1 /* $NetBSD: init_main.c,v 1.435 2011/08/07 13:33:01 rmind 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 1.435 2011/08/07 13:33:01 rmind 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/pserialize.h> 153 #include <sys/pset.h> 154 #include <sys/sysctl.h> 155 #include <sys/reboot.h> 156 #include <sys/event.h> 157 #include <sys/mbuf.h> 158 #include <sys/sched.h> 159 #include <sys/sleepq.h> 160 #include <sys/iostat.h> 161 #include <sys/vmem.h> 162 #include <sys/uuid.h> 163 #include <sys/extent.h> 164 #include <sys/disk.h> 165 #include <sys/msgbuf.h> 166 #include <sys/module.h> 167 #include <sys/event.h> 168 #include <sys/lockf.h> 169 #include <sys/once.h> 170 #include <sys/kcpuset.h> 171 #include <sys/ksyms.h> 172 #include <sys/uidinfo.h> 173 #include <sys/kprintf.h> 174 #ifdef FAST_IPSEC 175 #include <netipsec/ipsec.h> 176 #endif 177 #ifdef SYSVSHM 178 #include <sys/shm.h> 179 #endif 180 #ifdef SYSVSEM 181 #include <sys/sem.h> 182 #endif 183 #ifdef SYSVMSG 184 #include <sys/msg.h> 185 #endif 186 #include <sys/domain.h> 187 #include <sys/namei.h> 188 #if NRND > 0 189 #include <sys/rnd.h> 190 #endif 191 #include <sys/pipe.h> 192 #if NVERIEXEC > 0 193 #include <sys/verified_exec.h> 194 #endif /* NVERIEXEC > 0 */ 195 #ifdef KTRACE 196 #include <sys/ktrace.h> 197 #endif 198 #include <sys/kauth.h> 199 #ifdef KERN_SA 200 #include <sys/savar.h> 201 #endif 202 #include <net80211/ieee80211_netbsd.h> 203 #ifdef PTRACE 204 #include <sys/ptrace.h> 205 #endif /* PTRACE */ 206 207 #include <sys/syscall.h> 208 #include <sys/syscallargs.h> 209 210 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 211 #include <sys/pax.h> 212 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 213 214 #include <ufs/ufs/quota.h> 215 216 #include <miscfs/genfs/genfs.h> 217 #include <miscfs/syncfs/syncfs.h> 218 #include <miscfs/specfs/specdev.h> 219 220 #include <sys/cpu.h> 221 222 #include <uvm/uvm.h> /* extern struct uvm uvm */ 223 224 #if NSYSMON_TASKQ > 0 225 #include <dev/sysmon/sysmon_taskq.h> 226 #endif 227 228 #include <dev/cons.h> 229 230 #if NSYSMON_ENVSYS > 0 || NSYSMON_POWER > 0 || NSYSMON_WDOG > 0 231 #include <dev/sysmon/sysmonvar.h> 232 #endif 233 234 #include <net/bpf.h> 235 #include <net/if.h> 236 #include <net/raw_cb.h> 237 238 #include <prop/proplib.h> 239 240 #ifdef COMPAT_50 241 #include <compat/sys/time.h> 242 struct timeval50 boottime50; 243 #endif 244 245 #include <sys/userconf.h> 246 247 extern struct lwp lwp0; 248 extern time_t rootfstime; 249 250 #ifndef curlwp 251 struct lwp *curlwp = &lwp0; 252 #endif 253 struct proc *initproc; 254 255 struct vnode *rootvp, *swapdev_vp; 256 int boothowto; 257 int cold = 1; /* still working on startup */ 258 struct timespec boottime; /* time at system startup - will only follow settime deltas */ 259 260 int start_init_exec; /* semaphore for start_init() */ 261 262 static void check_console(struct lwp *l); 263 static void start_init(void *); 264 static void configure(void); 265 static void configure2(void); 266 static void configure3(void); 267 void main(void); 268 269 /* 270 * System startup; initialize the world, create process 0, mount root 271 * filesystem, and fork to create init and pagedaemon. Most of the 272 * hard work is done in the lower-level initialization routines including 273 * startup(), which does memory initialization and autoconfiguration. 