At 06:39 PM 3/6/00 , Anders Magnusson wrote:
> > Reads are now working (a RZ23L gets about 730KB/s).  
> > Writes seems to never complete.
> > 
>Wow, congrats!

Most writes complete but some don't.

> > I had to add a vmapbuf/vunmapbuf routines but hopefully
> > they won't effect the other disk drivers too much.
> >
>Actually they do... Removing them gave ~200k/s extra speed while
>doing raw disk I/O on a VS3100/m30. Also adding them may break the
>raw disk DMA on the NCR5380.
>It would be nice if they could be avoided. There is actually no reason
>to use those routines at all on a vax; the only situation they are 
>needed are on archs where a) all data must be copied and b) the whole
>memory isn't accessible directly. On those archs it could be solved in
>a different way anyway nowadays.

The problem is that the ncr53c9x MI code doesn't pass a ptr to the proc
so I can't pass it to bus_dmamap_load.  I didn't feel like changing the 
MI interface.
-- 
Matt Thomas               Internet:   matt@3am-software.com
3am Software Foundry      WWW URL:    http://www.3am-software.com/bio/matt/
Cupertino, CA             Disclaimer: I avow all knowledge of this message