WO2021064887A1

WO2021064887A1 - Impact analysis device, impact analysis method, and impact analysis program

Info

Publication number: WO2021064887A1
Application number: PCT/JP2019/038885
Authority: WO
Inventors: 僚柏木
Original assignee: 三菱電機株式会社
Priority date: 2019-10-02
Filing date: 2019-10-02
Publication date: 2021-04-08
Also published as: JPWO2021064887A1; JP7258171B2

Abstract

The present invention aims to provide a technology capable of displaying the impact of differences in conditions when software is executed. This impact analysis device comprises: an execution trace accumulation unit that accumulates a plurality of execution traces collected under each of a plurality of different conditions; a representative execution trace calculation unit that calculates a representative execution trace that includes a plurality of events and the average and variation in time of each of the plurality of events, calculating same on the basis of the plurality of execution traces collected by the execution trace accumulation unit; and a representative execution trace accumulation unit that accumulates the representative execution traces calculated by the representative execution trace calculation unit.

Description

Impact analyzer, impact analysis method, and impact analysis program

The present invention relates to a software impact analyzer, impact analysis method, and impact analysis program.

Various technologies have been proposed for impact analyzers that analyze the impact of changes in source code on software. For example, Patent Document 1 proposes a technique for extracting an influence range from the dependency of a function or a variable. For example, Patent Document 2 proposes a technique of measuring an execution trace obtained by inserting a tag into a source code and collecting the execution time of a function.

For example, Patent Document 3 proposes a technique for holding a function name in a memory, and Patent Document 4, for example, proposes a technique for comparing function call histories. For example, Patent Document 5 proposes a sampling trace that collects a large number of machine instruction sequences at appropriate intervals. For example, Patent Document 6 proposes a technique for evaluating the degree of deviation of a job history including a job type and its execution time.

Japanese Unexamined Patent Publication No. 2015-11661 Japanese Unexamined Patent Publication No. 11-316696 Japanese Unexamined Patent Publication No. 2001-282576 Japanese Unexamined Patent Publication No. 2006-053788 Japanese Unexamined Patent Publication No. 2004-078338 International Publication No. 2018/146714

However, the conventional impact analyzer uses the dependency obtained by static analysis without considering the execution of software. Therefore, there is a problem that even if the influence of the change of the processing content can be confirmed, the influence of the difference in the conditions at the time of software execution cannot be confirmed.

Therefore, the present invention has been made in view of the above problems, and an object of the present invention is to provide a technique capable of displaying the influence of a difference in conditions during software execution.

The impact analyzer according to the present invention collects a plurality of execution traces including a plurality of events including the entrance event and the exit event of the function in the software and the times of the plurality of events under different plurality of conditions. The plurality of events are based on the trace collection unit, the execution trace storage unit that stores the plurality of execution traces collected by the execution trace collection unit, and the plurality of execution traces accumulated in the execution trace storage unit. A representative execution trace calculation unit that calculates a representative execution trace including the average and variation of the time of each of the plurality of events, and a representative execution that accumulates the representative execution trace calculated by the representative execution trace calculation unit. It includes a trace storage unit and an output unit that outputs the representative execution trace accumulated in the representative execution trace storage unit.

According to the present invention, a representative execution trace including the average and variation of each time of a plurality of events is calculated and displayed based on a plurality of execution traces collected under a plurality of different conditions. With such a configuration, it is possible to display the influence of the difference in the conditions at the time of software execution.

The object, features, aspects and advantages of the present invention will be made clearer by the following detailed description and accompanying drawings.

It is a block diagram which shows the structure of the influence analysis apparatus which concerns on Embodiment 1. FIG. It is a figure which shows the example of the execution trace which concerns on Embodiment 1. FIG. It is a figure which shows the example of the representative execution trace which concerns on Embodiment 1. FIG. It is a figure which shows the display example of the influence analysis apparatus which concerns on Embodiment 1. FIG. It is a flowchart which shows the operation of the influence analysis apparatus which concerns on Embodiment 1. FIG. It is a flowchart which shows the operation of the influence analysis apparatus which concerns on Embodiment 1. FIG. It is a block diagram which shows the structure of the influence analysis apparatus which concerns on Embodiment 2. FIG. It is a flowchart which shows the operation of the influence analysis apparatus which concerns on Embodiment 2. It is a flowchart which shows the operation of the influence analysis apparatus which concerns on Embodiment 2. It is a figure for demonstrating the operation example of the influence analysis apparatus which concerns on Embodiment 2. FIG. It is a figure for demonstrating the operation example of the influence analysis apparatus which concerns on Embodiment 2. FIG. It is a figure which shows the display example of the influence analysis apparatus which concerns on Embodiment 2. FIG. It is a block diagram which shows the structure of the influence analysis apparatus which concerns on Embodiment 3. It is a figure which shows the example of the start event and the end event which concerns on Embodiment 3. It is a figure for demonstrating the operation example of the influence analysis apparatus which concerns on Embodiment 3. FIG. It is a flowchart which shows the operation of the influence analysis apparatus which concerns on Embodiment 3. It is a block diagram which shows the structure of the influence analysis apparatus which concerns on Embodiment 4. FIG. It is a figure for demonstrating an example of the mapping rule which concerns on Embodiment 4. FIG. It is a figure for demonstrating the operation example of the influence analysis apparatus which concerns on Embodiment 4. FIG. It is a flowchart which shows the operation of the influence analysis apparatus which concerns on Embodiment 4. FIG.

<Embodiment 1>
FIG. 1 is a block diagram showing a configuration of an impact analysis device according to a first embodiment of the present invention. This impact analyzer is a device that analyzes the impact on software performance due to changes in the source code or the like. Hereinafter, the effect will be described as "change effect", the software before change will be described as "software before change", and the software after change will be described as "software after change".

