JPH08315547A - Block replacement position identification method - Google Patents

Abstract

(57) [Abstract] [Purpose] In the block alternation process of the storage controller that employs the fixed-length block recording method, the position of the block to be alternated is specified even in the defective track where the block identifier of the alternate target block cannot be detected Allow them to take turns. [Structure] A replacement history is newly added for each entry of the defect list and registered in the storage controller. In a defective track where the target replacement logical block corresponding to the block replacement request from the host cannot be detected, the logical block arrangement can be estimated by referring to the defect list, and the physical position of the target block specified by the host can be detected. So that the replacement process can be executed. If there is a block for which the identifier cannot be detected even for blocks other than the purpose of the block replacement as a completion report of the block replacement process, the fact and address are reported as a warning to the host.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a block replacement method, and more particularly to a method for specifying a replacement target block position when a block identifier to be replaced cannot be detected.

[0002]

2. Description of the Related Art A is a typical example of an interface between an external storage device adopting a fixed block recording method and a host.
NSI Committee (American National Standed Ins
SCSI standard established by titute) (Small Computer)
There is a System Interface, but in the SCSI standard, a block replacement command (Reassign Blocks C) is used as a command for assigning a defective block that cannot be read / written due to a defect on the recording medium to a normal block.
ommend) is specified.

There are a logical block address and a physical block address as the address of the block. The logical block address has a one-to-one correspondence with the block identifier on the recording medium, and is used as an address specified during normal reading / writing from the host. The physical block address is a sequential number assigned to each block in ascending order from the index that is the start position of one track,
It is used as an address for managing the position of the bad block.

In the block replacement command, the host sends the logical block address of the defective block to be replaced to the storage controller as a command parameter.

A block replacement command processing method is, for example, as described in Japanese Patent Laid-Open No. 1-105069, a primary defect list in which addresses of defective blocks detected before user shipment are recorded and defects generated after user shipment. There is a method of managing a replaced block by providing a secondary defect list in which the address of the block is recorded.

The physical block address of the defective block is stored in each defect list.

The control program of the storage control device refers to the defect list at the time of format writing on the recording medium, and performs formatting assuming that the defective block is not used.
At the end of the unit (for example, one track) that eliminates the change,
Since it is decided to provide a predetermined spare block,
The bad blocks registered in the defect list are not directly assigned to the spare blocks, but the blocks are not used, and the entire block identifiers after the bad blocks are shifted backward to the spare blocks one by one, and format writing is performed. To do.

After a format write, even if a new defective block is detected by the verify operation for checking whether the format write is normally performed, the entire block after the defective block is shifted to the spare block and the format is again written. This is a technique for preventing an undesired rotation wait due to a direct block change in a continuous read / write operation performed after formatting. A unit that eliminates alternation in format writing and verification (for example, one track)
If more than the number of spare blocks is registered in the defect list, the spare blocks are replaced until the spare blocks are exhausted, and the defective blocks that cannot be completely shifted are replaced by another track. Directly to. The defective block address detected by the verification is added to the secondary defect list.

In the block replacement command after user shipment, the designated block is directly assigned to the spare block. In order to perform the alternation process of shifting the defective block and the subsequent blocks to the spare block, it is necessary to reformat the track. In the format operation, all data on the formatted track will be lost,
In order for the storage control device to shift the replacement process after shipment from the user, it is necessary to temporarily save the data written by the user to some area and restore it after formatting. Moving the physical storage location of the data written by the user involves the risk of losing the user's data. Therefore, in the replacement processing after the user is shipped, the replacement is performed by directly allocating it to a spare block or a replacement track. I am trying. The block address specified by the block replacement command is also added to the secondary defect list.

That is, in the defect list, the block address shifted and replaced and the block address directly replaced are mixed and registered.

[0011]

In the block replacement command processing after user shipment as described above,
In particular, in the block replacement process of the magnetic disk control device, since the defective block is a block that cannot be accessed again, the defective block is recorded in the identifier of the block in the normal state. To do this, it is necessary to detect the physical block address in the track where the bad block resides.

However, when defective blocks whose block identifiers cannot be detected are concentrated on a certain track, the disk control device cannot specify the block to be replaced and abnormally terminates the block replacement process. Was there.
The cause of the abnormal end is that the disk controller cannot detect a block identifier other than the host designation to the host, and the additional information of the abnormal end indicates that the block identifier other than the block address specified by the host cannot detect the block identifier. Report the address. The host stipulates in the SCSI standard that a new bad block address reported from the disk controller is added to the parameter of the alternate block command and the block alternate command is reissued. However, some hosts do not support the function of adding a new bad block and reissuing the block replacement command with a block replacement command, and the bad block cannot be replaced and the disk storage device is replaced. There was a problem I had to do.

