JP2008123438A - Computer system, program information collection method, and computer program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To collect proper information for repairing a problem occurred in a multiprocessor system. <P>SOLUTION: This computer system mounted with a plurality of CPUs is provided with a collection point setting part 104 for setting sequentially acquisition points serving as a trigger for the start of acquisition processing for acquiring program information including a register value and process information under execution of a program, within the program, a program information collection part 107 for conducting the acquisition processing in the set acquisition point, a counter part 17 for outputting a count value incremented in each of the accesses, a count value acquisition part 101 for acquiring the count value from the counter part 17 in every acquisition processing, and a history information output means 110 for outputting the acquired program information together with the corresponding count value, to be stored. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a method for collecting program information used for analyzing a problem that occurs in a computer having a plurality of CPUs.

  When repairing a problem that occurred in a server machine such as a general-purpose server, generally, analysis is performed based on the dump information collected in a snapshot when the problem occurs, and a reproduction test to reproduce the problem is repeated. The cause is pursued.

  As a method for collecting information necessary for analysis, generally, the memory state at the time of occurrence of the problem is frozen, and information about the memory state and the CPU (processor) state is collected and stored as dump information. To be done. However, since the information collected in this way is static information, the location where the problem is detected can be identified, but it is often difficult to find the point that causes the problem. Therefore, it is required to collect history information (running history information) that can trace (trace) the progress of execution of instructions. As a method for that purpose, a method as disclosed in Patent Document 1 has been proposed.

  According to Patent Document 1, a log area composed of a header area and a log data area is allocated on a memory, information necessary for acquiring log data is stored in the header area, and the header area is referred to when executing a log routine. Based on the information stored there, the log data area is made to acquire log data.

  By the way, the current general-purpose server is generally a multiprocessor system in which a plurality of CPUs execute instructions in parallel. In a multiprocessor system, there may be a problem that data is destroyed due to processing conflicts among a plurality of CPUs. Since such a problem depends on the timing of competition, the reproducibility is poor and the analysis is difficult. In order to investigate the cause of a problem that occurs in a multiprocessor system, it is necessary to collect sufficient history information, but it is difficult to do so with the conventional method as described above.

  As a method for collecting history information in a multiprocessor system, a method as disclosed in Patent Document 2 has been proposed.

According to Patent Document 2, a trace device that traces a program executed by one microprocessor system of a monitoring target microprocessor system having a multiprocessor configuration is configured to execute each instruction at each instruction stage of the program. The information on the time of execution and the history of the trace result information are recorded, and the information on the time and the trace result information are output in correspondence with each other. Then, the history is edited and displayed on the output unit.
JP-A-8-171501 Japanese Patent Laid-Open No. 10-260864

  However, according to Patent Document 2, the information on the memory bus is copied and output to make it history information. In other words, since only the memory read or write is targeted, there is a case where information necessary for analysis when a problem occurs cannot be collected.

  Conventionally, there has been a method of using dedicated hardware for outputting history information. However, with this method, when a general-purpose board provided for general-purpose servers is used, the technology is updated quickly, so it is necessary to frequently upgrade the board itself. You will have to add hardware. As a result, extra costs are incurred. In addition, there is a risk of quality instability.

  The present invention has been made in view of the above problems, and an object of the present invention is to collect information necessary for repairing problems that occur in a multiprocessor system.

  A computer system according to the present invention is a computer system equipped with a plurality of CPUs, and triggers for starting an acquisition process for acquiring program information including a register value and process information of the CPU during execution (starting) of the program Acquisition point (program stop position) to be set in order in the program, acquisition point setting means, program information acquisition means for performing the acquisition process each time the program processing comes to the acquisition point, A counter that outputs a count value that is incremented or decremented each time it is accessed, and a count value acquisition unit that accesses the counter and acquires the count value each time the acquisition process is performed by the program information acquisition unit And acquired by the acquisition process The serial program information, having a program information storage means for storing together with the count value corresponding to the acquired the program information by the count value obtaining means.

  Preferably, the program information acquisition unit and the count value acquisition unit are provided for each CPU, and the counter is commonly accessed from the count value acquisition units provided for each CPU. The program information acquisition means acquires the register value from the corresponding CPU.

  Further preferably, when the acquisition target receiving unit that receives the acquisition target of the program information designated by the user and the acquisition target received by the acquisition target receiving unit is a function, the acquisition point in the function In-function setting means for sequentially setting.

