JPH04178740A - Task scheduling system - Google Patents

Abstract

PURPOSE:To prevent tasks from being left unexecuted and to shorten the processing time by changing the present value of a task executed based on the execution priority level after task generation and increasing the present values of the other tasks, which are not executed, in such range that they do not exceed an upper limit value. CONSTITUTION:Tasks are scheduled based on task control blocks of tasks generated by task management. Execution priority levels of tasks T1 to T4 are (n), (m), (l), and (k) respectively, and initial values and present values are equal to each other, and the task T1 has the highest execution priority level (m). Present values of the other tasks T2 to T4 in the state waiting for execution are increased by one and are (n), (m), and (l) respectively. When the task T2 goes to the execution state after the task T1, present values of the other tasks T3 and T4 are increased by one and are (n) and (m) respectively. Thus, tasks having lower execution priority levels are prevented from being left unexecuted, and the turn-around time of the task of urgency is shortened.

Description

DETAILED DESCRIPTION OF THE INVENTION OBJECTS OF THE INVENTION (Field of Industrial Application) The present invention relates to improvements in the scheduling of tasks in a multitasking operating system (hereinafter referred to as O8) of a computer.

(Prior Art) In a computer system, a job given to the system by a user is divided into units (tasks) that are processed within the system and executed. In general, an O8 that has multiple tasks has a function to create and delete tasks (hereinafter referred to as the task management function), and a function to manage the status of the created tasks and control their execution (hereinafter referred to as the dispatcher function). ). Furthermore, as a scheduling method for controlling execution by managing the state of tasks, a method using execution priorities of tasks is generally put into practical use.

In this method, first, when a task is generated, an execution priority is set by a task management function, and a dispatcher function performs scheduling based on this execution priority. Each task is executed sequentially in accordance with this execution priority order, starting from the one with the highest priority.

In addition, when scheduling tasks, the so-called time slicing method is used, in which tasks with the same priority are executed sequentially using FIFO, etc., and each task is executed while limiting the execution time to prevent a specific task from occupying the CPU. Some are used together.

(Problem to be Solved by the Invention) However, in this kind of scheduling based on execution priority, there is a problem of task slump, that is, a state in which tasks with lower priority are not executed because tasks with higher priority are executed first. This may occur.

As a means to prevent this, there is a method of lowering the execution priority and scheduling after a task has been executed for a certain period of time.

However, this approach can deprioritize urgent tasks and increase turnaround time.

Similarly, by limiting the execution time of tasks by time slicing in order to prevent a specific task from occupying the CPU, there is a problem of increasing the turnaround time of urgent tasks.

The present invention has been made to solve the above-mentioned problems, and it is an object of the present invention to provide a task scheduling method that can prevent tasks from sagging and shorten the turnaround time of urgent tasks.

[Configuration of the Invention (Means for Solving the Problem) In order to solve the above problem, in an operating system that schedules multiple tasks and manages the execution of each task, when a task is created, the execution priority of each task is determined. As such, an initial value and a current value to be changed during scheduling are set, and the initial value can be given an execution priority that exceeds a predetermined upper limit of the allowable range of the current value that is sequentially changed. At the time of scheduling, the current value of the task being executed is changed to the initial value based on the execution precedence order after the task is created, and the current value of other tasks that are not executed is increased within a range that does not exceed the upper limit. The method is to change.

(Operation) By taking the above-described measures, the present invention raises the current value of even a task with a low execution priority, and returns the current value of the task to be executed to its initial value, thereby increasing the current value of the task. It is possible to prevent sinking. In addition, by raising the current value within a range that does not exceed the upper limit and making it possible to assign an initial value that exceeds the upper limit of the current value, urgent tasks can be executed with priority and turnaround time can be reduced. can be made smaller.

(Example) An example of the present invention will be described below with reference to FIGS. 1 to 3.

