JPH0477931A - Prevention device for deadlock of computer system - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Purpose of the Invention (Industrial Application Field) The present invention provides a deadlock prevention device for computer systems used in various computer systems that require the use of shared resources in multiple tasks. In particular, the present invention relates to a deadlock prevention device for a computer system that prevents deadlocks that may occur during access to shared resources by tasks.

(Prior Art) Generally, in a computer system, a certain series of processing is divided into multiple tasks, and these tasks sequentially access shared resources and obtain the necessary resources while performing tasks under a predetermined division of roles. When a method is adopted to execute a predetermined process, each task may make a mistake in the process of sequentially accessing shared resources, such as accessing shared resource B instead of accessing shared resource A. At that point, the other systems are in a waiting state for accessing the shared resource A. As a result, there is a concern that the damage to one task will spread to other tasks, causing the entire system to stop and developing into a put-lock state that cannot be easily recovered. Here, deadlock means that another task is in a waiting state and cannot operate.

Conventionally, as a means to prevent the occurrence of the above-mentioned deadlock state, it has been considered to accurately design a program for starting and ending exclusive control over shared resources for each task.

However, it is extremely difficult to design a complete program for each task because the program designer must accurately understand the behavior of the entire system. The reality is that they are trying to improve the problems little by little.

Moreover, complicated execution conditions are set for each task to access shared resources, but once a deadlock situation occurs during actual online operation, the problematic task cannot be completed and other tasks also stop. I have no choice but to do it. At this time, it takes a great deal of time and effort to investigate the cause of the put-lock.

(Problem to be solved by the invention) Therefore, when each task with complicated execution conditions set as described above accesses a shared resource while controlling it exclusively, a deadlock of the shared resource may occur due to a mistake in the procedure. Then, other tasks accessing the shared resource will also stop, causing a problem that the entire system will fall into a deadlock.

The present invention has been made in view of the above-mentioned circumstances, and by releasing the access when a deadlock occurs while accessing a shared resource, the occurrence of the deadlock phenomenon can be avoided and the damage can be reduced. The purpose of the present invention is to provide a deadlock prevention device for a computer system that minimizes deadlock.

[Configuration of the Invention (Means for Solving the Problem) In order to solve the above problem, the deadlock prevention device for a computer system according to the present invention is provided with a plurality of tasks, and a plurality of shared resources are shared among these tasks. In a computer system that accesses under exclusive control, there is an access time measuring means for measuring the elapsed time of access to a shared resource by a task, and an access limit time set in advance for the current elapsed access time obtained by this access time measuring means. an access forced termination means for forcibly terminating the access of the task when the access time is exceeded; and an access limit time learning means for updating the access limit time to an appropriate access limit time based on the access elapsed time obtained by the access time measuring means. This configuration avoids deadlock phenomena during access to shared resources.

(Operation) Therefore, by taking the above-described measures, the present invention learns the access limit time each time access to the shared resource is completed by the access limit time learning means, and, for example, sets the access limit in the memory after a certain period of time has elapsed. Update time. Thereafter, when a new access is started by a task, the access forced termination means forcibly terminates the access of the corresponding task when the current access elapsed time obtained by the access time measurement means exceeds the access limit time. By releasing access to other tasks, the execution conditions for other tasks are relaxed, and deadlock phenomena can be avoided. On the other hand, the access limit time learning means adds an appropriate value to the access limit time based on the fact that the access limit time has been exceeded to set the optimum access limit time.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention. In the figure, reference numeral 1 denotes a real-time operating system in a computer system, and this system 1 stores execution conditions such as mutual exclusive exclusive control conditions for each task 2, and performs exclusive control based on the stored contents. It has a function of accessing other shared resources 4 via the deadlock control unit 3 while performing the same operation.

The deadlock control unit 3 is composed of a type of task that has a function to prevent a deadlock phenomenon from occurring and to minimize the influence of a deadlock phenomenon when it is likely to occur.

This deadlock control unit 3 includes independent control memory blocks 5 corresponding to each shared resource 4, etc., as well as the name of the task and various other statuses when a deadlock of the shared resource 4 occurs. An auxiliary storage device 6 is provided to store the information and provide it for later factor analysis.

This control memory block 5 is provided with an access start time memory 51 for storing an access start time, an access limit time memory 52 for storing an access limit time, and an access limit learning time memory 53 for storing a learned value of an access limit value. ing.

On the other hand, the deadlock control section 3 functionally includes an access time measuring means 31 for determining the elapsed access time from the access start time of the access start time memory 51 to the present time when the shared resource 4 is accessed; an access forced termination means 32 for forcibly terminating the access to the shared resource 4 by the task when the access elapsed time obtained by the access time measuring means 3 exceeds an access limit time which is successively updated and set; and the access time measuring means 3
When the access elapsed time obtained in step 1 exceeds the access limit time, the access limit time learning means 32 learns the optimum access limit time and updates and registers the access limit time.

Next, while executing exclusive control, the operating system 1
The operation flow of the deadlock control unit 3 will be described separately for cases in which there is a request to start access to another shared resource 4 from the task, and cases in which there is a request to end access to the task currently being accessed.

(1) When there is a request to start access (see Figure 2)
.

Now, when there is a request to start accessing another shared resource 4 from the operating system 1, the deadlock control unit 3 determines whether there is a shared resource 4 currently being accessed for all shared resources 4 that have been accessed in the past. If the shared resource 4 currently being accessed does not exist, the access start time TO is stored in the access start time memory 51 corresponding to the shared resource 4 to be accessed this time (step S1).
is set (step S8), and the request to start accessing the corresponding shared resource 4 is terminated.

