**** BEGIN LOGGING AT Thu Jun 17 02:59:57 2010 Jun 17 06:54:13 03Steffen Sledz  07org.openembedded.dev * r7d940cb1aa 10openembedded.git/recipes/busybox/ (4 files in 3 dirs): (log message trimmed) Jun 17 06:54:13 busybox-1.11.3/busybox-1.13.2: enable backgrounding for udhcpc Jun 17 06:54:13 * The former default behaviour was to exit with failure if lease is not Jun 17 06:54:13 immediately obtained. This results in permanent network disconnect if DHCP Jun 17 06:54:13 server is (accidentally) not available at boot time. :( Jun 17 06:54:14 * Since bb 1.14 this is no longer hardcoded and can be configured. Jun 17 06:54:15 Signed-off-by: Steffen Sledz Jun 17 06:59:57 morning Jun 17 07:00:07 mickey_office: are your slides available somewhere? Jun 17 07:03:49 hi Jun 17 07:09:40 morning folks Jun 17 07:10:17 hrw: http://www.linuxtogo.org/~mickeyl/misc/LinuxTag10.pdf Jun 17 07:14:46 thx Jun 17 07:16:38 mickey_office: I see 2 bonus slides ;D Jun 17 07:17:28 hehe, ah right Jun 17 07:17:44 didn't have a chance to show these and noone asked :) Jun 17 07:30:10 mickey_office: very cool talk. too bad I had to be at the office for linuxtag Jun 17 07:32:04 mickey_office: and good that you went into detail on what is possible on the 'open' phones and what not. until now I had the idea that although having proprietary bits in it, maemo is the best approach yet (with out of the box root access and kernel with all drivers in mainline) Jun 17 07:32:19 s/idea/impression/ Jun 17 07:48:47 mickey_office: nice presentation! Jun 17 07:54:26 thanks Jun 17 07:54:39 it lacks a bit without me talking, but better than nothing ;) Jun 17 07:57:17 mickey_office: I could have given you a webos-internals logo .... Jun 17 07:57:26 yeah, but I got the idea Jun 17 07:57:51 actually found two sheets missed a caption the ones withi nternals (21 and 22 or so) Jun 17 07:57:56 would have been nice if there was one Jun 17 07:58:36 mickey_office: is it ok to share that link? Jun 17 07:59:06 rwhitby: absolutely. slides were composed in a rush; usually i have them ready before I go to the conference, this year i have so much pressure in some projects that I had to do them in the train :/ Jun 17 07:59:26 that's why some pics lack my usual quality control ;) Jun 17 07:59:53 btw anyone ever seen this behaviour: kernel boots, init=/bin/sh, I get a shell prompt (after the message /bin/sh: can't access tty; job control turned off Jun 17 07:59:53 ) then I can execute one command, that command executes, but the shell prompt never reappears; shell and apps are busybox Jun 17 07:59:56 mickey_office: what's the "Custom code on existing userland" line mean? Jun 17 08:00:28 Writing native code with "insert-random-compatible-toolchain here" and being able to run it Jun 17 08:00:49 (not touching the existing system) Jun 17 08:02:25 mickey_office: we do that on webOS Jun 17 08:02:59 (I agree with all the other lines, BTW) Jun 17 08:06:14 rwhitby: right, this was on the borderline of getting a green checkmark, IIRC i said something along the lines of 'not officially supported until the NDK and only recently some more familiar display subsystems" Jun 17 08:06:41 btw., did anyone have luck with X cooperating on a dedicated VT yet? Jun 17 08:07:01 mickey_office: oh, you're talking officially supported Jun 17 08:07:14 mickey_office: but early access for developers to PDK has been available for months Jun 17 08:07:39 with SDL, which is standard, since Jan Jun 17 08:10:36 hm Jun 17 08:10:57 libcap2 should use libattr1 instead of libattr from xfsprogs Jun 17 08:10:58 morning Jun 17 08:11:00 hit ant Jun 17 08:11:03 ups hi ant Jun 17 08:18:11 03Koen Kooi  07org.openembedded.dev * r9e52c6686d 10openembedded.git/recipes/fontconfig/ (fontconfig-native_2.6.0.bb fontconfig_2.6.0.bb): fontconfig-native 2.6.0: convert to new style staging Jun 17 08:18:12 03Koen Kooi  07org.openembedded.dev * r676cdc0a10 10openembedded.git/conf/distro/angstrom-2008.1.conf: angstrom: pin eglibc version to the same as the glibc version Jun 17 08:19:15 good morning Jun 17 08:19:43 hi florian Jun 17 08:31:24 hi pwgen Jun 17 08:31:32 pwgen I found a flaw in oe Jun 17 08:31:37 with libattr and attr Jun 17 08:31:50 which actually prevents us to install systemd Jun 17 08:33:18 I posted a fix to oe-dev ml Jun 17 08:36:36 . i am actual stuck at cgroups. my dev kernel had disabled it .. backed new kernel . but systemd fails . Jun 17 08:36:54 hm Jun 17 08:37:52 strange they want mount /cgroup/systemd, but the /cgroup is mounted as tmpfs with no systemd directory. Jun 17 08:38:22 hm Jun 17 08:38:28 special debian stuff? Jun 17 08:39:04 . hmm maybe, i had choosen debian as distribution on systemd compile time . Jun 17 08:39:34 because, the configure script cannot detect angstrom, and i think angstrom is quite debian orientated Jun 17 08:41:10 . the funny thing , this is the first time a have seen someting like cgroups, never used it before. i have to learn something Jun 17 08:41:17 hm what is systemd-install Jun 17 08:42:07 ? sounds like a binary install tool ? Jun 17 08:47:45 hm