1/* $NetBSD: init_main.c,v 1.549 2024/03/05 20:59:41 thorpej Exp $ */ 2 3/*- 4 * Copyright (c) 2008, 2009, 2019, 2023 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 1.549 2024/03/05 20:59:41 thorpej Exp $"); 101 102#include "opt_cnmagic.h" 103#include "opt_ddb.h" 104#include "opt_inet.h" 105#include "opt_ipsec.h" 106#include "opt_modular.h" 107#include "opt_ntp.h" 108#include "opt_pipe.h" 109#include "opt_syscall_debug.h" 110#include "opt_sysv.h" 111#include "opt_fileassoc.h" 112#include "opt_ktrace.h" 113#include "opt_pax.h" 114#include "opt_compat_netbsd.h" 115#include "opt_ptrace.h" 116#include "opt_splash.h" 117#include "opt_kernhist.h" 118#include "opt_gprof.h" 119 120#if defined(SPLASHSCREEN) && defined(makeoptions_SPLASHSCREEN_IMAGE) 121extern void *_binary_splash_image_start; 122extern 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/lwp.h> 142#include <sys/kthread.h> 143#include <sys/resourcevar.h> 144#include <sys/signalvar.h> 145#include <sys/systm.h> 146#include <sys/vnode.h> 147#include <sys/fstrans.h> 148#include <sys/tty.h> 149#include <sys/conf.h> 150#include <sys/disklabel.h> 151#include <sys/buf.h> 152#include <sys/device.h> 153#include <sys/exec.h> 154#include <sys/socketvar.h> 155#include <sys/protosw.h> 156#include <sys/percpu.h> 157#include <sys/pserialize.h> 158#include <sys/pset.h> 159#include <sys/sysctl.h> 160#include <sys/reboot.h> 161#include <sys/event.h> 162#include <sys/mbuf.h> 163#include <sys/sched.h> 164#include <sys/sleepq.h> 165#include <sys/ipi.h> 166#include <sys/iostat.h> 167#include <sys/vmem.h> 168#include <sys/uuid.h> 169#include <sys/extent.h> 170#include <sys/disk.h> 171#include <sys/msgbuf.h> 172#include <sys/module.h> 173#include <sys/module_hook.h> 174#include <sys/event.h> 175#include <sys/lockf.h> 176#include <sys/once.h> 177#include <sys/kcpuset.h> 178#include <sys/ksyms.h> 179#include <sys/uidinfo.h> 180#include <sys/kprintf.h> 181#include <sys/bufq.h> 182#include <sys/threadpool.h> 183#include <sys/futex.h> 184#ifdef IPSEC 185#include <netipsec/ipsec.h> 186#endif 187#include <sys/domain.h> 188#include <sys/namei.h> 189#include <sys/rnd.h> 190#include <sys/pipe.h> 191#if NVERIEXEC > 0 192#include <sys/verified_exec.h> 193#endif /* NVERIEXEC > 0 */ 194#ifdef KTRACE 195#include <sys/ktrace.h> 196#endif 197#include <sys/kauth.h> 198#include <net80211/ieee80211_netbsd.h> 199#include <sys/cprng.h> 200#include <sys/psref.h> 201#include <sys/radixtree.h> 202#include <sys/heartbeat.h> 203 204#include <sys/syscall.h> 205#include <sys/syscallargs.h> 206 207#include <sys/pax.h> 208 209#include <dev/clock_subr.h> 210 211#include <secmodel/secmodel.h> 212 213#include <ufs/ufs/quota.h> 214 215#include <miscfs/genfs/genfs.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#include <dev/cons.h> 223#include <dev/splash/splash.h> 224 225#include <net/bpf.h> 226#include <net/if.h> 227#include <net/pfil.h> 228#include <net/raw_cb.h> 229#include <net/if_llatbl.h> 230 231#include <prop/proplib.h> 232 233#include <sys/userconf.h> 234 235extern time_t rootfstime; 236 237#ifndef curlwp 238struct lwp *curlwp = &lwp0; 239#endif 240struct proc *initproc; 241 242struct vnode *rootvp, *swapdev_vp; 243int boothowto; 244int cold __read_mostly = 1; /* still working on startup */ 245int shutting_down __read_mostly; /* system is shutting down */ 246 247int start_init_exec; /* semaphore