What is a File?

A file is a named collection of related information stored on secondary storage. From the user's view it is a logical, contiguous sequence of bytes; from the OS's view it is a set of disk blocks linked by metadata.

File Attributes

Every file has metadata that the OS keeps in a directory entry / inode:

• Name — human-readable identifier.
• Identifier — inode number, unique within the filesystem.
• Type — regular, directory, symbolic link, device, FIFO, socket.
• Location — pointer(s) to data blocks.
• Size — current size in bytes.
• Protection — owner, group, permission bits.
• Time stamps — created, modified, accessed.

File Operations (System Calls)

File Descriptor Table

When a process opens a file, the kernel assigns a small integer file descriptor (fd). Three are pre-opened: 0 stdin, 1 stdout, 2 stderr. The fd indexes a per-process table that points to system-wide open-file entries.

File Types in Linux (ls -l first character)

Internal File Structure

• None (byte stream) — Unix philosophy; the program imposes structure (e.g., text lines).
• Simple records — fixed/variable-length lines.
• Complex — formatted documents, executables (ELF, PE).

File Sharing

Multiple processes can share a file. Issues:

• Locking — flock/fcntl for exclusive ranges.
• Open across fork — children inherit parent's fds.
• Network sharing — NFS, SMB.

Summary

• Files are named byte streams with metadata in inodes.
• The file-descriptor + open-file + inode triple decouples per-process state from the on-disk data.
• "Everything is a file" extends the abstraction to devices, pipes and sockets.