JPS60221853A - Swapping method of computer operating system - Google Patents

Abstract

PURPOSE:To decrease the number of times of swap-in as small as possible when a control has reached the process again, by leaving a swap unit executed at the time of swap-out, without swapping out said unit to the end. CONSTITUTION:For instance, a swap unit A3 of a process P1 of a main storage process 4 is swapped out in a secondary storage device 1. Also, in case a restart address of the process P1 exists in a swap unit A1, a managing program 2 restarts the execution of the process P1 through a swapper 3, in a state that the swap unit A3 remains swapped out. Subsequently, the swapper 3 selects a process P2, and in case its executing address is in a swap unit B1, the swapper 3 selects a part of a unit B3 separated most distantly from the unit B1, and it is swapped out to the device 1. In such a case, the program 2 commands swap-in of the process P1 again to the swapper 3, and if a dead area required for a process 4 cannot be secured, a part of the units B2, B1 is selected.

Description

TECHNICAL FIELD OF THE INVENTION The present invention relates to a swapping method for a computer operating system, and more particularly to a swapping method with demand paging R11 capability.

[Prior Art j] Conventionally, there has been a swapping method of this type as shown in FIG. In the figure, 1 is a secondary storage device that stores the image of the operating system execution [process], 2 is a management program that controls multiple execution processes, 3 is a swapper 5, and 4 is a process in the main storage device. It is.

The swapper 3 is a swap in/out control that controls reading of an execution image from the secondary storage device 1 to the main storage device process 4 and writing of an execution image from the main storage device process 4 to the secondary storage device 1. program, main memory process 4 indicates a process that was swapped at a certain time, and Pl is a swapping target process.
1" 2 is the process to be swapped out, A1 and A2 are one swap unit of process P1, and B1-84 are each of process P2.
represents one swap unit.

Next, the swapping operation of the computer operating system shown in Figure f51 will be explained.

Step a) The management program 2 executes a swap unit A1 of the process P1 which has already been partially swapped into the secondary storage device 1 by a predetermined command (the broken line part for the process PI, the solid line part for the two-dimensional 1! device 1) attempt to run the process.

Step b) The management program 2 instructs the swapper 3 to swap in the swap unit A1 of the process P1 from the secondary storage device 1 to the main storage device process 4.

Step C) The swapper 3 inquires whether the main memory process 4 has enough space to read the swap unit A1.

Step d) As a result, if it is determined that the area can be secured, the data is read from the secondary storage device 1 to the main storage device process 4, the swap-in processing is completed, and execution of the process P1 is resumed.

Step e) If it is found that the fj area cannot be secured, the swapper 3 notifies the management program 2 of the failure of the swap-in process.

Step f) The management program 2 now runs the swapper 3
command to swap out running processes.

Step g) Swamper 3 searches for running processes occupying main memory process 4 and selects one running process,
For example, select process P2. 'Stepped) The swapper 3 reads the first swap unit (B1) of the process P2 and writes it to the secondary storage device 1 (swap out operation).

Step i) Swapper 3 notifies management program 2 of the completion of swap out and returns to step et al. And if it turns out that the necessary area cannot be secured again in step C,
In step g, the next swap unit (B2) will be swapped out.

In the conventional method, as described above, the process is unconditionally swapped out from the top 4, and when that process gains execution rights, the part that has been swapped out is the part that will not be executed after being swapped in. Even if it is, swap-in processing is always performed.In other words, in the case of Figure 1, the restart address of process 1-1 is swap unit A2.
, and even if the secondary and swap unit A1 are not accessed, the swap unit At is always swapped.
The process P1 will be restarted after the process is read in.

Therefore, there is a drawback that unnecessary input/output processing is performed.

[Summary of the Invention 1 The present invention was made with the purpose of eliminating the drawbacks of the conventional method as described above, and the technical means of the present invention to achieve this purpose is a swap-out method. When the process is swapped out, the swap unit is swapped out in sequence starting from the swap unit farthest from the process being executed, and when control is passed to a process that has already been swapped out, the swap unit is swapped out.
Swap execution of said process while it is out.
1m again from the restart address after in, and the swap
A computer operating system swapping method characterized in that a swap-in process is performed for the first time when an out process is executed;
is in Therefore, it is as if 81 of demand paging
Swapping has the ability to realize functionality, reduce unnecessary input/output processing as much as possible, and enable effective system operation.

1 Embodiment J of the invention A preferred embodiment of the invention will be described below with reference to the drawings.

