**** BEGIN LOGGING AT Tue Jun 16 03:06:38 2020 Jun 16 03:28:11 im seeing below error with do_packagedata Jun 16 03:28:12 ERROR: package-release-0.1-r0 do_package: Error executing a python function in exec_python_func() autogenerated: Jun 16 03:28:13 '/home/sikumar3/zeus/sources/poky/meta/classes/package_pkgdata.bbclass', lineno: 44, function: package_prepare_pkgdata Jun 16 03:28:16 0046: Jun 16 03:29:10 the recipe was working fine with Jethro, Sumo. But, I am getting the above error with ZEUS Jun 16 09:20:01 jonmason: there's a team in intel that was actually using it (for a demo) Jun 16 11:09:32 Hey, I've got following problem: I've got a program that is compiled using qmake, it works fine, binary is in "B", then I `install -d ${D}/${bindir}` and `install -m 0755 ${B}/program-name ${D}/${bindir}` Jun 16 11:10:03 I've also added FILES_${PN} += "${bindir}", however the binaries are not installed Jun 16 11:10:25 I've copied them out from B using cp to my home directory and it works Jun 16 11:10:37 what could I be missing? Jun 16 11:18:44 kpo: look into $WORKDIR/image of your recipe, is the binary there? Jun 16 11:19:54 kpo: then, look into $WORKDIR/packages-split and check in which directory it is (normal package? -dev package? etc...) Jun 16 11:20:35 just to be sure the install is done in do_install right? Also check that your do_install has the correct content by checking $WORKDIR/temp/run.do_install Jun 16 11:37:57 qschulz: thanks, I'll look into that and get back with some info Jun 16 11:50:10 qschulz: ${WORKDIR}/image/${bindir} holds the built binary with 755 permissions, ${WORKDIR}/packages-split however does not exist Jun 16 11:51:15 I'm getting this info from do_install(), I've added some ls'es and echo'es that print me info to the logfile Jun 16 11:51:44 is it possible, that packages-split does not exist yet during do_install? Jun 16 11:56:46 kpo, maybe try specifying the binary explicitly in FILES_${PN} instead of just ${bindir} ? Jun 16 11:57:32 but I think ${bindir}/* is already covered by default Jun 16 11:58:26 so you should probably not need to specify it at all Jun 16 11:59:30 bitbake -e | grep ^FILES_ Jun 16 12:00:56 kroon, yes, I've tried to add explicit binary name to no avail. Will post ^FILES_ in a second Jun 16 12:01:25 kpo, post the recipe somewhere perhaps ? Jun 16 12:02:09 sure Jun 16 12:05:29 kroon: https://pastebin.com/EYe8se6r Jun 16 12:06:42 kpo: packages-split is created and used after do_install, in do_package probably? do_install installs into the image, so your do_install seems correct Jun 16 12:07:24 require recipes-qt/qt5/qt5.inc what is this :) ? Jun 16 12:08:23 just use `inherit qmake5` instead of that line and start from there Jun 16 12:08:28 kpo: If you just run `bitbake ` do you see packages-split? Jun 16 12:10:43 kpo: recipes-qt/qt5/qt5.inc is a common file for *the* qt recipes (e.g. qtbase, etc...) If you want to build a software based on qt (probably using qmake?) inherit qmake5. Jun 16 12:18:35 require recipes-qt5/qt5/qt5.inc <- thats copied from another recipe that builds much more complex qt5 app, yes, I should be using just inherit qmake, I'll try that :) Jun 16 12:19:26 and it's probably incorrect for that recipe as well? Jun 16 12:20:56 well, could be Jun 16 12:34:25 qschulz: well, turns out that after switching to inherit qmake5 it started to work as espected! thanks a lot! Jun 16 12:34:52 kpo: :) happy to have helped! Jun 16 12:35:37 Now it's time to investigate