This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] - Use of powerpc 64bit instructions in 32bit ABI
- From: Richard Henderson <rth at redhat dot com>
- To: Fariborz Jahanian <fjahanian at apple dot com>
- Cc: David Edelsohn <dje at watson dot ibm dot com>, Ian Lance Taylor <ian at wasabisystems dot com>, "David S. Miller" <davem at redhat dot com>, gcc-patches at gcc dot gnu dot org
- Date: Wed, 15 Oct 2003 16:44:55 -0700
- Subject: Re: [PATCH] - Use of powerpc 64bit instructions in 32bit ABI
- References: <20031015224107.GB28205@redhat.com> <7996C2E2-FF64-11D7-96DA-000393B9ED88@apple.com>
On Wed, Oct 15, 2003 at 04:08:23PM -0700, Fariborz Jahanian wrote:
> Apple patch keeps 64bit values in GPRs. If this can be accomplished without
> changes to common part please let us know how sparc port accomplishes this.
By letting HARD_REGNO_MODE_OK return true for DImode.
Now lets stop arguing about this. Your patch is *not* going
to go in as-is. You have two choices:
(1) Arrange for BITS_PER_WORD to remain unchanged for G5 in
32-bit mode, and rearrange a bit of the backend to let it
know that 64-bit quantities may be held in registers.
I really suspect that you want to persue this option.
(2) Have BITS_PER_WORD==64, but rewrite your patch such that
you do not break middle-end constraints. I've already
pointed out two parts that are wrong and how to fix them.
In either case, do not give us another 5000 line patch. It
must be broken up into smaller logically coherent parts.