**** BEGIN LOGGING AT Thu Dec 31 02:59:57 2009 Dec 31 03:28:32 getting this error Dec 31 03:28:33 malloc.c: In function '__libc_calloc': Dec 31 03:28:33 malloc.c:5947: internal compiler error: Segmentation fault Dec 31 03:28:46 happens when its compiling glibc Dec 31 03:28:59 is this like a one-off thing that might resolve itself if I rerun the bitbake? Dec 31 03:29:08 cause it doesn't seem like a code/typo error Dec 31 03:46:55 Aditya1: which platform is that? Dec 31 03:47:11 Aditya1: in general these errors are sticky.. Dec 31 03:47:16 the host I am compiling on? Ubuntu 9.04 Dec 31 03:47:18 err Dec 31 03:47:18 wait Dec 31 03:47:19 no Dec 31 03:47:20 Debian Dec 31 03:47:32 I recently switched after my Ubuntu VM died Dec 31 03:47:36 "sticky" ? Dec 31 03:47:59 Aditya1: What are you building? What is your MACHINE, DISTRO? Dec 31 03:48:09 oh sorry Dec 31 03:48:12 beagleboard Dec 31 03:48:21 angstrom stable Dec 31 03:49:03 Aditya1: wow, it might be the best to take it to the angstrom mailinglist. Dec 31 03:49:30 hmm Dec 31 03:49:34 what did you mean by "Sticky" Dec 31 03:51:08 Aditya1: if it happens once it will happen again Dec 31 03:51:15 ah Dec 31 03:52:01 Aditya1: the (cross)compiler tries to do something and then it either crashes, or runs into a condition it doesn't like and prints this message. And it is deterministic. Dec 31 03:52:46 oh Dec 31 03:52:47 bleh Dec 31 05:24:40 Anyone working on an updated chromium bb that uses GYP? Dec 31 07:50:56 hi, I have built midori using 'bitbake midori' and have a problem installing it via my feed: Dec 31 07:50:59 * ERROR: Cannot satisfy the following dependencies for midori: Dec 31 07:51:01 * libsoup-2.4-1 (>= 2.27.91) * gtk+ (>= 2.14.2) * libatk-1.0-0 (>= 1.28.0) * libsqlite3-0 (>= 3.6.20) * libwebkit-1.0-2 (>= 1.1.16+svnr50081-r50081) * Dec 31 07:51:21 how is that possible at all? package-inde is up to date Dec 31 07:56:49 filip: which distribution? Dec 31 07:57:06 filip: maybe preferring some older libwebkit? (webkit-gtk) Dec 31 07:57:59 filip: ah nvm there is just one webkit-gtk_git in oe tree Dec 31 07:58:16 filip: but it worked for me in shr and angstrom Dec 31 07:59:18 JaMa: jlime-2009.1 Dec 31 07:59:23 JaMa: but it built fine Dec 31 08:03:00 filip: and did bitbake midori also upgrade libsoup and webkit-gtk for you? Dec 31 08:06:08 NOTE: Running task 1424 of 1433 (ID: 153, /home/filip/pda/oe/openembedded/recipes/webkit/webkit-gtk_svn.bb, do_package) Dec 31 08:06:46 but I can't see the 'Packaged contents of ...' message Dec 31 08:08:00 and no package file... Dec 31 08:11:26 Hi all, i want add -pg and -lc_p parameter to qt4 embedded building, but tried for one day and failed, anybody knows how to do that? Dec 31 08:12:59 i have no idear where is the right position to add those two parameters and have oe correctly building with them Dec 31 08:34:50 leslie__: give me some time, I can show it to you today... Dec 31 08:35:22 filip: bitbake -c listtasks midori. You should see a task that (e.g. buildall) that will also create the packages Dec 31 08:37:24 zecke: thanks zecke and happy new year ;) Dec 31 08:41:32 zecke: I have tried add QMAKE_CFLAGS += -pg in linux-gnueabi-oe-g++/qmake.conf before configure begins, but in log of compile, it doesnt use this setting Dec 31 08:44:02 leslie__: well, CFLAGS is only for C. :) Dec 31 08:46:34 zecke: also the CXXFLAGS Dec 31 08:47:05 zecke: i add QMAKE_CXXFLAGS += -pg, and QMAKE_LFLAGS += -lc_p Dec 31 08:47:12 zecke: it seems no use add here Dec 31 08:47:19 leslie__: why do you use gprof in the first place? Dec 31 08:47:51 zecke: i want use gprof for Qt libs Dec 31 08:48:17 leslie__: do you know oprofile? :) Dec 31 08:48:27 zecke: no idear... Dec 31 