JP4387863B2 - Disturbance occurrence detection program and disturbance occurrence detection method - Google Patents

Description

  The present invention relates to a disturbance occurrence detection technique for detecting a problem program in a computer system, and in particular, a disturbance occurrence detection program for rapidly and reliably detecting a disturbance generated in a system without modifying an existing control program structure, and The present invention relates to a disturbance detection method.

  In online operations (ATM control, etc.), CPU resources are used / run in application programs to achieve the desired processing. However, due to inadequate control program structure or other application programs, the CPU resources May continue to be used carelessly, CPU resources may not be used by application programs, and a serious problem may occur where online stop occurs. A program that keeps using CPU resources carelessly and does not pass the CPU execution control right to other programs is called a “disturbance program”.

  Here, the disturbance means that the timing at which the CPU execution control right (CPU resource use right) is passed to an application program that needs to use the CPU resource is delayed. One cause of the disturbance is that a program with a high execution priority uses CPU resources for a long time without generating an interrupt. Disturbances occur when the program structure continues to use CPU resources without generating an interrupt.

  Even when interrupts are not prohibited, if a number of programs are looped in vain, the timing at which the CPU execution control right is passed to an application program that requires the use of CPU resources will be delayed and disturbed. May occur.

  Even if an attempt is made to identify (detect) a disturbance program after a disturbance has occurred, the information at that time does not already exist as investigation data. For this reason, in the past, a method was used to detect disturbance programs in anticipation.

Further, as a conventional technique for detecting the occurrence of a disturbance by a function incorporated in a control program, there is one described in Patent Document 1 below, which includes a counter mechanism in a thread control device, A technique is described in which a counter value is checked at a random monitoring time to detect a thread that performs an abnormal operation.
Japanese Patent Laid-Open No. 2000-10800

  The conventional technology that takes the above-mentioned prediction and detects the disturbance program requires a lot of time and man-hours, and may not solve the problem.

  Further, in the prior art described in Patent Document 1, a problem is solved by using a detection method based on modification of a control program (OS). When an existing control program is used, The program developer cannot deal with it. Further, when an existing control program is modified, there is a problem that the influence is large including re-installation of the control program itself, and a great number of man-hours are spent on the man-hour / period.

  An object of the present invention is to solve the above-described problems of the prior art and to provide a technique capable of detecting a disturbance generated in the system at high speed and without altering the existing control program structure. .

  In order to detect a disturbance program having a structure that continues to use CPU resources without generating an interrupt for a long time, when one process is finished in the disturbance program and an interrupt (execution program switching) occurs, It is necessary to detect whether or not other programs that must use the resources cannot operate.

  Therefore, the disturbance detection program of the present invention sets the execution priority of itself higher, acquires the current time X1, sets the timer interrupt time, monitors the time, and enters the execution state by the timer interrupt. The occurrence of the disturbance is detected by verifying whether the difference between the acquired time X2 and the time X1 exceeds the disturbance detection limit time, which is a time with a margin in the execution period of the disturbance occurrence detection program.

As a premise of the disturbance occurrence detection processing of the present invention, the following system environment setting is performed in advance.
(1) For the disturbance occurrence detection program, CPU allocation is set so that each CPU resource can be executed when there are a plurality of CPUs. In addition, a running mode in which space expulsion (page out) is prohibited is set.

  The CPU allocation setting that can be executed for each CPU resource is performed in the case of multiprocessor control. For example, if the disturbance detection program is not executable for each CPU resource, the CPU has an execution control right for one CPU. This is because the disturbance occurrence detection program can be executed by running on other CPUs even if the disturbance program does not pass the CPU execution control right to the disturbance occurrence detection program.

The CPU allocation setting and the page-out prohibited travel mode setting that can be executed for each of the plurality of CPU resources are set as environment setting parameters for each program, for example.
(2) Also, set the disturbance detection point time, disturbance detection limit time, and execution priority according to the startup parameters of the disturbance detection program.

  The disturbance detection point time is an execution cycle of the disturbance occurrence detection program, in other words, a time corresponding to the time for allocating CPU resources to each program.

  Disturbance detection limit time is a time with a certain margin in the disturbance detection point time. When the disturbance detection program exceeds the time from the time monitoring start time to the execution time, this time is exceeded. It is time to judge that a disturbance has occurred.

  The startup parameters are set by, for example, specifying each parameter [PARM = 'xxxx, yyyyyy, 00pppp'] in the EXEC statement of the job control statement. If the disturbance detection point time is A milliseconds (ms), the disturbance detection limit time is B milliseconds (ms), and the execution priority is C, for example, [PARM = 'A, B, C'] is designated. This execution priority can be set up to 255.

