JPH09185556A - Method for managing duplex memory and duplex memory management device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a duplex memory management method where a shadow memory is effectively used as a memory for holding data different from a primary memory and the performance is improved by reducing the issuing quantity of I/O, in kernel memory management. SOLUTION: As to the duplex memory system, two memory modules are set to be a pair and are used as the memory modules of one group, one which is usually accessed is set to be the primary memory 11 and the other is set to be the shadow memory 12 used when a fault occurs in the primary memory 11 in a computer 10 having the plural memory modules. Data is written into both memories and data is read from either memory. Thus, same data is always held. When the primary memory 11 is filled with single data, the content of the primary memory 11 is restored through the data structure when the fault occurs in the primary memory 11.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a kernel memory management system in a computer system using a dual memory, that is, a dual memory management method and a dual memory management device.

[0002]

2. Description of the Related Art In a conventional dual memory system, the contents of a primary memory that is normally accessed when managing kernel memory and the contents of a shadow memory that fails when the primary memory fails always match. Was the one used in. However, depending on the state of the memory, the contents may be represented by simple information, and at that time, it is not necessary to back up the contents of the primary memory with the shadow memory. By using this, it is possible to keep a record of the initial state when the primary memory is initialized, and to keep the contents of the shadow memory old. When writing is performed to the initialized primary memory, the contents of the shadow memory may be matched with the primary memory. The advantage is that when the data for any access does not exist in the plinary memory, I / O must be issued to read the data, but when the data exists in the shadow memory, I / O is not issued. Since it is only necessary to copy the data in the shadow memory to, the amount of I / O issued can be reduced.

Further, when a failure occurs in the primary memory, if the primary memory and the shadow memory are in the same state, the contents of the shadow memory can be used and the processing can be continued. If the primary memory and the shadow memory do not match, the contents of the shadow memory should be constructed according to the recorded state of the primary memory and the processing should be continued.

[0004]

However, as described above, in the conventional dual memory management system, the shadow memory is used only as an exact copy of the primary memory.

Therefore, the present invention has been made in consideration of the above circumstances. In order to solve the above-mentioned problems, in addition to the above-mentioned conventional technique, the shadow memory is different from the primary memory in the kernel memory management. An object of the present invention is to provide a duplicated memory management method and a duplicated memory management device, which are effectively used as a memory for holding data and whose performance is improved by reducing the amount of I / O issued.

[0006]

In order to achieve the above object, the present invention provides an operating system operating on a computer system having a plurality of memory modules, in which two memory modules are paired to form a set of memory modules. As a primary memory, the one that is normally accessed is used as the primary memory, and the one that is used when a failure occurs in the primary memory is used as the shadow memory. Writing is performed for both, and reading is performed for either. In a dual memory system that always holds the same data, when the primary memory is filled with a single data, the contents of the primary memory are restored from the data structure when a failure occurs in the primary memory A dual memory management method characterized by the above.

Further, in order to achieve the above object, the present invention provides the above dual memory management method, wherein a data structure for storing whether a content of the primary memory and a content of the shadow memory are in a matched state or a mismatched state, and the primary memory. And a data structure for restoring the contents of the primary memory are created and stored so that the contents of the primary memory can be restored with a single data. If a failure occurs in the primary memory, the contents of the shadow memory are used if the prime memory and the shadow memory match, and if they do not match, the contents of the primary memory are restored to the shadow memory to protect the data. I have tried to do it.

Further, the present invention has the following objects to attain the above objects.
In the above redundant memory management method, when the primary memory and the shadow memory do not match and there is no primary memory, but the data that exists in any shadow memory is accessed, the contents of any shadow memory are It is to copy to the primary memory of.

Further, in order to achieve the above object, the present invention provides
In the above redundant memory management method, when the primary memory and the shadow memory do not match and they do not exist in any primary memory, when accessing the data existing in any shadow memory, the contents of the shadow memory are valid or invalid. And a data structure that indicates which data in the external storage device the contents of the shadow memory are equal to, and if the data to be accessed is known in the external storage device, the data is stored. Searched for the existence of the shadow memory to be retained, and when the corresponding shadow memory exists, the contents of the shadow memory are copied to the primary memory prepared for reading the data from the external storage device. Especially.

