STEWART_R_WYATT at HP-Boise-om2.om.hp.com
STEWART_R_WYATT at HP-Boise-om2.om.hp.com
Thu Sep 17 13:27:54 PDT 1998
* From the T10 (formerly SCSI) Reflector (t10 at symbios.com), posted by:
* STEWART_R_WYATT at HP-Boise-om2.om.hp.com
Tape Profile Meeting minutes,
St. Petersburg Beach, Florida T11/98-473v0
September 15, 1998
1. Introductions: Group
Dale LaFollette, StorageTek, called the group to order and began the
meeting at 9AM. He had the participants introduce themselves.
2. Approval of this Agenda: T11/98-469v0 Group.
Changed Dave Baldwin to Dave Peterson on item 7. The FC-Tape draft is
124vA instead of 124v10. Added a discussion about CRNs. The minutes
were approved as modified.
3. Approval of 8/12 Minutes: T11/98-413v0 Stewart Wyatt
4. Old Action Items: Stewart Wyatt, HP
#1 Dave Baldwin, Emulex, was to revise the ABTS proposal, obtain a
document number and post the updated proposal: Dave Baldwin was not
able to attend, but Dave Peterson, StorageTek, came prepared with the
updated proposal. Agenda item 6. Ongoing.
#2 Brian Smith, Crossroads, to review Media Changer commands: Brian
was not in attendance, but Neal Wanamaker, Crossroads, reported that
Brian did not find anything to report.
#3 Stewart Wyatt to post a copy of the TapeAlert proposal: Completed
the document number is T11/98-412v0.
#4 Dave Peterson to include TapeAlert in the next version of the SSC:
#5 Stephen Gold, HP, to post a comment for inclusion of TapeAlert in
the SMC: Completed by Stewart Wyatt, HP.
#6 Participants to review the public/private loop addressing issue.
Agenda item 5. Ongoing.
#7 Rich Taborek/Dal Allan to review what letter ballots are required
before going for public review: Dal and Rich were both in attendance.
Dal reported that the initial letter ballot would be through T11
instead of T11.3 since the T11 ballot is required. Completed.
5. Public/Private Addressing: Jim Coomes, Seagate
Jim Coomes reviewed the issue of having public devices respond to an
alias of 00 00 AL_PA as well as their public address of DD AA AL_PA
during PLOGI for the benefit of the private devices that are not aware
of the public address. Afterwards the private device shall use the DD
AA AL_PA to address the public device that is returned in the S_ID of
Bob Snively, SUN, questioned this decision. His preference was that
the public devices not respond to private devices. But given this
situation he preferred to see the public devices continue to respond
to the 00 00 AL_PA. Dal Allan, ENDL, and Bob debated the issue in
detail. Jim Coomes reported that Bill Martin, Gaddzoox, had responded
to his posting, requesting that devices have only have one address to
respond to after login. Dal agreed as did others in the room.
Bob felt like this was requiring a change in private port behavior.
Dal and Horst Truestedt, ENDL, argued that the address field was
always a 24 bit field. Ed Gardner, Ophidian Designs, argued that this
change belongs in the PLDA or the FLA. Jim countered that those
documents are closed and done. Jim pointed out that this issue has
been publicized on the reflector for months and does not appear to
break any existing implementations.
An extension requested by Dave Baldwin was to allow public devices to
send PLOGI to all devices on the same loop with a D_ID of 00 00 AL_PA.
The public devices will use their DD AA AL_PA address as the S_ID in
the ACC. This extension would allow one process for login. Charles
Binford, Symbios, pointed out that after accepting the original
behavior, it would make the implementation easier to also accept the
Ed Gardiner suggests that this behavior may need to be generalized to
Jim agreed to look at the PLDA, FLA and FC-Tape documents to see if
other changes are required. He will update his proposal and post it to
Bob brought up additional discussion followed about the effect of
"Stealth Mode" and the effect of remote initiators logging in with
Jim Coomes saw the private/public addressing issue as a short term
need as he expects all devices to be public soon. Dal Allan assumed
that there would be along term need for private loop devices. He
suggested that someone needed to define a way (mode page bit) to have
the device remain private. Jim thought that was unnecessary. He said
that Seagate has had no requests for private only behavior from their
6. ABTS Enhancement Dave Peterson for Dave Baldwin
Ed asked why ABTS came to abort an exchange instead of a sequence. Bob
Snively explained that it came out of the FCSI development.
This proposal allows either the sequence initiator or recipient to
issue the ABTS. In class 2 the sequence can be aborted with the ACK.
This change was thought necessary for class 3. Some examples were
given that suggested that this function may be useful in class 2 such
as when an ACK is not returned to the target.
