-- Signal the end of options and disables further option
processing.
--allow=dirname_or_filename
Allow access to a directory or file. A temporary file
system is mounted on the top directory, and the allowed
files are mount-binded inside. Modifications to allowed
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: with the exception of user home,
both the link and the real file should be in the same top
directory. For user home, 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 --allow=~/.mozilla
$ firejail --allow=/tmp/.X11-unix --allow=/dev/null
$ firejail "--allow=/home/username/My Virtual Machines"
$ firejail --allow=~/work* --allow=/var/backups*
--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
--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 --appimage --private --profile=krita
krita-3.0-x86_64.appimage
$ firejail --appimage --net=none --x11 --profile=krita
krita-3.0-x86_64.appimage
--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
--build
The command builds a whitelisted profile. The profile is
printed on the screen. If /usr/bin/strace is installed on
the system, it also builds a whitelisted seccomp profile.
The program is run in a very relaxed sandbox, with only
--caps.drop=all and --nonewprivs. Programs that raise user
privileges are not supported in order to allow strace to
run. Chromium and Chromium-based browsers will not work.
Example:
$ firejail --build vlc ~/Videos/test.mp4
--build=profile-file
The command builds a whitelisted profile, and saves it in
profile-file. If /usr/bin/strace is installed on the
system, it also builds a whitelisted seccomp profile. The
program is run in a very relaxed sandbox, with only
--caps.drop=all and --nonewprivs. Programs that raise user
privileges are not supported in order to allow strace to
run. Chromium and Chromium-based browsers will not work.
Example:
$ firejail --build=vlc.profile vlc ~/Videos/test.mp4
-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.
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 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.
--cgroup=tasks-file
Place the sandbox in the specified control group. tasks-
file is the full path of cgroup tasks file.
Example:
# firejail --cgroup=/sys/fs/cgroup/g1/tasks
--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
--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
theindicated 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
theindicated 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-allow
Debug file system access.
Example:
$ firejail --debug-allow firefox
--debug-caps
Print all recognized capabilities in the current Firejail
software build and exit.
Example:
$ firejail --debug-caps
--debug-deny
Debug file access.
Example:
$ firejail --debug-deny firefox
--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.
--defaultgw=address
Use this address as default gateway in the new network
namespace.
Example:
$ firejail --net=eth0 --defaultgw=10.10.20.1 firefox
--deny=dirname_or_filename
Deny access to directory or file. File globbing is
supported, see FILE GLOBBING section for more details.
Example:
$ firejail --deny=/sbin --deny=/usr/sbin
$ firejail --deny=~/.mozilla
$ firejail "--deny=/home/username/My Virtual Machines"
$ firejail --deny=/home/username/My\ Virtual\ Machines
--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.
--disable-mnt
Deny access to /mnt, /media, /run/mount and /run/media.
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
--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
--ignore=command
Ignore command in profile file.
Example:
$ firejail --ignore=shell --ignore=seccomp firefox
$ firejail --ignore="net eth0" firefox
--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, cgroups
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, cgroups 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, cgroups 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-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|ipvaln
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
--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
--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
--noallow=dirname_or_filename
Disable --allow for this directory or file.
--noautopulse (deprecated)
See --keep-config-pulse.
--nodeny=dirname_or_filename
Disable --deny for this directory or file.
Example:
$ firejail
$ nc dict.org 2628
bash: /bin/nc: Permission denied
$ exit
$ firejail --nodeny=/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.
--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.
--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
--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. 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.
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.
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
$
--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.
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
--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
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
The default list can be customized, see --seccomp= for a
description. It can be customized also globally in
/etc/firejail/firejail.config file.
--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 --shell=none --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 --shell=none --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.
--shell=none
Run the program directly, without a user shell.
Example:
$ firejail --shell=none script.sh
--shell=program
Set default user shell. Use this shell to run the
application using -c shell option. For example "firejail
--shell=/bin/dash firefox" will start Mozilla Firefox as
"/bin/dash -c firefox". By default the user's preferred
shell is used.
Example: $firejail --shell=/bin/dash script.sh
--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
--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
[...]
--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
--tunnel[=devname]
Connect the sandbox to a network overlay/VPN tunnel
created by firetunnel utility. This options tries first
the client side of the tunnel. If this fails, it tries the
server side. If multiple tunnels are active, please
specify the tunnel device using --tunnel=devname.
The available tunnel devices are listed in /etc/firetunnel
directory, one file for each device. The files are
regular firejail profile files containing the network
configuration, and are created and managed by firetunnel
utility. By default ftc is the client-side device and fts
is the server-side device. For more information please see
man 1 firetunnel.
Example:
$ firejail --tunnel firefox
--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
--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
