i'm starting to figure out what is going on. the short story, I think, is that all the other boards are using soft float and it works fine because they are using a separate and newer branch of uboot that fixed a build-system related bug [1].
in other words: it's totally fine (desirable, actually) to build u-boot with -msoft-float and then boot a kernel and userspace with hard float. i think the issue I was running into with the imx6 dongle uboot has to do with something being erroneously pulled in from the hosts libgcc (which was compiled with hard float support) [1]. the build fails because it refuses to mix the soft-float u-boot binaries with the the hard-float stuff erroneously pulled from libgcc. i still don't know what the exact fix is because nobody mentions a particular commit but i am on the trail. i've done a little digging already but didn't get very far (USE_PRIVATE_LIBGCC doesn't appear to be used anywhere in any recipes so that couldn't be the fix). part of me wonders how hard it would be to port the board from the older uboot-imx6dongle to the official uboot-imx because this would probably solve the problem. more details to follow, i hope. --christian 1. http://u-boot.10912.n7.nabble.com/U-Boot-RFC-PATCH-ARMV7-Patch-to-fix-hard-float-build-issues-td22107.html On Fri, Aug 30, 2013 at 3:23 AM, Christian Betz <christian.b...@gmail.com> wrote: > On Thu, Aug 29, 2013 at 4:14 PM, Otavio Salvador > <ota...@ossystems.com.br> wrote: >> On Tue, Aug 27, 2013 at 10:25 PM, Christian Betz >> <christian.b...@gmail.com> wrote: >>> I found the answer and I was able to build u-boot (and >>> core-image-base) successfully after modifying >>> arch/arm/cpu/armv7/config.mk as described at the link below [1] >> >> This is probably something which has been fixed with a specific patch >> in mainline so the better way would be to 'export' the patch and use >> it here. >> >> Can you try git blame and check it? > > I tried, but I'm having a really hard time figuring out how the fpu > selection works for other boards or finding any code related to > hard-float that I can blame. since config.mk has the -msoft-float in > u-boot master and there no patches I can find that would change it. > > in other words, mainline u-boot appears to still have the following > line in arch/arm/cpu/armv7/config.mk: > > PLATFORM_RELFLAGS += -fno-common -ffixed-r8 -msoft-float > > and I can't find any recipes/patches that are changing it. so i'm > scratching my head wondering how the heck u-boot builds with hard > float for all the other boards. > > I'm hoping someone else has some pointers. I would like to do this the > "right" way but i'm currently lost. Or maybe i've just been awake too > long ;) > > --christian > > > >> >> -- >> Otavio Salvador O.S. Systems >> http://www.ossystems.com.br http://code.ossystems.com.br >> Mobile: +55 (53) 9981-7854 Mobile: +1 (347) 903-9750 > > > > -- > "the new garbage collector will be an arena-based, quad-color > incremental, generational, non-copying, high-speed, cache-optimized > garbage collector" -- LuaJIT Roadmap -- "the new garbage collector will be an arena-based, quad-color incremental, generational, non-copying, high-speed, cache-optimized garbage collector" -- LuaJIT Roadmap _______________________________________________ meta-freescale mailing list meta-freescale@yoctoproject.org https://lists.yoctoproject.org/listinfo/meta-freescale