From owner-freebsd-hackers  Thu Sep 18 17:32:10 1997
Return-Path: <owner-freebsd-hackers>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id RAA21563
          for hackers-outgoing; Thu, 18 Sep 1997 17:32:10 -0700 (PDT)
Received: from word.smith.net.au (ppp20.portal.net.au [202.12.71.120])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id RAA21552
          for <hackers@freebsd.org>; Thu, 18 Sep 1997 17:32:02 -0700 (PDT)
Received: from word.smith.net.au (localhost.smith.net.au [127.0.0.1])
	by word.smith.net.au (8.8.7/8.8.5) with ESMTP id JAA02973;
	Fri, 19 Sep 1997 09:59:42 +0930 (CST)
Message-Id: <199709190029.JAA02973@word.smith.net.au>
X-Mailer: exmh version 2.0zeta 7/24/97
To: joerg_wunsch@uriah.heep.sax.de (Joerg Wunsch)
cc: hackers@freebsd.org
Subject: Re: INB question 
In-reply-to: Your message of "Thu, 18 Sep 1997 22:18:39 +0200."
             <19970918221839.VL10449@uriah.heep.sax.de> 
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Date: Fri, 19 Sep 1997 09:59:39 +0930
From: Mike Smith <mike@smith.net.au>
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by hub.freebsd.org id RAA21558
Sender: owner-freebsd-hackers@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

> As Mike Smith wrote:
> > 
> > The ISA specification explicitly requires bus pullup resistors.  It may 
> > be unwise to depend on reading 0xff back-to-back with a previous read/
> > write operation, ...
> 
> That's why i wrote ``unspecified, with a tendency to 0xff''.

The implication (as an english speaker) from your claim was 
"unspecified but sometimes 0xff".  It would be civilised to qualify the 
"tendency" under the circumstances.

> > but the reader is welcome to calculate the RC time constant for a
> > transmission line with a few pF of capacitance and a 10K (or less)
> > pullup.
> 
> j@uriah 132% perl -e 'print 50e-12 * 10e3; print "\n"'
> 5e-07
> 
> I think 50 pF is rather an understatement.  0.5 µs doesn't seem to be
> terribly short, but IIRC, ISA inb's are artificially deferred by 1.25
> µs, so chances are good to actually see 0xff.  I wouldn't rely on it
> for a back-to-back read, however.

50pf is a little on the low side, granted.  I don't have my copy of 
Solari handy to check, but a quick eyeball of the boards to hand 
indicates that something lower like 4k7 is common practice, so 500ns
drift-up time is not unreasonable.

OBTW, see my trailing comment wrt. transfer rates; if ISA read cycles 
are deferred by 1.25us, how do I manage 1.3MW/sec from a user-space 
process?  (This is with a P166 on an HX board; nothing special.)

mike