for start_init() */ 248 249static void check_console(struct lwp *l); 250static void start_init(void *); 251static void configure(void); 252static void configure2(void); 253static void configure3(void); 254void main(void); 255 256/* 257 * System startup; initialize the world, create process 0, mount root 258 * filesystem, and fork to create init and pagedaemon. Most of the 259 * hard work is done in the lower-level initialization routines including 260 * startup(), which does memory initialization and autoconfiguration. 261 */ 262void 263main(void) 264{ 265 struct timespec time; 266 struct lwp *l; 267 struct proc *p; 268 int s, error; 269#ifdef NVNODE_IMPLICIT 270 int usevnodes; 271#endif 272 CPU_INFO_ITERATOR cii; 273 struct cpu_info *ci; 274 275#ifdef DIAGNOSTIC 276 /* 277 * Verify that CPU_INFO_FOREACH() knows about the boot CPU 278 * and only the boot CPU at this point. 279 */ 280 int cpucount = 0; 281 for (CPU_INFO_FOREACH(cii, ci)) { 282 KASSERT(ci == curcpu()); 283 cpucount++; 284 } 285 KASSERT(cpucount == 1); 286#endif 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#ifdef CNMAGIC 300 cn_set_magic(CNMAGIC); 301#endif 302 303 kernel_lock_init(); 304 once_init(); 305 todr_init(); 306 307 mi_cpu_init(); 308 kernconfig_lock_init(); 309 kthread_sysinit(); 310 311 /* Initialize the device switch tables. */ 312 devsw_init(); 313 314 /* Initialize event counters. */ 315 evcnt_init(); 316 317 uvm_init(); 318 ubchist_init(); 319 kcpuset_sysinit(); 320 321 prop_kern_init(); 322 323#if ((NKSYMS > 0) || (NDDB > 0) || (NMODULAR > 0)) 324 ksyms_init(); 325#endif 326 kprintf_init(); 327 328 percpu_init(); 329 330 /* Initialize radix trees (used by numerous subsystems). */ 331 radix_tree_init(); 332 333 /* Passive serialization. */ 334 pserialize_init(); 335 336 /* Initialize the extent manager. */ 337 extent_init(); 338 339 /* Do machine-dependent initialization. */ 340 cpu_startup(); 341 342 /* Initialize the sysctl subsystem. */ 343 sysctl_init(); 344 345 /* Initialize callouts, part 1. */ 346 callout_startup(); 347 348 /* Initialize the kernel authorization subsystem. */ 349 kauth_init(); 350 351 secmodel_init(); 352 353 spec_init(); 354 355 /* 356 * Set BPF op vector. Can't do this in bpf attach, since 357 * network drivers attach before bpf. 358 */ 359 bpf_setops(); 360 361 /* Initialize what we can in ipi(9) before CPUs are detected. */ 362 ipi_sysinit(); 363 364 /* Start module system. */ 365 module_init(); 366 module_hook_init(); 367 368 /* 369 * Initialize the kernel authorization subsystem and start the 370 * default security model, if any. We need to do this early 371 * enough so that subsystems relying on any of the aforementioned 372 * can work properly. Since the security model may dictate the 373 * credential inheritance policy, it is needed at least before 374 * any process is created, specifically proc0. 375 */ 376 module_init_class(MODULE_CLASS_SECMODEL); 377 378 /* Initialize the buffer cache */ 379 bufinit(); 380 biohist_init(); 381 382#ifdef KERNHIST 383 sysctl_kernhist_init(); 384#endif 385 386 387#if defined(SPLASHSCREEN) && defined(SPLASHSCREEN_IMAGE) 388 size_t splash_size = (&_binary_splash_image_end - 389 &_binary_splash_image_start) * sizeof(void *); 390 splash_setimage(&_binary_splash_image_start, splash_size); 391#endif 392 393 /* Initialize sockets. */ 394 soinit(); 395 396 /* 397 * The following things must be done before autoconfiguration. 