JP2006127377A - Fat file system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method for operating a buffer cache capable of improving disk reading/writing speed in a FAT file system. <P>SOLUTION: The FAT file system comprises a file operating means for managing a file descriptor and operating a file pointer, etc.; a directory entry operating means for analyzing directory entry information of each file and operating the editing of the constitution; a FAT operating means for acquiring a cluster address with data of a position to be designated in the file described therein or correctly forming the cluster chain of the file, so as to perform a writing processing in a disk, etc.; and a buffer cache operating means for managing the plurality of buffer caches which are divided by sector size, returning the buffer cache including data of a designated sector, or selecting unused buffer cache so as to read/write data to the buffer cache. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a FAT file system for handling a file group recorded on a FAT type disc. In particular, the present invention relates to a buffer cache operating method for speeding up access to data on a FAT disk.

(FAT file system)
A file system that manages files using a file allocation table (FAT) is generally called a FAT file system. A disk medium formatted on the assumption that it is used in the FAT file system is generally called a FAT type disk.

  FIG. 5 is based on the assumption that it is mainly used in MS-DOS (a disk-based operating system of Microsoft Corporation, a registered trademark of Microsoft Corporation), which is an operating system (OS) provided by Microsoft Corporation. This shows an example of the recording format of the FAT type disc.

  A medium called a “disk” is not limited to a FAT type disk, and is divided into a plurality of blocks in units called sectors, and data recorded on the disk is read or written in units of sectors.

  Reference numeral 501 denotes the first sector (logical sector number 0), which is generally called a boot sector. In the boot sector (501), a boot program for loading an operating system, the number of sectors included in one cluster (to be described later) (hereinafter referred to as "sector / cluster number"), the number of root directory entry items, Information such as the number of sectors used in a FAT area to be described later is recorded.

  Reference numeral 502 denotes a FAT area, which is an area for recording the allocation (allocation) status of clusters to files (details will be described later). In a FAT type disk that is currently widely used, 12 bits, 16 bits, or 32 bits are generally used to represent one cluster of FAT. The FAT area (502) is duplicated as shown in FIG. 5 (FAT1 and FAT2 in FIG. 5), and is continuously arranged on the disk. That is, the same contents are always recorded in FAT1 and FAT2.

  Reference numeral 503 denotes a root directory entry described later.

  Reference numeral 504 denotes a data area in which information on the actual file contents (data body) and directories belonging to the root directory and below is recorded.

  The data area (504) is managed in units called clusters. A cluster has a plurality of consecutive sectors as one unit, and one cluster is composed of a number of sectors represented by powers of 2 (1, 2, 4, 8, 16,...). Hereinafter, in order to simplify the description, it is assumed that one cluster is composed of one sector.

  FIG. 6 is a diagram in which one item of the directory entry in the root directory entry (503) of FIG. 5 is extracted and the configuration is described in detail. One item of the root directory entry (503) is composed of 32 bytes, and each item includes the name, extension, attribute, and date of last update of a file or directory (subdirectory) existing on the root directory of the FAT type disk. , The size of the file (the size of the file expressed in bytes), and the cluster group in which the actual file contents (data body) corresponding to the file described on the directory entry in the data area to be described later are recorded Among them, information such as a cluster number (hereinafter referred to as a start cluster number) in which data corresponding to the head of the file is recorded is recorded.

  FIG. 7 conceptually shows the contents recorded in the FAT area (502) of FIG. Each item (702) in the FAT area (502) has a one-to-one correspondence with the cluster address (703) in the data area (504). Each item (702) in the FAT area (502) is used to record the cluster group arrangement order (cluster chain) so that the cluster groups constituting the file can be read in the correct order when the file is read. . That is, the contents of each item (702) in the FAT area (502) are sequentially recorded with the value of the cluster address to be read next, and the last cluster of data indicates that it is the last cluster. A value (for example, FFFFh if one cluster is expressed by 16 bits) is recorded.

  As described above, the order of the cluster groups constituting the file can be correctly connected.

  In the FAT item (702) corresponding to a cluster that is not used in any file (unused cluster) among the clusters in the data area (504), 0000h is written as a value indicating that it is unused. ing.

