xref: /freebsd/sys/kern/init_main.c (revision 41582f28ddf7b42b713b270d41e8da80160b69a3)
19454b2d8SWarner Losh /*-
2df57947fSPedro F. Giffuni  * SPDX-License-Identifier: BSD-4-Clause
3df57947fSPedro F. Giffuni  *
42b14f991SJulian Elischer  * Copyright (c) 1995 Terrence R. Lambert
52b14f991SJulian Elischer  * All rights reserved.
62b14f991SJulian Elischer  *
7df8bae1dSRodney W. Grimes  * Copyright (c) 1982, 1986, 1989, 1991, 1992, 1993
8df8bae1dSRodney W. Grimes  *	The Regents of the University of California.  All rights reserved.
9df8bae1dSRodney W. Grimes  * (c) UNIX System Laboratories, Inc.
10df8bae1dSRodney W. Grimes  * All or some portions of this file are derived from material licensed
11df8bae1dSRodney W. Grimes  * to the University of California by American Telephone and Telegraph
12df8bae1dSRodney W. Grimes  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
13df8bae1dSRodney W. Grimes  * the permission of UNIX System Laboratories, Inc.
14df8bae1dSRodney W. Grimes  *
15df8bae1dSRodney W. Grimes  * Redistribution and use in source and binary forms, with or without
16df8bae1dSRodney W. Grimes  * modification, are permitted provided that the following conditions
17df8bae1dSRodney W. Grimes  * are met:
18df8bae1dSRodney W. Grimes  * 1. Redistributions of source code must retain the above copyright
19df8bae1dSRodney W. Grimes  *    notice, this list of conditions and the following disclaimer.
20df8bae1dSRodney W. Grimes  * 2. Redistributions in binary form must reproduce the above copyright
21df8bae1dSRodney W. Grimes  *    notice, this list of conditions and the following disclaimer in the
22df8bae1dSRodney W. Grimes  *    documentation and/or other materials provided with the distribution.
23df8bae1dSRodney W. Grimes  * 3. All advertising materials mentioning features or use of this software
24df8bae1dSRodney W. Grimes  *    must display the following acknowledgement:
25df8bae1dSRodney W. Grimes  *	This product includes software developed by the University of
26df8bae1dSRodney W. Grimes  *	California, Berkeley and its contributors.
27df8bae1dSRodney W. Grimes  * 4. Neither the name of the University nor the names of its contributors
28df8bae1dSRodney W. Grimes  *    may be used to endorse or promote products derived from this software
29df8bae1dSRodney W. Grimes  *    without specific prior written permission.
30df8bae1dSRodney W. Grimes  *
31df8bae1dSRodney W. Grimes  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
32df8bae1dSRodney W. Grimes  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
33df8bae1dSRodney W. Grimes  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
34df8bae1dSRodney W. Grimes  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
35df8bae1dSRodney W. Grimes  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
36df8bae1dSRodney W. Grimes  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
37df8bae1dSRodney W. Grimes  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
38df8bae1dSRodney W. Grimes  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
39df8bae1dSRodney W. Grimes  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
40df8bae1dSRodney W. Grimes  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
41df8bae1dSRodney W. Grimes  * SUCH DAMAGE.
42df8bae1dSRodney W. Grimes  *
43df8bae1dSRodney W. Grimes  *	@(#)init_main.c	8.9 (Berkeley) 1/21/94
44df8bae1dSRodney W. Grimes  */
45df8bae1dSRodney W. Grimes 
46677b542eSDavid E. O'Brien #include <sys/cdefs.h>
47677b542eSDavid E. O'Brien __FBSDID("$FreeBSD$");
48677b542eSDavid E. O'Brien 
4977fe4438SBenno Rice #include "opt_ddb.h"
50f7a0bb0dSBjoern A. Zeeb #include "opt_kdb.h"
5167481196SJohn Birrell #include "opt_init_path.h"
529541307fSJohn-Mark Gurney #include "opt_verbose_sysinit.h"
534bd49128SPeter Wemm 
54df8bae1dSRodney W. Grimes #include <sys/param.h>
550aa9ffcdSMitchell Horne #include <sys/systm.h>
565a8fceb3SMitchell Horne #include <sys/boottrace.h>
570aa9ffcdSMitchell Horne #include <sys/conf.h>
580aa9ffcdSMitchell Horne #include <sys/cpuset.h>
590aa9ffcdSMitchell Horne #include <sys/dtrace_bsd.h>
60905837ebSGleb Smirnoff #include <sys/epoch.h>
61e2e050c8SConrad Meyer #include <sys/eventhandler.h>
62f36ba452SJake Burkholder #include <sys/exec.h>
63b3c5c18dSBruce Evans #include <sys/file.h>
64df8bae1dSRodney W. Grimes #include <sys/filedesc.h>
653a325decSBrooks Davis #include <sys/imgact.h>
660304c731SJamie Gritton #include <sys/jail.h>
670aa9ffcdSMitchell Horne #include <sys/kernel.h>
680384fff8SJason Evans #include <sys/ktr.h>
691005a129SJohn Baldwin #include <sys/lock.h>
702bfc50bcSEdward Tomasz Napierala #include <sys/loginclass.h>
710aa9ffcdSMitchell Horne #include <sys/malloc.h>
72ac0ad63fSBruce Evans #include <sys/mount.h>
7335e0e5b3SJohn Baldwin #include <sys/mutex.h>
74df8bae1dSRodney W. Grimes #include <sys/proc.h>
75097055e2SEdward Tomasz Napierala #include <sys/racct.h>
76df8bae1dSRodney W. Grimes #include <sys/reboot.h>
770aa9ffcdSMitchell Horne #include <sys/resourcevar.h>
78de028f5aSJeff Roberson #include <sys/sched.h>
790aa9ffcdSMitchell Horne #include <sys/signalvar.h>
801005a129SJohn Baldwin #include <sys/sx.h>
810aa9ffcdSMitchell Horne #include <sys/syscallsubr.h>
820aa9ffcdSMitchell Horne #include <sys/sysctl.h>
830aa9ffcdSMitchell Horne #include <sys/sysent.h>
84ad7507e2SSteven Wallace #include <sys/sysproto.h>
8574b2192aSJohn Dyson #include <sys/unistd.h>
860aa9ffcdSMitchell Horne #include <sys/vmmeter.h>
870aa9ffcdSMitchell Horne #include <sys/vnode.h>
88df8bae1dSRodney W. Grimes 
89df8bae1dSRodney W. Grimes #include <machine/cpu.h>
90df8bae1dSRodney W. Grimes 
913683665bSRobert Watson #include <security/audit/audit.h>
92aed55708SRobert Watson #include <security/mac/mac_framework.h>
933683665bSRobert Watson 
94df8bae1dSRodney W. Grimes #include <vm/vm.h>
95efeaf95aSDavid Greenman #include <vm/vm_param.h>
96e2068d0bSJeff Roberson #include <vm/vm_extern.h>
97efeaf95aSDavid Greenman #include <vm/pmap.h>
98efeaf95aSDavid Greenman #include <vm/vm_map.h>
996faaa756SWolfram Schneider #include <sys/copyright.h>
100df8bae1dSRodney W. Grimes 
10126ab616fSBenno Rice #include <ddb/ddb.h>
10226ab616fSBenno Rice #include <ddb/db_sym.h>
10326ab616fSBenno Rice 
10437b087a6SPeter Wemm void mi_startup(void);				/* Should be elsewhere */
105df8bae1dSRodney W. Grimes 
106df8bae1dSRodney W. Grimes /* Components of the first process -- never freed. */
107154c04e5SPoul-Henning Kamp static struct session session0;
108154c04e5SPoul-Henning Kamp static struct pgrp pgrp0;