274 */ 275 void 276 main(void) 277 { 278 struct timespec time; 279 struct lwp *l; 280 struct proc *p; 281 int s, error; 282 #ifdef NVNODE_IMPLICIT 283 int usevnodes; 284 #endif 285 CPU_INFO_ITERATOR cii; 286 struct cpu_info *ci; 287 288 l = &lwp0; 289 #ifndef LWP0_CPU_INFO 290 l->l_cpu = curcpu(); 291 #endif 292 l->l_pflag |= LP_RUNNING; 293 294 /* 295 * Attempt to find console and initialize 296 * in case of early panic or other messages. 297 */ 298 consinit(); 299 300 kernel_lock_init(); 301 once_init(); 302 303 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 304 kernconfig_lock_init(); 305 kthread_sysinit(); 306 307 /* Initialize the device switch tables. */ 308 devsw_init(); 309 310 /* Initialize event counters. */ 311 evcnt_init(); 312 313 uvm_init(); 314 kcpuset_sysinit(); 315 316 prop_kern_init(); 317 318 #if ((NKSYMS > 0) || (NDDB > 0) || (NMODULAR > 0)) 319 ksyms_init(); 320 #endif 321 kprintf_init(); 322 323 percpu_init(); 324 325 /* Initialize lock caches. */ 326 mutex_obj_init(); 327 rw_obj_init(); 328 329 /* Passive serialization. */ 330 pserialize_init(); 331 332 /* Initialize the extent manager. */ 333 extent_init(); 334 335 /* Do machine-dependent initialization. */ 336 cpu_startup(); 337 338 /* Initialize the sysctl subsystem. */ 339 sysctl_init(); 340 341 /* Initialize callouts, part 1. */ 342 callout_startup(); 343 344 /* Initialize the kernel authorization subsystem. */ 345 kauth_init(); 346 347 spec_init(); 348 349 /* 350 * Set BPF op vector. Can't do this in bpf attach, since 351 * network drivers attach before bpf. 352 */ 353 bpf_setops(); 354 355 /* Start module system. */ 356 module_init(); 357 358 /* 359 * Initialize the kernel authorization subsystem and start the 360 * default security model, if any. We need to do this early 361 * enough so that subsystems relying on any of the aforementioned 362 * can work properly. Since the security model may dictate the 363 * credential inheritance policy, it is needed at least before 364 * any process is created, specifically proc0. 365 */ 366 module_init_class(MODULE_CLASS_SECMODEL); 367 368 /* Initialize the buffer cache */ 369 bufinit(); 370 371 /* Initialize sockets. */ 372 soinit(); 373 374 /* 375 * The following things must be done before autoconfiguration. 376 */ 377 #if NRND > 0 378 rnd_init(); /* initialize random number generator */ 379 #endif 380 381 /* Initialize process and pgrp structures. */ 382 #ifdef KERN_SA 383 sa_init(); 384 #endif 385 procinit(); 386 lwpinit(); 387 388 /* Initialize signal-related data structures. */ 389 signal_init(); 390 391 /* Initialize resource management. */ 392 resource_init(); 393 394 /* Create process 0. */ 395 proc0_init(); 396 lwp0_init(); 397 398 /* Disable preemption during boot. */ 399 kpreempt_disable(); 400 401 /* Initialize the UID hash table. */ 402 uid_init(); 403 404 /* Charge root for one process. */ 405 (void)chgproccnt(0, 1); 406 407 /* Initialize timekeeping. */ 408 time_init(); 409 410 /* Initialize the run queues, turnstiles and sleep queues. */ 411 sched_rqinit(); 412 turnstile_init(); 413 sleeptab_init(&sleeptab); 414 415 sched_init(); 416 417 /* Initialize processor-sets */ 418 psets_init(); 419 420 /* MI initialization of the boot cpu */ 421 error = mi_cpu_attach(curcpu()); 422 KASSERT(error == 0); 423 424 /* Initialize timekeeping, part 2. */ 425 time_init2(); 426 427 /* 428 * Initialize mbuf's. Do this now because we might attempt to 429 * allocate mbufs or mbuf clusters during autoconfiguration. 