KR20080038613A - The method of managing flash memory - Google Patents

Abstract

A method for managing a flash memory is provided to reduce a required memory amount of an address table for remapping, manage the flash memory efficiently by remapping data of a physical block to a logical block, and manage remapping information in the flash memory without any additional device. A super logical block including more than one logical block is generated. Pages of the logic block included in the super logical block are remapped to a physical block included in a super physical block. Remapping information for the pages of the logical block included in the super logical block is stored to a flash memory together with data by checking whether the remapping information for the logical block is stored in a ram, finding a location of the remapping information when the remapping information is not found, reading the remapping information stored in the found location, and storing the read remapping information to the RAM.

Description

How to manage flash memory {THE METHOD OF MANAGING FLASH MEMORY}

1 is a block diagram of a system based on a flash memory management method according to an embodiment of the present invention;

2 is a diagram illustrating a superblock in a flash memory management method according to an exemplary embodiment of the present invention.

3 is a view for explaining a re-imaging method in a flash memory management method according to an embodiment of the present invention.

4 is a view showing a specific example of the re-imaging method of FIG.

5 is a diagram illustrating a structure of re-imaging information stored together with data in a flash memory according to the present invention.

FIG. 6 is a diagram for explaining a method of collecting pages having similar overwriting frequencies into specific physical blocks. FIG.

The present invention relates to a flash memory management method.

Flash memory is a nonvolatile memory device that can erase or write data electrically. Compared to magnetic disk memory-based storage devices, flash memory-based storage devices consume less power and are smaller in size, and are being actively researched and developed as an alternative to magnetic disk memory. In particular, flash memory is expected to be in the spotlight as storage for mobile computing devices such as digital cameras, mobile phones, and personal digital assistants (PDAs).

However, unlike a magnetic disk memory in which a flash memory is free to overwrite data, the flash memory cannot be overwritten. To overwrite the data in the flash memory, you must first erase the data. That is, the memory cells must be returned to the initial state in which they can be written. This operation is called erasing. Erasing generally takes much longer than writing. Moreover, since the erasure is performed in units of blocks larger than the write, it may result in erasing even the portion where the write is not requested. Inevitable erasure must be restored by writing back, so in the worst case, a write request for one data requires one erase and one write for the erased unit. As such, the performance of writes is significantly lower than that of reads due to mismatches in the execution units of erase and writes, and even lower than that of magnetic disk-based storage, which inevitably incurs delays due to machine operation. Therefore, the improvement of write performance is a key technology in the design of storage devices based on flash memory.

U.S. Patent No. 5,845,313 proposes a method of constructing a linear address translation table that scans logical addresses stored in flash memory and initializes the address directly in separate RAM when the storage device is initialized. However, a large amount of RAM is required to store the address translation table. For example, to store the address translation table of a flash memory-based storage device with a total storage capacity of 32MB and a page size of 512B, 2B per 65,536 pages is required, which requires a total of 128KB of RAM. This requirement is too large for small systems with invaluable resources such as mobile devices.

U.S. Patent No. 5,404,485 proposes a method of allocating a new block (replacement block) for writing and writing to the allocated block. However, according to this, 1) new blocks are continuously allocated for writing, and thus there are a plurality of blocks of different versions in which the same page is written. That is, there is a problem that the capacity of the flash memory is considerably reduced by implicitly requiring at least one replacement block for every block. Also, 2) pages written in a new block should always be written in the same position as that of the previous block, so if frequent updates are made only for a specific page, and the remaining pages are rarely updated, only the content of a specific page is different. The content of the page is that the same replacement block is present in a plurality of waste of the storage space of the flash memory is severe, there is a problem that is not suitable for small systems such as mobile devices.

Korean Patent Laid-Open No. 2002-0092487 proposed a flash memory management technique that can freely re-think pages within a log block in order to reduce the waste of flash memory resources. When overwriting data belonging to one logical block, other logical blocks having log blocks are merged to allocate log blocks to the logical block. The merging process copies the valid data in the logical block and the log block into a new block, and then reuses the block created by the erase operation as the log block. However, there is a problem in that the data belonging to one logical block is inferior in performance due to the limitation of storing data only in one physical block (data block) and a log block allocated to the logical block.

