**** BEGIN LOGGING AT Tue Feb 07 03:00:01 2017 Feb 07 07:32:00 khem, this is a snippet of the recipe, https://bpaste.net/show/6f90e439e877 Feb 07 07:33:54 We have a postprocess which converts a sdcard image into a vsi image (a sparse dd image of sorts). To access the right input, it uses ${DEPLOY_DIR_IMAGE}/${IMAGE_NAME} which I believe is the ones containing DATETIME indirectly Feb 07 07:34:47 How can I check what variable contain DATETIME exactly? Feb 07 08:13:00 Hello everyone, I'm getting this error when compiling yocto "Task 73 (virtual: native: /home/ahmet/krogoht/sources/poky/meta/recipes-devtools/rpm/rpm_5.4.16.bb, do_compile) failed with exit code ' one' "Error rpm_5.4.16.bb, how can I resolve this source error? Feb 07 08:15:27 Hello everyone, I'm getting this error when compiling yocto "Task 73 (virtual: native: /home/ahmet/krogoht/sources/poky/meta/recipes-devtools/rpm/rpm_5.4.16.bb, do_compile) failed with exit code ' one' "Error rpm_5.4.16.bb, how can I resolve this source error? Feb 07 08:21:09 Are there anybody who can help? Feb 07 08:37:33 IMAGE_POSTPROCESS_COMMAND[vardepsexclude] = "DATETIME" is apparently not the fix for IMAGE_POST_PROCESS_COMMAND_append = " vsi_compress;" vsi_compress () { ... Feb 07 08:44:38 sveinse: perhaps vsi_compress[vardepsexclude ] += "DATETIME" ? Feb 07 08:47:07 pohly: testing as we speak Feb 07 09:02:52 Failed as well Feb 07 09:11:47 does someone know if is it possibile to install the "locale-gen" through Yocto? Feb 07 09:52:13 huh Feb 07 09:52:29 khem: for some reason, rebuilding glibc with your 2.25 upgrade meant all the locales disappeared Feb 07 09:52:30 packages/corei7-64-poky-linux/glibc-locale/glibc-locale-ca: FILELIST: removed all items "/usr/share/locale/ca/LC_MESSAGES/libc.mo" Feb 07 10:00:20 rburton: hi Feb 07 10:02:14 maxin: you're still the expert on useradd right? :) Feb 07 10:05:50 rburton: if you insist :) Feb 07 10:09:19 Any pointers in how I can find where I get the taskhash mismatch from? I'm running out of options. The log complains about mismatch in "vsi-image.bb.do_rootfs_wicenv" Feb 07 10:10:05 maxin: turns out useradd.bbclass uses SYSROOT_NATIVE in the PSEUDO_PASSWD assignment that gets into the postinst. which is recipe-specific now. which means every package that uses this class has a pointer to the recipe sysroot in the postinsts. Feb 07 10:15:06 rburton: will check that.. Feb 07 10:19:39 maxin: not sure what the best approach is tbh Feb 07 10:21:02 rburton: what does useradd.bbclass do? Feb 07 10:22:01 handle the creation of users Feb 07 10:24:26 rburton: PSEUDO_PASSWD should not be hardcoded then, and set from environment Feb 07 10:25:48 well, there's no central sysroot anymore at all, so it doesn't really make sense Feb 07 10:28:04 rburton: so why is it set at all? Feb 07 10:39:31 I'm giving up. :( I'll force-run bitbake twice on the build server, ignoring the taskhash error on the first run. Feb 07 10:45:06 hi Feb 07 10:46:14 sveinse: most commonly that's because the task includes DATETIME and that needs to be added vardepsexclude for the task Feb 07 10:46:49 I have a `bitbake` based build system that generates some `qemu` compatible images (kernel, rootfs, initrd), and I'd like to use `wic` instead, to get full images. Any suggestion about how to import `wic` over from yocto without resolving dependency related issues (missing classes, options etc) not too painfully? Feb 07 10:46:56 sveinse: bitbake-diffsigs can help you debug further Feb 07 10:48:32 sveinse: this is possibly incomplete, but https://wiki.yoctoproject.org/wiki/TipsAndTricks/Understanding_what_changed_(diffsigs_etc) Feb 07 10:51:03 khem: definitely something weird, my glibc-locale package here hasn't build any glibc-locale-[code] packages: they're all empty. Feb 07 11:26:48 also `MACHINE=qemux86-64 bitbake core-image-minimal` fails with `./.libs/libgmp.so: error adding symbols: File in wrong format`, apparently `g++` keeps trying to link 32bits objects with `libgmp.so` (which is built on a 64bits host) Feb 07 11:26:57 am I missing something ? Feb 07 11:36:11 joshuagl: Thanks. I'll check it out. I have tested numerous *[vardepsexlude]="DATETIME" without any success