init_main.c revision 1.334
1 /* $NetBSD: init_main.c,v 1.334 2007/12/27 13:34:43 elad 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 1.334 2007/12/27 13:34:43 elad Exp $"); 75 76 #include "opt_ipsec.h" 77 #include "opt_ntp.h" 78 #include "opt_pipe.h" 79 #include "opt_posix.h" 80 #include "opt_syscall_debug.h" 81 #include "opt_sysv.h" 82 #include "opt_systrace.h" 83 #include "opt_fileassoc.h" 84 #include "opt_ktrace.h" 85 #include "opt_pax.h" 86 87 #include "rnd.h" 88 #include "sysmon_envsys.h" 89 #include "sysmon_power.h" 90 #include "sysmon_taskq.h" 91 #include "sysmon_wdog.h" 92 #include "veriexec.h" 93 94 #include <sys/param.h> 95 #include <sys/acct.h> 96 #include <sys/filedesc.h> 97 #include <sys/file.h> 98 #include <sys/errno.h> 99 #include <sys/callout.h> 100 #include <sys/cpu.h> 101 #include <sys/kernel.h> 102 #include <sys/kmem.h> 103 #include <sys/mount.h> 104 #include <sys/proc.h> 105 #include <sys/kthread.h> 106 #include <sys/resourcevar.h> 107 #include <sys/signalvar.h> 108 #include <sys/systm.h> 109 #include <sys/vnode.h> 110 #include <sys/fstrans.h> 111 #include <sys/tty.h> 112 #include <sys/conf.h> 113 #include <sys/disklabel.h> 114 #include <sys/buf.h> 115 #include <sys/device.h> 116 #include <sys/exec.h> 117 #include <sys/socketvar.h> 118 #include <sys/protosw.h> 119 #include <sys/reboot.h> 120 #include <sys/user.h> 121 #include <sys/sysctl.h> 122 #include <sys/event.h> 123 #include <sys/mbuf.h> 124 #include <sys/sched.h> 125 #include <sys/sleepq.h> 126 #include <sys/iostat.h> 127 #include <sys/vmem.h> 128 #include <sys/uuid.h> 129 #include <sys/extent.h> 130 #include <sys/disk.h> 131 #include <sys/mqueue.h> 132 #include <sys/msgbuf.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/kauth.h> 167 #include <net80211/ieee80211_netbsd.h> 168 169 #include <sys/syscall.h> 170 #include <sys/syscallargs.h> 171 172 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 173 #include <sys/pax.h> 174 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 175 176 #include <ufs/ufs/quota.h> 177 178 #include <miscfs/genfs/genfs.h> 179 #include <miscfs/syncfs/syncfs.h> 180 181 #include <sys/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 */ 214 215 volatile int start_init_exec; /* semaphore for start_init() */ 216 217 static void check_console(struct lwp *l); 218 static void start_init(void *); 219 void main(void); 220 221 #if defined(__SSP__) || defined(__SSP_ALL__) 222 long __stack_chk_guard[8] = {0, 0, 0, 0, 0, 0, 0, 0}; 223 void __stack_chk_fail(void); 224 225 void 226 __stack_chk_fail(void) 227 { 228 panic("stack overflow detected; terminated"); 229 } 230 #endif 231 232 void __secmodel_none(void); 233 __weak_alias(secmodel_start,__secmodel_none); 234 void 235 __secmodel_none(void) 236 { 237 return; 238 } 239 240 /* 241 * System startup; initialize the world, create process 0, mount root 242 * filesystem, and fork to create init and pagedaemon. Most of the 243 * hard work is done in the lower-level initialization routines including 244 * startup(), which does memory initialization and autoconfiguration. 245 */ 246 void 247 main(void) 248 { 249 #ifdef __HAVE_TIMECOUNTER 250 struct timeval time; 251 #endif 252 struct lwp *l; 253 struct proc *p; 254 struct pdevinit *pdev; 255 int s, error; 256 extern struct pdevinit pdevinit[]; 257 #ifdef NVNODE_IMPLICIT 258 int usevnodes; 259 #endif 260 CPU_INFO_ITERATOR cii; 261 struct cpu_info *ci; 262 263 l = &lwp0; 264 #ifndef LWP0_CPU_INFO 265 l->l_cpu = curcpu(); 266 #endif 267 268 /* 269 * XXX This is a temporary check to be removed before 270 * NetBSD 5.0 is released. 271 */ 272 #if !defined(__i386__ ) && !defined(__x86_64__) 273 if (curlwp != l) { 274 printf("NOTICE: curlwp should be set before main()\n"); 275 DELAY(250000); 276 curlwp = l; 277 } 278 #endif 279 280 /* 281 * Attempt to find console and initialize 282 * in case of early panic or other messages. 