08:48:50 zecke: i can google it, so use that one i dont have to rebuild qt to support it? Dec 31 08:49:25 leslie__: you still need to rebuild everything with framepointers on ARM Dec 31 08:50:42 zecke pasted "Compile with framepointers" at http://paste.lisp.org/display/92856 Dec 31 08:51:16 leslie__: ^^ that is what I use for OProfile.. Dec 31 08:52:03 it seems the compilation defaultly have -fno-omit-frame-pointer option? Dec 31 08:53:02 leslie__: really, not a couple of months back. :) Dec 31 08:54:32 leslie__: ah right, in debug build it is enabled :) Dec 31 08:57:23 zecke: i will google for oprofile and use it on qt libs, thanks a lot zecke Dec 31 09:00:47 zecke annotated #92856 "use pg..." at http://paste.lisp.org/display/92856#1 Dec 31 09:01:36 leslie__: this seems to work (if you have a -lc_p then also uncomment the TARGET_LDFLAGS Dec 31 09:01:37 zecke: it is there (do_buildall Dec 31 09:01:44 ) Dec 31 09:01:58 filip: bitbake -cbuildall midori then Dec 31 09:02:28 -pg conflict with -fno-omit-frame-pointer? Dec 31 09:02:43 leslie__: not that I know of Dec 31 09:02:54 03Thomas Zimmermann  07org.openembedded.dev * r4670e963fe 10openembedded.git/recipes/gtk-webcore/midori_0.2.2.bb: Dec 31 09:02:58 midori: 0.2.2 remove retain-navigationbar-in-fullwindow.patch because it's included upstream Dec 31 09:03:00 Signed-off-by: Martin Jansa Dec 31 09:03:08 zecke: but when i directly add -pg in Makefile, it reports they are conflict Dec 31 09:03:11 03Benjamin Schieder  07org.openembedded.dev * ra07f23e6b9 10openembedded.git/recipes/openmoko-3rdparty/neomis_svn.bb: Dec 31 09:03:14 neomis: prevent screen blanking in SHR via fsoraw Dec 31 09:03:16 Signed-off-by: Martin Jansa Dec 31 09:03:44 zecke: you can have a test, maybe need to disable debug mode when use -pg and lc_p Dec 31 09:03:49 zecke annotated #92856 "untitled" at http://paste.lisp.org/display/92856#2 Dec 31 09:03:55 leslie__: not here.. Dec 31 09:08:14 /home/filip/pda/oe/build_dir/tmp/staging/i686-linux/usr/bin/python: error while loading shared libraries: /home/filip/pda/oe/build_dir/tmp/work/armv4-linux/python-2.6.2-ml8.0/Python-2.6.2/libpython2.6.so.1.0: ELF file OS ABI invalid Dec 31 09:08:18 huh Dec 31 09:08:59 filip: aeh? did you manually play with paths? Dec 31 09:09:24 zecke: i'm trying use -pg now, to see whether it will work with the default settings in oe Dec 31 09:09:52 zecke: nothing that I am aware o Dec 31 09:09:53 zecke: en, it's conflict Dec 31 09:09:56 of Dec 31 09:09:56 zecke: arm-angstrom-linux-gnueabi-g++: -pg and -fomit-frame-pointer are incompatible Dec 31 09:10:29 leslie__: not here :) Dec 31 09:11:23 zecke: ... Dec 31 09:11:44 leslie__: at least you know how to set it now :) Dec 31 09:11:54 zecke: yep Dec 31 09:12:18 zecke: thanks man, i'm going home now, it's the last night of 2009 in China this time... Dec 31 09:12:24 zecke: thanks a lot for your help Dec 31 09:15:06 take care Dec 31 09:31:47 zecke: I have errors using @base_version_less_or_equal with strings like 2.6.32+2.6.33-rc2. I see in bitbake vercmp is defined twice (lib/bb/utils.py, lib/bb/__init__.py)? Dec 31 09:39:14 argh, I can't build python because python-native gets called and tries to load arm libpython :/ Dec 31 09:41:51 filip: how does this happen? Does "python-native" have a rpath on the lib? or is LD_LIBRARY_PATH set Dec 31 09:42:02 filip: does it happen when you call the "native" python directly as well? Dec 31 09:45:11 zecke: it works directly: Dec 31 09:45:12 filip@euron-host ~/pda/oe/build_dir $ /home/filip/pda/oe/build_dir/tmp/staging/i686-linux/usr/bin/python Dec 31 09:45:15 Python 2.6.1 (r261:67515, Dec 31 2009, 10:36:47) Dec 31 09:45:56 zecke: should I try bitbake python in a