JP2008090699A - Trace logging method, apparatus and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique capable of controlling trace log output according to a rare situation that a programmer could not predict in advance or a load state of a computer system.
In a trace logging method for outputting a trace log at the time of program execution, runtime information indicating dynamic information is measured at the time of program execution, and the measured runtime information is stored in a storage device. And reading the measurement value, which is runtime information stored in the storage device, and calculating an evaluation value for evaluating the measurement value, and the calculated evaluation value satisfies a threshold condition specified in the policy definition table And a step of outputting a trace log to a storage device.
[Selection] Figure 1

Description

  The present invention relates to a trace logging technique for outputting a trace log during program execution.

  Conventionally, when outputting a trace log indicating a state at the time of execution of a program, as described in Patent Document 1, only data to be processed by a program has been used as a criterion for outputting a trace log. That is, in the data trace method of Patent Document 1, it is determined whether or not to output trace data based on whether or not the data during program execution satisfies a predetermined trace output condition. The trace data is collected when a certain data corresponds to a value that can be assumed in advance, for example, if the input data of the transaction is equal to “A” or “B”.

JP-A-5-241890

  In the above prior art, since the program targets only data in a range that can be assumed in advance, this method limits the conditions for outputting the trace log to a range that can be assumed in advance. There is a problem that the trace log cannot be output if a situation occurs that the programmer could not assume in advance.

  Further, in the above prior art, the output amount of the trace log depends only on the data to be processed, and ignores the load state of the computer system. For this reason, when the computer system is in a high load state, there is also a problem that the output amount of the trace log cannot be suppressed in order to reduce the load.

  An object of the present invention is to solve the above problems and provide a technique capable of controlling trace log output according to a rare situation that a programmer could not predict in advance or a load state of a computer system.

  The present invention outputs a trace log in accordance with dynamic runtime information of a program in a trace logging apparatus that outputs a trace log during program execution.

  In the trace logging device according to the present invention, when executing a program for which trace logging is performed, the execution time information indicating the dynamic information is measured as to how many times the program branches to which branch destination by a branch instruction. . Then, the ratio of the number of branches for each branch destination, that is, the branch probability is calculated as an evaluation value, and a trace log is output when the calculated branch probability is less than a certain value.

  In the trace logging apparatus according to the present invention, the number of execution steps spent for executing the target process of the program and the number of execution steps spent for executing the trace log output process are measured. Trace log output is suppressed when the ratio between the two is greater than a certain value determined by the load state of the system.

  As described above, in the present invention, since the trace log is output when the branching probability at the time of execution of the program is less than a certain value, the trace log can be output even in a rare situation where the programmer could not predict in advance. Is possible. This trace log makes it possible to correctly grasp the execution history of the program even in rare circumstances, and to detect signs of abnormality in advance. In addition, it is possible to easily investigate the cause when a failure occurs.

  Further, according to the present invention, the overhead of trace log output is reduced by suppressing the amount of trace log output according to the load state of the computer system. As a result, the high load state of the computer system can be alleviated.

  According to the present invention, it is possible to control trace log output in accordance with a rare situation in which it is difficult for a programmer to predict in advance or a load state of a computer system.

  A trace logging apparatus according to an embodiment that outputs a trace log during program execution will be described below.

  FIG. 1 is a diagram showing a configuration of a trace logging apparatus with a profiling function according to the present embodiment. As shown in FIG. 1, the computer 100 according to this embodiment includes a profiler unit 112 and a trace log output unit 113.

  The profiler unit 112 is a processing unit that measures runtime information indicating the dynamic information when executing a program and stores the measured runtime information in a storage device. The trace log output unit 113 reads the measurement value, which is runtime information stored in the storage device, calculates an evaluation value for evaluating the measurement value, and the calculated evaluation value is specified in the policy definition table. A processing unit that outputs a trace log to a storage device when a threshold condition is satisfied.

  A program for causing the computer 100 to function as the profiler unit 112 and the trace log output unit 113 is recorded on a recording medium such as a CD-ROM and stored in a magnetic disk or the like, and then loaded into a memory and executed. . The recording medium for recording the program may be a recording medium other than the CD-ROM. The program may be used by installing it from the recording medium into the information processing apparatus, or the program may be used by accessing the recording medium through a network.

  As shown in FIG. 1, a computer 100 that is a trace logging apparatus according to this embodiment includes a CPU 101, a keyboard 102, a display 103, storage devices 104 to 106, and a main storage device 108, which are connected by a bus 107. ing.

