JP2003006996A - Disk read/write control program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To eliminate verification processing during write by lowering an uncorrectable error occurrence probability during the read-out of data from a disk unit. SOLUTION: This disk read/write control program manages prescribed n-piece of data sector in one group as a unit (sector) of prescribed L-byte of data, and makes a computer carry out the write and read of a storage medium. This program makes the computer function as a parity generating means which forms m-piece of parity sector consisting of L-byte by repeating processing to generate m-byte of error detection correction code to n-byte of data formed by extracting 1 byte from the data sector of the same group for all L-byte of data, and a read/write control means which distributes the n-piece of data sector of the same group and the corresponding m-piece of parity sector as one parity group and writes them into the storage medium so that sectors belonging to the same parity group are not adjoined with each other.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to data writing / reading control of a disk device, and more particularly to a data writing / reading control program for improving error correction capability when reading data.

[0002]

2. Description of the Related Art In a storage device such as a floppy (registered trademark) disk or a magneto-optical disk, tracks for recording data are provided on a storage medium in a concentric pattern as shown in FIG. Data is recorded by being divided into units (sectors). For example, in the case of a magneto-optical disk having a diameter of 90 mm and a storage capacity of 540 MB, the track interval is 1.1 μm and the sector length is about 4 mm. In a normal office environment, there is dust with a diameter of several μm to 10 μm. When such dust adheres to the surface of the storage medium as shown in FIG. 15, a read error occurs.
The following methods are available as measures against this error.

1. Data is read immediately after writing the data, verification is performed to verify the written data with the read data, and if the correct data cannot be read, the write / read / verify is performed again, and the data is written correctly. Make sure that. 2. As a measure against an error that occurs after writing, an error detection / correction unit (ECC unit) is provided, and an error detection / correction code is added to each sector for recording. In this method, some errors (within the error correction capability range of the ECC means) that occur during data reading are corrected by the ECC means and output.

However, each of these methods has the following problems. 1. The method of performing verification immediately after writing data involves reading the same portion at the time of writing data for writing and verifying, so even if a small amount of data is written, waiting for the rotation of the storage medium for the verification process after writing is involved. , The actual writing time is longer than the net writing time.

2. When an error that exceeds the error correction capability of the ECC means occurs in one sector, a read error still occurs. As an example of measures against this problem, for example, "Japanese Patent Laid-Open No. 5-89609" is known. In the publicly known example, as shown in the table of FIG. 16, a byte string in which the bytes at the same position are extracted from the data of a predetermined group in which the byte positions of each data sector are aligned and arranged is generated, and the data sector and the generated bytes are generated. A sector composed of a parity code or an error detection and correction code generated for a column, that is, a parity sector is written as one parity group in a storage medium.

By doing this, the EC in a certain sector is
Even if an error exceeding the error correction capability of the C means occurs, the value of each byte of the data sector in which the error occurred is
The data can be corrected by performing back calculation from the other data sector of the same party group and the value of the byte position of the parity sector. However, since the method of the publicly known example does not consider the physical arrangement of the sectors forming one parity group on the storage medium, the sectors forming one parity group are physically adjacent to each other. It may be placed.

However, when dust having a diameter of several μm to 10 μm adheres to the surface of the storage medium, the dust extends over adjacent sectors or adjacent tracks at intervals of about 1 μm, and is added to the data sector within this range. An error that cannot be corrected by the error detection and correction code is generated. In FIG. 17, dust adheres to the conventional example in which sectors forming one parity group are continuously arranged in the sectors.
An example in which an error that cannot be corrected only by the error detection and correction code added to the sector is generated will be shown. In this case, dust adheres to the first sector and the second sector of one parity group, and uncorrectable errors occur simultaneously in the first sector and the second sector. It becomes impossible.

FIG. 18 shows an example in which dust adheres to a plurality of tracks on a storage medium having a sector arrangement similar to that shown in FIG. Also in this case, error correction becomes impossible as in the example of FIG.

[0009]

SUMMARY OF THE INVENTION It is an object of the present invention to reduce the probability of an uncorrectable error occurring at the time of reading data so that the verification process at the time of writing can be omitted.

