**** BEGIN LOGGING AT Tue Feb 14 03:00:03 2017 Feb 14 08:55:59 Is there a simple way to strip the deploy/ipk directory to only include the packages that are installed in the image (to generate a minimal package feed)? Feb 14 09:18:55 Hi all, Feb 14 09:19:15 Has anybody knows how to enable --python support in receipe meta-toolchain ? Feb 14 09:19:45 The gdb is built, but I cannot use kernel python scripts Feb 14 09:20:38 if there is an event handler set on "bb.event.BuildCompleted", is there a native sysroot available during the event handler run? Feb 14 09:26:44 what's the applying order of .bbappend files? Feb 14 09:27:14 I mean, if I have pn_1.0.bbappend and pn_%.bbappend in the same layer, which one gets applied first? Feb 14 09:54:18 hello Feb 14 09:54:28 by default, yocto installs rpm5 ? Feb 14 09:54:57 when rpm is enabled, that is Feb 14 09:57:33 cornel: for now at least Feb 14 09:57:43 jku, thanks. Feb 14 09:57:53 anybody knows the rationale behind this? Feb 14 09:58:22 cornel: behind what? Feb 14 09:58:33 or at least can anybody telll me how to prevent several packages installing same file in a random order? Feb 14 09:58:46 jku, behind choosing rpm5 as the default Feb 14 10:00:19 i know that here in our project at least procps and busybox install same file, and sometimes procps is the last, but some other times is busybox Feb 14 10:00:35 giving us a random state system Feb 14 10:01:02 is there a way to make rpm5 behave in deterministic way? Feb 14 10:01:40 you don't want two packages trying to install the same file, even if they do so in a deterministic order. Best to fix the recipes to use alternatives Feb 14 10:02:03 joshuagl, sure, but redhat's rpm will tell me if this is the case Feb 14 10:02:08 unlike rpm5 Feb 14 10:02:24 so i don't really know how many other files are in this situation Feb 14 10:02:53 is there a way to use redhat's rpm instead of rpm5? Feb 14 10:03:26 cornel: kanavin has a huge set of patches on the list that does taht among other things... Feb 14 10:03:28 joshuagl, will tell me because it will not allow it :) Feb 14 10:03:48 of course, but how easy that is depends on which version you're using Feb 14 10:04:01 2.0 Feb 14 10:04:56 http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta-yocto/conf/local.conf.sample.extended?h=jethro#n375 Feb 14 10:05:47 jku, i would choose rrpm4 if possibel Feb 14 10:06:12 cornel this is almost certainly a wrong way to solve the problem -- switching rpm version is a huge change Feb 14 10:06:28 bound to give you lots more head aches Feb 14 10:06:45 jku, right, but using non-deterministic rpm is a big problem for me Feb 14 10:07:05 joshuagl, thank you Feb 14 10:19:34 ok, found an lwn article from 2011 Feb 14 10:49:26 Hi, I need help. I try to cross compile a rust binary, which itself need some compiled C. It looks the objcopy is wrong or missing. This is the output https://gist.github.com/zzeroo/c8fdc52f5c8d289b53c3eb1c3d778077 Feb 14 10:50:10 I use the meta-rust-bin receipes Feb 14 11:15:19 cornel: this has nothing do with rpm5 or rpm4 (but rather the order in which repo manager such as smart or dnf chooses to install the packages), and if two recipes are installing the same file, you need to fix the recipes Feb 14 11:16:14 dnf will probably not give you any guarantees about it either Feb 14 11:19:55 Hello, I have a problem with yocto and the QA error installed but not shiped. I have different a lot of cmake files which an install( TARGETS.... ) a few of the files are working, for the other I get the error installed but not shipped. Do you have any idea how I can solve