KR100483490B1 - Doubly Journaling Store Method For Storing Data in Storage Medium - Google Patents

Abstract

In the present invention, when configuring a journaling file system for a storage medium for a hard disk or flash memory, the dual journaling to speed up the boot time of the file system and to facilitate file system recovery. (Doubly Journaling) A storage method, wherein one type of data is stored in a journaling manner from an initial position of a storage medium, and the other type of data is stored in a journaling manner from an end position of a storage medium in an initial direction. .

Description

Doubly Journaling Store Method For Storing Data in Storage Medium}

The present invention relates to a file system that operates a flash memory and a hard disk as a storage medium for storing, managing, and processing data.

With the recent development of the information industry and mobile computing technology, many personal digital assistants (PDAs), hand-held PCs (HPCs), mobile phones, e-books, etc. have been developed for the purpose of storing data therein. It is easy to carry, fast access time, and uses a lot of flash memory that requires low power.

Flash memory has different characteristics from normal random access memory (RAM). Flash memory is non-volatile and robust compared to hard disks. It can operate at low power and has fast access times similar to RAM. Its small size makes it ideal for mobile devices.

However, it is 5-10 times more expensive than a hard disk, and there is a disadvantage in that when a new data is written to an already existing space, it can be stored again after a cleaning process.

Based on the 28F640J3A model flash memory developed by Intel co., Ltd., the read speed is 100-150 nsec (nanosecond), which is similar to RAM, but the write and erase speeds are relatively high. slow. The write time using a buffer takes about 218 μsec (microseconds) when using a 32byte buffer, and the write time in an erase block takes about 0.8 sec per block. In addition, the size of a block that can be erased at one time is constant at 128 Kbytes, and the number of times that can be erased and reused at room temperature is set to about 100,000 times. The space in the flash memory that can be erased at one time is called an erase block or segment.

Flash memory can be classified into NOR, NAND, AND type according to the structure of a cell, but mainly NOR or NAND flash memory is used. NOR flash is used primarily for the purpose of storing code executed by the CPU directly connected to the memory address space due to the fast read speed and easy access per bit for random access, while NAND flash is relatively slow Because of random access time, it is mainly used for storing relatively large data such as music file or image file at once. In order to explain the method of the present invention, a description will mainly be made based on an OR type flash memory.

A conventional technique for constructing a file system using such a flash memory includes a patent entitled Flash File System of US Pat. No. 5,404,485. The A Flash Memory Based File System, published on page 155 to page 155 of the 1995 USENIX conference There is a paper titled.

In addition, the log space file system invented in Journal space release for log-structured storage systems of US Pat. No. 6,128,630 and Data storage library array with log-structured file system which allows simultaneous write and garbage collection of US Pat. No. 6,128,630. There is a prior art of a journaling flash file system (JFFS) that applies a structured file system to a flash memory.

A log system file system sequentially stores data generated when a file system is configured and stored on a hard disk. By doing so, you can maintain versions of old and newly modified data in log format, and you can recover by returning to the previous data for the current problem. JFFS uses a log-structured file system to configure the file system for flash memory and to store the data in sequence.

JFFS was developed by Axis Communications, Inc. (http://developer.axis.com/software/jffs/) in the United States, and version 2 was developed by RedHat, Inc. (http://sources.redhat.com/jffs2/). Is being developed under the GNU Public Licenses (GPL) of the Free Software Foundation (FSF).

1 is a diagram illustrating an embodiment of a storage method in JFFS2 constituting a file system in a flash memory.

Referring to the drawings, an example in which data is stored in a flash memory in the JFFS is shown. In some file systems, for example when there is a directory structure in the EXT2 file system on Linux, the way it is stored in the JFFS is a directory, as in (a) of FIG. Save the entry (Dir 1 entry). The information stored therein is the type of the directory node, the total length of the node, the cyclic redundancy check (CRC) of the head, the parent inode number, the version value, the node CRC, the name CRC, and the directory name. Immediately afterwards we store the inode for this directory. In FIG. 1, it is a Dir 1 inode. Information stored includes node type, total length, various types of CRC, version value, user ID, group ID, creation time, access time and modification time.

