JP2001265653A - Data processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten the access time of a direct access request to an external storage device. SOLUTION: To the direct access request for which cache control in an OS 52 is not performed in a main body 50, the cache control is performed by using the storage control part cache memory 64 of a disk controller 54. Thus, while keeping the merit of the direct access request of evading the inconsistency of the cache of the OS 52 and a magnetic disk device 56 due to the fault of the main body 50 or the like, the access time is shortened. For the write request of data for which the cache control is performed in the OS 52, the cache control is not performed in the disk controller 54. Thus, the capacity of the storage control part cache memory 64 is suppressed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data processing apparatus having an external storage unit, for example, a hard disk drive, and more particularly to a delayed write to the external storage unit.

[0002]

2. Description of the Related Art A magnetic disk drive has an advantage that the cost per unit capacity is lower than that of a memory.
There is a disadvantage that the access time is longer than that of the memory due to the mechanical operation. For this reason, computer systems use hierarchical storage in which the main memory is formed of an expensive but high-speed semiconductor memory, while the magnetic disk device is used as an external storage device.
In hierarchical storage, a difference in access speed between a memory and a magnetic disk device is absorbed by using a disk cache mechanism.

The conventionally used disk cache is provided by an operating system (OS) on a main body including a central processing unit. The disk cache mechanism stores a copy of the data at the accessed address in the disk cache on the main memory when the magnetic disk device is accessed. Then, when a magnetic disk access request to the same address is made, the read / write to the disk cache on the memory is performed without accessing the actual magnetic disk. However, if there is a write request to the magnetic disk, the actual data on the magnetic disk will be inconsistent with the data in the disk cache. I do.

Here, if the main body stops unexpectedly,
There is a possibility that the contents of the disk cache are not reflected on the magnetic disk, causing inconsistency between the two. So, O
The type of disk I / O request that does not pass through the disk cache is provided in the application interface of S, and important data can be directly written to the magnetic disk device to protect it.

FIG. 5 is a block diagram for explaining cache control in a conventional computer. FIG. 6 shows a disk I / O in an OS of a conventional computer.
It is a flowchart of O processing, and is a figure for demonstrating cache control. When an application program operating on the OS 4 in the main body 2 requests access to the magnetic disk device 6, the type of the request is determined by the request determination logic 8 (S30). Request judgment logic 8
Determines whether the request type is a normal I / O request made via a disk cache or a direct access request that issues an I / O request directly to the disk controller 10 without using it. For example, if the access request issued by the application program APP1 is a normal I / O request, the disk cache memory on the memory 14 via the disk cache logic 12 provided by the OS based on the determination S30 in the request determination logic 8. Reading and writing of data are performed between these steps (S32), and upon completion of this, APP1 is notified of the completion of READ / WRITE (S34), and the processing of the READ / WRITE request from APP1 ends (S36). The disk cache logic 12 reflects the contents of the disk cache in the memory 14 to the magnetic disk device 6 at a predetermined timing.

On the other hand, for example, when the access request issued by the application program APP2 is a direct access request, the OS issues an I / O request directly to the disk controller 10 based on the determination S30 in the request determination logic 8, and Data is directly read and written between the APP 2 and the magnetic disk device 6 (S38). Upon the completion, the completion of READ / WRITE is notified to APP2 (S34), and the READ / WRITE from APP2 is notified.
The TE request process ends (S36).

[0007]

Conventionally, from the viewpoint of protecting important data as described above, there has been a case where the main unit directly accesses the magnetic disk device without using the disk cache on the memory. In this case, the actual access time to the magnetic disk device can be seen from the application of the main unit, and there is a problem that the processing is not performed quickly.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and has been made to avoid an inconsistency in the storage contents between a main unit and a magnetic disk device due to the influence of a failure of the main unit, and to prevent the access from occurring. An object of the present invention is to provide a data processing device in which a reduction in time is prevented.

[0009]

According to the data processing apparatus of the present invention, the storage control unit determines whether the data requested to be written to the storage control unit cache memory and the external storage unit is from the cache of the main processing unit. Determination means for determining whether the data is not from the cache by the determination means, and write control means for temporarily storing the data determined to be not from the cache in the storage control unit cache memory and performing a delayed write to the external storage unit. Have

In the data processing device according to the present invention,
The determination unit makes the determination based on the fact that the data requested to be written from the cache is in block units.

[0011] In the data processing device according to the present invention,
The determination means may perform the determination further in consideration of a data size of the data.

