> > > > Any further progress on the DMA SCSI code?
[...]
> > How difficult would DMA support be to code? Could you give me some hints
> > how this could be done? I think the NCR chips are quite well documented,
> > aren't they? Or I could grab some code from another platform of NetBSD or
> > Linux or something...
> > 
> The only problem with it is that there needs to be some smart-locking
> so that 5380 registers on any controller is touched while a DMA transfer
> is on-going. This isn't actually a big problem, but it hasn't been 
> fixed yet.

I'd love to offer help on this.  Is it a matter of "we just haven't
had time to fix it yet."?  Or is there some dark voodoo that hasn't
been figured out?

If you could point us at the specific area of code that needs
attention, I'd love to take a look at it.  I'll grab -current this
weekend and put it on my MicroVAX 3100/38 to play with it.  Hopefully I
can get a kernel compile done before Monday (though I don't know how
long it will realisticly take, I've never compiled a kernel on a one
of my VAXen before).  I'll also need to look up documentation on
the 5380, but that's not hard to find.

My main "production" MicroVAX is running Ultrix right now, and I'm
increasingly fed up with it.  The _only_ thing holding me back from
switching to NetBSD is lack of DMA SCSI.  We'll be glad to fix it,
honest we will!

> -- Ragge

-Seth

(P.S. -- I've never properly said THANKS!! for all the hard work you've
put into this project, Ragge.  Your efforts are truly appreciated by
many, many people.)