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 1.77 2007/04/30 07:18:53 dillon 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>
67 #include <sys/sysref2.h>
69 #include <machine/cpu.h>
72 #include <vm/vm_param.h>
75 #include <vm/vm_map.h>
76 #include <vm/vm_extern.h>
78 #include <sys/copyright.h>
80 /* Components of the first process -- never freed. */
81 static struct session session0;
82 static struct pgrp pgrp0;
83 static struct sigacts sigacts0;
84 static struct filedesc filedesc0;
85 static struct plimit limit0;
86 static struct vmspace vmspace0;
87 struct proc *initproc;
90 struct thread thread0;
93 extern struct user *proc0paddr;
94 extern int fallback_elf_brand;
96 int boothowto = 0; /* initialized so that it can be patched */
97 SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, "");
100 * This ensures that there is at least one entry so that the sysinit_set
101 * symbol is not undefined. A sybsystem ID of SI_SPECIAL_DUMMY is never
104 SYSINIT(placeholder, SI_SPECIAL_DUMMY, SI_ORDER_ANY, NULL, NULL)
107 * The sysinit table itself. Items are checked off as the are run.
108 * If we want to register new sysinit types, add them to newsysinit.
110 SET_DECLARE(sysinit_set, struct sysinit);
111 struct sysinit **sysinit, **sysinit_end;
112 struct sysinit **newsysinit, **newsysinit_end;
116 * Merge a new sysinit set into the current set, reallocating it if
117 * necessary. This can only be called after malloc is running.
120 sysinit_add(struct sysinit **set, struct sysinit **set_end)
122 struct sysinit **newset;
123 struct sysinit **sipp;
124 struct sysinit **xipp;
127 count = set_end - set;
129 count += newsysinit_end - newsysinit;
131 count += sysinit_end - sysinit;
132 newset = kmalloc(count * sizeof(*sipp), M_TEMP, M_WAITOK);
134 panic("cannot malloc for sysinit");
137 for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
140 for (sipp = sysinit; sipp < sysinit_end; sipp++)
143 for (sipp = set; sipp < set_end; sipp++)
146 kfree(newsysinit, M_TEMP);
148 newsysinit_end = newset + count;
152 * Callbacks from machine-dependant startup code (e.g. init386) to set
153 * up low level entities related to cpu #0's globaldata.
155 * Called from very low level boot code.
158 mi_proc0init(struct globaldata *gd, struct user *proc0paddr)
160 lwkt_init_thread(&thread0, proc0paddr, LWKT_THREAD_STACK, 0, gd);
161 lwkt_set_comm(&thread0, "thread0");
162 LIST_INIT(&proc0.p_lwps);
163 LIST_INSERT_HEAD(&proc0.p_lwps, &lwp0, lwp_list);
164 lwp0.lwp_thread = &thread0;
165 lwp0.lwp_proc = &proc0;
166 proc0.p_usched = usched_init();
167 lwp0.lwp_cpumask = 0xFFFFFFFF;
168 varsymset_init(&proc0.p_varsymset, NULL);
169 thread0.td_flags |= TDF_RUNNING;
170 thread0.td_proc = &proc0;
171 thread0.td_lwp = &lwp0;
172 thread0.td_switch = cpu_heavy_switch; /* YYY eventually LWKT */
176 * System startup; initialize the world, create process 0, mount root
177 * filesystem, and fork to create init and pagedaemon. Most of the
178 * hard work is done in the lower-level initialization routines including
179 * startup(), which does memory initialization and autoconfiguration.
181 * This allows simple addition of new kernel subsystems that require
182 * boot time initialization. It also allows substitution of subsystem
183 * (for instance, a scheduler, kernel profiler, or VM system) by object
184 * module. Finally, it allows for optional "kernel threads".
189 struct sysinit *sip; /* system initialization*/
190 struct sysinit **sipp; /* system initialization*/
191 struct sysinit **xipp; /* interior loop of sort*/
192 struct sysinit *save; /* bubble*/
194 if (sysinit == NULL) {
195 sysinit = SET_BEGIN(sysinit_set);
196 sysinit_end = SET_LIMIT(sysinit_set);
201 * Perform a bubble sort of the system initialization objects by
202 * their subsystem (primary key) and order (secondary key).
204 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
205 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
206 if ((*sipp)->subsystem < (*xipp)->subsystem ||
207 ((*sipp)->subsystem == (*xipp)->subsystem &&
208 (*sipp)->order <= (*xipp)->order))
217 * Traverse the (now) ordered list of system initialization tasks.
218 * Perform each task, and continue on to the next task.
220 * The last item on the list is expected to be the scheduler,
221 * which will not return.
