xref: /netbsd-src/sys/kern/init_main.c (revision 1394f01b4a9e99092957ca5d824d67219565d9b5)
1 /*	$NetBSD: init_main.c,v 1.101 1997/06/14 04:18:34 thorpej Exp $	*/
2 
3 /*
4  * Copyright (c) 1995 Christopher G. Demetriou.  All rights reserved.
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.
12  *
13  * Redistribution and use in source and binary forms, with or without
14  * modification, are permitted provided that the following conditions
15  * are met:
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.
28  *
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
39  * SUCH DAMAGE.
40  *
41  *	@(#)init_main.c	8.9 (Berkeley) 1/21/94
42  */
43 
44 #include <sys/param.h>
45 #include <sys/filedesc.h>
46 #include <sys/errno.h>
47 #include <sys/exec.h>
48 #include <sys/kernel.h>
49 #include <sys/mount.h>
50 #include <sys/map.h>
51 #include <sys/proc.h>
52 #include <sys/resourcevar.h>
53 #include <sys/signalvar.h>
54 #include <sys/systm.h>
55 #include <sys/vnode.h>
56 #include <sys/tty.h>
57 #include <sys/conf.h>
58 #include <sys/disklabel.h>
59 #include <sys/buf.h>
60 #ifdef REAL_CLISTS
61 #include <sys/clist.h>
62 #endif
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/namei.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 #include <vm/vm_pageout.h>
89 
90 #include <net/if.h>
91 #include <net/raw_cb.h>
92 
93 char	copyright[] =
94 "Copyright (c) 1982, 1986, 1989, 1991, 1993\n    The Regents of the University of California.  All rights reserved.\n\n";
95 
96 /* Components of the first process -- never freed. */
97 struct	session session0;
98 struct	pgrp pgrp0;
99 struct	proc proc0;
100 struct	pcred cred0;
101 struct	filedesc0 filedesc0;
102 struct	plimit limit0;
103 struct	vmspace vmspace0;
104 struct	proc *curproc = &proc0;
105 struct	proc *initproc, *pageproc;
106 
107 int	cmask = CMASK;
108 extern	struct user *proc0paddr;
109 
110 struct	vnode *rootvp, *swapdev_vp;
111 int	boothowto;
112 struct	timeval boottime;
113 struct	timeval runtime;
114 
115 static void check_console __P((struct proc *p));
116 static void start_init __P((struct proc *));
117 static void start_pagedaemon __P((struct proc *));
118 void main __P((void *));
119 
120 #ifdef cpu_set_init_frame
121 void *initframep;				/* XXX should go away */
122 #endif
123 
124 extern char sigcode[], esigcode[];
125 #ifdef SYSCALL_DEBUG
126 extern char *syscallnames[];
127 #endif
128 
129 struct emul emul_netbsd = {
130 	"netbsd",
131 	NULL,
132 	sendsig,
133 	SYS_syscall,
134 	SYS_MAXSYSCALL,
135 	sysent,
136 #ifdef SYSCALL_DEBUG
137 	syscallnames,
138 #else
139 	NULL,
140 #endif
141 	0,
142 	copyargs,
143 	setregs,
144 	sigcode,
145 	esigcode,
146 };
147 
148 /*
149  * System startup; initialize the world, create process 0, mount root
150  * filesystem, and fork to create init and pagedaemon.  Most of the
151  * hard work is done in the lower-level initialization routines including
152  * startup(), which does memory initialization and autoconfiguration.
153  */
154 void
155 main(framep)
156 	void *framep;				/* XXX should go away */
157 {
158 	register struct proc *p;
159 	register struct pdevinit *pdev;
160 	register int i;
161 	int s, error;
162 	register_t rval[2];
163 	extern struct pdevinit pdevinit[];
164 	extern void roundrobin __P((void *));
165 	extern void schedcpu __P((void *));
166 	extern void disk_init __P((void));
167 #if defined(NFSSERVER) || defined(NFS)
168 	extern void nfs_init __P((void));
169 #endif
170 
171 	/*
172 	 * Initialize the current process pointer (curproc) before
173 	 * any possible traps/probes to simplify trap processing.
174 	 */
175 	p = &proc0;
176 	curproc = p;
177 	/*
178 	 * Attempt to find console and initialize
179 	 * in case of early panic or other messages.
