**** BEGIN LOGGING AT Wed Jun 25 02:59:57 2008 Jun 25 03:19:17 use -march=i486 for glibc and rest does not matter what opt you use Jun 25 06:25:27 03Laibsch 07org.oe.dev * r5fa082df... 10/ (1 packages/cmake/cmake.inc): cmake: fix minor spelling error. Jun 25 07:01:11 hi all, where can i get more info on these errors, lines 86-89... http://dpaste.com/58853/ Jun 25 07:02:39 morning Jun 25 07:02:55 methril, morning Jun 25 07:03:01 Sup3rkiddo, why do you have a .debug directory? Jun 25 07:03:57 mbuf, i don't know where that came from... i thought it would be because of the -ggdb flag but that was not the case Jun 25 07:04:37 shall i pastebin my recipe? Jun 25 07:04:43 Sup3rkiddo, from the output, only a "debug" package is allowed to have a .debug directory; yes pastebin your .bb file Jun 25 07:04:57 then remove the -ggdb Jun 25 07:05:43 the error tells that odeviced is not a debug package, and you have debug Jun 25 07:06:12 pastebin your receipe Jun 25 07:06:49 mbuf, methril, http://dpaste.com/58996/ Jun 25 07:07:21 Sup3rkiddo, you can also look at the insane.bbclass file for these QA checks openembedded/classes/insane.bbclass (in OM) Jun 25 07:07:56 methril, oh, so how do i make odeviced a debug package, i was of the impression that the build systems gives a separate -dev and -dbg package automatically Jun 25 07:08:04 mbuf, thanks, i will take a look at that Jun 25 07:09:21 5th bullet. http://oe.linuxtogo.org/more-quality-assurance-tests is this somehow related? Jun 25 07:10:35 Sup3rkiddo, yes Jun 25 07:10:52 hmm... Jun 25 07:11:33 lemme try again commenting out the rdepends flag, it compiles fine, so i guess there shouldn' be an issue with my code Jun 25 07:13:31 is a QA check (then insane.bbclass one) Jun 25 07:13:43 i see, will take a look Jun 25 07:24:45 from package_qa_write_error in insane.bbclass it seems that .so are not allowed in normal packages...should i move my .so to a separate package? Jun 25 08:14:09 hi all, is there a good "hello world" tutorial on how to make oe recipes and add them to an image ? Jun 25 08:14:21 (using a local overlay, possibly) Jun 25 08:24:08 03Laibsch 07org.oe.dev * r5ebc68ac... 10/ (4 files in 2 dirs): olsrd: introduce MAJ_VER variable. Downloads are split into MAJ_VER directories. Jun 25 10:44:31 morning Jun 25 10:44:57 morning hrw Jun 25 11:15:48 Crofton|work: ping? Jun 25 11:16:04 anyone has pictures from my flat at OEDEM2007? Jun 25 11:21:57 your empty flat? Jun 25 11:22:43 yes Jun 25 11:22:56 I should have Jun 25 11:23:36 can you put one picture somewhere? Jun 25 11:25:59 zecke: sorry - removed them ;( Jun 25 11:29:19 morning all Jun 25 11:33:04 zecke, pong Jun 25 11:35:15 Crofton|work: do you have a picture of my empty flat? Jun 25 11:35:28 let me see Jun 25 11:38:36 I have a couple Jun 25 11:40:31 hmm, task-base wants to build the kernel ... Jun 25 11:42:26 where should I put the picture? Jun 25 12:18:58 hi mickeyl, zecke, all Jun 25 12:19:15 master! Jun 25 12:19:33 zecke, I have a couple of photos Jun 25 13:19:25 hmm Jun 25 13:20:03 quick question: BB_NUMBER_THREADS>1 can be considered "safe", as opposed to PARALLEL_MAKE? Jun 25 13:20:19 ie, the latter can actually break builds, while the former shouldn't Jun 25 13:20:34 oxo: well, when breakage i regularily have with do_fetch Jun 25 13:20:55 oxo: e.g. if you update 6 sources from one cvs server but they limit the number of connections Jun 25 13:21:10 ah ok Jun 25 13:21:14 but apart from fetching? Jun 25 13:21:36 not seen a breakage yet, the deps seems to be right nowadays Jun 25 13:21:38 oxo: save Jun 25 13:21:44 because I suspect my recent glibc-errors were result of parallel building (threads + "-j") Jun 25 13:21:59 oxo: so the only possible breakage is with wrong DEPENDS Jun 25 13:22:07 ah I see Jun 25 13:22:12 BB_NUMBER_THREADS = "8" Jun 25 13:22:16 PARALLEL_MAKE = "-j 8" Jun 25 13:22:22 are my configs for quadcore Jun 25 13:22:33 thing is, my glibc-errors are gone when building without -j and threads Jun 25 13:22:59 this is dual quadcore, even, so 16/8 could do Jun 25 13:23:53 hrw: does glibc 2.6.1 build with these flags? Jun 25 13:24:03 in poky yes Jun 25 