  According to another aspect of the computer system of the present invention, when a predetermined data is accessed during execution of a program, break setting means for setting the program to break, and a break occurs during execution of the program Program information acquisition means for performing acquisition processing for acquiring program information including the register value and process information of the CPU, a counter that outputs a count value incremented or decremented each time there is an access, and the program Each time the acquisition process is performed by the information acquisition unit, the count value acquisition unit that accesses the counter to acquire the count value, and the program information acquired by the acquisition process is acquired by the count value acquisition unit. Corresponding to the program information Having a program information storage means for storing together with the serial count value.

  Information necessary for repairing problems that occur in a multiprocessor system can be collected.

  FIG. 1 is a diagram showing an example of a hardware configuration of a computer 1 according to the present invention, and FIG. 2 is a diagram showing a process stored in a memory space.

  As shown in FIG. 1, the computer 1 includes a plurality of CPUs (Central Processing Units) 11 a, 11 b,..., A memory 12, a hard disk 13, and a communication interface 14. The computer 1 is a personal computer or a workstation compatible with a multiprocessor, and is used as a server machine such as a general-purpose server. Hereinafter, each or all of the CPUs 11a, 11b,... May be referred to as “CPU 11”. The same applies to other codes.

  Various programs relating to an OS (Operating System) and various tools are installed in the hard disk 13. These programs are read at the time of execution, and stored in the memory space of the memory 12 in units of execution such as processes. The process is configured by a process structure including a process ID and stack information (process stack information) as shown in FIG. 2, for example. Program processing is performed by sequentially executing such processes.

  The communication interface 14 is an interface for communicating with other devices via a communication line such as a LAN, WAN, or the Internet. The communication interface 14 includes a network interface card (NIC), a modem, a USB, and other various interfaces.

  The computer 1 is connected to a keyboard 2 that is an input device for a user to input a command, a mouse 3, and a display 4 that is a display device.

In the computer 1, logs (history information) relating to resource usage and the like are output and collected during execution of a program designated in advance by the user. In the present embodiment, information related to CPU registers (register information), information related to processes held in the memory (process information), and the like are collected as logs. It is also possible to specify a specific function or specific data as an object (log collection object) for collecting logs. In that case, a log is output in the processing related to the specified function or data. Hereinafter, processing related to log collection will be described.
[Description of functional configuration]
3 is a diagram showing an example of the functional configuration of the computer 1, FIG. 4 is a diagram for explaining the process information acquisition processing, FIG. 5 is a diagram showing an example of the configuration of the history information output processing unit 15, and FIG. It is a figure for demonstrating the process of register information acquisition.

  As shown in FIG. 3, the computer 1 includes a history information output processing unit 15, a process information acquisition unit 16, a counter unit 17, and the like.

  The history information output processing units 15a, 15b,... Are provided in association with the CPUs 11a, 11b,. That is, the history information output processing unit 15 and the CPU 11 are associated one-to-one like the history information output processing unit 15a and the CPU 11a, and the history information output processing unit 15b and the CPU 11b. The history information output processing unit 15 performs processing for outputting a log during execution of a predetermined program. Hereinafter, this log may be referred to as “history information DR”. The output history information DR is stored in a predetermined area of a storage device such as the memory 12 or the hard disk 13.

  As shown in FIG. 4, the process information acquisition unit 16 performs a process for acquiring a process ID and stack information stored in the memory space. The process ID and the stack information acquired here are output to the history information output processing unit 15 and used to generate the history information DR.

  The process ID can be collected by referring to the process structure on the process stack. Further, by collecting stack information, it is possible to know process backtrace information. Hereinafter, the process ID and stack information may be collectively referred to as “process information”.

  The counter unit 17 holds a numerical value (count value) for counting log output, and increments (or decrements) the count value by one each time it is accessed from the history information output processing unit 15. Output. Note that the count value may be incremented after output. Since the counter unit 17 is provided in common (only one) for the plurality of CPUs 11a, 11b,..., An exclusive count value is output to the CPU 11. The output count value is used by the history information output processing unit 15 to generate history information DR.

  As shown in FIG. 5, the history information output processing unit 15 includes a count value acquisition unit 101, a register value acquisition unit 102, a collection target designation reception unit 103, a collection point setting unit 104, a collection target function setting unit 105, and a collection target. The data setting unit 106, the program information collection unit 107, the history information generation unit 108, the output destination designation reception unit 109, the history information output unit 110, and the like.

  Note that the program and data for realizing the processing of each unit in FIGS. 3 and 5 are installed in the hard disk 13. These programs and data are read into the memory 12 as necessary, and the CPU 11 executes the programs.

