Todd Whitesel <toddpw@best.com> writes:
> I think you are correct, although conceivably one day they might be required
> for quirk table purposes.

Hmm.  If it's desirable to make a 'generic' quirks table that needs to
differentiate on subsystem vendor ID and subsystem ID, then i think
that it should match all 4 IDs (product/vendor/subsystem/subsystem
vendor), as the puc driver currently does.


> It is also possible that some perverse hardware vendor will make their card
> extremely difficult, and the only method we have for distinguishing their
> hack is to check the subsystem ID.

There apparently exist cases where people have whacked others' chips
to do their bidding, and use subsystem/subsystem vendor to
differentiate.  In at least one case that i'm aware of, a 'normal'
driver for an existing relatively generic chip will match (based on
vendor/product ID), but not function correctly because, well, there's
a lot of different stuff.  8-)

In that specific case, if the generic driver is to do anything at all,
it should just reject if subsystem/subsystem vendor match the
'whacked' device's values.  Another reasonable approach is to just let
the driver for the 'whacked' device match 'better' than the generic
driver.


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