1INITRD(4) Linux Programmer's Manual INITRD(4)
2
6 initrd - boot loader initialized RAM disk
7
9 /dev/initrd is a read-only block device assigned major number 1 and mi‐
10 nor number 250. Typically /dev/initrd is owned by root:disk with mode
11 0400 (read access by root only). If the Linux system does not have
12 /dev/initrd already created, it can be created with the following com‐
13 mands:
14 mknod -m 400 /dev/initrd b 1 250
16 chown root:disk /dev/initrd
17 Also, support for both "RAM disk" and "Initial RAM disk" (e.g., CON‐
19 FIG_BLK_DEV_RAM=y and CONFIG_BLK_DEV_INITRD=y) must be compiled di‐
20 rectly into the Linux kernel to use /dev/initrd. When using /dev/ini‐
21 trd, the RAM disk driver cannot be loaded as a module.
22
24 The special file /dev/initrd is a read-only block device. This device
25 is a RAM disk that is initialized (e.g., loaded) by the boot loader be‐
26 fore the kernel is started. The kernel then can use /dev/initrd's con‐
27 tents for a two-phase system boot-up.
28 In the first boot-up phase, the kernel starts up and mounts an initial
30 root filesystem from the contents of /dev/initrd (e.g., RAM disk ini‐
31 tialized by the boot loader). In the second phase, additional drivers
32 or other modules are loaded from the initial root device's contents.
33 After loading the additional modules, a new root filesystem (i.e., the
34 normal root filesystem) is mounted from a different device.
35 Boot-up operation
37 When booting up with initrd, the system boots as follows:
38 1. The boot loader loads the kernel program and /dev/initrd's contents
40 into memory.
41 2. On kernel startup, the kernel uncompresses and copies the contents
43 of the device /dev/initrd onto device /dev/ram0 and then frees the
44 memory used by /dev/initrd.
45 3. The kernel then read-write mounts the device /dev/ram0 as the ini‐
47 tial root filesystem.
48 4. If the indicated normal root filesystem is also the initial root
50 filesystem (e.g., /dev/ram0) then the kernel skips to the last step
51 for the usual boot sequence.
52 5. If the executable file /linuxrc is present in the initial root
54 filesystem, /linuxrc is executed with UID 0. (The file /linuxrc
55 must have executable permission. The file /linuxrc can be any valid
56 executable, including a shell script.)
57 6. If /linuxrc is not executed or when /linuxrc terminates, the normal
59 root filesystem is mounted. (If /linuxrc exits with any filesystems
60 mounted on the initial root filesystem, then the behavior of the
61 kernel is UNSPECIFIED. See the NOTES section for the current kernel
62 behavior.)
63 7. If the normal root filesystem has a directory /initrd, the device
65 /dev/ram0 is moved from / to /initrd. Otherwise, if the directory
66 /initrd does not exist, the device /dev/ram0 is unmounted. (When
67 moved from / to /initrd, /dev/ram0 is not unmounted and therefore
68 processes can remain running from /dev/ram0. If directory /initrd
69 does not exist on the normal root filesystem and any processes re‐
70 main running from /dev/ram0 when /linuxrc exits, the behavior of the
71 kernel is UNSPECIFIED. See the NOTES section for the current kernel
72 behavior.)
73 8. The usual boot sequence (e.g., invocation of /sbin/init) is per‐
75 formed on the normal root filesystem.
76 Options
78 The following boot loader options, when used with initrd, affect the
79 kernel's boot-up operation:
80 initrd=filename
82 Specifies the file to load as the contents of /dev/initrd. For
83 LOADLIN this is a command-line option. For LILO you have to use
84 this command in the LILO configuration file /etc/lilo.config.
85 The filename specified with this option will typically be a
86 gzipped filesystem image.
87 noinitrd
89 This boot option disables the two-phase boot-up operation. The
90 kernel performs the usual boot sequence as if /dev/initrd was
91 not initialized. With this option, any contents of /dev/initrd
92 loaded into memory by the boot loader contents are preserved.
93 This option permits the contents of /dev/initrd to be any data
94 and need not be limited to a filesystem image. However, device
95 /dev/initrd is read-only and can be read only one time after
96 system startup.
97 root=device-name
99 Specifies the device to be used as the normal root filesystem.
100 For LOADLIN this is a command-line option. For LILO this is a
101 boot time option or can be used as an option line in the LILO
102 configuration file /etc/lilo.config. The device specified by
103 this option must be a mountable device having a suitable root
104 filesystem.
105 Changing the normal root filesystem
107 By default, the kernel's settings (e.g., set in the kernel file with
108 rdev(8) or compiled into the kernel file), or the boot loader option
109 setting is used for the normal root filesystems. For an NFS-mounted
110 normal root filesystem, one has to use the nfs_root_name and
111 nfs_root_addrs boot options to give the NFS settings. For more infor‐
112 mation on NFS-mounted root see the kernel documentation file Documenta‐
113 tion/filesystems/nfs/nfsroot.txt (or Documentation/filesystems/nfs‐
114 root.txt before Linux 2.6.33). For more information on setting the
115 root filesystem see also the LILO and LOADLIN documentation.
116 It is also possible for the /linuxrc executable to change the normal
118 root device. For /linuxrc to change the normal root device, /proc must
119 be mounted. After mounting /proc, /linuxrc changes the normal root de‐
120 vice by writing into the proc files /proc/sys/kernel/real-root-dev,
121 /proc/sys/kernel/nfs-root-name, and /proc/sys/kernel/nfs-root-addrs.
