JPH04367035A - Real time processor for plural tasks - Google Patents

Abstract

PURPOSE:To change the priority order of a prescribed task when the prescribed conditions are satisfied and also to optimize the priority order changing conditions by a learning function in response to the working environment, in regard of a real-time multi-task OS which processes plural tasks in real time. CONSTITUTION:An executing task switch means 2 switches the executing tasks based on their priority order stored in a task priority order storage means 1. When the conditions of a condition deciding means 3 are satisfied, the task priority order is changed by a priority order changing means 4. The changing state of the priority order is leant by a condition learning means 5 so that the priority order changing conditions are optimized in accordance with the working environment. Thus the executing order of tasks is optimized in accordance with the working environment and the real-time responsiveness is improved.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to real-time multi-task processing apparatus, and more particularly to improvements in real-time multi-tasking operating systems for computers.

0002

[Prior Art] Conventionally, operating systems (O
S) is widely used. Of these, real-time
A multitasking OS is an OS that has a function of processing multiple tasks (multitasking) in real time, and is used in factory process control and building management systems. In real-time multitasking OS,
As an algorithm for determining the execution order of each task, either a priority method or a round robin method, or a combination of both are generally used. In the priority system, tasks with preset higher priorities are executed first, and tasks with lower priorities are executed later. In addition, in the round-robin method, tasks that are to be executed in parallel are given equal C
A PU time is allocated, and any task that does not finish within that time is forcibly terminated and placed in a waiting state for execution, and the next task is executed.

FIG. 4 illustrates an algorithm for determining the task processing order in a conventional standard real-time multitasking OS. Now, multiple tasks T1, T2, T3, T
The priorities of 4, T5, T6,...
Assume that it is as shown in FIG. That is, task T1,
Tasks T2 and T3 have the same rank and are ranked first, task T4 has the second rank, and tasks T5 and T6 have the same rank and have the third rank. The tasks T1, T2, and T3 in the first order are sequentially executed in a time-sharing manner as shown by arrow A. Then, when the execution of the tasks T1, T2, and T3 in the first order is completed, the task T4 in the second order is executed next, as shown by arrow B. When the execution of the task T4 in the second order is completed, the tasks T5 and T6 in the third order are then executed as shown by the arrow C.
are executed sequentially in a time-sharing manner.

[0004] In this way, in a conventional real-time multitasking OS, when attempting to process multiple tasks in real time, the execution order is determined by the priority assigned to each task in advance at the time of generation. Things are processed sequentially. Furthermore, between tasks with the same priority, FIFO
According to the (First In First Out) algorithm, tasks are sequentially processed so that the task input first is executed first. Note that the processing wait order of tasks is determined when a system call or interrupt processing occurs.

[0005]

[Problem to be solved by the invention] Conventional real-time
In a multitasking OS, when a plurality of tasks issue operation requests at the same time, the processing is consistently executed according to the above-described procedure. However, in reality, when certain conditions are met,
The subsequent processing can be predicted empirically. for example,
In a building management system, when an alarm occurs, the user should first investigate the location of the alarm, read a manual for countermeasures, and then take the appropriate countermeasures. In such a case, a processing task to check the location of the alarm, a processing task to view the countermeasure manual, and a processing task to execute the countermeasure are
It is desirable to be in a waiting state with a higher level of priority than normal. However, in the conventional OS, the execution order of tasks is determined according to the priority at the time of generation, so the response may seem slow.

[0006] The present invention has been made in view of the above points, and its purpose is to provide a real-time multitasking OS that processes multiple tasks in real time, when predetermined conditions are met. The objective is to improve real-time responsiveness by changing the priority of certain tasks.

[0007]

[Means for Solving the Problems] In order to solve the above problems, in the real-time processing device for multiple tasks according to the present invention, as shown in FIG. Task priority storage means 1 for storing priorities
and an execution task that refers to the memory contents of the task priority storage means 1, executes tasks with higher priorities before tasks with lower priorities, and executes tasks with the same priority cyclically in a predetermined order. In a CPU control device comprising a switching means 2, a condition determining means 3 for determining a predetermined condition;
a priority changing means 4 for changing the priority stored in the task priority storing means 1 when a predetermined condition is determined by the condition determining means 3; and a condition learning means 5 for learning the conditions for changing the priority. It is characterized by having the following.

