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 * $FreeBSD: head/sys/kern/init_main.c 83366 2001-09-12 08:38:13Z julian $
43 */
44
45#include "opt_init_path.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/ktr.h>
52#include <sys/lock.h>
53#include <sys/mount.h>
54#include <sys/mutex.h>
55#include <sys/sysctl.h>
56#include <sys/proc.h>
57#include <sys/resourcevar.h>
58#include <sys/systm.h>
59#include <sys/signalvar.h>
60#include <sys/vnode.h>
61#include <sys/sysent.h>
62#include <sys/reboot.h>
63#include <sys/sx.h>
64#include <sys/sysproto.h>
65#include <sys/vmmeter.h>
66#include <sys/unistd.h>
67#include <sys/malloc.h>
68#include <sys/conf.h>
69
70#include <machine/cpu.h>
71#include <machine/globals.h>
72
73#include <vm/vm.h>
74#include <vm/vm_param.h>
75#include <vm/pmap.h>
76#include <vm/vm_map.h>
77#include <sys/user.h>
78#include <sys/copyright.h>
79
80void mi_startup(void); /* Should be elsewhere */
81
82/* Components of the first process -- never freed. */
83static struct session session0;
84static struct pgrp pgrp0;
85struct proc proc0;
86struct thread *thread0;
87static struct procsig procsig0;
88static struct filedesc0 filedesc0;
89static struct plimit limit0;
90static struct vmspace vmspace0;
91struct proc *initproc;
92
93int cmask = CMASK;
94extern int fallback_elf_brand;
95
96struct vnode *rootvp;
97int boothowto = 0; /* initialized so that it can be patched */
98SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, "");
99int bootverbose;
100SYSCTL_INT(_debug, OID_AUTO, bootverbose, CTLFLAG_RW, &bootverbose, 0, "");
101
102/*
103 * This ensures that there is at least one entry so that the sysinit_set
104 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never
105 * executed.
106 */
107SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL)
108
109/*
110 * The sysinit table itself. Items are checked off as the are run.
111 * If we want to register new sysinit types, add them to newsysinit.
112 */
113SET_DECLARE(sysinit_set, struct sysinit);
114struct sysinit **sysinit, **sysinit_end;
115struct sysinit **newsysinit, **newsysinit_end;
116
117/*
118 * Merge a new sysinit set into the current set, reallocating it if
119 * necessary. This can only be called after malloc is running.
120 */
121void
122sysinit_add(struct sysinit **set, struct sysinit **set_end)
123{
124 struct sysinit **newset;
125 struct sysinit **sipp;
126 struct sysinit **xipp;
127 int count;
128
129 count = set_end - set;
130 if (newsysinit)
131 count += newsysinit_end - newsysinit;
132 else
133 count += sysinit_end - sysinit;
134 newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
135 if (newset == NULL)
136 panic("cannot malloc for sysinit");
137 xipp = newset;
138 if (newsysinit)
139 for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
140 *xipp++ = *sipp;
141 else
142 for (sipp = sysinit; sipp < sysinit_end; sipp++)
143 *xipp++ = *sipp;
144 for (sipp = set; sipp < set_end; sipp++)
145 *xipp++ = *sipp;
146 if (newsysinit)
147 free(newsysinit, M_TEMP);
148 newsysinit = newset;
149 newsysinit_end = newset + count;
150}
151
152/*
153 * System startup; initialize the world, create process 0, mount root
154 * filesystem, and fork to create init and pagedaemon. Most of the
155 * hard work is done in the lower-level initialization routines including
156 * startup(), which does memory initialization and autoconfiguration.
157 *
158 * This allows simple addition of new kernel subsystems that require
159 * boot time initialization. It also allows substitution of subsystem
160 * (for instance, a scheduler, kernel profiler, or VM system) by object
161 * module. Finally, it allows for optional "kernel threads".
162 */
163void
164mi_startup(void)
165{
166
167 register struct sysinit **sipp; /* system initialization*/
168 register struct sysinit **xipp; /* interior loop of sort*/
169 register struct sysinit *save; /* bubble*/
170
171 if (sysinit == NULL) {
172 sysinit = SET_BEGIN(sysinit_set);
173 sysinit_end = SET_LIMIT(sysinit_set);
174 }
175
176restart:
177 /*
178 * Perform a bubble sort of the system initialization objects by
179 * their subsystem (primary key) and order (secondary key).
180 */
181 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
182 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
183 if ((*sipp)->subsystem < (*xipp)->subsystem ||
184 ((*sipp)->subsystem == (*xipp)->subsystem &&
185 (*sipp)->order <= (*xipp)->order))
186 continue; /* skip*/
187 save = *sipp;
188 *sipp = *xipp;
189 *xipp = save;
190 }
191 }
192
193 /*
194 * Traverse the (now) ordered list of system initialization tasks.