109df8bae1dSRodney W. Grimes struct	proc proc0;
110f743ea96SMateusz Guzik struct thread0_storage thread0_st __aligned(32);
11123734995SJake Burkholder struct	vmspace vmspace0;
1122b14f991SJulian Elischer struct	proc *initproc;
113df8bae1dSRodney W. Grimes 
114983ed4f9SMatt Macy int
115983ed4f9SMatt Macy linux_alloc_current_noop(struct thread *td __unused, int flags __unused)
116983ed4f9SMatt Macy {
117983ed4f9SMatt Macy 	return (0);
118983ed4f9SMatt Macy }
119983ed4f9SMatt Macy int (*lkpi_alloc_current)(struct thread *, int) = linux_alloc_current_noop;
120983ed4f9SMatt Macy 
12156fddc5dSBrooks Davis #ifndef BOOTHOWTO
12256fddc5dSBrooks Davis #define	BOOTHOWTO	0
12356fddc5dSBrooks Davis #endif
12456fddc5dSBrooks Davis int	boothowto = BOOTHOWTO;	/* initialized so that it can be patched */
125a0c87b74SGavin Atkinson SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0,
126a0c87b74SGavin Atkinson 	"Boot control flags, passed from loader");
12756fddc5dSBrooks Davis 
12856fddc5dSBrooks Davis #ifndef BOOTVERBOSE
12956fddc5dSBrooks Davis #define	BOOTVERBOSE	0
13056fddc5dSBrooks Davis #endif
13156fddc5dSBrooks Davis int	bootverbose = BOOTVERBOSE;
132a0c87b74SGavin Atkinson SYSCTL_INT(_debug, OID_AUTO, bootverbose, CTLFLAG_RW, &bootverbose, 0,
133a0c87b74SGavin Atkinson 	"Control the output of verbose kernel messages");
134946bb7a2SPoul-Henning Kamp 
135c7962400SKyle Evans #ifdef VERBOSE_SYSINIT
136c7962400SKyle Evans /*
137c7962400SKyle Evans  * We'll use the defined value of VERBOSE_SYSINIT from the kernel config to
138c7962400SKyle Evans  * dictate the default VERBOSE_SYSINIT behavior.  Significant values for this
139c7962400SKyle Evans  * option and associated tunable are:
140c7962400SKyle Evans  * - 0, 'compiled in but silent by default'
141c7962400SKyle Evans  * - 1, 'compiled in but verbose by default' (default)
142c7962400SKyle Evans  */
143c7962400SKyle Evans int	verbose_sysinit = VERBOSE_SYSINIT;
144c7962400SKyle Evans TUNABLE_INT("debug.verbose_sysinit", &verbose_sysinit);
145c7962400SKyle Evans #endif
146c7962400SKyle Evans 
14713534277SWarner Losh #ifdef INVARIANTS
1483f272816SWarner Losh FEATURE(invariants, "Kernel compiled with INVARIANTS, may affect performance");
14913534277SWarner Losh #endif
15013534277SWarner Losh 
1512b14f991SJulian Elischer /*
1522b14f991SJulian Elischer  * This ensures that there is at least one entry so that the sysinit_set
1532b14f991SJulian Elischer  * symbol is not undefined.  A sybsystem ID of SI_SUB_DUMMY is never
1542b14f991SJulian Elischer  * executed.
15526f9a767SRodney W. Grimes  */
156237fdd78SRobert Watson SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL);
1578a129caeSDavid Greenman 
15894e9d7c1SPeter Wemm /*
15994e9d7c1SPeter Wemm  * The sysinit table itself.  Items are checked off as the are run.
16094e9d7c1SPeter Wemm  * If we want to register new sysinit types, add them to newsysinit.
16194e9d7c1SPeter Wemm  */
162f41325dbSPeter Wemm SET_DECLARE(sysinit_set, struct sysinit);
163f41325dbSPeter Wemm struct sysinit **sysinit, **sysinit_end;
164f41325dbSPeter Wemm struct sysinit **newsysinit, **newsysinit_end;
16594e9d7c1SPeter Wemm 
16694e9d7c1SPeter Wemm /*
16794e9d7c1SPeter Wemm  * Merge a new sysinit set into the current set, reallocating it if
16894e9d7c1SPeter Wemm  * necessary.  This can only be called after malloc is running.
16994e9d7c1SPeter Wemm  */
17094e9d7c1SPeter Wemm void
171f41325dbSPeter Wemm sysinit_add(struct sysinit **set, struct sysinit **set_end)
17294e9d7c1SPeter Wemm {
17394e9d7c1SPeter Wemm 	struct sysinit **newset;
17494e9d7c1SPeter Wemm 	struct sysinit **sipp;
17594e9d7c1SPeter Wemm 	struct sysinit **xipp;
176f41325dbSPeter Wemm 	int count;
17794e9d7c1SPeter Wemm 
178f41325dbSPeter Wemm 	count = set_end - set;
179ddd62546SPeter Wemm 	if (newsysinit)
180f41325dbSPeter Wemm 		count += newsysinit_end - newsysinit;
181ddd62546SPeter Wemm 	else
182f41325dbSPeter Wemm 		count += sysinit_end - sysinit;
183ac2fffa4SPedro F. Giffuni 	newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
18494e9d7c1SPeter Wemm 	if (newset == NULL)
18594e9d7c1SPeter Wemm 		panic("cannot malloc for sysinit");
18694e9d7c1SPeter Wemm 	xipp = newset;
187ddd62546SPeter Wemm 	if (newsysinit)
188f41325dbSPeter Wemm 		for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
189ddd62546SPeter Wemm 			*xipp++ = *sipp;
190ddd62546SPeter Wemm 	else
191f41325dbSPeter Wemm 		for (sipp = sysinit; sipp < sysinit_end; sipp++)
19294e9d7c1SPeter Wemm 			*xipp++ = *sipp;
193f41325dbSPeter Wemm 	for (sipp = set; sipp < set_end; sipp++)
19494e9d7c1SPeter Wemm 		*xipp++ = *sipp;
195ddd62546SPeter Wemm 	if (newsysinit)
196ddd62546SPeter Wemm 		free(newsysinit, M_TEMP);
19794e9d7c1SPeter Wemm 	newsysinit = newset;
198f41325dbSPeter Wemm 	newsysinit_end = newset + count;
19994e9d7c1SPeter Wemm }
20026f9a767SRodney W. Grimes 
201b871e661SJohn Baldwin #if defined (DDB) && defined(VERBOSE_SYSINIT)
202b871e661SJohn Baldwin static const char *
203b871e661SJohn Baldwin symbol_name(vm_offset_t va, db_strategy_t strategy)
204b871e661SJohn Baldwin {
205b871e661SJohn Baldwin 	const char *name;
206b871e661SJohn Baldwin 	c_db_sym_t sym;
207b871e661SJohn Baldwin 	db_expr_t  offset;
208b871e661SJohn Baldwin 
209b871e661SJohn Baldwin 	if (va == 0)
210b871e661SJohn Baldwin 		return (NULL);
211b871e661SJohn Baldwin 	sym = db_search_symbol(va, strategy, &offset);
212b871e661SJohn Baldwin 	if (offset != 0)
213b871e661SJohn Baldwin 		return (NULL);
214b871e661SJohn Baldwin 	db_symbol_values(sym, &name, NULL);
215b871e661SJohn Baldwin 	return (name);
216b871e661SJohn Baldwin }
217b871e661SJohn Baldwin #endif
218b871e661SJohn Baldwin 
219df8bae1dSRodney W. Grimes /*
220df8bae1dSRodney W. Grimes  * System startup; initialize the world, create process 0, mount root
221df8bae1dSRodney W. Grimes  * filesystem, and fork to create init and pagedaemon.  Most of the
222df8bae1dSRodney W. Grimes  * hard work is done in the lower-level initialization routines including
223df8bae1dSRodney W. Grimes  * startup(), which does memory initialization and autoconfiguration.
2242b14f991SJulian Elischer  *
2252b14f991SJulian Elischer  * This allows simple addition of new kernel subsystems that require
2262b14f991SJulian Elischer  * boot time initialization.  It also allows substitution of subsystem
2272b14f991SJulian Elischer  * (for instance, a scheduler, kernel profiler, or VM system) by object
228c5b193bfSPoul-Henning Kamp  * module.  Finally, it allows for optional "kernel threads".
