init_main.c revision 1.381
1 /* $NetBSD: init_main.c,v 1.381 2009/02/12 18:24:18 christos 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 1.381 2009/02/12 18:24:18 christos Exp $"); 101 102 #include "opt_ddb.h" 103 #include "opt_ipsec.h" 104 #include "opt_ntp.h" 105 #include "opt_pipe.h" 106 #include "opt_syscall_debug.h" 107 #include "opt_sysv.h" 108 #include "opt_fileassoc.h" 109 #include "opt_ktrace.h" 110 #include "opt_pax.h" 111 #include "opt_compat_netbsd.h" 112 #include "opt_wapbl.h" 113 114 #include "ksyms.h" 115 #include "rnd.h" 116 #include "sysmon_envsys.h" 117 #include "sysmon_power.h" 118 #include "sysmon_taskq.h" 119 #include "sysmon_wdog.h" 120 #include "veriexec.h" 121 122 #include <sys/param.h> 123 #include <sys/acct.h> 124 #include <sys/filedesc.h> 125 #include <sys/file.h> 126 #include <sys/errno.h> 127 #include <sys/callout.h> 128 #include <sys/cpu.h> 129 #include <sys/kernel.h> 130 #include <sys/mount.h> 131 #include <sys/proc.h> 132 #include <sys/kthread.h> 133 #include <sys/resourcevar.h> 134 #include <sys/signalvar.h> 135 #include <sys/systm.h> 136 #include <sys/vnode.h> 137 #include <sys/fstrans.h> 138 #include <sys/tty.h> 139 #include <sys/conf.h> 140 #include <sys/disklabel.h> 141 #include <sys/buf.h> 142 #include <sys/device.h> 143 #include <sys/exec.h> 144 #include <sys/socketvar.h> 145 #include <sys/protosw.h> 146 #include <sys/percpu.h> 147 #include <sys/pset.h> 148 #include <sys/sysctl.h> 149 #include <sys/reboot.h> 150 #include <sys/user.h> 151 #include <sys/sysctl.h> 152 #include <sys/event.h> 153 #include <sys/mbuf.h> 154 #include <sys/sched.h> 155 #include <sys/sleepq.h> 156 #include <sys/iostat.h> 157 #include <sys/vmem.h> 158 #include <sys/uuid.h> 159 #include <sys/extent.h> 160 #include <sys/disk.h> 161 #include <sys/mqueue.h> 162 #include <sys/msgbuf.h> 163 #include <sys/module.h> 164 #include <sys/event.h> 165 #include <sys/lockf.h> 166 #include <sys/once.h> 167 #include <sys/ksyms.h> 168 #include <sys/uidinfo.h> 169 #include <sys/kprintf.h> 170 #ifdef FAST_IPSEC 171 #include <netipsec/ipsec.h> 172 #endif 173 #ifdef SYSVSHM 174 #include <sys/shm.h> 175 #endif 176 #ifdef SYSVSEM 177 #include <sys/sem.h> 178 #endif 179 #ifdef SYSVMSG 180 #include <sys/msg.h> 181 #endif 182 #include <sys/domain.h> 183 #include <sys/namei.h> 184 #if NRND > 0 185 #include <sys/rnd.h> 186 #endif 187 #include <sys/pipe.h> 188 #if NVERIEXEC > 0 189 #include <sys/verified_exec.h> 190 #endif /* NVERIEXEC > 0 */ 191 #ifdef KTRACE 192 #include <sys/ktrace.h> 193 #endif 194 #include <sys/kauth.h> 195 #ifdef WAPBL 196 #include <sys/wapbl.h> 197 #endif 198 #include <net80211/ieee80211_netbsd.h> 199 200 #include <sys/syscall.h> 201 #include <sys/syscallargs.h> 202 203 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 204 #include <sys/pax.h> 205 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 206 207 #include <ufs/ufs/quota.h> 208 209 #include <miscfs/genfs/genfs.h> 210 #include <miscfs/syncfs/syncfs.h> 211 212 #include <sys/cpu.h> 213 214 #include <uvm/uvm.h> 215 216 #if NSYSMON_TASKQ > 0 217 #include <dev/sysmon/sysmon_taskq.h> 218 #endif 219 220 #include <dev/cons.h> 221 222 #if NSYSMON_ENVSYS > 0 || NSYSMON_POWER > 0 || NSYSMON_WDOG > 0 223 #include <dev/sysmon/sysmonvar.h> 224 #endif 225 226 #include <net/if.h> 227 #include <net/raw_cb.h> 228 229 #include <secmodel/secmodel.h> 230 231 #ifdef COMPAT_50 232 #include <compat/sys/time.h> 