JPH0277927A - Control system for semiconductor disk subsystem - Google Patents

Abstract

PURPOSE:To effectively use the information stored in a semiconductor disk device regardless of the presence or absence of the fault of a track control table by adding a decision table to the disk device of all logical tracks. CONSTITUTION:A decision table 22 containing plural entries corresponding to all logical tracks respectively is added to a part of a semiconductor disk device 2. Then a flag is set at each entry of the table 22 in response to the presence or absence of the access failure of the disk 2 against a track control table 21a set on each track 21. Based on this flag, the table 21a is not used hereafter when it is decided that the table 21a and a reading or reconstituting failure. Then a desired record is successively searched at and after the head of a desired track 21 and the prescribed input/output process is continued. Thus the information stored in the disk 2 can be effectively used regardless of the presence or absence of the fault of the table 21a.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a control technique for a semiconductor disk subsystem, and particularly to a technique that is effective when applied to a recovery technique when a failure occurs.

[Conventional technology]

In recent years, due to the decline in the cost of semiconductor memory, the storage function of large-capacity magnetic disk devices can be replaced with semiconductor memory that can be accessed at high speed. A semiconductor disk device that achieves high-speed data input/output without unnecessary waiting time caused by mechanical operations such as waiting for disk rotation is put into practical use as an external storage device for computer systems. It has reached this point.

That is, the semiconductor memory is divided into logical tracks corresponding to tracks in a magnetic disk device, etc., and a track control table is provided in which storage addresses of individual records in each logical track are recorded. By referring to
This system enables direct access to the target record in each logical track and allows high-speed response to input/output requests from higher-level channel devices.

In this case, if the information in the track control table becomes unreadable, conventionally, for example, the former TACM manual published by Hitachi, Ltd. /3 Semiconductor storage device.

)1-6918-1 battery device (Nα8080-2-
As described in 078-30), a sense byte in which failure conditions and the like were recorded was reported to an upper level channel device, indicating an error.

[Problem to be solved by the invention]

However, in the above-mentioned conventional technology, even if the storage state of each record in the logical track corresponding to the track control table that has become unreadable is healthy,
The upper channel device that detects the sense byte regards the entire logical track as defective and terminates input/output processing, resulting in a problem that subsequent access to records within the logical track becomes impossible.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide a control technology for a semiconductor disk subsystem that can effectively utilize information stored in a semiconductor disk device without being affected by the occurrence of a failure in the track control table. It is about providing.

The above and other objects and novel features of the present invention will become apparent from the description of this specification and the accompanying drawings.

[Means to solve the problem]

A brief overview of typical inventions disclosed in this application is as follows.

That is, the control method of the semiconductor disk subsystem according to the present invention corresponds to each track in a disk device, and has a logical track in which predetermined information is stored in a format equivalent to that of the disk device, and a logical track of this logical track. A semiconductor disk device is constructed by building a track control table in a semiconductor memory in which management information is stored, and by referring to the track control table, information in the semiconductor disk device can be controlled in response to input/output requests from a higher-level processing device. In a semiconductor disk subsystem consisting of a semiconductor disk control device that controls access, if the track control table corresponding to the target logical track cannot be referenced, the information is accessed sequentially from the beginning of the logical track, or The track control table is reconstructed based on all the information valid within the logical track to continue input/output operations.

[Effect]

According to the above-mentioned means, even if it becomes impossible to read management information from the track control table, access to information stored in the semiconductor disk device can be continued without immediately notifying the higher-level processing device of the error. By doing so, the information stored in the semiconductor disk device can be effectively utilized without being affected by the presence or absence of a failure in the track control table.

〔Example〕

FIG. 1 is a flow chart showing an example of a control method for a semiconductor disk subsystem according to an embodiment of the present invention, and FIG. 2 is a flow chart showing an example of a control method for a semiconductor disk subsystem according to an embodiment of the present invention. FIG. 2 is a block diagram showing an example.

First, as shown in FIG. 2, the computer system in this embodiment includes a channel device 1 that controls input/output with the outside in place of the central processing unit based on instructions from the central processing unit (not shown). , a semiconductor disk device 2 consisting of a semiconductor memory or the like driven by an uninterruptible power supply, etc., and a semiconductor disk control interposed between the two to control the output of data between the channel device l and the semiconductor disk device 2. It is equipped with a device 3.

A part of the semiconductor disk device 2 constituted by a semiconductor memory is divided into a plurality of logical tracks 21 corresponding to the tracks in a normal magnetic disk device (not shown), and each of the logical tracks 21 includes the It is configured so that predetermined information is stored in a format completely equivalent to that of a magnetic disk device.

In the case of this embodiment, as shown in FIG. 3, a plurality of records R8 to R1 are stored in each logical track 21 according to the so-called CKD method, which is well known in the logical track control table magnetic disk device. .

That is, each logical track 21 records a cylinder number to which the logical track 21 belongs, a head number indicating which head of the cylinder the logical track 21 corresponds to, error correction information, etc. A home address section HA, a plurality of subsequent records RO-Ro, and a gap G in which predetermined control information and the like are recorded are provided.

