**** BEGIN LOGGING AT Fri Oct 11 03:00:00 2019 Oct 11 07:33:19 New news from stackoverflow: How to install files in Yocto Oct 11 07:58:41 Good morning again everybody Oct 11 08:02:41 I've a noob question today, but it's making me loosing a lot of time and I'm not able to formulate the right question to Google. I'm customizing boot2qt image for my machine, but I don't get how to reassemple image after i change some included recipe; say I just remove a dependency in a included recipe: how can I make Yocto regenerare image, without recompiling nothing, but exluding the no-more-needed software? Oct 11 08:03:25 New news from stackoverflow: All QT5 recipes are failing on compilation Oct 11 08:03:56 mous16: bitbake my-funny-image Oct 11 08:04:10 mous16: just rerun the image building. Oct 11 08:18:51 I'm trying to create a toolchain for windows for a multilib image (ARM 32-bit and aarch64). The toolchain for Linux works fine, but when building one for Windows (with the meta-mingw layer), the 32-bit version fails to compile binaries. The 64-bit version works fine, but when I try to compile a simple hello world program with the 32-bit compiler I get: Oct 11 08:18:56 c:/tagmas~1/vm_share/kodkod~1.0/sysroots/x86_64-kkdistsdk-mingw32/usr/bin/arm-pokymllib32-linux-gnueabi/../../libexec/arm-poky-linux-gnueabi/gcc/arm-poky-linux-gnueabi/7.3.0/real-ld.exe: cannot find -lgcc Oct 11 08:20:26 any ideas? Oct 11 08:25:31 none that you would accept :) Oct 11 08:33:15 LetoThe2nd: I hoped that 'bitbake my-funny-recipe' do the work, but it actually don't. bitbake starts, thinks for a minute, and successfully terminate without reassembling the image, because "there are no tasks to do". Using 'bitbake -c clean my-funny-image' do the work, and with sstate-cache it's pretty fast to reassemple everithing, but it doesn't feel right Oct 11 08:33:29 New news from stackoverflow: How to add support for screen timeout while running a qt5.9 app on linuxfb platform? Oct 11 08:34:32 mous16: then you did change something in some way that bitbake cannot detect, which is a strong indication that its wrong. Oct 11 08:57:49 Good morning everyone! I have a quick question. I have a recipe where the packageconfig is set like this "PACKAGECONFIG ??= 'package1 package2 package3". Now I want to add an additional package in the bbappend file. I first tried "PACKAGECONFIG += package4", but that did not work. Now I just did overwrite the whole packageconfig in the bbappend and Oct 11 08:57:49 it works fine, but I wonder what the correct workflow would be Oct 11 09:00:19 Hi everyone,I tried to create an image with WIC.I have a layer with my main recipe and an update layer for the update.When I call wic with my update recipe the rootfs is missing.How can I tell WIC where is the rootfs? Oct 11 09:03:36 New news from stackoverflow: No internet access without running 'ifup eth0' or restarting /etc/init.d/networking on jetson nano with yocto/poky Oct 11 09:12:37 hoxy: https://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-PACKAGECONFIG. tldr, use _append Oct 11 09:12:49 (if you just want to say 'whatever the default was, and this') Oct 11 09:13:20 Thank you. I will try that Oct 11 09:13:33 when in doubt the docs are actually useful Oct 11 09:53:30 linux-firmware includes a lot of firmware files for different hardware - what is the "right" way to remove unneeded firmware files? Oct 11 09:59:01 Ohh, I just noticed that linux-firmware-* is a thing. I'll have a look at that Oct 11 10:02:39 bojones, yes, most the firmware files are split between packages, you can install just the ones you need Oct 11 10:06:18 mihai, yes thats great. Thanks Oct 11 10:21:22 hello folks. when defining the target-os in one's machine definition file, what should one assign to? TARGET_OS or HOST_OS ? Oct 11 10:21:38 for normal software the distinction is irrelevant, and is only relevant for compilers Oct 11 10:22:01 so, for gcc-cross, the right value to set is TARGET_OS, so I would suspect one needs to set TARGET_OS. but I'm not completely sure. Oct 11 10:23:02 my machine config would then override bitbake.conf's TARGET_OS = "linux${LIBCEXTENSION}${ABIEXTENSION}" Oct 11 10:23:12 well HOST_OS is defined as ${TARGET_OS} in bitbake.conf so just set target and it sets both Oct 11 10:23:45 hi Oct 11 10:34:15 rburton, glib, qemu and gcc in the core layer reference the mingw32 OS override Oct 11 10:34:34 Morning. Once a class has redefined a method/task, is it possible to reference the old version? Oct 11 10:34:39 Or is there a cleaner way to reach my aim, which is performing one command just before do_compile which adjusts the env that do_compile is about to use? Oct 11 10:35:02 Is environment shared between tasks? Oct 11 10:35:03 interesing. I think that means the mingw support is "officially" in part in the OE core layers. Oct 11 10:36:09 ChrisStuart: do_compile_prepend? or [before]? don't remember for sure, but theres a way to inject functions before and after calls. Oct 11 10:37:33 ChrisStuart: do_compile_prepend() to inject just a single command is the easiest way Oct 11 10:38:39 OK I'll look at that and do some testing about the environment sharing. That type of injection can be done multiple times right, it doesn't literally create a task called 'do_compile_prepend'? Oct 11 11:02:22 it literally glues your bit of shell to the beginning of do_compile Oct 11 11:02:44 oh but if you just want to set the environment, just use export FOO="bar" in the recipe outside of any tasks Oct 11 11:03:58 New news from stackoverflow: changing kernel configuration for different recipes based on a external variable in yocto Oct 11 11:21:33 if I have a multiconfig build, and then execute a target, say an image, without specifying a config using the "multiconfig:name:image", but If I say just "image" Oct 11 11:21:46 will it build the image for all configs, or only the first? Oct 11 11:29:03 the 'master' config Oct 11 11:29:13 whatever is in local.conf is the unscoped config Oct 11 11:29:46 multiconfig just *adds more* Oct 11 11:32:24 ah, so I really have one master config, and zero or more multiconfigs. one master config where I should/can set a MACHINE. and for my mingw32 target, I would add a single multiconfig file that sets MACHINE to device-mingw32 Oct 11 11:32:49 if i were you i'd focus on the windows machine, not multiconfig Oct 11 11:33:00 when the machine works, then figure out how multiconfig works Oct 11 11:33:04 and when I say "bitbake ourimage" it would build the bootable image for our firmware. and "bitbake multiconfig:master:customerapp", it would build the customer app Oct 11 11:33:16 yes Oct 11 11:34:03 rburton, in our case, the windows App will need to be bundled into the root file system of the device (so customers can download the .exe from the device). so I would have inter-config-dependencies Oct 11 11:34:12 but those seem to be possible, according to the config. Oct 11 11:34:22 thanks, I will try Oct 11 11:34:28 yeah mc can do that Oct 11 11:34:35 i've got images which pull files from other configs Oct 11 11:34:58 ah I see. you also use this multiconfig Oct 11 11:35:30 https://github.com/intel/meta-acrn/blob/master/docs/getting-started.md#setup Oct 11 11:36:16 hello Oct 11 11:36:23 the uos config builds images and the main config makes packages from those images to put them in the image Oct 11 11:37:47 I'm trying to downgrade the GCC version to 9.4 in the rocko branch, but i'm getting one error. Can anybody help me? Oct 11 11:39:01 honestly the first time someone has said they're downgrading gcc. normally its 'i'm upgrading gcc and the world is breaking' Oct 11 11:39:09 i won't ask why but just say what the error is Oct 11 11:39:13 "ask, don't ask to ask" Oct 11 11:42:54 what? what I need to do is "downgrade" the gcc version. Oct 11 11:43:01 LordXor: downgrade to 9.4 in rocko sounds pretty wrong in itself. Oct 11 11:43:28 LordXor, the point is to figure out whether you really need to downgrade Oct 11 11:43:38 perhaps you just need to pass other options or smth Oct 11 11:44:40 no, thats not what i meant. rocko is not exactly up to date, and comes with either gcc 6.4 or 7.3 Oct 11 11:44:58 Because the develop team needs to compile a software that is not compatible with the