init_main.c revision 46129 
1/*
2 * Copyright (c) 1995 Terrence R. Lambert
3 * All rights reserved.
4 *
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 * $Id: init_main.c,v 1.113 1999/04/24 18:50:48 dt Exp $
43 */
44
45#include "opt_devfs.h"
46
47#include <sys/param.h>
48#include <sys/file.h>
49#include <sys/filedesc.h>
50#include <sys/kernel.h>
51#include <sys/mount.h>
52#include <sys/sysctl.h>
53#include <sys/proc.h>
54#include <sys/resourcevar.h>
55#include <sys/signalvar.h>
56#include <sys/systm.h>
57#include <sys/vnode.h>
58#include <sys/sysent.h>
59#include <sys/reboot.h>
60#include <sys/sysproto.h>
61#include <sys/vmmeter.h>
62#include <sys/unistd.h>
63#include <sys/malloc.h>
64
65#include <machine/cpu.h>
66
67#include <vm/vm.h>
68#include <vm/vm_param.h>
69#include <vm/vm_prot.h>
70#include <sys/lock.h>
71#include <vm/pmap.h>
72#include <vm/vm_map.h>
73#include <sys/user.h>
74#include <sys/copyright.h>
75
76extern struct linker_set	sysinit_set;	/* XXX */
77
78extern void __main __P((void));
79extern void main __P((void *framep));
80
81/* Components of the first process -- never freed. */
82static struct session session0;
83static struct pgrp pgrp0;
84struct	proc proc0;
85static struct pcred cred0;
86static struct procsig procsig0;
87static struct filedesc0 filedesc0;
88static struct plimit limit0;
89static struct vmspace vmspace0;
90struct	proc *initproc;
91
92int cmask = CMASK;
93extern	struct user *proc0paddr;
94
95struct	vnode *rootvp;
96int	boothowto = 0;		/* initialized so that it can be patched */
97
98struct	timeval boottime;
99SYSCTL_STRUCT(_kern, KERN_BOOTTIME, boottime,
100	CTLFLAG_RD, &boottime, timeval, "");
101
102/*
103 * Promiscuous argument pass for start_init()
104 *
105 * This is a kludge because we use a return from main() rather than a call
106 * to a new routine in locore.s to kick the kernel alive from locore.s.
107 */
108static void	*init_framep;
109
110
111#if __GNUC__ >= 2
112void __main() {}
113#endif
114
115
116/*
117 * This ensures that there is at least one entry so that the sysinit_set
118 * symbol is not undefined.  A sybsystem ID of SI_SUB_DUMMY is never
119 * executed.
120 */
121SYSINIT(placeholder, SI_SUB_DUMMY,SI_ORDER_ANY, NULL, NULL)
122
123/*
124 * The sysinit table itself.  Items are checked off as the are run.
125 * If we want to register new sysinit types, add them to newsysinit.
126 */
127struct sysinit **sysinit = (struct sysinit **)sysinit_set.ls_items;
128struct sysinit **newsysinit;
129
130/*
131 * Merge a new sysinit set into the current set, reallocating it if
132 * necessary.  This can only be called after malloc is running.
133 */
134void
135sysinit_add(set)
136	struct sysinit **set;
137{
138	struct sysinit **newset;
139	struct sysinit **sipp;
140	struct sysinit **xipp;
141	int count = 0;
142
143	if (newsysinit)
144		for (sipp = newsysinit; *sipp; sipp++)
145			count++;
146	else
147		for (sipp = sysinit; *sipp; sipp++)
148			count++;
149	for (sipp = set; *sipp; sipp++)
150		count++;
151	count++;		/* Trailing NULL */
152	newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
153	if (newset == NULL)
154		panic("cannot malloc for sysinit");
155	xipp = newset;
156	if (newsysinit)
157		for (sipp = newsysinit; *sipp; sipp++)
158			*xipp++ = *sipp;
159	else
160		for (sipp = sysinit; *sipp; sipp++)
161			*xipp++ = *sipp;
162	for (sipp = set; *sipp; sipp++)
163		*xipp++ = *sipp;
164	*xipp = NULL;
165	if (newsysinit)
166		free(newsysinit, M_TEMP);
167	newsysinit = newset;
168}
169
170/*
171 * System startup; initialize the world, create process 0, mount root
172 * filesystem, and fork to create init and pagedaemon.  Most of the
173 * hard work is done in the lower-level initialization routines including
174 * startup(), which does memory initialization and autoconfiguration.
