init_main.c revision 1.342
1 /* $NetBSD: init_main.c,v 1.342 2008/03/09 15:39:14 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 1.342 2008/03/09 15:39:14 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/pset.h> 120 #include <sys/sysctl.h> 121 #include <sys/reboot.h> 122 #include <sys/user.h> 123 #include <sys/sysctl.h> 124 #include <sys/event.h> 125 #include <sys/mbuf.h> 126 #include <sys/sched.h> 127 #include <sys/sleepq.h> 128 #include <sys/iostat.h> 129 #include <sys/vmem.h> 130 #include <sys/uuid.h> 131 #include <sys/extent.h> 132 #include <sys/disk.h> 133 #include <sys/mqueue.h> 134 #include <sys/msgbuf.h> 135 #include <sys/module.h> 136 #ifdef FAST_IPSEC 137 #include <netipsec/ipsec.h> 138 #endif 139 #ifdef SYSVSHM 140 #include <sys/shm.h> 141 #endif 142 #ifdef SYSVSEM 143 #include <sys/sem.h> 144 #endif 145 #ifdef SYSVMSG 146 #include <sys/msg.h> 147 #endif 148 #ifdef P1003_1B_SEMAPHORE 149 #include <sys/ksem.h> 150 #endif 151 #include <sys/domain.h> 152 #include <sys/namei.h> 153 #if NRND > 0 154 #include <sys/rnd.h> 155 #endif 156 #include <sys/pipe.h> 157 #ifdef LKM 158 #include <sys/lkm.h> 159 #endif 160 #if NVERIEXEC > 0 161 #include <sys/verified_exec.h> 162 #endif /* NVERIEXEC > 0 */ 163 #ifdef KTRACE 164 #include <sys/ktrace.h> 165 #endif 166 #include <sys/kauth.h> 167 #include <net80211/ieee80211_netbsd.h> 168 169 #include <sys/syscall.h> 170 #include <sys/syscallargs.h> 171 172 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 173 #include <sys/pax.h> 174 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 175 176 #include <ufs/ufs/quota.h> 177 178 #include <miscfs/genfs/genfs.h> 179 #include <miscfs/syncfs/syncfs.h> 180 181 #include <sys/cpu.h> 182 183 #include <uvm/uvm.h> 184 185 #if NSYSMON_TASKQ > 0 186 #include <dev/sysmon/sysmon_taskq.h> 187 #endif 188 189 #include <dev/cons.h> 190 191 #if NSYSMON_ENVSYS > 0 || NSYSMON_POWER > 0 || NSYSMON_WDOG > 0 192 #include <dev/sysmon/sysmonvar.h> 193 #endif 194 195 #include <net/if.h> 196 #include <net/raw_cb.h> 197 198 #include <secmodel/secmodel.h> 199 200 extern struct proc proc0; 201 extern struct lwp lwp0; 202 extern struct cwdinfo cwdi0; 203 extern time_t rootfstime; 204 205 #ifndef curlwp 206 struct lwp *curlwp = &lwp0; 207 #endif 208 struct proc *initproc; 209 210 struct vnode *rootvp, *swapdev_vp; 211 int boothowto; 212 int cold = 1; /* still working on startup */ 213 struct timeval boottime; /* time at system startup - will only follow settime deltas */ 214 215 volatile int start_init_exec; /* semaphore for start_init() */ 216 217 static void check_console(struct lwp *l); 218 static void start_init(void *); 219 void main(void); 220 221 #if defined(__SSP__) || defined(__SSP_ALL__) 222 long __stack_chk_guard[8] = {0, 0, 0, 0, 0, 0, 0, 0}; 223 void __stack_chk_fail(void); 224 225 void 226 __stack_chk_fail(void) 227 { 228 panic("stack overflow detected; terminated"); 229 } 230 #endif 231 232 void __secmodel_none(void); 233 __weak_alias(secmodel_start,__secmodel_none); 234 void 235 __secmodel_none(void) 236 { 237 return; 238 } 239 240 /* 241 * System startup; initialize the world, create process 0, mount root 242 * filesystem, and fork to create init and pagedaemon. Most of the 243 * hard work is done in the lower-level initialization routines including 244 * startup(), which does memory initialization and autoconfiguration. 245 */ 246 void 247 main(void) 248 { 249 struct timeval time; 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 /* Start module system. */ 287 module_init(); 288 289 /* Initialize callouts, part 1. */ 290 callout_startup(); 291 292 /* 293 * Initialize the kernel authorization subsystem and start the 294 * default security model, if any. We need to do this early 295 * enough so that subsystems relying on any of the aforementioned 296 * can work properly. Since the security model may dictate the 297 * credential inheritance policy, it is needed at least before 298 * any process is created, specifically proc0. 