In these figures, 1 is a task ready-kinny for task scheduling, and the symbols therein indicate execution priority, with higher execution priority (faster execution) in the direction of the arrow. Also, M P R (
maxjm priority) indicates an upper limit value that is the upper limit of the allowable range of the current value that is successively changed. 2 indicates a task being executed.

T1 to T4 are diagrams showing tasks to be scheduled, in which symbols surrounded by 0 indicate execution priority (
(current value, initial value).

FIG. 1 is a diagram showing the initial state of task scheduling.
FIG. 2 is a diagram showing the state of task scheduling when task T1 is executed first, and FIG. 3 is a diagram showing the state of task scheduling when task T1 is executed first.
FIG. 2 is a diagram showing the state of task scheduling when 2 is executed.

First, each task targeted for task scheduling requires an area (hereinafter referred to as a task control block) in which information for controlling the task is stored. This task control block is secured in memory by a function called task management. Here, securing this task control block and initializing its contents is called task generation. Further, an area for storing the initial value and current value of the execution priority is reserved as a component in this task control block.

Task scheduling is performed based on the task control block of each task generated by task management. Also, task scheduling manages tasks, and this management is performed in an execution state in which a CPU is assigned and processing is performed, and in an execution wait state in which a task waits for a CPU to be assigned.

To manage the execution waiting state, the task is registered in a management area called task ready queue 1 in FIG. 1 (in this embodiment, tasks T1 to T4 are managed).

Each of the tasks T1 to T4 has an execution priority of m.

m,, lit, and the initial value and current value are the same.

Further, the execution priority of task T1 is highest, m.

Next, FIG. 2 shows a task scheduling state in which task T1 is assigned a CPU and is in an execution state. As shown in the figure, other tasks T2~
T4 is raised by one by each current value, each n
, m, p.

Furthermore, as shown in FIG. 3, when the task T2 or T] enters the execution state, the current values of the remaining tasks T3 and T4 are increased by one, and become n and m, respectively. Here, task T1, which has been in the execution state so far, is interrupted, its current value n is returned to the initial value m, and it is connected next to task T4.

Next, an example of the task scheduling method of the present invention will be explained according to the flowchart shown in FIG.

Incidentally, FIG. 4 is a diagram showing an example of the flow of a dispatcher that schedules tasks. Now, when a system interrupt occurs (software interrupts and hardware interrupts when a task in progress finishes or is interrupted by a time slice), the dispatcher's processing starts. .

At this time, the dispatcher first selects the current value of the task currently being executed (TI in Figure 2) and the upper limit MPR of the current value.
(Step Sl).

If the current value is greater than or equal to MPR, it is assumed that an urgent task is being executed, and the task is terminated without scheduling. As a result, the urgent task is put into execution again. On the other hand, if it is smaller than MPR,
The currently executed task is extracted, and the process moves to step S2 to perform task scheduling processing. That is, the dispatcher searches the task ready queue 1 that manages the status of tasks and retrieves the task to be executed next (T2 in FIG. 2).

Thereafter, in step S3, a CPU is assigned to the task to be executed (task T2 in FIG. 2) and brought into an execution state (disbatch). Furthermore, as shown in step S4, it is checked whether there are any other tasks in the task ready queue 1. If there are no other tasks, step S6 is executed, and if there are any other tasks, step S5 is executed (tasks T3 and T4 in FIG. 2).

In step S5, processing is performed to increase the current value of the task in the execution waiting state (this means changing the current values of T3 and T4 in FIG. 2 to n and m)'.

This change processing is performed for all remaining tasks in the task ready queue. Here, the current value is changed in descending order of execution priority for tasks. However, the upper limit MPR
Do not change the current value exceeding .

On the other hand, in step S6, the task taken out in step S1 is re-registered in the task ready queue and placed in a waiting state for execution (TI in FIG. 3). At this time, the current value of the re-registered task is returned to its initial value and registered in the corresponding execution priority queue (in the example in Figure 3, the current value is changed from n to m, and the task is placed next to task T4. Connecting).