okay Jun 17 08:47:53 seems its installing tool Jun 17 08:48:01 and dis/enable services Jun 17 08:51:21 i am not shure but i think is was run at the end when i had done native compiling Jun 17 08:52:12 after native compiling the /cgroups directory appeard and some systemd-* binarys appeard in /sbin .. Jun 17 08:52:55 systemd-install is in /usr/bin Jun 17 08:53:36 03Koen Kooi  07org.openembedded.dev * rb7f66c7347 10openembedded.git/conf/distro/include/ (toolchain-external.inc toolchain-internal.inc): Jun 17 08:53:37 toolchain-internal: define PREFERRED_PROVIDER for linux-libc-headers Jun 17 08:53:37 toolchain-external: define PREFERRED_PROVIDER for linux-libc-headers, but comment it. Someone actually using this needs to verify Jun 17 08:53:46 okay Jun 17 08:53:52 image is ready Jun 17 08:53:57 lets test it Jun 17 08:55:24 woglinde: let get it run native compiled. then lets think about getting it runnung with OE/Angstrom .. Jun 17 08:55:39 hehe Jun 17 08:55:48 no will test it on beagleboard now Jun 17 08:57:41 by the way. mplayer_svn seems only compiled with somewhere enabled omap stuff ( mplayer_1.0rc2 comiles fine ) Jun 17 09:05:09 ah Jun 17 09:05:14 I need cgroup to Jun 17 09:05:16 o Jun 17 09:05:23 lets compile the kernel again Jun 17 09:06:22 pwgen but this dont looks so bad Jun 17 09:06:24 http://pastebin.com/Ltrud8tb Jun 17 09:10:08 hm pwgen I think we should patch it to run without cgroup Jun 17 09:10:33 cgroup came with 2.6.27 , so i have luck Jun 17 09:11:15 but on zaurus its stuck to 2.6.26 . and maybe on some other platforms also Jun 17 09:11:25 yeah but cgroup isnt necessary on a one cpu system Jun 17 09:12:11 als i understand it its for process limiting and this may be for interest Jun 17 09:12:59 hm Jun 17 09:13:03 okay Jun 17 09:14:02 but the devtmps may be more problematic . actually i did not came so far in the bootprocess to hang on missing devnodes .. Jun 17 09:14:26 .. maybe . its hanging, and i am not quite shure why ... Jun 17 09:15:44 failed to mount /cgroup/systemd of type cgroup with options none,name=systemd. in mount-setup.c:93 .. Jun 17 09:16:19 i wish i could get a shell (:-(( Jun 17 09:25:15 03Martin Jansa  07org.openembedded.dev * ra3b49ea0b4 10openembedded.git/conf/distro/include/preferred-shr-versions.inc: Jun 17 09:25:16 preferred-shr-versions: set preferred version also for linux-libc-headers-native Jun 17 09:25:16 Signed-off-by: Martin Jansa Jun 17 10:08:02 03Koen Kooi  07org.openembedded.dev * r69d4f16058 10openembedded.git/recipes/fakeroot/ (fakeroot-native_1.12.4.bb fakeroot_1.12.4.bb): Jun 17 10:08:02 fakeroot(-native): convert to new style staging Jun 17 10:08:02 Staging looks really weird, but it matches the way it was. Jun 17 10:08:03 03Koen Kooi  07org.openembedded.dev * r689618f8e2 10openembedded.git/conf/distro/angstrom-2008.1.conf: angstrom: pin llh-native, just like SHR does Jun 17 10:08:04 03Koen Kooi  07org.openembedded.dev * r9181dc14ab 10openembedded.git/recipes/dbus/dbus-native_1.2.1.bb: dbus-native 1.2.1: convert to new style staging Jun 17 12:23:04 he stefan Jun 17 12:23:10 hi woglinde Jun 17 12:23:32 woglinde: fighting a not staged jni.h right now with your overlay changes from yesterday Jun 17 12:23:36 woglinde: any idea? Jun 17 12:23:45 hu? Jun 17 12:23:54 woglinde: ERROR: '/home/stefan/stable/meta-jalimo/packages/icedtea/openjdk-6_6b18.bb' failed Jun 17 12:23:54 hm Jun 17 12:24:09 hm robert cleaned up a lot today Jun 17 12:24:36 hm you need to use jamvm-native and jamvm-initial Jun 17 12:24:42 no cacao please Jun 17 12:24:50 and classpath 0.98 Jun 17 12:24:52 woglinde: culprint seems to be this one: http://pastebin.com/fs6AhnUL Jun 17 12:25:11 That should all be default when I build openjdk-6 on our setup, no? Jun 17 12:25:19 let me check Jun 17 12:25:50 hm look what the fake-jdk in the WOKRDIR looks like Jun 17 12:25:57 it has the headers Jun 17 12:26:09 woglinde: ok Jun 17 12:26:12 stefan and which stage was it? Jun 17 12:26:22 1/3 or 2/3 or 3/3 Jun 17 12:26:29 search in do_compile Jun 17 12:26:59 mom Jun 17 12:27:52 woglinde: my ssh connection to the buildhost is laggy, takes some time Jun 17 12:29:53 woglinde: The do_compile log does not have anything about stages. Do I look at the wrong place? Jun 17 12:30:29 sure Jun 17 12:30:38 search for //3 Jun 17 12:30:39 ups Jun 17 12:30:44 /3 Jun 17 12:30:51 ah, not for stage :) Jun 17 12:32:12 woglinde: there we go: 2/3 Building bootstrap JDK Jun 17 12:32:52 hm okay Jun 17 12:32:55 woglinde: building classpath from scratch now, checking the version and the javavm-native and initial Jun 17 12:33:33 whats bootstrap/jdk1.60/include has inside the icedtea dir Jun 17 12:38:40 woglinde: only a broken link there Jun 17 12:38:42 lrwxrwxrwx 1 stefan stefan 99 2010-06-16 17:28 * -> /home/stefan/stable/tmp/work/armv7a-angstrom-linux-gnueabi/openjdk-6-jre-6b18-r0/fake-jdk/include/* Jun 17 12:39:46 the fake-jdk include dir is empty Jun 17 12:40:30 stefan ah not good Jun 17 12:40:37 yeah, figured that :) Jun 17 12:41:03 no jni.h makes little babyseal cry ;) Jun 17 12:47:23 hms how