398 */ 399 rnd_init(); /* initialize entropy pool */ 400 401 cprng_init(); /* initialize cryptographic PRNG */ 402 403 /* Initialize process and pgrp structures. */ 404 procinit(); 405 lwpinit(); 406 407 /* Must be called after lwpinit (lwpinit_specificdata) */ 408 psref_init(); 409 410 /* Initialize exec structures */ 411 exec_init(1); /* signal_init calls exechook_establish() */ 412 413 /* Initialize signal-related data structures. */ 414 signal_init(); 415 416 /* Initialize resource management. */ 417 resource_init(); 418 419 /* Create process 0. */ 420 proc0_init(); 421 lwp0_init(); 422 423 /* Disable preemption during boot. */ 424 kpreempt_disable(); 425 426 /* Initialize the threadpool system. */ 427 threadpools_init(); 428 429 /* Initialize the UID hash table. */ 430 uid_init(); 431 432 /* Charge root for one process. */ 433 (void)chgproccnt(0, 1); 434 435 /* Initialize the run queues, turnstiles and sleep queues. */ 436 sched_rqinit(); 437 turnstile_init(); 438 sleeptab_init(&sleeptab); 439 440 sched_init(); 441 442 /* Initialize processor-sets */ 443 psets_init(); 444 445 /* Initialize cpufreq(9) */ 446 cpufreq_init(); 447 448 /* MI initialization of the boot cpu */ 449 error = mi_cpu_attach(curcpu()); 450 KASSERT(error == 0); 451 452 /* Initialize timekeeping. */ 453 time_init(); 454 455 /* 456 * Initialize mbuf's. Do this now because we might attempt to 457 * allocate mbufs or mbuf clusters during autoconfiguration. 458 */ 459 mbinit(); 460 461 /* Initialize I/O statistics. */ 462 iostat_init(); 463 464 /* Initialize the log device. */ 465 loginit(); 466 467 /* Second part of module system initialization. */ 468 module_start_unload_thread(); 469 470 /* Initialize autoconf data structures before any modules are loaded */ 471 config_init_mi(); 472 473 /* Initialize the file systems. */ 474#ifdef NVNODE_IMPLICIT 475 /* 476 * If maximum number of vnodes in namei vnode cache is not explicitly 477 * defined in kernel config, adjust the number such as we use roughly 478 * 10% of memory for vnodes and associated data structures in the 479 * assumed worst case. Do not provide fewer than NVNODE vnodes. 480 */ 481 usevnodes = calc_cache_size(vmem_size(kmem_arena, VMEM_FREE|VMEM_ALLOC), 482 10, VNODE_KMEM_MAXPCT) / VNODE_COST; 483 if (usevnodes > desiredvnodes) 484 desiredvnodes = usevnodes; 485#endif /* NVNODE_IMPLICIT */ 486#ifdef MAXFILES_IMPLICIT 487 /* 488 * If maximum number of files is not explicitly defined in 489 * kernel config, adjust the number so that it is somewhat 490 * more reasonable on machines with larger memory sizes. 491 * Arbitrary numbers are 20,000 files for 16GB RAM or more 492 * and 10,000 files for 1GB RAM or more. 493 * 494 * XXXtodo: adjust this and other values totally dynamically 495 */ 496 if (ctob((uint64_t)physmem) >= 16ULL * 1024 * 1024 * 1024) 497 maxfiles = MAX(maxfiles, 20000); 498 if (ctob((uint64_t)physmem) >= 1024 * 1024 * 1024) 499 maxfiles = MAX(maxfiles, 10000); 500#endif /* MAXFILES_IMPLICIT */ 501 502 /* Initialize fstrans. */ 503 fstrans_init(); 504 505 vfsinit(); 506 lf_init(); 507 508 /* Initialize the file descriptor system. */ 509 fd_sys_init(); 510 511 /* Initialize kqueue. */ 512 kqueue_init(); 513 514 inittimecounter(); 515 ntp_init(); 516 517 /* Initialize tty subsystem. */ 518 tty_init(); 