JPH0731617B2 - Program runaway detection method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Outline] In a computer system, a method for detecting a program runaway when performing a linkage process between a main routine and a subroutine or between subroutines is described. By providing a means for checking the validity of the return address, it is possible to detect a program runaway early and to minimize the damage caused by a program runaway. In a computer system that saves and manages return addresses in a stack and performs linkage processing between multiple programs, the called program reads the return address from the stack and copies it to a specified area of memory. Perform the original process, After the original processing is completed, the return address is read again from the stack, and this return address is compared with the return address copied in a predetermined area of the memory. If they do not match, it is determined that the program is out of control.

[Industrial application field]

The present invention relates to a program runaway detection method in a computer system when a main routine and a subroutine or a subroutine and a subroutine are linked.

[Conventional technology]

FIG. 4 (a) shows an example of program linkage processing.
In the illustrated example, the main routine A calls the subroutine B during the processing, and the subroutine B further calls the subroutine C in the processing.

When calling these subroutines, the restoration information is saved at the time of calling. The return information includes return address and register information, and when the processing is completed in the calling subroutine, the previously saved return information is taken out and the processing of the calling main routine or the subroutine is restarted based on it.

When subroutines are called multiple times as in the linkage processing shown in FIG. 4 (a), the return information to be saved also becomes multiple, so it is necessary to control the saving and restoration quickly. is there. Conventionally, management of these return information is
This is performed under the control of the OS using the stack and stack pointer as shown in FIG. 4 (b). The stack is constructed in a last-in first-out method.

In FIG. 4B, the state of the stack during execution of the main routine A is empty and is empty. Indicates the state of the stack when the main routine A calls the subroutine B, and "A" indicates the restored information of the saved main routine A. Subroutine B is subroutine C
Is the state of the stack when is called, and "B" indicates the restored information to the saved subroutine B. Is the state of the stack when returning from the subroutine C to the subroutine B, and "B" is taken out from the stack and used. And from subroutine B to main routine A
This is the state of the stack when returning to, and "A" is taken from the stack and used.

[Problems to be Solved by the Invention]

In a conventional computer system, in a program linkage process, a program error may cause a program runaway and the stack pointer may shift or the stack value may be destroyed. In this state, a return instruction is executed from the called program. As a result, the callee's program was erroneously restored to the caller's program, which caused damage such as memory content destruction due to program runaway, and made it difficult to find bugs.

The present invention enables early detection of program runaway by providing means for checking the validity of the return address from the callee program to the caller program in the program linkage processing, and minimizes damage caused by program runaway. The purpose is to limit it.

[Means for Solving the Problems]

According to the present invention, when a program is called, the return address saved in the stack of the called program is copied to the memory, and at the end of the process, the return address fetched from the stack is copied to the memory. The return address of the stack is judged to be valid when the two match and the runaway is recognized when the two do not match.

FIG. 1 is a principle diagram of the present invention.

In the figure, 1 is a calling program such as a main routine or a subroutine.

Reference numeral 2 is a called program such as a subroutine.

2a, 2b, 2c are components of the called program 2, 2a is a return address copy process placed at the beginning of the program, 2b is the original process, and 2c is the return address validity placed immediately before the return instruction. This is a check process.

Reference numeral 3 is a stack for saving the return address.

Reference numeral 4 is a return address copy area of the memory.

In the return address copying process of the called program 2, the return address saved in the stack 3 is once taken out, copied in the return address copying area 4 of the memory, and the return address is loaded back in the stack 3.

In the return address validity check process, stack 3
The return address fetched from is compared with the return address copied in the return address copy area 4 to judge the validity of the return address of the stack.
If they match, it is judged as valid, and the return to the calling program 1 is executed. If they do not match, it is recognized that the program is in a runaway state as an error in the return address, and the program is stopped and error processing by the error processing routine is performed. Let it be done.

[Action]

During processing by calling a program, the return address changes if the program runs out of control due to a bug and the stack pointer shifts or the contents of the stack are destroyed. The present invention can easily detect this by checking the validity using the return address copied at the start of the processing of the called program.

When multiple program calls are made, the validity of the return address is checked in each called program.

[Embodiment] FIG. 2 is a block diagram of a computer system according to an embodiment of the present invention, in which only the elements related to the present invention are simplified and shown.

In the figure, 5 is a CPU, 6 is a register [R1], 7 is a stack pointer, 8 is a memory with a stack area [1], 9 is a memory with a data storage area [2], and 10 is a return address copy area [M1 ].

FIG. 3 shows a flow of a linked program of the main routine 11 and the subroutine 12 according to the embodiment of the present invention. These are executed under the computer system of FIG. The operation will be described below.

The main routine 11 is composed of a process A, a process by a subroutine call, and a process B.

Not shown when a subroutine call is issued
The OS saves the return address in the stack area of the memory [1] pointed to by the stack pointer 7 and activates the subroutine 12. The subroutine executes the following steps.

． The subroutine 12 first executes the instruction POP R1. This instruction pops up the top return address from the stack and stores it in register [R1].

． Next, the subroutine 12 executes the instruction PUSH R1.
This instruction pushes the return address of register [R1] onto the stack and reloads it. At this time, register [R1]
Has the previous return address.

． Next, the return address of the register [R1] is stored in the return address copy area [M1].

． The subroutine 12 executes the original process C.

． After the end of the process C, the instruction POP R1 is executed to pop up the return address from the stack and store it in the register [R1].

． Then execute the instruction PUSH R1 to push the return address back onto the stack.

． The contents of the register [R1] and the contents of the return address copy area [M1] are compared.

． If R1 = M1 in the comparison, the return instruction (not shown) is executed, and the return control of the OS is performed using the return address of the stack.

． If R1 ≠ M1 in the comparison of, the program is stopped.

When the return to the main routine 11 is performed, the process B is executed from the return address.

〔The invention's effect〕

According to the present invention, it is possible to detect a bug in a program at an early stage by a relatively simple process while maintaining the independence of the program regarding the runaway detection process of the program in the linked process of a plurality of programs. This has the advantages that it is easy to identify the program that contains the bug, the efficiency of the bug analysis and investigation can be improved, the malfunction caused by the bug can be prevented from expanding, and the modification does not spread to other programs when the program is revised.

[Brief description of drawings]

FIG. 1 is a principle diagram of the present invention, FIG. 2 is a configuration diagram of a computer system according to one embodiment of the present invention, FIG. 3 is a flow chart of a program according to one embodiment of the present invention, and FIG. 4 is a conventional program. It is an explanatory view of cooperation processing. In Figure 1, 1: Calling source program 2: Calling destination program 2a: Return address copy processing 2b: Original processing 2c: Return address validity check processing 3: Stack 4: Return address copy area

Claims (1)

[Claims] 1. When a program being executed is called from another program, return information such as a return address is saved in a stack and managed, and in a computer system for performing linkage processing between a plurality of programs, the called program is , Read the return address from the above stack, copy it to a specified area of memory, and then execute the original processing,
After the original processing is completed, the return address is read again from the stack, and this return address is compared with the return address copied to a predetermined area of the memory. If they do not match, it is determined that the program has runaway. Program runaway detection method.