**** BEGIN LOGGING AT Wed Apr 22 02:59:57 2009 Apr 22 06:35:55 good morning Apr 22 06:37:20 morning Apr 22 06:57:55 good morning Apr 22 07:19:05 good morning Apr 22 07:29:14 morning. Apr 22 07:29:25 Anyone around able to help with a failing bitbake? Apr 22 07:29:36 I am trying to bb netsurf w/o luck. Apr 22 07:32:11 don't ask, just post the log (in pastebin) and people will react if they can help (or want to see more info) Apr 22 08:25:05 morning Apr 22 08:38:11 03Dirk Opfer  07org.openembedded.dev * ree7238c8c1 10openembedded.git/recipes/argp-standalone/argp-standalone_1.3.bb: Apr 22 08:38:11 argp-standalone: new recipe, version 1.3 Apr 22 08:38:11 Argp is an interface for parsing unix-style argument vectors. This is only needed for uclibc Apr 22 08:38:16 03Dirk Opfer  07org.openembedded.dev * r6a64f6e927 10openembedded.git/recipes/pthsem/ (files/ldflags.patch pthsem_2.0.7.bb): Apr 22 08:38:16 pthsem: new recipe, version 2.0.7 Apr 22 08:38:16 GNU Portable Threads replacement with semaphore support Apr 22 08:38:17 03Dirk Opfer  07org.openembedded.dev * rb2c58f438d 10openembedded.git/recipes/eibd/ (eibd_0.0.4.bb files/eibd): Apr 22 08:38:20 eibd: new recipe, version 0.0.4 Apr 22 08:38:22 Eibd provides an interface to the EIB / KNX bus Apr 22 08:38:24 03Dirk Opfer  07org.openembedded.dev * rfbc3617d2d 10openembedded.git/conf/checksums.ini: checksums.ini: add some more checksums Apr 22 08:38:27 03Dirk Opfer  07org.openembedded.dev * r8ad50430b0 10openembedded.git/recipes/linknx/ (files/configure-libcurl.patch linknx_0.0.1.26.bb): Apr 22 08:38:30 linknx: new recipe, version 0.0.1.26 Apr 22 08:38:32 Linknx is an automation platform providing high level functionalities to EIB/KNX installation Apr 22 08:38:36 03Dirk Opfer  07org.openembedded.dev * rb2bbfbf9df 10openembedded.git/recipes/owfs/ (files/owhttpd files/owserver owfs_2.7p21.bb): Apr 22 08:38:39 owfs: new recipe, version 2.7p21 Apr 22 08:38:41 OWFS is an easy way to use the powerful 1-wire system of Dallas/Maxim Apr 22 09:12:32 03Steffen Sledz  07org.openembedded.dev * rb3ee52abc5 10openembedded.git/recipes/initscripts/ (files/hipox/reboot initscripts_1.0.bb): Apr 22 09:12:32 initscripts: modifications for hipox machine Apr 22 09:12:32 - add kexec handling in /etc/init.d/reboot Apr 22 09:12:32 - make /sys available again for /etc/init.d/reboot Apr 22 10:07:49 03Michael 'Mickey' Lauer  07org.openembedded.dev * ra1c39a650d 10openembedded.git/recipes/freesmartphone/libgsm0710mux_git.bb: libgsm0710mux: needs libfsotransport now Apr 22 10:07:50 03Michael 'Mickey' Lauer  07org.openembedded.dev * rd0e0791eb9 10openembedded.git/ (2 files in 2 dirs): vala-dbus-binding-tool-native: new recipe; a dbus interface generator Apr 22 10:07:52 03Michael 'Mickey' Lauer  07org.openembedded.dev * r9cfc8432cd 10openembedded.git/recipes/freesmartphone/ (fso-term_git.bb libfsotransport_git.bb): libfsotransport: new recipe; transport library for FSO. add as dependency for fso-term Apr 22 10:25:16 03Koen Kooi  07org.openembedded.dev * r3812ebdea8 10openembedded.git/ (3 files in 2 dirs): Apr 22 10:25:16 glibc: add 2.9 Apr 22 10:25:16 * runtime tested on beagleboard/angstrom Apr 22 10:25:23 03Koen Kooi  07org.openembedded.dev * r756b69f818 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Apr 22 11:01:30 03Florian Boor  07org.openembedded.dev * rb3ab078b3c 10openembedded.git/conf/machine/micro2440.conf: micro2440.conf: Improve jffs2 image parameters. Apr 22 11:01:32 03Florian Boor  07org.openembedded.dev * r6a1b13cfbc 10openembedded.git/recipes/lowlevel_topas910/lowlevel-topas910_1.0.bb: New recipe: lowlevel-topas910, a mini bootloader for the Topas910 Apr 22 11:01:32 03Florian Boor  07org.openembedded.dev * rafa8c86350 10openembedded.git/recipes/linux/linux-2.6.29/micro2440/defconfig: linux-2.6.29: Tune micro2440 defconfig a little bit, build in loop for loop root. Apr 22 11:01:33 03Florian Boor  07org.openembedded.dev * r470ba9654f 10openembedded.git/recipes/linux/linux-2.6.26/topas910/defconfig: linux: Add topas910 defconfig for 2.6.26. Apr 22 11:01:36 03Florian Boor  07org.openembedded.dev * rc09525bd6d 10openembedded.git/conf/machine/topas910.conf: New machine: Topas910 - Toshiba TMPA910CRAXBG Starter Kit Apr 22 11:01:39 03Florian Boor  07org.openembedded.dev * r695bef5347 10openembedded.git/recipes/linux/linux_2.6.26.bb: linux: 2.6.26 Add inital support for Topas910, we'll add a better kernel later. Apr 22 11:01:42 03Florian Boor  07org.openembedded.dev * rdd8a979f49 10openembedded.git/conf/checksums.ini: checksums.ini: Add original Topas patch and bootloader. Apr 22 11:39:02 Hello Apr 22 11:39:18 why do i get this message when i want to compile gettext for my arm? Apr 22 11:39:22 http://vim.pastey.net/112829 Apr 22 11:42:09 hi, all! Apr 22 11:44:23 how could I find reason for error ERROR: Multiple .bb files are due ti be built, which each provide ... ? Apr 22 11:45:13 how could I find where I am wrong with my custom .bb? Apr 22 11:48:32 is opkg_svn currently unbuildable in .dev? Apr 22 11:50:45 slapin: probably setting PREFERRED_PROVIDER Apr 22 12:02:07 do anybody know how to make bitbake print out environment of all these PREFERRED_PROVIDER_x vars, I've set providers in several places but I'm not sure if it is not overrided somewhere Apr 22 12:11:43 gremlin[it]: hi, I completed x11-image build with your config and I got bluez-utils error again Apr 22 12:12:05 gremlin[it]: I think you are not really using stable Apr 22 12:13:50 mckoan, how can i do a diff between two branch (local and remote) ? Apr 22 12:16:49 slapin_: bitbake -b recipe.bb -e|grep ^PREF Apr 22 12:19:39 hrw, thanks a lot! Apr 22 12:32:26 gremlin[it]: I would start with git status :-) Apr 22 12:32:43 gremlin[it]: what git branch says there ? Apr 22 12:35:32 gremlin[it]: BTW git diff , or git diff --cached Apr 22 12:35:47 on branch stable ... and my stable is a checkout of origin/stable/2009 Apr 22 12:35:59 git diff --cached ... report nothing Apr 22 12:36:34 what git branch says there ? Apr 22 12:36:37 either `git diff origin/stable/2009` report nothing Apr 22 12:37:02 show me output Apr 22 12:37:06 just "stable" ... i did a checkout from origin/stable/2009 to origin Apr 22 12:37:16 just "stable" ... i did a checkout from "origin/stable/2009" to "stable" Apr 22 12:37:39 [gremlin@sesha openembedded]$ git status Apr 22 12:37:39 # On branch stable Apr 22 12:37:39 nothing to commit (working directory clean) Apr 22 12:38:14 how can you be sure that your named 'stable' is equal to "origin/stable/2009" ? Apr 22 12:39:38 i try with `git diff origin/stable/2009` give no output ... Apr 22 12:40:09 if i try with other, like `git diff origin/org.openembedded.dev`, i obtain lot of differences Apr 22 12:40:31 ok me too Apr 22 12:40:51 I don't understand what is different Apr 22 12:41:24 me too :( Apr 22 12:42:31 64bit debian kernel? I couldn't believe! Apr 22 12:43:52 hahahah Apr 22 12:44:11 is the only diff beetween us Apr 22 12:44:55 mckoan, i use fedor aon both machines ..one is an old P4 (32bit) other is an AMD 64bit ... Apr 22 12:47:29 gremlin[it]: git log origin/stable/2009..HEAD Apr 22 12:47:57 hrw thanks .... empy output Apr 22 12:47:58 * mckoan wonders if anyone else would try building qemux86/x11-image in stable Apr 22 12:48:24 mckoan: it builds here - for glibc and uclibc Apr 22 13:32:23 65 minutes to build console-image from scratch :-) Apr 22 13:33:13 I don't think build time can be raised as a serious objection to OE any more Apr 22 13:33:52 cbrake: http://blog.windfluechter.net/index.php?/archives/420-guid.html - and people complain that OE takes lot of time Apr 22 13:35:57 am I doing something completely wrong, as it took me over 7 hours to build the console image? Apr 22 13:36:34 I think it depends a little bit on what machine you are using Apr 22 13:36:50 and settings of build Apr 22 13:37:05 hrw: interesting link -- I never have found native build solutions practical for real work, where I need to support multiple people building, etc Apr 22 13:37:44 koobe: my machine did build for 4 machines (4 archs) with console-image x11-image base-image in 3.5h Apr 22 13:37:49 http://home.haerwu.biz:8010/ Apr 22 13:38:30 koobe: I'm running on a core-i7 with PARALLEL_MAKE = "-j 5", and BB_NUMBER_THREADS = "4" Apr 22 13:39:05 koobe: CPU meter probably averages 50% Apr 22 13:39:13 it was a vmware virtual machine on a 2.4ghz intel core2 cpu Apr 22 13:39:24 BB_NUMBER_THREADS = "6" Apr 22 13:39:29 PARALLEL_MAKE = "-j 8" Apr 22 13:39:36 on my core2quad Apr 22 13:39:40 but I think that did include the downloads as well, next time would have been faster without them, right Apr 22 13:39:50 koobe: yep Apr 22 13:40:19 cbrake: I'm still using BB_DEFAULT_TASK = "buildall" Apr 22 13:40:29 dunno if it slows down more Apr 22 13:40:30 does the downloading mechanism have proper locking, so that I could share the downloads directory without conflicts? Apr 22 13:40:40 koobe: can share Apr 22 13:40:52 koobe: I have one DL_DIR for all users/builds Apr 22 13:41:03 ok, great. Apr 22 13:41:10 koobe: you really get the speedup from running things in parallel. Its easy to imagine a core-i7 being 5-6 times faster than a single core VM Apr 22 13:41:16 hrw: canyou comment on http://article.gmane.org/gmane.comp.handhelds.openembedded/16818 Apr 22 13:41:41 is buildall still necessary or got merged...I perhaps missed it Apr 22 13:42:16 I do not have that set and it works Apr 22 13:42:43 mm, I think I should worry, having a console-image from scratch in about 2 - 3 hours ( on 6-core virtual machine, with parallel_make = "-j 4" and BB_NUMBER_THREADS = "6") Apr 22 13:44:36 tsjsieb: virtual... Apr 22 13:44:43 tsjsieb: what was real hardware? Apr 22 13:46:04 It's Intel(R) Xeon(R) CPU E5345 @ 2.33GHz Apr 22 13:46:27 are there any things under /tmp that I could share or re-use like the DL_DIR. So that all users wouldn't have to build from scratch, but only their application and then the image. Apr 22 13:46:57 koobe: DL_DIR can be anywhere - I have it in ~/devel/sources/dl_dir Apr 22 13:47:05 and it takes 24GB here Apr 22 13:47:26 ok, I see. but how about the intermediate files, like the staging and packages. Apr 22 13:47:41 koobe: better not Apr 22 13:48:15 ok, so it's better to bite the bullet and have everyone build from scratch for the first time? Apr 22 13:48:30 gremlin[it]: I started a completely new build on a different i686 ubuntu machine Apr 22 13:50:57 mckoan, if then don't fail is debian ! ... ghghghghgh Apr 22 13:55:41 ah, about rebuild from packaged staging... Apr 22 13:55:59 strange thing..compiler is not found... Apr 22 13:56:10 from scratch /build/tmp/angstrom/cross/armv5te/ Apr 22 13:56:19 pstage is /build/tmp/angstrom/cross Apr 22 13:56:49 seems like 'armv5te' suffix is not kept Apr 22 14:23:29 re Apr 22 14:23:30 * denix|away is away: I'm not here Apr 22 14:23:35 * denix|away is back (gone 00:00:02) Apr 22 14:38:57 mickeyl: good morning Apr 22 14:39:06 good morning pb_ ! Apr 22 14:39:21 morning Apr 22 15:20:42 03Michael 'Mickey' Lauer  07org.openembedded.dev * r82ede6b24b 10openembedded.git/recipes/bustle/bustle-dbus-monitor_0.2.0.bb: Apr 22 15:20:42 bustle-dbus-monitor: new recipe; monitoring utility for the Bustle D-Bus suite Apr 22 15:20:42 TODO: add eavesdropping magic in postinstall/prerm Apr 22 15:20:52 03Michael 'Mickey' Lauer  07org.openembedded.dev * r71f70d7be4 10openembedded.git/recipes/dbus/ (dbus.inc dbus_1.2.1.bb): dbus: bump default timeout value Apr 22 15:27:39 alphaone: here we are Apr 22 15:27:43 minus eavesdropping Apr 22 15:28:47 hi mickeyl Apr 22 15:28:51 morning Apr 22 15:29:01 hi kergoth Apr 22 15:29:27 yo florian , hi kergoth Apr 22 15:30:25 hi kergoth mickeyl florian Apr 22 15:32:45 mickeyl: ? Apr 22 15:32:49 Ah, great :-) Apr 22 15:33:17 mickeyl: I think I found the kobject notify issue with frameworkd Apr 22 15:33:26 alphaone: ah, good. what's it? Apr 22 15:33:38 mickeyl: let me check first Apr 22 15:35:47 never mind Apr 22 15:37:49 mickeyl: a console-bild from yesterday had something to complain about user 'messagebus' Apr 22 15:38:06 cannot be precise here, was on first boot however Apr 22 15:38:51 hmm, need more info :) Apr 22 15:39:34 eh, gerne, later, in the meantime google 'groups passwd dbus' Apr 22 15:40:01 debian + ubuntu bugs Apr 22 15:41:02 are the groups created by a post_install script? never checked this Apr 22 15:41:46 yes, they are Apr 22 15:41:54 iirc it was a missing entry in passwd or groups Apr 22 15:42:03 will check later on Apr 22 15:42:41 for sure was about mesagebus :-) Apr 22 15:42:51 with 2 s :-) Apr 22 15:43:43 I think these are udev bugs, perhaps. Apr 22 15:44:18 Much discussion about udev attempting to resolve group names when rules are parsed, or when rules are resolved. Apr 22 15:44:31 ah, for sure I have one.. the dev tarball is read-only... Apr 22 15:44:49 But I think messagebus should be in the passwd and/or group file once dbus is in the image. Apr 22 15:45:30 mwester: seems sane thing Apr 22 15:46:09 The udev workaround I