JPH07152551A - Computer system and program executing method - Google Patents

Abstract

PURPOSE:To realize the on-line version up by correcting a function without discontinuing the execution of a program. CONSTITUTION:An address control table 133 contains the entry addresses for each function, and an entry address is selected out of the table 133 when the corresponding function is carried out. Meanwhile a corrected function is loaded in a memory 13 and an entry address of the table 133 is rewritten into the entry address of the loaded function. Thus a program can be corrected without discontinuing the program. Therefore it is just required for revision of a certain function to add a new version of the function with its old version left in a program as it is and to rewrite the address corresponding to the function into the address of new version of the function. In such a way, the on-line version up can easily be attained.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a computer system and a program executing method, and more particularly to a computer system and a program executing method suitable for correcting and correcting a program in the system.

[0002]

2. Description of the Related Art In recent years, the performance of computers has continued to improve dramatically, and as a result, the scale of programs has increased.
The speed is increasing. Therefore, bugs in the program are inevitably increasing.

Computers and their programs often play an important role in corporate information systems, and it is important to correct bugs in such programs as soon as they are discovered.

Further, there are many cases where it is desired to change / improve the function of the program due to factors such as a change in the request to the program. In the conventional program execution method, when calling a subroutine, the address of the subroutine determined at the time of linking is used directly. In this method, when the program is changed, the address of the subroutine is also generally changed, and therefore, it is necessary to change the location of calling the subroutine (entry address of the subroutine). For this reason, it is very difficult to revise the running program (referred to as online version upgrade), and it is necessary to temporarily stop the program and recompile it. This can result in a significant loss, especially if you revise a program that manages account deposits and withdrawals on computers used in online systems such as banks, the entire online system will be shut down. The problem that it has to occur occurs.

[0005]

In the past, when the program was changed, the entry address of the subroutine also changed, making it difficult to upgrade the version online. To correct the program, the system is temporarily stopped and the program is compiled. The drawback was that it had to be redone.

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a computer system and a program execution method capable of easily performing an online version upgrade without stopping the operation of the system. .

[0007]

In a computer system according to the present invention, an address management for each subroutine included in a program to be executed holds address information indicating an entry address on a memory in which the subroutine is stored. By referring to the table, this address management table, the entry address in the memory corresponding to the subroutine to be called is detected, the subroutine calling means for calling and executing the subroutine of the entry address, and the new subroutine corresponding to the subroutine to be corrected Load the appropriate subroutine on the memory and set the corresponding entry address in the address management table
Table updating means for rewriting the entry address into which the new subroutine is loaded is provided.

In this computer system, an address management table for holding an entry address is provided for each subroutine. When the subroutine is executed, the corresponding entry address is obtained from the address management table and the subroutine is modified. In this case, by loading the modified subroutine on the memory and rewriting the entry address of the address management table to the entry address of the loaded subroutine,
The program is modified without stopping the program.

Therefore, when a sub-routine is revised, the new version is added to the program while the old version of the sub-routine remains in the program, and the address corresponding to the sub-routine need only be rewritten to that of the new sub-routine. . As a result, online version upgrade can be easily realized.

Further, the address management table is provided with an area in which call prohibition information indicating prohibition of call of a subroutine is set, and during the rewriting period of the entry address, the old version of the subroutine is called by the subroutine calling means. Call barring information is set to barred. As a result, the consistency of the program execution contents between threads can be guaranteed.

[0011]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows the configuration of a computer system according to an embodiment of the present invention. The computer system 11 is one of a plurality of computer systems that form an online system, and the network 21
Connected to other systems via.

As shown in the figure, the computer system 11 is provided with a CPU 12, a main memory 13, a communication control unit 14, and a magnetic disk device 15, which are interconnected via a system bus 15.

The CPU 12 controls the entire computer system 11, and executes the operating system 131 of the main memory 13 and various application programs. The main memory 13 is
The operating system 131 and the application program 132 to be executed are stored.

The operating system 131 is composed of various programs for system management.
Here, a subroutine load management program 131a necessary for realizing the program execution method of the present invention is provided.
And a table management program 131b are incorporated.

Subroutine load management program 131
The a loads the revised new subroutine from the magnetic disk device 15 or the network 21 into the empty area of the main memory 13. The table management program 131b stores the address management table 133 in the memory 13
It is secured above and the contents of the table 133 are managed. When the new subroutine is loaded into the main memory 13 by the subroutine load management program 131a, the table management program 131b rewrites the entry address corresponding to the old version subroutine of the address management table 133 to the load address.

The application program 132 is for managing the withdrawal / receipt of an account, and includes a plurality of subroutines. The application program 132 is written in the language C. In language C,
Subroutines are called functions. Therefore, in the following description, all the subroutines will be referred to as functions.

