JPS59161758A - Control system for disk cache - Google Patents

Abstract

PURPOSE:To prevent wrong data from being transferring by providing a comparing circuit, etc., and checking whether transfer data is a data that a CPU requests or not, then transferring it. CONSTITUTION:When the data that the CPU1 requests is present in a track buffer 13 and to be transferred, a table searching circuit 9 searches a hush table 11. At this time, the number of a magnetic disk, etc., are stored in a register 16 while compressed by a hush algorithm circuit 10. A comparing circuit 17 compares the contents of the register 16 with the compressed number of the magnetic dish device, etc., read out of the buffer 13, and when they are not coincident, a transfer circuit 14 is controlled to stop the transmission of the data. When they coincide with each other, the data is sent out. Consequently, whether the transfer data is the data that the CPU1 requests or not is confirmed at every time, so no wrong data is transferred to the CPU1.

Description

DETAILED DESCRIPTION OF THE INVENTION (a) Technical Field of the Invention The present invention relates to a disk which is provided in a magnetic disk subsystem and has a built-in high-speed buffer for storing data in units of trunks in order to improve the access time of the magnetic disk. Related to the cache device, in particular, a disk cache control method that makes it possible to read the take of the high-speed buffer requested by the central processing unit from the high-speed buffer at the correct address (b) Prior art and problems Magnetic disk cache The subsystem is equipped with a disk cache device between the magnetic disk control device and the magnetic disk device, and the disk cache device has a built-in high-speed buffer called a trunk 8 buffer to transfer data on the magnetic disk requested by the central processing unit to the trunk. The data is read out and stored in advance in the high-speed buffer for each unit (this allows data to be transferred at high speed when the central processing unit needs the data).

FIG. 1 is a block diagram of a circuit explaining operation 0 of a conventional disk cache device. The microprocessor 5 of the disk cache device 4 controls each circuit using microinstructions based on the microprogram stored in the control memory 196. iij in which the central processing unit 1 accesses the magnetic disk of the magnetic disk device 3 via the magnetic disk control device 2;

In order to determine whether the data is target data for the device 4, the interface control port 1e15
The location where the data exists on the magnetic disk is loaded into the register 7 through the process. The location is determined by the device number of the magnetic disk device 3, the cylinder number of the magnetic disk,
It is indicated by the number of magnetic heno 1. The table search circuit 9 searches the extensions registered in the extension 1-table 8 [
. . and the contents of the register 7, if it is determined that the data is data to be handled by the disk cache device 4, the hash table 11 is searched using the hash algorithm of the hash algorithm circuit 10.

Here, extension 1 - table 8 and hash table 1
1 will be explained. The extension 1-table 8 contains the cylinder number of the magnetic disk and the magnetic disk 7 in order to indicate data that has been determined in advance to be targeted by the disk cache device 4 in a certain continuous area on the magnetic disk.
It is registered in a continuous width indicated by the address number 1.

In addition, in order to reduce the storage capacity of the hash table 11, 1 hide data that does not represent the device number, 2 hide data that represents the cylinder number, and 1 hide data that represents the magnetic head number are stored in the hash table 11. It is compressed and registered using the hashish algorithm, and then the trunk buffer 1
The start address of the data stored in No. 3 is also registered in the same way. Therefore, if the contents loaded into the register 7 are within the continuous area of the extension I-table 8, the data becomes the data targeted by the disk cache device 4. Also, the contents of register 7 and the hash table 11
Report the matching contents.

It means that it exists in the above-mentioned Dataka River Rack Bathosopha 13, and if there is no match, it means that it does not exist. As a result of this search, if the data exists in the trunk buffer 13 and the central processing unit 1 does not read the data, in order to transfer the data to the central processing unit 1, the・F1
Row F' of the start address of the data stored in 3.
The data is read out by the transfer circuit 14. The data is sent out via the interface control circuit 15. If the rank buffer 13 does not contain the data desired by the central processing unit 1, the data is read from the magnetic disk and stored. When the central processing unit 1 writes data, it writes the data in the trunk buffer 13 regardless of whether or not the corresponding data exists in the 1-rack bath sofa 13.

When storing data in the trunk buffer 13, one trunk is stored on the magnetic disk. When the trunk buffer 13 overflows, the oldest stored data is replaced with the newest data under the control of the microprogram in the control memory 6. Each time the above operation is repeated, the probability that the data required by the central processing unit 1 exists in the trunk buffer 13 increases, and the take can be transferred from the trunk buffer 13.

When storing data in the trunk buffer 13, the microprogram 5 searches for an unused block in the trunk buffer 13 and stores its leading address in the register 12. If there is no unused block, the start address of the oldest stored data is stored in the register 12. Further, in the hash table 11, the device number, cylinder number, and magnetic head number of the magnetic disk device 3 of the data to be stored are processed by the hash algorithm circuit 10, and the result is registered as an address. 1
Register the start address of 2 cents and the contents of register 7.

