**** BEGIN LOGGING AT Wed Aug 14 02:59:56 2019 Aug 14 05:46:07 FailDev: aorry, but why do your compile, package and install soemthign first just to remove it again then? just build your image without ofono Aug 14 05:51:42 whoa, i just checked inbox. and i feel VERY INVITED! Aug 14 05:56:28 LetoThe2nd: Thanks for the session yesterday! Aug 14 05:56:48 Chrusel: you're welcome, hope you could learn something and have fun Aug 14 05:57:56 LetoThe2nd: I could. configs are global, recipes are local ;-) Aug 14 05:58:15 Chrusel: thats really essential! Aug 14 05:58:34 LetoThe2nd: (y) Aug 14 05:58:42 i'm not kidding you, i have every class on yocto/OE repeat that aloud. Aug 14 07:13:19 LetoThe2nd: RP: Weekly Status from August 13, 2019: "Twitch - Next event is Sept. 3rd at 8am PDT" - is the announcement correct or takes the session place an Sept, 10th as expected? Aug 14 07:14:21 Can't wait for brainfood ;-) Aug 14 07:15:44 Chrusel: shall be Sept 10th as expected. Aug 14 07:16:11 seems RP was off-by-one (week) :) Aug 14 07:21:30 LetoThe2nd: sorry, will fix next week Aug 14 07:33:21 RP: no problem Aug 14 07:41:31 LetoThe2nd: thanks for clarifying Aug 14 07:44:41 np Aug 14 08:00:30 RP: is it me or bitbake just became considerably slower at Initialising tasks? Aug 14 08:02:02 RP: theres a considerable pause after Initialising tasks: 100% Aug 14 08:03:24 I think I might have not performed a git pull in about a week, so the change could be somewhere around there Aug 14 08:04:06 too late at night for me to debug haha Aug 14 08:04:16 aehs29: you are maybe experiencing the issues in this thread? http://lists.openembedded.org/pipermail/openembedded-core/2019-August/285692.html Aug 14 08:05:51 aehs29: yes, there are some performance problems :( Aug 14 08:06:06 aehs29: perhaps a few different ones :/ Aug 14 08:08:09 nrossi: yeah that definitely sounds like it Aug 14 08:09:19 at least were aware of it haha Aug 14 08:09:36 aehs29: there is a patch in -next for part of the problem, not sure its all of it Aug 14 08:10:14 RP: cool, I'll pull that and test Aug 14 08:10:17 but perhaps tomorrow Aug 14 08:10:32 is rburton on vacation or something? I wanted to run something by him Aug 14 08:10:58 regarding the gobject introspectin on newer intel tunes Aug 14 08:13:21 ok g2g to bed, gnight guys Aug 14 08:20:50 aehs29: he's away this week Aug 14 09:01:22 Hello someone here knows a good wrbsite/doc/book to explain how to manage several project with several applications, in order to keep the clean setup ? Aug 14 09:11:01 I recommend a vacuum cleaner for large spaces, and a swiffer cloth for smaller ones. Aug 14 09:18:22 RP: I tested the latest bitbake from master-next, but I am still seeing a 1-2 minutes delay between "Initialising tasks" and "Checking sstate mirror object availability" and then of course another pause between "Executing Tasks" and it actually starting to show any tasks being processed... Aug 14 09:33:10 Saur: ok, I think I fixed a problem but perhaps not the only one. How long is the second pause? Aug 14 09:33:24 Saur: I think the second pause is where its figuring out which tasks already ran Aug 14 09:33:44 Saur: that second pause should be better, the first one, less sure about Aug 14 09:34:47 RP: Didn't measure the second pause as that at least used to be a number of minutes. I'll run it again and see... Aug 14 10:53:41 kanavin_: I'm testing a patch to just drop that gcc-cross-canadian shlibs exclusion fwiw Aug 14 11:03:28 I had to drop working on yocto for a while sadly, but now I'm back on the saddle. Someone here was kind enough to point me to how to structure a yocto OS project to be git-friendly that was not described in the manual but I forgot who it was :/ Aug 14 11:11:21 could it be this one https://github.com/rossburton/meta-ross Aug 14 11:22:32 Ad0: if it was ross, then probably https://github.com/rossburton/customdistro Aug 14 11:22:41 Ad0: Perhaps. Aug 14 11:22:41 (10:10:32 AM) aehs29: is rburton on vacation or something? I wanted to run something by him Aug 14 11:22:41 (10:20:50 AM) RP: aehs29: he's away this week Aug 14 11:26:47 ah ! Aug 14 11:26:49 thank you! Aug 14 11:29:17 I see he refers openembedded directly instead of poky Aug 14 11:29:43 I'm trying to run SECURITY_PIC_CFLAGS := "${SECURITY_CFLAGS}" in a layer of mine in an attempt to re-use and extend poky's SECURITY_CFLAGS, but I only get bb.data_smart.ExpansionError: Failure expanding variable lcl_maybe_fortify, expression was ${@oe.utils.conditional('DEBUG_BUILD','1','','-D_FORTIFY_SOURCE=2',d)} which triggered exception NameError: name 'oe' is not defined. Aug 14 11:30:04 Do I have to do some magic in the layer config to be able to use oe.utils.conditional? Aug 14 11:33:50 neverpanic: I think it may be too early, lib/oe probably isn't in the path yet when parsing the layer.conf Aug 14 11:34:15 neverpanic: that seems like something that ought to be in the distro config rather than layer.conf Aug 14 11:35:40 I'm not doing this in the layer.conf, I'm doing it in meta--bsp/conf/distro/include/security_flags.inc, but your point of being parsed very early is a good thought; I probably shouldn't be using := then Aug 14 11:37:12 Unfortunately not using := gets me "Failure expanding variable SECURITY_PIC_CFLAGS, expression was ${SECURITY_CFLAGS} -fPIC which triggered exception RecursionError: maximum recursion depth exceeded", but I guess I'll just have to start looking why that happens then. Aug 14 11:38:43 RP: yes, I saw - thanks Aug 14 11:43:20 SOME_VARIABLE_machine-name, what is this pattern called again? Aug 14 11:43:53 When assignment of SOME_VARIABLE depends on what your MACHINE is? Aug 14 11:48:48 n.m. "Overrides" Aug 14 11:50:17 My yocto build manifest branch is topic/yocto, so I'm getting below error while do_rootfs() of image with Thud based build - https://pastebin.com/ZGK8BLUg Aug 14 11:51:45 psiva87: file not found IMX8MQ_topic/yocto_20190813063206.testdata.json Aug 14 11:52:02 if i had to guess, then its the '/' in the name breaking the build. Aug 14 11:53:29 LetoThe2nd: yes, I think. How should I go abt it as I can't change branch name and all? Aug 14 11:54:21 psiva87: no idea, ask the one who provided it. yet you can always checkout the branch under a different local name Aug 14 11:54:41 git checkout topic/yocto -b local_branch Aug 14 11:54:44 It's git. You are always on your own branch. So checkout a local branch that tracks the remote one. Aug 14 12:22:57 New news from stackoverflow: Configuration files meta data usage in recipes Aug 14 12:44:55 Guys, glib-2.0-target build is broken on master branch (native still works). I am able to build glib-2.0 on warrior and bake the image eventually. I have a Ubuntu 18.04 with gcc 7.4 and gcc 9.1 as build machine (tried with both compilers). It has something to do with cross-compilation with meson build tool. Aug 14 12:48:52 nabokov, we are not able to help you if you do not provide details. master branch is built many times over in varous configurations by the autobuilder machines, so it is not broken and general, and probably breaks only in your specific setup. Aug 14 12:49:58 you could try with plain poky (no other layers) and default configuration for a qemu machine, and see if that builds or not. Aug 14 12:50:27 I see,,, the error is "meson.build:1:0: ERROR: Unknown compiler(s):" in do_configure() task Aug 14 12:51:23 @kanavin_ I will give it a try Aug 14 12:53:38 @kanavin the upper layers do not modify neither qmeu nor glib-2.0 recipes as I have verified this. And the wired thing is that it fails when it checks the compilers and their cmd-line options Aug 14 12:56:27 the cross compiler is aarch64 gcc Aug 14 13:34:14 LetoThe2nd: thanks, i got the file. i will upload! Aug 14 13:44:49 ndec: :) Aug 14 13:45:19 ndec: was a strange session. not many viewers, but more questions. plus more f**kups on my side. Aug 14 13:45:41 it's the Summer effect! Aug 14 13:46:17 possibly. we'll see :) Aug 14 13:55:18 How can I see what OVERRIDES are available? Aug 14 13:58:15 jofr: bitbake -e |grep '^OVERRIDES=' Aug 14 13:58:42 jofr: To be correct, those are the overrides that are active. Aug 14 14:01:22 Saur: Thanks. I actually ended up just piping them to a file from my recipe Aug 14 14:05:14 Does anyone how to troubleshoot do_package_qa QA Issue "systemd: The compile log indicates that host include and/or library paths were used."