**** BEGIN LOGGING AT Thu Jan 09 02:59:58 2020 Jan 09 03:04:51 Marex: ideally I am thinking if we can set XDG_RUNTIME_DIR in service file as a backfill that will be good Jan 09 03:05:01 for meta-qt5 send a pull request to github Jan 09 03:12:27 khem: OK Jan 09 03:12:41 khem: and yes, that would be good Jan 09 03:12:58 khem: I kinda have to wonder why did you ditch the weston-start script though Jan 09 03:18:36 Marex: it does too much for systemd context, especially specifying user etc can be inbuilt with systemd Jan 09 03:18:54 sorting out system stuff atuomagically is always good Jan 09 03:19:04 btw. if you dont use pam then systemd-user might not work Jan 09 03:19:21 I hope you know that Jan 09 03:21:32 khem: I am using this for debugging/development, so no worries Jan 09 03:25:28 ok Jan 09 03:34:01 khem: anyway, I'm calling it a night, thanks for such prompt help! Jan 09 03:41:14 np gn **** BEGIN LOGGING AT Thu Jan 09 05:32:28 2020 **** BEGIN LOGGING AT Thu Jan 09 06:04:39 2020 Jan 09 06:55:52 New news from stackoverflow: Yocto sato image is running extremely slow, is there a way to improve its performance? Jan 09 10:52:55 kanavin: I had to drop elfutils as the rpm fix isn't in current rpm, only the upgraded version Jan 09 10:53:18 kanavin: I do appreciate the patches btw, just trying to give you all the feedback about where the issues are... Jan 09 10:53:31 there are a few unfortunately :( Jan 09 11:01:38 RP, of course. These are tricky upgrades, I expected there will be issues. Jan 09 11:01:54 anyone know why lxc 3.0.x LTS branch recipe is not used in meta-virtualization? following bleeding edge lxc results in, well, bleeding Jan 09 11:06:00 specifically, executing shell commands with env variables etc fails with lxc-attach. lxc is know to break command line compatibility even in minor releases Jan 09 11:06:20 mcfrisk: you need to ask zeddii Jan 09 11:07:09 mcfrisk: btw, we did publish a new stable release policy. I've been meaning to talk to you about that in connection with the sumo patchset Jan 09 11:07:33 the new policy should give us somewhere to place it and a mechanism to collaborate around it Jan 09 11:08:02 sorry these things take time :/ Jan 09 11:09:26 RP: link? I should move all my patches from custom layer to poky and i could send them out.. Jan 09 11:12:45 mcfrisk: https://wiki.yoctoproject.org/wiki/LTS - I know the title says LTS but it covers stable maintenance in general Jan 09 11:15:23 * RP tweaks the intro section there to make this clear Jan 09 11:22:20 RP: like, proper tweaking with a sledge hammer? Jan 09 11:23:09 LetoThe2nd: "percussive engineering" Jan 09 11:26:07 LetoThe2nd: I save those for bitbake internals Jan 09 11:28:39 paulbarker: ++ Jan 09 11:29:11 paulbarker: "chipping data processing" Jan 09 11:35:54 * RP wonders if its normal to have trouble counting the different types of hammer you own Jan 09 11:38:18 RP: Just read that LTS wiki page and it looks good Jan 09 11:57:46 RP: review the LTS as well, good document! Jan 09 11:58:28 though I would consider gcc point release updates ok for stable, major version updates not Jan 09 12:00:36 paulbarker, mcfrisk: glad it makes sense :) Jan 09 12:01:15 mcfrisk: That is one to make judgement calls on I suspect given the circumstances... Jan 09 12:56:57 New news from stackoverflow: WORKDIR in a yocto recipe Jan 09 13:00:42 I don't see any obvious update in this SO page, what triggers this message ^? Jan 09 13:01:14 qschulz: edits in the question. lately folks have been on a question formatting spree, it seems. Jan 09 13:02:32 LetoThe2nd: ok thanks, so I don't have to jump on those to answer before you do :p Jan 09 13:02:53 qschulz: we should perhaps try and fix the bot to better handle that Jan 09 13:27:03 New news from stackoverflow: Update Configuration files in Yocto? Jan 09 14:04:17 I have to cp a