JPH0926898A - Device and method for detection of program error - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect a program error by utilizing the one function of a paging means to be originally used for realizing a virtual storage system. SOLUTION: The source program prepared by a C language is translated into an object program and is executed. If a program error exists in the source program at the time, the effect is displayed on a CRT display device 60. In this detection, a paging unit 13 is utilized. Namely, when the initialization of a pointer value is not performed, a state that a virtual address is not allocated to all the parts of a physical address is generated, a page fault is generated and a system program is called. In this case, because the generation of a reading command or a writing command is performed for a set error detection area without fail, the correspondened processing at the time of a prescribed error detection is to be executed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an error in which a pointer variable is made an automatic variable and a pointer value is forgotten to be initialized when a source program that takes an automatic variable in a stack area is created like C language or Pascal language. (Less than,
It is simply called a program error. ) For detecting a program error and a method for detecting the same.

[0002]

2. Description of the Related Art A program error detecting device of this kind is already known. In a conventional program error detection device, when a source program is translated into an object program, the program error is detected by analyzing the syntax of the source program.

In Japanese Patent Application No. 5-253082, an attempt is made to detect a program error using a hardware error detection circuit. In other words, a translation means for translating the source program into an object program, and an automatic initialization that automatically adds an instruction to set the pointer value of the pointer variable to a predetermined address value in the object program when the pointer variable is an automatic variable. And an error detection circuit that detects read / write to a memory area in a predetermined range including the address value set by the automatic initialization means and outputs an error signal when the object program is executed, And a notification means for notifying the operator that the pointer value of the pointer variable in the source program has been forgotten to be initialized according to the error signal. In this case,
Since the detection of the program error is performed by the error detection circuit without relying on the above-mentioned syntax analysis, the detection speed can be improved and the reliable detection can be realized.

[0004]

By the way, there is known a paging means having an address conversion function in order to realize a so-called virtual memory system. The paging means for this virtual storage system has an address translation function of translating a virtual address into a physical address in units of a predetermined page based on a page table, and when the virtual address is not assigned anywhere in the physical address, It also has an exception handling request function that generates a page fault and calls a system program. For example, among the 8086 CPUs manufactured by Intel, 80386
More versions (80386, 80486, Pent
The CPU of (i.m.) has a built-in paging means and is called a paging unit.

In the above-mentioned Japanese Patent Application No. 5-253082, the program error is detected based on the error signal output from the error detection circuit. However, even if the page fault generated by the paging means is used, The applicant considered that a program error could be detected. If a program error can be detected by using one function of the paging means originally used for realizing such a virtual memory system, there is no need to add special hardware, which is very advantageous.

However, for example, in the above-mentioned paging means having a built-in CPU, it can be controlled only by a special system program, and the paging means cannot be used by a general program. Generally, the right to control the paging means is held by a system program such as an OS (operating system) and cannot be directly controlled by a general application.

In view of the above, the present invention intends to perform a predetermined process from a general application on a system program having the right to control the paging means, and use one function of the paging means to detect a program error. To aim.

[0008]

Means for Solving the Problems and Effects of the Invention
The invention described in (1) is a device for detecting a program error that forgets to initialize the pointer value while making the pointer variable an automatic variable when a source program that takes an automatic variable in the stack area is created. A compiler having a translation function for translating into a program and an automatic initialization function for automatically adding an instruction for setting a pointer value of a pointer variable to a predetermined address value to an object program when the pointer variable is an automatic variable And an address translation function that translates a virtual address into a physical address on a page-by-page basis based on a page table, and generates a page fault if the virtual address is not assigned anywhere in the physical address. Exception handling request function to call Paging means for a virtual memory system and advance setting of an error detection area address and a corresponding process at the time of error detection for a system program at the start of the object program translated by the compiler or before the program starts. When the cause of the page fault generated by the paging means at the time of executing the processing execution means and the object program is due to reading / writing from / to the error detection area, a predetermined value set by the preprocessing execution means. And an error detection time response process execution means for executing the error detection time response process.

