init_main.c revision 1.530
1 /* $NetBSD: init_main.c,v 1.530 2020/09/07 03:50:41 thorpej 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 1.530 2020/09/07 03:50:41 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_wapbl.h" 116 #include "opt_ptrace.h" 117 #include "opt_splash.h" 118 #include "opt_kernhist.h" 119 #include "opt_gprof.h" 120 121 #if defined(SPLASHSCREEN) && defined(makeoptions_SPLASHSCREEN_IMAGE) 122 extern void *_binary_splash_image_start; 123 extern void *_binary_splash_image_end; 124 #endif 125 126 #include "ksyms.h" 127 128 #include "veriexec.h" 129 130 #include <sys/param.h> 131 #include <sys/acct.h> 132 #include <sys/filedesc.h> 133 #include <sys/file.h> 134 #include <sys/errno.h> 135 #include <sys/callout.h> 136 #include <sys/cpu.h> 137 #include <sys/cpufreq.h> 138 #include <sys/spldebug.h> 139 #include <sys/kernel.h> 140 #include <sys/mount.h> 141 #include <sys/proc.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> 202 203 #include <sys/syscall.h> 204 #include <sys/syscallargs.h> 205 206 #include <sys/pax.h> 207 208 #include <dev/clock_subr.h> 209 210 #include <secmodel/secmodel.h> 211 212 #include <ufs/ufs/quota.h> 213 214 #include <miscfs/genfs/genfs.h> 215 #include <miscfs/specfs/specdev.h> 216 217 #include <sys/cpu.h> 218 219 #include <uvm/uvm.h> /* extern struct uvm uvm */ 220 221 #include <dev/cons.h> 222 #include <dev/splash/splash.h> 223 224 #include <net/bpf.h> 225 #include <net/if.h> 226 #include <net/pfil.h> 227 #include <net/raw_cb.h> 228 #include <net/if_llatbl.h> 229 230 #include <prop/proplib.h> 231 232 #include <sys/userconf.h> 233 234 extern struct lwp lwp0; 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 rw_obj_init(); 333 334 /* Initialize radix trees (used by numerous subsystems). */ 335 radix_tree_init(); 336 337 /* Passive serialization. */ 338 pserialize_init(); 339 340 /* Initialize the extent manager. */ 341 extent_init(); 342 343 /* Do machine-dependent initialization. */ 344 cpu_startup(); 345 346 /* Initialize the sysctl subsystem. */ 347 sysctl_init(); 348 349 /* Initialize callouts, part 1. */ 350 callout_startup(); 351 352 /* Initialize the kernel authorization subsystem. */ 353 kauth_init(); 354 355 secmodel_init(); 356 357 spec_init(); 358 359 /* 360 * Set BPF op vector. Can't do this in bpf attach, since 361 * network drivers attach before bpf. 362 */ 363 bpf_setops(); 364 365 /* Start module system. */ 366 module_init(); 367 module_hook_init(); 368 369 /* 370 * Initialize the kernel authorization subsystem and start the 371 * default security model, if any. We need to do this early 372 * enough so that subsystems relying on any of the aforementioned 373 * can work properly. Since the security model may dictate the 374 * credential inheritance policy, it is needed at least before 375 * any process is created, specifically proc0. 376 */ 377 module_init_class(MODULE_CLASS_SECMODEL); 378 379 /* Initialize the buffer cache */ 380 bufinit(); 381 biohist_init(); 382 383 #ifdef KERNHIST 384 sysctl_kernhist_init(); 385 #endif 386 387 388 #if defined(SPLASHSCREEN) && defined(SPLASHSCREEN_IMAGE) 389 size_t splash_size = (&_binary_splash_image_end - 390 &_binary_splash_image_start) * sizeof(void *); 391 splash_setimage(&_binary_splash_image_start, splash_size); 392 #endif 393 394 /* Initialize sockets. */ 395 soinit(); 396 397 /* 398 * The following things must be done before autoconfiguration. 