did was to add group names that udev complained about to the group file in the image creation postprocessing step; hacky, but it works with the current udevs. Apr 22 15:46:52 * mickeyl considers following the OpenWRT folks advice and forgetting about udev Apr 22 15:47:02 they seem to have much success with hotplug2 Apr 22 15:47:05 I'll untar a fresh rootfs and log the first boot Apr 22 15:47:49 about the device tarball, this is created on first run, isnt? Apr 22 15:50:47 ya Apr 22 15:53:04 huh, hotplug2 seems interesting indeed Apr 22 15:54:59 eep, local root priv esc bug in udev eh, fun Apr 22 15:55:36 h, that's why Gentoo updates 2 days ago... Apr 22 15:59:10 kergoth: would you pls check where do your cross stuff go after rebuild from pstage? I have .../cross/ instead of .../cross/armv5te (like from scratch) Apr 22 15:59:28 so no crosscompiler is found... Apr 22 15:59:35 but image is rebuilt ;-) Apr 22 16:08:32 going now, bbl Apr 22 16:59:19 re Apr 22 17:06:58 Bah Apr 22 17:07:16 More stuff failed in my big build loop on stable than I think shoul have Apr 22 17:07:21 time to repeat on dev Apr 22 17:07:45 Tartarus: my buildbot has more luck Apr 22 17:08:23 03Marcin Juszkiewicz  07org.openembedded.dev * r9219f40b58 10openembedded.git/recipes/linux/linux.inc: (log message trimmed) Apr 22 17:08:23 linux.inc: unset CONFIG_LOCALVERSION_AUTO to not get 2.6.27.2-svn8826-dirty26 like names Apr 22 17:08:23 Year ago we removed CONFIG_LOCALVERSION from .config to get more sane names of Apr 22 17:08:23 kernels. Today with 2.6.27.2 I got 2.6.27.2-svn8826-dirty26 kernel which was Apr 22 17:08:23 2.6.27.2-svn8826-dirty9 on device so no modules ;( Apr 22 17:08:27 This patch changes this by unsetting CONFIG_LOCALVERSION_AUTO variable. Apr 22 17:08:29 Signed-off-by: Marcin Juszkiewicz Apr 22 17:09:17 hrw, I doubt it :) Apr 22 17:09:38 Tartarus: http://home.haerwu.biz:8010/ Apr 22 17:10:13 You aren't building any meta-toolchain targets :) Apr 22 17:10:42 otavio: will you Ack linux.inc change for stable/2009 branch? Apr 22 17:11:01 Tartarus: right. will add them later Apr 22 17:11:10 bb in ~1h Apr 22 17:15:46 hrw|gone, when you get back, can I have your buildbot config stuff? :) Apr 22 17:25:29 03Michael 'Mickey' Lauer  07fso/milestone5.5 * rde24993430 10openembedded.git/ (2 files in 2 dirs): vala-dbus-binding-tool-native: new recipe; a dbus interface generator Apr 22 17:25:30 03Angus Ainslie  07fso/milestone5.5 * rd18c949df0 10openembedded.git/recipes/dillo/dillo_2.0.bb: Apr 22 17:25:30 dillo: add ssl support Apr 22 17:25:30 fix RDEPENDS Apr 22 17:25:31 03Angus Ainslie  07fso/milestone5.5 * r75b6c47d3c 10openembedded.git/recipes/fltk/fltk2_svn.bb: fltk2 : ship all of the libs Apr 22 17:55:21 hrw|gone: I can ack it Apr 22 18:02:06 otavio: I ask because I would like to just push it without sending to oeml. I already have Koen's ack Apr 22 18:02:11 re Apr 22 18:03:21 hrw: cool :-) Apr 22 18:04:05 so I have your ack Apr 22 18:10:34 03Marcin Juszkiewicz  07stable/2009 * r0265185802 10openembedded.git/recipes/linux/linux.inc: (log message trimmed) Apr 22 18:10:34 linux.inc: unset CONFIG_LOCALVERSION_AUTO to not get 2.6.27.2-svn8826-dirty26 like names Apr 22 18:10:34 Year ago we removed CONFIG_LOCALVERSION from .config to get more sane names of Apr 22 18:10:34 kernels. Today with 2.6.27.2 I got 2.6.27.2-svn8826-dirty26 kernel which was Apr 22 18:10:34 2.6.27.2-svn8826-dirty9 on device so no modules ;( Apr 22 18:10:38 This patch changes this by unsetting CONFIG_LOCALVERSION_AUTO variable. Apr 22 18:10:40 Signed-off-by: Marcin Juszkiewicz Apr 22 18:10:54 now I can continue with BUG stuff Apr 22 18:15:53 can't build meta-toolchain since recent sync with openebedded.org.dev last week: ERROR: QA Issue: non -dev package contains symlink .so: gcc path '/work/i686-commonos-linux/gcc-4.2.4-r3/install/gcc/usr/lib/gcc/i686-commonos-linux/4.2.4/libgcc_s.so' Apr 22 18:15:59 that something expected? Apr 22 18:19:57 hrw, buildbot config please, thanks :) Apr 22 18:19:59 off for a bit Apr 22 18:20:04 ok Tartarus Apr 22 18:20:15 fuck off, ncurses. when i say --disable-relink, that doesn't mean go ahead and relink when installing anyway. Apr 22 18:20:27 heh Apr 22 18:25:40 03Marcin Juszkiewicz  07org.openembedded.dev * r6cf8a2695b 10openembedded.git/contrib/buildbot/ (README master.cfg scripts/full-oe-autobuild): buildbot: config of my buildbot running on http://home.haerwu.biz:8010/ Apr 22 18:25:50 Tartarus: here you hae Apr 22 18:27:45 wow, and it uses -rpath even when i pass --disable-rpath Apr 22 18:27:51 * kergoth growls Apr 22 18:49:36 03Michael 'Mickey' Lauer  07fso/milestone5.5 * rd02b2d551e 10openembedded.git/recipes/ (2 files in 2 dirs): Apr 22 18:49:36 python-dateutil: split out zoneinfo subpackage Apr 22 18:49:36 also add to task-python-everything Apr 22 19:01:02 03Roman I Khimov  07org.openembedded.dev * r57cd858103 10openembedded.git/ (8 files in 3 dirs): Apr 22 19:01:02 dialog: update version 1.0-20050306 -> 1.1-20080819 Apr 22 19:01:02 Previous version is obsolete and can't be found on official FTP, it is also Apr 22 19:01:02 removed from Debian archives so the packages failed to build if not using Apr 22 19:01:02 some distro-specific mirror. Apr 22 19:01:03 Hence, update version removing the old one along the way. Apr 22 19:01:08 Acked-by: Koen Kooi Apr 22 19:04:36 hi Roman_Khimov Apr 22 19:04:47 Hi there Apr 22 19:10:46 have a nice rest of day Apr 22 19:12:37 03Petri Lehtinen  07stable/2009 * r60dea07419 10openembedded.git/classes/base.bbclass: Apr 22 19:12:37 base.bbclass: Combine all checksum.ini files found in BBPATH Apr 22 19:12:37 This makes it possible for overlays to have checksum.ini. Apr 22 19:12:37 Signed-off-by: Denys Dmytriyenko Apr 22 19:12:37 Acked-by: Koen Kooi Apr 22 19:12:38 Acked-by: Tom Rini Apr 22 19:13:19 hi all! Apr 22 19:14:03 for err.. some architecture I need several toolchains installed at the same time, e.g bfin-elf, bfin-uclinux and bfin-linux-uclibc with bfin-linux-uclibc being main for applications, bfin-elf for some firmwares and bfin-uclinux for kernel. While I see it technically possible, what approach is the best for it? I mainly address STAGING_DIR_TARGET issue and some minor ones... Apr 22 19:21:37 Er Apr 22 19:21:42 Do you really? Apr 22 19:21:47 ie Apr 22 19:21:52 does each toolchain have different patches? Apr 22 19:22:05 Normally you should be able to use one for any and firmware just needs to pass the right options, etc Apr 22 19:22:23 Also, are you fixing up the