223 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
225 if (sip->subsystem == SI_SPECIAL_DUMMY)
226 continue; /* skip dummy task(s)*/
228 if (sip->subsystem == SI_SPECIAL_DONE)
232 (*(sip->func))(sip->udata);
234 /* Check off the one we're just done */
235 sip->subsystem = SI_SPECIAL_DONE;
237 /* Check if we've installed more sysinit items via KLD */
238 if (newsysinit != NULL) {
239 if (sysinit != SET_BEGIN(sysinit_set))
240 kfree(sysinit, M_TEMP);
241 sysinit = newsysinit;
242 sysinit_end = newsysinit_end;
244 newsysinit_end = NULL;
249 panic("Shouldn't get here!");
255 ***************************************************************************
257 **** The following SYSINIT's belong elsewhere, but have not yet
260 ***************************************************************************
263 print_caddr_t(void *data __unused)
265 kprintf("%s", (char *)data);
267 SYSINIT(announce, SI_BOOT1_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
270 * Leave the critical section that protected us from spurious interrupts
271 * so device probes work.
274 leavecrit(void *dummy __unused)
277 KKASSERT(!IN_CRITICAL_SECT(curthread));
279 kprintf("Leaving critical section, allowing interrupts\n");
281 SYSINIT(leavecrit, SI_BOOT2_LEAVE_CRIT, SI_ORDER_ANY, leavecrit, NULL)
284 ***************************************************************************
286 **** The two following SYSINT's are proc0 specific glue code. I am not
287 **** convinced that they can not be safely combined, but their order of
288 **** operation has been maintained as the same as the original init_main.c
291 **** These probably belong in init_proc.c or kern_proc.c, since they
292 **** deal with proc0 (the fork template process).
294 ***************************************************************************
298 proc0_init(void *dummy __unused)
307 * Initialize process and pgrp structures.
312 * additional VM structures
317 * Create process 0 (the swapper).
319 LIST_INSERT_HEAD(&allproc, p, p_list);
321 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
322 LIST_INIT(&pgrp0.pg_members);
323 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
325 pgrp0.pg_session = &session0;
326 session0.s_count = 1;
327 session0.s_leader = p;
329 p->p_sysent = &aout_sysvec;
331 p->p_flag = P_SYSTEM;
333 lp->lwp_stat = LSRUN;
335 p->p_rtprio.type = RTP_PRIO_NORMAL;
336 p->p_rtprio.prio = 0;
337 lp->lwp_rtprio = p->p_rtprio;
342 bcopy("swapper", p->p_comm, sizeof ("swapper"));
343 bcopy("swapper", thread0.td_comm, sizeof ("swapper"));
345 /* Create credentials. */
346 p->p_ucred = crget();
347 p->p_ucred->cr_ruidinfo = uifind(0);
348 p->p_ucred->cr_ngroups = 1; /* group 0 */
349 p->p_ucred->cr_uidinfo = uifind(0);
352 p->p_ucred->cr_prison = NULL;
354 /* Create sigacts. */
355 p->p_sigacts = &sigacts0;
356 p->p_sigacts->ps_refcnt = 1;
358 /* Initialize signal state for process 0. */
361 /* Create the file descriptor table. */
362 fdinit_bootstrap(p, &filedesc0, cmask);
364 /* Create the limits structures. */
365 plimit_init0(&limit0);
366 p->p_limit = &limit0;
368 /* Allocate a prototype map so we have something to fork. */
369 pmap_pinit0(vmspace_pmap(&vmspace0));
370 p->p_vmspace = &vmspace0;
371 sysref_init(&vmspace0.vm_sysref, &vmspace_sysref_class);
372 vm_map_init(&vmspace0.vm_map,
373 round_page(VM_MIN_USER_ADDRESS),
374 trunc_page(VM_MAX_USER_ADDRESS),
375 vmspace_pmap(&vmspace0));
376 sysref_activate(&vmspace0.vm_sysref);
379 * Charge root for one process.
381 (void)chgproccnt(p->p_ucred->cr_uidinfo, 1, 0);
384 SYSINIT(p0init, SI_BOOT2_PROC0, SI_ORDER_FIRST, proc0_init, NULL)
386 static int proc0_post_callback(struct proc *p, void *data __unused);
390 proc0_post(void *dummy __unused)
395 * Now we can look at the time, having had a chance to verify the
396 * time from the file system. Pretend that proc0 started now.
398 allproc_scan(proc0_post_callback, NULL);
401 * Give the ``random'' number generator a thump.
402 * XXX: Does read_random() contain enough bits to be used here ?
405 skrandom(ts.tv_sec ^ ts.tv_nsec);
409 proc0_post_callback(struct proc *p, void *data __unused)
411 microtime(&p->p_start);
415 SYSINIT(p0post, SI_SUB_PROC0_POST, SI_ORDER_FIRST, proc0_post, NULL)
418 ***************************************************************************
420 **** The following SYSINIT's and glue code should be moved to the
421 **** respective files on a per subsystem basis.
423 ***************************************************************************
428 ***************************************************************************
430 **** The following code probably belongs in another file, like
431 **** kern/init_init.c.
433 ***************************************************************************
437 * List of paths to try when searching for "init".
