1 /* $NetBSD: init_main.c,v 1.428 2011/04/14 16:20:52 yamt 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 1.428 2011/04/14 16:20:52 yamt 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_sa.h" 108 #include "opt_syscall_debug.h" 109 #include "opt_sysv.h" 110 #include "opt_fileassoc.h" 111 #include "opt_ktrace.h" 112 #include "opt_pax.h" 113 #include "opt_compat_netbsd.h" 114 #include "opt_wapbl.h" 115 #include "opt_ptrace.h" 116 117 #include "drvctl.h" 118 #include "ksyms.h" 119 #include "rnd.h" 120 #include "sysmon_envsys.h" 121 #include "sysmon_power.h" 122 #include "sysmon_taskq.h" 123 #include "sysmon_wdog.h" 124 #include "veriexec.h" 125 126 #include <sys/param.h> 127 #include <sys/acct.h> 128 #include <sys/filedesc.h> 129 #include <sys/file.h> 130 #include <sys/errno.h> 131 #include <sys/callout.h> 132 #include <sys/cpu.h> 133 #include <sys/spldebug.h> 134 #include <sys/kernel.h> 135 #include <sys/mount.h> 136 #include <sys/proc.h> 137 #include <sys/kthread.h> 138 #include <sys/resourcevar.h> 139 #include <sys/signalvar.h> 140 #include <sys/systm.h> 141 #include <sys/vnode.h> 142 #include <sys/fstrans.h> 143 #include <sys/tty.h> 144 #include <sys/conf.h> 145 #include <sys/disklabel.h> 146 #include <sys/buf.h> 147 #include <sys/device.h> 148 #include <sys/exec.h> 149 #include <sys/socketvar.h> 150 #include <sys/protosw.h> 151 #include <sys/percpu.h> 152 #include <sys/pset.h> 153 #include <sys/sysctl.h> 154 #include <sys/reboot.h> 155 #include <sys/event.h> 156 #include <sys/mbuf.h> 157 #include <sys/sched.h> 158 #include <sys/sleepq.h> 159 #include <sys/iostat.h> 160 #include <sys/vmem.h> 161 #include <sys/uuid.h> 162 #include <sys/extent.h> 163 #include <sys/disk.h> 164 #include <sys/msgbuf.h> 165 #include <sys/module.h> 166 #include <sys/event.h> 167 #include <sys/lockf.h> 168 #include <sys/once.h> 169 #include <sys/ksyms.h> 170 #include <sys/uidinfo.h> 171 #include <sys/kprintf.h> 172 #ifdef FAST_IPSEC 173 #include <netipsec/ipsec.h> 174 #endif 175 #ifdef SYSVSHM 176 #include <sys/shm.h> 177 #endif 178 #ifdef SYSVSEM 179 #include <sys/sem.h> 180 #endif 181 #ifdef SYSVMSG 182 #include <sys/msg.h> 183 #endif 184 #include <sys/domain.h> 185 #include <sys/namei.h> 186 #if NRND > 0 187 #include <sys/rnd.h> 188 #endif 189 #include <sys/pipe.h> 190 #if NVERIEXEC > 0 191 #include <sys/verified_exec.h> 192 #endif /* NVERIEXEC > 0 */ 193 #ifdef KTRACE 194 #include <sys/ktrace.h> 195 #endif 196 #include <sys/kauth.h> 197 #ifdef KERN_SA 198 #include <sys/savar.h> 199 #endif 200 #include <net80211/ieee80211_netbsd.h> 201 #ifdef PTRACE 202 #include <sys/ptrace.h> 203 #endif /* PTRACE */ 204 205 #include <sys/syscall.h> 206 #include <sys/syscallargs.h> 207 208 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 209 #include <sys/pax.h> 210 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 211 212 #include <ufs/ufs/quota.h> 213 214 #include <miscfs/genfs/genfs.h> 215 #include <miscfs/syncfs/syncfs.h> 216 #include <miscfs/specfs/specdev.h> 217 218 #include <sys/cpu.h> 219 220 #include <uvm/uvm.h> /* extern struct uvm uvm */ 221 222 #if NSYSMON_TASKQ > 0 223 #include <dev/sysmon/sysmon_taskq.h> 224 #endif 225 226 #include <dev/cons.h> 227 228 #if NSYSMON_ENVSYS > 0 || NSYSMON_POWER > 0 || NSYSMON_WDOG > 0 229 #include <dev/sysmon/sysmonvar.h> 230 #endif 231 232 #include <net/bpf.h> 233 #include <net/if.h> 234 #include <net/raw_cb.h> 235 236 #include <prop/proplib.h> 237 238 #ifdef COMPAT_50 239 #include <compat/sys/time.h> 240 struct timeval50 boottime50; 241 #endif 242 243 #ifdef _KERNEL_OPT 244 #include "opt_userconf.h" 245 #endif 246 #ifdef USERCONF 247 #include <sys/userconf.h> 248 #endif 249 250 extern struct lwp lwp0; 251 extern time_t rootfstime; 252 253 #ifndef curlwp 254 struct lwp *curlwp = &lwp0; 255 #endif 256 struct proc *initproc; 257 258 struct vnode *rootvp, *swapdev_vp; 259 int boothowto; 260 int cold = 1; /* still working on startup */ 261 struct timespec boottime; /* time at system startup - will only follow settime deltas */ 262 263 int start_init_exec; /* semaphore for start_init() */ 264 265 static void check_console(struct lwp *l); 266 static void start_init(void *); 267 static void configure(void); 268 static void configure2(void); 269 static void configure3(void); 270 void main(void); 271 272 /* 273 * System startup; initialize the world, create process 0, mount root 274 * filesystem, and fork to create init and pagedaemon. Most of the 275 * hard work is done in the lower-level initialization routines including 276 * startup(), which does memory initialization and autoconfiguration. 277 */ 278 void 279 main(void) 280 { 281 struct timespec time; 282 struct lwp *l; 283 struct proc *p; 284 int s, error; 285 #ifdef NVNODE_IMPLICIT 286 int usevnodes; 287 #endif 288 CPU_INFO_ITERATOR cii; 289 struct cpu_info *ci; 290 291 l = &lwp0; 292 #ifndef LWP0_CPU_INFO 293 l->l_cpu = curcpu(); 294 #endif 295 l->l_pflag |= LP_RUNNING; 296 297 /* 298 * Attempt to find console and initialize 299 * in case of early panic or other messages. 300 */ 301 consinit(); 302 303 kernel_lock_init(); 304 once_init(); 305 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 306 kernconfig_lock_init(); 307 308 /* Initialize the device switch tables. */ 309 devsw_init(); 310 311 uvm_init(); 312 313 prop_kern_init(); 314 315 #if ((NKSYMS > 0) || (NDDB > 0) || (NMODULAR > 0)) 316 ksyms_init(); 317 #endif 318 kprintf_init(); 319 320 percpu_init(); 321 322 /* Initialize lock caches. */ 323 mutex_obj_init(); 324 rw_obj_init(); 325 326 /* Initialize the extent manager. */ 327 extent_init(); 328 329 /* Initialize event counters */ 330 evcnt_init(); 331 332 /* Do machine-dependent initialization. */ 333 cpu_startup(); 334 335 /* Initialize the sysctl subsystem. */ 336 sysctl_init(); 337 338 /* Initialize callouts, part 1. */ 339 callout_startup(); 340 341 /* Initialize the kernel authorization subsystem. */ 342 kauth_init(); 343 344 spec_init(); 345 346 /* 347 * Set BPF op vector. Can't do this in bpf attach, since 348 * network drivers attach before bpf. 349 */ 350 bpf_setops(); 351 352 /* Start module system. */ 353 module_init(); 354 355 /* 356 * Initialize the kernel authorization subsystem and start the 357 * default security model, if any. We need to do this early 358 * enough so that subsystems relying on any of the aforementioned 359 * can work properly. Since the security model may dictate the 360 * credential inheritance policy, it is needed at least before 361 * any process is created, specifically proc0. 362 */ 363 module_init_class(MODULE_CLASS_SECMODEL); 364 365 /* Initialize the buffer cache */ 366 bufinit(); 367 368 /* Initialize sockets. */ 369 soinit(); 370 371 /* 372 * The following things must be done before autoconfiguration. 