Re: [2.6 patch] offer CC_OPTIMIZE_FOR_SIZE only if EXPERIMENTAL



On Wed, 14 Dec 2005, Adrian Bunk wrote:
>
> My patch has the advantage that it doesn't allow the broken
> CC_OPTIMIZE_FOR_SIZE=y setting on ARM if !EXPERIMENTAL.

That isn't how it was before either.

Before, it _asked_ you if EMBEDDED was set, and "y" was just the default
(but you could select "n" if you wanted to). I don't think it's
necessarily wrong to allow a -O2 ARM or H8300 kernel, although apparently
there are compilers that are broken that way too..

So my patch should give the old behaviour for the EMBEDDED platforms, and
_allow_ it for non-embedded unless SPARC64 is set, or EXPERIMENTAL isn't
set.

That sounds like the right thing to do to me..

Of course, the really right thing would be to chase down what goes wrong
with -Os. It might be a compiler bug, but it might be a real kernel bug
that just happens to bite us (-Os works fine for me on ppc64, and
apparently Fedora has used it at least on x86-64, but it might be
something subtle).

Linus
-
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/

Relevant Pages

  • Re: Latest git oopses during boot
    ... I'll try applying the patch to a freshly downloaded git-tree. ... I would suggest a patch mis-application problem first (or possibly even ... That does make a compiler bug at least more likely. ... want to let the gcc people know. ...
    (Linux-Kernel)
  • Re: Latest git oopses during boot
    ... I'll try applying the patch to a freshly downloaded git-tree. ... That does make a compiler bug at least more likely. ... want to let the gcc people know. ... because the size of Scsi_Host structure was multiples of 8. ...
    (Linux-Kernel)
  • Re: Latest git oopses during boot
    ... I'll try applying the patch to a freshly downloaded git-tree. ... I would suggest a patch mis-application problem first (or possibly even ... That does make a compiler bug at least more likely. ... want to let the gcc people know. ...
    (Linux-Kernel)