Information processing apparatus and information processing method

The information processing apparatus uses thread state transitions and a trained classifier to identify and locate program malfunctions, addressing the challenge of prolonged service interruptions in complex software systems.

JP2024162655AActive Publication Date: 2024-11-21INTERNET INITIATIVE JAPAN INC

Patent Information

Application Number
JP2023078382
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-11
Publication Date
2024-11-21
Estimated Expiration
2043-05-11

AI Technical Summary

Technical Problem

Conventional methods struggle to easily identify the cause and location of program malfunctions in complex software systems, leading to prolonged service interruptions.

Method used

An information processing apparatus and method that records transitions of execution states of multiple threads, uses a trained classifier to analyze variable values, and identifies the cause and location of program abnormalities by correlating thread execution states with predefined labels.

Benefits of technology

Facilitates quick and accurate identification of program malfunctions, enabling rapid recovery and minimizing service disruptions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024162655000001_ABST
    Figure 2024162655000001_ABST
Patent Text Reader

Abstract

To easily identify a cause of a malfunction that occurred in a program.SOLUTION: An information processing apparatus includes: a setting unit 10 configured to set variables for recording transitions of execution state of a plurality of threads in a process which is an execution unit of a program; an acquisition unit 13 configured to acquire values of the variables indicating the transitions of execution state of the threads, the transitions being recorded in accordance with execution of the program; a classifying unit 14 configured to provide the values of the variables indicating the transitions of execution state of the threads, acquired by the acquisition unit 13, as unknown inputs, to a trained classifier, and output a class related to a cause of a malfunction of the program by executing calculation using the trained classifier; and a presentation unit 16 configured to present the cause of the malfunction of the program, on the basis of the class output by the classifying unit 14.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to an information processing device and an information processing method, and more particularly to a technique for identifying an operational abnormality in a program. [Background technology]

[0002] In recent years, the demands on software have become more sophisticated and complex in order to provide high-performance systems and services. Software implemented in high-performance systems consists of programs with huge amounts of source code, and the execution of programs has become even more complex due to parallelization such as multithreading.

[0003] When a process failure or operational abnormality occurs in an operational system, it takes time to identify the part of the program that is causing the failure or operational abnormality. If it takes a long time to recover from the failure, the service will be interrupted for a long time, which will have a significant impact on users.

[0004] Conventionally, when an error occurs during process execution that makes it impossible to continue processing due to a system failure or abnormal processing of a program, and the program is forcibly terminated, a configuration is known in which the memory state is dumped to a core file in the system. For example, Patent Document 1 discloses a technique in which a core file output to a hard disk is read into a debugger to retroactively analyze the location where the error occurred and the values ​​of variables at that time.

[0005] However, with the technology disclosed in Patent Document 1, it is not possible to identify the cause of a program's abnormal operation and the location where the error occurred from the log information, so it is necessary to further analyze the source code to identify the cause and the location where the error occurred. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2005-301570 A Summary of the Invention [Problem to be solved by the invention]

[0007] Thus, in the conventional technology, when an abnormality occurs in a program's operation, it is not possible to easily identify the cause of the abnormality.

[0008] The present invention has been made to solve the above-mentioned problems, and has an object to easily identify the cause of an operational abnormality in a program when the abnormality occurs. [Means for solving the problem]

[0009] In order to solve the above-mentioned problems, the information processing device of the present invention comprises a setting unit configured to set variables for recording transitions of the execution states of multiple threads within a process, which is an execution unit of a program; an acquisition unit configured to acquire values ​​of variables indicating the transitions of the execution states of the multiple threads recorded in accordance with the execution of the program; a classification unit configured to provide the values ​​of the variables indicating the transitions of the execution states of the multiple threads acquired by the acquisition unit as unknown inputs to a trained classifier, perform calculations of the trained classifier, and output a classification class related to the cause of the operation abnormality of the program; and a presentation unit configured to present the cause of the operation abnormality of the program based on the classification class output by the classification unit.

[0010] In addition, the information processing device of the present invention may further include an identification unit configured to identify a location where the operational abnormality occurred in the program based on the value of the variable indicating the transition of the execution state of the multiple threads acquired by the acquisition unit, and the presentation unit may be configured to present information indicating the location where the identified operational abnormality occurred.

