1 /* $NetBSD: init_main.c,v 1.489 2017/01/05 03:22:20 pgoyette 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 1.489 2017/01/05 03:22:20 pgoyette Exp $"); 101 102 #include "opt_ddb.h" 103 #include "opt_inet.h" 104 #include "opt_ipsec.h" 105 #include "opt_modular.h" 106 #include "opt_ntp.h" 107 #include "opt_pipe.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 #include "opt_rnd_printf.h" 117 #include "opt_splash.h" 118 #include "opt_kernhist.h" 119 120 #if defined(SPLASHSCREEN) && defined(makeoptions_SPLASHSCREEN_IMAGE) 121 extern void *_binary_splash_image_start; 122 extern void *_binary_splash_image_end; 123 #endif 124 125 #include "ksyms.h" 126 127 #include "veriexec.h" 128 129 #include <sys/param.h> 130 #include <sys/acct.h> 131 #include <sys/filedesc.h> 132 #include <sys/file.h> 133 #include <sys/errno.h> 134 #include <sys/callout.h> 135 #include <sys/cpu.h> 136 #include <sys/cpufreq.h> 137 #include <sys/spldebug.h> 138 #include <sys/kernel.h> 139 #include <sys/mount.h> 140 #include <sys/proc.h> 141 #include <sys/kthread.h> 142 #include <sys/resourcevar.h> 143 #include <sys/signalvar.h> 144 #include <sys/systm.h> 145 #include <sys/vnode.h> 146 #include <sys/fstrans.h> 147 #include <sys/tty.h> 148 #include <sys/conf.h> 149 #include <sys/disklabel.h> 150 #include <sys/buf.h> 151 #include <sys/device.h> 152 #include <sys/exec.h> 153 #include <sys/socketvar.h> 154 #include <sys/protosw.h> 155 #include <sys/percpu.h> 156 #include <sys/pserialize.h> 157 #include <sys/pset.h> 158 #include <sys/sysctl.h> 159 #include <sys/reboot.h> 160 #include <sys/event.h> 161 #include <sys/mbuf.h> 162 #include <sys/sched.h> 163 #include <sys/sleepq.h> 164 #include <sys/ipi.h> 165 #include <sys/iostat.h> 166 #include <sys/vmem.h> 167 #include <sys/uuid.h> 168 #include <sys/extent.h> 169 #include <sys/disk.h> 170 #include <sys/msgbuf.h> 171 #include <sys/module.h> 172 #include <sys/event.h> 173 #include <sys/lockf.h> 174 #include <sys/once.h> 175 #include <sys/kcpuset.h> 176 #include <sys/ksyms.h> 177 #include <sys/uidinfo.h> 178 #include <sys/kprintf.h> 179 #include <sys/bufq.h> 180 #ifdef IPSEC 181 #include <netipsec/ipsec.h> 182 #endif 183 #include <sys/domain.h> 184 #include <sys/namei.h> 185 #include <sys/rnd.h> 186 #include <sys/pipe.h> 187 #if NVERIEXEC > 0 188 #include <sys/verified_exec.h> 189 #endif /* NVERIEXEC > 0 */ 190 #ifdef KTRACE 191 #include <sys/ktrace.h> 192 #endif 193 #include <sys/kauth.h> 194 #include <net80211/ieee80211_netbsd.h> 195 #include <sys/cprng.h> 196 197 #include <sys/syscall.h> 198 #include <sys/syscallargs.h> 199 200 #include <sys/pax.h> 201 202 #include <secmodel/secmodel.h> 203 204 #include <ufs/ufs/quota.h> 205 206 #include <miscfs/genfs/genfs.h> 207 #include <miscfs/specfs/specdev.h> 208 209 #include <sys/cpu.h> 210 211 #include <uvm/uvm.h> /* extern struct uvm uvm */ 212 213 #include <dev/cons.h> 214 #include <dev/splash/splash.h> 215 216 #include <net/bpf.h> 217 #include <net/if.h> 218 #include <net/raw_cb.h> 219 #include <net/if_llatbl.h> 220 221 #include <prop/proplib.h> 222 223 #include <sys/userconf.h> 224 225 extern struct lwp lwp0; 226 extern time_t rootfstime; 227 228 #ifndef curlwp 229 struct lwp *curlwp = &lwp0; 230 #endif 231 struct proc *initproc; 232 233 struct vnode *rootvp, *swapdev_vp; 234 int boothowto; 235 int cold = 1; /* still working on startup */ 236 struct timespec boottime; /* time at system startup - will only follow settime deltas */ 237 238 int start_init_exec; /* semaphore for start_init() */ 239 240 cprng_strong_t *kern_cprng; 241 242 static void check_console(struct lwp *l); 243 static void start_init(void *); 244 static void configure(void); 245 static void configure2(void); 246 static void configure3(void); 247 void main(void); 248 249 /* 250 * System startup; initialize the world, create process 0, mount root 251 * filesystem, and fork to create init and pagedaemon. Most of the 252 * hard work is done in the lower-level initialization routines including 253 * startup(), which does memory initialization and autoconfiguration. 