**** BEGIN LOGGING AT Wed Nov 19 02:59:59 2014 Nov 19 07:26:32 good morning Nov 19 07:35:09 morning all Nov 19 08:11:48 good morning Nov 19 08:37:31 morning all Nov 19 08:43:22 hi bluelightning Nov 19 08:43:37 hi mckoan Nov 19 09:18:46 anyone who knows how to make a tarball with e.g. my kernel-image package and only that? Nov 19 10:13:21 Hi all, I've just started looking at the bootloaders in Yocto and couldn't find what the LABEL=boot or LABEL=install-efi was used for in grub for the kernel command line. Would anyone be able to explain/point me? Nov 19 10:28:17 <_valle_> Having some problems after moving from daisy to dizzy branch. Anyone have any clue of my problems? http://pastebin.com/Ncvw6wrC Nov 19 10:28:57 _valle_: did you keep the same TMPDIR? Nov 19 10:29:18 <_valle_> bluelightning: No i deleted the tmp folder Nov 19 10:29:47 <_valle_> bluelightning: Is there anything else you should do when changing branch? Nov 19 10:33:49 hey.. how to tell bitbake only to setup the rfs -- so take all things from sysroot and pack into a rfs (ext3..)? i did bitbake myimage -c rootfs Nov 19 10:34:06 _valle_: there is the migration guide: http://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#moving-to-the-yocto-project-1.7-release Nov 19 10:34:18 but bitbake builds things again that are already built and in sysroot Nov 19 10:34:26 _valle_: but to be honest I don't see any errors in the list that would be solved by anything in that migration documentation Nov 19 10:35:04 _valle_: the gcc-cross-initial failure would be the first one to look at - can you check config.log to see if there is an actual error other than "cannot compile"? Nov 19 10:35:11 but maybe it is because i just added IMAGE_INSTALL_append to my local.conf before Nov 19 10:35:54 ericbutters: something else must have changed between the last time you built and now, that would have triggered re-execution of tasks Nov 19 10:44:18 <_valle_> bluelightning: Unfortunately there is no more error in the log... Nov 19 10:49:50 _valle_: which host distro are you building on? Nov 19 10:50:08 <_valle_> bluelightning: Ubuntu 14.04 Nov 19 10:50:39 hmm ok Nov 19 10:50:58 I've definitely successfully built gcc-cross-initial on that distro Nov 19 10:57:02 <_valle_> bluelightning: Worked when I used qemux86 as machine. Will start on clean slate and erase everything and test one more time with my machine. Nov 19 10:58:16 _valle_: can you please check for other config.log files? I've been told there is more than one for gcc-cross-initial Nov 19 10:58:41 _valle_: it would really help if rather than erasing everything we could try to debug the issue Nov 19 10:59:05 jabk: you mean if Yocto supports a way to output only the kernel image when building? Nov 19 10:59:24 <_valle_> bluelightning: Okay I wait with the erasing, but I'm going away for lunch now. Get back to you later Nov 19 10:59:38 _valle_: ok, thanks Nov 19 11:00:19 kimrrh: well sort-a: suppose I have some files e.g. u-boot.imx and uImage that I want in a tarball and I would like to automate it in Yocto Nov 19 11:01:46 kimrrh: I think I need to create my own class e.g. "archive.bbclass" that does this (simple) thing - I was just wondering if not this was supported in Yocto already Nov 19 11:17:05 Hi guys. It is now possible to match any version using % symbol in bbappen, for example: https://git.congatec.com/yocto/meta-openembedded/commit/5b8498b53dcdf2ed797cbd2049dd3856ef17e2ff Nov 19 11:17:41 but is it possible to match only major version, something like util-linux_2.