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 */
43
44 #include <sys/cdefs.h>
45 __FBSDID("$FreeBSD: stable/10/sys/kern/init_main.c 337471 2018-08-08 17:49:35Z bdrewery $");
46
47 #include "opt_ddb.h"
48 #include "opt_init_path.h"
49
50 #include <sys/param.h>
51 #include <sys/kernel.h>
52 #include <sys/exec.h>
53 #include <sys/file.h>
54 #include <sys/filedesc.h>
55 #include <sys/jail.h>
56 #include <sys/ktr.h>
57 #include <sys/lock.h>
58 #include <sys/loginclass.h>
59 #include <sys/mount.h>
60 #include <sys/mutex.h>
61 #include <sys/syscallsubr.h>
62 #include <sys/sysctl.h>
63 #include <sys/proc.h>
64 #include <sys/racct.h>
65 #include <sys/resourcevar.h>
66 #include <sys/systm.h>
67 #include <sys/signalvar.h>
68 #include <sys/vnode.h>
69 #include <sys/sysent.h>
70 #include <sys/reboot.h>
71 #include <sys/sched.h>
72 #include <sys/sx.h>
73 #include <sys/sysproto.h>
74 #include <sys/vmmeter.h>
75 #include <sys/unistd.h>
76 #include <sys/malloc.h>
77 #include <sys/conf.h>
78 #include <sys/cpuset.h>
79
80 #include <machine/cpu.h>
81
82 #include <security/audit/audit.h>
83 #include <security/mac/mac_framework.h>
84
85 #include <vm/vm.h>
86 #include <vm/vm_param.h>
87 #include <vm/pmap.h>
88 #include <vm/vm_map.h>
89 #include <sys/copyright.h>
90
91 #include <ddb/ddb.h>
92 #include <ddb/db_sym.h>
93
94 void mi_startup(void); /* Should be elsewhere */
95
96 /* Components of the first process -- never freed. */
97 static struct session session0;
98 static struct pgrp pgrp0;
99 struct proc proc0;
100 struct thread thread0 __aligned(16);
101 struct vmspace vmspace0;
102 struct proc *initproc;
103
104 #ifndef BOOTHOWTO
105 #define BOOTHOWTO 0
106 #endif
107 int boothowto = BOOTHOWTO; /* initialized so that it can be patched */
108 SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0,
109 "Boot control flags, passed from loader");
110
111 #ifndef BOOTVERBOSE
112 #define BOOTVERBOSE 0
113 #endif
114 int bootverbose = BOOTVERBOSE;
115 SYSCTL_INT(_debug, OID_AUTO, bootverbose, CTLFLAG_RW, &bootverbose, 0,
116 "Control the output of verbose kernel messages");
117
118 #ifdef INVARIANTS
119 FEATURE(invariants, "Kernel compiled with INVARIANTS, may affect performance");
120 #endif
121
122 /*
123 * This ensures that there is at least one entry so that the sysinit_set
124 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never
125 * executed.
126 */
127 SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL);
128
129 /*
130 * The sysinit table itself. Items are checked off as the are run.
131 * If we want to register new sysinit types, add them to newsysinit.
132 */
133 SET_DECLARE(sysinit_set, struct sysinit);
134 struct sysinit **sysinit, **sysinit_end;
135 struct sysinit **newsysinit, **newsysinit_end;
136
137 /*
138 * Merge a new sysinit set into the current set, reallocating it if
139 * necessary. This can only be called after malloc is running.
