JPH02165352A - Dump processing system - Google Patents

Abstract

PURPOSE:To execute rapid dump processing by extracting only an area specified by the contents of a register in a CPU at the time of accessing a program in a dump file. CONSTITUTION:When an abnormality detecting processing part 13 detects the generation of abnormality, a fault information extracting processing part 15 extracts an area corresponding to the contents of the register in the CPU at the time of generating the abnormality and an area corresponding to the contents of the register in the CPU receded to a register receding area 21 in each program working area 20. A file storage processing part 16 dumps the extracted program part in the dump file 2 as information for fault inspection. Since only the minimum information necessary for fault inspection is dumped, the dump processing can be rapidly executed.

Description

[Detailed Description of the Invention] [Summary] The object of the present invention is to enable high-speed collection processing regarding a dump processing method for collecting data necessary for failure investigation when an abnormality occurs during program execution. The work area required for program execution is configured to be provided according to the stack format in the order of called programs, and when the next program is called, the work area of the program is placed in the work area of the calling program. a register save processing unit that saves the current contents of registers prepared for execution; a pointer setting processing unit that places a pointer between a work area of a called program and a work area of a called program; When an error occurs in a program, the program part corresponding to the contents of the register at the time of the error is collected in a dump file, and the contents of the register saved in accordance with the pointer set by the pointer setting processing section are read out and the contents are updated. The configuration is such that a dump processing unit that processes the corresponding area to be collected into a dump file can be used.

[Industrial application field]

The present invention relates to a dump processing method for collecting data necessary for fault investigation when an abnormality occurs in program execution, and particularly relates to a dump processing method that enables high-speed collection processing.

In a data processing device, a dump process is executed to investigate a program abnormality that has occurred.

Various levels of processing methods have been proposed for this dump processing depending on the nature of the program and the purpose of fault investigation, but simple dump processing methods are often adopted to speed up the collection process. In such a simple dump processing method, it is necessary to configure the system so that the necessary data can be collected as quickly as possible.

[Conventional technology]

In conventional dump processing methods, when an abnormality occurs during program execution, the status of resources such as the contents of the work area owned by the task and the program modules used are all collected in a dump file. Ta. In other words, as shown in Figure 7, when an abnormality is detected, the contents of all work areas owned by that task are written to a dump file by following the task work area management queue. At the same time, by following the program management queue, all program modules used by the task were written to a dump file.

It is something that

[Problem to be solved by the invention]

However, with such conventional technology, when a large number of programs are called to execute a task, the amount of data to be written to the dump file increases, so it is necessary to prepare a large capacity dump file. In addition to this, there was a problem in that it took a long time to collect data.

The present invention has been made in view of the above circumstances, and it is possible to investigate program failures by finding the program part specified by the contents of the CPU register at the time the program is called. Focusing on the fact that technology dumps even redundant information that is not used for fault investigation, only the areas specified by the contents of the CPU's registers at the time the program is called are collected in the dump file. Therefore, it is an object of the present invention to provide a dump processing method that realizes high-speed dump processing.

In the figure, l is a data processing device equipped with the present invention, 2 is a dump file that stores information collected for failure investigation, and lO is a program execution unit that executes a program. , a register save processing section 11, a pointer setting processing section 12, an abnormality detection processing section 13, and a dump processing section 14, 20 is a work area used for executing the program, and 20 is a work area used for executing the program. They are configured to be stacked in order in a stack format. This work area 20
is the register save area 21 and the backward pointer area 2.
2.

When the next program is called, the register save processing section 11 processes to save the contents of the register of the CPU at the time of the call to the register save area 21 of the work area 20 of the calling program, and the pointer setting processing section 12 By setting the address information of the calling program's work area 20 in the backward pointer area 22 of the calling program's work area 20, The abnormality detection processing unit 13 performs processing to detect the occurrence of an abnormality during execution of the program, and the dump processing unit 14 performs processing to place a pointer between the two when an abnormality occurs in the program.
The area corresponding to the contents of the register of the CPU at the time of the abnormality is sampled, and the CPU of the register save area 21 is read out by the pointer of the bank word pointer area 22.
a fault information collection processing unit 15 that performs processing to collect an area corresponding to the contents of the register;
and a file storage processing unit 16 that processes the information collected for fault investigation to be stored in the dump file 2.

