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("$FreeBSD$"); 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/mount.h> 59 #include <sys/mutex.h> 60 #include <sys/syscallsubr.h> 61 #include <sys/sysctl.h> 62 #include <sys/proc.h> 63 #include <sys/resourcevar.h> 64 #include <sys/systm.h> 65 #include <sys/signalvar.h> 66 #include <sys/vnode.h> 67 #include <sys/sysent.h> 68 #include <sys/reboot.h> 69 #include <sys/sched.h> 70 #include <sys/sx.h> 71 #include <sys/sysproto.h> 72 #include <sys/vmmeter.h> 73 #include <sys/unistd.h> 74 #include <sys/malloc.h> 75 #include <sys/conf.h> 76 #include <sys/cpuset.h> 77 78 #include <machine/cpu.h> 79 80 #include <security/audit/audit.h> 81 #include <security/mac/mac_framework.h> 82 83 #include <vm/vm.h> 84 #include <vm/vm_param.h> 85 #include <vm/pmap.h> 86 #include <vm/vm_map.h> 87 #include <sys/copyright.h> 88 89 #include <ddb/ddb.h> 90 #include <ddb/db_sym.h> 91 92 void mi_startup(void); /* Should be elsewhere */ 93 94 /* Components of the first process -- never freed. */ 95 static struct session session0; 96 static struct pgrp pgrp0; 97 struct proc proc0; 98 struct thread thread0 __aligned(16); 99 struct vmspace vmspace0; 100 struct proc *initproc; 101 102 int boothowto = 0; /* initialized so that it can be patched */ 103 SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, ""); 104 int bootverbose; 105 SYSCTL_INT(_debug, OID_AUTO, bootverbose, CTLFLAG_RW, &bootverbose, 0, ""); 106 107 /* 108 * This ensures that there is at least one entry so that the sysinit_set 109 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never 110 * executed. 111 */ 112 SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL); 113 114 /* 115 * The sysinit table itself. Items are checked off as the are run. 116 * If we want to register new sysinit types, add them to newsysinit. 117 */ 118 SET_DECLARE(sysinit_set, struct sysinit); 119 struct sysinit **sysinit, **sysinit_end; 120 struct sysinit **newsysinit, **newsysinit_end; 121 122 /* 123 * Merge a new sysinit set into the current set, reallocating it if 124 * necessary. This can only be called after malloc is running. 125 */ 126 void 127 sysinit_add(struct sysinit **set, struct sysinit **set_end) 128 { 129 struct sysinit **newset; 130 struct sysinit **sipp; 131 struct sysinit **xipp; 132 int count; 133 134 count = set_end - set; 135 if (newsysinit) 136 count += newsysinit_end - newsysinit; 137 else 138 count += sysinit_end - sysinit; 139 newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT); 140 if (newset == NULL) 141 panic("cannot malloc for sysinit"); 142 xipp = newset; 143 if (newsysinit) 144 for (sipp = newsysinit; sipp < newsysinit_end; sipp++) 145 *xipp++ = *sipp; 146 else 147 for (sipp = sysinit; sipp < sysinit_end; sipp++) 148 *xipp++ = *sipp; 149 for (sipp = set; sipp < set_end; sipp++) 150 *xipp++ = *sipp; 151 if (newsysinit) 152 free(newsysinit, M_TEMP); 153 newsysinit = newset; 154 newsysinit_end = newset + count; 155 } 156 157 /* 158 * System startup; initialize the world, create process 0, mount root 159 * filesystem, and fork to create init and pagedaemon. Most of the 160 * hard work is done in the lower-level initialization routines including 161 * startup(), which does memory initialization and autoconfiguration. 162 * 163 * This allows simple addition of new kernel subsystems that require 164 * boot time initialization. It also allows substitution of subsystem 165 * (for instance, a scheduler, kernel profiler, or VM system) by object 166 * module. Finally, it allows for optional "kernel threads". 