xref: /freebsd/sys/kern/init_main.c (revision aed557087269cd052aa76cc15af4a1fd70cbbf24)
19454b2d8SWarner Losh /*-
22b14f991SJulian Elischer  * Copyright (c) 1995 Terrence R. Lambert
32b14f991SJulian Elischer  * All rights reserved.
42b14f991SJulian Elischer  *
5df8bae1dSRodney W. Grimes  * Copyright (c) 1982, 1986, 1989, 1991, 1992, 1993
6df8bae1dSRodney W. Grimes  *	The Regents of the University of California.  All rights reserved.
7df8bae1dSRodney W. Grimes  * (c) UNIX System Laboratories, Inc.
8df8bae1dSRodney W. Grimes  * All or some portions of this file are derived from material licensed
9df8bae1dSRodney W. Grimes  * to the University of California by American Telephone and Telegraph
10df8bae1dSRodney W. Grimes  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
11df8bae1dSRodney W. Grimes  * the permission of UNIX System Laboratories, Inc.
12df8bae1dSRodney W. Grimes  *
13df8bae1dSRodney W. Grimes  * Redistribution and use in source and binary forms, with or without
14df8bae1dSRodney W. Grimes  * modification, are permitted provided that the following conditions
15df8bae1dSRodney W. Grimes  * are met:
16df8bae1dSRodney W. Grimes  * 1. Redistributions of source code must retain the above copyright
17df8bae1dSRodney W. Grimes  *    notice, this list of conditions and the following disclaimer.
18df8bae1dSRodney W. Grimes  * 2. Redistributions in binary form must reproduce the above copyright
19df8bae1dSRodney W. Grimes  *    notice, this list of conditions and the following disclaimer in the
20df8bae1dSRodney W. Grimes  *    documentation and/or other materials provided with the distribution.
21df8bae1dSRodney W. Grimes  * 3. All advertising materials mentioning features or use of this software
22df8bae1dSRodney W. Grimes  *    must display the following acknowledgement:
23df8bae1dSRodney W. Grimes  *	This product includes software developed by the University of
24df8bae1dSRodney W. Grimes  *	California, Berkeley and its contributors.
25df8bae1dSRodney W. Grimes  * 4. Neither the name of the University nor the names of its contributors
26df8bae1dSRodney W. Grimes  *    may be used to endorse or promote products derived from this software
27df8bae1dSRodney W. Grimes  *    without specific prior written permission.
28df8bae1dSRodney W. Grimes  *
29df8bae1dSRodney W. Grimes  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
30df8bae1dSRodney W. Grimes  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
31df8bae1dSRodney W. Grimes  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
32df8bae1dSRodney W. Grimes  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
33df8bae1dSRodney W. Grimes  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
34df8bae1dSRodney W. Grimes  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
35df8bae1dSRodney W. Grimes  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
36df8bae1dSRodney W. Grimes  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
37df8bae1dSRodney W. Grimes  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
38df8bae1dSRodney W. Grimes  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
39df8bae1dSRodney W. Grimes  * SUCH DAMAGE.
40df8bae1dSRodney W. Grimes  *
41df8bae1dSRodney W. Grimes  *	@(#)init_main.c	8.9 (Berkeley) 1/21/94
42df8bae1dSRodney W. Grimes  */
43df8bae1dSRodney W. Grimes 
44677b542eSDavid E. O'Brien #include <sys/cdefs.h>
45677b542eSDavid E. O'Brien __FBSDID("$FreeBSD$");
46677b542eSDavid E. O'Brien 
4777fe4438SBenno Rice #include "opt_ddb.h"
4867481196SJohn Birrell #include "opt_init_path.h"
4940244964SRobert Watson #include "opt_mac.h"
504bd49128SPeter Wemm 
51df8bae1dSRodney W. Grimes #include <sys/param.h>
529e209b12SAlfred Perlstein #include <sys/kernel.h>
53f36ba452SJake Burkholder #include <sys/exec.h>
54b3c5c18dSBruce Evans #include <sys/file.h>
55df8bae1dSRodney W. Grimes #include <sys/filedesc.h>
560384fff8SJason Evans #include <sys/ktr.h>
571005a129SJohn Baldwin #include <sys/lock.h>
58ac0ad63fSBruce Evans #include <sys/mount.h>
5935e0e5b3SJohn Baldwin #include <sys/mutex.h>
608f19eb88SIan Dowse #include <sys/syscallsubr.h>
61946bb7a2SPoul-Henning Kamp #include <sys/sysctl.h>
62df8bae1dSRodney W. Grimes #include <sys/proc.h>
63df8bae1dSRodney W. Grimes #include <sys/resourcevar.h>
64df8bae1dSRodney W. Grimes #include <sys/systm.h>
6536240ea5SDoug Rabson #include <sys/signalvar.h>
66df8bae1dSRodney W. Grimes #include <sys/vnode.h>
67f3f0ca60SSøren Schmidt #include <sys/sysent.h>
68df8bae1dSRodney W. Grimes #include <sys/reboot.h>
69de028f5aSJeff Roberson #include <sys/sched.h>
701005a129SJohn Baldwin #include <sys/sx.h>
71ad7507e2SSteven Wallace #include <sys/sysproto.h>
72efeaf95aSDavid Greenman #include <sys/vmmeter.h>
7374b2192aSJohn Dyson #include <sys/unistd.h>
7494e9d7c1SPeter Wemm #include <sys/malloc.h>
75db901281SPoul-Henning Kamp #include <sys/conf.h>
76df8bae1dSRodney W. Grimes 
77df8bae1dSRodney W. Grimes #include <machine/cpu.h>
78df8bae1dSRodney W. Grimes 
793683665bSRobert Watson #include <security/audit/audit.h>
80aed55708SRobert Watson #include <security/mac/mac_framework.h>
813683665bSRobert Watson 
82df8bae1dSRodney W. Grimes #include <vm/vm.h>
83efeaf95aSDavid Greenman #include <vm/vm_param.h>
84efeaf95aSDavid Greenman #include <vm/pmap.h>
85efeaf95aSDavid Greenman #include <vm/vm_map.h>
866faaa756SWolfram Schneider #include <sys/copyright.h>
87df8bae1dSRodney W. Grimes 
8826ab616fSBenno Rice #include <ddb/ddb.h>
8926ab616fSBenno Rice #include <ddb/db_sym.h>
9026ab616fSBenno Rice 
9137b087a6SPeter Wemm void mi_startup(void);				/* Should be elsewhere */
92df8bae1dSRodney W. Grimes 
93df8bae1dSRodney W. Grimes /* Components of the first process -- never freed. */
94154c04e5SPoul-Henning Kamp static struct session session0;
95154c04e5SPoul-Henning Kamp static struct pgrp pgrp0;
96df8bae1dSRodney W. Grimes struct	proc proc0;
972a3b1065SOlivier Houchard struct	thread thread0 __aligned(8);
984f0db5e0SJulian Elischer struct	ksegrp ksegrp0;
9923734995SJake Burkholder struct	vmspace vmspace0;
1002b14f991SJulian Elischer struct	proc *initproc;
101df8bae1dSRodney W. Grimes 
102c463cf1cSBruce Evans int	boothowto = 0;		/* initialized so that it can be patched */
1038d0bf3d6SJordan K. Hubbard SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, "");
1047a08bae6SJohn Baldwin int	bootverbose;
1057a08bae6SJohn Baldwin SYSCTL_INT(_debug, OID_AUTO, bootverbose, CTLFLAG_RW, &bootverbose, 0, "");
106946bb7a2SPoul-Henning Kamp 
1072b14f991SJulian Elischer /*
1082b14f991SJulian Elischer  * This ensures that there is at least one entry so that the sysinit_set
1092b14f991SJulian Elischer  * symbol is not undefined.  A sybsystem ID of SI_SUB_DUMMY is never
1102b14f991SJulian Elischer  * executed.
