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>
80 #include <vm/vm_param.h>
82 #include <vm/vm_map.h>
83 #include <sys/copyright.h>
85 void mi_startup(void); /* Should be elsewhere */
87 /* Components of the first process -- never freed. */
88 static struct session session0;
89 static struct pgrp pgrp0;
91 struct thread thread0;
92 struct ksegrp ksegrp0;
93 struct vmspace vmspace0;
94 struct proc *initproc;
96 int boothowto = 0; /* initialized so that it can be patched */
97 SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, "");
99 SYSCTL_INT(_debug, OID_AUTO, bootverbose, CTLFLAG_RW, &bootverbose, 0, "");
102 * This ensures that there is at least one entry so that the sysinit_set
103 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never
106 SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL)
109 * The sysinit table itself. Items are checked off as the are run.
110 * If we want to register new sysinit types, add them to newsysinit.
112 SET_DECLARE(sysinit_set, struct sysinit);
113 struct sysinit **sysinit, **sysinit_end;
114 struct sysinit **newsysinit, **newsysinit_end;
117 * Merge a new sysinit set into the current set, reallocating it if
118 * necessary. This can only be called after malloc is running.
121 sysinit_add(struct sysinit **set, struct sysinit **set_end)
123 struct sysinit **newset;
124 struct sysinit **sipp;
125 struct sysinit **xipp;
128 count = set_end - set;
130 count += newsysinit_end - newsysinit;
132 count += sysinit_end - sysinit;
133 newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
135 panic("cannot malloc for sysinit");
138 for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
141 for (sipp = sysinit; sipp < sysinit_end; sipp++)
143 for (sipp = set; sipp < set_end; sipp++)
146 free(newsysinit, M_TEMP);
148 newsysinit_end = newset + count;
152 * System startup; initialize the world, create process 0, mount root
153 * filesystem, and fork to create init and pagedaemon. Most of the
154 * hard work is done in the lower-level initialization routines including
155 * startup(), which does memory initialization and autoconfiguration.
157 * This allows simple addition of new kernel subsystems that require
158 * boot time initialization. It also allows substitution of subsystem
159 * (for instance, a scheduler, kernel profiler, or VM system) by object
160 * module. Finally, it allows for optional "kernel threads".
166 register struct sysinit **sipp; /* system initialization*/
167 register struct sysinit **xipp; /* interior loop of sort*/
168 register struct sysinit *save; /* bubble*/
170 if (sysinit == NULL) {
171 sysinit = SET_BEGIN(sysinit_set);
172 sysinit_end = SET_LIMIT(sysinit_set);
177 * Perform a bubble sort of the system initialization objects by
178 * their subsystem (primary key) and order (secondary key).
180 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
181 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
182 if ((*sipp)->subsystem < (*xipp)->subsystem ||
183 ((*sipp)->subsystem == (*xipp)->subsystem &&
184 (*sipp)->order <= (*xipp)->order))
193 * Traverse the (now) ordered list of system initialization tasks.
194 * Perform each task, and continue on to the next task.
196 * The last item on the list is expected to be the scheduler,
197 * which will not return.
199 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
201 if ((*sipp)->subsystem == SI_SUB_DUMMY)
202 continue; /* skip dummy task(s)*/
204 if ((*sipp)->subsystem == SI_SUB_DONE)
208 (*((*sipp)->func))((*sipp)->udata);
210 /* Check off the one we're just done */
211 (*sipp)->subsystem = SI_SUB_DONE;
213 /* Check if we've installed more sysinit items via KLD */
214 if (newsysinit != NULL) {
215 if (sysinit != SET_BEGIN(sysinit_set))
216 free(sysinit, M_TEMP);
217 sysinit = newsysinit;
218 sysinit_end = newsysinit_end;
220 newsysinit_end = NULL;
225 panic("Shouldn't get here!");
231 ***************************************************************************
233 **** The following SYSINIT's belong elsewhere, but have not yet
236 ***************************************************************************
239 print_caddr_t(void *data __unused)
241 printf("%s", (char *)data);
243 SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
244 SYSINIT(version, SI_SUB_COPYRIGHT, SI_ORDER_SECOND, print_caddr_t, version)
247 static char wit_warn[] =
248 "WARNING: WITNESS option enabled, expect reduced performance.\n";
249 SYSINIT(witwarn, SI_SUB_COPYRIGHT, SI_ORDER_SECOND + 1,
250 print_caddr_t, wit_warn)
254 static char diag_warn[] =
255 "WARNING: DIAGNOSTIC option enabled, expect reduced performance.\n";
256 SYSINIT(diagwarn, SI_SUB_COPYRIGHT, SI_ORDER_SECOND + 2,
257 print_caddr_t, diag_warn)
261 set_boot_verbose(void *data __unused)
264 if (boothowto & RB_VERBOSE)
267 SYSINIT(boot_verbose, SI_SUB_TUNABLES, SI_ORDER_ANY, set_boot_verbose, NULL)
269 struct sysentvec null_sysvec = {
299 ***************************************************************************
301 **** The two following SYSINIT's are proc0 specific glue code. I am not
302 **** convinced that they can not be safely combined, but their order of
303 **** operation has been maintained as the same as the original init_main.c
306 **** These probably belong in init_proc.c or kern_proc.c, since they
307 **** deal with proc0 (the fork template process).
