Provided by: firejail_0.9.72-2_amd64 
NAME
Firejail - Linux namespaces sandbox program
SYNOPSIS
Start a sandbox:
firejail [OPTIONS] [program and arguments]
Start an AppImage program:
firejail [OPTIONS] --appimage [OPTIONS] [appimage-file and arguments]
File transfer from an existing sandbox
firejail {--ls | --get | --put | --cat} dir_or_filename
Network traffic shaping for an existing sandbox:
firejail --bandwidth={name|pid} bandwidth-command
Monitoring:
firejail {--list | --netstats | --top | --tree}
Miscellaneous:
firejail {-? | --debug-caps | --debug-errnos | --debug-syscalls | --debug-
syscalls32 | --debug-protocols | --help | --version}
DESCRIPTION
Firejail is a SUID sandbox program that reduces the risk of security breaches by
restricting the running environment of untrusted applications using Linux namespaces,
seccomp-bpf and Linux capabilities. It allows a process and all its descendants to have
their own private view of the globally shared kernel resources, such as the network stack,
process table, mount table. Firejail can work in a SELinux or AppArmor environment, and
it is integrated with Linux Control Groups.
Written in C with virtually no dependencies, the software runs on any Linux computer with
a 3.x kernel version or newer. It can sandbox any type of processes: servers, graphical
applications, and even user login sessions.
Firejail allows the user to manage application security using security profiles. Each
profile defines a set of permissions for a specific application or group of applications.
The software includes security profiles for a number of more common Linux programs, such
as Mozilla Firefox, Chromium, VLC, Transmission etc.
Firejail is currently implemented as an SUID binary, which means that if a malicious or
compromised user account manages to exploit a bug in Firejail, that could ultimately lead
to a privilege escalation to root. To mitigate this, it is recommended to only allow
trusted users to run firejail (see firejail-users(5) for details on how to achieve that).
For more details on the security/usability tradeoffs of Firejail, see: #4601
⟨https://github.com/netblue30/firejail/discussions/4601⟩
Alternative sandbox technologies like snap (https://snapcraft.io/) and flatpak
(https://flatpak.org/) are not supported. Snap and flatpak packages have their own native
management tools and will not work when sandboxed with Firejail.
USAGE
Without any options, the sandbox consists of a filesystem build in a new mount namespace,
and new PID and UTS namespaces. IPC, network and user namespaces can be added using the
command line options. The default Firejail filesystem is based on the host filesystem with
the main system directories mounted read-only. These directories are /etc, /var, /usr,
/bin, /sbin, /lib, /lib32, /libx32 and /lib64. Only /home and /tmp are writable.
Upon execution Firejail first looks in ~/.config/firejail/ for a profile and if it doesn't
find one, it looks in /etc/firejail/. For profile resolution detail see
https://github.com/netblue30/firejail/wiki/Creating-Profiles#locations-and-types. If an
appropriate profile is not found, Firejail will use a default profile. The default
profile is quite restrictive. In case the application doesn't work, use --noprofile option
to disable it. For more information, please see SECURITY PROFILES section below.
If a program argument is not specified, Firejail starts the user's preferred shell.
Examples:
$ firejail [OPTIONS] # starting the program specified in $SHELL, usually
/bin/bash
$ firejail [OPTIONS] firefox # starting Mozilla Firefox
# sudo firejail [OPTIONS] /etc/init.d/nginx start
OPTIONS
-- Signal the end of options and disables further option processing.
--allow-debuggers
Allow tools such as strace and gdb inside the sandbox by whitelisting system calls
ptrace and process_vm_readv. This option is only available when running on Linux
kernels 4.8 or newer - a kernel bug in ptrace system call allows a full bypass of
the seccomp filter.
Example:
$ firejail --allow-debuggers --profile=/etc/firejail/firefox.profile strace -f
firefox
--allusers
All directories under /home are visible inside the sandbox. By default, only
current user home directory is visible.
Example:
$ firejail --allusers
--apparmor
Enable AppArmor confinement with the "firejail-default" AppArmor profile. For more
information, please see APPARMOR section below.
--apparmor=profile_name
Enable AppArmor confinement with a custom AppArmor profile. Note that profile in
question must already be loaded into the kernel. For more information, please see
APPARMOR section below.
--apparmor.print=name|pid
Print the AppArmor confinement status for the sandbox identified by name or by PID.
Example:
$ firejail --apparmor.print=browser
5074:netblue:/usr/bin/firejail /usr/bin/firefox-esr
AppArmor: firejail-default enforce
--appimage
Sandbox an AppImage (https://appimage.org/) application. If the sandbox is started
as a regular user, nonewprivs and a default capabilities filter are enabled.
private-bin and private-lib are disabled by default when running appimages.
Example:
$ firejail --appimage --profile=krita krita-3.0-x86_64.appimage
$ firejail --quiet --appimage --private --profile=krita krita-3.0-x86_64.appimage
$ firejail --appimage --net=none --x11 --profile=krita krita-3.0-x86_64.appimage
Note: When using both --appimage and --profile, it is recommended to always specify
the former before the latter, so that any ?HAS_APPIMAGE conditionals inside of the
profile evaluate to true (see ?CONDITIONAL in firejail-profile(5)).
--bandwidth=name|pid
Set bandwidth limits for the sandbox identified by name or PID, see TRAFFIC SHAPING
section for more details.
--bind=filename1,filename2
Mount-bind filename1 on top of filename2. This option is only available when
running as root.
Example:
# firejail --bind=/config/etc/passwd,/etc/passwd
--blacklist=dirname_or_filename
Blacklist directory or file. File globbing is supported, see FILE GLOBBING section
for more details.
Symbolic link handling: Blacklisting a path that is a symbolic link will also
blacklist the path that it points to. For example, if ~/foo is blacklisted and it
points to /bar, then /bar will also be blacklisted.
Example:
$ firejail --blacklist=/sbin --blacklist=/usr/sbin
$ firejail --blacklist=~/.mozilla
$ firejail "--blacklist=/home/username/My Virtual Machines"
$ firejail --blacklist=/home/username/My\ Virtual\ Machines
--build
The command builds a whitelisted profile. The profile is printed on the screen. The
program is run in a very relaxed sandbox, with only --caps.drop=all and
--seccomp=!chroot. Programs that raise user privileges are not supported.
Example:
$ firejail --build vlc ~/Videos/test.mp4
$ firejail --build --appimage ~/Downloads/Subsurface.AppImage
--build=profile-file
The command builds a whitelisted profile, and saves it in profile-file. The program
is run in a very relaxed sandbox, with only --caps.drop=all and --seccomp=!chroot.
Programs that raise user privileges are not supported.
Example:
$ firejail --build=vlc.profile vlc ~/Videos/test.mp4
$ firejail --build=Subsurface.profile --appimage ~/Downloads/Subsurface.AppImage
-c Login shell compatibility option. This option is use by some login programs when
executing the login shell, such as when firejail is used as a restricted login
shell. It currently does not change the execution of firejail.
--caps Linux capabilities is a kernel feature designed to split up the root privilege into
a set of distinct privileges. These privileges can be enabled or disabled
independently, thus restricting what a process running as root can do in the
system. See capabilities(7) for details.
By default root programs run with all capabilities enabled. --caps option disables
the following capabilities: CAP_SYS_MODULE, CAP_SYS_RAWIO, CAP_SYS_BOOT,
CAP_SYS_NICE, CAP_SYS_TTY_CONFIG, CAP_SYSLOG, CAP_MKNOD, CAP_SYS_ADMIN. The filter
is applied to all processes started in the sandbox.
Example:
$ sudo firejail --caps /etc/init.d/nginx start
--caps.drop=all
Drop all capabilities for the processes running in the sandbox. This option is
recommended for running GUI programs or any other program that doesn't require root
privileges. It is a must-have option for sandboxing untrusted programs installed
from unofficial sources - such as games, Java programs, etc.
Example:
$ firejail --caps.drop=all warzone2100
--caps.drop=capability,capability,capability
Define a custom blacklist Linux capabilities filter.
Example:
$ firejail --caps.drop=net_broadcast,net_admin,net_raw
--caps.keep=capability,capability,capability
Define a custom whitelist Linux capabilities filter.
Example:
$ sudo firejail --caps.keep=chown,net_bind_service,setgid,\ setuid
/etc/init.d/nginx start
--caps.print=name|pid
Print the caps filter for the sandbox identified by name or by PID.
Example:
$ firejail --name=mygame --caps.drop=all warzone2100 &
$ firejail --caps.print=mygame
Example:
$ firejail --list
3272:netblue::firejail --private firefox
$ firejail --caps.print=3272
--cat=name|pid filename
Print content of file from sandbox container, see FILE TRANSFER section for more
details.
--chroot=dirname
Chroot the sandbox into a root filesystem. Unlike the regular filesystem container,
the system directories are mounted read-write. If the sandbox is started as a
regular user, nonewprivs and a default capabilities filter are enabled.
Example:
$ firejail --chroot=/media/ubuntu warzone2100
For automatic mounting of X11 and PulseAudio sockets set environment variables
FIREJAIL_CHROOT_X11 and FIREJAIL_CHROOT_PULSE.
Note: Support for this command is controlled in firejail.config with the chroot
option.
--cpu=cpu-number,cpu-number,cpu-number
Set CPU affinity.
Example:
$ firejail --cpu=0,1 handbrake
--cpu.print=name|pid
Print the CPU cores in use by the sandbox identified by name or by PID.
Example:
$ firejail --name=mygame --caps.drop=all warzone2100 &
$ firejail --cpu.print=mygame
Example:
$ firejail --list
3272:netblue::firejail --private firefox
$ firejail --cpu.print=3272
--dbus-log=file
Specify the location for the DBus log file.
The log file contains events for both the system and session buses if both of the
--dbus-system.log and --dbus-user.log options are specified. If no log file path is
given, logs are written to the standard output instead.
Example:
$ firejail --dbus-system=filter --dbus-system.log \
--dbus-log=dbus.txt
--dbus-system=filter|none
Set system DBus sandboxing policy.
The filter policy enables the system DBus filter. This option requires installing
the xdg-dbus-proxy utility. Permissions for well-known can be specified with the
--dbus-system.talk and --dbus-system.own options.
The none policy disables access to the system DBus.
Only the regular system DBus UNIX socket is handled by this option. To disable the
abstract sockets (and force applications to use the filtered UNIX socket) you would
need to request a new network namespace using --net command. Another option is to
remove unix from the --protocol set.
