xref: /netbsd-src/sbin/init/init.8 (revision de1dfb1250df962f1ff3a011772cf58e605aed11)
1.\"	$NetBSD: init.8,v 1.31 2004/02/19 13:24:31 lukem Exp $
2.\"
3.\" Copyright (c) 1980, 1991, 1993
4.\"	The Regents of the University of California.  All rights reserved.
5.\"
6.\" This code is derived from software contributed to Berkeley by
7.\" Donn Seeley at Berkeley Software Design, Inc.
8.\"
9.\" Redistribution and use in source and binary forms, with or without
10.\" modification, are permitted provided that the following conditions
11.\" are met:
12.\" 1. Redistributions of source code must retain the above copyright
13.\"    notice, this list of conditions and the following disclaimer.
14.\" 2. Redistributions in binary form must reproduce the above copyright
15.\"    notice, this list of conditions and the following disclaimer in the
16.\"    documentation and/or other materials provided with the distribution.
17.\" 3. Neither the name of the University nor the names of its contributors
18.\"    may be used to endorse or promote products derived from this software
19.\"    without specific prior written permission.
20.\"
21.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31.\" SUCH DAMAGE.
32.\"
33.\"     @(#)init.8	8.6 (Berkeley) 5/26/95
34.\"
35.Dd February 19, 2004
36.Dt INIT 8
37.Os
38.Sh NAME
39.Nm init
40.Nd process control initialization
41.Sh SYNOPSIS
42.Nm
43.Sh DESCRIPTION
44The
45.Nm
46program is the last stage of the boot process (after the kernel loads
47and initializes all the devices).
48It normally begins multi-user operation.
49.Pp
50The following table describes the state machine used by
51.Nm :
52.Bl -enum
53.It
54Single user shell.
55.Nm
56may be passed
57.Fl s
58from the boot program to prevent the system from going multi-user and
59to instead execute a single user shell without starting the normal
60daemons.
61The system is then quiescent for maintenance work and may
62later be made to go to state 2 (multi-user) by exiting the single-user
63shell (with ^D).
64.It
65Multi-user boot (default operation).
66Executes
67.Pa /etc/rc
68(see
69.Xr rc 8 ) .
70If this was the first state entered (as opposed to entering here after
71state 1), then
72.Pa /etc/rc
73will be invoked with its first argument being
74.Sq autoboot .
75If
76.Pa /etc/rc
77exits with a non-zero (error) exit code, commence single user
78operation by giving the super-user a shell on the console by going
79to state 1 (single user).
80Otherwise, proceed to state 3.
81.It
82Set up ttys as specified in
83.Xr ttys 5 .
84See below for more information.
85On completion, continue to state 4.
86.It
87Multi-user operation.
88Depending upon the signal received, change state appropriately;
89on
90.Dv SIGTERM ,
91go to state 7;
92on
93.Dv SIGHUP ,
94go to state 5;
95on
96.Dv SIGTSTP ,
97go to state 6.
98.It
99Clean-up mode; re-read
100.Xr ttys 5 ,
101killing off the controlling processes on lines that are now
102.Sq off ,
103and starting processes that are newly
104.Sq on .
105On completion, go to state 4.
106.It
107.Sq Boring
108mode; no new sessions.
109Signals as per state 4.
110.It
111Shutdown mode.
112Send
113.Dv SIGHUP
114to all controlling processes, reap the processes for 30 seconds,
115and the go to state 1 (single user); warning if not all the processes died.
116.El
117.Pp
118If the
119.Sq console
120entry in the
121.Xr ttys 5
122file is marked
123.Dq insecure ,
124then
125.Nm
126will require that the superuser password be
127entered before the system will start a single-user shell.
128The password check is skipped if the
129.Sq console
130is marked as
131.Dq secure .
132.Pp
133The kernel runs with four different levels of security.
134Any superuser process can raise the security level, but only
135.Nm
136can lower it.
137.Pp
138The security level mechanism is intended to allow the administrator
139to protect the persistent code and data on the system, or a subset
140thereof, from modification, even by the superuser.