[Effect]

In the present invention, when the next program is called, the register save processing unit 11 saves the CP at the time of the call.
The contents of the register U are saved to the register save area 21 of the work area 20 of the caller, and then the pointer setting processing unit 12 saves the contents of the register U to the buffer quote pointer area 22 of the work area 20 of the called program that has been stacked up. Address information of the calling work area 20 is set. When the program execution unit 10 executes the program while executing such processing, when the abnormality detection processing unit 13 detects the occurrence of an abnormality, the failure information collection processing unit 15
An area corresponding to the contents of the CPU register at the time of the abnormality occurrence and a register save area 21 in the work area 20 of each program
The area corresponding to the content of the register of the CPU to be saved is sampled, and the file storage processing unit 16 processes the sampled program portion to be dumped into the dump file 2 as information for fault investigation.

In this way, in the present invention, only the minimum amount of information necessary for fault investigation is dumped, so the capacity of the dump file 2 can be reduced, and the dump processing can be performed at extremely high speed. This will make it possible to achieve this goal.

〔Example〕

Hereinafter, the present invention will be explained in detail according to examples.

FIG. 2 shows a system configuration for realizing the present invention. As shown in this figure, the system configuration of the present invention is the same as that of the prior art, including a central processing unit 30 having, for example, 16 register groups 31, and a main memory in which execution programs and work areas are expanded. 32, a storage control device 33 that controls data flow, an external storage device 34 that serves as an auxiliary storage device or a storage device for dump files, an input/output device 35, and an input/output control device 36. It turns out.

FIG. 3 shows a method of configuring the work area 20 of the present invention. As shown in FIG. 3 <a>, assume an example in which program A calls program B in accordance with a "CALL command", and program B calls program C in accordance with the "CALL command". In this way, when the programs are sequentially called, the present invention has a method as shown in FIG. 3 (b).
), a work area for program C to be called next is set above the work area for program A, and a work area for program C to be called next is set above the work area for program C. In this way, the work area 20 for each program is set up by piling up the work area 20 for each program in a stack format according to the order of the programs to be called.

Next, the dump processing method of the present invention that will be executed using the work area 20 configured as described above will be explained according to the flowcharts shown in FIGS. 4 and 5. Here, the flowchart in FIG. 4 shows the processing that is executed when there is no abnormality in the program, and the flowchart in FIG. 5 shows the processing that is executed when there is an abnormality in the program.

First, the flowchart in FIG. 4(a) will be explained. When the next program is called without any abnormality in the program, the called program will be placed in the stack pointer register ( In the work area 20 of the calling program, which is the area pointed to by (sometimes abbreviated as "SPR" in the figure),
A process is performed to save the contents of the register group 31 at the time of the call.The contents of the register group 31 to be saved in this way are saved to the register save area 21 shown in ■ in FIG. 4(b). That will happen.

Next, in step 12, update processing is performed so that the contents of the stack pointer register point to the start address of the work area 20 of the called program.
The size of the work area 20 of the called program, shown schematically in ■ in Figure (b), has been determined by the compiler, so this process is executed by obtaining information about that size. Next, in step 13, the contents of the updated stack pointer register are stored in the forward pointer area 23 of the work area 20 of the calling program, and finally, in step 14, the contents of the stack pointer register before the update are stored. It is stored in the backward pointer area 22 of the work area 20 of the called program. The processing of steps 13 and 14 is shown in (1) and (2) in FIG. 4(b).

In this way, according to the flowchart in FIG. 4, if there is no abnormality in the program, the contents of the register group 31 at the time of the call are saved in the work area 20 of the calling program, and the contents of the register group 31 at the time of the calling program are saved. A pointer is placed between the program's work area 20 and the called program's work area 20.

The called program then executes a predetermined process using the set work area 20, and when the next program is newly called, the same process will be executed.