399 */ 400 rnd_init(); /* initialize entropy pool */ 401 402 cprng_init(); /* initialize cryptographic PRNG */ 403 404 /* Initialize process and pgrp structures. */ 405 procinit(); 406 lwpinit(); 407 408 /* Must be called after lwpinit (lwpinit_specificdata) */ 409 psref_init(); 410 411 /* Initialize signal-related data structures. */ 412 signal_init(); 413 414 /* Initialize resource management. */ 415 resource_init(); 416 417 /* Create process 0. */ 418 proc0_init(); 419 lwp0_init(); 420 421 /* Disable preemption during boot. */ 422 kpreempt_disable(); 423 424 /* Initialize the threadpool system. */ 425 threadpools_init(); 426 427 /* Initialize the UID hash table. */ 428 uid_init(); 429 430 /* Charge root for one process. */ 431 (void)chgproccnt(0, 1); 432 433 /* Initialize timekeeping. */ 434 time_init(); 435 436 /* Initialize the run queues, turnstiles and sleep queues. */ 437 sched_rqinit(); 438 turnstile_init(); 439 sleeptab_init(&sleeptab); 440 441 sched_init(); 442 443 /* Initialize processor-sets */ 444 psets_init(); 445 446 /* Initialize cpufreq(9) */ 447 cpufreq_init(); 448 449 /* MI initialization of the boot cpu */ 450 error = mi_cpu_attach(curcpu()); 451 KASSERT(error == 0); 452 453 /* Initialize timekeeping, part 2. */ 454 time_init2(); 455 456 /* 457 * Initialize mbuf's. Do this now because we might attempt to 458 * allocate mbufs or mbuf clusters during autoconfiguration. 459 */ 460 mbinit(); 461 462 /* Initialize I/O statistics. */ 463 iostat_init(); 464 465 /* Initialize the log device. */ 466 loginit(); 467 468 /* Second part of module system initialization. */ 469 module_start_unload_thread(); 470 471 /* Initialize autoconf data structures before any modules are loaded */ 472 config_init_mi(); 473 474 /* Initialize the file systems. */ 475 #ifdef NVNODE_IMPLICIT 476 /* 477 * If maximum number of vnodes in namei vnode cache is not explicitly 478 * defined in kernel config, adjust the number such as we use roughly 479 * 10% of memory for vnodes and associated data structures in the 480 * assumed worst case. Do not provide fewer than NVNODE vnodes. 481 */ 482 usevnodes = calc_cache_size(vmem_size(kmem_arena, VMEM_FREE|VMEM_ALLOC), 483 10, VNODE_KMEM_MAXPCT) / VNODE_COST; 484 if (usevnodes > desiredvnodes) 485 desiredvnodes = usevnodes; 486 #endif 487 488 /* Initialize fstrans. */ 489 fstrans_init(); 490 491 vfsinit(); 492 lf_init(); 493 494 /* Initialize the file descriptor system. */ 495 fd_sys_init(); 496 497 /* Initialize cwd structures */ 498 cwd_sys_init(); 499 500 /* Initialize kqueue. */ 501 kqueue_init(); 502 503 inittimecounter(); 504 ntp_init(); 505 506 /* Initialize tty subsystem. */ 507 tty_init(); 508 ttyldisc_init(); 509 510 /* Initialize the buffer cache, part 2. */ 511 bufinit2(); 512 513 /* Initialize the disk wedge subsystem. */ 514 dkwedge_init(); 515 516 /* Initialize pfil */ 517 pfil_init(); 518 519 /* Initialize interfaces. */ 520 ifinit1(); 521 522 spldebug_start(); 523 524 /* Initialize sockets thread(s) */ 525 soinit1(); 526 527 /* 528 * Initialize the bufq strategy sub-system and any built-in 