init_main.c revision 1.339
1 /* $NetBSD: init_main.c,v 1.339 2008/01/15 03:37:10 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 1.339 2008/01/15 03:37:10 rmind 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_fileassoc.h" 83 #include "opt_ktrace.h" 84 #include "opt_pax.h" 85 86 #include "rnd.h" 87 #include "sysmon_envsys.h" 88 #include "sysmon_power.h" 89 #include "sysmon_taskq.h" 90 #include "sysmon_wdog.h" 91 #include "veriexec.h" 92 93 #include <sys/param.h> 94 #include <sys/acct.h> 95 #include <sys/filedesc.h> 96 #include <sys/file.h> 97 #include <sys/errno.h> 98 #include <sys/callout.h> 99 #include <sys/cpu.h> 100 #include <sys/kernel.h> 101 #include <sys/kmem.h> 102 #include <sys/mount.h> 103 #include <sys/proc.h> 104 #include <sys/kthread.h> 105 #include <sys/resourcevar.h> 106 #include <sys/signalvar.h> 107 #include <sys/systm.h> 108 #include <sys/vnode.h> 109 #include <sys/fstrans.h> 110 #include <sys/tty.h> 111 #include <sys/conf.h> 112 #include <sys/disklabel.h> 113 #include <sys/buf.h> 114 #include <sys/device.h> 115 #include <sys/exec.h> 116 #include <sys/socketvar.h> 117 #include <sys/protosw.h> 118 #include <sys/percpu.h> 119 #include <sys/sysctl.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 #include <sys/msgbuf.h> 134 #ifdef FAST_IPSEC 135 #include <netipsec/ipsec.h> 136 #endif 137 #ifdef SYSVSHM 138 #include <sys/shm.h> 139 #endif 140 #ifdef SYSVSEM 141 #include <sys/sem.h> 142 #endif 143 #ifdef SYSVMSG 144 #include <sys/msg.h> 145 #endif 146 #ifdef P1003_1B_SEMAPHORE 147 #include <sys/ksem.h> 148 #endif 149 #include <sys/domain.h> 150 #include <sys/namei.h> 151 #if NRND > 0 152 #include <sys/rnd.h> 153 #endif 154 #include <sys/pipe.h> 155 #ifdef LKM 156 #include <sys/lkm.h> 157 #endif 158 #if NVERIEXEC > 0 159 #include <sys/verified_exec.h> 160 #endif /* NVERIEXEC > 0 */ 161 #ifdef KTRACE 162 #include <sys/ktrace.h> 163 #endif 164 #include <sys/kauth.h> 165 #include <net80211/ieee80211_netbsd.h> 166 167 #include <sys/syscall.h> 168 #include <sys/syscallargs.h> 169 170 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 171 #include <sys/pax.h> 172 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 173 174 #include <ufs/ufs/quota.h> 175 176 #include <miscfs/genfs/genfs.h> 177 #include <miscfs/syncfs/syncfs.h> 178 179 #include <sys/cpu.h> 180 181 #include <uvm/uvm.h> 182 183 #if NSYSMON_TASKQ > 0 184 #include <dev/sysmon/sysmon_taskq.h> 185 #endif 186 187 #include <dev/cons.h> 188 189 #if NSYSMON_ENVSYS > 0 || NSYSMON_POWER > 0 || NSYSMON_WDOG > 0 190 #include <dev/sysmon/sysmonvar.h> 191 #endif 192 193 #include <net/if.h> 194 #include <net/raw_cb.h> 195 196 #include <secmodel/secmodel.h> 197 198 extern struct proc proc0; 199 extern struct lwp lwp0; 200 extern struct cwdinfo cwdi0; 201 extern time_t rootfstime; 202 203 #ifndef curlwp 204 struct lwp *curlwp = &lwp0; 205 #endif 206 struct proc *initproc; 207 208 struct vnode *rootvp, *swapdev_vp; 209 int boothowto; 210 int cold = 1; /* still working on startup */ 211 struct timeval boottime; /* time at system startup - will only follow settime deltas */ 212 213 volatile int start_init_exec; /* semaphore for start_init() */ 214 215 static void check_console(struct lwp *l); 216 static void start_init(void *); 217 void main(void); 218 219 #if defined(__SSP__) || defined(__SSP_ALL__) 220 long __stack_chk_guard[8] = {0, 0, 0, 0, 0, 0, 0, 0}; 221 void __stack_chk_fail(void); 222 223 void 224 __stack_chk_fail(void) 225 { 226 panic("stack overflow detected; terminated"); 227 } 228 #endif 229 230 void __secmodel_none(void); 231 __weak_alias(secmodel_start,__secmodel_none); 232 void 233 __secmodel_none(void) 234 { 235 return; 236 } 237 238 /* 239 * System startup; initialize the world, create process 0, mount root 240 * filesystem, and fork to create init and pagedaemon. Most of the 241 * hard work is done in the lower-level initialization routines including 242 * startup(), which does memory initialization and autoconfiguration. 