bfin support? It wasn't quite there a few months ago :) Apr 22 19:33:36 03Roman I Khimov  07org.openembedded.dev * r934ef803ea 10openembedded.git/ (6 files in 2 dirs): clamav: add 0.94.2 version recipe Apr 22 19:49:59 Tartarus, I've got main toolchain working, but whole stuff is very hacky at the moment, I think I'll submit it into some branch tonight Apr 22 19:50:26 Tartarus, all toolchains produce different kinds of binaries with different ABIs. Apr 22 19:52:25 slapin_nb, right, that's not the question Apr 22 19:52:34 The question is do you need different compilers for that? Apr 22 19:53:07 ie you can build u-boot and the kernel and userland for most targets with one compiler Apr 22 19:53:30 Heck, you can use a "normal" compiler to build Android, in some respects :) Apr 22 19:54:08 Tartarus, doe to analog's modifications to targets, it won't understand some flags depending on target it was built for Apr 22 19:54:40 Tartarus, I know it is brain damage, but I have too little time/knowledge to fix that Apr 22 19:54:44 03Roman I Khimov  07org.openembedded.dev * rd070641ef2 10openembedded.git/ (5 files in 3 dirs): Apr 22 19:54:44 siproxd: new recipe Apr 22 19:54:44 New recipe for siproxd 0.7.1 Apr 22 19:54:44 Acked-by: Koen Kooi Apr 22 19:55:11 Ah Apr 22 19:55:20 Well, I guess take a look at the gcc-cross-kernel stuff Apr 22 19:55:26 and adapt for gcc-cross-firmware too? Apr 22 19:55:42 Tartarus, thanks for hint! Apr 22 19:55:48 np :) Apr 22 19:58:37 03Frans Meulenbroeks  07org.openembedded.dev * r8fc533bcb9 10openembedded.git/ (conf/checksums.ini recipes/flickcurl/flickcurl_1.9.bb): Apr 22 19:58:37 flickcurl: created recipe Apr 22 19:58:37 flickcurl is a C interface to retrieve flickr data Apr 22 19:58:38 03Frans Meulenbroeks  07org.openembedded.dev * ref0a43982a 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Apr 22 20:31:47 yeah gcc-4.4 is released Apr 22 20:33:39 yay Apr 22 20:34:13 eabi for armv4 Apr 22 20:34:14 haha Apr 22 20:49:04 hrw|gone ? Apr 22 21:04:47 anyone run into libtool using the -rpath and searching that path for the libs, choking on libc.a in /usr/lib? i *know* i saw that at one point, and resolved it, but now I'm hitting it and can't remember Apr 22 21:06:35 hm sorry Apr 22 21:10:19 hi ph5 Apr 22 21:10:52 jo ant Apr 22 21:11:03 hi woglinde Apr 22 21:12:02 ant_home: morning :) Apr 22 21:12:10 hi woglinde :) Apr 22 21:12:14 hello Jay7 Apr 22 21:15:01 hi jay7 Apr 22 21:16:12 anyone have recent experience with building libgles-omap3? Apr 22 21:16:31 Anyone other than hrw|gone have a buildbot setup for OE? Apr 22 21:16:43 * Jay7 thinking about building new 'firmware' for his akita Apr 22 21:16:59 hoj I will test it next week Apr 22 21:17:05 for beagleboard Apr 22 21:17:11 to get it run with qt Apr 22 21:17:18 ant_home: whant image did you using now on your Z? Apr 22 21:17:40 actually I'm testing 3 images... Apr 22 21:17:53 console-image will be fixed in a few mins :-) Apr 22 21:18:01 is x11-image usable? :) Apr 22 21:18:05 opie-image and x11-image the others Apr 22 21:18:15 well, boots Apr 22 21:18:22 or may be switch to stable/2009? Apr 22 21:18:30 did not check keyboard yet Apr 22 21:18:32 no Apr 22 21:18:38 nonsense for Z Apr 22 21:18:49 we are waiting for 2.6.29 Apr 22 21:18:57 ah, well.. Apr 22 21:19:14 some interesting post in linux-arm Apr 22 21:19:22 woglinde: I'm hitting a missing kernel header mach/display.h Error is just like this http://tinderbox.openembedded.net/public/logs/4227413.txt Apr 22 21:19:24 Omegamoon is reading all ML ;-) Apr 22 21:20:09 Oops, I mean http://tinderbox.openembedded.net/public/logs/4227413.txt Apr 22 21:24:00 hoj which kernelversion you are using? Apr 22 21:25:21 "bitbake virtual/kernel" builds 2.6.27 Apr 22 21:26:05 for machine omap3evm Apr 22 21:27:04 I see the patch: http://marc.info/?l=linux-fbdev-devel&m=122581507722475&w=2, but the kernel source doesn't even have a arch/arm/plat-omap/include/mach dir Apr 22 21:28:01 so I assume that patch is not part of the default omap3evm kernel Apr 22 21:28:27 hm I think 2.6.27 is reallay old Apr 22 21:28:31 in this case Apr 22 21:28:40 hoj do you have a beagleboard? Apr 22 21:29:02 no, just an omap35xEVM Apr 22 21:29:28 hoj hm try the kernel for beagleboard Apr 22 21:30:13 new to oe, so which file would I touch to change the kernel version (in FAQ?) Apr 22 21:31:52 local.conf Apr 22 21:32:48 Will try, thanks! Apr 22 21:32:53 hoj are you using stable? Apr 22 21:34:08 woglinde, you mean for DISTRO? I set to "angstrom-2008.1" Apr 22 21:34:31 nope Apr 22 21:34:38 which branch did you checkout Apr 22 21:34:50 and I think this kernel shoudl work Apr 22 21:34:51 linux-omap_2.6.29 Apr 22 21:36:08 I got oe source with "git clone git://git.openembedded.net/openembedded" as suggested on http://wiki.openembedded.net/index.php/Getting_started Apr 22 21:36:45 hm ah Apr 22 21:36:46 sorry Apr 22 21:38:12 you could try the following too Apr 22 21:38:18 Haven't tried specifying a certain kernel yet Apr 22 21:38:32 edit recipes/linux/linux-omap_2.6.29.bb Apr 22 21:38:35 and add Apr 22 21:38:57 DEFAULT_PREFERENCE_omap3evm = "1" Apr 22 21:39:36 but then you have to commit it in your local branch Apr 22 21:39:49 otherwise you cannt run git pull --rebase Apr 22 21:39:56 for updates Apr 22 21:40:46 ok Apr 22 21:40:48 trying to figure out what recipe populates /etc/ld.so.conf and why it doesn't simply contain an 'include /etc/ld.so.conf.d/*.conf' on my distro? Apr 22 21:41:47 in local.conf PREFFERED_VERSION_linux-omap = "2.6.29" may work Apr 22 21:42:16 hoj then you dont have to check in someting Apr 22 21:43:20 tharvey hm basefiles maybee Apr 22 21:43:25 otherwise libc itself Apr 22 21:43:48 woglinde: so, maybe the PREFFERED_VERSION _or_ DEFAULT_PREFERENCE? then 'bitbake virtual/kernel'? Apr 22 21:43:52 hmm... actually looks like something builds ld.so.conf on the fly? I don't see it in my rootfs Apr 22 21:44:01 hoj right Apr 22 21:44:22 I'll try the local.conf way. Thanks! Apr 22 21:45:24 hoj yes but I am not sure it will work Apr 22 21:45:45 * Jay7 -> sleep Apr 22 21:45:49 woglinde Got it; trying now... Apr 22 22:00:52 mickey|bbl: ping Apr 22 22:00:59 mickey|bbl: http://fr.pastebin.ca/1400229 Apr 22 22:01:14 chown: unknown user messagebus Apr 22 22:02:05 Stopping system message bus: start-stop-daemon: unknown user messagebus Apr 22 22:05:22 woglinde: after the PREFFERED_VERSION_linux-omap addition to local.conf, bitbake didn't seem to compile anything (just parsed files?), so I'll try the other way Apr 