529 * strategy modules - they may be needed by some devices during 530 * auto-configuration 531 */ 532 bufq_init(); 533 module_init_class(MODULE_CLASS_BUFQ); 534 535 /* Configure the system hardware. This will enable interrupts. */ 536 configure(); 537 538 /* Once all CPUs are detected, initialize the per-CPU cprng_fast. */ 539 cprng_fast_init(); 540 541 ssp_init(); 542 543 ubc_init(); /* must be after autoconfig */ 544 545 mm_init(); 546 547 configure2(); 548 549 ipi_sysinit(); 550 551 futex_sys_init(); 552 553 /* Now timer is working. Enable preemption. */ 554 kpreempt_enable(); 555 556 /* Get the threads going and into any sleeps before continuing. */ 557 yield(); 558 559 /* Enable deferred processing of RNG samples */ 560 rnd_init_softint(); 561 562 vmem_rehash_start(); /* must be before exec_init */ 563 564 /* Initialize exec structures */ 565 exec_init(1); /* seminit calls exithook_establish() */ 566 567 #if NVERIEXEC > 0 568 /* 569 * Initialise the Veriexec subsystem. 570 */ 571 veriexec_init(); 572 #endif /* NVERIEXEC > 0 */ 573 574 pax_init(); 575 576 #ifdef IPSEC 577 /* Attach network crypto subsystem */ 578 ipsec_attach(); 579 #endif 580 581 /* 582 * Initialize protocols. Block reception of incoming packets 583 * until everything is ready. 584 */ 585 s = splnet(); 586 ifinit(); 587 #if defined(INET) || defined(INET6) 588 lltableinit(); 589 #endif 590 domaininit(true); 591 ifinit_post(); 592 if_attachdomain(); 593 splx(s); 594 595 #ifdef GPROF 596 /* Initialize kernel profiling. */ 597 kmstartup(); 598 #endif 599 600 /* Initialize system accounting. */ 601 acct_init(); 602 603 #ifndef PIPE_SOCKETPAIR 604 /* Initialize pipes. */ 605 pipe_init(); 606 #endif 607 608 #ifdef KTRACE 609 /* Initialize ktrace. */ 610 ktrinit(); 611 #endif 612 613 machdep_init(); 614 615 procinit_sysctl(); 616 617 scdebug_init(); 618 619 /* 620 * Create process 1 (init(8)). We do this now, as Unix has 621 * historically had init be process 1, and changing this would 622 * probably upset a lot of people. 623 * 624 * Note that process 1 won't immediately exec init(8), but will 625 * wait for us to inform it that the root file system has been 626 * mounted. 627 */ 628 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL)) 629 panic("fork init"); 630 631 /* 632 * The initproc variable cannot be initialized in start_init as there 633 * is a race between vfs_mountroot and start_init. 634 */ 635 mutex_enter(&proc_lock); 636 initproc = proc_find_raw(1); 637 mutex_exit(&proc_lock); 638 639 /* 640 * Load any remaining builtin modules, and hand back temporary 641 * storage to the VM system. Then require force when loading any 642 * remaining un-init'ed built-in modules to avoid later surprises. 643 */ 644 module_init_class(MODULE_CLASS_ANY); 645 module_builtin_require_force(); 646 647 /* 648 * Finalize configuration now that all real devices have been 649 * found. This needs to be done before the root device is 650 * selected, since finalization may create the root device. 651 */ 652 config_finalize(); 653 654 sysctl_finalize(); 655 656 /* 657 * Now that autoconfiguration has completed, we can determine 658 * the root and dump devices. 