[Ffmpeg-devel-irc] ffmpeg-devel.log.20120114

burek burek021 at gmail.com
Sun Jan 15 02:05:03 CET 2012

[00:37] <funman> ok finally got it
[00:37] <funman> http://pastie.org/3181237
[00:37] <funman> the .exe and dlls all have debug symbols
[00:38] <funman> i can insert breakpoints and have complete backtraces when they happen
[00:38] <Compn> lol
[00:38] <Compn> ok that is a useless backtrace
[00:38] <funman> indeed but the other binary i have (without symbols) actually show function name in the bt
[00:38] <Compn> no function names
[00:40] <uau> funman: do you have function names when you set a breakpoint elsewhere?
[00:40] <funman> it just says it's in DXVA2CreateVideoService() from Windows' dxva2.dll
[00:40] <uau> could be that the issue is that the stack is corrupted
[00:40] <funman> uau: yeah, just checked with WinMain
[00:40] <funman> likely
[00:40] <uau> so it would not be an issue looking up the names, but the addresses might not correspond to functions
[00:42] <funman> even the last one?
[00:42] <funman> shouldn't it be resolved from IP/PC ?
[00:43] <uau> well maybe it crashed after jumping to a return address read from corrupted stack
[00:43] <funman> true
[00:43] <funman> it's Windows code so we dont know
[00:43] <uau> you could check whether the asm makes any sense
[00:45] <funman> the address doesn't make sense at all
[00:45] <funman> "No function contains program counter for selected frame."
[00:46] <funman> VLC looks if that function exist but doesn't call it
[00:46] <uau> i think that message you quoted only means there's no debug info
[00:46] <uau> so it can't automatically show asm for exactly the whole function
[00:46] <funman> ah but it comes from 'disassemble'
[00:46] <uau> you need to give it more arguments probablly
[00:47] <funman> ooh ok
[00:47] <funman> well it makes sense
[00:47] <funman> it is NULL dereference
[00:47] <uau> what is?
[00:48] <funman> http://pastie.org/3181294
[00:48] <funman> rdx = 0
[00:49] <uau> that looks like meaningful code though
[00:49] <funman> ffmpeg has no reference to DXVA2CreateVideoService
[00:49] <uau> so lack of debug info rather than a random address
[00:49] <funman> uau: yeah, the first line is *(rdx) = rax; iiuc (translated to C)
[00:50] <uau> well that alone could happen to occur in random memory
[00:50] <funman> yeah
[00:50] <uau> i mean that overall the asm looks like a meaningful function
[00:51] <funman> the second function in the backtrace doesnt point to valid memory though
[00:52] <funman> in any case it's annoying to debug
[00:55] <funman> ah and i dont see symbols defined in vlc .dll s
[00:56] <funman> http://pastie.org/3181327
[00:57] <funman> enough for today :)
[00:57] <uau> funman: was that after starting the process?
[00:58] <funman> yes
[00:59] <funman> same thing if i don't run the process
[00:59] <funman> i'll ask help on #mingw-w64 they are helpful - just need some timezone concordance
[01:00] <uau> so the libraries should already have been loaded?
[01:00] <funman> yeah
[01:00] <funman> well maybe
[01:00] <funman> DLLs are special
[01:03] <funman> ramiro left a long time ago? last time i saw him was 1 year ago or something
[01:04] <michaelni> last post from ramiro on ffmpeg-dev was 5 Nov 2011, thats not so long ago
[01:13] <Compn> ramiro left irc and stopped his autobuilds site after the fork
[01:13] <Compn> but i saw him once or twice on the list
[01:13] <Compn> or something
[01:17] <CIA-31> ffmpeg: 03Carl Eugen Hoyos 07master * r7c10d3ca2e 10ffmpeg/libavcodec/ (pamenc.c version.h): 
[01:17] <CIA-31> ffmpeg: Simplify 32bit RGB pam encoding.
