JPH02178748A - Virtual storage control method for virtual storage computer system - Google Patents

Abstract

PURPOSE:To improve the efficiency of a page absence processing and a swapping processing by swapping in only the maximum interval of a page allocated to a virtual memory which includes page absence, which includes only the page that does not exist in a present main storage means and which continues on a secondary storage means. CONSTITUTION:A virtual storage computer system is composed of the virtual memory 1, the secondary storage device 2, a real memory 3, a virtual memory data base 4 and a central processing unit 5. When page absence occurs in a logical block which has previously been set, only the interval of the page allocated to the virtual memory 1 which includes the page where page absence concerned is included, which includes only the page that does not exist in the real memory at present and which continues on the secondary storage device is swapped in from the secondary storage device 2 to the real memory 3. Thus, the efficiency of a paging processing is improved and overhead can be reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a virtual memory control method for a virtual memory computing system that performs page absence processing, swapping processing, and the like.

[Conventional technology]

In a conventional virtual memory calculation system, a method described in Japanese Patent Application Laid-Open No. 60-221854 is known as a method for improving the efficiency of paging processing due to page absence.

In this invention, when processing a page fault interrupt, based on a management list of a collection of pages classified according to preset relationships, a plurality of pages including the page that caused the interrupt, that is, logically meaningful By page-ining all pages of a block that are not currently in main memory from secondary storage to main memory, the number of page-ins can be reduced. This was done to improve overhead.

[Problem to be solved by the invention]

By the way, in the conventional page fault interrupt control method described above, it is true that the page (Pi) that caused the page fault interrupt and all the pages (Pjn) that have a preset relationship with that page (Pi) are Although there is a logical relationship, these pages may only be contiguous in virtual memory and may not be elements of a single physical sequence on the secondary storage device. In other words, even if all of the plurality of pages (PJ n) described above are continuous on the virtual memory, they may be discontinuous on the secondary storage device.

Therefore, for example, among pages PI to P5 that are read by executing the page absence processing process, pages P1 to P4 are consecutive on the same disk board of a secondary storage device (for example, a magnetic disk storage device). If only page P5 is allocated on another disk board, when accessing pages P1 to P5 by executing the page absence processing process, pages P1 to P4 are first accessed, and then page P5 is allocated on another disk board. It is necessary to move the access arm in the magnetic disk storage device to access page P5, and it takes time to access the page in the paging process, and the efficiency of the paging process does not necessarily improve.

In addition, as the virtual memory space increases, the amount of information in the database (management list) that contains information on logical connections also increases, so it takes time to search within the management list, resulting in a large overhead. Therefore, there is a drawback that the program execution efficiency decreases.

SUMMARY OF THE INVENTION An object of the present invention is to provide a virtual memory control method for a virtual memory computing system that can eliminate the above drawbacks, improve the efficiency of paging processing, and reduce overhead.

[Means to solve the problem]

In the present invention, only pages that include the page that caused the page fault and that do not currently exist in the main storage means are included;
The maximum interval of pages allocated in contiguous virtual memory on the secondary storage means is swapped and
It was designed so that it would be included in the system.

[Effect]

According to the present invention, when a page fault occurs in a preset logical block, the page contains only the page that caused the page fault and does not currently exist in the main storage means, and Only intervals of pages allocated to continuous virtual memory on the secondary storage means are swapped in from the secondary storage means to the main storage means.

〔Example〕

Embodiments of a virtual memory control method for a virtual memory computing system according to the present invention will be described below with reference to the accompanying drawings.

FIG. 1 is a schematic configuration diagram showing an embodiment of a virtual memory computing system to which a virtual memory control method according to the present invention is applied. In the figure, the virtual memory computing system includes a virtual memory 1, a secondary storage device 2, a real memory 3, and a virtual memory
It has a database 4 and a central processing unit 5.

Programs are allocated to the virtual memory 1 in units of logically significant blocks (hereinafter referred to as logical blocks).

The secondary storage device 2 stores programs and the like in units of logical blocks allocated to the virtual memory 1.

FIG. 2 shows the relationship between the state of the program allocation on the virtual memory 1 and the state of the program when the program is stored on the secondary storage device 2. In FIG. 2, for example, logical blocks L are allocated on the virtual memory 1 in correspondence with physical blocks. The logical block is composed of physical blocks A, B, C and D, and these physical blocks are consecutively allocated on the virtual memory 1.

In addition, in the embodiment, since each of the above physical blocks is set to match a swap unit consisting of logically meaningful unilateral pages, the above physical blocks A, B, C and D is hereinafter referred to as swap units A, B, C, and D; swap units A, B, and C among these swap units;
includes pages 1-2, pages 3-8, and pages 9-10, respectively, as shown in FIG.

As shown in FIG. 1, the virtual memory database (hereinafter referred to as VM database) 4 is configured to be able to associate virtual memory and physical memory as shown in FIG. 2 above. For example, as shown in FIG. 3, it is composed of logical block attribute information 4A and swap unit attribute information 4B.

