**** BEGIN LOGGING AT Mon May 09 02:59:59 2016 May 09 07:37:02 Linkandzelda: you do have the service manuals? :) May 09 08:34:07 my new sim has just arrived at home May 09 08:34:12 i'm at work, though May 09 08:34:28 and i've slept quite bad today, so maybe i won't start again with yappari until tomorrow May 09 08:34:33 but at least i've got the tools now :) May 09 08:49:23 you prolla should have gotten 2 dozen, no? May 09 08:49:32 prolly* May 09 08:50:34 wait... 3 dozen, so statistically you always have a number that just got unlocked after 60 days May 09 11:04:24 let's hope i only need this one :) May 09 11:05:29 :) May 09 11:07:52 i dont use crapsapp so i can lend you my number too May 09 11:08:20 hi May 09 11:08:21 and 2 numbers of my gf May 09 11:08:32 i've tried to convince facebook that my mobile number is the same as their amsterdam offices May 09 11:08:37 and there are starters for ~0.7eur May 09 11:08:41 thusfar they've been refusing it :P May 09 11:12:35 lol buzz May 09 11:12:53 i'll get them to do it ;) May 09 11:13:09 maybe i can just put a random recruiter thats hunting for FB employees in it May 09 11:33:48 So yeah in reference to the discussion in the channel a few days ago, no, I am not interested in reverse engineering WhatsApp or anything else to do with Facebook :) May 09 11:34:22 Although it sounds like progress is being made and reverse engineering is not needed right now May 09 11:43:28 Oh and in regards to whoever it was that was asking about GPRS before, the GPRS stuff is handled by the cellular services daemon and csd-gprs. The dbus interface to this module is actually documented (likely unintentionally though) May 09 11:43:40 The thing that talks to it is libicd-network-gprs May 09 11:47:51 * jonwil wonders if there are still parts of the various cellular standards that are not available to the public (I know the GSM encryption used to be top secret May 09 11:49:56 but then someone figured it out and its not secret anymore May 09 11:51:54 @kerio: ping May 09 11:54:49 jonwil: sup May 09 12:02:40 So yeah about openssl in maemo and what you said the other day (reading from chanlogs) it looks like there are less dependencies on openssl than it looks at first glance May 09 12:03:19 and especially less closed source dependencies May 09 13:04:24 new number registered May 09 13:05:04 :) May 09 13:05:05 nice May 09 13:05:52 let's see what happens now May 09 13:07:45 Segmentation fault May 09 13:07:47 well May 09 13:07:55 that was unexpected May 09 13:07:59 i didn't do anything May 09 13:17:35 it segfaults when i try to send a message to someone May 09 13:17:43 cipher() method from axolotl is segfaulting May 09 13:17:54 there's a lot of work here :( May 09 13:18:22 i don't quite understand how can this even happen. libwa was supposedly working for coderus on jolla May 09 13:19:37 LUL May 09 13:19:42 ceene: is it using libcrypto May 09 13:23:08 libaxolotl.pro:PKGCONFIG += openssl libssl libcrypto May 09 13:23:12 somewhere, yes May 09 13:23:37 someone hasn't been checking their NULLs May 09 13:23:50 i guess so May 09 13:23:58 libaxolotl.pro ? is that for axoloti? May 09 13:24:02 hmm, no that cant be May 09 13:24:42 buZz: it's libaxolotl, now called simply Signal May 09 13:24:46 implementation of textsecure May 09 13:24:49 ah ok May 09 13:24:51 i don't know why the salad of names May 09 13:25:01 creativity :P May 09 13:26:14 i don't know if i'm gonna be able to solve this May 09 13:26:20 