New Log page for Information Exceptions (with note)

[George Penokie]

* From the SCSI Reflector, posted by:
* "George Penokie" <GOP at RCHVMP3.VNET.IBM.COM>
*
I received the attached comment from our drive developers.  I believe
it points out several reason for NOT implementing a new log page for
information exception reporting.

Bye for now,
George





 Subject: New Log Page and Parm for Information Exceptions

 The need for such a Log parm is a bit mind boggling. I can't conceive
 of a hardware or software reason why a system developer couldn't design
 their code to issue a Request Sense command rather than a Log Sense
 command. Either will probably cause them to redesign their software.

 The other reason that this is mind boggling is that the only "Method"
 this would help in is when MRIE = 6h. If a system can't issue Request
 Sense commands without getting a Check Condition for some goofy
 architecture reason, they shouldn't set MRIE = 6h !

  When MRIE = 0h, the system is in all likelihood NOT going to go looking
                  for failures anyway.

  When MRIE = 1h, Asynchronous Event reporting is on. Therefore Sense
                  Data is generated by definition.

  When MRIE = 2h, 3h, 4h, and 5h = The target returns Check Condition.
                                   Thus no problem. It is not
                                   "unsolicited".

 Thus if a system STILL wants to use these features but can't handle
 asynchronous event notification, and can't handle Sense data from check
 condition (highly unlikely) then this proposal looks innocuous to me.
 It sure looks like a useless cost adder to me though. Especially when
 folks are trying to keep the cost of SCSI down.

 But if for some reason it does go through the committee we may want
 to consider allowing the target to return a larger parameter.  The
 code/qualifier doesn't really give the whole picture with our drives.
 I would suggest allowing Targets to return more bytes, if they have
 them.  That way we could return data as to which PFA test actually
 failed to aid in failure analysis.  There are only 2 qualifiers and
 we have many more tests that constitute our "Information Exceptions".