On Tue, 18 Dec 2001, Mike Pumford wrote:

> > [ Sent here since we don't have a port-acorn32 mailing list yet ]
> >
> > Anyone trying to run NetBSD/acorn32 on an Acorn NC is likely to find the
> > work I've done over the weekend useful.  Specifically, I've modified the
> > "ne at podulebus" driver to work around a bug in the MX98905 chip used on
> > EtherI and EtherN cards, so they can now be used with NetBSD.]
>
> What bug is this?

At least some MX98905s seem to ignore RBCR1, so transfers of more that 255
bytes to/from the CPU fail.  I found a description of the bug in
<URL:http://mail-index.netbsd.org/port-arm32/1996/06/23/0005.html>.

> I didn't notice any problems when I did the original
> development work.

What chips do your cards have on them?

> On a related note my current acorn32 kernel detects my
> Etherlan 602 card as an NE1000 rather than an NE2000.

Is this a consequence of my recent fixes, do you know?  Putting
if_ne_pbus* back to a week ago should give you something self-consistent
to compare against.

> Do you know why this is or should I do some digging? My first guess
> is something has gone bad in the 16bit bus_space routines.

I've no idea.  I've changed the if_media routines for E600 cards (so they
now report whether they're getting link beat), and generally made the code
use #defined constants, but that shouldn't have broken anything.  I didn't
enable the MX98905 workaround on the E600 since I only had indirect
evidence that it was needed there.

-- 
Ben Harris                                                   <bjh21@netbsd.org>
Portmaster, NetBSD/arm26               <URL:http://www.netbsd.org/Ports/arm26/>