175 *
176 * This allows simple addition of new kernel subsystems that require
177 * boot time initialization.  It also allows substitution of subsystem
178 * (for instance, a scheduler, kernel profiler, or VM system) by object
179 * module.  Finally, it allows for optional "kernel threads".
180 */
181void
182main(framep)
183	void *framep;
184{
185
186	register struct sysinit **sipp;		/* system initialization*/
187	register struct sysinit **xipp;		/* interior loop of sort*/
188	register struct sysinit *save;		/* bubble*/
189
190	/*
191	 * Copy the locore.s frame pointer for proc0, this is forked into
192	 * all other processes.
193	 */
194	init_framep = framep;
195
196restart:
197	/*
198	 * Perform a bubble sort of the system initialization objects by
199	 * their subsystem (primary key) and order (secondary key).
200	 */
201	for (sipp = sysinit; *sipp; sipp++) {
202		for (xipp = sipp + 1; *xipp; xipp++) {
203			if ((*sipp)->subsystem < (*xipp)->subsystem ||
204			     ((*sipp)->subsystem == (*xipp)->subsystem &&
205			      (*sipp)->order < (*xipp)->order))
206				continue;	/* skip*/
207			save = *sipp;
208			*sipp = *xipp;
209			*xipp = save;
210		}
211	}
212
213	/*
214	 * Traverse the (now) ordered list of system initialization tasks.
215	 * Perform each task, and continue on to the next task.
216	 *
217	 * The last item on the list is expected to be the scheduler,
218	 * which will not return.
219	 */
220	for (sipp = sysinit; *sipp; sipp++) {
221
222		if ((*sipp)->subsystem == SI_SUB_DUMMY)
223			continue;	/* skip dummy task(s)*/
224
225		if ((*sipp)->subsystem == SI_SUB_DONE)
226			continue;
227
228		switch( (*sipp)->type) {
229		case SI_TYPE_DEFAULT:
230			/* no special processing*/
231			(*((*sipp)->func))((*sipp)->udata);
232			break;
233
234		case SI_TYPE_KTHREAD:
235			/* kernel thread*/
236			if (fork1(&proc0, RFMEM|RFFDG|RFPROC))
237				panic("fork kernel thread");
238			cpu_set_fork_handler(pfind(proc0.p_retval[0]),
239			    (*sipp)->func, (*sipp)->udata);
240			break;
241
242		case SI_TYPE_KPROCESS:
243			if (fork1(&proc0, RFFDG|RFPROC))
244				panic("fork kernel process");
245			cpu_set_fork_handler(pfind(proc0.p_retval[0]),
246			    (*sipp)->func, (*sipp)->udata);
247			break;
248
249		default:
250			panic("init_main: unrecognized init type");
251		}
252
253		/* Check off the one we're just done */
254		(*sipp)->subsystem = SI_SUB_DONE;
255
256		/* Check if we've installed more sysinit items via KLD */
257		if (newsysinit != NULL) {
258			if (sysinit != (struct sysinit **)sysinit_set.ls_items)
259				free(sysinit, M_TEMP);
260			sysinit = newsysinit;
261			newsysinit = NULL;
262			goto restart;
263		}
264	}
265
266	panic("Shouldn't get here!");
267	/* NOTREACHED*/
268}
269
270
271/*
272 * Start a kernel process.  This is called after a fork() call in
273 * main() in the file kern/init_main.c.
274 *
275 * This function is used to start "internal" daemons.
