JPH09319596A - Device and method for scheduling task - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a device and a method for scheduling task with which a lot of processings can be performed and the use efficiency of a CPU is improved. SOLUTION: The time and state of limitation for each task are set to a limit time managing table 500. A time priority scheduling mechanism 300 is activated each time an event is generated or a task is finished and while referring to the limit time managing table 500, among tasks not in standstill state, the task having minimum limit time is dispatched as the next execution task. Thus, a lot of processings are performed without generating any CPU idle time.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus and method for scheduling a multitask program.

[0002]

2. Description of the Related Art As conventional task scheduling methods, there are the following two methods.

The first method is called fixed priority scheduling, and is a method in which the scheduler determines the execution order of a plurality of tasks in an event driven manner in accordance with the fixed priority specified by the user.

The second method is timed priority (deadline).
This is called scheduling, and a time limit is set in the time management table for each task, and the time limit is decremented by a certain fixed interval. At that time,
If a new event has occurred, newly set the time limit of the task corresponding to the process, check the time management table, and check the time management table to find out the remaining time of all tasks. This is a method in which the scheduler determines the execution order of a plurality of tasks at a time slice (constant time interval) with the smallest number as the next execution task.

[0005]

However, the above-described conventional scheduling method has the following problems.

(First Method) a. Problem 1: Since the priority order is fixed, it lacks adaptability to changes in the frequency, interval, and type of processing requests from the outside world to the system.

B. Problem 2: Even in the case of an a priori processing request from the outside world, the CPU usage efficiency cannot be increased to 100%.

[0008] There is a problem.

For example, task (1) [time limit 20 m
s, processing time 10 ms] and task (2) [time limit 30
ms, processing time 15 ms], consider scheduling of two periodic tasks (time limit = task activation period). Task (1) has a shorter time limit, so if you schedule tasks with higher priority, task (1)
[10 ms] → task (2) [10 ms] → task (1) [10 ms] is executed, and although the processing of task (2) remains 5 ms, the time limit of task (2) is 30 ms. You can see that it has passed. Therefore, scheduling cannot be done in time and the system usage efficiency cannot be raised to 100%. Therefore, in order to realize a hard real-time system that attaches importance to time limits, the second method is used for scheduling.

(Second Method) The scheduler executes a dynamic priority change in which the remaining time limit of a plurality of tasks is managed by a time management table, and the task with the smallest remaining time limit is set as the next execution task. By doing so, the task can be scheduled more efficiently than the fixed priority. However, it is more efficient than fixed priority scheduling, but has the following problem 3.

C. Problem 3: Even if there is a processing request because the scheduling timing is a time slice, there is a possibility that a CPU idle time up to the time slice interval will occur.

For example, the same example as described above, task (1) [time limit 20 ms, processing time 10 ms] and task (2)
Consider scheduling of two periodic tasks of [time limit 30 ms, processing time 15 ms] (time limit = task activation period). Assuming that the time slice interval at which the scheduler is activated is 10 ms, task (1) [10
ms] → task (2) [10 ms] → task (2) [5
ms] → idle [5 ms] → task (1) [10 m
s] → task (1) [10 ms] → task (2) [10
[ms]], it can be seen that the control time of 30 ms in the second cycle of task (2) elapses even though the processing of task (2) remains for 5 ms. Since the time slice interval is 10 ms, CP
The idle time not using U is 5 ms, and the scheduling cannot be done in time. Even with this method,
The usage efficiency of the system cannot be raised to 100%.

(Common to the first and second methods) d. Problem 4: It is not possible to predict at the time when the processing request occurs whether or not the processing request can be performed within the time limit.

The present invention has been made under such circumstances, and an object of the present invention is to provide a scheduling device and method capable of performing a large number of processes and improving the CPU usage efficiency. To do.

[0015]

[Means for Solving the Problems] The Problems 1, 2, 3
Can be solved by setting the scheduling timing to be event-driven instead of time slice in the time-priority scheduling in the second method of the related art.

Problem 4 is the system based on the mechanism of making the timing of scheduling in the time-priority scheduling in the second method of the prior art not the time slice but the event, and the history of the processing time of each task. This is accomplished by providing a mechanism to monitor the load.

More specifically, the task scheduling device is configured as in the following (1) to (4) and the task scheduling method is configured as in the following (5).

(1) A time management table in which a time limit and a status are set for each task, and the time management table that is activated each time an event occurs and the task ends is referred to, and the time limit table of the tasks that are not in the sleep state A task scheduling device provided with a timed priority scheduling mechanism for dispatching the least number of tasks as the next execution task.

