>>I think it's also worth looking at what BSDI did; I don't think it's
>>completely correct but it had certain useful behaivors..  for
>>instance, popping a card and reinserting it doesn't require you to
>>re-ifconfig the interface.
>Hmmmm .... is this a useful semantic to preserve?  I could see it
>getting rather complicated trying to do this in practice.

	Although I do not believe there's sole "ultimate" solution for pcmcia
	device model, I still believe BSDI's approach is good.

	Traditionally UNIX assumes there will be no changes made to device
	attach/detach during its runtime.  Because of it UNIX do not really
	have a good way to notify of changes to the userland programs, and
	not many userland programs are designed to be adaptable to those
	changes.  I think that we may need to hide those dynamic changes to
	network devices (interface removal, pcmcia slot power down and
	other things) from the userland programs to "preserve" traditional,
	not-too-many-changes-to-list-of-devices view from uesrland.

	if_detach and ifnet-refcnt is good approach from kernel designer's
	standpoint, but they do not really help existing userland codebase.
	Most of them assume that list of network interface will not change
	after it starts
	(you are free to argue that they are not well written, but this is
	the way it was for a long years, until notebook PCs are around)

itojun