TWI459196B - Log-structured file system - Google Patents

Description

Login for Flash Memory with Improved Data Access Performance - Structured File System

The invention relates to a login structured file system (LOG-STRUCTURED FILE SYSTEM) for flash memory, in particular to a login structured file system for improving data access performance in a flash memory (LOG-STRUCTURED FILE SYSTEM) ).

The hard disk in the computer host is a computer data storage center. As the hard disk capacity increases, the operating system is also equipped with different file management systems to maximize the access efficiency of the hard disk, such as the early FAT16 and FAT32 file management systems. Today's NTFS file management system; the biggest difference between the NTFS and the FAT series file management system is that the FAT system uses the "File Allocation Table" to manage disks and files, but the NTFS file management system is Use "Master File Table (MFT)". As shown in the fifth figure, the management file table mainly contains the header, file or folder name, file data, file or folder related information, creation date, etc. . When the operating system generates new data to be stored on the hard disk, the NTFS file management system generates a file management table for the newly added data, and stores the file or folder on the hard disk together. Generally speaking, in order to improve the storage efficiency of the hard disk file, the innermost circle of the hard disk will be planned as an index file storage area, so that the head can quickly find the index data of the file to be read by the operating system, and then according to the index data storage bit. Address, read the data file corresponding to this index data.

However, the management file list is also a file on the hard disk. As the number of files or folders increases, the number of management file tables becomes relatively large, which increases the file capacity of the record management file table. The configuration unit of the disc (which can be set to 512B to 4096KB in NTFS) may face split storage processing and cause file fragmentation to be discontinuous; thus, it is possible to reduce the performance of the operating system to access hard disk files or folders. .

As for the data capacity of the file or folder exceeds the hard disk setting configuration unit, there will be a chance to store in the discontinuous configuration unit after the split, resulting in broken and non-continuous files, but because the information is stored in the management file table, Therefore, the hard disk cooperates with the NTFS file management system, and the read performance of the non-contiguous files or folders is not significantly reduced.

It can be seen from the above description that the NTFS file management system cooperates with the hard disk structure and the control mode of the read/write head to provide the best way to read and write files or folders in terms of hard disk access performance, but it cannot be applied to the current It is quite popular for the access management of files or folders of flash memory devices. The main reason is that the data storage structure of the flash memory device is completely different from that of the hard disk. The actual memory structure is composed of array memory cells, so the actual memory structure is quite different from that of the hard disk, although it is thinner than the hard disk. The size of the flash memory device, due to the special memory structure of the flash memory device, also caused two congenital restrictions: one can not directly overwrite the data, and the other is the limited number of erasures (about 100,000 times, but each flash Depending on the type of memory, for example, MLC is 10,000 times and SLC is 100,000 times. Therefore, flash memory devices must use different file management systems.

The following is a further description of the current method of accessing files or folders by a flash memory device.

As shown in FIG. 6A, when the operating system requests the flash memory device to store the new data command, the file management system generates an index data table including the file name or the folder name and related information, such as NTFS. The index data sheet is stored with the file or folder in the flash memory area (20) of the flash memory device. When the operating system wants to delete one or several files (files B and E) or folders of the flash memory device, the index data table will also be read first, and after the data page stored in the corresponding file or folder is obtained, The block (21) and its index data table of the data page can be deleted, so that the blocks (21) form free blocks for the storage of new files or folders, as shown in Figure B. In this way, if the file size (files G and H) or the folder capacity is larger than the blocks, it must be divided and stored in other free blocks, resulting in discontinuous files (files H1 and H2) or folders. .

Since the flash memory device is arranged in an array of a plurality of blocks and has no read head design, although the management system cooperates with the establishment of an index data table for accessing files or folders, it does not contribute to improvement. Read the performance of discrete files or folders. As the capacity of flash memory devices increases, the chances of users storing files or folders with a single large data capacity become high. How to establish an effective file management system to improve file access performance will help The development of higher capacity flash memory devices.

In view of the above-mentioned deficiencies of the file management system of the existing flash memory device, the main object of the present invention is to provide a LOG-STRUCTURED FILE SYSTEM for improving the data access performance of the flash memory.