373 */ 374 #if NRND > 0 375 rnd_init(); /* initialize random number generator */ 376 #endif 377 378 /* Initialize process and pgrp structures. */ 379 #ifdef KERN_SA 380 sa_init(); 381 #endif 382 procinit(); 383 lwpinit(); 384 385 /* Initialize signal-related data structures. */ 386 signal_init(); 387 388 /* Initialize resource management. */ 389 resource_init(); 390 391 /* Create process 0. */ 392 proc0_init(); 393 lwp0_init(); 394 395 /* Disable preemption during boot. */ 396 kpreempt_disable(); 397 398 /* Initialize the UID hash table. */ 399 uid_init(); 400 401 /* Charge root for one process. */ 402 (void)chgproccnt(0, 1); 403 404 /* Initialize timekeeping. */ 405 time_init(); 406 407 /* Initialize the run queues, turnstiles and sleep queues. */ 408 sched_rqinit(); 409 turnstile_init(); 410 sleeptab_init(&sleeptab); 411 412 sched_init(); 413 414 /* Initialize processor-sets */ 415 psets_init(); 416 417 /* MI initialization of the boot cpu */ 418 error = mi_cpu_attach(curcpu()); 419 KASSERT(error == 0); 420 421 /* Initialize timekeeping, part 2. */ 422 time_init2(); 423 424 /* 425 * Initialize mbuf's. Do this now because we might attempt to 426 * allocate mbufs or mbuf clusters during autoconfiguration. 427 */ 428 mbinit(); 429 430 /* Initialize I/O statistics. */ 431 iostat_init(); 432 433 /* Initialize the log device. */ 434 loginit(); 435 436 /* Second part of module system initialization. */ 437 module_start_unload_thread(); 438 439 /* Initialize the file systems. */ 440 #ifdef NVNODE_IMPLICIT 441 /* 442 * If maximum number of vnodes in namei vnode cache is not explicitly 443 * defined in kernel config, adjust the number such as we use roughly 444 * 10% of memory for vnodes and associated data structures in the 445 * assumed worst case. Do not provide fewer than NVNODE vnodes. 446 */ 447 usevnodes = 448 calc_cache_size(kernel_map, 10, VNODE_VA_MAXPCT) / VNODE_COST; 449 if (usevnodes > desiredvnodes) 450 desiredvnodes = usevnodes; 451 #endif 452 vfsinit(); 453 lf_init(); 454 455 /* Initialize fstrans. */ 456 fstrans_init(); 457 458 /* Initialize the file descriptor system. */ 459 fd_sys_init(); 460 461 /* Initialize cwd structures */ 462 cwd_sys_init(); 463 464 /* Initialize kqueue. */ 465 kqueue_init(); 466 467 /* Initialize the system monitor subsystems. */ 468 #if NSYSMON_TASKQ > 0 469 sysmon_task_queue_preinit(); 470 #endif 471 472 #if NSYSMON_ENVSYS > 0 473 sysmon_envsys_init(); 474 #endif 475 476 #if NSYSMON_POWER > 0 477 sysmon_power_init(); 478 #endif 479 480 #if NSYSMON_WDOG > 0 481 sysmon_wdog_init(); 482 #endif 483 484 inittimecounter(); 485 ntp_init(); 486 487 /* Initialize tty subsystem. */ 488 tty_init(); 489 ttyldisc_init(); 490 491 /* Initialize the buffer cache, part 2. */ 492 bufinit2(); 493 494 /* Initialize the disk wedge subsystem. */ 495 dkwedge_init(); 496 497 /* Initialize interfaces. */ 498 ifinit1(); 499 500 spldebug_start(); 501 502 /* Configure the system hardware. This will enable interrupts. */ 503 configure(); 504 505 ssp_init(); 506 507 ubc_init(); /* must be after autoconfig */ 508 509 configure2(); 510 /* Now timer is working. Enable preemption. */ 511 kpreempt_enable(); 512 513 #ifdef SYSVSHM 514 /* Initialize System V style shared memory. */ 515 shminit(); 516 #endif 517 518 vmem_rehash_start(); /* must be before exec_init */ 519 520 /* Initialize exec structures */ 521 exec_init(1); /* seminit calls exithook_establish() */ 522 523 #ifdef SYSVSEM 524 /* Initialize System V style semaphores. */ 525 seminit(); 526 #endif 527 528 #ifdef SYSVMSG 529 /* Initialize System V style message queues. */ 530 msginit(); 531 #endif 532 533 #if NVERIEXEC > 0 534 /* 535 * Initialise the Veriexec subsystem. 