**** BEGIN LOGGING AT Tue Apr 24 03:00:02 2018 Apr 24 06:56:35 not a best day for dear bitbake Apr 24 06:56:36 [Tue Apr 24 05:31:40 2018] Killed process 23781 (Worker) total-vm:405792kB, anon-rss:274604kB, file-rss:0kB, shmem-rss:0kB Apr 24 06:56:40 [Tue Apr 24 08:46:15 2018] Killed process 6988 (Worker) total-vm:406252kB, anon-rss:275056kB, file-rss:0kB, shmem-rss:0kB Apr 24 06:56:43 [Tue Apr 24 08:46:29 2018] Killed process 6100 (PRServ) total-vm:322660kB, anon-rss:180652kB, file-rss:0kB, shmem-rss:0kB Apr 24 07:08:15 * LetoThe2nd kills some more bitbake processes, just becuase. Apr 24 07:48:09 New news from stackoverflow: YOCTO - MTD drivers implementation Apr 24 08:09:22 Hello! Why is the field "Filename" disallowed in opkg-build? Does it break something? Apr 24 08:10:33 Hello ! Apr 24 08:12:59 I'm trying to add an user in a bbappend file with EXTRA_USERS_PARAM, I get not error but the user is not added.does someone have any clues ? Apr 24 08:13:14 ops I forgot my bbappend file : http://dpaste.com/38AA408 Apr 24 08:49:21 Good morning. Apr 24 08:50:52 Been a while since I used Yocto, so I a little rusty. Is there an easy way to list out all the URIs used for a target? I need to give IT a list of protocols and ports I want allowed through a firewall. I can guess SVN, Git, etc but an exhaustive list would be good. Apr 24 08:51:22 oh nvm I found that useradd-example.bb, I should not be using EXTRA_USERS_PARAM. Apr 24 08:55:49 TafThorne1: usually you should be fine with git svn http ftp Apr 24 08:59:34 LetoThe2nd: I'll add https :) Apr 24 09:00:41 jww: i personally don't make a distinction there, but technically you're absolutely right. Apr 24 09:12:17 Thank you Apr 24 09:12:38 I ended up doing: Apr 24 09:12:38 rgrep -ho '\w*://' | sort | uniq -c Apr 24 09:12:38 In my unbuild repo Apr 24 09:12:55 Little crude but it seems to have given some hints: Apr 24 09:12:55 rgrep -ho '\(f\(ile\|tp\)\|git\|svn\|http\(s\)*\)://' | sort | uniq -c Apr 24 09:13:00 27 :// Apr 24 09:13:00 1 addr:// Apr 24 09:13:00 1 Binary file bitbake/doc/template/VeraMoBd.ttf matches Apr 24 09:13:00 1 Binary file bitbake/doc/template/VeraMono.ttf matches Apr 24 09:13:00 1 Binary file bitbake/doc/template/Vera.ttf matches Apr 24 09:14:02 oops Apr 24 09:14:06 Sorry about that. Apr 24 11:06:20 Hi all, wanted a suggestion what would be the best way to do the following, I want to add a dependency on image which I did do_install[depends] += "xx-image:do_image_complete" now I want add some additional files to the tar.gz which was created by xx-image recipe, what is the best approach for it ? Or should I just manually extract and add the file.. Apr 24 11:37:09 prabhakarlad: why would you not just create a new image recipe and include the files in that? Apr 24 11:37:24 Or bbappend the image and append some code to a phase where the tarball is created? Apr 24 11:55:39 neverpanic: I just wanted a proper way to do things, which I was not aware of the correct way to do it, thanks for the suggestion, Ill create a new image recipe and include file in there, I want to encrypt the fs (fs.tar.gz) which is created by the image recipe so should I be using pkg_postinst_${PN} to do it ? Apr 24 11:56:56 I don't know, sorry. You'll have to read the image recipe and its bbclasses to find a suitable place. Apr 24 12:09:09 so the image generation code is flexible. if you want to encrypt the image, then make a new image type which does the encryption Apr 24 12:10:02 see IMAGE_TYPES in image_types.bbclass Apr 24 12:10:21 that lists all the image types that are available Apr 24 12:19:47 you can look https://github.com/sbabic/meta-swupdate/blob/master/classes/swupdate.bbclass for example, a bit more complex than just encryption Apr 24 12:35:24 nayfe: rburton_: thanks for the pointers :), I want to do something similar to swupdate Apr 24 12:49:04 New news from stackoverflow: Font size and icons are very small in Yocto Image Apr 24 12:56:20 prabhakarlad: you want to create your own system update? Apr 24 12:57:35 please not another system update tool Apr 24 12:58:13 :D Apr 24 12:58:33 there are 15 competing standards... Apr 24 13:11:08 is it possible that bitbake refuses to start on a Ubuntu 14.04 host? I tried now rocko and sumo, bitbake