The impact analysis device shown in FIG. 1 includes an execution trace collection unit 1, an execution trace storage unit 2, a representative execution trace calculation unit 3, a representative execution trace storage unit 4, and a display unit 5. The input of the execution trace storage unit 2 is connected to the output of the execution trace collection unit 1. The input of the representative execution trace calculation unit 3 is connected to the output of the execution trace storage unit 2. The input of the representative execution trace storage unit 4 is connected to the output of the representative execution trace calculation unit 3. The input of the display unit 5 is connected to the output of the representative execution trace storage unit 4.

The execution trace collection unit 1 is a tool that collects a plurality of execution traces under different plurality of conditions. Although details will be described later, an execution trace is an event sequence including a plurality of events including entrance events and exit events of one or more functions in software and times of the plurality of events. The entrance event and the exit event can also be referred to as an entrance passage event and an exit passage event, respectively. The time is the measurement time from a certain point in time to a certain point in time. For example, the time of an event is a measurement time from a reference time point to a time point in which the event occurs.

The execution trace collection unit 1 may be, for example, a tool of a method of inserting additional code into the source code, that is, a tool of a method of instrumenting. Alternatively, the execution trace collection unit 1 may be, for example, a tool having a debugger for collecting instruction traces obtained from a macro cell of a microprocessor and having a log generation function in which a logging function is manually embedded in the source code. Alternatively, the execution trace collection unit 1 may be, for example, a sampling trace tool capable of acquiring a history at a shorter sampling interval than the event to be monitored.

FIG. 2 is a diagram showing an example of execution trace 6. The execution trace 6 includes a plurality of events including the entrance event and the exit event of the function, and the time when the plurality of events occur. The time may be, for example, the elapsed time from the start of the trace, or the difference time between the event and the event immediately before the event. In the example of FIG. 2, the elapsed time of the funcA function, the funcB function called from the funcA function, and the funcC function called from the funcA function is shown as the time of the entrance event and the exit event, and the difference time is shown in parentheses. ing.

Funca @ enter indicates the entrance of the funcaA function, and funca @ leave indicates the exit of the funcaA function. Similarly, funcB @ enter indicates the entrance of the funcB function, funcB @ leave indicates the exit of the funcB function, func @ enter indicates the entrance of the funcC function, and funcC @ leave indicates the exit of the funcC function.

In the example of FIG. 2, the funca function first starts processing at 0 ns (0 ns), and the funca function calls the func B function at 70 ns (70 ns). Then, the funcB function finishes processing at 280ns (210ns), the funcaA function processes again, and the funcA function calls the funcC function at 330ns (50ns). Then, the funcC function finishes the process at 500 ns (170 ns), and the funcA function performs the process again and finishes the process at 600 ns (100 ns).

The plurality of execution traces 6 are collected by operating the software to be processed under a plurality of conditions, and are substantially statistically processed by the representative execution trace calculation unit 3 as described later. The plurality of conditions differ from each other, for example, in continuous operation time, the number of other processes and other tasks, and at least one of the startup environments of other processes and other tasks.

The execution trace storage unit 2 stores a plurality of execution traces 6 collected by the execution trace collection unit 1. A plurality of execution traces 6 may be stored in the execution trace storage unit 2 independently of each other, or may be continuously connected or combined in a state in which they can be distinguished from each other and stored in the execution trace storage unit 2. Good.

The representative execution trace calculation unit 3 calculates the representative execution trace by performing statistical processing based on the plurality of execution traces 6 accumulated in the execution trace storage unit 2.

FIG. 3 is a diagram showing an example of the representative execution trace 7. The representative execution trace 7 is an event series including a plurality of events and the average and variation of the time of each of the plurality of events. The plurality of events of the representative execution trace 7 are the same as the plurality of events of the execution trace 6 used for calculating the representative execution trace 7. The time may be, for example, an elapsed time or a difference time, as described above. The variability may be, for example, a variance or a standard deviation. In FIG. 3, the average difference time and the variance difference time are shown as examples of the average and variation of the time of the representative execution trace 7. Hereinafter, the case where the time is the difference time and the variation is the variance will be mainly described, but the same applies to the case where the time is the elapsed time and the case where the variation is the standard deviation.

In the example of FIG. 3, the funca function first starts processing with 0 ns and a variance of 0 ns, and the funca function calls the func B function with an average of 70 ns and a variance of 10 ns. Then, the funcB function finishes processing with an average of 200 ns and a variance of 20 ns, the funca function performs processing again, and the funca function calls the funcC function with an average of 30 ns and a variance of 15 ns. Then, the funcC function finishes the process with an average of 170 ns and a variance of 13 ns, and the funca function performs the process again and finishes the process with 110 ns and a variance of 10 ns.

The representative execution trace storage unit 4 stores the representative execution trace 7 calculated by the representative execution trace calculation unit 3.

The display unit 5 selects the representative execution trace 7 accumulated in the representative execution trace storage unit 4, and displays the selected representative execution trace 7 in the manner of the timing diagram. Hereinafter, the output unit will be described as being a display unit 5 that displays the representative execution trace 7, but the present invention is not limited to this, and may be, for example, a communication unit that transmits the representative execution trace 7. Since the representative execution trace 7 has an average and variation statistically processed for each event, the event transition, which is a transition of the function, is displayed with an event occurrence interval corresponding to the average and variation.

