**** BEGIN LOGGING AT Thu Feb 06 02:59:57 2020
Feb 06 03:11:46 <kiwi_29> khem thanks. but there is no other way to single step?    the devpyshell looks promising.
Feb 06 08:14:07 <LetoThe2nd> as i can't spot it from glaring at the recipes: how do i disable (not remove!) systemd-networkd?
Feb 06 08:14:59 <qschulz> LetoThe2nd: SYSTEMD_AUTO_ENABLE?
Feb 06 08:15:46 <qschulz> (set to disable)
Feb 06 08:16:14 <LetoThe2nd> qschulz: i would agree for about any recipe that is kind of freestanding, but networkd is part of systemd itself and the recipe doesn't seem to use that mechanism
Feb 06 08:21:11 <kriive> LetoThe2nd: PACKAGECONFIG_remove = " networkd" in a systemd bbappend?
Feb 06 08:21:34 <kriive> Sorry, read too late
Feb 06 08:21:46 <kriive> You meant disable
Feb 06 08:21:51 <LetoThe2nd> yep
Feb 06 08:24:33 <qschulz> LetoThe2nd: worst case scenario, a postinst script which runs systemctl disable systemd-networkd?
Feb 06 08:24:57 <LetoThe2nd> qschulz: sure. but i though i'd ask if can avoid that.
Feb 06 08:32:00 <kriive> If I have to provision a initial network config for all of my devices (i.e. a standard static ip on eth0), what would be the "correct" solution?
Feb 06 08:32:40 <LetoThe2nd> kriive: write a recipe that just installs the configuration files as needed.
Feb 06 08:32:50 <kriive> I am currently bbappending systemd to copy a systemd-netowrd config, but every time I need to recompile systemd
Feb 06 08:33:58 <LetoThe2nd> like i said, write a recipe that just installs the .network files to /etc/systemd/network, and have it RDEPEND_${PN} = "systemd" for correctness.
Feb 06 08:34:03 <kriive> That makes sense ty
Feb 06 08:38:40 <LetoThe2nd> is it possible to have pkg_postinst_append() :P
Feb 06 08:39:58 <qschulz> LetoThe2nd: meta/classes/manpages.bbclass says yes?
Feb 06 09:06:00 <yocti> New news from stackoverflow: Yocto project Bitbake error : ERROR: ParseError in poky/meta/conf/machine/: not a BitBake file <https://stackoverflow.com/questions/60091093/yocto-project-bitbake-error-error-parseerror-in-poky-meta-conf-machine-not>
Feb 06 09:58:32 <creich> attieg: maybe you already explained, but what exactly was the point of having the module out-of-tree again?
Feb 06 10:36:18 <yocti> New news from stackoverflow: how to add recipe for yocto systemd service <https://stackoverflow.com/questions/54660983/how-to-add-recipe-for-yocto-systemd-service>
Feb 06 11:48:05 <RP> rburton: gstreamer floating dependency on libcap2 :/
Feb 06 11:48:50 <RP> rburton: iputils floating suid bits
Feb 06 12:13:46 <rburton> floating suid bits, how?
Feb 06 12:13:57 <rburton> at least the cap2 one is easy
Feb 06 12:56:53 <nameclash> LetoThe2nd, thanks for the externalsrc hint -- that was exactly what I was looking for
Feb 06 12:59:20 <LetoThe2nd> nameclash: :)
Feb 06 13:00:46 <nameclash> I have another (unrelated) question though: is there any standard way to package installs to mountpoints? Say, my fstab defines /dev/mmcblk0p3 being mounted at /home/root/foo and I have some recipe populating the rootfs at /home/root/foo/bar
Feb 06 13:02:12 <nameclash> what I observe is that /home/root/foo is empty when running the image on the rig so I suspect there's a conflict
Feb 06 13:02:21 <LetoThe2nd> nameclash: not sure i understand the question
Feb 06 13:03:20 <nameclash> the rootfs image that falls out of the yocto build contains the following paths:
Feb 06 13:03:31 <LetoThe2nd> nameclash: you mean, you want some parts of the image to be on seperate partitions?
Feb 06 13:03:49 <nameclash> yes
Feb 06 13:04:06 <LetoThe2nd> nameclash: that can be done through a .wks file, utilizing wic.
Feb 06 13:04:25 <nameclash> ok
Feb 06 13:04:26 <LetoThe2nd> nameclash: the storage layout is beyond the rootfs building stage.
