**** BEGIN LOGGING AT Thu Dec 13 03:00:02 2018 Dec 13 03:36:49 bluelightning: thanks! Dec 13 03:36:59 np Dec 13 05:37:10 New news from stackoverflow: Is there any options in yocto bitbake to make the command autocomplete? Dec 13 08:30:56 I set a path in recipe file, but it seems trying to access another path. why is that? Dec 13 08:31:40 cp -f ${WORKDIR}/file ${D}/${bindir}/file Dec 13 08:32:00 cp: cannot stat '/home/amit/fsl-community-bsp/build/tmp/work/armv7at2hf-neon-fslc-linux-gnueabi/file/5.31-r0/file': No such file or directory Dec 13 08:32:12 that's not the path I tried to copy to Dec 13 08:33:56 hello, I am using poky master and I have a problem when generating an SDK for my image: https://pastebin.com/ELT88XWa Dec 13 08:34:22 I use both util-linux and toybox Dec 13 08:34:41 the images build file Dec 13 08:35:03 on the image losetup, swapoff, umount and dmesg is provided by toybox Dec 13 08:35:22 such things as util-linux-umount are not installed on the image Dec 13 08:35:58 but when generating the SDK util-linux-umount wants to get installed to the target sysroot for some reason Dec 13 08:36:16 and gets in conflict with toybox Dec 13 09:26:52 does anyone know how can I forcefully exclude a recipe from the SDK target sysroot? Dec 13 09:42:26 nacknick: no, that's the value of WORKDIR/file Dec 13 09:46:58 rburton: hello, why does SDK generation try to include system utilities like util-linux, busybox/toybox in target sysroot? Dec 13 09:47:16 presumably because they're needed via dependencies Dec 13 09:47:25 you'll have to look at your dependency tree Dec 13 09:47:45 rburton: but as far as I understand they contribute nothing for the purpose of cross-compilation Dec 13 09:47:59 well util-linux ships a number of libraries Dec 13 09:48:07 possibly, bad dependencies Dec 13 09:48:13 [09:47:25] you'll have to look at your dependency tree Dec 13 09:49:43 rburton: not sure why update-alternatives does not handle this properly for me during SDK generation: https://pastebin.com/ELT88XWa Dec 13 09:50:00 because iirc alternatives don't run during sdk generation Dec 13 09:50:48 rburton: well, the snippet was taken from do_populate_sdk error log, so ... Dec 13 09:51:11 and it even says "update-alternatives" Dec 13 09:51:25 this is poky master, by the way Dec 13 09:54:07 dmesg is implemented by toybox on my target image, for example Dec 13 09:54:28 and I do not ship the util-linux package that has the util-linux version of dmesg Dec 13 09:54:36 so I image builds fine Dec 13 09:54:39 pass. i've got hr breathing down my neck about mandatory training so free tech support will be limited today. Dec 13 09:55:29 understandable Dec 13 09:55:40 thanks anyway for replying Dec 13 09:55:51 rburton: :-) just to cheer you up: yesterday i learned that the new word for HR is "Employee Experience" Dec 13 10:07:56 ∕!\ ΑΤTN: Tһіs channeⅼ һаs moved tഠ irc․freenode.nеt #oѕirіѕⅼаb ∕﹗\ Dec 13 10:18:13 rburton, the file is under /home/, how to set it currectly? Dec 13 10:18:43 nacknick: it should have been copied to workdir by being in SRC_URI Dec 13 10:21:48 rburton, thanks Dec 13 10:22:19 another question please. I want to change the value of CC for all recipes, how can I do it? Dec 13 10:25:57 if I write: "CC = ...." in .bb files; in some files I can see that it compiled differently, and in others - not (I check by the generated executable...) Dec 13 10:26:15 in other words, I want to change the compiler for all recipes Dec 13 10:27:03 nacknick: that only works if the Makefile or whatever equivalent inside the recipe actually cares for the value of CC Dec 13 10:27:11 if it doesn't, its a bug anyways, but hey.... Dec 13 10:58:55 nacknick: local.conf