At the end of the last meeting I agreed to raise the following
issue on the reflector...
One of the areas still requiring clarification is the mapping
of transport and target transport clients into SBP-2 Units and
Logical Units. This is an area where the SCSI roots of SBP-2
have really impacted our efforts.
- Looking at the draft SCSI-3 ARCHITECTURE MODEL -2 (SAM-2) rev
0.3, I find that "Each SCSI-3 protocol standard shall specify
the response to a target reset event including the conditions
under which a target hard reset shall be executed.
- Looking at SBP-2, rev 4, we find TARGET RESET affects all
logged-in initiators (to any logical unit within the Unit).
Some questions which need to be answered are:
a) How is TARGET RESET used by drivers n commercial operating
systems? (Both non-embedded and embedded?)
b) Can we define how broadly (or narrowly) the TARGET RESET event
affects a device using our transport?
If so, how should it be defined?
If not, does this mean that we must use one Unit directory
for each possible instance of a service?
Ideas, anyone?
-- Greg Shue Hewlett-Packard Company All-in-One Division gregs@sdd.hp.com ----------------------------------------------------------------