**** BEGIN LOGGING AT Thu Dec 13 03:00:02 2018 Dec 13 09:27:03 does anyone know how can I forcefully exclude a recipe from the SDK target sysroot? Dec 13 09:27:39 why would I even want to have util-linux and/or busybox/toybox installed there anyway? Dec 13 09:27:59 as far as I can tell those things are not needed for cross-compilation Dec 13 09:28:36 but because of update-alternatives, etc. conflicts between util-linux and toybox symlinks are preventing SDK generation Dec 13 09:29:09 this is with poky master: https://pastebin.com/ELT88XWa Dec 13 15:02:49 Hello. I am having some issues with multiconfig and native recipes. When I run `bitbake multiconfig:*:erlang-native ` bitbake runs all the configs on parallel. Dec 13 15:03:23 and since all the builds are run on the same folder the build crashes. Is this expected behavior? Dec 13 15:54:32 https://lwn.net/Articles/774734/ <-- "The x32 subarchitecture may be removed" unless people speak up. Dec 13 15:54:55 * vmeson reads https://lkml.org/lkml/2018/12/10/1151 and points RP and others to it. Dec 13 15:55:12 alt.x32.die.die.die Dec 13 16:13:16 vmeson: I've replied to the thread Dec 13 16:13:22 vmeson: thanks for the headsup Dec 13 16:15:14 RP: cool and thanks. The bit about 10% faster kernel builds was interesting. Supercomputing folks would likely also give up their first born for a 10% perf boost but I don't know if they use x32 in that domain. Dec 13 16:16:00 vmeson: I think there are genuine memory wins for it and it is useful in niche use cases. I suspect if there were commonly available desktop support for it, it would be popular Dec 13 16:25:12 khem, have you maybe still kernels 3.1x around to build? Back to gcc7 it eems I need an older binutils. With 2.31 I get Dec 13 16:25:53 mips/kernel/r4k_fpu.S:83: Error: opcode not supported on this processor: mips3 (mips3) `sdc1 $f 0,272+0($4)' Dec 13 16:26:30 this issue is around since 2015 Dec 13 16:30:34 OLDEST_KERNEL = "3.2.0" no no, we need a _mips override Dec 13 16:30:49 I'l test for fun Dec 13 16:31:30 3.13.5 bcm-tainted here Dec 13 17:05:12 khem, seems binutils 2.31 is the cuplrit... rechecking Dec 13 17:25:45 vmeson: how does one specify an ilp32 build (x86 or aarch)? Dec 13 17:26:08 ilp32 so people do use it hah Dec 13 17:27:03 tlwoerner: I guess: https://www.yoctoproject.org/docs/current/mega-manual/mega-manual.html#using-x32-psabi Dec 13 17:27:07 I haven't done it recently. Dec 13 17:27:07 use MACHINE=qemux86-64 and set DEFAULTTUNE = "core2-64-x32" Dec 13 17:27:22 right Dec 13 17:27:24 okay Dec 13 17:27:40 is there ilp32 support for arm? Dec 13 17:27:55 kernel folks are discussing about dropping x32 Dec 13 17:28:03 any arm ilp32 fans here ? Dec 13 17:28:16 do we still need: baselib = "${@d.getVar('BASE_LIB_tune-' + (d.getVar('DEFAULTTUNE', True) or 'INVALID'), True) or 'lib'}" Dec 13 17:28:20 as per the docs? Dec 13 17:30:10 Dont think so Dec 13 17:30:22 I am not sure how one would do it for arm though Dec 13 17:30:39 Dont remember if we accepted patches for it in core Dec 13 17:46:28 khem, pls have a look at the python3 powerpc-linux-musl patch. I took the short path, I know... Dec 13 17:47:18 khem, about kernels 3.1x, I'll try binutils 2.30, I bet it's that Dec 13 17:47:50 bbl Dec 13 20:22:12 ant_home: ideally we get something upstream to stop patching one by one Dec 13 20:22:34 ant_home: upstream are nice, and i've even gone through the CLA pain so if we have a patch i don't mind filing the patch Dec 13 20:22:59 at least start with filing a bug. i'm not sure what that bit of the configure is trying to do... Dec 13 20:36:41 rburton, the configure is gnu/glibc-biased Dec 13 20:36:52 one patch added -spe for ppc Dec 13 20:37:05 otherwise it defaults to -gnu Dec 13 20:37:43 we could add a big __GLIBC__ guard before and after and add another case Dec 13 20:38:12 but what about openwrt folks? they are on uclibc afaik Dec 13 20:39:13 I really hate these spaghetti if-else, just got another one with kodi/mips/musl recently Dec 13 20:40:12 you are right, we open a bug and show our OE-specific solution Dec 13 21:48:29 anyone going to LCA in January? Dec 13 22:04:14 You are :) Dec 13 22:06:06 I just noticed it is in NZ Dec 13 22:15:15 bluelightning: i wish! (LCA) Dec 13 22:15:28 bluelightning: behanw and ka6sox are going Dec 13 22:15:40 nice Dec 13 22:15:55 Crofton: well, only if management approves it... can't afford to pay for it myself Dec 13 22:15:56 * tlwoerner watches the videos afterwards Dec 13 22:16:47 ant_home: http://www.literateprogramming.com/ifdefs.pdf "#ifdef Considered Harmful" Dec 13 22:16:51 NZ is a small place, can't you walk? Dec 13 22:16:57 lol Dec 13 22:17:12 or ride an elf Dec 13 22:18:07 or hobbit Dec 13 22:59:53 tlwoerner, that towers of code...even INTERCAL has better syntax ;) Dec 13 23:04:51 Crofton: it's a short 218 hour walk ;) Dec 13 23:17:14 tlwoerner: giant eagle, surely Dec 13 23:17:49 lol Dec 13 23:19:55 khem, same kernel builds with glibc...maybe binutils-cross or gcc-cross is miscompiled for mipsel? Dec 13 23:20:06 (when using musl) Dec 13 23:26:14 bluelightning, but no swimming? Dec 13 23:26:26 Crofton: yes, a ferry in the middle of that Dec 13 23:26:41 well, ferry or you swim :D Dec 13 23:26:41 ah good, no wetsuit needed Dec 13 23:33:11 khem, it's the first time I see kernel build failing switching libc. Poor mipsel... Dec 14 02:53:24 ant_home: kernel should not matter if userspace uses glibc or musl Dec 14 02:53:37 lets see what error do you see **** ENDING LOGGING AT Fri Dec 14 03:00:01 2018