141In order for this protection to be effective, the administrator
142must ensure that no program that is run while the security level
143is 0 or lower, nor any data or configuration file used by any such
144program, can be modified while the security level is greater than
1450.
146This may be achieved through the careful use of the
147.Dq immutable
148file flag to define and protect a Trusted Computing Base (TCB)
149consisting of all such programs and data, or by ensuring that all
150such programs and data are on filesystems that are mounted read-only
151and running at security level 2 or higher.
152.Em Particular care must be taken to ensure, if relying upon
153.Em security level 1 and the use of file flags, that the integrity of the
154.Em TCB cannot be compromised through the use of modifications to the
155.Em disklabel or access to overlapping disk partitions, including the
156.Em raw partition .
157.Pp
158Do not overlook the fact that shell scripts (or anything else fed to an
159interpreter, through any mechanism) and the kernel itself are "programs
160that run while the security level is 0" and must be considered part of
161the TCB.
162.Pp
163Security levels are defined as follows:
164.Bl -tag -width flag
165.It Ic -1
166Permanently insecure mode \- always run system in level 0 mode.
167.It Ic 0
168Insecure mode \- immutable and append-only flags may be changed.
169All devices may be read or written subject to their permissions.
170.It Ic 1
171Secure mode \- system immutable and system append-only flags may not
172be turned off; disks for mounted filesystems,
173.Pa /dev/mem ,
174and
175.Pa /dev/kmem
176are read-only.
177.Pp
178The verified exec in-kernel fingerprint table may not be changed
179(see
180.Xr veriexecctl 8 ) .
181.It Ic 2
182Highly secure mode \- same as secure mode, plus disks are always
183read-only whether mounted or not, new disks may not be mounted,
184and existing mounts may only be downgraded from read-write to read-only.
185This level precludes tampering with filesystems by unmounting them,
186but also inhibits running
187.Xr newfs 8
188while the system is multi-user.
189.Pp
190The
191.Xr settimeofday 2
192system call can only advance the time.
193.Pp
194The state of
195.Xr ipf 8
196(the in-kernel IP filtering facility) may not be changed.
197.Pp
198Users may not change the per-process core name template format, only the
199default can be changed.
200.Pp
201Downgrading from highly secure mode to insecure mode (that is, to single-user
202mode) always requires the root password to be entered on the console, whether
203the console is marked as
204.Dq secure
205in
206.Pa /etc/ttys
207or not.
208.El
209.Pp
210Normally, the system runs in level 0 mode while single user
211and in level 1 mode while multi-user.
212If the level 2 mode is desired while running multi-user,
213it can be set in the startup script
214.Pa /etc/rc
215using
216.Xr sysctl 8 .
217If it is desired to run the system in level 0 mode while multi-user,
218the administrator must build a kernel with
219.Sy options INSECURE
220in the kernel configuration file, which initializes the kernel's
221.Va securelevel
222variable to -1.
223See
224.Xr options 4
225and
226.Xr config 8
227for details.
228.Pp
229In multi-user operation,
230.Nm
231maintains
232processes for the terminal ports found in the file
233.Xr ttys 5 .
234.Nm
235reads this file, and executes the command found in the second field.
236This command is usually
237.Xr getty 8 ;
238it opens and initializes the tty line and executes the
239.Xr login 1
240program.
241The
242.Xr login 1
243program, when a valid user logs in, executes a shell for that user.
244When this shell dies, either because the user logged out or an
245abnormal termination occurred (a signal), the
246.Nm
247program wakes up, deletes the user from the
248.Xr utmp 5
249file of current users and records the logout in the
250.Xr wtmp 5
251file.
252The cycle is
253then restarted by
254.Nm
255executing a new
256.Xr getty 8
257for the line.
258.pl +1
259.Pp
260Line status (on, off, secure, getty, or window information)
261may be changed in the
262.Xr ttys 5
263file without a reboot by sending the signal
264.Dv SIGHUP
265to
266.Nm
267with the command
268.Dq Li "kill \-s HUP 1" .
