Draft minutes of SAS Protocol WG -- May 5-6, 2003
John Lohmeyer
lohmeyer at t10.org
Wed May 7 08:03:40 PDT 2003
* From the T10 Reflector (t10 at t10.org), posted by:
* John Lohmeyer <lohmeyer at t10.org>
*
Accredited Standards Committee*
InterNational Committee for Information Technology Standards (INCITS)
Doc. No.: T10/03-177r0
Date: May 7, 2003
Reply to: John Lohmeyer
To: T10 Membership
From: Ralph Weber and John Lohmeyer
Subject: SAS Protocol WG -- May 5-6, 2003
Nashua, NH
Agenda
1. Opening Remarks
2. Approval of Agenda
3. Attendance and Membership
4. Old Business
4.1 SAS Working Draft Review (SAS) [Elliott]
4.1.1 BREAK message upstream (Elliott)
4.1.2 DATA frame before XFER_RDY (Evans)
4.1.3 DONE clarifications (Elliott)
4.1.4 Set Rate (Elliott)
4.1.5 Unrecognized Target Port Transfer Tags
4.1.6 I_T Nexus Loss timer [Coomes]
4.2 SAS Revision 3 T10 letter ballot comment resolution (03-055r6)
[Elliott]
4.3 SAS-1.1: CRC on Frame Header [Roberts]
4.4 SAS-1.1: Transport Layer Retries [Jones]
4.5 SAS Requests and Confirmations (03-023r0) [Evans]
5. New Business
5.1 SAS 1.1: SATA II Port Selector Interoperability (03-155r0) [Sheffield]
5.2 SAS: Initiator Response Timeout timer (03-164r3) [Penokie]
5.3 SAS: SATA information to Annex (03-168r0) [Penokie]
5.4 Initiator Connection Tag Changes Detected by Target [Krantz]
6. Review of Recommendations
7. Meeting Schedule
8. Adjournment
Results of Meeting
1. Opening Remarks
John Lohmeyer called the meeting to order at 9:00 a.m. Monday, May 5, 2003. He
thanked Zane Daggett of Hitachi Cable Manchester for hosting the meeting. As
usual, the people present introduced themselves.
2. Approval of Agenda
The draft agenda was approved with the following changes:
4.1.1 BREAK message upstream (Elliott)
4.1.2 DATA frame before XFER_RDY (Evans)
4.1.3 DONE clarifications (Elliott)
4.1.4 Set Rate (Elliott)
The following items was added or revised during the course of the meeting:
4.1.5 Unrecognized Target Port Transfer Tags
4.1.6 I_T Nexus Loss timer [Coomes]
5.1 SAS 1.1: SATA II Port Selector Interoperability (03-155r0) [Sheffield]
5.4 Initiator Connection Tag Changes Detected by Target [Krantz]
3. Attendance and Membership
Attendance at working group meetings does not count toward minimum attendance
requirements for T10 membership. Working group meetings are open to any person
or organization directly and materially affected by T10's scope of work. The
following people attended the meeting:
Name S Organization Electronic Mail Address
---------------------- -- ------------------------- -------------------------
Mr. Ron Roberts P Adaptec, Inc. rkroberts at aol.com
Ms. Pat Thaler V Agilent Technologies pat_thaler at agilent.com
Mr. John Tyndall A Crossroads Systems, Inc. jtyndall at crossroads.com
Mr. Titkwan Hui A Dallas Semiconductor tk.hui at dalsemi.com
Mr. Robert H. Nixon P Emulex bob.nixon at emulex.com
Mr. Ralph O. Weber P ENDL Texas roweber at acm.org
Mr. Douglas Wagner P FCI dwagner at fciconnect.com
Mr. Mike Fitzpatrick P Fujitsu mfitzpatrick at fcpa.
fujitsu.com
Mr. Nathan Hastad P General Dynamics nathan.hastad at gd-ais.com
Mr. Rob Elliott P Hewlett Packard Co. Elliott at hp.com
Mr. Steven Fairchild V Hewlett Packard Co. Steve.Fairchild at hp.com
Mr. Zane Daggett P Hitachi Cable Manchester zdaggett at hcm.hitachi.com
Mr. George O. Penokie P IBM / Tivoli Systems gop at us.ibm.com
Mr. Robert Sheffield A Intel Corp. robert.l.sheffield at intel.
com
Mr. John Lohmeyer P LSI Logic Corp. lohmeyer at t10.org
Mr. Brian Day V LSI Logic Corp. brian.day at lsil.com
Mr. Tim Hoglund V LSI Logic Corp. tim.hoglund at lsil.com
Mr. Steven Schauer V LSI Logic Corp. steve.schauer at lsil.com
Mr. Mark Evans P Maxtor Corp. mark_evans at maxtor.com
Mr. Martin Czekalski V Maxtor Corp. marty_czekalski at maxtor.
