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 1.23 2003/09/24 18:37:48 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>
68 #include <machine/cpu.h>
71 #include <vm/vm_param.h>
74 #include <vm/vm_map.h>
76 #include <sys/copyright.h>
78 extern struct linker_set sysinit_set; /* XXX */
80 void mi_startup(void); /* Should be elsewhere */
82 /* Components of the first process -- never freed. */
83 static struct session session0;
84 static struct pgrp pgrp0;
85 static struct procsig procsig0;
86 static struct filedesc0 filedesc0;
87 static struct plimit limit0;
88 static struct vmspace vmspace0;
89 struct proc *initproc;
91 struct thread thread0;
94 extern struct user *proc0paddr;
95 extern int fallback_elf_brand;
98 int boothowto = 0; /* initialized so that it can be patched */
99 SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 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 struct sysinit **sysinit = (struct sysinit **)sysinit_set.ls_items;
113 struct sysinit **newsysinit;
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)
122 struct sysinit **newset;
123 struct sysinit **sipp;
124 struct sysinit **xipp;
128 for (sipp = newsysinit; *sipp; sipp++)
131 for (sipp = sysinit; *sipp; sipp++)
133 for (sipp = set; *sipp; sipp++)
135 count++; /* Trailing NULL */
136 newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
138 panic("cannot malloc for sysinit");
141 for (sipp = newsysinit; *sipp; sipp++)
144 for (sipp = sysinit; *sipp; sipp++)
146 for (sipp = set; *sipp; sipp++)
150 free(newsysinit, M_TEMP);
155 * System startup; initialize the world, create process 0, mount root
156 * filesystem, and fork to create init and pagedaemon. Most of the
157 * hard work is done in the lower-level initialization routines including
158 * startup(), which does memory initialization and autoconfiguration.
160 * This allows simple addition of new kernel subsystems that require
161 * boot time initialization. It also allows substitution of subsystem
162 * (for instance, a scheduler, kernel profiler, or VM system) by object
163 * module. Finally, it allows for optional "kernel threads".
168 struct sysinit *sip; /* system initialization*/
169 struct sysinit **sipp; /* system initialization*/
170 struct sysinit **xipp; /* interior loop of sort*/
171 struct sysinit *save; /* bubble*/
175 * Perform a bubble sort of the system initialization objects by
176 * their subsystem (primary key) and order (secondary key).
178 for (sipp = sysinit; *sipp; sipp++) {
179 for (xipp = sipp + 1; *xipp; xipp++) {
180 if ((*sipp)->subsystem < (*xipp)->subsystem ||
181 ((*sipp)->subsystem == (*xipp)->subsystem &&
182 (*sipp)->order <= (*xipp)->order))
191 * Traverse the (now) ordered list of system initialization tasks.
192 * Perform each task, and continue on to the next task.
194 * The last item on the list is expected to be the scheduler,
195 * which will not return.
197 for (sipp = sysinit; (sip = *sipp) != NULL; 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 != (struct sysinit **)sysinit_set.ls_items)
213 free(sysinit, M_TEMP);
214 sysinit = newsysinit;
220 panic("Shouldn't get here!");
226 ***************************************************************************
228 **** The following SYSINIT's belong elsewhere, but have not yet
231 ***************************************************************************
234 print_caddr_t(void *data __unused)
236 printf("%s", (char *)data);
238 SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
241 * Leave the critical section that protected us from spurious interrupts
242 * so device probes work.
245 leavecrit(void *dummy __unused)
248 KKASSERT(!IN_CRITICAL_SECT(curthread));
250 printf("Leaving critical section, allowing interrupts\n");
252 SYSINIT(leavecrit, SI_SUB_LEAVE_CRIT, SI_ORDER_ANY, leavecrit, NULL)
255 ***************************************************************************
257 **** The two following SYSINT's are proc0 specific glue code. I am not
258 **** convinced that they can not be safely combined, but their order of
259 **** operation has been maintained as the same as the original init_main.c
262 **** These probably belong in init_proc.c or kern_proc.c, since they
263 **** deal with proc0 (the fork template process).