  An OS 109 and application programs 110 are expanded in the main storage device 108. The application program 110 includes a main processing unit 111, a profiler unit 112, and a trace log output unit 113.

  The application program 110 determines whether or not to output a trace log and a main processing unit 111 that performs target processing of the program, a profiler unit 112 that collects and records program execution information, and outputs a trace log based on the result. The trace log output unit 113 is repeatedly executed in order.

  FIG. 2 is a diagram showing an outline of information stored in the storage devices 104 to 106 according to the present embodiment. As shown in FIG. 2, runtime information such as branch probabilities is stored in each storage device.

  The statistical information storage device 104 stores a branch probability table 300, an execution step number table 400, and a subroutine information table 500. The profiler unit 112 records program execution information in these tables 300 to 500. The trace log output unit 113 acquires information from these tables 300 to 500, and determines whether to output a trace log based on the information.

  The policy definition storage device 105 stores a policy definition table 600. The trace log output unit 113 determines whether to output a trace log according to the rules recorded in the policy definition table 600.

  The trace log storage device 106 stores a trace log table 700. The trace log output unit 113 outputs the trace log to the trace log table 700 and stores it.

  FIG. 3 is a diagram illustrating an example of a branch probability table according to the present embodiment. The branch probability table 300 records the branch of the process that occurred in the execution of the program and the number of times.

  The branch probability table 300 includes a list of branch locations (branch sources) 301, a list of branch destinations 302 corresponding to each branch location, and the number of branches 303 that is the number of branches to each branch destination. Here, when the application program 110 is executed for the first time, information on the branch location, branch destination, and branch count collected in the test stage is stored as initial values, and thereafter the application program 110 is executed. Assume that information on the number of branches is accumulated every time.

  FIG. 4 is a diagram showing an example of the execution step number table of the present embodiment. The execution step number table 400 records the number of execution steps that the program has spent executing each process.

  The execution step number table 400 includes a process type 401 executed by the program and a step number 402 required for the execution.

  The total number of execution steps represents the total number of steps required for execution of each process by the program. The number of trace log output execution steps represents the total number of steps required for the process of determining whether to output the trace log and the process of outputting the trace log. The number of main process continuous execution steps represents the total number of steps required to execute the main processing unit 111 as the target process of the program since the trace log was last output. When the trace log output unit 113 outputs the trace log, the number of main process continuous execution steps is returned to zero.

  FIG. 5 is a diagram showing an example of the subroutine information table 500 of the present embodiment. The subroutine information table 500 records information on the subroutine currently being executed and the execution history of the subroutine.

  The subroutine information table 500 includes a subroutine name 501 that is a unique name of a subroutine, parameter information 502 that determines subroutine processing, and a subroutine execution start time 503. The call history of the main routine and the subroutine called from the main routine is recorded in the order of call.

  The profiler unit 112 additionally registers subroutine information at the top of the subroutine information table 500 when the execution of the subroutine is started. Also, when the execution of the subroutine is finished and the process returns to the caller, the subroutine information that has been executed from the top of the subroutine information table 500 is deleted. By this processing, the subroutine information table 500 holds information on the currently executed subroutine at the top.

  FIG. 6 is a diagram showing an example of the policy definition table of this embodiment. The policy definition table 600 defines a determination rule for determining whether to output a trace log.

  The policy definition table 600 includes a setting item 601 indicating the type of determination rule, a reference value for the setting item or a setting value / setting equation 602 that is an expression for giving a reference value, and a priority 603 of the determination rule.

  A user or administrator of this system can control when the trace log is output by changing the setting value / setting expression 602 and the priority order 603.

  FIG. 7 is a diagram showing an example of the trace log table of the present embodiment. The trace log table 700 stores the trace log output by the trace log output unit 113.

  The trace log table 700 includes a corresponding item 701 that indicates which setting item 601 of the policy definition table 600 is output because each trace log satisfies the setting value / setting expression 602, a value 702 taken by the setting item 601 at that time, It consists of a trace log text 703 determined in correspondence with the corresponding item 701 and additional information 704.

  The first line of FIG. 7 is an example of a trace log output when the condition according to the branching probability of the first line of the policy definition table 600 of FIG. 6 is satisfied. The branch probability is output as the corresponding item, the branch probability is output as the value, the branch location and branch destination are output as the trace log body, and the name and value of the condition variable used for the branch condition are added as additional information.

