**** BEGIN LOGGING AT Wed May 26 03:00:34 2021 May 26 07:09:31 In an ongoing build, is there a way to terminate a specific ongoing task ? May 26 07:11:30 I guess killing the log.do_xxx. should work.. May 26 07:14:13 hello yoctoers ! May 26 07:14:34 One question, I have a .vimrc that I want to add to each user in the distro May 26 07:15:07 usualkly I would create a vim_%.bbappend and install the .vimrc in ${D}/home/root May 26 07:15:21 but how to manage it if I have multiple users ? May 26 07:16:01 thekappe, isnt that what /etc/skel/ is used for ? May 26 07:16:25 thekappe: you could install it in /etc/vim/vimrc, it will apply for all users on that system May 26 07:17:28 or /usr/share/vim/vimrc May 26 07:19:46 hello guys May 26 07:19:57 didn't know about that May 26 07:20:17 I'll try your solutions May 26 08:29:44 good morning May 26 08:46:47 Hello all, if i call oe_runmake or make in a custom do_compile() function, I have to pass the number of thread or bitbake set it from BB_NUMBER_THREADS/PARALLEL_MAKE variable ? Thx! May 26 09:05:37 Ninic0c0: do_compile() already calls oe_runmake therefore make by default. You don't need to call it again May 26 09:07:54 Ninic0c0: and it should use PARALLEL_MAKE variable setting May 26 09:10:25 mckoan thank you for support, it's a custom recipe :) May 26 10:14:57 @Ninic0c0 if you call oe_runmake instead of make it should do it for you May 26 10:29:33 * RP looks at https://autobuilder.yocto.io/pub/non-release/20210526-2/testresults/qemuarm/2021-05-26--08-58/host_stats_1_top.txt and worries about the cpu/memory various things are using May 26 10:30:21 7g RES, 16g VIRT for cmake's rpmbuild for example May 26 10:30:33 or the 100% cpu usage of the many bitbake-workers May 26 10:48:04 is there some chance to use `bitbake -g -u taskexp` from the crops/poky docker container? May 26 10:49:50 RP: 16gb virtual memory for building the rpms of cmake? May 26 10:53:51 oh, I can use it with ncurses ... 😁 May 26 10:54:12 then what's taskexp? 🤔 May 26 10:54:42 taskexp is a graphical UI for exploring tasks and dependencies May 26 10:55:56 oh, I see ... so I guess ncurses is the appropriate library to use into a docker container 🤔 May 26 10:58:59 rburton: yes. See the log May 26 11:00:21 would that be rpm's xz going full on cookie monster? May 26 11:00:22 now I'm wondering why in the --help output of bitbake I can see only knotty or ncurses as values for the -u argument, why it doesn't show also taskexp? May 26 11:00:49 most likely because someone forgot to add it to the help output May 26 11:02:13 Hello, on the master branch, with CROPS ubuntu20.04, I have some do_package core dumped issues with glibc, I observe this since 2 May: ERROR: glibc-2.33-r0 do_package: dump_syms failed with exit code 134 (cmd was dump_syms '/home/yoctouser/build/tmp/work/core2-32-sbr-linux/glibc/2.33-r0/package/usr/sbin/zic' 2>/dev/null): May 26 11:03:24 /bin/sh: line 1: 1025927 Aborted (core dumped) dump_syms '/home/yoctouser/build/tmp/work/core2-32-sbr-linux/glibc/2.33-r0/package/usr/sbin/zic' 2> /dev/null May 26 11:04:56 is this normal? https://imgur.com/KxkiASY.png it's apparently stuck in here ... should I expect something more? May 26 11:05:55 I guess that ncurses UI should use that .dot generated files to show me something interactive, right? May 26 11:06:30 the curses ui isn't for exploring dependencies May 26 11:07:21 rburton, then how could I explore dependencies on cli? May 26 11:07:42 grep on the dot file is easy to be honest May 26 11:07:50 dot files are human readable, and easily grepped May 26 11:08:21 you might be able to get crops to forward the X sockets so you can use the ui app from inside the container, dunno May 26 11:15:23 khem: you broke non-linux builds QA Issue: gcc-runtime: Files/directories were installed but not shipped in any package: /usr/lib/-gdb.py