]> git.ipfire.org Git - thirdparty/dracut.git/blob - dracut.asc
fix(zfcp_rules): correct shellcheck regression when parsing ccw args
[thirdparty/dracut.git] / dracut.asc
1 dracut
2 ======
3 Harald Hoyer <harald@redhat.com>
4 v3.0, October 2013
5
6 :language: bash
7
8 = Introduction
9 This section is a modified version of
10 http://en.wikipedia.org/wiki/Initrd which is licensed under the
11 Creative Commons Attribution/Share-Alike License.
12
13 == Definition
14 An _initial ramdisk_ is a temporary file system used in the boot process of the
15 Linux kernel. _initrd_ and _initramfs_ refer to slightly different schemes for
16 loading this file system into memory. Both are commonly used to make
17 preparations before the real root file system can be mounted.
18
19 == Rationale
20 Many Linux distributions ship a single, generic kernel image that is intended to
21 boot as wide a variety of hardware as possible. The device drivers for this
22 generic kernel image are included as loadable modules, as it is not possible to
23 statically compile them all into the one kernel without making it too large to
24 boot from computers with limited memory or from lower-capacity media like floppy
25 disks.
26
27 This then raises the problem of detecting and loading the modules necessary to
28 mount the root file system at boot time (or, for that matter, deducing where or
29 what the root file system is).
30
31 To further complicate matters, the root file system may be on a software RAID
32 volume, LVM, NFS (on diskless workstations), or on an encrypted partition. All
33 of these require special preparations to mount.
34
35 Another complication is kernel support for hibernation, which suspends the
36 computer to disk by dumping an image of the entire system to a swap partition or
37 a regular file, then powering off. On next boot, this image has to be made
38 accessible before it can be loaded back into memory.
39
40 To avoid having to hardcode handling for so many special cases into the kernel,
41 an initial boot stage with a temporary root file system
42 —now dubbed early user space— is used. This root file system would contain
43 user-space helpers that would do the hardware detection, module loading and
44 device discovery necessary to get the real root file system mounted.
45
46 == Implementation
47 An image of this initial root file system (along with the kernel image) must be
48 stored somewhere accessible by the Linux bootloader or the boot firmware of the
49 computer. This can be:
50
51 * The root file system itself
52 * A boot image on an optical disc
53 * A small ext2/ext3 or FAT-formatted partition on a local disk
54 (a _boot partition_)
55 * A TFTP server (on systems that can boot from Ethernet)
56
57 The bootloader will load the kernel and initial root file system image into
58 memory and then start the kernel, passing in the memory address of the image.
59
60 Depending on which algorithms were compiled statically into it, the kernel can
61 currently unpack initrd/initramfs images compressed with gzip, bzip2 and LZMA.
62
63 == Mount preparations
64 dracut can generate a customized initrams image which contains only whatever is
65 necessary to boot some particular computer, such as ATA, SCSI and filesystem
66 kernel modules (host-only mode).
67
68 dracut can also generate a more generic initramfs image (default mode).
69
70 dracut's initramfs starts only with the device name of the root file system (or
71 its UUID) and must discover everything else at boot time. A complex cascade of
72 tasks must be performed to get the root file system mounted:
73
74 * Any hardware drivers that the boot process depends on must be loaded. All
75 kernel modules for common storage devices are packed onto the initramfs and then
76 udev pulls in modules matching the computer's detected hardware.
77
78 * On systems which display a boot rd.splash screen, the video hardware must be
79 initialized and a user-space helper started to paint animations onto the display
80 in lockstep with the boot process.
81
82 * If the root file system is on NFS, dracut does then:
83 ** Bring up the primary network interface.
84 ** Invoke a DHCP client, with which it can obtain a DHCP lease.
85 ** Extract the name of the NFS share and the address of the NFS server from the
86 lease.
87 ** Mount the NFS share.
88
89 * If the root file system appears to be on a software RAID device, there is no
90 way of knowing which devices the RAID volume spans; the standard MD utilities
91 must be invoked to scan all available block devices with a raid signature and
92 bring the required ones online.
93
94 * If the root file system appears to be on a logical volume, the LVM utilities
95 must be invoked to scan for and activate the volume group containing it.
96
97 * If the root file system is on an encrypted block device:
98 ** Invoke a helper script to prompt the user to type in a passphrase and/or
99 insert a hardware token (such as a smart card or a USB security dongle).
100
101 * Create a decryption target with the device mapper.
102
103 dracut uses udev, an event-driven hotplug agent, which invokes helper programs
104 as hardware devices, disk partitions and storage volumes matching certain rules
105 come online. This allows discovery to run in parallel, and to progressively
106 cascade into arbitrary nestings of LVM, RAID or encryption to get at the root
107 file system.
108
109 When the root file system finally becomes visible:
110
111 * Any maintenance tasks which cannot run on a mounted root file system
112 are done.
113 * The root file system is mounted read-only.
114 * Any processes which must continue running (such as the rd.splash screen helper
115 and its command FIFO) are hoisted into the newly-mounted root file system.
116
117 The final root file system cannot simply be mounted over /, since that would
118 make the scripts and tools on the initial root file system inaccessible for any
119 final cleanup tasks. On an initramfs, the initial root file system cannot be
120 rotated away. Instead, it is simply emptied and the final root file system
121 mounted over the top.
122
123 If the systemd module is used in the initramfs, the ordering of the services
124 started looks like <<dracutbootup7>>.
125
126 == Dracut on shutdown
127
128 On a systemd driven system, the dracut initramfs is also used for the shutdown
129 procedure.
130
131 The following steps are executed during a shutdown:
132
133 * systemd switches to the shutdown.target
134 * systemd starts
135 $prefix/lib/systemd/system/shutdown.target.wants/dracut-shutdown.service
136 * dracut-shutdown.service executes /usr/lib/dracut/dracut-initramfs-restore
137 which unpacks the initramfs to /run/initramfs
138 * systemd finishes shutdown.target
139 * systemd kills all processes
140 * systemd tries to unmount everything and mounts the remaining read-only
141 * systemd checks, if there is a /run/initramfs/shutdown executable
142 * if yes, it does a pivot_root to /run/initramfs and executes ./shutdown.
143 The old root is then mounted on /oldroot.
144 /usr/lib/dracut/modules.d/99shutdown/shutdown.sh is the shutdown executable.
145 * shutdown will try to umount every /oldroot mount and calls the various
146 shutdown hooks from the dracut modules
147
148 This ensures, that all devices are disassembled and unmounted cleanly.
149
150 = User Manual
151
152 :leveloffset: 1
153 include::dracut.8.asc[]
154
155 :leveloffset: 1
156 [[dracutconf5]]
157 include::dracut.conf.5.asc[]
158
159 [[dracutcmdline7]]
160 include::dracut.cmdline.7.asc[]
161
162 [[lsinitrd1]]
163 include::lsinitrd.1.asc[]
164
165 [[mkinitrd8]]
166 include::mkinitrd.8.asc[]
167
168 = Developer Manual
169
170 :leveloffset: 1
171 [[dracutmodules7]]
172 include::dracut.modules.7.asc[]
173
174 [[dracutbootup7]]
175 include::dracut.bootup.7.asc[]
176
177 :leveloffset: 0
178 [appendix]
179 License
180 -------
181 This work is licensed under the Creative Commons Attribution/Share-Alike
182 License. To view a copy of this license, visit
183 http://creativecommons.org/licenses/by-sa/3.0/ or send a letter to Creative
184 Commons, 559 Nathan Abbott Way, Stanford, California 94305, USA.
185