Ticket #1692 (closed defect: fixed)
Using atempo filter causes segfault with AVX
| Reported by: | takis | Owned by: | |
|---|---|---|---|
| Priority: | important | Component: | avfilter |
| Version: | git-master | Keywords: | avx crash SIGSEGV atempo |
| Cc: | Blocked By: | ||
| Blocking: | Reproduced by developer: | yes | |
| Analyzed by developer: | yes |
Description
Summary of the bug:
Using the atempo filter causes a segmentation fault.
How to reproduce:
./audiogen original.wav
% ffmpeg -i original.wav -af atempo=0.8 test.wav
ffmpeg version N-43928-g60924df Copyright (c) 2000-2012 the FFmpeg developers
built on Aug 27 2012 14:19:01 with gcc 4.6 (Ubuntu/Linaro 4.6.3-1ubuntu5)
configuration: --enable-avfilter --enable-frei0r --enable-gpl --enable-libfaac --enable-libmp3lame --enable-libschroedinger --enable-libspeex --enable-libtheora --enable-libvorbis --enable-libvpx --enable-libx264 --enable-libxvid --enable-nonfree --enable-pic --enable-pthreads --enable-shared --enable-version3 --enable-x11grab
libavutil 51. 70.100 / 51. 70.100
libavcodec 54. 54.100 / 54. 54.100
libavformat 54. 25.104 / 54. 25.104
libavdevice 54. 2.100 / 54. 2.100
libavfilter 3. 13.101 / 3. 13.101
libswscale 2. 1.101 / 2. 1.101
libswresample 0. 15.100 / 0. 15.100
libpostproc 52. 0.100 / 52. 0.100
[wav @ 0x2172240] max_analyze_duration 5000000 reached at 5015510
Guessed Channel Layout for Input Stream #0.0 : mono
Input #0, wav, from 'original.wav':
Duration: 00:00:06.00, bitrate: 705 kb/s
Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 44100 Hz, mono, s16, 705 kb/s
File 'test.wav' already exists. Overwrite ? [y/N] y
Output #0, wav, to 'test.wav':
Metadata:
encoder : Lavf54.25.104
Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 44100 Hz, mono, s16, 705 kb/s
Stream mapping:
Stream #0:0 -> #0:0 (pcm_s16le -> pcm_s16le)
Press [q] to stop, [?] for help
Segmentation fault (core dumped)
Change History
comment:2 Changed 9 months ago by takis
(gdb) r -y -i audiogen.wav -af atempo=0.8 asdf.wav
Starting program: /usr/local/src/ff/ffmpeg.org/build-gcc/ffmpeg_g -y -i audiogen.wav -af atempo=0.8 asdf.wav
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
ffmpeg version N-43928-g60924df Copyright (c) 2000-2012 the FFmpeg developers
built on Aug 27 2012 14:19:01 with gcc 4.6 (Ubuntu/Linaro? 4.6.3-1ubuntu5)
configuration: --enable-avfilter --enable-frei0r --enable-gpl --enable-libfaac --enable-libmp3lame --enable-libschroedinger --enable-libspeex --enable-libtheora --enable-libvorbis --enable-libvpx --enable-libx264 --enable-libxvid --enable-nonfree --enable-pic --enable-pthreads --enable-shared --enable-version3 --enable-x11grab
libavutil 51. 70.100 / 51. 70.100
libavcodec 54. 54.100 / 54. 54.100
libavformat 54. 25.104 / 54. 25.104
libavdevice 54. 2.100 / 54. 2.100
libavfilter 3. 13.101 / 3. 13.101
libswscale 2. 1.101 / 2. 1.101
libswresample 0. 15.100 / 0. 15.100
libpostproc 52. 0.100 / 52. 0.100
[wav @ 0x632240] max_analyze_duration 5000000 reached at 5015510
Guessed Channel Layout for Input Stream #0.0 : stereo
Input #0, wav, from 'audiogen.wav':
Duration: 00:00:06.00, bitrate: 1411 kb/s
Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 44100 Hz, stereo, s16, 1411 kb/s
Output #0, wav, to 'asdf.wav':
Metadata:
encoder : Lavf54.25.104
Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 44100 Hz, stereo, s16, 1411 kb/s
Stream mapping:
Press [q] to stop, ? for help
Program received signal SIGSEGV, Segmentation fault.