In addition, each record R, 1 has management information such as the cylinder number, head number, record number, key length that indicates the length of the data section, and data length in the key part described later. It consists of a count section C, a key section in which index information for a predetermined key search for a record, etc. is recorded, and a data section consisting of normal data.

Furthermore, in the case of the semiconductor disk device 2 as in this embodiment, the head side of the home address section HA of each logical track 21 contains the storage location address of each record R1 in the logical track 2'1. A track control table 21a in which the following information is recorded is arranged.

The semiconductor disk control device 3 includes a buffer 31.
is provided, and when the upper channel device 1 requests access to an arbitrary record R7, the semiconductor disk control device 3 reads the contents of the track control table 21a into this buffer 31 and accesses the target record R7.
By directly accessing the record R,l in a predetermined logical track 21 based on the address of 0, higher speed access is realized in conjunction with the high speed of the semiconductor memory itself.

In this case, a part of the semiconductor disk device 2 includes a plurality of IJs corresponding to each of all the logical tracks 21 provided in the semiconductor disk device 2, as shown in FIG.
A determination table 22 is provided in which each entry contains a determination table 22 that determines whether or not there is a failure in the access of the semiconductor disk controller 2 to the track control table 21a provided for each of the individual logical tracks 21. The configuration is such that the flag 22a that is set is recorded.

In the case of this embodiment, if it is determined by this flag 22a that there has been a failure in reading and rebuilding the track control table 21a as described below, from now on, the track control table 21a (hereinafter referred to as unused mode), the target record is sequentially searched from the beginning of the target logical track 21 and predetermined input/output processing is continued.

Furthermore, FIG. 5 shows a series of commands issued from the channel device 1 to the semiconductor disk control device 3 when outputting predetermined data, a channel command word (hereinafter simply referred to as a command) constituting a chain 4. This is an example of a combination.

That is, the 5EEK command 41 has cylinder number A, head number B, etc. as parameters, and instructs positioning to the target logical track 21 specified by the cylinder number A and head number B.

In addition, the 5EARCHID EQUAL command 42 has cylinder number A, head number B, and record number C as parameters, and selects the target record R, whose cylinder number, head number, and record number recorded in count section C match. Command the action to find out.

Generally speaking, the TIC command 43 performs a conditional branch within the command chain 4, and in the case of this embodiment, the TIC command 43 executes a conditional branch within the command chain 4.
This is repeated until the L command 42 is successful.

Further, the WRITE D command 44 instructs writing of new data into the data portion of the target record R1 indicated by the record number C.

The operation of this embodiment will be explained below.

First, when a predetermined input/output command is given to the channel device 1 from a central processing unit (not shown), the channel device 1
starts reading out a series of command chains 4 shown in the above-mentioned figure 5 stored in a part of the main memory (not shown) shared with the central processing unit, and sequentially sends the read commands to the semiconductor disk controller 3. Issued on.

Here, in the conventional case, the semiconductor disk control device 2 that receives the 5EEK command 41 and the <5EARCHID EQUAL command 42 subsequently selects the corresponding logic based on information such as the cylinder number A and the head B. The control information stored in the track control table 21a of the track 21 is read into the buffer 31, and the address in the logical cylinder 21 of the target record Rn that matches the record number C is specified, and the following W
Prepare for write processing using the RrTE D command 44.

If the semiconductor disk control device 3 fails to read the track control table 21H of the logical track 21 that is the target of the semiconductor disk device 2, the read operation is repeated a predetermined number of times, and if the read operation still fails, a failure occurs. A sense byte indicating that the track control table 21a
The input/output processing for the logical track 21 was terminated, and the records of the logical track 21 could not be accessed thereafter.

Therefore, in the case of this embodiment, the above-mentioned inconvenience is avoided by performing a series of control operations as shown in FIG.

First, prior to the read operation of the track control table 21a, the flag 22a of the determination table 22 is read (step 501), and it is determined whether or not the track control table 21H can be read, that is, whether or not it is in an unused mode (step 502). .

Here, if the flag 22a corresponding to the track control table 21a is in the unused mode, the count section C of each record is sequentially counted from the head of the logical track 21 without using the track control table 21a. , and performs an operation to search for the target record R1, and performs a predetermined update process using the WRITE D command 44 on the target record R6 (step 515).

If the mode is not in use, the contents of the target track control table 21H are read into the buffer 31 (5
03).

At this time, it is determined whether or not the read operation has been performed normally (step 504). If the read operation has been performed normally, the individual Using the record storage address etc., execute the WRITE D command 44.
A predetermined update process is executed (step 514).

On the other hand, if an abnormality occurs in the readout operation of the track control table 21a, the number of retries for the readout operation is set (step 505), and a try counter (not shown) is incremented. It is determined whether or not the set value has been exceeded (step 506), and the steps 503 to 3 are repeated a predetermined set number of times.
Repeat operation 505.

