[FFmpeg-cvslog] r14758 - in trunk/libavcodec: dsputil.h mdct.c wma.c

Robert Swain robert.swain
Fri Aug 15 10:10:13 CEST 2008


2008/8/15 Mike Melanson <mike at multimedia.cx>:
> superdump wrote:
>> Author: superdump
>> Date: Thu Aug 14 16:40:16 2008
>> New Revision: 14758
>>
>> Log:
>> Add declarations for the sine tables used in wma.c (half window sizes: 128,
>> 256, 512, 1024 and 2048) to mdct.c. Make them accessible via dsputil.h. Make
>> wma.c use these shared tables.
>>
>>
>> Modified:
>>    trunk/libavcodec/dsputil.h
>>    trunk/libavcodec/mdct.c
>>    trunk/libavcodec/wma.c
>
> Try running the regression suite. This really makes the WMA test fail in
> a big way:
>
> ./build-cache/ffmpeg-14758 -y -flags +bitexact -dct fastint -idct simple
> -ab 128k -ac 2 -f s16le -i ffmpeg-build/tests/asynth1.sw -acodec wmav1
> a-wmav1.asf
>
>
> Check this out:
>
>
> *** glibc detected *** ./build-cache/ffmpeg-14758: free(): invalid
> pointer: 0x0877e6c0 ***
> ======= Backtrace: =========
> /lib/tls/i686/cmov/libc.so.6[0xb7dd9a85]
> /lib/tls/i686/cmov/libc.so.6(cfree+0x90)[0xb7ddd4f0]
> ./build-cache/ffmpeg-14758(ff_wma_end+0x5a)[0x83bc24a]
> ======= Memory map: ========
> 08048000-08518000 r-xp 00000000 08:01 96671
> /home/melanson/ffmpeg-regression/build-cache/ffmpeg-14758
> 08518000-08522000 rw-p 004cf000 08:01 96671
> /home/melanson/ffmpeg-regression/build-cache/ffmpeg-14758
> 08522000-0880a000 rw-p 08522000 00:00 0          [heap]
> b7b00000-b7b21000 rw-p b7b00000 00:00 0
> b7b21000-b7c00000 ---p b7b21000 00:00 0
> b7c05000-b7c0f000 r-xp 00000000 08:01 123318     /lib/libgcc_s.so.1
> b7c0f000-b7c10000 rw-p 0000a000 08:01 123318     /lib/libgcc_s.so.1
> b7c15000-b7d6e000 rw-p b7c15000 00:00 0
> b7d6e000-b7eb7000 r-xp 00000000 08:01 123595
> /lib/tls/i686/cmov/libc-2.7.so
> b7eb7000-b7eb8000 r--p 00149000 08:01 123595
> /lib/tls/i686/cmov/libc-2.7.so
> b7eb8000-b7eba000 rw-p 0014a000 08:01 123595
> /lib/tls/i686/cmov/libc-2.7.so
> b7eba000-b7ebd000 rw-p b7eba000 00:00 0
> b7ebd000-b7ebf000 r-xp 00000000 08:01 123603
> /lib/tls/i686/cmov/libdl-2.7.so
> b7ebf000-b7ec1000 rw-p 00001000 08:01 123603
> /lib/tls/i686/cmov/libdl-2.7.so
> b7ec1000-b7ec2000 rw-p b7ec1000 00:00 0
> b7ec2000-b7ee5000 r-xp 00000000 08:01 123609
> /lib/tls/i686/cmov/libm-2.7.so
> b7ee5000-b7ee7000 rw-p 00023000 08:01 123609
> /lib/tls/i686/cmov/libm-2.7.so
> b7ee7000-b7efb000 r-xp 00000000 08:01 18472      /usr/lib/libz.so.1.2.3.3
> b7efb000-b7efc000 rw-p 00013000 08:01 18472      /usr/lib/libz.so.1.2.3.3
> b7f01000-b7f03000 rw-p b7f01000 00:00 0
> b7f03000-b7f04000 r-xp b7f03000 00:00 0          [vdso]
> b7f04000-b7f1e000 r-xp 00000000 08:01 122925     /lib/ld-2.7.so
> b7f1e000-b7f20000 rw-p 00019000 08:01 122925     /lib/ld-2.7.so
> bf96e000-bf983000 rw-p bffeb000 00:00 0          [stack]
> Aborted
>
>
> Happens on x86_32 and PowerPC and is probably why x86_64 regressions are
> failing as well. The crash emanates from ff_wma_end().

It's probably because it's trying to free these tables... comitting fix.

Rob




More information about the ffmpeg-cvslog mailing list