**** BEGIN LOGGING AT Mon Jan 20 02:59:57 2020 Jan 20 06:16:45 New news from stackoverflow: How to upgrade opkg packages from 2014.12 to 2018.12 when postinst script returns status 1 and opkg fails? Jan 20 06:26:08 The python2 removal commit really got to me :( Jan 20 06:39:07 Is anyone successfully using meta-java and openjre-8 in either warrior, zeus or master ? Jan 20 07:17:07 is it possible to apply bbappend somehow conditionally? If package X, depending on Y, is excluded from image, don't apply Y.bbappend Jan 20 07:20:38 stuom1, sounds problematic to change behavior of a recipe build depending on whether a package is included in an image or not Jan 20 07:21:08 since a package can be installed later on Jan 20 07:24:08 what about based on machine type? I tried to add COMPATIBLE_MACHINE but it of course just fails to build Jan 20 07:24:43 is only way to manually delete the bbappend when i dont want it, sounds tedious Jan 20 07:25:06 stuom1, there is a mask you can play with Jan 20 07:25:44 BBMASK Jan 20 07:29:11 ok, I'll look into it, thanks Jan 20 07:53:10 good morning Jan 20 07:55:56 hi. Jan 20 08:47:13 New news from stackoverflow: glibc is failing while doing bitbake[SUMO version] Jan 20 09:20:13 Hi guys, my board has support for sumo branch only. Does it make sense to build a product out of it, even if sumo is out of official support? Jan 20 09:20:40 Hi all, I'm trying to add an entry to https://wiki.yoctoproject.org/wiki/Project_Users but I want to be sure to know where to put our company name. What does "OSV" stand for? Non-native here, have never seen that one before and a quick googling => /me shrug Jan 20 09:21:04 OSV = operating system vendor Jan 20 09:21:28 kriive: thanks, will add that to the wiki :) Jan 20 09:23:09 kriive: it makes sense to forward port the BSP Jan 20 09:23:56 Should I do it by myself? Is it feasible to port to a newer Yocto release? Jan 20 09:24:02 port it* Jan 20 09:24:34 kriive: it depends a bit, but usually a BSP needs little more than bootloader and kernel to be functional. and this is really no magic Jan 20 09:25:23 kriive: unless its super custom HW, jsut name the board respectively show the BSP layer, then we can make proper statements. Jan 20 09:27:11 It's the up-board, https://github.com/emutex/meta-up-board Jan 20 09:29:27 kriive: looks highly unproblematic Jan 20 09:30:00 kriive: maybe the firmware needs a bit of fine-tuning, but getting the board up and running shouldn't be much of a problem on warrior/zeus Jan 20 09:32:10 LetoThe2nd: thank you, forking it right now Jan 20 09:34:48 kriive: I'm using the meta-intel layer and the 'intel-corei7-64' machine to support the Up Squared I've got here Jan 20 09:35:14 It should also work with 'genericx86' and 'genericx86-64' from meta-yocto-bsp Jan 20 09:36:22 I'm wondering if that layer is needed Jan 20 09:36:47 paulbarker: "special sauce" Jan 20 09:37:01 Yeah, I guess it's needed to have the GPIOs and onboard LEDs working Jan 20 09:37:15 Ah ok Jan 20 09:37:43 Unfortunately I need those three little blinky bastards c: Jan 20 09:37:52 Looks like it just adds a few things on top of the BSP provided by meta-intel. The readme could be a bit clearer Jan 20 09:38:15 "little blinky bastards" - I'm stealing that Jan 20 09:38:45 * LetoThe2nd blinks Jan 20 09:39:31 paulbarker: haha it's all yours Jan 20 09:40:27 LetoThe2nd any updates on the GitHub Repository with the Yocto Cheatsheet? Jan 20 09:41:36 shan1: not yet, sorry. i have to admit i totally forgot about it. Jan 20 09:42:49 also your personal `meta-heavymetal` repo can be of much help if you pushed some code onto it? Jan 20 09:43:56 shan1: it can be once i get around to push stuff, yes. unfortunately i'm the kind of guy who habitually tinkers a lot of stuff simultaneously and only finishes urgent things. Jan 20 09:44:07 like the preparations for todays session :) Jan 20 09:44:40 :D Like a true engineer! Jan 20 09:46:00 shan1: even more, my spare allocatable time does not seem to improve soon. Jan 20 09:46:29 Sorry for the small spamming on the Wiki, was hesitating between