180 	 */
181 	consinit();
182 	printf(copyright);
183 
184 	vm_mem_init();
185 	kmeminit();
186 	disk_init();		/* must come before autoconfiguration */
187 	tty_init();		/* initialise tty list */
188 	config_init();		/* init autoconfiguration data structures */
189 	cpu_startup();
190 
191 	/*
192 	 * Initialize process and pgrp structures.
193 	 */
194 	procinit();
195 
196 	/*
197 	 * Create process 0 (the swapper).
198 	 */
199 	LIST_INSERT_HEAD(&allproc, p, p_list);
200 	p->p_pgrp = &pgrp0;
201 	LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
202 	LIST_INIT(&pgrp0.pg_members);
203 	LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
204 
205 	pgrp0.pg_session = &session0;
206 	session0.s_count = 1;
207 	session0.s_leader = p;
208 
209 	p->p_flag = P_INMEM | P_SYSTEM;
210 	p->p_stat = SRUN;
211 	p->p_nice = NZERO;
212 	p->p_emul = &emul_netbsd;
213 	bcopy("swapper", p->p_comm, sizeof ("swapper"));
214 
215 	/* Create credentials. */
216 	cred0.p_refcnt = 1;
217 	p->p_cred = &cred0;
218 	p->p_ucred = crget();
219 	p->p_ucred->cr_ngroups = 1;	/* group 0 */
220 
221 	/* Create the file descriptor table. */
222 	p->p_fd = &filedesc0.fd_fd;
223 	filedesc0.fd_fd.fd_refcnt = 1;
224 	filedesc0.fd_fd.fd_cmask = cmask;
225 	filedesc0.fd_fd.fd_ofiles = filedesc0.fd_dfiles;
226 	filedesc0.fd_fd.fd_ofileflags = filedesc0.fd_dfileflags;
227 	filedesc0.fd_fd.fd_nfiles = NDFILE;
228 
229 	/* Create the limits structures. */
230 	p->p_limit = &limit0;
231 	for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++)
232 		limit0.pl_rlimit[i].rlim_cur =
233 		    limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
234 	limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur = NOFILE;
235 	limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur = MAXUPRC;
236 	i = ptoa(cnt.v_free_count);
237 	limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i;
238 	limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
239 	limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
240 	limit0.p_refcnt = 1;
241 
242 	/* Allocate a prototype map so we have something to fork. */
243 	p->p_vmspace = vmspace_alloc(round_page(VM_MIN_ADDRESS),
244 				     trunc_page(VM_MAX_ADDRESS), TRUE);
245 
246 	p->p_addr = proc0paddr;				/* XXX */
247 
248 	/*
249 	 * We continue to place resource usage info and signal
250 	 * actions in the user struct so they're pageable.
251 	 */
252 	p->p_stats = &p->p_addr->u_stats;
253 	p->p_sigacts = &p->p_addr->u_sigacts;
254 
255 	/*
256 	 * Charge root for one process.
257 	 */
258 	(void)chgproccnt(0, 1);
259 
260 	rqinit();
261 
262 	/* Configure virtual memory system, set vm rlimits. */
263 	vm_init_limits(p);
264 
265 	/* Initialize the file systems. */
266 #if defined(NFSSERVER) || defined(NFS)
267 	nfs_init();			/* initialize server/shared data */
268 #endif
269 	vfsinit();
270 
271 	/* Start real time and statistics clocks. */
272 	initclocks();
273 
274 	/* Initialize mbuf's. */
275 	mbinit();
276 
277 #ifdef REAL_CLISTS
278 	/* Initialize clists. */
279 	clist_init();
280 #endif
281 
282 #ifdef SYSVSHM
283 	/* Initialize System V style shared memory. */
284 	shminit();
285 #endif
286 
287 #ifdef SYSVSEM
288 	/* Initialize System V style semaphores. */
289 	seminit();
290 #endif
291 
292 #ifdef SYSVMSG
293 	/* Initialize System V style message queues. */
294 	msginit();
295 #endif
296 
297 	/* Attach pseudo-devices. */
298 	for (pdev = pdevinit; pdev->pdev_attach != NULL; pdev++)
299 		(*pdev->pdev_attach)(pdev->pdev_count);
300 
301 	/*
302 	 * Initialize protocols.  Block reception of incoming packets
303 	 * until everything is ready.
