1/*-
2 * Copyright (c) 1995 Terrence R. Lambert
3 * All rights reserved.
4 *
5 * Copyright (c) 1982, 1986, 1989, 1991, 1992, 1993
6 *	The Regents of the University of California.  All rights reserved.
7 * (c) UNIX System Laboratories, Inc.
8 * All or some portions of this file are derived from material licensed
9 * to the University of California by American Telephone and Telegraph
10 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
11 * the permission of UNIX System Laboratories, Inc.
12 *
13 * Redistribution and use in source and binary forms, with or without
14 * modification, are permitted provided that the following conditions
15 * are met:
16 * 1. Redistributions of source code must retain the above copyright
17 *    notice, this list of conditions and the following disclaimer.
18 * 2. Redistributions in binary form must reproduce the above copyright
19 *    notice, this list of conditions and the following disclaimer in the
20 *    documentation and/or other materials provided with the distribution.
21 * 3. All advertising materials mentioning features or use of this software
22 *    must display the following acknowledgement:
23 *	This product includes software developed by the University of
24 *	California, Berkeley and its contributors.
25 * 4. Neither the name of the University nor the names of its contributors
26 *    may be used to endorse or promote products derived from this software
27 *    without specific prior written permission.
28 *
29 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
30 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
31 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
32 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
33 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
34 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
35 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
36 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
37 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
38 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
39 * SUCH DAMAGE.
40 *
41 *	@(#)init_main.c	8.9 (Berkeley) 1/21/94
42 */
43
44#include <sys/cdefs.h>
45__FBSDID("$FreeBSD$");
46
47#include "opt_ddb.h"
48#include "opt_init_path.h"
49
50#include <sys/param.h>
51#include <sys/kernel.h>
52#include <sys/exec.h>
53#include <sys/file.h>
54#include <sys/filedesc.h>
55#include <sys/jail.h>
56#include <sys/ktr.h>
57#include <sys/lock.h>
58#include <sys/loginclass.h>
59#include <sys/mount.h>
60#include <sys/mutex.h>
61#include <sys/syscallsubr.h>
62#include <sys/sysctl.h>
63#include <sys/proc.h>
64#include <sys/racct.h>
65#include <sys/resourcevar.h>
66#include <sys/systm.h>
67#include <sys/signalvar.h>
68#include <sys/vnode.h>
69#include <sys/sysent.h>
70#include <sys/reboot.h>
71#include <sys/sched.h>
72#include <sys/sx.h>
73#include <sys/sysproto.h>
74#include <sys/vmmeter.h>
75#include <sys/unistd.h>
76#include <sys/malloc.h>
77#include <sys/conf.h>
78#include <sys/cpuset.h>
79
80#include <machine/cpu.h>
81
82#include <security/audit/audit.h>
83#include <security/mac/mac_framework.h>
84
85#include <vm/vm.h>
86#include <vm/vm_param.h>
87#include <vm/pmap.h>
88#include <vm/vm_map.h>
89#include <sys/copyright.h>
90
91#include <ddb/ddb.h>
92#include <ddb/db_sym.h>
93
94void mi_startup(void);				/* Should be elsewhere */
95
96/* Components of the first process -- never freed. */
97static struct session session0;
98static struct pgrp pgrp0;
99struct	proc proc0;
100struct	thread thread0 __aligned(16);
101struct	vmspace vmspace0;
102struct	proc *initproc;
103
104int	boothowto = 0;		/* initialized so that it can be patched */
105SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0,
106	"Boot control flags, passed from loader");
107int	bootverbose;
108SYSCTL_INT(_debug, OID_AUTO, bootverbose, CTLFLAG_RW, &bootverbose, 0,
109	"Control the output of verbose kernel messages");
110
111/*
112 * This ensures that there is at least one entry so that the sysinit_set
113 * symbol is not undefined.  A sybsystem ID of SI_SUB_DUMMY is never
114 * executed.
115 */
116SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL);
117
118/*
119 * The sysinit table itself.  Items are checked off as the are run.
120 * If we want to register new sysinit types, add them to newsysinit.