229df8bae1dSRodney W. Grimes  */
23026f9a767SRodney W. Grimes void
23137b087a6SPeter Wemm mi_startup(void)
232df8bae1dSRodney W. Grimes {
2332b14f991SJulian Elischer 
23461f481fbSKevin Lo 	struct sysinit **sipp;	/* system initialization*/
23561f481fbSKevin Lo 	struct sysinit **xipp;	/* interior loop of sort*/
23661f481fbSKevin Lo 	struct sysinit *save;	/* bubble*/
2372b14f991SJulian Elischer 
23826ab616fSBenno Rice 	int last;
2395a8fceb3SMitchell Horne #if defined(VERBOSE_SYSINIT)
24026ab616fSBenno Rice 	int verbose;
24126ab616fSBenno Rice #endif
24226ab616fSBenno Rice 
2436032e088SColin Percival 	TSENTER();
2446032e088SColin Percival 
245b94e6f0eSJohn Baldwin 	if (boothowto & RB_VERBOSE)
246b94e6f0eSJohn Baldwin 		bootverbose++;
247b94e6f0eSJohn Baldwin 
248f41325dbSPeter Wemm 	if (sysinit == NULL) {
249f41325dbSPeter Wemm 		sysinit = SET_BEGIN(sysinit_set);
250f41325dbSPeter Wemm 		sysinit_end = SET_LIMIT(sysinit_set);
251f41325dbSPeter Wemm 	}
252f41325dbSPeter Wemm 
25394e9d7c1SPeter Wemm restart:
2542b14f991SJulian Elischer 	/*
2552b14f991SJulian Elischer 	 * Perform a bubble sort of the system initialization objects by
2562b14f991SJulian Elischer 	 * their subsystem (primary key) and order (secondary key).
2572b14f991SJulian Elischer 	 */
25840b28705SColin Percival 	TSENTER2("bubblesort");
259f41325dbSPeter Wemm 	for (sipp = sysinit; sipp < sysinit_end; sipp++) {
260f41325dbSPeter Wemm 		for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
2612b14f991SJulian Elischer 			if ((*sipp)->subsystem < (*xipp)->subsystem ||
2622b14f991SJulian Elischer 			     ((*sipp)->subsystem == (*xipp)->subsystem &&
263af4b2d2dSPeter Wemm 			      (*sipp)->order <= (*xipp)->order))
2642b14f991SJulian Elischer 				continue;	/* skip*/
2652b14f991SJulian Elischer 			save = *sipp;
2662b14f991SJulian Elischer 			*sipp = *xipp;
2672b14f991SJulian Elischer 			*xipp = save;
2682b14f991SJulian Elischer 		}
2692b14f991SJulian Elischer 	}
27040b28705SColin Percival 	TSEXIT2("bubblesort");
2712b14f991SJulian Elischer 
27226ab616fSBenno Rice 	last = SI_SUB_COPYRIGHT;
2735a8fceb3SMitchell Horne #if defined(VERBOSE_SYSINIT)
27426ab616fSBenno Rice 	verbose = 0;
27526ab616fSBenno Rice #if !defined(DDB)
27626ab616fSBenno Rice 	printf("VERBOSE_SYSINIT: DDB not enabled, symbol lookups disabled.\n");
27726ab616fSBenno Rice #endif
27826ab616fSBenno Rice #endif
27926ab616fSBenno Rice 
2802b14f991SJulian Elischer 	/*
2812b14f991SJulian Elischer 	 * Traverse the (now) ordered list of system initialization tasks.
2822b14f991SJulian Elischer 	 * Perform each task, and continue on to the next task.
2832b14f991SJulian Elischer 	 */
284f41325dbSPeter Wemm 	for (sipp = sysinit; sipp < sysinit_end; sipp++) {
2852b14f991SJulian Elischer 		if ((*sipp)->subsystem == SI_SUB_DUMMY)
2862b14f991SJulian Elischer 			continue;	/* skip dummy task(s)*/
2872b14f991SJulian Elischer 
28894e9d7c1SPeter Wemm 		if ((*sipp)->subsystem == SI_SUB_DONE)
28994e9d7c1SPeter Wemm 			continue;
29094e9d7c1SPeter Wemm 
2915a8fceb3SMitchell Horne 		if ((*sipp)->subsystem > last)
2925a8fceb3SMitchell Horne 			BOOTTRACE_INIT("sysinit 0x%7x", (*sipp)->subsystem);
2935a8fceb3SMitchell Horne 
29426ab616fSBenno Rice #if defined(VERBOSE_SYSINIT)
295c7962400SKyle Evans 		if ((*sipp)->subsystem > last && verbose_sysinit != 0) {
29626ab616fSBenno Rice 			verbose = 1;
29726ab616fSBenno Rice 			printf("subsystem %x\n", last);
29826ab616fSBenno Rice 		}
29926ab616fSBenno Rice 		if (verbose) {
30026ab616fSBenno Rice #if defined(DDB)
301b871e661SJohn Baldwin 			const char *func, *data;
30226ab616fSBenno Rice 
303b871e661SJohn Baldwin 			func = symbol_name((vm_offset_t)(*sipp)->func,
304b871e661SJohn Baldwin 			    DB_STGY_PROC);
305b871e661SJohn Baldwin 			data = symbol_name((vm_offset_t)(*sipp)->udata,
306b871e661SJohn Baldwin 			    DB_STGY_ANY);
307b871e661SJohn Baldwin 			if (func != NULL && data != NULL)
308b871e661SJohn Baldwin 				printf("   %s(&%s)... ", func, data);
309b871e661SJohn Baldwin 			else if (func != NULL)
310b871e661SJohn Baldwin 				printf("   %s(%p)... ", func, (*sipp)->udata);
31126ab616fSBenno Rice 			else
31226ab616fSBenno Rice #endif
31326ab616fSBenno Rice 				printf("   %p(%p)... ", (*sipp)->func,
31426ab616fSBenno Rice 				    (*sipp)->udata);
31526ab616fSBenno Rice 		}
31626ab616fSBenno Rice #endif
31726ab616fSBenno Rice 
3189c8b8baaSPeter Wemm 		/* Call function */
3192b14f991SJulian Elischer 		(*((*sipp)->func))((*sipp)->udata);
32094e9d7c1SPeter Wemm 
32126ab616fSBenno Rice #if defined(VERBOSE_SYSINIT)
32226ab616fSBenno Rice 		if (verbose)
32326ab616fSBenno Rice 			printf("done.\n");
32426ab616fSBenno Rice #endif
32526ab616fSBenno Rice 
32694e9d7c1SPeter Wemm 		/* Check off the one we're just done */
3275a8fceb3SMitchell Horne 		last = (*sipp)->subsystem;
32894e9d7c1SPeter Wemm 		(*sipp)->subsystem = SI_SUB_DONE;
32994e9d7c1SPeter Wemm 
33094e9d7c1SPeter Wemm 		/* Check if we've installed more sysinit items via KLD */
33194e9d7c1SPeter Wemm 		if (newsysinit != NULL) {
332f41325dbSPeter Wemm 			if (sysinit != SET_BEGIN(sysinit_set))
33394e9d7c1SPeter Wemm 				free(sysinit, M_TEMP);
33494e9d7c1SPeter Wemm 			sysinit = newsysinit;
335f41325dbSPeter Wemm 			sysinit_end = newsysinit_end;
33694e9d7c1SPeter Wemm 			newsysinit = NULL;
337f41325dbSPeter Wemm 			newsysinit_end = NULL;
33894e9d7c1SPeter Wemm 			goto restart;
33994e9d7c1SPeter Wemm 		}
3402b14f991SJulian Elischer 	}
3412b14f991SJulian Elischer 
3426032e088SColin Percival 	TSEXIT();	/* Here so we don't overlap with start_init. */
3435a8fceb3SMitchell Horne 	BOOTTRACE("mi_startup done");
3446032e088SColin Percival 
345785797c3SAndriy Gapon 	mtx_assert(&Giant, MA_OWNED | MA_NOTRECURSED);
346785797c3SAndriy Gapon 	mtx_unlock(&Giant);
347785797c3SAndriy Gapon 
348785797c3SAndriy Gapon 	/*
349785797c3SAndriy Gapon 	 * Now hand over this thread to swapper.