  The count value acquisition unit 101 accesses the counter unit 17 and acquires the count value therefrom. As illustrated in FIG. 6, the register value acquisition unit 102 acquires a register value (register value) as register information of the corresponding CPU 11. The “corresponding CPU 11” is the CPU 11 corresponding to the history information output processing unit 15 to which the register value acquisition unit 102 belongs. The other parts included in the history information output processing unit 15 are expressed in the same manner.

  The collection target designation receiving unit (collection target selection unit) 103 receives designation from the user regarding the log collection target, determines whether the specified log collection target is a program, a function, or data. Accordingly, the designated log output target is notified to the collection point setting unit 104, the collection target function setting unit 105, or the collection target data setting unit 106. That is, when a program is designated, the program is notified to the collection point setting unit 104. When a function is designated, the function is notified to the collection target function setting unit 105. When data is designated, the data is notified to the collection target data setting unit 106.

  The collection point setting unit 104, the collection target function setting unit 105, and the collection target data setting unit 106 temporarily stop processing of the program being executed and start collecting history information DR, that is, log collection. Set a trigger position (hereinafter sometimes referred to as “program stop position”). The collection point setting unit 104 sets the program stop position so that all history information DR can be collected in the designated program. The collection target function setting unit 105 sets a program stop position in the designated function. The collection target data setting unit 106 sets a program stop position for the specified data.

  The program information collection unit 107 acquires a register value from the register value acquisition unit 102 and acquires process information from the process information acquisition unit. Hereinafter, register values and process information may be collectively referred to as “program information”.

  The history information generation unit 108 generates the history information DR by processing the program information collected by the program information collection unit 107 as necessary and converting the format.

  It is designated in advance by the user which method (output method) the generated history information DR is output to which location (output destination). For example, the following contents are specified.

    (1) A data area is secured in the memory space and output there.

    (2) Output to a standard output destination provided in advance.

    (3) Output to a specific file.

    (4) Transmit (packet transmission) to a designated IP address (server, etc.) via a communication line.

  It is also possible to designate a plurality of output destinations for one history information DR. In that case, the history information DR is output to each of the designated output destinations.

  The output destination designation receiving unit (output selecting unit) 109 receives designation from the user about the output destination of the history information DR and notifies the history information output unit 110 of it.

  The history information output unit (history holding unit) 110 outputs the history information DR generated by the history information generation unit 108 to the output destination notified from the output destination designation receiving unit 109.

Hereinafter, the processing of each unit when collecting the history information DR will be described in more detail by broadly dividing into the first method, the second method, and the third method. In the following, it is assumed that an area in the memory 12 provided for each CPU 11 is designated as the output destination of the history information DR.
[First method]
FIG. 7 is a flowchart for explaining the flow of processing for collecting history information DR by the first method, FIG. 8 is a diagram for explaining processing for setting a program stop position, and FIG. 9 is an example of history information DR. FIG. 10 is a diagram showing history information DR output for each CPU 11.

  In order to collect the history information DR, the user designates a log collection target program (hereinafter referred to as “program P1”) (# 100). The collection target designation receiving unit 103 receives the designation.

  Prior to the start of the program P1, the collection point setting unit 104 sets a program stop position for the program P1 based on the designation from the user received by the collection target designation receiving unit 103. Here, the setting of the program stop position will be described with reference to FIG.

  In FIG. 8, a program P 1 is stored in the memory space of the memory 12. When setting the program stop position, the collection point setting unit 104 first acquires the head address of the program P1 (# 101). As shown in FIG. 8I, the instruction M1 stored at the head address is saved by copying it to an instruction saving area (data saving area) prepared in advance in the memory space ( # 102). Further, as shown in (ii) of FIG. 8, the contents of the head address of the program P1 are replaced with a trap instruction (# 103). As a result, the head address of the program P1 is set as the program stop position. An instruction is an execution unit such as a process.

  When the processing of the program P1 is started, the trap instruction stored at the head address is executed (# 104). As a result, the program information collection unit 107 is activated and starts operating. The trap instruction includes, for example, an instruction for starting the program information collecting unit 107 and an address of an instruction (jump destination) to be processed next.

  When starting the operation, the program information collection unit 107 temporarily stops the processing of the program P1. Then, processing for collecting program information at that time is performed (# 105).

  When collecting program information, the program information collection unit 107 instructs the register value acquisition unit 102 to acquire the register information of the CPU 11 from the corresponding CPU 11. In parallel or before or after this, the process information acquisition unit 16 is instructed to acquire process information. The register information and process information are received from the register value acquisition unit 102 and the process information acquisition unit 16, respectively.

