init_main.c revision 1.545
1 /* $NetBSD: init_main.c,v 1.545 2023/09/12 16:17:21 ad Exp $ */ 2 3 /*- 4 * Copyright (c) 2008, 2009, 2019 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 1.545 2023/09/12 16:17:21 ad 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) 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/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> 234 235 extern time_t rootfstime; 236 237 #ifndef curlwp 238 struct lwp *curlwp = &lwp0; 239 #endif 240 struct proc *initproc; 241 242 struct vnode *rootvp, *swapdev_vp; 243 int boothowto; 244 int cold __read_mostly = 1; /* still working on startup */ 245 int shutting_down __read_mostly; /* system is shutting down */ 246 247 int start_init_exec; /* semaphore for start_init() */ 248 249 static void check_console(struct lwp *l); 250 static void start_init(void *); 251 static void configure(void); 252 static void configure2(void); 253 static void configure3(void); 254 void 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 */ 262 void 263 main(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 lock caches. */ 331 mutex_obj_init(); 332 333 /* Initialize radix trees (used by numerous subsystems). */ 334 radix_tree_init(); 335 336 /* Passive serialization. */ 337 pserialize_init(); 338 339 /* Initialize the extent manager. */ 340 extent_init(); 341 342 /* Do machine-dependent initialization. */ 343 cpu_startup(); 344 345 /* Initialize the sysctl subsystem. */ 346 sysctl_init(); 347 348 /* Initialize callouts, part 1. */ 349 callout_startup(); 350 351 /* Initialize the kernel authorization subsystem. */ 352 kauth_init(); 353 354 secmodel_init(); 355 356 spec_init(); 357 358 /* 359 * Set BPF op vector. Can't do this in bpf attach, since 360 * network drivers attach before bpf. 361 */ 362 bpf_setops(); 363 364 /* Initialize what we can in ipi(9) before CPUs are detected. */ 365 ipi_sysinit(); 366 367 /* Start module system. */ 368 module_init(); 369 module_hook_init(); 370 371 /* 372 * Initialize the kernel authorization subsystem and start the 373 * default security model, if any. We need to do this early 374 * enough so that subsystems relying on any of the aforementioned 375 * can work properly. Since the security model may dictate the 376 * credential inheritance policy, it is needed at least before 377 * any process is created, specifically proc0. 378 */ 379 module_init_class(MODULE_CLASS_SECMODEL); 380 381 /* Initialize the buffer cache */ 382 bufinit(); 383 biohist_init(); 384 385 #ifdef KERNHIST 386 sysctl_kernhist_init(); 387 #endif 388 389 390 #if defined(SPLASHSCREEN) && defined(SPLASHSCREEN_IMAGE) 391 size_t splash_size = (&_binary_splash_image_end - 392 &_binary_splash_image_start) * sizeof(void *); 393 splash_setimage(&_binary_splash_image_start, splash_size); 394 #endif 395 396 /* Initialize sockets. */ 397 soinit(); 398 399 /* 400 * The following things must be done before autoconfiguration. 