A second aspect of the present invention is the program error detecting apparatus according to the first aspect, wherein the paging means is built in a CPU. Claim 3
In the program error detection device according to claim 1 or 2, the error detection response processing execution means is a predetermined forced termination set as a part of the object program translated by the compiler. The object program itself is configured to be forcibly terminated based on a processing routine.

According to a fourth aspect of the present invention, in the program error detection apparatus according to the first or second aspect, the error detection response processing execution means is a part of the object program translated by the compiler or a system program. Based on a predetermined error occurrence notification processing routine that is set as a part of itself, executes a predetermined processing that notifies the operator of a program error occurrence due to forgetting to initialize the pointer value of the pointer variable in the source program. It is characterized by being configured as follows.

According to a fifth aspect of the invention, a translation function for translating a source program that takes automatic variables in a stack area into an object program, and a pointer of a pointer variable to an object program when the pointer variable is an automatic variable A compiler with an automatic initialization function that automatically adds an instruction to set a value to a predetermined address value, and an address that converts an address in the virtual address space into an address in the physical address space in predetermined page units A virtual memory system having a translation function and an exception handling request function for invoking a system program by generating a paging fault when the virtual address does not exist in the memory; , Pointing the pointer variable as an automatic variable A method for detecting a program error that forgets to initialize a value, which is performed when the object program translated by the compiler is started or before the program is started, when the address of the error detection area is set and the error is detected in the system program. If the cause of the paging fault generated by the paging means at the time of execution of the object program is due to reading / writing to the error detection area, the corresponding processing is set by the preprocessing execution means. It is characterized in that the specified predetermined error response processing is executed.

According to a sixth aspect of the present invention, in the program error detection method according to the fifth aspect, the predetermined processing set as a part of the object program translated by the compiler is performed as the error detection response processing. The object program itself is forcibly terminated based on a forced termination processing routine.

According to a seventh aspect of the present invention, in the program error detecting method according to the fifth aspect, a part of the object program translated by the compiler or a part of the system program itself is used as the error detection response processing. Based on a predetermined error occurrence notification processing routine set as, the operator is notified of a program error occurrence caused by forgetting to initialize the pointer value of the pointer variable in the source program.

In the program error detection device of the present invention,
In the compiler, the source program that takes an automatic variable in the stack area is translated into an object program by the translation means. However, if the source program uses a pointer variable as an automatic variable, the pointer value of the pointer variable is added to the object program. Is automatically added by the automatic initialization means.

Then, at the start of the object program translated by the compiler or before the start of the program, the preprocessing execution means sets the address of the error detection area and the corresponding processing at the time of error detection for the system program. . On the other hand, the paging means is for a virtual storage system, has an address conversion function of converting a virtual address into a physical address in units of a predetermined page based on a page table, and the virtual address is allocated to any physical address. If not, it has an exception handling request function to generate a page fault and call a system program.

Therefore, when the cause of the page fault generated by the paging means at the time of execution of the object program is due to the reading / writing of the error detection area set by the preprocessing execution means, the preprocessing execution means is also used. The error detection response processing execution means executes the predetermined error detection response processing set by the above.

If the pointer value of the pointer variable is initialized by the programmer (program), the error detection area set at the time of execution of the object program is not read or written, and the address set by the programmer is not used. Exception handling does not occur because the operation is performed on the value address.

However, when the pointer value is not initialized by the programmer, the virtual address is not allocated to any physical address,
A page fault occurs and the system program is called. In this case, the read command or the write command is always issued to the error detection area set as described above, so that the predetermined error detection response process is executed.

Even if the pointer value is not initialized, a page fault occurs and the system program is called when the virtual address is not assigned to any physical address. In this case, since the read command or the write command is generated outside the error detection area, normal page fault exception processing in the virtual memory system will be executed.

As described above, according to the program error detecting apparatus of the present invention, a general application performs a predetermined process on a system program having the right to control the paging means, and the program is executed by using one function of the paging means. Can detect errors. That is, since the detection of the program error is performed by using the paging means configured by hardware, not by the parsing,
Since it is possible to improve the detection speed and realize reliable detection, and to detect a program error by using one function of the paging means originally used for realizing the virtual memory system, special hardware is added. Is unnecessary.

