1 /* $NetBSD: init_main.c,v 1.382 2009/02/13 22:41:04 apb Exp $ */ 2 3 /*- 4 * Copyright (c) 2008 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 1.382 2009/02/13 22:41:04 apb Exp $"); 101 102 #include "opt_ddb.h" 103 #include "opt_ipsec.h" 104 #include "opt_modular.h" 105 #include "opt_ntp.h" 106 #include "opt_pipe.h" 107 #include "opt_syscall_debug.h" 108 #include "opt_sysv.h" 109 #include "opt_fileassoc.h" 110 #include "opt_ktrace.h" 111 #include "opt_pax.h" 112 #include "opt_compat_netbsd.h" 113 #include "opt_wapbl.h" 114 115 #include "ksyms.h" 116 #include "rnd.h" 117 #include "sysmon_envsys.h" 118 #include "sysmon_power.h" 119 #include "sysmon_taskq.h" 120 #include "sysmon_wdog.h" 121 #include "veriexec.h" 122 123 #include <sys/param.h> 124 #include <sys/acct.h> 125 #include <sys/filedesc.h> 126 #include <sys/file.h> 127 #include <sys/errno.h> 128 #include <sys/callout.h> 129 #include <sys/cpu.h> 130 #include <sys/kernel.h> 131 #include <sys/mount.h> 132 #include <sys/proc.h> 133 #include <sys/kthread.h> 134 #include <sys/resourcevar.h> 135 #include <sys/signalvar.h> 136 #include <sys/systm.h> 137 #include <sys/vnode.h> 138 #include <sys/fstrans.h> 139 #include <sys/tty.h> 140 #include <sys/conf.h> 141 #include <sys/disklabel.h> 142 #include <sys/buf.h> 143 #include <sys/device.h> 144 #include <sys/exec.h> 145 #include <sys/socketvar.h> 146 #include <sys/protosw.h> 147 #include <sys/percpu.h> 148 #include <sys/pset.h> 149 #include <sys/sysctl.h> 150 #include <sys/reboot.h> 151 #include <sys/user.h> 152 #include <sys/sysctl.h> 153 #include <sys/event.h> 154 #include <sys/mbuf.h> 155 #include <sys/sched.h> 156 #include <sys/sleepq.h> 157 #include <sys/iostat.h> 158 #include <sys/vmem.h> 159 #include <sys/uuid.h> 160 #include <sys/extent.h> 161 #include <sys/disk.h> 162 #include <sys/mqueue.h> 163 #include <sys/msgbuf.h> 164 #include <sys/module.h> 165 #include <sys/event.h> 166 #include <sys/lockf.h> 167 #include <sys/once.h> 168 #include <sys/ksyms.h> 169 #include <sys/uidinfo.h> 170 #include <sys/kprintf.h> 171 #ifdef FAST_IPSEC 172 #include <netipsec/ipsec.h> 173 #endif 174 #ifdef SYSVSHM 175 #include <sys/shm.h> 176 #endif 177 #ifdef SYSVSEM 178 #include <sys/sem.h> 179 #endif 180 #ifdef SYSVMSG 181 #include <sys/msg.h> 182 #endif 183 #include <sys/domain.h> 184 #include <sys/namei.h> 185 #if NRND > 0 186 #include <sys/rnd.h> 187 #endif 188 #include <sys/pipe.h> 189 #if NVERIEXEC > 0 190 #include <sys/verified_exec.h> 191 #endif /* NVERIEXEC > 0 */ 192 #ifdef KTRACE 193 #include <sys/ktrace.h> 194 #endif 195 #include <sys/kauth.h> 196 #ifdef WAPBL 197 #include <sys/wapbl.h> 198 #endif 199 #include <net80211/ieee80211_netbsd.h> 200 201 #include <sys/syscall.h> 202 #include <sys/syscallargs.h> 203 204 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 205 #include <sys/pax.h> 206 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 207 208 #include <ufs/ufs/quota.h> 209 210 #include <miscfs/genfs/genfs.h> 211 #include <miscfs/syncfs/syncfs.h> 212 213 #include <sys/cpu.h> 214 215 #include <uvm/uvm.h> 216 217 #if NSYSMON_TASKQ > 0 218 #include <dev/sysmon/sysmon_taskq.h> 219 #endif 220 221 #include <dev/cons.h> 222 223 #if NSYSMON_ENVSYS > 0 || NSYSMON_POWER > 0 || NSYSMON_WDOG > 0 224 #include <dev/sysmon/sysmonvar.h> 225 #endif 226 227 #include <net/if.h> 228 #include <net/raw_cb.h> 229 230 #include <secmodel/secmodel.h> 231 232 #ifdef COMPAT_50 