FIG. 4 is a display example of the representative execution trace 7 by the display unit 5. FIG. 4 shows an operation in which the funcA function calls the funcB function and the funcC function in this order. The display unit 5 displays the transition of the function based on the representative execution trace 7 which is the result of the statistical processing. For example, the display unit 5 displays the transition of the function with an event occurrence interval of “average difference time ± 2 × variance difference time”. In the example of FIG. 4, the display unit 5 represents the function transition from the funcaA function or the like to the funcB function or the like with two arrows. The arrow on the left indicates "average difference time-2 x variance difference time", and the arrow on the right indicates "average difference time + 2 x variance difference time".

The execution trace collection unit 1, the execution trace storage unit 2, the representative execution trace calculation unit 3, and the representative execution trace storage unit 4 process each of the pre-change software and the post-change software. Then, the display unit 5 displays the representative execution trace 7 of the software before the change and the representative execution trace 7 of the software after the change, so that the user can confirm the influence of the change.

<Operation>
FIG. 5 is a flowchart showing the operation of the impact analysis device according to the first embodiment to visualize the effect of the change. Hereinafter, this operation will be described with reference to FIG.

In steps S1 to S3, processing is performed on the software before the change. First, in step S1, the execution trace collecting unit 1 acquires the execution trace 6 of the software before the change under certain conditions and stores it in the execution trace accumulating unit 2. After the accumulation, the process proceeds to step S2.

In step S2, the execution trace collecting unit 1 determines whether or not the number of times the execution trace 6 has been acquired has reached the specified number of times specified in advance. If it is determined that the value has not been reached, the process returns to step S1, and in step S1, the execution trace collecting unit 1 acquires the execution trace 6 under different conditions and accumulates the execution trace 6 in the execution trace storage unit 2. If it is determined that the value has been reached, the process proceeds to step S3. As described above, a plurality of execution traces 6 collected under a plurality of different conditions are accumulated in the execution trace storage unit 2.

In step S3, the representative execution trace calculation unit 3 calculates the representative execution trace 7 of the pre-change software based on the plurality of execution traces 6 of the pre-change software accumulated in the execution trace storage unit 2, and the representative execution trace is calculated. Accumulate in the storage unit 4. After the accumulation, the process proceeds to step S4.

In steps S4 to S6, the same processing as in steps S1 to S3 is performed for the changed software. First, in step S4, the execution trace collection unit 1 acquires the execution trace 6 of the modified software under certain conditions and accumulates it in the execution trace storage unit 2. After the accumulation, the process proceeds to step S5.

In step S5, the execution trace collecting unit 1 determines whether or not the number of times the execution trace 6 has been acquired has reached the specified number of times specified in advance. If it is determined that the value has not been reached, the process returns to step S4, and in step S4, the execution trace collection unit 1 acquires the execution trace 6 under different conditions and accumulates it in the execution trace storage unit 2. If it is determined that the value has been reached, the process proceeds to step S6.

In step S6, the representative execution trace calculation unit 3 calculates the representative execution trace 7 of the modified software based on the plurality of execution traces 6 of the modified software accumulated in the execution trace storage unit 2, and the representative execution trace is calculated. Accumulate in the storage unit 4. After the accumulation, the process proceeds to step S7.

In step S7, the display unit 5 displays the representative execution trace 7 accumulated in the representative execution trace storage unit 4. The display unit 5 selectively or simultaneously displays, for example, the representative execution trace 7 of the software before the change and the representative execution trace 7 of the software after the change. This allows the user to see the effect of changes made between the pre-change software and the post-change software on the performance of the software. After the display, the operation of FIG. 5 ends.

FIG. 6 is a flowchart showing the calculation operation of the representative execution trace 7 in steps S3 and S6. Hereinafter, this operation will be described with reference to FIG.

First, in step S11, the representative execution trace calculation unit 3 empties the event series list. After that, the process proceeds to step S12.

In step S12, the representative execution trace calculation unit 3 acquires the execution trace 6 from the execution trace storage unit 2. After the acquisition, the process proceeds to step S13.

In step S13, the representative execution trace calculation unit 3 confirms whether or not the event series of the execution trace 6 acquired in step S12 exists in the event series list. If it does not exist, the process proceeds to step S14, and if it exists, the process proceeds to step S15.

In step S14, the representative execution trace calculation unit 3 adds the event sequence that did not exist to the event sequence list. After the addition, the process proceeds to step S15.

In step S15, the representative execution trace calculation unit 3 calculates the difference time from the immediately preceding event for the event of the event series in the event series list. After the calculation, the process proceeds to step S16.

In step S16, the representative execution trace calculation unit 3 determines whether or not there is an unprocessed execution trace 6 among the plurality of execution traces 6 accumulated in the execution trace storage unit 2. If it is determined that it exists, the process returns to step S12, and in step S12, the representative execution trace calculation unit 3 acquires the unprocessed execution trace 6 from the execution trace storage unit 2. If it is determined that it does not exist, the process proceeds to step S17. By the above operation, a maximum of the same number of events of the same type as a plurality of execution traces 6 are obtained, and the difference time is calculated for each of them. For example, when N (N is a natural number of 1 or more) of execution traces 6 as shown in FIG. 2 are collected, N events of one type called funca @ enter are obtained, and the difference time is calculated for each of them. To. That is, the difference time of one type of N (N is a natural number of 1 or more) events is calculated.

In step S17, the representative execution trace calculation unit 3 acquires one or more events of one type from the event series list. After the acquisition, the process proceeds to step S18.

In step S18, the representative execution trace calculation unit 3 calculates the mean value and the variance value from the difference time of one or more events of one type. The representative execution trace calculation unit 3 generates a representative execution trace 7 including one type of event and the average difference time and the distributed difference time of the one type of event. After that, the process proceeds to step S19.