519 ttyldisc_init(); 520 521 /* Initialize the buffer cache, part 2. */ 522 bufinit2(); 523 524 /* Initialize the disk wedge subsystem. */ 525 dkwedge_init(); 526 527 /* Initialize pfil */ 528 pfil_init(); 529 530 /* Initialize interfaces. */ 531 ifinit1(); 532 533 spldebug_start(); 534 535 /* Initialize sockets thread(s) */ 536 soinit1(); 537 538 /* 539 * Initialize the bufq strategy sub-system and any built-in 540 * strategy modules - they may be needed by some devices during 541 * auto-configuration 542 */ 543 bufq_init(); 544 module_init_class(MODULE_CLASS_BUFQ); 545 546 /* Configure the system hardware. This will enable interrupts. */ 547 configure(); 548#ifdef __HAVE_LEGACY_INTRCNT 549 evcnt_attach_legacy_intrcnt(); 550#endif 551 552 /* Enable deferred processing of RNG samples */ 553 rnd_init_softint(); 554 555 /* Once all CPUs are detected, initialize the per-CPU cprng_fast. */ 556 cprng_fast_init(); 557 558 /* 559 * Now that softints can be established, start monitoring 560 * system heartbeat on all CPUs. 561 */ 562 heartbeat_start(); 563 564 ssp_init(); 565 566 ubc_init(); /* must be after autoconfig */ 567 568 mm_init(); 569 570 configure2(); 571 572 /* Initialize the rest of ipi(9) after CPUs have been detected. */ 573 ipi_percpu_init(); 574 575 futex_sys_init(); 576 577 /* Now timer is working. Enable preemption. */ 578 kpreempt_enable(); 579 580 /* Get the threads going and into any sleeps before continuing. */ 581 yield(); 582 583 vmem_rehash_start(); /* must be before exec_init */ 584 585#if NVERIEXEC > 0 586 /* 587 * Initialise the Veriexec subsystem. 588 */ 589 veriexec_init(); 590#endif /* NVERIEXEC > 0 */ 591 592 pax_init(); 593 594#ifdef IPSEC 595 /* Attach network crypto subsystem */ 596 ipsec_attach(); 597#endif 598 599 /* 600 * Initialize protocols. Block reception of incoming packets 601 * until everything is ready. 602 */ 603 s = splnet(); 604 ifinit(); 605#if defined(INET) || defined(INET6) 606 lltableinit(); 607#endif 608 domaininit(true); 609 ifinit_post(); 610 if_attachdomain(); 611 splx(s); 612 613#ifdef GPROF 614 /* Initialize kernel profiling. */ 615 kmstartup(); 616#endif 617 618 /* Initialize system accounting. */ 619 acct_init(); 620 621#ifndef PIPE_SOCKETPAIR 622 /* Initialize pipes. */ 623 pipe_init(); 624#endif 625 626#ifdef KTRACE 627 /* Initialize ktrace. */ 628 ktrinit(); 629#endif 630 631 machdep_init(); 632 633 procinit_sysctl(); 634 635 scdebug_init(); 636 637 /* 638 * Create process 1 (init(8)). We do this now, as Unix has 639 * historically had init be process 1, and changing this would 640 * probably upset a lot of people. 641 * 642 * Note that process 1 won't immediately exec init(8), but will 643 * wait for us to inform it that the root file system has been 644 * mounted. 645 */ 646 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL)) 647 panic("fork init"); 648 649 /* 650 * The initproc variable cannot be initialized in start_init as there 651 * is a race between vfs_mountroot and start_init. 652 */ 653 mutex_enter(&proc_lock); 654 initproc = proc_find_raw(1); 655 mutex_exit(&proc_lock); 656 657 /* 658 * Load any remaining builtin modules, and hand back temporary 659 * storage to the VM system. Then require force when loading any 660 * remaining un-init'ed built-in modules to avoid later surprises. 