1 /* $NetBSD: init_main.c,v 1.396 2009/07/27 17:40:57 mbalmer Exp $ */ 2 3 /*- 4 * Copyright (c) 2008, 2009 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 17 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 18 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 19 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 20 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 21 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 22 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 23 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 24 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 25 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 26 * POSSIBILITY OF SUCH DAMAGE. 27 */ 28 29 /* 30 * Copyright (c) 1982, 1986, 1989, 1991, 1992, 1993 31 * The Regents of the University of California. All rights reserved. 32 * (c) UNIX System Laboratories, Inc. 33 * All or some portions of this file are derived from material licensed 34 * to the University of California by American Telephone and Telegraph 35 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 36 * the permission of UNIX System Laboratories, Inc. 37 * 38 * Redistribution and use in source and binary forms, with or without 39 * modification, are permitted provided that the following conditions 40 * are met: 41 * 1. Redistributions of source code must retain the above copyright 42 * notice, this list of conditions and the following disclaimer. 43 * 2. Redistributions in binary form must reproduce the above copyright 44 * notice, this list of conditions and the following disclaimer in the 45 * documentation and/or other materials provided with the distribution. 46 * 3. Neither the name of the University nor the names of its contributors 47 * may be used to endorse or promote products derived from this software 48 * without specific prior written permission. 49 * 50 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 51 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 52 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 53 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 54 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 55 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 56 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 57 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 58 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 59 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 60 * SUCH DAMAGE. 61 * 62 * @(#)init_main.c 8.16 (Berkeley) 5/14/95 63 */ 64 65 /* 66 * Copyright (c) 1995 Christopher G. Demetriou. All rights reserved. 67 * 68 * Redistribution and use in source and binary forms, with or without 69 * modification, are permitted provided that the following conditions 70 * are met: 71 * 1. Redistributions of source code must retain the above copyright 72 * notice, this list of conditions and the following disclaimer. 73 * 2. Redistributions in binary form must reproduce the above copyright 74 * notice, this list of conditions and the following disclaimer in the 75 * documentation and/or other materials provided with the distribution. 76 * 3. All advertising materials mentioning features or use of this software 77 * must display the following acknowledgement: 78 * This product includes software developed by the University of 79 * California, Berkeley and its contributors. 80 * 4. Neither the name of the University nor the names of its contributors 81 * may be used to endorse or promote products derived from this software 82 * without specific prior written permission. 