init_main.c revision 1.317
1 /* $NetBSD: init_main.c,v 1.317 2007/09/25 22:33:59 ad 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 1.317 2007/09/25 22:33:59 ad Exp $"); 75 76 #include "opt_ipsec.h" 77 #include "opt_multiprocessor.h" 78 #include "opt_ntp.h" 79 #include "opt_pipe.h" 80 #include "opt_posix.h" 81 #include "opt_syscall_debug.h" 82 #include "opt_sysv.h" 83 #include "opt_systrace.h" 84 #include "opt_fileassoc.h" 85 #include "opt_ktrace.h" 86 #include "opt_pax.h" 87 88 #include "rnd.h" 89 #include "sysmon_envsys.h" 90 #include "sysmon_power.h" 91 #include "sysmon_taskq.h" 92 #include "sysmon_wdog.h" 93 #include "veriexec.h" 94 95 #include <sys/param.h> 96 #include <sys/acct.h> 97 #include <sys/filedesc.h> 98 #include <sys/file.h> 99 #include <sys/errno.h> 100 #include <sys/callout.h> 101 #include <sys/cpu.h> 102 #include <sys/kernel.h> 103 #include <sys/kmem.h> 104 #include <sys/mount.h> 105 #include <sys/proc.h> 106 #include <sys/kthread.h> 107 #include <sys/resourcevar.h> 108 #include <sys/signalvar.h> 109 #include <sys/systm.h> 110 #include <sys/vnode.h> 111 #include <sys/fstrans.h> 112 #include <sys/tty.h> 113 #include <sys/conf.h> 114 #include <sys/disklabel.h> 115 #include <sys/buf.h> 116 #include <sys/device.h> 117 #include <sys/exec.h> 118 #include <sys/socketvar.h> 119 #include <sys/protosw.h> 120 #include <sys/reboot.h> 121 #include <sys/user.h> 122 #include <sys/sysctl.h> 123 #include <sys/event.h> 124 #include <sys/mbuf.h> 125 #include <sys/sched.h> 126 #include <sys/sleepq.h> 127 #include <sys/iostat.h> 128 #include <sys/vmem.h> 129 #include <sys/uuid.h> 130 #include <sys/extent.h> 131 #include <sys/disk.h> 132 #include <sys/mqueue.h> 133 #ifdef FAST_IPSEC 134 #include <netipsec/ipsec.h> 135 #endif 136 #ifdef SYSVSHM 137 #include <sys/shm.h> 138 #endif 139 #ifdef SYSVSEM 140 #include <sys/sem.h> 141 #endif 142 #ifdef SYSVMSG 143 #include <sys/msg.h> 144 #endif 145 #ifdef SYSTRACE 146 #include <sys/systrace.h> 147 #endif 148 #ifdef P1003_1B_SEMAPHORE 149 #include <sys/ksem.h> 150 #endif 151 #include <sys/domain.h> 152 #include <sys/namei.h> 153 #if NRND > 0 154 #include <sys/rnd.h> 155 #endif 156 #include <sys/pipe.h> 157 #ifdef LKM 158 #include <sys/lkm.h> 159 #endif 160 #if NVERIEXEC > 0 161 #include <sys/verified_exec.h> 162 #endif /* NVERIEXEC > 0 */ 163 #ifdef KTRACE 164 #include <sys/ktrace.h> 165 #endif 166 #include <sys/debug.h> 167 #include <sys/kauth.h> 168 #include <net80211/ieee80211_netbsd.h> 169 170 #include <sys/syscall.h> 171 #include <sys/syscallargs.h> 172 173 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) 174 #include <sys/pax.h> 175 #endif /* PAX_MPROTECT || PAX_SEGVGUARD */ 176 #include <ufs/ufs/quota.h> 177 178 #ifdef DDB 179 #include <machine/db_machdep.h> 180 #include <ddb/db_command.h> 181 #endif 182 183 #include <miscfs/genfs/genfs.h> 184 #include <miscfs/syncfs/syncfs.h> 185 186 #include <machine/cpu.h> 187 188 #include <uvm/uvm.h> 189 190 #if NSYSMON_TASKQ > 0 191 #include <dev/sysmon/sysmon_taskq.h> 192 #endif 193 194 #include <dev/cons.h> 195 196 #if NSYSMON_ENVSYS > 0 || NSYSMON_POWER > 0 || NSYSMON_WDOG > 0 197 #include <dev/sysmon/sysmonvar.h> 198 #endif 199 200 #include <net/if.h> 201 #include <net/raw_cb.h> 202 203 #include <secmodel/secmodel.h> 204 205 extern struct proc proc0; 206 extern struct lwp lwp0; 207 extern struct cwdinfo cwdi0; 208 