JPS6196575A - Erroneous read correction of magnetic disk device - Google Patents

Abstract

PURPOSE:To correct reading without a central processing unit by providing a channel which ORs exclusively data in memory and an error pattern when an erroneous pattern is received from a controller, and re-stores the result in the memory. CONSTITUTION:When an erroneous read arises in the data area of a magnetic disk 50, the erroneous pattern from a magnetic disk controller 40 is transferred to a data buffer 23 of a channel 20. Data read out from a memory 30 specified by a memory address register 22 and the erroneous pattern are calculated by an exclusive OR circuit 27 and re-stored in the memory 30. With respect to zero data from the controller 40, data in the memory 30 will not be re-written, but only the subtraction of a byte counter 24 and the addition of the register 22 are executed. Thus the erroneous read is corrected without the central processing unit and the processing efficiency can be improved.

Description

DETAILED DESCRIPTION OF THE INVENTION (Technical Field) The present invention relates to a read error correction method for a magnetic disk device, and more particularly to an error correction method when a correctable read error occurs in a data area of a magnetic disk device.

(Conventional technology) Conventional magnetic disk device of this type σ) Reading error.

In this correction method, when a correctable read error occurs in the data area of the magnetic disk subsystem, the magnetic disk subsystem performs a central processing
Under the control of CPtJ)io', the contents of the memory 30' are read out via the channel 20' and the magnetic disk controller [
40' for multiple magnetic disks 1 [DKU) 50'
One of the devices is selected, and the data to be written is read out and stored in the memory 3 o I via the reverse route.

In this case, the data 7 format in the magnetic disk drive is as shown in Figure 2, where the area HA is the physical address of the disk and the home address, and the next area RO is used for special purposes such as dealing with defective tracks. Furthermore, the following areas RIN are used for user purposes, and are divided into count i (Q), key part (6), and data part ■.These are error correction codes (EC
C) t-Additional error correction checking is performed by well-known logic. ``If this error that can be corrected occurs, the magnetic disk controller 40'' issues a retry instruction through the channel 20'' for the count section and the key part, and the error pattern is confirmed by the magnetic disk controller 40. The data is stored in an internal buffer and corrected, and the contents are rewritten to the memory 30 through channel 20rt-. However, since the data area usually has a long number of bytes, it is impossible for the magnetic disk controller 40' to perform correction processing. Therefore, if a correctable read error occurs in a long data area, the magnetic disk controller 40' abnormally terminates the instruction. Thereafter, the central processing unit 10' issues a sense command to obtain the error pattern and its position li', and as shown in FIG. up to (in bytes)? The magnetic disk controller ff uses the error display stat shown as the sense byte.
Have it transferred from 40'.

Then, based on the contents of this sense byte, the central processing unit lO' intervenes and corrects the data on the memo IJ 30'. Thereafter, execution is resumed from the next instruction in the instruction chain that was interrupted due to the abnormal termination of the magnetic disk controller 40', and subsequent reading operations are continued.

In this way, correction of read errors in the conventional data area involves interrupting the read operation due to abnormal termination of the magnetic disk controller 40', resulting in the intervention of the central processing unit 10'. This has the disadvantage of reducing processing efficiency.

(Objective of the Invention) An object of the present invention is to eliminate such drawbacks of the conventional read error correction method of a magnetic disk device, and to perform a central processing when correcting read errors 11) in the data area of the magnetic disk device. The objective is to provide an error correction method that does not require processing equipment.

(Structure of the Invention) According to the present invention, in a read error correction method of a magnetic disk device that corrects the contents of a memory by intervention of a central processing unit through a control device and a channel, an error display obtained from a data area of a magnetic disk is used. The control device transfers the error pattern to the channel in the read error part of the data area according to the stat and the error pattern, and transfers zero data to the channel in the other part.

When receiving the error pattern from the control @, performs an exclusive OR operation on the data in the memory indicated by the memory address and the error pattern data, stores the result again in the memory, and A read error correction system characterized in that the channel only updates a byte counter and a memory address when the zero data is received.

(Example) Next, embodiments of the present invention will be described with reference to all the drawings.

FIG. 3 shows the configuration of an embodiment of the present invention. Third
In the figure, it is assumed that the magnetic disk device 50 has a data area 51t- of length t, and a read error of t2 bytes occurs at t1 bytes from the beginning of this data area 51. The serial data bits from the data area 51 are converted into data in byte units through the serial/parallel conversion circuit 43 in the magnetic disk controller 40, and an error correction byte is created by the error detection circuit 44, as shown in FIG. The erroneous byte is stored in the error pattern register 45 as shown.
The number of bytes from the first byte to the last byte of the error pattern is stored in the displacement register 46. When the magnetic disk controller 40 receives a retry command from the channel 20, these registers store the error pattern in the detected part of the error indicated by error displacement, and the zero data for all channels in other parts. Used to transfer to 2.