If the read operation is not successful even after repeating the above read operation a predetermined number of times, or in the case of a tryout, so-called surface analysis is performed on all records of the logical track 21 (step 507).

In other words, when a failure occurs in a storage medium such as a normal magnetic disk, surface analysis is used to record data on the storage medium by reading the data in each byte in the track, unlike normal access. This is an operation for checking the status, and the semiconductor disk device 2 of this embodiment also performs an operation equivalent to this.

Then, based on the result of this surface analysis, the track control table 21a is edited again (step 508), and the editing result is written at the beginning of the logical track 21 (step 509).

At this time, it is determined whether the write operation was performed normally (step 510), and if it is normal, step 514)
Branch to WRITE D for the desired record.
A predetermined update process using the command 44 is executed.

On the other hand, if it is determined that there is an abnormality in the write operation, a retry counter (not shown) in which the number of retries of the write operation is set is incremented (step 51).
1) Further, it is determined whether the value of the try counter exceeds a predetermined set value or not (step 512), and the operations of steps 509 to 511 are repeated a predetermined set number of times.

If the write operation is not successful even after repeating the above write operation a predetermined number of times, or in the case of a tryout, the cylinder number m and head (track) number corresponding to the logical track 21 are determined in the determination table 22. The flag 22a corresponding to n is set to an unused mode indicating that the track control table 21a is unusable.

Thereafter, in step 515, without using the track control table 21a, the count section C of each record is sequentially checked from the head side of the logical track 21 to search for the target record R7, and the target record R7 is searched for. A predetermined update process using the WRITE D command 44 is executed on the record R11.

As described above, according to the control operation of this embodiment, even if a failure occurs in the read operation of the track control table 21H provided for each of the individual logical tracks 21 in the semiconductor disk device 2, the control operation as in the conventional method can be performed. It becomes possible to continue accessing the record stored in the logical track 21 without immediately notifying the upper channel device 1 of the error using the sense byte and terminating the input/output operation.

As a result, all records stored in the semiconductor disk device 2 can be effectively utilized without being affected by the occurrence of a failure in the track control table 21H.

Although the invention made by the present inventor has been specifically explained above based on Examples, it goes without saying that the present invention is not limited to the Examples and can be modified in various ways without departing from the gist thereof. Nor.

〔Effect of the invention〕

Among the inventions disclosed in this application, the effects obtained by typical inventions are briefly described below.

That is, a logical track that corresponds to each track of a disk device and stores predetermined information in a format equivalent to that of the disk device, and a track control table that stores management information of this logical track are provided on a semiconductor device. a semiconductor disk device configured in a memory; and a semiconductor disk control device that controls access to the information in the semiconductor disk device in response to an input/output request from a higher-level processing device by referring to the track control table. If the track control table corresponding to the target logical track cannot be referenced in a semiconductor disk subsystem consisting of By reconstructing the track control table based on all the valid information in the track control table, input/output operations can be continued, so even if it becomes impossible to read management information from the track control table, for example. By continuing to access the target information without immediately notifying the higher-level processing unit of the error, the information stored in the semiconductor disk device is made valid regardless of whether or not a failure occurs in the track control table. It can be used for

[Brief explanation of the drawing]

FIG. 1 is a flowchart showing an example of a control method for a semiconductor disk subsystem according to an embodiment of the present invention, and FIG. 2 shows an example of a computer system constructed using the semiconductor disk subsystem according to this embodiment. Block diagram; Figure 3 is a conceptual diagram showing the recording format of information on each logical track in a semiconductor disk device; Figure 4 is a conceptual diagram showing an example of the configuration of a determination table; Figure 5 is the configuration of a command chain. It is an explanatory view showing an example. DESCRIPTION OF SYMBOLS 1... Channel device (upper processing device), 2... Semiconductor disk device, 21... Logical track, 21a.
... Track control table, 22... Judgment table,
22a...Flag, 3...Semiconductor disk control device, 31...Buffer, 4...Command chain, 5
01-515...Steps showing an example of control operation.

Claims (1)

[Claims] 1. A logical track that corresponds to each track in a disk device and stores predetermined information in a format equivalent to that of the disk device, and a track that stores management information for this logical track. a semiconductor disk device configured with a control table constructed in a semiconductor memory; and a semiconductor disk device that controls access to the information in the semiconductor disk device in response to an input/output request from a higher-level processing device by referring to the track control table. In a semiconductor disk subsystem consisting of a semiconductor disk control device, if the track control table corresponding to the target logical track cannot be referenced, the information is accessed sequentially from the beginning of the logical track, or A control method for a semiconductor disk subsystem, characterized in that input/output operations are continued by reconstructing the track control table based on all the information valid within a logical track. 2. A determination table is provided to record success or failure in the reconstruction of each of the track control tables, and if the determination table determines that the previous reconstruction was a failure, the track control table 2. A control method for a semiconductor disk subsystem according to claim 1, wherein the information is accessed sequentially from the beginning of the logical track without using the logical track.