  More specifically, referring to FIG. 7, for example, the start cluster address (701) of a file existing in the root directory entry (503) is 0004h from the information recorded in the root directory entry (503). If it is an address, since 0005h is written in the 0004h-th item of the FAT item (702), data subsequent to the data indicated in the cluster indicated by the 'a' part of 703 ('b' of 703) Is recorded at the 0005h-th cluster address in the data area (504). Similarly, since 000Ah is written in the 0005h-th item of the FAT item (702), the data following the 'b' portion ('c' portion of 703) is recorded at the 000Ah-th cluster address. I understand that. The 000Ah-th item of the FAT item (702) is written with FFFFh indicating the end of the cluster chain, so the 000Ah-th cluster ('c' portion of 703) of the data area (504) is the file of interest. This is the cluster that contains the last data.

  That is, in general, when focusing on a certain file and comparing the contents of the file and the allocation status of the cluster, it is as shown in FIG. 8 (in the case of 1 sector / cluster). This indicates that when the data position from the beginning of the file is known, it is possible to know which cluster it contains.

(Data read / write processing in the conventional FAT file system)
In the following, an outline of the processing when a file is opened on a FAT type disk in the conventional FAT file system and the data of the file is read and written will be described. In order to simplify the description, the file to be read / written from now on is specified as existing on the root directory entry (503), and the explanation of error processing is also omitted. Further, it is assumed that the sector size of the disk and the number of sectors / clusters are read from the boot sector (501) in advance and the numerical values are stored in the memory.

(Software configuration example)
FIG. 1 shows a configuration example of software when a user program reads / writes data on a FAT type disk. As shown in the figure, it can be roughly divided into 101 user program, 103 file operation unit, 104 directory operation unit, 105 FAT operation unit, 106 buffer cache operation unit, and 107 disk driver unit. Normally, a buffer cache memory used by the buffer cache operation unit (106) is secured on a semiconductor memory for speeding up. The user program (101) uses the interface provided by the FAT file system program (102) to perform operations such as reading from and writing to files in the storage device under the disk driver unit (107).

(Reading data from a file)
First, a procedure for reading data from a file will be described with reference to FIG.

  First, in step S901, an instruction to open a file from which data is to be read is issued from the user program (101). As a result of reading the root directory entry (503) into the buffer cache memory and analyzing it, if the specified file exists and the file is successfully opened, the file is represented as a unique numerical value for each execution unit of the program. A numerical value (file descriptor) is returned to the user program (101). In the file open process, the number of the start cluster of the file (denoted as “next”) is obtained from the corresponding root directory entry (503).

  In step S902, the user program (101) designates the file descriptor obtained in step S901, the buffer for storing the read data, the number of bytes to be read, and the read position of the file data.

  In the next step S903, what number cluster in the cluster chain corresponds to the data to be read from the read position of the file data designated in step S902, the sector size of the disk stored in advance, and the number of sectors / clusters. Ask for. In this description, this is expressed as n.

  Next, in step S904, the cumulative m (initial value is 1) of the cluster sequence numbers is compared with n obtained in step S903. If m and n match, the process proceeds to step S907, and if not, the process proceeds to step S905.

  In step S905, the sector in which the FAT item corresponding to the next is written is read from the FAT area (502) by one sector to the buffer cache memory, and in step S906, the FAT item next is focused on. As described above, the next cluster number constituting the cluster chain of the file is obtained. This result is substituted into next again, and 1 is added to the cumulative m of the cluster sequence numbers.

  The processing from step S904 to step S906 is repeatedly executed until the accumulated m of cluster sequence numbers matches the cluster number n obtained in step S903, and finally proceeds to step S907.

  Through the above steps, the cluster on the disk where the data at the read position specified in step S902 is assigned is substituted for next. Accordingly, in step S907, the data of the corresponding cluster address (next) is read from the disk to the buffer cache.

  In step S908, target data is extracted from the data read in step S907, and is copied in the same designated number of bytes to the buffer designated by the user program (101) in step S902.

  Finally, the file opened in step S909 is closed.

  Thus, the file data reading process is completed.

(Data writing process to file)
The processing procedure for writing data to the file is as shown in FIG.

