**** BEGIN LOGGING AT Fri Jun 30 03:00:04 2017 Jun 30 09:07:49 kergoth: I actually did bitbake formatting for pygments a while back, never got around to pushing it upstream. It's probably a bit out of date now, no support for unset for instance. We used the pygments support for markdown syntax highlighting. I'll see if I can push a branch somewhere. Jun 30 09:12:04 rburton: emacs multi-mode for bitbake: https://bitbucket.org/olanilsson/bitbake-modes , elpa packages from my personal repo https://olanilsson.bitbucket.io/packages/ : (add-to-list 'package-archives '("coan" . "https://olanilsson.bitbucket.io/packages/")) Jun 30 09:12:08 Morning..! Im writing a bunch of scripts for automating deployment of images generated by yocto ; I can see that oe-init-build-env sets a few variables, but I cant seem to inherit useful things such as DEPLOY_DIR etc... Is there a quick and easy way to get these into a shell script? Jun 30 09:13:01 rburton: Unfortunately it is not very good at indentation. No folding support I'm afraid. Jun 30 09:16:42 pev: basically, the env gets filtered by BB_ENV_WHITELIST. you can add things through BB_ENV_EXTRAWHITE, but usually there are reasons (mainly reprocutibility) why things are filtered out. so you have to use it with care (and not complain later about unexpected behaviour) Jun 30 09:18:54 pev: you can probably run "bitbake -e " and then grep for DEPLOY_DIR in the output Jun 30 09:19:30 LetoThe2nd: Ah, thanks. Looking under scripts there's a load of python scripts that seem to be a lot cleverer things such as wic having a get_bitbake_var() function. Is that to be preferred? Jun 30 09:19:53 abelal: Yeah, I can do that happily, it's more about automating deployment for multiple boards... Jun 30 09:21:16 pev: i'm not competent in those areas, sorry. our approach is a little different, as we automatically set up the build, run it, copy the artifacts that we want out, and then delete it again. Jun 30 10:53:00 "SendErrorReport Sending error reports ( running 8 hrs, 3 mins, 35 secs )" - YIKES! Jun 30 11:01:23 Hallo. I have a problem in do_populate_sysroot with several packages. The message: The recipe *** is trying to install files into shared area ... I don't know what I can do. I tried the mentioned solutions in the error message. But this doesn't solve my problem. As example zlib: There is only one recipe. I don't know what conflicts here. It want to install libz.so.1.2.8 and the symlink libz.so.1. In the mentioned manifest these two file Jun 30 11:01:47 First in: ../sysroots/dra7xx-evm/lib Jun 30 11:02:59 Second in: ../sysroots/dra7xx-evm/usr/include/glibc-locale-internal-arm7ahf-neon-linux-gnueabi/lib Jun 30 11:03:16 is this the conflict? And how I could solve this? Jun 30 11:46:24 Hello guys, is there a way to install an ipk file to the built image ? Jun 30 11:47:17 this is in case I can't install it from a package feed server Jun 30 12:04:41 kbo: yes, copy it manually and run opkg Jun 30 12:10:07 mckoan: Ok I was hoping to find something already included in some bb class Jun 30 12:10:19 or function Jun 30 12:28:17 hi, if upgraded my kernel but now the splash screen shows tux with blue feet and my gui application is also showing blue and red invers Jun 30 12:36:35 hello everyone. i'm trying to get a yocto build running in docker and have run into a bit of a problem Jun 30 12:37:21 bitbake execs some child tasks, and in the process tries to access /proc/$pid/io, which is not found. i think this has to do with dockers process namespacing, but i'm not sure Jun 30 12:38:19 this is the error output: https://hastebin.com/fedibugodo.log Jun 30 12:40:33 seems to be the buildstats system trying to access the processes to measure progress. is there a way to disable them? Jun 30 12:40:41 *it Jun 30 13:10:01 Hi, I try to integrate an executable to my image. Recipe by