yet, but I do believe that it is the solution. You say for the task. A IMAGE_POST_PROCESS_COMMAND runs under do_image_complete, so it should be do_image_complete[vardepsexclude]+="DATETIME" perhaps. It's worth a shot Feb 07 11:39:12 hmm, what happened to https://autobuilder.yocto.io/builders/nightly-ppc/builds/173 ? :( Feb 07 12:01:34 unhappy patchelf? Feb 07 12:04:49 joshuagl: could be. Just not sure why there/now :/ Feb 07 12:06:13 RP: somehow reusing an already stripped binary? Feb 07 12:07:25 joshuagl: its changing the interpreter so it should work regardless Feb 07 12:09:10 fedora24 tmp]$ ./sysroots-uninative/x86_64-linux/usr/bin/patchelf-uninative --print-interpreter /home/pokybuild/yocto-autobuilder/yocto-worker/nightly-ppc/build/build/tmp/sysroots-uninative/x86_64-linux/lib/ld-linux-x86-64.so.2 Feb 07 12:09:10 cannot find section .interp Feb 07 12:14:06 test_postinst_rootfs_and_boot: build an image three times with rpm, deb and ipk... then build it another three times but with systemd instead of sysvinit! Feb 07 12:14:18 * kanavin1 makes note to not run this test when in a hurry again Feb 07 13:20:24 I'm getting emails about an open issue I have that I dont think I can do anything about, could anyone help me out on what to do next? https://bugzilla.yoctoproject.org/show_bug.cgi?id=10669 Feb 07 13:20:25 Bug 10669: normal, Medium, 2.3 M2, linus.wallgren, IN PROGRESS REVIEW , apt package missing systemd files Feb 07 13:22:24 ecksun: is that fixed in master? Feb 07 13:23:04 (looks like oe-core cde67af8ce9b8e72b5818233a7ef4c2aa181404a fixes it to me) Feb 07 13:24:15 rburton, uhm, so no review then? :D Feb 07 13:24:56 ecksun: patch was on the list for review Feb 07 13:25:02 nobody said no Feb 07 13:25:07 very well :D Feb 07 13:25:17 the question is, does that fix the entire bug report Feb 07 13:25:21 if it does then you can close the bug Feb 07 13:25:30 if it is my patch, then I consider my issue fixed :D Feb 07 13:25:46 that sha is ' apt-package: Include maintenance scripts' Feb 07 13:25:52 I havent found that commit yet though (I'm trying to remember how to navigate the yocto project again) Feb 07 13:26:08 joshuagl: if you want these things fixed soon-ish, I'm probably the wrong person Feb 07 13:26:20 http://git.openembedded.org/openembedded-core/commit/?id=cde67af8ce9b8e72b5818233a7ef4c2aa181404a Feb 07 13:26:46 thanks Feb 07 13:26:59 well, that is indeed my commit, alright, so if I think thats fine I just close the issue then, I presume? Feb 07 13:27:01 kanavin1: sure. I'm just resetting to default assignee. They are medium so not *required* for 2.3. You could triage with rburton or he can punt to the triage team? Feb 07 13:27:28 ecksun: yeah, move to resolved fixed, and put the oe-core commit id in the message. Feb 07 13:27:44 cool Feb 07 13:27:51 ecksun: thanks Feb 07 13:28:02 joshuagl: it's also possible that the problem will fix itself by updating those to latest upstream Feb 07 13:28:06 have some buns to eat, will do it straight after that :P Feb 07 13:28:09 right* Feb 07 13:28:36 kanavin1: yes, that's possible. I did retest all of the current recipes before re-assigning the bugs so can be certain they are currently valid Feb 07 13:33:16 joshuagl: it looks increasingly certain we will not continue with rpm5, and rpm4 will replace it btw Feb 07 13:33:38 kanavin1: OK, neat Feb 07 13:33:52 I keep meaning to play with your dnf branch but -ETIME Feb 07 13:37:19 cve-check-tool-native has a task that needs to run a binary produced by cve-check-tool-native. It used to work with "addtask populate_cve_db after do_populate_sysroot" and then assuming the tool is now in path... Feb 07 13:38:19 what location of the tool would be the right choice? Feb 07 13:39:17 joshuagl: the branch is now up to 35 commits Feb 07 13:39:21 I wonder if I can hit 50 Feb 07 13:39:51 jku: does it add cve-check-tool-native to DEPENDS? Feb 07 13:40:27 rburton: does that actually work, adding yourself to depends? Feb 07 13:42:04 for context, this what the cve-check class does: do_cve_check[depends] = "cve-check-tool-native:do_populate_cve_db" Feb 07 13:43:31 oh hu Feb 07 13:43:39 sorry didn't read Feb 07 15:09:12 hi Feb 07 15:09:55 i