init_main.c revision 1.327
1 /* $NetBSD: init_main.c,v 1.327 2007/11/07 16:51:28 matt 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 1.327 2007/11/07 16:51:28 matt 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 #include <sys/msgbuf.h> 134 #ifdef FAST_IPSEC 135 #include <netipsec/ipsec.h> 136 #endif 137 #ifdef SYSVSHM 138 #include <sys/shm.h> 139 #endif 140 #ifdef SYSVSEM 141 #include <sys/sem.h> 142 #endif 143 #ifdef SYSVMSG 144 #include <sys/msg.h> 145 #endif 146 #ifdef SYSTRACE 147 #include <sys/systrace.h> 148 #endif 149 #ifdef P1003_1B_SEMAPHORE 150 #include <sys/ksem.h> 151 #endif 152 #include <sys/domain.h> 153 #include <sys/namei.h> 154 #if NRND > 0 155 #include <sys/rnd.h> 156 #endif 157 #include <sys/pipe.h> 158 #ifdef LKM 159 #include <sys/lkm.h> 160 #endif 161 #if NVERIEXEC > 0 162 #include <sys/verified_exec.h> 163 #endif /* NVERIEXEC > 0 */ 164 #ifdef KTRACE 165 #include <sys/ktrace.h> 166 #endif 167 #include <sys/debug.h> 168 #include <sys/kauth.h> 169 #include <net80211/ieee80211_netbsd.h> 170 171 #include <sys/syscall.h> 172 #include <sys/syscallargs.h> 173 174 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) 175 #include <sys/pax.h> 176 #endif /* PAX_MPROTECT || PAX_SEGVGUARD */ 177 #include <ufs/ufs/quota.h> 178 179 #include <miscfs/genfs/genfs.h> 180 #include <miscfs/syncfs/syncfs.h> 181 182 #include <sys/cpu.h> 183 184 #include <uvm/uvm.h> 185 186 #if NSYSMON_TASKQ > 0 187 #include <dev/sysmon/sysmon_taskq.h> 188 #endif 189 190 #include <dev/cons.h> 191 192 #if NSYSMON_ENVSYS > 0 || NSYSMON_POWER > 0 || NSYSMON_WDOG > 0 193 #include <dev/sysmon/sysmonvar.h> 194 #endif 195 196 #include <net/if.h> 197 #include <net/raw_cb.h> 198 199 #include <secmodel/secmodel.h> 200 201 extern struct proc proc0; 202 extern struct lwp lwp0; 203 extern struct cwdinfo cwdi0; 204 extern time_t rootfstime; 205 206 #ifndef curlwp 207 struct lwp *curlwp = &lwp0; 208 #endif 209 struct proc *initproc; 210 211 struct vnode *rootvp, *swapdev_vp; 212 int boothowto; 213 int cold = 1; /* still working on startup */ 214 struct timeval boottime; /* time at system startup - will only follow settime deltas */ 215 216 volatile int start_init_exec; /* semaphore for start_init() */ 217 218 static void check_console(struct lwp *l); 219 static void start_init(void *); 220 void main(void); 221 222 #if defined(__SSP__) || defined(__SSP_ALL__) 223 long __stack_chk_guard[8] = {0, 0, 0, 0, 0, 0, 0, 0}; 224 void __stack_chk_fail(void); 225 226 void 227 __stack_chk_fail(void) 228 { 229 panic("stack overflow detected; terminated"); 230 } 231 #endif 232 233 void __secmodel_none(void); 234 __weak_alias(secmodel_start,__secmodel_none); 235 void 236 __secmodel_none(void) 237 { 238 return; 239 } 240 241 /* 242 * System startup; initialize the world, create process 0, mount root 243 * filesystem, and fork to create init and pagedaemon. Most of the 244 * hard work is done in the lower-level initialization routines including 245 * startup(), which does memory initialization and autoconfiguration. 246 */ 247 void 248 main(void) 249 { 250 #ifdef __HAVE_TIMECOUNTER 251 struct timeval time; 252 #endif 253 struct lwp *l; 254 struct proc *p; 255 struct pdevinit *pdev; 256 int s, error; 257 extern struct pdevinit pdevinit[]; 258 #ifdef NVNODE_IMPLICIT 259 int usevnodes; 260 #endif 261 CPU_INFO_ITERATOR cii; 262 struct cpu_info *ci; 263 264 l = &lwp0; 265 #ifndef LWP0_CPU_INFO 266 l->l_cpu = curcpu(); 267 #endif 268 269 /* 270 * XXX This is a temporary check to be removed before 271 * NetBSD 5.0 is released. 272 */ 273 #if !defined(__i386__ ) && !defined(__x86_64__) 274 if (curlwp != l) { 275 printf("NOTICE: curlwp should be set before main()\n"); 276 DELAY(250000); 277 curlwp = l; 278 } 279 #endif 280 281 /* 282 * Attempt to find console and initialize 283 * in case of early panic or other messages. 