140 */
141 void
sysinit_add(struct sysinit ** set,struct sysinit ** set_end)142 sysinit_add(struct sysinit **set, struct sysinit **set_end)
143 {
144 struct sysinit **newset;
145 struct sysinit **sipp;
146 struct sysinit **xipp;
147 int count;
148
149 count = set_end - set;
150 if (newsysinit)
151 count += newsysinit_end - newsysinit;
152 else
153 count += sysinit_end - sysinit;
154 newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
155 if (newset == NULL)
156 panic("cannot malloc for sysinit");
157 xipp = newset;
158 if (newsysinit)
159 for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
160 *xipp++ = *sipp;
161 else
162 for (sipp = sysinit; sipp < sysinit_end; sipp++)
163 *xipp++ = *sipp;
164 for (sipp = set; sipp < set_end; sipp++)
165 *xipp++ = *sipp;
166 if (newsysinit)
167 free(newsysinit, M_TEMP);
168 newsysinit = newset;
169 newsysinit_end = newset + count;
170 }
171
172 #if defined (DDB) && defined(VERBOSE_SYSINIT)
173 static const char *
symbol_name(vm_offset_t va,db_strategy_t strategy)174 symbol_name(vm_offset_t va, db_strategy_t strategy)
175 {
176 const char *name;
177 c_db_sym_t sym;
178 db_expr_t offset;
179
180 if (va == 0)
181 return (NULL);
182 sym = db_search_symbol(va, strategy, &offset);
183 if (offset != 0)
184 return (NULL);
185 db_symbol_values(sym, &name, NULL);
186 return (name);
187 }
188 #endif
189
190 /*
191 * System startup; initialize the world, create process 0, mount root
192 * filesystem, and fork to create init and pagedaemon. Most of the
193 * hard work is done in the lower-level initialization routines including
194 * startup(), which does memory initialization and autoconfiguration.
195 *
196 * This allows simple addition of new kernel subsystems that require
197 * boot time initialization. It also allows substitution of subsystem
198 * (for instance, a scheduler, kernel profiler, or VM system) by object
199 * module. Finally, it allows for optional "kernel threads".
200 */
201 void
mi_startup(void)202 mi_startup(void)
203 {
204
205 register struct sysinit **sipp; /* system initialization*/
206 register struct sysinit **xipp; /* interior loop of sort*/
207 register struct sysinit *save; /* bubble*/
208
209 #if defined(VERBOSE_SYSINIT)
210 int last;
211 int verbose;
212 #endif
213
214 if (boothowto & RB_VERBOSE)
215 bootverbose++;
216
217 if (sysinit == NULL) {
218 sysinit = SET_BEGIN(sysinit_set);
219 sysinit_end = SET_LIMIT(sysinit_set);
220 }
221
222 restart:
223 /*
224 * Perform a bubble sort of the system initialization objects by
225 * their subsystem (primary key) and order (secondary key).
226 */
227 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
228 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
229 if ((*sipp)->subsystem < (*xipp)->subsystem ||
230 ((*sipp)->subsystem == (*xipp)->subsystem &&
231 (*sipp)->order <= (*xipp)->order))
232 continue; /* skip*/
233 save = *sipp;
234 *sipp = *xipp;
235 *xipp = save;
236 }
237 }
238
239 #if defined(VERBOSE_SYSINIT)
240 last = SI_SUB_COPYRIGHT;
241 verbose = 0;
242 #if !defined(DDB)
243 printf("VERBOSE_SYSINIT: DDB not enabled, symbol lookups disabled.\n");
244 #endif
245 #endif
246
247 /*
248 * Traverse the (now) ordered list of system initialization tasks.
249 * Perform each task, and continue on to the next task.
250 */
251 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
252
253 if ((*sipp)->subsystem == SI_SUB_DUMMY)
254 continue; /* skip dummy task(s)*/
255
256 if ((*sipp)->subsystem == SI_SUB_DONE)
257 continue;
258
259 #if defined(VERBOSE_SYSINIT)
260 if ((*sipp)->subsystem > last) {
261 verbose = 1;
262 last = (*sipp)->subsystem;
263 printf("subsystem %x\n", last);
264 }
265 if (verbose) {
266 #if defined(DDB)
267 const char *func, *data;
268
269 func = symbol_name((vm_offset_t)(*sipp)->func,
270 DB_STGY_PROC);
271 data = symbol_name((vm_offset_t)(*sipp)->udata,
272 DB_STGY_ANY);
273 if (func != NULL && data != NULL)
274 printf(" %s(&%s)... ", func, data);
275 else if (func != NULL)
276 printf(" %s(%p)... ", func, (*sipp)->udata);
277 else
278 #endif
279 printf(" %p(%p)... ", (*sipp)->func,
280 (*sipp)->udata);
281 }
282 #endif
283
284 /* Call function */
285 (*((*sipp)->func))((*sipp)->udata);
286
287 #if defined(VERBOSE_SYSINIT)
288 if (verbose)
289 printf("done.\n");
290 #endif
291
292 /* Check off the one we're just done */
293 (*sipp)->subsystem = SI_SUB_DONE;
294
295 /* Check if we've installed more sysinit items via KLD */
296 if (newsysinit != NULL) {
297 if (sysinit != SET_BEGIN(sysinit_set))
298 free(sysinit, M_TEMP);
299 sysinit = newsysinit;
300 sysinit_end = newsysinit_end;
301 newsysinit = NULL;
302 newsysinit_end = NULL;
303 goto restart;
304 }
305 }
306
307 mtx_assert(&Giant, MA_OWNED | MA_NOTRECURSED);
308 mtx_unlock(&Giant);
309
310 /*
311 * Now hand over this thread to swapper.