The present invention to solve this problem is to reduce the memory requirements of the address table for re-imaging, flash memory that can efficiently manage the flash memory through the re-imaging method that can freely store the data of the logical block in the physical block It aims to provide a management method.

In addition, an object of the present invention is to provide a flash memory management method capable of managing re-imaging information in a flash memory without an additional device.

In accordance with an aspect of the present invention, there is provided a method of managing a flash memory, the method comprising: generating a logical super block including one or more logical blocks and rethinking the logical super block into a corresponding physical super block; It includes.

Preferably, pages of a logical block included in the logical super block are re-imagined into physical blocks included in the physical super block.

It is preferable to manage re-imaging information on pages of a logical block included in the logical super block together with data in the flash memory.

Checking whether re-throne information for the logical block is stored in RAM, and if re-throne information for the logical block is not in the RAM, finding a location of the rethrone information for the logical block; And reading the re-imaging information stored in the location and storing in the RAM.

When the location of the data is changed, the method may further include changing the re-throne information and storing the changed re-throne information together with the data.

Re-imaging information on the pages of the logical block included in the logical super block is preferably divided into several pages according to the tree structure.

When re-imaging information on pages of a logical block included in the logical super block is stored in the flash memory according to the tree structure, intermediate nodes of the tree structure are preferably stored together.

The physical super block includes a number of physical blocks greater than or equal to the number of logical blocks included in the logical super block, and the number of physical blocks included in the physical super block may be increased or decreased.

If there is no free page in which no data is written in the physical super block, merging the physical blocks included in the physical super block to generate a free physical block including free pages in which no data is written. desirable.

Selecting a physical super block to perform the merging, selecting a physical block to perform an erase operation among physical blocks included in the selected physical super block, and valid data stored in the selected physical block Copying to a free page in another physical block included in the physical super block, and changing the selected physical block to a free physical block to which data is not written through an erase operation after the copying; It is preferable to include the step of removing the free physical block.

Hereinafter, with reference to the accompanying drawings will be described a preferred embodiment of the present invention;

1 is a block diagram of a system based on a flash memory management method according to an exemplary embodiment.

Referring to FIG. 1, a system based on a flash memory management method according to an exemplary embodiment includes a flash memory 101, a ROM 102, a RAM 103, and a CPU 104.

The ROM 102 stores a program for controlling the flash memory 101.

The RAM 103 is used as a buffer for reading and writing rethink information and data.

The CPU 104 executes a program stored in the ROM 102.

FIG. 2 is a diagram for describing a superblock in a flash memory management method according to an exemplary embodiment.

 Referring to FIG. 2, one logical super block consists of two consecutive logical blocks, and one physical super block consists of two or more physical blocks. This will be described in more detail as follows. That is, the first logical super block 201 is composed of two logical blocks (Logical Block 0, Logical Block 1), and the second logical super block 204 is composed of two other logical blocks (Logical Block 2, Logical Block 3). ) In addition, the first physical super block 203 corresponding to the first logical super block 201 is composed of four physical blocks 1 (Physical Block 1, Physical Block 7, Physical Block 2, Physical Block 3), and the second logic. The second physical super block 205 corresponding to the super block 204 is composed of two other physical blocks (Physical Block 4 and Physical Block 0).

The number of logical blocks constituting the logical super block may vary, and the number of physical blocks constituting the physical super block corresponding to the logical super block is greater than or equal to the number of logical blocks. The logical block constituting the logical super block is fixed, but the physical block constituting the physical super block corresponding to the logical super block may be any physical block in the flash memory, and may be a physical block through the re-imaging method of the present invention described below. The position of may be changed, the number of physical blocks may be increased or decreased, and pages belonging to a logical block forming one logical super block may be located at any position of the physical block forming a physical super block corresponding to the logical super block. Can be stored in.

3 is a view for explaining a re-imaging method in a flash memory management method according to an embodiment of the present invention.

