**** BEGIN LOGGING AT Tue Feb 15 02:59:56 2011 Feb 15 06:37:06 gm Feb 15 08:10:39 good morning Feb 15 08:35:32 moin Feb 15 08:37:07 03Martin Jansa  07master * rd4f2f04f74 10openembedded.git/classes/rootfs_ipk.bbclass: (log message trimmed) Feb 15 08:37:07 rootfs_ipk: don't emit error code when ONLINE_PACKAGE_MANAGEMENT is really none Feb 15 08:37:07 * currently do_rootfs fails when it's "none", because that test returns Feb 15 08:37:07 exit code '1': Feb 15 08:37:07 | log_check: Using /OE/tmpdir-shr/work/spitz-oe-linux-gnueabi/initramfs-kexecboot-klibc-image-1.0-r0/temp/log.do_rootfs.21324 as logfile Feb 15 08:37:07 | Logfile is clean Feb 15 08:37:08 | + rm -rf /OE/tmpdir-shr/rootfs/initramfs-kexecboot-klibc-image-tmp Feb 15 08:52:03 I have collecting some build time benchmarks here: https://spreadsheets.google.com/ccc?key=0AhY8lmfkCabTdGMtNWQ5UENxQ2VjZjQ4djNKdmZXdnc&authkey=CJGpv4MB Feb 15 08:52:25 build times per bb/make threads number Feb 15 09:00:27 good morning Feb 15 09:20:26 hey ,, maybe someone clarify how i can make u-boot keep some env variable like ${something} Feb 15 09:20:43 as asoon as a saveenv this it extracts ${something} Feb 15 09:26:14 rob_w: try to escape somehow.. Feb 15 09:26:26 e.g. like \${var} Feb 15 09:26:35 or may be $${var} Feb 15 09:26:35 ah ok .. thx Feb 15 09:26:43 or '${var}' Feb 15 09:27:28 just store 3 vars with different escaping and look :) Feb 15 09:31:50 hmm Feb 15 09:32:12 it works .. but i need something different Feb 15 10:04:55 rob_w: may be I misunderstand you then :) Feb 15 10:05:42 * Jay7 gone Feb 15 10:06:27 mornin Feb 15 10:10:46 03Stefan Schmidt  07master * r3885847bc7 10openembedded.git/recipes/bug-desktop-icons/files/ (4 files): Feb 15 10:10:46 bug-desktop-icons: Fix QA error in desktop files. Feb 15 10:10:46 Its not allowed to use the file extension when not using a full path. Just use Feb 15 10:10:46 the image name here. Feb 15 10:11:37 morning Feb 15 10:18:37 hi mickey|office Feb 15 10:18:38 hi mickey|office Feb 15 10:36:29 XorA|gone: where are you? Feb 15 10:56:26 gm Feb 15 10:57:10 ericben: Hello Feb 15 10:57:24 ericben: you're the one who replied to the qmake issue? Feb 15 11:02:50 hey likewise Feb 15 11:07:23 Jin^eLD: Hey Feb 15 11:07:29 Jin^eLD: you at embedded world 2011? Feb 15 11:09:04 likewise: nope :( Feb 15 11:09:32 not doing any expos lately... Feb 15 11:10:12 03Andreas Oberritter  07master * rdb341fa906 10openembedded.git/recipes/autofs/autofs.inc: Feb 15 11:10:12 autofs: set license field to GPLv2+ Feb 15 11:10:12 Signed-off-by: Andreas Oberritter Feb 15 11:18:37 anybody at Embedded World 2011? Feb 15 11:19:45 likewise: not this year, sorry Feb 15 11:21:30 http://www.linaro.org/snowball-announcement/ - nice dualcore A9 board Feb 15 11:27:44 03Henning Heinold  07org.openembedded.dev * rc8e377aebf 10openembedded.git/recipes/libsdl/ (libsdl-mixer/configure.patch libsdl-mixer_1.2.11.bb): Feb 15 11:27:44 libsdl-mixer: fix compiling for angstroem 2008.1 Feb 15 11:27:44 * libtool don't honors AC_CONFIG_AUX_DIRS in this Feb 15 11:27:44 case so not all libtool stuff is deleted, fixed it Feb 15 11:27:44 by copy the stuff from the build-script dir Feb 15 11:27:44 and remove the dir afterwards Feb 15 11:27:45 * bump PR Feb 15 11:53:10 hi bing Feb 15 11:53:12 hi ericben Feb 15 12:47:30 hi otavio Feb 15 12:47:33 yes I am Feb 15 12:47:39 ericben: nice. Feb 15 12:47:51 ericben: I got some improvement but it still fails Feb 15 12:48:05 ericben: well, qt now seems fine but it fails to identify the x11 headers Feb 15 12:48:16 otavio: I'm slowly working on trying to get our sdk to use the same tree as nokia's one but I lack time actually Feb 15 12:48:53 ericben: I can help on that since I really need a full qt sdk working for our development team Feb 15 12:49:12 otavio: which sdk are you using ? Feb 15 12:49:18 which sdk recipe Feb 15 12:49:32 ericben: we built on our Feb 15 12:49:53 ericben: because we have extra stuff to put on it. Feb 15 12:50:02 ericben: except from the qt all works fine Feb 15 12:50:27 otavio: ok I 'm not using qt for X, only qt embedded so for the X11 part I can't help a lot Feb 15 12:51:09 otavio: what is the error you get concernign the x11 headers ? Feb 15 12:51:14 ericben: I am now checking how to fix it. Feb 15 12:51:24 CMake Error at /opt/sdk/osnetboot-terminal/i586/share/cmake-2.8/Modules/FindX11.cmake:381 (MESSAGE): Could not find X11 Feb 15 12:53:14 the X11_INC_SEARCH_PATH does not include the SDK path Feb 15 13:01:18 ericben: oh; it seems I got it working Feb 15 13:29:15 do_populate_sdk_append() does not properly seem to properly executed herecya Feb 15 13:30:27 can't you guys let me get caught up n my email ..... Feb 15 13:55:01 03Koen Kooi  07org.openembedded.dev * rd9f3073b57 10openembedded.git/recipes/alsa/ (alsa-state.bb alsa-state/omap4430-panda/asound.conf): Feb 15 13:55:02 alsa-state: enable HDMI audio by default Feb 15 13:55:02 Signed-off-by: Koen Kooi Feb 15 14:18:48 ericben, can you look at http://patches.openembedded.org/patch/384/ and if happy make it apply again maybe? thanks Feb 15 14:23:47 I'm a little bit lost in the manual trying to figure out how to specify the packages that go into an image (jffs2, etc) that gets burned to flash. Is there any receipe that's particularly easy to follow? I guess I was expecting to see some IMAGE_* with a huge list of packages Feb 15 14:24:04 demigod2k, you can do that to Feb 15 14:24:09 but most of the time it seems to be more hierarchical where it references some task-* that