JP2546075B2 - Disk controller retry method - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk control device for transferring data between a disk device and a host device, and more particularly to a plurality of records recorded in the same track of the disk device by one command. The present invention relates to a retry method of a disk control device which prevents a long processing time when a correctable error is detected in a read record in continuous reading.

A disk control device for controlling data transfer of a disk device used as an external storage device of a computer system has a record recorded on the same track by one command such as a dump command from the central processing unit. In some cases, an instruction is issued to instruct continuous reading.

When an error occurs in the record read from the disk device during execution of this dump command, the operating system (O) of the central processing unit (O
Although the retry process is performed under the control of S), since the retry process by the OS takes time, it is desired to realize the efficient retry process without the intervention of the OS.

[0004]

2. Description of the Related Art FIG. 7 is a block diagram illustrating an example of a conventional technique. The processor 8 of the disk control device 2 reads the program stored in the control memory 7 and operates to control the interface circuits 4 and 6 and the data transfer control circuit 5 so that the processor 8 between the host device 1 and the disk device 3 can operate. Cause data transfer.

That is, when the host device 1 designates the address of the disk device 3, that is, the machine number, the cylinder number, the head number and the record number of the disk device 3, and instructs the writing of data, the processor 8 When the head of the disk device 3 is positioned at the designated cylinder via the interface circuit 6 and the designated head approaches the designated record, data is sent from the host device 1 via the interface circuit 4, and the data transfer control circuit 5 and the interface circuit 6 to transfer to the disk device 3,
Causes data to be written from the specified record.

When data reading is designated, when the designated head is positioned at the designated record of the designated cylinder of the disk device 3, the data is read and the interface circuit 6 and the data transfer control are performed. It is transferred to the higher-level device 1 via the circuit 5 and the interface circuit 4.

While the data transfer is being executed, the ECC circuit 9 provided in the data transfer control circuit 5 detects an error in each record of the transferred data, and corrects a correctable one.

FIG. 8 is a diagram for explaining the recording format of the disk device. In each track of the disk device 3,
A home address is recorded following the index indicating the beginning of the track, and then a plurality of records are recorded.

Each record is a cylinder address,
A count part for recording a head address, a record address, a data length, etc., a key part for recording logical identification information of a data part that follows, and a data part for recording data. The count part, the key part, the data part. An ECC code for error detection and correction is added to the end of the.

The ECC circuit 9 reads this ECC code to detect and correct errors in the count section and the key section.
As for the data section, the buffer memories provided in the interface circuits 4 and 6 have a small capacity and cannot store all the data in the data section, so that error correction cannot be performed.

When a dump command is issued to the disk controller 2 from the host device 1, the disk controller 2 records on the same track from the record of the designated address of the disk device 3. The specified records are continuously read for the designated number of records and transferred to the host device 1.

Then, the ECC circuit 9 detects an error in the count part, the key part, and the data part for each record transferred, and corrects a correctable error in the count part and the key part. An error is only detected and notified to the processor 8, and no error correction is performed.

When the processor 8 is notified of the error occurrence, the processor 8 notifies the upper device 1 of the error occurrence, and since the upper device 1 issues the dump command, the OS control is performed instead of the normal command retry. The command chain is recombined by and the retry is instructed. That is, the reading is instructed from the record next to the record in which the error has occurred.

This is because in normal command retry, the head position is shifted by the offset of the servo circuit from the data read start address of the disk device 3 to read the data again, but in the case of the dump command, This is because an error occurs in a plurality of records, and even if the data is read from the beginning, the head position is displaced, so that an error may occur in another record and the data may not be repaired.

[0015]

As described above, since the command chain is recombined under the control of the OS in the conventional retry of the error in the dump command,
Due to the heavy load on the host device 1 and the rearrangement of the command chains, the disk device 3 has to redo the head positioning operation, which causes a problem that the processing takes time.

In view of such a problem, the present invention provides a buffer memory for storing data of one track or more so that the ECC circuit 9 can correct the error of the data, and does not perform the retry by the OS but normally. It is intended to shorten the processing time by transferring the data stored in the buffer memory to the host device 1 again by the command retry.