233 #include <compat/sys/time.h> 234 struct timeval50 boottime50; 235 #endif 236 237 extern struct proc proc0; 238 extern struct lwp lwp0; 239 extern struct cwdinfo cwdi0; 240 extern time_t rootfstime; 241 242 #ifndef curlwp 243 struct lwp *curlwp = &lwp0; 244 #endif 245 struct proc *initproc; 246 247 struct vnode *rootvp, *swapdev_vp; 248 int boothowto; 249 int cold = 1; /* still working on startup */ 250 struct timespec boottime; /* time at system startup - will only follow settime deltas */ 251 252 int start_init_exec; /* semaphore for start_init() */ 253 254 static void check_console(struct lwp *l); 255 static void start_init(void *); 256 void main(void); 257 258 void __secmodel_none(void); 259 __weak_alias(secmodel_start,__secmodel_none); 260 void 261 __secmodel_none(void) 262 { 263 return; 264 } 265 266 /* 267 * System startup; initialize the world, create process 0, mount root 268 * filesystem, and fork to create init and pagedaemon. Most of the 269 * hard work is done in the lower-level initialization routines including 270 * startup(), which does memory initialization and autoconfiguration. 271 */ 272 void 273 main(void) 274 { 275 struct timespec time; 276 struct lwp *l; 277 struct proc *p; 278 int s, error; 279 #ifdef NVNODE_IMPLICIT 280 int usevnodes; 281 #endif 282 CPU_INFO_ITERATOR cii; 283 struct cpu_info *ci; 284 285 l = &lwp0; 286 #ifndef LWP0_CPU_INFO 287 l->l_cpu = curcpu(); 288 #endif 289 290 /* 291 * Attempt to find console and initialize 292 * in case of early panic or other messages. 293 */ 294 consinit(); 295 296 kernel_lock_init(); 297 once_init(); 298 299 uvm_init(); 300 301 #if ((NKSYMS > 0) || (NDDB > 0) || (NMODULAR > 0)) 302 ksyms_init(); 303 #endif 304 kprintf_init(); 305 306 percpu_init(); 307 308 /* Initialize lock caches. */ 309 mutex_obj_init(); 310 311 /* Initialize the extent manager. */ 312 extent_init(); 313 314 /* Do machine-dependent initialization. */ 315 cpu_startup(); 316 317 /* Initialize the sysctl subsystem. */ 318 sysctl_init(); 319 320 /* Initialize callouts, part 1. */ 321 callout_startup(); 322 323 /* 324 * Initialize the kernel authorization subsystem and start the 325 * default security model, if any. We need to do this early 326 * enough so that subsystems relying on any of the aforementioned 327 * can work properly. Since the security model may dictate the 328 * credential inheritance policy, it is needed at least before 329 * any process is created, specifically proc0. 330 */ 331 kauth_init(); 332 secmodel_start(); 333 334 /* Initialize the buffer cache */ 335 bufinit(); 336 337 /* Initialize sockets. */ 338 soinit(); 339 340 /* 341 * The following things must be done before autoconfiguration. 342 */ 343 evcnt_init(); /* initialize event counters */ 344 #if NRND > 0 345 rnd_init(); /* initialize random number generator */ 346 #endif 347 348 /* Initialize process and pgrp structures. */ 349 procinit(); 350 lwpinit(); 351 352 /* Initialize signal-related data structures. */ 353 signal_init(); 354 355 /* Initialize resource management. */ 356 resource_init(); 357 358 /* Create process 0 (the swapper). */ 359 proc0_init(); 360 361 /* Initialize the UID hash table. */ 362 uid_init(); 363 364 /* Charge root for one process. */ 365 (void)chgproccnt(0, 1); 366 367 /* Initialize timekeeping. */ 368 time_init(); 369 370 /* Initialize the run queues, turnstiles and sleep queues. */ 371 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 372 sched_rqinit(); 373 turnstile_init(); 374 sleeptab_init(&sleeptab); 375 376 /* Initialize processor-sets */ 377 psets_init(); 378 379 /* MI initialization of the boot cpu */ 380 error = mi_cpu_attach(curcpu()); 381 KASSERT(error == 0); 382 383 /* Initialize timekeeping, part 2. */ 384 time_init2(); 385 386 /* 387 * Initialize mbuf's. Do this now because we might attempt to 388 * allocate mbufs or mbuf clusters during autoconfiguration. 