JPH06162689A - Magnetic disk controller - Google Patents

Abstract

PURPOSE:To early discover generation of an abnormal state in a magnetic disk controller by providing a patrol read control unit, an ECC control unit and a read error reporter. CONSTITUTION:When a command from a host system 10 is not executed, a track number storage unit 24 transmits a track number to a patrol read control unit 25. The unit 25 transmits to a processor 23 to execute a patrol reading operation. The processor 23 instructs a data transfer control unit 22 to execute the patrol reading operation. The unit 22 operates a magnetic disk apparatus 30 to read data. An ECC control unit 26 detects whether a read error is present or not according to input data and ECC, and reports read error to a read error reporter 27. The reporter 27 reports it to the system 10 through the processor 23.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a magnetic disk controller for controlling read / write operation of information in a magnetic disk device.

[0002]

2. Description of the Related Art In a magnetic disk device, if the magnetic head stays at the same position on the surface of the magnetic disk (medium), when an abnormal flying state of the magnetic head occurs, the track at that position may be damaged. Therefore, the service life of the magnetic disk device may be shortened due to the high value. For this reason, the conventional magnetic disk control device uses the dummy seek operation by the patrol seek operation when the information read or write operation is not performed (idle time) so that the magnetic head does not stay at the same position on the medium surface. The magnetic head is moved uniformly over the entire area of the medium surface.

Further, when the axis of the magnetic disk device is deviated due to aging, the read operation margin (read margin) is lowered and a read error is likely to occur, and the read / write characteristics of the magnetic disk device are deteriorated. However, since the read margin is reduced and a read error is likely to occur, the reliability of the magnetic disk device is reduced in any case. Therefore, if there is a track that cannot be read normally due to misalignment of the axis of the magnetic disk device, deterioration of read / write characteristics, defect of the magnetic disk, etc., operation to read information from that track by a software command. The process for detecting and recovering from the abnormality is performed.

[0004]

As described above, the conventional magnetic disk control device detects and recovers an abnormality by a software command when there is a track in the magnetic disk device that cannot perform a normal read operation. Since processing is performed, it takes a long time for execution processing, and there is a problem that the input / output performance of the magnetic disk device and the entire information processing system is low.

Further, the above-described processing has a problem that it is difficult to quickly find an abnormality in the magnetic disk device because it depends on a software instruction.

Further, when a correctable read error occurs, if it is left as it is, it becomes an uncorrectable read error and the data is lost.

[0007]

A first magnetic disk control device of the present invention controls transfer of data between a processor whose operation is controlled by firmware and a host device by control of the processor. A first data transfer control unit, a second data transfer control unit that controls transfer of data and ECC to and from a magnetic disk device under the control of the processor, and a write operation is actually performed by the processor during a write operation. A track number storage unit for storing the number of the track that has been deleted, and a patrol read operation for the track of the number transmitted from the track number storage unit when a command from the host device is not being executed. A patrol read control unit for transmitting to the processor, and the second data transfer from the first data transfer control unit A read error detection / correction code for the data transferred to the control unit is generated, and a read error is detected by detecting whether or not there is a read error in the data transferred from the second data transfer control unit and the read error detection / correction code. If there is, an ECC control unit that determines whether or not correction is possible and, if correctable, generates the correction information;
When the CC control unit determines that there is a read error, it is provided with a read error reporting unit for transmitting the fact to the processor.

The second magnetic disk controller of the present invention is
Between the processor whose operation is controlled by the firmware, the first data transfer control unit which controls the transfer of data to and from the host device under the control of the processor, and the magnetic disk device under the control of the processor. A second data transfer control unit that controls the transfer of data and ECC, a track number storage unit that stores the number of the track actually written by the processor during a write operation, and an instruction from the host device. A patrol off-rollack marginal read control unit that transmits to the processor to perform a patrol off-roll rack marginal read operation on the track of the number transmitted from the track number storage unit when not executing;
A read error detection and correction code for the data transferred from the first data transfer control unit to the second data transfer control unit is generated, and the read error detection and correction data transferred from the second data transfer control unit An ECC control unit that detects whether or not there is a read error in the code, determines whether or not the read error is correctable, and generates correction information when the read error is correctable, and the ECC control And a read error reporting unit for transmitting the read error to the processor when the unit determines that there is a read error.