401 */ 402 rnd_init(); /* initialize entropy pool */ 403 404 cprng_init(); /* initialize cryptographic PRNG */ 405 406 /* Initialize process and pgrp structures. */ 407 procinit(); 408 lwpinit(); 409 410 /* Must be called after lwpinit (lwpinit_specificdata) */ 411 psref_init(); 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 cwd structures */ 512 cwd_sys_init(); 513 514 /* Initialize kqueue. */ 515 kqueue_init(); 516 517 inittimecounter(); 518 ntp_init(); 519 520 /* Initialize tty subsystem. */ 521 tty_init(); 522 ttyldisc_init(); 523 524 /* Initialize the buffer cache, part 2. */ 525 bufinit2(); 526 527 /* Initialize the disk wedge subsystem. */ 528 dkwedge_init(); 529 530 /* Initialize pfil */ 531 pfil_init(); 532 533 /* Initialize interfaces. */ 534 ifinit1(); 535 536 spldebug_start(); 537 538 /* Initialize sockets thread(s) */ 539 soinit1(); 540 541 /* 542 * Initialize the bufq strategy sub-system and any built-in 543 * strategy modules - they may be needed by some devices during 544 * auto-configuration 545 */ 546 bufq_init(); 547 module_init_class(MODULE_CLASS_BUFQ); 548 549 /* Configure the system hardware. This will enable interrupts. */ 550 configure(); 551 #ifdef __HAVE_LEGACY_INTRCNT 552 evcnt_attach_legacy_intrcnt(); 553 #endif 554 555 /* Enable deferred processing of RNG samples */ 556 rnd_init_softint(); 557 558 /* Once all CPUs are detected, initialize the per-CPU cprng_fast. */ 559 cprng_fast_init(); 560 561 /* 562 * Now that softints can be established, start monitoring 563 * system heartbeat on all CPUs. 564 */ 565 heartbeat_start(); 566 567 ssp_init(); 568 569 ubc_init(); /* must be after autoconfig */ 570 571 mm_init(); 572 573 configure2(); 574 575 /* Initialize the rest of ipi(9) after CPUs have been detected. */ 576 ipi_percpu_init(); 577 578 futex_sys_init(); 579 580 /* Now timer is working. Enable preemption. */ 581 kpreempt_enable(); 582 583 /* Get the threads going and into any sleeps before continuing. */ 584 yield(); 585 586 vmem_rehash_start(); /* must be before exec_init */ 587 588 /* Initialize exec structures */ 589 exec_init(1); /* seminit calls exithook_establish() */ 590 591 #if NVERIEXEC > 0 592 /* 593 * Initialise the Veriexec subsystem. 594 */ 595 veriexec_init(); 596 #endif /* NVERIEXEC > 0 */ 597 598 pax_init(); 599 600 #ifdef IPSEC 601 /* Attach network crypto subsystem */ 602 ipsec_attach(); 603 #endif 604 605 /* 606 * Initialize protocols. Block reception of incoming packets 607 * until everything is ready. 608 */ 609 s = splnet(); 610 ifinit(); 611 #if defined(INET) || defined(INET6) 612 lltableinit(); 613 #endif 614 domaininit(true); 615 ifinit_post(); 616 if_attachdomain(); 617 splx(s); 618 619 #ifdef GPROF 620 /* Initialize kernel profiling. */ 621 kmstartup(); 622 #endif 623 624 /* Initialize system accounting. */ 625 acct_init(); 626 627 #ifndef PIPE_SOCKETPAIR 628 /* Initialize pipes. */ 629 pipe_init(); 630 #endif 631 632 #ifdef KTRACE 633 /* Initialize ktrace. */ 634 ktrinit(); 635 #endif 636 637 machdep_init(); 638 639 procinit_sysctl(); 640 641 scdebug_init(); 642 643 /* 644 * Create process 1 (init(8)). We do this now, as Unix has 645 * historically had init be process 1, and changing this would 646 * probably upset a lot of people. 647 * 648 * Note that process 1 won't immediately exec init(8), but will 649 * wait for us to inform it that the root file system has been 650 * mounted. 651 */ 652 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL)) 653 panic("fork init"); 654 655 /* 656 * The initproc variable cannot be initialized in start_init as there 657 * is a race between vfs_mountroot and start_init. 658 */ 659 mutex_enter(&proc_lock); 660 initproc = proc_find_raw(1); 661 mutex_exit(&proc_lock); 662 663 /* 664 * Load any remaining builtin modules, and hand back temporary 665 * storage to the VM system. Then require force when loading any 666 * remaining un-init'ed built-in modules to avoid later surprises. 