Referring to FIG. 3, a page global directory (PGD) 302, which is re-imaging information in units of logical blocks, is stored in a RAM and includes a page middle directory 303 (PMD) having re-imaging information in a logical super block. Has a location. The size of the PGD 302 is equal to the size of the logical superblock. The PMD 303 is stored in a spare area of a flash memory, which will be described later, and has four PT (Page Table, 304) positions having positions where actual data is stored. The PT 304 is stored in the spare area of the flash memory along with the PMD 303 and has 16 page addresses. Accordingly, 64 page addresses may be indicated using one PMD 303 and four PTs 304.

According to the present invention, re-imaging information on pages of a logical block included in a logical super block is managed in a flash memory. To this end, the flash memory re-imaging method according to an embodiment of the present invention 1) check whether the re-imaging information for the logical block included in the logical super block is stored in the RAM, and 2) included in the logical super block If the re-imaging information for the logical block is not in RAM, find the location of the re-imaging information for the logical block included in the logical super block, and 3) the re-imaging information for the logical block included in the logical super block stored at the found location. After reading, store in RAM, 4) if the location of the data is changed, change the re-imaging information for the logical block included in the logical super block, and 5) store the changed re-imaging information with the data.

This will be described in more detail as follows.

First, when re-imaging information of a logical block included in a logical super block is not present in RAM, the logical address may be assigned to a superblock number, a PGD number, Divide into PMD number and PTE number, find the corresponding PGD in PGD using superblock number, and read the position of corresponding PMD using PGD number. After reading the PMD from the flash memory using the location information of the PMD stored in the PGD, the location of the PT is found using the PMD number, and the PT is read from the flash memory again. From the read PT, the PTE number is used to find the physical address where the data corresponding to the logical address is stored.

Once read, the PMD 303 and the PT 304 are temporarily stored in the RAM 103. When the logical address 301 using the same PMD 303 or the PT 304 comes, the PMD 303 and the PT 304 are stored immediately without reading from the flash memory. The information stored temporarily in the RAM 103 is used. Since the size of the RAM 103 is limited, the temporarily stored PMD 303 and PT 304 are replaced with PMD 303 and PT 304 corresponding to different logical addresses. Since the present invention does not specifically define a replacement algorithm, it is also possible to use a generalized replacement algorithm such as Least Recently Used (LRU) to implement the present invention.

4 is a diagram illustrating a specific example of the re-imaging method described with reference to FIG. 3.

Referring to FIG. 4, when the size of the logical superblock is 4 and there are 64 logical pages in one logical block, when referring to a logical address having a logical block number of 17 and a logical page number of 12, the superblock number Is 4, PGD number is 1, PMD number is 0, and PTE number is 12. In the PGD, there is information that the corresponding PMD 303 is stored in the 24th page of the physical block number 6. Therefore, the spare area of the 24th page of the physical block number 6 is read in order to read the PMD. Use the PMD number to locate the PT. At this time, the PMD is always stored with one PT. Therefore, if the PT is stored together with the PMD, the information already exists in the RAM. If the corresponding PT is stored in another page as shown in FIG. 4, the PT is read from the flash memory again. Use PT number from PT to find the physical address where data is stored. Read the desired data using the physical address found.

When data with logical block number 17 and logical page number 12 attempts to overwrite, the corresponding PMD and PT are read by the method described above, and then a free page is found in which data existing in the physical super block is not written. If the free page does not exist, the physical blocks included in the physical super block are merged to generate a free physical block including free pages in which no data is written, and add the free physical block to the physical super block. To this end, 1) select a physical superblock to be merged, 2) select a physical block to perform an erase operation among physical blocks included in the selected physical superblock, and 3) validate the stored physical block. 4) Copy the data to a free page in another physical block included in the physical super block, and 4) change the selected physical block to a free physical block to which no data is written through the erase operation, and 5) free the physical super block. Remove the physical block.

After changing the value of the corresponding PT stored in RAM at the address of the free page, the corresponding PMD also stored in RAM is also changed to indicate an empty free page. Store the changed PMD and PT with the data to be overwritten on the free page. Change the entry of the PGD indicating the PMD to the newly stored address.

5 is a diagram showing the structure of re-imaging information stored with data in a flash memory according to the present invention.