even if i solve this particular problem May 09 13:26:27 what about all the future ones i don't even know about? May 09 13:26:27 :( May 09 13:26:35 :( May 09 13:26:54 coderus has gone completely silent May 09 13:41:09 Its a pitty no-one is still working on whatsapp plugin for libpurple, if that was still maintained then we could just compile libpurple on maemo and use that (which gets us all the other libpurple plugins too I guess) May 09 13:43:04 not that I care about Whatsapp or anything else connected to Facebook :P May 09 13:43:25 problem with purple is not whatsapp plugin i think May 09 13:43:35 but purple versions and libc compatibility May 09 13:43:37 or something like that May 09 13:43:47 also, maemo messaging interface is awful May 09 14:17:52 next step is gonna make axolotl and its children compile statically May 09 14:17:56 easier debugging May 09 14:18:05 and i guess final .deb package should have them statically compiled May 09 14:18:15 no sense on distributing 3 .deb files for just one application May 09 14:18:24 and then maybe gdb step by step May 09 14:18:26 :/ May 09 14:19:41 kerio: http://talk.maemo.org/showthread.php?t=96288 May 09 14:29:46 jonwil: \o/ May 09 14:29:49 who da man May 09 14:29:50 you da man May 09 14:30:30 jonwil: have you considered keeping openssl 0.9.8 instead? May 09 14:30:52 i don't think ABI shims are going to work May 09 14:30:59 I never suggested abi shims May 09 14:31:03 0x42c1d7a8 in AES_encrypt () from /usr/lib/libcrypto.so.0.9.8 May 09 14:31:22 by the way... that's the origin of this segfault May 09 14:31:24 I suggested we port newest openssl/whatever to Maemo and recompile the things that make sense May 09 14:31:33 The things that we can recompile rather May 09 14:31:37 jonwil: what do you mean by "For the packages that link to libssl/libcrypto but dont actually call any functions in them" May 09 14:31:45 you mean packages that have them as a dependency? May 09 14:32:04 remember that dlopen is fucking weird May 09 14:32:12 and sometimes you might need to link in openssl to dlopen() a so that needs openssl May 09 14:32:27 oh nvm you also checked the .sos May 09 14:32:34 Yeah I checked May 09 14:32:38 qt dlopens ssl, for example, and is ready to use any upper version May 09 14:32:49 as it searches for version numbers May 09 14:33:03 QT is one of the weird ones yea May 09 14:33:13 i have a not-yet-ready patch for qt to use newer openssl and use latest tls as default May 09 14:33:26 These particular binaries link to libssl/libcrypto but have no functions in their import tables from either library May 09 14:33:39 jonwil: for the binary stuff we probably want to keep openssl 0.9.8 around tho May 09 14:33:44 jonwil: I just wish you'd use a tad more formatting on your posts in http://talk.maemo.org/showthread.php?t=96288 May 09 14:33:44 Yes we do May 09 14:33:45 maybe with the RHEL patches? May 09 14:33:48 We want to keep both May 09 14:33:59 We can keep both May 09 14:34:04 hell, let's just use RHEL's openssl May 09 14:34:09 what's the worst that can happen May 09 14:34:27 We have the newest openssl and recompile everything that we can recompile so it uses that May 09 14:34:32 redhat sneaking into your dreams? May 09 14:34:34 ye ye May 09 14:34:37 and we keep openssl 0.9.8 for the things we cant recompile May 09 14:34:44 DocScrutinizer05: RHEL 5 or something stupid like that has openssl 0.9.8 May 09 14:34:49 and it's LTE until 2020 or so May 09 14:34:54 LTS May 09 