Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 Apollo; site apollo.UUCP
Path: utzoo!linus!decvax!wivax!apollo!alan
From: alan@apollo.UUCP (Alan Lehotsky)
Newsgroups: net.flame
Subject: Re: Godless 68000 instruction set
Message-ID: <201@apollo.UUCP>
Date: Mon, 12-Dec-83 09:34:05 EST
Article-I.D.: apollo.201
Posted: Mon Dec 12 09:34:05 1983
Date-Received: Tue, 13-Dec-83 02:37:51 EST
References: <1933@fortune.UUCP>
Organization: Apollo Computer, Chelmsford, Mass.
Lines: 18

The fact that the "MOVE.B <ea>,-(SP)" instruction pushes a WORD on the
stack "isn't a bug, it's a feature".

    [[BTW, we used to use that as an SPR response at DEC...]]

The problem is that the 68K (like it's spiritual ancestor the PDP-11)
has a WORD addressable memory architecture.  So imagine what happens if
you pushed a BYTE on the stack (in supervisor mode...) and then took
some sort of interrupt (which will cause the PC, etc to be pushed on
the stack.

    BOOM!
    ODD-ADDRESS-TRAP....

Of course, what I really don't like about this is that the architecture
manual is rather vague about which byte contains my data after the
push has completed.