**** BEGIN LOGGING AT Tue Jul 07 03:00:06 2009 **** BEGIN LOGGING AT Tue Jul 07 04:18:13 2009 Jul 07 05:10:17 Tartarus: libunwind-exceptions did not work on arm Jul 07 05:10:30 Tartarus: and they are not in initial because it does not matter there Jul 07 05:19:11 moreover I think we do not ship libunwind Jul 07 06:27:32 Goodmorning Jul 07 06:28:47 I currently have to use about 2 or 3 times the space on my harddisk for OpenEmbedded as about 1 or 2 months ago, is it something I missed that causes this? Jul 07 08:52:29 morning folks Jul 07 09:00:06 yo BlindMan Jul 07 09:09:11 morning Jul 07 09:10:04 hi blindman Jul 07 09:10:14 good morning Jul 07 09:10:18 florian: good morning Jul 07 09:58:53 gm Jul 07 10:03:26 hey Crofton|work Jul 07 10:05:21 hi Crofton|work Jul 07 10:18:32 how can I specify that bitbake should use toolchain X for an image except that package N should use toolchain Y? Jul 07 10:18:42 morning all Jul 07 10:18:49 hey RP Jul 07 10:18:50 hello Jul 07 10:19:00 RP, greetings Jul 07 10:19:37 I tried following this guide http://wiki.openmoko.org/wiki/User:CesarB/Compiling_for_EABI as I have exactly the same problem but that seems to be a bit old Jul 07 10:26:39 hello, has anybody experienced problems enabling OMAP MULTIPLEXING in linux kernel? Jul 07 10:26:57 If I enable it I got a kernel panic (cannot mount rootfs)? Jul 07 10:34:55 Viltapi: OE supports EABI all the way through for most devices Jul 07 10:43:53 XorA: I need to build U-boot 1.3.4 and that causes and EABI version error "arm-angstrom-linux-gnueabi-ld: ERROR: Source object .../build/tmp/cross/arm/lib/gcc/arm-angstrom-linux-gnueabi/4.3.3/libgcc.a(_udivdi3.o) has EABI version 4, but target u-boot has EABI version 0" Jul 07 10:44:40 Viltapi: 1.3.4? thats quiet ancient isnt it? Jul 07 10:45:24 XorA: hehe, very probably Jul 07 10:46:00 but i'm trying to build for EA3131 and it's been supplied with an u-boot patch for version 1.3.4 Jul 07 10:46:10 http://www.embeddedartists.com/products/kits/lpc3131_kit.php Jul 07 10:48:39 Viltapi: I think the OE u-boot eabi patches were pretty generic, try applying them Jul 07 10:49:19 XorA: i tried two of them but didn't get them to work Jul 07 10:50:55 to me, the two (fix-arm920t-eabi.patch and uboot-eabi-fix-HACK.patch) seemed very machine-specific Jul 07 10:51:38 03ghost  07org.openembedded.dreambox * rc430ba2598 10openembedded.git/packages/dreambox/dreambox-dvb-modules.bb: Jul 07 10:51:38 dreambox-dvb-modules.bb: update dm800 and dm8000 drivers Jul 07 10:51:38 dm8000 changes: Jul 07 10:51:39 - fixed no more working ts playback after MPEG4 Part2 (Divx) playback Jul 07 10:51:39 - CI fix for Dre Crypt Module Jul 07 10:51:40 dm800 and dm8000 changes: Jul 07 10:51:44 - /proc/stb/audio/audio_delay_pcm/bitstream is now working in both directions.. negative offset now delays video Jul 07 10:51:46 03ghost  07org.openembedded.dreambox * r2173d9603d 10openembedded.git/packages/gstreamer/gst-plugin-dvbmediasink.bb: gst-plugin-dvbmediasink.bb: push srcdate Jul 07 10:51:49 03Felix Domke  07org.openembedded.dreambox * ra640ebe624 10openembedded.git/: Merge branch 'org.openembedded.dreambox' of git://git.opendreambox.org/git/ghost/opendreambox-1.5 into org.openembedded.dreambox Jul 07 10:52:22 Viltapi: ah crap your right, I forgot how it was organised Jul 07 10:53:49 i managed to build u-boot manually but when using OE, I would need a specific toolchain for it, or avoid the EABI version problem Jul 07 10:54:11 Viltapi: can the equivalent patches be ported to your arch? Jul 07 10:54:31 Viltapi: or does mainsteam u-boot now support your board? thats the two directions I would go in Jul 07 10:54:41 Viltapi: from experience Jul 07 10:55:34 i doubt the mainstream u-boot as that board is very new and the patch is even newer Jul 07 10:55:45 why it has been made for 1.3.4...no idea Jul 07 10:56:13 but i may have to consider porting the patches/hacks Jul 07 10:57:09 Viltapi: commercial stuff gets stuck in a timewarp normally, due to the work being started years before it ends Jul 07 10:57:57 XorA: that's true, but I understood that the patch was 3rd