QErr bit and Multi-Initiator

Charles Binford chas at smtplink.wichitaks.ncr.com
Tue Aug 16 10:26:43 PDT 1994


Date:         Fri Aug 12 08:59:21 1994
From: Gene Milligan <Gene_Milligan at notes.seagate.com>
Sender: Gateway <gateway at notes.seagate.com>
Message-ID: <2e4b722a.seagate at notes.seagate.com>
Organization: Seagate Technology
Reply-To: Gene Milligan <Gene_Milligan at notes.seagate.com>
To: scsi at WichitaKS.NCR.COM
Subject:      Re: QErr bit and Multi-Initiator

What Gene Milligan wrote is marked with '>':
------------------------------------------------------------------------------
>Referring to Charles Monia's original questions 8/11/94.
              ^^^^^^^^^^^^^^^ You mean Charles Binford

>I think your literal assumption of how QErr works is one that would result from

I am not sure what you mean by my "literal assumption of how QErr works".  SCSI 
says "...all remaining suspended I/O processes shall be aborted...".  It goes on
to say that other initiators which had I/O processes aborted should get a unit 
attention.  I read that to mean the target should basically perform the 
equivalent of a SCSI-2 CLEAR_QUEUE operation when the contingent allegiance is 
cleared.
 
>artificial intelligence. I would say it works analogously to what would happen
>if rather than an error the Initiator that experienced the troublesome error
>had sent down an instantaneous Abort Task Set. Only the I/O Processes present
>in the Task Set at the time the instigating Unit Attention Condition occurred
>due to real errors from real I/O processing are aborted. Pending reports due
>other initiators have no clearing function relative to fresh new I/O processes.

Are you suggesting that the target should keep track of whether this contingent 
allegiance is a result of "real errors from real I/O processing" verses a unit 
attention, and then perform the QErr function only on "real errors"?  If I 
misread your comments, please correct me.  If I understood you, I disagree.
Charles Binford

> 
>As I recall the  QErr function came about due to the desire to have an auto
>reservation upon error similar to the way (I am told) the block mux channel
>works. This then would have given the Initiator the opportunity to issue  an
>Abort Task Set rather than have a QErr function.
>--
>Gene Milligan -- Gene_Milligan at notes.seagate.com
>-------------------------------------------------------------------------
>Seagate Technology   -   920 Disc Drive   -   Scotts Valley, CA 95066 USA
>Main Phone 408-438-6550   -   Email Problems postmaster at notes.seagate.com
>Technical Support: BBS 408-438-8771  Fax 408-438-8137  Voice 408-438-8222  
>-------------------------------------------------------------------------




More information about the T10 mailing list