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 #include "opt_mac.h" 50 51 #include <sys/param.h> 52 #include <sys/kernel.h> 53 #include <sys/exec.h> 54 #include <sys/file.h> 55 #include <sys/filedesc.h> 56 #include <sys/ktr.h> 57 #include <sys/lock.h> 58 #include <sys/mac.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/resourcevar.h> 65 #include <sys/systm.h> 66 #include <sys/signalvar.h> 67 #include <sys/vnode.h> 68 #include <sys/sysent.h> 69 #include <sys/reboot.h> 70 #include <sys/sched.h> 71 #include <sys/sx.h> 72 #include <sys/sysproto.h> 73 #include <sys/vmmeter.h> 74 #include <sys/unistd.h> 75 #include <sys/malloc.h> 76 #include <sys/conf.h> 77 78 #include <machine/cpu.h> 79 80 #include <security/audit/audit.h> 81 82 #include <vm/vm.h> 83 #include <vm/vm_param.h> 84 #include <vm/pmap.h> 85 #include <vm/vm_map.h> 86 #include <sys/copyright.h> 87 88 #include <ddb/ddb.h> 89 #include <ddb/db_sym.h> 90 91 void mi_startup(void); /* Should be elsewhere */ 92 93 /* Components of the first process -- never freed. */ 94 static struct session session0; 95 static struct pgrp pgrp0; 96 struct proc proc0; 97 struct thread thread0 __aligned(8); 98 struct ksegrp ksegrp0; 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 __unused) 288 { 289 printf("%s", (char *)data); 290 } 291 SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright) 292 SYSINIT(version, SI_SUB_COPYRIGHT, SI_ORDER_SECOND, print_caddr_t, version) 293 294 #ifdef WITNESS 295 static char wit_warn[] = 296 "WARNING: WITNESS option enabled, expect reduced performance.\n"; 297 SYSINIT(witwarn, SI_SUB_COPYRIGHT, SI_ORDER_SECOND + 1, 298 print_caddr_t, wit_warn) 299 #endif 300 301 #ifdef DIAGNOSTIC 302 static char diag_warn[] = 303 "WARNING: DIAGNOSTIC option enabled, expect reduced performance.\n"; 304 SYSINIT(diagwarn, SI_SUB_COPYRIGHT, SI_ORDER_SECOND + 2, 305 print_caddr_t, diag_warn) 306 #endif 307 308 static void 309 set_boot_verbose(void *data __unused) 310 { 311 312 if (boothowto & RB_VERBOSE) 313 bootverbose++; 314 } 315 SYSINIT(boot_verbose, SI_SUB_TUNABLES, SI_ORDER_ANY, set_boot_verbose, NULL) 316 317 struct sysentvec null_sysvec = { 318 0, 319 NULL, 320 0, 321 0, 322 NULL, 323 0, 324 NULL, 325 NULL, 326 NULL, 327 NULL, 328 NULL, 329 NULL, 330 NULL, 331 "null", 332 NULL, 333 NULL, 334 0, 335 PAGE_SIZE, 336 VM_MIN_ADDRESS, 337 VM_MAXUSER_ADDRESS, 338 USRSTACK, 339 PS_STRINGS, 340 VM_PROT_ALL, 341 NULL, 342 NULL, 343 NULL 344 }; 345 346 /* 347 *************************************************************************** 348 **** 349 **** The two following SYSINIT's are proc0 specific glue code. I am not 350 **** convinced that they can not be safely combined, but their order of 351 **** operation has been maintained as the same as the original init_main.c 352 **** for right now. 353 **** 354 **** These probably belong in init_proc.c or kern_proc.c, since they 355 **** deal with proc0 (the fork template process). 356 **** 357 *************************************************************************** 358 */ 359 /* ARGSUSED*/ 360 static void 361 proc0_init(void *dummy __unused) 362 { 363 struct proc *p; 364 unsigned i; 365 struct thread *td; 366 struct ksegrp *kg; 367 368 GIANT_REQUIRED; 369 p = &proc0; 370 td = &thread0; 371 kg = &ksegrp0; 372 373 /* 374 * Initialize magic number. 375 */ 376 p->p_magic = P_MAGIC; 377 378 /* 379 * Initialize thread, process and ksegrp structures. 