233 struct timeval50 boottime50; 234 #endif 235 236 extern struct proc proc0; 237 extern struct lwp lwp0; 238 extern struct cwdinfo cwdi0; 239 extern time_t rootfstime; 240 241 #ifndef curlwp 242 struct lwp *curlwp = &lwp0; 243 #endif 244 struct proc *initproc; 245 246 struct vnode *rootvp, *swapdev_vp; 247 int boothowto; 248 int cold = 1; /* still working on startup */ 249 struct timespec boottime; /* time at system startup - will only follow settime deltas */ 250 251 int start_init_exec; /* semaphore for start_init() */ 252 253 static void check_console(struct lwp *l); 254 static void start_init(void *); 255 void main(void); 256 257 void __secmodel_none(void); 258 __weak_alias(secmodel_start,__secmodel_none); 259 void 260 __secmodel_none(void) 261 { 262 return; 263 } 264 265 /* 266 * System startup; initialize the world, create process 0, mount root 267 * filesystem, and fork to create init and pagedaemon. Most of the 268 * hard work is done in the lower-level initialization routines including 269 * startup(), which does memory initialization and autoconfiguration. 270 */ 271 void 272 main(void) 273 { 274 struct timespec time; 275 struct lwp *l; 276 struct proc *p; 277 int s, error; 278 #ifdef NVNODE_IMPLICIT 279 int usevnodes; 280 #endif 281 CPU_INFO_ITERATOR cii; 282 struct cpu_info *ci; 283 284 l = &lwp0; 285 #ifndef LWP0_CPU_INFO 286 l->l_cpu = curcpu(); 287 #endif 288 289 /* 290 * Attempt to find console and initialize 291 * in case of early panic or other messages. 292 */ 293 consinit(); 294 295 kernel_lock_init(); 296 once_init(); 297 298 uvm_init(); 299 300 #if ((NKSYMS > 0) || (NDDB > 0) || (NMODULAR > 0)) 301 ksyms_init(); 302 #endif 303 kprintf_init(); 304 305 percpu_init(); 306 307 /* Initialize lock caches. */ 308 mutex_obj_init(); 309 310 /* Initialize the extent manager. */ 311 extent_init(); 312 313 /* Do machine-dependent initialization. */ 314 cpu_startup(); 315 316 /* Initialize the sysctl subsystem. */ 317 sysctl_init(); 318 319 /* Initialize callouts, part 1. */ 320 callout_startup(); 321 322 /* 323 * Initialize the kernel authorization subsystem and start the 324 * default security model, if any. We need to do this early 325 * enough so that subsystems relying on any of the aforementioned 326 * can work properly. Since the security model may dictate the 327 * credential inheritance policy, it is needed at least before 328 * any process is created, specifically proc0. 329 */ 330 kauth_init(); 331 secmodel_start(); 332 333 /* Initialize the buffer cache */ 334 bufinit(); 335 336 /* Initialize sockets. */ 337 soinit(); 338 339 /* 340 * The following things must be done before autoconfiguration. 341 */ 342 evcnt_init(); /* initialize event counters */ 343 #if NRND > 0 344 rnd_init(); /* initialize random number generator */ 345 #endif 346 347 /* Initialize process and pgrp structures. */ 348 procinit(); 349 lwpinit(); 350 351 /* Initialize signal-related data structures. */ 352 signal_init(); 353 354 /* Initialize resource management. */ 355 resource_init(); 356 357 /* Create process 0 (the swapper). */ 358 proc0_init(); 359 360 /* Initialize the UID hash table. */ 361 uid_init(); 362 363 /* Charge root for one process. */ 364 (void)chgproccnt(0, 1); 365 366 /* Initialize timekeeping. */ 367 time_init(); 368 369 /* Initialize the run queues, turnstiles and sleep queues. */ 370 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 371 sched_rqinit(); 372 turnstile_init(); 373 sleeptab_init(&sleeptab); 374 375 /* Initialize processor-sets */ 376 psets_init(); 377 378 /* MI initialization of the boot cpu */ 379 error = mi_cpu_attach(curcpu()); 380 KASSERT(error == 0); 381 382 /* Initialize timekeeping, part 2. */ 383 time_init2(); 384 385 /* 386 * Initialize mbuf's. Do this now because we might attempt to 387 * allocate mbufs or mbuf clusters during autoconfiguration. 