304 	 */
305 	s = splimp();
306 	ifinit();
307 	domaininit();
308 	splx(s);
309 
310 #ifdef GPROF
311 	/* Initialize kernel profiling. */
312 	kmstartup();
313 #endif
314 
315 	/* Kick off timeout driven events by calling first time. */
316 	roundrobin(NULL);
317 	schedcpu(NULL);
318 
319 	/* Determine the root and dump devices. */
320 	cpu_rootconf();
321 	cpu_dumpconf();
322 
323 	/* Mount the root file system. */
324 	do {
325 		domountroothook();
326 		if ((error = vfs_mountroot())) {
327 			printf("cannot mount root, error = %d\n", error);
328 			boothowto |= RB_ASKNAME;
329 			setroot(root_device,
330 			    (rootdev != NODEV) ? DISKPART(rootdev) : 0, NULL);
331 		}
332 	} while (error != 0);
333 	mountroothook_destroy();
334 
335 	mountlist.cqh_first->mnt_flag |= MNT_ROOTFS;
336 	mountlist.cqh_first->mnt_op->vfs_refcount++;
337 
338 	/* Get the vnode for '/'.  Set filedesc0.fd_fd.fd_cdir to reference it. */
339 	if (VFS_ROOT(mountlist.cqh_first, &rootvnode))
340 		panic("cannot find root vnode");
341 	filedesc0.fd_fd.fd_cdir = rootvnode;
342 	VREF(filedesc0.fd_fd.fd_cdir);
343 	VOP_UNLOCK(rootvnode);
344 	filedesc0.fd_fd.fd_rdir = NULL;
345 	swapinit();
346 
347 	/*
348 	 * Now can look at time, having had a chance to verify the time
349 	 * from the file system.  Reset p->p_rtime as it may have been
350 	 * munched in mi_switch() after the time got set.
351 	 */
352 	p->p_stats->p_start = runtime = mono_time = boottime = time;
353 	p->p_rtime.tv_sec = p->p_rtime.tv_usec = 0;
354 
355 	/* Initialize signal state for process 0. */
356 	siginit(p);
357 
358 	/* Create process 1 (init(8)). */
359 	if (sys_fork(p, NULL, rval))
360 		panic("fork init");
361 #ifdef cpu_set_init_frame			/* XXX should go away */
362 	if (rval[1]) {
363 		/*
364 		 * Now in process 1.
365 		 */
366 		initframep = framep;
367 		start_init(curproc);
368 		return;
369 	}
370 #else
371 	cpu_set_kpc(pfind(1), start_init);
372 #endif
373 
374 	/* Create process 2 (the pageout daemon). */
375 	if (sys_fork(p, NULL, rval))
376 		panic("fork pager");
377 #ifdef cpu_set_init_frame			/* XXX should go away */
378 	if (rval[1]) {
379 		/*
380 		 * Now in process 2.
381 		 */
382 		start_pagedaemon(curproc);
383 	}
384 #else
385 	cpu_set_kpc(pfind(2), start_pagedaemon);
386 #endif
387 
388 	/* The scheduler is an infinite loop. */
389 	scheduler();
390 	/* NOTREACHED */
391 }
392 
393 static void
394 check_console(p)
395 	struct proc *p;
396 {
397 	struct nameidata nd;
398 	int error;
399 
400 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console", p);
401 	error = namei(&nd);
402 	if (error == 0)
403 		vrele(nd.ni_vp);
404 	else if (error == ENOENT)
405 		printf("warning: no /dev/console\n");
406 	else
407 		printf("warning: lookup /dev/console: error %d\n", error);
408 }
409 
410 /*
411  * List of paths to try when searching for "init".
412  */
413 static char *initpaths[] = {
414 	"/sbin/init",
415 	"/sbin/oinit",
416 	"/sbin/init.bak",
417 	NULL,
418 };
419 
420 /*
421  * Start the initial user process; try exec'ing each pathname in "initpaths".
422  * The program is invoked with one argument containing the boot flags.
423  */
424 static void
425 start_init(p)
426 	struct proc *p;
427 {
428 	vm_offset_t addr;
429 	struct sys_execve_args /* {
430 		syscallarg(char *) path;
431 		syscallarg(char * const *) argp;
432 		syscallarg(char * const *) envp;
433 	} */ args;
434 	int options, i, error;
435 	register_t retval[2];
436 	char flags[4], *flagsp;
437 	char **pathp, *path, *ucp, **uap, *arg0, *arg1 = NULL;
438 
439 	/*
440 	 * Now in process 1.
