**** BEGIN LOGGING AT Thu Nov 03 03:00:00 2016 Nov 03 03:05:16 that isn't what i was thinking... Nov 03 06:01:36 I'm trying to get yocto to run from a hard disk. It's not finishing the boot process. Research suggests that there is some problem around the rootfs. Nov 03 06:01:56 Here is a screenshot of where the boot is stuck Nov 03 06:02:00 http://imgur.com/a/pf9Cy Nov 03 06:02:52 grub.cfg looks like this: Nov 03 06:02:55 serial --speed=115200 --word=8 --parity=no --stop=1 terminal_input --append serial terminal_output --append serial set timeout=1 GRUB_TIMEOUT=1 menuentry 'yocto' { linux /boot/yocto_bzImage root=/dev/xvda1 rw console=ttyS0,115200 } Nov 03 06:03:36 I'm not sure what I can do to take a next step in resolving this. Does anyone have any suggestions? thanks. Nov 03 06:39:29 hi ... if I trigger a bitbake build of our image I always get a Note "There are 2 recipes to be removed from sysroot" anyone know what this means? Nov 03 08:39:45 khem: so does weston actually use gbm on bcm? Nov 03 08:50:25 rburton: sry but I think I need to come back to the static lib package topic ... I changed now to have a SDK recipe and break the dep via RDEPENDS_${PN}-dev = "" ... then I thought its maybe smarter to do RDEPENDS_${PN}-dev_remove = "${PN}" but this has no effect? Nov 03 09:00:06 nerdboy: thx for your answer 12h ago (i was afk) Nov 03 09:04:16 rburton: it asks for gbm Nov 03 09:18:28 rburton: hey Nov 03 09:18:58 rburton: i'm curious, what tool do you use for combining layers? Nov 03 09:19:12 rburton: i-e different layers from different git repos Nov 03 09:19:19 git submodule? Nov 03 09:31:35 zeenix: poky is built with combo-layer but there is no single widely accepted solution here (at least git submodules and repo are used as well) Nov 03 09:32:08 jku: ah ok Nov 03 09:32:21 jku: we use repo here and i'm starting to hate it Nov 03 09:32:28 hence the question Nov 03 09:34:02 we once developed https://github.com/carosio/redomat which has a declaration format similar to repo, and which is some docker based tooling for building a completely specified "layer-combo" Nov 03 09:46:54 th: cool Nov 03 09:46:59 th: how does it compare to repo? Nov 03 09:49:27 zeenix: we're using submodules, and we hate it as well. Nov 03 09:51:42 zeenix: well i do not know repo very well. but redomat is what we use for building our releases. reproducability and having a full declaration was the top priority, which we implemented by it. Nov 03 09:52:33 does anybody here ever used a virtual keyboard on DirectFB ? Nov 03 09:52:43 zeenix: effectively you create a declaration and just call `redomat.py my_declaraion.xml` in the end you will have a docker image of a success or failure. Nov 03 10:03:48 th: sounds good Nov 03 10:37:39 zeenix: i hear theres a fork of repo coming out soonish which may be better. basically there's no perfect solution... Nov 03 10:38:43 rburton: ah ok Nov 03 10:52:26 i have a strange behaviour (or at least, not the one i'm waiting) with the following recipe: http://pastebin.com/1aB7x0kd Nov 03 10:52:34 in my understanding Nov 03 10:52:46 this recipe is supposed to create two packages: base-users and base-users-dev Nov 03 10:53:07 actually, it only create base-users-dev Nov 03 10:53:18 base-users is missing, and my image generation failed because of it Nov 03 10:53:23 because by default packages with no content don't get generated Nov 03 10:53:37 ALLOW_EMPTY_${PN}="1" Nov 03 10:53:40 hmmm Nov 03 10:53:48 but, is "base-users" supposed to be empty ? Nov 03 10:53:57 you didnt put any files in it Nov 03 10:53:58 so yes Nov 03 10:54:03 hm Nov 03 10:54:10 but ... i didn't in base-users-dev too Nov 03 10:54:14 and it's generated Nov 03 10:54:15 :) Nov 03 10:54:18 i've given you the hints you need ;) Nov 03 10:54:26 ok thanks :) Nov 03 10:54:33 grep bitbake.conf for ALLOW_EMPTY :) Nov 03 10:54:42 i noticed an other strange behaviour Nov 03 10:55:39 if, for base-users-dev, i use group == 1000 for my