[FFmpeg-devel] [PATCH 1/2] avcodec/v4l2: fix single plane decoding

Mark Thompson sw at jkqxz.net
Sat Oct 7 01:53:10 EEST 2017


On 06/10/17 22:25, Jorge Ramirez-Ortiz wrote:
> On 10/06/2017 09:47 PM, Mark Thompson wrote:
>> On 06/10/17 08:51, Jorge Ramirez-Ortiz wrote:
>>> ---
>>>   libavcodec/v4l2_buffers.c | 14 ++++++++++++--
>>>   1 file changed, 12 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/libavcodec/v4l2_buffers.c b/libavcodec/v4l2_buffers.c
>>> index ef7d040..ba70c5d 100644
>>> --- a/libavcodec/v4l2_buffers.c
>>> +++ b/libavcodec/v4l2_buffers.c
>>> @@ -244,13 +244,23 @@ static int v4l2_buf_to_bufref(V4L2Buffer *in, int plane, AVBufferRef **buf)
>>>     static int v4l2_bufref_to_buf(V4L2Buffer *out, int plane, const uint8_t* data, int size, AVBufferRef* bref)
>>>   {
>>> +    unsigned int bytesused, length;
>>> +
>>>       if (plane >= out->num_planes)
>>>           return AVERROR(EINVAL);
>>>   +    bytesused = FFMIN(size, out->plane_info[plane].length);
>>> +    length = out->plane_info[plane].length;
>>> +
>>>       memcpy(out->plane_info[plane].mm_addr, data, FFMIN(size, out->plane_info[plane].length));
>>>   -    out->planes[plane].bytesused = FFMIN(size, out->plane_info[plane].length);
>>> -    out->planes[plane].length = out->plane_info[plane].length;
>>> +    if (V4L2_TYPE_IS_MULTIPLANAR(out->buf.type)) {
>>> +        out->planes[plane].bytesused = bytesused;
>>> +        out->planes[plane].length = length;
>>> +    } else {
>>> +        out->buf.bytesused = bytesused;
>>> +        out->buf.length = length;
>>> +    }
>>>         return 0;
>>>   }
>>>
>> Can you clarify what devices are being fixed here?  I believe there are two modes (single-plane, multi-plane) - of the devices known to work, do we know which modes they use?
> 
> venus driver in Qualcomm SoC - 96Bords db410c[1]  - > multiplanar
> coda driver for Imx6 - cubox[2]  -> single plane
> 
> [1] https://www.96boards.org/product/dragonboard410c/
> [2] https://www.solid-run.com/freescale-imx6-family/cubox-i/
> 
> the patch above fixes single plan decoding

Ok, good to know that this case is now tested.  LGTM, and applied.

Thanks,

- Mark


More information about the ffmpeg-devel mailing list