have been using bash syntax for conditionals in some places in .bb files. But sometimes it looks like python. What is preferred method for top level conditonals where its not inside functions identified by {P}? Feb 07 15:54:21 I have a target package that generates some files that another target package wants during build, but these files are never needed on the target Feb 07 15:54:35 I guess I should just put them in a -dev package...? Feb 07 15:54:54 yes Feb 07 15:55:50 YPTM: Ready-Access Number: 8007302996 Access Code: 2705751 Feb 07 15:55:55 YPTM: Stephen joined Feb 07 15:56:11 YPTM - armin is on Feb 07 15:58:45 YPTM - Armin is on Feb 07 15:59:09 YPTM - Stephano is on Feb 07 15:59:10 YPTM - Sona is on Feb 07 15:59:14 Hi Armin Feb 07 15:59:26 hello Feb 07 15:59:45 YPTM: ross on Feb 07 15:59:51 YPTM — Joshua joined Feb 07 16:00:05 Can anyone (like me) join silently out of curiosity? Feb 07 16:00:25 Yes Feb 07 16:00:38 All interested in YP are welcome to join Feb 07 16:02:19 YPTM: Jussi here Feb 07 16:02:55 sveinse is on Feb 07 16:03:12 YPTM: Saul is here Feb 07 16:03:55 YPTM: Richard joined Feb 07 16:05:17 cve check tool Feb 07 16:05:22 I have problem with cve-check tool Feb 07 16:27:00 YPTM is over. Thanks! Feb 07 16:27:25 Sona: so the rewrite of cve-check-tool was abandoned. the author is now looking at using https://github.com/coreos/clair instead for their purposes. Feb 07 16:27:37 Thank you, this was very interesting Feb 07 16:29:15 Sona: which interestingly has backends for redhat, debian, canonical and more Feb 07 16:31:31 pohly: I'm seeing a failure cve-check-tool failure where the fetch fails with CURLE_SSL_CACERT_BADFILE, is this similar to what your patch in meta-security-isafw fixes? Feb 07 16:31:56 jku: yes, that might be it. Feb 07 16:32:54 I kept wondering why no-one seemed to care when I mentioned that problem when cve-check-tool was added to OE-core ;-} Feb 07 16:33:28 because i have the bug to solve it once and for all, but haven't got around to it :/ Feb 07 16:35:14 rburton, pohly: I'm not sure if this is the same thing though. I'm just running the binary from terminal (on my build host) and it still fails Feb 07 16:35:38 probably because its looking for a sysroot that doesn't exist? Feb 07 16:36:08 maybe so , I'll take a look Feb 07 16:36:13 True, now with recipe specific sysroots it's probably failing even harder. Feb 07 16:36:37 Still, explicitly passing a path for the CA database should work. Feb 07 16:38:12 yeah you're right: Feb 07 16:38:20 open("/mnt/extra-ssd/tmp/work/x86_64-linux/curl-native/7.52.1-r0/recipe-sysroot-native/etc/ssl/certs/ca-certificates.crt", O_RDONLY) = -1 ENOENT (No such file or directory) Feb 07 16:41:25 would it be wrong for curl-native to use host certificates? Feb 07 16:41:44 less wrong than a path that doesn't exist Feb 07 16:41:50 that was a patch i had locally for a bit tbh Feb 07 16:49:07 rburton, RP: some broken symlink for a recipe specific sysroot: tmp-glibc/work/*/initramfs-framework-refkit-dm-verity/1.0-r0/recipe-sysroot-native/usr/bin/../../usr/lib/ssl/openssl.cnf -> ./../../../../../../work/x86_64-linux/openssl-native/1.0.2k-r0/recipe-sysroot-native/etc/ssl/openssl.cnf Feb 07 17:03:17 pohly: right, jku has a bug Feb 07 17:05:09 RP: oh about that... what would you like me to do with it? you had a WIP patch in your branch that added a bbclass for this Feb 07 17:08:12 jku: you didn't find any other changes needed? Feb 07 17:10:20 can a bblayers.conf.sample inherit from another conf? Feb 07 17:11:17 RP: no, those three (and the gcc ones you handled) were the only ones. Now that you ask I'm not sure if I built world or not... I can do that and double-check Feb 07 17:39:44 rburton: If the author of cve check tool abandoned, shouldn't we also use another tool? Feb 07 17:40:22 justanotherboy: suggestions welcome Feb 07 17:40:54 personally i'd basically clone the fetcher bit of clair in python to write a new tool Feb 07 17:41:00 but that assumes i have the time Feb 07 17:42:05 rburton: Definitely we need to check on clair Feb 07 17:42:17 rburton: As you said the problem is time Feb 07 17:48:15 jku: thanks, I think I got confused about who was doing what