388 */ 389 mbinit(); 390 391 /* Initialize I/O statistics. */ 392 iostat_init(); 393 394 /* Initialize the log device. */ 395 loginit(); 396 397 /* Start module system. */ 398 module_init(); 399 400 /* Initialize the file systems. */ 401 #ifdef NVNODE_IMPLICIT 402 /* 403 * If maximum number of vnodes in namei vnode cache is not explicitly 404 * defined in kernel config, adjust the number such as we use roughly 405 * 10% of memory for vnodes and associated data structures in the 406 * assumed worst case. Do not provide fewer than NVNODE vnodes. 407 */ 408 usevnodes = 409 calc_cache_size(kernel_map, 10, VNODE_VA_MAXPCT) / VNODE_COST; 410 if (usevnodes > desiredvnodes) 411 desiredvnodes = usevnodes; 412 #endif 413 vfsinit(); 414 lf_init(); 415 416 /* Initialize fstrans. */ 417 fstrans_init(); 418 419 /* Initialize the file descriptor system. */ 420 fd_sys_init(); 421 422 /* Initialize cwd structures */ 423 cwd_sys_init(); 424 425 /* Initialize kqueue. */ 426 kqueue_init(); 427 428 /* Initialize message queues. */ 429 mqueue_sysinit(); 430 431 /* Initialize the system monitor subsystems. */ 432 #if NSYSMON_TASKQ > 0 433 sysmon_task_queue_preinit(); 434 #endif 435 436 #if NSYSMON_ENVSYS > 0 437 sysmon_envsys_init(); 438 #endif 439 440 #if NSYSMON_POWER > 0 441 sysmon_power_init(); 442 #endif 443 444 #if NSYSMON_WDOG > 0 445 sysmon_wdog_init(); 446 #endif 447 448 inittimecounter(); 449 ntp_init(); 450 451 /* Initialize the device switch tables. */ 452 devsw_init(); 453 454 /* Initialize tty subsystem. */ 455 tty_init(); 456 ttyldisc_init(); 457 458 /* Initialize the buffer cache, part 2. */ 459 bufinit2(); 460 461 /* Initialize the disk wedge subsystem. */ 462 dkwedge_init(); 463 464 /* Initialize interfaces. */ 465 ifinit1(); 466 467 /* Configure the system hardware. This will enable interrupts. */ 468 configure(); 469 470 ssp_init(); 471 472 configure2(); 473 474 ubc_init(); /* must be after autoconfig */ 475 476 #ifdef SYSVSHM 477 /* Initialize System V style shared memory. */ 478 shminit(); 479 #endif 480 481 #ifdef SYSVSEM 482 /* Initialize System V style semaphores. */ 483 seminit(); 484 #endif 485 486 #ifdef SYSVMSG 487 /* Initialize System V style message queues. */ 488 msginit(); 489 #endif 490 491 #if NVERIEXEC > 0 492 /* 493 * Initialise the Veriexec subsystem. 494 */ 495 veriexec_init(); 496 #endif /* NVERIEXEC > 0 */ 497 498 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 499 pax_init(); 500 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 501 502 #ifdef FAST_IPSEC 503 /* Attach network crypto subsystem */ 504 ipsec_attach(); 505 #endif 506 507 /* 508 * Initialize protocols. Block reception of incoming packets 509 * until everything is ready. 