1 /*- 2 * SPDX-License-Identifier: BSD-4-Clause 3 * 4 * Copyright (c) 1995 Terrence R. Lambert 5 * All rights reserved. 6 * 7 * Copyright (c) 1982, 1986, 1989, 1991, 1992, 1993 8 * The Regents of the University of California. All rights reserved. 9 * (c) UNIX System Laboratories, Inc. 10 * All or some portions of this file are derived from material licensed 11 * to the University of California by American Telephone and Telegraph 12 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 13 * the permission of UNIX System Laboratories, Inc. 14 * 15 * Redistribution and use in source and binary forms, with or without 16 * modification, are permitted provided that the following conditions 17 * are met: 18 * 1. Redistributions of source code must retain the above copyright 19 * notice, this list of conditions and the following disclaimer. 20 * 2. Redistributions in binary form must reproduce the above copyright 21 * notice, this list of conditions and the following disclaimer in the 22 * documentation and/or other materials provided with the distribution. 23 * 3. All advertising materials mentioning features or use of this software 24 * must display the following acknowledgement: 25 * This product includes software developed by the University of 26 * California, Berkeley and its contributors. 27 * 4. Neither the name of the University nor the names of its contributors 28 * may be used to endorse or promote products derived from this software 29 * without specific prior written permission. 30 * 31 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 32 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 33 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 34 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 35 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 36 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 37 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 38 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 39 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 40 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 41 * SUCH DAMAGE. 42 */ 43 44 #include "opt_ddb.h" 45 #include "opt_kdb.h" 46 #include "opt_init_path.h" 47 #include "opt_verbose_sysinit.h" 48 49 #include <sys/param.h> 50 #include <sys/systm.h> 51 #include <sys/boottrace.h> 52 #include <sys/conf.h> 53 #include <sys/cpuset.h> 54 #include <sys/dtrace_bsd.h> 55 #include <sys/epoch.h> 56 #include <sys/eventhandler.h> 57 #include <sys/exec.h> 58 #include <sys/file.h> 59 #include <sys/filedesc.h> 60 #include <sys/imgact.h> 61 #include <sys/jail.h> 62 #include <sys/kernel.h> 63 #include <sys/ktr.h> 64 #include <sys/lock.h> 65 #include <sys/loginclass.h> 66 #include <sys/malloc.h> 67 #include <sys/mount.h> 68 #include <sys/mutex.h> 69 #include <sys/proc.h> 70 #include <sys/racct.h> 71 #include <sys/reboot.h> 72 #include <sys/resourcevar.h> 73 #include <sys/queue.h> 74 #include <sys/queue_mergesort.h> 75 #include <sys/sched.h> 76 #include <sys/signalvar.h> 77 #include <sys/sx.h> 78 #include <sys/syscallsubr.h> 79 #include <sys/sysctl.h> 80 #include <sys/sysent.h> 81 #include <sys/sysproto.h> 82 #include <sys/unistd.h> 83 #include <sys/vmmeter.h> 84 #include <sys/vnode.h> 85 86 #include <machine/cpu.h> 87 88 #include <security/audit/audit.h> 89 #include <security/mac/mac_framework.h> 90 91 #include <vm/vm.h> 92 #include <vm/vm_param.h> 93 #include <vm/vm_extern.h> 94 #include <vm/pmap.h> 95 #include <vm/vm_map.h> 96 #include <sys/copyright.h> 97 