avcodec/apedec: Fix multiple integer overflows and undefined behaviorin filter_3800()
authorMichael Niedermayer <michael@niedermayer.cc>
Sun, 16 Jun 2019 09:39:15 +0000 (11:39 +0200)
committerMichael Niedermayer <michael@niedermayer.cc>
Wed, 31 Jul 2019 18:35:07 +0000 (20:35 +0200)
Fixes: left shift of negative value -4
Fixes: signed integer overflow: -15091694 * 167 cannot be represented in type 'int'
Fixes: signed integer overflow: 1898547155 + 453967445 cannot be represented in type 'int'
Fixes: 15258/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-5759095564402688
Fixes: signed integer overflow: 962196438 * 31 cannot be represented in type 'int'
Fixes: 15364/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-5718799845687296

Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
libavcodec/apedec.c

index 7cf99a0..27425b1 100644 (file)
@@ -859,9 +859,9 @@ static av_always_inline int filter_3800(APEPredictor *p,
         return predictionA;
     }
     d2 =  p->buf[delayA];
-    d1 = (p->buf[delayA] - p->buf[delayA - 1]) << 1;
-    d0 =  p->buf[delayA] + ((p->buf[delayA - 2] - p->buf[delayA - 1]) << 3);
-    d3 =  p->buf[delayB] * 2 - p->buf[delayB - 1];
+    d1 = (p->buf[delayA] - p->buf[delayA - 1]) * 2U;
+    d0 =  p->buf[delayA] + ((p->buf[delayA - 2] - p->buf[delayA - 1]) * 8U);
+    d3 =  p->buf[delayB] * 2U - p->buf[delayB - 1];
     d4 =  p->buf[delayB];
 
     predictionA = d0 * p->coeffsA[filter][0] +
@@ -881,7 +881,7 @@ static av_always_inline int filter_3800(APEPredictor *p,
     p->coeffsB[filter][1] -= (((d4 >> 30) & 2) - 1) * sign;
 
     p->filterB[filter] = p->lastA[filter] + (predictionB >> shift);
-    p->filterA[filter] = p->filterB[filter] + ((p->filterA[filter] * 31) >> 5);
+    p->filterA[filter] = p->filterB[filter] + ((int)(p->filterA[filter] * 31U) >> 5);
 
     return p->filterA[filter];
 }
@@ -902,7 +902,7 @@ static void long_filter_high_3800(int32_t *buffer, int order, int shift, int len
         dotprod = 0;
         sign = APESIGN(buffer[i]);
         for (j = 0; j < order; j++) {
-            dotprod += delay[j] * coeffs[j];
+            dotprod += delay[j] * (unsigned)coeffs[j];
             coeffs[j] += ((delay[j] >> 31) | 1) * sign;
         }
         buffer[i] -= dotprod >> shift;