**** BEGIN LOGGING AT Wed Apr 03 02:59:57 2019 Apr 03 05:17:19 New news from stackoverflow: Why qtbase is failing during compilation phase in yocto thud? Apr 03 05:47:25 New news from stackoverflow: How to clean, free space? Apr 03 07:33:23 good morning Apr 03 09:02:50 I have a package that was updated, and a dependent package that prints "Skipping as already exists in sysroot" for its dependencies Apr 03 09:03:16 so it doesn't get the updated things in the sysroot Apr 03 09:46:20 ernstp: and... Apr 03 09:46:41 morn Apr 03 09:48:54 mourn Apr 03 10:12:57 RP: I have a bit of free cycles, so if there's anything, I could look into that Apr 03 10:13:07 RP: otherwise, will work on some long-standing package upgrades :) Apr 03 10:13:16 e.g. python-numpy Apr 03 10:13:49 kanavin: do you happen to have an opensuse150 machine handy? Apr 03 10:14:14 kanavin: the main thing right now is to get things ready for release and the bug list from M3 was quite long Apr 03 10:14:37 RP: I have opensuse tumbleweed NUC at home, that's a rolling release though Apr 03 10:14:38 mckoan: I mean I update recipe A and require those things in recipe B, B depends on A Apr 03 10:15:11 but the sysroot is not updated with the new things from A when building B, it just says "Skipping as already exists in sysroot" Apr 03 10:15:11 kanavin: I'm just staring at https://bugzilla.yoctoproject.org/show_bug.cgi?id=13237 and wondering what to do with it Apr 03 10:15:12 Bug 13237: normal, Medium+, 2.7 M4, ross.burton, NEW , [2.7 M3 RC1][OE-selftest][Opensuseleap 15.0][Public AB]:"test_wic_cp_ext" failing for Opensuseleap 15.0. Apr 03 10:15:20 using the old sysroot Apr 03 10:15:32 something is not invalidated properly... Apr 03 10:17:11 kanavin: The other area we have known issues is ptest right now, more at the runner level afaict Apr 03 10:20:43 kanavin: by totally guessing found a revert which "fixes" that wic bug Apr 03 10:23:34 * RP wonders if anyone knows printf vs echo -n in shell behaviour Apr 03 10:23:45 I mean echo -e Apr 03 10:25:33 RP: I could build core-image-sato-sdk-ptest and see about fixing some of this then? https://bugzilla.yoctoproject.org/buglist.cgi?quicksearch=2.7%20m3&list_id=611623 Apr 03 10:25:41 RP: i'm aware that the idiom is to just use printf because echo is so badly defined Apr 03 10:26:02 rburton: so why would http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=89d4a8df074598cfb3a76e41db7c45d845afd961 break :/ Apr 03 10:26:34 kanavin: I'd appreciate more eyes/help on ptest, yes please. I do have a little more data Apr 03 10:26:58 RP: missing \n at the end of the printf expression? Apr 03 10:27:00 kanavin: https://autobuilder.yocto.io/pub/non-release/20190402-8/testresults/qemux86-64-ptest/testresults.json was a ptest run with latest master and the log results are "interesting" Apr 03 10:27:11 $ echo -e hello Apr 03 10:27:11 hello Apr 03 10:27:11 rburton: hmm, could be Apr 03 10:27:12 ross@flashheart ~/Yocto/poky (mut %) Apr 03 10:27:12 $ printf hello Apr 03 10:27:14 helloross@flashheart ~/Yocto/poky (mut %) Apr 03 10:27:43 depends how cmd is handled as to whether that makes a difference Apr 03 10:27:45 hello Apr 03 10:27:47 kanavin: a bit of a brain dump. Grab http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/t222&id=821530d23aa3323edebcb7d5cbd6651c88f3d23c and http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/t222&id=65227988fe1d9ff4a0e2c458fa1393b3374b8328 Apr 03 10:28:09 RP: ha, I guess we're not aiming to get a 100% pass on those :) the best we can do is to make sure the pass rate doesn't decrease, and tests do not timeout? Apr 03 10:28:16 kanavin: "wget https://autobuilder.yocto.io/pub/non-release/20190402-8/testresults/qemux86-64-ptest/testresults.json", "resulttool store testresult.json /tmp/something/" then you can see ptest log files for each ptest Apr 03 10:28:35 kanavin: you can also get a summary of that data with the resulttool report command Apr 03 10:29:20 kanavin: I really want to know a) why several tests timeout b) why several of the later tests see disk space issues c) why python3's log has stap runtime test output in the middle of it Apr 03 10:29:35 kanavin: I'd be a lot happier if this was giving reliable results Apr 03 10:29:41 RP: so which branch/commit should I take for building