Referring to FIG. 5A, the spare area stores data information DI, physical block mapping information PBMT, PMD 303, and PT 304.

Referring to FIG. 5B, the data information DI stores bad block information, an error correction code ECC for stability of data, and a logical address of data for recovery.

Referring to FIG. 5C, seven physical block numbers are recorded in the physical block mapping information PBMT.

 Referring to FIGS. 5D and 5E, in order to store physical addresses in PMDs and PTs, numbers of physical block mapping information (PBMT) and page numbers within blocks are recorded.

FIG. 6 is a diagram for describing a method of collecting pages having similar overwriting frequencies into specific physical blocks.

Referring to FIG. 6, an example of a superblock having a size of 3 and receiving two additional log blocks to play a superblock composed of five physical blocks is shown. Since the present invention does not specifically define the selection algorithm except in special cases, it is also possible to use a generalized selection algorithm such as LRU to implement the present invention.

Select a physical block including an invalid page in the selected physical super block and copy it to a free page in the physical super block. If a valid page exists in the selected physical block even after the free page is exhausted, the valid page allocates a new empty free physical block and copies it to the free physical block. After copying all valid pages, the selected physical block is deleted into an empty free physical block and then removed from the physical super block. This process is repeated until the number of physical blocks in the physical super block is equal to the number of logical blocks in the logical super block.

As described above, those skilled in the art will appreciate that the present invention can be implemented in other specific forms without changing the technical spirit or essential features. Therefore, the exemplary embodiments described above are to be understood as illustrative and not restrictive in all respects, and the scope of the present invention is indicated by the following claims rather than the detailed description, and the meaning and scope of the claims and All changes or modifications derived from the equivalent concept should be interpreted as being included in the scope of the present invention.

As described above in detail, the present invention manages a plurality of logical blocks continuously while minimizing additional resources for a flash storage device by managing a block-based reimaging table in a RAM and managing a page-based reimaging table in a flash memory. It provides a way to improve the performance of flash memory storage device by managing blocks as one super block.

In addition, there is an advantage of providing flexibility that can freely store data in multiple logical blocks belonging to a superblock in a plurality of different physical blocks.

In addition, it is possible to reduce the number of write operations and erase operations through a reproducing method utilizing flexibility, thereby extending the overall life of the flash memory having a limited number of erase operations.

Claims (10)

Generating a logical super block including one or more logical blocks; And Rethinking the logical superblock into a corresponding physical superblock; Flash memory management method comprising a. According to claim 1, And re-imaging pages of a logical block included in the logical super block into physical blocks included in the physical super block. The method of claim 2, And re-imaging information on pages of a logical block included in the logical super block in the flash memory together with data. The method of claim 3, wherein Checking whether re-imaging information for the logical block is stored in a RAM; Finding relocation information for the logical block if the re-imaging information for the logical block is not present in the RAM; Reading re-imaging information stored in the location and storing in the RAM; Flash memory management method comprising a. The method of claim 4, wherein Changing the re-imaging information when the position of data is changed; Storing the changed re-imaging information with the data; Flash memory management method further comprising. The method of claim 3, wherein Re-imaging information on the pages of the logical block included in the logical super block is divided into several pages according to the tree structure management method. The method of claim 6, And storing intermediate images of the tree structure in the flash memory according to the tree structure, when the re-imaging information on the pages of the logical block included in the logical super block is stored in the flash memory. According to claim 1, The physical super block includes a number of physical blocks greater than or equal to the number of logical blocks included in the logical super block, The number of physical blocks included in the physical super block may increase or decrease. The method of claim 8, If there is no free page in which no data is written in the physical super block, merging the physical blocks included in the physical super block to generate a free physical block including free pages in which no data is written. Memory management method. The method of claim 9, Selecting a physical super block to perform the merging; Selecting a physical block to perform an erase operation among physical blocks included in the selected physical super block; Copying valid data stored in the selected physical block to a free page in another physical block included in the physical super block; Changing the selected physical block to a free physical block in which data is not written through the erase operation after the copying; Removing the free physical block from the physical super block; Flash memory management method comprising a.

Images