Difference between revisions of "Mkinitcpio"

Revision as of 19:19, 6 October 2013

zh-CN:Mkinitcpio Template:Article summary start Template:Article summary text Template:Article summary heading Template:Article summary text Template:Article summary end mkinitcpio is the next generation of initramfs creation.

Overview

mkinitcpio is a Bash script used to create an initial ramdisk environment. From the mkinitcpio man page:

The initial ramdisk is in essence a very small environment (early userspace) which loads various kernel modules and sets up necessary things before handing over control to init. This makes it possible to have, for example, encrypted root filesystems and root filesystems on a software RAID array. mkinitcpio allows for easy extension with custom hooks, has autodetection at runtime, and many other features.

Traditionally, the kernel was responsible for all hardware detection and initialization tasks early in the boot process before mounting the root filesystem and passing control to init. However, as technology advances, these tasks have become increasingly complex.

Nowadays, the root filesystem may be on a wide range of hardware, from SCSI to SATA to USB drives, controlled by a variety of drive controllers from different manufacturers. Additionally, the root filesystem may be encrypted or compressed; within a software RAID array or a logical volume group. The simple way to handle that complexity is to pass management into userspace: an initial ramdisk.

See also: /dev/brain0 » Blog Archive » Early Userspace in Arch Linux.

mkinitcpio is a modular tool for building an initramfs CPIO image, offering many advantages over alternative methods; these advantages include:

• The use of BusyBox to provide a small and lightweight base for early userspace.
• Support for udev for hardware auto-detection at runtime, thus preventing the loading of unnecessary modules.
• Using an extendable hook-based init script, which supports custom hooks that can easily be included in pacman packages.
• Support for LVM2, dm-crypt for both legacy and LUKS volumes, mdadm, and swsusp and suspend2 for resuming and booting from USB mass storage devices.
• The ability to allow many features to be configured from the kernel command line without needing to rebuild the image.

mkinitcpio has been developed by the Arch Linux developers and from community contributions. See the public Git repository.

Installation

The mkinitcpio package is available in the official repositories and is installed by default as a member of the base group.

Advanced users may wish to install the latest development version of mkinitcpio from Git:

$ git clone git://projects.archlinux.org/mkinitcpio.git

Image creation and activation

By default, the mkinitcpio script generates two images after kernel installation or upgrades: /boot/initramfs-linux.img and /boot/initramfs-linux-fallback.img. The fallback image utilizes the same configuration file as the default image, except the autodetect hook is skipped during creation, thus including a full range of modules. The autodetect hook detects required modules and tailors the image for specific hardware, shrinking the initramfs.

Users may create any number of initramfs images with a variety of different configurations. The desired image must be specified for the bootloader, often in its configuration file. After changes are made to the configuration file, the image must be regenerated. For the stock Arch Linux kernel, linux, this is done by running this command with root privileges:

# mkinitcpio -p linux

The -p switch specifies a preset to utilize; most kernel packages provide a related mkinitcpio preset file, found in /etc/mkinitcpio.d (e.g. /etc/mkinitcpio.d/linux.preset for linux). A preset is a predefined definition of how to create an initramfs image instead of specifying the configuration file and output file every time.

Users can manually create an image using an alternative configuration file. For example, the following will generate an initramfs image according to the directions in /etc/mkinitcpio-custom.conf and save it at /boot/linux-custom.img.

# mkinitcpio -c /etc/mkinitcpio-custom.conf -g /boot/linux-custom.img

If creating an image for a kernel other than the one currently running, add the kernel version to the command line. You can see available kernel versions in /usr/lib/modules.

# mkinitcpio -g /boot/linux.img -k 3.3.0-ARCH

Configuration

The primary configuration file for mkinitcpio is /etc/mkinitcpio.conf. Additionally, preset definitions are provided by kernel packages in the /etc/mkinitcpio.d directory (e.g. /etc/mkinitcpio.d/linux.preset).

Users can modify six variables within the configuration file:

MODULES

Kernel modules to be loaded before any boot hooks are run.

BINARIES

Additional binaries to be included in the initramfs image.

FILES

Additional files to be included in the initramfs image.

HOOKS

Hooks are scripts that execute in the initial ramdisk.

COMPRESSION

Used to compress the initramfs image.

COMPRESSION_OPTIONS

Command line options to pass to the COMPRESSION program.