312 */
313 swapper();
314 /* NOTREACHED*/
315 }
316
317
318 /*
319 ***************************************************************************
320 ****
321 **** The following SYSINIT's belong elsewhere, but have not yet
322 **** been moved.
323 ****
324 ***************************************************************************
325 */
326 static void
print_caddr_t(void * data)327 print_caddr_t(void *data)
328 {
329 printf("%s", (char *)data);
330 }
331
332 static void
print_version(void * data __unused)333 print_version(void *data __unused)
334 {
335 int len;
336
337 /* Strip a trailing newline from version. */
338 len = strlen(version);
339 while (len > 0 && version[len - 1] == '\n')
340 len--;
341 printf("%.*s %s\n", len, version, machine);
342 printf("%s\n", compiler_version);
343 }
344
345 SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t,
346 copyright);
347 SYSINIT(trademark, SI_SUB_COPYRIGHT, SI_ORDER_SECOND, print_caddr_t,
348 trademark);
349 SYSINIT(version, SI_SUB_COPYRIGHT, SI_ORDER_THIRD, print_version, NULL);
350
351 #ifdef WITNESS
352 static char wit_warn[] =
353 "WARNING: WITNESS option enabled, expect reduced performance.\n";
354 SYSINIT(witwarn, SI_SUB_COPYRIGHT, SI_ORDER_THIRD + 1,
355 print_caddr_t, wit_warn);
356 SYSINIT(witwarn2, SI_SUB_LAST, SI_ORDER_THIRD + 1,
357 print_caddr_t, wit_warn);
358 #endif
359
360 #ifdef DIAGNOSTIC
361 static char diag_warn[] =
362 "WARNING: DIAGNOSTIC option enabled, expect reduced performance.\n";
363 SYSINIT(diagwarn, SI_SUB_COPYRIGHT, SI_ORDER_THIRD + 2,
364 print_caddr_t, diag_warn);
365 SYSINIT(diagwarn2, SI_SUB_LAST, SI_ORDER_THIRD + 2,
366 print_caddr_t, diag_warn);
367 #endif
368
369 static int
null_fetch_syscall_args(struct thread * td __unused,struct syscall_args * sa __unused)370 null_fetch_syscall_args(struct thread *td __unused,
371 struct syscall_args *sa __unused)
372 {
373
374 panic("null_fetch_syscall_args");
375 }
376
377 static void
null_set_syscall_retval(struct thread * td __unused,int error __unused)378 null_set_syscall_retval(struct thread *td __unused, int error __unused)
379 {
380
381 panic("null_set_syscall_retval");
382 }
383
384 struct sysentvec null_sysvec = {
385 .sv_size = 0,
386 .sv_table = NULL,
387 .sv_mask = 0,
388 .sv_sigsize = 0,
389 .sv_sigtbl = NULL,
390 .sv_errsize = 0,
391 .sv_errtbl = NULL,
392 .sv_transtrap = NULL,
393 .sv_fixup = NULL,
394 .sv_sendsig = NULL,
395 .sv_sigcode = NULL,
396 .sv_szsigcode = NULL,
397 .sv_prepsyscall = NULL,
398 .sv_name = "null",
399 .sv_coredump = NULL,
400 .sv_imgact_try = NULL,
401 .sv_minsigstksz = 0,
402 .sv_pagesize = PAGE_SIZE,
403 .sv_minuser = VM_MIN_ADDRESS,
404 .sv_maxuser = VM_MAXUSER_ADDRESS,
405 .sv_usrstack = USRSTACK,
406 .sv_psstrings = PS_STRINGS,
407 .sv_stackprot = VM_PROT_ALL,
408 .sv_copyout_strings = NULL,
409 .sv_setregs = NULL,
410 .sv_fixlimit = NULL,
411 .sv_maxssiz = NULL,
412 .sv_flags = 0,
413 .sv_set_syscall_retval = null_set_syscall_retval,
414 .sv_fetch_syscall_args = null_fetch_syscall_args,
415 .sv_syscallnames = NULL,
416 .sv_schedtail = NULL,
417 .sv_thread_detach = NULL,
418 .sv_trap = NULL,
419 };
420
421 /*
422 ***************************************************************************
423 ****
424 **** The two following SYSINIT's are proc0 specific glue code. I am not
425 **** convinced that they can not be safely combined, but their order of
426 **** operation has been maintained as the same as the original init_main.c
427 **** for right now.