22 22:05:22 mickey|bbl: that was first boot Apr 22 22:05:39 mickey|bbl: on second things are better:-) http://fr.pastebin.ca/1400233 Apr 22 22:06:01 hoj hm try bitbake -c clean linux Apr 22 22:06:06 bitbake linux Apr 22 22:07:01 not virtual/kernel target? Apr 22 22:07:24 hm or linux-omap Apr 22 22:07:26 sorry Apr 22 22:08:19 ok Apr 22 22:11:52 woglinde: bitbake is getting 2.6.29 source; good so far... Apr 22 22:11:56 cool Apr 22 22:23:42 good nite Apr 22 22:24:11 bye Apr 22 22:37:48 03Andrea Adami  07org.openembedded.dev * r1a36b908f1 10openembedded.git/recipes/images/ (console-base-image.bb console-image.bb gnuradio-image.bb): console-base-image,console-image,gnuradio-image: don't hardcode psplash-zap Apr 22 22:44:37 hello Apr 22 22:45:48 I would like help with $ bitbake task-sdk-host Apr 22 22:45:51 anybody here? Apr 22 22:46:43 probably can't help (I'm new :) ), but what's your specific question? Apr 22 22:47:20 To be specific I would like to paste the results of the error log, ok? Apr 22 22:48:18 if it's long, you might try pastebin Apr 22 22:52:01 thanks hoj, that site seems slow/busy, but I hope to have something posted soon, it is still "Waiting for pastebin.com.." Apr 22 23:11:21 http://pastebin.com/d321543f9 Apr 22 23:12:04 this is the bottom fragment of the log file. Notice that several files have compiled OK before the failure. Apr 22 23:12:55 filename: build/tmp/work/i686-armv4t-sdk-linux-gnueabi/gcc-cross-sdk-4.3.3-r2/temp/log.do_compile.22221 Apr 22 23:16:01 Looks like glibc headers are missing, but I don't have any theories. I installed a toolchain separeately, so I skipped the cross tool build.... Apr 22 23:17:04 where is "/diskb" ? remote? Apr 22 23:21:46 no it is a local drive right off the root Apr 22 23:29:26 dickelbeck: not sure, but you might look for stdio.h in the paths listed in the last compile line (to make sure libc was built) Apr 22 23:38:05 looking at the gcc-cross-sdk failure too Apr 22 23:38:09 was that stable or dev? Apr 22 23:43:23 hot: that was it, /build/tmp/staging/armv4t-linux-gnueabi/usr/include is empty Apr 22 23:43:46 how can that be empty Apr 22 23:45:54 dickelbeck, you're using an external toolchain? Apr 22 23:46:04 If so, how did you plug that into OE? That would probably explain it Apr 22 23:46:37 I've done SDK + external toolchain before, but step 0 was "shove external toolchain into ${TMPDIR}" so that everything would work Apr 22 23:47:11 But... you're building gcc for an SDK, which isn't sensical unless you're trying to use an external libc, but OE-built compiler Apr 22 23:53:20 Tartarus: thanks! Apr 22 23:54:09 No external toolchain, this is from $bitbake gcc-cross-sdk Apr 22 23:54:55 http://pastebin.com/d321543f9 Apr 22 23:55:06 hmm, maybe i got confused Apr 22 23:55:20 this is dev Apr 22 23:55:41 dickelbeck, yeah, ok confused what you said with what hoj said Apr 22 23:55:47 dickelbeck, so, lets reset the converstation :) Apr 22 23:55:51 What MACHINE/DISTRO are you using? Apr 22 23:56:11 And I assume you tried to build task-sdk-host because meta-toolchain (or m-t-something) failed Apr 22 23:56:35 machine=ks8695px in an overlay file distro is my own also, in the same overlay Apr 22 23:56:52 distro based on angstrom, minimal or totally custom? Apr 22 23:57:22 But, either way Apr 22 23:57:45 totally custom Apr 22 23:57:51 Try building meta-toolchain. It _shouldn't_ be the case, but possibly, somehow is, that you don't have the target libc built yet. Apr 22 23:58:06 OK Apr 22 23:58:08 ok, give me a minute Apr 22 23:58:13 np :) Apr 22 23:58:20 I've got a buncha debug ideas, but they'll take a bit to run Apr 22 23:59:00 Also, if you can, try doing DISTRO=minimal and building that. It's possible you don't have some virtuals setup right. Does bitbake -g task-sdk-host, for your distro, give any warnings? Apr 22 23:59:33 I've got a boat-load of SDK builds going on right now on .dev, but I just build meta-toolchain*, not task-sdk-host directly Apr 22 23:59:44 That could only matter in that m-t has target stuff explicitly in it Apr 23 00:00:23 But even still, bitbake gcc-cross-sdk -g should show glibc (or eglibc or uclibc) in the list of stuff it needs first Apr 23 00:01:18 dick@dick-desktop:/diskb/oestuff/build$ bitbake -g task-sdk-host Apr 23 00:01:18 NOTE: Handling BitBake files: \ (6717/6717) [100 %] Apr 23 00:01:18 NOTE: Parsing finished. 6425 cached, 0 parsed, 292 skipped, 0 masked. Apr 23 00:01:18 NOTE: Cache is clean, not saving. Apr 23 00:01:18 NOTE: Resolving any missing task queue dependencies Apr 23 00:01:18 NOTE: Preparing runqueue Apr 23 00:01:20 NOTE: Dependencies saved to 'depends.dot' Apr 23 00:01:22 NOTE: Task dependencies saved to 'task-depends.dot' Apr 23 00:01:24 dick@dick-desktop:/diskb/oestuff/build$ Apr 23 00:01:40 clean as a whistle, now trying meta-toolchain Apr 23 00:02:28 * denix|away is away: I'm not here Apr 23 00:02:40 k Apr 23 00:03:15 fwiw, and I say this as someone who maintains a custom distro too, it's best when something fails to see if minimal or angstrom also fails Apr 23 00:03:32 If so, it's a generic bug, if not, start seeing what they do differently w/ bitbake -g Apr 23 00:03:43 The .dot files can be turned into pretty graphs, or just grep'd about Apr 23 00:04:53 great advice. Lots to learn. I spent a month in oe 6 months ago, and forgot everything I learned. Updated to dev, and much to learn again. Apr 23 00:05:45 comes right back around to the same error: Apr 23 00:05:49 NOTE: Task failed: /diskb/oestuff/build/tmp/work/i686-armv4t-sdk-linux-gnueabi/gcc-cross-sdk-4.3.3-r2/temp/log.do_compile.26924 Apr 23 00:05:50 NOTE: package gcc-cross-sdk-4.3.3-r2: task do_compile: failed Apr 23 00:05:50 ERROR: TaskFailed event exception, aborting Apr 23 00:05:50 ERROR: Build of /diskb/oestuff/openembedded/recipes/gcc/gcc-cross-sdk_4.3.3.bb do_compile failed Apr 23 00:05:50 ERROR: Task 248 (/diskb/oestuff/openembedded/recipes/gcc/gcc-cross-sdk_4.3.3.bb, do_compile) failed Apr 23 00:05:51 NOTE: Tasks Summary: Attempted 397 tasks of which 387 didn't need to be rerun and 1 failed. Apr 23 00:05:53 ERROR: '/diskb/oestuff/openembedded/recipes/gcc/gcc-cross-sdk_4.3.3.bb' failed Apr 23 00:06:53 this was bitbake meta-toolchain Apr 23 00:06:57 great name Apr 23 00:07:06 i think you need to change it Apr 23 00:07:31 ok Apr 23 00:07:44 Now, time to go a different route Apr 23 00:07:46 but then again one would think wierd things if they focused on anyones nicks Apr 23 00:07:47 but first Apr 23 00:07:53 if you have a multi core