> Basicly, Apple has a timeout routine, and in the timeout routine
> they check if it is wedged and they restart.

:-)  Thanks.  Instead of figuring out the cause, they
beat the device up to recover.  Well, I have a little more
experience with this now.  I does not seem to actually hang.
It just becomes confused.  My patch will let the driver
adjust to the confusion without dropping packets unnecessarily.
It might be useful to add a timeout to reinitialize like
Apple does, but at this point I don't think it necessary.
If someone with write privs would apply my patch to the
active branches, I would appreciate (Matt T.?).  Meanwhile,
I applied them to the sources where we work.

Cheers & Thanks,
	Bruce