This is the mail archive of the
mailing list for the GCC project.
Re: rs6000 stack boundary
- From: Alan Modra <amodra at bigpond dot net dot au>
- To: Geoff Keating <geoffk at desire dot geoffk dot org>
- Cc: dje at watson dot ibm dot com, gcc-patches at gcc dot gnu dot org
- Date: Sat, 1 May 2004 11:38:22 +0930
- Subject: Re: rs6000 stack boundary
- References: <20040430143702.GO2565@bubble.modra.org> <firstname.lastname@example.org> <200404302212.i3UMC6T26024@makai.watson.ibm.com> <200404302355.i3UNtw61022391@desire.geoffk.org>
If I understand your objection, you're concerned that using -maltivec
with a target that happens to compile libraries with -meabi, will no
longer work. That's true.
However, such targets won't work with -mabi=altivec either. Except
for t-spe, I see no multilib selection for mabi=altivec. This, along
with the fact that the altivec support in gcc currently doesn't doesn't
work without 16 byte alignment, means that -maltivec and -mabi=altivec
on such targets is not usable anyway.
What I'd like to see is a -maltivec option that works for linux. It
doesn't currently unless -mabi=altivec is also used, and even that
combination is buggy. Please don't erect roadblocks. I can't fix
all the problems in one go.
IBM OzLabs - Linux Technology Centre