SAS directions on 15 issues

Elliott, Robert (Server Storage) Elliott at
Tue Feb 11 14:56:07 PST 2003

* From the T10 Reflector (t10 at, posted by:
* "Elliott, Robert (Server Storage)" <Elliott at>
Here is the direction on various SAS issues from today's letter ballot
resolution call.  These will be implemented in sas-r03d.  Please note
that IDENTIFY gets a new field, link rate field values change
everywhere, and the DISCOVER fields are rearranged.

* Transport protocol error retries - several backup vendors have no
problem using READ POSITION and LOCATE to recover from failed tape
commands.  Do nothing for SAS-1.

* STP affilation clearing - clear whenever the link reset is begun on
the SATA physical link except when requested by PHY CONTROL function
LINK RESET operation.

* SMP frame type errors - if target doesn't see 40h, generate BREAK
(like bad CRC); if initiator doesn't see 41h, send Frame Received (SMP
Failure) confirmation

* SMP state machine corrections - as proposed in the email (withe SMP
frame type errors handled as noted above)

* SSP Extra bytes in CDB fields - "shall be ignored" by the target

* SSP Incorrect LUN in TASK frame - return RESPONSE frame with new
response code of INVALID LOGICAL UNIT.  Propose SAM-3 changes to model
this (FUNCTION REJECTED service response)

* SSP sense data not multiple of 4 - let SENSE DATA LENGTH be any value.
Use the frame header's NUMBER OF FILL BYTES field to indicate extra
bytes to be ignored

* Link rate fields - change OPEN address frame, SMP DISCOVER, and SMP
PHY CONTROL to use 8h for 1.5 Gbps and 9h for 3.0 Gbps

* SMP PHY CONTROL operations on incoming phy - Reject requests to

* SMP function support by non-expanders - Make REPORT GENERAL and
DISCOVER mandatory for all devices supporting SMP, allow REPORT PHY SATA
in end devices

* Identification of phys in wide ports - Add the PHY IDENTIFIER to the
the SAS mode page descriptor, and the SAS log page descriptor

* SMP Phy Change Count (03-089) - add this new field to DISCOVER

* Field layout - to incorporate 03-089 and the phy identifer, make these
IDENTIFY address frame: 
put phy identifier in byte 20
mark 21-27 as reserved not restricted

OPEN address frame: no change

SMP DISCOVER function response: 
move 32-39 to 40-47 to keep the "incoming IDENTIFY" region reasonably
Swap Attached SAS address and SAS address fields.
put Attached Phy Identifier in byte 32.
40 Minimum Link Rates
41 Maximum Link Rates
42 Phy Change Count 
43 Internal Phy and Partial Pathway Timeout Value
44 Routing Attribute
45 reserved
46 reserved
47 reserved
48 reserved 
49 reserved
50 vendor-specific 
51 vendor-specific 

* SSP Target port transfer tag (03-091) - allow target to check for
FFFFh in COMMAND and TASK frames.  The target shall return a RESPONSE
frame with a response code of INVALID FRAME if it chooses to check any
of the reserved fields in the frame header or the IU structures, or the
FFFFh in the target port transfer tag field.

* Primitives inside frames - debate on whether a BREAK primitive
sequence within the first 8 data dwords (or non-ALIGN/NOTIFY dwords?)
after a SOAF shall be detected or shall not be detected.  May be
detected is unacceptable.  Transmitters shall not insert BREAKs inside
address frames.  Prepare to debate on next week's conference call and in

Rob Elliott, elliott at
Hewlett-Packard Industry Standard Server Storage Advanced Technology

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

More information about the T10 mailing list