98 #include <ddb/ddb.h> 99 #include <ddb/db_sym.h> 100 101 void mi_startup(void); /* Should be elsewhere */ 102 103 /* Components of the first process -- never freed. */ 104 static struct session session0; 105 static struct pgrp pgrp0; 106 struct proc proc0; 107 struct thread0_storage thread0_st __aligned(32); 108 struct vmspace vmspace0; 109 struct proc *initproc; 110 111 int 112 linux_alloc_current_noop(struct thread *td __unused, int flags __unused) 113 { 114 return (0); 115 } 116 int (*lkpi_alloc_current)(struct thread *, int) = linux_alloc_current_noop; 117 118 #ifndef BOOTHOWTO 119 #define BOOTHOWTO 0 120 #endif 121 int boothowto = BOOTHOWTO; /* initialized so that it can be patched */ 122 SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, 123 "Boot control flags, passed from loader"); 124 125 #ifndef BOOTVERBOSE 126 #define BOOTVERBOSE 0 127 #endif 128 int bootverbose = BOOTVERBOSE; 129 SYSCTL_INT(_debug, OID_AUTO, bootverbose, CTLFLAG_RW, &bootverbose, 0, 130 "Control the output of verbose kernel messages"); 131 132 #ifdef VERBOSE_SYSINIT 133 /* 134 * We'll use the defined value of VERBOSE_SYSINIT from the kernel config to 135 * dictate the default VERBOSE_SYSINIT behavior. Significant values for this 136 * option and associated tunable are: 137 * - 0, 'compiled in but silent by default' 138 * - 1, 'compiled in but verbose by default' (default) 139 */ 140 int verbose_sysinit = VERBOSE_SYSINIT; 141 #endif 142 143 #ifdef INVARIANTS 144 FEATURE(invariants, "Kernel compiled with INVARIANTS, may affect performance"); 145 #endif 146 147 /* 148 * The sysinit linker set compiled into the kernel. These are placed onto the 149 * sysinit list by mi_startup; sysinit_add can add (e.g., from klds) additional 150 * sysinits to the linked list but the linker set here does not change. 151 */ 152 SET_DECLARE(sysinit_set, struct sysinit); 153 154 /* 155 * The sysinit lists. Items are moved to sysinit_done_list when done. 156 */ 157 static STAILQ_HEAD(sysinitlist, sysinit) sysinit_list; 158 static struct sysinitlist sysinit_done_list = 159 STAILQ_HEAD_INITIALIZER(sysinit_done_list); 160 161 /* 162 * Compare two sysinits; return -1, 0, or 1 if a comes before, at the same time 163 * as, or after b. 164 */ 165 static int 166 sysinit_compar(struct sysinit *a, struct sysinit *b, void *thunk __unused) 167 { 168 169 if (a->subsystem < b->subsystem) 170 return (-1); 171 if (a->subsystem > b->subsystem) 172 return (1); 173 if (a->order < b->order) 174 return (-1); 175 if (a->order > b->order) 176 return (1); 177 return (0); 178 } 179 180 static void 181 sysinit_mklist(struct sysinitlist *list, struct sysinit **set, 182 struct sysinit **set_end) 183 { 184 struct sysinit **sipp; 185 186 TSENTER(); 187 TSENTER2("listify"); 188 STAILQ_INIT(list); 189 for (sipp = set; sipp < set_end; sipp++) 190 STAILQ_INSERT_TAIL(list, *sipp, next); 191 TSEXIT2("listify"); 192 TSENTER2("mergesort"); 193 STAILQ_MERGESORT(list, NULL, sysinit_compar, sysinit, next); 194 TSEXIT2("mergesort"); 195 TSEXIT(); 196 } 197 198 /* 199 * Merge a new sysinit set into the sysinit list. 200 */ 201 void 202 sysinit_add(struct sysinit **set, struct sysinit **set_end) 203 { 204 struct sysinitlist new_list; 205 206 TSENTER(); 207 208 /* Construct a sorted list from the new sysinits. */ 209 sysinit_mklist(&new_list, set, set_end); 210 211 /* Merge the new list into the existing one. */ 212 TSENTER2("STAILQ_MERGE"); 213 STAILQ_MERGE(&sysinit_list, &new_list, NULL, sysinit_compar, sysinit, next); 