chiark / gitweb /
fix spelling of privilege
[elogind.git] / man / systemd.exec.xml
index 5721dc15537e9e1901264ccbeccae930d86fe87e..3f27d13c38201471e5ee2f2a9c76ce3eac931d8b 100644 (file)
@@ -57,7 +57,7 @@
                 <title>Description</title>
 
                 <para>Unit configuration files for services, sockets,
                 <title>Description</title>
 
                 <para>Unit configuration files for services, sockets,
-                mount points and swap devices share a subset of
+                mount points, and swap devices share a subset of
                 configuration options which define the execution
                 environment of spawned processes.</para>
 
                 configuration options which define the execution
                 environment of spawned processes.</para>
 
                 configuration options are configured in the [Service],
                 [Socket], [Mount], or [Swap] sections, depending on the unit
                 type.</para>
                 configuration options are configured in the [Service],
                 [Socket], [Mount], or [Swap] sections, depending on the unit
                 type.</para>
-
-                <para>Processes started by the system systemd instance
-                are executed in a clean environment in which only the
-                <varname>$PATH</varname> and <varname>$LANG</varname>
-                variables are set by default. In order to add
-                additional variables, see the
-                <varname>Environment=</varname> and
-                <varname>EnvironmentFile=</varname> options below. To
-                specify variables globally, see
-                <varname>DefaultEnvironment=</varname> in
-                <citerefentry><refentrytitle>systemd-system.conf</refentrytitle><manvolnum>5</manvolnum></citerefentry>
-                or the kernel option
-                <varname>systemd.setenv=</varname> in
-                <citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry>. Processes
-                started by the user systemd instances inherit all
-                environment variables from the user systemd instance,
-                and have <varname>$HOME</varname>,
-                <varname>$USER</varname>,
-                <varname>$XDG_RUNTIME_DIR</varname> defined, among
-                others. In addition, <varname>$MANAGERPID</varname>
-                contains the PID of the user systemd instance.</para>
         </refsect1>
 
         <refsect1>
         </refsect1>
 
         <refsect1>
                                 <listitem><para>Controls the CPU
                                 affinity of the executed
                                 processes. Takes a space-separated
                                 <listitem><para>Controls the CPU
                                 affinity of the executed
                                 processes. Takes a space-separated
-                                list of CPU indexes. This option may
+                                list of CPU indices. This option may
                                 be specified more than once in which
                                 case the specificed CPU affinity masks
                                 are merged. If the empty string is
                                 be specified more than once in which
                                 case the specificed CPU affinity masks
                                 are merged. If the empty string is
                                 for the assignment.</para>
 
                                 <para>Example:
                                 for the assignment.</para>
 
                                 <para>Example:
-                                <programlisting>Environment="VAR1=word1 word2" VAR2=word3 "VAR3=word 5 6"</programlisting>
+                                <programlisting>Environment="VAR1=word1 word2" VAR2=word3 "VAR3=$word 5 6"</programlisting>
                                 gives three variables <literal>VAR1</literal>,
                                 gives three variables <literal>VAR1</literal>,
-                                <literal>VAR2</literal>, <literal>VAR3</literal>.
+                                <literal>VAR2</literal>, <literal>VAR3</literal>
+                                with the values <literal>word1 word2</literal>,
+                                <literal>word3</literal>, <literal>$word 5 6</literal>.
                                 </para>
 
                                 <para>
                                 </para>
 
                                 <para>
 
                                 <para>The files listed with this
                                 directive will be read shortly before
 