A third magnetic disk controller of the present invention is
Between the processor whose operation is controlled by the firmware, the first data transfer control unit which controls the transfer of data to and from the host device under the control of the processor, and the magnetic disk device under the control of the processor. A second data transfer control unit that controls the transfer of data and ECC, a track number storage unit that stores the number of the track actually written by the processor during a write operation, and an instruction from the host device. A patrol read control unit for transmitting to the processor so as to execute a patrol read operation for the track of the number transmitted from the track number storage unit when not executing; and the first data transfer control unit for transmitting the patrol read control unit to the processor. Generates a read error detection and correction code for the data transferred to the second data transfer control unit. The data transferred from the second data transfer control unit and the read error detection / correction code are detected whether or not there is a read error, and when there is a read error, it is determined whether or not the correction is possible and the correction is made. An ECC control unit that generates correction information when possible, a read error reporting unit that notifies the processor when there is a read error in the ECC control unit, and the second data transfer control A data recovery buffer for temporarily holding data transferred from the second data transfer controller and correcting the data in accordance with the correction information input from the ECC controller, and data transferred from the second data transfer controller or the data recovery buffer transfer First read data path control unit for selecting any one of the received data and transmitting it to the first data transfer control unit A second data transfer control unit that selects one of the data transferred from the first data transfer control unit and the data transferred from the data recovery buffer and sends the selected data to the second data transfer control unit. And a second read data path control unit.

A fourth magnetic disk controller according to the present invention is
Between the processor whose operation is controlled by the firmware, the first data transfer control unit which controls the transfer of data to and from the host device under the control of the processor, and the magnetic disk device under the control of the processor. A second data transfer control unit that controls the transfer of data and ECC, a track number storage unit that stores the number of the track actually written by the processor during a write operation, and an instruction from the host device. A patrol read control unit for transmitting to the processor so as to execute a patrol read operation for the track of the number transmitted from the track number storage unit when not executing; Generates a read error detection and correction code for the data transferred to the second data transfer control unit. The data transferred from the second data transfer control unit and the read error detection / correction code are detected whether or not there is a read error, and when there is a read error, it is determined whether or not the correction is possible and the correction is made. An ECC control unit that generates correction information when possible, a read error reporting unit that notifies the processor when there is a read error in the ECC control unit, and the second data transfer control A data recovery buffer for temporarily holding data transferred from the second data transfer controller and correcting the data in accordance with the correction information input from the ECC controller, and data transferred from the second data transfer controller or the data recovery buffer transfer First read data path control unit for selecting any one of the received data and transmitting it to the first data transfer control unit A second data transfer control unit that selects one of the data transferred from the first data transfer control unit and the data transferred from the data recovery buffer and sends the selected data to the second data transfer control unit. A second read data path control unit, and E
And an offset read control unit for slightly moving the magnetic head of the magnetic disk device to the inner and outer circumference sides of the track by the uncorrectable information input from the CC control unit to execute the read operation again. ing.

A fifth magnetic disk controller according to the present invention is
A processor whose operation is controlled by firmware, a first data transfer control unit that controls transfer of data to and from a host device by control of the processor, and a plurality of magnetic disk devices by control of the processor And a second data transfer control unit for controlling the transfer of data and ECC between the magnetic disk device and the plurality of magnetic disk devices, which are actually written by the processor during a write operation to the corresponding magnetic disk devices. A plurality of track number storage units that store the numbers of tracks performed and a patrol read to any one of the plurality of magnetic disk devices by monitoring the plurality of track number storage units. The patrol read priority control unit that decides whether to execute the operation and the command from the host device. And a patrol read control unit for transmitting to the processor to execute a patrol read operation for a track of a designated number of the magnetic disk device when determined by the patrol read priority control unit when not executing A read error detection / correction code is generated for the data transferred from one data transfer control unit to the second data transfer control unit, and the read error detection / correction code and the data transferred from the second data transfer control unit are generated. Whether or not there is a read error is detected, and if there is a read error, it is determined whether or not it can be corrected, and if it is correctable, correction information is generated.
The CC control unit, the read error reporting unit that notifies the processor when there is a read error in the ECC control unit, and the data transferred from the second data transfer control unit temporarily. Hold the EC
The data recovery buffer for correcting according to the correction information input from the C control unit, and either the data transferred from the second data transfer control unit or the data transferred to the data recovery buffer is selected. Of the first read data path control unit to be sent to the first data transfer control unit and the data transferred from the first data transfer control unit or the data transferred from the data recovery buffer Of the magnetic field of the magnetic disk device according to the second read data path control unit for selecting and transmitting the selected data to the second data transfer control unit and the uncorrectable information input from the ECC control unit. An offset read control unit that slightly moves the head to the inner and outer circumference sides of the track and performs the read operation again. It is provided.