extern time_t rootfstime; 209 210 #ifndef curlwp 211 struct lwp *curlwp = &lwp0; 212 #endif 213 struct proc *initproc; 214 215 struct vnode *rootvp, *swapdev_vp; 216 int boothowto; 217 int cold = 1; /* still working on startup */ 218 struct timeval boottime; /* time at system startup - will only follow settime deltas */ 219 int ncpu = 0; /* number of CPUs configured, assume 1 */ 220 221 volatile int start_init_exec; /* semaphore for start_init() */ 222 223 static void check_console(struct lwp *l); 224 static void start_init(void *); 225 void main(void); 226 227 #if defined(__SSP__) || defined(__SSP_ALL__) 228 long __stack_chk_guard[8] = {0, 0, 0, 0, 0, 0, 0, 0}; 229 void __stack_chk_fail(void); 230 231 void 232 __stack_chk_fail(void) 233 { 234 panic("stack overflow detected; terminated"); 235 } 236 #endif 237 238 void __secmodel_none(void); 239 __weak_alias(secmodel_start,__secmodel_none); 240 void 241 __secmodel_none(void) 242 { 243 return; 244 } 245 246 /* 247 * System startup; initialize the world, create process 0, mount root 248 * filesystem, and fork to create init and pagedaemon. Most of the 249 * hard work is done in the lower-level initialization routines including 250 * startup(), which does memory initialization and autoconfiguration. 251 */ 252 void 253 main(void) 254 { 255 #ifdef __HAVE_TIMECOUNTER 256 struct timeval time; 257 #endif 258 struct lwp *l; 259 struct proc *p; 260 struct pdevinit *pdev; 261 int s, error; 262 extern struct pdevinit pdevinit[]; 263 #ifdef NVNODE_IMPLICIT 264 int usevnodes; 265 #endif 266 CPU_INFO_ITERATOR cii; 267 struct cpu_info *ci; 268 269 l = &lwp0; 270 l->l_cpu = curcpu(); 271 l->l_proc = &proc0; 272 l->l_lid = 1; 273 274 /* 275 * XXX This is a temporary check to be removed before 276 * NetBSD 5.0 is released. 277 */ 278 #if !defined(__i386__ ) && !defined(__x86_64__) 279 if (curlwp != l) { 280 printf("NOTICE: curlwp should be set before main()\n"); 281 DELAY(250000); 282 curlwp = l; 283 } 284 #endif 285 286 /* 287 * Attempt to find console and initialize 288 * in case of early panic or other messages. 289 */ 290 consinit(); 291 292 KERNEL_LOCK_INIT(); 293 294 uvm_init(); 295 296 #ifdef DEBUG 297 debug_init(); 298 #endif 299 300 kmem_init(); 301 302 /* Initialize the extent manager. */ 303 extent_init(); 304 305 /* Do machine-dependent initialization. */ 306 cpu_startup(); 307 308 /* Initialize callouts, part 1. */ 309 callout_startup(); 310 311 /* 312 * Initialize the kernel authorization subsystem and start the 313 * default security model, if any. We need to do this early 314 * enough so that subsystems relying on any of the aforementioned 315 * can work properly. Since the security model may dictate the 316 * credential inheritance policy, it is needed at least before 317 * any process is created, specifically proc0. 318 */ 319 kauth_init(); 320 secmodel_start(); 321 322 /* Initialize the buffer cache */ 323 bufinit(); 324 325 /* 326 * Initialize mbuf's. Do this now because we might attempt to 327 * allocate mbufs or mbuf clusters during autoconfiguration. 328 */ 329 mbinit(); 330 331 /* Initialize sockets. */ 332 soinit(); 333 334 /* 335 * The following things must be done before autoconfiguration. 