[01:17] <CIA-31> ffmpeg: Reviewed-by: Paul B Mahol
[01:17] <CIA-31> ffmpeg: 03Carl Eugen Hoyos 07master * r93af38ca93 10ffmpeg/ffmpeg.c: 
[01:17] <CIA-31> ffmpeg: Increase maximum encoding video buffer size.
[01:17] <CIA-31> ffmpeg: libavutil now supports 64bit RGB formats.
[01:17] <CIA-31> ffmpeg: 03Carl Eugen Hoyos 07master * r3fe4540b08 10ffmpeg/libavcodec/ (pamenc.c version.h): Support 64bit pam encoding.
[01:18] <CIA-31> ffmpeg: 03Carl Eugen Hoyos 07master * rc714cd3bd4 10ffmpeg/libavcodec/ (tiff.c tiffenc.c version.h): 
[01:18] <CIA-31> ffmpeg: Support 64bit tiff images.
[01:18] <CIA-31> ffmpeg: Fixes a part of ticket #503.
[01:18] <CIA-31> ffmpeg: Reviewed-by: Paul B Mahol
[01:18] <CIA-31> ffmpeg: Reviewed-by: Jean First
[01:18] <CIA-31> ffmpeg: 03Carl Eugen Hoyos 07master * r0b1f20e23c 10ffmpeg/libavcodec/sgidec.c: 
[01:18] <CIA-31> ffmpeg: Support reading 64bit sgi images.
[01:18] <CIA-31> ffmpeg: Reviewed-by: Jean First
[01:18] <CIA-31> ffmpeg: 03Carl Eugen Hoyos 07master * rdd55adb1a7 10ffmpeg/libavcodec/dpx.c: 
[01:18] <CIA-31> ffmpeg: Support reading 64bit dpx images.
[01:18] <CIA-31> ffmpeg: Reviewed-by: Jean First
[01:19] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r8d95eb6702 10ffmpeg/libavcodec/utils.c: 
[01:19] <CIA-31> ffmpeg: lavc: Prevent opening of experimental codecs if strict_std_compliance > FF_COMPLIANCE_EXPERIMENTAL.
[01:19] <CIA-31> ffmpeg: Some of these encoders may produce invalid bitstreams, which should not
[01:19] <CIA-31> ffmpeg: be done without the user knowing.
[01:19] <CIA-31> ffmpeg: Some of these decoders may be unfinished and may contain security issues.
[01:19] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:19] <CIA-31> ffmpeg: 03Baptiste Coudurier 07master * r89fc7e36c7 10ffmpeg/libavcodec/alac.c: 
[01:19] <CIA-31> ffmpeg: alacdec: support 32 bps
[01:19] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:24] <funman> Compn: :(
[01:24] <funman> there are no windows build on fate ?
[01:25] <funman> running tests would require wine but build-only works on linux
[01:41] <CIA-31> ffmpeg: 03Carl Eugen Hoyos 07master * rffd1017fb8 10ffmpeg/libavcodec/ (libopenjpegenc.c version.h): 
[01:41] <CIA-31> ffmpeg: Gray16 encoding with libopenjpeg.
[01:41] <CIA-31> ffmpeg: Reviewed-by: Michael Bradshaw
[01:41] <CIA-31> ffmpeg: 03Carl Eugen Hoyos 07master * r36397ea1c7 10ffmpeg/libavcodec/ (libopenjpegenc.c version.h): 
[01:41] <CIA-31> ffmpeg: RGBA64 encoding with libopenjpeg.