file to a directory that I know exists in rootfs, but if I cp it in do_install_append, I get error that no such directory. What do I have to do? Jan 09 14:04:18 hey, my uboot is not booting. i flashed it into QSPI NOR, it looks ok. UART is not echoing one byte. Iam using the IMX6ULLEVK board Jan 09 14:06:05 hello Jan 09 14:06:35 is this the right place to ask about errors with bitbake core-image-weston Jan 09 14:08:10 yes Jan 09 14:09:24 ok, i'm trying to build Yocto v3.9.0 and i got error ERROR: Task (/home/xxx/ff42/build/build/../poky/meta/recipes-core/glibc/glibc-initial_2.26.bb:do_configure) failed with exit code '1' Jan 09 14:09:34 3.9? Jan 09 14:09:39 yes Jan 09 14:09:43 that's not a release we've made Jan 09 14:10:01 what's your release then? Jan 09 14:10:01 anyway, what's the actual error Jan 09 14:10:05 nopeman: Could you post the full bitbake output via something like pastebin Jan 09 14:10:24 latest release was 3.0 Jan 09 14:10:44 ah, ok Jan 09 14:10:49 ok, one sec Jan 09 14:12:37 https://www.paste.org/102241 Jan 09 14:13:44 nopeman: please post the initial logs sent out by bitbake Jan 09 14:14:06 where i can find them? Jan 09 14:14:29 *** These critical programs are missing or too old: GNU ld Jan 09 14:14:36 nopeman: when you call bitbake something it start and shows the details Jan 09 14:14:39 what linux are you building on? Jan 09 14:15:06 mckoan: is asking for the bit that goes BUILD_SYS = "x86_64-linux" Jan 09 14:15:06 NATIVELSBSTRING = "debian-10" Jan 09 14:15:47 ubuntu 18.04.3 Jan 09 14:16:01 you *do* have ld installed right Jan 09 14:16:14 $ ld --version Jan 09 14:16:14 GNU ld (GNU Binutils for Debian) 2.31.1 Jan 09 14:17:11 $ ld --version Jan 09 14:22:43 ? Jan 09 14:25:25 i got GNU ld (GNU Binutils for Ubuntu) 2.30 Jan 09 14:25:35 do i need newer version? Jan 09 14:27:13 New news from stackoverflow: YOCTO - A Issue: nativesdk-qtbase: Files/directories were installed but not shipped in any package Jan 09 14:27:35 inside tmp/work/x86*/glibc-initial/[version]/build there will be a config.log file Jan 09 14:27:44 that most likely has more clues as to why it doesn't like your ld Jan 09 14:30:57 As a headups I've decided to do some major surgery on the sstate cache on the autobuilder which may break builds a little if they're running whilst this happens. We have a tmp file explosion problem in sstate :( Jan 09 14:31:00 i got it in aarch64-poky-linux catalog Jan 09 14:31:10 nopeman: and this is a real ubuntu? not WSL? Jan 09 14:31:23 real Ubuntu on VM Jan 09 14:34:36 nopeman: plus, 3.0 has glibc 2.30 Jan 09 14:35:20 rocko shipped 2.26 Jan 09 14:35:33 master doesn't even have glibc-initial Jan 09 14:36:19 nopeman: so, please take a minute to describe in detail what this thing that you are actually trying to build is, really. Jan 09 14:37:33 well i got this thing i'm trying to build:https://elinux.org/R-Car/Boards/Yocto-Gen3 Jan 09 14:38:10 harass renesas? Jan 09 14:38:25 nopeman: build it on a host from the era this stuff was created in Jan 09 14:38:39 god its horrible that they're calling their releases "Yocto 3.21.0" Jan 09 14:38:49 nopeman: you need an older OS in your VM Jan 09 14:38:50 Renesas R-Car-M3 Starter Kit Pro to be exact Jan 09 14:38:52 at least they didn't call it Yocto Linux 3.21.1 Jan 09 14:39:05 rburton: that would have probably been better :/ Jan 09 14:39:10 that article is awfulin some ways i instantly want to get drunk. Jan 09 14:39:11 nopeman: 'Yocto v3.x.y(Ex Yocto v3.21.0) : Ubuntu 16.04 LTS (64bit) is recommended as OS' Jan 09 14:39:22 and people wonder why I insist on codenames for the branches Jan 09 14:39:26 RP: If I wanted to get 1/2 wic changes backported to zeus/warrior, how long should they have been in master for? Jan 09 14:39:28 lol Jan 09 14:39:41 RP: not my, i love codenames. Jan 09 14:40:28 Tartarus: people seem to love doing