might include a bunch of packages Feb 15 14:24:26 those will work for more machines Feb 15 14:27:23 one thing that confuses me currently in recipes/image/* every file lists IMAGE_INSTALL but that keyword doesn't appear in the manual. under the image section it talks about PACKAGE_INSTALL Feb 15 14:27:42 did that change, or do I just need to follow the lead of the image-*.bb and use that variable? Feb 15 14:35:12 Tartarus: I add this on the todo list Feb 15 14:54:26 thanks ericben Feb 15 15:39:53 trying to understand alternative update, I have an image w/ busybox for /bin/sh, I add bash and bash is in /bin/bash, but now how do I get the symbolic link /bin/sh to update to the correct value? Feb 15 15:40:34 khem: while building newer (not-yet-in-OE) webkit-efl for armv4t I noticed another binutils-2.21 segfaults (which doesn't happen on arvm7a) http://paste.pocoo.org/show/338981/ Feb 15 15:44:38 Jay7: nice TestBuilder script(s) :) Feb 15 15:45:38 atl19711: why would you want /bin/sh to point to bash? unless you're doing builds in the rootfs its not necessary Feb 15 15:46:34 atl19711: I only have the manual to work from, but did you run the update_alternatives which is what is documented to modify the symlink? Feb 15 15:47:22 kergoth: rt, I want ${D}/bin/sh to be ${D}/bin/bash in my target root Feb 15 15:47:24 kergoth: it would also happen if you had a rootfs which pointed to busybox and then on a writeable rootfs you decided to install a bash pacakge... right? Feb 15 15:48:00 but bash doesn't provide alternative to /bin/sh Feb 15 15:48:07 it's not safe to manage /bin/sh with update-alternatives, at least last time i tried it, because update-alternatives is written in sh Feb 15 15:48:13 see 'cat /usr/lib/opkg/alternatives/sh' Feb 15 15:49:22 As I haven't gotten that far with OE yet I cannot comment, but theyre used in the manual as the example of update-alternatives. if it doesn't work that way it might be worth reworking that section of the manual Feb 15 15:50:03 rt, all opkg / ipk packages work. However, what about an image build that has it correct from the start, is that possible? Feb 15 15:51:47 is oe or friends at embedded world ? Feb 15 15:53:03 demigod2k: I don't know what you're currently reading in manually... just saying that bash is not supposed to change /bin/sh link and bash recipe doesn't use u-a at all Feb 15 15:53:11 s/manually/manual Feb 15 15:53:34 JaMa|wrk: http://docs.openembedded.org/usermanual/usermanual.html#update-alternatives_class Feb 15 15:54:01 thats where it talks about the /bin/sh symlink pointing to busybox or bash. it then follows and works the example of /bin/ping Feb 15 15:54:07 ah ok, so there is wrong example Feb 15 15:55:10 ya its not a huge deal since it's not the worked example. should I put that into the bugtracker? Feb 15 15:55:54 demigod2k: can you send patch for manuall changing that /bin/sh example to ping example used later and also change s#/usr/lib/ipkg/alternatives/#/usr/lib/opkg/alternatives/# in later text? Feb 15 15:56:19 demigod2k: or just s#ipkg#opkg#g :) Feb 15 15:58:01 03Martin Jansa  07master * r5f3bc5b423 10openembedded.git/recipes/openmoko-3rdparty/spojegui_svn.bb: Feb 15 15:58:01 spojegui: bump SRCREV and remove upstream applied desktop file fixes for QA issues Feb 15 15:58:01 Signed-off-by: Martin Jansa Feb 15 16:01:08 JaMa|wrk: is there anything on the site that explains how to calculate a diff using git (this is my first time using that) and then does the patch get sent to the mailing list? Feb 15 16:01:38 the wiki page on patches only explains how to apply them not how to make and submit them so I'm a bit lost Feb 15 16:01:52 http://progit.org/book/ Feb 15 16:01:59 google git documentation Feb 15 16:03:01 ok ya I can do it but it'll be a while. does this project have a patch submission system or do they go straight to the mailing list? Feb 15 16:06:35 mailing list Feb 15 16:10:57 JaMa|wrk: Did you really mean the ipkg/opkg a simple search and replace? Having never used either one, I have no idea if the commandlines needs to change too Feb 15 16:13:16 Hmm Feb 15 16:13:27 I guess I didn't post to the ML about qt4 + uclibc + arm failure, heh Feb 15 16:15:03 JaMa|wrk: thats a compiler problem mostly it seems to me Feb 15 16:15:25 demigod2k: it's the same tool, new name Feb 15 16:15:34 JaMa|wrk: please preprocess that file and with compiler options send it to me Feb 15 16:15:35 "new" Feb 15 16:15:50 there are 48 other occurences in the rest of the manual. should they all get replaced too, or only that one section? Feb 15 16:16:32 sorry for the questions but I just walked up to this last friday and haven't even build a working filesystem yet so I just dont know these things Feb 15 16:17:51 JaMa|wrk: these two files preprocessed is what I need JIT.cpp.o and JITOpcodes32_64.cpp.o Feb 15 16:24:23 has someone saw a build-failure of gcc-cross-sdk? Feb 15 16:24:31 of yestarday's OE Feb 15 16:24:47 Hi, there! I've got one little problem: Feb 15 16:24:49 http://pastebin.com/DK14GGp9 Feb 15 16:24:51 Any idea? Feb 15 16:27:45 hrm, wonder how to convince vim into properly highlighting ${@} within shell functions Feb 15 16:28:08 it highlights shell functions with the shell syntax, is the problem, when ${} and ${@} vars in there are actually expanded by bitbake, not shell Feb 15 16:28:11 hmm Feb 15 16:31:16 khem: seems to happen only with thumb enabled Feb 15 16:31:25 khem: when I force ARM mode again it builds fine Feb 15 16:31:39 khem: will provide more info if