3 <!DOCTYPE refentry PUBLIC
"-//OASIS//DTD DocBook XML V4.5//EN"
4 "http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd">
5 <!-- SPDX-License-Identifier: LGPL-2.1-or-later -->
6 <refentry id=
"systemd-fstab-generator">
9 <title>systemd-fstab-generator
</title>
10 <productname>systemd
</productname>
14 <refentrytitle>systemd-fstab-generator
</refentrytitle>
15 <manvolnum>8</manvolnum>
19 <refname>systemd-fstab-generator
</refname>
20 <refpurpose>Unit generator for /etc/fstab
</refpurpose>
24 <para><filename>/usr/lib/systemd/system-generators/systemd-fstab-generator
</filename></para>
28 <title>Description
</title>
30 <para><filename>systemd-fstab-generator
</filename> is a generator
31 that translates
<filename>/etc/fstab
</filename> (see
32 <citerefentry project='man-pages'
><refentrytitle>fstab
</refentrytitle><manvolnum>5</manvolnum></citerefentry>
33 for details) into native systemd units early at boot and when
34 configuration of the system manager is reloaded. This will
35 instantiate mount and swap units as necessary.
</para>
37 <para>The
<varname>passno
</varname> field is treated like a simple
38 boolean, and the ordering information is discarded. However, if
39 the root file system is checked, it is checked before all the
40 other file systems.
</para>
43 <citerefentry><refentrytitle>systemd.mount
</refentrytitle><manvolnum>5</manvolnum></citerefentry>
45 <citerefentry><refentrytitle>systemd.swap
</refentrytitle><manvolnum>5</manvolnum></citerefentry>
46 for more information about special
<filename>/etc/fstab
</filename>
47 mount options this generator understands.
</para>
49 <para>One special topic is handling of symbolic links. Historical init
50 implementations supported symlinks in
<filename>/etc/fstab
</filename>.
51 Because mount units will refuse mounts where the target is a symbolic link,
52 this generator will resolve any symlinks as far as possible when processing
53 <filename>/etc/fstab
</filename> in order to enhance backwards compatibility.
54 If a symlink target does not exist at the time that this generator runs, it
55 is assumed that the symlink target is the final target of the mount.
</para>
57 <para><filename>systemd-fstab-generator
</filename> implements
58 <citerefentry><refentrytitle>systemd.generator
</refentrytitle><manvolnum>7</manvolnum></citerefentry>.
</para>
62 <title>Kernel Command Line
</title>
64 <para><filename>systemd-fstab-generator
</filename> understands the
65 following kernel command line parameters:
</para>
67 <variablelist class='kernel-commandline-options'
>
70 <term><varname>fstab=
</varname></term>
71 <term><varname>rd.fstab=
</varname></term>
73 <listitem><para>Takes a boolean argument. Defaults to
<literal>yes
</literal>. If
74 <literal>no
</literal>, causes the generator to ignore any mounts or swap devices configured in
75 <filename>/etc/fstab
</filename>.
<varname>rd.fstab=
</varname> is honored only in the initrd, while
76 <varname>fstab=
</varname> is honored by both the main system and the initrd.
</para></listitem>
80 <term><varname>root=
</varname></term>
82 <listitem><para>Configures the operating system's root filesystem to mount when running in the
83 initrd. This accepts a device node path (usually
<filename>/dev/disk/by-uuid/…
</filename> or
84 <filename>/dev/disk/by-label/…
</filename> or similar), or the special values
<literal>gpt-auto
</literal>
85 and
<literal>tmpfs
</literal>.
</para>
87 <para>Use
<literal>gpt-auto
</literal> to explicitly request automatic root file system discovery via
88 <citerefentry><refentrytitle>systemd-gpt-auto-generator
</refentrytitle><manvolnum>8</manvolnum></citerefentry>.
