Re: Denys Dmytriyenko : beaglebone: disable THUMB2_KERNEL in defconfig for newer binutils

[Denys Dmytriyenko <denis@denix.org>]

On Thu, May 03, 2012 at 01:34:31PM -0400, Denys Dmytriyenko wrote:
> On Thu, May 03, 2012 at 07:12:05PM +0200, Koen Kooi wrote:
> > 
> > 
> > Op 3 mei 2012 om 18:21 heeft Denys Dmytriyenko <denis@denix.org> het volgende geschreven:
> > 
> > > On Thu, May 03, 2012 at 08:25:11AM +0200, Koen Kooi wrote:
> > >> 
> > >> Op 3 mei 2012, om 08:09 heeft Denys Dmytriyenko het volgende geschreven:
> > >> 
> > >>> On Thu, May 03, 2012 at 08:02:00AM +0200, Koen Kooi wrote:
> > >>>> 
> > >>>> Op 3 mei 2012, om 02:43 heeft Arago Project git het volgende geschreven:
> > >>>> 
> > >>>>> Module: meta-ti
> > >>>>> Branch: master
> > >>>>> Commit: 3c693c26087aa148b944166083b5336f05d1c15b
> > >>>>> URL:    http://arago-project.org/git/meta-ti.git?a=commit;h=3c693c26087aa148b944166083b5336f05d1c15b
> > >>>>> 
> > >>>>> Author: Denys Dmytriyenko <denys@ti.com>
> > >>>>> Date:   Wed May  2 20:40:58 2012 -0400
> > >>>>> 
> > >>>>> beaglebone: disable THUMB2_KERNEL in defconfig for newer binutils
> > >>>> 
> > >>>> Wait, what? Can we actually send these changes for review first? T2 is 
> > >>>> needed to meet the boot time goal, so please revert this one!
> > >>> 
> > >>> Have another solution to get it building under recent binutils?
> > >> 
> > >> Patch binutils or patch the kernel.
> > > 
> > > Like I said, do you have a solution? Do you have said patches?
> > > 
> > > Since gcc-4.5 and binutils-2.20 are scheduled to be removed from 
> > > meta-oe/toolchain-layer in a few days, I had to fix it some way. That's the 
> > > best I could do under this circumstances. I can't have beaglebone broken with 
> > > any  oher toolchain, to only gain few milliseconds of boot time.
> > 
> > It's in the range of half a second when I last tested it. 
> 
> If you care to provide binutils-2.20 with it forever, the change can be 
> reverted.
> 
> Again, your decision to drop gcc-4.5 and binutils-2.20 from meta-oe drove me 
> to push this fix.

Well, I might have found another way to fix it w/o changing the defconfig. 
Testing it now on a clean build - will let you know the result...

-- 
Denys