This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: discouraging register use
- From: DJ Delorie <dj at redhat dot com>
- To: rth at redhat dot com
- Cc: gcc at gcc dot gnu dot org
- Date: Mon, 3 Jan 2005 15:22:06 -0500
- Subject: Re: discouraging register use
- References: <200412302354.iBUNsiBH002125@greed.delorie.com> <20041231062116.GA12583@redhat.com>
> > [(set (match_operand:SI 0 "" "c,c,?m,?m")
> > (plus:SI (match_operand:SI 1 "" "0,0,0,0")
> > (match_operand:SI 1 "" "c,?m,c,?m")))]
>
> Yes, that looks plausible. The other way isn't going to work because
> you penalize c at the same time as m by using "?cm".
>
> Does it actually work?
One case I tried seems to work, but it will mean some REALLY long
constraint lists for some of the common insns. I suppose I can skip
the generic move insns, as they shouldn't have too much bearing on
register selection, if I get all the other insns (which have shorter
constraint lists).
But it's a lot of work to fix up everything, I wanted to make sure I
wasn't missing something obvious before I started on it.