user adm and then declare GROUPADD_PARAMS_${PN}-dev = "-g 1000 adm" Nov 03 10:55:47 it fails with "group doesn't exist" Nov 03 10:56:36 which is strange since it correctly execute the GROUPADD_PARAMS_${PN} command set before USERADD_PARAMS_${PN} Nov 03 10:56:48 but looks like ${PN}-dev break this behaviour Nov 03 10:56:55 gonna dig a bit more btw, thx rburton Nov 03 11:33:05 hello guys Nov 03 11:33:57 Ox4: Hello Nov 03 11:35:30 Ox4: hi Nov 03 11:52:13 hi Nov 03 11:52:44 I think that there is a bug in conf/distro/include/yocto-uninative.inc Nov 03 11:52:59 UNINATIVE_URL = "http://downloads.yoctoproject.org/releases/uninative/1.0.1/" Nov 03 11:53:19 Shouldn't this be a conditional asigment? Nov 03 11:53:56 it is not possible to overwrite it in e.g., local.conf to point to a local download mirror Nov 03 11:54:29 which is an issue if you want to build without intenet connnection Nov 03 12:00:04 can i use yocto to generate a toolchain/sdk which is hosted on windows? if yes, how do i do this? Nov 03 12:01:57 melonipoika: in master and latest release it is overridable Nov 03 12:01:59 oh... meta-fsl-arm is missing a morty branch :/ Nov 03 12:02:04 otavio, ^^^ Nov 03 12:03:01 fmeerkoetter: yesterday there was a discussion on meta-mingw which should give one aV_V played around with it Nov 03 12:03:34 oh thank you jku. I am using krogoth and forgot to check master :-( Nov 03 12:03:44 mdnneo: thanks. "with there was" you mean here on IRC? so i can check the logs? Nov 03 12:04:03 I'm a noob to yocto. I cannot see where to get class files from. I get "ERROR: ParseError in configuration INHERITs: Could not inherit file classes/boot-directdisk.bbclass" Nov 03 12:04:50 fmeerkoetter: I tried but got a build error Nov 03 12:05:21 now I'm with other issue, but the thing is: Nov 03 12:05:39 floob: recipe directories have a "classes" folder Nov 03 12:05:40 1) download and add the layer meta-mingw Nov 03 12:06:48 2) set on local.conf: SDKMACHINE = "check at conf/machine-sdk/" Nov 03 12:07:11 aV_V, mdnneo: thanks for the hint on the general direction. didn't knew about meta-mingw Nov 03 12:07:37 floob: http://lists.openembedded.org/pipermail/openembedded-core/2016-March/119663.html - boot-directdisk.bbclass: merge it into image-vm.bbclass Nov 03 12:07:50 JoiF: find /mnt/poky | grep boot-directdisk.bbclass returns no files, which seems to indicate there is no such file in the poky hierarchy. Do I need to download the classes from somewhere? Nov 03 12:07:54 fmeerkoetter: u welcome, kergoth helped me with this Nov 03 12:11:04 thanks joif Nov 03 12:13:41 floob: np Nov 03 12:14:40 When I boot, the boot process does not complete - this image shows where it gets up to http://imgur.com/a/OucrU - any guesses as to what might be wrong? I built x86-64 core-image-minimal-initramfs Nov 03 12:20:18 fmeerkoetter, aV_V: let me know if you have success basically I'm also interested in just never had the time to look ;) Nov 03 12:21:53 mdnneo: ok, right now I'm trying to cross-compile from linux Nov 03 12:22:20 floob: Seems like your kernel booted just fine. Nov 03 12:24:17 Joif: oh. I guess I would have expected it to give some sort of message to say it had finished. It sort of looks like it has just stopped. Nov 03 12:27:42 floob: Does your image contain a rootsf? Nov 03 12:27:50 or even a rootfs? Nov 03 12:28:27 label yocto kernel /boot/yocto_bzImage initrd /boot/yocto_initramfs_rootfs.cpio.gz append root=/dev/ram0 rw console=tty0 console=ttyS0,115200 Nov 03 12:28:58 joif: I believe so, yes. The network is not responding to me. Nov 03 12:29:47 I have dropbear sshd and nginx installed and neither respond. Nov 03 12:30:06 floob: Well, my comment about your kernel booting fine was a bit tongue-in-cheek comment .. don't expect networking if you have ONLY the kernel running. ;) Nov 03 12:30:43 ha! Nov 03 12:40:34 floob: I actually haven't done this on Yocto myself. Nov 03 12:42:24 I haven't Yoctoed this myself .... Nov 03 12:44:22 