MODULES

The MODULES array is used to specify modules to load before anything else is done.

Modules suffixed with a ? will not throw errors if they are not found. This might be useful for custom kernels that compile in modules which are listed explicitly in a hook or config file.

BINARIES and FILES

These options allow users to add files to the image. Both BINARIES and FILES are added before hooks are run, and may be used to override files used or provided by a hook. BINARIES are auto-located within a standard PATH and dependency-parsed, meaning any required libraries will also be added. FILES are added as-is. For example:

FILES="/etc/modprobe.d/modprobe.conf"

BINARIES="kexec"

For both, BINARIES and FILES, multiple entries can be added delimited with spaces.

HOOKS

The HOOKS setting is the most important setting in the file. Hooks are small scripts which describe what will be added to the image. For some hooks, they will also contain a runtime component which provides additional behavior, such as starting a daemon, or assembling a stacked block device. Hooks are referred to by their name, and executed in the order they exist in the HOOKS setting in the config file.

The default HOOKS setting should be sufficient for most simple, single disk setups. For root devices which are stacked or multi-block devices such as LVM, mdadm, or LUKS, see the respective wiki pages for further necessary configuration.

Build hooks

Build hooks are found in /usr/lib/initcpio/install. These files are sourced by the bash shell during runtime of mkinitcpio and should contain two functions: build and help. The build function describes the modules, files, and binaries which will be added to the image. An API, documented by mkinitcpio(8), serves to facilitate the addition of these items. The help function outputs a description of what the hook accomplishes.

For a list of all available hooks:

$ mkinitcpio -L

Use mkinitcpio's -H option to output help for a specific hook, for example:

$ mkinitcpio -H udev

Runtime hooks

Runtime hooks are found in /usr/lib/initcpio/hooks. For any runtime hook, there should always be a build hook of the same name, which calls add_runscript to add the runtime hook to the image. These files are sourced by the busybox ash shell during early userspace. With the exception of cleanup hooks, they will always be run in the order listed in the HOOKS setting. Runtime hooks may contain several functions:

run_earlyhook: Functions of this name will be run once the API filesystems have been mounted and the kernel command line has been parsed. This is generally where additional daemons, such as udev, which are needed for the early boot process are started from.

run_hook: Functions of this name are run shortly after the early hooks. This is the most common hook point, and operations such as assembly of stacked block devices should take place here.

run_latehook: Functions of this name are run after the root device has been mounted. This should be used, sparingly, for further setup of the root device, or for mounting other filesystems, such as /usr.

run_cleanuphook: Functions of this name are run as late as possible, and in the reverse order of how they are listed in the HOOKS setting in the config file. These hooks should be used for any last minute cleanup, such as shutting down any daemons started by an early hook.

Common hooks

A table of common hooks and how they affect image creation and runtime follows. Note that this table is not complete, as packages can provide custom hooks.

Deprecated hooks

As of mkinitcpio 0.13.0, the usbinput hook is deprecated in favor of the keyboard hook.

As of mkinitcpio 0.12.0, the following hooks are deprecated. If you use any of these hooks, you need to replace them with a single instance of the block hook.

• fw
• mmc
• pata
• sata
• scsi
• usb
• virtio

For more information, you can review Git commit 97368c0e78 or see the arch-projects mailing list.

COMPRESSION

The kernel supports several formats for compression of the initramfs - gzip, bzip2, lzma, xz (also known as lzma2), and lzo. For most use cases, gzip or lzop provide the best balance of compressed image size and decompression speed.

COMPRESSION="gzip"
COMPRESSION="bzip2"  # since kernel 2.6.30
COMPRESSION="lzma"   # since kernel 2.6.30
COMPRESSION="lzop"   # since kernel 2.6.34
COMPRESSION="xz"     # since kernel 2.6.38
COMPRESSION="lz4c"   # since kernel 3.11

Specifying no COMPRESSION will result in a gzip compressed initramfs file. To create an uncompressed image, specify COMPRESSION=cat in the config or use -z cat on the command line.

Make sure you have the correct file compression utility installed for the method you wish to use.

COMPRESSION_OPTIONS

These are additional flags passed to the program specified by COMPRESSION, such as:

COMPRESSION_OPTIONS='-9'

In general these should never be needed as mkinitcpio will make sure that any supported compression method has the necessary flags to produce a working image.