659 */ 660 cpu_rootconf(); 661 cpu_dumpconf(); 662 663 /* Mount the root file system. */ 664 do { 665 domountroothook(root_device); 666 if ((error = vfs_mountroot())) { 667 printf("cannot mount root, error = %d\n", error); 668 boothowto |= RB_ASKNAME; 669 setroot(root_device, 670 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 671 } 672 } while (error != 0); 673 mountroothook_destroy(); 674 675 configure3(); 676 677 /* 678 * Initialise the time-of-day clock, passing the time recorded 679 * in the root filesystem (if any) for use by systems that 680 * don't have a non-volatile time-of-day device. 681 */ 682 inittodr(rootfstime); 683 684 /* 685 * Now can look at time, having had a chance to verify the time 686 * from the file system. Reset l->l_rtime as it may have been 687 * munched in mi_switch() after the time got set. 688 */ 689 getnanotime(&time); 690 691 mutex_enter(&proc_lock); 692 LIST_FOREACH(p, &allproc, p_list) { 693 KASSERT((p->p_flag & PK_MARKER) == 0); 694 mutex_enter(p->p_lock); 695 TIMESPEC_TO_TIMEVAL(&p->p_stats->p_start, &time); 696 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 697 lwp_lock(l); 698 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 699 lwp_unlock(l); 700 } 701 mutex_exit(p->p_lock); 702 } 703 mutex_exit(&proc_lock); 704 binuptime(&curlwp->l_stime); 705 706 for (CPU_INFO_FOREACH(cii, ci)) { 707 ci->ci_schedstate.spc_lastmod = time_second; 708 } 709 710 /* Create the pageout daemon kernel thread. */ 711 uvm_swap_init(); 712 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 713 NULL, NULL, "pgdaemon")) 714 panic("fork pagedaemon"); 715 716 /* Create the filesystem syncer kernel thread. */ 717 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 718 NULL, NULL, "ioflush")) 719 panic("fork syncer"); 720 721 /* Wait for final configure threads to complete. */ 722 config_finalize_mountroot(); 723 724 /* 725 * Okay, now we can let init(8) exec! It's off to userland! 726 */ 727 mutex_enter(&proc_lock); 728 start_init_exec = 1; 729 cv_broadcast(&lbolt); 730 mutex_exit(&proc_lock); 731 732 /* The scheduler is an infinite loop. */ 733 uvm_scheduler(); 734 /* NOTREACHED */ 735 } 736 737 /* 738 * Configure the system's hardware. 739 */ 740 static void 741 configure(void) 742 { 743 744 /* 745 * XXX 746 * callout_setfunc() requires mutex(9) so it can't be in config_init() 747 * on amiga and atari which use config_init() and autoconf(9) fucntions 748 * to initialize console devices. 749 */ 750 config_twiddle_init(); 751 752 pmf_init(); 753 754 /* Initialize driver modules */ 755 module_init_class(MODULE_CLASS_DRIVER); 756 757 userconf_init(); 758 if (boothowto & RB_USERCONF) 759 userconf_prompt(); 760 761 if ((boothowto & (AB_SILENT|AB_VERBOSE)) == AB_SILENT) { 762 printf_nolog("Detecting hardware..."); 763 } 764 765 /* 766 * Do the machine-dependent portion of autoconfiguration. This 767 * sets the configuration machinery here in motion by "finding" 768 * the root bus. When this function returns, we expect interrupts 769 * to be enabled. 770 */ 771 cpu_configure(); 772 } 773 774 static void 775 configure2(void) 776 { 777 CPU_INFO_ITERATOR cii; 778 struct cpu_info *ci; 779 int s; 780 781 /* Fix up CPU topology info, which has all been collected by now. */ 782 cpu_topology_init(); 783 784 /* 785 * Now that we've found all the hardware, start the real time 786 * and statistics clocks. 