JPH0619787A - Storage management device - Google Patents

Abstract

PURPOSE:To improve the processing efficiency of an operating system by making a copy processing be irreducible minimum in copying at the time of writing. CONSTITUTION:This system is provided with a cache memory device 3 having a function which does not let cached data be the object of replacement when the address of the cached data is included in the page address held by the address holding means 5. by mapping the page of a copy source at the time of reading, by mapping the page of a copy destination at the time of writing and by setting the page address of the copy destination in an address holding means 5 for an address outputted from a processor 10 by an address converter 4. Thus, effect equal to copying can be obtained without immediately copying whole data of the page and data transfer at the time of copying can be reduced.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a storage management system in an operating system.

[0002]

2. Description of the Related Art In recent operating systems, as shown in the document "Maekawa, Tokoro, Shimizu Edition, Distributed Operating System, Kyoritsu Shuppan, pp.34",
It provides a copy-on-write function. This is because when the address space is duplicated, the entire space is not immediately copied, but the address space is write-protected and a page in the space is shared so that any page in the space can be shared. When a change due to writing occurs, a new page is allocated and only that page is copied. According to this method, only the data included in a required page needs to be copied, so that it is widely used in various places of storage management.

FIG. 4 is a diagram for explaining the conventional storage management system as described above. In the figure, 20 is an address map in the main memory, and 21 is an address map in the processor. Reference numerals 41 and 42 are pages provided in the address map of the processor. Reference numerals 51 and 52 are pages provided in the address map in the main memory.

Next, the operation will be described. Pages 41 to 42 of the address map 21 in the processor
Description will be made regarding a case where an instruction to copy is issued to. When the page 41 is copied to the page 42, the entire area is not immediately copied and the page 51 provided in the main memory corresponding to the page 41 is set as the write-protected page. At this point, the actual data is not copied from page 51 to another page. After that, the read request to the page 41 and the read request to the page 42 can be dealt with by mapping the page 51 in which writing is prohibited.

On the other hand, when a write request is made to page 41 or page 42, page 52 is assigned as a new page to be copied corresponding to page 42. Then, the data of page 51 is actually copied to the newly allocated page 52. After that, when there is a write request to page 41, the write is performed to page 51, and when there is a write request to page 42, the write is performed to page 52.

[0006]

In the conventional copy-on-write method, it is necessary to copy the entire page even when only part of the page is changed. Since the copy process is performed by the CPU, the effect of this on the processing efficiency of the operating system cannot be ignored. Further, from the viewpoint of the efficiency of input / output processing, it is desirable that the size of one page is large, but this also increases the load of copy processing.

[0007]

A storage management system according to the present invention has the following elements. (A) A data storage unit that manages and stores data in units of predetermined areas, (b) a processor that accesses data in the data storage unit, and (c) data that the processor accesses between the data storage unit and the processor. (D) When there is an instruction to copy the data in the first area of the data storage section to the second area,
When the second area is allocated and the processor performs a write operation on the address in the second area, the corresponding address in the second area is allocated, and the processor allocates the address in the second area. When performing a read operation, the first address corresponding to the address of the second area
Address conversion means for allocating the address of the area
(E) When the processor accesses an address in the second area of the data storage unit based on a predetermined process,
Access data storage means for storing the data accessed by the address converted by the address conversion means in the temporary storage portion, (f) after the predetermined processing of the processor is completed, the access data storage means stores the data in the temporary storage portion. Matching means for generating data in the second area allocated by the address converting means using the stored data and the data in the first area.

[0008]

The storage management system according to the present invention maps the data in the first area when the data in the second area is read by the address conversion means. Meanwhile, the second
The data in the second area is mapped to the writing in the area. Then, the access data storage means stores the data read from the first area in the temporary storage unit when the second area is read. Similarly, when writing is performed in the second area, a portion related to the writing is read from the first area into the temporary storage unit, and writing is performed in the temporary storage unit.
The written data is saved in the temporary storage unit. The data read from the first area and the data read from the first area and rewritten are stored in the temporary storage unit and are not erased. Then, the matching means creates the latest data in the second area using the latest data in the temporary storage section and the data in the first area after the processor finishes the predetermined processing. By the above operation, it is not necessary to immediately copy the data in the first area as the copy source to the second area as the copy destination. Further, when the data stored in the temporary storage unit is released after the predetermined process of the processor is completed, the data stored in the temporary storage unit and the data in the first region are used for the second area. There is no need to create a new work, and the operation of copying work is unnecessary.