[0012]

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

FIG. 1 is a block diagram showing a first embodiment of the present invention.

The magnetic disk controller 20 of the embodiment shown in FIG.
Controls the operation of the magnetic disk device 30 in accordance with a command from the host device 10.

The magnetic disk controller 20 controls data transfer between the processor 23 which is controlled by firmware and operates according to an instruction from the host device 10 and the control signal 200 from the processor 23. The magnetic disk device 3 is controlled by the data transfer control unit 21 and the control signal 201 from the processor 23.
The second data transfer control unit 22 that controls the data transfer with 0, and the number of the track actually written by the control signal 202 from the processor 23 at the time of the write operation (write operation) is stored. Track number storage 24
When the instruction from the higher-level device 10 is not executed (during idle), the processor for executing the patrol read operation for the track having the number transmitted by the track number transmission signal 203 from the track number storage unit 24. And a first data transfer control unit 21 for sending a read execution signal 204 to the patrol read control unit 25.
ECC (error information) is generated from the data transferred from the data path 205 to the data path 205, and whether there is a read error in the data transferred from the second data transfer control unit 22 via the data path 206. If there is a read error, it is determined whether it can be corrected, and if it is correctable, an ECC control unit 26 that generates correction information and an error transmission signal 20 from the ECC control unit 26.
7, a read error reporting unit 27 is provided which receives a notification of the presence or absence of a read error and reports it to the processor 23 by an error report signal 208 when there is no read error.

The magnetic disk control device 20 configured as described above performs the patrol read operation as follows when the command from the host device 10 is not being executed.

The track number storage unit 24 is used for the host device 10.
Information from the control signal 20
When input from the processor 23 by 2, the track number storage unit 24 sends the track number transmission signal 203 to the patrol read control unit 25 to transmit the track number. The patrol read control unit 25 having input the track number sends the read execution signal 2 to the processor 23.
04 is transmitted and the patrol read operation is performed for the track of that number.

Receiving this notification, the processor 23 issues a command to the second data transfer control unit 22 to execute the patrol read operation for the corresponding track, and receives the command, the data transfer control unit 22. 22 operates the magnetic disk device 30 to read (read) information (data) from the corresponding track. The data transfer control unit 22 that receives the data and the ECC from the magnetic disk device 30 sends them to the ECC via the data path 206.
It is transferred to the control unit 26.

The ECC control unit 26 detects whether or not there is a read error (read error) based on the input data and ECC, and when there is a read error, determines whether or not it can be corrected. When it is correctable, correction information is generated and the presence or absence of a read error is reported to the read error reporting unit 27 by the error transmission signal 207.

When the read error reporting unit 27 receives the report from the ECC control unit 26, when there is a read error, the read error reporting unit 27 informs the processor 23 of the fact by the error reporting signal 208.
Report to. Processor 23 that received this report
Reports this to the data transfer control unit 21 by the control signal 200, and the data transfer control unit 21 sends the control signal 1
00 to report this to the upper level device 10.

By the above operation, the magnetic disk device 30
It becomes possible to detect the occurrence of the abnormal state in the early stage.

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

The magnetic disk controller 40 of the embodiment shown in FIG.
1 is the one in which the patrol read control unit 25 of the magnetic disk control device 20 of FIG. 1 is replaced with a patrol off-roll rack marginal read control unit 45. The configuration and operation of the other parts are the same as those of the embodiment of FIG. is there.

The patrol off-roll margin marginal read control unit 45 reduces the read margin due to the axial deviation of the magnetic disk device due to aging, or the read / write characteristics of the magnetic disk device deteriorate and the read margin decreases. In some cases, the patrol off roll rack marginal read operation is performed to deal with them. As a result, when a track in which a normal read operation cannot be performed occurs due to the misalignment of the axis of the magnetic disk device or the deterioration of the read / write characteristics, it becomes possible to detect it early.

FIG. 3 is a block diagram showing a third embodiment of the present invention.