com
Mr. Bruce Leshay V Maxtor Corp. bruce_leshay at maxtor.com
Mr. Bill Galloway P Pivot3, Inc. billg at pivot3.com
Dr. Heng Liao V PMC-Sierra, Inc. Liaoheng at pmc-sierra.com
Mr. Craig W. Carlson A# QLogic Corp. craig.carlson at qlogic.com
Mr. Ting Li Chan V QLogic Corp. ting.chan at qlogic.com
Mr. Arie Krantz V QLogic Corp. l_krantz at qlc.com
Mr. Don Carothers V QLogic Corp. don.carothers at qlogic.com
Mr. Jim Jones A Quantum Corp. jim.jones at quantum.com
Mr. Gerald Houlder P Seagate Technology gerry.houlder at seagate.com
Mr. Jim Coomes V Seagate Technology jim.coomes at seagate.com
Mr. Vit Novak P Sun Microsystems, Inc. vit.novak at sun.com
Mr. Paul D. Aloisi P Texas Instruments Paul_Aloisi at ti.com
Mr. Robert Kando A. Texas Instruments Robert_Kando at ti.com
Mr. Ron Mathews P UNISYS Corporation ronald.mathews at unisys.com
Mr. David Woolf V University of New djwoolf at iol.unh.edu
Hampshire
Mr. Martin Braff V Velio Communications martin at velio.com
Mr. Zoran Miljanic V Velio Communications zoran at velio.com
Mr. Kenneth Hirata P Vixel Corp. Ken.Hirata at Vixel.com
38 People Present
Status Key: P - Principal
A,A# - Alternate
AV - Advisory Member
L - Liaison
V - Visitor
4. Old Business
4.1 SAS Working Draft Review (SAS) [Elliott]
4.1.1 BREAK message upstream [Elliott]
Rob Elliott reviewed a problem where BREAK messages that should be passed
upstream are not necessarily received. He noted that a complete solution would
require numerous changes. George Penokie noted that only one condition really
needs to be covered in the standard.
In the absence of any objections, the group instructed the editor to make the
minimal change covering the one change George described.
4.1.2 DATA frame before XFER_RDY [Evans]
Bruce Leshay questioned what a target should do when it receives DATA frame
that is not covered by a previous XFER_RDY. Bruce, Jim Coomes, and George
Penokie agreed that the DATA frame should be discarded. Rob Elliott pointed
out that other parts of SAS require returning a CHECK CONDITION in such cases.
The group identified several cases similar to the specific case identified by
Bruce but differing in the details regarding why the target views the DATA
frame as unsolicited.
Rob asked that the group conduct a straw poll on the following question:
How should a target respond to a DATA frame that it has not requested
(regardless of the form in which the condition occurs)?
a) Terminate the command with a CHECK CONDITION status, with sense key of
ABORTED COMMAND and a new additional sense code;
b) Drop the DATA frame without reporting anything at any layer; or
c) Treat the condition as an overlapped command (i.e., abort all commands for
that initiator)
In the first straw poll the count was a8:b7:c2. John Lohmeyer eliminated
option c. In the second straw poll the count was a4:b5.
Jim Coomes moved that the SAS editor be instructed to require targets to
discard all DATA frames that have not been solicited with a XFER_RDY (or First
Burst) with that requirement taking precedence over all other requirements
listed at the end of 9.2.5.1. George Penokie seconded the motion. The motion
passed on a vote of 9:3:1.
4.1.3 DONE clarifications [Elliott]
Rob Elliott proposed adding clarifying text regarding timers in the DONE
primitive description and the group supported the changes.
4.1.4 Set Rate [Elliott]
Rob Elliott presented a problem concerning the "window" for 1.5 Gbs
transmissions. The group agreed to changing SAS to clarify that the 1.5 Gbs
"window" must always appear first.
4.1.5 Unrecognized Target Port Transfer Tags
Concerns were raised about a potential requirement on targets that do not
implement the target port transfer tag. Based on some wording in 9.2.5.1
inexperienced readers of SAS might believe that the target is required to test
incoming target port transfer tags even though they do not implement support
for the feature.
George Penokie moved that the SAS editor be instructed to require that targets
to discard all DATA frames with unrecognized Target Port Transfer Tags. Bill
Galloway seconded the motion. The motion passed on a 13:0:3.
4.1.6 I_T Nexus Loss timer [Coomes]
At the request of Jim Coomes, the group agreed to instruct the SAS editor to
change SAS to specify that an I_T Nexus Loss timer value of FFFF in the
Protocol-Specific Port mode page means that I_T Nexus Loss conditions will be
retried infinitely, 0000 means vendor specific retry behavior, and all other
values mean the number of milliseconds during which retries shall be
attempted.