389 */ 390 mbinit(); 391 392 /* Initialize I/O statistics. */ 393 iostat_init(); 394 395 /* Initialize the log device. */ 396 loginit(); 397 398 /* Start module system. */ 399 module_init(); 400 401 /* Initialize the file systems. */ 402 #ifdef NVNODE_IMPLICIT 403 /* 404 * If maximum number of vnodes in namei vnode cache is not explicitly 405 * defined in kernel config, adjust the number such as we use roughly 406 * 10% of memory for vnodes and associated data structures in the 407 * assumed worst case. Do not provide fewer than NVNODE vnodes. 408 */ 409 usevnodes = 410 calc_cache_size(kernel_map, 10, VNODE_VA_MAXPCT) / VNODE_COST; 411 if (usevnodes > desiredvnodes) 412 desiredvnodes = usevnodes; 413 #endif 414 vfsinit(); 415 lf_init(); 416 417 /* Initialize fstrans. */ 418 fstrans_init(); 419 420 /* Initialize the file descriptor system. */ 421 fd_sys_init(); 422 423 /* Initialize cwd structures */ 424 cwd_sys_init(); 425 426 /* Initialize kqueue. */ 427 kqueue_init(); 428 429 /* Initialize message queues. */ 430 mqueue_sysinit(); 431 432 /* Initialize the system monitor subsystems. */ 433 #if NSYSMON_TASKQ > 0 434 sysmon_task_queue_preinit(); 435 #endif 436 437 #if NSYSMON_ENVSYS > 0 438 sysmon_envsys_init(); 439 #endif 440 441 #if NSYSMON_POWER > 0 442 sysmon_power_init(); 443 #endif 444 445 #if NSYSMON_WDOG > 0 446 sysmon_wdog_init(); 447 #endif 448 449 inittimecounter(); 450 ntp_init(); 451 452 /* Initialize the device switch tables. */ 453 devsw_init(); 454 455 /* Initialize tty subsystem. */ 456 tty_init(); 457 ttyldisc_init(); 458 459 /* Initialize the buffer cache, part 2. */ 460 bufinit2(); 461 462 /* Initialize the disk wedge subsystem. */ 463 dkwedge_init(); 464 465 /* Initialize interfaces. */ 466 ifinit1(); 467 468 /* Configure the system hardware. This will enable interrupts. */ 469 configure(); 470 471 ssp_init(); 472 473 configure2(); 474 475 ubc_init(); /* must be after autoconfig */ 476 477 #ifdef SYSVSHM 478 /* Initialize System V style shared memory. */ 479 shminit(); 480 #endif 481 482 #ifdef SYSVSEM 483 /* Initialize System V style semaphores. */ 484 seminit(); 485 #endif 486 487 #ifdef SYSVMSG 488 /* Initialize System V style message queues. */ 489 msginit(); 490 #endif 491 492 #if NVERIEXEC > 0 493 /* 494 * Initialise the Veriexec subsystem. 495 */ 496 veriexec_init(); 497 #endif /* NVERIEXEC > 0 */ 498 499 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 500 pax_init(); 501 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 502 503 #ifdef FAST_IPSEC 504 /* Attach network crypto subsystem */ 505 ipsec_attach(); 506 #endif 507 508 /* 509 * Initialize protocols. Block reception of incoming packets 510 * until everything is ready. 