11126f9a767SRodney W. Grimes  */
1122b14f991SJulian Elischer SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL)
1138a129caeSDavid Greenman 
11494e9d7c1SPeter Wemm /*
11594e9d7c1SPeter Wemm  * The sysinit table itself.  Items are checked off as the are run.
11694e9d7c1SPeter Wemm  * If we want to register new sysinit types, add them to newsysinit.
11794e9d7c1SPeter Wemm  */
118f41325dbSPeter Wemm SET_DECLARE(sysinit_set, struct sysinit);
119f41325dbSPeter Wemm struct sysinit **sysinit, **sysinit_end;
120f41325dbSPeter Wemm struct sysinit **newsysinit, **newsysinit_end;
12194e9d7c1SPeter Wemm 
12294e9d7c1SPeter Wemm /*
12394e9d7c1SPeter Wemm  * Merge a new sysinit set into the current set, reallocating it if
12494e9d7c1SPeter Wemm  * necessary.  This can only be called after malloc is running.
12594e9d7c1SPeter Wemm  */
12694e9d7c1SPeter Wemm void
127f41325dbSPeter Wemm sysinit_add(struct sysinit **set, struct sysinit **set_end)
12894e9d7c1SPeter Wemm {
12994e9d7c1SPeter Wemm 	struct sysinit **newset;
13094e9d7c1SPeter Wemm 	struct sysinit **sipp;
13194e9d7c1SPeter Wemm 	struct sysinit **xipp;
132f41325dbSPeter Wemm 	int count;
13394e9d7c1SPeter Wemm 
134f41325dbSPeter Wemm 	count = set_end - set;
135ddd62546SPeter Wemm 	if (newsysinit)
136f41325dbSPeter Wemm 		count += newsysinit_end - newsysinit;
137ddd62546SPeter Wemm 	else
138f41325dbSPeter Wemm 		count += sysinit_end - sysinit;
13994e9d7c1SPeter Wemm 	newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
14094e9d7c1SPeter Wemm 	if (newset == NULL)
14194e9d7c1SPeter Wemm 		panic("cannot malloc for sysinit");
14294e9d7c1SPeter Wemm 	xipp = newset;
143ddd62546SPeter Wemm 	if (newsysinit)
144f41325dbSPeter Wemm 		for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
145ddd62546SPeter Wemm 			*xipp++ = *sipp;
146ddd62546SPeter Wemm 	else
147f41325dbSPeter Wemm 		for (sipp = sysinit; sipp < sysinit_end; sipp++)
14894e9d7c1SPeter Wemm 			*xipp++ = *sipp;
149f41325dbSPeter Wemm 	for (sipp = set; sipp < set_end; sipp++)
15094e9d7c1SPeter Wemm 		*xipp++ = *sipp;
151ddd62546SPeter Wemm 	if (newsysinit)
152ddd62546SPeter Wemm 		free(newsysinit, M_TEMP);
15394e9d7c1SPeter Wemm 	newsysinit = newset;
154f41325dbSPeter Wemm 	newsysinit_end = newset + count;
15594e9d7c1SPeter Wemm }
15626f9a767SRodney W. Grimes 
157df8bae1dSRodney W. Grimes /*
158df8bae1dSRodney W. Grimes  * System startup; initialize the world, create process 0, mount root
159df8bae1dSRodney W. Grimes  * filesystem, and fork to create init and pagedaemon.  Most of the
160df8bae1dSRodney W. Grimes  * hard work is done in the lower-level initialization routines including
161df8bae1dSRodney W. Grimes  * startup(), which does memory initialization and autoconfiguration.
1622b14f991SJulian Elischer  *
1632b14f991SJulian Elischer  * This allows simple addition of new kernel subsystems that require
1642b14f991SJulian Elischer  * boot time initialization.  It also allows substitution of subsystem
1652b14f991SJulian Elischer  * (for instance, a scheduler, kernel profiler, or VM system) by object
166c5b193bfSPoul-Henning Kamp  * module.  Finally, it allows for optional "kernel threads".
167df8bae1dSRodney W. Grimes  */
16826f9a767SRodney W. Grimes void
16937b087a6SPeter Wemm mi_startup(void)
170df8bae1dSRodney W. Grimes {
1712b14f991SJulian Elischer 
1722b14f991SJulian Elischer 	register struct sysinit **sipp;		/* system initialization*/
1732b14f991SJulian Elischer 	register struct sysinit **xipp;		/* interior loop of sort*/
1742b14f991SJulian Elischer 	register struct sysinit *save;		/* bubble*/
1752b14f991SJulian Elischer 
17626ab616fSBenno Rice #if defined(VERBOSE_SYSINIT)
17726ab616fSBenno Rice 	int last;
17826ab616fSBenno Rice 	int verbose;
17926ab616fSBenno Rice #endif
18026ab616fSBenno Rice 
181f41325dbSPeter Wemm 	if (sysinit == NULL) {
182f41325dbSPeter Wemm 		sysinit = SET_BEGIN(sysinit_set);
183f41325dbSPeter Wemm 		sysinit_end = SET_LIMIT(sysinit_set);
184f41325dbSPeter Wemm 	}
185f41325dbSPeter Wemm 
18694e9d7c1SPeter Wemm restart:
1872b14f991SJulian Elischer 	/*
1882b14f991SJulian Elischer 	 * Perform a bubble sort of the system initialization objects by
1892b14f991SJulian Elischer 	 * their subsystem (primary key) and order (secondary key).