needed later, but now gtg Feb 15 16:32:16 khem: btw testing your last gcc patch and sofar everything works (except maybe-unrelated build issue with webkit-efl) Feb 15 16:39:04 http://wiki.openembedded.org/index.php/Release-2011.03#Test_combinations updated Feb 15 16:41:19 khem, do you recall me asking before about qt4 + uclibc + arm? Feb 15 16:51:44 do the extra rdepends bits in package_*.bbclass bug anyone else? just because we enable *emission* of a given type of package doesn't mean we'll be using it for a given image.. Feb 15 16:53:50 Tartarus: apropos qt4 + uclibc + arm is it a working combination with OE? Feb 15 16:54:08 mckoan, no, link failure in qt Feb 15 16:54:12 other uclibcs are fine Feb 15 16:55:09 hmmm Feb 15 16:55:19 sec Feb 15 16:56:19 http://pastebin.com/zwiBKf40 Feb 15 17:04:54 Tartarus: which other uclibcs are you talking to? Feb 15 17:11:00 mckoan, er? see the Release-2011.03 table Feb 15 17:11:23 Tartarus: yes, reading it just now, thx Feb 15 17:11:27 (and fwiw, mips64+uclibc+qt4 will work next time, i checked after the postgresql change) Feb 15 17:11:55 Tartarus: BTW looks not a uclibc related problem Feb 15 17:12:23 hm? Feb 15 17:12:31 mips64 + uclibc seems like an odd combination to me Feb 15 17:12:33 arm+glibc/eglibc+qt is fine Feb 15 17:12:48 khem: any idea about this build failure of gcc-cross-sdk? Feb 15 17:12:49 http://paste.debian.net/107724/ Feb 15 17:12:57 fray, I won't argue that Feb 15 17:13:35 fray, did you look at the crazy table I posted from imgur the other day? :) Feb 15 17:13:51 no id on't think I saw it.. was it to the OE list? if so which topic? Feb 15 17:14:27 I'd rework the qt4e-*images and switch to 4.7.1 Feb 15 17:14:43 http://i.imgur.com/Cbakj.png Feb 15 17:14:49 half the raw data for my table entry, heh Feb 15 17:15:08 mckoan, heh? If you switch to 4.7, the images work Feb 15 17:15:15 I am looking at trying to fix them for 4.6.x tho Feb 15 17:15:34 wow Feb 15 17:16:16 Tartarus: build report (png) looks nice :-) Feb 15 17:16:28 yeah Feb 15 17:16:51 for next week, I figured out how to not split the job into 3 hosts and not have IT hunt me down for having way too many VMs up at once Feb 15 17:19:40 hi to all ... ;) Feb 15 17:20:00 Anyone has compiled openembedded + opie in mini2440 ? Feb 15 17:25:42 have a nice rest of the day Feb 15 17:31:16 I'm trying to build console-image and for some reason this decided to try building glibc (which of course it cannot do for my processor, with an external toolchain). How can I track down why glibc got added as a dependency in this case? Feb 15 17:31:44 I'm thinking to start with the "bitbake -g" to create a *.dot file and try to figure out where it came from Feb 15 17:34:17 Yesterday i have compiled openembedded for my mini2440 with opie . All working . But it is possible to install multimedia player for opie ? Feb 15 17:42:03 ah sorry make that glib. If that's a GNOME library I'm not exactly sure why that's building for a console-image but I suppose the question stands Feb 15 17:48:28 03Henning Heinold  07org.openembedded.dev * rc719c6c185 10openembedded.git/recipes/flac/flac.inc: Feb 15 17:48:28 flac: fix build problems with rpath on older libtool Feb 15 17:48:28 * bump INC_PR Feb 15 18:04:53 hi obi Feb 15 18:12:30 hi woglinde! what do you think about http://git.opendreambox.org/?p=obi/openembedded.git;a=log;h=refs/heads/upstart ? Feb 15 18:13:56 hm Feb 15 18:14:04 we need to support both Feb 15 18:14:11 indeed Feb 15 18:14:13 but I wonder Feb 15 18:14:18 that's what my patches do ;) Feb 15 18:14:20 what does look better in bootchart? Feb 15 18:14:31 upstart or clasic init? Feb 15 18:14:53 i'm not finished converting our scripts, so i can't tell yet Feb 15 18:15:10 with upstart we can start the UI much earlier Feb 15 18:16:07 ok Feb 15 18:16:09 I would think we'd be better off encoding the information into the metadata and generating the scripts from that Feb 15 18:16:09 service start, stop, etc.. Feb 15 18:16:09 * kergoth shrugs Feb 15 18:16:15 I know the advantages of upstart on desktop Feb 15 18:16:27 but is it faster on slower embedded devices Feb 15 18:16:50 of course things get rather more complex when you look at inits that spawn based on sockets and the like Feb 15 18:17:02 systemd support is of course very explicit, no avoiding that.. Feb 15 18:17:03 heh Feb 15 18:18:19 I'm interested in looking into systemd stuff.. Feb 15 18:18:24 systemd compiles not with uclibc Feb 15 18:18:46 and you make sure kernel uses cgroup Feb 15 18:18:46 the combination is init script run and control, with process accounting, resource management and other life cycle control infrastructure looks really interesting to me for embedded systems Feb 15 18:18:46 considering you have to patch many of the daemons to fully support it.. it is very interesting stuff Feb 15 18:18:46 * kergoth nods Feb 15 18:19:05 is systemd compatible with sysvinit scripts? Feb 15 18:19:05 woglinde I've only done basic investigation, but for many of the devices I work on.. it would solve a huge control problem that I have Feb 15 18:19:05 obi, it can be.. but doesn't have to be Feb 15 18:19:05 fray I have started some worked Feb 15 18:19:06 systemd can replace init, cron and inetd type functionality.. Feb 15 18:19:57 with init scripts that tell systemd what to do (vs doing it themselves) you have a lot more resource management and life cycle (health) control for the system.. Feb 15 18:20:08 suddenly the init is capable of watching if a task should be running and