do I an ack to a commit? Jun 17 12:49:14 add an ack you mean? Jun 17 12:49:19 yes Jun 17 12:49:23 sorry Jun 17 12:49:34 either by adding it to the pagtch file by hand or getting it from patchwork Jun 17 12:49:42 I normally use the ölater one Jun 17 12:50:06 just go on mbox file and you get the patch with all ACKs and SOBs included Jun 17 12:50:28 woglinde: would be thtis one for you: http://patchwork.openembedded.org/patch/2221/mbox/ Jun 17 12:53:32 hm forogt signed off anyway Jun 17 12:54:21 :) Jun 17 12:54:39 you can still do an git am -s when applying it Jun 17 12:56:42 or use contrib/patchwork/pw-am.sh to get patch Jun 17 12:57:02 hrw: yeah, good point Jun 17 12:57:03 pw-am.sh fetches patches, signs and integrated Jun 17 12:57:11 stefan_schmidt: I am one of authors ;D Jun 17 12:57:19 hrw: :) Jun 17 12:57:38 hrw: Does your new job leave you anytime for OE? Jun 17 12:57:46 or is it sucked by work and family? Jun 17 12:57:53 Which of course have prio :) Jun 17 12:58:12 stefan_schmidt: I do not have customers for OE now and none of devices here runs OE derivates so no builds Jun 17 12:58:29 stefan_schmidt: but I look at OE recipes for hints Jun 17 12:58:36 hrw: heh Jun 17 13:05:53 03Henning Heinold  07org.openembedded.dev * rfad5af9a1c 10openembedded.git/recipes/xfsprogs/libattr_2.4.43.bb: Jun 17 13:05:53 libattr: remove libattr because it is clashing with udev/attr and is not used by any other recipe Jun 17 13:05:53 Signed-off-by: Henning Heinold Jun 17 13:05:53 Acked-by: Koen Kooi Jun 17 13:05:53 Acked-By: Frans Meulenbroeks Jun 17 13:06:02 03Henning Heinold  07org.openembedded.dev * r3784cd3860 10openembedded.git/recipes/libcap/libcap2_2.16.bb: Jun 17 13:06:02 libcap2: use attr instead of libattr as dependency Jun 17 13:06:02 * convert to new staging system, by removing do_stage Jun 17 13:06:02 * bump PR Jun 17 13:06:02 Signed-off-by: Henning Heinold Jun 17 13:06:03 Acked-by: Koen Kooi Jun 17 13:06:03 Acked-By: Frans Meulenbroeks Jun 17 13:10:26 stefan_schmidt: life Jun 17 13:10:55 hrw: good, keeping your small family happy is important Jun 17 13:47:54 yeag Jun 17 13:48:00 booted into angstroem with systemd Jun 17 13:48:05 and getty Jun 17 13:48:06 cool Jun 17 13:48:13 yay, nice one Jun 17 13:48:45 okay time for check in Jun 17 13:48:47 hm Jun 17 13:49:00 hm will fix the getty stuff first Jun 17 13:49:05 woglinde: just finished a new build here. classpath is 0.98 and still the same problem. Which package should take care of filling in fake-jdk includes? Jun 17 13:50:27 our oe recipe Jun 17 13:50:30 in jalimo Jun 17 13:50:31 hm Jun 17 13:50:41 did you update jalimo as well? Jun 17 13:50:50 I did yetserday Jun 17 13:50:55 hm Jun 17 13:50:58 let me check what new stuff is in there Jun 17 13:51:01 update to todays Jun 17 13:51:44 hmm, 9 new commits. you guys are busy Jun 17 13:51:58 ah robert updated just in the moment Jun 17 13:52:13 r839 jetzt Jun 17 13:52:39 shcon hier Jun 17 13:52:46 err, already on it Jun 17 13:53:32 hm next commit Jun 17 13:53:44 maybe you should wait 10 minutes or so Jun 17 13:53:50 03Joshua Judson Rosen  07org.openembedded.dev * raefd3cbac5 10openembedded.git/recipes/ (3 files in 2 dirs): Jun 17 13:53:50 foxtrotgps: add recipe for FoxtrotGPS 0.99.4 Jun 17 13:53:50 * also add it to shr feeds Jun 17 13:53:50 Signed-off-by: Martin Jansa Jun 17 13:54:11 woglinde: will do Jun 17 13:58:32 stefan_schmidt: its all fine now Jun 17 13:58:52 thebohemian: yay :) Jun 17 13:59:13 thebohemian: what was it? Jun 17 13:59:37 stefan_schmidt: just built openjdk6 and openjdk6-shark and it did fine. Jun 17 14:00:09 stefan_schmidt: "classpath is 0.98 and still the same problem." -> which one is it? Jun 17 14:02:11 thebohemian: no jni.h in fake-jdk/include in the workdir Jun 17 14:02:26 thebohemian: means openjdk-6 was failing due to missing header Jun 17 14:03:26 thebohemian: but let me try to finish a build with 840 first Jun 17 14:03:44 maybe it was fixed in one of your commits since yesterday Jun 17 14:04:41 stefan_schmidt: unfortunately I havent had this issue before so I think it is unlikely Jun 17 14:04:53 stefan_schmidt: but I'll look it up Jun 17 14:05:10 thebohemian: hmm Jun 17 14:05:31 thebohemian: its in do_compile right here. Either top or flop. will let you knoe Jun 17 14:05:32 know Jun 17 14:06:27 thebohemian: wrt the gcc patches. I can look into handling them but only from 27th (back from vacation) Jun 17 14:07:14 stefan_schmidt: ok, but the patches are in meta-oe already, right? so the buglabs builds have it? Jun 17 14:07:50 thebohemian: thats what I got from woglinde. I saw a big gcc commit yesterday so they should be in. Jun 17 14:08:05 thebohemian: I will look into getting this into OE dev and stable then Jun 17 14:08:19 thebohemian: For all the other jalimo recipes you guys take care? Jun 17 14:08:38 stefan_schmidt: yes Jun 17 14:08:43 thebohemian: ok, good Jun 17 14:09:19 -> A com.buglabs.build.oe/meta-oe/recipes/gcc/gcc-4.3.1/debian/armel-atomic-builtins.dpatch Jun 17 14:09:19 so its in Jun 17 14:10:19 yup Jun 17 14:23:35 thebohemian: it