**** BEGIN LOGGING AT Tue Sep 16 02:59:58 2014 Sep 16 07:20:21 good morning Sep 16 07:21:43 good morning to you too Sep 16 07:26:24 Hi,I'm trying to enable LVDS on wandboard solo .I'm using yocto with kernel 3.10.17. I've followed http://pastebin.com/bDvMiehZ to make the device tree for ldb. but the display seems to be not working can anyone please help Sep 16 08:18:38 hi. how to start with yocto? can i build an image for the raspberry pi for example? Sep 16 08:21:00 qknight: refer this: http://git.yoctoproject.org/cgit/cgit.cgi/meta-raspberrypi/tree/README Sep 16 08:21:44 qknight: also this documentation: http://www.yoctoproject.org/docs/current/yocto-project-qs/yocto-project-qs.html Sep 16 08:23:39 ... especially the latter one, actually. Sep 16 08:42:38 maxin, LetoThe2nd thanks! Sep 16 08:53:24 Hi all, updated Yocto Daisy from repository and now grub loads before showing "error: /initrd not found" After continuing I'm finding that the kernel panics because of an invalid user config and can't find a valid rootfs! Does anyone know where about in the system I can set it to not load initrd? Sep 16 08:54:05 morning all Sep 16 09:17:18 Hi,I'm trying to enable LVDS on wandboard solo .I'm using yocto with kernel 3.10.17. I've followed http://pastebin.com/bDvMiehZ to make the device tree for ldb. but the display seems to be not working can anyone please see the steps and let me know whats wrong? Sep 16 10:13:01 Hi, I'm wanting to get the latest chromium into my yocto image for ARM. I'd like to get an idea of effort I'd have to put in to do this - any ideas? Sep 16 10:17:06 karooga: add meta-browser, bitbake chromium Sep 16 10:17:19 (then wait all day for it to compile) Sep 16 10:19:25 rburton: I was using hob but didn't see chromium in the packages menu? Sep 16 10:21:27 you need to download meta-browser, and add that layer Sep 16 10:23:40 rburton: Sorry I'm still a bit of noob with yocto. How do I download meta-browser? Sep 16 10:25:26 not sure how to do it with hob. you'd use git to clone it from https://github.com/OSSystems/meta-browser Sep 16 10:30:10 rburton: It's doesn't have to be with hob. It was just the only way I could see from the documentation how to find out the package list. Sep 16 10:30:46 rburton: Where in the tree should I clone it into? Just ./build ? Sep 16 10:30:52 find . -name *.bb will list all recipes :) Sep 16 10:31:07 no, put it next to poky or whatever you cloned in the first place Sep 16 10:31:14 then add the path to bblayer.conf Sep 16 10:31:27 bblayers.conf, that is Sep 16 10:31:45 Hob does have a way to add a layer in as well once it's cloned somewhere on your machine Sep 16 10:31:55 (alternatively) Sep 16 10:39:55 Does it matter which image I run bitbake on? Sep 16 10:40:46 I'm trying to include mplayer2 in an fsl-image-x11 target for freescale imx6 hardware. It needs libav_9 with I therefore built with putting PREFERRED_VERSION_libav = "9.10" in the local.conf. I can build mplayer2, but when it comes to creating the image it throws an error in do_rootfs: ls: cannot access /home/fs/fsl-3.10.31-alpha-bsp/x11/tmp/sysroots/mx6sexceet/pkgdata/runtime-reverse/libavcodec53: No such file or directory. Any ideas? Sep 16 10:43:50 bluelightning: So if I have bblayers.conf set up with chromium layer then bitbake -i core-image-minimal will handle all the dependencies even if minimal doesn't have all the xserver gui stuff? Sep 16 10:45:25 karooga: no, there's an additional step required to add chromium to your image, and even then it wouldn't start up X on boot unless you also add that in Sep 16 10:48:23 frsc: I guess that's a bug Sep 16 10:48:47 frsc: did you perhaps build a different version of libav first? Sep 16 10:49:58 bluelightning: How do I add chromium to the image, once