Our APM code doesn't do any direct manipulation of the display,
especially when the kernel is stopped at a breakpoint.
My best guess as to the cause is that your machine's BIOS has given up
on cooperative management.  In typical operation, the OS polls the BIOS
every second to see if there are any events to handle, such as battery
alarms, inactivity timeouts, etc.  I suspect your machine went idle, the
BIOS tried to tell the OS, and when the OS didn't listen it gave up and
turned off devices itself.  However, I would then expect the BIOS to
wake up the machine again when activity resumes (e.g., keyboard input)

Solutions?  Hmm.  Perhaps the debugger entry code could disable power
saving, but that doesn't seem very "pure."  If you're expecting to do
debugging, disabling power management in the BIOS and not configuring
APM into the kernel may be a reasonable choice.

-- 
==John Kohl <jtk@kolvir.arlington.ma.us>, <john_kohl@alum.mit.edu>
Happiness is having a song dedicated to you at a show :)
Home page: <URL:http://jtk.ne.mediaone.net/~jtk/>