In the application program 132, the function call instruction 132a is an instruction to call the function (f1), but the entry address of the called function is not incorporated in the function call instruction 132a.
The entry address is obtained by referring to the address management table 133. The address management table 133 indicates "A" as the entry address of the function (f1).
When it indicates 0 ", the function (f1) 132b is called and executed, but when" A1 "is indicated as the entry address of the function (f1), the function (f1) 132c is called and executed. It

The address management table 133 holds, for each function included in the application program 132, its identifier (function name), the entry address of the memory 13, and the entry prohibition flag. When modifying the function (f1), the function name f1 of the address management table 133 is loaded after the revised function (f1) 132c is loaded.
The entry addresses corresponding to are from A0 to A as shown.
Can be rewritten as 1. The entry prohibition flag is set to "H" during the function modification period, thereby preventing the old version function from being called at the time of function modification.

The communication control unit 14 is connected to the network 2
1 to perform communication with other systems, and is used, for example, to download a revised function from the network 21. Magnetic disk drive 1
Reference numeral 5 is used as a secondary storage device of the computer system 11 and stores various programs and files.

Next, an example of specific contents of the address management table 133 will be described with reference to FIG. As shown, the function read of the application program 132
When the file has a file (), check data (), and write file (), the address management table 133 holds the correspondence between the function name and the entry address. Here, function read file (), function check data (), function write fi
le () has addresses 0x70000000 and 0, respectively
It is assumed that it exists at x70001000 and 0x70005000.

When the function read file () is revised,
The revised version of the function read file () is loaded into the memory, and the entry address corresponding to the function read file is from 0x70000000 corresponding to the old version function read file () to the entry address of the revised version read file (here, 0x70006000). Is changed to. After that, the function
Address 0x700 when calling read file ().
06000 is called.

FIG. 3 shows a comparison between the function call instruction of the conventional C program and the function call instruction adopted in the present invention. In a normal C program, when calling the function read file (), a code fragment that calls the function read file () with an argument file is used as a call instruction. In this case, the address 0x70000000 of the function read file () is embedded in the object code at the time of linking.

On the other hand, in this embodiment, the character string of the function name is used as an argument, and the address management table 133 is used.
Function that returns the address of the function with reference to fn name to ad
A piece of code that calls the function read file () via rs (),
I.e. fn name to adrs ("read file")) (file)
Is used as the call instruction for the function read file ().

Next, referring to the flowchart of FIG.
The operation procedure to update the function read file () is explained. First, the table management program 131b of the operating system 131 uses the old version of the read file () function.
To prevent the new execution of
Set the entry prohibition flag corresponding to le () to "H",
Calling this function is temporarily prohibited (step S1).
1). Next, the load management program 131a executes the function
The new version of read file () object code is stored in memory 1
A new version of the function read file () is added by loading in the empty area 3 (address 0x70006000 or later) (step S12).

After that, the table management program 131b
Is the function read file () of the address management table 133.
The entry address corresponding to the new version address 0x7
Rewrite to 006000 (step S13), and
By resetting the flag from "H" to "L", the calling of the function is permitted again (step S14).

As a result, the new version of read file () can be read without changing the object code of the old version of read file ().
It is possible to call (). Note that the procedure of FIG. 4 suspends only new calls during the revision of the function read file (), so in multiprogramming such as multithreading and multitasking, the old and new versions of the read file can be read after the revision. There may be a situation where () is operated at the same time. This situation does not necessarily cause inconvenience, but as shown in the flowchart of FIG. 5, step S1 is added to the flowchart of FIG.
By adding 00, it is possible to easily prevent the occurrence of such a situation.

That is, in FIG. 5, steps S11 to S13 are executed similarly to the flowchart of FIG. 4, but it is waited for all threads executing the function read file () to finish processing in step S100. After that, permission to call the new version of the function is granted (step S14).

This step 100 can be realized by preparing a thread counter for counting the number of threads (or tasks) executing the function for each function. This thread counter can be realized as a part of the function of the operating system 131. In this case, the address management table 133 preferably has a field for holding the value of the thread counter, as shown in FIG.

Next, the procedure for calling the function read file () will be described with reference to the flowcharts of FIGS. 7 and 8. First, in order to obtain the address of the function read file () to be called, the function fn name to addr () is called with the character string "read file" of the function name as an argument (step S21).

In step S21, as shown in FIG. 8, first, the function name "read" given as an argument is read.
In order to check whether the function read file () corresponding to "file" is currently being revised, the entry prohibition flag is checked (step S211). If the flag is set to "H", it is being revised. Therefore, the function call is temporarily stopped by looping and waiting until it is reset, and the address corresponding to the function name "read file" is obtained by asserting the address management table 133. The obtained address is returned to the calling side (steps S212 and S213).