1902b14f991SJulian Elischer 	 */
191f41325dbSPeter Wemm 	for (sipp = sysinit; sipp < sysinit_end; sipp++) {
192f41325dbSPeter Wemm 		for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
1932b14f991SJulian Elischer 			if ((*sipp)->subsystem < (*xipp)->subsystem ||
1942b14f991SJulian Elischer 			     ((*sipp)->subsystem == (*xipp)->subsystem &&
195af4b2d2dSPeter Wemm 			      (*sipp)->order <= (*xipp)->order))
1962b14f991SJulian Elischer 				continue;	/* skip*/
1972b14f991SJulian Elischer 			save = *sipp;
1982b14f991SJulian Elischer 			*sipp = *xipp;
1992b14f991SJulian Elischer 			*xipp = save;
2002b14f991SJulian Elischer 		}
2012b14f991SJulian Elischer 	}
2022b14f991SJulian Elischer 
20326ab616fSBenno Rice #if defined(VERBOSE_SYSINIT)
20426ab616fSBenno Rice 	last = SI_SUB_COPYRIGHT;
20526ab616fSBenno Rice 	verbose = 0;
20626ab616fSBenno Rice #if !defined(DDB)
20726ab616fSBenno Rice 	printf("VERBOSE_SYSINIT: DDB not enabled, symbol lookups disabled.\n");
20826ab616fSBenno Rice #endif
20926ab616fSBenno Rice #endif
21026ab616fSBenno Rice 
2112b14f991SJulian Elischer 	/*
2122b14f991SJulian Elischer 	 * Traverse the (now) ordered list of system initialization tasks.
2132b14f991SJulian Elischer 	 * Perform each task, and continue on to the next task.
2142b14f991SJulian Elischer 	 *
2152b14f991SJulian Elischer 	 * The last item on the list is expected to be the scheduler,
2162b14f991SJulian Elischer 	 * which will not return.
2172b14f991SJulian Elischer 	 */
218f41325dbSPeter Wemm 	for (sipp = sysinit; sipp < sysinit_end; sipp++) {
21974b2192aSJohn Dyson 
2202b14f991SJulian Elischer 		if ((*sipp)->subsystem == SI_SUB_DUMMY)
2212b14f991SJulian Elischer 			continue;	/* skip dummy task(s)*/
2222b14f991SJulian Elischer 
22394e9d7c1SPeter Wemm 		if ((*sipp)->subsystem == SI_SUB_DONE)
22494e9d7c1SPeter Wemm 			continue;
22594e9d7c1SPeter Wemm 
22626ab616fSBenno Rice #if defined(VERBOSE_SYSINIT)
22726ab616fSBenno Rice 		if ((*sipp)->subsystem > last) {
22826ab616fSBenno Rice 			verbose = 1;
22926ab616fSBenno Rice 			last = (*sipp)->subsystem;
23026ab616fSBenno Rice 			printf("subsystem %x\n", last);
23126ab616fSBenno Rice 		}
23226ab616fSBenno Rice 		if (verbose) {
23326ab616fSBenno Rice #if defined(DDB)
23426ab616fSBenno Rice 			const char *name;
23526ab616fSBenno Rice 			c_db_sym_t sym;
23626ab616fSBenno Rice 			db_expr_t  offset;
23726ab616fSBenno Rice 
23826ab616fSBenno Rice 			sym = db_search_symbol((vm_offset_t)(*sipp)->func,
23926ab616fSBenno Rice 			    DB_STGY_PROC, &offset);
24026ab616fSBenno Rice 			db_symbol_values(sym, &name, NULL);
24126ab616fSBenno Rice 			if (name != NULL)
24226ab616fSBenno Rice 				printf("   %s(%p)... ", name, (*sipp)->udata);
24326ab616fSBenno Rice 			else
24426ab616fSBenno Rice #endif
24526ab616fSBenno Rice 				printf("   %p(%p)... ", (*sipp)->func,
24626ab616fSBenno Rice 				    (*sipp)->udata);
24726ab616fSBenno Rice 		}
24826ab616fSBenno Rice #endif
24926ab616fSBenno Rice 
2509c8b8baaSPeter Wemm 		/* Call function */
2512b14f991SJulian Elischer 		(*((*sipp)->func))((*sipp)->udata);
25294e9d7c1SPeter Wemm 
25326ab616fSBenno Rice #if defined(VERBOSE_SYSINIT)
25426ab616fSBenno Rice 		if (verbose)
25526ab616fSBenno Rice 			printf("done.\n");
25626ab616fSBenno Rice #endif
25726ab616fSBenno Rice 
25894e9d7c1SPeter Wemm 		/* Check off the one we're just done */
25994e9d7c1SPeter Wemm 		(*sipp)->subsystem = SI_SUB_DONE;
26094e9d7c1SPeter Wemm 
26194e9d7c1SPeter Wemm 		/* Check if we've installed more sysinit items via KLD */
26294e9d7c1SPeter Wemm 		if (newsysinit != NULL) {
263f41325dbSPeter Wemm 			if (sysinit != SET_BEGIN(sysinit_set))
26494e9d7c1SPeter Wemm 				free(sysinit, M_TEMP);
26594e9d7c1SPeter Wemm 			sysinit = newsysinit;
266f41325dbSPeter Wemm 			sysinit_end = newsysinit_end;
26794e9d7c1SPeter Wemm 			newsysinit = NULL;
268f41325dbSPeter Wemm 			newsysinit_end = NULL;
26994e9d7c1SPeter Wemm 			goto restart;
27094e9d7c1SPeter Wemm 		}
2712b14f991SJulian Elischer 	}
2722b14f991SJulian Elischer 
273477a642cSPeter Wemm 	panic("Shouldn't get here!");
2742b14f991SJulian Elischer 	/* NOTREACHED*/
2752b14f991SJulian Elischer }
2762b14f991SJulian Elischer 
2772b14f991SJulian Elischer 
2782b14f991SJulian Elischer /*
2792b14f991SJulian Elischer  ***************************************************************************
2802b14f991SJulian Elischer  ****
2812b14f991SJulian Elischer  **** The following SYSINIT's belong elsewhere, but have not yet
2822b14f991SJulian Elischer  **** been moved.
