JPH02220155A - Magnetic disk controller - Google Patents

Abstract

PURPOSE:To shorten the processing time required for the hit check, etc., with use of a disk cache memory of a comparatively small capacity by selecting a disk cache memory having a proper block size out of plural disk cache memories based on a data reading request and the contents of a cache tag memory. CONSTITUTION:The disk cache memories 45X-45Z store the data having higher using frequencies out of those data of a magnetic disk device. The cache tag memories 43X-43Z control the states of memories 45X-45Z. A disk controller 42A selects one of memories 45X-45Z that has a proper block size based on a data reading request and the contents of memories 43X-43Z. As a result, the processing time required for the hit check, etc., is shortened with use of memories 45X-45Z having comparatively small capacities.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a magnetic disk control device that controls a magnetic disk device.

In particular, it relates to a disk cache memory of a magnetic disk control device.

[Conventional technology] The configuration of a conventional example will be explained with reference to FIG.

FIG. 2 is a block diagram showing a conventional magnetic disk control device.

In FIG. 2, the system includes a central processing unit (1),
The system bus (2) connected to this central processing unit (1)
), a main storage device (3) connected to this system bus (2), a magnetic disk control device (4) connected to the system bus (2), and this magnetic disk control device (4)
It consists of a magnetic disk device (5) connected to.

In addition, the conventional magnetic disk control device (4) has a system bus interface (2) connected to the system bus (2).
41), a disk controller (42) connected to this system bus interface (41), a cache tag memory (43) connected to this disk controller (42), a magnetic disk device (5) and a system bus interface. A disk interface (44) connected to the NiceF 41) and a disk controller (42)
) and a disk cache memory (45) whose input side is connected to the disk interface (44) and whose output side is connected to the system bus interface (41).

Note that the magnetic disk device (5) is also connected to a disk controller (42) of the magnetic disk control device (4).

Next, the operation of the above-mentioned conventional example will be explained with reference to FIG.

FIG. 3 is an explanatory diagram showing the data structure of a cache tag memory (43) of a conventional magnetic disk control device.

In FIG. 3, the cache tag memory (43) is composed of n entries having addresses O to (n-1), and each entry contains a valid bit V, physical sector address PSA, reference counter RF, etc. There is.

First, the central processing unit (1) starts the magnetic disk drive (5).
) is given to the magnetic disk controller (4) via the system bus (2). That is, the disk controller (42) receives the input/output command as the input/output command A from the central processing unit (1) via the system bus interface (41).

The disk controller (42) determines that the contents of the input/output command A are the physical sector address PS of the magnetic disk device (5),
When reading data from A(p), the physical sector address P S A (p) is given to the cache tag memory (43) as disk sector address information B.

The cache tag memory (43) checks whether information regarding the physical sector address P S A (p) exists in its own data. That is, entry 0~
Check whether the contents of the physical sector address PSA of (n-1) match the disk sector address information B.0 If they match and the valid bit V=1, make the hit signal C significant and The matching entry address is returned to the disk controller (42) as entry information. In the example shown in Figure 3, the entry address is “1”.
” is.

The disk controller (42) determines that the data (contents) at the physical sector address P S A (p) for which data reading was requested has been read into the disk cache memory (45) from the magnetic disk device (5) in the past. ,
I know it still exists. Therefore, the address of the disk cache memory (45) is calculated based on the entry information, and the data at the physical sector address PSA (p) read into the disk cache memory (45) is transferred to the system bus interface (41). The data is transferred to the main storage device (3) via the computer.

In this way, the conventional magnetic disk control device (4) directly determines the physical sector address P S of the magnetic disk device (5).
Since there is no need to access A (p), input/output processing time can be significantly shortened.

Here, the disk cache memory (45) will be explained.

The disk cache memory (45) is entirely divided into n sections, all of which have the same size. The size of one of these partitions is called the (block size), and is generally set to an integral multiple of the size of one physical sector of the magnetic disk device (5). When a certain part of the disk device (5) is accessed, there is a very high probability that the vicinity will be accessed subsequently, and the cache tag memory (43)
This is due to the need to reduce the number of entries.

For example, the block size of the disk cache memory (45) is 8, which is the sector size of the magnetic disk device (5).
The case of double the number will be explained.