what the other recipe that builds qt5 app uses this non-proper way of building using requires /inc/blabla :) Jun 16 12:38:52 gl :) Jun 16 13:15:19 Guys need some help creating an ISO file using wic...I have created a core-image-full-cmdline for genericx86-64. And then added the following line WKS_FILE = "mkhybridiso.wks" in conf/local.conf and the create an ISO using wic create ../scripts/lib/wic/canned-wks/mkhybridiso.wks -e core-image-full-cmdline. It gives me two files Jun 16 13:15:20 HYBRIS_ISO_IMG.direct and .iso. I tried booting using .iso file but it came up with an error message "Cannot find rootfs.img file in /dev/media/hdc, dropping to shell....can progress from there. Any assistance to the right direction will be appreciated.. Jun 16 13:16:26 The following build artifacts were used to create the image(s): ROOTFS_DIR["/boot"]: build/tmp/work/genericx86_64-poky-linux/core-image-full-cmdline/1.0-r0/rootfs BOOTIMG_DIR: build/tmp/work/genericx86_64-poky-linux/core-image-full-cmdline/1.0-r0/recipe-sysroot/usr/share KERNEL_DIR: Jun 16 13:16:26 build/tmp/deploy/images/genericx86-64 NATIVE_SYSROOT: build/tmp/work/core2-64-poky-linux/wic-tools/1.0-r0/recipe-sysroot-native Jun 16 13:17:45 srijan: tbh, that rather sounds like some hardcoded path is injected somwhere ("/dev/media/hdc") Jun 16 13:19:26 I also figure out that my images are all around 400MB in size, but the HYBRID_ISO_IMG is only 44 MB in size. Seems some where it is not picking with the ROOTFS Jun 16 13:19:43 very well possible, yes. Jun 16 13:22:58 Is the BOOTIMG_DIR directory correctly set? Becasue all the other three ROOTFS_DIR, KERNEL_DIR and NATIVE_SYSROOT are the one that have the right files Jun 16 13:23:55 inspect bitbake -e to find out, i'd say. Jun 16 13:32:05 Seems they are in the correct place, but still wic cannot compile the iso with the rootfs Jun 16 13:35:08 The mkhybridiso.wks....: Jun 16 13:35:10 part /boot --source isoimage-isohybrid --sourceparams="loader=grub-efi,image_name=HYBRID_ISO_IMG" --ondisk cd --label HYBRIDISObootloader --timeout=10 --append="" Jun 16 13:35:45 Is this OK, or do we need to add something else, here.. Jun 16 14:29:42 hello guys Jun 16 14:44:31 In my Yocto project, I have a shared library that was successfully created and installed (via cmake). Now i want to use this library in an application and i use in CMakeLists.txt: 'findlabry(MYVAR mylib)'. When building devtool tells me that MYVAR is set to 'NOTFOUND'. Any suggestions? Jun 16 14:44:53 find_library Jun 16 14:46:10 chris_ber: the recipe that wants to use it has a DEPENDS? Jun 16 14:46:10 chris_ber: missing DEPENDS on the recipe providing the shared library? Jun 16 14:46:30 Letothe2nd: one too many words for me otherwise I'd have been quicker. Damn it. Jun 16 14:46:55 I added (manually) 'DEPENDS += " mylib"' Jun 16 14:47:03 ^^ Jun 16 14:47:07 qschulz: :) Jun 16 14:47:56 chris_ber: where do you install your lib in your package (not the one needing it, the one providing it) Jun 16 14:51:11 https://bitbin.it/DzdHHJqB/ Jun 16 14:51:34 qschulz: everything should be done by cmake? Jun 16 14:52:53 do you have your shared lib in the recipe-sysroot directory of your second (devtool'ed) recipe? Jun 16 14:53:22 no Jun 16 14:53:30 oh wait Jun 16 14:55:02 tmp/work/armv7vet2hf-neon-poky-linux-gnueabi/monitoring-service/1.0+git999-r0/recipe-sysroot/lib: does not contain my shared library Jun 16 14:55:57 can you check if it's in recipe-sysroot and not only in