i've added the layer? Sep 16 10:50:38 bluelightning: Should i be using a specific target image? or is core-image-minimal sufficient? Sep 16 10:50:46 karooga: http://www.yoctoproject.org/docs/current/dev-manual/dev-manual.html#usingpoky-extend-customimage Sep 16 10:51:05 karooga: the simplest thing would be to start with an image that already contains X Sep 16 10:54:54 bluelightning: yes, I built libav_0.8.9 before. Sep 16 10:56:11 karooga: perhaps core-image-x11 Sep 16 10:56:26 bluelightning: Sweet, ok I'll do that. Think if I'm going to go through the process of building X then I may as well get my target platform (rpi) correct. :-) Sep 16 10:56:40 frsc: ok, so as a workaround I would suggest commenting out the PREFERRED_VERSION_libav line; then bitbake -c clean libav; then uncomment it again and continue Sep 16 10:56:55 bluelightning: I don't see a core-image-x11? Sep 16 10:57:17 karooga: which version of the build system are you using? Sep 16 10:57:53 daisy for x86 linux. Sep 16 10:58:14 64bit i mean Sep 16 11:00:02 Hi,I'm trying to enable LVDS on wandboard solo .I'm using yocto with kernel 3.10.17. I've followed http://pastebin.com/bDvMiehZ to make the device tree for ldb. but the display seems to be not working can anyone please see the steps and let me know whats wrong? Sep 16 11:01:17 bluelightning: Thanks! I'll have a try! Sep 16 11:04:18 karooga: hmm, that's strange then - core-image-x11 should be there... Sep 16 11:05:10 Nilesh_: if you don't get an answer here I'd suggest contacting the maintainer of the BSP you are using (or the mailing list specific to that BSP, if there is one) Sep 16 11:05:13 bluelightning: I did pulled it off git two days ago so I don't think it's outdated. Seems git HEAD was all i needed for daisy. Sep 16 11:05:30 right, and I'm pretty sure core-image-x11 has existed for a while Sep 16 11:06:06 bluelightning:yep sure...thanks :-) Sep 16 11:13:54 bluelightning: I tried cleaning libav as you said, but still get the same error in do_rootfs Sep 16 11:16:46 frsc: hmm, ok Sep 16 11:16:55 frsc: would you mind filing a bug at https://bugzilla.yoctoproject.org/ ? Sep 16 11:20:37 bluelightning: I will try one more thing and if it doesn't work I will file a bug. Sep 16 12:10:02 bluelightning: bug filed at: https://bugzilla.yoctoproject.org/show_bug.cgi?id=6729 Sep 16 12:10:03 Bug 6729: normal, Undecided, ---, cristian.iorga, NEW , Integrating mplayer2 in fsl-image-x11 with libav_9 Sep 16 12:27:45 * mcfrisk is wondering how many end users and developers yocto has, compared to say Debian & Ubuntu? Sep 16 12:32:50 mcfrisk: it's hard to get an exact count - lots of folks don't talk to us about what they're doing with our tools (and that's perfectly fine) Sep 16 12:33:37 I think it's fair to say it's pretty widely used in the embedded Linux field by now though Sep 16 12:33:37 I'm getting an error saying not recipes for "/gstreamer1.0-plugins-bad_1.4.0.bbappend" however I've included "./meta-openembedded/meta-multimedia" and "meta-raspberrypi" layers so this doesn't make sense. Sep 16 12:34:35 karooga: if I had to guess I'd say you're using the master branch of one or more of those layers when you should probably be using daisy (assuming you're using the daisy release of poky) Sep 16 12:34:47 bluelightning: in Debian each package has a named maintainer, then release and security teams etc, I guess such don't exists in yocto Sep 16 12:35:34 and then enduser based testing also doesn't really happen in yocto, and test automation is I guess pretty basic as well. Sep 16 12:36:07 mcfrisk: we have a maintainer for individual recipes in the core (see meta-yocto/conf/distro/include/maintainers.inc), and each layer outside of that has a maintainer (see http://layers.openembedded.org ) Sep 16 12:36:59 we do have a dedicated QA team, and a number of organisations involved in the project also have their own QA teams that do regular testing Sep 16 12:37:38 security patch support too? Sep 16 12:37:49 test automation is only a couple of releases old to be fair but we do have it Sep 16 12:38:06 security patches are backported, but we do not have a dedicated security team, no Sep 16 12:38:25 bluelightning: yip, you're correct, but the rpi layer doesn't seem to be in the daisy release. Sep 16 12:39:10 in docs I see qemu etc testing, are there details what tests are run and on which hw archs? Like running ltp tests on all supported platforms.. Sep 16 12:40:01 karooga: http://git.yoctoproject.org/cgit/cgit.cgi/meta-raspberrypi/ has a daisy branch... Sep 16 12:40:46 mcfrisk: in terms of what the base project tests, we test on the hardware reference platforms for the project - there is one per architecture Sep 16 12:42:17 http://git.yoctoproject.org/cgit.cgi/poky/plain/README.hardware Sep 16 12:42:28 regular QA reports are sent out to our mailing list Sep 16 12:42:32 I'm wondering where line-in-sand when to use Debian/Ubuntu/Fedora as base and add apps and possibly tune some parts like kernel, or when to use yocto Sep 16 12:44:23 well, it depends on what you are doing Sep 16 12:44:53 but reproducibility and upgrades while preserving your customisations can be tricky with an off-the-shelf distro Sep 16 12:45:47 well, I've seen them be hard with yocto too since people are touching everything without thinking of maintainability... Sep 16 12:46:43 can you be more specific? Sep 16 12:48:02 well, not really :) just inside some companies updating between yocto releases has shown how much changes are all over Sep 16 12:48:08 generally, we do try to provide the means to keep your customisations separate - but of course if you just go ahead and modify recipes directly instead of using a custom layer with bbappends, then yes you will end up with something that is difficult to maintain Sep 16 12:48:17 exactly Sep 16 12:48:36 right, but there's not a lot we can do about that other than document best practices, which I think we do Sep 16 12:48:48 true Sep 16 12:49:03 we're open to suggestions on how to improve, naturally Sep 16 12:49:48 are there some details about the current testin setup? Sep 16 12:51:45 ok, so firstly here is one of our full-pass QA reports: https://wiki.yoctoproject.org/wiki/WW36_-_2014-09-04_-_Full_Pass_-_1.7_M3.rc1 Sep 16 12:51:52 here's our test plan: https://wiki.yoctoproject.org/wiki/Yocto_1.7_Overall_Test_Plan Sep 16 12:52:16 ah, google just found the old 1.4 test plan Sep 16 13:01:47 bluelightning: thanks for your help. I'm going to have to go through the docs and get a better understanding of how everything fits together. Cheers Sep 16 13:01:55 karooga: no problem Sep 16 13:03:56 hmm, test run logs are open only for registered users? Or some specific user group? https://bugzilla.yoctoproject.org/tr_show_run.cgi?run_id=2832 Sep 16 13:04:28 quit Sep 16 13:12:41 mcfrisk: I think you probably just need an account which you can add yourself; that's probably a default quirk of testopia rather than a conscious decision we've made though Sep 16 13:16:49 yep, added an account and access works. Sep 16 13:17:11 So the tests are running simple commands on target via serial/ssh like rpm -qa or 'gcc hello.c' Sep 16 13:18:08 and no existing test suites like ltp Sep 16 13:20:28 we do have ltp, Wind River runs it regularly and publishes the results: https://wiki.yoctoproject.org/wiki/LTP_result Sep 16 13:21:06 cool Sep 16 13:21:46 btw, so I count yocto has 20 developers, or at least package maintainers: git grep MAINTAINER ./meta-yocto/conf/distro/include/maintainers.inc | egrep -o '".