new build_dir? Dec 31 09:50:29 filip: so someone, somewhere is setting a LD_LIBRARY_PATH? Dec 31 09:52:29 zecke: yes, python recipe or python makefiles Dec 31 09:53:03 zecke: is it possible that bitbake python will fail for everybody? or is it something specific to my setup? Dec 31 09:53:28 zecke: I don't have any weird stuff done afaik Dec 31 09:54:17 filip: most likely your setup, FSO is using python... so i would assume it to work Dec 31 10:08:26 zecke: but I have no idea what I could have done wrong :/ Dec 31 10:09:47 filip: use strace and figure out why it is trying this lib Dec 31 10:12:47 zecke: well I think it HAS to be LD_LIBRARY_PATH, will strace tell me anything more? Dec 31 10:15:10 zecke: found why: we want to use utils.vercmp as providers.py. Unfortunately @base_version_less_or_equal uses bb.vercmp Dec 31 10:17:44 and fails comparing some kernel versions Dec 31 10:32:22 morning everyone Dec 31 10:33:47 likewise: do you still feel the owner of the calamari machine stuff in oe? I tried building angstrom for calamari the other day but somewhere along the line building gcc failed for me Dec 31 10:34:04 good morning Dec 31 10:34:17 florian: hi I read your email thx Dec 31 10:45:10 zecke: in /home/filip/pda/oe/build_dir/tmp/work/armv4-linux/python-2.6.2-ml8.0/Python-2.6.2/Makefile in rule sharedmods there is: *s*) $(RUNSHARED) CC='$(CC)' LDSHARED='$(BLDSHARED)' OPT='$(OPT)' $(HOSTPYTHON) -E $(srcdir)/setup.py -q build;; \ Dec 31 10:45:36 zecke: RUNSHARED contains: LD_LIBRARY_PATH=/home/filip/pda/oe/build_dir/tmp/work/armv4-linux/python-2.6.2-ml8.0/Python-2.6.2: Dec 31 10:45:59 zecke: can this result from my misconfiguration? Dec 31 10:51:03 #RUNSHARED= LD_LIBRARY_PATH=/home/filip/pda/oe/build_dir/tmp/work/armv4-linux/python-2.6.2-ml8.0/Python-2.6.2: Dec 31 10:51:06 RUNSHARED= Dec 31 10:51:12 zecke: this made it work Dec 31 10:59:52 florian: good morning Dec 31 10:59:54 zecke: good afternoon Dec 31 11:00:52 eFfeM: yes Dec 31 11:07:01 likewise: does it build for you (I can retry here, i've removed my tmp dir afterwards so don't have the log handy) Dec 31 11:07:36 (restarted the build right now, will keep you posted) Dec 31 11:08:04 (triple tasking as usual) Dec 31 11:15:08 hello to all. does anybody know why my git repo allways break when i try to sync it with the following error message. Dec 31 11:15:09 error: Entry 'classes/kernel.bbclass' not uptodate. Cannot merge. Dec 31 11:15:11 fatal: merging of trees 9263e12da49a97b38eb18332450cb884ea4d2ef3 and 2d6d4fa223c7025f6a67513222847db3a5944f57 failed Dec 31 11:15:12 Merge with strategy recursive failed. Dec 31 11:16:12 xperia: you somehow modified classes/kernel.bbclass Dec 31 11:16:30 you can use git checkout classes/kernel.bbclass to revert to the version in git Dec 31 11:16:37 and after that git pull works again Dec 31 11:17:04 eFfeM: i didnt change anything in this file till yet. will try your suggestion now with checkout ! Dec 31 11:18:52 okay have executed it now. the command returned without any output. will try now again to pull. Dec 31 11:24:47 eFfeM: I think I put a last-known-good revision in your local wiki Dec 31 11:25:16 eFfeM: I used to write down the revision that built from scratch for exactly this purpose Dec 31 11:27:11 eFfeM: I have too many builds going as well, and especially too little disk space to run another. Dec 31 11:33:38 likewise: have just buyed a usb hardisk with 1TB space for under 100 USD. now i have plenty of space to play :-) before i had the same problem like you. needed allways to delete some files before i was able to build something. Dec 31 11:34:50 problem with this 1TB however is that it is one huge fat partition and everytime i do