[0010]

According to a first aspect of the present invention, a predetermined number of L bytes of data are used as a unit (sector).
A disk read / write control program that manages a predetermined number n of data sectors as one group to execute writing and reading to and from a storage medium. The computer extracts one byte each from data sectors of the same group. A process of generating an m-byte error detection and correction code for n-byte data is repeated for all L-byte data to form m parity sectors consisting of L bytes, and n parity groups of the same group. The data sector and the corresponding m parity sectors are set as one parity group, and the sectors belonging to the same parity group are distributedly arranged so as not to be physically adjacent to each other and function as a read / write control unit for writing to the storage medium.

According to a second aspect of the present invention, a predetermined number of L bytes of data are used as a unit (sector), and a predetermined number of n data sectors are managed as one group to perform writing and reading on the storage medium. In the disk read / write control device, a process of generating an error detection / correction code of m bytes is repeated for all data of L bytes for n bytes of data extracted from the data sectors of the same group, one byte each, from L bytes. Parity generating means for forming m parity sectors and m parity sectors corresponding to n data sectors of the same group are regarded as one parity group, and sectors belonging to the same parity group are physically connected to each other. It is characterized in that it has a read / write control means which is arranged in a distributed manner so as not to be adjacent to each other and writes in a storage medium.

As a result, even if a physically localized data error occurs due to adhesion of dust or the like, the probability that uncorrectable errors concentrate in the sectors of the same parity group is reduced. Therefore, the verification process at the time of writing can be omitted. As a method of physically distributively disposing sectors of the same parity group in a storage medium in which sectors are arranged concentrically and the number of sectors per track is N, there is a method of making (n + m) equal to or smaller than N.
As a result, it is possible to prevent the sectors of the same parity group from being arranged at the same position on the adjacent tracks.

As another method of physically distributing the sectors of the same parity group, the following method may be used. Sectors of one parity group are arranged every k sectors by a number k that is relatively prime to N. That is, sector numbers s = i, i + k, i + 2 × k, i + 3 ×
k. ． ． Let i + (n + m−1) × k be one parity group.

FIG. 1 shows an example of sector arrangement when N = 4 and k = 3. Here, s1 to s12 in order from the inner circumference
Sector numbers are assigned up to s1, s4, s7,
s10 constitutes the parity group 1. The invention described in claim 3 is a disk read / write control program for reading data written by using the disk read / write control program according to claim 1, wherein the read / write control means cannot correct in the sector when reading the data sector. When an error is detected, the parity sector is read to perform error correction.

That is, when there is no error in the read data sector, or when the error detection and correction code is added to the sector of the read data and the error correction is possible only by its content, the parity sector is read. Omit it. Therefore, unless an uncorrectable error occurs, the time required to read data remains the same as when the parity sector is not provided. When an uncorrectable error occurs, more powerful error correction is possible.

According to a fourth aspect of the present invention, there is provided the disk read / write control program according to the first aspect, wherein the storage medium is divided into predetermined areas, and a data read error rate and seek are performed for each area. The read / write control means has an error rate measuring means for measuring an error rate, and the read / write control means performs a verification process after writing when the error rate is larger than a predetermined value.

Thus, the error countermeasure is dynamically updated and corrected not only for a temporary error caused by the adhesion of foreign matter to the surface of the storage medium but also for a read error caused by a change in the surface condition caused by the adherence of foreign matter. It is possible to reduce the probability of occurrence of an impossible error.

[0018]

FIG. 2 shows a hard disk (HD
A configuration example of the first embodiment in which the data of D) 1 is copied onto the magneto-optical disk (MO) 7 at high speed is shown. Read / write control means 5
Is set to H for the buffer 2 according to an instruction from the host device or the like
The data reading of DD1 is instructed, and the buffer 2 is HD
The data read from D1 is temporarily stored.

The parity generation means 4 initializes the parity buffer 3 in accordance with an instruction from the read / write control means 5, and every time the buffer 2 reads data for one sector from the HDD 1, a new reference is made to the read data. Parity is generated and the data in the parity buffer 3 is updated. The parity buffer 3 stores the parity data generated at that time, and has a buffer for g sectors, where g is the number of parity groups handled in one process. Since the buffer 2 is also used as a work area for creating parity data, the parity buffer 3
It is desirable to have the same capacity as.

