1 /* $NetBSD: init_main.c,v 1.420 2010/06/10 20:54:53 pooka 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 1.420 2010/06/10 20:54:53 pooka 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> 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 void main(void); 270 271 /* 272 * System startup; initialize the world, create process 0, mount root 273 * filesystem, and fork to create init and pagedaemon. Most of the 274 * hard work is done in the lower-level initialization routines including 275 * startup(), which does memory initialization and autoconfiguration. 276 */ 277 void 278 main(void) 279 { 280 struct timespec time; 281 struct lwp *l; 282 struct proc *p; 283 int s, error; 284 #ifdef NVNODE_IMPLICIT 285 int usevnodes; 286 #endif 287 CPU_INFO_ITERATOR cii; 288 struct cpu_info *ci; 289 290 l = &lwp0; 291 #ifndef LWP0_CPU_INFO 292 l->l_cpu = curcpu(); 293 #endif 294 l->l_pflag |= LP_RUNNING; 295 296 /* 297 * Attempt to find console and initialize 298 * in case of early panic or other messages. 299 */ 300 consinit(); 301 302 kernel_lock_init(); 303 once_init(); 304 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 305 306 /* Initialize the device switch tables. */ 307 devsw_init(); 308 309 uvm_init(); 310 311 prop_kern_init(); 312 313 #if ((NKSYMS > 0) || (NDDB > 0) || (NMODULAR > 0)) 314 ksyms_init(); 315 #endif 316 kprintf_init(); 317 318 percpu_init(); 319 320 /* Initialize lock caches. */ 321 mutex_obj_init(); 322 rw_obj_init(); 323 324 /* Initialize the extent manager. */ 325 extent_init(); 326 327 /* Do machine-dependent initialization. */ 328 cpu_startup(); 329 330 /* Initialize the sysctl subsystem. */ 331 sysctl_init(); 332 333 /* Initialize callouts, part 1. */ 334 callout_startup(); 335 336 /* Initialize the kernel authorization subsystem. */ 337 kauth_init(); 338 339 spec_init(); 340 341 /* 342 * Set BPF op vector. Can't do this in bpf attach, since 343 * network drivers attach before bpf. 344 */ 345 bpf_setops(); 346 347 /* Start module system. */ 348 module_init(); 349 350 /* 351 * Initialize the kernel authorization subsystem and start the 352 * default security model, if any. We need to do this early 353 * enough so that subsystems relying on any of the aforementioned 354 * can work properly. Since the security model may dictate the 355 * credential inheritance policy, it is needed at least before 356 * any process is created, specifically proc0. 357 */ 358 module_init_class(MODULE_CLASS_SECMODEL); 359 360 /* Initialize the buffer cache */ 361 bufinit(); 362 363 /* Initialize sockets. */ 364 soinit(); 365 366 /* 367 * The following things must be done before autoconfiguration. 368 */ 369 evcnt_init(); /* initialize event counters */ 370 #if NRND > 0 371 rnd_init(); /* initialize random number generator */ 372 #endif 373 374 /* Initialize process and pgrp structures. */ 375 #ifdef KERN_SA 376 sa_init(); 377 #endif 378 procinit(); 379 lwpinit(); 380 381 /* Initialize signal-related data structures. */ 382 signal_init(); 383 384 /* Initialize resource management. */ 385 resource_init(); 386 387 /* Create process 0. */ 388 proc0_init(); 389 lwp0_init(); 390 391 /* Disable preemption during boot. */ 392 kpreempt_disable(); 393 394 /* Initialize the UID hash table. */ 395 uid_init(); 396 397 /* Charge root for one process. */ 398 (void)chgproccnt(0, 1); 399 400 /* Initialize timekeeping. */ 401 time_init(); 402 403 /* Initialize the run queues, turnstiles and sleep queues. */ 404 sched_rqinit(); 405 turnstile_init(); 406 sleeptab_init(&sleeptab); 407 408 sched_init(); 409 410 /* Initialize processor-sets */ 411 psets_init(); 412 413 /* MI initialization of the boot cpu */ 414 error = mi_cpu_attach(curcpu()); 415 KASSERT(error == 0); 416 417 /* Initialize timekeeping, part 2. */ 418 time_init2(); 419 420 /* 421 * Initialize mbuf's. Do this now because we might attempt to 422 * allocate mbufs or mbuf clusters during autoconfiguration. 