From owner-cvs-sys  Wed Jul 19 01:58:24 1995
Return-Path: cvs-sys-owner
Received: (from majordom@localhost)
          by freefall.cdrom.com (8.6.11/8.6.6) id BAA16281
          for cvs-sys-outgoing; Wed, 19 Jul 1995 01:58:24 -0700
Received: from gndrsh.aac.dev.com (gndrsh.aac.dev.com [198.145.92.241])
          by freefall.cdrom.com (8.6.11/8.6.6) with ESMTP id BAA16256
          ; Wed, 19 Jul 1995 01:58:08 -0700
Received: (from rgrimes@localhost) by gndrsh.aac.dev.com (8.6.11/8.6.9) id BAA06594; Wed, 19 Jul 1995 01:58:27 -0700
From: "Rodney W. Grimes" <rgrimes@gndrsh.aac.dev.com>
Message-Id: <199507190858.BAA06594@gndrsh.aac.dev.com>
Subject: Re: cvs commit: src/sys/i386/include pmap.h
To: davidg@Root.COM
Date: Wed, 19 Jul 1995 01:58:27 -0700 (PDT)
Cc: CVS-commiters@freefall.cdrom.com, cvs-sys@freefall.cdrom.com
In-Reply-To: <199507190713.AAA00404@corbin.Root.COM> from "David Greenman" at Jul 19, 95 00:13:01 am
X-Mailer: ELM [version 2.4 PL24]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Content-Length: 1888      
Sender: cvs-sys-owner@freebsd.org
Precedence: bulk

> 
> >> 
> >> davidg      95/07/18 23:37:13
> >> 
> >>   Modified:    sys/i386/i386  machdep.c
> >>                sys/i386/include  pmap.h
> >>   Log:
> >>   Rewrote memory sizing code to generally deal with holes in extended memory.
> >>   This code change should allow certain Compaq machines with a 128K hole
> >>   at 16MB to work.
> >
> >And probably just fixed a problem for all PS/2 ABIOS class machines
> >with a simliar ``hole'', but often larger (15->16MB is missing on
> >a true blue PS/2 ABIOS machine.)
> 
>    Unfortunately, it doesn't yet have a mechanism that allows you to say "I
> have a memory hole in this range of addresses, so please skip it". The code
> tries to figure it out by itself and this will obviously fail if there are
> some shared memory devices that are addressed in the hole. I'm sure you agree
> that it would be bad to add that shared memory into the global page pool. :-)

:-) yea, and the PS/2 has some really fun stuff in there!

>    I don't know how to deal with this problem. It almost needs to be a
> userconfig'able sort of thing. Perhaps two configurable holes that can be
> specified to map out a given area would be sufficient for most cases.

For the PS/2 it would be detectable very early on, you simply need to
look for a few magic tokens and find out if you have an ABIOS system,
if you do 15M->16M is off limits to you.

I'll agree that this should be configurable in some way, but I think your
idea of 2 holes would cover 90% of what we are likely to ever run into,
and the other 10% is going to be so bizzare we won't ever be able to easily
deal with it.

> Anyway,
> my point is only that the mechanism doesn't yet exist.

Well, okay, but at least a start has been made!!!


-- 
Rod Grimes                                      rgrimes@gndrsh.aac.dev.com
Accurate Automation Company                 Reliable computers for FreeBSD