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 1.72 2007/02/03 17:05:57 corecode 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;
88 struct thread thread0;
91 extern struct user *proc0paddr;
92 extern int fallback_elf_brand;
94 int boothowto = 0; /* initialized so that it can be patched */
95 SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, "");
98 * This ensures that there is at least one entry so that the sysinit_set
99 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never
102 SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL)
105 * The sysinit table itself. Items are checked off as the are run.
106 * If we want to register new sysinit types, add them to newsysinit.
108 SET_DECLARE(sysinit_set, struct sysinit);
109 struct sysinit **sysinit, **sysinit_end;
110 struct sysinit **newsysinit, **newsysinit_end;
114 * Merge a new sysinit set into the current set, reallocating it if
115 * necessary. This can only be called after malloc is running.
118 sysinit_add(struct sysinit **set, struct sysinit **set_end)
120 struct sysinit **newset;
121 struct sysinit **sipp;
122 struct sysinit **xipp;
125 count = set_end - set;
127 count += newsysinit_end - newsysinit;
129 count += sysinit_end - sysinit;
130 newset = kmalloc(count * sizeof(*sipp), M_TEMP, M_WAITOK);
132 panic("cannot malloc for sysinit");
135 for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
138 for (sipp = sysinit; sipp < sysinit_end; sipp++)
141 for (sipp = set; sipp < set_end; sipp++)
144 kfree(newsysinit, M_TEMP);
146 newsysinit_end = newset + count;
150 * Callbacks from machine-dependant startup code (e.g. init386) to set
151 * up low level entities related to cpu #0's globaldata.
153 * Called from very low level boot code.
156 mi_proc0init(struct globaldata *gd, struct user *proc0paddr)
158 lwkt_init_thread(&thread0, proc0paddr, LWKT_THREAD_STACK, 0, gd);
159 lwkt_set_comm(&thread0, "thread0");
160 proc0.p_addr = (void *)thread0.td_kstack;
161 LIST_INIT(&proc0.p_lwps);
162 LIST_INSERT_HEAD(&proc0.p_lwps, &lwp0, lwp_list);
163 lwp0.lwp_thread = &thread0;
164 lwp0.lwp_proc = &proc0;
165 proc0.p_usched = usched_init();
166 lwp0.lwp_cpumask = 0xFFFFFFFF;
167 varsymset_init(&proc0.p_varsymset, NULL);
168 thread0.td_flags |= TDF_RUNNING;
169 thread0.td_proc = &proc0;
170 thread0.td_lwp = &lwp0;
171 thread0.td_switch = cpu_heavy_switch; /* YYY eventually LWKT */
175 * System startup; initialize the world, create process 0, mount root
176 * filesystem, and fork to create init and pagedaemon. Most of the
177 * hard work is done in the lower-level initialization routines including
178 * startup(), which does memory initialization and autoconfiguration.
180 * This allows simple addition of new kernel subsystems that require
181 * boot time initialization. It also allows substitution of subsystem
182 * (for instance, a scheduler, kernel profiler, or VM system) by object
183 * module. Finally, it allows for optional "kernel threads".
188 struct sysinit *sip; /* system initialization*/
189 struct sysinit **sipp; /* system initialization*/
190 struct sysinit **xipp; /* interior loop of sort*/
191 struct sysinit *save; /* bubble*/
193 if (sysinit == NULL) {
194 sysinit = SET_BEGIN(sysinit_set);
195 sysinit_end = SET_LIMIT(sysinit_set);
200 * Perform a bubble sort of the system initialization objects by
201 * their subsystem (primary key) and order (secondary key).
203 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
204 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
205 if ((*sipp)->subsystem < (*xipp)->subsystem ||
206 ((*sipp)->subsystem == (*xipp)->subsystem &&
207 (*sipp)->order <= (*xipp)->order))
216 * Traverse the (now) ordered list of system initialization tasks.
217 * Perform each task, and continue on to the next task.
219 * The last item on the list is expected to be the scheduler,
220 * which will not return.
222 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
224 if (sip->subsystem == SI_SUB_DUMMY)
225 continue; /* skip dummy task(s)*/
227 if (sip->subsystem == SI_SUB_DONE)
231 (*(sip->func))(sip->udata);
233 /* Check off the one we're just done */
234 sip->subsystem = SI_SUB_DONE;
236 /* Check if we've installed more sysinit items via KLD */
237 if (newsysinit != NULL) {
238 if (sysinit != SET_BEGIN(sysinit_set))
239 kfree(sysinit, M_TEMP);
240 sysinit = newsysinit;
241 sysinit_end = newsysinit_end;
243 newsysinit_end = NULL;
248 panic("Shouldn't get here!");
254 ***************************************************************************
256 **** The following SYSINIT's belong elsewhere, but have not yet
259 ***************************************************************************
262 print_caddr_t(void *data __unused)
264 kprintf("%s", (char *)data);
266 SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
269 * Leave the critical section that protected us from spurious interrupts
270 * so device probes work.
