init_main.c revision 1.315
1 /* $NetBSD: init_main.c,v 1.315 2007/09/24 00:00:52 martin 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 1.315 2007/09/24 00:00:52 martin 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 /* 270 * Initialize the current LWP pointer (curlwp) before 271 * any possible traps/probes to simplify trap processing. 272 */ 273 l = &lwp0; 274 curlwp = l; 275 l->l_cpu = curcpu(); 276 l->l_proc = &proc0; 277 l->l_lid = 1; 278 279 /* 280 * Attempt to find console and initialize 281 * in case of early panic or other messages. 282 */ 283 consinit(); 284 285 KERNEL_LOCK_INIT(); 286 287 uvm_init(); 288 289 #ifdef DEBUG 290 debug_init(); 291 #endif 292 293 kmem_init(); 294 295 /* Initialize the extent manager. */ 296 extent_init(); 297 298 /* Do machine-dependent initialization. */ 299 cpu_startup(); 300 301 /* Initialize callouts, part 1. */ 302 callout_startup(); 303 304 /* 305 * Initialize the kernel authorization subsystem and start the 306 * default security model, if any. We need to do this early 307 * enough so that subsystems relying on any of the aforementioned 308 * can work properly. Since the security model may dictate the 309 * credential inheritance policy, it is needed at least before 310 * any process is created, specifically proc0. 311 */ 312 kauth_init(); 313 secmodel_start(); 314 315 /* Initialize the buffer cache */ 316 bufinit(); 317 318 /* 319 * Initialize mbuf's. Do this now because we might attempt to 320 * allocate mbufs or mbuf clusters during autoconfiguration. 321 */ 322 mbinit(); 323 324 /* Initialize sockets. */ 325 soinit(); 326 327 /* 328 * The following things must be done before autoconfiguration. 329 */ 330 evcnt_init(); /* initialize event counters */ 331 #if NRND > 0 332 rnd_init(); /* initialize RNG */ 333 #endif 334 335 /* Initialize process and pgrp structures. */ 336 procinit(); 337 lwpinit(); 338 339 /* Initialize signal-related data structures. */ 340 signal_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 the run queues, turnstiles and sleep queues. */ 352 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 353 sched_rqinit(); 354 turnstile_init(); 355 sleeptab_init(&sleeptab); 356 357 /* MI initialization of the boot cpu */ 358 error = mi_cpu_attach(curcpu()); 359 KASSERT(error == 0); 360 361 /* Initialize the sysctl subsystem. */ 362 sysctl_init(); 363 364 /* Initialize I/O statistics. */ 365 iostat_init(); 366 367 /* Initialize the file systems. */ 368 #ifdef NVNODE_IMPLICIT 369 /* 370 * If maximum number of vnodes in namei vnode cache is not explicitly 371 * defined in kernel config, adjust the number such as we use roughly 372 * 1.0% of memory for vnode cache (but not less than NVNODE vnodes). 373 */ 374 usevnodes = (ptoa((unsigned)physmem) / 100) / sizeof(struct vnode); 375 if (usevnodes > desiredvnodes) 376 desiredvnodes = usevnodes; 377 #endif 378 vfsinit(); 379 380 /* Initialize fstrans. */ 381 fstrans_init(); 382 383 /* Initialize the select()/poll() system calls. */ 384 selsysinit(); 385 386 /* Initialize asynchronous I/O. */ 387 aio_sysinit(); 388 389 /* Initialize message queues. */ 390 mqueue_sysinit(); 391 392 /* Initialize the system monitor subsystems. */ 393 #if NSYSMON_TASKQ > 0 394 sysmon_task_queue_preinit(); 395 #endif 396 397 #if NSYSMON_ENVSYS > 0 398 sysmon_envsys_init(); 399 #endif 400 401 #if NSYSMON_POWER > 0 402 sysmon_power_init(); 403 #endif 404 405 #if NSYSMON_WDOG > 0 406 sysmon_wdog_init(); 407 #endif 408 409 #ifdef __HAVE_TIMECOUNTER 410 inittimecounter(); 411 ntp_init(); 412 #endif /* __HAVE_TIMECOUNTER */ 413 414 #ifdef DDB 415 db_init_commands(); 416 #endif 417 418 /* Initialize the device switch tables. */ 419 devsw_init(); 420 421 /* Initialize the disk wedge subsystem. */ 422 dkwedge_init(); 423 424 /* Configure