661 */ 662 module_init_class(MODULE_CLASS_ANY); 663 module_builtin_require_force(); 664 665 /* 666 * Finalize configuration now that all real devices have been 667 * found. This needs to be done before the root device is 668 * selected, since finalization may create the root device. 669 */ 670 config_finalize(); 671 672 sysctl_finalize(); 673 674 /* 675 * Now that autoconfiguration has completed, we can determine 676 * the root and dump devices. 677 */ 678 cpu_rootconf(); 679 cpu_dumpconf(); 680 681 /* Mount the root file system. */ 682 do { 683 domountroothook(root_device); 684 if ((error = vfs_mountroot())) { 685 printf("cannot mount root, error = %d\n", error); 686 boothowto |= RB_ASKNAME; 687 setroot(root_device, 688 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 689 } 690 } while (error != 0); 691 mountroothook_destroy(); 692 693 configure3(); 694 695 /* 696 * Initialise the time-of-day clock, passing the time recorded 697 * in the root filesystem (if any) for use by systems that 698 * don't have a non-volatile time-of-day device. 699 */ 700 inittodr(rootfstime); 701 702 /* 703 * Now can look at time, having had a chance to verify the time 704 * from the file system. Reset l->l_rtime as it may have been 705 * munched in mi_switch() after the time got set. 706 */ 707 getnanotime(&time); 708 709 mutex_enter(&proc_lock); 710 LIST_FOREACH(p, &allproc, p_list) { 711 KASSERT((p->p_flag & PK_MARKER) == 0); 712 mutex_enter(p->p_lock); 713 TIMESPEC_TO_TIMEVAL(&p->p_stats->p_start, &time); 714 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 715 lwp_lock(l); 716 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 717 lwp_unlock(l); 718 } 719 mutex_exit(p->p_lock); 720 } 721 mutex_exit(&proc_lock); 722 binuptime(&curlwp->l_stime); 723 724 for (CPU_INFO_FOREACH(cii, ci)) { 725 ci->ci_schedstate.spc_lastmod = time_second; 726 } 727 728 /* Create the pageout daemon kernel thread. */ 729 uvm_swap_init(); 730 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 731 NULL, NULL, "pgdaemon")) 732 panic("fork pagedaemon"); 733 734 /* Create the filesystem syncer kernel thread. */ 735 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 736 NULL, NULL, "ioflush")) 737 panic("fork syncer"); 738 739 /* Wait for final configure threads to complete. */ 740 config_finalize_mountroot(); 741 742 /* 743 * Okay, now we can let init(8) exec! It's off to userland! 744 */ 745 mutex_enter(&proc_lock); 746 start_init_exec = 1; 747 cv_broadcast(&lbolt); 748 mutex_exit(&proc_lock); 749 750 /* The scheduler is an infinite loop. */ 751 uvm_scheduler(); 752 /* NOTREACHED */ 753} 754 755/* 756 * Configure the system's hardware. 757 */ 758static void 759configure(void) 760{ 761 762 /* 763 * XXX 764 * callout_setfunc() requires mutex(9) so it can't be in config_init() 765 * on amiga and atari which use config_init() and autoconf(9) functions 766 * to initialize console devices. 767 */ 768 config_twiddle_init(); 769 770 pmf_init(); 771 772 /* Initialize driver modules */ 773 module_init_class(MODULE_CLASS_DRIVER); 774 775 userconf_init(); 776 if (boothowto & RB_USERCONF) 777 userconf_prompt(); 778 779 if ((boothowto & (AB_SILENT|AB_VERBOSE)) == AB_SILENT) { 780 printf_nolog("Detecting hardware..."); 781 } 782 783 /* 784 * Do the machine-dependent portion of autoconfiguration. This 785 * sets the configuration machinery here in motion by "finding" 786 * the root bus. When this function returns, we expect interrupts 787 * to be enabled. 