> >  * The "ne" driver should be able to use the vax SGEC "ze" without
> >    hopefully too much work.
> > 
> The SGEC should already be MI. I have put the code in dev/ic/sgec*.

When I ported an allegedly-MI Ultrix SGEC driver from MIPS to VAX many
years ago I discovered that there were some obscure machine dependencies.
I have forgotten details, but I recall that where there was a write
buffer flush on MIPS something also needed to be done on VAX even though
the SOC chip I was using didn't have a store buffer and didn't reorder
memory accesses.  So I can't offer any specific advice, but don't be too
optimistic until you see it work.

    --John Carr (jfc@mit.edu)