667 */ 668 module_init_class(MODULE_CLASS_ANY); 669 module_builtin_require_force(); 670 671 /* 672 * Finalize configuration now that all real devices have been 673 * found. This needs to be done before the root device is 674 * selected, since finalization may create the root device. 675 */ 676 config_finalize(); 677 678 sysctl_finalize(); 679 680 /* 681 * Now that autoconfiguration has completed, we can determine 682 * the root and dump devices. 683 */ 684 cpu_rootconf(); 685 cpu_dumpconf(); 686 687 /* Mount the root file system. */ 688 do { 689 domountroothook(root_device); 690 if ((error = vfs_mountroot())) { 691 printf("cannot mount root, error = %d\n", error); 692 boothowto |= RB_ASKNAME; 693 setroot(root_device, 694 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 695 } 696 } while (error != 0); 697 mountroothook_destroy(); 698 699 configure3(); 700 701 /* 702 * Initialise the time-of-day clock, passing the time recorded 703 * in the root filesystem (if any) for use by systems that 704 * don't have a non-volatile time-of-day device. 705 */ 706 inittodr(rootfstime); 707 708 /* 709 * Now can look at time, having had a chance to verify the time 710 * from the file system. Reset l->l_rtime as it may have been 711 * munched in mi_switch() after the time got set. 712 */ 713 getnanotime(&time); 714 715 mutex_enter(&proc_lock); 716 LIST_FOREACH(p, &allproc, p_list) { 717 KASSERT((p->p_flag & PK_MARKER) == 0); 718 mutex_enter(p->p_lock); 719 TIMESPEC_TO_TIMEVAL(&p->p_stats->p_start, &time); 720 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 721 lwp_lock(l); 722 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 723 lwp_unlock(l); 724 } 725 mutex_exit(p->p_lock); 726 } 727 mutex_exit(&proc_lock); 728 binuptime(&curlwp->l_stime); 729 730 for (CPU_INFO_FOREACH(cii, ci)) { 731 ci->ci_schedstate.spc_lastmod = time_second; 732 } 733 734 /* Create the pageout daemon kernel thread. */ 735 uvm_swap_init(); 736 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 737 NULL, NULL, "pgdaemon")) 738 panic("fork pagedaemon"); 739 740 /* Create the filesystem syncer kernel thread. */ 741 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 742 NULL, NULL, "ioflush")) 743 panic("fork syncer"); 744 745 /* Wait for final configure threads to complete. */ 746 config_finalize_mountroot(); 747 748 /* 749 * Okay, now we can let init(8) exec! It's off to userland! 750 */ 751 mutex_enter(&proc_lock); 752 start_init_exec = 1; 753 cv_broadcast(&lbolt); 754 mutex_exit(&proc_lock); 755 756 /* The scheduler is an infinite loop. */ 757 uvm_scheduler(); 758 /* NOTREACHED */ 759 } 760 761 /* 762 * Configure the system's hardware. 763 */ 764 static void 765 configure(void) 766 { 767 768 /* 769 * XXX 770 * callout_setfunc() requires mutex(9) so it can't be in config_init() 771 * on amiga and atari which use config_init() and autoconf(9) functions 772 * to initialize console devices. 773 */ 774 config_twiddle_init(); 775 776 pmf_init(); 777 778 /* Initialize driver modules */ 779 module_init_class(MODULE_CLASS_DRIVER); 780 781 userconf_init(); 782 if (boothowto & RB_USERCONF) 783 userconf_prompt(); 784 785 if ((boothowto & (AB_SILENT|AB_VERBOSE)) == AB_SILENT) { 786 printf_nolog("Detecting hardware..."); 787 } 788 789 /* 790 * Do the machine-dependent portion of autoconfiguration. This 791 * sets the configuration machinery here in motion by "finding" 792 * the root bus. When this function returns, we expect interrupts 793 * to be enabled. 