**** BEGIN LOGGING AT Tue Jun 26 03:00:02 2018 Jun 26 06:07:32 tttims: did you add -dbg package? Jun 26 07:06:54 Hi everyone, I am doing bitbake -c populate_sdk image_name and it throwing some error Jun 26 07:07:01 ERROR: gcc-cross-canadian-arm-7.3.0-r0 do_compile: oe_runmake failed ERROR: gcc-cross-canadian-arm-7.3.0-r0 do_compile: Function failed: do_compile Jun 26 07:07:11 anyone has faced this before ? Jun 26 07:18:49 karthik_: can you post the entire log on pastebin for example? Jun 26 10:36:51 hi, i'm getting an error where one file is installed by two different recipes, not when building a system, but when building the external SDK Jun 26 10:36:55 "ERROR: mg-hd-image-1.0-r0 do_sdk_depends: The file /usr/share/X11/xorg.conf.d/10-evdev.conf is installed by both xserver-xorg and xf86-input-evdev, aborting" Jun 26 10:37:07 is there any way around this? or likely causes? Jun 26 11:07:04 I think I'm about to go insane..! I'm trying to do fitImage and signing/verified boot and all that good stuff, but as soon as I add UBOOT_SIGN_ENABLE = "1" to my local.conf, bitbake fails and starts complaning about circular dependencies.. :/ Jun 26 11:07:45 Anyone know about a good source of information/howto? Jun 26 11:13:36 jof: are you using kernel-fitimage? Jun 26 11:15:51 nayfe: Yes. https://hastebin.com/ucarejibuw Jun 26 11:58:18 Hi, folks! Jun 26 12:00:22 I've built image and started it with runqemu script, but I got a problem - there is no eth0 device. I can see tap0 on host, but only localhost is on qemu. Jun 26 12:02:04 I am using custom kernel with my own defconfig. May be I just skipped some options here. Does somebody know what it can be? Jun 26 12:15:46 astrunin[m]> is your ethernet driver is here? with ifconfig -a do you see eth0 ? maybe you need to tweek qb_network_device from qemuboot.conf file? Jun 26 12:59:25 nayfe: I've made it work! I've started qemu with my own tweaked parameters. Jun 26 13:00:48 ok nice Jun 26 13:01:42 works for me: -net nic -net tap,ifname=tap0,script=no,downscript=no Jun 26 13:05:53 when doing swupdate, there is a single-copy approach and a double-copy approach. if we assume that a) the new image was already tested sufficiently and known to work, and b) any problems (e.g., a power glitch) at update time can still be handled with the single-copy approach by restarting the install, what advantage does double-copy buy you? Jun 26 13:14:02 astrunin[m]> you can persist that conf in qemuboot.conf and with QB_ variables i think Jun 26 13:14:56 yates> downtime is quicker Jun 26 13:15:29 yates> and fallback too :) Jun 26 13:25:33 nayfe: yes, downtime could be an issue in some scenarios. it's not that critical in ours. Jun 26 13:26:12 also, re: fallback, simply have available both the previous update (e.g., on usb thumbdrive) as well as the new. if the new fails, reinstall the old. Jun 26 13:33:23 yates: both approaches are valid with pros and cons, if memory space is more important than downtime then go with single-layout Jun 26 13:34:56 hello Jun 26 13:35:16 nayfe: yes, i'm thinking along those lines. Jun 26 13:35:52 hello Jun 26 13:36:23 the environment setup script from SDK should include some variables for cmake? Jun 26 13:36:52 i have added TOOLCHAIN_HOST_TASK += "nativesdk-cmake" to my image Jun 26 13:38:06 i ahve this file in sysroot OEToolchainConfig.cmake Jun 26 13:38:11 but no alias for cmake Jun 26 13:44:55 nayfe: I know how to cook it ☺! But I prefer to use my own scripts instead of runqemu. Jun 26 13:46:10 right :) Jun 26 13:52:33 ionutp> maybe TOOLCHAIN_HOST_TASK += "nativesdk-cmake-dev" Jun 26 13:53:24 thanks nayfe ... i will try Jun 26 13:56:49 ionutp: see http://lists.openembedded.org/pipermail/openembedded-devel/2018-April/117908.html for example Jun 26 14:11:24 RP, the bind update needs https://patchwork.openembedded.org/patch/150929/. I thought it was in a /mut but is gone now Jun 26 14:35:02 Still working on becoming insane.. :P https://hastebin.com/equnuvefus Trying to do fitImage+u-boot+signing Jun 26 14:35:39 So my current problem is that u-boot.dtb can't be found. But I can't figure out what I need to do in order to create it? Jun 26 14:42:50 armpit: I thought we removed the need for it by default? Jun 26 14:43:08 armpit: test builds were ok without it? Jun 26 14:43:09 correct. Jun 26 14:43:55 RP, ok.. no worries Jun 26 14:46:06 are there any good resources for learning