From owner-freebsd-current  Tue Apr 15 23:18:11 1997
Return-Path: <owner-current>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id XAA13919
          for current-outgoing; Tue, 15 Apr 1997 23:18:11 -0700 (PDT)
Received: from rah.star-gate.com (rah.star-gate.com [204.188.121.18])
          by freefall.freebsd.org (8.8.5/8.8.5) with ESMTP id XAA13901
          for <freebsd-current@freefall.freebsd.org>; Tue, 15 Apr 1997 23:18:09 -0700 (PDT)
Received: from rah.star-gate.com (localhost.star-gate.com [127.0.0.1]) by rah.star-gate.com (8.8.5/8.7.3) with ESMTP id XAA00525; Tue, 15 Apr 1997 23:17:25 -0700 (PDT)
Message-Id: <199704160617.XAA00525@rah.star-gate.com>
X-Mailer: exmh version 1.6.9 8/22/96
To: David Dawes <dawes@rf900.physics.usyd.edu.au>
cc: "Louis A. Mamakos" <louie@TransSys.COM>,
        Kazutaka YOKOTA <yokota@zodiac.mech.utsunomiya-u.ac.jp>,
        freebsd-current@freefall.freebsd.org
Subject: Re: moused enhancement (was: Re: make world) 
In-reply-to: Your message of "Wed, 16 Apr 1997 15:07:20 +1000."
             <19970416150720.17257@rf900.physics.usyd.edu.au> 
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Date: Tue, 15 Apr 1997 23:17:25 -0700
From: Amancio Hasty <hasty@rah.star-gate.com>
Sender: owner-current@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk


Ah, just one of the guys I was thinking about 8)

On my PPRO 200mhz with the Natoma chipset, i enabled "uc" memory mode for
my Dimond Stealth S3 968 and I noticed that the X server tends to lock up
solid sometimes  when the cursor changes. 

The following obscure instructions enables "uncached mode" across the
S3's linear frame buffer which is 0xf0000000 and is 4mb big.

wrmsr(0x204, 0xf0000001  );
wrmsr(0x205, 0xf8000000 | 0x800);

In essence this is what fastvid does and the codes needs to be in 
a driver -- the iopl trick at user level will not work.

Most of you will probably not run into this problem with a PPRO because
you don't have fast access to the S3 card's frame buffer.

For sure it speeds memory access to the card like about 4 times at least
thats what I measure in DOS. With mpeg players I get about a third better
thruput.


At any rate, it looks like the cursor for the S3 stuff needs a little bit
of work. My current X server version is 3.2a. 

It feels weird reporting a bug on something that I work a long time ago 8)


	Tnks!
	Amancio