Re: [2.6 patch] fix the EMBEDDED menu

From: Adrian Bunk
Date: Wed Dec 14 2005 - 17:23:35 EST


On Wed, Dec 14, 2005 at 10:17:03PM +0000, Russell King wrote:
> On Wed, Dec 14, 2005 at 02:05:31PM -0800, Andrew Morton wrote:
> > Adrian Bunk <bunk@xxxxxxxxx> wrote:
> > >
> > > Hi Linus,
> > >
> > > your patch to allow CC_OPTIMIZE_FOR_SIZE even for EMBEDDED=n has broken
> > > the EMBEDDED menu.
> >
> > It looks like that patch needs to be reverted or altered anyway. sparc64
> > machines are failing all over the place, possibly due to newly-exposed
> > compiler bugs.
> >
> > Whether it's the compiler or it's genuine kernel bugs, the same problems
> > are likely to bite other architectures.
>
> I believe there are instances where ARM fails if CC_OPTIMIZE_FOR_SIZE
> is not set. Luckily we have assertions in the generated assembly to
> flag these as assembly errors when they happen, rather than silently
> continuing to build.
>
> Maybe CC_OPTIMIZE_FOR_SIZE should be:
>
> bool "..." if BROKEN || (!ARM && !SPARC64)
>
> ? 8)
>
> Note also that the help text:
>
> WARNING: some versions of gcc may generate incorrect code with this
> option. If problems are observed, a gcc upgrade may be needed.
>
> is reversed for the situation we have with ARM. Hence, I propose we
> change this to something like:
>
> WARNING: some versions of gcc may generate incorrect code if this
> option is changed form the platform default. If problems are
> observed, either a gcc upgrade may be needed or alternatively
> the platform default should be selected (=y for ARM and Sparc64,
> n for others.)

The patch

[2.6 patch] offer CC_OPTIMIZE_FOR_SIZE only if EXPERIMENTAL

I sent a few minutes ago should handle this:
It allows changing the platform default only if EXPERIMENTAL is enabled.

> Russell King

cu
Adrian

BTW: The second platform defaulting to CC_OPTIMIZE_FOR_SIZE is H8300,
not SPARC64.
--

"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/