floob: But you need to configure the kernel correctly, which I assume happens for you when you're using the meta-initramfs recipe, and configure where to locate the initramfs image. Nov 03 12:44:40 floob: Also, I don't know if bitbake takes care of the dual-build process for you? Nov 03 12:45:06 floob: Someone else here probably knows a lot more about this than I do :) Nov 03 12:47:12 Crofton|work: Good point. ;) Nov 03 12:50:21 LocutusOfBorg: it won't have; we moved to meta-freescale Nov 03 13:15:41 how can I get a package built with the sdk into the package feed? Nov 03 13:19:32 ok otavio thanks Nov 03 13:19:51 I wish you had pushed an empty branch with a file README inside, telling us to move :p Nov 03 13:28:36 Hi! does anybody know how the useradd class is supposed to work? I was able to add a user to the system, but now every new build fails because the user is already there: Nov 03 13:28:36 WARNING: useradd command did not succeed. Retrying... Nov 03 13:28:36 | ERROR: Tried running useradd command 10 times without success, giving up Nov 03 13:28:36 | WARNING: exit code 1 from a shell command. Nov 03 13:28:36 | ERROR: Function failed: useradd_sysroot Nov 03 13:44:52 fl0v01: can you explain what you wnt to accomplish and maybe show how you tried to do it? Nov 03 13:46:56 jku: i wanted to add a user for the mosquitto mqtt broker Nov 03 13:47:10 so i added inherit useradd to the recipe Nov 03 13:47:37 USERADD_PACKAGES = "${PN}-user" Nov 03 13:47:37 USERADD_PARAM_${PN}-user = "-r mosquitto -s /sbin/nologin" Nov 03 13:48:17 when i build the image the first time with the new package everything worked and the user is in the system Nov 03 13:50:10 sure something else isn't already creating the user? Nov 03 13:54:39 fl0v01: try moving username to the end of USERADD_PARAMS... Nov 03 13:55:05 or just dropping it Nov 03 13:59:33 fl0v01: yeah, pretty sure the name needs to be last: the class is not very clever about that Nov 03 14:01:00 jku: ok ill try Nov 03 14:03:52 jku: Nice! It worked. Thank you! Nov 03 14:22:31 hello Nov 03 14:24:16 hello Nov 03 14:36:32 hellp Nov 03 16:01:40 something bizarre just happened. Nov 03 16:01:41 I did a bitbake linux-yocto -c cleanall Nov 03 16:01:41 And then tried to make it again. Nov 03 16:01:41 It failed in do_unpack Nov 03 16:01:41 And now my poky directory git remote origin seems to think its a linux kernel repo. Nov 03 16:01:41 And the repository is broken. Don't see branches etc. Nov 03 16:01:41 :s Nov 03 16:01:41 the only thing I can fathom. is I moved between morty to master. Nov 03 16:01:41 didn't source a fresh environment. And just kept on working in another terminal window. Nov 03 16:17:50 ZubairLK: wow, that's weird Nov 03 16:21:40 Thankfully, I haven't lost any work in progress Nov 03 16:21:46 but still quite bizarre.. Nov 03 17:09:23 @pohly, thanks for the fix. I am apparently more under-the-weather than I thought. Nov 03 17:14:11 seebs: no problem. Once I noticed the regression, I started gathering more debug output in an attempt to understand the problem better, and that then led to a solution relatively quickly ;-} Nov 03 17:14:11 yeah. I probably should have thought more about it. In retrospect, a less-minimal test case (with one file that worked) would have probably been a better thing to investigate. Nov 03 17:32:26 Hi all. For local development how would i specify that bitbake uses a local working copy of my repository AND that it wouldn't use the last commited state but the current state of the repository (including uncommited changes) ? Nov 03 17:33:49 e.g. could i use another fetcher that doesn't care about git without messing up my recipe? Nov 03 17:35:13 I mean i could just do the commit and reset the repo to head when i'm not happy with the results. However this is qutie cumbersome and error prone Nov 03 17:37:48 use a file:// uri, or use exterrnalsrc Nov 03 17:45:52 good usecase for devtool Nov 03 17:46:25 Anticom: if you are on