510 */ 511 s = splnet(); 512 ifinit(); 513 domaininit(); 514 if_attachdomain(); 515 splx(s); 516 517 #ifdef GPROF 518 /* Initialize kernel profiling. */ 519 kmstartup(); 520 #endif 521 522 /* Initialize system accounting. */ 523 acct_init(); 524 525 #ifndef PIPE_SOCKETPAIR 526 /* Initialize pipes. */ 527 pipe_init(); 528 #endif 529 530 #ifdef KTRACE 531 /* Initialize ktrace. */ 532 ktrinit(); 533 #endif 534 535 /* Initialize the UUID system calls. */ 536 uuid_init(); 537 538 #ifdef WAPBL 539 /* Initialize write-ahead physical block logging. */ 540 wapbl_init(); 541 #endif 542 543 /* 544 * Create process 1 (init(8)). We do this now, as Unix has 545 * historically had init be process 1, and changing this would 546 * probably upset a lot of people. 547 * 548 * Note that process 1 won't immediately exec init(8), but will 549 * wait for us to inform it that the root file system has been 550 * mounted. 551 */ 552 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 553 panic("fork init"); 554 555 /* 556 * Load any remaining builtin modules, and hand back temporary 557 * storage to the VM system. 558 */ 559 module_init_class(MODULE_CLASS_ANY); 560 561 /* 562 * Finalize configuration now that all real devices have been 563 * found. This needs to be done before the root device is 564 * selected, since finalization may create the root device. 565 */ 566 config_finalize(); 567 568 /* 569 * Now that autoconfiguration has completed, we can determine 570 * the root and dump devices. 571 */ 572 cpu_rootconf(); 573 cpu_dumpconf(); 574 575 /* Mount the root file system. */ 576 do { 577 domountroothook(); 578 if ((error = vfs_mountroot())) { 579 printf("cannot mount root, error = %d\n", error); 580 boothowto |= RB_ASKNAME; 581 setroot(root_device, 582 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 583 } 584 } while (error != 0); 585 mountroothook_destroy(); 586 587 /* 588 * Initialise the time-of-day clock, passing the time recorded 589 * in the root filesystem (if any) for use by systems that 590 * don't have a non-volatile time-of-day device. 591 */ 592 inittodr(rootfstime); 593 594 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 595 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 596 597 /* 598 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 599 * reference it. 600 */ 601 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 602 if (error) 603 panic("cannot find root vnode, error=%d", error); 604 cwdi0.cwdi_cdir = rootvnode; 605 VREF(cwdi0.cwdi_cdir); 606 VOP_UNLOCK(rootvnode, 0); 607 cwdi0.cwdi_rdir = NULL; 608 609 /* 610 * Now that root is mounted, we can fixup initproc's CWD 611 * info. All other processes are kthreads, which merely 612 * share proc0's CWD info. 613 */ 614 initproc->p_cwdi->cwdi_cdir = rootvnode; 615 VREF(initproc->p_cwdi->cwdi_cdir); 616 initproc->p_cwdi->cwdi_rdir = NULL; 617 618 /* 619 * Now can look at time, having had a chance to verify the time 620 * from the file system. Reset l->l_rtime as it may have been 621 * munched in mi_switch() after the time got set. 622 */ 623 getnanotime(&time); 624 boottime = time; 625 #ifdef COMPAT_50 626 { 627 struct timeval tv; 628 TIMESPEC_TO_TIMEVAL(&tv, &time); 629 timeval_to_timeval50(&tv, &boottime50); 630 } 631 #endif 632 mutex_enter(proc_lock); 633 LIST_FOREACH(p, &allproc, p_list) { 634 KASSERT((p->p_flag & PK_MARKER) == 0); 635 mutex_enter(p->p_lock); 636 TIMESPEC_TO_TIMEVAL(&p->p_stats->p_start, &time); 637 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 638 lwp_lock(l); 639 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 640 lwp_unlock(l); 641 } 642 mutex_exit(p->p_lock); 643 } 644 mutex_exit(proc_lock); 645 binuptime(&curlwp->l_stime); 