JPS5940309A - Memory controller - Google Patents

Abstract

PURPOSE:To realize the high-speed reading and transfer regardless of the quality of media, by giving an integrated retrial process successively to the codes having errors for each designated read data length. CONSTITUTION:When a read instruction is applied from a CPU, etc. of a main system, a memory controller 2 sets the data length L to be read, the write address to the 2nd memory 3, etc. and then transmits a read command to a magnetic disk device 1. With this command data #A-#D are read out successively from the device 1 for each record, and the read error information is set to a control table 4 for each record in reponse to the error check codes added to the records #A-#D. The read data are written successively to the memory 3 regardless of the presence or absence of a read error. When the transfer is over with all data of designated length L, the controller 2 checks the table 4 and decides to complete the operation or to start a retry operation in accordance with the contents of the table 4.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a memory controller that reads stored data from a first memory such as a magnetic disk and transfers it to a second memory.

[Technical background of the invention and its problems]

When recording data on a magnetic disk, etc., Fig. 1(a)
As shown in the figure, even if the data writing position on the medium is on the same track T, each record #A to
It is generally known that the position shifts every #D. For example, in the case of Figure 1, record #A is normal and record #B
and #D are shifted upward in the drawing, and #C is shifted downward. Such a data position shift causes a read error in the read operation of the magnetic disk.

Therefore, conventional magnetic disk drives are provided with an offset function to enable complete reading even when there is such a positional shift. The offset function is a function that allows the position of the reading head to be slightly moved in the left and right directions (up and down in the figure) of the track 1. This offset function is used in conjunction with a read attempt (hereinafter referred to as retry) operation for a record in which an error has occurred.

In other words, conventionally, as shown in Fig. 1(b), if L is taken sequentially from the record position to #-D, first, the 6-fuko removal for record #A is normal, so the next #Read B. If an error occurs for #B, the head is temporarily returned to the position of the head of #B, the head is moved to the left of track 1 (hereinafter referred to as a direction), and a retry is made. As a result of the retry, if there is no error, the spatula 1' is returned to the reference position 4N, and the next record #C is advanced to read. If an error occurs regarding #C, the head is temporarily returned to the 11th position at the head of #C, and the head is slightly moved in the direction (■) to perform an IJ) lie. the result,
If another error occurs, move the head to track 1.
Make a slight movement to the right (hereinafter referred to as the e direction) and retry. Thereafter, retry and offset operations are performed in the same manner until the specified read data length (#A to #D in this example) is reached.

As described above, in conventional memory controllers, when data in a certain record cannot be read normally, retry and offset operations are repeated until reading becomes normal at that record position, so to speak, retrying is performed on a record-by-record basis.

The disadvantage of this method is that each offset operation results in wasted time equal to the disk rotation waiting time.
The speed will be slower. This will depend on the quality of the media, and will be more noticeable on low quality media with a lot of misalignment.

[Purpose of the invention]

Therefore, the present invention solves the above-mentioned conventional drawbacks and 1. It is an object of the present invention to provide a memory controller that enables high-speed reading and transfer regardless of the quality of media.

[Summary of the invention]

In order to achieve the above object, a memory controller according to the present invention first reads all data existing within a specified read data length from a first memory in record units and writes it to a second memory.

Next, for each record read, it is detected whether there is a reading error or not, and the internal memory (for example,
Read error information is set in a management table of RAM, etc., in correspondence with each record.

For example, if it is normal, the logic is ゝも1'', if there is an error, it is ゝゝ0''.
Set it like this.

Next, based on the error information in the management table above, the first
The retry data for the record in error is written again to the second memory to update its contents. At this time, the presence or absence of a reading error for the retried record is detected, and the error information is set in the management table again to update its contents.

The above operations are repeated in the same way until there is no more error information in the management table (until there are no more error records). Instead of retrying each record and proceeding to the next record when the reading of that record becomes normal, it uniformly retries records that have errors in sequence based on the specified read data length. It is something to be processed. Therefore, in the case of the present invention, when focusing on the movement of the reading head, the head does not perform a retry operation in record units, but moves in units of data length corresponding to all error records.

〔Effect of the invention〕

According to the configuration of the present invention, the reading time can be increased regardless of the quality of the storage medium (first memory) to be read. That is, the reading operation is not interrupted and retried for each record as in the prior art, and the waiting time can therefore be reduced.

[Embodiments of the invention]

Hereinafter, the present invention will be described in detail based on illustrated embodiments.

FIG. 2 shows the relationship between the memory controller according to the present invention, the first memory, and the second memory. In FIG. 2, the memory controller 2 is a magnetic disk device f which is a first memory.
Read data length L specified from iifl (Figure 1 (a)
)) for each record (or sector) (Fig. 1(a) 1. #A, #B.

