1 /* $NetBSD: init_main.c,v 1.430 2011/05/26 04:25:26 uebayasi 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 1.430 2011/05/26 04:25:26 uebayasi 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 kthread_sysinit(); 308 309 /* Initialize the device switch tables. */ 310 devsw_init(); 311 312 uvm_init(); 313 314 prop_kern_init(); 315 316 #if ((NKSYMS > 0) || (NDDB > 0) || (NMODULAR > 0)) 317 ksyms_init(); 318 #endif 319 kprintf_init(); 320 321 percpu_init(); 322 323 /* Initialize lock caches. */ 324 mutex_obj_init(); 325 rw_obj_init(); 326 327 /* Initialize the extent manager. */ 328 extent_init(); 329 330 /* Initialize event counters */ 331 evcnt_init(); 332 333 /* Do machine-dependent initialization. */ 334 cpu_startup(); 335 336 /* Initialize the sysctl subsystem. */ 337 sysctl_init(); 338 339 /* Initialize callouts, part 1. */ 340 callout_startup(); 341 342 /* Initialize the kernel authorization subsystem. */ 343 kauth_init(); 344 345 spec_init(); 346 347 /* 348 * Set BPF op vector. Can't do this in bpf attach, since 349 * network drivers attach before bpf. 350 */ 351 bpf_setops(); 352 353 /* Start module system. */ 354 module_init(); 355 356 /* 357 * Initialize the kernel authorization subsystem and start the 358 * default security model, if any. We need to do this early 359 * enough so that subsystems relying on any of the aforementioned 360 * can work properly. Since the security model may dictate the 361 * credential inheritance policy, it is needed at least before 362 * any process is created, specifically proc0. 363 */ 364 module_init_class(MODULE_CLASS_SECMODEL); 365 366 /* Initialize the buffer cache */ 367 bufinit(); 368 369 /* Initialize sockets. */ 370 soinit(); 371 372 /* 373 * The following things must be done before autoconfiguration. 374 */ 375 #if NRND > 0 376 rnd_init(); /* initialize random number generator */ 377 #endif 378 379 /* Initialize process and pgrp structures. */ 380 #ifdef KERN_SA 381 sa_init(); 382 #endif 383 procinit(); 384 lwpinit(); 385 386 /* Initialize signal-related data structures. */ 387 signal_init(); 388 389 /* Initialize resource management. */ 390 resource_init(); 391 392 /* Create process 0. */ 393 proc0_init(); 394 lwp0_init(); 395 396 /* Disable preemption during boot. */ 397 kpreempt_disable(); 398 399 /* Initialize the UID hash table. */ 400 uid_init(); 401 402 /* Charge root for one process. */ 403 (void)chgproccnt(0, 1); 404 405 /* Initialize timekeeping. */ 406 time_init(); 407 408 /* Initialize the run queues, turnstiles and sleep queues. */ 409 sched_rqinit(); 410 turnstile_init(); 411 sleeptab_init(&sleeptab); 412 413 sched_init(); 414 415 /* Initialize processor-sets */ 416 psets_init(); 417 418 /* MI initialization of the boot cpu */ 419 error = mi_cpu_attach(curcpu()); 420 KASSERT(error == 0); 421 422 /* Initialize timekeeping, part 2. */ 423 time_init2(); 424 425 /* 426 * Initialize mbuf's. Do this now because we might attempt to 427 * allocate mbufs or mbuf clusters during autoconfiguration. 428 */ 429 mbinit(); 430 431 /* Initialize I/O statistics. */ 432 iostat_init(); 433 434 /* Initialize the log device. */ 435 loginit(); 436 437 /* Second part of module system initialization. */ 438 module_start_unload_thread(); 439 440 /* Initialize the file systems. */ 441 #ifdef NVNODE_IMPLICIT 442 /* 443 * If maximum number of vnodes in namei vnode cache is not explicitly 444 * defined in kernel config, adjust the number such as we use roughly 445 * 10% of memory for vnodes and associated data structures in the 446 * assumed worst case. Do not provide fewer than NVNODE vnodes. 