646 647 for (CPU_INFO_FOREACH(cii, ci)) { 648 ci->ci_schedstate.spc_lastmod = time_second; 649 } 650 651 /* Create the pageout daemon kernel thread. */ 652 uvm_swap_init(); 653 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 654 NULL, NULL, "pgdaemon")) 655 panic("fork pagedaemon"); 656 657 /* Create the filesystem syncer kernel thread. */ 658 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 659 NULL, NULL, "ioflush")) 660 panic("fork syncer"); 661 662 /* Create the aiodone daemon kernel thread. */ 663 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 664 uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE)) 665 panic("fork aiodoned"); 666 667 vmem_rehash_start(); 668 669 /* Initialize exec structures */ 670 exec_init(1); 671 672 /* 673 * Okay, now we can let init(8) exec! It's off to userland! 674 */ 675 mutex_enter(proc_lock); 676 start_init_exec = 1; 677 cv_broadcast(&lbolt); 678 mutex_exit(proc_lock); 679 680 /* The scheduler is an infinite loop. */ 681 uvm_scheduler(); 682 /* NOTREACHED */ 683 } 684 685 static void 686 check_console(struct lwp *l) 687 { 688 struct nameidata nd; 689 int error; 690 691 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console"); 692 error = namei(&nd); 693 if (error == 0) 694 vrele(nd.ni_vp); 695 else if (error == ENOENT) 696 printf("warning: no /dev/console\n"); 697 else 698 printf("warning: lookup /dev/console: error %d\n", error); 699 } 700 701 /* 702 * List of paths to try when searching for "init". 703 */ 704 static const char * const initpaths[] = { 705 "/sbin/init", 706 "/sbin/oinit", 707 "/sbin/init.bak", 708 NULL, 709 }; 710 711 /* 712 * Start the initial user process; try exec'ing each pathname in "initpaths". 713 * The program is invoked with one argument containing the boot flags. 714 */ 715 static void 716 start_init(void *arg) 717 { 718 struct lwp *l = arg; 719 struct proc *p = l->l_proc; 720 vaddr_t addr; 721 struct sys_execve_args /* { 722 syscallarg(const char *) path; 723 syscallarg(char * const *) argp; 724 syscallarg(char * const *) envp; 725 } */ args; 726 int options, i, error; 727 register_t retval[2]; 728 char flags[4], *flagsp; 729 const char *path, *slash; 730 char *ucp, **uap, *arg0, *arg1 = NULL; 731 char ipath[129]; 732 int ipx, len; 733 734 /* 735 * Now in process 1. 736 */ 737 strncpy(p->p_comm, "init", MAXCOMLEN); 738 739 /* 740 * Wait for main() to tell us that it's safe to exec. 741 */ 742 mutex_enter(proc_lock); 743 while (start_init_exec == 0) 744 cv_wait(&lbolt, proc_lock); 745 mutex_exit(proc_lock); 746 747 /* 748 * This is not the right way to do this. We really should 749 * hand-craft a descriptor onto /dev/console to hand to init, 750 * but that's a _lot_ more work, and the benefit from this easy 751 * hack makes up for the "good is the enemy of the best" effect. 752 */ 753 check_console(l); 754 755 /* 756 * Need just enough stack to hold the faked-up "execve()" arguments. 