The read / write control means 5 writes the data read from the HDD 1 into the buffer 2 and the data in the parity buffer 3 to the MO 7, and initializes (zero-clears) the parity buffer. An example of the sector arrangement on the storage medium of the parity group data shown in FIG. 1 is realized by the first embodiment, and the processing of the example shown in FIG. 6 in which dust is attached will be described with n = 3 and m = 1. In this example, the three data sectors forming the parity group G1 are respectively s
At 1, s4 and s7, one parity sector generated from the data of each data sector is written at s10.

The general processing procedure is as follows: Data of a predetermined 3 sectors read from the HDD 1 is held in the buffer 2 as a parity group. 2. Parity generation means 4
Generates a parity sector by taking the exclusive OR (xor) of the corresponding n bytes of the predetermined three sectors and sends it to the parity buffer 3. 3. The read / write control means 5 writes the data for 3 sectors of the buffer 2 and the generated parity sector to s1, s4, s7, and s10, respectively.

As a result, even if dusts are spread over a plurality of tracks as shown in FIG. 6 and a data error that cannot be corrected on a sector-by-sector basis occurs in a plurality of sectors, data cannot be written in the error-uncorrectable sector. However, in the parity group G1, there are only s1 sectors. Therefore, using the bytes of the s4, s7, and s10 sectors, s1
Sector error correction is possible.

The error rate measuring means 6 receives the verification information at the time of reading and the position information of the read data from the read / write control means 5, calculates the error rate associated with reading / writing to the MO 7 from the error occurrence state at the position, and parity. The generation unit 4 updates the value of n (the size of one parity group) according to the error rate. For example, when the error rate is high, n is set small, and when the error rate is low, n is set large.

The calculated error rate may be written in MO7 and stored in a storage medium, and this information may be read and processed when the storage medium is inserted. The number g of parity groups may be fixed for each MO disk, or M
It may be changed depending on each zone of the O disk. In this example, it is assumed that the number of parity groups at a certain point of time is g, and that m-sector parity data is generated and added to n-sector data.

FIG. 3 shows a schematic flow of processing when writing data. The read / write control means 5 performs a seek (positioning) operation according to a write request from the host device, and re-executes when a seek error occurs. Further, the error rate measuring means 6 counts the number of seek operation errors. FIG. 4 shows a schematic flow of processing when reading data. The error rate measuring means 6 counts the number of errors as in the case of writing data.

The error rate measuring means 6 stores the seek error in the above-mentioned seek operation and read operation and the occurrence frequency of the read error together with the sector number, and when the read error rate becomes larger than a predetermined value. , Perform one of the following error processes. (1) The read / write control means 5 forcibly performs verification after writing.

(2) The read / write control means 5 performs the verification after writing a predetermined number of times, and performs the sector replacement process if the read error does not disappear. (3) The parity generation unit 4 enhances the error correction capability by reducing the number n of sectors forming one parity group or increasing the number of error detection and correction codes generated for one parity group.

The above (1) is a relatively simple process, and since the difference from the conventional control is small, it is easy to implement. (2)
Judges that "errors occur frequently at the relevant position, so there is a high possibility that an error will occur even if re-execution", and the sector is replaced. In the case of (3), the error rate is expected to increase to a range of several times, and the error correction capability is increased to the extent commensurate with it.

The error detection / correction code is described, for example, in "Digital Transmission of Information", Morikita Publishing ISBN4-627-82350-9, "Computer Basic Course 18 Code Theory", Hiroshi Miyagawa et al., Shokoido et al. By using the Reed-Solomon code, stronger correction becomes possible. Further, the parity generation means 4 utilizes the characteristic that "in which sector an error has occurred can be almost specified by error detection / correction in sector units", and m parities are calculated by m independent calculation formulas. The sector can be calculated. That is,
The value of the parity sector is calculated in advance by m expressions that are independent of each other regardless of which m term is removed. If this causes an error in any sector,
It is possible to construct m relational expressions in which the value of the sector is unknown, and by solving this, the value of the sector in which the error has occurred can be obtained.