430 */ 431 mbinit(); 432 433 /* Initialize I/O statistics. */ 434 iostat_init(); 435 436 /* Initialize the log device. */ 437 loginit(); 438 439 /* Second part of module system initialization. */ 440 module_start_unload_thread(); 441 442 /* Initialize the file systems. */ 443 #ifdef NVNODE_IMPLICIT 444 /* 445 * If maximum number of vnodes in namei vnode cache is not explicitly 446 * defined in kernel config, adjust the number such as we use roughly 447 * 10% of memory for vnodes and associated data structures in the 448 * assumed worst case. Do not provide fewer than NVNODE vnodes. 449 */ 450 usevnodes = 451 calc_cache_size(kernel_map, 10, VNODE_VA_MAXPCT) / VNODE_COST; 452 if (usevnodes > desiredvnodes) 453 desiredvnodes = usevnodes; 454 #endif 455 vfsinit(); 456 lf_init(); 457 458 /* Initialize fstrans. */ 459 fstrans_init(); 460 461 /* Initialize the file descriptor system. */ 462 fd_sys_init(); 463 464 /* Initialize cwd structures */ 465 cwd_sys_init(); 466 467 /* Initialize kqueue. */ 468 kqueue_init(); 469 470 /* Initialize the system monitor subsystems. */ 471 #if NSYSMON_TASKQ > 0 472 sysmon_task_queue_preinit(); 473 #endif 474 475 #if NSYSMON_ENVSYS > 0 476 sysmon_envsys_init(); 477 #endif 478 479 #if NSYSMON_POWER > 0 480 sysmon_power_init(); 481 #endif 482 483 #if NSYSMON_WDOG > 0 484 sysmon_wdog_init(); 485 #endif 486 487 inittimecounter(); 488 ntp_init(); 489 490 /* Initialize tty subsystem. */ 491 tty_init(); 492 ttyldisc_init(); 493 494 /* Initialize the buffer cache, part 2. */ 495 bufinit2(); 496 497 /* Initialize the disk wedge subsystem. */ 498 dkwedge_init(); 499 500 /* Initialize interfaces. */ 501 ifinit1(); 502 503 spldebug_start(); 504 505 /* Initialize sockets thread(s) */ 506 soinit1(); 507 508 509 /* Configure the system hardware. This will enable interrupts. */ 510 configure(); 511 512 ssp_init(); 513 514 ubc_init(); /* must be after autoconfig */ 515 516 mm_init(); 517 518 configure2(); 519 /* Now timer is working. Enable preemption. */ 520 kpreempt_enable(); 521 522 #ifdef SYSVSHM 523 /* Initialize System V style shared memory. */ 524 shminit(); 525 #endif 526 527 vmem_rehash_start(); /* must be before exec_init */ 528 529 /* Initialize exec structures */ 530 exec_init(1); /* seminit calls exithook_establish() */ 531 532 #ifdef SYSVSEM 533 /* Initialize System V style semaphores. */ 534 seminit(); 535 #endif 536 537 #ifdef SYSVMSG 538 /* Initialize System V style message queues. */ 539 msginit(); 540 #endif 541 542 #if NVERIEXEC > 0 543 /* 544 * Initialise the Veriexec subsystem. 545 */ 546 veriexec_init(); 547 #endif /* NVERIEXEC > 0 */ 548 549 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 550 pax_init(); 551 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 552 553 #ifdef FAST_IPSEC 554 /* Attach network crypto subsystem */ 555 ipsec_attach(); 556 #endif 557 558 /* 559 * Initialize protocols. Block reception of incoming packets 560 * until everything is ready. 