**** BEGIN LOGGING AT Mon Mar 01 02:59:57 2021 Mar 01 07:51:43 good morning Mar 01 08:43:42 I need to add a c flag for a package recipe that make it pass gcc header file checking Mar 01 08:44:05 that package is using cmake and I try to add CFLAGS or TARGET_CFLAGS none of them would work Mar 01 08:47:47 ayaka: look at the cmakelists.txt if it actually honours passed flags. if they are completely overriden... Mar 01 08:48:11 LetoThe2nd, yes it does, but just bitbake -e won't show those variable Mar 01 08:51:48 ayaka: then you might be doing something wrong, like not appending correctly or such. Mar 01 08:52:19 ok, then I would try _append instead of += Mar 01 08:53:06 ayaka: it depends, really. if nothign is showing up in bitbake -e at all (it should give you the complete evaluation history per variable) then the error might be something else, too. Mar 01 08:54:32 OK, I think I figure out why, I have two recipe with the same PN and PV in the a different layer Mar 01 08:54:59 then I modify the lower one but not the one in the upper layer Mar 01 08:55:40 :) Mar 01 10:05:04 hey guys, please how can i add systemd without networkd and make connman automatically connect to ethernet Mar 01 10:05:38 as i checked both connman and networkd cannot coexist Mar 01 10:21:44 @JPEW: good morning Mar 01 10:23:08 JPEW:  it at all possible to launch the pyrex driven build from a bash script? Mar 01 10:26:00 @jpew: Problem wiith sourcing pyrex-init-build-env in a bash script results in: ERROR: /tmp/test/poky-amd/oe-init-build-env not bound in container (File doesn't exist). Mar 01 10:26:22 @jpew: when the file actually exists Mar 01 10:32:05 intera91: jpew is usian, don't expect an answer in the upcoming 4 or 5 hours. Mar 01 10:32:40 medaliyou: remove networkd from systemd's packageconfig? networkd is in there with a weak default (??=) through poky/meta/recipes-core/systemd. Mar 01 10:33:58 LetoThe2nd, and thank you Mar 01 10:34:51 not from packag config, but i need systemd support but without networkd Mar 01 10:34:56 i want to use connman in place Mar 01 10:35:10 PACKAGECONFIG[networkd] = "-Dnetworkd=true,-Dnetworkd=false" Mar 01 10:35:28 i ddin't know how to write in local.conf Mar 01 10:35:31 LetoThe2nd: Thank you ;-) Mar 01 10:36:40 ls Mar 01 10:37:09 ayaka: have fun. Mar 01 10:37:37 medaliyou: either write a bbappend for systemd, or PACKAGECONFIG_remove_pn-systemd = "networkd" in local.conf Mar 01 10:38:20 madaliyou: https://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#var-PACKAGECONFIG last Paragraph is about changing the value via a configuration file Mar 01 10:38:31 rburton: blergh! Mar 01 10:38:37 "in your distro conf" Mar 01 10:38:40 thank you rburton and dwagenk Mar 01 10:39:18 you're welcome. Mar 01 10:56:17 Down to 9 non-reproducible package exclusions in the reproducibility test Mar 01 10:56:25 (ignoring go) Mar 01 11:02:18 Anyone who understands ruby able to look at the ruby does one? Mar 01 12:11:07 any ideas why i could be seeing a lot of D states in a container when starting processes? Mar 01 12:14:22 e.g. even a git clone or such blocks for a couple of seconds Mar 01 12:29:16 RP: do we enable wayland DISTRO_FEATURE by default also in the selftests ? Mar 01 12:39:32 dl9pf: depend whether poky does Mar 01 12:40:25 POKY_DEFAULT_DISTRO_FEATURES includes it Mar 01 12:40:51 dl9pf: so yes Mar 01 12:41:34 ok, I've a change to meta-selftest/files/static-[group,passwd] then to add the wayland groups Mar 01 12:41:47 dl9pf: fair enough Mar 01 12:41:59 shows up of you add core-image-weston to the repro build for example Mar 01 12:42:53 dl9pf: right, makes sense. I think the test only checks the packages, not the image fwiw Mar 01 12:43:30 hm. building latest dunfell qemuarm core-image-minimal with cold tmp cold sstate in 19 minutes. ok? Mar 01 12:43:48 LetoThe2nd: seems ok to me Mar 01 12:44:06 LetoThe2nd: image-sato is about an hour on the performance benchmark machines Mar 01 12:44:16 RP: cold tmp, cold sstate? Mar 01 12:44:24 LetoThe2nd: yes Mar 01 12:44:26 but hot dl Mar 01 12:44:31 right Mar 01 12:44:32 ok, lemme give it a spin. Mar 01 12:45:22 RP: ipk and qemuarm are fine? or something else for better