The parity sector is generated by the method described above, and when the read error occurs, the parity sector is read and the error can be corrected. FIG. 5 shows an example of pseudo code for performing error processing according to the error rate. FIG. 7 shows a configuration example of a second embodiment in which real-time required data is read from a digital video (DV) device or the like and written in MO. If real time is required,
If verification / rewriting after writing occurs, the writing time that should be guaranteed as the worst value may be exceeded.
By applying the present invention, verification after writing can be omitted, and the above possibility can be avoided.

In this example, the input data from the DV 11 is taken as the DMA processing, and the size n of the parity group is set.
Is fixed. In either case, parallel processing is facilitated and higher speed processing is possible. Also, the parity generation logic is x
Not only a simple logic such as or, but also a logic having a high error correction capability such as a Reed-Solomon code with m> 1 may be used.

FIG. 8 shows the error rate measuring means 16 of the second embodiment.
The internal configuration and the operation conceptual diagram of FIG. 4 are shown, and the data verification operation when the data acquisition from the DV 11 (writing to the MO 17) is completed will be described. In FIG. 8, the error map 62 is a table or a bit map showing the location of the sector where the uncorrectable read error occurs. The parity sector address 61 holds the number g of parity groups when the data is stored and the number n of sectors forming the parity group, and writes this value in MO17.

The read / write control means 15 reads the data just written from the MO 17 and stores it in the buffer 12. At this time, if an uncorrectable error occurs during reading, the fact is recorded in the error map 62. g × (n +
m) When the reading of the sector is completed, the error correction means 63 is activated. The error correction means 63 refers to the error map 62, and when an uncorrectable read error is recorded, tries the error correction processing shown in FIG.

By performing the above processing, the probability that unreadable sectors will remain can be made extremely small. Note that the process of FIG. 8 may be provided with a start instruction means for execution, display that the process is being executed, and start the process in response to an instruction from the operator.

In some cases, the process of FIG. 8 finds a sector that cannot be read correctly, and even if the sector is written back in the error correction process, a read error may occur. In this case, a replacement sector is usually assigned from the secondary defect list (SDL) and written to the replacement sector. If the replacement sector cannot be assigned because the replacement sector area has been used up, an error may be notified to the operator.

FIG. 10 is a block diagram of the parity sector area, the secondary parity area and the secondary parity management, as a means for reducing the predetermined number n when the error rate is larger than the predetermined value by applying the present invention. The layout example of the logical data in a disk of the file system of Example 3 provided with the rate management area is shown. In the third embodiment, the parity sector is assigned in units of clusters consisting of n sectors in the data area, that is, in units of one parity group. The number of parity sectors in the parity sector area is approximately 32000 to 64000 sectors. The secondary parity sector area is an area of a parity sector assigned to a smaller unit and is used according to the error occurrence status. Therefore, it is possible to manage which portion the secondary parity sector is assigned to. The following parity management is provided.
The error rate management area holds the position of the error frequently occurring area on the storage medium.

In this file system, when writing data to the disk, parity data is created in cluster units and written in the parity sector area. Further, referring to the secondary parity management, if it is an area where an error frequently occurs, for example, the parity for the first half of the cluster is obtained and this is written in the secondary parity sector area. This is substantially equivalent to halving n.

A pseudo code example of the secondary parity sector processing is shown in FIG. In the figure, the process of is 8 KByte (= 64 Kbi
Bitmap of t), that is, a bit map of "1" if the cluster in which the secondary parity sector is to be recorded, and a bit map of "0" otherwise. In the process of, the structure of the target position of the secondary parity sector, that is, the “cluster number” (2 bytes), the “position within the cluster”, and the “sector number” (each 4 bits) is arranged. If {1203, 4, 4}, it means that 4 sectors of parity are created from the 4 / 16th portion from the beginning of the cluster 1203.