265 ***************************************************************************
269 proc0_init(void *dummy __unused)
272 struct filedesc0 *fdp;
278 * Initialize process and pgrp structures.
283 * Initialize sleep queue hash table
288 * additional VM structures
293 * Create process 0 (the swapper).
295 LIST_INSERT_HEAD(&allproc, p, p_list);
297 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
298 LIST_INIT(&pgrp0.pg_members);
299 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
301 pgrp0.pg_session = &session0;
302 session0.s_count = 1;
303 session0.s_leader = p;
305 p->p_sysent = &aout_sysvec;
307 p->p_flag = P_INMEM | P_SYSTEM | P_CP_RELEASED;
310 p->p_rtprio.type = RTP_PRIO_NORMAL;
311 p->p_rtprio.prio = 0;
316 bcopy("swapper", p->p_comm, sizeof ("swapper"));
318 /* Create credentials. */
319 p->p_ucred = crget();
320 p->p_ucred->cr_ruidinfo = uifind(0);
321 p->p_ucred->cr_ngroups = 1; /* group 0 */
322 p->p_ucred->cr_uidinfo = uifind(0);
325 p->p_ucred->cr_prison = NULL;
327 /* Create procsig. */
328 p->p_procsig = &procsig0;
329 p->p_procsig->ps_refcnt = 1;
331 /* Initialize signal state for process 0. */
334 /* Create the file descriptor table. */
336 p->p_fd = &fdp->fd_fd;
338 fdp->fd_fd.fd_refcnt = 1;
339 fdp->fd_fd.fd_cmask = cmask;
340 fdp->fd_fd.fd_ofiles = fdp->fd_dfiles;
341 fdp->fd_fd.fd_ofileflags = fdp->fd_dfileflags;
342 fdp->fd_fd.fd_nfiles = NDFILE;
344 /* Create the limits structures. */
345 p->p_limit = &limit0;
346 for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++)
347 limit0.pl_rlimit[i].rlim_cur =
348 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
349 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur =
350 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
351 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur =
352 limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
353 i = ptoa(vmstats.v_free_count);
354 limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i;
355 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
356 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
357 limit0.p_cpulimit = RLIM_INFINITY;
360 /* Allocate a prototype map so we have something to fork. */
361 pmap_pinit0(vmspace_pmap(&vmspace0));
362 p->p_vmspace = &vmspace0;
363 vmspace0.vm_refcnt = 1;
364 vm_map_init(&vmspace0.vm_map, round_page(VM_MIN_ADDRESS),
365 trunc_page(VM_MAXUSER_ADDRESS));
366 vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0);
369 * We continue to place resource usage info and signal
370 * actions in the user struct so they're pageable.
372 p->p_stats = &p->p_addr->u_stats;
373 p->p_sigacts = &p->p_addr->u_sigacts;
376 * Charge root for one process.
378 (void)chgproccnt(p->p_ucred->cr_uidinfo, 1, 0);
381 SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL)
385 proc0_post(void *dummy __unused)
391 * Now we can look at the time, having had a chance to verify the
392 * time from the file system. Pretend that proc0 started now.
394 FOREACH_PROC_IN_SYSTEM(p) {
395 microtime(&p->p_stats->p_start);
399 * Give the ``random'' number generator a thump.
400 * XXX: Does read_random() contain enough bits to be used here ?
403 srandom(ts.tv_sec ^ ts.tv_nsec);
405 SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL)
408 ***************************************************************************
410 **** The following SYSINIT's and glue code should be moved to the
411 **** respective files on a per subsystem basis.
413 ***************************************************************************
418 ***************************************************************************
420 **** The following code probably belongs in another file, like
421 **** kern/init_init.c.
423 ***************************************************************************
427 * List of paths to try when searching for "init".
429 static char init_path[MAXPATHLEN] =
431 __XSTRING(INIT_PATH);
433 "/sbin/init:/sbin/oinit:/sbin/init.bak:/stand/sysinstall";
435 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, "");
438 * Start the initial user process; try exec'ing each pathname in init_path.
