JPH10269110A - Method for avoiding hang-up of computer system, and computer system using the same method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent the execution efficiency of a system from deteriorating, by providing a mechanism that detects a process operating continuously for a time longer than a certain specified time according to operation information of a CPU and avoiding a hang-up of the computer system. SOLUTION: The computer 1 is provided with a virtual device driver 3, which does not have hardware for actual control, and CPU operation information 2 is gathered periodically. Then, a monitor process 4 is connected to the virtual driver 3 and receives information from the virtual device driver and operates on the same privilege level with an application program to detect a hang-up process 5, send a signal to the hang-up process 5 in such a case, and control its operation, thereby evading the hang-up of the computer 1. Consequently, the execution efficiency of the system can be prevented from deteriorating.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for avoiding a hang-up of a computer system for monitoring and controlling process scheduling of a computer and its operation, and a computer system using this method.

[0002]

2. Description of the Related Art A conventional computer system detects a runaway of a central computer (CPU) by a function called a watchdog timer (program stop monitoring time limiter). In many cases, the system is restarted by performing a dynamic restart.

In order to monitor and control the operation of a process, it is necessary to modify an existing operating system (OS) process, and it is impossible to implement the process without a source code. is there.

When a computer adopts a multiprocessor configuration having a plurality of CPUs, and when a process runs away on one CPU, the other CPU performs the processing of the entire system. Although there is no problem in terms of continuity of operation, performance (execution efficiency)
In this respect, deterioration has occurred, and there is no method for detecting this state.

[0005]

In the scheduling of a non-TSS policy in a conventional computer system (setting of an operation schedule without time-sharing use), when one process is continuously operated by using a CPU, it is higher. Until a process having a priority attempts to operate, a process having a low priority or a TSS process which also exists in the computer system cannot operate.

In addition, when a process that occupies the CPU does not release the CPU due to a bug in the program or the like, the process becomes inoperable in most cases, and there is no other way to stop the system. . Furthermore, the system stop processing also requires that a plurality of TSS processes operate.
Under the above-mentioned situation, even the stop of the system is not confused. Such a situation often occurs when the programmer is immature.

[0007] Monitoring using a watchdog timer that has been used in the past is often implemented so as to detect, as an error, a case where a CPU goes out of control in hardware. Therefore, for example, in a computer system that performs multi-priority scheduling such as UNIX (registered trademark of X / OPEN Company Limited), runaway of a high-priority process is not detected as hardware runaway of the CPU. Not possible (O as CPU)
S is not abnormal).

In the watchdog function, the system is reset after determining that an error has occurred.
For systems that perform advanced file management such as speeding up (input / output), forced resetting is highly likely to cause file system destruction and should not be performed.

On the other hand, when the computer system has a multiprocessor configuration, by monitoring, detecting, and eliminating the runaway of the process for each CPU, even if runaway occurs on one CPU and the system hangs up (stops). It is necessary to avoid this and prevent the performance of the system from deteriorating.

An object of the present invention is to provide a method for avoiding a hang-up of a computer system which enables the following items.

(1) A sound operation of the system is obtained by detecting a situation where a high-priority process is occupying the CPU and operating, and forcibly stopping or temporarily stopping the process.

(2) Even when the computer system has a multiprocessor configuration, it is possible to detect a situation in which a high-priority process occupies the CPU and operates.

(3) The above mechanism can be realized without modifying the OS.

[0014]

According to a first aspect of the present invention, there is provided a computer system having a non-TSS scheduling policy.
A feature is provided that detects a process that operates continuously for a certain period of time or more based on PU operation information, thereby avoiding a hang-up of the computer system.

According to a second aspect of the present invention, in a computer system having a non-TSS scheduling policy, when a multi-processor configuration is adopted, a process that operates continuously for a predetermined time or more for each processor is provided. It is equipped with a detection mechanism to avoid computer system hang-up.
The method for preventing a hang-up of a computer system according to claim 3 is characterized in that, in a computer system having a non-TSS scheduling policy, when a process that operates continuously for a predetermined time or more is detected, the process is forcibly terminated or temporarily stopped. Avoid inoperable state,
The feature is to avoid the hang-up of the computer system.

According to a fourth aspect of the present invention, in the computer system having a non-TSS scheduling policy, the inoperable state avoiding mechanism is realized not by modifying the OS but by an external mechanism. The configuration is characterized in that a hang-up of the computer system is avoided.

According to a fifth aspect of the present invention, there is provided a computer system having a non-TSS scheduling policy, wherein a computer system having a non-TSS scheduling policy does not use a service of an OS based on availability of CPU operation information.
It is characterized in that the online / offline can be detected to avoid a hang-up of the computer system.

The computer system according to claim 6 is a non-T computer system.
In a computer system having an SS scheduling policy, a virtual device driver that periodically collects operation information of a CPU, and a CP from the device driver
It is characterized by comprising runaway process detecting means for receiving U operation information and detecting a runaway process, and monitoring means for forcibly terminating or temporarily stopping the runaway process by a signal from the runaway process detecting means.

