xref: /netbsd-src/sys/kern/init_main.c (revision 817cd315412cb2ce0bc4e6bf3f12a866fd948396)
1 /*	$NetBSD: init_main.c,v 1.412 2009/12/17 01:25:10 rmind Exp $	*/
2 
3 /*-
4  * Copyright (c) 2008, 2009 The NetBSD Foundation, Inc.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  *
16  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
17  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
18  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
19  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
20  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
21  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
22  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
23  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
24  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
25  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
26  * POSSIBILITY OF SUCH DAMAGE.
27  */
28 
29 /*
30  * Copyright (c) 1982, 1986, 1989, 1991, 1992, 1993
31  *	The Regents of the University of California.  All rights reserved.
32  * (c) UNIX System Laboratories, Inc.
33  * All or some portions of this file are derived from material licensed
34  * to the University of California by American Telephone and Telegraph
35  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
36  * the permission of UNIX System Laboratories, Inc.
37  *
38  * Redistribution and use in source and binary forms, with or without
39  * modification, are permitted provided that the following conditions
40  * are met:
41  * 1. Redistributions of source code must retain the above copyright
42  *    notice, this list of conditions and the following disclaimer.
43  * 2. Redistributions in binary form must reproduce the above copyright
44  *    notice, this list of conditions and the following disclaimer in the
45  *    documentation and/or other materials provided with the distribution.
46  * 3. Neither the name of the University nor the names of its contributors
47  *    may be used to endorse or promote products derived from this software
48  *    without specific prior written permission.
49  *
50  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
51  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
52  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
53  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
54  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
55  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
56  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
57  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
58  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
59  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
60  * SUCH DAMAGE.
61  *
62  *	@(#)init_main.c	8.16 (Berkeley) 5/14/95
63  */
64 
65 /*
66  * Copyright (c) 1995 Christopher G. Demetriou.  All rights reserved.
67  *
68  * Redistribution and use in source and binary forms, with or without
69  * modification, are permitted provided that the following conditions
70  * are met:
71  * 1. Redistributions of source code must retain the above copyright
72  *    notice, this list of conditions and the following disclaimer.
73  * 2. Redistributions in binary form must reproduce the above copyright
74  *    notice, this list of conditions and the following disclaimer in the
75  *    documentation and/or other materials provided with the distribution.
76  * 3. All advertising materials mentioning features or use of this software
77  *    must display the following acknowledgement:
78  *	This product includes software developed by the University of
79  *	California, Berkeley and its contributors.
80  * 4. Neither the name of the University nor the names of its contributors
81  *    may be used to endorse or promote products derived from this software
82  *    without specific prior written permission.
83  *
84  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
85  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
86  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
87  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
88  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
89  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
90  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
91  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
92  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
93  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
94  * SUCH DAMAGE.
95  *
96  *	@(#)init_main.c	8.16 (Berkeley) 5/14/95
97  */
98 
99 #include <sys/cdefs.h>
100 __KERNEL_RCSID(0, "$NetBSD: init_main.c,v 1.412 2009/12/17 01:25:10 rmind Exp $");