  The history information generation unit 108 receives the program information (register information and process information) collected by the program information collection unit 107, and performs processing for generating history information DR using the program information. In this process, the history information generation unit 108 instructs the count value acquisition unit 101 to acquire a count value. Based on the instruction, the count value acquisition unit 101 accesses the counter unit 17 and acquires the count value (# 106).

  At this time, the counter unit 17 increases the count value held by itself by one. As a result, when accessed next time, a count value that is one greater than the count value output this time is output. That is, a sequential count value that is unique as a whole is paid out for each program information.

  When the count value is acquired, the count value acquisition unit 101 outputs it to the history information generation unit 108. The history information generation unit 108 gives the count value to the program information received from the program information collection unit 107. Furthermore, the history information DR as shown in FIG. 9 is generated by rearranging the contents shown in the program information, making it redundant so that it is easy to understand, or performing unnecessary processing such as deleting unnecessary information. To do.

  As shown in FIG. 9, the history information DR includes “CPU identification information” and “instruction” in addition to the count value, register value, process ID, and stack information.

  The “CPU identification information” indicates information for identifying each of the CPUs 11a, 11b,. In the above example, CPU identification information of the CPU that executes the instruction M1 may be acquired. The “command” indicates a command sentence executed by the CPU. Here, for simplicity of explanation, a value as shown in FIG. 9 is used as the count value, but a value of about 32 bits is actually used.

  The history information output unit 110 outputs and stores the history information DR generated by the history information generation unit 108 in an area (history storage area) in the memory 12 provided for the corresponding CPU 11 (# 107).

  Thereafter, the program information collection unit 107 executes the instruction M1 saved in the instruction saving area (# 108), and after execution, the instruction M1 is stored with the original address of the memory space, that is, the trap instruction. (# 109). In this case, the program information collection unit 107 may support execution of the instruction M1. That is, the address of the instruction M1 may be passed to the CPU 11, and the actual process of the instruction M1 may be executed by the CPU 11. The same applies to step # 208 of FIG. 11 in the [second method] described later.

  Further, the program information collection unit 107 instructs the collection point setting unit 104 to set the next program stop position. Based on the instruction, the collection point setting unit 104 obtains the address of the instruction (next instruction) to be executed next to the instruction M1 in the program P1.

  The address of the next instruction is normally the address next to the address where the instruction M1 is stored, but when the instruction M1 is an instruction related to a function call, the address to be called is the address of the next instruction. . In the case of a branch instruction, the address of the next instruction is obtained by referring to the register information.

  When the address of the next instruction (herein referred to as “instruction M2”) is obtained (Yes in # 110 and # 111), the instruction M2 is saved in the instruction saving area (# 112), and the address The contents are replaced with a trap instruction (# 113).

  When the processing up to step # 113 is completed, the processing of the program P1 is continued. When the process reaches (reaches) the instruction M2, the trap instruction stored at the address of the instruction M2 is executed in step # 113 (# 104). Then, the process from step # 105 to step # 109 for the instruction M2 is performed. Thereafter, in step # 110, the address of the next instruction is further determined, and the same processing is repeated.

Through the above processing, as shown in FIG. 10, the history information DR for all the instructions included in the program P1 is sequentially output for each CPU 11. That is, it is possible to obtain a program running history that can sequentially trace executed instructions across a plurality of CPUs 11.
[Second method]
FIG. 11 is a flowchart for explaining the flow of processing for collecting history information DR by the second method. Here, a case will be described in which history information DR is collected only when a function arbitrarily designated by the user is executed. Note that detailed description of the same processing as that described in [First Method] is omitted.

  When the user designates a log collection target function (here, described as “function K1”), the collection target designation receiving unit 103 receives the designation (# 200).

  The collection target function setting unit 105 receives the function name of the function K1 from the collection target designation receiving unit 103, and obtains the start address and range of the function K1 in the memory space based on the symbol information on the program (# 201). . Furthermore, the instruction indicated by the obtained start address (herein referred to as “instruction M3”) is saved in the instruction save area (# 202), and the contents of the start address are replaced with a trap instruction (# 203). . As a result, the start address is set to the program stop position.

  When a program including the function K1 (herein referred to as “program P2”) is started and the process reaches the function K1, the trap instruction stored at the start address of the function K1 is executed (# 204). . As a result, the program information collection unit 107 is activated and starts operating. When starting the operation, the program information collecting unit 107 temporarily stops the processing of the program P2. Then, program information at that time is collected (# 205).

  The history information generation unit 108 receives the program information collected by the program information collection unit 107, and performs processing for generating history information DR using the program information. In this process, the history information generation unit 108 instructs the count value acquisition unit 101 to acquire a count value. Based on the instruction, the count value acquisition unit 101 accesses the counter unit 17 and acquires a count value (# 206).