May 26 11:15:37 rburton: I suspect rpm's xz usage May 26 11:15:48 * RP is still hoping kanavin might poke at it May 26 11:17:43 hm, rpm has limits on xz in 32-bit environments but not 64 :) May 26 11:18:45 rburton, apparently the container is lacking gtk 🤔 https://bpa.st/HSCQ May 26 11:18:49 okay someone was on something good when they wrote this May 26 11:19:08 rburton: rpm code? May 26 11:19:25 wyre: you'll also need to shuttle the X connection between the container and the host. obviously containers don't do that by default. you could just not use a container... May 26 11:19:27 RP: yes May 26 11:20:32 rburton, apparently using -e argument in the docker run should be enough http://fabiorehm.com/blog/2014/09/11/running-gui-apps-with-docker/ shouldn't it? May 26 11:20:54 wyre: rebuild crops with the extra dependencies and find out May 26 11:21:35 maybe switching to zstd might be a good idea for rpm anyway May 26 11:22:39 rburton: can we hack it somehow? May 26 11:24:50 ah if i update my clone the really horrible code disappears May 26 11:24:52 which is good May 26 11:26:19 ffs rpm why May 26 11:26:33 * RP decides he probably doesn't want to look May 26 11:26:45 rburton, I cannot find anything about rebuild crops with extra dependencies May 26 11:26:59 its just a docker container, get the dockerfile, add the extra deps, rebuild May 26 11:27:08 oh, you mean manually May 26 11:27:12 yes May 26 11:27:16 rburton, which part specifically? May 26 11:27:22 but I'd need to know what extra deps I need May 26 11:27:38 I'm not sure what gtk package I need May 26 11:27:48 wyre: gtk3 and python-gobject should cover it May 26 11:28:27 kanavin: how rpm uses T[integer] open modes to control the number of threads used May 26 11:34:37 kanavin: where do we tell rpm to use something other than gzip to compress? May 26 11:35:22 RP: well its a two line patch to enable the thread count cap on all hosts, which caps to 4 threads May 26 11:39:52 rburton: does that limit the memory? May 26 11:40:07 rburton: we should put a limit on the threads if it is currently unlimited May 26 11:40:21 it uses the processor count May 26 11:40:51 rburton, what gtk3 package? http://ix.io/3nUD May 26 11:43:08 RP: so there's a macro for xz memory limit May 26 11:43:16 I guess libgtk-3-0 May 26 11:43:17 and i've a patch to swap ncpus for 4 May 26 11:43:39 but what about libgtk-3-bin? May 26 12:03:07 rburton: sounds good, thanks! May 26 12:03:24 doing a build test now to see if i can reproduce the crazy memory usage May 26 12:09:04 rburton, well, apparently all is right, I've installed the necessary dependencies and now I'm not having any python traceback, but GTK still doesn't init despite DISPLAY variable is properly configured 🤔 https://bpa.st/6HPA May 26 12:11:20 rburton: FWIW, bitbake cmake -c package_write_rpm -f shows fairly high mem use here May 26 12:38:30 rburton, package.bbclass May 26 12:38:40 rburton, and it already uses xz, not gz May 26 12:42:30 Yocto seems to support HTTP basic authentication for using an sstate-cache mirror. https://patchwork.openembedded.org/patch/130333/ Where do I need to put the credentials? May 26 13:31:52 Hi all, I am using eSDK, trying to build kernel (linux-yocto) and see task do_unpack_and_patch, which actually does NOT apply any patch in .kernel-meta/patches. Does anyone of you have any tip? May 26 13:51:17 hi there guys, anyone that built wpewebkit+cog for imx6? May 26 13:51:45 i've build with rdk backend for imx6 on framebuffer, but when i call cog there's nothing showing on the screen May 26 14:09:17 How does one adjust the timeout for QEMU to boot so that slow boots don't fail with "oeqa.core.exception.OEQATimeoutError: Timed out after 30 seconds of execution" ? May 26 14:11:39 I tried using TEST_QEMUBOOT_TIMEOUT = "180" in local.conf, but it still gave the above error