**** BEGIN LOGGING AT Thu Nov 28 02:59:58 2013 Nov 28 08:55:26 Hi all, Why does rootfs fails when appending dhcp to my image? Nov 28 08:57:06 morning Nov 28 09:09:39 What is causing this error: error: Failed dependencies: netbase < 1:5.0 conflicts with init-ifupdown-1.0-r3.armv7a_vfp_neon Nov 28 09:10:04 I am trying to add init-ifdowm recipe Nov 28 09:10:11 down* Nov 28 09:12:47 rburton: Hi, How do I clean my sysroot manually? Nov 28 09:18:23 Zaif: there's a wipe-sysroot script that will just wipe away the current sysroot. or just delete all of tmp/ Nov 28 09:18:41 as long as your sstate is outside of tmp, you'll b able to rebuild it all quickly Nov 28 09:36:14 rburton: Do I run it as source ./poky/scripts/wipe-sysroot? Nov 28 09:41:40 When I run repo sync, I get uncommited changes error-> error: Cannot remove project "poky/meta-ti": uncommitted changes are present Nov 28 09:53:48 Zaif: no, just run it normally if you've already ran oe-init-build-env Nov 28 09:55:31 morning all Nov 28 09:57:37 rburton: ok. Nov 28 09:57:42 When I run bitbake -c cleansstate I get timeout error with bitbake server? Nov 28 09:59:21 bitbake -ccleansstate [recipename]? Nov 28 10:09:59 Good Day Everyone. I have a question considering replacing udev with mdev. VIRTUAL-RUNTIME_dev_manager = "mdev" leads to an error (see http://permalink.gmane.org/gmane.linux.embedded.yocto.general/12273). However, simply using poky-tiny distro (as suggested on the mailing list by Khem Raj) is not really an option for us as far as I understand. Nov 28 10:10:34 Does anyone have any pointers as how to achieve this? Nov 28 10:14:13 Dennis__: did you enable CONFIG_MDEV in busybox? Nov 28 10:18:08 ndec: I did create my own defconf where I did set CONFIG_MDEV. However, I'm not sure how to make busybox read defconf from my layer instead of reading form meta/. What I tried was adding a busybox_1.20.2.bbappend to my layer with "FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}:" Nov 28 10:19:35 Dennis__: that should do it. do you have the busybox-mdev package created? Nov 28 10:20:03 and added to the image? Nov 28 10:22:36 Ok, I don't know if the package was created. How can I check? There isn't any mdev in my rootfs image. Nov 28 10:24:29 Dennis__: you can check the .ipk (if using ipk)generated in tmp/deploy/ Nov 28 10:24:57 you can check the busybox defconfig too in tmp/work/../busybox to see if yours was taken into account Nov 28 10:26:03 rburton: giving recipe name solved it. I am struggling with init-ifupdown recipe. Unable to add it to my Image. Nov 28 10:26:34 Zaif: did you remove all your local pieces, as it's in oe-core Nov 28 10:26:56 and the oe-core netbase recipe has a conflicts on older versions of ifupdown because a file moved between them Nov 28 10:28:32 rburton: when i run the script I get this error -> Could not determine SSTATE_MANIFESTS/STAGING_DIR/STAMPS_DIR, check above for errors Nov 28 10:29:27 Zaif: presumably you didn't run it after oe-init-build-env, as it ir runs bitbake to determine what to delete Nov 28 10:29:35 Zaif: the alternative is to just delete your tmp/ Nov 28 10:33:51 ndec: I think busybox wasn't built again since I changed the configuration to use my defconf. I did bitbake busybox -c clean and bitbake busybox again and now the package is created. I have to leave now but I will check later if it was also added to my image. Nov 28 10:34:31 rburton: I did run it after oe-init-build-env but still i get this error. If i try to update by repo sync I get these conflicts http://pastebin.com/Lz9r2JiH Nov 28 10:34:35 ndec: Thank you very much for your clarifications. Nov 28 10:35:12 Dennis__: ok. iirc, you might have to add it into your image.. Nov 28 10:35:27 Zaif: never used repo. if you want to do a clean build, just wipe tmp/ Nov 28 10:35:40 rburton: ok. thank you Nov 28 10:45:47 Hi! I'm using