Questions about RESERVATION CONFLICT

Gerry_Houlder at notes.seagate.com Gerry_Houlder at notes.seagate.com
Wed Oct 27 07:39:54 PDT 1999


* From the T10 Reflector (t10 at t10.org), posted by:
* Gerry_Houlder at notes.seagate.com
*
I didn't see any other replies, so I'll take a stab at answering the questions.

>1) Why does MODE SENSE encounter a RESERVATION CONFLICT from the
>non-reservation holding initiator?   I would have thought that this
>is simply a "question" and not a "directive", on a par with INQUIRY
>and so should not be blocked.

A1. The reservation conflict rules are a carryover from the early days of SCSI.
MODE SENSE wasn't considered important enough to break through a reservation
conflict. At that time, a Mode Sense often used the same data buffer space as
read and write data, so a target may have to purge cached data to make room for
the mode data. This disturbance of one initiator's work by another is intended
to be prevented by reservations.

>2) Why is SET LIMITS (for SBC) not subject to RESERVATION CONFLICT?
>I would think that this would cause a conflict since any subsequent
>command within the SET LIMITS link set will fail with RESERVATION
>CONFLICT.

A2. This looks like a mistake to me. When I look up SBC document, Set limits(10)
and (12) state that they are not allowed through reservations. The table in
Annex B of SPC-2 states that these commands are allowed through reservations.
This is a serious contradiction. I presume whoever put together the table in
Annex B thought it was reasonable to let set limits through a reservation even
though SBC requires reservation conflict in its wording. I presume we haven't
encountered this problem sooner because very few targets implement this command.


*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org






More information about the T10 mailing list