%.bbappend ? Nov 19 11:18:22 diego_r: yes that should also work AFAIK Nov 19 11:18:42 I believe its solely a string match on the filename Nov 19 11:18:43 thanks omnipresent bluelightning Nov 19 11:19:30 bluelightning: btw, I've patched watchdog initscript and it now in testing in a bbappend in my layer Nov 19 11:20:00 I will send in the patch for the watchdog recipe when I have time Nov 19 11:21:07 diego_r: ok, that would be great, thanks Nov 19 11:22:47 bluelightning: could you pls take a look here? http://paste.ubuntu.com/9093537/ Nov 19 11:23:29 ericbutters: that is strange... Nov 19 11:23:45 otavio_: JaMa: ^ Nov 19 11:33:26 Hi. I'm debugging a cross build and noticed some binaries are linked against the wrong dynamic linker. I used -v as argument to gcc and noticed the linker is not given -dynamic-linker. What could cause -dynamic-linker not to be given as argument to the linker? Nov 19 11:35:24 bluelightning: otavio_: JaMa: i think it is related to the branch i checkout out with meta-qt5.. so here i use dizzy, but my yocto/poky is 1.5.1 -- i did because i need to use qt-5.3 -- how to fix the md5 issue? Nov 19 11:36:37 ericbutters: revert the patches which updated LIC_FILES_CHKSUM to match new LICENSE file from daisy Nov 19 11:38:46 JaMa: okay thanks! i give a try Nov 19 11:40:07 Wrt the linker issue, I also noticed COLLECT_LTO_WRAPPER is not set. Don't know if it is related to the problem. Nov 19 12:16:28 JaMa: one other thing, i i say populate_sdk, does bibake than also build the sdk? Nov 19 12:17:50 how to rebuild the sdk? Nov 19 13:06:30 <_valle_> bluelightning: Here is a paste of config.log - http://pastebin.com/DY2Jy53q Nov 19 13:16:25 ericbutters: yes Nov 19 13:19:38 bluelightning: thanks, i can not find qmake in the sdk, so i have to do 'bitbake meta-toolchain-qt5' ? Nov 19 13:22:07 ericbutters: right it will not add the tools automatically unfortunately, but you can do TOOLCHAIN_HOST_TASK_append = " " Nov 19 13:22:19 (that would be from your image recipe) Nov 19 13:22:54 _valle_: can you send a copy of /poky/build/tmp/work/x86_64-poky-linux-gnueabi/gcc-cross-initial-arm/4.9.1-r0/gcc-4.9.1/build.x86_64-linux.arm-poky-linux-gnueabi/./gcc/as ? Nov 19 13:23:11 _valle_: you can email it to me if it's too large for pastebin Nov 19 13:23:24 what is then for qt5? to have all needed libs/header/bin (qmake) there? Nov 19 13:24:34 ericbutters: I'm not sure but I'd be looking at what the meta-toolchain-qt5 recipe adds to that variable to find out Nov 19 13:24:45 (i.e. that's what I'd recommend) Nov 19 13:24:46 thanks Nov 19 13:33:42 <_valle_> bluelightning: Sure, here you go http://pastebin.com/arBCankf Nov 19 13:44:39 ericbutters: yes; -c populate_sdk does rebuild it Nov 19 14:07:01 Hi, everyone, I am compiling Qt5 envir with no-xcb and eglfs, now it shows an compile error as Qt/5.3/Src/qtbase/lib/libQt5Quick.a(qsgdefaultdistancefieldglyphcache.o):qsgdefaultdistancefieldglyphcache.cpp:function QSGDefaultDistanceFieldGlyphCache::resizeTexture(QSGDefaultDistanceFieldGlyphCache::TextureInfo*, int, int): error: undefined reference to 'QOpenGLFunctions_3_2_Core::versionProfile()' Nov 19 14:07:24 does anyone meet the problem before? Nov 19 14:08:05 This is configure command line I used : ./configure -opensource -confirm-license -no-xcb -eglfs -opengl es2 Nov 19 14:09:12 _valle_: ok, so I've just run it past someone else and apparently those ORIGINAL_*_FOR_TARGET values in that script should not be empty, so that is a clue to where the problem lies Nov 19 14:32:35 Hello! Question about SDK tar.bz2 files. Why does it not contain SDK name and version in main directory name and instead has everything from .? Why is there a separate .sh installer/wrapper which then tries to install to correct path? I'd expect the tar.bz2 also have the name and version in a directory name so that multiple SDK version can be installed on same machine. Nov 19 14:33:24 ^ details from tar_sdk() in meta/classes/populate_sdk_base.bbclass Nov 19 14:40:57 or another way to ask, what would break if the tar ball had the SDK name and version in it. I assume the .sh installer. Nov 19 14:41:06 Hi, I'm writing a small recipe like helloworld, I can't figure out how can I install the resulting executable in /home/root? Nov 19 14:41:59 