>>>>> "Charles" == mycroft  <mycroft@gnu.ai.mit.edu> writes:

Charles> Actually, the com probe sequence is a bit braindead.  It
Charles> would be better to tickle the chip and see if it interrupts.
Charles> I'm not sure if there's a general way to do that without
Charles> outputting a byte, though.  (Loopback mode doesn't work on
Charles> some clones.)

After speaking to several people I've got the conclusion that the ISA
bus *cannot* guarantee a 0xff value when reading an unexistant port.
Apparently, some busses keep a voltage in the middle of a logical zero
and one (for a fast switch time), some busses are pulled up, some are
pulled down.  Do you have the 16450/550 specs?  If we not do the tickle
probe, we might maybe go for a better kind of passive probe.  Anyway,
is it so bad to output something during probe time?  A break signal
maybe?  I understand this is low priority, but maybe sometime...

Niklas