party Jul 07 10:58:27 at least the website says "resources contributed by customers and users" Jul 07 10:58:51 but that doesn't stop the customers from being commercial as well Jul 07 10:59:54 morning Jul 07 11:00:03 hello Jul 07 11:47:34 good afternoon Jul 07 12:00:38 would be possible to push the following patches please ? Jul 07 12:00:41 http://patchwork.openembedded.org/patch/808/ Jul 07 12:00:45 http://patchwork.openembedded.org/patch/809/ Jul 07 12:06:14 sure Jul 07 12:06:54 hi hrw Jul 07 12:07:58 hrw: what is the passage of being allowed to OE write access? is it enough to ask to ML? Jul 07 12:08:52 03Marco Cavallini  07org.openembedded.dev * r246e641c1a 10openembedded.git/conf/distro/ (3 files in 2 dirs): Jul 07 12:08:52 New conf/distro/kaeilos.conf with customized include files Jul 07 12:08:52 Signed-off-by: Marco Cavallini Jul 07 12:09:01 03Marco Cavallini  07org.openembedded.dev * ref1048dc46 10openembedded.git/recipes/psplash/files/ronetix-pm9263/psplash-default: Jul 07 12:09:01 Added recipes/psplash/files/ronetix-pm9263/psplash-default Jul 07 12:09:01 Signed-off-by: Marco Cavallini Jul 07 12:09:22 mckoan: it is 'show your patches, we will grant r/w if they ok' or someone suggests or person asks Jul 07 12:09:31 mckoan: ask - I will agree Jul 07 12:11:23 hrw: how do I 'show' my patches ? Jul 07 12:11:36 mckoan, you have been doing that alrady :) Jul 07 12:15:59 mckoan: you are listed as stable/2009 maintainer - do you plan to give some acks for [STABLE] patches? Jul 07 12:16:00 Crofton|work: ok ;-) Jul 07 12:16:50 hrw: I would be pleased to do that, we should discuss later about how doing that Jul 07 12:17:37 -> bb in 2hrs Jul 07 12:28:12 hi all Jul 07 12:30:21 hi Thomas Jul 07 12:40:20 morning Jul 07 12:52:40 hi Jul 07 12:52:47 hi woglinde Jul 07 12:52:53 hi florian Jul 07 12:53:05 hm oh vlc 1.0 is out Jul 07 12:53:18 hi henning, chouimat Jul 07 12:53:45 hi hrw Jul 07 12:54:39 florian: I need to check my phone - should have some LT photos Jul 07 12:55:02 hrw: oh... that would be good Jul 07 12:55:27 hm plam pre will be released in october in germany Jul 07 12:56:50 woglinde: how much €€€? Jul 07 12:56:52 re Crofton|work Jul 07 12:57:48 hrw, gm Jul 07 12:57:52 got back last night Jul 07 12:58:20 hrw hm only with a plan from o2 Jul 07 12:58:20 florian: http://marcin.juszkiewicz.com.pl/download/lt/ Jul 07 12:58:50 woglinde: likely to be expensive at least... Jul 07 12:58:57 florian: crappy quality (phone taken) Jul 07 12:59:09 hm price without plan is not setteld Jul 07 12:59:39 hrw: i'll take a look - many thanks Jul 07 13:00:48 hrw: are these the slides from the emlix talk? Jul 07 13:02:58 yes Jul 07 13:04:31 heh.... do i read "> 100 packages?" Jul 07 13:04:43 03Robert Schuster  07org.openembedded.dev * r0784d75461 10openembedded.git/recipes/antlr/antlr-native_2.7.7.bb: antlr-native 2.7.7: Use proper variable. Jul 07 13:04:43 03Robert Schuster  07org.openembedded.dev * r9e01be7741 10openembedded.git/recipes/classpath/ (5 files): classpath: Depend on antlr-native, increased resp. PRs. Jul 07 13:04:56 yes Jul 07 13:05:14 florian: and recently they added X11 Jul 07 13:05:16 ;D Jul 07 13:05:28 * florian is impressed ;) Jul 07 13:07:02 he you photograph my son Jul 07 13:07:33 I will sue you *g* Jul 07 13:11:33 today is the day of bad jokes right? ;) Jul 07 13:12:46 hrw: the top sales argument seems to be the ability to reproduce the same results for a long time. i guess ai'll write a short guide how to do this with oe :-) Jul 07 13:13:47 ;) Jul 07 13:15:21 florian the coffee machine was form buero2.0 Jul 07 13:15:24 not from tarent :) Jul 07 13:15:55 woglinde: sure? Jul 07 13:16:48 florian jepp Jul 07 13:16:56 I am 100% sure Jul 07 13:17:07 florian: hrm, have you been part of the e2factory talk or why do you say that? Jul 07 13:18:42 hm zecke is working on moc Jul 07 13:19:02 BlindMan: no i didn't manage to attend unluckily. Jul 07 13:19:46 BlindMan: I just took a look at the slides on the photos and the website. Jul 07 13:20:27 okay I have to leave Jul 07 13:20:30 bye all Jul 07 13:35:13 bye woglinde Jul 07 13:36:19 