The magnetic disk controller 50 of the embodiment shown in FIG.
1 temporarily holds the data transferred from the data transfer control unit 22 via the read path 209 to the magnetic disk control device 20 of FIG.
A data recovery buffer 58 that corrects data including a read error according to the correction information input by the path 211,
Either the data transferred from the data transfer control unit 22 via the read path 209 or the data transferred from the data recovery buffer 58 via the data recovery path 210 is selected, and the data is transferred via the read path 212. A first read data path control unit 59a to be sent to the transfer control unit 21 and data transferred from the data transfer control unit 21 via the read path 205 or transferred from the data recovery buffer 58 via the data recovery path 210. A second read data path controller 59b for selecting one of the received data and sending it to the data transfer controller 22 via the read path 213 is added. The configuration and operation of the other parts are the same as those of the embodiment shown in FIG.

The magnetic disk controller 50 configured as described above, when detecting a correctable read error during the patrol read operation, processes as follows.

When the data recovery buffer 58 learns from the ECC control unit 26 that a correctable read error is detected via the ECC path 211, the data recovery buffer 58 stores the data already stored according to the correction information input from the ECC control unit 26. Correct the error in.

The first read data path control unit 59a and the second read data path control unit 59b are set so as to select the data transferred from the data recovery buffer 58 via the data recovery path 210, and read. The data path control unit 59a transfers the data to the data transfer control unit 21.
The read data path control unit 59b sends the data to the data transfer control unit 22.

In this way, the occurrence of an abnormal state in the magnetic disk device 30 can be detected at an early stage, and the data in which the read error has occurred can be restored to the correct data without the intervention of software. When the data is read from that track, the read error does not occur.

FIG. 4 is a block diagram showing a fourth embodiment of the present invention.

The magnetic disk controller 60 of the embodiment shown in FIG.
3 causes the magnetic disk controller 50 of FIG. 3 to slightly move the magnetic head of the magnetic disk device 30 inward and outward with respect to the track by the uncorrectable information input from the ECC controller 26. (This operation is referred to as an offset operation), and an offset read control unit 65 for executing the read operation again is added. The structure and operation of the other parts are the same as those of the embodiment shown in FIG.

When the magnetic disk control device 60 having the above-described structure detects an uncorrectable read error during the patrol read operation, it processes as follows.

When the offset read controller 65 knows that an uncorrectable read error has occurred based on the information from the ECC controller 26, it controls the magnetic head of the magnetic disk device 30 to perform an offset operation. It is transmitted to the processor 23 via the path 215. Accordingly, the processor 23 causes the magnetic disk device 30 to read the data by the offset operation via the data transfer control unit 22, and when the read error does not occur, the magnetic disk device 30 has an axis deviation. Then, the offset is released, and the processor 23 rewrites the data to the track.

In this way, the occurrence of an abnormal state in the magnetic disk device 30 can be detected at an early stage, and the data in which the read error caused by the axis deviation has occurred can be restored to the correct data without the intervention of software. After that, when the data is read from the track by the command from the higher-level device 10, the read error does not occur.

FIG. 5 is a block diagram showing a fifth embodiment of the present invention.

The magnetic disk controller 70 of the embodiment shown in FIG.
Enables three magnetic disk devices 31 to 33 to be connected to the second data transfer control unit 22, and replaces the track number storage unit 24 of the magnetic disk control device 60 of FIG.
The three track number storages 74 to 76 and the three track number storages 74 to 76 are monitored by the priority monitoring paths 263a to 263c, and three units connected to the second data transfer control unit 22 are monitored. Magnetic disk unit 31
33 to 33, a patrol read priority control unit 77 is provided to determine which of the magnetic disk devices to execute the patrol read operation and notify the patrol read control unit 25. The configuration and operation of the other parts are the same as in the embodiment of FIG.

When the three magnetic disk devices 31 to 33 are controlled by the magnetic disk control device 60 configured as described above, the track numbers actually written during the write operation are set to the three magnetic disk devices. Devices 31-33
The tracks are stored in the corresponding track number storages 74 to 76, respectively, and the patrol read priority control unit 77 determines the tracks to be accessed during the patrol read operation.

As a result, the occurrence of an abnormal state can be detected early in all three magnetic disk devices.

