Commit b47a019471 for ffmpeg
commit b47a0194715034ce292f29510a5dfb45e5250bd5
Author: Michael Niedermayer <michael@niedermayer.cc>
Date: Mon Nov 16 06:56:38 2026 +0100
avcodec/proresdsp: avoid signed overflow in the ProRes RAW dequantization
Signed integer overflow Replicated through API with UBSAN
Fixes: signed integer overflow: -154665 * 16575 cannot be represented in type 'int'
Fixes: 572632128/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_PRORES_RAW_DEC_fuzzer-6177287729643520
Fixes: T1p6OXTgBhLz
Regression since: bc1cf21643f6cb63c0756268f61bf33ebd0b8712
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
diff --git a/libavcodec/proresdsp.c b/libavcodec/proresdsp.c
index 919a5d6797..9a5cba598c 100644
--- a/libavcodec/proresdsp.c
+++ b/libavcodec/proresdsp.c
@@ -88,7 +88,7 @@ static void prores_idct_12(int16_t *restrict block, const int16_t *restrict qmat
static void prores_idct_bayer_16(int32_t *restrict block, const int16_t *restrict qmat)
{
for (int i = 0; i < 64; i++)
- block[i] = av_clip_int16(block[i] * qmat[i]);
+ block[i] = av_clip_int16(block[i] * (unsigned)qmat[i]);
for (int i = 0; i < 8; i++)
idctRowCondDC_int32_16bit(block + i*8, 0);