[FFmpeg-devel] [PATCH 7/8] avcodec/vc1_block: Fix undefined behavior in ac prediction rescaling

[Michael Niedermayer]

The intermediates are required to fit in 12bit (8.1.3.9 Coefficient Scaling)
See SMPTE 421M-2006 and Amendment 1-2007

Fixes: signed integer overflow: -20691 * 262144 cannot be represented in type 'int'
Fixes: 18479/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_VC1_fuzzer-5128912371187712

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/vc1_block.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/libavcodec/vc1_block.c b/libavcodec/vc1_block.c
index 4c6dfaf6bf..da0eee8cc7 100644
--- a/libavcodec/vc1_block.c
+++ b/libavcodec/vc1_block.c
@@ -1036,10 +1036,10 @@ static int vc1_decode_intra_block(VC1Context *v, int16_t block[64], int n,
             if (q2 && q1 != q2) {
                 if (dc_pred_dir) { // left
                     for (k = 1; k < 8; k++)
-                        block[k << v->left_blk_sh] += (ac_val[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18;
+                        block[k << v->left_blk_sh] += (int)(ac_val[k] * q2 * (unsigned)ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18;
                 } else { //top
                     for (k = 1; k < 8; k++)
-                        block[k << v->top_blk_sh] += (ac_val[k + 8] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18;
+                        block[k << v->top_blk_sh] += (int)(ac_val[k + 8] * q2 * (unsigned)ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18;
                 }
             } else {
                 if (dc_pred_dir) { // left
-- 
2.23.0