comparability (is that even a word?) Mar 01 12:46:27 LetoThe2nd: https://autobuilder.yoctoproject.org/typhoon/#/builders/91/builds/4238/steps/11/logs/stdio Mar 01 12:46:39 LetoThe2nd: that is the config we benchmark Mar 01 12:47:05 ok, thx Mar 01 12:51:56 kicked off, now lets see :) Mar 01 12:53:11 also gave it INHERIT "testimage report-error" Mar 01 13:09:34 RP: seen this one before ? Mar 01 13:09:38 https://www.irccloud.com/pastebin/IAIOm91U/ Mar 01 13:20:50 dl9pf: hmm, no Mar 01 13:21:59 ok, will watch for it and file a bug Mar 01 13:27:34 RP: wooohoooo! https://pastebin.com/dFidmwTM Mar 01 13:28:27 LetoThe2nd: nice :) Mar 01 13:29:01 i think it qualifies as fast. :) Mar 01 13:29:09 RP: quick question. When I try to re-run the reproducibility test (and hopefully reuse the sstate for a faster build), it tells me that build-st already exists and exits. But when I look in that directory, I see and sstate dir. If I remove it, won't I toss out the hot sstate ? Mar 01 13:29:45 and yah, when I removed perf from the exclusions. I see it failing now, so can poke at it. Mar 01 13:49:49 @jpew: are you there sir? and good morning by the way Mar 01 14:04:51 zeddii: there is a keep build dir option to oe-selftest Mar 01 14:19:35 Anyone an idea which recipe provides a full-fledged readlink? Mar 01 14:19:35 resolvconf uses readlink -m but the busybox readlink does not support -m Mar 01 14:20:26 Hi Everyone, I ran into this while trying to run my yocto build, I wonder what could be the problem Mar 01 14:20:38 https://www.irccloud.com/pastebin/VEWcrKQW/ Mar 01 14:21:49 what I've tried `rm -rf /home/dorinda/Poky/build/tmp/work/qemux86_64-poky-linux/core-image-minimal/1.0-r0/rootfs` but still same error Mar 01 14:22:48 RP: as in '-K' to the selftest ? yah. I tried that, but it still doesn't like my build dir. Maybe its an interaction of the new patch you did (and that I applied to my local branch for sstate re-use in testing perf). Mar 01 14:24:49 zeddii: I think -K not working is a genuine bug. Please do hack around it (or fix it). Also, setting SSTATE_DIR somewhere before running oe-selftest will probably work Mar 01 14:25:45 dorinda: bitbake core-image-minimal -c clean will probably fix it. Before you do, what does the end of WORKDIR/pseudo/pseudo.log say? Mar 01 14:26:59 acked. I'll dig a bit into it now. Mar 01 14:28:56 zeddii: I think with my two pending patches in -next, we're down to a list of 6 repro issues outside of go, perf being one. no pressure ;-) Mar 01 14:29:04 :D Mar 01 14:31:57 zeddii: is anyone already working on golang breakage in meta-virtualization? the docker-moby and containerd-opencontainers failing to build? Mar 01 14:32:23 RP: Alright thank you, trying that now. the end of pseudo.log says Mar 01 14:32:30 https://www.irccloud.com/pastebin/YcbzdPAw/ Mar 01 14:32:35 they have shown up in my world builds after go was upgraded on friday Mar 01 14:33:15 JaMa: I haven't done an oe-core update and rebuild yet, but will trigger one today and look into it. Mar 01 14:33:31 zeddii: thanks! :0 Mar 01 14:34:10 dorinda: what does the rest of it say? That is odd but may be caused by the deletion of rootfs you did :/ Mar 01 14:35:43 https://www.irccloud.com/pastebin/9fvDn5Fh/ Mar 01 14:36:31 the rest....and the error is gone after bitbake clean thanks! :) Mar 01 14:46:43 rburton: you'd be a good person to help fix the ovmf reproducibity issue, right? Mar 01 14:47:37 intera91: I'm here. That should work, you may need to adjust what's getting bound into the container Mar 01 14:49:37 dorinda: thanks. I believe the latest patches on master should fix that (related to PSEUDO_IGNORE_PATHS) Mar 01 14:54:10 RP: maybe. is there a bug yet? Mar 01 14:55:21 rburton: no, https://autobuilder.yocto.io/pub/repro-fail/2021-2-25-rp/ovmf-shell-efi/diff/ Mar 01 14:55:31 rburton: one stray path getting in... Mar 01 14:55:59 RP: you mean the reproducibility issues fix? Alright. Mar 01 14:56:31 dorinda: no, I mean http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=6066d67c216a3f475b5df262c9c89967a02e7834 Mar 01 14:56:51 dorinda: your failure mentioned oe-rootfs-repo Mar 01 14:57:09 rburton: I'll admit I don't know a