428 ****
429 **** These probably belong in init_proc.c or kern_proc.c, since they
430 **** deal with proc0 (the fork template process).
431 ****
432 ***************************************************************************
433 */
434 /* ARGSUSED*/
435 static void
proc0_init(void * dummy __unused)436 proc0_init(void *dummy __unused)
437 {
438 struct proc *p;
439 struct thread *td;
440 struct ucred *newcred;
441 vm_paddr_t pageablemem;
442 int i;
443
444 GIANT_REQUIRED;
445 p = &proc0;
446 td = &thread0;
447
448 /*
449 * Initialize magic number and osrel.
450 */
451 p->p_magic = P_MAGIC;
452 p->p_osrel = osreldate;
453
454 /*
455 * Initialize thread and process structures.
456 */
457 procinit(); /* set up proc zone */
458 threadinit(); /* set up UMA zones */
459
460 /*
461 * Initialise scheduler resources.
462 * Add scheduler specific parts to proc, thread as needed.
463 */
464 schedinit(); /* scheduler gets its house in order */
465
466 /*
467 * Create process 0 (the swapper).
468 */
469 LIST_INSERT_HEAD(&allproc, p, p_list);
470 LIST_INSERT_HEAD(PIDHASH(0), p, p_hash);
471 mtx_init(&pgrp0.pg_mtx, "process group", NULL, MTX_DEF | MTX_DUPOK);
472 p->p_pgrp = &pgrp0;
473 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
474 LIST_INIT(&pgrp0.pg_members);
475 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
476
477 pgrp0.pg_session = &session0;
478 mtx_init(&session0.s_mtx, "session", NULL, MTX_DEF);
479 refcount_init(&session0.s_count, 1);
480 session0.s_leader = p;
481
482 p->p_sysent = &null_sysvec;
483 p->p_flag = P_SYSTEM | P_INMEM | P_KTHREAD;
484 p->p_flag2 = 0;
485 p->p_state = PRS_NORMAL;
486 knlist_init_mtx(&p->p_klist, &p->p_mtx);
487 STAILQ_INIT(&p->p_ktr);
488 p->p_nice = NZERO;
489 /* pid_max cannot be greater than PID_MAX */
490 td->td_tid = PID_MAX + 1;
491 LIST_INSERT_HEAD(TIDHASH(td->td_tid), td, td_hash);
492 td->td_state = TDS_RUNNING;
493 td->td_pri_class = PRI_TIMESHARE;
494 td->td_user_pri = PUSER;
495 td->td_base_user_pri = PUSER;
496 td->td_lend_user_pri = PRI_MAX;
497 td->td_priority = PVM;
498 td->td_base_pri = PVM;
499 td->td_oncpu = 0;
500 td->td_flags = TDF_INMEM;
501 td->td_pflags = TDP_KTHREAD;
502 td->td_cpuset = cpuset_thread0();
503 prison0_init();
504 p->p_peers = 0;
505 p->p_leader = p;
506 p->p_reaper = p;
507 LIST_INIT(&p->p_reaplist);
508
509 strncpy(p->p_comm, "kernel", sizeof (p->p_comm));
510 strncpy(td->td_name, "swapper", sizeof (td->td_name));
511
512 callout_init_mtx(&p->p_itcallout, &p->p_mtx, 0);
513 callout_init_mtx(&p->p_limco, &p->p_mtx, 0);
514 callout_init(&td->td_slpcallout, 1);
515
516 /* Create credentials. */
517 newcred = crget();
518 newcred->cr_ngroups = 1; /* group 0 */
519 newcred->cr_uidinfo = uifind(0);
520 newcred->cr_ruidinfo = uifind(0);
521 newcred->cr_prison = &prison0;
522 newcred->cr_loginclass = loginclass_find("default");
523 proc_set_cred_init(p, newcred);
524 #ifdef AUDIT
525 audit_cred_kproc0(newcred);
526 #endif
