Re: [PATCH v2 10/11] media: cedrus: Adjust buffer size based on control values

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Jernej,

On 02/11/2022 19:08, Jernej Skrabec wrote:
> In some cases decoding engine needs extra space in capture buffers. This
> is the case for decoding 10-bit HEVC frames into 8-bit capture format.
> This commit only adds infrastructure for such cases.
> 
> Signed-off-by: Jernej Skrabec <jernej.skrabec@xxxxxxxxx>
> ---
>  drivers/staging/media/sunxi/cedrus/cedrus.c       | 14 ++++++++++++++
>  drivers/staging/media/sunxi/cedrus/cedrus.h       |  2 ++
>  drivers/staging/media/sunxi/cedrus/cedrus_video.c |  4 ++++
>  3 files changed, 20 insertions(+)
> 
> diff --git a/drivers/staging/media/sunxi/cedrus/cedrus.c b/drivers/staging/media/sunxi/cedrus/cedrus.c
> index 6a2c08928613..c36999e47591 100644
> --- a/drivers/staging/media/sunxi/cedrus/cedrus.c
> +++ b/drivers/staging/media/sunxi/cedrus/cedrus.c
> @@ -68,8 +68,22 @@ static int cedrus_try_ctrl(struct v4l2_ctrl *ctrl)
>  	return 0;
>  }
>  
> +static int cedrus_s_ctrl(struct v4l2_ctrl *ctrl)
> +{
> +	struct cedrus_ctx *ctx = container_of(ctrl->handler,
> +					      struct cedrus_ctx, hdl);
> +	struct vb2_queue *vq = v4l2_m2m_get_vq(ctx->fh.m2m_ctx,
> +					       V4L2_BUF_TYPE_VIDEO_CAPTURE);
> +
> +	if (!vb2_is_busy(vq) && !vb2_is_streaming(vq))

You can drop the !vb2_is_streaming part. If you are streaming, then
it will also be 'busy'.

> +		cedrus_reset_cap_format(ctx);

This is weird. I would expect that this is only called for specific controls,
since presumably not all will change the sizeimage value. It looks like
it applies only to V4L2_CID_STATELESS_HEVC_SPS, so I see no reason why
you would call cedrus_reset_cap_format() for other controls as well.

And what happens if someone makes such a change *while streaming*?
Shouldn't cedrus_try_ctrl detect that and fail with -EBUSY in that case?

Regardless, this is unexpected behavior, so some explanatory comments
would be useful.

I'm also not sure whether it isn't better to add cedrus_s_ctrl in the next
patch, I think it would be more understandable when seen with an actual
control.

Regards,

	Hans

> +
> +	return 0;
> +}
> +
>  static const struct v4l2_ctrl_ops cedrus_ctrl_ops = {
>  	.try_ctrl = cedrus_try_ctrl,
> +	.s_ctrl = cedrus_s_ctrl,
>  };
>  
>  static const struct cedrus_control cedrus_controls[] = {
> diff --git a/drivers/staging/media/sunxi/cedrus/cedrus.h b/drivers/staging/media/sunxi/cedrus/cedrus.h
> index 5904294f3108..774fe8048ce3 100644
> --- a/drivers/staging/media/sunxi/cedrus/cedrus.h
> +++ b/drivers/staging/media/sunxi/cedrus/cedrus.h
> @@ -162,6 +162,8 @@ struct cedrus_dec_ops {
>  	int (*start)(struct cedrus_ctx *ctx);
>  	void (*stop)(struct cedrus_ctx *ctx);
>  	void (*trigger)(struct cedrus_ctx *ctx);
> +	unsigned int (*extra_cap_size)(struct cedrus_ctx *ctx,
> +				       struct v4l2_pix_format *pix_fmt);
>  };
>  
>  struct cedrus_variant {
> diff --git a/drivers/staging/media/sunxi/cedrus/cedrus_video.c b/drivers/staging/media/sunxi/cedrus/cedrus_video.c
> index dec5d3ae4871..dc67940f1ade 100644
> --- a/drivers/staging/media/sunxi/cedrus/cedrus_video.c
> +++ b/drivers/staging/media/sunxi/cedrus/cedrus_video.c
> @@ -250,6 +250,10 @@ static int cedrus_try_fmt_vid_cap_p(struct cedrus_ctx *ctx,
>  	pix_fmt->height = ctx->src_fmt.height;
>  	cedrus_prepare_format(pix_fmt);
>  
> +	if (ctx->current_codec->extra_cap_size)
> +		pix_fmt->sizeimage +=
> +			ctx->current_codec->extra_cap_size(ctx, pix_fmt);
> +
>  	return 0;
>  }
>  





[Index of Archives]     [Linux Driver Development]     [Linux Driver Backports]     [DMA Engine]     [Linux GPIO]     [Linux SPI]     [Video for Linux]     [Linux USB Devel]     [Linux Coverity]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [Yosemite Backpacking]
  Powered by Linux