4.2 SAS Revision 3 T10 letter ballot comment resolution (03-055r6) [Elliott]
Rob Elliott noted that the following documents are relevant:
03-055r6.pdf - comments sorted by author
03-055r6.zip - contains a zip file with sas-r03 with comments (PDF database)
sas-r03g.pdf - latest version of SAS
sas-r03g.zip - Frame source for SAS rev 3g
Rob also noted that 03-055r7 will be posted to reflect the results of this
meeting.
The group discussed handling outstanding transfers upon receipt of a hard
reset. It was agreed that SAS discussion of hard resets on other target ports
should include a reference to SAM-3.
In addition, the SAM-3 editor agreed to clarify SAM-3 to reflect the fact that
SAM-3 requirements regarding hard resets affecting other ports do not apply to
the effects caused by the logical unit reset required during the processing of
a hard reset.
Rob Elliott moved that 03-055r7 be recommended as resolving the SAS Letter
Ballot comments and that SAS r04 be recommended as containing the SAS Letter
Ballot comments resolution. Mark Evans seconded the motion. The motion passed
on a vote of 16:0:0.
4.3 SAS-1.1: CRC on Frame Header [Roberts]
Ron Roberts asked that discussion of this topic be deferred to the next
meeting.
4.4 SAS-1.1: Transport Layer Retries [Jones]
Rob Elliott led a discussion of the currently defined transport layer error
detection and recovery mechanisms (03-186r0).
Rob asked Jim Jones which of the current behaviors need to be addressed. Jim
responded that backup applications will terminate a backup whenever a command
is aborted. Any changes that move in the direction of eliminating aborted
commands are valuable.
The group walked through the various cases listed by Rob and tentatively
proposed error recovery mechanisms that do not result in commands being
aborted.
Rob Elliott and Jim Jones agreed to prepare a detailed proposal for
consideration at the next meeting. Jim Jones agreed to prepare a set of ladder
of diagrams matching the proposal.
4.5 SAS Requests and Confirmations (03-023r0) [Evans]
Mark Evans requested that this item be removed from future agendas.
5. New Business
5.1 SAS 1.1: SATA II Port Selector Interoperability (03-155r0) [Sheffield]
Bob Sheffield presented information from a draft SATA II specification (page 5
in 03-155r0) and asked the group if there was any interest in adding text to
SAS 1.1 to handle SATA II Port Selectors. The group asked questions about how
Port Selectors are expected to work.
Bob agreed to prepare a detailed proposal for consideration at the next
meeting.
5.2 SAS: Initiator Response Timeout timer (03-164r3) [Penokie]
As the group reviewed the issues under contention in 03-164r3, it was
discovered that all the proposed changes have already been incorporated in SAS
r03g.
5.3 SAS: SATA information to Annex (03-168r0) [Penokie]
George Penokie proposed that all the SATA-only text be moved to a SAS annex.
George Penokie moved that 03-168r0 for inclusion in SAS as resolution for
several rejected Letter Ballot comments. Ron Roberts seconded the motion.
Ralph Weber began reviewing the proposed changes individually.
Tim Hoglund moved to call the question. Jim Coomes seconded the motion to call
the question. The motion to call the question failed to achieve a two-thirds
vote with a count of 5:4.
When Ralph completed his review there were no further objections to voting on
the motion.
The motion to included 03-168r0 failed on a vote of 6:7:3
George Penokie moved that every occurrence of "(Informative)" be removed and
that any remaining occurrences of "See SATA" be changed to "See ATA/ATAPI-7
Volume 3". Ron Robert seconded the motion. The motion passed on a vote of
11:1:4.
5.4 Initiator Connection Tag Changes Detected by Target [Krantz]
John Lohmeyer reported on a question from Arie Krantz on the action to be
taken by an initiator if a change is detected in the Initiator Connection Tag.
Bill Galloway responded that the target can do anything that is convenient
except aborting all the outstanding commands.
The group discussed ways in which the wording could be improved to clarify the
allowed behaviors. Rob Elliott took notes for resolving the issue as part of
the SAS Letter Ballot.
6. Review of Recommendations
The following documents were recommended for approval during this meeting:
03-055r7 - SAS Letter Ballot comments resolution
SAS r04 - SAS revision containing the Letter Ballot comments resolution
7. Meeting Schedule
SAS Protocol Working Group meetings are scheduled for:
Monday-Tuesday, July 7-8, 2003 commencing at 1:30 p.m. in Colorado Springs, CO
at the Wyndham Colorado Springs Hotel (719-260-1800), hosted by LSI Logic. The
group discussed the possibility of adjourning the meeting at noon on Tuesday
and starting the CAP meeting a half-day early.
Additional teleconference calls may be announced on the T10 reflector as
needed.
8. Adjournment
The meeting was adjourned at 12:10 p.m. on Tuesday May 6, 2003.
*
* 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