101 
102 #include "opt_ddb.h"
103 #include "opt_ipsec.h"
104 #include "opt_modular.h"
105 #include "opt_ntp.h"
106 #include "opt_pipe.h"
107 #include "opt_sa.h"
108 #include "opt_syscall_debug.h"
109 #include "opt_sysv.h"
110 #include "opt_fileassoc.h"
111 #include "opt_ktrace.h"
112 #include "opt_pax.h"
113 #include "opt_compat_netbsd.h"
114 #include "opt_wapbl.h"
115 #include "opt_ptrace.h"
116 
117 #include "drvctl.h"
118 #include "ksyms.h"
119 #include "rnd.h"
120 #include "sysmon_envsys.h"
121 #include "sysmon_power.h"
122 #include "sysmon_taskq.h"
123 #include "sysmon_wdog.h"
124 #include "veriexec.h"
125 
126 #include <sys/param.h>
127 #include <sys/acct.h>
128 #include <sys/filedesc.h>
129 #include <sys/file.h>
130 #include <sys/errno.h>
131 #include <sys/callout.h>
132 #include <sys/cpu.h>
133 #include <sys/spldebug.h>
134 #include <sys/kernel.h>
135 #include <sys/mount.h>
136 #include <sys/proc.h>
137 #include <sys/kthread.h>
138 #include <sys/resourcevar.h>
139 #include <sys/signalvar.h>
140 #include <sys/systm.h>
141 #include <sys/vnode.h>
142 #include <sys/fstrans.h>
143 #include <sys/tty.h>
144 #include <sys/conf.h>
145 #include <sys/disklabel.h>
146 #include <sys/buf.h>
147 #include <sys/device.h>
148 #include <sys/exec.h>
149 #include <sys/socketvar.h>
150 #include <sys/protosw.h>
151 #include <sys/percpu.h>
152 #include <sys/pset.h>
153 #include <sys/sysctl.h>
154 #include <sys/reboot.h>
155 #include <sys/sysctl.h>
156 #include <sys/event.h>
157 #include <sys/mbuf.h>
158 #include <sys/sched.h>
159 #include <sys/sleepq.h>
160 #include <sys/iostat.h>
161 #include <sys/vmem.h>
162 #include <sys/uuid.h>
163 #include <sys/extent.h>
164 #include <sys/disk.h>
165 #include <sys/msgbuf.h>
166 #include <sys/module.h>
167 #include <sys/event.h>
168 #include <sys/lockf.h>
169 #include <sys/once.h>
170 #include <sys/ksyms.h>
171 #include <sys/uidinfo.h>
172 #include <sys/kprintf.h>
173 #ifdef FAST_IPSEC
174 #include <netipsec/ipsec.h>
175 #endif
176 #ifdef SYSVSHM
177 #include <sys/shm.h>
178 #endif
179 #ifdef SYSVSEM
180 #include <sys/sem.h>
181 #endif
182 #ifdef SYSVMSG
183 #include <sys/msg.h>
184 #endif
185 #include <sys/domain.h>
186 #include <sys/namei.h>
187 #if NRND > 0
188 #include <sys/rnd.h>
189 #endif
190 #include <sys/pipe.h>
191 #if NVERIEXEC > 0
192 #include <sys/verified_exec.h>
193 #endif /* NVERIEXEC > 0 */
194 #ifdef KTRACE
195 #include <sys/ktrace.h>
196 #endif
197 #include <sys/kauth.h>
198 #ifdef WAPBL
199 #include <sys/wapbl.h>
200 #endif
201 #ifdef KERN_SA
202 #include <sys/savar.h>
203 #endif
204 #include <net80211/ieee80211_netbsd.h>
205 #ifdef PTRACE
206 #include <sys/ptrace.h>
207 #endif /* PTRACE */
208 
209 #include <sys/syscall.h>
210 #include <sys/syscallargs.h>
211 
212 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR)
213 #include <sys/pax.h>
214 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */
215 
216 #include <ufs/ufs/quota.h>
217 
218 #include <miscfs/genfs/genfs.h>
219 #include <miscfs/syncfs/syncfs.h>
220 #include <miscfs/specfs/specdev.h>
221 
222 #include <sys/cpu.h>
223 
224 #include <uvm/uvm.h>
225 
226 #if NSYSMON_TASKQ > 0
227 #include <dev/sysmon/sysmon_taskq.h>
228 #endif
229 
230 #include <dev/cons.h>
231 
232 #if NSYSMON_ENVSYS > 0 || NSYSMON_POWER > 0 || NSYSMON_WDOG > 0
233 #include <dev/sysmon/sysmonvar.h>
234 #endif
235 
236 #include <net/if.h>
237 #include <net/raw_cb.h>
238 
239 #include <prop/proplib.h>
240 
241 #ifdef COMPAT_50
242 #include <compat/sys/time.h>
243 struct timeval50 boottime50;
244 #endif
245 
246 #ifdef _KERNEL_OPT
247 #include "opt_userconf.h"
248 #endif
249 #ifdef USERCONF
250 #include <sys/userconf.h>
251 #endif
252 
253 extern struct proc proc0;
254 extern struct lwp lwp0;
255 extern time_t rootfstime;
256 
257 #ifndef curlwp
258 struct	lwp *curlwp = &lwp0;
259 #endif
260 struct	proc *initproc;
261 
262 struct	vnode *rootvp, *swapdev_vp;
263 int	boothowto;
264 int	cold = 1;			/* still working on startup */
265 struct timespec boottime;	        /* time at system startup - will only follow settime deltas */
266 
267 int	start_init_exec;		/* semaphore for start_init() */
268 
269 static void check_console(struct lwp *l);
270 static void start_init(void *);
271 static void configure(void);
272 static void configure2(void);
273 void main(void);
274 
275 /*
276  * System startup; initialize the world, create process 0, mount root
277  * filesystem, and fork to create init and pagedaemon.  Most of the
278  * hard work is done in the lower-level initialization routines including
279  * startup(), which does memory initialization and autoconfiguration.