794 */ 795 cpu_configure(); 796 } 797 798 static void 799 configure2(void) 800 { 801 CPU_INFO_ITERATOR cii; 802 struct cpu_info *ci; 803 int s; 804 805 /* Fix up CPU topology info, which has all been collected by now. */ 806 cpu_topology_init(); 807 808 /* 809 * Now that we've found all the hardware, start the real time 810 * and statistics clocks. 811 */ 812 initclocks(); 813 814 cold = 0; /* clocks are running, we're warm now! */ 815 s = splsched(); 816 curcpu()->ci_schedstate.spc_flags |= SPCF_RUNNING; 817 splx(s); 818 819 /* Setup the runqueues and scheduler. */ 820 runq_init(); 821 synch_init(); 822 823 /* Boot the secondary processors. */ 824 for (CPU_INFO_FOREACH(cii, ci)) { 825 uvm_cpu_attach(ci); 826 } 827 828 /* Decide how to partition free memory. */ 829 uvm_page_rebucket(); 830 831 mp_online = true; 832 #if defined(MULTIPROCESSOR) 833 cpu_boot_secondary_processors(); 834 #endif 835 836 /* 837 * Bus scans can make it appear as if the system has paused, so 838 * twiddle constantly while config_interrupts() jobs are running. 839 */ 840 config_twiddle_fn(NULL); 841 842 /* 843 * Create threads to call back and finish configuration for 844 * devices that want interrupts enabled. 845 */ 846 config_create_interruptthreads(); 847 } 848 849 static void 850 configure3(void) 851 { 852 853 /* 854 * Create threads to call back and finish configuration for 855 * devices that want the mounted root file system. 856 */ 857 config_create_mountrootthreads(); 858 859 /* Get the threads going and into any sleeps before continuing. */ 860 yield(); 861 } 862 863 static void 864 rootconf_handle_wedges(void) 865 { 866 struct disklabel label; 867 struct partition *p; 868 struct vnode *vp; 869 daddr_t startblk; 870 uint64_t nblks; 871 device_t dev; 872 int error; 873 874 if (booted_nblks) { 875 /* 876 * bootloader passed geometry 877 */ 878 dev = booted_device; 879 startblk = booted_startblk; 880 nblks = booted_nblks; 881 882 /* 883 * keep booted_device and booted_partition 884 * in case the kernel doesn't identify a wedge 885 */ 886 } else { 887 /* 888 * bootloader passed partition number 889 * 890 * We cannot ask the partition device directly when it is 891 * covered by a wedge. Instead we look up the geometry in 892 * the disklabel. 893 */ 894 vp = opendisk(booted_device); 895 896 if (vp == NULL) 897 return; 898 899 VOP_UNLOCK(vp); 900 error = VOP_IOCTL(vp, DIOCGDINFO, &label, FREAD, NOCRED); 901 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 902 VOP_CLOSE(vp, FREAD, NOCRED); 903 vput(vp); 904 if (error) 905 return; 906 907 KASSERT(booted_partition >= 0 908 && booted_partition < MAXPARTITIONS); 909 910 p = &label.d_partitions[booted_partition]; 911 912 dev = booted_device; 913 startblk = p->p_offset; 914 nblks = p->p_size; 915 } 916 917 dev = dkwedge_find_partition(dev, startblk, nblks); 918 if (dev != NULL) { 919 booted_device = dev; 920 booted_partition = 0; 921 } 922 } 923 924 void 925 rootconf(void) 926 { 927 if (booted_device != NULL) 928 rootconf_handle_wedges(); 929 930 setroot(booted_device, booted_partition); 931 } 932 933 static void 934 check_console(struct lwp *l) 935 { 936 struct vnode *vp; 937 int error; 938 939 error = namei_simple_kernel("/dev/console", 940 NSM_FOLLOW_NOEMULROOT, &vp); 941 if (error == 0) { 942 vrele(vp); 943 } else if (error == ENOENT) { 944 if (boothowto & (AB_VERBOSE|AB_DEBUG)) 945 printf("warning: no /dev/console\n"); 946 } else { 947 printf("warning: lookup /dev/console: error %d\n", error); 948 } 949 } 950 951 /* 952 * List of paths to try when searching for "init". 