350785797c3SAndriy Gapon 	 */
351785797c3SAndriy Gapon 	swapper();
3522b14f991SJulian Elischer 	/* NOTREACHED*/
3532b14f991SJulian Elischer }
3542b14f991SJulian Elischer 
3559ef6c28aSBruce Evans static void
35678edb09eSAndriy Gapon print_caddr_t(void *data)
3579ef6c28aSBruce Evans {
3589ef6c28aSBruce Evans 	printf("%s", (char *)data);
3599ef6c28aSBruce Evans }
3607e936cefSAndriy Gapon 
3617e936cefSAndriy Gapon static void
3627e936cefSAndriy Gapon print_version(void *data __unused)
3637e936cefSAndriy Gapon {
3647e936cefSAndriy Gapon 	int len;
3657e936cefSAndriy Gapon 
3667e936cefSAndriy Gapon 	/* Strip a trailing newline from version. */
3677e936cefSAndriy Gapon 	len = strlen(version);
3687e936cefSAndriy Gapon 	while (len > 0 && version[len - 1] == '\n')
3697e936cefSAndriy Gapon 		len--;
3707e936cefSAndriy Gapon 	printf("%.*s %s\n", len, version, machine);
371bfdcb3bcSAndriy Gapon 	printf("%s\n", compiler_version);
3727e936cefSAndriy Gapon }
3737e936cefSAndriy Gapon 
374237fdd78SRobert Watson SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t,
375237fdd78SRobert Watson     copyright);
376237fdd78SRobert Watson SYSINIT(trademark, SI_SUB_COPYRIGHT, SI_ORDER_SECOND, print_caddr_t,
377237fdd78SRobert Watson     trademark);
3787e936cefSAndriy Gapon SYSINIT(version, SI_SUB_COPYRIGHT, SI_ORDER_THIRD, print_version, NULL);
3797a08bae6SJohn Baldwin 
3802cf6bdacSPoul-Henning Kamp #ifdef WITNESS
3812cf6bdacSPoul-Henning Kamp static char wit_warn[] =
3822cf6bdacSPoul-Henning Kamp      "WARNING: WITNESS option enabled, expect reduced performance.\n";
3835a01eca6SBrooks Davis SYSINIT(witwarn, SI_SUB_COPYRIGHT, SI_ORDER_FOURTH,
384237fdd78SRobert Watson    print_caddr_t, wit_warn);
3855a01eca6SBrooks Davis SYSINIT(witwarn2, SI_SUB_LAST, SI_ORDER_FOURTH,
386237fdd78SRobert Watson    print_caddr_t, wit_warn);
3872cf6bdacSPoul-Henning Kamp #endif
3882cf6bdacSPoul-Henning Kamp 
3892cf6bdacSPoul-Henning Kamp #ifdef DIAGNOSTIC
3902cf6bdacSPoul-Henning Kamp static char diag_warn[] =
3912cf6bdacSPoul-Henning Kamp      "WARNING: DIAGNOSTIC option enabled, expect reduced performance.\n";
3925a01eca6SBrooks Davis SYSINIT(diagwarn, SI_SUB_COPYRIGHT, SI_ORDER_FIFTH,
393237fdd78SRobert Watson     print_caddr_t, diag_warn);
3945a01eca6SBrooks Davis SYSINIT(diagwarn2, SI_SUB_LAST, SI_ORDER_FIFTH,
395237fdd78SRobert Watson     print_caddr_t, diag_warn);
3962cf6bdacSPoul-Henning Kamp #endif
3972cf6bdacSPoul-Henning Kamp 
398*41582f28SJohn Baldwin #if __SIZEOF_LONG__ == 4
399*41582f28SJohn Baldwin static char ilp32_warn[] =
400*41582f28SJohn Baldwin     "WARNING: 32-bit kernels are deprecated and may be removed in FreeBSD 15.0.\n";
401*41582f28SJohn Baldwin SYSINIT(ilp32warn, SI_SUB_COPYRIGHT, SI_ORDER_FIFTH,
402*41582f28SJohn Baldwin     print_caddr_t, ilp32_warn);
403*41582f28SJohn Baldwin SYSINIT(ilp32warn2, SI_SUB_LAST, SI_ORDER_FIFTH,
404*41582f28SJohn Baldwin     print_caddr_t, ilp32_warn);
405*41582f28SJohn Baldwin #endif
406*41582f28SJohn Baldwin 
407afe1a688SKonstantin Belousov static int
4082d88da2fSKonstantin Belousov null_fetch_syscall_args(struct thread *td __unused)
409afe1a688SKonstantin Belousov {
410afe1a688SKonstantin Belousov 
411afe1a688SKonstantin Belousov 	panic("null_fetch_syscall_args");
412afe1a688SKonstantin Belousov }
413afe1a688SKonstantin Belousov 
414afe1a688SKonstantin Belousov static void
415afe1a688SKonstantin Belousov null_set_syscall_retval(struct thread *td __unused, int error __unused)
416afe1a688SKonstantin Belousov {
417afe1a688SKonstantin Belousov 
418afe1a688SKonstantin Belousov 	panic("null_set_syscall_retval");
419afe1a688SKonstantin Belousov }
420afe1a688SKonstantin Belousov 
421de8374dfSDmitry Chagin static void
422de8374dfSDmitry Chagin null_set_fork_retval(struct thread *td __unused)
423de8374dfSDmitry Chagin {
424de8374dfSDmitry Chagin 
425de8374dfSDmitry Chagin }
426de8374dfSDmitry Chagin 
427f36ba452SJake Burkholder struct sysentvec null_sysvec = {
428a8d403e1SKonstantin Belousov 	.sv_size	= 0,
429a8d403e1SKonstantin Belousov 	.sv_table	= NULL,
430a8d403e1SKonstantin Belousov 	.sv_fixup	= NULL,
431a8d403e1SKonstantin Belousov 	.sv_sendsig	= NULL,
432a8d403e1SKonstantin Belousov 	.sv_sigcode	= NULL,
433a8d403e1SKonstantin Belousov 	.sv_szsigcode	= NULL,
434a8d403e1SKonstantin Belousov 	.sv_name	= "null",
435a8d403e1SKonstantin Belousov 	.sv_coredump	= NULL,
436a8d403e1SKonstantin Belousov 	.sv_minsigstksz	= 0,
437a8d403e1SKonstantin Belousov 	.sv_minuser	= VM_MIN_ADDRESS,
438a8d403e1SKonstantin Belousov 	.sv_maxuser	= VM_MAXUSER_ADDRESS,
439a8d403e1SKonstantin Belousov 	.sv_usrstack	= USRSTACK,
440a8d403e1SKonstantin Belousov 	.sv_psstrings	= PS_STRINGS,
4413fc21fddSMark Johnston 	.sv_psstringssz	= sizeof(struct ps_strings),
442a8d403e1SKonstantin Belousov 	.sv_stackprot	= VM_PROT_ALL,
443a8d403e1SKonstantin Belousov 	.sv_copyout_strings	= NULL,
444a8d403e1SKonstantin Belousov 	.sv_setregs	= NULL,
445a8d403e1SKonstantin Belousov 	.sv_fixlimit	= NULL,
446afe1a688SKonstantin Belousov 	.sv_maxssiz	= NULL,
447afe1a688SKonstantin Belousov 	.sv_flags	= 0,
448afe1a688SKonstantin Belousov 	.sv_set_syscall_retval = null_set_syscall_retval,
449afe1a688SKonstantin Belousov 	.sv_fetch_syscall_args = null_fetch_syscall_args,
450afe1a688SKonstantin Belousov 	.sv_syscallnames = NULL,
451e5d81ef1SDmitry Chagin 	.sv_schedtail	= NULL,
45291d1786fSDmitry Chagin 	.sv_thread_detach = NULL,
453038c7205SDmitry Chagin 	.sv_trap	= NULL,
454de8374dfSDmitry Chagin 	.sv_set_fork_retval = null_set_fork_retval,
455548a2ec4SAndrew Turner 	.sv_regset_begin = NULL,
456548a2ec4SAndrew Turner 	.sv_regset_end  = NULL,
457f36ba452SJake Burkholder };
4583ebc1248SPeter Wemm 
4592b14f991SJulian Elischer /*
4607c686853SKevin Lo  * The two following SYSINIT's are proc0 specific glue code.  I am not
4617c686853SKevin Lo  * convinced that they can not be safely combined, but their order of
4627c686853SKevin Lo  * operation has been maintained as the same as the original init_main.c
4637c686853SKevin Lo  * for right now.
4642b14f991SJulian Elischer  */
4652b14f991SJulian Elischer /* ARGSUSED*/
466154c04e5SPoul-Henning Kamp static void
46737b087a6SPeter Wemm proc0_init(void *dummy __unused)
4682b14f991SJulian Elischer {
4698ec21e3aSPoul-Henning Kamp 	struct proc *p;
470b40ce416SJulian Elischer 	struct thread *td;
471daf63fd2SMateusz Guzik 	struct ucred *newcred;
4725949c7e5SMateusz Guzik 	struct uidinfo tmpuinfo;
473c0b5261bSMateusz Guzik 	struct loginclass tmplc = {
474c0b5261bSMateusz Guzik 		.lc_name = "",
475c0b5261bSMateusz Guzik 	};
476ac45ee97SAlan Cox 	vm_paddr_t pageablemem;
477ac45ee97SAlan Cox 	int i;
478df8bae1dSRodney W. Grimes 
4790cddd8f0SMatthew Dillon 	GIANT_REQUIRED;
480df8bae1dSRodney W. Grimes 	p = &proc0;
481079b7badSJulian Elischer 	td = &thread0;
482df8bae1dSRodney W. Grimes 
483df8bae1dSRodney W. Grimes 	/*
484f231de47SKonstantin Belousov 	 * Initialize magic number and osrel.
