Re: [FFmpeg-devel] [PATCH 3/6] avcodec/h264_refs: Rewrite code to make control flow clearer

[Andreas Rheinhardt <andreas.rheinhardt@outlook.com>]

Andreas Rheinhardt:
> While this change IMO makes the control flow clearer
> for the human reader, it is especially important for
> GCC: It erroneously believes that it is possible to
> enter the SHORT2(UNUSED|LONG) cases without having
> entered the preceding block that initializes pic,
> frame_num, structure and j; it would emit -Wmaybe-uninitialized
> warnings for these variables if they were not pseudo-
> initialized with av_uninit(). This patch allows to remove
> the latter.
> 
> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
> ---
>  libavcodec/h264_refs.c | 22 ++++++++++------------
>  1 file changed, 10 insertions(+), 12 deletions(-)
> 
> diff --git a/libavcodec/h264_refs.c b/libavcodec/h264_refs.c
> index 9bc7b20988..c2ba4b9b9e 100644
> --- a/libavcodec/h264_refs.c
> +++ b/libavcodec/h264_refs.c
> @@ -646,8 +646,9 @@ int ff_h264_execute_ref_pic_marking(H264Context *h)
>              av_log(h->avctx, AV_LOG_DEBUG, "mmco:%d %d %d\n", h->mmco[i].opcode,
>                     h->mmco[i].short_pic_num, h->mmco[i].long_arg);
>  
> -        if (mmco[i].opcode == MMCO_SHORT2UNUSED ||
> -            mmco[i].opcode == MMCO_SHORT2LONG) {
> +        switch (mmco[i].opcode) {
> +        case MMCO_SHORT2UNUSED:
> +        case MMCO_SHORT2LONG:
>              frame_num = pic_num_extract(h, mmco[i].short_pic_num, &structure);
>              pic       = find_short(h, frame_num, &j);
>              if (!pic) {
> @@ -659,16 +660,12 @@ int ff_h264_execute_ref_pic_marking(H264Context *h)
>                  }
>                  continue;
>              }
> -        }
> -
> -        switch (mmco[i].opcode) {
> -        case MMCO_SHORT2UNUSED:
> -            if (h->avctx->debug & FF_DEBUG_MMCO)
> -                av_log(h->avctx, AV_LOG_DEBUG, "mmco: unref short %d count %d\n",
> -                       h->mmco[i].short_pic_num, h->short_ref_count);
> -            remove_short(h, frame_num, structure ^ PICT_FRAME);
> -            break;
> -        case MMCO_SHORT2LONG:
> +            if (mmco[i].opcode == MMCO_SHORT2UNUSED) {
> +                if (h->avctx->debug & FF_DEBUG_MMCO)
> +                    av_log(h->avctx, AV_LOG_DEBUG, "mmco: unref short %d count %d\n",
> +                           h->mmco[i].short_pic_num, h->short_ref_count);
> +                remove_short(h, frame_num, structure ^ PICT_FRAME);
> +            } else {
>                  if (h->long_ref[mmco[i].long_arg] != pic)
>                      remove_long(h, mmco[i].long_arg, 0);
>  
> @@ -678,6 +675,7 @@ int ff_h264_execute_ref_pic_marking(H264Context *h)
>                      h->long_ref[mmco[i].long_arg]->long_ref = 1;
>                      h->long_ref_count++;
>                  }
> +            }
>              break;
>          case MMCO_LONG2UNUSED:
>              j   = pic_num_extract(h, mmco[i].long_arg, &structure);

Will apply the rest of this patchset tomorrow unless there are objections.

- Andreas

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".