                                 <para>The files listed with this
                                 directive will be read shortly before
-                                the process is executed. Settings from
-                                these files override settings made
-                                with
+                                the process is executed (more
+                                specifically, after all
+                                processes from a previous unit state
+                                terminated. This means you can
+                                generate these files in one unit
+                                state, and read it with this option in
+                                the next). Settings from these files
+                                override settings made with
                                 <varname>Environment=</varname>. If
                                 the same variable is set twice from
                                 these files, the files will be read in
                                 <varname>Environment=</varname>. If
                                 the same variable is set twice from
                                 these files, the files will be read in
                         <varlistentry>
                                 <term><varname>StandardError=</varname></term>
                                 <listitem><para>Controls where file
                         <varlistentry>
                                 <term><varname>StandardError=</varname></term>
                                 <listitem><para>Controls where file
-                                descriptor 2 (STDERR) of the executed
-                                processes is connected to. The
-                                available options are identical to
+                                descriptor 2 (STDERR) of the
+                                executed processes is connected to.
+                                The available options are identical to
                                 those of
                                 <varname>StandardOutput=</varname>,
                                 with one exception: if set to
                                 those of
                                 <varname>StandardOutput=</varname>,
                                 with one exception: if set to
                         <varlistentry>
                                 <term><varname>TTYPath=</varname></term>
                                 <listitem><para>Sets the terminal
                         <varlistentry>
                                 <term><varname>TTYPath=</varname></term>
                                 <listitem><para>Sets the terminal
-                                device node to use if standard input,
-                                output or stderr are connected to a
+                                device node to use if standard input, output,
+                                or error are connected to a
                                 TTY (see above). Defaults to
                                 <filename>/dev/console</filename>.</para></listitem>
                         </varlistentry>
                                 TTY (see above). Defaults to
                                 <filename>/dev/console</filename>.</para></listitem>
                         </varlistentry>
                                 for details.</para></listitem>
                         </varlistentry>
 
                                 for details.</para></listitem>
                         </varlistentry>
 
-                        <varlistentry>
-                                <term><varname>TCPWrapName=</varname></term>
-                                <listitem><para>If this is a
-                                socket-activated service, this sets the
-                                tcpwrap service name to check the
-                                permission for the current connection
-                                with. This is only useful in
-                                conjunction with socket-activated
-                                services, and stream sockets (TCP) in
-                                particular. It has no effect on other
-                                socket types (e.g. datagram/UDP) and
-                                on processes unrelated to socket-based
-                                activation. If the tcpwrap
-                                verification fails, daemon start-up
-                                will fail and the connection is
-                                terminated. See
-                                <citerefentry><refentrytitle>tcpd</refentrytitle><manvolnum>8</manvolnum></citerefentry>
-                                for details. Note that this option may
-                                be used to do access control checks
-                                only. Shell commands and commands
-                                described in
-                                <citerefentry><refentrytitle>hosts_options</refentrytitle><manvolnum>5</manvolnum></citerefentry>
-                                are not supported.</para></listitem>
-                        </varlistentry>
-
                         <varlistentry>
                                 <term><varname>CapabilityBoundingSet=</varname></term>
 
                         <varlistentry>
                                 <term><varname>CapabilityBoundingSet=</varname></term>
 
                                 may be prefixed with
                                 <literal>-</literal>, in which case
                                 they will be ignored when they do not
                                 may be prefixed with
                                 <literal>-</literal>, in which case
                                 they will be ignored when they do not
-                                exist.</para></listitem>
+                                exist. Note that using this
+                                setting will disconnect propagation of
+                                mounts from the service to the host
+                                (propagation in the opposite direction
+                                continues to work). This means that
+                                this setting may not be used for
+                                services which shall be able to
+                                install mount points in the main mount
+                                namespace.</para></listitem>
                         </varlistentry>
 
