When decoding non-keyframes, the decoding process expects
there to be two reference frames, the last one and the golden
one. The existence of the golden one is checked and in case
it is there, it is presumed that the last one exists as well.
This assumption is wrong in case of memory allocation failure,
namely in case the call to ff_thread_ref_frame() that sets
the last frame fails.
Fix this by actually performing a shuffle without creating
new references. This can't fail and has the advantage of
fewer implicit allocations.
Signed-off-by: Andreas Rheinhardt <andreas.rheinha...@outlook.com>
---
libavcodec/vp3.c | 11 +++--------
1 file changed, 3 insertions(+), 8 deletions(-)
diff --git a/libavcodec/vp3.c b/libavcodec/vp3.c
index 5ce1ecfce7..f12bd497b9 100644
--- a/libavcodec/vp3.c
+++ b/libavcodec/vp3.c
@@ -2498,19 +2498,14 @@ static int update_frames(AVCodecContext *avctx)
Vp3DecodeContext *s = avctx->priv_data;
int ret = 0;
- /* shuffle frames (last = current) */
- ff_thread_release_ext_buffer(avctx, &s->last_frame);
- ret = ff_thread_ref_frame(&s->last_frame, &s->current_frame);
- if (ret < 0)
- goto fail;
-
if (s->keyframe) {
ff_thread_release_ext_buffer(avctx, &s->golden_frame);
ret = ff_thread_ref_frame(&s->golden_frame, &s->current_frame);
}
+ /* shuffle frames */
+ ff_thread_release_ext_buffer(avctx, &s->last_frame);
+ FFSWAP(ThreadFrame, s->last_frame, s->current_frame);
-fail:
- ff_thread_release_ext_buffer(avctx, &s->current_frame);
return ret;
}
--
2.34.1
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