[0009]

【Example】

Example 1. Next, the present invention will be described with reference to the drawings. FIG. 1 is a diagram showing the configuration of a computer system related to the storage management system of the present invention. In FIG. 1, 1 is a CPU, 2 is a first address translation device, 3 is a cache memory, 4 is a second address translation device, 5 is an address holding means, and 6 is a main Memory,
10 is a processor. FIG. 2 is a diagram illustrating an address map according to an embodiment of the present invention. Reference numeral 20 is an address map in the main memory, and 21 is an address map in the processor. FIG. 3 shows a processing procedure of the storage management method of the present invention.

An embodiment will be described below with reference to FIG.
FIG. 3 shows a case where a copy command is issued from page 41 to page 42, and shows a state in which page 51 corresponding to page 41 is write-protected. That is, the write-protected page 51 is mapped to both the read command for the page 41 and the read command for the page 42. In the flow chart of FIG. 3, the start time is when the writing to the write-protected page 51 occurs in such a state.

When a write operation is made to a page 51 which is write-protected by the operating system to copy the address space, the copy-on-write process is started from step 31. First, in step 32, a new page 52 to be copied is allocated. Step 33
Then, the address translation device 4 is set to map the page 51 for reading the page 52 and map the page 52 for writing the page 52.
The address map at this time is shown in FIG. At this point, only mapping is done and data is not copied. Subsequently, in step 34, the address 62 of the page 52 is set in the address holding unit 5, and the data of the page 52 is excluded from the cache replacement target in the cache memory 3. Here, the meaning of excluding it from the cache replacement target means that when the data of the page 52 is read into the cache memory 3, the data is continuously stored in the cache. Normally, the data read into the cache memory is replaced by the new data by the old order or other algorithm, but in this case, the data of page 52 is erased by other data once read from the main memory into the cache memory. It means that there is no chance that it will end up. In step 35, address 63 in page 52
When a write to the cache memory 3 occurs, the cache memory 3 reads the data at the corresponding address 55 in the page 51 from the main memory 6, and the processor 10 writes to the cache memory 3. Since this data is handled as data included in page 52 in cache memory 3, it is not the target of cache replacement due to the setting in step 34.

When a read operation is performed on the address 64 in the page 52, the cache memory 3 reads the data at the corresponding address 56 in the page 51 from the main memory 6, and the processor 10 reads from the cache memory 3. . This data is stored in the cache memory 3
Since it is treated as data included in page 52,
Due to the setting in step 34, the cache is not replaced. When writing and reading with respect to the address in the page 52 occur in this way, the data corresponding to the address is read into the cache memory, and in the case of reading, the data is saved as it is in the cache memory. In the case of writing, writing is performed in the cache memory. At this point, the actual data has not yet been copied to page 52.

After that, the process in the case where the page 52 is released in step 36 or the cache memory 3 and the main memory 6 need to be matched with each other will be described below. Here, the release means a case where it is not necessary to write back the data of the page 52 in the cache memory 3 to the main memory 6. Also, here, the matching means page 5 in the cache memory 3.
This is a case where it is necessary to write back the data of No. 2 to the main memory 6. The cases of release and matching will be described separately below.

First, when the page 52 is released, in step 37, the page 52 to the address holding means 5 is released.
Of the cache memory 3 in step 38.
In, the data of page 52 is invalidated. Step 3
At 9, the address translator 4 is canceled with the settings so far and the same page is mapped for reading and writing.

Next, when it is required to match the cache of the page 52 with the main memory, in step 37,
The setting of the page 52 for the address holding unit 5 is released. Then, in step 41, the latest data is created for all the addresses of page 52. That is, the data before the processor performs the predetermined work is left in the page 51, and the data read from the page 51 and the written data are present in the cache memory 3. By combining the data stored in the cache memory and the data stored in the cache memory, the latest data as the page 52 can be configured. Step 41
Then, the original data on page 51 is stored in the cache memory 3
It will read in and create the latest data. At step 42, the data of page 52 in the cache memory 3 is flushed. Here, flushing means writing the data of page 52 in the cache memory 3 to the main memory 6. Then, in step 39,
The same page is mapped to the address translation device 4 for reading and writing, and the previous settings are canceled.