  The second line in FIG. 7 is an example of a trace log output when the condition based on the number of main processing continuous execution steps in the third line of the policy definition table 600 in FIG. 6 is satisfied. The corresponding item is the number of main process continuous execution steps, the value is the number of execution steps that executed the main process continuously, the name of the subroutine currently being executed as the trace log text, and the current execution as additional information Print the names and values of subroutine parameters.

  The third line in FIG. 7 is an example of trace log output when the condition according to the trace log output command in the fourth line of the policy definition table 600 in FIG. 6 is satisfied. An arbitrary message given by an explicit trace log output command is output as a trace log body, indicating that the item is a trace log output command.

  The fourth line of FIG. 7 is an example of a trace log output when the condition according to the subroutine execution time of the fifth line of the policy definition table 600 of FIG. 6 is satisfied. The corresponding item is the subroutine execution time, the subroutine execution time as the value, the subroutine name that has been executed as the trace log body, and the name and value of the subroutine parameter that has been executed as the load information.

  FIG. 11 is a diagram showing a flow of processing executed by the application program 110 of this embodiment. FIG. 11 shows processing in which the application program 110 repeatedly executes the main processing unit 111, the profiler unit 112, and the trace log output unit 113 in order.

  The main processing unit 111 executes the target process of the program until the end instruction, the explicit trace log output instruction, the branch instruction, the execution of the subroutine is completed, or a predetermined number of steps are executed (step 901). When the end command is reached (step 902), the program execution is ended (step 912). When other conditions are satisfied, control is transferred to the profiler unit 112.

If the profiler unit 112 has reached an explicit trace log output command (step 903), the profiler unit 112 shifts control to step 907 of the trace log output unit 113. If it has reached the branch instruction (step 904), the number of branches 303 corresponding to the branch point x- branch destination x _m of the branch probability table 300 is increased by one (step 906). In any of these cases, if the control is transferred to the profiler unit because the execution of the subroutine has been completed or a certain number of steps have been executed, the total number of execution steps in the execution step number table 400 and the main process continuous execution The number of steps is integrated (step 905).

  The trace log output unit 113 applies the determination rule of the policy definition table 600 according to the priority order and determines whether or not to output the trace log (step 907). Details of the determination method will be described with reference to FIG. If it is determined to output the trace log (step 908), the trace log is output to the trace log table 700 (step 909).

When outputting the trace log, the profiler unit 112 integrates the total execution step number and the trace log output execution step number in the execution step number table 400, and sets the main process continuous execution step number to zero (step 911). When the trace log is not output, the total number of execution steps in the execution step number table 400 is integrated (step 910).
As described above, the series of processing is repeatedly executed by returning to step 901 until the program ends.

  FIG. 12 is a diagram showing a flow of determination processing of the trace log output unit 113 of this embodiment. FIG. 12 illustrates processing in which the trace log output unit 113 determines whether to output a trace log according to the policy definition table 600.

  First, the setting item with the highest priority is acquired from the policy definition table 600 (step 1001). By the evaluation method 1100-1500 corresponding to this setting item, it is determined whether to output the trace log or not, or whether to hold the determination and determine by evaluation of other setting items (step 1002).

If it is determined whether to output the trace log (step 1003), the determination is completed based on the result (step 908).
If the determination is suspended and there is no setting item with the next highest priority in the policy definition table 600 (step 1004), it is determined not to output the trace log (step 908).
If the determination is suspended and there is a setting item with the next highest priority in the policy definition table 600, the setting item is acquired (step 1005), and the process returns to step 1002 to repeat the evaluation and determination corresponding to the setting item. .

  FIG. 13 is a diagram showing a flow of processing for determining whether or not to output a trace log according to the branch probability according to the present embodiment. When the main processing unit 111 reaches the branch instruction, the main processing unit 111 passes a parameter indicating that the branch instruction has been reached to the profiler unit 112. The profiler unit 112 delivers the parameters when transferring control to the trace log output unit 113. The trace log output unit 113 determines whether or not the branch instruction has been reached based on this parameter.

  If the execution of the program has not reached the branch instruction (step 1101), the determination is suspended, and the next priority setting item is evaluated.

  When the execution of the program reaches the branch instruction, the branch probability is calculated and it is determined whether or not to output the trace log.