The above directory entries and inodes are not visible to the user, but are additional information that is used only by the file system. This is called meta data. When you save a file in a directory, you save a file entry (File 1 entry) and a file inode (File 1 inode), just like a directory. Directories and files are viewed in the same format in the file system. Directories have no actual data, and file data (File 1 data) is stored after the inode.

In this way, JFFS uses a method of journaling directories and files for flash memory. As shown in (b) of FIG. 1, when the Dir 1 inode and the File 1 inode are changed and updated to a new value, the original metadata is invalidated, and the new data is sequentially stored in the storage location. The invalidation state does not actually erase data, it merely marks it as unnecessary data. The version value of the new data is increased by one step. Therefore, if there is a problem with the new data, it is easy to recover as long as there is a previous version of the data.

When the data is updated and the new data is stored, the storage space of the flash reaches its limit and the flash memory must be erased to secure the storage space. However, since the flash memory has to be erased only in units of erase blocks, As in c), the valid data (Dir 1 entry) is moved, and when the erase block is filled with invalid spaces, the erase process is performed to erase one erase block as shown in FIG.

Garbage Collection (GC) is a process in which invalid space is collected due to lack of storage space, and then new blocks are deleted by erasing blocks. In addition, a space for storing new data created by the erasing process is called a free space. Now that you have a lot of free space, you can continue to store your data.

When using a hard disk as a storage medium, it is stored in a similar manner. However, in order to secure space, the hard disk does not move invalid data and make an invalidation block with an erase block size. However, in order to speed up access to a file on a hard disk, there are cases where multiple pieces of the same file are collected and moved to physically neighboring locations. The hard disk does not need to be erased, so if the space to store new data is invalid, save it immediately.

In the JFFS storage method, the entry and inode locations are mixed with the data in the file, so when the flash memory is connected to the operating system to mount the file system to form a file system, In order to make the directory structure in memory, the entire structure of the flash memory must be read and metadata must be found to construct the directory structure.

The same problem occurs when configuring a log file system or a journaling file system using a hard disk instead of a flash memory. You do not need to move the data to fit the erased blocks like flash memory, but if the metadata is mixed with the file data and scattered throughout the space, the access speed is 40-50 times slower than flash, and the size is over gigabyte. Configuring a file system by reading the entire disk from the server can be very time consuming.

In addition, using a log file system or a journaling file system for a hard disk is mainly used to store and play a large amount of multimedia data. However, its utility as a file system for multimedia is poor.

In order to solve this problem, the present invention separates the meta data from the general file data, and stores the file data from the first position of the storage medium, and stores the metadata in the first direction from the end position of the storage medium. The purpose is to provide.

The present invention for achieving the above object includes a method of storing the metadata in the direction from the beginning to the end of the storage medium, and the file data from the end position to the beginning. The data stored from the beginning of the storage medium and the data stored from the end to the first direction may be data of information other than metadata or file data.

The method of storing data in the dual journaling format proposed by the present invention is to store one type of data from the beginning of the storage device and the other type from the end to the front. Data stored from the beginning is called front journaling, and data stored from the end to the front is called rear journaling.

The location where the storage takes place is called the head, and the location where the erase process is performed at the back of the journaling is called the tail. That is, there is a head and tail of the first half journaling and a head and tail of the second half journaling. In addition, the point where the head meets in the first half and the second half is called the center point.

When data to be stored enters the file system, the storage medium needs space to physically store this data and to store metadata such as file entries and inodes needed to form a file for this data. In this case, the data in the file can be stored in the first half journaling, and the metadata can be stored in the second half journaling. The opposite is also true.