ways to present the company's solutions and products :/ Jan 20 09:54:03 LetoThe2nd: thanks for the heads-up with meta-gplv2 Jan 20 09:55:25 qschulz: :) Jan 20 09:55:55 i feel very influence-ish lately. Jan 20 10:04:51 LetoThe2nd: you can't call yourself an influencer if you don't have an instagram. Work on that first if not already ;) Jan 20 10:09:10 qschulz: i've been told that nowadays i need somethink called "tuk-tuk", or simliar? Jan 20 10:11:52 tik-tok, I think we might be already a little bit above the age target though :) Jan 20 10:11:55 List of Influences for Influencers: Instagram, YouTube, Tik Tok and a whole lot of Narcissism. LetoThe2nd you always forget to say "Like Share and Subscribe to the Yocto Project" at the end of the twitch sessions Jan 20 10:12:19 shan1: he's a beginner don't judge Jan 20 10:13:20 i think my narcissm levels are ok. after all, god emperor the known universe. Jan 20 10:14:11 -> https://en.wikipedia.org/wiki/Leto_II_Atreides#God_Emperor_of_Dune Jan 20 10:15:17 :D Jan 20 10:17:33 New news from stackoverflow: How should the sstate-cache directory be deleted in Yocto? Jan 20 10:26:03 Where can we report issues with yocti? It's digging up old issues again which leads to me just ignoring it and defeats the purpose Jan 20 10:27:59 paulbarker: actually its not really a bug, its usually that somebody changes the formatting or such on SO Jan 20 10:29:33 but i agree, it would be good if we could filter that out. yet its probably non-trivial to distinguish unimportant formatting edits from those actually modifying the content Jan 20 10:29:37 Yes but it means yocti is just spamming the channel. I don't bother checking the links often as they're most often old questions which have been edited slightly and were already answered Jan 20 10:30:27 It just needs to ignore any questions over ~3 months old. A change on a 2 month old question may still be relevant but a 2 year old question isn't Jan 20 10:30:49 paulbarker: good point. Jan 20 10:31:01 paulbarker: maybe poke mr.admin in the afternoon? Jan 20 10:31:09 Will do Jan 20 10:31:24 \o/ Jan 20 11:04:51 Guys, are there are any important differences between linux-yocto and linux-intel? Jan 20 11:07:25 Because I see that linux-yocto dropped 4.14 in 2.7, but I still found 4.14 in linux-intel and I was wondering if it's the same Jan 20 11:09:17 kriive: we already used and supported such board Jan 20 11:09:53 kriive: you should be able to get the proper support from Up community forum Jan 20 11:11:59 Yeah, but Up board support is frozen to Yocto 2.5.3 and I would like to have the latest patches available Jan 20 11:13:04 Btw they are pretty unresponsive lol. But I use this opportunity to learn a bit more about Yocto internals! Win-win c: Jan 20 11:14:02 kriive: Unless you need any out-of-tree kernel modules which only support 4.14 you should be able to use a later kernel Jan 20 11:15:00 kriive: linux-intel is the kernel provided by meta-intel. its based on the Intel LTS kernel, which is a fork of mainline. if you're using a x86 then use linux-intel for sure as its full of IA-fixes and improvements Jan 20 11:16:00 Thank you very very much guys! Jan 20 11:16:00 (https://github.com/intel/linux-intel-lts) Jan 20 11:17:44 New news from stackoverflow: Use llvmpipe instead of softpipe in yocto Jan 20 11:47:51 New news from stackoverflow: yocto-bsp command not found Jan 20 11:50:52 does "bitbake -g" output in recipe-depends.dot list full recipe dependencies for each recipe? Also transient ones where x depends on y depends on z depends on ö? Jan 20 12:17:46 mcfrisk: i think bitbake -g does only output the full expanded task-depends these days. for everything else, use buildhistory. Jan 20 12:31:26 I'm little confused about KERNEL_FEATURES, how is it respected? In manual I found how to turn on sound on qemux86 and other examples. Where can I search for full list of respected options or I just missing the point? Jan 20 12:32:59 dev1990: https://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#var-KERNEL_FEATURES Jan 20 12:33:22 dev1990: basically this