yocto as someone who has done a lot of linux system integration already? Jun 26 14:47:22 does yocto use the host's c compiler to build the kernel? Jun 26 14:49:24 joseppc to -build- the kernel no, to build some tools to build the kernel, maybe Jun 26 14:52:27 for an arm target, it presumably either builds a cross compiler from a host toolchain, or uses a pre-built cross compiler Jun 26 14:52:43 it takes host gcc to boostrap gcc native stuff Jun 26 14:55:40 jof> do you define a custom devicetree ? any dtb in u-boot work dir? Jun 26 14:56:06 fray: thanks, that's what I thought. Jun 26 15:06:18 fray: I asked because I seem to get errors related to gcc 8.x, which is what's installed on my system Jun 26 15:09:19 joseppc> indeed it's not fully supported yet Jun 26 15:16:26 nayfe: but that should only show when building "host" recipes, right? Jun 26 15:18:53 joseppc: apart from some recipes which build native tools they need to build Jun 26 15:19:35 rburton: ah true, but not the kernel I assume Jun 26 15:19:49 big assumption that i suspect is wrong Jun 26 15:19:59 i'd *expect* the kernel to build tooling Jun 26 15:20:23 ah! Jun 26 15:21:31 ok, I get it now, the errors I got are when building user space tools I think, but these should still be built with the gcc provided by yocto, no? Jun 26 15:21:50 no Jun 26 15:21:52 I had to patch the kernel with this: https://github.com/torvalds/linux/commit/854e55ad289ef8888e7991f0ada85d5846f5afb9 (and another similar patch) Jun 26 15:22:07 anything that runs on the host is built using the host compiler Jun 26 15:22:23 obviously the host has a working compiler, as we build a cross compiler with it... Jun 26 15:31:00 is the git cloning of stuff done per build-dir (. oe-init-build-dir build-dir), or "globally"? Jun 26 15:35:48 yates: sources recipes and build folders are linked by parameters in build folders conf/local.conf conf/bblayers.conf etc Jun 26 15:36:55 yates: recipes are fetched into DL_DIR. whether htat's global or not depends on what you set it to in local.conf. generally it's placed outside of TMPDIR, and ideally outside of TOPDIR as well to be shared amongst all your builds, but that's up to you Jun 26 15:40:48 i see. thanks. Jun 26 15:41:10 what is TOPDIR? Jun 26 15:41:41 e.g., sources/poky/ ? Jun 26 15:43:43 TOPDIR == your build directory, where conf/local.conf and TMPDIR live Jun 26 15:44:54 most main variables are defined in bitbake.conf, TOPDIR however is usually defined either by bitbake itself or via bblayers.conf Jun 26 15:45:14 i would recommend perusing bitbake.conf, bblayers.conf, and local.conf, should be informative Jun 26 15:46:06 ok Jun 26 15:49:08 once a repo has been initially cloned into DL_DIR, and a build image generated, is there anything that subsequently triggers an update of the repo? Jun 26 15:49:18 that depends on the recipe Jun 26 15:49:32 recipes can either specify a specific commit, or track the tip of a branch Jun 26 15:49:47 the former is far more common, it helps with build reproducibility Jun 26 15:50:07 of course this only applies to recipes fetching sources, layers are fetched via other means Jun 26 15:50:38 * yates facepalms Jun 26 15:50:48 LAYERS are also fetched? OMG! Jun 26 15:54:33 where would the specific kernel git commit tag be specified? Jun 26 15:55:19 what search string could i use to "find . -name "*" -exec grep -Hn " ? Jun 26 15:55:42 i've looked severl places and can't find it. Jun 26 15:57:33 https://www.yoctoproject.org/docs/2.5/overview-manual/overview-manual.html#sources-dev-environment Jun 26 15:57:41 specifically https://www.yoctoproject.org/docs/2.5/overview-manual/overview-manual.html#scms Jun 26 15:58:11 kergoth: thank you. very helpful. Jun 26 15:58:44 no problem Jun 26 15:59:13 the "h=" option to get gets the specified tag? Jun 26 15:59:20 to git Jun 26 16:00:40 as the latter of the above two links states, SRCREV is how the commit to be checked out is specified. tag= or branch= can be used in concert with SRCREV="${AUTOREV}" to track the tip of a git ref rather than specifying a commit hash Jun 26 16:01:02 tags can change, so specifying a tag rather than the underling tag commit will result in bitbake having to contact the remote server every time it does a recipe parse to map it to a commit Jun 26 16:01:15 so we generally recommend setting a specific commit in the recipe and updating as needed Jun 26 