First, the number of branches 303 of the branch destination x ₁ ... X _n corresponding to the reached branch location x is acquired from the branch probability table 300 and totaled (step 1102). The branch probability x _m / Σx _i (i = 1... N) for the branch destination x _m determined by the evaluation of the branch condition is calculated (step 1103).

Next, a setting value for the branch probability is acquired from the policy definition table 600 (1104).
When the branching probability calculated in steps 1102 to 1103 is equal to or less than the set value (step 1105), it is determined that the trace log is output. If the branching probability exceeds the set value, the determination is suspended and the next priority setting item is evaluated.

  With the above processing, it is possible to output a trace log for all conditional branches in the program when the branch probability is less than or equal to the set value. That is, it is possible to automatically detect a rare situation that a programmer cannot assume in advance and output a trace log.

  FIG. 14 is a diagram showing a flow of processing for determining whether to output a trace log from the ratio of the number of execution steps required for outputting the trace log and the total number of execution steps according to this embodiment. First, the total number of execution steps and the number of trace log output execution steps are acquired from the execution step number table 400 (step 1201). The ratio of the number of execution steps required for outputting the trace log is calculated as Trace log output execution rate = Number of trace log output execution steps / Total number of execution steps (Step 1202).

  Next, a trace log output execution rate setting formula is acquired from the policy definition table 600 (step 1203). Using the load factor of the computer system as a variable, the value of the setting formula is calculated (step 1204).

  When the trace log output execution rate calculated in steps 1201 to 1202 exceeds the value of the setting formula (step 1205), it is determined that the trace log is not output. If the trace log output execution rate is less than or equal to the value of the setting expression, the determination is suspended and the next priority setting item is evaluated.

  As an example, let us consider the case where the load factor of a computer system is represented by L and the setting formula is 0.1 × (1-L). If the load factor of the computer system is 100%, L = 1, setting formula 0.1 x (1-L) = 0.1 x (1-1) = 0, so the trace log output rate does not depend on the trace log output execution rate. Is not output. When the load factor is 50%, L = 0.5, setting formula 0.1 x (1-L) = 0.1 x (1-0.5) = 0.05, so the trace log output execution rate exceeds 0.05 (= 5%) If it is, the trace log is not output.

  With the above processing, the execution rate of the trace log output can be suppressed to a certain ratio determined by the load state of the computer system. That is, the trace log output amount can be suppressed according to the load state of the computer system, and the high load state of the computer system can be alleviated.

  FIG. 15 is a diagram showing a flow of processing for determining whether or not to output a trace log every time the target processing of the program of this embodiment is executed for a certain number of steps. First, from the execution step number table 400, the main process continuous execution step number, which is the number of steps in which the target process of the program has been executed without interposing the trace log output process, is obtained (step 1301).

  Next, a setting value for the number of main processing continuous execution steps is acquired from the policy definition table 600 (step 1302). If the number of main process continuous execution steps is greater than or equal to the set value (step 1303), it is determined that a trace log is output. When the number of main process continuous execution steps is less than the set value, the determination is suspended and the next set item of priority is evaluated.

  With the above processing, it is possible to output a trace log for each predetermined number of execution steps without explicitly instructing to output a trace log, and it is possible to grasp the operation of the application program 110 from the trace log. .

  FIG. 16 is a diagram showing the flow of processing for determining whether or not to output a trace log when the explicit trace log output command of this embodiment is reached. When the main processing unit 111 reaches the trace log output command, the main processing unit 111 passes a parameter indicating that the trace log output command has been reached to the profiler unit 112. The profiler unit 112 delivers the parameters when transferring control to the trace log output unit 113. The trace log output unit 113 determines whether or not the trace log output command has been reached based on this parameter.

  When the execution of the program reaches an explicit trace log output command (step 1401), it is determined that the trace log is output. If the trace log output command has not been reached, the determination is suspended and the next priority setting item is evaluated.

  FIG. 17 is a diagram showing a flow of processing for determining whether to output a trace log when execution of the subroutine of the present embodiment is completed. When the main processing unit 111 completes execution of the subroutine, the main processing unit 111 passes a parameter indicating completion of execution of the subroutine to the profiler unit 112. The profiler unit 112 delivers the parameters when transferring control to the trace log output unit 113. The trace log output unit 113 determines whether the execution of the subroutine is completed based on this parameter.

If the program has not completed execution of the subroutine (step 1501), the determination is suspended and the next priority setting item is evaluated.
If the execution of the subroutine has been completed, the execution start time 503 of the subroutine currently being executed is acquired from the top of the subroutine information table 500, and the execution time of the subroutine is calculated from the difference from the current time (step 1502). .