  At this time, the counter unit 17 increases the count value held by itself by one. As a result, when accessed next time, a count value that is one greater than the count value output this time is output.

  The history information generation unit 108 generates history information DR using the count value acquired by the count value acquisition unit 101 and the program information received from the program information collection unit 107.

  The history information output unit 110 outputs and stores the history information DR generated by the history information generation unit 108 in an area in the memory 12 provided for the corresponding CPU 11 (# 207).

  Thereafter, the program information collection unit 107 executes the instruction M3 stored in the instruction saving area (# 208), and sets the instruction M3 to the original address of the memory space, that is, the address where the trap instruction is stored. Return (# 209).

  Furthermore, the program information collection unit 107 instructs the collection target function setting unit 105 to set the next program stop position. Based on the instruction, the collection target function setting unit 105 obtains the address of the instruction (next instruction) to be executed next to the instruction M3 in the function K1.

  The address of the next instruction is normally the address next to the address of the current instruction, but in the case of a branch instruction or the like, the register information is referred to determine the address of the next instruction. When the current instruction is a function call instruction, the address is executed when returning from the function call.

  When the address of the next instruction (herein referred to as “instruction M4”) is obtained (Yes in # 210 and # 211), the instruction M4 is saved in the instruction saving area (# 212), and the address The contents are replaced with a trap instruction (# 213).

  When the processing up to step # 213 is completed, the processing of the program P2, which is the program body, is continued. When the processing of the instruction M4 is started, the trap instruction stored at the address of the instruction M4 is executed in step # 213 (# 204). Then, processing from step # 205 to step # 209 for the instruction M4 is performed. Thereafter, in step # 210, the address of the next instruction is further determined, and the same processing is repeated.

  If the function K1 is called from another function, when the execution of the function K1 ends and the function returns to the calling function, the instruction at the head address of the function K1 is saved in the instruction saving area and trapped. Replace with an instruction. This is because the program information collecting unit 107 is operated when the function K1 is operated next time.

  Through the above processing, history information DR within the range of the function K1 designated by the user is accumulated, and information that can trace the operation of the function K1 can be obtained.

In the second method, a plurality of functions may be designated as log collection targets. Also in this case, a unique and sequential count value is associated with each history information DR in the order in which the history information DR is output (instruction execution order). The execution order of instructions and the CPU 11 that executed them can be specified.
[Third method]
FIG. 12 is a flowchart for explaining the flow of processing for collecting history information DR by the third method, and FIG. 13 is a diagram for explaining processing for setting and generating a break. Here, a case will be described in which the history information DR is collected only when data arbitrarily designated by the user is accessed. Note that detailed description of the same processing as that described in [First Method] and [Second Method] is omitted.

  When the user designates log collection target data (herein described as “data D1”), the collection target designation receiving unit 103 receives the designation (# 300).

  The collection target data setting unit 106 receives the name (data name) of the data D1 from the collection target designation receiving unit 103, and as shown in (i) of FIG. 13, “read break” and “write break” are performed on the data D1. "Is set (# 301).

  The “read break” is a function that breaks (suspends) a program that is being executed when the set data is about to be read. The “write break” is a function for causing a program that is being executed to break when a write (update) to the set data is attempted. In this embodiment, when a break occurs due to a read break or a write break, the program information collection unit 107 is activated and starts operating. Alternatively, when the data D1 is accessed, the processing of the program being executed by the collection target data setting unit 106 may be interrupted, and the program information collection unit 107 may be activated. By setting the read break and the write break, the address of the data D1 is set at the program stop position.

  If there is an access for reading or writing to data for which a read break and a write break are set during the execution of the program, a break occurs (# 302). As a result, the program information collection unit 107 is activated ((ii) in FIG. 13).

  When the operation starts, the program information collection unit 107 collects program information at that time (# 303).

  The history information generation unit 108 receives the program information collected by the program information collection unit 107, and performs processing for generating history information DR using the program information. In this process, the history information generation unit 108 instructs the count value acquisition unit 101 to acquire a count value. Based on the instruction, the count value acquisition unit 101 accesses the counter unit 17 and acquires the count value (# 304).

  At this time, the counter unit 17 increases the count value held by itself by one. As a result, when accessed next time, a count value that is one greater than the count value output this time is output.

  The history information generation unit 108 generates history information DR using the count value acquired by the count value and the program information received from the program information collection unit 107.

  The history information output unit 110 outputs and stores the history information DR generated by the history information generation unit 108 in an area in the memory 12 provided for the corresponding CPU 11 (# 305).

