init_main.c revision 1.319
1 /* $NetBSD: init_main.c,v 1.319 2007/10/08 15:12:06 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 1.319 2007/10/08 15:12:06 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 #include <miscfs/genfs/genfs.h> 179 #include <miscfs/syncfs/syncfs.h> 180 181 #include <machine/cpu.h> 182 183 #include <uvm/uvm.h> 184 185 #if NSYSMON_TASKQ > 0 186 #include <dev/sysmon/sysmon_taskq.h> 187 #endif 188 189 #include <dev/cons.h> 190 191 #if NSYSMON_ENVSYS > 0 || NSYSMON_POWER > 0 || NSYSMON_WDOG > 0 192 #include <dev/sysmon/sysmonvar.h> 193 #endif 194 195 #include <net/if.h> 196 #include <net/raw_cb.h> 197 198 #include <secmodel/secmodel.h> 199 200 extern struct proc proc0; 201 extern struct lwp lwp0; 202 extern struct cwdinfo cwdi0; 203 extern time_t rootfstime; 204 205 #ifndef curlwp 206 struct lwp *curlwp = &lwp0; 207 #endif 208 struct proc *initproc; 209 210 struct vnode *rootvp, *swapdev_vp; 211 int boothowto; 212 int cold = 1; /* still working on startup */ 213 struct timeval boottime; /* time at system startup - will only follow settime deltas */ 214 int ncpu = 0; /* number of CPUs configured, assume 1 */ 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 l->l_cpu = curcpu(); 266 l->l_proc = &proc0; 267 l->l_lid = 1; 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 file systems. */ 370 #ifdef NVNODE_IMPLICIT 371 /* 372 * If maximum number of vnodes in namei vnode cache is not explicitly 373 * defined in kernel config, adjust the number such as we use roughly 374 * 1.0% of memory for vnode cache (but not less than NVNODE vnodes). 375 */ 376 usevnodes = (ptoa((unsigned)physmem) / 100) / sizeof(struct vnode); 377 if (usevnodes > desiredvnodes) 378 desiredvnodes = usevnodes; 379 #endif 380 vfsinit(); 381 382 /* Initialize fstrans. */ 383 fstrans_init(); 384 385 /* Initialize the file descriptor system. */ 386 filedesc_init(); 387 388 /* Initialize the select()/poll() system calls. */ 389 selsysinit(); 390 391 /* Initialize asynchronous I/O. */ 392 aio_sysinit(); 393 394 /* Initialize message queues. */ 395 mqueue_sysinit(); 396 397 /* Initialize the system monitor subsystems. */ 398 #if NSYSMON_TASKQ > 0 399 sysmon_task_queue_preinit(); 400 #endif 401 402 #if NSYSMON_ENVSYS > 0 403 sysmon_envsys_init(); 404 #endif 405 406 #if NSYSMON_POWER > 0 407 sysmon_power_init(); 408 #endif 409 410 #if NSYSMON_WDOG > 0 411 sysmon_wdog_init(); 412 #endif 413 414 #ifdef __HAVE_TIMECOUNTER 415 inittimecounter(); 416 ntp_init(); 417 #endif /* __HAVE_TIMECOUNTER */ 418 419 /* Initialize the device switch tables. */ 420 devsw_init(); 421 422 /* Initialize the disk wedge subsystem. */ 423 dkwedge_init(); 424 425 /* Configure the system hardware. This will enable interrupts. */ 426 configure(); 427 428 #if defined(__SSP__) || defined(__SSP_ALL__) 429 { 430 #ifdef DIAGNOSTIC 431 printf("Initializing SSP:"); 432 #endif 433 /* 434 * We initialize ssp here carefully: 435 * 1. after we got some entropy 436 * 2. without calling a function 437 */ 438 size_t i; 439 long guard[__arraycount(__stack_chk_guard)]; 440 441 arc4randbytes(guard, sizeof(guard)); 442 for (i = 0; i < __arraycount(guard); i++) 443 __stack_chk_guard[i] = guard[i]; 444 #ifdef DIAGNOSTIC 445 for (i = 0; i < __arraycount(guard); i++) 446 printf("%lx ", guard[i]); 447 printf("\n"); 448 #endif 449 } 450 #endif 451 ubc_init(); /* must be after autoconfig */ 452 453 /* Lock the kernel on behalf of proc0. */ 454 KERNEL_LOCK(1, l); 455 456 #ifdef SYSTRACE 457 systrace_init(); 458 #endif 459 460 #ifdef SYSVSHM 461 /* Initialize System V style shared memory. */ 462 shminit(); 463 #endif 464 465 #ifdef SYSVSEM 466 /* Initialize System V style semaphores. */ 467 seminit(); 468 #endif 469 470 #ifdef SYSVMSG 471 /* Initialize System V style message queues. */ 472 msginit(); 473 #endif 474 475 #ifdef P1003_1B_SEMAPHORE 476 /* Initialize posix semaphores */ 477 ksem_init(); 478 #endif 479 480 #if NVERIEXEC > 0 481 /* 482 * Initialise the Veriexec subsystem. 