336 */ 337 evcnt_init(); /* initialize event counters */ 338 #if NRND > 0 339 rnd_init(); /* initialize RNG */ 340 #endif 341 342 /* Initialize process and pgrp structures. */ 343 procinit(); 344 lwpinit(); 345 346 /* Initialize signal-related data structures. */ 347 signal_init(); 348 349 /* Create process 0 (the swapper). */ 350 proc0_init(); 351 352 /* Initialize the UID hash table. */ 353 uid_init(); 354 355 /* Charge root for one process. */ 356 (void)chgproccnt(0, 1); 357 358 /* Initialize the run queues, turnstiles and sleep queues. */ 359 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 360 sched_rqinit(); 361 turnstile_init(); 362 sleeptab_init(&sleeptab); 363 364 /* MI initialization of the boot cpu */ 365 error = mi_cpu_attach(curcpu()); 366 KASSERT(error == 0); 367 368 /* Initialize the sysctl subsystem. */ 369 sysctl_init(); 370 371 /* Initialize I/O statistics. */ 372 iostat_init(); 373 374 /* Initialize the file systems. */ 375 #ifdef NVNODE_IMPLICIT 376 /* 377 * If maximum number of vnodes in namei vnode cache is not explicitly 378 * defined in kernel config, adjust the number such as we use roughly 379 * 1.0% of memory for vnode cache (but not less than NVNODE vnodes). 380 */ 381 usevnodes = (ptoa((unsigned)physmem) / 100) / sizeof(struct vnode); 382 if (usevnodes > desiredvnodes) 383 desiredvnodes = usevnodes; 384 #endif 385 vfsinit(); 386 387 /* Initialize fstrans. */ 388 fstrans_init(); 389 390 /* Initialize the select()/poll() system calls. */ 391 selsysinit(); 392 393 /* Initialize asynchronous I/O. */ 394 aio_sysinit(); 395 396 /* Initialize message queues. */ 397 mqueue_sysinit(); 398 399 /* Initialize the system monitor subsystems. */ 400 #if NSYSMON_TASKQ > 0 401 sysmon_task_queue_preinit(); 402 #endif 403 404 #if NSYSMON_ENVSYS > 0 405 sysmon_envsys_init(); 406 #endif 407 408 #if NSYSMON_POWER > 0 409 sysmon_power_init(); 410 #endif 411 412 #if NSYSMON_WDOG > 0 413 sysmon_wdog_init(); 414 #endif 415 416 #ifdef __HAVE_TIMECOUNTER 417 inittimecounter(); 418 ntp_init(); 419 #endif /* __HAVE_TIMECOUNTER */ 420 421 #ifdef DDB 422 db_init_commands(); 423 #endif 424 425 /* Initialize the device switch tables. */ 426 devsw_init(); 427 428 /* Initialize the disk wedge subsystem. */ 429 dkwedge_init(); 430 431 /* Configure the system hardware. This will enable interrupts. */ 432 configure(); 433 434 #if defined(__SSP__) || defined(__SSP_ALL__) 435 { 436 #ifdef DIAGNOSTIC 437 printf("Initializing SSP:"); 438 #endif 439 /* 440 * We initialize ssp here carefully: 441 * 1. after we got some entropy 442 * 2. without calling a function 443 */ 444 size_t i; 445 long guard[__arraycount(__stack_chk_guard)]; 446 447 arc4randbytes(guard, sizeof(guard)); 448 for (i = 0; i < __arraycount(guard); i++) 449 __stack_chk_guard[i] = guard[i]; 450 #ifdef DIAGNOSTIC 451 for (i = 0; i < __arraycount(guard); i++) 452 printf("%lx ", guard[i]); 453 printf("\n"); 454 #endif 455 } 456 #endif 457 ubc_init(); /* must be after autoconfig */ 458 459 /* Lock the kernel on behalf of proc0. */ 460 KERNEL_LOCK(1, l); 461 462 #ifdef SYSTRACE 463 systrace_init(); 464 #endif 465 466 #ifdef SYSVSHM 467 /* Initialize System V style shared memory. */ 468 shminit(); 469 #endif 470 471 #ifdef SYSVSEM 472 /* Initialize System V style semaphores. */ 473 seminit(); 474 #endif 475 476 #ifdef SYSVMSG 477 /* Initialize System V style message queues. */ 478 msginit(); 479 #endif 480 481 #ifdef P1003_1B_SEMAPHORE 482 /* Initialize posix semaphores */ 483 ksem_init(); 484 #endif 485 486 #if NVERIEXEC > 0 487 /* 488 * Initialise the Veriexec subsystem. 