  Thereafter, the reading or writing process for the original data D1 is performed as usual, and the subsequent process is continued (# 306). When the data D1 is accessed again during the execution of the process, the processes after step # 302 are performed.

  It is also possible to specify a plurality of data for log collection. In that case, a read break and a write break are set for the plurality of pieces of data by the collection target data setting unit 106. Then, a break occurs when any of these data is accessed, whereby the processing from step # 302 is performed.

  Through the above processing, the history information DR is accumulated when referring to or updating the data D1 designated by the user, and information that can trace the processing on the data D1 can be obtained.

  Note that the user may be able to specify whether to set a read break or a write break.

  By the way, when a problem occurs in a computer having a plurality of CPUs, in order to pursue an operation (or process) that causes the problem, a program running history is collected for each CPU, and executed instructions are spread over a plurality of CPUs. Need to be traced in time series.

  According to the present embodiment, program information (register information and process information) effective for analysis when a problem occurs (failure analysis) is output. In addition, since the history information output processing unit 15 is provided for each CPU 11 and one counter unit 17 is provided for all the CPUs 11, a unique and sequential count value can be obtained for each program information to be output. Is granted. Thereby, program information is accumulated as time-series history information in units of execution instructions. That is, it is possible to collect information that can specify the execution order of instructions and the CPU 11 that has executed the instructions.

  In some cases, key information (functions, data, etc.) when analyzing a problem that has occurred can be estimated in advance from the contents of an existing dump file. In such a case, if program information is collected for all executed instructions, information that is already known to be irrelevant to the analysis will be output, which wastes memory space (resources). End up. Moreover, there is a possibility that the analysis work may be hindered because the information necessary for the analysis is truly lost.

  In order to reduce the amount of information to be collected, it is conceivable to use a timer or the like and perform collection at regular time intervals. However, in this method, there is a case where necessary information is lost because the collection is not performed at the time when the collection should be performed. This will hinder the analysis work.

  According to the second or third method described above, a function or data that requires collection of program information for analysis is designated in advance by the user, and program information is collected in units of each instruction or data in the function. The By collecting program information for only necessary parts in this way, it is possible to prevent resources from being wasted while collecting information necessary for analysis. Also, efficient tracing is possible.

  Further, according to the present embodiment, the user himself / herself is a user environment such as a high-speed medium, a large-capacity medium, or a medium of another device via a communication line (network) as an output destination of the history information DR. Various locations (recording areas) can be designated. Therefore, the history information DR can be output to an appropriate place according to the resource usage status and the convenience is improved.

  Furthermore, according to the present embodiment, information necessary for analysis when a problem occurs can be collected by software, so that it is not necessary to add special hardware, and costs can be reduced.

  That is, according to this embodiment, the analysis work at the time of the problem occurrence can be dramatically improved, and the possibility that the problem that could not be solved can be solved. Furthermore, by narrowing down the information to be collected, speeding up and efficiency can be realized, and in addition, the efficiency of analysis work can be improved.

  FIG. 14 is a diagram for explaining processing for setting a program stop position when the first to third methods are used in combination, and FIG. 15 is a flowchart for explaining the flow of processing for collecting history information according to another example. It is.

  In this embodiment, although the case where the 1st method, the 2nd method, and the 3rd method were performed individually was demonstrated, these can also be used together. That is, when the program P4, the function K2, and the data D2 are designated as the log collection target, as shown in FIG. 14, the trap instruction is stored at the start address of the program P4 and the function K2 by the collection point setting unit 104, A read break and a write break are set for data 2. The history information DR is collected by the first method triggered by the execution of the trap instruction stored at the head address of the program P4. The history information DR is collected by the second method triggered by the execution of the trap instruction stored at the head address of the function K2. Further, history information DR is collected by the third method triggered by access to data D2.

  In the present embodiment, a computer 1 having a plurality of CPUs is used, but a computer having a single CPU may be used. In that case, only one history information output processing unit 15 may be provided.

  Further, the count value may not be included in the history information DR, but may be output and stored together with the history information DR at a specified output destination.

  In the first method and the second method, the program stop position is set only for an instruction designated in advance by a user or the like, and thereby only the history information DR for the designated instruction is collected. Good.

  Alternatively, the history information DR may be collected as follows. All instructions to be collected in the history information DR are saved in advance in the instruction saving area (# 400 in FIG. 15A), and each is replaced with a trap instruction (# 401). Then, when the program is started (# 500 in FIG. 15B), a process for collecting history information DR at the location of each trap instruction replaced in step # 401, that is, steps from step # 104 in FIG. The processing up to # 109 or the processing from step # 204 to step # 209 in FIG. 11 is executed. Note that each step shown in FIG. 15A may be manually performed by the user himself / herself.