953 */ 954 static const char * const initpaths[] = { 955 "/sbin/init", 956 "/sbin/oinit", 957 "/sbin/init.bak", 958 "/rescue/init", 959 NULL, 960 }; 961 962 /* 963 * Start the initial user process; try exec'ing each pathname in "initpaths". 964 * The program is invoked with one argument containing the boot flags. 965 */ 966 static void 967 start_init(void *arg) 968 { 969 struct lwp *l = arg; 970 struct proc *p = l->l_proc; 971 vaddr_t addr; 972 struct sys_execve_args /* { 973 syscallarg(const char *) path; 974 syscallarg(char * const *) argp; 975 syscallarg(char * const *) envp; 976 } */ args; 977 int options, i, error; 978 register_t retval[2]; 979 char flags[4], *flagsp; 980 const char *path, *slash; 981 char *ucp, **uap, *arg0, *arg1, *argv[3]; 982 char ipath[129]; 983 int ipx, len; 984 985 /* 986 * Now in process 1. 987 */ 988 strncpy(p->p_comm, "init", MAXCOMLEN); 989 990 /* 991 * Wait for main() to tell us that it's safe to exec. 992 */ 993 mutex_enter(&proc_lock); 994 while (start_init_exec == 0) 995 cv_wait(&lbolt, &proc_lock); 996 mutex_exit(&proc_lock); 997 998 /* 999 * This is not the right way to do this. We really should 1000 * hand-craft a descriptor onto /dev/console to hand to init, 1001 * but that's a _lot_ more work, and the benefit from this easy 1002 * hack makes up for the "good is the enemy of the best" effect. 1003 */ 1004 check_console(l); 1005 1006 /* 1007 * Need just enough stack to hold the faked-up "execve()" arguments. 1008 */ 1009 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 1010 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 1011 NULL, UVM_UNKNOWN_OFFSET, 0, 1012 UVM_MAPFLAG(UVM_PROT_RW, UVM_PROT_RW, UVM_INH_COPY, 1013 UVM_ADV_NORMAL, 1014 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 1015 panic("init: couldn't allocate argument space"); 1016 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 1017 1018 ipx = 0; 1019 while (1) { 1020 if (boothowto & RB_ASKNAME) { 1021 printf("init path"); 1022 if (initpaths[ipx]) 1023 printf(" (default %s)", initpaths[ipx]); 1024 printf(": "); 1025 len = cngetsn(ipath, sizeof(ipath)-1); 1026 if (len == 4 && strcmp(ipath, "halt") == 0) { 1027 kern_reboot(RB_HALT, NULL); 1028 } else if (len == 6 && strcmp(ipath, "reboot") == 0) { 1029 kern_reboot(0, NULL); 1030 #if defined(DDB) 1031 } else if (len == 3 && strcmp(ipath, "ddb") == 0) { 1032 console_debugger(); 1033 continue; 1034 #endif 1035 } else if (len > 0 && ipath[0] == '/') { 1036 ipath[len] = '\0'; 1037 path = ipath; 1038 } else if (len == 0 && initpaths[ipx] != NULL) { 1039 path = initpaths[ipx++]; 1040 } else { 1041 printf("use absolute path, "); 1042 #if defined(DDB) 1043 printf("\"ddb\", "); 1044 #endif 1045 printf("\"halt\", or \"reboot\"\n"); 1046 continue; 1047 } 1048 } else { 1049 if ((path = initpaths[ipx++]) == NULL) { 1050 ipx = 0; 1051 boothowto |= RB_ASKNAME; 1052 continue; 1053 } 1054 } 1055 1056 ucp = (char *)USRSTACK; 1057 1058 /* 1059 * Construct the boot flag argument. 1060 */ 1061 flagsp = flags; 1062 *flagsp++ = '-'; 1063 options = 0; 1064 1065 if (boothowto & RB_SINGLE) { 1066 *flagsp++ = 's'; 1067 options = 1; 1068 } 1069 #ifdef notyet 1070 if (boothowto & RB_FASTBOOT) { 1071 *flagsp++ = 'f'; 1072 options = 1; 1073 } 1074 #endif 1075 1076 /* 1077 * Move out the flags (arg 1), if necessary. 