511 */ 512 s = splnet(); 513 ifinit(); 514 domaininit(); 515 if_attachdomain(); 516 splx(s); 517 518 #ifdef GPROF 519 /* Initialize kernel profiling. */ 520 kmstartup(); 521 #endif 522 523 /* Initialize system accounting. */ 524 acct_init(); 525 526 #ifndef PIPE_SOCKETPAIR 527 /* Initialize pipes. */ 528 pipe_init(); 529 #endif 530 531 #ifdef KTRACE 532 /* Initialize ktrace. */ 533 ktrinit(); 534 #endif 535 536 /* Initialize the UUID system calls. */ 537 uuid_init(); 538 539 #ifdef WAPBL 540 /* Initialize write-ahead physical block logging. */ 541 wapbl_init(); 542 #endif 543 544 /* 545 * Create process 1 (init(8)). We do this now, as Unix has 546 * historically had init be process 1, and changing this would 547 * probably upset a lot of people. 548 * 549 * Note that process 1 won't immediately exec init(8), but will 550 * wait for us to inform it that the root file system has been 551 * mounted. 552 */ 553 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 554 panic("fork init"); 555 556 /* 557 * Load any remaining builtin modules, and hand back temporary 558 * storage to the VM system. 559 */ 560 module_init_class(MODULE_CLASS_ANY); 561 562 /* 563 * Finalize configuration now that all real devices have been 564 * found. This needs to be done before the root device is 565 * selected, since finalization may create the root device. 566 */ 567 config_finalize(); 568 569 /* 570 * Now that autoconfiguration has completed, we can determine 571 * the root and dump devices. 572 */ 573 cpu_rootconf(); 574 cpu_dumpconf(); 575 576 /* Mount the root file system. */ 577 do { 578 domountroothook(); 579 if ((error = vfs_mountroot())) { 580 printf("cannot mount root, error = %d\n", error); 581 boothowto |= RB_ASKNAME; 582 setroot(root_device, 583 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 584 } 585 } while (error != 0); 586 mountroothook_destroy(); 587 588 /* 589 * Initialise the time-of-day clock, passing the time recorded 590 * in the root filesystem (if any) for use by systems that 591 * don't have a non-volatile time-of-day device. 592 */ 593 inittodr(rootfstime); 594 595 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 596 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 597 598 /* 599 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 600 * reference it. 601 */ 602 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 603 if (error) 604 panic("cannot find root vnode, error=%d", error); 605 cwdi0.cwdi_cdir = rootvnode; 606 VREF(cwdi0.cwdi_cdir); 607 VOP_UNLOCK(rootvnode, 0); 608 cwdi0.cwdi_rdir = NULL; 609 610 /* 611 * Now that root is mounted, we can fixup initproc's CWD 612 * info. All other processes are kthreads, which merely 613 * share proc0's CWD info. 614 */ 615 initproc->p_cwdi->cwdi_cdir = rootvnode; 616 VREF(initproc->p_cwdi->cwdi_cdir); 617 initproc->p_cwdi->cwdi_rdir = NULL; 618 619 /* 620 * Now can look at time, having had a chance to verify the time 621 * from the file system. Reset l->l_rtime as it may have been 622 * munched in mi_switch() after the time got set. 623 */ 624 getnanotime(&time); 625 boottime = time; 626 #ifdef COMPAT_50 627 { 628 struct timeval tv; 629 TIMESPEC_TO_TIMEVAL(&tv, &time); 630 timeval_to_timeval50(&tv, &boottime50); 631 } 632 #endif 633 mutex_enter(proc_lock); 634 LIST_FOREACH(p, &allproc, p_list) { 635 KASSERT((p->p_flag & PK_MARKER) == 0); 636 mutex_enter(p->p_lock); 637 TIMESPEC_TO_TIMEVAL(&p->p_stats->p_start, &time); 638 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 639 lwp_lock(l); 640 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 641 lwp_unlock(l); 642 } 643 mutex_exit(p->p_lock); 644 } 645 mutex_exit(proc_lock); 