  The description of the processing from step S1001 to step S1007 is the same as the corresponding processing when data is read from the file, and is therefore omitted.

  Through the above steps, the data at the target cluster address is read out to the buffer cache memory. In step S1008, the data stored in the buffer designated by the user program (101) in step S1002 is overwritten on part or all of this buffer cache.

  Finally, the file opened in step S1009 is closed.

  Thus, the file data writing process is completed.

(Processing of the buffer cache operation unit)
As described above, when reading / writing data on the disk, it is necessary to read the data on the normal disk once into the buffer cache memory. The role of the buffer cache operation unit (106) is to cut out a necessary portion of the data read into the buffer cache memory with the sector size of the disk as a delimiter and copy it to the user's buffer or to change it. Is to copy only or all of this from the user's buffer to the buffer cache. Furthermore, by preparing several buffer caches divided by sector size and managing them for each sector number, once read data is stored for a certain period of time, and the user program (101) or FAT operation unit (105) When trying to read again the data of the same sector number as the data read in the past, the data is read from a medium (semiconductor memory) having a higher access speed than the disk medium, thereby speeding up the operation of the user program (101). That is.

  This state will be briefly described with reference to FIG.

  Suppose now that there is a buffer cache delimited by several sector sizes. At this time, if a sector to be read or written from the file system is designated, first, in step S1101, a buffer cache including data of the corresponding sector is searched from all buffer caches.

  If there is a corresponding one in the buffer cache (YES in step S1102), this is returned. If there is no corresponding item (NO in step S1102), a buffer that is not used in steps S1103 and S1104, that is, not referenced by another program is searched, and one of them is selected. At this time, an algorithm called LRU (Least Recently Used) is generally used as a method for searching for a buffer that is not referenced by another program.

  Finally, a read or write operation from the disk is performed on the selected buffer (step S1105).

As another conventional example, a cache buffer for storing a FAT (file management table) is provided in the magneto-optical disk apparatus to shorten the access time for recording and reproducing information, thereby improving the working efficiency of the information recording / reproducing apparatus. An apparatus (for example, refer to Patent Document 1) can be used.
JP-A-8-110868

  The LRU algorithm described above as an example of a method for selecting an appropriate buffer from unused buffers described in the processing of the buffer cache operation unit is “discarding in order from the buffer with the oldest reference time”. Is.

  When the data of the same sector number is read again after the buffer cache data is discarded by the RU algorithm, the data is read from the disk having a low access speed. However, in the case of the FAT file system, it can be easily imagined that the number of times of referring to the FAT area increases due to its nature. Therefore, when the data in the FAT area is frequently discarded from the buffer cache, there is a disadvantage that the read / write speed of the file is reduced accordingly. In particular, in an embedded device in which the memory capacity for program execution is limited and a large amount of memory cannot be used for the buffer cache, there is a high probability that the FAT area buffer cache will be discarded within a short time.

  The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a buffer cache operating method for improving the read / write speed of a disk in a FAT file system.

  The present invention relates to a disk operation means for acquiring the sector size and sector / cluster size of a data disk to be subjected to file operation, a request from a user program, a file specified by a file name by managing a file descriptor File handling means for opening or closing files, creating new files, or manipulating file pointers to indicate from which position in the file space a read or write operation is performed, and by instructions from the user program Reads the directory entry information of each file or directory stored on the disk and analyzes it to search for the existence of the file instructed by the user program or to include the data at the beginning of the file The latest directory entry information, such as the cluster address on the disk, the file size, or the file attributes, file size, and last access date updated by file or directory operations The directory on which the data exists based on the directory entry operation means for performing the directory entry configuration editing operation such as writing to the disk, the position of the data on the file to be read / written, and the FAT area information read from the disk The FAT operation means that obtains the above cluster address, or correctly forms the cluster chain of the file and writes it to the disk, and a plurality of buffer caches delimited by the sector size, the reference number, the sector number, Past access history and data It is managed separately, and a buffer cache including data of a sector designated by an instruction from the disk operation means, file operation means, directory entry operation means, or FAT operation means is returned, or an unused buffer cache is selected. Or buffer cache operating means for reading data from the disk to the buffer cache and writing data to the buffer cache.

