xref: /openbsd-src/sys/kern/init_main.c (revision b2ea75c1b17e1a9a339660e7ed45cd24946b230e)
1 /*	$OpenBSD: init_main.c,v 1.73 2001/07/27 09:55:07 niklas Exp $	*/
2 /*	$NetBSD: init_main.c,v 1.84.4.1 1996/06/02 09:08:06 mrg Exp $	*/
3 
4 /*
5  * Copyright (c) 1995 Christopher G. Demetriou.  All rights reserved.
6  * Copyright (c) 1982, 1986, 1989, 1991, 1992, 1993
7  *	The Regents of the University of California.  All rights reserved.
8  * (c) UNIX System Laboratories, Inc.
9  * All or some portions of this file are derived from material licensed
10  * to the University of California by American Telephone and Telegraph
11  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
12  * the permission of UNIX System Laboratories, Inc.
13  *
14  * Redistribution and use in source and binary forms, with or without
15  * modification, are permitted provided that the following conditions
16  * are met:
17  * 1. Redistributions of source code must retain the above copyright
18  *    notice, this list of conditions and the following disclaimer.
19  * 2. Redistributions in binary form must reproduce the above copyright
20  *    notice, this list of conditions and the following disclaimer in the
21  *    documentation and/or other materials provided with the distribution.
22  * 3. All advertising materials mentioning features or use of this software
23  *    must display the following acknowledgement:
24  *	This product includes software developed by the University of
25  *	California, Berkeley and its contributors.
26  * 4. Neither the name of the University nor the names of its contributors
27  *    may be used to endorse or promote products derived from this software
28  *    without specific prior written permission.
29  *
30  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
31  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
32  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
33  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
34  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
35  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
36  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
37  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
38  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
39  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
40  * SUCH DAMAGE.
41  *
42  *	@(#)init_main.c	8.9 (Berkeley) 1/21/94
43  */
44 
45 #include <sys/param.h>
46 #include <sys/filedesc.h>
47 #include <sys/file.h>
48 #include <sys/errno.h>
49 #include <sys/exec.h>
50 #include <sys/kernel.h>
51 #include <sys/kthread.h>
52 #include <sys/mount.h>
53 #include <sys/map.h>
54 #include <sys/proc.h>
55 #include <sys/resourcevar.h>
56 #include <sys/signalvar.h>
57 #include <sys/systm.h>
58 #include <sys/namei.h>
59 #include <sys/vnode.h>
60 #include <sys/tty.h>
61 #include <sys/conf.h>
62 #include <sys/buf.h>
63 #include <sys/device.h>
64 #include <sys/protosw.h>
65 #include <sys/reboot.h>
66 #include <sys/user.h>
67 #ifdef SYSVSHM
68 #include <sys/shm.h>
69 #endif
70 #ifdef SYSVSEM
71 #include <sys/sem.h>
72 #endif
73 #ifdef SYSVMSG
74 #include <sys/msg.h>
75 #endif
76 #include <sys/domain.h>
77 #include <sys/mbuf.h>
78 #include <sys/pipe.h>
79 
80 #include <sys/syscall.h>
81 #include <sys/syscallargs.h>
82 
83 #include <ufs/ufs/quota.h>
84 
85 #include <machine/cpu.h>
86 
87 #include <vm/vm.h>
88 
89 #include <uvm/uvm.h>
90 
91 #include <net/if.h>
92 #include <net/raw_cb.h>
93 
94 #if defined(CRYPTO)
95 #include <crypto/cryptodev.h>
96 #include <crypto/cryptosoft.h>
97 #endif
98 
99 #if defined(NFSSERVER) || defined(NFSCLIENT)