121 */
122SET_DECLARE(sysinit_set, struct sysinit);
123struct sysinit **sysinit, **sysinit_end;
124struct sysinit **newsysinit, **newsysinit_end;
125
126/*
127 * Merge a new sysinit set into the current set, reallocating it if
128 * necessary.  This can only be called after malloc is running.
129 */
130void
131sysinit_add(struct sysinit **set, struct sysinit **set_end)
132{
133	struct sysinit **newset;
134	struct sysinit **sipp;
135	struct sysinit **xipp;
136	int count;
137
138	count = set_end - set;
139	if (newsysinit)
140		count += newsysinit_end - newsysinit;
141	else
142		count += sysinit_end - sysinit;
143	newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
144	if (newset == NULL)
145		panic("cannot malloc for sysinit");
146	xipp = newset;
147	if (newsysinit)
148		for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
149			*xipp++ = *sipp;
150	else
151		for (sipp = sysinit; sipp < sysinit_end; sipp++)
152			*xipp++ = *sipp;
153	for (sipp = set; sipp < set_end; sipp++)
154		*xipp++ = *sipp;
155	if (newsysinit)
156		free(newsysinit, M_TEMP);
157	newsysinit = newset;
158	newsysinit_end = newset + count;
159}
160
161#if defined (DDB) && defined(VERBOSE_SYSINIT)
162static const char *
163symbol_name(vm_offset_t va, db_strategy_t strategy)
164{
165	const char *name;
166	c_db_sym_t sym;
167	db_expr_t  offset;
168
169	if (va == 0)
170		return (NULL);
171	sym = db_search_symbol(va, strategy, &offset);
172	if (offset != 0)
173		return (NULL);
174	db_symbol_values(sym, &name, NULL);
175	return (name);
176}
177#endif
178
179/*
180 * System startup; initialize the world, create process 0, mount root
181 * filesystem, and fork to create init and pagedaemon.  Most of the
182 * hard work is done in the lower-level initialization routines including
183 * startup(), which does memory initialization and autoconfiguration.
184 *
185 * This allows simple addition of new kernel subsystems that require
186 * boot time initialization.  It also allows substitution of subsystem
187 * (for instance, a scheduler, kernel profiler, or VM system) by object
188 * module.  Finally, it allows for optional "kernel threads".
189 */
190void
191mi_startup(void)
192{
193
194	register struct sysinit **sipp;		/* system initialization*/
195	register struct sysinit **xipp;		/* interior loop of sort*/
196	register struct sysinit *save;		/* bubble*/
197
198#if defined(VERBOSE_SYSINIT)
199	int last;
200	int verbose;
201#endif
202
203	if (boothowto & RB_VERBOSE)
204		bootverbose++;
205
206	if (sysinit == NULL) {
207		sysinit = SET_BEGIN(sysinit_set);
208		sysinit_end = SET_LIMIT(sysinit_set);
209	}
210
211restart:
212	/*
213	 * Perform a bubble sort of the system initialization objects by
214	 * their subsystem (primary key) and order (secondary key).
215	 */
216	for (sipp = sysinit; sipp < sysinit_end; sipp++) {
217		for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
218			if ((*sipp)->subsystem < (*xipp)->subsystem ||
219			     ((*sipp)->subsystem == (*xipp)->subsystem &&
220			      (*sipp)->order <= (*xipp)->order))
221				continue;	/* skip*/
222			save = *sipp;
223			*sipp = *xipp;
224			*xipp = save;
225		}
226	}
227
228#if defined(VERBOSE_SYSINIT)
229	last = SI_SUB_COPYRIGHT;
230	verbose = 0;
231#if !defined(DDB)
232	printf("VERBOSE_SYSINIT: DDB not enabled, symbol lookups disabled.\n");
233#endif
234#endif
235
236	/*
237	 * Traverse the (now) ordered list of system initialization tasks.
238	 * Perform each task, and continue on to the next task.
