2 * Copyright (c) 1995 Terrence R. Lambert
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.
13 * Redistribution and use in source and binary forms, with or without
14 * modification, are permitted provided that the following conditions
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.
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
41 * @(#)init_main.c 8.9 (Berkeley) 1/21/94
44 #include <sys/cdefs.h>
45 __FBSDID("$FreeBSD$");
47 #include "opt_init_path.h"
50 #include <sys/param.h>
51 #include <sys/kernel.h>
54 #include <sys/filedesc.h>
58 #include <sys/mount.h>
59 #include <sys/mutex.h>
60 #include <sys/syscallsubr.h>
61 #include <sys/sysctl.h>
63 #include <sys/resourcevar.h>
64 #include <sys/systm.h>
65 #include <sys/signalvar.h>
66 #include <sys/vnode.h>
67 #include <sys/sysent.h>
68 #include <sys/reboot.h>
69 #include <sys/sched.h>
71 #include <sys/sysproto.h>
72 #include <sys/vmmeter.h>
73 #include <sys/unistd.h>
74 #include <sys/malloc.h>
77 #include <machine/cpu.h>
79 #include <security/audit/audit.h>
82 #include <vm/vm_param.h>
84 #include <vm/vm_map.h>
85 #include <sys/copyright.h>
87 void mi_startup(void); /* Should be elsewhere */
89 /* Components of the first process -- never freed. */
90 static struct session session0;
91 static struct pgrp pgrp0;
93 struct thread thread0 __aligned(8);
94 struct ksegrp ksegrp0;
95 struct vmspace vmspace0;
96 struct proc *initproc;
98 int boothowto = 0; /* initialized so that it can be patched */
99 SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, "");
101 SYSCTL_INT(_debug, OID_AUTO, bootverbose, CTLFLAG_RW, &bootverbose, 0, "");
104 * This ensures that there is at least one entry so that the sysinit_set
105 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never
108 SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL)
111 * The sysinit table itself. Items are checked off as the are run.
112 * If we want to register new sysinit types, add them to newsysinit.
114 SET_DECLARE(sysinit_set, struct sysinit);
115 struct sysinit **sysinit, **sysinit_end;
116 struct sysinit **newsysinit, **newsysinit_end;
119 * Merge a new sysinit set into the current set, reallocating it if
120 * necessary. This can only be called after malloc is running.
123 sysinit_add(struct sysinit **set, struct sysinit **set_end)
125 struct sysinit **newset;
126 struct sysinit **sipp;
127 struct sysinit **xipp;
130 count = set_end - set;
132 count += newsysinit_end - newsysinit;
134 count += sysinit_end - sysinit;
135 newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
137 panic("cannot malloc for sysinit");
140 for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
143 for (sipp = sysinit; sipp < sysinit_end; sipp++)
145 for (sipp = set; sipp < set_end; sipp++)
148 free(newsysinit, M_TEMP);
150 newsysinit_end = newset + count;
154 * System startup; initialize the world, create process 0, mount root
155 * filesystem, and fork to create init and pagedaemon. Most of the
156 * hard work is done in the lower-level initialization routines including
157 * startup(), which does memory initialization and autoconfiguration.
159 * This allows simple addition of new kernel subsystems that require
160 * boot time initialization. It also allows substitution of subsystem
161 * (for instance, a scheduler, kernel profiler, or VM system) by object
162 * module. Finally, it allows for optional "kernel threads".
168 register struct sysinit **sipp; /* system initialization*/
169 register struct sysinit **xipp; /* interior loop of sort*/
170 register struct sysinit *save; /* bubble*/
172 if (sysinit == NULL) {
173 sysinit = SET_BEGIN(sysinit_set);
174 sysinit_end = SET_LIMIT(sysinit_set);
179 * Perform a bubble sort of the system initialization objects by
180 * their subsystem (primary key) and order (secondary key).
