From owner-freebsd-hackers  Mon Feb  5 04:49:58 1996
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id EAA24738
          for hackers-outgoing; Mon, 5 Feb 1996 04:49:58 -0800 (PST)
Received: from relay.hp.com (relay.hp.com [15.255.152.2])
          by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id EAA24732
          for <hackers@freebsd.org>; Mon, 5 Feb 1996 04:49:57 -0800 (PST)
Received: from fakir.india.hp.com by relay.hp.com with ESMTP
	(1.37.109.16/15.5+ECS 3.3) id AA299074496; Mon, 5 Feb 1996 04:48:21 -0800
Received: from localhost by fakir.india.hp.com with SMTP
	(1.37.109.16/15.5+ECS 3.3) id AA155974303; Mon, 5 Feb 1996 18:15:03 +0530
Message-Id: <199602051245.AA155974303@fakir.india.hp.com>
To: gjennejohn@frt.dec.com
Cc: hackers@freebsd.org
Subject: Re: Willows 
In-Reply-To: Your message of "Mon, 05 Feb 1996 12:47:16 +0100."
             <9602051147.AA25141@cssmuc.frt.dec.com> 
Date: Mon, 05 Feb 1996 18:15:02 +0530
From: A JOSEPH KOSHY <koshy@india.hp.com>
Sender: owner-hackers@freebsd.org
Precedence: bulk


gj> I didn't think of looking there. I think that the X server uses LDTs,

I'm curious about this.  Does the X server really need specific LDT
support on FreeBSD?  AFAIK we need access to the ioports on the machine 
(thus an IOPL change is in order) and mmaping of framebuffer memory on some
of the cards (which is standard, taking care to make the FB non-cacheable).  

However there shouldn't be a need for anything more machine specific than that.

On this topic; do we support the P5's 1MB page size?  This could reduce
the number of TLB misses when mmaping the video framebuffer (1-4MB typ) and
when the CPU does typical drawing (bresenham lines get hit particularly 
by tlb stepping).  Even the accelarators which use sparse addressing
of their registers (Mach64, P9000, P9100 etc) this could be a win.

Koshy