Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!bloom-beacon!athena.mit.edu!tada
From: tada@athena.mit.edu (Michael J Zehr)
Newsgroups: comp.arch
Subject: Re: Self-modifying code
Summary: example code
Message-ID: <15000@bloom-beacon.MIT.EDU>
Date: 10 Oct 89 21:21:35 GMT
References: <1080@mipos3.intel.com> <4415@bd.sei.cmu.edu> <33334@ames.arc.nasa.gov>
Sender: daemon@bloom-beacon.MIT.EDU
Reply-To: tada@athena.mit.edu (Michael J Zehr)
Organization: Massachusetts Institute of Technology
Lines: 20


I found an example of self-modifying code back in the middle ages (c.
1981) in the operating system on a Z-80 chip.  As i recall, the Z-80 had
bit set, clear, and test op-codes which needed an assembly-time bit
number rather than using a register to determine which bit to
set/clear/test.  it took less space to determine the right bit and load
the opcode (i think a shift would give you the opcode for a bit, or
maybe shift and add an offset) into the right place in memory and then
execute it.  the other option was to have a 8 branch statements.

but the way to really save space was to call the routine with a
parameter which indicated which operation (set, clear, test) and after
computing the opcode to indicate the right bit, also select the opcode
to indicate the kind of operation and load that into memory and execute
it. 

*very* kludgy, but it took a surprisingly small amount of space with
only a small penalty in speed.

-michael j zehr