Feb 06 13:04:45 <nameclash> yeah, that's what I was thought
Feb 06 13:04:53 <nameclash> thanks, I'll look into
Feb 06 13:08:20 <BMWE> Hi all,
Feb 06 13:09:21 <BMWE> My solution vendor has provided me some HW with Yocto-Poky
Feb 06 13:10:15 <BMWE> I need there port forwarding via ssh, but can't find the sshd_config file. Any help would be appreciated
Feb 06 13:24:41 <mckoan> is it supposed to inherit something special to use .cfg kernel fragments?
Feb 06 13:25:03 <mckoan> looks like that inherit kernel is not enough
Feb 06 13:26:18 <LetoThe2nd> mckoan: i think it must be yocto-kernel style, but not sure
Feb 06 13:30:53 <mckoan> LetoThe2nd: as I suspected, thx
Feb 06 13:32:00 <qschulz> mckoan: linux-yocto iirc
Feb 06 13:32:36 <qschulz> and I was wrong :) LetoThe2nd has the correct answer :D
Feb 06 13:40:45 <LetoThe2nd> qschulz: but...but... but... you're the kernel guy!
Feb 06 13:46:39 <qschulz> LetoThe2nd: but... but... but... I don't make Yocto kernel recipes :D (and we don't use fragments at the moment)
Feb 06 13:49:45 <mckoan> LOL
Feb 06 13:51:17 <jonmason> I'm seeing failures in the gcc test on testimage on zeus.  Is this a known issue?
Feb 06 13:51:20 <LetoThe2nd> qschulz: hi5
Feb 06 13:51:51 <LetoThe2nd> jonmason: failures are a known issue. generally.
Feb 06 13:52:46 <jonmason> I didn't see an open bug.  So I just wanted to make sure I'm not being stupid (or more stupid than normal)
Feb 06 13:54:53 <LetoThe2nd> jonmason: so rest assured, you are just being averagely sutpid.
Feb 06 13:55:30 <LetoThe2nd> nah, just being the inevitable jester. no idea, seriously.
Feb 06 14:01:04 <dev1990> Did someone managed to configure bitbake with vscode?
Feb 06 14:01:11 <dev1990> I found this plugin https://github.com/EugenWiens/vscode-bitbake
Feb 06 14:02:57 <jonmason> LetoThe2nd: you are king of the yocto trolls ;-)
Feb 06 14:11:47 * LetoThe2nd bows
Feb 06 14:20:20 <mckoan> LetoThe2nd: actually not-yocto kernels can't support fragments, so we managed it using a config_prepend
Feb 06 14:20:21 <RP> jonmason: not a known issue, no. it works in all our autobuilder tests
Feb 06 14:20:47 <mckoan> LetoThe2nd: just FYI
Feb 06 14:22:20 <LetoThe2nd> mckoan: thanks. i'll try to remember (and fail in 5 minutes, but hey, its the effort that counts right?)
Feb 06 14:28:26 <mckoan> LetoThe2nd: indeed it's very appreciated ;-)
Feb 06 14:54:12 <jonmason> Are people doing xen using multiconfig for the dom0 and domUs?
Feb 06 15:00:18 <jonmason> RP: I'll post email to the list to confirm the error and make sure I'm seeing up things properly.  If so, then I'll open a bug
Feb 06 15:21:02 <attieg> creich: the driver is very closely related to the firmware running on a co-processor, and I'd like to be able to keep them as a pair, in their repository...
Feb 06 15:21:21 <attieg> though now you question it, that's not impossible to do in-tree... hmm
Feb 06 15:21:51 <attieg> ideally I'd like to avoid maintaining a set of patches along side the repository, and risk them going out of sync, etc...
Feb 06 15:22:55 <Ad0> I have trouble finding the task / where to _append() when it installs stuff to uboot partition on raspberry pi image build. I need to rename 2 files before they are put on that partition in the image
Feb 06 15:38:26 <fullstop> is there a recipe for lsgpio (it's in the kernel tree) ?
Feb 06 16:07:19 <yocti> New news from stackoverflow: How to setup bootargs to load nfs in beaglebone through eth0 <https://stackoverflow.com/questions/57948311/how-to-setup-bootargs-to-load-nfs-in-beaglebone-through-eth0>
Feb 06 16:11:12 <fabian81> is it possible to modify a recipe with devtool that is based on multiple git repositories? i am trying and devtool is extracting only one repo ignoring destsuffix and second repo
Feb 06 16:53:36 <angelo__> how can i set a preferred gcc to cross compile from a machine conf ?
Feb 06 17:01:12 <qschulz> angelo__: PREFERRED_VERSION_gcc = "xyz" ?
