SATA Port Selector

Sheffield, Robert L robert.l.sheffield at intel.com
Tue Feb 10 12:42:36 PST 2004


* From the T10 Reflector (t10 at t10.org), posted by:
* "Sheffield, Robert L" <robert.l.sheffield at intel.com>
*
Kido-san,

Your comment #2 below is the subject of a working proposal discussed at
the January SAS WG meeting, 03-029r0. That proposal suggests fixing the
problem by adding a field to the DISCOVER response to indicate the state
of the SP state machine. The SAS WG felt it was unnecessary and thought
it too restrictive for the manaagement application to be concerned about
the state of the SP state machine at such a low level to ascertain the
validity of the ATTACHED SATA PORT SELECTOR bit. The SAS WG recommended
instead to add a state to the NEGOTIATED PHYSICAL LINK RATE field to
indicate when the link rate has not yet been negotiated, and to specify
that the SP state machine causes a BROADCAST(CHANGE) to be generated
when a port selector is positively detected. This way, as long as the
DISCOVER command is issued (in response to BROADCAST(CHANGE)) prior to
the next HOT-PLUG Timeout, the SMP initiator knows the ATTACHED SATA
PORT SELECTOR bit should still be valid.

Regarding your comment #3, it was felt important to recognize the
difference between a SATA device attached directly to an expander phy
versus a SATA device attached behind a SATA Port Selector. This is
important when STP initiators exchange ID information (side-band) about
SATA devices and the SAS expander ports in the SAS domain that have port
selectors so that the STP initiators can decide which STP initiator
should access the device when.

If we agree regarding comment #3 that it's important to recognize the
SATA Port Selector, whether it's attached to an expander port via the
ACTIVE or INACTIVE phy, then we have to consider the case where the SP
state machine must recognize the SATA Port Selector on the ACTIVE phy.
In this case the Port Selector will respond with a COMINIT immediately
after sending the initial COMWAKE. It is also possible that the phy is
already active to begin with, and that the first COMINIT seen by the
expander phy in SP0 is the passthrough COMINIT from the SATA device
behind the SATA port selector. Given the possible sequences of events,
it is possible that COMWAKE detection could occur in the SP3 or SP4
states. This is why those states contain the text about setting the
ATTACHED SATA PORT SELECTOR bit upon detection of COMWAKE.

I still haven't completed drafting 03-029r1, so if you have further
suggestions, I'd be happy to consider them.

Thanks,
Bob Sheffield
Intel Corporation

-----Original Message-----
From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf Of Kido
Shogo
Sent: Tuesday, February 10, 2004 5:40 AM
To: t10 at t10.org
Subject: SATA Port Selector


* From the T10 Reflector (t10 at t10.org), posted by:
* "Kido Shogo" <kido-sxa at necst.nec.co.jp>
*
Hi all,

I have some proposals and questions about the SATA Port Selector 
in SAS-1.1 rev3.

1. I think the following text is unnecessary for SP3 and SP4 state.
  "If this receives COMWAKE Detected message and phy supports
  attachment to s SATA device and supports attachment to a 
  SATA port selector, it shall set the ATTACHED SATA PORT 
  SELECTOR bit in the DISCOVER response to one."

  I think the inactive port of the SATA Port Selector sends 
  COMWAKE only,  and, the phy which SATA Port Selector attached 
  does not move to SP3 or SP4 state.  


2. The following text is described in the SP0 state. 
   "Any transition to this state shall set to the ATTACHED SATA PORT 
   SELECTOR bit in the DISCOVER response to zero. "

   It means that the phy which SATA Port Selector attached repeats 
   the ATTACHED SATA PORT SELECTOR bit zero and one.
   I think this action is the problem because SMP initiator detects     
   the phy as attached to the SATA Port selector or as not attached 
   to the SATA Port selector depending on the timing to transmit 
   the Discover. 


3. I do not understand the situation that  ATTACHED SATA PORT 
   SELECTOR bit is one and ATTACHED SATA DEVICE bit is one 
   in the Discover response. 

   I think the ATTACHED SATA PORT SELECTOR bit is always zero 
   when the phy detects the SATA deivce. 


Thanks,
Shogo Kido
 NEC System Technologies, Ltd.
 External:(+81)89-947-7901
*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org
*
* 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