</para>
90 <para>Use
<literal>tmpfs
</literal> in order to mount a
<citerefentry
91 project='man-pages'
><refentrytitle>tmpfs
</refentrytitle><manvolnum>5</manvolnum></citerefentry> file
92 system as root file system of the OS. This is useful in combination with
93 <varname>mount.usr=
</varname> (see below) in order to combine a volatile root file system with a
94 separate, immutable
<filename>/usr/
</filename> file system. Also see
95 <varname>systemd.volatile=
</varname> below.
</para></listitem>
99 <term><varname>rootfstype=
</varname></term>
101 <listitem><para>Takes the root filesystem type that will be
102 passed to the mount command.
<varname>rootfstype=
</varname> is
103 honored by the initrd.
</para></listitem>
107 <term><varname>rootflags=
</varname></term>
109 <listitem><para>Takes the root filesystem mount options to use.
<varname>rootflags=
</varname> is
110 honored by the initrd.
</para>
112 <para>Note that unlike most kernel command line options this setting does not override settings made
113 in configuration files (specifically: the mount option string in
114 <filename>/etc/fstab
</filename>). See
115 <citerefentry><refentrytitle>systemd-remount-fs.service
</refentrytitle><manvolnum>8</manvolnum></citerefentry>.
</para></listitem>
119 <term><varname>mount.usr=
</varname></term>
121 <listitem><para>Takes the
<filename>/usr/
</filename> filesystem
122 to be mounted by the initrd. If
123 <varname>mount.usrfstype=
</varname> or
124 <varname>mount.usrflags=
</varname> is set, then
125 <varname>mount.usr=
</varname> will default to the value set in
126 <varname>root=
</varname>.
</para>
128 <para>Otherwise, this parameter defaults to the
129 <filename>/usr/
</filename> entry found in
130 <filename>/etc/fstab
</filename> on the root filesystem.
</para>
132 <para><varname>mount.usr=
</varname> is honored by the initrd.
137 <term><varname>mount.usrfstype=
</varname></term>
139 <listitem><para>Takes the
<filename>/usr/
</filename> filesystem
140 type that will be passed to the mount command. If
141 <varname>mount.usr=
</varname> or
142 <varname>mount.usrflags=
</varname> is set, then
143 <varname>mount.usrfstype=
</varname> will default to the value
144 set in
<varname>rootfstype=
</varname>.
</para>
146 <para>Otherwise, this value will be read from the
147 <filename>/usr/
</filename> entry in
148 <filename>/etc/fstab
</filename> on the root filesystem.
</para>
150 <para><varname>mount.usrfstype=
</varname> is honored by the
151 initrd.
</para></listitem>
155 <term><varname>mount.usrflags=
</varname></term>
157 <listitem><para>Takes the
<filename>/usr/
</filename> filesystem
158 mount options to use. If
<varname>mount.usr=
</varname> or
159 <varname>mount.usrfstype=
</varname> is set, then
160 <varname>mount.usrflags=
</varname> will default to the value
161 set in
<varname>rootflags=
</varname>.
</para>
163 <para>Otherwise, this value will be read from the
164 <filename>/usr/
</filename> entry in
165 <filename>/etc/fstab
</filename> on the root filesystem.
</para>
167 <para><varname>mount.usrflags=
</varname> is honored by the
168 initrd.
</para></listitem>
172 <term><varname>roothash=
</varname></term>
173 <term><varname>usrhash=
</varname></term>
175 <listitem><para>These options are primarily read by
176 <citerefentry><refentrytitle>systemd-veritysetup-generator
</refentrytitle><manvolnum>8</manvolnum></citerefentry>. When
177 set this indicates that the root file system (or
<filename>/usr/
</filename>) shall be mounted from
178 Verity volumes with the specified hashes. If these kernel command line options are set the root (or
179 <filename>/usr/
</filename>) file system is thus mounted from a device mapper volume
180 <filename>/dev/mapper/root
</filename> (or
<filename>/dev/mapper/usr
</filename>).
</para></listitem>
184 <term><varname>systemd.volatile=
</varname></term>
186 <listitem><para>Controls whether the system shall boot up in volatile mode. Takes a boolean argument or the
187 special value
<option>state
</option>.
