JPH04102118A - Buffer flash mechanism - Google Patents

Abstract

PURPOSE:To reduce the danger of erasing data by providing a means to retrieve a buffer, which is allocated to a file to be closed and stores data not reflected on an auxiliary storage device yet, from a table and to output the data of all the buffers like this to the auxiliary storage device. CONSTITUTION:A managing means 2 manages the allocating condition of buffers and the reflecting condition showing whether the data stored in the respective buffers are outputted to an auxiliary storage device 8 or not for each file according to management information on a table 10 and when a file close request is outputted since a user process completes the use of the file, a file close time buffer flash means retrieves the buffer, which is allocated to the file to be closed and stores data not reflected on the auxiliary storage device yet, from the table 10 and outputs the data of all the buffers like this to the auxiliary storage device 8. Thus, the danger of erasing data is reduced.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a file management system for an information processing device,
In particular, in a file management system that divides the data storage area in an auxiliary storage device into blocks of a fixed size as the unit of file area allocation and input/output, and performs delayed writing using a buffer for each block.
The present invention relates to a buffer flush mechanism that outputs unoutput data in a buffer to an auxiliary storage device.

[Conventional technology]

In a file management system that divides the data storage area in an auxiliary storage device such as a magnetic disk device into blocks of a fixed size as the unit of file area allocation and input/output unit, and uses a buffer for each block, data is output. In this case, the process does not output to the auxiliary storage device at that point, but notifies the user process of the end of output processing when the data is written to the buffer, and actually outputs the data in the buffer to the auxiliary storage device. Delayed writing is performed to speed up data output processing to the user process, with the intention of performing it asynchronously with the user process at a later time.

By performing such delayed writing, it is possible to speed up the output processing, but on the other hand, there is a situation where the data requested by the user process is not output to the auxiliary storage device and exists only in the buffer. It can occur.

If a failure such as a system shutdown occurs under such circumstances, data that would have been written by the user process will be lost.

Therefore, in conventional file management systems of this type, in addition to the times when the buffer is basically needed, such as when the buffer is allocated to another block and its contents are rewritten, or when the system is terminated, etc. The configuration is such that even when a user issues a request to write data on the buffer, the data on the buffer can be output to the auxiliary storage device, that is, the buffer can be flushed.

[Problem to be solved by the invention]

According to the above configuration, the buffer can be flushed upon request from the user, so that the user can properly issue a request to write data on the buffer, and the data requested to be output by the user process will be flushed on the buffer. This makes it possible to shorten as much as possible the period during which a situation occurs that only exists. However, this requires a large burden on the user as it is necessary to issue data write requests appropriately, and if such a request is forgotten, data may be left out of the buffer for a long time after the user process finishes outputting. A situation may occur where data is not being output to the auxiliary storage device, increasing the risk of data loss.

The present invention was made in view of the above circumstances, and its purpose is to automatically delete only those buffers that store data of files that are no longer in use and for which no data has been written to the auxiliary storage device. The objective is to provide a buffer flush mechanism that detects and flushes the buffer.

[Means to solve the problem]

In order to achieve the above object, the present invention divides a data storage area in an auxiliary storage device into blocks of a fixed size to serve as a unit of file area allocation and a unit of input/output, and uses a buffer for each block to delay In an information processing device that speeds up input/output by writing data, the status of buffer allocation and the status of reflection of data stored in each buffer on the auxiliary storage device are determined for each file using management information on a table. A management means for managing each file, and a management means that searches the table for buffers that are allocated to files to be closed and that store data that has not yet been reflected in the auxiliary storage device, and retrieves the data of all those buffers. and buffer flush means for outputting the file to the auxiliary storage device.

[Effect]

In the buffer flush mechanism of the present invention, the data storage area in the auxiliary storage device is divided into blocks of a fixed size, which are used as the unit of file area allocation and the unit of input/output, and delayed writing is performed using a buffer in block units. When the data is stored in the auxiliary storage device, the management means reflects the buffer allocation status, which buffer is allocated to each file, and whether the data stored in each buffer has been output to the auxiliary storage device or has not been output yet. The status is managed for each file using management information on a table, and when a user process issues a file close request after finishing using the file, when the file is closed, the faranonyu means handles the closing process. A buffer that is allocated to a file that should be saved and stores data that has not yet been reflected in the auxiliary storage device (
7 buffers are searched from the table and the data of all those buffers is output to the auxiliary storage device.