[01:41] <CIA-31> ffmpeg: Reviewed-by: Michael Bradshaw
[02:26] <Compn> funman : ramiro ran the win32 fate boxes :(
[02:27] <Compn> funman : libav runs some mingw fate
[02:27] <Compn> http://fate.libav.org/
[02:28] <Compn> so at least thats pretty close to ffmpeg :)
[02:30] <Compn> with a normal gcc version mingw seems ok
[02:42] <funman> yeah mingw is just the runtime and standard library
[02:42] <funman> all gcc patches are upstream these days
[02:53] <CIA-31> ffmpeg: 03Diego Biurrun 07master * r64d779f2f7 10ffmpeg/libavcodec/ (cabac.c cabac.h): 
[02:53] <CIA-31> ffmpeg: cabac: Remove ff_h264_lps_state array.
[02:53] <CIA-31> ffmpeg: It was only ever used in the cabac test program, but never initialized.
[02:53] <CIA-31> ffmpeg: 03Diego Biurrun 07master * r0a60780c7f 10ffmpeg/libavcodec/ (cabac.c cabac.h): cabac: Mark ff_h264_mps_state array as static, it is only used within cabac.c.
[02:53] <CIA-31> ffmpeg: 03Ronald S. Bultje 07master * re7843db3df 10ffmpeg/libswscale/swscale_unscaled.c: (log message trimmed)
[02:53] <CIA-31> ffmpeg: swscale: fix invalid conversions and memory problems.
[02:53] <CIA-31> ffmpeg: Fixes problems where rgbToRgbWrapper() is called even though it doesn't
[02:53] <CIA-31> ffmpeg: support this particular conversion (e.g. converting from RGB444 to
[02:53] <CIA-31> ffmpeg: anything). Thirdly, fixes issues where rgbToRgbWrapper() is called for
[02:53] <CIA-31> ffmpeg: non-native endiannness conversions (e.g. RGB555BE on a LE system).
[02:53] <CIA-31> ffmpeg: Fourthly, fixes crashes when converting from e.g. monowhite to
[02:53] <CIA-31> ffmpeg: 03Mans Rullgard 07master * r5c5e1ea3cd 10ffmpeg/libavcodec/arm/asm.S: (log message trimmed)
[02:53] <CIA-31> ffmpeg: ARM: 4-byte align start of all asm functions
[02:53] <CIA-31> ffmpeg: Due to apprent bugs in the GNU assembler and/or linker, relocations
[02:53] <CIA-31> ffmpeg: can be incorrectly processed if the alignment of a Thumb instruction
[02:53] <CIA-31> ffmpeg: is changed in the output file compared to the input object.
[02:53] <CIA-31> ffmpeg: This fixes crashes in h264 decoding with Thumb enabled. No effect in
[02:53] <CIA-31> ffmpeg: ARM mode since everything is 4-byte aligned there.
[02:53] <CIA-31> ffmpeg: 03Paul B Mahol 07master * r0cc1a86dc3 10ffmpeg/libswscale/ (rgb2rgb.c rgb2rgb.h swscale_unscaled.c): 
[02:54] <CIA-31> ffmpeg: rgb2rgb: rgb12to15()
[02:54] <CIA-31> ffmpeg: Signed-off-by: Ronald S. Bultje <rsbultje at gmail.com>
[02:54] <CIA-31> ffmpeg: 03Justin Ruggles 07master * r68d6012c72 10ffmpeg/tests/ (10 files in 2 dirs): 
[03:14] <durandal_1707> build is failling for me
[03:15] <Compn> pastebin you have not made
[03:15] <durandal_1707> lps/mlps for cabac
[03:16] <durandal_1707> libavcodec/cabac.c:149:13: error: use of undeclared identifier 'ff_h264_lps_state'; did you mean 'ff_h264_mlps_state'?