As described above, when the arbitrary data is to be stored in the storage medium, the dual journaling storage method of the present invention determines which data to store and starts storing. When the data is updated, the old data is invalidated and the new data is stored at the head position. When the deletion of data occurs, it is only necessary to invalidate the data.

If this storage, update, and deletion is repeated, the first and second journaling meet and center points for the storage medium. If both the first half and the second half return to the initial position and continue processing the data, then either the head of the first half or the head of the second half meets the center point first, in which case there is a lot of data to store in the first journaling. Push the points to the other side to get more journaling on the data-rich side.

In the case of flash memory, one of the purposes of using the journaling storage method is to distribute the erase count of erased blocks evenly. Moving the center point increases the journaling of the data-rich side, and thus the erase count of the first half and the second half. You can arrange evenly.

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

FIG. 2 is a diagram illustrating an embodiment in which the heads of the first half journaling and the second half journaling meet each other to determine a center point in the dual journaling storage method according to the present invention.

Referring to FIG. 2A, the first half head is denoted by H1, and the first half tail is denoted by T1. In addition, the head of the second half journaling was marked with H2 and tail as T2, and the center point was marked with C.

When data Data1 enters such a file system, it is determined whether the first half or the second half is stored in the head according to the type of data. In Fig. 2A, Data1 is the first half journaling.

2B shows a state when six data are stored. Data1 through Data4 are data for the first half journaling, and Data5 and Data6 are data for the second half journaling. In the next run, Data2 and Data6 were cleared. Due to the nature of the journaling storage method, the data is not actually erased, but only marked as invalid.

In FIG. 2C, a garbage collection (GC) is generated due to insufficient storage space for the storage medium. Since Data1 is valid data, it moves to the head and then invalidates the previous Data1, and exceeds the size of the erase block with the previously invalidated Data2 to free up space by actually performing the erase process.

This actual erasing process is performed in flash memory, and there is no need for the erasing process when using a disk. When free space is secured, the tail T1, which is the end of the first half journaling, is moved to the position of Data3. Similarly, when Data6 is invalidated and erased to the erase block size, the tail T2 of the second half of journaling is also moved.

In this state, we want to store new data Data7. This data is intended to be stored in the first half head H1 as data of first half journaling, but there is insufficient space. Therefore, as shown in (d) of FIG. 2, only D7-1, which can be stored, is stored, and the first half head H1 returns to the beginning to store the remaining D7-2.

In this way, when the first half journaling encounters the second half journaling, it returns to the original position in a circular manner and saves it. In addition, the point where the head H1 of the first half and the head H2 of the second half met is set as the center point C. FIG. The latter head H2 also returns to the end position and waits for storage. In this way, data is stored in the first half journaling and the second half journaling, and the center point (C) is defined.

If you keep storing data in this way, one of the first half head H1 or the second half head H2 reaches the position of the center point C first, in which case the first journaling part has a lot of data to store. You can judge. Therefore, the center point is moved to the other side to increase the space of journaling first arriving at the center point (C).

FIG. 3 is a diagram illustrating an embodiment of a process in which a center point is newly determined by first increasing the head of the rear half journaling and meeting the center point and moving the center point toward the front half journaling in the dual journaling storage method.

Referring to FIG. 3 (a), Data3 and Data5 are invalidated, data storage in the latter half of the journaling is active, Data8 and Data9 are stored, and data10 is about to be stored, but there is insufficient space. In the latter part, the center point (C) is reached first. Since the head H2 of the latter half does not meet the head H1 of the first half, the center point C is moved toward the first half as long as there is enough space to store the first half.

The determination that there is enough data in the first half is made by performing GC of FIG. 5. To do this, as shown in FIG. 5 (b), D7-2, which is the first half of data, is moved to the first half head H1, and the center point C is also moved. At this time, the unit of data moving in the first half is the erase block unit in the case of flash memory. Because you can move and erase to create a free space and then save new data. When free space is created, new data Data10 is stored. The movement of the center point C is moved until it meets the head H1 of the first half or until there is enough free data storage space in the first half.

