**** BEGIN LOGGING AT Thu Feb 27 02:59:58 2020 Feb 27 03:50:16 pole Feb 27 03:55:42 north Feb 27 05:51:28 hi all. i was hitting the below error while running "bitbake core-image-base". Feb 27 05:51:30 /nobackup/sikumar3/mynode_sumo/build-ul/tmp/work-shared/imx6uloib/kernel-source/scripts/gcc-version.sh: line 26: arm-poky-linux-gnueabi-gcc: command not found Feb 27 05:52:09 this is getting resolved when i comment out "DEPENDS += "bridge-utils"" in my linux-imx_%.bbappend Feb 27 05:52:25 what could be the reason? Feb 27 05:53:17 *sorry i was using "DEPENDS = "bridge-utils" Feb 27 07:17:26 hello everybody Feb 27 07:17:45 I am running qml app on yocto Feb 27 07:17:49 with visibility: "FullScreen" Feb 27 07:18:16 but it didn't full screen when I run it on target machine and screen Feb 27 07:18:20 any help ? Feb 27 07:51:52 New news from stackoverflow: Gsteamer conflicting declaration in opencv in Yocto build Feb 27 09:00:26 hi, gm Feb 27 09:01:14 in a "ro" rootfs looks like ssh is not working, inspecting it, looks like ssh port is owned by pid 1 (init) Feb 27 09:01:18 tcp6 0 0 [::]:ssh [::]:* LISTEN 1/init Feb 27 09:01:39 if any hint, welcome Feb 27 09:02:48 angelo__: on which release? Feb 27 09:03:52 is a zeus Feb 27 09:04:23 a couple of observations: you are looking at a tcp6 port. and by default, ssh is socket activated by systemd Feb 27 09:06:18 LetoThe2nd, oh ok ... Feb 27 09:08:14 LetoThe2nd, issue seems connected with RO fs Feb 27 09:08:27 if [ ! -f /etc/ssh/ssh_host_key ] Feb 27 09:08:27 then Feb 27 09:08:27 echo "Generating keys for the ssh server: " Feb 27 09:08:27 ssh-keygen -q -t rsa1 -f /etc/ssh/ssh_host_key -C '' -N '' Feb 27 09:08:43 so, no ssh folder is there and not created too Feb 27 09:09:16 uhm, i can create it manually btw .. Feb 27 09:09:30 angelo__: unless you did mess with it, the read-only-rootfs class should take measures so the key generation works. i've personally used it. Feb 27 09:10:09 LetoThe2nd, ok Feb 27 09:12:09 ok, there seems to be a good reason for sshd not working Feb 27 09:12:31 "/usr/sbin/sshd" is not there, sorry for chaos Feb 27 09:13:15 hehehe Feb 27 09:37:37 rburton, hey, do you know if Juro Bystricky still works at intel ? Feb 27 09:39:14 Hi all, could you please point me how to avoid Yocto to parse specific bb file ? because off openjdk-7 looks boggus with arm64 ARCH :S Feb 27 09:45:02 PinkSnake: https://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-BBMASK Feb 27 09:46:37 Hi guys, does anyone here use Gitlab CD/CI for their Yocto project? Feb 27 09:47:13 Alternatively, what CD/CI solution do you use/have found effective for your workflows? Feb 27 09:48:24 kroon: he does Feb 27 09:48:49 kriive: any ci works you just need to persist sstate/downloads between runs Feb 27 09:49:05 rburton, thanks Feb 27 09:50:14 paulbarker: did you ever write something about that CI setup you had? Feb 27 09:50:31 rburton: oh ok, thanks Feb 27 09:51:37 erbo thx for the link ;) Feb 27 09:52:17 New news from stackoverflow: Jenkins setup to work with different projects managed with the repo tool Feb 27 09:52:42 erbo: I got dragged off onto other things :( Should have some more time in March Feb 27 09:53:17 paulbarker: I know exactly how it is, just wanted to check if I had missed it :) Feb 27 09:54:24 erbo: It's working very well, the CI side is all public on GitLab I just need to write up how to configure Cloudflare & Backblaze B2 Feb 27 10:27:19 how are structs mapped to /sys/kernel/debug in Linux? Feb 27 10:28:21 perdmann: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/Documentation/filesystems/debugfs.txt Feb 27 11:12:59 exit Feb 27 11:52:40 New news from stackoverflow: Yocto Package have unmet dependencies while installing deb file Feb 27 13:18:00 When running "bitbake PACKAGE package-index" - is it guaranteed that package-index will be build after PACKAGE got built? Feb 27 13:18:08 no Feb 27 13:18:18 wow - that was quick :) Feb 27 13:18:20 both will be built in parallel Feb 27 13:19:00 i actually wrote a class to regenerate the package index after every bitbake if you want to test it :) Feb 27 13:19:08 calling bitbake twice then - but how would I (easily) prevent "bitbake package-index" from creating an (empty) commit to buildhistory? Feb 27 13:19:51 is this class already in zeus? Feb 27 13:20:12 no its not merged, thus the 'test it' comment Feb 27 13:20:57 top 3 commits on http://git.yoctoproject.org/cgit.cgi/poky-contrib/log/?h=ross/index if you're interested Feb 27 13:26:40 mmmh, it seems like in 99% of the cases I do not wish the index to be updated Feb 27 13:26:55 actually, it needs to be updated only during CI runs Feb 27 13:28:34 you only need the index updated if you want to use the deploy directory as a package feed outside of bitbake Feb 27 13:28:44 eg upgrading on the target using opkg/dnf/ap Feb 27 13:28:46 t Feb 27 13:40:19 I know - my CI pushes them right after building to a server. Feb 27 13:41:14 I resolved the problem now by fetching the buildhistory repo, building the images, then pushing the updated build history to the Git repo and just after this building the index Feb 27 14:25:31 hello using yocto zeus im trying to build a (meta-freescale) layer depending on git://git.openembedded.org/openembedded-core. I cannot figure out how to add it to my bblayer.conf. there is no conf/local.conf at the root. Feb 27 14:27:47 mauz555: in your build directory, there is a conf directory. this holds local.conf and bblayers.conf Feb 27 14:28:01 mauz555: you do not have to modify anything inside one of the layers. Feb 27 14:29:11 mauz555: plus, core is included in poky Feb 27 14:29:21 LetoThe2nd: I have my poky folder with my build folder inside I have a conf/local.conf (where I set the MACHINE) and a bblayer.conf (where I set the absolute path of my meta-freescale layer) Feb 27 14:31:22 LetoThe2nd: when I run `bitbake core-image-minimal` it says Nothing PROVIDES 'python-pycrypto-native' so I was thinking that meta-freescale is missing a layer, checking meta-freescale README it says it depends solely on git://git.openembedded.org/openembedded-core Feb 27 14:32:24 so is poky/meta, equivalent to openembedded-core/meta ? Feb 27 14:32:44 it is. Feb 27 14:32:44 mauz555: poky/meta is oe-core, yes. Feb 27 14:33:04 mauz555: i sugegst two things 1) start over with the build outside Feb 27 14:33:35 mauz555: 2) try building without meta-freescale first, maybe for qemuarm, so you get used to the process. Feb 27 14:34:40 LetoThe2nd: ty i'll try this and keep trying Feb 27 14:35:23 mauz555: 3) or 0), you decide: watch at least #0, and #8 of https://www.youtube.com/playlist?list=PLD4M5FoHz-TxMfBFrDKfIS_GLY25Qsfyj Feb 27 14:35:23 btw are poky/meta-poky and poky/meta-yocto-bsp mandatory (meta-freescale does not indicate depending on that)? Feb 27 14:36:24 they are probably not mandatory in themselves, but you need some distro Feb 27 14:46:34 mm, looks like the READ_ONLY feature is not installing at all sshd ... is this possible ? Feb 27 14:47:02 angelo__: i am very certain that it works nicely at least with dropbear. Feb 27 14:47:41 anyone considered uses of debuginfod in the yocto/oe context ? Feb 27 14:48:30 yann: RP added a feature request bug on that Feb 27 14:49:00 yann: https://bugzilla.yoctoproject.org/show_bug.cgi?id=13807 Feb 27 14:49:01 Bug 13807: enhancement, Undecided, ---, newcomer, NEW , Add support for elfutils debuginfod Feb 27 14:49:24 Oh, not RP Feb 27 14:49:52 erbo: https://twitter.com/TheYoctoJester/status/1230815094151446529 Feb 27 14:50:09 erbo: was rburton :) Feb 27 14:51:25 RP: almost the same initials :) Feb 27 14:51:33 when i use sudo umount /dev/sdb to umount my sd card it didn't work Feb 27 14:51:50 the contents on sd card still exist Feb 27 14:51:51 ? Feb 27 14:52:04 moustafa: maybe because the partitions was called /dev/sdb? Feb 27 14:52:10 moustafa: *what* are you trying to do? Feb 27 14:52:32 try to umount sd card Feb 27 14:52:55 @erbo no without partition number Feb 27 14:53:00 I've used df -h Feb 27 14:53:18 New news from stackoverflow: NOTHING PROVIDES 'virtual/x86_64-oesdk-linux-compilerlibs' Feb 27 14:53:27 When you say it still exist, do you mean that it's still mounted? Feb 27 14:53:33 Unmounting wont delete anything Feb 27 14:53:56 and you'd usually umount the mount point, not the device Feb 27 14:54:06 umount /media/sd or something Feb 27 14:54:42 maybe something still holds a file opne? Feb 27 14:55:07 rburton: you can point it to a device too Feb 27 14:55:15 in that case it should delay unmounting until no more things arein use. Feb 27 14:55:18 umount should complain if it could unmount Feb 27 14:55:21 sure but if you point it at the wrong device it doesn't work :) Feb 27 14:55:36 rburton: true :) Feb 27 14:55:41 ok thx Feb 27 14:55:45 erbo: nice, thx! Feb 27 16:05:28 can I add multiple files in SRC_URI using some wildcart characters? i.e. Feb 27 16:05:29 FILESEXTRAPATHS_prepend := "${THISDIR}/firmware:" Feb 27 16:05:31 SRC_URI += "file://*.bin file://*.elf" Feb 27 16:12:29 it was supported, not sure if it still is. it's almost always a bad idea, though Feb 27 16:30:24 kpo_: i think you can just put a directory name and it will copy the entire contents Feb 27 16:30:28 (if thats easier) Feb 27 16:33:48 Hey y'all, I'm new to Yocto (a few weeks now) and have recently built for a recipe which installs directly from a whl file using similar instructions to: https://stackoverflow.com/questions/48660051/yocto-recipe-python-whl-package Feb 27 16:34:39 But when I build I get errors that NOTHING PROVIDES 'virtual/x86_64-oesdk-linux-gcc'... I'm presuming the whl file does some compilation - how do I force it to the correct CC flags and so forth. I.e. how do I fix my error :( Feb 27 16:36:40 And the whl is: https://github.com/lhelontra/tensorflow-on-arm Feb 27 16:38:04 hello I could not find a way to build meta-freescale on zeus (https://pastebin.com/SaMyeGzs) python-pycrypto-native is missing. where can I find a layer that provides it ? Feb 27 16:40:03 mauz555: Are you sure your not on master of poky? Feb 27 16:42:09 JPEW: im on poky zeus Feb 27 16:42:16 git branch Feb 27 16:44:13 JPEW: 5e1f52edb7a9f790fb6cb5d96502f3690267c1b1 Feb 27 16:45:23 JPEW: same for meta-freescale im on "zeus" branch 94f4f086c6014cbcfd10bda3540d19558c8bf0b0 Feb 27 16:47:08 mauz555: Looks like its provided by meta-python (part of meta-openembedded): http://git.openembedded.org/meta-openembedded/tree/meta-python/recipes-devtools/python/python-pycrypto_2.6.1.bb?h=zeus Feb 27 16:47:24 Thats probably missing as a dependency in meta-freescale Feb 27 16:48:23 JPEW: on meta-freescale (MASTER) they seems to have updated this dependency: https://github.com/Freescale/meta-freescale/search?q=python3-pycrypto-native&type=Commits Feb 27 16:49:04 mauz555: Right, python2 is deprecated :) Feb 27 16:49:10 JPEW: https://github.com/Freescale/meta-freescale/pull/276 YES Feb 27 16:49:50 so python2 is drop after zeus ? Feb 27 16:50:22 mauz555: From oe-core. It's moved to meta-python2 Feb 27 17:11:50 