**** BEGIN LOGGING AT Tue Sep 27 02:59:57 2011 Sep 27 06:44:49 hello i m compiling gsoap for mini2440 ....using but I m getting an errors while compiling Sep 27 06:45:02 can anyone help me... Sep 27 06:48:05 stdsoap2_ssl_cpp.cpp:4161: error: invalid conversion from 'const X509V3_EXT_METHOD*' to 'X509V3_EXT_METHOD* Sep 27 08:48:22 sanket: perhaps you need another openssl version Sep 27 09:01:51 what's a good way to search my build directory to find out if a package is already part of my image, so I don't add it twice? Sep 27 09:02:45 I require console-base-image.bb, but don't know all it adds, and am in the process of adding back the packages we used in oe-classic. Sep 27 09:09:30 tasslehoff: maybe inherit testlab and compare the installed*.txt Sep 27 09:10:35 ant_work: thanks. Sep 27 09:10:49 anyone got a working mirror for cpufrequtils? Sep 27 09:56:58 where can I find a good mirror of linux-firmware.git now that kernel.org is down? Sep 27 10:00:26 doesn't angstrom have mirrors for most of this? how can we make our recipes check those? Sep 27 10:06:04 some Sep 27 10:06:09 not everything Sep 27 10:06:24 morning mickey|office Sep 27 10:06:29 It looks at http://www.angstrom-distribution.org/unstable/sources/git_git.kernel.org.pub.scm.linux.kernel.git.dwmw2.linux-firmware.git_a3ffb1a98408f713ff82d59ae1fd203d650974c4.tar.gz, but gives a 404 Sep 27 10:35:12 morning ka6sox Sep 27 10:37:03 morning who has the keys to openembedded.net's DNS? Sep 27 10:37:18 there are some changes we need to do to re-aim for new server. Sep 27 10:55:27 .net is under my control Sep 27 11:51:16 what's the meaning of "referenced by DSO" ? Sep 27 11:51:18 errorlog --> http://pastebin.com/yDEwTktr Sep 27 12:13:48 gdb-cross doesn't work. It waits forever after this line: Sep 27 12:13:53 NOTE: package gdb-cross-7.2-r9.0: task do_fetch: Started Sep 27 12:14:04 The file gnu/gdb/gdb-7.2.tar.gz doesn't exist on the mirrors. Sep 27 12:14:48 Cubi_, i had a similar problem with binutils, just vistit the server and change the bb Sep 27 12:18:23 same problem with all gdb recipes Sep 27 12:18:39 I'll modify my .bb Sep 27 12:18:56 Cubi_, maybe they point to kernel.org ? Sep 27 12:19:40 and some were renamed to ie 7.2a.. Sep 27 12:23:03 lautriv, same problem with all mirrors, somebody added an 'a' to the files Sep 27 12:24:06 And the md5 sum has changed Sep 27 12:24:54 exact the reverse of my problem, missed the a in the filename :-) looks like there was a serious leak in the toolchain, removed in the a version. Sep 27 12:26:37 maybe a git pull solves it....... Sep 27 12:27:26 it wasn't leak http://nickclifton.livejournal.com/9067.html Sep 27 12:28:03 lautriv, git pull doesn't help Sep 27 12:30:42 it was fixed in oe-core.. if you're using classic OE you have to adapt something like http://git.openembedded.org/cgit.cgi/openembedded-core/commit/?id=c27690a04acf3870f6e0133cd05ba93d6c66126e and send it to oe-devel list for someone to merge it Sep 27 14:05:36 hello. Sep 27 14:10:40 why is bitbake downloading and compiling programs that i already have installed? like, it seems to be downloading and building git-1.7.3.5 when 1.7.1 is installed and works perfectly fine Sep 27 14:19:21 genew, define installed Sep 27 14:19:57 $ git --version Sep 27 14:19:57 git version 1.7.1 Sep 27 14:20:12 installed on host with apt-get? Sep 27 14:20:17 correct Sep 27 14:20:23 ok Sep 27 14:20:47 for git I don't know why it needs it Sep 27 14:20:51 maybe for autorev Sep 27 