the system hardware. This will enable interrupts. */ 425 configure(); 426 427 #if defined(__SSP__) || defined(__SSP_ALL__) 428 { 429 #ifdef DIAGNOSTIC 430 printf("Initializing SSP:"); 431 #endif 432 /* 433 * We initialize ssp here carefully: 434 * 1. after we got some entropy 435 * 2. without calling a function 436 */ 437 size_t i; 438 long guard[__arraycount(__stack_chk_guard)]; 439 440 arc4randbytes(guard, sizeof(guard)); 441 for (i = 0; i < __arraycount(guard); i++) 442 __stack_chk_guard[i] = guard[i]; 443 #ifdef DIAGNOSTIC 444 for (i = 0; i < __arraycount(guard); i++) 445 printf("%lx ", guard[i]); 446 printf("\n"); 447 #endif 448 } 449 #endif 450 ubc_init(); /* must be after autoconfig */ 451 452 /* Lock the kernel on behalf of proc0. */ 453 KERNEL_LOCK(1, l); 454 455 #ifdef SYSTRACE 456 systrace_init(); 457 #endif 458 459 #ifdef SYSVSHM 460 /* Initialize System V style shared memory. */ 461 shminit(); 462 #endif 463 464 #ifdef SYSVSEM 465 /* Initialize System V style semaphores. */ 466 seminit(); 467 #endif 468 469 #ifdef SYSVMSG 470 /* Initialize System V style message queues. */ 471 msginit(); 472 #endif 473 474 #ifdef P1003_1B_SEMAPHORE 475 /* Initialize posix semaphores */ 476 ksem_init(); 477 #endif 478 479 #if NVERIEXEC > 0 480 /* 481 * Initialise the Veriexec subsystem. 482 */ 483 veriexec_init(); 484 #endif /* NVERIEXEC > 0 */ 485 486 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) 487 pax_init(); 488 #endif /* PAX_MPROTECT || PAX_SEGVGUARD */ 489 490 /* Attach pseudo-devices. */ 491 for (pdev = pdevinit; pdev->pdev_attach != NULL; pdev++) 492 (*pdev->pdev_attach)(pdev->pdev_count); 493 494 #ifdef FAST_IPSEC 495 /* Attach network crypto subsystem */ 496 ipsec_attach(); 497 #endif 498 499 /* 500 * Initialize protocols. Block reception of incoming packets 501 * until everything is ready. 502 */ 503 s = splnet(); 504 ifinit(); 505 domaininit(); 506 if_attachdomain(); 507 splx(s); 508 509 #ifdef GPROF 510 /* Initialize kernel profiling. */ 511 kmstartup(); 512 #endif 513 514 /* Initialize system accouting. */ 515 acct_init(); 516 517 #ifndef PIPE_SOCKETPAIR 518 /* Initialize pipes. */ 519 pipe_init(); 520 #endif 521 522 /* Setup the scheduler */ 523 sched_init(); 524 525 #ifdef KTRACE 526 /* Initialize ktrace. */ 527 ktrinit(); 528 #endif 529 530 /* Initialize the UUID system calls. */ 531 uuid_init(); 532 533 /* 534 * Create process 1 (init(8)). We do this now, as Unix has 535 * historically had init be process 1, and changing this would 536 * probably upset a lot of people. 537 * 538 * Note that process 1 won't immediately exec init(8), but will 539 * wait for us to inform it that the root file system has been 540 * mounted. 541 */ 542 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 543 panic("fork init"); 544 545 /* 546 * Now that device driver threads have been created, wait for 547 * them to finish any deferred autoconfiguration. Note we don't 548 * need to lock this semaphore, since we haven't booted any 549 * secondary processors, yet. 550 */ 551 while (config_pending) 552 (void) tsleep(&config_pending, PWAIT, "cfpend", 0); 553 554 /* 555 * Finalize configuration now that all real devices have been 556 * found. This needs to be done before the root device is 557 * selected, since finalization may create the root device. 558 */ 559 config_finalize(); 560 561 /* 562 * Now that autoconfiguration has completed, we can determine 563 * the root and dump devices. 564 */ 565 cpu_rootconf(); 566 cpu_dumpconf(); 567 568 /* Mount the root file system. */ 569 do { 570 domountroothook(); 571 if ((error = vfs_mountroot())) { 572 printf("cannot mount root, error = %d\n", error); 573 boothowto |= RB_ASKNAME; 574 setroot(root_device, 575 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 576 } 577 } while (error != 0); 578 mountroothook_destroy(); 579 580 /* 581 * Initialise the time-of-day clock, passing the time recorded 582 * in the root filesystem (if any) for use by systems that 583 * don't have a non-volatile time-of-day device. 