561 */ 562 s = splnet(); 563 ifinit(); 564 domaininit(true); 565 if_attachdomain(); 566 splx(s); 567 568 #ifdef GPROF 569 /* Initialize kernel profiling. */ 570 kmstartup(); 571 #endif 572 573 /* Initialize system accounting. */ 574 acct_init(); 575 576 #ifndef PIPE_SOCKETPAIR 577 /* Initialize pipes. */ 578 pipe_init(); 579 #endif 580 581 #ifdef KTRACE 582 /* Initialize ktrace. */ 583 ktrinit(); 584 #endif 585 586 #ifdef PTRACE 587 /* Initialize ptrace. */ 588 ptrace_init(); 589 #endif /* PTRACE */ 590 591 /* Initialize the UUID system calls. */ 592 uuid_init(); 593 594 machdep_init(); 595 596 procinit_sysctl(); 597 598 /* 599 * Create process 1 (init(8)). We do this now, as Unix has 600 * historically had init be process 1, and changing this would 601 * probably upset a lot of people. 602 * 603 * Note that process 1 won't immediately exec init(8), but will 604 * wait for us to inform it that the root file system has been 605 * mounted. 606 */ 607 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 608 panic("fork init"); 609 610 /* 611 * Load any remaining builtin modules, and hand back temporary 612 * storage to the VM system. Then require force when loading any 613 * remaining un-init'ed built-in modules to avoid later surprises. 614 */ 615 module_init_class(MODULE_CLASS_ANY); 616 module_builtin_require_force(); 617 618 /* 619 * Finalize configuration now that all real devices have been 620 * found. This needs to be done before the root device is 621 * selected, since finalization may create the root device. 622 */ 623 config_finalize(); 624 625 sysctl_finalize(); 626 627 /* 628 * Now that autoconfiguration has completed, we can determine 629 * the root and dump devices. 630 */ 631 cpu_rootconf(); 632 cpu_dumpconf(); 633 634 /* Mount the root file system. */ 635 do { 636 domountroothook(root_device); 637 if ((error = vfs_mountroot())) { 638 printf("cannot mount root, error = %d\n", error); 639 boothowto |= RB_ASKNAME; 640 setroot(root_device, 641 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 642 } 643 } while (error != 0); 644 mountroothook_destroy(); 645 646 configure3(); 647 648 /* 649 * Initialise the time-of-day clock, passing the time recorded 650 * in the root filesystem (if any) for use by systems that 651 * don't have a non-volatile time-of-day device. 652 */ 653 inittodr(rootfstime); 654 655 /* 656 * Now can look at time, having had a chance to verify the time 657 * from the file system. Reset l->l_rtime as it may have been 658 * munched in mi_switch() after the time got set. 659 */ 660 getnanotime(&time); 661 boottime = time; 662 #ifdef COMPAT_50 663 { 664 struct timeval tv; 665 TIMESPEC_TO_TIMEVAL(&tv, &time); 666 timeval_to_timeval50(&tv, &boottime50); 667 } 668 #endif 669 mutex_enter(proc_lock); 670 LIST_FOREACH(p, &allproc, p_list) { 671 KASSERT((p->p_flag & PK_MARKER) == 0); 672 mutex_enter(p->p_lock); 673 TIMESPEC_TO_TIMEVAL(&p->p_stats->p_start, &time); 674 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 675 lwp_lock(l); 676 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 677 lwp_unlock(l); 678 } 679 mutex_exit(p->p_lock); 680 } 681 mutex_exit(proc_lock); 682 binuptime(&curlwp->l_stime); 683 684 for (CPU_INFO_FOREACH(cii, ci)) { 685 ci->ci_schedstate.spc_lastmod = time_second; 686 } 687 688 /* Create the pageout daemon kernel thread. */ 689 uvm_swap_init(); 690 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 691 NULL, NULL, "pgdaemon")) 692 panic("fork pagedaemon"); 693 694 /* Create the filesystem syncer kernel thread. */ 695 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 696 NULL, NULL, "ioflush")) 697 panic("fork syncer"); 698 699 /* Create the aiodone daemon kernel thread. */ 700 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 701 uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE)) 702 panic("fork aiodoned"); 703 704 /* 705 * Okay, now we can let init(8) exec! It's off to userland! 