380 */ 381 procinit(); /* set up proc zone */ 382 threadinit(); /* set up thead, upcall and KSEGRP zones */ 383 384 /* 385 * Initialise scheduler resources. 386 * Add scheduler specific parts to proc, ksegrp, thread as needed. 387 */ 388 schedinit(); /* scheduler gets its house in order */ 389 /* 390 * Initialize sleep queue hash table 391 */ 392 sleepinit(); 393 394 /* 395 * additional VM structures 396 */ 397 vm_init2(); 398 399 /* 400 * Create process 0 (the swapper). 401 */ 402 LIST_INSERT_HEAD(&allproc, p, p_list); 403 LIST_INSERT_HEAD(PIDHASH(0), p, p_hash); 404 mtx_init(&pgrp0.pg_mtx, "process group", NULL, MTX_DEF | MTX_DUPOK); 405 p->p_pgrp = &pgrp0; 406 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash); 407 LIST_INIT(&pgrp0.pg_members); 408 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist); 409 410 pgrp0.pg_session = &session0; 411 mtx_init(&session0.s_mtx, "session", NULL, MTX_DEF); 412 session0.s_count = 1; 413 session0.s_leader = p; 414 415 p->p_sysent = &null_sysvec; 416 p->p_flag = P_SYSTEM; 417 p->p_sflag = PS_INMEM; 418 p->p_state = PRS_NORMAL; 419 knlist_init(&p->p_klist, &p->p_mtx, NULL, NULL, NULL); 420 STAILQ_INIT(&p->p_ktr); 421 p->p_nice = NZERO; 422 td->td_state = TDS_RUNNING; 423 kg->kg_pri_class = PRI_TIMESHARE; 424 kg->kg_user_pri = PUSER; 425 kg->kg_base_user_pri = PUSER; 426 td->td_priority = PVM; 427 td->td_base_pri = PUSER; 428 td->td_oncpu = 0; 429 p->p_peers = 0; 430 p->p_leader = p; 431 432 433 bcopy("swapper", p->p_comm, sizeof ("swapper")); 434 435 callout_init(&p->p_itcallout, CALLOUT_MPSAFE); 436 callout_init(&td->td_slpcallout, CALLOUT_MPSAFE); 437 438 /* Create credentials. */ 439 p->p_ucred = crget(); 440 p->p_ucred->cr_ngroups = 1; /* group 0 */ 441 p->p_ucred->cr_uidinfo = uifind(0); 442 p->p_ucred->cr_ruidinfo = uifind(0); 443 p->p_ucred->cr_prison = NULL; /* Don't jail it. */ 444 #ifdef AUDIT 445 audit_proc_alloc(p); 446 audit_proc_kproc0(p); 447 #endif 448 #ifdef MAC 449 mac_create_proc0(p->p_ucred); 450 #endif 451 td->td_ucred = crhold(p->p_ucred); 452 453 /* Create sigacts. */ 454 p->p_sigacts = sigacts_alloc(); 455 456 /* Initialize signal state for process 0. */ 457 siginit(&proc0); 458 459 /* Create the file descriptor table. */ 460 p->p_fd = fdinit(NULL); 461 p->p_fdtol = NULL; 462 463 /* Create the limits structures. */ 464 p->p_limit = lim_alloc(); 465 for (i = 0; i < RLIM_NLIMITS; i++) 466 p->p_limit->pl_rlimit[i].rlim_cur = 467 p->p_limit->pl_rlimit[i].rlim_max = RLIM_INFINITY; 468 p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_cur = 469 p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles; 470 p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_cur = 471 p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc; 472 i = ptoa(cnt.v_free_count); 473 p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_max = i; 474 p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i; 475 p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3; 476 p->p_cpulimit = RLIM_INFINITY; 477 478 p->p_stats = pstats_alloc(); 479 480 /* Allocate a prototype map so we have something to fork. */ 481 pmap_pinit0(vmspace_pmap(&vmspace0)); 482 p->p_vmspace = &vmspace0; 483 vmspace0.vm_refcnt = 1; 484 vm_map_init(&vmspace0.vm_map, p->p_sysent->sv_minuser, 485 p->p_sysent->sv_maxuser); 486 vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0); 487 488 /* 489 * Charge root for one process. 