14:21:26 it is now downloading the latest version of bzip2 Sep 27 14:22:07 the biggest wtf was when it downloaded the full source for linux kernel 2.6.31 Sep 27 14:22:22 im running 2.6.35 and trying to cross compile 2.6.39 Sep 27 14:22:51 it's probably getting the linux headers from 2.6.31 Sep 27 14:23:07 Yes, for libc. Sep 27 14:23:23 In general OE tries to minimize the dependencies on the host system. Sep 27 14:23:35 This means it downloads a lot of stuff. Sep 27 14:23:44 i can see how it seems like a good idea in theory Sep 27 14:23:45 genew: if you don't want to compile these you have to add ASUME_PROVIDED or so to your local.conf Sep 27 14:23:49 for these packages Sep 27 14:23:51 It also ends up installing lots of things (like bzip) on the target so it needs the source. Sep 27 14:24:43 http://www.openembedded.org/index.php/Advanced_configuration#ASSUME_PROVIDED Sep 27 14:26:34 not just to minimise the dependencies, in some cases native versions of some utilities are built because additional patches are applied/options enabled or the version generally available in most distros is too old Sep 27 14:27:17 so, it builds a whole 12gb linux system within a linux system just to be extra sure that everything will work? Sep 27 14:27:28 bluelightning: That's why the dependencies are minimized, yes. Sep 27 14:28:20 i'd probably be praising it, except it ran into an error when building Sep 27 14:28:20 :3 Sep 27 14:29:39 Actually I wish the dependencies got cut down a bit more, things like svn are annyonig. Sep 27 14:31:09 genew: what was the error? Sep 27 14:31:23 1 sec Sep 27 14:32:04 /bin/sh: /home/gene/junk/ooe/tmp/work/armv7a-angstrom-linux-gnueabi/freetype-2.4.3-r1/freetype-2.4.3/builds/unix/arm-angstrom-linux-gnueabi-libtool: No such file or directory Sep 27 14:32:35 not entirely sure what it is trying to do... Sep 27 14:32:54 clean toolchain and restart your build Sep 27 14:33:03 ah no Sep 27 14:33:04 sorry Sep 27 14:33:16 clean freetypee rather Sep 27 14:33:24 with -c cleansstate Sep 27 14:37:19 hmm, it seems that freetype worked this time. thanks Sep 27 14:44:59 okay, it is attempting a git clone now, but it doesnt seem to be working Sep 27 14:48:48 actually, i think it is working, there's just no status information :( Sep 27 15:14:32 Hello Sep 27 15:16:38 We found a quite ugly bug and we have no idea where to begin to look at it ... when we run bitbake inside a bash shell it /fail/ to pass PSEUDO_DISABLE=1 deep in the shell ... so when it tries to ssh to a remote repo (git in this case) it try to use /root/.ssh and gets stuck waiting for type 'yes' to the new host hash. If I run it inside zsh shell it works. Sep 27 15:16:57 This happens with today's oe-core Sep 27 15:17:06 and today's bitbake Sep 27 15:20:30 otavio: what host distro is this on? when did it start happening? Sep 27 15:28:25 is it simple to set up native compilation of kernel modules on the target system? i asked the gentleman who is meant to be maintaining the system i am working with, and he seems to think that all the header files are in place, but the kernel makefile (and other unknown files) are not present Sep 27 15:30:40 bluelightning: debian Sep 27 15:30:59 bluelightning: we're preparing an autobuilder and then it showed up Sep 27 15:32:14 bluelightning: squeeze Sep 27 15:32:50 bluelightning: Sep 27 15:32:51 (devel)~/hacking/ossystems/embedded-linux% dpkg -s bash | grep Version Sep 27 15:32:51 Version: 4.1-3 Sep 27 15:33:21 