214 TSEXIT2("STAILQ_MERGE"); 215 216 TSEXIT(); 217 } 218 219 #if defined (DDB) && defined(VERBOSE_SYSINIT) 220 static const char * 221 symbol_name(vm_offset_t va, db_strategy_t strategy) 222 { 223 const char *name; 224 c_db_sym_t sym; 225 db_expr_t offset; 226 227 if (va == 0) 228 return (NULL); 229 sym = db_search_symbol(va, strategy, &offset); 230 if (offset != 0) 231 return (NULL); 232 db_symbol_values(sym, &name, NULL); 233 return (name); 234 } 235 #endif 236 237 /* 238 * System startup; initialize the world, create process 0, mount root 239 * filesystem, and fork to create init and pagedaemon. Most of the 240 * hard work is done in the lower-level initialization routines including 241 * startup(), which does memory initialization and autoconfiguration. 242 * 243 * This allows simple addition of new kernel subsystems that require 244 * boot time initialization. It also allows substitution of subsystem 245 * (for instance, a scheduler, kernel profiler, or VM system) by object 246 * module. Finally, it allows for optional "kernel threads". 247 */ 248 void 249 mi_startup(void) 250 { 251 struct sysinit *sip; 252 int last; 253 #if defined(VERBOSE_SYSINIT) 254 int verbose; 255 #endif 256 257 TSENTER(); 258 259 if (boothowto & RB_VERBOSE) 260 bootverbose++; 261 262 /* Construct and sort sysinit list. */ 263 sysinit_mklist(&sysinit_list, SET_BEGIN(sysinit_set), SET_LIMIT(sysinit_set)); 264 265 last = SI_SUB_DUMMY; 266 #if defined(VERBOSE_SYSINIT) 267 TUNABLE_INT_FETCH("debug.verbose_sysinit", &verbose_sysinit); 268 verbose = 0; 269 #if !defined(DDB) 270 printf("VERBOSE_SYSINIT: DDB not enabled, symbol lookups disabled.\n"); 271 #endif 272 #endif 273 274 /* 275 * Perform each system initialization task from the ordered list. Note 276 * that if sysinit_list is modified (e.g. by a KLD) we will nonetheless 277 * always perform the earlist-sorted sysinit at each step; using the 278 * STAILQ_FOREACH macro would result in items being skipped if inserted 279 * earlier than the "current item". 280 */ 281 while ((sip = STAILQ_FIRST(&sysinit_list)) != NULL) { 282 STAILQ_REMOVE_HEAD(&sysinit_list, next); 283 STAILQ_INSERT_TAIL(&sysinit_done_list, sip, next); 284 285 if (sip->subsystem == SI_SUB_DUMMY) 286 continue; /* skip dummy task(s)*/ 287 288 if (sip->subsystem > last) 289 BOOTTRACE_INIT("sysinit 0x%7x", sip->subsystem); 290 291 #if defined(VERBOSE_SYSINIT) 292 if (sip->subsystem != last && verbose_sysinit != 0) { 293 verbose = 1; 294 printf("subsystem %x\n", sip->subsystem); 295 } 296 if (verbose) { 297 #if defined(DDB) 298 const char *func, *data; 299 300 func = symbol_name((vm_offset_t)sip->func, 301 DB_STGY_PROC); 302 data = symbol_name((vm_offset_t)sip->udata, 303 DB_STGY_ANY); 304 if (func != NULL && data != NULL) 305 printf(" %s(&%s)... ", func, data); 306 else if (func != NULL) 307 printf(" %s(%p)... ", func, sip->udata); 308 else 309 #endif 310 printf(" %p(%p)... ", sip->func, 311 sip->udata); 312 } 313 #endif 314 315 /* Call function */ 316 (*(sip->func))(sip->udata); 317 318 #if defined(VERBOSE_SYSINIT) 319 if (verbose) 320 printf("done.\n"); 321 #endif 322 323 /* Check off the one we're just done */ 324 last = sip->subsystem; 325 } 326 327 TSEXIT(); /* Here so we don't overlap with start_init. */ 328 BOOTTRACE("mi_startup done"); 329 330 mtx_assert(&Giant, MA_OWNED | MA_NOTRECURSED); 331 mtx_unlock(&Giant); 332 333 /* 334 * We can't free our thread structure since it is statically allocated. 