527 #ifdef MAC
528 mac_cred_create_swapper(newcred);
529 #endif
530 td->td_ucred = crhold(newcred);
531
532 /* Create sigacts. */
533 p->p_sigacts = sigacts_alloc();
534
535 /* Initialize signal state for process 0. */
536 siginit(&proc0);
537
538 /* Create the file descriptor table. */
539 p->p_fd = fdinit(NULL);
540 p->p_fdtol = NULL;
541
542 /* Create the limits structures. */
543 p->p_limit = lim_alloc();
544 for (i = 0; i < RLIM_NLIMITS; i++)
545 p->p_limit->pl_rlimit[i].rlim_cur =
546 p->p_limit->pl_rlimit[i].rlim_max = RLIM_INFINITY;
547 p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_cur =
548 p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
549 p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_cur =
550 p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
551 p->p_limit->pl_rlimit[RLIMIT_DATA].rlim_cur = dfldsiz;
552 p->p_limit->pl_rlimit[RLIMIT_DATA].rlim_max = maxdsiz;
553 p->p_limit->pl_rlimit[RLIMIT_STACK].rlim_cur = dflssiz;
554 p->p_limit->pl_rlimit[RLIMIT_STACK].rlim_max = maxssiz;
555 /* Cast to avoid overflow on i386/PAE. */
556 pageablemem = ptoa((vm_paddr_t)cnt.v_free_count);
557 p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_cur =
558 p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_max = pageablemem;
559 p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = pageablemem / 3;
560 p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_max = pageablemem;
561 p->p_cpulimit = RLIM_INFINITY;
562
563 /* Initialize resource accounting structures. */
564 racct_create(&p->p_racct);
565
566 p->p_stats = pstats_alloc();
567
568 /* Allocate a prototype map so we have something to fork. */
569 pmap_pinit0(vmspace_pmap(&vmspace0));
570 p->p_vmspace = &vmspace0;
571 vmspace0.vm_refcnt = 1;
572
573 /*
574 * proc0 is not expected to enter usermode, so there is no special
575 * handling for sv_minuser here, like is done for exec_new_vmspace().
576 */
577 vm_map_init(&vmspace0.vm_map, vmspace_pmap(&vmspace0),
578 p->p_sysent->sv_minuser, p->p_sysent->sv_maxuser);
579
580 /*
581 * Call the init and ctor for the new thread and proc. We wait
582 * to do this until all other structures are fairly sane.
583 */
584 EVENTHANDLER_INVOKE(process_init, p);
585 EVENTHANDLER_INVOKE(thread_init, td);
586 EVENTHANDLER_INVOKE(process_ctor, p);
587 EVENTHANDLER_INVOKE(thread_ctor, td);
588
589 /*
590 * Charge root for one process.
591 */
592 (void)chgproccnt(p->p_ucred->cr_ruidinfo, 1, 0);
593 PROC_LOCK(p);
594 racct_add_force(p, RACCT_NPROC, 1);
595 PROC_UNLOCK(p);
596 }
597 SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL);
598
599 /* ARGSUSED*/
600 static void
proc0_post(void * dummy __unused)601 proc0_post(void *dummy __unused)
602 {
603 struct timespec ts;
604 struct proc *p;
605 struct rusage ru;
606 struct thread *td;
607
608 /*
609 * Now we can look at the time, having had a chance to verify the
610 * time from the filesystem. Pretend that proc0 started now.