BlindMan: do you have more experience with e2factory? Jul 07 13:37:37 * mario-goulart is curious about e2 too Jul 07 13:38:07 florian: no, heard it for the first time Jul 07 13:38:22 looks nice, but quite a lot of work for beginning the project Jul 07 13:39:02 the funny thing is, he said, more or less, "omg, we need the old compile chain, where is it?" and they use that as a solution Jul 07 13:39:15 but on the other hand, pack it up and stuff it in an SCM works prolly as well ;) Jul 07 13:39:16 for that point Jul 07 13:39:29 but at least you are helped with doing exact rebuilds Jul 07 13:39:38 which is nice and all Jul 07 13:39:55 and prolly also something I would want to deploy here Jul 07 13:40:14 (at work, which is acutally kind of a requirements nonetheless) Jul 07 13:40:23 BlindMan: what about the development of recipes? Jul 07 13:40:30 mario-goulart: what do you mean? Jul 07 13:40:56 BlindMan: I heard e2 doesn't have many recipes. Is it hard to make new ones? Jul 07 13:41:44 BlindMan: to solve 'exact rebuild on same tools as it was' problem is easy - you need to keep whole system which was used to build Jul 07 13:41:55 mario-goulart: prolly not, but it looks likeyou still have to do much of the setup you want to have for your (embedded) project Jul 07 13:42:29 hrw: yep, that was the major selling point according to the presenter Jul 07 13:42:38 hrw: at least his narrative built on that Jul 07 13:43:19 the benefit is prolly more in being able to backtrack specific configurations and being able to built from taht on Jul 07 13:43:19 BlindMan: and then I asked few questions which nearly killed this advance Jul 07 13:43:54 "how it differ from keeping that 8 years old system archived with copy of metadata/sources used to build'? Jul 07 13:43:57 etc Jul 07 13:46:04 re Jul 07 13:46:09 hrw: ah, you were that :) Jul 07 13:46:35 hrw: i asked smething like that a couple of days ago, but smoehow nobody answered that here ;) Jul 07 13:47:33 hrw: but it's with oe and the recipies the same, pack everything up + toolchain you got and stuff it in an SCM or tag it or whatever and you are set, i would assume wrt. e2 Jul 07 13:47:46 at least it's kind of achieving the same for that Jul 07 13:48:29 not quite Jul 07 13:48:50 rather pack OE metadata + sources + archive of operating system used for build Jul 07 13:48:59 then use that OS in vbox/vmware etc Jul 07 13:49:06 thats how I would solve that Jul 07 13:49:23 hrw: yeah, something like that ... or the changeroot setup Jul 07 13:49:47 I do not like chroot ideas Jul 07 13:50:42 how comes? Jul 07 13:54:41 I do not like chroot basically Jul 07 13:55:58 but chroot is faster in the cases where you don't need to care about making sure you have the other distro kernel and daemons Jul 07 13:56:15 and with bind mounts you've got all your data around anyhow Jul 07 13:57:15 sure Jul 07 13:58:24 03Koen Kooi  07stable/2009 * r4796ca8d5b 10openembedded.git/conf/distro/include/angstrom.inc: Jul 07 13:58:24 angstrom: stop disabling binary localegen for >=armv6 per RFC on the ml Jul 07 13:58:24 Signed-off-by: Marcin Juszkiewicz Jul 07 13:58:24 Acked-by: Koen Kooi Jul 07 14:00:58 So, is there BK or MTN history in some git repo somewhere? Jul 07 14:01:06 Or something I can search on the web? Jul 07 14:01:49 Tartarus: I think the main git repo should have the history all the way back to bk, though it is a bit garbled in places. Jul 07 14:01:57 bkbits.net is still online, you can see the old tree through their web interface Jul 07 14:02:21 Er, der, yes, mtn is in git :) Jul 07 14:02:26 I need bk stuff, heh Jul 07 14:02:34 I don't think the mtn archive is online anywhere but the mtn->git conversion should have kept it mostly intact. Jul 07 14:02:36 trying to see where --disable-libunwind-exceptions came from Jul 07 14:03:29 try http://openembedded.bkbits.net:8080/openembedded/packages/gcc/?PAGE=dir as a starting point Jul 07 14:04:14 k, thanks Jul 07 14:18:11 Hmm, if bk searches are to believed, it did come in post BK Jul 07 14:18:17 So it should be in our git history right? Jul 07 14:22:51 should Jul 07 14:23:04 but