The process of step (2) obtains the parity of the secondary parity sector, that is, the portion pointed to by. The data read operates almost as a conventional file system except for the following points. That is, (1) When a read error occurs, its location and frequency are counted and registered in the error rate management unit. (2) When an uncorrectable read error occurs (the normal file system reports the error as it is), the error occurrence location is recorded as system information. As a result, the software started immediately after that can know the place where the error occurred and can repair the error from the corresponding parity information.

Although the FAT file system is assumed here, other file systems can have the same configuration as long as it has a concept corresponding to a cluster or allocation unit. FIG. 12 shows an example of a cluster data write processing flow according to the third embodiment. As Example 4,
An example in which the present invention is applied to the magneto-optical disk device as its own function will be described. An MO disk of 230 MByte to 640 MByte adopts a method called ZCLV. This divides the recording area of the disc into 10 to 17 zones and changes the number of sectors per track for each zone. The number of sectors per track is small on the inner circumference, and increases toward the outer circumference. For example, M of 540 MByte
The O disk consists of 18 zones. Innermost is 87
There are 5 tracks with 50 sectors per track, and the outermost area is 875 tracks with 84 sectors per track.

Based on the number of sectors N per track for each zone, one parity group is provided every k = N−3 sectors. That is, regarding the innermost part, the sector with the sector number s is set to the (s mod 47) th parity group. One parity sector is prepared for each parity group, a parity value is calculated for each parity group when writing data, and the parity sector is written.

FIG. 13 is a conceptual diagram of the sector arrangement of the fourth embodiment. In the figure, only the sectors arranged as one parity group on the storage medium are indicated by lines. For example, the width of the hatched portion is about 20 μm, that is, 2
Although it is 0 track, even if dust of about 10 μmφ is attached to this track, the sectors belonging to the same parity group, which are arranged in the area of 10 μm in the circumferential direction for 10 tracks, remain within 1 sector, and error correction is performed. Is possible.

In the fourth embodiment, a parity sector area for storing the parity value of the sector in the same zone is provided for each zone. In the innermost example, k is N-3
= 47 sectors. The size of the parity group in the zone is the total number of sectors / k = 875 × 50/47 = 9.
81, and the number of parity sectors is 47. Further, an area for managing the state of each sector is provided for each parity group. It manages three states of {released state / written / overwritten} for each sector. Since 2-bit information is sufficient for state management, one sector, that is, k state management sectors, is sufficient for each parity group.

To make the disc compatible with conventional discs,
The initial defect list (PDL) is registered as if there were 2k defective sectors in the low level format. The sector registered as the defective sector is used as the parity sector. Hereinafter, for the sake of explanation, when the linearly numbered “sector number s” is k × j + (i−1), this sector belongs to “parity group i”, and “parity sector i” and “state management” It is assumed to be managed by the sector i ".

FIG. 14 shows the data writing procedure of the fourth embodiment. An example of the operation of the user application program when writing a large amount of data such as image data is shown below. 1. Initialize the disk in advance. 2. The size of the data to be written is estimated (with a margin), an area for that is secured on the disk, and the area is released as necessary. The estimation method is estimated, for example, from the approximate time and bit rate of moving image data. 3. Write data sequentially. 4. Close the file. 5. Eject the disc.

Along with this, the following processing is internally performed. 1. Initialize the state management sector and parity sector. 2. Initialize the status management sector and parity sector of the write zone. 3. The state management sector is updated according to the writing, and the parity sector is updated. If a zone switch occurs, write these to disk. 4. Write the parity sector and state management sector held in the buffer to the disk.

In the above processing, the contents of the state management sector do not indicate {overwritten}. Regarding the data thus written, even if the data becomes unreadable after the writing due to adhesion of dust or the like, as long as it remains in one sector in the parity group, it can be restored from the values of other sectors. It should be noted that although it has been described here that one zone is collectively managed for convenience of explanation, one parity group can be made smaller by making k larger or subdividing the zone into the first half and the second half. .