299 */ 300 kauth_init(); 301 secmodel_start(); 302 303 /* Initialize the buffer cache */ 304 bufinit(); 305 306 /* Initialize sockets. */ 307 soinit(); 308 309 /* 310 * The following things must be done before autoconfiguration. 311 */ 312 evcnt_init(); /* initialize event counters */ 313 #if NRND > 0 314 rnd_init(); /* initialize RNG */ 315 #endif 316 317 /* Initialize process and pgrp structures. */ 318 procinit(); 319 lwpinit(); 320 321 /* Initialize signal-related data structures. */ 322 signal_init(); 323 324 /* Initialize resource management. */ 325 resource_init(); 326 327 /* Create process 0 (the swapper). */ 328 proc0_init(); 329 330 /* Initialize the UID hash table. */ 331 uid_init(); 332 333 /* Charge root for one process. */ 334 (void)chgproccnt(0, 1); 335 336 /* Initialize timekeeping. */ 337 time_init(); 338 339 /* Initialize the run queues, turnstiles and sleep queues. */ 340 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 341 sched_rqinit(); 342 turnstile_init(); 343 sleeptab_init(&sleeptab); 344 345 /* Initialize processor-sets */ 346 psets_init(); 347 348 /* MI initialization of the boot cpu */ 349 error = mi_cpu_attach(curcpu()); 350 KASSERT(error == 0); 351 352 /* 353 * Initialize mbuf's. Do this now because we might attempt to 354 * allocate mbufs or mbuf clusters during autoconfiguration. 355 */ 356 mbinit(); 357 358 /* Initialize the sysctl subsystem. */ 359 sysctl_init(); 360 361 /* Initialize I/O statistics. */ 362 iostat_init(); 363 364 /* Initialize the log device. */ 365 loginit(); 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 file descriptor system. */ 384 filedesc_init(); 385 386 /* Initialize the select()/poll() system calls. */ 387 selsysinit(); 388 389 /* Initialize asynchronous I/O. */ 390 aio_sysinit(); 391 392 /* Initialize message queues. */ 393 mqueue_sysinit(); 394 395 /* Initialize the system monitor subsystems. */ 396 #if NSYSMON_TASKQ > 0 397 sysmon_task_queue_preinit(); 398 #endif 399 400 #if NSYSMON_ENVSYS > 0 401 sysmon_envsys_init(); 402 #endif 403 404 #if NSYSMON_POWER > 0 405 sysmon_power_init(); 406 #endif 407 408 #if NSYSMON_WDOG > 0 409 sysmon_wdog_init(); 410 #endif 411 412 inittimecounter(); 413 ntp_init(); 414 415 /* Initialize the device switch tables. */ 416 devsw_init(); 417 418 /* Initialize tty subsystem. */ 419 tty_init(); 420 ttyldisc_init(); 421 422 /* Initialize the buffer cache, part 2. */ 423 bufinit2(); 424 425 /* Initialize the disk wedge subsystem. */ 426 dkwedge_init(); 427 428 /* Configure the system hardware. This will enable interrupts. */ 429 configure(); 430 431 #if defined(__SSP__) || defined(__SSP_ALL__) 432 { 433 #ifdef DIAGNOSTIC 434 printf("Initializing SSP:"); 435 #endif 436 /* 437 * We initialize ssp here carefully: 438 * 1. after we got some entropy 439 * 2. without calling a function 440 */ 441 size_t i; 442 long guard[__arraycount(__stack_chk_guard)]; 443 444 arc4randbytes(guard, sizeof(guard)); 445 for (i = 0; i < __arraycount(guard); i++) 446 __stack_chk_guard[i] = guard[i]; 447 #ifdef DIAGNOSTIC 448 for (i = 0; i < __arraycount(guard); i++) 449 printf("%lx ", guard[i]); 450 printf("\n"); 451 #endif 452 } 453 #endif 454 ubc_init(); /* must be after autoconfig */ 455 456 /* Lock the kernel on behalf of proc0. */ 457 KERNEL_LOCK(1, l); 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) || defined(PAX_ASLR) 487 pax_init(); 488 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 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. 548 */ 549 while (config_pending) 550 (void) tsleep(&config_pending, PWAIT, "cfpend", hz); 551 552 /* 553 * Load any remaining builtin modules, and hand back temporary 554 * storage to the VM system. 555 */ 556 module_init_class(MODULE_CLASS_ANY); 557 module_jettison(); 558 559 /* 560 * Finalize configuration now that all real devices have been 561 * found. This needs to be done before the root device is 562 * selected, since finalization may create the root device. 