init_main.c revision 1.314
1 /* $NetBSD: init_main.c,v 1.314 2007/09/07 18:56:08 rmind 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 1.314 2007/09/07 18:56:08 rmind 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 /* 265 * Initialize the current LWP pointer (curlwp) before 266 * any possible traps/probes to simplify trap processing. 267 */ 268 l = &lwp0; 269 curlwp = l; 270 l->l_cpu = curcpu(); 271 l->l_proc = &proc0; 272 l->l_lid = 1; 273 274 /* 275 * Attempt to find console and initialize 276 * in case of early panic or other messages. 277 */ 278 consinit(); 279 280 KERNEL_LOCK_INIT(); 281 282 uvm_init(); 283 284 #ifdef DEBUG 285 debug_init(); 286 #endif 287 288 kmem_init(); 289 290 /* Initialize the extent manager. */ 291 extent_init(); 292 293 /* Do machine-dependent initialization. */ 294 cpu_startup(); 295 296 /* Initialize callouts, part 1. */ 297 callout_startup(); 298 299 /* 300 * Initialize the kernel authorization subsystem and start the 301 * default security model, if any. We need to do this early 302 * enough so that subsystems relying on any of the aforementioned 303 * can work properly. Since the security model may dictate the 304 * credential inheritance policy, it is needed at least before 305 * any process is created, specifically proc0. 306 */ 307 kauth_init(); 308 secmodel_start(); 309 310 /* Initialize the buffer cache */ 311 bufinit(); 312 313 /* 314 * Initialize mbuf's. Do this now because we might attempt to 315 * allocate mbufs or mbuf clusters during autoconfiguration. 316 */ 317 mbinit(); 318 319 /* Initialize sockets. */ 320 soinit(); 321 322 /* 323 * The following things must be done before autoconfiguration. 324 */ 325 evcnt_init(); /* initialize event counters */ 326 #if NRND > 0 327 rnd_init(); /* initialize RNG */ 328 #endif 329 330 /* Initialize process and pgrp structures. */ 331 procinit(); 332 lwpinit(); 333 334 /* Initialize signal-related data structures. */ 335 signal_init(); 336 337 /* Create process 0 (the swapper). */ 338 proc0_init(); 339 340 /* Initialize the UID hash table. */ 341 uid_init(); 342 343 /* Charge root for one process. */ 344 (void)chgproccnt(0, 1); 345 346 /* Initialize the run queues, turnstiles and sleep queues. */ 347 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 348 sched_rqinit(); 349 turnstile_init(); 350 sleeptab_init(&sleeptab); 351 352 /* MI initialization of the boot cpu */ 353 error = mi_cpu_attach(curcpu()); 354 KASSERT(error == 0); 355 356 /* Initialize the sysctl subsystem. */ 357 sysctl_init(); 358 359 /* Initialize I/O statistics. */ 360 iostat_init(); 361 362 /* Initialize the file systems. */ 363 #ifdef NVNODE_IMPLICIT 364 /* 365 * If maximum number of vnodes in namei vnode cache is not explicitly 366 * defined in kernel config, adjust the number such as we use roughly 367 * 1.0% of memory for vnode cache (but not less than NVNODE vnodes). 368 */ 369 usevnodes = (ptoa((unsigned)physmem) / 100) / sizeof(struct vnode); 370 if (usevnodes > desiredvnodes) 371 desiredvnodes = usevnodes; 372 #endif 373 vfsinit(); 374 375 /* Initialize fstrans. */ 376 fstrans_init(); 377 378 /* Initialize the select()/poll() system calls. */ 379 selsysinit(); 380 381 /* Initialize asynchronous I/O. */ 382 aio_sysinit(); 383 384 /* Initialize message queues. */ 385 mqueue_sysinit(); 386 387 /* Initialize the system monitor subsystems. */ 388 #if NSYSMON_TASKQ > 0 389 sysmon_task_queue_preinit(); 390 #endif 391 392 #if NSYSMON_ENVSYS > 0 393 sysmon_envsys_init(); 394 #endif 395 396 #if NSYSMON_POWER > 0 397 sysmon_power_init(); 398 #endif 399 400 #if NSYSMON_WDOG > 0 401 sysmon_wdog_init(); 402 #endif 403 404 #ifdef __HAVE_TIMECOUNTER 405 inittimecounter(); 406 ntp_init(); 407 #endif /* __HAVE_TIMECOUNTER */ 408 409 /* Initialize the device switch tables. */ 410 devsw_init(); 