[FFmpeg-devel] [PATCH 4/6] avcodec/mobiclip: Bound planar prediction value

[Michael Niedermayer]

Alternatively we could wrap around or error out. Wrap around would look
most messy as downstream computations would need to wrap as well.
I did not find a official specification which would list what to do
with huge values.

Fixes: signed integer overflow: 2 * 1073741952 cannot be represented in type 'int'
Fixes: 26765/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MOBICLIP_fuzzer-6594926936326144

Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>
---
 libavcodec/mobiclip.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/libavcodec/mobiclip.c b/libavcodec/mobiclip.c
index 9fa88e84a0..87a9a47da9 100644
--- a/libavcodec/mobiclip.c
+++ b/libavcodec/mobiclip.c
@@ -900,7 +900,7 @@ static int predict_intra(AVCodecContext *avctx, AVFrame *frame, int ax, int ay,
             uint8_t *left = frame->data[plane] + ay * frame->linesize[plane] + FFMAX(ax - 1, 0);
             int bottommost = frame->data[plane][(ay + size - 1) * frame->linesize[plane] + FFMAX(ax - 1, 0)];
             int rightmost = frame->data[plane][FFMAX(ay - 1, 0) * frame->linesize[plane] + ax + size - 1];
-            int avg = (bottommost + rightmost + 1) / 2 + 2 * get_se_golomb(gb);
+            int avg = (bottommost + rightmost + 1) / 2 + 2 * av_clip(get_se_golomb(gb), -(1<<16), 1<<16);
             int r6 = adjust(avg - bottommost, size);
             int r9 = adjust(avg - rightmost, size);
             int shift = adjust(size, size) == 8 ? 3 : 2;
-- 
2.17.1