273 leavecrit(void *dummy __unused)
276 KKASSERT(!IN_CRITICAL_SECT(curthread));
278 kprintf("Leaving critical section, allowing interrupts\n");
280 SYSINIT(leavecrit, SI_SUB_LEAVE_CRIT, SI_ORDER_ANY, leavecrit, NULL)
283 ***************************************************************************
285 **** The two following SYSINT's are proc0 specific glue code. I am not
286 **** convinced that they can not be safely combined, but their order of
287 **** operation has been maintained as the same as the original init_main.c
290 **** These probably belong in init_proc.c or kern_proc.c, since they
291 **** deal with proc0 (the fork template process).
293 ***************************************************************************
297 proc0_init(void *dummy __unused)
306 * Initialize process and pgrp structures.
311 * additional VM structures
316 * Create process 0 (the swapper).
318 LIST_INSERT_HEAD(&allproc, p, p_list);
320 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
321 LIST_INIT(&pgrp0.pg_members);
322 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
324 pgrp0.pg_session = &session0;
325 session0.s_count = 1;
326 session0.s_leader = p;
328 p->p_sysent = &aout_sysvec;
330 p->p_flag = P_SYSTEM;
333 p->p_rtprio.type = RTP_PRIO_NORMAL;
334 p->p_rtprio.prio = 0;
335 lp->lwp_rtprio = p->p_rtprio;
340 bcopy("swapper", p->p_comm, sizeof ("swapper"));
341 bcopy("swapper", thread0.td_comm, sizeof ("swapper"));
343 /* Create credentials. */
344 p->p_ucred = crget();
345 p->p_ucred->cr_ruidinfo = uifind(0);
346 p->p_ucred->cr_ngroups = 1; /* group 0 */
347 p->p_ucred->cr_uidinfo = uifind(0);
350 p->p_ucred->cr_prison = NULL;
352 /* Create procsig. */
353 p->p_procsig = &procsig0;
354 p->p_procsig->ps_refcnt = 1;
356 /* Initialize signal state for process 0. */
359 /* Create the file descriptor table. */
360 fdinit_bootstrap(p, &filedesc0, cmask);
362 /* Create the limits structures. */
363 plimit_init0(&limit0);
364 p->p_limit = &limit0;
366 /* Allocate a prototype map so we have something to fork. */
367 pmap_pinit0(vmspace_pmap(&vmspace0));
368 p->p_vmspace = &vmspace0;
369 vmspace0.vm_refcnt = 1;
370 vm_map_init(&vmspace0.vm_map,
371 round_page(VM_MIN_USER_ADDRESS),
372 trunc_page(VM_MAX_USER_ADDRESS),
373 vmspace_pmap(&vmspace0));
376 * We continue to place signal
377 * actions in the user struct so they're pageable.
379 * XXX old + cufty. will be removed
381 p->p_sigacts = &p->p_addr->u_sigacts;
384 * Charge root for one process.
386 (void)chgproccnt(p->p_ucred->cr_uidinfo, 1, 0);
389 SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL)
391 static int proc0_post_callback(struct proc *p, void *data __unused);
395 proc0_post(void *dummy __unused)
400 * Now we can look at the time, having had a chance to verify the
401 * time from the file system. Pretend that proc0 started now.
403 allproc_scan(proc0_post_callback, NULL);
406 * Give the ``random'' number generator a thump.
407 * XXX: Does read_random() contain enough bits to be used here ?
410 skrandom(ts.tv_sec ^ ts.tv_nsec);
414 proc0_post_callback(struct proc *p, void *data __unused)
416 microtime(&p->p_start);
420 SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL)
423 ***************************************************************************
425 **** The following SYSINIT's and glue code should be moved to the
426 **** respective files on a per subsystem basis.
428 ***************************************************************************
433 ***************************************************************************
435 **** The following code probably belongs in another file, like
436 **** kern/init_init.c.
438 ***************************************************************************
442 * List of paths to try when searching for "init".
444 static char init_path[MAXPATHLEN] =
446 __XSTRING(INIT_PATH);
448 "/sbin/init:/sbin/oinit:/sbin/init.bak:/stand/sysinstall";
450 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, "");
453 * Start the initial user process; try exec'ing each pathname in init_path.
454 * The program is invoked with one argument containing the boot flags.