423 */ 424 mbinit(); 425 426 /* Initialize I/O statistics. */ 427 iostat_init(); 428 429 /* Initialize the log device. */ 430 loginit(); 431 432 /* Second part of module system initialization. */ 433 module_init2(); 434 435 /* Initialize the file systems. */ 436 #ifdef NVNODE_IMPLICIT 437 /* 438 * If maximum number of vnodes in namei vnode cache is not explicitly 439 * defined in kernel config, adjust the number such as we use roughly 440 * 10% of memory for vnodes and associated data structures in the 441 * assumed worst case. Do not provide fewer than NVNODE vnodes. 442 */ 443 usevnodes = 444 calc_cache_size(kernel_map, 10, VNODE_VA_MAXPCT) / VNODE_COST; 445 if (usevnodes > desiredvnodes) 446 desiredvnodes = usevnodes; 447 #endif 448 vfsinit(); 449 lf_init(); 450 451 /* Initialize fstrans. */ 452 fstrans_init(); 453 454 /* Initialize the file descriptor system. */ 455 fd_sys_init(); 456 457 /* Initialize cwd structures */ 458 cwd_sys_init(); 459 460 /* Initialize kqueue. */ 461 kqueue_init(); 462 463 /* Initialize the system monitor subsystems. */ 464 #if NSYSMON_TASKQ > 0 465 sysmon_task_queue_preinit(); 466 #endif 467 468 #if NSYSMON_ENVSYS > 0 469 sysmon_envsys_init(); 470 #endif 471 472 #if NSYSMON_POWER > 0 473 sysmon_power_init(); 474 #endif 475 476 #if NSYSMON_WDOG > 0 477 sysmon_wdog_init(); 478 #endif 479 480 inittimecounter(); 481 ntp_init(); 482 483 /* Initialize tty subsystem. */ 484 tty_init(); 485 ttyldisc_init(); 486 487 /* Initialize the buffer cache, part 2. */ 488 bufinit2(); 489 490 /* Initialize the disk wedge subsystem. */ 491 dkwedge_init(); 492 493 /* Initialize interfaces. */ 494 ifinit1(); 495 496 spldebug_start(); 497 498 /* Configure the system hardware. This will enable interrupts. */ 499 configure(); 500 501 ssp_init(); 502 503 configure2(); 504 /* Now timer is working. Enable preemption. */ 505 kpreempt_enable(); 506 507 ubc_init(); /* must be after autoconfig */ 508 509 #ifdef SYSVSHM 510 /* Initialize System V style shared memory. */ 511 shminit(); 512 #endif 513 514 vmem_rehash_start(); /* must be before exec_init */ 515 516 /* Initialize exec structures */ 517 exec_init(1); /* seminit calls exithook_establish() */ 518 519 #ifdef SYSVSEM 520 /* Initialize System V style semaphores. */ 521 seminit(); 522 #endif 523 524 #ifdef SYSVMSG 525 /* Initialize System V style message queues. */ 526 msginit(); 527 #endif 528 529 #if NVERIEXEC > 0 530 /* 531 * Initialise the Veriexec subsystem. 532 */ 533 veriexec_init(); 534 #endif /* NVERIEXEC > 0 */ 535 536 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 537 pax_init(); 538 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 539 540 #ifdef FAST_IPSEC 541 /* Attach network crypto subsystem */ 542 ipsec_attach(); 543 #endif 544 545 /* 546 * Initialize protocols. Block reception of incoming packets 547 * until everything is ready. 548 */ 549 s = splnet(); 550 ifinit(); 551 domaininit(true); 552 if_attachdomain(); 553 splx(s); 554 555 #ifdef GPROF 556 /* Initialize kernel profiling. */ 557 kmstartup(); 558 #endif 559 560 /* Initialize system accounting. */ 561 acct_init(); 562 563 #ifndef PIPE_SOCKETPAIR 564 /* Initialize pipes. */ 565 pipe_init(); 566 #endif 567 568 #ifdef KTRACE 569 /* Initialize ktrace. */ 570 ktrinit(); 571 #endif 572 573 #ifdef PTRACE 574 /* Initialize ptrace. */ 575 ptrace_init(); 576 #endif /* PTRACE */ 577 578 /* Initialize the UUID system calls. */ 579 uuid_init(); 580 581 machdep_init(); 582 583 /* 584 * Create process 1 (init(8)). We do this now, as Unix has 585 * historically had init be process 1, and changing this would 586 * probably upset a lot of people. 587 * 588 * Note that process 1 won't immediately exec init(8), but will 589 * wait for us to inform it that the root file system has been 590 * mounted. 591 */ 592 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 593 panic("fork init"); 594 595 /* 596 * Load any remaining builtin modules, and hand back temporary 597 * storage to the VM system. 