inexistent LUN

[Chuck Micalizzi <chuck.micalizzi@qlogic.com>]

Julian,

      Regarding version 5 of the iSCSI draft:

      On page 71, in Section 2.20 it states:

            It may happen that a target receives a message with a
            format error (e.g., inconsistent fields, reserved fields
            not 0, inexistent LUN etc.) or a digest error (e.g.,
            invalid payload or header)...

      I'm concerned about the iSCSI layer having the responsibility of
      validating the LUN. In my opinion this is the responsibility of the 
      SCSI layer. The target iSCSI layer should ignore the LUN field
      and let the SCSI layer validate the LUN. In SCSI, a command
      directed to a non-existent or missing LUN is not always an error.
      The SCSI Command Inquiry and Request Sense won't return
      "check condition" status if addressed to a non-existent LUN. 

chuck micalizzi