CN1822230A - Method for realizing memory file system based on NAND Flash - Google Patents

Abstract

The invention discloses a method for realizing a file system based on a NAND Flash memory. It establishes the NAND Flash memory file system control block index, uses the index to allocate linked list blocks and recover linked list garbage data, and reads and writes data with burst input and output, and uses the garbage block recovery mechanism to recover those that are no longer included. A block of valid data. The file system can significantly reduce the loading time, perform fast reading and writing with low power consumption, and has excellent read and write performance for large files, evenly erases and writes the storage space, can realize efficient management of blocks, and has high space utilization efficiency. The file system can effectively manage large-capacity NAND Flash memory and realize fast startup, which is suitable for various large-capacity mobile data storage and multimedia devices.

Description

Implementation method of file system based on NAND Flash memory

technical field

The invention relates to the field of embedded system file systems, in particular to a method for realizing a file system based on NAND Flash memory.

Background technique

Flash memory is a kind of memory that has developed rapidly in recent years and belongs to non-volatile memory. It has the characteristics of EEPROM electrical erasability, and also has the advantages of low power consumption, high density, small size, high reliability, erasable, rewritable, and reprogrammable. Flash memory also has some disadvantages, including limited erasing and writing, and any Flash device must perform erasing before writing. There are mainly two types of Flash memory, NAND and NOR. NAND Flash reads and writes in units of blocks, and it is very simple to perform erasing operations; while NOR reads and writes in units of bytes, requiring all bits in the target block to be written to 0 before erasing. When erasing NOR devices, it is performed in blocks of 64-128KB, and the time to perform a write/erase operation is 1-5s; when erasing NAND devices, it is performed in blocks of 8-32KB. It takes 4ms. When erasing is performed, the difference in block size further widens the performance gap between NOR and NAND. Therefore, to say that the traditional file system is applied to Flash memory, it must be improved. Generally speaking, it is realized through FTL, but the use of FTL requires payment of patent fees.

At present, JFFS2 and YAFFS are mostly used as the file system of the basic Flash memory. JFFS2 and YAFFS startup needs to scan the whole storage interval to find index nodes and file control blocks. For a large storage capacity, there is a problem that the loading time is too long; both of them consume a lot of memory.

Contents of the invention

The object of the present invention is to provide a kind of realization method based on NAND Flash storage file system.

The technical scheme that the present invention solves its technical problem adopts is as follows:

1) Flash file system loading

The loading of the Flash file system is started by the booting of the Flash file system control block FlashFSCB, the latest FlashFSCB is found from the FlashFSCB data interval, the file system is loaded, the file system data is initialized, and the garbage cleaning kernel thread is started to complete the initialization work;

2) Flash file system block allocation and associated data collection

The Flash file system uses a method that associates data, that is, multiple blocks of content from the same file,

Space allocation strategies grouped into a contiguous region within a single unit of erase;

3) Flash file system garbage block recovery

Flash file system garbage block collection copies all valid data blocks from a pre-erase unit to another new erase unit called a transfer unit, then updates the block mapping table and then erases the obsolete pre-erase unit ;

4) Flash file system error recovery

The Flash file system uses the "write first, then erase" method for error recovery. When updating the data of a sector of Flash, the previous data will only be allowed to be erased after the update operation is completed and the newly stored data is verified successfully. If the operation is successful, the data in the new sector is valid, otherwise the data in the old sector is valid.

Compared with the background technology, the present invention has the beneficial effects that:

The present invention is a kind of realization based on NAND Flash memory file system, and its main function is to improve file system loading speed by establishing file system control block FlashFSCB index in the beginning area of NAND Flash interval; Manage and schedule kernel threads for garbage cleaning to achieve uniform erasing and writing of storage areas; realize burst input and output, and reduce power consumption. The file system can effectively manage large-capacity NAND Flash memory and realize fast startup, which is suitable for various large-capacity mobile data storage and multimedia embedded devices.

(1) Accelerate the loading speed of the Flash file system. The present invention establishes the FlashFSCB index of the file system control block in the beginning area of the NAND Flash interval, finds the super block through the pointer pointing to the super block interval in the index area, and performs loading and initialization of the file system without scanning the entire Flash storage space, and the speed advantage is obvious. Certain real-time.

(2) Uniform erasing and writing of the Flash file system. The Flash memory cannot be reused indefinitely. The present invention utilizes all the storage units in a balanced manner to apply circularly to the entire interval without allowing a certain unit to be overused, thereby achieving uniform erasing and writing, and prolonging the service life of the NAND Flash.

(3) Reduce memory consumption. The present invention accesses Flash from a small storage window, and can reduce the number of calls to map physical blocks to the window through the assembled associated data, speed up the continued access of files, and reduce the generation of garbage blocks, thereby reducing memory usage. usage of.

(4) Reduce the power consumption of the Flash file system. All access to file data is indirectly through the SRam buffer. When accessing file data, the corresponding page is read into the SRAM buffer according to the instruction of the index node, and then read and write access is performed. When data needs to be modified, modify the SRAM buffer page corresponding to the page. The data write-back adopts the delayed write technology. When the memory buffer page is discarded, it is written into the Flash page. This burst type I/O access can effectively reduce power consumption.

Description of drawings

Figure 1 is a flow chart of the file system;

Fig. 2 is a flowchart of file system initialization;

Fig. 3 is a flow chart of free block allocation;

Fig. 4 is a flow chart of garbage block recovery;

Fig. 5 is a flow chart of file data error detection and recovery.

Detailed ways

The specific use of the file system based on NAND Flash memory is as follows.