598 */ 599 module_init_class(MODULE_CLASS_ANY); 600 601 /* 602 * Finalize configuration now that all real devices have been 603 * found. This needs to be done before the root device is 604 * selected, since finalization may create the root device. 605 */ 606 config_finalize(); 607 608 sysctl_finalize(); 609 610 /* 611 * Now that autoconfiguration has completed, we can determine 612 * the root and dump devices. 613 */ 614 cpu_rootconf(); 615 cpu_dumpconf(); 616 617 /* Mount the root file system. */ 618 do { 619 domountroothook(root_device); 620 if ((error = vfs_mountroot())) { 621 printf("cannot mount root, error = %d\n", error); 622 boothowto |= RB_ASKNAME; 623 setroot(root_device, 624 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 625 } 626 } while (error != 0); 627 mountroothook_destroy(); 628 629 /* 630 * Initialise the time-of-day clock, passing the time recorded 631 * in the root filesystem (if any) for use by systems that 632 * don't have a non-volatile time-of-day device. 633 */ 634 inittodr(rootfstime); 635 636 /* 637 * Now can look at time, having had a chance to verify the time 638 * from the file system. Reset l->l_rtime as it may have been 639 * munched in mi_switch() after the time got set. 640 */ 641 getnanotime(&time); 642 boottime = time; 643 #ifdef COMPAT_50 644 { 645 struct timeval tv; 646 TIMESPEC_TO_TIMEVAL(&tv, &time); 647 timeval_to_timeval50(&tv, &boottime50); 648 } 649 #endif 650 mutex_enter(proc_lock); 651 LIST_FOREACH(p, &allproc, p_list) { 652 KASSERT((p->p_flag & PK_MARKER) == 0); 653 mutex_enter(p->p_lock); 654 TIMESPEC_TO_TIMEVAL(&p->p_stats->p_start, &time); 655 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 656 lwp_lock(l); 657 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 658 lwp_unlock(l); 659 } 660 mutex_exit(p->p_lock); 661 } 662 mutex_exit(proc_lock); 663 binuptime(&curlwp->l_stime); 664 665 for (CPU_INFO_FOREACH(cii, ci)) { 666 ci->ci_schedstate.spc_lastmod = time_second; 667 } 668 669 /* Create the pageout daemon kernel thread. */ 670 uvm_swap_init(); 671 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 672 NULL, NULL, "pgdaemon")) 673 panic("fork pagedaemon"); 674 675 /* Create the filesystem syncer kernel thread. */ 676 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 677 NULL, NULL, "ioflush")) 678 panic("fork syncer"); 679 680 /* Create the aiodone daemon kernel thread. */ 681 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 682 uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE)) 683 panic("fork aiodoned"); 684 685 /* 686 * Okay, now we can let init(8) exec! It's off to userland! 687 */ 688 mutex_enter(proc_lock); 689 start_init_exec = 1; 690 cv_broadcast(&lbolt); 691 mutex_exit(proc_lock); 692 693 /* The scheduler is an infinite loop. */ 694 uvm_scheduler(); 695 /* NOTREACHED */ 696 } 697 698 /* 699 * Configure the system's hardware. 700 */ 701 static void 702 configure(void) 703 { 704 705 /* Initialize autoconf data structures. */ 706 config_init_mi(); 707 /* 708 * XXX 709 * callout_setfunc() requires mutex(9) so it can't be in config_init() 710 * on amiga and atari which use config_init() and autoconf(9) fucntions 711 * to initialize console devices. 712 */ 713 config_twiddle_init(); 714 715 pmf_init(); 716 #if NDRVCTL > 0 717 drvctl_init(); 718 #endif 719 720 #ifdef USERCONF 721 if (boothowto & RB_USERCONF) 722 user_config(); 723 #endif 724 725 if ((boothowto & (AB_SILENT|AB_VERBOSE)) == AB_SILENT) { 726 printf_nolog("Detecting hardware..."); 727 } 728 729 /* 730 * Do the machine-dependent portion of autoconfiguration. This 731 * sets the configuration machinery here in motion by "finding" 732 * the root bus. When this function returns, we expect interrupts 733 * to be enabled. 734 */ 735 cpu_configure(); 736 } 737 738 static void 739 configure2(void) 740 { 741 CPU_INFO_ITERATOR cii; 742 struct cpu_info *ci; 743 int s; 744 745 /* 746 * Now that we've found all the hardware, start the real time 747 * and statistics clocks. 