  With the above configuration, the buffer cache including the data in the FAT area is handled separately from the buffer cache for other data, and the data in the FAT area is included when a data read or write operation is performed on a certain file. The buffer cache is stored in the buffer cache for a longer period than normal data.

  As described above, in the FAT file system and its buffer cache operation method according to the present invention, a buffer cache including data in the FAT area is handled separately from a buffer cache including other data. When the data is read or written, the buffer cache including the data in the FAT area is stored in the buffer cache for a longer period than the normal data. As a result, there is an effect that the cache hit rate of the data in the FAT area is improved, and the read / write speed of the disk can be increased without using a large memory capacity for the buffer cache.

(System configuration)
The recording format of the FAT disk, the directory entry structure, and the FAT area structure used in the description of this embodiment are the same as those shown in FIGS. Omitted.

  FIG. 1 is a software configuration diagram of the FAT file system used in the description of this embodiment. 101 is a user program that uses the FAT file system, and 102 is a FAT file system program that implements the FAT file system according to the present invention.

  103 receives a request from the user program (101), manages file descriptors, opens or closes a file specified by a file name, creates a new file, or reads from any position in the file space. Or a file operation unit for managing a file pointer indicating whether to perform a write operation.

  104 analyzes the directory entry information read from the disk, searches for the existence of a file instructed by the user program (101), acquires a cluster including data at the head of the file, The latest directory entry, such as obtaining the file size or updating the file size that has changed due to operations such as writing to the file, the date of the last access to the file, and recording these in the directory entry A directory entry operation unit 105 that analyzes, manages, and updates a directory structure such as writing information to a disk, and 105 stores the target data from the position of data on a file to be read / written and the FAT item read from the disk. On the recorded disc Or get the raster address, a FAT operation unit that performs such processing to be written to the disk to properly form a cluster chain of the file.

  106 manages a plurality of buffer caches separated by sector size based on the number of references, sector number, past access history, and data type, and is designated by an instruction from the file operation unit, directory entry operation unit, or FAT operation unit If there is a buffer cache containing the data of the selected sector, return it, or if there is not, select a buffer cache that is not referenced by the program using the LRU algorithm and the data type of the buffer cache together. This is a buffer cache operation unit that discards the contents of the cache and newly reads data into the buffer cache or writes data into the buffer cache.

  Reference numeral 107 denotes a disk driver unit that reads data from a disk storage medium such as a floppy (registered trademark) disk, hard disk, or flash memory card (not shown) as a data storage area, or writes data to the medium. is there.

  FIG. 2 is a block diagram showing a schematic hardware configuration for realizing the functional configuration described above. In the figure, reference numeral 201 denotes a CPU which loads and executes the above-described user program (101) and FAT file system program (102). A RAM 202 stores the above-described user program (101) and FAT file system program (102), and is used as a buffer cache area (203) when reading or writing data on the disk. It is used as a work area for saving or saving various variables used during the operation of each program including the buffer cache management unit according to the present invention. Reference numeral 204 denotes a disk device serving as a data storage area. A system bus 205 connects the CPU (201), the RAM (202), and the disk device (204).

  FIG. 3 shows a configuration example of the buffer cache used in the description of this embodiment. 301 is a reference number, 302 is a sector number, and 303 is a flag indicating a data type indicating whether the buffer cache is data in the FAT area (1 in this description) or not (0 in this description). .

  In 304, a value such as a memory address specifying the next buffer cache address of the buffer cache forming the chain structure is recorded, and the access history to the buffer cache is managed by changing the chain structure of the buffer cache.

  A part 305 stores actual data.

(Detailed operation explanation)
Next, the operation of the FAT file system in this embodiment will be described. In order to simplify the explanation, the file to be read or written is intended for a file on the root directory of the FAT type disk, and detailed explanation of error processing is also omitted. Further, it is assumed that the sector size of the disk and the sector / cluster number are read from the boot sector (501) in advance and the numerical values are stored in the memory (202). Also, the data read / write processing from the file is the same as that of the conventional example, so the description thereof will be omitted. Hereinafter, the processing of the buffer cache management unit will be described with reference to the flowchart of FIG.