terminates... either performing an action, logging and event or doing "something" reasonable with that info Feb 15 18:20:15 but it becomes a system design philosophy.. not just a drop in replacement Feb 15 18:21:01 minor thing, but i do like that it avoids the use of shell in the startup process, no bashisms to fix there at least :) Feb 15 18:21:01 * fray also notes the systems he's been working on need more then traditional "run levels" Feb 15 18:21:16 they truely have different states they must be running in.. add to that power management capabilities.. and it can get interesting Feb 15 18:21:38 shell can be a controller -- but it's required.. that I like Feb 15 18:22:10 socket and dbus based is my current preference.. (but I understand why using dbus isn't necessarily the best answer -- but it helps simplify notifications and remote actions) Feb 15 18:22:50 remote actions? Feb 15 18:22:50 * fray just hasn't had the time to explore any of that stuff in the last 6 months though.. :( Feb 15 18:23:16 you need to care about security Feb 15 18:24:29 woglinde many fo the devices I work with are part of a larger system. The controlling device needs to recive information or control additional devices within that system.. Feb 15 18:24:37 and yes, you absolutely need to control security! Feb 15 18:24:38 and dbus hasnt build in much cryptography Feb 15 18:24:44 obi hm they are working? Feb 15 18:24:44 I'm not worried about intra system security.. Feb 15 18:24:46 (again for my devices.. generically I am worried about security on all levels) Feb 15 18:25:01 but I would only use dbus for command and control -- not for large data set processing or anything that needs to be "encrypted".. Feb 15 18:25:07 dbus is completely the wrong channel for that data Feb 15 18:25:13 fray I wrote dbus remote shutdown/reboot stuff Feb 15 18:25:31 with sender as shell script and dbus-send Feb 15 18:25:35 (other concern with dbus is operations per unit of time can be slow depending on implementation -- I've been told it's much better now) Feb 15 18:25:42 reciver in c with lowlevel api Feb 15 18:25:55 ya, I did some prototyping of that.. we used the standard system permissions to control access.. Feb 15 18:25:56 I hate the glib capsulation Feb 15 18:26:23 if you were not in specific groups then you didn't have access to the interfaces in order to control the state changes.. (state changes are roughly equivalent to run level changes) Feb 15 18:26:51 fray na best would be cert bases stuff Feb 15 18:27:04 than control groups Feb 15 18:29:21 woglinde: yes, it's working with upstart 0.6.7 + libnih + mountall from ubuntu Feb 15 18:29:39 I'm not sure what's changed recently.. but control groups has an advantage of being able to control resource usage, as well as monitor for overage and alert health monitoring systems who could take actions (such as change states -- kill apps, etc) Feb 15 18:30:20 mountall creates events on mounting filesystems. it's not needed, but i like it Feb 15 18:30:24 the monitor on process death -- or monitor on processes accounting indicating larger then certain memory or i/o usage have been really useful in past demo situations.. Feb 15 18:32:03 I looked into upstart before and dismissed it for my projects, but I can't remember why anymore Feb 15 18:33:33 I think at the time it was due to it being GPLv3 or containing components that were GPLv3.. but I don't see that in the current package anymore.. Feb 15 18:34:16 (I've got customers who refuse to use any GPLv3 software due to specific clauses that they can not meet due to regulatory issues in various parts of the world) Feb 15 18:35:16 obi send to ml Feb 15 18:35:18 (they're considered to be part of the safety equipment, and by law safety equipment can only be modified by the manufacturer.. if the manufacturer makes it "easy" to modify they are liable for any damages as a result of modification) Feb 15 18:35:22 I will ack them Feb 15 18:36:30 fray: on this subject, gplv2 or gplv3 whouldn't change much, whould it? Feb 15 18:36:35 would it? Feb 15 18:36:58 woglinde: i'm going to finish converting our stuff first (and to test it a little more) Feb 15 18:37:02 their interpretation is GPLv2 requires source code and such to be given away.. but does not require the end user to be able to modifyt he device.. while GPLv3's "tivoization clause" does require that.. Feb 15 18:37:05 obi okay Feb 15 18:37:20 thus they have made the decision they can not use GPLv3 in this class of products.. Feb 15 18:37:30 fray: Ah I see Feb 15 18:37:34 fray: agreed Feb 15 18:37:55 (these customers have absolutely no problem with source code requiremetns, etc.. but they just can't allow modifications due to legal requirements in certain parts of the world..) Feb 15 18:38:39 (they also believe if they make it difficult to modify -- they have met the burden that the end customer is now liable if it's been modified.. GPLv3 requirements as they read them, requrie it to be "easy") Feb 15 18:38:48 khem: I got your mail about the gcc-cross-sdk ... do you have any clue about the reason? Feb 15 18:44:41 otavio: no clue Feb 15 18:44:53 otavio: you said you had it building Feb 15 18:45:01 I would suggest to do a git bisect Feb 15 18:45:10 from your known spot Feb 15 18:51:28 otavio: whats your distro/machin/image Feb 15 18:53:55 khem: micro/i586 (building on amd64)/internal image Feb 15 18:54:05 ah micro Feb 15 18:54:26 khem: the problem is that I don't know the "good" one Feb 15 18:54:28 otavio: can you try same with