In step S19, the representative execution trace calculation unit 3 determines whether or not there is an unprocessed event in the event series list. If it is determined that the event exists, the process returns to step S17, and in step S17, the representative execution trace calculation unit 3 acquires an unprocessed event from the event sequence list. If it is determined that it does not exist, the operation of FIG. 6 ends.

<Summary of Embodiment 1>
According to the impact analyzer according to the first embodiment as described above, the average and variation of the time of each of the plurality of events are included based on the plurality of execution traces 6 collected under different plurality of conditions. The representative execution trace 7 is calculated and displayed. According to such a configuration, it is possible to display for confirming the influence of the difference in the conditions at the time of software execution on the performance of the software, for example, the influence of other tasks, other processes, interruptions, and the like. Therefore, the user can confirm the performance impact in consideration of the impact.

<Embodiment 2>
According to the first embodiment, the user compares the average time, which is the average of the times of the representative execution trace 7, with the time of the execution trace 6, and identifies the execution trace 6 having a certain difference from the representative execution trace 7. Therefore, the cause of the difference can be investigated. However, when the event sequence of the representative execution trace 7 is long or when the number of the plurality of execution traces 6 is large, the work becomes difficult. In view of this, as will be clarified in the following description, the impact analyzer according to the second embodiment of the present invention automatically extracts the execution trace 6 having a certain difference from the representative execution trace 7. Is possible.

FIG. 7 is a block diagram showing the configuration of the impact analysis device according to the second embodiment. Hereinafter, among the components according to the second embodiment, the components that are the same as or similar to the above-mentioned components are designated by the same or similar reference numerals, and different components will be mainly described.

The configuration of the impact analysis device shown in FIG. 7 is the same as the configuration in which the execution trace outlier extraction unit 8 which is the extraction unit is added to the configuration of the impact analysis device according to the first embodiment (FIG. 1). The input of the execution trace storage unit 2 is connected to the output of the execution trace collection unit 1. The input of the representative execution trace calculation unit 3 is connected to the output of the execution trace storage unit 2. The input of the representative execution trace calculation unit 3 is connected to the output of the execution trace storage unit 2. The input of the representative execution trace storage unit 4 is connected to the output of the representative execution trace calculation unit 3. The input of the execution trace outlier extraction unit 8 is connected to the output of the execution trace storage unit 2 and the output of the representative execution trace storage unit 4. The input of the display unit 5 is connected to the output of the representative execution trace storage unit 4 and the output of the execution trace outlier extraction unit 8.

In the second embodiment, the representative execution trace storage unit 4 stores a plurality of representative execution traces 7 (for example, traces similar to the traces A to D in FIG. 10) in which combinations of a plurality of events are different from each other. The difference in the combination of the plurality of events referred to here includes that at least one of the types and the number of events is different.

The execution trace outlier extraction unit 8 is the difference between the combination of the plurality of execution traces 6 stored in the execution trace storage unit 2 and the combination of the plurality of events among the plurality of representative execution traces 7 stored in the representative execution trace storage unit 4. Gets the smallest representative execution trace 7. Hereinafter, the representative execution trace 7 acquired by the execution trace outlier extraction unit 8 may be referred to as the “representative execution trace 7 of the nearest neighbor series”.

The execution trace outlier extraction unit 8 defines an allowable range based on the average time, which is the average included in the representative execution trace 7 of the nearest neighbor series, and is a plurality of execution traces 6 accumulated in the execution trace storage unit 2. Among them, the execution trace 6 whose time is out of the permissible range is extracted. Hereinafter, the execution trace 6 extracted by the execution trace outlier extraction unit 8 may be referred to as “outlier execution trace 6”.

The display unit 5 superimposes and displays the execution trace 6 of the outliers extracted from the execution trace outlier extraction unit 8 on the representative execution trace 7 of the nearest neighbor series used for the extraction. As a result, the user can confirm the representative execution trace 7 and the execution trace 6 having a certain difference from the representative execution trace 7.

<Operation>
FIG. 8 is a flowchart showing an operation in which the impact analysis device according to the second embodiment visualizes the effect of change. Hereinafter, this operation will be described with reference to FIG.

In steps S21 to S23, the same processing as the processing of steps S1 to S3 of FIG. 5 described in the first embodiment is performed on the software before change. First, in step S21, the execution trace collecting unit 1 acquires the execution trace 6 of the software before the change under certain conditions and accumulates it in the execution trace accumulating unit 2. After the accumulation, the process proceeds to step S22.

In step S22, the execution trace collecting unit 1 determines whether or not the number of times the execution trace 6 has been acquired has reached the specified number of times specified in advance. If it is determined that the result has not been reached, the process returns to step S21, and in step S21, the execution trace collecting unit 1 acquires the execution trace 6 under different conditions and accumulates it in the execution trace storage unit 2. If it is determined that the value has been reached, the process proceeds to step S23. As a result, a plurality of execution traces 6 of the software before the change are accumulated in the execution trace storage unit 2.

In step S23, the representative execution trace calculation unit 3 calculates the representative execution trace 7 of the pre-change software based on the plurality of execution traces 6 of the pre-change software accumulated in the execution trace storage unit 2, and the representative execution trace is calculated. Accumulate in the storage unit 4.

The above processing is performed on a plurality of execution traces 6 (for example, traces similar to traces A to D in FIG. 10) in which combinations of a plurality of events are different from each other. As a result, with respect to the software before change, a plurality of representative execution traces 7 having different combinations of a plurality of events are accumulated in the representative execution trace storage unit 4. After the accumulation, the process proceeds to step S24.

In step S24 and step S25, the same processing as in step S21 and step S22 is performed for the changed software. As a result, a plurality of execution traces 6 of the modified software are accumulated in the execution trace storage unit 2. After the accumulation, the process proceeds to step S26.