                         <varlistentry>
                         </varlistentry>
 
                         <varlistentry>
                                 system namespace for the executed
                                 processes and mounts private
                                 <filename>/tmp</filename> and
                                 system namespace for the executed
                                 processes and mounts private
                                 <filename>/tmp</filename> and
-                                <filename>/var/tmp</filename> directories
-                                inside it, that are not shared by
-                                processes outside of the
+                                <filename>/var/tmp</filename>
+                                directories inside it that is not
+                                shared by processes outside of the
                                 namespace. This is useful to secure
                                 access to temporary files of the
                                 process, but makes sharing between
                                 processes via
                                 <filename>/tmp</filename> or
                                 <filename>/var/tmp</filename>
                                 namespace. This is useful to secure
                                 access to temporary files of the
                                 process, but makes sharing between
                                 processes via
                                 <filename>/tmp</filename> or
                                 <filename>/var/tmp</filename>
-                                impossible. All temporary data created
-                                by service will be removed after service
-                                is stopped. Defaults to
-                                false.</para></listitem>
+                                impossible. If this is enabled, all
+                                temporary files created by a service
+                                in these directories will be removed
+                                after the service is stopped. Defaults
+                                to false. It is possible to run two or
+                                more units within the same private
+                                <filename>/tmp</filename> and
+                                <filename>/var/tmp</filename>
+                                namespace by using the
+                                <varname>JoinsNamespaceOf=</varname>
+                                directive, see
+                                <citerefentry><refentrytitle>systemd.unit</refentrytitle><manvolnum>5</manvolnum></citerefentry>
+                                for details. Note that using this
+                                setting will disconnect propagation of
+                                mounts from the service to the host
+                                (propagation in the opposite direction
+                                continues to work). This means that
+                                this setting may not be used for
+                                services which shall be able to install
+                                mount points in the main mount
+                                namespace.</para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>PrivateDevices=</varname></term>
+
+                                <listitem><para>Takes a boolean
+                                argument. If true, sets up a new /dev
+                                namespace for the executed processes
+                                and only adds API pseudo devices such
+                                as <filename>/dev/null</filename>,
+                                <filename>/dev/zero</filename> or
+                                <filename>/dev/random</filename> (as
+                                well as the pseudo TTY subsystem) to
+                                it, but no physical devices such as
+                                <filename>/dev/sda</filename>. This is
+                                useful to securely turn off physical
+                                device access by the executed
+                                process. Defaults to false. Enabling
+                                this option will also remove
+                                <constant>CAP_MKNOD</constant> from
+                                the capability bounding set for the
+                                unit (see above), and set
+                                <varname>DevicePolicy=closed</varname>
+                                (see
+                                <citerefentry><refentrytitle>systemd.resource-control</refentrytitle><manvolnum>5</manvolnum></citerefentry>
+                                for details). Note that using this
+                                setting will disconnect propagation of
+                                mounts from the service to the host
+                                (propagation in the opposite direction
+                                continues to work). This means that
+                                this setting may not be used for
+                                services which shall be able to
+                                install mount points in the main mount
+                                namespace.</para></listitem>
                         </varlistentry>
 
                         <varlistentry>
                         </varlistentry>
 
                         <varlistentry>
                                 available to the executed process.
                                 This is useful to securely turn off
                                 network access by the executed
                                 available to the executed process.
                                 This is useful to securely turn off
                                 network access by the executed
-                                process. Defaults to
-                                false.</para></listitem>
+                                process. Defaults to false. It is
+                                possible to run two or more units
+                                within the same private network
+                                namespace by using the
+                                <varname>JoinsNamespaceOf=</varname>
+                                directive, see
+                                <citerefentry><refentrytitle>systemd.unit</refentrytitle><manvolnum>5</manvolnum></citerefentry>
+                                for details. Note that this option
+                                will disconnect all socket families
+                                from the host, this includes
+                                AF_NETLINK and AF_UNIX. The latter has
+                                the effect that AF_UNIX sockets in the
+                                abstract socket namespace will become
+                                unavailable to the processes (however,
+                                those located in the file system will
+                                continue to be
+                                accessible).</para></listitem>
                         </varlistentry>
 
                         <varlistentry>
                         </varlistentry>
 
                         <varlistentry>
                                 <option>shared</option>,
                                 <option>slave</option> or
                                 <option>private</option>, which
                                 <option>shared</option>,
                                 <option>slave</option> or
                                 <option>private</option>, which
-                                control whether the file system
-                                namespace set up for this unit's
-                                processes will receive or propagate
-                                new mounts. See
+                                control whether mounts in the file
+                                system namespace set up for this
+                                unit's processes will receive or
+                                propagate mounts or unmounts. See
                                 <citerefentry><refentrytitle>mount</refentrytitle><manvolnum>2</manvolnum></citerefentry>
                                 <citerefentry><refentrytitle>mount</refentrytitle><manvolnum>2</manvolnum></citerefentry>
-                                for details. Default to
-                                <option>shared</option>.</para></listitem>
+                                for details. Defaults to
+                                <option>shared</option>. Use
+                                <option>shared</option> to ensure that
+                                mounts and unmounts are propagated
+                                from the host to the container and
+                                vice versa. Use <option>slave</option>
+                                to run processes so that none of their
+                                mounts and unmounts will propagate to
+                                the host. Use <option>private</option>
+                                to also ensure that no mounts and
+                                unmounts from the host will propagate
+                                into the unit processes'
+                                namespace. Note that
+                                <option>slave</option> means that file
+                                systems mounted on the host might stay
+                                mounted continously in the unit's
+                                namespace, and thus keep the device
+                                busy. Note that the file system
+                                namespace related options
+                                (<varname>PrivateTmp=</varname>,
+                                <varname>PrivateDevices=</varname>,
+                                <varname>ReadOnlyDirectories=</varname>,
+                                <varname>InaccessibleDirectories=</varname>
+                                and
+                                <varname>ReadWriteDirectories=</varname>)
+                                require that mount and unmount
+                                propagation from the unit's file
+                                system namespace is disabled, and
+                                hence downgrade
+                                <option>shared</option> to
+                                <option>slave</option>.
+                                </para></listitem>
                         </varlistentry>
 