254 */ 255 void 256 main(void) 257 { 258 struct timespec time; 259 struct lwp *l; 260 struct proc *p; 261 int s, error; 262 #ifdef NVNODE_IMPLICIT 263 int usevnodes; 264 #endif 265 CPU_INFO_ITERATOR cii; 266 struct cpu_info *ci; 267 268 l = &lwp0; 269 #ifndef LWP0_CPU_INFO 270 l->l_cpu = curcpu(); 271 #endif 272 l->l_pflag |= LP_RUNNING; 273 274 /* 275 * Attempt to find console and initialize 276 * in case of early panic or other messages. 277 */ 278 consinit(); 279 280 kernel_lock_init(); 281 once_init(); 282 283 mi_cpu_init(); 284 kernconfig_lock_init(); 285 kthread_sysinit(); 286 287 /* Initialize the device switch tables. */ 288 devsw_init(); 289 290 /* Initialize event counters. */ 291 evcnt_init(); 292 293 uvm_init(); 294 ubchist_init(); 295 kcpuset_sysinit(); 296 297 prop_kern_init(); 298 299 #if ((NKSYMS > 0) || (NDDB > 0) || (NMODULAR > 0)) 300 ksyms_init(); 301 #endif 302 kprintf_init(); 303 304 percpu_init(); 305 306 /* Initialize lock caches. */ 307 mutex_obj_init(); 308 rw_obj_init(); 309 310 /* Passive serialization. */ 311 pserialize_init(); 312 313 /* Initialize the extent manager. */ 314 extent_init(); 315 316 /* Do machine-dependent initialization. */ 317 cpu_startup(); 318 319 /* Initialize the sysctl subsystem. */ 320 sysctl_init(); 321 322 /* Initialize callouts, part 1. */ 323 callout_startup(); 324 325 /* Initialize the kernel authorization subsystem. */ 326 kauth_init(); 327 328 secmodel_init(); 329 330 spec_init(); 331 332 /* 333 * Set BPF op vector. Can't do this in bpf attach, since 334 * network drivers attach before bpf. 335 */ 336 bpf_setops(); 337 338 /* Start module system. */ 339 module_init(); 340 341 /* 342 * Initialize the kernel authorization subsystem and start the 343 * default security model, if any. We need to do this early 344 * enough so that subsystems relying on any of the aforementioned 345 * can work properly. Since the security model may dictate the 346 * credential inheritance policy, it is needed at least before 347 * any process is created, specifically proc0. 348 */ 349 module_init_class(MODULE_CLASS_SECMODEL); 350 351 /* Initialize the buffer cache */ 352 bufinit(); 353 biohist_init(); 354 355 #ifdef KERNHIST 356 sysctl_kernhist_init(); 357 #endif 358 359 360 #if defined(SPLASHSCREEN) && defined(SPLASHSCREEN_IMAGE) 361 size_t splash_size = (&_binary_splash_image_end - 362 &_binary_splash_image_start) * sizeof(void *); 363 splash_setimage(&_binary_splash_image_start, splash_size); 364 #endif 365 366 /* Initialize sockets. */ 367 soinit(); 368 369 /* 370 * The following things must be done before autoconfiguration. 371 */ 372 rnd_init(); /* initialize entropy pool */ 373 374 cprng_init(); /* initialize cryptographic PRNG */ 375 376 /* Initialize process and pgrp structures. */ 377 procinit(); 378 lwpinit(); 379 380 /* Initialize signal-related data structures. */ 381 signal_init(); 382 383 /* Initialize resource management. */ 384 resource_init(); 385 386 /* Create process 0. */ 387 proc0_init(); 388 lwp0_init(); 389 390 /* Disable preemption during boot. */ 391 kpreempt_disable(); 392 393 /* Initialize the UID hash table. */ 394 uid_init(); 395 396 /* Charge root for