**** BEGIN LOGGING AT Tue Jul 07 02:59:58 2015 Jul 07 08:16:45 guys, how to mix a recipe to have part of it building a host tool and part of it building something for the MACHINE? Jul 07 08:17:32 I know that I suppose to use BBCLASSEXTEND =+ "native", but what else should I do? Jul 07 08:20:20 ftonello: depends on the recipe Jul 07 08:20:37 ftonello: some use bbclassexted and depend on PN-native, then tell the cross build to use tooling in the sysroot Jul 07 08:21:03 I am starting a recipe from scratch. Jul 07 08:21:10 s/recipe/upstream/ Jul 07 08:21:24 autotools is slowly moving towards an idiom of have CC_FOR_BUILD etc which are used to compile host tools in cross builds, which will just work out of the box Jul 07 08:22:08 The thing is that this recipe is to build a program for the TARGET, but it requires a tool to generate Makefiles which is included in its source code. Jul 07 08:22:30 So I basically created a task for building this tool, but I just need to build locally, I don't have to export to any sysroot at the moment. Jul 07 08:28:35 joshuagl: i think "[OE-core] [PATCH] cross-localedef-native_2.20.bb: fix for gcc5" is for fido Jul 07 08:29:53 rburton: ooh, thanks - will look at that during my upstream time later Jul 07 08:50:08 morning guys Jul 07 08:50:36 is there a way to build a kernel only by external toolchain? Jul 07 08:50:58 I know about EXTERNAL_TOOLCHAIN variable, but it is used for the whole project Jul 07 09:04:19 good morning Jul 07 09:25:56 mckoan: good morning Jul 07 09:32:21 hello all, im trying to use the lttng and when i build in dizzy i'm having a build error in lttng-modules Jul 07 09:33:17 3.14 kernel, 1.7.2. Besides the inhibit strip and (or using the extra features tools-profile) debug split, am i missing something? Jul 07 10:25:28 http://www.openembedded.org/wiki/Adding_a_secondary_toolchain very nice :) Jul 07 10:29:49 i have better luck with 3.17 building the lttng modules in Dizzy 12.0.2. Jul 07 13:30:00 i got do_compile error about ltrace for arm64: http://paste.ubuntu.com/11835838/ Jul 07 13:40:40 RP: rburton: AB appears idle, OK to queue a build? Jul 07 13:43:26 guys, what does mean virtual/TOOLCHAIN-EXAMPLE in this document http://www.openembedded.org/wiki/Adding_a_secondary_toolchain ? Jul 07 14:08:21 Hi all, how do I set autologin for a user in image recipe ? I tried inittab + login but got duplicated ID Jul 07 14:55:44 YPTM: Ready-Access Number: 8007302996/9139049836 Access Code: 2705751 Jul 07 14:57:12 YPTM: Stephen Joined Jul 07 15:00:16 YPTM: Michael joined. Jul 07 15:01:07 YPTM: Randy MacLeod joined. Jul 07 15:03:07 YPTM: belen joined the called and has one open about the state of Toaster Jul 07 15:03:58 joined yocto call Jul 07 15:04:55 YPTM: Bruce Joined Jul 07 15:08:42 YPTM: Alex V here! Jul 07 15:08:44 YPTM ross joined late Jul 07 15:09:03 (sorry) Jul 07 15:09:06 sory for being late :D Jul 07 15:09:11 sorry* Jul 07 15:14:54 the recent removal of libmodule-build-perl from cpan-build.bbclass cripples the class Jul 07 15:15:18 instead, libmodule-build-perl should be built from CPAN and brought into oe-core Jul 07 15:17:21 YPTM is done! Jul 07 15:46:24 halstead: ping Jul 07 15:48:56 ulf`, pong Jul 07 16:00:08 halstead: nevermind :) Jul 07 16:00:22 ulf`, okay! Jul 07 16:01:20 halstead: I wanted to ask if you can launch an eggdrop bot for #automotive to enable logging, but the community there has already found one Jul 07 16:01:44 Excellent. :) Jul 07 17:55:51 Is there a way to change the GCC version used by bitbake? I'm getting a failure in gcc-cross-initial-4.8 and I'm reading that it's related to gcc-4.8 - can I switch to 4.7? Jul 07 17:56:58 oe-core doesn't retain old versions like that except for particular cases, generally. currently there's 4.8, 4.9, and 5.1. 