13:24:18 ever tried angstrom? Jun 25 13:24:26 let Jun 25 13:24:27 some time ago - moment Jun 25 13:24:49 *let me put it in other words: is it likely that glibc/binutils/gcc breaks when using -j ? Jun 25 13:24:59 same settings for angstrom-2008.1 used and worked Jun 25 13:30:37 oxo: maybe Jun 25 13:32:41 is it possible to instruct bitbake/oe to build glibc without -j? Jun 25 13:32:57 ie, override parallel_make for certain packages Jun 25 13:34:23 oxo: grep for PARALLEL_MAKE in packages Jun 25 13:34:30 ah k, tnx Jun 25 13:34:34 but not in local.conf? Jun 25 13:34:45 well, local.conf is global Jun 25 13:34:52 in the bb it is per package Jun 25 13:35:35 hmm Jun 25 13:42:26 hi Jun 25 13:45:19 for a channel with 145 people it's fairly quiet.. Jun 25 13:47:11 ;) Jun 25 13:48:01 i was just wondering (hoping? praying?) that someone here had got oe to work with an acer n30 pda Jun 25 13:49:56 n30... Jun 25 13:50:30 it's a near pda, letux project started working on linux for it then one of the main devs had a car accident iirc Jun 25 13:50:33 neat* Jun 25 13:50:46 it was merged with oe supposedly Jun 25 13:50:57 iirc we have this one Jun 25 13:51:32 with very old kernel... Jun 25 13:51:54 i don't really mind what kernel if i can get X running on it lol Jun 25 13:54:33 other main constraint is acer don't have the wince image on their site any more so i want to just work off and sd card without flashing the main rom in the devoce Jun 25 13:54:37 device* Jun 25 13:56:35 use haret? Jun 25 13:57:03 i read somewhere that haret didn't work but yeah that's the idea Jun 25 13:57:16 i've found a couple of bootloaders people have written than work Jun 25 13:57:42 so i've got a basic kernel booting with usb keyboard support Jun 25 13:58:04 but that's just one i dug up not one i've managed to produce myself Jun 25 14:04:34 bitbake or something about oe doesn't like me much :/ Jun 25 14:06:20 uafter doing what the "getting started" page says should ">bitbake nano" work? Jun 25 14:12:50 need help with changing the ctrl + tab way of switching tabs in dreamweaver to look more like the windows one Jun 25 14:19:39 hmmm, task-boot has DEPENDS = "virtual/kernel" Jun 25 14:19:49 Maybe commeting this out while stop it from trying to make a kernel Jun 25 14:20:00 I can't see why ASSUME_PROVIDED fails though Jun 25 14:21:50 Good morning. for those going to guadec hope to see you there... Jun 25 14:23:01 kgilmer: I will be at OH booth Jun 25 14:23:20 where is it? Jun 25 14:23:50 Istanbul Jun 25 14:23:52 http://guadec.expectnation.com/public/content/main Jun 25 14:24:22 Sounds fun, wish I was good enough at UI to justify going :) Jun 25 14:24:27 http://guadec.org/ is simpler Jun 25 14:41:01 Crofton|work well I'm certainly not good enought yet still going :) Jun 25 14:49:47 Can I put a RDEPENDS in a machine.conf file? The davinci machines have a kernel driver that has some sample ARM side programs I would like to install in the image Jun 25 14:50:22 MACHINE_EXTRA_RDEPENDS Jun 25 14:50:29 thanks Jun 25 14:50:40 task-base use it Jun 25 14:50:47 ok Jun 25 15:55:13 is the first build supposed to take forever? Jun 25 15:55:33 yeah, it takes awhile :) Jun 25 15:55:39 good :) Jun 25 15:55:42 building the toolchain Jun 25 15:55:53 ah ok Jun 25 15:56:31 it was pretty sure it was doing what it's meant to because there's still lots of positive text coming every so often Jun 25 15:58:30 s/it/I Jun 25 15:59:12 do you know what the support's like for the sdio wifi cards? Jun 25 15:59:35 sorry mm0zct, I do not. Jun 25 16:00:12 it's ok, it's probably never going to happen anyway since i'm planning to just run of an sd card anyway.. which kinda stops me being able to use one lol Jun 25 16:01:42 :) Jun 25 16:03:37 can anyone explain the error here: http://www.pastebin.org/46098 - 'ln: creating symbolic link `ipkg`: File exists' Jun 25 16:04:06 I get that when trying to build targets for certain machine's but not others Jun 25 16:11:39 tharvey_ I've seen that before too. I think I've just removed the symbolic link to get it fixed. Jun 25 16:11:52 or perhaps cleaned the package in question Jun 25 16:13:53 ok, I'll try that - thanks Jun 25 16:16:51 03mickeyl 07org.oe.dev * r85a1e5d8... 