14:35:09 (fucking 4g lol) May 09 14:35:57 I suspect most of the closed-source things calling libssl/libcrypto are low-risk anyway May 09 14:36:09 that's how the NSA sneaks in! May 09 14:37:18 osso-backup for example only needs libssl for AES and for SHA1 May 09 14:37:18 [2016-05-09 Mon 16:36:21] Centurion_Dan: does devuan have a armel build now? May 09 14:37:19 [2016-05-09 Mon 16:36:39] yes... but our current buildhosts are offline... May 09 14:37:24 jonwil: do you know if AES_encrypt from libcrypto.so.0.9.8 may be buggy? May 09 14:37:33 is there a centos 5 armel May 09 14:37:40 DocScrutinizer05: \o/ May 09 14:37:54 am i a bad enough dude to switch my sheevaplug to devuan May 09 14:38:02 I know nothing about possible issues in openssl May 09 14:38:24 ceene: check man pages from different versions of openssl for AES_encrypt May 09 14:38:29 see if anything changed May 09 14:39:29 DocScrutinizer05: should i switch to devuan May 09 14:39:55 if builds were reproducible, i bet the binaries i have would be bit-by-bit the same May 09 14:40:03 no systemd, no udev May 09 14:41:28 of course none of this helps in any way with the really big problem, namely how to deal with microb and nss May 09 14:42:03 forget about microb and find a replacement browser May 09 14:42:03 something I haven't a clue about May 09 14:42:05 such as midori May 09 14:42:20 which under easy debian works wonderfully May 09 14:42:32 jonwil: well, that's a completely orthogonal issue May 09 14:43:15 in the grand scheme of things, the number of packages using openssl to talk to the internet at large is very small and the number of closed packages doing that is even smaller. May 09 14:43:54 I always thought we could upgrade the browserd under microB all we want, since it's a generic rendering engine and completely FOSS May 09 14:44:16 DocScrutinizer05: yes, but that would require someone who understands how the bloody thing works May 09 14:44:27 timeless: ^^^ May 09 14:44:47 The biggest problems with upgrading microb are:1.Finding an engine May 09 14:45:38 1.Finding an engine to use that is suitable and that can be ported to the Neo900 May 09 14:45:57 2.Dealing with Flash Player (which is fairly well coupled to microb-engine/Gecko) May 09 14:46:38 and mostly useless nowadays anyway? May 09 14:47:03 3.Dealing with any specific functionality in microb that the new engine needs to handle (for example on microb, I can click on a phone number link on certain web pages and get a dialer window ready to dial) May 09 14:47:26 that's mimetypes May 09 14:47:39 no, its special URL formats May 09 14:47:45 tel:123456 or something May 09 14:48:00 well and that is handled in mimetypes afaik May 09 14:48:09 i shall insist on midori once again :P May 09 14:48:11 probably May 09 14:48:18 not that i'm associated with midori in any way May 09 14:48:35 ceene: you can get midori any time you want May 09 14:49:02 just like Opera and FF and dunno what May 09 14:49:12 i know May 09 14:49:39 however none of those will replace browserd May 09 14:49:42 what if we pay mozilla May 09 14:49:44 don't know why May 09 14:50:34 jonwil: fuck flash player May 09 14:50:39 we don't need that anymore, thanks to apple May 09 14:50:44 Yeah I say "screw Flash Player" as well May 09 14:50:51 any website has to have a non-flash version May 09 14:50:52 but there are still plenty of sites out there that need it May 09 14:50:53 [2016-05-09 Mon 16:40:11] DocScrutinizer05: there you go, you have your