447 */ 448 usevnodes = 449 calc_cache_size(kernel_map, 10, VNODE_VA_MAXPCT) / VNODE_COST; 450 if (usevnodes > desiredvnodes) 451 desiredvnodes = usevnodes; 452 #endif 453 vfsinit(); 454 lf_init(); 455 456 /* Initialize fstrans. */ 457 fstrans_init(); 458 459 /* Initialize the file descriptor system. */ 460 fd_sys_init(); 461 462 /* Initialize cwd structures */ 463 cwd_sys_init(); 464 465 /* Initialize kqueue. */ 466 kqueue_init(); 467 468 /* Initialize the system monitor subsystems. */ 469 #if NSYSMON_TASKQ > 0 470 sysmon_task_queue_preinit(); 471 #endif 472 473 #if NSYSMON_ENVSYS > 0 474 sysmon_envsys_init(); 475 #endif 476 477 #if NSYSMON_POWER > 0 478 sysmon_power_init(); 479 #endif 480 481 #if NSYSMON_WDOG > 0 482 sysmon_wdog_init(); 483 #endif 484 485 inittimecounter(); 486 ntp_init(); 487 488 /* Initialize tty subsystem. */ 489 tty_init(); 490 ttyldisc_init(); 491 492 /* Initialize the buffer cache, part 2. */ 493 bufinit2(); 494 495 /* Initialize the disk wedge subsystem. */ 496 dkwedge_init(); 497 498 /* Initialize interfaces. */ 499 ifinit1(); 500 501 spldebug_start(); 502 503 /* Configure the system hardware. This will enable interrupts. */ 504 configure(); 505 506 ssp_init(); 507 508 ubc_init(); /* must be after autoconfig */ 509 510 configure2(); 511 /* Now timer is working. Enable preemption. */ 512 kpreempt_enable(); 513 514 #ifdef SYSVSHM 515 /* Initialize System V style shared memory. */ 516 shminit(); 517 #endif 518 519 vmem_rehash_start(); /* must be before exec_init */ 520 521 /* Initialize exec structures */ 522 exec_init(1); /* seminit calls exithook_establish() */ 523 524 #ifdef SYSVSEM 525 /* Initialize System V style semaphores. */ 526 seminit(); 527 #endif 528 529 #ifdef SYSVMSG 530 /* Initialize System V style message queues. */ 531 msginit(); 532 #endif 533 534 #if NVERIEXEC > 0 535 /* 536 * Initialise the Veriexec subsystem. 537 */ 538 veriexec_init(); 539 #endif /* NVERIEXEC > 0 */ 540 541 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 542 pax_init(); 543 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 544 545 #ifdef FAST_IPSEC 546 /* Attach network crypto subsystem */ 547 ipsec_attach(); 548 #endif 549 550 /* 551 * Initialize protocols. Block reception of incoming packets 552 * until everything is ready. 553 */ 554 s = splnet(); 555 ifinit(); 556 domaininit(true); 557 if_attachdomain(); 558 splx(s); 559 560 #ifdef GPROF 561 /* Initialize kernel profiling. */ 562 kmstartup(); 563 #endif 564 565 /* Initialize system accounting. */ 566 acct_init(); 567 568 #ifndef PIPE_SOCKETPAIR 569 /* Initialize pipes. */ 570 pipe_init(); 571 #endif 572 573 #ifdef KTRACE 574 /* Initialize ktrace. */ 575 ktrinit(); 576 #endif 577 578 #ifdef PTRACE 579 /* Initialize ptrace. */ 580 ptrace_init(); 581 #endif /* PTRACE */ 582 583 /* Initialize the UUID system calls. */ 584 uuid_init(); 585 586 machdep_init(); 587 588 procinit_sysctl(); 589 590 /* 591 * Create process 1 (init(8)). We do this now, as Unix has 592 * historically had init be process 1, and changing this would 593 * probably upset a lot of people. 594 * 595 * Note that process 1 won't immediately exec init(8), but will 596 * wait for us to inform it that the root file system has been 597 * mounted. 598 */ 599 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 600 panic("fork init"); 601 602 /* 603 * Load any remaining builtin modules, and hand back temporary 604 * storage to the VM system. Then require force when loading any 605 * remaining un-init'ed built-in modules to avoid later surprises. 606 */ 607 module_init_class(MODULE_CLASS_ANY); 608 module_builtin_require_force(); 609 610 /* 611 * Finalize configuration now that all real devices have been 612 * found. This needs to be done before the root device is 613 * selected, since finalization may create the root device. 614 */ 615 config_finalize(); 616 617 sysctl_finalize(); 618 619 /* 620 * Now that autoconfiguration has completed, we can determine 621 * the root and dump devices. 