This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: SH Linux and multilib
- To: Andrew Haley <aph at cambridge dot redhat dot com>
- Subject: Re: SH Linux and multilib
- From: "M. R. Brown" <mrbrown at 0xd6 dot org>
- Date: Mon, 30 Jul 2001 11:03:43 -0500
- Cc: gcc at gcc dot gnu dot org
* Andrew Haley <aph@cambridge.redhat.com> on Mon, Jul 30, 2001:
>
> Sure, I agree. It's important that the user can build whatever they
> need -- the disagreement here is about what the default should do. As
> you say, the default is at present unusable.
>
Huh? From going over the thread, it seems like the disagreement is about
the ability to *pick* a default, rather than being forced to use SH3,
little-endian. NIIBE's patches (although IMHO the wrong approach) allow
you to do that - and it's very convenient if you only target one particular
board. The sh-unknown-linux-gnu still supports multilib, so you have the
option of using the -m[lb] and -m[34] options to your whim to pull in the
correct multilibs (for libgcc, libstdc++, whatever).
It's irrelevant once you get to building glibc, since you can only pick one
target, but I don't think that's the issue at hand - for whatever board you
need glibc to run on, you just pass the -m* options anyway.
My other point of contention w/ the sh[34] notation is that it looks
totally contrived when building sh-elf/sh-coff targets, as there's no
conformity with specifying targets. But I still believe a mechanism needs
to be in place to pick a default other than sh3el, preferably from the
configure line, so the -m* options become invisible.
BTW, sorry for the malformed e-mail I sent earlier :).
M. R.