[FFmpeg-devel-irc] IRC log for 2010-08-19

irc at mansr.com irc at mansr.com
Fri Aug 20 02:00:34 CEST 2010


[00:45:19] <Dark_Shikari> 10:43 < fbarchard> i'm seeing a ~20% slowdown in ff vp8 decoding now compared to aug 2.  is that a known issue/change?
[00:45:22] <Dark_Shikari> 10:44 <@Dark_Shikari> Yes.  your computer got 20% slower
[00:47:21] <kierank> maybe he's talking about ffmpeg-mt
[00:49:23] <Dark_Shikari> 10:48 < booi> the bits are 20% slower on wednesdays because of the electrons
[00:59:02] <_skal_> will check locally
[01:34:39] <_skal_> unconfirmed.
[02:21:57] <Compn> hey its skal again :)
[02:22:09] <Compn> should we buy google stock? anything new in the pipeline ?
[03:31:55] <mchinen> anyone get "xargs: illegal option -- d" when running patcheck on mac?
[03:32:09] <mchinen> after it outputs some feedback I always hit this
[04:12:46] <astrange> vp8 won't be any different in ffmpeg-mt
[04:13:17] <astrange> i actually did change that file... i added EDGE_TOP to mpegvideo and it conflicts with EDGE_TOP in vp8. but it was just a renaming
[04:13:20] <astrange> i should send it anyway
[05:24:53] <saintdev> got long block analysis ported over.
[05:25:08] <saintdev> no i just need to make sure it actually works
[05:25:12] <saintdev> *now
[05:26:15] <kshishkov> can you also port MP3 encoder so we can get rid of libmp3lame support?
[05:26:46] <saintdev> mp3 is dead
[05:27:02] <drv> long live mp3
[05:27:07] <saintdev> and what's wrong with lame?
[05:27:19] <kshishkov> mp3 has braindead bitstream format
[05:27:34] <kshishkov> (and designed for coding pop music)
[06:23:32] <saintdev> anybody have a guess on the next values in this sequence: http://pastebin.org/605838
[06:23:49] <saintdev> not at all sure how these were derived :/
[06:56:42] <siretart> morning
[07:07:03] <kshishkov> god morgon
[07:10:22] <andoma> hej
[07:15:26] <av500> ho
[07:25:18] <mchinen> is there anything like a ring buffer in libav*?
[07:26:04] <pJok> god morgon, kshishkov
[07:26:13] <astrange> libavutil/fifo.h
[07:26:18] <mchinen> ah I forgot its god morgon time
[07:27:34] <mchinen> astrange: great, thanks!
[07:33:37] <kshishkov> pJok: you may not understand that but it's really good morning for me - I drank Trocadero and has a bit of Prästost in the fridge
[07:37:35] <pJok> kshishkov, so you have your little slice of sweden in the fridge then?
[07:37:53] <kshishkov> pJok: indeed
[07:46:46] <pJok> kshishkov, congrats... now you just need to move there :)
[07:46:52] <astrange> i found an unstripped official vp6 decoder
[07:46:56] <astrange> is it too late to RE that?
[07:47:08] <kshishkov> probably yes
[07:47:29] <kshishkov> you'd better take VP4 or VP7
[07:47:32] <kshishkov> or TM2X
[07:51:37] <superdump> kshishkov: i have prästost in my fridge too
[07:51:45] <superdump> but i've never had trocadero
[07:52:11] <kshishkov> well, you're not from Norrlands
[08:08:09] <astrange> do we know why vp6 .flv has one byte of extradata? i don't see vp6.c using it
[08:19:15] <spaam> superdump: so you like sweden also? :)
[08:22:21] <kshishkov> astrange: it's used for width and height coding IIRC and is skipped in flv demuxer since it's FLV-specific
[08:27:32] <superdump> spaam: yeah, i do
[08:27:38] * superdump is in stockholm
[08:29:54] <spaam> superdump: cool :)
[08:33:11] <kshishkov> spaam: and what's your opinion on Trocadero?