the image? I guess Ross fixed some of those issues, so I want those fixes Apr 03 10:30:02 kanavin: the above was master and I think has ross' patches. I need to check I didn't miss something from master Apr 03 10:30:49 kanavin: I will teach the autobuilder to generate the log files in future, patches just haven't been cleaned up/submitted yet Apr 03 10:31:24 kanavin: util-linux has a known issue that its not using update-alternatives so tests the wrong binaries. Mariano was looking into that Apr 03 10:31:50 kanavin: less worried about specific failures and most worried about the systemic problems and the timeouts Apr 03 10:32:18 kanavin: lttng has patches pending to fix that timeout so don't worry about that, psrcode is looking at that in isolation Apr 03 10:33:47 RP: right, will take a moment to build the image here though Apr 03 10:33:53 RP, are we really supporting 3 kernels in warrior? Apr 03 10:35:24 * armpit why do we do this to our elves? Apr 03 10:35:35 armpit: Hmm, may need zeddii's input on that Apr 03 10:36:01 4.18, 4.19 and 5.0 Apr 03 10:36:01 kanavin: no problem Apr 03 10:36:12 armpit: I'm sure the plan was to remove at least one Apr 03 10:39:51 armpit: I think we need a list of things to sort before M4 is built Apr 03 10:40:27 RP: just to make sure, the ptests are run in sequence, and not in parallel? Apr 03 10:42:03 kanavin: oe-core 7ef8f96941ed52b2a00cbe8f57511a8891b39698 is a way of working around the update-alternatives problems Apr 03 10:42:09 (that was for busybox, same issue) Apr 03 10:43:03 * armpit tiny uses 4.18, lsb uses 4.19 and default is 5.0 Apr 03 11:00:17 how was the turnout @ the tech meeting? Apr 03 11:01:04 you were not there so everyone left Apr 03 11:01:46 * armpit sees RP all over the planning sheet Apr 03 11:02:02 Crofton, I doubt that Apr 03 11:03:22 Crofton, what are you doing up so early? Apr 03 11:04:55 I was going to ask the same .... Apr 03 11:06:12 traveling.. in UK heading to India Apr 03 11:06:29 ah Apr 03 11:06:51 slap RP around for me while there Apr 03 11:07:25 what about rburton ?? he may be closer Apr 03 11:08:02 if you're in london then RP is likely closer Apr 03 11:08:04 slap him! Apr 03 11:08:11 Too bad they are at opposite ends of the contry Apr 03 11:08:19 country Apr 03 11:08:51 or has Britain split back into many kingdoms again? Apr 03 11:09:15 look its not funny we're living in this carnage Apr 03 11:10:06 the live updating news site i'm following for brexit is up to three pages of events today and its barely midday Apr 03 11:13:00 jesus Apr 03 11:14:45 we have a zombie for president, I don't know what's worse :) Apr 03 11:15:04 can you secede your home to an EU Apr 03 11:15:56 JaMa, don't zombies have brains? Apr 03 11:17:19 at least you're not in brexit for cccpenter Apr 03 11:17:42 armpit: only small one to obtain alcohol or some food Apr 03 11:17:59 the rest is already eaten from inside Apr 03 11:18:36 rburton, I also find it funny how BBC says 'another decisive week for the brexit process' Apr 03 11:18:54 totally decisive Apr 03 11:19:51 oh, and mildly annoyed that they have in fact taken a side in all this :( they give a disproportionate amount of time to various brexiteer flavours, and almost no time to those who think it was all a colossal mistake Apr 03 11:20:18 kanavin: i think they meant divisive Apr 03 11:21:55 * armpit needs a nap Apr 03 11:22:39 RP, looks like I need to backport manual test for sumo Apr 03 11:23:02 maybe even thud Apr 03 11:25:13 * Crofton waits for Trump to close the Mexican border and the supply of Avacados to dry, thus sticking it the libs again Apr 03 11:26:51 hi guys i have a problem :) i am using bitbake and my production code compiled fine... but when i try to bitbake unittest which uses the same .inc as my production code, i get linker error undefined reference for references inside library which worked for production... this library has LIB without append for target... i tried rearranging order of libs, but no result... i hope i explained clearly enough :/ Apr 03 11:27:53 lib in question is protobuf from google proto buffer.. and all undefined references are for some functions envolving strings Apr 03 11:28:25 oh god, back to work everyone! Apr 03 11:29:07 o_o Apr 03 11:42:29 Crofton: it is weird how