This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Use-and-clobber insns and some ancient combine.c code
- From: Michael Matz <matz at suse dot de>
- To: Ulrich Weigand <weigand at i1 dot informatik dot uni-erlangen dot de>
- Cc: Richard Earnshaw <rearnsha at gcc dot gnu dot org>,kenner at vlsi1 dot ultra dot nyu dot edu, gcc at gcc dot gnu dot org
- Date: Fri, 18 Jun 2004 15:04:09 +0200 (CEST)
- Subject: Re: Use-and-clobber insns and some ancient combine.c code
- References: <200406171608.SAA05852@faui1m.informatik.uni-erlangen.de>
Hi,
On Thu, 17 Jun 2004, Ulrich Weigand wrote:
> I had not been aware of this interpretation of CLOBBER. This is
> unfortunate, as it appears to make the use of use-and-clobber
> type patterns unfeasible ...
Yes. A CLOBBER without any constraints is implicitely an early-clobber,
and yes that's unfortunate. The way I was working around this issue in
some local patches was to implement a kind of non-early-clobber, which is
simply a SET with an UNSPEC source:
parallel([
(set X Y)
(set Y (unspec blabla)) ])
Ciao,
Michael.