one process. */ 397 (void)chgproccnt(0, 1); 398 399 /* Initialize timekeeping. */ 400 time_init(); 401 402 /* Initialize the run queues, turnstiles and sleep queues. */ 403 sched_rqinit(); 404 turnstile_init(); 405 sleeptab_init(&sleeptab); 406 407 sched_init(); 408 409 /* Initialize processor-sets */ 410 psets_init(); 411 412 /* Initialize cpufreq(9) */ 413 cpufreq_init(); 414 415 /* MI initialization of the boot cpu */ 416 error = mi_cpu_attach(curcpu()); 417 KASSERT(error == 0); 418 419 /* Initialize timekeeping, part 2. */ 420 time_init2(); 421 422 /* 423 * Initialize mbuf's. Do this now because we might attempt to 424 * allocate mbufs or mbuf clusters during autoconfiguration. 425 */ 426 mbinit(); 427 428 /* Initialize I/O statistics. */ 429 iostat_init(); 430 431 /* Initialize the log device. */ 432 loginit(); 433 434 /* Second part of module system initialization. */ 435 module_start_unload_thread(); 436 437 /* Initialize the file systems. */ 438 #ifdef NVNODE_IMPLICIT 439 /* 440 * If maximum number of vnodes in namei vnode cache is not explicitly 441 * defined in kernel config, adjust the number such as we use roughly 442 * 10% of memory for vnodes and associated data structures in the 443 * assumed worst case. Do not provide fewer than NVNODE vnodes. 444 */ 445 usevnodes = calc_cache_size(vmem_size(kmem_arena, VMEM_FREE|VMEM_ALLOC), 446 10, VNODE_KMEM_MAXPCT) / VNODE_COST; 447 if (usevnodes > desiredvnodes) 448 desiredvnodes = usevnodes; 449 #endif 450 vfsinit(); 451 lf_init(); 452 453 /* Initialize fstrans. */ 454 fstrans_init(); 455 456 /* Initialize the file descriptor system. */ 457 fd_sys_init(); 458 459 /* Initialize cwd structures */ 460 cwd_sys_init(); 461 462 /* Initialize kqueue. */ 463 kqueue_init(); 464 465 inittimecounter(); 466 ntp_init(); 467 468 /* Initialize tty subsystem. */ 469 tty_init(); 470 ttyldisc_init(); 471 472 /* Initialize the buffer cache, part 2. */ 473 bufinit2(); 474 475 /* Initialize the disk wedge subsystem. */ 476 dkwedge_init(); 477 478 /* Initialize the kernel strong PRNG. */ 479 kern_cprng = cprng_strong_create("kernel", IPL_VM, 480 CPRNG_INIT_ANY|CPRNG_REKEY_ANY); 481 482 /* Initialize interfaces. */ 483 ifinit1(); 484 485 spldebug_start(); 486 487 /* Initialize sockets thread(s) */ 488 soinit1(); 489 490 /* 491 * Initialize the bufq strategy sub-system and any built-in 492 * strategy modules - they may be needed by some devices during 493 * auto-configuration 494 */ 495 bufq_init(); 496 module_init_class(MODULE_CLASS_BUFQ); 497 498 /* Configure the system hardware. This will enable interrupts. */ 499 configure(); 500 501 /* Once all CPUs are detected, initialize the per-CPU cprng_fast. */ 502 cprng_fast_init(); 503 504 ssp_init(); 505 506 ubc_init(); /* must be after autoconfig */ 507 508 mm_init(); 509 510 configure2(); 511 512 ipi_sysinit(); 513 514 /* Now timer is working. Enable preemption. */ 515 kpreempt_enable(); 516 517 /* Get the threads going and into any sleeps before continuing. */ 518 yield(); 519 520 /* Enable deferred processing of RNG samples */ 521 rnd_init_softint(); 522 523 #ifdef RND_PRINTF 524 /* Enable periodic injection of console output into entropy pool */ 525 kprintf_init_callout(); 526 #endif 527 528 vmem_rehash_start(); /* must be before exec_init */ 529 530 /* Initialize exec structures */ 531 exec_init(1); /* seminit calls exithook_establish() */ 532 533 #if NVERIEXEC > 0 534 /* 535 * Initialise the Veriexec subsystem. 536 */ 537 veriexec_init(); 538 #endif /* NVERIEXEC > 0 */ 539 540 pax_init(); 541 542 #ifdef IPSEC 543 /* Attach network crypto subsystem */ 544 ipsec_attach(); 545 #endif 546 547 /* 548 * Initialize protocols. Block reception of incoming packets 549 * until everything is ready. 