276 */
277/* ARGSUSED*/
278void
279kproc_start(udata)
280	const void *udata;
281{
282	const struct kproc_desc	*kp = udata;
283	struct proc		*p = curproc;
284
285#ifdef DIAGNOSTIC
286	printf("Start pid=%d <%s>\n",p->p_pid, kp->arg0);
287#endif
288
289	/* save a global descriptor, if desired*/
290	if( kp->global_procpp != NULL)
291		*kp->global_procpp	= p;
292
293	/* this is a non-swapped system process*/
294	p->p_flag |= P_INMEM | P_SYSTEM;
295
296	/* set up arg0 for 'ps', et al*/
297	strcpy( p->p_comm, kp->arg0);
298
299	/* call the processes' main()...*/
300	(*kp->func)();
301
302	/* NOTREACHED */
303	panic("kproc_start: %s", kp->arg0);
304}
305
306
307/*
308 ***************************************************************************
309 ****
310 **** The following SYSINIT's belong elsewhere, but have not yet
311 **** been moved.
312 ****
313 ***************************************************************************
314 */
315#ifdef OMIT
316/*
317 * Handled by vfs_mountroot (bad idea) at this time... should be
318 * done the same as 4.4Lite2.
319 */
320SYSINIT(swapinit, SI_SUB_SWAP, SI_ORDER_FIRST, swapinit, NULL)
321#endif	/* OMIT*/
322
323static void print_caddr_t __P((void *data));
324static void
325print_caddr_t(data)
326	void *data;
327{
328	printf("%s", (char *)data);
329}
330SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
331
332
333/*
334 ***************************************************************************
335 ****
336 **** The two following SYSINT's are proc0 specific glue code.  I am not
337 **** convinced that they can not be safely combined, but their order of
338 **** operation has been maintained as the same as the original init_main.c
339 **** for right now.
340 ****
341 **** These probably belong in init_proc.c or kern_proc.c, since they
342 **** deal with proc0 (the fork template process).
343 ****
344 ***************************************************************************
345 */
346/* ARGSUSED*/
347static void proc0_init __P((void *dummy));
348static void
349proc0_init(dummy)
350	void *dummy;
351{
352	register struct proc		*p;
353	register struct filedesc0	*fdp;
354	register unsigned i;
355
356	/*
357	 * Initialize the current process pointer (curproc) before
358	 * any possible traps/probes to simplify trap processing.
359	 */
360	p = &proc0;
361
362	/*
363	 * Initialize process and pgrp structures.
364	 */
365	procinit();
366
367	/*
368	 * Initialize sleep queue hash table
369	 */
370	sleepinit();
371
372	/*
373	 * additional VM structures
374	 */
375	vm_init2();
376
377	/*
378	 * Create process 0 (the swapper).
379	 */
380	LIST_INSERT_HEAD(&allproc, p, p_list);
381	p->p_pgrp = &pgrp0;
382	LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
383	LIST_INIT(&pgrp0.pg_members);
384	LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
385
386	pgrp0.pg_session = &session0;
387	session0.s_count = 1;
388	session0.s_leader = p;
389
390	p->p_sysent = &aout_sysvec;
391
392	p->p_flag = P_INMEM | P_SYSTEM;
393	p->p_stat = SRUN;
394	p->p_nice = NZERO;
395	p->p_rtprio.type = RTP_PRIO_NORMAL;
396	p->p_rtprio.prio = 0;
397
398/*
399 * Link for kernel based threads
400 */
401	p->p_peers = 0;
402	p->p_leader = p;
403
404	bcopy("swapper", p->p_comm, sizeof ("swapper"));
405
406	/* Create credentials. */
407	cred0.p_refcnt = 1;
408	p->p_cred = &cred0;
409	p->p_ucred = crget();
410	p->p_ucred->cr_ngroups = 1;	/* group 0 */
411
412	/* Create procsig. */
413	p->p_procsig = &procsig0;
414	p->p_procsig->ps_refcnt = 1;
415
416	/* Create the file descriptor table. */
417	fdp = &filedesc0;
418	p->p_fd = &fdp->fd_fd;
419	fdp->fd_fd.fd_refcnt = 1;
420	fdp->fd_fd.fd_cmask = cmask;
421	fdp->fd_fd.fd_ofiles = fdp->fd_dfiles;
422	fdp->fd_fd.fd_ofileflags = fdp->fd_dfileflags;
423	fdp->fd_fd.fd_nfiles = NDFILE;
424
425	/* Create the limits structures. */
426	p->p_limit = &limit0;
427	for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++)
428		limit0.pl_rlimit[i].rlim_cur =
429		    limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
430	limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur =
431	    limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
432	limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur =
433	    limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
434	i = ptoa(cnt.v_free_count);
435	limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i;
436	limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
437	limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
438	limit0.p_cpulimit = RLIM_INFINITY;
439	limit0.p_refcnt = 1;
440
441
442	/* Allocate a prototype map so we have something to fork. */
443	pmap_pinit0(vmspace_pmap(&vmspace0));
444	p->p_vmspace = &vmspace0;
445	vmspace0.vm_refcnt = 1;
446	vm_map_init(&vmspace0.vm_map, round_page(VM_MIN_ADDRESS),
447	    trunc_page(VM_MAXUSER_ADDRESS));
448	vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0);
449	p->p_addr = proc0paddr;				/* XXX */
450
451#ifndef __alpha__		/* XXX what is this? */
452#define INCOMPAT_LITES2
453#ifdef INCOMPAT_LITES2
454	/*
455	 * proc0 needs to have a coherent frame base in its stack.