[04:14] <durandal_1707> answer to clang question is No
[04:50] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * rc834c7e39c 10ffmpeg/libavcodec/cabac.c: 
[04:50] <CIA-31> ffmpeg: 1000l commit forgotten cabac fix
[04:50] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:51] <michaelni> darn i had fixed it but forgot commit --amend 
[04:55] <durandal_1707> hmm, i fixed it differently, reusing h264_mps_state
[05:01] <michaelni> well i just reverted the 2 line removials that broke it
[05:01] <michaelni> seemed the most straightforward
[05:04] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r2422e8271c 10ffmpeg/libavcodec/cabac_functions.h: 
[05:04] <CIA-31> ffmpeg: cabac_functions: fix license header, this wasnt developed in libav
[05:04] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:04] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r44e191457a 10ffmpeg/libavcodec/utils.c: 
[05:04] <CIA-31> ffmpeg: utils: check that parameters havnt changed in reget_buffer()
[05:04] <CIA-31> ffmpeg: Fixes Ticket902
[05:04] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[07:29] <CIA-31> ffmpeg: 03Oana Stratulat 07master * ra63a86fd6f 10ffmpeg/libavformat/avidec.c: 
[07:29] <CIA-31> ffmpeg: Fixes issue 890: OOM with zmbv file
[07:29] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[10:41] <CIA-31> ffmpeg: 03Carl Eugen Hoyos 07master * r6ba74be511 10ffmpeg/libavcodec/utils.c: Add CR/LF to a reget_buffer warning message.
[10:41] <CIA-31> ffmpeg: 03Carl Eugen Hoyos 07master * r28002030f2 10ffmpeg/libavcodec/iff.c: 
[10:41] <CIA-31> ffmpeg: Support transparency in iff images.
[10:41] <CIA-31> ffmpeg: Fixes ticket #707.
[10:41] <CIA-31> ffmpeg: Reviewed-by: Peter Ross
[12:37] <CIA-31> ffmpeg: 03Stefano Sabatini 07master * r4057505e28 10ffmpeg/doc/bitstream_filters.texi: doc/bistream_filters.texi: add documentation for the h264_mp4toannexb filter
[14:21] <j-b> 'morning
[14:37] <pross-au> very early morning here
[15:20] <kirarav2> hello :)
[15:47] <pross-au> slow news day
[15:52] <Compn> pross-au : your comment on mikes blog, what did you mean ffmpeg merge with fmpeg, what fmpeg ?
[15:52] <pross-au> oops
[15:53] <pross-au> ffmpeg merge with libav
[15:53] <Compn> ah
[15:53] <pross-au> reunification to the motherland
[15:53] <Compn> :D
[15:53] <pross-au> dare i say, tawain will merge first
[15:53] <Compn> and the dalai lama will turn catholic
[15:53] <Compn> :D
[15:54] <Compn> china got to him
[15:56] <pross-au> oops^2
[15:59] <Compn> wb
[16:01] <pross-au> slip of the finger
[16:55] <CIA-31> ffmpeg: 03Marton Balint 07master * r1348420b8a 10ffmpeg/ffplay.c: 
[16:55] <CIA-31> ffmpeg: ffplay: silence buffer size must be a multiple of frame size
[16:55] <CIA-31> ffmpeg: Otherwise the audio channels sometimes would be shifted on output.
[16:55] <CIA-31> ffmpeg: Signed-off-by: Marton Balint <cus at passwd.hu>
[16:55] <CIA-31> ffmpeg: 03Marton Balint 07master * r6f06545b26 10ffmpeg/ffplay.c: (log message trimmed)
[16:55] <CIA-31> ffmpeg: ffplay: use swr_set_compensation for audio synchronization
[16:55] <CIA-31> ffmpeg: Also change synchronize_audio to only calculate the wanted number of samples
[16:55] <CIA-31> ffmpeg: instead of doing the actual synchronization, and make swr_convert handle the
[16:55] <CIA-31> ffmpeg: sample addition or deletion.
[16:55] <CIA-31> ffmpeg: This new method replaces the old buggy way which simply deleted or
[16:55] <CIA-31> ffmpeg: multiplied samples.