thing about what is going on in that build, I did look :/ Mar 01 14:59:10 RP: please let me know your view before eventually resending the series: https://lists.openembedded.org/g/openembedded-core/message/148708 Mar 01 14:59:34 RP: the edk build is a horror show Mar 01 15:00:10 RP: oh okay I see, Thanks. Mar 01 15:00:32 rburton: yes, I'm backing away carefully :/ Mar 01 15:01:13 vdl: I was unclear how broken it is without the --numeric-owner option :/ Mar 01 15:01:38 vdl: if it doesn't work at all it is an RDEPEND. I'm unclear on how much it works (or not) Mar 01 15:03:26 RP: systemd-container include /sbin/machinectl. machinectl has many subcommands, which work, except "import-tar" which requires a version of tar with --numeric-owner support. So the package isn't broken per-se, but one subcommand from one tool is, so it's a grey area for me. Mar 01 15:05:21 vdl: ok. that it is a RRECOMMENDS Mar 01 15:05:59 ok Mar 01 15:07:15 RP: more specifically, support for machinectl import-* subcommands require the "importd" PACKAGECONFIG entry. Mar 01 15:08:10 RP: so the series apply cleanly on latest master (systemd recipe renaming is detected without problem). Do you still prefer me to resend? Mar 01 15:09:36 @jpew: hiya, it works fine the first time, but trying to reproduce build in anbother directory and you end up with : ERROR: /home/patrick/workspace/test/poky-amd/oe-init-build-env not bound in container (File doesn't exist). Mar 01 15:10:22 Out of curiosity, can a PACKAGECONFIG entry pull in a recommended package? Mar 01 15:10:23 @jpew: even after docker image rm and docker container prune Mar 01 15:10:34 vdl: please as its difficult for me to start forcing patches to apply Mar 01 15:11:04 intera91: You shouldn't need to remove the image... can send me a log the the commands you're trying to run? Mar 01 15:11:12 @jpew: retry in the original directory and it works, so the original bindng is stored somewhere Mar 01 15:11:12 intera91: and your pyrex.ini Mar 01 15:12:21 @jpews: always  the orignal as built by : meta-pyrex/mkconfig > ./pyrex.ini Mar 01 15:13:03 @jpew really the question is where is the original binding(s) stored? Mar 01 15:13:23 pardon my ignorance but what is nowadays the proper way to submit a fix (in my case in meta-connectivity, resolvconf package) Mar 01 15:15:55 intera91: The binds aren't really "stored" anywhere; they parsed from pyrex.ini and the environment everytime the container starts Mar 01 15:22:08 eFfeM: http://cgit.openembedded.org/openembedded-core/tree/README.OE-Core Mar 01 15:24:19 qschulz: thanks Mar 01 15:31:06 @jpew: ok I just have to rewrite the environmrnt variable as the bind section of th epyrex ini is empty Mar 01 15:31:15 @jpew: thanks ;-) Mar 01 15:34:00 intera91: Ya there are two variables you can use: PYREX_BIND is the "end user" variable for the user to bind in whatever they want (e.g. sstate, downloads, extra layers, etc.). Mar 01 15:41:39 intera91: Then there is PYREX_CONFIG_BIND that you can set at a "project" level (e.g. if you have a custom layout or init script) Mar 01 15:44:23 JPEW: here is the bash script to launch the build from nothing: https://pastebin.com/hunKKGHm Mar 01 15:52:55 RP: I'm starting to look at swatbot for file new bugs, https://swatbot.yoctoproject.org/mainindex/failed/ Mar 01 16:04:09 intera91: That's the one that works? FWIW, you shouldn't do a one-liner to set PYREXCONFFILE and PYREX_CONFIG_BIND when sourcing (see https://github.com/garmin/pyrex#setup) Mar 01 16:05:26 alimon: you want https://swatbot.yoctoproject.org/mainindex/swat/ Mar 01 16:06:13 RP: my guess is that ones that already has owner are already done? Mar 01 16:06:44 RP: like for example 168849 needs some debug Mar 01 16:10:56 Hello All i have a problem including lxc to my mage using meta-virtuaization Mar 01 16:11:10 IMAGE_INSTALL_append = " network-systemd \ Mar 01 16:11:11  kernel-module-bridge \ Mar 01 16:11:11  lxc \ Mar 01 16:11:12  lxc-networking \ Mar 01 16:11:12  lxc-setup \ Mar 01 16:11:13  lxc-templates \ Mar 01 16:11:13     " Mar 01 16:11:31 the following error raises Mar 01 16:11:32 Failed: Mar 01 16:11:33   lxc-setup.aarch64 2.0.8-r0 Mar 01 16:11:33 Error: Transaction failed Mar 01 16:11:39 RP: Have we done anything (or plan to do) in regards to having the build layers reuse compliant? https://reuse.software/ Is this something we are interested in at least? Mar 01 16:12:13 alimon: abelloni can probably help with the current status. I think the ones listed as Pending still need to be looked at to determine what to do with them (new bugs, add to existing bug or report on mailing list) Mar 01 16:36:34 rburton: https://bugzilla.yoctoproject.org/show_bug.cgi?id=14264 filed so we now have a bug too Mar 01 16:38:18 alephan: sounds reasonable. We've been trying to switch to license identifiers. Copyright identifiers are newer and we've not looked at that yet Mar 01 16:42:48 alephan: That's interesting. If that was ubiquitous, it might be able to reduce a lot of OE's license tracking code. It seems to be solving a similar problem Mar 01 16:43:35 It might not tackle the same problem though JPEW Mar 01 16:43:47 This is more of the metadata license (and patches) Mar 01 16:44:30 RP: It shouldn't be very complicated to do but it would add contributions requirements. Mar 01 16:44:55 For example each recipe will have to have its correct header and there are discussions about how to handle patches for example. Mar 01 16:44:56 alephan: Ya, I realized that :) Mar 01 16:46:12 khem: do you happen to run the ptests for lua in your CI ? I do lua-ptest crashing with a general protection fault here (dunfell 3.1.5 + meta-oe) Mar 01 16:46:49 thats lua 5.3.6 Mar 01 16:47:39 JPEW: That being said though, if we have packages reuse compliant it would definitely help to autonate the OE license tracking code Mar 01 16:47:40 automate* Mar 01 16:47:41 It would actually make it really easy Mar 01 16:48:29 You can also compute the spdx file directly this way Mar 01 16:48:59 (spdx bom) Mar 01 16:49:34 alephan: I certainly like the idea of improving things, the exact best way to do that is tricky. I know how much work just adding license identifiers was Mar 01 16:49:59 This is actually not hard as well - there is tooling that does it automatically Mar 01 16:51:13 So for the bulk of stuff is not hard (like recipes, distros, layer confs etc). The discussion needs to be held around patches Mar 01 16:52:36 Is it reasonable to start requiring a .license counterpart with the headers for each patch? Should we embedded the license metadata as part of each patch? Stuff like that. But once we have we can include CI that lints the players (`reuse lint`) and never worry about it. Mar 01 16:53:32 I'm going to try it out on an internal layer - and see how it looks. I was interested to check the general feeling of it, Mar 01 16:53:46 which looks positive. Mar 01 17:06:28 RP: I've resent, I didn't Cc you, maybe I should've, sorry. Mar 01 17:10:09 vdl: its fine, thanks! Mar 01 17:10:35 alephan: the hard part is knowing who has the copyright Mar 01 17:13:16 RP: That is indeed true. And I need to get some legal knowledge there - I'm planning to have a discussion with our legal team soon. For existing shared stage (like oe-core) that will be indeed a challenge. Mar 01 17:27:21 khem: would I be correct in my guess that updating mozjs is quite non-trivial? Mar 01 18:39:19 JPEW: ltp is so bad at reproducibility, diffoscope can't cope :( Mar 01 18:40:47 RP: oof Mar 01 18:41:52 Hmm, I added resolvconf to my image but now when building I get: Mar 01 18:41:52 packagegroup-core-boot : Depends: systemd but it is not going to be installed Mar 01 18:41:52 If I remove resolvconf from the image recipe it builds. Mar 01 18:42:05 anyone an idea what is wrong? Mar 01 18:42:17 JPEW: I've ended up with meld to try and get handle on where to start Mar 01 18:43:21 (Oh and I have several other packages using systemd, (but maybe not during boot) Mar 01 19:19:58 RP: I'm a little suprised the diffoscope per-file diff limit didn't kick in Mar 01 19:20:29 RP: Maybe that's per file in the package, and there are a lot of package files? Mar 01 19:33:58 JPEW: there are a lot of files Mar 01 20:47:27 JPEW: I'm starting to wonder if this explains why ltp is so unreliable on the autobuilder Mar 01 22:31:29 * RP has filed bugs for all the