622 */ 623 cpu_rootconf(); 624 cpu_dumpconf(); 625 626 /* Mount the root file system. */ 627 do { 628 domountroothook(root_device); 629 if ((error = vfs_mountroot())) { 630 printf("cannot mount root, error = %d\n", error); 631 boothowto |= RB_ASKNAME; 632 setroot(root_device, 633 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 634 } 635 } while (error != 0); 636 mountroothook_destroy(); 637 638 configure3(); 639 640 /* 641 * Initialise the time-of-day clock, passing the time recorded 642 * in the root filesystem (if any) for use by systems that 643 * don't have a non-volatile time-of-day device. 644 */ 645 inittodr(rootfstime); 646 647 /* 648 * Now can look at time, having had a chance to verify the time 649 * from the file system. Reset l->l_rtime as it may have been 650 * munched in mi_switch() after the time got set. 651 */ 652 getnanotime(&time); 653 boottime = time; 654 #ifdef COMPAT_50 655 { 656 struct timeval tv; 657 TIMESPEC_TO_TIMEVAL(&tv, &time); 658 timeval_to_timeval50(&tv, &boottime50); 659 } 660 #endif 661 mutex_enter(proc_lock); 662 LIST_FOREACH(p, &allproc, p_list) { 663 KASSERT((p->p_flag & PK_MARKER) == 0); 664 mutex_enter(p->p_lock); 665 TIMESPEC_TO_TIMEVAL(&p->p_stats->p_start, &time); 666 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 667 lwp_lock(l); 668 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 669 lwp_unlock(l); 670 } 671 mutex_exit(p->p_lock); 672 } 673 mutex_exit(proc_lock); 674 binuptime(&curlwp->l_stime); 675 676 for (CPU_INFO_FOREACH(cii, ci)) { 677 ci->ci_schedstate.spc_lastmod = time_second; 678 } 679 680 /* Create the pageout daemon kernel thread. */ 681 uvm_swap_init(); 682 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 683 NULL, NULL, "pgdaemon")) 684 panic("fork pagedaemon"); 685 686 /* Create the filesystem syncer kernel thread. */ 687 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 688 NULL, NULL, "ioflush")) 689 panic("fork syncer"); 690 691 /* Create the aiodone daemon kernel thread. */ 692 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 693 uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE)) 694 panic("fork aiodoned"); 695 696 /* 697 * Okay, now we can let init(8) exec! It's off to userland! 698 */ 699 mutex_enter(proc_lock); 700 start_init_exec = 1; 701 cv_broadcast(&lbolt); 702 mutex_exit(proc_lock); 703 704 /* The scheduler is an infinite loop. */ 705 uvm_scheduler(); 706 /* NOTREACHED */ 707 } 708 709 /* 710 * Configure the system's hardware. 711 */ 712 static void 713 configure(void) 714 { 715 716 /* Initialize autoconf data structures. */ 717 config_init_mi(); 718 /* 719 * XXX 720 * callout_setfunc() requires mutex(9) so it can't be in config_init() 721 * on amiga and atari which use config_init() and autoconf(9) fucntions 722 * to initialize console devices. 723 */ 724 config_twiddle_init(); 725 726 pmf_init(); 727 #if NDRVCTL > 0 728 drvctl_init(); 729 #endif 730 731 #ifdef USERCONF 732 userconf_init(); 733 if (boothowto & RB_USERCONF) 734 userconf_prompt(); 735 #endif 736 737 if ((boothowto & (AB_SILENT|AB_VERBOSE)) == AB_SILENT) { 738 printf_nolog("Detecting hardware..."); 739 } 740 741 /* 742 * Do the machine-dependent portion of autoconfiguration. This 743 * sets the configuration machinery here in motion by "finding" 744 * the root bus. When this function returns, we expect interrupts 745 * to be enabled. 746 */ 747 cpu_configure(); 748 } 749 750 static void 751 configure2(void) 752 { 753 CPU_INFO_ITERATOR cii; 754 struct cpu_info *ci; 755 int s; 756 757 /* 758 * Now that we've found all the hardware, start the real time 759 * and statistics clocks. 760 */ 761 initclocks(); 762 763 cold = 0; /* clocks are running, we're warm now! */ 764 s = splsched(); 765 curcpu()->ci_schedstate.spc_flags |= SPCF_RUNNING; 766 splx(s); 767 768 /* Boot the secondary processors. */ 769 for (CPU_INFO_FOREACH(cii, ci)) { 770 uvm_cpu_attach(ci); 771 } 772 mp_online = true; 773 #if defined(MULTIPROCESSOR) 774 cpu_boot_secondary_processors(); 775 #endif 776 777 /* Setup the runqueues and scheduler. */ 778 runq_init(); 779 synch_init(); 780 781 /* 782 * Bus scans can make it appear as if the system has paused, so 783 * twiddle constantly while config_interrupts() jobs are running. 