From owner-freebsd-current  Fri Jan  2 11:17:13 1998
Return-Path: <owner-freebsd-current>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id LAA25111
          for current-outgoing; Fri, 2 Jan 1998 11:17:13 -0800 (PST)
          (envelope-from owner-freebsd-current)
Received: from flea.best.net (root@flea.best.net [206.184.139.131])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id LAA25085;
          Fri, 2 Jan 1998 11:16:57 -0800 (PST)
          (envelope-from dillon@flea.best.net)
Received: (from dillon@localhost) by flea.best.net (8.8.8/8.7.3) id LAA14499; Fri, 2 Jan 1998 11:16:33 -0800 (PST)
Date: Fri, 2 Jan 1998 11:16:33 -0800 (PST)
From: Matt Dillon <dillon@best.net>
Message-Id: <199801021916.LAA14499@flea.best.net>
To: "John S. Dyson" <dyson@freebsd.org>
Cc: current@freebsd.org, dg@root.com
Subject: Interesting page table methodology
Sender: owner-freebsd-current@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

    I've recently been studying the MMU on the PowerPC and MIPS cpus.  The
    PowerPC's methodology in particular appeals to me.  It's probably not
    possible to do something similar with the Pentium's hardwired MMU, but
    here is the basic idea.

    With the PowerPC & MIPS cpus you can take an interrupt on a TLB miss


    (on the MIPS you don't have a choice.  On the PowerPC it's possible to
    do it either way depending on the chip0.  The interrupt routine nominally
    obtains the TLB