2016-07-22 23:31:57 +02:00
2016-07-22 20:00:13 +02:00
2016-07-21 14:04:37 +02:00
2016-07-21 00:29:49 +02:00
…
2016-03-21 15:41:51 +01:00
2016-03-21 15:41:51 +01:00
2016-07-21 20:43:23 +02:00
2016-07-22 22:44:39 +02:00
2016-07-18 17:30:14 +02:00
…
2016-03-21 15:41:51 +01:00
2016-07-21 11:43:09 +02:00
…
2016-07-20 17:30:29 +02:00

                      _______  __    _____  ___    __  ___________  _______
                     /"     "||" \  (\"   \|"  \  |" \("     _   ")/" __   )
                    (: ______)||  | |.\\   \    | ||  |)__/  \\__/(__/ _) ./
                     \/    |  |:  | |: \.   \\  | |:  |   \\_ /       /  //
                     // ___)  |.  | |.  \    \. | |.  |   |.  |    __ \_ \\
                    (:  (     /\  |\|    \    \ | /\  |\  \:  |   (: \__) :\
                     \__/    (__\_|_)\___|\____\)(__\_|_)  \__|    \_______)

Fast init for Linux

Travis Status Coverity Status

Table of Contents

Introduction

Finit is a plugin-based init with process supervision similar to that of D.J. Bernstein's daemontools and Gerrit Pape's runit. The main focus of Finit is on small and embedded GNU/Linux systems, yet fully functional on standard server and desktop installations as well.

Traditional SysV init style systems are scripted. For low-resource embedded systems this is quite resource intensive and often leads to long boot times. Finit reduces context switches and forking of shell scripts to provide a system bootstrap written entirely in C.

There is no /etc/init.d/rcS script, or similar. Instead configuration is read from the main /etc/finit.conf, which details kernel modules to load and bootstrap services to start. After initial bootstrap, including setting up networking, /etc/finit.d/ and the familiar /etc/rc.local are run.

Example /etc/finit.conf:

    # Fallback if /etc/hostname is missing
    host myhostname
    
    # Devices to fsck at boot
    check /dev/vda1
    
    # Runlevel to start after bootstrap, runlevel 'S'
    runlevel 2
    
    # Network bringup, not needed if /etc/network/interfaces exist
    #network service networking start
    #network /sbin/ifup -a

    # Services to be monitored and respawned as needed
    service [S12345] /sbin/watchdogd -L -f                       -- System watchdog daemon
    service [S12345] /sbin/syslogd -n -b 3 -D                    -- System log daemon
    service [S12345] /sbin/klogd -n                              -- Kernel log daemon
    service    2345] /sbin/lldpd -d -c -M1 -H0 -i                -- LLDP daemon (IEEE 802.1ab)
    
    # For multiple instances of the same service, add :ID somewhere between
    # the service/run/task keyword and the command.
    service :1 [2345] /sbin/httpd -f -h /http -p 80   -- Web server
    service :2 [2345] /sbin/httpd -f -h /http -p 8080 -- Old web server

    # Alternative method instead of below runparts, can also use /etc/rc.local
    #task [S] /etc/init.d/keyboard-setup start -- Setting up preliminary keymap
    #task [S] /etc/init.d/acpid start          -- Starting ACPI Daemon
    #task [S] /etc/init.d/kbd start            -- Preparing console
    #run [2] /etc/init.d/networking start      -- Start networking

    # Inetd services to start on demand, with alternate ports and filtering
    inetd ftp/tcp          nowait [2345] /sbin/uftpd -i -f       -- FTP daemon
    inetd tftp/udp           wait [2345] /sbin/uftpd -i -y       -- TFTP daemon
    inetd time/udp           wait [2345] internal                -- UNIX rdate service
    inetd time/tcp         nowait [2345] internal                -- UNIX rdate service
    inetd 3737/tcp         nowait [2345] internal.time           -- UNIX rdate service
    inetd telnet/tcp       nowait [2345] /sbin/telnetd -i -F     -- Telnet daemon
    inetd 2323/tcp         nowait [2345] /sbin/telnetd -i -F     -- Telnet daemon
    inetd 222/tcp@eth0     nowait [2345] /sbin/dropbear -i -R -F -- SSH service
    inetd ssh/tcp@*,!eth0  nowait [2345] /sbin/dropbear -i -R -F -- SSH service
    
    # Run start scripts from this directory
    # runparts /etc/start.d
    
    # Virtual consoles to start getty on
    tty [12345] /dev/tty1    115200 linux
    tty [12345] /dev/ttyAMA0 115200 vt100

For an example of a full blown embedded Linux, see TroglOS or the contrib/ section with Alpine Linux support

Features

Process Supervision

Start, monitor and restart services should they fail.

