init_main.c revision 1.267
1 /* $NetBSD: init_main.c,v 1.267 2006/06/07 22:33:39 kardel Exp $ */ 2 3 /* 4 * Copyright (c) 1982, 1986, 1989, 1991, 1992, 1993 5 * The Regents of the University of California. All rights reserved. 6 * (c) UNIX System Laboratories, Inc. 7 * All or some portions of this file are derived from material licensed 8 * to the University of California by American Telephone and Telegraph 9 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 10 * the permission of UNIX System Laboratories, Inc. 11 * 12 * Redistribution and use in source and binary forms, with or without 13 * modification, are permitted provided that the following conditions 14 * are met: 15 * 1. Redistributions of source code must retain the above copyright 16 * notice, this list of conditions and the following disclaimer. 17 * 2. Redistributions in binary form must reproduce the above copyright 18 * notice, this list of conditions and the following disclaimer in the 19 * documentation and/or other materials provided with the distribution. 20 * 3. Neither the name of the University nor the names of its contributors 21 * may be used to endorse or promote products derived from this software 22 * without specific prior written permission. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 34 * SUCH DAMAGE. 35 * 36 * @(#)init_main.c 8.16 (Berkeley) 5/14/95 37 */ 38 39 /* 40 * Copyright (c) 1995 Christopher G. Demetriou. All rights reserved. 41 * 42 * Redistribution and use in source and binary forms, with or without 43 * modification, are permitted provided that the following conditions 44 * are met: 45 * 1. Redistributions of source code must retain the above copyright 46 * notice, this list of conditions and the following disclaimer. 47 * 2. Redistributions in binary form must reproduce the above copyright 48 * notice, this list of conditions and the following disclaimer in the 49 * documentation and/or other materials provided with the distribution. 50 * 3. All advertising materials mentioning features or use of this software 51 * must display the following acknowledgement: 52 * This product includes software developed by the University of 53 * California, Berkeley and its contributors. 54 * 4. Neither the name of the University nor the names of its contributors 55 * may be used to endorse or promote products derived from this software 56 * without specific prior written permission. 57 * 58 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 59 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 60 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 61 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 62 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 63 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 64 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 65 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 66 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 67 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 68 * SUCH DAMAGE. 69 * 70 * @(#)init_main.c 8.16 (Berkeley) 5/14/95 71 */ 72 73 #include <sys/cdefs.h> 74 __KERNEL_RCSID(0, "$NetBSD: init_main.c,v 1.267 2006/06/07 22:33:39 kardel Exp $"); 75 76 #include "opt_ipsec.h" 77 #include "opt_kcont.h" 78 #include "opt_multiprocessor.h" 79 #include "opt_ntp.h" 80 #include "opt_pipe.h" 81 #include "opt_posix.h" 82 #include "opt_syscall_debug.h" 83 #include "opt_sysv.h" 84 #include "opt_verified_exec.h" 85 86 #include "rnd.h" 87 88 #include <sys/param.h> 89 #include <sys/acct.h> 90 #include <sys/filedesc.h> 91 #include <sys/file.h> 92 #include <sys/errno.h> 93 #include <sys/callout.h> 94 #include <sys/kernel.h> 95 #include <sys/kcont.h> 96 #include <sys/mount.h> 97 #include <sys/proc.h> 98 #include <sys/kthread.h> 99 #include <sys/resourcevar.h> 100 #include <sys/signalvar.h> 