83 * 84 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 85 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 86 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 87 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 88 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 89 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 90 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 91 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 92 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 93 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 94 * SUCH DAMAGE. 95 * 96 * @(#)init_main.c 8.16 (Berkeley) 5/14/95 97 */ 98 99 #include <sys/cdefs.h> 100 __KERNEL_RCSID(0, "$NetBSD: init_main.c,v 1.396 2009/07/27 17:40:57 mbalmer Exp $"); 101 102 #include "opt_ddb.h" 103 #include "opt_ipsec.h" 104 #include "opt_modular.h" 105 #include "opt_ntp.h" 106 #include "opt_pipe.h" 107 #include "opt_syscall_debug.h" 108 #include "opt_sysv.h" 109 #include "opt_fileassoc.h" 110 #include "opt_ktrace.h" 111 #include "opt_pax.h" 112 #include "opt_compat_netbsd.h" 113 #include "opt_wapbl.h" 114 115 #include "ksyms.h" 116 #include "rnd.h" 117 #include "sysmon_envsys.h" 118 #include "sysmon_power.h" 119 #include "sysmon_taskq.h" 120 #include "sysmon_wdog.h" 121 #include "veriexec.h" 122 123 #include <sys/param.h> 124 #include <sys/acct.h> 125 #include <sys/filedesc.h> 126 #include <sys/file.h> 127 #include <sys/errno.h> 128 #include <sys/callout.h> 129 #include <sys/cpu.h> 130 #include <sys/kernel.h> 131 #include <sys/mount.h> 132 #include <sys/proc.h> 133 #include <sys/kthread.h> 134 #include <sys/resourcevar.h> 135 #include <sys/signalvar.h> 136 #include <sys/systm.h> 137 #include <sys/vnode.h> 138 #include <sys/fstrans.h> 139 #include <sys/tty.h> 140 #include <sys/conf.h> 141 #include <sys/disklabel.h> 142 #include <sys/buf.h> 143 #include <sys/device.h> 144 #include <sys/exec.h> 145 #include <sys/socketvar.h> 146 #include <sys/protosw.h> 147 #include <sys/percpu.h> 148 #include <sys/pset.h> 149 #include <sys/sysctl.h> 150 #include <sys/reboot.h> 151 #include <sys/user.h> 152 #include <sys/sysctl.h> 153 #include <sys/event.h> 154 #include <sys/mbuf.h> 155 #include <sys/sched.h> 156 #include <sys/sleepq.h> 157 #include <sys/iostat.h> 158 #include <sys/vmem.h> 159 #include <sys/uuid.h> 160 #include <sys/extent.h> 161 #include <sys/disk.h> 162 #include <sys/msgbuf.h> 163 #include <sys/module.h> 164 #include <sys/event.h> 165 #include <sys/lockf.h> 166 #include <sys/once.h> 167 #include <sys/ksyms.h> 168 #include <sys/uidinfo.h> 169 #include <sys/kprintf.h> 170 #ifdef FAST_IPSEC 171 #include <netipsec/ipsec.h> 172 #endif 173 #ifdef SYSVSHM 174 #include <sys/shm.h> 175 #endif 176 #ifdef SYSVSEM 177 #include <sys/sem.h> 178 #endif 179 #ifdef SYSVMSG 180 #include <sys/msg.h> 181 #endif 182 #include <sys/domain.h> 183 #include <sys/namei.h> 184 #if NRND > 0 185 #include <sys/rnd.h> 186 #endif 187 #include <sys/pipe.h> 188 #if NVERIEXEC > 0 189 #include <sys/verified_exec.h> 190 #endif /* NVERIEXEC > 0 */ 191 #ifdef KTRACE 192 #include <sys/ktrace.h> 193 #endif 194 #include <sys/kauth.h> 195 #ifdef WAPBL 196 #include <sys/wapbl.h> 197 #endif 198 #include <net80211/ieee80211_netbsd.h> 199 200 #include <sys/syscall.h> 201 #include <sys/syscallargs.h> 202 203 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 204 #include <sys/pax.h> 205 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 206 207 #include <ufs/ufs/quota.h> 208 209 #include <miscfs/genfs/genfs.h> 210 #include <miscfs/syncfs/syncfs.h> 211 212 #include <sys/cpu.h> 213 214 #include <uvm/uvm.h> 215 216 #if NSYSMON_TASKQ > 0 217 #include <dev/sysmon/sysmon_taskq.h> 218 #endif 219 220 #include <dev/cons.h> 221 222 #if NSYSMON_ENVSYS > 0 || NSYSMON_POWER > 0 || NSYSMON_WDOG > 0 223 #include <dev/sysmon/sysmonvar.h> 224 #endif 225 226 #include <net/if.h> 227 #include <net/raw_cb.h> 228 229 #include <secmodel/secmodel.h> 230 231 #ifdef COMPAT_50 232 #include <compat/sys/time.h> 233 struct timeval50 boottime50; 234 #endif 235 236 extern struct proc proc0; 237 extern struct lwp lwp0; 238 extern struct cwdinfo cwdi0; 239 extern time_t rootfstime; 240 241 #ifndef curlwp 242 struct lwp *curlwp = &lwp0; 243 #endif 244 struct proc *initproc; 245 246 struct vnode *rootvp, *swapdev_vp; 247 int boothowto; 248 int cold = 1; /* still working on startup */ 249 struct timespec boottime; /* time at system startup - will only follow settime deltas */ 250 251 int start_init_exec; /* semaphore for start_init() */ 252 253 static void check_console(struct lwp *l); 254 static void start_init(void *); 255 void main(void); 256 257 void __secmodel_none(void); 258 __weak_alias(secmodel_start,__secmodel_none); 259 void 260 __secmodel_none(void) 261 { 262 return; 263 } 264 265 /* 266 * System startup; initialize the world, create process 0, mount root 267 * filesystem, and fork to create init and pagedaemon. Most of the 268 * hard work is done in the lower-level initialization routines including 269 * startup(), which does memory initialization and autoconfiguration. 270 */ 271 void 272 main(void) 273 { 274 struct timespec time; 275 struct lwp *l; 276 struct proc *p; 277 int s, error; 278 #ifdef NVNODE_IMPLICIT 279 int usevnodes; 280 #endif 281 CPU_INFO_ITERATOR cii; 282 struct cpu_info *ci; 283 284 l = &lwp0; 285 #ifndef LWP0_CPU_INFO 286 l->l_cpu = curcpu(); 287 #endif 288 l->l_pflag |= LP_RUNNING; 289 290 /* 291 * Attempt to find console and initialize 292 * in case of early panic or other messages. 293 */ 294 consinit(); 295 296 kernel_lock_init(); 297 once_init(); 298 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 299 300 uvm_init(); 301 302 #if ((NKSYMS > 0) || (NDDB > 0) || (NMODULAR > 0)) 303 ksyms_init(); 304 #endif 305 kprintf_init(); 306 307 percpu_init(); 308 309 /* Initialize lock caches. */ 310 mutex_obj_init(); 311 rw_obj_init(); 312 313 /* Initialize the extent manager. */ 314 extent_init(); 315 316 /* Do machine-dependent initialization. */ 317 cpu_startup(); 318 319 /* Initialize the sysctl subsystem. */ 320 sysctl_init(); 321 322 /* Initialize callouts, part 1. */ 323 callout_startup(); 324 325 /* 326 * Initialize the kernel authorization subsystem and start the 327 * default security model, if any. We need to do this early 328 * enough so that subsystems relying on any of the aforementioned 329 * can work properly. Since the security model may dictate the 330 * credential inheritance policy, it is needed at least before 331 * any process is created, specifically proc0. 332 */ 333 kauth_init(); 334 secmodel_start(); 335 336 /* Initialize the buffer cache */ 337 bufinit(); 338 339 /* Initialize sockets. */ 340 soinit(); 341 342 /* 343 * The following things must be done before autoconfiguration. 344 */ 345 evcnt_init(); /* initialize event counters */ 346 #if NRND > 0 347 rnd_init(); /* initialize random number generator */ 348 #endif 349 350 /* Initialize process and pgrp structures. */ 351 procinit(); 352 lwpinit(); 353 354 /* Initialize signal-related data structures. */ 355 signal_init(); 356 357 /* Initialize resource management. */ 358 resource_init(); 359 360 /* Create process 0 (the swapper). */ 361 proc0_init(); 362 363 /* Disable preemption during boot. */ 364 kpreempt_disable(); 365 366 /* Initialize the UID hash table. */ 367 uid_init(); 368 369 /* Charge root for one process. */ 370 (void)chgproccnt(0, 1); 371 372 /* Initialize timekeeping. */ 373 time_init(); 374 375 /* Initialize the run queues, turnstiles and sleep queues. */ 376 sched_rqinit(); 377 turnstile_init(); 378 sleeptab_init(&sleeptab); 379 380 /* Initialize processor-sets */ 381 psets_init(); 382 383 /* MI initialization of the boot cpu */ 384 error = mi_cpu_attach(curcpu()); 385 KASSERT(error == 0); 386 387 /* Initialize timekeeping, part 2. */ 388 time_init2(); 389 390 /* 391 * Initialize mbuf's. Do this now because we might attempt to 392 * allocate mbufs or mbuf clusters during autoconfiguration. 