From owner-freebsd-current  Fri May 15 05:48:33 1998
Return-Path: <owner-freebsd-current@FreeBSD.ORG>
Received: (from majordom@localhost)
          by hub.freebsd.org (8.8.8/8.8.8) id FAA16805
          for freebsd-current-outgoing; Fri, 15 May 1998 05:48:33 -0700 (PDT)
          (envelope-from owner-freebsd-current@FreeBSD.ORG)
Received: from rf900.physics.usyd.edu.au (rf900.physics.usyd.edu.au [129.78.129.109])
          by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id FAA16779;
          Fri, 15 May 1998 05:48:22 -0700 (PDT)
          (envelope-from dawes@rf900.physics.usyd.edu.au)
Received: (from dawes@localhost) by rf900.physics.usyd.edu.au (8.8.5/8.8.2) id WAA16578; Fri, 15 May 1998 22:46:37 +1000 (EST)
Message-ID: <19980515224637.08480@rf900.physics.usyd.edu.au>
Date: Fri, 15 May 1998 22:46:37 +1000
From: David Dawes <dawes@rf900.physics.usyd.edu.au>
To: Poul-Henning Kamp <phk@critter.freebsd.dk>
Cc: dyson@FreeBSD.ORG, sos@FreeBSD.ORG, current@FreeBSD.ORG
Subject: Re: People having problems with X windows?
References: <199805150738.CAA03075@dyson.iquest.net> <2028.895222610@critter.freebsd.dk>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 0.69
In-Reply-To: <2028.895222610@critter.freebsd.dk>; from Poul-Henning Kamp on Fri, May 15, 1998 at 10:56:50AM +0200
Sender: owner-freebsd-current@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

On Fri, May 15, 1998 at 10:56:50AM +0200, Poul-Henning Kamp wrote:
>In message <199805150738.CAA03075@dyson.iquest.net>, "John S. Dyson" writes:
>
>>I just removed my WC "improvements", by adding the return to the
>>subroutine in the pmap code.  See if it is more friendly.  The
>>symptoms appear to be those associated with improper setting of
>>WC mode for a display adaptor.  I had to add a heuristic to make
>>it work correctly on one of my cards, but I guess it was bad.
>>
>>The code is still in there, but I guess there is the need for more
>>research into the issue.
>
>I would expect WC to break any and all accelerated adapters that
>communicate with the graphics engine through the ram window...

For PCI cards, memory windows used for MMIO shouldn't be marked
pre-fetchable.  Memory windows that are purely for framebuffer use could
be.

>What is needed is to XFree86 to be able to madvise the true frame
>buffer as "Write-combine" and leave the rest of the ram window 
>unaffected.

Having such a mechanism would be useful.

David

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message