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
42 * $FreeBSD: src/sys/kern/init_main.c,v 1.134.2.8 2003/06/06 20:21:32 tegge Exp $
43 * $DragonFly: src/sys/kern/init_main.c,v 1.68 2006/12/23 00:35:03 swildner Exp $
46 #include "opt_init_path.h"
48 #include <sys/param.h>
50 #include <sys/filedesc.h>
51 #include <sys/kernel.h>
52 #include <sys/mount.h>
53 #include <sys/sysctl.h>
55 #include <sys/resourcevar.h>
56 #include <sys/signalvar.h>
57 #include <sys/systm.h>
58 #include <sys/vnode.h>
59 #include <sys/sysent.h>
60 #include <sys/reboot.h>
61 #include <sys/sysproto.h>
62 #include <sys/vmmeter.h>
63 #include <sys/unistd.h>
64 #include <sys/malloc.h>
65 #include <sys/file2.h>
66 #include <sys/thread2.h>
68 #include <machine/cpu.h>
71 #include <vm/vm_param.h>
74 #include <vm/vm_map.h>
76 #include <sys/copyright.h>
78 /* Components of the first process -- never freed. */
79 static struct session session0;
80 static struct pgrp pgrp0;
81 static struct procsig procsig0;
82 static struct filedesc filedesc0;
83 static struct plimit limit0;
84 static struct vmspace vmspace0;
85 struct proc *initproc;
87 struct thread thread0;
90 extern struct user *proc0paddr;
91 extern int fallback_elf_brand;
93 int boothowto = 0; /* initialized so that it can be patched */
94 SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, "");
97 * This ensures that there is at least one entry so that the sysinit_set
98 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never
101 SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL)
104 * The sysinit table itself. Items are checked off as the are run.
105 * If we want to register new sysinit types, add them to newsysinit.
107 SET_DECLARE(sysinit_set, struct sysinit);
108 struct sysinit **sysinit, **sysinit_end;
109 struct sysinit **newsysinit, **newsysinit_end;
113 * Merge a new sysinit set into the current set, reallocating it if
114 * necessary. This can only be called after malloc is running.
117 sysinit_add(struct sysinit **set, struct sysinit **set_end)
119 struct sysinit **newset;
120 struct sysinit **sipp;
121 struct sysinit **xipp;
124 count = set_end - set;
126 count += newsysinit_end - newsysinit;
128 count += sysinit_end - sysinit;
129 newset = kmalloc(count * sizeof(*sipp), M_TEMP, M_WAITOK);
131 panic("cannot malloc for sysinit");
134 for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
137 for (sipp = sysinit; sipp < sysinit_end; sipp++)
140 for (sipp = set; sipp < set_end; sipp++)
143 kfree(newsysinit, M_TEMP);
145 newsysinit_end = newset + count;
149 * System startup; initialize the world, create process 0, mount root
150 * filesystem, and fork to create init and pagedaemon. Most of the
151 * hard work is done in the lower-level initialization routines including
152 * startup(), which does memory initialization and autoconfiguration.
154 * This allows simple addition of new kernel subsystems that require
155 * boot time initialization. It also allows substitution of subsystem
156 * (for instance, a scheduler, kernel profiler, or VM system) by object
157 * module. Finally, it allows for optional "kernel threads".
162 struct sysinit *sip; /* system initialization*/
163 struct sysinit **sipp; /* system initialization*/
164 struct sysinit **xipp; /* interior loop of sort*/
165 struct sysinit *save; /* bubble*/
167 if (sysinit == NULL) {
168 sysinit = SET_BEGIN(sysinit_set);
169 sysinit_end = SET_LIMIT(sysinit_set);
174 * Perform a bubble sort of the system initialization objects by
175 * their subsystem (primary key) and order (secondary key).
177 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
178 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
179 if ((*sipp)->subsystem < (*xipp)->subsystem ||
180 ((*sipp)->subsystem == (*xipp)->subsystem &&
181 (*sipp)->order <= (*xipp)->order))
190 * Traverse the (now) ordered list of system initialization tasks.
191 * Perform each task, and continue on to the next task.
193 * The last item on the list is expected to be the scheduler,
194 * which will not return.
196 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
198 if (sip->subsystem == SI_SUB_DUMMY)
199 continue; /* skip dummy task(s)*/
201 if (sip->subsystem == SI_SUB_DONE)
205 (*(sip->func))(sip->udata);
207 /* Check off the one we're just done */
208 sip->subsystem = SI_SUB_DONE;
210 /* Check if we've installed more sysinit items via KLD */
211 if (newsysinit != NULL) {
212 if (sysinit != SET_BEGIN(sysinit_set))
213 kfree(sysinit, M_TEMP);
214 sysinit = newsysinit;
215 sysinit_end = newsysinit_end;
217 newsysinit_end = NULL;
222 panic("Shouldn't get here!");
228 ***************************************************************************
230 **** The following SYSINIT's belong elsewhere, but have not yet
233 ***************************************************************************
236 print_caddr_t(void *data __unused)
238 kprintf("%s", (char *)data);
240 SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
243 * Leave the critical section that protected us from spurious interrupts
244 * so device probes work.