FIG. 2 is a block diagram showing a swapping method for a computer operating system according to the present invention. It is basically the same as the first kl, but for ease of understanding, the process P1 of the main memory process 4 is Assume that a new swap unit A3 is provided.

The swapping operation of such a system is performed as follows.

Step a) Now, the process P1 of the main memory process 4 is swapped to the secondary storage l. - It is assumed that position A3 has been swapped out.

Step b) If process 1]1's I11 open address is in swap unit A1, then management program 2
causes the swapper 3 to resume execution of the process P1 with the swap unit A3 swapped out.

Step C) If swap unit A3 is not executed, processing continues and will eventually be swapped out or process P1 will terminate.

Step d) If you try to change swap unit A3,
Since the management program 2 receives a command that the demand page does not exist, at that point, the management program 2 commands the swapper 3 to swap in the process P1.

Step e) If the swapper 3 finds that the required amount of space (swap unit A3) can be secured in the main memory process 4, it reads from the secondary memory 1 into the main memory process 4 and performs swap-in. End management program 2
The line will be resumed after notification.

Step f) In step C, if it is determined that the required amount of space cannot be secured in the main memory process 4, the management program 2 is notified of the swap-in failure. ? Embedded program 2 instructs swapper 3 to swap execution process 1〕1.
Command out.

Step g) If the process selected by swapper 3 at this time is P2 and the execution address at that time is within swap unit B1, then swapper 3
selects the portion of swap unit B3 that is farthest from swap unit B1 and swaps it out to secondary storage device 1.

Step h) The management program 2 is notified of the completion of the swap out by the swapper 3, instructs the swapper 3 to swap in the process P1 again, and returns to step e. If you still cannot secure enough free space in main memory process 4, then swap unit B2
, and then the swap unit 131 is selected and swapped out.

Step i) In this way process P1 swaps
The program will be read in and executed again (1j).

As described in -1- above, the present invention allows the swap unit being executed at the time of swap out to remain unswapped out until the end, thereby reducing the number of swap ins when control is handed over to that process again. I tried to do it with as few steps as possible.

In the present invention, a simple algorithm is used to select the swap unit to be swapped out. (1) A reference bit is provided for each swap unit, and this reference bit is set each time the swap unit is executed. (2) An algorithm is added to remember the order in which swap units are referenced, and when swapping out, the one with the reference bit of 0 is swapped out first. More effective swapping can be achieved by using techniques such as swapping out from swap units that have been swapped.

[Advantageous Effects of the Invention 1] As described above, according to the present invention, unnecessary input/output processing during swapping is eliminated as much as possible, thereby enabling high-speed and efficient system operation.

[Brief explanation of the drawing]

FIG. 1 is an overall configuration diagram of a conventional method, and FIG. 2 is an overall configuration diagram showing an embodiment of the present invention. 1 is a secondary storage device, 2 is a management program, 3 is a swapper 3, 4 is a main storage process, A1-A3 . B1-B
5 is the swap unit. In the drawings, the same reference numerals indicate the same or corresponding parts. Procedural amendment “Voluntary” Il’$059 M 2. Ha9B Mr. Commissioner of the Japan Patent Office 1, Display of the incident Showa! Contents of amendment to the swapping method of Patent Application No. '144410 of 1999 (1) The scope of claims is amended as shown in the attached sheet. (2) Akira #Iiwa, page 3, lines 4-7 [Partially swapped out... Attempts to execute the process. ' shall be amended as follows. "Attempt to execute process P1 whose swap unit AI (broken line part for process P1, solid line part for secondary storage device 1) has been swapped out." (3) "Process" on page 5, line 20 of the same page. Correct as "swap unit". (4) On page 6, line 3, "It is a swapping method." is amended to "It is a swapping method." (5) Delete "Pl" on page 8, line 5. 2. When swapping out claims, swap out sequentially from the swap unit farthest from the one being executed,
When control is passed to a process that has already been swapped out, the execution of the process that is swapped out is restarted from the rJII address after being swapped in.
n, the swapped out y = mm L1
A computer operating system characterized by performing swap-in processing for the first time when a program is executed.
How to swat the system.

Claims (1)

[Claims] When swapping out, the swap unit is swapped out in sequence starting from the swap unit furthest from where it was being executed.
When control is transferred to a process that has already been swapped out, the execution of the process is resumed from the restart address after the swap-in while being swapped out;
A method for swapping a computer operating system, characterized in that a swap-in process is performed for the first time when the swapped-out process is executed.