[08:34:31] <spaam> The best :)
[08:35:32] <spaam> when i lived in norrland.. i always had trocadero to my pizza ,D  here in south they dont have it :(
[08:35:51] <kshishkov> south?
[08:36:09] <spaam> yeah.. Ronneby is in south part of sweden ;D
[08:36:52] <kshishkov> Götaland, it's south indeed
[08:38:12] <kshishkov> personally I've never been to south Sweden much, only one day in Göteborg
[08:38:31] <kshishkov> and somehow I want to see northern parts more
[08:38:45] * kshishkov has never been to the north of Gävle
[08:39:37] <spaam> then you have never been to norrland but some ppl say that gävle is a part of it
[08:40:02] <kshishkov> it's a border, yes
[08:40:17] <kshishkov> but Sundsvall will definitely count ;)
[08:40:27] <spaam> mm :)
[09:36:00] <mru> hey, stop splitting my net
[09:37:36] <thresh> come on, it is freenode
[09:37:48] <av500> netsplits are free?
[09:37:52] <kshishkov> so get your money back
[09:39:51] <av500> who do i complain to? al goreß
[09:39:53] <av500> ?
[09:40:08] <spaam> do it
[09:40:50] <kshishkov> well, IRC was here even before he invented Internet
[09:54:15] <KotH> moin boys
[09:55:35] <spaam> time to wake up now KotH ?
[09:56:12] <KotH> nah.. time to start up the computer and log in
[09:59:50] <cartman> morning
[10:41:26] * KotH hands cartman a glas of osaft 
[10:41:55] <cartman> what to do with it ;)
[10:42:08] <KotH> drink?
[10:43:07] <cartman> I don't drink what I can't spell ;>
[10:43:11] <kshishkov> and what juice is that?
[10:43:18] <osaft> Oo
[10:43:53] * kshishkov is suspicious about alphabet juices. 42saft would be better
[10:44:05] <cartman> osaft: are you drinkable? :)
[10:44:31] <osaft> this takes a disgusting direction in my mind
[10:44:38] <cartman> LOL
[10:44:43] <cartman> osaft is new around here I guess
[10:45:20] <osaft> :-)
[10:46:22] <KotH> cartman: yes, he apeared on 2009-06-06T20:14+0200
[10:46:31] <cartman> KotH: oh :)
[10:46:57] <osaft> Good to know :-D
[10:47:30] <KotH> kshishkov: ask the guys around you what osaft means :)
[11:02:18] <twnqx> hey, could someone please apply my patch for the sanitycheck on bitrates from 13 days ago? :X
[11:37:47] * KotH applies the sanity check to twnqx 
[11:37:52] <KotH> hmm.. it somehow fails
[11:56:41] <twnqx> KotH: not to me! of course that would fail!
[11:57:26] * KotH puts a [FAIL] stamp on twnqx's forehead
[14:10:18] <jhuntwork> KotH, got it working
[14:31:29] <kierank_> so will we get a working ffmidi player soon?
[14:31:47] <twnqx> the more interesting question would be
[14:31:54] <twnqx> where do you take the samples from?
[14:32:17] <twnqx> and: would it support that great 4G steinberg piano sample set?
[14:32:41] <kshishkov> kierank_: just add good soundbank loading or good ffsynth
[14:32:42] <av500> ffkaraoke ftw!
[14:33:04] * kshishkov would like to have adplug.sf.net integrated into FFmpeg though
[14:33:10] <kierank_> i was hoping we'd get ffmidi with libavsequencer but all the discussion is going way over my head
[14:33:21] <CIA-93> ffmpeg: stefano * r24830 /trunk/libavfilter/avfilter.c: Cosmetics: apply misc spacing style fixes.
[14:33:22] <CIA-93> ffmpeg: stefano * r24831 /trunk/libavfilter/avfilter.c:
[14:33:22] <CIA-93> ffmpeg: Enable and change the log level to DEBUG for the message printed in
[14:33:22] <CIA-93> ffmpeg: case of picref copy.