441 	 */
442 	initproc = p;
443 
444 #ifdef cpu_set_init_frame			/* XXX should go away */
445 	/*
446 	 * We need to set the system call frame as if we were entered through
447 	 * a syscall() so that when we call sys_execve() below, it will be able
448 	 * to set the entry point (see setregs) when it tries to exec.  The
449 	 * startup code in "locore.s" has allocated space for the frame and
450 	 * passed a pointer to that space as main's argument.
451 	 */
452 	cpu_set_init_frame(p, initframep);
453 #endif
454 
455 	/*
456 	 * This is not the right way to do this.  We really should
457 	 * hand-craft a descriptor onto /dev/console to hand to init,
458 	 * but that's a _lot_ more work, and the benefit from this easy
459 	 * hack makes up for the "good is the enemy of the best" effect.
460 	 */
461 	check_console(p);
462 
463 	/*
464 	 * Need just enough stack to hold the faked-up "execve()" arguments.
465 	 */
466 	addr = USRSTACK - PAGE_SIZE;
467 	if (vm_allocate(&p->p_vmspace->vm_map, &addr, (vm_size_t)PAGE_SIZE,
468 	    FALSE) != 0)
469 		panic("init: couldn't allocate argument space");
470 	p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
471 
472 	for (pathp = &initpaths[0]; (path = *pathp) != NULL; pathp++) {
473 		ucp = (char *)(addr + PAGE_SIZE);
474 
475 		/*
476 		 * Construct the boot flag argument.
477 		 */
478 		flagsp = flags;
479 		*flagsp++ = '-';
480 		options = 0;
481 
482 		if (boothowto & RB_SINGLE) {
483 			*flagsp++ = 's';
484 			options = 1;
485 		}
486 #ifdef notyet
487 		if (boothowto & RB_FASTBOOT) {
488 			*flagsp++ = 'f';
489 			options = 1;
490 		}
491 #endif
492 
493 		/*
494 		 * Move out the flags (arg 1), if necessary.
495 		 */
496 		if (options != 0) {
497 			*flagsp++ = '\0';
498 			i = flagsp - flags;
499 #ifdef DEBUG
500 			printf("init: copying out flags `%s' %d\n", flags, i);
501 #endif
502 			(void)copyout((caddr_t)flags, (caddr_t)(ucp -= i), i);
503 			arg1 = ucp;
504 		}
505 
506 		/*
507 		 * Move out the file name (also arg 0).
508 		 */
509 		i = strlen(path) + 1;
510 #ifdef DEBUG
511 		printf("init: copying out path `%s' %d\n", path, i);
512 #endif
513 		(void)copyout((caddr_t)path, (caddr_t)(ucp -= i), i);
514 		arg0 = ucp;
515 
516 		/*
517 		 * Move out the arg pointers.
518 		 */
519 		uap = (char **)((long)ucp & ~ALIGNBYTES);
520 		(void)suword((caddr_t)--uap, 0);	/* terminator */
521 		if (options != 0)
522 			(void)suword((caddr_t)--uap, (long)arg1);
523 		(void)suword((caddr_t)--uap, (long)arg0);
524 
525 		/*
526 		 * Point at the arguments.
527 		 */
528 		SCARG(&args, path) = arg0;
529 		SCARG(&args, argp) = uap;
530 		SCARG(&args, envp) = NULL;
531 
532 		/*
533 		 * Now try to exec the program.  If can't for any reason
534 		 * other than it doesn't exist, complain.
535 		 */
536 		if ((error = sys_execve(p, &args, retval)) == 0)
537 			return;
538 		if (error != ENOENT)
539 			printf("exec %s: error %d\n", path, error);
540 	}
541 	printf("init: not found\n");
542 	panic("no init");
543 }
544 
545 static void
546 start_pagedaemon(p)
547 	struct proc *p;
548 {
549 
550 	/*
551 	 * Now in process 2.
552 	 */
553 	pageproc = p;
554 	p->p_flag |= P_INMEM | P_SYSTEM;	/* XXX */
555 	bcopy("pagedaemon", curproc->p_comm, sizeof ("pagedaemon"));
556 	vm_pageout();
557 	/* NOTREACHED */
558 }
559