  Specifically, when the CPU resource allocation time (that is, the disturbance detection point time) to each program is 100 ms, the disturbance detection limit time is 500 ms, and the execution priority is 255, [PARM = '00100, 00500,00255'] Is specified. If the disturbance detection program is a type of system that is started by a command instead of a job control statement, specify the same in the start parameter of that command.

In the present invention, after the system environment is set, the occurrence of a disturbance is detected as follows.
(3) The disturbance occurrence detection program sets its execution priority with reference to the startup parameters.
(4) The disturbance occurrence detection program performs time monitoring after acquiring the current time X1.
(5) The disturbance occurrence detection program acquires the time X2 when the execution state is entered by the timer interruption after the monitoring time has elapsed.
(6) The difference between the time X2 and the time X1 is compared with the disturbance detection limit time.
(7) When the difference between the time X2 and the time X1 is greater than the disturbance detection limit time, it is determined that a disturbance has occurred because the CPU resources are continuously used by the disturbance program, and the space dump process is immediately performed. That is, for example, the trace information in the OS in which the execution history of each program is recorded is read and stored in dump information storage means such as a disk. Based on the dump information stored in the dump information storage means, it is possible to detect which program caused the disturbance.

  The processes (4) to (7) will be described with reference to FIGS. FIG. 6 is an operation time chart of the disturbance occurrence detection program. If the set disturbance detection point time is Ams, the disturbance occurrence detection program sets the timer interruption after Ams and waits for time monitoring. As a result, the disturbance occurrence detection program should normally be executed in an Ams cycle.

Specifically, disturbance occurrence detection program, the time t ₀ ~t first run between _1, first wait between times _{t 1 ~t 2 (WAIT),} 2 time between time t ₂ ~t ₃ waiting for execution, the time t ₃ ~t ₄ a second time (wAIT), the time t ₄ ~t ₅ for the third time of execution, time t ₅ ~t ₆ the third time of waiting (wAIT), the time t ₆ ~t ₇ is the fourth execution, and so on.

  7A and 7B are diagrams for explaining the detection of the occurrence of a disturbance. FIG. 7A shows a normal case, and FIG. 7B shows a case where a disturbance has occurred. In the example shown in FIG. 7, it is assumed that the set disturbance detection point time is Ams and the disturbance detection limit time is Bms.

  In FIG. 7A, the disturbance occurrence detection program acquires the time X1 at the start point shown in the figure. Set the disturbance detection point time Ams as a timer and monitor the time. Then, the time X2 is acquired when the execution state is caused by the timer interrupt.

  The disturbance occurrence detection program determines that a disturbance has occurred if the difference between the time X2 when the execution state is caused by the timer interrupt and the time X1 acquired at the start point is greater than the disturbance detection limit time Bms.

  Even if the difference between the time X2 and the time X1 is greater than the disturbance detection point time Ams, it is determined that no disturbance has occurred if it is smaller than the disturbance detection limit time Bms. For example, in the case of FIG. 7A, since the difference between time X2 and time X1 is smaller than Bms, it is determined that no disturbance has occurred. On the other hand, in the case of FIG. 7B, since the difference between the time X2 and the time X1 is larger than Bms, it is determined that a disturbance has occurred.

  That is, the present invention is a disturbance occurrence detection program for detecting the occurrence of a disturbance caused by a disturbance program, which is a program that continues to use CPU resources carelessly, and sets the execution priority of the disturbance occurrence detection program as a disturbance detection target. Disturbance detection so that the disturbance occurrence detection program obtains the CPU execution control right and enters the execution state at every predetermined disturbance detection point time and the execution priority setting procedure set to a value equal to or higher than the execution priority of the program to be Set the point time to the timer and monitor the time in the timer interrupt wait state, and when the disturbance detection program enters the execution state due to the timer interrupt, the time monitoring starts and the execution state Is compared with a predetermined disturbance detection limit time, and the time monitoring is started after the time monitoring is started. Time until the line state, if exceeding a predetermined disturbance detection limit time, a disturbance verification procedure determines that the disturbance has occurred, and characterized by causing a computer to execute.

  Further, the disturbance occurrence detection program of the present invention provides a computer with a spatial dump processing procedure for outputting information of a predetermined memory area of the computer to an external storage device when it is determined that a disturbance has occurred in the disturbance verification procedure. Includes a program to be executed.

