**** BEGIN LOGGING AT Wed May 30 02:59:58 2012 May 30 03:10:08 Greetings all. Thanks to nerdboy and vmeson, I got my kernel compiling. May 30 03:11:08 I'm still not sure in which file task virtual/kernel is defined. How can I get bitbake to list all the tasks it knows about? Or list its config options? May 30 03:11:15 Maybe a verbose dry run is best? May 30 03:12:28 what do you want to? May 30 03:12:46 bump the kernel rev or define your own kernel recipe? May 30 03:13:36 i think my kurobox hard disk is finally dying... May 30 03:14:04 nerdboy, I want to know the pathname of all the files the tasks are using. May 30 03:14:27 This information seems fundamental, beyond the current task. May 30 03:15:40 i'd start with `find . -type d -name tasks` and look there May 30 03:16:11 but i still don't see the point... May 30 03:16:26 is this quest for information kernel-related? May 30 03:18:02 tasks are pretty high level May 30 03:18:21 it least in the arago version we're using... May 30 03:18:37 For kernel, yes, and in general. I'm not crazy to want to know how the tool I'm using works, am I? May 30 03:18:58 in principle, no... May 30 03:20:16 but in this case i'd spend more time building and working at the package/image level May 30 03:20:56 get a feel for it first and then delving into the internals will make more sense May 30 03:21:58 you must have more concrete goals than just satisfying your curiosity? May 30 03:23:17 Yes, nerdboy, in addition to satisfying my curiosity, I want to become proficient at using the tool and configuring it at a high level. May 30 03:27:03 then local.conf and the other conf/[machine|include] files, then package and image recipes May 30 03:28:16 and depending on where you start (branch/tag wise) and what repo(s) you're using, you may have to fix a few things to get consistent clean builds May 30 03:30:16 i had to upgrade gstreamer for our build recently back-porting the recipes was actually easier than getting the right stuff in our images... May 30 04:41:41 hello all May 30 05:46:18 morning. I want to insert a couple of folders into my oe-core built rootfs.tar, without messing up permissions. anyone know how to do that? May 30 06:04:55 * tasslehoff found the correct fakeroot-incantation, and shuffles back to his corner May 30 07:26:57 good morning May 30 07:53:00 one question...the SRC_URI[md5sum] in the recipes is the sum of the md5 for every files in the SRC_URI? May 30 07:58:53 stuk_gen: just for the downloaded tarball (or whatever), not for the patches and files in the openembedded repos May 30 07:59:18 tasslehoff: oooh good! May 30 08:00:09 tasslehoff: so i have a problem :) i'm trying build a systemd-gnome-image but there isn't on a server the netbase_4.47 May 30 08:00:34 so i download the file and calc checksum and put on the SRC_URI[md5sum] of the recipes May 30 08:00:54 but every want i run the build...i have an error that said the checksum is bad May 30 08:01:25 and everytimes i change the checksum that is display, its change! May 30 08:03:06 -.- May 30 08:03:29 f**k my error i haven't create a done file so every times its download the error files May 30 08:14:49 puh :( facing same supid problem again building xfce-nm-image http://paste.ubuntu.com/1014376/ May 30 08:15:44 do_rootfs failed with cannot symlink from .var/vache to volotile/cache May 30 08:38:38 morning all May 30 08:40:19 hi :) May 30 09:16:39 bluelightning any idea? http://paste.ubuntu.com/1014376/ May 30 09:23:55 ericbutt1rs: not really no I'm afraid May 30 09:27:28 bluelightning see here: http://patches.openembedded.org/patch/21343/ May 30 09:27:35 similar problem May 30 09:28:07 but i can not find the cause for xfce-nm-image May 30 09:28:53 but i removed CONMANPKGS from xfce-nm-image.bb May 30 10:05:05 bluelightning i got a console-image running. i need to run firefox. so i can install with opkg. what do you think is best way? using gpe feeds from /next and xserver-xorg from v2012.05 ? May 30 10:05:29 ericbutt1rs: mixing feeds is likely to give you pain May 30 10:05:44 ok.. i need a minimal x11 system May 30 10:06:29 only firefox controlled by mouse May 30 10:22:30 where do i find v2012.xx-core feed? http://www.angstrom-distribution.org/repo/?pkgname=twm