>Yah .. I'm not convinced that we're really in any condition to deal with 
>most forms of loadable drivers at the moment.  This example is a shining one.

We're not.  The way FreeBSD does things is somewhat ``better'' in this
area, although I still don't like it all that much.

FreeBSD seems to follow this method:

	if in kernel, follow usual probe / attach rules.

	if lkm, the probe routine calls a function which will
	call dev_attach.  NetBSD is different enough in this
	area that we don't have this -- perhaps ours is called
	a bus specific name?  I've not looked into this.

It seems that there should simply be a way to call the device's usual
match/attach routines, let it decide if it should be attached, and do
so using the usual method.  That would require no changes to the
existing code or kernel.

Perhaps the LKM module for devices should take the addresses of the
match/attach routines, set up what appears to be a configuration data
structure, and call the match/attach?  I don't know, and I've not
poked at this much yet.

--Michael

--
Michael Graff <explorer@flame.org>        NetBSD is the way to go!
PGP key on a key-server near you!         Netshade the world!