**** BEGIN LOGGING AT Wed Sep 28 02:59:56 2011 Sep 28 06:42:09 good morning Sep 28 08:22:04 I have created a recipe with a valid git-repo and a valid SRCREV, but bitbake says that git ls-remote gives emtpy output. What do I do? Sep 28 08:22:58 and: when I can't find a recipe I used in oe-classic, is it because not all recipes are ported/copied to oe-core yet, or could it just be removed? Sep 28 09:34:38 tasslehoff: The first reason is true. It has not been ported yet. But you should also look into meta-openembedded. Sep 28 09:35:07 But if you do not need the recipe you can also remove it or you do not need to port it. Sep 28 09:52:54 hello. I'm trying to build for 486sx and building eglibc I get undefined reference to Sep 28 09:52:54 `__i686.get_pc_thunk.bx ; How do I fix it? Sep 28 09:54:47 I fould that there's a patch called tls_i486.patch for glibc but I don't know how to apply it Sep 28 09:56:13 PaulePanter: it is stuff I need. libjson-dev, python-simplejson and libconfig. Sep 28 09:56:53 I've taken the recipes from the classic tree and added them to my meta-layer, and they compile fine. Sep 28 10:00:45 what does the porting process consist of? a critical look, a cleanup and a decision on where to place the recipe? Sep 28 10:17:07 tasslehoff: I guess that is correct. In my opinion the clean up should take place in OE-classic. Then import it to meta-oe, update/reset some variables like `PR` and add `LIC_FILES_CHECKSUM`, write good commit message and send it to the list. Sep 28 10:20:41 PaulePanter: ok. I'll try to clean up the ones I use and send the patches. Sep 28 10:21:50 must I RTFS to learn about the LIC_FILES_CHECKSUM? one of the recipes is LGPLv2, and I have no idea what I'm supposed to set the checksum to. Sep 28 10:32:01 tasslehoff: LIC_FILES_CHKSUM isn't really dependent on LICENSE Sep 28 10:32:22 tasslehoff: it's just a way of forcing an error if the upstream changes the license notice Sep 28 10:32:51 tasslehoff: so LIC_FILES_CHKSUM should refer to file(s) or portions thereof that should not change unless the license is changed Sep 28 10:33:04 i.e. COPYING, source file header, etc. Sep 28 10:33:25 the yocto docs are probably the most comprehensive available on this variable Sep 28 10:33:54 bluelightning: understood Sep 28 10:34:22 tasslehoff: http://www.yoctoproject.org/docs/poky-ref-manual/poky-ref-manual.html#usingpoky-specifying-LIC_FILES_CHKSUM Sep 28 10:35:04 I think I sent some patches for that section in the next version of the docs (to come with the yocto 1.1 release) Sep 28 10:46:20 Q: If I want to run my application with hardfp, while everything else was compiled with softfp, must I recompile the whole system or it's fine to recompile just my application? Sep 28 10:46:36 I would expect to recompile the whole system Sep 28 11:31:58 hm. after I've added all my stuff from oe-classic to oe-core, my image.tar.bz2 has gone from 45MB to 15MB. I'm kinda suspicious :) Sep 28 11:32:57 tasslehoff: indeed that is a rather large reduction Sep 28 12:29:41 should one not use init.d scripts anymore? Sep 28 12:45:20 tasslehoff: there is work about replacing initscripts with systemd Sep 28 12:49:00 ant_work: ok. thought that was my issue, but I have worse problems: http://pastebin.com/BshdeNS5. still, my new MLO, u-boot and rootfs+kernel actually managed to boot and reach the prompt :) Sep 28 13:24:40 how do I install my images into SD card? Is there any tutorial? Sep 28 13:25:07 script maybe? Sep 28 13:28:21 just extract them? Sep 28 13:28:35 but do it as root Sep 28 13:28:55 extract .tar.gz into SD card? Sep 28 13:28:58 will it be bootable? Sep 28 13:30:04 no, you have to make it bootable, by installing a bootloader, or using the bootloader of your device Sep 28 13:35:21 and how to install bootloader? Sep 28 13:35:52 depends on your device Sep 28 13:36:07 x86 device Sep 28 13:36:19 like normal pc Sep 28 13:36:55 there you can install e.g. syslinux Sep 28 13:38:08 how? Sep 28 13:42:36 for syslinux it's just "syslinux /dev/...." like discribed in the manpage: http://linux.die.net/man/1/syslinux Sep 28 13:47:09 and what should be the structure of sd card? Sep 28 13:47:18 like 64 MB fat16 + ext2? Sep 28 13:49:43 why that for x86? Sep 28 13:49:52 dunno Sep 28 13:50:25 the fat partition is only required by some OMAP afaik Sep 28 13:50:41 so ext2 for whole device? Sep 28 13:50:56 consider the sd as a normal disk (well, almost...) Sep 28 13:52:30 syslinux: invalid media signature (not a FAT filesystem?) Sep 28 13:52:36 so syslinux requires FAT? Sep 28 13:53:45 syslinux was just the first bootloader i thought of, you can use any other too. Sep 28 13:54:07 but for syslinux you need