**** BEGIN LOGGING AT Thu May 25 03:00:03 2017 May 25 05:40:34 denix: bluelightning: twall is gone, i'm too late to the party :-( May 25 05:40:58 i ran into this same problem a couple weeks ago as i tried (in vain) to get accelerated graphics working on the bbb May 25 05:41:39 what i didn't realize is that the binary graphics stuff only works with wayland, there's no support for x11 with the PowerVR binaries May 25 06:09:37 morning all May 25 06:35:16 tlwoerner: interesting, I'd have assumed it was the other way around May 25 06:35:28 apparently twall wanted wayland support so that would be good though May 25 06:36:25 bluelightning: :-) May 25 06:36:32 i guess wayland is the future May 25 06:57:08 if recipe A (xf86-video-armada) DEPENDS on package B (libdrm-armada) but A expect's B to have an underscore in its name (libdrm_armada) instead of a hyphen, how would this be handled? May 25 07:00:13 what is the proper way to create a mingw/windows toolchain/sdk with yocto? May 25 07:00:26 i've added meta-mingw to my bblayers May 25 07:00:49 and set SDK_OS to mingw32 May 25 07:00:58 is there something else to do? May 25 07:01:18 on krogoth i get this error: https://pastebin.com/vSPgEH2C May 25 07:01:29 on other branches i get other errors... May 25 07:04:00 fmeerkoetter: my understanding is that meta-mingw is mostly abandoned (?) May 25 07:04:33 fmeerkoetter: i think the current suggested method is to use CROPS -> https://git.yoctoproject.org/cgit/cgit.cgi/crops/tree/README.md May 25 07:05:07 basically build an eSDK and run it in a container with some glue so it integrates with eclipse or the cmdline May 25 07:05:34 tlwoerner: thanks! that is good to know. i wasn't aware that meta-mingw is dead May 25 07:05:56 fmeerkoetter: i *think* it's dead :-) May 25 07:06:11 ack May 25 07:06:35 fmeerkoetter: actually, the link i sent you is a really bad link (really old CROPS stuff) May 25 07:06:57 tlwoerner: why do you think meta-mingw is dead? May 25 07:07:08 they don't use ceed anymore May 25 07:08:27 nrossi: i thought it was dead because my impression is that it's broken more often than not, and instead of trying to maintain it, the SDK gurus have decided to use containers instead May 25 07:08:45 maybe i'm thinking of the macos native work? May 25 07:09:32 in any case, i *believe* the official position of The Yocto Project is to use CROPS and containers to develop with Windows and MacOS hosts (?) May 25 07:09:50 tlwoerner: ah ok, it does get broken easily. Although I think it is still very much used, probably doesn't get enough attention though... May 25 07:10:35 https://www.yoctoproject.org/sites/default/files/improving_the_yocto_developer_experience.pdf May 25 07:12:19 i think its for example in use by the meta-boot2qt/qtfordevice creation distro done by The Qt Company May 25 07:12:20 afaik May 25 07:12:46 fmeerkoetter: ^^ this looks more up-to-date May 25 07:13:51 https://www.youtube.com/watch?v=d3xanDJuXRA May 25 07:18:18 fmeerkoetter: maybe https://wiki.yoctoproject.org/wiki/TipsAndTricks/CropsCLIContainers May 25 07:18:22 i'll check May 25 07:19:09 a bit of input from me. it might become less important with windows10. colleguages are having good results with using a regular linux sdk in the windows10 linux environment May 25 08:47:12 fmeerkoetter: cool! who would have guessed?! May 25 08:49:57 hi, when i add in conf/local.conf KERNEL_IMAGETYPE+="uImage" it produces errors "disk full" while building the rpi-sdimg, when i use KERNEL_IMAGETYPE="uImage" it runs without errors and builds u-boot in tmp/deploy/images/raspberrypi/ but when i dd and load the image, there is no u-boot booting ? what do i miss May 25 08:51:32 the only two lines in conf/local.conf are MACHINE ??