788 */ 789 cpu_configure(); 790} 791 792static void 793configure2(void) 794{ 795 CPU_INFO_ITERATOR cii; 796 struct cpu_info *ci; 797 int s; 798 799 /* Fix up CPU topology info, which has all been collected by now. */ 800 cpu_topology_init(); 801 802 /* 803 * Now that we've found all the hardware, start the real time 804 * and statistics clocks. 805 */ 806 initclocks(); 807 808 cold = 0; /* clocks are running, we're warm now! */ 809 s = splsched(); 810 curcpu()->ci_schedstate.spc_flags |= SPCF_RUNNING; 811 splx(s); 812 813 /* Setup the runqueues and scheduler. */ 814 runq_init(); 815 synch_init(); 816 817 /* Boot the secondary processors. */ 818 for (CPU_INFO_FOREACH(cii, ci)) { 819 uvm_cpu_attach(ci); 820 } 821 822 /* Decide how to partition free memory. */ 823 uvm_page_rebucket(); 824 825 mp_online = true; 826#if defined(MULTIPROCESSOR) 827 cpu_boot_secondary_processors(); 828#endif 829 830 /* 831 * Bus scans can make it appear as if the system has paused, so 832 * twiddle constantly while config_interrupts() jobs are running. 833 */ 834 config_twiddle_fn(NULL); 835 836 /* 837 * Create threads to call back and finish configuration for 838 * devices that want interrupts enabled. 839 */ 840 config_create_interruptthreads(); 841} 842 843static void 844configure3(void) 845{ 846 847 /* 848 * Create threads to call back and finish configuration for 849 * devices that want the mounted root file system. 850 */ 851 config_create_mountrootthreads(); 852 853 /* Get the threads going and into any sleeps before continuing. */ 854 yield(); 855} 856 857static void 858rootconf_handle_wedges(void) 859{ 860 struct disklabel label; 861 struct partition *p; 862 struct vnode *vp; 863 daddr_t startblk; 864 uint64_t nblks; 865 device_t dev; 866 int error; 867 868 if (booted_nblks) { 869 /* 870 * bootloader passed geometry 871 */ 872 dev = booted_device; 873 startblk = booted_startblk; 874 nblks = booted_nblks; 875 876 /* 877 * keep booted_device and booted_partition 878 * in case the kernel doesn't identify a wedge 879 */ 880 } else { 881 /* 882 * bootloader passed partition number 883 * 884 * We cannot ask the partition device directly when it is 885 * covered by a wedge. Instead we look up the geometry in 886 * the disklabel. 887 */ 888 vp = opendisk(booted_device); 889 890 if (vp == NULL) 891 return; 892 893 VOP_UNLOCK(vp); 894 error = VOP_IOCTL(vp, DIOCGDINFO, &label, FREAD, NOCRED); 895 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 896 VOP_CLOSE(vp, FREAD, NOCRED); 897 vput(vp); 898 if (error) 899 return; 900 901 KASSERT(booted_partition >= 0 902 && booted_partition < MAXPARTITIONS); 903 904 p = &label.d_partitions[booted_partition]; 905 906 dev = booted_device; 907 startblk = p->p_offset; 908 nblks = p->p_size; 909 } 910 911 dev = dkwedge_find_partition(dev, startblk, nblks); 912 if (dev != NULL) { 913 booted_device = dev; 914 booted_partition = 0; 915 } 916} 917 918void 919rootconf(void) 920{ 921 if (booted_device != NULL) 922 rootconf_handle_wedges(); 923 924 setroot(booted_device, booted_partition); 925} 926 927static void 928check_console(struct lwp *l) 929{ 930 struct vnode *vp; 931 int error; 932 933 error = namei_simple_kernel("/dev/console", 934 NSM_FOLLOW_NOEMULROOT, &vp); 935 if (error == 0) { 936 vrele(vp); 937 } else if (error == ENOENT) { 938 if (boothowto & (AB_VERBOSE|AB_DEBUG)) 939 printf("warning: no /dev/console\n"); 940 } else { 941 printf("warning: lookup /dev/console: error %d\n", error); 942 } 943} 944 945/* 946 * List of paths to try when searching for "init". 