Example:
$ firejail --dbus-system=none
--dbus-system.broadcast=name=[member][@path]
Allows the application to receive broadcast signals from the indicated interface
member at the indicated object path exposed by the indicated bus name on the system
DBus. The name may have a .* suffix to match all names underneath it, including
itself. The interface member may have a .* to match all members of an interface,
or be * to match all interfaces. The path may have a /* suffix to indicate all
objects underneath it, including itself. Omitting the interface member or the
object path will match all members and object paths, respectively.
Example:
$ firejail --dbus-system=filter --dbus-system.broadcast=\
org.freedesktop.Notifications=\
org.freedesktop.Notifications.*@/org/freedesktop/Notifications
--dbus-system.call=name=[member][@path]
Allows the application to call the indicated interface member at the indicated
object path exposed by the indicated bus name on the system DBus. The name may
have a .* suffix to match all names underneath it, including itself. The interface
member may have a .* to match all members of an interface, or be * to match all
interfaces. The path may have a /* suffix to indicate all objects underneath it,
including itself. Omitting the interface member or the object path will match all
members and object paths, respectively.
Example:
$ firejail --dbus-system=filter --dbus-system.call=\
org.freedesktop.Notifications=\
org.freedesktop.Notifications.*@/org/freedesktop/Notifications
--dbus-system.log
Turn on DBus logging for the system DBus. This option requires --dbus-
system=filter.
Example:
$ firejail --dbus-system=filter --dbus-system.log
--dbus-system.own=name
Allows the application to own the specified well-known name on the system DBus.
The name may have a .* suffix to match all names underneath it, including itself
(e.g. "foo.bar.*" matches "foo.bar", "foo.bar.baz" and "foo.bar.baz.quux", but not
"foobar").
Example:
$ firejail --dbus-system=filter --dbus-system.own=\
org.gnome.ghex.*
--dbus-system.see=name
Allows the application to see, but not talk to the specified well-known name on the
system DBus. The name may have a .* suffix to match all names underneath it,
including itself (e.g. "foo.bar.*" matches "foo.bar", "foo.bar.baz" and
"foo.bar.baz.quux", but not "foobar").
Example:
$ firejail --dbus-system=filter --dbus-system.see=\
org.freedesktop.Notifications
--dbus-system.talk=name
Allows the application to talk to the specified well-known name on the system DBus.
The name may have a .* suffix to match all names underneath it, including itself
(e.g. "foo.bar.*" matches "foo.bar", "foo.bar.baz" and "foo.bar.baz.quux", but not
"foobar").
Example:
$ firejail --dbus-system=filter --dbus-system.talk=\
org.freedesktop.Notifications
--dbus-user=filter|none
Set session DBus sandboxing policy.
The filter policy enables the session DBus filter. This option requires installing
the xdg-dbus-proxy utility. Permissions for well-known names can be added with the
--dbus-user.talk and --dbus-user.own options.
The none policy disables access to the session DBus.
Only the regular session DBus UNIX socket is handled by this option. To disable the
abstract sockets (and force applications to use the filtered UNIX socket) you would
need to request a new network namespace using --net command. Another option is to
remove unix from the --protocol set.
Example:
$ firejail --dbus-user=none
--dbus-user.broadcast=name=[member][@path]
Allows the application to receive broadcast signals from the indicated interface
member at the indicated object path exposed by the indicated bus name on the
session DBus. The name may have a .* suffix to match all names underneath it,
including itself. The interface member may have a .* to match all members of an
interface, or be * to match all interfaces. The path may have a /* suffix to
indicate all objects underneath it, including itself. Omitting the interface
member or the object path will match all members and object paths, respectively.
Example:
$ firejail --dbus-user=filter --dbus-user.broadcast=\
org.freedesktop.Notifications=\
org.freedesktop.Notifications.*@/org/freedesktop/Notifications
--dbus-user.call=name=[member][@path]
Allows the application to call the indicated interface member at the indicated
object path exposed by the indicated bus name on the session DBus. The name may
have a .* suffix to match all names underneath it, including itself. The interface
member may have a .* to match all members of an interface, or be * to match all
interfaces. The path may have a /* suffix to indicate all objects underneath it,
including itself. Omitting the interface member or the object path will match all
members and object paths, respectively.
Example:
$ firejail --dbus-user=filter --dbus-user.call=\
org.freedesktop.Notifications=\
org.freedesktop.Notifications.*@/org/freedesktop/Notifications
--dbus-user.log
Turn on DBus logging for the session DBus. This option requires --dbus-user=filter.
Example:
$ firejail --dbus-user=filter --dbus-user.log
--dbus-user.own=name
Allows the application to own the specified well-known name on the session DBus.
The name may have a .* suffix to match all names underneath it, including itself
(e.g. "foo.bar.*" matches "foo.bar", "foo.bar.baz" and "foo.bar.baz.quux", but not
"foobar").
Example:
$ firejail --dbus-user=filter --dbus-user.own=org.gnome.ghex.*
--dbus-user.talk=name
Allows the application to talk to the specified well-known name on the session
DBus. The name may have a .* suffix to match all names underneath it, including
itself (e.g. "foo.bar.*" matches "foo.bar", "foo.bar.baz" and "foo.bar.baz.quux",
but not "foobar").
Example:
$ firejail --dbus-user=filter --dbus-user.talk=\
org.freedesktop.Notifications
--dbus-user.see=name
Allows the application to see, but not talk to the specified well-known name on the
session DBus. The name may have a .* suffix to match all names underneath it,
including itself (e.g. "foo.bar.*" matches "foo.bar", "foo.bar.baz" and
"foo.bar.baz.quux", but not "foobar").
Example:
$ firejail --dbus-user=filter --dbus-user.see=\
org.freedesktop.Notifications
--debug
Print debug messages.
Example:
$ firejail --debug firefox
--debug-blacklists
Debug blacklisting.
Example:
$ firejail --debug-blacklists firefox
--debug-caps
Print all recognized capabilities in the current Firejail software build and exit.
Example:
$ firejail --debug-caps
--debug-errnos
Print all recognized error numbers in the current Firejail software build and exit.
Example:
$ firejail --debug-errnos
--debug-private-lib
Debug messages for --private-lib option.
--debug-protocols
Print all recognized protocols in the current Firejail software build and exit.
Example:
$ firejail --debug-protocols
--debug-syscalls
Print all recognized system calls in the current Firejail software build and exit.
Example:
$ firejail --debug-syscalls
--debug-syscalls32
Print all recognized 32 bit system calls in the current Firejail software build and
exit.
--debug-whitelists
Debug whitelisting.
Example:
$ firejail --debug-whitelists firefox
--defaultgw=address
Use this address as default gateway in the new network namespace.
Example:
$ firejail --net=eth0 --defaultgw=10.10.20.1 firefox
--deterministic-exit-code
Always exit firejail with the first child's exit status. The default behavior is to
use the exit status of the final child to exit, which can be nondeterministic.
--deterministic-shutdown
Always shut down the sandbox after the first child has terminated. The default
behavior is to keep the sandbox alive as long as it contains running processes.
--disable-mnt
Blacklist /mnt, /media, /run/mount and /run/media access.
Example:
$ firejail --disable-mnt firefox
--dns=address
Set a DNS server for the sandbox. Up to three DNS servers can be defined. Use this
option if you don't trust the DNS setup on your network.
Example:
$ firejail --dns=8.8.8.8 --dns=8.8.4.4 firefox
Note: this feature is not supported on systemd-resolved setups.
--dns.print=name|pid
Print DNS configuration for a sandbox identified by name or by PID.
Example:
$ firejail --name=mygame --caps.drop=all warzone2100 &
$ firejail --dns.print=mygame
Example:
$ firejail --list
3272:netblue::firejail --private firefox
$ firejail --dns.print=3272
--dnstrace[=name|pid]
Monitor DNS queries. The sandbox can be specified by name or pid. Only networked
sandboxes created with --net are supported. This option is only available when
running the sandbox as root.
Without a name/pid, Firejail will monitor the main system network namespace.
Example:
$ sudo firejail --dnstrace
11:31:43 9.9.9.9 linux.com (type 1)
11:31:45 9.9.9.9 fonts.googleapis.com (type 1) NXDOMAIN
11:31:45 9.9.9.9 js.hs-scripts.com (type 1) NXDOMAIN
11:31:45 9.9.9.9 www.linux.com (type 1)
11:31:45 9.9.9.9 fonts.googleapis.com (type 1) NXDOMAIN
11:31:52 9.9.9.9 js.hs-scripts.com (type 1) NXDOMAIN
11:32:05 9.9.9.9 secure.gravatar.com (type 1)
11:32:06 9.9.9.9 secure.gravatar.com (type 1)
11:32:08 9.9.9.9 taikai.network (type 1)
11:32:08 9.9.9.9 cdn.jsdelivr.net (type 1)
11:32:08 9.9.9.9 taikai.azureedge.net (type 1)
11:32:08 9.9.9.9 www.youtube.com (type 1)
--env=name=value
Set environment variable in the new sandbox.
Example:
$ firejail --env=LD_LIBRARY_PATH=/opt/test/lib
--fs.print=name|pid
Print the filesystem log for the sandbox identified by name or by PID.
Example:
$ firejail --name=mygame --caps.drop=all warzone2100 &
$ firejail --fs.print=mygame
Example:
$ firejail --list
3272:netblue::firejail --private firefox
$ firejail --fs.print=3272
--get=name|pid filename
Get a file from sandbox container, see FILE TRANSFER section for more details.
-?, --help
Print options end exit.
--hostname=name
Set sandbox hostname.
Example:
$ firejail --hostname=officepc firefox
--hosts-file=file
Use file as /etc/hosts.
Example:
$ firejail --hosts-file=~/myhosts firefox
--ids-check
Check file hashes previously generated by --ids-check. See INTRUSION DETECTION
SYSTEM section for more details.
Example:
$ firejail --ids-check
--ids-init
Initialize file hashes. See INTRUSION DETECTION SYSTEM section for more details.
Example:
$ firejail --ids-init
--ignore=command
Ignore command in profile file.
Example:
$ firejail --ignore=seccomp --ignore=caps firefox
$ firejail --ignore="net eth0" firefox
--icmptrace[=name|pid]
Monitor ICMP traffic. The sandbox can be specified by name or pid. Only networked
sandboxes created with --net are supported. This option is only available when
running the sandbox as root.