otavio: hmm, that is concerning... Sep 27 15:34:23 otavio: you know the last working rev then you could bisect it I would say Sep 27 15:34:42 khem: no Sep 27 15:34:47 hmmm Sep 27 15:34:50 khem: I am a zsh user Sep 27 15:34:58 khem: so for me it works and keeps working Sep 27 15:35:06 so when you change from zsh to bash you see it Sep 27 15:35:20 khem: but the builder would use bash and then this was discovered Sep 27 15:35:27 I see Sep 27 15:35:32 khem: i did the inverse, from bash to zsh Sep 27 15:35:38 khem: on the builder Sep 27 15:35:47 * otavio test in his home Sep 27 15:36:14 I guess bash is most used shell amongst devs here so its surprising Sep 27 15:36:55 khem: moving from zsh to bash doesn't show it ... Sep 27 15:37:13 khem: so it must be something related but not bash itself ... Sep 27 15:38:38 * otavio changes his shell in passwd Sep 27 15:39:39 * bluelightning is confused... Sep 27 15:39:43 * otavio too Sep 27 15:39:57 * otavio testing with a clean shell and it works grr... Sep 27 15:40:13 'PSEUDO_DISABLED': '1' Sep 27 15:43:22 In which case FAKEROOTENV is used? Sep 27 15:43:36 it sets PSEUDO_DISABLED=0 and this is what I see Sep 27 15:45:51 otavio: Dont do anything with PSEUDO_DISABLED Sep 27 15:46:00 instead use PSEUDO_ENABLED Sep 27 15:46:17 err I mean when you dont want to set it Sep 27 15:46:22 dont set it to 0 even Sep 27 15:46:34 khem: I am not messing with it; the point is the builder had it as 0 and I am trying to figure WHY Sep 27 15:47:57 that should be from sanity.bbclass Sep 27 15:48:18 khem: I don't think so as we got it when it was running git ls-remote Sep 27 15:48:32 khem: so something on the env made it decide to use fakerootenv Sep 27 15:48:42 * otavio went to have lunch Sep 27 16:36:14 Where could I locate full BitBake documentation instead of doing RTFC? Sep 27 16:41:54 * otavio is back Sep 27 16:42:09 khem: do you have any clue what might be messing with this? Sep 27 16:42:12 bluelightning: ? Sep 27 16:43:42 otavio: there must be something unusual going on on the system you're seeing this issue on, we'd have found this issue already otherwise Sep 27 16:52:58 do you guys have a workaround/mirror I can use to get cpufrequtils to build? Sep 27 16:54:51 bluelightning: yes ... I agree ... but any clue how to debug it? Sep 27 16:59:27 otavio: without digging into it myself, no... Sep 27 17:01:17 Cannot register service: RPC: Authentication error; why = Client credential too weak Sep 27 17:01:30 when using runqemu with nfs build Sep 27 17:01:37 root Sep 27 17:01:52 the solutions I find Sep 27 17:02:03 are for fedora Sep 27 17:02:54 this is ubuntu 11.10 server Sep 27 17:27:35 bluelightning: can you give us a hint how we can trace why this is set ? mario-goulart and I are working on this Sep 27 17:27:53 bluelightning: it seems git ls-remtoe is running with fakeroot, what makes no sense et all Sep 27 17:30:06 bluelightning: the env seems like it is with fakerootenv set; dunno why though Sep 27 17:39:54 is there actually a problem known in conjunction with ldd and debian x86_64 as build-host ?