Ultrasparc 3 support

[Jason George]

>What would be interesting from my point of view is to see how well the
>OpenBSD code works.  I've heard a few reports that it is unusable, and a
>few reports that it sort of works.  Knowing whether you can use DMA
>under OpenBSD may be useful as it would give us another code base to
>gain information from.

I supplied the original pile of Blade 1000 and 2000 machines to Theo and the 
2006 OpenBSD hackthon.

You should take an OpenBSD sparc64 snapshot and try it.  It is extremely 
usable and stable.  Mark Kettenis has made huge leaps in working around the 
myriad of bugs in the processor and associated glue logic.  Ultrasparc III 
running OpenBSD is solid and keeps getting faster.

Originally, the initial patches and commits had extensively IFDEF work to 
ensure that earlier versions of Ultrasparc wouldn't break.  The code has been 
fully integrated into the tree for a number of months.

Also, a few days ago Mark committed a driver for the Cassini network 
controller chips.

--Jason