In step S26, the execution trace outlier extraction unit 8 acquires the representative execution trace 7 of the nearest neighbor series of the software before change from the plurality of representative execution traces 7 of the software before change accumulated in the representative execution trace storage unit 4. To do. Then, the execution trace outlier extraction unit 8 compares the representative execution trace 7 of the nearest series of the software before the change with the plurality of execution traces 6 of the software after the change accumulated in the execution trace storage unit 2, and after the change. Extract the outlier execution trace 6 for the software. Then, the display unit 5 superimposes and displays the extracted execution trace 6 of the outliers on the representative execution trace 7 of the nearest neighbor series. As a result, the user can confirm the representative execution trace 7 of the software before the change and the execution trace 6 of the software after the change having a certain difference from the representative execution trace 7.

FIG. 9 is a flowchart showing the acquisition operation of the representative execution trace 7 of the nearest neighbor series and the extraction operation and the display operation of the outlier execution trace 6 in step S26. Hereinafter, these operations will be described with reference to FIG.

First, in step S31, the execution trace outlier extraction unit 8 acquires the representative execution trace 7 from the representative execution trace storage unit 4. After the acquisition, the process proceeds to step S32.

In step S32, the execution trace outlier extraction unit 8 combines a plurality of events between the representative execution trace 7 acquired in step S31 and the plurality of execution traces 6 accumulated in the execution trace storage unit 2. Calculate the difference between. After the calculation, the process proceeds to step S33.

In step S33, the execution trace outlier extraction unit 8 determines whether or not the difference calculated in step S32 is smaller than the difference calculated similarly for the representative execution trace 7 of the nearest neighbor series. If it is determined that it is small, the process proceeds to step S34, and if it is determined that it is not small, the process proceeds to step S35.

In step S34, the execution trace outlier extraction unit 8 updates the representative execution trace 7 of the nearest neighbor series to the representative execution trace 7 acquired in step S31. After the update, the process proceeds to step S35.

In step S35, the execution trace outlier extraction unit 8 determines whether or not an unprocessed representative execution trace 7 exists among the plurality of representative execution traces 7 accumulated in the representative execution trace storage unit 4. If it is determined that it exists, the process returns to step S31, and in step S31, the execution trace outlier extraction unit 8 acquires the unprocessed representative execution trace 7 from the representative execution trace storage unit 4. If it is determined that it does not exist, the process proceeds to step S36.

FIG. 10 is a diagram for explaining an operation example of steps S32 to S35. In the example of FIG. 10, the reference trace is the execution trace 6 accumulated in the execution trace storage unit 2, the trace A is the representative execution trace 7 acquired in the first step S31, and ..., the trace D is This is the representative execution trace 7 acquired in the fourth step S31. Hereinafter, the difference in the combination of a plurality of events will be abbreviated as “event difference”.

In the event series of the reference trace, the funcA function is called, and the funcB function and the funcC function are called in the funcaA function. Since the event series of the trace A is equal to the event series of the reference trace, the difference between the events is "0".

In the event series of trace B, only the funcC function is called in the funcA function. That is, in the event series of trace B, since there is no entrance event and exit event of the funcB function accompanying the call of the funcB function, the difference between the events is "2".

In the event series of trace C, the funcB function and the funcD function are called in the funcA function. That is, in the event series of trace C, the funcD function is called instead of the funcC function, and the entrance event and exit event of the funcC function and the entrance event and exit event of the funcD function are different, so that the difference between the events is "2". ..

In the event series of trace D, the funcB function, the funcC function, and the funcD function are called in the funcA function. Since the entrance event and exit event of the funcD function are increased by the amount that the funcD function is called, the difference between the events is "2".

When the execution trace outlier extraction unit 8 performs the processing of steps S32 to S35 for the above example, the execution trace outlier extraction unit 8 sets the trace A having the smallest event difference from the execution trace 6 among the traces A to D as the nearest neighbor series. Acquired as the representative execution trace 7.

In the above description, the combination of a plurality of events of a plurality of execution traces 6 compared with the representative execution trace 7 has been described as being the same as each other, but they may be different from each other. In such a case, for example, the execution trace outlier extraction unit 8 obtains the difference between the events of each execution trace 6 and the representative execution trace 7, and the value obtained by summing the differences for the plurality of execution traces 6 is the largest. The smaller representative execution trace 7 may be acquired as the representative execution trace 7 of the nearest neighbor series. For example, when there is one first reference trace in FIG. 11 and five second reference traces in FIG. 11 as a plurality of execution traces 6, the difference between the events of trace A, trace B, trace C, and trace D is , 2, 10, 14, and 14, respectively. In such a case, the execution trace outlier extraction unit 8 may acquire the trace A having the smallest event difference from the execution trace 6 among the traces A to D as the representative execution trace 7 of the nearest neighbor series.

The processing after step S36 in FIG. 9 is performed for each of the plurality of execution traces 6 stored in the execution trace storage unit 2.

First, in step S36, the execution trace outlier extraction unit 8 acquires an event from the representative execution trace 7 of the nearest neighbor series. After that, the process proceeds to step S37.

In step S37, the execution trace outlier extraction unit 8 defines an allowable range based on the average time of the events acquired in step S36. For example, when the variance included in the representative execution trace 7 is the variance time, the execution trace outlier extraction unit 8 defines the permissible range as the range of the average time ± (dispersion time × 2). The execution trace outlier extraction unit 8 determines whether or not the time of the event of the execution trace 6 is within the permissible range. If it is determined that the amount is settled, the process proceeds to step S38. If it is determined that it does not fit, the process proceeds to step S39.