411 412 /* Initialize the disk wedge subsystem. */ 413 dkwedge_init(); 414 415 /* Configure the system hardware. This will enable interrupts. */ 416 configure(); 417 418 #if defined(__SSP__) || defined(__SSP_ALL__) 419 { 420 #ifdef DIAGNOSTIC 421 printf("Initializing SSP:"); 422 #endif 423 /* 424 * We initialize ssp here carefully: 425 * 1. after we got some entropy 426 * 2. without calling a function 427 */ 428 size_t i; 429 long guard[__arraycount(__stack_chk_guard)]; 430 431 arc4randbytes(guard, sizeof(guard)); 432 for (i = 0; i < __arraycount(guard); i++) 433 __stack_chk_guard[i] = guard[i]; 434 #ifdef DIAGNOSTIC 435 for (i = 0; i < __arraycount(guard); i++) 436 printf("%lx ", guard[i]); 437 printf("\n"); 438 #endif 439 } 440 #endif 441 ubc_init(); /* must be after autoconfig */ 442 443 /* Lock the kernel on behalf of proc0. */ 444 KERNEL_LOCK(1, l); 445 446 #ifdef SYSTRACE 447 systrace_init(); 448 #endif 449 450 #ifdef SYSVSHM 451 /* Initialize System V style shared memory. */ 452 shminit(); 453 #endif 454 455 #ifdef SYSVSEM 456 /* Initialize System V style semaphores. */ 457 seminit(); 458 #endif 459 460 #ifdef SYSVMSG 461 /* Initialize System V style message queues. */ 462 msginit(); 463 #endif 464 465 #ifdef P1003_1B_SEMAPHORE 466 /* Initialize posix semaphores */ 467 ksem_init(); 468 #endif 469 470 #if NVERIEXEC > 0 471 /* 472 * Initialise the Veriexec subsystem. 473 */ 474 veriexec_init(); 475 #endif /* NVERIEXEC > 0 */ 476 477 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) 478 pax_init(); 479 #endif /* PAX_MPROTECT || PAX_SEGVGUARD */ 480 481 /* Attach pseudo-devices. */ 482 for (pdev = pdevinit; pdev->pdev_attach != NULL; pdev++) 483 (*pdev->pdev_attach)(pdev->pdev_count); 484 485 #ifdef FAST_IPSEC 486 /* Attach network crypto subsystem */ 487 ipsec_attach(); 488 #endif 489 490 /* 491 * Initialize protocols. Block reception of incoming packets 492 * until everything is ready. 493 */ 494 s = splnet(); 495 ifinit(); 496 domaininit(); 497 if_attachdomain(); 498 splx(s); 499 500 #ifdef GPROF 501 /* Initialize kernel profiling. */ 502 kmstartup(); 503 #endif 504 505 /* Initialize system accouting. */ 506 acct_init(); 507 508 #ifndef PIPE_SOCKETPAIR 509 /* Initialize pipes. */ 510 pipe_init(); 511 #endif 512 513 /* Setup the scheduler */ 514 sched_init(); 515 516 #ifdef KTRACE 517 /* Initialize ktrace. */ 518 ktrinit(); 519 #endif 520 521 /* Initialize the UUID system calls. */ 522 uuid_init(); 523 524 /* 525 * Create process 1 (init(8)). We do this now, as Unix has 526 * historically had init be process 1, and changing this would 527 * probably upset a lot of people. 528 * 529 * Note that process 1 won't immediately exec init(8), but will 530 * wait for us to inform it that the root file system has been 531 * mounted. 532 */ 533 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 534 panic("fork init"); 535 536 /* 537 * Now that device driver threads have been created, wait for 538 * them to finish any deferred autoconfiguration. Note we don't 539 * need to lock this semaphore, since we haven't booted any 540 * secondary processors, yet. 541 */ 542 while (config_pending) 543 (void) tsleep(&config_pending, PWAIT, "cfpend", 0); 544 545 /* 546 * Finalize configuration now that all real devices have been 547 * found. This needs to be done before the root device is 548 * selected, since finalization may create the root device. 549 */ 550 config_finalize(); 551 552 /* 553 * Now that autoconfiguration has completed, we can determine 554 * the root and dump devices. 