                         <varlistentry>
                         </varlistentry>
 
                         <varlistentry>
                                 this service.</para></listitem>
                         </varlistentry>
 
                                 this service.</para></listitem>
                         </varlistentry>
 
+                        <varlistentry>
+                                <term><varname>SELinuxContext=</varname></term>
+
+                                <listitem><para>Set the SELinux
+                                security context of the executed
+                                process. If set, this will override
+                                the automated domain
+                                transition. However, the policy still
+                                needs to autorize the transition. This
+                                directive is ignored if SELinux is
+                                disabled. If prefixed by
+                                <literal>-</literal>, all errors will
+                                be ignored. See
+                                <citerefentry><refentrytitle>setexeccon</refentrytitle><manvolnum>3</manvolnum></citerefentry>
+                                for details.</para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>AppArmorProfile=</varname></term>
+
+                                <listitem><para>Takes a profile name as argument.
+                                The process executed by the unit will switch to
+                                this profile when started. Profiles must already
+                                be loaded in the kernel, or the unit will fail.
+                                This result in a non operation if AppArmor is not
+                                enabled. If prefixed by <literal>-</literal>, all errors
+                                will be ignored.
+                                </para></listitem>
+                        </varlistentry>
+
                         <varlistentry>
                                 <term><varname>IgnoreSIGPIPE=</varname></term>
 
                         <varlistentry>
                                 <term><varname>IgnoreSIGPIPE=</varname></term>
 
                         <varlistentry>
                                 <term><varname>SystemCallFilter=</varname></term>
 
                         <varlistentry>
                                 <term><varname>SystemCallFilter=</varname></term>
 