2832b14f991SJulian Elischer  ****
2842b14f991SJulian Elischer  ***************************************************************************
2852b14f991SJulian Elischer  */
2869ef6c28aSBruce Evans static void
28737b087a6SPeter Wemm print_caddr_t(void *data __unused)
2889ef6c28aSBruce Evans {
2899ef6c28aSBruce Evans 	printf("%s", (char *)data);
2909ef6c28aSBruce Evans }
291d841aaa7SBruce Evans SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
2925add74b4SRobert Watson SYSINIT(trademark, SI_SUB_COPYRIGHT, SI_ORDER_SECOND, print_caddr_t, trademark)
2935add74b4SRobert Watson SYSINIT(version, SI_SUB_COPYRIGHT, SI_ORDER_THIRD, print_caddr_t, version)
2947a08bae6SJohn Baldwin 
2952cf6bdacSPoul-Henning Kamp #ifdef WITNESS
2962cf6bdacSPoul-Henning Kamp static char wit_warn[] =
2972cf6bdacSPoul-Henning Kamp      "WARNING: WITNESS option enabled, expect reduced performance.\n";
2985add74b4SRobert Watson SYSINIT(witwarn, SI_SUB_COPYRIGHT, SI_ORDER_THIRD + 1,
2992cf6bdacSPoul-Henning Kamp    print_caddr_t, wit_warn)
3002cf6bdacSPoul-Henning Kamp #endif
3012cf6bdacSPoul-Henning Kamp 
3022cf6bdacSPoul-Henning Kamp #ifdef DIAGNOSTIC
3032cf6bdacSPoul-Henning Kamp static char diag_warn[] =
3042cf6bdacSPoul-Henning Kamp      "WARNING: DIAGNOSTIC option enabled, expect reduced performance.\n";
30588b85279SRobert Watson SYSINIT(diagwarn, SI_SUB_COPYRIGHT, SI_ORDER_THIRD + 2,
3062cf6bdacSPoul-Henning Kamp     print_caddr_t, diag_warn)
3072cf6bdacSPoul-Henning Kamp #endif
3082cf6bdacSPoul-Henning Kamp 
3097a08bae6SJohn Baldwin static void
3107a08bae6SJohn Baldwin set_boot_verbose(void *data __unused)
3117a08bae6SJohn Baldwin {
3127a08bae6SJohn Baldwin 
3137a08bae6SJohn Baldwin 	if (boothowto & RB_VERBOSE)
3147a08bae6SJohn Baldwin 		bootverbose++;
3157a08bae6SJohn Baldwin }
3167a08bae6SJohn Baldwin SYSINIT(boot_verbose, SI_SUB_TUNABLES, SI_ORDER_ANY, set_boot_verbose, NULL)
3172b14f991SJulian Elischer 
318f36ba452SJake Burkholder struct sysentvec null_sysvec = {
319f36ba452SJake Burkholder 	0,
320f36ba452SJake Burkholder 	NULL,
321f36ba452SJake Burkholder 	0,
322f36ba452SJake Burkholder 	0,
323f36ba452SJake Burkholder 	NULL,
324f36ba452SJake Burkholder 	0,
325f36ba452SJake Burkholder 	NULL,
326f36ba452SJake Burkholder 	NULL,
327f36ba452SJake Burkholder 	NULL,
328f36ba452SJake Burkholder 	NULL,
329f36ba452SJake Burkholder 	NULL,
330f36ba452SJake Burkholder 	NULL,
331f36ba452SJake Burkholder 	NULL,
332f36ba452SJake Burkholder 	"null",
333f36ba452SJake Burkholder 	NULL,
334f36ba452SJake Burkholder 	NULL,
335f36ba452SJake Burkholder 	0,
336f36ba452SJake Burkholder 	PAGE_SIZE,
337f36ba452SJake Burkholder 	VM_MIN_ADDRESS,
338f36ba452SJake Burkholder 	VM_MAXUSER_ADDRESS,
339f36ba452SJake Burkholder 	USRSTACK,
340f36ba452SJake Burkholder 	PS_STRINGS,
341f36ba452SJake Burkholder 	VM_PROT_ALL,
342f36ba452SJake Burkholder 	NULL,
343c460ac3aSPeter Wemm 	NULL,
344f36ba452SJake Burkholder 	NULL
345f36ba452SJake Burkholder };
3463ebc1248SPeter Wemm 
3472b14f991SJulian Elischer /*
3482b14f991SJulian Elischer  ***************************************************************************
3492b14f991SJulian Elischer  ****
3500c920c0dSGiorgos Keramidas  **** The two following SYSINIT's are proc0 specific glue code.  I am not
3512b14f991SJulian Elischer  **** convinced that they can not be safely combined, but their order of
3522b14f991SJulian Elischer  **** operation has been maintained as the same as the original init_main.c
3532b14f991SJulian Elischer  **** for right now.
3542b14f991SJulian Elischer  ****
3552b14f991SJulian Elischer  **** These probably belong in init_proc.c or kern_proc.c, since they
3562b14f991SJulian Elischer  **** deal with proc0 (the fork template process).
3572b14f991SJulian Elischer  ****
3582b14f991SJulian Elischer  ***************************************************************************
3592b14f991SJulian Elischer  */
3602b14f991SJulian Elischer /* ARGSUSED*/
361154c04e5SPoul-Henning Kamp static void
36237b087a6SPeter Wemm proc0_init(void *dummy __unused)
3632b14f991SJulian Elischer {
3648ec21e3aSPoul-Henning Kamp 	struct proc *p;
3658ec21e3aSPoul-Henning Kamp 	unsigned i;
366b40ce416SJulian Elischer 	struct thread *td;
367079b7badSJulian Elischer 	struct ksegrp *kg;
368df8bae1dSRodney W. Grimes 
3690cddd8f0SMatthew Dillon 	GIANT_REQUIRED;
370df8bae1dSRodney W. Grimes 	p = &proc0;
371079b7badSJulian Elischer 	td = &thread0;
3724f0db5e0SJulian Elischer 	kg = &ksegrp0;
373df8bae1dSRodney W. Grimes 
374df8bae1dSRodney W. Grimes 	/*
3750384fff8SJason Evans 	 * Initialize magic number.