say minimal Feb 15 18:54:41 we can debug it Feb 15 18:58:02 khem: but do you have any guess why it would change something? I can use minimal but this would require a build from scratch :-( Feb 15 18:58:34 khem: but I can do that if necessary Feb 15 18:58:39 this could happen if you have a linker script stub in /usr/lib Feb 15 18:58:56 for libc.so Feb 15 18:59:38 khem: right. The distro I am using is based on micro but with /usr prefix/exec_prefix/prefix_native Feb 15 19:00:16 khem: so in theory this should work as in minimal, no? Feb 15 19:02:37 03Michael 'Mickey' Lauer  07org.openembedded.dev * r9993897bd4 10openembedded.git/recipes/freesmartphone/libgisi_git.bb: libgisi: packaging++ Feb 15 19:17:06 03Tom Rini  07master * reaff6a9b68 10openembedded.git/recipes/u-boot/ (2 files in 2 dirs): Feb 15 19:17:06 u-boot 2009.01: Fix 'weak' issues Feb 15 19:17:06 Based on existing patches for other boards. Feb 15 19:17:06 Signed-off-by: Tom Rini Feb 15 19:25:07 hi, I've an issue debugging stuff with gdb...again Feb 15 19:25:32 http://www.pastie.org/1567574 Feb 15 19:25:35 see the from Feb 15 19:26:09 and the "warning: no loadable sections found in added symbol-file" Feb 15 19:26:55 khem, hi Feb 15 19:31:30 hmm Feb 15 19:32:03 Where did you add -g to the cflags? Feb 15 19:32:32 And did you also add them for pn-glibc (or pn-eglibc) ? Feb 15 19:34:52 khem: doing a full rebuild to check Feb 15 19:39:11 GNUtoo|laptop: that means there is no debug info Feb 15 19:39:24 yes I know Feb 15 19:39:30 that's why I pinged you Feb 15 19:39:35 I've no idea why debug info is not there Feb 15 19:42:16 03Khem Raj  07master * rf76026a6c1 10openembedded.git/recipes/gcc/ (11 files in 3 dirs): Feb 15 19:42:16 gcc-4.5: Bring latest from linaro 4.5 and bump svn SRCREV for upstream Feb 15 19:42:16 Please test this patch out in your respective combinations and report Feb 15 19:42:16 and regressions you see. Feb 15 19:42:16 Signed-off-by: Khem Raj Feb 15 19:42:25 03Khem Raj  07master * rf4211a7dca 10openembedded.git/recipes/binutils/ (binutils-2.21/libtool-2.4-update.patch binutils_2.21.bb): Feb 15 19:42:25 binutils_2.21: Fix QA errors due to wrong RPATH Feb 15 19:42:25 Signed-off-by: Khem Raj Feb 15 19:43:10 oh I did not mean to push that gcc stuff Feb 15 19:43:31 khem: hehe too late ;-) Feb 15 19:44:18 03Michael 'Mickey' Lauer  07org.openembedded.dev * r595b1c1538 10openembedded.git/recipes/vala/ (vala_0.11.4.2.bb vala_0.11.6.1.bb): vala: 0.11.4.2 -> 0.11.6.1 Feb 15 19:50:45 evening Feb 15 19:50:47 hehe.. Feb 15 19:50:57 khem: seems we got gcc updates :) Feb 15 19:51:33 Jay7: unpredicted ones ;-) Feb 15 19:51:34 lol Feb 15 19:51:34 ynezz: thanks :) Feb 15 19:51:39 ynezz: feel free to contact me about ideas how to extend testbuilder :) Feb 15 19:51:45 otavio: yeah :) Feb 15 19:52:16 well.. I'll run set of builds tonight on master then to check how bad is it :) Feb 15 19:52:48 * mwester prepares for practically everything to rebuild next time he updates OE... :o Feb 15 19:52:56 hehe Feb 15 19:54:19 Its been reasonably tested but I still wanted to keep it out Feb 15 19:54:19 Hey Mr. khem -- can you try a BE build of xinetd, and comment on the correct fix? (I have a patch to config.sub that will fix it, but that might not be the best way to do so)... Feb 15 19:54:26 what will khem do about the unwanted gcc push? revert? Feb 15 19:54:31 and now my machine has given up on ssh connection Feb 15 19:55:10 GNUtoo|laptop: yes Feb 15 19:55:13 eventually Feb 15 19:55:32 let's try to leave it and look :) Feb 15 19:56:13 I think it will not cause any uproar as I have already tested it on most of arches but you never know when it gets wider testing Feb 15 19:56:31 * Jay7 is optimistic about it :) Feb 15 19:56:32 * mwester ssh's remotely to his build server, moves tmpdir aside, pats it gently on the head for encouragement, and kicks off a full build. Feb 15 19:56:42 mwester: did u send the fix to ml ? Feb 15 19:57:03 Nope. I'm asking advice before I do that. Feb 15 19:57:05 ah Feb 15 19:57:13 tell me what the error is Feb 15 19:57:15 I have lost my OE box Feb 15 19:57:19 so I can try it Feb 15 19:57:24 (Don't want to look more foolish than I am if there is some very simple fix) Feb 15 19:57:25 s/can/cannt/ Feb 15 19:57:42 khem, no hurry. I'll pastebin the log this evening when I'm back from work. Feb 15 19:57:48 mwester: somethings can not be hidden if they exist :) Feb 15 19:57:54 :p :D Feb 15 19:58:22 like recklessness that I have Feb 15 19:58:49 it's an SCM tool - we can always revert. :) Feb 15 19:59:15 I know may be thats somewhere at back of mind Feb 15 19:59:18 hmmm I'm not sure if the revert is the best thing Feb 15 19:59:27 maybe we test and report? Feb 15 19:59:37 GNUtoo|laptop: I am only worried if release gets impacted Feb 15 19:59:41 Gah! I'm gonna be late for my meeting -- back in a bit. I'll report if there are any problems with the build. Feb 15 19:59:42 ahhh ok Feb 15 19:59:49 mwester: ok Feb 15 19:59:59 Tartarus: around ? Feb 15 20:00:23 btw when you'll have time, ping me for the debugging issue I have Feb 15 20:00:56 Tartarus: when does your next testing cycle begin ? Feb 15 20:01:09 GNUtoo|laptop: sure thing Feb 15 20:01:20 thanks a lot Feb 15 20:03:16 GNUtoo|laptop: can you sow me log.do_compile of say libz Feb 15 20:03:41 Does anyone use OE on avr32 here Feb 15 20:04:09 ok Feb 15 20:06:19 http://pastebin.com/D7ddkJjM Feb 15 20:06:25 no I've no avr32 Feb 15 20:06:39 maybe ulf on the mailing list? Feb 15 20:07:04 khem: once you squash