passed! Jun 17 14:23:43 stefan_schmidt: :) Jun 17 14:23:54 thebohemian: seems something between 830 and 840 fixed it even if you never had the problem Jun 17 14:24:29 stefan_schmidt: I was cleaning up the files today. removing cruft for supporting older openjdk releases Jun 17 14:24:50 thebohemian: maybe something like this Jun 17 14:25:12 at least it build again. If the production image goes well I can finally test shark :) Jun 17 14:27:59 stefan_schmidt: for shark you'll have to install a bunch of llvm libraries to get it to work Jun 17 14:28:48 which the recipe unfortunately does not pull in automatically (because of disabled shlibs mechanism - see oe-dev ML) Jun 17 14:30:14 robert dod you see the sugesstion from koen Jun 17 14:30:20 about libjvm Jun 17 14:30:56 thebohemian: yeah, woglinde told me already. thanks Jun 17 14:37:52 woglinde: that looks good. exactly what I need Jun 17 14:55:07 hmmmm Jun 17 14:55:42 Exception RuntimeError: 'maximum recursion depth exceeded while calling a Python object' in ignored Jun 17 14:56:23 anyone have any thoughts on this? wonder if causing an exception in an exception handler or something could result in that Jun 17 14:56:23 hmm Jun 17 15:25:20 anyone feel like reviewing the bitbake logging patchset? Jun 17 15:25:27 (branch) Jun 17 15:58:10 hi Jun 17 15:59:06 hey Jun 17 16:00:42 khem, may i ask you to test my big nptl simplification? I'll test i386 and x86_64, perhaps you could check arm? I can try to briefly check ppc32 and build-test mips (will ask austin to check on his sparc but i think i handle that one fine already) Jun 17 16:01:19 khem, do you have a mips to test on or a readily useable emulator, by chance? Jun 17 16:04:18 khem, overall i like it pretty much. From the feelings, i've removed about 60% with the additional benefit that it handles subarches properly now, supports O=, is more parallel-save and last but not least turned into something that is actually maintainable :) Jun 17 16:16:24 why are acl and attr recipes in udev/ anyway? Jun 17 16:18:26 * kergoth_ was wondering about that too Jun 17 16:29:36 03Mike Westerhof  07org.openembedded.dev * r5a29fcc38a 10openembedded.git/conf/distro/include/slugos.inc: Jun 17 16:29:36 SlugOS: slugos.inc - Fix bluez dependencies Jun 17 16:29:36 No need to build all of bluez if we only depend on the libs. Jun 17 16:30:45 khem, is it just me or is librt handled inelegantly? I mean, isn't it insane to link against threads if i just want to link my test-timer_create binary? Jun 17 16:31:19 i mean either i'm just blind and dumb or whoever did that is an idiot Jun 17 16:32:58 freaking mess. I'll fix that in a separate iteration later on. ARGH Jun 17 16:36:27 blindvt donr know Jun 17 16:36:36 git log Jun 17 17:09:50 hi Jun 17 17:11:51 I've the kernel modules that are not included in an images despite of having: Jun 17 17:12:27 * MACHINE_EXTRA_RRECOMMENDS in the IMAGE_INSTALL Jun 17 17:13:05 *kernel-modules in MACHINE_EXTRA_RRECOMMENDS in the machine config Jun 17 17:13:20 *cleaned and rebuilt the image and looked at the time in the files Jun 17 17:13:26 check sanity Jun 17 17:13:32 bitbake -e image | grep \^IMAGE_INSTALL Jun 17 17:13:37 ok thanks Jun 17 17:14:37 kernel-modules is in Jun 17 17:15:23 checked do_rootfs log? Jun 17 17:15:28 I'll look Jun 17 17:16:35 03Mike Westerhof  07org.openembedded.dev * r4ef5087ed7 10openembedded.git/ (2 files in 2 dirs): Jun 17 17:16:35 SlugOS: package feed changes to drop x11, and unbuildable packages Jun 17 17:16:35 This set of changes removes packages that pull in libx11, and Jun 17 17:16:35 (temporarily) disables packages that won't build, mostly due to Jun 17 17:16:35 gcc 4.4.4 issues. Only tested on LE for the moment. Jun 17 17:16:51 Downloading file:/home/gnutoo/embedded/oe/oetmps/bug/deploy/glibc/ipk/bug/kernel-modules_2.6.27.2+svnr10746-r32.5_bug.ipk. Jun 17 17:16:55 I'll check that Jun 17 17:17:49 all I use is stock oe but the image Jun 17 17:17:55 s/oe/oe.dev Jun 17 17:20:13 http://pastebin.com/mW68E2mB : Depends have all the necessary things Jun 17 17:20:22 I end up only with vfat module Jun 17 17:21:35 I've some collected errors at configuring with opkg Jun 17 17:21:36 http://pastebin.com/bHJRyexM Jun 17 17:21:46 but I don't think it's the cause Jun 17 17:23:51 mmm Jun 17 17:23:58 I bet the answer is in log_do_rootfs Jun 17 17:25:20 because there are occurences of the modules in this log file: Jun 17 17:25:22 + echo 'kernel-modules OPP kernel-module-bmi-lcd-core,;' Jun 17 17:25:43 blindvt: where is the patch Jun 17 17:27:02 mmm maybe that occur after the image creation Jun 17 17:27:10 mwester: can you let me know the build failures that you are seeing on slug using gcc 4.4.4 Jun 17 17:27:17 ah no Jun 17 17:27:18 khem, i'll pass it to you once i'm happy. factoring out unrelated parts ATM (it's big enough as it is) Jun 17 17:27:19 it does a rm Jun 17 17:28:01 + du -k => du: write error Jun 17 17:28:02 khem, they'll appear in tinderbox next time the "blocks" autobuilder runs. Jun 17 17:28:20 * khem does not know to parse tinderbox Jun 17 17:28:58 * GNUtoo|laptop wonder what to do now Jun 17 17:30:29 khem, see this just committed gem, for example (plus ranting about austin a tiny bit while at it ;) Jun 17 17:34:43 btw bitbake -i doesn't work on bitbake master Jun 17 17:36:47 kergoth_, thanks a lot,what should I do now? Jun 17 17:36:52 mwester: which builder should I look into Jun 17 17:37:29 DJWillis, hi Jun 17 17:38:02 Hi GNUtoo|laptop Jun 17 17:38:16 nslu2-linux seems relavent but last build was on April end Jun 17 17:38:29 DJWillis, how is pandora things going? Jun 17 17:38:57 I saw on the blog that people complain about the lack of software => games recipes are more than welcomed Jun 17 17:39:24 I started like this...I wanted wesnoth on my freerunner Jun 17 17:39:54 (I tried oe before but I wanted to do things that are too difficult and failed) Jun 17 17:40:05 oh for fuck's sake, lib/bb/build.py's exception handling really, really needs to be reworked Jun 17 17:40:33 kergoth_, ah it should have failled and didn't? Jun 17 17:41:02 GNUtoo|laptop: next ;-), lots of people moaning, very few 'doing' and I am not a gamer so dont expect many games recipes from me ;). Core stuff mostly from me, the image is Angstrom/Xfce with a few small mods really, try and keep it close to mainline. Jun 17 17:41:16 just doing odd things. catching and re-raising other things and losing the useful information along the way Jun 17 17:41:26 and i'm sick of seeing like 47 error messages when one task fails Jun 17 17:41:33 yes, i realize the function failed, and that means the task failed,a nd... Jun 17 17:41:34 bite me Jun 17 17:42:00 DJWillis, ok Jun 17 17:42:26 wesnoth needs to be updated...it's waiting in my todo list.... Jun 17 17:42:55 hmmm Jun 17 17:43:33 GNUtoo|laptop: people have 1.8 on the Pandora but it's not built using OE :( Jun 17 17:43:38 so what should I do? fix the python problem,or fix the error,or am I off topic and you weren't talking to me Jun 17 17:43:45 DJWillis, ah ok Jun 17 17:43:57 maybe a cp will do the trick Jun 17 17:45:03 DJWillis, I've an idea, create something to make pnds with oe Jun 17 17:45:16 like a class or something like that Jun 17 17:45:23 and with BBCLASS_EXTEND Jun 17 17:45:33 you would add pnd functionality Jun 17 17:45:53 PND are static right? Jun 17 17:46:09 * GNUtoo|laptop must fix his image Jun 17 17:49:13 GNUtoo|laptop: I did consider that, PND's are quite basic, there main reason to exist is to deal with removable media and being pulled without any 'nice stuff' like unmounting ;). Jun 17 17:49:36 ok Jun 17 17:50:05 if they are basic it shouldn't be too difficult Jun 17 17:51:16 GNUtoo|laptop: no, it would be very easy to do the basics, I guess it needs work but the PND daemon runs fine on most hardware I run OE on ;) Jun 17 17:51:42 ah there is a daemon Jun 17 17:52:37 GNUtoo|laptop: yep, that does the iso/squashfs and read/write munging using AUFS. Jun 17 17:52:45 ok Jun 17 17:53:04 AUFS is the successor of unionfs? Jun 17 17:53:36 ah ok yes Jun 17 17:56:27 GNUtoo|laptop: sort of 'upgraded unionfs' ;), well in some circles ;) Jun 17 17:56:44 ok Jun 17 17:57:59 khem: blocks Jun 17 17:59:01 but it will (falsely) claim the builds are good, because we use the -k flag to bitbake on the package feeds. I'll post the URL to one of the failed packages when I get a few moments and can find it, but I'd like to get the package feeds up and running before I re-enable the broken packages. :) Jun 17 18:02:30 sure makes sense Jun 17 18:06:47 kergoth_, was that message for me? is it the bug I'm hitting? oh for fuck's sake, lib/bb/build.py's exception handling really, really needs to be reworked Jun 17 18:11:55 maybe the best way is to boot it and figure on target Jun 17 18:12:00 hopefully I've serial Jun 17 18:12:36 hi Jun 17 18:12:56 because kernel-modules should bring the modules Jun 17 18:12:57 hi Jun 17 18:15:26 hi woglinde Jun 17 18:16:45 no, that was referring to what i'm working on for work at the moment, giant pain in the ass Jun 17 18:16:52 ah ok Jun 17 18:17:15 kergoth_, do you know what I should look for in the log_do_rootfs? Jun 17 18:17:19 khem, how does your configury verification go btw? Any news or delta to apply them or poke pb_ to ACK them, too? :) Jun 17 18:20:19 blindvt: I have pulled them in Jun 17 18:20:27 blindvt: building with them Jun 17 18:20:52 but fighting other daemons along way I want to get them in my build successfully Jun 17 18:21:03 consider them in sometimes soon Jun 17 18:22:08 basically it does: Jun 17 18:22:11 Installing kernel-modules (2.6.27.2+svnr10746-r32.5) to root... Jun 17 18:22:13 and then Jun 17 18:22:22 it continues with the other stuff like initscripts Jun 17 18:22:29 I really don't know why Jun 17 18:22:37 maybe I should give -DDD a try? Jun 17 18:23:00 khem, k. I'll wait and poke you and you folks again. I take it that i desparately need this to get a sane state to base off, i take it Jun 17 18:23:25 bah Jun 17 18:23:43 i take it that you see that i desparately need this to get a sane state to base off Jun 17 18:23:47 hi can some one suggest me how