| 1 | /* $NetBSD: init_main.c,v 1.487 2016/11/16 12:31:33 pgoyette Exp $ */ |
| 2 | |
| 3 | /*- |
| 4 | * Copyright (c) 2008, 2009 The NetBSD Foundation, Inc. |
| 5 | * All rights reserved. |
| 6 | * |
| 7 | * Redistribution and use in source and binary forms, with or without |
| 8 | * modification, are permitted provided that the following conditions |
| 9 | * are met: |
| 10 | * 1. Redistributions of source code must retain the above copyright |
| 11 | * notice, this list of conditions and the following disclaimer. |
| 12 | * 2. Redistributions in binary form must reproduce the above copyright |
| 13 | * notice, this list of conditions and the following disclaimer in the |
| 14 | * documentation and/or other materials provided with the distribution. |
| 15 | * |
| 16 | * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS |
| 17 | * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED |
| 18 | * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR |
| 19 | * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS |
| 20 | * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR |
| 21 | * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF |
| 22 | * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS |
| 23 | * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN |
| 24 | * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) |
| 25 | * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE |
| 26 | * POSSIBILITY OF SUCH DAMAGE. |
| 27 | */ |
| 28 | |
| 29 | /* |
| 30 | * Copyright (c) 1982, 1986, 1989, 1991, 1992, 1993 |
| 31 | * The Regents of the University of California. All rights reserved. |
| 32 | * (c) UNIX System Laboratories, Inc. |
| 33 | * All or some portions of this file are derived from material licensed |
| 34 | * to the University of California by American Telephone and Telegraph |
| 35 | * Co. or Unix System Laboratories, Inc. and are reproduced herein with |
| 36 | * the permission of UNIX System Laboratories, Inc. |
| 37 | * |
| 38 | * Redistribution and use in source and binary forms, with or without |
| 39 | * modification, are permitted provided that the following conditions |
| 40 | * are met: |
| 41 | * 1. Redistributions of source code must retain the above copyright |
| 42 | * notice, this list of conditions and the following disclaimer. |
| 43 | * 2. Redistributions in binary form must reproduce the above copyright |
| 44 | * notice, this list of conditions and the following disclaimer in the |
| 45 | * documentation and/or other materials provided with the distribution. |
| 46 | * 3. Neither the name of the University nor the names of its contributors |
| 47 | * may be used to endorse or promote products derived from this software |
| 48 | * without specific prior written permission. |
| 49 | * |
| 50 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
| 51 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 52 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 53 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
| 54 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
| 55 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
| 56 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
| 57 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
| 58 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
| 59 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 60 | * SUCH DAMAGE. |
| 61 | * |
| 62 | * @(#)init_main.c 8.16 (Berkeley) 5/14/95 |
| 63 | */ |
| 64 | |
| 65 | /* |
| 66 | * Copyright (c) 1995 Christopher G. Demetriou. All rights reserved. |
| 67 | * |
| 68 | * Redistribution and use in source and binary forms, with or without |
| 69 | * modification, are permitted provided that the following conditions |
| 70 | * are met: |
| 71 | * 1. Redistributions of source code must retain the above copyright |
| 72 | * notice, this list of conditions and the following disclaimer. |
| 73 | * 2. Redistributions in binary form must reproduce the above copyright |
| 74 | * notice, this list of conditions and the following disclaimer in the |
| 75 | * documentation and/or other materials provided with the distribution. |
| 76 | * 3. All advertising materials mentioning features or use of this software |
| 77 | * must display the following acknowledgement: |
| 78 | * This product includes software developed by the University of |
| 79 | * California, Berkeley and its contributors. |
| 80 | * 4. Neither the name of the University nor the names of its contributors |
| 81 | * may be used to endorse or promote products derived from this software |
| 82 | * without specific prior written permission. |
| 83 | * |