947 */ 948static const char * const initpaths[] = { 949 "/sbin/init", 950 "/sbin/oinit", 951 "/sbin/init.bak", 952 "/rescue/init", 953 NULL, 954}; 955 956/* 957 * Start the initial user process; try exec'ing each pathname in "initpaths". 958 * The program is invoked with one argument containing the boot flags. 959 */ 960static void 961start_init(void *arg) 962{ 963 struct lwp *l = arg; 964 struct proc *p = l->l_proc; 965 vaddr_t addr; 966 struct sys_execve_args /* { 967 syscallarg(const char *) path; 968 syscallarg(char * const *) argp; 969 syscallarg(char * const *) envp; 970 } */ args; 971 int options, i, error; 972 register_t retval[2]; 973 char flags[4], *flagsp; 974 const char *path, *slash; 975 char *ucp, **uap, *arg0, *arg1, *argv[3]; 976 char ipath[129]; 977 int ipx, len; 978 979 /* 980 * Now in process 1. 981 */ 982 strncpy(p->p_comm, "init", MAXCOMLEN); 983 984 /* 985 * Wait for main() to tell us that it's safe to exec. 986 */ 987 mutex_enter(&proc_lock); 988 while (start_init_exec == 0) 989 cv_wait(&lbolt, &proc_lock); 990 mutex_exit(&proc_lock); 991 992 /* 993 * This is not the right way to do this. We really should 994 * hand-craft a descriptor onto /dev/console to hand to init, 995 * but that's a _lot_ more work, and the benefit from this easy 996 * hack makes up for the "good is the enemy of the best" effect. 997 */ 998 check_console(l); 999 1000 /* 1001 * Need just enough stack to hold the faked-up "execve()" arguments. 1002 */ 1003 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 1004 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 1005 NULL, UVM_UNKNOWN_OFFSET, 0, 1006 UVM_MAPFLAG(UVM_PROT_RW, UVM_PROT_RW, UVM_INH_COPY, 1007 UVM_ADV_NORMAL, 1008 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 1009 panic("init: couldn't allocate argument space"); 1010 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 1011 1012 ipx = 0; 1013 while (1) { 1014 if (boothowto & RB_ASKNAME) { 1015 printf("init path"); 1016 if (initpaths[ipx]) 1017 printf(" (default %s)", initpaths[ipx]); 1018 printf(": "); 1019 len = cngetsn(ipath, sizeof(ipath)-1); 1020 if (len == 4 && strcmp(ipath, "halt") == 0) { 1021 kern_reboot(RB_HALT, NULL); 1022 } else if (len == 6 && strcmp(ipath, "reboot") == 0) { 1023 kern_reboot(0, NULL); 1024#if defined(DDB) 1025 } else if (len == 3 && strcmp(ipath, "ddb") == 0) { 1026 console_debugger(); 1027 continue; 1028#endif 1029 } else if (len > 0 && ipath[0] == '/') { 1030 ipath[len] = '\0'; 1031 path = ipath; 1032 } else if (len == 0 && initpaths[ipx] != NULL) { 1033 path = initpaths[ipx++]; 1034 } else { 1035 printf("use absolute path, "); 1036#if defined(DDB) 1037 printf("\"ddb\", "); 1038#endif 1039 printf("\"halt\", or \"reboot\"\n"); 1040 continue; 1041 } 1042 } else { 1043 if ((path = initpaths[ipx++]) == NULL) { 1044 ipx = 0; 1045 boothowto |= RB_ASKNAME; 1046 continue; 1047 } 1048 } 1049 1050 ucp = (char *)USRSTACK; 1051 1052 /* 1053 * Construct the boot flag argument. 1054 */ 1055 flagsp = flags; 1056 *flagsp++ = '-'; 1057 options = 0; 1058 1059 if (boothowto & RB_SINGLE) { 1060 *flagsp++ = 's'; 1061 options = 1; 1062 } 1063#ifdef notyet 1064 if (boothowto & RB_FASTBOOT) { 1065 *flagsp++ = 'f'; 1066 options = 1; 1067 } 1068#endif 1069 1070 /* 1071 * Move out the flags (arg 1), if necessary. 