= "raspberrypi" and KERNEL_IMAGETYPE = "uImage" May 25 08:51:52 the only two lines i added May 25 09:34:55 If anyone would like to send their first OE patch then https://bugzilla.yoctoproject.org/show_bug.cgi?id=11575 is trivial and i'll happily give guidance May 25 09:34:56 Bug 11575: normal, Undecided, 2.4 M1, ross.burton, ACCEPTED , ERROR: cryptodev-tests-1.8-r0 do_checkuri: Fetcher failure for URL: 'http://download.gna.org/cryptodev-linux/cryptodev-linux-1.8.tar.gz' May 25 09:46:18 why is xterm in fsl-image-machine-test missing ? is it a bug May 25 09:50:45 vm__: i guess a good followup question is should it be in there, or are you expecting it to be there but it isn't May 25 09:54:10 rburton, my mistake, i missed meta-openembedded May 25 10:17:47 when i do bitbake -e rpi-hwup-image|grep UBOOT_MACHINE it returns a line, but not in which file this line is, is that possible May 25 10:58:11 vm__: don't grep, use less and you'll see the variable history above including where it gets set May 25 11:02:36 rburton, how do i run this command with less? pipe into less didnt work May 25 11:03:24 bitbake -e rpi-hwup-image | less May 25 11:03:57 grep is great and all but you don't want to see the line that sets UBOOT_MACHINE, you want to see the one above it May 25 11:04:37 yeah ,thx May 25 11:07:44 bitbake -e rpi-hwup-image | grep -B 10 UBOOT_MACHINE is even more convienient May 25 11:09:36 if 10 is enough May 25 11:09:45 complicated logic may have more May 25 11:09:53 so i just | less and then search from there May 25 11:10:04 ie /^UBOOT_MACHINE= May 25 11:10:53 yes, it depends on, in this case it was enough May 25 13:04:28 rburton: hi, can you merge this fix: http://lists.openembedded.org/pipermail/openembedded-core/2017-May/137114.html ? I should fix meta-intel build I hope. May 25 13:24:12 good morning folks May 25 13:25:49 i have my raspi configured wrong still, how do I get it to install package management (smart), xfce, and sshd ? May 25 13:30:47 berndhs: mostly by setting IMAGE_FEATURES as appropriate: package-management will give you a package manager, ssh-server-(openssh or dropbear) will get a sshd. for xfce, you need to add meta-xfce to layers and look at the same images that layer has. May 25 13:32:36 hmm, let me check if I did that. I thought i did, but perhaps not. May 25 14:00:49 hello May 25 14:01:29 any odroid C2 owner is hear? May 25 15:43:37 rburton: | /home/pokybuild/yocto-autobuilder/yocto-worker/nightly-mips-lsb/build/build/tmp/work-shared/edgerouter/kernel-source/fs/aufs/dir.c:88: undefined reference to `____ilog2_NaN' May 25 15:44:06 rburton: it seems your gcc7 testing did not pick up bruce's changes or bruce did not backport the fix May 25 15:44:17 i re-kicked the lsb builds as i forgot to pick the 4.1 fix May 25 15:44:28 ah ok May 25 15:44:37 yeah they're still running May 25 15:45:17 https://autobuilder.yoctoproject.org/main/builders/nightly-mips-lsb/builds/1113 May 25 15:45:19 failed May 25 15:45:31 nuts May 25 15:45:44 zeddii: ^ May 25 15:45:51 and its with the error I mentioned May 25 15:45:57 yeah May 25 15:47:30 I meet a problem with the recently poky, the sdk I build won't install pc files for pkg-config May 25 15:47:43 rburton: meta-yocto-bsp = "heads/ross/gcc7:49d649a7f50cc078b4916b71f39cf02fc3e86168" May 25 15:47:44 meta-qt4 = "heads/master:033509ebc1eb0d13290a7a3f0c2772ed58712c34" May 25 15:47:53 any idea with that May 25 15:47:55 it did not pick your kernel cherry pick May 25 15:48:08 hm May 25 15:48:14 ok i'll try again May 25 15:48:36 rburton: wait no it did May 25 15:48:41 SDK_INCLUDE_PKGDATA = "1" is not default set today? May 25 15:48:45 http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/commit/?h=ross/gcc7&id=49d649a7f50cc078b4916b71f39cf02fc3e86168 May 25 15:49:01 so the SHAs match May 25 15:49:14 since linux-yocto uses branches May 25 15:49:15 ayaka: that variable has nothing to do with pkg-config May 25 15:49:26 I am thinking zeddii forgot about mips May 25 15:49:35 rburton, so what is the problem with that ? May 25 15:50:50 here it seems to be in malta-32 br http://git.yoctoproject.org/cgit/cgit.cgi/linux-yocto-4.1/commit/?h=standard/mti-malta32&id=398d1785c6bde0085b99f00f346e199a15fbdcef May 25 15:52:42 his SRCREV bump is asking for 7420516ff545f8ab3e2471cf08bf851744264a6a May 25 15:52:54 and that does include the backport May 25 15:53:53 but the build says linux-yocto-4.1.38 May 25 15:53:59 whereas it should be linux-yocto-4.1.39 May 25 15:57:19 same for mips64 May 25 15:57:33 rburton: so I think it did not get the right kernel srcrev May 25 15:57:35 somehow May 25 15:58:49 is there a way to go into workspace on AB May 25 15:58:56 and inspect the metadata May 25 16:13:57 rburton, it looks all the dev packages are not installed from the target.manifest May 25 16:14:06 