Path: utzoo!mnetor!uunet!lll-winken!lll-lcc!ames!ucsd!sdcsvax!ucsdhub!hp-sdd!hplabs!sdcrdcf!ism780c!tim
From: tim@ism780c.UUCP (T.W."Tim" Smith, Knowledgian)
Newsgroups: comp.sys.atari.st
Subject: Re: About Atari's, Mac's, and IBM's
Message-ID: <9790@ism780c.UUCP>
Date: 26 Apr 88 16:08:29 GMT
References: <5419@ucbvax.Berkeley.EDU> <226500010@trsvax> <213@bdt.UUCP> <4738@cup.portal.com>
Reply-To: tim@ism780c.UUCP (T.W."Tim" Smith, Knowledgian)
Organization: Suction and Pressure Lab, California Institute of Lawsonomy
Lines: 26

In article <4738@cup.portal.com> Thomas_E_Zerucha@cup.portal.com writes:
< The 386 has a 'virtual real memory mode' (no, that is not a contradiction,
< it emulates an 8086 address space in paged memory, so you can still play
< segment overlap games (theoretically)), and there may be one or more products
< out that take advantage of this, but I don't know how "protected" it would
< be.

On many ( most? ) implementations of Unix for the 386, there are products
available that use this mode ( called 'virtual 86 mode' ) to provide for
the execution of msdos programs under Unix.

When in virtual 86 mode, the 386 computes addresses in the same way an
8086 would ( (segment number << 4 ) + offset ), but instead of this being
a physical address, as it would be on an 8086, it is sent to the paging
unit for address translation.

A process running in this mode is as "protected" as any other process
running in protected mode.

Combine this with a Unix that implements the dos filesystem under the
System V3 file system switch, and you have a quite reasonable union of
dos and Unix ( why you would want to slime your Unix with dos is
another matter...)
-- 
Tim Smith				tim@ism780c.isc.com
"I don't practice what I preach because I'm not the
                       kind of person I'm preaching to" -- J.R. "Bob" Dobbs