JPH07182250A - Automatic alternation processing method for magnetic disks - Google Patents

Abstract

PURPOSE:To improve reliability and to evade the delay of a command end report to a master host by preventing data in a magnetic disk device from being erased. CONSTITUTION:Command unreceived time for managing an idle state in which a reading/writing request is not received from the master host 1, is prepared, and in the idle state, diagnostic reading for a magnetic disk device 4 is executed. At the time of detecting a defective sector, a stand-by sector in the device 4 is used and substituting processing is executed. A management table 3 manages the frequency of defective writing and the number of defective sectors, and when each value exceeds its threshold, the host 1 is informed of the dangerousness of the device 4. Thereby, a recovery disabled error is generated at the time when the host 1, instructs to perform reading, so that the generation of data erasing can be prevented.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic alternation processing method suitable for preventing data loss in a magnetic disk device used in a computer system or the like which requires high reliability.

[0002]

2. Description of the Related Art Conventionally, in a magnetic disk drive control system, when a recoverable error (a read error occurs, but data is recovered by retrying or error correction by ECC) due to a read request from a host, Since there is a high possibility that the sector will become unreadable, there is a method of using the recovered data to automatically replace the spare sector inside the control system to prevent data loss. For a write request, if writing is disabled due to an ID error,
Alternatively, there is a method of performing automatic replacement when it is determined that the sector is defective in the read check after writing. That is, in the conventional technique, the trigger for the automatic replacement processing for both reading and writing is the access request of the upper host, and is a method of performing the replacement processing and then reporting the end.

Further, as the alternation processing of the magnetic disk,
As described in Japanese Patent Laid-Open No. 63-257962, there is also a method of temporarily saving data of a track to be replaced to a spare track.

[0004]

In the above-mentioned conventional technique, if the read request is not executed from the host,
If the automatic replacement process is not started and there is no access for a long time, a next read start may cause an unrecoverable error, which may lead to data loss. Also in terms of performance, in the conventional technology, after performing the alternation process,
Since the command completion is reported, there is a problem that throughput is reduced.

According to the present invention, prior to a read instruction from the host, the time during which there is no access from the host is utilized to detect a defective sector and save the data to a good sector. Also,
Data can be guaranteed before an unrecoverable error occurs. Also in terms of performance, there is no problem of delaying the command completion report.

[0006]

In order to achieve the above object, a command non-reception time table is provided for command issuance management from a host, and the drive itself is managed in an IDLE state.

Further, by managing the diagnostic read cylinder on the memory and on the disk, the entire magnetic disk can be sequentially diagnosed.

Also, the drive itself is at risk because the number of times that a read error has occurred once, but the number of times when it became normal by rewriting, is actually judged as a bad sector and is changed to manage the number of sectors. It becomes possible to judge.

Further, by providing the write address address and the read error occurrence table, it is possible to realize the automatic alternation processing when the access from the upper host is interrupted.

[0010]

When the command issuance from the host is interrupted, it is determined that the IDLE state has occurred due to the command non-reception time, and the diagnostic read and automatic replacement processing can be executed.
In addition, by managing the diagnostic read cylinder, even if a command is received from the host during the diagnostic read, or even if the power is turned off, the entire magnetic disk
It becomes possible to diagnose.

Further, it is possible to report to the host that the drive itself is in danger, based on the number of times normalization has been made by rewriting and the number of defective sectors generated.

[0012]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of a disk control system according to the present invention.

The magnetic disk drive 4 of FIG. 1 has one or several spare sectors and one cylinder or several cylinders of spare cylinder area for each track. The magnetic disk controller 2 has a function that can be assigned to a cylinder.

In order to measure the command non-reception time, the command non-reception time table 11 shown in FIG. In addition, the CPU 10 counts up the command non-reception time table 11 by a timer interrupt every one second.

The magnetic disk controller 2 judges that the command is not received from the host 1 at the time when the command non-reception time table 11 reaches N seconds (N is different depending on the system), so that it is in the IDLE state. Activate the diagnostic lead for automatic replacement.

In the diagnostic read process, the diagnostic read is started from the physical cylinder shown in the diagnostic read cylinder NO table 13. In addition, the initial value of the diagnostic read cylinder table should start with the user cylinder,
The diagnostic read is performed from the head cylinder of the user cylinder.

If the diagnostic read of the diagnostic lead cylinder is completed normally, the diagnostic read cylinder NO table 13 is counted up and the diagnostic read of the next cylinder is performed.

When a command is received from the host 1 during the diagnostic read, the diagnostic read is immediately stopped and I
At the time of entering the DLE state, the diagnostic read is restarted from the head of the stopped cylinder. Also, diagnostic lead table 1
3 is also saved in the diagnostic cylinder NO8 of the magnetic disk device 4 at regular intervals, and the value of the diagnostic cylinder NO8 is transferred to the diagnostic read table 13 at the time of restarting after the power is cut off, so that the value is continuously stored. The diagnostic read can be performed, and when the diagnostic read is completed up to the last cylinder, the process returns to the leading cylinder again. Then, when a recoverable error occurs during the diagnostic read, the automatic replacement process is started.