The main technical means used to achieve the above purpose is to enable the login structure. The file system includes: a command decoding module, which interprets instructions from the upper operating system, including new and delete commands; wherein the command decoding module replaces the delete command with a move command when determining that the operating system is a delete command; An index data table establishing module is connected to the command decoding module, and when determining the new command, establishing an index data table with the newly added file or folder, the index data table includes the file or folder name and Relevant information, including relevant date and source of establishment; a sequential access module is linked to the command decoding module and the index data table to create a module, when the new command is judged, a new file or folder will be added. The data and its index data table are stored together in the last block of the flash memory area; when the delete command is a move command, the source of the index data table of the target file or folder is changed; a free block is integrated. The module is linked to the sequential access module, and the corresponding file or folder whose source of the changed index table is changed is deleted.

The above invention provides a flash memory device capable of managing files or folders stored in a continuous block, mainly when the operating system releases and deletes any of the files or folders, first changing the source of the index data table to make the operation The system cannot find the file or folder in the original directory, and the flash memory device is not stored in the file or folder for the user; furthermore, the free block splicing module of the present invention will be specific Time, delete the changed index data table and its corresponding file or folder to release more consecutive free blocks; therefore, the file management system of the present invention does not release files or folders directly Free block, When the new file or folder is saved, the free block cannot be stored in the contiguous block because the free block capacity is insufficient. Moreover, because the present invention does not release the delete command in the operating system, the block is removed. Therefore, the flash memory device of the present invention achieves a slower upper limit of the number of erasures, and relatively increases the service life of the flash memory device.

First, please refer to the first figure and the second figure, which is a functional block diagram and a flow chart of a login-structured file system (LOG-STRUCTURED FILE SYSTEM) for improving data access performance of the flash memory of the present invention. The login-structured file system (10) is coupled to a flash memory area (20) consisting of an array free block, and includes: a command decoding module (11), which is interpreted from the upper layer operating system ( 30) The instruction (S1) includes a new and delete command; wherein the command decoding module (11) determines that the operating system (30) is a delete command (S2), that is, replaces the delete command with the move command (S3); The index data table establishing module (12) is connected to the command decoding module (11), and when determining the new command (S4), establishing an index data table with the newly added file or folder, the index data table Contains the file or folder name and related information (S7), wherein the related data includes the date of creation and the source of the establishment; a sequential access module (13) is linked to the command decoding module (11) and the index data table is established. Module (12), when judging the new command (S4), when there is a free block The state (S5), add a file or folder information and the index table from the last free memory blocks of the flash zone (20) to begin storing; Analyzing deleted in order When the command is moved (S2), change the source of the index data table of the target file or folder (S3), for example, change the original E:\B.txt to E:\FILE1\B.txt; a free block The merging module (14) is linked to the sequential access module (13), and deletes the corresponding file or folder whose source of the index data table is changed. In this embodiment, the free block is completely When the new command occurs, the module (14) executes the flash memory area (20) without a free area (S6).

Please refer to the third A to C diagrams to further illustrate the storage state of the internal block of the flash memory device when the invention adds a new and delete command to the operating system.

First, as shown in the third figure A, when the operating system releases the delete file A and E commands, the command decoding module will judge the delete command, and the delete command into the move command to the file B and E. The source of the file related information in the index data table is changed, but the block (21) storing the files B and E is not deleted, so the flash memory area (20) will not receive the delete command every time. That is, many continuous free blocks (21) are released. In this way, the user cannot find the file or folder in the original directory through the operating system, and the user does not have the flash memory device stored in the file or folder.

As shown in the third figure B, when the operating system releases a new file G and H commands, the command decoding mode combination is judged to be a new command, and the index data table of the files G and H is established, due to the flash memory. There is no discontinuous free block (21) in the area (20), so the files G and H can be stored in consecutive free blocks in sequence; and unless the file G or H data capacity is larger than the current free block capacity, it is added. Files can be stored continuously in the flash memory area. Therefore, if there is no free block in the flash data area (representing full), it can be used as a free block block to start the time of the free block, or to reach the preset time point, such as As shown in Figure 3C, the index data table with the changed source and its corresponding file B and E or folder will be deleted. The blocks for storing files B and E will be deleted, so that there will be discontinuous freedom. Area, so further discontinuous free blocks are consolidated into contiguous blocks.