393 */ 394 mbinit(); 395 396 /* Initialize I/O statistics. */ 397 iostat_init(); 398 399 /* Initialize the log device. */ 400 loginit(); 401 402 /* Start module system. */ 403 module_init(); 404 405 /* Initialize the file systems. */ 406 #ifdef NVNODE_IMPLICIT 407 /* 408 * If maximum number of vnodes in namei vnode cache is not explicitly 409 * defined in kernel config, adjust the number such as we use roughly 410 * 10% of memory for vnodes and associated data structures in the 411 * assumed worst case. Do not provide fewer than NVNODE vnodes. 412 */ 413 usevnodes = 414 calc_cache_size(kernel_map, 10, VNODE_VA_MAXPCT) / VNODE_COST; 415 if (usevnodes > desiredvnodes) 416 desiredvnodes = usevnodes; 417 #endif 418 vfsinit(); 419 lf_init(); 420 421 /* Initialize fstrans. */ 422 fstrans_init(); 423 424 /* Initialize the file descriptor system. */ 425 fd_sys_init(); 426 427 /* Initialize cwd structures */ 428 cwd_sys_init(); 429 430 /* Initialize kqueue. */ 431 kqueue_init(); 432 433 /* Initialize the system monitor subsystems. */ 434 #if NSYSMON_TASKQ > 0 435 sysmon_task_queue_preinit(); 436 #endif 437 438 #if NSYSMON_ENVSYS > 0 439 sysmon_envsys_init(); 440 #endif 441 442 #if NSYSMON_POWER > 0 443 sysmon_power_init(); 444 #endif 445 446 #if NSYSMON_WDOG > 0 447 sysmon_wdog_init(); 448 #endif 449 450 inittimecounter(); 451 ntp_init(); 452 453 /* Initialize the device switch tables. */ 454 devsw_init(); 455 456 /* Initialize tty subsystem. */ 457 tty_init(); 458 ttyldisc_init(); 459 460 /* Initialize the buffer cache, part 2. */ 461 bufinit2(); 462 463 /* Initialize the disk wedge subsystem. */ 464 dkwedge_init(); 465 466 /* Initialize interfaces. */ 467 ifinit1(); 468 469 /* Configure the system hardware. This will enable interrupts. */ 470 configure(); 471 472 ssp_init(); 473 474 configure2(); 475 /* Now timer is working. Enable preemption. */ 476 kpreempt_enable(); 477 478 ubc_init(); /* must be after autoconfig */ 479 480 #ifdef SYSVSHM 481 /* Initialize System V style shared memory. */ 482 shminit(); 483 #endif 484 485 #ifdef SYSVSEM 486 /* Initialize System V style semaphores. */ 487 seminit(); 488 #endif 489 490 #ifdef SYSVMSG 491 /* Initialize System V style message queues. */ 492 msginit(); 493 #endif 494 495 #if NVERIEXEC > 0 496 /* 497 * Initialise the Veriexec subsystem. 498 */ 499 veriexec_init(); 500 #endif /* NVERIEXEC > 0 */ 501 502 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 503 pax_init(); 504 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 505 506 #ifdef FAST_IPSEC 507 /* Attach network crypto subsystem */ 508 ipsec_attach(); 509 #endif 510 511 /* 512 * Initialize protocols. Block reception of incoming packets 513 * until everything is ready. 514 */ 515 s = splnet(); 516 ifinit(); 517 domaininit(true); 518 if_attachdomain(); 519 splx(s); 520 521 #ifdef GPROF 522 /* Initialize kernel profiling. */ 523 kmstartup(); 524 #endif 525 526 /* Initialize system accounting. */ 527 acct_init(); 528 529 #ifndef PIPE_SOCKETPAIR 530 /* Initialize pipes. */ 531 pipe_init(); 532 #endif 533 534 #ifdef KTRACE 535 /* Initialize ktrace. */ 536 ktrinit(); 537 #endif 538 539 /* Initialize the UUID system calls. */ 540 uuid_init(); 541 542 #ifdef WAPBL 543 /* Initialize write-ahead physical block logging. */ 544 wapbl_init(); 545 #endif 546 547 /* 548 * Create process 1 (init(8)). We do this now, as Unix has 549 * historically had init be process 1, and changing this would 550 * probably upset a lot of people. 