-                                <listitem><para>Takes a space-separated
-                                list of system call
+                                <listitem><para>Takes a
+                                space-separated list of system call
                                 names. If this setting is used, all
                                 system calls executed by the unit
                                 names. If this setting is used, all
                                 system calls executed by the unit
-                                process except for the listed ones
+                                processes except for the listed ones
                                 will result in immediate process
                                 termination with the
                                 <constant>SIGSYS</constant> signal
                                 will result in immediate process
                                 termination with the
                                 <constant>SIGSYS</constant> signal
                                 the effect is inverted: only the
                                 listed system calls will result in
                                 immediate process termination
                                 the effect is inverted: only the
                                 listed system calls will result in
                                 immediate process termination
-                                (blacklisting). If this option is used,
+                                (blacklisting). If running in user
+                                mode and this option is used,
                                 <varname>NoNewPrivileges=yes</varname>
                                 <varname>NoNewPrivileges=yes</varname>
-                                is implied. This feature makes use of
-                                the Secure Computing Mode 2 interfaces
-                                of the kernel ('seccomp filtering')
-                                and is useful for enforcing a minimal
+                                is implied. This feature makes use of the
+                                Secure Computing Mode 2 interfaces of
+                                the kernel ('seccomp filtering') and
+                                is useful for enforcing a minimal
                                 sandboxing environment. Note that the
                                 <function>execve</function>,
                                 <function>rt_sigreturn</function>,
                                 sandboxing environment. Note that the
                                 <function>execve</function>,
                                 <function>rt_sigreturn</function>,
                                 merged. If the empty string is
                                 assigned, the filter is reset, all
                                 prior assignments will have no
                                 merged. If the empty string is
                                 assigned, the filter is reset, all
                                 prior assignments will have no
-                                effect.</para></listitem>
+                                effect.</para>
+
+                                <para>If you specify both types of
+                                this option (i.e. whitelisting and
+                                blacklisting), the first encountered
+                                will take precedence and will dictate
+                                the default action (termination or
+                                approval of a system call). Then the
+                                next occurrences of this option will
+                                add or delete the listed system calls
+                                from the set of the filtered system
+                                calls, depending of its type and the
+                                default action. (For example, if you have started
+                                with a whitelisting of
+                                <function>read</function> and
+                                <function>write</function>, and right
+                                after it add a blacklisting of
+                                <function>write</function>, then
+                                <function>write</function> will be
+                                removed from the set.)
+                                </para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>SystemCallErrorNumber=</varname></term>
+
+                                <listitem><para>Takes an
+                                <literal>errno</literal> error number
+                                name to return when the system call
+                                filter configured with
+                                <varname>SystemCallFilter=</varname>
+                                is triggered, instead of terminating
+                                the process immediately. Takes an
+                                error name such as
+                                <constant>EPERM</constant>,
+                                <constant>EACCES</constant> or
+                                <constant>EUCLEAN</constant>. When this
+                                setting is not used, or when the empty
+                                string is assigned, the process will be
+                                terminated immediately when the filter
+                                is triggered.</para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>SystemCallArchitectures=</varname></term>
+
+                                <listitem><para>Takes a space
+                                separated list of architecture
+                                identifiers to include in the system
+                                call filter. The known architecture
+                                identifiers are
+                                <constant>x86</constant>,
+                                <constant>x86-64</constant>,
+                                <constant>x32</constant>,
+                                <constant>arm</constant> as well as
+                                the special identifier
+                                <constant>native</constant>. Only
+                                system calls of the specified
+                                architectures will be permitted to
+                                processes of this unit. This is an
+                                effective way to disable compatibility
+                                with non-native architectures for
+                                processes, for example to prohibit
+                                execution of 32-bit x86 binaries on
+                                64-bit x86-64 systems. The special
+                                <constant>native</constant> identifier
+                                implicitly maps to the native
+                                architecture of the system (or more
+                                strictly: to the architecture the
+                                system manager is compiled for). If
+                                running in user mode and this option
+                                is used,