that? Feb 14 11:20:04 rburton: a little help with isafw / cve-check-tool? https://github.com/01org/meta-security-isafw/issues/79#issuecomment-279624850 elena wants to run cve-check-tool from a event handler for "bb.event.BuildCompleted". Is there a way to ensure there is a proper sysroot with cve-check-tool-native available? Feb 14 11:20:26 jku: have a look at ross/index on contrib Feb 14 11:22:13 ross so staging_populate_sysroot_dir(), set STAGING_DIR_NATIVE... thanks Feb 14 11:26:06 kanavin, ok, i'll check if yum is used to install the packages (i believe yes) Feb 14 11:26:11 kanavin, thank you Feb 14 11:28:39 cornel: your problem is something else entirely: two packages should not install the same file in the first place Feb 14 11:28:57 nothing to do with yum or smart, or dpkg, or whatever is the backend Feb 14 11:29:24 kanavin, i'm pretty sure rpm4 does not allow this but rpm5 does allow it. therefore i have a hard time accepting this rpm5 thing Feb 14 11:29:47 rburton, what has for your final slanf Feb 14 11:30:03 still finding the oneXXX I HATE SLANG Feb 14 11:30:04 I really thought we had a warning for installing files multiple times when building rootfs... Feb 14 11:30:24 jku: I think so too Feb 14 11:30:48 jku, can you tell me in what log would i find such warnings? Feb 14 11:31:49 cornel: printed on stdout when running bitbake Feb 14 11:31:59 kanavin, ah, thank you Feb 14 11:32:31 i would definitely do not sit for hours looking at a display hoping to see a warning :) Feb 14 11:34:20 cornel: you'll probably find it in workdir/arch/image/version/temp/log.do_rootfs too (substitute with your local values as needed, or issue a find command) Feb 14 11:34:30 Hunk: installed-not-shipped means 'make install' is installing the fieles, but nothing in FILES_* matches them Feb 14 11:34:33 kanavin, thank you Feb 14 11:34:36 gonna check it Feb 14 11:35:10 Crofton|work: try ross/mut2 now, i think an earlier push failed Feb 14 11:41:38 jku: awww, mesa dropped its comedy sha wrapper Feb 14 11:41:54 :) Feb 14 11:44:52 rburton: the actual sha1 implementation is ~200 loc. The wrapper was 250 lines (plus 150 lines of configure) Feb 14 11:45:49 yeah Feb 14 11:45:53 been reading the commit Feb 14 11:46:02 and some of the "fun" bugs Feb 14 12:00:52 rburton: once more please :) (branch been force pushed to clear the clutter a bit) Feb 14 12:07:07 kanavin: https://autobuilder.yoctoproject.org/main/builders/nightly-x86-64/builds/1123 Feb 14 12:26:39 Hello, I have a problem with installing my packages with cmake. I've a lot of cmake packages whith Install definitions. Because of that the shipping to the iage should work. But I get the error installed but not shipped on a few cmake recipes ( not all ). But the cmake files are always the same. Anyone has an idea how what could be the problem with this problem? Feb 14 12:27:09 rburton: one of last night's autobulders went green, but the other one didn't - and I found issues to fix in the logs even in the green one Feb 14 12:27:17 rburton: so now it should be all sorted Feb 14 12:38:26 Hello, I have a problem with installing my packages with cmake. I've a lot of cmake packages whith Install definitions. Because of that the shipping to the iage should work. But I get the error installed but not shipped on a few cmake recipes ( not all ). But the cmake files are always the same. Anyone has an idea how what could be the problem with this problem? Feb 14 12:43:00 rburton, yep I see it now Feb 14 12:47:38 Hey, I have a question regarding multiple kernel images: Is it possible to include 2 kernel images (fit) in one image (ext4 rootfs)? I'd like the bootloader to decide which kernel to boot, depending on some configurations. Feb 14 13:28:34 Hi, I have a patch to send to oe-core but it also need documentation, how to proceed since documentation can be found in poky repository but not oe-core one ? Feb 14 13:46:44 rburton: ugh, bad breakage :-/ Feb 14 13:46:52 rburton: I've reproduced locally Feb 14 13:47:37 geoffrey_l: open a documentation bug in yocto bugzilla, post your suggested change there Feb 14 13:48:05 kanavin: Thanks Feb 14 13:48:15 geoffrey_l: the docs are http://git.yoctoproject.org/cgit/cgit.cgi/yocto-docs Feb 14 13:48:31 geoffrey_l: provide a link to the patch that the documentation change is against Feb 14 14:02:12 Hello, I've a question about the QA error : [ installed vs shipped]. I'm using a cmake file to install my compiled targets, but this targets isn't shipped to the image. Anyone has an idea what could be the difference on install vs shipped Feb 14 14:03:30 the files installed by your CMakeLists.txt are not completely assigned to packages by the FILES_* variables in your recipe Feb 14 14:05:17 Thank you for the response I Will check this Feb 14 14:05:27 but shoudln't be this happen automicly ? Feb 14 14:05:48 Hunnk: there's no way to make that fully automatic Feb 14 14:06:49 Hunnk: also sometimes you really don't want to ship all the files -- then you can add a do_install_append() where you delete those files instead Feb 14 14:06:50 ok just the file or with directory ? Feb 14 14:07:15 I just added now Feb 14 14:07:26 FILES_ += myExecutable Feb 14 14:11:10 Hunnk: if you grep for FILES_ in meta/ you will find lots of examples. If you post a link to your recipe maybe someone will advice on your specific case Feb 14 14:18:22 hi everyone Feb 14 14:18:35 i have errors like : ERROR: Taskhash mismatch dc3543e66b52acc379ec340f0c3f7703 verses 581ae57e28e43d6c41e72240a2c01f04 for ... Feb 14 14:19:06 i there a way to cleanup those hash? I can't get through these erros... Feb 14 14:19:57 I would like to cleanup the hash Feb 14 14:29:59 @jku ty I will check this Feb 14 14:48:06 Is it possible to check the value of the FILE_* variable ? Feb 14 14:48:10 for a package? Feb 14 14:48:43 bitbake -e $recipe | grep -E '^FILES_' Feb 14 14:55:45 <_Ben> What is the best way to modify the L2 cache `AUX_CTRL` register using Yocto? I'm building for an ARM A9 Feb 14 14:58:20 _Ben: should be not related to yocto at all. this is something that the kernel, respective its init is supposed to take care of. Feb 14 15:03:16 <_Ben> LetoThe2nd: Yes, I'm wondering if there is a way to use Yocto to modify this instead of modifying the kernel source directly Feb 14 15:04:16 _Ben: it is actually supposed to be the other way round. you modify your kernel source, create a path file out of the modifications, and then include that in your OE-powered buildprocess. Feb 14 15:05:41 <_Ben> LetoThe2nd: I modify the defconfig using a .bbappend, there's nothing similar for this? Feb 14 15:06:15 _Ben: you apply a patch file using a bbappend. Feb 14 15:06:20 (for example) Feb 14 15:07:58 <_Ben> Thank you :) Feb 14 15:14:24 kanavin: i've tested your dnf-rpm4 patchset a bit Feb 14 15:15:00 kanavin: it has worked fine for me, no issues Feb 14 15:16:18 kanavin: however, i tried enabling rpm signing and run into issues Feb 14 15:18:20 marquiz: rpm signing is not supported yet, working on it Feb 14 15:21:25 Hello, I've a question about the gitsm fetcher. When I use this fetcher the source files get copied in my Workdir in the unpack step, which takes really long. Is it possible to generate an