280  */
281 void
282 main(void)
283 {
284 	struct timespec time;
285 	struct lwp *l;
286 	struct proc *p;
287 	int s, error;
288 #ifdef NVNODE_IMPLICIT
289 	int usevnodes;
290 #endif
291 	CPU_INFO_ITERATOR cii;
292 	struct cpu_info *ci;
293 
294 	l = &lwp0;
295 #ifndef LWP0_CPU_INFO
296 	l->l_cpu = curcpu();
297 #endif
298 	l->l_pflag |= LP_RUNNING;
299 
300 	/*
301 	 * Attempt to find console and initialize
302 	 * in case of early panic or other messages.
303 	 */
304 	consinit();
305 
306 	kernel_lock_init();
307 	once_init();
308 	mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE);
309 
310 	/* Initialize the device switch tables. */
311 	devsw_init();
312 
313 	uvm_init();
314 
315 	prop_kern_init();
316 
317 #if ((NKSYMS > 0) || (NDDB > 0) || (NMODULAR > 0))
318 	ksyms_init();
319 #endif
320 	kprintf_init();
321 
322 	percpu_init();
323 
324 	/* Initialize lock caches. */
325 	mutex_obj_init();
326 	rw_obj_init();
327 
328 	/* Initialize the extent manager. */
329 	extent_init();
330 
331 	/* Do machine-dependent initialization. */
332 	cpu_startup();
333 
334 	/* Initialize the sysctl subsystem. */
335 	sysctl_init();
336 
337 	/* Initialize callouts, part 1. */
338 	callout_startup();
339 
340 	/* Initialize the kernel authorization subsystem. */
341 	kauth_init();
342 
343 	spec_init();
344 
345 	/* Start module system. */
346 	module_init();
347 
348 	/*
349 	 * Initialize the kernel authorization subsystem and start the
350 	 * default security model, if any. We need to do this early
351 	 * enough so that subsystems relying on any of the aforementioned
352 	 * can work properly. Since the security model may dictate the
353 	 * credential inheritance policy, it is needed at least before
354 	 * any process is created, specifically proc0.
355 	 */
356 	module_init_class(MODULE_CLASS_SECMODEL);
357 
358 	/* Initialize the buffer cache */
359 	bufinit();
360 
361 	/* Initialize sockets. */
362 	soinit();
363 
364 	/*
365 	 * The following things must be done before autoconfiguration.
366 	 */
367 	evcnt_init();		/* initialize event counters */
368 #if NRND > 0
369 	rnd_init();		/* initialize random number generator */
370 #endif
371 
372 	/* Initialize process and pgrp structures. */
373 #ifdef KERN_SA
374 	sa_init();
375 #endif
376 	procinit();
377 	lwpinit();
378 
379 	/* Initialize signal-related data structures. */
380 	signal_init();
381 
382 	/* Initialize resource management. */
383 	resource_init();
384 
385 	/* Create process 0. */
386 	proc0_init();
387 
388 	/* Disable preemption during boot. */
389 	kpreempt_disable();
390 
391 	/* Initialize the UID hash table. */
392 	uid_init();
393 
394 	/* Charge root for one process. */
395 	(void)chgproccnt(0, 1);
396 
397 	/* Initialize timekeeping. */
398 	time_init();
399 
400 	/* Initialize the run queues, turnstiles and sleep queues. */
401 	sched_rqinit();
402 	turnstile_init();
403 	sleeptab_init(&sleeptab);
404 
405 	sched_init();
406 
407 	/* Initialize processor-sets */
408 	psets_init();
409 
410 	/* MI initialization of the boot cpu */
411 	error = mi_cpu_attach(curcpu());
412 	KASSERT(error == 0);
413 
414 	/* Initialize timekeeping, part 2. */
415 	time_init2();
416 
417 	/*
418 	 * Initialize mbuf's.  Do this now because we might attempt to
419 	 * allocate mbufs or mbuf clusters during autoconfiguration.