means "if you use a yocto-style kernel, it should work. otherwise, not." Jan 20 12:53:42 LetoThe2nd: yep, it looks very much like expanded dependency list and that's just what I need.. Jan 20 13:09:42 LetoThe2nd: thanks, I found that before, but I'm not sure how to take adventage of it. For example statement "cfg/sound" will enable every module that is in kernel-source-dir/sound/ ? or it just expand to CONFIG_SND=y? or it will be searching for sound.cfg kernel fragment? Jan 20 13:10:29 dev1990: it will enable whats stated in cfg/sound.scc file. Jan 20 13:12:02 but I get empty results with: find poky/ -iname 'sound.scc' Jan 20 13:16:31 I checked master branch and zeus branch, sound.scc is not to be found there Jan 20 13:18:23 dev1990: https://git.yoctoproject.org/cgit/cgit.cgi/yocto-kernel-cache/tree/cfg Jan 20 13:27:56 LetoThe2nd: ah ok, now "yocto-style kernel" makes to me more sense :) Jan 20 13:27:58 thx Jan 20 13:34:37 RP: I’m not quite sure the best way to go with my 5.4 series. I fixed a couple of problems that showed up on my first AB run, and triggered it again last night. mips64 wont’ boot under sysvinit (I’ve been testing under systemd and it works fine for minimal/sato). and qemux64-musl is throwing protection faults on some commands. everything else was green. Jan 20 13:35:00 I have to put this down for the next few days, so I’m not sure if it would be best for me to send it, or to just let it sit longer. Jan 20 13:36:36 zeddii: Best to let it sit I think, we can't afford red builds Jan 20 13:36:53 unless we find someone who know mips Jan 20 13:37:01 I don’t think I can fix that problem. Jan 20 13:37:11 so to be completely honest, it’s after FOSDEM Jan 20 13:37:36 and x86 musl gpf. I have zero idea what to do about that. Jan 20 13:37:41 hi there, i am currently learning some basics about yocto and use the bootlin documents for that. so far the material is very helpful (so if anyone from bootlin is here: thx guys :) Jan 20 13:37:54 so I’m mainly wondering about how I can get some people who care about those arches to help. Jan 20 13:38:17 zeddii: khem has hacked on those things, it may be worth asking him Jan 20 13:38:24 zeddii: approach that usually works: "throwing money" Jan 20 13:38:36 zeddii: not sure what else we can do, we're struggling for help in various areas atm Jan 20 13:38:50 LetoThe2nd: who's money? :) Jan 20 13:38:59 i added a new recipe to the recipe-etended folder within poky (i know that's not the clean way... just for testing) which can be built, but for some reason fails to be included within a core-image-minimal Jan 20 13:39:02 yah. I’ve pushed all my other things last week, and I need to do them this week, or I’ll be the one looking for $$ and a job Jan 20 13:39:19 is there anything necessary besides adding it to IMAGE_INSTALL_append in my local.conf? Jan 20 13:39:33 RP: do i care, as long as its not mine? Jan 20 13:39:41 zeddii: I understand. Do we want to add what you have but not make it the default? Jan 20 13:39:50 the mips64 issue on sysvinit is related to the clock stuff I pinged mainline about. hwclock seems to be hanging the boot. Jan 20 13:39:59 zeddii: that way people could reproduce more easily Jan 20 13:40:03 but systemd can deal with it. Jan 20 13:40:13 creich: for quick tests, its CORE_IMAGE_EXTRA_INSTALL Jan 20 13:40:28 LetoThe2nd: thank you! Jan 20 13:40:44 RP: that’s possible. I’ll see if I can detangle the parts, I can’t say 100% it isn’t the libc-headers, but I’ll spend some background cycles today trying to sort that out. Jan 20 13:41:01 creich: but if you already know that both the recipe placement AND the inclusion mechanisms are hacky and wrong, why do you stick to them? a layer setup costs you roughly 2 minutes. Jan 20 13:41:01 zeddii: happy to run a build to test the headers in isolation Jan 20 13:42:08 LetoThe2nd: just walking through the course material.. thought it might bring some insights.. so to be honest, not sure ^^ Jan 20 13:42:54 just wanted to understand it. and since that _append worked for other packages i thought there might be another 