  The execution priority, the disturbance detection point time, and the disturbance detection limit time are given by parameters at the start of the disturbance occurrence detection program.

  The disturbance occurrence detection program is configured as a processing program different from the control program constituting the operating system and operates separately from the control program.

  Also, the disturbance occurrence detection method according to the present invention assigns the disturbance occurrence detection program to each CPU constituting the multiprocessor system when the occurrence of the disturbance in the multiprocessor system is detected, and operates only on a specific CPU. It is characterized by making it.

  According to the present invention, it is possible to detect a problem program in the system at the level of a processing program without altering the existing control program structure, and to shorten the period until the disturbance program is detected in the system and to reduce the man-hours. Can do. Therefore, according to the present invention, it is possible to detect a disturbance generated in the system at high speed and with certainty, which is highly effective in improving workability. In addition, since it is not necessary to modify the control program structure of the OS, it is excellent in portability to other systems.

  Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing an example of the configuration of the disturbance occurrence detection apparatus of the present invention. Reference numeral 1 denotes a disturbance occurrence detection device including a CPU and a memory. The disturbance occurrence detection apparatus 1 in this embodiment is obtained by installing a disturbance occurrence detection program 11 in a data processing apparatus using a general computer.

  In the disturbance occurrence detection apparatus 1, 10 is an OS (operating system), 11 is a disturbance occurrence detection program, 12 is an application program X, 13 is an application program Y, and 14 is an application program Z. In the OS 10, 100 is a control program A, 101 is a control program B, and 102 is a control program C, and each implements various functions of the OS 10.

  In the disturbance occurrence detection program 11, 110 is an execution priority setting unit that sets the execution priority of the disturbance occurrence detection program 11, 111 is a time monitoring unit that sets the disturbance detection point time as a timer, and performs time monitoring, and 112 is time monitoring. A disturbance verification unit 113 that verifies the occurrence of a disturbance based on a time monitoring result by the unit 111, and a spatial dump processing unit 113 that performs a spatial dump process when the disturbance verification unit 112 determines that a disturbance has occurred. Reference numeral 200 denotes dump information storage means for storing dump information as a result of the space dump process. Spatial dumping means outputting the data of the memory of the disturbance occurrence detection device 1 to an external storage device.

  FIG. 2 is a diagram showing a disturbance occurrence detection program and execution priorities of the respective disturbance detection targets. In FIG. 2, 103 is a trace information storage unit in the OS 10, and 201 is trace information stored in the trace information storage unit. It is assumed that the disturbance occurrence detection program 11 has a CPU allocation setting that can be executed for each of a plurality of CPU resources.

  In the example shown in FIG. 2, the program at the top of the figure has a higher execution priority. In FIG. 2, the execution priority is higher in the order of the disturbance occurrence detection program 11, each control program in the OS 10, and each application program.

  In the embodiment of the present invention, the execution priority of each program can be set up to a maximum of 255, and the execution priority of the disturbance occurrence detection program 11 is set to a higher value than the program that is the target of the disturbance occurrence detection. . If there is another program whose execution priority is higher than that of the disturbance occurrence detection program 11, even if the execution of the disturbance detection target program ends after the monitoring time has elapsed, the disturbance occurrence detection program 11 has immediate CPU execution control right. This is because there is a possibility that the disturbance cannot be detected.

  For example, by setting the execution priority of the disturbance occurrence detection program 11 to the maximum value of 255, occurrence of disturbance can be detected even when each control program of the OS 10 is a disturbance program. Although the execution priority of the disturbance occurrence detection program 11 is preferably higher than the execution priority of the disturbance detection target program, the occurrence of disturbance can be detected even if the execution priority is the same.

  When the disturbance detection program 11 determines that a disturbance has occurred, it reads the trace information 201 in the trace information storage unit 103 in the OS 10 and stores it in the dump information storage means 200. Therefore, the user can detect the disturbance program based on the dump information stored in the dump information storage means 200.

  FIG. 3 is a diagram illustrating a configuration example of the trace information. The trace information 201 includes a program name, interrupt information, the time when the interrupt occurred, and other items. Each number recorded in the item of interrupt information indicates, for example, the type of interrupt.

  In the example of FIG. 3, the control program A100 enters an execution state at interrupt “2” (eg, timer interrupt) at time “1000001000001”, and the disturbance detection program 11 executes at interrupt “2” at time “1000001000011” and time “1000001000028”. It shows that it became a state.