551 * 552 * Note that process 1 won't immediately exec init(8), but will 553 * wait for us to inform it that the root file system has been 554 * mounted. 555 */ 556 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 557 panic("fork init"); 558 559 /* 560 * Load any remaining builtin modules, and hand back temporary 561 * storage to the VM system. 562 */ 563 module_init_class(MODULE_CLASS_ANY); 564 565 /* 566 * Finalize configuration now that all real devices have been 567 * found. This needs to be done before the root device is 568 * selected, since finalization may create the root device. 569 */ 570 config_finalize(); 571 572 /* 573 * Now that autoconfiguration has completed, we can determine 574 * the root and dump devices. 575 */ 576 cpu_rootconf(); 577 cpu_dumpconf(); 578 579 /* Mount the root file system. */ 580 do { 581 domountroothook(); 582 if ((error = vfs_mountroot())) { 583 printf("cannot mount root, error = %d\n", error); 584 boothowto |= RB_ASKNAME; 585 setroot(root_device, 586 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 587 } 588 } while (error != 0); 589 mountroothook_destroy(); 590 591 /* 592 * Initialise the time-of-day clock, passing the time recorded 593 * in the root filesystem (if any) for use by systems that 594 * don't have a non-volatile time-of-day device. 595 */ 596 inittodr(rootfstime); 597 598 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 599 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 600 601 /* 602 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 603 * reference it. 604 */ 605 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 606 if (error) 607 panic("cannot find root vnode, error=%d", error); 608 cwdi0.cwdi_cdir = rootvnode; 609 VREF(cwdi0.cwdi_cdir); 610 VOP_UNLOCK(rootvnode, 0); 611 cwdi0.cwdi_rdir = NULL; 612 613 /* 614 * Now that root is mounted, we can fixup initproc's CWD 615 * info. All other processes are kthreads, which merely 616 * share proc0's CWD info. 617 */ 618 initproc->p_cwdi->cwdi_cdir = rootvnode; 619 VREF(initproc->p_cwdi->cwdi_cdir); 620 initproc->p_cwdi->cwdi_rdir = NULL; 621 622 /* 623 * Now can look at time, having had a chance to verify the time 624 * from the file system. Reset l->l_rtime as it may have been 625 * munched in mi_switch() after the time got set. 626 */ 627 getnanotime(&time); 628 boottime = time; 629 #ifdef COMPAT_50 630 { 631 struct timeval tv; 632 TIMESPEC_TO_TIMEVAL(&tv, &time); 633 timeval_to_timeval50(&tv, &boottime50); 634 } 635 #endif 636 mutex_enter(proc_lock); 637 LIST_FOREACH(p, &allproc, p_list) { 638 KASSERT((p->p_flag & PK_MARKER) == 0); 639 mutex_enter(p->p_lock); 640 TIMESPEC_TO_TIMEVAL(&p->p_stats->p_start, &time); 641 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 642 lwp_lock(l); 643 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 644 lwp_unlock(l); 645 } 646 mutex_exit(p->p_lock); 647 } 648 mutex_exit(proc_lock); 649 binuptime(&curlwp->l_stime); 650 651 for (CPU_INFO_FOREACH(cii, ci)) { 652 ci->ci_schedstate.spc_lastmod = time_second; 653 } 654 655 /* Create the pageout daemon kernel thread. */ 656 uvm_swap_init(); 657 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 658 NULL, NULL, "pgdaemon")) 659 panic("fork pagedaemon"); 660 661 /* Create the filesystem syncer kernel thread. */ 662 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 663 NULL, NULL, "ioflush")) 664 panic("fork syncer"); 665 666 /* Create the aiodone daemon kernel thread. */ 667 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 668 uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE)) 669 panic("fork aiodoned"); 670 671 vmem_rehash_start(); 672 673 /* Initialize exec structures */ 674 exec_init(1); 675 676 /* 677 * Okay, now we can let init(8) exec! It's off to userland! 