JP2008129709A - Task control method, data processing system, task control program, and its storage medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a task control method capable of efficiently processing tasks by detecting failure with the tasks, a data processing system, a task control program, and its storage medium. <P>SOLUTION: The task control method for processing data by allocating the process resources of a central processing unit 11 to a plurality of tasks during slice time includes a step (S2) for causing a task monitoring means 25 to detect the workload of data processing by the tasks, a step (S4) for detecting failure with the tasks based on the workload, a step (S6) for determining whether or not the slime time should be extended according to the amount of data waiting to be processed, if no failure with the tasks is detected, and a step (S7) for requesting a system management means 21 for allocating the slice time to extend the slice time if it is determined that the slice time should be extended. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a task control method, a data processing system, a task control program, and a storage medium for allocating processing resources of a central processing unit to a plurality of tasks during a slice time, and in particular, detecting a task abnormality. The present invention relates to a task control method for performing appropriate processing, a data processing system, a task control program, and a storage medium thereof.

  In multi-programming where multiple programs share the processing resources of one central processing unit (CPU), the appropriate slice time is set so that the time that the task that is a unit of processing occupies the CPU does not deviate extremely. Assigned to. As the slice time becomes shorter, it becomes easier to supply processing resources equally to each task. However, if the slice time is too short, task switching overhead occurs and efficiency decreases.

  Also, if the slice time is too long, it may not be preferable because a specific task may occupy the CPU. Therefore, if an appropriate slice time is set and the processing is completed during the slice time, the CPU is released from the task side. However, if the task is not completed in time, the CPU is occupied up to that slice time. It allows multiple tasks to be processed. FIG. 1 is a schematic diagram of this conventional task assignment. In FIG. 1, task A and task B are switched by an OS (Operating System) and processed by one CPU. Initially, the CPU is processing task A, but when an event occurs, the processing is switched to task B. When the processing of task B is completed, task B releases processing resources and switches to task A (E1 in FIG. 1), but if the processing is not completed, the CPU is occupied with the slice time as the upper limit, and the slice time has elapsed. Then, the OS forcibly switches to task A (E2).

  In addition, techniques for variably controlling the slice time have been proposed (see, for example, Patent Documents 1 to 3). The technique described in Patent Document 1 determines the system load state based on the CPU idle ratio, and optimizes processing efficiency by extending the slice time when the load is small and shortening when the load is large. Yes.

  The technique described in Patent Document 2 stores dispatch information for each task during a certain period, estimates the characteristics of each task from the dispatch information, and adjusts the slice time based on the estimation result. Since the dispatch information includes information indicating whether each task has used up the slice time or not, depending on the task characteristics, the task with a small number of times of use shortens the slice time and the task with a large number of times of use increases the slice time. It can be adjusted to the slice time.

In the technique described in Patent Document 3, a priority is assigned to each task, CPU processing resources are assigned according to the priority, and a task with a low load is assigned to a target load distribution determined from the total load distribution. By reducing the processing resources, the task is processed in such a manner that the frequency with which the processing resources are assigned to the task with the higher priority is increased.
JP-A-5-066954 JP-A-5-241860 JP-A-11-249913

  However, since all of the techniques of Patent Documents 1 to 3 extend the slice time for a task with a large load, there is a problem that the slice time is extended even if an abnormality occurs during task processing. . For example, when a task runs away, the slice time is extended to the maximum although data is not processed, and the runaway cannot be detected. That is, in the techniques of Patent Documents 1 to 3, even when the slice time is increased according to the load or the like, the processing is not always performed efficiently, and whether or not the processing is performed efficiently is detected. There is a problem that you can't.

  As a means of detecting system runaway, a watchdog timer that executes a program while detecting that it has not runaway due to a periodic timer interrupt is known, but a timer interrupt is generated even if a task runs away Therefore, the watchdog timer cannot detect task runaway.

  In view of the above problems, an object of the present invention is to provide a task control method, a data processing system, a task control program, and a storage medium capable of efficiently processing a task by detecting a task abnormality.