283 */ 284 consinit(); 285 286 kernel_lock_init(); 287 288 uvm_init(); 289 290 kmem_init(); 291 292 /* Initialize the extent manager. */ 293 extent_init(); 294 295 /* Do machine-dependent initialization. */ 296 cpu_startup(); 297 298 /* Initialize callouts, part 1. */ 299 callout_startup(); 300 301 /* 302 * Initialize the kernel authorization subsystem and start the 303 * default security model, if any. We need to do this early 304 * enough so that subsystems relying on any of the aforementioned 305 * can work properly. Since the security model may dictate the 306 * credential inheritance policy, it is needed at least before 307 * any process is created, specifically proc0. 308 */ 309 kauth_init(); 310 secmodel_start(); 311 312 /* Initialize the buffer cache */ 313 bufinit(); 314 315 /* 316 * Initialize mbuf's. Do this now because we might attempt to 317 * allocate mbufs or mbuf clusters during autoconfiguration. 318 */ 319 mbinit(); 320 321 /* Initialize sockets. */ 322 soinit(); 323 324 /* 325 * The following things must be done before autoconfiguration. 326 */ 327 evcnt_init(); /* initialize event counters */ 328 #if NRND > 0 329 rnd_init(); /* initialize RNG */ 330 #endif 331 332 /* Initialize process and pgrp structures. */ 333 procinit(); 334 lwpinit(); 335 336 /* Initialize signal-related data structures. */ 337 signal_init(); 338 339 /* Initialize resource management. */ 340 resource_init(); 341 342 /* Create process 0 (the swapper). */ 343 proc0_init(); 344 345 /* Initialize the UID hash table. */ 346 uid_init(); 347 348 /* Charge root for one process. */ 349 (void)chgproccnt(0, 1); 350 351 /* Initialize timekeeping. */ 352 time_init(); 353 354 /* Initialize the run queues, turnstiles and sleep queues. */ 355 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 356 sched_rqinit(); 357 turnstile_init(); 358 sleeptab_init(&sleeptab); 359 360 /* MI initialization of the boot cpu */ 361 error = mi_cpu_attach(curcpu()); 362 KASSERT(error == 0); 363 364 /* Initialize the sysctl subsystem. */ 365 sysctl_init(); 366 367 /* Initialize I/O statistics. */ 368 iostat_init(); 369 370 /* Initialize the log device. */ 371 loginit(); 372 373 /* Initialize the file systems. */ 374 #ifdef NVNODE_IMPLICIT 375 /* 376 * If maximum number of vnodes in namei vnode cache is not explicitly 377 * defined in kernel config, adjust the number such as we use roughly 378 * 1.0% of memory for vnode cache (but not less than NVNODE vnodes). 379 */ 380 usevnodes = (ptoa((unsigned)physmem) / 100) / sizeof(struct vnode); 381 if (usevnodes > desiredvnodes) 382 desiredvnodes = usevnodes; 383 #endif 384 vfsinit(); 385 386 /* Initialize fstrans. */ 387 fstrans_init(); 388 389 /* Initialize the file descriptor system. */ 390 filedesc_init(); 391 392 /* Initialize the select()/poll() system calls. */ 393 selsysinit(); 394 395 /* Initialize asynchronous I/O. */ 396 aio_sysinit(); 397 398 /* Initialize message queues. */ 399 mqueue_sysinit(); 400 401 /* Initialize the system monitor subsystems. */ 402 #if NSYSMON_TASKQ > 0 403 sysmon_task_queue_preinit(); 404 #endif 405 406 #if NSYSMON_ENVSYS > 0 407 sysmon_envsys_init(); 408 #endif 409 410 #if NSYSMON_POWER > 0 411 sysmon_power_init(); 412 #endif 413 414 #if NSYSMON_WDOG > 0 415 sysmon_wdog_init(); 416 #endif 417 418 #ifdef __HAVE_TIMECOUNTER 419 inittimecounter(); 420 ntp_init(); 421 #endif /* __HAVE_TIMECOUNTER */ 422 423 /* Initialize the device switch tables. */ 424 devsw_init(); 425 426 /* Initialize tty subsystem. */ 427 tty_init(); 428 ttyldisc_init(); 429 430 /* Initialize the disk wedge subsystem. */ 431 dkwedge_init(); 432 433 /* Configure the system hardware. This will enable interrupts. */ 434 configure(); 435 436 #if defined(__SSP__) || defined(__SSP_ALL__) 437 { 438 #ifdef DIAGNOSTIC 439 printf("Initializing SSP:"); 440 #endif 441 /* 442 * We initialize ssp here carefully: 443 * 1. after we got some entropy 444 * 2. without calling a function 445 */ 446 size_t i; 447 long guard[__arraycount(__stack_chk_guard)]; 448 449 arc4randbytes(guard, sizeof(guard)); 450 for (i = 0; i < __arraycount(guard); i++) 451 __stack_chk_guard[i] = guard[i]; 452 #ifdef DIAGNOSTIC 453 for (i = 0; i < __arraycount(guard); i++) 454 printf("%lx ", guard[i]); 455 printf("\n"); 456 #endif 457 } 458 #endif 459 ubc_init(); /* must be after autoconfig */ 460 461 /* Lock the kernel on behalf of proc0. */ 462 KERNEL_LOCK(1, l); 463 464 #ifdef SYSTRACE 465 systrace_init(); 466 #endif 467 468 #ifdef SYSVSHM 469 /* Initialize System V style shared memory. */ 470 shminit(); 471 #endif 472 473 #ifdef SYSVSEM 474 /* Initialize System V style semaphores. */ 475 seminit(); 476 #endif 477 478 #ifdef SYSVMSG 479 /* Initialize System V style message queues. */ 480 msginit(); 481 #endif 482 483 #ifdef P1003_1B_SEMAPHORE 484 /* Initialize posix semaphores */ 485 ksem_init(); 486 #endif 487 488 #if NVERIEXEC > 0 489 /* 490 * Initialise the Veriexec subsystem. 491 */ 492 veriexec_init(); 493 #endif /* NVERIEXEC > 0 */ 494 495 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 496 pax_init(); 497 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 498 499 /* Attach pseudo-devices. */ 500 for (pdev = pdevinit; pdev->pdev_attach != NULL; pdev++) 501 (*pdev->pdev_attach)(pdev->pdev_count); 502 503 #ifdef FAST_IPSEC 504 /* Attach network crypto subsystem */ 505 ipsec_attach(); 506 #endif 507 508 /* 509 * Initialize protocols. Block reception of incoming packets 510 * until everything is ready. 511 */ 512 s = splnet(); 513 ifinit(); 514 domaininit(); 515 if_attachdomain(); 516 splx(s); 517 518 #ifdef GPROF 519 /* Initialize kernel profiling. */ 520 kmstartup(); 521 #endif 522 523 /* Initialize system accouting. */ 524 acct_init(); 525 526 #ifndef PIPE_SOCKETPAIR 527 /* Initialize pipes. */ 528 pipe_init(); 529 #endif 530 531 /* Setup the scheduler */ 532 sched_init(); 533 534 #ifdef KTRACE 535 /* Initialize ktrace. */ 536 ktrinit(); 537 #endif 538 539 /* Initialize the UUID system calls. */ 540 uuid_init(); 541 542 /* 543 * Create process 1 (init(8)). We do this now, as Unix has 544 * historically had init be process 1, and changing this would 545 * probably upset a lot of people. 546 * 547 * Note that process 1 won't immediately exec init(8), but will 548 * wait for us to inform it that the root file system has been 549 * mounted. 550 */ 551 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 552 panic("fork init"); 553 554 /* 555 * Now that device driver threads have been created, wait for 556 * them to finish any deferred autoconfiguration. 557 */ 558 while (config_pending) 559 (void) tsleep(&config_pending, PWAIT, "cfpend", hz); 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 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 637 lwp_unlock(l); 638 } 639 mutex_exit(&p->p_smutex); 640 } 641 mutex_exit(&proclist_lock); 642 binuptime(&curlwp->l_stime); 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(PRI_PGDAEMON, 0, NULL, uvm_pageout, 651 NULL, NULL, "pgdaemon")) 652 panic("fork pagedaemon"); 653 654 /* Create the filesystem syncer kernel thread. */ 655 if (kthread_create(PRI_IOFLUSH, 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, PRI_VM, IPL_BIO, 0)) 661 panic("fork aiodoned"); 662 663 vmem_rehash_start(); 664 665 /* Initialize exec structures */ 666 exec_init(1); 667 668 /* 669 * Okay, now we can let init(8) exec! It's off to userland! 670 */ 671 start_init_exec = 1; 672 wakeup(&start_init_exec); 673 674 /* The scheduler is an infinite loop. */ 675 uvm_scheduler(); 676 /* NOTREACHED */ 677 } 678 679 static void 680 check_console(struct lwp *l) 681 { 682 struct nameidata nd; 683 int error; 684 685 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console"); 686 error = namei(&nd); 687 if (error == 0) 688 vrele(nd.ni_vp); 689 else if (error == ENOENT) 690 printf("warning: no /dev/console\n"); 691 else 692 printf("warning: lookup /dev/console: error %d\n", error); 693 } 694 695 /* 696 * List of paths to try when searching for "init". 697 */ 698 static const char * const initpaths[] = { 699 "/sbin/init", 700 "/sbin/oinit", 701 "/sbin/init.bak", 702 NULL, 703 }; 704 705 /* 706 * Start the initial user process; try exec'ing each pathname in "initpaths". 