[FFmpeg-devel] [PATCH 3/3] avcodec/x86/diracdsp: Fix incorrect src addressing in dequant_subband_32()

James Almer jamrial at gmail.com
Thu Jan 30 22:14:18 EET 2020


On 1/29/2020 6:55 PM, Michael Niedermayer wrote:
> Fixes: Segfault (not reproducable with asm, which made this hard to debug)
> Fixes: decoding errors
> Fixes: 19854/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DIRAC_fuzzer-5729372837511168
> 
> 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/x86/diracdsp.asm | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/libavcodec/x86/diracdsp.asm b/libavcodec/x86/diracdsp.asm
> index cc8a26fca5..a18bda113e 100644
> --- a/libavcodec/x86/diracdsp.asm
> +++ b/libavcodec/x86/diracdsp.asm
> @@ -294,8 +294,9 @@ cglobal dequant_subband_32, 7, 7, 4, src, dst, stride, qf, qs, tot_v, tot_h
>  
>      add    srcq, mmsize
>      add    dstq, mmsize
> -    sub    tot_hd, 4
> +    sub    tot_hq, 4

tot_h comes from stack, and on Windows x86_64 the higher 32 bits will be
garbage. This should remain as tot_hd, so the sub instruction will
implicitly clear said high bits (if the value could be negative, then
you'd have to sign extend it instead).

>      jg     .loop_h
> +    lea    srcq, [srcq + 4*tot_hq]
>  
>      add    r3, strideq
>      dec    tot_vd
> 



More information about the ffmpeg-devel mailing list