JPS62191933A - Register control system - Google Patents

Abstract

PURPOSE:To facilitate a call to the modules of different types of language processing systems by giving an instruction modifying the contents of a mapping register. CONSTITUTION:A processor is roughly constituted with three parts. A man memory device 7 has the module 9, and an instruction controller 6 fetches the instruction, translates and transmits it to an arithmetic unit 8 through the main memory device 7 as it is. After the arithmetic unit completes the arithmetic, it stores the result in the main memory device 7. When an ALU 10 to execute the actual arithmetic in the arithmetic unit handles a register, it is converted by the mapping register 4 to handle a physical register 5.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field 1] The present invention relates to a register control method in a general-purpose register type processing device, and the present invention is directed to a register control method for facilitating module calls of different language processing systems. Regarding.

[Conventional technology]

Overlap register window method (vLSI by Tatsu Motooka et al.
Computer IJ, Iwanami Shoten, pp. 44-45), but it requires large-capacity registers, the number of global registers is fixed, and it does not take into account module calls from different language processing systems. .

[Problem that the invention seeks to solve]

An object of the present invention is to provide a register control method in a general-purpose register type processing device, and particularly to provide a register control method suitable for module calls between different language processing systems.

[Means for solving problems]

The method of lane mating, which is the key to the present invention, will be explained with reference to FIG. Operand 2 of instruction 1 specifies rl of logical register 3 (rO, rl . . . rN) and assumes that it is t. At this time, which physical register 5 (l(,
Conversion from the logical register 3 to the physical register 5 is performed using the mapping register 4 (MO, , Ml . In this figure, Ml points to 几N, so it is converted as r1→R,N, and the instruction is executed.

Generally, when Mi = j, it is converted to ri-+Rj and the instruction is executed.

In the present invention, by equipping each module with an instruction to change the value of the mapping register 4, the logic register 3
to enable operation.

In general, different language processing systems often have different module call linkages (for example, work registers, return value storage registers, etc.).

Therefore, when calling modules between different language processing systems, registers must be exchanged. This register exchange can be performed at high speed by changing the contents of the mapping register.

Further, if the contents of the mapping register are automatically changed when the module is called and returned, the speed can be further increased and the module call can be made more efficient.

[Implementation row]

Hereinafter, the content of the present invention will be explained using examples.

FIG. 2 shows the configuration of the embodiment. The processing device is roughly divided into three parts.

There is a module 9 on the main memory 7 , the instruction is fetched by the instruction control unit 6 , interpreted, and the operands are sent via the main memory 7 to the arithmetic unit 8 as they are. . After the arithmetic operation is completed, the arithmetic device stores the result in the main storage device 7. In the arithmetic unit, when handling ALUIOI-j and registers that perform actual calculations, they are converted by the mapping register 4 and used as the physical registers 5.

Next, the configuration of module 9 and module calling will be explained with reference to FIG. Each module 9 has a module disk 11 containing a mapping register value (MR) 12, a base pointer (SB) 13 to a static variable area 16, and a program code 15.
It has a base pointer (PB) 14 to. MOD19 pointer to the module disc libter 11 that is being executed
Put it in. The linkage when module B is called in the module person is shown below.

Here, it is assumed that the module person and the module B are different language processing systems, and the module linkage is as follows.

Module A: Work registers nrO, rl, return value storage register is r7. Module B: Work registers are r5, r7, return value storage register is rOo. The processing is shown in FIGS. 4 and 5.

1. When you call the module, as shown in Figure 4,
Save the contents of MOD (pointer value to A's module descriptor) and return address to the stack, and
The disc libter address of is stored in MOD. As a result, the MR in the disc libter of B is loaded into the mapping register, and the correspondence between the physical register and the logical register is changed for module B. Then, the value of PB is set in the program counter and execution of module B begins.

On the other hand, when returning from module B to module A, as shown in FIG. 5, the return address is taken out from the stack and set in the program counter, and the MOD value saved from the stack is taken out and restored. Furthermore, module A's MRf and mapping register are loaded, and the correspondence between physical registers and logical registers is determined by module A.
and returns to execution of module A.

In this way, as shown in FIG.
, 7(A), and the logical registers during module B execution are rO(B), (B), . . . r7(8) tosult, the following equation holds true.

ro(A):r7(I!I):几.

r 1(A) = r 6(” = R, 1r7(A+
= , □CB) = B 7 Therefore, it can be seen that the usage of the work register and the usage of the return value storage register between modules A and B are well linked.

Similarly, the language processing systems of modules 8 and B are different.
Even if there is a module C with a different module linkage, when calling module Bt- from module C,
If the MR capacity of the module C disc libter is properly set, the linkage will be correct. In other words, if the MR value of the module discriminator is properly set, differences in linkage between modules can be absorbed and calls can be made freely.

As can be seen from the above, according to this implementation sequence, differences in register usage due to module calls in the language processing system can be absorbed into the MR value in the module descriptor, and modules can be called freely. becomes.

〔Effect of the invention〕

According to the present invention, calls to modules of different language processing systems can be easily made by mapping registers, which reduces the burden on the program and improves efficiency, which has the effect of improving software productivity. .

[Brief explanation of drawings]

Fig. 1 is a diagram showing register mapping of the present invention, Fig. 2 is a block diagram in an embodiment of the present invention, Fig. 3 is a diagram showing the module structure and module call, and Fig. 4 is a diagram showing the module call. FIG. 5 is a flowchart at the time of return from the module, and FIG. 6 is a diagram showing the correspondence between physical registers and logical registers. 1... Instruction, 2... Operand, 3... Logical register, 4... Mapping register, 5... Physical register, 6 Figure 1 Figure 2 Raku 3

Claims (1)

[Claims] 1. A general-purpose register type processing device has a mapping register that performs mapping between a general-purpose register number in an operand of an instruction and a corresponding general-purpose register, and stores the contents of the mapping register. A register control method characterized by providing a changing instruction. 2. The register control system according to claim 1, characterized in that the system has a function of automatically changing the contents of the mapping register when calling and returning a module.