Without a name/pid, Firejail will monitor the main system network namespace.
Example
$ sudo firejail --icmptrace
20:53:54 192.168.1.60 -> 142.250.65.174 - 98 bytes - Echo request/0
20:53:54 142.250.65.174 -> 192.168.1.60 - 98 bytes - Echo reply/0
20:53:55 192.168.1.60 -> 142.250.65.174 - 98 bytes - Echo request/0
20:53:55 142.250.65.174 -> 192.168.1.60 - 98 bytes - Echo reply/0
20:53:55 192.168.1.60 -> 1.1.1.1 - 154 bytes - Destination unreachable/Port
unreachable
--include=file.profile
Include a profile file before the regular profiles are used.
Example:
$ firejail --include=/etc/firejail/disable-devel.inc gedit
--interface=interface
Move interface in a new network namespace. Up to four --interface options can be
specified. Note: wlan devices are not supported for this option.
Example:
$ firejail --interface=eth1 --interface=eth0.vlan100
--ip=address
Assign IP addresses to the last network interface defined by a --net option. A
default gateway is assigned by default.
Example:
$ firejail --net=eth0 --ip=10.10.20.56 firefox
--ip=none
No IP address and no default gateway are configured for the last interface defined
by a --net option. Use this option in case you intend to start an external DHCP
client in the sandbox.
Example:
$ firejail --net=eth0 --ip=none
If the corresponding interface doesn't have an IP address configured, this option
is enabled by default.
--ip=dhcp
Acquire an IP address and default gateway for the last interface defined by a --net
option, as well as set the DNS servers according to the DHCP response. This option
requires the ISC dhclient DHCP client to be installed and will start it
automatically inside the sandbox.
Example:
$ firejail --net=br0 --ip=dhcp
This option should not be used in conjunction with the --dns option if the DHCP
server is set to configure DNS servers for the clients, because the manually
specified DNS servers will be overwritten.
The DHCP client will NOT release the DHCP lease when the sandbox terminates. If
your DHCP server requires leases to be explicitly released, consider running a DHCP
client and releasing the lease manually in conjunction with the --net=none option.
--ip6=address
Assign IPv6 addresses to the last network interface defined by a --net option.
Example:
$ firejail --net=eth0 --ip6=2001:0db8:0:f101::1/64 firefox
Note: you don't need this option if you obtain your ip6 address from router via
SLAAC (your ip6 address and default route will be configured by kernel
automatically).
--ip6=dhcp
Acquire an IPv6 address and default gateway for the last interface defined by a
--net option, as well as set the DNS servers according to the DHCP response. This
option requires the ISC dhclient DHCP client to be installed and will start it
automatically inside the sandbox.
Example:
$ firejail --net=br0 --ip6=dhcp
This option should not be used in conjunction with the --dns option if the DHCP
server is set to configure DNS servers for the clients, because the manually
specified DNS servers will be overwritten.
The DHCP client will NOT release the DHCP lease when the sandbox terminates. If
your DHCP server requires leases to be explicitly released, consider running a DHCP
client and releasing the lease manually.
--iprange=address,address
Assign an IP address in the provided range to the last network interface defined by
a --net option. A default gateway is assigned by default.
Example:
$ firejail --net=eth0 --iprange=192.168.1.100,192.168.1.150
--ipc-namespace
Enable a new IPC namespace if the sandbox was started as a regular user. IPC
namespace is enabled by default for sandboxes started as root.
Example:
$ firejail --ipc-namespace firefox
--join=name|pid
Join the sandbox identified by name or by PID. By default a /bin/bash shell is
started after joining the sandbox. If a program is specified, the program is run
in the sandbox. If --join command is issued as a regular user, all security filters
are configured for the new process the same they are configured in the sandbox. If
--join command is issued as root, the security filters and cpus configurations are
not applied to the process joining the sandbox.
Example:
$ firejail --name=mygame --caps.drop=all warzone2100 &
$ firejail --join=mygame
Example:
$ firejail --list
3272:netblue::firejail --private firefox
$ firejail --join=3272
--join-filesystem=name|pid
Join the mount namespace of the sandbox identified by name or PID. By default a
/bin/bash shell is started after joining the sandbox. If a program is specified,
the program is run in the sandbox. This command is available only to root user.
Security filters and cpus configurations are not applied to the process joining the
sandbox.
--join-network=name|pid
Join the network namespace of the sandbox identified by name. By default a
/bin/bash shell is started after joining the sandbox. If a program is specified,
the program is run in the sandbox. This command is available only to root user.
Security filters and cpus configurations are not applied to the process joining the
sandbox. Example:
# start firefox
$ firejail --net=eth0 --name=browser firefox &
# change netfilter configuration
$ sudo firejail --join-network=browser bash -c "cat /etc/firejail/nolocal.net |
/sbin/iptables-restore"
# verify netfilter configuration
$ sudo firejail --join-network=browser /sbin/iptables -vL
# verify IP addresses
$ sudo firejail --join-network=browser ip addr
Switching to pid 1932, the first child process inside the sandbox
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: eth0-1931: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state
UNKNOWN group default
link/ether 76:58:14:42:78:e4 brd ff:ff:ff:ff:ff:ff
inet 192.168.1.158/24 brd 192.168.1.255 scope global eth0-1931
valid_lft forever preferred_lft forever
inet6 fe80::7458:14ff:fe42:78e4/64 scope link
valid_lft forever preferred_lft forever
--join-or-start=name
Join the sandbox identified by name or start a new one. Same as "firejail
--join=name" if sandbox with specified name exists, otherwise same as "firejail
--name=name ..."
Note that in contrary to other join options there is respective profile option.
--keep-config-pulse
Disable automatic ~/.config/pulse init, for complex setups such as remote pulse
servers or non-standard socket paths.
Example:
$ firejail --keep-config-pulse firefox
--keep-dev-shm
/dev/shm directory is untouched (even with --private-dev)
Example:
$ firejail --keep-dev-shm --private-dev
--keep-fd=all
Inherit all open file descriptors to the sandbox. By default only file descriptors
0, 1 and 2 are inherited to the sandbox, and all other file descriptors are closed.
Example:
$ firejail --keep-fd=all
--keep-fd=file_descriptor
Don't close specified open file descriptors. By default only file descriptors 0, 1
and 2 are inherited to the sandbox, and all other file descriptors are closed.
Example:
$ firejail --keep-fd=3,4,5
--keep-var-tmp
/var/tmp directory is untouched.
Example:
$ firejail --keep-var-tmp
--list List all sandboxes, see MONITORING section for more details.
Example:
$ firejail --list
7015:netblue:browser:firejail firefox
7056:netblue:torrent:firejail --net=eth0 transmission-gtk
7064:netblue::firejail --noroot xterm
--ls=name|pid dir_or_filename
List files in sandbox container, see FILE TRANSFER section for more details.
--mac=address
Assign MAC addresses to the last network interface defined by a --net option. This
option is not supported for wireless interfaces.
Example:
$ firejail --net=eth0 --mac=00:11:22:33:44:55 firefox
--machine-id
Spoof id number in /etc/machine-id file - a new random id is generated inside the
sandbox. Note that this breaks audio support. Enable it when sound is not
required.
Example:
$ firejail --machine-id
--mkdir=dirname
Create a directory in user home. Parent directories are created as needed.
Example:
$ firejail --mkdir=~/work/project
--mkfile=filename
Create an empty file in user home.
Example:
$ firejail --mkfile=~/work/project/readme
--memory-deny-write-execute
Install a seccomp filter to block attempts to create memory mappings that are both
writable and executable, to change mappings to be executable, or to create
executable shared memory. The filter examines the arguments of mmap, mmap2,
mprotect, pkey_mprotect, memfd_create and shmat system calls and returns error
EPERM to the process (or kills it or log the attempt, see --seccomp-error-action
below) if necessary.
Note: shmat is not implemented as a system call on some platforms including i386,
and it cannot be handled by seccomp-bpf.
--mtu=number
Assign a MTU value to the last network interface defined by a --net option.
Example:
$ firejail --net=eth0 --mtu=1492
--name=name
Set sandbox name. Several options, such as --join and --shutdown, can use this name
to identify a sandbox.
In case the name supplied by the user is already in use by another sandbox,
Firejail will assign a new name as "name-PID", where PID is the process ID of the
sandbox. This functionality can be disabled at run time in
/etc/firejail/firejail.config file, by setting "name-change" flag to "no".
Example:
$ firejail --name=browser firefox &
$ firejail --name=browser --private firefox --no-remote &
$ firejail --list
1198:netblue:browser:firejail --name=browser firefox
1312:netblue:browser-1312:firejail --name=browser --private firefox --no-remote
--net=bridge_interface
Enable a new network namespace and connect it to this bridge interface. Unless
specified with option --ip and --defaultgw, an IP address and a default gateway
will be assigned automatically to the sandbox. The IP address is verified using ARP
before assignment. The address configured as default gateway is the bridge device
IP address. Up to four --net options can be specified.
Example:
$ sudo brctl addbr br0
$ sudo ifconfig br0 10.10.20.1/24
$ sudo brctl addbr br1
$ sudo ifconfig br1 10.10.30.1/24
$ firejail --net=br0 --net=br1
--net=ethernet_interface|wireless_interface
Enable a new network namespace and connect it to this ethernet interface using the
standard Linux macvlan|ipvlan driver. Unless specified with option --ip and
--defaultgw, an IP address and a default gateway will be assigned automatically to
the sandbox. The IP address is verified using ARP before assignment. The address
configured as default gateway is the default gateway of the host. Up to four --net
options can be specified. Support for ipvlan driver was introduced in Linux kernel
3.19.
Example:
$ firejail --net=eth0 --ip=192.168.1.80 --dns=8.8.8.8 firefox
$ firejail --net=wlan0 firefox
--net=none
Enable a new, unconnected network namespace. The only interface available in the
new namespace is a new loopback interface (lo). Use this option to deny network
access to programs that don't really need network access.
Example:
$ firejail --net=none vlc
Note: --net=none can crash the application on some platforms. In these cases, it
can be replaced with --protocol=unix.
--net=tap_interface
Enable a new network namespace and connect it to this ethernet tap interface using
the standard Linux macvlan driver. If the tap interface is not configured, the
sandbox will not try to configure the interface inside the sandbox. Please use
--ip, --netmask and --defaultgw to specify the configuration.
