JPH01255918A - Control system for disk system with mixed sector length - Google Patents

Abstract

PURPOSE:To make the correcting manday of software and an application program unnecessary by absorbing the different in a sector length with a disk cache memory. CONSTITUTION:When a disk cache control device 3 recognizes a magnetic disk device a magnetic disk having a different sector length based on the instruction of an instructing means 13 and writes a data block for a magnetic disk device having a large sector length, since the block is rewritten to the magnetic disk device after only a part needed by the said data block is rewritten on a disk cache memory 4, the erasing of the data due to the partial writing is prevented. Even in a virtual memory system, since the data block is transferred to the buffer area of a main memory 2 from the disk cache memory 4, it is not necessary to change the software and correct, etc., an application program prepared by a user regardless of the input output action of a magnetic disk device side.

Description

[Detailed Description of the Invention] [Summary] Modification of software and application programs regarding a mixed sector length disk system control method that does not require software changes when disk devices with different sector lengths are connected to a computer system. In order to eliminate the need for man-hours, it is equipped with a disk cache memory, executes data transfer between the main storage device and the disk device via the disk cache memory, and also allows disk devices with different sector lengths to coexist. In a connected computer system, a disk cache control device that controls the disk cache memory is made to recognize the sector length of the disk device, and processes data transferred from the main storage device based on the minimum sector length. If a data block is written to a disk device other than the disk device having the minimum sector length, the data block is written to the disk cache memory for each minimum sector length. After the data block is completed by rewriting processing, the data block is written to the corresponding disk device.

[Industrial application field]

The present invention relates to a mixed sector length disk system control method that eliminates the need to change software when disk devices with different sector lengths are connected to a computer system.

As a core external storage device for various computer systems,
Magnetic disk devices are used, and these magnetic disk devices are broadly classified into fixed length magnetic disk devices and variable length magnetic disk devices. This internal fixed-length magnetic disk device is widely used in computer systems because of its ease of data handling.

This fixed-length magnetic disk device records data in areas called sectors, which are divided tracks of a magnetic disk at equal intervals. The size of this sector area, that is, the sector length, is 256 bytes, 512 bytes,
There are 1024 bytes, etc., and in various computer systems, one or more magnetic disk devices of a single type of sector length are generally connected to form a system based on the factors at the time. That is, for example, if it is a 512-byte sector magnetic disk device, only the 512-byte sector magnetic disk device is connected.

By the way, as the storage capacity of magnetic disk devices increases, the sector length also increases, and magnetic disk devices with small sector lengths are no longer being manufactured. However, when adding magnetic disk devices to an existing computer system, It is necessary to ensure that no inconvenience occurs.

[Conventional technology]

As mentioned above, conventionally, a computer system is constructed by connecting magnetic disk devices with the same sector length.

However, until a few years ago, 256-byte sector magnetic disk drives were the only supply market in the market, but recently, with the increase in storage capacity, 512-byte sector magnetic disk drives have shifted to the center of the market. are doing. And 256
Byte sector magnetic disk drives are no longer manufactured.

Therefore, when adding a magnetic disk device to a computer system built using a 256-byte sector magnetic disk device, it is recommended to replace it with a 512-byte sector magnetic disk device, or replace it with a 256-byte sector magnetic disk device and a 512-byte sector magnetic disk device. This means that a magnetic disk device will be used in the mix.

However, when a system is constructed using a magnetic disk device with 256-byte sectors, the data is created based on the sector length, so this 256-byte data is
When simply copying to a byte sector magnetic disk device, logically different data is written within one sector.

That is, a magnetic disk control device that controls a magnetic disk device usually has a machine number specifying the magnetic disk device, a cylinder number of the magnetic disk device, a magnetic head number,
A data address consisting of a sector number and a data length are given, and data read/write is instructed.

Therefore, data of a specified data length is written or read from the specified sector of the track where the specified magnetic head is positioned in the cylinder specified by the address on the magnetic disk.

FIG. 4 is a diagram illustrating copying of data based on sector length.