195 * Perform each task, and continue on to the next task.
196 *
197 * The last item on the list is expected to be the scheduler,
198 * which will not return.
199 */
200 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
201
202 if ((*sipp)->subsystem == SI_SUB_DUMMY)
203 continue; /* skip dummy task(s)*/
204
205 if ((*sipp)->subsystem == SI_SUB_DONE)
206 continue;
207
208 /* Call function */
209 (*((*sipp)->func))((*sipp)->udata);
210
211 /* Check off the one we're just done */
212 (*sipp)->subsystem = SI_SUB_DONE;
213
214 /* Check if we've installed more sysinit items via KLD */
215 if (newsysinit != NULL) {
216 if (sysinit != SET_BEGIN(sysinit_set))
217 free(sysinit, M_TEMP);
218 sysinit = newsysinit;
219 sysinit_end = newsysinit_end;
220 newsysinit = NULL;
221 newsysinit_end = NULL;
222 goto restart;
223 }
224 }
225
226 panic("Shouldn't get here!");
227 /* NOTREACHED*/
228}
229
230
231/*
232 ***************************************************************************
233 ****
234 **** The following SYSINIT's belong elsewhere, but have not yet
235 **** been moved.
236 ****
237 ***************************************************************************
238 */
239static void
240print_caddr_t(void *data __unused)
241{
242 printf("%s", (char *)data);
243}
244SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
245SYSINIT(version, SI_SUB_COPYRIGHT, SI_ORDER_SECOND, print_caddr_t, version)
246
247static void
248set_boot_verbose(void *data __unused)
249{
250
251 if (boothowto & RB_VERBOSE)
252 bootverbose++;
253}
254SYSINIT(boot_verbose, SI_SUB_TUNABLES, SI_ORDER_ANY, set_boot_verbose, NULL)
255
256/*
257 ***************************************************************************
258 ****
259 **** The two following SYSINT's are proc0 specific glue code. I am not
260 **** convinced that they can not be safely combined, but their order of
261 **** operation has been maintained as the same as the original init_main.c
262 **** for right now.
263 ****
264 **** These probably belong in init_proc.c or kern_proc.c, since they
265 **** deal with proc0 (the fork template process).
266 ****
267 ***************************************************************************
268 */
269/* ARGSUSED*/
270static void
271proc0_init(void *dummy __unused)
272{
273 register struct proc *p;
274 register struct filedesc0 *fdp;
275 register unsigned i;
276 struct thread *td;
277
278 GIANT_REQUIRED;
279 /*
280 * This assumes the proc0 struct has already been linked
281 * using proc_linkup() in the machine specific initialisation
282 * e.g. i386_init()
283 */
284 p = &proc0;
285 td = thread0;
286 mtx_init(&p->p_mtx, "process lock", MTX_DEF);
287
288 /*
289 * Initialize magic number.
290 */
291 p->p_magic = P_MAGIC;
292
293 /*
294 * Initialize process and pgrp structures.
295 */
296 procinit();
297
298 /*
299 * Initialize sleep queue hash table
300 */
301 sleepinit();
302
303 /*
304 * additional VM structures
305 */
306 vm_init2();
307
308 /*
309 * Create process 0 (the swapper).