In step S38, the execution trace outlier extraction unit 8 determines whether or not an unprocessed event exists in the representative execution trace 7 of the nearest neighbor series. If it is determined that the event exists, the process returns to step S36, and in step S36, the execution trace outlier extraction unit 8 acquires an unprocessed event from the representative execution trace 7 of the nearest neighbor series. If it is determined that it does not exist, the operation of FIG. 9 ends.

In step S39, the execution trace outlier extraction unit 8 determines the execution trace 6 determined in step S37 as the outlier execution trace 6, and the display unit 5 uses the outlier execution trace 6 for extraction. It is superimposed and displayed on the representative execution trace 7 of the nearest neighbor series. After the display, the operation of FIG. 9 ends.

FIG. 12 is a diagram showing a display example of step S39. The dotted line shows the representative execution trace 7 of the nearest neighbor series, and the solid line shows the execution trace 6 of outliers. According to the display as shown in FIG. 12, of the outlier execution traces 6, the timing of transition from the funca function to the funcB function and the timing of transition from the funca function to the funcC function are allowed by the representative execution trace 7. It can be read that it is out of range.

<Summary of Embodiment 2>
According to the impact analyzer according to the second embodiment as described above, the execution trace 6 whose time is out of the permissible range is extracted and displayed from the plurality of execution traces 6 stored in the execution trace storage unit 2. To do. According to such a configuration, the user can confirm the representative execution trace 7 and the execution trace 6 having a certain difference from the representative execution trace 7. Therefore, it can be expected that the investigation of the cause of the difference between the execution trace 6 and the representative execution trace 7 can be facilitated.

In the above description, the execution trace outlier extraction unit 8 extracts the execution trace 6 in which the difference from the average time of the representative execution trace 7 of the nearest neighbor series is equal to or more than a certain value, and the display unit 5 extracts the execution trace 6 Was displayed, but it is not limited to this. For example, the execution trace outlier extraction unit 8 extracts the execution trace 6 in which the difference between the combinations of a plurality of events is a certain number or more with respect to the representative execution trace 7 of the nearest neighbor series, and the display unit 5 extracts the execution trace 6 May be displayed. Even with such a configuration, it can be expected that the investigation of different causes can be facilitated.

<Embodiment 3>
In the first embodiment and the second embodiment, it is assumed that the execution trace 6 collected by the execution trace collecting unit 1 has the same start event and end event. If the start event and end event of the execution trace 6 collected by the execution trace collection unit 1 are not the same, there is a possibility that the execution traces 6 cannot be compared correctly with each other. In view of this, as will be clarified in the following description, the impact analyzer according to the third embodiment of the present invention can align the start and end of the events of the plurality of execution traces 6. ..

FIG. 13 is a block diagram showing the configuration of the impact analysis device according to the third embodiment. Hereinafter, among the components according to the third embodiment, the components that are the same as or similar to the above-mentioned components are designated by the same or similar reference numerals, and different components will be mainly described.

The impact analysis device shown in FIG. 13 is the same as the configuration in which the event restriction unit 10 which is a restriction unit is added to the configuration (FIG. 7) of the impact analysis device according to the second embodiment.

The input of the event restriction unit 10 is connected to the output of the execution trace collection unit 1. The input of the execution trace storage unit 2 is connected to the output of the event restriction unit 10. The input of the representative execution trace calculation unit 3 is connected to the output of the execution trace storage unit 2. The input of the representative execution trace storage unit 4 is connected to the output of the representative execution trace calculation unit 3. The input of the execution trace outlier extraction unit 8 is connected to the output of the execution trace storage unit 2 and the output of the representative execution trace storage unit 4. The input of the display unit 5 is connected to the output of the representative execution trace storage unit 4 and the output of the execution trace outlier extraction unit 8.

The event limiting unit 10 uses the execution trace 6 accumulated in the execution trace accumulating unit 2 as a predetermined start event and a pre-designated end event among the execution traces 6 collected by the execution trace collecting unit 1. It is limited to the execution trace 6 which has. When the execution trace 6 collected by the execution trace collection unit 1 has one or more event sequences from the same event as the start event to the same event as the end event, the event restriction unit 10 sets the event sequence as the execution trace 6. , Stored in the execution trace storage unit 2. If the execution trace 6 collected by the execution trace collection unit 1 does not have any event sequence from the same event as the start event to the same event as the end event, the event restriction unit 10 executes the execution trace 6 and the execution trace storage unit 2 Do not store in.

FIG. 14 is a diagram showing an example of a start event and an end event specified by the event restriction unit 10. In FIG. 14, the entrance event of the funca function is specified as the start event, and the exit event of the funca function is specified as the end event.

FIG. 15 is a diagram for explaining an example of event restriction using the start event and the exit event of FIG. In the input event series, that is, the execution trace 6 collected by the execution trace collecting unit 1, the funcaE function calls the funcaA function, and then the funcaA function calls the funcB function and the funcC function in this order. As shown in FIG. 14, when the entrance event and the exit event of the funca function are specified as the start event and the end event, the event restriction unit 10 limits only the event sequence in which the funca function calls the func B function and the func C function in this order. , Output as an output event series. As a result, in the examples of FIGS. 14 and 15, an event sequence that does not include the entrance event and the exit event of the funcE function is accumulated in the execution trace storage unit 2 as the execution trace 6.

<Operation>
FIG. 16 is a flowchart showing the operation of the event limiting unit 10. Hereinafter, this operation will be described with reference to FIG.

First, in step S41, the event restriction unit 10 sets the output flag to invalid. After that, the process proceeds to step S42.

In step S42, the event limiting unit 10 acquires an event from the execution trace 6 collected by the execution trace collecting unit 1. After that, the process proceeds to step S43.