version 6.4, only with 9.4 Oct 11 11:44:59 so you either got the version, the release or the grading-direction wrong :) Oct 11 11:45:05 lol Oct 11 11:46:08 LordXor: you can always try to rip the gcc parts out of a recent release and inject them into rocko, but that certainly cause *MUCH* pain Oct 11 11:46:59 LordXor: so the two more viable options are, adding an additional gcc version just for your problematic package, or updating to warrior/zeus, what ever brings the desired gcc Oct 11 11:47:17 yet both still mean quite a bit of work, its usually not trivial. Oct 11 11:47:22 LetoThe2nd: 9.2 is the current versionin master so it's not even supproted (tested?) yet Oct 11 11:47:23 rburton, "Note that thanks to a bug in bitbake if you go straight to acrn-image-base from an empty sstate then it will build a lot of recipes twice." Oct 11 11:47:29 How can I add an additional gcc version?? Oct 11 11:47:47 rburton, so this is considered a bug now. In the manual, it is described merely as a missing feature :) Oct 11 11:48:29 yeah bug feature, details Oct 11 11:48:37 LordXor: 9.4 does not even exist: https://gcc.gnu.org/ Oct 11 11:48:44 LordXor: same as before: rip out the parts from a newer release, inject them, but nail down the common compiler, then make it available under a presumable different name, depend on its -native version. roughly. Oct 11 11:48:52 not trivial :P Oct 11 11:49:40 why is this a downgrade rather than an upgrade? Oct 11 11:50:06 litb: there's so much wrong in the approach that up vs. down is pretty much "whatever" :) Oct 11 11:52:48 Thank you all Oct 11 11:59:50 bye Oct 11 12:00:04 hrhrhr Oct 11 12:03:26 that was strange Oct 11 12:03:47 Sounds like misinformation Oct 11 12:03:57 i read it as "you didn't tell me the answers i wanted to hear." Oct 11 12:04:02 possibly both Oct 11 12:04:23 was this the tech equivilant to being sent to get some dehydrated water? Oct 11 12:05:24 maybe. Oct 11 12:06:08 Possibly. :p But then again, he did "downgrade" to *something* .. but he just had an error with it... Oct 11 12:28:22 Hi, I'm new to Yocto - I am struggling to point devtool to a subdirectory in a git repo (elastic beats repo, I want only metricbeat) - is there any docs pointing me to the relevant steps to take? (didn't find any preexisting oe-recipes for that purpose unfortunately) Oct 11 12:43:03 small_son_oliver: did you try --src-subdir ? Oct 11 12:43:40 (with devtool add) Oct 11 12:44:51 Not yet, but thanks - I'll try that. (I tried appending subpath in the recipe as described in bitbake user manual under 4.3.5 git fetcher ( https://www.yoctoproject.org/docs/1.6/bitbake-user-manual/bitbake-user-manual.html#git-fetcher )) Oct 11 12:45:18 hm, i'm going to copy the things out of meta-mingw into my own layer and remove all the "nativesdk-" prefixes. a better approach would be to factor out all the things in meta-mingw, so it's usable for both cross and nativesdk builds Oct 11 12:48:24 rburton : I'm about to work on opkg-utils SPDX identifier this week end, I see that for instance : http://git.yoctoproject.org/cgit/cgit.cgi/opkg-utils/tree/opkg.py has a text explaining the license type (GPLV2 or later) are the SPDX supposed to replace those text ? Or just add a quick way to know the license ? Oct 11 13:12:01 hey, anyone know how to register to yocto project summit in Lyon now that only late registration is possible? https://www.yoctoproject.org/yocto-project-summit-2019-2/ has link to registration page but it shows the event as closed. Oct 11 13:19:29 mcfrisk: I got this link in my mailbox: https://events19.linuxfoundation.org/events/open-source-summit-europe-2019/program/co-located-events/#yocto-project-summit Oct 11 13:19:46 I'm already registered so I can't really help more thant that I think Oct 11 13:20:32 mcfrisk: ah i see, you can ad yocto project to an OSS registration but not register only for it? Oct 11 13:25:59 qschulz: yes, exactly Oct 11 13:29:42 LetoThe2nd: RP: ^ ? I know you're going so you most likely know who to ask for this Oct 11 14:06:20 ndec: ^^^ Oct 11 14:06:35 mcfrisk: you should talk to dnec Oct 11 14:06:38 ndec Oct 