CN111324574A

CN111324574A - Block mapped file system supporting dynamic management

Info

Publication number: CN111324574A
Application number: CN202010116311.XA
Authority: CN
Inventors: 王博; 王夕臣; 王波; 章诗晨
Original assignee: China Aeronautical Radio Electronics Research Institute
Current assignee: China Aeronautical Radio Electronics Research Institute
Priority date: 2020-02-25
Filing date: 2020-02-25
Publication date: 2020-06-23
Anticipated expiration: 2040-02-25
Also published as: CN111324574B

Abstract

The invention discloses a block mapping file system supporting dynamic management, which is used for managing a disk space, when the block mapping file system receives a file operation request transmitted from an upper layer, the block mapping file system firstly indexes DBI to obtain a corresponding DBI data structure, then positions the corresponding BBM data structure through a first BBM index number distributed in the DBI data structure, and determines which BBM indexes a file is composed of through a later BBM index number in the BBM data structure, thereby determining which data units DBU the file is composed of, and then updates the BBM data structure after completing read-write operation on the data units DBU; one of the DBI data structures corresponds to a file, and one of the BBM data structures corresponds to a DBU. The invention adopts the mode of the mapping table to manage the disk space, and the addition, deletion and release of the file are simple and flexible.

Description

Block mapped file system supporting dynamic management

Technical Field

The invention relates to a file system in an embedded real-time operating system, in particular to a block mapping file system supporting dynamic management in the embedded file system for reliable data storage in an airborne environment.

Background

The embedded file system is an important component of an embedded real-time operating system, and provides a method for organizing files of mass data on storage equipment or partitions. The method and the corresponding software interface are responsible for providing operations of establishing, reading, writing, retrieving, deleting files and the like for users.

The embedded file system can realize the structured management of the data files and the catalogues, determine the storage mode of the data on the storage equipment and the logical relationship between the data files, improve the space utilization efficiency of the storage equipment, complete various operations on the data files and provide a transparent file operation interface for users.

The FAT32 file system and partially customized file system are commonly used in the current airborne environment. The FAT32 file system has standard interfaces and complex data structure, and also has the problems of poor reliability, easy damage of files and the like in the actual use process; aiming at the problem that FAT32 is exposed, part of customized file systems adopt a lightweight design and have the characteristics of high reliability, non-volatility of file power failure and the like, but the file systems divide specific storage regions for data types to correspond to the specific storage regions, different types of data cannot be stored across the regions, and the use of storage space is not flexible. In the face of rapid development and higher requirements in the field of aviation, the embedded file systems cannot meet the requirements of high reliability and high flexibility at the same time.

On the premise of ensuring reliability, a Block Mapping File System (Dynamic Data Block Mapping File System) supporting Dynamic management can realize Dynamic management of disk space, is not limited to a specific area of a disk and can only store fixed types of Data, various types of Data can be flexibly stored, and the space after the File is deleted can be recycled by the File System to store other types of Data, so that the Dynamic management of the disk space is realized; the DDBM file system interface standard can be seamlessly coupled into the existing embedded real-time operating system, and the standard communication protocol can be directly connected and used without changing the interface; the DDBM file system is used as an aviation customized high-reliability file system developed aiming at an embedded environment, can meet the product support of an aviation onboard environment, and has the characteristics of safe and reliable data storage, flexible disk space management, standard interface and the like.

Disclosure of Invention

The invention aims to provide a block mapping file system supporting dynamic management, which breaks through the mode that one original data type corresponds to one data partition, uniformly manages the whole disk space, manages the disk space in a mapping table mode, is simple and flexible in file addition, deletion and disk space release, and provides an aviation embedded file system with high reliability, high flexibility and standard interfaces for the existing and future aviation onboard data storage equipment.