maemo upstream :) May 09 14:50:56 not really May 09 14:50:56 for whatever version May 09 14:51:11 jonwil: have you tried switching your user agent to safari/ipad May 09 14:51:54 Not talking about my N900 (where I have flash player turned off because its a resource hog) but on my desktop plenty of sites still use Flash unfortunately May 09 14:52:19 But yeah I am sure if we said to CSSU users "flash player wont work anymore" there would be complaints from people who do need it May 09 14:52:26 I noticed they do only until you unistall flash May 09 14:53:15 but yeah, ack for complaints about compatibility in CSSU May 09 14:53:47 for sure needs to be an opt-in package in CSSU May 09 14:54:55 Plenty of news websites still use Flash to deliver video clips for example (and rely on dedicated "apps" on mobile to avoid the "no flash on mobile browsers" problem) May 09 14:55:13 DocScrutinizer05: a handful of porn sites^W^Wsites require you to switch to an iphoney user agent string May 09 14:55:15 https://botbot.me/freenode/devuan/2016-05-09/?msg=65725085&page=2 May 09 14:55:30 jonwil: switch your user agent and use a modern browser May 09 14:57:11 jonwil: as mentioned above I never ran into problems with HTML5 instead of flash when I simply uninstalled flashlibs... until some embedded systems insist in using flash in their HTML GUI May 09 14:57:22 :-/ May 09 14:57:26 DocScrutinizer05: do you reckon that the community could pay for someone to update microb to a (more) recent gecko+nss? May 09 14:57:44 I doubt that May 09 14:58:01 it's pretty hard to find affordale devels for such stuff May 09 14:58:09 affordable May 09 14:58:46 actually it's hard to find any devels at all May 09 15:02:36 Ok so mainstream sites like twitter seem to serve up HTML audio/video these days May 09 15:03:10 but I can assure you that no matter what user-agent you use or what browser you have or whether you have Flash installed, a site like CouchTuner will not serve up anything other than Flash May 09 15:04:38 then again, no-one sane is going to be using a site like that on the tiny screen and low-bandwidth network connection of a N900 :) May 09 15:07:13 uhm May 09 15:07:18 can somebody explain me this? May 09 15:07:27 (gdb) n May 09 15:07:27 282 if (version >= 3) { May 09 15:07:27 (gdb) p version May 09 15:07:27 $6 = 3 May 09 15:07:27 (gdb) s May 09 15:07:29 [LWP 2803 exited] May 09 15:07:32 Breakpoint 1, SessionCipher::getCiphertext (this=, version=3, messageKeys=..., plaintext=...) at src/sessioncipher.cpp:303 May 09 15:07:35 303 qDebug() << "IV:" << iv.constData(); May 09 15:07:42 instead of getting inside the version>=3 May 09 15:07:48 this thing has jumped to the } else { May 09 15:08:01 well, further than that in fact! May 09 15:08:30 me, I dont care about Flash on my N900, I just want an up-to-date browser engine that can run sites like Slack and Trello and can pass the SSLLabs browser test May 09 15:15:51 jonwil: i can assure you, i don't have flash player installed May 09 15:15:53 http://i.imgur.com/j2UCvV3.jpg May 09 15:16:02 have you considered that maybe your browser is just shit May 09 15:16:42 (that's with the default user agent string, mind you) May 09 15:18:41 i mean May 09 15:38:03 ceene: is the binary optimized? May 09 15:39:58 ceene: i see you've seen the power of the static side.. muahahahaha May 09 16:02:58 -Obla? May 09 16:03:21 yes, that could cause such funny effects May 09 16:03:33 -Obladi -Oblada May 09 16:04:57 