563 */ 564 config_finalize(); 565 566 /* 567 * Now that autoconfiguration has completed, we can determine 568 * the root and dump devices. 569 */ 570 cpu_rootconf(); 571 cpu_dumpconf(); 572 573 /* Mount the root file system. */ 574 do { 575 domountroothook(); 576 if ((error = vfs_mountroot())) { 577 printf("cannot mount root, error = %d\n", error); 578 boothowto |= RB_ASKNAME; 579 setroot(root_device, 580 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 581 } 582 } while (error != 0); 583 mountroothook_destroy(); 584 585 /* 586 * Initialise the time-of-day clock, passing the time recorded 587 * in the root filesystem (if any) for use by systems that 588 * don't have a non-volatile time-of-day device. 589 */ 590 inittodr(rootfstime); 591 592 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 593 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 594 595 /* 596 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 597 * reference it. 598 */ 599 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 600 if (error) 601 panic("cannot find root vnode, error=%d", error); 602 cwdi0.cwdi_cdir = rootvnode; 603 VREF(cwdi0.cwdi_cdir); 604 VOP_UNLOCK(rootvnode, 0); 605 cwdi0.cwdi_rdir = NULL; 606 607 /* 608 * Now that root is mounted, we can fixup initproc's CWD 609 * info. All other processes are kthreads, which merely 610 * share proc0's CWD info. 611 */ 612 initproc->p_cwdi->cwdi_cdir = rootvnode; 613 VREF(initproc->p_cwdi->cwdi_cdir); 614 initproc->p_cwdi->cwdi_rdir = NULL; 615 616 /* 617 * Now can look at time, having had a chance to verify the time 618 * from the file system. Reset l->l_rtime as it may have been 619 * munched in mi_switch() after the time got set. 620 */ 621 getmicrotime(&time); 622 boottime = time; 623 mutex_enter(&proclist_lock); 624 LIST_FOREACH(p, &allproc, p_list) { 625 KASSERT((p->p_flag & PK_MARKER) == 0); 626 mutex_enter(&p->p_smutex); 627 p->p_stats->p_start = time; 628 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 629 lwp_lock(l); 630 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 631 lwp_unlock(l); 632 } 633 mutex_exit(&p->p_smutex); 634 } 635 mutex_exit(&proclist_lock); 636 binuptime(&curlwp->l_stime); 637 638 for (CPU_INFO_FOREACH(cii, ci)) { 639 ci->ci_schedstate.spc_lastmod = time_second; 640 } 641 642 /* Create the pageout daemon kernel thread. */ 643 uvm_swap_init(); 644 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 645 NULL, NULL, "pgdaemon")) 646 panic("fork pagedaemon"); 647 648 /* Create the filesystem syncer kernel thread. */ 649 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 650 NULL, NULL, "ioflush")) 651 panic("fork syncer"); 652 653 /* Create the aiodone daemon kernel thread. */ 654 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 655 uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE)) 656 panic("fork aiodoned"); 657 658 vmem_rehash_start(); 659 660 /* Initialize exec structures */ 661 exec_init(1); 662 663 /* 664 * Okay, now we can let init(8) exec! It's off to userland! 665 */ 666 start_init_exec = 1; 667 wakeup(&start_init_exec); 668 669 /* The scheduler is an infinite loop. */ 670 uvm_scheduler(); 671 /* NOTREACHED */ 672 } 673 674 static void 675 check_console(struct lwp *l) 676 { 677 struct nameidata nd; 678 int error; 679 680 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console"); 681 error = namei(&nd); 682 if (error == 0) 683 vrele(nd.ni_vp); 684 else if (error == ENOENT) 685 printf("warning: no /dev/console\n"); 686 else 687 printf("warning: lookup /dev/console: error %d\n", error); 688 } 689 690 /* 691 * List of paths to try when searching for "init". 692 */ 693 static const char * const initpaths[] = { 694 "/sbin/init", 695 "/sbin/oinit", 696 "/sbin/init.bak", 697 NULL, 698 }; 699 700 /* 701 * Start the initial user process; try exec'ing each pathname in "initpaths". 702 * The program is invoked with one argument containing the boot flags. 