[0040]

As described above, the magnetic disk control device of the present invention controls the transfer of data between the processor whose operation is controlled by the firmware and the host device by the control of the processor. And EC between the data transfer control unit and the magnetic disk unit
When the second data transfer control unit that controls the transfer of C, the track number storage unit that stores the track number actually written by the processor at the time of the write operation, and the instruction from the host device are not executed A patrol read control unit for transmitting to the processor to execute the patrol read operation for the track of the number transmitted from the track number storage unit, and the first data transfer control unit transferred to the second data transfer control unit. Generate a read error detection and correction code for the data to be read, and detect whether there is a read error in the data transferred from the second data transfer control unit and the read error detection and correction code, and correct if there is a read error. If it is possible to correct it by judging whether it is possible or not, an ECC control unit that generates correction information and a read error in the ECC control unit By providing a read error reporting unit for transmitting this to the processor when it is determined that Tsu, there is an effect that it is possible to find the occurrence of the abnormal condition in the magnetic disk apparatus at an early stage.

[Brief description of drawings]

FIG. 1 is a block diagram showing a first embodiment of the present invention.

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

FIG. 3 is a block diagram showing a third embodiment of the present invention.

FIG. 4 is a block diagram showing a fourth embodiment of the present invention.

FIG. 5 is a block diagram showing a fifth embodiment of the present invention.

[Explanation of symbols]

10 Higher-level device 20/40/50/60/70 Magnetic disk controller 21/71 Data transfer controller 22/72 Data transfer controller 23/73 Processor 24 / 74-76 Track number storage 25 / 75a Patrol read controller 26/76 ECC control unit 27/77 Read error reporting unit 30/31/32/33 Magnetic disk unit 45 Patrol off-roll rack marginal read control unit 58/78 Data recovery buffer 59a / 59b / 79a / 79b Read data path control unit 65 -75b Offset read control unit 77 Patrol read priority control unit 100-200-201-202 Control signal 203 Track number transmission signal 204 Read execution signal 205-206 Data path 207 Error transmission signal 208 Error report signal 09, 212, 213 read path 210 data recovery path 211 ECC path 215 control path 263a~263c priority monitoring path

Claims (5)

[Claims] 1. A processor whose operation is controlled by firmware, a first data transfer control unit which controls transfer of data to and from a host device under the control of the processor, and a magnetic disk under the control of the processor. A second data transfer control unit that controls transfer of data and ECC to and from a device, a track number storage unit that stores the number of the track actually written by the processor during a write operation, and the upper order A patrol read control unit for transmitting to the processor to perform a patrol read operation for the track of the number transmitted from the track number storage unit when not executing an instruction from the device; and the first data. Read error detection and correction for data transferred from the transfer control unit to the second data transfer control unit Detecting whether or not there is a read error in the code generated and the data transferred from the second data transfer control unit and the read error detection and correction code, and if there is a read error, whether it is correctable or not And an ECC control unit for generating correction information when it is correctable, and a read error reporting unit for transmitting the error to the processor when it is determined that there is a read error in the ECC control unit. Characteristic magnetic disk controller. 2. A processor whose operation is controlled by firmware, a first data transfer control unit which controls transfer of data to and from a host device by control of the processor, and a magnetic disk which is controlled by the processor. A second data transfer control unit that controls transfer of data and ECC to and from a device, a track number storage unit that stores the number of the track actually written by the processor during a write operation, and the upper order A patrol off-roll rack marginal read control unit for transmitting to the processor to execute a patrol off-roll rack marginal read operation for the track of the number transmitted from the track number storage unit when not executing an instruction from the device; From the first data transfer control unit to the second data transfer control unit A read error detection and correction code for the data transferred to the second data transfer control unit, and whether or not there is a read error in the data transferred from the second data transfer control unit and the read error detection and correction code is detected. When it is correctable, an ECC control unit that determines whether it is correctable and generates correction information when correctable;
A magnetic disk control device comprising: a read error reporting unit for transmitting the read error to the processor when the control unit determines that there is a read error. 3. A processor whose operation is controlled by firmware, a first data transfer control unit which controls transfer of data to and from a host device under the control of the processor, and a magnetic disk under the control of the processor. A second data transfer control unit that controls transfer of data and ECC to and from a device, a track number storage unit that stores the number of the track actually written by the processor during a write operation, and the upper order A patrol read control unit for transmitting to the processor to perform a patrol read operation for the track of the number transmitted from the track number storage unit when not executing an instruction from the device; and the first data. Read error detection and correction for data transferred from the transfer control unit to the second data transfer control unit Detecting whether or not there is a read error in the code generated and the data transferred from the second data transfer control unit and the read error detection and correction code, and if there is a read error, whether it is correctable or not And an ECC control unit that generates correction information when the error is correctable, a read error reporting unit that notifies the processor when there is a read error in the ECC control unit, and the second A data recovery buffer that temporarily holds the data transferred from the data transfer control unit and corrects the data in accordance with the correction information input from the ECC control unit; and the data transferred from the second data transfer control unit or Data recovery buffer The first read data that selects one of the transferred data and sends it to the first data transfer control unit. And path control unit,
A second one of the data transferred from the first data transfer control unit and the data transferred from the data recovery buffer is selected and transmitted to the second data transfer control unit. And a read data path control unit. 4. A processor whose operation is controlled by firmware, a first data transfer control section which controls transfer of data to and from a host device under the control of the processor, and a magnetic disk under the control of the processor. A second data transfer control unit that controls transfer of data and ECC to and from a device, a track number storage unit that stores the number of the track actually written by the processor during a write operation, and the upper order A patrol read control unit for transmitting to the processor to perform a patrol read operation for the track of the number transmitted from the track number storage unit when not executing an instruction from the device; and the first data. Read error detection and correction for data transferred from the transfer control unit to the second data transfer control unit Detecting whether or not there is a read error in the code generated and the data transferred from the second data transfer control unit and the read error detection and correction code, and if there is a read error, whether it is correctable or not And an ECC control unit that generates correction information when the error is correctable, a read error reporting unit that notifies the processor when there is a read error in the ECC control unit, and the second A data recovery buffer that temporarily holds the data transferred from the data transfer control unit and corrects the data in accordance with the correction information input from the ECC control unit; and the data transferred from the second data transfer control unit or Data recovery buffer The first read data that selects one of the transferred data and sends it to the first data transfer control unit. And path control unit,
A second one of the data transferred from the first data transfer control unit and the data transferred from the data recovery buffer is selected and transmitted to the second data transfer control unit. Of the read data path control unit and the uncorrectable information input from the ECC control unit, the magnetic head of the magnetic disk device is slightly moved to the inner and outer circumference sides with respect to the track, and the read operation is performed again. A magnetic disk control device comprising: an offset read control unit to be executed. 5. A processor whose operation is controlled by firmware, a first data transfer control unit which controls transfer of data to and from a host device under the control of the processor, and a plurality of units under the control of the processor. Second data transfer control unit for controlling transfer of data and ECC to and from the magnetic disk device, and the processor provided for the plurality of magnetic disk devices at the time of writing operation to the corresponding magnetic disk devices. A plurality of track number storage units that store the numbers of the tracks that have actually been written by, and which one of the plurality of magnetic disk devices is monitored by monitoring the plurality of track number storage units. A patrol read priority control unit for deciding whether to execute a patrol read operation for A patrol read control unit that transmits to the processor to execute a patrol read operation for a track of a designated number of the magnetic disk device if determined by the patrol read priority control unit when not executing an instruction from the device. And a read error detection and correction code for the data transferred from the first data transfer control unit to the second data transfer control unit, and the data transferred from the second data transfer control unit and the read error. An ECC control unit that detects whether or not the detection / correction code has a read error, determines whether or not the read error is correctable, and generates correction information when the read error is correctable; When the ECC control unit determines that there is a read error, the read error reporting unit that notifies the processor of the error. A data recovery buffer that temporarily holds the data transferred from the second data transfer control unit and corrects the data in accordance with correction information input from the ECC control unit; and a data recovery buffer transferred from the second data transfer control unit. A first read data path control unit that selects either the incoming data or the data that is transferred to the data recovery buffer and sends the selected data to the first data transfer control unit; and the first data A second read data path control for selecting one of the data transferred from the transfer control unit and the data transferred from the data recovery buffer and sending the selected data to the second data transfer control unit. Section and the uncorrectable information input from the ECC control section, the magnetic head of the magnetic disk device is moved to the inner circumference side of the track. Magnetic disk controller, characterized in that it comprises an offset read control unit for re-execute the read operation slightly moved to the fine outer peripheral side.