🚀 go-pugleaf

Hi.

I'm new to device driver programming, so excuse me if my
terminology is not 100% correct.

I have taken over a GP Fault device driver written by someone
else. It almost always works ok.

When in 386 enhanced mode in windows 3.1, the driver is hooked in with 
Hook_PM_... and defined to handle interrupt "0dh".

Normally, when I get a GP fault, my handler kicks in and throws my
application (lisp) into the debugger, instead of windows putting up
the "Do you want to close" screen.

I have a repeatable case though, where I'm not sure why its GP faulting,
with the following behavior:
   1) If I don't install my handler, there is no GP fault (at least there 
      is no "Do you want to close" screen and CVW doesn't detect the fault).
   2) If my handler handles the fault, I get thrown into my debugger
      in an almost-dead state.
   3) If my handler declines the fault by chaining to the next handler
      in the chain, there is no GP fault (at least there is no "Do you
      want to close" screen and CVW doesn't detect the fault.

Is it possible that there are some classes (error codes?) of GP faults
that my device driver is supposed to know to ignore?

Is there any good way to debug this (I've never used WDEB386 or WINICE,
but I could learn them if necessary)?

This is a real bitch to debug because the error only happens when I'm
set up so that CVW won't catch the error :-(

Thanks in advance for any help or ideas.

Jeff Eisen
jne@ai.mit.edu