Therefore, according to the method of the embodiment described above, by raising the priority of the task in the execution waiting state and returning the current value of the task whose execution has been interrupted to the initial value and placing it in the execution waiting state, execution priority can be given. It is possible to prevent tasks with lower rankings from sinking.

Note that in this embodiment, the task being executed in step S1 is M
Is it judged whether or not it is greater than or equal to PR? This is a process to exclude time slices from being applied to urgent tasks that have a current value greater than or equal to MPR. When applying, this step S1 may be deleted. It is also possible to change the current value of each task in the execution waiting state by task management or the like.

Next, FIGS. 5 and 6 are diagrams illustrating an example of processing an urgent task.

First, FIG. 5 is a diagram showing a state in which an urgent task TX is generated by the task management and registered in the task ready queue 1 by the dispatcher in the scheduling state shown in FIG. 2, and its execution priority is given. The order is a value X greater than the current upper limit value MPR, and is connected to a queue higher than the execution priority MPR.

Next, FIG. 6 is a diagram showing a scheduling state when an urgent task TX is executed. As shown in the figure, the urgent task overtakes task T2, which had previously had the highest execution priority, and enters the execution state preferentially. In this way, by assigning an execution priority that exceeds MPR, which is the upper limit of the current value, as an initial value, urgent tasks are executed preferentially. On the other hand, the current value of task T2 remains at the same value n as in FIG. This is because the current value n of task T2 exceeds the upper limit value MPR, so it is maintained at the same value n. By controlling the current value so that it does not exceed the upper limit value MPR in this way, it is possible to prevent the influence on urgent tasks.

Therefore, according to the embodiments described above, priority is given to urgent tasks by controlling the current value so that it does not exceed the upper limit value MPR and by allowing the initial value to be set to a value that exceeds the upper limit value MPR. This makes it possible to implement the system in a timely manner, thereby shortening the turnaround time.

Note that the present invention is not limited to the above-mentioned embodiments, and can be implemented with various modifications without departing from the gist thereof.

[Effects of the Invention] As explained above, according to the present invention, during task scheduling, the current value of the task being executed is returned to the initial value based on the execution priority after task generation, and the current value of other tasks that are not executed is By taking measures to raise the value within a range that does not exceed the upper limit, it is possible to prevent tasks with low execution priority from becoming sluggish.

In addition, the current value has an upper limit value indicating the upper limit of the allowable range, and by making it possible to set a value that exceeds the upper limit value as the initial value,
Reduce turnaround time for urgent tasks
It can be reduced.

[Brief explanation of the drawing]

FIGS. 1 to 4 are diagrams for explaining two embodiments of the present invention, in which FIG. 1 shows the initial state of task scheduling, and FIG. 2 shows when task T1 is executed for the first time. 3 is a diagram showing the state of task scheduling when task T2 is executed next. FIG. 4 is a diagram showing an example of the flow of a dispatcher that schedules tasks. 5 and 6 are diagrams for explaining other embodiments of the present invention, and FIG. 5 shows a state in which an urgent task is registered in the scheduling state shown in FIG. The diagram shown in FIG. 6 is a diagram showing a task scheduling state. 1...Task ready queue for task scheduling, M, PR...Upper limit of current value, 2...Task in progress, T1-T4...Task to be scheduled, TX...Urgent tasks that require. Applicant's agent Patent attorney Takehiko Suzue-11112□ r 7

Claims (1)

[Claims] In an operating system that schedules a plurality of tasks and manages the execution of each task, when a task is generated, an initial value and a current value to be changed during scheduling are set as the execution priority of each task. It is possible to give an execution priority that exceeds a predetermined upper limit of the allowable range of the current value, which is set and the initial value is changed sequentially, and when task scheduling, the task is executed based on the execution priority after the task is created. A task scheduling method characterized in that the current value of a task currently being executed is changed to an initial value, and the current value of another task that is not executed is changed in the direction of raising it within a range that does not exceed an upper limit value.