*"'| sort | uniq -c| wc -l Sep 16 13:22:28 and that some developers are quite busy maintainin over 190 packages in yocto, or then support is really minimal :) Sep 16 13:22:33 we do get a lot of contributions from the wider community as well, so that's not really representative Sep 16 13:25:37 true, and there are other meta and bsp layers with different maintainers too. Sep 16 14:39:39 how does svn module expect to find a tgz? downloading faild behind a proxy so I have a tar ball from svn co and I need to feed it somehow to bitbake. Sep 16 14:47:19 why kmod builds are not being stripped even if there is no ptest in use? Sep 16 14:59:46 is there a YPTM this week.. Sep 16 15:02:43 to query if current package being built has FEATURE in PACKAGECONFIG, is bb.utils.contains("PACKAGECONFIG", "FEATURE", "XXXX", "YYYY", d)? Sep 16 15:04:24 k-s: correct yes Sep 16 15:04:30 * zeddii suspects that fray got his answer by the silence Sep 16 15:04:52 I think we had one last week, which means no meeting this week Sep 16 15:22:37 hah yup Sep 16 15:22:51 I didn't make last weeks meeting, so I couldn't remember if we had one or not.. :P Sep 16 15:26:28 ERROR: Index creation command '/home/tiago/yocto/poky/build/tmp/sysroots/x86_64-linux/usr/bin/createrepo --update -q /home/tiago/yocto/poky/build/tmp/deploy/rpm/all' failed with return code 1: Sep 16 15:26:32 error: db_init:/home/tiago/yocto/poky/build/tmp/work/x86_64-linux/rpm-native/5.4.9-r63/rpm-5.4.9/rpmdb/db3.c:1087: dbenv->open(2): No such file or directory Sep 16 15:26:35 error: cannot open Packages(0) index: No such file or directory(2) DB: Berkeley DB 5.3.21: (May 11, 2012) Sep 16 15:26:38 error: cannot open Packages database in /home/tiago/yocto/poky/build/tmp/sysroots/x86_64-linux/var/lib/rpm Sep 16 15:26:41 sorry the flood Sep 16 15:27:06 but does anyone seen this type of error? Sep 16 15:28:41 no, but that is definitely a bug, it should never be trying to open an rpm db in that location Sep 16 15:28:58 vignatti: which version of the build system are you using? Sep 16 15:29:36 bluelightning: hmm.. build system/ Sep 16 15:29:50 I'm running daisy just Sep 16 15:31:10 ok, could you please file a bug at http://bugzilla.yoctoproject.org/ ? Sep 16 15:31:29 bluelightning: well, sure Sep 16 15:31:48 bluelightning: but we are building our own and custom build image Sep 16 15:32:22 tmpsantos could get that image working in his environment, but I couldn't Sep 16 15:32:40 unless you've modified createrepo, rpm or the code that calls them that should not matter Sep 16 15:42:06 bluelightning: previously I was building a different image for a different machine type and now I change this but I didn't manually clean the sstate caches and etc from build/ Sep 16 15:42:23 do you think this could interfere in my most recent build? Sep 16 15:46:25 vignatti: that should not cause an error like this, no Sep 16 15:48:05 bluelightning: but then there was another difference also: before I was using Ubuntu 12.04 and now I'm on 14.04 Sep 16 15:48:29 and as I said, I didn't clean my cache whatsoever Sep 16 15:49:35 so, we do keep the native parts separate per host distro, so again that should not be a problem Sep 16 15:50:22 fair enough Sep 16 17:00:58 how to say "mount" is provided by busybox and util-linux-mount so when using systemd I can depend on "mount" and get one of those (both are being installed now, since systemd DEPENDS on util-linux-mount directly?) Sep 16 17:02:40 Hi folks - is it possible to craft a bbappend such that it will only apply to the machine recipe, but not the native/nativesdk recipe? (the recipe in question has bbclassextend="nativesdk") Sep 16 17:08:53 k-s: I seem to recall that systemd expects functionality that busybox's mount doesn't