Feb 06 17:01:41 <qschulz> (might have to check there isn't a virtual package for gcc but it should be preferred_version for sure)
Feb 06 17:07:32 <yocti> New news from stackoverflow: How to add python 2.7 in yocto sdk <https://stackoverflow.com/questions/60100155/how-to-add-python-2-7-in-yocto-sdk>
Feb 06 17:21:43 <zeddii> RP: did you toss that queue of tool changes in the bin ? If not, i'll send a single patch on top of it that fixes things up.
Feb 06 17:21:57 <zeddii> or, I'll send a v2 with them all.,
Feb 06 17:28:05 <RP> zeddii: I dropped for now, I can add back easily
Feb 06 17:28:35 <RP> zeddii: wasn't sure if it was an incremental fix or not
Feb 06 17:31:54 <zeddii> neither was I :D I'll send the whole queue as a v2, save you rummaging around for the old one.
Feb 06 20:00:31 <Nathan22> Any idea why a 'install -d -m0755 ${D}${datadir}/${BPN}/examples' wouldn't generate the directory in a do_install_append?
Feb 06 20:01:56 <emg> I'm buildingan install image using sdimage-bootpart.wks which uses the bootimg-partition plugin. One of the files I had in my IMAGE_BOOT_FILES_append is now larger than 4GB so it can't be added to the boot partition which is vfat. How can I add that file to the second rootfs partition instead in my image.bb?
Feb 06 20:07:25 <emg> more accurately, I'm trying to include an image.wic as rootfs.img (instead of the image.wic.xz) and because it's big I want to put it in the other partition which the install script will still find as it looks for /run/media/*/rootfs.img
Feb 06 20:17:38 <emg> while the first line in sdimage-bootpart.wks is: part /boot --source bootimg-partition ...   the second line is   part / --source rootfs ...   so I think then my question is how do I add the file I want to the rootfs instead (currently it's in IMAGE_BOOT_FILES for bootimg-partition)
Feb 06 20:19:03 <emg> where "the file I want" is an image.wic, instead of adding a normal package (do_rootfs[depends] += "my-image:do_image_complete" is earlier in the file so I know the image exists)
Feb 06 20:44:58 <fullstop> I forced a kernel build and now it's telling me that my build is tainted from doing this.  It doesn't go away after a clean.  How do I clean my, uhh, taint?
Feb 06 20:49:32 <Crofton|road> any advice on cloning linux-yocto from git.yp.org nd not having it fail?
Feb 06 20:52:04 <palate> Hej :)
Feb 06 20:56:42 <emg> fullstop: perhaps do a bitbake -c cleanall <yourkernelpackage>
Feb 06 20:59:44 <fullstop> emg: let me try that
Feb 06 20:59:53 <fullstop> I did a clean, but not a cleanall
Feb 06 21:03:26 <fullstop> (still fetching -- that was never tainted)
Feb 06 21:06:00 <fullstop> emg: that seems to have worked.  Thanks!
Feb 06 21:06:21 <emg> ok I'm trying to just make a recipe that copies a file to /rootfs.img but adding that to IMAGE_INSTALL doesn't seem to add it to that second partition (rootfs partition)
Feb 06 21:06:24 <emg> fullstop: awesome!
Feb 06 21:33:45 <rburton> RP: kanavin is a genius for spotting the obvious
Feb 06 22:01:28 <RP> rburton: but we're talking about config.h, not config.h
Feb 06 22:01:34 <RP> not xconfig.h
Feb 06 22:01:42 <rburton> look at the diffoscope
Feb 06 22:01:46 <rburton> the variadic was config.h
Feb 06 22:01:48 <rburton> the rest were xconfig.h
Feb 06 22:03:24 <RP> rburton: ah, yes. Ahhrg
Feb 06 22:03:27 <rburton> indeed
Feb 06 22:03:29 <rburton> thanks alex :)
Feb 06 22:03:42 <rburton> i somehow totally missed the change in filename too
Feb 06 22:03:56 <rburton> packages/corei7-64-poky-linux/perl/perl-dev: FILELIST: removed "/usr/lib/perl5/5.30.1/x86_64-linux/CORE/xconfig.h"
Feb 06 22:03:59 <rburton> \o/
Feb 06 22:04:47 <RP> rburton: that explains a lot
Feb 06 22:04:54 <rburton> yes!
Feb 06 22:05:03 <rburton> with many eyes etc
Feb 06 22:05:21 <rburton> neither the perl-cross maintainer or us noticed :)
Feb 06 22:07:13 <RP> rburton: you have a patch you're testing?