rburton and I are literally at opposite ends of the country Apr 03 11:42:45 * RP thinks we should bring back the kingdom of Northumbria Apr 03 11:43:12 Ruled by Uthred of Babbenburg Apr 03 11:44:13 Crofton: indeed. Bambrugh (Babbenburn in modern English) still has a castle ready to act as capital again :) Apr 03 11:44:37 I've looked at it using the Google real time camera Apr 03 11:44:49 Well, looks like Britain expelled armpit Apr 03 11:45:12 Lots of good castles around here. Something to do with a Viking problem Apr 03 11:45:53 Pesky Danes Apr 03 11:51:55 rburton: you're right about the trailing \n, that fixes it Apr 03 12:00:19 RP: cookie for me! Apr 03 12:04:06 rburton: yes, one problem down :) Apr 03 12:40:03 I'm writing a recipe for a package which has its source in a git repo with submodules. I don't see support in the fetcher for that Apr 03 12:40:26 u1106: gitsm:// Apr 03 12:40:52 https://www.yoctoproject.org/docs/latest/bitbake-user-manual/bitbake-user-manual.html#gitsm-fetcher Apr 03 12:40:57 rburton beat me to it :) Apr 03 12:41:12 but links to docs +++ Apr 03 12:44:05 ah, bitbake manual, I searched the mega manual... Apr 03 12:46:13 hmm... Apr 03 12:46:13 > The fetcher has known issues where it does not use the normal source mirroring infrastructure properly Apr 03 12:46:36 so an offline build from my local source mirror might not work? Apr 03 12:53:14 sorry, new to yocto. please can any one suggest how can I check the yocto version. Apr 03 13:00:13 from the source? git describe From inside an image? cat /etc/version (Not sure whether all image recipes put something useful there, mine does) Apr 03 13:02:25 anshulma: where did you get your yocto from? Apr 03 13:05:33 rburton, its from intel.. Apr 03 13:06:42 I want to read it from inside the image and cat /etc/version doesn't contain much useful.. Apr 03 13:07:18 anshulma: *where* from intel Apr 03 13:07:54 * Crofton loves the image-buildinfo bbclass Apr 03 13:07:56 if it exists /etc/os-release is useful Apr 03 13:09:32 looks like something I forgot to turn on, thanks for the tip Apr 03 13:09:48 rburton, yes it exists and provide the required info. Thanks. Apr 03 13:24:22 Hello, I have from our supplier a kernel folder and u-boot folder. They are git clones of freescale uboot-imx.git and linux-imx.git with boards specific files in them. I would like to integrate them into "meta-mylayer" to make everything more integrated. Iknow there are tons of guides out there but I'm hessitant to pick a random guide without having a plan, does anyone have suggestions on the best way to implement this or any tips? Apr 03 13:28:30 i have a static library with .a files for many different architectures. how can i find all the architectures for which my bitbake is compiling for Apr 03 13:29:11 Absentee: its compiling for the architecture you set MACHINE to Apr 03 13:30:19 TARGET_ARCH might be close to what you want Apr 03 13:30:25 well... its for many machines :D Apr 03 13:30:54 thanks for the direction where to look for Apr 03 14:09:07 should this one be applied to sumo and thud? http://lists.openembedded.org/pipermail/openembedded-core/2019-March/279839.html rocko and older don't have these 2 patches **** BEGIN LOGGING AT Wed Apr 03 14:10:19 2019 Apr 03 14:14:20 willie2: in a nutshell, do not do. Apr 03 14:15:14 willie2: put the folders into some kind of source management or at least to a known location as a tarball. then reference that from the metadata Apr 03 14:17:52 LetoThe2nd: And only use yocto for the filesystem? Apr 03 14:18:19 Letothe2nd: That does not feel like a long-term solution Apr 03 14:22:20 Letothe2nd: would it be much work to copy the dts files and create my own machine or something like that? Apr 03 14:38:00 willie2: Freescale maintains a Yocto-layer (meta-freescale).. does that not suit your needs? Apr 03 14:42:18 jofr: The board I'm using is custom but based on imx6sabresd. I'm trying to figure out how to implement my custom u-boot and kernel into my yocto layer Apr 03 14:47:23 Then you could, like LetoThe2nd said, but this into version control, and then create you layer with recipes that pulls those changes in Apr 03 14:47:34 put* Apr 03 14:49:02 New news from stackoverflow: tiovx-app-host error on arago yocto project on cunstom AM57xx board