Example:
$ firejail --net=tap0 --ip=10.10.20.80 --netmask=255.255.255.0
--defaultgw=10.10.20.1 firefox
--net.print=name|pid
If a new network namespace is enabled, print network interface configuration for
the sandbox specified by name or PID. Example:
$ firejail --net.print=browser
Switching to pid 1853, the first child process inside the sandbox
Interface MAC IP Mask Status
lo 127.0.0.1 255.0.0.0 UP
eth0-1852 5e:fb:8e:27:29:26 192.168.1.186 255.255.255.0 UP
--netfilter
Enable a default firewall if a new network namespace is created inside the sandbox.
This option has no effect for sandboxes using the system network namespace.
The default firewall is optimized for regular desktop applications. No incoming
connections are accepted:
*filter
:INPUT DROP [0:0]
:FORWARD DROP [0:0]
:OUTPUT ACCEPT [0:0]
-A INPUT -i lo -j ACCEPT
-A INPUT -m state --state RELATED,ESTABLISHED -j ACCEPT
# allow ping
-A INPUT -p icmp --icmp-type destination-unreachable -j ACCEPT
-A INPUT -p icmp --icmp-type time-exceeded -j ACCEPT
-A INPUT -p icmp --icmp-type echo-request -j ACCEPT
# drop STUN (WebRTC) requests
-A OUTPUT -p udp --dport 3478 -j DROP
-A OUTPUT -p udp --dport 3479 -j DROP
-A OUTPUT -p tcp --dport 3478 -j DROP
-A OUTPUT -p tcp --dport 3479 -j DROP
COMMIT
Example:
$ firejail --net=eth0 --netfilter firefox
--netfilter=filename
Enable the firewall specified by filename if a new network namespace is created
inside the sandbox. This option has no effect for sandboxes using the system
network namespace.
Please use the regular iptables-save/iptables-restore format for the filter file.
The following examples are available in /etc/firejail directory:
webserver.net is a webserver firewall that allows access only to TCP ports 80 and
443. Example:
$ firejail --netfilter=/etc/firejail/webserver.net --net=eth0 \
/etc/init.d/apache2 start
nolocal.net/nolocal6.net is a desktop client firewall that disable access to local
network. Example:
$ firejail --netfilter=/etc/firejail/nolocal.net \
--net=eth0 firefox
--netfilter=filename,arg1,arg2,arg3 ...
This is the template version of the previous command. $ARG1, $ARG2, $ARG3 ... in
the firewall script are replaced with arg1, arg2, arg3 ... passed on the command
line. Up to 16 arguments are supported. Example:
$ firejail --net=eth0 --ip=192.168.1.105 \
--netfilter=/etc/firejail/tcpserver.net,5001 server-program
--netfilter.print=name|pid
Print the firewall installed in the sandbox specified by name or PID. Example:
$ firejail --name=browser --net=eth0 --netfilter firefox &
$ firejail --netfilter.print=browser
--netfilter6=filename
Enable the IPv6 firewall specified by filename if a new network namespace is
created inside the sandbox. This option has no effect for sandboxes using the
system network namespace. Please use the regular iptables-save/iptables-restore
format for the filter file.
--netfilter6.print=name|pid
Print the IPv6 firewall installed in the sandbox specified by name or PID. Example:
$ firejail --name=browser --net=eth0 --netfilter firefox &
$ firejail --netfilter6.print=browser
--netlock
Several type of programs (email clients, multiplayer games etc.) talk to a very
small number of IP addresses. But the best example is tor browser. It only talks to
a guard node, and there are two or three more on standby in case the main one
fails. During startup, the browser contacts all of them, after that it keeps
talking to the main one... for weeks!
Use the network locking feature to build and deploy a custom network firewall in
your sandbox. The firewall allows only the traffic to the IP addresses detected
during the program startup. Traffic to any other address is quietly dropped. By
default the network monitoring time is one minute.
A network namespace (--net=eth0) is required for this feature to work. Example:
$ firejail --net=eth0 --netlock \
--private=~/tor-browser_en-US ./start-tor-browser.desktop
--netmask=address
Use this option when you want to assign an IP address in a new namespace and the
parent interface specified by --net is not configured. An IP address and a default
gateway address also have to be added. By default the new namespace interface comes
without IP address and default gateway configured. Example:
$ sudo /sbin/brctl addbr br0
$ sudo /sbin/ifconfig br0 up
$ firejail --ip=10.10.20.67 --netmask=255.255.255.0 --defaultgw=10.10.20.1
--netns=name
Run the program in a named, persistent network namespace. These can be created and
configured using "ip netns".
--netstats
Monitor network namespace statistics, see MONITORING section for more details.
Example:
$ firejail --netstats
PID User RX(KB/s) TX(KB/s) Command
1294 netblue 53.355 1.473 firejail --net=eth0 firefox
7383 netblue 9.045 0.112 firejail --net=eth0 transmission
--nettrace[=name|pid]
Monitor received TCP. UDP, and ICMP traffic. The sandbox can be specified by name
or pid. Only networked sandboxes created with --net are supported. This option is
only available when running the sandbox as root.
Without a name/pid, Firejail will monitor the main system network namespace.
Example:
$ sudo firejail --nettrace
95 KB/s geoip 457, IP database 4436
52 KB/s *********** 64.222.84.207:443 United States
33 KB/s ******* 89.147.74.105:63930 Hungary
0 B/s 45.90.28.0:443 NextDNS
0 B/s 94.70.122.176:52309(UDP) Greece
339 B/s 104.26.7.35:443 Cloudflare
If /usr/bin/geoiplookup is installed (geoip-bin package in Debian), the country the
traffic originates from is added to the trace. We also use the static IP map in
/usr/lib/firejail/static-ip-map to print the domain names for some of the more
common websites and cloud platforms. No external services are contacted for
reverse IP lookup.
--nice=value
Set nice value for all processes running inside the sandbox. Only root may specify
a negative value.
Example:
$ firejail --nice=2 firefox
--no3d Disable 3D hardware acceleration.
Example:
$ firejail --no3d firefox
--noautopulse (deprecated)
See --keep-config-pulse.
--noblacklist=dirname_or_filename
Disable blacklist for this directory or file.
Example:
$ firejail
$ nc dict.org 2628
bash: /bin/nc: Permission denied
$ exit
$ firejail --noblacklist=/bin/nc
$ nc dict.org 2628
220 pan.alephnull.com dictd 1.12.1/rf on Linux 3.14-1-amd64
--nodbus (deprecated)
Disable D-Bus access (both system and session buses). Equivalent to --dbus-
system=none --dbus-user=none.
Example:
$ firejail --nodbus --net=none
--nodvd
Disable DVD and audio CD devices.
Example:
$ firejail --nodvd
--noinput
Disable input devices.
Example:
$ firejail --noinput
--noexec=dirname_or_filename
Remount directory or file noexec, nodev and nosuid. File globbing is supported, see
FILE GLOBBING section for more details.
Example:
$ firejail --noexec=/tmp
/etc and /var are noexec by default if the sandbox was started as a regular user.
--nogroups
Disable supplementary groups. Without this option, supplementary groups are enabled
for the user starting the sandbox. For root user supplementary groups are always
disabled.
Note: By default all regular user groups are removed with the exception of the
current user. This can be changed using --allusers command option.
Example:
$ id
uid=1000(netblue) gid=1000(netblue)
groups=1000(netblue),24(cdrom),25(floppy),27(sudo),29(audio)
$ firejail --nogroups
Parent pid 8704, child pid 8705
Child process initialized
$ id
uid=1000(netblue) gid=1000(netblue) groups=1000(netblue)
$
--nonewprivs
Sets the NO_NEW_PRIVS prctl. This ensures that child processes cannot acquire new
privileges using execve(2); in particular, this means that calling a suid binary
(or one with file capabilities) does not result in an increase of privilege. This
option is enabled by default if seccomp filter is activated.
--noprinters
Disable printers.
--noprofile
Do not use a security profile.
Example:
$ firejail
Reading profile /etc/firejail/default.profile
Parent pid 8553, child pid 8554
Child process initialized
[...]
$ firejail --noprofile
Parent pid 8553, child pid 8554
Child process initialized
[...]
--noroot
Install a user namespace with a single user - the current user. root user does not
exist in the new namespace. This option requires a Linux kernel version 3.8 or
newer. The option is not supported for --chroot and --overlay configurations, or
for sandboxes started as root.
Example:
$ firejail --noroot
Parent pid 8553, child pid 8554
Child process initialized
$ ping google.com
ping: icmp open socket: Operation not permitted
$
--nosound
Disable sound system.
Example:
$ firejail --nosound firefox
--notv Disable DVB (Digital Video Broadcasting) TV devices.
Example:
$ firejail --notv vlc
--nou2f
Disable U2F devices.
Example:
$ firejail --nou2f
--novideo
Disable video devices.
--nowhitelist=dirname_or_filename
Disable whitelist for this directory or file.
--oom=value
Configure kernel's OutOfMemory-killer score for this sandbox. The acceptable score
values are between 0 and 1000 for regular users, and -1000 to 1000 for root. For
more information on OOM kernel feature see man choom.
Example:
$ firejail --oom=300 firefox
--output=logfile
stdout logging and log rotation. Copy stdout to logfile, and keep the size of the
file under 500KB using log rotation. Five files with prefixes .1 to .5 are used in
rotation.
Example:
$ firejail --output=sandboxlog /bin/bash
[...]
$ ls -l sandboxlog*
-rw-r--r-- 1 netblue netblue 333890 Jun 2 07:48 sandboxlog
-rw-r--r-- 1 netblue netblue 511488 Jun 2 07:48 sandboxlog.1
-rw-r--r-- 1 netblue netblue 511488 Jun 2 07:48 sandboxlog.2
-rw-r--r-- 1 netblue netblue 511488 Jun 2 07:48 sandboxlog.3
-rw-r--r-- 1 netblue netblue 511488 Jun 2 07:48 sandboxlog.4
-rw-r--r-- 1 netblue netblue 511488 Jun 2 07:48 sandboxlog.5
--output-stderr=logfile
Similar to --output, but stderr is also stored.
--private
Mount new /root and /home/user directories in temporary filesystems. All
modifications are discarded when the sandbox is closed.
Example:
$ firejail --private firefox
--private=directory
Use directory as user home. --private and --private=directory cannot be used
together.
