1 /* $OpenBSD: init_main.c,v 1.150 2008/06/09 07:07:16 djm Exp $ */ 2 /* $NetBSD: init_main.c,v 1.84.4.1 1996/06/02 09:08:06 mrg Exp $ */ 3 4 /* 5 * Copyright (c) 1995 Christopher G. Demetriou. All rights reserved. 6 * Copyright (c) 1982, 1986, 1989, 1991, 1992, 1993 7 * The Regents of the University of California. All rights reserved. 8 * (c) UNIX System Laboratories, Inc. 9 * All or some portions of this file are derived from material licensed 10 * to the University of California by American Telephone and Telegraph 11 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 12 * the permission of UNIX System Laboratories, Inc. 13 * 14 * Redistribution and use in source and binary forms, with or without 15 * modification, are permitted provided that the following conditions 16 * are met: 17 * 1. Redistributions of source code must retain the above copyright 18 * notice, this list of conditions and the following disclaimer. 19 * 2. Redistributions in binary form must reproduce the above copyright 20 * notice, this list of conditions and the following disclaimer in the 21 * documentation and/or other materials provided with the distribution. 22 * 3. Neither the name of the University nor the names of its contributors 23 * may be used to endorse or promote products derived from this software 24 * without specific prior written permission. 25 * 26 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 27 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 28 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 29 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 30 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 31 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 32 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 34 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 35 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 36 * SUCH DAMAGE. 37 * 38 * @(#)init_main.c 8.9 (Berkeley) 1/21/94 39 */ 40 41 #include <sys/param.h> 42 #include <sys/filedesc.h> 43 #include <sys/file.h> 44 #include <sys/errno.h> 45 #include <sys/exec.h> 46 #include <sys/kernel.h> 47 #include <sys/kthread.h> 48 #include <sys/mount.h> 49 #include <sys/proc.h> 50 #include <sys/resourcevar.h> 51 #include <sys/signalvar.h> 52 #include <sys/systm.h> 53 #include <sys/namei.h> 54 #include <sys/vnode.h> 55 #include <sys/tty.h> 56 #include <sys/conf.h> 57 #include <sys/buf.h> 58 #include <sys/device.h> 59 #include <sys/socketvar.h> 60 #include <sys/lockf.h> 61 #include <sys/protosw.h> 62 #include <sys/reboot.h> 63 #include <sys/user.h> 64 #ifdef SYSVSHM 65 #include <sys/shm.h> 66 #endif 67 #ifdef SYSVSEM 68 #include <sys/sem.h> 69 #endif 70 #ifdef SYSVMSG 71 #include <sys/msg.h> 72 #endif 73 #include <sys/domain.h> 74 #include <sys/mbuf.h> 75 #include <sys/pipe.h> 76 #include <sys/workq.h> 77 78 #include <sys/syscall.h> 79 #include <sys/syscallargs.h> 80 81 #include <dev/rndvar.h> 82 83 #include <ufs/ufs/quota.h> 84 85 #include <machine/cpu.h> 86 87 #include <uvm/uvm.h> 88 89 #include <net/if.h> 90 #include <net/raw_cb.h> 91 92 #if defined(CRYPTO) 93 #include <crypto/cryptodev.h> 94 #include <crypto/cryptosoft.h> 95 #endif 96 97 #if defined(NFSSERVER) || defined(NFSCLIENT) 98 extern void nfs_init(void); 99 #endif 100 101 #include "softraid.h" 102 103 const char copyright[] = 104 "Copyright (c) 1982, 1986, 1989, 1991, 1993\n" 105 "\tThe Regents of the University of California. All rights reserved.