| 84 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
| 85 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 86 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 87 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
| 88 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
| 89 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
| 90 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
| 91 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
| 92 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
| 93 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 94 | * SUCH DAMAGE. |
| 95 | * |
| 96 | * @(#)init_main.c 8.16 (Berkeley) 5/14/95 |
| 97 | */ |
| 98 | |
| 99 | #include <sys/cdefs.h> |
| 100 | __KERNEL_RCSID(0, "$NetBSD: init_main.c,v 1.487 2016/11/16 12:31:33 pgoyette Exp $" ); |
| 101 | |
| 102 | #include "opt_ddb.h" |
| 103 | #include "opt_inet.h" |
| 104 | #include "opt_ipsec.h" |
| 105 | #include "opt_modular.h" |
| 106 | #include "opt_ntp.h" |
| 107 | #include "opt_pipe.h" |
| 108 | #include "opt_syscall_debug.h" |
| 109 | #include "opt_sysv.h" |
| 110 | #include "opt_fileassoc.h" |
| 111 | #include "opt_ktrace.h" |
| 112 | #include "opt_pax.h" |
| 113 | #include "opt_compat_netbsd.h" |
| 114 | #include "opt_wapbl.h" |
| 115 | #include "opt_ptrace.h" |
| 116 | #include "opt_rnd_printf.h" |
| 117 | #include "opt_splash.h" |
| 118 | |
| 119 | #if defined(SPLASHSCREEN) && defined(makeoptions_SPLASHSCREEN_IMAGE) |
| 120 | extern void *_binary_splash_image_start; |
| 121 | extern void *_binary_splash_image_end; |
| 122 | #endif |
| 123 | |
| 124 | #include "ksyms.h" |
| 125 | |
| 126 | #include "veriexec.h" |
| 127 | |
| 128 | #include <sys/param.h> |
| 129 | #include <sys/acct.h> |
| 130 | #include <sys/filedesc.h> |
| 131 | #include <sys/file.h> |
| 132 | #include <sys/errno.h> |
| 133 | #include <sys/callout.h> |
| 134 | #include <sys/cpu.h> |
| 135 | #include <sys/cpufreq.h> |
| 136 | #include <sys/spldebug.h> |
| 137 | #include <sys/kernel.h> |
| 138 | #include <sys/mount.h> |
| 139 | #include <sys/proc.h> |
| 140 | #include <sys/kthread.h> |
| 141 | #include <sys/resourcevar.h> |
| 142 | #include <sys/signalvar.h> |
| 143 | #include <sys/systm.h> |
| 144 | #include <sys/vnode.h> |
| 145 | #include <sys/fstrans.h> |
| 146 | #include <sys/tty.h> |
| 147 | #include <sys/conf.h> |
| 148 | #include <sys/disklabel.h> |
| 149 | #include <sys/buf.h> |
| 150 | #include <sys/device.h> |
| 151 | #include <sys/exec.h> |
| 152 | #include <sys/socketvar.h> |
| 153 | #include <sys/protosw.h> |
| 154 | #include <sys/percpu.h> |
| 155 | #include <sys/pserialize.h> |
| 156 | #include <sys/pset.h> |
| 157 | #include <sys/sysctl.h> |
| 158 | #include <sys/reboot.h> |
| 159 | #include <sys/event.h> |
| 160 | #include <sys/mbuf.h> |
| 161 | #include <sys/sched.h> |
| 162 | #include <sys/sleepq.h> |
| 163 | #include <sys/ipi.h> |
| 164 | #include <sys/iostat.h> |
| 165 | #include <sys/vmem.h> |
| 166 | #include <sys/uuid.h> |
| 167 | #include <sys/extent.h> |
| 168 | #include <sys/disk.h> |
| 169 | #include <sys/msgbuf.h> |
| 170 | #include <sys/module.h> |
| 171 | #include <sys/event.h> |
| 172 | #include <sys/lockf.h> |
| 173 | #include <sys/once.h> |
| 174 | #include <sys/kcpuset.h> |
| 175 | #include <sys/ksyms.h> |
| 176 | #include <sys/uidinfo.h> |
| 177 | #include <sys/kprintf.h> |
| 178 | #include <sys/bufq.h> |
| 179 | #ifdef IPSEC |
| 180 | #include <netipsec/ipsec.h> |
| 181 | #endif |
| 182 | #include <sys/domain.h> |
| 183 | #include <sys/namei.h> |
| 184 | #include <sys/rnd.h> |
| 185 | #include <sys/pipe.h> |
| 186 | #if NVERIEXEC > 0 |
| 187 | #include <sys/verified_exec.h> |
| 188 | #endif /* NVERIEXEC > 0 */ |
| 189 | #ifdef KTRACE |
| 190 | #include <sys/ktrace.h> |
| 191 | #endif |
| 192 | #include <sys/kauth.h> |
| 193 | #include <net80211/ieee80211_netbsd.h> |
| 194 | #include <sys/cprng.h> |
| 195 | |
| 196 | #include <sys/syscall.h> |
| 197 | #include <sys/syscallargs.h> |
| 198 | |
| 199 | #include <sys/pax.h> |
| 200 | |
| 201 | #include <secmodel/secmodel.h> |
| 202 | |
| 203 | #include <ufs/ufs/quota.h> |
| 204 | |
| 205 | #include <miscfs/genfs/genfs.h> |
| 206 | #include <miscfs/specfs/specdev.h> |
| 207 | |
| 208 | #include <sys/cpu.h> |
| 209 | |
| 210 | #include <uvm/uvm.h> /* extern struct uvm uvm */ |
| 211 | |
| 212 | #include <dev/cons.h> |
| 213 | #include <dev/splash/splash.h> |
| 214 | |
| 215 | #include <net/bpf.h> |
| 216 | #include <net/if.h> |
| 217 | #include <net/raw_cb.h> |
| 218 | #include <net/if_llatbl.h> |
| 219 | |
| 220 | #include <prop/proplib.h> |
| 221 | |
| 222 | #include <sys/userconf.h> |
| 223 | |