4850384fff8SJason Evans 	 */
4860384fff8SJason Evans 	p->p_magic = P_MAGIC;
487f231de47SKonstantin Belousov 	p->p_osrel = osreldate;
4880384fff8SJason Evans 
4898460a577SJohn Birrell 	/*
4908460a577SJohn Birrell 	 * Initialize thread and process structures.
4918460a577SJohn Birrell 	 */
4928460a577SJohn Birrell 	procinit();	/* set up proc zone */
4938460a577SJohn Birrell 	threadinit();	/* set up UMA zones */
4948460a577SJohn Birrell 
4958460a577SJohn Birrell 	/*
4968460a577SJohn Birrell 	 * Initialise scheduler resources.
4978460a577SJohn Birrell 	 * Add scheduler specific parts to proc, thread as needed.
4988460a577SJohn Birrell 	 */
499ed062c8dSJulian Elischer 	schedinit();	/* scheduler gets its house in order */
500b1508c72SDavid Greenman 
501b1508c72SDavid Greenman 	/*
502df8bae1dSRodney W. Grimes 	 * Create process 0 (the swapper).
503df8bae1dSRodney W. Grimes 	 */
504a3bfb996SJeffrey Hsu 	LIST_INSERT_HEAD(&allproc, p, p_list);
5057b531e60SJohn Baldwin 	LIST_INSERT_HEAD(PIDHASH(0), p, p_hash);
5066008862bSJohn Baldwin 	mtx_init(&pgrp0.pg_mtx, "process group", NULL, MTX_DEF | MTX_DUPOK);
5073360b485SKonstantin Belousov 	sx_init(&pgrp0.pg_killsx, "killpg racer");
508df8bae1dSRodney W. Grimes 	p->p_pgrp = &pgrp0;
509a3bfb996SJeffrey Hsu 	LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
510a3bfb996SJeffrey Hsu 	LIST_INIT(&pgrp0.pg_members);
511a3bfb996SJeffrey Hsu 	LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
512a3bfb996SJeffrey Hsu 
513df8bae1dSRodney W. Grimes 	pgrp0.pg_session = &session0;
5146008862bSJohn Baldwin 	mtx_init(&session0.s_mtx, "session", NULL, MTX_DEF);
515bc093719SEd Schouten 	refcount_init(&session0.s_count, 1);
516df8bae1dSRodney W. Grimes 	session0.s_leader = p;
517df8bae1dSRodney W. Grimes 
5183ebc1248SPeter Wemm 	p->p_sysent = &null_sysvec;
519db57c70aSKonstantin Belousov 	p->p_flag = P_SYSTEM | P_INMEM | P_KPROC;
52055648840SJohn Baldwin 	p->p_flag2 = 0;
521e602ba25SJulian Elischer 	p->p_state = PRS_NORMAL;
5229e590ff0SKonstantin Belousov 	p->p_klist = knlist_alloc(&p->p_mtx);
5232c255e9dSRobert Watson 	STAILQ_INIT(&p->p_ktr);
524fa885116SJulian Elischer 	p->p_nice = NZERO;
525cf31cadeSMateusz Guzik 	td->td_tid = THREAD0_TID;
526aae3547bSMateusz Guzik 	tidhash_add(td);
527fa2528acSAlex Richardson 	TD_SET_STATE(td, TDS_RUNNING);
5288460a577SJohn Birrell 	td->td_pri_class = PRI_TIMESHARE;
5298460a577SJohn Birrell 	td->td_user_pri = PUSER;
53060d48235SDavid Xu 	td->td_base_user_pri = PUSER;
531acbe332aSDavid Xu 	td->td_lend_user_pri = PRI_MAX;
5322c100766SJulian Elischer 	td->td_priority = PVM;
533fd058078SJohn Baldwin 	td->td_base_pri = PVM;
5348c636a11SNathan Whitehorn 	td->td_oncpu = curcpu;
5359b9a0179SPawel Jakub Dawidek 	td->td_flags = TDF_INMEM;
5369b9a0179SPawel Jakub Dawidek 	td->td_pflags = TDP_KTHREAD;
537d7f687fcSJeff Roberson 	td->td_cpuset = cpuset_thread0();
5383f289c3fSJeff Roberson 	td->td_domain.dr_policy = td->td_cpuset->cs_domain;
539b96bd95bSIan Lepore 	prison0_init();
5402c1011f7SJohn Dyson 	p->p_peers = 0;
5412c1011f7SJohn Dyson 	p->p_leader = p;
542237623b0SKonstantin Belousov 	p->p_reaper = p;
5430fb9f29bSBjoern A. Zeeb 	p->p_treeflag |= P_TREE_REAPER;
544237623b0SKonstantin Belousov 	LIST_INIT(&p->p_reaplist);
5452c1011f7SJohn Dyson 
5467ab24ea3SJulian Elischer 	strncpy(p->p_comm, "kernel", sizeof (p->p_comm));
5477ab24ea3SJulian Elischer 	strncpy(td->td_name, "swapper", sizeof (td->td_name));
548df8bae1dSRodney W. Grimes 
549f7e50ea7SKonstantin Belousov 	callout_init_mtx(&p->p_itcallout, &p->p_mtx, 0);
5501c4bcd05SJeff Roberson 	callout_init_mtx(&p->p_limco, &p->p_mtx, 0);
551fd90e2edSJung-uk Kim 	callout_init(&td->td_slpcallout, 1);
5522fd1ffefSKonstantin Belousov 	TAILQ_INIT(&p->p_kqtim_stop);
5534f559836SJake Burkholder 
554df8bae1dSRodney W. Grimes 	/* Create credentials. */
555daf63fd2SMateusz Guzik 	newcred = crget();
556daf63fd2SMateusz Guzik 	newcred->cr_ngroups = 1;	/* group 0 */
5575949c7e5SMateusz Guzik 	/* A hack to prevent uifind from tripping over NULL pointers. */
5585949c7e5SMateusz Guzik 	curthread->td_ucred = newcred;
5595949c7e5SMateusz Guzik 	tmpuinfo.ui_uid = 1;
5605949c7e5SMateusz Guzik 	newcred->cr_uidinfo = newcred->cr_ruidinfo = &tmpuinfo;
561daf63fd2SMateusz Guzik 	newcred->cr_uidinfo = uifind(0);
562daf63fd2SMateusz Guzik 	newcred->cr_ruidinfo = uifind(0);
563c0b5261bSMateusz Guzik 	newcred->cr_loginclass = &tmplc;
564c0b5261bSMateusz Guzik 	newcred->cr_loginclass = loginclass_find("default");
5655949c7e5SMateusz Guzik 	/* End hack. creds get properly set later with thread_cow_get_proc */
5665949c7e5SMateusz Guzik 	curthread->td_ucred = NULL;
567daf63fd2SMateusz Guzik 	newcred->cr_prison = &prison0;
5681724c563SMateusz Guzik 	newcred->cr_users++; /* avoid assertion failure */
569ffb34484SMateusz Guzik 	proc_set_cred_init(p, newcred);
5701724c563SMateusz Guzik 	newcred->cr_users--;
5711724c563SMateusz Guzik 	crfree(newcred);
5723683665bSRobert Watson #ifdef AUDIT
573daf63fd2SMateusz Guzik 	audit_cred_kproc0(newcred);
5743683665bSRobert Watson #endif
57540244964SRobert Watson #ifdef MAC
576daf63fd2SMateusz Guzik 	mac_cred_create_swapper(newcred);
57740244964SRobert Watson #endif
57890af4afaSJohn Baldwin 	/* Create sigacts. */
57990af4afaSJohn Baldwin 	p->p_sigacts = sigacts_alloc();
5806626c604SJulian Elischer 
58137b087a6SPeter Wemm 	/* Initialize signal state for process 0. */
58237b087a6SPeter Wemm 	siginit(&proc0);
58337b087a6SPeter Wemm 
584df8bae1dSRodney W. Grimes 	/* Create the file descriptor table. */
58585078b85SConrad Meyer 	p->p_pd = pdinit(NULL, false);
586893d20c9SMateusz Guzik 	p->p_fd = fdinit();
587ad05d580STor Egge 	p->p_fdtol = NULL;
588df8bae1dSRodney W. Grimes 
589df8bae1dSRodney W. Grimes 	/* Create the limits structures. */
59091d5354aSJohn Baldwin 	p->p_limit = lim_alloc();
59191d5354aSJohn Baldwin 	for (i = 0; i < RLIM_NLIMITS; i++)
59291d5354aSJohn Baldwin 		p->p_limit->pl_rlimit[i].rlim_cur =