in do_install() I have install -d "${D}${sbindir}", then I set FILES_${PN} = "${sbindir}" Nov 19 14:42:15 this works but put the executable in /usr/sbin Nov 19 14:43:04 I need something like this: install -D -m 0755 ${S}/testapp ${D}/home/root/testapp Nov 19 14:44:04 yoooc: typically recipes don't install files to a home directory Nov 19 14:44:22 yoooc: but if you must, you'd just need to ensure the directory gets added to FILES_${PN} Nov 19 14:44:52 the home directory for the root user is defined by the ROOT_HOME variable, FWIW Nov 19 14:46:54 bluelightning: if i change DISTRO_FEATURES in my local.conf, is it needed to remove tmp/work?? Nov 19 14:47:26 ericbutters: you probably should remove tmp if you do that yes, and not just tmp/work Nov 19 14:47:35 okay Nov 19 14:47:53 bluelightning:FILES_${PN} = "${D}/home/root" worked Nov 19 14:48:20 bluelightning:so easy. I didn't find anything about it googling Nov 19 14:50:03 yoooc: hmm, we do have http://www.yoctoproject.org/docs/current/dev-manual/dev-manual.html#new-recipe-writing-a-new-recipe but it does look like we don't actually describe the package splitting process in sufficient detail Nov 19 14:51:58 bluelightning:actuall now I have problems with the deployment in the image: * opkg_install_cmd: Cannot install package testapp Nov 19 14:52:22 cannot install why? does it say? Nov 19 14:53:06 yoooc: oh, btw that FILES value is invalid... drop the ${D} there Nov 19 14:53:24 didn't it warn you about unpackaged files? Nov 19 14:57:48 bluelightning:I set FILES_${PN} = "/home/root" but is the same opkg_install_cmd: Cannot install package testapp Nov 19 14:58:06 yoooc: yes but what does it say after that? Nov 19 15:01:18 _valle_: I think at this point it would be great to have a bug filed to track this - would you mind filing one and attach the log / script pasted earlier? Nov 19 15:05:08 bluelightning: one more thing ;) after installing poky-eglibc-i686-meta-toolchain-qt5-cortexa9hf-vfp-neon-toolchain-1.5.1.sh there is export OE_QMAKE_AR="arm-poky-linux-gnueabi-ar" Nov 19 15:05:37 ey.. OE_QMAKE_* missing in environment-setup-cortexa9hf-vfp-neon-poky-linux-gnueabi Nov 19 15:06:12 right, yes, that is another thing - you'll need to do whatever the meta-toolchain-qt5 recipe does in order to handle that Nov 19 15:07:02 why? i managed to setup by hand.. right now i added all qt5 things i need to IMAGE_INSTALL_append in local.conf and removed tmp/ to do a clean build again Nov 19 15:10:27 hi, since today I started getting errors in yocto that mkfs.ext3 can't recognize -d switch Nov 19 15:11:10 so build fails and image can't be created Nov 19 15:11:57 I have opened poky/meta/classes/image_types.bbclass and what I see Nov 19 15:12:39 mkfs.$fstype -F $extra_imagecmd ${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.$fstype -d ${IMAGE_ROOTFS} Nov 19 15:13:46 but I didn't found any documentation that could explain -d switch Nov 19 15:14:04 how that could be? Nov 19 15:15:40 until today I was able to build everything Nov 19 15:16:17 today I added DISTRO_FEATURES_append = " x11 pam systemd opengl alsa" and VIRTUAL-RUNTIME_init_manager = "systemd" Nov 19 15:16:29 to local.conf Nov 19 15:16:50 any ideas? Nov 19 15:17:06 -d is "populate the new rootfs with this content" Nov 19 15:17:27 so its likely that your build isn't using our e2fsutils-native, but your hosts Nov 19 15:17:52 have you been poking at the sysroots? Nov 19 15:17:52 ericbutters: well it's up to you, if you're happy to set up manually you can - but you could also make the changes required to add the appropriate lines to the env setup script by default Nov 19 15:18:16 rburton, good input Nov 19 15:18:46 how I can return it back into my build? Nov 19 15:19:02 well, did you mess with the sysroot? Nov 19 15:19:24 deleting tmp and letting it re-populate from sstate will be one solution, hopefully Nov 19 15:19:43 don't delete your