symlink or something like this ? Feb 14 15:44:53 rburton: ... and once more :) Feb 14 15:46:55 Hi guys! How to disable check - "do_package: QA Issue: split_and_strip_files:" ? Feb 14 15:47:12 I tried INSANE_SKIP_${PN} += "magic_file" INSANE_SKIP_${PN} += "split_and_strip_files" , but it doesn't help me Feb 14 15:48:57 I got error - ERROR: chromium-58.0.3007.3-r0 do_package: QA Issue: split_and_strip_files: 'file /u02/embb/build/tmp-glibc/work/embb_r5-oe-linux/chromium/58.0.3007.3-r0/package/usr/bin/chromium/chrome' failed [split-strip] Feb 14 15:49:16 but anyway I want to check this image Feb 14 15:49:42 file command output is " /u02/tavolga/build/tmp-glibc/work/tavolga_r5-oe-linux/chromium/58.0.3007.3-r0/package/usr/bin/chromium/chrome: ERROR: ELF 32-bit LSB shared object, MIPS, MIPS32 rel2 version 1 (SYSV), dynamically linked, interpreter /lib/ld.so.1error reading (Invalid argument)" Feb 14 15:50:06 z Feb 14 15:59:14 INHIBIT_PACKAGE_STRIP and INHIBIT_PACKAGE_DEBUG_SPLIT =1 skip it Feb 14 16:31:19 * reanguiano looks into EGL initialization Feb 14 17:01:49 zeddii: kernel 4.9 breaks mips Feb 14 17:07:30 well i wish i knew what changed between yesterday and today which means that harfbuzz doesn't build for me Feb 14 17:21:59 I'm trying to patch this file http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/recipes-core/udev/udev-extraconf/mount.sh with bitbake and I get "can't find file to patch". I think it is an issue of paths because the file is sourced from the recipe repo rather than a source package. But the recipe has S = "${WORKDIR}", which I thought would redefine the patchdir appropriately... How do I target this file? Feb 14 17:37:53 ah! got it working with SRC_URI_append = " file://xxx.patch;patchdir=${WORKDIR}/" Feb 14 17:56:03 :q Feb 14 19:11:28 How do I enable thumb when building with MACHINE=qemuarm (using just OE-core and bitbake, no DISTRO set)? Feb 14 19:12:15 pohly: #PREFERRED_ARM_INSTRUCTION_SET ?= "thumb" Feb 14 19:12:15 #ARM_INSTRUCTION_SET = "${PREFERRED_ARM_INSTRUCTION_SET}" Feb 14 19:12:16 i think Feb 14 19:12:25 at least i found that in my local.conf :) Feb 14 19:13:15 rburton: that's your long-term memory, right? ;-) Feb 14 19:13:55 yeah Feb 14 22:05:51 Hello - I have two different initramfs recipes, for booting in different circumstances. I'm only seeing one of them in build/tmp/deploy/images though. Feb 14 22:05:55 How can I get both generated? Feb 14 22:06:17 I don't understand the question Feb 14 22:06:39 it's an image. either you explicitly bitbake it, or something depends on it (generally via the INITRAMFS variables) Feb 14 22:06:45 it's not going to magically build every image it finds Feb 14 22:07:38 So, in my normal image, there's an INITRD_IMAGE = "${BPN}-initramfs". Can I force it to also build ${BPN}-initramfs-wifi somehow? Feb 14 22:07:59 I tried to bitbake my-client-image-initramfs-wifi, and it did stuff and completed. Feb 14 22:09:34 that depends on your goals, and what release version you're using, but you can use the 'depends' flag of a task to depend on anything else. Feb 14 22:09:49 i.e. do_image_complete[depends] += "some-other-image:do_image_complete" or whatever Feb 14 22:10:01 Ah, right. That target. Feb 14 22:10:22 but if this image isn't going to use the other image until runtime, i'd just have my automated builds run the appropriate bitbake commands explicitly, rather than implicitly requiring that they all build when you bitbake the main image Feb 14 22:16:44 Got it. I forgot to change IMAGE_BASENAME when I copied the initramfs recipe. **** ENDING LOGGING AT Wed Feb 15 03:00:01 2017