Proposal to add a new modifier to SCSI Write commands. -Reply

[Joseph Glider/Almaden/IBM]

* From the SCSI Reflector (scsi at symbios.com), posted by:
* "Joseph Glider/Almaden/IBM" <gliderj at almaden.ibm.com>
*
The job of anyone (host driver, array controller or drive) who reassigns a
block after an unrecovered read error should be to get a good surface for
somebody to write on, and to make sure it's current contents (or lack of) are
not read as good data. That is the essence of the need for 'logically' bad
data, and why it needs be differentiated from ECC error (in an ideal world).

I would agree that putting 'forced error' controls into the Write command is
not needed unless it IMPROVES the current situation. As I understand it, the
READ LONG and WRITE LONG commands are being used for the purpose, but do not
entirely satisfy the the requirement, of signalling 'logically' bad data. Any
change, if useful and possible, should put in more functionality, and not just
reproduce the current function of simulating ECC errors.

* From the SCSI Reflector (scsi at symbios.com), posted by:
* PAT LaVarre <LAVARRE at IOMEGA.COM>
*
> The host should only Reassign when it knows what
> to put in the new block (that is, after a failed
> Write or a Recovered Read).

As another "should" not (yet) captured in the
standard, this rule holds true for the drive, too,
yes?  (Except in response to an explicit 07h
REASSIGN command?)

> Maybe your system works differently?

Hmmm.

Seems like a system that reassigned any LBA that
ever had trouble would be more reliable - up until
the point when it ran out of reallocation space.
(WIBNI we had a standard way to predict exhaustion
of reallocation space.)

I can't say for sure I know of a system that
reallocates unrecovered reads myself ...

.... but I'm thinking I've seen host people
implement reallocation on a retry count.

Try writes ten times, try reads twenty, reassign
otherwise, loop?

Pat LaVarre

*
* For SCSI Reflector information, send a message with
* 'info scsi' (no quotes) in the message body to majordomo at symbios.com



*
* For SCSI Reflector information, send a message with
* 'info scsi' (no quotes) in the message body to majordomo at symbios.com