( i get errors about "referenced by DSO" ) Sep 27 17:39:59 mario-goulart: try to run bitbake -D to see if it outputs something regarding fakeroot on the git recipes Sep 27 17:40:23 mario-goulart: i'd say: bitbake -D liboshwcollect or so Sep 27 17:43:15 just a quick general question here: anyone know if there are plans on getting a glibc recipe > 2.10 in one of the OE repos? Sep 27 17:44:07 otavio: ok Sep 27 17:45:41 bmouring: no there are no plans Sep 27 17:45:49 bmouring: actually glibc has been removed Sep 27 17:46:14 its eglibc thats used now and in fore seeable future Sep 27 17:46:39 khem: eesh, well thx for the info Sep 27 17:47:39 otavio: nothing new, apparently. Gets stuck on "git ls-remote ..." Sep 27 17:47:40 is there any reason you want glibc Sep 27 17:47:54 mario-goulart: is it kernel recipe ? Sep 27 17:48:19 ko dot org is down for few weeks now Sep 27 17:48:28 khem: no. It's an ossystems git repository. Sep 27 17:48:58 khem: it works when fakeroot is not into play. Sep 27 17:49:12 hmm Sep 27 17:49:26 khem: just what I had been using and had been OK with, I'll just try to switch stuff over anyway Sep 27 17:49:28 you should use pseudo Sep 27 17:49:31 for reliable results Sep 27 17:49:37 We're trying to find out why it works for otavio and it doesn't work on the environment I'm using. Sep 27 17:50:43 mario-goulart: but the log shows anything regarding fakeroot? Sep 27 17:51:15 khem: I'm actually using pseudo, but for some reason FAKEROOTENV is setting PSEUDO_DISABLED=0 for that particular task. Sep 27 17:52:51 khem: we are using pseudo and didn't mess up with this but seems like something is using fakerootenv Sep 27 17:52:56 khem: dunno why yet Sep 27 17:54:45 khem: being "that particular task" "git ls-remote ...". :-) Sep 27 17:56:04 So ssh tries to get the ssh configuration from root, which is not what we expect. Thus, ssh hangs waiting for valid credentials. Sep 27 17:56:58 What we are trying to find out is why PSEUDO_DISABLED=0 is coming into play. Sep 27 17:58:08 ah Sep 27 18:03:45 FAKEROOTENV though define in bb.conf is not used in oe-core Sep 27 18:03:58 so make sure how this is being read in Sep 27 18:07:47 Doesn't bitbake use it? Sep 27 18:08:34 bitbake's cache.py has: Sep 27 18:08:36 self.fakerootenv = self.getvar('FAKEROOTENV', metadata) Sep 27 18:09:14 Maybe you mean I should be sure what config file is being read? Sep 27 18:09:32 sure about* Sep 27 18:33:19 khem: it seems we managed to restrict how it is happening, not why yet Sep 27 18:34:06 khem: it seems package_setscene runs with fakeroot set, and for unkown reason it calls git lsremote on one machine but not in another Sep 27 18:35:05 where is the default kernel configuration saved in the openembedded file tree? Sep 27 18:35:47 flyer-xplorer, it's saved in the file://defconfig Sep 27 18:36:19 i might need to back up a bit. where is file://defconfig mapped to? Sep 27 18:36:20 so for instance if your kenrel is linux-foo Sep 27 18:36:26 it will be in the linux-foo or linux-foo-3.0(if your version is 3.0) Sep 27 18:36:43 or if it's simply linux in linux Sep 27 18:36:53 use find to find it once in the correct folder Sep 27 18:46:09 oh found Sep 27 18:57:53 khem: OK; we isolated the issue ... what changes the behaviour is the sstate usage ... Sep 27 18:58:32 when we use gitpkgv, for unknown reason, it calls git lastest rev Sep 27 19:01:26 hmmm Sep 27 19:01:30 sstate Sep 27 19:02:54 mario-goulart: yes find out why bitbake executes that code path which uses fakerootenv Sep 27 19:04:13 khem: we get GITPKGV and this seems to be run when unpacking sstate Sep 27 19:05:05 GITPKGV is in oe-core Sep 27 19:05:12 I doubt that Sep 27 19:05:16 khem: nops, meta-oe Sep 27 19:05:22 hmmm Sep 27 19:05:27 khem: gitpkgv class Sep 27 19:05:40 correct Sep 27 19:06:15 khem: it seems we should have add the headrev in the persistent database