[libcamera-devel] [PATCH v2] v4l2: V4L2CameraProxy: Add EXPBUF as one of the supported ioctl

Laurent Pinchart laurent.pinchart at ideasonboard.com
Sun Nov 28 17:37:18 CET 2021


On Sun, Nov 28, 2021 at 03:31:40PM +0200, Laurent Pinchart wrote:
> Hi Vedant,
> 
> Thank you for the patch.
> 
> On Sun, Nov 28, 2021 at 11:45:56AM +0530, Vedant Paranjape wrote:
> > To support DMABUF as one of the memory buffer, we need to implement
> > EXPBUF in the v4l2 compat layer.
> > 
> > This patch implements vidioc_expbuf as one of the supported ioctls.
> > 
> > Bug: https://bugs.libcamera.org/show_bug.cgi?id=89
> > 
> > Signed-off-by: Vedant Paranjape <vedantparanjape160201 at gmail.com>
> > ---
> >  src/v4l2/v4l2_camera_proxy.cpp | 37 ++++++++++++++++++++++++++++++++++
> >  src/v4l2/v4l2_camera_proxy.h   |  1 +
> >  2 files changed, 38 insertions(+)
> > 
> > diff --git a/src/v4l2/v4l2_camera_proxy.cpp b/src/v4l2/v4l2_camera_proxy.cpp
> > index 3610e63cade3..cb2f93631ab9 100644
> > --- a/src/v4l2/v4l2_camera_proxy.cpp
> > +++ b/src/v4l2/v4l2_camera_proxy.cpp
> > @@ -624,6 +624,39 @@ int V4L2CameraProxy::vidioc_dqbuf(V4L2CameraFile *file, struct v4l2_buffer *arg,
> >  	return 0;
> >  }
> >  
> > +int V4L2CameraProxy::vidioc_expbuf(V4L2CameraFile *file, struct v4l2_exportbuffer *arg)
> > +{
> > +	LOG(V4L2Compat, Debug) << "Servicing vidioc_expbuf fd = " << file->efd();
> > +
> > +	if (!hasOwnership(file))
> > +		return -EBUSY;
> > +
> > +	/* \todo Verify that the memory type is MMAP when adding DMABUF support */
> > +	if (!validateBufferType(arg->type))
> > +		return -EINVAL;
> > +
> > +	if (arg->index >= bufferCount_)
> > +		return -EINVAL;
> > +
> > +	if (arg->flags & ~(O_CLOEXEC | O_ACCMODE))
> > +		return -EINVAL;
> > +
> > +	LOG(V4L2Compat, Debug) << arg->index << " : index of the buffer";
> 
> I'd either drop this, or combine it with the message at the top of the
> function.
> 
> > +
> > +	if (file->priority() < maxPriority())
> > +		return -EBUSY;
> 
> Why is this chech needed ?
> 
> > +
> > +	memset(arg->reserved, 0, sizeof(arg->reserved));
> > +
> > +	arg->fd = fcntl(buffers_[arg->index].m.fd,
> > +			arg->flags & O_CLOEXEC ? F_DUPFD_CLOEXEC : F_DUPFD, 0);

Actually I don't think this is correct. .m.fd is for importing dmabufs,
it's always 0 at the moment.

If this has passed v4l2-compliance, it means EXPBUF isn't properly
tested by it.

> > +	arg->fd = fcntl(arg->fd, F_SETFL, 0);
> 
> 0 isn't the right value.
> 
> > +
> > +	LOG(V4L2Compat, Debug) << "Exported buffer at index: " << arg->index;
> 
> I'd drop this too.
> 
> > +
> > +	return 0;
> > +}
> > +
> >  int V4L2CameraProxy::vidioc_streamon(V4L2CameraFile *file, int *arg)
> >  {
> >  	LOG(V4L2Compat, Debug) << "Servicing vidioc_streamon fd = " << file->efd();
> > @@ -685,6 +718,7 @@ const std::set<unsigned long> V4L2CameraProxy::supportedIoctls_ = {
> >  	VIDIOC_QUERYBUF,
> >  	VIDIOC_QBUF,
> >  	VIDIOC_DQBUF,
> > +	VIDIOC_EXPBUF,
> >  	VIDIOC_STREAMON,
> >  	VIDIOC_STREAMOFF,
> >  };
> > @@ -755,6 +789,9 @@ int V4L2CameraProxy::ioctl(V4L2CameraFile *file, unsigned long request, void *ar
> >  	case VIDIOC_DQBUF:
> >  		ret = vidioc_dqbuf(file, static_cast<struct v4l2_buffer *>(arg), &locker);
> >  		break;
> > +	case VIDIOC_EXPBUF:
> > +		ret = vidioc_expbuf(file, static_cast<struct v4l2_exportbuffer *>(arg));
> > +		break;
> >  	case VIDIOC_STREAMON:
> >  		ret = vidioc_streamon(file, static_cast<int *>(arg));
> >  		break;
> > diff --git a/src/v4l2/v4l2_camera_proxy.h b/src/v4l2/v4l2_camera_proxy.h
> > index fccec241879d..81ef7788e9fe 100644
> > --- a/src/v4l2/v4l2_camera_proxy.h
> > +++ b/src/v4l2/v4l2_camera_proxy.h
> > @@ -58,6 +58,7 @@ private:
> >  	int vidioc_qbuf(V4L2CameraFile *file, struct v4l2_buffer *arg);
> >  	int vidioc_dqbuf(V4L2CameraFile *file, struct v4l2_buffer *arg,
> >  			 libcamera::MutexLocker *locker);
> > +	int vidioc_expbuf(V4L2CameraFile *file, struct v4l2_exportbuffer *arg);
> >  	int vidioc_streamon(V4L2CameraFile *file, int *arg);
> >  	int vidioc_streamoff(V4L2CameraFile *file, int *arg);
> >  

-- 
Regards,

Laurent Pinchart


More information about the libcamera-devel mailing list