  In view of the above problems, the present invention provides a task control method for processing data by allocating processing resources of a central processing unit to a plurality of tasks during a slice time. A step of detecting an abnormality of the task based on a work amount, a step of determining whether or not to extend a slice time based on an amount of data waiting to be processed if a task abnormality is not detected, and an extension of a slice time If it is determined, the system management means for allocating the slice time requests to extend the slice time.

  According to the present invention, since the workload of data processing by a task is detected, it is possible to detect a task abnormality, and using this, it is possible to detect bugs, detect abnormal operations, and quantitatively limit processing limits. It becomes possible to know.

  It is possible to provide a task control method, system, task control program, and storage medium thereof capable of efficiently processing a task by detecting a task abnormality.

The best mode for carrying out the present invention will be described below with reference to the drawings.
First, an outline of the task control method of this embodiment will be described. FIG. 2 shows a schematic diagram of extending the slice time according to the present embodiment. An OS (Operating System) initially executes task A, but assigns processing resources of the central processing unit to task B when an event occurs (hereinafter simply referred to as “switching processing to task B”). Task A is, for example, interactive processing that waits for input or operation from the network, and the event that occurs is data reception or some user operation. The task B whose processing has been switched starts data processing corresponding to the storage and operation of the received data.

  In the present embodiment, the OS allocates a slice time min as the initial slice time of task B, and extends the slice time stepwise with the slice time max as the upper limit according to the data processing status. The slice time min is the shortest slice time given when the task is switched. If the processing of the task B does not end during the slice time min, the OS sets the slice time when the data processing condition satisfies a predetermined condition. Extend in stages. If the processing of task B does not end even if it is extended to the slice time max, the OS switches the processing to task A. As described above, since it is possible to detect an abnormal state of task B such as runaway by determining whether or not the data processing status satisfies a predetermined condition, the slice time is extended only when data processing is normally executed. can do.

  FIG. 3 is a hardware configuration diagram of a computer 10 (corresponding to the data processing system in the claims) 10 that implements the task control method of the present embodiment. The computer 10 is a microcomputer incorporated in, for example, a portable electronic device. The computer 10 is configured to include a CPU 11, a RAM 12, a ROM 13, a memory 14, an input device 15, a display control unit 16, and a memory card mounting unit 17 that are mutually connected by a bus.

  The ROM 13 stores programs and parameters for starting up the OS such as BIOS (Basic Input Output System). The memory 14 is a nonvolatile memory such as a flash memory, and stores various files such as an OS, a program, and a data file. The CPU 11 provides various functions by reading the OS and programs from the memory 14 and executing them, and comprehensively controls the processing performed by the computer 10. The RAM 12 is a work area for the CPU 11 to execute the OS and programs and temporarily store data. The input device 15 is used for inputting various operation instructions from a user such as a keyboard and a touch panel. The display control unit 16 draws on a display such as a liquid crystal with a predetermined resolution, number of colors, and the like based on screen information instructed by the program. For example, a GUI (Graphical User Interface) screen is formed, and various windows and data necessary for operation are displayed on the display. The memory card loading unit 17 is an interface that loads a storage medium 18 such as a flash memory, reads a program or data file stored in the storage medium 18, and writes a program or data file in the memory 14.

  The task control method of the present embodiment is realized by the CPU 11 executing a task control program. The task control program is stored in the storage medium 18 and distributed, or downloaded from the server by communication means and installed in the memory 14.

  FIG. 4 shows a functional block diagram of the data processing system in which the task control program is executed. When the CPU 11 executes the task control program, the system management unit 21, the task monitoring unit 25, and the task B23 (hereinafter simply referred to as task B) are realized. The system management means 21 is, for example, a dispatcher that determines a task to be executed and controls starting and stopping of the task, and can use the function of the OS. When the system management unit 21 activates task B, slice time Min is assigned to task B. Therefore, the processing start time of task B and the elapsed time after the assignment are known to the system management means 21.