[14:33:49] <kierank_> wow that adplug thing plays loads of formats
[14:33:53] <kierank_> i've never heard of most of them
[14:34:29] <BBB> kshishkov: why does rv40 not have ssse3 ops?
[14:34:45] <BBB> kshishkov: would the rounding differences make h264 significantly slower?
[14:34:58] <kshishkov> kierank_: most of those come from DOS games, the rest is from demo scene
[14:35:39] <kshishkov> BBB: nobody bothered to write them.
[14:36:05] <BBB> oh that's all?
[14:36:07] <BBB> hm...
[14:36:16] <twnqx> kshishkov: hardware midi support!
[14:36:30] <Dark_Shikari> So, what will we do about soundfont libraries?
[14:36:34] * twnqx has a terratec maestro 32/96 ISA souncard with rom sample banks
[14:36:40] <kshishkov> twnqx: I won't object, go on
[14:36:41] <Dark_Shikari> libavsequencer is nothing without soundfonts
[14:36:44] <twnqx> just for the purpose of playing midi files.
[14:36:53] <av500> twnqx: terratec :)
[14:37:08] <twnqx> kshishkov: noone even applies my simple crashfix patches, do you think some more code would go easier?
[14:37:18] <kshishkov> Dark_Shikari: well, you can generate some generic sine wave for all instruments ;)
[14:37:22] <Dark_Shikari> lol
[14:37:39] <kshishkov> actually I used suck player for a while
[14:37:43] <kshishkov> *such
[14:37:48] <kshishkov> (yes, it sucked too)
[14:38:04] <twnqx> av500: that card is the best thing you can get for playing the old duke nukem 3D
[14:38:10] <twnqx> i bought it on ebay even...
[14:38:24] <twnqx> it's in my classic PC though.
[14:38:24] <Dark_Shikari> http://x264.nl/developers/Dark_Shikari/UNOwenWasHer.mp3 <---touhou midi rendered with a good sequencer
[14:38:29] <kshishkov> twnqx: Creative AWE64?
[14:38:30] <Dark_Shikari> I want ffmpeg to sound at least that good.
[14:38:45] <twnqx> kshishkov: terratec maestro 32/96 + gravis ultrasound ace
[14:39:00] <BBB> Dark_Shikari: you're into midi now?
[14:39:06] <kshishkov> twnqx: gravis is good!
[14:39:10] <twnqx> paired with a matrox millenium 2 + diamond voodoo 3d for graphics
[14:39:24] <kshishkov> BBB: since Touhou games use midi music (some of them at least) he is
[14:39:44] <Dark_Shikari> BBB: well, ZUN still does everything with mod+tracker stuff, I think
[14:39:49] <twnqx> powered by a... pentium mmx 233!
[14:39:51] <Dark_Shikari> therefore, mod == cool
[14:39:55] <BBB> heh :)
[14:40:04] <kshishkov> twnqx: looks like you got quite awesome vintage hardware indeed
[14:40:17] * pJok has some Xeon vintage hardware
[14:40:27] <twnqx> kshishkov: took quite some time to get all the pieces again
[14:40:30] <pJok> dual xeon, 2.8ghz with HT
[14:40:34] <BBB> Dark_Shikari: if you want cheap optimization credit, apply my patches and write h264 chroma mc2/mc4 1D/0D functions, should make ffh264dec a little faster
[14:40:37] <pJok> but aparantly no 64bit
[14:40:40] * BBB lazy
[14:40:45] <BBB> and I want to go back to xvp8 also
[14:40:51] <twnqx> still need the extended cache ram for the gigabyte mainboard so i can cache above 64MB ram
[14:40:54] <Dark_Shikari> BBB: are those chroma positions really used often enough to be worth the branch?
[14:41:08] <Dark_Shikari> oh you just mean normal 1D
[14:41:10] <BBB> it's branchless
[14:41:12] <Dark_Shikari> not mpeg2 shortcut 1D
[14:41:25] <Dark_Shikari> an unpredictable function call is a branch too
[14:41:31] <BBB> hm...