4 is a flowchart illustrating an overall process of performing a dual journaling storage method according to the present invention.

Referring to the drawings, the present invention shows a process of storing data, a process of determining a center point, and a process of moving a center point through a flowchart.

Initially, the initial values of the head H1 of the first half, the head H2 of the second half, the tail T1 of the first half, the tail T2 of the second half, and the center point C are 0. (S101)

 The data storage request for the file system is made through a buffer. When the storage request (S102) is made in the storage medium in this state, it is checked whether the storage space is sufficient through the GC of FIG. 5 (S103).

If the storage medium of the storage medium is sufficient by the GC and the storage space of the storage medium is sufficient, the flow returns to the flowchart of FIG. 4 and continues. If not, the storage space is secured, and then the flow returns to the flowchart of FIG. Continue to run.

In this state, it is checked whether the center point C is 0 (S104), in order to confirm whether the center point is first determined. When the center point (C) is first determined, it is necessary to confirm that the head meets the opposing head, but when the center point is already formed, it is necessary to check whether the head meets the center point first.

When the center point C is 0 or non-O as a result of the determination of step 104, the present invention checks whether the data is the first half journaling data (S105), and stores the data (H1 + S) in the first half. do. In this state, it is checked whether the storage data H1 + S meets the rear half head H2 (S106). On the contrary, if the data H1 + S is not the first half journaling data, the second half is symmetrically performed with the first half. (S107)

If the data (H1 + S) does not exceed the rear half head (H2), it can be stored as it is (S108), if it exceeds, the possible size of the data size (S) is stored, the first half head (H1) to the first position and the rest The place where you save and meet is defined as the center point (C). When the data is stored in this way, the center point C is determined, so the center point C value is no longer zero.

In addition, in step 104, even when the center point C is not 0, it is checked whether the data is the first half journaling data as in the case of 0 (S110). If it is the first half, the data (H1 + S) is stored. In this state, check whether the stored data (H1 + S) meets the center point (C), and if the opposite data (H1 + S) is not the first half journaling data, the second half is symmetrically with the first half as in step S106. (S113)

If the data H1 + S does not exceed the center point C in step 111, the data is stored in the first half head H1 and space is secured (S112).

Conversely, if the data H1 + S exceeds the center point C in step 111, the data is stored between the first half head H1 and the center point C as much as it can be stored (S114), and then the center point C ), Move it to the second half of journaling and save.

In this state, it is checked whether there is enough space in the second half (S115). If there is enough space in the second half, the data is transferred to the second half head H2 and the space is erased (S116), and the data is stored in the erased space (S117).

On the other hand, if there is not enough space for the second half journaling in step 115, the possible space in the second half is erased, the center point C is moved by that amount, and the remaining data that cannot be stored and the first half head H1 is stored. Save to a location (S118), and when this storage process is finished, the call to the GC process again as shown in FIG.

5 is a flowchart illustrating a process of performing garbage collection (GC) as part of a dual journaling storage method.

In the process of performing GC of FIG. 5, once a request for data storage is received, GC is performed to check whether there is enough free space to store data. The data to be stored once exists in the storage buffer. If there is no free space of the data size to be stored, the data is moved by the journaling method and the invalidation space is erased as much as the erased block. After the minimum space for new data is secured, the storage is performed first. This is the case where the first partial GC is called in Fig. 4 (S101-S103 in Fig. 4).

After checking whether there is data in the buffer in a state in which the erase block free block exists (S202) for a predetermined data size (S202), if there is data in the buffer, it is checked whether the data is the first half journaling (S203). In step 203, if the data is not the first half journaling, the second half is also performed in the same manner as the first half (S204). On the contrary, if the data is the first half journaling, it is checked whether there is enough free space (S205).

In step 205, if the free space is sufficient, the GC is terminated. If the free space is not sufficient, the effective space is moved from the first half tail T1 to the first half head H1 and the invalidation space is confirmed and erased in the erase block (S206).