? Aug 14 14:07:12 chinhuat: i'd try to find out what modifications are in your build concerning the standard systemd recipes, because those do not cause a QA issue :) Aug 14 14:14:02 LetoThe2nd: I'm using systemd from tags/yocto-2.7.1, a minimal bbappend which adds a patch via SRC_URI. Aug 14 14:15:33 the strange thing is that it compiles fine on one Ubuntu 16.04 LTS but not the other Aug 14 14:16:26 at least I can diff the ~1400 lines of log.do_compile, was hoping there's more obvious way to do so Aug 14 14:30:51 and maybe that patch causes it? Aug 14 14:32:02 and what is that tags/yocto-2.7.1 thing, it seems to en vogue today. Aug 14 14:39:22 I can't overwrite default kernel Kconfig value using meta. how can I force it? Aug 14 14:55:49 chinhuat: that usually means it looked at the host system and pulled in something it shouldn't have done. If you grep the log you should find the host reference that test searches for somewhere Aug 14 14:57:08 RP: Nice find abut removing that debug message. With that removed my build times are almost back where they used to be. :) Aug 14 14:58:25 Saur: I'm somewhat relieved its that simple! Aug 14 14:58:32 RP: I've sent a mail to the list with some timings. Aug 14 14:58:51 Saur: thanks for testing and reporting back Aug 14 14:59:29 Saur: I think I'm going to merge all these queued bitbake changes to master and then see where we go from there. I have another three patches locally for more bugs :/ Aug 14 15:01:26 Good thing you didn't merge 46ce56ef (referred to in my mail as 19a88c68, which it was before some rebase of master-next). There it really didn't work at all. :P Aug 14 15:03:27 Saur: I think that commit is still in the queue? Aug 14 15:04:20 RP: Yes, 46ce56ef is its current ref in Poky. But the later commits fixes it. Aug 14 15:04:53 RP: But I think it was the head of master-next yesterday. Aug 14 15:08:25 RP: Oh, and I'm seeing something like this on every build now: WARNING: .../bitbake/lib/bb/cache.py:446: ResourceWarning: unclosed Aug 14 15:09:20 I think I saw someone report this to the list (JaMa?), but I cannot find the mail... Aug 14 15:16:54 Saur: hashequiv or not? Aug 14 15:36:19 RP: thanks, turns out log already contains cc1 warnings about -isystem /usr/include/efi -isystem /usr/include/efi/x86_64 Aug 14 16:01:03 hello all Aug 14 16:01:17 I see that in $CC there are the compiler binary name + some compiler flags Aug 14 16:01:25 and in $CFLAGS, there are some other flags for the compiler Aug 14 16:01:45 what's the rule for a flag to be included in CC or in CFLAGS? Aug 14 16:09:05 I can't overwrite default kernel Kconfig value using meta. how can I force it? Aug 14 16:16:50 RP: I see that on failure it use to emit the location of log.do_* task which ever task was failing but lately I am not seeing it Aug 14 16:22:37 khem: hmm, that may be a side effect of the removal of FuncFailed Aug 14 16:22:48 khem: we should track it down and fix it, maybe need a bug Aug 14 16:23:07 chinhuat: that would do it, it shouldn't be looking there Aug 14 16:53:44 New news from stackoverflow: bash script fails under bitbake due to process substitution Aug 14 17:34:44 so. I have a weak default value (??