This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: Testsuite failure ?
- To: David Ronis <ronis at onsager dot chem dot mcgill dot ca>
- Subject: Re: Testsuite failure ?
- From: Manfred Hollstein <manfredh at redhat dot com>
- Date: Fri, 25 Aug 2000 22:58:33 +0200 (MEST)
- Cc: law at cygnus dot com, aoliva at redhat dot com, gcc-bugs at gcc dot gnu dot org
- References: <200008251937.PAA21129@onsager.chem.mcgill.ca>
- Reply-To: Manfred Hollstein <manfred dot h at gmx dot net>
On Friday, 25 August 2000, 15:37:09 -0400, ronis@onsager.chem.mcgill.ca wrote:
>
> Jeffrey A. Law replied:
>
> malign-double CHANGES THE ABI.
>
> This means that if you use it, you have to make sure everything is
> compiled
> with that flag. Otherwise bad things will happen. Depending on the
> code
> you write, it might go unnoticed for weeks, months or even years, but
> eventually you'll get nailed.
>
> And when we say everything, we mean it. It includes glibc, every
> library,
> and every application.
>
> -malign-double is evil. Do not use it.
>
> I think I understand. However, I may have added this flag when I built libc
> [I do to everything else]. Also:
>
> 1) I thought alignment was a big performace win on 686's?
Yeah, but I believe, current versions of the tools (gcc/g++ and binutils)
are much smarter in figuring out what is most appropriate.
>
> 2) What about the other -malign- ix86 options? For example, the default ix86
> makefile used to build the linux kernel has loads of these.
Well, there's still the recommendation to use gcc-2.7.2.3 for building
the v2.0/v2.2 Linux kernels. And, gcc-2.7.2.3 wasn't smart enough to figure
out what I described above.
>
> David
>
>
>
Cheers.
l8er
manfred