'hidden mechanism' but i guess it's only some typo somewhere on my side Jan 20 13:43:21 maybe i just skip that wrong path and continue with the layer stuff, since that's the way to go anyway :) Jan 20 13:43:27 you're right :) Jan 20 13:43:35 creich: do you happen to know our outright *AWESOME* and *EXCELLENT* live coding sessions? https://www.youtube.com/playlist?list=PLD4M5FoHz-TxMfBFrDKfIS_GLY25Qsfyj Jan 20 13:44:07 creich: #2 is just exactly about what you are learning at the moemnt. Jan 20 13:46:08 LetoThe2nd: nope, i didn't know them. i'll have a look. Thank you again! Jan 20 13:47:47 :) Jan 20 13:54:31 creich: I taught that one training (ex-employee). If you've got questions, you can ask here. It's an introduction so anybody here should be able to answer you if I'm not here anyway. Jan 20 13:55:12 creich: quick and dumb question: you didn't forget to add a space at the beginning of you string in the content of _append right? (otherwise, what LetoThe2nd said ;) ) Jan 20 13:55:48 creich: and remember put as little as possible in conf/local.conf. So fine for debugging but you want most in an image recipe/machine conf/distro conf/SW recipe/layer, etc. :) Jan 20 14:02:41 Is it important to drink beer during live coding session? Jan 20 14:03:44 dev1990: it is. Jan 20 14:05:52 dev1990: but if you don't, you're welcome anyway. Jan 20 14:06:19 qschulz: are you sure? Jan 20 14:06:26 * LetoThe2nd gives the undecided look. Jan 20 14:07:49 LetoThe2nd: are you actually starting a sect? Jan 20 14:11:24 :) Jan 20 14:15:09 qschulz: probably :) Jan 20 14:19:52 LetoThe2nd: Anyway, thanks for great content :-) Jan 20 14:20:15 dev1990: YW. Jan 20 14:20:23 sadly I cannot drink since I broke my leg (well probably I'll do small exception in near feature) Jan 20 14:20:41 as usual, remember to follow, everybody, and join us later. next session is in about 100minutes. Jan 20 14:20:55 but I can spend more time with hobby project and yocto :) Jan 20 14:51:04 zeddii: want me to try a cherry-pick of just the headers? Jan 20 14:52:29 sounds good. that'll help me with the 32bit x86 case and narrowing down where it may be. Jan 20 14:53:19 zeddii: running Jan 20 14:59:48 Hello, really basic question : I have 2 custom boards and one application (inherit autotools) linked to opencv. For one board I have to use opencv2, opencv4 for the other one. How can I write a recipe to pass correct include path ( /usr/include/opencv4/opencv2/core/core.hpp vs /usr/include/opencv2/core/core.hpp) ? Or I have to manage that at Jan 20 14:59:48 Autotools level ? Thx. Jan 20 15:01:27 PinkSnake: autotools level Jan 20 15:01:35 pkg-config or whatever opencv supports. Jan 20 15:01:50 LetoThe2nd Ok thx ;) Jan 20 15:02:56 LetoThe2nd Yocto will be smart to choose the correct lib to install on sysroot ? Jan 20 15:04:12 PinkSnake: it will install whatever you tell it to. usually different boards run different MACHINEs at least, so you could easily switch via PREFERRED_PROVIDER there. its not exactly super clean, DISTRP would be better, but.. it depends, in the end. Jan 20 15:17:35 What's the best way to perform syntax checks (e.g. shellcheck/shfmt) on bash snippets in OE recipes? Jan 20 15:19:37 stacktrust: If I'm not mistaken, it should't be bash first of :) Jan 20 15:20:18 stacktrust: rburton (I think?) told me to use scripts/verify-bashisms to check the scritps are not using some bashism Jan 20 15:21:53 how can i add a task to run before all the other tasks? Jan 20 15:22:11 milloni: depends exactly what you want to do Jan 20 15:22:43 milloni: Is that before all tasks in a single recipe? Or before any task from any recipe? Jan 20 15:22:49 milloni: anonymous tasks can be run at parse time. Otherwise you addtask maybe before do_fetch? (is there something run even before?) Jan 20 15:22:59 qschulz: i want a recipe to error (fatal error) every time someone wants to run it (but not yet when parsed) Jan 20 15:23:10 qschulz: thank you for the hints :) Jan 20 15:23:22 milloni: I'm failing to see the usecase? Jan 20 15:23:23 paulbarker: before all tasks in a single recipe Jan 