Re: [PATCH 0/5][v4] Make multi-machine toolchains to co-exist (Alternate approach)

[Martin Jansa <martin.jansa@gmail.com>]

On Tue, Aug 03, 2010 at 10:20:42AM -0700, Khem Raj wrote:
> On Tue, Aug 3, 2010 at 8:36 AM, Koen Kooi <k.kooi@student.utwente.nl> wrote:
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> >
> > On 03-08-10 15:35, Martin Jansa wrote:
> >> On Tue, Aug 03, 2010 at 12:13:58PM +0200, Koen Kooi wrote:
> >>> -----BEGIN PGP SIGNED MESSAGE-----
> >>> Hash: SHA1
> >>>
> >>> Acked-by: Koen Kooi <k-kooi@ti.com>
> >>>
> >>> If noone objects, can we get this in ASAP?
> >>
> >> Building from scratch armv4t then armv6 was ok, building few more
> >> packages for armv4t again still worked, but now I've upgraded gcc/eglibc
> >> and building eglibc-initial for armv4t first fails:
> >>
> >> http://tinderbox.openembedded.org/packages/665289/
> >> http://build.shr-project.org/tests/jama/config.log
> >>
> >> see
> >> configure: error: compiler support for __thread is require
> >> and
> >> Fatal error: Invalid -march= option: `armv4t'
> >
> > I had that with a previous incarnation of that patchset, but it seems to
> > have gone away.
> > Do we want to solve this problem first or apply the patches and solve it
> > later?
> I would suggest that I push these changes and meanwhile
> I will try to reproduce Martin's problem it will take few hours on my
> box. It seems that the binutils got wiped out
> when Martin cleaned stuff and then building eglibc resorted to
> assembler on build machine
> and that bombs later.

I think I had last incarnation (at last latest on patchwork - pw-am.sh
2521 2524 2525 2523 2522).

Khem was right about binutils issue.

I did something like this:
MACHINE=om-gta02 bitbake shr-image
MACHINE=htcdream bitbake shr-image
git pull (for newer gcc+eglibc)
MACHINE=om-gta02 bitbake shr-image - and now it failed to build eglibc-initial
MACHINE=htcdream bitbake shr-image - this upgraded gcc/eglibc fine

export MACHINE=om-gta02
for i in openembedded/recipes/gcc/gcc*4.5*bb \
         openembedded/recipes/eglibc/eglibc*2.12*.bb; do 
  bitbake -c clean -b $i;
done
MACHINE=om-gta02 bitbake shr-image - still failing in eglibc-initial

export MACHINE=htcdream
for i in openembedded/recipes/gcc/gcc*4.5*bb \
         openembedded/recipes/eglibc/eglibc*2.12*.bb; do
  bitbake -c clean -b $i;
done
MACHINE=om-gta02 bitbake shr-image - still failing in eglibc-initial

export MACHINE=om-gta02
for i in openembedded/recipes/gcc/gcc*4.5*bb
         openembedded/recipes/eglibc/eglibc*2.12*.bb
         openembedded/recipes/binutils/binutils*2.20.1*.bb ; do 
  bitbake -c clean -b $i; 
done
MACHINE=om-gta02 bitbake shr-image - seems to pass now

Regards,

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com