May 30 10:52:53 hi, all! How could I using OE-core force some recipe to be built using thumb instruction set? May 30 10:55:11 slapin: look at the tune files. For armv5te thumb interworking is default May 30 10:55:34 then some recipes can disable that and opt for arm code May 30 10:56:43 see e.g. http://patches.openembedded.org/patch/26155/ May 30 10:57:52 df -h May 30 10:58:53 ant_work: I want binary to be built with -mthumb, thumb-interwork allows interoperability of thumb and arm code, not binary in thumb May 30 11:00:50 I meant the site files do set both -mthumb -mthumb-interworking May 30 11:01:18 s/site/tune/ May 30 11:02:17 http://cgit.openembedded.org/openembedded-core/tree/meta/conf/machine/include/tune-thumb.inc May 30 11:02:40 look at the ones you're using May 30 11:10:53 slapin: now that I look at it, I see arm is the default. Thumb was maybe in Angstrom long ago, can't remember. May 30 11:34:10 ant_work: I want a list of recipes to be built with thumb. with classic, it was possible, with ARM_INSTRUCTION_SET_pn-foo = "thumb", is it possible with oe-core? May 30 12:22:25 slapin: I'd say yes, why not? May 30 12:25:31 ant_work: what does _pn-foo mean, is it package or recipe name? May 30 12:28:47 slapin: irc _pn is only appropriate in global configurations (e.g. distro config) May 30 12:29:52 ant_work: local.conf? May 30 12:30:15 yes, local.conf is fine May 30 12:30:46 this is rather obscure override May 30 12:31:01 ant_work: is it package name or recipe name? May 30 12:31:33 found! pls look at this thread: http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-July/022114.html May 30 12:33:06 so it is the same as ${PN}, so it is recipe name, thanks! May 30 12:33:46 yes, package name May 30 13:00:41 added gpg to my image, but have problems decrypting a file with a passphrase. http://pastebin.com/VLJnaD9R May 30 16:14:34 bitbake -c devshell is failing for me. Here's the error: | (gnome-terminal:24456): GLib-GIO-ERROR **: No GSettings schemas are installed on the system May 30 16:14:40 Anyone have any clues for me. May 30 16:14:43 ? May 30 16:30:28 awozniak: hmm, I wonder if theres an environment variable that gnome-terminal expects to be set that we're not letting through... May 30 16:31:18 this is after upgrading to Ubuntu 12.04 LTS. I think I've seen it before, but I forgot how I worked around it last time. May 30 16:31:51 I can run this just fine from the command line: gnome-terminal --disable-factory -t 'Bitbake Developer Shell' May 30 16:56:08 khem, hi May 30 16:56:45 bluelightning: any clues which env variables? =) May 30 16:58:07 awozniak: I'm not sure I'm afraid, I don't use gnome-terminal... but odds are it's something beginning GNOME or GTK if it is that May 30 16:58:52 awozniak: you could try clearing them to see if you could reproduce the failure running outside of a build May 30 16:59:25 Tried that already with GNOME_* env vars (there are only three). No Joy. Trying now with GTK_MODULES (the only GTK_* var I have) May 30 17:00:17 no joy unsetting GTK_MODULES either May 30 17:00:27 can't get the same failure May 30 17:09:13 awozniak: hmm, well I'm not sure what would make it fail other than the environment being different May 30 17:09:51 awozniak: if it helps at all, you can set OE_TERMINAL to use a different terminal application May 30 17:16:32 i am trying to build the lzo package from openembedded using gcc 4.5 and eglibc 2.10. However, the build fails with a series of messages: Error: offset out of range. Does anyone know how to fix this? May 30 17:17:12 khem, ping me when you're available May 30 17:17:27 in the meantime I'll try to narrow more the issue May 30 17:19:05 GNUtoo-desktop: whats up May 30 17:19:07 I am here May 30 17:19:16 for sometime May 30 17:19:22 I've strange results May 30 17:19:31 I built denzil with angstrom and that patch: May 30 17:19:34 hmm May 30 17:19:38 ok May 30 17:20:22 http://pastie.org/private/apwia3hzculcuvci5iuxmq May 30 17:20:38 normally it should have used gcc 4.7 May 30 17:20:42 I'll verify that May 30 17:20:48 but it worked with gcc 4.7 May 30 17:21:00 I mean firefox and chromium ran May 30 17:21:04 ok May 30 17:21:06 *ran fine May 30 17:21:09 so you are all set May 30 17:21:11 I'm now checking