239	 */
240	for (sipp = sysinit; sipp < sysinit_end; sipp++) {
241
242		if ((*sipp)->subsystem == SI_SUB_DUMMY)
243			continue;	/* skip dummy task(s)*/
244
245		if ((*sipp)->subsystem == SI_SUB_DONE)
246			continue;
247
248#if defined(VERBOSE_SYSINIT)
249		if ((*sipp)->subsystem > last) {
250			verbose = 1;
251			last = (*sipp)->subsystem;
252			printf("subsystem %x\n", last);
253		}
254		if (verbose) {
255#if defined(DDB)
256			const char *func, *data;
257
258			func = symbol_name((vm_offset_t)(*sipp)->func,
259			    DB_STGY_PROC);
260			data = symbol_name((vm_offset_t)(*sipp)->udata,
261			    DB_STGY_ANY);
262			if (func != NULL && data != NULL)
263				printf("   %s(&%s)... ", func, data);
264			else if (func != NULL)
265				printf("   %s(%p)... ", func, (*sipp)->udata);
266			else
267#endif
268				printf("   %p(%p)... ", (*sipp)->func,
269				    (*sipp)->udata);
270		}
271#endif
272
273		/* Call function */
274		(*((*sipp)->func))((*sipp)->udata);
275
276#if defined(VERBOSE_SYSINIT)
277		if (verbose)
278			printf("done.\n");
279#endif
280
281		/* Check off the one we're just done */
282		(*sipp)->subsystem = SI_SUB_DONE;
283
284		/* Check if we've installed more sysinit items via KLD */
285		if (newsysinit != NULL) {
286			if (sysinit != SET_BEGIN(sysinit_set))
287				free(sysinit, M_TEMP);
288			sysinit = newsysinit;
289			sysinit_end = newsysinit_end;
290			newsysinit = NULL;
291			newsysinit_end = NULL;
292			goto restart;
293		}
294	}
295
296	mtx_assert(&Giant, MA_OWNED | MA_NOTRECURSED);
297	mtx_unlock(&Giant);
298
299	/*
300	 * Now hand over this thread to swapper.
301	 */
302	swapper();
303	/* NOTREACHED*/
304}
305
306
307/*
308 ***************************************************************************
309 ****
310 **** The following SYSINIT's belong elsewhere, but have not yet
311 **** been moved.
312 ****
313 ***************************************************************************
314 */
315static void
316print_caddr_t(void *data)
317{
318	printf("%s", (char *)data);
319}
320
321static void
322print_version(void *data __unused)
323{
324	int len;
325
326	/* Strip a trailing newline from version. */
327	len = strlen(version);
328	while (len > 0 && version[len - 1] == '\n')
329		len--;
330	printf("%.*s %s\n", len, version, machine);
331	printf("%s\n", compiler_version);
332}
333
334SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t,
335    copyright);
336SYSINIT(trademark, SI_SUB_COPYRIGHT, SI_ORDER_SECOND, print_caddr_t,
337    trademark);
338SYSINIT(version, SI_SUB_COPYRIGHT, SI_ORDER_THIRD, print_version, NULL);
339
340#ifdef WITNESS
341static char wit_warn[] =
342     "WARNING: WITNESS option enabled, expect reduced performance.\n";
343SYSINIT(witwarn, SI_SUB_COPYRIGHT, SI_ORDER_THIRD + 1,
344   print_caddr_t, wit_warn);
345SYSINIT(witwarn2, SI_SUB_LAST, SI_ORDER_THIRD + 1,
346   print_caddr_t, wit_warn);
347#endif
348
349#ifdef DIAGNOSTIC
350static char diag_warn[] =
351     "WARNING: DIAGNOSTIC option enabled, expect reduced performance.\n";
352SYSINIT(diagwarn, SI_SUB_COPYRIGHT, SI_ORDER_THIRD + 2,
353    print_caddr_t, diag_warn);
354SYSINIT(diagwarn2, SI_SUB_LAST, SI_ORDER_THIRD + 2,
355    print_caddr_t, diag_warn);
356#endif
357
358static int
359null_fetch_syscall_args(struct thread *td __unused,
360    struct syscall_args *sa __unused)
361{
362
363	panic("null_fetch_syscall_args");
364}
365
366static void