489 */ 490 veriexec_init(); 491 #endif /* NVERIEXEC > 0 */ 492 493 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) 494 pax_init(); 495 #endif /* PAX_MPROTECT || PAX_SEGVGUARD */ 496 497 /* Attach pseudo-devices. */ 498 for (pdev = pdevinit; pdev->pdev_attach != NULL; pdev++) 499 (*pdev->pdev_attach)(pdev->pdev_count); 500 501 #ifdef FAST_IPSEC 502 /* Attach network crypto subsystem */ 503 ipsec_attach(); 504 #endif 505 506 /* 507 * Initialize protocols. Block reception of incoming packets 508 * until everything is ready. 509 */ 510 s = splnet(); 511 ifinit(); 512 domaininit(); 513 if_attachdomain(); 514 splx(s); 515 516 #ifdef GPROF 517 /* Initialize kernel profiling. */ 518 kmstartup(); 519 #endif 520 521 /* Initialize system accouting. */ 522 acct_init(); 523 524 #ifndef PIPE_SOCKETPAIR 525 /* Initialize pipes. */ 526 pipe_init(); 527 #endif 528 529 /* Setup the scheduler */ 530 sched_init(); 531 532 #ifdef KTRACE 533 /* Initialize ktrace. */ 534 ktrinit(); 535 #endif 536 537 /* Initialize the UUID system calls. */ 538 uuid_init(); 539 540 /* 541 * Create process 1 (init(8)). We do this now, as Unix has 542 * historically had init be process 1, and changing this would 543 * probably upset a lot of people. 544 * 545 * Note that process 1 won't immediately exec init(8), but will 546 * wait for us to inform it that the root file system has been 547 * mounted. 548 */ 549 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 550 panic("fork init"); 551 552 /* 553 * Now that device driver threads have been created, wait for 554 * them to finish any deferred autoconfiguration. Note we don't 555 * need to lock this semaphore, since we haven't booted any 556 * secondary processors, yet. 557 */ 558 while (config_pending) 559 (void) tsleep(&config_pending, PWAIT, "cfpend", 0); 560 561 /* 562 * Finalize configuration now that all real devices have been 563 * found. This needs to be done before the root device is 564 * selected, since finalization may create the root device. 565 */ 566 config_finalize(); 567 568 /* 569 * Now that autoconfiguration has completed, we can determine 570 * the root and dump devices. 571 */ 572 cpu_rootconf(); 573 cpu_dumpconf(); 574 575 /* Mount the root file system. */ 576 do { 577 domountroothook(); 578 if ((error = vfs_mountroot())) { 579 printf("cannot mount root, error = %d\n", error); 580 boothowto |= RB_ASKNAME; 581 setroot(root_device, 582 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 583 } 584 } while (error != 0); 585 mountroothook_destroy(); 586 587 /* 588 * Initialise the time-of-day clock, passing the time recorded 589 * in the root filesystem (if any) for use by systems that 590 * don't have a non-volatile time-of-day device. 591 */ 592 inittodr(rootfstime); 593 594 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 595 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 596 597 /* 598 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 599 * reference it. 600 */ 601 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 602 if (error) 603 panic("cannot find root vnode, error=%d", error); 604 cwdi0.cwdi_cdir = rootvnode; 605 VREF(cwdi0.cwdi_cdir); 606 VOP_UNLOCK(rootvnode, 0); 607 cwdi0.cwdi_rdir = NULL; 608 609 /* 610 * Now that root is mounted, we can fixup initproc's CWD 611 * info. All other processes are