550 */ 551 s = splnet(); 552 ifinit(); 553 #if defined(INET) || defined(INET6) 554 lltableinit(); 555 #endif 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 machdep_init(); 579 580 procinit_sysctl(); 581 582 scdebug_init(); 583 584 /* 585 * Create process 1 (init(8)). We do this now, as Unix has 586 * historically had init be process 1, and changing this would 587 * probably upset a lot of people. 588 * 589 * Note that process 1 won't immediately exec init(8), but will 590 * wait for us to inform it that the root file system has been 591 * mounted. 592 */ 593 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 594 panic("fork init"); 595 596 /* 597 * Load any remaining builtin modules, and hand back temporary 598 * storage to the VM system. Then require force when loading any 599 * remaining un-init'ed built-in modules to avoid later surprises. 600 */ 601 module_init_class(MODULE_CLASS_ANY); 602 module_builtin_require_force(); 603 604 /* 605 * Finalize configuration now that all real devices have been 606 * found. This needs to be done before the root device is 607 * selected, since finalization may create the root device. 608 */ 609 config_finalize(); 610 611 sysctl_finalize(); 612 613 /* 614 * Now that autoconfiguration has completed, we can determine 615 * the root and dump devices. 616 */ 617 cpu_rootconf(); 618 cpu_dumpconf(); 619 620 /* Mount the root file system. */ 621 do { 622 domountroothook(root_device); 623 if ((error = vfs_mountroot())) { 624 printf("cannot mount root, error = %d\n", error); 625 boothowto |= RB_ASKNAME; 626 setroot(root_device, 627 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 628 } 629 } while (error != 0); 630 mountroothook_destroy(); 631 632 configure3(); 633 634 /* 635 * Initialise the time-of-day clock, passing the time recorded 636 * in the root filesystem (if any) for use by systems that 637 * don't have a non-volatile time-of-day device. 638 */ 639 inittodr(rootfstime); 640 641 /* 642 * Now can look at time, having had a chance to verify the time 643 * from the file system. Reset l->l_rtime as it may have been 644 * munched in mi_switch() after the time got set. 645 */ 646 getnanotime(&time); 647 boottime = time; 648 649 mutex_enter(proc_lock); 650 LIST_FOREACH(p, &allproc, p_list) { 651 KASSERT((p->p_flag & PK_MARKER) == 0); 652 mutex_enter(p->p_lock); 653 TIMESPEC_TO_TIMEVAL(&p->p_stats->p_start, &time); 654 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 655 lwp_lock(l); 656 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 657 lwp_unlock(l); 658 } 659 mutex_exit(p->p_lock); 660 } 661 mutex_exit(proc_lock); 662 binuptime(&curlwp->l_stime); 663 664 for (CPU_INFO_FOREACH(cii, ci)) { 665 ci->ci_schedstate.spc_lastmod = time_second; 666 } 667 668 /* Create the pageout daemon kernel thread. */ 669 uvm_swap_init(); 670 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 671 NULL, NULL, "pgdaemon")) 672 panic("fork pagedaemon"); 673 674 /* Create the filesystem syncer kernel thread. */ 675 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 676 NULL, NULL, "ioflush")) 677 panic("fork syncer"); 678 679 /* Create the aiodone daemon kernel thread. */ 680 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 681 uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE)) 682 panic("fork aiodoned"); 683 684 /* Wait for final configure threads to complete. */ 685 config_finalize_mountroot(); 686 687 /* 688 * Okay, now we can let init(8) exec! It's off to userland! 