584 */ 585 inittodr(rootfstime); 586 587 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 588 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 589 590 /* 591 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 592 * reference it. 593 */ 594 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 595 if (error) 596 panic("cannot find root vnode, error=%d", error); 597 cwdi0.cwdi_cdir = rootvnode; 598 VREF(cwdi0.cwdi_cdir); 599 VOP_UNLOCK(rootvnode, 0); 600 cwdi0.cwdi_rdir = NULL; 601 602 /* 603 * Now that root is mounted, we can fixup initproc's CWD 604 * info. All other processes are kthreads, which merely 605 * share proc0's CWD info. 606 */ 607 initproc->p_cwdi->cwdi_cdir = rootvnode; 608 VREF(initproc->p_cwdi->cwdi_cdir); 609 initproc->p_cwdi->cwdi_rdir = NULL; 610 611 /* 612 * Now can look at time, having had a chance to verify the time 613 * from the file system. Reset l->l_rtime as it may have been 614 * munched in mi_switch() after the time got set. 615 */ 616 #ifdef __HAVE_TIMECOUNTER 617 getmicrotime(&time); 618 #else 619 mono_time = time; 620 #endif 621 boottime = time; 622 mutex_enter(&proclist_lock); 623 LIST_FOREACH(p, &allproc, p_list) { 624 KASSERT((p->p_flag & PK_MARKER) == 0); 625 mutex_enter(&p->p_smutex); 626 p->p_stats->p_start = time; 627 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 628 lwp_lock(l); 629 l->l_cpu->ci_schedstate.spc_runtime = time; 630 l->l_rtime.tv_sec = l->l_rtime.tv_usec = 0; 631 lwp_unlock(l); 632 } 633 mutex_exit(&p->p_smutex); 634 } 635 mutex_exit(&proclist_lock); 636 637 for (CPU_INFO_FOREACH(cii, ci)) { 638 ci->ci_schedstate.spc_lastmod = time_second; 639 } 640 641 /* Create the pageout daemon kernel thread. */ 642 uvm_swap_init(); 643 if (kthread_create(PVM, 0, NULL, uvm_pageout, 644 NULL, NULL, "pgdaemon")) 645 panic("fork pagedaemon"); 646 647 /* Create the filesystem syncer kernel thread. */ 648 if (kthread_create(PINOD, 0, NULL, sched_sync, NULL, NULL, "ioflush")) 649 panic("fork syncer"); 650 651 /* Create the aiodone daemon kernel thread. */ 652 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 653 uvm_aiodone_worker, NULL, PVM, IPL_BIO, 0)) 654 panic("fork aiodoned"); 655 656 vmem_rehash_start(); 657 658 #if defined(MULTIPROCESSOR) 659 /* Boot the secondary processors. */ 660 cpu_boot_secondary_processors(); 661 #endif 662 663 /* Initialize exec structures */ 664 exec_init(1); 665 666 /* 667 * Okay, now we can let init(8) exec! It's off to userland! 668 */ 669 start_init_exec = 1; 670 wakeup(&start_init_exec); 671 672 /* The scheduler is an infinite loop. */ 673 uvm_scheduler(); 674 /* NOTREACHED */ 675 } 676 677 static void 678 check_console(struct lwp *l) 679 { 680 struct nameidata nd; 681 int error; 682 683 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console", l); 684 error = namei(&nd); 685 if (error == 0) 686 vrele(nd.ni_vp); 687 else if (error == ENOENT) 688 printf("warning: no /dev/console\n"); 689 else 690 printf("warning: lookup /dev/console: error %d\n", error); 691 } 692 693 /* 694 * List of paths to try when searching for "init". 695 */ 696 static const char *initpaths[] = { 697 "/sbin/init", 698 "/sbin/oinit", 699 "/sbin/init.bak", 700 NULL, 701 }; 702 703 /* 704 * Start the initial user process; try exec'ing each pathname in "initpaths". 705 * The program is invoked with one argument containing the boot flags. 