In channel 20 of this embodiment, subchannel buffer 2
1 stores the instruction code, byte counter value, and start memory address in advance, and its output a is stored in the memory.
The address register 22 and byte counter 24 are set. This byte counter 24 operates in synchronization with the byte counter 41 of the magnetic disk controller 40, and the contents of the data buffer 42 are transferred to the memory write bus via the data buffer 23 of the channel, the AND circuit 28, 28', and the OR circuit 29. fk and is written to the memory 30.

When the magnetic disk controller 40 detects a read error as described above, the status byte of the retry request is sent to the channel 20, so the status byte determination circuit 25 of the channel 20 detects the retry request and sends the status byte using the signal 6. Reset initial values from subchannel buffer 21 to byte counter 24 and memory address register 22? During normal transfer, the bit serial data sent from the magnetic disk device 50 is converted into byte serial data by the serial/parallel conversion circuit 43 of the magnetic disk system # device 40, and the error detection circuit 44t Data buffer 42Il? : is transferred to the data buffer 23 of the channel 20. The channel 20 is sequentially written in units of 4 bytes from the data buffer 23 into the memory indicated by the memory address register 22 via the AND circuit 28' (normal transfer signal e=1) and the memory write bus f. During data transfer after the magnetic disk controller 40 detects a correctable read error and requests a retry to the channel 20, the data creation circuit 47 converts the serial/parallel converter 4
3 to 1 byte, and in synchronization with the signal, the byte position indicated by the error display stat register 46 and the contents of the error pattern register 45 are transferred to the data buffer 23. In the portion where no error has occurred (part A, 0 in FIG. 3), the data creation circuit 47 creates all zero data and transfers it to the data buffer 23 of the channel 20 via the data buffer 42.

The data sent from the magnetic disk controller 40 to the data buffer 23 in this way is detected by the non-zero data detection circuit 26.
checks each byte to see if it is zero data. With zero data, the signals e and d are zero, so the AND circuits 28 and 28' do not function, and the memory 3
A read of 0 is also not performed. - If the non-zero data detection circuit 26 detects non-zero data, the memory address
The data in the memory 30 indicated by the register 22 is once read and input to the exclusive OR circuit 27 through the memory read bus g.

The data is again exclusive-ORed with the non-zero data in the data buffer 23 and rewritten from the AND circuit 28 (signal d=1) into the memory 30 indicated by the memory address register 22 through the memory write bus/vf-. Ru.

As described above, in this embodiment, during retry transfer after a correctable read error occurs, when the data from the magnetic disk control device 40 is zero, the data in the memory 30 is not replaced; The byte counter 24 is simply subtracted and the memory address register 22 is added.

However, only when the data from DKC,i is non-zero, then the memory 3 indicated by the memory address register 22
The data once read from 0° is stored in the data buffer 23.
Exclusive OR is performed with the non-zero data (error pattern) in the data, and writing to the memory 30 indicated by the memory address register 22 is performed again.

Therefore, in this embodiment, erroneous data in the memory 30 can be corrected while performing a retry transfer operation between the channel 20 and the magnetic disk controller.

(Effects of the Invention) As described above, the present invention eliminates the need for the control device to abnormally terminate all instructions when a reading error occurs in the data area, eliminates the need for central processing unit intervention, and improves processing efficiency. There is an effect.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing an embodiment of the present invention, FIG. 2 is a diagram showing the system configuration of a conventional magnetic disk device, and FIG. 3 is a diagram showing the data area of a conventional magnetic disk. ...Central processing bag fit (CPU), 20.
... Channel, 21 ... Sub channel buffer, 22 ... Memory address register, 2
3... Data buffer, 24... Byte counter, 25... Status byte decoder,
26...Non-zero data detection circuit, 27...
・・Exclusive OR circuit, 28.28'・・・・・・AN
D circuit, 29...OR circuit, 30...
Memory, 40...Magnetic disk controller (DK
C), 41... Byte counter %42...
...Data fist buffer, 43...Direct/parallel conversion circuit, 44...Error detection Oi, 45...
...Error pattern register, 46...Displacement register, 47...Data creation circuit, 50...Magnetic disk unit (DKU). PI-1

Claims (1)

[Claims] In the read error correction method of magnetic disk drives, which corrects the contents of memory through the intervention of the central processing unit through a control unit and a channel, the data area is read using the error display stat and error pattern determined from the data area of the magnetic disk. The control device transfers the error pattern as data in the error part and zero data to the channel in the other part, and the data in the memory indicated by the memory address and its data when receiving the error pattern from the control device. The channel performs an exclusive OR operation with the error pattern data, stores the result again in the memory, and simply updates only the byte counter and memory address when the zero data is received. A reading error correction method characterized by: