A Modest Proposal: Protocol Services for Link Events

Ralph Weber ralphoweber at compuserve.com
Mon Sep 15 16:34:57 PDT 2003


* From the T10 Reflector (t10 at t10.org), posted by:
* Ralph Weber <ralphoweber at compuserve.com>
*
Paul,

Before you go too far with this, please verify that SAM-3 clause 6
fails to address you issue.

Rob Elliott, Mallikarjun Chadalapaka, and several others worked pretty
hard on clause 6 in general and the I_T Nexus Loss concept in particular.
To first order, the problem described in you message looks a lot like
an I_T Nexus Loss.

Regards,

.Ralph

Paul.A.Suhler at certance.com wrote:

> 
>* From the T10 Reflector (t10 at t10.org), posted by:
>* Paul.A.Suhler at certance.com
>*
>A number of SCSI transport protocols specify the effects that link events
>have upon tasks.  For example, in login-based protocols like FCP, a login
>or logout will terminate all tasks.  It doesn't seem correct to represent
>this as an invocation of Task Management Request Received (Abort Task /
>Abort Task Set / etc.) in the target and (for every affected task) an
>invocation of Command Complete Received in the initiator.
>
>Everyone who implements a SCSI port has to deal with these events in their
>internal APIs.  Is there value to standardizing the behavior by adding new
>protocol services to SAM and using them in transport protocols?  Will this
>help us avoid incorrect implementations?
>
>Or is this an unnecessary foray into Formalism Hell?
>
>Thanks,
>
>Paul Suhler
>Senior Staff Firmware Engineer
>Certance
>1650 Sunflower Avenue
>Costa Mesa, California  92626
>(714) 641-2485 (voice)
>(714) 966-7328 (fax)
>
>
>*
>* 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