[FFmpeg-devel] [PATCH 2/3] avcodec/h264idct_template: fix multiple runtime error: signed integer overflow

Michael Niedermayer michael at niedermayer.cc
Wed Mar 15 05:12:50 EET 2017


Fixes: 857/clusterfuzz-testcase-5319093760557056

Benchmark changes from 335->333 (so if its not a random fluctuation then it would be faster)

Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg
Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>
---
 libavcodec/h264idct_template.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/libavcodec/h264idct_template.c b/libavcodec/h264idct_template.c
index c62716090c..229a9ac36b 100644
--- a/libavcodec/h264idct_template.c
+++ b/libavcodec/h264idct_template.c
@@ -304,7 +304,7 @@ void FUNCC(ff_h264_chroma422_dc_dequant_idct)(int16_t *_block, int qmul){
 void FUNCC(ff_h264_chroma_dc_dequant_idct)(int16_t *_block, int qmul){
     const int stride= 16*2;
     const int xStride= 16;
-    int a,b,c,d,e;
+    SUINT a,b,c,d,e;
     dctcoef *block = (dctcoef*)_block;
 
     a= block[stride*0 + xStride*0];
@@ -317,8 +317,8 @@ void FUNCC(ff_h264_chroma_dc_dequant_idct)(int16_t *_block, int qmul){
     b= c-d;
     c= c+d;
 
-    block[stride*0 + xStride*0]= ((a+c)*qmul) >> 7;
-    block[stride*0 + xStride*1]= ((e+b)*qmul) >> 7;
-    block[stride*1 + xStride*0]= ((a-c)*qmul) >> 7;
-    block[stride*1 + xStride*1]= ((e-b)*qmul) >> 7;
+    block[stride*0 + xStride*0]= (int)((a+c)*qmul) >> 7;
+    block[stride*0 + xStride*1]= (int)((e+b)*qmul) >> 7;
+    block[stride*1 + xStride*0]= (int)((a-c)*qmul) >> 7;
+    block[stride*1 + xStride*1]= (int)((e-b)*qmul) >> 7;
 }
-- 
2.11.0



More information about the ffmpeg-devel mailing list