immediate backports, I try and at least let it be there for a week or two Jan 09 14:40:40 Tartarus: armpit is probably the one to ask as he handles it Jan 09 14:40:48 RP: OK, I'll come back to this in about a week :) Thanks Jan 09 14:40:51 ok. so i should start with ubuntu 16.04 Jan 09 14:41:14 yeah try that. and ask renesas if they're going to backport the fixes or make a new release. Jan 09 14:41:34 thanks for your help i got to build another setup now. Probably i'll be back Jan 09 14:42:27 rule of thumb: if it involves cars, assume you need to use a really old distribution Jan 09 14:43:29 rburton: rule of real life: if it involves reneses, you better.... Jan 09 14:45:24 ha Jan 09 14:45:39 better leave? :) Jan 09 14:46:30 insert what you feel suitable, personally :) Jan 09 14:49:44 Hello :) Jan 09 14:50:52 I have to cp a file to a directory that I know exists in rootfs, but if I cp it in do_install_append, I get error that no such directory. What do I have to do? Jan 09 14:51:28 stuom1: Create the directory first Jan 09 14:51:30 each do_install is run independently of the others.. so you need to mkdir -p the directory in each of them Jan 09 14:54:09 Anyone know what code gets executed on the "server thread" as a result of a cli command being passed via the connection.send here? https://github.com/openembedded/bitbake/blob/master/lib/bb/server/process.py#L332 Jan 09 14:56:35 paulbarker thanks, I dont know why i didnt try mkdir...i just assumed the toolchain makes the directory since it exists on rootfs Jan 09 14:57:15 stuom1: It can easily catch you out when learning Yocto. Each recipe's do_install is writing to a staging directory not directly to the rootfs Jan 09 14:57:45 The rootfs is assembled at the end Jan 09 14:57:48 (this is done to ensure that each chunk of the system can be standalone if necessary) Jan 09 14:58:29 ok, i understand Jan 09 15:50:32 kpo: hi Jan 09 16:57:42 New news from stackoverflow: Poky-sumo : nothing provides kernel-module-hello... error Jan 09 17:10:24 hi Jan 09 17:10:26 getting Jan 09 17:10:29 NOTE: Reconnecting to bitbake server... Jan 09 17:10:29 NOTE: Retrying server connection (#1)... Jan 09 17:10:36 what could be the reason ? Jan 09 17:10:45 you have a bitbake session that is still running and hasn't completed.. Jan 09 17:10:47 check ps Jan 09 17:11:09 angelo__: Or, bitbake exited but bitbake.sock didn't get removed Jan 09 17:11:17 oh Jan 09 17:11:52 fray, JPEW thanks, killed. Let's see Jan 09 17:12:09 angelo__: Are you building with a container? Jan 09 17:17:04 schroot Jan 09 17:17:17 was exited and reentered and issue was still there Jan 09 17:17:26 killed python3 instance, all works now Jan 09 17:17:36 thanks ! Jan 09 17:17:56 angelo__: I've seen that when the container exits without giving bitbake a chance to cleanup Jan 09 18:00:59 ohai o/ Jan 09 19:48:05 hmm. perf on 5.5+ kernels is failing due to missing modules in our native python. Jan 09 19:48:09 * zeddii looks at the recipe Jan 09 19:49:58 * zeddii sees the do_install_append messing around with the right bits . hmmm. Jan 09 20:30:45 If I would like to develop/debug inside the bitbake script... is it necessary to source the bitbake environment? Otherwise... is there a developement readme for bitbake? Jan 09 20:39:25 what exactly are you trying to debug? Jan 09 20:39:33 how you go about it depends on what you're trying ot do Jan 09 20:40:13 kergoth: I would like to try to implement a feature inside devtool Jan 09 20:40:28 devtool is its own script Jan 09 20:40:36 not bitbake itself Jan 09 20:40:39 but devtool needs bitbake env right? Jan 09 20:45:32 khem: heh, seems the solution wasn't as easy afterall :) Jan 09 20:45:54 khem: I'll test it ASAP Jan 09 20:55:21 Hmm something off in the toolchain build between x86_64 and x86_64-x32, it seems. If you do a multiconfig build of those two and