**** BEGIN LOGGING AT Sun Jun 17 02:59:59 2012 Jun 17 18:38:09 khem, hi Jun 17 18:41:12 GNUtoo-hplaptop, have you also registeres GNUtoo-shower ? Jun 17 18:41:24 lol no why? Jun 17 18:41:41 I don't have a rugged laptop that goes under the watter Jun 17 18:41:48 GNUtoo-hplaptop, just kidding because i saw at least 4 nicks from you ;) Jun 17 18:42:49 yes that is a laptop that I recuperated thanks to flashrom like that: http://www.flashrom.org/File:Serduino_laptop.jpeg Jun 17 18:43:08 I also wrote the code for the arduino Jun 17 18:44:18 GNUtoo-hplaptop: heelo Jun 17 18:44:32 * khem is busy watching Netherlands Vs. Portugal Jun 17 18:44:56 khem, hi, do you know what hash I should backport to get unaligned loads on armv7 with a 2.6.35 kernel? Jun 17 18:45:01 ok Jun 17 18:45:25 hmm did you try the option I suggested last time ? Jun 17 18:45:27 I tried 8428e84d42179c2a00f5f6450866e70d802d1d05 and it still had the same problem Jun 17 18:45:34 yes of course, I know how to fix Jun 17 18:45:44 however the problem is making it compatible with various compilers Jun 17 18:46:03 because I suspect that buglabs don't use oe-core for production images Jun 17 18:46:23 so If I add -mno-unaligned-access it would break their builds Jun 17 18:46:36 however it would make it work with gcc4.7 Jun 17 18:46:45 I tried passing it as argument, that failed too Jun 17 18:47:00 so or I patch the kernel for -mno-unaligned-access in a recipe Jun 17 18:47:10 or I fix it properly abckporting the right patches Jun 17 18:47:15 I prefer the second option Jun 17 18:47:29 s/a recipe/the recipe/ Jun 17 18:47:48 so you have two different compilers compiling same kernel ? Jun 17 18:47:50 right Jun 17 18:48:19 I suppose so, personally I use oe-core but buglabs don't use oe-core, they use their fork of oe-classic I think Jun 17 18:48:27 ok Jun 17 18:48:29 so they would refuse my patch if I add that Jun 17 18:48:49 because it would break the build because the flag woulnd't work on older gcc.... Jun 17 18:48:52 then you can add the patch so that it enquires gcc before it enables the option Jun 17 18:49:17 no need, I just add the patch to meta-bug which is the oe-core version Jun 17 18:49:42 I'll do that Jun 17 18:49:58 but I wanted to get unaligned loads so it would become way faster tough.... Jun 17 18:50:20 thanks a lot Jun 17 18:50:36 hmm, you need to backport quite a bit Jun 17 18:50:48 if you want to make older kernel do unaligned loads Jun 17 18:51:04 ah ok Jun 17 18:51:14 and still there might be issue or two lurking Jun 17 18:51:26 so I'll just -mno-unaligned-access then Jun 17 18:51:28 thanks a lot Jun 17 18:51:40 yes for older kenrels I would just do that Jun 17 18:51:45 unless you have perf issues Jun 17 18:51:57 and would like to consider unaligned loads Jun 17 18:52:07 it could be a bigger project Jun 17 18:52:11 perf means performances I guess Jun 17 18:52:15 ok Jun 17 18:52:15 yes Jun 17 18:52:22 because there is the perf uttility too Jun 17 18:52:30 heh yes Jun 17 18:56:23 KBUILD_CFLAGS += $(call cc-option, -fno-unaligned-access) Jun 17 18:56:31 you could do something like that in makefile Jun 17 18:56:40 that will be valid in both compilers Jun 17 18:56:53 netherland scores !! Jun 17 19:01:26 GNUtoo-hplaptop: something like this will help you Jun 17 19:01:28 https://groups.google.com/forum/?fromgroups#!topic/meta-fsl-arm/XlQKeJBp_uI Jun 17 19:01:45 khem, yes possible too Jun 17 19:02:00 cc-option is the one checking the option? Jun 17 19:02:10 like in Documentation/Kconfig/ ? Jun 17 19:02:14 or something like that? Jun 17 19:03:20 I'll look in the doc Jun 17 19:04:38 yes it it Jun 17 19:05:41 thanks a lot!!!! Jun 17 19:07:46 this is the committed patch https://github.com/Freescale/meta-fsl-arm/blob/master/recipes-kernel/linux/linux-imx-2.6.35.3/no-unaligned-access.patch Jun 17 19:08:36 thanks a lot Jun 17 23:47:34 hey. is there any recipe outthere that builds stunnel? **** ENDING LOGGING AT Mon Jun 18 02:59:58 2012