=) in a recipe. I want to add and remove a few things. What's the way of doing that in my bbappend file ? Aug 14 17:46:23 you can add/remove to the value, but not to the default. Aug 14 17:50:21 so I basically need to copy it and adapt it (and update it whenever defaults changes) ? Aug 14 17:52:10 http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-core/systemd/systemd_242.bb?h=master#n67 - it is that one I'd like to have changes for Aug 14 17:52:55 yep, copy and modify the line. technically it's possible to manipulate the default, but it'd be fragile and built on knowledge of bitbake's internals, which wouldn't be ideal Aug 14 17:57:13 ok. Aug 14 17:57:31 thanks. had hoped for something smarter, but obviously my attempts didn't work. Aug 14 17:59:16 honestly i'd think about just appending/prepending to packageconfig directly. i'd expect most distros to append/prepend to a recipe packageconfig rather than *setting* it, so the distinction lowers in value there Aug 14 17:59:24 depends on the requirements, of course Aug 14 18:24:00 New news from stackoverflow: bash script fails under bitbake due to process substitution, NOT missing /dev/fd link Aug 14 19:16:41 kergoth: I've an interesting problem with the siggen code. I want to select a siggen which isn't available until after the first data store init as its in lib/oe but if you do it from the environment it gives errors :/ Aug 14 19:16:56 * RP does't quite know the best solution for that Aug 14 19:24:29 RP: beer + heavy metal. Aug 14 20:38:40 <__angelo> have a kernel recipe bbappend Aug 14 20:38:54 <__angelo> with do_kernel_configme_prepend() that seems not parsed/executed Aug 14 20:39:05 <__angelo> (i am in sumo) Aug 14 20:53:34 <__angelo> used do_configure_prepend, seems to work Aug 14 21:28:37 RP: Added statistics to the hashserver: http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=jpew/hashserve-stats&id=b57bda1261fcffb81e50eaecc610c12b11f73c78 Aug 14 21:31:06 JPEW: very cool. Let me put that on the autobuilder! Aug 14 21:32:45 JPEW: added. Now to try a build :) Aug 14 21:32:53 * RP has a couple of patches to sort then I'll try one Aug 14 21:36:52 RP: Sounds good. In my testing, there doesn't appear to be any appreciable difference in performance from using the autobuilder database vs. using an empty database, so unless it's running into a more pathological case, the database contents don't appear to be the issue Aug 14 21:38:36 JPEW: any idea where the issue could be? Aug 14 21:39:02 JPEW: I managed a build which was mainly warnings from hashserv connection timeouts earlier which was good and bad :) Aug 14 21:40:11 JPEW: patches needed for testing pushed, build fired Aug 14 21:41:22 RP: Well... some math on the stats would be good. For example, my average connection + request time is about 1 millisecond, so that's 6 seconds for 6101 requests. Aug 14 21:41:41 If the connection time is high but the request time is low, adding more threads might help Aug 14 21:41:56 If the request time is high also, it's probably sqlite's fault Aug 14 21:49:39 JPEW: I messed up and am restarting everything for better stats. Aug 14 21:49:45 JPEW: that run I saw {"connections": {"max_time": 0.05390169005841017, "stdev": 0.0032631353720951077, "num": 91947, "total_time": 266.7830419270322, "average": 0.002901487182040003}, "requests": {"max_time": 0.017891400959342718, "stdev": 0.0002913867554612365, "num": 91946, "total_time": 63.40855172695592, "average": 0.0006896281700884859}} Aug 14 22:00:33 RP: I suspect that the single thread can't keep up and it backs up the queue, delaying the time until connections get processed. Aug 14 22:02:05 {"connections": {"total_time": 1436.2524371920153, "max_time": 0.25648579513654113, "num": 436014, "average": 0.0032940511937506944, "stdev": 0.0057608324730484125}, "requests": {"total_time": 294.0030629877001, "max_time": 0.254582560621202, "num": 436013, "average": 0.0006742988465658137, "stdev": 0.0005197994332214296}} Aug 14 22:02:13 JPEW: seems probable Aug 14 22:02:37 Anyway, time to go home Aug 14 22:03:54 JPEW: data definitely helps, thanks! **** ENDING LOGGING AT Thu Aug 15 02:59:58 2019