335 * Just sleep forever. This thread could be repurposed for something if 336 * the need arises. 337 */ 338 for (;;) 339 tsleep(__builtin_frame_address(0), PNOLOCK, "parked", 0); 340 } 341 342 static void 343 print_caddr_t(const void *data) 344 { 345 printf("%s", (const char *)data); 346 } 347 348 static void 349 print_version(const void *data __unused) 350 { 351 int len; 352 353 /* Strip a trailing newline from version. */ 354 len = strlen(version); 355 while (len > 0 && version[len - 1] == '\n') 356 len--; 357 printf("%.*s %s\n", len, version, machine); 358 printf("%s\n", compiler_version); 359 } 360 361 C_SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, 362 copyright); 363 C_SYSINIT(trademark, SI_SUB_COPYRIGHT, SI_ORDER_SECOND, print_caddr_t, 364 trademark); 365 C_SYSINIT(version, SI_SUB_COPYRIGHT, SI_ORDER_THIRD, print_version, NULL); 366 367 #ifdef WITNESS 368 static const char wit_warn[] = 369 "WARNING: WITNESS option enabled, expect reduced performance.\n"; 370 C_SYSINIT(witwarn, SI_SUB_COPYRIGHT, SI_ORDER_FOURTH, 371 print_caddr_t, wit_warn); 372 C_SYSINIT(witwarn2, SI_SUB_LAST, SI_ORDER_FOURTH, 373 print_caddr_t, wit_warn); 374 #endif 375 376 #ifdef DIAGNOSTIC 377 static const char diag_warn[] = 378 "WARNING: DIAGNOSTIC option enabled, expect reduced performance.\n"; 379 C_SYSINIT(diagwarn, SI_SUB_COPYRIGHT, SI_ORDER_FIFTH, 380 print_caddr_t, diag_warn); 381 C_SYSINIT(diagwarn2, SI_SUB_LAST, SI_ORDER_FIFTH, 382 print_caddr_t, diag_warn); 383 #endif 384 385 #if __SIZEOF_LONG__ == 4 386 static const char ilp32_warn[] = 387 "WARNING: 32-bit kernels are deprecated and may be removed in FreeBSD 15.0.\n"; 388 C_SYSINIT(ilp32warn, SI_SUB_COPYRIGHT, SI_ORDER_FIFTH, 389 print_caddr_t, ilp32_warn); 390 C_SYSINIT(ilp32warn2, SI_SUB_LAST, SI_ORDER_FIFTH, 391 print_caddr_t, ilp32_warn); 392 #endif 393 394 static int 395 null_fetch_syscall_args(struct thread *td __unused) 396 { 397 398 panic("null_fetch_syscall_args"); 399 } 400 401 static void 402 null_set_syscall_retval(struct thread *td __unused, int error __unused) 403 { 404 405 panic("null_set_syscall_retval"); 406 } 407 408 static void 409 null_set_fork_retval(struct thread *td __unused) 410 { 411 412 } 413 414 struct sysentvec null_sysvec = { 415 .sv_size = 0, 416 .sv_table = NULL, 417 .sv_fixup = NULL, 418 .sv_sendsig = NULL, 419 .sv_sigcode = NULL, 420 .sv_szsigcode = NULL, 421 .sv_name = "null", 422 .sv_coredump = NULL, 423 .sv_minsigstksz = 0, 424 .sv_minuser = VM_MIN_ADDRESS, 425 .sv_maxuser = VM_MAXUSER_ADDRESS, 426 .sv_usrstack = USRSTACK, 427 .sv_psstrings = PS_STRINGS, 428 .sv_psstringssz = sizeof(struct ps_strings), 429 .sv_stackprot = VM_PROT_ALL, 430 .sv_copyout_strings = NULL, 431 .sv_setregs = NULL, 432 .sv_fixlimit = NULL, 433 .sv_maxssiz = NULL, 434 .sv_flags = 0, 435 .sv_set_syscall_retval = null_set_syscall_retval, 436 .sv_fetch_syscall_args = null_fetch_syscall_args, 437 .sv_syscallnames = NULL, 438 .sv_schedtail = NULL, 439 .sv_thread_detach = NULL, 440 .sv_trap = NULL, 441 .sv_set_fork_retval = null_set_fork_retval, 442 .sv_regset_begin = NULL, 443 .sv_regset_end = NULL, 444 }; 445 446 /* 447 * The two following SYSINIT's are proc0 specific glue code. I am not 448 * convinced that they can not be safely combined, but their order of 449 * operation has been maintained as the same as the original init_main.c 450 * for right now. 451 */ 452 /* ARGSUSED*/ 453 static void 454 proc0_init(void *dummy __unused) 455 { 456 struct proc *p; 457 struct thread *td; 458 struct ucred *newcred; 459 struct uidinfo tmpuinfo; 460 struct loginclass tmplc = { 461 .lc_name = "", 462 }; 463 vm_paddr_t pageablemem; 464 int i; 465 466 GIANT_REQUIRED; 467 p = &proc0; 468 td = &thread0; 469 470 /* 471 * Initialize magic number and osrel. 