Next, a set value of subroutine execution time is acquired from the policy definition table 600 (step 1503).
If the execution time of the subroutine that has completed execution is less than the set value (step 1504), the determination is suspended and the next priority setting item is evaluated.
If the execution time of the subroutine that has completed execution is equal to or greater than the set value, the execution start time of all subroutines in the subroutine information table 500 is advanced by the execution time (step 1505), and it is determined that the trace log is output. By advancing the execution start time by the execution time, it is possible to prevent the trace log from being output from all callers of the subroutine.

  FIG. 8 is a diagram showing an example of generation of the application program 110 of the present embodiment. FIG. 8 shows a process of generating an application program 110 that is an execution program having a profiler unit 112 and a trace log output unit 113 from a source program 114 including only a main processing unit 111 that is a target process of the program.

  The language processing system 115 analyzes the contents of the given source program 114, generates the profiler unit 112 and the trace log output unit 113, and embeds them in the execution program, thereby generating the application program 110.

  As one of the embodiments, there is a method in which the profiler unit 112 and the trace log output unit 113 are embedded in a program as a subroutine, and the calling process of these subroutines is embedded in a place where profile processing or trace log output processing is required in the program. is there.

  As another embodiment, there is a method of directly embedding a corresponding processing portion of the profiler unit 112 to the trace log output unit 113 in a program where necessary for profile processing or trace log output processing.

  FIG. 9 is a diagram showing an example of the source program 114 before being processed by the language processing system 115 of this embodiment. The number at the left end indicates the line number, and the contents of the program are indicated by pseudo code based on the programming language C.

  In FIG. 9, lines 3 to 6 are subroutine definitions, lines 8 to 23 are main routine definitions, lines 11 to 16 are branch instructions, line 18 is a subroutine execution, and line 20 is an explicit trace log output instruction.

  FIG. 10 is a diagram showing an example of the application program 110 after the example of the source program shown in FIG. 9 is processed by the language processing system 115 of the present embodiment. The number at the left end indicates the line number, and the contents of the program are indicated by pseudo code based on the programming language C. The process written in uppercase letters is a process embedded in the language processing system 115.

  In FIG. 10, lines 1 to 12 are subroutine definitions, lines 14 to 48 are main routine definitions, lines 24 to 37 are branch instructions, line 39 is a subroutine execution, and lines 41 to 43 are explicit trace log output instructions. is there.

  Lines 4 and 16 in FIG. 10 are processes for storing the name of the currently executed subroutine, the names and values of the parameters in the subroutine information table 500, and line 11 is a process for returning the subroutine information to the caller information.

  Lines 10 and 46 in FIG. 10 are processes for transferring control to the profiler unit 112 when execution of the subroutine is completed. The completion of subroutine execution is passed as a parameter. Since the name of the subroutine that has completed execution, the parameter name and value, and the execution start time are recorded in the subroutine information table 500, they are not passed as parameters.

  Lines 18 to 21 in FIG. 10 are processes for registering a process for transferring control to the profiler unit 112 every time a certain number of steps are executed as an interrupt process. It is set to pass as a parameter that a certain number of steps have been executed.

  Lines 28 and 34 in FIG. 10 are processes for transferring control to the profiler unit 112 when a branch instruction is reached. The fact that the branch instruction has been reached, the branch location, the branch destination, and the condition variable name and value used to determine the branch are passed as parameters. The language processing system 115 automatically generates and assigns branch locations and branch destination names.

  Line 43 in FIG. 10 is a process for transferring control to the profiler unit 112 when an explicit trace log output command is reached. An explicit trace log output command has been reached, and any specified message is passed as a parameter. This process corresponds to the 20th line in FIG. 9 and replaces the process of the source program 114.

  As described above, according to the trace logging apparatus of the present embodiment, the trace log is output according to the dynamic runtime information of the program. It is possible to control the trace log output according to the load state.

