init_main.c revision 1.333
1 /* $NetBSD: init_main.c,v 1.333 2007/12/26 16:01:35 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 1.333 2007/12/26 16:01:35 ad 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) 173 #include <sys/pax.h> 174 #endif /* PAX_MPROTECT || PAX_SEGVGUARD */ 175 #include <ufs/ufs/quota.h> 176 177 #include <miscfs/genfs/genfs.h> 178 #include <miscfs/syncfs/syncfs.h> 179 180 #include <sys/cpu.h> 181 182 #include <uvm/uvm.h> 183 184 #if NSYSMON_TASKQ > 0 185 #include <dev/sysmon/sysmon_taskq.h> 186 #endif 187 188 #include <dev/cons.h> 189 190 #if NSYSMON_ENVSYS > 0 || NSYSMON_POWER > 0 || NSYSMON_WDOG > 0 191 #include <dev/sysmon/sysmonvar.h> 192 #endif 193 194 #include <net/if.h> 195 #include <net/raw_cb.h> 196 197 #include <secmodel/secmodel.h> 198 199 extern struct proc proc0; 200 extern struct lwp lwp0; 201 extern struct cwdinfo cwdi0; 202 extern time_t rootfstime; 203 204 #ifndef curlwp 205 struct lwp *curlwp = &lwp0; 206 #endif 207 struct proc *initproc; 208 209 struct vnode *rootvp, *swapdev_vp; 210 int boothowto; 211 int cold = 1; /* still working on startup */ 212 struct timeval boottime; /* time at system startup - will only follow settime deltas */ 213 214 volatile int start_init_exec; /* semaphore for start_init() */ 215 216 static void check_console(struct lwp *l); 217 static void start_init(void *); 218 void main(void); 219 220 #if defined(__SSP__) || defined(__SSP_ALL__) 221 long __stack_chk_guard[8] = {0, 0, 0, 0, 0, 0, 0, 0}; 222 void __stack_chk_fail(void); 223 224 void 225 __stack_chk_fail(void) 226 { 227 panic("stack overflow detected; terminated"); 228 } 229 #endif 230 231 void __secmodel_none(void); 232 __weak_alias(secmodel_start,__secmodel_none); 233 void 234 __secmodel_none(void) 235 { 236 return; 237 } 238 239 /* 240 * System startup; initialize the world, create process 0, mount root 241 * filesystem, and fork to create init and pagedaemon. Most of the 242 * hard work is done in the lower-level initialization routines including 243 * startup(), which does memory initialization and autoconfiguration. 244 */ 245 void 246 main(void) 247 { 248 #ifdef __HAVE_TIMECOUNTER 249 struct timeval time; 250 #endif 251 struct lwp *l; 252 struct proc *p; 253 struct pdevinit *pdev; 254 int s, error; 255 extern struct pdevinit pdevinit[]; 256 #ifdef NVNODE_IMPLICIT 257 int usevnodes; 258 #endif 259 CPU_INFO_ITERATOR cii; 260 struct cpu_info *ci; 261 262 l = &lwp0; 263 #ifndef LWP0_CPU_INFO 264 l->l_cpu = curcpu(); 265 #endif 266 267 /* 268 * XXX This is a temporary check to be removed before 269 * NetBSD 5.0 is released. 270 */ 271 #if !defined(__i386__ ) && !defined(__x86_64__) 272 if (curlwp != l) { 273 printf("NOTICE: curlwp should be set before main()\n"); 274 DELAY(250000); 275 curlwp = l; 276 } 277 #endif 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 kmem_init(); 290 291 /* Initialize the extent manager. */ 292 extent_init(); 293 294 /* Do machine-dependent initialization. */ 295 cpu_startup(); 296 297 /* Initialize callouts, part 1. */ 298 callout_startup(); 299 300 /* 301 * Initialize the kernel authorization subsystem and start the 302 * default security model, if any. We need to do this early 303 * enough so that subsystems relying on any of the aforementioned 304 * can work properly. Since the security model may dictate the 305 * credential inheritance policy, it is needed at least before 306 * any process is created, specifically proc0. 307 */ 308 kauth_init(); 309 secmodel_start(); 310 311 /* Initialize the buffer cache */ 312 bufinit(); 313 314 /* 315 * Initialize mbuf's. Do this now because we might attempt to 316 * allocate mbufs or mbuf clusters during autoconfiguration. 317 */ 318 mbinit(); 319 320 /* Initialize sockets. */ 321 soinit(); 322 323 /* 324 * The following things must be done before autoconfiguration. 