provide, but I may be wrong Sep 16 17:09:20 wotte: depends, what does the bbappend do? Sep 16 17:09:58 bluelightning: I want to test that :-) Sep 16 17:10:24 bluelightning: that's why to choose one or another, so if user wants just the basic he could use busybox Sep 16 17:10:32 k-s: then you'd just remove the dependency altogether, since busybox (and hence the mount symlink) will always be there by default Sep 16 17:10:34 bluelightning: how to do it in .bb? Sep 16 17:10:40 ok Sep 16 17:14:56 bluelightning: The recipe in question is the linux-libc-headers. (Yes, yes. I read the warning. Our kernel (and eglibc) is extensively modified and the headers need to be present in the SDK for users not leveraging Yocto). I've got a bbappend that pulls headers from our repository, which works swimmingly so long as you're using a 3.x series kernel as the SDK host. If you take the SDK and try to use it on an older distro (CentOS 6.5 and your 2.x Sep 16 17:14:56 kernel, I'm looking at you) the tunctl provided by the SDK craps out because the "kernel is too old". My hypothesis at this point is the kernel headers used to build nativesdk packages is to blame. Sep 16 17:18:09 wotte: ok, well I was thinking that for whatever variables you're setting in the bbappend, you'd just use _class-target overrides so that they only apply for the target Sep 16 17:18:49 Aha. I'll give that a shot. Thanks! Sep 16 17:18:50 I'm wondering if whatever customisations you're applying are really the source of the issue here or not Sep 16 17:19:16 I suspect that they won't be, but I guess you can only try... Sep 16 17:19:20 I'm really not sure either. At this point, I'm grasping at straws. Sep 16 17:20:31 I've been banging my head against this on and off for the last couple months, and an end user that can only use RHEL 6.4 will be ready to use a release soon. Sep 16 17:21:35 FWIW I build regularly on a CentOS machine, but I think I may not have ever tried running a tunctl binary from the SDK on it Sep 16 17:23:07 I tried a vanilla SDK build on the CentOS machine and "tunctl -h" didn't freak out, so I suspect its something we're doing Sep 16 17:23:10 I'm just not sure what Sep 16 17:32:16 was versioned DEPENDS possible in yocto? trying have both gtest 1.6.0 and 1.7.0 in build dependencies.. Sep 16 17:32:52 no Sep 16 17:33:12 sigh Sep 16 17:33:17 if you want both to be built you will need to namespace them, so that the version actually becomes part of PN Sep 16 17:33:34 yep Sep 16 17:33:52 what's the standard for that? Sep 16 17:34:00 for adding version number to the package name? Sep 16 17:34:58 it's not very common so I wouldn't say there is a standard, but for clutter we have done e.g. clutter-0.8 / clutter-1.0 Sep 16 17:35:16 heading out, bbl Sep 16 17:36:53 Random other question: Is it possible to have devshell launch not in pseudo/fakeroot? Emacs and other stuff freaks out, so I've been doing echo $PATH > .path; /bin/su -p wotte; export PATH=`cat .path`, but I'd like to avoid that step. Sep 16 17:38:16 wotte: export PSEUDO_UNLOAD=1 or so Sep 16 17:38:19 afaik Sep 16 17:38:37 kergoth: Thanks, I'll give it a shot. Sep 16 17:49:37 Can someone help me with the Yocto ADT eclipse plugin? Sep 16 18:04:43 Hey all, I have what I'm sure is a basic yocto issue. Bitbake is saying BBPATH and BBFILES aren't configured properly when building a stock freescale community image. Here's the error with the environment variables printed at the end: http://pastebin.com/qPiJFBZU Sep 16 18:07:37 So far as I can tell those are set properly... Sep 16 18:40:25 Is there a nice way to use buildbot (yocto-autobuilder or otherwise) to rebuild an ${AUTOREV}ed package if the package's source is updated? Sep 16 19:02:41 Resolved my issue from earlier, though