483 */ 484 veriexec_init(); 485 #endif /* NVERIEXEC > 0 */ 486 487 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) 488 pax_init(); 489 #endif /* PAX_MPROTECT || PAX_SEGVGUARD */ 490 491 /* Attach pseudo-devices. */ 492 for (pdev = pdevinit; pdev->pdev_attach != NULL; pdev++) 493 (*pdev->pdev_attach)(pdev->pdev_count); 494 495 #ifdef FAST_IPSEC 496 /* Attach network crypto subsystem */ 497 ipsec_attach(); 498 #endif 499 500 /* 501 * Initialize protocols. Block reception of incoming packets 502 * until everything is ready. 503 */ 504 s = splnet(); 505 ifinit(); 506 domaininit(); 507 if_attachdomain(); 508 splx(s); 509 510 #ifdef GPROF 511 /* Initialize kernel profiling. */ 512 kmstartup(); 513 #endif 514 515 /* Initialize system accouting. */ 516 acct_init(); 517 518 #ifndef PIPE_SOCKETPAIR 519 /* Initialize pipes. */ 520 pipe_init(); 521 #endif 522 523 /* Setup the scheduler */ 524 sched_init(); 525 526 #ifdef KTRACE 527 /* Initialize ktrace. */ 528 ktrinit(); 529 #endif 530 531 /* Initialize the UUID system calls. */ 532 uuid_init(); 533 534 /* 535 * Create process 1 (init(8)). We do this now, as Unix has 536 * historically had init be process 1, and changing this would 537 * probably upset a lot of people. 538 * 539 * Note that process 1 won't immediately exec init(8), but will 540 * wait for us to inform it that the root file system has been 541 * mounted. 542 */ 543 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 544 panic("fork init"); 545 546 /* 547 * Now that device driver threads have been created, wait for 548 * them to finish any deferred autoconfiguration. Note we don't 549 * need to lock this semaphore, since we haven't booted any 550 * secondary processors, yet. 551 */ 552 while (config_pending) 553 (void) tsleep(&config_pending, PWAIT, "cfpend", 0); 554 555 /* 556 * Finalize configuration now that all real devices have been 557 * found. This needs to be done before the root device is 558 * selected, since finalization may create the root device. 559 */ 560 config_finalize(); 561 562 /* 563 * Now that autoconfiguration has completed, we can determine 564 * the root and dump devices. 565 */ 566 cpu_rootconf(); 567 cpu_dumpconf(); 568 569 /* Mount the root file system. */ 570 do { 571 domountroothook(); 572 if ((error = vfs_mountroot())) { 573 printf("cannot mount root, error = %d\n", error); 574 boothowto |= RB_ASKNAME; 575 setroot(root_device, 576 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 577 } 578 } while (error != 0); 579 mountroothook_destroy(); 580 581 /* 582 * Initialise the time-of-day clock, passing the time recorded 583 * in the root filesystem (if any) for use by systems that 584 * don't have a non-volatile time-of-day device. 