kthreads, which merely 612 * share proc0's CWD info. 613 */ 614 initproc->p_cwdi->cwdi_cdir = rootvnode; 615 VREF(initproc->p_cwdi->cwdi_cdir); 616 initproc->p_cwdi->cwdi_rdir = NULL; 617 618 /* 619 * Now can look at time, having had a chance to verify the time 620 * from the file system. Reset l->l_rtime as it may have been 621 * munched in mi_switch() after the time got set. 622 */ 623 #ifdef __HAVE_TIMECOUNTER 624 getmicrotime(&time); 625 #else 626 mono_time = time; 627 #endif 628 boottime = time; 629 mutex_enter(&proclist_lock); 630 LIST_FOREACH(p, &allproc, p_list) { 631 KASSERT((p->p_flag & PK_MARKER) == 0); 632 mutex_enter(&p->p_smutex); 633 p->p_stats->p_start = time; 634 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 635 lwp_lock(l); 636 l->l_cpu->ci_schedstate.spc_runtime = time; 637 l->l_rtime.tv_sec = l->l_rtime.tv_usec = 0; 638 lwp_unlock(l); 639 } 640 mutex_exit(&p->p_smutex); 641 } 642 mutex_exit(&proclist_lock); 643 644 for (CPU_INFO_FOREACH(cii, ci)) { 645 ci->ci_schedstate.spc_lastmod = time_second; 646 } 647 648 /* Create the pageout daemon kernel thread. */ 649 uvm_swap_init(); 650 if (kthread_create(PVM, 0, NULL, uvm_pageout, 651 NULL, NULL, "pgdaemon")) 652 panic("fork pagedaemon"); 653 654 /* Create the filesystem syncer kernel thread. */ 655 if (kthread_create(PINOD, 0, NULL, sched_sync, NULL, NULL, "ioflush")) 656 panic("fork syncer"); 657 658 /* Create the aiodone daemon kernel thread. */ 659 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 660 uvm_aiodone_worker, NULL, PVM, IPL_BIO, 0)) 661 panic("fork aiodoned"); 662 663 vmem_rehash_start(); 664 665 #if defined(MULTIPROCESSOR) 666 /* Boot the secondary processors. */ 667 cpu_boot_secondary_processors(); 668 #endif 669 670 /* Initialize exec structures */ 671 exec_init(1); 672 673 /* 674 * Okay, now we can let init(8) exec! It's off to userland! 675 */ 676 start_init_exec = 1; 677 wakeup(&start_init_exec); 678 679 /* The scheduler is an infinite loop. */ 680 uvm_scheduler(); 681 /* NOTREACHED */ 682 } 683 684 static void 685 check_console(struct lwp *l) 686 { 687 struct nameidata nd; 688 int error; 689 690 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console", l); 691 error = namei(&nd); 692 if (error == 0) 693 vrele(nd.ni_vp); 694 else if (error == ENOENT) 695 printf("warning: no /dev/console\n"); 696 else 697 printf("warning: lookup /dev/console: error %d\n", error); 698 } 699 700 /* 701 * List of paths to try when searching for "init". 702 */ 703 static const char *initpaths[] = { 704 "/sbin/init", 705 "/sbin/oinit", 706 "/sbin/init.bak", 707 NULL, 708 }; 709 710 /* 711 * Start the initial user process; try exec'ing each pathname in "initpaths". 712 * The program is invoked with one argument containing the boot flags. 713 */ 714 static void 715 start_init(void *arg) 716 { 717 struct lwp *l = arg; 718 struct proc *p = l->l_proc; 719 vaddr_t addr; 720 struct sys_execve_args /* { 721 syscallarg(const char *) path; 722 syscallarg(char * const *) argp; 723 syscallarg(char * const *) envp; 724 } */ args; 725 int options, i, error; 726 register_t retval[2]; 727 char flags[4], *flagsp; 728 const char *path, *slash; 729 char *ucp, **uap, *arg0, *arg1 = NULL; 730 char ipath[129]; 731 int ipx, len; 732 733 /* 734 * Now in process 1. 