Discovering which PHY caused the CHANGE

Reif, Jim Jim.Reif at hp.com
Fri Jan 31 15:40:10 PST 2003


* From the T10 Reflector (t10 at t10.org), posted by:
* "Reif, Jim" <Jim.Reif at hp.com>
*
Couldn't you issue the REPORT PHY SATA command to all phys of the
expander which was determined to have it's CHANGE COUNT incremented,
and then check the AFFILIATION VALID bit? If this bit is cleared, chances 
are that there was a link reset on this phy. The bit should have been set after
any I/O's were issued to that phy. 

Jim Reif
Hewlett-Packard 
281-514-8237
email: jim.reif at hp.com


-----Original Message-----
From: Johnson, Stephen B. [mailto:sjohnson at lsil.com]
Sent: Friday, January 31, 2003 4:29 PM
To: t10 at t10.org
Subject: Discovering which PHY caused the CHANGE


* From the T10 Reflector (t10 at t10.org), posted by:
* "Johnson, Stephen B." <sjohnson at lsil.com>
*

Link Reset can occur on loss of dword sync and other causes. This will cause
an expander to broadcast CHANGE as it should.
If a SATA device is attached to the culprit PHY, the command set will be
cleared.
As an initiator I would like to know which drives command set was cleared
before IO's start timing out.

Using SMP commands how does the initiator know which PHY caused the CHANGE
if the link rate has not changed?

I can't seem to find anything definitive using SMP requests.

The REPORT GENERAL gives the EXPANDER CHANGE COUNT, so we know which
expander broadcast the change.

I don't see any fields in the DISCOVER response that help.
I don't see any fields in the REPORT PHY SATA that help.

The REPORT ERROR LOG response does have the "LOSS OF DWORD SYNCHRONIZATION
COUNT" 
but I don't think that was the intent of this request and it does not cover
all the cases.

SAS drives could possibly continue along since the command set is not
cleared and link rate hasn't changed. 
But, it will still kick off a discovery that does not find anything. This is
OK behavior.

So, what we really could use is a 16 bit PHY CHANGE COUNT field in the
DISCOVER response that behaves similar to the EXPANDER CHANGE COUNT.


Steve Johnson
LSI Logic
719 533 7511
sjohnson at lsil.com



*
* 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