On the other hand, if there are other shared resources 4 that are being accessed in step S1, the access time measuring means 31 sets the start time TO in the access start time memory 51 of the control memory block 5 for each of those shared resources 4.
The elapsed access time T from the time to the present time is calculated (step S2).

After that, the access forced termination means 32 performs step S3.
After reading the access limit time TL from the access limit time memory 52, it is determined whether the access elapsed time T exceeds the access limit time TL (step S4). Here, if the access elapsed time T has not yet exceeded, it is determined that there is no possibility of deadlock, and the process moves to step S8, and the access start time To is stored in the access start time memory 51 corresponding to the shared resource 4 to be accessed this time. is set (step S8).

Furthermore, when it is determined in step S4 that the access elapsed time T exceeds the access limit time TL,
When it is determined that there is a possibility of deadlock occurring in the shared resource 4 due to some kind of hindrance to exclusive control, the names of the relevant control memory block 5 and the task 2 being accessed are edited as appropriate and then stored in the auxiliary storage device 6. . The reason why names and the like are stored in the auxiliary storage device 6 is to analyze the cause of the task 2 and the shared resource 4 that may cause a deadlock later.

After that, the access forced termination means 32 executes step S6.
In this step, a forced termination request signal for the relevant task is sent to the first operating system. Here, after the corresponding task is forcibly terminated by the operating system 1 or the access forced termination means 32, the access limit time learning means 33 calculates a certain ratio of about N1% of the access limit time TL of the corresponding shared resource 4. is added and stored in the access limit time memory 52 as a new access limit time TL, and the previous access limit time TL is stored in the access limit learning time memory 53 as an access limit learning time TA (step S7). Thereafter, the process moves to step S8, where the access start time TO is set in the access start time memory 51 corresponding to the shared resource 4 that has newly started accessing this time, and this operation procedure is ended.

(2) When there is a request to start access (see Figure 3)
.

When the operating system 1 issues a request to terminate access to the shared resource 4, the access time measuring means 3 of the dead-end/tsuku control unit 3 performs an access termination request on the control memory block 5 corresponding to the shared resource 4 to be accessed in step Sll. While reading the start time To from the access start time memory 51, the current time T1 is obtained from the operating system or an internal clock device, and the current access time Ts is calculated by the operation (T, -To).

Thereafter, the access limit time TL is learned under the following conditions based on the current access time Ts and stored in the access limit time memory 52.

That is, the learning of the access limit time TL is based on a certain access limit time T when the shared resource 4 is initially created.
After setting L in the access limit time memory 52, when the relationship of access time Ts > access limit learning time TA is established at regular intervals (for example, 10) from the start of the computer system, the access time Ts is learned as an access limit. It is stored in the access limit learning memory 53 as time TA. After the fixed period ends, a fixed ratio of about N2% is added to the access limit learning time TA, and this is stored in the access limit time memory 52 as a new access limit time TL (slit S2).

Therefore, according to the configuration of the embodiment configured as described above, after the access from the operating system 1 is started,
The deadlock control unit 3 monitors the elapsed time T of the shared resource 4, and when this elapsed time T exceeds the access limit time TL for each shared resource 4, the task 2 that is accessing the shared resource is forcibly terminated. By canceling access to task 4 and releasing the corresponding shared resource 4 to another task 2, the entire computer system can maintain normal operation, and a large-scale system deadlock can be prevented. can. Further, by comparing the elapsed time and the access limit time of each shared resource 4 and updating the access limit time every - period, the access limit time of each shared resource 4 can be appropriately set. Furthermore, when the elapsed time of each shared resource 4 exceeds the access limit time TL, information such as the control memory block of the shared resource 4 and the name of the task currently being accessed is stored in the auxiliary storage device 6. information can be easily analyzed, the cause of the abnormality can be quickly investigated, and system development work can be carried out efficiently.

Note that the present invention can be implemented with various modifications without departing from the gist thereof.

[Effects of the Invention] As explained above, according to the present invention, the access limit time for each shared resource is set in advance, and when the elapsed time from the access start time of the shared resource exceeds the access limit time of the target shared resource. Since the task being accessed is forcibly terminated and the shared resources are released to other tasks, it is possible to prevent deadlock phenomena from occurring in the computer system.

[Brief explanation of the drawing]

1 to 3 are shown to explain an embodiment of a deadlock prevention device for a computer system according to the present invention, and FIG. 1 is a block diagram showing an embodiment of the device of the present invention, and FIG. FIG. 2 is an operation flowchart when an access start request is received from the operating system, and FIG. 3 is an operation flowchart when an access termination request is received from the operating system. 1... Operating system, 2... Task,
3...Deadlock control unit, 4...File and other shared resources, 5...Control memory block, 6...Auxiliary storage device, 31...Access time measurement means, 32...
- Access forced termination means, 33...Access limit time learning means, 51...Access start time memory, 52
...Access limit time memory, 53...Access limit learning time memory. Applicant's agent Patent attorney Takehikoen Suzue - 1~ Figure 1

Claims (1)

[Claims] In a computer system in which a plurality of tasks are provided and these tasks access a plurality of files and other resources (hereinafter collectively referred to as shared resources) under exclusive control, An access time measuring means for measuring time; and an access forced termination for forcibly terminating the access of the task when the current elapsed access time obtained by the access time measuring means exceeds a preset access limit time. and access limit time learning means for updating the access limit time to an appropriate access limit time based on the access elapsed time obtained by the access time measuring means, to avoid deadlock phenomenon during access to a shared resource. Deadlock prevention device for computer systems.