Hi Hans, On Tue, Jun 06, 2017 at 11:08:07AM +0200, Hans Verkuil wrote: > > + /* Initialize vdev */ > > + strlcpy(vdev->name, node->name, sizeof(vdev->name)); > > + vdev->release = video_device_release_empty; > > + vdev->fops = &m2m2->v4l2_file_ops; > > + vdev->ioctl_ops = &ipu3_v4l2_ioctl_ops; > > + vdev->lock = &node->lock; > > + vdev->v4l2_dev = &m2m2->v4l2_dev; > > + vdev->queue = &node->vbq; > > + vdev->vfl_dir = node->output ? VFL_DIR_TX : VFL_DIR_RX; > > Why have two video nodes (one tx, one rx) instead of a single m2m device > node? > > I'm not saying this is wrong, I just like to know the rationale for this > design. There are a bunch of outputs from the same input stream. Also the parameters are needed to process the frame, I think there are two OUTPUT devices and five CAPTURE devices. Yong: could you send the media graph of the device, please? -- Regards, Sakari Ailus e-mail: sakari.ailus@xxxxxx XMPP: sailus@xxxxxxxxxxxxxx