Please refer to the fourth figure, which is a second preferred embodiment of the login-structured file system (10a) for improving data access performance for the flash memory of the present invention, which is compared with the first preferred embodiment. The example includes a cache memory area (22). The cache memory area (22) is divided into one block from the cache memory area (20), and is used for storing the index data table, and will cache the memory when the computer is turned off. The index data table in the area (22) is restored to the flash memory area (20), thereby speeding up the performance of the operating system for reading or deleting files, and since the cache memory area (22) is not stored as a file or folder, It is also possible to reduce the number of times that the memory area (22) must be restored to the flash memory area (20) because the cache memory area (22) is full.

Since the computer may not be properly shut down, the index data table of the cache memory area may not be saved back to the flash memory area, so the present invention further adds a working state filtering module, which is connected to the flash memory area to filter the current An index data table containing a specific file format (.doc/.exl/.txt/.ppt, etc.), if judged to be those file formats, is stored in the flash memory area from the cache memory area after being turned off; In this way, abnormal shutdown can be avoided, and the file created in the job disappears.

In summary, the present invention is applied to a file management system of a flash memory device, and does not release a discontinuous free block every time a file is deleted, but can effectively manage files or folders stored in a continuous block; Because the present invention does not delete the block when the delete command is issued in the operating system, the flash memory device of the present invention achieves a lower limit of the number of erasures and relatively increases the service life of the flash memory device.

(10) (10a). . . File management system

(11). . . Command decoding module

(12). . . Index data table creation module

(13). . . Sequential access module

(14) ‧‧‧Free block consolidation module

(15)‧‧‧Work status filter module

(20)‧‧‧Flash memory area

(21) ‧‧‧ Blocks

(22)‧‧‧Cache memory area

(30) ‧‧‧Operating system

The first figure is a functional block diagram of a first preferred embodiment of the login-structured file system of the present invention.

The second figure is a flow chart corresponding to each functional block diagram of the first figure.

The third A to C diagrams are schematic diagrams of the present invention for deleting, new, and non-continuous free blocks in the operating system.

Figure 4 is a functional block diagram of a second preferred embodiment of the login-structured file system of the present invention.

The fifth picture is a schematic diagram of the structure of the management file table of the NTFS file management system with both hard disks.

Figure 6A: Schematic diagram of deleting files from the flash memory device file management system.

Figure 6B: Schematic diagram of adding files to the flash memory device file management system.

(10). . . File management system

(11). . . Command decoding module

(12). . . Index data table creation module

(13). . . Sequential access module

(14). . . Free block consolidation module

(20). . . Flash memory area

(30). . . working system

Claims (5)

A login structured file system for improving data access performance in a flash memory is connected to a flash memory area composed of an array free block, and includes: a command decoding module, which is interpreted from the upper layer operation The command of the system includes a new and delete command; wherein the command decode module replaces the delete command with a move command when determining that the operating system is a delete command; and an index data table creation module is coupled to the command decode module, and When judging the new command, the index data table of the newly added file or folder is created, and the index data table includes the file or folder name and related data; a sequential access module is connected to the command decoding module. When judging the new command, the newly added file or folder data is sequentially stored in the continuous free block; when the delete command is the move command, the source of the index data table of the target file or folder is changed; The block merging module is linked to the sequential access module, and the corresponding files or folders of the source of the changed index data table are deleted in full. For example, in the login structured file system described in claim 1, the free block merging module is adapted to the sequential access module to determine that there is no free block, and the corresponding file of the source of the index data table is changed. Or delete all folders. For example, in the login structured file system described in claim 1 or 2, the flash memory area further divides a cache memory area for use as a storage index data table. For example, in the login structured file system described in claim 3, the flash memory area returns the index data table in the cache memory area to the flash memory area when the computer system sends a computer shutdown signal. For example, in the login structured file system described in claim 4, the flash memory system further includes a working status filtering module for filtering an index data table currently containing a specific file format, and has been established at present. Files that match these file formats are stored in the flash memory area from the cache memory area after being turned off.