3760384fff8SJason Evans 	 */
3770384fff8SJason Evans 	p->p_magic = P_MAGIC;
3780384fff8SJason Evans 
3790384fff8SJason Evans 	/*
380ed062c8dSJulian Elischer 	 * Initialize thread, process and ksegrp structures.
381a3bfb996SJeffrey Hsu 	 */
382ed062c8dSJulian Elischer 	procinit();	/* set up proc zone */
383ed062c8dSJulian Elischer 	threadinit();	/* set up thead, upcall and KSEGRP zones */
384a3bfb996SJeffrey Hsu 
385a3bfb996SJeffrey Hsu 	/*
386ed062c8dSJulian Elischer 	 * Initialise scheduler resources.
387ed062c8dSJulian Elischer 	 * Add scheduler specific parts to proc, ksegrp, thread as needed.
388ed062c8dSJulian Elischer 	 */
389ed062c8dSJulian Elischer 	schedinit();	/* scheduler gets its house in order */
390ed062c8dSJulian Elischer 	/*
391b1508c72SDavid Greenman 	 * Initialize sleep queue hash table
392b1508c72SDavid Greenman 	 */
393b1508c72SDavid Greenman 	sleepinit();
394b1508c72SDavid Greenman 
395b1508c72SDavid Greenman 	/*
3963075778bSJohn Dyson 	 * additional VM structures
3973075778bSJohn Dyson 	 */
3983075778bSJohn Dyson 	vm_init2();
3993075778bSJohn Dyson 
4003075778bSJohn Dyson 	/*
401df8bae1dSRodney W. Grimes 	 * Create process 0 (the swapper).
402df8bae1dSRodney W. Grimes 	 */
403a3bfb996SJeffrey Hsu 	LIST_INSERT_HEAD(&allproc, p, p_list);
4047b531e60SJohn Baldwin 	LIST_INSERT_HEAD(PIDHASH(0), p, p_hash);
4056008862bSJohn Baldwin 	mtx_init(&pgrp0.pg_mtx, "process group", NULL, MTX_DEF | MTX_DUPOK);
406df8bae1dSRodney W. Grimes 	p->p_pgrp = &pgrp0;
407a3bfb996SJeffrey Hsu 	LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
408a3bfb996SJeffrey Hsu 	LIST_INIT(&pgrp0.pg_members);
409a3bfb996SJeffrey Hsu 	LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
410a3bfb996SJeffrey Hsu 
411df8bae1dSRodney W. Grimes 	pgrp0.pg_session = &session0;
4126008862bSJohn Baldwin 	mtx_init(&session0.s_mtx, "session", NULL, MTX_DEF);
413df8bae1dSRodney W. Grimes 	session0.s_count = 1;
414df8bae1dSRodney W. Grimes 	session0.s_leader = p;
415df8bae1dSRodney W. Grimes 
4163ebc1248SPeter Wemm 	p->p_sysent = &null_sysvec;
417f202965eSJohn Baldwin 	p->p_flag = P_SYSTEM;
418f202965eSJohn Baldwin 	p->p_sflag = PS_INMEM;
419e602ba25SJulian Elischer 	p->p_state = PRS_NORMAL;
420571dcd15SSuleiman Souhlal 	knlist_init(&p->p_klist, &p->p_mtx, NULL, NULL, NULL);
4212c255e9dSRobert Watson 	STAILQ_INIT(&p->p_ktr);
422fa885116SJulian Elischer 	p->p_nice = NZERO;
423e602ba25SJulian Elischer 	td->td_state = TDS_RUNNING;
4242c100766SJulian Elischer 	kg->kg_pri_class = PRI_TIMESHARE;
4252c100766SJulian Elischer 	kg->kg_user_pri = PUSER;
42643863138SDavid Xu 	kg->kg_base_user_pri = PUSER;
4272c100766SJulian Elischer 	td->td_priority = PVM;
4282c100766SJulian Elischer 	td->td_base_pri = PUSER;
429060563ecSJulian Elischer 	td->td_oncpu = 0;
4302c1011f7SJohn Dyson 	p->p_peers = 0;
4312c1011f7SJohn Dyson 	p->p_leader = p;
432e602ba25SJulian Elischer 
4332c1011f7SJohn Dyson 
434df8bae1dSRodney W. Grimes 	bcopy("swapper", p->p_comm, sizeof ("swapper"));
435df8bae1dSRodney W. Grimes 
436c06eb4e2SSam Leffler 	callout_init(&p->p_itcallout, CALLOUT_MPSAFE);
437c06eb4e2SSam Leffler 	callout_init(&td->td_slpcallout, CALLOUT_MPSAFE);
4384f559836SJake Burkholder 
439df8bae1dSRodney W. Grimes 	/* Create credentials. */
440df8bae1dSRodney W. Grimes 	p->p_ucred = crget();
441df8bae1dSRodney W. Grimes 	p->p_ucred->cr_ngroups = 1;	/* group 0 */
442f535380cSDon Lewis 	p->p_ucred->cr_uidinfo = uifind(0);
443b1fc0ec1SRobert Watson 	p->p_ucred->cr_ruidinfo = uifind(0);
44491421ba2SRobert Watson 	p->p_ucred->cr_prison = NULL;	/* Don't jail it. */
4453683665bSRobert Watson #ifdef AUDIT
4463683665bSRobert Watson 	audit_proc_alloc(p);
4473683665bSRobert Watson 	audit_proc_kproc0(p);
4483683665bSRobert Watson #endif
44940244964SRobert Watson #ifdef MAC
45040244964SRobert Watson 	mac_create_proc0(p->p_ucred);
45140244964SRobert Watson #endif
4528e2e767bSJohn Baldwin 	td->td_ucred = crhold(p->p_ucred);
45375c13541SPoul-Henning Kamp 
45490af4afaSJohn Baldwin 	/* Create sigacts. */
45590af4afaSJohn Baldwin 	p->p_sigacts = sigacts_alloc();
4566626c604SJulian Elischer 
45737b087a6SPeter Wemm 	/* Initialize signal state for process 0. */
45837b087a6SPeter Wemm 	siginit(&proc0);
45937b087a6SPeter Wemm 
460df8bae1dSRodney W. Grimes 	/* Create the file descriptor table. */
4618ec21e3aSPoul-Henning Kamp 	p->p_fd = fdinit(NULL);
462ad05d580STor Egge 	p->p_fdtol = NULL;
463df8bae1dSRodney W. Grimes 
464df8bae1dSRodney W. Grimes 	/* Create the limits structures. */
46591d5354aSJohn Baldwin 	p->p_limit = lim_alloc();
46691d5354aSJohn Baldwin 	for (i = 0; i < RLIM_NLIMITS; i++)
46791d5354aSJohn Baldwin 		p->p_limit->pl_rlimit[i].rlim_cur =
46891d5354aSJohn Baldwin 		    p->p_limit->pl_rlimit[i].rlim_max = RLIM_INFINITY;
46991d5354aSJohn Baldwin 	p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_cur =