706 */ 707 static void 708 start_init(void *arg) 709 { 710 struct lwp *l = arg; 711 struct proc *p = l->l_proc; 712 vaddr_t addr; 713 struct sys_execve_args /* { 714 syscallarg(const char *) path; 715 syscallarg(char * const *) argp; 716 syscallarg(char * const *) envp; 717 } */ args; 718 int options, i, error; 719 register_t retval[2]; 720 char flags[4], *flagsp; 721 const char *path, *slash; 722 char *ucp, **uap, *arg0, *arg1 = NULL; 723 char ipath[129]; 724 int ipx, len; 725 726 /* 727 * Now in process 1. 728 */ 729 strncpy(p->p_comm, "init", MAXCOMLEN); 730 731 /* 732 * Wait for main() to tell us that it's safe to exec. 733 */ 734 while (start_init_exec == 0) 735 (void) tsleep(&start_init_exec, PWAIT, "initexec", 0); 736 737 /* 738 * This is not the right way to do this. We really should 739 * hand-craft a descriptor onto /dev/console to hand to init, 740 * but that's a _lot_ more work, and the benefit from this easy 741 * hack makes up for the "good is the enemy of the best" effect. 742 */ 743 check_console(l); 744 745 /* 746 * Need just enough stack to hold the faked-up "execve()" arguments. 747 */ 748 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 749 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 750 NULL, UVM_UNKNOWN_OFFSET, 0, 751 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 752 UVM_ADV_NORMAL, 753 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 754 panic("init: couldn't allocate argument space"); 755 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 756 757 ipx = 0; 758 while (1) { 759 if (boothowto & RB_ASKNAME) { 760 printf("init path"); 761 if (initpaths[ipx]) 762 printf(" (default %s)", initpaths[ipx]); 763 printf(": "); 764 len = cngetsn(ipath, sizeof(ipath)-1); 765 if (len == 0) { 766 if (initpaths[ipx]) 767 path = initpaths[ipx++]; 768 else 769 continue; 770 } else { 771 ipath[len] = '\0'; 772 path = ipath; 773 } 774 } else { 775 if ((path = initpaths[ipx++]) == NULL) 776 break; 777 } 778 779 ucp = (char *)USRSTACK; 780 781 /* 782 * Construct the boot flag argument. 783 */ 784 flagsp = flags; 785 *flagsp++ = '-'; 786 options = 0; 787 788 if (boothowto & RB_SINGLE) { 789 *flagsp++ = 's'; 790 options = 1; 791 } 792 #ifdef notyet 793 if (boothowto & RB_FASTBOOT) { 794 *flagsp++ = 'f'; 795 options = 1; 796 } 797 #endif 798 799 /* 800 * Move out the flags (arg 1), if necessary. 801 */ 802 if (options != 0) { 803 *flagsp++ = '\0'; 804 i = flagsp - flags; 805 #ifdef DEBUG 806 printf("init: copying out flags `%s' %d\n", flags, i); 807 #endif 808 arg1 = STACK_ALLOC(ucp, i); 809 ucp = STACK_MAX(arg1, i); 810 (void)copyout((void *)flags, arg1, i); 811 } 812 813 /* 814 * Move out the file name (also arg 0). 815 */ 816 i = strlen(path) + 1; 817 #ifdef DEBUG 818 printf("init: copying out path `%s' %d\n", path, i); 819 #else 820 if (boothowto & RB_ASKNAME || path != initpaths[0]) 821 printf("init: trying %s\n", path); 822 #endif 823 arg0 = STACK_ALLOC(ucp, i); 824 ucp = STACK_MAX(arg0, i); 825 (void)copyout(path, arg0, i); 826 827 /* 828 * Move out the arg pointers. 829 */ 830 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 831 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 832 SCARG(&args, path) = arg0; 833 SCARG(&args, argp) = uap; 834 SCARG(&args, envp) = NULL; 835 slash = strrchr(path, '/'); 836 if (slash) 837 (void)suword((void *)uap++, 838 (long)arg0 + (slash + 1 - path)); 839 else 840 (void)suword((void *)uap++, (long)arg0); 841 if (options != 0) 842 (void)suword((void *)uap++, (long)arg1); 843 (void)suword((void *)uap++, 0); /* terminator */ 844 845 /* 846 * Now try to exec the program. If can't for any reason 847 * other than it doesn't exist, complain. 848 */ 849 error = sys_execve(l, &args, retval); 850 if (error == 0 || error == EJUSTRETURN) { 851 KERNEL_UNLOCK_LAST(l); 852 return; 853 } 854 printf("exec %s: error %d\n", path, error); 855 } 856 printf("init: not found\n"); 857 panic("no init"); 858 } 859
Indexes created Thu Sep 25 02:09:54 GMT 2025