367null_set_syscall_retval(struct thread *td __unused, int error __unused)
368{
369
370	panic("null_set_syscall_retval");
371}
372
373struct sysentvec null_sysvec = {
374	.sv_size	= 0,
375	.sv_table	= NULL,
376	.sv_mask	= 0,
377	.sv_sigsize	= 0,
378	.sv_sigtbl	= NULL,
379	.sv_errsize	= 0,
380	.sv_errtbl	= NULL,
381	.sv_transtrap	= NULL,
382	.sv_fixup	= NULL,
383	.sv_sendsig	= NULL,
384	.sv_sigcode	= NULL,
385	.sv_szsigcode	= NULL,
386	.sv_prepsyscall	= NULL,
387	.sv_name	= "null",
388	.sv_coredump	= NULL,
389	.sv_imgact_try	= NULL,
390	.sv_minsigstksz	= 0,
391	.sv_pagesize	= PAGE_SIZE,
392	.sv_minuser	= VM_MIN_ADDRESS,
393	.sv_maxuser	= VM_MAXUSER_ADDRESS,
394	.sv_usrstack	= USRSTACK,
395	.sv_psstrings	= PS_STRINGS,
396	.sv_stackprot	= VM_PROT_ALL,
397	.sv_copyout_strings	= NULL,
398	.sv_setregs	= NULL,
399	.sv_fixlimit	= NULL,
400	.sv_maxssiz	= NULL,
401	.sv_flags	= 0,
402	.sv_set_syscall_retval = null_set_syscall_retval,
403	.sv_fetch_syscall_args = null_fetch_syscall_args,
404	.sv_syscallnames = NULL,
405	.sv_schedtail	= NULL,
406};
407
408/*
409 ***************************************************************************
410 ****
411 **** The two following SYSINIT's are proc0 specific glue code.  I am not
412 **** convinced that they can not be safely combined, but their order of
413 **** operation has been maintained as the same as the original init_main.c
414 **** for right now.
415 ****
416 **** These probably belong in init_proc.c or kern_proc.c, since they
417 **** deal with proc0 (the fork template process).
418 ****
419 ***************************************************************************
420 */
421/* ARGSUSED*/
422static void
423proc0_init(void *dummy __unused)
424{
425	struct proc *p;
426	struct thread *td;
427	vm_paddr_t pageablemem;
428	int i;
429
430	GIANT_REQUIRED;
431	p = &proc0;
432	td = &thread0;
433
434	/*
435	 * Initialize magic number and osrel.
436	 */
437	p->p_magic = P_MAGIC;
438	p->p_osrel = osreldate;
439
440	/*
441	 * Initialize thread and process structures.
442	 */
443	procinit();	/* set up proc zone */
444	threadinit();	/* set up UMA zones */
445
446	/*
447	 * Initialise scheduler resources.
448	 * Add scheduler specific parts to proc, thread as needed.
449	 */
450	schedinit();	/* scheduler gets its house in order */
451	/*
452	 * Initialize sleep queue hash table
453	 */
454	sleepinit();
455
456	/*
457	 * additional VM structures
458	 */
459	vm_init2();
460
461	/*
462	 * Create process 0 (the swapper).
463	 */
464	LIST_INSERT_HEAD(&allproc, p, p_list);
465	LIST_INSERT_HEAD(PIDHASH(0), p, p_hash);
466	mtx_init(&pgrp0.pg_mtx, "process group", NULL, MTX_DEF | MTX_DUPOK);
467	p->p_pgrp = &pgrp0;
468	LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
469	LIST_INIT(&pgrp0.pg_members);
470	LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
471
472	pgrp0.pg_session = &session0;
473	mtx_init(&session0.s_mtx, "session", NULL, MTX_DEF);
474	refcount_init(&session0.s_count, 1);
475	session0.s_leader = p;
476
477	p->p_sysent = &null_sysvec;
478	p->p_flag = P_SYSTEM | P_INMEM;
479	p->p_flag2 = 0;
480	p->p_state = PRS_NORMAL;
481	knlist_init_mtx(&p->p_klist, &p->p_mtx);
482	STAILQ_INIT(&p->p_ktr);
483	p->p_nice = NZERO;
484	/* pid_max cannot be greater than PID_MAX */
485	td->td_tid = PID_MAX + 1;
486	LIST_INSERT_HEAD(TIDHASH(td->td_tid), td, td_hash);
487	td->td_state = TDS_RUNNING;
488	td->td_pri_class = PRI_TIMESHARE;