47091d5354aSJohn Baldwin 	    p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
47191d5354aSJohn Baldwin 	p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_cur =
47291d5354aSJohn Baldwin 	    p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
473df8bae1dSRodney W. Grimes 	i = ptoa(cnt.v_free_count);
47491d5354aSJohn Baldwin 	p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_max = i;
47591d5354aSJohn Baldwin 	p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
47691d5354aSJohn Baldwin 	p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
4775715307fSJohn Baldwin 	p->p_cpulimit = RLIM_INFINITY;
478df8bae1dSRodney W. Grimes 
4798b059651SDavid Schultz 	p->p_stats = pstats_alloc();
4808b059651SDavid Schultz 
481df8bae1dSRodney W. Grimes 	/* Allocate a prototype map so we have something to fork. */
482b1028ad1SLuoqi Chen 	pmap_pinit0(vmspace_pmap(&vmspace0));
483df8bae1dSRodney W. Grimes 	p->p_vmspace = &vmspace0;
484df8bae1dSRodney W. Grimes 	vmspace0.vm_refcnt = 1;
48505ba50f5SJake Burkholder 	vm_map_init(&vmspace0.vm_map, p->p_sysent->sv_minuser,
48605ba50f5SJake Burkholder 	    p->p_sysent->sv_maxuser);
487b1028ad1SLuoqi Chen 	vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0);
488df8bae1dSRodney W. Grimes 
48939fb8e6bSJulian Elischer 	/*
490a3bfb996SJeffrey Hsu 	 * Charge root for one process.
491df8bae1dSRodney W. Grimes 	 */
492b1fc0ec1SRobert Watson 	(void)chgproccnt(p->p_ucred->cr_ruidinfo, 1, 0);
49326f9a767SRodney W. Grimes }
4942b14f991SJulian Elischer SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL)
4952b14f991SJulian Elischer 
4962b14f991SJulian Elischer /* ARGSUSED*/
497154c04e5SPoul-Henning Kamp static void
49837b087a6SPeter Wemm proc0_post(void *dummy __unused)
4992b14f991SJulian Elischer {
50000af9731SPoul-Henning Kamp 	struct timespec ts;
50137b087a6SPeter Wemm 	struct proc *p;
502a6fc8288SPeter Wemm 
5032b14f991SJulian Elischer 	/*
5041b0b259eSBruce Evans 	 * Now we can look at the time, having had a chance to verify the
5051b0b259eSBruce Evans 	 * time from the filesystem.  Pretend that proc0 started now.
5062b14f991SJulian Elischer 	 */
5071005a129SJohn Baldwin 	sx_slock(&allproc_lock);
50837b087a6SPeter Wemm 	LIST_FOREACH(p, &allproc, p_list) {
50987ccef7bSDag-Erling Smørgrav 		microuptime(&p->p_stats->p_start);
5105b1a8eb3SPoul-Henning Kamp 		p->p_rux.rux_runtime = 0;
51137b087a6SPeter Wemm 	}
5121005a129SJohn Baldwin 	sx_sunlock(&allproc_lock);
5135b1a8eb3SPoul-Henning Kamp 	PCPU_SET(switchtime, cpu_ticks());
5140384fff8SJason Evans 	PCPU_SET(switchticks, ticks);
5152b14f991SJulian Elischer 
516a6fc8288SPeter Wemm 	/*
517a6fc8288SPeter Wemm 	 * Give the ``random'' number generator a thump.
518a6fc8288SPeter Wemm 	 */
51900af9731SPoul-Henning Kamp 	nanotime(&ts);
52000af9731SPoul-Henning Kamp 	srandom(ts.tv_sec ^ ts.tv_nsec);
5212b14f991SJulian Elischer }
5222b14f991SJulian Elischer SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL)
5232b14f991SJulian Elischer 
524df8bae1dSRodney W. Grimes /*
5252b14f991SJulian Elischer  ***************************************************************************
5262b14f991SJulian Elischer  ****
5272b14f991SJulian Elischer  **** The following SYSINIT's and glue code should be moved to the
5282b14f991SJulian Elischer  **** respective files on a per subsystem basis.
5292b14f991SJulian Elischer  ****
5302b14f991SJulian Elischer  ***************************************************************************
531df8bae1dSRodney W. Grimes  */
532df8bae1dSRodney W. Grimes 
533df8bae1dSRodney W. Grimes 
534df8bae1dSRodney W. Grimes /*
5352b14f991SJulian Elischer  ***************************************************************************
5362b14f991SJulian Elischer  ****
5372b14f991SJulian Elischer  **** The following code probably belongs in another file, like
53837b087a6SPeter Wemm  **** kern/init_init.c.
5392b14f991SJulian Elischer  ****
5402b14f991SJulian Elischer  ***************************************************************************
541df8bae1dSRodney W. Grimes  */
542df8bae1dSRodney W. Grimes 
543df8bae1dSRodney W. Grimes /*
544df8bae1dSRodney W. Grimes  * List of paths to try when searching for "init".
545df8bae1dSRodney W. Grimes  */
5465f967b24SDag-Erling Smørgrav static char init_path[MAXPATHLEN] =
54767481196SJohn Birrell #ifdef	INIT_PATH
54867481196SJohn Birrell     __XSTRING(INIT_PATH);
54967481196SJohn Birrell #else
550f3f4baf0SDag-Erling Smørgrav     "/sbin/init:/sbin/oinit:/sbin/init.bak:/rescue/init:/stand/sysinstall";
55167481196SJohn Birrell #endif
552af1408e3SLuigi Rizzo SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0,
553af1408e3SLuigi Rizzo 	"Path used to search the init process");
554df8bae1dSRodney W. Grimes 
555df8bae1dSRodney W. Grimes /*
556724447acSRalf S. Engelschall  * Shutdown timeout of init(8).