325 */ 326 evcnt_init(); /* initialize event counters */ 327 #if NRND > 0 328 rnd_init(); /* initialize RNG */ 329 #endif 330 331 /* Initialize process and pgrp structures. */ 332 procinit(); 333 lwpinit(); 334 335 /* Initialize signal-related data structures. */ 336 signal_init(); 337 338 /* Initialize resource management. */ 339 resource_init(); 340 341 /* Create process 0 (the swapper). */ 342 proc0_init(); 343 344 /* Initialize the UID hash table. */ 345 uid_init(); 346 347 /* Charge root for one process. */ 348 (void)chgproccnt(0, 1); 349 350 /* Initialize timekeeping. */ 351 time_init(); 352 353 /* Initialize the run queues, turnstiles and sleep queues. */ 354 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 355 sched_rqinit(); 356 turnstile_init(); 357 sleeptab_init(&sleeptab); 358 359 /* MI initialization of the boot cpu */ 360 error = mi_cpu_attach(curcpu()); 361 KASSERT(error == 0); 362 363 /* Initialize the sysctl subsystem. */ 364 sysctl_init(); 365 366 /* Initialize I/O statistics. */ 367 iostat_init(); 368 369 /* Initialize the log device. */ 370 loginit(); 371 372 /* Initialize the file systems. */ 373 #ifdef NVNODE_IMPLICIT 374 /* 375 * If maximum number of vnodes in namei vnode cache is not explicitly 376 * defined in kernel config, adjust the number such as we use roughly 377 * 1.0% of memory for vnode cache (but not less than NVNODE vnodes). 378 */ 379 usevnodes = (ptoa((unsigned)physmem) / 100) / sizeof(struct vnode); 380 if (usevnodes > desiredvnodes) 381 desiredvnodes = usevnodes; 382 #endif 383 vfsinit(); 384 385 /* Initialize fstrans. */ 386 fstrans_init(); 387 388 /* Initialize the file descriptor system. */ 389 filedesc_init(); 390 391 /* Initialize the select()/poll() system calls. */ 392 selsysinit(); 393 394 /* Initialize asynchronous I/O. */ 395 aio_sysinit(); 396 397 /* Initialize message queues. */ 398 mqueue_sysinit(); 399 400 /* Initialize the system monitor subsystems. */ 401 #if NSYSMON_TASKQ > 0 402 sysmon_task_queue_preinit(); 403 #endif 404 405 #if NSYSMON_ENVSYS > 0 406 sysmon_envsys_init(); 407 #endif 408 409 #if NSYSMON_POWER > 0 410 sysmon_power_init(); 411 #endif 412 413 #if NSYSMON_WDOG > 0 414 sysmon_wdog_init(); 415 #endif 416 417 #ifdef __HAVE_TIMECOUNTER 418 inittimecounter(); 419 ntp_init(); 420 #endif /* __HAVE_TIMECOUNTER */ 421 422 /* Initialize the device switch tables. */ 423 devsw_init(); 424 425 /* Initialize tty subsystem. */ 426 tty_init(); 427 ttyldisc_init(); 428 429 /* Initialize the disk wedge subsystem. */ 430 dkwedge_init(); 431 432 /* Configure the system hardware. This will enable interrupts. */ 433 configure(); 434 435 #if defined(__SSP__) || defined(__SSP_ALL__) 436 { 437 #ifdef DIAGNOSTIC 438 printf("Initializing SSP:"); 439 #endif 440 /* 441 * We initialize ssp here carefully: 442 * 1. after we got some entropy 443 * 2. without calling a function 444 */ 445 size_t i; 446 long guard[__arraycount(__stack_chk_guard)]; 447 448 arc4randbytes(guard, sizeof(guard)); 449 for (i = 0; i < __arraycount(guard); i++) 450 __stack_chk_guard[i] = guard[i]; 451 #ifdef DIAGNOSTIC 452 for (i = 0; i < __arraycount(guard); i++) 453 printf("%lx ", guard[i]); 454 printf("\n"); 455 #endif 456 } 457 #endif 458 ubc_init(); /* must be after autoconfig */ 459 460 /* Lock the kernel on behalf of proc0. */ 461 KERNEL_LOCK(1, l); 462 463 #ifdef SYSTRACE 464 systrace_init(); 465 #endif 466 467 #ifdef SYSVSHM 468 /* Initialize System V style shared memory. */ 469 shminit(); 470 #endif 471 472 #ifdef SYSVSEM 473 /* Initialize System V style semaphores. */ 474 seminit(); 475 #endif 476 477 #ifdef SYSVMSG 478 /* Initialize System V style message queues. */ 479 msginit(); 480 #endif 481 482 #ifdef P1003_1B_SEMAPHORE 483 /* Initialize posix semaphores */ 484 ksem_init(); 485 #endif 486 487 #if NVERIEXEC > 0 488 /* 489 * Initialise the Veriexec subsystem. 490 */ 491 veriexec_init(); 492 #endif /* NVERIEXEC > 0 */ 493 494 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) 495 pax_init(); 496 #endif /* PAX_MPROTECT || PAX_SEGVGUARD */ 497 498 /* Attach pseudo-devices. */ 499 for (pdev = pdevinit; pdev->pdev_attach != NULL; pdev++) 500 (*pdev->pdev_attach)(pdev->pdev_count); 501 502 #ifdef FAST_IPSEC 503 /* Attach network crypto subsystem */ 504 ipsec_attach(); 505 #endif 506 507 /* 508 * Initialize protocols. Block reception of incoming packets 509 * until everything is ready. 