Brian D Chase writes:
[...]
> When I first starting inquiring about the VS3100 support for NetBSD, I had
> some discussions with Bertram about the SCSI/ST506 controllers since he
> didn't have access to one.  All the machines he's working with apparently

Nicely put! In this case "All" means "Exactly one 3100" ;-)

> have SCSI/SCSI controllers.  Sooooo it sort of looks like the present
> version of the VS3100 SCSI driver is expecting the dual SCSI controller 
> configuration -- if it sees one or none, it gets lost.
>
> Calling upon the assitance of Bertram... How difficult would it be to
> modify the SCSI driver to allow for configurations with one or zero SCSI
> controllers?  I'd be happy to assist in the testing process.

It's not that difficult. The old idea was to have a pre-configured list of
devices derived from the SID/SIDEX. Since I had/have access to 2000 and
3100/76 this sounded reasonably and thus there is no real probe() routine
yet, all devices in the list are just assumed to be there. In the meantime
these assumptions have been proved wrong :-/

This explains the problems with other models of 3100 where the device
configuration (esp. SCSI/ST506) is not fix but variable.

I think the easiest solution would be to add a real probe() routine
which checks for the existence of SCSI and ST506 controllers at the
same addresses for all models. I hope to have some spare time at the
end of this week, then I'll look for that. This won't cure the SCSI
and DMA problems with other models of 3100, but at least it should 
help net-booting most of the 3100s...

Ciao,
	bertram