[0011] In addition, the information processing device according to the present invention may further include a learning unit configured to learn the relationship between the values ​​of the variables indicating the transitions of the execution states of the multiple threads and causes of the operational abnormality using training data in which a predetermined cause of the operational abnormality is assigned as a correct answer label to the values ​​of the variables indicating the transitions of the execution states of the multiple threads, and to construct the trained classifier.

[0012] In addition, in the information processing device of the present invention, the value of the variable indicating the transition of the execution states of the multiple threads may include either a first value indicating that a first thread of the multiple threads has transitioned to a second execution state in which a second thread executes a second instruction after executing a first instruction in a first execution state, or in parallel, and a second value indicating that the first execution state of the first thread did not end normally and did not transition to the second execution state because the first thread was unable to execute the first instruction.

[0013] In addition, in the information processing device of the present invention, when the value of the variable indicating the transition of the execution states of the multiple threads acquired by the acquisition unit includes the second value, the identification unit may determine that the first instruction that could not be executed by the first thread in the first execution state that did not terminate normally, as indicated by the second value, corresponds to the location where the operational abnormality occurred in the program.

[0014] In order to solve the above-mentioned problems, the information processing method of the present invention includes a first step of setting variables for recording transitions of execution states of multiple threads within a process, which is an execution unit of a program; a second step of acquiring values ​​of variables indicating the transitions of the execution states of the multiple threads recorded according to the execution of the program; a third step of providing the values ​​of the variables indicating the transitions of the execution states of the multiple threads acquired in the second step as unknown inputs to a trained classifier, performing calculations of the trained classifier, and outputting a classification class regarding a cause of the operation abnormality of the program; and a fourth step of presenting the cause of the operation abnormality of the program based on the classification class output in the third step.

[0015] In addition, the information processing method of the present invention may further include a fifth step of identifying a location in the program where the operational abnormality occurred based on the value of the variable indicating the transition of the execution state of the multiple threads obtained in the second step, and the fourth step may present information indicating the location where the identified operational abnormality occurred.

[0016] In addition, the information processing method according to the present invention may further include a sixth step of learning a relationship between the values ​​of the variables indicating the transitions of the execution states of the multiple threads and the causes of the operational abnormality using training data in which a preset cause of the operational abnormality is assigned as a correct answer label to the values ​​of the variables indicating the transitions of the execution states of the multiple threads, and constructing the trained classifier.

[0017] Furthermore, in the information processing method of the present invention, the value of the variable indicating the transition of the execution states of the multiple threads may include either a first value indicating that a first thread of the multiple threads has transitioned to a second execution state in which a second thread executes a second instruction after, or in parallel with, executing a first instruction in a first execution state, and a second value indicating that the first execution state of the first thread did not end normally and did not transition to the second execution state because the first thread was unable to execute the first instruction.

[0018] Furthermore, in the information processing method of the present invention, the fifth step may, when the value of the variable indicating the transition of the execution states of the multiple threads obtained in the second step includes the second value, determine that the first instruction that could not be executed by the first thread in the first execution state that did not terminate normally, as indicated by the second value, corresponds to the location where the operational abnormality occurred in the program. Effect of the Invention

[0019] According to the present invention, the values ​​of variables indicating the transition of the execution states of multiple threads recorded in response to the execution of a program are provided as unknown inputs to a trained classifier, and the trained classifier is operated to output a classification class relating to the cause of an abnormality in program operation. Therefore, when an abnormality in program operation occurs, the cause of the abnormality can be easily identified. [Brief description of the drawings]

[0020] [Figure 1] FIG. 1 is a block diagram showing a configuration of an information processing system including an information processing device according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a diagram for explaining an overview of the information processing device according to the present embodiment. [Diagram 3] FIG. 3 is a block diagram showing a hardware configuration of the information processing device according to the present embodiment. [Figure 4]FIG. 4 is a diagram for explaining the learning process performed by the learning unit according to the present embodiment. [Diagram 5] FIG. 5 is a flowchart showing the operation of the information processing device according to the present embodiment. [Figure 6] FIG. 6 is a flowchart showing the operation of the information processing device according to the present embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0021] A preferred embodiment of the present invention will be described in detail below with reference to Figures 1 to 6. In the following description, as an example, a program executed in a core network device 2 that performs communication control and is provided in a core network that complies with a predetermined communication standard such as 5G is treated as a processing target of the information processing device 1. However, a device in which the information processing device 1 implements a program for identifying the cause of an operational abnormality is not limited to the core network device 2.