0x00007ffff6c691d0 in ?? () from /usr/local/lib/libavcodec.so.54
(gdb) bt
#0 0x00007ffff6c691d0 in ?? () from /usr/local/lib/libavcodec.so.54
#1 0x00007ffff6c69315 in ?? () from /usr/local/lib/libavcodec.so.54
#2 0x00007ffff6c69d15 in ?? () from /usr/local/lib/libavcodec.so.54
#3 0x00007ffff6c69d55 in ?? () from /usr/local/lib/libavcodec.so.54
#4 0x00007ffff6c69d95 in ?? () from /usr/local/lib/libavcodec.so.54
#5 0x00007ffff6c69dd5 in ?? () from /usr/local/lib/libavcodec.so.54
#6 0x00007ffff6c69e15 in ?? () from /usr/local/lib/libavcodec.so.54
#7 0x00007ffff6c6a185 in ?? () from /usr/local/lib/libavcodec.so.54
#8 0x00007ffff6c699d0 in ?? () from /usr/local/lib/libavcodec.so.54
#9 0x00007ffff6b20fdb in ?? () from /usr/local/lib/libavcodec.so.54
#10 0x00007ffff793d673 in ?? () from /usr/local/lib/libavfilter.so.3
#11 0x00007ffff794149b in ?? () from /usr/local/lib/libavfilter.so.3
#12 0x00007ffff7946f4d in ?? () from /usr/local/lib/libavfilter.so.3
#13 0x00007ffff7947176 in av_buffersrc_add_ref () from /usr/local/lib/libavfilter.so.3
#14 0x00007ffff79472e8 in av_buffersrc_add_frame () from /usr/local/lib/libavfilter.so.3
#15 0x0000000000417aa3 in decode_audio (got_output=0x7fffffffd70c, pkt=0x7fffffffd600, ist=0x71b3c0)
at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:1524
#16 output_packet (ist=<optimized out>, pkt=<optimized out>) at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:1753
#17 0x00000000004188d9 in process_input (file_index=<optimized out>) at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:2818
#18 0x0000000000407512 in transcode_step () at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:2914
#19 transcode () at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:2966
#20 main (argc=<optimized out>, argv=<optimized out>) at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:3146
(gdb)
comment:3 Changed 9 months ago by takis
Valgrind shows no errors:
/usr/local/src/ff/ffmpeg.org/tests$ valgrind ../build-gcc/ffmpeg_g -y -i audiogen.wav -af atempo=0.8 asdfasd.wav
==27154== Memcheck, a memory error detector
==27154== Copyright (C) 2002-2011, and GNU GPL'd, by Julian Seward et al.
==27154== Using Valgrind-3.7.0 and LibVEX; rerun with -h for copyright info
==27154== Command: ../build-gcc/ffmpeg_g -y -i audiogen.wav -af atempo=0.8 asdfasd.wav
==27154==
ffmpeg version N-43928-g60924df Copyright (c) 2000-2012 the FFmpeg developers
built on Aug 27 2012 14:19:01 with gcc 4.6 (Ubuntu/Linaro? 4.6.3-1ubuntu5)
configuration: --enable-avfilter --enable-frei0r --enable-gpl --enable-libfaac --enable-libmp3lame --enable-libschroedinger --enable-libspeex --enable-libtheora --enable-libvorbis --enable-libvpx --enable-libx264 --enable-libxvid --enable-nonfree --enable-pic --enable-pthreads --enable-shared --enable-version3 --enable-x11grab
libavutil 51. 70.100 / 51. 70.100
libavcodec 54. 54.100 / 54. 54.100
libavformat 54. 25.104 / 54. 25.104
libavdevice 54. 2.100 / 54. 2.100
libavfilter 3. 13.101 / 3. 13.101
libswscale 2. 1.101 / 2. 1.101
libswresample 0. 15.100 / 0. 15.100
libpostproc 52. 0.100 / 52. 0.100
[wav @ 0xd6c1380] max_analyze_duration 5000000 reached at 5015510
Guessed Channel Layout for Input Stream #0.0 : stereo
Input #0, wav, from 'audiogen.wav':
Duration: 00:00:06.00, bitrate: 1411 kb/s
Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 44100 Hz, stereo, s16, 1411 kb/s
Output #0, wav, to 'asdfasd.wav':
Metadata:
encoder : Lavf54.25.104
Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 44100 Hz, stereo, s16, 1411 kb/s
Stream mapping:
Press [q] to stop, ? for help
size= 1291kB time=00:00:07.49 bitrate=1411.2kbits/s
video:0kB audio:1291kB subtitle:0 global headers:0kB muxing overhead 0.003480%
==27154==
==27154== HEAP SUMMARY:
==27154== in use at exit: 0 bytes in 0 blocks
==27154== total heap usage: 4,677 allocs, 4,677 frees, 5,608,663 bytes allocated
==27154==
==27154== All heap blocks were freed -- no leaks are possible
==27154==
==27154== For counts of detected and suppressed errors, rerun with: -v
==27154== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 2 from 2)
comment:4 Changed 9 months ago by Cigaes
The gdb backtrace you show was produced using dynamically-linked installed libraries: they lack the debugging symbols, and therefore the backtrace is unusable.
