JPS62259266A - Exchange block assigning system - Google Patents

Abstract

PURPOSE:To be limited with only a defect block processing time and to remove excessive rotation waiting by setting up a logical address shear generated by assigning logically continuous good blocks to defective blocks to set a value less than a previously set value. CONSTITUTION:When an instruction for writing of continuous two blocks from a block 1, e.g., is arrived from a CPU3 by using a logical address, a controller 33 converts a specified logical address into a physical address, and at the time of detecting that a request block of the CPU31 is the block 1, reads out the ID part 11 of the block 1. The controller 33 recognizes that the block 1 is a defective block and exchange blocks are continuous blocks 2 and writes data obtained from the CPU31 in the block 2 and the block 3 to be a good block to end the processing. Since the block 2 follows the processing can be restricted only to the processing time of the block 1 to be a defective block.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a replacement block allocation method for a disk-shaped recording medium, and in particular to a fixed-length storage medium in which the data transfer start position is indicated by a logical address such as a relative block number. In a data recording device, when data is transferred from a block containing a defect,
The present invention relates to a replacement block allocation method that can avoid unnecessary rotational waiting.

[Conventional technology]

Disk-shaped recording media used in external storage devices include magnetic disks, flexible disks, and optical disks.In order to record information on these media, a collection of information called a record is usually physically stored. Store as a unit. Now, talking about magnetic tape, there is a fixed length method (FBA 1'Fixed Block Ar) to determine the recording unit (record length) mentioned above.
The recording unit of chiLture), that is, the sector method, and the variable length method (CK D... Coune Key
There is a recording unit (recording unit of a device), that is, a format method. In the variable length method, the length changes depending on the user's application. These two methods also require different read/write instructions from the central processing unit (CPU).
In the case of the variable length method, CKD, access is made by a physical address (record number) indicating the number of records from the beginning, whereas in the case of the fixed length method, FB
In the case of A, access is made using a logical address (logical block number) indicating the number of blocks from the beginning of the device.

Conventionally, in external storage devices such as FBA, in which read/write instructions from the CPU are performed using logical addresses, and data is recorded in fixed block units,
If there is a defective block (a block containing a defect in it), an appropriate replacement block is selected from a dedicated replacement block area provided for each cylinder or track.

The block is allocated as a replacement for the defective block, and data to be written on the defective block is written in the replacement block. In that case, the defective block and replacement block are linked to each other by pointers.

Specifically, in this pointer, the number of the other party is written in an ID (identification) section provided for each block. Note that this type of device is, for example, disclosed in Japanese Patent Application Laid-Open No. 58-31.
This is shown in Publication No. 13.

That is, in the conventional method, as shown in FIG. 2, when block 6 of blocks 6, 7, and 8 is a defective block, any block 10 of dedicated replacement block area 9 is selected. This is designated as a replacement block for block 6.

In order to indicate that the defective block 6 is connected to the block 10 in the replacement block area 9, the address of the block 10 is written in the ID part of the defective block 6, and read/write is not performed from the CPU side. When an instruction command arrives, it repositions to block 10 instead of processing block 6, and writes data from the CPU to block 10.

After reading the data in block 10, the process returns to block 7, writes the data in block 7 to the external storage device, and ends the process. Therefore, an extra time equivalent to one rotation is definitely spent during this time.

[Problem that the invention seeks to solve]

In this way, in the conventional replacement block allocation method, replacement blocks are selected from a dedicated replacement block area, so when multiple blocks including defective blocks are read/written consecutively, a defective block is encountered. In this case, it is necessary to reposition the defective block to the replacement block connected by the pointer and read/write data, and then return to the block next to the defective block to continue processing. There was a problem in that a rotational wait was always required to process a defective block. In other words, when the target block for reading/writing is one block and the corresponding block is a defective block, the average rotation is 1/2, and when reading/writing multiple blocks including defective blocks, the unnecessary rotation is made for one rotation. Each rotation wait time was required.

SUMMARY OF THE INVENTION An object of the present invention is to solve such conventional problems and provide a replacement block allocation method that does not require extra rotational waiting and takes only the time required to process defective blocks.

[Means for solving problems]

In order to achieve the above object, the replacement block allocation method of the present invention, when allocating a replacement block to avoid a defective block, allocates a good block that is logically contiguous to the defective block as a replacement block, and A feature of this method is that the deviation of logical addresses resulting from the allocation is reduced to a predetermined number or less.

