Hi Niklas,
On 10/03/2021 17:45, Niklas Söderlund wrote:
> When a subdevice signals end of stream stop the VIN in addition to
> informing user-space of the event.
>
> Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@xxxxxxxxxxxx>
> Reviewed-by: Jacopo Mondi <jacopo+renesas@xxxxxxxxxx>
> ---
> * Changes since v2
> - Log using vin_dbg() instead of v4l2_info().
> ---
> drivers/media/platform/rcar-vin/rcar-v4l2.c | 16 +++++++++++++++-
> 1 file changed, 15 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/media/platform/rcar-vin/rcar-v4l2.c b/drivers/media/platform/rcar-vin/rcar-v4l2.c
> index 457a65bf6b664f05..76f6f46799e95791 100644
> --- a/drivers/media/platform/rcar-vin/rcar-v4l2.c
> +++ b/drivers/media/platform/rcar-vin/rcar-v4l2.c
> @@ -969,9 +969,23 @@ void rvin_v4l2_unregister(struct rvin_dev *vin)
> static void rvin_notify_video_device(struct rvin_dev *vin,
> unsigned int notification, void *arg)
> {
> + const struct v4l2_event *event;
> +
> switch (notification) {
> case V4L2_DEVICE_NOTIFY_EVENT:
> - v4l2_event_queue(&vin->vdev, arg);
> + event = arg;
> +
> + switch (event->type) {
> + case V4L2_EVENT_EOS:
This is not the right event for this. EVENT_EOS is specific to codecs and does
not signal an error condition.
I think we need a new event for this, I've seen similar situations with HDMI
receivers. The SOURCE_CHANGE event was (ab)used for that, but I think a new
event (V4L2_EVENT_XFER_ERROR?) should be created for these situations.
> + rvin_stop_streaming(vin);
In addition to this you should call vb2_queue_error(), that's typically used
in situations where there is an unrecoverable transfer error.
Regards,
Hans
> + vin_dbg(vin,
> + "Subdevice signaled end of stream, stopping.\n");
> + break;
> + default:
> + break;
> + }
> +
> + v4l2_event_queue(&vin->vdev, event);
> break;
> default:
> break;
>