By operating as described above, it becomes possible for the central processing unit 1 to transfer desired data at high speed, but the data stored in the trunk buffer 13 is the same as that recorded on the magnetic disk. Information on the various tables mentioned above is not included. Therefore registers 7, 12 . Extent table 8. Table search circuit 9. Hatshu algorithm circuit 10. When the address set in the register 12 becomes inaccurate due to a failure in the hash table 11 or the like, correct data cannot be transferred. However, the conventional disk cache device has a serious drawback in that it transfers an incorrect take in response to a request from the central processing unit 1.

(C) Object of the Invention The object of the present invention is to eliminate the above-mentioned drawbacks when data required by a central processing unit is transferred from an L-rack buffer of a disk cache device.

By checking whether the transferred data is the data required by the central processing unit and then transferring it.

An object of the present invention is to provide a disk canosh control method that does not transfer erroneous data.

(d) Structure of the Invention The structure of the present invention is to write necessary data to the high-speed buffer in a disk cache device that has a built-in high-speed buffer for storing data in trunk units in order to improve the access time of the magnetic disk. /Means for searching a table for reading.

means for recording the search results of the search means; means for comparing the output of the means for storing the search results with read data of the high-speed buffer; and reading of the high-speed buffer based on the comparison result of the comparison means. means for preventing data transmission;
means for adding the search result of the search means to a part of the data and writing it into the high-speed buffer, and when writing data to the high-speed buffer, the search result of the search means is added and written 1. When reading data from the high speed buffer, the search result of the search means is compared with the search result added to the data in the high speed buffer, and data is sent only when they match.

(e) Embodiment of the Invention FIG. 2 is a block diagram of a circuit showing an embodiment of the invention. FIG. 2 shows a register 16 for recording 7 tres of the hash table 11 obtained by the hash algorithm of the hash algorithm circuit 10 in the circuit of FIG.
A line for writing the address of the register 16 into the track buffer 13, a comparison circuit 17 that compares the address of the register 16 with the address read from the trunk buffer 13, and a line for writing the address of the register 16 into the track buffer 13, Time transfer circuit 14
A line for stopping data transfer is added to the above.

As explained in FIG. 1, when data is stored in the track buffer 13, the microprocessor uses the output of the hash algorithm 10 as an address.

The number 5 is stored in the hash table 11 including the number of the magnetic disk device, the cylinder number, the magnetic disk number, and the number! ~Register the start address of the rack buffer 13. At this time register 1G
simultaneously stores the compressed values of the magnetic disk device number, cylinder number, and magnetic head number from the output of the hash algorithm circuit 10, and when data is written to the track buffer 13, the magnetic disk drive number is stored before the data. A compressed value of the disk device number, print printer number, and magnetic heel number (1) is written. Therefore, all the data in the trunk buffer 13 is added to the hash table 11 address (-1). .

When data requested by the central processing unit 1 exists in the track buffer 13 and the data is transferred to the central processing unit 1, the cable search circuit 9 searches the hash table 11 as described above.

At this time, the register 16 contains the number of the magnetic tisf device.

The cylinder number and magnetic P number are compressed and stored by the hash algorithm circuit 10. Comparison circuit 7 compares the contents of the register 16 with the compressed magnetic disk drive number, cylinder number, and magnetic disk number read from the track buffer 13, and if they do not match, controls the transfer circuit 14 to transfer the data. Prevent sending. If they match, the following data will be sent. By operating as described above, it is possible to confirm each time that the data is the same as the data required by the central processing unit 1, thereby preventing 18 incorrect data from being sent to the central processing unit 1.

In this embodiment, the address of the hash table 11 is written in front of the data of the trunk function I3, but it is also possible to use the output of the table search circuit 9 that corresponds to each data of the trunk function function I3. .

(f) Effects of the Invention" 2. As explained above, 1. In the present invention, in response to a take request from the central processing unit, a check is performed every time whether the data transferred from the disk cache device is correct. 2. The benefit is that it prevents the data from being sent to the central processing unit for processing and causing serious harm.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a circuit explaining the operation of a conventional disk storage device, and FIG. 2 is a block diagram of a circuit showing an embodiment of the present invention. (2) is the central processing unit, and (2) is the magnetic tisf control unit. 3 is a magnetic disk device, 4 is a disk cache device,
5 is the microprocessor, 6 is the control record 1. a. 7.12.16 is Resisk, 8 is Exken)-table, 9 is table search circuit, 10 is Panosh algorithm circuit, 11 is Hash table, 13 is Trasokhasofa, 14 is transfer circuit, 15 is interface control circuit, 17 is a comparison circuit.

Claims (1)

[Claims] To improve the access time of a magnetic disk. ] - In a disk cache device having a built-in high-speed buffer for storing takes in units of ranks, a means for searching a table for writing/reading data necessary for the couch into the high-speed buffer, and a search result of the search means are provided. a means of remembering,
means for comparing the output of the means for storing the search result with the read data of the high speed buffer; means for blocking the sending of the sofa read data to the high speed based on the comparison result of the comparison means; and searching of the search means. means for adding the result to a part of the data and writing it into the high-speed buffer, and when writing data to the high-speed buffer, the search result of the search means is added and written, and the data is read from the high-speed buffer. A disk cache control system characterized in that when outputting data, the search result of the search means is compared with the search result added to the data of the sofa to the high speed, and the data is sent only when they match.