bk history in git is borked at places Jul 07 14:24:36 /nick cbrake Jul 07 14:25:35 * cbrake tries e2factory, but runs into several problems, and does not really understand the advantages over OE Jul 07 14:25:49 one thing about OE is it is fairly self contained Jul 07 14:26:54 So, AFAICS, --disable-libunwind-exceptions just came in w/ 4.1.2, uncommented upon Jul 07 14:26:57 cbrake: isn't e2 self contained? What aditional tools do you need? Jul 07 14:26:58 and got spread around Jul 07 14:27:42 mario-goulart: it may be from the sense that things are done in a chroot Jul 07 14:27:48 mario-goulart: but where does the chroot come from? Jul 07 14:29:11 cbrake, I added an entry for melo.oe.org Jul 07 14:29:24 back from vacation Jul 07 14:29:26 Crofton: ok, thanks -- can you reply to Tom? Jul 07 14:29:34 will do Jul 07 14:30:13 Tartarus: [OT] Jul 07 14:30:19 any idea about cfgloopmanip.c:1480: internal compiler error: in get_loop_body, at cfgloop.c:830 Jul 07 14:30:41 that's Gentoo's gcc-4.3.2 Jul 07 14:32:15 Google doesn't fully help..I'm already using -O2 Jul 07 14:32:52 odd Jul 07 14:33:03 sigh.. I'll have to read the full gccint.pdf.... Jul 07 14:34:40 heh, I guess any option below about -O9997 is probably untested with gentoo. Jul 07 14:35:08 just below -O11 ;) Jul 07 14:35:32 he..that's an old server here... CFLAGS="-march=pentium3 -pipe -O2" Jul 07 14:35:52 a glorious Supermicro DualPIII Jul 07 14:45:35 mickeyl: good morning Jul 07 14:47:17 morning pb_ Jul 07 14:47:19 hi folks Jul 07 14:47:24 hi mickeyl Jul 07 14:47:28 hey chouimat! Jul 07 14:47:33 how are things for you? Jul 07 14:47:51 mickeyl: got laid off 2 weeks ago ... beside this I will say goof Jul 07 14:47:54 err good Jul 07 14:48:00 ya, read about that here Jul 07 14:48:02 anything new in sight? Jul 07 14:48:32 not yet ... pretty dead around here and I'm on vacations for another week or 2 Jul 07 14:48:51 *nod* Jul 07 14:48:55 this must be the year of the layoffs Jul 07 14:48:59 so many folks Jul 07 14:49:01 even in IT Jul 07 14:49:37 also got a nice stack of devices that I can play with them now that I have the time ... the sheevaplug and the openrd-client come to mind Jul 07 14:51:01 mickeyl: and how are things for you? Jul 07 14:52:10 can't complain. i'd love to do more paid work on linux (as opposed to iPhone which is paying my bills atm.), but that will come Jul 07 14:52:30 the struggle for open hardware continues Jul 07 14:52:36 and i'm still a part of it Jul 07 14:52:47 (somewhat) Jul 07 14:52:53 mickeyl: iphone what a good idea ;) Jul 07 14:53:10 well... can't say it's much fun, but there's tons of demand... Jul 07 14:53:23 actually i like objective-c Jul 07 14:53:38 and learning some new things is always refreshening Jul 07 14:53:43 so true Jul 07 14:54:17 hmmm maybe I should try to port darwin to the sheevaplug ... Jul 07 15:00:00 hehe Jul 07 15:02:05 or GNUstep? Jul 07 15:15:23 re Jul 07 15:17:12 I hate moments when I need to hack in kernel Jul 07 15:27:34 hey guys Jul 07 15:27:57 so is there a cute way I can build my native toolchain such that I can install it *natively* on a system for development outside of OE? Jul 07 15:28:12 it seems a damn shame to waste the fact that I have it build Jul 07 15:31:39 "bitbake meta-toolchain" Jul 07 15:44:40 03Ihar Hrachyshka  07org.openembedded.dev * r7ad718cfd6 10openembedded.git/ (10 files in 3 dirs): Jul 07 15:44:40 qt4: updated version to 4.5.2. Jul 07 15:44:40 Signed-off-by: Ihar Hrachyshka Jul 07 15:44:40 Signed-off-by: Koen Kooi Jul 07 15:45:44 re Jul 07 16:00:29 http://tree.celinuxforum.org/CelfPubWiki/ELC2009Presentations?action=AttachFile&do=view&target=ELC.klaasvangend.openembedded.v4.pdf Jul 07 16:06:15 worth reading Jul 07 16:10:45 hrw: we all already know those pains Jul 07 16:11:38 mckoan: sure, I know Jul 07 16:12:09 * florian prits the second last slide and tapes it to the office wall Jul 07 16:13:20 yep Jul 07 16:13:23 hrw: this is quite good indeed Jul 07 16:13:40 he missed the painful environment setup Jul 07 16:13:50 florian: I think that mvista has experience in embedded Jul 07 16:14:05 florian: you mean something other then local.conf? Jul 07 16:14:05 hrw: sure... Jul 07 16:14:25 hrw: local.conf + shell environment Jul 07 16:14:32 BBPATH suxx Jul 07 16:14:59 any environment variable setting for oe must die... Jul 07 16:15:05 ok, I need to finish Jul 07 16:15:11 have a nice rest of day Jul 07 16:17:03 hrw|gone: nice presentations, thanks Jul 07 16:18:31 * florian -> home, bbiab Jul 07 16:40:53 hi rkirti Jul 07 16:45:25 hello woglinde Jul 07 16:48:57 hmmnn.... how can I go about generating my own checksums.ini file? is there some automated way to do this? Jul 07 16:50:07 nekoxp nope Jul 07 16:50:12 it's just a file with checksums in, right? you could easily enough write a script to call sha256sum and m5sum. Jul 07 16:50:21 there's no real magic involved. Jul 07 16:50:37 so how does it get generated in git? just by hand? Jul 07 16:51:02 more or less. there are some semi-automated tools for adding to it but it's fundamentally a manual process. Jul 07 16:51:40 git has nothing to do with it Jul 07 16:52:04 well someone is checking in the changes I wondered if they did it with a script or just added the entry by hand Jul 07 16:52:13 sorting it is automated.. sure.. .:D Jul 07 16:52:33 NekoXP: base.bbclass will print a suggested entry for you if it notices one is missing. Jul 07 16:52:49 I never noticed that Jul 07 16:52:53 it's easy enough to just copy that into the file, and then run the sorter if you were too lazy to put it in the right place. Jul 07 16:52:58 hehe :D Jul 07 16:53:19 that's approximately what I do, though in my case with added pauses to shake my fist at the monitor and curse the existence of checksums.ini Jul 07 16:54:15 pb *g* Jul 07 17:00:18 * pb_ go home now Jul 07 17:00:19 later all Jul 07 17:00:24 now that this topic is on, I have a rather naive query. I have been working on packages which are mostly pulled from only one repo (Maemo) so adding checksums is easy as pie, but sometimes there are generic packages which can be pulled from any debian mirror Jul 07 17:00:32 do I need to add checksums for all ? Jul 07 17:01:03 yeah, welcome to the sucky world of checksums.ini. Jul 07 17:03:56 rkirti i think those usually have a cname Jul 07 17:04:02 that you could use Jul 07 17:04:51 er i guess its an a record Jul 07 17:05:02 but yeah, ie. ftp.us.debian.org points to 5x ips Jul 07 17:05:45 m4t: :) Jul 07 17:06:21 also, to get sha256; shasum -a 256 Jul 07 17:09:54 m4t: nice. I had been using gtkhash or some openssl based command so far. Jul 07 17:11:09 i couldnt figure out how to get openssl to do sha256 Jul 07 17:11:16 it does md5 and sha1 Jul 07 17:14:35 03Ihar Hrachyshka  07org.openembedded.dev * rf3d6071476 10openembedded.git/recipes/qt4/ (qt4-embedded.inc qt4-x11-free.inc qt4.inc): Jul 07 17:14:35 qt4: added postgres, mysql, and sqlite2/3 support. Jul 07 17:14:35 Signed-off-by: Ihar Hrachyshka Jul 07 17:14:35 Signed-off-by: Koen Kooi Jul 07 17:15:01 m4t: openssl dgst -sha256 Jul 07 17:15:34 o Jul 07 17:15:45 easy enough Jul 07 17:44:31 fighting with pulseaudio again Jul 07 17:55:09 yes Jul 07 17:55:11 now it works Jul 07 17:55:38 pb with uclibc and linuxthread.old we have one problem Jul 07 17:55:54 gcc detects that __thread is available Jul 07 17:56:32 but if a library is linked of course the tls-functions are unkown Jul 07 17:56:58 a fix is suggest here I adapet it for pulseaudio where the problem occurs Jul 07 17:57:07 http://gcc.gnu.org/ml/libstdc++/2009-07/msg00001.html Jul 07 18:10:52 Is there a do_clean task defined for image recipes ? I tried it but it doesnt seem to do be doing the right thing. Jul 07 18:17:35 woglinde: that looks like a good patch, though of course it would be even better to add tls support to uclibc Jul 07 18:34:47 pb__: uclibc has TLS support on certain arches Jul 07 18:34:56 pb__: arm mips sh Jul 07 18:35:04 urg, jetlag is settling in Jul 07 18:35:38 woglinde: I think gcc only checks for TLS support though assembler Jul 07 18:36:02 we should probably build gcc with --disable-tls for uclibc/lt combo Jul 07 18:38:18 Does anybody know how to "glue" a kernel and a cpio file (for