167 */ 168 void 169 mi_startup(void) 170 { 171 172 register struct sysinit **sipp; /* system initialization*/ 173 register struct sysinit **xipp; /* interior loop of sort*/ 174 register struct sysinit *save; /* bubble*/ 175 176 #if defined(VERBOSE_SYSINIT) 177 int last; 178 int verbose; 179 #endif 180 181 if (sysinit == NULL) { 182 sysinit = SET_BEGIN(sysinit_set); 183 sysinit_end = SET_LIMIT(sysinit_set); 184 } 185 186 restart: 187 /* 188 * Perform a bubble sort of the system initialization objects by 189 * their subsystem (primary key) and order (secondary key). 190 */ 191 for (sipp = sysinit; sipp < sysinit_end; sipp++) { 192 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) { 193 if ((*sipp)->subsystem < (*xipp)->subsystem || 194 ((*sipp)->subsystem == (*xipp)->subsystem && 195 (*sipp)->order <= (*xipp)->order)) 196 continue; /* skip*/ 197 save = *sipp; 198 *sipp = *xipp; 199 *xipp = save; 200 } 201 } 202 203 #if defined(VERBOSE_SYSINIT) 204 last = SI_SUB_COPYRIGHT; 205 verbose = 0; 206 #if !defined(DDB) 207 printf("VERBOSE_SYSINIT: DDB not enabled, symbol lookups disabled.\n"); 208 #endif 209 #endif 210 211 /* 212 * Traverse the (now) ordered list of system initialization tasks. 213 * Perform each task, and continue on to the next task. 214 * 215 * The last item on the list is expected to be the scheduler, 216 * which will not return. 217 */ 218 for (sipp = sysinit; sipp < sysinit_end; sipp++) { 219 220 if ((*sipp)->subsystem == SI_SUB_DUMMY) 221 continue; /* skip dummy task(s)*/ 222 223 if ((*sipp)->subsystem == SI_SUB_DONE) 224 continue; 225 226 #if defined(VERBOSE_SYSINIT) 227 if ((*sipp)->subsystem > last) { 228 verbose = 1; 229 last = (*sipp)->subsystem; 230 printf("subsystem %x\n", last); 231 } 232 if (verbose) { 233 #if defined(DDB) 234 const char *name; 235 c_db_sym_t sym; 236 db_expr_t offset; 237 238 sym = db_search_symbol((vm_offset_t)(*sipp)->func, 239 DB_STGY_PROC, &offset); 240 db_symbol_values(sym, &name, NULL); 241 if (name != NULL) 242 printf(" %s(%p)... ", name, (*sipp)->udata); 243 else 244 #endif 245 printf(" %p(%p)... ", (*sipp)->func, 246 (*sipp)->udata); 247 } 248 #endif 249 250 /* Call function */ 251 (*((*sipp)->func))((*sipp)->udata); 252 253 #if defined(VERBOSE_SYSINIT) 254 if (verbose) 255 printf("done.\n"); 256 #endif 257 258 /* Check off the one we're just done */ 259 (*sipp)->subsystem = SI_SUB_DONE; 260 261 /* Check if we've installed more sysinit items via KLD */ 262 if (newsysinit != NULL) { 263 if (sysinit != SET_BEGIN(sysinit_set)) 264 free(sysinit, M_TEMP); 265 sysinit = newsysinit; 266 sysinit_end = newsysinit_end; 267 newsysinit = NULL; 268 newsysinit_end = NULL; 269 goto restart; 270 } 271 } 272 273 panic("Shouldn't get here!"); 274 /* NOTREACHED*/ 275 } 276 277 278 /* 279 *************************************************************************** 280 **** 281 **** The following SYSINIT's belong elsewhere, but have not yet 282 **** been moved. 283 **** 284 *************************************************************************** 285 */ 286 static void 287 print_caddr_t(void *data) 288 { 289 printf("%s", (char *)data); 290 } 291 292 static void 293 print_version(void *data __unused) 294 { 295 int len; 296 297 /* Strip a trailing newline from version. */ 298 len = strlen(version); 299 while (len > 0 && version[len - 1] == '\n') 300 len--; 301 printf("%.*s %s\n", len, version, machine); 302 } 303 304 SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, 305 copyright); 306 SYSINIT(trademark, SI_SUB_COPYRIGHT, SI_ORDER_SECOND, print_caddr_t, 307 trademark); 308 SYSINIT(version, SI_SUB_COPYRIGHT, SI_ORDER_THIRD, print_version, NULL); 309 310 #ifdef WITNESS 311 static char wit_warn[] = 312 "WARNING: WITNESS option enabled, expect reduced performance.