310 */
311 LIST_INSERT_HEAD(&allproc, p, p_list);
312 LIST_INSERT_HEAD(PIDHASH(0), p, p_hash);
313 p->p_pgrp = &pgrp0;
314 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
315 LIST_INIT(&pgrp0.pg_members);
316 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
317
318 pgrp0.pg_session = &session0;
319 session0.s_count = 1;
320 session0.s_leader = p;
321
322#ifdef __ELF__
323 p->p_sysent = &elf_freebsd_sysvec;
324#else
325 p->p_sysent = &aout_sysvec;
326#endif
327
328 p->p_flag = P_SYSTEM;
329 p->p_sflag = PS_INMEM;
330 p->p_stat = SRUN;
331 p->p_ksegrp.kg_nice = NZERO;
332 p->p_ksegrp.kg_pri.pri_class = PRI_TIMESHARE;
333 p->p_ksegrp.kg_pri.pri_level = PVM;
334 p->p_ksegrp.kg_pri.pri_native = PUSER;
335 p->p_ksegrp.kg_pri.pri_user = PUSER;
336
337 p->p_peers = 0;
338 p->p_leader = p;
339
340 bcopy("swapper", p->p_comm, sizeof ("swapper"));
341
342 callout_init(&p->p_itcallout, 0);
343 callout_init(&td->td_slpcallout, 1);
344
345 /* Create credentials. */
346 p->p_ucred = crget();
347 p->p_ucred->cr_ngroups = 1; /* group 0 */
348 p->p_ucred->cr_uidinfo = uifind(0);
349 p->p_ucred->cr_ruidinfo = uifind(0);
350 p->p_ucred->cr_prison = NULL; /* Don't jail it. */
351
352 /* Create procsig. */
353 p->p_procsig = &procsig0;
354 p->p_procsig->ps_refcnt = 1;
355
356 /* Initialize signal state for process 0. */
357 siginit(&proc0);
358
359 /* Create the file descriptor table. */
360 fdp = &filedesc0;
361 p->p_fd = &fdp->fd_fd;
362 fdp->fd_fd.fd_refcnt = 1;
363 fdp->fd_fd.fd_cmask = cmask;
364 fdp->fd_fd.fd_ofiles = fdp->fd_dfiles;
365 fdp->fd_fd.fd_ofileflags = fdp->fd_dfileflags;
366 fdp->fd_fd.fd_nfiles = NDFILE;
367
368 /* Create the limits structures. */
369 p->p_limit = &limit0;
370 for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++)
371 limit0.pl_rlimit[i].rlim_cur =
372 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
373 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur =
374 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
375 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur =
376 limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
377 i = ptoa(cnt.v_free_count);
378 limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i;
379 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
380 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
381 limit0.p_cpulimit = RLIM_INFINITY;
382 limit0.p_refcnt = 1;
383
384 /* Allocate a prototype map so we have something to fork. */
385 pmap_pinit0(vmspace_pmap(&vmspace0));
386 p->p_vmspace = &vmspace0;
387 vmspace0.vm_refcnt = 1;
388 vm_map_init(&vmspace0.vm_map, round_page(VM_MIN_ADDRESS),
389 trunc_page(VM_MAXUSER_ADDRESS));
390 vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0);
391
392 /*
393 * We continue to place resource usage info and signal
394 * actions in the user struct so they're pageable.
395 */
396 p->p_stats = &p->p_uarea->u_stats;
397 p->p_sigacts = &p->p_uarea->u_sigacts;
398
399 /*
400 * Charge root for one process.
401 */
402 (void)chgproccnt(p->p_ucred->cr_ruidinfo, 1, 0);
403}
404SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL)
405
406/* ARGSUSED*/
407static void
408proc0_post(void *dummy __unused)
409{
410 struct timespec ts;
411 struct proc *p;
412
413 /*
414 * Now we can look at the time, having had a chance to verify the
415 * time from the file system. Pretend that proc0 started now.
416 */
417 sx_slock(&allproc_lock);
418 LIST_FOREACH(p, &allproc, p_list) {
419 microtime(&p->p_stats->p_start);
420 p->p_runtime = 0;
421 }
422 sx_sunlock(&allproc_lock);
423 microuptime(PCPU_PTR(switchtime));
424 PCPU_SET(switchticks, ticks);
425
426 /*
427 * Give the ``random'' number generator a thump.
428 */
429 nanotime(&ts);
430 srandom(ts.tv_sec ^ ts.tv_nsec);
431}
432SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL)
433
434/*
435 ***************************************************************************
436 ****
437 **** The following SYSINIT's and glue code should be moved to the
438 **** respective files on a per subsystem basis.
439 ****
440 ***************************************************************************
441 */
442
443
444/*
445 ***************************************************************************
446 ****
447 **** The following code probably belongs in another file, like
448 **** kern/init_init.c.
449 ****
450 ***************************************************************************
451 */
452
453/*
454 * List of paths to try when searching for "init".
455 */
456static char init_path[MAXPATHLEN] =
457#ifdef INIT_PATH
458 __XSTRING(INIT_PATH);
459#else
460 "/sbin/init:/sbin/oinit:/sbin/init.bak:/stand/sysinstall";
461#endif
462SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, "");
463
464/*
465 * Start the initial user process; try exec'ing each pathname in init_path.
466 * The program is invoked with one argument containing the boot flags.