to build a custom kerenl (with own kernel souces ) for oe (DIST =Angstrom MACHINE=omap3evm ) Jun 17 18:24:13 sreddy_, there are several approach, does the kernel change often? Jun 17 18:24:40 sreddy: srctree and gitver classes help Jun 17 18:24:59 no Jun 17 18:25:09 hi khem ;) Jun 17 18:25:09 kernel doesnt change often Jun 17 18:25:30 sreddy_, so make a recipe for it Jun 17 18:25:50 khem, and don't forget to also push your uClibc AUTOREV hunks on top of it (or i dig out mine if needed) Jun 17 18:26:40 * blindvt can't type anything but asm or C today, apparently Jun 17 18:26:50 mmm -DDD doesn't give that much infos on do_rootfs Jun 17 18:26:54 only the log do Jun 17 18:27:06 khem, you know what i mean Jun 17 18:29:22 blindvt: I pushed one of yours Jun 17 18:29:24 earlier Jun 17 18:31:07 khem, the retain "git", yes. Thanks for that. I was talking about auto-tracking master, short of a PREFERRED_VERSION_uclibc_git=0x0815 (if that works at all, didn't look nor would care TBH. master is master) Jun 17 18:33:07 i am a newbie to this ... I did the following steps placed my kernel sources (.tar.bz ) in the "$OEBASE/sources " and created "$OEBASE/local " directory with the directory structure like "local/classes" "local/conf" "local/packages" Jun 17 18:36:36 blindvt: I did not want to do that Jun 17 18:36:44 and only bump it manually Jun 17 18:36:48 after testing Jun 17 18:37:41 khem, i see, ok Jun 17 18:40:04 blindvt: sometimes uclibc master breaks Jun 17 18:40:20 and OE users might not be happy Jun 17 18:41:39 sreddy_, hmm so you don't have a git repository or something like that for your kernel? Jun 17 18:41:42 i am a newbie to this ... I did the following steps placed my kernel sources (.tar.bz ) in the "$OEBASE/sources " and created "$OEBASE/local " directory with the directory structure like "local/classes" "local/conf" "local/packages" and trying to modify the given recipe from "linux-omap-psp_2.6.32.bb" Jun 17 18:41:48 no Jun 17 18:41:58 as of now i dont have Jun 17 18:42:12 maybe you should create one then Jun 17 18:42:27 becuase how other would be able to fetch from it Jun 17 18:42:38 khem, i agree. It's better if there is an automated oe testbad that verifies against the major 6 arches before random master changes are pushed down anyones throat in OE Jun 17 18:42:41 * blindvt nods khem Jun 17 18:42:43 if there is not revision control repository Jun 17 18:42:58 do you have a tarball hosted somewhere Jun 17 18:43:03 i use svn Jun 17 18:43:04 or is it all local Jun 17 18:43:10 ah ok Jun 17 18:43:18 so point your recipe to the svn of your kernel Jun 17 18:43:26 but for testing currently I ma not suing Jun 17 18:43:33 ok Jun 17 18:43:41 basically there are 2 things: Jun 17 18:43:57 *make a recipe of a stable version of your kenrel that is hosted in a public place Jun 17 18:44:08 *use the openembedded toolchain to build and test it locally Jun 17 18:44:20 I bet you want both Jun 17 18:44:41 yes .. I would like to go for easy way first Jun 17 18:45:01 sreddy_, so build a toolchain with openembedded Jun 17 18:45:05 like this: Jun 17 18:45:09 bitbake devshell Jun 17 18:45:22 then you will get a toolchain + a devshell Jun 17 18:45:28 i did that for console-image Jun 17 18:45:33 ok Jun 17 18:45:39 but you've got the devshell? Jun 17 18:45:46 it's a script that you need to source Jun 17 18:45:55 it put the compiler etc... in your path Jun 17 18:46:11 then you can go into your kernel directory and do: Jun 17 18:46:41 make ARCH=something CROSS_COMPILE=arm-oe- LDFLAGS="" Jun 17 18:46:46 or things like this Jun 17 18:46:58 that would compile your kernel locally Jun 17 18:47:03 bitbase how to check if I gort devshell Jun 17 18:47:17 sreddy_, the devshell is in : Jun 17 18:47:42 deploy/glibc/addons/ Jun 17 18:47:48 but if you didn't bitbake it Jun 17 18:47:55 you won't have it Jun 17 18:48:10 if you've already got a toolchain just bitbake it now Jun 17 18:48:14 it won't take too long Jun 17 18:49:13 yes i dont have devshell .. i ll bitbake it now Jun 17 18:50:39 ahh I've got an idea...grep the tasks for this machine extra recommand Jun 17 18:51:34 ah ok I got it Jun 17 18:51:38 task-base...hmmm Jun 17 18:51:58 when i do console doesnt it include task-base Jun 17 18:52:14 i will do that Jun 17 18:52:19 sreddy_, I wasn't talking to you Jun 17 18:52:26 oh i am sorry Jun 17 18:52:26 sreddy_, I've an issue with an image Jun 17 18:52:28 np Jun 17 18:54:06 hmmm Jun 17 18:54:21 kergoth, maybe I've an idea Jun 17 18:54:31 it's RRECOMMANDS not RDEPENDS Jun 17 18:54:46 RRECOMMENDS_task-machine-base = "${MACHINE_EXTRA_RRECOMMENDS}" Jun 17 18:54:58 MACHINE_EXTRA_RRECOMMENDS = "marvell-gspi-fw marvell-sdio-fw kernel-modules" Jun 17 19:41:00 aha Jun 17 19:41:06 only newer images got that Jun 17 19:41:13 older ones got all modules Jun 17 19:42:44 so here's my question Jun 17 19:43:02 does recomands changed recently? Jun 17 19:45:39 any pulseaudio experts around? I built and did an "opkg install pulseaudio" but I'm not finding much in the way of installed binaries Jun 17 19:46:29 awozniak: with