689 */ 690 mutex_enter(proc_lock); 691 start_init_exec = 1; 692 cv_broadcast(&lbolt); 693 mutex_exit(proc_lock); 694 695 /* The scheduler is an infinite loop. */ 696 uvm_scheduler(); 697 /* NOTREACHED */ 698 } 699 700 /* 701 * Configure the system's hardware. 702 */ 703 static void 704 configure(void) 705 { 706 707 /* Initialize autoconf data structures. */ 708 config_init_mi(); 709 /* 710 * XXX 711 * callout_setfunc() requires mutex(9) so it can't be in config_init() 712 * on amiga and atari which use config_init() and autoconf(9) fucntions 713 * to initialize console devices. 714 */ 715 config_twiddle_init(); 716 717 pmf_init(); 718 719 /* Initialize driver modules */ 720 module_init_class(MODULE_CLASS_DRIVER); 721 722 userconf_init(); 723 if (boothowto & RB_USERCONF) 724 userconf_prompt(); 725 726 if ((boothowto & (AB_SILENT|AB_VERBOSE)) == AB_SILENT) { 727 printf_nolog("Detecting hardware..."); 728 } 729 730 /* 731 * Do the machine-dependent portion of autoconfiguration. This 732 * sets the configuration machinery here in motion by "finding" 733 * the root bus. When this function returns, we expect interrupts 734 * to be enabled. 735 */ 736 cpu_configure(); 737 } 738 739 static void 740 configure2(void) 741 { 742 CPU_INFO_ITERATOR cii; 743 struct cpu_info *ci; 744 int s; 745 746 /* 747 * Now that we've found all the hardware, start the real time 748 * and statistics clocks. 749 */ 750 initclocks(); 751 752 cold = 0; /* clocks are running, we're warm now! */ 753 s = splsched(); 754 curcpu()->ci_schedstate.spc_flags |= SPCF_RUNNING; 755 splx(s); 756 757 /* Boot the secondary processors. */ 758 for (CPU_INFO_FOREACH(cii, ci)) { 759 uvm_cpu_attach(ci); 760 } 761 mp_online = true; 762 #if defined(MULTIPROCESSOR) 763 cpu_boot_secondary_processors(); 764 #endif 765 766 /* Setup the runqueues and scheduler. */ 767 runq_init(); 768 synch_init(); 769 770 /* 771 * Bus scans can make it appear as if the system has paused, so 772 * twiddle constantly while config_interrupts() jobs are running. 773 */ 774 config_twiddle_fn(NULL); 775 776 /* 777 * Create threads to call back and finish configuration for 778 * devices that want interrupts enabled. 779 */ 780 config_create_interruptthreads(); 781 } 782 783 static void 784 configure3(void) 785 { 786 787 /* 788 * Create threads to call back and finish configuration for 789 * devices that want the mounted root file system. 790 */ 791 config_create_mountrootthreads(); 792 793 /* Get the threads going and into any sleeps before continuing. */ 794 yield(); 795 } 796 797 static void 798 rootconf_handle_wedges(void) 799 { 800 struct disklabel label; 801 struct partition *p; 802 struct vnode *vp; 803 daddr_t startblk; 804 uint64_t nblks; 805 device_t dev; 806 int error; 807 808 if (booted_nblks) { 809 /* 810 * bootloader passed geometry 811 */ 812 dev = booted_device; 813 startblk = booted_startblk; 814 nblks = booted_nblks; 815 816 /* 817 * keep booted_device and booted_partition 818 * in case the kernel doesn't identify a wedge 819 */ 820 } else { 821 /* 822 * bootloader passed partition number 823 * 824 * We cannot ask the partition device directly when it is 825 * covered by a wedge. Instead we look up the geometry in 826 * the disklabel. 