https://paste.fedoraproject.org/paste/qhXBLYAPnHpR4MV6MoQ2915M1UNdIGYhyRLivL9gydE=/ May 25 16:21:30 hi folks, how does one restart the network service, other than reboot? May 25 16:22:10 and how does one define a static interface, with constant ip for eth0 inet? May 25 16:22:34 berndhs: depends on what init system you use May 25 16:23:19 i'm using the out of the box one, whatever the default is. Not systemd. May 25 16:23:21 what the hell, this makes no sense. May 25 16:23:45 i have anonymous python setting the 'dirs' flag on a bunch of tasks. bitabke went to run the task, and it failed with the 'cd' that *bitbake added* saying the dir didn't exist May 25 16:23:51 kergoth: thats why ignorance is bliss .. May 25 16:23:53 so bitbake used the dir from dirs for the cd, but didn't create it May 25 16:24:10 berndhs: sysv, so /etc/init.d/whatevertheserviceiscalled restart May 25 16:24:12 the only thing that makes sense is some sort of race, since multiple tasks are using the dir, but nothing sets cleandirs anywhere May 25 16:24:59 kergoth: infact I used dirs to fix non existing dir issue just yesterday May 25 16:25:12 so I know it works *mostly* May 25 16:25:44 hrm May 25 16:26:12 this is a recipe that has a bunch of individual tasks all populating a single dir, then a master task that depends on those and deploys that output with sstate. basically exactly how do_image_complete is handled May 25 16:28:31 rburton: is there a good writeup on ip so I can configure /etc/network/interfaces correctly? Once I can ssh into my raspi i can make progress. May 25 16:28:48 but now I'm stuck, no network connection. May 25 16:28:59 berndhs: the debian documentation as its literally copied from debian May 25 16:29:18 or use connman as that will automatically bring up ethernet when you plug a wire in May 25 16:30:00 connman i have to put in the local.conf? The IMAGE_somethings? May 25 16:30:28 you install connman May 25 16:30:52 i have no network, hence no package installation May 25 16:31:56 install connman in your image May 25 16:43:36 kergoth: I see bitbake bail out on large loads like all meta-oe layers included ( typical world builds beyond just oe-core ) May 25 16:43:40 ERROR: Worker process (24195) exited unexpectedly (-9), shutting down... May 25 16:43:43 ERROR: Worker process (24195) exited unexpectedly (-9), shutting down... May 25 16:43:45 ERROR: Worker process (24195) exited unexpectedly (-9), shutting down... May 25 16:43:47 WARNING: /mnt/a/oe/sources/bitbake/lib/bb/runqueue.py:1179: ResourceWarning: unclosed file <_io.BufferedWriter name=71> May 25 16:43:50 check dmesg, i'd suggest OOM kill May 25 16:43:53 let me have a try on the latest poky May 25 16:43:54 s/suggest/suspect/ May 25 16:44:23 [ +0.000002] Out of memory: Kill process 23962 (Cooker) score 20 or sacrifice child May 25 16:44:26 [ +0.000015] Killed process 24195 (Worker) total-vm:332932kB, anon-rss:139760kB, file-rss:0kB, shmem-rss:0kB May 25 16:44:39 hmmm May 25 16:45:16 so may be I should limit my parallelism May 25 16:46:36 ah i have connman installed already May 25 16:47:15 the auto definition of eth0 or wlan0 is optimistic, but i have something i can work with at least :) May 25 16:49:21 khem: if you're on ubuntu 16.04, there was a bug that was resulting in me hitting on all the freaking time onn one of my vms May 25 16:49:22 khem: see https://blog.meteor.com/how-to-stop-ubuntu-xenial-from-randomly-killing-your-big-processes-4a3e2d09323f May 25 16:50:01 i think they fixed it, but worth checking to see if you have the update May 25 16:50:37 kergoth: I am on archlinux May 25 16:50:50 ah, then yeah, might want to back off a bit on the parrallelism :) May 25 16:51:06 kergoth: but I know my DRAM is not enough when you build world May 25 16:51:27 since I cant control if it will start building firefox, chormium and webkitgtk all at same time May 25 16:52:09 * kergoth nods May 25 16:52:13 i've had that problem too May 25 16:52:43 i have a conf/slow.conf in my ~/.oe, then i add that to my baseline BBPATh in my bblayers.conf, and do a bitbake -R conf/slow.conf webkitgtk:do_compile when i have to May 25 16:52:45 heh May 25 16:53:57 connmanctl does not support -v or -version or --version, they are lame people :P May 25 16:55:49 tempted to add a custom bitbake scheduler