starts with no messages and freezes Apr 24 13:11:54 I can't figuer out whether depends on the OS versione though Apr 24 13:14:24 nayfe: I already have swupdate system which works, the only thing is it takes in an encrypted (openssl) tar file which has all the files to update a system, which I wanted to create a recipe which creates a encrypted tar file (mlo+u-boot+rfs) Apr 24 13:15:15 after a while it complains about network, but network is working (ping and git clone worked) Apr 24 13:19:11 New news from stackoverflow: IMAGE_INSTALL and IMAGE_INSTALL_append fails to add packages in yocto build Apr 24 13:20:34 example.com not responding, doh! Apr 24 13:21:17 so if example.com is out of order bitbake freezes? Apr 24 13:49:05 Where can i find all the parameters that i can use in my wic files ? Apr 24 14:02:06 Is there a simple way to link a package's PACKAGECONFIG with a image feature? Or perhaps disto feature? Apr 24 14:02:55 Is is possible to link a recipe's config up with a image feature, or is image features just about the specific collection of packages and thus shouldn't image feature get linked to a packageconfig? Apr 24 14:15:15 osten: https://www.yoctoproject.org/docs/current/mega-manual/mega-manual.html#ref-kickstart Apr 24 14:20:58 sveinse: PACKAGECONFIG are modified at distro conf level, so you can do PACKAGECONFIG_append = "${@bb.utils.contains('DISTRO_FEATURES', 'XX', ' XX', '', d)}" i don't think you can change it for a specific image Apr 24 14:21:23 sveinse: with image feature, literally impossible Apr 24 14:21:40 distro features, lots of examples in oe-core Apr 24 14:21:49 in that the majority of packageconfigs respect distro features Apr 24 14:22:24 you can't do it per image because the workflow is recipe creates packages, packages are used to construct images Apr 24 14:22:40 if you've two images being built at the same time, they'll use the same packages Apr 24 14:42:39 Thanks guys, that's what I expected for image features. Apr 24 14:43:55 hello guys ! I want to start dhcpcd on eth0 at boot. I have the dhcp@.service , how can I instruct bitbake to enable dhcp@eth0.service on the target system ? Apr 24 14:47:09 Is it sufficient to add SYSTEMD_SERVICE_${PN} = "dhcpcd@eth0.service " ? Apr 24 14:47:12 rburton_: how can we have two images being built ? Apr 24 14:47:29 $ bitbake some-image some-other-image Apr 24 14:47:50 if you build a hddimg then you're already building two images as that has an initramfs in, which is an image Apr 24 14:48:13 rburton_: Great thanks I though it be only one image or one package per bitbake command Apr 24 14:48:22 nope, any number Apr 24 14:48:49 ok, thank you for that info Apr 24 15:36:48 Is PACKAGECONFIG_pn-xyz_append needed or is it implicit in the PACKAGECONFIG that you can do PACKAGECONFIG_pn-xyz = "additional-config" ? Apr 24 15:41:12 nayfe: regarding PACKAGECONFIG with DISTRO_FEATURES, what I don't like about selecting that featureset over distro features is that it forces a complete rebuild of a lot of things, including glibc, which makes changes here affect more than just selecting a global config for a few packages Apr 24 15:41:13 no append would just assign instead of extend Apr 24 15:41:22 also append before pn_xyz Apr 24 15:41:28 rburton_: great, thanks Apr 24 15:42:09 So it makes me think that using DISTRO_FEATURES is not the right hammer for that kind of settings Apr 24 15:43:38 sveinse: yeah don't abuse distro features for custom settings unless they affect *a lot* of recipes Apr 24 15:43:46 altering distro features doesn't rebuild everything that uses it, only recipes conditional upon the existence or non-existence of particular entries in it, due to use of contains() Apr 24 15:44:05 also changing DISTRO_FEATURES should only rebuild eg glibc if you change a feature it uses Apr 24 15:46:17 Strange, not what I'm seeing. So let me backup a second: I want a "official" flag which shall build a release image. It affects package composition and it shall affect PACKAGECONFIG config for a few packages which needs this information (like not displaying "BETA version"). Apr 24 15:46:57 The only thing I added from local.conf was DISTRO_FEATURES_append = " official-build" and now I'm seeing the glibc and qt is all being recompiled