initramfs)? Currently I'm using the kernel's CONFIG_CMDLINE parameter to point to the rootfs, but this approach requires kernel recompilation for each modification made to rootfs. I'd like to have a kernel file and a cpio file containing the rootfs, then use something to put them together as an ELF file, so I don't need to recompile the kernel if I modify the rootfs (I'd Jul 07 18:38:18 only need to "glue" them together). Jul 07 18:38:53 Crofton|work: how was europe Jul 07 18:39:02 great Jul 07 18:39:18 was is your first time ? Jul 07 18:39:23 no Jul 07 18:39:31 I see Jul 07 18:39:39 I've been a number of times Jul 07 18:40:15 what all did u visit Jul 07 18:40:39 berlin/linuxtag Jul 07 18:41:23 nice was it sponsored trip ? Jul 07 18:44:31 no Jul 07 18:44:57 basically vacation, but I can write off airfare do to Linux tag :) Jul 07 18:46:02 khem: oh, right. I wonder why woglinde is having trouble then. Jul 07 18:48:16 Crofton|work: it would be nice if you can send a summary of linux tag to ml if you have time to do so Jul 07 18:50:25 I should try Jul 07 18:58:08 morning kergoth Jul 07 18:58:55 hey Jul 07 19:06:50 g'day kergoth Jul 07 19:10:35 someone other than me also complained about the size bloat due to enabling debug info Jul 07 19:10:42 re Jul 07 19:11:03 I didn't complain so far but yes its annoying Jul 07 19:11:22 with gcc 4.4 it dumps even more Jul 07 19:27:56 does anyone know if anyone's looked into creating, alongside a normal image, the same image, but including the debug information for all included packages, to be used by gdb w/ a gdbserver on the target? Jul 07 19:29:09 kergoth: sounds interested Jul 07 19:29:13 *interesting Jul 07 19:30:00 kergoth: one other thing I've run into is needed binaries that can be used with oprofile (unless someone can suggest a better way to do profiling) Jul 07 19:30:10 well, it'd be cool if you could add gdbserver to an image, flash it, boot it, and fire up gdb and connect, and have the debug information available locally Jul 07 19:30:17 ah, hmm Jul 07 19:31:21 cbrake: gprof is also useful though its only stochastic Jul 07 19:32:18 kergoth: that's a pretty good idea Jul 07 19:32:44 okay, so i take it nobody's done it :) alrighty... Jul 07 19:33:16 the debug stuff makes it hard to build on my laptop Jul 07 19:33:21 kergoth: I've never gotten oprofile to work very well, but I'm pretty sure it has to frame-pointer options Jul 07 19:33:23 with gdbserver you dont need debug info on target you need it on host Jul 07 19:33:31 khem: yes, that's what I'm talking about. Jul 07 19:33:47 khem: right now its only available inside of a pile of -dbg packages in tmp/deploy/ipk, not exactly convenient Jul 07 19:33:52 kergoth: I did this with mamona and I guess you can aswell with other OE based distro's Jul 07 19:34:21 kergoth: I see Jul 07 19:35:31 * kergoth checks out mamona Jul 07 19:35:54 installers eh, cute Jul 07 19:38:32 re Jul 07 19:38:44 khem its on i486 iwth uclibc Jul 07 19:39:42 woglinde, try adding --disable-tls to configure options for gcc-cross Jul 07 19:40:44 khem its only problem for now with pulseaudio Jul 07 19:40:58 which has tls check on its own Jul 07 19:41:17 so I adjusted the macro Jul 07 19:41:17 woglinde: I think it will be problem in libstdc++ too I believe Jul 07 19:41:19 and may be other packages Jul 07 19:41:36 then we should use the patch I posted Jul 07 19:41:47 on the tls m4 macro Jul 07 19:41:55 http://gcc.gnu.org/ml/libstdc++/2009-07/msg00001.html Jul 07 19:45:14 woglinde: yeah that is ok. Jul 07 19:45:38 but disable-tls is better for older Jul 07 19:45:47 hm? Jul 07 19:46:04 the macro adjust that Jul 07 19:48:11 argssssssss Jul 07 19:48:52 woglinde: I think the patch works for gcc libraries only Jul 07 19:48:59 why the hell I dont understand git reset Jul 07 19:49:31 if we know for sure that TLS is not supported we should disable in toolchain IMP Jul 07 19:49:34 imo Jul 07 19:50:46 hm I cant undo a git reset --hard right? Jul 07 19:50:55 :) Jul 07 19:52:24 woglinde: not possible I think Jul 07 19:52:31 damnit Jul 07 19:53:54 hi laibsch Jul 07 19:54:08 hi Jul 07 