As described above, according to this embodiment, the processor is composed of only the CPU, or the CPU and the first address translation device connected to the CPU and used for normal virtual memory management, and the address output from the processor. On the other hand, it has a second address translation device that provides different mappings for write operation and read operation, a main memory that operates by the address output from the second address translation device, and an address holding unit for a designated page, A computer system having a write-back cache memory device having a function of not replacing the cache data when the address of the cached data is included in the address held by the holding means. Operating system, without immediately copying the data, Characterized in that to obtain the same effect as P..

According to this embodiment, the second address translation device maps the copy source page at the time of reading and maps the copy destination page at the time of writing. Furthermore, by setting the address of the copy destination in the address holding means, it is possible to prevent the data contained in the copy destination page from being excluded from the cache replacement target and to be written back to the copy destination page. Thereby, without immediately copying the data of the copy source page to the copy destination page, from the processor,
It is possible to change the data in the copy destination page. Further, according to this embodiment, the cache memory 3
Page 51 to Page 5
The actual copying work to No. 2 is not necessary, and the efficiency of the processing is improved.

Example 2. In the first embodiment, the case where the main storage is used as the data storage unit that stores the data has been described as an example. However, regardless of whether the data storage unit that stores the data is the main storage, another internal storage device or an external device is used. It may be a storage device.

Example 3. In the above-described first embodiment, the case where the cache memory is used as the temporary storage unit that temporarily stores the data has been described as an example, but the cache memory is an example of the temporary storage unit that temporarily stores the data. Any storage device may be used as long as it temporarily stores data.

Example 4. In the first embodiment, the first
Although the case where the page is set as one area unit as the area or the second area is shown, the page is an example of the area,
It does not matter even when copying between areas using other units.

Example 5. In the above-described first embodiment, the case where the cache memory 3 stores the data when the page designated by the address holding device 5 is accessed has been described. If the accessed data is stored in a temporary storage unit such as a memory, the data may be stored by another method.

Embodiment 6. In the first embodiment, the data of page 51 is stored in the cache memory 3 when the matching is obtained.
Although the case where the page 52 is created by copying the data into the page 52 and then flushing the cache memory 3 is shown, the data in the page 51 is directly copied into the page 52 without passing through the cache memory 3 and then in the cache memory 3. It is also possible to create new data in the page 52 by flushing the data in the page 52 to the page 52.

[0023]

As described above, according to the present invention, during the copy process at the time of writing of the operating system, the changed data is held in the temporary storage section so that the entire page data is immediately copied. It is possible to obtain the same effect as that of copying without using the data, reduce the data transfer at the time of copying, and improve the processing efficiency of the operating system.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of a computer system according to a storage management system of the present invention.

FIG. 2 is a diagram illustrating an address map according to an embodiment of the present invention.

FIG. 3 is a diagram showing a procedure of processing of a storage management system of the present invention.

FIG. 4 is a diagram illustrating a conventional address map.

[Explanation of symbols]

 1 CPU 2 First Address Translation Device 3 Cache Memory 4 Second Address Translation Device 5 Address Holding Means 6 Main Memory 10 Processor

Claims (1)

[Claims] 1. A storage management method having the following elements: (a) a data storage unit for managing and storing data in predetermined area units; (b) a processor for accessing the data in the data storage unit; (c) the data; A temporary storage unit between the storage unit and the processor for temporarily storing data accessed by the processor, and (d) an instruction to copy the data in the first area of the data storage unit to the second area. , The second area is allocated, and when the processor performs a write operation to the address in the second area, the corresponding address in the second area is allocated, and the processor allocates to the address in the second area. (E) processor for allocating an address in the first area corresponding to an address in the second area Access data storage means for storing, in the temporary storage portion, data accessed by the address converted by the address conversion means when accessing an address in the second area of the data storage portion based on a predetermined process; (F) After the predetermined processing of the processor is completed, the second data allocated by the address conversion means using the data stored in the temporary storage unit by the access data storage means and the data in the first area. A matching means that produces data in a region.