182 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
183 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
184 if ((*sipp)->subsystem < (*xipp)->subsystem ||
185 ((*sipp)->subsystem == (*xipp)->subsystem &&
186 (*sipp)->order <= (*xipp)->order))
195 * Traverse the (now) ordered list of system initialization tasks.
196 * Perform each task, and continue on to the next task.
198 * The last item on the list is expected to be the scheduler,
199 * which will not return.
201 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
203 if ((*sipp)->subsystem == SI_SUB_DUMMY)
204 continue; /* skip dummy task(s)*/
206 if ((*sipp)->subsystem == SI_SUB_DONE)
210 (*((*sipp)->func))((*sipp)->udata);
212 /* Check off the one we're just done */
213 (*sipp)->subsystem = SI_SUB_DONE;
215 /* Check if we've installed more sysinit items via KLD */
216 if (newsysinit != NULL) {
217 if (sysinit != SET_BEGIN(sysinit_set))
218 free(sysinit, M_TEMP);
219 sysinit = newsysinit;
220 sysinit_end = newsysinit_end;
222 newsysinit_end = NULL;
227 panic("Shouldn't get here!");
233 ***************************************************************************
235 **** The following SYSINIT's belong elsewhere, but have not yet
238 ***************************************************************************
241 print_caddr_t(void *data __unused)
243 printf("%s", (char *)data);
245 SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
246 SYSINIT(version, SI_SUB_COPYRIGHT, SI_ORDER_SECOND, print_caddr_t, version)
249 static char wit_warn[] =
250 "WARNING: WITNESS option enabled, expect reduced performance.\n";
251 SYSINIT(witwarn, SI_SUB_COPYRIGHT, SI_ORDER_SECOND + 1,
252 print_caddr_t, wit_warn)
256 static char diag_warn[] =
257 "WARNING: DIAGNOSTIC option enabled, expect reduced performance.\n";
258 SYSINIT(diagwarn, SI_SUB_COPYRIGHT, SI_ORDER_SECOND + 2,
259 print_caddr_t, diag_warn)
263 set_boot_verbose(void *data __unused)
266 if (boothowto & RB_VERBOSE)
269 SYSINIT(boot_verbose, SI_SUB_TUNABLES, SI_ORDER_ANY, set_boot_verbose, NULL)
271 struct sysentvec null_sysvec = {
301 ***************************************************************************
303 **** The two following SYSINIT's are proc0 specific glue code. I am not
304 **** convinced that they can not be safely combined, but their order of
305 **** operation has been maintained as the same as the original init_main.c
308 **** These probably belong in init_proc.c or kern_proc.c, since they
309 **** deal with proc0 (the fork template process).
311 ***************************************************************************
315 proc0_init(void *dummy __unused)
328 * Initialize magic number.
330 p->p_magic = P_MAGIC;
333 * Initialize thread, process and ksegrp structures.
335 procinit(); /* set up proc zone */
336 threadinit(); /* set up thead, upcall and KSEGRP zones */
339 * Initialise scheduler resources.
340 * Add scheduler specific parts to proc, ksegrp, thread as needed.
342 schedinit(); /* scheduler gets its house in order */
344 * Initialize sleep queue hash table
349 * additional VM structures
354 * Create process 0 (the swapper).
