JPH01217635A - Register saving system - Google Patents

Abstract

PURPOSE:To ensure the quick switch of tasks by saving and restoring a register with the use of a tag bit and a saving area address register and using the bus idle time during the execution of a switched task or a procedure or every time register reference and replacement instructions are issued. CONSTITUTION:The register reference/replacement instructions are carried out after a called task is started. In this case, the present register contents are saved into a saving area pointed by a calling side saving area address register 21 as long as the tag bit of the relevant register is equal to 0. In addition, the data are read out of a saving area pointed by a called side saving area address register 22 and the register is restored. In such a way, the register is not saved nor restored at the switch of registers but the tasks are first switched. Thus the register is saved and restored by means of an idle bus during the run of the switched task or at the time point when the register is referred to and replaced in a task execution mode. As a result, the tasks are switched quickly.

Description

[Detailed Description of the Invention] [Summary of the Invention] Regarding a method for saving registers when tasks are switched in a computer, an interrupt occurs, or a procedure (subroutine) is called, switching can be performed quickly, The purpose is to provide a register saving method that does not cause register contents to be lost due to programming errors. A tag bit is provided for the caller, and a save area address register called the caller save area address register is provided, and these tag bits and the save area address register are used to save and restore registers, and execute the task or procedure that was switched. The structure is such that the register reference and update commands are performed each time a register reference or update command is issued, using bus free time during the process.

[Industrial application field]

The present invention relates to a method for saving registers when a computer switches tasks, receives an interrupt, or calls a procedure (subroutine).

At the time of task switching, interrupt, or procedure call, in order to preserve the environment of the previously running task or calling procedure, in the conventional method, O8 or the called procedure is the program's responsibility to execute all operations using machine language instructions. Retirement of the register! / Must be restored.

[Conventional technology]

5 to 7 show the procedure for saving and restoring registers.

Figure 5 shows the situation when switching tasks. When a switching event to task B occurs while task A is being executed, the OS
ing System) reads the contents of all registers of the CPU and stores them in save area A, then reads the contents of the registers saved during task B processing from save area B and sets them in the corresponding register. , thus task B
After preparing the execution environment, task B is instructed to execute.

Figure 6 shows an interrupt. When an interrupt occurs while task A is being executed, O8 saves the contents of all registers to save area A.
After interrupt processing is completed, all registers are restored and task A is restarted.

Figure 7 shows when a procedure is called. When procedure 8 in execution calls procedure B, procedure B saves the contents of all registers to save area A, then starts executing procedure B, and completes its execution. Then, the save area A is read out, all registers are restored, and the execution of procedure A is resumed.

However, this conventional method has the following problems.

That is, (1) It takes time to save and restore registers.

This is especially noticeable in CPUs with a large number of registers. The time required to save and restore registers is extremely important, especially in interrupt processing, as it affects response time. In addition, the program is responsible for saving and restoring registers, so if there is an error in the program (for example, if the programmer forgets to add a set of instructions for saving and restoring), the contents of the registers are not guaranteed, and the self-task Or, it may interfere not only with its own procedures but also with other tasks or operations that continue in other years.

[Problem to be solved by the invention]

SUMMARY OF THE INVENTION An object of the present invention is to improve the above-mentioned problems and to provide a register saving method that allows rapid switching and prevents loss of register contents due to programming errors.

[Means to solve the problem]

As shown in FIG. 1, in the present invention, registers 11, 12.・・・・・・Tag Pi) 1
1a, 12a, . . . are added, and a calling side save area address register 21 and a called side save area address register 22 are provided so that the CPU can automatically save and restore registers. do.

[Effect]

The CPU uses the tag bits Ila, 12a, . . . and the save area address registers 21, 22 to save the registers in the following manner.

■Clear the tag bits of all registers to 0 when switching tasks. Also updates the save area address register. That is, the contents of the called side save area address register 22 are transferred to the calling side save area address register 21, and the save area address of the task that is about to be called is stored in the former (register 22).

■After the called task starts, refer to the register/
When executing an instruction to update, check the tag bit of that register and if it is 0 (the contents of the register are invalid, unsaved, and the contents are the same as when they were called), the current register contents are pointed to by register 21. The data is saved to a save area, and further data is read from the save area pointed to by the register 22 and the register is restored (this restoration process is unnecessary in the case of a procedure call). Once restored, set the register's tag bit to 1 (the contents of the register are valid, ie, have been updated, and the contents of the register are those of the called party).

■Before ■ above, in parallel with the execution of the instruction started by the called task, the operation of ■, that is, the saving of the register with tag O, was performed automatically using the bus free time.
Perform restoration.

In this way, the present invention saves registers and saves registers when switching tasks.
Rather than performing restoration, the task is first switched, and while the switched task is running, the bus is used for free space, and if there is a register reference or update during the execution of the task, the register is saved and restored at that point. , task switching can be performed quickly. Once the registers are saved and restored using the bus free time, the time required for this will no longer appear on the surface and will be as if it had never existed.