It is conceivable that the paging means is built in the CPU as described in claim 2 or is connected by attaching a dedicated terminal for communicating with the paging means LSI to the CPU. However, when the CPU and the paging means are separately provided, for example, it is necessary for the maker of the CPU to set and manufacture the paging means LSI as a family chip.

As the error detection response processing execution means, as described in claim 3, based on a predetermined forced termination processing routine set as a part of the object program translated by the compiler,
The object program itself may be configured to be terminated forcibly. In this case, for example, if the instruction manual or the like describes that the cause of the forced termination is a program error, the operator can know that there is a program error by looking at the instruction manual.

According to the fourth aspect, the pointer in the source program is based on a predetermined error occurrence notification processing routine set as a part of the object program translated by the compiler or a part of the system program itself. It is configured to execute a predetermined process for notifying the operator of the occurrence of a program error due to forgetting to initialize the pointer value of the variable. Examples of the predetermined process to be notified in this case include, for example,
It is conceivable to display the fact that a program error exists on a display such as a CRT, sound a notification buzzer, or light a notification lamp.

Also in these cases, the detection speed can be improved and reliable detection can be realized, and the program error can be detected by utilizing one function of the paging means originally used for realizing the virtual memory system. Therefore, there is an effect that it is not necessary to add special hardware.

[0025]

Next, an embodiment of the present invention will be described. In the following description, the case where the program error detection apparatus according to the embodiment of the present invention is applied to a personal computer will be taken as an example.

FIG. 1 is a block diagram showing a schematic configuration of a control unit of a personal computer, which includes a CPU 10, a system I / O area 20, a RAM 30, and a VRAM 40.
Are connected to each other by an address bus AB and a data bus DB. Further, the VRAM 40 has a CRT display device 6
0 is connected.

The RAM 30 has a memory capacity of 000 as shown in FIG.
It has an area of 640 Kbytes from addresses 00H to 9FFFFH, and the area is MS-DOS.
A (registered trademark name of Microsoft Corporation) area 23, a code area 24 for storing a program, and a data area 26 for storing data such as a plurality of numerical values are provided, and the addresses indicating the areas 24, 26 are variable. Has been done. Further, the RAM 30 is also provided with a stack area 28 capable of taking automatic variables. The RAM 30 is of a large model type and is represented as a virtual address space in FIG. The difference between the virtual address and the physical address will be described later.

In the system I / O area 20, the 4K-byte area at addresses D5000H to D5FFFH is an area (error detection area) 32 used exclusively for error detection, which will be described later. The error detection area 32 cannot be used as the code area 24, the data area 26, the stack area 28, or the like. In other words, this area is never usable by some programmers.

The personal computer in this embodiment is a PC-9801 series manufactured by NEC Corporation, and the CPU 10 is an 8086 series C manufactured by Intel Corporation.
80386 or higher version with PU (80386,80
486, Pentium). The CPUs of version 80386 and above have a built-in paging unit 13 as a paging means, which will be described later, and all programs cannot read or write by directly specifying a physical address.

That is, part of the address output from the CPU core 11 in the CPU 10 (upper 20 bits of A31 to A12 in this example) is output to the paging unit 13. The paging unit 13 converts the address output from the CPU core 11 into another address by referring to the page table 15 and outputs the address to the address bus AB. Further, the lower 12 bits of A11 to A0 are directly output to the address bus AB without passing through the paging unit 13. All data output (D31 ~
0) Input / output directly to / from the address bus AB.

The page table 15 shows the correspondence between virtual addresses and physical addresses, and one entry exists for each virtual page. This page table 15
Some have a register inside the CPU, and some have a first address in the register on a memory. When the page table 15 is stored in the memory, the paging unit 13 automatically accesses the page table 15 at a timing that the CPU core 11 does not know. Further, when the page table 15 is provided on the memory, a cache memory for the page table is always prepared inside the CPU 10. In this case, normally 64 ~
A cache for 256 entries is prepared.

For example, virtual address 0000C000H
4K bytes of ~ 0000CFFFH is the physical address 2
If the virtual address 0C123H is accessed from the program if it is allocated to FEC7000H to 2FEC7FFFH, the physical address 2FEC7123H will be accessed.