243 */ 244 void 245 main(void) 246 { 247 #ifdef __HAVE_TIMECOUNTER 248 struct timeval time; 249 #endif 250 struct lwp *l; 251 struct proc *p; 252 struct pdevinit *pdev; 253 int s, error; 254 extern struct pdevinit pdevinit[]; 255 #ifdef NVNODE_IMPLICIT 256 int usevnodes; 257 #endif 258 CPU_INFO_ITERATOR cii; 259 struct cpu_info *ci; 260 261 l = &lwp0; 262 #ifndef LWP0_CPU_INFO 263 l->l_cpu = curcpu(); 264 #endif 265 266 /* 267 * Attempt to find console and initialize 268 * in case of early panic or other messages. 269 */ 270 consinit(); 271 272 kernel_lock_init(); 273 274 uvm_init(); 275 276 kmem_init(); 277 278 percpu_init(); 279 280 /* Initialize the extent manager. */ 281 extent_init(); 282 283 /* Do machine-dependent initialization. */ 284 cpu_startup(); 285 286 /* Initialize callouts, part 1. */ 287 callout_startup(); 288 289 /* 290 * Initialize the kernel authorization subsystem and start the 291 * default security model, if any. We need to do this early 292 * enough so that subsystems relying on any of the aforementioned 293 * can work properly. Since the security model may dictate the 294 * credential inheritance policy, it is needed at least before 295 * any process is created, specifically proc0. 296 */ 297 kauth_init(); 298 secmodel_start(); 299 300 /* Initialize the buffer cache */ 301 bufinit(); 302 303 /* Initialize sockets. */ 304 soinit(); 305 306 /* 307 * The following things must be done before autoconfiguration. 308 */ 309 evcnt_init(); /* initialize event counters */ 310 #if NRND > 0 311 rnd_init(); /* initialize RNG */ 312 #endif 313 314 /* Initialize process and pgrp structures. */ 315 procinit(); 316 lwpinit(); 317 318 /* Initialize signal-related data structures. */ 319 signal_init(); 320 321 /* Initialize resource management. */ 322 resource_init(); 323 324 /* Create process 0 (the swapper). */ 325 proc0_init(); 326 327 /* Initialize the UID hash table. */ 328 uid_init(); 329 330 /* Charge root for one process. */ 331 (void)chgproccnt(0, 1); 332 333 /* Initialize timekeeping. */ 334 time_init(); 335 336 /* Initialize the run queues, turnstiles and sleep queues. */ 337 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 338 sched_rqinit(); 339 turnstile_init(); 340 sleeptab_init(&sleeptab); 341 342 /* Initialize processor-sets */ 343 psets_init(); 344 345 /* MI initialization of the boot cpu */ 346 error = mi_cpu_attach(curcpu()); 347 KASSERT(error == 0); 348 349 /* 350 * Initialize mbuf's. Do this now because we might attempt to 351 * allocate mbufs or mbuf clusters during autoconfiguration. 352 */ 353 mbinit(); 354 355 /* Initialize the sysctl subsystem. */ 356 sysctl_init(); 357 358 /* Initialize I/O statistics. */ 359 iostat_init(); 360 361 /* Initialize the log device. */ 362 loginit(); 363 364 /* Initialize the file systems. */ 365 #ifdef NVNODE_IMPLICIT 366 /* 367 * If maximum number of vnodes in namei vnode cache is not explicitly 368 * defined in kernel config, adjust the number such as we use roughly 369 * 1.0% of memory for vnode cache (but not less than NVNODE vnodes). 