[0022] [Information processing system configuration] First, an overview of an information processing system including an information processing device 1 according to an embodiment of the present invention will be described. As shown in Fig. 1, the information processing system includes the information processing device 1 and a core network device 2, and the information processing device 1 and the core network device 2 are connected via a network NW such as a Wide Area Network (WAN), a Local Area Network (LAN), or the Internet.

[0023] The information processing system according to the present embodiment infers the cause of an operational abnormality in a program executed on a computer of the core network device 2, and identifies the location of the operational abnormality in the program. The core network device 2 is provided in a 5G core network and performs communication control.

[0024] The core network device 2 can be realized by a computer equipped with a processor, a main memory device, a communication interface, an auxiliary memory device, and an input / output I / O, and a program that controls these hardware resources. The core network device 2 is, for example, a device that applies a general-purpose multitasking OS such as Linux (registered trademark) or Windows (registered trademark).

[0025] The programs executed on the computer of the core network device 2 include multi-threaded processes. A process, which is the execution unit of a program, is managed by the OS, holds variables and states used within the program, and is composed of one or more threads.

[0026] A thread is a unit of program execution from the CPU's perspective, and one thread is assigned to each CPU core. Since the OS assigns an independent memory area to each process, processes do not usually share memory areas. On the other hand, threads share the same memory area within a process.

[0027] In this embodiment, a case will be described as an example in which a program for which an operational abnormality is inferred executes one process using a combination of multi-threading and a single thread, but the process may be multiple processes.

[0028] In this embodiment, the state of a thread refers to the state of a CPU core assigned to the thread and the state of memory holding various variable values. The transition of a thread's execution state refers to a state transition in which a thread executes one or more predetermined instructions according to source code when instructions are executed sequentially, in parallel, or in parallel within a process.

[0029] For example, a thread executes an instruction by reserving the memory space required for the CPU core to execute the instruction. When this thread executes an instruction, the CPU calculates the address of the next instruction to be executed, and switches to the same thread or a different thread to execute the next instruction. Such a transition of the thread state is called an execution state transition.

[0030] When a thread is operating normally, after the thread executes a predetermined instruction, for example, the same or another thread sequentially executes the next instruction, or another thread executing simultaneously executes the next instruction. On the other hand, when an operational abnormality occurs in a thread, the thread does not end normally and enters a state in which the next instruction cannot be executed. Thus, in this embodiment, the transition of the execution state of a thread indicates a state transition during normal operation, or a state transition in which the thread cannot execute all of the predetermined instructions due to some operational abnormality and cannot transition to the next execution state.

[0031] In addition, in this embodiment, the position of the corresponding source code can be identified from the instruction executed in each execution state of the thread. Here, the source code refers to an instruction to a computer consisting of a character string such as text written in a programming language or a text file. The source code defines the structure and operation of a program, and describes instructions such as functions and classes to be executed by the thread. The generation of a thread can be clearly indicated in advance in the source code. Note that the thread actually executes instructions obtained by converting the source code into machine code.

[0032] Figure 2 shows the states of each thread, S1, S2, . . ., S t-1 ,S t In FIG. 2, states S1, S2, and S t Each of the states S1 to S t In each state, a thread executes one or more predetermined instructions. t-1 and S3',...,St-1 ' indicates each state in the multithread operation. The states S3,...,S t-1 and S3',...,S t-1 Each ' constitutes a transaction.

[0033] For example, when a process in a program is running normally, the execution of each thread in the process changes the state to S1, S2, . . . , S t-1 ,S t The execution states of the threads change sequentially or in parallel. On the other hand, if some operational abnormality occurs, the execution state of a thread does not change to the next execution state. In this embodiment, the normal operation and operational abnormality of a program or process particularly refer to the operation of a thread within a process.

