This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Mechanism to configure default scheduling model on PA
- From: law at redhat dot com
- To: "John David Anglin" <dave at hiauly1 dot hia dot nrc dot ca>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Fri, 31 May 2002 12:15:25 -0600
- Subject: Re: Mechanism to configure default scheduling model on PA
- Reply-to: law at redhat dot com
In message <200205311754.g4VHs8gF006581@hiauly1.hia.nrc.ca>, "John David
Anglin" writes:
> It's easier to use PA7100LC for hppa1.0 and hppa1.1 targets, and PA8000
> for the rest. A target with an unspecified architecture would use
> PA8000. The idea was to force people to specify their target
> completely if they want something other than the default, and default
> to the newer architecture. Of course, config.guess generates a complete
> specification if no target option is provided.
>
> Does this approach sound ok?
Seems reasonable.
> In a similar vein, Stefan Becker was
> interested in being able to change the default architecture without
> hacking config.gcc.
I've been pondering this kind of thing recently as well. Specifically, I've
been thinking about requesting comments on changing GCC's behavior when
configured/built on a PA2.0 machine.
Right now if config.guess returns hppa1.0*, then you get PA1.0. If it
returns hppa1.1* or hppa2.0*, then you get PA1.1. What I've been pondering
is changing the behavior of hppa2.0* so that you'd get PA2.0 code generation.
(note you'd still have to configure hppa64 to get PA64).
Thoughts?
jeff