309 ***************************************************************************
313 proc0_init(void *dummy __unused)
326 * Initialize magic number.
328 p->p_magic = P_MAGIC;
331 * Initialize thread, process and ksegrp structures.
333 procinit(); /* set up proc zone */
334 threadinit(); /* set up thead, upcall and KSEGRP zones */
337 * Initialise scheduler resources.
338 * Add scheduler specific parts to proc, ksegrp, thread as needed.
340 schedinit(); /* scheduler gets its house in order */
342 * Initialize sleep queue hash table
347 * additional VM structures
352 * Create process 0 (the swapper).
354 LIST_INSERT_HEAD(&allproc, p, p_list);
355 LIST_INSERT_HEAD(PIDHASH(0), p, p_hash);
356 mtx_init(&pgrp0.pg_mtx, "process group", NULL, MTX_DEF | MTX_DUPOK);
358 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
359 LIST_INIT(&pgrp0.pg_members);
360 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
362 pgrp0.pg_session = &session0;
363 mtx_init(&session0.s_mtx, "session", NULL, MTX_DEF);
364 session0.s_count = 1;
365 session0.s_leader = p;
367 p->p_sysent = &null_sysvec;
368 p->p_flag = P_SYSTEM;
369 p->p_sflag = PS_INMEM;
370 p->p_state = PRS_NORMAL;
371 knlist_init(&p->p_klist, &p->p_mtx, NULL, NULL, NULL);
372 STAILQ_INIT(&p->p_ktr);
374 td->td_state = TDS_RUNNING;
375 kg->kg_pri_class = PRI_TIMESHARE;
376 kg->kg_user_pri = PUSER;
377 td->td_priority = PVM;
378 td->td_base_pri = PUSER;
384 bcopy("swapper", p->p_comm, sizeof ("swapper"));
386 callout_init(&p->p_itcallout, CALLOUT_MPSAFE);
387 callout_init(&td->td_slpcallout, CALLOUT_MPSAFE);
389 /* Create credentials. */
390 p->p_ucred = crget();
391 p->p_ucred->cr_ngroups = 1; /* group 0 */
392 p->p_ucred->cr_uidinfo = uifind(0);
393 p->p_ucred->cr_ruidinfo = uifind(0);
394 p->p_ucred->cr_prison = NULL; /* Don't jail it. */
396 mac_create_proc0(p->p_ucred);
398 td->td_ucred = crhold(p->p_ucred);
400 /* Create sigacts. */
401 p->p_sigacts = sigacts_alloc();
403 /* Initialize signal state for process 0. */
406 /* Create the file descriptor table. */
407 p->p_fd = fdinit(NULL);
410 /* Create the limits structures. */
411 p->p_limit = lim_alloc();
412 for (i = 0; i < RLIM_NLIMITS; i++)
413 p->p_limit->pl_rlimit[i].rlim_cur =
414 p->p_limit->pl_rlimit[i].rlim_max = RLIM_INFINITY;
415 p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_cur =
416 p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
417 p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_cur =
418 p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
419 i = ptoa(cnt.v_free_count);
420 p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_max = i;
421 p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
422 p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
423 p->p_cpulimit = RLIM_INFINITY;
425 p->p_stats = pstats_alloc();
427 /* Allocate a prototype map so we have something to fork. */
428 pmap_pinit0(vmspace_pmap(&vmspace0));
429 p->p_vmspace = &vmspace0;
430 vmspace0.vm_refcnt = 1;
431 vm_map_init(&vmspace0.vm_map, p->p_sysent->sv_minuser,
432 p->p_sysent->sv_maxuser);
433 vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0);
436 * Charge root for one process.
438 (void)chgproccnt(p->p_ucred->cr_ruidinfo, 1, 0);
440 SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL)
444 proc0_post(void *dummy __unused)
450 * Now we can look at the time, having had a chance to verify the
451 * time from the filesystem. Pretend that proc0 started now.
453 sx_slock(&allproc_lock);
454 LIST_FOREACH(p, &allproc, p_list) {
455 microuptime(&p->p_stats->p_start);
456 p->p_rux.rux_runtime.sec = 0;
457 p->p_rux.rux_runtime.frac = 0;
459 sx_sunlock(&allproc_lock);
460 binuptime(PCPU_PTR(switchtime));
461 PCPU_SET(switchticks, ticks);
464 * Give the ``random'' number generator a thump.