784 */ 785 config_twiddle_fn(NULL); 786 787 /* 788 * Create threads to call back and finish configuration for 789 * devices that want interrupts enabled. 790 */ 791 config_create_interruptthreads(); 792 793 /* Get the threads going and into any sleeps before continuing. */ 794 yield(); 795 } 796 797 static void 798 configure3(void) 799 { 800 801 /* 802 * Create threads to call back and finish configuration for 803 * devices that want the mounted root file system. 804 */ 805 config_create_mountrootthreads(); 806 807 /* Get the threads going and into any sleeps before continuing. */ 808 yield(); 809 } 810 811 static void 812 check_console(struct lwp *l) 813 { 814 struct vnode *vp; 815 int error; 816 817 error = namei_simple_kernel("/dev/console", 818 NSM_FOLLOW_NOEMULROOT, &vp); 819 if (error == 0) 820 vrele(vp); 821 else if (error == ENOENT) 822 printf("warning: no /dev/console\n"); 823 else 824 printf("warning: lookup /dev/console: error %d\n", error); 825 } 826 827 /* 828 * List of paths to try when searching for "init". 829 */ 830 static const char * const initpaths[] = { 831 "/sbin/init", 832 "/sbin/oinit", 833 "/sbin/init.bak", 834 NULL, 835 }; 836 837 /* 838 * Start the initial user process; try exec'ing each pathname in "initpaths". 839 * The program is invoked with one argument containing the boot flags. 840 */ 841 static void 842 start_init(void *arg) 843 { 844 struct lwp *l = arg; 845 struct proc *p = l->l_proc; 846 vaddr_t addr; 847 struct sys_execve_args /* { 848 syscallarg(const char *) path; 849 syscallarg(char * const *) argp; 850 syscallarg(char * const *) envp; 851 } */ args; 852 int options, i, error; 853 register_t retval[2]; 854 char flags[4], *flagsp; 855 const char *path, *slash; 856 char *ucp, **uap, *arg0, *arg1 = NULL; 857 char ipath[129]; 858 int ipx, len; 859 860 /* 861 * Now in process 1. 862 */ 863 strncpy(p->p_comm, "init", MAXCOMLEN); 864 865 /* 866 * Wait for main() to tell us that it's safe to exec. 867 */ 868 mutex_enter(proc_lock); 869 while (start_init_exec == 0) 870 cv_wait(&lbolt, proc_lock); 871 mutex_exit(proc_lock); 872 873 /* 874 * This is not the right way to do this. We really should 875 * hand-craft a descriptor onto /dev/console to hand to init, 876 * but that's a _lot_ more work, and the benefit from this easy 877 * hack makes up for the "good is the enemy of the best" effect. 878 */ 879 check_console(l); 880 881 /* 882 * Need just enough stack to hold the faked-up "execve()" arguments. 883 */ 884 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 885 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 886 NULL, UVM_UNKNOWN_OFFSET, 0, 887 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 888 UVM_ADV_NORMAL, 889 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 890 panic("init: couldn't allocate argument space"); 891 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 892 893 ipx = 0; 894 while (1) { 895 if (boothowto & RB_ASKNAME) { 896 printf("init path"); 897 if (initpaths[ipx]) 898 printf(" (default %s)", initpaths[ipx]); 899 printf(": "); 900 len = cngetsn(ipath, sizeof(ipath)-1); 901 if (len == 4 && strcmp(ipath, "halt") == 0) { 902 cpu_reboot(RB_HALT, NULL); 903 } else if (len == 6 && strcmp(ipath, "reboot") == 0) { 904 cpu_reboot(0, NULL); 905 #if defined(DDB) 906 } else if (len == 3 && strcmp(ipath, "ddb") == 0) { 907 console_debugger(); 908 continue; 909 #endif 910 } else if (len > 0 && ipath[0] == '/') { 911 ipath[len] = '\0'; 912 path = ipath; 913 } else if (len == 0 && initpaths[ipx] != NULL) { 914 path = initpaths[ipx++]; 915 } else { 916 printf("use absolute path, "); 917 #if defined(DDB) 918 printf("\"ddb\", "); 919 #endif 920 printf("\"halt\", or \"reboot\"\n"); 921 continue; 922 } 923 } else { 924 if ((path = initpaths[ipx++]) == NULL) { 925 ipx = 0; 926 boothowto |= RB_ASKNAME; 927 continue; 928 } 929 } 930 931 ucp = (char *)USRSTACK; 932 933 /* 934 * Construct the boot flag argument. 