pulseaudio you may consider that a blessing ;-) Jun 17 19:47:10 Package pulseaudio (0.9.15-r8.4.5) is installed on root and has the following files: Jun 17 19:47:10 /usr/libexec/pulse/proximity-helper Jun 17 19:47:17 and nothing else Jun 17 19:47:28 opkg list | grep pulse Jun 17 19:47:48 woglinde, hi do you know recommands well? especially the machines ones? Jun 17 19:47:53 awozniak: that said, there is a lot more to getting pulseaudio working then just the pulseaudio package IIRC, I backed out of using it in my images as it was a pain in the rear. Jun 17 19:48:04 I've strange things Jun 17 19:48:08 woglinde: a mountain of things are listed. do I need to install them all? Jun 17 19:48:46 hmmm pulseaudio, I remember having to deal with too Jun 17 19:51:21 hmmm Jun 17 19:51:31 * GNUtoo|laptop thinks again Jun 17 19:57:07 DJWillis: any clues for me? Jun 17 20:01:27 awozniak: using alsa I guess? the packages that mention alsa and pulse would be my guess. Hmmm, pulseaudio-alsa-wrapper pulseaudio-esd-wrapper pulseaudio-module-gconf libasound-module-ctl-pulse libasound-module-pcm-pulse at a guess. Jun 17 20:02:20 thanks, I'll try those Jun 17 20:06:49 ah yes you need modules too Jun 17 20:06:58 like alsa,and a transport Jun 17 20:07:01 s/transport/ Jun 17 20:07:10 a way to connect your app to pulse Jun 17 20:13:25 # /etc/init.d/pulseaudio start Jun 17 20:13:25 Starting audio server: E: cmdline.c: Invalid resample method 'sinc-fastest'. Jun 17 20:13:25 E: main.c: Failed to parse command line. Jun 17 20:13:25 (failed.) Jun 17 20:13:35 any more clues? Jun 17 20:17:57 seems it likes "src-sinc-fastest" not "sinc-fastest" Jun 17 20:20:51 ah... modules Jun 17 20:50:49 mmm Jun 17 20:50:58 in newer images kernel-modules are not even installed Jun 17 20:50:59 but Jun 17 20:51:10 the strange thing is that marvell firmware is Jun 17 20:51:24 MACHINE_EXTRA_RRECOMMENDS="marvell-gspi-fw marvell-sdio-fw kernel-modules" Jun 17 21:06:31 03Chris Larson  07org.openembedded.dev * raa8bef28a2 10openembedded.git/ (7 files in 2 dirs): Jun 17 21:06:31 Indentation fixups, caught by python -t Jun 17 21:06:31 Signed-off-by: Chris Larson Jun 17 21:06:43 03Chris Larson  07org.openembedded.dev * ra84b681daa 10openembedded.git/conf/bitbake.conf: Jun 17 21:06:43 Drop old unused crufty MANIFEST variable Jun 17 21:06:43 Signed-off-by: Chris Larson Jun 17 21:08:29 sigh Jun 17 21:10:05 how does MACHINE_EXTRA_RRECOMMENDS works on the task/image/opkg side Jun 17 21:16:01 hmm, a 2to3 run and a few fixups and bitbake seems to be fine in python3, other than all the python in OE of course.. thats good Jun 17 21:16:20 kergoth: thats good news Jun 17 21:16:32 kergoth: I heard python 3 is sloow Jun 17 21:17:28 * kergoth_ hasn't looked at it much yet Jun 17 21:18:06 * woglinde_ is think how to get SERIAL_CONSOLE var into systemd getty stuff Jun 17 21:18:15 sigh.... Jun 17 21:18:28 GNUtoo|laptop: see task-base.bb Jun 17 21:18:35 that's what I'm in Jun 17 21:18:44 there is: Jun 17 21:18:54 woglinde: look into shadow :) Jun 17 21:18:55 task-base-extended -> Jun 17 21:19:06 khem hm? Jun 17 21:19:14 RDEPENDS -> task-base Jun 17 21:19:58 task-base->RDEPENDS on task-machine-base Jun 17 21:20:23 and finally: Jun 17 21:20:26 RRECOMMENDS_task-machine-base = "${MACHINE_EXTRA_RRECOMMENDS}" Jun 17 21:20:34 but I don't know how RRECOMMENDS work Jun 17 21:20:43 khem shadow dont helps Jun 17 21:21:12 * GNUtoo|laptop looks if the manual has something Jun 17 21:22:03 manual has only a description of RRECOMMENDS Jun 17 21:22:42 now with newer images IMAGE_INSTALL don't even has the MACHINE_EXTRA_RRECOMMENDS content Jun 17 21:22:49 anymore Jun 17 21:23:03 rrecs aren't magic Jun 17 21:23:17 I know but I don't know where to look for Jun 17 21:23:18 they're just rdepends that can be uninstalled later with opkg without removing the package that pulled them in Jun 17 21:23:33 ahh ok nice Jun 17 21:23:52 I also heard that they are optional Jun 17 21:23:58 at least in deb distros Jun 17 21:24:03 you can enable or disable them Jun 17 21:24:15 you can tell the package manager whether you want them pulled in, iirc Jun 17 21:24:24 its a funciton of the tool, not the metadata Jun 17 21:24:26 and in the case of the image how do I do that Jun 17 21:24:30 ? Jun 17 21:24:33 opkg always installs rrecs Jun 17 21:24:40 it isn't dpkg Jun 17 21:24:43 ah ok Jun 17 21:24:48 so there is a bug Jun 17 21:24:55 theses modules should be installed Jun 17 21:25:01 there's a separate variable that lets you make it remove them from the image if they get sucked in and you dont want them Jun 17 21:25:01 and they were some revs ago Jun 17 21:25:08 so use bisect and figure it out already Jun 17 21:25:08 ah ok Jun 17 21:25:17 git bisect, binary search of the commits between the working and nonworking Jun 17 21:25:23 you'll find the exact commit tha tbroke it Jun 17 21:25:26 ok thanks a lot Jun 17 21:25:31 see git help bisect Jun 17 21:25:32 np Jun 17 21:26:04 git bisect start , then test, then git bisect good or git bisect bad, depending on whether it worked or not, and keep doing