init_main.c revision 1.376
1 /* $NetBSD: init_main.c,v 1.376 2008/11/30 18:21:36 martin 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 1.376 2008/11/30 18:21:36 martin 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_wapbl.h" 112 113 #include "rnd.h" 114 #include "sysmon_envsys.h" 115 #include "sysmon_power.h" 116 #include "sysmon_taskq.h" 117 #include "sysmon_wdog.h" 118 #include "veriexec.h" 119 120 #include <sys/param.h> 121 #include <sys/acct.h> 122 #include <sys/filedesc.h> 123 #include <sys/file.h> 124 #include <sys/errno.h> 125 #include <sys/callout.h> 126 #include <sys/cpu.h> 127 #include <sys/kernel.h> 128 #include <sys/mount.h> 129 #include <sys/proc.h> 130 #include <sys/kthread.h> 131 #include <sys/resourcevar.h> 132 #include <sys/signalvar.h> 133 #include <sys/systm.h> 134 #include <sys/vnode.h> 135 #include <sys/fstrans.h> 136 #include <sys/tty.h> 137 #include <sys/conf.h> 138 #include <sys/disklabel.h> 139 #include <sys/buf.h> 140 #include <sys/device.h> 141 #include <sys/exec.h> 142 #include <sys/socketvar.h> 143 #include <sys/protosw.h> 144 #include <sys/percpu.h> 145 #include <sys/pset.h> 146 #include <sys/sysctl.h> 147 #include <sys/reboot.h> 148 #include <sys/user.h> 149 #include <sys/sysctl.h> 150 #include <sys/event.h> 151 #include <sys/mbuf.h> 152 #include <sys/sched.h> 153 #include <sys/sleepq.h> 154 #include <sys/iostat.h> 155 #include <sys/vmem.h> 156 #include <sys/uuid.h> 157 #include <sys/extent.h> 158 #include <sys/disk.h> 159 #include <sys/mqueue.h> 160 #include <sys/msgbuf.h> 161 #include <sys/module.h> 162 #include <sys/event.h> 163 #include <sys/lockf.h> 164 #include <sys/once.h> 165 #include <sys/ksyms.h> 166 #include <sys/uidinfo.h> 167 #ifdef FAST_IPSEC 168 #include <netipsec/ipsec.h> 169 #endif 170 #ifdef SYSVSHM 171 #include <sys/shm.h> 172 #endif 173 #ifdef SYSVSEM 174 #include <sys/sem.h> 175 #endif 176 #ifdef SYSVMSG 177 #include <sys/msg.h> 178 #endif 179 #include <sys/domain.h> 180 #include <sys/namei.h> 181 #if NRND > 0 182 #include <sys/rnd.h> 183 #endif 184 #include <sys/pipe.h> 185 #if NVERIEXEC > 0 186 #include <sys/verified_exec.h> 187 #endif /* NVERIEXEC > 0 */ 188 #ifdef KTRACE 189 #include <sys/ktrace.h> 190 #endif 191 #include <sys/kauth.h> 192 #ifdef WAPBL 193 #include <sys/wapbl.h> 194 #endif 195 #include <net80211/ieee80211_netbsd.h> 196 197 #include <sys/syscall.h> 198 #include <sys/syscallargs.h> 199 200 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 201 #include <sys/pax.h> 202 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 203 204 #include <ufs/ufs/quota.h> 205 206 #include <miscfs/genfs/genfs.h> 207 #include <miscfs/syncfs/syncfs.h> 208 209 #include <sys/cpu.h> 210 211 #include <uvm/uvm.h> 212 213 #if NSYSMON_TASKQ > 0 214 #include <dev/sysmon/sysmon_taskq.h> 215 #endif 216 217 #include <dev/cons.h> 218 219 #if NSYSMON_ENVSYS > 0 || NSYSMON_POWER > 0 || NSYSMON_WDOG > 0 220 #include <dev/sysmon/sysmonvar.h> 221 #endif 222 223 #include <net/if.h> 224 #include <net/raw_cb.h> 225 226 #include <secmodel/secmodel.h> 227 228 extern struct proc proc0; 229 extern struct lwp lwp0; 230 extern struct cwdinfo cwdi0; 231 extern time_t rootfstime; 232 233 #ifndef curlwp 234 struct lwp *curlwp = &lwp0; 235 #endif 236 struct proc *initproc; 237 238 struct vnode *rootvp, *swapdev_vp; 239 int boothowto; 240 int cold = 1; /* still working on startup */ 241 struct timeval boottime; /* time at system startup - will only follow settime deltas */ 242 243 int start_init_exec; /* semaphore for start_init() */ 244 245 static void check_console(struct lwp *l); 246 static void start_init(void *); 247 void main(void); 248 void ssp_init(void); 249 250 #if defined(__SSP__) || defined(__SSP_ALL__) 