yocto on a gumstix, and i'm trying to get a GPIO-Event driver to work. The driver was initially built for open-embedded, and hence it doesn't work out of the box on yocto. I've found a bb recipie for the file on http://wiki.gumstix.org/index.php?title=User_GPIO_Driver, but it didn't work out of the box for yocto. It uses a SVN repository, but doesnt have a COPYING license file in it. Can I bypass that check somehow? Nov 28 10:47:02 erija952_: it has no statement of license anywhere at all? not even in a source file header / README? Nov 28 10:48:25 Header file says GNU GPL2 or BSD, but the copying file is not in the repo Nov 28 10:50:38 there doesn't need to be a COPYING file, just a statement of the license you can point to Nov 28 10:50:48 in LIC_FILES_CHKSUM within the recipe that is Nov 28 10:51:00 http://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#usingpoky-configuring-LIC_FILES_CHKSUM Nov 28 10:51:27 the point is, the correct solution is to set this variable properly rather than trying to evade the check; it's there for your protection Nov 28 10:53:31 Yes, I understand that. I'll try to read up on that link... i'll be back. Thanks for the help! Nov 28 11:27:40 Hi ! I need gfortran to compile a lib from my SDK. I created a gcc_4.7.bbappend which contains FORTRAN += ",fortran" and RUNTIMETARGET += "libgfortran". Unfortunately, gfortran is not generated at all. Do you have any idea ? Nov 28 11:39:24 the following command gives me g++ location and tells me gfortran is missing : find build/tmp/work/i586-poky-linux/gcc-* \( -name "*-gfortran" -o -name "*-gcc" \) -executable -type f -follow Nov 28 13:02:12 Hello, can I overwrite variables that come from .inc file using .bbappend? Or to put it another way: which comes first .inc or .bbappend? Nov 28 13:04:35 kbart: not really qualified myself to answer your question, but you know how ?= works right? .... I guess that's what most .inc files use.... only set variable if not already set. so it should allow you to override. Nov 28 13:07:42 fatal, well, this one (namely qt4.inc) uses "OTHER_PACKAGES = ...", so it's not the case.. What I'm trying to achieve is to remove demos, examples, doc etc. from my installation. Nov 28 13:08:29 and I would like to have my changes in a separate meta layer, so I want to avoid modifying this file directly. Nov 28 13:13:54 kbart: inc; bbappends are processed after what is in the recipe itself Nov 28 13:14:17 (including any includes/requires) Nov 28 13:19:24 bluelightning, thank you for explanation. Now it makes me wonder, why it doesn't work then.. Is is possible to print a complete recipe with variables somehow? After all includes, bbappends etc. Nov 28 13:19:24 bluelightning: if I understand you correctly, its like this: if foo.inc contains FOO="bar", then foo.bb which includes foo.inc can't override FOO, but foo.bbappend can override with FOO="quux" ? Nov 28 13:21:15 kbart: it is indeed: bitbake -e recipename | less Nov 28 13:21:44 kbart: (where recipename is just the name, i.e. without any path or .bb) Nov 28 13:22:16 fatal: no, the recipe can override FOO after the include/require line as well Nov 28 13:28:16 bluelightning, thanks, I know -e command, somehow I'm missing needed info in it.. I'll try another approach. Nov 28 13:34:58 bluelightning: ah, ok... thanks. Nov 28 13:38:00 bluelightning: hmm. what did you mean by 'inc; bbappends are processed after what is in the recipe itself'. Nov 28 13:38:35 i thought require/include and inherit were processed 'in-place' when parsing the recipes Nov 28 13:38:50 ndec: because an inc file is processed at the point the include/require line appears in the recipe, and the bbappends are processed after what is in the recipe, then the bbappend will always be processed after any inc Nov 28 13:38:52 e.g. the exact place where the include/require or