recipe-sysroot/lib? Jun 16 14:56:48 YPTM: Jan-Simon is on Jun 16 14:57:29 i can find it in 'recipe-sysroot/usr/lib/' Jun 16 14:58:17 YPTM: armin in on Jun 16 14:58:42 chris_ber: which is correct (at least the default path where to expect libs) Jun 16 14:59:20 chris_ber: do you have `inherit cmake` in your recipes? Jun 16 14:59:50 ok, could be something with name? in my library-cmake-file i named it 'prom' and the 'lib' was added automatically to 'libprom' Jun 16 15:00:02 yes, it is inherit from cmake Jun 16 15:00:16 YPTM: Scott Murray is on Jun 16 15:00:35 chris_ber: that's a good lead Jun 16 15:00:37 YPTM: Joshua Watt here Jun 16 15:01:17 Come to YP Dev Day Virtual! Jun 16 15:01:17 YP Dev Day Virtual, home page: Jun 16 15:01:18 https://www.yoctoproject.org/yocto-project-dev-day-virtual-north-america-2020/ Jun 16 15:01:18 YP Dev Day Virtual, ELC registration page: Jun 16 15:01:18 https://events.linuxfoundation.org/yocto-dev-day-north-america/register/ Jun 16 15:01:21 YPTM: ross here Jun 16 15:01:45 YPTM: david here Jun 16 15:02:25 qschulz: if i remove the prefix 'lib' it fails and suggests 'Close matches: libprom' Jun 16 15:03:51 chris_ber: I'm not familiar with cmake. You should dump the directories which are traversed by cmake when looking up for this library Jun 16 15:03:53 this is the recipe created by devtool add: https://bitbin.it/tFK1WGlH/ Jun 16 15:04:14 and i added 'DEPENDS ...' by myself Jun 16 15:04:32 ok Jun 16 15:05:17 and check that you take all paths sent by Yocto in your cmake or find the one that isn't :) Jun 16 15:05:48 I can't help more but someone might. In any case, if you find the solution, please let me/us know Jun 16 15:06:03 ok, thx. i will let you know Jun 16 15:08:17 RP: link to these WDD (?) charts? Jun 16 15:08:31 tlwoerner: I think it's https://wiki.yoctoproject.org/charts/combo.html Jun 16 15:08:38 paulbarker: thanks Jun 16 15:15:28 tlwoerner: that is the one Jun 16 15:36:25 hi there ! I was giving a try to core-image-x11 running into qemu.. and it's damn sloooow.. do I have to check something on qemu conf, prior launching the stuff ? Jun 16 15:37:09 PaowZ: qemu is slow, and even wors if you're in a VM Jun 16 15:38:11 PaowZ: did you enable kvm? Jun 16 15:40:45 RP: well.. at first sight, -enable-kvm is not supplied to runtime.. https://pastebin.com/nvqGMq4w Jun 16 15:45:15 take this, youngsters: https://twitter.com/TheYoctoJester/status/1272918015722639361 Jun 16 15:46:08 Letothe2nd: ok boomer Jun 16 15:48:39 RP: woow.. that definitely did the trick.. how's that not appended by default.. Jun 16 15:49:53 qschulz: adding '.so' to the name of the shared library worked Jun 16 15:50:17 now i am struggled with missing headers, but i think i solved the find_library-issue Jun 16 15:50:34 Letothe2nd: shall became a mantra.. Jun 16 15:50:39 *ebecome Jun 16 15:51:26 chris_ber: it seems you need the .so indeed. https://cmake.org/cmake/help/latest/command/find_library.html is not very clear about that but it's weirdly implicit Jun 16 15:52:01 "When using this to specify names with and without a version suffix" I guess it means you have to give the whole filename until the version suffix (if there is one) Jun 16 15:53:12 yo, think so Jun 16 15:53:41 chris_ber: where are your headers installed? Jun 16 15:53:50 are they available in the sysroot? Jun 16 15:54:02 i am not sure. i am looking them up Jun 16 15:55:50 i have to use 2 headers, but 1 is missing Jun 16 15:56:20 chris_ber: missing from the