JPH113271A - Memory allocation system in computer system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten the time required for allocation when a zero-cleared memory page is required in a computer system. SOLUTION: A computer system manages a memory by a page unit and allocates pages corresponding to a request. A load monitoring mechanism 1 detects whether or not the load of the computer system is low and activates a page clear mechanism 2 when the load of the computer system is low. The page clear mechanism 2 zero-clears the page in a free state among the pages. When the zero-cleared page is requested, a page allocater 5 allocates the zero- cleared page.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a computer system, and more particularly, to a memory allocation method used in a computer system.

[0002]

2. Description of the Related Art Generally, in a computer system, an operating system is provided with a mechanism (function) for managing a memory in page units and allocating the memory in response to a request. On the other hand, when the operating system operates, the page that is zero-cleared is often required. Thus, when the page that is zero-cleared is required, the operating system performs zero-clearing when allocating the page. .

When the entire page is cleared to zero, it is necessary to write data over a relatively wide range. For this reason, it takes time to perform the process itself when the entire page is cleared to zero.

[0004] In addition, when the entire page is cleared to zero, a large number of cache replacements of the processor occur. For this reason, data that has been cached before the page clearing process is replaced, and in a later process, reloading from the memory to the cache may occur, thereby deteriorating the performance of the computer.

[0005]

As described above, in the conventional computer system, the page clearing process takes a long time because the page clearing process is performed when a zero-cleared page is required. There is.

Further, in the present invention, when allocating a zero-cleared page, a page requiring a large amount of writing to the memory is cleared, so that the cache is frequently replaced when performing the zero-clear processing, When the subsequent processing is continued, there is a problem that re-reading from the memory is required and computer performance is reduced.

An object of the present invention is to provide a memory allocation method capable of reducing the time required for allocating when a zero-cleared memory page is required in a computer system.

[0008]

According to the present invention, in a computer system for managing a plurality of memory areas and allocating according to a request, the computer system detects whether the load on the computer system is low and detects the load. First means for zero-clearing a free memory area of the memory area when the load of the system is low, and second means for allocating the zero-cleared memory area when the zero-cleared memory area is requested A memory allocation method in a computer system having the second means.

Specifically, there are provided first pool means for pooling a memory area in a free state, and second pool means for pooling the memory area that has been zero-cleared. Extracting the memory area in the free state to be allocated from the first pool means, clearing the contents of the memory area to zero,
And a load monitoring mechanism that detects whether the load on the computer system is low and activates the clear mechanism when the load on the computer system is low. Means for allocating the zero-cleared memory area from the second pool means when it is determined that the zero-cleared memory area is required upon receiving the request, and the zero-cleared page is requested. If it is determined that there is no memory area, the memory area in the free state is allocated from the first pool means.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below with reference to the drawings.

Referring to FIG. 1, in a computer system (not shown), an operating system manages a memory in page units. The computer system is provided with a load monitoring mechanism (system load monitoring mechanism) 1, and the system load monitoring mechanism 1 monitors the load status of the computer system. Then, the system load monitoring mechanism 1 determines whether the current system load is high or low.

When the system load monitoring mechanism 1 determines that the system load is low, the page zero clearing mechanism 2 is activated. The page zero clearing mechanism 2 extracts a page in a free state which is to be allocated when the next page is requested from the free page pool 3, and clears its contents to zero. Then, this page is moved to the zero-cleared page pool 4.

When the page requester 6 needs a zero-cleared page, the page requester 6 requests the zero-cleared page from a page allocator (page allocator) 5. When determining that the zero-cleared page is needed, the page allocating mechanism 5 takes out the page from the pool of cleared pages prepared by the page zero clearing mechanism 5 and returns it to the requester 6. On the other hand, when the zero-cleared page is not requested, the page allocating mechanism 5 returns a page in a free state from the free page pool 3.

Here, a UNIX operating system will be described as a specific example.

In the UNIX operating system,
When there are no more runnable processes, a transition is made to a state called idle. This idle state means that the system load is low, and the transition to this state triggers the load monitoring mechanism 1 to reset the page zero clear mechanism 2
Start

The page zero clear mechanism 2 takes out a page from the free page pool 3 and clears the contents to zero. The cleared page is added to a dedicated pool (cleared free page pool) 4.

The above processing is continued until a runnable process appears.

When an application program (requester) requests a BSS (zero-cleared data area), the operating system (OS) allocates a page that has been cleared to zero to the application program. At this time, a page that has been cleared to zero is requested to the page allocation mechanism 5.

The page allocating mechanism 5 takes out the page whose contents have been zero-cleared by the page zero clearing mechanism 2 from the cleared free pool 4 and returns it to the requester 6. if,
If the cleared pool is empty, free page pool 3
The page in the free state is taken out from, the content is cleared to zero, and returned to the requester 6.

The above-mentioned memory (page) allocation can be applied to a system in which it is expected that a memory having a specific content is required. In this case, the page zero clear mechanism 2 shown in FIG. 1 is replaced with a mechanism for writing specific contents. This writing mechanism can speed up the process of allocating a memory of a specific content.

[0021]

As described above, according to the present invention, the zero clear processing of a page is performed only when the load on the system is low, and the zero clear processing of the page is not performed at the time of page allocation. Page clearing process at the time of page allocation like this,
This has the effect of speeding up the actual page allocation process.

Further, according to the present invention, the page zero-clear processing is performed only when the system load is low, and the page zero-clear processing is not performed at the time of page allocation. This has the effect of speeding up the page allocation process and subsequent processes.

[Brief description of the drawings]

FIG. 1 is a block diagram for explaining a memory allocation method according to the present invention.

[Explanation of symbols]

 1 load monitoring mechanism (system load monitoring mechanism) 2 page zero clear mechanism 3 free page pool 4 zero cleared page pool 5 page allocating mechanism (page allocator) 6 page requester

Claims (4)

[Claims] In a computer system that manages a plurality of memory areas and performs allocation in response to a request, it is detected whether or not the load on the computer system is low. A second means for zero-clearing a memory area in a free state, and a second means for allocating the zero-cleared memory area when the zero-cleared memory area is requested. And a memory allocation method in a computer system. 2. The memory allocation method in a computer system according to claim 1, wherein said first pooling means pools a memory area in a free state, and said second pooling means pools said zero-cleared memory area. Wherein the first means takes out the memory area in the free state which is to be allocated from the first pool means, clears the contents of the memory area to zero, and sends the memory area to the second pool means. Memory allocation in a computer system, comprising: a clear mechanism that can be shifted; and a load monitoring mechanism that detects whether the load on the computer system is low and activates the clear mechanism when the load on the computer system is low. method. 3. The memory allocation method according to claim 2, wherein said second means, upon receiving said request, determines that said zero-cleared memory area is required. Allocating the zero-cleared memory area from the pool means, and allocating the free memory area from the first pool means when it is determined that the zero-cleared page is not requested. Memory allocation method in a computer system. 4. In a computer system having a specific memory area having predetermined contents by managing a plurality of memory areas and allocating according to a request, it is detected whether or not the load of the computer system is low. A memory allocation method in a computer system, comprising: means for allocating the contents of the specific memory area in response to the request when the load on the computer system is low.