As shown in FIG. 4(a), in a magnetic disk drive with a 256-byte sector, data ■ and data ■ based on a sector length of 256 bytes are separated by physical sector addresses indicated by 〉 marks, and data ■ For example, if it is 200 bytes, the remaining 56 bytes of area (2) have been cleared. Due to a hardware characteristic called partial write in the magnetic disk control device that controls the magnetic disk device, when reading data, old data is left behind and cannot be read, so data with a number of bytes less than the sector length is This is because when the data is written, the remaining area is cleared based on the data length instructed by the host device to the magnetic disk control device.

If these data ■ and ■ are simply copied to a 512-byte sector magnetic disk device, they will be written in one sector as shown in Figure 4 (bl), and if only data ■ is rewritten, the magnetic disk will be transferred from the host device to the magnetic disk drive. Since the data length of the data (2) instructed to the control device remains 200 bytes, the data (2) is cleared by the partial write, as shown in FIG. 4 (C1).

FIG. 5 is a diagram illustrating copying of blocks.

In a magnetic disk device with a 6256-byte sector, which is n times the sector length, a block, which is a collection of records, is divided into 512-byte sectors, as shown in Figure 5(a), assuming that one block is three sectors. When multiple blocks are simply copied to a magnetic disk device, as shown in Figure 5 (bl), the first data block is written in sector ■ and 1/2 area of sector ■, but the remaining 1 data block of sector The data of the next block is written in the /2 area.

In this case as well, if you rewrite the previously written block,
Since the data length to be written to sector ■ as instructed by the magnetic disk controller remains 256 bytes, the next block written adjacent to the remaining 1/2 area of sector ■ by partial write. data will be cleared.

FIG. 6 is a diagram illustrating the layout of the buffer area of the virtual storage system.

In a computer system using a virtual storage method, data is transferred page by page between a buffer area provided in a main storage device and a magnetic disk device. 256
In a computer system that uses a byte sector magnetic disk device, this buffer area is also allocated at around 256 by ha, and for example, as shown in buffer area ■ and buffer area ■ in Figure 6(a), the buffer area is The area may be physically cut off.

In this case, if the buffer area ■ has a size of 256 bytes, 1/2 of the data [phase] of the data read from the 512-byte sector magnetic disk device will be transferred to the buffer area ■, but the remaining data will be transferred to the buffer area ■. Data ■ is a different command when the channel command is data [phase] and when it is data ■, so the input/output operation on the magnetic disk device side may shift to the beginning of the next sector ■. may not be read.

[Problem to be solved by the invention]

As mentioned above, conventionally, when magnetic disk drives with different sector lengths were mixed or replaced with a magnetic disk drive with a larger sector length, data could not be simply copied to the magnetic disk drive with a larger sector length. Therefore, data from a magnetic disk device with a small sector length,
When writing to a magnetic disk device with a large sector length, the software is changed so that it is cleared by partial write and does not disappear as described above. In order to prevent the data from being written, it is necessary to perform processing such as assigning an address to each sector and copying the data.

However, there is a problem in that this requires a large amount of man-hours to change the software.

In addition, particularly in computer systems using virtual memory, there is a problem in that it is necessary to modify a huge number of application programs created by users depending on the scale of the computer system.

In view of these problems, the present invention aims to make a disk cache memory absorb the difference in sector length, thereby eliminating the need for man-hours for modifying software and application programs.

[Means to solve the problem]

FIG. 1 is a block diagram of the principle of the present invention.

The central processing unit 1 sends an input/output request to the disk cache control device 3 without being aware of the sector length of the magnetic disk device 7 or 8. When the central processing unit 1 requests to write a data block to the magnetic disk device 7 or 8, the disk cache control device 3 writes the data block to the magnetic disk device 7 or 8 based on the system initialization information based on the instruction from the instruction means 13. Check the sector length of 8.

For example, if the magnetic disk device 7 has a small sector length and the magnetic disk device 8 has a large sector length, the disk cache control device 3 recognizes this difference in sector length,
If the magnetic disk device designated by the central processing unit 1 is 8, it is checked whether a data block corresponding to the data block has been read from the magnetic disk device 8 in the disk cache memory 4.

