JPH0520201A - Virtual storage management system - Google Patents

Abstract

PURPOSE:To reduce the generation of overhead due to the substitution of pages by effectively utilizing unused pages in a main storage device. CONSTITUTION:Since a cache memory space can be dynamically set up in the main storage device 11 by utilizing an idle area in the device 11, a page to be applied to page out is stored in the device 11 as it is without being saved to a secondary storage device 12. An unused page is newly allocated to a working set generating a page-out request. When page-in is required, whether its corresponding page exists in a page-out page group stored in the device 11 as a cache memory space or is checked, and when the corresponding page exists, the page is allocated to the working set issuing the page-in request without accessing a page from the device 12.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a virtual memory management system for realizing a multiple programming environment on an electronic computer, and in particular, a plurality of working sets respectively corresponding to a plurality of jobs are assigned to a main memory device so that the main memory is The present invention relates to a working set type virtual memory management system for performing page replacement between a device and a secondary storage device.

[0002]

2. Description of the Related Art Generally, in an electronic computer system, a logical address space larger than the real address space of the main memory is realized so that a program can be created without being aware of the upper limit of the real memory capacity on the main memory. A virtual memory system is used. In particular, a working set type virtual memory management system is most widely used as a virtual memory management system for realizing a multiple programming environment.

The working set type virtual memory management system allocates a working set to the main storage device for all jobs (active jobs) executed in parallel, and the working set is allocated between the main storage device and the secondary storage device. Page replacement is performed for each working set. Here, the working set is a storage range used for storing the page referred to by the corresponding job on the main storage device.

That is, a job using a specific working set allocated on the main storage device uses the working set as a window, reads a page to be referenced from the secondary storage device, and sets it as the working set. Store sequentially. In this case, if there is no unused page in the working set corresponding to the job, the page to be replaced is selected from the working set, paged out and saved in the secondary storage device. It Then, when the page saved in the secondary storage device becomes necessary, the page is recalled from the secondary storage device to the main storage device again.

Usually, accessing such a secondary storage device requires more time than accessing the main storage device, and thus becomes a large overhead for the virtual storage management system. For this reason, it is preferable that the number of page saves and page calls to the secondary storage device be minimized.

However, in the conventional virtual memory management system, page replacement is performed separately for each working set, so that even if an unused page exists in another working set on the main memory, When there is no unused page inside the corresponding working set, the page-out target page must be saved in the secondary storage device without fail. For this reason, there is a drawback in that the secondary storage device is frequently accessed due to the page replacement as described above, which increases the overhead.

[0007]

Conventionally, since page replacement is performed separately for each working set, even if an unused page exists in the main storage device, the page is stored in the secondary storage device. There is a drawback that it needs to be saved and the overhead is increased.

The present invention has been made in view of the above circumstances, and it is possible to effectively use an unused page in the main storage device and reduce the overhead by replacing the page. The purpose is to provide.

[0009]

Means and Actions for Solving the Problems
A page-out request is generated in a virtual memory management system of the working set system in which a plurality of working sets respectively corresponding to a plurality of jobs are assigned to the main storage device and pages are replaced between the main storage device and the secondary storage device. At this time, it is detected whether or not an unused page exists in the plurality of working sets, and when the presence of an unused page is detected, the page-out target page is stored in the main memory together with A means for allocating the detected unused page as a part of the working set that issued the page-out request, and a page-out target page group stored in the main memory when the page-in request occurs. Detects whether or not the page-in target page exists, and when the existence is detected, the page-in target The first; and a means for assigning a chromatography di as part of the working set that issued the page-in requests.

In this virtual memory management system, when there is an unused page in any of the plurality of working sets allocated on the main memory, the page-out target page is the secondary memory. It is stored in the main memory as it is without being saved in. Then, the unused page is newly allocated to the working set that issued the page-out request. Therefore, even when page-out is required, the overhead due to paging can be reduced by the amount of saving to the secondary storage device. When a page-in is required, it is first checked whether or not there is a corresponding page in the page-out target page group stored in the main storage device. If there is, a secondary storage device is found. The page is assigned to the working set that issued the page-in request without calling the page from. Therefore, even when page-in is required, if the page is in the page-out target page group, the call from the secondary storage device is unnecessary, and the overhead due to paging can be reduced accordingly.

Also, in the virtual memory management system of the present invention, when there are no unused pages in the plurality of working sets, the page-out target page stored in the main memory is stored in the secondary memory. It is characterized by further comprising means for page-out. In this virtual memory management system, since the accumulated page-out target pages can be released to each working set, the number of usable pages can be equally guaranteed for all working sets.