489	td->td_user_pri = PUSER;
490	td->td_base_user_pri = PUSER;
491	td->td_lend_user_pri = PRI_MAX;
492	td->td_priority = PVM;
493	td->td_base_pri = PVM;
494	td->td_oncpu = 0;
495	td->td_flags = TDF_INMEM|TDP_KTHREAD;
496	td->td_cpuset = cpuset_thread0();
497	prison0.pr_cpuset = cpuset_ref(td->td_cpuset);
498	p->p_peers = 0;
499	p->p_leader = p;
500
501
502	strncpy(p->p_comm, "kernel", sizeof (p->p_comm));
503	strncpy(td->td_name, "swapper", sizeof (td->td_name));
504
505	callout_init_mtx(&p->p_itcallout, &p->p_mtx, 0);
506	callout_init_mtx(&p->p_limco, &p->p_mtx, 0);
507	callout_init(&td->td_slpcallout, CALLOUT_MPSAFE);
508
509	/* Create credentials. */
510	p->p_ucred = crget();
511	p->p_ucred->cr_ngroups = 1;	/* group 0 */
512	p->p_ucred->cr_uidinfo = uifind(0);
513	p->p_ucred->cr_ruidinfo = uifind(0);
514	p->p_ucred->cr_prison = &prison0;
515	p->p_ucred->cr_loginclass = loginclass_find("default");
516#ifdef AUDIT
517	audit_cred_kproc0(p->p_ucred);
518#endif
519#ifdef MAC
520	mac_cred_create_swapper(p->p_ucred);
521#endif
522	td->td_ucred = crhold(p->p_ucred);
523
524	/* Create sigacts. */
525	p->p_sigacts = sigacts_alloc();
526
527	/* Initialize signal state for process 0. */
528	siginit(&proc0);
529
530	/* Create the file descriptor table. */
531	p->p_fd = fdinit(NULL);
532	p->p_fdtol = NULL;
533
534	/* Create the limits structures. */
535	p->p_limit = lim_alloc();
536	for (i = 0; i < RLIM_NLIMITS; i++)
537		p->p_limit->pl_rlimit[i].rlim_cur =
538		    p->p_limit->pl_rlimit[i].rlim_max = RLIM_INFINITY;
539	p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_cur =
540	    p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
541	p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_cur =
542	    p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
543	p->p_limit->pl_rlimit[RLIMIT_DATA].rlim_cur = dfldsiz;
544	p->p_limit->pl_rlimit[RLIMIT_DATA].rlim_max = maxdsiz;
545	p->p_limit->pl_rlimit[RLIMIT_STACK].rlim_cur = dflssiz;
546	p->p_limit->pl_rlimit[RLIMIT_STACK].rlim_max = maxssiz;
547	/* Cast to avoid overflow on i386/PAE. */
548	pageablemem = ptoa((vm_paddr_t)cnt.v_free_count);
549	p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_cur =
550	    p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_max = pageablemem;
551	p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = pageablemem / 3;
552	p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_max = pageablemem;
553	p->p_cpulimit = RLIM_INFINITY;
554
555	/* Initialize resource accounting structures. */
556	racct_create(&p->p_racct);
557
558	p->p_stats = pstats_alloc();
559
560	/* Allocate a prototype map so we have something to fork. */
561	pmap_pinit0(vmspace_pmap(&vmspace0));
562	p->p_vmspace = &vmspace0;
563	vmspace0.vm_refcnt = 1;
564
565	/*
566	 * proc0 is not expected to enter usermode, so there is no special
567	 * handling for sv_minuser here, like is done for exec_new_vmspace().
568	 */
569	vm_map_init(&vmspace0.vm_map, vmspace_pmap(&vmspace0),
570	    p->p_sysent->sv_minuser, p->p_sysent->sv_maxuser);
571
572	/*
573	 * Call the init and ctor for the new thread and proc.  We wait
574	 * to do this until all other structures are fairly sane.
575	 */
576	EVENTHANDLER_INVOKE(process_init, p);
577	EVENTHANDLER_INVOKE(thread_init, td);
578	EVENTHANDLER_INVOKE(process_ctor, p);
579	EVENTHANDLER_INVOKE(thread_ctor, td);
580
581	/*
582	 * Charge root for one process.