As a fifth embodiment, a storage device may be used in which n times the sector length on the recording medium, that is, one parity group is viewed as a logical sector length. For example, in a storage device using a storage medium having a sector length of 512 Bytes or 2 KBytes, it is possible to make a plurality of sectors managed as one parity group and notify the host side of the sector length as 16 KBytes or 32 KBytes, for example. The present invention can be applied by preparing one or more parity sectors for each parity group. In general, the positional relationship between a data sector and a parity sector that are put together as a parity group is arbitrary. For example, it may be located at a place continuous with the sectors grouped as a parity group, or only the parity sectors may be grouped in a management block or the like.

As a sixth embodiment, application software for collectively backing up hard disk data may be used. For example, backup software writes a large amount of data at once. Also, with this software, backup, or writing, is an important factor. Such software can perform its own file management. It is also possible to generate unique parity data for each backup file and write this. For example, parity data for m sectors may be generated and written each time backup writing reaches n sectors.

As the seventh embodiment, a file system, particularly a file system with data compression, may be used. For example, in the FAT16 format file system,
Disk areas are managed in units called clusters.
When the disk capacity is about 508 MBytes, the cluster size is 8 KBytes, and when the sector length is 512 Bytes, it is 16 sectors. Therefore, when formatting a disk and constructing a file system, about 1/16 of the entire area is allocated as a parity sector. Thereafter, each time a file is written (since this writing process can be handled in cluster units), parity sector data is generated for each cluster and this is written in a separately allocated area.

In a file system in which the apparent disk capacity is made larger than the physical disk capacity by performing data compression, data having a size of 8 KBytes to 32 KBytes is compressed as one unit, and this is physically It becomes possible to write and manage in a small area. (Supplementary Note 1) A disk read / write control program that manages a predetermined number n of data sectors as a group by using a predetermined number of L bytes of data as a unit (sector), and causes a computer to perform writing and reading to and from a storage medium. Therefore, for each n-byte data extracted by the computer from the data sector of the same group, one byte each, m
The process of generating the byte error detection and correction code is repeated for all the data of L bytes, and the parity generation means that forms m parity sectors of L bytes and the m number of data sectors corresponding to the n data sectors of the same group. A disk read / write control program for functioning as a read / write control unit that distributively arranges sectors that belong to the same parity group so as not to be physically adjacent to each other and uses them as a parity group and writes them to a storage medium.

(Supplementary Note 2) A disk read / write control device that manages a predetermined number n of data sectors as a group by using a predetermined number of L bytes of data as a unit (sector) and performs reading and writing from / to a storage medium. Therefore, a process of generating an error detection and correction code of m bytes is repeated for n bytes of data extracted by 1 byte each from data sectors of the same group for all L bytes of data, and m parity of L bytes is generated. Parity generating means forming a sector and m parity sectors corresponding to n data sectors of the same group are set as one parity group and distributed so that sectors belonging to the same parity group are not physically adjacent to each other. A disk read / write control device having a read / write control means arranged and writing in a storage medium. .

(Supplementary Note 3) A disk read / write control program for reading data written using the disk read / write control program according to claim 1, wherein the read / write control means detects an error that cannot be corrected in the sector when reading the data sector. A disk read / write control program which, when detected, reads the parity sector and performs error correction.

(Supplementary Note 4) The disk read / write control program according to claim 1, wherein the storage medium is divided into predetermined areas, and a data read error rate and a seek error rate are measured for each individual area. A disk read / write control program comprising error rate measuring means, wherein the read / write control means performs verification processing after writing when the error rate is larger than a predetermined value.

(Supplementary Note 5) A computer-readable recording medium in which data is written using the disk read / write control program according to claim 1. (Supplementary note 6) The disk read / write control program according to claim 1, wherein the storage medium is divided into predetermined areas, and an error rate measurement for measuring a data read error rate and a seek error rate for each individual area. When the error rate is larger than a predetermined value, the parity generation means increases m / n, that is, decreases the predetermined number n or increases the predetermined number m. Disk read / write control program.

(Supplementary Note 7) The disk read / write control program according to claim 1, wherein the sectors are arranged concentrically and the number of sectors per track is N, and the read / write control means is N A disk read / write control program characterized in that sectors of the same parity group are distributed and written in a storage medium for every k sectors by a prime number k.