706 */ 707 mutex_enter(proc_lock); 708 start_init_exec = 1; 709 cv_broadcast(&lbolt); 710 mutex_exit(proc_lock); 711 712 /* The scheduler is an infinite loop. */ 713 uvm_scheduler(); 714 /* NOTREACHED */ 715 } 716 717 /* 718 * Configure the system's hardware. 719 */ 720 static void 721 configure(void) 722 { 723 724 /* Initialize autoconf data structures. */ 725 config_init_mi(); 726 /* 727 * XXX 728 * callout_setfunc() requires mutex(9) so it can't be in config_init() 729 * on amiga and atari which use config_init() and autoconf(9) fucntions 730 * to initialize console devices. 731 */ 732 config_twiddle_init(); 733 734 pmf_init(); 735 #if NDRVCTL > 0 736 drvctl_init(); 737 #endif 738 739 userconf_init(); 740 if (boothowto & RB_USERCONF) 741 userconf_prompt(); 742 743 if ((boothowto & (AB_SILENT|AB_VERBOSE)) == AB_SILENT) { 744 printf_nolog("Detecting hardware..."); 745 } 746 747 /* 748 * Do the machine-dependent portion of autoconfiguration. This 749 * sets the configuration machinery here in motion by "finding" 750 * the root bus. When this function returns, we expect interrupts 751 * to be enabled. 752 */ 753 cpu_configure(); 754 } 755 756 static void 757 configure2(void) 758 { 759 CPU_INFO_ITERATOR cii; 760 struct cpu_info *ci; 761 int s; 762 763 /* 764 * Now that we've found all the hardware, start the real time 765 * and statistics clocks. 766 */ 767 initclocks(); 768 769 cold = 0; /* clocks are running, we're warm now! */ 770 s = splsched(); 771 curcpu()->ci_schedstate.spc_flags |= SPCF_RUNNING; 772 splx(s); 773 774 /* Boot the secondary processors. */ 775 for (CPU_INFO_FOREACH(cii, ci)) { 776 uvm_cpu_attach(ci); 777 } 778 mp_online = true; 779 #if defined(MULTIPROCESSOR) 780 cpu_boot_secondary_processors(); 781 #endif 782 783 /* Setup the runqueues and scheduler. */ 784 runq_init(); 785 synch_init(); 786 787 /* 788 * Bus scans can make it appear as if the system has paused, so 789 * twiddle constantly while config_interrupts() jobs are running. 790 */ 791 config_twiddle_fn(NULL); 792 793 /* 794 * Create threads to call back and finish configuration for 795 * devices that want interrupts enabled. 796 */ 797 config_create_interruptthreads(); 798 799 /* Get the threads going and into any sleeps before continuing. */ 800 yield(); 801 } 802 803 static void 804 configure3(void) 805 { 806 807 /* 808 * Create threads to call back and finish configuration for 809 * devices that want the mounted root file system. 810 */ 811 config_create_mountrootthreads(); 812 813 /* Get the threads going and into any sleeps before continuing. */ 814 yield(); 815 } 816 817 static void 818 check_console(struct lwp *l) 819 { 820 struct vnode *vp; 821 int error; 822 823 error = namei_simple_kernel("/dev/console", 824 NSM_FOLLOW_NOEMULROOT, &vp); 825 if (error == 0) 826 vrele(vp); 827 else if (error == ENOENT) 828 printf("warning: no /dev/console\n"); 829 else 830 printf("warning: lookup /dev/console: error %d\n", error); 831 } 832 833 /* 834 * List of paths to try when searching for "init". 