Inetd

Finit comes with a built-in inetd server. No need to maintain a separate config file for services that you want to start on demand.

All inetd services started can be filtered per port and inbound interface, reducing the need for a full blown firewall.

Built-in optional inetd services:

  • echo RFC862
  • chargen RFC864
  • daytime RFC867
  • discard RFC863
  • time (rdate) RFC868

For more information, see doc/inetd.md.

Runlevels

Runlevels are optional in Finit, but support for SysV runlevels is available if needed. All services in runlevel S(1) are started first, followed by the desired run-time runlevel. Runlevel S can be started in sequence by using run [S] cmd. Changing runlevels at runtime is done like any other init, e.g. init 4

Plugins

Finit plugins can be callbacks, boot hooks into different stages of the boot process, or pure extensions. Plugins are written in C and compiled into a dynamic library that is loaded automatically by finit at boot. A basic set of plugins that extend and modify the basic behavior are bundled in the plugins/ directory.

Capabilities:

  • Task/Run/Service callbacks
    Modify service arguments, override service monitor's decisions to start/stop/reload a service.
  • Hooks
    Hook into the boot at predefined points to extend Finit
  • I/O
    Listen to external events and control Finit behavior/services
  • Inetd
    Extend Finit with internal inetd services, for an example, see plugins/time.c

Extensions and functionality not purely related to what an /sbin/init needs to start a system are available as a set of plugins that either hook into the boot process or respond to various I/O.

For more information, see doc/plugins.md.

Runparts & /etc/rc.local

At the end of the boot, when networking and all services are up, finit calls its built-in run-parts(8) on the runparts <DIR> directory, and /etc/rc.local, in that order if they exist.

    runparts /etc/rc.d/

No configuration stanza in /etc/finit.conf is required for rc.local. If it exists and is an executable shell script, finit calls it at the very end of the boot, before calling the HOOK_SYSTEM_UP. See more on hooks in doc/plugins.md, and the system bootstrap below.

Bootstrap

  1. Populate /dev
  2. Parse /etc/finit.conf
  3. Load all .so plugins
  4. Remount/Pivot / to get R+W
  5. Call 1st level hooks, HOOK_ROOTFS_UP
  6. Mount /etc/fstab and swap, if available
  7. Cleanup stale files from /tmp/* et al
  8. Enable SysV init signals
  9. Call 2nd level hooks, HOOK_BASEFS_UP
  10. Start all 'S' runlevel tasks and services
  11. Load kernel parameters from /etc/sysctl.conf
  12. Set hostname and bring up loopback interface
  13. Call network script, if set in /etc/finit.conf
  14. Call 3rd level hooks, HOOK_NETWORK_UP
  15. Load all *.conf files in /etc/finit.d/ and switch to the active active runlevel, as set in /etc/finit.conf, default is 2. Here is where the rest of all tasks and inetd services are started.
  16. Call 4th level hooks, HOOK_SVC_UP
  17. If runparts <DIR> is set, run-parts(8) is called on <DIR>
  18. Call /etc/rc.local, if it exists and is an executable shell script
  19. Call 5th level (last) hooks, HOOK_SYSTEM_UP
  20. Start TTYs defined in /etc/finit.conf, or rescue on /dev/console

In (10) and (15) tasks and services defined in /etc/finit.conf are started. Remember, all service and task stanzas are started in parallel and run in sequence, and in the order listed. Hence, to emulate a SysV /etc/init.d/rcS one could write a long file with only run statements.

Notice the five hook points that are called at various point in the bootstrap process. This is where plugins can extend the boot in any way they please.

For instance, at HOOK_BASEFS_UP a plugin could read an XML file from a USB stick, convert/copy its contents to the system's /etc/ directory, well before all 'S' runlevel tasks are started. This could be used with system images that are created read-only and all configuration is stored on external media.

Runlevels

Basic support for runlevels is included in Finit from v1.8. By default all services, tasks, run commands and TTYs listed without a set of runlevels get a default set [234] assigned. The default runlevel after boot is 2.

