From: Stephen FINCH <steve.finch@st.com>
To: "'George Penokie'" <gop@us.ibm.com>
Cc: <t10@t10.org>
Subject: RE: BROADCAST ASYNCHRONOUS EVENT
Date: Fri, 7 Jul 2006 10:35:20 -0600
X-Message-Number: 7033
Formatted message: HTML-formatted message

George,
The description of state CC0 and CC1 explicitly discuss only three broadcast
messages:  
   BROADCAST Received (Change) message, 
   BROADCAST Received (Reserved Change 0) message, and 
   BROADCAST Received (Reserved Change 1)
The other BROADCAST types are not mentioned.
What I was proposing is that every BROADCAST type received cause an
appropriate confirmation to be generated.
I think this is what everyone does anyway, we just need to update the
proposed standard with appropriate language.
I'm even willing to have a statement to the effect that this confirmation
"may" occur.
But as it reads today, no one receives a confirmation of any type when a
BROADCAST (Asynchronous Event) is received.
See you next week,
Steve Finch
steve.finch@st.com
-----Original Message-----
From: owner-t10@t10.org [mailto:owner-t10@t10.org] On Behalf Of George
Penokie
Sent: Thursday, July 06, 2006 2:53 PM
To: Stephen FINCH
Cc: t10@t10.org
Subject: Re: BROADCAST ASYNCHRONOUS EVENT
Steve, 
The BROADCAST Received (x) messages that are received in by the SL_CC0:Idle
or the SL_CC1:ArbSel specifically result in a Change Received confirmation.
That confirmation will trigger a discovery of the topology. A Broadcast
(Asynchronous Event) should not be causing any confirmation will trigger a
discovery of the topology.  There are no other states in which a Broadcast
results in a confirmation to a management layer. 
So the question is; Should the Broadcast (Asynchronous Event) cause a
notification to the management layer or is it like other Broadcasts that are
handled in some other fashion. If the management layer does care then there
will have to be a new confirmation added to some state machines states
(probably SL_CC0:Idle or the SL_CC1:ArbSel). Then the next question is how
far up the stack does it have to go. If it only needs to go to the
management layer, then using the same method as change would work. But if
the information needs to go up to the application client then there is a
problem in that SAM has no definition on how to handle an asynchronous
event. 
Bye for now,
George Penokie
Dept 9A8 030-3 A410
E-Mail:    gop@us.ibm.com
Internal:  553-5208
External: 507-253-5208 
Stephen FINCH <steve.finch@st.com> 
Sent by: owner-t10@t10.org 
06/15/2006 04:05 PM 
To
<t10@t10.org> 
cc
Subject
BROADCAST ASYNCHRONOUS EVENT
* From the T10 Reflector (t10@t10.org), posted by:
* Stephen FINCH <steve.finch@st.com>
*
I noticed that in sas2r04 that we define BROADCAST (ASYNCHRONOUS EVENT) and
have a way to send it, but we don't allow anyone to receive it.  I think we
need to add to the SL_CC state machine, in states CC0, CC1 and CC2, the
requirement for the CC state machine to send a ASYNCHRONOUS EVENT RECEIVED
confirmation to the management layer.  This is how the reception of a
BROADCAST (CHANGE) is handled.
Or maybe we should make the change more generic and allow _any_ received
BROADCAST to cause a confirmation to be sent to the management layer with a
type argument?
Regards,
Steve Finch
STMicroelectronics
*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo@t10.org