or distro conf Dec 13 10:59:05 see bitbake.conf for the default Dec 13 10:59:12 better question: why are you changing CC? Dec 13 10:59:25 out of the box it points to the compiler that we just built Dec 13 11:07:40 nacknick: so continuing the fun with your file drop-in binary, did you remember to make the replacement binary target only? or are you ending up with a target binary replacing the native build of file? Dec 13 11:18:56 rburton, I want to change CC because it consists a flags that disrupt my script that modify the binary. the binary changing was just a test. actually I take the generated binary, modify it and then replace it Dec 13 11:19:10 CC doesn't contain flags Dec 13 11:19:33 well, it has a sysroot flag, but good luck building anything without that Dec 13 11:21:38 in case its relevant, remember that packaging will fiddle the binaries Dec 13 11:22:10 so if you're doing anything like adding signatures, you'll need to do it later than install Dec 13 11:23:38 I'm not. I'm changing a compiling flag Dec 13 11:24:37 i'm sure there's a good reason why you don't just change the compiler flags Dec 13 11:25:19 so what flag is breaking your script? Dec 13 11:27:44 -mthumb Dec 13 11:28:08 so yeah that's not in CC Dec 13 11:28:14 so where? Dec 13 11:28:18 just turn off thumb globally instead? Dec 13 11:28:23 ok Dec 13 11:28:24 how? Dec 13 11:29:40 ARM_THUMB_OPT=arm says the BSP comments Dec 13 11:31:15 rburton: I ran into a wall with virglrenderer sadly :( Dec 13 11:31:28 just asked for help on #virgil3d Dec 13 11:31:47 kanavin_: what's up? i can't help but i'm curious Dec 13 11:32:06 runqemu - ERROR - Failed to run qemu: gl_version 0 - compat profile Dec 13 11:32:07 WARNING: running without ARB/KHR robustness in place may crash Dec 13 11:32:07 qemu-system-i386: ../../../../../../worksp Dec 13 11:32:15 sorry Dec 13 11:32:18 like this: Dec 13 11:32:46 runqemu - INFO - Running /home/alexander/development/poky/build/tmp/work/x86_64-linux/qemu-helper-native/1.0-r1/recipe-sysroot-native/usr/bin/qemu-system-i386 -device virtio-net-pci,netdev=net0,mac=52:54:00:12:34:02 -netdev tap,id=net0,ifname=tap0,script=no,downscript=no -drive file=/home/alexander/development/poky/build/tmp/deploy/images/qemux86/core-image-minimal-qemux86-20181212165244.rootfs.ext4,if=virtio,format=raw -display sdl,gl=on -vga virtio Dec 13 11:32:46 -show-cursor -usb -device usb-tablet -object rng-random,filename=/dev/urandom,id=rng0 -device virtio-rng-pci,rng=rng0 -cpu pentium2 -enable-kvm -m 256 -serial mon:vc -serial null -kernel /home/alexander/development/poky/build/tmp/deploy/images/qemux86/bzImage--4.19-rc++git0+3b9c29018a_122d468967-r0-qemux86-20181212124959.bin -append 'root=/dev/vda rw highres=off clocksource=kvm-clock hpet=disable noapic nolapic mem=256M ip=192.168.7.2::192.168.7.1:255. Dec 13 11:32:46 255.255.0 oprofile.timer=1 uvesafb.task_timeout=-1 ' Dec 13 11:32:50 runqemu - ERROR - Failed to run qemu: gl_version 0 - compat profile Dec 13 11:32:52 WARNING: running without ARB/KHR robustness in place may crash Dec 13 11:32:54 qemu-system-i386: ../../../../../../workspace/sources/libepoxy-native/src/dispatch_common.c:861: epoxy_get_proc_address: Assertion `0 && "Couldn't find current GLX or EGL context.