May 26 14:23:36 ./meta/lib/oeqa/runtime/cases/ping.py: @OETimeout(30) May 26 14:36:29 hi all -- I can `git clone git@github.com:company/repo.git` from the kas shell, but bitbake fails to fetch the sources from my recipe. how should I format the SRC_URI for my private github repo? May 26 14:49:49 RP: Ugh, the MinGW build for ztd is a mess May 26 14:50:04 It rebuilds in do_install because all the make targets are wrong :( May 26 14:51:30 JPEW: ah :( May 26 14:52:21 JPEW: I suspect this is going to cause issues for anything using cmake :( May 26 14:56:43 Ya. I the fix to make it not fail isn't terrible.... not compiling in do_install is a little more involved May 26 14:59:22 v2d: git://git@github.com/company/repo.git;protocol=ssh should work May 26 15:08:04 Not to stir the pot but: https://twitter.com/bagder/status/1397433855817170949?s=20 https://twitter.com/jellyfin/status/1397408094691471364?s=20 https://lwn.net/Articles/857252/ May 26 15:08:52 Crofton|cloud: has been chasing people to libera lately :D May 26 15:09:02 Where are we on the topic of IRC server for #yocto? Are we staying here, moving to something else etc.. :) pinging ndec probably? May 26 15:09:30 I think the answer is wait and see what happens.. May 26 15:09:44 Shhhh May 26 15:10:07 Well, something's happening :) May 26 15:10:09 but the channels exist over there May 26 15:10:10 i believe the current trend is that we are moving to libera.chat.. May 26 15:10:23 for both YP and OE channels. May 26 15:10:23 I updated to OE wiki earlier May 26 15:10:46 the only thing that I thought was 'decided' was we don't want multiple channels on multiple IM services or we're all going to go crazy May 26 15:12:01 fray: that train left the station years ago :( May 26 15:14:33 yocton: I get "Host key verification failed." but I have my keys in /etc/skel/.ssh May 26 15:14:35 fray: i think halstead is planning to use bots to duplicate the discussions on old and new channels to ease the transition May 26 15:16:30 Basically, everyone I trust is moving May 26 15:17:15 v2d: bitbake needs to access your ssh-agent. Does that make sense to you? May 26 15:22:59 yocton: kinda. Are you using kas by any chance? May 26 15:23:24 v2d: no May 26 15:25:32 v2d: try "ssh-add" juste before launching the whole process? It might work! May 26 15:25:43 Some announcment on the ML that we are switching IRC servers would be good.... I don't really want to be squating on the channels on 3 different servers, but I don't know where everyone is going May 26 15:26:53 yocton: french I presume? ^^ May 26 15:30:45 v2d: Oui :) (this 'e' on "juste" is in my muscle memory, Could not help it!) May 26 15:35:31 yocton: I do the same ^^ May 26 15:39:52 RP: OK, ztd patch to fix the build in oe-core, MinGW build option bbappend in master-next of meta-mingw May 26 15:40:09 Hopefully upstream takes the patch May 26 15:40:21 JPEW: thanks! May 26 15:43:51 JPEW: yes, we will make an announcement.. but we need to make sure halstead is ready with the infrastructure changes/bot.. and he is a bit busy these days ;) May 26 15:45:16 https://twitter.com/IRCCloud/status/1397475744448663564 May 26 16:02:50 ndec: that sort of talk will get irccloud banned from connecting May 26 16:03:37 if they alienate enough people it'll make switching easier May 26 16:04:13 if yocto/oe move then i'm down to channels i don't care about on freenode May 26 16:08:57 rburton: share the unpackaged file, it seems it did not get named right May 26 16:09:17 yeah just doing a build locally now in fact May 26 16:24:52 so the file should be called /usr/lib/libstdc++.so.6.0.29-gdb.py May 26 16:38:05 what's the ubuntu version recommended to use bitbake? 