19:54:10 hi Jul 07 19:54:19 I haven't really signed in, yet ;-) Jul 07 19:55:47 okay Jul 07 19:56:00 now I have to do all pulseaudio patches agaib Jul 07 19:56:25 woglinde: always save diffs :) Jul 07 19:56:39 khem haha Jul 07 19:56:45 they was save in the commits Jul 07 19:56:59 were Jul 07 20:06:13 woglinde: local files in the working copy or index are completely gone when you do a reset, but if your changes were committed, you can get to the old ref before reset updated it by using git reflog Jul 07 20:06:19 s/do a reset/do a hard reset/ Jul 07 20:07:23 kergoth hm I did hard Jul 07 20:07:26 so all was gone Jul 07 20:07:30 but its okay Jul 07 20:07:35 learning by the hard way Jul 07 20:07:41 hehe Jul 07 20:07:51 well keep reflog in mind, can be useful on occasion Jul 07 20:07:52 the pulseaudio patches were trival Jul 07 20:07:59 but I dont remeber what I fixed else Jul 07 20:08:36 * kergoth started setting up *tons* of project directories, including the collections, to avoid those issues, so now everything he works on is completely independent of everything else.. hehe Jul 07 20:09:34 hm I hope the push will destroy nothing Jul 07 20:10:26 woglinde: reflog works till like a month of the hard reset, apparently Jul 07 20:10:49 woglinde: I have had success recovering my files lost due to hard reset thanks to that Jul 07 20:15:35 rkirti hm Jul 07 20:15:38 I will try Jul 07 20:17:36 woglinde: You most probably have already found this, but just in case : http://tinyurl.com/mlg9r2 - a comment on the same page links to a nice comprehensive tutorial on git recovery Jul 07 20:20:23 Is it possible to override the use of rm_work in a particular recipe, while having it in local.conf ? Jul 07 20:36:35 evening Jul 07 20:39:48 hi BlindMan Jul 07 20:41:37 re blinder Jul 07 20:44:36 rkirti: you can append and prepend it it Jul 07 20:45:51 khem: :) did that. very hacky in my case, but worked Jul 07 20:46:19 rkirti: what happens if you define do_rm_work in your recipe Jul 07 20:47:46 khem: nice one. Will check, Hopefully it should successfully override. Thanks :) Jul 07 20:47:56 rm_work_pn-foo = ":" ... rm_work[func] = "1" ? :) Jul 07 20:51:32 hello all Jul 07 20:51:40 gm Jul 07 20:53:19 hi likewise Jul 07 20:53:27 khem, kergoth, all: I would like to add a new architecture "nios2" to OpenEmbedded, currently being added by a colleague of mine. It's probably a uclinux + uclibc combo. Is this .dev material I can commit even when it's still in non-working state? Jul 07 20:54:01 as long as its isolated to a certain arch/machine, i don't see any reason why not. it won't affect anyone else Jul 07 20:54:23 likewise hm you could use a userbranch too Jul 07 20:54:28 if you are unsure Jul 07 20:54:46 kergoth: Agreed, it's isolated, we will start with external toolchain support. Once we touch common parts (toolchain) I might work on a branch or go through patches. Jul 07 20:55:19 sounds good Jul 07 20:55:52 florian, Crofton: how was the LinuxTag show? Jul 07 20:56:00 likewise was okay Jul 07 20:56:13 likewise: http://fl0rian.wordpress.com/2009/07/06/linuxtag-summary/ Jul 07 20:56:14 woglinde: sorry didn't know you where there too? Many OE devs? Jul 07 20:56:18 we give out many flyers Jul 07 20:56:42 gave Jul 07 20:56:42 A0 poster? Jul 07 20:56:43 args Jul 07 20:57:09 i need to try to talk MV into sending me to one of these conferences that have an OE presence Jul 07 20:57:30 hehe Jul 07 20:57:31 yes Jul 07 20:57:44 maybee you can come to oedem too Jul 07 20:57:45 I'm quite happy with it but we need to improve advertizing these events. Jul 07 20:58:00 kergoth, +1 Jul 07 20:58:08 woglinde: that would be nice. doubt itll happen this year, everybody's being tight fisted with the travel funds, stupid economy.. Jul 07 20:58:24 so that we get a higher number of our users to shows like linuxtag Jul 07 20:58:30 we should meet mid-ocean :-) Jul 07 20:58:36 khem, http://fl0rian.wordpress.com/2009/07/06/linuxtag-summary/ Jul 07 20:59:20 what of the friendly-arm devices is that specifically on the photo? I am interested in the MACHINE/IMAGE combo :-) Jul 07 20:59:35 Crofton: yeah thx read it just now Jul 07 20:59:37 isn't sealand somewhere in there ;) Jul 07 20:59:50 khem hm could you please look at uclibc-inital Jul 07 21:00:00 the first make now builds the whole uclibc Jul 07 21:00:08 woglinde hmm Jul 07 21:00:13 so the second make made no sin Jul 07 21:00:23 khem: long time no see, good to see you around again Jul 07 21:00:30 pregen install_dev Jul 07 21:00:45 likewise: the thingy with the large screen left is a micro2440 sdk with 7" screen Jul 07 21:00:52 likewise: I was on vacation and then work piled on : Jul 07 21:01:01 so no much cycles for OE lately Jul 07 21:01:24 whahahaa Jul 07 21:01:38 micro misscompiles i486 targets Jul 07 21:02:00 khem: I thought you had a multi-khem-kore upgrade recently? Jul 07 21:02:10 undefined reference to `__i686.get_pc_thunk.bx' Jul 07 21:02:12 heh yeah Jul 07 21:02:25 woglinde: that's not a miscompilation, it's just a glibc bug Jul 07 21:02:38 see the logs :-} Jul 07 21:03:57 woglinde what march is it using Jul 07 21:04:20 pb and wyh nobody fixed it? Jul 07 21:04:27 woglinde: does thsis patch helps http://www.eglibc.org/archives/patches/msg00073.html Jul 07 21:04:37 woglinde: well, I guess not many people care about i586 and i686. Jul 07 21:04:43 it's a trivial bug, it should be easy to fix Jul 07 21:05:20 er, not many people care about i486 and i586 Jul 07 21:05:22 khem aeh? Jul 07 21:05:28 I guess plenty still care about i686. Jul 07 21:05:36 khem I am trying i486 arch for wrap board Jul 07 21:05:39 khem: that patch doesn't look like it will help with this Jul 07 21:06:12 not least since it's all guarded by "#ifdef __i686" and that isn't going to be true on a 486 system Jul 07 21:06:31 oh i486 hmm but I think you need to use march=i486 while compiling glibc Jul 07 21:06:38 woglinde: glibc 2.6.1 should work, it got broken somewhere between there and 2.9. Jul 07 21:06:44 *sigh* why always I stumble about it Jul 07 21:06:57 seriously, shouldn't x86 be more supported now that we can consider intel/x86 going embedded? Jul 07 21:07:30 woglinde: or, as I say, see the logs. I can't remember exactly what you need to to do fix it but I did mention it to someone else yesterday or the day before Jul 07 21:07:43 dinner time now, bbiab Jul 07 21:08:07 pb__: I think with newer version of glibc some old x86 variants support is dropped Jul 07 21:08:19 I think i386 for sure Jul 07 21:08:34 hm Jul 07 21:09:13 hm maybee debian has a fix Jul 07 21:09:15 ah, not dinner time after all Jul 07 21:09:27 * pb__ checks the logs himself Jul 07 21:10:57 10:49.00 valhalla: I think it's probably just a case of making dl-tlsdesc.S use SETUP_PIC_REG(). Jul 07 21:11:06 woglinde: that's probably what you need to do, this is the same bug. Jul 07 21:11:43 it happens to compile on i686 as it stands but it will break for i586 and older archs. Jul 07 21:14:32 or just use an older libc, if you don't need 2.9 specifically Jul 07 21:18:02 woglinde: you need this patch http://sourceware.org/git/?p=glibc.git;a=commit;f=sysdeps/i386/dl-tlsdesc.S;h=71885b7a9149b5e54906204491b64d5ad0a48e9d Jul 07 21:18:31 http://sourceware.org/ml/libc-help/2008-12/msg00013.html Jul 07 21:18:32 hehe Jul 07 21:19:05 yeah, that looks like the right thing Jul 07 21:19:23 hm should I provide a patch? Jul 07 21:19:28 sure Jul 07 22:40:30 hi, are the server whitelisted for ssh? Jul 07 23:11:40 GNUtoo: which servers Jul 07 23:12:01 oe's ones...there are rumors about SSH exploits...but they are only rumors Jul 07 23:15:53 khem, I first knew about it in the openssh mailing list... Jul 07 23:16:50 I need to add some stuff to the pkg_postinst for gnome-panel Jul 07 23:17:44 when I add a pkg_postinst to the recipe it puts my additions *before* those that come from gconf class Jul 07 23:18:06 is there an easy way to have my stuff come *afte* the gconf stuff? Jul 07 23:21:31 sakoman: do_pkg_postinst_append? Jul 07 23:21:39 sakoman: just an assumption Jul 07 23:24:00 I tried pkg_postinst_append, but no joy Jul 07 23:25:14 it puts it at the start, just like pkg_postinst without the append **** ENDING LOGGING AT Wed Jul 08 02:59:59 2009