| 224 | extern struct lwp lwp0; |
| 225 | extern time_t rootfstime; |
| 226 | |
| 227 | #ifndef curlwp |
| 228 | struct lwp *curlwp = &lwp0; |
| 229 | #endif |
| 230 | struct proc *initproc; |
| 231 | |
| 232 | struct vnode *rootvp, *swapdev_vp; |
| 233 | int boothowto; |
| 234 | int cold = 1; /* still working on startup */ |
| 235 | struct timespec boottime; /* time at system startup - will only follow settime deltas */ |
| 236 | |
| 237 | int start_init_exec; /* semaphore for start_init() */ |
| 238 | |
| 239 | cprng_strong_t *kern_cprng; |
| 240 | |
| 241 | static void check_console(struct lwp *l); |
| 242 | static void start_init(void *); |
| 243 | static void configure(void); |
| 244 | static void configure2(void); |
| 245 | static void configure3(void); |
| 246 | void main(void); |
| 247 | |
| 248 | /* |
| 249 | * System startup; initialize the world, create process 0, mount root |
| 250 | * filesystem, and fork to create init and pagedaemon. Most of the |
| 251 | * hard work is done in the lower-level initialization routines including |
| 252 | * startup(), which does memory initialization and autoconfiguration. |
| 253 | */ |
| 254 | void |
| 255 | main(void) |
| 256 | { |
| 257 | struct timespec time; |
| 258 | struct lwp *l; |
| 259 | struct proc *p; |
| 260 | int s, error; |
| 261 | #ifdef NVNODE_IMPLICIT |
| 262 | int usevnodes; |
| 263 | #endif |
| 264 | CPU_INFO_ITERATOR cii; |
| 265 | struct cpu_info *ci; |
| 266 | |
| 267 | l = &lwp0; |
| 268 | #ifndef LWP0_CPU_INFO |
| 269 | l->l_cpu = curcpu(); |
| 270 | #endif |
| 271 | l->l_pflag |= LP_RUNNING; |
| 272 | |
| 273 | /* |
| 274 | * Attempt to find console and initialize |
| 275 | * in case of early panic or other messages. |
| 276 | */ |
| 277 | consinit(); |
| 278 | |
| 279 | kernel_lock_init(); |
| 280 | once_init(); |
| 281 | |
| 282 | mi_cpu_init(); |
| 283 | kernconfig_lock_init(); |
| 284 | kthread_sysinit(); |
| 285 | |
| 286 | /* Initialize the device switch tables. */ |
| 287 | devsw_init(); |
| 288 | |
| 289 | /* Initialize event counters. */ |
| 290 | evcnt_init(); |
| 291 | |
| 292 | uvm_init(); |
| 293 | ubchist_init(); |
| 294 | kcpuset_sysinit(); |
| 295 | |
| 296 | prop_kern_init(); |
| 297 | |
| 298 | #if ((NKSYMS > 0) || (NDDB > 0) || (NMODULAR > 0)) |
| 299 | ksyms_init(); |
| 300 | #endif |
| 301 | kprintf_init(); |
| 302 | |
| 303 | percpu_init(); |
| 304 | |
| 305 | /* Initialize lock caches. */ |
| 306 | mutex_obj_init(); |
| 307 | rw_obj_init(); |
| 308 | |
| 309 | /* Passive serialization. */ |
| 310 | pserialize_init(); |
| 311 | |
| 312 | /* Initialize the extent manager. */ |
| 313 | extent_init(); |
| 314 | |
| 315 | /* Do machine-dependent initialization. */ |
| 316 | cpu_startup(); |
| 317 | |
| 318 | /* Initialize the sysctl subsystem. */ |
| 319 | sysctl_init(); |
| 320 | |
| 321 | /* Initialize callouts, part 1. */ |
| 322 | callout_startup(); |
| 323 | |
| 324 | /* Initialize the kernel authorization subsystem. */ |
| 325 | kauth_init(); |
| 326 | |
| 327 | secmodel_init(); |
| 328 | |
| 329 | spec_init(); |
| 330 | |
| 331 | /* |
| 332 | * Set BPF op vector. Can't do this in bpf attach, since |
| 333 | * network drivers attach before bpf. |
| 334 | */ |
| 335 | bpf_setops(); |
| 336 | |
| 337 | /* Start module system. */ |
| 338 | module_init(); |
| 339 | |
| 340 | /* |
| 341 | * Initialize the kernel authorization subsystem and start the |
| 342 | * default security model, if any. We need to do this early |
| 343 | * enough so that subsystems relying on any of the aforementioned |
| 344 | * can work properly. Since the security model may dictate the |
| 345 | * credential inheritance policy, it is needed at least before |
| 346 | * any process is created, specifically proc0. |
| 347 | */ |
| 348 | module_init_class(MODULE_CLASS_SECMODEL); |
| 349 | |
| 350 | /* Initialize the buffer cache */ |
| 351 | bufinit(); |
| 352 | |
| 353 | |
| 354 | #if defined(SPLASHSCREEN) && defined(SPLASHSCREEN_IMAGE) |
| 355 | size_t splash_size = (&_binary_splash_image_end - |
| 356 | &_binary_splash_image_start) * sizeof(void *); |
| 357 | splash_setimage(&_binary_splash_image_start, splash_size); |
| 358 | #endif |
| 359 | |
| 360 | /* Initialize sockets. */ |
| 361 | soinit(); |
| 362 | |
| 363 | /* |
| 364 | * The following things must be done before autoconfiguration. |
| 365 | */ |
| 366 | rnd_init(); /* initialize entropy pool */ |
| 367 | |
| 368 | cprng_init(); /* initialize cryptographic PRNG */ |
| 369 | |
| 370 | /* Initialize process and pgrp structures. */ |
| 371 | procinit(); |
| 372 | lwpinit(); |
| 373 | |
| 374 | /* Initialize signal-related data structures. */ |