Feb 06 22:09:33 * RP notes patch on list
Feb 06 22:11:48 <emg> Think I figured this out, we'll see, the do_package and do_rootfs take forever with a 31G file in there :-/
Feb 06 22:11:59 <emg> first pass exploded because there's a set max rootfs size that I exceeded
Feb 06 22:57:16 <emg> holy shit it worked
Feb 06 22:57:40 <emg> now I can losetup/mount a partition from the image to do stupid stuff during the install
Feb 06 23:05:03 <RP> emg: we don't tend to do that as it needs root privs :(
Feb 06 23:13:32 <emg> RP: eh? I have a feeling I said something ambiguous as I don't understand what you mean by that
Feb 06 23:37:38 <mischief> i have a recipe in a layer that points to a private git repo. i want to share my layer with someone, but not give them access to that git repo. currently bitbake will fail at the recipe parsing step for that recipe. is there a simple way to avoid that besides putting it in a different layer?
Feb 06 23:40:07 <palate> I'm getting a build failure with Yocto 3.0.1: died with <Signals.SIGSYS: 31>
Feb 06 23:40:21 <palate> where could that come from?
Feb 06 23:40:57 <palate> actually it says that build/tmp/work/core2-32-poky-linux/initscripts/1.0-r155/package/etc/init.d/umountnfs.sh
Feb 06 23:41:03 <palate> died with sigsys
Feb 06 23:42:34 <JaMa> if "bitbake -k -c testimage core-image-sato-sdk-ptest" is running over 15000s without any output (other than "Bitbake still alive (15000s)") is it safe to assume that something went wrong? This is with poky DISTRO, with nodistro it failed much sooner
Feb 06 23:45:24 <khem> yes
Feb 06 23:46:05 <khem> which qemu arch
Feb 06 23:46:22 <khem> and perhaps use core-image-sato-sdk-ptest-fast first
Feb 06 23:47:44 <palate> Actually I'm trying to build an ubi image but with 2.7.1, ubi fails, and 3.0.1 just fails to build the minimal image in the first place
Feb 06 23:52:07 <khem> palate: those were questions for JaMa but lets see whats up here on your issue
Feb 06 23:52:41 <khem> palate: are you on archlinux by any chance ?
Feb 06 23:53:12 <khem> or another host distro which has file utility compiled with libseccomp enabled ?
Feb 06 23:56:29 <palate> khem: archlinux, yes :/
Feb 06 23:56:48 <palate> khem: not sure what libseccomp is, though
Feb 06 23:59:56 <RP> emg: using loop mounts and losetup need root permissions. We don't run builds as root
Feb 07 00:06:15 <RP> armpit: the zeus release build failed with the runpid bug I just reassigned to tgamblin this morning :/
Feb 07 00:06:25 <RP> (multilib)
Feb 07 00:07:57 <tgamblin> RP: I'll prioritize that one along with oeqa skipTests
Feb 07 00:10:00 <RP> tgamblin: its a nice easy one, just put a try/except in
Feb 07 00:10:18 <RP> tgamblin: the process is exiting at an unfortunate time between the code lines afaict
Feb 07 00:10:41 * RP isn't sure how we saw it twice in 24 hours 
Feb 07 00:20:23 <mischief> anyone?
Feb 07 00:28:50 <tgamblin> RP: Yeah, just recreating it that may take some time
Feb 07 00:30:22 <khem> palate: you need to pin your file package to 5.37
Feb 07 00:32:15 <khem> palate: sudo  pacman -U /var/cache/pacman/pkg/file-5.37-5-x86_64.pkg.tar.xz and then add IgnorePkg   = file to /etc/pacman.conf
Feb 07 00:32:45 <khem> that will downgrade it to 5.37 and then ignore future upgrades, its not best but thats a workaround for now
Feb 07 00:33:18 <khem> other option is to use docker container, if you are interested then I can tell you about that too
Feb 07 00:33:46 <creich> attieg: sorry for the late answer. came in very late today ^^
Feb 07 00:34:42 <creich> if there is no other reason than that, maybe you just find a way to grab the patch and the co-processor firmware within one bb-layer
Feb 07 00:35:10 <creich> even though it would be interesting why the out-of-tree build fails and how to maintain such a driver in general
Feb 07 00:35:46 <creich> did you make any progress? i'll be unavailable for the next to days at least. so maybe sunday night i might be available, but i can't promise
**** ENDING LOGGING AT Fri Feb 07 02:59:57 2020