Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ncar!tank!shamash!odeon!bga
From: bga@odeon.ahse.cdc.com (Bruce Albrecht)
Newsgroups: comp.arch
Subject: Re: Self-modifying code
Message-ID: <2441@odeon.ahse.cdc.com>
Date: 12 Oct 89 02:40:43 GMT
References: <1080@mipos3.intel.com> <48682@ricerca.UUCP> <236@bbxsda.UUCP>
Organization: Control Data Corporation
Lines: 18

In article <236@bbxsda.UUCP>, scott@bbxsda.UUCP (Scott Amspoker) writes:
> So, if you were maintaining some assembly code and enountered
> an instruction such as:
> 
>     bra  *-*
> 
> You knew that the operand would be determined at runtime. 
> Scary stuff, no?

It's equivalent to the FORTRAN assigned GOTO, or the PDP-11's JSR/JMP reg,
or other jumps calculated on the fly, it's just that the address is saved
in memory.

CDC's peripheral processors (PP's) had 4k bytes.  In order to squeeze 
everything into a PP, they relied heavily on overlays, and putting code
in buffer areas to preset constants, etc.  The program would execute the
code in the buffer, which set any jumps, constants, etc, that needed
changing, and then reuse the buffer in the main code area.