[14:41:39] <BBB> isn't a function pointer always unpredictable?
[14:41:47] <kshishkov> yes
[14:41:48] <Dark_Shikari> no
[14:41:52] <twnqx> not if it's inlined :P
[14:41:53] <BBB> lol :)
[14:41:53] <Dark_Shikari> modern CPUs predict indirect branches
[14:41:58] <twnqx> oh, indirect
[14:42:01] <BBB> really? that's amazing
[14:42:10] <BBB> anyway, I'll test and see if it's worth it
[14:42:14] * kshishkov waits for modern CPUs to predict women's behaviour
[14:42:26] <Dark_Shikari> I mean, even before indirect branch prediction
[14:42:31] <Dark_Shikari> they at least predicted "it will go the same place it did last time"
[14:42:35] <ohsix> no trace cache big enough
[14:42:36] <Dark_Shikari> even crappy CPUs do that
[14:42:50] <BBB> I'll test it and see if it actually improves it for real
[14:42:56] <BBB> note how I haven't applied any of these patches yet
[14:43:05] <BBB> that's because it all depends on whether the optimization actually works or not
[14:43:12] <BBB> if it doesn't, Ill throw it out b/c I wasted my time ;)
[14:43:27] <BBB> I think it will
[14:43:29] <BBB> but I'm not sure
[14:46:48] <BBB> and why are there so many ^&% on libav-user who want to use ffh264 for decoding, but insist on building their own rtp stack, and then ask "why doesn't it work" and refusing to test if ffplay can actually play it (i.e. if you use the built-in rtp stack)?
[14:47:15] <BBB> did they have a class assignment in some idiot university or so?
[14:47:25] <KotH> probably
[14:47:28] <Dark_Shikari> prolly
[14:48:08] <kshishkov> BBB: dunno, there were no classes in my university on misusing libav*
[14:49:00] <KotH> BBB: check if they all come from teh same region
[14:49:13] <BBB> let me check that
[14:50:20] <BBB> hotmail :-p
[14:50:22] <BBB> difficult to say
[14:50:51] <CIA-93> ffmpeg: mstorsjo * r24832 /trunk/libavformat/ (utils.c httpauth.c internal.h): Make parse_key_value from httpauth a common lavf internal function
[14:51:44] <CIA-93> ffmpeg: mstorsjo * r24833 /trunk/libavformat/ (http.c internal.h): Move the definition of the maximum url size for static buffers to internal.h
[14:51:55] <KotH> rotfl... the police raided the google office in zürich
[14:52:08] <mru> why?
[14:52:17] <KotH> reason: a swiss took his rifle to work, because he has to go to the shooting range this evening
[14:52:34] <KotH> damn foreigners, dont know the culture here!
[14:52:35] <KotH> :-)
[14:53:04] <ohsix> american office workers are all just waiting to shoot up the place
[14:53:25] <kierank_> KotH: google zurich doesn't have an armoury?
[14:53:27] <kierank_> pffft
[14:53:54] <KotH> kierank_: armoury? what for?
[14:54:06] <kierank_> storing weapons
[14:54:06] <KotH> we're in .ch, we dont need to hide our rifles
[14:54:29] <kierank_> why did the police raid then?
[14:54:34] <KotH> two years ago, i did the same. took the rifle to work in the morning, put it beside my desk
[14:54:50] <KotH> kierank_: somebody called the police, so they had to do something
[14:55:27] <CIA-93> ffmpeg: mstorsjo * r24834 /trunk/ (5 files in 3 dirs): Add Apple HTTP Live Streaming demuxer
[14:55:32] <KotH> the only thing my coworkers did, was that two of the foreigners here took pictures, nothing else ^^'
[14:55:36] <ohsix> did he have it in a scabbard or was he just carrying it in?