or other way Sep 27 19:06:50 yeah there needs to be a refrence Sep 27 19:07:08 khem: any clue where to look how to make it? Sep 27 19:07:53 can you guys compile oe-core now? do you have a solution/hack for the issues with the mirrors for kernel.org recipes? Sep 27 19:08:02 * tasslehoff won't leave this alone ;) Sep 27 19:08:36 tasslehoff: in local.conf: KERNELORG_MIRROR = "http://mirrors.arsc.edu/kernel" Sep 27 19:08:48 otavio: thanks! Sep 27 19:08:52 tasslehoff: np Sep 27 19:10:44 khem: what bothers me is that sstage supposely uses PV only, not PKGV Sep 27 19:10:46 otavio: GITPKGV's nature is that it should be run everytime right ? Sep 27 19:10:54 khem: yes Sep 27 19:11:09 khem: if you're using AUTOREV, yes ... Sep 27 19:11:25 khem: if you're using a tag and it is local already, not required to run Sep 27 19:11:38 you can make PKGV into PV then sstate will take care of it Sep 27 19:11:58 khem: ?? Sep 27 19:11:59 I need to get some food Sep 27 19:12:37 otavio: anything special I need to do to make it use the mirror? bitbake -c fetch cpufrequtils fails Sep 27 19:12:55 does it mirror all stuff fetched from kernel.org, or just ... kernels? Sep 27 19:13:19 tasslehoff: only tarballs, not git repos Sep 27 19:14:04 otavio: ok. no solution for the git repos? Sep 27 19:14:13 tasslehoff: not that I am aware of Sep 27 19:14:21 github? Sep 27 19:14:57 mario-goulart: a possible alternative if the required repo is there .. but this also involves you changing the .bb file Sep 27 19:15:23 I see. Sep 27 19:18:18 used github for the kernel recipes, but haven't found a decent repo for cpufrequtils Sep 27 19:32:59 tasslehoff: Don't know how current it is, or how reliable the site is, but http://mirror.nexcess.net/kernel.org/scm/utils/kernel/cpufreq/cpufrequtils.git/ Sep 27 19:36:05 kergoth: thanks. than I should just input that into SRC_URI with a protocol=git statement? Sep 27 19:36:59 replace http:// with git:// and protocol=http Sep 27 19:40:07 kergoth: ah. perhaps that was the error in my previous attempt with another mirror. thanks Sep 27 19:40:33 the *protocol* here is http, as that's how it gets the files, but the *fetcher* we need is git Sep 27 19:40:46 not sure if there's a way to specify the url in the reverse way, but i don't think so Sep 27 19:52:02 should I use gentoo's bitbake 1.13.3 over the stable release 1.12? Sep 27 19:54:41 yes. 1.13.3 is the best choice right now Sep 27 20:10:46 is beagleboard a seperate target from beagleboard-xmc? Sep 27 20:12:15 flyer-xplorer: its better to check it out than using it from distro installed Sep 27 20:12:20 after all its python Sep 27 20:17:24 would it be best to move the .oe profile stuff into a shell's normal init directory? Sep 27 20:20:39 I see that python-simplejson has not made it from oe-classic to oe-core. Anyone know why? Sep 27 20:36:34 what is the different between bitbake's base-image and console-image targets? Sep 27 21:07:00 how do I install applications into the console-image? Assuming I want to install gpsd: bitbake gpsd builds the ipk, but does it deploy it into the console-image? Sep 27 23:36:24 how can I get oe to re-download the sources from git if I accidentally messed them all up? Sep 27 23:43:55 I just throw away my downloads folder.. Sep 27 23:50:52 will that force it to refetch the git repo? Sep 28 01:00:03 starting to think that the oe i am using for this build is not playing by the normal rules **** ENDING LOGGING AT Wed Sep 28 02:59:56 2011