KR20110070656A - Method and apparatus for processing data of flash memory - Google Patents

Abstract

PURPOSE: Data processing method and device are provided to shorten mount time by reducing the occupancy of a main memory space. CONSTITUTION: A mapping table processor(10) generates a mapping table corresponding to a saving space of a flash memory. A write buffer(20) performs data write operation of the flash memory. If the data is recorded by a page unit, a block selector(30) selects an empty table by searching a page table from the mapping table. The write buffer records the data of the corresponding page.

Description

Method and device for processing data in flash memory {Method and Apparatus for processing data of Flash memory}

The present invention relates to a data processing method and apparatus of a flash memory, and in particular, in a file system for storing data in a flash memory used as a main storage device of an embedded system, to reduce the amount of memory consumed by reducing the mapping table space consumption. A data processing method and apparatus for a flash memory.

The present invention is derived from the research conducted as part of the technology development project of the Ministry of Knowledge Economy and the Korea Institute of Information and Telecommunications Promotion. .

Recently, the use of various mobile devices such as mobile phones, MP3s, digital cameras, portable multimedia players (PMPs), and the like, has increased the number of embedded systems equipped with large-capacity flash memory. Increasingly, these embedded systems use a combination of Linux and the Native Flash File System for flash memory to reduce costs.

In general, flash memory has characteristics such as nonvolatile, low power consumption, and fast access speed, and is classified into NAND type and NOR type flash memory according to the type of gate constituting the memory cell. Among them, NAND flash memory is used in most devices for storing large data due to its high density.

Types of flash memory file systems include a journaling flash file system (JFFS) that uses a memory technology device (MTD), and another another flash file system (YAFFS). The file system adopts a log-structured file system to cover the wear-out characteristic of the flash memory, which has a limited number of erase times, and uses a small size unit for efficient memory usage. The memory space is divided and managed, and a mapping table composed of these units is managed in main memory.

Due to the characteristics of the flash memory, the existing file system used for the hard disk cannot be directly applied to the flash memory. Therefore, research on the file system for the flash memory has been actively conducted.

However, in an embedded system using a large amount of flash memory, there are many large files such as multimedia data. For this, it is unnecessary to use a mapping table entry.

In addition, the log structure file system (LFS) takes a long mount time by reading and mounting the entire flash memory, and a bigger problem is that the time increases in proportion to the capacity of the flash. Of course, users of embedded systems also want a lot of storage, but they don't want the main memory space that is unnecessarily spent on the mapping tables and the slow boot time.

Therefore, when using a conventional LFS-style file system (JFFS, YAFFS) as a file system for flash memory, there is a problem that the mapping table occupies a large amount of space in main memory and a mount time for configuring it in main memory is long. .

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problem, and an object of the present invention is to reduce the size of the mapping table required for a large file in a log structure file system of a flash memory, thereby reducing the main memory space consumption, and thereby mounting time. Provided are a method and an apparatus for processing data in a flash memory to shorten the time required.

In order to achieve the above object, a data processing method of a flash memory according to an embodiment of the present invention is a data processing method of a flash memory having a mapping table configured to correspond to blocks including a plurality of pages. Among them, a block in which data is stored in page units constitutes a mapping table for each page, and a block in which data is stored in blocks per block constitutes a mapping table for each block.

According to the configuration of the present invention, by reducing the size of the mapping table required for large files to reduce the main memory space consumption, thereby providing a data processing method for reducing the mount time, as a result, the use of less flash memory resources used by the user and booting There is an advantage in reducing the waiting time of the user by reducing the time.

1 and 2 are views for explaining a storage structure of a flash memory according to an embodiment of the present invention;
3 is a diagram showing the configuration of a data processing apparatus of a flash memory according to the present invention;
4 is a diagram illustrating a data writing method of a flash memory according to the present invention;
5 illustrates a mounting scheme in accordance with the present invention.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings.

The data processing method of a flash memory according to the present invention is applied to a flash memory having a mapping table configured to correspond to blocks including a plurality of pages. In the data processing method of the flash memory according to the present invention, a block in which data is stored in page units among the blocks of the flash memory constitutes a mapping table for each page, and a block in which data is stored in block units among the blocks constitutes a mapping table for each block. Includes features to

First, a flash memory to which the present invention is applied is a nonvolatile device capable of electrically writing / erasing, such as a NOR flash memory or a NAND flash memory. For example, the flash memory adopts EEPROM (Electrically Eraseable and Programmable Read Only Memory), which is a floating gate transistor. The NAND flash memory is a data storage flash memory in which cells are arranged in series. The NAND flash memory has a relatively large page unit with a read / write unit and a large capacity, making it easier to increase the capacity of the NAND flash memory.

However, flash memory has the following disadvantages as compared to a storage medium such as a conventional hard disk. First, flash memory cannot overwrite its original address when data is modified. Since each bit of flash memory is only toggled in one direction, the write operation must be preceded by an initialization operation. That is, the memory space must be initialized before the write operation. Second, since each block of the flash memory has a limited number of initialization operations, the usable memory space is drastically reduced when the total space of the flash memory is not evenly used. Various cleaning policies and wear-leveling techniques are needed to overcome and effectively use the disadvantages of flash memory. In addition, the NAND flash memory has a disadvantage that only the page unit access is allowed, not the byte unit.