[14:55:55] <KotH> probably just carying on his back, as it is the usual way to carry it around
[14:56:19] <KotH> heck, in may, when i had to go shooting, i went saturday morning to the range, shot my rounds, then went shopping
[14:56:28] <ohsix> you aren't going to shoot up the place with a rifle :O at best you'd only get one or two kills
[14:56:39] <KotH> i think it was late afternoon when i returned home... all the time with my rifle on my back
[14:57:05] <ohsix> people are pretty spooked by pistols
[15:10:16] <BBB> hi spyfeng :)
[15:17:13] * KotH shoulders his rifle and leaves
[15:17:20] <KotH> have a nice evening boys
[15:17:28] <KotH> and dont call the police! :-)
[15:17:30] <kierank_> is the lone ranger music playing in the background KotH ?
[15:17:45] <KotH> kierank_: nope: madlax ost :)
[16:50:09] <CIA-93> ffmpeg: aurel * r24835 /trunk/libavformat/ (avio.c avio.h avformat.h): add FF_API_URL_CLASS define to enable usage of URLContext as a AVClass
[16:52:27] <CIA-93> ffmpeg: mru * r24836 /trunk/tests/fate-run.sh: fate: add -v 0 to ffmpeg flags
[16:53:08] <CIA-93> ffmpeg: mru * r24836 /trunk/tests/fate-run.sh: fate: add -v 0 to ffmpeg flags
[16:53:08] <CIA-93> ffmpeg: mru * r24837 /trunk/tests/fate-run.sh: fate-run: use 'run' function in 'ffmpeg' function
[16:53:10] <CIA-93> ffmpeg: mru * r24838 /trunk/libavcodec/imc.c:
[16:53:10] <CIA-93> ffmpeg: imc: fix undefined float to int conversion
[16:53:10] <CIA-93> ffmpeg: Conversion of an out of range float to int is undefined. Clipping to
[16:53:10] <CIA-93> ffmpeg: the final range first avoids such problems. This fixes decoding on
[16:53:10] <CIA-93> ffmpeg: MIPS, which handles these conversions differently from many other CPUs.
[16:54:22] * twnqx looks at mru with puppy eyes - apply my proposed patch, please?
[16:54:42] <mru> what's your patch?
[16:54:45] <mru> is it approved?
[16:55:32] <twnqx> i don't think i ever received even comments on it
[16:55:42] <mru> ping it
[16:55:47] <twnqx> 13 days ago, bitrate sanitycheckign
[16:55:53] <twnqx> i can't, i fragged my mail server >_>
[16:57:25] <J_Darnley> twnqx: I'll look for it and bump it if you want
[16:57:35] <twnqx> would be great
[16:58:42] <J_Darnley> Are you Schmidt?
[16:59:07] <twnqx> yeah
[17:00:09] <J_Darnley> done
[17:02:12] <twnqx> thank you
[17:08:05] <CIA-93> ffmpeg: mru * r24839 /trunk/libavformat/applehttp.c: applehttp: define _XOPEN_SOURCE=600 as required for usleep()
[18:06:56] <BBB> omg sometimes I think I'm talking to a wall there
[18:07:14] <kshishkov> talk to ceiling, it's friendlier
[18:07:32] <BBB> hi ceiling "hi ronald", see, at least ceiling listens
[18:07:56] <kshishkov> at least you have some people around to treat you
[18:07:59] <BBB> I also don't understand inline asm
[18:08:12] <BBB> how do I tell gcc that I want two registries just to do stuff with?
[18:08:42] * kshishkov tries not to touch inline asm
[18:08:50] <kshishkov> especially with GCC
[18:09:58] <kurosu> BBB: you mean, to add some registers to the clobber list or something like this?