827 */ 828 vp = opendisk(booted_device); 829 830 if (vp == NULL) 831 return; 832 833 error = VOP_IOCTL(vp, DIOCGDINFO, &label, FREAD, NOCRED); 834 VOP_CLOSE(vp, FREAD, NOCRED); 835 vput(vp); 836 if (error) 837 return; 838 839 KASSERT(booted_partition >= 0 840 && booted_partition < MAXPARTITIONS); 841 842 p = &label.d_partitions[booted_partition]; 843 844 dev = booted_device; 845 startblk = p->p_offset; 846 nblks = p->p_size; 847 } 848 849 dev = dkwedge_find_partition(dev, startblk, nblks); 850 if (dev != NULL) { 851 booted_device = dev; 852 booted_partition = 0; 853 } 854 } 855 856 void 857 rootconf(void) 858 { 859 if (booted_device != NULL) 860 rootconf_handle_wedges(); 861 862 setroot(booted_device, booted_partition); 863 } 864 865 static void 866 check_console(struct lwp *l) 867 { 868 struct vnode *vp; 869 int error; 870 871 error = namei_simple_kernel("/dev/console", 872 NSM_FOLLOW_NOEMULROOT, &vp); 873 if (error == 0) 874 vrele(vp); 875 else if (error == ENOENT) 876 printf("warning: no /dev/console\n"); 877 else 878 printf("warning: lookup /dev/console: error %d\n", error); 879 } 880 881 /* 882 * List of paths to try when searching for "init". 883 */ 884 static const char * const initpaths[] = { 885 "/sbin/init", 886 "/sbin/oinit", 887 "/sbin/init.bak", 888 "/rescue/init", 889 NULL, 890 }; 891 892 /* 893 * Start the initial user process; try exec'ing each pathname in "initpaths". 894 * The program is invoked with one argument containing the boot flags. 895 */ 896 static void 897 start_init(void *arg) 898 { 899 struct lwp *l = arg; 900 struct proc *p = l->l_proc; 901 vaddr_t addr; 902 struct sys_execve_args /* { 903 syscallarg(const char *) path; 904 syscallarg(char * const *) argp; 905 syscallarg(char * const *) envp; 906 } */ args; 907 int options, i, error; 908 register_t retval[2]; 909 char flags[4], *flagsp; 910 const char *path, *slash; 911 char *ucp, **uap, *arg0, *arg1, *argv[3]; 912 char ipath[129]; 913 int ipx, len; 914 915 /* 916 * Now in process 1. 917 */ 918 strncpy(p->p_comm, "init", MAXCOMLEN); 919 920 /* 921 * Wait for main() to tell us that it's safe to exec. 922 */ 923 mutex_enter(proc_lock); 924 while (start_init_exec == 0) 925 cv_wait(&lbolt, proc_lock); 926 mutex_exit(proc_lock); 927 928 /* 929 * This is not the right way to do this. We really should 930 * hand-craft a descriptor onto /dev/console to hand to init, 931 * but that's a _lot_ more work, and the benefit from this easy 932 * hack makes up for the "good is the enemy of the best" effect. 933 */ 934 check_console(l); 935 936 /* 937 * Need just enough stack to hold the faked-up "execve()" arguments. 938 */ 939 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 940 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 941 NULL, UVM_UNKNOWN_OFFSET, 0, 942 UVM_MAPFLAG(UVM_PROT_RW, UVM_PROT_RW, UVM_INH_COPY, 943 UVM_ADV_NORMAL, 944 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 945 panic("init: couldn't allocate argument space"); 946 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 947 948 ipx = 0; 949 while (1) { 950 if (boothowto & RB_ASKNAME) { 951 printf("init path"); 952 if (initpaths[ipx]) 953 printf(" (default %s)", initpaths[ipx]); 954 printf(": "); 955 len = cngetsn(ipath, sizeof(ipath)-1); 956 if (len == 4 && strcmp(ipath, "halt") == 0) { 957 cpu_reboot(RB_HALT, NULL); 958 } else if (len == 6 && strcmp(ipath, "reboot") == 0) { 959 cpu_reboot(0, NULL); 960 #if defined(DDB) 961 } else if (len == 3 && strcmp(ipath, "ddb") == 0) { 962 console_debugger(); 963 continue; 964 #endif 965 } else if (len > 0 && ipath[0] == '/') { 966 ipath[len] = '\0'; 967 path = ipath; 968 } else if (len == 0 && initpaths[ipx] != NULL) { 969 path = initpaths[ipx++]; 970 } else { 971 printf("use absolute path, "); 972 #if defined(DDB) 973 printf("\"ddb\", "); 974 #endif 975 printf("\"halt\", or \"reboot\"\n"); 976 continue; 977 } 978 } else { 979 if ((path = initpaths[ipx++]) == NULL) { 980 ipx = 0; 981 boothowto |= RB_ASKNAME; 982 continue; 983 } 984 } 985 986 ucp = (char *)USRSTACK; 987 988 /* 989 * Construct the boot flag argument. 