Feb 07 18:42:02 I've used yocto to build a system with systemd, I am trying to configure the network interfaces to come up at boot, however there is no /etc/network/interfaces file, so I assume systemd is handling this. Any suggestion on where to configure these interfaces? Feb 07 18:46:05 rburton: "multiple layers cannot agree between themselves about issues" should not be a valid reason for adding stuff to oe-core Feb 07 18:47:50 khem: ^^^ Feb 07 19:10:38 mcudev: idk exactly what your issue is, but there should be a network service file, somewhere in /usr/lib/systemd if Im not mistaken, it may be a good starting point Feb 07 19:13:35 rburton: nothing that I see, should impact locale Feb 07 19:15:40 kanavin_home: its not about agreeing, its about fragmentation and how it can help in avoiding that, if layers dont agree that can not be resolved in anycase. Feb 07 19:16:32 go is a formidable language of the web and we need to have better support for it Feb 07 19:18:27 using OE for docker containers is a common usecase, Feb 07 19:28:08 khem: by adding things to oe-core you swap one problem with another (bloat, maintenance burden) Feb 07 19:28:19 khem: if that can be avoided, it should be Feb 07 19:28:37 e.g. by making an officially blessed layer Feb 07 19:32:30 that goes unsaid, should be weighed with value Feb 07 19:33:24 rburton: is this only happening with the latest bump to 2.25 release or did snapshot also had this issue I wonder Feb 07 19:39:29 aehs29: thanks! Feb 07 20:07:31 how do i make bitbake to show the messages in "old" scrolling terminal form ? Feb 07 20:10:42 Could http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?id=eb937ee0a1ea1e695aa995c2d1a3561ba12335f8 (fix gdb in sdk when python is enabled) be backported to morty? Feb 07 20:11:37 rob_w: easiest is just pipe to cat. bitbake | cat, since bitbake goes based on stdout being a tty Feb 07 20:12:01 right i though there would be such way too but thats what i need Feb 07 20:12:08 thx kergoth Feb 07 20:12:26 usually tools provide an arg to force the matter, but i'm not sure we ever added one (and honestly, bitbake has too damn many cmdine options as it is :) Feb 07 20:12:31 jmesmon: send a backport patch, should be simple enough Feb 07 20:12:55 its just for some fancy presentation thingi .. so dont worry Feb 07 20:13:36 i got a tower of embedded devices with displays but vertically on top of each other Feb 07 20:14:20 so if make each of em look at the cat pipe of bitbake with their respected offset in lines, i should get a nice big scrolling screen to visualize bitbake@work Feb 07 20:15:43 ha, nice Feb 07 20:16:50 glad u like it ;-) its just a eyecatcher thingi for a local linux group community event ;-) Feb 07 20:25:40 rburton: I've just done this, but is there a standard format for the subject I should follow? Is doing "[morty][PATCH] " expected for OE-Core changes? Or something else? Feb 07 20:26:40 Also: if the patch needed no changes (direct cherry-pick from master worked), should I preserve the original commit author? (I've done so for this one as it is the default) Feb 07 20:29:12 jmesmon: yes that's the correct format, and yes preserve the author Feb 07 20:31:31 I also made the guess that including a new "(From OE-Core rev: ...)" line with the master sha would make sense. Is that correct to do for backports? Feb 07 20:32:45 jmesmon: that's helpful yes... when I was doing it I would say "From OE-Core master rev:" but that's not mandatory Feb 07 20:32:58 ah, that's a good point. Feb 07 21:00:32 omg .. i just had the perfect idea for a slogan for that tower showing bitbake messages .. i masquerade it as huge joint , to roll your own linux ?? he he he Feb 07 21:03:30 bonus points if it lets out the magic smoke ;) Feb 07 21:03:59 not in my country .. Feb 07 21:04:05 not yet at least Feb 07 21:04:08 ;_) Feb 07 21:05:13 but a simple smoke machine ontop should make some people nervous at that event Feb 07 21:07:41 I was referring of course to magic smoke out of the electronics, but then I guess it wouldn't work too well afterwards ;) Feb 07 21:57:17 bluelightning: it is hard to get it back in... **** ENDING LOGGING AT Wed Feb 08 03:00:02 2017