chiark / gitweb /
man: mention /proc/self/mountinfo in systemd.mount(5)
[elogind.git] / man / systemd.mount.xml
index 4bf652323ef3cde559a4be2c512119383419995f..f319cf7ce9d36f8921f23e724ced28efc9437764 100644 (file)
@@ -44,7 +44,7 @@
 
         <refnamediv>
                 <refname>systemd.mount</refname>
-                <refpurpose>systemd mount configuration files</refpurpose>
+                <refpurpose>Mount unit configuration</refpurpose>
         </refnamediv>
 
         <refsynopsisdiv>
                 <citerefentry><refentrytitle>systemd.exec</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
                 which define the execution environment the
                 <citerefentry><refentrytitle>mount</refentrytitle><manvolnum>8</manvolnum></citerefentry>
-                binary is executed in.</para>
+                binary is executed in, and in
+                <citerefentry><refentrytitle>systemd.kill</refentrytitle><manvolnum>5</manvolnum></citerefentry>
+                which define the way the processes are
+                terminated. Note that the User= and Group= options are
+                not particularly useful for mount units specifying a
+                <literal>Type=</literal> option or using configuration
+                not specified in <filename>/etc/fstab</filename>;
+                <citerefentry><refentrytitle>mount</refentrytitle><manvolnum>8</manvolnum></citerefentry>
+                will refuse options that aren't listed in
+                <filename>/etc/fstab</filename> if it is not run as
+                UID 0.</para>
 
                 <para>Mount units must be named after the mount point
                 directories they control. Example: the mount point
                 mounting. See
                 <citerefentry><refentrytitle>systemd.automount</refentrytitle><manvolnum>5</manvolnum></citerefentry>.</para>
 
-                <para>If an mount point is beneath another mount point
+                <para>If a mount point is beneath another mount point
                 in the file system hierarchy, a dependency between both
                 units is created automatically.</para>
 
-                <para>Mount points created at runtime independent on
-                unit files or <filename>/etc/fstab</filename> will be
+                <para>Mount points created at runtime (independently of
+                unit files or <filename>/etc/fstab</filename>) will be
                 monitored by systemd and appear like any other mount
-                unit in systemd.</para>
+                unit in systemd.
+                See <filename>/proc/self/mountinfo</filename> description
+                in <citerefentry><refentrytitle>proc</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
+                </para>
+
+                <para>Some file systems have special semantics as API
+                file systems for kernel-to-userspace and
+                userspace-to-userpace interfaces. Some of them may not
+                be changed via mount units, and cannot be disabled.
+                For a longer discussion see <ulink
+                url="http://www.freedesktop.org/wiki/Software/systemd/APIFileSystems">API
+                File Systems</ulink>.</para>
         </refsect1>
 
         <refsect1>
                 should wait for a device to show up before giving up
                 on an entry from
                 <filename>/etc/fstab</filename>. Specify a time in
-                seconds or explicitly specifiy a unit as
+                seconds or explicitly specify a unit as
                 <literal>s</literal>, <literal>min</literal>,
                 <literal>h</literal>, <literal>ms</literal>.</para>
 
                 <para>If a mount point is configured in both
-                <filename>/etc/fstab</filename> and a unit file, the
-                configuration in the latter takes precedence.</para>
+                <filename>/etc/fstab</filename> and a unit file that
+                is stored below <filename>/usr</filename> the former
+                will take precedence. If the unit file is stored below
+                <filename>/etc</filename> it will take
+                precedence. This means: native unit files take
+                precedence over traditional configuration files, but
+                this is superseded by the rule that configuration in
+                <filename>/etc</filename> will always take precedence
+                over configuration in
+                <filename>/usr</filename>.</para>
         </refsect1>
 
         <refsect1>
                 supervises. A number of options that may be used in
                 this section are shared with other unit types. These
                 options are documented in
-                <citerefentry><refentrytitle>systemd.exec</refentrytitle><manvolnum>5</manvolnum></citerefentry>. The
+                <citerefentry><refentrytitle>systemd.exec</refentrytitle><manvolnum>5</manvolnum></citerefentry>
+                and
+                <citerefentry><refentrytitle>systemd.kill</refentrytitle><manvolnum>5</manvolnum></citerefentry>. The
                 options specific to the [Mount] section of mount
                 units are the following:</para>
 
                                 <term><varname>Where=</varname></term>
                                 <listitem><para>Takes an absolute path
                                 of a directory of the mount point. If
-                                the mount point is not existing at
+                                the mount point does not exist at the
                                 time of mounting, it is created. This
                                 string must be reflected in the unit
                                 file name. (See above.) This option is
                                 will be terminated forcibly via
                                 SIGTERM, and after another delay of
                                 this time with SIGKILL. (See
-                                <option>KillMode=</option> below.)
+                                <option>KillMode=</option> in
+                                <citerefentry><refentrytitle>systemd.kill</refentrytitle><manvolnum>5</manvolnum></citerefentry>.)
                                 Takes a unit-less value in seconds, or
                                 a time span value such as "5min
                                 20s". Pass 0 to disable the timeout
                                 logic. Defaults to
                                 90s.</para></listitem>
                         </varlistentry>
-
-                        <varlistentry>
-                                <term><varname>KillMode=</varname></term>
-                                <listitem><para>Specifies how
-                                processes of this mount shall be
-                                killed. One of
-                                <option>control-group</option>,
-                                <option>process</option>,
-                                <option>none</option>.</para>
-
-                                <para>This option is mostly equivalent
-                                to the <option>KillMode=</option>
-                                option of service files. See
-                                <citerefentry><refentrytitle>systemd.service</refentrytitle><manvolnum>5</manvolnum></citerefentry>
-                                for details.</para></listitem>
-                        </varlistentry>
-
-                        <varlistentry>
-                                <term><varname>KillSignal=</varname></term>
-                                <listitem><para>Specifies which signal
-                                to use when killing a process of this
-                                mount. Defaults to SIGTERM.
-                                </para></listitem>
-                        </varlistentry>
-
-                        <varlistentry>
-                                <term><varname>SendSIGKILL=</varname></term>
-                                <listitem><para>Specifies whether to
-                                send SIGKILL to remaining processes
-                                after a timeout, if the normal
-                                shutdown procedure left processes of
-                                the mount around. Takes a boolean
-                                value. Defaults to "yes".
-                                </para></listitem>
-                        </varlistentry>
-
                 </variablelist>
+
+                <para>Check
+                <citerefentry><refentrytitle>systemd.exec</refentrytitle><manvolnum>5</manvolnum></citerefentry>
+                and
+                <citerefentry><refentrytitle>systemd.kill</refentrytitle><manvolnum>5</manvolnum></citerefentry>
+                for more settings.</para>
         </refsect1>
 
         <refsect1>
                           <citerefentry><refentrytitle>systemctl</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
                           <citerefentry><refentrytitle>systemd.unit</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
                           <citerefentry><refentrytitle>systemd.exec</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
+                          <citerefentry><refentrytitle>systemd.kill</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
                           <citerefentry><refentrytitle>systemd.service</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
                           <citerefentry><refentrytitle>systemd.device</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
+                          <citerefentry><refentrytitle>proc</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
                           <citerefentry><refentrytitle>mount</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
-                          <citerefentry><refentrytitle>systemd-fstab-generator</refentrytitle><manvolnum>8</manvolnum></citerefentry>
+                          <citerefentry><refentrytitle>systemd-fstab-generator</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
+                          <citerefentry><refentrytitle>systemd.directives</refentrytitle><manvolnum>7</manvolnum></citerefentry>
                   </para>
         </refsect1>