Path: utzoo!mnetor!uunet!lll-winken!lll-lcc!ames!amdahl!dlb!ardent!rap
From: rap@ardent.UUCP (Rob Peck)
Newsgroups: comp.sys.amiga
Subject: Resident programs
Message-ID: <317@ardent.UUCP>
Date: 19 Feb 88 17:25:23 GMT
Organization: Dana Computer, Inc., Sunnyvale, CA
Lines: 29
Keywords: RSN, neat, RESIDENT

Last night, at the BADGE meeting, Jim Goodnow II showed a few slides describing
a program he is working on to make most-any program "resident", even if
it was not designed to be resident in the first place.  Some programs,
like those in BCPL, are pure code and don't modify their code segment
while they run, allocating a data segment so to speak, when they run.
C programs are not usually that lucky, having both a segment of initialized
and noninitialized data.  Jim's program patches a few things in the
AmigaDOS tables, provides a custom loader for programs that runs as a
resident library, and effectively relinks the seglists of the loaded
program so that each invocation of the program can have a fresh copy of
the data segments, with the pure code repositioned at the end of the seglist
and the new data segments at the beginning, with a jump ahead of the data
(since the first segment first word is usally the beginning of the code
as far as AmigaDOS is concerned).  For now, the program works best with
code compiled with Aztec 3.4/3.6, but he is planning to extend the program
to recognized other compiler constructs (the program binary image has
to be patched in some areas to assure that the data is correctly referenced).
As I understood, small model (data area less than 64k), data referenced
through A4, works the best.

When?  Who knows, but it was certainly a stimulating lecture (slightly over
an hour, with "slides" showing where the patches would be wedged in).

Price?  I cannot speak for Jim, but I think I heard him say maybe PD.

Good job on the lecture, Jim.  I'm looking forward to it!  Hope the
slides are part of the docs you release for it, WELL DONE!


Rob Peck			...ihnp4!hplabs!ardent!rap