Path: utzoo!attcan!uunet!wuarchive!usc!jarthur!nntp-server.caltech.edu!rknop
From: rknop@nntp-server.caltech.edu (Robert Andrew Knop)
Newsgroups: comp.sys.cbm
Subject: Re: HELP: Assembly - Address Maintenance for code overlaying
Message-ID: <1990Dec10.192057.1239@nntp-server.caltech.edu>
Date: 10 Dec 90 19:20:57 GMT
References: <2373.27637c31@iccgcc.decnet.ab.com>
Organization: California Institute of Technology, Pasadena
Lines: 34

lockemer@iccgcc.decnet.ab.com writes:

>I am writing a program in assembly which will need to load different sections
>of code into the same areas of memory. These sections will have multiple
>routines in them. Does anyone have any suggestions... on how to access
>the routines from the main code (such as using labels somehow) so... I do not
>have to manually keep track of where each routine ends up any time it is
>reassembled? I am writing this on a 128 using MERLIN 128.

You could always use the GEOS/VLIR trick, which is include a jump table at the
beginning of the overlay module (the section of code that has multiple sets
of routines loaded into it).  For example, if you had three routines, the
beginning of the overlay module would look like

Overlay_start:
               jmp  routine1
               jmp  routine2
               jmp  routine3

Then, whenever you want to access these routines from your main code (the
code that is always memory resident), you would put in a statment like

   jsr Overlay_Start+(n-1)*3

where n is the number of the routine (1, 2, or 3).  As long as every overlay
module starts at the same place in memory (the location Overlay_Start), and
as long as you don't change that location, you can move the routnes in the
overlay module around, and only need to recompile that module; recompiling
that module will create the correct jump table, but the location of the jump
table is the same, so that no changes need to be made to the routine that
accesses the jump table.

-Rob Knop
rknop@juliet.caltech.edu