536 */ 537 veriexec_init(); 538 #endif /* NVERIEXEC > 0 */ 539 540 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 541 pax_init(); 542 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 543 544 #ifdef FAST_IPSEC 545 /* Attach network crypto subsystem */ 546 ipsec_attach(); 547 #endif 548 549 /* 550 * Initialize protocols. Block reception of incoming packets 551 * until everything is ready. 552 */ 553 s = splnet(); 554 ifinit(); 555 domaininit(true); 556 if_attachdomain(); 557 splx(s); 558 559 #ifdef GPROF 560 /* Initialize kernel profiling. */ 561 kmstartup(); 562 #endif 563 564 /* Initialize system accounting. */ 565 acct_init(); 566 567 #ifndef PIPE_SOCKETPAIR 568 /* Initialize pipes. */ 569 pipe_init(); 570 #endif 571 572 #ifdef KTRACE 573 /* Initialize ktrace. */ 574 ktrinit(); 575 #endif 576 577 #ifdef PTRACE 578 /* Initialize ptrace. */ 579 ptrace_init(); 580 #endif /* PTRACE */ 581 582 /* Initialize the UUID system calls. */ 583 uuid_init(); 584 585 machdep_init(); 586 587 procinit_sysctl(); 588 589 /* 590 * Create process 1 (init(8)). We do this now, as Unix has 591 * historically had init be process 1, and changing this would 592 * probably upset a lot of people. 593 * 594 * Note that process 1 won't immediately exec init(8), but will 595 * wait for us to inform it that the root file system has been 596 * mounted. 597 */ 598 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 599 panic("fork init"); 600 601 /* 602 * Load any remaining builtin modules, and hand back temporary 603 * storage to the VM system. Then require force when loading any 604 * remaining un-init'ed built-in modules to avoid later surprises. 605 */ 606 module_init_class(MODULE_CLASS_ANY); 607 module_builtin_require_force(); 608 609 /* 610 * Finalize configuration now that all real devices have been 611 * found. This needs to be done before the root device is 612 * selected, since finalization may create the root device. 613 */ 614 config_finalize(); 615 616 sysctl_finalize(); 617 618 /* 619 * Now that autoconfiguration has completed, we can determine 620 * the root and dump devices. 621 */ 622 cpu_rootconf(); 623 cpu_dumpconf(); 624 625 /* Mount the root file system. */ 626 do { 627 domountroothook(root_device); 628 if ((error = vfs_mountroot())) { 629 printf("cannot mount root, error = %d\n", error); 630 boothowto |= RB_ASKNAME; 631 setroot(root_device, 632 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 633 } 634 } while (error != 0); 635 mountroothook_destroy(); 636 637 configure3(); 638 639 /* 640 * Initialise the time-of-day clock, passing the time recorded 641 * in the root filesystem (if any) for use by systems that 642 * don't have a non-volatile time-of-day device. 643 */ 644 inittodr(rootfstime); 645 646 /* 647 * Now can look at time, having had a chance to verify the time 648 * from the file system. Reset l->l_rtime as it may have been 649 * munched in mi_switch() after the time got set. 650 */ 651 getnanotime(&time); 652 boottime = time; 653 #ifdef COMPAT_50 654 { 655 struct timeval tv; 656 TIMESPEC_TO_TIMEVAL(&tv, &time); 657 timeval_to_timeval50(&tv, &boottime50); 658 } 659 #endif 660 mutex_enter(proc_lock); 661 LIST_FOREACH(p, &allproc, p_list) { 662 KASSERT((p->p_flag & PK_MARKER) == 0); 663 mutex_enter(p->p_lock); 664 TIMESPEC_TO_TIMEVAL(&p->p_stats->p_start, &time); 665 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 