[0034] An example of a thread behavior abnormality is deadlock E2 shown in Figure 2. A multithreaded deadlock is an abnormal behavior in which two or more threads block each other, and each thread waits for the other to finish. A deadlock occurs when the order of exclusive memory control is not properly maintained in multithreaded operations.

[0035] Another example of an operational abnormality of a thread is a heap error E1 shown in FIG. 2. A heap error is an operational abnormality that occurs when the heap area, which is a memory area used by a program, is improperly released. In other words, a heap error occurs when the same memory number is used in different processes when a memory hunt is performed using a hash function. Other examples include stack overflow, infinite loop, starvation, etc., and the information processing device 1 according to this embodiment infers the cause of a known operational abnormality in a pre-set thread.

[0036] When a thread operation abnormality such as a deadlock E2 or a heap error E1 occurs in the core network device 2, the thread execution state cannot transition and the process is forcibly terminated. A signal indicating the forced termination of the process is sent to the information processing device 1 via the network NW. In response to the process termination signal, the information processing device 1 infers the cause of the operation abnormality based on the values ​​of variables indicating the transition of the execution states of multiple threads in the target process and identifies the location in the program where the operation abnormality occurred.

[0037] [Functional blocks of information processing equipment] As shown in FIG. 1, the information processing device 1 includes a setting unit 10, a learning unit 11, a storage unit 12, an acquisition unit 13, a classification unit 14, an identification unit 15, and a presentation unit 16.

[0038] The setting unit 10 sets variables for recording transitions of execution states of multiple threads in a process, which is an execution unit of a program. The values ​​of the variables set by the setting unit 10 include, for example, a first value indicating that a first thread among the multiple threads has executed a first instruction in a first execution state, or in parallel with the execution of a second instruction, transitioned to a second execution state in which a second thread executes a subsequent second instruction. Furthermore, the values ​​of the variables include a second value indicating that the first execution state of the first thread did not end normally and did not transition to the second execution state because the first thread cannot execute the first instruction.

[0039] Specifically, the setting unit 10 can set a flag for recording a transition of an execution state by a thread for each of one or more predetermined instructions in the source code of the target program. The one or more predetermined instructions defining each execution state can be set arbitrarily, and can be set according to, for example, the execution order between threads or the description order of the source code.

[0040] 2, when a process is operating normally, a thread is executed and transitions from state S1 to a specified state S2. This indicates that, for example, in accordance with the execution order between threads, a first thread transitions from state S1 where the first thread executes a first instruction to state S2 where the second thread executes a second instruction. In such normal operation, the setting unit 10 can define a flag so that the flag value in state S1 is "1" (first value).

[0041] On the other hand, if some operational abnormality occurs, the first thread cannot execute the first instruction, so state S1 does not end normally and state S1 does not transition to the next state S2 where the second thread executes the second instruction. In this case, the setting unit 10 defines the flag so that the flag value in state S1 is "0" (second value).

[0042] For example, as shown in Figure 2, when each thread in a process is operating normally, the thread states S1 to S t The flag value up to is "1".

[0043] For example, as shown in Figure 2, when deadlock E2 occurs in the multi-threaded section, t-1 ,S t-1 The flag value of each of the threads is set to "0". Then, the state of the subsequent threads S t The flag value of states S1 and S2 in the single-threaded section before the deadlock E2 occurs is also "0." On the other hand, the flag values ​​of states S1 and S2 in the single-threaded section before the deadlock E2 occurs are "1," which indicates normal operation.

[0044] As shown in Figure 2, a heap error E1 occurs in the single-threaded section where the state transitions from S1 to S2. In this case, the thread cannot transition from S1 to S2, so the flag value of S1 is "0" and the flag value of the subsequent state S2 is also "0". Furthermore, the states S3, S3', ..., S3 of all threads, including the subsequent multi-threaded section, are t-1 ,S t-1 ',St The flag value will also be "0".

[0045] The learning unit 11 uses training data in which preset causes of operational abnormalities are assigned as correct answer labels to flag values ​​indicating transitions of the execution states of multiple threads within a process in a program, to learn the relationship between flag values ​​indicating transitions of the execution states of multiple threads and causes of operational abnormalities, and constructs a learned classifier.