420 	 */
421 	mbinit();
422 
423 	/* Initialize I/O statistics. */
424 	iostat_init();
425 
426 	/* Initialize the log device. */
427 	loginit();
428 
429 	/* Second part of module system initialization. */
430 	module_init2();
431 
432 	/* Initialize the file systems. */
433 #ifdef NVNODE_IMPLICIT
434 	/*
435 	 * If maximum number of vnodes in namei vnode cache is not explicitly
436 	 * defined in kernel config, adjust the number such as we use roughly
437 	 * 10% of memory for vnodes and associated data structures in the
438 	 * assumed worst case.  Do not provide fewer than NVNODE vnodes.
439 	 */
440 	usevnodes =
441 	    calc_cache_size(kernel_map, 10, VNODE_VA_MAXPCT) / VNODE_COST;
442 	if (usevnodes > desiredvnodes)
443 		desiredvnodes = usevnodes;
444 #endif
445 	vfsinit();
446 	lf_init();
447 
448 	/* Initialize fstrans. */
449 	fstrans_init();
450 
451 	/* Initialize the file descriptor system. */
452 	fd_sys_init();
453 
454 	/* Initialize cwd structures */
455 	cwd_sys_init();
456 
457 	/* Initialize kqueue. */
458 	kqueue_init();
459 
460 	/* Initialize the system monitor subsystems. */
461 #if NSYSMON_TASKQ > 0
462 	sysmon_task_queue_preinit();
463 #endif
464 
465 #if NSYSMON_ENVSYS > 0
466 	sysmon_envsys_init();
467 #endif
468 
469 #if NSYSMON_POWER > 0
470 	sysmon_power_init();
471 #endif
472 
473 #if NSYSMON_WDOG > 0
474 	sysmon_wdog_init();
475 #endif
476 
477 	inittimecounter();
478 	ntp_init();
479 
480 	/* Initialize tty subsystem. */
481 	tty_init();
482 	ttyldisc_init();
483 
484 	/* Initialize the buffer cache, part 2. */
485 	bufinit2();
486 
487 	/* Initialize the disk wedge subsystem. */
488 	dkwedge_init();
489 
490 	/* Initialize interfaces. */
491 	ifinit1();
492 
493 	spldebug_start();
494 
495 	/* Configure the system hardware.  This will enable interrupts. */
496 	configure();
497 
498 	ssp_init();
499 
500 	configure2();
501 	/* Now timer is working.  Enable preemption. */
502 	kpreempt_enable();
503 
504 	ubc_init();		/* must be after autoconfig */
505 
506 #ifdef SYSVSHM
507 	/* Initialize System V style shared memory. */
508 	shminit();
509 #endif
510 
511 #ifdef SYSVSEM
512 	/* Initialize System V style semaphores. */
513 	seminit();
514 #endif
515 
516 #ifdef SYSVMSG
517 	/* Initialize System V style message queues. */
518 	msginit();
519 #endif
520 
521 #if NVERIEXEC > 0
522 	/*
523 	 * Initialise the Veriexec subsystem.
524 	 */
525 	veriexec_init();
526 #endif /* NVERIEXEC > 0 */
527 
528 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR)
529 	pax_init();
530 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */
531 
532 #ifdef	FAST_IPSEC
533 	/* Attach network crypto subsystem */
534 	ipsec_attach();
535 #endif
536 
537 	/*
538 	 * Initialize protocols.  Block reception of incoming packets
539 	 * until everything is ready.