666 lwp_lock(l); 667 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 668 lwp_unlock(l); 669 } 670 mutex_exit(p->p_lock); 671 } 672 mutex_exit(proc_lock); 673 binuptime(&curlwp->l_stime); 674 675 for (CPU_INFO_FOREACH(cii, ci)) { 676 ci->ci_schedstate.spc_lastmod = time_second; 677 } 678 679 /* Create the pageout daemon kernel thread. */ 680 uvm_swap_init(); 681 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 682 NULL, NULL, "pgdaemon")) 683 panic("fork pagedaemon"); 684 685 /* Create the filesystem syncer kernel thread. */ 686 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 687 NULL, NULL, "ioflush")) 688 panic("fork syncer"); 689 690 /* Create the aiodone daemon kernel thread. */ 691 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 692 uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE)) 693 panic("fork aiodoned"); 694 695 /* 696 * Okay, now we can let init(8) exec! It's off to userland! 697 */ 698 mutex_enter(proc_lock); 699 start_init_exec = 1; 700 cv_broadcast(&lbolt); 701 mutex_exit(proc_lock); 702 703 /* The scheduler is an infinite loop. */ 704 uvm_scheduler(); 705 /* NOTREACHED */ 706 } 707 708 /* 709 * Configure the system's hardware. 710 */ 711 static void 712 configure(void) 713 { 714 715 /* Initialize autoconf data structures. */ 716 config_init_mi(); 717 /* 718 * XXX 719 * callout_setfunc() requires mutex(9) so it can't be in config_init() 720 * on amiga and atari which use config_init() and autoconf(9) fucntions 721 * to initialize console devices. 722 */ 723 config_twiddle_init(); 724 725 pmf_init(); 726 #if NDRVCTL > 0 727 drvctl_init(); 728 #endif 729 730 #ifdef USERCONF 731 if (boothowto & RB_USERCONF) 732 user_config(); 733 #endif 734 735 if ((boothowto & (AB_SILENT|AB_VERBOSE)) == AB_SILENT) { 736 printf_nolog("Detecting hardware..."); 737 } 738 739 /* 740 * Do the machine-dependent portion of autoconfiguration. This 741 * sets the configuration machinery here in motion by "finding" 742 * the root bus. When this function returns, we expect interrupts 743 * to be enabled. 744 */ 745 cpu_configure(); 746 } 747 748 static void 749 configure2(void) 750 { 751 CPU_INFO_ITERATOR cii; 752 struct cpu_info *ci; 753 int s; 754 755 /* 756 * Now that we've found all the hardware, start the real time 757 * and statistics clocks. 758 */ 759 initclocks(); 760 761 cold = 0; /* clocks are running, we're warm now! */ 762 s = splsched(); 763 curcpu()->ci_schedstate.spc_flags |= SPCF_RUNNING; 764 splx(s); 765 766 /* Boot the secondary processors. */ 767 for (CPU_INFO_FOREACH(cii, ci)) { 768 uvm_cpu_attach(ci); 769 } 770 mp_online = true; 771 #if defined(MULTIPROCESSOR) 772 cpu_boot_secondary_processors(); 773 #endif 774 775 /* Setup the runqueues and scheduler. */ 776 runq_init(); 777 synch_init(); 778 779 /* 780 * Bus scans can make it appear as if the system has paused, so 781 * twiddle constantly while config_interrupts() jobs are running. 782 */ 783 config_twiddle_fn(NULL); 784 785 /* 786 * Create threads to call back and finish configuration for 787 * devices that want interrupts enabled. 788 */ 789 config_create_interruptthreads(); 790 791 /* Get the threads going and into any sleeps before continuing. */ 792 yield(); 793 } 794 795 static void 796 configure3(void) 797 { 798 799 /* 800 * Create threads to call back and finish configuration for 801 * devices that want the mounted root file system. 