WORRIES ABOUT OPTIONAL READ(6) AND RESERVE(6)

GARDNER at acm.org GARDNER at acm.org
Sun Aug 18 14:23:44 PDT 1996


* From the SCSI Reflector, posted by:
* GARDNER at ACM.ORG
*

Giles,

The intent (certainly my intent) of voting to make read(6)
optional/obsolete was to flag to host driver/adapter developers that
they should no longer use it in new products.  Stories of systems
going out of their way to choose read(6) vs. read(10) only reinforce
the desire to require read(10) support (already done in SCSI-2) and
encourage host driver and adapter developers to always use read(10).

Of course, nothing prohibits a disk from implementing read(6), and any
device that wishes to claim SCSI-2 compliance will do so.  For the
near future all drives will (de facto) need to claim SCSI-2 compliance
to be marketable.  They'll work fine with older hosts that support
SCSI-1 or SCSI-2.

It's only new hosts that claim SCSI-3 compliance that are affected.
Such a host must not issue a read(6), at least not without checking
the SCSI version in the Inquiry data.  More likely any presently
shipping host already requires SCSI-2, and certainly new hosts should.
If so, the host should use read(10) always instead of read(6).  The
only reason to use read(6) is to support SCSI-1 disk drives.  When was
the last time you saw a SCSI-1 disk offered for sale, that didn't also
support SCSI-2?  I'd be surprised if you could find one still in use,
let alone being sold in new products.  (We're discussing disk drives,
not other devices).

We're not all that far away from the day that all shipping SCSI disk
drives will be larger than read(6) can address.  Continuing to require
its support, or expecting such drives to work on older SCSI-1 host
systems, gets less reasonable every day.

Edward A. Gardner                  gardner at acm.org
Ophidian Designs                   719 593-8866 voice
1262 Hofstead Terrace              719 593-8989 fax
Colorado Springs, CO  80907




More information about the T10 mailing list