  Assume that there are n buffer caches configured as shown in FIG. If a read or write sector is specified from the user program (101), file operation unit (103), directory operation unit (104), FAT operation unit (105), etc., n buffers are first searched in step S401. Then, referring to the sector number (302) in FIG. 3, the search for the buffer cache including the data of the corresponding sector is started. If there is a corresponding one in the buffer cache (YES in step S402), this is returned.

  As a result of searching all the buffers, if there is no corresponding one (NO in step S402), an unused buffer cache (that is, not referenced by another program) is then processed by the processing from step S403. Move on to search for.

  In steps S403 and S404, referring to the reference number (301) and the data type (303), a buffer cache including data other than the FAT area and the reference number being zero is selected from the n buffer caches using the LRU algorithm. Search using in combination. If there is a buffer cache that meets the conditions (YES in step S404), this is returned.

  If there is no buffer cache that meets the condition in step S404 (NO in step S404), a buffer cache that satisfies the condition that the reference number is zero is searched from the n buffer caches in steps S405 and S406 by the LRU algorithm. Return this.

  Finally, with respect to the buffer cache selected by the processing in steps S401 to S406, the buffer cache chain indicated by the memory address (304) of the next buffer cache is reconfigured in step S407 according to the LRU algorithm, and the corresponding processing is performed. A read / write operation to the buffer cache is performed (step S408).

The software block diagram of the FAT file system in the Example of this invention, and a prior art example. FIG. 2 is a schematic block diagram of hardware for operating the FAT file system shown in FIG. 1. The block diagram of the buffer cache in the Example of this invention. The flowchart which shows the process sequence in the buffer cache operation part in the Example of this invention. 5 shows a recording format of the FAT type disk in the embodiment of the present invention and the conventional example. The block diagram of the directory entry corresponding to one file in the Example of this invention, and a prior art example. The conceptual diagram of the FAT area | region of the FAT type disc in the Example of this invention, and a prior art example. The example of the Example of this invention and the example of the cluster allocation to the file in a prior art example. The flowchart which shows the procedure at the time of the file data read in the Example of this invention, and a prior art example. The flowchart which shows the procedure at the time of the data writing in the Example in this invention, and a prior art example. 9 is a flowchart showing a processing procedure in a buffer cache operation unit in a conventional example.

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 User program 102 FAT file system program 103 File operation part 104 Directory operation part 105 FAT operation part 106 Buffer cache operation part 107 Disk driver part 201 CPU
202 RAM
203 Buffer cache memory 204 Disk device 205 System bus

Claims (3)

  In a FAT file system that manages a file using a file allocation table, a file operation unit that acquires the sector size or sector / cluster size of a data disk that is a file operation target, and receives a request from a user program, A file pointer that indicates whether to open or close a file specified by a file name by managing descriptors, create a new file, or read data from or write data to a location in the file space In response to instructions from the file operation means for performing the above operations and the user program, the directory entry information of each file or directory stored on the disk is read, analyzed, and analyzed by the user program. Search for the existence of the file or directory specified by, get the cluster address on the disk containing the data at the beginning of the file, get the file size, or operate on the file or directory Directory entry operation means for performing directory entry configuration editing operations such as writing the latest directory entry information updated by the disk to a disk, and editing various attributes of files and directories, and the position of data on the file to be read / written And FAT operation means for obtaining the cluster address on the disk where the target data exists from the FAT information read from the FAT area, or processing for correctly forming the cluster chain of the file data and writing it to the disk And Sek A plurality of buffer caches separated by size are managed by reference number, sector number, past access history, and data type, and according to instructions from the disk operation means, file operation means, directory entry operation means, and FAT operation means A buffer cache operation that returns a buffer cache containing data for the specified sector, selects an unused buffer cache, reads data from the disk to the buffer cache, or writes data to the buffer cache And a FAT file system.   In the buffer cache operation means of the FAT file system according to claim 1, when an unused buffer is searched, it is preferentially used from a buffer including data other than the FAT area in combination with the LRU algorithm. It is characterized by selecting.   In the buffer cache operation means of the FAT file system according to claim 1, when searching for unused buffers, if all the buffers are data in the FAT area, they are referred to by the LRU algorithm. A buffer including data having the oldest time is selected.

Images