That is, if there is a data block on the disk cache memory 4 starting from the address specified by the central processing unit 1 and having a specified data length or more, this disk cache memory 4
Starting from the specified address above, rewriting is performed using the data read from the main storage device 2 for each minimum sector length.

When the specified data length has been rewritten, the remaining data in the data block remains as it was, and channel 5
The data is transferred to the designated magnetic disk device 8 via the magnetic disk control device 6, and written into the area where the data block before rewriting was recorded.

If there is no data block on the disk cache memory 4 starting from the address specified by the central processing unit 1 and having the specified data length or more, the specified magnetic disk device 8
A data block of the specified data length is read from the specified address, expanded on the disk cache memory 4, and the same processing as described above is performed on this data block, and the original data block of the magnetic disk device 8 is restored again. Write to the recorded area.

When the central processing unit 1 specifies the magnetic disk device 7, the disk cache control device 3 performs conventional processing on the data block on the disk cache memory 4 to rewrite the data block, and then The magnetic disk device 7 is caused to write the rewritten data block.

[Effect]

By configuring as above, disk cache system? The Il device 3 recognizes magnetic disk devices with different sector lengths based on instructions from the instruction means 13, and when writing a data block to a magnetic disk device with a large sector length, writes the data block on the disk cache memory 4. After rewriting only the necessary part, it is rewritten to the magnetic disk device, which prevents erasure of data due to partial write, and even in the virtual storage method, data is transferred from the disk cache memory 4 to the buffer area of the main storage device 2. Since the blocks are transferred, there is no need to change the software or modify the application program created by the user, regardless of the input/output operations on the magnetic disk device side.

〔Example〕

FIG. 2 is a block diagram of a circuit showing an embodiment of the present invention.
FIG. 3 is a flowchart illustrating the operation of FIG. 2.

The central processing unit 1 provides the disk cache control unit 3 with the machine number, cylinder number, and magnetic An input/output request instructing a data block read/write is sent by specifying a head number, an address indicating a sector number in which the magnetic head starts reading/writing data, and a data length.

The disk cache control device 3 includes a magnetic disk device 7
A correspondence table 11 that associates the data blocks in the disks of ~9 with the data blocks in the disk cache memory 4, and a cache that has information such as the frequency of use and validity/invalidity of the data blocks in the disk cache memory 4. As shown in FIG. 3, the correspondence table 11 is created based on the input/output requests sent by the central processing unit 1.
Explore.

This search of the correspondence table 11 uses a hash algorithm or the like based on the machine number, address, and data length sent by the central processing unit 1 in order to speed up the processing. Correspondence table 11
The data block that the central processing unit 1 requested for input/output is
If it indicates that it exists in the disk cache memory 4, that is, if it hits, the usage frequency in the management table 12 is corrected to increase the priority, and then it is checked whether the request from the central processing unit 1 is a read or a write.

If the request from the central processing unit 1 is a read, the disk cache control device 3 reads the data block from the disk cache memory 4 and transfers it to the main storage device 2. or,
When the request from the central processing unit 1 is a write, the disk cache control unit 3 recognizes the sector length of the magnetic disk devices 7 to 9 based on the machine number specified by the central processing unit 1 based on the instruction from the instruction program 10.

For example, if the magnetic disk device 7 is a 256-byte sector magnetic disk device, the magnetic disk device 8 is a 512-byte sector magnetic disk device, and the magnetic disk device 9 is a 1024-byte sector magnetic disk device, the system Since the initial setting information indicates the sector length by machine number, if the machine number indicates the magnetic disk device 7, the data block read from the main storage device 2 is written to the cache memory 4 as in the conventional case.

If the machine number indicates magnetic disk device 8 or 9, the data block read from main storage device 2 is processed in units of 256 bytes and written to disk cache memory 4.

For example, if the data blocks read from main memory 2 are 2
00 bytes, 2 in disk cache memory 4
Write 00 bytes of data and clear the remaining 56 bytes. Further, if the data block is 500 bytes, 500 bytes of data are written to the disk cache memory 4, and the remaining 12 bytes are cleared.

