X3T9.2/88-143








To: John Lohmeyer, Chairman X3T9.2

From: Bob Pentecost

Subject: Comments, question on SCSI-2 Rev. 5


The following questions, comments and suggestions are the result of
detailed reviewing of SCSI-2 Rev. 5 done by a variety of people (some
with little SCSI experience, some with a lot of SCSI-1 experience) at
Hewlett Packard.



3.1  p. 3-1

      Did the term host bus adapter (HBA) originate with SCSI?  If so,
      should it be defined?


5.6.9  IGNORE WIDE RESIDUE  (p. 5-22)

      It isn't obvious why this message is used only after the DATA IN
      phase.  Perhaps an explanation would be in order here.


6.1  p. 6-1

      Is CDB LUN ignored if the Identify message was sent?  Found
      answer in 6.6.2 on page 6-6.  This is a little confusing.


6.2.4  p. 6-6, paragraph 3

      The sentence: "Commands that use multiple bytes for transfer
      length allow 65,535 or greater blocks of data..." seems to imply
      that these commands can only transfer 65,535 blocks or greater
      but not 1 through 65,534.  A following sentence clears it up but
      the first sentence is still confusing.


6.4.3  p. 6-11, last paragraph
      This seems to be the only place that comes close to explaining
      how relative addressing works.  A general explanation is needed
      (perhaps in the glossary).


6.5.1  p. 6-12, last sentence

      ASC "CHANGED DEFINITION" is actually "CHANGED OPERATING
      DEFINITION" (3Fh).


6.5.3  p. 6-13, paragraphs (1) and (2)

      Additional Sense Code Invalid LU is now called Unsupported LU
      (25h).

6.9.2  p. 6-18, paragraph 1

      Does SCSI-2 place any constraints on allocation of queue slots
      to LUN's?  Must each LUN be allocated at least one slot or can
      the entire pool of slots in a target be dynamically managed?


6.9.2  p. 6-20, paragraph 5

      What is the definition of "short" and "longer" periods of time
      with respect to rejection of tagged queuing for target resource
      allocation reasons?


6.9.3.2  p. 6-23, paragraph 1

      The first sentence doesn't make sense beginning with the phrase:
      "...and removed from...".


6.9.3.2  p. 6.23, paragraph 2

      This paragraph seems to be the submitters comments on the
      command queuing proposal.  Should they remain in the document?


7.2.5.1  p. 7-23, first paragraph

      The Queuing bit in Inquiry should state explicitly that Tagged
      Command queuing is supported.


7.2.6 Table 7-20:  Log Select command.

      PC field is missing.


7.2.10  p. 7-36, MODE SENSE

      Mode Sense (6) has the DBD bit while 7.2.11 Mode Sense (10) does
      not.

.pa
7.2.12  p. 7-43, very last sentence

      Are people still willing to live with the fact that the
      initiator cannot distinguish between unsupported mode parameters
      and a supported value of 0?


7.2.14  p. 7-48, Table 7-40 RECEIVE DIAGNOTIC RESULTS

      What is byte 1, bit 4?


7.2.15 Table 7-48,  p. 7-59.

      If a Check Condition status is sent by a device, is it valid to
      set the Sense bytes to No Sense, ASC=0, and ASQ=0 (setting only
      the Filemark and EOM bits and not using ASQ 01 or 02)?


7.2.15 Table 7-50,  p. 7-61

      Are ASCs => 80h Vendor Unique?  What about ASQs?


7.2.16  p. 7-65, SEND DIAGNOSTIC

      The PF bit is used in Send Diagnostic, but it isn't used in
      Receive Diagnostic.  If it was in Receive Diagnostic, then the
      device wouldn't need to remember the proper format for returning
      data.


7.4.4 Error Counters

      The counters for counting errors associated with Write, Read and
      Verify operations only provide for counting Corrected Errors.
      Why were Uncorrect Errors excluded?  They could be added as
      parameter code 0006h on each of the pages (Tables 7-68, 7-69,
      and 7-70).


7.6 VPD Parameter Pages

      Table 7-79 (p. 7-89) lists page 83h as Reserved.  Table 7-84 (p.
      7-94) has page 83h as the Unit Serial Number page (which should
      be page 80h according to Table 7-79).  If the Unit Serial Number
      page is 80h, then it should be moved ahead of pages 81h and 82h.


8.3 Table 8-45,  p. 8-57

      Mode Page 9 (Peripheral Device Parameters) is called reserved in
      the direct access section and is not listed in the Sequential
      Access section (Table 9-21,  p. 9-43).  They should be
      consistent.


8.3.1  p. 8-58, Mode Pages

      The mode pages for Direct Access devices are not in numerical
      order, while the Sequential Access list (9.3) and the generic
      list (7.5) are in numerical order.


8.4.1  p. 8-85, second paragraph from bottom

      Change "... in the Translated Format field" to "... in the
      translated format field".

8.4.2.1  p. 8-87, first and second paragraphs

      Change "(see table 8-_, Defect List Format)" to "(see table 8-_,
      FORMAT UNIT Defect Descriptor Format and Requirements)"

      This table has a lot of extra information.  Perhaps the table
      that was originally referred to (and was deleted in Rev. 5,
      p. 8-4) would be more appropriate here.