\n"' failed. Dec 13 11:33:16 this happens when the kernel is already booting, when it gets to initializing the driver for the virtual graphics card Dec 13 11:33:34 both epoxy-native and virglrenderer-native unit tests pass Dec 13 11:33:36 epoxy dlopen's the libgl so library depends won't find it, sure its in the native sysroot? Dec 13 11:34:11 slap a strace on that to check it opens the library at least? Dec 13 11:36:41 just looking at it now Dec 13 11:40:50 rburton: don't see anything suspicious there :-/ Dec 13 11:50:54 rburton: I placed extra logging into epoxy around that code Dec 13 11:52:53 rburton, I have two files that contain ARM_THUMB_OPT: sources/poky/meta/conf/machine/include/arm/feature-arm-thumb.inc and ./build/tmp/deploy/images/imx6ullevk/core-image-base-imx6ullevk-20181212073329.testdata.json Dec 13 11:53:10 which one should I change? Dec 13 11:54:44 inside 'feature-arm-thumb.inc': ARM_THUMB_OPT = "${@['arm', 'thumb'][d.getVar('ARM_INSTRUCTION_SET') == 'thumb']}" Dec 13 11:56:12 nacknick: neither, change your local.conf or distro config Dec 13 11:56:48 oh set ARM_INSTRUCTION_SET=arm instead Dec 13 11:57:06 thats the usual one recipes touch if they need to force not-thumb Dec 13 11:57:50 in local.conf 'set ARM_INSTRUCTION_SET=arm' or 'ARM_INSTRUCTION_SET=arm'? Dec 13 11:58:16 with or without 'set'? Dec 13 12:01:15 without Dec 13 12:01:16 ok. without Dec 13 12:01:18 hh Dec 13 12:03:41 if I want to install 'apt'? how should I do it? because adding 'apt' to 'CORE_IMAGE_EXTRA_INSTALL' did not allowed to me to use 'apt install' on the borad. do I miss something? Dec 13 12:09:52 in local.conf or distro config set PACKAGE_CLASSES to package_deb, in the image recipe add package-management to IMAGE_FEATURES Dec 13 12:11:05 then run a httpd on your build machine serving the deploy/deb directory, and set PACKAGE_FEED_URIS to the base URL Dec 13 12:11:12 voila, working feeds Dec 13 12:11:30 note that deb is the least-supported, rpm or ipkg is tested more Dec 13 12:13:52 rburton sorry, but I did not understand , "in the image recipe add package-management" - which 'image recipe'? Dec 13 12:14:21 rburton your second sentence "then run..." I did not understand at all :S Dec 13 12:14:35 the recipe for the image you're building. or just set EXTRA_IMAGE_FEATURES in your local.conf to package-management. Dec 13 12:14:44 ok Dec 13 12:14:47 that's clear Dec 13 12:14:47 apt-get install needs a feed and a server, so you need to run the server Dec 13 12:15:11 I need to run it? there is no global repository? Dec 13 12:15:15 global? Dec 13 12:15:16 no Dec 13 12:15:24 you just built the packages Dec 13 12:15:28 nobody else has them Dec 13 12:15:36 especially as you're swapping binaries in during do_install Dec 13 12:15:59 why? all the packages sources are on the internet Dec 13 12:16:03 *sources* Dec 13 12:16:09 yes Dec 13 12:16:19 where else on the internet is the binary that you just altered using your script Dec 13 12:16:39 put my script aside... I'm not talking about it now Dec 13 12:16:44 but that's key Dec 13 12:17:01 you boot your image and do apt-get install file, it should install the package that you just built Dec 13 12:17:16 if that doesn't work, then what's the point? Dec 13 12:17:57 ok. so I need to build the packages and use as server for the embbeded machine? Dec 13 12:27:19 if you want a feed to work, yes Dec 13 12:27:46 you need to build the packages (as otherwise they don't exist) and run a http daemon (so the board can download the packages from it) Dec 13 12:38:32 New news from stackoverflow: vlc add in yocto build Dec 13 13:21:16 hello! Dec 13 13:22:24 I accidentally deleted download directory. now the tasks are failing on do_unpack,... I have to do bitbake -c cleanall for offending package... how to tell bitbake to redownload missing files Dec 13 13:27:17 (I changed DL_DIR to another preexsting - semi filled directory) Dec 13 13:31:35 ak77: remove the do_fetch stamps from tmp/stamps/ Dec 13 13:32:07 RP: will this keep predownloaded archives? Dec 13 13:32:18 RP: i mean, already downloaded archives Dec 13 13:35:49 ak77: yes Dec 13 13:36:00 and by, keep, i