703 */ 704 static void 705 start_init(void *arg) 706 { 707 struct lwp *l = arg; 708 struct proc *p = l->l_proc; 709 vaddr_t addr; 710 struct sys_execve_args /* { 711 syscallarg(const char *) path; 712 syscallarg(char * const *) argp; 713 syscallarg(char * const *) envp; 714 } */ args; 715 int options, i, error; 716 register_t retval[2]; 717 char flags[4], *flagsp; 718 const char *path, *slash; 719 char *ucp, **uap, *arg0, *arg1 = NULL; 720 char ipath[129]; 721 int ipx, len; 722 723 /* 724 * Now in process 1. 725 */ 726 strncpy(p->p_comm, "init", MAXCOMLEN); 727 728 /* 729 * Wait for main() to tell us that it's safe to exec. 730 */ 731 while (start_init_exec == 0) 732 (void) tsleep(&start_init_exec, PWAIT, "initexec", 0); 733 734 /* 735 * This is not the right way to do this. We really should 736 * hand-craft a descriptor onto /dev/console to hand to init, 737 * but that's a _lot_ more work, and the benefit from this easy 738 * hack makes up for the "good is the enemy of the best" effect. 739 */ 740 check_console(l); 741 742 /* 743 * Need just enough stack to hold the faked-up "execve()" arguments. 744 */ 745 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 746 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 747 NULL, UVM_UNKNOWN_OFFSET, 0, 748 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 749 UVM_ADV_NORMAL, 750 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 751 panic("init: couldn't allocate argument space"); 752 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 753 754 ipx = 0; 755 while (1) { 756 if (boothowto & RB_ASKNAME) { 757 printf("init path"); 758 if (initpaths[ipx]) 759 printf(" (default %s)", initpaths[ipx]); 760 printf(": "); 761 len = cngetsn(ipath, sizeof(ipath)-1); 762 if (len == 0) { 763 if (initpaths[ipx]) 764 path = initpaths[ipx++]; 765 else 766 continue; 767 } else { 768 ipath[len] = '\0'; 769 path = ipath; 770 } 771 } else { 772 if ((path = initpaths[ipx++]) == NULL) 773 break; 774 } 775 776 ucp = (char *)USRSTACK; 777 778 /* 779 * Construct the boot flag argument. 780 */ 781 flagsp = flags; 782 *flagsp++ = '-'; 783 options = 0; 784 785 if (boothowto & RB_SINGLE) { 786 *flagsp++ = 's'; 787 options = 1; 788 } 789 #ifdef notyet 790 if (boothowto & RB_FASTBOOT) { 791 *flagsp++ = 'f'; 792 options = 1; 793 } 794 #endif 795 796 /* 797 * Move out the flags (arg 1), if necessary. 798 */ 799 if (options != 0) { 800 *flagsp++ = '\0'; 801 i = flagsp - flags; 802 #ifdef DEBUG 803 printf("init: copying out flags `%s' %d\n", flags, i); 804 #endif 805 arg1 = STACK_ALLOC(ucp, i); 806 ucp = STACK_MAX(arg1, i); 807 (void)copyout((void *)flags, arg1, i); 808 } 809 810 /* 811 * Move out the file name (also arg 0). 812 */ 813 i = strlen(path) + 1; 814 #ifdef DEBUG 815 printf("init: copying out path `%s' %d\n", path, i); 816 #else 817 if (boothowto & RB_ASKNAME || path != initpaths[0]) 818 printf("init: trying %s\n", path); 819 #endif 820 arg0 = STACK_ALLOC(ucp, i); 821 ucp = STACK_MAX(arg0, i); 822 (void)copyout(path, arg0, i); 823 824 /* 825 * Move out the arg pointers. 826 */ 827 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 828 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 829 SCARG(&args, path) = arg0; 830 SCARG(&args, argp) = uap; 831 SCARG(&args, envp) = NULL; 832 slash = strrchr(path, '/'); 833 if (slash) 834 (void)suword((void *)uap++, 835 (long)arg0 + (slash + 1 - path)); 836 else 837 (void)suword((void *)uap++, (long)arg0); 838 if (options != 0) 839 (void)suword((void *)uap++, (long)arg1); 840 (void)suword((void *)uap++, 0); /* terminator */ 841 842 /* 843 * Now try to exec the program. If can't for any reason 844 * other than it doesn't exist, complain. 845 */ 846 error = sys_execve(l, &args, retval); 847 if (error == 0 || error == EJUSTRETURN) { 848 KERNEL_UNLOCK_LAST(l); 849 return; 850 } 851 printf("exec %s: error %d\n", path, error); 852 } 853 printf("init: not found\n"); 854 panic("no init"); 855 } 856
Indexes created Sun Sep 21 20:09:37 GMT 2025