100 extern void nfs_init __P((void));
101 #endif
102 
103 char	copyright[] =
104 "Copyright (c) 1982, 1986, 1989, 1991, 1993\n"
105 "\tThe Regents of the University of California.  All rights reserved.\n"
106 "Copyright (c) 1995-2001 OpenBSD. All rights reserved.  http://www.OpenBSD.org\n";
107 
108 /* Components of the first process -- never freed. */
109 struct	session session0;
110 struct	pgrp pgrp0;
111 struct	proc proc0;
112 struct	pcred cred0;
113 struct	filedesc0 filedesc0;
114 struct	plimit limit0;
115 struct	vmspace vmspace0;
116 struct	sigacts sigacts0;
117 #ifndef curproc
118 struct	proc *curproc;
119 #endif
120 struct	proc *initproc;
121 
122 int	cmask = CMASK;
123 extern	struct user *proc0paddr;
124 
125 void	(*md_diskconf) __P((void)) = NULL;
126 struct	vnode *rootvp, *swapdev_vp;
127 int	boothowto;
128 struct	timeval boottime;
129 struct	timeval runtime;
130 
131 /* XXX return int so gcc -Werror won't complain */
132 int	main __P((void *));
133 void	check_console __P((struct proc *));
134 void	start_init __P((void *));
135 void	start_pagedaemon __P((void *));
136 void	start_update __P((void *));
137 void	start_reaper __P((void *));
138 void    start_crypto __P((void *));
139 
140 extern char sigcode[], esigcode[];
141 #ifdef SYSCALL_DEBUG
142 extern char *syscallnames[];
143 #endif
144 
145 struct emul emul_native = {
146 	"native",
147 	NULL,
148 	sendsig,
149 	SYS_syscall,
150 	SYS_MAXSYSCALL,
151 	sysent,
152 #ifdef SYSCALL_DEBUG
153 	syscallnames,
154 #else
155 	NULL,
156 #endif
157 	0,
158 	copyargs,
159 	setregs,
160 	NULL,
161 	sigcode,
162 	esigcode,
163 };
164 
165 /*
166  * System startup; initialize the world, create process 0, mount root
167  * filesystem, and fork to create init and pagedaemon.  Most of the
168  * hard work is done in the lower-level initialization routines including
169  * startup(), which does memory initialization and autoconfiguration.
170  */
171 /* XXX return int, so gcc -Werror won't complain */
172 int
173 main(framep)
174 	void *framep;				/* XXX should go away */
175 {
176 	register struct proc *p;
177 	register struct pdevinit *pdev;
178 	struct timeval rtv;
179 	register int i;
180 	int s;
181 	register_t rval[2];
182 	extern struct pdevinit pdevinit[];
183 	extern void scheduler_start __P((void));
184 	extern void disk_init __P((void));
185 	extern void endtsleep __P((void *));
186 	extern void realitexpire __P((void *));
187 
188 	/*
189 	 * Initialize the current process pointer (curproc) before
190 	 * any possible traps/probes to simplify trap processing.
191 	 */
192 	p = &proc0;
193 	curproc = p;
194 
195 	/*
196 	 * Attempt to find console and initialize
197 	 * in case of early panic or other messages.
198 	 */
199 	config_init();		/* init autoconfiguration data structures */
200 	consinit();
201 	printf(copyright);
202 	printf("\n");
203 
204 	uvm_init();
205 	disk_init();		/* must come before autoconfiguration */
206 	tty_init();		/* initialise tty's */
207 	cpu_startup();
208 
209 	/*
210 	 * Initialize mbuf's.  Do this now because we might attempt to
211 	 * allocate mbufs or mbuf clusters during autoconfiguration.
212 	 */
213 	mbinit();
214 
215 	cpu_configure();
216 
217 	/* Initialize sysctls (must be done before any processes run) */
218 	sysctl_init();
219 
220 	/*
221 	 * Initialize process and pgrp structures.
222 	 */
223 	procinit();
224 
225 	/*
226 	 * Initialize filedescriptors.
227 	 */
228 	filedesc_init();
229 
230 	/*
231 	 * Initialize pipes.
232 	 */
233 	pipe_init();
234 
235 	/*
236 	 * Create process 0 (the swapper).
