Subject: problems with scsi_test kernel on 3/75
To: None <dall@HFRD.DSTO.GOV.AU>
From: Gordon W. Ross <gwr@mc.com>
List: port-sun3
Date: 11/15/1995 18:37:10
> Date: Thu, 16 Nov 1995 09:16:29 +1030
> From: Ian Dall <dall@HFRD.DSTO.GOV.AU>

> Jason Thorpe <thorpej@SJ.Xenotropic.COM> writes:
> 
>   > The Sun 3 kernels by default use the old SunOS mapping:
> 
>   > sd0 at scsibus? targ 0 lun 0
>   > sd1 at scsibus? targ 0 lun 1
>   > sd2 at scsibus? targ 1 lun 0
>   > sd3 at scsibus? targ 1 lun 1
>   > [ . . . ]

Note that only those shown above are "hard coded", so any devices
with target/LUN not matching the above will be assigned to sd4,5,...
due to the "wild card" entry that follows:

      sd* at scsibus? target ? lun ?

This will probably break the way "boot -a" works if you boot from a
SCSI disk that does not match any of the above.  I figured people
could arrange for their boot disk to be one of those.  If not...

> Which raises a question - how many disks can I support without a kernel
> rebuild? I currently am using SunOS with root on disk 10 (scsi id 5)
> for reasons you don't want to know about. Can I boot (perhaps with
> -a) without recompiling. For SunOS I had to rebuild to support
> the extra disks (I think it only does 2 by default) and also made
> disk 10 the root disk. I have been lurking on this mailing list
> for some time and am considering taking the NetBSD plunge for my
> Sun 3/50.
> 
> Ian
> 

...you will need to make the scsi config lines something like this:

# Sun traditional device names...
sd0 at scsibus? target 0 lun 0
sd1 at scsibus? target 0 lun 1
sd2 at scsibus? target 1 lun 0
sd3 at scsibus? target 1 lun 1
sd10 at scsibus? target 5 lun 0
# Anything else...
sd* at scsibus? target ? lun ?


Gordon