[18:10:14] <BBB> yeah, I guess... I totally don't understand this
[18:10:15] <kshishkov> probably not to fail with ICE
[18:10:30] <astrange> declare two ints and set them as outputs
[18:11:08] <kurosu> at the end of the asm block, you usually declare the input on a line starting with :
[18:11:24] <kurosu> s/input/output or things you write
[18:11:33] <BBB> I took put_pixels4_mmx and am trying to convert it to something that writes two pixels, and in x86 simd that would be mov [in], r%dw followed by move r%d, [out]
[18:11:38] <kurosu> a second for the input (like a memory adress or something)
[18:11:52] <kurosu> then last line to what is used by the code
[18:11:55] <kurosu> for instance:
[18:11:56] <kurosu> : "%"REG_c, "memory"
[18:12:18] <kurosu> meaning it will use ecx/rcx and memory
[18:12:29] <BBB> what if I want 3 registries?
[18:12:43] <mru> kurosu: bad idea
[18:12:50] <BBB> "%" REG_a, "%"REG_b, "%"REG_c, "memory"?
[18:12:54] <mru> declare ints and request them as output regs
[18:13:16] <mru> asm ("stuff" : "=&r"(tmp1), "=&r"(tmp2) ...)
[18:13:57] * peloverde is amused by users patching out avcore
[18:14:48] <kurosu> depends on if he actually wants something meaningful inside at the end of the block, or is just using them
[18:14:51] <kurosu> but whatever
[18:14:56] <kshishkov> peloverde: aren't we all trying to patch it out?
[18:15:15] <mru> kurosu: you should never grab a named register like that
[18:15:29] <BBB> mru: dsputil does it all over the place...
[18:15:36] <mru> BBB: so it should be fixed
[18:15:57] <mru> appeal to past mistakes doesn't work on me
[18:17:48] <BBB> hey, I'm learning dude :-p
[18:17:54] <BBB> I'm taking examples around me ;)
[18:18:31] <mru> first you must learn to tell good examples from bad ones
[18:18:33] <BBB> how do I name word-size registries again?
[18:18:39] <mru> registers
[18:18:44] <BBB> REG_a appears to use 64-bit regs
[18:18:46] <mru> registries is something else
[18:18:56] <peloverde> HKEY_FFMPEG?
[18:19:12] <mru> peloverde: that's not the only form of registry
[18:19:17] <mru> but probably the worst
[18:19:26] * kshishkov reminds mru of gconf
[18:19:37] <BBB> maybe just word "mov [in], %%"REG_a"\n\t"?
[18:19:57] <BBB> I mean "mov word [in], %%"REG_a"\n\t"
[18:19:57] <mru> BBB: why are you messing with explicit named regs?
[18:20:13] <BBB> because I have no idea what I'm doing ? :-p
[18:20:23] <BBB> maybe I should just do this in yasm
[18:20:36] <BBB> at least then I know what I'm doing
[18:20:49] <mru> most things should be using yasm
[18:21:02] <mru> only fragments too small to be a separate function should use inline asm
[18:22:01] <BBB> I'm adding minor things to h264's simd set... it's all inline asm, I'm trying to stay in style with that
[18:22:14] <BBB> if you think I should just write all new code, even minor additions, in yasm, that's fine with me also
[18:23:08] <peloverde> sometimes things are so limited precisely because they are inline
[18:23:38] <BBB> I guess...
[18:23:40] <mru> I see no problem with adding yasm to a mostly-inline collection
[18:23:56] <BBB> ok, yasm it is then
[19:25:04] <BBB> hmm... so I unzip a zip file that contains an installer... now I have an exe that complains my cpu is not recognized
[19:25:26] <BBB> unzip/7zip don't recognize the exe as an archive
[19:25:30] <BBB> strings gives me nothing
[19:25:37] <BBB> how do I figure out what's in the exe/installer?
[19:25:48] <kshishkov> maybe it's not zip
[19:26:09] <kshishkov> usually segment names contain some hints
[19:26:34] <kshishkov> most popular are M$ installer (cabextract can unpack it)
[19:26:51] <kshishkov> InstallShield (unshield tool from synce project can unpack it)
[19:27:37] <kshishkov> NSIS and WISE (fat luck, windows-only unpackers)
[19:27:41] <BBB> it's installshield wizzard
[19:27:48] <BBB> I got wine, so that's ok
[19:27:56] <kshishkov> drunkard
[19:29:32] <roxfan> inno!