237 	 */
238 	LIST_INSERT_HEAD(&allproc, p, p_list);
239 	p->p_pgrp = &pgrp0;
240 	LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
241 	LIST_INIT(&pgrp0.pg_members);
242 	LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
243 
244 	pgrp0.pg_session = &session0;
245 	session0.s_count = 1;
246 	session0.s_leader = p;
247 
248 	p->p_flag = P_INMEM | P_SYSTEM | P_NOCLDWAIT;
249 	p->p_stat = SRUN;
250 	p->p_nice = NZERO;
251 	p->p_emul = &emul_native;
252 	bcopy("swapper", p->p_comm, sizeof ("swapper"));
253 
254 	/* Init timeouts. */
255 	timeout_set(&p->p_sleep_to, endtsleep, p);
256 	timeout_set(&p->p_realit_to, realitexpire, p);
257 
258 	/* Create credentials. */
259 	cred0.p_refcnt = 1;
260 	p->p_cred = &cred0;
261 	p->p_ucred = crget();
262 	p->p_ucred->cr_ngroups = 1;	/* group 0 */
263 
264 	/* Create the file descriptor table. */
265 	p->p_fd = &filedesc0.fd_fd;
266 	filedesc0.fd_fd.fd_refcnt = 1;
267 	filedesc0.fd_fd.fd_cmask = cmask;
268 	filedesc0.fd_fd.fd_ofiles = filedesc0.fd_dfiles;
269 	filedesc0.fd_fd.fd_ofileflags = filedesc0.fd_dfileflags;
270 	filedesc0.fd_fd.fd_nfiles = NDFILE;
271 	filedesc0.fd_fd.fd_himap = filedesc0.fd_dhimap;
272 	filedesc0.fd_fd.fd_lomap = filedesc0.fd_dlomap;
273 
274 	/* Create the limits structures. */
275 	p->p_limit = &limit0;
276 	for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++)
277 		limit0.pl_rlimit[i].rlim_cur =
278 		    limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
279 	limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur = NOFILE;
280 	limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = MIN(NOFILE_MAX,
281 	    (maxfiles - NOFILE > NOFILE) ?  maxfiles - NOFILE : NOFILE);
282 	limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur = MAXUPRC;
283 	i = ptoa(uvmexp.free);
284 	limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i;
285 	limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
286 	limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
287 	limit0.p_refcnt = 1;
288 
289 	/* Allocate a prototype map so we have something to fork. */
290 	uvmspace_init(&vmspace0, pmap_kernel(), round_page(VM_MIN_ADDRESS),
291 	    trunc_page(VM_MAX_ADDRESS), TRUE);
292 	p->p_vmspace = &vmspace0;
293 
294 	p->p_addr = proc0paddr;				/* XXX */
295 
296 	/*
297 	 * We continue to place resource usage info in the
298 	 * user struct so they're pageable.
299 	 */
300 	p->p_stats = &p->p_addr->u_stats;
301 
302 	/*
303 	 * Charge root for one process.
304 	 */
305 	(void)chgproccnt(0, 1);
306 
307 	rqinit();
308 
309 	/* Configure virtual memory system, set vm rlimits. */
310 	uvm_init_limits(p);
311 
312 	/* Initialize the file systems. */
313 #if defined(NFSSERVER) || defined(NFSCLIENT)
314 	nfs_init();			/* initialize server/shared data */
315 #endif
316 	vfsinit();
317 
318 	/* Start real time and statistics clocks. */
319 	initclocks();
320 
321 #ifdef SYSVSHM
322 	/* Initialize System V style shared memory. */
323 	shminit();
324 #endif
325 
326 #ifdef SYSVSEM
327 	/* Initialize System V style semaphores. */
328 	seminit();
329 #endif
330 
331 #ifdef SYSVMSG
332 	/* Initialize System V style message queues. */
333 	msginit();
334 #endif
335 
336 	/* Attach pseudo-devices. */
337 	randomattach();
338 	for (pdev = pdevinit; pdev->pdev_attach != NULL; pdev++)
339 		if (pdev->pdev_count > 0)
340 			(*pdev->pdev_attach)(pdev->pdev_count);
341 
342 #ifdef CRYPTO
343 	swcr_init();
344 #endif /* CRYPTO */
345 
346 	/*
347 	 * Initialize protocols.  Block reception of incoming packets
348 	 * until everything is ready.
349 	 */
350 	s = splimp();
351 	ifinit();
352 	domaininit();
353 	splx(s);
354 
355 #ifdef GPROF
356 	/* Initialize kernel profiling. */
357 	kmstartup();
358 #endif
359 
360 	/* Start the scheduler */
361 	scheduler_start();
362 
363 	dostartuphooks();
364 
365 	/* Configure root/swap devices */
366 	if (md_diskconf)
367 		(*md_diskconf)();
368 
369 	/* Mount the root file system. */
370 	if (vfs_mountroot())
371 		panic("cannot mount root");
372 	CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS;
373 
374 	/* Get the vnode for '/'.  Set filedesc0.fd_fd.fd_cdir to reference it. */
375 	if (VFS_ROOT(mountlist.cqh_first, &rootvnode))
376 		panic("cannot find root vnode");
377 	filedesc0.fd_fd.fd_cdir = rootvnode;
378 	VREF(filedesc0.fd_fd.fd_cdir);
379 	VOP_UNLOCK(rootvnode, 0, p);
380 	filedesc0.fd_fd.fd_rdir = NULL;
381 
382 	uvm_swap_init();
383 
384 	/*
385 	 * Now can look at time, having had a chance to verify the time
386 	 * from the file system.  Reset p->p_rtime as it may have been
387 	 * munched in mi_switch() after the time got set.