284 */ 285 consinit(); 286 287 kernel_lock_init(); 288 289 uvm_init(); 290 291 #ifdef DEBUG 292 debug_init(); 293 #endif 294 295 kmem_init(); 296 297 /* Initialize the extent manager. */ 298 extent_init(); 299 300 /* Do machine-dependent initialization. */ 301 cpu_startup(); 302 303 /* Initialize callouts, part 1. */ 304 callout_startup(); 305 306 /* 307 * Initialize the kernel authorization subsystem and start the 308 * default security model, if any. We need to do this early 309 * enough so that subsystems relying on any of the aforementioned 310 * can work properly. Since the security model may dictate the 311 * credential inheritance policy, it is needed at least before 312 * any process is created, specifically proc0. 313 */ 314 kauth_init(); 315 secmodel_start(); 316 317 /* Initialize the buffer cache */ 318 bufinit(); 319 320 /* 321 * Initialize mbuf's. Do this now because we might attempt to 322 * allocate mbufs or mbuf clusters during autoconfiguration. 323 */ 324 mbinit(); 325 326 /* Initialize sockets. */ 327 soinit(); 328 329 /* 330 * The following things must be done before autoconfiguration. 331 */ 332 evcnt_init(); /* initialize event counters */ 333 #if NRND > 0 334 rnd_init(); /* initialize RNG */ 335 #endif 336 337 /* Initialize process and pgrp structures. */ 338 procinit(); 339 lwpinit(); 340 341 /* Initialize signal-related data structures. */ 342 signal_init(); 343 344 /* Create process 0 (the swapper). */ 345 proc0_init(); 346 347 /* Initialize the UID hash table. */ 348 uid_init(); 349 350 /* Charge root for one process. */ 351 (void)chgproccnt(0, 1); 352 353 /* Initialize the run queues, turnstiles and sleep queues. */ 354 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 355 sched_rqinit(); 356 turnstile_init(); 357 sleeptab_init(&sleeptab); 358 359 /* MI initialization of the boot cpu */ 360 error = mi_cpu_attach(curcpu()); 361 KASSERT(error == 0); 362 363 /* Initialize the sysctl subsystem. */ 364 sysctl_init(); 365 366 /* Initialize I/O statistics. */ 367 iostat_init(); 368 369 /* Initialize the log device. */ 370 loginit(); 371 372 /* Initialize the file systems. */ 373 #ifdef NVNODE_IMPLICIT 374 /* 375 * If maximum number of vnodes in namei vnode cache is not explicitly 376 * defined in kernel config, adjust the number such as we use roughly 377 * 1.0% of memory for vnode cache (but not less than NVNODE vnodes). 378 */ 379 usevnodes = (ptoa((unsigned)physmem) / 100) / sizeof(struct vnode); 380 if (usevnodes > desiredvnodes) 381 desiredvnodes = usevnodes; 382 #endif 383 vfsinit(); 384 385 /* Initialize fstrans. */ 386 fstrans_init(); 387 388 /* Initialize the file descriptor system. */ 389 filedesc_init(); 390 391 /* Initialize the select()/poll() system calls. */ 392 selsysinit(); 393 394 /* Initialize asynchronous I/O. */ 395 aio_sysinit(); 396 397 /* Initialize message queues. */ 398 mqueue_sysinit(); 399 400 /* Initialize the system monitor subsystems. */ 401 #if NSYSMON_TASKQ > 0 402 sysmon_task_queue_preinit(); 403 #endif 404 405 #if NSYSMON_ENVSYS > 0 406 sysmon_envsys_init(); 407 #endif 408 409 #if NSYSMON_POWER > 0 410 sysmon_power_init(); 411 #endif 412 413 #if NSYSMON_WDOG > 0 414 sysmon_wdog_init(); 415 #endif 416 417 #ifdef __HAVE_TIMECOUNTER 418 inittimecounter(); 419 ntp_init(); 420 #endif /* __HAVE_TIMECOUNTER */ 421 422 /* Initialize the device switch tables. */ 423 devsw_init(); 424 425 /* Initialize tty subsystem. */ 426 tty_init(); 427 ttyldisc_init(); 428 429 /* Initialize the disk wedge subsystem. */ 430 dkwedge_init(); 431 432 /* Configure the system hardware. This will enable interrupts. */ 433 configure(); 