Responses to 05-141r1 SAS letter ballot comments

Gil Romo gil.romo at qlogic.com
Wed Apr 20 14:31:55 PDT 2005


* From the T10 Reflector (t10 at t10.org), posted by:
* Gil Romo <gil.romo at qlogic.com>
*
I have responses to three of the comments contained in 05-141r1.

part 1) Your statement that there are no arguments described in a Backoff 
Reverse confirmation is correct.  The arguments are implied, however, and this 
does not require a state machine change.

part 2) In 7.15.4.5, the standard states that the expander phy (ie, Port 0) 
transitions to XL5 to transmit the connection request associated with the 
Forward Open indication.  Then, the XL6 state performs arbitration between the 
received request and the forwarded request.  The XL1 to XL5 transition occurs 
when no ECM has been received, implying that the ECM has not received this 
request.  There is no discussion of the different timing conditions between 
crossing requests in the standard, other than the state machine definition 
itself.

part 6) After generating a Backoff Retry response, Port 0 transitions to XL1 
with its Retry Priority argument set to IGNORE AWT.  Port 0 then requests Port 1 
resources and wins according to the statement "the ECM shall select the winner 
|from the set of Request Path requests with Retry Priority Status arguments of 
IGNORE AWT" (section 7.12.4.1), while Port 1 loses.  Upon losing, Port 1 returns 
to the idle state where it immediately receives the Forward Open indication.  
There is no memory of the arbitrating (normal) confirmation that was received in 
the previous ECM request, thus the condition you describe does not occur.

Regards,
Gil
--------------------------------------------------------------------------
Gilbert Romo                         Circuits and Integration Engineer
Voice: (949) 389-6266                QLogic Corporation
E-mail: gil.romo at qlogic.com          Aliso Viejo, California


*
* 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