835 */ 836 static const char * const initpaths[] = { 837 "/sbin/init", 838 "/sbin/oinit", 839 "/sbin/init.bak", 840 NULL, 841 }; 842 843 /* 844 * Start the initial user process; try exec'ing each pathname in "initpaths". 845 * The program is invoked with one argument containing the boot flags. 846 */ 847 static void 848 start_init(void *arg) 849 { 850 struct lwp *l = arg; 851 struct proc *p = l->l_proc; 852 vaddr_t addr; 853 struct sys_execve_args /* { 854 syscallarg(const char *) path; 855 syscallarg(char * const *) argp; 856 syscallarg(char * const *) envp; 857 } */ args; 858 int options, i, error; 859 register_t retval[2]; 860 char flags[4], *flagsp; 861 const char *path, *slash; 862 char *ucp, **uap, *arg0, *arg1 = NULL; 863 char ipath[129]; 864 int ipx, len; 865 866 /* 867 * Now in process 1. 868 */ 869 strncpy(p->p_comm, "init", MAXCOMLEN); 870 871 /* 872 * Wait for main() to tell us that it's safe to exec. 873 */ 874 mutex_enter(proc_lock); 875 while (start_init_exec == 0) 876 cv_wait(&lbolt, proc_lock); 877 mutex_exit(proc_lock); 878 879 /* 880 * This is not the right way to do this. We really should 881 * hand-craft a descriptor onto /dev/console to hand to init, 882 * but that's a _lot_ more work, and the benefit from this easy 883 * hack makes up for the "good is the enemy of the best" effect. 884 */ 885 check_console(l); 886 887 /* 888 * Need just enough stack to hold the faked-up "execve()" arguments. 889 */ 890 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 891 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 892 NULL, UVM_UNKNOWN_OFFSET, 0, 893 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 894 UVM_ADV_NORMAL, 895 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 896 panic("init: couldn't allocate argument space"); 897 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 898 899 ipx = 0; 900 while (1) { 901 if (boothowto & RB_ASKNAME) { 902 printf("init path"); 903 if (initpaths[ipx]) 904 printf(" (default %s)", initpaths[ipx]); 905 printf(": "); 906 len = cngetsn(ipath, sizeof(ipath)-1); 907 if (len == 4 && strcmp(ipath, "halt") == 0) { 908 cpu_reboot(RB_HALT, NULL); 909 } else if (len == 6 && strcmp(ipath, "reboot") == 0) { 910 cpu_reboot(0, NULL); 911 #if defined(DDB) 912 } else if (len == 3 && strcmp(ipath, "ddb") == 0) { 913 console_debugger(); 914 continue; 915 #endif 916 } else if (len > 0 && ipath[0] == '/') { 917 ipath[len] = '\0'; 918 path = ipath; 919 } else if (len == 0 && initpaths[ipx] != NULL) { 920 path = initpaths[ipx++]; 921 } else { 922 printf("use absolute path, "); 923 #if defined(DDB) 924 printf("\"ddb\", "); 925 #endif 926 printf("\"halt\", or \"reboot\"\n"); 927 continue; 928 } 929 } else { 930 if ((path = initpaths[ipx++]) == NULL) { 931 ipx = 0; 932 boothowto |= RB_ASKNAME; 933 continue; 934 } 935 } 936 937 ucp = (char *)USRSTACK; 938 939 /* 940 * Construct the boot flag argument. 941 */ 942 flagsp = flags; 943 *flagsp++ = '-'; 944 options = 0; 945 946 if (boothowto & RB_SINGLE) { 947 *flagsp++ = 's'; 948 options = 1; 949 } 950 #ifdef notyet 951 if (boothowto & RB_FASTBOOT) { 952 *flagsp++ = 'f'; 953 options = 1; 954 } 955 #endif 956 957 /* 958 * Move out the flags (arg 1), if necessary. 