Finit supports runlevels 0-9, and S, with 0 reserved for halt, 6 reboot and S for services to only run at bootstrap. Runlevel 1 is the single user level, where usually no networking is enabled. In Finit this is more of a policy for the user to define. Normally only runlevels 1-6 are used, and even more commonly, only the default runlevel is used.

To specify an allowed set of runlevels for a service, run command, task, or tty, add [NNN] to your /etc/finit.conf, like this:

    service [S12345] /sbin/syslogd -n -x     -- System log daemon
    run     [S]      /etc/init.d/acpid start -- Starting ACPI Daemon
    task    [S]      /etc/init.d/kbd start   -- Preparing console
    service [S12345] /sbin/klogd -n -x       -- Kernel log daemon
    tty     [12345]  /dev/tty1
    tty     [2]      /dev/tty2
    tty     [2]      /dev/tty3
    tty     [2]      /dev/tty4
    tty     [2]      /dev/tty5
    tty     [2]      /dev/tty6

In this example syslogd is first started, in parallel, and then acpid is called using a conventional SysV init script. It is called with the run command, meaning the following task command to start the kbd script is not called until the acpid init script has fully completed. Then the keyboard setup script is called in parallel with klogd as a monitored service.

Again, tasks and services are started in parallel, while run commands are called in the order listed and subsequent commands are not started until a run command has completed.

Switching between runlevels can be done by calling init with a single argument, e.g. init 5 switches to runlevel 5. When changing runlevels Finit also automatically reloads all .conf files in the /etc/finit.d/ directory. So if you want to set a new system config, switch to runlevel 1, change all config files in the system, and touch all .conf files in /etc/finit.d before switching back to the previous runlevel again — that way Finit can both stop old services and start any new ones for you, without rebooting the system.

Rebooting & Halting

Finit handles SIGUSR1 and SIGUSR2 for reboot and halt, and listens to /dev/initctl so system reboot and halt commands also work. The latter is implemented in the optional initctl.so plugin and can be accessed with the traditional telinit command line tool, symlinked to finit. Hence, if finit is your system init, then init q will work as the UNIX beards intended.

    ~ $ telinit
    Usage: telinit [OPTIONS] [q | Q | 0-9]
    
    Options:
      -h, --help            This help text
      -v, --version         Show Finit version
    
    Commands:
      q | Q           Reload *.conf in /etc/finit.d/, like SIGHUP
      0 - 9           Change runlevel: 0 halt, 6 reboot

Commands & Status

Finit also implements a more modern API to query status, and start/stop services, called initctl. Unlike telinit the initctl tool does not return until the given command has fully completed.

    ~ $ initctl -h
    Usage: initctl [OPTIONS] <COMMAND>
    
    Options:
      -d, --debug               Debug initctl (client)
      -v, --verbose             Verbose output
      -h, --help                This help text
    
    Commands:
      debug                     Toggle Finit (daemon) debug
      help                      This help text
      reload                    Reload *.conf in /etc/finit.d/ and activate changes
      runlevel [0-9]            Show or set runlevel: 0 halt, 6 reboot
      status | show             Show status of services
      cond     set   <COND>     Set (assert) condition     => +COND
      cond     clear <COND>     Clear (deassert) condition => -COND
      cond     flux  <COND>     Emulate flux condition     => ~COND
      cond     show             Show condition status
      start    <JOB|NAME>[:ID]  Start service by job# or name, with optional ID
      stop     <JOB|NAME>[:ID]  Stop/Pause a running service by job# or name
      restart  <JOB|NAME>[:ID]  Restart (stop/start) service by job# or name
      reload   <JOB|NAME>[:ID]  Reload (SIGHUP) service by job# or name
      version                   Show Finit version

The <!> notation to a service stanza can be used empty, then it will apply to reload and runlevel commands. I.e., when a service's .conf file has been changed Finit will stop and start it instead. If a service does not have <!> declared, the service will be reloaded (SIGHUP:ed) in the START phase instead. The latter is the preferred behaviour, but not all daemons support this, unfortunately.

Note: even though it is possible to start services not belonging to the current runlevel these services will not be respawned automatically by Finit if they exit (crash). Hence, if the runlevel is 2, the below Dropbear SSH service will not be restarted if it is killed or exits.