now I have another question: is there an easy way to get Bitbake to use git or http wget rather than FTP when fetching files for a freescale community image? Sep 16 19:28:36 is there any clean way to uninstall shadow/shadow-securetty after image is ready? I don't ship package tools (rpm,opkg) with my image, so I don't need useradd et al in the final binary Sep 16 19:30:20 k-s: perhaps by not deploying them in the first place? Sep 16 19:30:46 fishey1: things that "inherit useradd" get it, for instance systemd and dbus to set their required users Sep 16 19:32:39 k-s: and they need ones that run on the target system for some reason? (forgive my ignorance, haven't been using yocto too long) Sep 16 19:33:17 no, that's why I want to remove it Sep 16 19:33:19 :-) Sep 16 19:33:55 fishey1: currently the useradd.bbclass generates post_inst/rm for you with useradd commands Sep 16 19:34:18 fishey1: but in my case, there is no package management in the final image, so I don't need these tools Sep 16 19:36:11 fishey1: actually for systemd if we use their sysusers.d one could avoid useradd and do that, see http://www.freedesktop.org/software/systemd/man/sysusers.d.html but I'm still far from this point Sep 16 20:10:05 k-s: try appending shadow etc to ROOTFS_BOOTSTRAP_INSTALL Sep 16 20:10:28 ok Sep 16 20:10:37 hang on, no, not that variable Sep 16 20:11:05 use ROOTFS_PKGMANAGE_BOOTSTRAP_append = " shadow shadow-securetty" Sep 16 20:11:15 that's a gross hack but might just work :) Sep 16 20:11:40 (that variable is the list of packages that are essential to bootstrap package management, and will be removed if you don't want a package manager on the target Sep 16 20:23:02 k-s: ah, I see. and it pulls them in so that it can remove the packages later >.> . rburton: doesn't sound like too much of a gross hack :) Sep 16 20:29:06 rburton: who would be a good guy to talk about the systemd changes I'm doing? Sep 16 20:29:23 rburton: particularly we need to discuss using PACKAGECONFIG x split packages Sep 16 20:29:31 k-s: probably rburton, by coincidence ;) Sep 16 20:29:37 great :-) Sep 16 20:30:01 split packages are more flexible, but the cost of making it right is too high Sep 16 20:30:12 and I don't see the flexibility being used Sep 16 20:30:29 ie: install only systemd without tmpfiles, then install tmpfiles Sep 16 20:31:10 OTOH if we just follow with PACKAGECONFIG, we leave it to systemd to fix in its own build, so *.service Wants/Requires= are right Sep 16 20:32:14 all in all I see people using systemd fully in non minimal builds, it adds a great deal of features in not that big space Sep 16 20:32:39 but if you compare with busybox, it shows as too big, but it is not that much Sep 16 20:50:38 Hello yocto folks! If I add IMAGE_INSTALL += "kernel-dev", will the kernel headers show up on my target image? If so, how do I remove them and have them only show up in the SDK? Sep 16 20:51:07 er, I should specify... into my image recipe Sep 16 20:52:26 Right now I don't have kernel-dev listed, and when I run bitbake -c populate_sdk image kernel headers don't show up. Sep 16 21:17:53 rburton: comments about the systemd as per above? Sep 16 21:22:49 Hey yoo .. dora, util-linux -- setterm ? Sep 16 21:23:00 Can't seem to see why it does not appear to be getting built. Sep 16 21:27:06 Dangit... adding IMAGE_FEATURES += "dev-pkgs" doesn't help... still no kernel headers in the SDK Sep 16 21:28:25 Unless I need to re-build the image first, then do "-c populate_sdk" Sep 16 21:29:20 that appeared to be a no-op... bitbake returned too quick. Sep 16 21:30:22 and..... nvm setterm. It's just not split off. Sep 16 21:39:50 k-s: probably best to send a mail to the list about this **** ENDING LOGGING AT Wed Sep 17 02:59:58 2014