System: Pentium 150 running NetBSD/i386 1.2, Adaptec 2940 PCI
controller, internal Seagate 2.1 Gb fixed disk, external Iomega Jaz
drive.

I've picked up an Iomega Jaz drive (SCSI, external) and played with it
a little.  I'm able to label the disk and write filesystems to it, so
it's working fine for most of what I plan to do with it.  Whenever I
access the disk for the first time, though, I get a console message
like this:

sd1(ahc1:4:0): illegal request, data = 35 82 01 3c ce 00 00 00 00 00
00 00 00 00 00 00 00
sd1: could not mode sense (4); using fictitious geometry

This happens when I run "disklabel -r" on the drive after I've
inserted the disk but haven't mounted anything, and it also happens
with the first filesystem mounted from the Jaz disk.  That is, if I
mount two filesystems in a row from the Jaz, the first one triggers
this error and the second doesn't.  If I unmount everything from the
Jaz disk and try again, the first filesystem mounted gives me the
error again, and the second doesn't.  These errors don't affect the
drive's function when I'm using it as a secondary device and running
from my Seagate fixed disk.

Stranger, though, is the system's behaviour when I try to boot from
the Jaz.  My system is definitely recognizing the disk, and loading a
kernel from it, but then, immediately after "changing root device to
sd0a", I get three repetitions of the above "illegal request" message
and then the machine hangs.  (By "repetitions", I mean that the
general gist of the messages is the same; the values in the "data"
field change between occurrences.)  I get this behaviour whether I
boot straight from the Jaz drive or use a bootblock from a floppy disk
and redirect to the Jaz drive at the boot prompt.

(In case anyone's wondering about the discrepancy between "sd1" and
"sd0" in the above descriptions, I'm remapping the Jaz drive from SCSI
ID 4 to 2 and back again in order to try booting from it.  My internal
disk is SDSI ID 3.)

Not knowing much of kernel internals, it looks to me like the Jaz
drive and the SCSI subsystem are disagreeing about something, but I'm
not sure what.  Can anyone suggest fixes or experiments to gather
better information on the problem?  Though I don't need to be able to
boot from this disk, I'd like to have that ability.  I'd appreciate
any suggestions anyone can offer.

Keith Browne
tuxedo@interlog.com