370 */ 371 usevnodes = (ptoa((unsigned)physmem) / 100) / sizeof(struct vnode); 372 if (usevnodes > desiredvnodes) 373 desiredvnodes = usevnodes; 374 #endif 375 vfsinit(); 376 377 /* Initialize fstrans. */ 378 fstrans_init(); 379 380 /* Initialize the file descriptor system. */ 381 filedesc_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 /* Initialize the device switch tables. */ 415 devsw_init(); 416 417 /* Initialize tty subsystem. */ 418 tty_init(); 419 ttyldisc_init(); 420 421 /* Initialize the buffer cache, part 2. */ 422 bufinit2(); 423 424 /* Initialize the disk wedge subsystem. */ 425 dkwedge_init(); 426 427 /* Configure the system hardware. This will enable interrupts. */ 428 configure(); 429 430 #if defined(__SSP__) || defined(__SSP_ALL__) 431 { 432 #ifdef DIAGNOSTIC 433 printf("Initializing SSP:"); 434 #endif 435 /* 436 * We initialize ssp here carefully: 437 * 1. after we got some entropy 438 * 2. without calling a function 439 */ 440 size_t i; 441 long guard[__arraycount(__stack_chk_guard)]; 442 443 arc4randbytes(guard, sizeof(guard)); 444 for (i = 0; i < __arraycount(guard); i++) 445 __stack_chk_guard[i] = guard[i]; 446 #ifdef DIAGNOSTIC 447 for (i = 0; i < __arraycount(guard); i++) 448 printf("%lx ", guard[i]); 449 printf("\n"); 450 #endif 451 } 452 #endif 453 ubc_init(); /* must be after autoconfig */ 454 455 /* Lock the kernel on behalf of proc0. */ 456 KERNEL_LOCK(1, l); 457 458 #ifdef SYSVSHM 459 /* Initialize System V style shared memory. */ 460 shminit(); 461 #endif 462 463 #ifdef SYSVSEM 464 /* Initialize System V style semaphores. */ 465 seminit(); 466 #endif 467 468 #ifdef SYSVMSG 469 /* Initialize System V style message queues. */ 470 msginit(); 471 #endif 472 473 #ifdef P1003_1B_SEMAPHORE 474 /* Initialize posix semaphores */ 475 ksem_init(); 476 #endif 477 478 #if NVERIEXEC > 0 479 /* 480 * Initialise the Veriexec subsystem. 481 */ 482 veriexec_init(); 483 #endif /* NVERIEXEC > 0 */ 484 485 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 486 pax_init(); 487 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 488 489 /* Attach pseudo-devices. */ 490 for (pdev = pdevinit; pdev->pdev_attach != NULL; pdev++) 491 (*pdev->pdev_attach)(pdev->pdev_count); 492 493 #ifdef FAST_IPSEC 494 /* Attach network crypto subsystem */ 495 ipsec_attach(); 496 #endif 497 498 /* 499 * Initialize protocols. Block reception of incoming packets 500 * until everything is ready. 501 */ 502 s = splnet(); 503 ifinit(); 504 domaininit(); 505 if_attachdomain(); 506 splx(s); 507 508 #ifdef GPROF 509 /* Initialize kernel profiling. */ 510 kmstartup(); 511 #endif 512 513 /* Initialize system accouting. */ 514 acct_init(); 515 516 #ifndef PIPE_SOCKETPAIR 517 /* Initialize pipes. */ 518 pipe_init(); 519 #endif 520 521 /* Setup the scheduler */ 522 sched_init(); 523 524 #ifdef KTRACE 525 /* Initialize ktrace. */ 526 ktrinit(); 527 #endif 528 529 /* Initialize the UUID system calls. */ 530 uuid_init(); 531 532 /* 533 * Create process 1 (init(8)). We do this now, as Unix has 534 * historically had init be process 1, and changing this would 535 * probably upset a lot of people. 536 * 537 * Note that process 1 won't immediately exec init(8), but will 538 * wait for us to inform it that the root file system has been 539 * mounted. 540 */ 541 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 542 panic("fork init"); 543 544 /* 545 * Now that device driver threads have been created, wait for 546 * them to finish any deferred autoconfiguration. 547 */ 548 while (config_pending) 549 (void) tsleep(&config_pending, PWAIT, "cfpend", hz); 550 551 /* 552 * Finalize configuration now that all real devices have been 553 * found. This needs to be done before the root device is 554 * selected, since finalization may create the root device. 555 */ 556 config_finalize(); 557 558 /* 559 * Now that autoconfiguration has completed, we can determine 560 * the root and dump devices. 561 */ 562 cpu_rootconf(); 563 cpu_dumpconf(); 564 565 /* Mount the root file system. */ 566 do { 567 domountroothook(); 568 if ((error = vfs_mountroot())) { 569 printf("cannot mount root, error = %d\n", error); 570 boothowto |= RB_ASKNAME; 571 setroot(root_device, 572 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 573 } 574 } while (error != 0); 575 mountroothook_destroy(); 576 577 /* 578 * Initialise the time-of-day clock, passing the time recorded 579 * in the root filesystem (if any) for use by systems that 580 * don't have a non-volatile time-of-day device. 