[0012]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows the system configuration of a virtual memory management system according to an embodiment of the present invention. This virtual memory management system is for efficiently realizing a multiple programming environment in a computer system, and a plurality of jobs corresponding to a plurality of jobs (job 1, job 2, job 3, ...) Working in parallel are respectively provided. The working sets WS1, WS2, WS3, ... Are reserved on the main storage device 11. Each job is stored in the main storage device 11
Using the corresponding working set above as a window,
A program or data in the virtual memory space defined in the secondary storage device 12 is referred to.

Working set WS on the main memory 11
The real memory space of each of 1, WS2, and WS3 is, for example, 25
It is 6 Kbytes, and the virtual memory space of each of the virtual working sets WS1 ', WS2', WS3 'on the secondary storage device 12 is 4 Mbytes, for example. The storage area of the real memory space is managed in an allocation unit called a page. One page is a continuous fixed-length area of 4 Kbytes, for example. The storage area of the virtual memory space is also managed by pages of the same size as the real memory space.

In FIG. 1, for simplification of description, each real memory space of the working sets WS1, WS2, WS3 has a storage area for four pages, and virtual working sets WS1 ', WS2', WS3. It is shown that each virtual memory space of 'has storage areas for n (n> 4) pages.

That is, for job 1, n pages of virtual address space from page P1-1 to page P1-n defined in the virtual working set WS1 'can be used for programming, and the n pages Pages required in the program are stored in the working set WS1 on the main storage device 11. The contents of page 4 of Working Set WS1 are Virtual Working Set WS
It is updated as necessary by replacing the page with 1 '. By this page replacement, arbitrary 4 pages out of n pages of the virtual working set WS1 'are stored in the working set WS1. Therefore, the job 1 uses the four pages of the working set WS1 as a window to generate the virtual working set WS1.
Reference can be made to all n pages of ‘′.

The page replacement is performed by reading a page from the virtual working set WS1 'of the secondary storage device 12 to the working set WS1 of the main storage device 11 (page-in).
And page writing (page out) from the working set WS1 of the main storage device 11 to the virtual working set WS1 ′ of the secondary storage device 12. The virtual memory management unit 13 controls such page replacement by page-in and page-out.

Also for job 2, as for job 1,
Page P2 defined in virtual working set WS2 '
It is possible to use the virtual address space for n pages from -1 to page P2-n, and the necessary page among these n pages is the working set WS on the main storage device 11.
Stored in 2. Also for job 3, n pages of virtual address space from page P3-1 to page P3-n defined in the virtual working set WS3 'can be used, and necessary pages among these n pages are available. Are stored in the working set WS3 on the main storage device 11.

In each of job 2 and job 3, page replacement by page-in and page-out is executed. This page replacement is controlled by the virtual memory management unit 13.

The virtual storage management unit 13 manages the working sets WS1 to WS3 on the main storage device 11 in page units and controls page rewriting in each of the working sets WS1 to WS3.

In this page rewriting control, the virtual memory management unit 13 is configured to reduce access to the secondary storage device as much as possible when there are unused pages in the working sets WS1 to WS3. The cache memory space is dynamically set on the main storage device 11 by using the unused pages. This cache memory space temporarily holds a page-out target page to the secondary storage device 12. The setting of the cache memory space by the virtual memory management unit 13 is performed as follows.

For example, as shown in FIG. 1, four pages of pages P1-1 to P1-4 are stored in the storage area of four pages of the working set WS1, and the storage area of all four pages is stored. Is used, only two pages of the storage area for four pages of the working set WS2 are in use, and the remaining two pages are unused.

In this situation, when the job 1 requests to refer to the n-th page P1-n, for example, a page fault occurs because the page P1-n is not stored in the working set WS1. When a page fault occurs, the page P1-n is transferred to the secondary storage device 1
It is necessary to page in from 2. However, since there is no free area in the working set WS1, in order to generate the free area, the page that is least likely to be used among the pages P1-1 to P1-4 (for example, page P1-
It is necessary to select 1) according to an algorithm such as LRU and page out the page to the secondary storage device 12. Normally, the page P1-1 targeted for page-out is the other working set WS2, WS3.
Secondary storage device 1 with or without unused pages in
Paged out to 2.

However, here, the working set W
Since there is an unused page in S2, the virtual memory management unit 1
When the page-out is required, the page-out target page P1-1 is directly pooled in the main storage device 11 without being paged out. Then, the virtual memory management unit 13 allocates, for example, a storage area corresponding to the fourth page P2-4 of the working set WS2 as a free area of the working set WS1 that needs to be paged out, and the page P1-n is set in the free area. Page in.