A discussion followed about the relevance of a sequence recipient
sending a ABTS. Since the ABTS doesn't have a payload there was a
question about how the sequence to be aborted is specified. Bob
suggested eliminating the last paragraph. This would allow setting the
bits to select aborting the exchange or sequence and eliminate the
proposed change allowing both recipients and initiator to send a ABTS.
No need for this functionality was identified. It will be dropped
until a need is identified.
Another discussion followed between Bob and Dal about how an initiator
can signal to a target an error in a long read and whether that need
7. FCP Model Enhancement: Dave Peterson
The enhancement is Annex C in Rev 1.09 of the profile. Dave had posted
the annex to the reflector independently.
Ed would like to request the FCP_CONF by NOT making the FCP_RSP not
the last frame of the sequence. After some discussion, he decided that
this would preclude linked commands.
Dave was reminded that this text must be normative and was suggested
to add a note that it will be replaced by FCP2.
Dal Allan was concerned about the text that allowed the target to
assume that the FCP_CONF for a given exchange was lost and proceed
without error recovery if it receives a command with the same OX_ID.
After some discussion, Dal Allan and Matthew Wakeley, HP, felt that
the target should not accept the new exchange. (Reject in class 2 or
drop the frame in class 3.) The last sentence of the proposal was
changed to read that..."the target shall follow the rules as defined
in the FC-PH in regards to duplicate exchange management".
Dave presented an overhead with the FCP_CONF model included in the
posting and the last draft which he had already started marking up.
The group proposed a number of other changes. Bob Snively wanted to
remove the reference to dual porting which Rob Basham, IBM, had
requested. He also questioned the value of item b and its paragraph
which suggested that FCP_CONF is needed for initialization. No one
present could see why this was an exceptional requirement. Rob was not
present to defend his position.
8. Required SCSI Commands Cont. T11/98-375v1 Dale Lafollette
Doug Hagerman, Digital now Compaq, sent a memo to Dave LaFollette
requesting that the SCSI Tape Device Commands follow the PLDA pattern
which includes an "allowable" category. Bob Snively has argued against
using allowable as he believes it defeats interoperability. A long
discussion followed about the meanings of the various words and the
effect of implementing a "prohibited" feature. The decision was to use
a single column for the SCSI Tape Commands table for both hosts and
Only two words will be used to describe a commands support in the
profile. The first is Invokable which is defined to mean that the
Initiator is permitted to execute and target is required to support
the command. The second is Prohibited using the FLA definition which
is currently in the profile.
The group reviewed the commands looking at which fields should be
supported. The results will be included in the next revision of the
8a CRN Review
Charles Binford noted that Annex D states that a LUN Reset Task
Management function will reset the CRN to 1. He asked how the LUN
reset is performed and did not get an answer as no one seemed to know.
He also thought this should be included in the clearing effects table.
9. FC-Tape Draft Review: T11/98-124vA
Charles Binford asked about the FCP_DL and residuals. After some
discussion it was decided that when executing a write command to
calculate the transfer length first and if it doesn't match the FCP_DL
length to reject the command. Eric Oetting noted that the behavior
required for residuals is described in the SSC read and write command
descriptions. Dave Peterson suggested requiring that the FCP_DL equal
the transfer length.
10. T10 New Business: Group
11. T11 New Business: Group
New schedule for FC-Tape:
Sept. 24: FC-Tape draft completed. Reflector posting stating our
intention to request a letter ballot through T11.
Oct 7: FC-Tape working group resolves any new issues with Sept. 24th
Oct 8: Request a letter ballot for the Oct 7th version of the draft
unless the 2 week rule is called. If it is, then request a letter
ballot on the Sept. 24th version.
Letter ballot starts some days latter.
30 days after the letter ballot closes, assumed to be after November
T10 meetings, the editor starts reviewing letter ballot comments.
Dec 16 T11 working group meeting reviews our editors comments on the
letter ballot comments
12. Review New Action Items: Stewart Wyatt
#1 Jim Coomes agreed to look at the PLDA, FLA and FC-Tape documents to
see if other changes are required to support the proposed changes in
public/private addressing. Jim will update proposal and post it to the
#2 Dave Peterson to get a document number for the ABTS proposal,
update it and upload it to the web site
#3 Dale LaFollette will complete the SMC command and parameter review
for input to the document.
#3 Dave Peterson will update FC-Tape with this meetings comments by
#4 Dale LaFollette will post the FC_Tape schedule and a request for
review in preparation for the letter ballot when Dave completes the
13. Adjournment: Group 5 PM
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at symbios.com
More information about the T10