439 static char init_path[MAXPATHLEN] =
441 __XSTRING(INIT_PATH);
443 "/sbin/init:/sbin/oinit:/sbin/init.bak:/stand/sysinstall";
445 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, "");
448 * Start the initial user process; try exec'ing each pathname in init_path.
449 * The program is invoked with one argument containing the boot flags.
451 * The MP lock is held on entry.
454 start_init(void *dummy, struct trapframe *frame)
457 struct execve_args args;
459 char *var, *path, *next, *s;
460 char *ucp, **uap, *arg0, *arg1;
468 lp = ONLY_LWP_IN_PROC(p);
470 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */
471 mp = mountlist_boot_getfirst();
472 if (VFS_ROOT(mp, &vp))
473 panic("cannot find root vnode");
474 if (mp->mnt_ncmountpt.ncp == NULL) {
475 cache_allocroot(&mp->mnt_ncmountpt, mp, vp);
476 cache_unlock(&mp->mnt_ncmountpt); /* leave ref intact */
478 p->p_fd->fd_cdir = vp;
479 vref(p->p_fd->fd_cdir);
480 p->p_fd->fd_rdir = vp;
481 vref(p->p_fd->fd_rdir);
482 vfs_cache_setroot(vp, cache_hold(&mp->mnt_ncmountpt));
483 vn_unlock(vp); /* leave ref intact */
484 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_ncdir);
485 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_nrdir);
488 * Need just enough stack to hold the faked-up "execve()" arguments.
490 addr = trunc_page(USRSTACK - PAGE_SIZE);
491 error = vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
494 VM_PROT_ALL, VM_PROT_ALL,
497 panic("init: couldn't allocate argument space");
498 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
499 p->p_vmspace->vm_ssize = 1;
501 if ((var = kgetenv("init_path")) != NULL) {
502 strncpy(init_path, var, sizeof init_path);
503 init_path[sizeof init_path - 1] = 0;
505 if ((var = kgetenv("kern.fallback_elf_brand")) != NULL)
506 fallback_elf_brand = strtol(var, NULL, 0);
508 for (path = init_path; *path != '\0'; path = next) {
513 for (next = path; *next != '\0' && *next != ':'; next++)
516 kprintf("start_init: trying %.*s\n", (int)(next - path),
520 * Move out the boot flag argument.
523 ucp = (char *)USRSTACK;
524 (void)subyte(--ucp, 0); /* trailing zero */
525 if (boothowto & RB_SINGLE) {
526 (void)subyte(--ucp, 's');
530 if (boothowto & RB_FASTBOOT) {
531 (void)subyte(--ucp, 'f');
537 (void)subyte(--ucp, 'C');
541 (void)subyte(--ucp, '-');
542 (void)subyte(--ucp, '-'); /* leading hyphen */
546 * Move out the file name (also arg 0).
548 (void)subyte(--ucp, 0);
549 for (s = next - 1; s >= path; s--)
550 (void)subyte(--ucp, *s);
554 * Move out the arg pointers.
556 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
557 (void)suword((caddr_t)--uap, (long)0); /* terminator */
558 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
559 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
562 * Point at the arguments.
569 * Now try to exec the program. If can't for any reason
570 * other than it doesn't exist, complain.
572 * Otherwise, return via fork_trampoline() all the way
573 * to user mode as init!
575 * WARNING! We may have been moved to another cpu after
576 * acquiring the current user process designation. The
577 * MP lock will migrate with us though so we still have to
580 if ((error = sys_execve(&args)) == 0) {
582 lp->lwp_proc->p_usched->acquire_curproc(lp);
586 kprintf("exec %.*s: error %d\n", (int)(next - path),
589 kprintf("init: not found in path %s\n", init_path);
594 * Like kthread_create(), but runs in it's own address space.
595 * We do this early to reserve pid 1.
597 * Note special case - do not make it runnable yet. Other work
598 * in progress will change this more.
601 create_init(const void *udata __unused)
607 error = fork1(&lwp0, RFFDG | RFPROC, &initproc);
609 panic("cannot fork init: %d", error);
610 initproc->p_flag |= P_SYSTEM;
611 lp = ONLY_LWP_IN_PROC(initproc);
612 cpu_set_fork_handler(lp, start_init, NULL);
615 SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
618 * Make it runnable now.
621 kick_init(const void *udata __unused)
623 start_forked_proc(&lwp0, initproc);
625 SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)
628 * Machine independant globaldata initialization
630 * WARNING! Called from early boot, 'mycpu' may not work yet.
633 mi_gdinit(struct globaldata *gd, int cpuid)
635 TAILQ_INIT(&gd->gd_tdfreeq); /* for pmap_{new,dispose}_thread() */
636 TAILQ_INIT(&gd->gd_systimerq);
637 gd->gd_sysid_alloc = cpuid; /* prime low bits for cpu lookup */
638 gd->gd_cpuid = cpuid;
639 gd->gd_cpumask = (cpumask_t)1 << cpuid;
641 vm_map_entry_reserve_cpu_init(gd);