joda@pdc.kth.se (Johan Danielsson) writes:
> cgd@netbsd.org (Chris G. Demetriou) writes:
> > is _NOT_ an adequate test to determine if a card works in the way
> > you're expecting it to work, at least for the PCI equivalent
> > class/subclass.
> 
> It would be possible to make a puc-like device for cardbus, but I'd
> rather wait until we know that there are devices that really require
> it.

i think you miss the point:

you've created a driver which blindly matches communications/serial as
a simple 'com' port.  you do some additional checking in the attach
routine, to be sure, but that's beside the point.

unless you have some documentation that says for cardbus that it'd OK,
matching that is bogus.  communications/modem, at least according to
the PCI spec, can (for some programming interface types) reasonably be
expected to work the way your driver expects.  communications/serial
CANNOT (unless you have some documentation to the contrary).  If you
want to do a table-based match (i.e. return 0 unless you manage to
look up the device from your table) that'd be reasonable, but it's not
right to just match communications/serial without any further
information (or documentation that indicates that for CardBus it's a
standardized class/subclass).

that's why i asked if you have some documentation that indicates that
what you're doing here is reasonable.  absent that, it's broken.


cgd
-- 
Chris Demetriou - cgd@netbsd.org - http://www.netbsd.org/People/Pages/cgd.html
Disclaimer: Not speaking for NetBSD, just expressing my own opinion.