**** BEGIN LOGGING AT Thu Sep 07 03:00:01 2017 Sep 07 03:00:35 kergoth: FWIW, I worked around by conditionally (if IMAGE_FSTYPE contains myiamge) adding it to do_image[prefuncs] in an anonymous python function. Not pretty, but I think it should hold for now :-) Sep 07 03:00:54 kergoth: thanks for your help! Sep 07 03:04:50 ah, indeed, nicely done Sep 07 03:04:52 np Sep 07 08:13:30 Hello all! Is it possible configure two u-boot for one machine in yocto? regards Sep 07 09:57:16 Hi everyone, How to bitbake a recipe as if it was the first time , I mean with files,packages and package-split working directory entirely rebuild from zero ? Sep 07 10:43:40 hmm Sep 07 10:43:55 how would yocto decide the host arch is "armv7l" rather than arm? Sep 07 10:44:34 I'm missing build deps on the host so I was trying a slightly weird trick to run yocto inside flatpak Sep 07 10:44:50 could that have done a "setarch" and upset the whole thing? Sep 07 10:45:02 yeah maybe. what is actually failing where? Sep 07 10:47:39 Loading cache: 100% |##########################################################################################################################################################| Time: 0:00:00 Sep 07 10:47:41 Loaded 3 entries from dependency cache. Sep 07 10:47:43 ERROR: /home/ramcq/freedesktop-sdk-base//meta-freedesktop/recipes-support/re2c/re2c_0.13.5.bb: Unable to determine endianness for architecture 'armv7l' | ETA: --:--:-- Sep 07 10:47:54 ERROR: /home/ramcq/freedesktop-sdk-base//meta-freedesktop/recipes-support/re2c/re2c_0.13.5.bb: Please add your architecture to siteinfo.bbclass Sep 07 10:48:02 it just spits out a bunch of those at the very start and dies Sep 07 10:48:48 I copied the siteinfo.bbclass line for arm then it got past that, then when it came to actually run ./configure on the first module, it was looking for armv7l-linux and died again Sep 07 10:49:45 right Sep 07 10:51:05 so thats your machine conf Sep 07 10:52:12 oh unless its failing for native stuff in which case does flatpak mangle uname? Sep 07 10:52:29 BUILD_ARCH := "${@os.uname()[4]}" Sep 07 10:53:17 Linux endless6 4.4.0-93-generic #116-Ubuntu SMP Fri Aug 11 16:31:48 UTC 2017 armv7l armv7l armv7l GNU/Linux Sep 07 10:54:00 compare that with outside flatpak? Sep 07 10:54:20 that is outside! Sep 07 10:54:23 ah Sep 07 10:54:40 set BUILD_ARCH=arm in your local.conf... Sep 07 10:54:46 same inside Sep 07 10:55:38 surprised this is breaking for you and not alex. new hardware/kernel/something? Sep 07 10:56:00 we're basically expecting uname to say arm/armeb/aarch64/aarch64_be Sep 07 10:56:57 tristan did the arm builds it turns out Sep 07 10:57:15 and his timezone is such that I never know when he's awake so essentially can never find him on IRC :P Sep 07 10:58:41 i'd hope that setting build_arch explicitly would solve this Sep 07 10:59:53 yeah, parsing again, lets see what happens Sep 07 11:00:24 all this just to find out "quickly" if config.gcc is generated because I most definitely don't have one Sep 07 11:00:32 maybe I should've spent more time grepping in gcc Sep 07 11:01:42 BUILD_SYS = "arm-linux" Sep 07 11:01:44 thats better Sep 07 11:03:40 when you can compare your uname with tristan and see if that is different, or if he has any other mojo Sep 07 11:58:26 hi, i have some issues getting gpsd to work. I can't get gpsd to work. gpsd is not finding my gpsd device Sep 07 11:59:41 hmwel: you'll have better luck with the gpsd support channels Sep 07 11:59:47 k Sep 07 13:35:03 im having some trouble building the extensible-sdk for a given machine where kernel,dtb,initramfs is bundled in a fitImage. The signatured computed when executing do_populate_sdk_ext for the kernel is not matching the signature that is locked in SIGGEN_LOCKEDSIGS. Sep 07 13:39:26 Is there a way to tell the bitbake fetcher to extract a zip archive that has a non standard file ending (.nupkg)? Adding unpack=1 doesn't seem to help Sep 07 13:42:34 I think you'd need to edit unpack() in bitbake/lib/bb/fetch2/__init__.py Sep 07 13:46:50 joshuagl: Ok I will have a look, thanks Sep 07 13:47:21 bachp: please send the patch upstream if you write one Sep 07 13:47:41 I will Sep 07 13:54:52 joshuagl: The question is what to do. I could just add nupkg to the list there, or add a generic mechanism that allows to configure extraction via some param in the url Sep 07 13:55:14 ;extract=zip or ;extract=tar.gz for example Sep 07 14:00:19 bachp: could always propose the generic mechanism, so long as you don't fear rejection :-) Sep 07 14:32:44 bitbake (current master = 1ca2eec459424892391f060442ef38cf28d6a54a) no longer starts up for me (ERROR: Unable to start bitbake server). It times out eventually, dumping bitbake-cookerdaemon.log, which contains an error about my bblayers.conf version. Sep 07 14:33:13 RP: should an error parsing the base configuration cause a failure to connect to the server? Sep 07 14:34:40 bachp: you can set downloadfilename parameter to rename it to .zip Sep 07 14:35:31 JaMa: So the extraction logic comes after the rename. I will try that thanks Sep 07 15:31:05 pohly: I saw the bug, it really means "the server failed to start" Sep 07 15:31:41 pohly: discuss here or on the bug? Sep 07 15:31:53 Bug is probably better, now that I have filed it. Sep 07 15:41:56 hello everybody. I'm trying to generate an yocto image without busybox because i want to use toybox instead. Does yocto allow to build image without busybox ? Sep 07 15:47:22 tcolin: As there is a recipe for toybox (http://cgit.openembedded.org/meta-openembedded/tree/meta-oe/recipes-core/toybox/toybox_0.7.3.bb?h=master) I assume somebody has used it. But I don't have any experience what it would take to get rid of busybox entirely. Sep 07 15:48:09 i have try to play with PREFERRED_PROVIDER_virtual/base-utils = "toybox" VIRTUAL-RUNTIME_base-utils= "toybox" but i still have busybox installed :/ Sep 07 15:56:33 tcolin: Did you grep for busybox in meta it might give you some hints where it gets pulled in Sep 07 15:56:59 In fact PREFERRED_PROVIDER_virtual/base-utils = "toybox" VIRTUAL-RUNTIME_base-utils= "toybox" did the trick. So my issue is solved Sep 07 15:58:54 anyone have issues cloning large repos? I'm seeing something like the following: Sep 07 15:58:55 fatal: read error: Connection reset by peer77 GiB | 4.52 MiB/s fatal: early EOF fatal: index-pack failed Sep 07 15:59:10 I've got a workaround, but it's still weird... wondering if it's been solved by someone Sep 07 15:59:45 mattsm: from git.yp.org or another server? Sep 07 15:59:53 github in this case Sep 07 16:00:04 git clone -vvv git@github.com:raspberrypi/firmware.git Sep 07 16:00:35 mattsm: sounds like either the remote server or something between you and it Sep 07 16:00:38 using http can work, but you need to update your git config to increase the postbuffer Sep 07 16:01:39 from what I've read it's possibly related to doing a gc during fetch Sep 07 16:02:03 RP: the only time it seems to fail is on large git repos so... Sep 07 16:13:50 is it possible to define a bitbake variable from a shell funtion? Sep 07 16:13:57 or a task Sep 07 16:14:38 Is USB pass through enabled in qemu as compiled by OE? I'm getting an error for qemuparams="-device usb-host,vendorid=046d,productid=c52b": Sep 07 16:14:38 qemu-system-x86_64: -device usb-host,vendorid=046d,productid=c52b: 'usb-host' is not a valid device model name Sep 07 16:16:10 i.e. I have some function calculate_stuff () which should set some variables which I need later in the recipe. Sep 07 16:16:50 I can do this with a pure python function which returns a value, but is it possible with a shell function? Sep 07 16:17:10 no Sep 07 16:17:25 shell functions are executed by a shell, so anything you set disappears when the shell exits Sep 07 16:17:31 no, a shell function can't define a bitbake variable. if your'e calling it from pyhton, you could have it echo something to stdout and get it that way, but you should probably take a step back and rethink how you're doing this and what you want to accomplish Sep 07 16:18:27 rburton, kergoth, thanks Sep 07 16:18:50 Yep, it's disabled: "libusb libusb (for usb passthrough)" and --disable-libusb is used in OE. Sep 