This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: hard register reload patch
- To: Joern Rennecke <amylaar at cygnus dot co dot uk>
- Subject: Re: hard register reload patch
- From: Bernd Schmidt <bernds at pathia dot cygnus dot co dot uk>
- Date: Wed, 1 Dec 1999 12:59:50 +0000 (GMT)
- cc: law at cygnus dot com, gcc-patches at gcc dot gnu dot org
On Wed, 1 Dec 1999, Joern Rennecke wrote:
> > > registers. Joerns patch is a step in that direction (combine_reloads goes
> > > away).
> > How related is this to your reload cleanup patch. It sounds like the same
> > thing, and I've already agreed that we're going in the right direction.
> > Continuing down this road seems like a reasonable thing to do.
>
> But that aspecty can't be simply broken out - combine_reload reuses hard
> registers that die. If you remove it without a replacement, register pressure
> will increase, resulting in worse code and more reload aborts when it runs
> out of spill registers.
>
> By patch replaces the functionality of combine_reloads because registers that
> die become available as spill registers, and the times they are not usable
> are encoded in the hard reloads.
It's possible that in the end we'll want to apply the whole patch (regardless
of what happens to sh.md). I just haven't looked at it closely enough to
determine which parts are doing what.
Regarding combine_reloads: didn't you say once that the first part of it is
useless even now? If so, let's delete it.
Bernd