[0046] Fig. 4 is a schematic diagram showing a neural network structure used as an example of a classifier learned by the learning unit 11. As shown in Fig. 4, for example, a machine learning model of a neural network structure is adopted as the classifier, and the numbers of nodes in the input layer, hidden layer, and output layer, the number of layers in the hidden layer, etc. are set. The learning unit 11 sets the number of input nodes in the input layer corresponding to the number of steps 1 to t of the transition of the execution state of a thread in a process in a program that infers the cause of an operational abnormality. The number of nodes in the output layer can be the number of nodes according to the classification class.

[0047] As shown in Figure 4, each node in the input layer represents the execution state of a thread, S1 to S t The size and elements of the classifier, including the number of hidden layers h and the density of connections between nodes of the neural network, are not limited as long as the design provides sufficient inference accuracy. For example, the connections between nodes may be fully connected or sparsely connected.

[0048] The learning unit 11 can use, as the teacher data, actual data of flag values ​​indicating transitions of the execution states of threads acquired from the core network device 2 by the acquisition unit 13 described later. In addition, the classification class of the output layer set by the learning unit 11 can be arbitrarily set to two-class or multi-class classification, etc., depending on the number of causes of operational abnormalities to be classified.

[0049] The neural network shown in Figure 4 applies an activation function to the weighted sum of inputs for flag values ​​indicating the transition of the execution states of multiple threads given to the input layer x, and passes the output determined by threshold processing to the output layer y. Each output node in the output layer indicates a classification class related to the cause of the operation abnormality. In addition, the output layer y can output the probability that the input data belongs to a classification class.

[0050] The learning unit 11 adjusts the weights w of the connections between nodes so that the output when a flag value indicating the transition of the execution states of multiple threads is given as an input becomes a value indicating the cause of the operation abnormality indicated in the label of the teacher data. The learning unit 11 can use, for example, backpropagation to compare the obtained output value with the given input value, examine the error of each weight w, and propagate it in the reverse direction, ultimately determining parameters such as the weight w. Through such learning processing, the learning unit 11 constructs a learned classifier.

[0051] The storage unit 12 stores the trained classifier, i.e., trained parameters, constructed by the learning unit 11. In this embodiment, the storage unit 12 stores trained weights of the neural network. The storage unit 12 also stores the classifier before training.

[0052] The acquiring unit 13 acquires flag values ​​indicating transitions of the execution states of multiple threads in a process of a target program from the core network device 2. The flag values ​​indicating transitions of the execution states of the threads are flag values ​​set by the setting unit 10. The acquiring unit 13 can acquire the flag values ​​indicating transitions of the execution states of multiple threads upon receiving a process termination signal from the core network device 2.

[0053] The classification unit 14 provides the learned classifier with the flag values ​​indicating the transition of the execution states of the multiple threads acquired by the acquisition unit 13 as unknown inputs, performs calculations on the learned classifier, and outputs a classification class related to the cause of the operation abnormality of the process. In the example of Fig. 4, the classification unit 14 can output, as the classification class, any one of the preset values ​​of "normal operation", "heap error", and "deadlock".

[0054] The identification unit 15 identifies a location where an operation abnormality has occurred in the program, based on the flag values ​​indicating the transition of the execution states of the multiple threads acquired by the acquisition unit 13. The flag values ​​indicating the transition of the execution states of the multiple threads acquired by the acquisition unit 13 are input values ​​to each input node of the trained classifier.

[0055] For example, when the flag value indicating the transition of the execution states of multiple threads acquired by the acquisition unit 13 includes "0", the identification unit 15 identifies the location in the source code of the first instruction that could not be executed by the first thread in the first execution state (e.g., state S1) that did not end normally, as indicated by the flag value "0", as the location where the operational abnormality occurred in the program.

[0056] Specifically, Figure 2 shows that when the classification result is heap error E1, the flag value of thread state S1 is "0," meaning that it is not possible to transition from state S1 to state S2. In this case, it is possible to identify that heap error E1 occurred at the location in the source code that corresponds to the step of the execution state indicated by thread state S1. In other words, the location in the source code where the flag for thread state S1 is set can be identified as the location where heap error E1 occurred.