646 binuptime(&curlwp->l_stime); 647 648 for (CPU_INFO_FOREACH(cii, ci)) { 649 ci->ci_schedstate.spc_lastmod = time_second; 650 } 651 652 /* Create the pageout daemon kernel thread. */ 653 uvm_swap_init(); 654 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 655 NULL, NULL, "pgdaemon")) 656 panic("fork pagedaemon"); 657 658 /* Create the filesystem syncer kernel thread. */ 659 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 660 NULL, NULL, "ioflush")) 661 panic("fork syncer"); 662 663 /* Create the aiodone daemon kernel thread. */ 664 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 665 uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE)) 666 panic("fork aiodoned"); 667 668 vmem_rehash_start(); 669 670 /* Initialize exec structures */ 671 exec_init(1); 672 673 /* 674 * Okay, now we can let init(8) exec! It's off to userland! 675 */ 676 mutex_enter(proc_lock); 677 start_init_exec = 1; 678 cv_broadcast(&lbolt); 679 mutex_exit(proc_lock); 680 681 /* The scheduler is an infinite loop. */ 682 uvm_scheduler(); 683 /* NOTREACHED */ 684 } 685 686 static void 687 check_console(struct lwp *l) 688 { 689 struct nameidata nd; 690 int error; 691 692 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console"); 693 error = namei(&nd); 694 if (error == 0) 695 vrele(nd.ni_vp); 696 else if (error == ENOENT) 697 printf("warning: no /dev/console\n"); 698 else 699 printf("warning: lookup /dev/console: error %d\n", error); 700 } 701 702 /* 703 * List of paths to try when searching for "init". 704 */ 705 static const char * const initpaths[] = { 706 "/sbin/init", 707 "/sbin/oinit", 708 "/sbin/init.bak", 709 NULL, 710 }; 711 712 /* 713 * Start the initial user process; try exec'ing each pathname in "initpaths". 714 * The program is invoked with one argument containing the boot flags. 715 */ 716 static void 717 start_init(void *arg) 718 { 719 struct lwp *l = arg; 720 struct proc *p = l->l_proc; 721 vaddr_t addr; 722 struct sys_execve_args /* { 723 syscallarg(const char *) path; 724 syscallarg(char * const *) argp; 725 syscallarg(char * const *) envp; 726 } */ args; 727 int options, i, error; 728 register_t retval[2]; 729 char flags[4], *flagsp; 730 const char *path, *slash; 731 char *ucp, **uap, *arg0, *arg1 = NULL; 732 char ipath[129]; 733 int ipx, len; 734 735 /* 736 * Now in process 1. 737 */ 738 strncpy(p->p_comm, "init", MAXCOMLEN); 739 740 /* 741 * Wait for main() to tell us that it's safe to exec. 742 */ 743 mutex_enter(proc_lock); 744 while (start_init_exec == 0) 745 cv_wait(&lbolt, proc_lock); 746 mutex_exit(proc_lock); 747 748 /* 749 * This is not the right way to do this. We really should 750 * hand-craft a descriptor onto /dev/console to hand to init, 751 * but that's a _lot_ more work, and the benefit from this easy 752 * hack makes up for the "good is the enemy of the best" effect. 753 */ 754 check_console(l); 755 756 /* 757 * Need just enough stack to hold the faked-up "execve()" arguments. 