467 */
468static void
469start_init(void *dummy)
470{
471 vm_offset_t addr;
472 struct execve_args args;
473 int options, error;
474 char *var, *path, *next, *s;
475 char *ucp, **uap, *arg0, *arg1;
476 struct thread *td;
477 struct proc *p;
478
479 mtx_lock(&Giant);
480
481 GIANT_REQUIRED;
482
483 td = curthread;
484 p = td->td_proc;
485
486 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */
487 if (VFS_ROOT(TAILQ_FIRST(&mountlist), &rootvnode))
488 panic("cannot find root vnode");
489 p->p_fd->fd_cdir = rootvnode;
490 VREF(p->p_fd->fd_cdir);
491 p->p_fd->fd_rdir = rootvnode;
492 VREF(p->p_fd->fd_rdir);
493 VOP_UNLOCK(rootvnode, 0, td);
494
495 /*
496 * Need just enough stack to hold the faked-up "execve()" arguments.
497 */
498 addr = trunc_page(USRSTACK - PAGE_SIZE);
499 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
500 FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0)
501 panic("init: couldn't allocate argument space");
502 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
503 p->p_vmspace->vm_ssize = 1;
504
505 if ((var = getenv("init_path")) != NULL) {
506 strncpy(init_path, var, sizeof init_path);
507 init_path[sizeof init_path - 1] = 0;
508 }
509 if ((var = getenv("kern.fallback_elf_brand")) != NULL)
510 fallback_elf_brand = strtol(var, NULL, 0);
511
512 for (path = init_path; *path != '\0'; path = next) {
513 while (*path == ':')
514 path++;
515 if (*path == '\0')
516 break;
517 for (next = path; *next != '\0' && *next != ':'; next++)
518 /* nothing */ ;
519 if (bootverbose)
520 printf("start_init: trying %.*s\n", (int)(next - path),
521 path);
522
523 /*
524 * Move out the boot flag argument.
525 */
526 options = 0;
527 ucp = (char *)USRSTACK;
528 (void)subyte(--ucp, 0); /* trailing zero */
529 if (boothowto & RB_SINGLE) {
530 (void)subyte(--ucp, 's');
531 options = 1;
532 }
533#ifdef notyet
534 if (boothowto & RB_FASTBOOT) {
535 (void)subyte(--ucp, 'f');
536 options = 1;
537 }
538#endif
539
540#ifdef BOOTCDROM
541 (void)subyte(--ucp, 'C');
542 options = 1;
543#endif
544 if (devfs_present) {
545 (void)subyte(--ucp, 'd');
546 options = 1;
547 }
548
549 if (options == 0)
550 (void)subyte(--ucp, '-');
551 (void)subyte(--ucp, '-'); /* leading hyphen */
552 arg1 = ucp;
553
554 /*
555 * Move out the file name (also arg 0).
556 */
557 (void)subyte(--ucp, 0);
558 for (s = next - 1; s >= path; s--)
559 (void)subyte(--ucp, *s);
560 arg0 = ucp;
561
562 /*
563 * Move out the arg pointers.
564 */
565 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
566 (void)suword((caddr_t)--uap, (long)0); /* terminator */
567 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
568 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
569
570 /*
571 * Point at the arguments.
572 */
573 args.fname = arg0;
574 args.argv = uap;
575 args.envv = NULL;
576
577 /*
578 * Now try to exec the program. If can't for any reason
579 * other than it doesn't exist, complain.
580 *
581 * Otherwise, return via fork_trampoline() all the way
582 * to user mode as init!
583 */
584 if ((error = execve(td, &args)) == 0) {
585 mtx_unlock(&Giant);
586 return;
587 }
588 if (error != ENOENT)
589 printf("exec %.*s: error %d\n", (int)(next - path),
590 path, error);
591 }
592 printf("init: not found in path %s\n", init_path);
593 panic("no init");
594}
595
596/*
597 * Like kthread_create(), but runs in it's own address space.
598 * We do this early to reserve pid 1.
599 *
600 * Note special case - do not make it runnable yet. Other work
601 * in progress will change this more.
602 */
603static void
604create_init(const void *udata __unused)
605{
606 int error;
607
608 error = fork1(thread0, RFFDG | RFPROC | RFSTOPPED, &initproc);
609 if (error)
610 panic("cannot fork init: %d\n", error);
611 PROC_LOCK(initproc);
612 initproc->p_flag |= P_SYSTEM;
613 PROC_UNLOCK(initproc);
614 mtx_lock_spin(&sched_lock);
615 initproc->p_sflag |= PS_INMEM;
616 mtx_unlock_spin(&sched_lock);
617 cpu_set_fork_handler(&initproc->p_thread, start_init, NULL);
618}
619SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
620
621/*
622 * Make it runnable now.