[0057] The presentation unit 16 presents the cause of the operation abnormality of the program based on the classification class output by the classification unit 14. The presentation unit 16 also presents the location of the operation abnormality in the source code identified by the identification unit 15. The presentation unit 16 can transmit the classification result and the identification result to an external server or the like via the network NW, for example. Alternatively, the presentation unit 16 may send the classification result and the identification result to the core network device 2.

[0058] [Hardware configuration of information processing device] Next, an example of a hardware configuration for realizing the information processing device 1 having the above-mentioned functions will be described with reference to FIG.

[0059] As shown in FIG. 2, the information processing device 1 can be realized, for example, by a computer having a processor 102, a main memory device 103, a communication interface 104, an auxiliary memory device 105, and an input / output (I / O) 106 connected via a bus 101, and a program that controls these hardware resources.

[0060] The main memory device 103 prestores programs for the processor 102 to perform various controls and calculations. The processor 102 and the main memory device 103 realize the various functions of the information processing device 1, such as the setting unit 10, the learning unit 11, the classification unit 14, and the identification unit 15 shown in FIG.

[0061] The communication interface 104 is an interface circuit for connecting the information processing device 1 to various external electronic devices via a network.

[0062] The auxiliary storage device 105 is composed of a readable / writable storage medium and a drive for reading and writing various information such as programs and data from and to the storage medium. The auxiliary storage device 105 can use a semiconductor memory such as a hard disk or a flash memory as the storage medium.

[0063] The auxiliary storage device 105 has a program storage area for storing an information processing program executed by the information processing device 1. The auxiliary storage device 105 also has an area for storing a learning program for learning a classifier. The auxiliary storage device 105 realizes the storage unit 12 described in FIG. 1. Furthermore, the auxiliary storage device 105 may have, for example, a backup area for backing up the above-mentioned data and programs.

[0064] The input / output I / O 106 is an input / output device that inputs signals from an external device and outputs signals to an external device.

[0065] [Operation of information processing device] Next, the operation of the information processing device 1 having the above-mentioned configuration will be described with reference to the flowcharts of Fig. 5 and Fig. 6. Fig. 5 is a flowchart showing a learning process by the information processing device 1. Fig. 6 is a flowchart showing a classification process and a specification process by the information processing device 1 using a trained classifier.

[0066] 5, the setting unit 10 sets flags for recording the transition of the execution state of multiple threads in the process of a program executed in the core network device 2 (step S1). Specifically, the setting unit 10 can set a flag for recording the transition of the execution state of a thread for each of one or more predetermined instructions in the source code of a program for which the cause of an operational abnormality is to be inferred in advance. The setting unit 10 can define the transition of the execution state of a thread for each instruction executed by each thread according to, for example, the execution order between the threads.

[0067] Next, the learning unit 11 uses as training data data in which a preset cause of an operational abnormality is assigned as a correct answer label to flag values ​​indicating the transition of the execution states of multiple threads in a process, learns the relationship between the flag values ​​indicating the transition of the execution states of multiple threads and the causes of the operational abnormality, and constructs a learned classifier (step S2).

[0068] For example, the learning unit 11 can use training data in which the causes of operational abnormalities that can occur in both single-threaded and multi-threaded systems, such as "normal operation," "heap error," and "deadlock," are assigned as correct answer labels to flag values ​​indicating the transition of the execution states of multiple threads.

[0069] Next, the trained classifier constructed by the training unit 11, that is, the trained parameters, are stored in the storage unit 12 (step S3).

[0070] Next, the classification process and the identification process will be described with reference to the flowchart shown in Fig. 6. First, the acquisition unit 13 acquires flag values ​​indicating the transition of the execution states of multiple threads in a process (step S10). Upon receiving a forced termination signal from the core network device 2, the acquisition unit 13 can acquire flag values ​​indicating the transition of the execution states of multiple threads from the core network device 2 via the network NW.