787 */ 788 initclocks(); 789 790 cold = 0; /* clocks are running, we're warm now! */ 791 s = splsched(); 792 curcpu()->ci_schedstate.spc_flags |= SPCF_RUNNING; 793 splx(s); 794 795 /* Setup the runqueues and scheduler. */ 796 runq_init(); 797 synch_init(); 798 799 /* Boot the secondary processors. */ 800 for (CPU_INFO_FOREACH(cii, ci)) { 801 uvm_cpu_attach(ci); 802 } 803 804 /* Decide how to partition free memory. */ 805 uvm_page_rebucket(); 806 807 mp_online = true; 808 #if defined(MULTIPROCESSOR) 809 cpu_boot_secondary_processors(); 810 #endif 811 812 /* 813 * Bus scans can make it appear as if the system has paused, so 814 * twiddle constantly while config_interrupts() jobs are running. 815 */ 816 config_twiddle_fn(NULL); 817 818 /* 819 * Create threads to call back and finish configuration for 820 * devices that want interrupts enabled. 821 */ 822 config_create_interruptthreads(); 823 } 824 825 static void 826 configure3(void) 827 { 828 829 /* 830 * Create threads to call back and finish configuration for 831 * devices that want the mounted root file system. 832 */ 833 config_create_mountrootthreads(); 834 835 /* Get the threads going and into any sleeps before continuing. */ 836 yield(); 837 } 838 839 static void 840 rootconf_handle_wedges(void) 841 { 842 struct disklabel label; 843 struct partition *p; 844 struct vnode *vp; 845 daddr_t startblk; 846 uint64_t nblks; 847 device_t dev; 848 int error; 849 850 if (booted_nblks) { 851 /* 852 * bootloader passed geometry 853 */ 854 dev = booted_device; 855 startblk = booted_startblk; 856 nblks = booted_nblks; 857 858 /* 859 * keep booted_device and booted_partition 860 * in case the kernel doesn't identify a wedge 861 */ 862 } else { 863 /* 864 * bootloader passed partition number 865 * 866 * We cannot ask the partition device directly when it is 867 * covered by a wedge. Instead we look up the geometry in 868 * the disklabel. 869 */ 870 vp = opendisk(booted_device); 871 872 if (vp == NULL) 873 return; 874 875 error = VOP_IOCTL(vp, DIOCGDINFO, &label, FREAD, NOCRED); 876 VOP_CLOSE(vp, FREAD, NOCRED); 877 vput(vp); 878 if (error) 879 return; 880 881 KASSERT(booted_partition >= 0 882 && booted_partition < MAXPARTITIONS); 883 884 p = &label.d_partitions[booted_partition]; 885 886 dev = booted_device; 887 startblk = p->p_offset; 888 nblks = p->p_size; 889 } 890 891 dev = dkwedge_find_partition(dev, startblk, nblks); 892 if (dev != NULL) { 893 booted_device = dev; 894 booted_partition = 0; 895 } 896 } 897 898 void 899 rootconf(void) 900 { 901 if (booted_device != NULL) 902 rootconf_handle_wedges(); 903 904 setroot(booted_device, booted_partition); 905 } 906 907 static void 908 check_console(struct lwp *l) 909 { 910 struct vnode *vp; 911 int error; 912 913 error = namei_simple_kernel("/dev/console", 914 NSM_FOLLOW_NOEMULROOT, &vp); 915 if (error == 0) { 916 vrele(vp); 917 } else if (error == ENOENT) { 918 if (boothowto & (AB_VERBOSE|AB_DEBUG)) 919 printf("warning: no /dev/console\n"); 920 } else { 921 printf("warning: lookup /dev/console: error %d\n", error); 922 } 923 } 924 925 /* 926 * List of paths to try when searching for "init". 