something newer than 2.0 then I think devtool is quite good fit for your usecase here Nov 03 19:30:43 does the rpi distro have hardware accelerated gstreamer enabled? Nov 03 19:55:47 seems like it does Nov 03 20:27:55 What is the AB trying to tell me /me my sanity tests keep failing :( Nov 03 20:53:40 mattsm: yes it does, it uses omax Nov 03 20:53:54 and dismanx Nov 03 20:54:06 khem: i cheated and looked in the repo, do you know any good reference for setting up hardware accelerated pipelines? Nov 03 20:54:19 internet documentation is sparse, and i'm a gstreamer newbie Nov 03 20:59:41 try this gst-launch-1.0 souphttpsrc location="http://download.blender.org/peach/bigbuckbunny_movies/big_buck_bunny_720p_h264.mov" ! typefind ! qtdemux name=demux demux. ! queue ! h264parse ! omxh264dec ! glimagesink demux. ! queue ! faad ! autoaudiosink Nov 03 21:04:36 khem: I will! I need to build an image first Nov 03 21:05:25 khem: actually planning to use an rpi to encode some HDMI IP stream to a lower bitrate to faciltate "remote" debugging of some displays Nov 03 21:05:57 just need to craft the right pipeline Nov 03 22:58:33 khem: what's a good image for a weston desktop on rpi3? Nov 03 22:59:56 core-image-weston should work i guess Nov 03 23:09:52 yeah Nov 03 23:10:21 mattsm: if you use meta-metrological then you can use westeros-wpe-image and use SME/EME and secure video stuff Nov 03 23:10:33 is there a distro that set's DISTRO_FEATURES? Nov 03 23:10:47 poky works out of box Nov 03 23:11:06 halstead: I was getting worried about the new autobuilder, then it disappeared offline? Nov 03 23:11:36 RP, I migrated the controller to another host. Nov 03 23:11:45 mattsm: RDK is out of box but them you need to be RDK member etc. Nov 03 23:12:02 halstead: there was a problem with the controller? Nov 03 23:12:28 RP, I am updating hypervisor kernels. There shouldn't be any downtime. Nov 03 23:13:19 RP, I just killed it. The migration wasn't clean and the controller rebooted. Nov 03 23:13:40 halstead: ok, no problem Nov 03 23:13:58 RP, I'm sorry. It will be back up very quickly. Nov 03 23:14:31 halstead: there were errors writing the auto.conf file in that build. We're not sure if its something to do with the buildhistory git repo or something autobuilder related Nov 03 23:16:09 RP, It's back. The workers should reconnect shortly. Nov 03 23:17:04 RP, They are retrying the builds that were interrupted in the shutdown. Nov 03 23:17:30 RP, Should I stop them and empty the queue? Nov 03 23:24:24 halstead: yes, I think we should stop and restart the whole thing Nov 03 23:24:38 RP, one moment. Nov 03 23:26:22 Everything down Nov 03 23:27:15 halstead: ok, thanks. Do you want me to restart? Nov 03 23:27:29 RP, All back up. Ready for a build. Nov 03 23:27:56 halstead: fired Nov 03 23:28:28 RP, My apologies. I've migrated the controller hundreds of times on the production cluster and not had this happen. Nov 03 23:28:47 halstead: I'm going to see if I can find out which process in the createAutoConf is taking too long Nov 03 23:28:59 halstead: these are worrying: https://autobuilder.yocto.io/builders/nightly-x86-64-lsb/builds/92/steps/CreateAutoConf/logs/stdio Nov 03 23:29:11 RP, The setup here is very similar just newer distros. I'll track down the issue and avoid migrations during builds. Nov 03 23:29:13 * RP suspects the git checkout command buried in there Nov 03 23:30:59 RP, likely if there are lots of small files. We may need to do that on local disk and copy to the nfs mount after. Nov 03 23:35:26 halstead: is the build happening on an nfs mount? Nov 03 23:36:59 RP, No but that repo is under /srv/autobuilder/ which is the nfs mountpoint Nov 03 23:37:13 halstead: ah, right Nov 03 23:37:53 halstead: something we may need to discuss with bavery/joshuagl Nov 03 23:38:32 * RP notes its not even through the fetch yet :( Nov 03 23:58:49 halstead: I just traced the single remaining quilt fetch process that is holding the build