Next, the flowchart of FIG. 5(a) will be explained. When an abnormality in the program is detected, as shown in step 21, for example, one page before and after the page in the area pointed to by the contents of the register group 31 at the time of the abnormality is collected as data to be dumped and written to the dump file 2. , this process is executed for all the registers included in the register group 31. Next, in step 22, the work area 20 of the program pointed to by the stack pointer register at the time of the abnormality occurrence (in the figure, the work area 20 of each program is
This process of determining the address of the work area 20 of the program that is the calling source for the program from the contents stored in the backward pointer area 22 of the backward pointer area 22 of the program (hereinafter referred to as "stack" for simplicity) is shown in FIG. b) ■
Diagrammatically shown.

Next, in step 23, for example, one page before and after the page of the area pointed to by the contents of the register saved in the register save area 21 of the work area 20 of the identified program is collected as data to be dumped, and the dump file 2 is created. The register saved in the register save area 21 shown in (2) in FIG. Next, in step 24, from the contents stored in the buffer quote pointer area 22 of the work area 20 of the program to be processed in step 23, a process is performed to obtain the address of the work area 20 of the program that is the caller of the program. . This process is shown in (■) in FIG. 5(b). Then, in the next step 25, it is determined whether or not the work area 20 of the program for three levels set in advance has been returned, and if it has returned, the process is terminated, and conversely,
If it has not returned, the process returns to step 23.

In this way, the work area 20 of the program for three levels is
The area pointed to by the contents of the register saved in is dumped to dump file 2. The reason why we stopped at three levels is because empirically, for normal failures, it is possible to find the cause by going back this many times.
Furthermore, this number may be increased, or
It may be arranged to return to the work area 2o of the program that is the first calling source.

FIG. 6 shows the overall flow of the processing in the flowcharts of FIGS. 4 and 5 described above, using FIG. 3(a) as a specific example. In the figure, 13 is the abnormality detection processing section explained in FIG. 1, and 15 is the failure information collection processing section also explained in FIG. 1.

Although the illustrated embodiment has been described above, the present invention is not limited thereto.For example, the page length of the program to be dumped is not limited to a total of three pages, one page before and one page before. , from the point of view of dump processing,
It is not necessary to provide a forward pointer.

〔Effect of the invention〕

As described above, in the present invention, program fault investigation can be realized by finding the area specified by the contents of the CPU register at the time of program call, and the redundant area that is not used for fault investigation in the conventional technology can be realized. Focusing on the point that even the most detailed information was dumped, we realized high-speed dump processing by collecting only the program parts specified by the contents of the CPU's registers when the program was called into a dump file. This makes it possible to

Furthermore, since unnecessary information is not collected, the size of the dump file that must be prepared can be reduced.

[Brief explanation of the drawing]

Figure 1 is a diagram of the principle configuration of the present invention, Figure 2 is a system configuration diagram of the present invention, Figure 3 is an explanatory diagram of the method of configuring the work area of the present invention, and Figures 4 and 5 are diagrams showing how the present invention is executed. FIG. 6 is an explanatory diagram of the processing flow of the present invention, and FIG. 7 is an explanatory diagram for explaining the prior art. In the figure, 1 is a data processing device, 2 is a dump file, lO
1 is a program execution unit, 11 is a register save processing unit, and 12 is a program execution unit.
13 is a pointer setting processing unit, 13 is an abnormality detection processing unit, 14 is a dump processing unit, 15 is a failure information collection processing unit, 16 is a file storage processing unit, 20 is a work area, 21 is a register save area, and 22 is a bank word pointer. Area 23 is a forward pointer area.

Claims (1)

[Claims] In a dump processing method for collecting data necessary for failure investigation into a dump file (2) when an abnormality occurs during program execution, the data is necessary for program execution. The work area (20) is configured to be provided according to the stack format in the order of called programs, and when the next program is called, the work area (20) of the calling program is prepared for program execution. A register save processing unit (11) that saves the contents of the register at that point in time, and a pointer setting that places a pointer between the work area (20) of the called program and the work area (20) of the called program. The processing unit (12), when an abnormality occurs in the program, collects the program part corresponding to the contents of the register at the time of the abnormality into the dump file (2),
The content of the register saved according to the pointer set by the pointer setting processing unit (12) is read out, and the area corresponding to this content is saved in the dump file (2).
A dump processing method characterized by comprising: a dump processing unit (14) that performs processing to collect data.