59391d5354aSJohn Baldwin 		    p->p_limit->pl_rlimit[i].rlim_max = RLIM_INFINITY;
59491d5354aSJohn Baldwin 	p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_cur =
59591d5354aSJohn Baldwin 	    p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
59691d5354aSJohn Baldwin 	p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_cur =
59791d5354aSJohn Baldwin 	    p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
598ac45ee97SAlan Cox 	p->p_limit->pl_rlimit[RLIMIT_DATA].rlim_cur = dfldsiz;
599ac45ee97SAlan Cox 	p->p_limit->pl_rlimit[RLIMIT_DATA].rlim_max = maxdsiz;
600ac45ee97SAlan Cox 	p->p_limit->pl_rlimit[RLIMIT_STACK].rlim_cur = dflssiz;
601ac45ee97SAlan Cox 	p->p_limit->pl_rlimit[RLIMIT_STACK].rlim_max = maxssiz;
602ac45ee97SAlan Cox 	/* Cast to avoid overflow on i386/PAE. */
603e2068d0bSJeff Roberson 	pageablemem = ptoa((vm_paddr_t)vm_free_count());
604ac45ee97SAlan Cox 	p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_cur =
605ac45ee97SAlan Cox 	    p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_max = pageablemem;
606ac45ee97SAlan Cox 	p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = pageablemem / 3;
607ac45ee97SAlan Cox 	p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_max = pageablemem;
6085715307fSJohn Baldwin 	p->p_cpulimit = RLIM_INFINITY;
609df8bae1dSRodney W. Grimes 
6104ea6a9a2SMateusz Guzik 	PROC_LOCK(p);
6114ea6a9a2SMateusz Guzik 	thread_cow_get_proc(td, p);
6124ea6a9a2SMateusz Guzik 	PROC_UNLOCK(p);
6134ea6a9a2SMateusz Guzik 
614097055e2SEdward Tomasz Napierala 	/* Initialize resource accounting structures. */
615097055e2SEdward Tomasz Napierala 	racct_create(&p->p_racct);
616097055e2SEdward Tomasz Napierala 
6178b059651SDavid Schultz 	p->p_stats = pstats_alloc();
6188b059651SDavid Schultz 
619df8bae1dSRodney W. Grimes 	/* Allocate a prototype map so we have something to fork. */
620df8bae1dSRodney W. Grimes 	p->p_vmspace = &vmspace0;
621f7db0c95SMark Johnston 	refcount_init(&vmspace0.vm_refcnt, 1);
622100ac78bSKonstantin Belousov 	pmap_pinit0(vmspace_pmap(&vmspace0));
623878adb85SBjoern A. Zeeb 
624878adb85SBjoern A. Zeeb 	/*
625878adb85SBjoern A. Zeeb 	 * proc0 is not expected to enter usermode, so there is no special
626878adb85SBjoern A. Zeeb 	 * handling for sv_minuser here, like is done for exec_new_vmspace().
627878adb85SBjoern A. Zeeb 	 */
62892351f16SAlan Cox 	vm_map_init(&vmspace0.vm_map, vmspace_pmap(&vmspace0),
62992351f16SAlan Cox 	    p->p_sysent->sv_minuser, p->p_sysent->sv_maxuser);
6307c7454feSRandall Stewart 
631175389cfSEdward Tomasz Napierala 	/*
632175389cfSEdward Tomasz Napierala 	 * Call the init and ctor for the new thread and proc.  We wait
633175389cfSEdward Tomasz Napierala 	 * to do this until all other structures are fairly sane.
6347c7454feSRandall Stewart 	 */
6352ca45184SMatt Joras 	EVENTHANDLER_DIRECT_INVOKE(process_init, p);
6362ca45184SMatt Joras 	EVENTHANDLER_DIRECT_INVOKE(thread_init, td);
637598f2b81SMateusz Guzik #ifdef KDTRACE_HOOKS
638598f2b81SMateusz Guzik 	kdtrace_proc_ctor(p);
639598f2b81SMateusz Guzik 	kdtrace_thread_ctor(td);
640598f2b81SMateusz Guzik #endif
6412ca45184SMatt Joras 	EVENTHANDLER_DIRECT_INVOKE(process_ctor, p);
6422ca45184SMatt Joras 	EVENTHANDLER_DIRECT_INVOKE(thread_ctor, td);
643df8bae1dSRodney W. Grimes 
64439fb8e6bSJulian Elischer 	/*
645a3bfb996SJeffrey Hsu 	 * Charge root for one process.
646df8bae1dSRodney W. Grimes 	 */
647b1fc0ec1SRobert Watson 	(void)chgproccnt(p->p_ucred->cr_ruidinfo, 1, 0);
64858c77a9dSEdward Tomasz Napierala 	PROC_LOCK(p);
64958c77a9dSEdward Tomasz Napierala 	racct_add_force(p, RACCT_NPROC, 1);
65058c77a9dSEdward Tomasz Napierala 	PROC_UNLOCK(p);
65126f9a767SRodney W. Grimes }
652237fdd78SRobert Watson SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL);
6532b14f991SJulian Elischer 
6542b14f991SJulian Elischer /* ARGSUSED*/
655154c04e5SPoul-Henning Kamp static void
65637b087a6SPeter Wemm proc0_post(void *dummy __unused)
6572b14f991SJulian Elischer {
65837b087a6SPeter Wemm 	struct proc *p;
659a140976eSAttilio Rao 	struct rusage ru;
660d92909c1SRobert Watson 	struct thread *td;
661a6fc8288SPeter Wemm 
6622b14f991SJulian Elischer 	/*
6631b0b259eSBruce Evans 	 * Now we can look at the time, having had a chance to verify the
6641b0b259eSBruce Evans 	 * time from the filesystem.  Pretend that proc0 started now.
6652b14f991SJulian Elischer 	 */
6661005a129SJohn Baldwin 	sx_slock(&allproc_lock);
6674f506694SXin LI 	FOREACH_PROC_IN_SYSTEM(p) {
66803bd1b69SBryan Drewery 		PROC_LOCK(p);
66903bd1b69SBryan Drewery 		if (p->p_state == PRS_NEW) {
67003bd1b69SBryan Drewery 			PROC_UNLOCK(p);
67103bd1b69SBryan Drewery 			continue;
67203bd1b69SBryan Drewery 		}
67387ccef7bSDag-Erling Smørgrav 		microuptime(&p->p_stats->p_start);
6745c7bebf9SKonstantin Belousov 		PROC_STATLOCK(p);
675a140976eSAttilio Rao 		rufetch(p, &ru);	/* Clears thread stats */
6765b1a8eb3SPoul-Henning Kamp 		p->p_rux.rux_runtime = 0;
677a140976eSAttilio Rao 		p->p_rux.rux_uticks = 0;
678a140976eSAttilio Rao 		p->p_rux.rux_sticks = 0;
679a140976eSAttilio Rao 		p->p_rux.rux_iticks = 0;
68003bd1b69SBryan Drewery 		PROC_STATUNLOCK(p);
681d92909c1SRobert Watson 		FOREACH_THREAD_IN_PROC(p, td) {
682d92909c1SRobert Watson 			td->td_runtime = 0;
683d92909c1SRobert Watson 		}
68403bd1b69SBryan Drewery 		PROC_UNLOCK(p);
68537b087a6SPeter Wemm 	}
6861005a129SJohn Baldwin 	sx_sunlock(&allproc_lock);
6875b1a8eb3SPoul-Henning Kamp 	PCPU_SET(switchtime, cpu_ticks());
6880384fff8SJason Evans 	PCPU_SET(switchticks, ticks);
6892b14f991SJulian Elischer }
690237fdd78SRobert Watson SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL);
6912b14f991SJulian Elischer 
692df8bae1dSRodney W. Grimes /*
6932b14f991SJulian Elischer  ***************************************************************************
6942b14f991SJulian Elischer  ****
6952b14f991SJulian Elischer  **** The following SYSINIT's and glue code should be moved to the
6962b14f991SJulian Elischer  **** respective files on a per subsystem basis.