Therefore, in the data processing method of the flash memory according to the present invention, in consideration of the characteristics of the flash memory, the flash memory (for example, NAND flash memory, etc.) is basically defined as one log, and this is from the beginning to the end. Log-Structure file method of logging is used. Since this structure is a method of writing data to a new area of the flash memory, it is possible to avoid the limitation of the erase operation before the write operation is performed.

1 and 2 are views for explaining a storage structure of a flash memory according to the present invention.

Referring to FIG. 1, a storage structure of a flash memory is divided into a plurality of blocks 2j blocks having a plurality of pages 2i pages. The flash memory stores and deletes data based on the configured pages and blocks. For example, if a block of flash memory is divided into m blocks, and each block is divided into n pages, the flash memory space has m x n pages.

Here, the size of the page is made small enough to be efficient when recording small sized data, and the size of the block is made large enough to be efficient when writing large sized data. In the case of the NAND flash memory shown in FIG. 1, read / write access is possible in units of 512 bytes, and data erasing can be accessed in blocks of 32 pages. Blocks may correspond to pages and blocks, respectively. For example, if a block size is 128KB, 64 2KB pages per block are configured according to the type of flash memory. If a block size is 16KB, 32 512B pages are configured per block. do.

The page of the flash memory includes a header, and actual data (hereinafter referred to as chunk) is stored in a space other than the header. In the case of NAND flash memory, a header is written to a spare area of 16 bytes per page, and the entire page can be used as a data space.

Referring to FIG. 2, the header includes a type, inode number, chunk number, and version fields. The inode number is a unique ID for each file stored in flash memory. The inode and data pages for a file stored in flash memory have the same inode number. A chunk number is a number assigned to each chunk by dividing a file stored in flash memory into a single fixed chunk. The chunk number is divided by the chunk size in a file and stored in order for each chunk. Chunk number is a valid field when the file size is relatively small or large. The version field is a page where data is stored. When a request for overwriting other data is received, the version field is written to another page. The version field is different from the page where the data is stored. This version field is a field for determining which pages are valid.

The type includes an inode, a data page, and a data block. An inode type is file metadata, commonly called an inode, that contains all information describing the file in addition to the actual data of the file, such as read / write permissions and modification time. However, the data corresponding to the inode does not include information about the location of the page / blocks. The data block type means that the entire block having the corresponding page has contiguous data belonging to only one file. In other words, the data block type means that the data stored in all pages constituting the block belongs to one file. Thus, the data block type indicates that only the first page of the block has a header, and that the data block is indicated in the header, without each page of the block having its own header. The data page type means that data stored in a page is data of a specific file.

Referring to FIG. 3, the data processing apparatus of the flash memory to which the data processing method of the flash memory according to the present invention is applied may include a mapping table processor 10, a write buffer unit 20, a block selector 30, and a page selector ( 40).

The mapping table processor 10 generates a mapping table corresponding to the storage space of the flash memory. For example, when the storage space of the flash memory includes m x n pages, the mapping table processor 10 generates a mapping table corresponding to the included storage space. In the present invention, the mapping table is used during the storage / read operation of the flash memory.

Here, the mapping table processing unit 10 creates a mapping table for each page only for blocks used to store data in units of pages of flash memory when generating the mapping table, and generates a mapping table for each large block of data to create unnecessary space. This should not be used.

In addition, the mapping table processor 10 determines a block unit or a page unit based on the type of the flash memory, and generates a mapping table accordingly. That is, the mapping table processor 10 creates a mapping table for each block in units of blocks on the main memory (not shown) when the flash memory is mounted. It then checks the type first by reading the header of the first page of each block (stored in spare area for NAND flash memory). If the type of the checked block is a data block type, the mapping table processor 10 determines that the headers of the remaining pages do not need to be read at all because all of the blocks belong to one file. Then, the mapping table processing unit 10 fills the block table entry of the mapping table for the corresponding block with only the read information.

Here, if the type of the identified block is a data page type or an inode type, the mapping table processor 10 determines that the corresponding block is used in units of pages rather than units of blocks. The mapping table processor 10 creates a page table entry of the mapping table corresponding to the determined block. The mapping table processing unit 10 then reads the header for each page in the block and fills in the page table entry. In this way, the mapping table processing unit 10 makes a page table entry of the mapping table only for blocks used in page units, not block units, so that a large amount of page table space is not unnecessarily used.

The write buffer unit 20 performs a data write operation of the flash memory. The write buffer unit 20 performs a data write operation to the flash memory using the above-described mapping table. That is, in the present invention, the write request to the flash memory is basically recorded in a log-structured file storage method such as JFFS or YAFFS. The write buffer unit 20 includes a write buffer for each file therein, and temporarily collects data requested for writing to the flash memory in the buffer. Then, the write buffer unit 20 periodically checks the buffer and writes a block if the collected data exceeds a block size, and otherwise writes a page.

