X3T9.2/88-152

Date:     November 8, 1988
To:       X3T9.2

From:     John Lohmeyer, NCR Corporation
Subject:  Proposed revision to the SAVE DATA POINTER message

I feel that the committee acted too hastily in accepting the new rules in 88-
120 as requirements.  I have no problem with accepting the rules as 
recommendations, but making them requirements causes unnecessary overhead for 
systems that do not check the final value of the data pointer.  It also causes 
a problem for the few systems that cannot save data pointers.

I propose the following wording to replace the current SAVE DATA POINTER 
section:



5.6.21. SAVE DATA POINTER

  This message is sent from a target to direct the initiator to save a copy of 
the current data pointer for the present nexus.  (See 5.4 for a definition of 
pointers.)

  It is recommended that if a target restores the status pointer, it send a 
SAVE DATA POINTER message before sending the RESTORE POINTERS message.  It is 
also recommended that if any data transfer has occurred for the I/O process, 
the target send a SAVE DATA POINTER message before sending the final 
DISCONNECT message.  These recommendations do not apply if the initiator has 
accepted a MODIFY DATA POINTER message for the logical unit of the I/O 
process.

  IMPLEMENTORS NOTE:  The above recommendations permit certain host adapters 
  to perform an integrity check on the final value of the current data 
  pointer.  Because the final value of the data pointer (saved or current) is 
  not guaranteed to be at the end of the actual data transferred, it is 
  recommended that device driver software obtain residue information from the 
  information field in the REQUEST SENSE data for those device types that 
  report residue information (e.g., sequential-access devices).