472 */ 473 p->p_magic = P_MAGIC; 474 p->p_osrel = osreldate; 475 476 /* 477 * Initialize thread and process structures. 478 */ 479 procinit(); /* set up proc zone */ 480 threadinit(); /* set up UMA zones */ 481 482 /* 483 * Initialise scheduler resources. 484 * Add scheduler specific parts to proc, thread as needed. 485 */ 486 schedinit(); /* scheduler gets its house in order */ 487 488 /* 489 * Create process 0 (the swapper). 490 */ 491 LIST_INSERT_HEAD(&allproc, p, p_list); 492 LIST_INSERT_HEAD(PIDHASH(0), p, p_hash); 493 mtx_init(&pgrp0.pg_mtx, "process group", NULL, MTX_DEF | MTX_DUPOK); 494 sx_init(&pgrp0.pg_killsx, "killpg racer"); 495 p->p_pgrp = &pgrp0; 496 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash); 497 LIST_INIT(&pgrp0.pg_members); 498 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist); 499 500 pgrp0.pg_session = &session0; 501 mtx_init(&session0.s_mtx, "session", NULL, MTX_DEF); 502 refcount_init(&session0.s_count, 1); 503 session0.s_leader = p; 504 505 p->p_sysent = &null_sysvec; 506 p->p_flag = P_SYSTEM | P_INMEM | P_KPROC; 507 p->p_flag2 = 0; 508 p->p_state = PRS_NORMAL; 509 p->p_klist = knlist_alloc(&p->p_mtx); 510 STAILQ_INIT(&p->p_ktr); 511 p->p_nice = NZERO; 512 td->td_tid = THREAD0_TID; 513 tidhash_add(td); 514 TD_SET_STATE(td, TDS_RUNNING); 515 td->td_pri_class = PRI_TIMESHARE; 516 td->td_user_pri = PUSER; 517 td->td_base_user_pri = PUSER; 518 td->td_lend_user_pri = PRI_MAX; 519 td->td_priority = PVM; 520 td->td_base_pri = PVM; 521 td->td_oncpu = curcpu; 522 td->td_flags = TDF_INMEM; 523 td->td_pflags = TDP_KTHREAD; 524 td->td_cpuset = cpuset_thread0(); 525 td->td_domain.dr_policy = td->td_cpuset->cs_domain; 526 prison0_init(); 527 p->p_peers = 0; 528 p->p_leader = p; 529 p->p_reaper = p; 530 p->p_treeflag |= P_TREE_REAPER; 531 LIST_INIT(&p->p_reaplist); 532 533 strncpy(p->p_comm, "kernel", sizeof (p->p_comm)); 534 strncpy(td->td_name, "swapper", sizeof (td->td_name)); 535 536 callout_init_mtx(&p->p_itcallout, &p->p_mtx, 0); 537 callout_init_mtx(&p->p_limco, &p->p_mtx, 0); 538 callout_init(&td->td_slpcallout, 1); 539 TAILQ_INIT(&p->p_kqtim_stop); 540 541 /* Create credentials. */ 542 newcred = crget(); 543 newcred->cr_ngroups = 1; /* group 0 */ 544 /* A hack to prevent uifind from tripping over NULL pointers. */ 545 curthread->td_ucred = newcred; 546 tmpuinfo.ui_uid = 1; 547 newcred->cr_uidinfo = newcred->cr_ruidinfo = &tmpuinfo; 548 newcred->cr_uidinfo = uifind(0); 549 newcred->cr_ruidinfo = uifind(0); 550 newcred->cr_loginclass = &tmplc; 551 newcred->cr_loginclass = loginclass_find("default"); 552 /* End hack. creds get properly set later with thread_cow_get_proc */ 553 curthread->td_ucred = NULL; 554 newcred->cr_prison = &prison0; 555 newcred->cr_users++; /* avoid assertion failure */ 556 p->p_ucred = crcowget(newcred); 557 newcred->cr_users--; 558 crfree(newcred); 559 #ifdef AUDIT 560 audit_cred_kproc0(newcred); 561 #endif 562 #ifdef MAC 563 mac_cred_create_swapper(newcred); 564 #endif 565 /* Create sigacts. */ 566 p->p_sigacts = sigacts_alloc(); 567 568 /* Initialize signal state for process 0. */ 569 siginit(&proc0); 570 571 /* Create the file descriptor table. */ 572 p->p_pd = pdinit(NULL, false); 573 p->p_fd = fdinit(); 574 p->p_fdtol = NULL; 