The paging unit 13 makes the programs actually loaded at different addresses (physical addresses) appear to be loaded at the same addresses (virtual addresses) when the programs are operated by so-called multitasking. Is the hardware used for. The above processing becomes possible by rewriting the page table 15 when switching the task.

Then, the paging unit 1 of this embodiment
Reference numeral 3 is a paging unit compatible with the virtual storage system. Virtual memory is generally used in a large-scale multitasking OS, and is a method in which a hard disk or the like, which is an auxiliary storage device, is virtually regarded as a memory and the apparent memory is made larger than the mounted memory. In order to realize the virtual memory, a paging unit compatible with the virtual memory system is required, and the paging unit 13 of this embodiment is a paging unit compatible with the virtual memory system.

The paging unit 13 compatible with the virtual memory system has a function of generating a page fault exception handling request signal to the CPU core 11 in addition to the above basic functions. This exception process is similar to the interrupt process, but the interrupt process terminates the process currently being executed and then transfers control to the interrupt processing routine, while the exception process cancels the process currently being executed. Control is transferred to the exception handling routine. Therefore, in the interrupt processing, the execution is started from the next instruction that is interrupted, but in the exception processing, the execution is restarted from the instruction that was being executed when the exception occurred.

In this way, the CPU core 11 can control the paging unit 13 to read or set the allocation state of the virtual address to the physical address. However, the 80386 CPU has what is called a privilege level. The paging unit 13 can be controlled only by a program of privilege level "0". When using MS-DOS,
Since a normal program including DOS operates at the privilege level "3", the paging unit 13 cannot be controlled.

When MS-DOS is used, D
Since a normal program including the OS operates in a mode that emulates the operation of 8086 called virtual 86 mode, only 20 bits of the virtual address of 32 bits can be used. This is the command form of 8086 is 20
This is because only bit addresses can be generated. That is, a normal program including DOS can access only virtual addresses 00000000H to 000FFFFFH. Virtual address 001000
00H to FFFFFFFFH can be accessed only in a mode unique to 80386 (called a protect mode) that is not compatible with 8086, but MS-DOS has 8
Since it is a 086 OS, this is not supported.

However, in practice, the paging unit 13
Since the management of the
When operating the S-DOS, a system program for managing the paging unit 13 is prepared separately from the DOS. This system program operates in an 80386-specific mode that is incompatible with the 8086, and operates at the privilege level “0”.

What is important here is that since the paging unit 13 of this embodiment is compatible with the virtual storage system, the page does not exist, that is, the setting that "4 Kbytes with a virtual address is not allocated to any physical address" is set. It is possible. Description will be made with reference to the page table entry (PTE) of the 80386 CPU shown in FIG. The page table of the 80386 CPU is placed on the memory, and the position on the page table memory is determined by a register called CR3. At the time of task switching, only CR3 is changed to change the entire page table.

One page table entry as shown in FIG. 3 is prepared for each virtual address 4K. In the upper 20 bits of the page table entry, the upper 20 bits (A32 to A12) of the physical address are written. The upper 20 bits of the virtual address replaces the upper 20 bits of the page table entry and is output as a physical address to the outside of the CPU. The lower 12 bits of the page table entry is an area for storing various system information, of which the least significant bit (bit 0) is P
When the P bit is "1", the page is considered to exist and the memory or the like is normally accessed. When the P bit is "0", the page is considered to be absent.

When an instruction to access the virtual address of page fault (P bit is 0) is executed,
Before the instruction is executed, an exception processing request called “page fault” is generated from the paging unit 13 to the CPU core 11, and control is transferred to the system program that manages the paging unit 13. In this case, a process generally called swap is executed.

The lower 12 pages of the page table entry
Bits are areas that store various system information,
Although it has been described that the least significant bit is P bit, there are A bit, D bit, S bit and W bit in addition to the above.
The D bit is a bit that becomes "1" when written to the virtual address, the S bit is a bit that can be accessed only by the OS when this is "1", and the W bit is "1" here. Is a bit that can be read but not written.

The A bit is a bit which changes from "0" to "1" when the virtual address for the entry is accessed from the CPU core 11. CPU core 11
Periodically checks the A bit to find out which virtual address is frequently used. The A bit automatically changes from "0" to "1", but does not change from "1" to "0" until the program returns.

