JPH0962520A - Infinite loop monitoring device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To monitor an infinite loop with a software and to automatically cancel that state. SOLUTION: For example, tasks A and B are started and when the property of the task A is higher, the task A starts operation. This is displayed on a ready list 5 of an OS part 2. A monitor task 3 periodically receives the instruction from a timer task 4, refers to the ready list 5 and compares the task, which bad been executed in the last reference with the task being executed this time. When the same task is contotnuiusly displayed on the ready list 5, the state of infinite loop is judged and system restart processing is automatically performed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an infinite loop monitoring apparatus which detects that a task has entered an infinite loop state due to some abnormality in an information processing apparatus and recovers it.

[0002]

2. Description of the Related Art In a system in which a plurality of tasks are activated in an information processing apparatus and each of these tasks executes a certain process, in an environment in which only one of the tasks is permitted to execute, a predetermined number is assigned to each task. Priority order is established. That is, each task is sequentially assigned a CP according to its priority order.
It occupies U and executes predetermined processing. While one task is being executed, another task is in a waiting state. Here, the task being executed may enter an infinite loop state for some reason, and so-called deadlock may prevent further processing. When the operator monitors the information processing apparatus and recognizes such an infinite loop state, the information processing apparatus is restarted by pressing the reset button and performing a hardware reset. A device that automatically detects such an infinite loop state by hardware is also known.

[0003]

The conventional monitoring of an infinite loop as described above has the following problems to be solved. If the infinite loop state is recognized by the operator, the operator can escape from the infinite loop state by performing a reset operation. However, if the operator does not constantly monitor the operating state of the information processing device,
The state may be left unattended, and processing will be delayed if other tasks are left in the waiting state forever.

On the other hand, a device that detects an infinite loop in terms of hardware increases the cost of the device itself and hinders the downsizing of the device. Further, it is not easy to incorporate this type of device into an information processing device that is already in operation.

[0005]

The present invention adopts the following configurations in order to solve the above points. (Structure) The infinite loop monitoring apparatus of the present invention is a monitoring task for monitoring whether or not a started task is in an infinite loop state in a case where a plurality of tasks are processed in the set priority order, respectively. The monitoring task includes a timer task that outputs a monitoring operation instruction at a fixed cycle in order to periodically execute the monitoring operation for the monitoring task, and a ready list that displays the currently operating task.
During the monitoring operation, the ready list is referenced to recognize the task being executed, and it is compared with the task that was being executed at the previous monitoring operation to see if it is the same. When the task is being executed, the system restart processing is executed by determining that the task is in an infinite loop state.

(Operation) In this information processing apparatus, only the task having the highest priority among the tasks waiting to be processed is allowed to execute. You can see what tasks are running by looking at the readylist. If the same task continues to be executed by periodically referring to the ready list, it is determined to be an infinite loop state. The timer task sets such a monitoring timing at an appropriate interval to operate the monitoring task. Therefore, if the cycle is too long, the timing of infinite loop detection is delayed. If it is too short, the monitoring operation of the monitoring task unnecessarily increases, so that an appropriate cycle is set. The interrupt handler executes the timekeeping process for causing the timer task to send a message on the OS unit side. This is because the interrupt handler on the OS side can time more accurately than the timer task itself, so if the timer task itself can time properly, such a timer It may have a function.

(Structure) In the above apparatus, the threshold value of a predetermined number of times for judging an infinite loop may be changed depending on the execution content of the task. (Function) Depending on the nature of the task, there are long and short execution times, so the threshold is changed to make appropriate and prompt judgment.

[0008]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to specific examples. FIG. 1 is a functional block diagram showing an example of the device of the present invention. This figure illustrates various tasks operating in the information processing apparatus and an OS (operation system) portion. The task unit 1 includes a task A and a task B to be monitored by this device.
And a monitoring task 3 provided for infinite loop monitoring,
A timer task 4 and the like are provided. Further, the OS section 2 is provided with a ready list 5 for displaying currently executing tasks and an interrupt handler 6 for measuring time.