10/ (1 conf/distro/include/preferred-om-2008-versions.inc): preferred-om-2008-versions.inc: bump python-pygobject Jun 25 16:28:03 by "a while" is 2hours normal? lol Jun 25 16:28:58 mm0zct: depends on machine speed Jun 25 16:29:14 my machine is able to build whole image in 2h Jun 25 16:29:27 my previous one was not able to Jun 25 16:29:31 1.7GHz core duo Jun 25 16:31:21 i just told it to botbake nano but it's the first time so it's got to do a lot of other stuff first Jun 25 16:33:07 bitbake* Jun 25 16:38:07 * * OE Bug 4390 has been created by gwossum(AT)acm.org Jun 25 16:38:09 * * Add AVR32 gcc patch from latest buildroot-avr32 Jun 25 16:38:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4390 Jun 25 17:00:06 thesing: morning Jun 25 17:10:25 03mickeyl 07org.oe.dev * ra71285da... 10/ (4 files in 4 dirs): frameworkd git specify modem type in configuration Jun 25 17:17:16 hi all .. was wondering .. do you guys use oe in your day-to-day application development, or do you use another cross compiler to do app dev and, when you are done, you create a bitbake recipe and integrate it into an oe build? Jun 25 17:18:50 i prefer latter Jun 25 17:22:12 kario, what are you using for day-to-day app dev .. standalone cross-compiler or a dev environment like scratchbox or something? Jun 25 17:28:58 03koen 07org.oe.dev * r5a8e680a... 10/ (1 packages/linux/linux-omap2_git.bb): linux-omap2 git: bump to 2.6.26-rc8 Jun 25 17:30:22 jjg, you can use a bb fiel to do dev builds, or yuo can try using toolchain from tmp Jun 25 17:30:30 there are several solutions Jun 25 17:30:44 there is a meta-toolchain target that works for some Jun 25 17:31:37 i'm looking for something that is simple to use ... as i understand it, oe doesn't delete some cache files or something, so i have to manually delete them between each build .. that a correct statement? Jun 25 17:32:53 standalone cross-compiler / host compiler Jun 25 17:34:03 start with a host compiler and move to cross compiler when i need to test the binary in the device Jun 25 17:34:48 kario, do you target arm for any of your apps? Jun 25 17:35:37 arm yes Jun 25 17:35:59 what cross compiler are you using? i've been using code sourcery's Jun 25 17:37:35 http://www.emdebian.org/ Jun 25 17:37:58 hi Jun 25 17:39:32 likewise, gm Jun 25 17:40:54 * Crofton|work curses that he cannot build an image without building a kernel Jun 25 17:41:36 Crofton: Did you try making a fake-kernel.bb ?\ Jun 25 17:41:46 no Jun 25 17:42:00 I suspect I will try making one that builds first Jun 25 17:44:37 thesing: Is tere a wiki somewhere that has instructions on how to make a initramfs kernel in OE? I'd like to see if I can get it working on poodle Jun 25 17:44:45 It is just annoying that I can't find the "problem" Jun 25 17:44:49 Crofton: let me guess: you cannot build kernels anymore? Jun 25 17:44:53 no Jun 25 17:45:08 I am trying to build a rootfs for a xilinx ml403 board Jun 25 17:45:19 building the kernel properly is "tricky" Jun 25 17:45:29 all I need is the rootfs, someone else made the kernel Jun 25 17:45:59 task-boot depends on virtual/kernel, but commenting that out doesn't stop bitbake from trying to build the kernel Jun 25 17:46:20 Crofton: but you can (if you want) specify and build the kernel using OE? Jun 25 17:46:49 Crofton: the task based stuff maybe includes kernel-modules? Jun 25 17:47:00 I could, but I am trying to put together the rootfs without a working OE recipe Jun 25 17:47:48 maybe Jun 25 17:48:02 I am having trouble decoding the dependency graphs Jun 25 17:48:12 Crofton: can you paste them? Jun 25 17:48:47 from head: a -> b: means a depends on b Jun 25 17:48:48 got it Jun 25 17:49:33 commenting out kernel from task-boot RDEPENDS Jun 25 17:50:44 task-base tries to make kernel though Jun 25 17:51:11 I hate the *task* stuff. That's what I throw out first when I start building s/w based on oe. Jun 25 17:51:34 I want a sparse dependency tree, deterministic without the task stuff. Jun 25 17:52:29 it is great when it works Jun 25 17:52:44 but finding stuff like this is hard