Jim,

In regards your query about:

>Do you think a single step program monitor could be implemented
>in an ISR?
>Could the '877 be a minimal ICD without the ICD?

I haven't looked at the ICD, as yet. Do you want to monitor internal
program operation (ie, by emulating the program in the '877), or just
monitor external pin signals, while single-stepping? Sounds like the
former. I haven't been going that route, but mostly building
"minimalist" h.w. debug tools using the PIC and Scenix - in my latest
device, the SX steps the clock of the target DUT, and monitors
external pins, until a trigger-point occurs.

The idea of using the '877 as a single-step, debug, 1-chip, cheapo ICE
sounds good, but I am not sure you can do it very easily. In the 8088,
for instance, they have a trap flag in the status reg that enables the
built-in (INT 1) single-step capability of the chip. This is what allows
you to run Debug.com and Codeview at the same time as executing a
target program. As far as I know, the PIC chips don't have this ability.
And since the PICs don't have s.w. interrupts, I am not sure an ISR
would help.

Myke Predko, who devised his minimalist EMU, might know more about
this stuff.
---------

In my own designs, I never use ICEs or simulators. Otherwise, I would
have a closet full of them. My preferred method is to first write a
basic monitor routine, accessible via RS-232, and which can display the
cpu registers, and read the external pins at high rates, and then I
write the app on top of that as small callable modules, rather than
a long chain of goto's. I then do "post-mortems" after running newly-
coded routines. This usually works quite well for rapid debugging.
(I realize this sounds very primitive to the ICE set).

Although I've never done it before, it just occurred to me that some-
thing even better would be to stick a < call reg_capture > statement at
strategic places in the code. It would save the reg values to a circular
buffer. Then you could do post-mortems on execution "inside" the
routines, rather than only at the end.

These methods work quite well in PICs that have multi-level stacks,
and more than minimal internal RAM.

- Dan Michaels
Oricom Technologies
http://www.sni.net/~oricom

=================
=================
[snip]
>
>Do you think a single step program monitor could be implemented in an ISR?
>Could the '877 be a minimal ICD without the ICD?
>
>Would you or your company be interested in pulling the pieces together and
>making the board and / or PC software?
>
>James Newton, PICList Admin #3
>mailto:jamesnewton@piclist.com
>1-619-652-0593 phone
>http://www.piclist.com
>