20 15:23:24 no, i didn't forget the space ;) Jan 20 15:23:56 milloni: Raise a SkipRecipe exception during parsing Jan 20 15:24:17 qschulz: there's an edge case where this class removes do_fetch from the list of tasks to be run Jan 20 15:24:18 E.g. https://github.com/agherzan/meta-raspberrypi/blob/master/recipes-kernel/linux/linux-raspberrypi-dev.bb Jan 20 15:24:27 so that's why im trying to be more general Jan 20 15:24:57 milloni: and this edge case is based on what? Jan 20 15:25:10 e.g. machine? distro? Jan 20 15:25:40 state of a local filesystem Jan 20 15:26:29 milloni: what exactly are you trying to perform, you've tickled my curiosity now :) Jan 20 15:27:31 there's a class that's utterly broken, we've ported a few recipes that were previously using this class to a different class that we wrote Jan 20 15:27:58 that should cover us, however there's multiple recipes that we don't use that i didnt bother to port Jan 20 15:28:19 i want them to error hard in case anyone tries to build them Jan 20 15:28:40 is anyone else seeing a missing python.inc in ross's python2 -> meta-python2 patch. I must be missing something. Jan 20 15:30:13 milloni: then I guess paulbarker's solution is nice enough for you? Jan 20 15:31:12 you could then override this class in your own layer (if your layer is before the layer introducing this class) with that anonymous function (provided it's not your class, otherwise trivial change) Jan 20 15:31:12 i was just going to say, thanks for the tip, but this will result in a "NOTE: recipe being skipped" somewhere up in our huuge build log, and then a "no recipe for: " Jan 20 15:31:46 The blacklist bbclass is another possibility Jan 20 15:32:00 paulbarker: how does that work? Jan 20 15:33:00 I just looked. It raises SkipRecipe haha. Sorry Jan 20 15:33:09 oops :) Jan 20 15:34:00 thanks for the suggestions anyway Jan 20 15:34:50 milloni: but... if there's no other provider for that recipe/package, the build will fail Jan 20 15:35:03 hm, is there a guaranteed order to tasks added using bb.build.addtask() ? Jan 20 15:35:24 milloni: I guess you could use BBMASK also if ever, but you need to maintain the list Jan 20 15:35:34 i.e addtask("A"); addtask("B"); does "B" depend on "A"? Jan 20 15:35:47 qschulz: that's exactly what i dont want to do Jan 20 15:35:52 AFAIK, no Jan 20 15:36:55 milloni: isn't there a parameter to addtask to request A after B or vice versa? Jan 20 15:36:57 qschulz: yes, it will fail, it's more about being user friendly - upon seeing this, one would have to scroll up to see the "NOTE:" up in the log to know why it failed Jan 20 15:36:57 milloni: BUT, bb.build.addtask takes 4 parameters Jan 20 15:37:24 milloni: def addtask(task, before, after, d): Jan 20 15:37:26 and our build log is full of NOTE's and warnings Jan 20 15:37:36 milloni: clean the warnings :D Jan 20 15:38:20 RP: sure, i can do "addtask whatever before whichever" Jan 20 15:38:52 i was wondering if there's an easy way to add my task before all the other tasks Jan 20 15:39:28 because the class in question is adding/removing tasks from the list based on complicated logic so it's not entirely clear which task goes first Jan 20 15:39:56 milloni: dumb question but why can't you fail in the anonymous function run at parsing time? Jan 20 15:40:22 milloni: we don't have such an API since then two people would want this and both would want to be first Jan 20 15:41:19 qschulz: if it fails when parsing, i'd have to bbmask all offending recipes (otherwise the entire build will fail) Jan 20 15:41:29 milloni: duh. Jan 20 15:41:33 (to myself) Jan 20 15:41:48 but if'm going to bbmask all offending recipes, that takes care of it (i dont to do anything more than that) Jan 20 15:41:52 RP: ack, makes sense Jan 20 15:42:11 milloni: COMPATIBLE_MACHINE with an non-existing machine? (COMPATIBLE_MACHINE = "youshouldreallynotbebuildingthisrecipe")? Jan 20 15:42:41 qschulz: hm, :D Jan 20 15:43:19 milloni: or a non-existing license and then put INCOMPATIBLE_LICENSES = "brokenrecipeneedsthisdumblicense"? Jan 20 15:43:32 milloni: I can find plenty