You write:
> Hmmm.  Let's try doing what we did before, send me the info on the drive
> (I *might* be able to find it around here...), and I'll recompile the
> kernel so it thinks that's a disk drive.  Last time we tried that, it
> was real confused about the geometry info, I think, but we'll start from
> there...  The docs you have should help.  Is it a SCSI-2 or SCSI-1
> device?

According to SCSIProbe, the vendor string is "EPSON", the product
string is "OMD-5010", and the version string is "3.08" (ignoring
trailing spaces).  It's a SCSI-2 device, but there's a jumper on the
back that can make it pretend it's a SCSI-1 device.  I'd rather not
have to change that for MacBSD, though.

I remember the confusion about the geometry info.  I have done some raw
SCSI work before, interfacing to Mac tape drives.  What command is used
to get the geometry info?  Inquiry (12H)?  What info is expected to be
there?  If you want to mail me the relevant sources, I can take a look
at the code that interprets the reply and see if it matches the info
the drive returns.

> > figured out that the root filesystem wasn't on /dev/sd0a anymore.
> > (It's now on /dev/sd1a, presumably because my MacBSD drive is at SCSI
> > ID 1.  Probably a good idea, but knowing about it before it bit me
> > would have made my life easier.  ;-)
> Hmmm.  I thought I did post that when I announced that kernel.  If I
> think about it, I'll put a readme near it...

I think you said something about partitions moving around within each
SCSI ID, but I don't remember anything about the SCSI ID's being treated
differently.

- monroe

------------------------------------------------------------------------------