434 435 #if defined(__SSP__) || defined(__SSP_ALL__) 436 { 437 #ifdef DIAGNOSTIC 438 printf("Initializing SSP:"); 439 #endif 440 /* 441 * We initialize ssp here carefully: 442 * 1. after we got some entropy 443 * 2. without calling a function 444 */ 445 size_t i; 446 long guard[__arraycount(__stack_chk_guard)]; 447 448 arc4randbytes(guard, sizeof(guard)); 449 for (i = 0; i < __arraycount(guard); i++) 450 __stack_chk_guard[i] = guard[i]; 451 #ifdef DIAGNOSTIC 452 for (i = 0; i < __arraycount(guard); i++) 453 printf("%lx ", guard[i]); 454 printf("\n"); 455 #endif 456 } 457 #endif 458 ubc_init(); /* must be after autoconfig */ 459 460 /* Lock the kernel on behalf of proc0. */ 461 KERNEL_LOCK(1, l); 462 463 #ifdef SYSTRACE 464 systrace_init(); 465 #endif 466 467 #ifdef SYSVSHM 468 /* Initialize System V style shared memory. */ 469 shminit(); 470 #endif 471 472 #ifdef SYSVSEM 473 /* Initialize System V style semaphores. */ 474 seminit(); 475 #endif 476 477 #ifdef SYSVMSG 478 /* Initialize System V style message queues. */ 479 msginit(); 480 #endif 481 482 #ifdef P1003_1B_SEMAPHORE 483 /* Initialize posix semaphores */ 484 ksem_init(); 485 #endif 486 487 #if NVERIEXEC > 0 488 /* 489 * Initialise the Veriexec subsystem. 490 */ 491 veriexec_init(); 492 #endif /* NVERIEXEC > 0 */ 493 494 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) 495 pax_init(); 496 #endif /* PAX_MPROTECT || PAX_SEGVGUARD */ 497 498 /* Attach pseudo-devices. */ 499 for (pdev = pdevinit; pdev->pdev_attach != NULL; pdev++) 500 (*pdev->pdev_attach)(pdev->pdev_count); 501 502 #ifdef FAST_IPSEC 503 /* Attach network crypto subsystem */ 504 ipsec_attach(); 505 #endif 506 507 /* 508 * Initialize protocols. Block reception of incoming packets 509 * until everything is ready. 510 */ 511 s = splnet(); 512 ifinit(); 513 domaininit(); 514 if_attachdomain(); 515 splx(s); 516 517 #ifdef GPROF 518 /* Initialize kernel profiling. */ 519 kmstartup(); 520 #endif 521 522 /* Initialize system accouting. */ 523 acct_init(); 524 525 #ifndef PIPE_SOCKETPAIR 526 /* Initialize pipes. */ 527 pipe_init(); 528 #endif 529 530 /* Setup the scheduler */ 531 sched_init(); 532 533 #ifdef KTRACE 534 /* Initialize ktrace. */ 535 ktrinit(); 536 #endif 537 538 /* Initialize the UUID system calls. */ 539 uuid_init(); 540 541 /* 542 * Create process 1 (init(8)). We do this now, as Unix has 543 * historically had init be process 1, and changing this would 544 * probably upset a lot of people. 545 * 546 * Note that process 1 won't immediately exec init(8), but will 547 * wait for us to inform it that the root file system has been 548 * mounted. 549 */ 550 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 551 panic("fork init"); 552 553 /* 554 * Now that device driver threads have been created, wait for 555 * them to finish any deferred autoconfiguration. Note we don't 556 * need to lock this semaphore, since we haven't booted any 557 * secondary processors, yet. 558 */ 559 while (config_pending) 560 (void) tsleep(&config_pending, PWAIT, "cfpend", 0); 561 562 /* 563 * Finalize configuration now that all real devices have been 564 * found. This needs to be done before the root device is 565 * selected, since finalization may create the root device. 566 */ 567 config_finalize(); 568 569 /* 570 * Now that autoconfiguration has completed, we can determine 571 * the root and dump devices. 572 */ 573 cpu_rootconf(); 574 cpu_dumpconf(); 575 576 /* Mount the root file system. */ 577 do { 578 domountroothook(); 579 if ((error = vfs_mountroot())) { 580 printf("cannot mount root, error = %d\n", error); 581 boothowto |= RB_ASKNAME; 582 setroot(root_device, 583 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 584 } 585 } while (error != 0); 586 mountroothook_destroy(); 587 588 /* 589 * Initialise the time-of-day clock, passing the time recorded 590 * in the root filesystem (if any) for use by systems that 591 * don't have a non-volatile time-of-day device. 