Apr 03 14:49:07 kanavin: any luck reproducing the ptest issues? I can reproduce problems here but its inconsistent :/ Apr 03 14:53:37 jofr: and this is where you lose me, I have additional .dts and .dtsi files, with new names etc. Can i create a "u-boot-imx_bbappend" recipe that uses them, localy or otherwise and how do i make sure it gets implemented correctly? I'm sure just appending "my_custom.dts" wont solve the problem Apr 03 14:55:27 jofr: i think i will try to create my own machine that uses these specific files, I'll probably come back crying later Apr 03 15:04:04 Please can anyone share any doc/links or suggestions regarding ACPI event handling in yocto. I have an embedded device with ubuntu on it. WHen I press power button I can see event in /etc/acpi/event/powerbtn and I can modify this script to disable OS shutdown. And it works. Now I installed yocto on this device, disabled shutdown in /etc/apci/event/powerbtn, but still the OS shutds down. Apr 03 15:04:51 I can see that event does gets delivered to /etc/apci/events/powerbtn (I had log stmt there),but even after disabling OS shutdown, still it happens. Apr 03 15:38:59 anshulma: are you using systemd? Apr 03 15:39:05 iirc that handles the power buttons itself Apr 03 15:45:55 rburton, yes, I am using systemd.. Thats a good point. Now let me figure out how can I handle power events with systemd. Thanks. Apr 03 15:50:20 anshulma: https://www.freedesktop.org/software/systemd/man/logind.conf.html Apr 03 15:55:28 rburton, yes, referred that and now I able to control the behavior using systemd. Thanks a lot for your help.. Apr 03 15:56:19 * power button behaviour.. Apr 03 16:16:31 willie2: sorry, i am not around much today. no, you would use yocto to build all of u-boot, filesystem, kernel. but the metadata (= layers) should not be bundled with the sources. Apr 03 16:49:23 kanavin: confirmed the strace issues are a space issue, going to add a patch to fix that. Doesn't explain the other issues though :/ Apr 03 16:55:53 RP: I just came back to check the testimage run here Apr 03 16:55:59 RP: one failure is NOTE: core-image-sato-sdk-ptest-1.0-r0 do_testimage: test_rpm_install (rpm.RpmInstallRemoveTest) Apr 03 16:55:59 NOTE: core-image-sato-sdk-ptest-1.0-r0 do_testimage: ... FAIL Apr 03 16:56:06 ... installing package base-passwd-doc-3.5.29-r0.core2_64 needs 44KB on the / filesystem Apr 03 16:56:25 so something in the tests is filling up the disk, but I guess in general we should provide lots and lots of space to the tests Apr 03 16:56:48 also: NOTE: core-image-sato-sdk-ptest-1.0-r0 do_testimage: test_ptestrunner (ptest.PtestRunnerTest) Apr 03 16:56:48 NOTE: core-image-sato-sdk-ptest-1.0-r0 do_testimage: ... unexpected success Apr 03 16:56:53 not sure what that means :D Apr 03 16:57:19 kanavin: we effectively ignore the test result of ptest runner and expect it to fail Apr 03 16:57:43 kanavin: you can poke into tmp/log/oeqa to see the actual ptest results Apr 03 16:58:05 resulttool report tmp/log/oeqa/ Apr 03 16:58:39 RP: thanks, I just promised I would be heading home though, so it'll have to wait until tomorrow Apr 03 16:58:46 kanavin: np Apr 03 16:59:12 kanavin: I'm just sharing anything I find so we don't duplicate work. There are still several odd things going on here :/ Apr 03 16:59:55 RP: to run ptests to begin with is an improvement compared to not running them :) Apr 03 17:01:02 $ resulttool report tmp/log/oeqa/ Apr 03 17:01:07 kanavin: totally! We're definitely getting more consistent results Apr 03 17:01:09 ... ModuleNotFoundError: No module named 'jinja2' Apr 03 17:01:17 kanavin: ah, it does need that :/ Apr 03 17:01:28 * RP did update the docs with the dependency Apr 03 17:01:53 right, as long as it's documented Apr 03 17:05:03 kanavin: patch on the list for the space issue. We can retest with this and see how things stand on the AB too Apr 03 17:08:08 RP: right, I'll re-run the tests using this Apr 03 17:16:54 hey all, I want to set the branch and git hash (SRCREV_pn- ??