757 */ 758 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 759 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 760 NULL, UVM_UNKNOWN_OFFSET, 0, 761 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 762 UVM_ADV_NORMAL, 763 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 764 panic("init: couldn't allocate argument space"); 765 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 766 767 ipx = 0; 768 while (1) { 769 if (boothowto & RB_ASKNAME) { 770 printf("init path"); 771 if (initpaths[ipx]) 772 printf(" (default %s)", initpaths[ipx]); 773 printf(": "); 774 len = cngetsn(ipath, sizeof(ipath)-1); 775 if (len == 4 && strcmp(ipath, "halt") == 0) { 776 cpu_reboot(RB_HALT, NULL); 777 } else if (len == 6 && strcmp(ipath, "reboot") == 0) { 778 cpu_reboot(0, NULL); 779 #if defined(DDB) 780 } else if (len == 3 && strcmp(ipath, "ddb") == 0) { 781 console_debugger(); 782 continue; 783 #endif 784 } else if (len > 0 && ipath[0] == '/') { 785 ipath[len] = '\0'; 786 path = ipath; 787 } else if (len == 0 && initpaths[ipx] != NULL) { 788 path = initpaths[ipx++]; 789 } else { 790 printf("use absolute path, "); 791 #if defined(DDB) 792 printf("\"ddb\", "); 793 #endif 794 printf("\"halt\", or \"reboot\"\n"); 795 continue; 796 } 797 } else { 798 if ((path = initpaths[ipx++]) == NULL) { 799 ipx = 0; 800 boothowto |= RB_ASKNAME; 801 continue; 802 } 803 } 804 805 ucp = (char *)USRSTACK; 806 807 /* 808 * Construct the boot flag argument. 809 */ 810 flagsp = flags; 811 *flagsp++ = '-'; 812 options = 0; 813 814 if (boothowto & RB_SINGLE) { 815 *flagsp++ = 's'; 816 options = 1; 817 } 818 #ifdef notyet 819 if (boothowto & RB_FASTBOOT) { 820 *flagsp++ = 'f'; 821 options = 1; 822 } 823 #endif 824 825 /* 826 * Move out the flags (arg 1), if necessary. 827 */ 828 if (options != 0) { 829 *flagsp++ = '\0'; 830 i = flagsp - flags; 831 #ifdef DEBUG 832 printf("init: copying out flags `%s' %d\n", flags, i); 833 #endif 834 arg1 = STACK_ALLOC(ucp, i); 835 ucp = STACK_MAX(arg1, i); 836 (void)copyout((void *)flags, arg1, i); 837 } 838 839 /* 840 * Move out the file name (also arg 0). 841 */ 842 i = strlen(path) + 1; 843 #ifdef DEBUG 844 printf("init: copying out path `%s' %d\n", path, i); 845 #else 846 if (boothowto & RB_ASKNAME || path != initpaths[0]) 847 printf("init: trying %s\n", path); 848 #endif 849 arg0 = STACK_ALLOC(ucp, i); 850 ucp = STACK_MAX(arg0, i); 851 (void)copyout(path, arg0, i); 852 853 /* 854 * Move out the arg pointers. 855 */ 856 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 857 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 858 SCARG(&args, path) = arg0; 859 SCARG(&args, argp) = uap; 860 SCARG(&args, envp) = NULL; 861 slash = strrchr(path, '/'); 862 if (slash) 863 (void)suword((void *)uap++, 864 (long)arg0 + (slash + 1 - path)); 865 else 866 (void)suword((void *)uap++, (long)arg0); 867 if (options != 0) 868 (void)suword((void *)uap++, (long)arg1); 869 (void)suword((void *)uap++, 0); /* terminator */ 870 871 /* 872 * Now try to exec the program. If can't for any reason 873 * other than it doesn't exist, complain. 874 */ 875 error = sys_execve(l, &args, retval); 876 if (error == 0 || error == EJUSTRETURN) { 877 KERNEL_UNLOCK_LAST(l); 878 return; 879 } 880 printf("exec %s: error %d\n", path, error); 881 } 882 printf("init: not found\n"); 883 panic("no init"); 884 } 885 886 /* 887 * calculate cache size from physmem and vm_map size. 888 */ 889 vaddr_t 890 calc_cache_size(struct vm_map *map, int pct, int va_pct) 891 { 892 paddr_t t; 893 894 /* XXX should consider competing cache if any */ 895 /* XXX should consider submaps */ 896 t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; 897 if (map != NULL) { 898 vsize_t vsize; 899 900 vsize = vm_map_max(map) - vm_map_min(map); 901 vsize = (uintmax_t)vsize * va_pct / 100; 902 if (t > vsize) { 903 t = vsize; 904 } 905 } 906 return t; 907 } 908
Indexes created Tue Sep 23 02:09:52 GMT 2025