592 */ 593 inittodr(rootfstime); 594 595 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 596 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 597 598 /* 599 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 600 * reference it. 601 */ 602 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 603 if (error) 604 panic("cannot find root vnode, error=%d", error); 605 cwdi0.cwdi_cdir = rootvnode; 606 VREF(cwdi0.cwdi_cdir); 607 VOP_UNLOCK(rootvnode, 0); 608 cwdi0.cwdi_rdir = NULL; 609 610 /* 611 * Now that root is mounted, we can fixup initproc's CWD 612 * info. All other processes are kthreads, which merely 613 * share proc0's CWD info. 614 */ 615 initproc->p_cwdi->cwdi_cdir = rootvnode; 616 VREF(initproc->p_cwdi->cwdi_cdir); 617 initproc->p_cwdi->cwdi_rdir = NULL; 618 619 /* 620 * Now can look at time, having had a chance to verify the time 621 * from the file system. Reset l->l_rtime as it may have been 622 * munched in mi_switch() after the time got set. 623 */ 624 #ifdef __HAVE_TIMECOUNTER 625 getmicrotime(&time); 626 #else 627 mono_time = time; 628 #endif 629 boottime = time; 630 mutex_enter(&proclist_lock); 631 LIST_FOREACH(p, &allproc, p_list) { 632 KASSERT((p->p_flag & PK_MARKER) == 0); 633 mutex_enter(&p->p_smutex); 634 p->p_stats->p_start = time; 635 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 636 lwp_lock(l); 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 curlwp->l_stime = time; 644 645 for (CPU_INFO_FOREACH(cii, ci)) { 646 ci->ci_schedstate.spc_lastmod = time_second; 647 } 648 649 /* Create the pageout daemon kernel thread. */ 650 uvm_swap_init(); 651 if (kthread_create(PRI_PGDAEMON, 0, NULL, uvm_pageout, 652 NULL, NULL, "pgdaemon")) 653 panic("fork pagedaemon"); 654 655 /* Create the filesystem syncer kernel thread. */ 656 if (kthread_create(PRI_IOFLUSH, 0, NULL, sched_sync, NULL, NULL, "ioflush")) 657 panic("fork syncer"); 658 659 /* Create the aiodone daemon kernel thread. */ 660 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 661 uvm_aiodone_worker, NULL, PRI_VM, IPL_BIO, 0)) 662 panic("fork aiodoned"); 663 664 vmem_rehash_start(); 665 666 #if defined(MULTIPROCESSOR) 667 /* Boot the secondary processors. */ 668 cpu_boot_secondary_processors(); 669 #endif 670 671 /* Initialize exec structures */ 672 exec_init(1); 673 674 /* 675 * Okay, now we can let init(8) exec! It's off to userland! 676 */ 677 start_init_exec = 1; 678 wakeup(&start_init_exec); 679 680 /* The scheduler is an infinite loop. */ 681 uvm_scheduler(); 682 /* NOTREACHED */ 683 } 684 685 static void 686 check_console(struct lwp *l) 687 { 688 struct nameidata nd; 689 int error; 690 691 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console", l); 692 error = namei(&nd); 693 if (error == 0) 694 vrele(nd.ni_vp); 695 else if (error == ENOENT) 696 printf("warning: no /dev/console\n"); 697 else 698 printf("warning: lookup /dev/console: error %d\n", error); 699 } 700 701 /* 702 * List of paths to try when searching for "init". 703 */ 704 static const char * const initpaths[] = { 705 "/sbin/init", 706 "/sbin/oinit", 707 "/sbin/init.bak", 708 NULL, 709 }; 710 711 /* 712 * Start the initial user process; try exec'ing each pathname in "initpaths". 713 * The program is invoked with one argument containing the boot flags. 