| 375 | signal_init(); |
| 376 | |
| 377 | /* Initialize resource management. */ |
| 378 | resource_init(); |
| 379 | |
| 380 | /* Create process 0. */ |
| 381 | proc0_init(); |
| 382 | lwp0_init(); |
| 383 | |
| 384 | /* Disable preemption during boot. */ |
| 385 | kpreempt_disable(); |
| 386 | |
| 387 | /* Initialize the UID hash table. */ |
| 388 | uid_init(); |
| 389 | |
| 390 | /* Charge root for one process. */ |
| 391 | (void)chgproccnt(0, 1); |
| 392 | |
| 393 | /* Initialize timekeeping. */ |
| 394 | time_init(); |
| 395 | |
| 396 | /* Initialize the run queues, turnstiles and sleep queues. */ |
| 397 | sched_rqinit(); |
| 398 | turnstile_init(); |
| 399 | sleeptab_init(&sleeptab); |
| 400 | |
| 401 | sched_init(); |
| 402 | |
| 403 | /* Initialize processor-sets */ |
| 404 | psets_init(); |
| 405 | |
| 406 | /* Initialize cpufreq(9) */ |
| 407 | cpufreq_init(); |
| 408 | |
| 409 | /* MI initialization of the boot cpu */ |
| 410 | error = mi_cpu_attach(curcpu()); |
| 411 | KASSERT(error == 0); |
| 412 | |
| 413 | /* Initialize timekeeping, part 2. */ |
| 414 | time_init2(); |
| 415 | |
| 416 | /* |
| 417 | * Initialize mbuf's. Do this now because we might attempt to |
| 418 | * allocate mbufs or mbuf clusters during autoconfiguration. |
| 419 | */ |
| 420 | mbinit(); |
| 421 | |
| 422 | /* Initialize I/O statistics. */ |
| 423 | iostat_init(); |
| 424 | |
| 425 | /* Initialize the log device. */ |
| 426 | loginit(); |
| 427 | |
| 428 | /* Second part of module system initialization. */ |
| 429 | module_start_unload_thread(); |
| 430 | |
| 431 | /* Initialize the file systems. */ |
| 432 | #ifdef NVNODE_IMPLICIT |
| 433 | /* |
| 434 | * If maximum number of vnodes in namei vnode cache is not explicitly |
| 435 | * defined in kernel config, adjust the number such as we use roughly |
| 436 | * 10% of memory for vnodes and associated data structures in the |
| 437 | * assumed worst case. Do not provide fewer than NVNODE vnodes. |
| 438 | */ |
| 439 | usevnodes = calc_cache_size(vmem_size(kmem_arena, VMEM_FREE|VMEM_ALLOC), |
| 440 | 10, VNODE_KMEM_MAXPCT) / VNODE_COST; |
| 441 | if (usevnodes > desiredvnodes) |
| 442 | desiredvnodes = usevnodes; |
| 443 | #endif |
| 444 | vfsinit(); |
| 445 | lf_init(); |
| 446 | |
| 447 | /* Initialize fstrans. */ |
| 448 | fstrans_init(); |
| 449 | |
| 450 | /* Initialize the file descriptor system. */ |
| 451 | fd_sys_init(); |
| 452 | |
| 453 | /* Initialize cwd structures */ |
| 454 | cwd_sys_init(); |
| 455 | |
| 456 | /* Initialize kqueue. */ |
| 457 | kqueue_init(); |
| 458 | |
| 459 | inittimecounter(); |
| 460 | ntp_init(); |
| 461 | |
| 462 | /* Initialize tty subsystem. */ |
| 463 | tty_init(); |
| 464 | ttyldisc_init(); |
| 465 | |
| 466 | /* Initialize the buffer cache, part 2. */ |
| 467 | bufinit2(); |
| 468 | |
| 469 | /* Initialize the disk wedge subsystem. */ |
| 470 | dkwedge_init(); |
| 471 | |
| 472 | /* Initialize the kernel strong PRNG. */ |
| 473 | kern_cprng = cprng_strong_create("kernel" , IPL_VM, |
| 474 | CPRNG_INIT_ANY|CPRNG_REKEY_ANY); |
| 475 | |
| 476 | /* Initialize interfaces. */ |
| 477 | ifinit1(); |
| 478 | |
| 479 | spldebug_start(); |
| 480 | |
| 481 | /* Initialize sockets thread(s) */ |
| 482 | soinit1(); |
| 483 | |
| 484 | /* |
| 485 | * Initialize the bufq strategy sub-system and any built-in |
| 486 | * strategy modules - they may be needed by some devices during |
| 487 | * auto-configuration |
| 488 | */ |
| 489 | bufq_init(); |
| 490 | module_init_class(MODULE_CLASS_BUFQ); |
| 491 | |
| 492 | /* Configure the system hardware. This will enable interrupts. */ |
| 493 | configure(); |
| 494 | |
| 495 | /* Once all CPUs are detected, initialize the per-CPU cprng_fast. */ |
| 496 | cprng_fast_init(); |
| 497 | |
| 498 | ssp_init(); |
| 499 | |
| 500 | ubc_init(); /* must be after autoconfig */ |
| 501 | |
| 502 | mm_init(); |
| 503 | |
| 504 | configure2(); |
| 505 | |
| 506 | ipi_sysinit(); |
| 507 | |
| 508 | /* Now timer is working. Enable preemption. */ |
| 509 | kpreempt_enable(); |
| 510 | |
| 511 | /* Get the threads going and into any sleeps before continuing. */ |
| 512 | yield(); |
| 513 | |
| 514 | /* Enable deferred processing of RNG samples */ |
| 515 | rnd_init_softint(); |
| 516 | |
| 517 | #ifdef RND_PRINTF |
| 518 | /* Enable periodic injection of console output into entropy pool */ |
| 519 | kprintf_init_callout(); |
| 520 | #endif |
| 521 | |
| 522 | vmem_rehash_start(); /* must be before exec_init */ |
| 523 | |
| 524 | /* Initialize exec structures */ |
| 525 | exec_init(1); /* seminit calls exithook_establish() */ |
| 526 | |
| 527 | #if NVERIEXEC > 0 |