</para>
189 <para>If false (the default), this generator makes no changes to the mount tree and the system is booted up in
192 <para>If true the generator ensures
193 <citerefentry><refentrytitle>systemd-volatile-root.service
</refentrytitle><manvolnum>8</manvolnum></citerefentry>
194 is run in the initrd. This service changes the mount table before transitioning to the host system,
195 so that a volatile memory file system (
<literal>tmpfs
</literal>) is used as root directory, with only
196 <filename>/usr/
</filename> mounted into it from the configured root file system, in read-only mode.
197 This way the system operates in fully stateless mode, with all configuration and state reset at boot
198 and lost at shutdown, as
<filename>/etc/
</filename> and
<filename>/var/
</filename> will be served
199 from the (initially unpopulated) volatile memory file system.
</para>
201 <para>If set to
<option>state
</option> the generator will leave the root directory mount point unaltered,
202 however will mount a
<literal>tmpfs
</literal> file system to
<filename>/var/
</filename>. In this mode the normal
203 system configuration (i.e. the contents of
<literal>/etc/
</literal>) is in effect (and may be modified during
204 system runtime), however the system state (i.e. the contents of
<literal>/var/
</literal>) is reset at boot and
205 lost at shutdown.
</para>
207 <para>If this setting is set to
<literal>overlay
</literal> the root file system is set up as
208 <literal>overlayfs
</literal> mount combining the read-only root directory with a writable
209 <literal>tmpfs
</literal>, so that no modifications are made to disk, but the file system may be modified
210 nonetheless with all changes being lost at reboot.
</para>
212 <para>Note that in none of these modes the root directory,
<filename>/etc/
</filename>,
<filename>/var/
</filename>
213 or any other resources stored in the root file system are physically removed. It's thus safe to boot a system
214 that is normally operated in non-volatile mode temporarily into volatile mode, without losing data.
</para>
216 <para>Note that with the exception of
<literal>overlay
</literal> mode, enabling this setting will
217 only work correctly on operating systems that can boot up with only
<filename>/usr/
</filename>
218 mounted, and are able to automatically populate
<filename>/etc/
</filename>, and also
219 <filename>/var/
</filename> in case of
<literal>systemd.volatile=yes
</literal>.
</para>
221 <para>Also see
<varname>root=tmpfs
</varname> above, for a method to combine a
222 <literal>tmpfs
</literal> file system with a regular
<filename>/usr/
</filename> file system (as
223 configured via
<varname>mount.usr=
</varname>). The main distinction between
224 <varname>systemd.volatile=yes
</varname>, and
<varname>root=tmpfs
</varname> in combination
225 <varname>mount.usr=
</varname> is that the former operates on top of a regular root file system and
226 temporarily obstructs the files and directories above its
<filename>/usr/
</filename> subdirectory,
227 while the latter does not hide any files, but simply mounts a unpopulated tmpfs as root file system
228 and combines it with a user picked
<filename>/usr/
</filename> file system.
</para></listitem>
232 <term><varname>systemd.swap
</varname></term>
234 <listitem><para>Takes a boolean argument or enables the option if specified
235 without an argument. If disabled, causes the generator to ignore
236 any swap devices configured in
<filename>/etc/fstab
</filename>.
237 Defaults to enabled.
</para></listitem>
243 <title>See Also
</title>
245 <citerefentry><refentrytitle>systemd
</refentrytitle><manvolnum>1</manvolnum></citerefentry>,
246 <citerefentry project='man-pages'
><refentrytitle>fstab
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
247 <citerefentry><refentrytitle>systemd.mount
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
248 <citerefentry><refentrytitle>systemd.swap
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
249 <citerefentry><refentrytitle>systemd-cryptsetup-generator
</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
250 <citerefentry><refentrytitle>systemd-gpt-auto-generator
</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
251 <citerefentry><refentrytitle>kernel-command-line
</refentrytitle><manvolnum>7</manvolnum></citerefentry>,
252 <ulink url=
"https://systemd.io/ENVIRONMENT/">Known Environment Variables
</ulink>