Example:
$ firejail --private=/home/netblue/firefox-home firefox
Bug: Even with this enabled, some commands (such as mkdir, mkfile and private-
cache) will still operate on the original home directory. Workaround: Disable the
incompatible commands, such as by using "ignore mkdir" and "ignore mkfile". For
details, see #903 ⟨https://github.com/netblue30/firejail/issues/903⟩
--private-bin=file,file
Build a new /bin in a temporary filesystem, and copy the programs in the list. The
files in the list must be expressed as relative to the /bin, /sbin, /usr/bin,
/usr/sbin, or /usr/local/bin directories. If no listed files are found, /bin
directory will be empty. The same directory is also bind-mounted over /sbin,
/usr/bin, /usr/sbin and /usr/local/bin. All modifications are discarded when the
sandbox is closed. Multiple private-bin commands are allowed and they accumulate.
File globbing is supported, see FILE GLOBBING section for more details.
Example:
$ firejail --private-bin=bash,sed,ls,cat
Parent pid 20841, child pid 20842
Child process initialized
$ ls /bin
bash cat ls sed
--private-cache
Mount an empty temporary filesystem on top of the .cache directory in user home.
All modifications are discarded when the sandbox is closed.
Example:
$ firejail --private-cache openbox
--private-cwd
Set working directory inside jail to the home directory, and failing that, the root
directory. Does not impact working directory of profile include paths.
Example:
$ pwd
/tmp
$ firejail --private-cwd
$ pwd
/home/user
--private-cwd=directory
Set working directory inside the jail. Full directory path is required. Symbolic
links are not allowed. Does not impact working directory of profile include paths.
Example:
$ pwd
/tmp
$ firejail --private-cwd=/opt
$ pwd
/opt
--private-dev
Create a new /dev directory. Only disc, dri, dvb, hidraw, null, full, zero, tty,
pts, ptmx, random, snd, urandom, video, log, shm and usb devices are available.
Use the options --no3d, --nodvd, --nosound, --notv, --nou2f and --novideo for
additional restrictions.
Example:
$ firejail --private-dev
Parent pid 9887, child pid 9888
Child process initialized
$ ls /dev
cdrom cdrw dri dvd dvdrw full log null ptmx pts random shm snd sr0
tty urandom zero
$
--private-etc=file,directory
Build a new /etc in a temporary filesystem, and copy the files and directories in
the list. The files and directories in the list must be expressed as relative to
the /etc directory (e.g., /etc/foo must be expressed as foo). If no listed file is
found, /etc directory will be empty. All modifications are discarded when the
sandbox is closed. Multiple private-etc commands are allowed and they accumulate.
Example:
$ firejail --private-etc=group,hostname,localtime, \
nsswitch.conf,passwd,resolv.conf
--private-home=file,directory
Build a new user home in a temporary filesystem, and copy the files and directories
in the list in the new home. The files and directories in the list must be
expressed as relative to the current user's home directory. All modifications are
discarded when the sandbox is closed.
Example:
$ firejail --private-home=.mozilla firefox
--private-lib=file,directory
This feature is currently under heavy development. Only amd64 platforms are
supported at this moment. The files and directories in the list must be expressed
as relative to the /lib directory. The idea is to build a new /lib in a temporary
filesystem, with only the library files necessary to run the application. It could
be as simple as:
$ firejail --private-lib galculator
but it gets complicated really fast:
$ firejail --private-lib=x86_64-linux-gnu/xed,x86_64-linux-gnu/gdk-
pixbuf-2.0,libenchant.so.1,librsvg-2.so.2 xed
The feature is integrated with --private-bin:
$ firejail --private-lib --private-bin=bash,ls,ps
$ ls /lib
ld-linux-x86-64.so.2 libgpg-error.so.0 libprocps.so.6 libsystemd.so.0
libc.so.6 liblz4.so.1 libpthread.so.0 libtinfo.so.5
libdl.so.2 liblzma.so.5 librt.so.1 x86_64-linux-gnu
libgcrypt.so.20 libpcre.so.3 libselinux.so.1
$ ps
PID TTY TIME CMD
1 pts/0 00:00:00 firejail
45 pts/0 00:00:00 bash
48 pts/0 00:00:00 ps
$
Note: Support for this command is controlled in firejail.config with the private-
lib option.
--private-opt=file,directory
Build a new /opt in a temporary filesystem, and copy the files and directories in
the list. The files and directories in the list must be expressed as relative to
the /opt directory, and must not contain the / character (e.g., /opt/foo must be
expressed as foo, but /opt/foo/bar -- expressed as foo/bar -- is disallowed). If
no listed file is found, /opt directory will be empty. All modifications are
discarded when the sandbox is closed.
Example:
$ firejail --private-opt=firefox /opt/firefox/firefox
--private-srv=file,directory
Build a new /srv in a temporary filesystem, and copy the files and directories in
the list. The files and directories in the list must be expressed as relative to
the /srv directory, and must not contain the / character (e.g., /srv/foo must be
expressed as foo, but /srv/foo/bar -- expressed as srv/bar -- is disallowed). If
no listed file is found, /srv directory will be empty. All modifications are
discarded when the sandbox is closed.
Example:
# firejail --private-srv=www /etc/init.d/apache2 start
--private-tmp
Mount an empty temporary filesystem on top of /tmp directory whitelisting X11 and
PulseAudio sockets.
Example:
$ firejail --private-tmp
$ ls -al /tmp
drwxrwxrwt 4 nobody nogroup 80 Apr 30 11:46 .
drwxr-xr-x 30 nobody nogroup 4096 Apr 26 22:18 ..
drwx------ 2 nobody nogroup 4096 Apr 30 10:52 pulse-PKdhtXMmr18n
drwxrwxrwt 2 nobody nogroup 4096 Apr 30 10:52 .X11-unix
--profile=filename_or_profilename
Load a custom security profile from filename. For filename use an absolute path or
a path relative to the current path. For more information, see SECURITY PROFILES
section below.
Example:
$ firejail --profile=myprofile
--profile.print=name|pid
Print the name of the profile file for the sandbox identified by name or or PID.
Example:
$ firejail --profile.print=browser
/etc/firejail/firefox.profile
--protocol=protocol,protocol,protocol
Enable protocol filter. The filter is based on seccomp and checks the first
argument to socket system call. Recognized values: unix, inet, inet6, netlink,
packet, and bluetooth. This option is not supported for i386 architecture.
Multiple protocol commands are allowed and they accumulate.
Example:
$ firejail --protocol=unix,inet,inet6 firefox
--protocol.print=name|pid
Print the protocol filter for the sandbox identified by name or PID.
Example:
$ firejail --name=mybrowser firefox &
$ firejail --protocol.print=mybrowser
unix,inet,inet6,netlink
Example:
$ firejail --list
3272:netblue::firejail --private firefox
$ firejail --protocol.print=3272
unix,inet,inet6,netlink
--put=name|pid src-filename dest-filename
Put a file in sandbox container, see FILE TRANSFER section for more details.
--quiet
Turn off Firejail's output.
The same effect can be obtained by setting an environment variable FIREJAIL_QUIET
to yes.
--read-only=dirname_or_filename
Set directory or file read-only. File globbing is supported, see FILE GLOBBING
section for more details.
Example:
$ firejail --read-only=~/.mozilla firefox
--read-write=dirname_or_filename
Set directory or file read-write. Only files or directories belonging to the
current user are allowed for this operation. File globbing is supported, see FILE
GLOBBING section for more details. Example:
$ mkdir ~/test
$ touch ~/test/a
$ firejail --read-only=~/test --read-write=~/test/a
--restrict-namespaces
Install a seccomp filter that blocks attempts to create new cgroup, ipc, net,
mount, pid, time, user or uts namespaces.
Example:
$ firejail --restrict-namespaces
--restrict-namespaces=cgroup,ipc,net,mnt,pid,time,user,uts
Install a seccomp filter that blocks attempts to create any of the specified
namespaces. The filter examines the arguments of clone, unshare and setns system
calls and returns error EPERM to the process (or kills it or logs the attempt, see
--seccomp-error-action below) if necessary. Note that the filter is not able to
examine the arguments of clone3 system calls, and always responds to these calls
with error ENOSYS.
Example:
$ firejail --restrict-namespaces=user,net
--rlimit-as=number
Set the maximum size of the process's virtual memory (address space) in bytes. Use
k(ilobyte), m(egabyte) or g(igabyte) for size suffix (base 1024).
--rlimit-cpu=number
Set the maximum limit, in seconds, for the amount of CPU time each sandboxed
process can consume. When the limit is reached, the processes are killed.
The CPU limit is a limit on CPU seconds rather than elapsed time. CPU seconds is
basically how many seconds the CPU has been in use and does not necessarily
directly relate to the elapsed time. Linux kernel keeps track of CPU seconds for
each process independently.
--rlimit-fsize=number
Set the maximum file size that can be created by a process. Use k(ilobyte),
m(egabyte) or g(igabyte) for size suffix (base 1024).
--rlimit-nofile=number
Set the maximum number of files that can be opened by a process.
--rlimit-nproc=number
Set the maximum number of processes that can be created for the real user ID of the
calling process.
--rlimit-sigpending=number
Set the maximum number of pending signals for a process.
--rmenv=name
Remove environment variable in the new sandbox.
Example:
$ firejail --rmenv=DBUS_SESSION_BUS_ADDRESS
--scan ARP-scan all the networks from inside a network namespace. This makes it possible
to detect macvlan kernel device drivers running on the current host.
Example:
$ firejail --net=eth0 --scan
--seccomp
Enable seccomp filter and blacklist the syscalls in the default list, which is
@default-nodebuggers unless --allow-debuggers is specified, then it is @default.
To help creating useful seccomp filters more easily, the following system call
groups are defined: @aio, @basic-io, @chown, @clock, @cpu-emulation, @debug,
@default, @default-nodebuggers, @default-keep, @file-system, @io-event, @ipc,
@keyring, @memlock, @module, @mount, @network-io, @obsolete, @privileged, @process,
@raw-io, @reboot, @resources, @setuid, @swap, @sync, @system-service and @timer.
More information about groups can be found in /usr/share/doc/firejail/syscalls.txt
The default list can be customized, see --seccomp= for a description. It can be
customized also globally in /etc/firejail/firejail.config file.
System architecture is strictly imposed only if flag --seccomp.block-secondary is
used. The filter is applied at run time only if the correct architecture was
detected. For the case of I386 and AMD64 both 32-bit and 64-bit filters are
installed.
