On Mon, 17 Aug 1998 13:44:59 +0100 
 Richard Earnshaw <rearnsha@arm.com> wrote:

 > Thanks, I compiled a kernel from a sup of your changes, but no dice.  It's 
 > getting closer, but the interface still fails to get installed because 
 > ofisa_reg_get is only returning one register (for the IO address), so the 
 > memory region does not get configured.  This causes the cs8900 code to 
 > fail because it currently doesn't support this mode of operation.

Sigh.  Ok, I will fix this RSN.

 > The old version of the driver used to find the memory region by other 
 > nefarious means (sorry, haven't worked out how yet -- unix device probing 
 > is still a bit new to me, and trying to get logs of a NC with no disk 
 > trying to boot over the interface it can't find is a bit of a trial) so I 
 > think we need a cs_ofisa_md_reg_fixup to fake up the missing register.

The old driver used to hard-code the information, ignoring what the
firmware told it.  That is wrong for a generic driver (the ofisa attachment
is independent of Shark'ness...)

I'll make the appropriate reg fixup hack today.

Jason R. Thorpe                                       thorpej@nas.nasa.gov
NASA Ames Research Center                            Home: +1 408 866 1912
NAS: M/S 258-5                                       Work: +1 650 604 0935
Moffett Field, CA 94035                             Pager: +1 650 940 5942