Since the CPU core 11 knows which virtual address is used a lot by the A bit, the virtual address having a low usage rate is written on the disk.
After saving the contents, the P bit of the page table is set to "0". As a result, the physical address used by that virtual address until now becomes unused, so the virtual address that the instruction that caused the page fault tried to access is assigned to that physical address. Then, the contents of the previously stored virtual address are read from the disk and the P bit is set to "1". This is the specific processing content of the swap processing. When returning from the exception handling routine, the instruction that caused the exception handling is executed again, but since the P bit of the virtual address to be accessed has changed to "1" this time, no page fault occurs.

The swap processing in the exception processing routine is O
Since the system program such as S performs the swap processing, the user program cannot recognize that the page processing occurs and the swap processing is being performed even if the swap processing is actually occurring. Therefore, it seems that the memory has increased, and this is the whole of the method called virtual memory.

In this embodiment, this function is utilized to start the program and to detect the error detection area 32 (D
Set the page absent setting (P bit is 0) in the area from 5000H to D5FFFH for 4K bytes, and set the error detection dedicated address and the jump destination when an error occurs to the system program when the program starts. Instruct. Conventionally, there is no system program having a function of receiving such an instruction.

As a result, when reading or writing is attempted to be executed in the error detection area 32, exception processing occurs before the instruction is executed, and control is transferred to the system program managing the paging unit 13. The system program investigates the address where the "page fault" occurred. This is because a "page fault" may occur in addition to the address in the error detection area 32. If the result of the investigation indicates that the address where the "page fault" occurred is an address within the error detection area 32, the address jumps to the address designated in advance.

At the jump destination, the fact that the pointer variable has been used without being initialized is displayed on the screen, and the forced termination processing of the program is performed. In addition, regardless of normal termination or error termination, it is necessary to instruct the system program to invalidate the error detection area and cancel the jump destination when an error occurs when the program ends.

The operation of the personal computer configured as above will be described. A source program created in C language is translated into an object program, and this object program is executed. If it is detected that the source program has a program error (forgetting to initialize the pointer value of the pointer variable) during the execution of the object program, the fact is displayed on the CRT display device 60 and the operator is notified.

FIG. 4 shows a program code representing a source program. Source program is PC
It was created on the MS-DOS for the 9801 series. Although a detailed description of the source program code is omitted, although two functions uekansu and sitakansu are defined in the figure, the function sitakansu is called when the function uekansu is executed.

First line, third line, m + 1th line, mth line
The code “int” described in the + 3rd line and the like is a code that declares that each variable, the return value of the function, and the like are of the integer type. As described above, when each variable, the return value of the function, or the like is simply declared as an int type, it indicates that these are automatic variables, and the stack area 28 in FIG. 2 is used. become. The pointer variable * poi declared to be an int type in the m + 3th line is an automatic variable.

The pointer variable "* poi" is the pointer value po.
Represents the data at the address specified by i, or the data to write at the address specified by the pointer value poi. The pointer variable * poi in the (m + n) th row represents the latter, that is, 123 is the data to be written at the address designated by the pointer value poi. Therefore, the programmer uses the pointer variable * po
Pointer value (address value) that represents the address to write i
poi must be set in the part R before the m + nth line is executed. In this way, setting the pointer value poi representing the address to write the pointer variable * poi is called initialization of the pointer value of the pointer variable.

Even if only the pointer variable * poi is set without initializing the pointer value poi, it is not known to which address the data will be written. Therefore, when the writing is performed to the address in the MS-DOS area 23, the OS is destroyed, and when the writing is performed to the address in the code area 24, the program is destroyed. There is a problem. Further, when the program is executed for the address in the unused area 50, the program is executed without any problem, but an unsatisfactory result is output.

In the source program code shown in FIG. 4, the pointer variable * poi is set, but it is forgotten to initialize the pointer value poi. In other words, it is a state that includes a program error. When this source program code is translated, it becomes the object program code shown in FIG.