Runtime customization

Runtime configuration options can be passed to init and certain hooks via the kernel command line. Kernel command line parameters are often supplied by the bootloader. The options discussed below can be appended to the kernel command line to alter default behavior. See Kernel parameters and Arch Boot Process for more information.

init

root

This is the most important parameter specified on the kernel command line, as it determines what device will be mounted as your proper root device. mkinitcpio is flexible enough to allow a wide variety of formats, for example:

root=/dev/sda1                                                # /dev node
root=LABEL=CorsairF80                                         # label
root=UUID=ea1c4959-406c-45d0-a144-912f4e86b207                # UUID
root=PARTUUID=14420948-2cea-4de7-b042-40f67c618660            # GPT partition UUID
root=/dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0-part1    # udev symlink (requires the udev hook)
root=801                                                      # hex-encoded major/minor number

break

If break or break=premount is specified, init pauses the boot process (after loading hooks, but before mounting the root filesystem) and launches an interactive shell which can be used for troubleshooting purposes. This shell can be launched after the root has been mounted by specifying break=postmount. Normal boot continues after exiting from the shell.

disablehooks

Disable hooks at runtime by adding disablehooks=hook1{,hook2,...}. For example:

disablehooks=resume

earlymodules

Alter the order in which modules are loaded by specifying modules to load early via earlymodules=mod1{,mod2,...}. (This may be used, for example, to ensure the correct ordering of multiple network interfaces.)

rootdelay=N

Pause for N seconds before mounting the root file system by appending rootdelay. (This may be used, for example, if booting from a USB hard drive that takes longer to initialize.)

See also: Debugging with GRUB and init

Using RAID

First, add the mdadm hook to the HOOKS array and any required RAID modules (raid456, ext4) to the MODULES array in /etc/mkinitcpio.conf.

Kernel Parameters: Using the mdadm hook, you no longer need to configure your RAID array in the GRUB parameters. The mdadm hook will either use your /etc/mdadm.conf file or automatically detect the array(s) during the init phase of boot.

Assembly via udev is also possible using the mdadm_udev hook. Upstream prefers this method of assembly. /etc/mdadm.conf will still be read for purposes of naming the assembled devices if it exists.

Using net

Required Packages:

net requires the mkinitcpio-nfs-utils package from the official repositories.

Kernel Parameters:

ip=

An interface spec can be either short form, which is just the name of an interface (eth0 or whatever), or long form. ( Kernel Documentation )

The long form consists of up to seven elements, separated by colons:

 ip=<client-ip>:<server-ip>:<gw-ip>:<netmask>:<hostname>:<device>:<autoconf>
 nfsaddrs= is an alias to ip= and can be used too.

Parameter explanation:

 <client-ip>   IP address of the client. If empty, the address will
               either be determined by RARP/BOOTP/DHCP. What protocol
               is used depends on the <autoconf> parameter. If this
               parameter is not empty, autoconf will be used.
 
 <server-ip>   IP address of the NFS server. If RARP is used to
               determine the client address and this parameter is NOT
               empty only replies from the specified server are
               accepted. To use different RARP and NFS server,
               specify your RARP server here (or leave it blank), and
               specify your NFS server in the `nfsroot' parameter
               (see above). If this entry is blank the address of the
               server is used which answered the RARP/BOOTP/DHCP
               request.
 
 <gw-ip>       IP address of a gateway if the server is on a different
               subnet. If this entry is empty no gateway is used and the
               server is assumed to be on the local network, unless a
               value has been received by BOOTP/DHCP.
 
 <netmask>     Netmask for local network interface. If this is empty,
               the netmask is derived from the client IP address assuming
               classful addressing, unless overridden in BOOTP/DHCP reply.
 
 <hostname>    Name of the client. If empty, the client IP address is
               used in ASCII notation, or the value received by
               BOOTP/DHCP.
 
 <device>      Name of network device to use. If this is empty, all
               devices are used for RARP/BOOTP/DHCP requests, and the
               first one we receive a reply on is configured. If you
               have only one device, you can safely leave this blank.
 
 <autoconf>	Method to use for autoconfiguration. If this is either
               'rarp', 'bootp', or 'dhcp' the specified protocol is
               used.  If the value is 'both', 'all' or empty, all
               protocols are used.  'off', 'static' or 'none' means
               no autoconfiguration.