\n"; 313 SYSINIT(witwarn, SI_SUB_COPYRIGHT, SI_ORDER_THIRD + 1, 314 print_caddr_t, wit_warn); 315 SYSINIT(witwarn2, SI_SUB_RUN_SCHEDULER, SI_ORDER_THIRD + 1, 316 print_caddr_t, wit_warn); 317 #endif 318 319 #ifdef DIAGNOSTIC 320 static char diag_warn[] = 321 "WARNING: DIAGNOSTIC option enabled, expect reduced performance.\n"; 322 SYSINIT(diagwarn, SI_SUB_COPYRIGHT, SI_ORDER_THIRD + 2, 323 print_caddr_t, diag_warn); 324 SYSINIT(diagwarn2, SI_SUB_RUN_SCHEDULER, SI_ORDER_THIRD + 2, 325 print_caddr_t, diag_warn); 326 #endif 327 328 static void 329 set_boot_verbose(void *data __unused) 330 { 331 332 if (boothowto & RB_VERBOSE) 333 bootverbose++; 334 } 335 SYSINIT(boot_verbose, SI_SUB_TUNABLES, SI_ORDER_ANY, set_boot_verbose, NULL); 336 337 struct sysentvec null_sysvec = { 338 .sv_size = 0, 339 .sv_table = NULL, 340 .sv_mask = 0, 341 .sv_sigsize = 0, 342 .sv_sigtbl = NULL, 343 .sv_errsize = 0, 344 .sv_errtbl = NULL, 345 .sv_transtrap = NULL, 346 .sv_fixup = NULL, 347 .sv_sendsig = NULL, 348 .sv_sigcode = NULL, 349 .sv_szsigcode = NULL, 350 .sv_prepsyscall = NULL, 351 .sv_name = "null", 352 .sv_coredump = NULL, 353 .sv_imgact_try = NULL, 354 .sv_minsigstksz = 0, 355 .sv_pagesize = PAGE_SIZE, 356 .sv_minuser = VM_MIN_ADDRESS, 357 .sv_maxuser = VM_MAXUSER_ADDRESS, 358 .sv_usrstack = USRSTACK, 359 .sv_psstrings = PS_STRINGS, 360 .sv_stackprot = VM_PROT_ALL, 361 .sv_copyout_strings = NULL, 362 .sv_setregs = NULL, 363 .sv_fixlimit = NULL, 364 .sv_maxssiz = NULL 365 }; 366 367 /* 368 *************************************************************************** 369 **** 370 **** The two following SYSINIT's are proc0 specific glue code. I am not 371 **** convinced that they can not be safely combined, but their order of 372 **** operation has been maintained as the same as the original init_main.c 373 **** for right now. 374 **** 375 **** These probably belong in init_proc.c or kern_proc.c, since they 376 **** deal with proc0 (the fork template process). 377 **** 378 *************************************************************************** 379 */ 380 /* ARGSUSED*/ 381 static void 382 proc0_init(void *dummy __unused) 383 { 384 struct proc *p; 385 struct thread *td; 386 vm_paddr_t pageablemem; 387 int i; 388 389 GIANT_REQUIRED; 390 p = &proc0; 391 td = &thread0; 392 393 /* 394 * Initialize magic number and osrel. 395 */ 396 p->p_magic = P_MAGIC; 397 p->p_osrel = osreldate; 398 399 /* 400 * Initialize thread and process structures. 401 */ 402 procinit(); /* set up proc zone */ 403 threadinit(); /* set up UMA zones */ 404 405 /* 406 * Initialise scheduler resources. 407 * Add scheduler specific parts to proc, thread as needed. 408 */ 409 schedinit(); /* scheduler gets its house in order */ 410 /* 411 * Initialize sleep queue hash table 412 */ 413 sleepinit(); 414 415 /* 416 * additional VM structures 417 */ 418 vm_init2(); 419 420 /* 421 * Create process 0 (the swapper). 422 */ 423 LIST_INSERT_HEAD(&allproc, p, p_list); 424 LIST_INSERT_HEAD(PIDHASH(0), p, p_hash); 425 mtx_init(&pgrp0.pg_mtx, "process group", NULL, MTX_DEF | MTX_DUPOK); 426 p->p_pgrp = &pgrp0; 427 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash); 428 LIST_INIT(&pgrp0.pg_members); 429 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist); 430 431 pgrp0.pg_session = &session0; 432 mtx_init(&session0.s_mtx, "session", NULL, MTX_DEF); 433 refcount_init(&session0.s_count, 1); 434 session0.s_leader = p; 435 436 p->p_sysent = &null_sysvec; 437 p->p_flag = P_SYSTEM | P_INMEM; 438 p->p_state = PRS_NORMAL; 439 knlist_init_mtx(&p->p_klist, &p->p_mtx); 440 STAILQ_INIT(&p->p_ktr); 441 p->p_nice = NZERO; 442 td->td_tid = PID_MAX + 1; 443 td->td_state = TDS_RUNNING; 444 td->td_pri_class = PRI_TIMESHARE; 445 td->td_user_pri = PUSER; 446 td->td_base_user_pri = PUSER; 447 td->td_priority = PVM; 448 td->td_base_pri = PUSER; 449 td->td_oncpu = 0; 450 td->td_flags = TDF_INMEM|TDP_KTHREAD; 451 td->td_cpuset = cpuset_thread0(); 452 prison0.pr_cpuset = cpuset_ref(td->td_cpuset); 453 p->p_peers = 0; 454 p->p_leader = p; 455 456 457 strncpy(p->p_comm, "kernel", sizeof (p->p_comm)); 458 strncpy(td->td_name, "swapper", sizeof (td->td_name)); 459 460 callout_init(&p->p_itcallout, CALLOUT_MPSAFE); 461 callout_init_mtx(&p->p_limco, &p->p_mtx, 0); 462 callout_init(&td->td_slpcallout, CALLOUT_MPSAFE); 463 464 /* Create credentials. */ 465 p->p_ucred = crget(); 466 p->p_ucred->cr_ngroups = 1; /* group 0 */ 467 p->p_ucred->cr_uidinfo = uifind(0); 468 p->p_ucred->cr_ruidinfo = uifind(0); 469 p->p_ucred->cr_prison = &prison0; 470 #ifdef AUDIT 471 audit_cred_kproc0(p->p_ucred); 472 #endif 473 #ifdef MAC 474 mac_cred_create_swapper(p->p_ucred); 475 #endif 476 td->td_ucred = crhold(p->p_ucred); 477 478 /* Create sigacts. */ 479 p->p_sigacts = sigacts_alloc(); 480 481 /* Initialize signal state for process 0. */ 482 siginit(&proc0); 483 484 /* Create the file descriptor table. */ 485 p->p_fd = fdinit(NULL); 486 p->p_fdtol = NULL; 487 488 /* Create the limits structures. */ 489 p->p_limit = lim_alloc(); 490 for (i = 0; i < RLIM_NLIMITS; i++) 491 p->p_limit->pl_rlimit[i].rlim_cur = 492 p->p_limit->pl_rlimit[i].rlim_max = RLIM_INFINITY; 493 p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_cur = 494 p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles; 495 p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_cur = 496 p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc; 497 p->p_limit->pl_rlimit[RLIMIT_DATA].rlim_cur = dfldsiz; 498 p->p_limit->pl_rlimit[RLIMIT_DATA].rlim_max = maxdsiz; 499 p->p_limit->pl_rlimit[RLIMIT_STACK].rlim_cur = dflssiz; 500 p->p_limit->pl_rlimit[RLIMIT_STACK].rlim_max = maxssiz; 501 /* Cast to avoid overflow on i386/PAE. */ 502 pageablemem = ptoa((vm_paddr_t)cnt.v_free_count); 503 p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_cur = 504 p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_max = pageablemem; 505 p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = pageablemem / 3; 506 p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_max = pageablemem; 507 p->p_cpulimit = RLIM_INFINITY; 508 509 p->p_stats = pstats_alloc(); 510 511 /* Allocate a prototype map so we have something to fork. */ 512 pmap_pinit0(vmspace_pmap(&vmspace0)); 513 p->p_vmspace = &vmspace0; 514 vmspace0.vm_refcnt = 1; 515 516 /* 517 * proc0 is not expected to enter usermode, so there is no special 518 * handling for sv_minuser here, like is done for exec_new_vmspace(). 519 */ 520 vm_map_init(&vmspace0.vm_map, vmspace_pmap(&vmspace0), 521 p->p_sysent->sv_minuser, p->p_sysent->sv_maxuser); 522 523 /*- 524 * call the init and ctor for the new thread and proc 525 * we wait to do this until all other structures 526 * are fairly sane. 527 */ 528 EVENTHANDLER_INVOKE(process_init, p); 529 EVENTHANDLER_INVOKE(thread_init, td); 530 EVENTHANDLER_INVOKE(process_ctor, p); 531 EVENTHANDLER_INVOKE(thread_ctor, td); 532 533 /* 534 * Charge root for one process. 535 */ 536 (void)chgproccnt(p->p_ucred->cr_ruidinfo, 1, 0); 537 } 538 SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL); 539 540 /* ARGSUSED*/ 541 static void 542 proc0_post(void *dummy __unused) 543 { 544 struct timespec ts; 545 struct proc *p; 546 struct rusage ru; 547 struct thread *td; 548 549 /* 550 * Now we can look at the time, having had a chance to verify the 551 * time from the filesystem. Pretend that proc0 started now. 552 */ 553 sx_slock(&allproc_lock); 554 FOREACH_PROC_IN_SYSTEM(p) { 555 microuptime(&p->p_stats->p_start); 556 PROC_SLOCK(p); 557 rufetch(p, &ru); /* Clears thread stats */ 558 PROC_SUNLOCK(p); 559 p->p_rux.rux_runtime = 0; 560 p->p_rux.rux_uticks = 0; 561 p->p_rux.rux_sticks = 0; 562 p->p_rux.rux_iticks = 0; 563 FOREACH_THREAD_IN_PROC(p, td) { 564 td->td_runtime = 0; 565 } 566 } 567 sx_sunlock(&allproc_lock); 568 PCPU_SET(switchtime, cpu_ticks()); 569 PCPU_SET(switchticks, ticks); 570 571 /* 572 * Give the ``random'' number generator a thump. 573 */ 574 nanotime(&ts); 575 srandom(ts.tv_sec ^ ts.tv_nsec); 576 } 577 SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL); 578 579 static void 580 random_init(void *dummy __unused) 581 { 582 583 /* 584 * After CPU has been started we have some randomness on most 585 * platforms via get_cyclecount(). For platforms that don't 586 * we will reseed random(9) in proc0_post() as well. 587 */ 588 srandom(get_cyclecount()); 589 } 590 SYSINIT(random, SI_SUB_RANDOM, SI_ORDER_FIRST, random_init, NULL); 591 592 /* 593 *************************************************************************** 594 **** 595 **** The following SYSINIT's and glue code should be moved to the 596 **** respective files on a per subsystem basis. 597 **** 598 *************************************************************************** 599 */ 600 601 602 /* 603 *************************************************************************** 604 **** 605 **** The following code probably belongs in another file, like 606 **** kern/init_init.c. 607 **** 608 *************************************************************************** 609 */ 610 611 /* 612 * List of paths to try when searching for "init". 613 */ 614 static char init_path[MAXPATHLEN] = 615 #ifdef INIT_PATH 616 __XSTRING(INIT_PATH); 617 #else 618 "/sbin/init:/sbin/oinit:/sbin/init.bak:/rescue/init:/stand/sysinstall"; 619 #endif 620 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, 621 "Path used to search the init process"); 622 623 /* 624 * Shutdown timeout of init(8). 625 * Unused within kernel, but used to control init(8), hence do not remove. 626 */ 627 #ifndef INIT_SHUTDOWN_TIMEOUT 628 #define INIT_SHUTDOWN_TIMEOUT 120 629 #endif 630 static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT; 631 SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout, 632 CTLFLAG_RW, &init_shutdown_timeout, 0, ""); 633 634 /* 635 * Start the initial user process; try exec'ing each pathname in init_path. 636 * The program is invoked with one argument containing the boot flags. 637 */ 638 static void 639 start_init(void *dummy) 640 { 641 vm_offset_t addr; 642 struct execve_args args; 643 int options, error; 644 char *var, *path, *next, *s; 645 char *ucp, **uap, *arg0, *arg1; 646 struct thread *td; 647 struct proc *p; 648 649 mtx_lock(&Giant); 650 651 GIANT_REQUIRED; 652 653 td = curthread; 654 p = td->td_proc; 655 656 vfs_mountroot(); 657 658 /* 659 * Need just enough stack to hold the faked-up "execve()" arguments. 