714 */ 715 static void 716 start_init(void *arg) 717 { 718 struct lwp *l = arg; 719 struct proc *p = l->l_proc; 720 vaddr_t addr; 721 struct sys_execve_args /* { 722 syscallarg(const char *) path; 723 syscallarg(char * const *) argp; 724 syscallarg(char * const *) envp; 725 } */ args; 726 int options, i, error; 727 register_t retval[2]; 728 char flags[4], *flagsp; 729 const char *path, *slash; 730 char *ucp, **uap, *arg0, *arg1 = NULL; 731 char ipath[129]; 732 int ipx, len; 733 734 /* 735 * Now in process 1. 736 */ 737 strncpy(p->p_comm, "init", MAXCOMLEN); 738 739 /* 740 * Wait for main() to tell us that it's safe to exec. 741 */ 742 while (start_init_exec == 0) 743 (void) tsleep(&start_init_exec, PWAIT, "initexec", 0); 744 745 /* 746 * This is not the right way to do this. We really should 747 * hand-craft a descriptor onto /dev/console to hand to init, 748 * but that's a _lot_ more work, and the benefit from this easy 749 * hack makes up for the "good is the enemy of the best" effect. 750 */ 751 check_console(l); 752 753 /* 754 * Need just enough stack to hold the faked-up "execve()" arguments. 755 */ 756 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 757 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 758 NULL, UVM_UNKNOWN_OFFSET, 0, 759 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 760 UVM_ADV_NORMAL, 761 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 762 panic("init: couldn't allocate argument space"); 763 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 764 765 ipx = 0; 766 while (1) { 767 if (boothowto & RB_ASKNAME) { 768 printf("init path"); 769 if (initpaths[ipx]) 770 printf(" (default %s)", initpaths[ipx]); 771 printf(": "); 772 len = cngetsn(ipath, sizeof(ipath)-1); 773 if (len == 0) { 774 if (initpaths[ipx]) 775 path = initpaths[ipx++]; 776 else 777 continue; 778 } else { 779 ipath[len] = '\0'; 780 path = ipath; 781 } 782 } else { 783 if ((path = initpaths[ipx++]) == NULL) 784 break; 785 } 786 787 ucp = (char *)USRSTACK; 788 789 /* 790 * Construct the boot flag argument. 791 */ 792 flagsp = flags; 793 *flagsp++ = '-'; 794 options = 0; 795 796 if (boothowto & RB_SINGLE) { 797 *flagsp++ = 's'; 798 options = 1; 799 } 800 #ifdef notyet 801 if (boothowto & RB_FASTBOOT) { 802 *flagsp++ = 'f'; 803 options = 1; 804 } 805 #endif 806 807 /* 808 * Move out the flags (arg 1), if necessary. 809 */ 810 if (options != 0) { 811 *flagsp++ = '\0'; 812 i = flagsp - flags; 813 #ifdef DEBUG 814 printf("init: copying out flags `%s' %d\n", flags, i); 815 #endif 816 arg1 = STACK_ALLOC(ucp, i); 817 ucp = STACK_MAX(arg1, i); 818 (void)copyout((void *)flags, arg1, i); 819 } 820 821 /* 822 * Move out the file name (also arg 0). 823 */ 824 i = strlen(path) + 1; 825 #ifdef DEBUG 826 printf("init: copying out path `%s' %d\n", path, i); 827 #else 828 if (boothowto & RB_ASKNAME || path != initpaths[0]) 829 printf("init: trying %s\n", path); 830 #endif 831 arg0 = STACK_ALLOC(ucp, i); 832 ucp = STACK_MAX(arg0, i); 833 (void)copyout(path, arg0, i); 834 835 /* 836 * Move out the arg pointers. 837 */ 838 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 839 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 840 SCARG(&args, path) = arg0; 841 SCARG(&args, argp) = uap; 842 SCARG(&args, envp) = NULL; 843 slash = strrchr(path, '/'); 844 if (slash) 845 (void)suword((void *)uap++, 846 (long)arg0 + (slash + 1 - path)); 847 else 848 (void)suword((void *)uap++, (long)arg0); 849 if (options != 0) 850 (void)suword((void *)uap++, (long)arg1); 851 (void)suword((void *)uap++, 0); /* terminator */ 852 853 /* 854 * Now try to exec the program. If can't for any reason 855 * other than it doesn't exist, complain. 856 */ 857 error = sys_execve(l, &args, retval); 858 if (error == 0 || error == EJUSTRETURN) { 859 KERNEL_UNLOCK_LAST(l); 860 return; 861 } 862 printf("exec %s: error %d\n", path, error); 863 } 864 printf("init: not found\n"); 865 panic("no init"); 866 } 867
Indexes created Sun Sep 21 18:09:52 GMT 2025