585 */ 586 inittodr(rootfstime); 587 588 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 589 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 590 591 /* 592 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 593 * reference it. 594 */ 595 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 596 if (error) 597 panic("cannot find root vnode, error=%d", error); 598 cwdi0.cwdi_cdir = rootvnode; 599 VREF(cwdi0.cwdi_cdir); 600 VOP_UNLOCK(rootvnode, 0); 601 cwdi0.cwdi_rdir = NULL; 602 603 /* 604 * Now that root is mounted, we can fixup initproc's CWD 605 * info. All other processes are kthreads, which merely 606 * share proc0's CWD info. 607 */ 608 initproc->p_cwdi->cwdi_cdir = rootvnode; 609 VREF(initproc->p_cwdi->cwdi_cdir); 610 initproc->p_cwdi->cwdi_rdir = NULL; 611 612 /* 613 * Now can look at time, having had a chance to verify the time 614 * from the file system. Reset l->l_rtime as it may have been 615 * munched in mi_switch() after the time got set. 616 */ 617 #ifdef __HAVE_TIMECOUNTER 618 getmicrotime(&time); 619 #else 620 mono_time = time; 621 #endif 622 boottime = time; 623 mutex_enter(&proclist_lock); 624 LIST_FOREACH(p, &allproc, p_list) { 625 KASSERT((p->p_flag & PK_MARKER) == 0); 626 mutex_enter(&p->p_smutex); 627 p->p_stats->p_start = time; 628 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 629 lwp_lock(l); 630 l->l_cpu->ci_schedstate.spc_runtime = time; 631 l->l_rtime.tv_sec = l->l_rtime.tv_usec = 0; 632 lwp_unlock(l); 633 } 634 mutex_exit(&p->p_smutex); 635 } 636 mutex_exit(&proclist_lock); 637 638 for (CPU_INFO_FOREACH(cii, ci)) { 639 ci->ci_schedstate.spc_lastmod = time_second; 640 } 641 642 /* Create the pageout daemon kernel thread. */ 643 uvm_swap_init(); 644 if (kthread_create(PVM, 0, NULL, uvm_pageout, 645 NULL, NULL, "pgdaemon")) 646 panic("fork pagedaemon"); 647 648 /* Create the filesystem syncer kernel thread. */ 649 if (kthread_create(PINOD, 0, NULL, sched_sync, NULL, NULL, "ioflush")) 650 panic("fork syncer"); 651 652 /* Create the aiodone daemon kernel thread. */ 653 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 654 uvm_aiodone_worker, NULL, PVM, IPL_BIO, 0)) 655 panic("fork aiodoned"); 656 657 vmem_rehash_start(); 658 659 #if defined(MULTIPROCESSOR) 660 /* Boot the secondary processors. */ 661 cpu_boot_secondary_processors(); 662 #endif 663 664 /* Initialize exec structures */ 665 exec_init(1); 666 667 /* 668 * Okay, now we can let init(8) exec! It's off to userland! 669 */ 670 start_init_exec = 1; 671 wakeup(&start_init_exec); 672 673 /* The scheduler is an infinite loop. */ 674 uvm_scheduler(); 675 /* NOTREACHED */ 676 } 677 678 static void 679 check_console(struct lwp *l) 680 { 681 struct nameidata nd; 682 int error; 683 684 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console", l); 685 error = namei(&nd); 686 if (error == 0) 687 vrele(nd.ni_vp); 688 else if (error == ENOENT) 689 printf("warning: no /dev/console\n"); 690 else 691 printf("warning: lookup /dev/console: error %d\n", error); 692 } 693 694 /* 695 * List of paths to try when searching for "init". 696 */ 697 static const char *initpaths[] = { 698 "/sbin/init", 699 "/sbin/oinit", 700 "/sbin/init.bak", 701 NULL, 702 }; 703 704 /* 705 * Start the initial user process; try exec'ing each pathname in "initpaths". 706 * The program is invoked with one argument containing the boot flags. 707 */ 708 static void 709 start_init(void *arg) 710 { 711 struct lwp *l = arg; 712 struct proc *p = l->l_proc; 713 vaddr_t addr; 714 struct sys_execve_args /* { 715 syscallarg(const char *) path; 716 syscallarg(char * const *) argp; 717 syscallarg(char * const *) envp; 718 } */ args; 719 int options, i, error; 720 register_t retval[2]; 721 char flags[4], *flagsp; 722 const char *path, *slash; 723 char *ucp, **uap, *arg0, *arg1 = NULL; 724 char ipath[129]; 725 int ipx, len; 726 727 /* 728 * Now in process 1. 