In step S43, the event restriction unit 10 determines whether or not the output flag is effectively set. If it is determined that the output flag is set to be valid, the process proceeds to step S47, and if it is determined that the output flag is set to be invalid, the process proceeds to step S44.

In step S44, the event restriction unit 10 determines whether or not the event acquired in step S42 matches the start event. If it is determined that they match, the process proceeds to step S45, and if it is determined that they do not match, the process proceeds to step S50.

In step S45, the event restriction unit 10 effectively sets the output flag. After that, the process proceeds to step S46.

In step S46, the event restriction unit 10 creates a new execution trace 6. After that, the process proceeds to step S47.

In step S47, the event restriction unit 10 includes the event acquired in step S42 in the created execution trace 6. After that, the process proceeds to step S48.

In step S48, the event restriction unit 10 determines whether or not the event acquired in step S42 matches the end event. If it is determined that they match, the process proceeds to step S49, and if it is determined that they do not match, the process proceeds to step S50.

In step S49, the event restriction unit 10 sets the output flag to invalid. After that, the process proceeds to step S50.

In step S50, the event limiting unit 10 determines whether or not there is an unprocessed event in the execution trace 6 collected by the execution trace collecting unit 1. If it is determined that the event exists, the process returns to step S42, and in step S42, the event limiting unit 10 acquires an unprocessed event from the execution trace 6 collected by the execution trace collecting unit 1. If it is determined that it does not exist, the operation of FIG. 16 ends.

<Summary of Embodiment 3>
According to the impact analyzer according to the third embodiment as described above, the execution trace 6 accumulated in the execution trace storage unit 2 is an execution trace having a predetermined start event and a predetermined end event. Limit to 6. According to such a configuration, it is possible to suppress the processing of the execution trace 6 and the representative execution trace 7 in which the event sequence is different from the desired event sequence. Therefore, it is possible to improve the processing efficiency of the impact analysis device and the work efficiency of the user.

<Embodiment 4>
In the third embodiment, the execution trace 6 and the representative execution trace 7 are classified only by the event series of the function transition, and the execution trace 6 and the representative execution trace 7 are not classified due to the difference in variables such as the state. In the third embodiment, when the execution trace 6 collected by the execution trace collection unit 1 includes a variable, the types of events are subdivided with respect to the discrete values of the variables. As a result, even if the contents of two events are substantially the same, the events may be treated as different from each other due to the different variables in the events. As a result, the number of execution traces 6 and thus the representative execution traces 7 may increase unnecessarily. In view of this, as will be clarified in the following description, the impact analyzer according to the fourth embodiment of the present invention can suppress the processing of the execution trace 6 and thus the representative execution trace 7 with respect to the variables. ing.

FIG. 17 is a block diagram showing the configuration of the impact analysis device according to the fourth embodiment. Hereinafter, among the components according to the fourth embodiment, the same or similar components as those described above will be designated by the same or similar reference numerals, and different components will be mainly described.

The impact analysis device shown in FIG. 17 is the same as the configuration in which the event conversion unit 12 which is a conversion unit is added to the configuration (FIG. 13) of the impact analysis device according to the third embodiment.

The input of the event conversion unit 12 is connected to the output of the execution trace collection unit 1. The input of the event limiting unit 10 is connected to the output of the event conversion unit 12. The input of the execution trace storage unit 2 is connected to the output of the event restriction unit 10. The input of the representative execution trace calculation unit 3 is connected to the output of the execution trace storage unit 2. The input of the representative execution trace storage unit 4 is connected to the output of the representative execution trace calculation unit 3. The input of the execution trace outlier extraction unit 8 is connected to the output of the execution trace storage unit 2 and the output of the representative execution trace storage unit 4. The input of the display unit 5 is connected to the output of the representative execution trace storage unit 4 and the output of the execution trace outlier extraction unit 8.

In the fourth embodiment, the execution trace 6 includes a variable. Then, when the variables of the execution trace 6 collected by the execution trace collection unit 1 include the variables specified in advance, the event conversion unit 12 uses a mapping that reduces the types of discrete values that the predetermined variables can take. Then, the variable of the execution trace 6 is converted. On the other hand, when the variables of the execution trace 6 collected by the execution trace collection unit 1 do not include the variables specified in advance, the event conversion unit 12 outputs the execution trace 6 as it is.

FIG. 18 is a diagram showing an example of mapping rules used in the event conversion unit 12. In FIG. 18, conversion conditions of x <0, 0 ≦ x <15, and 15 ≦ x are defined as rules for mapping the variable x to any of 0, 1, and 2, respectively. Further, as a rule for mapping the variable y to any of 0 and 1, conversion conditions of y <5, 5 ≦ y are defined, respectively.

FIG. 19 is a diagram showing an output example of the event conversion unit 12. In FIG. 19, as an input event sequence, x = 7 after calling the funcA function, y = 10 after calling the funcB function, y = -1 after calling the funcC function, and x = 30 after the funcC function is completed. It is shown. When the event conversion unit 12 converts the input event sequence of FIG. 19 using the rule of FIG. 18, the output event sequence of FIG. 19 is obtained. That is, an event sequence is obtained in which x = 1 after calling the funcA function, y = 1 after calling the funcB function, y = 0 after calling the funcC function, and x = 2 after the funcC function is completed.

<Operation>
FIG. 20 is a flowchart showing the operation of the event conversion unit 12. Hereinafter, this operation will be described with reference to FIG.

First, in step S61, the event conversion unit 12 acquires an event from the execution trace 6 collected by the execution trace collection unit 1. After that, the process proceeds to step S62.