The present invention is applied to an information processing apparatus in which a plurality of tasks are processed in the set priority order and the task being executed occupies the CPU. In the example of this figure, the task A has a higher priority than the task B and is configured to be executed first. Also, the monitoring task 3 is a part that performs processing for monitoring whether or not the task that is actually operating is in an infinite loop state after being activated. The timer task 4 is composed of a program that sends a message instructing the monitoring task 3 to execute the monitoring operation at a constant cycle. This message is stored in mailbox 3A,
When the monitoring task 3 reads the message, the monitoring operation instruction of the timer task 4 is transmitted to the monitoring task 3.

The ready list 5 of the OS section 2 is a section for acquiring and displaying the task currently being executed by the handling function of the OS. Therefore, the task A is displayed in the ready list 5 while the task A is being executed, and the task B is displayed in the ready list 5 when the processing of the task A is finished and the task B is started to be executed. Monitoring task 3 is timer task 4
Whenever a monitoring operation instruction is received from the ready list 5
Recognize the task being executed by referring to. Further, the monitoring task 3 holds the task name displayed by referring to the ready list 5 last time. And during the monitoring operation,
The previously operating task and the name of the task obtained by referring to the ready list 5 this time are compared, and if they are the same, it is recognized that the task is continuously operating. In this case, the timing of the monitoring operation instruction output from the timer task 4 is selected at an appropriate interval, and the infinite loop state is set when the same task is operating all the time by referring to the ready list 5 several times. to decide. When the monitoring task 3 makes this determination, it executes the system restart process of the information processing apparatus.

FIG. 2 shows a concrete operation explanatory diagram of the apparatus of the present invention. (A) of the figure is a time chart when the task A and the task B operate normally and end the data, and (b) is a time chart when the task A enters an infinite loop state. 2A, the monitoring task 3 shown in FIG. 1 has t1, t2, t3, t4, t shown in FIG.
5, the ready list 5 is referred to in accordance with the monitoring operation instruction. In this case, task A has time t1,
It was operating when the ready list 5 was referenced at t2. However, at time t3, the display is switched such that task B is operating. Then, at time t5, the processing of task B ends, and other task names are displayed in the ready list 5, for example. The monitoring task 3 recognizes the end of the processing of such a task and confirms that the information processing device is operating normally. On the other hand, as shown in (b), it is assumed that the monitoring task 3 refers to the times t1, t2, t3, t4, t5 and the repeat ready list 5, and in any case there is a display that the task A is being executed. In this case, for example, the sixth time is set as the threshold value, and if the task A is still operating at the sixth time t6, the monitoring task 3
Determines that task A is in an infinite loop state and performs system restart processing.

The above is the outline of the operation of the apparatus of the present invention, and the operation of the apparatus of the present invention will be described more specifically below.
FIG. 3 is a flow chart of the startup process of the device. First,
The monitoring task is activated along with the activation of the apparatus and enters a standby state (step S1). In the next step S2, the timer task 4 is started and the interrupt handler 6 of the OS section 2 is also started. Further, in steps S3 and S4, the task A and the task B are activated, and the information processing apparatus enters the operating state (step S5).

FIG. 4 is a flow chart of the infinite loop detection operation after the apparatus is activated. First, in step S1,
When the interrupt handler 6 of the OS unit 2 detects a timeout, it issues a timeout interrupt to the timer task 4 in step S2. Then, the timer task starts its operation, and sends a message for instructing the monitoring operation to the monitoring task 3. In step S3, the monitoring task 3
Receives this message and makes an infinite loop decision.

FIG. 5 shows a flowchart of such an infinite loop judgment processing subroutine. First, the monitoring task 3 refers to the task name of the ready list 5 in step S1. Next, in step S2, the RUN flag of the internal table prepared for the monitoring task 3 is checked.

FIG. 6 is an explanatory diagram of the internal table of the monitoring task. As shown in this figure, the internal table of the monitoring task 3 displays the task name, the RUN flag, the continuous RUN counter, and the threshold value. As the task name, the names of task A and task B to be monitored are written. Further, the RUN flag is "1" when operating and "0" when waiting. Here, since the task A is operating, the RUN flag of the task A is “1” and the R of the task B is R.
The UN flag is "0". The continuous RUN counter is a counter that is counted out by “1” each time the monitoring task 3 executes the monitoring operation once during the operation of the task A and recognizes the name of the task A in the ready list 5, for example.