439 * The program is invoked with one argument containing the boot flags.
441 * The MP lock is held on entry.
444 start_init(void *dummy)
447 struct execve_args args;
449 char *var, *path, *next, *s;
450 char *ucp, **uap, *arg0, *arg1;
455 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */
456 if (VFS_ROOT(TAILQ_FIRST(&mountlist), &rootvnode))
457 panic("cannot find root vnode");
458 p->p_fd->fd_cdir = rootvnode;
459 VREF(p->p_fd->fd_cdir);
460 p->p_fd->fd_rdir = rootvnode;
461 VREF(p->p_fd->fd_rdir);
462 vfs_cache_setroot(rootvnode);
463 VOP_UNLOCK(rootvnode, 0, curthread);
466 * Need just enough stack to hold the faked-up "execve()" arguments.
468 addr = trunc_page(USRSTACK - PAGE_SIZE);
469 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
470 FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0)
471 panic("init: couldn't allocate argument space");
472 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
473 p->p_vmspace->vm_ssize = 1;
475 if ((var = getenv("init_path")) != NULL) {
476 strncpy(init_path, var, sizeof init_path);
477 init_path[sizeof init_path - 1] = 0;
479 if ((var = getenv("kern.fallback_elf_brand")) != NULL)
480 fallback_elf_brand = strtol(var, NULL, 0);
482 for (path = init_path; *path != '\0'; path = next) {
487 for (next = path; *next != '\0' && *next != ':'; next++)
490 printf("start_init: trying %.*s\n", (int)(next - path),
494 * Move out the boot flag argument.
497 ucp = (char *)USRSTACK;
498 (void)subyte(--ucp, 0); /* trailing zero */
499 if (boothowto & RB_SINGLE) {
500 (void)subyte(--ucp, 's');
504 if (boothowto & RB_FASTBOOT) {
505 (void)subyte(--ucp, 'f');
511 (void)subyte(--ucp, 'C');
515 (void)subyte(--ucp, '-');
516 (void)subyte(--ucp, '-'); /* leading hyphen */
520 * Move out the file name (also arg 0).
522 (void)subyte(--ucp, 0);
523 for (s = next - 1; s >= path; s--)
524 (void)subyte(--ucp, *s);
528 * Move out the arg pointers.
530 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
531 (void)suword((caddr_t)--uap, (long)0); /* terminator */
532 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
533 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
536 * Point at the arguments.
543 * Now try to exec the program. If can't for any reason
544 * other than it doesn't exist, complain.
546 * Otherwise, return via fork_trampoline() all the way
547 * to user mode as init!
549 * WARNING! We may have been moved to another cpu after
550 * acquiring P_CURPROC. The MP lock will migrate with us
551 * though so we still have to release it.
553 if ((error = execve(&args)) == 0) {
559 printf("exec %.*s: error %d\n", (int)(next - path),
562 printf("init: not found in path %s\n", init_path);
567 * Like kthread_create(), but runs in it's own address space.
568 * We do this early to reserve pid 1.
570 * Note special case - do not make it runnable yet. Other work
571 * in progress will change this more.
574 create_init(const void *udata __unused)
580 error = fork1(&proc0, RFFDG | RFPROC, &initproc);
582 panic("cannot fork init: %d\n", error);
583 initproc->p_flag |= P_INMEM | P_SYSTEM;
584 cpu_set_fork_handler(initproc, start_init, NULL);
587 SYSINIT(init,SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
590 * Make it runnable now.
593 kick_init(const void *udata __unused)
595 start_forked_proc(&proc0, initproc);
597 SYSINIT(kickinit,SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)
600 * Machine independant globaldata initialization
602 * WARNING! Called from early boot, 'mycpu' may not work yet.
605 mi_gdinit(struct globaldata *gd, int cpuid)
607 TAILQ_INIT(&gd->gd_tdfreeq); /* for pmap_{new,dispose}_thread() */
608 gd->gd_cpuid = cpuid;