(2) The task scheduling device according to (1), further comprising a task load monitoring mechanism for displaying a message indicating that there is a risk that the task processing cannot be executed within the time limit.

(3) The task load monitoring mechanism uses the following (a)
The task scheduling device according to (2) above, which determines that there is a risk when the conditional expression of is satisfied.

(A) Σ [T (n) -t (n)] / L (n)> 1 Here, Σ is a task state of n = 1, 2, 3 ... Represents the total of non-states, L
(N) is the time limit of the task (n), t (n) is the processing time of the task (n) that is currently being executed, T (n) is the history of the past processing time of the task (n), and the past t It is a value obtained by calculating the average, maximum, etc. based on (n).

(4) The task load monitoring mechanism has the following (b)
The task scheduling device according to (2) above, which determines that there is a risk when the conditional expression of is satisfied.

(B) ΣT (n) / I (n)> 1 Here, Σ represents the total of n = 1, 2, 3, ..., Number of tasks, and I (n) is the number of tasks (n). Initial time limit, T
(N) is a history of past processing times of task (n), and is a value obtained by calculating an average, maximum, etc. based on past t (n).

(5) A task scheduling method in which the timing of the timed priority scheduling is event driven.

[0025]

BEST MODE FOR CARRYING OUT THE INVENTION The present invention will be described in detail below with reference to an embodiment of a task scheduling device.

(Embodiment 1) FIG. 1 is a diagram showing a configuration of a "task scheduling apparatus" which is Embodiment 1. The processing described below is performed by a CPU (not shown). In the system including the present embodiment, an interrupt process 20 is activated by an interrupt occurrence in FIG. 1, and upon receipt of an event occurrence notification, the task registration mechanism 100 receives an initial time limit value (I (n )) 600 is the time limit management table 50
0 is registered as a time limit (L (n)) 520, and the time priority scheduling mechanism 300 is activated. In the time limit priority scheduling mechanism 300, the time limit management table 5
Of the tasks registered in 00, the maximum time limit (L
(N)) 520 retrieves the remaining few tasks and dispatches them as the next execution tasks. If different events occur successively, a plurality of tasks are registered in the time limit management table 500, and the task with the smallest time limit (L (n)) 520 among them is the execution task. Asynchronously with this, the time limit (L (n)) 520 of each task registered in the time limit management table 500 is decremented by the timer mechanism activated by the timer interrupt processing 30. Each task issues a processing end notification to the task deletion mechanism 400 when the processing is completed. The task deletion mechanism 400 deletes the corresponding task from the scheduling target tasks managed by the time limit management table 500 (SUSPEND), and the time priority scheduling mechanism 3
00 is started. In this way, task scheduling is performed immediately after the occurrence of an event and the end of processing (that is, event-driven).

As shown in FIG. 6, the time limit management table 500 has a task (n) 510 corresponding to a task identifier.
And L representing the remaining time limit of the task (n)
(N) 520 and STATUS indicating the status of task (n)
It is stored including S (n) 530 information. STAT
A possible value of US (n) 530 is one of three states: EXECUTE (in execution), READY (executable), and SUSPEND (pause). STATUS
The initial state of (n) 530 is SUSPEND. S
In the TATUS (n) 530, when an event occurs, the task registration mechanism 100 assigns R to the task (n) related to the event.
When EADY is set and EXECUTION is set by the timed priority scheduling mechanism 300, EXECUTE is set, and when the task processing ends, SUS is set by the task deletion mechanism 400.
PEND is set. The L (n) 520 has a time limit initial value I by the task registration mechanism 100 when an event occurs.
(N) 620 is set (see FIG. 7), STATUS
(N) Decremented by timer mechanism 200 while 530 is READY or EXECUTE. The time-limited priority scheduling mechanism 300 is activated by the task registration mechanism 100 when an event occurs and by the task deletion mechanism 400 when task processing is completed so that a task with a small time limit L (n) 520 is always executed.

As shown in FIG. 7, a time limit initial value 600
Is stored including the task (n) 610 corresponding to the task identifier and I (n) 620 information indicating the initial value of the time limit of the task (n). Initial time limit value I (n) 6
If the task (n) is a periodical activation task, the periodical time interval is preset in 20. Further, if the task (n) is not a periodic task, the initial value of the time limit allowed for the user to arbitrarily execute the processing of the task is set. The time limit initial value I (n) 620 is
When an event occurs, it is referred to by the task registration mechanism 100 and set in the time limit L (n) 520 of the time limit management table 500.