execute a command like ls i have to wait about 1 minute till the full disk is scanned for some reason. Dec 31 11:35:55 eFfeM: hmmm it wont work. will try to delete it with git rm and check it out again Dec 31 11:36:22 xperia: I have multiple disks, one is 1 TB Dec 31 11:37:11 I think I need to use a SSD for the build, then move the build to a large slow disk when finished. Dec 31 11:37:57 wooow and still no free space availble. you are build big things likewise. your cpu must run the whole day at 100% :-) Dec 31 11:38:50 strange git rm classes/kernel.bbclass Dec 31 11:38:51 error: 'classes/kernel.bbclass' has local modifications Dec 31 11:38:52 how can this happen. i have never done some mods to this file. hmmm Dec 31 11:51:32 xperia: if the system thinks there are local mods it will refuse to checkout Dec 31 11:52:35 not sure if git rm is the proper solution; you might want to mv classes/kernel.bbclass away and attempt a git checkout classes/kernel.bbclass and see if the file is indeed made and the problem gone Dec 31 11:52:57 otherwise something is wrong in your git tree (and i'm out of ideas for that one, you might need help from #igt) Dec 31 11:53:41 xperia: wrt your 1TB disk: put multiple partitions on it Dec 31 11:53:48 eFfeM: thanks for the info. have just maked git diff classes/kernel.bbclass and it output zero modification to this file. will try now your mv suggestion. Dec 31 11:54:23 thinking of it, are you building on a fat or ntfs partition? (is your git tree on fat or ntfs) Dec 31 11:54:33 not sure whether that will work, never tried that Dec 31 11:56:10 likewise: i know about the last known good version, but I feel the head should build too, also from the main tree and without overlays, so I tried this at home, and I get http://pastebin.ca/1732940 Dec 31 11:56:36 actually might have one or two small changes pending for calamari but would love to be able to build from the head before modifying things Dec 31 11:57:19 (and currently I'm @ home and decided to be green and power off the system at work for the Xmas holiday so can't really compare) Dec 31 11:57:36 eFfeM: keeping toolchains working can be a time consuming task, especially for the rare machines. Dec 31 11:58:07 true Dec 31 11:59:14 eFfem: it is a one big fat partition at the moment. i have downloaded oe from trunk and think i will also not have success to build it without any problems. Dec 31 11:59:14 but I would expect this to be this suffix of object file thing to be very standard (but could not find anything in site or so for other platfroms either, although I would expect it to apply for everyone) Dec 31 11:59:39 xperia: i would strongly encourage to move to an ext3 partition Dec 31 12:00:14 eFfeM: did you look into the config.log? Dec 31 12:01:02 xperia it might well be that fat does not have certain metadata for files that is needed by git; eg fat has no notion of user and group. Dec 31 12:01:13 xperia this will also cause problems building things Dec 31 12:01:17 likewise: will do Dec 31 12:02:20 hmmm bad to hear this really. allready wasted more than 12 hours for setting up the full oe build environment and need now again to start from the beginning. :-( cry Dec 31 12:02:47 xperia: it'll also take several hours to compile, for me this is an overnight job Dec 31 12:03:23 but really having a single 1TB fat partition is not a good plan at all (also not for windows) Dec 31 12:04:06 one question i wanted to build this on a ubuntu server that run allways at home but have readed that i need bash and not dash. what for distros are you using for building oe normally ? Dec 31 12:04:38 ubuntu works fine, just go to /bin and ln /bin/sh to /bin/bash (instead of dash)