Can you find the exact revision of the libraries installed in /usr/local/lib? Running a recent ffmpeg binary with older libraries should not cause a segfault, but developers are not always as careful as when using the same version.
comment:5 Changed 9 months ago by takis
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
ffmpeg version N-43943-g4f92d31 Copyright (c) 2000-2012 the FFmpeg developers
built on Aug 28 2012 13:23:13 with gcc 4.6 (Ubuntu/Linaro? 4.6.3-1ubuntu5)
configuration: --enable-avfilter --enable-frei0r --enable-gpl --enable-libfaac --enable-libmp3lame --enable-libschroedinger --enable-libspeex --enable-libtheora --enable-libvorbis --enable-libvpx --enable-libx264 --enable-libxvid --enable-nonfree --enable-pic --enable-pthreads --disable-shared --enable-static --enable-version3 --enable-x11grab --disable-optimizations
libavutil 51. 70.100 / 51. 70.100
libavcodec 54. 54.100 / 54. 54.100
libavformat 54. 25.104 / 54. 25.104
libavdevice 54. 2.100 / 54. 2.100
libavfilter 3. 13.101 / 3. 13.101
libswscale 2. 1.101 / 2. 1.101
libswresample 0. 15.100 / 0. 15.100
libpostproc 52. 0.100 / 52. 0.100
[wav @ 0x1760240] max_analyze_duration 5000000 reached at 5015510
Guessed Channel Layout for Input Stream #0.0 : stereo
Input #0, wav, from '../tests/audiogen.wav':
Duration: 00:00:06.00, bitrate: 1411 kb/s
Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 44100 Hz, stereo, s16, 1411 kb/s
Output #0, wav, to 'asdf.wav':
Metadata:
encoder : Lavf54.25.104
Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 44100 Hz, stereo, s16, 1411 kb/s
Stream mapping:
Press [q] to stop, ? for help
Program received signal SIGSEGV, Segmentation fault.
fft16_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:321
321 mova m2, Z(2)
comment:6 Changed 9 months ago by takis
(gdb) bt
#0 fft16_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:321
#1 0x0000000000b59e75 in fft32_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:358
#2 0x0000000000b5a875 in fft64_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:799
#3 0x0000000000b5a8b5 in fft128_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:799
#4 0x0000000000b5a8f5 in fft256_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:799
#5 0x0000000000b5a935 in fft512_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:799
#6 0x0000000000b5a975 in fft1024_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:799
#7 0x0000000000b5ace5 in fft2048_interleave_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:800
#8 0x0000000000b5a530 in ff_fft_calc_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:575
#9 0x00000000009b3436 in ff_rdft_calc_c (s=0x1814200, data=0x18678d0)
at /usr/local/src/ff/ffmpeg.org/libavcodec/rdft.c:69
#10 0x0000000000605ddd in av_rdft_calc (s=0x1814200, data=0x18678d0)
at /usr/local/src/ff/ffmpeg.org/libavcodec/avfft.c:106
#11 0x000000000049ca55 in yae_apply (atempo=0x1766fa0, src_ref=0x7fffffffd3f0, src_end=0x1886ce0 "",
dst_ref=0x1767078, dst_end=0x1888320 "") at /usr/local/src/ff/ffmpeg.org/libavfilter/af_atempo.c:815
#12 0x000000000049d286 in filter_samples (inlink=0x1814060, src_buffer=0x1886e20)
at /usr/local/src/ff/ffmpeg.org/libavfilter/af_atempo.c:1066
#13 0x0000000000453e3d in ff_filter_samples_framed (link=0x1814060, samplesref=0x1886e20)
at /usr/local/src/ff/ffmpeg.org/libavfilter/audio.c:214
#14 0x0000000000453efe in ff_filter_samples (link=0x1814060, samplesref=0x1886e20)
at /usr/local/src/ff/ffmpeg.org/libavfilter/audio.c:229
#15 0x000000000045d35b in request_frame (link=0x1814060) at /usr/local/src/ff/ffmpeg.org/libavfilter/buffersrc.c:384
#16 0x000000000045c9c1 in av_buffersrc_add_ref (s=0x17e9800, buf=0x1886e20, flags=4)
at /usr/local/src/ff/ffmpeg.org/libavfilter/buffersrc.c:152