540 	 */
541 	s = splnet();
542 	ifinit();
543 	domaininit(true);
544 	if_attachdomain();
545 	splx(s);
546 
547 #ifdef GPROF
548 	/* Initialize kernel profiling. */
549 	kmstartup();
550 #endif
551 
552 	/* Initialize system accounting. */
553 	acct_init();
554 
555 #ifndef PIPE_SOCKETPAIR
556 	/* Initialize pipes. */
557 	pipe_init();
558 #endif
559 
560 #ifdef KTRACE
561 	/* Initialize ktrace. */
562 	ktrinit();
563 #endif
564 
565 #ifdef PTRACE
566 	/* Initialize ptrace. */
567 	ptrace_init();
568 #endif /* PTRACE */
569 
570 	/* Initialize the UUID system calls. */
571 	uuid_init();
572 
573 #ifdef WAPBL
574 	/* Initialize write-ahead physical block logging. */
575 	wapbl_init();
576 #endif
577 
578 	machdep_init();
579 
580 	/*
581 	 * Create process 1 (init(8)).  We do this now, as Unix has
582 	 * historically had init be process 1, and changing this would
583 	 * probably upset a lot of people.
584 	 *
585 	 * Note that process 1 won't immediately exec init(8), but will
586 	 * wait for us to inform it that the root file system has been
587 	 * mounted.
588 	 */
589 	if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc))
590 		panic("fork init");
591 
592 	/*
593 	 * Load any remaining builtin modules, and hand back temporary
594 	 * storage to the VM system.
595 	 */
596 	module_init_class(MODULE_CLASS_ANY);
597 
598 	/*
599 	 * Finalize configuration now that all real devices have been
600 	 * found.  This needs to be done before the root device is
601 	 * selected, since finalization may create the root device.
602 	 */
603 	config_finalize();
604 
605 	sysctl_finalize();
606 
607 	/*
608 	 * Now that autoconfiguration has completed, we can determine
609 	 * the root and dump devices.
610 	 */
611 	cpu_rootconf();
612 	cpu_dumpconf();
613 
614 	/* Mount the root file system. */
615 	do {
616 		domountroothook();
617 		if ((error = vfs_mountroot())) {
618 			printf("cannot mount root, error = %d\n", error);
619 			boothowto |= RB_ASKNAME;
620 			setroot(root_device,
621 			    (rootdev != NODEV) ? DISKPART(rootdev) : 0);
622 		}
623 	} while (error != 0);
624 	mountroothook_destroy();
625 
626 	/*
627 	 * Initialise the time-of-day clock, passing the time recorded
628 	 * in the root filesystem (if any) for use by systems that
629 	 * don't have a non-volatile time-of-day device.
630 	 */
631 	inittodr(rootfstime);
632 
633 	/*
634 	 * Now can look at time, having had a chance to verify the time
635 	 * from the file system.  Reset l->l_rtime as it may have been
636 	 * munched in mi_switch() after the time got set.
637 	 */
638 	getnanotime(&time);
639 	boottime = time;
640 #ifdef COMPAT_50
641 	{
642 		struct timeval tv;
643 		TIMESPEC_TO_TIMEVAL(&tv, &time);
644 		timeval_to_timeval50(&tv, &boottime50);
645 	}
646 #endif
647 	mutex_enter(proc_lock);
648 	LIST_FOREACH(p, &allproc, p_list) {
649 		KASSERT((p->p_flag & PK_MARKER) == 0);
650 		mutex_enter(p->p_lock);
651 		TIMESPEC_TO_TIMEVAL(&p->p_stats->p_start, &time);
652 		LIST_FOREACH(l, &p->p_lwps, l_sibling) {
653 			lwp_lock(l);
654 			memset(&l->l_rtime, 0, sizeof(l->l_rtime));
655 			lwp_unlock(l);
656 		}
657 		mutex_exit(p->p_lock);
658 	}
659 	mutex_exit(proc_lock);
660 	binuptime(&curlwp->l_stime);
661 
662 	for (CPU_INFO_FOREACH(cii, ci)) {
663 		ci->ci_schedstate.spc_lastmod = time_second;
664 	}
665 
666 	/* Create the pageout daemon kernel thread. */
667 	uvm_swap_init();
668 	if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout,
669 	    NULL, NULL, "pgdaemon"))
670 		panic("fork pagedaemon");
671 
672 	/* Create the filesystem syncer kernel thread. */
673 	if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync,
674 	    NULL, NULL, "ioflush"))
675 		panic("fork syncer");
676 
677 	/* Create the aiodone daemon kernel thread. */
678 	if (workqueue_create(&uvm.aiodone_queue, "aiodoned",
679 	    uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE))
680 		panic("fork aiodoned");
681 
682 	vmem_rehash_start();
683 
684 	/* Initialize exec structures */
685 	exec_init(1);
686 
687 	/*
688 	 * Okay, now we can let init(8) exec!  It's off to userland!