An object of the present invention is to newly provide a function of specifying the position of a target block to be replaced, so that one block designated by the host can be used even in a track in which defective blocks whose block identifiers cannot be detected are concentrated. The purpose of the present invention is to execute block replacement processing by specifying by inference and enabling block replacement.

[0014]

In order to achieve the above object, according to the present invention, each time a block is replaced after a format write, the replacement history of a defective block is determined by whether the defective physical block is shifted or directly replaced. Record in the defect list together with the address. When a block alternation command is received from the host, in order to specify the physical block address of the defective block, all columns of the block identifier of the track in which the defective block exists are read. If the bad block identifier specified by the host cannot be detected in the sequence of all read block identifiers, one of the read identifiers is used as a reference and the replacement history of the defect list is traced to create an array of block identifiers. I will do it. Here, by tracing the alternation history of the defect list, a sequence of all block identifiers is created, and the logic of estimating the block identifier of the unreadable portion is defined as inference.

A comparison check of the arrangement of block identifiers created by inference and the arrangement of all the read block identifiers is performed to verify the validity of the arrangement of block identifiers created by inference. When the comparison check results in disagreement, the reference block is switched to another readable block identifier, and the creation of the block identifier arrangement by inference and the comparison check are repeated. This operation is repeated until the comparison between the block identifier created by inference and all the block identifiers that can be read matches. As a result, if more block identifiers than the number of spare blocks can be read in the alternate-designated track, the physical block address of the defective block identifier for the alternate purpose can be specified and the alternate process can be executed.

[0016]

In a magnetic disk drive that has been read / written for a long time, the probability of defective blocks increasing. There are various causes for the generation of a bad block, but it is also conceivable that a plurality of block identifiers may be completely destroyed in the same track, making it impossible to read or write the block. The host attempts to replace a bad block and rewrite data, but if this attempt fails, the magnetic disk drive must be replaced.
In the block identifier identification method based on inference using the block replacement command, even if there are multiple defective blocks whose identifiers cannot be detected in addition to the defective blocks designated by the host on the same track, the defective blocks designated by the host must be identified and replaced. You can In such a case, a warning can be given to the host by reporting the address of a defective block other than the block which the host has alternately designated by the command end report. Since the block alternation command can be executed in any case, a bad block can be recovered even if the magnetic disk drive has been used for a long time without replacing the drive.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail below with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of a magnetic disk subsystem to which the present invention is applied.
In FIG. 1, 10 is a magnetic disk controller, 20 is a magnetic disk drive, and 30 is an interface cable for connecting the magnetic disk controller and the magnetic disk.

The magnetic disk controller 10 has a defect list 11 as a control table. In the defect list, a replacement physical identification is provided as a separation flag as to whether the defective physical block address and the defective block are shifted or replaced directly.

Details of the inside of the magnetic disk drive 20 are shown at 21.
Shown in

One track is equally divided into blocks, and a block identifier and a data part are separately recorded in one block. The block identifier has a logical block address of the block and a flag indicating whether the block is normal, defective, or spare. The data section records the data received at the time of a write request from the host. In the present example, the unit for eliminating the alternation is one track, and several spare blocks (two in this example) are provided at the end of the logical block address of one track.

An example in which the logical block address 3 is shifted and replaced by the format write or verify is shown in (1). One of the spare blocks is used, and the block identifiers after the logical block address 3 are shifted one by one to the spare block.

Further, it is assumed that the host issues a block alternation command to the logical block address 3 for the track state of (1). Since the logical block address 3 is directly changed to the spare block, the block identifier is arranged as shown in (2).

Now, from the host, the logical block address 3
Assuming that the block identifier is specified by the block alternation command, and the four blocks of the identifier including the logical block address 3 in the track intended for alternation cannot be detected, the procedure until the block identifier is specified is shown in FIG. 2 and FIG. A description will be given according to the flowchart of FIG. A block whose block identifier cannot be detected is indicated by x. In addition, α is added to the defect list as the defective physical block address of the replacement target track.
And β have been recorded. Physical block address α
It can be seen that they are offset and alternate, and β is directly alternate.

The track in which the logical block alternately designated by the host exists is obtained (step 101).

