On 04/04/2022 18:35, Philipp Zabel wrote:
> The V4L2 specification states:
>
> "If the application sets this to 0 for an output stream, then bytesused
> will be set to the size of the buffer (see the length field of this
> struct) by the driver."
>
> Since we set allow_zero_bytesused, we have to handle this ourselves.
>
> Signed-off-by: Philipp Zabel <p.zabel@xxxxxxxxxxxxxx>
> ---
> drivers/media/platform/chips-media/coda-bit.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/media/platform/chips-media/coda-bit.c b/drivers/media/platform/chips-media/coda-bit.c
> index c484c008ab02..705a179ea8f0 100644
> --- a/drivers/media/platform/chips-media/coda-bit.c
> +++ b/drivers/media/platform/chips-media/coda-bit.c
> @@ -381,6 +381,9 @@ void coda_fill_bitstream(struct coda_ctx *ctx, struct list_head *buffer_list)
> /* Dump empty buffers */
> if (!vb2_get_plane_payload(&src_buf->vb2_buf, 0)) {
> src_buf = v4l2_m2m_src_buf_remove(ctx->fh.m2m_ctx);
> + vb2_set_plane_payload(&src_buf->vb2_buf, 0,
> + vb2_plane_size(&src_buf->vb2_buf,
> + 0));
Would it be possible to stop using allow_zero_bytesused altogether?
Are there still applications that rely on zero-sized output buffers to stop the
decoder?
I'm not actually sure that I want this in the driver, perhaps v4l2-compliance
can be modified to turn a fail into a warn if the driver is the coda driver.
Patching the driver is hiding the fact that the coda driver does something
non-standard for legacy reasons. It doesn't make sense either to change
bytesused to the buffer size since there really is nothing in the buffer.
v4l2-compliance already has checks for two drivers, search for is_vivid and
is_uvcvideo.
I'm skipping this patch for now.
Regards,
Hans
> v4l2_m2m_buf_done(src_buf, VB2_BUF_STATE_DONE);
> continue;
> }