356 LIST_INSERT_HEAD(&allproc, p, p_list);
357 LIST_INSERT_HEAD(PIDHASH(0), p, p_hash);
358 mtx_init(&pgrp0.pg_mtx, "process group", NULL, MTX_DEF | MTX_DUPOK);
360 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
361 LIST_INIT(&pgrp0.pg_members);
362 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
364 pgrp0.pg_session = &session0;
365 mtx_init(&session0.s_mtx, "session", NULL, MTX_DEF);
366 session0.s_count = 1;
367 session0.s_leader = p;
369 p->p_sysent = &null_sysvec;
370 p->p_flag = P_SYSTEM;
371 p->p_sflag = PS_INMEM;
372 p->p_state = PRS_NORMAL;
373 knlist_init(&p->p_klist, &p->p_mtx, NULL, NULL, NULL);
374 STAILQ_INIT(&p->p_ktr);
376 td->td_state = TDS_RUNNING;
377 kg->kg_pri_class = PRI_TIMESHARE;
378 kg->kg_user_pri = PUSER;
379 td->td_priority = PVM;
380 td->td_base_pri = PUSER;
386 bcopy("swapper", p->p_comm, sizeof ("swapper"));
388 callout_init(&p->p_itcallout, CALLOUT_MPSAFE);
389 callout_init(&td->td_slpcallout, CALLOUT_MPSAFE);
391 /* Create credentials. */
392 p->p_ucred = crget();
393 p->p_ucred->cr_ngroups = 1; /* group 0 */
394 p->p_ucred->cr_uidinfo = uifind(0);
395 p->p_ucred->cr_ruidinfo = uifind(0);
396 p->p_ucred->cr_prison = NULL; /* Don't jail it. */
399 audit_proc_kproc0(p);
402 mac_create_proc0(p->p_ucred);
404 td->td_ucred = crhold(p->p_ucred);
406 /* Create sigacts. */
407 p->p_sigacts = sigacts_alloc();
409 /* Initialize signal state for process 0. */
412 /* Create the file descriptor table. */
413 p->p_fd = fdinit(NULL);
416 /* Create the limits structures. */
417 p->p_limit = lim_alloc();
418 for (i = 0; i < RLIM_NLIMITS; i++)
419 p->p_limit->pl_rlimit[i].rlim_cur =
420 p->p_limit->pl_rlimit[i].rlim_max = RLIM_INFINITY;
421 p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_cur =
422 p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
423 p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_cur =
424 p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
425 i = ptoa(cnt.v_free_count);
426 p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_max = i;
427 p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
428 p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
429 p->p_cpulimit = RLIM_INFINITY;
431 p->p_stats = pstats_alloc();
433 /* Allocate a prototype map so we have something to fork. */
434 pmap_pinit0(vmspace_pmap(&vmspace0));
435 p->p_vmspace = &vmspace0;
436 vmspace0.vm_refcnt = 1;
437 vm_map_init(&vmspace0.vm_map, p->p_sysent->sv_minuser,
438 p->p_sysent->sv_maxuser);
439 vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0);
442 * Charge root for one process.
444 (void)chgproccnt(p->p_ucred->cr_ruidinfo, 1, 0);
446 SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL)
450 proc0_post(void *dummy __unused)
456 * Now we can look at the time, having had a chance to verify the
457 * time from the filesystem. Pretend that proc0 started now.
459 sx_slock(&allproc_lock);
460 LIST_FOREACH(p, &allproc, p_list) {
461 microuptime(&p->p_stats->p_start);
462 p->p_rux.rux_runtime = 0;
464 sx_sunlock(&allproc_lock);
465 PCPU_SET(switchtime, cpu_ticks());
466 PCPU_SET(switchticks, ticks);
469 * Give the ``random'' number generator a thump.
472 srandom(ts.tv_sec ^ ts.tv_nsec);
474 SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL)
477 ***************************************************************************
479 **** The following SYSINIT's and glue code should be moved to the
480 **** respective files on a per subsystem basis.
482 ***************************************************************************
487 ***************************************************************************
489 **** The following code probably belongs in another file, like
490 **** kern/init_init.c.
492 ***************************************************************************
496 * List of paths to try when searching for "init".
498 static char init_path[MAXPATHLEN] =
500 __XSTRING(INIT_PATH);
502 "/sbin/init:/sbin/oinit:/sbin/init.bak:/rescue/init:/stand/sysinstall";
504 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0,
505 "Path used to search the init process");
508 * Shutdown timeout of init(8).
509 * Unused within kernel, but used to control init(8), hence do not remove.