#17 0x000000000045c67c in av_buffersrc_add_frame (buffer_src=0x17e9800, frame=0x1884a40, flags=4)
at /usr/local/src/ff/ffmpeg.org/libavfilter/buffersrc.c:91
#18 0x000000000043e9f3 in decode_audio (ist=0x18493c0, pkt=0x7fffffffd780, got_output=0x7fffffffd838)
at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:1524
#19 0x000000000043fcb0 in output_packet (ist=0x18493c0, pkt=0x7fffffffd8e0)
---Type <return> to continue, or q <return> to quit---
at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:1753
#20 0x0000000000444a7a in process_input (file_index=0) at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:2818
#21 0x0000000000444e0c in transcode_step () at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:2914
#22 0x0000000000444f26 in transcode () at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:2966
#23 0x000000000044551c in main (argc=7, argv=0x7fffffffdfa8) at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:3146
comment:7 Changed 9 months ago by takis
(gdb) disass $pc-32,$pc+32 Dump of assembler code from 0xb59d10 to 0xb59d50: 0x0000000000b59d10 <fft8_avx+112>: vblendps $0x55,%ymm0,%ymm4,%ymm0 0x0000000000b59d16 <fft8_avx+118>: vsubps %ymm0,%ymm3,%ymm1 0x0000000000b59d1a <fft8_avx+122>: vaddps %ymm0,%ymm3,%ymm0 0x0000000000b59d1e <fft8_avx+126>: vmovaps %ymm0,(%rdi) 0x0000000000b59d22 <fft8_avx+130>: vmovaps %ymm1,0x20(%rdi) 0x0000000000b59d27 <fft8_avx+135>: retq 0x0000000000b59d28 <fft8_avx+136>: nopl 0x0(%rax,%rax,1) => 0x0000000000b59d30 <fft16_avx+0>: vmovaps 0x40(%rdi),%ymm2 0x0000000000b59d35 <fft16_avx+5>: vmovaps 0x60(%rdi),%ymm3 0x0000000000b59d3a <fft16_avx+10>: vsubps %ymm3,%ymm2,%ymm7 0x0000000000b59d3e <fft16_avx+14>: vaddps %ymm3,%ymm2,%ymm2 0x0000000000b59d42 <fft16_avx+18>: vxorps 0x347196(%rip),%ymm7,%ymm7 # 0xea0ee0 <ps_p1p1m1p1> 0x0000000000b59d4a <fft16_avx+26>: vshufps $0xbe,%ymm7,%ymm2,%ymm3 0x0000000000b59d4f <fft16_avx+31>: vshufps $0x44,%ymm7,%ymm2,%ymm2 End of assembler dump.
comment:8 Changed 9 months ago by takis
(gdb) info all-registers
rax 0x18678d0 25589968
rbx 0x1000 4096
rcx 0xb59ca0 11902112
rdx 0xb5ace0 11906272
rsi 0xb 11
rdi 0x18678d0 25589968
rbp 0x7fffffffd370 0x7fffffffd370
rsp 0x7fffffffd2c8 0x7fffffffd2c8
r8 0x1764b20 24529696
r9 0x1873920 25639200
r10 0x0 0
r11 0x7ffff426db1d 140737289575197
r12 0x42cb00 4377344
r13 0x7fffffffdfa0 140737488347040
r14 0x0 0
r15 0x0 0
rip 0xb59d30 0xb59d30 <fft16_avx>
eflags 0x10202 [ IF RF ]
cs 0x33 51
ss 0x2b 43
ds 0x0 0
es 0x0 0
fs 0x0 0
gs 0x0 0
st0 0 (raw 0x00000000000000000000)
st1 0 (raw 0x00000000000000000000)
st2 0 (raw 0x00000000000000000000)
st3 0 (raw 0x00000000000000000000)
st4 0 (raw 0x00000000000000000000)
---Type <return> to continue, or q <return> to quit---
st5 0 (raw 0x00000000000000000000)
st6 0 (raw 0x00000000000000000000)
st7 0 (raw 0x00000000000000000000)
fctrl 0x37f 895
fstat 0x0 0
ftag 0xffff 65535
fiseg 0x0 0
fioff 0x0 0
foseg 0x0 0
fooff 0x0 0
fop 0x0 0
mxcsr 0x1fa0 [ PE IM DM ZM OM UM PM ]
ymm0 {v8_float = {0x0, 0x0, 0xffffdba0, 0xffffde04, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x0,
0x8000000000000000, 0x0, 0x0}, v32_int8 = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x80, 0x11, 0xc6, 0x0,
0xf0, 0x7, 0xc6, 0x0 <repeats 16 times>}, v16_int16 = {0x0, 0x0, 0x0, 0x0, 0x8000, 0xc611, 0xf000, 0xc607, 0x0,
0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v8_int32 = {0x0, 0x0, 0xc6118000, 0xc607f000, 0x0, 0x0, 0x0, 0x0},
v4_int64 = {0x0, 0xc607f000c6118000, 0x0, 0x0}, v2_int128 = {0xc607f000c61180000000000000000000,
0x00000000000000000000000000000000}}
ymm1 {v8_float = {0x0, 0x0, 0x127, 0x6af, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x0, 0x8000000000000000, 0x0,
0x0}, v32_int8 = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x80, 0x93, 0x43, 0x0, 0xe0, 0xd5, 0x44,