\n" 106 "Copyright (c) 1995-2008 OpenBSD. All rights reserved. http://www.OpenBSD.org\n"; 107 108 /* Components of the first process -- never freed. */ 109 struct session session0; 110 struct pgrp pgrp0; 111 struct proc proc0; 112 struct process process0; 113 struct pcred cred0; 114 struct plimit limit0; 115 struct vmspace vmspace0; 116 struct sigacts sigacts0; 117 struct proc *initproc; 118 119 int cmask = CMASK; 120 extern struct user *proc0paddr; 121 122 struct vnode *rootvp, *swapdev_vp; 123 int boothowto; 124 struct timeval boottime; 125 int ncpus = 1; 126 __volatile int start_init_exec; /* semaphore for start_init() */ 127 128 #if !defined(NO_PROPOLICE) 129 long __guard[8]; 130 #endif 131 132 /* XXX return int so gcc -Werror won't complain */ 133 int main(void *); 134 void check_console(struct proc *); 135 void start_init(void *); 136 void start_cleaner(void *); 137 void start_update(void *); 138 void start_reaper(void *); 139 void init_crypto(void); 140 void init_exec(void); 141 void kqueue_init(void); 142 void workq_init(void); 143 144 extern char sigcode[], esigcode[]; 145 #ifdef SYSCALL_DEBUG 146 extern char *syscallnames[]; 147 #endif 148 149 struct emul emul_native = { 150 "native", 151 NULL, 152 sendsig, 153 SYS_syscall, 154 SYS_MAXSYSCALL, 155 sysent, 156 #ifdef SYSCALL_DEBUG 157 syscallnames, 158 #else 159 NULL, 160 #endif 161 0, 162 copyargs, 163 setregs, 164 NULL, 165 sigcode, 166 esigcode, 167 EMUL_ENABLED | EMUL_NATIVE, 168 }; 169 170 171 /* 172 * System startup; initialize the world, create process 0, mount root 173 * filesystem, and fork to create init and pagedaemon. Most of the 174 * hard work is done in the lower-level initialization routines including 175 * startup(), which does memory initialization and autoconfiguration. 176 */ 177 /* XXX return int, so gcc -Werror won't complain */ 178 int 179 main(void *framep) 180 { 181 struct proc *p; 182 struct pdevinit *pdev; 183 struct timeval rtv; 184 quad_t lim; 185 int s, i; 186 extern struct pdevinit pdevinit[]; 187 extern void scheduler_start(void); 188 extern void disk_init(void); 189 extern void endtsleep(void *); 190 extern void realitexpire(void *); 191 192 /* 193 * Initialize the current process pointer (curproc) before 194 * any possible traps/probes to simplify trap processing. 195 */ 196 curproc = p = &proc0; 197 p->p_cpu = curcpu(); 198 199 /* 200 * Initialize timeouts. 201 */ 202 timeout_startup(); 203 204 /* 205 * Attempt to find console and initialize 206 * in case of early panic or other messages. 207 */ 208 config_init(); /* init autoconfiguration data structures */ 209 consinit(); 210 211 printf("%s\n", copyright); 212 213 KERNEL_LOCK_INIT(); 214 215 uvm_init(); 216 disk_init(); /* must come before autoconfiguration */ 217 tty_init(); /* initialise tty's */ 218 cpu_startup(); 219 220 /* 221 * Initialize mbuf's. Do this now because we might attempt to 222 * allocate mbufs or mbuf clusters during autoconfiguration. 223 */ 224 mbinit(); 225 226 /* Initialize sockets. */ 227 soinit(); 228 229 /* 230 * Initialize process and pgrp structures. 231 */ 232 procinit(); 233 234 /* Initialize file locking. */ 235 lf_init(); 236 237 /* 238 * Initialize filedescriptors. 239 */ 240 filedesc_init(); 241 242 /* 243 * Initialize pipes. 244 */ 245 pipe_init(); 246 247 /* 248 * Initialize kqueues. 249 */ 250 kqueue_init(); 251 252 /* 253 * Create process 0 (the swapper). 254 */ 255 256 process0.ps_mainproc = p; 257 TAILQ_INIT(&process0.ps_threads); 258 TAILQ_INSERT_TAIL(&process0.ps_threads, p, p_thr_link); 259 p->p_p = &process0; 260 261 LIST_INSERT_HEAD(&allproc, p, p_list); 262 p->p_pgrp = &pgrp0; 263 LIST_INSERT_HEAD(PIDHASH(0), p, p_hash); 264 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash); 265 LIST_INIT(&pgrp0.pg_members); 266 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist); 267 268 pgrp0.pg_session = &session0; 269 session0.s_count = 1; 270 session0.s_leader = p; 271 272 atomic_setbits_int(&p->p_flag, P_SYSTEM | P_NOCLDWAIT); 273 p->p_stat = SONPROC; 274 p->p_nice = NZERO; 275 p->p_emul = &emul_native; 276 bcopy("swapper", p->p_comm, sizeof ("swapper")); 277 278 /* Init timeouts. */ 279 timeout_set(&p->p_sleep_to, endtsleep, p); 280 