or otherwise fast box Apr 23 00:08:15 PARALLEL_MAKE = "-j 6" Apr 23 00:08:15 BB_NUMBER_THREADS = "4" Apr 23 00:08:19 in your local.conf Apr 23 00:08:26 (That's for my quad-core box, adjust to taste) Apr 23 00:08:35 Then try building for angstrom or minimal Apr 23 00:08:39 but still your machine Apr 23 00:08:53 i need to change branches Apr 23 00:08:59 oh, and do just bitbake gcc-cross-sdk Apr 23 00:09:00 or somethin cause nothin builds Apr 23 00:09:06 that'll skip out a few other things that are building for you Apr 23 00:09:34 then again experimental branches can be that way Apr 23 00:12:23 single core, 1 gig RAM, 2.6 gHz, need a new computer soon I guess, but fast SCSI disks, started now. Apr 23 00:12:58 hmm Apr 23 00:13:04 maybe just BB_NUMBER_THREADS=2 then Apr 23 00:13:35 unrelated, have you speced out a new machine? I was surprised how cheap stuff is Apr 23 00:13:49 built a monster of a box for ~$1300USD shipped Apr 23 00:14:36 The biggest "slow" part of OE is that you build from scratch gcc a few times, and your libc too Apr 23 00:14:56 But once you're doing a make -j N on that with a fast box, you don't notice anymore Apr 23 00:15:02 It also means less coffee making time :( Apr 23 00:16:49 I am all ears. Everything I know about a desktop PC is obsolete, and I've been at a computer 40 hours per week for 26 years writing software. Apr 23 00:17:17 shit my disk just filled up. I will have to scoot a partition boundary over a little, that will set us back 10 minutes. Apr 23 00:17:28 owch Apr 23 00:17:40 I don't keep up anymore like I used to, either. Apr 23 00:17:46 But it's amazing how cheap a 1.5TB disk is Apr 23 00:18:19 I got 4 of the things, raid0+1 together for some sort of protection against failure (yes, I know, need real backups too) Apr 23 00:19:33 http://www.newegg.com/Product/Product.aspx?Item=N82E16822148337 Apr 23 00:20:13 And I'm sure everyone here has paid 2x or more for some tiny fraction of the size in disk space Apr 23 00:25:07 yes that price is insane value. Apr 23 00:25:45 seagate is about all I trust, from the old days. Apr 23 00:26:37 where is a good place to get a quad core mini tower or desktop pc, somewhat partial to AMD, but its a minor preference. Apr 23 00:27:07 built mine from newegg.com :) Apr 23 00:27:21 just looked at the invoice to find that link, the 1300 included a new UPS too Apr 23 00:27:24 1150 otherwise Apr 23 00:27:47 happily stock CPU cooler are fine again Apr 23 00:28:11 say again? Apr 23 00:28:28 retail/OEM amd cpus come with a good heatsink/fan combo Apr 23 00:28:38 i recall that not being the case before Apr 23 00:29:00 I get it, CPUs. Apr 23 00:41:43 the partition expansion is probably nearing completion, been running for awhile. Apr 23 00:42:53 what is the name of your distro? and have you made it public? Apr 23 00:43:16 nope, not public, but I work for Embedded Alley Apr 23 00:43:30 you may, or may not, have heard of us, which is ok :) Apr 23 00:44:46 http://dl.softplc.com/ds_Smartboard.pdf Apr 23 00:45:13 if you ever need an ARM board with some decent comm ports.... Apr 23 00:45:30 this is my target. Apr 23 00:45:39 and my design Apr 23 00:45:53 neat Apr 23 00:46:57 Is there a reason you ruled out using angstrom or similar? Apr 23 00:48:00 Well we have our own initialization scripts, coming from prior work. Apr 23 00:48:53 Frankly, it is difficult to characterize these distros. We need a ram disk at /tmp and invested a good bit of time in our init scripts and runlevel support. Apr 23 00:49:06 We also run a firewall normally, and are headless most always. Apr 23 00:49:34 Ah Apr 23 00:49:53 angstrom can do headless, but if you're providing your own set of initscripts, totally, that makes a bit more sense Apr 23 00:50:18 fwiw, i'd recommend basing your distro conf files off of angstrom, or at least looking good and hard at their preferred versions/provider files Apr 23 00:50:48 This is us having fun: Apr 23 00:50:50 Things are better than they used to be, but still not 100%, about defaults being sane if you aren't specifying versions and providers for packages Apr 23 00:50:51 http://softplc.com/prusace.php Apr 23 00:51:44 cool Apr 23 00:52:39 And, ah yes Apr 23 00:52:57 since 6 months ago, there is now a stable branch that is going to be usable long term, really, we mean it this time :) Apr 23 00:53:02 http://wiki.openembedded.net/index.php/Stable Apr 23 00:53:27 And i've got a number of SDK updates i'm testing that'll be in there shortly Apr 23 00:54:31 Do you think stable is better for us to be looking at? I build my own kernel, and generally tend to want to push the ball forward as far as possible each kick. Apr 23 00:55:01 This pertains to gcc version and glibc versions. I only want to have to do this about once every year or two. Apr 23 00:55:21 Well, I think the wiki page has the policy goals and such Apr 23 00:56:07 But the general idea is that it'll be easier for people to make commerical products off us as the 2009 branch won't change wildly in 2009, or ever Apr 23 00:56:50 ie if you pull .dev tomorrow, a bunch of things could change, but not on stable Apr 23 00:57:15 happily right now at least, the bugfixing mode that's going on for stable means fixing bugs in .dev first then pulling into stable Apr 23 00:58:06 I see our use case being: get everything to build, then stick with that. This could conform to dev, as long as a person does not git pull after things are working, right? Apr 23 00:58:14 Yeah Apr 23 00:58:43 It's just that for the time peroid inbetween when you start working and when you're ready to snap things, stable won't change as much as dev Apr 23 00:58:52 and more importantly, no big changes Apr 23 00:59:22 right now at least, dev and stable aren't really different in terms of what does and doesn't build / work Apr 23 00:59:38 But for stable, there's things that must always be working, and that's not strictly the case for dev Apr 23 01:00:35 There's safety in numbers, what are most "users" using I wonder? Apr 23 01:00:40 The other side of things is that possibly at least, you could do more frequent updates to your userland side of things as the only stuff that should change for you, assuming you use preferred versions/preferred providers for everything, is just bugfixes Apr 23 01:00:43 OK gparted just finished. Apr 23 01:01:25 There's definately a push to make stable work and work well for the defined list of things that need to work Apr 23 01:02:52 And I do think the "ooh, there's a stable now!" has pulled