File Systems

1
File Systems

• CSCI 4534

2
What is a file?

• A file is information that is stored on disks or
  other external media

3
Why do we use files?

• We often need for several applications or users
  to share data
• That data needs to be stored and retrieved

4
What is a file system?

• The part of the operating system that deals with
  files is called a file system
• This section discusses how files are structured,
  named, accessed, used, protected, and implemented

5
File Systems

• Consist of two parts
• A collection of files with stored data
• A directory structure
• With organization of and information about all
  the files in the system
• There may be a third part
• Partitions which separate directories

6
File Concepts

• The Operating System provides a uniform logical
  view of how information is stored
• Physical stored on nonvolatile media
• Logical named collection of related data
• Types of files
• Data numeric, alphabetic, alphanumeric, binary,
  ASCII
• Programs source, object, executable

7
File Concepts

• Meaning of a file is defined by the creator (or
  the user?)
• Files have structure according to their type
• Text sequence of characters organized into lines
• Source sequence of subprograms consisting of
  data and instructions
• Object sequence of bytes organized into blocks
  understandable by linker
• Executable binary code and data to be loaded
  into memory and executed

8
File Types

• Regular files contain user information
• ASCII good for being edited with a text editor
• Binary have an internal structure that is
  meaningful to the programs that use them
• Directories system files for maintaining the
  structure of the file system

9
File Attributes

• Typical Attributes see also Figure 6-4
• Name
• Type
• Location
• Size
• Protection and access control
• Time, date and user ID for creation, last
  modification, last use

10
File Operations

• System calls are provided for
• Create a file find space, add to directory
• Delete release file space, delete dir entry
• Open system fetches attributes and disk
  addresses
• Close space is freed
• Read name and location
• Write name and data
• Append add data to end of file
• Seek repositions file pointer

11
File Operations

• Other operations
• Get attributes see Make p. 388
• Set attributes e.g. protection modes
• Rename
• Copy

12
File Operations

• Before files are used, an open( ) system call is
  usually made
• Avoids searching for the file each time
• O.S. keeps a table about all open files
• May have a count of users if shareable
• File name is associated with index into table,
  open( ) returns a pointer (we have seen an int)
• The open( ) call can specify access-mode
  information and can check for permissions

13
File Operations

• What happens when a file (such as in UNIX) is
  shared by several users?
• The O.S. uses two tables
• A system-wide table location, access dates, size
• A per-process table
• Current file pointer for this process
• Each entry points to the system-wide table

14
File Types

• What file types should an O.S. recognize?
• See types in Figure 6-1
• File names are followed by an extension
• Extension specifies type of file and type of
  operations that can be done on the file
• .com, .bat, .exe can all be executed
• .asm is extension that an assembler is expecting

15
File Structure

• O.S.s may support different types of files
• Some may support many or a few
• However, all must support executable files
• Files can be seen as
• A sequence of 8-bit bytes
• no interpretation is made flexible
• A sequence of records
• A tree (see Figure 6-2)

16
File Structure

• How can a specific logical record in a file be
  found?
• Disk I/O uses blocks
• All blocks are the same size
• Logical records are packed into physical blocks
• In UNIX, files are simply a stream of bytes
• Those bytes are stored into a specific of bytes
  per block (eg. 512, 1024)

17
File Structure

• What determines how many logical records are in
  each block?
• Logical record size
• Physical block size
• A file, then, is a sequence of blocks
• Some part of the last block is wasted (internal
  fragmentation)

18
Access Methods

• Sequential Access
• One record is processed after the other
• Examples compilers, editors
• Mostly reading and writing is done
• Works on sequential-access devices and on
  random-access devices
• Operations read next, write next

19
Access Methods

• Random Access
• Records are accessed by key, not position
• File seen as a ed sequence of blocks or records
• Useful in databases
• Provides immediate access to large amounts of
  data
• Where to start reading?
• Each Read operation gives reading position
• Or Seek finds the first position, then reads
  sequentially from there

20
Directory Structure

• File systems are organized
• Disks are split into one or more partitions
• IBM calls them minidisks, others are called
  volumes
• Usually each disk has a partition that stores
  files and directories
• Each partition contains information about the
  files within it
• Called a device directory or a volume table of
  contents which holds the file attributes for
  files on this partition

21
Directory Structure

• The directory can be viewed as a symbol table
• Translates file names onto directory entries
• What do we want to do with directories?
• Search for a file
• Create, delete
• Rename, list files
• Traverse the file system

22
Directory Structure

• Single-Level Directory
• All files have unique names
• Two-Level Directory
• Each user has a UFD (user file directory)
• A higher-level MFD (master) is indexed by user
  name or account number

23
Directory Structure

• Hierarchical Directories
• Must specify pathname
• Absolute begins at root
• Relative begins at current directory
• Directory contains pointers to files (0) or
  subdirectory (1)
• Must have current directory, be able to change
  directories, be able to delete directories
• Allow access to other users files if given access

24
Directory Operations

• Create
• Delete
• Opendir
• Closedir
• Readdir
• Rename
• Link

25
File-System Implementation

• Layout (see Figure 6-11)
• Master Boot Record
• Boot Block
• Superblock

26
File-System Implementation

• Contiguous Allocation
• Linked List Allocation
• File Allocation Table
• I-nodes

27
Implementing Directories

• Directories provide information to find disk
  blocks
• Disk address of file contiguous
• Number of the first block linked list schemes
• Number of the i-node
• Where are attributes stored?
• In directory entry
• In the i-nodes
• See Figure 6-17 for ways to handle long file
  names in directories

28
File Sharing

• Directed Acyclic Graph (DAG)

29
File Protection

• We want to keep information safe from
• physical damage (reliability)
• improper access (protection)
• We need protection for these operations
• Read, write, execute, append, list, delete

30
File Protection

• Access control
• Use an access control list
• specifies user name and types of access allowed
• O.S. checks access list when a request is made
• lists are tough to maintain
• Three classifications
• Owner, Group, Universe
• Example of Sara