Thus, for example, if the hit data block is 1024
If it is a byte, 512 bytes are rewritten,
The remaining 512 bytes remain as they were.

The disk cache control device 3 sends the thus rewritten data block on the disk cache memory 4 to the magnetic disk control device 6 via the channel 5 based on the elements of the correspondence table 11, and identifies the hit data block. The hit data block of the magnetic disk device is rewritten in the magnetic disk area where the hit data block was written.

Correspondence table 1 indicates that the data block requested for input/output by the central processing unit 1 does not exist in the disk cache memory 4.
1, that is, when there is a miss, the disk cache control device 3 writes the data block in the disk of the magnetic disk devices 7-9 at the time when the specified data block was written to the disk cache memory 4. ,
Elements that correspond to data blocks in the disk cache memory 4 are added to empty items in the correspondence table 11.

If there is no free item in the correspondence table 11, the management table 12 is referred to, and the item of the element of the data block designated as invalid is deleted to make it a free item. Then, elements indicating the addition of a new data block, the validity of this data block, the frequency of use, etc. are added to the management table 12. At this time, old data blocks that are used less frequently are determined to have a lower priority and are invalidated.

Then, based on the specified machine number, address, and data length, the data block is read from the magnetic disk unit having the specified machine number of the magnetic disk units 7 to 9 via the channel 5 and the magnetic disk control unit 6, and the data block is The routine moves to a routine that writes data to the cache memory 4 and checks whether the request from the central processing unit 1 is a read or write request.

In this embodiment, the disk cache control device is provided between the central processing unit and the channel, but there is a method in which a disk cache control device and a disk cache memory are provided in the central processing unit, and a method in which a disk cache control device and a disk cache memory are provided in the magnetic disk control device, respectively. Of course, the same applies when a disk cache memory is provided.

〔Effect of the invention〕

As described above, the present invention can eliminate the need to modify software or user-created application programs in a computer system in which disk devices with different sector lengths are connected together.

[Brief explanation of the drawing]

FIG. 1 is a principle block diagram of the present invention, FIG. 2 is a block diagram of a circuit showing an embodiment of the present invention, FIG. 3 is a flowchart explaining the operation of FIG. 2, and FIG.
FIG. 5 is a diagram for explaining data copying based on sector length, FIG. 5 is a diagram for explaining block copying, and FIG. 6 is a diagram for explaining the layout of a buffer area in a virtual storage system. In the figure, 1 is a central processing unit, 2 is a main storage device, 3 is a disk cache control device, 4 is a disk cache memory, 5 is a channel, 6 is a magnetic disk control device, 7
+8+9 is a magnetic disk device, 10 is an instruction program, 11 is a correspondence table, 12 is a management table,
13 is an instruction means. ,Tiri sun and moon mogadai p1fu"mouth・baro乎 1 lower P-first υth one time")n (col 4's furotsuno m sheep ? (2) 茅2 figure's φ Kakushita and Toraguchi 1 yen slot flow + yard matter 3 diagram (2) (α') Cb)', (C) mouthpiece that can be said to be a copy of Chi 2 who goes to Cef 9shi 4 It can be said that it is a modification of Figure 1/So 2, and it can be said that it is Sun Moon and Satoru. 7F and Otsuguchi

Claims (1)

[Claims] A disk cache memory (4) is provided, and a main storage device (4) is provided.
Data transfer between 2) and the disk devices (7) and (8) is performed via the disk cache memory (4), and disk devices (7) and (8) with different sector lengths are connected together. The computer system includes a disk cache control device (3) that controls the disk cache memory (4), and a disk device (7) (
instruction means (8) for recognizing the sector length and instructing to process the data transferred from the main storage device (2) and write it into the disk cache memory (4) based on the minimum sector length; 13) is provided, and when a data block is written to a disk device other than the disk device with the minimum sector length, the data block is completed by performing rewriting processing for each minimum sector length on the disk cache memory (4). After that, the data block is written to the corresponding disk device.