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