0x0 <repeats 16 times>}, v16_int16 = {0x0, 0x0, 0x0, 0x0, 0x8000, 0x4393, 0xe000, 0x44d5, 0x0, 0x0, 0x0, 0x0,
0x0, 0x0, 0x0, 0x0}, v8_int32 = {0x0, 0x0, 0x43938000, 0x44d5e000, 0x0, 0x0, 0x0, 0x0}, v4_int64 = {0x0,
0x44d5e00043938000, 0x0, 0x0}, v2_int128 = {0x44d5e000439380000000000000000000,
0x00000000000000000000000000000000}}
ymm2 {v8_float = {0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x0, 0x0, 0x0, 0x0}, v32_int8 = {
0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0xe0, 0x3f, 0x0 <repeats 24 times>}, v16_int16 = {0x0, 0x0, 0x0, 0x3fe0,
0x0 <repeats 12 times>}, v8_int32 = {0x0, 0x3fe00000, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_int64 = {
0x3fe0000000000000, 0x0, 0x0, 0x0}, v2_int128 = {0x00000000000000003fe0000000000000,
0x00000000000000000000000000000000}}
---Type <return> to continue, or q <return> to quit---
ymm3 {v8_float = {0x0, 0x0, 0x0, 0xa0000000, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x8000000000000000,
0x8000000000000000, 0x0, 0x0}, v32_int8 = {0xb, 0xe0, 0xb, 0xe0, 0x2e, 0xdd, 0x2e, 0xdd, 0x4, 0xdb, 0x4, 0xdb,
0x9b, 0xd9, 0x9b, 0xd9, 0x0 <repeats 16 times>}, v16_int16 = {0xe00b, 0xe00b, 0xdd2e, 0xdd2e, 0xdb04, 0xdb04,
0xd99b, 0xd99b, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v8_int32 = {0xe00be00b, 0xdd2edd2e, 0xdb04db04,
0xd99bd99b, 0x0, 0x0, 0x0, 0x0}, v4_int64 = {0xdd2edd2ee00be00b, 0xd99bd99bdb04db04, 0x0, 0x0}, v2_int128 = {
0xd99bd99bdb04db04dd2edd2ee00be00b, 0x00000000000000000000000000000000}}
ymm4 {v8_float = {0x38000000, 0xf0000000, 0x40000000, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_double = {
0x8000000000000000, 0x8000000000000000, 0x0, 0x0}, v32_int8 = {0xf9, 0xd8, 0xf9, 0xd8, 0x21, 0xd9, 0x21, 0xd9,
0x13, 0xda, 0x13, 0xda, 0xca, 0xdb, 0xca, 0xdb, 0x0 <repeats 16 times>}, v16_int16 = {0xd8f9, 0xd8f9, 0xd921,
0xd921, 0xda13, 0xda13, 0xdbca, 0xdbca, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v8_int32 = {0xd8f9d8f9,
0xd921d921, 0xda13da13, 0xdbcadbca, 0x0, 0x0, 0x0, 0x0}, v4_int64 = {0xd921d921d8f9d8f9, 0xdbcadbcada13da13, 0x0,
0x0}, v2_int128 = {0xdbcadbcada13da13d921d921d8f9d8f9, 0x00000000000000000000000000000000}}
ymm5 {v8_float = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x0, 0x0, 0x0, 0x0}, v32_int8 = {
0x10, 0x27, 0x10, 0x27, 0xaa, 0x26, 0xaa, 0x26, 0x7c, 0x25, 0x7c, 0x25, 0x8c, 0x23, 0x8c, 0x23,
0x0 <repeats 16 times>}, v16_int16 = {0x2710, 0x2710, 0x26aa, 0x26aa, 0x257c, 0x257c, 0x238c, 0x238c, 0x0, 0x0,
0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v8_int32 = {0x27102710, 0x26aa26aa, 0x257c257c, 0x238c238c, 0x0, 0x0, 0x0, 0x0},
v4_int64 = {0x26aa26aa27102710, 0x238c238c257c257c, 0x0, 0x0}, v2_int128 = {0x238c238c257c257c26aa26aa27102710,
0x00000000000000000000000000000000}}
ymm6 {v8_float = {0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x0, 0x0, 0x0, 0x0}, v32_int8 = {
0x2f, 0x87, 0x29, 0xe8, 0x10, 0x11, 0x81, 0x3f, 0x0 <repeats 24 times>}, v16_int16 = {0x872f, 0xe829, 0x1110,
0x3f81, 0x0 <repeats 12 times>}, v8_int32 = {0xe829872f, 0x3f811110, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_int64 = {
0x3f811110e829872f, 0x0, 0x0, 0x0}, v2_int128 = {0x00000000000000003f811110e829872f,
0x00000000000000000000000000000000}}
ymm7 {v8_float = {0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x0, 0x0, 0x0, 0x0}, v32_int8 = {
0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0xe0, 0x3f, 0x0 <repeats 24 times>}, v16_int16 = {0x0, 0x0, 0x0, 0x3fe0,
0x0 <repeats 12 times>}, v8_int32 = {0x0, 0x3fe00000, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_int64 = {
0x3fe0000000000000, 0x0, 0x0, 0x0}, v2_int128 = {0x00000000000000003fe0000000000000,
0x00000000000000000000000000000000}}
ymm8 {v8_float = {0x53500000, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x0, 0x0, 0x0, 0x0},