18.04? May 26 16:38:35 well, what version is crops based on? May 26 16:38:51 crops is based on a slew of distros depending on what one you pick May 26 16:39:00 but the documentation lists the supported distros May 26 16:39:08 a recent LTS is good May 26 16:40:59 rburton, I've created a VM with 20.04 and apparently installed all dependencies but I still cannot run taskexp UI ... 😞 https://bpa.st/G57Q May 26 16:41:46 wyre: using dunfell? May 26 16:41:55 warrior May 26 16:42:33 https://wiki.yoctoproject.org/wiki/Releases <-- that's been EOL since june last year May 26 16:42:44 you need to cherry-pick befe6d2cecacf8d43f2ae0cd697d4479af6fef95 May 26 16:43:02 or use dunfell, I guess May 26 16:43:39 at least May 26 16:44:11 rburton, cherry pick form what repo? 🤔 May 26 16:44:18 that's a poky commit May 26 16:44:20 poky? May 26 16:44:23 oh, I see May 26 16:48:21 rburton, could I just do `git checkout befe6d2cecacf8d43f2ae0cd697d4479af6fef95` in the poky repo? 🤔 May 26 16:50:41 no May 26 16:50:50 that would checkout, you want to cherry-pick May 26 16:53:49 khem: so the problem is that for some reason libname isn't set when installing May 26 16:53:52 libname=`sed -ne "/^library_names=/{s/.*='//;s/'$$//;s/ .*//;p;}" \ May 26 16:54:04 in libstdc++-v3/python/Makefile May 26 16:57:21 rburton: seems to be related to https://github.com/gcc-mirror/gcc/commit/dbb87d6a9c37c8eca80d77782717a8144515c16d May 26 16:58:32 rburton: can you try reverting this for tests ? May 26 16:59:47 and quicky see if it fixes the issue May 26 17:02:24 and do we have the appropriate .la files around when this is generating or are they deleted before do_install May 26 17:03:15 rburton: also share libstdc++.la if you can May 26 17:08:30 ty rburton it was that 😁 May 26 17:08:32 has anyone ever encountered an error that rpm package (creation) never stop for some large package? May 26 17:09:47 like for infinite time ? May 26 17:10:16 dev1990: some can take a very long time, I've not seen it hang May 26 17:10:38 memory/thread usage with xz is bad May 26 17:11:07 well if I just rerun, it will finish in about 1-3mins May 26 17:11:30 dev1990: I've not seen that locally or on the autobuilder May 26 17:12:25 khem: library_names='' says the .la May 26 17:12:29 that would be it :) May 26 17:12:58 khem: this is for the zephyr build May 26 17:16:34 khem: so i expect reverting will solve the problem but looks like there's an underlying issue May 26 17:17:54 khem: this is the .la file that got generated https://www.irccloud.com/pastebin/jTzS4Wk6/ May 26 17:18:59 rburton: why is `library_names` empty for zephyr May 26 17:19:18 rburton: hi -- have you succeed to fetch a private github repository with kas? May 26 17:19:47 and can you tell me the target tuple for zephyr gcc ? May 26 17:19:47 khem: no idea :) May 26 17:20:23 checking host system type... arm-poky-eabi May 26 17:20:32 that tuple? May 26 17:20:41 ok so any baremetal toolchain will have this issue May 26 17:21:02 is .la different beofore and after the patch May 26 17:25:31 building with the gcc bump reverted May 26 17:40:04 khem: library_names is "" with your point release reverted May 26 17:42:54 quick question, whenever I make a build my rpm directories don't seem to contain repomd.xml. I'm on gatesgarth May 26 17:45:01 that file is only generated on demand May 26 17:45:06 (when an image is built) May 26 17:45:13 if you want it to exist, bitbake package-index May 26 17:46:55 interesting, i have been building images and it did not appear. Let me try package-index May 26 17:49:06 yea.. that worked.. surprising it wouldn't work when an image is being built. It just happened on Krogoth. Maybe something I'm doing here. May 26 17:50:52 well, the repomd.xml is put in a temporary directory, not deploy/rpm May 26 17:51:02 when building an image, that is May 26 17:51:46 basically if you want to use deploy/* as a normal feed outside of bitbake, then bitbake package-index first to ensure the feed index is up to date May 26 17:54:04 yea, I was