388 	 */
389 	p->p_stats->p_start = runtime = mono_time = boottime = time;
390 	p->p_rtime.tv_sec = p->p_rtime.tv_usec = 0;
391 
392 	/* Initialize signal state for process 0. */
393 	signal_init();
394 	p->p_sigacts = &sigacts0;
395 	siginit(p);
396 
397 	/* Create process 1 (init(8)). */
398 	if (fork1(p, SIGCHLD, FORK_FORK, NULL, 0, rval))
399 		panic("fork init");
400 	cpu_set_kpc(pfind(rval[0]), start_init, pfind(rval[0]));
401 
402 	/* Create process 2, the pageout daemon kernel thread. */
403 	if (kthread_create(start_pagedaemon, NULL, NULL, "pagedaemon"))
404 		panic("fork pagedaemon");
405 
406 	/* Create process 3, the reaper daemon kernel thread. */
407 	if (kthread_create(start_reaper, NULL, NULL, "reaper"))
408 		panic("fork reaper");
409 
410 	/* Create process 4, the update daemon kernel thread. */
411 	if (kthread_create(start_update, NULL, NULL, "update")) {
412 #ifdef DIAGNOSTIC
413 		panic("fork update");
414 #endif
415 	}
416 
417 #ifdef CRYPTO
418 	/* Create process 5, the crypto kernel thread. */
419 	if (kthread_create(start_crypto, NULL, NULL, "crypto"))
420 		panic("crypto thread");
421 #endif /* CRYPTO */
422 
423 	/* Create any other deferred kernel threads. */
424 	kthread_run_deferred_queue();
425 
426 	microtime(&rtv);
427 	srandom((u_long)(rtv.tv_sec ^ rtv.tv_usec));
428 
429 	randompid = 1;
430 	/* The scheduler is an infinite loop. */
431 	uvm_scheduler();
432 	/* NOTREACHED */
433 }
434 
435 /*
436  * List of paths to try when searching for "init".
437  */
438 static char *initpaths[] = {
439 	"/sbin/init",
440 	"/sbin/oinit",
441 	"/sbin/init.bak",
442 	NULL,
443 };
444 
445 void
446 check_console(p)
447 	struct proc *p;
448 {
449 	struct nameidata nd;
450 	int error;
451 
452 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console", p);
453 	error = namei(&nd);
454 	if (error) {
455 		if (error == ENOENT)
456 			printf("warning: /dev/console does not exist\n");
457 		else
458 			printf("warning: /dev/console error %d\n", error);
459 	} else
460 		vrele(nd.ni_vp);
461 }
462 
463 /*
464  * Start the initial user process; try exec'ing each pathname in "initpaths".
465  * The program is invoked with one argument containing the boot flags.
466  */
467 void
468 start_init(arg)
469 	void *arg;
470 {
471 	struct proc *p = arg;
472 	vaddr_t addr;
473 	struct sys_execve_args /* {
474 		syscallarg(char *) path;
475 		syscallarg(char **) argp;
476 		syscallarg(char **) envp;
477 	} */ args;
478 	int options, i, error;
479 	register_t retval[2];
480 	char flags[4], *flagsp;
481 	char **pathp, *path, *ucp, **uap, *arg0, *arg1 = NULL;
482 
483 	/*
484 	 * Now in process 1.
485 	 */
486 	initproc = p;
487 
488 	check_console(p);
489 
490 	/*
491 	 * Need just enough stack to hold the faked-up "execve()" arguments.