758 */ 759 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 760 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 761 NULL, UVM_UNKNOWN_OFFSET, 0, 762 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 763 UVM_ADV_NORMAL, 764 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 765 panic("init: couldn't allocate argument space"); 766 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 767 768 ipx = 0; 769 while (1) { 770 if (boothowto & RB_ASKNAME) { 771 printf("init path"); 772 if (initpaths[ipx]) 773 printf(" (default %s)", initpaths[ipx]); 774 printf(": "); 775 len = cngetsn(ipath, sizeof(ipath)-1); 776 if (len == 4 && strcmp(ipath, "halt") == 0) { 777 cpu_reboot(RB_HALT, NULL); 778 } else if (len == 6 && strcmp(ipath, "reboot") == 0) { 779 cpu_reboot(0, NULL); 780 #if defined(DDB) 781 } else if (len == 3 && strcmp(ipath, "ddb") == 0) { 782 console_debugger(); 783 continue; 784 #endif 785 } else if (len > 0 && ipath[0] == '/') { 786 ipath[len] = '\0'; 787 path = ipath; 788 } else if (len == 0 && initpaths[ipx] != NULL) { 789 path = initpaths[ipx++]; 790 } else { 791 printf("use absolute path, "); 792 #if defined(DDB) 793 printf("\"ddb\", "); 794 #endif 795 printf("\"halt\", or \"reboot\"\n"); 796 continue; 797 } 798 } else { 799 if ((path = initpaths[ipx++]) == NULL) { 800 ipx = 0; 801 boothowto |= RB_ASKNAME; 802 continue; 803 } 804 } 805 806 ucp = (char *)USRSTACK; 807 808 /* 809 * Construct the boot flag argument. 810 */ 811 flagsp = flags; 812 *flagsp++ = '-'; 813 options = 0; 814 815 if (boothowto & RB_SINGLE) { 816 *flagsp++ = 's'; 817 options = 1; 818 } 819 #ifdef notyet 820 if (boothowto & RB_FASTBOOT) { 821 *flagsp++ = 'f'; 822 options = 1; 823 } 824 #endif 825 826 /* 827 * Move out the flags (arg 1), if necessary. 828 */ 829 if (options != 0) { 830 *flagsp++ = '\0'; 831 i = flagsp - flags; 832 #ifdef DEBUG 833 printf("init: copying out flags `%s' %d\n", flags, i); 834 #endif 835 arg1 = STACK_ALLOC(ucp, i); 836 ucp = STACK_MAX(arg1, i); 837 (void)copyout((void *)flags, arg1, i); 838 } 839 840 /* 841 * Move out the file name (also arg 0). 842 */ 843 i = strlen(path) + 1; 844 #ifdef DEBUG 845 printf("init: copying out path `%s' %d\n", path, i); 846 #else 847 if (boothowto & RB_ASKNAME || path != initpaths[0]) 848 printf("init: trying %s\n", path); 849 #endif 850 arg0 = STACK_ALLOC(ucp, i); 851 ucp = STACK_MAX(arg0, i); 852 (void)copyout(path, arg0, i); 853 854 /* 855 * Move out the arg pointers. 856 */ 857 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 858 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 859 SCARG(&args, path) = arg0; 860 SCARG(&args, argp) = uap; 861 SCARG(&args, envp) = NULL; 862 slash = strrchr(path, '/'); 863 if (slash) 864 (void)suword((void *)uap++, 865 (long)arg0 + (slash + 1 - path)); 866 else 867 (void)suword((void *)uap++, (long)arg0); 868 if (options != 0) 869 (void)suword((void *)uap++, (long)arg1); 870 (void)suword((void *)uap++, 0); /* terminator */ 871 872 /* 873 * Now try to exec the program. If can't for any reason 874 * other than it doesn't exist, complain. 875 */ 876 error = sys_execve(l, &args, retval); 877 if (error == 0 || error == EJUSTRETURN) { 878 KERNEL_UNLOCK_LAST(l); 879 return; 880 } 881 printf("exec %s: error %d\n", path, error); 882 } 883 printf("init: not found\n"); 884 panic("no init"); 885 } 886 887 /* 888 * calculate cache size from physmem and vm_map size. 889 */ 890 vaddr_t 891 calc_cache_size(struct vm_map *map, int pct, int va_pct) 892 { 893 paddr_t t; 894 895 /* XXX should consider competing cache if any */ 896 /* XXX should consider submaps */ 897 t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; 898 if (map != NULL) { 899 vsize_t vsize; 900 901 vsize = vm_map_max(map) - vm_map_min(map); 902 vsize = (uintmax_t)vsize * va_pct / 100; 903 if (t > vsize) { 904 t = vsize; 905 } 906 } 907 return t; 908 } 909