  In the present embodiment, in the first method and the second method, the history information DR for a certain instruction is collected and then the program stop position is set for the next instruction. However, the program stop position is set. May be performed in advance for all instructions for which the history information DR is collected.

  The history information generation unit 108 that supports history management by generating the history information DR can be referred to as a “history management unit”. The history information output unit 110 that once holds and outputs data about the history can be referred to as a “history holding unit”.

  In addition, the configuration and function of each unit of the computer 1, the contents indicated by each data, the contents or order of processing, and the like can be appropriately changed in accordance with the spirit of the present invention.

In the embodiment described above, the following notes are also disclosed.
(Appendix 1)
A computer system having a plurality of CPUs,
An acquisition point setting means for sequentially setting an acquisition point that triggers the start of an acquisition process for acquiring program information including the CPU register value and process information during execution of the program;
Program information acquisition means for performing the acquisition processing each time the processing of the program comes to the location of the acquisition point;
A counter that outputs an incremented or decremented count value each time it is accessed;
Each time the acquisition process is performed by the program information acquisition unit, the count value acquisition unit that accesses the counter and acquires the count value;
Program information storage means for storing the program information acquired in the acquisition process together with the count value corresponding to the program information acquired by the count value acquisition means;
A computer system comprising:
(Appendix 2)
The program information acquisition unit and the count value acquisition unit are provided for each CPU,
The counter is provided so as to be commonly accessed from each count value acquisition means provided for each CPU.
The program information acquisition means acquires the register value from the corresponding CPU.
The computer system according to appendix 1.
(Appendix 3)
An acquisition target receiving means for receiving an acquisition target of the program information designated by a user;
In-function setting means for sequentially setting the acquisition points in the function when the acquisition target received by the acquisition target receiving means is a function;
The computer system according to Supplementary Note 1 or 2, further comprising:
(Appendix 4)
When the acquisition target received by the acquisition target reception means is data, the program has break setting means for performing settings for causing the program to break when the data is accessed during execution of the program,
The program information acquisition means performs the acquisition process when a break occurs during the execution of the program.
The computer system according to attachment 3.
(Appendix 5)
A storage destination receiving means for receiving a storage destination of the program information designated by the user;
The program information storage unit stores the program information in the storage destination received by the storage destination reception unit.
The computer system according to any one of appendices 1 to 4.
(Appendix 6)
A computer system having a plurality of CPUs,
Break setting means for making a setting to break the program when predetermined data is accessed during execution of the program;
Program information acquisition means for performing an acquisition process for acquiring program information including a register value of the CPU and process information when a break occurs during execution of the program;
A counter that outputs an incremented or decremented count value each time it is accessed;
Each time the acquisition process is performed by the program information acquisition unit, the count value acquisition unit that accesses the counter and acquires the count value;
Program information storage means for storing the program information acquired in the acquisition process together with the count value corresponding to the program information acquired by the count value acquisition means;
A computer system comprising:
(Appendix 7)
The acquisition point that triggers the start of an acquisition process for acquiring program information including the register value and process information of the CPU during execution of the program in a computer system equipped with a plurality of CPUs is set for the program. One step,
A second step of performing the acquisition process when the processing of the program comes to the location of the acquisition point;
A third step of obtaining a count value by accessing a counter that outputs a count value that is incremented or decremented each time access is made, when the acquisition process is performed at the acquisition point;
A fourth step of storing the program information acquired in the acquisition process together with the count value corresponding to the program information acquired in the third step;
The second step, the third step, and the fourth step are performed for each acquisition point that is sequentially set in the program by the processing of the first step.
The program information collection method characterized by the above-mentioned.
(Appendix 8)
To a computer with multiple CPUs,
A first step of setting, for the program, an acquisition point that triggers the start of an acquisition process for acquiring program information including the register value and process information of the CPU during execution of the program;
A second step of performing the acquisition process when the processing of the program comes to the location of the acquisition point;
A third step of obtaining a count value by accessing a counter that outputs a count value that is incremented or decremented each time access is made, when the acquisition process is performed at the acquisition point;
Storing the program information acquired in the acquisition process together with the count value corresponding to the program information acquired in the third step; and
At that time,
Causing the second step, the third step, and the fourth step to be executed for each of the acquisition points sequentially set in the program by the processing of the first step;
A computer program characterized by the above.