| 528 | /* |
| 529 | * Initialise the Veriexec subsystem. |
| 530 | */ |
| 531 | veriexec_init(); |
| 532 | #endif /* NVERIEXEC > 0 */ |
| 533 | |
| 534 | pax_init(); |
| 535 | |
| 536 | #ifdef IPSEC |
| 537 | /* Attach network crypto subsystem */ |
| 538 | ipsec_attach(); |
| 539 | #endif |
| 540 | |
| 541 | /* |
| 542 | * Initialize protocols. Block reception of incoming packets |
| 543 | * until everything is ready. |
| 544 | */ |
| 545 | s = splnet(); |
| 546 | ifinit(); |
| 547 | #if defined(INET) || defined(INET6) |
| 548 | lltableinit(); |
| 549 | #endif |
| 550 | domaininit(true); |
| 551 | if_attachdomain(); |
| 552 | splx(s); |
| 553 | |
| 554 | #ifdef GPROF |
| 555 | /* Initialize kernel profiling. */ |
| 556 | kmstartup(); |
| 557 | #endif |
| 558 | |
| 559 | /* Initialize system accounting. */ |
| 560 | acct_init(); |
| 561 | |
| 562 | #ifndef PIPE_SOCKETPAIR |
| 563 | /* Initialize pipes. */ |
| 564 | pipe_init(); |
| 565 | #endif |
| 566 | |
| 567 | #ifdef KTRACE |
| 568 | /* Initialize ktrace. */ |
| 569 | ktrinit(); |
| 570 | #endif |
| 571 | |
| 572 | machdep_init(); |
| 573 | |
| 574 | procinit_sysctl(); |
| 575 | |
| 576 | scdebug_init(); |
| 577 | |
| 578 | /* |
| 579 | * Create process 1 (init(8)). We do this now, as Unix has |
| 580 | * historically had init be process 1, and changing this would |
| 581 | * probably upset a lot of people. |
| 582 | * |
| 583 | * Note that process 1 won't immediately exec init(8), but will |
| 584 | * wait for us to inform it that the root file system has been |
| 585 | * mounted. |
| 586 | */ |
| 587 | if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) |
| 588 | panic("fork init" ); |
| 589 | |
| 590 | /* |
| 591 | * Load any remaining builtin modules, and hand back temporary |
| 592 | * storage to the VM system. Then require force when loading any |
| 593 | * remaining un-init'ed built-in modules to avoid later surprises. |
| 594 | */ |
| 595 | module_init_class(MODULE_CLASS_ANY); |
| 596 | module_builtin_require_force(); |
| 597 | |
| 598 | /* |
| 599 | * Finalize configuration now that all real devices have been |
| 600 | * found. This needs to be done before the root device is |
| 601 | * selected, since finalization may create the root device. |
| 602 | */ |
| 603 | config_finalize(); |
| 604 | |
| 605 | sysctl_finalize(); |
| 606 | |
| 607 | /* |
| 608 | * Now that autoconfiguration has completed, we can determine |
| 609 | * the root and dump devices. |
| 610 | */ |
| 611 | cpu_rootconf(); |
| 612 | cpu_dumpconf(); |
| 613 | |
| 614 | /* Mount the root file system. */ |
| 615 | do { |
| 616 | domountroothook(root_device); |
| 617 | if ((error = vfs_mountroot())) { |
| 618 | printf("cannot mount root, error = %d\n" , error); |
| 619 | boothowto |= RB_ASKNAME; |
| 620 | setroot(root_device, |
| 621 | (rootdev != NODEV) ? DISKPART(rootdev) : 0); |
| 622 | } |
| 623 | } while (error != 0); |
| 624 | mountroothook_destroy(); |
| 625 | |
| 626 | configure3(); |
| 627 | |
| 628 | /* |
| 629 | * Initialise the time-of-day clock, passing the time recorded |
| 630 | * in the root filesystem (if any) for use by systems that |
| 631 | * don't have a non-volatile time-of-day device. |
| 632 | */ |
| 633 | inittodr(rootfstime); |
| 634 | |
| 635 | /* |
| 636 | * Now can look at time, having had a chance to verify the time |
| 637 | * from the file system. Reset l->l_rtime as it may have been |
| 638 | * munched in mi_switch() after the time got set. |
| 639 | */ |
| 640 | getnanotime(&time); |
| 641 | boottime = time; |
| 642 | |
| 643 | mutex_enter(proc_lock); |
| 644 | LIST_FOREACH(p, &allproc, p_list) { |
| 645 | KASSERT((p->p_flag & PK_MARKER) == 0); |
| 646 | mutex_enter(p->p_lock); |
| 647 | TIMESPEC_TO_TIMEVAL(&p->p_stats->p_start, &time); |
| 648 | LIST_FOREACH(l, &p->p_lwps, l_sibling) { |
| 649 | lwp_lock(l); |
| 650 | memset(&l->l_rtime, 0, sizeof(l->l_rtime)); |
| 651 | lwp_unlock(l); |
| 652 | } |
| 653 | mutex_exit(p->p_lock); |
| 654 | } |
| 655 | mutex_exit(proc_lock); |
| 656 | binuptime(&curlwp->l_stime); |
| 657 | |
| 658 | for (CPU_INFO_FOREACH(cii, ci)) { |
| 659 | ci->ci_schedstate.spc_lastmod = time_second; |
| 660 | } |
| 661 | |
| 662 | /* Create the pageout daemon kernel thread. */ |
| 663 | uvm_swap_init(); |
| 664 | if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, |
| 665 | NULL, NULL, "pgdaemon" )) |
| 666 | panic("fork pagedaemon" ); |
| 667 | |
| 668 | /* Create the filesystem syncer kernel thread. */ |
| 669 | if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, |
| 670 | NULL, NULL, "ioflush" )) |
| 671 | panic("fork syncer" ); |
| 672 | |
| 673 | /* Create the aiodone daemon kernel thread. */ |