456	 */
457	cpu_set_init_frame(p, init_framep);			/* XXX! */
458#endif	/* INCOMPAT_LITES2*/
459#endif
460
461	/*
462	 * We continue to place resource usage info and signal
463	 * actions in the user struct so they're pageable.
464	 */
465	p->p_stats = &p->p_addr->u_stats;
466	p->p_sigacts = &p->p_addr->u_sigacts;
467
468	/*
469	 * Charge root for one process.
470	 */
471	(void)chgproccnt(0, 1);
472
473	/*
474	 * Initialize the procfs flags (to 0, of course)
475	 */
476	p->p_stops = p->p_stype = p->p_step = 0;
477
478}
479SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL)
480
481/* ARGSUSED*/
482static void proc0_post __P((void *dummy));
483static void
484proc0_post(dummy)
485	void *dummy;
486{
487	struct timespec ts;
488
489	/*
490	 * Now we can look at the time, having had a chance to verify the
491	 * time from the file system.  Pretend that proc0 started now.
492	 */
493	microtime(&proc0.p_stats->p_start);
494	proc0.p_runtime = 0;
495	microuptime(&switchtime);
496	switchticks = ticks;
497
498	/*
499	 * Give the ``random'' number generator a thump.
500	 * XXX: Does read_random() contain enough bits to be used here ?
501	 */
502	nanotime(&ts);
503	srandom(ts.tv_sec ^ ts.tv_nsec);
504
505	/* Initialize signal state for process 0. */
506	siginit(&proc0);
507}
508SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL)
509
510
511
512
513/*
514 ***************************************************************************
515 ****
516 **** The following SYSINIT's and glue code should be moved to the
517 **** respective files on a per subsystem basis.
518 ****
519 ***************************************************************************
520 */
521
522/* ARGSUSED */
523static void root_conf __P((void *dummy));
524static void
525root_conf(dummy)
526	void *dummy;
527{
528	cpu_rootconf();
529}
530SYSINIT(root_conf, SI_SUB_ROOT_CONF, SI_ORDER_FIRST, root_conf, NULL)
531
532/* ARGSUSED*/
533static void xxx_vfs_root_fdtab __P((void *dummy));
534static void
535xxx_vfs_root_fdtab(dummy)
536	void *dummy;
537{
538	register struct filedesc0	*fdp = &filedesc0;
539
540	/* Get the vnode for '/'.  Set fdp->fd_fd.fd_cdir to reference it. */
541	if (VFS_ROOT(mountlist.cqh_first, &rootvnode))
542		panic("cannot find root vnode");
543	fdp->fd_fd.fd_cdir = rootvnode;
544	VREF(fdp->fd_fd.fd_cdir);
545	VOP_UNLOCK(rootvnode, 0, &proc0);
546	fdp->fd_fd.fd_rdir = rootvnode;
547}
548SYSINIT(retrofit, SI_SUB_ROOT_FDTAB, SI_ORDER_FIRST, xxx_vfs_root_fdtab, NULL)
549
550
551/*
552 ***************************************************************************
553 ****
554 **** The following code probably belongs in another file, like
555 **** kern/init_init.c.  It is here for two reasons only:
556 ****
557 ****	1)	This code returns to startup the system; this is
558 ****		abnormal for a kernel thread.