After this, as shown in FIG. 7, fn
The function read file () is called using the address of read file () returned from name te addr () (step S22).

As described above, in this embodiment, the address management table 133 holding the entry address for each function is provided, and when executing the function, the corresponding entry address is obtained from the address management table 133. When the function is modified, the modified function is loaded on the memory, and the address management table 133
By rewriting the entry address of <1> to the entry address of the loaded function, the program is corrected without stopping the program.

Therefore, when revising a certain function,
It is only necessary to add the new version to the program while leaving the old version of the function in the program, and rewrite the address corresponding to the function to that of the new version function. As a result, online version upgrade can be easily realized.

In the address management table 133,
An area for setting a call prohibition flag indicating prohibition of function call is provided, and the call prohibition flag is set so that the call of the old function is prohibited during the rewriting of the entry address. This allows
It is possible to guarantee the consistency of program execution contents between threats.

[0035]

As described above, according to the present invention, when a subroutine is revised, the new version is added to the program while the old version of the subroutine remains in the program, and the subroutine identifier and its address are added. In the correspondence table, it is only necessary to rewrite the address corresponding to the subroutine to that of the new version subroutine. Therefore, the online version upgrade can be easily performed without stopping the operation of the system.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a computer system according to an embodiment of the present invention.

2 is a diagram showing an example of a specific configuration of an address management table provided in the computer system of FIG.

3 is a diagram showing an example of a function call instruction used in the computer system of FIG.

4 is a function rea in the computer system of FIG.
The flowchart explaining an example of the operation procedure for updating d file ().

5 is a function rea in the computer system of FIG.
The flowchart explaining other operation procedures for updating d file ().

FIG. 6 is a diagram showing an example of another specific configuration of an address management table provided in the computer system of FIG.

7 is a flowchart illustrating a procedure for calling a function read file () using the address management table of FIG. 2 in the computer system of FIG.

8 is a flowchart illustrating a table reference operation in the flowchart of FIG.

[Explanation of symbols]

11 ... Computer system, 12 ... CPU, 13 ... Main memory, 14 ... Communication control unit, 15 ... Magnetic disk device, 131 ... Operating system, 13
1a ... Subroutine load management program, 131b ...
Table management program, 132 ... Application program, 132a ... Function call instruction, 132b, 13
2c ... Function, 133 ... Address management table.

Claims (4)

[Claims] 1. An address management table that holds, for each subroutine included in a program to be executed, address information indicating an entry address on a memory where the subroutine is stored, and a target to be called by referring to this address management table Subroutine calling means for detecting the entry address in the memory corresponding to the sub-routine, calling the subroutine of the entry address and executing it, and loading a new subroutine corresponding to the subroutine to be modified on the memory And a table updating unit that rewrites the corresponding entry address of (1) to the entry address loaded with the new subroutine. 2. The address management table has an area in which call prohibition information indicating call prohibition of the subroutine is set for each of the subroutines, and the table updating means writes an entry address of the address management table. During the replacement period, there is provided means for setting the calling prohibition information in the address management table prior to rewriting the entry address so that the calling of the correction target subroutine by the subroutine calling means is prohibited. The computer system according to claim 1. 3. A call control having, for each subroutine included in a program to be executed, address information indicating an entry address on a memory where the subroutine is stored, and two states indicating prohibition and permission of the call of the subroutine. An address management table holding information, a thread counter indicating the number of threads executing the subroutine for each of the subroutines, and an entry address on the memory corresponding to the called subroutine by referring to the address management table Is detected, and a subroutine calling means for calling and executing the subroutine of the entry address and a new subroutine corresponding to the subroutine to be modified are loaded on the memory, and the corresponding entry address of the address management table Table updating means for rewriting an entry address into which a new subroutine has been loaded, and a call to the address management table prior to rewriting the entry address so that the calling of the correction target subroutine by the subroutine calling means is prohibited. Means for setting call control information of the prohibited state, and detecting that the number of threats executing the correction target subroutine is exhausted by referring to the thread counter,
And a means for rewriting the call control information to a call permission state in response to the detection. 4. For each subroutine included in a program to be executed, address information indicating an entry address on the memory where the subroutine is stored is registered in the memory, and responds to a subroutine call instruction of the program to be executed. Then, referring to the address information, the entry address in the memory corresponding to the subroutine to be called is detected, the subroutine at the detected entry address is called and executed, and a new subroutine corresponding to the subroutine to be corrected is executed. Is loaded on the memory, and the entry address corresponding to the address information is rewritten to the entry address loaded with the new subroutine.