The object program shown in FIG. 5 will be briefly described. The portion 60 represents an instruction executed at the start of execution of the function uekansu. Base pointer "b
The p "and stack pointer" sp "are saved. The line 62 shows an instruction to secure an automatic variable area of 6 bytes, and the portion 64 shows an instruction to call the function sitakansu. In line 66 the function uekansu
Is stored in the AX register, and the portion 68 shows the instruction executed at the end of the execution of the function uekansu. The saved base pointer "bp" and stack pointer "sp" are returned.

Similarly, in the portion 70, an instruction for saving the base pointer "bp" and the stack pointer "sp" is shown below. 1 automatic variable area on line 72
The instruction to reserve 2 bytes is the function sitakansu in part 74.
In the line 76, an instruction for executing the execution of the above is stored, and an instruction for storing the return value of the function sitakansu in the AX register is shown. In addition, the portion 78 represents an instruction for returning the base pointer "bp" and the stack pointer "sp".

As is apparent from FIG. 4, the pointer variable * poi is an automatic variable in the source program. Therefore, as shown by a portion S surrounded by a solid line in FIG.
An instruction to set the pointer value poi of poi as "D500: 0800" is automatically added. "D500:
As shown in FIG. 6, "0800 address" is a pointer value in the case of being expressed in the segment space. Therefore, if it is corrected to a pointer value in the virtual address space, it becomes address D5800H, which is included in the error detection area 32 described above. Is clear. In this embodiment, the pointer value poi is set in the middle of the error detection area 32. Further, SS in FIG. 6 represents a segment. FIG.
When the program in the C language shown in FIG. 5 is compiled into the assembler program shown in FIG. 5, the part S in FIG. 5 is not added by the normal compiler, but the part S in FIG. 5 is added by the C compiler of this embodiment. To be done.

Next, the execution of the object program code shown in FIG. 5 will be described with reference to the flowchart shown in FIG.
In FIG. 5, the object program execution process in the user mode is shown on the left side of the broken line, and the system program execution process in the supervisor mode is shown on the right side of the broken line.

In the first step 1 in the object program execution processing (hereinafter, simply represented by S1; the same applies to other steps), initialization is performed. In this initialization, the system program in the supervisor mode is called, and the process of S101 is performed.

At S101, the following processes 1 to 3 are performed. That is, the error detection area is designated, the valid flag is set to 1, the address of the area is stored, the P bit is cleared (that is, the P bit is set to 0), and the jump address when an error occurs is set. This S101
When the process of (1) is completed, the process returns to S1 of the object program execution process.

In the subsequent S2, the function uekansu is called,
In S3, the function sitakansu is called. Then, in S4, the portion 74 in FIG. 5 is executed, and in S5, the write command (WT) of the pointer variable * poi is issued. In the present embodiment, since the pointer value poi of the pointer variable * poi is set to the address D5800H in the error detection area 32 in the portion S in FIG. 5, the writing process to the virtual address D5800H is about to be executed. Then, a page fault exception occurs.

That is, the pointer value poi is set to the address D5800H in the part S, but the pointer value is not set by the programmer (since there is no instruction for setting the pointer value in the source program), this value is Not changed Therefore, a page fault exception is issued because an instruction to write to an address in the error detection area 32 is about to be executed. When this page fault exception is issued, the processing of S111 in the system program execution processing is executed. In S111, it is determined whether or not the cause of the exception processing is due to the error detection area. If the cause of the exception processing is not due to the error detection area, the process proceeds to S112 and normal page fault exception processing is performed. Run. The content of the normal page fault exception processing in S112 is the swap processing described above, and therefore will not be repeated here.

On the other hand, when the cause of the exception processing is due to the error detection area, the process proceeds to S113, and the return address is rewritten to the jump address at the time of error occurrence set in S101. After the processing of S113 is completed, the program is returned to the rewritten genre address.
Move to.

In S11, the execution of the object program is forcibly terminated, and in the subsequent S12, the CRT display device 6
The error display "The pointer variable is not initialized" is displayed in 0. The operator can see from this display that there is a program error.

Then, in S13, a setting cancellation process is performed. In this setting cancellation processing, the system program in the supervisor mode is called, and the processing of S121 is performed. In S121, the error detection area is invalidated (the valid flag is set to 0) and the P bit is set (that is, the P bit is set to 1). When the process of S121 ends, the process returns to S13 of the object program execution process.

