Re: [PATCH v2 3/3] media: docs: dev-stateless-decoder: Document frame type flags

[Nicolas Dufresne <nicolas@ndufresne.ca>]

Le dimanche 16 janvier 2022 à 21:55 +0300, Dmitry Osipenko a écrit :
> Document that frame type flags must be set for the coded buffer.
> Decoders, like NVIDIA Tegra h264 decoder for example, won't work
> properly without these flags.
> 
> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
> ---
>  .../userspace-api/media/v4l/dev-stateless-decoder.rst          | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/Documentation/userspace-api/media/v4l/dev-stateless-decoder.rst b/Documentation/userspace-api/media/v4l/dev-stateless-decoder.rst
> index 4a26646eeec5..845f4481d34f 100644
> --- a/Documentation/userspace-api/media/v4l/dev-stateless-decoder.rst
> +++ b/Documentation/userspace-api/media/v4l/dev-stateless-decoder.rst
> @@ -271,6 +271,9 @@ A typical frame would thus be decoded using the following sequence:
>            we are not sure that the current decode request is the last one needed
>            to produce a fully decoded frame, then
>            ``V4L2_BUF_FLAG_M2M_HOLD_CAPTURE_BUF`` must also be set.
> +          ``V4L2_BUF_FLAG_KEYFRAME``, ``V4L2_BUF_FLAG_PFRAME`` and
> +          ``V4L2_BUF_FLAG_BFRAME`` must be set if relevant to the coded
> +          format.

I think I would rather keep this in the CODEC specific APIs (BFrame being an
MPEG specific concept). If I knew about this before we made the API final, I
would have added the slice_type bitstream parameter in the
v4l2_h264_decode_params structure. But as the types are rather limited, I would
preserve our "reservered" space and make that part of the
V4L2_H264_DECODE_PARAM_FLAGS_{} instead. Do you also need that for HEVC ? (do
you have HEVC on that generation of tegra ?)

>  
>        ``request_fd``
>            must be set to the file descriptor of the decoding request.