---Type <return> to continue, or q <return> to quit---
v32_int8 = {0x35, 0x55, 0x55, 0x55, 0x55, 0x55, 0xa5, 0x3f, 0x0 <repeats 24 times>}, v16_int16 = {0x5535, 0x5555,
0x5555, 0x3fa5, 0x0 <repeats 12 times>}, v8_int32 = {0x55555535, 0x3fa55555, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0},
v4_int64 = {0x3fa5555555555535, 0x0, 0x0, 0x0}, v2_int128 = {0x00000000000000003fa5555555555535,
0x00000000000000000000000000000000}}
ymm9 {v8_float = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x0, 0x0, 0x0, 0x0}, v32_int8 = {
0x39, 0xe2, 0xd9, 0xed, 0x6b, 0xc1, 0x56, 0x3f, 0x0 <repeats 24 times>}, v16_int16 = {0xe239, 0xedd9, 0xc16b,
0x3f56, 0x0 <repeats 12 times>}, v8_int32 = {0xedd9e239, 0x3f56c16b, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_int64 = {
0x3f56c16bedd9e239, 0x0, 0x0, 0x0}, v2_int128 = {0x00000000000000003f56c16bedd9e239,
0x00000000000000000000000000000000}}
ymm10 {v8_float = {0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x1, 0x0, 0x0, 0x0}, v32_int8 = {
0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0xf0, 0x3f, 0x0 <repeats 24 times>}, v16_int16 = {0x0, 0x0, 0x0, 0x3ff0,
0x0 <repeats 12 times>}, v8_int32 = {0x0, 0x3ff00000, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_int64 = {
0x3ff0000000000000, 0x0, 0x0, 0x0}, v2_int128 = {0x00000000000000003ff0000000000000,
0x00000000000000000000000000000000}}
ymm11 {v8_float = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x0, 0x0, 0x0, 0x0}, v32_int8 = {
0x7, 0x5c, 0x14, 0x33, 0x26, 0xa6, 0xb1, 0x3c, 0x0 <repeats 24 times>}, v16_int16 = {0x5c07, 0x3314, 0xa626,
0x3cb1, 0x0 <repeats 12 times>}, v8_int32 = {0x33145c07, 0x3cb1a626, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_int64 = {
0x3cb1a62633145c07, 0x0, 0x0, 0x0}, v2_int128 = {0x00000000000000003cb1a62633145c07,
0x00000000000000000000000000000000}}
ymm12 {v8_float = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x0, 0x0, 0x0, 0x0}, v32_int8 = {
0x0 <repeats 32 times>}, v16_int16 = {0x0 <repeats 16 times>}, v8_int32 = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0,
0x0}, v4_int64 = {0x0, 0x0, 0x0, 0x0}, v2_int128 = {0x00000000000000000000000000000000,
0x00000000000000000000000000000000}}
ymm13 {v8_float = {0x0, 0x64, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x300000000000, 0x0, 0x0, 0x0},
v32_int8 = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0xc8, 0x42, 0x0 <repeats 24 times>}, v16_int16 = {0x0, 0x0, 0x0, 0x42c8,
0x0 <repeats 12 times>}, v8_int32 = {0x0, 0x42c80000, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_int64 = {
0x42c8000000000000, 0x0, 0x0, 0x0}, v2_int128 = {0x000000000000000042c8000000000000,
0x00000000000000000000000000000000}}
ymm14 {v8_float = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x0, 0x0, 0x0, 0x0}, v32_int8 = {
---Type <return> to continue, or q <return> to quit---
0x9d, 0x9f, 0x7a, 0xe2, 0x66, 0xf4, 0x63, 0xb9, 0x0 <repeats 24 times>}, v16_int16 = {0x9f9d, 0xe27a, 0xf466,
0xb963, 0x0 <repeats 12 times>}, v8_int32 = {0xe27a9f9d, 0xb963f466, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_int64 = {
0xb963f466e27a9f9d, 0x0, 0x0, 0x0}, v2_int128 = {0x0000000000000000b963f466e27a9f9d,
0x00000000000000000000000000000000}}
ymm15 {v8_float = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, v4_double = {0x0, 0x0, 0x0, 0x0}, v32_int8 = {
0x0 <repeats 32 times>}, v16_int16 = {0x0 <repeats 16 times>}, v8_int32 = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0,
0x0}, v4_int64 = {0x0, 0x0, 0x0, 0x0}, v2_int128 = {0x00000000000000000000000000000000,
0x00000000000000000000000000000000}}
comment:9 Changed 9 months ago by takis
Sure, I used a static build now, as I find that easier to debug. Furthermore, it is not related to the specific sample, using any audiofile gives the same result.