490 */ 491 (void)chgproccnt(p->p_ucred->cr_ruidinfo, 1, 0); 492 } 493 SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL) 494 495 /* ARGSUSED*/ 496 static void 497 proc0_post(void *dummy __unused) 498 { 499 struct timespec ts; 500 struct proc *p; 501 502 /* 503 * Now we can look at the time, having had a chance to verify the 504 * time from the filesystem. Pretend that proc0 started now. 505 */ 506 sx_slock(&allproc_lock); 507 LIST_FOREACH(p, &allproc, p_list) { 508 microuptime(&p->p_stats->p_start); 509 p->p_rux.rux_runtime = 0; 510 } 511 sx_sunlock(&allproc_lock); 512 PCPU_SET(switchtime, cpu_ticks()); 513 PCPU_SET(switchticks, ticks); 514 515 /* 516 * Give the ``random'' number generator a thump. 517 */ 518 nanotime(&ts); 519 srandom(ts.tv_sec ^ ts.tv_nsec); 520 } 521 SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL) 522 523 /* 524 *************************************************************************** 525 **** 526 **** The following SYSINIT's and glue code should be moved to the 527 **** respective files on a per subsystem basis. 528 **** 529 *************************************************************************** 530 */ 531 532 533 /* 534 *************************************************************************** 535 **** 536 **** The following code probably belongs in another file, like 537 **** kern/init_init.c. 538 **** 539 *************************************************************************** 540 */ 541 542 /* 543 * List of paths to try when searching for "init". 544 */ 545 static char init_path[MAXPATHLEN] = 546 #ifdef INIT_PATH 547 __XSTRING(INIT_PATH); 548 #else 549 "/sbin/init:/sbin/oinit:/sbin/init.bak:/rescue/init:/stand/sysinstall"; 550 #endif 551 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, 552 "Path used to search the init process"); 553 554 /* 555 * Shutdown timeout of init(8). 556 * Unused within kernel, but used to control init(8), hence do not remove. 557 */ 558 #ifndef INIT_SHUTDOWN_TIMEOUT 559 #define INIT_SHUTDOWN_TIMEOUT 120 560 #endif 561 static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT; 562 SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout, 563 CTLFLAG_RW, &init_shutdown_timeout, 0, ""); 564 565 /* 566 * Start the initial user process; try exec'ing each pathname in init_path. 567 * The program is invoked with one argument containing the boot flags. 568 */ 569 static void 570 start_init(void *dummy) 571 { 572 vm_offset_t addr; 573 struct execve_args args; 574 int options, error; 575 char *var, *path, *next, *s; 576 char *ucp, **uap, *arg0, *arg1; 577 struct thread *td; 578 struct proc *p; 579 580 mtx_lock(&Giant); 581 582 GIANT_REQUIRED; 583 584 td = curthread; 585 p = td->td_proc; 586 587 vfs_mountroot(); 588 589 /* 590 * Need just enough stack to hold the faked-up "execve()" arguments. 591 */ 592 addr = p->p_sysent->sv_usrstack - PAGE_SIZE; 593 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE, 594 FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0) 595 panic("init: couldn't allocate argument space"); 596 p->p_vmspace->vm_maxsaddr = (caddr_t)addr; 597 p->p_vmspace->vm_ssize = 1; 598 599 if ((var = getenv("init_path")) != NULL) { 600 strlcpy(init_path, var, sizeof(init_path)); 601 freeenv(var); 602 } 603 604 for (path = init_path; *path != '\0'; path = next) { 605 while (*path == ':') 606 path++; 607 if (*path == '\0') 608 break; 609 for (next = path; *next != '\0' && *next != ':'; next++) 610 /* nothing */ ; 611 if (bootverbose) 612 printf("start_init: trying %.