575 576 /* Create the limits structures. */ 577 p->p_limit = lim_alloc(); 578 for (i = 0; i < RLIM_NLIMITS; i++) 579 p->p_limit->pl_rlimit[i].rlim_cur = 580 p->p_limit->pl_rlimit[i].rlim_max = RLIM_INFINITY; 581 p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_cur = 582 p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles; 583 p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_cur = 584 p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc; 585 p->p_limit->pl_rlimit[RLIMIT_DATA].rlim_cur = dfldsiz; 586 p->p_limit->pl_rlimit[RLIMIT_DATA].rlim_max = maxdsiz; 587 p->p_limit->pl_rlimit[RLIMIT_STACK].rlim_cur = dflssiz; 588 p->p_limit->pl_rlimit[RLIMIT_STACK].rlim_max = maxssiz; 589 /* Cast to avoid overflow on i386/PAE. */ 590 pageablemem = ptoa((vm_paddr_t)vm_free_count()); 591 p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_cur = 592 p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_max = pageablemem; 593 p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = pageablemem / 3; 594 p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_max = pageablemem; 595 p->p_cpulimit = RLIM_INFINITY; 596 597 PROC_LOCK(p); 598 thread_cow_get_proc(td, p); 599 PROC_UNLOCK(p); 600 601 /* Initialize resource accounting structures. */ 602 racct_create(&p->p_racct); 603 604 p->p_stats = pstats_alloc(); 605 606 /* Allocate a prototype map so we have something to fork. */ 607 p->p_vmspace = &vmspace0; 608 refcount_init(&vmspace0.vm_refcnt, 1); 609 pmap_pinit0(vmspace_pmap(&vmspace0)); 610 611 /* 612 * proc0 is not expected to enter usermode, so there is no special 613 * handling for sv_minuser here, like is done for exec_new_vmspace(). 614 */ 615 vm_map_init(&vmspace0.vm_map, vmspace_pmap(&vmspace0), 616 p->p_sysent->sv_minuser, p->p_sysent->sv_maxuser); 617 618 /* 619 * Call the init and ctor for the new thread and proc. We wait 620 * to do this until all other structures are fairly sane. 621 */ 622 EVENTHANDLER_DIRECT_INVOKE(process_init, p); 623 EVENTHANDLER_DIRECT_INVOKE(thread_init, td); 624 #ifdef KDTRACE_HOOKS 625 kdtrace_proc_ctor(p); 626 kdtrace_thread_ctor(td); 627 #endif 628 EVENTHANDLER_DIRECT_INVOKE(process_ctor, p); 629 EVENTHANDLER_DIRECT_INVOKE(thread_ctor, td); 630 631 /* 632 * Charge root for one process. 633 */ 634 (void)chgproccnt(p->p_ucred->cr_ruidinfo, 1, 0); 635 PROC_LOCK(p); 636 racct_add_force(p, RACCT_NPROC, 1); 637 PROC_UNLOCK(p); 638 } 639 SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL); 640 641 /* ARGSUSED*/ 642 static void 643 proc0_post(void *dummy __unused) 644 { 645 struct proc *p; 646 struct rusage ru; 647 struct thread *td; 648 649 /* 650 * Now we can look at the time, having had a chance to verify the 651 * time from the filesystem. Pretend that proc0 started now. 652 */ 653 sx_slock(&allproc_lock); 654 FOREACH_PROC_IN_SYSTEM(p) { 655 PROC_LOCK(p); 656 if (p->p_state == PRS_NEW) { 657 PROC_UNLOCK(p); 658 continue; 659 } 660 microuptime(&p->p_stats->p_start); 661 PROC_STATLOCK(p); 662 rufetch(p, &ru); /* Clears thread stats */ 663 p->p_rux.rux_runtime = 0; 664 p->p_rux.rux_uticks = 0; 665 p->p_rux.rux_sticks = 0; 666 p->p_rux.rux_iticks = 0; 667 PROC_STATUNLOCK(p); 668 FOREACH_THREAD_IN_PROC(p, td) { 669 td->td_runtime = 0; 670 } 671 PROC_UNLOCK(p); 672 } 673 sx_sunlock(&allproc_lock); 674 PCPU_SET(switchtime, cpu_ticks()); 675 PCPU_SET(switchticks, ticks); 676 } 677 SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL); 678 679 /* 680 *************************************************************************** 681 **** 682 **** The following SYSINIT's and glue code should be moved to the 683 **** respective files on a per subsystem basis. 