The invention aims to be realized by the following technical scheme:

a block mapping file system supporting dynamic management is used for managing a disk space, when the block mapping file system receives a file operation request transmitted from an upper layer, the block mapping file system firstly indexes DBI to obtain a corresponding DBI data structure, then positions the corresponding BBM data structure through a first BBM index number distributed in the DBI data structure, and determines which BBM indexes a file is composed of through a later BBM index number in the BBM data structure, thereby determining which data units DBU the file is composed of, and then updates the BBM data structure after completing read-write operation on the data units DBU; one of the DBI data structures corresponds to a file, and one of the BBM data structures corresponds to a DBU.

The invention has the beneficial effects that:

the invention greatly improves the reliability of the embedded file system in the airborne environment, effectively improves the defect that a general file system and a part of special file systems cannot simultaneously take high reliability and high flexibility into consideration, uniformly manages the whole disk space, manages the disk space by adopting a mapping table mode, and provides a key technical route map for the research and development of the high-reliability file system in the airborne embedded environment, wherein the addition, deletion and release of the file are simple and flexible.

Drawings

FIG. 1 is a partitioning diagram of a block mapped file system versus disk space.

Fig. 2 is a DBI data structure diagram of a block mapping file system.

FIG. 3 is a diagram of block mappings in a block mapping file system.

FIG. 4 is a BBM data structure diagram of a block mapped file system.

FIG. 5 is a block diagram of the relationship between a block mapped file system and an I/O system.

FIG. 6 is a hierarchical diagram of a block mapped file system.

FIG. 7 is a schematic diagram of the composition of individual files in a block mapped file system

FIG. 8 is a block map file system data software architecture diagram.

FIG. 9 is a data structure relationship diagram of a block mapped file system.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and examples.

A Block mapping file system supporting dynamic management shown in this embodiment adopts a DOS partition specification, see a partition diagram of the Block mapping file system shown in fig. 1 for a disk space, where a first sector of a disk is a master boot record mbr (main bootrecord), and on this basis, a data index area dbi (data Block index), a disk Block mapping area bbm (Block to Block map) and a data area dba (data Block operator) are added. The data area DBA is further divided into a plurality of consecutive data units DBU (data Block unit) of the same size, and the basic data unit of the Block mapping file system is DBU. The size of the DBU in the same product is fixed, and the DBU is adjustable between a range of 64MBytes to 2GBytes according to different specific applications, but is fixed in the same product.

And storing related information of the block mapping file system in the MBR, wherein the related information comprises the total number of DBI data structures, the total number of BBM data structures, magnetic disc marks, magic numbers and the like. And the related information of the file system in the MBR is read only and not written, the information is written by a ground tool when the disk is formatted, and the storage device reads the information in the initialization process of the disk insertion device. The MBR is the uppermost layer data structure, records key information of the block mapping file system, and can determine information such as DBI position, BBM position, DBI total number, BBM total number and the like. The end of the disk stores backup MBR information 2.

The DBI data structure is stored in the DBI, the backup DBI2 is stored at the tail end of the disk and is provided with a checksum, data corruption caused by abnormal power failure of equipment or other reasons is prevented, and therefore reliability of the block mapping file system is guaranteed. The DBI data structure is established under the condition that a DBI data structure is established every time a file is newly established, and folder information in the DBI data structure under the same folder is kept consistent; when the DBI data structure is newly created, the DBI space is full, which initiates a loop record, deleting the temporally oldest DBI data structure and releasing the corresponding BBM data structure and DBU space. Referring to fig. 2, each DBI data structure corresponds to a first BBM index number allocated to a file, and is used to quickly find a BBM data structure of actual content corresponding to the file. Because frequent reading and writing are not needed in the DBI data structure, the DBI data structure is directly written into a disk after being established and updated. Specific meanings in the DBI data structure are defined by table 1:

TABLE 1

The disk block mapping area BBM is a mapping of data units DBU in the DBA, as shown in fig. 3, each DBU corresponds to a BBM data structure in the BBM, the total number of the BBM data structures and the total number of the DBUs are consistent, the DBU1 corresponds to a BBM1 data structure, the DBU2 corresponds to a BBM2 data structure, and so on. The DBU only stores data, the data type stored in the DBU belongs to which file, the complete path of the file, the read-write position of the file and other key information are stored in the corresponding BBM data structure, and the index sequence corresponds to the DBU division sequence one by one. Referring to fig. 4, a BBM data structure defines a start address and an end address corresponding to a DBU, each time a block mapping file system operates the DBU, the corresponding BBM needs to be queried first to obtain key information, and after the operation is completed, the corresponding BBM information needs to be updated in time. To ensure the consistency of data, the update sequence of the BBM data structure and the DBU is defined as writing data first, writing DBU first, and then updating the BBM data structure. By the writing mode, even if the storage device is abnormally powered down, the situation that the data indicated in the BBM data structure is inconsistent with the data stored in the actual DBU can not occur. A magnetic disk block mapping area BBM is stored next to a data index area DBI, a backup BBM2 is stored at the tail end of a magnetic disk and provided with a checksum, data damage caused by abnormal power failure of equipment or other reasons is prevented, and therefore reliability of a block mapping file system is guaranteed. Specific meanings in the BBM data structure are defined by table 2:

TABLE 2

According to the capacity of 1TBytes of the disk, one DBI data structure occupies 1 sector (512 bytes), one BBM data structure occupies 1 sector (512 bytes), each data unit DBU is calculated to be 64MBytes in size, the required total capacity of the DBI is 10240 (10240 files are allowed to be established and circular recording) x 512 bytes, 5242880 bytes in total and 5 Mbytes in total, and the total capacity of the BBM is (1024x1024/64) x 512 bytes, 8388608 bytes in total and 8 Mbytes in total. It can be seen from the above calculations that the space occupied by the disk mapping management layer information is small, when the block mapping file system is initialized, this part of information will be read into the memory, because the operation is in the memory, the speed is fast, and because the whole data volume is not large, the overhead of one time of disk dropping is also within the acceptable range, the influence on the efficiency of the whole file system is very limited. In order to ensure the reliability of the file system, the DBI and BBM information can be set to be changed for many times and then be changed or be changed every time, and the strategy is flexible.

Referring to FIG. 5, the block mapped file system provides a standard set of POSIX compliant file system interfaces to users. The user can complete the operations of creating, reading, writing, deleting and the like of data on the storage device through 7 interfaces of the card (), open (), read (), write (), close (), ioctl (), remove (). The block mapping file system provides a standard I/O interface of the file system upwards, and calls a block device driver layer (XBD, CBIO) downwards to complete the operation on the actual block physical device (such as a solid state disk SSD).

Referring to fig. 6, the block map file system is largely divided into five layers from top to bottom. The first layer is a file system standard I/O layer, a block mapping file system considers the hooking problem with an embedded real-time operating system at the beginning of design, a block mapping file system interface conforms to the specification of the file system interface under the POSIX standard, belongs to the category of the file system standard I/O layer, and mainly comprises seven standard interfaces including a care (), an open (), a read (), a write (), a close (), an ioctl (), and a remove (), can seamlessly connect a mainstream embedded real-time operating system and a standard transmission protocol, and can be directly applied without changing or adjusting the interface when the operating system is changed. The second layer is a disk map management layer, and mainly relates to operations of the data index area DBI and the block map area BBM. Referring to fig. 9, when a user calls a file, the disk mapping management layer indexes the DBI data structure in the DBI area by using the file name as a condition, determines whether the file exists according to the index result, thereby determining whether to allocate a new DBI data structure or open an existing DBI data structure, finds a corresponding BBM data structure according to the first BBM index number information in the DBI, and can quickly determine which BBM data structures a file is composed of by using the previous and the next BBM index numbers in the BBM data structure. For example, in the file shown in fig. 7, the file is composed of DBU2, DBU10, DBU16, DBU30 and DBU36, and the specific file size of the file can be calculated by the write location pointer in the BBM36 data structure corresponding to the last DBU 36. The third layer is a disk logic layer, the BBM data structure determined by the disk mapping management layer is transmitted to the third layer, and a specific DBU is positioned for operation through the mapping relation between the BBM and the DBU. The fourth layer of block device driver layer is standard implementation of an operating system, and for different embedded real-time operating systems, the driving specifications of the block devices are different, such as the Vxworks real-time operating system of the wind and river company, CBIO driving specifications of version 5.5 and below, and XBD driving specifications of version 6.8 and above. The block mapping file system directly calls the drive read-write interface of the layer, and does not care about the specific implementation of the layer. The fifth layer is a physical layer of the block device, which mainly defines the specific physical form of the block device, for example, the block device is a storage device such as a NAND Flash-based solid state disk SSD or an SD card.