people into using that over .dev Apr 23 01:03:13 But really, today, there's not much difference between the two as stable is still fairly new Apr 23 01:03:21 There hasn't yet been a big change in .dev Apr 23 01:03:29 Only because I need more time in the day :) Apr 23 01:03:53 Are you one of the main developers? Is that what Embedded Alley does? Apr 23 01:04:44 (bitbake gcc-cross-sdk) churning Apr 23 01:05:08 DISTRO=minimal Apr 23 01:08:36 I'm not part of the core team, no, but I'm the most active sdk guy Apr 23 01:09:08 and we do a lotta stuff :) Apr 23 01:09:18 http://www.embeddedalley.com/ Apr 23 01:09:39 is meta-sdk what I need to build for a cross toolchain? I need basic console type libraries such as ncurses and probably the most advance library is libxml2 Apr 23 01:09:46 meta-toolchain Apr 23 01:10:01 oops, meant that. Apr 23 01:10:03 that'll give you a tarball you can install and that'll have some stuff in it Apr 23 01:10:22 then see meta-toolchain-fso for example for how to change the list of stuff that gets installed Apr 23 01:10:45 you'll probably want to make your own meta-toolchain-something that on the host uses the defaults, but on the target does task-sdk-bare + whatever stuff you want to support Apr 23 01:11:11 task-sdk-base is a bit large for my tastes, but -bare is really the bare minimum for c/c++ dev and no more Apr 23 01:11:20 Tartarus, this is awesomely helpful. We don't build anything on the target at this time. Apr 23 01:11:33 need libxml2 Apr 23 01:12:36 and I understand the I can use meta-toolchain-fso as an example. Apr 23 01:13:19 Last time I used openembedded, I had a few derivations like this, and put them into an overlay. Apr 23 01:14:00 But what I found was that some of the includes or referenced files in the custom recipe also had to be in the same overlay, Apr 23 01:14:18 they could not be found in the original tree. That was a bit unexpected. Apr 23 01:14:30 Yeah Apr 23 01:14:38 That's still true and somewhat annoying Apr 23 01:14:53 What you might want to consider is not overlaying but instead doing an internal git tree Apr 23 01:15:05 Is that policy, or simply lack of python support? Apr 23 01:15:20 I think it's a limitation of how this all works Apr 23 01:15:40 kergoth ? Apr 23 01:15:45 Or anyone else that knows bitbake? :) Apr 23 01:16:08 internal git tree gives me the flexibility to simply hack on the working directory to my heart's content? Apr 23 01:16:21 How well do you know git? Apr 23 01:16:43 very little, I am comfortable with svn Apr 23 01:16:59 Then yeah, you'd wanna get good at git before doing this kind of thing Apr 23 01:17:01 But yes Apr 23 01:17:09 need to run through a tutorial and start using it. Apr 23 01:17:29 Being another SCM, yes, these kind of things are possible Apr 23 01:18:54 the build it still going....without errors. Apr 23 01:19:27 in my distro file, named gatecraft.conf I have these lines: Apr 23 01:19:31 GATECRAFT_GCC_VERSION = "4.3.3" Apr 23 01:19:32 PREFERRED_VERSION_gcc = "${GATECRAFT_GCC_VERSION}" Apr 23 01:19:32 PREFERRED_VERSION_gcc-cross = "${GATECRAFT_GCC_VERSION}" Apr 23 01:19:32 PREFERRED_VERSION_gcc-cross-initial = "${GATECRAFT_GCC_VERSION}" Apr 23 01:19:32 PREFERRED_VERSION_gcc-cross-sdk = "${GATECRAFT_GCC_VERSION}" Apr 23 01:19:51 based on what angstrom has, yeah :) Apr 23 01:20:02 PREFERRED_VERSION_glibc = "2.9" Apr 23 01:20:02 PREFERRED_VERSION_glibc-intermediate = "2.9" Apr 23 01:20:16 fwiw, I'd use eglibc not glibc Apr 23 01:20:29 glibc not-a-fork aimed at embedded support Apr 23 01:20:38 The versions are newer, and the assignment is not ?= but = Apr 23 01:21:36 Ok, I will look into that. Thanks. That is a change to a provider somewhere? Apr 23 01:21:53 yeah Apr 23 01:22:01 see angstrom's ANGSTROMLIBC hook Apr 23 01:22:19 eglibc is in conf/distro/include/angstrom-eglibc.inc Apr 23 01:23:22 I hope you don't mind that I am using this compile time to maximize the brain dump. Apr 23 01:23:59 When this is done, the task will be to figure out what is different between DISTRO=minimal and DISTRO=gatecraft no? Apr 23 01:26:02 correct Apr 23 01:27:40 distro minimal seems to be not so minimal .... Apr 23 01:28:36 03Tom Rini  07org.openembedded.dev * rc681b0610f 10openembedded.git/recipes/busybox/ (8 files in 2 dirs): Apr 23 01:28:36 busybox 1.13.2: Add more upstream patches, bump PR Apr 23 01:28:36 This adds the awk, killall, printf, syslogd, top24, unzip and wget patches Apr 23 01:28:36 from http://busybox.net/downloads/fixes-1.13.2/ Apr 23 01:28:36 03Tom Rini  07org.openembedded.dev * r3c741fd199 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Apr 23 01:28:41 03Tom Rini  07org.openembedded.dev * r49a720ff85 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Apr 23 01:28:44 03Tom Rini  07org.openembedded.dev * rc8b3cc5ef8 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Apr 23 01:28:47 03Tom Rini  07org.openembedded.dev * r1e6c98c990 10openembedded.git/recipes/binutils/binutils-canadian-sdk_2.18.50.0.7.bb: binutils-canadian-sdk: Add 2.18.50.7 (as this is a popular default) Apr 23 01:28:52 03Tom Rini  07org.openembedded.dev * r98dd5d2a73 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Apr 23 01:28:55 03Tom Rini  07org.openembedded.dev * r3889422c46 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Apr 23 01:28:58 bah, crap Apr 23 01:29:07 Didn't think that particular checkout was so far out :( Apr 23 01:51:04 packaging the glibc now into ipks Apr 23 01:54:22 Tartarus: is there something like meta-toolchain for windows mingw? Apr 23 01:54:42 canadian-sdk Apr 23 01:54:49 but no example expansions yet Apr 23 01:55:31 need to google up a few extra conf variables for that Apr 23 01:56:24 that is exciting. Do you think canadian-sdk should build without error on linux? Apr 23 01:56:43 man if this could be used to generate a toolkit for window for my target that would be awesome. Apr 23 01:57:00 oops windows Apr 23 01:57:20 yeah Apr 23 01:57:31 canadian-sdk is for makign a mingw toolchain on linux Apr 23 01:57:43 people have made building on cygwin work but it's not up to date nor merged Apr 23 01:57:52 canadian-sdk has a caveat tho Apr 23 01:57:56 mingw + gcc 4.3.x don't mix Apr 23 01:58:04 so you'd have to drop down to 4.2.x Apr 23 01:58:17 or you can debug the problem, i'd apprecate it :) Apr 23 01:58:42 cygwin is dead to me. Understand about 4.2 vs. 4.3. Apr 23 01:59:56 when does the problem occur with 4.3? And are there other build tools that are doing it successfully that you know. I have