In the automatic alternation processing, in order to check whether or not a read error occurs due to a defect (noise, head position shift) at the time of writing, the recovered data is used and the writing processing is performed to the error occurrence sector. . After that, re-reading is performed several times, and if no read error occurs, it is determined as a failure at the time of writing, the number of write failures in the diagnostic read management table 14 is updated, and the replacement process is not executed. Perform diagnostic lead again.
Further, when executing the writing process, in order to prevent the data loss due to the power interruption, the data and the write address are saved in the automatic substitution save cylinder 5 of the magnetic disk device 4, and when the power is turned on, the automatic substitution save is performed. Cylinder 5
If the information is saved, the data recovery is executed. Also, if the number of write failures exceeds the threshold,
A drive write error is reported to the upper host 1.

Further, even in the case of re-reading, if a read error occurs, it is judged that the sector concerned is a defective sector, and if the spare sector of the track concerned or the spare sector is already used, a spare sector or spare cylinder in the vicinity thereof. To the replacement process. At the time of replacement allocation, information such as addresses and the number of replacement sectors by diagnostic read are saved in the replacement management table 16. At this time, if the number of diagnostic read replacement sectors exceeds the threshold value, a drive abnormality is reported to the upper host.

Further, because of the nature of the magnetic disk, it is highly possible that a flaw near a sector in which a defective sector has occurred has a flaw in the neighborhood. Therefore, the diagnostic read of the neighboring track is executed several times. .

In the above process, by detecting a defective sector and performing the replacement process, the reliability of the entire disk device is improved, and an unrecoverable error occurs when a read request is received from the upper host, resulting in data loss. It can be prevented from becoming. In addition, it is possible to prevent the throughput from being reported to the upper host to be terminated after the alternation process is executed in response to an access request from the upper host.

Further, in the above-mentioned example, the automatic replacement process in the diagnostic read is performed by utilizing the time when the host 1 is not accessed. However, even if the read request from the host 1 is used as a trigger, It is also possible to prevent performance degradation due to delay in reporting the command end to 1. That is, when a recoverable error occurs in response to a read request from the host 1, the automatic replacement management table 12
Then, the generated address is registered in advance, and when the access from the upper host 1 is interrupted, the automatic substitution processing is executed.
Regarding the write request from the upper host, the address where the writing is executed is managed in advance, and the diagnostic read of the area where the writing is executed is preferentially executed when the access from the upper host 1 is interrupted. Due to
It is possible to improve reliability and performance.

In the above description, all the automatic replacement processing was performed inside the magnetic disk control device 2, but these methods can be realized by the host system 1. That is, this is a method of performing diagnostic read processing to the magnetic disk control device 2 by utilizing the time during which an access request to the magnetic disk control device 2 is not made by the multiple processing management in the host system 1. By managing all the management tables 3 inside the magnetic disk control device 2 in the above description on the host 1 side, automatic replacement processing can be realized and the reliability of the magnetic disk device 4 can be improved.

Further, in the above description, all of the above cases were used when the magnetic disk device was used by the upper host 1 as a system, but this method is used for the shipping inspection of the magnetic disk device at the time of manufacturing and shipping, and there is a defect. It can also be applied as a sector detection and registration tool. In other words, if read margin test and write processing are executed in addition to diagnostic read of the entire magnetic disk device, shipment acceptance / replacement processing as a magnetic disk, replacement processing, and bad sector management are all performed even if the upper host 1 is not connected. Can be achieved with.

[0027]

According to the present invention, a defective sector is automatically eliminated without an access request from a host, so that replacement processing can be performed, and data is lost at the time of a read instruction from the host. Can be prevented. In addition, there is no delay in reporting the command completion to the host, which improves performance.

Further, by controlling the number of write failures and the number of defective sectors, it becomes possible to notify the upper host that the magnetic disk is dangerous, and high reliability can be obtained.

Further, manufacturing, shipping inspection, defective sector detection, replacement processing, and defective sector registration can be performed by the magnetic disk itself.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a disk control system according to the present invention.

FIG. 2 is a memory configuration diagram in the embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Host, 2 ... Magnetic disk control device, 3 ... Management table, 4 ... Magnetic disk device, 21 ... Host interface line, 22 ... Device interface line.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Office reference number FI Technical display location G11B 20/10 C 7736-5D 20/12 9295-5D 20/18 501 C 9074-5D F 9074- 5D 550 D 9074-5D N20, N28 552 A 9074-5D 574 E 9074-5D 27/10 A 8224-5D

Claims (3)

[Claims] 1. In a magnetic disk device that replaces a defective sector with a spare sector, a diagnostic read is performed to a medium by using a time when there is no read / write request from a host, and it is determined that the sector is defective. An automatic alternation processing method characterized by performing alternation processing. 2. When an access request is issued from the host, an error address is stored and access from the host is interrupted once in order to prevent a delay in reporting the end to the host side even when automatic replacement is required. A processing method characterized by preventing performance degradation by executing automatic replacement processing again. 3. In a computer system that uses a magnetic disk as an external storage device, a host reads out a diagnostic read from the host on a regular basis by utilizing the time when access to the magnetic disk is not required. The automatic replacement processing method is characterized by executing replacement processing when a defective sector is found.