That is, a virtual device driver that operates at the same privilege level as the OS and holds information on processes operating on each CPU by referring to CPU operation information managed by the OS at regular intervals; At the privilege level, run at the highest priority in the system,
And a monitoring process for receiving information on the operation process of each CPU from the virtual device driver.

The monitoring process operates at regular time intervals. That is, process information is received and each C
The number of process switching (switching) for each PU is compared with the state at the time of the previous operation. If the number of times of switching is at least two times greater than the number of times of switching to oneself, it means that two other processes have been dispatched and that the device is operating normally.

Even when the number of times of switching is only one, there is no problem because there is no other process running. The state in which one process is dispatched in addition to itself is the state in which one process is operating. If this state continues for a certain period of time or longer, it is determined that the running process is running away.

When the OS detects a runaway process based on the determination result, the OS appropriately transmits a signal to the process according to the reason for the runaway, and forcibly terminates the process depending on the type of the signal to be transmitted. , Can be paused.

With the above means, an unintended CPU occupation state due to a bug or the like caused by a specific process can be prevented, and the operation of a general process becomes possible. Further, the above configuration is
It can be realized without any modification to S.

[0024]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, an embodiment of a method for avoiding a hang-up of a computer system according to the present invention will be described. FIG.
, The CPU operation information 2 is the CP managed by the OS.
U operation information, and exists for each CPU in the case of a multiprocessor configuration.

The virtual device driver 3 is a device driver that operates at the same privilege level as the OS. The virtual device driver 3 is a device driver in the computer 1 that does not have hardware for periodically collecting the CPU operation information 2 and actually controlling it. .

The monitoring process 4 includes the virtual device driver 3
, Receives information from the virtual device driver 3, operates at the same privilege level as the application process, detects the runaway process 5, and when a runaway process is detected, sends a signal to the process. This is the process of transmitting and controlling the operation.

When a process corresponding to the monitoring process is executed at the same privilege level as the OS, the load increases and the performance of the system is affected. Application operating space is a normal OS
Then, since the privilege level is different from the space in which the OS operates, the information managed by the OS cannot be directly referred to. For this purpose, a virtual device driver 3 is provided. Further, even when the operation information can be referred to, the operation of performing the reference makes the operating process itself, and thus it is not possible to obtain useful information. Therefore, it is necessary to refer to the information with a driver that does not have the process context.

That is, the virtual device driver 3 executes the OS
Externally refers to the CPU operation information 2 managed by the server, and copies necessary data to its own local variable. And at the time of reference, so that the variable during reference is not changed,
Locks variables and performs exclusion.

The CPU operation information 2 managed by the OS includes information such as the number of times the process has been switched and the current operating process on the CPU. Can be specified.

Further, by referring to the CPU information, the online / offline status of the CPU can be monitored, which is effective from the viewpoint of preventive maintenance of the system.

As described above, according to the present invention, (1) unexpected CPU occupation of a high-priority process can be detected.

(2) Even in a multiprocessor configuration, the occupation status of one process can be detected.

(3) When a system hang-up due to CPU occupation is detected, the situation can be recovered by forcibly terminating or temporarily stopping the corresponding process.

(4) The online / offline status of the CPU can be monitored at the application level.

[0035]

According to the present invention, the reliability of the method for avoiding a hang-up of a computer system can be improved.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram of a method for avoiding a hang-up of a computer system according to an embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Computer 2 CPU operation information 3 Virtual device driver 4 Monitoring process 5 Runaway process

Claims (6)

[Claims] 1. A computer system having a non-TSS scheduling policy, comprising: a mechanism for detecting a process that operates continuously for a predetermined time or more based on operation information of a CPU to avoid a hang-up of the computer system. How to avoid system hang. 2. In a computer system having a non-TSS scheduling policy, when a multiprocessor configuration is adopted, a mechanism is provided for each processor to detect a process that operates continuously for a predetermined time or more, so that a computer system hangs up. How to avoid computer system hangup. 3. In a computer system having a non-TSS scheduling policy, when a process that operates continuously for a certain period of time or more is detected, the process is forcibly terminated or temporarily stopped to avoid an inoperable state. How to avoid computer system hangup. 4. A computer system having a non-TSS scheduling policy, wherein the inoperable state avoiding mechanism is realized by an external mechanism, not by modifying an OS.
How to avoid hang-up of computer system described in. 5. In a computer system having a non-TSS scheduling policy, online / offline of a CPU is detected based on availability of operation information of a CPU without using an OS service, and a hang-up of the computer system is performed. How to avoid computer system hangup. 6. In a computer system having a non-TSS scheduling policy, a virtual device driver that periodically collects CPU operation information, and a runaway process that receives the CPU operation information from the device driver and detects a runaway process A computer system comprising: detecting means; and monitoring means for forcibly terminating or temporarily stopping the runaway process based on a signal from the runaway process detecting means.