555 */ 556 cpu_rootconf(); 557 cpu_dumpconf(); 558 559 /* Mount the root file system. */ 560 do { 561 domountroothook(); 562 if ((error = vfs_mountroot())) { 563 printf("cannot mount root, error = %d\n", error); 564 boothowto |= RB_ASKNAME; 565 setroot(root_device, 566 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 567 } 568 } while (error != 0); 569 mountroothook_destroy(); 570 571 /* 572 * Initialise the time-of-day clock, passing the time recorded 573 * in the root filesystem (if any) for use by systems that 574 * don't have a non-volatile time-of-day device. 575 */ 576 inittodr(rootfstime); 577 578 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 579 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 580 581 /* 582 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 583 * reference it. 584 */ 585 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 586 if (error) 587 panic("cannot find root vnode, error=%d", error); 588 cwdi0.cwdi_cdir = rootvnode; 589 VREF(cwdi0.cwdi_cdir); 590 VOP_UNLOCK(rootvnode, 0); 591 cwdi0.cwdi_rdir = NULL; 592 593 /* 594 * Now that root is mounted, we can fixup initproc's CWD 595 * info. All other processes are kthreads, which merely 596 * share proc0's CWD info. 597 */ 598 initproc->p_cwdi->cwdi_cdir = rootvnode; 599 VREF(initproc->p_cwdi->cwdi_cdir); 600 initproc->p_cwdi->cwdi_rdir = NULL; 601 602 /* 603 * Now can look at time, having had a chance to verify the time 604 * from the file system. Reset l->l_rtime as it may have been 605 * munched in mi_switch() after the time got set. 606 */ 607 #ifdef __HAVE_TIMECOUNTER 608 getmicrotime(&time); 609 #else 610 mono_time = time; 611 #endif 612 boottime = time; 613 mutex_enter(&proclist_lock); 614 LIST_FOREACH(p, &allproc, p_list) { 615 KASSERT((p->p_flag & PK_MARKER) == 0); 616 mutex_enter(&p->p_smutex); 617 p->p_stats->p_start = time; 618 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 619 lwp_lock(l); 620 l->l_cpu->ci_schedstate.spc_runtime = time; 621 l->l_rtime.tv_sec = l->l_rtime.tv_usec = 0; 622 lwp_unlock(l); 623 } 624 mutex_exit(&p->p_smutex); 625 } 626 mutex_exit(&proclist_lock); 627 628 for (CPU_INFO_FOREACH(cii, ci)) { 629 ci->ci_schedstate.spc_lastmod = time_second; 630 } 631 632 /* Create the pageout daemon kernel thread. */ 633 uvm_swap_init(); 634 if (kthread_create(PVM, 0, NULL, uvm_pageout, 635 NULL, NULL, "pgdaemon")) 636 panic("fork pagedaemon"); 637 638 /* Create the filesystem syncer kernel thread. */ 639 if (kthread_create(PINOD, 0, NULL, sched_sync, NULL, NULL, "ioflush")) 640 panic("fork syncer"); 641 642 /* Create the aiodone daemon kernel thread. */ 643 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 644 uvm_aiodone_worker, NULL, PVM, IPL_BIO, 0)) 645 panic("fork aiodoned"); 646 647 vmem_rehash_start(); 648 649 #if defined(MULTIPROCESSOR) 650 /* Boot the secondary processors. */ 651 cpu_boot_secondary_processors(); 652 #endif 653 654 /* Initialize exec structures */ 655 exec_init(1); 656 657 /* 658 * Okay, now we can let init(8) exec! It's off to userland! 659 */ 660 start_init_exec = 1; 661 wakeup(&start_init_exec); 662 663 /* The scheduler is an infinite loop. */ 664 uvm_scheduler(); 665 /* NOTREACHED */ 666 } 667 668 static void 669 check_console(struct lwp *l) 670 { 671 struct nameidata nd; 672 int error; 673 674 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console", l); 675 error = namei(&nd); 676 if (error == 0) 677 vrele(nd.ni_vp); 678 else if (error == ENOENT) 679 printf("warning: no /dev/console\n"); 680 else 681 printf("warning: lookup /dev/console: error %d\n", error); 682 } 683 684 /* 685 * List of paths to try when searching for "init". 686 */ 687 static const char *initpaths[] = { 688 "/sbin/init", 689 "/sbin/oinit", 690 "/sbin/init.bak", 691 NULL, 692 }; 693 694 /* 695 * Start the initial user process; try exec'ing each pathname in "initpaths". 696 * The program is invoked with one argument containing the boot flags. 