623 */
624static void
625kick_init(const void *udata __unused)
626{
627
628 mtx_lock_spin(&sched_lock);
629 initproc->p_stat = SRUN;
630 setrunqueue(&initproc->p_thread); /* XXXKSE */
631 mtx_unlock_spin(&sched_lock);
632}
633SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)
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 * $FreeBSD: head/sys/kern/init_main.c 83366 2001-09-12 08:38:13Z julian $
43 */
44
45#include "opt_init_path.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/ktr.h>
52#include <sys/lock.h>
53#include <sys/mount.h>
54#include <sys/mutex.h>
55#include <sys/sysctl.h>
56#include <sys/proc.h>
57#include <sys/resourcevar.h>
58#include <sys/systm.h>
59#include <sys/signalvar.h>
60#include <sys/vnode.h>
61#include <sys/sysent.h>
62#include <sys/reboot.h>
63#include <sys/sx.h>
64#include <sys/sysproto.h>
65#include <sys/vmmeter.h>
66#include <sys/unistd.h>
67#include <sys/malloc.h>
68#include <sys/conf.h>
69
70#include <machine/cpu.h>
71#include <machine/globals.h>
72
73#include <vm/vm.h>
74#include <vm/vm_param.h>
75#include <vm/pmap.h>
76#include <vm/vm_map.h>
77#include <sys/user.h>
78#include <sys/copyright.h>
79
80void mi_startup(void); /* Should be elsewhere */
81
82/* Components of the first process -- never freed. */
83static struct session session0;
84static struct pgrp pgrp0;
85struct proc proc0;
86struct thread *thread0;
87static struct procsig procsig0;
88static struct filedesc0 filedesc0;
89static struct plimit limit0;
90static struct vmspace vmspace0;
91struct proc *initproc;
92
93int cmask = CMASK;
94extern int fallback_elf_brand;
95
96struct vnode *rootvp;
97int boothowto = 0; /* initialized so that it can be patched */
98SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, "");
99int bootverbose;
100SYSCTL_INT(_debug, OID_AUTO, bootverbose, CTLFLAG_RW, &bootverbose, 0, "");
101
102/*
103 * This ensures that there is at least one entry so that the sysinit_set
104 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never
105 * executed.
106 */
107SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL)
108
109/*
110 * The sysinit table itself. Items are checked off as the are run.
111 * If we want to register new sysinit types, add them to newsysinit.
112 */
113SET_DECLARE(sysinit_set, struct sysinit);
114struct sysinit **sysinit, **sysinit_end;
115struct sysinit **newsysinit, **newsysinit_end;
116
117/*
118 * Merge a new sysinit set into the current set, reallocating it if
119 * necessary. This can only be called after malloc is running.
120 */
121void
122sysinit_add(struct sysinit **set, struct sysinit **set_end)
123{
124 struct sysinit **newset;
125 struct sysinit **sipp;
126 struct sysinit **xipp;
127 int count;
128
129 count = set_end - set;
130 if (newsysinit)
131 count += newsysinit_end - newsysinit;
132 else
133 count += sysinit_end - sysinit;
134 newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
135 if (newset == NULL)
136 panic("cannot malloc for sysinit");
137 xipp = newset;
138 if (newsysinit)
139 for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
140 *xipp++ = *sipp;
141 else
142 for (sipp = sysinit; sipp < sysinit_end; sipp++)
143 *xipp++ = *sipp;
144 for (sipp = set; sipp < set_end; sipp++)
145 *xipp++ = *sipp;
146 if (newsysinit)
147 free(newsysinit, M_TEMP);
148 newsysinit = newset;
149 newsysinit_end = newset + count;
150}
151
152/*
153 * System startup; initialize the world, create process 0, mount root
154 * filesystem, and fork to create init and pagedaemon. Most of the
155 * hard work is done in the lower-level initialization routines including
156 * startup(), which does memory initialization and autoconfiguration.
157 *
158 * This allows simple addition of new kernel subsystems that require
159 * boot time initialization. It also allows substitution of subsystem
160 * (for instance, a scheduler, kernel profiler, or VM system) by object
161 * module. Finally, it allows for optional "kernel threads".
162 */
163void
164mi_startup(void)
165{
166
167 register struct sysinit **sipp; /* system initialization*/
168 register struct sysinit **xipp; /* interior loop of sort*/
169 register struct sysinit *save; /* bubble*/
170
171 if (sysinit == NULL) {
172 sysinit = SET_BEGIN(sysinit_set);
173 sysinit_end = SET_LIMIT(sysinit_set);
174 }
175
176restart:
177 /*
178 * Perform a bubble sort of the system initialization objects by
179 * their subsystem (primary key) and order (secondary key).
180 */
181 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
182 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
183 if ((*sipp)->subsystem < (*xipp)->subsystem ||
184 ((*sipp)->subsystem == (*xipp)->subsystem &&
185 (*sipp)->order <= (*xipp)->order))
186 continue; /* skip*/
187 save = *sipp;
188 *sipp = *xipp;
189 *xipp = save;
190 }
191 }
192
193 /*
194 * Traverse the (now) ordered list of system initialization tasks.