583	 */
584	(void)chgproccnt(p->p_ucred->cr_ruidinfo, 1, 0);
585	PROC_LOCK(p);
586	racct_add_force(p, RACCT_NPROC, 1);
587	PROC_UNLOCK(p);
588}
589SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL);
590
591/* ARGSUSED*/
592static void
593proc0_post(void *dummy __unused)
594{
595	struct timespec ts;
596	struct proc *p;
597	struct rusage ru;
598	struct thread *td;
599
600	/*
601	 * Now we can look at the time, having had a chance to verify the
602	 * time from the filesystem.  Pretend that proc0 started now.
603	 */
604	sx_slock(&allproc_lock);
605	FOREACH_PROC_IN_SYSTEM(p) {
606		microuptime(&p->p_stats->p_start);
607		PROC_SLOCK(p);
608		rufetch(p, &ru);	/* Clears thread stats */
609		PROC_SUNLOCK(p);
610		p->p_rux.rux_runtime = 0;
611		p->p_rux.rux_uticks = 0;
612		p->p_rux.rux_sticks = 0;
613		p->p_rux.rux_iticks = 0;
614		FOREACH_THREAD_IN_PROC(p, td) {
615			td->td_runtime = 0;
616		}
617	}
618	sx_sunlock(&allproc_lock);
619	PCPU_SET(switchtime, cpu_ticks());
620	PCPU_SET(switchticks, ticks);
621
622	/*
623	 * Give the ``random'' number generator a thump.
624	 */
625	nanotime(&ts);
626	srandom(ts.tv_sec ^ ts.tv_nsec);
627}
628SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL);
629
630static void
631random_init(void *dummy __unused)
632{
633
634	/*
635	 * After CPU has been started we have some randomness on most
636	 * platforms via get_cyclecount().  For platforms that don't
637	 * we will reseed random(9) in proc0_post() as well.
638	 */
639	srandom(get_cyclecount());
640}
641SYSINIT(random, SI_SUB_RANDOM, SI_ORDER_FIRST, random_init, NULL);
642
643/*
644 ***************************************************************************
645 ****
646 **** The following SYSINIT's and glue code should be moved to the
647 **** respective files on a per subsystem basis.
648 ****
649 ***************************************************************************
650 */
651
652
653/*
654 ***************************************************************************
655 ****
656 **** The following code probably belongs in another file, like
657 **** kern/init_init.c.
658 ****
659 ***************************************************************************
660 */
661
662/*
663 * List of paths to try when searching for "init".
664 */
665static char init_path[MAXPATHLEN] =
666#ifdef	INIT_PATH
667    __XSTRING(INIT_PATH);
668#else
669    "/sbin/init:/sbin/oinit:/sbin/init.bak:/rescue/init:/stand/sysinstall";
670#endif
671SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0,
672	"Path used to search the init process");
673
674/*
675 * Shutdown timeout of init(8).
676 * Unused within kernel, but used to control init(8), hence do not remove.
677 */
678#ifndef INIT_SHUTDOWN_TIMEOUT
679#define INIT_SHUTDOWN_TIMEOUT 120
680#endif
681static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT;
682SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout,
683	CTLFLAG_RW, &init_shutdown_timeout, 0, "Shutdown timeout of init(8). "
684	"Unused within kernel, but used to control init(8)");
685
686/*
687 * Start the initial user process; try exec'ing each pathname in init_path.
688 * The program is invoked with one argument containing the boot flags.
689 */
690static void
691start_init(void *dummy)
692{
693	vm_offset_t addr;
694	struct execve_args args;
695	int options, error;
696	char *var, *path, *next, *s;
697	char *ucp, **uap, *arg0, *arg1;
698	struct thread *td;
699	struct proc *p;
700
701	mtx_lock(&Giant);
702
703	GIANT_REQUIRED;
704
705	td = curthread;
706	p = td->td_proc;
707
708	vfs_mountroot();
709
710	/*
711	 * Need just enough stack to hold the faked-up "execve()" arguments.