of ways to not do it the proper way, just ask :D Jan 20 15:44:40 i think i'm going to stick it "before" some reasonable tasks Jan 20 15:45:00 is it legal to addtask before more than one task? Jan 20 15:45:11 addtask asdf before do_fetch Jan 20 15:45:17 addtask asdf before do_somethingelse Jan 20 15:45:42 ofcourse, but addtask asdf before do_fetch do_somethingelse is fine too Jan 20 15:46:01 perfect, thanks Jan 20 15:47:16 kergoth: ooooh I didn't know this one :) Jan 20 15:47:40 or you could grep oe-core for addtask, pretty sure examples of that are there already :) Jan 20 15:47:41 np Jan 20 15:48:16 milloni: you're basically creating a tree of dependencies with addtask. It's fine if your task appears in different "branches" of that tree (that visual example helps me remembering how things are working, maybe not perfect but seems ok in my mind) Jan 20 15:55:53 is there an ML for doc patches? I figured this would be an interesting ML to follow? Jan 20 15:57:38 qschulz: just the general yocto list Jan 20 15:59:41 I do wonder if we should create one? Jan 20 16:00:05 I think that makes sense Jan 20 16:00:33 General Yocto ML seems to be where many people asking for more high-level and/or beginner help send their emails Jan 20 16:02:35 While we're talking about docs in general, I find the ref-manual/dev-manual/mega-manual split/meaning very non-straightforward. I don't know where to look at generally. Is there any tip you have? Jan 20 16:04:35 qschulz: meta-manual contains all the separate manuals Jan 20 16:04:48 er, mega-manual Jan 20 16:05:02 I can't seem to type mega it seems :) Jan 20 16:05:32 RP: fingers wired to cd-ing into layers I guess :) Jan 20 16:06:48 RP: is mega-manual a concatenation of all the other things or is it a doc on its own (i.e. same content but duplicated or an include from dev-manual in mega-manual?) Jan 20 16:07:24 qschulz: its a concatenation Jan 20 16:08:29 RP: ok thanks. I was confused at one point because I didn't find what I wanted in one doc but in another. So it might just have been a brain fart after all.. Jan 20 16:18:50 New news from stackoverflow: avoid to use 'use_scm_version' in python-jsonchema Jan 20 16:49:28 LetoThe2nd: echo $USER Jan 20 16:53:45 You're welcome :) Jan 20 17:44:47 LetoThe2nd, thanks for hosting the twich session Jan 20 17:47:13 so... setuptools.bbclass isn't in meta-python2, or oe-core? i'm confused. Jan 20 17:47:34 probably should've made sure meta-python2 actually works before removing the files from oe-core Jan 20 17:48:15 kergoth: blame RP :) patch on the list for meta-py2 Jan 20 17:50:05 Hmm, any objection to trying to make examples/samples more consistent? thinking along the lines of manpages.bbclass. use 'examples' packageconfig everywhere, split into ${PN}-examples, possibly be more consistent in the paths on disk too Jan 20 17:57:20 zeddii: its mips or mips64, secondly, on musl/x86 crash is interesting perhaps its better to see if its API issue of kernel problem issue Jan 20 17:58:06 nothing points to the kernel. but as I was saying to RP, I have to put this down until after FOSDEM, I’ve done all I can for the past two weeks. Jan 20 17:58:27 mips64 is what is busted for clocks, I asked upstream. no help there. Jan 20 17:58:39 so I need to find someone that cares about mips64, otherwise, I’m out of ideas. Jan 20 17:59:39 oh, and mips64 issues are only with sysvinit, systemd is fine. Jan 20 18:01:23 that sounds perhaps syscall related, and if its clock_gettime etc. then 64bit time_t changes in kernel might have some changes not propagated to mips64 just a wild guess Jan 20 18:01:37 khem: sounds quite likely Jan 20 18:01:52 khem: build running atm with the libc headers change only Jan 20 18:01:54 yup. they broke VDSO related to gettime, but I wasn’t able to sort it out. Jan 20 18:02:07 zeddii: I assume if the libc piece builds we can merge that? Jan 20 18:02:08 musl/x86 I doubt, since alpine/docker guys would notice it too if its a generic problem Jan 20 18:02:49 zeddii: try with glibc 2.31-rc that we added to core/master lately Jan 20 18:02:52 RP: yup. it’s tested here. Jan 20 18:02:56 khem: I’m fully up to daqte Jan 20 18:02:58 date Jan 20 18:03:03 I see Jan 20 18:03:52 VDSO on mips64 would be needed since folks do run 32bit binaries there Jan 20 18:04:37 mips-kernel was zero help, I’ve debugged it and logged the details in a commit, but it’s just not something I’ll get back to for several weeks unfortunately. Jan 20 18:05:10 I was not happy with the answer upstream, since it has always worked before. I reverted their garbage and it worked again. so the bisect shows what broke it. Jan 20 18:05:24 I see Jan 20 19:30:08 zeddii, maybe a YP bug should be opened so details can be include so anyone who wants to help can Jan 20 19:42:47 armpit: heh, yw @ twitch Jan 20 19:47:09 Hey guys, is there a way to run a qemu built with Yocto using more than 1 core of my CPU ? Jan 20 19:48:33 GrimSleepless: it depends (TM) Jan 20 19:49:33 LetoThe2nd: I think I found the solution using runqemu. I need to add -qemuparams="-smp 2,maxcpus=2" Jan 20 19:50:14 GrimSleepless: that should emulate more than one cpu Jan 20 19:50:29 GrimSleepless: however running one more than one thread should be https://wiki.qemu.org/Features/tcg-multithread Jan 20 19:50:51 LetoThe2nd: oh great! Thank you! Jan 20 19:51:08 (thats at least my understanding) Jan 20 19:51:29 LetoThe2nd: It is more than enough for my current needs! Jan 20 20:16:12 Hello my hard-rocking Amigos! I am trying to add some files to my image's rootfs (I am starting with one single file). What I am doing: modifying this receipe https://github.com/GENIVI/meta-ivi/blob/master/meta-ivi/recipes-extended/common-api/capicxx-core-native_3.1.12.3.bb so that I get locally this -> https://pastebin.com/C87XiJvq (artifacts.xml is one of the files contained in the .zip file, which I Jan 20 20:16:18 can find already decompressed in $7WORKDIR}) Jan 20 20:17:25 nemgti-og: without knowing that capic thing, the cp is misplaced. Jan 20 20:17:28 Still my file is not included in my final build... could anybody please point me in the right direction to get this file in my rootfs? I have already spent a couple of days going through the documentation but havent been able to figure it out Jan 20 20:18:16 hey LetoThe2nd, you mean the copy command inside the do_install_append function? Jan 20 20:18:36 nemgti-og: 1) install instead of cp 2) extend FILES_${PN}, although you should get a install-but-not-shipped warning, theoretically Jan 20 20:18:46 hum is there another cp? :P Jan 20 20:19:17 hehehe.... no there isn't Jan 20 20:19:27 Wouldn't everything in ${bindir} be packaged anyway, so extending FILES_${PN} wouldn't be necessary – it's uncommon to put non-executable files into /usr/bin, though. Jan 20 20:20:04 neverpanic: yup, you're right. my bad. Jan 20 20:20:37 I thought everything would be included as neverpanic said... but then do you think the problem is that I am trying to move an xml file instead of an executable? Jan 20 20:20:38 I'd write this to read install -m0644 ${WORKDIR}/artifacts.xml ${D}${bindir}/testing_jah.xml, but other than that, the approach of using do_install_append() looks fine to me Jan 20 20:21:05 ok.... I think I will try that Jan 20 20:21:06 Although you might want to check ${WORKDIR}/temp/run.do_install for surprises, like somebody doing do_install_append() { exit 0 } Jan 20 20:21:29 ok. I will do that first then :) Thanks a lot! Jan 20 20:22:14 Aren't there available recipes for the CommonAPI generators already? Jan 20 20:24:07 the problem i am facing, is that i want to have the generator in the final image and this is normally not the case Jan 20 20:24:32 Right, normally you'd have the generator in the SDK associated with the image. Jan 20 20:24:49 Well, the generator is also not part of the SDK Jan 20 20:26:08 But even if you want the generator in the final image, wouldn't the appropriate solution to use IMAGE_INSTALL += "capicxx-core" and ensure that there is a non-native version of the recipe? Jan 20 20:26:17 Can I apply patches from a 4.14 kernel to a later kernel? Jan 20 20:26:26 kriive: depends (TM) Jan 20 20:26:32 