247 leavecrit(void *dummy __unused)
250 KKASSERT(!IN_CRITICAL_SECT(curthread));
252 kprintf("Leaving critical section, allowing interrupts\n");
254 SYSINIT(leavecrit, SI_SUB_LEAVE_CRIT, SI_ORDER_ANY, leavecrit, NULL)
257 ***************************************************************************
259 **** The two following SYSINT's are proc0 specific glue code. I am not
260 **** convinced that they can not be safely combined, but their order of
261 **** operation has been maintained as the same as the original init_main.c
264 **** These probably belong in init_proc.c or kern_proc.c, since they
265 **** deal with proc0 (the fork template process).
267 ***************************************************************************
271 proc0_init(void *dummy __unused)
277 lp = &proc0.p_lwp; /* XXX lwp to be: lwp0 */
280 * Initialize process and pgrp structures.
285 * additional VM structures
290 * Create process 0 (the swapper).
292 LIST_INSERT_HEAD(&allproc, p, p_list);
294 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
295 LIST_INIT(&pgrp0.pg_members);
296 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
298 pgrp0.pg_session = &session0;
299 session0.s_count = 1;
300 session0.s_leader = p;
302 p->p_sysent = &aout_sysvec;
304 p->p_flag = P_SYSTEM;
307 p->p_rtprio.type = RTP_PRIO_NORMAL;
308 p->p_rtprio.prio = 0;
309 p->p_lwp.lwp_rtprio = p->p_rtprio;
314 bcopy("swapper", p->p_comm, sizeof ("swapper"));
315 bcopy("swapper", thread0.td_comm, sizeof ("swapper"));
317 /* Create credentials. */
318 p->p_ucred = crget();
319 p->p_ucred->cr_ruidinfo = uifind(0);
320 p->p_ucred->cr_ngroups = 1; /* group 0 */
321 p->p_ucred->cr_uidinfo = uifind(0);
324 p->p_ucred->cr_prison = NULL;
326 /* Create procsig. */
327 p->p_procsig = &procsig0;
328 p->p_procsig->ps_refcnt = 1;
330 /* Initialize signal state for process 0. */
333 /* Create the file descriptor table. */
334 fdinit_bootstrap(p, &filedesc0, cmask);
336 /* Create the limits structures. */
337 plimit_init0(&limit0);
338 p->p_limit = &limit0;
340 /* Allocate a prototype map so we have something to fork. */
341 pmap_pinit0(vmspace_pmap(&vmspace0));
342 p->p_vmspace = &vmspace0;
343 vmspace0.vm_refcnt = 1;
344 vm_map_init(&vmspace0.vm_map, round_page(VM_MIN_USER_ADDRESS),
345 trunc_page(VM_MAX_USER_ADDRESS));
346 vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0);
349 * We continue to place resource usage info and signal
350 * actions in the user struct so they're pageable.
352 p->p_stats = &p->p_addr->u_stats;
353 p->p_sigacts = &p->p_addr->u_sigacts;
356 * Charge root for one process.
358 (void)chgproccnt(p->p_ucred->cr_uidinfo, 1, 0);
361 SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL)
363 static int proc0_post_callback(struct proc *p, void *data __unused);
367 proc0_post(void *dummy __unused)
372 * Now we can look at the time, having had a chance to verify the
373 * time from the file system. Pretend that proc0 started now.
375 allproc_scan(proc0_post_callback, NULL);
378 * Give the ``random'' number generator a thump.
379 * XXX: Does read_random() contain enough bits to be used here ?
382 skrandom(ts.tv_sec ^ ts.tv_nsec);
386 proc0_post_callback(struct proc *p, void *data __unused)
388 microtime(&p->p_start);
392 SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL)
395 ***************************************************************************
397 **** The following SYSINIT's and glue code should be moved to the
398 **** respective files on a per subsystem basis.
400 ***************************************************************************
405 ***************************************************************************
407 **** The following code probably belongs in another file, like
408 **** kern/init_init.c.
410 ***************************************************************************
414 * List of paths to try when searching for "init".
416 static char init_path[MAXPATHLEN] =
418 __XSTRING(INIT_PATH);
420 "/sbin/init:/sbin/oinit:/sbin/init.bak:/stand/sysinstall";
422 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, "");
425 * Start the initial user process; try exec'ing each pathname in init_path.
