File System Interface

File Concept

如何使用大规模存储和 I/O？后来有了文件系统，对磁盘提供了抽象。

• File system presents abstraction of disk。

• File \(\rightarrow\) Track/sector

• CPU is abstracted to process

• Memory is abstracted to address space
• Storage is abstracted to file system

File is a contiguous logical space for storing information.

• data:
• character: 记事本
• binary: 一段内存的dump
• application-specific: ppt
• program
• special one: proc file system - use file-system interface to retrieve system information.

File Attributes

• Name – only information kept in human-readable form
• Identifier – unique tag (number) identifies file within file system
• Type – needed for systems that support different types
• Location – pointer to file location on device
• Size – current file size
• Protection – controls who can do reading, writing, executing
• Time, date, and user identification – data for protection, security, and usage monitoring

这些信息是目录结构 (directory structure) 的一部分，都是文件的元信息，也存在磁盘上。 可能有其他属性，例如 checksum，这些会存到 extended file attributes 里。

Linux中可以使用file命令读取文件信息，也可以使用stat命令，statistic的缩写

modify 是修改文件内容 content data 的时间，change 是修改文件 Metadata 的时间。

File Operations

• create: touch, mkdir in Linux

• space in the file system should be found

• an entry must be allocated in the directory

• open: most operations need to file to be opened first

• return a handler for other operations

• Open-file table: tracks open files
• File pointer: pointer to last read/write location, per process that has the file open
• File-open count: counter of number of times a file is open – to allow removal of data from open-file table when last processes closes it
• Disk location of the file: cache of data access information
• Access rights: per-process access mode information

• 文件可能被并发访问，我们需要锁。有 Shared lock 和 Exclusive lock，以及两种锁的机制 mandatory lock（一旦进程获取了独占锁，操作系统就阻止任何其他进程访问对应文件）和 advisory lock（进程可以自己得知锁的状态然后决定要不要坚持访问）。

• read/write: need to maintain a pointer

• reposition within file – seek

  将 current-file-position pointer 的位置重新定位到给定值，例如文件开头或结尾。

• close

• delete

• Release file space

• Hard link: maintain a counter - delete the file until the last link is deleted

• truncate: empty a file but maintains its attributes

把文件的所有 content 清空，但保留 metadata。

其他操作可以通过上面这些操作实现。如拷贝就是 create+read&write。

File Types

识别不同的文件类型：

• as part of the file names - file extension

例如规定只有扩展名是 .com, .exe, .sh 的文件才能执行。

• magic number of the file

在文件开始部分放一些 magic number 来表明文件类型。例如 7f45 4c46 是 ASCII 字符，表示 ELF，代表 elf 文件格式。

File Structure

A file can have different structures, determined by OS or program

• No structure: a stream of bytes or words
• Simple record structure
• Lines of records, fixed length or variable length
• 如数据库
• Complex structures，如word

Access Methods

• Sequential access

• a group of elements is access in a predetermined order

  每次都只能从头开始访问。

• Direct access

• access an element at an arbitrary position in a sequence in (roughly) equal time, independent of sequence size.

  可以跳到任意的位置访问，也称为随机访问。

在直接访问的方法之上，还有可能提供索引，即先在索引中得知所需访问的内容在哪里，然后去访问。也有可能使用多层索引表。

Directory structure

Disk can be subdivided into partitions

• partitions also known as minidisks, slices

• different partitions can have different file systems

一个文件系统可以有多个 disk，一个 disk 可以有多个 partition，一个 partition 又有自己的文件系统。

• disk or partition can be used raw. (without a file system)

partition 也可以不对应一个文件系统。

Directory is a collection of nodes containing information about all files. 文件名的集合

Operations Performed on Directory

• Create a file: new files need to be created and added to directory
• delete a file: remove a file from directory
• List a directory: list all files in directory
• Search for a file: pattern matching
• Traverse the file system: access every directory and file within a directory

Directory Organization

• Efficiency: to locate a file quickly
• Naming: organize the directory structure to be convenient to users

Single-Level Directory

我们设计的 directory，要能快速定位文件；要兼顾效率、便于使用、便于按一些属性聚合。

A single directory for all users:

存在 Naming problems and grouping problems，如果两个用户想用相同的文件名，无法实现。

Two-Level Directory

Separate directory for each user

• Different user can have the same name for different files
• Each user has his own user file directory (UFD), it is in the master file directory (MFD).
• Efficient to search

Tree-Structured Directories

Files organized into trees

• efficient in searching, can group files, convenient naming

如果所需目录不在当前目录，那么用户就必须提供一个路径名 (path name) 来指定。 File can be accessed using absolute or relative path name

• absolute path name: /home/alice/..
• relative path is relative to the current directory (pwd)

操作：

• Creating a new file: touch
• Delete a file: rm
• Creating a new subdirectory: mkdir <dir-name>
• Delete directory:
• If directory is empty, then it’s easy to handle
• If not
  • Option I: directory cannot be deleted, unless it’s empty
  • Option II: delete all the files, directories and sub-directories
  • sudo rm -rf /

这里不能 share 一个文件（即多个指针指向同一个文件），因为这样就会形成一个图而不是树。

Acyclic-Graph Directories

allow links to a directory entry/files for aliasing (no longer a tree)

• Dangling pointer problem:

• e.g., if delete file /dict/all, /dict/w/list and /spell/words/list are d~angling pointers.

• Solution: back pointers/reference counter

• Back pointers record all the pointers to the entity, a variable size record

• Or count # of links to it and only (physically) delete it when counter is zero

  如果一个文件被删除，那么它的 reference counter 就会减一，当减到 0 时，才真正删除。

General Graph Directory

Allowing arbitrary links may generate cycles in the directory structure.

允许目录中有环。

• allow cycles, but use garbage collection to reclaim disk spaces

如果没有外界目录指向一个环，那么就把这个环都回收了。

• every time a new link is added use a cycle detection algorithm

File System Mounting

A file system must be mounted before it can be accessed.

• mounting links a file system to the system, usually forms a single name space.
• the location of the file system being mounted is call the mount point.
• a mounted file system makes the old directory at the mount point invisible.

Mounting a file system

File Sharing

share 文件需要有一定的保护。

• User IDs identify users, allowing protections to be per-user.

允许某些用户访问。

• Group IDs allow users to be in groups, permitting group access rights.

允许某些组的用户访问。

在分布式系统里，文件可以通过网络来共享。

Protection

文件的所有者/创建者应该能控制文件可以被谁访问，能被做什么。

Types of access

• read, write, append
• execute
• delete
• list

给每个文件和目录维护一个 Access Control List (ACL)，指定每个用户及其允许的访问类型。 优点是可以提供细粒度的控制，缺点是如何构建这个列表，以及如何将这个列表存在目录里。

评论