If the page fault exception handling does not occur in S5, the function sita is used in S6 and S7.
kansu and the function uekansu are returned, and the setting is canceled in S8. The contents of the setting release in S8 are the same as the setting release in S13, and the system program in the supervisor mode is also called, and in S131,
Disable error detection area (set valid flag to 0)
And set P bit (that is, set P bit to 1)
Is performed and the process returns to S8.

As described above, in this embodiment, before the pointer variable * poi is actually written, that is, the signal for designating the address in the error detection area 32 and the signal for instructing the writing in FIG. Since the execution of the object program is suspended at the point of time when it is issued, the pointer variable * poi is not actually written. However, even if writing is actually performed, since the pointer value poi is set in the error detection area 32 for error detection that is not used by the programmer as described above, other program code, etc. Is prevented from being destroyed. Also, the pointer value
Since poi is set in the middle of the error detection area 32, an error signal is issued when a read / write command is issued to the area of 2 Kbytes before and after the pointer value poi.

On the other hand, when the pointer value a is set by the programmer in the portion R (see FIG. 4) of the source program, the pointer to the address indicated by the set pointer value a is set in S5 of FIG. variable*
The poi will be written. Therefore, the write process for designating the pointer value indicating the address in the error detection area 32 is not issued, so that the page fault exception process is not issued.

As described above, in this embodiment, a program error (specifically, forgetting to initialize the pointer value poi of the pointer variable * poi) can be detected by using the paging unit 13 inside the CPU 10. That is,
A general application can perform a predetermined process on a system program having a right to control the paging unit 13, and a function of the paging unit 13 can be used to detect a program error.

Further, since the detection of the program error is performed by utilizing the paging unit 13 made up of hardware, not by the syntax analysis, the detection speed can be improved and the reliable detection can be realized. ,
Since a program error can be detected by utilizing one function of the paging unit 13 which is originally used for realizing the virtual memory system, it is not necessary to add special hardware.

In the above-described example, the object program itself is forcibly terminated based on a predetermined forcible termination processing routine set as a part of the object program translated by the compiler, as a response process when an error is detected. The CRT display device 60 is configured to display an error, and the CRT display device 60 is configured to display an error. However, if the cause of the forced end is a program error, for example, if the cause of the forced end is a program error, It is conceivable that the operator can know that there is a program error by looking at the instruction manual.

Further, based on a predetermined error occurrence notification processing routine which is set as a part of the system program itself, the operator is notified of the program error occurrence caused by forgetting to initialize the pointer value of the pointer variable in the source program. You may comprise so that a predetermined process may be performed. Then, as the predetermined process of notifying,
As in the above embodiment, the fact that a program error exists is indicated by C
In addition to displaying on the RT display device 60, it is conceivable to sound the notification buzzer or turn on the notification lamp.

In the above embodiment, the portion for converting the source program code into the object program code corresponds to the translation means, and an instruction for automatically setting the pointer value of the pointer variable * poi to the address D58000H (D500: 0800) is automatically generated. The part added to corresponds to the automatic initialization means. And the part which combined this translation means and the automatic initialization means corresponds to a compiler. Further, the memory area 32 (D) in a predetermined range including the pointer value (address D5800H) set by the automatic initialization means
Addresses 5000H to D5FFFH) correspond to the error detection area. Further, in FIG. 7, the part that calls the system program in the supervisor mode in the initial setting of S1 and executes the process of S101 corresponds to the execution of the process by the preprocess executing means. If a page fault exception is issued in S5, the processes of S111 and S113 in the system program execution process are executed, and the process proceeds to S11 of the object program,
The part that executes the processes of 11 to S13 corresponds to the execution of the process by the error detection response process executing means.

In the above embodiment, the error detection area and the jump destination when an error occurs are instructed to the system program in the initial setting at the start of the program, but the error detection area is set before the program starts. You can also However, in this case, since the position of the program has not yet been determined before loading the program, it is necessary to incorporate processing for displaying an error in the system program and processing for forcibly ending the program.