[0017]

FIG. 1 is a block diagram for explaining the principle of the present invention. A disk control device 10 has an error detection / correction function 12 for detecting and correcting an error in data read from a disk device 3. Equipped with this disk device 3
And data transfer between the host device 1 and the host device 1.

A buffer memory 11 for storing data read from the disk device 3 for one track or more is provided, and a plurality of records recorded in the same track of the disk device 3 can be consecutively written by one command. In the case of reading the data, the read records are sequentially transferred to the host device 1 and are sequentially stored in the buffer memory 11, and when a correctable error is detected in the read records, the host device 1 is requested to retransmit the data. , A record is read from the buffer memory 11 and retransmitted from the transferred first record to the higher-level device 1, the error-corrected record is transferred on the buffer memory 11, and then a record in which an error is detected by the disk device 3 is transferred. Reading is started from the record next to, transferred to the higher-level device 1, and stored in the buffer memory 11.

Further, when the disk control device 10 continuously reads a plurality of records recorded in the same track of the disk device 3 with one command, the read records are sequentially transferred to the host device 1. Along with buffer memory 1
1 is sequentially stored, and when a correctable error is detected, each time the error is detected, the host device 1 is requested to retransmit the data, a record is read from the buffer memory 11, and the transferred first record is read. After retransmitting to the host device 1, the error-corrected record is transferred on the buffer memory 11,
Then, reading is started from the record next to the record in which an error is detected by the disk device 3, the data is transferred to the host device 1 and stored in the buffer memory 11.

[0020]

With the above configuration, the record read by the start of execution of the dump command is stored and left in the buffer memory 11, and the record in which an error has occurred is corrected in the buffer memory 11. , A normal command retry is requested to the host device 1, the data in the buffer memory 11 is transferred again, and the data is read from the record next to the record in which the error occurred and the host device 1 Since it is sufficient to transfer to the device 1, the dump command processing time can be shortened.

[0021]

2 is a block diagram of a circuit showing an embodiment of the present invention, and FIGS. 3 to 6 are diagrams for explaining the operation of FIG.

The same reference numerals as those in FIG. 7 indicate those having the same functions.
The processor 8 of the disk controller 10 uses the control memory 14
The program stored in is read and operated, and as shown in FIG.
As shown in (A), when a dump command is received from the host device 1 via the interface circuit 4, the disk device 3 is controlled via the interface circuit 6 and the address of the disk device 3 designated by the dump command, that is, Records are continuously read from record m.

Then, through the data transfer control circuit 13 and the interface circuit 4, as shown in the data of FIG. 3A, the read record is transferred to the higher-level device 1, and the record read to the buffer memory 11 is transferred. Figure 3
As shown in (A), records m to m + 1 are sequentially stored.

When the processor 8 is notified by the ECC circuit 9 that an error has been detected as shown in FIG. 3B when the record n is read out, as shown in the unit check of FIG. 3B. Host device 1 via interface circuit 4
Report unit check to and initiate internal retry.

That is, it is checked whether or not the generated error can be corrected, and if it can be corrected, the ECC circuit 9 is made to correct it.
The corrected record n is stored in the buffer memory 11. The higher-level device 1 that has received the unit check is shown in FIG.
Since the command retry is instructed as shown in the retry of (C), the processor 8 changes the record stored in the buffer memory 11 as shown in the data of FIG.
The records m to m + 1 are sequentially read and transferred to the higher-level device 1.

When the corrected record n is transferred from the buffer memory 11 as shown in the data of FIG. 3 (D), the record n + 1 is read from the disk device 3 as shown in the data of FIG. 4 (A). Then, it is transferred to the higher-level device 1 and stored in the buffer memory 11.

If the error in the record n is uncorrectable, the processor 8 reads the record m from the buffer memory 11 and transfers it when the retry is instructed by the host device 1, as shown in FIG. As shown in FIG. 4 (B), the record n is read again from the disk device 3, and if an error does not occur or if it can be corrected, the next record n + 1 is recorded in the disk device as shown in FIG. 4 (A). Read from 3 and transfer.