4LC, 4#D), transferred to the second memory 3, and written. The memory controller 2 is provided with a management team 4 using RAM etc., and each record #A to #
Each reading error information is set corresponding to D. In this embodiment, the read error information is set as a logic signal ゝも1'' when the reading is normal, and as ``j□II'' when there is an error. Frj?, Judgment whether the reading is normal or not. is performed by the memory controller 2 based on the error check code added to each record #A to #D when the stored data is removed from the magnetic disk device 1.

Next, the operation will be explained. When a read command is given to the memory controller 2 from, for example, the CPU of the main system, the memory controller 2 performs necessary setup such as setting the length of data to be read and the write address to the second memory 3, and writes the data to the magnetic disk. Send a read command to device 1.

With this command, data 4#A to #D are sequentially read out from the magnetic disk device 1 in record units. The position of the reading head during this reading is the reference position N as shown in FIG. 1(C). Since an error check code is added to each record #A to #D, reading error information is set in the management table 4 for each record based on the error check code. In the case of Figure 1 (a), the only record that could be read normally is #A. Therefore, as shown in Figure 3 (a), the contents of the management table 4 are only #A as "1", and # and B. -4pD are all “
0" is set. On the other hand, each read data is sequentially written to the second memory 3 regardless of the presence or absence of a read error. When the transfer operation of all the data of the specified length L is completed, the memory controller 2 inspects the management table 4,
Depending on the content, it is determined whether the operation should be ended or whether the IJ) lie operation should be started.

As a result of inspection of management table 4, all contents are ゝゝ1''
In the case of , it means that all data has been read and transferred normally, so the read operation is ended.

However, as described above, in this embodiment, all records except record #A are errors, so the memory controller 2 enters a retry operation.

IJ) The lie operation is performed together with the offset operation described above. That is, as shown in FIG. 1(C), the reading head is returned to the position of record 4 #B, and is moved slightly in the direction ① to read records #B to #D again. An error check code is also added to the data resulting from this retry, and the reading error information for records #B to #4 #D is set again in the management table 4 and its contents are updated. At the same time, the data resulting from the retry is also rewritten to the second memory 3, and the contents are updated (the third
Figure (b)). In this first retry, records #B and #D were read normally, so the content of the management table 4 is 0'' only for record #C.

Therefore, the memory controller 2 enters a retry operation.

That is, as shown in FIG. 1(C), the reading head is returned to the position of record #C, and the reading operation is performed while being slightly moved in the e direction. As a result of checking the error check code added to the data by this retry, if the data has been read normally, the management table 4
``1'' is set in the area of record #, C, and the retry data is also rewritten to the second memory (Figure 3 (C)). This means that the data has been read successfully.

In this way, the retry operation is not performed for each record, but by performing it for the specified data length. Whereas it would be necessary to wait for the rotation of the magnetic disk four times, according to the present invention (Figure (C)), it is only necessary to wait for two times, which significantly shortens the time it takes to obtain normal data. .

In the above explanation, to retry, the reading head was returned to record #B, which had a smaller J ridge among the error records, and the retry was performed from record #B. However, other IJ l-lie methods are as follows. There are several possible ways.

One is that regardless of the position of the error record, it always starts from the first record of the specified read data length (for example, 4#A
~OD) Retry at t and write only the 1'1'' record as valid data based on the contents of the data t%r-filled table 4 to be sent to the second memory 3, and control to prohibit the others. Another method is to read only records in which 110'' is set in the management table 4. In this case, the memory controller 2 may set the IT disk address and the write address to the second memory 3.

[Brief explanation of the drawing]

FIG. 1 is an explanatory diagram showing the relationship between record misalignment and offset operation, where (a) shows the record misalignment, and (b)
2(C) shows the offset operation of the reading head using the conventional retry method, and FIG. 2(C) shows the offset operation of the reading head using the retry method according to the present invention. FIGS. , is a block diagram showing the relationship with a second memory. 1... first memory (magnetic disk device), 2...
memory controller, 3... second memory, 4.
...Management table.

Claims (1)

[Scope of Claims] A memory controller that reads stored data from a first memory and transfers it to a second memory, the controller reads all data within a specified read data length record by record and transfers the stored data to the second memory. Detects the presence or absence of a read error for each record, sets the error information in an internal management team memory, and reads the first memory based on the error information in the management table. By retrying, the retry data for the record with sheller is updated and recorded in the second memory, and the presence or absence of a read error for the retried record is detected, and the error information is stored in the management. A memory controller characterized by updating and recording in a table and repeating the above operation until all error records are eliminated.