6972b14f991SJulian Elischer  ****
6982b14f991SJulian Elischer  ***************************************************************************
699df8bae1dSRodney W. Grimes  */
700df8bae1dSRodney W. Grimes 
701df8bae1dSRodney W. Grimes /*
702df8bae1dSRodney W. Grimes  * List of paths to try when searching for "init".
703df8bae1dSRodney W. Grimes  */
7045f967b24SDag-Erling Smørgrav static char init_path[MAXPATHLEN] =
70567481196SJohn Birrell #ifdef	INIT_PATH
70667481196SJohn Birrell     __XSTRING(INIT_PATH);
70767481196SJohn Birrell #else
7083bedc940SSergey Kandaurov     "/sbin/init:/sbin/oinit:/sbin/init.bak:/rescue/init";
70967481196SJohn Birrell #endif
710af1408e3SLuigi Rizzo SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0,
711af1408e3SLuigi Rizzo 	"Path used to search the init process");
712df8bae1dSRodney W. Grimes 
713df8bae1dSRodney W. Grimes /*
714724447acSRalf S. Engelschall  * Shutdown timeout of init(8).
715724447acSRalf S. Engelschall  * Unused within kernel, but used to control init(8), hence do not remove.
716724447acSRalf S. Engelschall  */
717724447acSRalf S. Engelschall #ifndef INIT_SHUTDOWN_TIMEOUT
718724447acSRalf S. Engelschall #define INIT_SHUTDOWN_TIMEOUT 120
719724447acSRalf S. Engelschall #endif
720724447acSRalf S. Engelschall static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT;
721724447acSRalf S. Engelschall SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout,
722a0c87b74SGavin Atkinson 	CTLFLAG_RW, &init_shutdown_timeout, 0, "Shutdown timeout of init(8). "
723a0c87b74SGavin Atkinson 	"Unused within kernel, but used to control init(8)");
724724447acSRalf S. Engelschall 
725724447acSRalf S. Engelschall /*
7265f967b24SDag-Erling Smørgrav  * Start the initial user process; try exec'ing each pathname in init_path.
727df8bae1dSRodney W. Grimes  * The program is invoked with one argument containing the boot flags.
728df8bae1dSRodney W. Grimes  */
729df8bae1dSRodney W. Grimes static void
73037b087a6SPeter Wemm start_init(void *dummy)
731df8bae1dSRodney W. Grimes {
7323a325decSBrooks Davis 	struct image_args args;
733827c3852SBrooks Davis 	int error;
734dedc82aeSBrooks Davis 	char *var, *path;
735dedc82aeSBrooks Davis 	char *free_init_path, *tmp_init_path;
736b40ce416SJulian Elischer 	struct thread *td;
7379c8b8baaSPeter Wemm 	struct proc *p;
7383a325decSBrooks Davis 	struct vmspace *oldvmspace;
739df8bae1dSRodney W. Grimes 
7406032e088SColin Percival 	TSENTER();	/* Here so we don't overlap with mi_startup. */
7416032e088SColin Percival 
742b40ce416SJulian Elischer 	td = curthread;
743b40ce416SJulian Elischer 	p = td->td_proc;
744df8bae1dSRodney W. Grimes 
745563af2ecSMaxime Henrion 	vfs_mountroot();
746fb92273bSPoul-Henning Kamp 
74743d4f8c4SDevin Teske 	/* Wipe GELI passphrase from the environment. */
74843d4f8c4SDevin Teske 	kern_unsetenv("kern.geom.eli.passphrase");
74943d4f8c4SDevin Teske 
750ab3f5b6eSWarner Losh 	/* For Multicons, report which console is primary to both */
751ab3f5b6eSWarner Losh 	if (boothowto & RB_MULTIPLE) {
752ab3f5b6eSWarner Losh 		if (boothowto & RB_SERIAL)
753ab3f5b6eSWarner Losh 			printf("Dual Console: Serial Primary, Video Secondary\n");
754ab3f5b6eSWarner Losh 		else
755ab3f5b6eSWarner Losh 			printf("Dual Console: Video Primary, Serial Secondary\n");
756ab3f5b6eSWarner Losh 	}
757ab3f5b6eSWarner Losh 
7582be111bfSDavide Italiano 	if ((var = kern_getenv("init_path")) != NULL) {
759e80fb434SRobert Drehmel 		strlcpy(init_path, var, sizeof(init_path));
760d786139cSMaxime Henrion 		freeenv(var);
7615f967b24SDag-Erling Smørgrav 	}
762dedc82aeSBrooks Davis 	free_init_path = tmp_init_path = strdup(init_path, M_TEMP);
7635f967b24SDag-Erling Smørgrav 
764dedc82aeSBrooks Davis 	while ((path = strsep(&tmp_init_path, ":")) != NULL) {
7655f967b24SDag-Erling Smørgrav 		if (bootverbose)
766dedc82aeSBrooks Davis 			printf("start_init: trying %s\n", path);
7675f967b24SDag-Erling Smørgrav 
7683a325decSBrooks Davis 		memset(&args, 0, sizeof(args));
7693a325decSBrooks Davis 		error = exec_alloc_args(&args);
7703a325decSBrooks Davis 		if (error != 0)
7713a325decSBrooks Davis 			panic("%s: Can't allocate space for init arguments %d",
7723a325decSBrooks Davis 			    __func__, error);
7733a325decSBrooks Davis 
7743a325decSBrooks Davis 		error = exec_args_add_fname(&args, path, UIO_SYSSPACE);
7753a325decSBrooks Davis 		if (error != 0)
7763a325decSBrooks Davis 			panic("%s: Can't add fname %d", __func__, error);
7773a325decSBrooks Davis 		error = exec_args_add_arg(&args, path, UIO_SYSSPACE);
7783a325decSBrooks Davis 		if (error != 0)
7793a325decSBrooks Davis 			panic("%s: Can't add argv[0] %d", __func__, error);
780827c3852SBrooks Davis 		if (boothowto & RB_SINGLE)
781827c3852SBrooks Davis 			error = exec_args_add_arg(&args, "-s", UIO_SYSSPACE);
7823a325decSBrooks Davis 		if (error != 0)
7833a325decSBrooks Davis 			panic("%s: Can't add argv[0] %d", __func__, error);
784df8bae1dSRodney W. Grimes 
785df8bae1dSRodney W. Grimes 		/*
786df8bae1dSRodney W. Grimes 		 * Now try to exec the program.  If can't for any reason
787df8bae1dSRodney W. Grimes 		 * other than it doesn't exist, complain.
7882b14f991SJulian Elischer 		 *
78937b087a6SPeter Wemm 		 * Otherwise, return via fork_trampoline() all the way
7909c8b8baaSPeter Wemm 		 * to user mode as init!
791df8bae1dSRodney W. Grimes 		 */
7923a325decSBrooks Davis 		KASSERT((td->td_pflags & TDP_EXECVMSPC) == 0,
7933a325decSBrooks Davis 		    ("nested execve"));
79486608131SKonstantin Belousov 		oldvmspace = p->p_vmspace;
795aaf78c16SKonstantin Belousov 		error = kern_execve(td, &args, NULL, oldvmspace);
7963a325decSBrooks Davis 		KASSERT(error != 0,
7973a325decSBrooks Davis 		    ("kern_execve returned success, not EJUSTRETURN"));
7983a325decSBrooks Davis 		if (error == EJUSTRETURN) {
799aaf78c16SKonstantin Belousov 			exec_cleanup(td, oldvmspace);
800dedc82aeSBrooks Davis 			free(free_init_path, M_TEMP);
8016032e088SColin Percival 			TSEXIT();
802df8bae1dSRodney W. Grimes 			return;
803db72809dSJohn Baldwin 		}
804df8bae1dSRodney W. Grimes 		if (error != ENOENT)
805dedc82aeSBrooks Davis 			printf("exec %s: error %d\n", path, error);
806df8bae1dSRodney W. Grimes 	}
807dedc82aeSBrooks Davis 	free(free_init_path, M_TEMP);
808580e7e5aSGreg Lehey 	printf("init: not found in path %s\n", init_path);
809df8bae1dSRodney W. Grimes 	panic("no init");
810df8bae1dSRodney W. Grimes }
81137b087a6SPeter Wemm 
81237b087a6SPeter Wemm /*
813b706be23SKonstantin Belousov  * Like kproc_create(), but runs in its own address space.  We do this
814b706be23SKonstantin Belousov  * early to reserve pid 1.  Note special case - do not make it
815b706be23SKonstantin Belousov  * runnable yet, init execution is started when userspace can be served.