511 #ifndef INIT_SHUTDOWN_TIMEOUT
512 #define INIT_SHUTDOWN_TIMEOUT 120
514 static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT;
515 SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout,
516 CTLFLAG_RW, &init_shutdown_timeout, 0, "");
519 * Start the initial user process; try exec'ing each pathname in init_path.
520 * The program is invoked with one argument containing the boot flags.
523 start_init(void *dummy)
526 struct execve_args args;
528 char *var, *path, *next, *s;
529 char *ucp, **uap, *arg0, *arg1;
543 * Need just enough stack to hold the faked-up "execve()" arguments.
545 addr = p->p_sysent->sv_usrstack - PAGE_SIZE;
546 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
547 FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0)
548 panic("init: couldn't allocate argument space");
549 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
550 p->p_vmspace->vm_ssize = 1;
552 if ((var = getenv("init_path")) != NULL) {
553 strlcpy(init_path, var, sizeof(init_path));
557 for (path = init_path; *path != '\0'; path = next) {
562 for (next = path; *next != '\0' && *next != ':'; next++)
565 printf("start_init: trying %.*s\n", (int)(next - path),
569 * Move out the boot flag argument.
572 ucp = (char *)p->p_sysent->sv_usrstack;
573 (void)subyte(--ucp, 0); /* trailing zero */
574 if (boothowto & RB_SINGLE) {
575 (void)subyte(--ucp, 's');
579 if (boothowto & RB_FASTBOOT) {
580 (void)subyte(--ucp, 'f');
586 (void)subyte(--ucp, 'C');
591 (void)subyte(--ucp, '-');
592 (void)subyte(--ucp, '-'); /* leading hyphen */
596 * Move out the file name (also arg 0).
598 (void)subyte(--ucp, 0);
599 for (s = next - 1; s >= path; s--)
600 (void)subyte(--ucp, *s);
604 * Move out the arg pointers.
606 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
607 (void)suword((caddr_t)--uap, (long)0); /* terminator */
608 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
609 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
612 * Point at the arguments.
619 * Now try to exec the program. If can't for any reason
620 * other than it doesn't exist, complain.
622 * Otherwise, return via fork_trampoline() all the way
623 * to user mode as init!
625 if ((error = execve(td, &args)) == 0) {
630 printf("exec %.*s: error %d\n", (int)(next - path),
633 printf("init: not found in path %s\n", init_path);
638 * Like kthread_create(), but runs in it's own address space.
639 * We do this early to reserve pid 1.
641 * Note special case - do not make it runnable yet. Other work
642 * in progress will change this more.
645 create_init(const void *udata __unused)
647 struct ucred *newcred, *oldcred;
650 error = fork1(&thread0, RFFDG | RFPROC | RFSTOPPED, 0, &initproc);
652 panic("cannot fork init: %d\n", error);
653 KASSERT(initproc->p_pid == 1, ("create_init: initproc->p_pid != 1"));
654 /* divorce init's credentials from the kernel's */
657 initproc->p_flag |= P_SYSTEM;
658 oldcred = initproc->p_ucred;
659 crcopy(newcred, oldcred);
661 mac_create_proc1(newcred);
664 audit_proc_init(initproc);
666 initproc->p_ucred = newcred;
667 PROC_UNLOCK(initproc);
669 cred_update_thread(FIRST_THREAD_IN_PROC(initproc));
670 mtx_lock_spin(&sched_lock);
671 initproc->p_sflag |= PS_INMEM;
672 mtx_unlock_spin(&sched_lock);
673 cpu_set_fork_handler(FIRST_THREAD_IN_PROC(initproc), start_init, NULL);
675 SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
678 * Make it runnable now.
681 kick_init(const void *udata __unused)
685 td = FIRST_THREAD_IN_PROC(initproc);
686 mtx_lock_spin(&sched_lock);
688 setrunqueue(td, SRQ_BORING); /* XXXKSE */
689 mtx_unlock_spin(&sched_lock);
691 SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)