    ~ $ initctl status -v
    1       running  476     [S12345]   /sbin/watchdog -T 16 -t 2 -F /dev/watchdog
    2       running  477     [S12345]   /sbin/syslogd -n -b 3 -D
    3       running  478     [S12345]   /sbin/klogd -n
    4:1       inetd  0       [2345]     internal time allow *:37
    4:2       inetd  0       [2345]     internal time allow *:37
    4:3       inetd  0       [2345]     internal 3737 allow *:3737
    5:1       inetd  0       [2345]     /sbin/telnetd allow *:23 deny eth0,eth1
    5:2       inetd  0       [2345]     /sbin/telnetd allow eth0:2323,eth2:2323,eth1:2323
    6:1       inetd  0       [345]      /sbin/dropbear allow eth0:222
    6:2       inetd  0       [345]      /sbin/dropbear allow *:22 deny eth0

Building

Finit comes with a traditional configure script to control features and optional plugins to enable. It does however depend on two external libraries that provide some frog DNA needed:

Like most free/open source software that uses configure they deafult to install to /usr/local. However, some Linux distributions do no longer search that path for installed software, e.g. Fedora and Alpine Linux. To get finit's configure script to find its dependencies you have to help the pkg-config tool a bit if you do not change the default prefix path:

PKG_CONFIG_LIBDIR=/usr/local/lib/pkgconfig ./configure

Below are a few of the main switches to configure:

  • --disable-inetd: Disable the built-in inetd server.

  • --enable-embedded: Target finit for BusyBox getty and mdev instead of a standard Linux distribution with GNU tools and udev.

  • --enable-rw-rootfs: Most desktop and server systems boot with the root file stystem read-only. With this setting Finit will remount it as read-write early at boot so the bootmisc.so plugin can run. Usually not needed on embedded systems.

  • --enable-static: Build Finit statically. The plugins will be built-ins (.o files) and all external libraries, except the C library will be linked statically.

  • --enable-alsa-utils: Enable the optional alsa-utils.so sound plugin.

  • --enable-dbus: Enable the optional D-Bus dbus.so plugin.

  • --enable-lost: Enable noisy example plugin for HOOK_SVC_LOST.

  • --enable-resolvconf: Enable the resolvconf.so optional plugin.

  • --enable-x11-common: Enable the optional X Window x11-common.so plugin.

For more configure flags, see ./configure --help

Example

First, unpack the archive:

    $ tar xf finit-3.0.tar.xz
    $ cd finit-3.0/

Then configure, build and install:

    $ ./configure --enable-embedded --enable-rw-rootfs --enable-inetd-echo-plugin \
                  --enable-inetd-chargen-plugin --enable-inetd-daytime-plugin     \
                  --enable-inetd-discard-plugin --enable-inetd-time-plugin        \
                  --with-heading="Alpine Linux 3.4" --with-hostname=alpine
    $ make
    .
    .
    .
    $ DESTDIR=/tmp/finit make install

In this example the finit-3.0.tar.xz archive is unpacked to the user's home directory, configured, built and installed to a temporary staging directory. The environment variable DESTDIR controls the destination directory when installing, very useful for building binary standalone packages.

To target an embedded Linux system, usally a system that use BusyBox tools instead of udev & C:o, add --enable-embedded to the configure command above. This enables mdev instead of udev and the BusyBox getty syntax. Remember to also change the Linux config to:

CONFIG_UEVENT_HELPER_PATH="/sbin/mdev"

Note: If you run into problems starting Finit, take a look at finit.c. One of the most common problems is a custom Linux kernel build that lack CONFIG_DEVTMPFS. Another is too much cruft in the system /etc/fstab.

Running

The default install does not setup finit as the system default /sbin/init, neither does it setup an initial /etc/finit.conf.

It is assumed that users of finit are competent enough to either setup finit as their default /sbin/init or alter their respective GRUB, LOADLIN, LILO, U-Boot/Barebox or RedBoot boot loader configuration to give the kernel the following extra command line:

    init=/sbin/finit

Finit starting Debian 6.0

Debugging

Add finit_debug, or --debug, to the kernel command line to enable debug messages. See the output from configure --help for more on how to assist debugging.

    init=/sbin/finit --debug

Origin & References

This project is based on the original finit by Claudio Matsuoka which was reverse engineered from syscalls of the EeePC fastinit — "gaps filled with frog DNA …"

Finit is developed and maintained by Joachim Nilsson at GitHub. Please file bug reports, clone it, or send pull requests for bug fixes and proposed extensions.

Languages
C 84.3%
Shell 11.9%
Makefile 2.1%
M4 1.7%