1072 */ 1073 if (options != 0) { 1074 *flagsp++ = '\0'; 1075 i = flagsp - flags; 1076#ifdef DEBUG 1077 aprint_normal("init: copying out flags `%s' %d\n", flags, i); 1078#endif 1079 arg1 = STACK_ALLOC(ucp, i); 1080 ucp = STACK_MAX(arg1, i); 1081 if ((error = copyout((void *)flags, arg1, i)) != 0) 1082 goto copyerr; 1083 } else 1084 arg1 = NULL; 1085 1086 /* 1087 * Move out the file name (also arg 0). 1088 */ 1089 i = strlen(path) + 1; 1090#ifdef DEBUG 1091 aprint_normal("init: copying out path `%s' %d\n", path, i); 1092#else 1093 if (boothowto & RB_ASKNAME || path != initpaths[0]) 1094 printf("init: trying %s\n", path); 1095#endif 1096 arg0 = STACK_ALLOC(ucp, i); 1097 ucp = STACK_MAX(arg0, i); 1098 if ((error = copyout(path, arg0, i)) != 0) 1099 goto copyerr; 1100 1101 /* 1102 * Move out the arg pointers. 1103 */ 1104 ucp = (void *)STACK_ALIGN(ucp, STACK_ALIGNBYTES); 1105 uap = (char **)STACK_ALLOC(ucp, sizeof(argv)); 1106 SCARG(&args, path) = arg0; 1107 SCARG(&args, argp) = uap; 1108 SCARG(&args, envp) = NULL; 1109 slash = strrchr(path, '/'); 1110 1111 argv[0] = slash ? arg0 + (slash + 1 - path) : arg0; 1112 argv[1] = arg1; 1113 argv[2] = NULL; 1114 if ((error = copyout(argv, uap, sizeof(argv))) != 0) 1115 goto copyerr; 1116 1117 /* 1118 * Now try to exec the program. If it can't for any reason 1119 * other than it doesn't exist, complain. 1120 */ 1121 error = sys_execve(l, &args, retval); 1122 if (error == 0 || error == EJUSTRETURN) { 1123 KERNEL_UNLOCK_LAST(l); 1124 return; 1125 } 1126 printf("exec %s: error %d\n", path, error); 1127 } 1128 printf("init: not found\n"); 1129 panic("no init"); 1130copyerr: 1131 panic("copyout %d", error); 1132} 1133 1134/* 1135 * calculate cache size (in bytes) from physmem and vsize. 1136 */ 1137vaddr_t 1138calc_cache_size(vsize_t vsize, int pct, int va_pct) 1139{ 1140 paddr_t t; 1141 1142 /* XXX should consider competing cache if any */ 1143 /* XXX should consider submaps */ 1144 t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; 1145 if (vsize != 0) { 1146 vsize = (uintmax_t)vsize * va_pct / 100; 1147 if (t > vsize) { 1148 t = vsize; 1149 } 1150 } 1151 return t; 1152} 1153 1154/* 1155 * Print the system start up banner. 1156 * 1157 * - Print a limited banner if AB_SILENT. 1158 * - Always send normal banner to the log. 1159 */ 1160#define MEM_PBUFSIZE sizeof("99999 MB") 1161 1162void 1163banner(void) 1164{ 1165 static char notice[] = " Notice: this software is " 1166 "protected by copyright"; 1167 char pbuf[81]; 1168 void (*pr)(const char *, ...) __printflike(1, 2); 1169 int i; 1170 1171 if ((boothowto & AB_SILENT) != 0) { 1172 snprintf(pbuf, sizeof(pbuf), "%s %s (%s)", 1173 ostype, osrelease, kernel_ident); 1174 printf_nolog("%s", pbuf); 1175 for (i = 80 - strlen(pbuf) - sizeof(notice); i > 0; i--) 1176 printf(" "); 1177 printf_nolog("%s\n", notice); 1178 pr = aprint_normal; 1179 } else { 1180 pr = printf; 1181 } 1182 1183 memset(pbuf, 0, sizeof(pbuf)); 1184 (*pr)("%s%s", copyright, version); 1185 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)physmem)); 1186 (*pr)("total memory = %s\n", pbuf); 1187 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)uvm_availmem(false))); 1188 (*pr)("avail memory = %s\n", pbuf); 1189} 1190