This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] set -mabi=altivec with -ftree-vectorize
- From: Daniel Jacobowitz <drow at false dot org>
- To: Zdenek Dvorak <rakdver at kam dot mff dot cuni dot cz>
- Cc: gcc-patches at gcc dot gnu dot org, dje at watson dot ibm dot com
- Date: Tue, 31 Jul 2007 07:08:21 -0400
- Subject: Re: [patch] set -mabi=altivec with -ftree-vectorize
- References: <20070731071457.GB32641@kam.mff.cuni.cz>
On Tue, Jul 31, 2007 at 09:14:57AM +0200, Zdenek Dvorak wrote:
> Hello,
>
> using -O2 -maltivec -ftree-vectorize on ppc-linux leads to
> misscompilations, as -mabi=altivec is not enabled there.
> It would be somewhat annoying and error-prone to require a normal
> user to know this and add the extra option. As David Edelsohn
> suggested, this patch makes us enable -mabi=altivec automatically
> with -ftree-vectorize.
Is that really a good idea? It makes sense that -mabi= is an
ABI-changing option, but I'd be surprised if adding -ftree-vectorize
changed the ABI.
BTW, we recently made the linker warn about mismatches between soft
and hard float PowerPC code. I have a patch to warn about mismatchs
between code with -mabi=altivec and code without it. I just haven't
posted it yet because PR 32919 breaks my test scripts.
--
Daniel Jacobowitz
CodeSourcery