101 #include <sys/systm.h> 102 #include <sys/vnode.h> 103 #include <sys/tty.h> 104 #include <sys/conf.h> 105 #include <sys/disklabel.h> 106 #include <sys/buf.h> 107 #include <sys/device.h> 108 #include <sys/exec.h> 109 #include <sys/socketvar.h> 110 #include <sys/protosw.h> 111 #include <sys/reboot.h> 112 #include <sys/user.h> 113 #include <sys/sysctl.h> 114 #include <sys/event.h> 115 #include <sys/mbuf.h> 116 #ifdef FAST_IPSEC 117 #include <netipsec/ipsec.h> 118 #endif 119 #ifdef SYSVSHM 120 #include <sys/shm.h> 121 #endif 122 #ifdef SYSVSEM 123 #include <sys/sem.h> 124 #endif 125 #ifdef SYSVMSG 126 #include <sys/msg.h> 127 #endif 128 #ifdef P1003_1B_SEMAPHORE 129 #include <sys/ksem.h> 130 #endif 131 #include <sys/domain.h> 132 #include <sys/namei.h> 133 #if NRND > 0 134 #include <sys/rnd.h> 135 #endif 136 #ifndef PIPE_SOCKETPAIR 137 #include <sys/pipe.h> 138 #endif 139 #ifdef LKM 140 #include <sys/lkm.h> 141 #endif 142 #ifdef VERIFIED_EXEC 143 #include <sys/verified_exec.h> 144 #endif 145 #include <sys/kauth.h> 146 #include <net80211/ieee80211_netbsd.h> 147 148 #include <sys/syscall.h> 149 #include <sys/sa.h> 150 #include <sys/syscallargs.h> 151 152 #include <ufs/ufs/quota.h> 153 154 #include <miscfs/genfs/genfs.h> 155 #include <miscfs/syncfs/syncfs.h> 156 157 #include <machine/cpu.h> 158 159 #include <uvm/uvm.h> 160 161 #include <dev/cons.h> 162 163 #include <net/if.h> 164 #include <net/raw_cb.h> 165 166 extern struct proc proc0; 167 extern struct lwp lwp0; 168 extern struct cwdinfo cwdi0; 169 170 #ifndef curlwp 171 struct lwp *curlwp = &lwp0; 172 #endif 173 struct proc *initproc; 174 175 struct vnode *rootvp, *swapdev_vp; 176 int boothowto; 177 int cold = 1; /* still working on startup */ 178 #ifndef __HAVE_TIMECOUNTER 179 struct timeval boottime; 180 #endif 181 time_t rootfstime; /* recorded root fs time, if known */ 182 183 volatile int start_init_exec; /* semaphore for start_init() */ 184 185 static void check_console(struct lwp *l); 186 static void start_init(void *); 187 void main(void); 188 189 /* 190 * System startup; initialize the world, create process 0, mount root 191 * filesystem, and fork to create init and pagedaemon. Most of the 192 * hard work is done in the lower-level initialization routines including 193 * startup(), which does memory initialization and autoconfiguration. 194 */ 195 void 196 main(void) 197 { 198 #ifdef __HAVE_TIMECOUNTER 199 struct timeval time; 200 #endif 201 struct lwp *l; 202 struct proc *p; 203 struct pdevinit *pdev; 204 int s, error; 205 extern struct pdevinit pdevinit[]; 206 extern void schedcpu(void *); 207 #ifdef NVNODE_IMPLICIT 208 int usevnodes; 209 #endif 210 211 /* 212 * Initialize the current LWP pointer (curlwp) before 213 * any possible traps/probes to simplify trap processing. 214 */ 215 l = &lwp0; 216 curlwp = l; 217 l->l_cpu = curcpu(); 218 l->l_proc = &proc0; 219 l->l_lid = 1; 220 221 /* 222 * Attempt to find console and initialize 223 * in case of early panic or other messages. 224 */ 225 consinit(); 226 227 KERNEL_LOCK_INIT(); 228 229 uvm_init(); 230 231 /* Do machine-dependent initialization. */ 232 cpu_startup(); 233 234 /* Initialize callouts. */ 235 callout_startup(); 236 237 /* Initialize the buffer cache */ 238 bufinit(); 239 240 /* 241 * Initialize mbuf's. Do this now because we might attempt to 242 * allocate mbufs or mbuf clusters during autoconfiguration. 243 */ 244 mbinit(); 245 246 /* Initialize sockets. */ 247 soinit(); 248 249 #ifdef KCONT 250 /* Initialize kcont. */ 251 kcont_init(); 252 #endif 253 254 /* 255 * The following things must be done before autoconfiguration. 