timeout_set(&p->p_realit_to, realitexpire, p); 281 282 /* Create credentials. */ 283 cred0.p_refcnt = 1; 284 p->p_cred = &cred0; 285 p->p_ucred = crget(); 286 p->p_ucred->cr_ngroups = 1; /* group 0 */ 287 288 /* Initialize signal state for process 0. */ 289 signal_init(); 290 p->p_sigacts = &sigacts0; 291 siginit(p); 292 293 /* Create the file descriptor table. */ 294 p->p_fd = fdinit(NULL); 295 296 /* Create the limits structures. */ 297 p->p_p->ps_limit = &limit0; 298 for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++) 299 limit0.pl_rlimit[i].rlim_cur = 300 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY; 301 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur = NOFILE; 302 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = MIN(NOFILE_MAX, 303 (maxfiles - NOFILE > NOFILE) ? maxfiles - NOFILE : NOFILE); 304 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur = MAXUPRC; 305 lim = ptoa(uvmexp.free); 306 limit0.pl_rlimit[RLIMIT_RSS].rlim_max = lim; 307 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = lim; 308 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = lim / 3; 309 limit0.p_refcnt = 1; 310 311 /* Allocate a prototype map so we have something to fork. */ 312 uvmspace_init(&vmspace0, pmap_kernel(), round_page(VM_MIN_ADDRESS), 313 trunc_page(VM_MAX_ADDRESS), TRUE, TRUE); 314 p->p_vmspace = &vmspace0; 315 316 p->p_addr = proc0paddr; /* XXX */ 317 318 /* 319 * We continue to place resource usage info in the 320 * user struct so they're pageable. 321 */ 322 p->p_stats = &p->p_addr->u_stats; 323 324 /* 325 * Charge root for one process. 326 */ 327 (void)chgproccnt(0, 1); 328 329 /* Initialize run queues */ 330 sched_init_runqueues(); 331 sleep_queue_init(); 332 sched_init_cpu(curcpu()); 333 334 /* Initialize work queues */ 335 workq_init(); 336 337 /* Configure the devices */ 338 cpu_configure(); 339 340 /* Configure virtual memory system, set vm rlimits. */ 341 uvm_init_limits(p); 342 343 /* Initialize the file systems. */ 344 #if defined(NFSSERVER) || defined(NFSCLIENT) 345 nfs_init(); /* initialize server/shared data */ 346 #endif 347 vfsinit(); 348 349 /* Start real time and statistics clocks. */ 350 initclocks(); 351 352 /* Lock the kernel on behalf of proc0. */ 353 KERNEL_PROC_LOCK(p); 354 355 #ifdef SYSVSHM 356 /* Initialize System V style shared memory. */ 357 shminit(); 358 #endif 359 360 #ifdef SYSVSEM 361 /* Initialize System V style semaphores. */ 362 seminit(); 363 #endif 364 365 #ifdef SYSVMSG 366 /* Initialize System V style message queues. */ 367 msginit(); 368 #endif 369 370 /* Attach pseudo-devices. */ 371 randomattach(); 372 for (pdev = pdevinit; pdev->pdev_attach != NULL; pdev++) 373 if (pdev->pdev_count > 0) 374 (*pdev->pdev_attach)(pdev->pdev_count); 375 376 #ifdef CRYPTO 377 swcr_init(); 378 #endif /* CRYPTO */ 379 380 /* 381 * Initialize protocols. Block reception of incoming packets 382 * until everything is ready. 383 */ 384 s = splnet(); 385 ifinit(); 386 domaininit(); 387 if_attachdomain(); 388 splx(s); 389 390 #ifdef GPROF 391 /* Initialize kernel profiling. */ 392 kmstartup(); 393 #endif 394 395 #if !defined(NO_PROPOLICE) 396 { 397 volatile long newguard[8]; 398 399 arc4random_buf((long *)newguard, sizeof(newguard)); 400 401 for (i = sizeof(__guard)/sizeof(__guard[0]) - 1; i; i--) 402 __guard[i] = newguard[i]; 403 } 404 #endif 405 406 /* init exec and emul */ 407 init_exec(); 408 409 /* Start the scheduler */ 410 scheduler_start(); 411 412 /* 413 * Create process 1 (init(8)). We do this now, as Unix has 414 * historically had init be process 1, and changing this would 415 * probably upset a lot of people. 416 * 417 * Note that process 1 won't immediately exec init(8), but will 418 * wait for us to inform it that the root file system has been 419 * mounted. 