In order to obtain the physical block address of the logical block address designated by the host, all the identifiers of the track are read from the magnetic disk drive (step 102). It is checked whether or not there is a logical block address designated by the host in the read block identifier (step 103). If the logical block address designated by the host exists, the physical position of the block is obtained and block replacement processing is executed (step 104).

The state of arrangement of block identifiers read from the magnetic disk drive is shown in (1) of FIG. In this case, since there is no block identifier corresponding to the logical block address 3 designated by the host, the process proceeds to step 105.

It is checked whether there are three or more block identifiers that can be read (step 105). If there are not three or more, the logical block address designated by the host cannot be specified, and the block alternation process is abnormally terminated (step 106).

In the state (1) of FIG. 2, since there are six block identifiers that can be detected, the process proceeds to step 107.

Using one of the read block identifiers as a reference, an array of block identifiers is speculatively created by referring to the defect list (step 107).

As shown in (2) of FIG. 2, a block identifier array is created based on the logical block address 0. Since the physical block address α and subsequent shifts occur,
A block identifier sequence is created by shifting the block identifier to the spare block. Next, in (3), since the physical block address β is directly assigned to the spare block, the block identifier of the logical block address 3 is directly assigned to the spare block.

A sequence of block identifiers created by inference, and
A comparison check is performed to see if the physical positions of the respective logical block addresses match all in the array of all the block identifiers that could be read (step 108). The comparison result is judged (step 109). If the comparison result does not match, the reference of the logical block address is switched to another (step 110) and the inference and the check are repeated until the comparison check is established. When the comparison between the block identifier created by inference and all the read block identifiers is in agreement, the physical position of the logical block designated by the host is obtained and the alternation process is executed (step 111).

In (4) of FIG. 2, since the read logical block addresses are 6, 0, 1, 2, 4, and 5,
Logical block address 6, 0, 1, created by guessing
Compare with 2, 4, and 5. Since the physical position of each block identifier created by the estimation and the physical position of each block identifier that could be read completely match, the estimation is determined to be correct, and the process proceeds to step 111. In (6) of FIG. 2, the physical position of the logical block address 3 designated by the host is 2
Since it was possible to specify that, the replacement process can be executed.

As a normal termination report of the alternation process, it is reported that there is a defective block on the track other than the block designated by the host (step 112).

In (5) of FIG. 2, it is possible to detect that the logical block address that could not be detected was 7 when reading the block identifier. That is, the logical block address 7 is reported as a defective block address other than the block address designated by the host.

[0036]

As described above, according to the present invention, three or more block identifiers are detected even in a defective track in which the block identifier of the logical block alternate-designated by the host cannot be detected in the block alternation processing. If possible, the physical position of the block to be replaced can be specified by the defect list and the inference, so that the block replacement process of only the logical block designated by the host can be executed without abnormally terminating the replacement process. In addition to the block designated by the host, it is possible to warn that there is a defective block whose block identifier cannot be detected.

[Brief description of drawings]

FIG. 1 is a system configuration block diagram showing an embodiment of the present invention.

FIG. 2 is a track diagram showing the process until the block replacement position is specified.

FIG. 3 is a flowchart showing a procedure for identifying a block alternation position in FIG.

[Explanation of symbols]

10 ... Magnetic disk controller 20 ... Magnetic disk drive 30 ... Interface cable connecting magnetic disk controller and magnetic disk drive 11 ... Defect list 21 ... Track format in magnetic disk drive

 ─────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Atsushi Takayasu 2880 Kokufu, Odawara-shi, Kanagawa Stock company Hitachi Storage Systems Division

Claims (2)

[Claims] 1. A defect list in which an address of a defective block is recorded in a replacement target block position specifying method of a replacement function for replacing a defective block with a normal block in a subsystem of an external storage device adopting a fixed-length block recording method. By registering the replacement history in, if more block identifiers than the number of spare blocks can be detected even if a plurality of block identifiers cannot be detected in one track, the purpose of replacement is to infer by referring to the defect list. Identifies blocks so that block replacement can be performed, and even when the abnormal end of block replacement was not reported to the host until now, it was possible to execute block replacement processing even if the block replacement abnormal termination was reported to the host. Method. 2. Including a block which is designated by the host in turn, 1
When multiple block identifiers cannot be detected in the track,
When the replacement target block is specified and the block replacement is performed by inference with reference to the defect list, the block replacement command completion report indicates that there is a bad block whose block identifier cannot be detected in addition to the block specified by the host replacement. How to warn.