Recipe, everything works fine, but when comes the time of building the image do_rootfs complains with an error Computing transaction...error: Can't install capector-0.1-r0@cortexa9hf_vfp_neon: no package provides libudev.so.1 Jun 30 13:10:56 My first idea was to add a udev.bbappend linking libudev.so.0 to libudev.so.1 Jun 30 13:12:38 with FILES_{PN] += "${libdir}/libudev.so.1" and the link in do_install_append Jun 30 13:13:13 does someone know Why libusdev.so.1 is not found ? or where bitbake search ! Jun 30 13:17:28 udev provides this library Jun 30 13:18:22 MarcWe: it probably needs debugging what went wrong Jun 30 13:18:29 my udev provides libudev.so.0 not 1 and the binary I have to integrate is linked with the .1 Jun 30 13:18:49 oh its libusdev ? Jun 30 13:19:33 no sorry, it's a typo, libudev is the good one Jun 30 13:21:26 My question is more : How to force yocto to recognize the link to libudev.so.1 that I've created. Jun 30 13:30:57 hello guys.. does anybody know why systemd is always trying serial-getty@ttyS0.service without success ? Jun 30 13:31:10 it's filling every inch of the syslog Jun 30 13:31:39 it says /dev/ttyS0 is not a tty and continuosly repeat Jun 30 13:37:34 fberg: this is just what you get if you include poky/meta/recipes-core/systemd/systemd-serialgetty but seems like in your system there is no ttyS0 so it just always restarts the service Jun 30 13:38:24 Yes, but if I do a "$ls /dev" I can seed ttyS0 listed Jun 30 13:38:58 form the systemd.bb: # Use the upstream systemd serial-getty@.service and rely on Jun 30 13:38:58 # systemd-getty-generator instead of using the OE-core specific Jun 30 13:38:58 # systemd-serialgetty.bb - not enabled by default. Jun 30 13:39:44 Would it be possible to get http://lists.openembedded.org/pipermail/openembedded-core/2017-June/138883.html in master? read-only-rootfs + systemd + openssh is broken without it. Jun 30 13:47:01 is there a way of do the stuff yocto is doing for the package-split "outside" of yocto? is it just some script or how is this done? Jun 30 13:59:00 mdnneo: Most of that is burried in package.bbclass, what are you trying to do? Jun 30 14:00:24 jpew: try to misuse this to create packages via cmake ;) ... the plain cpack is not working like I would expect Jun 30 14:01:10 jpew: but ok than maybe its not that easy Jun 30 14:02:15 aand i've hit another snag, mpfr fails to build. the error is basically the same as this post: http://forum.solid-run.com/your-first-steps-beginners-corner-f17/wrestling-with-bitbake-error-mpfr-errno-20-not-a-d-t3308.html Jun 30 14:06:25 Hi =) How is the best way to really clean up a total build ? The reason why is I guess i di'nt do it properly so that i've a broken build as for exmeple ipk packages already on deploy/ when doing bitbake. Jun 30 14:06:53 rm -rf tmp/ sstate-cache/ deploy/ is a good start? Jun 30 14:18:07 ChrysD: Ya, you can start with that Jun 30 16:38:38 Hi Jun 30 16:38:49 Can I override .bbclass file? Jun 30 16:41:36 nope Jun 30 16:41:37 well, yes Jun 30 16:41:48 misread, htought you were asking about appending Jun 30 16:42:06 but yeah, just put it in your layer in classes/ and put your layer before the other layer in BBLAYERS Jun 30 16:44:03 thx, I knew that way. I want to make sure there is no better way Jun 30 16:44:17 not sure what you'd think would be better Jun 30 16:44:27 that's how you override any config file or class Jun 30 16:44:37 actually any file that's found relative to BBPATH.. Jun 30 16:47:19 I want to modify way ubinize.cfg file is generated. It is done by function inside image_types.bbclass file Jun 30 16:47:59 okay, so create your own bbclass that inherits image_types and then defines your own function of that name, replacing it with your version Jun 30 16:48:32 then set/add your class in IMAGE_CLASSES Jun 30 16:48:47 no need to override the class as a whole