It is a figure which shows the example of the hardware constitutions of the computer concerning this invention. It is a figure which shows the process stored in the memory space. It is a figure which shows the example of a functional structure of a computer. It is a figure for demonstrating the process of process information acquisition. It is a figure which shows the example of a structure of a log | history information output process part. It is a figure for demonstrating the process of register information acquisition. It is a flowchart for demonstrating the flow of a process of collection of the historical information by a 1st method. It is a figure for demonstrating the process of a program stop position setting. It is a figure which shows the example of log | history information. It is a figure which shows the log | history information output about each CPU. It is a flowchart for demonstrating the flow of the process of the collection of the history information by a 2nd method. It is a flowchart for demonstrating the flow of the process of collection of the historical information by a 3rd method. It is a figure for demonstrating the process at the time of the setting and generation | occurrence | production of a break. It is a figure for demonstrating the process of a program stop position setting at the time of using together the 1st to 3rd method. It is a flowchart for demonstrating the flow of the process of the collection of the history information by another example.

Explanation of symbols

1 Computer (computer system, computer)
11 CPU
17 Counter section (counter)
101 Count value acquisition unit (count value acquisition means)
103 Collection target designation reception unit (acquisition target reception means)
104 Collection point setting section (acquisition point setting means)
105 Collection target function setting section (acquisition point setting means, function setting means)
106 Collection target data setting section (break setting means)
107 Program information collection unit (program information acquisition means)
110 History information output unit (program information storage means)
D1, D2 data (obtained)
K1, K2 function (program, acquisition target)
P1, P2 program (obtained)

Claims (6)

A computer system having a plurality of CPUs,
An acquisition point setting means for sequentially setting an acquisition point that triggers the start of an acquisition process for acquiring program information including the CPU register value and process information during execution of the program;
Program information acquisition means for performing the acquisition processing each time the processing of the program comes to the location of the acquisition point;
A counter that outputs an incremented or decremented count value each time it is accessed;
Each time the acquisition process is performed by the program information acquisition unit, the count value acquisition unit that accesses the counter and acquires the count value;
Program information storage means for storing the program information acquired in the acquisition process together with the count value corresponding to the program information acquired by the count value acquisition means;
A computer system comprising: The program information acquisition unit and the count value acquisition unit are provided for each CPU,
The counter is provided so as to be commonly accessed from each count value acquisition means provided for each CPU.
The program information acquisition means acquires the register value from the corresponding CPU.
The computer system according to claim 1. An acquisition target receiving means for receiving an acquisition target of the program information designated by a user;
In-function setting means for sequentially setting the acquisition points in the function when the acquisition target received by the acquisition target receiving means is a function;
The computer system according to claim 1, comprising: A computer system having a plurality of CPUs,
Break setting means for making a setting to break the program when predetermined data is accessed during execution of the program;
Program information acquisition means for performing an acquisition process for acquiring program information including a register value of the CPU and process information when a break occurs during execution of the program;
A counter that outputs an incremented or decremented count value each time it is accessed;
Each time the acquisition process is performed by the program information acquisition unit, the count value acquisition unit that accesses the counter and acquires the count value;
Program information storage means for storing the program information acquired in the acquisition process together with the count value corresponding to the program information acquired by the count value acquisition means;
A computer system comprising: The acquisition point that triggers the start of an acquisition process for acquiring program information including the register value and process information of the CPU during execution of the program in a computer system equipped with a plurality of CPUs is set for the program. One step,
A second step of performing the acquisition process when the processing of the program comes to the location of the acquisition point;
A third step of obtaining a count value by accessing a counter that outputs a count value that is incremented or decremented each time access is made, when the acquisition process is performed at the acquisition point;
A fourth step of storing the program information acquired in the acquisition process together with the count value corresponding to the program information acquired in the third step;
The second step, the third step, and the fourth step are performed for each acquisition point that is sequentially set in the program by the processing of the first step.
The program information collection method characterized by the above-mentioned. To a computer with multiple CPUs,
A first step of setting, for the program, an acquisition point that triggers the start of an acquisition process for acquiring program information including the register value and process information of the CPU during execution of the program;
A second step of performing the acquisition process when the processing of the program comes to the location of the acquisition point;
A third step of obtaining a count value by accessing a counter that outputs a count value that is incremented or decremented each time access is made, when the acquisition process is performed at the acquisition point;
Storing the program information acquired in the acquisition process together with the count value corresponding to the program information acquired in the third step; and
At that time,
Causing the second step, the third step, and the fourth step to be executed for each of the acquisition points sequentially set in the program by the processing of the first step;
A computer program characterized by the above.

Images