a FAT partition that stores your kernel and initrd Sep 28 13:55:02 maybe I can use some unetbootin tool or something to create bootloader + images + ... for me? Sep 28 13:55:26 anyone going to be upset if I commit a recipe for fftwf based on the fftwf-3.3.1-beta tarball and call it 3.3.1 Sep 28 13:55:33 in oe-classic Sep 28 13:59:21 mar: sorry, I forgot the FAT requirements of syslinux Sep 28 13:59:55 modern bootloaders just read in /boot Sep 28 14:01:17 so what bootloader should I use? Sep 28 14:13:27 many are using grub on x86 Sep 28 14:13:50 but for simplicity stay on Syslinux at the beginning Sep 28 14:15:08 ok, so using syslinux i need two partitions, right? Sep 28 14:17:06 yes Sep 28 14:21:40 Does it surprise people to find that ${@FOO} expands to the value of the variable FOO? Sep 28 14:23:10 I guess we use the data store as the local variable store so it makes sense, its just not what I'd expected Sep 28 14:35:44 should empty syslinux just reboot PC? Sep 28 15:30:41 Ugh, angstom's site is always so fricking slow. Sep 28 15:32:23 mewyn: too many people hammering the image builder, so I'm told Sep 28 15:37:08 bluelightning: *nod* They probably need download mirrors for some of the stuff they have. 10kB/sec transfer rates are *painful* Sep 28 15:37:28 mewyn: yes, I've experienced the same thing Sep 28 15:38:22 mewyn: you can comment out the lines in angstrom-mirrors.bbclass if it's the PREMIRRORS that are causing the issue Sep 28 15:38:35 then it will just go to the upstream source Sep 28 15:38:56 of course then you may run into problems with stuff on kernel.org... Sep 28 15:43:38 Yeah Sep 28 15:43:44 I am Sep 28 15:43:51 This is really getting annoying. Sep 28 15:47:32 what's the url of the angstrom mirror site? Sep 28 15:47:49 I forget if it is at osuosl, or linuxtogo Sep 28 15:49:03 www is at linuxtogo Sep 28 15:49:10 they may have benn bw throttled again Sep 28 15:50:16 Crofton|work: http://www.angstrom-distribution.org/unstable/sources/ Sep 28 15:52:17 yep, that is ltg Sep 28 15:52:55 check with florian, he may have some insight Sep 28 15:53:33 it will probably be better Oct 1 Sep 28 16:09:36 what package controls which modules are autoloaded at boot? Sep 28 16:14:01 flyer-xplorer: /etc/modules.conf Sep 28 16:15:18 that file has the names of modules that should be autoloaded Sep 28 16:15:20 at boot Sep 28 16:17:27 or is it /etc/modules I dont remember OTOH Sep 28 16:20:19 thanks Sep 28 16:24:11 khem: at one point we had both: 2.4 and 2.6 ;) Sep 28 16:38:30 if I have patched a broken .bb file, how should I push the changes upstream? Sep 28 16:40:02 flyer-xplorer: http://www.openembedded.org/index.php/How_to_submit_a_patch_to_OpenEmbedded Sep 28 17:11:50 Hi, I setup a build envornment following instructions on the oe website with the oebb.sh scrip. I need to build something not currently in a recipie, is this feasible with an environment setup like this? Sep 28 17:13:45 vcs: yes you need to write recipe for your app Sep 28 17:16:34 alright, thanks, thats what i thought Sep 28 18:51:31 I've been working on some OE recipes on a repo that was forked off of the 2011.03-maintenance branch. Is the transition to openembedded-core complete enough that I should consider switching to the new repos? Sep 28 18:51:47 Or is stuff still in flux? Sep 28 19:08:55 pingswept, I would start thinking about transitioning Sep 28 19:09:27 crofton: Stuff is more stable than it was a few months ago? Sep 28 19:09:44 well, I am not sure yet, but I am in the same position Sep 28 19:10:21 Has there been a stable release of any sort? A tagged version? Sep 28 19:10:26 not yet Sep 28 19:10:40 to the best of my knowledge Sep 28 19:11:01 if you are not having issues with your current system, you may want to wait Sep 28 19:11:08 Yeah, I don't see one in the git history, but I'm just scanning the web interface. Sep 28 19:11:13 but you should be thinking about how to transition Sep 28 19:11:29 Yeah, everything is working fine, but I'd like to avoid shipping the old system if I can. Sep 28 19:11:40 I am about to ship the old system Sep 28 19:11:41 (Because shipping means supporting . . .) Sep 28 19:11:44 yeah Sep 28 19:12:04 but I have been shipping based on the release branch for almost a year Sep 28 19:14:00 Crofton: thanks for the info. I think I'll wait a few more months (or whatever) for a maintenance branch of some sort to appear. Sep 28 19:50:29 provo Sep 28 19:50:34 ops sorry Sep 28 20:46:49 Has anyone compiled the econ32 camera driver against a custom kernel built with openembedded? (specifically version 2.6.39 with USB mass storage and advanced power management (APM) enabled) ? Sep 28 20:47:25 It seems