Path: csiph.com!x330-a1.tempe.blueboxinc.net!newsfeed.hal-mli.net!feeder3.hal-mli.net!newsfeed.hal-mli.net!feeder1.hal-mli.net!npeer01.iad.highwinds-media.com!news.highwinds-media.com!feed-me.highwinds-media.com!post01.iad.highwinds-media.com!newsfe11.iad.POSTED!00000000!not-for-mail
From: Stefan Monnier <monnier@iro.umontreal.ca>
Newsgroups: comp.arch
Subject: Re: M68k add to memory is not a mistake any more
Message-ID: <jwvty2y3clb.fsf-monnier+comp.arch@gnu.org>
References: <ggtgp-8D1AEA.03180231012012@netnews.mchsi.com> <jgeu27$9rg$1@dont-email.me> <a7csv8-emn1.ln1@ntp6.tmsw.no> <jgiibd$hkq$1@dont-email.me> <4F2CD4FA.4050004@SPAM.comp-arch.net> <jgjn1i$nqe$1@dont-email.me> <ave009-2ft1.ln1@ntp6.tmsw.no> <jgk6tk$kcb$1@dont-email.me> <j17209-p972.ln1@ntp6.tmsw.no> <jguq3a$rb6$1@dont-email.me> <4F341C7B.9070703@SPAM.comp-arch.net> <jh2f6v$6jg$1@dont-email.me> <4F352B0B.6040108@SPAM.comp-arch.net>
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/24.0.92 (gnu/linux)
Cancel-Lock: sha1:cuhmQfkRexKTb2Hlq5Z6iz0MvEg=
MIME-Version: 1.0
Content-Type: text/plain
Lines: 12
X-Complaints-To: abuse@UsenetServer.com
NNTP-Posting-Date: Fri, 10 Feb 2012 16:01:37 UTC
Date: Fri, 10 Feb 2012 11:01:37 -0500
X-Received-Bytes: 1623
Xref: x330-a1.tempe.blueboxinc.net comp.arch:5863

> What you propose - writing to one register higher than the input - 
> causes me to break out in screaming fits, but, based on what my compiler guy
> friends in the past have said, can *PROBABLY* be handled.

Agreed.  At least, for a Chaitin/Briggs style graph coloring register
allocator, it's pretty easy to deal with restrictions "this can only go
to this subset of registers" by simply adding a few constraints to the
input graph, but having the output reg depend on the choice of the input
reg seems a lot less straightforward to express.


        Stefan