420 */ 421 if (fork1(p, SIGCHLD, FORK_FORK, NULL, 0, start_init, NULL, NULL, 422 &initproc)) 423 panic("fork init"); 424 425 /* 426 * Create any kernel threads whose creation was deferred because 427 * initproc had not yet been created. 428 */ 429 kthread_run_deferred_queue(); 430 431 /* 432 * Now that device driver threads have been created, wait for 433 * them to finish any deferred autoconfiguration. Note we don't 434 * need to lock this semaphore, since we haven't booted any 435 * secondary processors, yet. 436 */ 437 while (config_pending) 438 (void) tsleep((void *)&config_pending, PWAIT, "cfpend", 0); 439 440 dostartuphooks(); 441 442 #if NSOFTRAID > 0 443 config_rootfound("softraid", NULL); 444 #endif 445 446 /* Configure root/swap devices */ 447 diskconf(); 448 449 if (mountroot == NULL || ((*mountroot)() != 0)) 450 panic("cannot mount root"); 451 452 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 453 454 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */ 455 if (VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode)) 456 panic("cannot find root vnode"); 457 p->p_fd->fd_cdir = rootvnode; 458 VREF(p->p_fd->fd_cdir); 459 VOP_UNLOCK(rootvnode, 0, p); 460 p->p_fd->fd_rdir = NULL; 461 462 /* 463 * Now that root is mounted, we can fixup initproc's CWD 464 * info. All other processes are kthreads, which merely 465 * share proc0's CWD info. 466 */ 467 initproc->p_fd->fd_cdir = rootvnode; 468 VREF(initproc->p_fd->fd_cdir); 469 initproc->p_fd->fd_rdir = NULL; 470 471 /* 472 * Now can look at time, having had a chance to verify the time 473 * from the file system. Reset p->p_rtime as it may have been 474 * munched in mi_switch() after the time got set. 475 */ 476 #ifdef __HAVE_TIMECOUNTER 477 microtime(&boottime); 478 #else 479 boottime = mono_time = time; 480 #endif 481 LIST_FOREACH(p, &allproc, p_list) { 482 p->p_stats->p_start = boottime; 483 microuptime(&p->p_cpu->ci_schedstate.spc_runtime); 484 p->p_rtime.tv_sec = p->p_rtime.tv_usec = 0; 485 } 486 487 uvm_swap_init(); 488 489 /* Create the pageout daemon kernel thread. */ 490 if (kthread_create(uvm_pageout, NULL, NULL, "pagedaemon")) 491 panic("fork pagedaemon"); 492 493 /* Create the reaper daemon kernel thread. */ 494 if (kthread_create(start_reaper, NULL, NULL, "reaper")) 495 panic("fork reaper"); 496 497 /* Create the cleaner daemon kernel thread. */ 498 if (kthread_create(start_cleaner, NULL, NULL, "cleaner")) 499 panic("fork cleaner"); 500 501 /* Create the update daemon kernel thread. */ 502 if (kthread_create(start_update, NULL, NULL, "update")) 503 panic("fork update"); 504 505 /* Create the aiodone daemon kernel thread. */ 506 if (kthread_create(uvm_aiodone_daemon, NULL, NULL, "aiodoned")) 507 panic("fork aiodoned"); 508 509 #ifdef CRYPTO 510 /* Create the crypto kernel thread. */ 511 init_crypto(); 512 #endif /* CRYPTO */ 513 514 microtime(&rtv); 515 srandom((u_long)(rtv.tv_sec ^ rtv.tv_usec)); 516 517 randompid = 1; 518 519 #if defined(MULTIPROCESSOR) 520 /* Boot the secondary processors. */ 521 cpu_boot_secondary_processors(); 522 #endif 523 524 domountroothooks(); 525 526 /* 527 * Okay, now we can let init(8) exec! It's off to userland! 528 */ 529 start_init_exec = 1; 530 wakeup((void *)&start_init_exec); 531 532 /* The scheduler is an infinite loop. */ 533 uvm_scheduler(); 534 /* NOTREACHED */ 535 } 536 537 /* 538 * List of paths to try when searching for "init". 539 */ 540 static char *initpaths[] = { 541 "/sbin/init", 542 "/sbin/oinit", 543 "/sbin/init.bak", 544 NULL, 545 }; 546 547 void 548 check_console(struct proc *p) 549 { 550 struct nameidata nd; 551 int error; 552 553 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console", p); 554 error = namei(&nd); 555 if (error) { 556 if (error == ENOENT) 557 printf("warning: /dev/console does not exist\n"); 558 else 559 printf("warning: /dev/console error %d\n", error); 560 } else 561 vrele(nd.ni_vp); 562 } 563 564 /* 565 * Start the initial user process; try exec'ing each pathname in "initpaths". 