07 16:21:56 pohly: its a packageconfig Sep 07 16:26:51 rburton: yes, rebuilding already. Sep 07 17:00:47 say I have two totally independent projects which both have a recipe that's derived from one git repo. can I devtool-modify this recipe from both projects but have it point to the same local git clone so that I can do my work locally in one area? Sep 07 17:36:18 rburton: is the uvesa issue only happening on lsb? Sep 07 17:37:23 rburton: nvm just saw it on another one Sep 07 19:21:43 RP, I think we need to re-queue https://autobuilder.yocto.io/builders/nightly/builds/523 as yocto-2.2.2.rc2 because publish_artifacts was false. Sep 07 19:30:30 Started at https://autobuilder.yocto.io/builders/nightly/builds/525 Sep 07 19:41:21 hey folks - what actually decides on TARGET_SYS ? Sep 07 19:42:52 I'd expect armv7ahf to get me arm-linux-gnueabihf Sep 07 19:43:25 but I get arm-unknown-linux-gnueabi passed to gcc as the machine spec, with a bunch of tunes to put callconvention=hard etc Sep 07 19:43:42 this seems more likely to misdirect everyone Sep 07 19:43:53 as arm-unknown-linux-gnueabi means armel to everyone Sep 07 19:45:17 halstead_beetlej, thanks Sep 07 19:46:47 halstead, I didn't say beetlej 3x Sep 07 19:46:49 armpit, All the hashes are the same so it should be a fast build. Sep 07 19:46:59 :) Sep 07 19:49:37 armpit, bully4u and I were joking that if I get pinged 3 times I always appear. So I changed my nickname for a laugh. :) Sep 07 20:52:43 someone has a working opkg with pyro? Sep 07 21:07:48 halstead: thanks, I clearly missed a setting when triggering it Sep 07 21:08:09 RP, And I missed the send QA e-mail field. So I'll send that at the end. Sep 07 21:42:56 FabKna: works out of the box and is tested constantly Sep 07 21:50:14 rburton: Thank you, I have selected the openwrt omap sources but have problems since pyro with it. Sep 07 21:52:12 I get error: Nothing provides libc needed by... Sep 07 21:54:28 anyone else seeing random QA issues like this one: Sep 07 21:54:30 ERROR: glibc-locale-2.26-r0 do_package_qa: QA Issue: /usr/bin/localedef contained in package localedef requires libc.so.6(GLIBC_2.15), but no providers found in RDEPENDS_localedef? [file-rdeps] Sep 07 21:54:34 ERROR: glibc-locale-2.26-r0 do_package_qa: QA Issue: /usr/bin/localedef contained in package localedef requires libc.so.6, but no providers found in RDEPENDS_localedef? [file-rdeps] Sep 07 22:12:11 JamA, I've not seen that Sep 07 22:29:46 JaMa, maybe build host related? Sep 07 22:32:17 JaMa: I was getting those a few days ago, but they went away after another pull on master Sep 07 22:32:42 JaMa: but if they're random like you say maybe I'm just getting lucky =/ Sep 07 23:41:28 halstead, ping Sep 07 23:42:12 * armpit beetle juice, beetle juice beetle juice Sep 07 23:43:28 Hi armpit. Sep 07 23:44:24 halstead, I just say an issue with the morty build. does this sound like a host issue ? rsync: chgrp "/srv/autobuilder/autobuilder.yoctoproject.org/pub/releases/yocto-2.2.2.rc2/.poky-60402978fe3648bf560b3386c6e9dd661cdf2083.tar.bz2.md5sum.Vy1vM1" Sep 07 23:45:29 or a burp Sep 07 23:46:05 armpit, I think the message was cut off. Which file/URL are you looking at? Sep 07 23:46:29 https://autobuilder.yocto.io/builders/nightly-x32/builds/456/steps/Publishing%20Layer%20Tarballs/logs/stdio Sep 07 23:46:45 Thanks Sep 07 23:47:00 I logged a bug in the mean time Sep 07 23:48:42 * armpit looks for more beer Sep 07 23:49:06 armpit, It looks like nightly-x32/build doesn't contain a poky directory on that server and others. So cd poky will fail... Sep 07 23:49:09 Hrm. Sep 07 23:50:09 armpit, Failed exactly the same way at https://autobuilder.yocto.io/builders/nightly-x32/builds/439/steps/Publishing%20Layer%20Tarballs/logs/stdio Sep 07 23:52:44 yeah, saw that Sep 07 23:53:43 not sure how to classify the failure, AB, infrastructure or core Sep 08 00:02:09 armpit, So the cd poky doesn't actually make it fail. I found the issue and updated the bug. Thank you for the ping and helping me track down the problem. Sep 08 00:15:02 k **** ENDING LOGGING AT Fri Sep 08 02:59:59 2017