581 */ 582 inittodr(rootfstime); 583 584 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 585 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 586 587 /* 588 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 589 * reference it. 590 */ 591 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 592 if (error) 593 panic("cannot find root vnode, error=%d", error); 594 cwdi0.cwdi_cdir = rootvnode; 595 VREF(cwdi0.cwdi_cdir); 596 VOP_UNLOCK(rootvnode, 0); 597 cwdi0.cwdi_rdir = NULL; 598 599 /* 600 * Now that root is mounted, we can fixup initproc's CWD 601 * info. All other processes are kthreads, which merely 602 * share proc0's CWD info. 603 */ 604 initproc->p_cwdi->cwdi_cdir = rootvnode; 605 VREF(initproc->p_cwdi->cwdi_cdir); 606 initproc->p_cwdi->cwdi_rdir = NULL; 607 608 /* 609 * Now can look at time, having had a chance to verify the time 610 * from the file system. Reset l->l_rtime as it may have been 611 * munched in mi_switch() after the time got set. 612 */ 613 #ifdef __HAVE_TIMECOUNTER 614 getmicrotime(&time); 615 #else 616 mono_time = time; 617 #endif 618 boottime = time; 619 mutex_enter(&proclist_lock); 620 LIST_FOREACH(p, &allproc, p_list) { 621 KASSERT((p->p_flag & PK_MARKER) == 0); 622 mutex_enter(&p->p_smutex); 623 p->p_stats->p_start = time; 624 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 625 lwp_lock(l); 626 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 627 lwp_unlock(l); 628 } 629 mutex_exit(&p->p_smutex); 630 } 631 mutex_exit(&proclist_lock); 632 binuptime(&curlwp->l_stime); 633 634 for (CPU_INFO_FOREACH(cii, ci)) { 635 ci->ci_schedstate.spc_lastmod = time_second; 636 } 637 638 /* Create the pageout daemon kernel thread. */ 639 uvm_swap_init(); 640 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 641 NULL, NULL, "pgdaemon")) 642 panic("fork pagedaemon"); 643 644 /* Create the filesystem syncer kernel thread. */ 645 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 646 NULL, NULL, "ioflush")) 647 panic("fork syncer"); 648 649 /* Create the aiodone daemon kernel thread. */ 650 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 651 uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE)) 652 panic("fork aiodoned"); 653 654 vmem_rehash_start(); 655 656 /* Initialize exec structures */ 657 exec_init(1); 658 659 /* 660 * Okay, now we can let init(8) exec! It's off to userland! 661 */ 662 start_init_exec = 1; 663 wakeup(&start_init_exec); 664 665 /* The scheduler is an infinite loop. */ 666 uvm_scheduler(); 667 /* NOTREACHED */ 668 } 669 670 static void 671 check_console(struct lwp *l) 672 { 673 struct nameidata nd; 674 int error; 675 676 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console"); 677 error = namei(&nd); 678 if (error == 0) 679 vrele(nd.ni_vp); 680 else if (error == ENOENT) 681 printf("warning: no /dev/console\n"); 682 else 683 printf("warning: lookup /dev/console: error %d\n", error); 684 } 685 686 /* 687 * List of paths to try when searching for "init". 688 */ 689 static const char * const initpaths[] = { 690 "/sbin/init", 691 "/sbin/oinit", 692 "/sbin/init.bak", 693 NULL, 694 }; 695 696 /* 697 * Start the initial user process; try exec'ing each pathname in "initpaths". 698 * The program is invoked with one argument containing the boot flags. 