some scripts around here that would build mingw gcc 4.x, that I wrote 3 years ago. Apr 23 02:00:16 One of the mingw support things blows up Apr 23 02:00:21 Didn't have time to debug at the time Apr 23 02:00:23 still haven't Apr 23 02:00:28 Don't know if they would be a good start or not. Apr 23 02:00:33 I really wish codesourcery would publish their mingw env Apr 23 02:00:40 since they do have mingw-based 4.3.x compilers Apr 23 02:00:55 but maybe they're just using a mingw env compiled w/ 4.2.x or oler Apr 23 02:01:02 *older Apr 23 02:02:55 Yes, as I recall, the initial compiler can be older. In my scripts I started with 3.4 or some such, the stock ubuntu mingw version. And then the scripts simply called gcc's configure with the correct options. Apr 23 02:03:05 yeah Apr 23 02:03:11 we use a specific 3.4 for some stuff Apr 23 02:03:27 But when it's valid to be using a more normal compiler is when it falls down Apr 23 02:03:37 like it might be gcc 4.3.x mingw host stuff at fault Apr 23 02:03:41 I really don't recall where it blew Apr 23 02:03:51 But going down to 4.2.4 for gcc-* in pref versions works Apr 23 02:05:10 I was at 4.2.4 prefs before I started today. So I may have to revert for mere pain avoidance. The mingw support is not urgent, that can come later. Apr 23 02:06:10 Right now I just need a toolkit on my linux host, not target for which I can compile some new *.so files I am writing for our runtime. Apr 23 02:06:53 NOTE: package gcc-cross-sdk-4.2.4-r5: task do_compile: started Apr 23 02:07:53 and I need libxml2, that was the need today. The old toolkit I made was with OE from 6-8 months ago, and it did not have libxml2 in it. Apr 23 02:08:21 And I have forgotten everything I know about OE from back then, and have had to relearn it. This time I guess I will document with scripts. Apr 23 02:13:42 bitbake is done now, and it failed. Apr 23 02:13:57 not from what i understand Apr 23 02:14:19 I'll be darned if it did not fail in the same place. Apr 23 02:14:41 for your machine and DISTRO=minimal? interesting.. Apr 23 02:14:41 | The directory that should contain system headers does not exist: Apr 23 02:14:41 | /diskb/oestuff/build/tmp/staging/armv4t-oe-linux-gnueabi/usr/include Apr 23 02:14:41 | make[2]: *** [stmp-fixinc] Error 1 Apr 23 02:14:41 | rm gfortran.pod Apr 23 02:14:41 | make[2]: Leaving directory `/diskb/oestuff/build/tmp/work/i686-armv4t-sdk-oe-linux-gnueabi/gcc-cross-sdk-4.2.4-r5/gcc-4.2.4/build.i686-linux.arm-oe-linux-gnueabi/gcc' Apr 23 02:14:43 | make[1]: *** [all-gcc] Error 2 Apr 23 02:14:45 | make[1]: Leaving directory `/diskb/oestuff/build/tmp/work/i686-armv4t-sdk-oe-linux-gnueabi/gcc-cross-sdk-4.2.4-r5/gcc-4.2.4/build.i686-linux.arm-oe-linux-gnueabi' Apr 23 02:14:46 pastebin the machine.conf Apr 23 02:14:48 | make: *** [all] Error 2 Apr 23 02:14:50 | FATAL: oe_runmake failed Apr 23 02:15:04 holy shit dude why you keep pasting madd lines Apr 23 02:15:34 coney: sorry, I am not a veteran here. Apr 23 02:15:51 or anywhere Apr 23 02:16:59 http://pastebin.com/d64506159 Apr 23 02:17:48 where are the logs Apr 23 02:19:01 now i'm puzzled Apr 23 02:19:43 build/tmp/ Apr 23 02:20:23 * coney is rivited at the sheer size of it http://wiki.openembedded.net/index.php/Getting_started Apr 23 02:20:24 http://pastebin.com/m25d60747 is the bottom the log Apr 23 02:21:20 * Tartarus curses, very loudly, the QA_HASH thing being fatal Apr 23 02:21:25 er, GNU_HASH Apr 23 02:22:35 stmp-fixinc wonder what thats about Apr 23 02:25:48 * Tartarus undoes 846330c96c871b0b4116581fa195a6ad1932a439 for now and restarts his big damn loop Apr 23 02:26:37 khem ? Apr 23 02:26:47 excuse you? Apr 23 02:27:05 http://pastebin.com/m7929ede6 is the local.conf Apr 23 02:27:56 nothing in particular pops out, sorry Apr 23 02:28:10 and time to stop before I go crazier :) Apr 23 02:39:05 Tartarus: whats up Apr 23 02:47:51 dickelbeck: do you have libc headers and kernel headers installed ? Apr 23 02:48:22 do you mean as an external install? no. Apr 23 02:49:08 dickelbeck: I dont want to go thru all history above briefly update me what were you bitbakin when it happened Apr 23 02:49:44 bitbake gcc-cross-sdk Apr 23 02:49:59 with distro=minimal as a test Apr 23 02:50:30 and machine is my own ks8695.conf but is is nearly identical to the dev repo. Apr 23 02:51:04 I started with distro=gatecraft which was my own, and Tartarus asked to simply for a test to minimal or angstrom Apr 23 02:51:17 . Apr 23 02:52:22 ok, it seems your build is not pulling libc-headers Apr 23 02:52:43 can you do bitbake -g gcc-cross-sdk Apr 23 02:52:51 Tartarus: it cant automatically find extra files across overlay boundaries because overlays dont necessarily share the same structure. it cant make that assumption. you can make it happen by leveraging the FILESPATH vars though. i.e. FILESPATHBASE .= "/path/to/original/collection/recipes/busybox" or what have you Apr 23 02:53:14 kergoth: howdy, Apr 23 02:53:35 kergoth: what is .= (dot) before = Apr 23 02:54:00 += and =+ use a space as a separator, they're only appropriate for space separated variables Apr 23 02:54:08 for manipulating things like BBPATH, FILESPATH, you dont want that Apr 23 02:54:16 .= is the same as +=, but iwth no separator Apr 23 02:54:19 its a concatenation Apr 23 02:54:23 =. is the same as =+ Apr 23 02:54:28 kergoth: hmmm ok thx Apr 23 02:54:31 np Apr 23 02:54:33 * khem wondered Apr 23 02:54:46 Tartarus: forgot the : seperator in that .=, but you get the point Apr 23 02:54:48 * khem is using it in couple of gcc recipes. Apr 23 02:54:59 khem: pretty sure its in the bitbake manual ;) Apr 23 02:55:26 kergoth: To be true, I never read it :) Apr 23 02:55:30 bitbake -g gcc-cross-sdk runs without error, you want the two dot files? Apr 23 02:55:37 dickelbeck: yes Apr 23 02:55:39 its only really useful for syntax and stuff Apr 23 02:55:43 but it is useful for that Apr 23 02:55:43 http://bitbake.berlios.de/manual/ Apr 23 02:55:48 http://bitbake.berlios.de/manual/ch02.html#id869030 Apr 23 02:56:37 kergoth: I am sure its all in there :) Apr 23 02:56:52 actually, its not Apr 23 02:56:55 ?= is missing :) Apr 23 02:56:57 for some reason Apr 23 02:57:03 heh, Apr 23 02:57:06 its on my huge list of things that need fixing.. Apr 23 02:57:06 http://pastebin.com/m33e3cd7f is the depends.dot Apr 23 02:57:10 bugs everywhere Apr 23 02:58:17 http://pastebin.com/m50d3214f is the task....dot Apr 23 02:58:43 dickelbeck: your dependencies looks correct **** ENDING LOGGING AT Thu Apr 23 02:59:57 2009