[0012]

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a schematic block diagram of a computer system for explaining cache control according to an embodiment of the present invention. The main body 5 of the computer system
0 (main processing unit) is basically the same as that conventionally used. That is, the OS 52 performs basically the same processing as the flow chart of FIG. 6, and the magnetic disk device 5 via the disk controller 54 (storage control unit).
6 (external storage unit). Specifically, when an application program operating on the OS 52 requests access to the magnetic disk device 56, the type of the request is determined by the request determination logic 58.
Is determined (S30). Request determination logic 58
Determines whether the request type is a normal I / O request made via a disk cache or a direct access request that issues an I / O request directly to the disk controller 54 without using it. For example, when the access request issued by the application program APP1 is a normal I / O request, based on the determination S30 in the request determination logic 58, the memory 62 is transmitted via the OS cache logic 60, which is a disk cache logic provided by the OS. Reading and writing of data is performed with the above disk cache memory (S32), and upon completion of this, REA is transmitted to APP1.
The D / WRITE completion is notified (S34), and the READ / WRITE request processing from APP1 ends (S3).
6). The contents of the disk cache in the memory 62
Based on the cache logic 60, the data is delayed and written to the magnetic disk device 56 at a predetermined timing, and the contents are reflected on the magnetic disk device 56.

On the other hand, for example, when the access request issued by the application program APP2 is a direct access request, the OS sends the I / O directly to the disk controller 54 based on the determination S30 in the request determination logic 58.
A request is issued to read and write data (S38). Upon the completion, the completion of READ / WRITE is notified to APP2 (S34), and the READ / W from APP2 is notified.
The RITE request process ends (S36).

As described above, under the control of the OS 52, the main body 50 issues a write request to the magnetic disk device 56. This write request is processed by the disk controller 54. The feature of the computer system resides in the disk controller 54.

The conventional disk controller 10 has an OS
When there was a write request from the disk drive 4, the data was sequentially written to the magnetic disk device 6. On the other hand, in the present system, the disk controller 54 includes a storage control unit cache memory 64. The control logic unit 66 of the disk controller 54 includes a feature determination logic 68 and a controller cache logic 70 for performing control using the storage control unit cache memory 64. Here, the feature determination logic 68 is a determination unit that determines whether the data requested to be written to the magnetic disk device 56 is from the cache of the OS 52, and the controller cache logic 70 is that the data is not from the cache of the OS 52. This is write control means for temporarily storing the determined data in the storage control unit cache memory 64 and performing a delayed write to the magnetic disk device 56, and these operations will be apparent from the following description.

FIG. 2 is a flowchart for explaining the write control to the magnetic disk device 56 by the disk controller 54. The control logic unit 66 starts processing upon receiving an access request to the magnetic disk device 56 from the OS 52 (S80). In the control logic unit 66, first, the characteristic determination logic 68 determines whether the request is a write request for data cached in the OS 52 based on the characteristics of the parameters of the access request (S82).

For example, an application program APP
When the data cached in the OS 52 is reflected on the magnetic disk device 56 in response to one normal I / O request, based on the determination S82 in the feature determination logic 68,
The control logic unit 66 immediately issues an I / O request and accesses the magnetic disk device 56 (S84). Upon the completion, the completion of READ / WRITE is notified to the OS 52 (S86), and the READ / WRITE from the OS 52 is notified.
The E request process ends (S88).

On the other hand, the application program APP
OS 52 responds to the direct access request of OS 2
If the request is made to write data directly to the magnetic disk device 56 without going through the cache of the storage controller cache memory 64 via the controller cache logic 70 based on the determination S82 of the feature determination logic 68. Is read and written (S90), and upon completion, the completion of READ / WRITE is notified to the OS 52 (S86), and the read / write from the OS 52 is performed.
The TE request process ends (S88). The contents of the storage control unit cache memory 64 are written to the magnetic disk device 56 at a predetermined timing with a delay based on the controller cache logic 70, and the contents are reflected on the magnetic disk device 56.

The disk controller 54 as described above, which is performed in addition to the conventional disk cache in the OS 52
With the cache, the data directly requested by the application program is reflected on the magnetic disk device 56 even if a failure occurs in the main body 50, and the access time to the magnetic disk device 56 can be shortened. On the other hand, for data cached in the OS 52, the access speed is secured by the cache in the OS 52, so the cache in the disk controller 54 is not particularly necessary. Therefore, by not caching the data cached in the OS 52 by the disk controller 54, the required amount of the storage control unit cache memory 64 can be suppressed.

Next, the feature determination logic 68 will be described. As described above, the feature determination logic 68
A determination unit that determines whether the request is a write request for data cached in the OS 52 based on the characteristics of the parameters of the access request from the server (S8).
2). The parameters of the access request include the start address of the area to be accessed in the magnetic disk device 56 and the data length. The access request related to reading and writing between the cache of the OS 52 and the magnetic disk device 56 has the following features.

(I) Since the request is made in block units,
The start address of the parameter of the access request is a multiple of the block.

(Ii) Since the request is made in block units,
The data length of the parameter of the access request is a multiple of the block.