467 srandom(ts.tv_sec ^ ts.tv_nsec);
469 SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL)
472 ***************************************************************************
474 **** The following SYSINIT's and glue code should be moved to the
475 **** respective files on a per subsystem basis.
477 ***************************************************************************
482 ***************************************************************************
484 **** The following code probably belongs in another file, like
485 **** kern/init_init.c.
487 ***************************************************************************
491 * List of paths to try when searching for "init".
493 static char init_path[MAXPATHLEN] =
495 __XSTRING(INIT_PATH);
497 "/sbin/init:/sbin/oinit:/sbin/init.bak:/rescue/init:/stand/sysinstall";
499 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0,
500 "Path used to search the init process");
503 * Shutdown timeout of init(8).
504 * Unused within kernel, but used to control init(8), hence do not remove.
506 #ifndef INIT_SHUTDOWN_TIMEOUT
507 #define INIT_SHUTDOWN_TIMEOUT 120
509 static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT;
510 SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout,
511 CTLFLAG_RW, &init_shutdown_timeout, 0, "");
514 * Start the initial user process; try exec'ing each pathname in init_path.
515 * The program is invoked with one argument containing the boot flags.
518 start_init(void *dummy)
521 struct execve_args args;
523 char *var, *path, *next, *s;
524 char *ucp, **uap, *arg0, *arg1;
538 * Need just enough stack to hold the faked-up "execve()" arguments.
540 addr = p->p_sysent->sv_usrstack - PAGE_SIZE;
541 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
542 FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0)
543 panic("init: couldn't allocate argument space");
544 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
545 p->p_vmspace->vm_ssize = 1;
547 if ((var = getenv("init_path")) != NULL) {
548 strlcpy(init_path, var, sizeof(init_path));
552 for (path = init_path; *path != '\0'; path = next) {
557 for (next = path; *next != '\0' && *next != ':'; next++)
560 printf("start_init: trying %.*s\n", (int)(next - path),
564 * Move out the boot flag argument.
567 ucp = (char *)p->p_sysent->sv_usrstack;
568 (void)subyte(--ucp, 0); /* trailing zero */
569 if (boothowto & RB_SINGLE) {
570 (void)subyte(--ucp, 's');
574 if (boothowto & RB_FASTBOOT) {
575 (void)subyte(--ucp, 'f');
581 (void)subyte(--ucp, 'C');
586 (void)subyte(--ucp, '-');
587 (void)subyte(--ucp, '-'); /* leading hyphen */
591 * Move out the file name (also arg 0).
593 (void)subyte(--ucp, 0);
594 for (s = next - 1; s >= path; s--)
595 (void)subyte(--ucp, *s);
599 * Move out the arg pointers.
601 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
602 (void)suword((caddr_t)--uap, (long)0); /* terminator */
603 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
604 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
607 * Point at the arguments.
614 * Now try to exec the program. If can't for any reason
615 * other than it doesn't exist, complain.
617 * Otherwise, return via fork_trampoline() all the way
618 * to user mode as init!
620 if ((error = execve(td, &args)) == 0) {
625 printf("exec %.*s: error %d\n", (int)(next - path),
628 printf("init: not found in path %s\n", init_path);
633 * Like kthread_create(), but runs in it's own address space.
634 * We do this early to reserve pid 1.
636 * Note special case - do not make it runnable yet. Other work
637 * in progress will change this more.
640 create_init(const void *udata __unused)
642 struct ucred *newcred, *oldcred;
645 error = fork1(&thread0, RFFDG | RFPROC | RFSTOPPED, 0, &initproc);
647 panic("cannot fork init: %d\n", error);
648 KASSERT(initproc->p_pid == 1, ("create_init: initproc->p_pid != 1"));
649 /* divorce init's credentials from the kernel's */
652 initproc->p_flag |= P_SYSTEM;
653 oldcred = initproc->p_ucred;
654 crcopy(newcred, oldcred);
656 mac_create_proc1(newcred);
658 initproc->p_ucred = newcred;
659 PROC_UNLOCK(initproc);
661 cred_update_thread(FIRST_THREAD_IN_PROC(initproc));
662 mtx_lock_spin(&sched_lock);
663 initproc->p_sflag |= PS_INMEM;
664 mtx_unlock_spin(&sched_lock);
665 cpu_set_fork_handler(FIRST_THREAD_IN_PROC(initproc), start_init, NULL);
667 SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
670 * Make it runnable now.
673 kick_init(const void *udata __unused)
677 td = FIRST_THREAD_IN_PROC(initproc);
678 mtx_lock_spin(&sched_lock);
680 setrunqueue(td, SRQ_BORING); /* XXXKSE */
681 mtx_unlock_spin(&sched_lock);
683 SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)