JPH0452491B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a work area management method for a reentrant program, and in particular to a GETMAIN/
This invention relates to a work area management method that can easily secure and manage a necessary work area from an unused area without using the FREEMAIN macro command.

[Technology background]

In a multiprogramming system or multiprocessor system, subroutines for processing commonly required by each program, such as functional calculations, are
If each program holds the subroutine separately, the utilization efficiency of the main memory will decrease, so it is desirable to place only one such subroutine on the main memory and share it with each program.

However, since shared subroutines may be used concurrently by multiple programs, the instructions or data within the shared subroutines must not be rewritten by any program, and the work area must be shared. It is necessary to separate it from the subroutine and manage it for each program that calls the shared subroutine. A subroutine with such a structure is called a reentrant program.

The work area of a reentrant program must be reserved by the calling program or dynamically by the reentrant program when the reentrant program is executed. FIG. 1 shows an example of the latter, and 1 and 2 are reentrant programs. Put a GETMAIN macro at the beginning of each reentrant program and at the end
Place the FREEMAIN macro. The GETMAIN macro requests the supervisor to secure an area within the region (user area of memory), and also
Use the FREEMAIN macro to request the supervisor to return the previously secured area. Although this method is usually used, it has the drawback that it requires the execution of several hundred instructions for one request, which places a burden on the supervisor and reduces processing performance.

In addition, in consideration of future performance improvements of the reentrant program itself, a sufficiently large area should be created in advance.
Secure it with the GETMAIN macro, and at runtime,
There are ways to secure the necessary space yourself. However, with this method, the interface between the programs in the calling relationship becomes complicated, and as shown in FIG. When X intervenes, it becomes difficult to manage the area to connect the areas between the reentrant programs A and B, which places restrictions on the programs that can be applied.

[Object and structure of the invention]

An object of the present invention is to manage a work area independently and easily, without depending on a supervisor, even when other programs with different area management methods are sandwiched between a plurality of reentrant programs in a calling relationship. An object of the present invention is to provide a work area management method for a reentrant program that can ensure the following.

To this end, the present invention has a configuration in which, in a system in which a plurality of programs including reentrant programs are connected in a multiple call relationship, when the first reentrant program is executed, there is sufficient memory for itself and subsequent programs. A work area pool is secured, a control table containing information on the start position of the work area pool is created, and free space information is stored in a predetermined position of the work area pool, and the control table is sequentially created by one program. When control is passed to the next program called from the program, when each called program is a reentrant program, the position information of the work area pool is obtained from the control table and stored in the pool. The program secures the necessary work area based on the free space information in the work area pool, updates the free space information, and when the program that has secured the work area in the work area pool completes its processing, returns the work area and updates the free space information. It is characterized by restoring area information.

[Embodiments of the invention]

The details of the present invention will be explained below with reference to Examples.

FIG. 3 shows the same program example as in FIG. 2 to which the present invention is applied. In the figure, 3 and 4 are reentrant programs A and B, respectively, 5 is a program X with a different area management system, 6 is a supervisor, 7 is a work area pool, and 8 is a control table T. First, reentrant program A secures a sufficiently large work area pool 7 using the GETMAIN macro when starting execution. At this time, a control table T for managing the work area is created. Control table T contains pointers pointing to reserved work area pools.

When a call is made from reentrant program A to, for example, non-reentrant program X, and from program X to reentrant program B, control table T is passed. In this case, since program X has a different area management system, control table T is not used and the program is handed over to the next reentrant program B as is. The reentrant program B learns the secured work area pool from the pointer in the control table T, and secures its own work area therein as will be described later.

Note that when passing control tables between programs,
The easiest way to do this is in the form of parameters in the program call, which is a commonly used technique, but other commonly used techniques such as setting the address of a control table at a specific location in job space are possible. can also be used.

FIG. 4 shows an example of a work area management method. In the figure, 7 shows a work area pool, 8 shows a control table T, and 9 shows a program X work area.

The first address of the work area pool 7 pointed to by the control table T stores the final address _P1 of the work area pool 7, and the next address stores the first address _P2 of the free area.

Setting of P ₁ and P ₂ is performed by each reentrant program A and B, and the program includes a description therefor. For example, reentrant program B obtains the start address of the work area pool 7 from the control table T, checks the size of the free area in the pool from P ₁ and P ₂ , sets its own work area therein, and ₂ is updated with the start address P′ ₂ of the new free area.

When reentrant program B completes its processing, it returns its work area, restores _P'2 to _P2 , and returns execution rights to program X. When program X completes its processing, the execution right is returned to reentrant program A. The reentrant program A executes the FREEMAIN macro at the end of processing and requests the supervisor 6 to return the work area pool 7 and control table 8.

In this way, no matter how many consecutive reentrant programs there are, each reentrant program can set its own work area in the initially reserved work area pool. In addition, the program description for passing the control table to sequentially called programs is done, for example, in the parameter description for program calling, but this description is automatically written to each reentrant program by the compiler during the compilation process. can be added to.

〔Effect of the invention〕

As described above, according to the present invention, GETMAIN/
Supervisor calls (SVCs) due to FREEMAIN are limited to the first reentrant program, so for these SVCs from the called reentrant program to the supervisor, the supervisor has to execute only a few dozen instructions instead of hundreds. This makes it possible to execute only the following instructions, significantly reducing the burden on the supervisor and improving execution performance.

[Brief explanation of drawings]

Figures 1 and 2 are diagrams showing different methods for securing a work area for a conventional reentrant program, respectively, Figure 3 is an embodiment of the method of the present invention, and Figure 4 is an explanation of its free space management. It is a diagram. In the figure, 3 and 4 are reentrant programs, 5
are programs with different area management methods, 6 is a supervisor, 7 is a work area pool, and 8 is a control table.

Claims (1)

[Claims] 1. In a system in which multiple programs including reentrant programs are linked in a multiple call relationship, when the first reentrant program is executed, it must ensure a sufficient work area pool for itself and subsequent programs, and A control table containing information on the start position of the work area pool is created, and free space information is stored at a predetermined position in the work area pool, and the control table is used to control the next program called from one program in sequence. is passed at the same time, and when each called program is a reentrant program, it obtains the location information of the work space pool from the control table and calculates the required space based on the free space information stored in the pool. It is characterized by securing a work area, updating free space information, and returning the work area and restoring the free space information when the program that has secured the work area in the work area pool completes processing. A work area management method for reentrant programs.