[0057]

As described above, according to the present invention, 1. The same or equivalent hardware as the conventional storage device can be used to provide a storage device with a higher writing speed and more stable writing speed. The overhead for this is the storage capacity of the disk used as the parity sector, that is, 1% of the storage capacity of all disks.
It is about several percent. 2. The error correction capability is enhanced, the verification after writing can be omitted, and the speed of the data writing process can be realized. Further, by omitting the verification after writing, the actual writing speed becomes stable, and the writing speed that can be guaranteed to the host or application becomes high. This makes it possible to expand the uses of the storage device.

[Brief description of drawings]

FIG. 1 Example of sector allocation

FIG. 2 is a configuration example of the first embodiment.

FIG. 3 is an example of a data write processing flow according to the first embodiment.

FIG. 4 is an example of a data read processing flow according to the first embodiment.

FIG. 5: Pseudocode example for error processing according to error rate

FIG. 6 is an example of sector arrangement and dust adhesion according to the first embodiment.

FIG. 7 is a configuration example of the second embodiment.

FIG. 8 is a conceptual diagram of the internal configuration and operation of error rate measuring means according to the second embodiment.

FIG. 9: Example of error correction processing

FIG. 10 is a layout example of logical data in a disk according to the third embodiment.

FIG. 11 is a pseudo code example of secondary parity sector processing.

FIG. 12 is an example of a cluster data write processing flow according to the third embodiment.

FIG. 13 is a conceptual diagram of sector arrangement according to the fourth embodiment.

FIG. 14 is a data writing procedure of the fourth embodiment.

FIG. 15 is an example of tracks and sectors for recording data concentrically.

FIG. 16 shows an example of parity sector generation.

FIG. 17 is an example 1 in which dust adheres to a conventional example in which sectors are continuously arranged.

FIG. 18 is an example 2 in which dust adheres to the conventional example in which sectors are continuously arranged.

[Explanation of symbols]

c, d: Number of times of seek retry g: Number of parity groups handled in one process G: Parity group number k: Interval at which sectors constituting one parity group are arranged m: Number of parity sectors added to one parity group n: Number of data sectors forming one parity group N: number of sectors on one track L: data length per sector s: sector number

─────────────────────────────────────────────────── ─── Continued Front Page (51) Int.Cl. ⁷ Identification Code FI Theme Coat (Reference) G11B 20/18 G11B 20/18 572F

Claims (5)

[Claims] 1. A disk read / write control program that manages a predetermined number n of data sectors as a group by using a predetermined number of L bytes of data as a unit (sector), and causes a computer to execute writing and reading to and from a storage medium. That is, the process of generating an error detection and correction code of m bytes is repeated for all data of L bytes for n bytes of data extracted by the computer from each data sector of the same group one byte at a time. Parity generating means for forming a plurality of parity sectors and m parity sectors corresponding to n data sectors of the same group are regarded as one parity group, and sectors belonging to the same parity group are not physically adjacent to each other. Function as a read / write control means for writing data in a storage medium Disk read / write control program. 2. A disk read / write control device that manages a predetermined number n of data sectors as a group by using a predetermined number of L bytes of data as a unit (sector) and performs writing and reading to and from a storage medium. , A process of generating an error detection and correction code of m bytes is repeated for n bytes of data extracted by 1 byte each from data sectors of the same group for all L bytes of data, and m parity sectors of L bytes are generated. The parity generation means to be formed and m parity sectors corresponding to n data sectors of the same group are defined as one parity group,
A disk read / write control device comprising: a read / write control unit that distributively arranges sectors belonging to the same parity group so that they are not physically adjacent to each other and writes them to a storage medium. 3. A disk read / write control program for reading data written using the disk read / write control program according to claim 1, wherein the read / write control means detects an uncorrectable error in the sector when reading the data sector. In this case, the disk read / write control program is characterized by reading the parity sector and performing error correction. 4. The disk read / write control program according to claim 1, wherein the storage medium is divided into predetermined areas and a data read error rate and a seek error rate are measured for each area. A disk read / write control program comprising a measuring means, wherein the read / write control means performs verification processing after writing when the error rate is larger than a predetermined value. 5. A computer-readable recording medium in which data is written using the disk read / write control program according to claim 1.