660 */ 661 addr = p->p_sysent->sv_usrstack - PAGE_SIZE; 662 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE, 663 FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0) 664 panic("init: couldn't allocate argument space"); 665 p->p_vmspace->vm_maxsaddr = (caddr_t)addr; 666 p->p_vmspace->vm_ssize = 1; 667 668 if ((var = getenv("init_path")) != NULL) { 669 strlcpy(init_path, var, sizeof(init_path)); 670 freeenv(var); 671 } 672 673 for (path = init_path; *path != '\0'; path = next) { 674 while (*path == ':') 675 path++; 676 if (*path == '\0') 677 break; 678 for (next = path; *next != '\0' && *next != ':'; next++) 679 /* nothing */ ; 680 if (bootverbose) 681 printf("start_init: trying %.*s\n", (int)(next - path), 682 path); 683 684 /* 685 * Move out the boot flag argument. 686 */ 687 options = 0; 688 ucp = (char *)p->p_sysent->sv_usrstack; 689 (void)subyte(--ucp, 0); /* trailing zero */ 690 if (boothowto & RB_SINGLE) { 691 (void)subyte(--ucp, 's'); 692 options = 1; 693 } 694 #ifdef notyet 695 if (boothowto & RB_FASTBOOT) { 696 (void)subyte(--ucp, 'f'); 697 options = 1; 698 } 699 #endif 700 701 #ifdef BOOTCDROM 702 (void)subyte(--ucp, 'C'); 703 options = 1; 704 #endif 705 706 if (options == 0) 707 (void)subyte(--ucp, '-'); 708 (void)subyte(--ucp, '-'); /* leading hyphen */ 709 arg1 = ucp; 710 711 /* 712 * Move out the file name (also arg 0). 713 */ 714 (void)subyte(--ucp, 0); 715 for (s = next - 1; s >= path; s--) 716 (void)subyte(--ucp, *s); 717 arg0 = ucp; 718 719 /* 720 * Move out the arg pointers. 721 */ 722 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1)); 723 (void)suword((caddr_t)--uap, (long)0); /* terminator */ 724 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1); 725 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0); 726 727 /* 728 * Point at the arguments. 729 */ 730 args.fname = arg0; 731 args.argv = uap; 732 args.envv = NULL; 733 734 /* 735 * Now try to exec the program. If can't for any reason 736 * other than it doesn't exist, complain. 737 * 738 * Otherwise, return via fork_trampoline() all the way 739 * to user mode as init! 740 */ 741 if ((error = execve(td, &args)) == 0) { 742 mtx_unlock(&Giant); 743 return; 744 } 745 if (error != ENOENT) 746 printf("exec %.*s: error %d\n", (int)(next - path), 747 path, error); 748 } 749 printf("init: not found in path %s\n", init_path); 750 panic("no init"); 751 } 752 753 /* 754 * Like kproc_create(), but runs in it's own address space. 755 * We do this early to reserve pid 1. 756 * 757 * Note special case - do not make it runnable yet. Other work 758 * in progress will change this more. 759 */ 760 static void 761 create_init(const void *udata __unused) 762 { 763 struct ucred *newcred, *oldcred; 764 int error; 765 766 error = fork1(&thread0, RFFDG | RFPROC | RFSTOPPED, 0, &initproc); 767 if (error) 768 panic("cannot fork init: %d\n", error); 769 KASSERT(initproc->p_pid == 1, ("create_init: initproc->p_pid != 1")); 770 /* divorce init's credentials from the kernel's */ 771 newcred = crget(); 772 PROC_LOCK(initproc); 773 initproc->p_flag |= P_SYSTEM | P_INMEM; 774 oldcred = initproc->p_ucred; 775 crcopy(newcred, oldcred); 776 #ifdef MAC 777 mac_cred_create_init(newcred); 778 #endif 779 #ifdef AUDIT 780 audit_cred_proc1(newcred); 781 #endif 782 initproc->p_ucred = newcred; 783 PROC_UNLOCK(initproc); 784 crfree(oldcred); 785 cred_update_thread(FIRST_THREAD_IN_PROC(initproc)); 786 cpu_set_fork_handler(FIRST_THREAD_IN_PROC(initproc), start_init, NULL); 787 } 788 SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL); 789 790 /* 791 * Make it runnable now. 792 */ 793 static void 794 kick_init(const void *udata __unused) 795 { 796 struct thread *td; 797 798 td = FIRST_THREAD_IN_PROC(initproc); 799 thread_lock(td); 800 TD_SET_CAN_RUN(td); 801 sched_add(td, SRQ_BORING); 802 thread_unlock(td); 803 } 804 SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL); 805