697 */ 698 static void 699 start_init(void *arg) 700 { 701 struct lwp *l = arg; 702 struct proc *p = l->l_proc; 703 vaddr_t addr; 704 struct sys_execve_args /* { 705 syscallarg(const char *) path; 706 syscallarg(char * const *) argp; 707 syscallarg(char * const *) envp; 708 } */ args; 709 int options, i, error; 710 register_t retval[2]; 711 char flags[4], *flagsp; 712 const char *path, *slash; 713 char *ucp, **uap, *arg0, *arg1 = NULL; 714 char ipath[129]; 715 int ipx, len; 716 717 /* 718 * Now in process 1. 719 */ 720 strncpy(p->p_comm, "init", MAXCOMLEN); 721 722 /* 723 * Wait for main() to tell us that it's safe to exec. 724 */ 725 while (start_init_exec == 0) 726 (void) tsleep(&start_init_exec, PWAIT, "initexec", 0); 727 728 /* 729 * This is not the right way to do this. We really should 730 * hand-craft a descriptor onto /dev/console to hand to init, 731 * but that's a _lot_ more work, and the benefit from this easy 732 * hack makes up for the "good is the enemy of the best" effect. 733 */ 734 check_console(l); 735 736 /* 737 * Need just enough stack to hold the faked-up "execve()" arguments. 738 */ 739 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 740 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 741 NULL, UVM_UNKNOWN_OFFSET, 0, 742 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 743 UVM_ADV_NORMAL, 744 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 745 panic("init: couldn't allocate argument space"); 746 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 747 748 ipx = 0; 749 while (1) { 750 if (boothowto & RB_ASKNAME) { 751 printf("init path"); 752 if (initpaths[ipx]) 753 printf(" (default %s)", initpaths[ipx]); 754 printf(": "); 755 len = cngetsn(ipath, sizeof(ipath)-1); 756 if (len == 0) { 757 if (initpaths[ipx]) 758 path = initpaths[ipx++]; 759 else 760 continue; 761 } else { 762 ipath[len] = '\0'; 763 path = ipath; 764 } 765 } else { 766 if ((path = initpaths[ipx++]) == NULL) 767 break; 768 } 769 770 ucp = (char *)USRSTACK; 771 772 /* 773 * Construct the boot flag argument. 774 */ 775 flagsp = flags; 776 *flagsp++ = '-'; 777 options = 0; 778 779 if (boothowto & RB_SINGLE) { 780 *flagsp++ = 's'; 781 options = 1; 782 } 783 #ifdef notyet 784 if (boothowto & RB_FASTBOOT) { 785 *flagsp++ = 'f'; 786 options = 1; 787 } 788 #endif 789 790 /* 791 * Move out the flags (arg 1), if necessary. 792 */ 793 if (options != 0) { 794 *flagsp++ = '\0'; 795 i = flagsp - flags; 796 #ifdef DEBUG 797 printf("init: copying out flags `%s' %d\n", flags, i); 798 #endif 799 arg1 = STACK_ALLOC(ucp, i); 800 ucp = STACK_MAX(arg1, i); 801 (void)copyout((void *)flags, arg1, i); 802 } 803 804 /* 805 * Move out the file name (also arg 0). 806 */ 807 i = strlen(path) + 1; 808 #ifdef DEBUG 809 printf("init: copying out path `%s' %d\n", path, i); 810 #else 811 if (boothowto & RB_ASKNAME || path != initpaths[0]) 812 printf("init: trying %s\n", path); 813 #endif 814 arg0 = STACK_ALLOC(ucp, i); 815 ucp = STACK_MAX(arg0, i); 816 (void)copyout(path, arg0, i); 817 818 /* 819 * Move out the arg pointers. 820 */ 821 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 822 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 823 SCARG(&args, path) = arg0; 824 SCARG(&args, argp) = uap; 825 SCARG(&args, envp) = NULL; 826 slash = strrchr(path, '/'); 827 if (slash) 828 (void)suword((void *)uap++, 829 (long)arg0 + (slash + 1 - path)); 830 else 831 (void)suword((void *)uap++, (long)arg0); 832 if (options != 0) 833 (void)suword((void *)uap++, (long)arg1); 834 (void)suword((void *)uap++, 0); /* terminator */ 835 836 /* 837 * Now try to exec the program. If can't for any reason 838 * other than it doesn't exist, complain. 839 */ 840 error = sys_execve(l, &args, retval); 841 if (error == 0 || error == EJUSTRETURN) { 842 KERNEL_UNLOCK_LAST(l); 843 return; 844 } 845 printf("exec %s: error %d\n", path, error); 846 } 847 printf("init: not found\n"); 848 panic("no init"); 849 } 850
Indexes created Wed Sep 24 15:09:51 GMT 2025