mauz555: you need to have meta-openembedded/meta-python in bblayers.conf for meta-freescale zeus branch. Its README probably should get updated to mention that Feb 27 17:20:29 RP: I may have done something terrible: http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=jpew/reproducible&id=e6d75ece864ae08615798466cddf7c2b08eb992c thoughts? Feb 27 17:20:58 ^^ or anyone else for that matter Feb 27 17:31:07 JPEW: it would work but its rather ugly :/ Feb 27 17:39:55 close your eyes Feb 27 18:02:52 I was going to suggest only looking at it in a dimly lit room, but realized that wouldn’t work :D so maybe, only look at it in direct sunlight on a glossy screen. Feb 27 18:36:35 Hmm, apparently the nativesdk glibc we ship in the sdk fails to find its own gconv modules since we bypass its ability to be relocatable as built for an external toolchain.. guessing we need to set at least GCONV_PATH in the environment setup Feb 27 18:55:34 LetoThe2nd, did some more investigation on ssh thing Feb 27 18:55:41 system uses dropbear Feb 27 18:55:51 launched by socket activation (systemd) Feb 27 18:56:19 then i see Feb 27 18:56:20 loading /var/lib/dropbear/dropbear_rsa_host_key Failed Feb 27 18:56:36 Early exit: No hostkeys available. 'dropbear - Feb 27 18:56:36 R' may be useful or run dropbearkey Feb 27 18:56:57 can this be effect of the READ_ONLY option ? Feb 27 19:50:32 angelo__: not by default. rather something in your stack affects it. but if you want to e completely sure, kick off a pure poky build wir read-only-rootfs for qemu, and test there. Feb 27 20:19:50 Why are all my -native tasks failing because they can't find gnu-configize? Feb 27 20:20:15 Doing bitbake -c cleanall -native fixes it, but it's annoying :) Feb 27 20:51:24 Hello. I m running an x86 program on x86 based Yocto. However the program gives me "No such file or directory" Feb 27 20:51:37 The binary is present with correct permissions Feb 27 20:51:44 I m running core-image-minimal Feb 27 20:51:48 what could the issue be Feb 27 20:56:18 kiwi_2996: we've had something like that lately, was sometheting with the wrong ldd in the binary or such. Feb 27 20:56:52 kiwi_2996: let me guess, the binary was *not* bult using yocto :P Feb 27 21:00:11 LetoThe2nd yes Feb 27 21:00:21 binary was build on some other x86 platform (ubuntu) Feb 27 21:00:44 so I have to build binary using yocto? Feb 27 21:00:56 as in create a recipe and build it as part of yocto? Feb 27 21:01:16 kiwi_2996: it is definitively the recommended way. Feb 27 21:01:35 its usually wrong ldso encoded in binary Feb 27 21:01:38 it will also help you find crosscompilation problems. Feb 27 21:01:47 ah khem , the man! Feb 27 21:02:13 so best is to build an image with multilib if x86 is your target Feb 27 21:02:25 that sets the ldso paths like other distros do Feb 27 21:02:31 I see... Feb 27 21:02:38 what is the image name? Feb 27 21:02:45 core-image-multilib? Feb 27 21:02:50 even then you might be bitten by issues e.g. this binary maybe build with newer libraries than yocto Feb 27 21:03:02 and it will gripe about symbols Feb 27 21:03:26 so...any source is better to build as part of yocto by creating a recipe? Feb 27 21:03:29 kiwi_2996: you need to enable multilib there is a good howto on yocto wiki for that Feb 27 21:03:41 yes build from source is always better Feb 27 21:04:03 ok..thanks khem and LetoThe2nd .. I will try both options Feb 27 23:09:14 zeddii: this mips issues are there without the kernel upgrade so its something else in -next Feb 27 23:09:31 zeddii: which makes it my problem to find out what I guess :( Feb 28 01:34:58 RP: which mips issues ? **** ENDING LOGGING AT Fri Feb 28 02:59:57 2020