recipe-sysroot as well? Jun 16 15:57:01 in CmakeLists.txt in my library i added: 'set_target_properties(prom PROPERTIES .... PUBLIC_HEADER prom/include/prom.h PUBLIC_HEADER promhttp/include/promhttp.h) but prom.h is missing Jun 16 15:57:58 before checking the CmakeLists (which i probably won't be able to help with) we need to check the headers are in the recipe-sysroot Jun 16 15:58:17 recipe-sysroot of the library? Jun 16 15:59:30 in the recipe that needs the header that is "missing" Jun 16 16:00:14 yes, 1 of 2 is missing Jun 16 16:00:54 sakoman: http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/t222&id=d94aac9c23c62c80fd4b5b88c05c8bcedd748295 is part of the storable issue Jun 16 16:04:39 chris_ber: ok, is it available in sysroot-destdir in the recipe providing the header? Jun 16 16:04:47 no Jun 16 16:05:01 chris_ber: ok, what's the path to this header? Jun 16 16:06:12 build-qemuarm/tmp/work/armv7vet2hf-neon-poky-linux-gnueabi/libprom/0.0.3-r0/git/prom/include/prom.h Jun 16 16:06:44 this is the only path Jun 16 16:07:36 chris_ber: two options... make this path part of SYSROOT_DESTDIR (meh), or the best one (I think?) would be to install this prom.h in ${D}${includedir}. Not entirely sure but that should work Jun 16 16:07:54 ok Jun 16 16:16:19 qschulz: found it in cmake: 'set_target_properties(mylib PROPERTIES PUBLIC_HEADER "some.h;another.h")' <-- i have to list all headers together instead of call every with PUBLIC_HEADER Jun 16 16:16:37 thats enough for today. Thank you for your help Jun 16 17:18:18 Hi, this has probably been reported and it might have already been solved.. but bitbake fails miserably if $(pwd) doesn't exist. Jun 16 17:18:43 go make it exist, then. Jun 16 17:31:13 fullstop: coreutils or other "pwd" provider isn't explicitly listed in https://www.yoctoproject.org/docs/3.1/ref-manual/ref-manual.html#required-packages-for-the-build-host , but it's hard to imagine a distro not having dependencies from those required packages that would pull it in Jun 16 17:31:32 smurray: no, it's not that.. Jun 16 17:31:59 smurray: an infinite loop of FileNotFoundError: [Errno 2] No such file or directory Jun 16 17:32:22 pwd exists, $(pwd) does not.. if that makes sense. The current directory of the shell has been removed. Jun 16 17:33:10 fullstop: how did you manage that, "rm -rf" while it was running? If so, I'm with Letothe2nd, don't do that Jun 16 17:35:21 bitbake clean target while I was in the work directory Jun 16 17:35:31 ^ it can happen if you launch bitbake from a tmp work directory and clean a recipe. just do cd $BUILDDIR before bitbake Jun 16 17:35:47 I understood how it happened and how to get around it, but an infinite loops seems like a bug, no? Jun 16 17:42:51 fullstop: perhaps. I've managed to do it here (within the last couple of days, actually), and not seen the loop, just an error, so a question for RP, I guess Jun 16 17:43:23 Maybe it gives up eventually Jun 16 17:43:30 I stopped it after 6 tries Jun 16 17:51:43 something like this: https://pastebin.com/raw/V9Y1qwfY Jun 16 17:51:49 although maybe with some sort of warning Jun 16 18:07:51 fullstop: can't hurt to file a Bugzilla. I could maybe see just error-ing out immediately as opposed to chdir-ing being preferable for some people Jun 16 18:44:35 kanavin_home, sno: Can either of you see why Storage.pm (or any .pm file) would sometimes end up in /usr/lib/perl5/5.30.X/$[TARGET_ARCH}-linux/ and sometimes not? Jun 16 18:45:20 I've sent a patch