class to let me mark certain tasks to run only by themselves, with nothing else, or something May 25 18:10:53 is it possible to build the esdk with meta-qt5 ?! May 25 18:11:41 I see that patchwork is not being cleaned - accepted patches aren't archived... May 25 18:27:56 is master now fully open? May 25 19:27:22 denix: You are correct, patchwork's patch state updating is pending. We'll perform a manual updating in the following days. May 25 19:31:17 lamego: thanks. it would be helpful for tracking pending patches May 25 19:32:07 rburton: is RP on vacation? May 25 19:51:53 what is better. buildroot or yocto? May 25 19:53:17 they're two different projects with entirely different goals and focuses May 25 19:53:45 buildroot is great and fast for board bringup and the like, if you just need a one-off rootfs, it works great, but for long term maintenance of your device and its distribution, not so much May 25 19:53:46 it's like asking what is better - beer or wine? May 25 19:53:51 indeed May 25 19:54:35 a lot of folks assume it's direct competition or antagonistic, but i see them as more complementary, depends on your goals and requirements May 25 19:55:21 Quick question, anyone know what happened to the Morty tag on github OpenEmbedded? May 25 19:56:41 MikeAtMT: which layer? and morty is not a tag, it's a branch May 25 19:56:59 Sorry, branch May 25 19:57:06 openembedded-core May 25 19:57:17 same with cgit.openembedded.org May 25 19:57:51 wow, indeed! May 25 19:58:00 morty branch is gone! May 25 19:58:36 rburton, RP1: ^^^ May 25 20:04:06 halstead: ^^^ May 25 20:04:30 denix, Eek. May 25 20:07:00 denix In the logs it looks like a push from RP1. Should be easy to fix. May 25 20:07:52 halstead: thanks! I figured it was an accident and nothing malicious... :) May 25 20:09:48 Thanks, denix, all May 25 20:09:57 denix, The repo itself still has refs/heads/morty pointed at ce06f3a0c9859df1f897583659f836234c47d2d7 May 25 20:10:17 I'm looking into why it's not showing in cgit. May 25 20:12:16 halstead: git pull also complains 'refs/heads/morty' does not exist May 25 20:12:55 halstead: that is from my local checkout that's on 55fa8b210139509859258c0ee11b3534f10fa509 May 25 20:14:42 denix, I see the same when cloning the public read-only url. A read-write clone does have the morty branch. May 25 20:21:06 denix, It wasn't RP1's push, it's a umask issue after security updates yesterday that the push revealed. May 25 20:22:53 kergoth: buildroot - not so great for long term maintenance of your device - why not? May 25 20:24:06 denix, MikeAtMT , fixed now for clones and the web view. May 25 20:24:33 Thanks much, halstead! Cheers! May 25 20:24:35 halstead: heh, all these security updates are causing all kinds of problems... :) May 25 20:24:43 halstead: thanks May 25 20:25:41 Thanks for the ping. Looks like it was broken for 3 hours. May 25 20:26:13 You bet May 25 20:30:12 halstead: it's funny, because I just had to deal with our IT dept past few days for a problem that was also caused by a security update... :) May 25 20:31:56 denix, Sounds like getting a fix from me was a faster process. May 25 20:32:35 halstead: absolutely! :) ours was email related and took some time to debug... May 25 20:32:51 Investigating this it seems the problem may have been laying in wait for awhile. Gotta make sure it doesn't happen again. May 25 20:33:44 halstead: thanks again for your help May 25 20:34:01 :) May 25 22:14:02 how do you at the qt5 meta layer to the morty/pocky structure May 25 23:40:26 i got this error: ERROR: Task 518 (virtual:native:/home/kcroot/embedded/yocto-2.2/poky/meta/recipes-devtools/elfutils/elfutils_0.164.bb, do_compile) failed with exit code '1' May 25 23:40:43 i got this error May 26 00:39:34 kcroot: whats the error message ? May 26 00:40:01 black_13: you can use bitbake-layers add-layer cmd May 26 00:40:41 first clone the matching branch e.g. git clone git://github.com/meta-qt5/meta-qt5 -b pyro May 26 00:41:16 then cd into poky and use bitbake-layers add-layer May 26 00:42:48 does the follow instructions make sense http://gpupowered.org/node/20/ May 26 00:43:27 khem where do clone into the poky directory? May 26 00:47:56 it doesnt matter but yeah into poky is ok **** ENDING LOGGING AT Fri May 26 03:00:03 2017