inherit statement is does matter. Nov 28 13:39:00 ndec: that's correct, they are Nov 28 13:39:03 ah. ok... Nov 28 13:39:18 i think it was the ';' which was misleading in your reply ;-) Nov 28 13:39:33 i read it as inc *and* bbappends! Nov 28 13:39:33 right, it was a bit cryptically worded :) Nov 28 13:40:21 hehe... that gets quite tricky sometimes with xxx_append vs xxx += ... since one might work while the other one does not... Nov 28 13:44:05 yes Nov 28 13:44:21 the basic rule I try to follow is use += or =+ unless you know you have to use _append/_prepend Nov 28 14:57:25 Hi! Is there a "simple" way to update glib in yocto? I get 2.0 when building dylan, but would need 2.28+ Nov 28 15:05:44 erija952_: take the glib recipe from dora Nov 28 15:06:23 mkey... is that a meta layer or? Nov 28 15:07:04 dora is the latest release Nov 28 15:07:42 right, I just discovered... thanks! Nov 28 15:19:16 Can one do the equivalent of VAR ?= "value" in a python() block? Nov 28 15:24:09 Saur: if not d.getVar('VAR', False): d.setVar('VAR', 'value') Nov 28 15:24:36 bluelightning: Ah, ok. :) Nov 28 18:13:20 is DEPENDS = "ncurses" the correct way if a recipe depends on ... ncurses? Nov 28 18:15:54 LetoThe2nd: yep. Nov 28 18:16:13 ndec: :) Nov 28 18:16:33 that will guarantee that ncruses is 'populated' in the sysroot, before your recipe do_configure is executed. Nov 28 18:16:48 ok sounds like what i want :) Nov 28 18:24:00 mh ok. another one. documentation says for systemd i should do DISTRO_FEATURES_append = "systemd" Nov 28 18:24:50 so i added it to my image, but it gives a ParseError when place after the inherit core-image, and seems to be ignore if placed above :( Nov 28 19:10:52 hm. systemd is in DISTRO_FEATURES, but seems to be neither built nor installed Nov 28 19:21:21 any ideas on how to debug that? Nov 28 19:25:34 LetoThe2nd: did you change the init system? Nov 28 19:26:00 rburton: just as documented: Nov 28 19:26:01 DISTRO_FEATURES_append = " systemd" Nov 28 19:26:01 VIRTUAL-RUNTIME_init_manager = "systemd" Nov 28 19:26:26 works for me, i built an image yesterday. Nov 28 19:26:35 i presume you're using a fairly recent release Nov 28 19:27:00 git head from a few days ago, IIRC. let me pull and rebuild Nov 28 19:27:07 what image did you build? Nov 28 19:27:32 a good test is "bitbake systemd". if systemd isn't actually enabled, it will refuse to build Nov 28 19:27:39 core-image-minimal, extended by the given lines after "inherit core-image" Nov 28 19:28:26 hm, should work. Nov 28 19:28:30 * rburton -> dinner, sorry Nov 28 19:28:39 np, bon appetite. Nov 28 20:29:14 git pull and rebuild resulted in rebuild considerable parts like gcclib and friends, but still no systemd Nov 28 20:35:23 LetoThe2nd: you should set the DISTRO_xx and VIRTUAL-RUNTIME in a .conf file instead, not an image .bb file. Nov 28 20:36:04 typically a .conf if you have one, otherwise local.conf Nov 28 20:37:21 well, it's not 'you should', but 'you have to' in fact... Nov 28 20:38:01 ndec: ah yes?!? Nov 28 20:38:05 if you have it in 1 .bb file only, these variables are set only when executing a task from that .bb file. however such settings are much wider than the image itself. Nov 28 20:38:29 many recipes would do things like "if ${@base_contains('DISTRO_FEATURES','systemd','true','false',d)}; then" Nov 28 20:38:49 and that would be 'false' if you set it in the image .bb file. Nov 28 20:39:10 aaaaah... i see. Nov 28 20:39:37 you should set DISTRO_FEATURES in a .conf file always, and IMAGE_FEATURES in an image .bb file. Nov 28 20:40:26 i think 'bitbake systemd' should tell you that systemd is not in DISTRO_FEATURES in your current situation Nov 28 20:43:00 ndec: fun fact: it says ecatly that, but when doing -e on the image it is set. thats why assumend it works via dependencies. Nov 28 20:43:09 hehe ;-) Nov 28 20:44:44 well, it's