456 * The MP lock is held on entry.
459 start_init(void *dummy)
462 struct execve_args args;
464 char *var, *path, *next, *s;
465 char *ucp, **uap, *arg0, *arg1;
473 lp = ONLY_LWP_IN_PROC(p);
475 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */
476 mp = mountlist_boot_getfirst();
477 if (VFS_ROOT(mp, &vp))
478 panic("cannot find root vnode");
479 if (mp->mnt_ncmountpt.ncp == NULL) {
480 cache_allocroot(&mp->mnt_ncmountpt, mp, vp);
481 cache_unlock(&mp->mnt_ncmountpt); /* leave ref intact */
483 p->p_fd->fd_cdir = vp;
484 vref(p->p_fd->fd_cdir);
485 p->p_fd->fd_rdir = vp;
486 vref(p->p_fd->fd_rdir);
487 vfs_cache_setroot(vp, cache_hold(&mp->mnt_ncmountpt));
488 vn_unlock(vp); /* leave ref intact */
489 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_ncdir);
490 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_nrdir);
493 * Need just enough stack to hold the faked-up "execve()" arguments.
495 addr = trunc_page(USRSTACK - PAGE_SIZE);
496 error = vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
499 VM_PROT_ALL, VM_PROT_ALL,
502 panic("init: couldn't allocate argument space");
503 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
504 p->p_vmspace->vm_ssize = 1;
506 if ((var = kgetenv("init_path")) != NULL) {
507 strncpy(init_path, var, sizeof init_path);
508 init_path[sizeof init_path - 1] = 0;
510 if ((var = kgetenv("kern.fallback_elf_brand")) != NULL)
511 fallback_elf_brand = strtol(var, NULL, 0);
513 for (path = init_path; *path != '\0'; path = next) {
518 for (next = path; *next != '\0' && *next != ':'; next++)
521 kprintf("start_init: trying %.*s\n", (int)(next - path),
525 * Move out the boot flag argument.
528 ucp = (char *)USRSTACK;
529 (void)subyte(--ucp, 0); /* trailing zero */
530 if (boothowto & RB_SINGLE) {
531 (void)subyte(--ucp, 's');
535 if (boothowto & RB_FASTBOOT) {
536 (void)subyte(--ucp, 'f');
542 (void)subyte(--ucp, 'C');
546 (void)subyte(--ucp, '-');
547 (void)subyte(--ucp, '-'); /* leading hyphen */
551 * Move out the file name (also arg 0).
553 (void)subyte(--ucp, 0);
554 for (s = next - 1; s >= path; s--)
555 (void)subyte(--ucp, *s);
559 * Move out the arg pointers.
561 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
562 (void)suword((caddr_t)--uap, (long)0); /* terminator */
563 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
564 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
567 * Point at the arguments.
574 * Now try to exec the program. If can't for any reason
575 * other than it doesn't exist, complain.
577 * Otherwise, return via fork_trampoline() all the way
578 * to user mode as init!
580 * WARNING! We may have been moved to another cpu after
581 * acquiring the current user process designation. The
582 * MP lock will migrate with us though so we still have to
585 if ((error = sys_execve(&args)) == 0) {
587 lp->lwp_proc->p_usched->acquire_curproc(lp);
591 kprintf("exec %.*s: error %d\n", (int)(next - path),
594 kprintf("init: not found in path %s\n", init_path);
599 * Like kthread_create(), but runs in it's own address space.
600 * We do this early to reserve pid 1.
602 * Note special case - do not make it runnable yet. Other work
603 * in progress will change this more.
606 create_init(const void *udata __unused)
612 error = fork1(&lwp0, RFFDG | RFPROC, &initproc);
614 panic("cannot fork init: %d", error);
615 initproc->p_flag |= P_SYSTEM;
616 lp = ONLY_LWP_IN_PROC(initproc);
617 cpu_set_fork_handler(lp, start_init, NULL);
620 SYSINIT(init,SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
623 * Make it runnable now.
626 kick_init(const void *udata __unused)
628 start_forked_proc(&lwp0, initproc);
630 SYSINIT(kickinit,SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)
633 * Machine independant globaldata initialization
635 * WARNING! Called from early boot, 'mycpu' may not work yet.
638 mi_gdinit(struct globaldata *gd, int cpuid)
640 TAILQ_INIT(&gd->gd_tdfreeq); /* for pmap_{new,dispose}_thread() */
641 TAILQ_INIT(&gd->gd_systimerq);
642 gd->gd_cpuid = cpuid;
643 gd->gd_cpumask = (cpumask_t)1 << cpuid;
645 vm_map_entry_reserve_cpu_init(gd);