to fix the erroneous module creation but I'm still struggling to see why perl-module-x86-64-linux-storage would sometimes be created in PACKAGES and sometimes not Jun 16 18:45:33 Sadly I don't have a reproducer, just failed builds which say this happened Jun 16 19:13:19 In yocto krogoth , is there a way to remove a DEPENDS package form the final roots ? Cryptsetup list lvm2 as a DEPENDS but still install the lvm2 package .Using PACKAGE_EXLUDE (with ipkg) break the rootfs task and IMAGE_INSTALL_remove have no effect. Am I missing something ? Jun 16 19:17:38 aleblanc: if you are convinced that it will work, bbappend cryptsetup to remove the dependency Jun 16 19:19:34 Letothe2nd: hum yeah, guess that would be my next step Jun 16 19:20:32 aleblanc: but what is your *actual* problem? Jun 16 19:20:44 aleblanc: (besides still doing work on krogoth) Jun 16 19:21:31 Letothe2nd: ahah, need to trim 10mb ... lvm2 take about 5, and we don't use it... Jun 16 19:22:32 aleblanc: yeah but is this a new problem? Jun 16 19:23:29 aleblanc: now stepping up with something like this on krogoth is like going to the car shop and saying "hey i bought this car 15 years ago, and now i realized that it has issues" Jun 16 19:26:25 Letothe2nd: as far as I can see we are force to use krogoth as is the only one still capable of building something compatible with linux 2.26.33. There might be other solution but to me they are not as straight forward. Jun 16 19:26:52 aleblanc: ah so you're into software archeoloy :) Jun 16 19:27:02 Letothe2nd: yes pretty much ... Jun 16 19:28:54 aleblanc: nevertheless, if a product is so old that it is stuck on 2.26 then this must be a newly arisen problem, otherwise the product could not exist. and if it is a new product on that versions, actually selling it is highly problematic due to massive security flaws. in either case, you are in trouble. Jun 16 19:31:03 Letothe2nd: true and I can't argue on that... but it's a reality I live in with this project ... Jun 16 19:31:30 aleblanc: have fun then. Jun 16 19:38:25 RP: I tried to grep for storage.pm, or files named *torage* in my perl build dir on ubuntu 18.04, and nothing came up Jun 16 19:38:48 so I have no idea, other than it's distro specific host contamination maybe Jun 16 19:41:15 kanavin_home: perl-module-storage has them by default ? Jun 16 19:43:47 RP: akanavin@ubuntu1804-ty-3:~/build/tmp/work/core2-64-poky-linux/perl/5.30.2-r0/packages-split$ ls perl-module-storage Jun 16 19:43:47 ls: cannot access 'perl-module-storage': No such file or directory Jun 16 19:44:01 $ corelist -a Storage.pm Jun 16 19:44:01 Data for 2019-12-20 Jun 16 19:44:01 Storage.pm was not in CORE (or so I think) Jun 16 19:44:25 So - if it's there - it must come from somewhere else Jun 16 19:46:31 Sorry, I mean storage, I can't read.type :( Jun 16 19:46:57 *storable* Jun 16 19:47:57 Yes, Storable should arch-dependent Jun 16 19:49:56 sno: I don't understand how it sometimes seems to have a .pm within /x86-64-linux/ and sometimes not :/ Jun 16 19:50:55 RP: with the issue you found few days ago and this, it smells for a Makefile problem Jun 16 19:51:30 sno: yes, could be :/ Jun 16 19:51:47 I remember, ages ago (don't know, maybe some years) I got a ticket for a module I maintain where the target: and target:: etc. where crippled Jun 16 19:52:12 RP: I send you a pribvate mail with the mail contact to the Makefile / Configure maintainer Jun 16 19:52:41 he told me last year in Riga, Linux