802 */ 803 config_create_mountrootthreads(); 804 805 /* Get the threads going and into any sleeps before continuing. */ 806 yield(); 807 } 808 809 static void 810 check_console(struct lwp *l) 811 { 812 struct vnode *vp; 813 int error; 814 815 error = namei_simple_kernel("/dev/console", 816 NSM_FOLLOW_NOEMULROOT, &vp); 817 if (error == 0) 818 vrele(vp); 819 else if (error == ENOENT) 820 printf("warning: no /dev/console\n"); 821 else 822 printf("warning: lookup /dev/console: error %d\n", error); 823 } 824 825 /* 826 * List of paths to try when searching for "init". 827 */ 828 static const char * const initpaths[] = { 829 "/sbin/init", 830 "/sbin/oinit", 831 "/sbin/init.bak", 832 NULL, 833 }; 834 835 /* 836 * Start the initial user process; try exec'ing each pathname in "initpaths". 837 * The program is invoked with one argument containing the boot flags. 838 */ 839 static void 840 start_init(void *arg) 841 { 842 struct lwp *l = arg; 843 struct proc *p = l->l_proc; 844 vaddr_t addr; 845 struct sys_execve_args /* { 846 syscallarg(const char *) path; 847 syscallarg(char * const *) argp; 848 syscallarg(char * const *) envp; 849 } */ args; 850 int options, i, error; 851 register_t retval[2]; 852 char flags[4], *flagsp; 853 const char *path, *slash; 854 char *ucp, **uap, *arg0, *arg1 = NULL; 855 char ipath[129]; 856 int ipx, len; 857 858 /* 859 * Now in process 1. 860 */ 861 strncpy(p->p_comm, "init", MAXCOMLEN); 862 863 /* 864 * Wait for main() to tell us that it's safe to exec. 865 */ 866 mutex_enter(proc_lock); 867 while (start_init_exec == 0) 868 cv_wait(&lbolt, proc_lock); 869 mutex_exit(proc_lock); 870 871 /* 872 * This is not the right way to do this. We really should 873 * hand-craft a descriptor onto /dev/console to hand to init, 874 * but that's a _lot_ more work, and the benefit from this easy 875 * hack makes up for the "good is the enemy of the best" effect. 876 */ 877 check_console(l); 878 879 /* 880 * Need just enough stack to hold the faked-up "execve()" arguments. 881 */ 882 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 883 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 884 NULL, UVM_UNKNOWN_OFFSET, 0, 885 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 886 UVM_ADV_NORMAL, 887 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 888 panic("init: couldn't allocate argument space"); 889 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 890 891 ipx = 0; 892 while (1) { 893 if (boothowto & RB_ASKNAME) { 894 printf("init path"); 895 if (initpaths[ipx]) 896 printf(" (default %s)", initpaths[ipx]); 897 printf(": "); 898 len = cngetsn(ipath, sizeof(ipath)-1); 899 if (len == 4 && strcmp(ipath, "halt") == 0) { 900 cpu_reboot(RB_HALT, NULL); 901 } else if (len == 6 && strcmp(ipath, "reboot") == 0) { 902 cpu_reboot(0, NULL); 903 #if defined(DDB) 904 } else if (len == 3 && strcmp(ipath, "ddb") == 0) { 905 console_debugger(); 906 continue; 907 #endif 908 } else if (len > 0 && ipath[0] == '/') { 909 ipath[len] = '\0'; 910 path = ipath; 911 } else if (len == 0 && initpaths[ipx] != NULL) { 912 path = initpaths[ipx++]; 913 } else { 914 printf("use absolute path, "); 915 #if defined(DDB) 916 printf("\"ddb\", "); 917 #endif 918 printf("\"halt\", or \"reboot\"\n"); 919 continue; 920 } 921 } else { 922 if ((path = initpaths[ipx++]) == NULL) { 923 ipx = 0; 924 boothowto |= RB_ASKNAME; 925 continue; 926 } 927 } 928 929 ucp = (char *)USRSTACK; 930 931 /* 932 * Construct the boot flag argument. 