(Example) Next, an example of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of the present invention. In the figure, a user interface 1 is an interface with a user process (not shown), and a file input/output management means 2 that manages file input/output according to requests from the user process is connected to this user interface 1. , buffer allocation means 3 and file data chain creation means 4 are sequentially connected to this file input/output management means 2. Further, a file closing means 5 for performing file closing processing is connected to the user ink face 1, a buffer management table search means 6 is connected to the file closing means 5, and the buffer management table search means 6 and the file input/output management means A buffer data input/output means 7 is connected to 2. The auxiliary storage device 8 connected to this buffer data input/output means 7 is a magnetic disk device or the like, and the data storage area is divided into blocks of a fixed size, which are used as the unit of file area allocation and the unit of input/output. Furthermore, two tables are provided, namely a file management table 9 and a buffer management table 10, which are referred to and updated by each means.

FIG. 2 shows an example of the contents of the file management table 9. The file management table 9 is a table that stores various management information regarding files used by users, and is composed of a plurality of entries. Each entry has a file identification number fnl~ assigned to each file.
fnn and various file management information are stored,
Buffer management table entry addresses ffl to ff
n, fbl to fbn are stored. In this embodiment, which buffer b< is currently allocated to the file being used is managed by a bidirectional chain (forward chain and backward chain) for each file, and the file management table 9 One address f of each entry
The addresses f1 to ffn are the starting points of the forward chain, and the other addresses fb1 to fbn are the starting points of the backward chain.

FIG. 3 shows an example of the contents of the buffer management table 10. The buffer management table 10 is a table that stores various management information related to buffers allocated to files, and is composed of a plurality of entries in one-to-one correspondence with the buffers. has been done. Each entry includes the block number bnl-bnm in which the buffer stores, and the data storage area address dtl-dtm, which is a pointer to the data storage area.
.

Buffer management table entry address bfl~b
f m, b b 1 to b bm, flag f indicates whether or not the data in 7a has been reflected in the auxiliary storage device 8.
1gl to f1gm and other buffer management information are stored. Here, the addresses bf1 to bfm of one buffer management table entry indicate the next entry in the forward chain, and the addresses bbl to bbm of the other buffer management table entry indicate the next entry in the backward chain. This is the address to be used.

Next, the operation of this embodiment configured as described above will be explained.

In FIG. 1, when a file data input/output request is made by a user process (not shown) through the user interface 1, the file input/output management means 2 divides the input/output request into blocks, and divides the input/output request into blocks. In order to secure the buffer to be allocated to the block number and the file to be allocated, file management table 9
The buffer allocation means 3 is called by specifying the address of the entry (the entry in which the file identification number of the file is stored).

The buffer allocation means 3 uses the buffer management table 10
Other buffer management information in the file is checked to search for a usable buffer, and one buffer is reserved for the block requested by the file input/output management means 2. That is,
If an empty buffer is found, the block number specified by the file input/output management means 2 is stored in the block number column of the entry in the buffer management table 10 corresponding to the buffer, and information such as usage is stored in other buffer management information. Store. In addition, if there is no free buffer, for example, the contents of the buffer storing the oldest data are output to the auxiliary storage device 8 to secure a free buffer, and predetermined information is added to the entry in the buffer management table 10 corresponding to the buffer. Zero-order buffer allocation means 3 for storing
calls the file data chain creation means 4 in order to chain the secured buffer from the specified entry in the file management table 9.

The file data chain creation means 4 connects the buffer allocated by the buffer allocation means 3 to the chain starting from the specified entry of the file management table 9 by updating the file management table 9 and the buffer management table 10. . For example, there is one buffer containing the data of the file with file identification number fnl.
block number b of the file that never exists
If a buffer is allocated for block n2,
The file data chain creation means 4 includes a file management table 9 in which file identification numbers fnl shown in FIG. 2 are stored.
The address of the buffer management table entry in the entry ff1. The address of the entry in the buffer management table 10 of FIG. 3 in which block number bn2 is stored is stored in bf2.fbl of the entry. bb2 stores the address of the entry in the file management table 9 having the file identification number fnl.

When the processes of the buffer allocation means 3 and file data chain creation means 4 described above are completed, the file input/output management means 2 performs input/output using the buffer allocated by the buffer allocation means 3. When performing input/output with the auxiliary storage device 8, the buffer data input/output means 7 is called, and input/output is performed between the allocated buffer and the auxiliary storage device 8. However, in the case of delayed writing, Only the process of writing data to the buffer is performed, and a flag indicating that data has not yet been output to the auxiliary storage device 8 is set to block number bn2 in FIG.
The flag f1g2 is set for the entry storing , and the write process ends. Note that when the data in the buffer is output to the auxiliary storage device 8 after delayed writing, the flag fIg2 is changed to indicate that the data has been output.