11 14:07:24 ;-) Oct 11 14:09:47 mcfrisk: i will check.. i think this is not expected.. Oct 11 14:10:59 mcfrisk: can you share an email address (in PM), i will send an email and cc you. Oct 11 14:23:03 * zeddii hasn’t registered either Oct 11 14:30:24 well... i haven't yet either.. so let me do that now! Oct 11 14:56:26 ndec, you got the request for late registrion? Oct 11 14:56:37 Wonder why it clsoed so early? Oct 11 14:56:43 we are looking into that. Oct 11 14:56:52 it's not expected. Oct 11 14:59:08 thanks Oct 11 15:34:47 New news from stackoverflow: Is it possible to pass in command line variables to a bitbake build? Oct 11 15:47:45 rburton, it feels not correct to set TCLIB and DISTRO_FEATURES in my machine config. so I should make a separate distribution for the mingw config Oct 11 15:48:37 and set TCLIB+friends in the distribution, + provide an distro/include/tclib-mingw.conf that defines the preferred providers for libc, libintl etc Oct 11 15:49:17 and only set TARGET_OS and the tune settings in my machine config Oct 11 17:05:04 New news from stackoverflow: Yocto add an overlay directory tree Oct 11 18:25:53 all that mess with STAGING_DIR_* is so confusing :( Oct 11 18:27:16 I expected that for cross-canadian.bbclass, it sets STAGING_DIR_HOST to something different than STAGING_DIR_TARGET . after all, for a mingw SDK, the host is windows, but the target most probably will be linux! Oct 11 18:28:29 but, as it seems, STAGING_DIR_HOST is RECIPE_SYSROOT, and STAGING_DIR_TARGET is RECIPE_SYSROOT aswell! Oct 11 19:05:18 New news from stackoverflow: populate_sdk fails on 'binutils-crosssdk_2.32.bb:do_configure' Oct 11 20:17:45 is there a good guide on sstate mirror setup? Oct 11 20:18:30 i think i'm confused about it.. i did a build, rsync sstate to a http server and then rm -rf tmp sstate-cache/ cache/ and try to build again Oct 11 20:18:48 but only some small % was matched by bitbake when i build again Oct 11 20:55:30 zeddii: is 5.5.20 gonna be the last update for zeus? Oct 11 21:07:48 Anybody have any experience building an image for the raspberrypi that includes the bcm2835 library? I'm trying to build core-image-minimal with the library added but it is throwing errors and I can't figure out what is going wrong. Building core-image-minimal works when not including the bcm2835 library. Oct 11 21:48:05 ragnoraq, what are the errors? Oct 11 21:50:52 aehs29: I’ll do .20 after 3.0 releases, but Wind River will pickup the maintenance after that. So there will be more -stable updates to it. Oct 11 21:52:48 Crofton|work, ERROR: core-image-minimal-1.0-r0 do_rootfs: Could not invoke dnf. Command '/home/tom/Projects/raspberrypi/Brewer/build/tmp/work/raspberrypi2-poky-linux-gnueabi/core-image-minimal/1.0-r0/recipe-sysroot-native/usr/bin/dnf -v --rpmverbosity=info -y -c Oct 11 21:52:48 /home/tom/Projects/raspberrypi/Brewer/build/tmp/work/raspberrypi2-poky-linux-gnueabi/core-image-minimal/1.0-r0/rootfs/etc/dnf/dnf.conf --setopt=reposdir=/home/tom/Projects/raspberrypi/Brewer/build/tmp/work/raspberrypi2-poky-linux-gnueabi/core-image-minimal/1.0-r0/rootfs/etc/yum.repos.d Oct 11 21:52:49 --installroot=/home/tom/Projects/raspberrypi/Brewer/build/tmp/work/raspberrypi2-poky-linux-gnueabi/core-image-minimal/1.0-r0/rootfs --setopt=logdir=/home/tom/Projects/raspberrypi/Brewer/build/tmp/work/raspberrypi2-poky-linux-gnueabi/core-image-minimal/1.0-r0/temp Oct 11 21:52:49 --repofrompath=oe-repo,/home/tom/Projects/raspberrypi/Brewer/build/tmp/work/raspberrypi2-poky-linux-gnueabi/core-image-minimal/1.0-r0/oe-rootfs-repo --nogpgcheck install bcm2835 packagegroup-core-boot run-postinsts' returned 1: Oct 11 21:52:50 /home/tom/Projects/raspberrypi/Brewer/build/tmp/work/raspberrypi2-poky-linux-gnueabi/core-image-minimal/1.0-r0/rootfs/var/cache/dnf Oct 11 21:52:51 modules for: Oct 11 21:52:56 failure stored in: /home/tom/Projects/raspberrypi/Brewer/build/tmp/work/raspberrypi2-poky-linux-gnueabi/core-image-minimal/1.0-r0/temp/log.do_rootfs.13624 Oct 11 21:52:58 didn't need to be rerun and 1 failed. Oct 11 22:01:15 zeddii: alright cool **** ENDING LOGGING AT Sat Oct 12 02:59:57 2019