Control system for tightly-coupled multiprocessing units

Claim 1

A control system for tightly-coupled multiprocessing units having a main storage for storing a shared program and shared data, a plurality of processing units each referencing the shared program and referencing/updating the shared data, a function which is such that when a program being executed by a first one of the processing units intends to reference the shared program and reference/update the shared data under a condition wherein a program being executed by a second one of the processing units is referencing the shared program and referencing/updating the shared data, the program being executed by the first processing unit waits for completion of referencing of the shared program and referencing/updating of the shared data by the program, which is being executed by the second processing unit, by spin lock, and an interrupt signal generating function which notifies one of the processing units of a stop in processing of the other processing unit, the control system comprising: processing unit stopping means for, after a register associated with a first program which is being executed by a first processing unit and a first interrupt flag have been saved in a particular area of the main storage, causing the first processing unit to produce the interrupt signal to thereby stop the first processing unit; saving means for temporarily interrupting a second program which is being executed by a second processing unit connected to the first processing unit which has accepted the interrupt signal and for saving a register associated with the second program and a second interrupt flag in the main storage; program resuming means for resetting the first interrupt flag stored in the main storage by the processing unit stopping means, for recovering the register associated with the first program in the second processing unit, and for causing the first program to be resumed; and program switching means for, when the first program resumed by the second processing unit has entered into a waiting condition by spin lock in order to reference the shared data in the main storage which is being referenced/updated by the second program, resetting the second interrupt flag when an interrupt flag is set in a particular area of the main storage and for, after saving in a particular area of the main storage the register associated with the first program being executed by the second processing unit and the first interrupt flag, recovering the register associated with the second program from the main storage to the second processing unit to continue processing the second program; whereby when the program being executed by the second processing unit has entered into a waiting condition by spin lock after the above procedure in order to reference the shared data which is being referenced/updated by another program, the first and second programs are switched from one to the other when an interrupt flag is set in a particular area of the main storage, so that the first and second programs are alternately executed while avoiding dead-lock until contention in referencing/updating the shared data terminates.