256 */ 257 evcnt_init(); /* initialize event counters */ 258 #if NRND > 0 259 rnd_init(); /* initialize RNG */ 260 #endif 261 /* Initialize the sysctl subsystem. */ 262 sysctl_init(); 263 264 /* Initialize process and pgrp structures. */ 265 procinit(); 266 267 /* Initialize signal-related data structures. */ 268 signal_init(); 269 270 /* Create process 0 (the swapper). */ 271 proc0_init(); 272 273 /* 274 * Charge root for one process. 275 */ 276 (void)chgproccnt(0, 1); 277 278 rqinit(); 279 280 /* Initialize the file systems. */ 281 #ifdef NVNODE_IMPLICIT 282 /* 283 * If maximum number of vnodes in namei vnode cache is not explicitly 284 * defined in kernel config, adjust the number such as we use roughly 285 * 1.0% of memory for vnode cache (but not less than NVNODE vnodes). 286 */ 287 usevnodes = (ptoa((unsigned)physmem) / 100) / sizeof(struct vnode); 288 if (usevnodes > desiredvnodes) 289 desiredvnodes = usevnodes; 290 #endif 291 vfsinit(); 292 293 294 #ifdef __HAVE_TIMECOUNTER 295 #ifdef NTP 296 ntp_init(); 297 #endif 298 #endif /* __HAVE_TIMECOUNTER */ 299 300 /* Configure the system hardware. This will enable interrupts. */ 301 configure(); 302 303 ubc_init(); /* must be after autoconfig */ 304 305 /* Lock the kernel on behalf of proc0. */ 306 KERNEL_PROC_LOCK(l); 307 308 #ifdef SYSVSHM 309 /* Initialize System V style shared memory. */ 310 shminit(); 311 #endif 312 313 #ifdef SYSVSEM 314 /* Initialize System V style semaphores. */ 315 seminit(); 316 #endif 317 318 #ifdef SYSVMSG 319 /* Initialize System V style message queues. */ 320 msginit(); 321 #endif 322 323 #ifdef P1003_1B_SEMAPHORE 324 /* Initialize posix semaphores */ 325 ksem_init(); 326 #endif 327 328 /* Initialize kauth. */ 329 kauth_init(); 330 331 #ifdef VERIFIED_EXEC 332 /* 333 * Initialise the fingerprint operations vectors before 334 * fingerprints can be loaded. 335 */ 336 veriexec_init_fp_ops(); 337 #endif 338 339 /* Attach pseudo-devices. */ 340 for (pdev = pdevinit; pdev->pdev_attach != NULL; pdev++) 341 (*pdev->pdev_attach)(pdev->pdev_count); 342 343 #ifdef FAST_IPSEC 344 /* Attach network crypto subsystem */ 345 ipsec_attach(); 346 #endif 347 348 /* 349 * Initialize protocols. Block reception of incoming packets 350 * until everything is ready. 351 */ 352 s = splnet(); 353 ifinit(); 354 domaininit(); 355 if_attachdomain(); 356 splx(s); 357 358 #ifdef GPROF 359 /* Initialize kernel profiling. */ 360 kmstartup(); 361 #endif 362 363 /* Initialize system accouting. */ 364 acct_init(); 365 366 /* Kick off timeout driven events by calling first time. */ 367 schedcpu(NULL); 368 369 /* 370 * Create process 1 (init(8)). We do this now, as Unix has 371 * historically had init be process 1, and changing this would 372 * probably upset a lot of people. 373 * 374 * Note that process 1 won't immediately exec init(8), but will 375 * wait for us to inform it that the root file system has been 376 * mounted. 377 */ 378 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 379 panic("fork init"); 380 381 /* 382 * Create any kernel threads who's creation was deferred because 383 * initproc had not yet been created. 384 */ 385 kthread_run_deferred_queue(); 386 387 /* 388 * Now that device driver threads have been created, wait for 389 * them to finish any deferred autoconfiguration. Note we don't 390 * need to lock this semaphore, since we haven't booted any 391 * secondary processors, yet. 392 */ 393 while (config_pending) 394 (void) tsleep(&config_pending, PWAIT, "cfpend", 0); 395 396 /* 397 * Finalize configuration now that all real devices have been 398 * found. This needs to be done before the root device is 399 * selected, since finalization may create the root device. 400 */ 401 config_finalize(); 402 403 /* 404 * Now that autoconfiguration has completed, we can determine 405 * the root and dump devices. 