[0008]

[Operation] In the present invention, when processing a plurality of tasks in real time, based on the priority of each task stored in the task priority storage means 1, a task with a high priority is executed first, and The execution task is switched by the execution task switching means 2 so that the task with a lower value is executed later. The execution task switching means 2 switches execution tasks so that tasks with the same priority are executed cyclically in a predetermined order.

The condition determining means 3 determines whether a predetermined condition is satisfied. If a predetermined condition is satisfied, the priority order of the predetermined task stored in the task priority order storage means 1 is changed by the priority order change means 4. In this way, the execution order is changed so that tasks that should be processed preferentially under predetermined conditions are executed first, and real-time responsiveness can be improved. Also,
The condition learning means 5 optimizes the conditions for changing the priority order according to the usage environment by learning the change status of the priority order.

0010

Embodiment FIG. 2 is a block diagram showing the hardware configuration of an embodiment of the present invention. In the figure, 10 is a bus line, which includes an address bus, a data bus, and a control bus. Reference numeral 11 denotes a CPU, which has a so-called Neumann type program execution function. Reference numeral 12 denotes an external I/O device, which has a function of inputting data from the outside and outputting data to the outside. A keyboard 13 and a CRT display 14 are used as a man-machine interface between an operator and a computer. 15 is a ROM, which stores an operating system (OS) and necessary programs. 16
is a RAM in which various variables, flags, tables, etc. are stored. In this embodiment, it is assumed that the CPU 11 includes a built-in timer 20. The built-in timer 20 is used to generate an interrupt at fixed or programmed intervals to switch tasks.

FIG. 3 is a block diagram showing more specifically the configuration of the real-time OS of this embodiment. In the figure, #1 is a dispatcher, which is a feature of this system. This is a block that determines the task processing order of #6 from the information from each management block #2 to #5 described below and the basic rule of #7. #2 is an interrupt management block that manages the type of interrupt currently occurring and its processing status, and generates a reorganization signal requesting the dispatcher to estimate the correlation and dynamically change the priority every time an interrupt occurs. #3 is a pool usage status management block that manages the usage status of each memory pool. The usage status of memory pools is managed based on the free space in each memory pool and the task that owns free blocks, and when the free block count falls below a set value, a reorganization signal is generated. #4 is a queue management block that manages the waiting status of each queue, and generates a reorganization signal when the number of waiting blocks of each queue exceeds a set value. #5 is a task status management block that manages the status of tasks that are currently waiting for processing at each level, and when the number exceeds a set number, it generates a reorganization signal. In this way, the OS of the present invention is equipped with a plurality of management blocks #2 to #5, and when determining the processing order of tasks, not only the conventional priority of each task but also the usage status of the memory pool, The optimal task is selected by dynamically changing the priority according to various parameters such as the waiting status of each queue, the status of tasks in and outside real memory, running tasks, stopped tasks, and interrupt status. This determines the processing order. In other words, if various parameters such as the amount of free space in the memory pool, the amount of waiting blocks in the queue, or the number of waiting tasks exceed the set ranges, it means that the priority of the task is not appropriate. Optimize by increasing or decreasing the priority of Then, by storing the states of all parameters before changing the priority and the states of all parameters after changing the priority in association with each other, and estimating the correlation before and after the change, the state of the same parameter will be changed next time. When this occurs, increase or decrease the priority of the task in the same way as before to ensure that various parameters such as the amount of free space in the memory pool, the amount of waiting blocks in the queue, the number of waiting tasks, etc. deviate from the set range. can be prevented.

Next, the basic rule #7 will be explained in detail. This is a rule for dynamic task priority changes created by the user when creating the target system. Here, it is assumed that the notation of this rule conforms to the C language, but the only functions that can be used are those defined below. The functions used in the basic rule #7 will be explained below.

(1) Pool This is a function for obtaining information regarding free space in a specified pool, and its format is as follows. re
t=Pool(Poolname, PoolLevel
); Here, Poolname is the pool name, PoolL
evel is the free pool amount, and the return value is ret = -1 if the specified pool does not exist, and ret if it exceeds the set free pool amount.
= 0, and if it is less than the set free pool amount, ret = 1. Furthermore, if Poolname is omitted, information regarding free space is obtained for all pool names, and the logical sum of the results is calculated.

