Browse Prior Art Database

Control Unit Timeout for Interface Failures

IP.com Disclosure Number: IPCOM000105258D
Original Publication Date: 1993-Jul-01
Included in the Prior Art Database: 2005-Mar-19
Document File: 4 page(s) / 136K

Publishing Venue

IBM

Related People

Lynch, KR: AUTHOR [+2]

Abstract

Control units which attach to multiple data processing systems through the use of multiple I/O interfaces may become "hung" (i.e., unavailable for use) due to malfunctions of one or more of the attaching data processing systems or their attaching I/O interface(s). However, circumvention of the hang condition is possible through the application of additional logic to the control unit which detects and recovers the hang condition.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 39% of the total text.

Control Unit Timeout for Interface Failures

      Control units which attach to multiple data processing systems
through the use of multiple I/O interfaces may become "hung" (i.e.,
unavailable for use) due to malfunctions of one or more of the
attaching data processing systems or their attaching I/O
interface(s).  However, circumvention of the hang condition is
possible through the application of additional logic to the control
unit which detects and recovers the hang condition.

      Typically, when a control unit is communicating with one of the
attached data processing systems via that systems attaching I/O
interface[1].  The control unit maintains allegiance to the I/O
interface and is unavailable to other attaching I/O interfaces until
the allegiance condition is cleared or reset.  However, if the system
or the attaching I/O interface malfunctions and is unable to complete
or terminate the communications sequence with the control unit, the
allegiance condition is not cleared and the control unit becomes
unavailable (i.e., hung) for use by the other attaching I/O
interfaces to either the same system or to other systems.  Typically,
recovery of this hang condition in the control unit requires either
recovery of the malfunctioning system, the malfunctioning I/O
interface, or resetting of the control unit.  All of these actions
usually have a negative impact to the other attaching systems in
terms of availability of the control unit and its attached I/O
devices.  Additionally, the integrity of the data on the attaching
I/O devices may be impacted if the control unit is reset in order to
clear the hang condition.

      Circumvention of the hang condition is possible if additional
logic is added to the control unit which detects the malfunction and
activates logic which reports the error and clears the allegiance
condition.  Specifically, logic which may be either microcode or
hardware can be added which performs the following functions:

      Monitors the communications sequences between the control unit
and the attaching I/O interface for a specified amount of time.  The
amount of time is greater than the I/O interface should take to
continue, complete, or terminate the I/O sequence in progress.

      If the I/O sequence has not completed in the given timeout
period established for the sequence, the control unit activates logic
which causes the allegiance condition to be reset.  Additionally, the
control unit sets initiative within itself to report the resetting of
the allegiance condition to the subject I/O interface when the
interface subsequently becomes operational.

      From a channel or channel subsystem viewpoint, the allegiance
conditions which can be recovered using this technique include Active
Allegiance, Dedicated Allegiance, and Working Allegiance conditions
as defined in [2].

      From a control unit viewpoint, the allegiance conditions which
can be recovered using this technique in...