comment:10 Changed 9 months ago by takis
I wanted to git-bisect it, but is seems it has been happing from the first revision of libavfilter/af_atempo.c.
pissaris@takis-Latitude-E6520:/usr/local/src/ff/ffmpeg.org/build-static$ ./ffmpeg_g -t 10 -i /opt/android-sdk-linux_x86/samples/android-14/ApiDemos/res/raw/test_cbr.mp3 -af atempo=0.8 asdf.wav
ffmpeg version N-41650-ga1aac8d Copyright (c) 2000-2012 the FFmpeg developers
built on Aug 28 2012 13:37:26 with gcc 4.6.3
configuration: --enable-avfilter --enable-frei0r --enable-gpl --enable-libfaac --enable-libmp3lame --enable-libschroedinger --enable-libspeex --enable-libtheora --enable-libvorbis --enable-libvpx --enable-libx264 --enable-libxvid --enable-nonfree --enable-pic --enable-pthreads --disable-shared --enable-static --enable-version3 --enable-x11grab --disable-optimizations
libavutil 51. 58.100 / 51. 58.100
libavcodec 54. 25.100 / 54. 25.100
libavformat 54. 6.101 / 54. 6.101
libavdevice 54. 0.100 / 54. 0.100
libavfilter 2. 81.100 / 2. 81.100
libswscale 2. 1.100 / 2. 1.100
libswresample 0. 15.100 / 0. 15.100
libpostproc 52. 0.100 / 52. 0.100
[mp3 @ 0x185a120] max_analyze_duration 5000000 reached at 5015510
[mp3 @ 0x185a120] Estimating duration from bitrate, this may be inaccurate
Input #0, mp3, from '/opt/android-sdk-linux_x86/samples/android-14/ApiDemos/res/raw/test_cbr.mp3':
Duration: 00:00:52.86, start: 0.000000, bitrate: 55 kb/s
Stream #0:0: Audio: mp3, 44100 Hz, mono, s16, 56 kb/s
File 'asdf.wav' already exists. Overwrite ? [y/N] y
Output #0, wav, to 'asdf.wav':
Metadata:
encoder : Lavf54.6.101
Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 44100 Hz, mono, s16, 705 kb/s
Stream mapping:
Stream #0:0 -> #0:0 (mp3 -> pcm_s16le)
Press [q] to stop, [?] for help
Segmentation fault (core dumped)
comment:11 Changed 9 months ago by takis
(gdb) bt
#0 fft16_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:315
#1 0x0000000000aa76d5 in fft32_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:352
#2 0x0000000000aa8515 in fft64_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:649
#3 0x0000000000aa8555 in fft128_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:649
#4 0x0000000000aa8595 in fft256_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:649
#5 0x0000000000aa85d5 in fft512_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:649
#6 0x0000000000aa8615 in fft1024_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:649
#7 0x0000000000aa89a5 in fft2048_interleave_avx () at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:650
#8 0x0000000000aa8ba8 in ff_fft_dispatch_interleave_avx ()
at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_mmx.asm:650
#9 0x0000000000aaa82d in ff_fft_calc_avx (s=0x177f640, z=0x17b9c30)
at /usr/local/src/ff/ffmpeg.org/libavcodec/x86/fft_sse.c:37
#10 0x000000000094854c in ff_rdft_calc_c (s=0x177f620, data=0x17b9c30)
at /usr/local/src/ff/ffmpeg.org/libavcodec/rdft.c:69
#11 0x000000000065778e in av_rdft_calc (s=0x177f620, data=0x17b9c30)
at /usr/local/src/ff/ffmpeg.org/libavcodec/avfft.c:106
#12 0x000000000049addd in yae_apply (atempo=0x17af760, src_ref=0x7fffffffd590, src_end=0x17d9f60 "",
dst_ref=0x17af838, dst_end=0x17dad40 "") at /usr/local/src/ff/ffmpeg.org/libavfilter/af_atempo.c:815
#13 0x000000000049b61e in filter_samples (inlink=0x177f020, src_buffer=0x17863e0)
at /usr/local/src/ff/ffmpeg.org/libavfilter/af_atempo.c:1066
#14 0x0000000000451ebc in ff_filter_samples (link=0x177f020, samplesref=0x17863e0)