basically sending the entire RPM directory over to a server and noticed that those files were missing. So now steps are build image, run bitbake package-index, then push to server May 26 17:54:13 thanks May 26 17:54:19 renegade: there's a presentation about runtime package management in ten minutes at the yocto summit May 26 17:54:30 hopefully that covers this :) May 26 17:55:21 thanks will check it out. May 26 18:09:14 rburton: there is upstream bug for this https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99453 so I have updated it with our regression on baremetal toolchains May 26 18:09:48 lets revert the patch I pointed to meanwhile, it adds no value to OE as such May 26 18:11:55 renegade: The indexes are places in the image WORKDIR rather than deploy/rpm May 26 18:12:16 renegade: "bitbake package-index" will cause indexes to be built in deploy May 26 18:14:13 was there a reason for this change? I couldn't find anything in documentation. Krogoth seemed to have just built it in deploy/rpm May 26 18:14:28 i've missed a lot of changes due to this huge jump in versions May 26 18:21:56 renegade: packages being added to tmp/deploy while images were being built May 26 18:22:17 renegade: "bitbake core-image-minimal core-image-sato" causes races May 26 18:25:49 thanks May 26 18:55:51 RP: rebasing ptests to core-image-minimal was fairly easy, patches coming May 26 18:56:07 RP: it does take out a whole lot of unneeded stuff May 26 19:01:50 kanavin: I hoped it might, I did do some work on fixing dependencies in the past May 26 19:03:37 RP: this is what I did http://git.yoctoproject.org/cgit.cgi/poky-contrib/log/?h=akanavin/package-version-updates May 26 19:03:53 RP: running x86/aarch ptests on AB now May 26 19:07:12 kanavin: you did find a few more dependencies missing! :) Not sure about the ptest dev dependency insane check change May 26 19:08:22 kanavin: those changes look awesome, thanks for doing them! May 26 19:08:37 RP: python ptests need it May 26 19:08:55 kanavin: I mean that the INSANE_SKIP thing may be the right solution May 26 19:09:47 kanavin: looks good though, its a nice change May 26 19:10:11 RP: but then it would mask some other python package wrongly depending on -dev May 26 19:12:11 kanavin: we've had problems with ptest packages depending on -dev packages when they shouldn't too :/ May 26 19:15:08 kanavin: you can skip this just for the python-ptest package May 26 19:15:18 kanavin: d.getVar('INSANE_SKIP_' + package) May 26 19:32:23 are we there yet May 26 20:17:09 Should taskexp still work? "bitbake -g -u taskexp quilt-native" gets me "Timeout while waiting for a reply from the bitbake server (60s)" May 26 20:17:32 hey randy May 26 20:17:57 *should* work but haven't tried it for ages, and my build machine is headless May 26 20:20:23 rburton: Yeah that's what I was hoping for is that someone could just try it. Someone says it isn't working in the containers, but it doesn't *seem* to be working on my remote workstation either. I guess I could try spinning up a VM locally. May 26 20:21:29 You get a pop-up when running this command so I guess you need to do some X forwarding/VNC or something similar? May 26 20:22:37 qschulz: Correct, but I can see the output from glxgears, but bitbake gives me the timeout. May 26 20:23:22 rewitt: try with hardknott and gatesgarth release branches :) May 26 20:23:41 i can test later when my other laptop is to hand May 26 20:23:51 check if there's isn't a singletask.lock somewhere or something stopping bitbake to actually start? (don't remove it, kill the processes using it first with lsof) May 26 20:27:38 rewitt: worked for me here May 26 20:28:33 rewitt: (with master) May 26 20:29:43 rewitt: when you see that timeout, have a look at bitbake-cookerdaemon.log May 26 20:30:02 RP: It looks like it works after first running "bitbake quilt-native", but starting with a clean build directory and "bitbake quilt-native -g -u taskexp" seems to hang May 26 20:31:11 rewitt: oh, interesting. I was in an existing builddir May 26 20:32:53 rewitt: Confirmed that breaks. Log says Exception in the main loop but doesn't say what May 26 20:33:48 RP: Well at least you got me to poke at it enough to realize where it seems the problem lies. :) May 26 20:34:35 RP: It's not urgent or anything, when I respond to the issue I can just point out the caveats for now. I'll add an issue to bugzilla if you would like May 26 20:36:20 rewitt: that would be good thanks. We should fix this. The log even looks odd, not sure what happens May 26 20:40:48 rewitt: I managed to get logs, its a failure about HOSTTOOLS not being built, i.e. sanity test failing May 26 20:41:53 RP: That's so weird that it would fail the sanity test on a clean build directory but not on an existing one May 26 20:42:05 RP: I wouldn't expect that to impact HOSTTOOLS May 26 20:43:47 rewitt: I think its more that HOSTTOOLS doesn't exist. I don't fully understand what is going on, I'm just happier I have an actual error now May 26 20:44:07 rewitt: it is raising a double exception which is why the original logging failed May 26 20:49:36 RP: https://bugzilla.yoctoproject.org/show_bug.cgi?id=14408 May 26 21:11:38 Does anyone understand how to use devtool? I built an eSDK, and I'm trying to build some packages.. but I can't figure out the commands.. May 26 21:11:57 everything says I need to add to my workspace.. I don't want to edit the packages, I just want to build them from the existing layers May 26 21:12:42 devtool add can't add htem cause they exist.. devtool modify fails complaining about No such file or directory for ..../devtooltmp-.../initial_rev' says devtool-source class isn't workign right May 26 21:42:13 fray: https://pretalx.com/media/yocto-project-summit-2021/submissions/SNL9HJ/resources/yp_summit_2021_paul_devtool_ws_M6_DMWvdg6.pdf May 26 21:48:09 as follow-up to our discussion on layer quality yesterday ... here the stats for master branch: May 26 21:50:21 layers tested known to the layerindex: 162 May 26 21:50:21 layers passing all 3 checks: 13 (!) May 26 21:50:21 layers not passing all 3 checks: 147 May 26 21:50:21 layers with fetch issues ... a lot. (aka repo no longer on github or the like) May 26 21:51:20 thanks dl9pf May 26 21:51:27 this looks a good start **** BEGIN LOGGING AT Wed May 26 22:03:25 2021 May 26 22:30:15 dl9pf: if we did run more layers on the autobuilder, it would mean wider mirroring of sources which could help protect against this kind of issue a bit more... May 26 22:37:19 some of them just vanished, but are entries in the layerindex still ... gotta clean them up May 26 23:03:30 dl9pf: ah, we should just clean those out then May 26 23:09:39 RP: I'm not 100% sure that bitbake patch I just sent is correct... May 26 23:10:23 But it does at least document how to reproduce the problem (which is surprisingly easy) May 26 23:15:13 JPEW: which issue was that for, the one above here? May 26 23:15:41 The one from the hands on (tony's issue :) ) May 26 23:15:54 Where you have to kill -9 bitbake-server to recover :/ May 26 23:17:17 JPEW: ok, I'm confusing two issues :) I didn't hear tony describe the problem he had May 27 00:13:36 rewitt: I have a fix which is to add params.updateToServer(server, os.environ.copy()) to the taskexp UI. It has raised a lot of questions which need other fixes in the code. I updated the bug. May 27 01:06:17 are the yocto summit slides somwhere here? https://pretalx.com/p/about/ May 27 01:06:26 if so i do not see them May 27 01:08:17 nm i see May 27 02:15:19 are the videos going to be available for d/l too? May 27 02:30:12 yates: IIRC they will be on YouTube **** ENDING LOGGING AT Thu May 27 02:59:56 2021