that until it finishes Jun 17 21:26:36 ok thanks a lot, Jun 17 21:26:45 I also heard that there is a function like a log Jun 17 21:27:06 that tells you when you pulled,more specifically Jun 17 21:27:17 last pull was at revision K Jun 17 21:27:26 the one before at K-10 Jun 17 21:27:27 etc... Jun 17 21:27:38 it doesn't care about pulls, exactly Jun 17 21:27:49 what you can look at is the log of a ref.. git reflog Jun 17 21:27:56 it shows each time master changed Jun 17 21:28:03 ok thanks a lot Jun 17 21:28:05 when you pull, that merges, which updates master to the new point Jun 17 21:28:18 so essentially thats it, yeah, but there are other things that could change a ref too Jun 17 21:28:21 i.e. git reset, git rebase Jun 17 21:28:58 thanks a lot!!!! Jun 17 21:29:20 np Jun 17 21:29:38 using reflog to find the 'good' commit for the bisect? :) Jun 17 21:29:53 yes that's what I want to do Jun 17 21:30:02 use reflog for finding the good range faster Jun 17 21:30:51 can also git bisect skip, for if your test fails for an unrelated reason, so you don't know whether its good or bad (i.e. someone broke something else :) Jun 17 21:30:53 handy tools Jun 17 21:30:54 <3 git Jun 17 21:31:20 I loves git too Jun 17 21:32:15 * kergoth_ wonders how much work itd be to add a prototype of lua scripting in recipes instead of python, at least for the ${@} thing.. Jun 17 21:33:56 kergoth_, big question: Jun 17 21:34:00 hmm.. alternatively, turn ${@} into a python based dsl, just lock down more of what it can do, and give it a different access mechanism to the metadata.. but it's difficult to fully sandbox eval Jun 17 21:34:03 * kergoth_ thinks Jun 17 21:34:06 should I wipe tmpdir? Jun 17 21:34:09 for git bisect Jun 17 21:34:12 and start from bad Jun 17 21:34:28 because PR,packages versions etc...will get in my way Jun 17 21:34:33 no? Jun 17 21:34:52 ah no sorry Jun 17 21:34:54 could, not sure how the rootfs generation code reacts to old versions of things showing up when newer ones already exist Jun 17 21:34:56 I must be tired Jun 17 21:35:04 they'll *build* for sure, since the stamps wont exist Jun 17 21:35:09 (for the old pv/pr) Jun 17 21:35:09 ah ok Jun 17 21:35:17 so it should work Jun 17 21:35:19 I'll try Jun 17 21:37:50 kergoth_: why lua Jun 17 21:38:26 it's a good fit for that sort of thing, and isn't as difficult to lock down Jun 17 21:46:43 hm let see if my magic worked Jun 17 21:53:48 cool Jun 17 21:53:55 only small glintches Jun 17 22:04:30 yeah Jun 17 22:04:32 okay Jun 17 22:04:35 systemd is comming Jun 17 22:27:23 it doesn't work Jun 17 22:27:27 every -native fails Jun 17 22:27:31 wth do_patch Jun 17 22:27:45 ERROR: Task 1117 (virtual:native:/home/gnutoo/embedded/oe/org.openembedded.dev/recipes/glib-2.0/glib-2.0_2.24.0.bb, do_patch) failed with 256 Jun 17 22:28:13 I've packaged staging Jun 17 22:28:29 that would take ages Jun 17 22:28:33 293 of 3303 Jun 17 22:29:35 hm 2 hours Jun 17 22:29:41 ? Jun 17 22:30:48 from 293 to 3303 Jun 17 22:31:01 ah no more Jun 17 22:31:06 GNUtoo|laptop: Does quilt exist on your system Jun 17 22:31:10 I've to do everything manually Jun 17 22:31:21 I'll look Jun 17 22:31:30 GNUtoo|laptop: I would suggest to rebuild from scratch once Jun 17 22:31:42 ok Jun 17 22:31:45 I'll do that Jun 17 22:31:53 I bet there is no db to save Jun 17 22:32:10 ahhh maybe I got it Jun 17 22:32:14 I know why Jun 17 22:32:25 NOTE: Task failed: [Errno 17] File exists Jun 17 22:32:29 it's packaged staging Jun 17 22:32:39 basically the patch is saved in the packaged workdir Jun 17 22:32:44 and get redeployed Jun 17 22:32:52 and when redeployed it finds another patch Jun 17 22:32:58 with the same name Jun 17 22:33:14 khem I do the last check for systemd Jun 17 22:33:19 than I will push it Jun 17 22:33:50 unfornatly reboot/halt with busybox isnt working Jun 17 22:33:57 but startup is fine Jun 17 22:34:19 woglinde: ok Jun 17 22:35:36 or I could clean all native stuff Jun 17 22:37:56 ah ok I'll do it by hand Jun 17 22:38:11 it seem not to have so much native packages to rebuild Jun 17 22:43:53 mmm now dropbear fails Jun 17 22:43:57 this is clearly not native Jun 17 23:15:40 woot Jun 17 23:15:43 it worked Jun 17 23:15:48 commiting now Jun 17 23:19:45 03Henning Heinold  07org.openembedded.dev * rb5fdd85422 10openembedded.git/recipes/libcgroup/libcgroup_0.36.2.bb: libcgroup: added libcgroup to openembedded Jun 17 23:19:47 03Henning Heinold  07org.openembedded.dev * r119349c48a 10openembedded.git/recipes/systemd/ (6 files in 2 dirs): Jun 17 23:19:47 systemd: add systemd to openembedded Jun 17 23:19:47 * not fully functional yet Jun 17 23:19:47 * needs udev-154 and cgroup enabled in kernel Jun 17 23:19:48 * halt/reboot don't work with busybox Jun 17 23:19:55 03Henning Heinold  07org.openembedded.dev * r1b78440c8d 10openembedded.git/recipes/udev/ (18 files in 8 dirs): Jun 17 23:19:55 udev: add version 154 Jun 17 23:19:55 * not fully testet yet so checking with DEFAULT_PREFERENCE = "-1" Jun 17 23:19:55 * systemd needs it Jun 18 01:04:29 meh **** ENDING LOGGING AT Fri Jun 18 02:59:57 2010