kick off meta-toolchain for both, get failures. guessing something is rebuilding that shouldn't be or vice versa Jan 09 20:55:23 * kergoth digs Jan 09 21:06:15 * armpit ooh, sound like a good selftest to add Jan 09 21:06:47 ouch: Connecting to downloads.yoctoproject.org (downloads.yoctoproject.org)|198.145.29.63|:80... failed: Connection refused. Jan 09 21:07:29 mcfrisk, I just checked and it is working for me Jan 09 21:07:46 mcfrisk: maybe it just doesn't like you, personally.... Jan 09 21:07:50 * LetoThe2nd ducks and runs! Jan 09 21:08:12 that was from my CI build machine.. seems like one machine doesn't have http://downloads.yoctoproject.org/releases/opkg/opkg-0.4.1.tar.gz in cache.. Jan 09 21:08:59 * mcfrisk git blames LetoThe2nd Jan 09 21:09:41 mcfrisk: :) Jan 09 21:09:43 mcfrisk, there was some possible NAS work that could have caused a burp Jan 09 21:10:23 mcfrisk, LetoThe2nd will blame LetoThe1st Jan 09 21:12:34 armpit: hey leave my granddad out, mkay?!? Jan 09 21:13:33 sigh, life would be easier without knowledge, and SW development easier without testing.. Jan 09 21:14:42 mcfrisk: ah then thats probably why my sw is always so easy to develop :) Jan 09 21:14:57 fired zeus update CI builds on many CI instances which have not been building the topic before to test things out, and caboom... I will manually add things to caches Jan 09 21:16:26 I was so hoping to have seen last download cache misses this week. was a pita with maven, eclipse and p2 this week. Jan 09 21:17:42 * armpit with zeus I normally see lighting bolts Jan 09 21:18:01 * mcfrisk hits opencv too Jan 09 21:22:08 Marex: yeah, although my purist mind wants to use just systemd unit file and no helper scripts Jan 09 21:22:43 Marex: weston disables systemd support in its meson build system when pam is not in DISTRO_FEATURES Jan 09 21:23:20 so weston with pamless systemd distro is not expected to work Jan 09 22:01:26 khem: so maybe weston should require DISTRO_FEATURES "pam" if there's systemd in DISTRO_FEATURES ? Jan 09 22:01:29 khem: then all is good ? Jan 09 22:09:33 armpit: nas issues will only impact autobuilder builds and sstate FWIW Jan 09 22:10:57 Marex: yes I guess that will be a good patch Jan 09 22:11:13 RP: is AB back? Jan 09 22:12:43 khem: no, need to do some work on sstate for a bit Jan 09 22:13:02 khem: its too big and broken :/ Jan 09 22:15:13 ok Jan 09 22:16:23 Marex: I am thinking to install a profile fragment in /etc/profile.d which will define XDG_RUNTIME_DIR if not defined and keep the current systemd service Jan 09 22:22:43 kergoth: Had a chance to look at the latest pyrex :) Jan 09 22:34:43 khem: well if it's flaky without PAM, maybe just ditching support for such configuration would be for the best ? Jan 09 22:35:09 khem: instead of just trying to bend things to work, while having all the inobvious problems Jan 09 22:49:16 Marex: yes Jan 09 22:52:59 khem: maybe we should do that then Jan 09 23:04:17 Marex: agreed fwiw Jan 09 23:20:34 I get many errors like "useradd command did not succeed." from my user recipe after minor changes, is this a known bug? Jan 09 23:21:33 dev1990: maybe your custom recipe is broken ? :) Jan 09 23:24:24 https://github.com/dev-0x7C6/meta-retro/blob/zeus/recipes-core/retroarch/retroarch-user.bb Jan 09 23:24:42 this is recipe that I'm working on Jan 09 23:26:43 heh, after posting I see minor issue with --user-group -G Jan 09 23:34:25 meh, and now it's working fine without any chagnes, maybe I should just cleanup my workspace Jan 09 23:51:56 Marex: I think that should be a separate patch on top of current one Jan 09 23:52:11 Marex, I think current patch is fine to live with for now Jan 09 23:53:42 khem: it kinda is, yep Jan 09 23:53:46 khem: I have a patch like that locally Jan 09 23:54:03 ooko Jan 10 00:13:53 weston/ **** ENDING LOGGING AT Fri Jan 10 02:59:58 2020