Examples:

 ip=127.0.0.1:::::lo:none  --> Enable the loopback interface.
 ip=192.168.1.1:::::eth2:none --> Enable static eth2 interface.
 ip=:::::eth0:dhcp --> Enable dhcp protocol for eth0 configuration.

BOOTIF=

If you have multiple network cards, this parameter can include the MAC address of the interface you are booting from. This is often useful as interface numbering may change, or in conjunction with pxelinux IPAPPEND 2 or IPAPPEND 3 option. If not given, eth0 will be used.

Example:

 BOOTIF=01-A1-B2-C3-D4-E5-F6  # Note the prepended "01-" and capital letters.

nfsroot=

If the nfsroot parameter is NOT given on the command line, the default /tftpboot/%s will be used.

 nfsroot=[<server-ip>:]<root-dir>[,<nfs-options>]

Parameter explanation:

 <server-ip>   Specifies the IP address of the NFS server. If this field
               is not given, the default address as determined by the
               `ip' variable (see below) is used. One use of this
               parameter is for example to allow using different servers
               for RARP and NFS. Usually you can leave this blank.
 
 <root-dir>    Name of the directory on the server to mount as root. If
               there is a "%s" token in the string, the token will be
               replaced by the ASCII-representation of the client's IP
               address.
 
 <nfs-options> Standard NFS options. All options are separated by commas.
               If the options field is not given, the following defaults
               will be used:
                       port            = as given by server portmap daemon
                       rsize           = 1024
                       wsize           = 1024
                       timeo           = 7
                       retrans         = 3
                       acregmin        = 3
                       acregmax        = 60
                       acdirmin        = 30
                       acdirmax        = 60
                       flags           = hard, nointr, noposix, cto, ac

root=/dev/nfs

If you do not use the nfsroot parameter, you need to set root=/dev/nfs to boot from an NFS root via automatic configuration.

Using LVM

If your root device is on LVM, you must add the lvm2 hook. You have to pass your root device on the kernel command line in the following format:

root=/dev/mapper/<volume group name>-<logical volume name>

for example:

root=/dev/mapper/myvg-root

In addition, if your root device might initialize slowly (e.g. a USB device) and/or you receive a "volume group not found" error during boot, you might need to add the following to the kernel command line:

lvmwait=/dev/mapper/<volume group name><logical volume name>

for example:

lvmwait=/dev/mapper/myvg-root

This lets the boot process wait until LVM manages to make the device available.

Using encrypted root

If your root volume is encrypted, you need to add the encrypt hook.

For an encrypted root, use something similar to:

root=/dev/mapper/root cryptdevice=/dev/sda5:root

In this case, /dev/sda5 is the encrypted device, and we give it an arbitrary name of root, which means our root device, once unlocked, is mounted as /dev/mapper/root. On bootup, you will be prompted for the passphrase to unlock it. See LUKS#Configuration_of_initcpio for more details about using encrypted root.

/usr as a separate partition

If you keep /usr as a separate partition, you must adhere to the following requirements:

• Add the shutdown hook. The shutdown process will pivot to a saved copy of the initramfs and allow for /usr (and root) to be properly unmounted from the VFS.
• Add the fsck hook, mark /usr with a passno of 0 in /etc/fstab. While recommended for everyone, it is mandatory if you want your /usr partition to be fsck'ed at boot-up. Without this hook, /usr will never be fsck'd.
• Add the usr hook. This will mount the /usr partition after root is mounted. Prior to 0.9.0, mounting of /usr would be automatic if it was found in the real root's /etc/fstab.

Troubleshooting

Extracting the image

If you are curious about what is inside the initrd image, you can extract it and poke at the files inside of it.

The initrd image is an SVR4 CPIO archive, generated via the find and bsdcpio commands, optionally compressed with a compression scheme understood by the kernel. For more information on the compression schemes, see #COMPRESSION.

mkinitcpio includes a utility called lsinitcpio which will list and extract the contents of initramfs images.

You can list the files in the image with:

$ lsinitcpio /boot/initramfs-linux.img

And to extract them all in the current directory:

$ lsinitcpio -x /boot/initramfs-linux.img

You can also get a more human-friendly listing of the important parts in the image:

$ lsinitcpio -a /boot/initramfs-linux.img

See also

• Boot Debugging - Debugging with GRUB

References

1. Linux Kernel documentation on initramfs
2. Linux Kernel documentation on initrd
3. Wikipedia article on initrd