195 * Perform each task, and continue on to the next task.
196 *
197 * The last item on the list is expected to be the scheduler,
198 * which will not return.
199 */
200 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
201
202 if ((*sipp)->subsystem == SI_SUB_DUMMY)
203 continue; /* skip dummy task(s)*/
204
205 if ((*sipp)->subsystem == SI_SUB_DONE)
206 continue;
207
208 /* Call function */
209 (*((*sipp)->func))((*sipp)->udata);
210
211 /* Check off the one we're just done */
212 (*sipp)->subsystem = SI_SUB_DONE;
213
214 /* Check if we've installed more sysinit items via KLD */
215 if (newsysinit != NULL) {
216 if (sysinit != SET_BEGIN(sysinit_set))
217 free(sysinit, M_TEMP);
218 sysinit = newsysinit;
219 sysinit_end = newsysinit_end;
220 newsysinit = NULL;
221 newsysinit_end = NULL;
222 goto restart;
223 }
224 }
225
226 panic("Shouldn't get here!");
227 /* NOTREACHED*/
228}
229
230
231/*
232 ***************************************************************************
233 ****
234 **** The following SYSINIT's belong elsewhere, but have not yet
235 **** been moved.
236 ****
237 ***************************************************************************
238 */
239static void
240print_caddr_t(void *data __unused)
241{
242 printf("%s", (char *)data);
243}
244SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
245SYSINIT(version, SI_SUB_COPYRIGHT, SI_ORDER_SECOND, print_caddr_t, version)
246
247static void
248set_boot_verbose(void *data __unused)
249{
250
251 if (boothowto & RB_VERBOSE)
252 bootverbose++;
253}
254SYSINIT(boot_verbose, SI_SUB_TUNABLES, SI_ORDER_ANY, set_boot_verbose, NULL)
255
256/*
257 ***************************************************************************
258 ****
259 **** The two following SYSINT's are proc0 specific glue code. I am not
260 **** convinced that they can not be safely combined, but their order of
261 **** operation has been maintained as the same as the original init_main.c
262 **** for right now.
263 ****
264 **** These probably belong in init_proc.c or kern_proc.c, since they
265 **** deal with proc0 (the fork template process).
266 ****
267 ***************************************************************************
268 */
269/* ARGSUSED*/
270static void
271proc0_init(void *dummy __unused)
272{
273 register struct proc *p;
274 register struct filedesc0 *fdp;
275 register unsigned i;
276 struct thread *td;
277
278 GIANT_REQUIRED;
279 /*
280 * This assumes the proc0 struct has already been linked
281 * using proc_linkup() in the machine specific initialisation
282 * e.g. i386_init()
283 */
284 p = &proc0;
285 td = thread0;
286 mtx_init(&p->p_mtx, "process lock", MTX_DEF);
287
288 /*
289 * Initialize magic number.
290 */
291 p->p_magic = P_MAGIC;
292
293 /*
294 * Initialize process and pgrp structures.
295 */
296 procinit();
297
298 /*
299 * Initialize sleep queue hash table
300 */
301 sleepinit();
302
303 /*
304 * additional VM structures
305 */
306 vm_init2();
307
308 /*
309 * Create process 0 (the swapper).