707 * The program is invoked with one argument containing the boot flags. 708 */ 709 static void 710 start_init(void *arg) 711 { 712 struct lwp *l = arg; 713 struct proc *p = l->l_proc; 714 vaddr_t addr; 715 struct sys_execve_args /* { 716 syscallarg(const char *) path; 717 syscallarg(char * const *) argp; 718 syscallarg(char * const *) envp; 719 } */ args; 720 int options, i, error; 721 register_t retval[2]; 722 char flags[4], *flagsp; 723 const char *path, *slash; 724 char *ucp, **uap, *arg0, *arg1 = NULL; 725 char ipath[129]; 726 int ipx, len; 727 728 /* 729 * Now in process 1. 730 */ 731 strncpy(p->p_comm, "init", MAXCOMLEN); 732 733 /* 734 * Wait for main() to tell us that it's safe to exec. 735 */ 736 while (start_init_exec == 0) 737 (void) tsleep(&start_init_exec, PWAIT, "initexec", 0); 738 739 /* 740 * This is not the right way to do this. We really should 741 * hand-craft a descriptor onto /dev/console to hand to init, 742 * but that's a _lot_ more work, and the benefit from this easy 743 * hack makes up for the "good is the enemy of the best" effect. 744 */ 745 check_console(l); 746 747 /* 748 * Need just enough stack to hold the faked-up "execve()" arguments. 749 */ 750 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 751 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 752 NULL, UVM_UNKNOWN_OFFSET, 0, 753 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 754 UVM_ADV_NORMAL, 755 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 756 panic("init: couldn't allocate argument space"); 757 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 758 759 ipx = 0; 760 while (1) { 761 if (boothowto & RB_ASKNAME) { 762 printf("init path"); 763 if (initpaths[ipx]) 764 printf(" (default %s)", initpaths[ipx]); 765 printf(": "); 766 len = cngetsn(ipath, sizeof(ipath)-1); 767 if (len == 0) { 768 if (initpaths[ipx]) 769 path = initpaths[ipx++]; 770 else 771 continue; 772 } else { 773 ipath[len] = '\0'; 774 path = ipath; 775 } 776 } else { 777 if ((path = initpaths[ipx++]) == NULL) 778 break; 779 } 780 781 ucp = (char *)USRSTACK; 782 783 /* 784 * Construct the boot flag argument. 785 */ 786 flagsp = flags; 787 *flagsp++ = '-'; 788 options = 0; 789 790 if (boothowto & RB_SINGLE) { 791 *flagsp++ = 's'; 792 options = 1; 793 } 794 #ifdef notyet 795 if (boothowto & RB_FASTBOOT) { 796 *flagsp++ = 'f'; 797 options = 1; 798 } 799 #endif 800 801 /* 802 * Move out the flags (arg 1), if necessary. 803 */ 804 if (options != 0) { 805 *flagsp++ = '\0'; 806 i = flagsp - flags; 807 #ifdef DEBUG 808 printf("init: copying out flags `%s' %d\n", flags, i); 809 #endif 810 arg1 = STACK_ALLOC(ucp, i); 811 ucp = STACK_MAX(arg1, i); 812 (void)copyout((void *)flags, arg1, i); 813 } 814 815 /* 816 * Move out the file name (also arg 0). 817 */ 818 i = strlen(path) + 1; 819 #ifdef DEBUG 820 printf("init: copying out path `%s' %d\n", path, i); 821 #else 822 if (boothowto & RB_ASKNAME || path != initpaths[0]) 823 printf("init: trying %s\n", path); 824 #endif 825 arg0 = STACK_ALLOC(ucp, i); 826 ucp = STACK_MAX(arg0, i); 827 (void)copyout(path, arg0, i); 828 829 /* 830 * Move out the arg pointers. 831 */ 832 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 833 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 834 SCARG(&args, path) = arg0; 835 SCARG(&args, argp) = uap; 836 SCARG(&args, envp) = NULL; 837 slash = strrchr(path, '/'); 838 if (slash) 839 (void)suword((void *)uap++, 840 (long)arg0 + (slash + 1 - path)); 841 else 842 (void)suword((void *)uap++, (long)arg0); 843 if (options != 0) 844 (void)suword((void *)uap++, (long)arg1); 845 (void)suword((void *)uap++, 0); /* terminator */ 846 847 /* 848 * Now try to exec the program. If can't for any reason 849 * other than it doesn't exist, complain. 850 */ 851 error = sys_execve(l, &args, retval); 852 if (error == 0 || error == EJUSTRETURN) { 853 KERNEL_UNLOCK_LAST(l); 854 return; 855 } 856 printf("exec %s: error %d\n", path, error); 857 } 858 printf("init: not found\n"); 859 panic("no init"); 860 } 861
Indexes created Mon Sep 22 05:09:51 GMT 2025