Next, after the input/output to the block with block number bn2 is completed, if the file input/output management means 2 requests the buffer allocation means 3 to allocate a buffer to the block with block number bn3, for example, then block number b
A buffer is secured in the same way as when a buffer is secured in block n2, and the block number b is entered in the entry corresponding to the secured buffer in the buffer management table 10.
At the same time as n3 is written, other necessary information for buffer management information is written. Furthermore, when performing delayed writing, a flag indicating that buffer data has not been output is set in the flag f1g3 of the same entry.

Furthermore, the file data chain creation means 4 stores the address of the entry in the buffer management table 10 having the block number bn3 in ffl of the entry with the file identification number fnl in the file management table 9, and the address of the entry in the buffer management table 10 having the block number bn3 is stored in bf3 of the entry. The address of the entry in the buffer management table 10 with block number bn2 is stored in bb3, the address of the entry in the file management table 9 with the file identification number fn1 is stored in bb3, and the address of the entry in the buffer management table 10 with block number bn2 is stored in bb3. The address of the entry in the buffer management table 10 having the block number bn3 is stored in bb2.

As a result of the above processing, the entry of the file management table 9 with the file identification number fnl and the block number bn2. b
A bidirectional chain with the entry of the buffer management table 10 of n3 is created. FIG. 4 shows the state of the chain of the file management table 9 and buffer management table 10 at this time.

After that, when a file close request is issued from the user process through the user interface 1 to close the file with the file identification number fnl,
File closing means 5 is called.

The file closing means 5 calls the buffer management table retrieval means 6 during the file usage termination process.

The buffer management table search means 6 refers to the address ffl of the buffer management table entry in the file management table 9 entry having the file identification number fnl, and finds the entry in the buffer management table 10 that it points to, that is, the block with the program number bn3. Flag f1g in the entry that manages the buffer that stores
Determine the contents of 3. If the flag f1g3 indicates that the contents of the buffer have not yet been output, an output request for the buffer managed by this entry is sent to the buffer data input/output means 7, and the buffer data input/output means 7 outputs the data in the buffer to the corresponding block in the auxiliary storage device 8. The buffer management table search means 6 determines whether the output of the data of block number bn3 has been completed or not.
Alternatively, if the flag f1g3 indicates that the output has been completed, refer to the buffer management table IO entry (the entry with the block number bn2) pointed to by bf3 of the entry with the block number bn3, and select the flag f1g2.
After that, repeat the same process as above, and then
The buffer management table search means 6 refers to bf2,
An attempt is made to obtain the next entry, but bf2 points to the entry in file management table 9, so when the file is closed by 0 or more processes that end the search for buffer management table entries, The contents of all buffers that have been used and have not yet been flushed are written to the auxiliary storage device 8. Therefore, even if a system down occurs after a file is closed, file data is guaranteed. [Effect of the Invention] As explained above, according to the buffer flush mechanism of the present invention, when a file close request is issued, Buffers that have been allocated to files and that store data that has not yet been reflected in the auxiliary storage device will be automatically flushed and all buffer data will be output to the auxiliary storage device. , it becomes possible to reliably flush only the buffers having data of the file whose data has not been written to the auxiliary storage device when the use of the file ends, without imposing a burden on the user. Therefore, the risk of data loss can be reduced and the reliability of file data can be increased.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of an embodiment of the present invention, Fig. 2 is a diagram showing an example of the contents of a file management table, Fig. 3 is a diagram showing an example of the contents of a buffer management table, and Fig. 4 is a diagram showing an example of the contents of a file management table. FIG. 4 is a diagram showing an example of a buffer management chain created between the buffer management table and the buffer management table. In the figure, 1...User ink face 2...File input/output management means 3...Buffer allocation means 4...File data engine creation means 5...File closing means 6...Buffer 1 Management table search means 7... Buffer 1 Data input/output means 8... Auxiliary storage device 9... File management table 10... Buffer management table

Claims (2)

[Claims] (1) The data storage area in the auxiliary storage device is divided into blocks of a fixed size, which are used as the unit of file area allocation and the unit of input/output, and a buffer is used for each block to perform delayed writing, thereby increasing the speed of input/output. In an information processing device that is designed to be configurable, there is provided a management means for managing the buffer allocation status and the reflection status of data stored in each buffer in the auxiliary storage device for each file using management information on a table; A file that searches the table for buffers that are allocated to the file to be processed and stores data that has not yet been reflected in the auxiliary storage device, and outputs the data of all those buffers to the auxiliary storage device. A buffer flush mechanism characterized by comprising: buffer flush means when closed. (2) The management means chains an entry corresponding to a buffer allocated to the same file in a buffer management table having an entry corresponding to each buffer and an entry corresponding to a file in the file management table, and 2. The buffer flush mechanism according to claim 1, further comprising a configuration for managing whether or not data stored in each buffer is reflected in the auxiliary storage device based on a flag stored in an entry.