Everything is a file in UNIX ?

You most likely heard that everything in UNIX is a file, but it never really clicked for you. Or you get it, but couldn't find the technical details behind it. Or you have no clues about what this even means - like me before this.

In any cases, you are at the right place. Got your coffee ? ☕ Let's dive in.

Table of contents

• Is everything really a file ?
• What is a file, if everything is a file ?
• Inode
  • Basics
  • Inode number
• Directories
• Data

Is everything really a file ?

Yup.

This yet simple but quite complex sentence resume the entire architecture of the UNIX Operating System (OS).

Basically, every features your OS is able to do, rely on files. Processes, devices (such as a printer, or your mouse), networks, and even directories.

Wait, directories ?

Yup. There are 7 different types of files, you can find a complete list here. For now, we'll focus on regular files, and directories.

What is a file, if everything is a file ?

From an OS point of view, a file is actually nothing more than a specific data structure called an inode. You may visualize an inode such as a table containaing file related informations.

The inode itself contain all the metadatas (extra datas that describes data) relatives to the file it describe. Informations such as the creation date, the last update date, the ownership, the permissions, the file size ...

⚠ The file name and the actual file data are not part of the inode, keep it in mind, we'll cover this later.

Inode

Basics

You can visualize an inode informations directly in your terminal, by typing the ls -l command.

Here's the kind of output you should receive :

-rw-r--r--  1 user  root  13 10 mai 22:10 index.txt

In order :

• File type and permissions. It's quite hard to read, let's break it down together.

  • The first character specify the what kind of file this is (the 7 types we mentionned earlier).
    • - Regular file.
    • d Directory.
    • l Symbolic link.
    • b Block special file.
    • c Character special file.
    • s Socket link.
    • p FIFO.
  • The next three characters are related to the owner permissions for the file.
  • The next three characters are related to the group permissions for the file.

  • The last three characters are related to the others permissions for the file.

    All permissions fields can be read as follow :

    • Is the permission allowed to read the file ? - for no, r for yes.
    • Is the permission allowed to write the file ? - for no, w for yes.
    • Is the permission allowed to execute the file ? - for no, x for yes.

• Number of hard links. You may find more informations here.

• Owner name.

• Group name.

• Number of bytes in your file.

• Date of last modification.

• File name. Not a part of the inode, but still in the output. More on it later.

📌 You may find all these informations - and more, directly from your terminal in an UNIX environnement with the command man ls

If you're willing to get a bit more informations about a specific file, you may want to use the stat command directly.

Inode number

If you execute the ls -li command you will find an extra column at the beginning of the output.

12345678 -rw-r--r--  1 user  root  13 10 mai 22:10 index.txt

It contain a very specific integer, the inode number.

💡 This number is the unique identifier for this specific file in your entire file system.

Every files will have a different inode number (with the exception of hard links), and there's a maximum amount of inode number your file system can handle. Afterward, you will not be able to create anymore files.

File name and directories

If the filename is not specified by the inode, where is it stored ?

You most likely already guessed it by the title, but the file name is stored within the parent directory.

What exactly is a directory then ? It's actually a pretty simple thing. Type ls -l command somewhere with a directory.

drw-r--r--  1 user  root  13 10 mai 23:17 directoryName

Noticed the d at the beginning ?

The directories files actually only contain a mapping table, between a file inode number, and his name.

The list goes on for all the file or other directories it may contain.

A directory is nothing but a specific file, it also have an inode number, and his name is saved within his parent directory inode.

📌 ls -li for a directory to see the inode number

899 drw-r--r--  1 user  root  13 10 mai 23:17 directoryName

Here's a complete representation of the content of a directory.

💡 The first two entries are references to itself, and to his parent directory.
📌 To see them, use the -a option on the ls command to see hidden files.

The file data

So, where is the file data ?

What make the inode so special is that it kept references (pointers) toward the memory blocks that are actually containing the data in disk. By doing so, when we ask to open the file, it go through all of them and recover the informations needed.

The pointers part is oversimplified for the sake of this article, indirect pointers are purposely omitted.

Let's get a visual recap for a file inode.

I hope you enjoyed the read, and learned as much as I did. If you have any questions or comments, feel free to reach to me on Twitter or in the comments below. Have a nice day !