Skip to content


Description of common structure found across various AppArmor profiles

Programs to not confine

Some programs should not be confined by themselves. For example, tools such as ls, rm, diff or cat do not have profiles in this project. Let's see why.

These are general tools that in a general context can legitimately access any file in the system. Therefore, the confinement of such tools by a global profile would at best be minimal at worst be a security theater.

It gets even worse. Let's say, we write a profile for cat. Such a profile would need access to /etc/. We will add the following rule:

  /etc/{,**} rw,

However, as /etc can contain sensitive files, we now want to explicitly prevent access to these sensitive files. Problems:

  1. How do we know the exhaustive list of sensitive files in /etc?
  2. How do we ensure access to these sensitive files is not required?
  3. This breaks the principle of mandatory access control. See the first rule of this project which is to only allow what is required. Here we allow everything and blacklist some paths.

It creates even more issues when we want to use this profile in other profiles. Let's take the example of diff. Using this rule: @{bin}/diff rPx, this will restrict access to the very generic and not very confined diff profile. Whereas most of the time, we want to restrict diff to some specific file in our profile:

  • In dpkg, an internal child profile (rCx -> diff), allows diff to only access etc config files:


profile diff {
  include <abstractions/base>
  include <abstractions/consoles>

  @{bin}/       r,
  @{bin}/pager mr,
  @{bin}/less  mr,
  @{bin}/more  mr,
  @{bin}/diff  mr,

  owner @{HOME}/.lesshs* rw,

  # Diff changed config files
  /etc/** r,

  # For shell pwd
  /root/ r,


  • As it is a dependency of pass, diff inherits the pass' profile and has the same access as the pass profile, so it will be allowed to diff password files because more than a genericdiff, it is adiff` "version" for the pass password manager:

What if I still want to protect these programs?

You do not protect these programs. Protect the usage you have of these programs. In practice, it means that you should put your terminal in a sandbox managed environment with a sandboxing tool such as Toolbox.

To sum up

  1. Do not create a profile for programs such as: rm, ls, diff, cd, cat
  2. Do not create a profile for the shell: bash, sh, dash, zsh
  3. Use Toolbox.


This project and the apparmor-profiles official project provide a large selection of abstractions to be included in profiles. They should be used.

For instance, to allow download directory access, instead of writing:

owner @{HOME}/@{XDG_DOWNLOAD_DIR}/{,**} rw,

You should write:

include <abstractions/user-download-strict>

Children profiles

Usually, a child profile is in the children group. They have the following note:


Note: This profile does not specify an attachment path because it is intended to be used only via "Px -> child-open" exec transitions from other profiles.

Here is an overview of the current children profile:

  1. child-open: To open resources. Instead of allowing the ability to run all software in @{bin}/, the purpose of this profile is to list all GUI programs that can open resources. Ultimately, only sandbox manager programs such as bwrap, snap, flatpak, firejail should be present here. Until this day, this profile will be a controlled mess.

  2. child-pager: Simple access to pagers such as pager, less and more. This profile assumes the pager is reading its data from stdin, not from a file on disk.

  3. child-systemctl: Common systemctl action. Do not use it too much as most of the time you will need more privilege than what this profile is giving you.


Chromium based browsers share a similar structure. Therefore, they share the same abstraction: abstractions/chromium that includes most of the profile content.

This abstraction requires the following variables defined in the profile header:

@{name} = chromium
@{domain} = org.chromium.Chromium
@{lib_dirs} = @{lib}/chromium
@{config_dirs} = @{user_config_dirs}/chromium
@{cache_dirs} = @{user_cache_dirs}/chromium

If your application requires chromium to run (like electron) use abstractions/chromium-common instead.

Udev rules

See the kernel docs to check the major block and char numbers used in /run/udev/data/.

Special care must be given as sometimes udev numbers are allocated dynamically by the kernel. Therefore, the full range must be allowed:


  @{run}/udev/data/c@{dynamic}:@{int} r,  # For dynamic assignment range 234 to 254, 384 to 511

No New Privileges

No New Privileges is a flag preventing a newly started program to get more privileges than its parent process. This is a good thing for security. And it is commonly used in systemd unit files (when possible). This flag also prevents transitions to other profiles because it could be less restrictive than the parent profile (no Px or Ux allowed).

The possible solutions are:

  • The easiest (and unfortunately less secure) workaround is to ensure the programs do not run with no new privileges flag by disabling NoNewPrivileges in the systemd unit (or any other options implying it).
  • Inherit the current confinement (ix)
  • Stacking