If the operation of task A is completed, task B
Is switched to the operating state, the continuous RUN counter is cleared to zero. Therefore, the continuous RUN counter being incremented in sequence indicates that the task A is continuously operating. In the illustrated example, the continuous RUN counter is "5". Since the task B is not operating, it is “0”. The threshold value serves as a reference for indicating when the numerical value of the continuous RUN counter is reached to determine the infinite loop state. The reason for providing each task is that, due to the nature of the task, there are those that do not process for too long and those that continue for a relatively long time.

In step S2 of FIG. 5, the RUN flag of such an internal table is checked, and when it is ON, that is, in the state of "1", the continuous RUN counter is incremented by "1" in step S3. That is, if the continuous RUN counter is operating and is not "0", "1" is added to it. Then, in step S4, the value of the continuous RUN counter is compared with the value of the threshold value. In step S4, if the value exceeds the threshold value, it is determined that an infinite loop has occurred, and this determination process ends. If the continuous RUN counter is less than the threshold value, it is determined that the infinite loop is not yet performed, and this determination processing is exited.

On the other hand, in step S2, if the RUN flag in the internal table is off, that is, if the RUN flag was not being executed last time, the process proceeds to step S5, and the RUN flag corresponding to the task name being executed this time is turned on. That is, it is set to "1" and the continuous RUN counter is incremented by "1". Since the continuous RUN counter should have been originally "0", the continuous RUN counter is unconditionally "1" here. In step S6, the RUN flag of another task, for example task A, is turned off, that is, "0", and the continuous RUN counter is cleared to "0".

When the endless loop determination process as shown in FIG. 5 is completed, the process proceeds to step S4 in FIG. In step S4, the process branches depending on the result of the endless loop determination. If the loop is not an infinite loop, step S is performed again.
Returning to step 1, the processing from step S1 to step S3 is repeated. On the other hand, if it is an infinite loop, step S5
Then, the monitoring task 3 executes the system restart process. This system restart processing is processing for automatically resetting the information processing device by software.

As shown in FIGS. 3 to 5, the monitoring task 3 itself is composed of a program for executing a relatively simple process. The timer task 4 is also composed of a simple program that performs a timer interrupt at a fixed cycle. Therefore, as compared with hardware for detecting an infinite loop, it can be easily applied to any information processing device at a low price, and an infinite loop can be reliably detected and escaped therefrom. Further, when using hardware, a failure is likely to occur due to various causes. On the other hand, in the case of the present invention, as long as the information processing device normally operates due to the monitoring by software, there is no risk of failure. In addition, since the time of infinite loop judgment can be changed depending on the type and content of the task, the judgment can be optimized.

Due to the above-mentioned configuration, the present invention is configured to release the deadlock and automatically restart the system when an infinite loop state occurs. Therefore, when a deadlock state is left for a long time, such as a network relay device, it can be effectively used particularly for a system or the like which has a great influence on communication using the network.

[Brief description of drawings]

FIG. 1 is a functional block diagram of an apparatus of the present invention.

FIG. 2 is an explanatory diagram of the operation of the device of the present invention.

FIG. 3 is a flowchart of a startup process of the device.

FIG. 4 is an infinite loop detection operation flowchart.

FIG. 5 is an endless loop determination processing flowchart.

FIG. 6 is an explanatory diagram of a monitoring task internal table.

[Explanation of symbols]

 1 task part 2 OS part 3 monitoring task 4 timer task 5 ready list 6 interrupt handler

Claims (2)

[Claims] 1. A monitoring task that monitors whether or not a started task is in an infinite loop state in which a plurality of tasks are processed in the order of priority set respectively, and a periodic task for this monitoring task. It is equipped with a timer task that outputs a monitoring operation instruction at a fixed cycle in order to execute a monitoring operation, and a ready list that displays the task that is currently running. The monitoring task refers to the ready list during the monitoring operation. Then, it recognizes the task being executed, compares it with the task that was being executed at the time of the previous monitoring operation, and if the same task is still running even after repeating the monitoring operation a predetermined number of times, An infinite loop monitoring device characterized by executing a system restart process by judging an infinite loop state. 2. The endless loop monitoring device according to claim 1, wherein a threshold value of a predetermined number of times for judging an endless loop is varied depending on the execution content of the task.