Deferred Error vs. Unit Attention
erich_oetting at stortek.com
Tue Dec 15 14:14:21 PST 1998
* From the T10 Reflector (t10 at symbios.com), posted by:
* Erich Oetting <erich_oetting at stortek.com>
>* From the T10 Reflector (t10 at symbios.com), posted by:
>* Gerry_Houlder at notes.seagate.com
>The deferred error reporting mechanism is intended for reporting an error
>with a command that had previously been ended with GOOD status. It is not
>used for any other situation. An example of the deferred error case:
>(a) Initiator sends a WRITE command with Immediate bit set (tape devices do
>this) or with WCE active on mode page Ah (direct access devices). The
>target will take the data and return GOOD status right away.
>(b) Later the target encounters an error while writing the data. It stops
>command execution and prepares a deferred error report.
>(c) The target receives the next command from the same initiator and
>reports CHECK CONDITION status. The deferred error sense data is sent to
>the initiator in REQUEST SENSE command.
Thanks for the description Gerry.
I would like to add one more item to this list. This behaviour is not
specified in the SCSI standards, but probably should be.
(d) When the deferred error bit in sense data is one, the command that
received CHECK CONDITION status has not been executed.
If this contraint is not applied, it becomes nearly impossible for the
initiator to determine what happened. Some deferred errors, such as a
permanent write error, may cause all following write commands to fail.
If the following command is allowed to execute, do we return sense
describing the deferred error, or sense describing the error on the
... Erich Oetting
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at symbios.com
More information about the T10