Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!usc!elroy.jpl.nasa.gov!ncar!ico!onecom!wldrdg!hans
From: hans@wldrdg.UUCP (Johann Ruegg)
Newsgroups: comp.sys.atari.st.tech
Subject: Sozobon hcc and add.w 'bug'
Message-ID: <420@wldrdg.UUCP>
Date: 3 Nov 90 14:13:35 GMT
Organization: Wildridge Consulting, Boulder, CO
Lines: 20

Someone recently claimed that the code generated by hcc for the stack
is wrong.  I.E.
		add.w	#4,a6
should be
		add.l	#4,a6
because you might cross a 64k boundary.

This is NOT a bug.  There is no 64k boundary.
I quote from the M68000 programmer's reference manual (ADDA page):

	Op-Mode field -- Specifies the size of the operation:
		011 -- word operation.  The source operand is sign-extended to
			a long operand and the operation is performed on the 
			address register using all 32 bits.

Using 'add.l' instead will only make the instruction 2 bytes longer for no
reason.

	Johann Ruegg
	Sozobon Ltd.