1) Flash file system loading

In Figure 1, the file system is implemented through file system initialization, reading/writing file data, data block processing, and garbage collection. Since error handling is distributed throughout the process, 4) Flash file system error recovery explained in .

In Fig. 2, the loading of the Flash file system starts by the boot of the Flash file system control block FlashFSCB, and FlashFSCB is a data structure residing in a certain Flash memory interval and indicating the usage of the Flash file system on this interval, including index nodes, Directories, files, free interval allocation, dirty data intervals and garbage cleaning information, implement object-oriented management methods for various index nodes, directories, files and other file system data; start garbage collection kernel threads.

Construct the Flash file system control block index area in the interval start area, which stores pointers to the intervals storing FlashFSCB data. When the file system is loaded, the effective pointer is determined by comparing the size of the pointer Version field, and then by comparing the FlashFSCB Version field, the FlashFSCB data is obtained. Find the latest FlashFSCB in the interval, load the file system, initialize the index node table, file table and directory structure, and start the garbage cleaning kernel thread to complete the initialization work. Because the FlashFSCB control block is searched through the index to avoid the need to scan the entire interval in the traditional method. The practice of determining the directory structure of the file system significantly shortens the initialization time.

2) Flash file system block allocation and associated data collection

In Fig. 3, after the free interval allocation request appears, first check the current available block pointer to determine whether the current block can meet the requirements, if the requirements can be met, modify the current block pointer data information, and allow the allocation; if the requirements cannot be met, Check whether the head of the free block pointer has been reached, if it has arrived, start the garbage cleaning thread to perform garbage cleaning, otherwise, turn the pointer to the next free block.

The Flash file system uses a space allocation strategy: the associated data, that is, multiple blocks composed of the contents of the same file, are assembled into a continuous area in the same single erasing unit, and the Flash file system is as far as possible in the same area. A storage pool consisting of multiple physically continuous free blocks is maintained in the erase unit. If such a continuous storage pool cannot be realized, the Flash file system will try to ensure that all the blocks in the pool are in the same erase unit. If Even if this is not possible, the Flash file system will try to allocate the block pool to an erase unit with the most available space to achieve even erase.

3) Flash file system garbage block recovery

In Fig. 4, at first the garbage block recovery of the Flash file system determines the garbage block to be reclaimed, and judges whether it contains valid data, if the block does not contain valid data, then directly erase the block; if the target block contains If valid data is obtained, write data into other blocks, and perform data verification at the same time. If the result is correct, copy all valid data blocks from a pre-erased unit to another new erased unit called a transfer unit. , and then update the block mapping table and then erase the obsolete pre-erased unit. In this way, although these data are now stored in other spaces of the Flash memory, the original block still contains the original data when it appears to the outside world, otherwise the error processing will be performed . Finally, hang all free blocks into the free block list.

In the Flash file system, garbage block collection is only triggered when the block allocation algorithm needs it. The block allocation algorithm tries to keep a continuous free block pool in the same erase unit. If the block pool becomes too small, The block allocation algorithm will immediately use the garbage block recovery algorithm for garbage block recovery, and then the algorithm will find and reclaim the erasure unit that best matches the following criteria: the area with the most waste blocks, the least erasure cycle data, or the most static area ;

In addition, the garbage block recycling algorithm will also use a random selection processing method to ensure that the recycling process can evenly cover the entire storage space without deviation due to the way the application uses data;

4) Flash file system error recovery

In Figure 5, writing data to Flash may go wrong when responding to a file system write request, during garbage block collection, or even when formatting or erasing Flash, in which case the Flash file system is able to recover from the error come over.

The Flash file system uses the method of "write first and then erase". If an error occurs during the operation, when updating the data of a sector of Flash, the previous data will be deleted only after the update operation is completed and the newly stored data is verified successfully. If the operation is successful, the data in the new sector is valid, otherwise the data in the old sector is valid. At the same time, the Flash file system will not only put the critical mapping information in RAM. If there is a power failure when updating the mapping information Or if the Flash media is removed, the original mapping information is still available. When the power is restored or the Flash is reconnected, the Flash file system can use the information stored on the Flash to rebuild or verify the RAM-resident version of the Flash mapping table. . All access to file data is indirectly buffered. When accessing file data, the corresponding page is read into the SRAM buffer according to the instruction of the index node, and then read and write access is performed. When the data needs to be modified, the SRAM buffer page corresponding to the modified page is modified, and the data is written back using the delayed write technology. After the correct data is found, the memory buffer page is discarded and written into the Flash page. This burst type I/O access can effectively reduce power consumption.

Claims (1)

1. a kind of realization method based on NAND Flash memory file system, it is characterized in that: 1) Flash file system loading The loading of the Flash file system is started by the booting of the Flash file system control block FlashFSCB, the latest FlashFSCB is found from the FlashFSCB data interval, the file system is loaded, the file system data is initialized, and the garbage cleaning kernel thread is started to complete the initialization work; 2) Flash file system block allocation and associated data collection The Flash file system uses a space allocation strategy that aggregates associated data, that is, multiple blocks composed of the contents of the same file, into a contiguous area within a single erase unit; 3) Flash file system garbage block recovery Flash file system garbage block collection copies all valid data blocks from a pre-erase unit to another new erase unit called a transfer unit, then updates the block mapping table and then erases the obsolete pre-erase unit ; 4) Flash file system error recovery The Flash file system uses the "write first, then erase" method for error recovery. When updating the data of a sector of Flash, the previous data will only be allowed to be erased after the update operation is completed and the newly stored data is verified successfully. If the operation is successful, the data in the new sector is valid, otherwise the data in the old sector is valid.

Images