[19:29:49] <kshishkov> yes, that beast too
[19:32:11] <BBB> cabextract and unshield od not recognize the file
[19:33:18] <BBB> I'll try on a windows machine maybe
[19:33:36] <kshishkov> probably you should really follow Carl Eugen's advise and make FFmpedia sponsor installation extractor development
[19:49:42] <Compn> BBB : i use windows 'uniextract' program to extract exe installers
[19:49:48] <Compn> but it doesnt work on them all...
[19:50:00] <kshishkov> nothing does
[19:50:11] <kshishkov> (sometimes including original installers)
[19:50:42] <Compn> original installers are the pains
[20:00:28] <BBB> Compn: will try, thanks for the suggestion
[20:05:33] <BBB> woohoo
[20:05:42] <BBB> uniextract worked, I think
[20:05:49] <BBB> more compression mess
[20:05:52] <BBB> this is easy though
[20:05:54] <BBB> a msi file
[20:05:57] <BBB> containing a cab
[20:06:01] <BBB> containing actual data files
[20:06:03] <BBB> what a mess
[20:13:11] <BBB> unfortunately negative :(
[20:17:35] <roxfan> msis usually support /a switch which usually extracts all the important stuff
[20:35:15] <BBB> 7zip can handle msis
[20:35:20] <BBB> 7zip could also handle the cab
[20:35:22] <BBB> 7zip is amazing
[20:37:20] <roxfan> yeah but sometimes filenames in cabs are different from final ones
[20:37:37] <roxfan> msiexec /a takes care of that
[20:44:52] <peloverde> BBB: I don't see any yasm in that latest patch
[20:45:30] <BBB> oops
[20:45:32] <BBB> old version
[20:45:35] <BBB> forgot to refresh
[20:46:09] <BBB> fixed
[20:53:11] <roxfan> BBB: that firmware image has inside an inno installer exe which installs an "IMM4 codec"
[20:54:01] <BBB> roxfan: I don't have the fw image handy atm, but are there any ffmpeg symbols in the package?
[20:54:25] <roxfan> haven't seen any so far
[20:54:44] <BBB> (strings .. | grep -i {avc,ff,x264,h264,xvid} should give you something if there's any)
[20:55:39] <roxfan> nope
[20:57:45] <roxfan> hmm there's another exe
[20:57:52] <roxfan> this one's packed
[21:12:20] <BBB> roxfan: packed = compressed? yeah, I'm having those problems also, many of their binaries are 10-fold-compressed
[21:12:42] <roxfan> packed with some exe packer
[21:13:46] <BBB> can you send it to me?
[21:14:25] <roxfan> dcc?
[21:14:37] <BBB> sure
[21:15:12] <roxfan> sending...
[21:15:20] <BBB> it might be blocked here
[21:15:20] <BBB> ...
[21:15:37] <BBB> can you email it to me privatly as an attachment?
[21:15:55] <BBB> 4.pe is a weird name for an exe
[21:16:07] <roxfan> well, i don't know the actual name
[21:16:23] <roxfan> i just found raw data in the fw binary
[21:16:24] <BBB> ok, email it (rsbultje () gmailcom)
[21:16:45] <roxfan> gmail doesn't like exes
[21:17:32] <BBB> can you privmsg that link again?