if all you need to do is replace one function Jun 30 16:48:53 at least for something like this Jun 30 16:49:20 Thx, I'll try that Jun 30 16:57:23 Hey guys - anyone got some tips for debugging kernel hangs? Jun 30 16:57:35 Aside from sprinkling printks all around the place? Jun 30 17:13:59 any one familiar with testimage tests here ? Jun 30 17:45:54 RP: good news is that I am able to reproduce the error seen in AB with hardening patchset Jun 30 17:46:13 RP: Looking at it it seems its flagging it wrong Jun 30 17:47:17 since same error is in syslog in passing case Jun 30 17:47:51 so its possible that its faulting on something else Jun 30 17:48:01 and reporting something else Jun 30 18:23:36 in case anyone else would find it handy, https://github.com/kergoth/dotfiles/blob/master/scripts/git-mvd, https://gist.github.com/kergoth/2c89fffb507618a97254e1af39725ca9 Jun 30 18:23:48 little thing, but was annoying me Jun 30 18:24:20 erm, obvious littel bug there, fixing Jun 30 18:24:39 still :) Jun 30 18:27:04 checking it out now Jun 30 18:33:59 fixed. Jun 30 18:34:14 * kergoth has to relocate appends and whatnot all the damn time and got sick of mkdir -p + git-mv Jun 30 18:37:13 indeed. Any time I find myself typing a command more than once/hour I script it Jun 30 19:40:10 Hey guys, I'm making a recipe for a package that has CMakeLists.txt in the root folder, how do I tell cmake where to look? Jun 30 19:40:26 *that DOESN'T have Jun 30 19:42:55 sorry, i just thought to search the IRC logs... OECMAKE_SOURCEPATH Jun 30 19:43:22 wasn't having luck with the manual or google. Thanks IRC! Jun 30 20:39:16 nathani_: look at cmake.bbclass Jun 30 20:55:23 this is odd, when i build with one local.conf, ovmf builds fine, with another, it fails, but bitbake-diffsigs -t after doing a -S on both shows no differences Jun 30 20:55:24 :| Jun 30 20:56:57 gah Jun 30 21:02:05 is that because of some variable being ignored by BB_HASHBASE_WHITELIST (or some other more fundamental flaw)? Jun 30 21:15:52 ah ha, the ovmf failure is a race, crank make jobs up to 48 and it explodes Jun 30 21:16:03 er, 64 Jun 30 21:17:05 That must be a nice problem to have :) Jun 30 21:17:15 Hey - I'm building for a board based off the NXP Sabre colibri iMX7d Jun 30 21:17:36 I get into u-boot fine, and the kernel gets up to the point of "starting UTP" Jun 30 21:17:44 then it just stays there. Jun 30 21:18:18 I tried disabling everything having to do with UTP in it, but it's still happening. Jun 30 21:18:42 Does anyone have some insight into how I should troubleshoot this, or what is going on? Jun 30 21:20:19 UTP appears to only be used in by the usb mass storage gadget, so I disabled usb gadget support altogether but it's still happening :( Jun 30 21:27:22 huh, weird, i just noticed /var/volatile/tmp/tmp is a symlink pointing to /var/tmp which in turn is a symlink pointing to /var/volatile/tmp. how very circular Jun 30 21:29:27 tmp/tmp ? that is kind of lame Jun 30 21:31:26 I've noticed those when symlinks are made by volatiles (from /etc/default/volatiles) Jun 30 21:36:15 i.e. we run tmpfs on /var/run, but redirect /var/run/crash to a persistent location (e.g. /opt/crash) via a volatile. Invarible that ends up with an extra symlink /var/run/crash/crash -> /opt/crash Jun 30 21:37:00 huh, yeah, sounds like a bug in the volatiles script for sysvinit builds then Jun 30 21:37:27 ya, its been that way since at least 1.6, but it hasn't bothered me enough to track it down Jun 30 21:37:52 * kergoth nods Jun 30 21:37:53 minor annoyance Jun 30 22:20:22 So I discovered that uuc is a standalone program that handles UTP type stuff Jun 30 22:21:13 I don't find it anywhere in my rootfs, so I'm assuming it's in the initramfs. Is that true? And if so, how do I build that? I don't see it in my images dir in yocto **** ENDING LOGGING AT Sat Jul 01 03:00:01 2017