Re: altivec triplets: comments?

[Geoff Keating <geoffk at geoffk dot org>]

> Cc: Aldy Hernandez <aldyh@redhat.com>, gcc@gcc.gnu.org
> From: Alexandre Oliva <aoliva@redhat.com>
> Organization: GCC Team, Red Hat
> Date: 18 Nov 2001 09:56:35 -0200
> User-Agent: Gnus/5.0805 (Gnus v5.8.5) Emacs/20.7
> 
> On Nov  7, 2001, Geoff Keating <geoffk@geoffk.org> wrote:
> 
> > Daniel commented that
> 
> >> For instance, there aren't triplets for every new MIPS
> >> processor supported by GCC
> 
> > but actually, when we build a special version of Linux for some weird
> > MIPS ABI, a different triplet _is_ used.  For instance,
> 
> > mips-linux-gnu
> > mipsisa32-linux-gnu
> 
> In this case, wouldn't it make sense for powerpc-linux-gnualtivec be
> spelled powerpcaltivec-linux-gnu?  Otherwise, you wouldn't be able to
> do something like powerpc-linux-gnulibc2.2.

That means something different.  The triplet we want isn't saying that
the processor has the altivec instructions, but that the libc has been
compiled for the altivec ABI.

> I don't like powerpcaltivec much, but I think it's better than
> gnualtivec, and, at least, there's more tradition to encode this kind
> of change in the machine type than in the kernel-OS type.
> 
> powerpc-eabi* are more of an exception than a rule, IMO.
> 
> > and in fact there are a bunch of triplets for particular MIPS
> > processors, although most of these don't change the ABI they just
> > change the ISA.
> 
> Indeed.  I don't recall any precedent to changing the triplet to
> accomodate an ABI change.  Most often, trailing text has been added to
> indicate a change in object format, not ABI.

That's exactly what the '2.2' means in 'powerpc-linux-gnulibc2.2'; it
indicates an ABI change.

-- 
- Geoffrey Keating <geoffk@geoffk.org> <geoffk@redhat.com>