= ) of multiple recipes from a single include file I call manifest.inc....the problem is some of our libs live in gitlab, some in SVN, and I'm getting this error a lot "The SRCREV_FORMAT variable must be set when multiple SCMs are used." whenever I bake a single recipe Apr 03 17:19:11 is there a way to set the git hash or svn revision of multiple recipes from an include without corrupting the SRCREV setting of each individual recipe with all the others Apr 03 17:20:45 ebolton: SRCREV_pn- = "" ? Apr 03 17:21:12 @RP yea, that's what I'm doing now.... Apr 03 17:22:47 ebolton: I think you've set a SRCREV for something which doesn't support it and its that which is causing problems Apr 03 18:32:52 anyone here contribute to the gobject-instrospection? Apr 03 18:33:27 I have run into an issue and know a per recipe fix but I am curious if there is a more global workaround Apr 03 18:34:20 g-ir-scanner is blowing up because my $HOME is on a separate mount, so I started getting Apr 03 18:34:22 OSError: [Errno 18] Invalid cross-device link: '/tmp/g-ir-scanner-cache-1ybcs5xj' -> '/folk/masselst/.cache/g-ir-scanner/b81f3d909aae112686664c4f2f515145b23639cb' Apr 03 18:35:04 I can workaround this on a per recipe way, such as Apr 03 18:35:05 +do_compile_prepend() { Apr 03 18:35:07 + # prevents g-ir-scanner from writing cache data to $HOME Apr 03 18:35:09 + export GI_SCANNER_DISABLE_CACHE=1 Apr 03 18:35:10 +} Apr 03 18:36:09 I am thinking using the cache is ripe for host contamination, so is there a way to bake this into g-ir-scanner? Apr 03 18:46:02 digging more it looks like modifying the gobject-introspection.bbclass is the only/best option Apr 03 18:49:49 marka: I suspect we could add it into a wrapper around g-ir-scanner Apr 03 18:50:57 ya, that would make sense Apr 03 18:51:54 I find it odd that they wouldn't account for this scenario Apr 03 18:52:17 or maybe they do and it is just busted which is why I only started to see this recently Apr 03 18:54:39 * marka pokes at their git tree to see if there are any hints Apr 03 18:55:49 marka: I didn't reply about that go patch did I? :/ Apr 03 18:55:54 RP, it is in the wrapper for the target Apr 03 18:56:10 RP, no the GO patches are still waiting for to be merged Apr 03 18:56:49 OK, so my comment about possible host contamination are void as we don't use the cache for class-target Apr 03 18:56:54 marka: sorry, its slipped my radar :/ Apr 03 18:57:06 RP, I was debating pinging you again Apr 03 18:57:34 zeddii: said he was going to put them in his next merge request if they weren't merged as he needs them too Apr 03 19:01:49 it looks like the scanner respects XDG_CACHE_HOME, so for -native I could potentially use a directory in the build for scanner cache Apr 03 19:02:25 I don't see anything in their tree discussing this, looking for bug reports now Apr 03 19:04:28 nothing Apr 03 19:28:17 marka: how important is the cache? Apr 03 19:35:17 not having poked at this myself I don't have the information to say Apr 03 19:35:33 my worry is that it looks like you guys took the effort to keep it enabled for -native Apr 03 19:36:07 so it gave the impression that nobody wanted to disable for all Apr 03 19:36:53 I am looking for the change that caused my builds to fail, hoping that will help me with deciding how to fix this Apr 03 19:37:25 this is of course what I was trying to avoid, having stayed away from the gobject-introspection 'fun' previously Apr 03 19:38:21 let me do more investigation, I will get back to you with any findings and my proposed solution later today or in the morning Apr 03 20:04:08 marka: you should be able to do that workaround in the class Apr 03 20:04:18 cache disabling seems sensible Apr 03 20:04:46 rburton: agreed, it does work, since I am into it now I am doing more poking around before I apply the hammer Apr 03 20:05:23 as you say, cross-build infection is quite a risk so unless there's a lot of safety the hammer seems good to em Apr 03 20:05:25 me Apr 03 20:05:33 (especially as we're close to release) Apr 03 20:05:42 kanavin is the g-i expert fwiw Apr 03 20:05:52 rburton: as noted above the hammer is already in place for cross building, class-target Apr 03 20:05:57 so no worries there Apr 03 20:06:00 oh is it? Apr 03 20:06:03 it is Apr 03 20:06:30 thus less rush and I can try to pinpoint why this changed Apr 04 02:13:37 zeddii, you here ? Apr 04 02:14:32 * armpit maybe I'll send an email Apr 04 02:21:07 New news from stackoverflow: How can I use bitbake to set /etc/timestamp? **** ENDING LOGGING AT Thu Apr 04 02:59:57 2019