Further, in the above embodiment, the error detection area 32 is set to the addresses D5000H to D5FFFH, but other areas may be set and the capacity is 4 as well.
Other sizes may be used instead of K bytes. Further, the pointer value set by the automatic initialization means is the address D5800H in the error detection area 32, but any pointer value may be used as long as it is in the error detection area 32.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a control unit of a personal computer including a program error detection device according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram showing a RAM represented as a virtual address space.

FIG. 3 is an explanatory diagram of a page table entry.

FIG. 4 is an explanatory diagram showing a source program code translated by a CPU.

5 is an explanatory diagram showing an object program code obtained by translating the source program code of FIG.

FIG. 6 is an explanatory diagram of a RAM.

FIG. 7 is a flowchart showing an execution process of an object program, in which an object program execution process in a user mode is shown on the left side and a system program execution process in a supervisor mode is shown on the right side.

[Explanation of symbols]

10 ... CPU 11 ... CPU core 13 ... Paging unit 15 ... Page table 20 ... System I / O area 23 ... MS-DOS
Area 24 ... Code area 26 ... Data area 28 ... Stack area 32 ... Error detection area 60 ... CRT display device AB ... Address bus DB ... Data bus

Claims (7)

[Claims] 1. A device for detecting a program error that forgets to initialize a pointer value while using a pointer variable as an automatic variable when creating a source program that takes an automatic variable in a stack area, wherein the source program is an object program. And a compiler having an automatic initialization function for automatically adding an instruction for setting a pointer value of a pointer variable to a predetermined address value to an object program when the pointer variable is an automatic variable. , Has an address conversion function for converting a virtual address into a physical address in a predetermined page unit based on a page table,
And when the virtual address is not allocated to any of the physical addresses, a paging means for a virtual memory system having an exception handling request function that causes a page fault and calls a system program, and an object program translated by the compiler. At the start or before the start of the program, pre-processing execution means for setting the address of the error detection area and setting of the corresponding processing at the time of error detection for the system program, and generated by the paging means when the object program is executed When the cause of the page fault is due to reading / writing from / to the error detection area, an error detection response processing execution means for executing a predetermined error detection response processing set by the preprocessing execution means is provided. Prepare Program error detecting apparatus characterized by. 2. The program error detection device according to claim 1, wherein the paging means is built in a CPU. 3. The program error detection device according to claim 1, wherein the error detection response processing execution means is a predetermined forced termination process set as a part of the object program translated by the compiler. A program error detection device characterized in that it is configured to forcibly terminate the object program itself based on a routine. 4. The program error detection device according to claim 1, wherein the error detection response processing execution means is set as a part of the object program translated by the compiler or a part of the system program itself. Based on a predetermined error occurrence notification processing routine, it is configured to execute a predetermined process of notifying an operator of the occurrence of a program error caused by forgetting to initialize the pointer value of the pointer variable in the source program. Characteristic program error detection device. 5. A translation function for translating a source program that takes automatic variables in a stack area into an object program, and when the pointer variable is an automatic variable, sets the pointer value of the pointer variable to a predetermined address value in the object program. A compiler that has an automatic initialization function that automatically adds instructions and an address conversion function that converts an address in the virtual address space into an address in the physical address space in units of a specified page, and also has a virtual address And a paging means for a virtual memory system having an exception handling request function that calls a system program by generating a paging fault when the source variable is created, while the pointer variable is an automatic variable. A program that forgets to initialize its pointer value Method for detecting a system error, at the time of starting the object program translated by the compiler or before the program starts, the address setting of the error detection area and the corresponding processing at the time of error detection are performed for the system program, When the cause of the paging fault generated by the paging means is due to reading and writing in the error detection area during execution of the object program, a predetermined error detection response set by the preprocessing execution means is handled. A program error detection method characterized by executing processing. 6. The program error detection method according to claim 5, wherein, as the error detection response processing, a predetermined forced termination processing routine is set as a part of the object program translated by the compiler, A program error detection device characterized in that the object program itself is forcibly terminated. 7. The program error detection method according to claim 5, wherein as the error detection response processing, a predetermined value set as a part of the object program translated by the compiler or a part of the system program itself is set. A program error detecting method for notifying an operator of a program error occurrence caused by forgetting to initialize a pointer value of a pointer variable in the source program based on an error occurrence notification processing routine.