712	 */
713	addr = p->p_sysent->sv_usrstack - PAGE_SIZE;
714	if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
715			FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0)
716		panic("init: couldn't allocate argument space");
717	p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
718	p->p_vmspace->vm_ssize = 1;
719
720	if ((var = getenv("init_path")) != NULL) {
721		strlcpy(init_path, var, sizeof(init_path));
722		freeenv(var);
723	}
724
725	for (path = init_path; *path != '\0'; path = next) {
726		while (*path == ':')
727			path++;
728		if (*path == '\0')
729			break;
730		for (next = path; *next != '\0' && *next != ':'; next++)
731			/* nothing */ ;
732		if (bootverbose)
733			printf("start_init: trying %.*s\n", (int)(next - path),
734			    path);
735
736		/*
737		 * Move out the boot flag argument.
738		 */
739		options = 0;
740		ucp = (char *)p->p_sysent->sv_usrstack;
741		(void)subyte(--ucp, 0);		/* trailing zero */
742		if (boothowto & RB_SINGLE) {
743			(void)subyte(--ucp, 's');
744			options = 1;
745		}
746#ifdef notyet
747                if (boothowto & RB_FASTBOOT) {
748			(void)subyte(--ucp, 'f');
749			options = 1;
750		}
751#endif
752
753#ifdef BOOTCDROM
754		(void)subyte(--ucp, 'C');
755		options = 1;
756#endif
757
758		if (options == 0)
759			(void)subyte(--ucp, '-');
760		(void)subyte(--ucp, '-');		/* leading hyphen */
761		arg1 = ucp;
762
763		/*
764		 * Move out the file name (also arg 0).
765		 */
766		(void)subyte(--ucp, 0);
767		for (s = next - 1; s >= path; s--)
768			(void)subyte(--ucp, *s);
769		arg0 = ucp;
770
771		/*
772		 * Move out the arg pointers.
773		 */
774		uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
775		(void)suword((caddr_t)--uap, (long)0);	/* terminator */
776		(void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
777		(void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
778
779		/*
780		 * Point at the arguments.
781		 */
782		args.fname = arg0;
783		args.argv = uap;
784		args.envv = NULL;
785
786		/*
787		 * Now try to exec the program.  If can't for any reason
788		 * other than it doesn't exist, complain.
789		 *
790		 * Otherwise, return via fork_trampoline() all the way
791		 * to user mode as init!
792		 */
793		if ((error = sys_execve(td, &args)) == 0) {
794			mtx_unlock(&Giant);
795			return;
796		}
797		if (error != ENOENT)
798			printf("exec %.*s: error %d\n", (int)(next - path),
799			    path, error);
800	}
801	printf("init: not found in path %s\n", init_path);
802	panic("no init");
803}
804
805/*
806 * Like kproc_create(), but runs in it's own address space.
807 * We do this early to reserve pid 1.
808 *
809 * Note special case - do not make it runnable yet.  Other work
810 * in progress will change this more.
811 */
812static void
813create_init(const void *udata __unused)
814{
815	struct ucred *newcred, *oldcred;
816	int error;
817
818	error = fork1(&thread0, RFFDG | RFPROC | RFSTOPPED, 0, &initproc,
819	    NULL, 0);
820	if (error)
821		panic("cannot fork init: %d\n", error);
822	KASSERT(initproc->p_pid == 1, ("create_init: initproc->p_pid != 1"));
823	/* divorce init's credentials from the kernel's */
824	newcred = crget();
825	PROC_LOCK(initproc);
826	initproc->p_flag |= P_SYSTEM | P_INMEM;
827	oldcred = initproc->p_ucred;
828	crcopy(newcred, oldcred);
829#ifdef MAC
830	mac_cred_create_init(newcred);
831#endif
832#ifdef AUDIT
833	audit_cred_proc1(newcred);
834#endif
835	initproc->p_ucred = newcred;
836	PROC_UNLOCK(initproc);
837	crfree(oldcred);
838	cred_update_thread(FIRST_THREAD_IN_PROC(initproc));
839	cpu_set_fork_handler(FIRST_THREAD_IN_PROC(initproc), start_init, NULL);
840}
841SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL);
842
843/*
844 * Make it runnable now.
845 */
846static void
847kick_init(const void *udata __unused)
848{
849	struct thread *td;
850
851	td = FIRST_THREAD_IN_PROC(initproc);
852	thread_lock(td);
853	TD_SET_CAN_RUN(td);
854	sched_add(td, SRQ_BORING);
855	thread_unlock(td);
856}
857SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL);
858