Firejail will print seccomp violations to the audit log if the kernel was compiled
with audit support (CONFIG_AUDIT flag).
Example:
$ firejail --seccomp
--seccomp=syscall,@group,!syscall2
Enable seccomp filter, blacklist the default list and the syscalls or syscall
groups specified by the command, but don't blacklist "syscall2". On a 64 bit
architecture, an additional filter for 32 bit system calls can be installed with
--seccomp.32.
Example:
$ firejail --seccomp=utime,utimensat,utimes firefox
$ firejail --seccomp=@clock,mkdir,unlinkat transmission-gtk
$ firejail '--seccomp=@ipc,!pipe,!pipe2' audacious
Syscalls can be specified by their number if prefix $ is added, so for example $165
would be equal to mount on i386.
Instead of dropping the syscall by returning EPERM, another error number can be
returned using syscall:errno syntax. This can be also changed globally with
--seccomp-error-action or in /etc/firejail/firejail.config file. The process can
also be killed by using syscall:kill syntax, or the attempt may be logged with
syscall:log.
Example:
$ firejail --seccomp=unlinkat:ENOENT,utimensat,utimes
Parent pid 10662, child pid 10663
Child process initialized
$ touch testfile
$ ls testfile
testfile
$ rm testfile
rm: cannot remove `testfile': No such file or directory
If the blocked system calls would also block Firejail from operating, they are
handled by adding a preloaded library which performs seccomp system calls later.
However, this is incompatible with 32 bit seccomp filters.
Example:
$ firejail --noprofile --seccomp=execve sh
Parent pid 32751, child pid 32752
Post-exec seccomp protector enabled
list in: execve, check list: @default-keep prelist: (null), postlist: execve
Child process initialized in 46.44 ms
$ ls
Operation not permitted
--seccomp.block-secondary
Enable seccomp filter and filter system call architectures so that only the native
architecture is allowed. For example, on amd64, i386 and x32 system calls are
blocked as well as changing the execution domain with personality(2) system call.
--seccomp.drop=syscall,@group
Enable seccomp filter, and blacklist the syscalls or the syscall groups specified
by the command. On a 64 bit architecture, an additional filter for 32 bit system
calls can be installed with --seccomp.32.drop.
Example:
$ firejail --seccomp.drop=utime,utimensat,utimes,@clock
Instead of dropping the syscall by returning EPERM, another error number can be
returned using syscall:errno syntax. This can be also changed globally with
--seccomp-error-action or in /etc/firejail/firejail.config file. The process can
also be killed by using syscall:kill syntax, or the attempt may be logged with
syscall:log.
Example:
$ firejail --seccomp.drop=unlinkat:ENOENT,utimensat,utimes
Parent pid 10662, child pid 10663
Child process initialized
$ touch testfile
$ ls testfile
testfile
$ rm testfile
rm: cannot remove `testfile': No such file or directory
--seccomp.keep=syscall,@group,!syscall2
Enable seccomp filter, blacklist all syscall not listed and "syscall2". The system
calls needed by Firejail (group @default-keep: prctl, execve, execveat) are handled
with the preload library. On a 64 bit architecture, an additional filter for 32 bit
system calls can be installed with --seccomp.32.keep.
Example:
$ firejail --seccomp.keep=poll,select,[...] transmission-gtk
--seccomp.print=name|pid
Print the seccomp filter for the sandbox identified by name or PID.
Example:
$ firejail --name=browser firefox &
$ firejail --seccomp.print=browser
line OP JT JF K
=================================
0000: 20 00 00 00000004 ld data.architecture
0001: 15 01 00 c000003e jeq ARCH_64 0003 (false 0002)
0002: 06 00 00 7fff0000 ret ALLOW
0003: 20 00 00 00000000 ld data.syscall-number
0004: 35 01 00 40000000 jge X32_ABI true:0006 (false 0005)
0005: 35 01 00 00000000 jge read 0007 (false 0006)
0006: 06 00 00 00050001 ret ERRNO(1)
0007: 15 41 00 0000009a jeq modify_ldt 0049 (false 0008)
0008: 15 40 00 000000d4 jeq lookup_dcookie 0049 (false 0009)
0009: 15 3f 00 0000012a jeq perf_event_open 0049 (false 000a)
000a: 15 3e 00 00000137 jeq process_vm_writev 0049 (false 000b)
000b: 15 3d 00 0000009c jeq _sysctl 0049 (false 000c)
000c: 15 3c 00 000000b7 jeq afs_syscall 0049 (false 000d)
000d: 15 3b 00 000000ae jeq create_module 0049 (false 000e)
000e: 15 3a 00 000000b1 jeq get_kernel_syms 0049 (false 000f)
000f: 15 39 00 000000b5 jeq getpmsg 0049 (false 0010)
0010: 15 38 00 000000b6 jeq putpmsg 0049 (false 0011)
0011: 15 37 00 000000b2 jeq query_module 0049 (false 0012)
0012: 15 36 00 000000b9 jeq security 0049 (false 0013)
0013: 15 35 00 0000008b jeq sysfs 0049 (false 0014)
0014: 15 34 00 000000b8 jeq tuxcall 0049 (false 0015)
0015: 15 33 00 00000086 jeq uselib 0049 (false 0016)
0016: 15 32 00 00000088 jeq ustat 0049 (false 0017)
0017: 15 31 00 000000ec jeq vserver 0049 (false 0018)
0018: 15 30 00 0000009f jeq adjtimex 0049 (false 0019)
0019: 15 2f 00 00000131 jeq clock_adjtime 0049 (false 001a)
001a: 15 2e 00 000000e3 jeq clock_settime 0049 (false 001b)
001b: 15 2d 00 000000a4 jeq settimeofday 0049 (false 001c)
001c: 15 2c 00 000000b0 jeq delete_module 0049 (false 001d)
001d: 15 2b 00 00000139 jeq finit_module 0049 (false 001e)
001e: 15 2a 00 000000af jeq init_module 0049 (false 001f)
001f: 15 29 00 000000ad jeq ioperm 0049 (false 0020)
0020: 15 28 00 000000ac jeq iopl 0049 (false 0021)
0021: 15 27 00 000000f6 jeq kexec_load 0049 (false 0022)
0022: 15 26 00 00000140 jeq kexec_file_load 0049 (false 0023)
0023: 15 25 00 000000a9 jeq reboot 0049 (false 0024)
0024: 15 24 00 000000a7 jeq swapon 0049 (false 0025)
0025: 15 23 00 000000a8 jeq swapoff 0049 (false 0026)
0026: 15 22 00 000000a3 jeq acct 0049 (false 0027)
0027: 15 21 00 00000141 jeq bpf 0049 (false 0028)
0028: 15 20 00 000000a1 jeq chroot 0049 (false 0029)
0029: 15 1f 00 000000a5 jeq mount 0049 (false 002a)
002a: 15 1e 00 000000b4 jeq nfsservctl 0049 (false 002b)
002b: 15 1d 00 0000009b jeq pivot_root 0049 (false 002c)
002c: 15 1c 00 000000ab jeq setdomainname 0049 (false 002d)
002d: 15 1b 00 000000aa jeq sethostname 0049 (false 002e)
002e: 15 1a 00 000000a6 jeq umount2 0049 (false 002f)
002f: 15 19 00 00000099 jeq vhangup 0049 (false 0030)
0030: 15 18 00 000000ee jeq set_mempolicy 0049 (false 0031)
0031: 15 17 00 00000100 jeq migrate_pages 0049 (false 0032)
0032: 15 16 00 00000117 jeq move_pages 0049 (false 0033)
0033: 15 15 00 000000ed jeq mbind 0049 (false 0034)
0034: 15 14 00 00000130 jeq open_by_handle_at 0049 (false 0035)
0035: 15 13 00 0000012f jeq name_to_handle_at 0049 (false 0036)
0036: 15 12 00 000000fb jeq ioprio_set 0049 (false 0037)
0037: 15 11 00 00000067 jeq syslog 0049 (false 0038)
0038: 15 10 00 0000012c jeq fanotify_init 0049 (false 0039)
0039: 15 0f 00 00000138 jeq kcmp 0049 (false 003a)
003a: 15 0e 00 000000f8 jeq add_key 0049 (false 003b)
003b: 15 0d 00 000000f9 jeq request_key 0049 (false 003c)
003c: 15 0c 00 000000fa jeq keyctl 0049 (false 003d)
003d: 15 0b 00 000000ce jeq io_setup 0049 (false 003e)
003e: 15 0a 00 000000cf jeq io_destroy 0049 (false 003f)
003f: 15 09 00 000000d0 jeq io_getevents 0049 (false 0040)
0040: 15 08 00 000000d1 jeq io_submit 0049 (false 0041)
0041: 15 07 00 000000d2 jeq io_cancel 0049 (false 0042)
0042: 15 06 00 000000d8 jeq remap_file_pages 0049 (false 0043)
0043: 15 05 00 00000116 jeq vmsplice 0049 (false 0044)
0044: 15 04 00 00000087 jeq personality 0049 (false 0045)
0045: 15 03 00 00000143 jeq userfaultfd 0049 (false 0046)
0046: 15 02 00 00000065 jeq ptrace 0049 (false 0047)
0047: 15 01 00 00000136 jeq process_vm_readv 0049 (false 0048)
0048: 06 00 00 7fff0000 ret ALLOW
0049: 06 00 01 00000000 ret KILL
$
--seccomp-error-action= kill | ERRNO | log
By default, if a seccomp filter blocks a system call, the process gets EPERM as the
error. With --seccomp-error-action=error, another error number can be returned, for
example ENOSYS or EACCES. The process can also be killed (like in versions <0.9.63
of Firejail) by using --seccomp-error-action=kill syntax, or the attempt may be
logged with --seccomp-error-action=log. Not killing the process weakens Firejail
slightly when trying to contain intrusion, but it may also allow tighter filters if
the only alternative is to allow a system call.
--shutdown=name|pid
Shutdown the sandbox identified by name or PID.
Example:
$ firejail --name=mygame --caps.drop=all warzone2100 &
$ firejail --shutdown=mygame
Example:
$ firejail --list
3272:netblue::firejail --private firefox
$ firejail --shutdown=3272
--snitrace[=name|pid]
Monitor Server Name Indication (TLS/SNI). The sandbox can be specified by name or
pid. Only networked sandboxes created with --net are supported. This option is only
available when running the sandbox as root.