1078 */ 1079 if (options != 0) { 1080 *flagsp++ = '\0'; 1081 i = flagsp - flags; 1082 #ifdef DEBUG 1083 aprint_normal("init: copying out flags `%s' %d\n", flags, i); 1084 #endif 1085 arg1 = STACK_ALLOC(ucp, i); 1086 ucp = STACK_MAX(arg1, i); 1087 if ((error = copyout((void *)flags, arg1, i)) != 0) 1088 goto copyerr; 1089 } else 1090 arg1 = NULL; 1091 1092 /* 1093 * Move out the file name (also arg 0). 1094 */ 1095 i = strlen(path) + 1; 1096 #ifdef DEBUG 1097 aprint_normal("init: copying out path `%s' %d\n", path, i); 1098 #else 1099 if (boothowto & RB_ASKNAME || path != initpaths[0]) 1100 printf("init: trying %s\n", path); 1101 #endif 1102 arg0 = STACK_ALLOC(ucp, i); 1103 ucp = STACK_MAX(arg0, i); 1104 if ((error = copyout(path, arg0, i)) != 0) 1105 goto copyerr; 1106 1107 /* 1108 * Move out the arg pointers. 1109 */ 1110 ucp = (void *)STACK_ALIGN(ucp, STACK_ALIGNBYTES); 1111 uap = (char **)STACK_ALLOC(ucp, sizeof(argv)); 1112 SCARG(&args, path) = arg0; 1113 SCARG(&args, argp) = uap; 1114 SCARG(&args, envp) = NULL; 1115 slash = strrchr(path, '/'); 1116 1117 argv[0] = slash ? arg0 + (slash + 1 - path) : arg0; 1118 argv[1] = arg1; 1119 argv[2] = NULL; 1120 if ((error = copyout(argv, uap, sizeof(argv))) != 0) 1121 goto copyerr; 1122 1123 /* 1124 * Now try to exec the program. If it can't for any reason 1125 * other than it doesn't exist, complain. 1126 */ 1127 error = sys_execve(l, &args, retval); 1128 if (error == 0 || error == EJUSTRETURN) { 1129 KERNEL_UNLOCK_LAST(l); 1130 return; 1131 } 1132 printf("exec %s: error %d\n", path, error); 1133 } 1134 printf("init: not found\n"); 1135 panic("no init"); 1136 copyerr: 1137 panic("copyout %d", error); 1138 } 1139 1140 /* 1141 * calculate cache size (in bytes) from physmem and vsize. 1142 */ 1143 vaddr_t 1144 calc_cache_size(vsize_t vsize, int pct, int va_pct) 1145 { 1146 paddr_t t; 1147 1148 /* XXX should consider competing cache if any */ 1149 /* XXX should consider submaps */ 1150 t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; 1151 if (vsize != 0) { 1152 vsize = (uintmax_t)vsize * va_pct / 100; 1153 if (t > vsize) { 1154 t = vsize; 1155 } 1156 } 1157 return t; 1158 } 1159 1160 /* 1161 * Print the system start up banner. 1162 * 1163 * - Print a limited banner if AB_SILENT. 1164 * - Always send normal banner to the log. 1165 */ 1166 #define MEM_PBUFSIZE sizeof("99999 MB") 1167 1168 void 1169 banner(void) 1170 { 1171 static char notice[] = " Notice: this software is " 1172 "protected by copyright"; 1173 char pbuf[81]; 1174 void (*pr)(const char *, ...) __printflike(1, 2); 1175 int i; 1176 1177 if ((boothowto & AB_SILENT) != 0) { 1178 snprintf(pbuf, sizeof(pbuf), "%s %s (%s)", 1179 ostype, osrelease, kernel_ident); 1180 printf_nolog("%s", pbuf); 1181 for (i = 80 - strlen(pbuf) - sizeof(notice); i > 0; i--) 1182 printf(" "); 1183 printf_nolog("%s\n", notice); 1184 pr = aprint_normal; 1185 } else { 1186 pr = printf; 1187 } 1188 1189 memset(pbuf, 0, sizeof(pbuf)); 1190 (*pr)("%s%s", copyright, version); 1191 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)physmem)); 1192 (*pr)("total memory = %s\n", pbuf); 1193 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)uvm_availmem(false))); 1194 (*pr)("avail memory = %s\n", pbuf); 1195 } 1196
Indexes created Sun Sep 21 15:09:59 GMT 2025