684 **** 685 *************************************************************************** 686 */ 687 688 /* 689 * List of paths to try when searching for "init". 690 */ 691 static char init_path[MAXPATHLEN] = 692 #ifdef INIT_PATH 693 __XSTRING(INIT_PATH); 694 #else 695 "/sbin/init:/sbin/oinit:/sbin/init.bak:/rescue/init"; 696 #endif 697 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, 698 "Path used to search the init process"); 699 700 /* 701 * Shutdown timeout of init(8). 702 * Unused within kernel, but used to control init(8), hence do not remove. 703 */ 704 #ifndef INIT_SHUTDOWN_TIMEOUT 705 #define INIT_SHUTDOWN_TIMEOUT 120 706 #endif 707 static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT; 708 SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout, 709 CTLFLAG_RW, &init_shutdown_timeout, 0, "Shutdown timeout of init(8). " 710 "Unused within kernel, but used to control init(8)"); 711 712 /* 713 * Start the initial user process; try exec'ing each pathname in init_path. 714 * The program is invoked with one argument containing the boot flags. 715 */ 716 static void 717 start_init(void *dummy) 718 { 719 struct image_args args; 720 int error; 721 char *var, *path; 722 char *free_init_path, *tmp_init_path; 723 struct thread *td; 724 struct proc *p; 725 struct vmspace *oldvmspace; 726 727 TSENTER(); /* Here so we don't overlap with mi_startup. */ 728 729 td = curthread; 730 p = td->td_proc; 731 732 vfs_mountroot(); 733 734 /* Wipe GELI passphrase from the environment. */ 735 kern_unsetenv("kern.geom.eli.passphrase"); 736 737 /* For Multicons, report which console is primary to both */ 738 if (boothowto & RB_MULTIPLE) { 739 if (boothowto & RB_SERIAL) 740 printf("Dual Console: Serial Primary, Video Secondary\n"); 741 else 742 printf("Dual Console: Video Primary, Serial Secondary\n"); 743 } 744 745 if ((var = kern_getenv("init_path")) != NULL) { 746 strlcpy(init_path, var, sizeof(init_path)); 747 freeenv(var); 748 } 749 free_init_path = tmp_init_path = strdup(init_path, M_TEMP); 750 751 while ((path = strsep(&tmp_init_path, ":")) != NULL) { 752 if (bootverbose) 753 printf("start_init: trying %s\n", path); 754 755 memset(&args, 0, sizeof(args)); 756 error = exec_alloc_args(&args); 757 if (error != 0) 758 panic("%s: Can't allocate space for init arguments %d", 759 __func__, error); 760 761 error = exec_args_add_fname(&args, path, UIO_SYSSPACE); 762 if (error != 0) 763 panic("%s: Can't add fname %d", __func__, error); 764 error = exec_args_add_arg(&args, path, UIO_SYSSPACE); 765 if (error != 0) 766 panic("%s: Can't add argv[0] %d", __func__, error); 767 if (boothowto & RB_SINGLE) 768 error = exec_args_add_arg(&args, "-s", UIO_SYSSPACE); 769 if (error != 0) 770 panic("%s: Can't add argv[0] %d", __func__, error); 771 772 /* 773 * Now try to exec the program. If can't for any reason 774 * other than it doesn't exist, complain. 