From owner-freebsd-hardware  Sun Nov  1 00:51:51 1998
Return-Path: <owner-freebsd-hardware@FreeBSD.ORG>
Received: (from majordom@localhost)
          by hub.freebsd.org (8.8.8/8.8.8) id AAA29891
          for freebsd-hardware-outgoing; Sun, 1 Nov 1998 00:51:51 -0800 (PST)
          (envelope-from owner-freebsd-hardware@FreeBSD.ORG)
Received: from math.berkeley.edu (math.Berkeley.EDU [128.32.183.94])
          by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id AAA29885
          for <hardware@FreeBSD.ORG>; Sun, 1 Nov 1998 00:51:50 -0800 (PST)
          (envelope-from dan@math.berkeley.edu)
Received: (from dan@localhost)
	by math.berkeley.edu (8.8.7/8.8.7) id AAA18074;
	Sun, 1 Nov 1998 00:51:35 -0800 (PST)
Date: Sun, 1 Nov 1998 00:51:35 -0800 (PST)
From: dan@math.berkeley.edu (Dan Strick)
Message-Id: <199811010851.AAA18074@math.berkeley.edu>
To: grog@lemis.com
Subject: Re: Mixing 8- and 16-bit shared memory ISA cards
Cc: dan@math.berkeley.edu, dg@root.com, hardware@FreeBSD.ORG, jdp@polstra.com
Sender: owner-freebsd-hardware@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.org

> >    It affects all machines with ISA busses, not just early x86.
> 
> What's the reason for it?

I don't remember the details, but I think one of the ISA signal lines
is asserted by the ISA card to indicate if it wants to do 8 or 16 bit
transfers.  For some reason that I don't remember, all the devices in
a single segment of the I/O space may simultaneously assert their
preference/requirement when any one of them is queried.  This is
probably just a consequence of cheap bus design.

Therefore you want all the cards in the same I/O space segment to
use the same transfer width.  I vaguely recall that the penalty
for mixing types is that the 16 bit cards end up doing 8 bit bus
transfers (i.e. slowly).

Dan Strick
dan@math.berkeley.edu

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