Details of the task registration mechanism 100 will be described with reference to FIG. First, the internal table is searched for a task corresponding to the generated event (step 110). The initial value I (n) 620 of the task is obtained from the initial value 600 of the time limit based on the task (n) (step 12).
0). I (n) 620 is set in L (n) 520 of the time limit management table 500, and at the same time READY is set in STATUS (n) of the time limit management table 500 (step 130). Finally, the time-limited priority scheduling mechanism 300 is activated (step 140) to execute scheduling.

Details of the timer mechanism 200 will be described with reference to FIG. First, the index for searching the time limit management table 500 is initialized to n = 1 (step 21).
0). The subsequent processing is repeated until the task (n) == END (branch 220). STATUS = SUSPEN
If it is not D (branch 221), the time limit management table 50
The time limit L (n) 520 of 0 is decremented (step 222). The index n is incremented (step 230).

Details of the timed priority scheduling mechanism 300 will be described with reference to FIG. First, the index for searching the time limit management table 500 is initialized to n = 0 (step 310). The temporary variable Min for obtaining the minimum value of the time limit is initialized to the maximum value that the variable can take (step 320). A temporary variable NextTask for storing the identifier of the next execution task is initialized to IDLE (idle task identifier) (step 330). The index n is incremented (step 340). The subsequent processing steps 342 to 346 are repeated until the task (n) == END (branch 341).
If STATUS ≠ SUSPEND (branch 342), STATUS (n) 530 of the time limit management table 500 is set to READY (step 343). further,
If Min> L (n) (branch 344), Min = L
(N) (step 345) and NextTask = n (step 346) are set to search for a task having a shorter time limit L (n). When task (n) == END (branch 341), the loop is ended, and EXECUTE is set in STATUS (n) 530 of the time limit management table 500.
And set the mark of running (step 35
0). Finally, the task (NextTask) is dispatched as the next execution task (step 360).

Details of the task deletion mechanism 400 will be described with reference to FIG. First, the index for searching the time limit management table 500 is initialized to n = 1 (step 4).
10). The subsequent processing is repeated until task (n) == END (branch 420). If the task (n) is not the end task (branch 421), the index n is incremented to retrieve the next task (step 422). If the task (n) == end task (branch 421), the status of the task (n) is STATUS (n) = SUSPEND
Then, the task is removed from the scheduling target task and the loop is exited (step 430). Finally, the time priority scheduling mechanism 300 is activated to execute scheduling (step 440).

The pattern in which tasks are actually scheduled by providing each of the above mechanisms will be described using the same example as described in "Problems to be Solved by the Invention". Based on the above example, consider scheduling of two periodic tasks, task (1) [limit different time 20 ms, processing time 10 ms] and task (2) [limit time 30 ms, processing time 15 ms] (limit time = task start cycle) To). A timing chart in this case is shown in FIG. The task execution order is as follows: task (1) [10 ms] → task (2) [10 ms] → task (2) [5 ms] → task (1) [5 ms] → task (1) [5 ms] → task (2) [5 ms] → task (1) [10 ms] → task (2) [10 ms] and 60
Execution is repeated in the ms cycle, and there is no wasteful idle time when the CPU is not used.

As described above, according to this embodiment,
It is possible to eliminate a wasteful idle time of the CPU, execute a large amount of processing, and improve the usage efficiency of the CPU.

(Embodiment 2) In Embodiment 2, the problem 4
An example of a method for realizing the means for solving the above will be described. A configuration diagram at this time is shown in FIG. FIG. 9 shows FIG.
Task load monitoring mechanism 700 and processing time management table 8
00 is added. In FIG. 9, the system starts the interrupt process 20 when an interrupt occurs, receives an event occurrence notification, and starts the task load monitoring mechanism 700. The task load monitoring mechanism 700 predicts the risk of system overflow by the following conditional expression as a condition that the processing of the task being executed which is currently registered in the time limit management table 500 cannot be executed within the time limit.

Σ (T (n) -t (n)) / L (n)> 1 (Σ is the number of tasks n = 1, 2, 3 ... However, the fact that the task is not in the sleep state means that STATUS (n) ≠ SUSPE in the time limit management table 500.
Judged by ND, t (n) is the currently executing task (n)
, T (n), is a value obtained by calculating a weighted average based on past t (n) in the past processing time history of task (n), and is managed by the processing time management table 800.
After the conditional expression is calculated, t (n) is cleared.

Further, the event occurrence also activates the task registration mechanism 100 as in the first embodiment. In the task registration mechanism 100, the initial time limit value (I (n)) 600 of the task (n) corresponding to the event is registered as the time limit (L (n)) 520 in the time limit management table 500, and the time priority scheduling mechanism 300 is set. to start. In the time-limited priority scheduling mechanism 300, of the tasks registered in the time limit management table 500, the most time limit (L (n))
520 retrieves the remaining few tasks and dispatches them as the next execution tasks. If different events occur successively, a plurality of tasks are registered in the time limit management table 500, and the task with the smallest time limit (L (n)) 520 among them is the next execution task. Asynchronously to this,
Timer mechanism 200 activated by timer interrupt processing 30
Causes the time limit (L (n)) 520 of each task registered in the time limit management table 500 to be decremented, and STATUS (n) = EXECUTE
Processing time management table 8 only for tasks (during task execution)
T (n) 820 of 00 is incremented. When the processing of each task is completed, the task load monitoring mechanism 700
And issue a processing end notification to the task deletion mechanism 400. In the task load monitoring mechanism 700, the processing time management table 8
Weighted average T (n) 8 based on t (n) 820 of 00
Calculate 30. Furthermore, the task deletion mechanism 400 deletes the corresponding task from the scheduling target tasks managed by the time limit management table 500, and activates the timed priority scheduling mechanism 300. In this way,
It is possible to predict the system load when an event occurs.

Referring to FIG. 10, the processing time management table 800 has a task (n) 810 corresponding to a task identifier.
And t (n) 820 that represents the cumulative time during which the task (n) 810 has executed processing up to the present as processing time,
In the past processing time history of task (n) 810, the past t
(N) Value T calculated by weighted average based on 820
(N) Stored including 830 information. T (n) 8
The initial state of 30 is 0. The t (n) 820 is cleared by the task load monitoring mechanism 700 when an event occurs and is incremented by the timer mechanism 200 only while the time priority scheduling mechanism 300 is in the execution task (STATUS (n) = EXECUTE). At the end of task processing, the task monitoring mechanism 700 calculates a weighted average T (n) 830 based on t (n) 820.

Based on FIG. 11, the task load monitoring mechanism 70
0 will be described in detail. The task load monitoring mechanism 700 is activated when an event occurs and when the task ends. Therefore, first of all, it is determined which factor has caused the activation (branch 710). If activated due to the occurrence of an event, the risk of system overflow is predicted and displayed by the following conditional expression as a condition that the processing of the mid-execution task currently registered in the time limit management table 500 cannot be executed within the time limit. (Step 720).

Σ (T (n) -t (n)) / L (n)> 1 (Σ is the sum of n = 1, 2, 3, ... At this time, the fact that the task is not in the sleep state means that STATUS (n) ≠ SUS in the time limit management table 500.
Judged by PEND, t (n) is the processing time of the task (n) currently being executed, T (n) is a history of past processing times of the task (n), and weighted based on the past t (n). The calculated average is the calculated value. After the conditional expression is calculated, t (n) 820 is cleared (step 730).

At the branch 710, if the task end notification is activated, t of the processing time management table 800 is reached.
A weighted average T (n) 830 is calculated based on (n) 820 by the following formula (step 740).

Told = T (n) (step 741) α = 0.6 (step 742) T (n) = α ^* Told + (1-α) ^* t (n) (step 743) where α is a weight, Told is a variable for temporarily storing the immediately preceding T (n), and t (n) is the current processing time of task (n). Finally, T (n) is set to the processing time management table 80.
It is set to 0 (step 750).

The details of the timer mechanism 200 will be described with reference to FIG. The processing procedure according to the present embodiment is the processing procedure of the timer mechanism 200 of FIG. 3 according to the first embodiment with the operation of the processing time management table 800 added. First, the index for searching the time limit management table 500 is initialized to n = 1 (step 210). The subsequent processing is repeated until the task (n) == END (branch 220). STATUS of the time limit management table 500
If (n) == SUSPEND is not satisfied (branch 221), L (n) 520 of the time limit management table 500 is decremented (step 222). Time limit management table 5
If STATUS (n) of 00 = EXECUTE (branch 223), t (n) 82 of the processing time management table 800
0 is incremented (step 224). The index n is incremented (step 230).

As described above, according to this embodiment,
In addition to the effects of the first embodiment, the risk of system overflow is predicted and displayed, so that appropriate measures can be taken. In this example, the danger is displayed,
It is also possible to record the frequency of risk occurrence and use it as information such as hardware improvement.