729 */ 730 strncpy(p->p_comm, "init", MAXCOMLEN); 731 732 /* 733 * Wait for main() to tell us that it's safe to exec. 734 */ 735 while (start_init_exec == 0) 736 (void) tsleep(&start_init_exec, PWAIT, "initexec", 0); 737 738 /* 739 * This is not the right way to do this. We really should 740 * hand-craft a descriptor onto /dev/console to hand to init, 741 * but that's a _lot_ more work, and the benefit from this easy 742 * hack makes up for the "good is the enemy of the best" effect. 743 */ 744 check_console(l); 745 746 /* 747 * Need just enough stack to hold the faked-up "execve()" arguments. 748 */ 749 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 750 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 751 NULL, UVM_UNKNOWN_OFFSET, 0, 752 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 753 UVM_ADV_NORMAL, 754 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 755 panic("init: couldn't allocate argument space"); 756 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 757 758 ipx = 0; 759 while (1) { 760 if (boothowto & RB_ASKNAME) { 761 printf("init path"); 762 if (initpaths[ipx]) 763 printf(" (default %s)", initpaths[ipx]); 764 printf(": "); 765 len = cngetsn(ipath, sizeof(ipath)-1); 766 if (len == 0) { 767 if (initpaths[ipx]) 768 path = initpaths[ipx++]; 769 else 770 continue; 771 } else { 772 ipath[len] = '\0'; 773 path = ipath; 774 } 775 } else { 776 if ((path = initpaths[ipx++]) == NULL) 777 break; 778 } 779 780 ucp = (char *)USRSTACK; 781 782 /* 783 * Construct the boot flag argument. 784 */ 785 flagsp = flags; 786 *flagsp++ = '-'; 787 options = 0; 788 789 if (boothowto & RB_SINGLE) { 790 *flagsp++ = 's'; 791 options = 1; 792 } 793 #ifdef notyet 794 if (boothowto & RB_FASTBOOT) { 795 *flagsp++ = 'f'; 796 options = 1; 797 } 798 #endif 799 800 /* 801 * Move out the flags (arg 1), if necessary. 802 */ 803 if (options != 0) { 804 *flagsp++ = '\0'; 805 i = flagsp - flags; 806 #ifdef DEBUG 807 printf("init: copying out flags `%s' %d\n", flags, i); 808 #endif 809 arg1 = STACK_ALLOC(ucp, i); 810 ucp = STACK_MAX(arg1, i); 811 (void)copyout((void *)flags, arg1, i); 812 } 813 814 /* 815 * Move out the file name (also arg 0). 816 */ 817 i = strlen(path) + 1; 818 #ifdef DEBUG 819 printf("init: copying out path `%s' %d\n", path, i); 820 #else 821 if (boothowto & RB_ASKNAME || path != initpaths[0]) 822 printf("init: trying %s\n", path); 823 #endif 824 arg0 = STACK_ALLOC(ucp, i); 825 ucp = STACK_MAX(arg0, i); 826 (void)copyout(path, arg0, i); 827 828 /* 829 * Move out the arg pointers. 830 */ 831 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 832 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 833 SCARG(&args, path) = arg0; 834 SCARG(&args, argp) = uap; 835 SCARG(&args, envp) = NULL; 836 slash = strrchr(path, '/'); 837 if (slash) 838 (void)suword((void *)uap++, 839 (long)arg0 + (slash + 1 - path)); 840 else 841 (void)suword((void *)uap++, (long)arg0); 842 if (options != 0) 843 (void)suword((void *)uap++, (long)arg1); 844 (void)suword((void *)uap++, 0); /* terminator */ 845 846 /* 847 * Now try to exec the program. If can't for any reason 848 * other than it doesn't exist, complain. 849 */ 850 error = sys_execve(l, &args, retval); 851 if (error == 0 || error == EJUSTRETURN) { 852 KERNEL_UNLOCK_LAST(l); 853 return; 854 } 855 printf("exec %s: error %d\n", path, error); 856 } 857 printf("init: not found\n"); 858 panic("no init"); 859 } 860
Indexes created Sat Sep 20 22:09:52 GMT 2025