mean not re-download Dec 13 13:40:47 222 Dec 13 13:43:02 RP: thank you Dec 13 14:08:50 New news from stackoverflow: hostapd on yocto krogoth Dec 13 14:38:58 New news from stackoverflow: Read a specific memory address via /dev/mem from the command line Dec 13 15:34:28 anyone actively using vulkan? Dec 13 15:39:10 New news from stackoverflow: "File format not recognized" when building Petalinux app Dec 13 15:43:21 aehs29, ^^^^ Dec 13 15:47:44 is defining SRC_URI supported in a .bbclass? Dec 13 15:53:11 any variable can be manipulated in a class.. Dec 13 15:53:15 but it's not always a good idea. Dec 13 15:53:34 for something like SRC_URI, you'll have to manipulate it in an anonymous python section most likely for it to take effect Dec 13 15:53:40 OpenEmbedded lets yo do many thigns that are bad ideas :) Dec 13 15:54:15 I've seen two cases of it.. one is an inherited class for recipes where a standard SRC_URI can be defined for all recipes of a certaint ype... Dec 13 15:54:28 the other is where a complex SRC_URI replacement was needed where using hte existing mirror system was not possible Dec 13 15:59:26 fray: thanks Dec 13 16:05:07 apteryx: see eg gnomebase or pypi classes Dec 13 16:51:38 rburton: OK :-) Dec 13 17:06:34 * RP wonders if he can get counselling after reading distrodata.bbclass Dec 13 17:12:52 hello! Dec 13 17:12:53 I am still struggling with distro, image, and their features... does this logic holds: I defien several machines, and several images (one of them being upgrade-ready-image)... to support two project, should I define two distros? Dec 13 17:22:28 what is the difference between the distros? Dec 13 17:45:26 Crofton: bunch of packages Dec 13 17:46:43 a distro is basically a distro.conf file Dec 13 17:46:55 Crofton: I am aware of that Dec 13 17:47:01 bunch of packages is usually hanled at image lvel Dec 13 17:47:08 level :) Dec 13 17:48:40 ok, so..... an upgrade-artefact-image-type is an image class. and I should define images per project? Dec 13 17:49:34 it is really hard to know without knowing details Dec 13 18:04:06 maybe I am missing something but there should be few use-cases written in manual, when to do distro/image Dec 13 18:40:09 Hi. Can I use "runqemu" to test a kernel with a bundled initramfs ? I can't really figure out the syntax on how to start it .. Dec 13 19:06:50 qemu-system-i386 -kernel is probably easier.. Dec 13 19:17:37 Crofton: haha thanks Dec 13 20:54:52 has anyone seen a recipe for bedrock? https://bedrockdb.com/ Dec 13 23:02:38 yates: http://layers.openembedded.org/layerindex/branch/master/recipes/?q=bedrock says no Dec 13 23:03:33 yates: eeek blockchain based run away Dec 13 23:27:01 hi all, is it possible to use a for loop in a BitBake recipe? Kind of like a shell for loop? as in: for file in file1 file2 file3; do .... ; done Dec 13 23:29:53 robbawebba: the functions are written in python or shell, so yes Dec 13 23:30:12 the shell functions are literally ran by bash Dec 13 23:38:42 khem: could you glance at the python3 musl powerpc patch on the list please? I suspect it could be improved with your domain knowledge :) Dec 13 23:57:30 rburton: ahhh okay. Thank you for the comfirmation, I should've known it was a shell function! I appreciate the info Dec 14 00:12:53 inet_pton(AF_INET, "99||SDKTD[0", &(addr.sin_addr)) Dec 14 00:13:06 This API is returning 1 (that means succesfull) Dec 14 00:13:56 Linux cb4f0b42-30a5-e162-dbb0-cbe8577b0c28 4.10.17-yocto-standard #14 SMP PREEMPT Thu Nov 8 19:44:47 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux Dec 14 00:14:06 Where should I file bug for this Dec 14 00:14:49 inet_aton() works fine **** ENDING LOGGING AT Fri Dec 14 03:00:01 2018