>Date: Tue, 21 Mar 1995 15:29:52 -0500
>From: "Charles M. Hannum" <mycroft@ai.mit.edu>

>   Is the time inb's from port 0x84 last constant by
>   design/agreement/hardware ?
>
>In theory, reading from a non-existant ISA I/O port should cause an
>I/O timeout, which would take 1.25us (10 cycles of an 8MHz clock).  Or
>something like that; I don't have ISA specs.  In practice, it seems to
>work on many machines, but I wouldn't place any bets.

How can you be sure that I/O port 0x84 is (and will remain) unused?
According to a list I have, port 0x84 is used for Compaq POST
diagnostics, EISA bus cycle synchronization, and is part of the DMA
controller on some kind of oddball system (74612?).  The list I am
using came with version 44 of Ralf Brown's interrupt list.

-- 
Mike Long <mike.long@analog.com>       PGP 2.6.2 public key signature:
VLSI Design Engineer                  CCBF225E7D3F7ECB2C8F7ABB15D9BE7B
Analog Devices, CPD Division
Norwood, MA 02062 USA                assert(*this!=opinionof(Analog));