959 */ 960 if (options != 0) { 961 *flagsp++ = '\0'; 962 i = flagsp - flags; 963 #ifdef DEBUG 964 aprint_normal("init: copying out flags `%s' %d\n", flags, i); 965 #endif 966 arg1 = STACK_ALLOC(ucp, i); 967 ucp = STACK_MAX(arg1, i); 968 (void)copyout((void *)flags, arg1, i); 969 } 970 971 /* 972 * Move out the file name (also arg 0). 973 */ 974 i = strlen(path) + 1; 975 #ifdef DEBUG 976 aprint_normal("init: copying out path `%s' %d\n", path, i); 977 #else 978 if (boothowto & RB_ASKNAME || path != initpaths[0]) 979 printf("init: trying %s\n", path); 980 #endif 981 arg0 = STACK_ALLOC(ucp, i); 982 ucp = STACK_MAX(arg0, i); 983 (void)copyout(path, arg0, i); 984 985 /* 986 * Move out the arg pointers. 987 */ 988 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 989 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 990 SCARG(&args, path) = arg0; 991 SCARG(&args, argp) = uap; 992 SCARG(&args, envp) = NULL; 993 slash = strrchr(path, '/'); 994 if (slash) 995 (void)suword((void *)uap++, 996 (long)arg0 + (slash + 1 - path)); 997 else 998 (void)suword((void *)uap++, (long)arg0); 999 if (options != 0) 1000 (void)suword((void *)uap++, (long)arg1); 1001 (void)suword((void *)uap++, 0); /* terminator */ 1002 1003 /* 1004 * Now try to exec the program. If can't for any reason 1005 * other than it doesn't exist, complain. 1006 */ 1007 error = sys_execve(l, &args, retval); 1008 if (error == 0 || error == EJUSTRETURN) { 1009 KERNEL_UNLOCK_LAST(l); 1010 return; 1011 } 1012 printf("exec %s: error %d\n", path, error); 1013 } 1014 printf("init: not found\n"); 1015 panic("no init"); 1016 } 1017 1018 /* 1019 * calculate cache size (in bytes) from physmem and vm_map size. 1020 */ 1021 vaddr_t 1022 calc_cache_size(struct vm_map *map, int pct, int va_pct) 1023 { 1024 paddr_t t; 1025 1026 /* XXX should consider competing cache if any */ 1027 /* XXX should consider submaps */ 1028 t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; 1029 if (map != NULL) { 1030 vsize_t vsize; 1031 1032 vsize = vm_map_max(map) - vm_map_min(map); 1033 vsize = (uintmax_t)vsize * va_pct / 100; 1034 if (t > vsize) { 1035 t = vsize; 1036 } 1037 } 1038 return t; 1039 } 1040 1041 /* 1042 * Print the system start up banner. 1043 * 1044 * - Print a limited banner if AB_SILENT. 1045 * - Always send normal banner to the log. 1046 */ 1047 #define MEM_PBUFSIZE sizeof("99999 MB") 1048 1049 void 1050 banner(void) 1051 { 1052 static char notice[] = " Notice: this software is " 1053 "protected by copyright"; 1054 char pbuf[81]; 1055 void (*pr)(const char *, ...); 1056 int i; 1057 1058 if ((boothowto & AB_SILENT) != 0) { 1059 snprintf(pbuf, sizeof(pbuf), "%s %s (%s)", 1060 ostype, osrelease, kernel_ident); 1061 printf_nolog("%s", pbuf); 1062 for (i = 80 - strlen(pbuf) - sizeof(notice); i > 0; i--) 1063 printf(" "); 1064 printf_nolog("%s\n", notice); 1065 pr = aprint_normal; 1066 } else { 1067 pr = printf; 1068 } 1069 1070 memset(pbuf, 0, sizeof(pbuf)); 1071 (*pr)("%s%s", copyright, version); 1072 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)physmem)); 1073 (*pr)("total memory = %s\n", pbuf); 1074 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)uvmexp.free)); 1075 (*pr)("avail memory = %s\n", pbuf); 1076 } 1077