927 */ 928 static const char * const initpaths[] = { 929 "/sbin/init", 930 "/sbin/oinit", 931 "/sbin/init.bak", 932 "/rescue/init", 933 NULL, 934 }; 935 936 /* 937 * Start the initial user process; try exec'ing each pathname in "initpaths". 938 * The program is invoked with one argument containing the boot flags. 939 */ 940 static void 941 start_init(void *arg) 942 { 943 struct lwp *l = arg; 944 struct proc *p = l->l_proc; 945 vaddr_t addr; 946 struct sys_execve_args /* { 947 syscallarg(const char *) path; 948 syscallarg(char * const *) argp; 949 syscallarg(char * const *) envp; 950 } */ args; 951 int options, i, error; 952 register_t retval[2]; 953 char flags[4], *flagsp; 954 const char *path, *slash; 955 char *ucp, **uap, *arg0, *arg1, *argv[3]; 956 char ipath[129]; 957 int ipx, len; 958 959 /* 960 * Now in process 1. 961 */ 962 strncpy(p->p_comm, "init", MAXCOMLEN); 963 964 /* 965 * Wait for main() to tell us that it's safe to exec. 966 */ 967 mutex_enter(&proc_lock); 968 while (start_init_exec == 0) 969 cv_wait(&lbolt, &proc_lock); 970 mutex_exit(&proc_lock); 971 972 /* 973 * This is not the right way to do this. We really should 974 * hand-craft a descriptor onto /dev/console to hand to init, 975 * but that's a _lot_ more work, and the benefit from this easy 976 * hack makes up for the "good is the enemy of the best" effect. 977 */ 978 check_console(l); 979 980 /* 981 * Need just enough stack to hold the faked-up "execve()" arguments. 982 */ 983 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 984 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 985 NULL, UVM_UNKNOWN_OFFSET, 0, 986 UVM_MAPFLAG(UVM_PROT_RW, UVM_PROT_RW, UVM_INH_COPY, 987 UVM_ADV_NORMAL, 988 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 989 panic("init: couldn't allocate argument space"); 990 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 991 992 ipx = 0; 993 while (1) { 994 if (boothowto & RB_ASKNAME) { 995 printf("init path"); 996 if (initpaths[ipx]) 997 printf(" (default %s)", initpaths[ipx]); 998 printf(": "); 999 len = cngetsn(ipath, sizeof(ipath)-1); 1000 if (len == 4 && strcmp(ipath, "halt") == 0) { 1001 kern_reboot(RB_HALT, NULL); 1002 } else if (len == 6 && strcmp(ipath, "reboot") == 0) { 1003 kern_reboot(0, NULL); 1004 #if defined(DDB) 1005 } else if (len == 3 && strcmp(ipath, "ddb") == 0) { 1006 console_debugger(); 1007 continue; 1008 #endif 1009 } else if (len > 0 && ipath[0] == '/') { 1010 ipath[len] = '\0'; 1011 path = ipath; 1012 } else if (len == 0 && initpaths[ipx] != NULL) { 1013 path = initpaths[ipx++]; 1014 } else { 1015 printf("use absolute path, "); 1016 #if defined(DDB) 1017 printf("\"ddb\", "); 1018 #endif 1019 printf("\"halt\", or \"reboot\"\n"); 1020 continue; 1021 } 1022 } else { 1023 if ((path = initpaths[ipx++]) == NULL) { 1024 ipx = 0; 1025 boothowto |= RB_ASKNAME; 1026 continue; 1027 } 1028 } 1029 1030 ucp = (char *)USRSTACK; 1031 1032 /* 1033 * Construct the boot flag argument. 1034 */ 1035 flagsp = flags; 1036 *flagsp++ = '-'; 1037 options = 0; 1038 1039 if (boothowto & RB_SINGLE) { 1040 *flagsp++ = 's'; 1041 options = 1; 1042 } 1043 #ifdef notyet 1044 if (boothowto & RB_FASTBOOT) { 1045 *flagsp++ = 'f'; 1046 options = 1; 1047 } 1048 #endif 1049 1050 /* 1051 * Move out the flags (arg 1), if necessary. 