(2) Que This is a function for obtaining information about the amount of waiting blocks in a specified queue, and its format is as follows. ret=Que(Quename, QueLevel)
; Here, Quename is the queue name, QueLevel
is the waiting block amount, and the return value is ret=-1 if the specified queue does not exist, and ret=0 if it exceeds the set waiting block amount.
If the amount of waiting blocks is less than the set amount of waiting blocks, ret=1. Furthermore, if Quename is omitted, information on the amount of waiting blocks is obtained for all queue names, and the logical sum of the results is calculated.

(3) IntI This is a function for obtaining information regarding a specified interrupt, and its format is as follows. ret=IntI(IntIname); where, In
tIname is the interrupt type name, and the return value is ret = 0 if there is no set interrupt, and r if there is a set interrupt and it is unprocessed.
et=1. Also, if IntIname is omitted, information will be obtained for all interrupt names,
Take the logical OR of the results.

(4) Task This is a function for obtaining information regarding a task at a specified level, and its format is as follows. ret=Task(TaskPri, TaskLevel
l); Here, TaskPri is the priority level of the task, TaskLevel is the number of waiting tasks, and the return value is ret=0 if the number of waiting tasks of the specified priority does not exceed the set amount. Then,
If the number of waiting tasks of the designated priority exceeds the set amount, ret=1. Also,
If TaskPri is omitted, determination is made for all levels, and the results are logically summed.

#7 expressed using the above functions
The #1 dispatcher registers the basic rules of #1 in the internal table immediately after system startup, and references them during execution.
6 dynamic priorities are created in an optimal state.

Next, a description will be given of how the #1 dispatcher performs dynamic prioritization. First, when a reorganization signal is received from one of the management blocks, this signal is used as a signal to change the contents of the current state (Now-State) of the parameter state table to the previous state (Old-State).
ate), and the contents of the next state (Next-State) are moved to the current state. Then, the states of all parameters are saved in the next state. Then, according to the basic rules, the next state, current state, and previous state patterns are given evaluation values for dynamic prioritization that will be performed the next time the current state and previous state of this pattern appear, and are registered in the table. . lastly,
If the current state and previous state patterns exist in the table and the evaluation value is greater than or equal to the set value, the priority is dynamically changed and the next reorganization signal is waited for.

[0019]

Effects of the Invention The present invention provides a real-time processing device for multiple tasks in which a task with a higher priority is executed before a task with a lower priority, and tasks with the same priority are executed cyclically in a predetermined order. , a condition determining means for determining a predetermined condition, and a priority changing means for changing the priority stored in the task priority storage means when the predetermined condition is determined by the condition determining means. Therefore, if a predetermined condition is met, the priority of a predetermined task can be changed, which has the effect of increasing real-time responsiveness and enabling optimal control with a high degree of freedom. be. Furthermore, in the present invention, since a condition learning means is provided to learn the change status of the priority order, it is possible to optimize the conditions for changing the priority order according to the usage environment. be.

[Brief explanation of drawings]

FIG. 1 is a claim-compatible block diagram showing the basic configuration of the present invention.

FIG. 2 is a block diagram showing the hardware configuration of an embodiment of the present invention.

FIG. 3 is a block diagram showing the system configuration of the present invention.

FIG. 4 is an explanatory diagram of the operation of a conventional example.

[Explanation of symbols]

1. Task priority storage means 2 Execution task switching means 3 Condition determination means 4. Priority change means 5 Conditional learning means

Claims (1)

[Claims] 1. Task priority storage means for storing priorities assigned to each of a plurality of tasks, and tasks with higher priority than tasks with lower priority by referring to the stored contents of the task priority storage means. Also run first,
In a CPU control device comprising an execution task switching means for cyclically executing tasks of the same priority in a predetermined order, a condition determining means for determining a predetermined condition; and when the predetermined condition is determined by the condition determining means. 1. A real-time processing device for a plurality of tasks, comprising: priority order changing means for changing the priority order stored in the task priority order storage means; and condition learning means for learning conditions for changing the priority order.