(Third Embodiment) In the third embodiment, the problem 4 is described.
An example of a method for realizing the means for solving the above will be described. A configuration diagram at this time is shown in FIG. FIG.
A task load monitoring mechanism 710 and a processing time management table 800 are added to the configuration of FIG. In FIG. 13, the system starts an interrupt process 20 when an interrupt occurs, receives an event occurrence notification, and receives a task load monitoring mechanism 710.
Is started. The task load monitoring mechanism 710 predicts the risk of system overflow by the following conditional expression as a condition that cannot be executed within the time limit including tasks that may be activated in the future due to event occurrence.

ΣT (n) / I (n)> 1 (Σ represents the total of n = 1, 2, 3, ..., Number of tasks) where I (n) is the initial value of the time limit and t (N) is the processing time of the task (n) currently being executed, T (n) is a history of past processing times of the task (n), and a weighted average is calculated based on the past t (n). The value is used as a value and managed by the processing time management table 800. After the conditional expression is calculated, t (n) is cleared.

Further, the event occurrence also activates the task registration mechanism 100 as in the first embodiment. In the task registration mechanism 100, the initial time limit value (I (n)) 600 of the task (n) corresponding to the event is registered as the time limit (L (n)) 520 in the time limit management table 500, and the time priority scheduling mechanism 300 is set. to start. In the time-limited priority scheduling mechanism 300, of the tasks registered in the time limit management table 500, the most time limit (L (n))
520 retrieves the remaining few tasks and dispatches them as the next execution tasks. If different events occur successively, a plurality of tasks are registered in the time limit management table 500, and the task with the smallest time limit (L (n)) 520 among them is the execution task. Asynchronously with this, the timer mechanism 200 activated by the timer interrupt processing 30 decrements the time limit (L (n)) 520 of each task registered in the time limit management table 500, and further, STATUS (n ) = EXECUTE (task is being executed), t (n) 820 of the processing time management table 800 is incremented. Each task is
When the processing is completed, a processing end notification is issued to the task load monitoring mechanism 710 and the task deletion mechanism 400. The task load monitoring mechanism 710 calculates a weighted average T (n) 830 based on t (n) 820 of the processing time management table 800. Furthermore, the task deletion mechanism 400 deletes the corresponding task from the scheduling target tasks managed by the time limit management table 500, and activates the timed priority scheduling mechanism 300. In this way, it becomes possible to predict the system load when an event occurs.

Since the third embodiment is the same as the second embodiment except for the internal operation of the task load monitoring mechanism, only different parts will be described.

Referring to FIG. 11, task load monitoring mechanism 71
0 will be described in detail. The task load monitoring mechanism 710 is activated when an event occurs and when the task ends. Therefore, first of all, it is determined which factor has caused the activation (branch 710). If it is activated by an event occurrence, the risk of system overflow is predicted and displayed by the following conditional expression as a condition that cannot be executed within the time limit including tasks that may be activated by an event occurrence in the future (step 72
0).

ΣT (n) / I (n)> 1 (Σ represents the sum of n = 1, 2, 3, ..., Number of tasks) where I (n) is the initial value of the time limit, t (n) is the processing time of the currently executing task (n), T (n) is the history of the past processing time of the task (n), and the weighted average is calculated based on the past t (n). Value. After calculating the conditional expression,
The t (n) 820 is cleared (step 730).

At the branch 710, if the task is started by the task end notification, t in the processing time management table 800.
A weighted average T (n) 830 is calculated based on (n) 820 by the following formula (step 740).

Told = T (n) (step 741) α = 0.6 (step 742) T (n) = α ^* Told + (1-α) ^* t (n) (step 743) where α is a weight, Told is a variable for temporarily storing the immediately preceding T (n), and t (n) is the current processing time of task (n). Finally, T (n) is set to the processing time management table 80.
It is set to 0 (step 750).

As described above, the same effect as in the second embodiment can be obtained in this embodiment as well.

(Fourth Embodiment) In the fourth embodiment, an example of a method for realizing means for solving the problem 4 will be described.
A configuration diagram at this time is shown in FIG. FIG.
The task load monitoring mechanism 720 and the processing time management table 800 are added to the above configuration. Example 4
In the task load monitoring mechanism 700 according to the second embodiment,
This is realized by replacing the calculation method of T (n) 830 corresponding to step 740 of FIG. 11 with a calculation formula for obtaining the following maximum value instead of the weighted average.