maintainer shall contact him when they have problems :) Jun 16 19:58:36 sno: thanks. I'll see if I can describe what we're seeing and see if he has any ideas Jun 16 20:02:17 RP: you can add me as Cc or Bcc ;) Jun 16 20:02:59 sno: will do Jun 16 20:03:53 RP: maybe I can convince him to accept patches and I see that I find my module issue with broken parallel make Jun 16 20:11:05 sno: should I mention that other issue or stick to the storable issue? Jun 16 20:16:54 RP: when you also think that it could be a makefile issue - I'd mention both Jun 16 20:17:10 sno: ok Jun 16 20:32:55 RP: installation is performed by installperl script in the perl source tree, which is written in perl Jun 16 20:33:09 figuring out why it sometimes puts things in different places may take a while :-/ Jun 16 20:36:33 kanavin_home: another weird race: https://autobuilder.yoctoproject.org/typhoon/#/builders/87/builds/1030 - I can't even spot the error exactly in that :( Jun 16 20:37:08 kanavin_home: (its gobject-introspection) Jun 16 20:37:36 kanavin_home: right, I don't understand perl enough to know where to start with its makefiles and installer :( Jun 16 20:40:06 RP: I think none of us want to look at installperl, which is 30KB of line noise or close to that Jun 16 20:42:07 kanavin_home: the intermittent failures on the autobuilder are getting out of control :( Jun 16 20:44:59 RP: the g-i failure seems like the qemu in the wrapper script refused to run, but there's nothing printed Jun 16 20:46:04 kanavin_home: I think we may have the builddir that failed for this one if that helps? Jun 16 20:46:25 RP: it could help, yes Jun 16 20:48:05 installperl doesn't look that terrible - but however, installperl isn't parallelized Jun 16 20:48:49 last change on installperl was Thu Oct 4 14:41:03 2018 +1000 Jun 16 20:48:56 RP: can't log in though: Jun 16 20:48:56 k@linux-f9zs:~/development/poky/build> ssh akanavin@ubuntu2004-ty-2.yocto.io Jun 16 20:48:56 Warning: Permanently added 'ubuntu2004-ty-2.yocto.io' (ECDSA) to the list of known hosts. Jun 16 20:48:56 akanavin@ubuntu2004-ty-2.yocto.io: Permission denied (publickey). Jun 16 20:50:13 halstead: around? can you help kanavin_home ? Jun 16 20:50:32 kanavin_home: I just saved that builddir to /home/pokybuild/saved since the worker is still live Jun 16 20:51:05 RP: right, I'll have a look tomorrow then Jun 16 20:51:06 kanavin_home: it has sakoman1's selftest patch which means the builddir is there Jun 16 20:51:30 22:51 is not a good time for this kind of thing :) Jun 16 20:52:21 kanavin_home: no, I'm just ensuring we save it! Jun 16 20:52:34 kanavin_home: tomorrow is fine now we have the data Jun 16 20:53:03 kanavin_home: getting you access is something we need halstead for though Jun 16 21:22:46 kanavin_home, I'm adding your account to the 20.04 LTS workers now. Jun 16 21:23:26 kanavin_home, It's ready to test now. Jun 16 21:50:09 halstead: thanks! Jun 16 21:54:52 sno: I'm not sure I understand the reply :/ Jun 17 00:06:07 I have posted the question in stack overflow on how to build the bootgen utility natively , I have tried a few different options like BBCLASS Extend - native nativesdk but still I get the errors. I am not sure what am I missing. Please help, I have not written the recipes in yocto. Jun 17 00:06:19 https://stackoverflow.com/questions/62401357/how-to-build-bootgen-utility-in-yocto-makefile-system?noredirect=1#comment110364337_62401357 **** ENDING LOGGING AT Wed Jun 17 02:59:57 2020