Apr 24 15:48:07 sveinse: just use a variable for that., much easier Apr 24 15:53:58 a variable would work, yeah, though adding something to distro features shouldn't rebuild anything if it's not used, if that's happening, and you can reproduce the behavior, please open a bug, and potentially use bitbake-diffsigs or bitbake -S printdiff to examine the delta Apr 24 16:10:09 sveinse: of course if you make a really low-level library respect the feature then everything that depends on that will rebuild Apr 24 16:21:54 kergoth: 1) bitbake image, 2) make conf changes, 3) run bitbake -S printdiff image ? Apr 24 16:30:48 pretty much Apr 24 16:31:03 khem: are you running a host with gcc8? Apr 24 17:09:37 I don't get this: If I run bitbake image through to completion, and then run bitbake -S printdiff image straight after it complains about one of my packages like "Variable PV value changed from '6.4.0+hgAUTOINC-7ebe100557a0+' to '6.4.0+hgAUTOINC-e8bedaa2912c+'". However I checked the image manifest from the first build and it does already contain the correct latter version. Why then is bitbake considering Apr 24 17:09:43 the former? Apr 24 17:15:20 If I add DISTRO_FEATURES_append = " official" to local.conf, I get a change which triggers glib-2.0:do_install due to hash change, see https://bpaste.net/show/25bd0e2e5020 . What can I do next to compare why and what in the glibc-2.0:do_install taskhash which has changed? Apr 24 17:19:10 sveinse: are you using AUTOREV for that compoennt? Apr 24 17:20:02 sveinse: I'm using openembedded-core/scripts/sstate-diff-machines.sh to create a "snapshot" of all the signatures before and after, then you can use bitbake-diffsigs to compare any unexpectedly changed task signatures Apr 24 17:20:49 bitbake-diffsig sais its only the DISTRO_FEATURES value that has changed in glib-2.0:do_install Apr 24 17:21:43 JaMa: Not precisely AUTOREV, but somewhat similar feature Apr 24 17:23:15 that reminds me, i'd really like to add another -S command like printdiff, but which *only* considers STAMPS_DIR and doesn't compare against any possibly existing sstate checksums, to make it more like bitbake-whatchanged Apr 24 17:23:43 SRCPV := "AUTOINC-${@get_hgid('${R}',d)}" and PV = "${BASE_VERSION}+hg${SRCPV}". It a part of a longer story. Its a larger in-source layer Apr 24 17:28:55 once I reran bitbake image after setting DISTRO_FEATURES_append, it started recompiling glib-2.0. do_compile is earlier in the process than do_install. Something fishy is going on Apr 24 17:31:18 And if I abort the compilation (of everything) and run bitbake-diffsigs -t glib-2.0 do_compile, I get an extensive list https://bpaste.net/show/1bb0c0e0bd5b Apr 24 17:32:36 For some reason my target seems to be have been configured as TUNE armv5e, while I'm having a cortexa9hf-neon Apr 24 17:33:33 Either bitbake-diffsigs is giving me junk, or there is something really incorrect with this machine definition, alternating between values Apr 24 17:35:22 If I build an image, make a change to local.conf, rerun bitbake image and abort it in the compilation, can I then trust that bitbake-diffsigs will give me sane values? Apr 24 17:37:45 Hi, I have a eventhandler in one of my bbclass file in my layer. I have set the mask to bb.event.BuildStarted. But, I do not see my event handler getting triggered. I put debug prints sanity.bbclass and I see that getting triggered for the same event handler Apr 24 17:38:10 Can someone please tell me if there is anything else I need to do in my layer to register event handlers? Apr 24 17:39:08 Can it be that this machine does not like that I put DISTRO_FEATURES_append=" something" in local.conf and that that ruins some default setting which alters the behaviour completely? Apr 24 18:06:16 I'm trying to run a bash script relying on here commands. This script runs under fakeroot. I get /dev/fd/62: No such file or directory. Is that because of the run environment? Apr 24 18:10:47 all you need to do for diffsigs is have sigs to compare. bitbake -S none someimage; : make some change; bitbake -S none someimage or bitbake-diffsigs. afaik Apr 24 18:14:23 rburton: not host but I have gcc8 recipes for OE Apr 24 18:17:04 so pseudo seems to be using __builtin_apply which is a gcc specific