something May 30 17:21:13 no May 30 17:21:22 in SHR we don't use denzil May 30 17:21:24 but master May 30 17:21:28 ok May 30 17:21:35 you told me to use gcc 4.7 with denzil to make a test May 30 17:21:53 gcc 4.7 is not in denzil at all May 30 17:21:58 did you backport it to denzil May 30 17:22:23 let me look May 30 17:22:37 I've a toolchain-layer May 30 17:22:47 in meta-oe May 30 17:22:53 also look at my diff May 30 17:23:01 *the pastie May 30 17:23:39 ok. is that diff with angstrom ? May 30 17:23:46 yes May 30 17:23:55 I think easiest is that you check the output of bitabke -g May 30 17:24:05 to see which recipe it picked for gcc cross May 30 17:24:22 angstrom uses yet another variable to pin compiler May 30 17:24:50 OE-Core uses GCCVERSION May 30 17:24:54 arm-angstrom-linux-gnueabi-gcc (GCC) 4.5.4 20120305 (prerelease) May 30 17:24:58 indeed you're right May 30 17:24:59 and angstrom uses ANGSTROMGCCVERSION or somesuch May 30 17:25:19 so it works ok with gcc 4.5 atleast now we know that May 30 17:25:33 yes we already knew yesterday May 30 17:34:36 khem, now I try angstrom-denzil with 4.7? May 30 17:35:10 well, I think what I would like is if you can reproduce it say on oe-core+meta-browser May 30 17:35:23 that way we eliminate a lot of layers May 30 17:35:43 and we still have gcc 4.6 in OE-Core so we can then try once with 4.6 and once with 4.7 May 30 17:35:53 and make sure that its just gcc thats causing it May 30 17:36:29 khem, meta-browser depends on a lot more than just oe-core May 30 17:36:41 GNUtoo-desktop: ok what does it depend on May 30 17:36:46 add those layers too May 30 17:36:54 I mean have minimal changes May 30 17:37:29 once thing I want to make sure is that master has changed far and more than gcc May 30 17:37:36 and its not something else May 30 17:37:47 lot of packages are upgraded that can impact too May 30 17:38:18 I tried using the same rootfs May 30 17:38:28 and just using the firefox or chromium package May 30 17:38:35 ok May 30 17:38:38 and what was the result May 30 17:39:00 firefox+chromium only worked when buit with denzil May 30 17:39:07 that is to saay gcc 4.6 May 30 17:39:36 beside it's clear that it's the compiler in chromium and firefox cases May 30 17:39:44 the backtraces prooves it May 30 17:39:53 _init() is only in the compiler May 30 17:40:04 and the javascript trampoline stuff was in firefox May 30 17:40:07 not in some lib May 30 17:42:12 ok can you try using gcc-4.6 on the setup you have ? May 30 17:43:23 set ANGSTROM_GCC_VERSION = "4.6%" in your local.conf May 30 17:43:32 and see if it still works May 30 17:43:48 in the same setup where it failed when using gcc 4.7 May 30 17:43:51 ok May 30 17:44:01 I located where ANGSTRIOM_GCC_VERSION is May 30 17:44:07 and it doesn't use a % May 30 17:44:12 but I'll use 4.6 May 30 17:45:20 thanks a lot I'll do that May 30 17:45:24 and try the result May 30 17:50:33 ok it does not hurt to use % btw May 30 17:51:56 what's the % for? May 30 17:52:00 like a wildcard? May 30 17:53:25 yes May 30 17:53:49 ok May 30 17:55:15 it'll take a long time for testing tough May 30 17:55:28 because I'm already compiling chromium to do a test May 30 17:56:02 because I didn't pay attention to the hash in the file for fetching the layers May 30 17:56:09 but it's an old version of meta-browser May 30 18:56:15 how can i build a meta-toolchain that has more then one possible target arch? May 30 19:05:59 otavio: you a project manager? May 30 19:14:23 Here's the answer to my repeated question as to how to tell where tasks are defined: May 30 19:14:51 "bitbake -e virtual/kernel | grep virtual | grep kernel" May 30 19:15:39 And here's the answer to my question, how can I tell which directory to put a modified linux.config into for kernel rebuild? May 30 19:16:01 otavio: more than 1 arch ? May 30 19:16:03 Answer: "bitbake -e virtual/kernel | grep -e ^S=" May 30 19:16:16 Thanks to challinan for the answer on different channel. May 30 19:16:24 ccssnet: yes, I am May 30 19:16:29 otavio: you can build multiple meta-toolchains and install them on same machine May 30 19:16:31 khem: yes May 30 19:16:43 I was gonig to ask khem at his upcoming talk, but now he's off the hook. May 30 19:16:44 khem: but it can't be done