Here, when the write buffer unit 20 writes a block, the block selection unit 30 searches for an empty block next to the last recorded block and writes the block. That is, when recording data in block units, the block selector 30 searches for a block table in the mapping table and selects an empty block. The write buffer unit 20 then writes data in the corresponding block.

When writing a page, the write buffer unit 20 finds a blank page through the page selector 40 and writes data to the corresponding page. That is, when recording data in units of pages, the block selector 30 searches for a page table in the mapping table and selects an empty table. The write buffer unit 20 then writes data to the corresponding page.

Here, when writing a page, the write buffer unit 20 does not select an empty block after the last written block, but finds a block in which only some pages are used among blocks already used. For example, the write buffer unit 20 is a case in which it is determined that the write does not exceed the block size (for example, when the data of more than the block size does not accumulate in the buffer for a predetermined time or the file is closed before stacking). You will need a blank page. If there is a corresponding block, the write buffer unit 20 may write a blank page in the block. That is, the write buffer unit 20 searches for a target among blocks used in page units instead of block units. In addition, if the block used in the page unit is not full, the write buffer unit 20 may select the next empty block as in the case of writing a block.

4 is a diagram illustrating a file writing method of a flash memory according to the present invention.

Referring to FIG. 4, the write buffer unit 20 receives a write request to flash memory (S10). Then, the write buffer unit 20 temporarily collects data requested for writing to the internal buffer.

Next, the write buffer unit 20 periodically checks the internal buffer to determine whether the collected data has exceeded the block size (S20).

In operation S10, when the data collected in the buffer exceeds one block size as a result of the determination, the write buffer unit 20 performs block writing. When writing a block, when writing data in block units, the block selector 30 selects an empty block by searching the block table in the mapping table. Then, the write buffer unit 20 finds an empty block next to the last written block and writes data requested to be written to the corresponding block.

On the other hand, if the data collected in the buffer does not exceed one block size as a result of the determination in step S20 (S20; No), the write buffer unit 20 writes an empty block after the last written block. Rather than selecting, a block in which only some pages are used among the blocks already used is first found (S30). That is, if it is determined that the write buffer unit 20 does not exceed the block size, the write buffer unit 20 needs a blank page instead of an empty block.

Here, in step S30, if there is a block having a blank page as a result of the determination (S30; Yes), the write buffer unit 20 writes data to the blank page in the block (S40). That is, when writing a page, the block selector 30 searches a page table in the mapping table and selects an empty page table. Then, the write buffer unit 20 finds a blank page and writes data to the page.

If, in step S30, the write buffer unit 20 determines that the blocks used in units of pages are not full (S30; No), the write buffer unit 20 selects the next empty block and writes the blocks as in the case of block writing (S50).

5 is a diagram illustrating a file mounting method according to the present invention.

Referring to FIG. 5, first, the mapping table processing unit 10 creates a mapping table for each block in units of blocks on a main memory (not shown) when the flash memory is mounted (S60).

Next, the mapping table processing unit 10 checks the type of the flash memory (S70). That is, the mapping table processing unit 10 first checks the type by reading the header of the first page of each block (stored in the spare area if the NAND flash memory).

Subsequently, if the type of the checked block is a data block type (S80: Yes), the mapping table processor 10 determines that the headers of the remaining pages do not need to be read at all because all of the blocks belong to one file (S100).

Then, the mapping table processing unit 10 fills the block table entry of the mapping table for the corresponding block with only the read information (S120).

In step S80, the mapping table processing unit 10 is not a table block type (S80; No), and the mating table processing unit 10 determines that the type of the checked block is a data page type or an inode type (S90; Yes). Is determined not to be used in units of blocks but in units of pages (S110). The mapping table processor 10 creates a page table entry of the mapping table corresponding to the determined block. The mapping table processing unit 10 reads the header for each page in the block and fills the page table entry (S120). In step S80, the mapping table processing unit 10 is not a table block type (S80; No),

In addition, if the type of the checked block is not a data page type or an inode type (S90; No), the mapping table processing unit 10 performs the mapping table in a block method or a page method according to a default value. Fill (S120). The default value is a preset value.

As described above, in the present invention, the page table entry of the mapping table is made only for blocks used in page units, not block units, so that unnecessarily large page table spaces are not used for large data.

Therefore, according to the configuration of the present invention, by reducing the size of the mapping table required for large files to reduce the main memory space consumption, thereby providing a data processing method for reducing the mount time, as a result of the resource consumption of the flash memory used by the user It has the advantage of reducing the user's wait time by reducing boot time.

Although a preferred embodiment according to the present invention has been described above, it can be modified in various forms, and those skilled in the art can make various modifications and modifications without departing from the claims of the present invention. It is understood that it may be practiced.

10: mapping table processing unit 20: write buffer unit
30: block selector 40: block selector

Claims (1)

A data processing method of a flash memory having a mapping table configured to correspond to blocks including a plurality of pages, the method comprising:
The block in which data is stored in page units among the blocks constitutes a mapping table for each page, and the block in which data is stored in block units among the blocks constitutes a mapping table for each block.

Images