16:01:32 (note: tag= and branch= are url parameters, not variablels) Jun 26 16:31:37 Just curious, where can I find the validation tests used to mark a distrobution as "tested" ? Jun 26 16:31:58 is there anything about an extensible SDK build, that might cause a bunch of errors to crop up in a build that builds a normal system fine? Jun 26 16:37:10 is there a common recipe for putting together the rootfs? I can't see where its coming from Jun 26 16:38:23 yates: its a class Jun 26 16:39:33 yates: Perhaps take a look at how these recipes use core-image.bbclass (and inherently, image.bbclass): https://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/recipes-core/images Jun 26 16:40:12 it's definitely useful to pick a recipe and read it, following all the includes/inherits Jun 26 17:41:12 jynik: the autobuilder is basically the validation tests Jun 26 20:50:26 does the "zImage" that my version of yocto creates do the "create an initial initrd" to load the full kernel? Jun 26 20:50:57 i'm using a vendor-supplied yocto build (variscite) which is in turn a derivative of the meta-fslc for the i.MX6, fyi. Jun 26 20:51:20 my version of the image, rather. Jun 26 20:53:04 i was just looking around and, other than spl and u-boot on an unformatted partition, there is fat32 with the .dtb files and zImage, and there is ext4 with the rootfs. Jun 26 20:54:54 zImage is the full kernel already as far as I know. Jun 26 20:56:20 dennism: yes, but i thought the standard sequence for booting linux is to first create an initrd, then boot the full kernel from there. Jun 26 20:56:35 initrd == init ramdisk Jun 26 20:57:35 I think those are optional when booting linux. You put stuff in them like lots of kernel modules. If your kernel has support for what it needs already built in then no need for an initrd image. Jun 26 20:57:46 is it the case that in the typical OpenEmbedded system, the initrd phase is skpped, and u-boot loads zImage directly? Jun 26 20:58:48 i see. Jun 26 20:59:25 Yeah you can type help in the u-boot menu and find all the commands for loading a kernel. Jun 26 21:09:41 Hello guys , i hope i'm right here , i need your help in yocto :) i'm trying to build an image with squashfs fstype but now i want it to create a sdcard image and it says image_cmd is not defined for fstype sdcard .. i search already about it but didnt find anything to declare image cmd .. btw yocto branch morty Jun 26 21:23:57 anyone can help ? Jun 26 21:25:36 chronos: https://www.yoctoproject.org/docs/2.1/ref-manual/ref-manual.html#var-IMAGE_CMD Jun 26 21:27:39 Is squashfs a custom type? Jun 26 21:31:08 no Jun 26 21:31:31 it will be used for mmc cards as read only fs Jun 26 21:35:48 i trying to built for an sdcard an image which include boot rootfs so i thought image_fstyüe "sdcard" is a way to go is this not right ? Jun 26 21:37:25 try "squashfs tar.bz2 sdcard" Jun 26 21:39:35 dennism same No IMAGE_CMD defined for IMAGE_FSTYPES entry 'sdcard' Jun 26 21:39:55 maybe the morty branch dont support fstype sdcard Jun 26 21:41:18 afraid not something I've tackled yet. someone else will have to chime in that knows more. Jun 26 21:43:44 i'm getting this error when trying to generate the initrd for meta-swupdate: https://paste.fedoraproject.org/paste/W1H3ukqfjThyRy4J9afs9w Jun 26 21:43:56 as described here: http://sbabic.github.io/swupdate/swupdate.html Jun 26 21:44:46 any thoughts? Jun 26 21:59:11 yates or denism you know how to use grub instead of grub efi , cause grub efi give me always compile error Jun 26 22:11:35 Dear all Jun 26 22:12:30 Is there any neat way to remove a feature from for example RDEPENDS_${PN}-utils = "xauth xrandr xmodmap ...... " without the need to copy the long list in bbappend? Jun 26 22:12:39 maybe some python magic? Jun 26 22:12:52 I do need to remove for example xrandr from the above example Jun 26 23:02:02 lukma: Looks like RDEPENDS_thing-stuff_remove = Jun 26 23:03:50 https://www.yoctoproject.org/docs/1.6/bitbake-user-manual/bitbake-user-manual.html ; 3.1.9. Removal (Override Style Syntax) Jun 26 23:41:21 lukma: majuk _remove should be avoided when possible Jun 26 23:42:11 lukma: copying the whole list is the right way Jun 27 01:38:59 I get compile error for qtbase_git.bb. My yocto version is Morty. How do I resolve this error? Jun 27 02:05:53 User__: youll have to be more specific **** ENDING LOGGING AT Wed Jun 27 03:00:01 2018