678 */ 679 mutex_enter(proc_lock); 680 start_init_exec = 1; 681 cv_broadcast(&lbolt); 682 mutex_exit(proc_lock); 683 684 /* The scheduler is an infinite loop. */ 685 uvm_scheduler(); 686 /* NOTREACHED */ 687 } 688 689 static void 690 check_console(struct lwp *l) 691 { 692 struct vnode *vp; 693 int error; 694 695 error = namei_simple_kernel("/dev/console", 696 NSM_FOLLOW_NOEMULROOT, &vp); 697 if (error == 0) 698 vrele(vp); 699 else if (error == ENOENT) 700 printf("warning: no /dev/console\n"); 701 else 702 printf("warning: lookup /dev/console: error %d\n", error); 703 } 704 705 /* 706 * List of paths to try when searching for "init". 707 */ 708 static const char * const initpaths[] = { 709 "/sbin/init", 710 "/sbin/oinit", 711 "/sbin/init.bak", 712 NULL, 713 }; 714 715 /* 716 * Start the initial user process; try exec'ing each pathname in "initpaths". 717 * The program is invoked with one argument containing the boot flags. 718 */ 719 static void 720 start_init(void *arg) 721 { 722 struct lwp *l = arg; 723 struct proc *p = l->l_proc; 724 vaddr_t addr; 725 struct sys_execve_args /* { 726 syscallarg(const char *) path; 727 syscallarg(char * const *) argp; 728 syscallarg(char * const *) envp; 729 } */ args; 730 int options, i, error; 731 register_t retval[2]; 732 char flags[4], *flagsp; 733 const char *path, *slash; 734 char *ucp, **uap, *arg0, *arg1 = NULL; 735 char ipath[129]; 736 int ipx, len; 737 738 /* 739 * Now in process 1. 740 */ 741 strncpy(p->p_comm, "init", MAXCOMLEN); 742 743 /* 744 * Wait for main() to tell us that it's safe to exec. 745 */ 746 mutex_enter(proc_lock); 747 while (start_init_exec == 0) 748 cv_wait(&lbolt, proc_lock); 749 mutex_exit(proc_lock); 750 751 /* 752 * This is not the right way to do this. We really should 753 * hand-craft a descriptor onto /dev/console to hand to init, 754 * but that's a _lot_ more work, and the benefit from this easy 755 * hack makes up for the "good is the enemy of the best" effect. 756 */ 757 check_console(l); 758 759 /* 760 * Need just enough stack to hold the faked-up "execve()" arguments. 761 */ 762 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 763 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 764 NULL, UVM_UNKNOWN_OFFSET, 0, 765 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 766 UVM_ADV_NORMAL, 767 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 768 panic("init: couldn't allocate argument space"); 769 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 770 771 ipx = 0; 772 while (1) { 773 if (boothowto & RB_ASKNAME) { 774 printf("init path"); 775 if (initpaths[ipx]) 776 printf(" (default %s)", initpaths[ipx]); 777 printf(": "); 778 len = cngetsn(ipath, sizeof(ipath)-1); 779 if (len == 4 && strcmp(ipath, "halt") == 0) { 780 cpu_reboot(RB_HALT, NULL); 781 } else if (len == 6 && strcmp(ipath, "reboot") == 0) { 782 cpu_reboot(0, NULL); 783 #if defined(DDB) 784 } else if (len == 3 && strcmp(ipath, "ddb") == 0) { 785 console_debugger(); 786 continue; 787 #endif 788 } else if (len > 0 && ipath[0] == '/') { 789 ipath[len] = '\0'; 790 path = ipath; 791 } else if (len == 0 && initpaths[ipx] != NULL) { 792 path = initpaths[ipx++]; 793 } else { 794 printf("use absolute path, "); 795 #if defined(DDB) 796 printf("\"ddb\", "); 797 #endif 798 printf("\"halt\", or \"reboot\"\n"); 799 continue; 800 } 801 } else { 802 if ((path = initpaths[ipx++]) == NULL) { 803 ipx = 0; 804 boothowto |= RB_ASKNAME; 805 continue; 806 } 807 } 808 809 ucp = (char *)USRSTACK; 810 811 /* 812 * Construct the boot flag argument. 