  As shown in FIG. 3, since the trace information 201 indicates the execution history of each program, it is possible to detect which program caused the disturbance based on the trace information 201. The trace information 201 is automatically collected in a predetermined trace information storage unit 103 in the memory when the execution control of the program is switched by the OS 10 due to an interrupt or the like. When the OS 10 does not have such a trace function, other predetermined memory areas may be dumped instead of the trace information storage unit 103.

  FIG. 4 is a diagram illustrating an example of a disturbance occurrence detection process flow. In this example, it is assumed that the disturbance detection point time is set to Ams, the disturbance detection limit time is Bms, and the execution priority is set to 255, which is the maximum value, as startup parameters.

  When the startup parameter is set, the disturbance detection program 11 first sets its execution priority with reference to the startup parameter (step S1). Here, the disturbance occurrence detection program 11 sets the maximum value 255 as the execution priority.

  Next, the disturbance occurrence detection program 11 acquires and stores the current time X1 using a time acquisition macro instruction provided by the OS 10 (step S2). Next, referring to the disturbance detection point time (Ams) in the startup parameter, the monitoring time is set in the timer using a macro instruction of the OS 10 (step S3), and time monitoring (WAIT) is performed (step S4).

  The disturbance occurrence detection program 11 acquires the current time X2 when the monitoring time elapses and enters the execution state due to the timer interruption (step S5), and determines whether X2−X1> B, that is, whether a disturbance has occurred. (Step S6). If X2-X1> B, a space dump process is performed (step S7). The user can detect the disturbance program based on the dump information stored in the dump information storage means 200 by the space dump process. This is because it can be seen from the trace information as shown in FIG. 3 which program has been using the CPU abnormally long.

  The space dump process in step S7 is performed by, for example, the following SDUMP macro instruction. Since the description of such a macro instruction depends on the type of the OS 10, only a notation example is shown here.

SDUMP HDR = (DISTURBANCE ABNORMAL END)
If X2-X1> B is not satisfied in step S6, X2 is stored as X1 (step S8), and the process returns to step S3.

  FIG. 5 is a diagram illustrating another example of the disturbance occurrence detection processing flow. It is substantially the same as the flow of FIG. 4 described above except that the acquisition methods of the current times X1 and X2 are different. Also in this example, it is assumed that the disturbance detection point time is set to Ams, the disturbance detection limit time is Bms, and the execution priority is set to 255, which is the maximum value, as startup parameters.

  When the startup parameter is set, the disturbance detection program 11 first sets its own execution priority with reference to the startup parameter (step S11). The disturbance occurrence detection program 11 sets the maximum value 255 as the execution priority.

  Next, the disturbance occurrence detection program 11 acquires the current time X1 (step S12). With reference to the disturbance detection point time (Ams) in the startup parameter, the monitoring time is set in the timer (step S13), and the time monitoring (WAIT) is performed (step S14).

  When the disturbance occurrence detection program 11 enters an execution state due to a timer interrupt, it acquires the current time X2 (step S15), and determines whether X2−X1> B, that is, whether a disturbance has occurred (step S16). If X2-X1> B, a space dump process is performed (step S17). The user can detect the disturbance program based on the dump information stored in the dump information storage means 200 by the space dump process.

  If X2-X1> B is not satisfied in step S16, the process returns to step S11 and the execution priority is reset. This is to keep the execution priority of the disturbance occurrence detection program 11 at the initially set maximum value even when the OS 10 automatically lowers the execution priority of the disturbance occurrence detection program 11.

  If the disturbance program falls into an infinite loop with interrupts disabled, the disturbance detection program 11 will not be executed even after the monitoring time has elapsed, and the spatial dump processing by the spatial dump processing unit 113 is not performed. However, since such a disturbance program can be recognized immediately without using the disturbance occurrence detection program 11, there is no problem even if the disturbance occurrence detection program 11 cannot detect the disturbance program.

  As can be understood from the above, the features of the embodiment of the present invention are described as follows.

(Supplementary note 1) A disturbance detection program for detecting the occurrence of a disturbance by a disturbance program, which is a program that continues to use CPU resources carelessly,
An execution priority setting procedure for setting the execution priority of the disturbance occurrence detection program to a value equal to or higher than the execution priority of the program whose disturbance is to be detected;
Time monitoring procedure for setting the disturbance detection point time in the timer and monitoring the time in the timer interrupt waiting state so that the disturbance occurrence detection program gets the CPU execution control right and enters the execution state at every predetermined disturbance detection point time When,
When the disturbance occurrence detection program enters the execution state due to a timer interrupt, the time from the start of the time monitoring to the execution state is compared with a predetermined disturbance detection limit time, and the time monitoring is performed. A disturbance verification procedure for determining that a disturbance has occurred if the time from the start to the execution state exceeds a predetermined disturbance detection limit time,
Disturbance detection program to be executed by a computer.