In this case, the virtual memory management unit 13 manages the pages P1-2, P1-3, P1-4, and P2-4 as pages of the working set WS1. Therefore, pages P1-2 and P1-3 on the secondary storage device 12,
The virtual addresses of P1-4 and P1-n are respectively assigned to pages P1-
2, P1-3, P1-4, P2-4 are converted into real addresses. Therefore, the job 1 can refer to the page P2-4 on the main storage device 11 by designating the virtual address for the page P1-n.

The page P1-1 to be paged out
Is managed as a cache memory space by the virtual memory management unit 13. That is, when page-in is required, it is checked whether or not the corresponding page exists in the page-out target page group managed as the cache memory, and if there is, the secondary storage device. The page to be paged in is fetched from the cache memory space without page-in from 12, and it is allocated to the working set which needs page-in.

Further, the virtual memory management unit 13 has a function of releasing the cache memory space set in the main memory device 11. That is, since the cache memory space is set by utilizing the existence of unused pages, when there are no unused pages in the main storage device 11, the page-out target pooled in the main storage device 11 is set. The page group is paged out to the secondary storage device 12. For example, pages P2-3 and P2 of working set WS2
-4 is used for another working set, when the page-in of a new page is required by the request of job 2, the page is not managed in the working set WS2 but is managed as the cache memory space. The existing page is preferentially paged out, and the area is set as an empty area of the working set WS2. As a result, the storage area used as the cache memory space can be released to each working set, and the number of usable pages can be equally guaranteed for all working sets.

FIG. 2 shows an example of a specific configuration of the virtual memory management unit 13. As illustrated, the virtual memory management unit 13 includes a cache memory management table 21, an unused page management table 22, a working set WS1 management table 23, a working set WS2 management table 24, and a working set WS3 management table 25.
Is equipped with.

The cache memory management table 21 is for managing the correspondence relationship between the virtual address and the real address for each page group targeted for page-out pooled in the main storage device 11 as a cache memory space in page units. is there. Unused page management table 22
Manages, in page units, the correspondence between virtual addresses and real addresses of unused pages in all working sets WS1 to WS3 ... Allocated on the main storage device 11.

Management table 2 for working set WS1
3, the working set WS2 management table 24, and the working set WS3 management table 25 have a storage capacity (256 Kbytes) in the main storage device 11 that can be used for each corresponding working set and a storage capacity that is actually being used. And the storage area in use holds the virtual address and the real address of each page.

Here, as in FIG. 1, the usable storage capacity of the working sets WS1 to WS3 is 4 pages, and the working set WS1 has page P1-1.
To P1-4 are all in use, and for working set WS2, pages P2-1, P2-
Consider the case where 2 is in use and the remaining pages P2-3 and P2-4 are unused.

In this case, the management table 23 for the working set WS1 manages the addresses of all the pages P1-1 to P1-4 in use, and the working set W1.
In the S2 management table 24, addresses are managed only for the pages P2-1 and P2-2 in use, and addresses of unused pages P2-3 and P2-4 are managed in the unused page management table 22.

Under these circumstances, the working set WS
When it becomes necessary to page out page P1-1 of No. 1, the page out target page P1-1 is transferred from the working set WS1 management table 23 to the cache memory management table 21, and the cache memory space Managed as. Also, unused pages P2-4
Is transferred from the unused page management table 22 to the working set WS1 management table 23, and is managed as a storage area of the working set WS1.

FIG. 3 shows an example of the usage pattern of the main storage device 11 in the virtual storage management system of FIG. As shown in the figure, the working sets WS1 to WS3, ... Are allotted to the main storage device 11, and unused pages in these working sets WS1 to WS3 are managed as free areas 110. The cache memory space 111 temporarily holds a page-out target page of a working set that uses a part of the free area 110.

Here, the working sets WS2, WS
The page P1-n is stored in the empty area 110 composed of three unused pages, and the page P1-n is used by the working set WS1. Then, the page P1-1 targeted for page-out of the working set WS1
Are registered and managed in the cache memory space 111.

As described above, in this virtual memory management system, by effectively using the free area 110 on the main memory device 11, the page-out target page is set to two.
It is possible to manage it as a cache memory space 111 by pooling it in the main storage device 11 without saving it in the next storage device 12.

Therefore, even when page-out is required, saving to the secondary storage device 12 is unnecessary,
The overhead due to paging can be reduced. Also,
When page-in is required, it is first checked whether or not there is a corresponding page in the page-out target page group pooled in the cache memory space 111, and if there is, the secondary storage device 12 Is unnecessary and the paging overhead can be reduced accordingly. Next, the operation of the virtual memory management unit 13 of FIG. 1 will be described with reference to the flowcharts of FIGS. First, referring to the flowchart of FIG.
The operation when a page out to the next storage device 12 is required will be described.