813 */ 814 flagsp = flags; 815 *flagsp++ = '-'; 816 options = 0; 817 818 if (boothowto & RB_SINGLE) { 819 *flagsp++ = 's'; 820 options = 1; 821 } 822 #ifdef notyet 823 if (boothowto & RB_FASTBOOT) { 824 *flagsp++ = 'f'; 825 options = 1; 826 } 827 #endif 828 829 /* 830 * Move out the flags (arg 1), if necessary. 831 */ 832 if (options != 0) { 833 *flagsp++ = '\0'; 834 i = flagsp - flags; 835 #ifdef DEBUG 836 printf("init: copying out flags `%s' %d\n", flags, i); 837 #endif 838 arg1 = STACK_ALLOC(ucp, i); 839 ucp = STACK_MAX(arg1, i); 840 (void)copyout((void *)flags, arg1, i); 841 } 842 843 /* 844 * Move out the file name (also arg 0). 845 */ 846 i = strlen(path) + 1; 847 #ifdef DEBUG 848 printf("init: copying out path `%s' %d\n", path, i); 849 #else 850 if (boothowto & RB_ASKNAME || path != initpaths[0]) 851 printf("init: trying %s\n", path); 852 #endif 853 arg0 = STACK_ALLOC(ucp, i); 854 ucp = STACK_MAX(arg0, i); 855 (void)copyout(path, arg0, i); 856 857 /* 858 * Move out the arg pointers. 859 */ 860 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 861 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 862 SCARG(&args, path) = arg0; 863 SCARG(&args, argp) = uap; 864 SCARG(&args, envp) = NULL; 865 slash = strrchr(path, '/'); 866 if (slash) 867 (void)suword((void *)uap++, 868 (long)arg0 + (slash + 1 - path)); 869 else 870 (void)suword((void *)uap++, (long)arg0); 871 if (options != 0) 872 (void)suword((void *)uap++, (long)arg1); 873 (void)suword((void *)uap++, 0); /* terminator */ 874 875 /* 876 * Now try to exec the program. If can't for any reason 877 * other than it doesn't exist, complain. 878 */ 879 error = sys_execve(l, &args, retval); 880 if (error == 0 || error == EJUSTRETURN) { 881 KERNEL_UNLOCK_LAST(l); 882 return; 883 } 884 printf("exec %s: error %d\n", path, error); 885 } 886 printf("init: not found\n"); 887 panic("no init"); 888 } 889 890 /* 891 * calculate cache size from physmem and vm_map size. 892 */ 893 vaddr_t 894 calc_cache_size(struct vm_map *map, int pct, int va_pct) 895 { 896 paddr_t t; 897 898 /* XXX should consider competing cache if any */ 899 /* XXX should consider submaps */ 900 t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; 901 if (map != NULL) { 902 vsize_t vsize; 903 904 vsize = vm_map_max(map) - vm_map_min(map); 905 vsize = (uintmax_t)vsize * va_pct / 100; 906 if (t > vsize) { 907 t = vsize; 908 } 909 } 910 return t; 911 } 912 913 /* 914 * Print the system start up banner. 915 * 916 * - Print a limited banner if AB_SILENT. 917 * - Always send normal banner to the log. 918 */ 919 #define MEM_PBUFSIZE sizeof("99999 MB") 920 921 void 922 banner(void) 923 { 924 static char notice[] = " Notice: this software is " 925 "protected by copyright"; 926 char pbuf[81]; 927 void (*pr)(const char *, ...); 928 int i; 929 930 if ((boothowto & AB_SILENT) != 0) { 931 snprintf(pbuf, sizeof(pbuf), "%s %s (%s)", 932 ostype, osrelease, kernel_ident); 933 printf_nolog("%s", pbuf); 934 for (i = 80 - strlen(pbuf) - sizeof(notice); i > 0; i--) 935 printf(" "); 936 printf_nolog("%s\n", notice); 937 pr = aprint_normal; 938 } else { 939 pr = printf; 940 } 941 942 memset(pbuf, 0, sizeof(pbuf)); 943 (*pr)("%s%s", copyright, version); 944 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)physmem)); 945 (*pr)("total memory = %s\n", pbuf); 946 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)uvmexp.free)); 947 (*pr)("avail memory = %s\n", pbuf); 948 } 949