and its blocked on an NFS lock :( Nov 04 00:00:36 halstead: (pid 28816 on fedora23 stuck in nfs_flock) Nov 04 00:00:52 halstead: stale nfs data from before the reboots? Nov 04 00:01:18 * armpit what is RP still doing awake? Nov 04 00:04:14 armpit: trying to get a test build? trying to figure out some of the various problems going on? :/ Nov 04 00:05:17 * RP runs the build on the main AB Nov 04 00:05:22 * RP heads for bed Nov 04 00:05:59 armpit: krogoth is looking improved btw :) Nov 04 00:06:30 cool Nov 04 00:10:10 RP: I didn't check for stuck child processes when I restarted the workers. Nov 04 00:15:20 children should be set free otherwise they stay in your basement Nov 04 00:20:08 i'm writing a recipe for a cmake tool for a raspberry pi machine, and i have the __curl_rule_01__ that raise because it uses the host (aka x86_64) curl includes Nov 04 00:20:11 why ? Nov 04 00:20:37 s/cmake tool/tool that use cmake for build/ Nov 04 00:20:40 :/ Nov 04 00:21:07 hard to say, that tool's CMakeLists.txt might explicitly add /usr/include to the include path, for example Nov 04 00:21:16 hmm Nov 04 00:21:23 ok, so probably upstream bug Nov 04 00:21:29 because of hard coded path Nov 04 00:21:35 but Nov 04 00:21:42 it's really yocto path Nov 04 00:21:43 Hard to say, unless you share a recipe and a link to the source. Nov 04 00:21:46 not my debian one Nov 04 00:22:19 hmm Nov 04 00:22:32 if i launch `bitbake ettercap` alone Nov 04 00:22:43 it gonna build for my rasp or for my host ? Nov 04 00:22:57 * binarym doubting Nov 04 00:23:02 that depends on the contents of the recipe Nov 04 00:23:30 here is the recipe : http://pastebin.com/fk8hXHUn Nov 04 00:23:36 usually, recipes that don't start with nativesdk- or end with -native are built for the target architecture; but recipes might violate that rule, the name does not determine what the build target is Nov 04 00:23:51 That's a target build Nov 04 00:24:02 i added the PACKAGE_ARCHS="arm" later Nov 04 00:24:08 but doesn't solved Nov 04 00:24:12 That doesn't make a difference Nov 04 00:24:33 and file included is host one: Nov 04 00:24:34 /home/binarym/Work/poky/poky-krogoth-15.0.1/build/tmp/sysroots/x86_64-linux/usr/include/curl/curlrules.h:142:3 Nov 04 00:25:24 You should probably set -DBUNDLED_LIBS=Off Nov 04 00:25:44 hmm Nov 04 00:25:47 really ? Nov 04 00:26:01 ok, gonna try without Nov 04 00:26:16 at first i thought that they brings .a with sources Nov 04 00:26:30 that's why i set it off Nov 04 00:27:55 neverpanic: still the same error and wrong path Nov 04 00:29:41 binarym: what's your $WORKDIR for this build? Nov 04 00:41:53 neverpanic: WORKDIR isn't set in my shell env Nov 04 00:42:25 neverpanic: btw, i used -DSYSTEM_LIBS=OFF Nov 04 00:42:27 to my recipe Nov 04 00:42:39 and now, it doesn't fail on the same error message Nov 04 00:42:42 compilation failed Nov 04 00:42:47 because of -pthread missing Nov 04 00:42:56 on CFLAGS Nov 04 00:43:01 err, that's interesting… Nov 04 00:43:05 CMake should take care of that… Nov 04 00:43:23 You can always CFLAGS_append = " -pthread", though Nov 04 00:43:31 on recipe ? Nov 04 00:43:34 directly ? Nov 04 00:43:35 yes Nov 04 00:43:46 nice :) Nov 04 00:44:30 That's probably a hack, though, and you should figure out what's wrong Nov 04 00:47:45 hmmm Nov 04 00:47:51 now, a new error message ... Nov 04 00:49:10 neverpanic: near 2 am here, i abandon for tonight Nov 04 00:49:17 thks for the help :) Nov 04 00:49:36 Same for me. Nov 04 02:51:02 How do I get my yocto build to display login prompt on serial and VGA console when it boots? Nov 04 02:57:29 floob: enable serial console support in kernel and set SERIAL_CONSOLES variable correctly in yocto metadata for your machine Nov 04 02:59:32 khem: >>"enable serial console support in kernel" would you mind pointing me in the right direction for where to find how to do that please? I'll do the research given a good start point thanks. **** ENDING LOGGING AT Fri Nov 04 03:00:00 2016