775 * 776 * Otherwise, return via fork_trampoline() all the way 777 * to user mode as init! 778 */ 779 KASSERT((td->td_pflags & TDP_EXECVMSPC) == 0, 780 ("nested execve")); 781 memset(td->td_frame, 0, sizeof(*td->td_frame)); 782 oldvmspace = p->p_vmspace; 783 error = kern_execve(td, &args, NULL, oldvmspace); 784 KASSERT(error != 0, 785 ("kern_execve returned success, not EJUSTRETURN")); 786 if (error == EJUSTRETURN) { 787 exec_cleanup(td, oldvmspace); 788 free(free_init_path, M_TEMP); 789 TSEXIT(); 790 return; 791 } 792 if (error != ENOENT) 793 printf("exec %s: error %d\n", path, error); 794 } 795 free(free_init_path, M_TEMP); 796 printf("init: not found in path %s\n", init_path); 797 panic("no init"); 798 } 799 800 /* 801 * Like kproc_create(), but runs in its own address space. We do this 802 * early to reserve pid 1. Note special case - do not make it 803 * runnable yet, init execution is started when userspace can be served. 804 */ 805 static void 806 create_init(const void *udata __unused) 807 { 808 struct fork_req fr; 809 struct ucred *newcred, *oldcred; 810 struct thread *td; 811 int error; 812 813 bzero(&fr, sizeof(fr)); 814 fr.fr_flags = RFFDG | RFPROC | RFSTOPPED; 815 fr.fr_procp = &initproc; 816 error = fork1(&thread0, &fr); 817 if (error) 818 panic("cannot fork init: %d\n", error); 819 KASSERT(initproc->p_pid == 1, ("create_init: initproc->p_pid != 1")); 820 /* divorce init's credentials from the kernel's */ 821 newcred = crget(); 822 sx_xlock(&proctree_lock); 823 PROC_LOCK(initproc); 824 initproc->p_flag |= P_SYSTEM | P_INMEM; 825 initproc->p_treeflag |= P_TREE_REAPER; 826 oldcred = initproc->p_ucred; 827 crcopy(newcred, oldcred); 828 #ifdef MAC 829 mac_cred_create_init(newcred); 830 #endif 831 #ifdef AUDIT 832 audit_cred_proc1(newcred); 833 #endif 834 proc_set_cred(initproc, newcred); 835 td = FIRST_THREAD_IN_PROC(initproc); 836 crcowfree(td); 837 td->td_realucred = crcowget(initproc->p_ucred); 838 td->td_ucred = td->td_realucred; 839 PROC_UNLOCK(initproc); 840 sx_xunlock(&proctree_lock); 841 crfree(oldcred); 842 cpu_fork_kthread_handler(FIRST_THREAD_IN_PROC(initproc), 843 start_init, NULL); 844 } 845 SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL); 846 847 /* 848 * Make it runnable now. 849 */ 850 static void 851 kick_init(const void *udata __unused) 852 { 853 struct thread *td; 854 855 td = FIRST_THREAD_IN_PROC(initproc); 856 thread_lock(td); 857 TD_SET_CAN_RUN(td); 858 sched_add(td, SRQ_BORING); 859 } 860 SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_MIDDLE, kick_init, NULL); 861 862 /* 863 * DDB(4). 864 */ 865 #ifdef DDB 866 static void 867 db_show_print_syinit(struct sysinit *sip, bool ddb) 868 { 869 const char *sname, *funcname; 870 c_db_sym_t sym; 871 db_expr_t offset; 872 873 #define xprint(...) \ 874 if (ddb) \ 875 db_printf(__VA_ARGS__); \ 876 else \ 877 printf(__VA_ARGS__) 878 879 if (sip == NULL) { 880 xprint("%s: no sysinit * given\n", __func__); 881 return; 882 } 883 884 sym = db_search_symbol((vm_offset_t)sip, DB_STGY_ANY, &offset); 885 db_symbol_values(sym, &sname, NULL); 886 sym = db_search_symbol((vm_offset_t)sip->func, DB_STGY_PROC, &offset); 887 db_symbol_values(sym, &funcname, NULL); 888 xprint("%s(%p)\n", (sname != NULL) ? sname : "", sip); 889 xprint(" %#08x %#08x\n", sip->subsystem, sip->order); 890 xprint(" %p(%s)(%p)\n", 891 sip->func, (funcname != NULL) ? funcname : "", sip->udata); 892 #undef xprint 893 } 894 895 DB_SHOW_COMMAND_FLAGS(sysinit, db_show_sysinit, DB_CMD_MEMSAFE) 896 { 897 struct sysinit *sip; 898 899 db_printf("SYSINIT vs Name(Ptr)\n"); 900 db_printf(" Subsystem Order\n"); 901 db_printf(" Function(Name)(Arg)\n"); 902 STAILQ_FOREACH(sip, &sysinit_done_list, next) { 903 db_show_print_syinit(sip, true); 904 if (db_pager_quit) 905 return; 906 } 907 STAILQ_FOREACH(sip, &sysinit_list, next) { 908 db_show_print_syinit(sip, true); 909 if (db_pager_quit) 910 break; 911 } 912 } 913 #endif /* DDB */ 914