Without a name/pid, Firejail will monitor the main system network namespace.
Example:
$ sudo firejail --snitrace
07:49:51 23.185.0.3 linux.com
07:49:51 23.185.0.3 www.linux.com
07:50:05 192.0.73.2 secure.gravatar.com
07:52:35 172.67.68.93 www.howtoforge.com
07:52:37 13.225.103.59 sf.ezoiccdn.com
07:52:42 142.250.176.3 www.gstatic.com
07:53:03 173.236.250.32 www.linuxlinks.com
07:53:05 192.0.77.37 c0.wp.com
07:53:08 192.0.78.32 jetpack.wordpress.com
07:53:09 192.0.77.32 s0.wp.com
07:53:09 192.0.77.2 i0.wp.com
07:53:10 192.0.77.2 i0.wp.com
07:53:11 192.0.73.2 1.gravatar.com
--tab Enable shell tab completion in sandboxes using private or whitelisted home
directories.
$ firejail --private --tab
--timeout=hh:mm:ss
Kill the sandbox automatically after the time has elapsed. The time is specified in
hours/minutes/seconds format.
$ firejail --timeout=01:30:00 firefox
--tmpfs=dirname
Mount a writable tmpfs filesystem on directory dirname. Directories outside user
home or not owned by the user are not allowed. Sandboxes running as root are exempt
from these restrictions. File globbing is supported, see FILE GLOBBING section for
more details.
Example:
$ firejail --tmpfs=~/.local/share
--top Monitor the most CPU-intensive sandboxes, see MONITORING section for more details.
Example:
$ firejail --top
--trace[=filename]
Trace open, access and connect system calls. If filename is specified, log trace
output to filename, otherwise log to console.
Example:
$ firejail --trace wget -q www.debian.org
Reading profile /etc/firejail/wget.profile
3:wget:fopen64 /etc/wgetrc:0x5c8e8ce6c0
3:wget:fopen /etc/hosts:0x5c8e8cfb70
3:wget:socket AF_INET SOCK_DGRAM IPPROTO_IP:3
3:wget:connect 3 8.8.8.8 port 53:0
3:wget:socket AF_INET SOCK_STREAM IPPROTO_IP:3
3:wget:connect 3 130.89.148.14 port 80:0
3:wget:fopen64 index.html:0x5c8e8d1a60
parent is shutting down, bye...
--tracelog
This option enables auditing blacklisted files and directories. A message is sent
to syslog in case the file or the directory is accessed.
Example:
$ firejail --tracelog firefox
Sample messages:
$ sudo tail -f /var/log/syslog
[...]
Dec 3 11:43:25 debian firejail[70]: blacklist violation - sandbox 26370, exe
firefox, syscall open64, path /etc/shadow
Dec 3 11:46:17 debian firejail[70]: blacklist violation - sandbox 26370, exe
firefox, syscall opendir, path /boot
[...]
Note: Support for this command is controlled in firejail.config with the tracelog
option.
--tree Print a tree of all sandboxed processes, see MONITORING section for more details.
Example:
$ firejail --tree
11903:netblue:firejail iceweasel
11904:netblue:iceweasel
11957:netblue:/usr/lib/iceweasel/plugin-container
11969:netblue:firejail --net=eth0 transmission-gtk
11970:netblue:transmission-gtk
--version
Print program version/compile time support and exit.
Example:
$ firejail --version
firejail version 0.9.27
Compile time support:
- AppArmor support is enabled
- AppImage support is enabled
- chroot support is enabled
- file and directory whitelisting support is enabled
- file transfer support is enabled
- firetunnel support is enabled
- networking support is enabled
- overlayfs support is enabled
- private-home support is enabled
- seccomp-bpf support is enabled
- user namespace support is enabled
- X11 sandboxing support is enabled
--veth-name=name
Use this name for the interface connected to the bridge for --net=bridge_interface
commands, instead of the default one.
Example:
$ firejail --net=br0 --veth-name=if0
--whitelist=dirname_or_filename
Whitelist directory or file. A temporary file system is mounted on the top
directory, and the whitelisted files are mount-binded inside. Modifications to
whitelisted files are persistent, everything else is discarded when the sandbox is
closed. The top directory can be all directories in / (except /proc and /sys),
/sys/module, /run/user/$UID, $HOME and all directories in /usr.
Symbolic link handling: Whitelisting a path that is a symbolic link will also
whitelist the path that it points to. For example, if ~/foo is whitelisted and it
points to ~/bar, then ~/bar will also be whitelisted. Restrictions: With the
exception of the user home directory, both the link and the real file should be in
the same top directory. For symbolic links in the user home directory, both the
link and the real file should be owned by the user.
File globbing is supported, see FILE GLOBBING section for more details.
Example:
$ firejail --noprofile --whitelist=~/.mozilla
$ firejail --whitelist=/tmp/.X11-unix --whitelist=/dev/null
$ firejail "--whitelist=/home/username/My Virtual Machines"
$ firejail --whitelist=~/work* --whitelist=/var/backups*
--writable-etc
Mount /etc directory read-write.
Example:
$ sudo firejail --writable-etc
--writable-run-user
Disable the default blacklisting of /run/user/$UID/systemd and
/run/user/$UID/gnupg.
Example:
$ sudo firejail --writable-run-user
--writable-var
Mount /var directory read-write.
Example:
$ sudo firejail --writable-var
--writable-var-log
Use the real /var/log directory, not a clone. By default, a tmpfs is mounted on top
of /var/log directory, and a skeleton filesystem is created based on the original
/var/log.
Example:
$ sudo firejail --writable-var-log
--x11 Sandbox the application using Xpra, Xephyr, Xvfb or Xorg security extension. The
sandbox will prevent screenshot and keylogger applications started inside the
sandbox from accessing clients running outside the sandbox. Firejail will try Xpra
first, and if Xpra is not installed on the system, it will try to find Xephyr. If
all fails, Firejail will not attempt to use Xvfb or X11 security extension.
Xpra, Xephyr and Xvfb modes require a network namespace to be instantiated in order
to disable X11 abstract Unix socket. If this is not possible, the user can disable
the abstract socket by adding "-nolisten local" on Xorg command line at system
level.
Example:
$ firejail --x11 --net=eth0 firefox
--x11=none
Blacklist /tmp/.X11-unix directory, ${HOME}/.Xauthority and the file specified in
${XAUTHORITY} environment variable. Remove DISPLAY and XAUTHORITY environment
variables. Stop with error message if X11 abstract socket will be accessible in
jail.
--x11=xephyr
Start Xephyr and attach the sandbox to this server. Xephyr is a display server
implementing the X11 display server protocol. A network namespace needs to be
instantiated in order to deny access to X11 abstract Unix domain socket.
Xephyr runs in a window just like any other X11 application. The default window
size is 800x600. This can be modified in /etc/firejail/firejail.config file.
The recommended way to use this feature is to run a window manager inside the
sandbox. A security profile for OpenBox is provided.
Xephyr is developed by Xorg project. On Debian platforms it is installed with the
command sudo apt-get install xserver-xephyr. This feature is not available when
running as root.
Example:
$ firejail --x11=xephyr --net=eth0 openbox
--x11=xorg
Sandbox the application using the untrusted mode implemented by X11 security
extension. The extension is available in Xorg package and it is installed by
default on most Linux distributions. It provides support for a simple
trusted/untrusted connection model. Untrusted clients are restricted in certain
ways to prevent them from reading window contents of other clients, stealing input
events, etc.
The untrusted mode has several limitations. A lot of regular programs assume they
are a trusted X11 clients and will crash or lock up when run in untrusted mode.
Chromium browser and xterm are two examples. Firefox and transmission-gtk seem to
be working fine. A network namespace is not required for this option.
Example:
$ firejail --x11=xorg firefox
--x11=xpra
Start Xpra (https://xpra.org) and attach the sandbox to this server. Xpra is a
persistent remote display server and client for forwarding X11 applications and
desktop screens. A network namespace needs to be instantiated in order to deny
access to X11 abstract Unix domain socket.
On Debian platforms Xpra is installed with the command sudo apt-get install xpra.
This feature is not available when running as root.
Example:
$ firejail --x11=xpra --net=eth0 firefox
--x11=xvfb
Start Xvfb X11 server and attach the sandbox to this server. Xvfb, short for X
virtual framebuffer, performs all graphical operations in memory without showing
any screen output. Xvfb is mainly used for remote access and software testing on
headless servers.
On Debian platforms Xvfb is installed with the command sudo apt-get install xvfb.
This feature is not available when running as root.
Example: remote VNC access
On the server we start a sandbox using Xvfb and openbox window manager. The default
size of Xvfb screen is 800x600 - it can be changed in /etc/firejail/firejail.config
(xvfb-screen). Some sort of networking (--net) is required in order to isolate the
abstract sockets used by other X servers.
$ firejail --net=none --x11=xvfb openbox
*** Attaching to Xvfb display 792 ***
Reading profile /etc/firejail/openbox.profile
Reading profile /etc/firejail/disable-common.inc
Reading profile /etc/firejail/disable-common.local
Parent pid 5400, child pid 5401
On the server we also start a VNC server and attach it to the display handled by
our Xvfb server (792).
$ x11vnc -display :792
On the client machine we start a VNC viewer and use it to connect to our server:
$ vncviewer
--xephyr-screen=WIDTHxHEIGHT
Set screen size for --x11=xephyr. The setting will overwrite the default set in
/etc/firejail/firejail.config for the current sandbox. Run xrandr to get a list of
supported resolutions on your computer.
Example:
$ firejail --net=eth0 --x11=xephyr --xephyr-screen=640x480 firefox
APPARMOR
AppArmor support is disabled by default at compile time. Use --enable-apparmor
configuration option to enable it:
$ ./configure --prefix=/usr --enable-apparmor
During software install, a generic AppArmor profile file, firejail-default, is placed in
/etc/apparmor.d directory. The local customizations must be placed in
/etc/apparmor.d/local/firejail-local. The profile needs to be loaded into the kernel by
reloading apparmor.service, rebooting the system or running the following command as root:
# apparmor_parser -r /etc/apparmor.d/firejail-default
The installed profile is supplemental for main firejail functions and among other things
does the following:
- Disable ptrace. With ptrace it is possible to inspect and hijack running
programs. Usually this is needed only for debugging. You should have no problems
running Chromium or Firefox. This feature is available only on Ubuntu kernels.