[For production]

In the present invention, instead of selecting a replacement block from a dedicated replacement block area, a good block following a defective block is allocated as a replacement block, and
By writing in the ID part (identification information part) of the block recorded for each block where in the cylinder or track containing the defective block the defective block or replacement block is connected, or by writing all By writing the logical block number corresponding to each block into the ID section in the block, extra rotational waiting is avoided. Therefore, in the present invention, since the position of the replacement block is not fixed and the replacement block is continuous to the connected defective block, even if multiple blocks including the defective block are read/written continuously, the defect Processing can be performed without waiting for rotation when returning from a block to a replacement block or from a replacement block to a block next to a defective block. However, when such a defective/replacement connection relationship is established, the blocks following the defective block are shifted one by one, so the CPU uses the logical address without knowing that this connection relationship is established. When a read/-IF write command comes, the CPU determines which block is actually to be read/written.
It is necessary to know the control device that exists between the computer and the external storage device and controls data transfer. In such a case, the control device reads the ID part in the block and determines which block in the cylinder or track containing the defective block the defect/replacement connection relationship written in the ID part is applied to. In response to a command from the CPU, there is a method to check whether the storage device is actually It is possible to recognize which block above should be read/written.

〔Example〕

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram of one track in a storage medium showing an embodiment of the present invention, and FIG. 3 is a block diagram of an information processing system to which the present invention is applied.

In FIG. 1, 1 is a block with an internal defect, that is, a defective block, 2, 3.4 are good blocks that follow the defective block and have no internal defects, 5 is a spare block area, and 11 and 21 are IDs. Section 12.22 is the data section. In addition, in FIG. 3, 31 is a CPU;
32 is a channel, 34 is an external storage drive device, and 33 is a control device 1, for example, a magnetic disk control device when the drive device is a magnetic disk device. When an input/output command is issued from the CPU 31, the channel 32 reads /;! with a logical address added to the control device 133. Sends a command containing F. The control device 33 decodes the command from the channel 32 and instructs the corresponding external storage device to perform the operation. For example, in the case of a magnetic disk control device, during a seek operation, the magnetic head is positioned at a desired track.

During a search operation, the specified address is compared with the address written on the magnetic disk to find the desired record, and when a write instruction is issued, the contents of the main storage device are stored in the specified location. When a write or read instruction is issued, it performs functions such as reading information from a specified location and sending it to the main storage device.

In this embodiment, when the control device detects that there is a defective block 1 during initialization of the external storage device, it allocates a good block 2 following the defective block 1 as a replacement block for the defective block 1, and also I
Physical position information (
For example, cylinder, head, sector) and connect them with this pointer.

Furthermore, in the ID section of all blocks within one track, the above-mentioned defect/
Whether or not alternate connections have been taken is written using relative block No. 8 from the beginning of the track. As a result, from the defective block onwards, the numbers are obtained by subtracting the number of defective blocks from the apparent relative block number (that is, the logical address number). The block number can be accessed. Therefore, in the case of FBA, the CPU issues an instruction to read 71g using the block number from the beginning of the device, that is, the logical address, so the above defect/
After the alternate connection is taken, the controller 33 accesses the specified block by converting its logical address into a relative block number and searching for the block whose relative block number is written in the 10th copy. be able to.

Now, when a write instruction for two consecutive blocks from block 1 arrives from the CPU using a logical address, for example, the control device 33 converts the specified logical address into a physical address using a certain predetermined conversion formula. As a result of conversion, C
When the control device 33 learns that the block requested by the PU is block 1, it reads the ID part of block l,
It is recognized that block 1 is a defective block and that its replacement block is block 2 which is continuous to defective block 1. In this case, the control table 33 writes data from the CPU 31 to block 2, which is a replacement block for the defective block 1, and block 3, which is a good block, and ends the process. During this time, since block 2, which is a replacement block, is continuous with block l, which is a defective block, the extra time spent in the middle of processing is only the time to process block 1, which is a defective block.

Next, when the CPU 31 issues a writing instruction for one block to the block 3 in FIG. 1 using the logical address, the control device 33 reads the ID section 31 of the block 3 and determines where in the corresponding track there is a defect. / Find out whether alternate connections have been made. As a result, it is recognized that the data that should originally be written in block 3 should be written in the next consecutive block 4, that is, the blocks are shifted by one, and the data is written in block 4 and the process ends. This block shift is made to the end of the cylinder or track so that it does not affect the primary cylinder or the next track. Further, a spare block area 5 is provided at the end of the cylinder or track to absorb block displacement. This number of spare blocks is
The number of defective blocks is made to match the number of defective blocks estimated to exist within one cylinder or one track. For example, in an external storage device in which there is only one defective block within an average of l cylinders, the number of spare blocks is set to one. If there are defective blocks close to this number, an appropriate replacement block is selected from a dedicated replacement block area prepared for each cylinder or track, and connected to the defective block. In this case, a rotational wait occurs for a read/write command to a defective block.

FIG. 4 is an operation flowchart of control 'ii showing one embodiment of the present invention. The control device 33 is configured as shown in FIG. 4(a) (
b) By executing the program shown in (c), read/write commands from the CPU to the external storage device are processed. In FIG. 4, RBNC is the relative block number of the currently positioned block from the beginning of the track, RBNO is the expected value of the relative block number (from the beginning of the track) of the block to be processed, and RBN
E is the relative block number from the beginning of the track of the block to which data is actually transferred to the CPU, excluding the number of shifted blocks, N is the number of shifted blocks display counter, and M
is the maximum number of shifted blocks allowed within one track. Also, 0VER is a flag indicating that the number of defective/replacement block concatenation relationships in which a replacement block is assigned to the next block of a defective block has overflowed M, and when this flag is on and a defective block is encountered, , indicates that the corresponding spare block is allocated in a dedicated spare area.