1052 */ 1053 if (options != 0) { 1054 *flagsp++ = '\0'; 1055 i = flagsp - flags; 1056 #ifdef DEBUG 1057 aprint_normal("init: copying out flags `%s' %d\n", flags, i); 1058 #endif 1059 arg1 = STACK_ALLOC(ucp, i); 1060 ucp = STACK_MAX(arg1, i); 1061 if ((error = copyout((void *)flags, arg1, i)) != 0) 1062 goto copyerr; 1063 } else 1064 arg1 = NULL; 1065 1066 /* 1067 * Move out the file name (also arg 0). 1068 */ 1069 i = strlen(path) + 1; 1070 #ifdef DEBUG 1071 aprint_normal("init: copying out path `%s' %d\n", path, i); 1072 #else 1073 if (boothowto & RB_ASKNAME || path != initpaths[0]) 1074 printf("init: trying %s\n", path); 1075 #endif 1076 arg0 = STACK_ALLOC(ucp, i); 1077 ucp = STACK_MAX(arg0, i); 1078 if ((error = copyout(path, arg0, i)) != 0) 1079 goto copyerr; 1080 1081 /* 1082 * Move out the arg pointers. 1083 */ 1084 ucp = (void *)STACK_ALIGN(ucp, STACK_ALIGNBYTES); 1085 uap = (char **)STACK_ALLOC(ucp, sizeof(argv)); 1086 SCARG(&args, path) = arg0; 1087 SCARG(&args, argp) = uap; 1088 SCARG(&args, envp) = NULL; 1089 slash = strrchr(path, '/'); 1090 1091 argv[0] = slash ? arg0 + (slash + 1 - path) : arg0; 1092 argv[1] = arg1; 1093 argv[2] = NULL; 1094 if ((error = copyout(argv, uap, sizeof(argv))) != 0) 1095 goto copyerr; 1096 1097 /* 1098 * Now try to exec the program. If it can't for any reason 1099 * other than it doesn't exist, complain. 1100 */ 1101 error = sys_execve(l, &args, retval); 1102 if (error == 0 || error == EJUSTRETURN) { 1103 KERNEL_UNLOCK_LAST(l); 1104 return; 1105 } 1106 printf("exec %s: error %d\n", path, error); 1107 } 1108 printf("init: not found\n"); 1109 panic("no init"); 1110 copyerr: 1111 panic("copyout %d", error); 1112 } 1113 1114 /* 1115 * calculate cache size (in bytes) from physmem and vsize. 1116 */ 1117 vaddr_t 1118 calc_cache_size(vsize_t vsize, int pct, int va_pct) 1119 { 1120 paddr_t t; 1121 1122 /* XXX should consider competing cache if any */ 1123 /* XXX should consider submaps */ 1124 t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; 1125 if (vsize != 0) { 1126 vsize = (uintmax_t)vsize * va_pct / 100; 1127 if (t > vsize) { 1128 t = vsize; 1129 } 1130 } 1131 return t; 1132 } 1133 1134 /* 1135 * Print the system start up banner. 1136 * 1137 * - Print a limited banner if AB_SILENT. 1138 * - Always send normal banner to the log. 1139 */ 1140 #define MEM_PBUFSIZE sizeof("99999 MB") 1141 1142 void 1143 banner(void) 1144 { 1145 static char notice[] = " Notice: this software is " 1146 "protected by copyright"; 1147 char pbuf[81]; 1148 void (*pr)(const char *, ...) __printflike(1, 2); 1149 int i; 1150 1151 if ((boothowto & AB_SILENT) != 0) { 1152 snprintf(pbuf, sizeof(pbuf), "%s %s (%s)", 1153 ostype, osrelease, kernel_ident); 1154 printf_nolog("%s", pbuf); 1155 for (i = 80 - strlen(pbuf) - sizeof(notice); i > 0; i--) 1156 printf(" "); 1157 printf_nolog("%s\n", notice); 1158 pr = aprint_normal; 1159 } else { 1160 pr = printf; 1161 } 1162 1163 memset(pbuf, 0, sizeof(pbuf)); 1164 (*pr)("%s%s", copyright, version); 1165 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)physmem)); 1166 (*pr)("total memory = %s\n", pbuf); 1167 format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)uvm_availmem(false))); 1168 (*pr)("avail memory = %s\n", pbuf); 1169 } 1170
Indexes created Sun Sep 21 20:09:37 GMT 2025