611 */
612 sx_slock(&allproc_lock);
613 FOREACH_PROC_IN_SYSTEM(p) {
614 PROC_LOCK(p);
615 if (p->p_state == PRS_NEW) {
616 PROC_UNLOCK(p);
617 continue;
618 }
619 microuptime(&p->p_stats->p_start);
620 PROC_STATLOCK(p);
621 rufetch(p, &ru); /* Clears thread stats */
622 p->p_rux.rux_runtime = 0;
623 p->p_rux.rux_uticks = 0;
624 p->p_rux.rux_sticks = 0;
625 p->p_rux.rux_iticks = 0;
626 PROC_STATUNLOCK(p);
627 FOREACH_THREAD_IN_PROC(p, td) {
628 td->td_runtime = 0;
629 }
630 PROC_UNLOCK(p);
631 }
632 sx_sunlock(&allproc_lock);
633 PCPU_SET(switchtime, cpu_ticks());
634 PCPU_SET(switchticks, ticks);
635
636 /*
637 * Give the ``random'' number generator a thump.
638 */
639 nanotime(&ts);
640 srandom(ts.tv_sec ^ ts.tv_nsec);
641 }
642 SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL);
643
644 static void
random_init(void * dummy __unused)645 random_init(void *dummy __unused)
646 {
647
648 /*
649 * After CPU has been started we have some randomness on most
650 * platforms via get_cyclecount(). For platforms that don't
651 * we will reseed random(9) in proc0_post() as well.
652 */
653 srandom(get_cyclecount());
654 }
655 SYSINIT(random, SI_SUB_RANDOM, SI_ORDER_FIRST, random_init, NULL);
656
657 /*
658 ***************************************************************************
659 ****
660 **** The following SYSINIT's and glue code should be moved to the
661 **** respective files on a per subsystem basis.
662 ****
663 ***************************************************************************
664 */
665
666
667 /*
668 ***************************************************************************
669 ****
670 **** The following code probably belongs in another file, like
671 **** kern/init_init.c.
672 ****
673 ***************************************************************************
674 */
675
676 /*
677 * List of paths to try when searching for "init".
678 */
679 static char init_path[MAXPATHLEN] =
680 #ifdef INIT_PATH
681 __XSTRING(INIT_PATH);
682 #else
683 "/sbin/init:/sbin/oinit:/sbin/init.bak:/rescue/init";
684 #endif
685 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0,
686 "Path used to search the init process");
687
688 /*
689 * Shutdown timeout of init(8).
690 * Unused within kernel, but used to control init(8), hence do not remove.
691 */
692 #ifndef INIT_SHUTDOWN_TIMEOUT
693 #define INIT_SHUTDOWN_TIMEOUT 120
694 #endif
695 static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT;
696 SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout,
697 CTLFLAG_RW, &init_shutdown_timeout, 0, "Shutdown timeout of init(8). "
698 "Unused within kernel, but used to control init(8)");
699
700 /*
701 * Start the initial user process; try exec'ing each pathname in init_path.
702 * The program is invoked with one argument containing the boot flags.
703 */
704 static void
start_init(void * dummy)705 start_init(void *dummy)
706 {
707 vm_offset_t addr;
708 struct execve_args args;
709 int options, error;
710 char *var, *path, *next, *s;
711 char *ucp, **uap, *arg0, *arg1;
712 struct thread *td;
713 struct proc *p;
714
715 mtx_lock(&Giant);
716
717 GIANT_REQUIRED;
718
719 td = curthread;
720 p = td->td_proc;
721
722 vfs_mountroot();
723
724 /* Wipe GELI passphrase from the environment. */
725 unsetenv("kern.geom.eli.passphrase");
726
727 /*
728 * Need just enough stack to hold the faked-up "execve()" arguments.
729 */
730 addr = p->p_sysent->sv_usrstack - PAGE_SIZE;
731 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE, 0,
732 VMFS_NO_SPACE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0)
733 panic("init: couldn't allocate argument space");
734 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
735 p->p_vmspace->vm_ssize = 1;
736
737 if ((var = getenv("init_path")) != NULL) {
738 strlcpy(init_path, var, sizeof(init_path));
739 freeenv(var);
740 }
741
742 for (path = init_path; *path != '\0'; path = next) {
743 while (*path == ':')
744 path++;
745 if (*path == '\0')
746 break;
747 for (next = path; *next != '\0' && *next != ':'; next++)
748 /* nothing */ ;
749 if (bootverbose)
750 printf("start_init: trying %.*s\n", (int)(next - path),
751 path);
752
753 /*
754 * Move out the boot flag argument.
