|
HTML rendering created 2024-06-26 by Michael Kerrisk, author of The Linux Programming Interface. For details of in-depth Linux/UNIX system programming training courses that I teach, look here. Hosting by jambit GmbH. |
|
|
NAME | SYNOPSIS | DESCRIPTION | USAGE | OPTIONS | NAME | DESKTOP INTEGRATION | EXAMPLES | FILE GLOBBING | FILE TRANSFER | MONITORING | RESTRICTED SHELL | SECURITY PROFILES | TRAFFIC SHAPING | LICENSE | SEE ALSO | COLOPHON |
|
|
|
FIREJAIL(1) firejail man page FIREJAIL(1)
Firejail - Linux namespaces sandbox program
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}
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.
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
-- 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
--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-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.
For valid names, see the NAME VALIDATION section.
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=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 ...".
See --name for details.
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-shell-rc
By default, when using a private home directory, firejail
copies files from the system's user home template
(/etc/skel) into it, which overrides attempts to whitelist
the original files (such as ~/.bashrc and ~/.zshrc). This
option disables this feature, and enables the user to
whitelist the original files.
--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. The
name cannot contain only digits, as that is treated as a
PID in the other options, such as in --join.
For valid names, see the NAME VALIDATION section.
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, --private-etc=file,directory,@group
The files installed by --private-etc are copies of the
original system files from /etc directory. By default,
the command brings in a skeleton of files and directories
used by most console tools:
$ firejail --private-etc dig debian.org
For X11/GTK/QT/Gnome/KDE programs add @x11 group as a
parameter. Example:
$ firejail --private-etc=@x11,gcrypt,python* gimp
gcrypt and /etc/python* directories are not part of the
generic @x11 group. File globbing is supported.
For games, add @games group:
$ firejail --private-etc=@games,@x11 warzone2100
Sound and networking files are included automatically,
unless --nosound or --net=none are specified. Files for
encrypted TLS/SSL protocol are in @tls-ca group.
$ firejail --private-etc=@tls-ca,wgetrc wget
https://debian.org
Note: The easiest way to extract the list of /etc files
accessed by your program is using strace utility:
$ strace /usr/bin/transmission-qt 2>&1 | grep open | grep
etc
--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-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
VALIDATION For simplicity, the same name validation is used for
multiple options. Rules:
The name must be 1-253 characters long. The name can only
contain ASCII letters, digits and the special characters "-._"
(that is, the name cannot contain spaces or control characters).
The name cannot contain only digits. The first and last
characters must be an ASCII letter or digit and the name may
contain special characters in the middle.
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.
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.
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]
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
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.
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.
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.
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
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
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⟩
This page is part of the Firejail (Firejail security sandbox)
project. Information about the project can be found at
⟨https://firejail.wordpress.com⟩. If you have a bug report for
this manual page, see ⟨https://firejail.wordpress.com/support/⟩.
This page was obtained from the project's upstream Git repository
⟨https://github.com/netblue30/firejail.git⟩ on 2024-06-14. (At
that time, the date of the most recent commit that was found in
the repository was 2024-06-14.) If you discover any rendering
problems in this HTML version of the page, or you believe there
is a better or more up-to-date source for the page, or you have
corrections or improvements to the information in this COLOPHON
(which is not part of the original manual page), send a mail to
man-pages@man7.org
0.9.73 Jun 2023 FIREJAIL(1)
Pages that refer to this page: firecfg(1), firemon(1), jailcheck(1), firejail-login(5), firejail-profile(5), firejail-users(5)
|
HTML rendering created 2024-06-26 by Michael Kerrisk, author of The Linux Programming Interface. For details of in-depth Linux/UNIX system programming training courses that I teach, look here. Hosting by jambit GmbH. |
|