remaining reproducible issues Mar 01 22:45:53 abelloni: hi, i reviewed the AB job failures today and only found a new issue in meta-mingw regarding a bbappend, any other issue you know that needs some debug/care? Mar 01 22:47:16 alimon: sure, swatbot still shows many: https://swatbot.yoctoproject.org/mainindex/swat/ Mar 01 22:47:32 I'll handle anything older than 168578 Mar 01 22:48:51 alimon: the bbappend has a fix on the list, I forgot to include the master-next branch of mingw for testing, sorry Mar 01 22:49:28 I did try and clean up a couple of disastrous builds from bad patches on the list Mar 01 22:52:49 alimon: Ah, I see you did update some bugs but you didn't update the status on swatbot, did you properly receive the credentials ? Mar 01 23:33:37 abelloni: nop, i don't have credentials Mar 01 23:38:10 Given the log output here, should I see "u-boot" appear in those few lines if my system was using u-boot? (it's a pocketbeagle): https://pastebin.com/wQ7QLqyn Mar 01 23:39:51 I am new to using u-boot (and go yocto in general), but my machine conf does mention u-boot: https://github.com/JonasVautherin/meta-pocketbeagle/blob/master/conf/machine/pocketbeagle.conf#L29-L37. I'm just not completely sure how to interact with it (and it starts with verifying that it is actually used at boot time) Mar 01 23:42:56 For instance, I have been setting the bootdelay using `bitbake -c menuconfig u-boot`, hoping to get a few seconds to start into u-boot, but it does not seem to wait anywhere. I have `/boot/u-boot.img` on my rootfs, but that does not say that it does load through u-boot Mar 01 23:45:26 also, I'm based on core-image-minimal, which I am not sure is a good idea (e.g. it's ignoring DISTRO_FEATURES and stuff like that, which was a bit confusing at first. Not sure if there is an influence on u-boot) Mar 02 00:05:53 jonesv[m]: that's definitely using u-boot Mar 02 00:06:58 jonesv[m]: I'd expect it to print the U-Boot version header on start, though Mar 02 00:09:37 smurray: what tells you it's u-boot, then? Mar 02 00:10:12 smurray: also, how could I get into the u-boot prompt from there? I am connected over serial, is there a key to press at some point? Mar 02 00:10:17 jonesv[m]: those messages about loading things match what u-boot would print Mar 02 00:11:33 jonesv[m]: usually you'd see a prompt that would let you break in, unless it has been configured off Mar 02 00:12:53 smurray: yeah so I tried to set CONFIG_BOOTDELAY=2, but it does not seem to stop and wait Mar 02 00:13:13 jonesv[m]: tlwoerner that did a bunch of the work on that pocketbeagle layer hangs out in here, he might have some ideas Mar 02 00:14:09 jonesv[m]: and iirc, moto-timo that's here also has a layer for it somewhere Mar 02 00:15:36 Right. Would be glad to get insights from tlwoerner_ or moto-timo then 🙂. I suspect tlwoerner_ is not super active, at least not on github (I opened a PR for gatesgarth but never got any answer) Mar 02 00:16:15 I'm trying to disable CONFIG_AUTOBOOT, but at this point I'm really guessing. I haven't found much u-boot documentation 😕 Mar 02 00:18:06 jonesv[m]: there's also #u-boot Mar 02 00:18:13 Yeah nothing I do in `bitbake -c menuconfig u-boot` seem to have any effect Mar 02 00:18:38 oh I was not aware of #u-boot, thanks! Mar 02 00:20:58 jonesv[m]: I've not tried that menuconfig task in quite a while, I know it did work in the past Mar 02 00:23:29 smurray: yeah it does change the .config, and for the kernel it's working. I just don't really know what to set to achieve what I want, I guess... Mar 02 00:24:36 I would like to enable fastboot and have a way to boot in a fastboot mode, and then flash my image through that (if that's possible at all) 🙂 Mar 02 00:25:20 Again, for learning. On the pocketbeagle it's easy enough to remove the sdcard and flash it directly 😆 Mar 02 01:29:33 jonesv[m] I haven't touched u-boot or pocketbeagle since the original e-ale: https://github.com/e-ale/yocto-e-ale Mar 02 01:30:33 https://github.com/e-ale/Slides/tree/master/yocto-images Mar 02 01:33:51 Yeah I think I had read that already, but that was not helping me for u-boot :( **** ENDING LOGGING AT Tue Mar 02 02:59:58 2021