310 */
311 LIST_INSERT_HEAD(&allproc, p, p_list);
312 LIST_INSERT_HEAD(PIDHASH(0), p, p_hash);
313 p->p_pgrp = &pgrp0;
314 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
315 LIST_INIT(&pgrp0.pg_members);
316 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
317
318 pgrp0.pg_session = &session0;
319 session0.s_count = 1;
320 session0.s_leader = p;
321
322#ifdef __ELF__
323 p->p_sysent = &elf_freebsd_sysvec;
324#else
325 p->p_sysent = &aout_sysvec;
326#endif
327
328 p->p_flag = P_SYSTEM;
329 p->p_sflag = PS_INMEM;
330 p->p_stat = SRUN;
331 p->p_ksegrp.kg_nice = NZERO;
332 p->p_ksegrp.kg_pri.pri_class = PRI_TIMESHARE;
333 p->p_ksegrp.kg_pri.pri_level = PVM;
334 p->p_ksegrp.kg_pri.pri_native = PUSER;
335 p->p_ksegrp.kg_pri.pri_user = PUSER;
336
337 p->p_peers = 0;
338 p->p_leader = p;
339
340 bcopy("swapper", p->p_comm, sizeof ("swapper"));
341
342 callout_init(&p->p_itcallout, 0);
343 callout_init(&td->td_slpcallout, 1);
344
345 /* Create credentials. */
346 p->p_ucred = crget();
347 p->p_ucred->cr_ngroups = 1; /* group 0 */
348 p->p_ucred->cr_uidinfo = uifind(0);
349 p->p_ucred->cr_ruidinfo = uifind(0);
350 p->p_ucred->cr_prison = NULL; /* Don't jail it. */
351
352 /* Create procsig. */
353 p->p_procsig = &procsig0;
354 p->p_procsig->ps_refcnt = 1;
355
356 /* Initialize signal state for process 0. */
357 siginit(&proc0);
358
359 /* Create the file descriptor table. */
360 fdp = &filedesc0;
361 p->p_fd = &fdp->fd_fd;
362 fdp->fd_fd.fd_refcnt = 1;
363 fdp->fd_fd.fd_cmask = cmask;
364 fdp->fd_fd.fd_ofiles = fdp->fd_dfiles;
365 fdp->fd_fd.fd_ofileflags = fdp->fd_dfileflags;
366 fdp->fd_fd.fd_nfiles = NDFILE;
367
368 /* Create the limits structures. */
369 p->p_limit = &limit0;
370 for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++)
371 limit0.pl_rlimit[i].rlim_cur =
372 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
373 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur =
374 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
375 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur =
376 limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
377 i = ptoa(cnt.v_free_count);
378 limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i;
379 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
380 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
381 limit0.p_cpulimit = RLIM_INFINITY;
382 limit0.p_refcnt = 1;
383
384 /* Allocate a prototype map so we have something to fork. */
385 pmap_pinit0(vmspace_pmap(&vmspace0));
386 p->p_vmspace = &vmspace0;
387 vmspace0.vm_refcnt = 1;
388 vm_map_init(&vmspace0.vm_map, round_page(VM_MIN_ADDRESS),
389 trunc_page(VM_MAXUSER_ADDRESS));
390 vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0);
391
392 /*
393 * We continue to place resource usage info and signal
394 * actions in the user struct so they're pageable.
395 */
396 p->p_stats = &p->p_uarea->u_stats;
397 p->p_sigacts = &p->p_uarea->u_sigacts;
398
399 /*
400 * Charge root for one process.
401 */
402 (void)chgproccnt(p->p_ucred->cr_ruidinfo, 1, 0);
403}
404SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL)
405
406/* ARGSUSED*/
407static void
408proc0_post(void *dummy __unused)
409{
410 struct timespec ts;
411 struct proc *p;
412
413 /*
414 * Now we can look at the time, having had a chance to verify the
415 * time from the file system. Pretend that proc0 started now.
416 */
417 sx_slock(&allproc_lock);
418 LIST_FOREACH(p, &allproc, p_list) {
419 microtime(&p->p_stats->p_start);
420 p->p_runtime = 0;
421 }
422 sx_sunlock(&allproc_lock);
423 microuptime(PCPU_PTR(switchtime));
424 PCPU_SET(switchticks, ticks);
425
426 /*
427 * Give the ``random'' number generator a thump.
428 */
429 nanotime(&ts);
430 srandom(ts.tv_sec ^ ts.tv_nsec);
431}
432SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL)
433
434/*
435 ***************************************************************************
436 ****
437 **** The following SYSINIT's and glue code should be moved to the
438 **** respective files on a per subsystem basis.
439 ****
440 ***************************************************************************
441 */
442
443
444/*
445 ***************************************************************************
446 ****
447 **** The following code probably belongs in another file, like
448 **** kern/init_init.c.
449 ****
450 ***************************************************************************
451 */
452
453/*
454 * List of paths to try when searching for "init".
455 */
456static char init_path[MAXPATHLEN] =
457#ifdef INIT_PATH
458 __XSTRING(INIT_PATH);
459#else
460 "/sbin/init:/sbin/oinit:/sbin/init.bak:/stand/sysinstall";
461#endif
462SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, "");
463
464/*
465 * Start the initial user process; try exec'ing each pathname in init_path.
466 * The program is invoked with one argument containing the boot flags.