For example, when a page out is required for the working set WS1, the virtual memory management unit 13
Refers to the unused page management table 22 to check whether unused pages exist in the other working sets WS2 and WS3 (step S11). If there is no unused page, the page to be paged out is saved in the secondary storage device 12, and the working set WS1
A free area is generated in the area (step S12).

On the other hand, when there is an unused page,
The pages to be paged out are directly pooled in the main storage device 11, and the management thereof is transferred from the working set WS1 management table 23 to the cache memory management table 21 (step S13). Then, by registering the unused page in the management table 23 for the working set WS1, the unused page is stored in the working set WS1.
Is assigned as a new page (step S14).
Next, the operation when page-in is required will be described with reference to the flowchart of FIG.

For example, when a page fault occurs in the working set WS1, the virtual memory management unit 13
Refers to the cache memory management table 21 and checks whether or not a page-in target page exists in the cache memory space 111 (step S21). If the page-in target page does not exist, the page-out target page is called from the secondary storage device 12 and paged in to the working set WS1 (step S).
22).

On the other hand, if a page-in target page exists in the cache memory space 111, the page-in target page is managed from the cache memory management table 21 to the working set WS1 management table 23.
The page to be paged in is assigned to the working set WS1 (step S23). Next, FIG.
The operation in the case where the free space in the main storage device 11 is insufficient will be described with reference to the flowchart of FIG.

The storage area used as the cache memory space 111 is originally a storage area of another working set having an unused page. For this reason,
If there is not enough free space, cache memory space 1
It is necessary to release the storage area used as 11.

Therefore, when the free space is insufficient, the virtual memory management unit 13 first refers to the cache memory management table 21 to refer to the main memory device 1.
It is checked whether or not there is a page managed as the cache memory space 111 on 1 (step S3).
1). When there is a page managed as the cache memory space 111, the virtual memory management unit 13
The page is saved to the secondary storage device 12 (step S
32). Then, the virtual memory management unit 13 manages the pages managed as the cache memory space 111,
It moves from the cache memory management table 21 to the unused page management table 22 and allocates it as an unused page (step S33).

As described above, in this embodiment, the free space on the main storage device 11 is utilized to make the main storage device 1
1, the cache memory space 111 can be dynamically set, and the page-out target page is directly stored in the main storage device 12 without being saved in the secondary storage device 12. Then, the unused page is newly allocated to the working set that issued the page-out request. Therefore, even when page-out is required, the overhead due to paging can be reduced as much as saving to the secondary storage device 12 is unnecessary.

When page-in is required,
First, it is checked whether or not there is a corresponding page in the page-out target page group stored in the main storage device as the cache memory space 111, and if there is, the page is called from the secondary storage device 12. None, the page is assigned to the working set that issued the page-in request. Therefore, even when page-in is necessary, if the page is in the page-out target page group, the call from the secondary storage device 12 is unnecessary, and the overhead due to paging can be reduced accordingly.

[0046]

As described above, according to the present invention, unused pages on the main storage device can be effectively used, and the overhead due to page replacement can be reduced.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a virtual memory management system according to an embodiment of the present invention.

FIG. 2 is a block diagram showing an example of a specific configuration of a virtual storage management unit provided in the virtual storage management system of the same embodiment.

FIG. 3 is an exemplary block diagram showing an example of a usage pattern of a main storage device in the virtual storage management system of the embodiment.

FIG. 4 is a flowchart illustrating an operation when a page-out is required in the embodiment.

FIG. 5 is a flowchart for explaining an operation when a page-in is required in the embodiment.

FIG. 6 is a flowchart for explaining the operation when the unused memory is insufficient in the embodiment.

[Explanation of symbols]

11 ... Main storage device, 12 ... Secondary storage device, 13 ... Virtual storage management device, 21 ... Cache memory management table, 2
2 ... Unused page management table, 23-25 ... Working set management table.

Claims (2)

[Claims] 1. A working set type virtual storage management system for allocating a plurality of working sets respectively corresponding to a plurality of jobs to a main storage device and performing page replacement between a main storage device and a secondary storage device, When a page-out request occurs, it is detected whether or not an unused page exists in the plurality of working sets, and when the existence of an unused page is detected, the page-out target page is set to the main storage device. Accumulated on top of
A means for allocating the detected unused page as a part of the working set that issued the page-out request, and a page-in target page among the pages accumulated in the main memory when the page-in request occurs. Virtual memory management, comprising: means for detecting whether or not the page exists, and allocating the page-in target page as a part of the working set that issued the page-in request when the existence is detected. system. 2. When the unused pages in the plurality of working sets are exhausted, a page is stored in the main storage device and is paged out to the secondary storage device. The virtual memory management system according to claim 1.