v4l1-compt.ko modules are missing from 2.6.39, even though they are enabled in the defconfig file.. Sep 28 20:53:26 JaMa|Off: what's holding the -contib jansa/x11 patchset finally? Sep 28 22:06:30 does anyone know what package contains the debug symbols for libstdc++? Sep 28 22:07:15 i didn't see a -dbg in the gcc recipe Sep 28 22:07:37 is this for oe.dev or oe-core Sep 28 22:07:50 for oe-core it should be part of gcc-runtime Sep 28 22:07:54 package Sep 28 22:08:11 2011.03-maintenance branch, so ... ? Sep 28 22:08:21 hmmm Sep 28 22:08:37 did u install -dev package ? Sep 28 22:09:01 i did not. figured that was headers and the like. but you think it may contain debug symbols? Sep 28 22:09:19 not sure but it might have packaged it Sep 28 22:09:26 if there is no -dbg Sep 28 22:09:27 ok let me check the FILES for it Sep 28 22:09:45 the gcc runtime was a bit special in oe.dev Sep 28 22:10:13 "special," huh? :-) Sep 28 22:10:29 yes it got built with cross-gcc Sep 28 22:10:55 and gcc had some stuff to stop stripping stuff like any other normal target package would do Sep 28 22:11:27 in oe.core its separated out so its a bit cleaner Sep 28 22:11:33 although complex Sep 28 22:12:01 looks like the -dev package only contains some libraries and some python files. nothing that appears to be symbols. Sep 28 22:12:18 ok Sep 28 22:12:22 did u try to debug ? Sep 28 22:12:27 afte installing -dev Sep 28 22:12:43 not yet, i'll try now. Sep 28 22:15:51 khem timing? Sep 28 22:22:24 how do I get oe to install the python gpsd bindings? it appears that gpsd provides (and builds) python-pygps, but I cannot install it Sep 28 22:25:11 @flyer-xplorer: Have you added it to your IMAGE_INSTALL and/or pulled it in via RRECOMMENDS in another recipe? Sep 28 22:26:39 kg4ysn: neigher. where should I put IMAGE_INSTALL ? Sep 28 22:27:51 i copied console-image.bb in recipes/images/ and stuck some stuff in there Sep 28 22:28:00 flyer-xplorer: yup, in console-images.bb then Sep 28 22:28:08 IMAGE_INSTALL += "python-pygps" for example Sep 28 22:28:13 after your other includes / requires Sep 28 22:28:35 ok, but will it work if I cannot directly call "bitbake python-pygps"? Sep 28 22:29:22 flyer-xplorer: no, because python-pygps is a subpackage under the gpsd higher level package. Sep 28 22:29:38 flyer-xplorer: you bitbake on gpsd, then opkg install python-pygps on your loaded unit Sep 28 22:29:46 so sticking it in IMAGE_INSTALL will catch it and load it then Sep 28 22:31:20 flyer-xplorer: yes, when it builds the console-image Sep 28 22:31:29 ok cool Sep 28 22:32:00 flyer-xplorer: there are some limitations involving upgrades and "production" systems ... RRECOMMENDS can be a workaround for that, but it doesn't sound like you're at the point where you need to worry about that yet :-) Sep 28 22:33:01 what exactly are those limitations? RRECOMMENDS sounds like you SHOULD install a preferial package, but it is not absolutely necessary Sep 28 22:33:43 flyer-xplorer: let's say you want to upgrade a production system and ADD a new package to the system. Unless that package has an RRECOMMENDS hook from an existing package, opkg update/upgrade won't know to add the package. Sep 28 22:33:52 flyer-xplorer: so the workaround is to IMAGE_INSTALL += "wrapper-recipe" Sep 28 22:34:07 flyer-xplorer: and then in wrapper-recipe.bb, you RRECOMMENDS all of the stuff that you want in your image. Sep 28 22:34:22 flyer-xplorer: then if you change what's in wrapper-recipe, the opkg update/upgrade knows what to do Sep 28 22:34:29 rather than doing IMAGE_INSTALL += all the stuff Sep 28 22:34:51 flyer-xplorer: RRECOMMENDS *does* cause things to be installed ... yeah it's not a recommendation, more like a smack to the head Sep 28 22:35:14 lol *upgrade dummy* Sep 28 22:38:15 khem: took me awhile to load up my system. doesn't look like libstdc++-dev is being built, which is odd. it's definitely defined in gcc-package-cross.inc and gcc-package-target.inc Sep 28 22:38:36 khem: Hi, do you have patch for meta-oe gcc-4.6 to apply after oe-core update? Sep 28 22:42:17 khem: Patch gcc-4.6-linaro-r106741.patch does not apply (enforce with -f) Sep 28 23:52:38 hi. is there anyway to change the boot process to just process the rcS file rather than runlevel scripts? my embedded device is application specific, so there will be no need for dynamic boot scripts. Sep 29 00:08:45 darkopium, don't put anything in the rc.X dir? Sep 29 00:09:38 but when i compile apps, it by defaults put something there Sep 29 00:10:08 would i need a script to do post processing to remove the rc.X files, and redo the rcS before packaging it up? **** ENDING LOGGING AT Thu Sep 29 02:59:56 2011