There is an input/output command A to read data for any sector whose physical sector address PSA of the magnetic disk device (5) is 81+1 to 8+7, and at that point, that address is still in the disk cache memory (45). If the data does not exist, the magnetic disk control device (4) reads 8 sectors worth of data from the physical sector address PSA of 8111 of the magnetic disk device (5) and stores it in the disk cache memory (45). It turns out. The data (content) read into one block of the disk cache memory (45) has a physical sector address PSA of 8ffl of the magnetic disk device (5).
, 5ILl+1.8m+2, . . . , 8m+7.

Cache tag memory (4
The content of 3) is that V=1 and PSA=8.

Therefore, after that, even if there is an input/output command A to read data to any sector whose physical sector address PSA is from 8 to 8 + 7, the magnetic disk control device (4) will not receive a hit signal C. Since it is significant, there is no need to directly access the magnetic disk device (5).

[Problems to be Solved by the Invention] In the conventional magnetic disk control device as described above, when a data read request is made to the vicinity of a sector where data has been read once, the processing is extremely quickly due to the cache effect. On the other hand, since multiple sectors in the vicinity are imported into the disk cache memory at once,
In the case where information scattered all over the magnetic disk is collected one sector at a time, there is a problem in that the processing time becomes longer and the usage efficiency of the disk cache memory decreases.

This invention was made to solve the above-mentioned problems, and it is possible to shorten processing time such as human inspection time by using a disk cache memory with a relatively small capacity, and to improve its usage efficiency. The purpose is to obtain a magnetic disk control device that can perform the following steps.

[Means for Solving the Problems] A magnetic disk control device according to the present invention includes the following means.

(i) A plurality of disk cache memories with different statically predetermined block sizes.

<ii) A cache tag memory corresponding to the plurality of disk cache memories.

(ii) A disk controller that selects one having an appropriate block size from among the plurality of disk cache memories based on a data read request and the contents of the cache tag memory.

[Operation] In the present invention, the disk cache memory stores frequently used data among the data of the magnetic disk device.

Further, the state of the disk cache memory is managed by the cache tag memory.

Further, the disk controller selects one having an appropriate block size from among the plurality of disk cache memories based on the data read request and the contents of the cache tag memory.

Examples] The configuration of the embodiment will be explained with reference to FIG.

FIG. 1 is a block diagram showing one embodiment of the present invention, and the system bus interface (41) and disk interface (44) are exactly the same as those of the conventional device.

In FIG. 1, one embodiment of the present invention has exactly the same components as those of the conventional device described above, and a disk controller (41) connected to a system bus interface (41).
28) and the cache tag memory (43X), (43Y) and <432) connected to this disk controller (42Δ),! :,, disk controller (42
^) and a disk cache memory (45X) whose input side is connected to the disk interface (44) and whose output side is connected to the system bus interface (41),
<45Y) and (452).

The magnetic disk device (5) is also connected to the disk controller (42^) of the magnetic disk control device (4^), and has a sector size of IK bytes.

The capacity of disk cache memory (45X) is 256
Similarly, the capacity of disk cache memory (45Y) and (45Z) is 256 bytes each, and the block size is 4 Kbytes and 8 bytes. have 64 and 32 byte blocks.

Cache tag memory (43x), (43Y) and (4
3Z) has the same data structure as the conventional cache tag memory (43), and each size is the same as the number of corresponding disk cache memories (45x), (45Y) and (452) blocks.

Next, the operation of the above embodiment will be explained.

First, the central processing bag! (1) is a magnetic disk device (5
) is given to the magnetic disk controller (4^) via the system bus (2). That is, the disk controller (42^) receives the input/output command as the input/output command A from the central processing unit (1) via the system bus interface (41).

The disk controller (42A) determines that the content of the input/output command A is the physical sector address PS of the magnetic disk device (5).
When reading data from A, its physical sector address P
SA is given as disk sector address information B to cache tag memories (43X), (43Y) and (43Z).

Each cache tag memory (43X), (43Y) and (
43Z) each checks whether information regarding the physical sector address PSA is included in its own data. That is, each entry (0 to 12), 0 to 63.0
0 to check whether the contents of the physical sector address PSA of ~31) match the disk sector address information B
If there is a match and a valid bit = 1, make the hit signal Cx, Cy or Cz significant;
And the matching entry address is added to the entry information Dx, D
It is returned to the disk controller (42^) as y or Dz.