In step S62, the event conversion unit 12 determines whether or not the variable in the event acquired in step S61 matches the variable specified by the mapping rule as shown in FIG. If it is determined that they match, the process proceeds to step S63, and if it is determined that they do not match, the process proceeds to step S64.

In step S63, the event conversion unit 12 converts the value of the variable in the event acquired in step S61 according to the mapping rule, rewrites it to the converted value, and outputs it. After that, the process proceeds to step S65.

In step S64, the event conversion unit 12 outputs the event acquired in step S61 as it is. After that, the process proceeds to step S65.

In step S65, the event conversion unit 12 determines whether or not there is an unprocessed event in the execution trace 6 collected by the execution trace collection unit 1. If it is determined that the event exists, the process returns to step S61, and in step S61, the event conversion unit 12 acquires an unprocessed event from the execution trace 6 collected by the execution trace collection unit 1. If it is determined that it does not exist, the operation of FIG. 20 ends.

<Summary of Embodiment 4>
According to the impact analyzer according to the fourth embodiment as described above, the variables of the execution trace 6 collected by the execution trace collection unit 1 are converted by using the mapping that reduces the types of variables. According to such a configuration, not only the function transition but also the performance influence considering the conversion condition of the variable can be compared and the cause can be visualized. At this time, since the number of execution traces 6 and thus the number of representative execution traces 7 can be suppressed by reducing the types of variables, it is possible to improve the processing efficiency of the impact analysis device and the work efficiency of the user.

<About others>
The execution trace collection unit 1, the representative execution trace calculation unit 3, the execution trace deviation value extraction unit 8, the event restriction unit 10, and the event conversion unit 12 are, for example, CPUs (Central Processing Units) (not shown) of a computer corresponding to the impact analyzer. ) Etc. are realized as the functions of the CPU by executing a program stored in a storage device such as a semiconductor memory (not shown) of the impact analyzer.

In the present invention, each embodiment can be freely combined, and each embodiment can be appropriately modified or omitted within the scope of the invention.

Although the present invention has been described in detail, the above description is exemplary in all embodiments and the present invention is not limited thereto. It is understood that innumerable variations not illustrated can be assumed without departing from the scope of the present invention.

1 Execution trace collection unit, 2 Execution trace storage unit, 3 Representative execution trace calculation unit, 4 Representative execution trace storage unit, 5 Display unit, 6 Execution trace, 7 Representative execution trace, 8 Execution trace outlier extraction unit, 10 Event restriction Department, 12 Event conversion department.

Claims

An execution trace collection unit that collects a plurality of events including an entry event and an exit event of a function in software and a plurality of execution traces including the times of the plurality of events under different plurality of conditions.
An execution trace storage unit that accumulates the plurality of execution traces collected by the execution trace collection unit, and an execution trace storage unit.
A representative execution trace calculation that calculates a representative execution trace including the plurality of events and the average and variation of the time of each of the plurality of events based on the plurality of execution traces accumulated in the execution trace storage unit. Department and
A representative execution trace storage unit that stores the representative execution trace calculated by the representative execution trace calculation unit, and a representative execution trace storage unit.
An impact analysis apparatus including an output unit that outputs the representative execution trace accumulated in the representative execution trace storage unit.
The impact analyzer according to claim 1.
A permissible range based on the average included in the representative execution trace stored in the representative execution trace storage unit is defined, and among the plurality of execution traces stored in the execution trace storage unit, the time is the time. Further provided with an extraction unit for extracting the execution trace that is out of the permissible range,
The output unit is an impact analyzer that superimposes and displays the execution trace extracted by the extraction unit on the representative execution trace used for extraction.
The impact analyzer according to claim 2.
The representative execution trace storage unit is
Accumulate a plurality of the representative execution traces in which the combination of the plurality of events is different from each other,
The extraction unit
Among the plurality of representative execution traces stored in the representative execution trace storage unit, the representative execution trace in which the difference between the combination of the plurality of execution traces stored in the execution trace storage unit and the plurality of events is the smallest is selected. Impact analyzer used for extraction.
The impact analyzer according to any one of claims 1 to 3.
The execution traces accumulated in the execution trace storage unit are limited to the execution traces having a predetermined start event and a predetermined end event among the execution traces collected by the execution trace collection unit. An impact analyzer that further comprises a restriction section.
The impact analyzer according to any one of claims 1 to 4.
The execution trace contains more variables
An impact analyzer further comprising a conversion unit that converts the variable of the execution trace collected by the execution trace collection unit by using a mapping that reduces the types of variables.
The execution trace collection unit collects a plurality of events including the entrance event and the exit event of the function in the software and a plurality of execution traces including the times of the plurality of events under different conditions.
The execution trace storage unit accumulates the plurality of execution traces collected by the execution trace collection unit, and the execution trace storage unit accumulates the plurality of execution traces.
Based on the plurality of execution traces accumulated in the execution trace storage unit, the representative execution trace calculation unit includes the plurality of events and the average and variation of the time of each of the plurality of events. Is calculated and
The representative execution trace accumulating unit accumulates the representative execution trace calculated by the representative execution trace calculation unit, and then accumulates the representative execution trace.
An impact analysis method in which the output unit outputs the representative execution trace accumulated in the representative execution trace storage unit.
A means for collecting a plurality of events including an entrance event and an exit event of a function in software and a plurality of execution traces including the times of the plurality of events under different conditions.
A means of accumulating the collected execution traces and
A means for calculating a representative execution trace including the plurality of events and the average and variation of the time of each of the plurality of events based on the accumulated plurality of execution traces.
Means for accumulating the calculated representative execution trace and
An impact analysis program that causes a computer to execute a means for outputting the accumulated representative execution trace.