748 */ 749 initclocks(); 750 751 cold = 0; /* clocks are running, we're warm now! */ 752 s = splsched(); 753 curcpu()->ci_schedstate.spc_flags |= SPCF_RUNNING; 754 splx(s); 755 756 /* Boot the secondary processors. */ 757 for (CPU_INFO_FOREACH(cii, ci)) { 758 uvm_cpu_attach(ci); 759 } 760 mp_online = true; 761 #if defined(MULTIPROCESSOR) 762 cpu_boot_secondary_processors(); 763 #endif 764 765 /* Setup the runqueues and scheduler. */ 766 runq_init(); 767 synch_init(); 768 769 /* 770 * Bus scans can make it appear as if the system has paused, so 771 * twiddle constantly while config_interrupts() jobs are running. 772 */ 773 config_twiddle_fn(NULL); 774 775 /* 776 * Create threads to call back and finish configuration for 777 * devices that want interrupts enabled. 778 */ 779 config_create_interruptthreads(); 780 781 /* Get the threads going and into any sleeps before continuing. */ 782 yield(); 783 } 784 785 static void 786 check_console(struct lwp *l) 787 { 788 struct vnode *vp; 789 int error; 790 791 error = namei_simple_kernel("/dev/console", 792 NSM_FOLLOW_NOEMULROOT, &vp); 793 if (error == 0) 794 vrele(vp); 795 else if (error == ENOENT) 796 printf("warning: no /dev/console\n"); 797 else 798 printf("warning: lookup /dev/console: error %d\n", error); 799 } 800 801 /* 802 * List of paths to try when searching for "init". 803 */ 804 static const char * const initpaths[] = { 805 "/sbin/init", 806 "/sbin/oinit", 807 "/sbin/init.bak", 808 NULL, 809 }; 810 811 /* 812 * Start the initial user process; try exec'ing each pathname in "initpaths". 813 * The program is invoked with one argument containing the boot flags. 814 */ 815 static void 816 start_init(void *arg) 817 { 818 struct lwp *l = arg; 819 struct proc *p = l->l_proc; 820 vaddr_t addr; 821 struct sys_execve_args /* { 822 syscallarg(const char *) path; 823 syscallarg(char * const *) argp; 824 syscallarg(char * const *) envp; 825 } */ args; 826 int options, i, error; 827 register_t retval[2]; 828 char flags[4], *flagsp; 829 const char *path, *slash; 830 char *ucp, **uap, *arg0, *arg1 = NULL; 831 char ipath[129]; 832 int ipx, len; 833 834 /* 835 * Now in process 1. 836 */ 837 strncpy(p->p_comm, "init", MAXCOMLEN); 838 839 /* 840 * Wait for main() to tell us that it's safe to exec. 841 */ 842 mutex_enter(proc_lock); 843 while (start_init_exec == 0) 844 cv_wait(&lbolt, proc_lock); 845 mutex_exit(proc_lock); 846 847 /* 848 * This is not the right way to do this. We really should 849 * hand-craft a descriptor onto /dev/console to hand to init, 850 * but that's a _lot_ more work, and the benefit from this easy 851 * hack makes up for the "good is the enemy of the best" effect. 852 */ 853 check_console(l); 854 855 /* 856 * Need just enough stack to hold the faked-up "execve()" arguments. 857 */ 858 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 859 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 860 NULL, UVM_UNKNOWN_OFFSET, 0, 861 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 862 UVM_ADV_NORMAL, 863 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 864 panic("init: couldn't allocate argument space"); 865 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 866 867 ipx = 0; 868 while (1) { 869 if (boothowto & RB_ASKNAME) { 870 printf("init path"); 871 if (initpaths[ipx]) 872 printf(" (default %s)", initpaths[ipx]); 873 printf(": "); 874 len = cngetsn(ipath, sizeof(ipath)-1); 875 if (len == 4 && strcmp(ipath, "halt") == 0) { 876 cpu_reboot(RB_HALT, NULL); 877 } else if (len == 6 && strcmp(ipath, "reboot") == 0) { 878 cpu_reboot(0, NULL); 879 #if defined(DDB) 880 } else if (len == 3 && strcmp(ipath, "ddb") == 0) { 881 console_debugger(); 882 continue; 883 #endif 884 } else if (len > 0 && ipath[0] == '/') { 885 ipath[len] = '\0'; 886 path = ipath; 887 } else if (len == 0 && initpaths[ipx] != NULL) { 888 path = initpaths[ipx++]; 889 } else { 890 printf("use absolute path, "); 891 #if defined(DDB) 892 printf("\"ddb\", "); 893 #endif 894 printf("\"halt\", or \"reboot\"\n"); 895 continue; 896 } 897 } else { 898 if ((path = initpaths[ipx++]) == NULL) { 899 ipx = 0; 900 boothowto |= RB_ASKNAME; 901 continue; 902 } 903 } 904 905 ucp = (char *)USRSTACK; 906 907 /* 908 * Construct the boot flag argument. 