Next, the case where an error occurs from a plurality of records will be described. As shown in FIG. 5A, record m + 1
If an error is detected in step 1, the unit check is reported to the higher-level device 1, and internal retry is started.

If the error can be corrected, the corrected record m + 1 is stored in the buffer memory 11 and transferred from the record m in the buffer memory 11 to the host device 1,
As shown in (B), when the corrected record m + 1 is transferred, the next record m + 2 is read from the disk device 3, transferred to the host device 1, and stored in the buffer memory 11.

As shown in FIG. 5 (C), when an error occurs in the record n, a unit check is reported to the host device 1,
Start internal retry. If the error can be corrected, the corrected record n is stored in the buffer memory 11, and
As shown in (D), the record m in the buffer memory 11 is again recorded.
From the buffer memory 11 to the upper device 1, the corrected record m + 1 is transferred, and the next record m + 2 to record n-1 are read from the buffer memory 11 and transferred to the upper device 1. Then, as shown in FIG. 6 (B), the corrected record n is read from the buffer memory 11 and the upper device 1
6C, the record n + 1 is read from the disk device 3 and transferred to the host device 1 and stored in the buffer memory 11 as shown in FIG. 6C.

[0031]

As described above, according to the present invention, an error is detected in a read record during execution of an instruction such as a dump command for continuously reading a plurality of records in the same track from a disk device. In this case, the command chain is not recombined by the OS and the data is retransmitted by retrying the command, which reduces the load on the host device and eliminates the need to redo the head positioning operation of the disk device, thereby correcting the error. The subsequent record reading can be processed only by waiting for the disk to rotate, so that the processing time of the dump command can be shortened.

[Brief description of drawings]

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

FIG. 2 is a block diagram of a circuit showing an embodiment of the present invention.

FIG. 3 is a diagram for explaining the operation of FIG. 2 (No. 1)

FIG. 4 is a diagram for explaining the operation of FIG. 2 (No. 2)

FIG. 5 is a diagram (part 3) for explaining the operation of FIG. 2;

FIG. 6 is a diagram (part 4) for explaining the operation of FIG. 2;

FIG. 7 is a block diagram illustrating an example of a conventional technique.

FIG. 8 is a diagram illustrating a recording format of a disk device.

[Explanation of symbols]

 1 host device 2, 10 disk control device 3 disk device 4, 6 interface circuit 5, 13 data transfer control circuit 7, 14 control memory 8 processor 9 ECC circuit 11 buffer memory 12 error detection and correction function

Claims (2)

(57) [Claims] 1. A function (12) for detecting and correcting an error in data read from a disk device (3), and performing data transfer between the disk device (3) and a host device (1). The disk controller (10) is provided with a buffer memory (11) for storing the data read from the disk device (3) for one track or more, and in the same track of the disk device (3) by one command. When reading a plurality of recorded records in succession, the read records are sequentially transferred to the host device (1) and are sequentially stored in the buffer memory (11), and correctable errors can be added to the read records. When this is detected, the upper device (1) is requested to retransmit the data, the record is read from the buffer memory (11), the transferred first record is retransmitted to the upper device (1), and then the buffer is retransmitted. Transfer the corrected record on the memory (11). Then, the reading is started from the record next to the record in which the error is detected by the disk device (3), the data is transferred to the host device (1), and the data is stored in the buffer memory (11). A method of retrying a disk control device characterized. 2. In the disk control device (10), when a plurality of records recorded in the same track of the disk device (3) are continuously read by one command, the read records are read as described above. The data is sequentially transferred to the host device (1) and sequentially stored in the buffer memory (11), and when a correctable error is detected, data is retransmitted to the host device (1) each time the error is detected. The buffer memory (1
After reading the record from 1) and retransmitting it from the transferred first record to the higher-level device (1), the error-corrected record on the buffer memory (11) is transferred, and then the disk device (3). A retry method for a disk control device, characterized in that reading is started from a record next to a record in which an error is detected, the read is started, the data is transferred to the host device (1) and stored in the buffer memory (11).