particularly on ARM where you think you are wazching the completely wrong binary with your Lauterbach May 09 16:05:35 the source often has no faintest resemblance with what you think should be there, in debugger May 09 16:05:59 DocScrutinizer05: yeah, -02 usually results in impossibility to debug the binary in a sane way :) May 09 16:06:05 -Og May 09 16:06:27 for when you feel gangsta May 09 16:06:46 for debugging, -O0 is a must, with --no-skip-frame-pointers (or whatever the flag is) May 09 16:06:54 hrm May 09 16:07:08 :nod: May 09 16:07:14 surely -fno-omit-frame-pointer is included in -O0 May 09 16:07:17 who the hell omits that anyway May 09 16:07:30 kerio: you don't really need that for release May 09 16:07:40 n8x0 cpu instruction set should be subset of a8/a7 right? May 09 16:07:41 it's going to be a while before we can collectively heal the damage that gentoo has caused ._. May 09 16:08:00 KotCzarny: afaik, yes May 09 16:08:08 like, don't get me wrong May 09 16:08:21 freemangordon: funnything i get: May 09 16:08:21 [I have no name!@n8x0chr 16:06:15 /]# /usr/sbin/iconvconfig May 09 16:08:22 Illegal instruction May 09 16:08:23 it's fun to compile something like openssl with -O3 -fomit-frame-pointers -funroll-all-loops May 09 16:08:24 to run the benchmarks May 09 16:08:40 unroll loops BWAHAHAHA May 09 16:08:40 kerio: you're kidding by blaming this on gentoo, right? May 09 16:09:11 Wizzup: maybe :> May 09 16:09:33 KotCzarny: and? what cou was that compiled for? May 09 16:09:33 can't spell -funroll-loops without FUN May 09 16:09:47 *CPU May 09 16:10:29 freemangordon: dont remember exact cpu of n8x0 but its a6 i think May 09 16:10:56 hmm, I meant - what CPU was /usr/sbin/iconvconfig compiled for? May 09 16:11:17 freemangordon: packages were from n8x0 repos and n8x0 sdk repo May 09 16:11:17 KotCzarny: OMAP2 May 09 16:11:28 at least, n800 May 09 16:11:30 so definitely not miscompiled May 09 16:11:38 buZz: this is arm6, correct? May 09 16:11:43 for (long i = MAXLONG; i > 0; i--) --- unroll DAT!! sucker! May 09 16:11:44 no, its armv6 May 09 16:12:01 yah, that's what I meant May 09 16:12:10 oh ok, cause arm6 is something else May 09 16:12:18 :) May 09 16:12:43 its actually arm9 ;) May 09 16:12:47 oh eh, arm11 May 09 16:12:51 :) May 09 16:12:58 OMAP1 is arm9 May 09 16:12:59 DocScrutinizer05: what is that loop doing May 09 16:13:00 KotCzarny: it has some alignment requirements, iirc May 09 16:13:09 whatever you like May 09 16:13:23 i can't imagine 2**64 different things to do ._. May 09 16:13:25 freemangordon: so even with subset there still might be problem running older binaries on a7? May 09 16:13:29 it's a stub to be finished by you ;-) May 09 16:13:43 KotCzarny: what are you trying to do? May 09 16:13:47 because other things run fine (ie. did apt-get update, ran bash etc) May 09 16:13:58 fmg: chroot for compiling packages for n8x0 May 09 16:14:01 could it be thumb-compiled? May 09 16:14:23 fmg: my a7 kernel has thumb enabled May 09 16:14:47 KotCzarny: so, you do chroot on n900? May 09 16:15:00 fmg: nope, on banana pi m1 May 09 16:15:04 aah May 09 16:15:08 and compile things for n900 in chroot May 09 16:15:09 arm11? May 09 16:15:27 and it works fine, and now i wanted to do the same for n8x0 May 09 16:15:33 * DocScrutinizer05 idly wonders if and/or why not there's a parameter to -funroll-all-loops to define the max loop size May 09 16:15:37 got it May 09 16:15:50 not that iconvconfig is essential, but wondered if there are some binary gotchas May 09 16:16:22 but you rather do that in sourcecode with metainstructions anyway May 09 16:16:46 global unroll is an idiocy May 09 16:17:41 KotCzarny: do readelf -A on the binary May 09 16:17:52 and compare it with some working binary May 09 16:17:54 bash: readelf: command not found May 09 16:18:00 thich package has it? May 09 16:18:01 binutils? May 09 16:18:08 DocScrutinizer05: clang: warning: optimization flag '-funroll-all-loops' is not supported May 09 16:18:09 ;-; May 09 16:18:11 do this in scratchbox May 09 16:18:18 hehe ok May 09 16:18:32 fmg: no x86 machine at the moment May 09 16:18:38 yes, binutils May 09 16:18:38 good boy clang May 09 16:18:52 honestly tho May 09 16:19:02 uint64_t x; for (int64_t i = INT64_MAX; i > 0; i--) x = i; May 09 16:19:09 why isn't it optimized to x = 0; May 09 16:19:18 or x = 1, i can't think at the moment May 09 16:19:37 Attribute Section: aeabi May 09 16:19:37 File Attributes May 09 16:19:47 KotCzarny: not here May 09 16:19:48 to teach you keep your brain in training? May 09 16:19:58 KotCzarny: use pastebin or something May 09 16:20:01 why do i have to think when compilers can do that for me May 09 16:20:04 same for /bin/bash May 09 16:20:13 no other output May 09 16:20:20 KotCzarny: weird May 09 16:20:25 I think such stuff gets optinized out at -O2 the latest May 09 16:20:56 and bash works May 09 16:21:54 since you don't use x as source of values (on 'right side of =') inside the loop May 09 16:22:16 http://pastebin.com/raw/dCbri4gG May 09 16:22:26 which actually sometimes sucks when compiler doesn't get that x is mapped to a IO or whatever May 09 16:22:46 KotCzarny: can you upload those 2 binaries so I can test with "my" readelf? May 09 16:22:59 sure, just a moment May 09 16:23:39 isn't readelf a messybox plugin? May 09 16:24:55 DocScrutinizer05: it least in SB it is not May 09 16:25:09 messybox-bauer May 09 16:25:18 ;-) May 09 16:26:00 wouldn't be surprised May 09 16:26:16 readelf --version ? May 09 16:26:23 https://transfer.sh/6TB6g/iconvconfig May 09 16:26:45 https://transfer.sh/EWN9d/bash May 09 16:27:00 IroN900:~# readelf --version May 09 16:27:01 GNU readelf (GNU Binutils for Debian) 2.18.50.20070820 May 09 16:30:16 KotCzarny: both bash and iconvconfig work in SB, unfortunately my readelf gives the same output May 09 16:30:41 yes, i know they work because they are from official repos May 09 16:33:12 KotCzarny: well, start it in gdb, to see which instruction is problematic May 09 16:34:54 http://paste.opensuse.org/45708992 FWIW May 09 16:35:24 fmg: how to run it in gdb? May 09 16:35:38 Program received signal SIGILL, Illegal instruction. May 09 16:35:38 0x0000b8d0 in ?? () May 09 16:35:50 disassemble May 09 16:36:02 No function contains program counter for selected frame. May 09 16:36:06 yeah, sure May 09 16:36:13 freemangordon: what means >>Tag_THUMB_ISA_use: Thumb-1<< ? May 09 16:37:05 KotCzarny: disassemble 0x0000b8d0 * or somesuch May 09 16:37:25 (gdb) disassemble 0x0000b8d0 May 09 16:37:25 No function contains specified address. May 09 16:37:52 "disassemble * 0x0000b8d0" May 09 16:37:53 * DocScrutinizer05 also glares at >>Tag_ABI_HardFP_use: SP and DP<< on an armel platform - though not really having the faintest clue May 09 16:38:04 A syntax error in expression, near `'. May 09 16:38:13 without the quotes May 09 16:38:18 was without May 09 16:38:29 just a second May 09 16:39:00 (gdb) disassemble * 0 May 09 16:39:00 Cannot