In the logical block attribute information 4A, corresponding to the information indicating the logical block allocated on the virtual memory (X, Y in FIG. 3), the corresponding logical block is stored on the physical memory. Information indicating where the information is stored is registered. In addition, in the swirler 1 unit attribute information 114B,
Information indicating each swap unit (for example, A, B, C, D, etc. shown in FIG. 3) and page information that constitutes the swap unit (for example, pages 1 to 10 shown in FIG. 2)
Corresponding to this, information indicating where the corresponding swap unit and page are stored in the physical memory is registered.

The page absence processing process that executes the page absence processing by referring to the VM database 4 as described above includes, for example, the page absence processing process A shown in FIG. It consists of b) and (c).

That is, the process (a>) is set to execute VM database processing.In other words, ■Refer to the VM database based on the corresponding page that caused the page failure;
Get swap interval information. ■It is set to execute preprocessing such as 0, which determines the swap interval based on the swap interval information.

Processing (b) is set to execute an I10 operation (input/output operation).

Processing (c) is set to execute post-processing such as real memory acquisition and mapping.

A central processing unit (hereinafter referred to as CPU) 5 executes an operating system program, the page absence processing process, and the like.

Next, the page absence processing operation of the virtual memory computing system will be explained.

Now, an attempt is made to etch the code data included in page 5 in swap unit B in the logical program shown in FIG. 2 from real memory 3, but the page 5
does not exist in the real memory 3 and is page absent, the above-described page absent processing process is executed.

That is, when a page failure occurs, the CPU 5
With reference to the VM database 4 shown in FIG. 3, information indicating the logical block L, information indicating the swap unit B,
At the same time as searching for an entry that contains information indicating page 5, after searching for a corresponding entry, it is found that the page contains page P5 that caused the page absence, contains only pages that do not currently exist in real memory 3, and 2 Next, determine the maximum interval of pages allocated to contiguous virtual memory on the storage device 2.

That is, as shown in FIG. 2, only pages 3 to 8 of swap unit B are selected. These selected pages 3 to 8 are swapped in from the secondary storage device 2 to the real memory 3.

Next, the CPU 5 allocates the real memory 3, reads pages 3 to 8 to be swapped in from the secondary storage device 2, and swaps them into the storage area allocated to the real memory 3. When a page that has been page-faulted in this way is paged in, the scheduler is activated and the page-fault handling process enters a standby state.

The page absence processing when one process is running in the system has been described above, but the page absence processing when a plurality of processes are running in the system will now be described.

FIG. 4 is a state diagram showing the overlay state of the page fault processing process. The page absence processing processes A and B are composed of page absence processing processes (a), (b), and (c), as explained in the page absence processing process A above.

As shown in Figure 4, process X that caused the page fault
(not shown), the page absence processing process A is executed, and I10 output operations such as swap-in from the secondary storage device 2 to the real memory 3 (process (b) shown in FIG. 4) are performed. If the process Y (not shown) causes a page fault while the process is running, the above process (a) of the page fault process, that is, the preprocessing of the page fault process, is executed by executing the page fault processing process B. It can be carried out.

By providing a plurality of page absence processing processes in this way, it is possible to reduce the overhead required to execute the page absence processing processes.

Also, if the swap out process also references the swap database table, i.e. I1010
0 requires a lot of preprocessing, for example, ρ1 is proportional to the size of the virtual memory space, and the virtual memory and physical memory (
If the amount of data that must be handled during swap-out processing increases, such as data indicating correspondence with secondary storage (secondary storage), multiple swap-out processes should be established to perform swapping in units of logical blocks. You can do it like this.

According to the embodiment described above, the virtual memory database is referenced and the corresponding page is swapped in according to the reference result, so that it is possible to improve the efficiency of page absence processing.

Further, according to the above embodiment, overhead can be reduced by providing a plurality of page fault processing processes and a plurality of swap out processes.

〔Effect of the invention〕

As explained above, according to the present invention, the virtual memory that includes only the page that caused the page fault and does not currently exist in the main memory, and that is continuous on the secondary memory Since only the maximum interval of pages allocated to the page is swapped in from the secondary storage means to the main storage means, it is possible to improve the efficiency of page absent processing and swapping processing.

[Brief explanation of the drawing]

FIG. 1 is a schematic configuration diagram showing an example of a virtual memory computing system to which the virtual memory control method according to the present invention is applied, and FIG. 2 shows the state in which programs are allocated to virtual memory and secondary storage devices. Fig. 3 is a diagram showing the relationship between virtual memory and virtual memory database; Fig. 4 is a diagram showing the relationship between virtual memory and virtual memory database;
The figure is a state diagram showing the state of the overlay of the page absence handling process. 1... Virtual memory, 2... Secondary storage device, 3...
Real memory, 4...Virtual memory database, 5...
・Central processing unit.

Claims (1)

[Claims] In a virtual memory computing system having a main storage means for storing a program divided into predetermined page units and a secondary storage means, there is provided a virtual memory computing system in which a page absence occurs within a preset logical block. When a page fault occurs, the page is allocated on virtual memory that includes only the page that caused the page fault and does not currently exist in the main memory, and that is continuous on the secondary memory. 1. A virtual memory control method for a virtual memory computing system, characterized in that the maximum interval of pages read is subject to swap-in.