On Thu, 19 Jul 2001, Bruce Lane wrote:

> 	Those facts are: I have had a number of cases of SPARC IPC's, IPX's,
> Classics, and even SPARC 5's where I've installed a non-Sun SCSI drive,
> had it work perfectly in the OS, and fail to respond properly to
> 'probe-scsi.'
>
> 	Perhaps my experiences are the result of firmware oddities. Whatever the
> case, the possibility that 'probe-scsi' may be failing on Jack's drive
> because it's not equipped with Sun firmware cannot be ignored.

Respectfully, I have to differ.

Very unfortunately, there is no special "Sun firmware" on SCSI devices
that makes a device respond to a probe-scsi.

My perhaps weak understanding of the way that probe-scsi works is
that it resets the SCSI bus, then marchs up the target numbers from
0 to 6.  There's a SCSI command in the SCSI standard that should
cause a device to return something like a 13-byte ASCII string
identifying the device. The probe-scsi sends that command to each
target in progression, and reports the 13-byte string it (may) get
back.

Regretably, I personally have never had a SCSI device work on any
SPARC-based machine that didn't respond to a probe-scsi.

My small and brief experience includes SS-1, SS-1+, IPC, SS-2 clones
(Tatung), SS-10 and SS-5.  The devices: Fujitsu, Connor, Seagate,
and HP disk drives, Exabyte 8mm tape drives, Magneto-optical drives,
8mm tape robots, MO robots from HP and others, and data-grade VHS
SCSI drives.

In my small experience, irregularities in probe-scsi (devices that
don't report back, devices that show up more than once, low-target
device reports, hi-target does not) mean that something's very
wrong, usually cabling, connectors or termination.

My humble advice: if you don't get probe-scsi information back,
fiddle around until you do.