sstate-cache and you'll rebuild in minutes Nov 19 15:19:50 ah Nov 19 15:20:16 recently I had deleted my sstate-cache manually Nov 19 15:20:26 seems that the reason Nov 19 15:20:40 doubt it Nov 19 15:20:51 thank you! I'll try this right now Nov 19 15:21:12 fwiw http://git.kernel.org/cgit/fs/ext2/e2fsprogs.git/tree/misc/mke2fs.c#n126 <— -d is upstream, but upstream is slow at making releases Nov 19 15:21:43 anyway, I'll try from scratch Nov 19 15:27:40 thank you for the link. I had idea that yocto uses custom mke2fs, but I was not sure. Nov 19 15:38:50 rburton, around? Nov 19 15:42:19 ikey: hey Nov 19 15:42:55 heya, pm? :) Nov 19 16:14:13 bluelightning:no way, I have to stay in /usr/sbin Nov 19 16:15:24 yoooc: sorry I'm not following - I was asking about the error message Nov 19 16:16:35 again stupid question from me but couldnt find in google: mainline vs staging version Nov 19 16:27:02 JustasMika: not sure I understand the question, can you explain? Nov 19 16:27:22 I mean whats the difference in kernel between staging and mainline Nov 19 16:27:36 meta-ti provides linux-ti-staging and linux-mainline for example, whats the difference? Nov 19 16:28:57 JustasMika: mainline is plain kernel.org kernel, while staging is with all the work on top of it that is being developed and pending acceptance in upstream/mainline Nov 19 16:29:40 okay. cool, thank you. weird that sometimes i cannot find such simple things using google :) Nov 19 16:30:13 JustasMika: check the recipe - it's rather self explanatory :) Nov 19 17:33:51 Got segfault using prelink-rtld "prelink-rtld[39678]: segfault at 200" is this a know issue ? Nov 19 17:35:24 need more info.. where was it running, what host/target/sdk, etc Nov 19 17:35:54 See https://bugzilla.yoctoproject.org/show_bug.cgi?id=6983 Nov 19 17:35:55 Bug 6983: normal, Undecided, ---, mark.hatle, NEEDINFO , prelink-rtld segfault Nov 19 17:36:40 This is the same issue Nov 19 17:37:17 like I said, I need more info to make any conclusion as to whats going on Nov 19 17:43:41 thanks guy, I'll let my colleagu sgurung the one who file the bug update the bug Nov 19 17:43:46 thanks for your time guys Nov 19 17:46:37 no problem Nov 19 17:46:49 sgurun will come back to you soon for sure Nov 19 18:39:54 with yocto, when I get an error like this is there a log file I can read to understand more ? ERROR: Task 93 (/home/vagrant/ti-glsdk_dra7xx-evm_6_10_00_02/yocto-layers/sources/meta-glsdk/recipes-kernel/linux/linux-ti-glsdk_3.12.bb, do_compile) failed with exit code '1' Nov 19 18:40:23 read the log file it pointed you to Nov 19 18:40:34 that message is useless, it only tells us a task failed, but not why Nov 19 18:40:50 it pointed you to a log.do_compile file either on that line or near it Nov 19 18:41:15 ah yeah, good thx Nov 19 18:41:58 found the culprit : /bin/sh: 1: bc: not found Nov 19 18:41:59 =) Nov 19 18:42:01 thx Nov 19 19:11:08 ah. add bc-native to DEPENDS Nov 19 19:13:41 pyqt has a configure-ng that allegedly supports cross compiling Nov 19 21:41:33 denix: I baked 3.12 kernel and apparently yup, cryptodev works nicely, there is not much speed increase, but mainly there is drastic cpu load drop! :) Nov 19 21:43:46 JustasMika: good to know. but it also means we have a regression in 3.14... :( Nov 19 21:44:32 it does not mean, maybe its me :) im newbie here so cant be sure about it :) Nov 19 21:51:55 JustasMika: either way additional testing on our side would not hurt to either confirm or deny :) Nov 19 22:11:38 JustasMika: in the meantime, you might want to poke at some of our openssl/cryptodev patches here - http://arago-project.org/git/?p=meta-arago.git;a=tree;f=meta-arago-distro/recipes-connectivity/openssl/openssl;hb=HEAD Nov 19 22:12:23 will do Nov 19 22:32:11 paulbarker: is there a branch for your opkg changes Nov 19 22:33:25 ah found it on patchwork, no matter **** ENDING LOGGING AT Thu Nov 20 02:59:58 2014