269This is referenced in the table above as state 5.
270On receipt of this signal,
271.Nm
272re-reads the
273.Xr ttys 5
274file.
275When a line is turned off in
276.Xr ttys 5 ,
277.Nm
278will send a
279.Dv SIGHUP
280signal to the controlling process
281for the session associated with the line.
282For any lines that were previously turned off in the
283.Xr ttys 5
284file and are now on,
285.Nm
286executes a new
287.Xr getty 8
288to enable a new login.
289If the getty or window field for a line is changed,
290the change takes effect at the end of the current
291login session (e.g., the next time
292.Nm
293starts a process on the line).
294If a line is commented out or deleted from
295.Xr ttys 5 ,
296.Nm
297will not do anything at all to that line.
298However, it will complain that the relationship between lines
299in the
300.Xr ttys 5
301file and records in the
302.Xr utmp 5
303file is out of sync,
304so this practice is not recommended.
305.Pp
306.Nm
307will terminate multi-user operations and resume single-user mode
308if sent a terminate
309.Pq Dv TERM
310signal, for example,
311.Dq Li "kill \-s TERM 1" .
312If there are processes outstanding that are deadlocked (because of
313hardware or software failure),
314.Nm
315will not wait for them all to die (which might take forever), but
316will time out after 30 seconds and print a warning message.
317.Pp
318.Nm
319will cease creating new
320.Xr getty 8 Ns 's
321and allow the system to slowly die away, if it is sent a terminal stop
322.Pq Dv TSTP
323signal, i.e.
324.Dq Li "kill \-s TSTP 1" .
325A later hangup will resume full
326multi-user operations, or a terminate will start a single user shell.
327This hook is used by
328.Xr reboot 8
329and
330.Xr halt 8 .
331.Pp
332The role of
333.Nm
334is so critical that if it dies, the system will reboot itself
335automatically.
336If, at bootstrap time, the
337.Nm
338process cannot be located, the system will panic with the message
339.Dq panic: init died (signal %d, exit %d) .
340.Sh FILES
341.Bl -tag -width /var/log/wtmp -compact
342.It Pa /dev/console
343System console device.
344.It Pa /dev/tty*
345Terminal ports found in
346.Xr ttys 5 .
347.It Pa /var/run/utmp
348Record of Current users on the system.
349.It Pa /var/log/wtmp
350Record of all logins and logouts.
351.It Pa /etc/ttys
352The terminal initialization information file.
353.It Pa /etc/rc
354System startup commands.
355.El
356.Sh DIAGNOSTICS
357.Bl -diag
358.It "getty repeating too quickly on port %s, sleeping"
359A process being started to service a line is exiting quickly
360each time it is started.
361This is often caused by a ringing or noisy terminal line.
362.Em "Init will sleep for 10 seconds" ,
363.Em "then continue trying to start the process" .
364.Pp
365.It "some processes would not die; ps axl advised."
366A process is hung and could not be killed when the system was
367shutting down.
368This condition is usually caused by a process that is stuck in a
369device driver because of a persistent device error condition.
370.El
371.Sh SEE ALSO
372.Xr kill 1 ,
373.Xr login 1 ,
374.Xr sh 1 ,
375.Xr options 4 ,
376.Xr ttys 5 ,
377.Xr config 8 ,
378.Xr getty 8 ,
379.Xr halt 8 ,
380.Xr rc 8 ,
381.Xr reboot 8 ,
382.Xr shutdown 8
383.Sh HISTORY
384A
385.Nm
386command appeared in
387.At v6 .
388.Sh BUGS
389Systems without
390.Xr sysctl 8
391behave as though they have security level \-1.
392.Pp
393The security level 2 restrictions relating to TCB integrity protection
394should be enforced at security level 1.
395Restrictions dependent upon security level but not relating to TCB
396integrity protection should be selected by
397.Xr sysctl 8
398settings available only at security level 0 or lower.
399