Jun 25 17:53:55 Crofton: how about a .bb that takes the external kernel and just packages it? Jun 25 17:54:43 the problem is, I do not want to screw with the kernel recipe at all Jun 25 17:55:09 the problem is task-boot builds a kernel, and task-bae adds modules Jun 25 17:55:58 now that I understand, I may be able to solve with assume provided .... Jun 25 17:56:05 * Crofton is cranky today Jun 25 17:56:53 me too Jun 25 17:57:10 and then also I just payed my lifetime-3rd speeding ticket Jun 25 17:57:21 for a lousy 4 km/h too fast :-) Jun 25 17:58:18 bbiab (or maybe not) Jun 25 18:04:36 hvontres|work: just set INITRAMFS_IMAGE somewhere to the initramfs-image you want to use. Jun 25 18:04:40 bye. Jun 25 18:48:46 now building base-image with MACHINE=x86-prescott fails with multiple 'ERROR: duplicate src declaration. Skipping' msgs: http://www.pastebin.org/46138 - ideas? Jun 25 19:15:11 tharvey: I'm getting the same thing. You'll notice, though, that tmp/root_fs is fully populated. Jun 25 19:18:05 yes, but the rootfs image isn't built... Jun 25 20:02:07 tharvey: What kind of image are you looking for? Jun 25 22:23:43 Does anyone know about debian repositories? Jun 25 22:24:08 Calling all debian repositories mavens. Jun 25 23:11:24 Does anyone know about debian repositories? Jun 25 23:36:32 http://linuxdevices.com/news/NS9978560959.html Jun 25 23:36:46 openmoko is making strides Jun 25 23:39:04 awesome Jun 25 23:43:34 oho linuxtogo related stuff on linuxdevices.com? Jun 26 00:02:09 trying to understand how to iteratively tweak a kernel config. Shouldn't I do 'bitbake -c devshell virtual/kernel' then 'make menuconfig' in that shell? I find that it performs the patch all over again and perhaps doesn't do the do_configure action Jun 26 00:04:07 fwiw try -c menuconfig, then configure your kernel, then run -c compile Jun 26 00:08:57 There's no automatic way to save work/..../linux-.../linux-2.6.N/.config to the OE file that was used Jun 26 00:08:58 tharvey: or, after running make menu config in devshell, copy the new .config file back to the source dir Jun 26 00:09:04 Need to do that part on your own Jun 26 00:10:48 I don't mind copying the .config back myself (expected that) just couldn't figure out why it wasn't copied there to begin with Jun 26 00:14:04 dcordes_, didn't know about 'bitbake -c menuconfig' that makes sense... guess kernel package adds that action - but it still doesn't configure the kernel beforehand - perhaps I need to 'bitbake -c configure' first? Jun 26 00:17:07 I think I understand - devshell.bbclass implements the devshell task, but inserts it after patch (before configure) so that explains why devshell does not copy in the config right? Jun 26 00:17:19 tharvey: hm no clue, sorry. I do the kernel editing stuff manually. I let OE build the virtual/kernel just as it likes, then I set CROSS_COMPILE=/oe/toolchain and ARCH=arm and edit and build the kernek in the workdir Jun 26 00:17:33 good old kernek Jun 26 00:18:40 ya... I think I'll need to resort to that... nothing wrong with that method Jun 26 00:19:20 looking at kernel.bbclass, the menuconfig task is inserted after do_patch - again... before any configure is done Jun 26 00:23:55 why would 'bitbake -c configure virtual/kernel' perform 270 tasks including building of automake, libtool, bison, flex etc when they have already been built? Jun 26 00:24:56 tharvey: err did you change DISTRO or so? Jun 26 00:27:22 no... Jun 26 00:28:57 no clue then Jun 26 00:29:15 its like it's rebuilding the toolchain correct? Jun 26 00:29:40 looks like it Jun 26 00:31:01 looks like I need to look at the dependencies... I can see that with 'bitbake -v -D -g virtual/kernel' correct? Jun 26 00:31:47 never used that Jun 26 00:32:11 hmm... I'll attach it tomorrow. At least OE is starting to make some sense to me now (mostly b/c of the awesome documentation on the poky site) Jun 26 00:32:14 night... Jun 26 00:32:33 tharvey: gnight Jun 26 02:45:50 03rwhitby 07org.oe.dev * rdd815ee8... 10/ (1 conf/distro/include/slugos.inc): slugos.inc: Set preferred providers for avahi and gconf **** ENDING LOGGING AT Thu Jun 26 02:59:57 2008