122 For a physical root device, the root device is changed by having /lin‐
123 uxrc write the new root filesystem device number into /proc/sys/ker‐
124 nel/real-root-dev. For an NFS root filesystem, the root device is
125 changed by having /linuxrc write the NFS setting into files
126 /proc/sys/kernel/nfs-root-name and /proc/sys/kernel/nfs-root-addrs and
127 then writing 0xff (e.g., the pseudo-NFS-device number) into file
128 /proc/sys/kernel/real-root-dev. For example, the following shell com‐
129 mand line would change the normal root device to /dev/hdb1:
130 echo 0x365 >/proc/sys/kernel/real-root-dev
132 For an NFS example, the following shell command lines would change the
134 normal root device to the NFS directory /var/nfsroot on a local net‐
135 worked NFS server with IP number 193.8.232.7 for a system with IP num‐
136 ber 193.8.232.2 and named "idefix":
137 echo /var/nfsroot >/proc/sys/kernel/nfs-root-name
139 echo 193.8.232.2:193.8.232.7::255.255.255.0:idefix \
140 >/proc/sys/kernel/nfs-root-addrs
141 echo 255 >/proc/sys/kernel/real-root-dev
142 Note: The use of /proc/sys/kernel/real-root-dev to change the root
144 filesystem is obsolete. See the Linux kernel source file Documenta‐
145 tion/admin-guide/initrd.rst (or Documentation/initrd.txt before Linux
146 4.10) as well as pivot_root(2) and pivot_root(8) for information on the
147 modern method of changing the root filesystem.
148 Usage
150 The main motivation for implementing initrd was to allow for modular
151 kernel configuration at system installation.
152 A possible system installation scenario is as follows:
154 1. The loader program boots from floppy or other media with a minimal
156 kernel (e.g., support for /dev/ram, /dev/initrd, and the ext2
157 filesystem) and loads /dev/initrd with a gzipped version of the ini‐
158 tial filesystem.
159 2. The executable /linuxrc determines what is needed to (1) mount the
161 normal root filesystem (i.e., device type, device drivers, filesys‐
162 tem) and (2) the distribution media (e.g., CD-ROM, network, tape,
163 ...). This can be done by asking the user, by auto-probing, or by
164 using a hybrid approach.
165 3. The executable /linuxrc loads the necessary modules from the initial
167 root filesystem.
168 4. The executable /linuxrc creates and populates the root filesystem.
170 (At this stage the normal root filesystem does not have to be a com‐
171 pleted system yet.)
172 5. The executable /linuxrc sets /proc/sys/kernel/real-root-dev, un‐
174 mounts /proc, the normal root filesystem and any other filesystems
175 it has mounted, and then terminates.
176 6. The kernel then mounts the normal root filesystem.
178 7. Now that the filesystem is accessible and intact, the boot loader
180 can be installed.
181 8. The boot loader is configured to load into /dev/initrd a filesystem
183 with the set of modules that was used to bring up the system.
184 (e.g., device /dev/ram0 can be modified, then unmounted, and fi‐
185 nally, the image is written from /dev/ram0 to a file.)
186 9. The system is now bootable and additional installation tasks can be
188 performed.
189 The key role of /dev/initrd in the above is to reuse the configuration
191 data during normal system operation without requiring initial kernel
192 selection, a large generic kernel or, recompiling the kernel.
193 A second scenario is for installations where Linux runs on systems with
195 different hardware configurations in a single administrative network.
196 In such cases, it may be desirable to use only a small set of kernels
197 (ideally only one) and to keep the system-specific part of configura‐
198 tion information as small as possible. In this case, create a common
199 file with all needed modules. Then, only the /linuxrc file or a file
200 executed by /linuxrc would be different.
201 A third scenario is more convenient recovery disks. Because informa‐
203 tion like the location of the root filesystem partition is not needed
204 at boot time, the system loaded from /dev/initrd can use a dialog
205 and/or auto-detection followed by a possible sanity check.
206 Last but not least, Linux distributions on CD-ROM may use initrd for
208 easy installation from the CD-ROM. The distribution can use LOADLIN to
209 directly load /dev/initrd from CD-ROM without the need of any floppies.
210 The distribution could also use a LILO boot floppy and then bootstrap a
211 bigger RAM disk via /dev/initrd from the CD-ROM.
212
214 /dev/initrd
215 /dev/ram0
216 /linuxrc
217 /initrd
218
220 1. With the current kernel, any filesystems that remain mounted when
221 /dev/ram0 is moved from / to /initrd continue to be accessible.
222 However, the /proc/mounts entries are not updated.
223 2. With the current kernel, if directory /initrd does not exist, then
225 /dev/ram0 will not be fully unmounted if /dev/ram0 is used by any
226 process or has any filesystem mounted on it. If /dev/ram0 is not
227 fully unmounted, then /dev/ram0 will remain in memory.
228 3. Users of /dev/initrd should not depend on the behavior given in the
230 above notes. The behavior may change in future versions of the
231 Linux kernel.
232
234 chown(1), mknod(1), ram(4), freeramdisk(8), rdev(8)
235 Documentation/admin-guide/initrd.rst (or Documentation/initrd.txt be‐
237 fore Linux 4.10) in the Linux kernel source tree, the LILO documenta‐
238 tion, the LOADLIN documentation, the SYSLINUX documentation
239
241 This page is part of release 5.12 of the Linux man-pages project. A
242 description of the project, information about reporting bugs, and the
243 latest version of this page, can be found at
244 https://www.kernel.org/doc/man-pages/.
245Linux 2021-03-22 INITRD(4)