Further, in order to achieve the above object, the present invention uses two memory modules as a pair as a set of memory modules in an operating system operating in a computer system having a plurality of memory modules. The one to be accessed is the primary memory and the one to be used when a failure occurs in the primary memory is configured to be the shadow memory. Writing is performed for both, and reading is performed from either one. A dual memory system for holding data, comprising means for restoring the contents of the primary memory from a data structure if a failure occurs in the primary memory when the primary memory is filled with a single data. The dual memory management device is characterized in that

Further, in order to achieve the above object, the present invention provides:
In the above dual memory management device, a data structure for storing whether the contents of the primary memory and the contents of the shadow memory match or do not match, and a data structure for restoring the contents of the primary memory are provided. A means for creating and storing data capable of restoring the contents of the primary memory with one data; a means for matching the primary memory with the primary memory when the primary memory and the shadow memory do not match; and a case where a failure occurs in the primary memory In addition, if the prime memory and the shadow memory are in a matched state, the contents of the shadow memory are used, and if they are not matched, means for restoring the contents of the primary memory from the restored data to the jadow memory and protecting the data are provided Especially.

Further, in order to achieve the above object, the present invention provides
In the above redundant memory management device, when the primary memory and the shadow memory do not match and there is no primary memory, but the data existing in any shadow memory is accessed, the content of any shadow memory is changed There is a means for copying to the primary memory of.

Further, in order to achieve the above object, the present invention provides
In the above redundant memory management device, when the primary memory and the shadow memory do not match and there is no primary memory, but the data that exists in any shadow memory is accessed, the contents of the shadow memory are valid or invalid. And a data structure that indicates which data in the external storage device the contents of the shadow memory are equal to, and when the data to be accessed is known in the external storage device, the data is stored. A way to find out if there is shadow memory to hold,
When the corresponding shadow memory exists, a means for copying the contents of the shadow memory to the primary memory prepared for reading the data from the external storage device is provided.

According to the above structure, the shadow memory is used not only as a copy of the primary memory but also as a memory for caching the past contents of the primary memory, and if the shadow memory is hit, I / O is issued. Instead, since the memory-to-memory copy from the shadow memory to the primary memory may be performed, the I / O issuance amount can be reduced and the performance improvement can be expected.

[0015]

BEST MODE FOR CARRYING OUT THE INVENTION The outline of the present invention is as follows (1) to (4). (1) In an operating system that operates on a computer system having a plurality of memory modules, a primary memory that uses two memory modules as a pair as a set of memory modules and normally accesses one memory module,
When the other memory is used as a shadow memory when a failure occurs in the primary memory, the data structure that manages whether the contents of the primary memory and the contents of the shadow memory match, and the contents of the primary memory are managed. And a data structure for managing the contents of the shadow memory, and a means for making the contents of the shadow memory corresponding to an arbitrary primary memory coincide with each other.

(2) In the above (1), when initializing the primary memory when invalidating the memory used for arbitrary data access, the contents and shadow of the primary memory Means for storing information indicating that the contents of the memories are inconsistent, means for storing information indicating the contents of the primary memory after initialization, and means for storing information indicating the contents of the shadow memory before initialization And a dual memory management system.

(3) In the above (2), in the case of a data access in which the contents of the primary memory and the shadow memory do not match, the writing to the primary memory is detected, and the contents of the primary memory and the corresponding shadow memory are detected. Is provided, and means for storing information indicating that the contents of the primary memory and the shadow memory are in a matched state is provided, which is a dual memory management system.

(4) In (3) above, when a new memory is used for arbitrary data access, it is searched whether or not there is a shadow memory holding the data to be accessed, and the corresponding shadow is searched. This is a dual memory management system characterized by including means for copying the contents of a page to an appropriate primary memory if there is a page.

According to the above (1) to (4), not only is the shadow memory used as a copy of the primary memory,
It is also used as a memory that caches the past contents of the primary memory. If the shadow memory is hit, I / O can be copied from the shadow memory to the primary memory instead of issuing I / O.
It is possible to reduce the amount of issuance and to expect performance improvement.

An embodiment of the present invention will be described below with reference to the drawings. As an example of the above (1), FIG. 1 shows a block diagram and an example of a data structure used therein. Next, as an example of the above (2), FIG. 2 shows a block diagram and an example of a data structure used therein. Similarly, as an example of the above (3), FIG.
Is shown. FIG. 4 shows an example of the above (4), and FIG. 5 shows an example of the algorithm used there.

First, in the configuration diagram of FIG.
A plurality of memory modules are provided inside, one is a primary memory 11 and the other is a shadow memory 12. Normally, the primary memory 11 and the shadow memory 12 are both written to, and read from either one, so that the same contents are stored.

When the primary memory 11 is filled with a single data, that data is written only in the primary memory 11. After that, when writing is generated in the primary memory 11, the contents of the shadow memory 12 are matched with those of the primary memory 11.