510 */ 511 s = splnet(); 512 ifinit(); 513 domaininit(); 514 if_attachdomain(); 515 splx(s); 516 517 #ifdef GPROF 518 /* Initialize kernel profiling. */ 519 kmstartup(); 520 #endif 521 522 /* Initialize system accouting. */ 523 acct_init(); 524 525 #ifndef PIPE_SOCKETPAIR 526 /* Initialize pipes. */ 527 pipe_init(); 528 #endif 529 530 /* Setup the scheduler */ 531 sched_init(); 532 533 #ifdef KTRACE 534 /* Initialize ktrace. */ 535 ktrinit(); 536 #endif 537 538 /* Initialize the UUID system calls. */ 539 uuid_init(); 540 541 /* 542 * Create process 1 (init(8)). We do this now, as Unix has 543 * historically had init be process 1, and changing this would 544 * probably upset a lot of people. 545 * 546 * Note that process 1 won't immediately exec init(8), but will 547 * wait for us to inform it that the root file system has been 548 * mounted. 549 */ 550 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 551 panic("fork init"); 552 553 /* 554 * Now that device driver threads have been created, wait for 555 * them to finish any deferred autoconfiguration. 556 */ 557 while (config_pending) 558 (void) tsleep(&config_pending, PWAIT, "cfpend", hz); 559 560 /* 561 * Finalize configuration now that all real devices have been 562 * found. This needs to be done before the root device is 563 * selected, since finalization may create the root device. 564 */ 565 config_finalize(); 566 567 /* 568 * Now that autoconfiguration has completed, we can determine 569 * the root and dump devices. 570 */ 571 cpu_rootconf(); 572 cpu_dumpconf(); 573 574 /* Mount the root file system. */ 575 do { 576 domountroothook(); 577 if ((error = vfs_mountroot())) { 578 printf("cannot mount root, error = %d\n", error); 579 boothowto |= RB_ASKNAME; 580 setroot(root_device, 581 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 582 } 583 } while (error != 0); 584 mountroothook_destroy(); 585 586 /* 587 * Initialise the time-of-day clock, passing the time recorded 588 * in the root filesystem (if any) for use by systems that 589 * don't have a non-volatile time-of-day device. 590 */ 591 inittodr(rootfstime); 592 593 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 594 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 595 596 /* 597 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 598 * reference it. 599 */ 600 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 601 if (error) 602 panic("cannot find root vnode, error=%d", error); 603 cwdi0.cwdi_cdir = rootvnode; 604 VREF(cwdi0.cwdi_cdir); 605 VOP_UNLOCK(rootvnode, 0); 606 cwdi0.cwdi_rdir = NULL; 607 608 /* 609 * Now that root is mounted, we can fixup initproc's CWD 610 * info. All other processes are kthreads, which merely 611 * share proc0's CWD info. 612 */ 613 initproc->p_cwdi->cwdi_cdir = rootvnode; 614 VREF(initproc->p_cwdi->cwdi_cdir); 615 initproc->p_cwdi->cwdi_rdir = NULL; 616 617 /* 618 * Now can look at time, having had a chance to verify the time 619 * from the file system. Reset l->l_rtime as it may have been 620 * munched in mi_switch() after the time got set. 621 */ 622 #ifdef __HAVE_TIMECOUNTER 623 getmicrotime(&time); 624 #else 625 mono_time = time; 626 #endif 627 boottime = time; 628 mutex_enter(&proclist_lock); 629 LIST_FOREACH(p, &allproc, p_list) { 630 KASSERT((p->p_flag & PK_MARKER) == 0); 631 mutex_enter(&p->p_smutex); 632 p->p_stats->p_start = time; 633 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 634 lwp_lock(l); 635 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 636 lwp_unlock(l); 637 } 638 mutex_exit(&p->p_smutex); 639 } 640 mutex_exit(&proclist_lock); 641 binuptime(&curlwp->l_stime); 642 643 for (CPU_INFO_FOREACH(cii, ci)) { 644 ci->ci_schedstate.spc_lastmod = time_second; 645 } 646 647 /* Create the pageout daemon kernel thread. */ 648 uvm_swap_init(); 649 if (kthread_create(PRI_PGDAEMON, 0, NULL, uvm_pageout, 650 NULL, NULL, "pgdaemon")) 651 panic("fork pagedaemon"); 652 653 /* Create the filesystem syncer kernel thread. */ 654 if (kthread_create(PRI_IOFLUSH, 0, NULL, sched_sync, NULL, NULL, "ioflush")) 655 panic("fork syncer"); 656 657 /* Create the aiodone daemon kernel thread. */ 658 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 659 uvm_aiodone_worker, NULL, PRI_VM, IPL_BIO, 0)) 660 panic("fork aiodoned"); 661 662 vmem_rehash_start(); 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"); 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 * const 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 Sun Sep 21 20:09:37 GMT 2025