735 */ 736 strncpy(p->p_comm, "init", MAXCOMLEN); 737 738 /* 739 * Wait for main() to tell us that it's safe to exec. 740 */ 741 while (start_init_exec == 0) 742 (void) tsleep(&start_init_exec, PWAIT, "initexec", 0); 743 744 /* 745 * This is not the right way to do this. We really should 746 * hand-craft a descriptor onto /dev/console to hand to init, 747 * but that's a _lot_ more work, and the benefit from this easy 748 * hack makes up for the "good is the enemy of the best" effect. 749 */ 750 check_console(l); 751 752 /* 753 * Need just enough stack to hold the faked-up "execve()" arguments. 754 */ 755 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 756 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 757 NULL, UVM_UNKNOWN_OFFSET, 0, 758 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 759 UVM_ADV_NORMAL, 760 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 761 panic("init: couldn't allocate argument space"); 762 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 763 764 ipx = 0; 765 while (1) { 766 if (boothowto & RB_ASKNAME) { 767 printf("init path"); 768 if (initpaths[ipx]) 769 printf(" (default %s)", initpaths[ipx]); 770 printf(": "); 771 len = cngetsn(ipath, sizeof(ipath)-1); 772 if (len == 0) { 773 if (initpaths[ipx]) 774 path = initpaths[ipx++]; 775 else 776 continue; 777 } else { 778 ipath[len] = '\0'; 779 path = ipath; 780 } 781 } else { 782 if ((path = initpaths[ipx++]) == NULL) 783 break; 784 } 785 786 ucp = (char *)USRSTACK; 787 788 /* 789 * Construct the boot flag argument. 790 */ 791 flagsp = flags; 792 *flagsp++ = '-'; 793 options = 0; 794 795 if (boothowto & RB_SINGLE) { 796 *flagsp++ = 's'; 797 options = 1; 798 } 799 #ifdef notyet 800 if (boothowto & RB_FASTBOOT) { 801 *flagsp++ = 'f'; 802 options = 1; 803 } 804 #endif 805 806 /* 807 * Move out the flags (arg 1), if necessary. 808 */ 809 if (options != 0) { 810 *flagsp++ = '\0'; 811 i = flagsp - flags; 812 #ifdef DEBUG 813 printf("init: copying out flags `%s' %d\n", flags, i); 814 #endif 815 arg1 = STACK_ALLOC(ucp, i); 816 ucp = STACK_MAX(arg1, i); 817 (void)copyout((void *)flags, arg1, i); 818 } 819 820 /* 821 * Move out the file name (also arg 0). 822 */ 823 i = strlen(path) + 1; 824 #ifdef DEBUG 825 printf("init: copying out path `%s' %d\n", path, i); 826 #else 827 if (boothowto & RB_ASKNAME || path != initpaths[0]) 828 printf("init: trying %s\n", path); 829 #endif 830 arg0 = STACK_ALLOC(ucp, i); 831 ucp = STACK_MAX(arg0, i); 832 (void)copyout(path, arg0, i); 833 834 /* 835 * Move out the arg pointers. 836 */ 837 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 838 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 839 SCARG(&args, path) = arg0; 840 SCARG(&args, argp) = uap; 841 SCARG(&args, envp) = NULL; 842 slash = strrchr(path, '/'); 843 if (slash) 844 (void)suword((void *)uap++, 845 (long)arg0 + (slash + 1 - path)); 846 else 847 (void)suword((void *)uap++, (long)arg0); 848 if (options != 0) 849 (void)suword((void *)uap++, (long)arg1); 850 (void)suword((void *)uap++, 0); /* terminator */ 851 852 /* 853 * Now try to exec the program. If can't for any reason 854 * other than it doesn't exist, complain. 855 */ 856 error = sys_execve(l, &args, retval); 857 if (error == 0 || error == EJUSTRETURN) { 858 KERNEL_UNLOCK_LAST(l); 859 return; 860 } 861 printf("exec %s: error %d\n", path, error); 862 } 863 printf("init: not found\n"); 864 panic("no init"); 865 } 866
Indexes created Sun Sep 21 16:09:51 GMT 2025