those issues ping me and I update mine to see if it keeps building Feb 15 20:07:15 khem: I am building using gcc before your push, with minimal. Feb 15 20:07:23 khem: well report if it works in some minutes Feb 15 20:11:46 otavio: you can update and build too it should work Feb 15 20:11:49 GNUtoo|laptop: I think your build is not emitting any debug info at all Feb 15 20:11:49 khem: I prefer to build before and update later so we know if it changes something or not Feb 15 20:11:49 otavio: yes sure Feb 15 20:11:49 hehe.. Nokia have plan B Feb 15 20:11:49 I might end up reverting in the evening Feb 15 20:11:49 revolution in company Feb 15 20:11:50 Jay7: and plan is ? Feb 15 20:11:51 khem, ouch Feb 15 20:11:51 khem: gcc-cross-intermediate-4.5-r31.1+svnr168622 <= at it Feb 15 20:11:51 http://nokiaplanb.com/ Feb 15 20:11:52 otavio: yes thats before my stuff Feb 15 20:11:55 Jay7: ah the 9 good souls Feb 15 20:11:55 indeed no -ggdb3 Feb 15 20:11:55 strange Feb 15 20:11:55 GNUtoo|laptop: whats your distro Feb 15 20:12:03 this f14 box is slow its keeps swapping Feb 15 20:12:05 I will redo it with natty Feb 15 20:12:09 Jay7: funny. hmm, this is interesting: "Dramatically increase efficiency by eliminating outdated and bureaucratic R&D practices like geographically distributed software development" Feb 15 20:12:10 SHR Feb 15 20:12:36 http://esr.ibiblio.org/?p=2961 Feb 15 20:12:37 hollisb: I think that last few words are not ok Feb 15 20:13:14 but I have seen that it can be very inefficient if you dont distribute the projects properly Feb 15 20:13:16 so they are right in some aspects Feb 15 20:13:41 I have seen it in two places where their indian and russian outsourcing bombed Feb 15 20:13:53 bitbake.conf: DEBUG_OPTIMIZATION = "-O -fno-omit-frame-pointer -g" Feb 15 20:14:17 why is it -g and not -ggdb3? Feb 15 20:14:26 because they thought hmm lets distribute the same project so the same code is being worked on 24/7 Feb 15 20:14:26 I know ggdb3 is huge....but Feb 15 20:14:50 GNUtoo|laptop: we optimize too much that -g3 tends to be useless Feb 15 20:15:06 ok Feb 15 20:15:16 -g2 is ok then? Feb 15 20:15:23 http://nokiaplans.com/ Feb 15 20:15:27 hehehe Feb 15 20:15:44 nokia could have done better with android Feb 15 20:15:50 nokia plan 9 is to port plan9 on the phones? Feb 15 20:15:58 haha Feb 15 20:21:53 anyone know why ltrace stops emitting useful output once glib programs went into the mainloop? Feb 15 20:35:06 mickeyl: you mean glib or glibc ? Feb 15 20:37:14 anyway ltrace would trace the calls from binary into dynamic libraries Feb 15 20:37:30 so once its inside libc calls within libc wont get traced Feb 15 20:39:58 argh hit another snag. for some reason the cairo package tries to compile with an -march=armv4t instead of an -march=armv5te and that won't work. What controls that flag, especially something that might be overridden on a package by package basis? Feb 15 20:40:37 demigod2k maybee they have some build in stuff? Feb 15 20:40:55 cairo has support for arm vfp and arm-neon Feb 15 20:41:11 demigod2k: which distro ? Feb 15 20:41:47 khem: minimal, except I modified that by commenting out the arm-thumb include because thumb doesn't work for this chip Feb 15 20:41:57 demigod2k: ok Feb 15 20:42:00 here is the problem Feb 15 20:42:18 TARGET_CC_ARCH_pn-cairo_armv5te = "-march=armv4t" Feb 15 20:42:24 in distro/include/sane-toolchain.inc Feb 15 20:42:31 khem, I think he meant glib, not glibc Feb 15 20:42:41 like glib-2.0 Feb 15 20:42:44 ok Feb 15 20:43:06 demigod2k: you can get rid of that line Feb 15 20:43:27 khem, so what should I do to get debugging in SHR? I add -ggdb2 in shr.conf overriding bitbake.conf? Feb 15 20:43:31 or in local.conf? Feb 15 20:43:35 khem looks some line from old ages? Feb 15 20:43:39 oh. I was checking in recipes, totally missed that. thanks! I'll check there in the future :) Feb 15 20:43:49 woglinde: could be Feb 15 20:44:04 demigod2k: if removing that line works for you Feb 15 20:44:06 report back Feb 15 20:44:12 so we can take care of it Feb 15 20:44:46 GNUtoo|laptop: ah glib hmmm no idea Feb 15 20:44:55 unless its not a shared library Feb 15 20:45:02 khem: building right now. My earlier question about glib was worked-around by removing arm-thumb. It was complaining that swp isn't supported. Rather than debug I just pulled it out entirely Feb 15 20:45:24 demigod2k: what is your arch Feb 15 20:45:24 -rwxr-xr-x 1 root root 745192 Jan 8 15:03 /usr/lib/libglib-2.0.so.0.2600.1 => shared Feb 15 20:45:28 it seems fairly new Feb 15 20:45:37 basically here's what we are trying to do Feb 15 20:45:52 mickeyl has a progam that segfaults( fsogsmd) Feb 15 20:45:56 I tried gdb Feb 15 20:45:58 didn't work Feb 15 20:46:01 khem: Marvell 78200 processor, external-toolchain Feb 15 20:46:05 he seem to have tried ltrace Feb 15 20:46:08 didn't work either Feb 15 20:46:21 gnutoo why didnt gdb worked? Feb 15 20:46:35 I bet because no -ggdb was included in the libraries I used Feb 15 20:46:42 yeah Feb 15 20:46:46 where should I add that? Feb 15 20:46:49 bitbaske is reporting a url for tinderbox.openembedded.net Feb 15 20:46:52 local.conf could work Feb 15 20:46:56 where do we fix that? Feb 15 20:47:07 but maybe It need to be fixed somewhere else Feb 15 20:47:11 like in shr.conf Feb 15 20:47:16 or bitbake.conf Feb 15 20:47:34 crofton glib should be compiled on angstroem with -ggdb too Feb 15 20:47:39 install -dbg package Feb 15 20:47:46 Crofton|work: grep tinderbox conf/* Feb 15 20:47:53 demigod2k: hmmm 78200 is armv5te