557724447acSRalf S. Engelschall  * Unused within kernel, but used to control init(8), hence do not remove.
558724447acSRalf S. Engelschall  */
559724447acSRalf S. Engelschall #ifndef INIT_SHUTDOWN_TIMEOUT
560724447acSRalf S. Engelschall #define INIT_SHUTDOWN_TIMEOUT 120
561724447acSRalf S. Engelschall #endif
562724447acSRalf S. Engelschall static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT;
563724447acSRalf S. Engelschall SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout,
564724447acSRalf S. Engelschall 	CTLFLAG_RW, &init_shutdown_timeout, 0, "");
565724447acSRalf S. Engelschall 
566724447acSRalf S. Engelschall /*
5675f967b24SDag-Erling Smørgrav  * Start the initial user process; try exec'ing each pathname in init_path.
568df8bae1dSRodney W. Grimes  * The program is invoked with one argument containing the boot flags.
569df8bae1dSRodney W. Grimes  */
570df8bae1dSRodney W. Grimes static void
57137b087a6SPeter Wemm start_init(void *dummy)
572df8bae1dSRodney W. Grimes {
573df8bae1dSRodney W. Grimes 	vm_offset_t addr;
574df8bae1dSRodney W. Grimes 	struct execve_args args;
5755f967b24SDag-Erling Smørgrav 	int options, error;
5765f967b24SDag-Erling Smørgrav 	char *var, *path, *next, *s;
5775f967b24SDag-Erling Smørgrav 	char *ucp, **uap, *arg0, *arg1;
578b40ce416SJulian Elischer 	struct thread *td;
5799c8b8baaSPeter Wemm 	struct proc *p;
580df8bae1dSRodney W. Grimes 
5819ed346baSBosko Milekic 	mtx_lock(&Giant);
5820384fff8SJason Evans 
5830cddd8f0SMatthew Dillon 	GIANT_REQUIRED;
5840cddd8f0SMatthew Dillon 
585b40ce416SJulian Elischer 	td = curthread;
586b40ce416SJulian Elischer 	p = td->td_proc;
587df8bae1dSRodney W. Grimes 
588563af2ecSMaxime Henrion 	vfs_mountroot();
589fb92273bSPoul-Henning Kamp 
590b7116168SPeter Wemm 	/*
591df8bae1dSRodney W. Grimes 	 * Need just enough stack to hold the faked-up "execve()" arguments.
592df8bae1dSRodney W. Grimes 	 */
59305ba50f5SJake Burkholder 	addr = p->p_sysent->sv_usrstack - PAGE_SIZE;
5945f967b24SDag-Erling Smørgrav 	if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
5955f967b24SDag-Erling Smørgrav 			FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0)
596df8bae1dSRodney W. Grimes 		panic("init: couldn't allocate argument space");
597df8bae1dSRodney W. Grimes 	p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
59826f9a767SRodney W. Grimes 	p->p_vmspace->vm_ssize = 1;
599df8bae1dSRodney W. Grimes 
6005f967b24SDag-Erling Smørgrav 	if ((var = getenv("init_path")) != NULL) {
601e80fb434SRobert Drehmel 		strlcpy(init_path, var, sizeof(init_path));
602d786139cSMaxime Henrion 		freeenv(var);
6035f967b24SDag-Erling Smørgrav 	}
6045f967b24SDag-Erling Smørgrav 
60567481196SJohn Birrell 	for (path = init_path; *path != '\0'; path = next) {
606e4082284SJohn Birrell 		while (*path == ':')
6075f967b24SDag-Erling Smørgrav 			path++;
60867481196SJohn Birrell 		if (*path == '\0')
6095f967b24SDag-Erling Smørgrav 			break;
610e4082284SJohn Birrell 		for (next = path; *next != '\0' && *next != ':'; next++)
6115f967b24SDag-Erling Smørgrav 			/* nothing */ ;
6125f967b24SDag-Erling Smørgrav 		if (bootverbose)
613ba41a07dSDmitrij Tejblum 			printf("start_init: trying %.*s\n", (int)(next - path),
614ba41a07dSDmitrij Tejblum 			    path);
6155f967b24SDag-Erling Smørgrav 
616df8bae1dSRodney W. Grimes 		/*
617df8bae1dSRodney W. Grimes 		 * Move out the boot flag argument.
618df8bae1dSRodney W. Grimes 		 */
619df8bae1dSRodney W. Grimes 		options = 0;
62005ba50f5SJake Burkholder 		ucp = (char *)p->p_sysent->sv_usrstack;
621df8bae1dSRodney W. Grimes 		(void)subyte(--ucp, 0);		/* trailing zero */
622df8bae1dSRodney W. Grimes 		if (boothowto & RB_SINGLE) {
623df8bae1dSRodney W. Grimes 			(void)subyte(--ucp, 's');
624df8bae1dSRodney W. Grimes 			options = 1;
625df8bae1dSRodney W. Grimes 		}
626df8bae1dSRodney W. Grimes #ifdef notyet
627df8bae1dSRodney W. Grimes                 if (boothowto & RB_FASTBOOT) {
628df8bae1dSRodney W. Grimes 			(void)subyte(--ucp, 'f');
629df8bae1dSRodney W. Grimes 			options = 1;
630df8bae1dSRodney W. Grimes 		}
631df8bae1dSRodney W. Grimes #endif
63217755ac8SPoul-Henning Kamp 
63317755ac8SPoul-Henning Kamp #ifdef BOOTCDROM
63417755ac8SPoul-Henning Kamp 		(void)subyte(--ucp, 'C');
63517755ac8SPoul-Henning Kamp 		options = 1;
63617755ac8SPoul-Henning Kamp #endif
637db901281SPoul-Henning Kamp 
638df8bae1dSRodney W. Grimes 		if (options == 0)
639df8bae1dSRodney W. Grimes 			(void)subyte(--ucp, '-');
640df8bae1dSRodney W. Grimes 		(void)subyte(--ucp, '-');		/* leading hyphen */
641df8bae1dSRodney W. Grimes 		arg1 = ucp;
642df8bae1dSRodney W. Grimes 
643df8bae1dSRodney W. Grimes 		/*
644df8bae1dSRodney W. Grimes 		 * Move out the file name (also arg 0).
