SCSI-2/3 Question -- Format of Target I/D in "Read Element Status"

Erich Oetting erich_oetting at stortek.com
Tue Jul 30 13:15:58 PDT 1996


* From the SCSI Reflector, posted by:
* erich_oetting at stortek.com (Erich Oetting)
*
>* From the SCSI Reflector, posted by:
>* Charles Monia 237-6757 <monia at AM.SHRMSG.SHR.mts.dec.com>
>*
>
>In working with SCSI Medium changers, we have noticed two
>interpretations for the SCSI bus address field in the Read Element Status
>data for tape drives. As shown in table 339 of SCSI-2, draft revision 10k,
>this field indicates the drive's SCSI target ID when it is on the
>same bus as the medium changer. The description of the field in
>the draft standard is:
>
>        "The SCSI bus address field, if valid, provides the SCSI address
>        (binary representation) of the primary device served by the
>        medium changer at this element address"
>
>The question of interpretation is over "binary representation".
>
>Most robots that We've seen put the binary *value* of the
>address in the field.  One, has elected to treat the field as a
>bit mask, with the appropriate bit set for the address.  For example
>given a drive at SCSI I/D 3 most will use:
>
>        |  7  |  6  |  5  |  4  |  3  |  2  |  1  |  0  |
>        +-----+-----+-----+-----+-----+-----+-----+-----+
>        |  0  |  0  |  0  |  0  |  0  |  0  |  1  |  1  |
>        +-----+-----+-----+-----+-----+-----+-----+-----+
>
>But the drive in question uses:
>
>        |  7  |  6  |  5  |  4  |  3  |  2  |  1  |  0  |
>        +-----+-----+-----+-----+-----+-----+-----+-----+
>        |  0  |  0  |  0  |  0  |  1  |  0  |  0  |  0  |
>        +-----+-----+-----+-----+-----+-----+-----+-----+
>
>Which of the above is correct?

My interpretation is that the first example is correct.  This is based on
most other fields being interpreted as a binary numbers.  Example 2 also
does not work for a 16 or 32 bit wide bus.


>In researching this question, I discovered that the SCSI-3 SMC draft standard
>(rev 5, subclause 6.5.6, pp46) defines this field and the logical unit field as
>follows:
>
>"The SCSI bus address field, if valid, provides the SCSI address (binary
>representation) of the primary device served by the medium changer at this
>element address.

Changing "binary representation" to "binary value" might clarify the
standard.

>
>The logical unit number field, if valid, provides the logical unit number
>within the SCSI bus device of the primary device served by the medium changer
>at this element address.
>....
>Editors note: The SCSI bus address and logical unit number fields are too
>small for SCSI-3."
>
>Has there been a proposal to address the issues pointed out by the editor's
>note?

I do not have a specific proposal at this time to fix these fields.

A more general solution would be to use world wide unique names to
identify the attached transport, and make the SCSI bus address field,
logical unit number field, and associated flags obsolete.  This would
provide for device identification rather than path identification, and
nicely sidestep the question of multiple ports.

World wide unique names may also be appropriate for Import/Export elements
used for interchange between two media changers.

  ... Erich Oetting  (SMC Technical Editor)







More information about the T10 mailing list