Unit attention on FCP devices

James Smart james.smart at trebia.com
Mon Aug 28 11:49:36 PDT 2000


* From the T10 Reflector (t10 at t10.org), posted by:
* "James Smart" <james.smart at trebia.com>
*
The Unit Attention (w/ NACA=0) was cleared when sense data was returned (as
per your indicated clauses in the spec). This has been true of all FCP
devices I have encountered.

For the condition you are worried about - in flight packets at the time of
the CA condition .. that was the whole purpose of ACA. If you do indeed care
about such protecting such a condition, ACA should be supported.

-- james


> -----Original Message-----
> From: owner-t10 at t10.org [mailto:owner-t10 at t10.org]On Behalf Of Edward A.
> Gardner
> Sent: Monday, August 28, 2000 11:28 AM
> To: T10 Reflector
> Subject: Unit attention on FCP devices
>
>
> * From the T10 Reflector (t10 at t10.org), posted by:
> * "Edward A. Gardner" <eag at ophidian.com>
> *
> I recently encountered a question about the operation of unit attention on
> FCP devices.  While I think I know how it has to work, I cannot
> find a clear
> statement of this anywhere (SAM, SAM-2, FCP, FCP-2).
>
> The specific question is what event clears a unit attention
> condition on an
> FCP device (or more generally on any device that supports autosense).
> Consider the following events:
>
> 1.  An initiator is accessing a LUN on an FCP target.
>
> 2.  A unit attention condition occurs on the LUN during some
> period when no
> commands or tasks are outstanding to it.  Perhaps someone changes
> the media
> on a removable media device.  The event specifically does not disturb the
> connection (process login) between the initiator and target.
>
> 3.  The initiator sends a command (NACA=0) to the target.  This command
> encounters the unit attention condition, the target returns CHECK
> CONDITION
> and autosense data describing the unit attention condition.  Note that the
> CA condition that occurs by returning CHECK CONDITION is
> immediately cleared
> by the simultaneous return of autosense data (I would like to say
> that no CA
> condition occurs, but that isn't SAM terminology).
>
> At this time, has the return of autosense data describing the
> unit attention
> condition cleared the unit attention, or does it still exist until the
> initiator issues an explicit REQUEST SENSE to clear it?
>
> On the one hand I believe it is necessary that the unit attention
> condition
> still exist until an explicit REQUEST SENSE.  Otherwise commands
> that are in
> flight, issued before the initiator was aware of the unit attention
> condition, may be executed against the wrong media (or against whatever
> change the unit attention was reporting).
>
> On the other hand the words in SAM and SAM-2 appear to suggest
> that the unit
> attention condition will be cleared by the return of autosense data.  SAM
> and SAM-2 use the phrase "if a request for sense data is received" (clause
> 5.6.5).  One can interpret autosense as a request for sense data.  Clause
> 5.6.4.2 states "The return of sense data in this way [autosense] is
> equivalent to an explicit command from the application client requesting
> sense data".  That is, returning autosense data is equivalent to a REQUEST
> SENSE, and REQUEST SENSE clears unit attention.
>
> The one thing I have no doubt about is that the current wording describing
> unit attention handling is ambiguous and confusing; I intend to generate
> some comments to clarify it.  But first I want to make certain whether FCP
> devices work the way I think they do.
>
> Edward A. Gardner               eag at ophidian.com
> Ophidian Designs                719 593-8866 voice
> 1262 Hofstead Terrace           719 593-8989 fax
> Colorado Springs, CO  80907     719 210-7200 cell
>
> *
> * For T10 Reflector information, send a message with
> * 'info t10' (no quotes) in the message body to majordomo at t10.org

--------------------
James Smart
Trebia Networks, Inc                  Ph:   978-318-9547
35 Forest Ridge Rd                    Cell: 603-674-3687
Concord,  MA   01742                  james.smart at trebia.com

*
* 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