[0071] Next, the classification unit 14 loads the learned parameters from the storage unit 12 (step S11). Then, the classification unit 14 provides the flag values ​​indicating the transition of the execution states of the multiple threads acquired by the acquisition unit 13 in step S10 as unknown inputs to the trained classifier, performs calculations on the trained classifier, and outputs a classification class regarding the cause of the abnormal operation of the process (step S12). For example, the classification unit 14 outputs values ​​indicating "normal operation", "heap error", "deadlock", etc. as classification classes.

[0072] Next, the presentation unit 16 presents the classification class output in step S12 (step S13). For example, the presentation unit 16 can send the inferred cause of the operational abnormality to an external server or the like via the network NW.

[0073] Next, the identification unit 15 identifies a location where an operation abnormality occurs in the program based on flag values ​​indicating transitions of the execution states of the multiple threads, which were acquired in step S10 and given as input values ​​to the trained classifier (step S14). t If the flag value "0" indicating that the transition to the next execution state cannot be made is included among the flag values ​​of the states S1 to S t The position of the instruction corresponding to the state with the smallest index number can be identified as the location where the operational anomaly has occurred.

[0074] For example, consider the case where the flag value of the thread state S1 is "1", the flag value of the state S2 is "0", and the flag values ​​of states S3 and after are all "0", and the classification class obtained as a classification result is "heap error". In this case, it can be seen that a heap error occurred during the execution of an instruction corresponding to the thread state S2, and therefore it was not possible to transition to the operation of state S3 and after. Therefore, the identification unit 15 identifies the location in the source code where the instruction corresponding to the thread state S2 is written as the location where the heap error occurred.

[0075] Next, the presentation unit 16 presents information indicating the location of the operational abnormality identified in step S14 (step S15). For example, the presentation unit 16 can send the information indicating the location of the operational abnormality to an external server or the like via the network NW.

[0076] As described above, according to the information processing device 1 of the present embodiment, flag values ​​indicating transitions of execution states of multiple threads are provided as unknown inputs to a trained classifier, and the trained classifier is operated to output a classification class relating to the cause of an abnormality in program operation. Therefore, when an abnormality in program operation occurs, the cause of the abnormality can be easily identified.

[0077] In addition, according to the information processing device 1 of this embodiment, by referring to flag values ​​indicating the transition of the execution states of multiple threads, it is possible to easily and quickly identify the location in the source code of a program where an operational abnormality has occurred.

[0078] In the above embodiment, a case where a neural network is used as a classifier has been described. However, in addition to the above-mentioned neural network model, a classifier such as logistic regression can be used as the classifier. When logistic regression is used, a probability value for predicting each classification class can be obtained as many as the number of classification classes. In addition, SVM, Naive Bayes, Random Forest, Decision Tree, etc., and deep learning in which a neural network is multi-layered may also be used as a classifier.

[0079] In the above-described embodiment, a case has been described in which both the functional units, the learning unit 11 that performs the learning process and the classification unit 14 that performs the classification process, are mounted on the information processing device 1. However, the learning unit 11 and the classification unit 14 may be provided as the same hardware configuration, or the learning process and the classification process may be distributed by a plurality of servers or the like on the network NW.

[0080] The above describes embodiments of the information processing device and information processing method of the present invention. However, the present invention is not limited to the described embodiments, and various modifications that may be envisioned by a person skilled in the art can be made within the scope of the invention described in the claims. [Explanation of symbols]

[0081] 1...information processing device, 2...core network device, 10...setting unit, 11...learning unit, 12...memory unit, 13...acquisition unit, 14...classification unit, 15...identification unit, 16...presentation unit, 101...bus, 102...processor, 103...main memory device, 104...communication interface, 105...auxiliary memory device, 106...input / output I / O, NW...network.

Claims

1. A setting unit configured to set variables for recording transitions of execution states of multiple threads in a process, which is an execution unit of a program; an acquisition unit configured to acquire values ​​of variables indicating transitions of the execution states of the plurality of threads, the values ​​being recorded in response to execution of the program; a classification unit configured to provide the value of the variable indicating the transition of the execution state of the plurality of threads acquired by the acquisition unit as an unknown input to a trained classifier, perform an operation of the trained classifier, and output a classification class regarding a cause of the operation abnormality of the program; a presentation unit configured to present a cause of the operation abnormality of the program based on the classification class output by the classification unit; An information processing device comprising:

2. 2. The information processing device according to claim 1, The method further includes the steps of: providing a program that executes a program that executes a program on the basis of a value of the variable that indicates a transition of the execution state of the plurality of threads, the value being acquired by the acquisition unit; The presentation unit is configured to present information indicating the location of the identified operational abnormality.