406 */ 407 cpu_rootconf(); 408 cpu_dumpconf(); 409 410 /* Mount the root file system. */ 411 do { 412 domountroothook(); 413 if ((error = vfs_mountroot())) { 414 printf("cannot mount root, error = %d\n", error); 415 boothowto |= RB_ASKNAME; 416 setroot(root_device, 417 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 418 } 419 } while (error != 0); 420 mountroothook_destroy(); 421 422 /* 423 * Initialise the time-of-day clock, passing the time recorded 424 * in the root filesystem (if any) for use by systems that 425 * don't have a non-volatile time-of-day device. 426 */ 427 inittodr(rootfstime); 428 429 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 430 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 431 432 /* 433 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 434 * reference it. 435 */ 436 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 437 if (error) 438 panic("cannot find root vnode, error=%d", error); 439 cwdi0.cwdi_cdir = rootvnode; 440 VREF(cwdi0.cwdi_cdir); 441 VOP_UNLOCK(rootvnode, 0); 442 cwdi0.cwdi_rdir = NULL; 443 444 /* 445 * Now that root is mounted, we can fixup initproc's CWD 446 * info. All other processes are kthreads, which merely 447 * share proc0's CWD info. 448 */ 449 initproc->p_cwdi->cwdi_cdir = rootvnode; 450 VREF(initproc->p_cwdi->cwdi_cdir); 451 initproc->p_cwdi->cwdi_rdir = NULL; 452 453 /* 454 * Now can look at time, having had a chance to verify the time 455 * from the file system. Reset p->p_rtime as it may have been 456 * munched in mi_switch() after the time got set. 457 */ 458 proclist_lock_read(); 459 s = splsched(); 460 #ifdef __HAVE_TIMECOUNTER 461 getmicrotime(&time); 462 #else 463 mono_time = time; 464 #endif 465 boottime = time; 466 LIST_FOREACH(p, &allproc, p_list) { 467 KASSERT((p->p_flag & P_MARKER) == 0); 468 p->p_stats->p_start = time; 469 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 470 if (l->l_cpu != NULL) 471 l->l_cpu->ci_schedstate.spc_runtime = time; 472 } 473 p->p_rtime.tv_sec = p->p_rtime.tv_usec = 0; 474 } 475 splx(s); 476 proclist_unlock_read(); 477 478 /* Create the pageout daemon kernel thread. */ 479 uvm_swap_init(); 480 if (kthread_create1(uvm_pageout, NULL, NULL, "pagedaemon")) 481 panic("fork pagedaemon"); 482 483 /* Create the filesystem syncer kernel thread. */ 484 if (kthread_create1(sched_sync, NULL, NULL, "ioflush")) 485 panic("fork syncer"); 486 487 /* Create the aiodone daemon kernel thread. */ 488 if (kthread_create1(uvm_aiodone_daemon, NULL, &uvm.aiodoned_proc, 489 "aiodoned")) 490 panic("fork aiodoned"); 491 492 #if defined(MULTIPROCESSOR) 493 /* Boot the secondary processors. */ 494 cpu_boot_secondary_processors(); 495 #endif 496 497 /* Initialize exec structures */ 498 exec_init(1); 499 500 /* 501 * Okay, now we can let init(8) exec! It's off to userland! 502 */ 503 start_init_exec = 1; 504 wakeup(&start_init_exec); 505 506 /* The scheduler is an infinite loop. */ 507 uvm_scheduler(); 508 /* NOTREACHED */ 509 } 510 511 void 512 setrootfstime(time_t t) 513 { 514 rootfstime = t; 515 } 516 517 static void 518 check_console(struct lwp *l) 519 { 520 struct nameidata nd; 521 int error; 522 523 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console", l); 524 error = namei(&nd); 525 if (error == 0) 526 vrele(nd.ni_vp); 527 else if (error == ENOENT) 528 printf("warning: no /dev/console\n"); 529 else 530 printf("warning: lookup /dev/console: error %d\n", error); 531 } 532 533 /* 534 * List of paths to try when searching for "init". 535 */ 536 static const char *initpaths[] = { 537 "/sbin/init", 538 "/sbin/oinit", 539 "/sbin/init.bak", 540 NULL, 541 }; 542 543 /* 544 * Start the initial user process; try exec'ing each pathname in "initpaths". 545 * The program is invoked with one argument containing the boot flags. 