built-in I guess I will mark it non-clangable now Apr 24 18:17:34 rburton: see https://github.com/kraj/openembedded-core/commits/kraj/master Apr 24 18:22:45 I've given up. Something strange is happening if I set DISTRO_FEATURES_append from local.conf, but I'm out of time to fix this. Going to to use another variable for it Apr 24 18:24:13 It insists on rerunning glib-2.0:do_compile, but bitbake-diffsigs doesn't highlight what it is Apr 24 18:25:57 sveinse: Just generate bitbake -e output with and without DISTRO_FEATURE_append and see what changes in DISRO_FEATURES settings Apr 24 18:26:43 khem: I did, and the only difference was the setting I was appending Apr 24 18:28:03 As was said earlier: Changing DISTRO_FEATURES should only rebuild glibc if a feature it uses changes. But apparently something is trigging a rebuild of glibc which I'm unable to find Apr 24 18:28:27 what feature are you appending ? Apr 24 18:28:36 ' foobar' Apr 24 18:28:41 ' anything' Apr 24 18:30:13 oh anychange in DISTRO_FEATURE results in glibc rebuilt Apr 24 18:31:14 khem: ah, ok, that clears it up then. Good. The DISTRO_FEATURE is definitely not the right hammer for this functionality. Thanks Apr 24 18:31:49 no that was not a statement, but an assertion Apr 24 18:32:04 ok Apr 24 18:32:13 I am not expecting glibc to rebuilt if unrelated distro feature is changed Apr 24 18:32:26 so something is amiss here Apr 24 18:33:15 although if something that glibc depends on is changed due to DISTRO_FEATURE change then it will rebuild of course but that will not be direct dependency Apr 24 18:34:31 I got some strange diffsigs during testing: https://bpaste.net/show/1bb0c0e0bd5b . I have no idea why these values toggles like this. Apr 24 18:36:51 Anyhow, I've ditched setting the DISTRO_FEATURE, I had to move on to the next item Apr 24 18:37:16 Which is failure to run bash here commands under a fakeroot task Apr 24 18:46:30 khem: host gcc8 is a barrel of laughs Apr 24 18:46:58 rburton: how come? Apr 24 18:47:05 khem: python2 crashes on startup (got a fix from fedora28) and the kernel won't build (fixed upstream just need to extract the patches) Apr 24 18:59:28 quite: rburton hello Apr 24 18:59:31 oops Apr 24 19:04:05 sveinse: what are you changing ? it seems machine Apr 24 19:04:50 khem: I'm not doing that deliberately, but I suspect some funky stuff in the machine layer Apr 24 19:05:00 rburton: yeah was expecting some of it, surprisingly OE's python2 is working out to be ok with gcc8 Apr 24 19:05:36 sveinse: yeah some distro and machine features are changing might be that default tune is changing Apr 24 19:05:51 is this same build config done twice ? Apr 24 19:05:55 or are you changing anything Apr 24 19:06:13 this is me adding DISTRO_FEATURES_append=' foobar' Apr 24 19:06:26 otherwise equal and sequentially Apr 24 19:15:24 sveinse: you tune value is changing tune-armv5te to tune-cortexa9thf-neon Apr 24 19:15:29 what is foobar Apr 24 19:16:18 khem: nothing, just a placeholder proving that appending a value to distro features causes change to my system somehow Apr 24 19:19:03 hi ... is there an easy way to disable the license check for a packagegroup I added GPLv3 to the INCOMPATIBLE_LICENSE var but for debugging I want to add packagegroup-core-debug? Apr 24 19:20:20 New news from stackoverflow: Cannot edit or delete files with WebDav (Apache2) Apr 24 19:34:14 How do I add xml.etree.ElementTree as a RPENEDS in my .bb file? Is it just python-xml? Apr 24 21:41:26 Oiy, I'm really at a loss here.. I am trying to install pynacl, and it's causing so much grief and holding up a whole project. It finally compiles and installs and seems to have no error until runtime, when ImportError: No module named 'nacl._sodium' . Anybody here familiar enough to actually sit down and assist with this? I'm at bugger all for a loss... Apr 24 22:49:51 aarcane: that suggests that file didn't get packaged, so look at what your pynacl packages actually contain Apr 24 22:50:07 * rburton ->bed Apr 24 23:24:23 rburton, I have it open in devtool workspace right now, and it's not even being built as far as I can tell... like pynacl has a weird hybrid python+binary build system and it's sooo confusing. **** ENDING LOGGING AT Wed Apr 25 03:00:05 2018