492 	 */
493 #ifdef MACHINE_STACK_GROWS_UP
494 	addr = USRSTACK;
495 #else
496 	addr = USRSTACK - PAGE_SIZE;
497 #endif
498 	if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE,
499 	    NULL, UVM_UNKNOWN_OFFSET,
500 	    UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY,
501 	    UVM_ADV_NORMAL, UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW))
502 	    != KERN_SUCCESS)
503 		panic("init: couldn't allocate argument space");
504 #ifdef MACHINE_STACK_GROWS_UP
505 	p->p_vmspace->vm_maxsaddr = (caddr_t)addr + PAGE_SIZE;
506 #else
507 	p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
508 #endif
509 
510 	for (pathp = &initpaths[0]; (path = *pathp) != NULL; pathp++) {
511 #ifdef MACHINE_STACK_GROWS_UP
512 		ucp = (char *)addr;
513 #else
514 		ucp = (char *)(addr + PAGE_SIZE);
515 #endif
516 		/*
517 		 * Construct the boot flag argument.
518 		 */
519 		flagsp = flags;
520 		*flagsp++ = '-';
521 		options = 0;
522 
523 		if (boothowto & RB_SINGLE) {
524 			*flagsp++ = 's';
525 			options = 1;
526 		}
527 #ifdef notyet
528 		if (boothowto & RB_FASTBOOT) {
529 			*flagsp++ = 'f';
530 			options = 1;
531 		}
532 #endif
533 
534 		/*
535 		 * Move out the flags (arg 1), if necessary.
536 		 */
537 		if (options != 0) {
538 			*flagsp++ = '\0';
539 			i = flagsp - flags;
540 #ifdef DEBUG
541 			printf("init: copying out flags `%s' %d\n", flags, i);
542 #endif
543 #ifdef MACHINE_STACK_GROWS_UP
544 			arg1 = ucp;
545 			(void)copyout((caddr_t)flags, (caddr_t)ucp, i);
546 			ucp += i;
547 #else
548 			(void)copyout((caddr_t)flags, (caddr_t)(ucp -= i), i);
549 			arg1 = ucp;
550 #endif
551 		}
552 
553 		/*
554 		 * Move out the file name (also arg 0).
555 		 */
556 		i = strlen(path) + 1;
557 #ifdef DEBUG
558 		printf("init: copying out path `%s' %d\n", path, i);
559 #endif
560 #ifdef MACHINE_STACK_GROWS_UP
561 		arg0 = ucp;
562 		(void)copyout((caddr_t)path, (caddr_t)ucp, i);
563 		ucp += i;
564 		ucp = (caddr_t)ALIGN((u_long)ucp);
565 		uap = (char **)ucp + 3;
566 #else
567 		(void)copyout((caddr_t)path, (caddr_t)(ucp -= i), i);
568 		arg0 = ucp;
569 		uap = (char **)((u_long)ucp & ~ALIGNBYTES);
570 #endif
571 
572 		/*
573 		 * Move out the arg pointers.
574 		 */
575 		(void)suword((caddr_t)--uap, 0);	/* terminator */
576 		if (options != 0)
577 			(void)suword((caddr_t)--uap, (long)arg1);
578 		(void)suword((caddr_t)--uap, (long)arg0);
579 
580 		/*
581 		 * Point at the arguments.
582 		 */
583 		SCARG(&args, path) = arg0;
584 		SCARG(&args, argp) = uap;
585 		SCARG(&args, envp) = NULL;
586 
587 		/*
588 		 * Now try to exec the program.  If can't for any reason
589 		 * other than it doesn't exist, complain.
590 		 */
591 		if ((error = sys_execve(p, &args, retval)) == 0)
592 			return;
593 		if (error != ENOENT)
594 			printf("exec %s: error %d\n", path, error);
595 	}
596 	printf("init: not found\n");
597 	panic("no init");
598 }
599 
600 void
601 start_pagedaemon(arg)
602 	void *arg;
603 {
604 	uvm_pageout();
605 	/* NOTREACHED */
606 }
607 
608 void
609 start_update(arg)
610 	void *arg;
611 {
612 	sched_sync(curproc);
613 	/* NOTREACHED */
614 }
615 
616 void
617 start_reaper(arg)
618 	void *arg;
619 {
620 	reaper();
621 	/* NOTREACHED */
622 }
623 
624 #ifdef CRYPTO
625 void
626 start_crypto(arg)
627 	void *arg;
628 {
629 	crypto_thread();
630 	/* NOTREACHED */
631 }
632 #endif /* CRYPTO */
633