both Capture1 and RGB capture,
don't they represent two different pipes?

> 
> if (vcap->ved.ent->pipe)
> 	pipe = vcap->ved.ent->pipe;
> else
> 	pipe = &vcap->stream.pipe;
> 
> 
> (and no need the function vimc_cap_get_output)
> 
> Thanks,
> Dafna