access memory at address 0x0 May 09 16:39:02 * DocScrutinizer05 afk o/ <3 May 09 16:39:08 but did you see paste of strace? May 09 16:39:13 http://pastebin.com/raw/dCbri4gG May 09 16:39:42 hmm May 09 16:39:53 as a side note: GNU gdb 6.8-debian May 09 16:40:45 KotCzarny: anyway, disasm should work, it is 7.2 here, but should not make difference May 09 16:41:01 syntax probably changed 6 -> 7 May 09 16:41:05 because it doesnt May 09 16:41:20 and according to strace: May 09 16:41:20 cacheflush(0xbeb20a68, 0xbeb20a78, 0, 0x98e8, 0xbeb20a68) = -1095628184 May 09 16:41:20 --- SIGILL (Illegal instruction) @ 0 (0) --- May 09 16:41:23 I doubt, is stock maemo is 6. May 09 16:41:38 stock maemo in 2008 May 09 16:42:09 KotCzarny: comeon, don't tell me the debugger cannot dump the instructions May 09 16:42:23 you saw it, bad syntax May 09 16:42:41 command is available, just missing magic incantation May 09 16:43:14 (gdb) disas 0x401050 0x401075 May 09 16:43:24 according to some cygwin google May 09 16:44:01 disassemble 0x0000b8d0 0 May 09 16:44:01 Dump of assembler code from 0xb8d0 to 0x0: May 09 16:44:16 Dump of assembler code from 0xb8d0 to 0xb8d8: May 09 16:44:16 0x0000b8d0 : bx lr May 09 16:44:16 0x0000b8d4 : push {lr} ; (str lr, [sp, #-4]!) May 09 16:45:03 KotCzarny: hmm, yeah, I don;t think bx lr is avalable on arm11 May 09 16:45:20 um, but binaries were compiled for arm11 May 09 16:45:28 which library is that? May 09 16:45:32 and are run on a7 May 09 16:45:53 hmm? May 09 16:46:59 KotCzarny: this is in iconvconfig or in some .so? May 09 16:47:11 how can i tell? May 09 16:47:50 bt? May 09 16:49:04 (gdb) bt May 09 16:49:04 #0 0x0000b8d0 in ?? () May 09 16:49:54 funny how everything else works without a problem, unless that iconvconfig was miscompiled or just working magically over some bug May 09 16:50:47 KotCzarny: what is the output of "cat /proc/cpuinfo"? May 09 16:51:25 http://pastebin.com/raw/J5qseGLw May 09 16:54:23 and this is apparently cpuinfo of nokia n8x0 http://lists.openafs.org/pipermail/openafs-info/2007-December/028209.html May 09 16:57:00 brb May 09 16:57:47 KotCzarny: could you also dump the previous 2-3 instructions, when you're back? May 09 16:58:21 as maybe 0x0000b8d0 is the instruction AFTER the illegal one May 09 16:59:04 0x0000b8c8 : mov r2, #0 ; 0x0 May 09 16:59:04 0x0000b8cc : svc 0x009f0002 May 09 16:59:04 0x0000b8d0 : bx lr May 09 16:59:43 aaah May 09 16:59:46 here it is :D May 09 16:59:52 thumb? May 09 16:59:53 svc 0x009f0002 May 09 16:59:59 no, supervisor call May 09 17:00:07 krhrhr May 09 17:00:13 funny thing to be in iconvconfig May 09 17:00:19 yes May 09 17:00:41 but this is meant for n800, where there is Nokia BOOTROM May 09 17:00:48 k, brb (15 min) May 09 17:13:01 KotCzarny: here is your bugger http://osxr.org:8080/glibc/source/ports/sysdeps/unix/sysv/linux/arm/sysdep.h?v=glibc-2.17#0333 May 09 17:17:04 looks nasty May 09 17:18:03 DocScrutinizer05: well, what "svc 0x009f0002" does, is to call supervisor (or kernel, not sure) to flush the cache May 09 17:18:46 and that doesn't seem related to Nokia May 09 17:18:53 or n800 in that regard May 09 17:19:30 yeah, oh well it's glibc, such stuff expected to show up there I guess. *somewhere* it needs to get done May 09 17:20:34 flush cache is sth that's not exactly trivial I guess May 09 17:20:46 KotCzarny: check CONFIG_OABI_COMPAT is enabled in your kernel :) May 09 17:21:04 or rather - make sure CONFIG_OABI_COMPAT is enabled May 09 17:39:54