699 */ 700 static void 701 start_init(void *arg) 702 { 703 struct lwp *l = arg; 704 struct proc *p = l->l_proc; 705 vaddr_t addr; 706 struct sys_execve_args /* { 707 syscallarg(const char *) path; 708 syscallarg(char * const *) argp; 709 syscallarg(char * const *) envp; 710 } */ args; 711 int options, i, error; 712 register_t retval[2]; 713 char flags[4], *flagsp; 714 const char *path, *slash; 715 char *ucp, **uap, *arg0, *arg1 = NULL; 716 char ipath[129]; 717 int ipx, len; 718 719 /* 720 * Now in process 1. 721 */ 722 strncpy(p->p_comm, "init", MAXCOMLEN); 723 724 /* 725 * Wait for main() to tell us that it's safe to exec. 726 */ 727 while (start_init_exec == 0) 728 (void) tsleep(&start_init_exec, PWAIT, "initexec", 0); 729 730 /* 731 * This is not the right way to do this. We really should 732 * hand-craft a descriptor onto /dev/console to hand to init, 733 * but that's a _lot_ more work, and the benefit from this easy 734 * hack makes up for the "good is the enemy of the best" effect. 735 */ 736 check_console(l); 737 738 /* 739 * Need just enough stack to hold the faked-up "execve()" arguments. 740 */ 741 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 742 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 743 NULL, UVM_UNKNOWN_OFFSET, 0, 744 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 745 UVM_ADV_NORMAL, 746 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 747 panic("init: couldn't allocate argument space"); 748 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 749 750 ipx = 0; 751 while (1) { 752 if (boothowto & RB_ASKNAME) { 753 printf("init path"); 754 if (initpaths[ipx]) 755 printf(" (default %s)", initpaths[ipx]); 756 printf(": "); 757 len = cngetsn(ipath, sizeof(ipath)-1); 758 if (len == 0) { 759 if (initpaths[ipx]) 760 path = initpaths[ipx++]; 761 else 762 continue; 763 } else { 764 ipath[len] = '\0'; 765 path = ipath; 766 } 767 } else { 768 if ((path = initpaths[ipx++]) == NULL) 769 break; 770 } 771 772 ucp = (char *)USRSTACK; 773 774 /* 775 * Construct the boot flag argument. 776 */ 777 flagsp = flags; 778 *flagsp++ = '-'; 779 options = 0; 780 781 if (boothowto & RB_SINGLE) { 782 *flagsp++ = 's'; 783 options = 1; 784 } 785 #ifdef notyet 786 if (boothowto & RB_FASTBOOT) { 787 *flagsp++ = 'f'; 788 options = 1; 789 } 790 #endif 791 792 /* 793 * Move out the flags (arg 1), if necessary. 794 */ 795 if (options != 0) { 796 *flagsp++ = '\0'; 797 i = flagsp - flags; 798 #ifdef DEBUG 799 printf("init: copying out flags `%s' %d\n", flags, i); 800 #endif 801 arg1 = STACK_ALLOC(ucp, i); 802 ucp = STACK_MAX(arg1, i); 803 (void)copyout((void *)flags, arg1, i); 804 } 805 806 /* 807 * Move out the file name (also arg 0). 808 */ 809 i = strlen(path) + 1; 810 #ifdef DEBUG 811 printf("init: copying out path `%s' %d\n", path, i); 812 #else 813 if (boothowto & RB_ASKNAME || path != initpaths[0]) 814 printf("init: trying %s\n", path); 815 #endif 816 arg0 = STACK_ALLOC(ucp, i); 817 ucp = STACK_MAX(arg0, i); 818 (void)copyout(path, arg0, i); 819 820 /* 821 * Move out the arg pointers. 822 */ 823 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 824 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 825 SCARG(&args, path) = arg0; 826 SCARG(&args, argp) = uap; 827 SCARG(&args, envp) = NULL; 828 slash = strrchr(path, '/'); 829 if (slash) 830 (void)suword((void *)uap++, 831 (long)arg0 + (slash + 1 - path)); 832 else 833 (void)suword((void *)uap++, (long)arg0); 834 if (options != 0) 835 (void)suword((void *)uap++, (long)arg1); 836 (void)suword((void *)uap++, 0); /* terminator */ 837 838 /* 839 * Now try to exec the program. If can't for any reason 840 * other than it doesn't exist, complain. 841 */ 842 error = sys_execve(l, &args, retval); 843 if (error == 0 || error == EJUSTRETURN) { 844 KERNEL_UNLOCK_LAST(l); 845 return; 846 } 847 printf("exec %s: error %d\n", path, error); 848 } 849 printf("init: not found\n"); 850 panic("no init"); 851 } 852
Indexes created Mon Sep 22 10:09:38 GMT 2025