Unfortunately I'm not an expert of adding things to the Yocto SDK, you would have to consult the documentation on how to do that. Jan 20 20:27:13 I believe TOOLCHAIN_HOST_TASK += "nativesdk-capicxx-core" may be what you're looking for, though. Jan 20 20:27:17 neverpanic: ok i get it. I started with yocto last week so I really have almost null idea of what i am trying to do.... Jan 20 20:27:34 i tried that last one without success Jan 20 20:27:38 .oO( we should really have open sourced our version of those recipes, too ) Jan 20 20:27:57 Well, that's probably because there is no nativesdk version of capicxx-core, so that's something you'd have to provide Jan 20 20:28:20 that is also right, there is no nativesdk-capicxx-core Jan 20 20:28:36 yocto is painfull Jan 20 20:28:47 At times, yes. Jan 20 20:29:16 LetoThe2nd: I'd file for a registration Jan 20 20:29:50 kriive: go ahead. my claims are 30% Jan 20 20:30:18 we have an agreement Jan 20 20:56:54 jsut found out my ${D} variable points to a /../../blahblah/iamge directory which does not exist.... is this normal? Jan 20 20:57:40 and varaible bindir seems to be unexistent Jan 20 21:04:50 http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/classes/base.bbclass#n336 should create ${D} before do_install runs Jan 20 21:05:16 and ${bindir} is defined at http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/conf/bitbake.conf#n44 Jan 20 21:05:27 that sounds highly obscure, yes Jan 20 21:05:46 could it be that i dont find this directory do to a difference between the poky tag i checkout out and the docs? Jan 20 21:06:31 i found out i have variables binddir_{cross, crossscripts, native, nativesdk} but not plain bindir Jan 20 21:06:32 The ${D} directory always gets created while the build runs. It's normal that it doesn't exist before the build, especially if you have rm_work enabled. Jan 20 21:08:22 And ${bindir} I think never gets redefined from the value it has in bitbake.conf; it's components, however, do change, for example ${exec_prefix} (used in the definition of bindir) gets redefined in nativesdk.bbclass at http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/classes/nativesdk.bbclass#n66 Jan 20 21:28:27 thanks guys. I need to stop and rest now. Have a good night you all. Jan 20 21:57:56 Its sad when you think oh, I made changes to this last year? Oh 2009, not 2019 :/ Jan 20 22:57:46 zeddii: that headers build was ok Jan 20 22:58:02 bluelightning: the headers build did show our mystery failure :( Jan 20 22:58:17 and the locked sigs one twice Jan 20 22:58:22 which seems excessive Jan 20 23:13:55 I'd like to have the uwsgi recipe in the meta-openstack layer updated - it's currently using a pretty old version from 2016. Is the only way to learn how to contribute and send a patch? Or is there a maintainer I can ask to update it? **** ENDING LOGGING AT Mon Jan 20 23:42:21 2020 **** BEGIN LOGGING AT Mon Jan 20 23:45:13 2020 **** ENDING LOGGING AT Mon Jan 20 23:48:17 2020 **** BEGIN LOGGING AT Mon Jan 20 23:51:11 2020 Jan 21 00:00:18 nm, I read the contributor guidelines and submitted a patch to the meta-virtualization mailing list/group. Jan 21 02:12:09 RP: cool, so that’s good to merge in isolation. I will post the rest of the series and note the known issues. Maybe I’ll get some interest in them. HA! Jan 21 02:22:02 Hello yocto friends! Jan 21 02:22:52 I would like to generate an image that has both a normal kernel zImage and a zImage with initramfs. Any suggestions on how to do that? Jan 21 02:23:27 The docs seem to indicate that the usual aproach involves chosing one or the other Jan 21 02:49:33 C-o-r-E: how do you plan to boot it ? Jan 21 02:49:48 perhaps look into wic image creator Jan 21 02:55:05 khem: with u-boot Jan 21 02:57:43 i'm more wondering about the recipe and task relationships... since you would need an image to be built before you can add it to the zImage Jan 21 02:58:30 but to build an image, you need virtual-kernel which implies that you already need to build zImage from some linux recipe **** ENDING LOGGING AT Tue Jan 21 02:59:56 2020