(Iii) Since the request is for a large amount of data according to the cache capacity of the OS 52, the data length of the parameter of the access request is relatively large.

FIG. 3 is a schematic diagram showing a storage area of the magnetic disk device 56 and an example of an access request. (I) above
The storage areas 100 to 104 of the magnetic disk device 56 having all of the features (i) to (iii) are examples of an address range accessed by an access request related to a cache of the OS 52. For example, the block length is a relatively large size such as 2048 bytes. On the other hand, the above (i) to (i)
The storage areas 110 to 114 of the magnetic disk device 56 lacking at least one of the characteristics ii) are examples of an address range accessed by an access request that does not pass through the cache of the OS 52.

FIG. 4 is a flowchart of the process of the feature determination logic 68. The feature determination logic 68 first determines whether the OS 52
It is determined whether or not the parameter of the access request from has the above characteristic (iii) (S120). If the data length is less than the predetermined threshold, it is determined that the data does not pass through the cache of the OS 52 (S122). ). On the other hand, if the data length is equal to or greater than the predetermined threshold, it is next determined whether or not the parameter has the above characteristic (i) (S124). If the access start address is not a multiple of the block length, It is determined that the data has not passed through the cache of the OS 52 (S122). On the other hand, if the start address is a multiple of the block length, then it is determined whether the parameter has the above characteristic (ii) (S126). If the data length is not a multiple of the block length, It is determined that the data does not pass through the cache of the OS 52 (S122). On the other hand, when the data length is a multiple of the block length, the above feature (i)
Since all of (iii) are satisfied, the feature determination logic 68 determines that the access request is a request to write data in the cache of the OS 52 (S12).
8). In the example shown in FIG. 3, the threshold of the data length of the condition (iii) is two blocks.

As described above, in the present embodiment, whether the data write request from the main unit 50 is related to data via the cache of the OS 52 or not is determined based on the characteristics of the parameter of the access request.
The determination can be made on the fourth side. In this embodiment,
The operation of the OS 52 of the main body 50 may be basically the same as the conventional one.

One feature of the present invention is that a direct access request to the magnetic disk device 56, that is, data not cached by the OS 52 is transferred to the magnetic disk device 56.
A means for caching the disk controller 54 in response to a request to write to The configuration having this feature can be realized by a main body OS different from the conventional one. For example, the OS gives information that can identify an access request to the magnetic disk device 56 and a direct access request regarding data from the cache of the OS to each of the access requests, and the disk controller 54
Based on the identification information, the storage control unit cache memory 6
4 can be configured to control whether or not temporary storage is performed.

[0029]

According to the data processing apparatus of the present invention, the storage control unit for controlling the external storage unit such as the magnetic disk device is provided with the storage control unit cache memory. When the data requested to be written to the external storage unit is not from the cache of the main body OS, the storage control unit cache memory temporarily stores the data and performs delayed writing to the external storage unit. As a result, regarding the data directly requested by the application program, the contents of the data are reflected in the external storage unit even if a failure or the like occurs in the main body, and the access time to the external storage unit can be reduced. That is, according to the present invention, there is an effect that both protection of data for which a direct access request to the external storage unit is made and reduction of the access time can be achieved.

[Brief description of the drawings]

FIG. 1 is a schematic block diagram of a computer system for explaining cache control according to an embodiment of the present invention.

FIG. 2 is a flowchart for explaining write control to a magnetic disk device in a disk controller.

FIG. 3 is a schematic diagram showing a storage area on a magnetic disk device and an example of an access request;

FIG. 4 is a flowchart of a process of a feature determination logic.

FIG. 5 is a block diagram for explaining cache control in a conventional computer.

FIG. 6 is a flowchart of a disk I / O process in an OS in a conventional computer, and is a diagram for explaining cache control.

[Explanation of symbols]

50 main unit, 52 operating system (O
S), 54 disk controller, 56 magnetic disk device, 62 memory, 64 storage control unit cache memory, 66 control logic unit, 68 feature determination logic, 70 controller cache logic.

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G06F 3/06 302 G06F 3/06 302A

Claims (3)

[Claims] 1. A data processing device comprising: an external storage unit controlled by a storage control unit; and a main processing unit controlled by an operating system and provided with a cache that performs a delayed write to the external storage unit. A storage control unit cache memory; a determination unit that determines whether data requested to be written to the external storage unit is from the cache of the main processing unit; and the determination unit determines that the data is not from the cache. And a write control unit for temporarily storing the data determined to be in the storage control unit cache memory and performing a delayed write to the external storage unit. 2. The data processing apparatus according to claim 1, wherein the determination unit makes the determination based on the fact that the data requested to be written from the cache is in block units. 3. The data processing apparatus according to claim 2, wherein the determination unit further performs the determination in consideration of a data size of the data.