[16:55] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r0b4f1e0d3d 10ffmpeg/: 
[16:55] <CIA-31> ffmpeg: Merge remote-tracking branch 'cus/stable'
[16:55] <CIA-31> ffmpeg: * cus/stable:
[16:55] <CIA-31> ffmpeg:  ffplay: silence buffer size must be a multiple of frame size
[16:55] <CIA-31> ffmpeg:  ffplay: use swr_set_compensation for audio synchronization
[17:22] <CIA-31> ffmpeg: 03Laurent BRULET 07master * r45c39e566f 10ffmpeg/libavformat/udp.c: 
[17:22] <CIA-31> ffmpeg: udp: fix segfault on closing
[17:22] <CIA-31> ffmpeg: Fixes ticket915
[17:24] <CIA-31> ffmpeg: 03Reimar Döffinger 07master * rc4ba5198ea 10ffmpeg/libavcodec/ (avcodec.h avpacket.c internal.h utils.c): (log message trimmed)
[17:24] <CIA-31> ffmpeg: Fix leaking of side data.
[17:24] <CIA-31> ffmpeg: While we correctly "register" the side data when we split it,
[17:24] <CIA-31> ffmpeg: the application (in this case FFmpeg) might not update the
[17:24] <CIA-31> ffmpeg: AVPacket pool it uses to finally free the packet, thus
[17:24] <CIA-31> ffmpeg: causing a leak.
[17:24] <CIA-31> ffmpeg: This also makes the av_dup_packet unnecessary which could
[17:44] <CIA-31> ffmpeg: 03Reimar Döffinger 07master * r83b12c16af 10ffmpeg/libavcodec/x86/rv34dsp.asm: 
[17:44] <CIA-31> ffmpeg: Use correct register size, fixes compilation with NASM.
[17:44] <CIA-31> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[20:12] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * rb26f0ee39f 10ffmpeg/libavformat/ (avidec.c riff.c): 
[20:12] <CIA-31> ffmpeg: avidec: XMPG support
[20:12] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:12] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r0d5c810bd5 10ffmpeg/libavcodec/mpeg12.c: 
[20:12] <CIA-31> ffmpeg: mpeg12dec: check uppercased codec_tag.
[20:12] <CIA-31> ffmpeg: We do this for all other codec_tag checks in mpegvideo*/h26*
[20:12] <CIA-31> ffmpeg: doing it here too makes the code more consistent.
[20:12] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:12] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r52889b543d 10ffmpeg/ (libavcodec/mpeg12.c libavformat/riff.c): 
[20:12] <CIA-31> ffmpeg: mpeg12dec: BW10 support
[20:12] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:23] <michaelni> ping to my ISPs web server: 652 packets transmitted, 135 received, 79% packet loss, time 651074ms
[22:23] Action: michaelni kicks his ISP very hard
[22:24] <Daemon404> why dont you ssh into somewhere faster?
[22:24] <Daemon404> rather, more reliable
[22:24] <nevcairiel> ssh isnt much fun when your connection is dropping data, either
[22:24] <Daemon404> true
[22:29] <ubitux> are you forced to keep your current isp?
[22:30] <ubitux> btw, Diego, Janne, Luca and Ronald didn't answer to me request about libav.org&
[22:31] <durandal_1707> ubitux: request about what?
[22:31] <Compn> yea whut request
[22:32] <ubitux> about the permission to use their work on libav.org for ffmpeg.org (web repository)
[22:32] <Compn> lol
[22:32] <ubitux> well i don't mind an explicit refusal
[22:32] <ubitux> but being ignored really is the worse
[22:32] <ubitux> you can't just move on
[22:33] <ubitux> it's not the first i'm deliberatly ignored from libav :(
[22:33] <Compn> i have not heard much from them since the fork. guess i'm on the 'list'.
[22:34] <Compn> ubitux : you tried in #libav-devel ?
[22:34] <Compn> or by email ?