990 */ 991 flagsp = flags; 992 *flagsp++ = '-'; 993 options = 0; 994 995 if (boothowto & RB_SINGLE) { 996 *flagsp++ = 's'; 997 options = 1; 998 } 999 #ifdef notyet 1000 if (boothowto & RB_FASTBOOT) { 1001 *flagsp++ = 'f'; 1002 options = 1; 1003 } 1004 #endif 1005 1006 /* 1007 * Move out the flags (arg 1), if necessary. 1008 */ 1009 if (options != 0) { 1010 *flagsp++ = '\0'; 1011 i = flagsp - flags; 1012 #ifdef DEBUG 1013 aprint_normal("init: copying out flags `%s' %d\n", flags, i); 1014 #endif 1015 arg1 = STACK_ALLOC(ucp, i); 1016 ucp = STACK_MAX(arg1, i); 1017 if ((error = copyout((void *)flags, arg1, i)) != 0) 1018 goto copyerr; 1019 } else 1020 arg1 = NULL; 1021 1022 /* 1023 * Move out the file name (also arg 0). 1024 */ 1025 i = strlen(path) + 1; 1026 #ifdef DEBUG 1027 aprint_normal("init: copying out path `%s' %d\n", path, i); 1028 #else 1029 if (boothowto & RB_ASKNAME || path != initpaths[0]) 1030 printf("init: trying %s\n", path); 1031 #endif 1032 arg0 = STACK_ALLOC(ucp, i); 1033 ucp = STACK_MAX(arg0, i); 1034 if ((error = copyout(path, arg0, i)) != 0) 1035 goto copyerr; 1036 1037 /* 1038 * Move out the arg pointers. 1039 */ 1040 ucp = (void *)STACK_ALIGN(ucp, STACK_ALIGNBYTES); 1041 uap = (char **)STACK_ALLOC(ucp, sizeof(argv)); 1042 SCARG(&args, path) = arg0; 1043 SCARG(&args, argp) = uap; 1044 SCARG(&args, envp) = NULL; 1045 slash = strrchr(path, '/'); 1046 1047 argv[0] = slash ? arg0 + (slash + 1 - path) : arg0; 1048 argv[1] = arg1; 1049 argv[2] = NULL; 1050 if ((error = copyout(argv, uap, sizeof(argv))) != 0) 1051 goto copyerr; 1052 1053 /* 1054 * Now try to exec the program. If it can't for any reason 1055 * other than it doesn't exist, complain. 1056 */ 1057 error = sys_execve(l, &args, retval); 1058 if (error == 0 || error == EJUSTRETURN) { 1059 KERNEL_UNLOCK_LAST(l); 1060 return; 1061 } 1062 printf("exec %s: error %d\n", path, error); 1063 } 1064 printf("init: not found\n"); 1065 panic("no init"); 1066 copyerr: 1067 panic("copyout %d", error); 1068 } 1069 1070 /* 1071 * calculate cache size (in bytes) from physmem and vsize. 1072 */ 1073 vaddr_t 1074 calc_cache_size(vsize_t vsize, int pct, int va_pct) 1075 { 1076 paddr_t t; 1077 1078 /* XXX should consider competing cache if any */ 1079 /* XXX should consider submaps */ 1080 t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; 1081 if (vsize != 0) { 1082 vsize = (uintmax_t)vsize * va_pct / 100; 1083 if (t > vsize) { 1084 t = vsize; 1085 } 1086 } 1087 return t; 1088 } 1089 1090 /* 1091 * Print the system start up banner. 1092 * 1093 * - Print a limited banner if AB_SILENT. 1094 * - Always send normal banner to the log. 1095 */ 1096 #define MEM_PBUFSIZE sizeof("99999 MB") 1097 1098 void 1099 banner(void) 1100 { 1101 static char notice[] = " Notice: this software is " 1102 "protected by copyright"; 1103 char pbuf[81]; 1104 void (*pr)(const char *, ...) __printflike(1, 2); 1105 int i; 1106 1107 if ((boothowto & AB_SILENT) != 0) { 1108 snprintf(pbuf, sizeof(pbuf), "%s %s (%s)", 1109 ostype, osrelease, kernel_ident); 1110 printf_nolog("%s", pbuf); 1111 for (i = 80 - strlen(pbuf) - sizeof(notice); i > 0; i--) 1112 printf(" "); 1113 printf_nolog("%s\n", notice); 1114 pr = aprint_normal; 1115 } else { 1116 pr = printf; 1117 } 1118 1119 memset(pbuf, 0, sizeof(pbuf)); 1120 (*pr)("%s%s", copyright, version); 1121 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)physmem)); 1122 (*pr)("total memory = %s\n", pbuf); 1123 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)uvmexp.free)); 1124 (*pr)("avail memory = %s\n", pbuf); 1125 } 1126