at same time? May 30 19:17:24 otavio: prolly not May 30 19:17:40 Alison_Chaiken: bitbake -e is like preprocessed output May 30 19:17:48 in different dir's :) May 30 19:17:49 of course, as advised, using a modified defconfig in your kernel recipe is not the "right" way ;) May 30 19:17:53 you can redirect it to a file May 30 19:17:55 by different bb's :) May 30 19:18:05 otavio: depends on computer resources for same time May 30 19:18:07 challinan: howdy May 30 19:18:18 hey khem, how's it going! May 30 19:18:18 is your dell netbook behaving now :) May 30 19:18:29 so far so good. May 30 19:18:36 yup, running a poky-derived sato image!! May 30 19:18:44 oh wow May 30 19:18:50 thats nice May 30 19:19:01 based on MACHINE=emenlow from meta-intel May 30 19:19:03 what do you use for graphics May 30 19:19:23 I have a USB install of atom based image for my acer machine May 30 19:19:24 i built 'core-image-sato' and it Just Worked(TM) May 30 19:19:29 yeah May 30 19:19:54 I am more interested in getting KDE/OE up on the machine May 30 19:20:23 anyone hitting 'attr' build failures? oddly, in make install it's failing to find libattr.lai May 30 19:20:24 hmm May 30 19:20:42 khem: it's certainly doable, just much more work ;) May 30 19:20:58 Alison_Chaiken: if you are using linux-yocto then the way to define your additions to defconfig May 30 19:21:02 is different May 30 19:21:12 I'm still using bitbake. May 30 19:21:15 there is document on yocto website May 30 19:21:19 khem: I want 2 more apps on my dell: firefox and skype and I'm happy May 30 19:21:32 challinan: ah, May 30 19:21:44 then you should look into layering meta-browser May 30 19:21:51 I should modify the kernel-building recipe (.bb) file then to point to the .config I want, in my own directory somewhere? May 30 19:22:09 Good luck with the skype, challinan. May 30 19:22:22 Maybe steve ballmer is listening on here and will help. May 30 19:22:23 khem: I built firefox successfully from meta-mozilla, but haven't tried it on the target yet. no time May 30 19:23:06 challinan: ok btw. meta-mozilla is merged into meta-browser now May 30 19:23:14 Alison_Chaiken: it's a great portable skype platform, but I've never had success with it under ubuntu that came with the machine May 30 19:23:39 khem: ah, ok. things move fast, this was just a couple weeks ago ;) May 30 19:23:46 otavio: check pm May 30 19:23:49 challinan: use meta-browser May 30 19:24:00 otavio, thanks May 30 19:24:12 * challinan also wants to play with webkit May 30 19:26:28 challinan: yeah, I have kde based images built few weeks ago May 30 19:26:41 with webkit+chromium May 30 19:26:50 but did nothin with it yet May 30 19:26:53 khem: nice! so I know who to ask ;) May 30 19:27:24 challinan: it was for arm arch though May 30 19:27:30 I need to try it on beaglebone May 30 19:27:34 if I can May 30 19:27:43 qtwebkit2 with WebGL and HTML5. I was amazed at how short HTML5 files have fast, seamless integrated videos. May 30 19:28:09 Alison_Chaiken: yes html5 amazes me May 30 19:28:49 the little brown truck just delivered my beaglebone - LCD panel right behind it May 30 19:29:54 challinan: hmm, I use vnc May 30 19:29:58 got no LCD May 30 19:31:20 * challinan likes vnc May 30 19:31:23 my old server/firewall uses vnc and enlightenment May 30 19:33:46 enlightenment! A #tizen mole. May 30 19:35:51 o no! a ptxdist mole May 30 19:36:07 gasp... a buildroot mole May 30 19:36:35 * ccssnet points at 20 more source distros May 30 19:37:25 * dm8tbr calls the pest-control emergency number May 30 19:38:24 bitbake -c virtual/enemieslist May 30 19:40:50 Ah, bitbake -c *clean* virtual/enemieslist. May 30 19:41:00 No wonder I can never get anything to work, even jokes. May 30 19:42:15 * ccssnet pokes the joke May 30 19:43:12 they where good May 30 21:33:38 I just noticed, none of the PowerPC executables produced by OE-core as pic.. they all appear to be non-pic May 30 21:34:15 I'm not sure why this is, all of the libraries though do appear to be pic.. so I'm wondering if the wrong crtbegin.o is being users, or if it's something more confusing then that May 30 21:34:28 Anyone have enough compiler knowledge to help me figure out what is wrong? **** ENDING LOGGING AT Thu May 31 02:59:58 2012