Method for time partitioned application scheduling in a computer operating system

A method for enforcing a schedule for a plurality of concurrently executing applications, each of which comprises at least one thread which consumes processing time overhead by invoking a computer operating system service, the method comprising the steps of:

• associating, with each said thread, a recurring corresponding time period;
• establishing, for each said thread, an initial corresponding time budget;
• accounting for the processing time overhead consumed by each said thread during each said corresponding time period by subtracting, from each said corresponding time budget, the sum of the time required to perform a context switch out of an executing said thread plus the greatest length of time for which the operating system disables interrupts while executing a critical section of instructions;
• terminating execution of each said thread when a thread's corresponding budget reaches a terminal value for the corresponding time period;
• and
• replenishing the initial corresponding time budget only at the first time each said thread is made ready to execute during each said corresponding time period.

A method for accounting for the processing time overhead consumed by responding to an asynchronous interrupt generated on behalf of one of a plurality of concurrently executing applications, wherein the interrupt is treated as a thread which consumes processing time overhead by invoking a computer operating system service, the method comprising the steps of:

• associating a recurring time period with the thread;
• establishing an initial time budget for the thread;
• subtracting, from the budget, the processing time overhead consumed by the thread during the time period;
• masking the interrupt unless the thread is waiting for an occurrence of the interrupt;
• if the thread's budget is depleted, then continuing to mask the interrupt until the beginning of the next said time period;
• and
• replenishing the initial time budget only during the first time the thread is made ready to execute during each subsequent said time period.

A method for accounting for processing time overhead consumed by a computer operating system in responding to an asynchronous interrupt, the method comprising the steps of:

• associating a recurring time period with the interrupt;
• establishing an initial time budget for the interrupt;
• subtracting, from said budget, the time utilized by the operating system in handling an occurrence of the interrupt;
• masking the interrupt until the beginning of the next said time period, if the budget is exhausted during a present time period;
• and
• replenishing the initial budget for the interrupt at the beginning of the next said time period associated with the interrupt.