| 674 | if (workqueue_create(&uvm.aiodone_queue, "aiodoned" , |
| 675 | uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE)) |
| 676 | panic("fork aiodoned" ); |
| 677 | |
| 678 | /* Wait for final configure threads to complete. */ |
| 679 | config_finalize_mountroot(); |
| 680 | |
| 681 | /* |
| 682 | * Okay, now we can let init(8) exec! It's off to userland! |
| 683 | */ |
| 684 | mutex_enter(proc_lock); |
| 685 | start_init_exec = 1; |
| 686 | cv_broadcast(&lbolt); |
| 687 | mutex_exit(proc_lock); |
| 688 | |
| 689 | /* The scheduler is an infinite loop. */ |
| 690 | uvm_scheduler(); |
| 691 | /* NOTREACHED */ |
| 692 | } |
| 693 | |
| 694 | /* |
| 695 | * Configure the system's hardware. |
| 696 | */ |
| 697 | static void |
| 698 | configure(void) |
| 699 | { |
| 700 | |
| 701 | /* Initialize autoconf data structures. */ |
| 702 | config_init_mi(); |
| 703 | /* |
| 704 | * XXX |
| 705 | * callout_setfunc() requires mutex(9) so it can't be in config_init() |
| 706 | * on amiga and atari which use config_init() and autoconf(9) fucntions |
| 707 | * to initialize console devices. |
| 708 | */ |
| 709 | config_twiddle_init(); |
| 710 | |
| 711 | pmf_init(); |
| 712 | |
| 713 | /* Initialize driver modules */ |
| 714 | module_init_class(MODULE_CLASS_DRIVER); |
| 715 | |
| 716 | userconf_init(); |
| 717 | if (boothowto & RB_USERCONF) |
| 718 | userconf_prompt(); |
| 719 | |
| 720 | if ((boothowto & (AB_SILENT|AB_VERBOSE)) == AB_SILENT) { |
| 721 | printf_nolog("Detecting hardware..." ); |
| 722 | } |
| 723 | |
| 724 | /* |
| 725 | * Do the machine-dependent portion of autoconfiguration. This |
| 726 | * sets the configuration machinery here in motion by "finding" |
| 727 | * the root bus. When this function returns, we expect interrupts |
| 728 | * to be enabled. |
| 729 | */ |
| 730 | cpu_configure(); |
| 731 | } |
| 732 | |
| 733 | static void |
| 734 | configure2(void) |
| 735 | { |
| 736 | CPU_INFO_ITERATOR cii; |
| 737 | struct cpu_info *ci; |
| 738 | int s; |
| 739 | |
| 740 | /* |
| 741 | * Now that we've found all the hardware, start the real time |
| 742 | * and statistics clocks. |
| 743 | */ |
| 744 | initclocks(); |
| 745 | |
| 746 | cold = 0; /* clocks are running, we're warm now! */ |
| 747 | s = splsched(); |
| 748 | curcpu()->ci_schedstate.spc_flags |= SPCF_RUNNING; |
| 749 | splx(s); |
| 750 | |
| 751 | /* Boot the secondary processors. */ |
| 752 | for (CPU_INFO_FOREACH(cii, ci)) { |
| 753 | uvm_cpu_attach(ci); |
| 754 | } |
| 755 | mp_online = true; |
| 756 | #if defined(MULTIPROCESSOR) |
| 757 | cpu_boot_secondary_processors(); |
| 758 | #endif |
| 759 | |
| 760 | /* Setup the runqueues and scheduler. */ |
| 761 | runq_init(); |
| 762 | synch_init(); |
| 763 | |
| 764 | /* |
| 765 | * Bus scans can make it appear as if the system has paused, so |
| 766 | * twiddle constantly while config_interrupts() jobs are running. |
| 767 | */ |
| 768 | config_twiddle_fn(NULL); |
| 769 | |
| 770 | /* |
| 771 | * Create threads to call back and finish configuration for |
| 772 | * devices that want interrupts enabled. |
| 773 | */ |
| 774 | config_create_interruptthreads(); |
| 775 | } |
| 776 | |
| 777 | static void |
| 778 | configure3(void) |
| 779 | { |
| 780 | |
| 781 | /* |
| 782 | * Create threads to call back and finish configuration for |
| 783 | * devices that want the mounted root file system. |
| 784 | */ |
| 785 | config_create_mountrootthreads(); |
| 786 | |
| 787 | /* Get the threads going and into any sleeps before continuing. */ |
| 788 | yield(); |
| 789 | } |
| 790 | |
| 791 | static void |
| 792 | rootconf_handle_wedges(void) |
| 793 | { |
| 794 | struct disklabel label; |
| 795 | struct partition *p; |
| 796 | struct vnode *vp; |
| 797 | daddr_t startblk; |
| 798 | uint64_t nblks; |
| 799 | device_t dev; |
| 800 | int error; |
| 801 | |
| 802 | if (booted_nblks) { |
| 803 | /* |
| 804 | * bootloader passed geometry |
| 805 | */ |
| 806 | dev = booted_device; |
| 807 | startblk = booted_startblk; |
| 808 | nblks = booted_nblks; |
| 809 | |
| 810 | /* |
| 811 | * keep booted_device and booted_partition |
| 812 | * in case the kernel doesn't identify a wedge |
| 813 | */ |
| 814 | } else { |
| 815 | /* |
| 816 | * bootloader passed partition number |
| 817 | * |
| 818 | * We cannot ask the partition device directly when it is |
| 819 | * covered by a wedge. Instead we look up the geometry in |
| 820 | * the disklabel. |
| 821 | */ |
| 822 | vp = opendisk(booted_device); |
| 823 | |
| 824 | if (vp == NULL) |
| 825 | return; |
| 826 | |
| 827 | error = VOP_IOCTL(vp, DIOCGDINFO, &label, FREAD, NOCRED); |