81637b087a6SPeter Wemm  */
81737b087a6SPeter Wemm static void
81837b087a6SPeter Wemm create_init(const void *udata __unused)
81937b087a6SPeter Wemm {
82033fd9b9aSMateusz Guzik 	struct fork_req fr;
8215a06cb0cSRobert Watson 	struct ucred *newcred, *oldcred;
822cd672ca6SMateusz Guzik 	struct thread *td;
82337b087a6SPeter Wemm 	int error;
82437b087a6SPeter Wemm 
82533fd9b9aSMateusz Guzik 	bzero(&fr, sizeof(fr));
82633fd9b9aSMateusz Guzik 	fr.fr_flags = RFFDG | RFPROC | RFSTOPPED;
82733fd9b9aSMateusz Guzik 	fr.fr_procp = &initproc;
82833fd9b9aSMateusz Guzik 	error = fork1(&thread0, &fr);
82937b087a6SPeter Wemm 	if (error)
83037b087a6SPeter Wemm 		panic("cannot fork init: %d\n", error);
831bafc8f25SRobert Watson 	KASSERT(initproc->p_pid == 1, ("create_init: initproc->p_pid != 1"));
8325a06cb0cSRobert Watson 	/* divorce init's credentials from the kernel's */
8335a06cb0cSRobert Watson 	newcred = crget();
834237623b0SKonstantin Belousov 	sx_xlock(&proctree_lock);
835f202965eSJohn Baldwin 	PROC_LOCK(initproc);
836b61ce5b0SJeff Roberson 	initproc->p_flag |= P_SYSTEM | P_INMEM;
837237623b0SKonstantin Belousov 	initproc->p_treeflag |= P_TREE_REAPER;
8385a06cb0cSRobert Watson 	oldcred = initproc->p_ucred;
8395a06cb0cSRobert Watson 	crcopy(newcred, oldcred);
84040244964SRobert Watson #ifdef MAC
841212ab0cfSRobert Watson 	mac_cred_create_init(newcred);
84240244964SRobert Watson #endif
8433683665bSRobert Watson #ifdef AUDIT
844faef5371SRobert Watson 	audit_cred_proc1(newcred);
8453683665bSRobert Watson #endif
846daf63fd2SMateusz Guzik 	proc_set_cred(initproc, newcred);
847cd672ca6SMateusz Guzik 	td = FIRST_THREAD_IN_PROC(initproc);
8481724c563SMateusz Guzik 	crcowfree(td);
8491724c563SMateusz Guzik 	td->td_realucred = crcowget(initproc->p_ucred);
8501724c563SMateusz Guzik 	td->td_ucred = td->td_realucred;
851f202965eSJohn Baldwin 	PROC_UNLOCK(initproc);
852237623b0SKonstantin Belousov 	sx_xunlock(&proctree_lock);
8535a06cb0cSRobert Watson 	crfree(oldcred);
8545c2cf818SKonstantin Belousov 	cpu_fork_kthread_handler(FIRST_THREAD_IN_PROC(initproc),
8555c2cf818SKonstantin Belousov 	    start_init, NULL);
85637b087a6SPeter Wemm }
857237fdd78SRobert Watson SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL);
85837b087a6SPeter Wemm 
85937b087a6SPeter Wemm /*
86037b087a6SPeter Wemm  * Make it runnable now.
86137b087a6SPeter Wemm  */
86237b087a6SPeter Wemm static void
86337b087a6SPeter Wemm kick_init(const void *udata __unused)
86437b087a6SPeter Wemm {
865079b7badSJulian Elischer 	struct thread *td;
866f202965eSJohn Baldwin 
867079b7badSJulian Elischer 	td = FIRST_THREAD_IN_PROC(initproc);
868982d11f8SJeff Roberson 	thread_lock(td);
86971fad9fdSJulian Elischer 	TD_SET_CAN_RUN(td);
870f0393f06SJeff Roberson 	sched_add(td, SRQ_BORING);
87137b087a6SPeter Wemm }
87210cb2424SMark Murray SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_MIDDLE, kick_init, NULL);
873f7a0bb0dSBjoern A. Zeeb 
874f7a0bb0dSBjoern A. Zeeb /*
875f7a0bb0dSBjoern A. Zeeb  * DDB(4).
876f7a0bb0dSBjoern A. Zeeb  */
877f7a0bb0dSBjoern A. Zeeb #ifdef DDB
878f7a0bb0dSBjoern A. Zeeb static void
879f7a0bb0dSBjoern A. Zeeb db_show_print_syinit(struct sysinit *sip, bool ddb)
880f7a0bb0dSBjoern A. Zeeb {
881f7a0bb0dSBjoern A. Zeeb 	const char *sname, *funcname;
882f7a0bb0dSBjoern A. Zeeb 	c_db_sym_t sym;
883f7a0bb0dSBjoern A. Zeeb 	db_expr_t  offset;
884f7a0bb0dSBjoern A. Zeeb 
885f7a0bb0dSBjoern A. Zeeb #define xprint(...)							\
886f7a0bb0dSBjoern A. Zeeb 	if (ddb)							\
887f7a0bb0dSBjoern A. Zeeb 		db_printf(__VA_ARGS__);					\
888f7a0bb0dSBjoern A. Zeeb 	else								\
889f7a0bb0dSBjoern A. Zeeb 		printf(__VA_ARGS__)
890f7a0bb0dSBjoern A. Zeeb 
891f7a0bb0dSBjoern A. Zeeb 	if (sip == NULL) {
892f7a0bb0dSBjoern A. Zeeb 		xprint("%s: no sysinit * given\n", __func__);
893f7a0bb0dSBjoern A. Zeeb 		return;
894f7a0bb0dSBjoern A. Zeeb 	}
895f7a0bb0dSBjoern A. Zeeb 
896f7a0bb0dSBjoern A. Zeeb 	sym = db_search_symbol((vm_offset_t)sip, DB_STGY_ANY, &offset);
897f7a0bb0dSBjoern A. Zeeb 	db_symbol_values(sym, &sname, NULL);
898f7a0bb0dSBjoern A. Zeeb 	sym = db_search_symbol((vm_offset_t)sip->func, DB_STGY_PROC, &offset);
899f7a0bb0dSBjoern A. Zeeb 	db_symbol_values(sym, &funcname, NULL);
900f7a0bb0dSBjoern A. Zeeb 	xprint("%s(%p)\n", (sname != NULL) ? sname : "", sip);
901f7a0bb0dSBjoern A. Zeeb 	xprint("  %#08x %#08x\n", sip->subsystem, sip->order);
902f7a0bb0dSBjoern A. Zeeb 	xprint("  %p(%s)(%p)\n",
903f7a0bb0dSBjoern A. Zeeb 	    sip->func, (funcname != NULL) ? funcname : "", sip->udata);
904f7a0bb0dSBjoern A. Zeeb #undef xprint
905f7a0bb0dSBjoern A. Zeeb }
906f7a0bb0dSBjoern A. Zeeb 
907c84c5e00SMitchell Horne DB_SHOW_COMMAND_FLAGS(sysinit, db_show_sysinit, DB_CMD_MEMSAFE)
908f7a0bb0dSBjoern A. Zeeb {
909f7a0bb0dSBjoern A. Zeeb 	struct sysinit **sipp;
910f7a0bb0dSBjoern A. Zeeb 
911f7a0bb0dSBjoern A. Zeeb 	db_printf("SYSINIT vs Name(Ptr)\n");
912f7a0bb0dSBjoern A. Zeeb 	db_printf("  Subsystem  Order\n");
913f7a0bb0dSBjoern A. Zeeb 	db_printf("  Function(Name)(Arg)\n");
914f7a0bb0dSBjoern A. Zeeb 	for (sipp = sysinit; sipp < sysinit_end; sipp++) {
915f7a0bb0dSBjoern A. Zeeb 		db_show_print_syinit(*sipp, true);
916f7a0bb0dSBjoern A. Zeeb 		if (db_pager_quit)
917f7a0bb0dSBjoern A. Zeeb 			break;
918f7a0bb0dSBjoern A. Zeeb 	}
919f7a0bb0dSBjoern A. Zeeb }
920f7a0bb0dSBjoern A. Zeeb #endif /* DDB */
921