+                                <varname>NoNewPrivileges=yes</varname>
+                                is implied. Note that setting this
+                                option to a non-empty list implies
+                                that <constant>native</constant> is
+                                included too. By default, this option
+                                is set to the empty list, i.e. no
+                                architecture system call filtering is
+                                applied.</para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>RestrictAddressFamilies=</varname></term>
+
+                                <listitem><para>Restricts the set of
+                                socket address families accessible to
+                                the processes of this unit. Takes a
+                                space-separated list of address family
+                                names to whitelist, such as
+                                <constant>AF_UNIX</constant>,
+                                <constant>AF_INET</constant> or
+                                <constant>AF_INET6</constant>. When
+                                prefixed with <constant>~</constant>
+                                the listed address families will be
+                                applied as blacklist, otherwise as
+                                whitelist. Note that this restricts
+                                access to the
+                                <citerefentry><refentrytitle>socket</refentrytitle><manvolnum>2</manvolnum></citerefentry>
+                                system call only. Sockets passed into
+                                the process by other means (for
+                                example, by using socket activation
+                                with socket units, see
+                                <citerefentry><refentrytitle>systemd.socket</refentrytitle><manvolnum>5</manvolnum></citerefentry>)
+                                are unaffected. Also, sockets created
+                                with <function>socketpair()</function>
+                                (which creates connected AF_UNIX
+                                sockets only) are unaffected. Note
+                                that this option has no effect on
+                                32-bit x86 and is ignored (but works
+                                correctly on x86-64). If running in user
+                                mode and this option is used,
+                                <varname>NoNewPrivileges=yes</varname>
+                                is implied. By default, no
+                                restriction applies, all address
+                                families are accessible to
+                                processes. If assigned the empty
+                                string, any previous list changes are
+                                undone.</para>
+
+                                <para>Use this option to limit
+                                exposure of processes to remote
+                                systems, in particular via exotic
+                                network protocols. Note that in most
+                                cases, the local
+                                <constant>AF_UNIX</constant> address
+                                family should be included in the
+                                configured whitelist as it is
+                                frequently used for local
+                                communication, including for
+                                <citerefentry><refentrytitle>syslog</refentrytitle><manvolnum>2</manvolnum></citerefentry>
+                                logging.</para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>Personality=</varname></term>
+
+                                <listitem><para>Controls which
+                                kernel architecture
+                                <citerefentry><refentrytitle>uname</refentrytitle><manvolnum>2</manvolnum></citerefentry>
+                                shall report, when invoked by unit
+                                processes. Takes one of
+                                <constant>x86</constant> and
+                                <constant>x86-64</constant>. This is
+                                useful when running 32-bit services on
+                                a 64-bit host system. If not specified,
+                                the personality is left unmodified and
+                                thus reflects the personality of the
+                                host system's
+                                kernel.</para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>RuntimeDirectory=</varname></term>
+                                <term><varname>RuntimeDirectoryMode=</varname></term>
+
+                                <listitem><para>Takes a list of
+                                directory names. If set, one or more
+                                directories by the specified names
+                                will be created below
+                                <filename>/run</filename> (for system
+                                services) or below
+                                <varname>$XDG_RUNTIME_DIR</varname>
+                                (for user services) when the unit is
+                                started, and removed when the unit is
+                                stopped. The directories will have the
+                                access mode specified in
+                                <varname>RuntimeDirectoryMode=</varname>,
+                                and will be owned by the user and
+                                group specified in
+                                <varname>User=</varname> and
+                                <varname>Group=</varname>. Use this to
+                                manage one or more runtime directories
+                                of the unit and bind their lifetime to
+                                the daemon runtime. The specified
+                                directory names must be relative, and
+                                may not include a
+                                <literal>/</literal>, i.e. must refer
+                                to simple directories to create or
+                                remove. This is particularly useful
+                                for unprivileged daemons that cannot
+                                create runtime directories in
+                                <filename>/run</filename> due to lack
+                                of privileges, and to make sure the
+                                runtime directory is cleaned up
+                                automatically after use. For runtime
+                                directories that require more complex
+                                or different configuration or lifetime
+                                guarantees, please consider using
+                                <citerefentry><refentrytitle>tmpfiles.d</refentrytitle><manvolnum>5</manvolnum></citerefentry>.</para></listitem>
                         </varlistentry>
 
