avcodec/hcadec: Check or bound indexes
authorMichael Niedermayer <michael@niedermayer.cc>
Wed, 13 May 2020 17:54:42 +0000 (19:54 +0200)
committerMichael Niedermayer <michael@niedermayer.cc>
Mon, 8 Jun 2020 18:45:56 +0000 (20:45 +0200)
This causes indexes into scale_conversion_table to wrap around, alternatively they
could be clipped, the table be enlarged or we can error out. I have not found a document that specifies
what is the correct way to handle this

Fixes: out of array access
Fixes: 21727/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HCA_fuzzer-5752477891952640.fuzz
Fixes: 22438/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HCA_fuzzer-5640717790871552

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

index f25d6c3..5fa8731 100644 (file)
@@ -286,8 +286,8 @@ static void reconstruct_hfr(HCAContext *s, ChannelContext *ch,
         return;
 
     for (int i = 0, k = start_band, l = start_band - 1; i < hfr_group_count; i++){
-        for (int j = 0; j < bands_per_hfr_group && k < total_band_count; j++, k++, l--){
-            ch->imdct_in[k] = scale_conversion_table[ch->hfr_scale[i] - ch->scale_factors[l]] * ch->imdct_in[l];
+        for (int j = 0; j < bands_per_hfr_group && k < total_band_count && l >= 0; j++, k++, l--){
+            ch->imdct_in[k] = scale_conversion_table[ (ch->hfr_scale[i] - ch->scale_factors[l]) & 63 ] * ch->imdct_in[l];
         }
     }