935 */ 936 flagsp = flags; 937 *flagsp++ = '-'; 938 options = 0; 939 940 if (boothowto & RB_SINGLE) { 941 *flagsp++ = 's'; 942 options = 1; 943 } 944 #ifdef notyet 945 if (boothowto & RB_FASTBOOT) { 946 *flagsp++ = 'f'; 947 options = 1; 948 } 949 #endif 950 951 /* 952 * Move out the flags (arg 1), if necessary. 953 */ 954 if (options != 0) { 955 *flagsp++ = '\0'; 956 i = flagsp - flags; 957 #ifdef DEBUG 958 aprint_normal("init: copying out flags `%s' %d\n", flags, i); 959 #endif 960 arg1 = STACK_ALLOC(ucp, i); 961 ucp = STACK_MAX(arg1, i); 962 (void)copyout((void *)flags, arg1, i); 963 } 964 965 /* 966 * Move out the file name (also arg 0). 967 */ 968 i = strlen(path) + 1; 969 #ifdef DEBUG 970 aprint_normal("init: copying out path `%s' %d\n", path, i); 971 #else 972 if (boothowto & RB_ASKNAME || path != initpaths[0]) 973 printf("init: trying %s\n", path); 974 #endif 975 arg0 = STACK_ALLOC(ucp, i); 976 ucp = STACK_MAX(arg0, i); 977 (void)copyout(path, arg0, i); 978 979 /* 980 * Move out the arg pointers. 981 */ 982 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 983 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 984 SCARG(&args, path) = arg0; 985 SCARG(&args, argp) = uap; 986 SCARG(&args, envp) = NULL; 987 slash = strrchr(path, '/'); 988 if (slash) 989 (void)suword((void *)uap++, 990 (long)arg0 + (slash + 1 - path)); 991 else 992 (void)suword((void *)uap++, (long)arg0); 993 if (options != 0) 994 (void)suword((void *)uap++, (long)arg1); 995 (void)suword((void *)uap++, 0); /* terminator */ 996 997 /* 998 * Now try to exec the program. If can't for any reason 999 * other than it doesn't exist, complain. 1000 */ 1001 error = sys_execve(l, &args, retval); 1002 if (error == 0 || error == EJUSTRETURN) { 1003 KERNEL_UNLOCK_LAST(l); 1004 return; 1005 } 1006 printf("exec %s: error %d\n", path, error); 1007 } 1008 printf("init: not found\n"); 1009 panic("no init"); 1010 } 1011 1012 /* 1013 * calculate cache size (in bytes) from physmem and vm_map size. 1014 */ 1015 vaddr_t 1016 calc_cache_size(struct vm_map *map, int pct, int va_pct) 1017 { 1018 paddr_t t; 1019 1020 /* XXX should consider competing cache if any */ 1021 /* XXX should consider submaps */ 1022 t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; 1023 if (map != NULL) { 1024 vsize_t vsize; 1025 1026 vsize = vm_map_max(map) - vm_map_min(map); 1027 vsize = (uintmax_t)vsize * va_pct / 100; 1028 if (t > vsize) { 1029 t = vsize; 1030 } 1031 } 1032 return t; 1033 } 1034 1035 /* 1036 * Print the system start up banner. 1037 * 1038 * - Print a limited banner if AB_SILENT. 1039 * - Always send normal banner to the log. 1040 */ 1041 #define MEM_PBUFSIZE sizeof("99999 MB") 1042 1043 void 1044 banner(void) 1045 { 1046 static char notice[] = " Notice: this software is " 1047 "protected by copyright"; 1048 char pbuf[81]; 1049 void (*pr)(const char *, ...); 1050 int i; 1051 1052 if ((boothowto & AB_SILENT) != 0) { 1053 snprintf(pbuf, sizeof(pbuf), "%s %s (%s)", 1054 ostype, osrelease, kernel_ident); 1055 printf_nolog("%s", pbuf); 1056 for (i = 80 - strlen(pbuf) - sizeof(notice); i > 0; i--) 1057 printf(" "); 1058 printf_nolog("%s\n", notice); 1059 pr = aprint_normal; 1060 } else { 1061 pr = printf; 1062 } 1063 1064 memset(pbuf, 0, sizeof(pbuf)); 1065 (*pr)("%s%s", copyright, version); 1066 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)physmem)); 1067 (*pr)("total memory = %s\n", pbuf); 1068 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)uvmexp.free)); 1069 (*pr)("avail memory = %s\n", pbuf); 1070 } 1071