| 828 | VOP_CLOSE(vp, FREAD, NOCRED); |
| 829 | vput(vp); |
| 830 | if (error) |
| 831 | return; |
| 832 | |
| 833 | KASSERT(booted_partition >= 0 |
| 834 | && booted_partition < MAXPARTITIONS); |
| 835 | |
| 836 | p = &label.d_partitions[booted_partition]; |
| 837 | |
| 838 | dev = booted_device; |
| 839 | startblk = p->p_offset; |
| 840 | nblks = p->p_size; |
| 841 | } |
| 842 | |
| 843 | dev = dkwedge_find_partition(dev, startblk, nblks); |
| 844 | if (dev != NULL) { |
| 845 | booted_device = dev; |
| 846 | booted_partition = 0; |
| 847 | } |
| 848 | } |
| 849 | |
| 850 | void |
| 851 | rootconf(void) |
| 852 | { |
| 853 | if (booted_device != NULL) |
| 854 | rootconf_handle_wedges(); |
| 855 | |
| 856 | setroot(booted_device, booted_partition); |
| 857 | } |
| 858 | |
| 859 | static void |
| 860 | check_console(struct lwp *l) |
| 861 | { |
| 862 | struct vnode *vp; |
| 863 | int error; |
| 864 | |
| 865 | error = namei_simple_kernel("/dev/console" , |
| 866 | NSM_FOLLOW_NOEMULROOT, &vp); |
| 867 | if (error == 0) |
| 868 | vrele(vp); |
| 869 | else if (error == ENOENT) |
| 870 | printf("warning: no /dev/console\n" ); |
| 871 | else |
| 872 | printf("warning: lookup /dev/console: error %d\n" , error); |
| 873 | } |
| 874 | |
| 875 | /* |
| 876 | * List of paths to try when searching for "init". |
| 877 | */ |
| 878 | static const char * const initpaths[] = { |
| 879 | "/sbin/init" , |
| 880 | "/sbin/oinit" , |
| 881 | "/sbin/init.bak" , |
| 882 | "/rescue/init" , |
| 883 | NULL, |
| 884 | }; |
| 885 | |
| 886 | /* |
| 887 | * Start the initial user process; try exec'ing each pathname in "initpaths". |
| 888 | * The program is invoked with one argument containing the boot flags. |
| 889 | */ |
| 890 | static void |
| 891 | start_init(void *arg) |
| 892 | { |
| 893 | struct lwp *l = arg; |
| 894 | struct proc *p = l->l_proc; |
| 895 | vaddr_t addr; |
| 896 | struct sys_execve_args /* { |
| 897 | syscallarg(const char *) path; |
| 898 | syscallarg(char * const *) argp; |
| 899 | syscallarg(char * const *) envp; |
| 900 | } */ args; |
| 901 | int options, i, error; |
| 902 | register_t retval[2]; |
| 903 | char flags[4], *flagsp; |
| 904 | const char *path, *slash; |
| 905 | char *ucp, **uap, *arg0, *arg1, *argv[3]; |
| 906 | char ipath[129]; |
| 907 | int ipx, len; |
| 908 | |
| 909 | /* |
| 910 | * Now in process 1. |
| 911 | */ |
| 912 | strncpy(p->p_comm, "init" , MAXCOMLEN); |
| 913 | |
| 914 | /* |
| 915 | * Wait for main() to tell us that it's safe to exec. |
| 916 | */ |
| 917 | mutex_enter(proc_lock); |
| 918 | while (start_init_exec == 0) |
| 919 | cv_wait(&lbolt, proc_lock); |
| 920 | mutex_exit(proc_lock); |
| 921 | |
| 922 | /* |
| 923 | * This is not the right way to do this. We really should |
| 924 | * hand-craft a descriptor onto /dev/console to hand to init, |
| 925 | * but that's a _lot_ more work, and the benefit from this easy |
| 926 | * hack makes up for the "good is the enemy of the best" effect. |
| 927 | */ |
| 928 | check_console(l); |
| 929 | |
| 930 | /* |
| 931 | * Need just enough stack to hold the faked-up "execve()" arguments. |
| 932 | */ |
| 933 | addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); |
| 934 | if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, |
| 935 | NULL, UVM_UNKNOWN_OFFSET, 0, |
| 936 | UVM_MAPFLAG(UVM_PROT_RW, UVM_PROT_RW, UVM_INH_COPY, |
| 937 | UVM_ADV_NORMAL, |
| 938 | UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) |
| 939 | panic("init: couldn't allocate argument space" ); |
| 940 | p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); |
| 941 | |
| 942 | ipx = 0; |
| 943 | while (1) { |
| 944 | if (boothowto & RB_ASKNAME) { |
| 945 | printf("init path" ); |
| 946 | if (initpaths[ipx]) |
| 947 | printf(" (default %s)" , initpaths[ipx]); |
| 948 | printf(": " ); |
| 949 | len = cngetsn(ipath, sizeof(ipath)-1); |
| 950 | if (len == 4 && strcmp(ipath, "halt" ) == 0) { |
| 951 | cpu_reboot(RB_HALT, NULL); |
| 952 | } else if (len == 6 && strcmp(ipath, "reboot" ) == 0) { |
| 953 | cpu_reboot(0, NULL); |
| 954 | #if defined(DDB) |
| 955 | } else if (len == 3 && strcmp(ipath, "ddb" ) == 0) { |
| 956 | console_debugger(); |
| 957 | continue; |
| 958 | #endif |
| 959 | } else if (len > 0 && ipath[0] == '/') { |
| 960 | ipath[len] = '\0'; |
| 961 | path = ipath; |
| 962 | } else if (len == 0 && initpaths[ipx] != NULL) { |
| 963 | path = initpaths[ipx++]; |
| 964 | } else { |
| 965 | printf("use absolute path, " ); |
| 966 | #if defined(DDB) |
| 967 | printf("\"ddb\", " ); |
| 968 | #endif |
| 969 | printf("\"halt\", or \"reboot\"\n" ); |
| 970 | continue; |
| 971 | } |
| 972 | } else { |