On the other hand, after the storage of the insertion data is performed, the GC is called once again to secure sufficient space in the storage space, in which case the number of erase blocks to be erased is determined by the graph of FIG. 6. This is when there is no data to store in the save buffer.

In this state, if the free erase block of N ₂ or more is secured in the decision graph as shown in FIG. 6, the erase process no longer occurs. N ₂ is usually set to about 10% of the total storage space.

When the number of free erasing blocks is between N ₁ and N ₂ , the number of free erasing blocks is deleted as many as the number of free erasing blocks in the graph in order to secure N ₂ or more free erasing blocks (S208).

If the number of free erased blocks is N ₁ or less in step 207, the erased blocks are secured in advance of any operation. If the number of free erased blocks is less than N ₁ even after the erase operation is performed, there is no more storage space. (S209-S211)

The reason for not getting enough free space from the first GC is that the erase process takes more time than reading or writing. When the demand for data storage comes in, the execution time for data storage ( This is because the latency time becomes longer.

In addition, since most flash memories perform a erase process and a read and write operation is required, the flash memory may suspend the erase process and perform a high priority operation.

As described above, an embodiment of the present invention describes a method of storing data in a dual journaling method when storing data in a flash memory or a hard disk.

When using the dual journaling storage method as in the embodiment, it is possible to achieve fast data access time because the same type of data is kept in a certain area, and the number of erases from the flash memory can be evenly distributed in the flash memory space.

In addition, according to the characteristics of the journaling storage method, if a data error occurs due to a power failure, data recovery to the previous version is easy, thereby ensuring the reliability of the data.

What has been described above is only one embodiment that describes a dual journaling storage method for storing data in a storage medium according to the present invention, and the present invention is not limited to the above-described embodiment, and is claimed in the following claims. Without departing from the gist of the present invention, any person having ordinary knowledge in the field of the present invention will have the technical idea of the present invention to the extent that various modifications can be made.

1 is a view showing an example of a storage method in JFFS2 constituting a file system in a general flash memory,

FIG. 2 is a diagram illustrating an embodiment in which a head of first half journaling and a second half journaling meet each other and a center point is determined in a dual journaling storage method according to the present invention;

3 is a view showing an embodiment of a process of determining a new center point by first increasing the head of the rear half journaling and then meeting the center point and moving the center point toward the front half journaling in the dual journaling storage method according to the present invention;

4 is a flowchart showing the overall performance of the dual journaling storage method of the present invention;

5 is a flowchart showing garbage collection (GC) performance as part of the dual journaling storage method of the present invention;

6 is a graph for determining the number of erased blocks in the GC of the dual journaling storage method of the present invention.

Explanation of symbols on the main parts of the drawings

H1; Half head H2; Half head

T1; first half T2; second half

C; center point

Claims (5)

A method for storing data on a storage medium, the method comprising: The dual journaling storage method for storing data on a storage medium, characterized in that for storing the data from the first position and the end position of the storage space of the storage medium toward the center. The method of claim 1, When data of different types or characteristics are stored in the storage medium, the first half journaling and the second half journaling are stored separately divided storage method for storing data on the storage medium, characterized in that. The method of claim 2, A dual journaling storage method for storing data on a storage medium, wherein the data is stored so that the first half journaling and the second half journaling meet at the center point. The method of claim 2, When the head of the first half journaling and the head of the second half journaling meet and a center point is formed first, and then, when the first half or the second half head meets the center point again for the case where the center point is formed a second time, the center point moves toward the opposite journaling Dual journaling storage method for storing data on a storage medium, characterized in that. In the journaling storage method when inserting data into the storage medium, Performing a GC when a request for data insertion is input to the storage medium to move the data and to erase the data if there is not enough storage space; And Checking whether there is enough space in the storage medium after data storage is completed to secure a space for storing the next inserted data. The dual journaling storage method of claim 2, further comprising: storing data in a storage medium.