467 */
468static void
469start_init(void *dummy)
470{
471 vm_offset_t addr;
472 struct execve_args args;
473 int options, error;
474 char *var, *path, *next, *s;
475 char *ucp, **uap, *arg0, *arg1;
476 struct thread *td;
477 struct proc *p;
478
479 mtx_lock(&Giant);
480
481 GIANT_REQUIRED;
482
483 td = curthread;
484 p = td->td_proc;
485
486 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */
487 if (VFS_ROOT(TAILQ_FIRST(&mountlist), &rootvnode))
488 panic("cannot find root vnode");
489 p->p_fd->fd_cdir = rootvnode;
490 VREF(p->p_fd->fd_cdir);
491 p->p_fd->fd_rdir = rootvnode;
492 VREF(p->p_fd->fd_rdir);
493 VOP_UNLOCK(rootvnode, 0, td);
494
495 /*
496 * Need just enough stack to hold the faked-up "execve()" arguments.
497 */
498 addr = trunc_page(USRSTACK - PAGE_SIZE);
499 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
500 FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0)
501 panic("init: couldn't allocate argument space");
502 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
503 p->p_vmspace->vm_ssize = 1;
504
505 if ((var = getenv("init_path")) != NULL) {
506 strncpy(init_path, var, sizeof init_path);
507 init_path[sizeof init_path - 1] = 0;
508 }
509 if ((var = getenv("kern.fallback_elf_brand")) != NULL)
510 fallback_elf_brand = strtol(var, NULL, 0);
511
512 for (path = init_path; *path != '\0'; path = next) {
513 while (*path == ':')
514 path++;
515 if (*path == '\0')
516 break;
517 for (next = path; *next != '\0' && *next != ':'; next++)
518 /* nothing */ ;
519 if (bootverbose)
520 printf("start_init: trying %.*s\n", (int)(next - path),
521 path);
522
523 /*
524 * Move out the boot flag argument.
525 */
526 options = 0;
527 ucp = (char *)USRSTACK;
528 (void)subyte(--ucp, 0); /* trailing zero */
529 if (boothowto & RB_SINGLE) {
530 (void)subyte(--ucp, 's');
531 options = 1;
532 }
533#ifdef notyet
534 if (boothowto & RB_FASTBOOT) {
535 (void)subyte(--ucp, 'f');
536 options = 1;
537 }
538#endif
539
540#ifdef BOOTCDROM
541 (void)subyte(--ucp, 'C');
542 options = 1;
543#endif
544 if (devfs_present) {
545 (void)subyte(--ucp, 'd');
546 options = 1;
547 }
548
549 if (options == 0)
550 (void)subyte(--ucp, '-');
551 (void)subyte(--ucp, '-'); /* leading hyphen */
552 arg1 = ucp;
553
554 /*
555 * Move out the file name (also arg 0).
556 */
557 (void)subyte(--ucp, 0);
558 for (s = next - 1; s >= path; s--)
559 (void)subyte(--ucp, *s);
560 arg0 = ucp;
561
562 /*
563 * Move out the arg pointers.
564 */
565 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
566 (void)suword((caddr_t)--uap, (long)0); /* terminator */
567 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
568 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
569
570 /*
571 * Point at the arguments.
572 */
573 args.fname = arg0;
574 args.argv = uap;
575 args.envv = NULL;
576
577 /*
578 * Now try to exec the program. If can't for any reason
579 * other than it doesn't exist, complain.
580 *
581 * Otherwise, return via fork_trampoline() all the way
582 * to user mode as init!
583 */
584 if ((error = execve(td, &args)) == 0) {
585 mtx_unlock(&Giant);
586 return;
587 }
588 if (error != ENOENT)
589 printf("exec %.*s: error %d\n", (int)(next - path),
590 path, error);
591 }
592 printf("init: not found in path %s\n", init_path);
593 panic("no init");
594}
595
596/*
597 * Like kthread_create(), but runs in it's own address space.
598 * We do this early to reserve pid 1.
599 *
600 * Note special case - do not make it runnable yet. Other work
601 * in progress will change this more.
602 */
603static void
604create_init(const void *udata __unused)
605{
606 int error;
607
608 error = fork1(thread0, RFFDG | RFPROC | RFSTOPPED, &initproc);
609 if (error)
610 panic("cannot fork init: %d\n", error);
611 PROC_LOCK(initproc);
612 initproc->p_flag |= P_SYSTEM;
613 PROC_UNLOCK(initproc);
614 mtx_lock_spin(&sched_lock);
615 initproc->p_sflag |= PS_INMEM;
616 mtx_unlock_spin(&sched_lock);
617 cpu_set_fork_handler(&initproc->p_thread, start_init, NULL);
618}
619SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
620
621/*
622 * Make it runnable now.
623 */
624static void
625kick_init(const void *udata __unused)
626{
627
628 mtx_lock_spin(&sched_lock);
629 initproc->p_stat = SRUN;
630 setrunqueue(&initproc->p_thread); /* XXXKSE */
631 mtx_unlock_spin(&sched_lock);
632}
633SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)