- Whitelist write access to several files under /run, /proc and /sys.
- Allow running programs only from well-known system paths, such as /bin, /sbin,
/usr/bin etc. Those paths are available as read-only. Running programs and scripts
from user home or other directories writable by the user is not allowed.
- Prevent using non-standard network sockets. Only unix, inet, inet6, netlink, raw
and packet are allowed.
- Deny access to known sensitive paths like .snapshots.
To enable AppArmor confinement on top of your current Firejail security features, pass
--apparmor flag to Firejail command line. You can also include apparmor command in a
Firejail profile file. Example:
$ firejail --apparmor firefox
DESKTOP INTEGRATION
A symbolic link to /usr/bin/firejail under the name of a program, will start the program
in Firejail sandbox. The symbolic link should be placed in the first $PATH position. On
most systems, a good place is /usr/local/bin directory. Example:
Make a firefox symlink to /usr/bin/firejail:
$ sudo ln -s /usr/bin/firejail /usr/local/bin/firefox
Verify $PATH
$ which -a firefox
/usr/local/bin/firefox
/usr/bin/firefox
Starting firefox in this moment, automatically invokes “firejail firefox”.
This works for clicking on desktop environment icons, menus etc. Use "firejail --tree" to
verify the program is sandboxed.
$ firejail --tree
1189:netblue:firejail firefox
1190:netblue:firejail firefox
1220:netblue:/bin/sh -c "/usr/lib/firefox/firefox"
1221:netblue:/usr/lib/firefox/firefox
We provide a tool that automates all this integration, please see firecfg(1) for more
details.
EXAMPLES
firejail
Sandbox a regular shell session.
firejail firefox
Start Mozilla Firefox.
firejail --debug firefox
Debug Firefox sandbox.
firejail --private firefox
Start Firefox with a new, empty home directory.
firejail --net=none vlc
Start VLC in an unconnected network namespace.
firejail --net=eth0 firefox
Start Firefox in a new network namespace. An IP address is assigned automatically.
firejail --net=br0 --ip=10.10.20.5 --net=br1 --net=br2
Start a shell session in a new network namespace and connect it to br0, br1, and
br2 host bridge devices. IP addresses are assigned automatically for the interfaces
connected to br1 and b2
firejail --list
List all sandboxed processes.
FILE GLOBBING
Globbing is the operation that expands a wildcard pattern into the
list of pathnames matching the pattern. This pattern is matched at firejail start,
and is NOT UPDATED at runtime. Files matching a blacklist, but created after
firejail start will be accessible within the jail. Matching is defined by:
- '?' matches any character
- '*' matches any string
- '[' denotes a range of characters
The globbing feature is implemented using glibc glob command. For
more information on the wildcard syntax see man 7 glob.
The following command line options are supported: --blacklist,
--private-bin, --noexec, --read-only, --read-write, --tmpfs, and --whitelist.
Examples:
$ firejail --private-bin=sh,bash,python*
$ firejail --blacklist=~/dir[1234]
$ firejail --read-only=~/dir[1-4]
FILE TRANSFER
These features allow the user to inspect the filesystem container of an existing sandbox
and transfer files between the container and the host filesystem.
--cat=name|pid filename
Write content of a container file to standard out. The container is specified by
name or PID. If standard out is a terminal, all ASCII control characters except
new line and horizontal tab are replaced.
--get=name|pid filename
Retrieve the container file and store it on the host in the current working
directory. The container is specified by name or PID.
--ls=name|pid dir_or_filename
List container files. The container is specified by name or PID.
--put=name|pid src-filename dest-filename
Put src-filename in sandbox container. The container is specified by name or PID.
Examples:
$ firejail --name=mybrowser --private firefox
$ firejail --ls=mybrowser ~/Downloads
drwxr-xr-x netblue netblue 4096 .
drwxr-xr-x netblue netblue 4096 ..
-rw-r--r-- netblue netblue 7847 x11-x305.png
-rw-r--r-- netblue netblue 6800 x11-x642.png
-rw-r--r-- netblue netblue 34139 xpra-clipboard.png
$ firejail --get=mybrowser ~/Downloads/xpra-clipboard.png
$ firejail --put=mybrowser xpra-clipboard.png ~/Downloads/xpra-clipboard.png
$ firejail --cat=mybrowser ~/.bashrc
INTRUSION DETECTION SYSTEM (IDS)
The host-based intrusion detection system tracks down and audits user and system file
modifications. The feature is configured using /etc/firejail/ids.config file, the
checksums are stored in /var/lib/firejail/USERNAME.ids, where USERNAME is the name of the
current user. We use BLAKE2 cryptographic function for hashing.
As a regular user, initialize the database:
$ firejail --ids-init
Opening config file /etc/firejail/ids.config
Loading config file /etc/firejail/ids.config
Opening config file /etc/firejail/ids.config.local
500 1000 1500 2000
2466 files scanned
IDS database initialized
The default configuration targets several system executables in directories such as /bin,
/sbin, /usr/bin, /usr/sbin, and several critical config files in user home directory such
as ~/.bashrc, ~/.xinitrc, and ~/.config/autostart. Several system config files in /etc
directory are also hashed.
Run --ids-check to audit the system:
$ firejail --ids-check
Opening config file /etc/firejail/ids.config
Loading config file /etc/firejail/ids.config
Opening config file /etc/firejail/ids.config.local
500 1000 1500
Warning: modified /home/netblue/.bashrc
2000
2466 files scanned: modified 1, permissions 0, new 0, removed 0
The program will print the files that have been modified since the database was created,
or the files with different access permissions. New files and deleted files are also
flagged.
Currently while scanning the file system, symbolic links are not followed, and files the
user doesn't have read access to are silently dropped. The program can also be run as
root (sudo firejail --ids-init/--ids-check).
MONITORING
Option --list prints a list of all sandboxes. The format for each process entry is as
follows:
PID:USER:Sandbox Name:Command
Option --tree prints the tree of processes running in the sandbox. The format for each
process entry is as follows:
PID:USER:Sandbox Name:Command
Option --top is similar to the UNIX top command, however it applies only to sandboxes.
Option --netstats prints network statistics for active sandboxes installing new network
namespaces.
Listed below are the available fields (columns) in alphabetical order for --top and
--netstats options:
Command
Command used to start the sandbox.
CPU% CPU usage, the sandbox share of the elapsed CPU time since the last screen update
PID Unique process ID for the task controlling the sandbox.
Prcs Number of processes running in sandbox, including the controlling process.
RES Resident Memory Size (KiB), sandbox non-swapped physical memory. It is a sum of
the RES values for all processes running in the sandbox.
RX(KB/s)
Network receive speed.
Sandbox Name
The name of the sandbox, if any.
SHR Shared Memory Size (KiB), it reflects memory shared with other processes. It is a
sum of the SHR values for all processes running in the sandbox, including the
controlling process.
TX(KB/s)
Network transmit speed.
Uptime Sandbox running time in hours:minutes:seconds format.
USER The owner of the sandbox.
RESTRICTED SHELL
To configure a restricted shell, replace /bin/bash with /usr/bin/firejail in /etc/passwd
file for each user that needs to be restricted. Alternatively, you can specify
/usr/bin/firejail in adduser command:
adduser --shell /usr/bin/firejail username
Additional arguments passed to firejail executable upon login are declared in
/etc/firejail/login.users file.
SECURITY PROFILES
Several command line options can be passed to the program using profile files. Firejail
chooses the profile file as follows:
1. If a profile file is provided by the user with --profile=FILE option, the profile FILE
is loaded. If a profile name is given, it is searched for first in the ~/.config/firejail
directory and if not found then in /etc/firejail directory. Profile names do not include
the .profile suffix. If there is a file with the same name as the given profile name, it
will be used instead of doing the profile search. To force a profile search, prefix the
profile name with a colon (:), eg. --profile=:PROFILE_NAME. Example:
$ firejail --profile=/home/netblue/icecat.profile icecat
Reading profile /home/netblue/icecat.profile
[...]
$ firejail --profile=icecat icecat-wrapper.sh
Reading profile /etc/firejail/icecat.profile
[...]
2. If a profile file with the same name as the application is present in
~/.config/firejail directory or in /etc/firejail, the profile is loaded.
~/.config/firejail takes precedence over /etc/firejail. Example:
$ firejail icecat
Command name #icecat#
Found icecat profile in /home/netblue/.config/firejail directory
Reading profile /home/netblue/.config/firejail/icecat.profile
[...]
3. Use default.profile file if the sandbox is started by a regular user, or server.profile
file if the sandbox is started by root. Firejail looks for these files in
~/.config/firejail directory, followed by /etc/firejail directory. To disable default
profile loading, use --noprofile command option. Example:
$ firejail
Reading profile /etc/firejail/default.profile
Parent pid 8553, child pid 8554
Child process initialized
[...]
$ firejail --noprofile
Parent pid 8553, child pid 8554
Child process initialized
[...]
See man 5 firejail-profile for profile file syntax information.
TRAFFIC SHAPING
Network bandwidth is an expensive resource shared among all sandboxes running on a system.
Traffic shaping allows the user to increase network performance by controlling the amount
of data that flows into and out of the sandboxes.
Firejail implements a simple rate-limiting shaper based on Linux command tc. The shaper
works at sandbox level, and can be used only for sandboxes configured with new network
namespaces.
Set rate-limits:
$ firejail --bandwidth=name|pid set network download upload
Clear rate-limits:
$ firejail --bandwidth=name|pid clear network
Status:
$ firejail --bandwidth=name|pid status
where:
name - sandbox name
pid - sandbox pid
network - network interface as used by --net option
download - download speed in KB/s (kilobyte per second)
upload - upload speed in KB/s (kilobyte per second)
Example:
$ firejail --name=mybrowser --net=eth0 firefox &
$ firejail --bandwidth=mybrowser set eth0 80 20
$ firejail --bandwidth=mybrowser status
$ firejail --bandwidth=mybrowser clear eth0
LICENSE
This program is free software; you can redistribute it and/or modify it under the terms of
the GNU General Public License as published by the Free Software Foundation; either
version 2 of the License, or (at your option) any later version.
Homepage: https://firejail.wordpress.com
SEE ALSO
firemon(1), firecfg(1), firejail-profile(5), firejail-login(5), firejail-users(5),
jailcheck(1)
⟨https://github.com/netblue30/firejail/wiki⟩, ⟨https://github.com/netblue30/firejail⟩