PERFORCE change 95068 for review

[M. Warner Losh]

In message: <014B1689-1A9A-4AC3-BE76-AF8E35FF0A9C at xcllnt.net>
            Marcel Moolenaar <marcel at xcllnt.net> writes:
: 
: On Apr 12, 2006, at 9:02 AM, John Baldwin wrote:
: 
: > On Wednesday 12 April 2006 10:50, Marcel Moolenaar wrote:
: >> http://perforce.freebsd.org/chv.cgi?CH=95068
: >>
: >> Change 95068 by marcel at marcel_nfs on 2006/04/12 14:49:30
: >>
: >> 	Replace the rmask and rval arrays with vendor, device, subvendor
: >> 	and subdevice variables. When subvendor is 0xffff, don't test the
: >> 	subvendor and subdevice values. This strips 24 bytes from the
: >> 	description.
: >> 	While here, sort the list on vendor, device, subvendor and
: >> 	subdevice.
: >
: > I guess you expanded any entries that were using masks other than
: > 0xffff?  (I think there were only a handful if any).
: 
: Not yet. I looked at the latest pucdata.c in NetBSD and more entries
: with masks other than 0xffff have been added. While the handful we
: had looked like mistakes (was 0xfffc, should probably be 0xfffe), the
: new ones don't. I need to investigate, because if the mask is used
: as a quick and dirty way to add multiple boards with a single entry
: then I'm not sure it's a good thing. For different boards, different
: descriptions are better. In that case I will expand. Otherwise I may
: end up bringing back the mask.
: 
: The entries in question have a /* MASK */ comment as a reminder.

You might want to take a look at some of what I did in my newcard
tree, since I reduce the data quite a bit.  I think I may have
committed it to head, but I might have forgotten.

The mask was indeed for multiple boards to make the table smaller, but
it really made it larger.

Warner