  The activated task B processes the process waiting data 22 used for the process of the task B (for example, data compression, data storage, etc.), and outputs processed data 24. The process waiting data 22 is temporarily stored in the RAM 12 as a work area and sequentially sent to the CPU 11 according to the number of bits of the data bus. The sent data is processed by the CPU 11, temporarily stored in the RAM 12, and finally stored in the memory 14. Therefore, the data processing amount of task B is detected by monitoring the processed data 24 in the RAM 12 or the memory 14.

  The task monitoring unit 25 monitors the processing status of the data of the task B, notifies the system management unit 21 of the abnormality of the task B, or requests to extend the slice time.

[Task abnormality detection processing and slice time extension processing]
Processing of the task monitoring unit 25 will be described. FIG. 5 is a flowchart showing the processing procedure of the task monitoring unit 25.
(S1)
Since the task monitoring unit 25 is notified of the activation of the task B from the system management unit 21, when the system management unit 21 starts processing of the task B, the task monitoring unit 25 acquires the processing waiting data amount of the processing waiting data 22 and stores it in the RAM 12. To do.
(S2)
Further, the task monitoring unit 25 calculates the work amount of the task B. The task monitoring means 25 measures the amount of processed data by monitoring the processed data 24 from the start to the end of the processing of task B. Since the processed data amount is detected every cycle time, for example, if the data amount processed every cycle time is divided, the work amount per unit time can be obtained for each cycle time. Further, since the elapsed time since the task B is activated is known, the work amount of the task B per unit time may be calculated by dividing the processed data amount by the elapsed time.
(S3)
Then, the task monitoring unit 25 calculates an average of work amounts per unit time calculated for each cycle time (hereinafter referred to as average work amount). The task monitoring means 25 stores the average work amount per unit time updated at any time in the RAM 12. Thereby, the average work amount required for the task B can be detected.
(S4) "Abnormality judgment"
The abnormality determination is a process for determining whether or not the task B process is proceeding normally. By performing the abnormality determination prior to the slice time extension process, it is possible to detect the abnormality before extending the slice time, thereby preventing a reduction in the efficiency of the data processing system.

  The task monitoring unit 25 determines whether or not the work amount of the task B is a work amount estimated to be normally processed by the following formula.

Average work volume-Allowable value Q> Current work volume ... I
Expression I is an expression for determining whether or not the current work amount is so small that it is estimated to be abnormal as compared to the average work amount. Therefore, when the current work amount becomes extremely smaller than the average work amount so far, it is determined that an abnormality has occurred in task B. The allowable value Q is a value for allowing an error in the current work amount, for example, a value of several percent to several tens percent of the average work amount or a predetermined value. For example, when task B runs out of control and an abnormality occurs, the amount of work is extremely reduced, so that the abnormality of the task can be detected by the determination of Formula I.

(S5)
When it is determined that there is an abnormality (Yes in S4), the task monitoring unit 25 notifies the system management unit 21 of the task abnormality. Upon receiving notification of task abnormality, the system management means 21 stops task B and prohibits activation after that, and outputs a code or message notifying abnormality. Alternatively, the task B may be initialized and the process waiting data 22 may be processed from the beginning. Therefore, when an abnormality is detected, the process of the task monitoring unit 25 ends in step S5, and thereafter, the process proceeds to the process of the system management unit 21.

(S6)
When it is not determined to be abnormal (No in S4), the task monitoring unit 25 determines whether to extend the slice time.

Amount of data waiting to be processed> Threshold value ... II
Average work amount-allowable value P <current work amount ... III
Expression II is an expression for determining whether or not the amount of the task B processing waiting data 22 stored in the processing waiting data 22 remains so that the slice time needs to be extended. Formula III is a formula for determining whether or not the current work amount is so large that it is necessary to extend the slice time compared to the average work amount. The allowable value P is a value for allowing an error in the current work amount, for example, a value of several percent to several tens percent of the average work amount or a predetermined value. Further, it may be the same value as the allowable value Q.