689 	 */
690 	mutex_enter(proc_lock);
691 	start_init_exec = 1;
692 	cv_broadcast(&lbolt);
693 	mutex_exit(proc_lock);
694 
695 	/* The scheduler is an infinite loop. */
696 	uvm_scheduler();
697 	/* NOTREACHED */
698 }
699 
700 /*
701  * Configure the system's hardware.
702  */
703 static void
704 configure(void)
705 {
706 
707 	/* Initialize autoconf data structures. */
708 	config_init_mi();
709 	/*
710 	 * XXX
711 	 * callout_setfunc() requires mutex(9) so it can't be in config_init()
712 	 * on amiga and atari which use config_init() and autoconf(9) fucntions
713 	 * to initialize console devices.
714 	 */
715 	config_twiddle_init();
716 
717 	pmf_init();
718 #if NDRVCTL > 0
719 	drvctl_init();
720 #endif
721 
722 #ifdef USERCONF
723 	if (boothowto & RB_USERCONF)
724 		user_config();
725 #endif
726 
727 	if ((boothowto & (AB_SILENT|AB_VERBOSE)) == AB_SILENT) {
728 		printf_nolog("Detecting hardware...");
729 	}
730 
731 	/*
732 	 * Do the machine-dependent portion of autoconfiguration.  This
733 	 * sets the configuration machinery here in motion by "finding"
734 	 * the root bus.  When this function returns, we expect interrupts
735 	 * to be enabled.
736 	 */
737 	cpu_configure();
738 }
739 
740 static void
741 configure2(void)
742 {
743 	CPU_INFO_ITERATOR cii;
744 	struct cpu_info *ci;
745 	int s;
746 
747 	/*
748 	 * Now that we've found all the hardware, start the real time
749 	 * and statistics clocks.
750 	 */
751 	initclocks();
752 
753 	cold = 0;	/* clocks are running, we're warm now! */
754 	s = splsched();
755 	curcpu()->ci_schedstate.spc_flags |= SPCF_RUNNING;
756 	splx(s);
757 
758 	/* Boot the secondary processors. */
759 	for (CPU_INFO_FOREACH(cii, ci)) {
760 		uvm_cpu_attach(ci);
761 	}
762 	mp_online = true;
763 #if defined(MULTIPROCESSOR)
764 	cpu_boot_secondary_processors();
765 #endif
766 
767 	/* Setup the runqueues and scheduler. */
768 	runq_init();
769 	synch_init();
770 
771 	/*
772 	 * Bus scans can make it appear as if the system has paused, so
773 	 * twiddle constantly while config_interrupts() jobs are running.
774 	 */
775 	config_twiddle_fn(NULL);
776 
777 	/*
778 	 * Create threads to call back and finish configuration for
779 	 * devices that want interrupts enabled.
780 	 */
781 	config_create_interruptthreads();
782 
783 	/* Get the threads going and into any sleeps before continuing. */
784 	yield();
785 }
786 
787 static void
788 check_console(struct lwp *l)
789 {
790 	struct vnode *vp;
791 	int error;
792 
793 	error = namei_simple_kernel("/dev/console",
794 				NSM_FOLLOW_NOEMULROOT, &vp);
795 	if (error == 0)
796 		vrele(vp);
797 	else if (error == ENOENT)
798 		printf("warning: no /dev/console\n");
799 	else
800 		printf("warning: lookup /dev/console: error %d\n", error);
801 }
802 
803 /*
804  * List of paths to try when searching for "init".
