[FFmpeg-devel] [PATCH 2/2] mjpegdec: fix handling of rstn in progressive jpegs.

[Alexander Strasser]

Hi Michael,

Michael Niedermayer wrote:
> mjpegdec: fix handling of rstn in progressive jpegs.

> Fixes Ticket2003

  I find the commit message a bit hard to deal with. Either I understood
the code correctly and RSTn handling was missing entirely for progressive
jpegs or you fixed some existing RSTn handling code.

  In the first case I would find "mjpegdec: Handle RSTn in progressive jpegs"
much clearer. In the second case it would be good to have a sentence or two
in the body of the commit message that say what was wrong before/is done
better with the patch.

  Patch seems fine to me. But you have better domain knowledge of jpeg
decoding and our jpeg decoder.

  Alexander

> Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
> ---
>  libavcodec/mjpegdec.c |    4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/libavcodec/mjpegdec.c b/libavcodec/mjpegdec.c
> index cf7dd8c..2ad1e2d 100644
> --- a/libavcodec/mjpegdec.c
> +++ b/libavcodec/mjpegdec.c
> @@ -1141,6 +1141,9 @@ static int mjpeg_decode_scan_progressive_ac(MJpegDecodeContext *s, int ss,
>          uint8_t *last_nnz    = &s->last_nnz[c][block_idx];
>          for (mb_x = 0; mb_x < s->mb_width; mb_x++, block++, last_nnz++) {
>                  int ret;
> +                if (s->restart_interval && !s->restart_count)
> +                    s->restart_count = s->restart_interval;
> +
>                  if (Ah)
>                      ret = decode_block_refinement(s, *block, last_nnz, s->ac_index[0],
>                                                    quant_matrix, ss, se, Al, &EOBRUN);
> @@ -1157,6 +1160,7 @@ static int mjpeg_decode_scan_progressive_ac(MJpegDecodeContext *s, int ss,
>                      s->dsp.idct_put(ptr, linesize, *block);
>                      ptr += 8 >> s->avctx->lowres;
>              }
> +            handle_rstn(s, 0);
>          }
>      }
>      return 0;
> -- 
> 1.7.9.5