If a failure occurs in the prime memory 11, if it matches the shadow memory 12, the contents of the shadow memory 12 can be used as they are to continue the processing. If they do not match, the contents of the primary memory from the restored data 13 can be returned to the shadow memory 12 and the processing can be continued.

Next, referring to the block diagram of FIG.
The primary memory 21 and the shadow memory 22 also have a memory module therein, and a data structure 23 for restoring the contents of the primary memory 21 and a prime memory 21.
And has a data structure 24 indicating whether the shadow memory 22 is in a matched state or a mismatched state.

Normally, since the primary memory 21 and the shadow memory 22 are in the same state, the data structure 2 showing the state
The matching state is recorded in 4. Primary memory 21
Is filled with a single data, the data is stored in the data structure 23 for restoring the contents of the primary memory 21, and is written only in the primary memory 21, and the inconsistent state is recorded in the data structure 24 indicating the state. To do.

After that, when a write occurs in the prime memory 21, the contents of the shadow memory 22 are matched with the primary memory 21, and the matched state is recorded in the data structure 24.

If a failure occurs in the prime memory 21, the data structure 24 indicating the state is referred to. If the coincidence state is recorded, the contents of the shadow memory 22 can be used as they are to continue the processing.

If the mismatched state is recorded, the contents of the primary memory are restored from the restored data 23 to the shadow memory 2
It can be returned to 2 and the processing can be continued. Subsequently, FIG.
In the configuration diagram of 1., the computer 30 has a memory module of a primary memory (1) 31, a shadow memory (1) 32, a primary memory (2) 33 and a shadow memory (2) 34, and the primary memory (1) 31 and a data structure (2) 38 indicating whether the shadow memory (1) 32 is in a matched state or a mismatched state.

Normally, the primary memory (1) 31 and the shadow memory (1) 32 are in a matched state, so the matched state is recorded in the data structure (1) 35 indicating the state. The data is written only in the primary memory (1) 32, and the mismatched state is recorded in the data structure (1) 36 indicating the state.

After that, when writing occurs in the primary memory (1) 31, the contents of the shadow memory (1) 32 are matched with the primary memory (1) 31, and the matching state is recorded in the data structure (1) 36.

Here, the former primary memory (1) 31
If an access occurs to the data held by, normally, a new primary memory (2) 33 and shadow memory (2) 34 are prepared, and I / O for reading the data required for access is issued. However, according to this embodiment, the contents of the shadow memory (1) 32 need only be copied to the primary memory (2) 33 and the shadow memory (2) 34. This allows I / O
Can reduce the issuance amount.

Finally, in the configuration diagram of FIG.
0 has a memory module of a primary memory (1) 41, a shadow memory (1) 42, a primary memory (2) 43, and a shadow memory (2) 44, for restoring the contents of the primary memory (1) 41. The data structure (1) 45, the primary memory (1) 41, and the shadow memory (1) 42 have a data structure (2) 46 indicating whether they are in a matched state or a mismatched state.

There is also a data structure (1) 49 indicating whether the contents of the shadow memory (1) 42 are valid or invalid and a similar data structure 4a for the shadow memory (2) 44. An external storage device 4b is connected to the computer 40 and stores programs and data.

Normally, since the primary memory (1) 41 and the shadow memory (1) 42 are in a matched state, the matched state is recorded in the data structure (1) 45 indicating the state. When filling the primary memory (1) 41 with a single data,
If the data before writing is valid, it is written to the external storage device 4b, and the fact that the shadow memory (1) 42 is valid is recorded in the data structure 49.

The data structure (1) indicating the state by storing the data written to the primary memory (1) 41 in the data structure (1) 46 for restoring the contents and writing the data only in the primary memory (1) 41. Record the inconsistency state at 46.

After that, the data is matched with the primary memory (1) 41, and the matching state is recorded in the data structure (1) 46.
Here, if an access occurs to the data previously held in the primary memory (1) 41, a new primary memory (2) 43 and a shadow memory (2) 44 are usually prepared and required for the access. I / O for reading data from the external storage device 4b must be issued, but according to the present embodiment, if an effective shadow memory (1) 42 is found by the algorithm as shown in FIG. Need only be copied to primary memory (2) 43 and shadow memory (2) 44. This makes it possible to reduce the amount of I / O issued.

[0037]

As described above in detail, according to the present invention, I / O issuance is achieved by using the shadow memory not only as a copy of the primary memory but also as a memory for caching the past contents of the primary memory. The amount can be reduced. That is, in addition to the conventional technique, in the kernel memory management, the shadow memory is effectively used as a memory for holding data different from the primary memory, and I / O
It is possible to provide a duplicated memory management method for improving the performance by reducing the issue amount of.