805  */
806 static const char * const initpaths[] = {
807 	"/sbin/init",
808 	"/sbin/oinit",
809 	"/sbin/init.bak",
810 	NULL,
811 };
812 
813 /*
814  * Start the initial user process; try exec'ing each pathname in "initpaths".
815  * The program is invoked with one argument containing the boot flags.
816  */
817 static void
818 start_init(void *arg)
819 {
820 	struct lwp *l = arg;
821 	struct proc *p = l->l_proc;
822 	vaddr_t addr;
823 	struct sys_execve_args /* {
824 		syscallarg(const char *) path;
825 		syscallarg(char * const *) argp;
826 		syscallarg(char * const *) envp;
827 	} */ args;
828 	int options, i, error;
829 	register_t retval[2];
830 	char flags[4], *flagsp;
831 	const char *path, *slash;
832 	char *ucp, **uap, *arg0, *arg1 = NULL;
833 	char ipath[129];
834 	int ipx, len;
835 
836 	/*
837 	 * Now in process 1.
838 	 */
839 	strncpy(p->p_comm, "init", MAXCOMLEN);
840 
841 	/*
842 	 * Wait for main() to tell us that it's safe to exec.
843 	 */
844 	mutex_enter(proc_lock);
845 	while (start_init_exec == 0)
846 		cv_wait(&lbolt, proc_lock);
847 	mutex_exit(proc_lock);
848 
849 	/*
850 	 * This is not the right way to do this.  We really should
851 	 * hand-craft a descriptor onto /dev/console to hand to init,
852 	 * but that's a _lot_ more work, and the benefit from this easy
853 	 * hack makes up for the "good is the enemy of the best" effect.
854 	 */
855 	check_console(l);
856 
857 	/*
858 	 * Need just enough stack to hold the faked-up "execve()" arguments.
859 	 */
860 	addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE);
861 	if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE,
862                     NULL, UVM_UNKNOWN_OFFSET, 0,
863                     UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY,
864 		    UVM_ADV_NORMAL,
865                     UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0)
866 		panic("init: couldn't allocate argument space");
867 	p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE);
868 
869 	ipx = 0;
870 	while (1) {
871 		if (boothowto & RB_ASKNAME) {
872 			printf("init path");
873 			if (initpaths[ipx])
874 				printf(" (default %s)", initpaths[ipx]);
875 			printf(": ");
876 			len = cngetsn(ipath, sizeof(ipath)-1);
877 			if (len == 4 && strcmp(ipath, "halt") == 0) {
878 				cpu_reboot(RB_HALT, NULL);
879 			} else if (len == 6 && strcmp(ipath, "reboot") == 0) {
880 				cpu_reboot(0, NULL);
881 #if defined(DDB)
882 			} else if (len == 3 && strcmp(ipath, "ddb") == 0) {
883 				console_debugger();
884 				continue;
885 #endif
886 			} else if (len > 0 && ipath[0] == '/') {
887 				ipath[len] = '\0';
888 				path = ipath;
889 			} else if (len == 0 && initpaths[ipx] != NULL) {
890 				path = initpaths[ipx++];
891 			} else {
892 				printf("use absolute path, ");
893 #if defined(DDB)
894 				printf("\"ddb\", ");
895 #endif
896 				printf("\"halt\", or \"reboot\"\n");
897 				continue;
898 			}
899 		} else {
900 			if ((path = initpaths[ipx++]) == NULL) {
901 				ipx = 0;
902 				boothowto |= RB_ASKNAME;
903 				continue;
904 			}
905 		}
906 
907 		ucp = (char *)USRSTACK;
908 
909 		/*
910 		 * Construct the boot flag argument.
911 		 */
912 		flagsp = flags;
913 		*flagsp++ = '-';
914 		options = 0;
915 
916 		if (boothowto & RB_SINGLE) {
917 			*flagsp++ = 's';
918 			options = 1;
919 		}
920 #ifdef notyet
921 		if (boothowto & RB_FASTBOOT) {
922 			*flagsp++ = 'f';
923 			options = 1;
924 		}
925 #endif
926 
927 		/*
928 		 * Move out the flags (arg 1), if necessary.