933 */ 934 flagsp = flags; 935 *flagsp++ = '-'; 936 options = 0; 937 938 if (boothowto & RB_SINGLE) { 939 *flagsp++ = 's'; 940 options = 1; 941 } 942 #ifdef notyet 943 if (boothowto & RB_FASTBOOT) { 944 *flagsp++ = 'f'; 945 options = 1; 946 } 947 #endif 948 949 /* 950 * Move out the flags (arg 1), if necessary. 951 */ 952 if (options != 0) { 953 *flagsp++ = '\0'; 954 i = flagsp - flags; 955 #ifdef DEBUG 956 aprint_normal("init: copying out flags `%s' %d\n", flags, i); 957 #endif 958 arg1 = STACK_ALLOC(ucp, i); 959 ucp = STACK_MAX(arg1, i); 960 (void)copyout((void *)flags, arg1, i); 961 } 962 963 /* 964 * Move out the file name (also arg 0). 965 */ 966 i = strlen(path) + 1; 967 #ifdef DEBUG 968 aprint_normal("init: copying out path `%s' %d\n", path, i); 969 #else 970 if (boothowto & RB_ASKNAME || path != initpaths[0]) 971 printf("init: trying %s\n", path); 972 #endif 973 arg0 = STACK_ALLOC(ucp, i); 974 ucp = STACK_MAX(arg0, i); 975 (void)copyout(path, arg0, i); 976 977 /* 978 * Move out the arg pointers. 979 */ 980 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 981 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 982 SCARG(&args, path) = arg0; 983 SCARG(&args, argp) = uap; 984 SCARG(&args, envp) = NULL; 985 slash = strrchr(path, '/'); 986 if (slash) 987 (void)suword((void *)uap++, 988 (long)arg0 + (slash + 1 - path)); 989 else 990 (void)suword((void *)uap++, (long)arg0); 991 if (options != 0) 992 (void)suword((void *)uap++, (long)arg1); 993 (void)suword((void *)uap++, 0); /* terminator */ 994 995 /* 996 * Now try to exec the program. If can't for any reason 997 * other than it doesn't exist, complain. 998 */ 999 error = sys_execve(l, &args, retval); 1000 if (error == 0 || error == EJUSTRETURN) { 1001 KERNEL_UNLOCK_LAST(l); 1002 return; 1003 } 1004 printf("exec %s: error %d\n", path, error); 1005 } 1006 printf("init: not found\n"); 1007 panic("no init"); 1008 } 1009 1010 /* 1011 * calculate cache size (in bytes) from physmem and vm_map size. 1012 */ 1013 vaddr_t 1014 calc_cache_size(struct vm_map *map, int pct, int va_pct) 1015 { 1016 paddr_t t; 1017 1018 /* XXX should consider competing cache if any */ 1019 /* XXX should consider submaps */ 1020 t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; 1021 if (map != NULL) { 1022 vsize_t vsize; 1023 1024 vsize = vm_map_max(map) - vm_map_min(map); 1025 vsize = (uintmax_t)vsize * va_pct / 100; 1026 if (t > vsize) { 1027 t = vsize; 1028 } 1029 } 1030 return t; 1031 } 1032 1033 /* 1034 * Print the system start up banner. 1035 * 1036 * - Print a limited banner if AB_SILENT. 1037 * - Always send normal banner to the log. 1038 */ 1039 #define MEM_PBUFSIZE sizeof("99999 MB") 1040 1041 void 1042 banner(void) 1043 { 1044 static char notice[] = " Notice: this software is " 1045 "protected by copyright"; 1046 char pbuf[81]; 1047 void (*pr)(const char *, ...); 1048 int i; 1049 1050 if ((boothowto & AB_SILENT) != 0) { 1051 snprintf(pbuf, sizeof(pbuf), "%s %s (%s)", 1052 ostype, osrelease, kernel_ident); 1053 printf_nolog("%s", pbuf); 1054 for (i = 80 - strlen(pbuf) - sizeof(notice); i > 0; i--) 1055 printf(" "); 1056 printf_nolog("%s\n", notice); 1057 pr = aprint_normal; 1058 } else { 1059 pr = printf; 1060 } 1061 1062 memset(pbuf, 0, sizeof(pbuf)); 1063 (*pr)("%s%s", copyright, version); 1064 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)physmem)); 1065 (*pr)("total memory = %s\n", pbuf); 1066 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)uvmexp.free)); 1067 (*pr)("avail memory = %s\n", pbuf); 1068 } 1069