T (n) = max (T (n), t (n)) (max () is a function that compares the first and second arguments and returns the larger value) The same effect as that of the second embodiment can be obtained by the embodiment.

(Fifth Embodiment) In a fifth embodiment, an example of a method for realizing means for solving the problem 4 will be described.
A configuration diagram at this time is shown in FIG. FIG.
The task load monitoring mechanism 730 and the processing time management table 800 are added to the above configuration. Example 5 is
In the task load monitoring mechanism 710 according to the third embodiment,
This is realized by replacing the calculation method of T (n) 830 corresponding to step 740 of FIG. 11 with a calculation formula for obtaining the following maximum value instead of the weighted average.

T (n) = max (T (n), t (n)) (max () is a function that compares the first and second arguments and returns the larger value) The same effect as that of the third embodiment can be obtained by the embodiment.

[0058]

As described above, according to the present invention,
Scheduling can be performed immediately at the time of event occurrence and processing end in time-limited priority scheduling, the overhead due to time slice is eliminated, and easy setting can improve system usage efficiency to 100% without consuming idle time. Will be possible.
Also, by giving the priority order by the time limit for each task, there is no case where the operation based on the specification such as priority inversion cannot be guaranteed. As a result, the system can accurately execute more processing based on the specifications, and a cost reduction effect can be expected.

Further, in the inventions according to claims 2 to 4, it is possible to predict the risk that the task processing cannot be executed within the time limit in the timed priority scheduling. It has become possible to monitor if it is working, and if the system is not expected to perform a process within the time limit, then appropriate action can be taken.
In addition, if the system is expected to be constantly under high load, it is possible to improve the hardware to an appropriate level and improve the system reliability without excessive equipment and cost. Is.

Further, in order to implement the present invention, it is not necessary to newly develop dedicated hardware, and it is possible to easily incorporate it into a fixed priority real-time OS. Reliability and performance can be improved at cost.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of a first embodiment.

FIG. 2 is a timing chart showing the processing of the task registration mechanism.

FIG. 3 is a timing chart showing the processing of the timer mechanism.

FIG. 4 is a flowchart showing the processing of the timed priority scheduling mechanism.

FIG. 5 is a flowchart showing the processing of the task deletion mechanism.

FIG. 6 is a diagram showing an example of stored information in a time limit management table.

FIG. 7 is a diagram showing an example of stored information of initial value of time limit.

FIG. 8 is a timing chart of task (1) and task (2)

FIG. 9 is a diagram showing a configuration of a second embodiment.

FIG. 10 is a diagram showing an example of stored information in a processing information management table.

FIG. 11 is a flowchart showing the processing of the task load monitoring mechanism.

FIG. 12 is a flowchart showing the processing of the timer mechanism.

FIG. 13 is a diagram showing a configuration of a third embodiment.

FIG. 14 is a diagram showing a configuration of a fourth embodiment.

FIG. 15 is a diagram showing the configuration of Example 5;

[Explanation of symbols]

 300 time priority scheduling mechanism 500 time limit management table

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Atsushi Hirahara 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc.

Claims (5)

[Claims] 1. A time management table in which a time limit and a status are set for each task, and the time management table that is activated each time an event occurs and the task ends and refers to the time management table is referred to. A task scheduling device comprising: a timed priority scheduling mechanism for dispatching a small number of tasks as a next execution task. 2. The task scheduling apparatus according to claim 1, further comprising a task load monitoring mechanism for displaying a message indicating that there is a risk that task processing cannot be executed within a time limit. . 3. The task scheduling device according to claim 2, wherein the task load monitoring mechanism determines that there is a risk when the following conditional expression (a) is satisfied. (A) Σ [T (n) -t (n)] / L (n)> 1 Here, Σ is n = 1, 2, 3, ... Represents the sum, L
(N) is the time limit of the task (n), t (n) is the processing time of the task (n) that is currently being executed, T (n) is the history of the past processing time of the task (n), and the past t It is a value obtained by calculating the average, maximum, etc. based on (n). 4. The task scheduling apparatus according to claim 2, wherein the task load monitoring mechanism determines that there is a risk when the following conditional expression (b) is satisfied. (B) ΣT (n) / I (n)> 1 Here, Σ represents the total of n = 1, 2, 3, ..., The number of tasks, and I (n) is the initial time limit of the task (n). Value, T
(N) is a history of past processing times of task (n), and is a value obtained by calculating an average, maximum, etc. based on past t (n). 5. A task scheduling method, wherein the timing of the timed priority scheduling is event driven.