Furthermore, since the CPU automatically performs the saving and restoring, there is no possibility that registers cannot be saved or restored due to a programming error, thereby causing a problem in operation.

〔Example〕

FIG. 2 shows processing related to saving and restoring registers during execution of a task switching (or interrupt, or procedure call, although these are omitted as appropriate) instruction. In this process, (1) the tags of all registers are cleared to O, and (2) the contents of the calling side save area register X are switched to the contents of the called save area register Y. Taking the case of switching task A to task B as an example, the save area address for the task is in register Y, and is moved to register X in the process (2) above.

Next, the save area address of task B is set in register Y, and this completes the task switching process.

FIG. 3 shows processing related to register saving and restoring when a register referencing/updating instruction is issued while a switched task is being executed. ■Usage for the register containing the instruction to be executed (
(Reference/Update) Check whether it is an instruction, and if it is a used instruction, ■ Check whether the tag is 0, and if it is 0, ■ Register
The contents of the register are saved to the address indicated by , and instead: 1) the memory contents of the address indicated by register Y are set in the register Y, and 2 the tag of the register Y is set to 1. This completes the saving and restoring of registers, and the execution of the relevant instruction begins.

The process shown in FIG. 3 is performed every time a register reference/update command is issued, and therefore for each register. Although not shown, even if there is free time on the bus (even if no register save/update command has been issued), each register is sequentially saved/restored and the tag bit is switched from 0 to 1. The register save/restore process is completed when the register save M/update command or the process using the empty bus is performed and the process reaches all registers. The save and restore addresses are in registers X and Y (21.22 in FIG. 1). This is common to a group of registers, but the addresses for individual registers in the group are fixedly determined by using the order of the second register, second register, etc.

FIG. 4 shows an example of register saving/restoring. Only one register (11) is shown, but the other registers 12°...
The same applies to 1n. (a) Before task switching, the contents of register 11 are 1111, the tag bit is 1, the contents of register 21 are 2222, and the contents of register 22 are 2222.
Assume that task A is being executed and is switched to task B. The contents 3333 of register 22 are task A
In this example, data 9999 is stored at this address in the memory 30.

(By executing the bl task switching instruction, the tag bit of register 11 is cleared to O, the contents of register 22 -3333 are written to register 21, and the save area address 5555 of task B is set to register 22.

(C) Content 1 of register 11 when register usage instruction is executed
111 is addressed to memory address 3333 indicated by register 21, memory data 6666 of save area address 5555 of task B is set in register 11, and the tag bit of register 11 is set to 1.

Although not shown, when switching from task B to task C, the state shown in FIG. 4(C) becomes the same as that shown in FIG. Set the contents 5555 of , set the address of the save area of task C in register 22, save the contents 6666 of register 11 to address 5555 indicated by register 21 by executing the register use instruction, register 1
The data at the address indicated by the register 22 is set to 1.

When a procedure is called, the register is saved, but it is not restored (the register data of the called procedure is saved to the register), so step 3 in FIG. 2 and step (2) in FIG. 3 are unnecessary.

〔Effect of the invention〕

As described above, in the present invention, registers are not saved or restored when switching tasks, so the switching time can be shortened. Since registers are saved and restored using free bus time, task switching becomes faster at best by the time it takes to execute the conventional save and restore instructions. The worst case is when the bus is not available and register save/restore is performed only by a register reference/update instruction, but even in this case, the required time is the same as in the conventional case.

Furthermore, in the present invention, since the CPU automatically performs the saving and restoring, errors in the program will not affect other tasks (continued by Nobuko).

[Brief explanation of the drawing]

Fig. 1 is an explanatory diagram of the principle of the present invention, Fig. 2 is a flowchart showing processing when switching tasks, Fig. 3 is a flowchart showing register saving and restoring operations, and Fig. 4 is register saving! /An explanatory diagram of an example of restoration processing. Figures 5 to 7 are explanatory diagrams of conventional register saving and restoration. Figure 5 is at the time of task switching, Figure 6 is at the time of interrupt, and Figure 7 is at the time of procedure call. be. 11.12 in Figure 1. . . . is the register of the CPU, 11a, 12a, . . . are its tag bits, 2
1 is a caller side save area address register, and 22 is a called side save area address register.

Claims (1)

[Claims] 1. In the data saving and restoring method of the registers of the central processing unit during task switching, interrupts, or procedure calls, tag bits (11a
. A register saving method is characterized in that it is carried out by utilizing bus free time during the execution of a switched task or procedure, and whenever a register reference or update instruction is issued. 2. Clear the tag bits of all registers to 0 when switching tasks, and also clear the called side save area address register (22)
The contents of are moved to the caller side save area address register (21), the save area address of the called task is set in the called side save area address register, and the register is referenced and updated while the called task is being executed. When an instruction is issued, check the tag bit of the register and if it is 0, save the contents of the register to the memory address indicated by the save area address register of the caller, and save the contents of the register to the memory address indicated by the save area register of the called side. 2. The register saving method according to claim 1, wherein data of a memory address is set and a tag bit of the register is set to 1.