23. An information processing apparatus comprising:

3. 3. The information processing device according to claim 2, The present invention further includes a learning unit configured to learn a relationship between the values ​​of the variables indicating the transition of the execution states of the plurality of threads and the causes of the operation abnormality by using teacher data in which a cause of the operation abnormality set in advance is assigned as a correct answer label to the values ​​of the variables indicating the transition of the execution states of the plurality of threads, and to construct the trained classifier.

23. An information processing apparatus comprising:

4. 4. The information processing device according to claim 3, The value of the variable indicating the transition of the execution state of the plurality of threads includes either a first value indicating that a first thread of the plurality of threads has transitioned to a second execution state in which a second thread executes a second instruction after the first thread executes a first instruction in a first execution state or in parallel with the first thread, and a second value indicating that the first execution state of the first thread did not end normally and did not transition to the second execution state because the first thread cannot execute the first instruction.

23. An information processing apparatus comprising:

5. 5. The information processing device according to claim 4, When the value of the variable indicating the transition of the execution state of the plurality of threads acquired by the acquisition unit includes the second value, the identification unit identifies that the first instruction that could not be executed by the first thread in the first execution state that did not end normally, which is indicated by the second value, corresponds to the occurrence location of the operation abnormality in the program.

23. An information processing apparatus comprising:

6. A first step of setting variables for recording transitions of execution states of multiple threads in a process, which is an execution unit of a program; a second step of acquiring values ​​of variables indicating transitions of the execution states of the plurality of threads, the values ​​being recorded in response to execution of the program; a third step of providing the values ​​of the variables indicating the transitions of the execution states of the plurality of threads acquired in the second step as unknown inputs to a trained classifier, performing an operation of the trained classifier, and outputting a classification class regarding a cause of the operation abnormality of the program; a fourth step of presenting a cause of the abnormal operation of the program based on the classification class output in the third step; An information processing method comprising:

7. 7. The information processing method according to claim 6, and a fifth step of identifying a location of the operation abnormality in the program based on the values ​​of the variables indicating the transition of the execution states of the plurality of threads acquired in the second step, The fourth step presents information indicating the location of the identified operational abnormality.

23. An information processing method comprising:

8. 8. The information processing method according to claim 7, Further, a sixth step of learning a relationship between the values ​​of the variables indicating the transition of the execution states of the plurality of threads and the causes of the operation abnormality by using training data in which a cause of the operation abnormality set in advance is assigned as a correct answer label to the values ​​of the variables indicating the transition of the execution states of the plurality of threads, and constructing the trained classifier.

23. An information processing method comprising:

9. 9. The information processing method according to claim 8, The value of the variable indicating the transition of the execution state of the plurality of threads includes either a first value indicating that a first thread of the plurality of threads has transitioned to a second execution state in which a second thread executes a second instruction after the first thread executes a first instruction in a first execution state or in parallel with the first thread, and a second value indicating that the first execution state of the first thread did not end normally and did not transition to the second execution state because the first thread cannot execute the first instruction.

23. An information processing method comprising:

10. 10. The information processing method according to claim 9, The fifth step is, when the value of the variable indicating the transition of the execution state of the plurality of threads obtained in the second step includes the second value, determining that the first instruction that could not be executed by the first thread in the first execution state that was not normally terminated and that is indicated by the second value corresponds to the occurrence location of the operation abnormality in the program.

23. An information processing method comprising:

Citation Information

Patent Citations

  • Multivariate detection of abnormal conditions in process plants

    JP2010522905A

  • Program verification device and method and program

    JP2011232814A

  • Processing state monitoring device

    JP2012138044A

  • Process dumping method, device and program

    JP2005301570A

Cited By

  • Abnormality management device and abnormality management method

    JP7813952B1

  • Anomaly detection device and anomaly detection method

    JP7833092B1