645df8bae1dSRodney W. Grimes 		 */
6465f967b24SDag-Erling Smørgrav 		(void)subyte(--ucp, 0);
6475f967b24SDag-Erling Smørgrav 		for (s = next - 1; s >= path; s--)
6485f967b24SDag-Erling Smørgrav 			(void)subyte(--ucp, *s);
649df8bae1dSRodney W. Grimes 		arg0 = ucp;
650df8bae1dSRodney W. Grimes 
651df8bae1dSRodney W. Grimes 		/*
652df8bae1dSRodney W. Grimes 		 * Move out the arg pointers.
653df8bae1dSRodney W. Grimes 		 */
654a20d7755SDoug Rabson 		uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
655c2da0fd9SBruce Evans 		(void)suword((caddr_t)--uap, (long)0);	/* terminator */
656c2da0fd9SBruce Evans 		(void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
657c2da0fd9SBruce Evans 		(void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
658df8bae1dSRodney W. Grimes 
659df8bae1dSRodney W. Grimes 		/*
660df8bae1dSRodney W. Grimes 		 * Point at the arguments.
661df8bae1dSRodney W. Grimes 		 */
662df8bae1dSRodney W. Grimes 		args.fname = arg0;
66326f9a767SRodney W. Grimes 		args.argv = uap;
66426f9a767SRodney W. Grimes 		args.envv = NULL;
665df8bae1dSRodney W. Grimes 
666df8bae1dSRodney W. Grimes 		/*
667df8bae1dSRodney W. Grimes 		 * Now try to exec the program.  If can't for any reason
668df8bae1dSRodney W. Grimes 		 * other than it doesn't exist, complain.
6692b14f991SJulian Elischer 		 *
67037b087a6SPeter Wemm 		 * Otherwise, return via fork_trampoline() all the way
6719c8b8baaSPeter Wemm 		 * to user mode as init!
672df8bae1dSRodney W. Grimes 		 */
673b40ce416SJulian Elischer 		if ((error = execve(td, &args)) == 0) {
6749ed346baSBosko Milekic 			mtx_unlock(&Giant);
675df8bae1dSRodney W. Grimes 			return;
676db72809dSJohn Baldwin 		}
677df8bae1dSRodney W. Grimes 		if (error != ENOENT)
678ba41a07dSDmitrij Tejblum 			printf("exec %.*s: error %d\n", (int)(next - path),
679ba41a07dSDmitrij Tejblum 			    path, error);
680df8bae1dSRodney W. Grimes 	}
681580e7e5aSGreg Lehey 	printf("init: not found in path %s\n", init_path);
682df8bae1dSRodney W. Grimes 	panic("no init");
683df8bae1dSRodney W. Grimes }
68437b087a6SPeter Wemm 
68537b087a6SPeter Wemm /*
68637b087a6SPeter Wemm  * Like kthread_create(), but runs in it's own address space.
68737b087a6SPeter Wemm  * We do this early to reserve pid 1.
68837b087a6SPeter Wemm  *
68937b087a6SPeter Wemm  * Note special case - do not make it runnable yet.  Other work
69037b087a6SPeter Wemm  * in progress will change this more.
69137b087a6SPeter Wemm  */
69237b087a6SPeter Wemm static void
69337b087a6SPeter Wemm create_init(const void *udata __unused)
69437b087a6SPeter Wemm {
6955a06cb0cSRobert Watson 	struct ucred *newcred, *oldcred;
69637b087a6SPeter Wemm 	int error;
69737b087a6SPeter Wemm 
698316ec49aSScott Long 	error = fork1(&thread0, RFFDG | RFPROC | RFSTOPPED, 0, &initproc);
69937b087a6SPeter Wemm 	if (error)
70037b087a6SPeter Wemm 		panic("cannot fork init: %d\n", error);
701bafc8f25SRobert Watson 	KASSERT(initproc->p_pid == 1, ("create_init: initproc->p_pid != 1"));
7025a06cb0cSRobert Watson 	/* divorce init's credentials from the kernel's */
7035a06cb0cSRobert Watson 	newcred = crget();
704f202965eSJohn Baldwin 	PROC_LOCK(initproc);
705f202965eSJohn Baldwin 	initproc->p_flag |= P_SYSTEM;
7065a06cb0cSRobert Watson 	oldcred = initproc->p_ucred;
7075a06cb0cSRobert Watson 	crcopy(newcred, oldcred);
70840244964SRobert Watson #ifdef MAC
70940244964SRobert Watson 	mac_create_proc1(newcred);
71040244964SRobert Watson #endif
7113683665bSRobert Watson #ifdef AUDIT
7123683665bSRobert Watson 	audit_proc_init(initproc);
7133683665bSRobert Watson #endif
7145a06cb0cSRobert Watson 	initproc->p_ucred = newcred;
715f202965eSJohn Baldwin 	PROC_UNLOCK(initproc);
7165a06cb0cSRobert Watson 	crfree(oldcred);
7174d1a4bb7SRobert Watson 	cred_update_thread(FIRST_THREAD_IN_PROC(initproc));
7189ed346baSBosko Milekic 	mtx_lock_spin(&sched_lock);
719f202965eSJohn Baldwin 	initproc->p_sflag |= PS_INMEM;
7209ed346baSBosko Milekic 	mtx_unlock_spin(&sched_lock);
721079b7badSJulian Elischer 	cpu_set_fork_handler(FIRST_THREAD_IN_PROC(initproc), start_init, NULL);
72237b087a6SPeter Wemm }
72337b087a6SPeter Wemm SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
72437b087a6SPeter Wemm 
72537b087a6SPeter Wemm /*
72637b087a6SPeter Wemm  * Make it runnable now.
72737b087a6SPeter Wemm  */
72837b087a6SPeter Wemm static void
72937b087a6SPeter Wemm kick_init(const void *udata __unused)
73037b087a6SPeter Wemm {
731079b7badSJulian Elischer 	struct thread *td;
732f202965eSJohn Baldwin 
733079b7badSJulian Elischer 	td = FIRST_THREAD_IN_PROC(initproc);
7349ed346baSBosko Milekic 	mtx_lock_spin(&sched_lock);
73571fad9fdSJulian Elischer 	TD_SET_CAN_RUN(td);
7362630e4c9SJulian Elischer 	setrunqueue(td, SRQ_BORING);	/* XXXKSE */
7379ed346baSBosko Milekic 	mtx_unlock_spin(&sched_lock);
73837b087a6SPeter Wemm }
73937b087a6SPeter Wemm SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)
740