909 */ 910 flagsp = flags; 911 *flagsp++ = '-'; 912 options = 0; 913 914 if (boothowto & RB_SINGLE) { 915 *flagsp++ = 's'; 916 options = 1; 917 } 918 #ifdef notyet 919 if (boothowto & RB_FASTBOOT) { 920 *flagsp++ = 'f'; 921 options = 1; 922 } 923 #endif 924 925 /* 926 * Move out the flags (arg 1), if necessary. 927 */ 928 if (options != 0) { 929 *flagsp++ = '\0'; 930 i = flagsp - flags; 931 #ifdef DEBUG 932 aprint_normal("init: copying out flags `%s' %d\n", flags, i); 933 #endif 934 arg1 = STACK_ALLOC(ucp, i); 935 ucp = STACK_MAX(arg1, i); 936 (void)copyout((void *)flags, arg1, i); 937 } 938 939 /* 940 * Move out the file name (also arg 0). 941 */ 942 i = strlen(path) + 1; 943 #ifdef DEBUG 944 aprint_normal("init: copying out path `%s' %d\n", path, i); 945 #else 946 if (boothowto & RB_ASKNAME || path != initpaths[0]) 947 printf("init: trying %s\n", path); 948 #endif 949 arg0 = STACK_ALLOC(ucp, i); 950 ucp = STACK_MAX(arg0, i); 951 (void)copyout(path, arg0, i); 952 953 /* 954 * Move out the arg pointers. 955 */ 956 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 957 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 958 SCARG(&args, path) = arg0; 959 SCARG(&args, argp) = uap; 960 SCARG(&args, envp) = NULL; 961 slash = strrchr(path, '/'); 962 if (slash) 963 (void)suword((void *)uap++, 964 (long)arg0 + (slash + 1 - path)); 965 else 966 (void)suword((void *)uap++, (long)arg0); 967 if (options != 0) 968 (void)suword((void *)uap++, (long)arg1); 969 (void)suword((void *)uap++, 0); /* terminator */ 970 971 /* 972 * Now try to exec the program. If can't for any reason 973 * other than it doesn't exist, complain. 974 */ 975 error = sys_execve(l, &args, retval); 976 if (error == 0 || error == EJUSTRETURN) { 977 KERNEL_UNLOCK_LAST(l); 978 return; 979 } 980 printf("exec %s: error %d\n", path, error); 981 } 982 printf("init: not found\n"); 983 panic("no init"); 984 } 985 986 /* 987 * calculate cache size from physmem and vm_map size. 988 */ 989 vaddr_t 990 calc_cache_size(struct vm_map *map, int pct, int va_pct) 991 { 992 paddr_t t; 993 994 /* XXX should consider competing cache if any */ 995 /* XXX should consider submaps */ 996 t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; 997 if (map != NULL) { 998 vsize_t vsize; 999 1000 vsize = vm_map_max(map) - vm_map_min(map); 1001 vsize = (uintmax_t)vsize * va_pct / 100; 1002 if (t > vsize) { 1003 t = vsize; 1004 } 1005 } 1006 return t; 1007 } 1008 1009 /* 1010 * Print the system start up banner. 1011 * 1012 * - Print a limited banner if AB_SILENT. 1013 * - Always send normal banner to the log. 1014 */ 1015 #define MEM_PBUFSIZE sizeof("99999 MB") 1016 1017 void 1018 banner(void) 1019 { 1020 static char notice[] = " Notice: this software is " 1021 "protected by copyright"; 1022 char pbuf[81]; 1023 void (*pr)(const char *, ...); 1024 int i; 1025 1026 if ((boothowto & AB_SILENT) != 0) { 1027 snprintf(pbuf, sizeof(pbuf), "%s %s (%s)", 1028 ostype, osrelease, kernel_ident); 1029 printf_nolog("%s", pbuf); 1030 for (i = 80 - strlen(pbuf) - sizeof(notice); i > 0; i--) 1031 printf(" "); 1032 printf_nolog("%s\n", notice); 1033 pr = aprint_normal; 1034 } else { 1035 pr = printf; 1036 } 1037 1038 memset(pbuf, 0, sizeof(pbuf)); 1039 (*pr)("%s%s", copyright, version); 1040 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)physmem)); 1041 (*pr)("total memory = %s\n", pbuf); 1042 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)uvmexp.free)); 1043 (*pr)("avail memory = %s\n", pbuf); 1044 } 1045