755 */
756 options = 0;
757 ucp = (char *)p->p_sysent->sv_usrstack;
758 (void)subyte(--ucp, 0); /* trailing zero */
759 if (boothowto & RB_SINGLE) {
760 (void)subyte(--ucp, 's');
761 options = 1;
762 }
763 #ifdef notyet
764 if (boothowto & RB_FASTBOOT) {
765 (void)subyte(--ucp, 'f');
766 options = 1;
767 }
768 #endif
769
770 #ifdef BOOTCDROM
771 (void)subyte(--ucp, 'C');
772 options = 1;
773 #endif
774
775 if (options == 0)
776 (void)subyte(--ucp, '-');
777 (void)subyte(--ucp, '-'); /* leading hyphen */
778 arg1 = ucp;
779
780 /*
781 * Move out the file name (also arg 0).
782 */
783 (void)subyte(--ucp, 0);
784 for (s = next - 1; s >= path; s--)
785 (void)subyte(--ucp, *s);
786 arg0 = ucp;
787
788 /*
789 * Move out the arg pointers.
790 */
791 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
792 (void)suword((caddr_t)--uap, (long)0); /* terminator */
793 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
794 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
795
796 /*
797 * Point at the arguments.
798 */
799 args.fname = arg0;
800 args.argv = uap;
801 args.envv = NULL;
802
803 /*
804 * Now try to exec the program. If can't for any reason
805 * other than it doesn't exist, complain.
806 *
807 * Otherwise, return via fork_trampoline() all the way
808 * to user mode as init!
809 */
810 if ((error = sys_execve(td, &args)) == 0) {
811 mtx_unlock(&Giant);
812 return;
813 }
814 if (error != ENOENT)
815 printf("exec %.*s: error %d\n", (int)(next - path),
816 path, error);
817 }
818 printf("init: not found in path %s\n", init_path);
819 panic("no init");
820 }
821
822 /*
823 * Like kproc_create(), but runs in it's own address space.
824 * We do this early to reserve pid 1.
825 *
826 * Note special case - do not make it runnable yet. Other work
827 * in progress will change this more.
828 */
829 static void
create_init(const void * udata __unused)830 create_init(const void *udata __unused)
831 {
832 struct ucred *newcred, *oldcred;
833 int error;
834
835 error = fork1(&thread0, RFFDG | RFPROC | RFSTOPPED, 0, &initproc,
836 NULL, 0);
837 if (error)
838 panic("cannot fork init: %d\n", error);
839 KASSERT(initproc->p_pid == 1, ("create_init: initproc->p_pid != 1"));
840 /* divorce init's credentials from the kernel's */
841 newcred = crget();
842 sx_xlock(&proctree_lock);
843 PROC_LOCK(initproc);
844 initproc->p_flag |= P_SYSTEM | P_INMEM;
845 initproc->p_treeflag |= P_TREE_REAPER;
846 LIST_INSERT_HEAD(&initproc->p_reaplist, &proc0, p_reapsibling);
847 oldcred = initproc->p_ucred;
848 crcopy(newcred, oldcred);
849 #ifdef MAC
850 mac_cred_create_init(newcred);
851 #endif
852 #ifdef AUDIT
853 audit_cred_proc1(newcred);
854 #endif
855 proc_set_cred(initproc, newcred);
856 PROC_UNLOCK(initproc);
857 sx_xunlock(&proctree_lock);
858 crfree(oldcred);
859 cred_update_thread(FIRST_THREAD_IN_PROC(initproc));
860 cpu_set_fork_handler(FIRST_THREAD_IN_PROC(initproc), start_init, NULL);
861 }
862 SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL);
863
864 /*
865 * Make it runnable now.
866 */
867 static void
kick_init(const void * udata __unused)868 kick_init(const void *udata __unused)
869 {
870 struct thread *td;
871
872 td = FIRST_THREAD_IN_PROC(initproc);
873 thread_lock(td);
874 TD_SET_CAN_RUN(td);
875 sched_add(td, SRQ_BORING);
876 thread_unlock(td);
877 }
878 SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL);
879