Browse Prior Art Database

Asynchronous Available Call Queue Replenish After Task Switched Out using Hardware Trap

IP.com Disclosure Number: IPCOM000110523D
Original Publication Date: 1992-Dec-01
Included in the Prior Art Database: 2005-Mar-25
Document File: 1 page(s) / 43K

Publishing Venue

IBM

Related People

Dickes, SM: AUTHOR [+5]

Abstract

This article describes the use of a semaphore to another task to keep information from a current task from being lost. Specifically, this article describes a means to prevent the system from hanging when another interrupt steals the last Available Call return element Queue (ACQ) element before the task can do a Supervisor Link (SVL) requiring an ACQ element to get more ACQ elements. In general, this is a method of converting a synchronous event (detecting of the monitored Call Return Element (CRE)) to an asynchronous event (trap).

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 76% of the total text.

Asynchronous Available Call Queue Replenish After Task Switched Out using Hardware Trap

       This article describes the use of a semaphore to another
task to keep information from a current task from being lost.
Specifically, this article describes a means to prevent the system
from hanging when another interrupt steals the last Available Call
return element Queue (ACQ) element before the task can do a
Supervisor Link (SVL) requiring an ACQ element to get more ACQ
elements.  In general, this is a method of converting a synchronous
event (detecting of the monitored Call Return Element (CRE)) to an
asynchronous event (trap).

      CRE replenish mechanisms consist of a monitored CRE near the
end of the available CRE chain.  A task using the monitored CRE would
be responsible for replenishing the CRE supply.

      To avoid the possibility that a task encountering the monitored
CRE then requests another CRE when there are no more CREs available
(which may cause a hang state), the task that receives the monitored
CRE sets flags (before the completion of the unit of operation and
while there is no chance that this task will request another CRE) to
force execution of separate replenish code after the next Endop.
(This Endop may be either normal continued execution or it may be
after the task dispatcher has been invoked and this task has just
been switched out).

      This can be accomplished as follows:
1.  The task that encounters the monitored CRE signals...