566 * The program is invoked with one argument containing the boot flags. 567 */ 568 void 569 start_init(void *arg) 570 { 571 struct proc *p = arg; 572 vaddr_t addr; 573 struct sys_execve_args /* { 574 syscallarg(const char *) path; 575 syscallarg(char *const *) argp; 576 syscallarg(char *const *) envp; 577 } */ args; 578 int options, error; 579 long i; 580 register_t retval[2]; 581 char flags[4], *flagsp; 582 char **pathp, *path, *ucp, **uap, *arg0, *arg1 = NULL; 583 584 /* 585 * Now in process 1. 586 */ 587 588 /* 589 * Wait for main() to tell us that it's safe to exec. 590 */ 591 while (start_init_exec == 0) 592 (void) tsleep((void *)&start_init_exec, PWAIT, "initexec", 0); 593 594 check_console(p); 595 596 /* 597 * Need just enough stack to hold the faked-up "execve()" arguments. 598 */ 599 #ifdef MACHINE_STACK_GROWS_UP 600 addr = USRSTACK; 601 #else 602 addr = USRSTACK - PAGE_SIZE; 603 #endif 604 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 605 NULL, UVM_UNKNOWN_OFFSET, 0, 606 UVM_MAPFLAG(UVM_PROT_RW, UVM_PROT_ALL, UVM_INH_COPY, 607 UVM_ADV_NORMAL, UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW))) 608 panic("init: couldn't allocate argument space"); 609 p->p_vmspace->vm_maxsaddr = (caddr_t)addr; 610 611 for (pathp = &initpaths[0]; (path = *pathp) != NULL; pathp++) { 612 #ifdef MACHINE_STACK_GROWS_UP 613 ucp = (char *)addr; 614 #else 615 ucp = (char *)(addr + PAGE_SIZE); 616 #endif 617 /* 618 * Construct the boot flag argument. 619 */ 620 flagsp = flags; 621 *flagsp++ = '-'; 622 options = 0; 623 624 if (boothowto & RB_SINGLE) { 625 *flagsp++ = 's'; 626 options = 1; 627 } 628 #ifdef notyet 629 if (boothowto & RB_FASTBOOT) { 630 *flagsp++ = 'f'; 631 options = 1; 632 } 633 #endif 634 635 /* 636 * Move out the flags (arg 1), if necessary. 637 */ 638 if (options != 0) { 639 *flagsp++ = '\0'; 640 i = flagsp - flags; 641 #ifdef DEBUG 642 printf("init: copying out flags `%s' %d\n", flags, i); 643 #endif 644 #ifdef MACHINE_STACK_GROWS_UP 645 arg1 = ucp; 646 (void)copyout((caddr_t)flags, (caddr_t)ucp, i); 647 ucp += i; 648 #else 649 (void)copyout((caddr_t)flags, (caddr_t)(ucp -= i), i); 650 arg1 = ucp; 651 #endif 652 } 653 654 /* 655 * Move out the file name (also arg 0). 656 */ 657 i = strlen(path) + 1; 658 #ifdef DEBUG 659 printf("init: copying out path `%s' %d\n", path, i); 660 #endif 661 #ifdef MACHINE_STACK_GROWS_UP 662 arg0 = ucp; 663 (void)copyout((caddr_t)path, (caddr_t)ucp, i); 664 ucp += i; 665 ucp = (caddr_t)ALIGN((u_long)ucp); 666 uap = (char **)ucp + 3; 667 #else 668 (void)copyout((caddr_t)path, (caddr_t)(ucp -= i), i); 669 arg0 = ucp; 670 uap = (char **)((u_long)ucp & ~ALIGNBYTES); 671 #endif 672 673 /* 674 * Move out the arg pointers. 675 */ 676 i = 0; 677 copyout(&i, (caddr_t)--uap, sizeof(register_t)); /* terminator */ 678 if (options != 0) 679 copyout(&arg1, (caddr_t)--uap, sizeof(register_t)); 680 copyout(&arg0, (caddr_t)--uap, sizeof(register_t)); 681 682 /* 683 * Point at the arguments. 684 */ 685 SCARG(&args, path) = arg0; 686 SCARG(&args, argp) = uap; 687 SCARG(&args, envp) = NULL; 688 689 /* 690 * Now try to exec the program. If can't for any reason 691 * other than it doesn't exist, complain. 692 */ 693 if ((error = sys_execve(p, &args, retval)) == 0) { 694 KERNEL_PROC_UNLOCK(p); 695 return; 696 } 697 if (error != ENOENT) 698 printf("exec %s: error %d\n", path, error); 699 } 700 printf("init: not found\n"); 701 panic("no init"); 702 } 703 704 void 705 start_update(void *arg) 706 { 707 sched_sync(curproc); 708 /* NOTREACHED */ 709 } 710 711 void 712 start_cleaner(void *arg) 713 { 714 buf_daemon(curproc); 715 /* NOTREACHED */ 716 } 717 718 void 719 start_reaper(void *arg) 720 { 721 reaper(); 722 /* NOTREACHED */ 723 } 724