| 973 | if ((path = initpaths[ipx++]) == NULL) { |
| 974 | ipx = 0; |
| 975 | boothowto |= RB_ASKNAME; |
| 976 | continue; |
| 977 | } |
| 978 | } |
| 979 | |
| 980 | ucp = (char *)USRSTACK; |
| 981 | |
| 982 | /* |
| 983 | * Construct the boot flag argument. |
| 984 | */ |
| 985 | flagsp = flags; |
| 986 | *flagsp++ = '-'; |
| 987 | options = 0; |
| 988 | |
| 989 | if (boothowto & RB_SINGLE) { |
| 990 | *flagsp++ = 's'; |
| 991 | options = 1; |
| 992 | } |
| 993 | #ifdef notyet |
| 994 | if (boothowto & RB_FASTBOOT) { |
| 995 | *flagsp++ = 'f'; |
| 996 | options = 1; |
| 997 | } |
| 998 | #endif |
| 999 | |
| 1000 | /* |
| 1001 | * Move out the flags (arg 1), if necessary. |
| 1002 | */ |
| 1003 | if (options != 0) { |
| 1004 | *flagsp++ = '\0'; |
| 1005 | i = flagsp - flags; |
| 1006 | #ifdef DEBUG |
| 1007 | aprint_normal("init: copying out flags `%s' %d\n" , flags, i); |
| 1008 | #endif |
| 1009 | arg1 = STACK_ALLOC(ucp, i); |
| 1010 | ucp = STACK_MAX(arg1, i); |
| 1011 | if ((error = copyout((void *)flags, arg1, i)) != 0) |
| 1012 | goto copyerr; |
| 1013 | } else |
| 1014 | arg1 = NULL; |
| 1015 | |
| 1016 | /* |
| 1017 | * Move out the file name (also arg 0). |
| 1018 | */ |
| 1019 | i = strlen(path) + 1; |
| 1020 | #ifdef DEBUG |
| 1021 | aprint_normal("init: copying out path `%s' %d\n" , path, i); |
| 1022 | #else |
| 1023 | if (boothowto & RB_ASKNAME || path != initpaths[0]) |
| 1024 | printf("init: trying %s\n" , path); |
| 1025 | #endif |
| 1026 | arg0 = STACK_ALLOC(ucp, i); |
| 1027 | ucp = STACK_MAX(arg0, i); |
| 1028 | if ((error = copyout(path, arg0, i)) != 0) |
| 1029 | goto copyerr; |
| 1030 | |
| 1031 | /* |
| 1032 | * Move out the arg pointers. |
| 1033 | */ |
| 1034 | ucp = (void *)STACK_ALIGN(ucp, STACK_ALIGNBYTES); |
| 1035 | uap = (char **)STACK_ALLOC(ucp, sizeof(argv)); |
| 1036 | SCARG(&args, path) = arg0; |
| 1037 | SCARG(&args, argp) = uap; |
| 1038 | SCARG(&args, envp) = NULL; |
| 1039 | slash = strrchr(path, '/'); |
| 1040 | |
| 1041 | argv[0] = slash ? arg0 + (slash + 1 - path) : arg0; |
| 1042 | argv[1] = arg1; |
| 1043 | argv[2] = NULL; |
| 1044 | if ((error = copyout(argv, uap, sizeof(argv))) != 0) |
| 1045 | goto copyerr; |
| 1046 | |
| 1047 | /* |
| 1048 | * Now try to exec the program. If it can't for any reason |
| 1049 | * other than it doesn't exist, complain. |
| 1050 | */ |
| 1051 | error = sys_execve(l, &args, retval); |
| 1052 | if (error == 0 || error == EJUSTRETURN) { |
| 1053 | KERNEL_UNLOCK_LAST(l); |
| 1054 | return; |
| 1055 | } |
| 1056 | printf("exec %s: error %d\n" , path, error); |
| 1057 | } |
| 1058 | printf("init: not found\n" ); |
| 1059 | panic("no init" ); |
| 1060 | copyerr: |
| 1061 | panic("copyout %d" , error); |
| 1062 | } |
| 1063 | |
| 1064 | /* |
| 1065 | * calculate cache size (in bytes) from physmem and vsize. |
| 1066 | */ |
| 1067 | vaddr_t |
| 1068 | calc_cache_size(vsize_t vsize, int pct, int va_pct) |
| 1069 | { |
| 1070 | paddr_t t; |
| 1071 | |
| 1072 | /* XXX should consider competing cache if any */ |
| 1073 | /* XXX should consider submaps */ |
| 1074 | t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; |
| 1075 | if (vsize != 0) { |
| 1076 | vsize = (uintmax_t)vsize * va_pct / 100; |
| 1077 | if (t > vsize) { |
| 1078 | t = vsize; |
| 1079 | } |
| 1080 | } |
| 1081 | return t; |
| 1082 | } |
| 1083 | |
| 1084 | /* |
| 1085 | * Print the system start up banner. |
| 1086 | * |
| 1087 | * - Print a limited banner if AB_SILENT. |
| 1088 | * - Always send normal banner to the log. |
| 1089 | */ |
| 1090 | #define MEM_PBUFSIZE sizeof("99999 MB") |
| 1091 | |
| 1092 | void |
| 1093 | banner(void) |
| 1094 | { |
| 1095 | static char notice[] = " Notice: this software is " |
| 1096 | "protected by copyright" ; |
| 1097 | char pbuf[81]; |
| 1098 | void (*pr)(const char *, ...) __printflike(1, 2); |
| 1099 | int i; |
| 1100 | |
| 1101 | if ((boothowto & AB_SILENT) != 0) { |
| 1102 | snprintf(pbuf, sizeof(pbuf), "%s %s (%s)" , |
| 1103 | ostype, osrelease, kernel_ident); |
| 1104 | printf_nolog("%s" , pbuf); |
| 1105 | for (i = 80 - strlen(pbuf) - sizeof(notice); i > 0; i--) |
| 1106 | printf(" " ); |
| 1107 | printf_nolog("%s\n" , notice); |
| 1108 | pr = aprint_normal; |
| 1109 | } else { |
| 1110 | pr = printf; |
| 1111 | } |
| 1112 | |
| 1113 | memset(pbuf, 0, sizeof(pbuf)); |
| 1114 | (*pr)("%s%s" , copyright, version); |
| 1115 | format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)physmem)); |
| 1116 | (*pr)("total memory = %s\n" , pbuf); |
| 1117 | format_bytes(pbuf, MEM_PBUFSIZE, ctob((uint64_t)uvmexp.free)); |
| 1118 | (*pr)("avail memory = %s\n" , pbuf); |
| 1119 | } |
| 1120 | |