FIG. 8 is a block map file system data software architecture diagram. The block mapping file system provides user layer interfaces, namely usrFsCreate (), usrFsOpen (), usrFsRed (), usrFsWrite (), usrFsClose (), usrFsIoctl (), and usrFsRegove () corresponding to the file system standard I/O interface under the POSIX standard. The core layer of the block mapping file system realizes operation interfaces for DBI and BBM, which are invisible to users and are called by the seven standard interfaces. The BSP driving layer provides an actual block device read-write interface, which is realized by an operating system, and the block mapping file system is only responsible for calling. The block device driver layer belongs to the standard implementation of the operating system, and the scheme is not detailed.

Based on the specific use mode of the block mapping file system, the data files are distributed and recovered on the disk by taking the DBU as a basic data unit instead of taking sectors or clusters as a unit, the use strategy is very flexible, and the dynamic management of the disk space is realized. Since the distribution of files on the disk will no longer be continuous. The sequencing of DBUs belonging to the same file will be bi-directionally indexed by the corresponding BBM.

It should be understood that equivalents and modifications of the technical solution and inventive concept thereof may occur to those skilled in the art, and all such modifications and alterations should fall within the scope of the appended claims.

Claims

1. A block mapping file system supporting dynamic management is used for managing a disk space and is characterized in that when the block mapping file system receives a file operation request transmitted from an upper layer, the block mapping file system firstly indexes DBI to obtain a corresponding DBI data structure, then positions the corresponding BBM data structure through a first BBM index number allocated in the DBI data structure, and determines which BBM indexes a file is composed of through a later BBM index number in the BBM data structure, thereby determining which data units DBU the file is composed of, and then updates the BBM data structure after completing read-write operation on the data units DBU; one of the DBI data structures corresponds to a file, and one of the BBM data structures corresponds to a DBU.

2. The system of claim 1, wherein the disk space comprises a Master Boot Record (MBR), a data index area (DBI), a disk block mapping area (BBM), and a data area (DBA), wherein the data area (DBA) is divided into a plurality of consecutive data units (DBU) of the same size, the data index area (DBI) stores a DBI data structure, the disk block mapping area (BBM) stores a BBM data structure, and the data units (DBU) store data.

3. A block mapped file system supporting dynamic management as claimed in claim 2 wherein one DBI data structure occupies 1 sector and one BBM data structure occupies 1 sector.

4. The system of claim 2, wherein the Master Boot Record (MBR), the data index area (DBI), and the disk block mapping area (BBM) are backed up at the end of the disk.

5. A block mapped file system supporting dynamic management as claimed in claim 1 wherein when the DBI data structure is created the DBI space is full to initiate a loop record, deleting the temporally oldest DBI data structure and freeing the corresponding BBM data structure and DBU space.

6. The system of claim 1, wherein the BBM data structure and the DBU are updated in a sequence such that data is written to the DBU first and then the BBM data structure is updated.