JPS62290958A - Task schedule system - Google Patents

Abstract

PURPOSE:To prevent an idle state of a processor to be caused by the conflict of resources by processing in parallel as much as possible when many tasks waiting for resources exist. CONSTITUTION:A task schedule 1 extracts tasks and schedules them with reference to a resource queue 3 as long as other processors are not using resources after checking a source busy flag 2. If the queue 3 is idle, the tasks are schedules by means of a normal task queue 4. When the occupation of resources is released via a task, the queue 3 is set in a resource busy mode. Then the procedure proceeds to the schedule 1 to schedule the tasks requesting other resources. In a release mode the one's own task is interrupted if a task kept in a resource queuing state is detected. Then the schedule 1 is activated to start the relevant task.

Description

Detailed Description of the Invention 3. Detailed Description of the Invention [Field of Industrial Application] The present invention relates to a task scheduling method in a multiprocessor system that shares memory.

[Conventional technology]

In task scheduling systems, tasks have traditionally been managed based on the priority at the time of task generation or the priority specified by the user. Also, even if there are a large number of tasks waiting for resources on a first-in, first-out basis, if they are at the back of the queue or have low priority, they will not be started immediately1.

[Problem that the invention seeks to solve]

Even if there are multiple central processing units, only one task that uses these resources can be executed at a time, and if the resource usage rate of each task is high, there is a possibility that it will become a resource bottleneck in the future even if it is fully used now. There is. Even if priority and order are ignored, it is possible to prevent future bottlenecks and improve throughput by ensuring that a task that uses resources is processed by one of the processors.

[Means for solving problems]

The present invention adds a resource waiting queue to the conventional task queue. When there is resource contention, tasks are placed in the resource waiting queue. When scheduling a task, check whether the resource is in use. If the resource is not in use, schedule the task in the resource waiting queue first. Additionally, when a task releases a resource, if there is another task waiting for the resource, the above rescheduling is performed to switch to that task.

[Effect]

According to the configuration of the present invention, when a task uses an exclusive resource at a high rate in a task scheduler on a processor that shares memory, when there is a task waiting for the resource, one of the processors uses the resource. By executing the task, it is possible to prevent the processor from being idle due to resource contention that may occur in the future.

[Embodiments of the invention]

An embodiment of the present invention will be described below. FIG. 1 is a schematic diagram of a task scheduler of the present invention. The task scheduler is activated to select and activate the next task to be processed, such as when a resource is released at the end of each task or when an input/output request is made. Requests for resources are made only from tasks that have been assigned a processor, and by releasing resources whenever a task leaves the processor, such as when a task ends or when making an input/output request, resources can only be used by the task that is currently assigned a processor. is occupied.

The task scheduler 1, whose processing flow is shown in FIG.
First, if the resource in use flag 2 is checked and the resource is not in use by another processor, the resource waiting queue 3 is checked and a task is taken out and scheduled. If the resource waiting queue 3 is empty, a task is scheduled from the normal task queue 4. If the resource is in use, scheduling is performed from the normal task queue 4.

When releasing resources from a task, do as follows. As shown in FIG. 3, if the resource is in use at the time of an occupancy request, it enters the resource waiting queue 3.

It then goes to the task scheduler to schedule the task that is requesting another resource. At the time of release, as shown in FIG. 4, if there is a task in the resource waiting queue, the current task is interrupted and the task goes to the task scheduler to start that task.

〔Effect of the invention〕

According to the present invention, when there are many tasks waiting for resources, by processing them in parallel to the maximum extent possible, it is possible to prevent processors from idling due to resource contention.

[Brief explanation of drawings]

FIG. 1 is a process outline diagram of the task scheduler, FIG. 2 is a flowchart of the process of the task scheduler, FIG. 3 is a flowchart of the process when resources are occupied, and FIG. 4 is a flowchart of the processes when resources are released. 1...Task scheduler, 2...Resource in use flag, 3...Resource waiting queue, 4...Normal task queue.

Claims (1)

[Claims] 1. In a schedule that schedules multiple tasks that exclusively use only one resource on a multiprocessor that shares main memory, one of the multiprocessors always uses the resource while at least one task is using the resource. A task scheduling method that is characterized by scheduling tasks that use resources to be executed.