                 </variablelist>
         </refsect1>
 
                         </varlistentry>
 
                 </variablelist>
         </refsect1>
 
+        <refsect1>
+                <title>Environment variables in spawned processes</title>
+
+                <para>Processes started by the system are executed in
+                a clean environment in which select variables
+                listed below are set. System processes started by systemd
+                do not inherit variables from PID 1, but processes
+                started by user systemd instances inherit all
+                environment variables from the user systemd instance.
+                </para>
+
+                <variablelist class='environment-variables'>
+                        <varlistentry>
+                                <term><varname>$PATH</varname></term>
+
+                                <listitem><para>Colon-separated list
+                                of directiories to use when launching
+                                executables. Systemd uses a fixed
+                                value of
+                                <filename>/usr/local/sbin</filename>:<filename>/usr/local/bin</filename>:<filename>/usr/sbin</filename>:<filename>/usr/bin</filename>:<filename>/sbin</filename>:<filename>/bin</filename>.
+                                </para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>$LANG</varname></term>
+
+                                <listitem><para>Locale. Can be set in
+                                <citerefentry><refentrytitle>locale.conf</refentrytitle><manvolnum>5</manvolnum></citerefentry>
+                                or on the kernel command line (see
+                                <citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry>
+                                and
+                                <citerefentry><refentrytitle>kernel-command-line</refentrytitle><manvolnum>7</manvolnum></citerefentry>).
+                                </para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>$USER</varname></term>
+                                <term><varname>$LOGNAME</varname></term>
+                                <term><varname>$HOME</varname></term>
+                                <term><varname>$SHELL</varname></term>
+
+                                <listitem><para>User name (twice), home
+                                directory, and the login shell.
+                                The variables are set for the units that
+                                have <varname>User=</varname> set,
+                                which includes user
+                                <command>systemd</command> instances.
+                                See
+                                <citerefentry><refentrytitle>passwd</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
+                                </para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>$XDG_RUNTIME_DIR</varname></term>
+
+                                <listitem><para>The directory for volatile
+                                state. Set for the user <command>systemd</command>
+                                instance, and also in user sessions.
+                                See
+                                <citerefentry><refentrytitle>pam_systemd</refentrytitle><manvolnum>8</manvolnum></citerefentry>.
+                                </para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>$XDG_SESSION_ID</varname></term>
+                                <term><varname>$XDG_SEAT</varname></term>
+                                <term><varname>$XDG_VTNR</varname></term>
+
+                                <listitem><para>The identifier of the
+                                session, the seat name, and
+                                virtual terminal of the session. Set
+                                by
+                                <citerefentry><refentrytitle>pam_systemd</refentrytitle><manvolnum>8</manvolnum></citerefentry>
+                                for login sessions.
+                                <varname>$XDG_SEAT</varname> and
+                                <varname>$XDG_VTNR</varname> will
+                                only be set when attached to a seat and a
+                                tty.</para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>$MAINPID</varname></term>
+
+                                <listitem><para>The PID of the units
+                                main process if it is known. This is
+                                only set for control processes as
+                                invoked by
+                                <varname>ExecReload=</varname> and
+                                similar.  </para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>$MANAGERPID</varname></term>
+
+                                <listitem><para>The PID of the user
+                                <command>systemd</command> instance,
+                                set for processes spawned by it.
+                                </para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>$LISTEN_FDS</varname></term>
+                                <term><varname>$LISTEN_PID</varname></term>
+
+                                <listitem><para>Information about file
+                                descriptors passed to a service for
+                                socket activation.  See
+                                <citerefentry><refentrytitle>sd_listen_fds</refentrytitle><manvolnum>3</manvolnum></citerefentry>.
+                                </para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>$TERM</varname></term>
+
+                                <listitem><para>Terminal type, set
+                                only for units connected to a terminal
+                                (<varname>StandardInput=tty</varname>,
+                                <varname>StandardOutput=tty</varname>,
+                                or
+                                <varname>StandardError=tty</varname>).
+                                See
+                                <citerefentry><refentrytitle>termcap</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
+                                </para></listitem>
+                        </varlistentry>
+                </variablelist>
+
+                <para>Additional variables may be configured by the
+                following means: for processes spawned in specific
+                units, use the <varname>Environment=</varname> and
+                <varname>EnvironmentFile=</varname> options above; to
+                specify variables globally, use
+                <varname>DefaultEnvironment=</varname> (see
+                <citerefentry><refentrytitle>systemd-system.conf</refentrytitle><manvolnum>5</manvolnum></citerefentry>)
+                or the kernel option
+                <varname>systemd.setenv=</varname> (see
+                <citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry>). Additional
+                variables may also be set through PAM,
+                cf. <citerefentry><refentrytitle>pam_env</refentrytitle><manvolnum>8</manvolnum></citerefentry>.</para>
+        </refsect1>
+
         <refsect1>
                   <title>See Also</title>
                   <para>
         <refsect1>
                   <title>See Also</title>
                   <para>
                           <citerefentry><refentrytitle>systemd.swap</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
                           <citerefentry><refentrytitle>systemd.mount</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
                           <citerefentry><refentrytitle>systemd.kill</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
                           <citerefentry><refentrytitle>systemd.swap</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
                           <citerefentry><refentrytitle>systemd.mount</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
                           <citerefentry><refentrytitle>systemd.kill</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
-                          <citerefentry><refentrytitle>systemd.cgroup</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
-                          <citerefentry><refentrytitle>systemd.directives</refentrytitle><manvolnum>7</manvolnum></citerefentry>
+                          <citerefentry><refentrytitle>systemd.resource-control</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
+                          <citerefentry><refentrytitle>systemd.directives</refentrytitle><manvolnum>7</manvolnum></citerefentry>,
+                          <citerefentry><refentrytitle>tmpfiles.d</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
+                          <citerefentry><refentrytitle>exec</refentrytitle><manvolnum>3</manvolnum></citerefentry>
                   </para>
         </refsect1>
 
                   </para>
         </refsect1>