929 		 */
930 		if (options != 0) {
931 			*flagsp++ = '\0';
932 			i = flagsp - flags;
933 #ifdef DEBUG
934 			printf("init: copying out flags `%s' %d\n", flags, i);
935 #endif
936 			arg1 = STACK_ALLOC(ucp, i);
937 			ucp = STACK_MAX(arg1, i);
938 			(void)copyout((void *)flags, arg1, i);
939 		}
940 
941 		/*
942 		 * Move out the file name (also arg 0).
943 		 */
944 		i = strlen(path) + 1;
945 #ifdef DEBUG
946 		printf("init: copying out path `%s' %d\n", path, i);
947 #else
948 		if (boothowto & RB_ASKNAME || path != initpaths[0])
949 			printf("init: trying %s\n", path);
950 #endif
951 		arg0 = STACK_ALLOC(ucp, i);
952 		ucp = STACK_MAX(arg0, i);
953 		(void)copyout(path, arg0, i);
954 
955 		/*
956 		 * Move out the arg pointers.
957 		 */
958 		ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES);
959 		uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3);
960 		SCARG(&args, path) = arg0;
961 		SCARG(&args, argp) = uap;
962 		SCARG(&args, envp) = NULL;
963 		slash = strrchr(path, '/');
964 		if (slash)
965 			(void)suword((void *)uap++,
966 			    (long)arg0 + (slash + 1 - path));
967 		else
968 			(void)suword((void *)uap++, (long)arg0);
969 		if (options != 0)
970 			(void)suword((void *)uap++, (long)arg1);
971 		(void)suword((void *)uap++, 0);	/* terminator */
972 
973 		/*
974 		 * Now try to exec the program.  If can't for any reason
975 		 * other than it doesn't exist, complain.
976 		 */
977 		error = sys_execve(l, &args, retval);
978 		if (error == 0 || error == EJUSTRETURN) {
979 			KERNEL_UNLOCK_LAST(l);
980 			return;
981 		}
982 		printf("exec %s: error %d\n", path, error);
983 	}
984 	printf("init: not found\n");
985 	panic("no init");
986 }
987 
988 /*
989  * calculate cache size from physmem and vm_map size.
990  */
991 vaddr_t
992 calc_cache_size(struct vm_map *map, int pct, int va_pct)
993 {
994 	paddr_t t;
995 
996 	/* XXX should consider competing cache if any */
997 	/* XXX should consider submaps */
998 	t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE;
999 	if (map != NULL) {
1000 		vsize_t vsize;
1001 
1002 		vsize = vm_map_max(map) - vm_map_min(map);
1003 		vsize = (uintmax_t)vsize * va_pct / 100;
1004 		if (t > vsize) {
1005 			t = vsize;
1006 		}
1007 	}
1008 	return t;
1009 }
1010 
1011 /*
1012  * Print the system start up banner.
1013  *
1014  * - Print a limited banner if AB_SILENT.
1015  * - Always send normal banner to the log.
1016  */
1017 #define MEM_PBUFSIZE	sizeof("99999 MB")
1018 
1019 void
1020 banner(void)
1021 {
1022 	static char notice[] = " Notice: this software is "
1023 	    "protected by copyright";
1024 	char pbuf[81];
1025 	void (*pr)(const char *, ...);
1026 	int i;
1027 
1028 	if ((boothowto & AB_SILENT) != 0) {
1029 		snprintf(pbuf, sizeof(pbuf), "%s %s (%s)",
1030 		    ostype, osrelease, kernel_ident);
1031 		printf_nolog("%s", pbuf);
1032 		for (i = 80 - strlen(pbuf) - sizeof(notice); i > 0; i--)
1033 			printf(" ");
1034 		printf_nolog("%s\n", notice);
1035 		pr = aprint_normal;
1036 	} else {
1037 		pr = printf;
1038 	}
1039 
1040 	memset(pbuf, 0, sizeof(pbuf));
1041 	(*pr)("%s%s", copyright, version);
1042 	format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)physmem));
1043 	(*pr)("total memory = %s\n", pbuf);
1044 	format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)uvmexp.free));
1045 	(*pr)("avail memory = %s\n", pbuf);
1046 }
1047