based and should support swp Feb 15 20:47:56 btw, I should fix it in testbuilder Feb 15 20:48:07 args dman Feb 15 20:48:07 yeah got it Feb 15 20:48:23 gnutoo and crofton looks similiar on irc Feb 15 20:48:37 I wonder how many more occurences of that are there :) Feb 15 20:49:00 woglinde, all -dbg are installed Feb 15 20:49:07 and gdb is setup correctly Feb 15 20:49:09 khem: I could throw it back in to get the exact error text, but the compiler choked on it being an unsupported instruction. I've hit a million other problems since the feroceon core is arm-like not actual arm so it didn't surprise me. but again I didn't research it deeper Feb 15 20:49:10 Crofton|work: fix it please Feb 15 20:49:11 the problem is that one: Feb 15 20:49:25 http://www.pastie.org/1567574 Feb 15 20:49:44 I changed it on my machine Feb 15 20:49:44 according to khem the .debug libs lack the debug infos Feb 15 20:49:45 khem: could also have been compiler issue. since its some marvell-provided thing god only knows what it does Feb 15 20:50:18 sigh maybe I change in local.conf and forget about it? Feb 15 20:50:27 or should I try to fix? Feb 15 20:51:52 khem: it's still cranking away on other packages, but bitbake got past do_configure and do_compile. Would have to get the binary onto a target to know for sure if the compiled result actually works Feb 15 20:51:59 demigod2k: on sheevaplug Feb 15 20:52:00 Sheevaplug 1.2 GHz Feb 15 20:52:00 Processor : Feroceon 88FR131 rev 1 (v5l) Feb 15 20:52:00 BogoMIPS : 1192.75 Feb 15 20:52:02 Features : swp half thumb fastmult edsp Feb 15 20:52:56 khem: two possible differences there -- that's a lot different series of chip, plus that could've been with the mainline gcc instead of the marvell provided gcc Feb 15 20:53:28 78200 is in the discovery innovation series. I dont understand the difference from the 88* series Feb 15 20:54:09 demigod2k: on the Manuals its based on ARM v5TE compliant Sheeva™ 88FR571-vd CPU Feb 15 20:54:28 03Yuri Bushmelev  07master * red2703741e 10openembedded.git/contrib/testing/ (testbuilder/setup-testbuilder testscript.sh): testbuilders: change oestats server to tinderbox.openembedded.org Feb 15 20:54:39 if it says its ARM v5te compliant then it can not punt swp Feb 15 20:54:57 eFfeM_work: I've updated oestats hostname in your script too Feb 15 20:56:31 gnutoo at least you know the function where to look Feb 15 21:01:05 Jay7, does that change you just pushed matter? Feb 15 21:01:08 Or just consistency? Feb 15 21:01:33 Tartarus: it changes only files in contrib to be consistent Feb 15 21:01:39 ok, thanks Feb 15 21:01:41 Tartarus: when will you fire your next build cycle Feb 15 21:01:47 somehow me and eFfeM have done same mistype :) Feb 15 21:02:18 03Martin Jansa  07master * r648bd1c785 10openembedded.git/conf/machine/nokia900.conf: Feb 15 21:02:18 nokia900: add kernel-module-ubifs to RRECOMMENDed modules Feb 15 21:02:18 Signed-off-by: Martin Jansa Feb 15 21:02:28 khem, friday evening Feb 15 21:02:38 branching on thurs again Feb 15 21:04:06 khem: ya, you've got me... gcc reports 4.3.2 but it complains. could be a bug, could be intentional. since the core isn't from arm my guess would've been that it's not quite compliant Feb 15 21:04:49 screwy marvell you can not register on their extranet if you are not using company email address Feb 15 21:05:04 so I can not access this freakin document Feb 15 21:05:06 correct. I'm on there and just looked but I didn't find SWP in a quick search through the errata Feb 15 21:05:37 ok as long as its not OE problem I dont care so much Feb 15 21:07:12 correct OE has no direct support for this chip (you have to use external toolchain, etc). seems like a glib issue really. worst case disabling thumb has gotten me around the issue -- I can dig deeper later Feb 15 21:11:45 03Melissa Watkins  07master * r5a1546f015 10openembedded.git/recipes/ti/ (4 files): Feb 15 21:11:45 ti-pru-sw: Update SRC_REV to 24 Feb 15 21:11:45 For the ti-pru-sw-examples, ti-pru-sw-app-loader, ti-pru-sw-edma-driver, Feb 15 21:11:45 ti-pru-sw-edma-library: Feb 15 21:11:45 * Update SRC_REV to 24 Feb 15 21:11:46 Signed-off-by: Melissa Watkins Feb 15 21:11:46 Signed-off-by: Khem Raj Feb 15 21:11:56 03Imran Mehmood  07master * rff02b94de7 10openembedded.git/conf/bitbake.conf: Feb 15 21:11:56 bitbake.conf: Fix wrong comparison Feb 15 21:11:56 The expression ('' or 'custom') will always result in 'custom' so the expression will become False when the variable (TARGET_OS, SDK_OS) is '' while its expected to be True in this case also. Feb 15 21:11:56 Signed-off-by: Imran Mehmood Feb 15 21:11:56 Acked-by: Christopher Larson Feb 15 21:11:56 Signed-off-by: Khem Raj Feb 15 21:22:29 well.. Feb 15 21:22:54 time to look what is broken with new gcc ;) Feb 15 21:23:12 khem: what distros are affected? angstrom-2010.x and minimal? Feb 15 21:28:29 I sent a pull-request with many small issues that were queued up in our internal tree Feb 15 21:28:36 Can someone take a look? Feb 15 21:40:35 mrmoku: hello, I finished german translation, http://wiki.freesmartphone.org/index.php/Feb11PRDraft/DE Feb 15 21:41:14 mrmoku: i did not correct the content, so i should change pig parts of the article .... wont burn a big discussion Feb 15 21:42:43 mrmoku: now the translation can be reviewed, can you inform vanous please ? he is not online i think Feb 15 21:42:48 nschle85: good, will take a look. Not sure if I manage today though Feb 15 21:44:05 mrmoku: no problem, now I did my task Feb 15 21:44:13 nschle85: yeah, thanks Feb 15 21:44:44 mrmoku: where will this article be releasd Feb 15 21:47:03 nschle85: the english original is on blog.shr-project.org and planet.openmoko.org Feb 15 21:47:53 wow om.org servers are still online Feb 15 21:54:35 mrmoku: how can the german translation be found ? there is no link on english wiki page Feb 15 21:59:12 http://pastebin.com/ZR3QB0Tn Feb 15 21:59:28 anyone have any thoughts why this dies so early and does not leave a log behind? Feb 15 21:59:54 (see /dev/null for further information) is not very helpful :) Feb 15 22:00:28 those have been showing up more often Feb 15 22:00:32 which is kinda bad, yeah Feb 15 22:01:35 heh Feb 15 22:01:41 understatement Feb 15 22:01:52 I know the autofoo has some issues Feb 15 22:02:02 and I would like to complain to the authors :) Feb 15 22:13:55 Tartarus, any thoughts on how to get some idea of hhow it is failing? Feb 15 22:15:00 Crofton|work: look ingo temp dir for log.do_configure.* Feb 15 22:15:06 s/ingo/into/ Feb 15 22:15:12 not off-hand, I had a "see /dev/null" issue pop up last night but it wasn't re-creatable Feb 15 22:15:27 no log Feb 15 22:16:05 I fear that issue with lzma-native was only top of iceberg Feb 15 22:17:07 Jay7, well, could you reliably recreate it? Feb 15 22:17:16 And if so, did you bisect bitbake perhaps? Feb 15 22:17:25 Tartarus: no, I have seen it only once after that 'fix' Feb 15 22:18:04 I have done about 16 builds after already w/o any problems Feb 15 22:20:43 florian: around? Feb 15 22:21:16 Jay7, but you could reliably make it happen before? Feb 15 22:21:29 Jay7, how about before that change and bitbake 1.10.2 ? Feb 15 22:21:47 Tartarus: not reliable but more frequently Feb 15 22:21:53 ka6sox-farfarawa: traveling, are you? Feb 15 22:22:23 Tartarus: I'm building always with bb master.. Feb 15 22:22:30 so, no 1.10 there Feb 15 22:22:59 Right Feb 15 22:23:07 But if this is a bitbake bug we need some point to bisect Feb 15 22:23:17 mwester: yup Feb 15 22:23:23 ah, did you mean to try reproduce it with 1.10? Feb 15 22:23:28 Yes Feb 15 22:23:31 well, good idea Feb 15 22:23:56 right now I have testbuilder running to check khem's gcc updates Feb 15 22:24:17 I'll run set of builds at morning then with 1.10 Feb 15 22:29:48 let me switch back to 1.10 Feb 15 22:32:29 That's probably way farther back than needed Feb 15 22:32:33 But it'll be a good test, heh Feb 15 22:41:19 Tartarus, I got messages from 1.10 bitbake Feb 15 22:41:50 ok Feb 15 22:41:56 Can you reliably _not_ get messages from master? Feb 15 22:49:15 otavio: send the link please Feb 15 22:49:21 to your tree Feb 15 22:49:38 Jay7: yes minimal angstrom-2010, micro Feb 15 22:51:00 mwester: it seems http://www.xinetd.org/ is down so I could not download xinetd tar Feb 15 22:54:55 03Khem Raj  07master * r638f45b390 10openembedded.git/recipes/gcc/ (11 files in 3 dirs): Feb 15 22:54:55 Revert "gcc-4.5: Bring latest from linaro 4.5 and bump svn SRCREV for upstream" Feb 15 22:54:55 This reverts commit f76026a6c1a00bd0b2f1ae2b045dee995293edc6. Feb 15 22:54:55 Pushed prematurely Feb 15 22:54:55 Signed-off-by: Khem Raj Feb 15 22:55:10 hm.. Feb 15 22:59:46 Jay7: you can still continue to test it but I did not want to push it yet Feb 15 22:59:59 Jay7: think you have pulled it from pw Feb 15 23:00:21 I've pulled it from master :) Feb 15 23:00:28 anyway Feb 15 23:01:03 I'll post results tomorrow :) Feb 15 23:02:01 collie builds is almost done (~7000/~9000 tasks) Feb 15 23:03:01 Jay7: sure it will help me Feb 15 23:11:28 angstrom-2010.x/collie/console-image+x11-image+opie-image is fine with new gcc Feb 15 23:13:34 * Jay7 -> sleep Feb 15 23:13:59 Jay7: I want someone to try non-arm/le Feb 15 23:14:12 khem: ben nanonote is mips Feb 15 23:14:20 MACHINE=nslu2be DISTRO slugos Feb 15 23:14:21 but it is at the end of queue :) Feb 15 23:14:26 Jay7: ok Feb 15 23:14:48 you can look for progress on oestats under Jay7-tb Feb 15 23:15:41 I can estimate that nanonote build will be started after ~4 hrs Feb 15 23:16:20 * Jay7 gone finally :) Feb 15 23:21:22 khem: Gentoo upgraded headers..let see...[ebuild U ] sys-kernel/linux-headers-2.6.36.1 [2.6.30-r1] Feb 16 00:03:53 03Chris Larson  07master * rdbc0cf9701 10bitbake.git/lib/bb/server/process.py: Feb 16 00:03:53 server: add profiling support to ProcessServer Feb 16 00:03:53 Based on poky's support, particularly the format of the processed log file and Feb 16 00:03:53 the log message shown to the user regarding the files written. Feb 16 00:03:53 Signed-off-by: Chris Larson Feb 16 00:07:47 03Chris Larson  07master * rcd43144f0f 10bitbake.git/lib/bb/server/process.py: Feb 16 00:07:47 server: clean up call of write_profile_stats Feb 16 00:07:47 Signed-off-by: Chris Larson Feb 16 00:28:03 Question on the upcoming release -- are commits to the mainline being merged into the release, or do I have to make arrangement to have that happen? Feb 16 00:29:59 We haven't branched, yet Feb 16 00:30:06 I guess I hadn't made that clear enough, sorry Feb 16 00:30:20 I probably didn't read closely enough. Feb 16 00:30:43 That's good news then -- I shall continue on stabilization rather than SlugOS features. :) Feb 16 00:30:50 Did you see [oe] 2011.03 release testing, starts soon! ? Feb 16 00:32:28 Yep - I checked the matrix you emailed out. Feb 16 00:37:27 I wasn't clear enough in there then, another email sent out :) Feb 16 00:37:34 You've got at least until the 24th ;) Feb 16 02:54:07 No one fix the db RPATH issues yet? **** ENDING LOGGING AT Wed Feb 16 02:59:57 2011