important to remember that all .conf files are 'included' when parsing 1 .bb file. but whatever you do in a .bb file cannot impact another .bb file since they are parsed 'independently' from each other... Nov 28 20:45:20 ok... Nov 28 20:46:11 so much to learn/know when you're new to OE Nov 28 20:47:00 you can replace OE by anything in fact ! Nov 28 20:48:21 hm, sometimes yes - sometimes no Nov 28 20:48:40 but yeah, most buildsystems comewith a certain amount of complexity Nov 28 21:00:11 yay... more and more rebuilding Nov 28 21:00:47 LetoThe2nd: what's being rebuilt is what depends on systemd, now that's enabled ;-) Nov 28 21:01:23 ndec: like, gcc-cross. obviously. :P Nov 28 21:01:58 oh, DISTRO_FEATURES has changed... Nov 28 21:02:06 that's generally costly... Nov 28 21:02:34 yeah, that was not really meant to be taken seriously. ;) Nov 28 21:03:37 LetoThe2nd: probably the good excuse to call it a day for today... it's probably as late for you as it is for me ;-) Nov 28 21:04:43 ndec: probably. but today is private tinkering day, so who cares ;) Nov 28 21:27:11 LetoThe2nd: i take it you're trying to pull systemd into your image? Nov 28 21:27:22 * nerdboy specifically kept it out Nov 28 21:27:38 nerdboy: yeah, trying to at the moment Nov 28 21:29:16 you should be able to just add systemd to DISTRO_FEATURES Nov 28 21:30:00 but if nothing really depends on it, i'm not sure how the rootfs would turn out... Nov 28 21:30:10 what kind of image recipe? Nov 28 21:31:11 core-image-minimal... yeah its just rebuilding Nov 28 21:31:28 but yeah, whenever i pull on master (every few weeks or so) and rebuild it usually has to rebuild a lot of stuff... Nov 28 21:32:06 so does your image recipe include core-image-minimal? Nov 28 21:32:57 ndec already pointed it out. the problem was that i did my changes to a new image recipe, instead of doing a new distro config. the latter seems to work now. Nov 28 21:33:09 core-image-minimal isn't fooling, it's *very* minimal... Nov 28 21:33:22 what's your target? Nov 28 21:34:04 yeah, i know what minimal means. Nov 28 21:34:17 it now specific target other that knowledge gathering ATM. Nov 28 21:34:34 well, i didn't realize how minimal until i based an image on it Nov 28 21:34:51 :) Nov 28 21:34:59 then i switched it to core-image instead Nov 28 21:35:20 i'm pretty much a requirements guy, so i need to ask Nov 28 21:35:50 why do you need systemd? is it just "knowledge gathering", or ? Nov 28 21:36:03 hehe yeah. really nothing more to it at the moment. Nov 28 21:36:21 why would someone not need systemd ;-) Nov 28 21:37:01 * nerdboy is still a little peeved at that one in current upstream Nov 28 21:37:16 switched all my desktops to xfce Nov 28 21:37:43 that doesn't prevent from using systemd. Nov 28 21:43:24 * LetoThe2nd is totally not interested in pro/contra anything discussions anyways. todays lesson learned is: set DISTRO_FEATURES in you distro.conf ;) Nov 28 21:55:42 i'm not saying anything either way, it's just the loss of choice Nov 28 21:56:00 at least for gnome, it's now a hard dep Nov 28 21:56:23 i have my reasons for crafting my rpi image that way Nov 28 21:56:58 and it's still a choice outside of current gnome3 afaik Nov 28 22:02:14 ndec: anyways - that did the trick. thanks, and now it's really a day ;) Nov 28 23:21:57 So I got to finally build core-image-minimal + linux-yocto-tiny using llvm+clang for almost all the packages (except kernel, elfutils and a few other packages). Nov 28 23:22:13 But when I boot the kernel, I basically get this error message: Nov 28 23:22:16 http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=fe77047364eac502904a33bcbc35a704be9acc9e Nov 28 23:22:22 But with exitcode 0. Nov 28 23:23:52 Any ideas? Here's the logs: http://sprunge.us/SLiH Nov 28 23:30:18 kergoth: ^ ? :) **** ENDING LOGGING AT Fri Nov 29 02:59:58 2013