This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [m68k] Fix and improve ColdFire function prologue/epilogue generation
- From: Gunther Nikl <gni at gecko dot de>
- To: Richard Henderson <rth at redhat dot com>
- Cc: Bernardo Innocenti <bernie at develer dot com>, Peter Barada <peter at baradas dot org>, gcc-patches at gcc dot gnu dot org
- Date: Thu, 4 Sep 2003 12:02:24 +0200
- Subject: Re: [m68k] Fix and improve ColdFire function prologue/epilogue generation
- References: <200308130240.03115.bernie@develer.com> <20030828231409.GE5175@redhat.com> <3F55D07A.9010801@develer.com> <20030903161051.GA26137@lorien.int.gecko.de> <20030903171304.GB6487@redhat.com>
On Wed, Sep 03, 2003 at 10:13:04AM -0700, Richard Henderson wrote:
> On Wed, Sep 03, 2003 at 06:10:51PM +0200, Gunther Nikl wrote:
> > Is %a6 %fp here?
>
> Yes. I was trying to be less confusing by only using the name %fp when
> it is used as a frame pointer.
Well, %a6 isn't necessarily the frame-pointer. My target uses %a5.
> > AFAICT, merging frame_size and saved_regs_size is only possible if
> > frame_size is > 8 and its < (32k - saved_regs_size).
>
> I'd forgotten about the subq case. I guess that could be moderately
> common -- something along the lines of
>
> int dummy;
> foo(&dummy);
No, a function conatining such code would use the frame-pointer on m68k
unless you pass the argument in a register. In that case GCC would use
%sp to access dummy.
> For the case of nearly 32k, I wouldn't think that would be common enough
> to be worth worrying about.
Thats probably true.
Gunther