*s\n", (int)(next - path), 613 path); 614 615 /* 616 * Move out the boot flag argument. 617 */ 618 options = 0; 619 ucp = (char *)p->p_sysent->sv_usrstack; 620 (void)subyte(--ucp, 0); /* trailing zero */ 621 if (boothowto & RB_SINGLE) { 622 (void)subyte(--ucp, 's'); 623 options = 1; 624 } 625 #ifdef notyet 626 if (boothowto & RB_FASTBOOT) { 627 (void)subyte(--ucp, 'f'); 628 options = 1; 629 } 630 #endif 631 632 #ifdef BOOTCDROM 633 (void)subyte(--ucp, 'C'); 634 options = 1; 635 #endif 636 637 if (options == 0) 638 (void)subyte(--ucp, '-'); 639 (void)subyte(--ucp, '-'); /* leading hyphen */ 640 arg1 = ucp; 641 642 /* 643 * Move out the file name (also arg 0). 644 */ 645 (void)subyte(--ucp, 0); 646 for (s = next - 1; s >= path; s--) 647 (void)subyte(--ucp, *s); 648 arg0 = ucp; 649 650 /* 651 * Move out the arg pointers. 652 */ 653 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1)); 654 (void)suword((caddr_t)--uap, (long)0); /* terminator */ 655 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1); 656 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0); 657 658 /* 659 * Point at the arguments. 660 */ 661 args.fname = arg0; 662 args.argv = uap; 663 args.envv = NULL; 664 665 /* 666 * Now try to exec the program. If can't for any reason 667 * other than it doesn't exist, complain. 668 * 669 * Otherwise, return via fork_trampoline() all the way 670 * to user mode as init! 671 */ 672 if ((error = execve(td, &args)) == 0) { 673 mtx_unlock(&Giant); 674 return; 675 } 676 if (error != ENOENT) 677 printf("exec %.*s: error %d\n", (int)(next - path), 678 path, error); 679 } 680 printf("init: not found in path %s\n", init_path); 681 panic("no init"); 682 } 683 684 /* 685 * Like kthread_create(), but runs in it's own address space. 686 * We do this early to reserve pid 1. 687 * 688 * Note special case - do not make it runnable yet. Other work 689 * in progress will change this more. 690 */ 691 static void 692 create_init(const void *udata __unused) 693 { 694 struct ucred *newcred, *oldcred; 695 int error; 696 697 error = fork1(&thread0, RFFDG | RFPROC | RFSTOPPED, 0, &initproc); 698 if (error) 699 panic("cannot fork init: %d\n", error); 700 KASSERT(initproc->p_pid == 1, ("create_init: initproc->p_pid != 1")); 701 /* divorce init's credentials from the kernel's */ 702 newcred = crget(); 703 PROC_LOCK(initproc); 704 initproc->p_flag |= P_SYSTEM; 705 oldcred = initproc->p_ucred; 706 crcopy(newcred, oldcred); 707 #ifdef MAC 708 mac_create_proc1(newcred); 709 #endif 710 #ifdef AUDIT 711 audit_proc_init(initproc); 712 #endif 713 initproc->p_ucred = newcred; 714 PROC_UNLOCK(initproc); 715 crfree(oldcred); 716 cred_update_thread(FIRST_THREAD_IN_PROC(initproc)); 717 mtx_lock_spin(&sched_lock); 718 initproc->p_sflag |= PS_INMEM; 719 mtx_unlock_spin(&sched_lock); 720 cpu_set_fork_handler(FIRST_THREAD_IN_PROC(initproc), start_init, NULL); 721 } 722 SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL) 723 724 /* 725 * Make it runnable now. 726 */ 727 static void 728 kick_init(const void *udata __unused) 729 { 730 struct thread *td; 731 732 td = FIRST_THREAD_IN_PROC(initproc); 733 mtx_lock_spin(&sched_lock); 734 TD_SET_CAN_RUN(td); 735 setrunqueue(td, SRQ_BORING); /* XXXKSE */ 736 mtx_unlock_spin(&sched_lock); 737 } 738 SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL) 739