The disk controller (42^) has each cache tag name 1: ! J (43X), (43Y) and (43Z
). Hit signal Cx, Cy or Cz
If either of these is significant, the data (contents) at the physical sector address PSA for which data reading was requested was previously transferred from the magnetic disk device (5) to the disk cache memory (45X), (45Y) or (452). ) and that it still exists. Therefore,
Based on the corresponding entry information Dx, Dy or Dz, the disk cache memory (45X), (45Y> or (
452) and save the disk cache memo! , J (45X), (45Y) or C (45Z) are transferred to the main storage device (3) via the system bus interface (41).

On the other hand, if none of the hit signals Cx, Cy, or Cz is significant, the disk controller (42^) reads the physical sector address PSA of the data read request.
The data (contents) is transferred from the magnetic disk device (5) to the disk cache memory (45X), (45Y) or (4
52) and found out that it did not exist (mishit).

In the above case, the disk controller (42^) is
Check the data count in input/output command A.

Find out how many sectors worth of data read is requested, compare the size of the data read request with the block size of each disk cache memory (45X), (45Y), and (45Z) from smallest to largest, and determine the requested size. Disk cache memory with the smallest block size that can contain (45X), (45Y) or (45
Select 2). Then, the data at the physical sector address PSA for which data reading was requested is transferred from the magnetic disk device (5) to the selected disk cache memory (4).
5X), (45Y) or C (45Z) and transferred to the main memory (3).

In other words, if the data read request is a miss, and the data count is less than or equal to 2 bytes, the disk controller (42^) stores the data in the disk cache memory (42^).
45X), and if the data count is greater than 2 bytes and less than 4 bytes, select disk cache memory (45Y), and if the data count is greater than 4 bytes, select disk cache memory (452).
Select.

Here, a comparison will be made between the conventional example and the above-mentioned embodiment.

If the block size of the conventional disk cache memory (45) is eight times the sector size of the magnetic disk device (5), then the sector size of the magnetic disk device (5) is IK.
Since it is a part-time job, the part-time job yells at 8. If the capacity of the disk cache memory (45) is, for example, 768 bytes, the number of entries in a block of 8 bytes in the conventional disk cache memory (45) is 96.

Each disk cache memory (45X), (4
The total capacity of 5Y) and (452) is 7, which is the same as the conventional example.
68 bytes (=256 bytes x 3).

The number of entries for a block of 2 bytes in the disk cache memory (45X) is 128, the number of entries for a block of 4 bytes in the disk cache memory (45Y) is 64, and
Since the number of entries in the 8-byte block in 2) is 32, the total number of entries is 224.

Therefore, in the embodiment described above, the hit rate for input/output commands with a small data count is more than doubled.

In addition, if the block size of the conventional disk cache memory (45) is set to 2 bytes, for example, the number of entries increases to 384, and the hit rate improves even in the conventional example, but the hit check time to check whether there is a hit or not increases. It has the disadvantage that it becomes long.

One embodiment of the present invention includes three disk cache memories (45X) with different block sizes, as described above.
Equipped with (45Y) and (452), tail note, disk cache memory (45x), (45Y) and (452)
) By making an appropriate selection based on system trends, that is, by examining the input/output command data count and making an appropriate selection, it is possible to shorten the human inspection time and increase the hit rate. be effective.

Although three disk cache memories have been described in the above embodiment, it is not necessary to limit the number to three, and even better effects can be expected by determining the type and size of the block size based on the system tendency. .

[Effects of the Invention] As explained above, the present invention includes a plurality of disk cache memories having different statically predetermined block sizes, a cache tag memory corresponding to the plurality of disk cache memories, Since it is equipped with a disk controller that selects one with an appropriate block size from among the plurality of disk cache memories based on a data read request and the contents of the cache tag memory, the disk cache memory has a relatively small capacity. This has the effect that processing time such as human testing time can be shortened and the efficiency of its use can be improved.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a block diagram showing a conventional magnetic disk control device, and FIG.
The figure is an explanatory diagram showing the data structure of a cache tag memory of a conventional magnetic disk control device. In the figure, (4A)...-Magnetic disk control device, (41)
... System bus interface, (42^)?
・・Disk controller, <43X), (43Y)
, (432) ... Cache tag memory, (44) ... Disk interface, <45
X), (45Y), (45Z)...Disk cache memory. In each figure, the same reference numerals indicate the same or equivalent parts.

Claims (1)

[Claims] A plurality of disk cache memories having different statically predetermined block sizes, a cache tag memory corresponding to the plurality of disk cache memories, and the above based on the data read request and the contents of the cache tag memory. A magnetic disk control device comprising a disk controller that selects one with an appropriate block size from among a plurality of disk cache memories.