[21:17:41] <BBB> my irc client decided to put notices in my systray
[21:17:44] <BBB> which I can't c/p :-p
[21:17:48] <BBB> thanks though, I appreciate this ;)
[21:22:25] <CIA-93> ffmpeg: aurel * r24840 /trunk/libavformat/ (avio.c avio.h avformat.h):
[21:22:25] <CIA-93> ffmpeg: add FF_API_REGISTER_PROTOCOL define to disable the deprecated
[21:22:25] <CIA-93> ffmpeg: register_protocol() function
[21:29:27] <BBB> roxfan: wine can run it and then unpack into a tempdir
[21:29:38] <CIA-93> ffmpeg: aurel * r24841 /trunk/libavformat/ (aviobuf.c avio.h avformat.h):
[21:29:38] <CIA-93> ffmpeg: add FF_API_URL_RESETBUF define to disable the deprecated url_resetbuf()
[21:29:38] <CIA-93> ffmpeg: public function
[21:29:41] <BBB> roxfan: I see the same symbols as in the other binary, but nothing ffmpeg-like
[21:29:49] <BBB> (other binary = the one I already analyzed)
[21:29:51] <roxfan> looks like it just installs stuff from the zip attached to the end
[21:29:56] <roxfan> some kind of player
[21:30:00] <BBB> yes
[21:30:03] <BBB> BackupPlayer.exe
[21:31:57] <BBB> the odd thing is that it gives me an error
[21:32:03] <BBB> but the error string is not in the binary
[21:32:23] <BBB> as if part of the binary is masked/encrypted or so
[21:32:47] <roxfan> the player gives an error?
[21:32:54] <BBB> here it does, yes
[21:33:04] <roxfan> check for unicode strings
[21:33:38] <BBB> "video overlay wasn't initialized So this program will shutdown. Please retry after system reset"
[21:33:43] <roxfan> hmm
[21:33:51] <BBB> how do I grep for unicode strings?
[21:33:57] <roxfan> no idea
[21:34:01] <roxfan> but it's there
[21:34:03] <roxfan> Video Overlay wasn't initialized
[21:34:18] <BBB> unicode?
[21:34:23] <roxfan> yep
[21:34:26] <BBB> hm...
[21:34:44] <roxfan> some interesting stuff... Verify Watermark Verification Successful Verification Failed
[21:36:17] <roxfan> some mmx code...
[21:36:35] <BBB> oh you're disassembling it?
[21:36:37] <BBB> :-p
[21:36:43] <BBB> I'm just going through it in hexedit and grep
[21:36:48] <BBB> maybe I should try your approach
[22:00:23] <CIA-93> libswscale: ramiro * r31982 /trunk/libswscale/swscale-test.c: swscale-test: add CRC output
[22:03:56] <roxfan> i think i figured out all parts of the firmware file
[22:04:19] <roxfan> besides the two exes the only code seems to be two dsp code blobs
[22:08:11] <kierank> roxfan: what file are you trying to disassemble
[22:08:31] <roxfan> some pvr(?) firmware BBB linked yesterday
[22:09:18] <kierank> http://www.engadget.com/2010/08/09/ipad-gets-xvid-avi-video-playback-with-cinexplayer-app-on-itunes/ --> likely a ffmpeg violation?
[22:10:32] <Dark_Shikari> prolly
[22:10:34] <Dark_Shikari> or xvid itself
[22:10:37] <Dark_Shikari> which would be gpl
[22:10:46] <Compn> heh
[22:11:03] <Compn> kshishkov : i got the MSA1 binary codec working. yet another microsoft screen codec
[22:11:07] <roxfan> "NXP Software B.V" o.O
[22:13:09] <roxfan> so could be using a commercial lib i guess
[22:14:50] <twnqx> NXP is philips, no?
[22:14:59] <twnqx> well, the semiconductor daughter
[22:15:05] <roxfan> yes
[22:15:59] <roxfan> http://www.software.nxp.com/?pageid=152
[22:22:00] <BBB> roxfan: is that the ipad app or the firmware?
[22:22:12] <roxfan> cinexplayer
[22:22:17] <BBB> ah, ok
[22:22:30] <roxfan> the firmware looks custom made
[22:23:42] <BBB> yeah that's my thinking also
[22:23:50] <BBB> note how it only supports avi files
[22:24:31] <roxfan> "DVR system AVI builder on DSP."
[22:24:47] <roxfan> hmm
[22:24:48] <roxfan> DATE STAMP [Jul  8 2003] [14:48:37]
[22:24:54] <roxfan> that's pretty old
[22:25:07] <BBB> it is


More information about the FFmpeg-devel-irc mailing list