Since it is not necessary to extend the slice time until the slice time min, the task monitoring unit 25 may determine whether or not to extend the slice time only after immediately before the slice time min.
(S7)
When the data processing status satisfies the conditions of Formula II and Formula III (Yes in S6), the task monitoring unit 25 requests the system management unit 21 to change the slice time. The system management means 21 that has received the request to change the slice time sets the time obtained by adding the predetermined time ΔT to the current slice time as the slice time. As a result, the slice time min is extended by ΔT. Therefore, when the conditions of the formulas II and III are not satisfied, the task monitoring unit 25 does not request the system management unit 21 to change the slice time.
(S8)
When the data processing status does not satisfy the conditions of Formula II and Formula III (No in S6), or after extending the slice time, the task monitoring unit 25 determines whether the processing of the task B is completed.

  When the processing of task B is completed, task B returns the processing to system management means 21. When the slice time before or after the extension has elapsed, the system management unit 21 stops the task B. Therefore, in these cases, the processing of the task monitoring unit 25 ends, the system management unit 21 switches the processing of the CPU 11 to the task A, and the CPU 11 resumes the processing of the task A.

  If task B has not ended (No in S8), task monitoring means 25 repeats the processing from step S1. Accordingly, even if the slice time is extended by ΔT, if the processing waiting data 22 remains more than the threshold value and the current work amount is larger than “average work amount−allowable value P”, the slice time is again determined in step S5. Is extended.

  As the slice time is extended, the judgment of the processing status according to Formula II and Formula III is less likely to be Yes. Therefore, the slice time is preferentially extended for tasks with a large amount of data to be processed, and the efficiency of the data processing system Will improve.

  As described above, according to the task control method of the present embodiment, when an abnormality of task B is detected, the slice time is not extended. Therefore, since the slice time is extended only for a task having a large amount of data to be processed without extending the slice time due to an abnormality, the efficiency of the data processing system can be improved.

  In addition, at the program development stage, it becomes easy to identify bugs hidden in the task by detecting an abnormality, and the system development progress speed increases. Further, if the program is in a practical stage, it is easy to detect when a task runs away, and after the abnormality is detected, the system management means 21 controls task B, thereby preventing other tasks from being affected and deadlocks and the like. Thus, the influence on the data processing system can be minimized. Further, since the task monitoring unit 25 detects the work amount of the task B, the limit value of the task B process can be known quantitatively.

It is the schematic of the adjustment of the conventional slice time. It is the schematic of adjustment of the slice time of this embodiment. It is a hardware block diagram of the computer which implement | achieves the task control method of this embodiment. It is a functional block diagram of a task control program. It is a flowchart figure which shows the process sequence of a task monitoring means.

Explanation of symbols

10 Computer 11 CPU
12 RAM
13 ROM
14 Memory 15 Input Device 16 Display Control Unit 17 Memory Card Mounting Unit 18 Storage Medium 21 System Management Means 22 Processing Wait Data 23 Task B
24 Processed data 25 Task monitoring means

Claims (5)

In a task control method for processing data by allocating processing resources of a central processing unit to a plurality of tasks during a slice time,
A task monitoring means for detecting a work amount of data processing by the task;
Detecting an abnormality of the task based on the workload;
If no abnormality of the task is detected, determining whether to extend the slice time based on the data processing status; and
If it is determined to extend the slice time, a step of requesting the system management means for assigning the slice time to extend the slice time;
A task control method comprising: In the step of detecting an abnormality of the task, when the work amount is smaller than an average of past work amounts, it is detected as abnormal.
The task control method according to claim 1, wherein: In a data processing system in which a central processing unit processes a plurality of tasks,
System management means for allocating processing resources of the central processing unit to a plurality of tasks during a slice time;
Task monitoring means for monitoring the task,
The task monitoring means detects an abnormality of the task based on the workload of data processing by the task, and if no abnormality is detected, requests the system management means to extend the slice time based on the data processing status.
A data processing system characterized by that. Detecting the work of a task to which processing resources of the central processing unit are allocated and processing data during a slice time;
Detecting an abnormality of the task based on the workload;
If no abnormality of the task is detected, determining whether to extend the slice time based on the data processing status; and
If it is determined to extend the slice time, requesting the system management means for allocating the slice time to extend the slice time;
Is realized by a computer.   A computer-readable storage medium storing the program according to claim 4.

Images