plunky@rya-online.net said:
> I think the solution (long-term) is that since the DeviceClass says
> IN_INTERFACE, if_aue should be looking at the interface class for the
> information and probably attaching by interface, also.. 

If device attachments and interface attachments get separated,
that would be just another cfattach.

If we take the specs seriously, we should try to match
interfaces only if the device class code is zero. The spec
says that a nonzero class means that the interfaces are
not independant. Windows inserts some "compound" or so
pseudo-device if there are multiple interfaces. In the
long term we might do so as well, but that would affect
all kernel config files, so I wouldn't it before everything
else is cleaned up and working.

best regards
Matthias