546 */ 547 static void 548 start_init(void *arg) 549 { 550 struct lwp *l = arg; 551 struct proc *p = l->l_proc; 552 vaddr_t addr; 553 struct sys_execve_args /* { 554 syscallarg(const char *) path; 555 syscallarg(char * const *) argp; 556 syscallarg(char * const *) envp; 557 } */ args; 558 int options, i, error; 559 register_t retval[2]; 560 char flags[4], *flagsp; 561 const char *path, *slash; 562 char *ucp, **uap, *arg0, *arg1 = NULL; 563 char ipath[129]; 564 int ipx, len; 565 566 /* 567 * Now in process 1. 568 */ 569 strncpy(p->p_comm, "init", MAXCOMLEN); 570 571 /* 572 * Wait for main() to tell us that it's safe to exec. 573 */ 574 while (start_init_exec == 0) 575 (void) tsleep(&start_init_exec, PWAIT, "initexec", 0); 576 577 /* 578 * This is not the right way to do this. We really should 579 * hand-craft a descriptor onto /dev/console to hand to init, 580 * but that's a _lot_ more work, and the benefit from this easy 581 * hack makes up for the "good is the enemy of the best" effect. 582 */ 583 check_console(l); 584 585 /* 586 * Need just enough stack to hold the faked-up "execve()" arguments. 587 */ 588 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 589 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 590 NULL, UVM_UNKNOWN_OFFSET, 0, 591 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 592 UVM_ADV_NORMAL, 593 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 594 panic("init: couldn't allocate argument space"); 595 p->p_vmspace->vm_maxsaddr = (caddr_t)STACK_MAX(addr, PAGE_SIZE); 596 597 ipx = 0; 598 while (1) { 599 if (boothowto & RB_ASKNAME) { 600 printf("init path"); 601 if (initpaths[ipx]) 602 printf(" (default %s)", initpaths[ipx]); 603 printf(": "); 604 len = cngetsn(ipath, sizeof(ipath)-1); 605 if (len == 0) { 606 if (initpaths[ipx]) 607 path = initpaths[ipx++]; 608 else 609 continue; 610 } else { 611 ipath[len] = '\0'; 612 path = ipath; 613 } 614 } else { 615 if ((path = initpaths[ipx++]) == NULL) 616 break; 617 } 618 619 ucp = (char *)USRSTACK; 620 621 /* 622 * Construct the boot flag argument. 623 */ 624 flagsp = flags; 625 *flagsp++ = '-'; 626 options = 0; 627 628 if (boothowto & RB_SINGLE) { 629 *flagsp++ = 's'; 630 options = 1; 631 } 632 #ifdef notyet 633 if (boothowto & RB_FASTBOOT) { 634 *flagsp++ = 'f'; 635 options = 1; 636 } 637 #endif 638 639 /* 640 * Move out the flags (arg 1), if necessary. 641 */ 642 if (options != 0) { 643 *flagsp++ = '\0'; 644 i = flagsp - flags; 645 #ifdef DEBUG 646 printf("init: copying out flags `%s' %d\n", flags, i); 647 #endif 648 arg1 = STACK_ALLOC(ucp, i); 649 ucp = STACK_MAX(arg1, i); 650 (void)copyout((caddr_t)flags, arg1, i); 651 } 652 653 /* 654 * Move out the file name (also arg 0). 655 */ 656 i = strlen(path) + 1; 657 #ifdef DEBUG 658 printf("init: copying out path `%s' %d\n", path, i); 659 #else 660 if (boothowto & RB_ASKNAME || path != initpaths[0]) 661 printf("init: trying %s\n", path); 662 #endif 663 arg0 = STACK_ALLOC(ucp, i); 664 ucp = STACK_MAX(arg0, i); 665 (void)copyout(path, arg0, i); 666 667 /* 668 * Move out the arg pointers. 669 */ 670 ucp = (caddr_t)STACK_ALIGN(ucp, ALIGNBYTES); 671 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 672 SCARG(&args, path) = arg0; 673 SCARG(&args, argp) = uap; 674 SCARG(&args, envp) = NULL; 675 slash = strrchr(path, '/'); 676 if (slash) 677 (void)suword((caddr_t)uap++, 678 (long)arg0 + (slash + 1 - path)); 679 else 680 (void)suword((caddr_t)uap++, (long)arg0); 681 if (options != 0) 682 (void)suword((caddr_t)uap++, (long)arg1); 683 (void)suword((caddr_t)uap++, 0); /* terminator */ 684 685 /* 686 * Now try to exec the program. If can't for any reason 687 * other than it doesn't exist, complain. 688 */ 689 error = sys_execve(l, &args, retval); 690 if (error == 0 || error == EJUSTRETURN) { 691 KERNEL_PROC_UNLOCK(l); 692 return; 693 } 694 printf("exec %s: error %d\n", path, error); 695 } 696 printf("init: not found\n"); 697 panic("no init"); 698 } 699
Indexes created Tue Sep 23 14:10:03 GMT 2025