It is a figure which shows the structure of the trace logging apparatus with a profiling function of this embodiment. It is a figure which shows the outline | summary of the information preserve | saved at the memory | storage devices 104-106 of this embodiment. It is a figure which shows an example of the branch probability table of this embodiment. It is a figure which shows an example of the execution step number table of this embodiment. It is a figure which shows an example of the subroutine information table 500 of this embodiment. It is a figure which shows an example of the policy definition table of this embodiment. It is a figure which shows an example of the trace log table of this embodiment. It is a figure which shows an example of the production | generation of the application program 110 of this embodiment. It is a figure which shows an example of the source program 114 before being processed by the language processing system 115 of this embodiment. It is a figure which shows an example of the application program 110 after processing the example of the source program shown in FIG. 9 by the language processing system 115 of this embodiment. It is a figure which shows the flow of the process which the application program 110 of this embodiment performs. It is a figure which shows the flow of the determination process of the trace log output part 113 of this embodiment. It is a figure which shows the flow of the process which determines whether a trace log is output by the branch probability of this embodiment. It is a figure which shows the flow of a process which determines whether a trace log is output from the ratio of the number of execution steps required for the output of the trace log of this embodiment, and the total number of execution steps. It is a figure which shows the flow of a process which determines whether a trace log is output whenever the target process of the program of this embodiment is performed a fixed number of steps. It is a figure which shows the flow of a process which determines whether a trace log is output when it reaches | attains the explicit trace log output command of this embodiment. It is a figure which shows the flow of a process which determines whether trace log is output when the execution of the subroutine of this embodiment is completed.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 100 ... Computer, 101 ... CPU, 102 ... Keyboard, 103 ... Display, 104 ... Statistical information storage device, 105 ... Policy definition storage device, 106 ... Trace log storage device, 107 ... Bus, 108 ... Main storage device, 109 ... OS , 110 ... Application program, 111 ... Main processing unit, 112 ... Profiler unit, 113 ... Trace log output unit, 300 ... Branch probability table, 400 ... Execution step number table, 500 ... Subroutine information table, 600 ... Policy definition table, 700 ... Trace log table, 301 ... Branch location (branch source), 302 ... Branch destination, 303 ... Number of branches, 401 ... Processing type, 402 ... Number of steps, 501 ... Subroutine name, 502 ... Parameter information, 503 ... Execution start time, 601: Setting item, 602: Setting Value / setting-type, 603 ... priority, 701 ... corresponding item, 702 ... value, 703 ... trace log text, 704 ... additional information, 114 ... source program, 115 ... language processor.

Claims (5)

In a trace logging method in a trace logging device that includes a storage device and outputs a trace log during program execution,
The execution time information indicating the dynamic information is measured when the program is executed, the measured execution time information is stored in the storage device, and the measurement value which is the execution time information stored in the storage device is read. A trace logging method characterized by calculating an evaluation value for evaluating the measured value and outputting a trace log to a storage device when the calculated evaluation value satisfies a threshold condition specified in a policy definition table.   As the execution information of the program, the number of branches for each branch instruction in each branch instruction is measured and stored in a storage device, and the information on the number of branches stored in the storage device is read to evaluate the branch probability for each branch destination. The trace log indicating the branch destination is output to the storage device when the branch probability is calculated as a value and the branch probability is equal to or less than a predetermined setting value specified in the policy definition table. Trace logging method.   As the execution information of the program, the total number of execution steps indicating the total number of execution steps at the time of execution of the program and the number of execution steps of the trace log output indicating the number of execution steps at the time of output of the trace log are measured and stored in the storage device. The information of the total number of execution steps and the number of execution steps of trace log output stored in the storage device is read out, the trace log output execution rate is calculated as the evaluation value, and the trace log output execution rate is calculated in the policy definition table. The trace logging method according to claim 1, wherein the output of the trace log is suppressed when the value is larger than a value of a predetermined setting expression specified. In a trace logging device that has a storage device and outputs a trace log during program execution,
A profiler unit that measures runtime information indicating dynamic information at the time of program execution and stores the measured runtime information in the storage device, and a measurement value that is runtime information stored in the storage device A trace log output unit that calculates an evaluation value for evaluating the measured value and outputs the trace log to the storage device when the calculated evaluation value satisfies a threshold condition specified in the policy definition table. A trace logging device comprising: In a program for causing a computer to execute a trace logging method in a trace logging device that includes a storage device and outputs a trace log during program execution,
Measuring execution time information indicating the dynamic information at the time of execution of the program and storing the measured execution time information in the storage device; and a measurement value which is the execution time information stored in the storage device. An evaluation value for reading and evaluating the measurement value is calculated, and when the calculated evaluation value satisfies a threshold condition specified in the policy definition table, a step of outputting a trace log to a storage device is executed by a computer. A program characterized by that.

Images