First, when a read/write command is issued from the CPU, the specified logical address is converted into a relative block number within one track, and this is set as the RBNE (step 100).
Next, it is assumed that the relative block number R[3NC and the relative block number RBNE obtained by subtracting the number of shifted blocks are equal (step 101). The head position is positioned at the processing start block (step 102). and,
The deviated block number counter N is reset to O (step 103), and the flag indicating that the maximum number of deviated blocks flftM allowed within one track has been exceeded is turned off (step 10G). In this way, the ID part of the located block is read (step 105). This block is called the current block. If the flag is turned on (step 106).

It is determined whether the current block is a replacement block (step 107), and if it is a replacement block, step 11)
7, and if it is not a replacement block, it is determined whether there is a connection relationship of defective/replacement blocks before the current 1-block (step 108). Note that before the current block, the current block Including yourself. When a connection relationship does not exist, the relative block number expected value RBNO is assumed to be equal to the relative block number RBNH obtained by subtracting the number of offset blocks (step 114), and the relative block number RBNC is equal to the relative block number expected value RB.
It is determined whether it is equal to NO (step 116). When a connection relationship exists, the counter value is assumed to be equal to the number of connection relationships (step 109), and it is checked whether N≧M or not (2), that is, whether the counter value is greater than or equal to the maximum value of the number of deviation blocks (step t t O), if it is smaller than the large value of i, considering that the relative block number expected value RBN○ is the sum of the relative block number RBNE obtained by subtracting the number of deviation blocks and the counter value N (step 115), RBN
Determine whether C=RBNO (step 11G
), and if the counter value N exceeds the maximum value M, it is checked whether the current block is a defective block (step 111), and if so, it is determined whether the replacement block is the next block. (step 112), otherwise turn on flag 0VER (step 11).
3). When detecting that flag 0VER is on,
Determine whether or not the current block is a defective block (Steps 117, 118). If not, read/write the data section of the block (Step 119). Next,
The relative block number RBNE is incremented by 1 (step 121). If the process is finished now, the flow is completed (step 122). If the current block is a defective block, it is positioned as a replacement block (step 120). ), the process returns to step 105 and reads the ID part of the located block. Furthermore, if the relative block is not equal to the expected value (step tte)
, check that the flag is on (step 12)
3) Check whether the current block is a replacement block (step 124), and if so, position it in the next block (step 125); if it is a replacement block, position it in the block next to the defective block. Position (
Step 126). Then, increment the relative block number by 1 (step 127), step 10
Return to step 5 and read the ID part of the positioned block.

In this way, in this embodiment, when reading/writing a block including a defective block, the next good block following the defective block is assigned as a replacement block for the defective block, so the process is performed from the defective block to the replacement block. , or from the replacement block to the defective block, there is no need to wait for one rotation as in the conventional method, and the only extra time required is the time for processing the defective block.

〔Effect of the invention〕

As explained above, according to the present invention, whether the target block for reading/V writing is one block and the corresponding block is a defective block, or when reading/writing multiple blocks including defective blocks, , there is no unnecessary waiting for rotation, and the unnecessary processing time is only the time required to process defective blocks.

[Brief explanation of drawings]

FIG. 1 is a block diagram of blocks within one track showing an embodiment of the present invention, FIG. 2 is an explanatory diagram of a conventional replacement block allocation method, and FIG. 3 is a diagram of the configuration of an information processing system to which the present invention is applied. , FIG. 4 is an operation flowchart of a control device showing one embodiment of the present invention. 1.6: defective block, 2,10: replacement block, 3,
4, 7, 8: Good block, 5: Reserve block area,
9: Replacement block area, 11: Defective block ID section,
12: Defective block data section, 21: Replacement block rD
Part, 22: Replacement block data part, 31: CPU, 32
: Channel, 33: Control device, 34: Drive device. Figure 4 (a) Figure 4 (C)

Claims (1)

[Claims] 1. Data read/write instructions from the central processing unit to the external storage device are performed using logical addresses such as relative block numbers from a reference point, and data is stored in fixed length blocks on the rotating storage medium. In an external storage device that records data in units, when allocating a replacement block to avoid a defective block, a good block that is logically contiguous to the defective block is allocated as a replacement block, and the logical address generated as a result of allocation of the replacement block is A replacement block allocation method is characterized in that the deviation between the blocks is set to a predetermined number or less. 2. As a result of allocating a replacement block to the above defective block,
A replacement block allocation method according to claim 1, characterized in that information indicating the amount of shift in logical addresses occurring in blocks subsequent to the replacement block is recorded in a block identification information area in each block. . 3. If the number of defective blocks exceeds the predetermined number, a replacement block is allocated to a spare block area prepared in advance. Alternate block allocation method.