99-285r0 - Changes made while incorporating Logical Unit Self-tests

Ralph O. Weber ralphoweber at CompuServe.COM
Tue Oct 5 18:38:33 PDT 1999


* From the T10 Reflector (t10 at t10.org), posted by:
* "Ralph O. Weber" <ralphoweber at compuserve.com>
*
A presentation for review at the November meetings has been placed on
the T10 FTP site as:

< ftp://ftp.t10.org/t10/document.99/99-285r0.pdf >

The text of the presentation is as follows.

Doc:  T10/99-285r0
Date: 05 October 1999
To:   T10 Technical Committee
From: Ralph Weber, LSI Logic Alternate Member of T10
Subj: Changes made while incorporating Logical Unit Self-tests

While incorporating the Logical Unit Self-tests proposal (99-179r5)
in SPC-2 revision 12, I encountered several problems mostly
concerning the relationship between the new proposal an the classic
self-tests defined by the SELFTEST bit.  This document describes the
changes made to give both self-test methods an identifiable
definition.

The classic self-test has been given the name "default self-test".
This name is consistent with the existing text: "A self-test
(SELFTEST) bit of one directs the device server to complete the
target's default self-test."  It has the additional advantage of
being different from all the names for the new self-test: "short",
"extended", "foreground", and "background".

The conditions stated in 99-179r5 for terminating other SEND
DIAGNOSTICS commands while a background mode self-test is in progress
allowed a command with the SELFTEST bit set to one to execute and
prohibited SES SEND DIAGNOSTICS commands (which are pretty much the
only commands SES has).  This was modified to prohibit SELFTEST=1 and
allow the SES uses of SEND DIAGNOSTICS.

The first paragraph in the description of the SEND DIAGNOSTICS
command was modified to clarify that implementation of the self test
features in 99-179r5 is optional.

99-179r5 requires that the response to an attempt to abort a
non-existent background self-test is NOT READY - LOGICAL UNIT NOT
READY, SELF-TEST IN PROGRESS.  This looks like a cut and paste error.
It has been changed to ILLEGAL REQUEST - INVALID FIELD IN CDB.

99-179r5 does not specify the interaction between the newly defined
function codes and the parameter list length.  Requirements have been
added that the parameter list length shall be zero when the function
code is non-zero.  This provides for future enhancements of the short
or extended self-test features.

Since 99-179r5 provides for returned self-test data in a log page,
numerous references to "returning data" that formerly were applicable
to diagnostic page data no longer fully apply, because the new
self-test feature returns data but not in response to a REQUEST
DIAGNOSTIC RESULTS command.  This technicality has been corrected by
changing "data" to "parameter data" wherever applicable in the SEND
DIAGNOSTIC command description.

The following requirements have been added to the ends of the
paragraphs describing the  DEFOFFL and UNITOFFL bits, to restrict
their usage to the default (classic) self-test:

"When the SELFTEST bit is zero, the DEFOFFL bit shall be ignored."
"When the SELFTEST bit is zero, the UNITOFFL bit shall be ignored."

It is believed that the two sentences above should have been added
when the SES usage of the SEND DIAGNOSTIC command was defined.

The table describing the exception commands for background self-tests
and the text describing the table were restructured to combine
identical situations and make the presentation of requirements more
direct.  This restructuring was intended to have no effect on the
requirements.


*
* 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