426 * The program is invoked with one argument containing the boot flags.
428 * The MP lock is held on entry.
431 start_init(void *dummy)
434 struct execve_args args;
436 char *var, *path, *next, *s;
437 char *ucp, **uap, *arg0, *arg1;
445 KKASSERT(p->p_nthreads == 1);
447 lp = LIST_FIRST(&p->p_lwps);
449 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */
450 mp = mountlist_boot_getfirst();
451 if (VFS_ROOT(mp, &vp))
452 panic("cannot find root vnode");
453 if (mp->mnt_ncmountpt.ncp == NULL) {
454 cache_allocroot(&mp->mnt_ncmountpt, mp, vp);
455 cache_unlock(&mp->mnt_ncmountpt); /* leave ref intact */
457 p->p_fd->fd_cdir = vp;
458 vref(p->p_fd->fd_cdir);
459 p->p_fd->fd_rdir = vp;
460 vref(p->p_fd->fd_rdir);
461 vfs_cache_setroot(vp, cache_hold(&mp->mnt_ncmountpt));
462 vn_unlock(vp); /* leave ref intact */
463 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_ncdir);
464 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_nrdir);
467 * Need just enough stack to hold the faked-up "execve()" arguments.
469 addr = trunc_page(USRSTACK - PAGE_SIZE);
470 error = vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
473 VM_PROT_ALL, VM_PROT_ALL,
476 panic("init: couldn't allocate argument space");
477 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
478 p->p_vmspace->vm_ssize = 1;
480 if ((var = kgetenv("init_path")) != NULL) {
481 strncpy(init_path, var, sizeof init_path);
482 init_path[sizeof init_path - 1] = 0;
484 if ((var = kgetenv("kern.fallback_elf_brand")) != NULL)
485 fallback_elf_brand = strtol(var, NULL, 0);
487 for (path = init_path; *path != '\0'; path = next) {
492 for (next = path; *next != '\0' && *next != ':'; next++)
495 kprintf("start_init: trying %.*s\n", (int)(next - path),
499 * Move out the boot flag argument.
502 ucp = (char *)USRSTACK;
503 (void)subyte(--ucp, 0); /* trailing zero */
504 if (boothowto & RB_SINGLE) {
505 (void)subyte(--ucp, 's');
509 if (boothowto & RB_FASTBOOT) {
510 (void)subyte(--ucp, 'f');
516 (void)subyte(--ucp, 'C');
520 (void)subyte(--ucp, '-');
521 (void)subyte(--ucp, '-'); /* leading hyphen */
525 * Move out the file name (also arg 0).
527 (void)subyte(--ucp, 0);
528 for (s = next - 1; s >= path; s--)
529 (void)subyte(--ucp, *s);
533 * Move out the arg pointers.
535 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
536 (void)suword((caddr_t)--uap, (long)0); /* terminator */
537 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
538 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
541 * Point at the arguments.
548 * Now try to exec the program. If can't for any reason
549 * other than it doesn't exist, complain.
551 * Otherwise, return via fork_trampoline() all the way
552 * to user mode as init!
554 * WARNING! We may have been moved to another cpu after
555 * acquiring the current user process designation. The
556 * MP lock will migrate with us though so we still have to
559 if ((error = sys_execve(&args)) == 0) {
561 lp->lwp_proc->p_usched->acquire_curproc(lp);
565 kprintf("exec %.*s: error %d\n", (int)(next - path),
568 kprintf("init: not found in path %s\n", init_path);
573 * Like kthread_create(), but runs in it's own address space.
574 * We do this early to reserve pid 1.
576 * Note special case - do not make it runnable yet. Other work
577 * in progress will change this more.
580 create_init(const void *udata __unused)
586 error = fork1(&proc0.p_lwp, RFFDG | RFPROC, &initproc);
588 panic("cannot fork init: %d", error);
589 initproc->p_flag |= P_SYSTEM;
590 lp = LIST_FIRST(&initproc->p_lwps);
591 cpu_set_fork_handler(lp, start_init, NULL);
594 SYSINIT(init,SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
597 * Make it runnable now.
600 kick_init(const void *udata __unused)
602 start_forked_proc(&proc0.p_lwp, initproc);
604 SYSINIT(kickinit,SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)
607 * Machine independant globaldata initialization
609 * WARNING! Called from early boot, 'mycpu' may not work yet.
612 mi_gdinit(struct globaldata *gd, int cpuid)
614 TAILQ_INIT(&gd->gd_tdfreeq); /* for pmap_{new,dispose}_thread() */
615 TAILQ_INIT(&gd->gd_systimerq);
616 gd->gd_cpuid = cpuid;
617 gd->gd_cpumask = (cpumask_t)1 << cpuid;
619 vm_map_entry_reserve_cpu_init(gd);