(Appendix 2) In the disturbance occurrence detection program described in Appendix 1,
A spatial dump processing procedure for outputting information of a predetermined memory area of the computer to an external storage device when it is determined that a disturbance has occurred in the disturbance verification procedure;
A disturbance detection program including a program for causing a computer to execute.

(Appendix 3) In the disturbance occurrence detection program described in Appendix 1 or Appendix 2,
The disturbance occurrence detection program, wherein the execution priority, the disturbance detection point time, and the disturbance detection limit time are given by parameters at the start of the disturbance occurrence detection program.

(Supplementary Note 4) The disturbance occurrence detection program according to Supplementary Note 1, Supplementary Note 2 or Supplementary Note 3, wherein the disturbance occurrence detection program is configured as a processing program different from the control program constituting the operating system. program.

(Supplementary note 5) A disturbance occurrence detection method using the disturbance occurrence detection program according to any one of supplementary notes 1 to 4,
A disturbance occurrence detection method, wherein, when the occurrence of a disturbance in a multiprocessor system is detected, the disturbance occurrence detection program is assigned to each CPU constituting the multiprocessor system and is operated only by a specific CPU.

  (Additional remark 6) The disturbance generation | occurrence | production detection program recording medium which recorded the disturbance generation | occurrence | production detection program of any one of Additional remark 1 to Additional remark 4 on the computer-readable recording medium.

It is a figure which shows an example of a structure of a disturbance generation | occurrence | production detection apparatus. It is a figure which shows the execution priority of each program used as the disturbance detection program and the disturbance detection target. It is a figure which shows the structural example of trace information. It is a figure which shows an example of a disturbance generation | occurrence | production detection processing flow. It is a figure which shows another example of a disturbance generation | occurrence | production detection processing flow. It is an operation | movement time chart of a disturbance generation | occurrence | production detection program. It is a figure explaining detection of disturbance generation.

Explanation of symbols

1 Disturbance detection device 10 OS
11 Disturbance detection program 12 Application program X
13 Application program Y
14 Application program Z
100 Control program A
101 Control program B
102 Control program C
103 Trace information storage unit 110 Execution priority setting unit 111 Time monitoring unit 112 Disturbance verification unit 113 Spatial dump processing unit 200 Dump information storage unit 201 Trace information

Claims (5)

A disturbance occurrence detection program for detecting the occurrence of a disturbance caused by a disturbance program that continues to use CPU resources carelessly,
An execution priority setting procedure for setting the execution priority of the disturbance occurrence detection program to a value equal to or higher than the execution priority of the program whose disturbance is to be detected;
Time monitoring procedure for setting the disturbance detection point time in the timer and monitoring the time in the timer interrupt waiting state so that the disturbance occurrence detection program gets the CPU execution control right and enters the execution state at every predetermined disturbance detection point time When,
When the disturbance occurrence detection program enters the execution state due to a timer interrupt, the time from the start of the time monitoring to the execution state is compared with a predetermined disturbance detection limit time, and the time monitoring is performed. A disturbance verification procedure for determining that a disturbance has occurred if the time from the start to the execution state exceeds a predetermined disturbance detection limit time,
Disturbance detection program to be executed by a computer. In the disturbance detection program according to claim 1,
A spatial dump processing procedure for outputting information of a predetermined memory area of the computer to an external storage device when it is determined that a disturbance has occurred in the disturbance verification procedure;
A disturbance detection program including a program for causing a computer to execute. In the disturbance occurrence detection program according to claim 1 or 2,
The disturbance occurrence detection program, wherein the execution priority, the disturbance detection point time, and the disturbance detection limit time are given by parameters at the start of the disturbance occurrence detection program. The disturbance occurrence detection program according to claim 1, 2 or 3, wherein the disturbance occurrence detection program is configured as a processing program different from a control program constituting an operating system. . A disturbance occurrence detection method using the disturbance occurrence detection program according to any one of claims 1 to 4,
A disturbance occurrence detection method, wherein, when the occurrence of a disturbance in a multiprocessor system is detected, the disturbance occurrence detection program is assigned to each CPU constituting the multiprocessor system and is operated only by a specific CPU.

Images