[Brief description of the drawings]

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

FIG. 2 is a block diagram showing a configuration in a state different from that of FIG. 1 according to the same embodiment.

FIG. 3 is a block diagram showing a configuration in a state different from that of FIG. 2 according to the embodiment.

FIG. 4 is a block diagram showing a configuration in a state different from that of FIG. 3 according to the embodiment.

FIG. 5 is a diagram showing an example of an algorithm according to the same embodiment.

[Explanation of symbols]

11, 21 ... Primer memory, 12, 22 ... Shadow memory, 13, 23 ... Restoration data, 24 ... Status data, 3
1, 41 ... Primer memory (1), 32, 42 ... Shadow memory (1), 33, 43 ... Primer memory (2),
34,44 ... Shadow memory (2), 35, 45 ... Restoration data (1), 36, 46 ... Status data (1), 37,
47 ... Restoration data (2), 38, 48 ... Status data (2), 49 ... Valid data (1), 4a ... Valid data (2), 4b ... External storage device.

Claims (8)

[Claims] 1. In an operating system operating on a computer system having a plurality of memory modules, two memory modules are used as a pair as one set of memory modules, and one normally accessed is used as a primary memory. One memory used when a failure occurs in the primary memory is used as a shadow memory, writing is performed for both, and reading is performed from either one so that the same data is always retained. When the primary memory is filled with the above data, when the failure occurs in the primary memory, the contents of the primary memory are restored from the data structure. 2. A data structure for storing whether the contents of the primary memory and the contents of the shadow memory match or do not match, and a data structure for restoring the contents of the primary memory are provided. The data that can restore the contents of the primary memory is created and stored, and the primary memory and the shadow memory are matched with the primary memory in a mismatched state, and when a failure occurs in the primary memory, the primary memory and the shadow memory 2. The redundant memory management method according to claim 1, wherein the contents of the shadow memory are used in the case of coincidence, and the contents of the primary memory are restored to the shadow memory from the restored data to protect the data in the case of disagreement. 3. The primary memory and the shadow memory are in a mismatched state, and when accessing data existing in any shadow memory that does not exist in any primary memory, the contents of any shadow memory are transferred to any primary memory. The duplicated memory management method according to claim 2, wherein the duplicated memory is copied. 4. Data indicating whether the contents of the shadow memory are valid or invalid when accessing data existing in any shadow memory that does not exist in any primary memory when the primary memory and the shadow memory do not match. A shadow memory that holds a structure and a data structure indicating which data in the external storage device is equal to the data in the external storage device, and holds the data when the data to be accessed is found in the external storage device Is searched, and if the corresponding shadow memory exists, the contents of the shadow memory are copied to the primary memory prepared for reading data from the external storage device. Item 4. The redundant memory management method according to Item 3. 5. In an operating system operating on a computer system having a plurality of memory modules, two memory modules are used as a pair as a set of memory modules, and one normally accessed is used as a primary memory, A redundant memory system in which one memory used when a failure occurs in the primary memory is configured as shadow memory, writing is performed for both, and reading is performed from either side to always retain the same data. A dual memory, characterized in that, when the primary memory is filled with a single data, a means is provided for restoring the contents of the primary memory from the data structure when a failure occurs in the primary memory. Management device. 6. A data structure for storing whether the contents of the primary memory and the contents of the shadow memory match or do not match, and a data structure for restoring the contents of the primary memory. Means for creating and storing data capable of restoring the contents of the primary memory; means for matching the primary memory with the primary memory when the shadow memory does not match; and a means for matching the primary memory when a failure occurs in the primary memory. The contents of the shadow memory are used if my memory and the shadow memory match, and if they do not match, means for restoring the contents of the primary memory from the restored data to the jadow memory and protecting the data are provided. 6. The duplicated memory management device according to claim 5. 7. When accessing data existing in any shadow memory that does not exist in any of the primary memories when the primary memory and the shadow memory do not match each other, the contents of the arbitrary shadow memory are transferred to the arbitrary primary memory. 7. The duplicated memory management device according to claim 6, further comprising means for copying. 8. Data indicating whether the contents of the shadow memory are valid or invalid when accessing data existing in any shadow memory that does not exist in any primary memory when the primary memory and the shadow memory do not match. A shadow memory that has a structure and a data structure that indicates which data in the external storage device is equal to the contents of the shadow memory, and holds the data when the data to be accessed is found in the external storage device. And a means for copying the contents of the shadow memory to the primary memory prepared for reading the data from the external storage device when the corresponding shadow memory exists. 8. The duplicated memory management device according to claim 7.