at /usr/local/src/ff/ffmpeg.org/libavfilter/audio.c:214
#15 0x00000000004595c0 in request_frame (link=0x177f020) at /usr/local/src/ff/ffmpeg.org/libavfilter/buffersrc.c:445
#16 0x0000000000453902 in ff_request_frame (link=0x177f020) at /usr/local/src/ff/ffmpeg.org/libavfilter/avfilter.c:325
#17 0x00000000004544ae in avfilter_request_frame (link=0x177f020)
at /usr/local/src/ff/ffmpeg.org/libavfilter/avfilter.c:594
#18 0x000000000049b6db in request_frame (outlink=0x177f400)
at /usr/local/src/ff/ffmpeg.org/libavfilter/af_atempo.c:1086
#19 0x0000000000453902 in ff_request_frame (link=0x177f400) at /usr/local/src/ff/ffmpeg.org/libavfilter/avfilter.c:325
---Type <return> to continue, or q <return> to quit---
#20 0x000000000045392d in ff_request_frame (link=0x177f6e0) at /usr/local/src/ff/ffmpeg.org/libavfilter/avfilter.c:327
#21 0x000000000045812d in av_buffersink_read (ctx=0x177ede0, buf=0x7fffffffd770)
at /usr/local/src/ff/ffmpeg.org/libavfilter/buffersink.c:109
#22 0x0000000000430614 in poll_filters () at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:1927
#23 0x0000000000437ef2 in transcode () at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:3608
#24 0x0000000000440e67 in main (argc=8, argv=0x7fffffffdf58) at /usr/local/src/ff/ffmpeg.org/ffmpeg.c:5917
comment:12 Changed 9 months ago by Cigaes
Thanks for the backtraces. It looks related to the AVX optimizations. I do not have the hardware to test. Does it make it work to build with --disable-avx?
comment:13 Changed 9 months ago by takis
From /proc/cpuinfo:
processor : 3 vendor_id : GenuineIntel cpu family : 6 model : 42 model name : Intel(R) Core(TM) i5-2520M CPU @ 2.50GHz stepping : 7 microcode : 0x15 cpu MHz : 2501.000 cache size : 3072 KB physical id : 0 siblings : 4 core id : 1 cpu cores : 2 apicid : 3 initial apicid : 3 fpu : yes fpu_exception : yes cpuid level : 13 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr pdcm pcid sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes xsave avx lahf_lm ida arat epb xsaveopt pln pts dtherm tpr_shadow vnmi flexpriority ept vpid bogomips : 4988.44 clflush size : 64 cache_alignment : 64 address sizes : 36 bits physical, 48 bits virtual
comment:14 Changed 9 months ago by takis
Yes, disabling AVX with the configure flag avoids the segfault.
comment:15 Changed 9 months ago by Cigaes
- Keywords avx added
- Summary changed from Using atempo filter causes segfault to Using atempo filter causes segfault with AVX
comment:16 Changed 9 months ago by cehoyos
- Priority changed from normal to important
- Status changed from new to open
comment:17 Changed 9 months ago by Cigaes
- Analyzed by developer set
- Reproduced by developer set
It is an alignment problem: AVX requires 32, the memory used is provided by av_realloc and only aligned to 16.
comment:18 Changed 9 months ago by Cigaes
Can you test if the following patch fixes the problem for you:
http://ffmpeg.org/pipermail/ffmpeg-devel/2012-August/130215.html
?
comment:19 Changed 9 months ago by takis
Yes, the patch works for me.
Thanks!
comment:20 Changed 9 months ago by Cigaes
- Status changed from open to closed
- Resolution set to fixed
Fixed in commit a5704659e3e41b7698812b89f67d9a7467a74d20.
Works for me with current git head. If it still happens for you, can you be more specific about your environment (looks like Ubuntu: 32 or 64 bits) and produce a valgrind backtrace (using ffmpeg_g and no shared libraries)?