4.6 is still available in meta-oe's toolchain-layer, though Jul 07 17:57:35 GCCVERSION = "4.6%" or whatever in local.conf should be sufficient if that version is available in your configuration Jul 07 17:59:00 so when it says gcc-cross-initial-4.8 - is it actually building that compiler? Jul 07 17:59:10 I don't understand the question Jul 07 17:59:11 it's not lying Jul 07 18:00:58 I'm still very new to yocto, and bitbake Jul 07 18:01:12 it is building the gcc compiler that targets my hardware platform, is what I'm asking Jul 07 18:01:21 that is what that step is doing Jul 07 18:02:11 when I do a build, it's building the cross compiler (if it hasn't been built already) and then using it to compile my kernel and all my packages Jul 07 18:02:22 and the gcc-cross-initial-4.8 recipe is the recipe for the gcc cross compiler Jul 07 18:03:00 building a cross-compiler toolchain is a multi-phase process. the initial toolchain is used to build the C library, which is then used in the non-initial gcc-cross build Jul 07 18:03:14 that's one of the recipes for the gcc cross compiler, yes Jul 07 18:03:45 Ok, roger. That's fine. I have done cross compiler builds for arm microcontrollers, so I'm familiar with the "bootstrap" compiler that you use to make your C libraries. Jul 07 18:03:57 This build system is just a new vocabulary for me. Jul 07 18:04:12 So needless to say, this is a failure very early in the overall build process. Jul 07 18:05:11 ryansturmer: if your cross-inital is failing, the solution might be to switch host compiler to something else Jul 07 18:05:34 ok, so that's what I thought, and so that gets back to my question Jul 07 18:05:48 well, ok hmm hang on a sec Jul 07 18:06:04 I've got two things going on here, the cross-initial-4.8 is obviously trying to build gcc-4.8 Jul 07 18:06:11 what gcc version is on your host? Jul 07 18:06:14 but my host also uses 4.8 as its default compiler Jul 07 18:06:17 so I got confused Jul 07 18:06:27 I switched my host compiler to 4.7 Jul 07 18:06:47 and I still fail the initial cross, but it doesn't look like the same error. Jul 07 18:07:04 I was doing this on archlinux, but I switched to xubuntu to play nice with the yocto instructions Jul 07 18:07:33 I had issues in the past building yocto 4.8 with debian 4.8 host Jul 07 18:07:47 but it always built fine with host 4.7 Jul 07 18:08:16 Well, so, I'm new to ubuntu as well, having always used arch, so maybe you can help me make sure I'm using the compiler version I think I am Jul 07 18:08:31 gcc --version Jul 07 18:08:45 ok that gets me 4.7 Jul 07 18:09:04 I used update-alternatives to configure multiple different compiler versions Jul 07 18:09:16 yeah, mine is pinned down to 4.7.4 Jul 07 18:09:23 I have 4.7 and 4.8 (the default for this version of xubuntu) installed Jul 07 18:09:33 and have set 4.7 as the default using update-alternatives Jul 07 18:09:52 4.7.3 is what it lists Jul 07 18:10:34 I am currently building yocto 4.8.2, 4.8.4 and 4.9.something successfully with the 4.7.4 Jul 07 18:14:45 I'm running a clean build again to see what the error is. Jul 07 18:21:38 man it really seems to do like a lot of tasks before it gets to the one that bombs Jul 07 18:21:51 I guess a lot of these look like they're dependencies of the cross compiler Jul 07 18:21:56 (flex, bison, gmp, etc) Jul 07 20:00:21 ha! I think this is actually a memory configuration thing Jul 07 20:00:32 it's not failing in the same place every time, and always on the big complicated builds Jul 07 20:00:53 I'm doing this in a VM and I think there's not quite enough memory assigned Jul 07 21:25:44 kergoth: i have a patch that sort-of appears to almost work for that long-standing "bitbake creates ${B} for you" bug Jul 07 21:26:11 running a world oe-core now to see how much breaks (so far, boost needed fixing) Jul 07 21:27:29 rburton: does it mean that bitbake wont create anymore Jul 07 21:28:20 not by default Jul 07 21:28:26 configure creates it Jul 07 21:28:33 * rburton just found the comment on boost.inc and giggled Jul 07 22:00:56 QA Issue: blablah relocations in .text Jul 07 22:01:00 anything to worry about? Jul 07 22:01:27 I get a couple of those in core, but my build seems to pass! Jul 07 22:10:48 ryansturmer: http://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#qa-issue-textrel Jul 07 22:33:52 is it weird to see that in a core package like systemd? Jul 07 22:34:08 I would figure the compiler options for those packages to be pretty well tuned out of the box Jul 07 22:34:32 I would have thought the same yes Jul 07 22:34:48 presumably something we've just missed somehow Jul 07 22:42:59 well, this is the intel edison build package Jul 07 22:43:14 so it could be pulling a different branch or an older tag than what is the current "latest and greatest" Jul 07 23:02:36 Hmm, need to brush up on the capabilities of the sstate locking stuff Jul 08 02:36:48 ryansturmer: its a performance issue if you have texrels Jul 08 02:37:05 and XIP may not work **** ENDING LOGGING AT Wed Jul 08 02:59:58 2015