Hi Micheal,
thanks for your work, I have some questions below ...
Le mardi 28 juin 2022 à 20:00 +0200, Michael Rodin a écrit :
> From: Niklas Söderlund <niklas.soderlund+renesas@xxxxxxxxxxxx>
>
> Add a new V4L2_EVENT_XFER_ERROR event to signal if an error happens during
> video transfer.
>
> The use-case that sparked this new event is to signal to the video
> device driver that an error has happen on the CSI-2 bus from the CSI-2
> receiver subdevice.
>
> Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@xxxxxxxxxxxx>
> [mrodin@xxxxxxxxxxxxxx: adapted information what to do if this new event is received]
> Signed-off-by: Michael Rodin <mrodin@xxxxxxxxxxxxxx>
> ---
> .../userspace-api/media/v4l/vidioc-dqevent.rst | 10 ++++++++++
> .../userspace-api/media/videodev2.h.rst.exceptions | 1 +
> include/uapi/linux/videodev2.h | 1 +
> 3 files changed, 12 insertions(+)
>
> diff --git a/Documentation/userspace-api/media/v4l/vidioc-dqevent.rst b/Documentation/userspace-api/media/v4l/vidioc-dqevent.rst
> index 6eb40073c906..3cf0b4859784 100644
> --- a/Documentation/userspace-api/media/v4l/vidioc-dqevent.rst
> +++ b/Documentation/userspace-api/media/v4l/vidioc-dqevent.rst
> @@ -182,6 +182,16 @@ call.
> the regions changes. This event has a struct
> :c:type:`v4l2_event_motion_det`
> associated with it.
> + * - ``V4L2_EVENT_XFER_ERROR``
I'm not sure why this event is specific to XFER. Is there uses cases were a
future implementation would have both XFER and RECEIVER error ?
> + - 7
> + - This event is triggered when an transfer error is detected while
> + streaming. For example if an error is detected on a video bus in
> + the pipeline. If a driver receives this event from an upstream
> + subdevice, it has to forward the event to userspace. The streaming
> + application has to check if the transfer error is unrecoverable,
> + i.e. no new buffers can be dequeued from the kernel after the
> + expected time. If the error is unrecoverable, the streaming
> + application should restart streaming if it wants to continue.
The process to determine if an error is recoverable or not isn't clear to me. As
an application developer, I would not know what to do here. Recoverable error
already have a designed mechanism, it consist of marking done a buffer with the
flag V4L2_BUF_FLAG_ERROR. I would like to understand what the existing mechanism
needed to be replaced, and the placement should be documented.
Nicolas
> * - ``V4L2_EVENT_PRIVATE_START``
> - 0x08000000
> - Base event number for driver-private events.
> diff --git a/Documentation/userspace-api/media/videodev2.h.rst.exceptions b/Documentation/userspace-api/media/videodev2.h.rst.exceptions
> index 9cbb7a0c354a..25bde61a1519 100644
> --- a/Documentation/userspace-api/media/videodev2.h.rst.exceptions
> +++ b/Documentation/userspace-api/media/videodev2.h.rst.exceptions
> @@ -500,6 +500,7 @@ replace define V4L2_EVENT_CTRL event-type
> replace define V4L2_EVENT_FRAME_SYNC event-type
> replace define V4L2_EVENT_SOURCE_CHANGE event-type
> replace define V4L2_EVENT_MOTION_DET event-type
> +replace define V4L2_EVENT_XFER_ERROR event-type
> replace define V4L2_EVENT_PRIVATE_START event-type
>
> replace define V4L2_EVENT_CTRL_CH_VALUE ctrl-changes-flags
> diff --git a/include/uapi/linux/videodev2.h b/include/uapi/linux/videodev2.h
> index 5311ac4fde35..44db724d4541 100644
> --- a/include/uapi/linux/videodev2.h
> +++ b/include/uapi/linux/videodev2.h
> @@ -2385,6 +2385,7 @@ struct v4l2_streamparm {
> #define V4L2_EVENT_FRAME_SYNC 4
> #define V4L2_EVENT_SOURCE_CHANGE 5
> #define V4L2_EVENT_MOTION_DET 6
> +#define V4L2_EVENT_XFER_ERROR 7
> #define V4L2_EVENT_PRIVATE_START 0x08000000
>
> /* Payload for V4L2_EVENT_VSYNC */