559 ****	2)	This code promiscuously uses init_frame
560 ****
561 ***************************************************************************
562 */
563
564static void kthread_init __P((const void *dummy));
565SYSINIT_KP(init,SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kthread_init, NULL)
566
567
568extern void prepare_usermode __P((void));
569static void start_init __P((struct proc *p));
570
571/* ARGSUSED*/
572static void
573kthread_init(dummy)
574	const void *dummy;
575{
576	/* Create process 1 (init(8)). */
577	start_init(curproc);
578
579	prepare_usermode();
580
581	/*
582	 * This returns to the fork trampoline, then to user mode.
583	 */
584	return;
585}
586
587
588/*
589 * List of paths to try when searching for "init".
590 */
591static char init_path[MAXPATHLEN] =
592    "/sbin/init;/sbin/oinit;/sbin/init.bak;/stand/sysinstall";
593SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, "");
594
595/*
596 * Start the initial user process; try exec'ing each pathname in init_path.
597 * The program is invoked with one argument containing the boot flags.
598 */
599static void
600start_init(p)
601	struct proc *p;
602{
603	vm_offset_t addr;
604	struct execve_args args;
605	int options, error;
606	char *var, *path, *next, *s;
607	char *ucp, **uap, *arg0, *arg1;
608
609	initproc = p;
610
611	/*
612	 * Need just enough stack to hold the faked-up "execve()" arguments.
613	 */
614	addr = trunc_page(USRSTACK - PAGE_SIZE);
615	if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
616			FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0)
617		panic("init: couldn't allocate argument space");
618	p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
619	p->p_vmspace->vm_ssize = 1;
620
621	if ((var = getenv("init_path")) != NULL) {
622		strncpy(init_path, var, MAXPATHLEN);
623		init_path[sizeof init_path - 1] = 0;
624	}
625
626	for (path = init_path; path != '\0'; path = next) {
627		while (*path == ';')
628			path++;
629		if (path == '\0')
630			break;
631		for (next = path; *next != '\0' && *next != ';'; next++)
632			/* nothing */ ;
633		if (bootverbose)
634			printf("start_init: trying %.*s\n", (int)(next - path),
635			    path);
636
637		/*
638		 * Move out the boot flag argument.
639		 */
640		options = 0;
641		ucp = (char *)USRSTACK;
642		(void)subyte(--ucp, 0);		/* trailing zero */
643		if (boothowto & RB_SINGLE) {
644			(void)subyte(--ucp, 's');
645			options = 1;
646		}
647#ifdef notyet
648                if (boothowto & RB_FASTBOOT) {
649			(void)subyte(--ucp, 'f');
650			options = 1;
651		}
652#endif
653
654#ifdef BOOTCDROM
655		(void)subyte(--ucp, 'C');
656		options = 1;
657#endif
658		if (options == 0)
659			(void)subyte(--ucp, '-');
660		(void)subyte(--ucp, '-');		/* leading hyphen */
661		arg1 = ucp;
662
663		/*
664		 * Move out the file name (also arg 0).
665		 */
666		(void)subyte(--ucp, 0);
667		for (s = next - 1; s >= path; s--)
668			(void)subyte(--ucp, *s);
669		arg0 = ucp;
670
671		/*
672		 * Move out the arg pointers.
673		 */
674		uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
675		(void)suword((caddr_t)--uap, (long)0);	/* terminator */
676		(void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
677		(void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
678
679		/*
680		 * Point at the arguments.
681		 */
682		args.fname = arg0;
683		args.argv = uap;
684		args.envv = NULL;
685
686		/*
687		 * Now try to exec the program.  If can't for any reason
688		 * other than it doesn't exist, complain.
689		 *
690		 * Otherwise return to main() which returns to btext
691		 * which completes the system startup.
692		 */
693		if ((error = execve(p, &args)) == 0)
694			return;
695		if (error != ENOENT)
696			printf("exec %.*s: error %d\n", (int)(next - path),
697			    path, error);
698	}
699	printf("init: not found\n");
700	panic("no init");
701}
702