Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Incorrect Initiator Connector Detection

IP.com Disclosure Number: IPCOM000117178D
Original Publication Date: 1996-Jan-01
Included in the Prior Art Database: 2005-Mar-31
Document File: 2 page(s) / 71K

Publishing Venue

IBM

Related People

Howe, SM: AUTHOR

Abstract

A method for efficiently detecting an incorrect initiator connection to a Small Computer System Interface (SCSI) device is disclosed.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Incorrect Initiator Connector Detection

      A method for efficiently detecting an incorrect initiator
connection to a Small Computer System Interface (SCSI) device is
disclosed.

      The SCSI standard defines a method for a host device to queue
I/O processes on a target device.  Each I/O process existing at the
target device must be uniquely identified.  The standard defines both
untagged and tagged queueing.

      Untagged queueing allow one I/O process to be queued from each
individual host (initiator).  Several initiators may exist.  The I/O
process is identified by the initiator that initiated the I/O
process.

      Tagged queueing allow multiple I/O processes to be queued from
each individual initiator.  Several initiators may exist.  A I/O
process is identified by a tag value and the initiator that initiated
the I/O process.

      An incorrect initiator connection occurs when an I/O process is
created that does not have a unique identification.  This may be the
result of an initiator attempted to create a second untagged I/O
process.  It may be the result of an initiator using a value for the
tag that already exists for that initiator.  It may also be the
result of an initiator mixing untagged and tagged I/O processes.

      The problem is to create a way to efficiently detect these
conditions.  Previous solutions to this problem take too long to
execute.  For these solutions, the execution time grows as the number
of queued I/O processes grows.

      The problem is solved by the following method.  The method
involves keeping track of what I/O processes exist for each initiator
in a table.

      A table is created containing a bit for each combination of
initiator and tag.  The SCSI standard allows for sixteen possible
initiators and 256 different values for the tag.  This results in a
512 byte table.

      When a tagged ...