[22:34] <ubitux> it was by mail
[22:34] <Compn> ouch
[22:34] <ubitux> i also got ignored several time on #libav-devel and in private on irc but well&
[22:35] <ubitux> (not for that matter)
[22:35] <Compn> like i tried explaining to people, there is a big communications problem
[22:36] <ohsix> that's too polite
[22:37] <Compn> oh, i mentioned communications problem in some ffmtech thread
[22:37] <ubitux> i think i don't have access to it
[22:37] <ubitux> i don't know at all what's happening there
[22:38] <Compn> i think it was on ffmpeg-devel
[22:38] <Compn> and yeah, i havent a clue whats going on, aside from who is on the board
[22:39] <Compn> lu_zero offered to make a website on ffmtech.org , but hes been busy i guess.
[22:39] <ubitux> anyway, i'll ping them with ffmpeg-devel in copy next time
[22:39] <ubitux> maybe next week.
[22:39] <Compn> and my idea to pay someone to create a quick website was ignored
[22:40] <Compn> merbanan said i may be able to help with ffmtech, i wonder how that went
[22:40] <Compn> have to ask him next time hes around
[22:45] <durandal_1707> how much bugs from old tracker have been cheched to be still alive?
[22:46] <michaelni> durandal_1707, dunno, best ask carl, he is the one knowing all the bugs to the finest detail
[22:47] <durandal_1707> our mjpeg decoder sucks
[22:47] <michaelni> why ?
[22:49] <durandal_1707> does not support some stuff
[22:50] <michaelni> some stuff ?
[22:59] <durandal_1707> #903 dunno what is cause of that and #907 32bpp pix_fmt come to mind
[23:01] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r5a840f6364 10ffmpeg/libswscale/x86/yuv2rgb_mmx.c: 
[23:01] <CIA-31> ffmpeg: x86/yuv2rgb_mmx: Remove yuv420 check.
[23:01] <CIA-31> ffmpeg: This check is not correct, the code supports more yuv variants.
[23:01] <CIA-31> ffmpeg: Fixes Ticket904
[23:01] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:01] <CIA-31> ffmpeg: 03Jean First 07master * rc48883163d 10ffmpeg/libavcodec/timecode.h: 
[23:01] <CIA-31> ffmpeg: timecode: fix typo
[23:01] <CIA-31> ffmpeg: Signed-off-by: Jean First <jeanfirst at gmail.com>
[23:01] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:07] <Compn> heh
[23:07] <Compn> durandal_1707 : carl and michael have been adding crazy amounts of mjpeg formats lately :D
[23:08] <Compn> and for some reason ami_stuff has been reporting all of the weird image formats 
[23:08] <Compn> and kostya did a bunch of them too
[23:08] <Compn> ffmpeg is going to take over imagemagick one day :D
[23:10] <Compn> i dont really know the reason why everyone is adding rare jpg/image formats
[23:10] <Compn> easy to reverse engineer i guess :)
[23:10] <CIA-31> ffmpeg: 03Reimar Döffinger 07master * r7a1723086a 10ffmpeg/libavcodec/x86/mpegaudiodec_mmx.c: (log message trimmed)
[23:10] <CIA-31> ffmpeg: Fix compilation without HAVE_AVX, HAVE_YASM etc.
[23:10] <CIA-31> ffmpeg: At the very least this should fix warnings about unused static
[23:10] <CIA-31> ffmpeg: functions if one or more of these is not defined.
[23:10] <CIA-31> ffmpeg: However even compilation might be broken if the compiler does
[23:10] <CIA-31> ffmpeg: not optimize the function away completely.
[23:10] <CIA-31> ffmpeg: This actually happens in case of the AVX function, since the
[23:37] <michaelni> durandal_1707, fix for #903 is on its way to the ML
[23:38] <durandal_1707> omg, how you do that?
[23:52] <michaelni> durandal_1707, patch for #907 is also on its way to the ML
[00:00] --- Sun Jan 15 2012

More information about the Ffmpeg-devel-irc mailing list