Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!mips!daver!bungi.com!news
From: garyj@neptun.pcs.com (Gary Jennejohn)
Newsgroups: comp.sys.nsc.32k
Subject: setjmp/longjmp
Message-ID: <9010170928.AA03120@neptun.pcs.com>
Date: 18 Oct 90 10:39:43 GMT
Sender: news@daver.bungi.com
Lines: 23
Approved: news@daver.bungi.com


Hi Bruce, et al.,

Thanks to Bruce for his thoughts on setjmp/longjmp. I tend to agree, but I'm
also rather conservative and take the approach that the general purpose
registers, at least, should be saved. This is what I did in my setjmp/
longjmp. I didn't do anything with the floating point registers. In my
opinion setjmp/longjmp shouldn't be misused as though they were save/resume
in the kernel.

What I mean by standalone is: a copy of the monitor is downloaded to
memory and then I jump to it. The monitor in memory then takes the place
of the one in EPROM. All traps, interrupts, etc., are handled by the
memory-resident monitor. It's with the memory-resident monitor that I was
having problems, but they've since disappeared. Apparently I was doing
something somewhere in the memory test code that was leading to problems
when I did a longjmp.

I now have the memory tests integrated into the monitor which I've tested
memory-resident. I fear that the monitor may now be too big to fit into
our 32k EPROM, but I'm not 100% sure.

Gary