251 long __stack_chk_guard[8] = {0, 0, 0, 0, 0, 0, 0, 0}; 252 void __stack_chk_fail(void); 253 254 void 255 __stack_chk_fail(void) 256 { 257 panic("stack overflow detected; terminated"); 258 } 259 260 void 261 ssp_init(void) 262 { 263 int s; 264 265 #ifdef DIAGNOSTIC 266 printf("Initializing SSP:"); 267 #endif 268 /* 269 * We initialize ssp here carefully: 270 * 1. after we got some entropy 271 * 2. without calling a function 272 */ 273 size_t i; 274 long guard[__arraycount(__stack_chk_guard)]; 275 276 arc4randbytes(guard, sizeof(guard)); 277 s = splhigh(); 278 for (i = 0; i < __arraycount(guard); i++) 279 __stack_chk_guard[i] = guard[i]; 280 splx(s); 281 #ifdef DIAGNOSTIC 282 for (i = 0; i < __arraycount(guard); i++) 283 printf("%lx ", guard[i]); 284 printf("\n"); 285 #endif 286 } 287 #else 288 void 289 ssp_init(void) 290 { 291 292 } 293 #endif 294 295 void __secmodel_none(void); 296 __weak_alias(secmodel_start,__secmodel_none); 297 void 298 __secmodel_none(void) 299 { 300 return; 301 } 302 303 /* 304 * System startup; initialize the world, create process 0, mount root 305 * filesystem, and fork to create init and pagedaemon. Most of the 306 * hard work is done in the lower-level initialization routines including 307 * startup(), which does memory initialization and autoconfiguration. 308 */ 309 void 310 main(void) 311 { 312 struct timeval time; 313 struct lwp *l; 314 struct proc *p; 315 int s, error; 316 #ifdef NVNODE_IMPLICIT 317 int usevnodes; 318 #endif 319 CPU_INFO_ITERATOR cii; 320 struct cpu_info *ci; 321 322 l = &lwp0; 323 #ifndef LWP0_CPU_INFO 324 l->l_cpu = curcpu(); 325 #endif 326 327 /* 328 * Attempt to find console and initialize 329 * in case of early panic or other messages. 330 */ 331 consinit(); 332 333 kernel_lock_init(); 334 once_init(); 335 336 uvm_init(); 337 338 ksyms_init(); 339 340 percpu_init(); 341 342 /* Initialize lock caches. */ 343 mutex_obj_init(); 344 345 /* Initialize the extent manager. */ 346 extent_init(); 347 348 /* Do machine-dependent initialization. */ 349 cpu_startup(); 350 351 /* Initialize callouts, part 1. */ 352 callout_startup(); 353 354 /* 355 * Initialize the kernel authorization subsystem and start the 356 * default security model, if any. We need to do this early 357 * enough so that subsystems relying on any of the aforementioned 358 * can work properly. Since the security model may dictate the 359 * credential inheritance policy, it is needed at least before 360 * any process is created, specifically proc0. 361 */ 362 kauth_init(); 363 secmodel_start(); 364 365 /* Initialize the buffer cache */ 366 bufinit(); 367 368 /* Initialize sockets. */ 369 soinit(); 370 371 /* 372 * The following things must be done before autoconfiguration. 373 */ 374 evcnt_init(); /* initialize event counters */ 375 #if NRND > 0 376 rnd_init(); /* initialize random number generator */ 377 #endif 378 379 /* Initialize process and pgrp structures. */ 380 procinit(); 381 lwpinit(); 382 383 /* Initialize signal-related data structures. */ 384 signal_init(); 385 386 /* Initialize resource management. */ 387 resource_init(); 388 389 /* Create process 0 (the swapper). */ 390 proc0_init(); 391 392 /* Initialize the UID hash table. */ 393 uid_init(); 394 395 /* Charge root for one process. */ 396 (void)chgproccnt(0, 1); 397 398 /* Initialize timekeeping. */ 399 time_init(); 400 401 /* Initialize the run queues, turnstiles and sleep queues. */ 402 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 403 sched_rqinit(); 404 turnstile_init(); 405 sleeptab_init(&sleeptab); 406 407 /* Initialize processor-sets */ 408 psets_init(); 409 410 /* MI initialization of the boot cpu */ 411 error = mi_cpu_attach(curcpu()); 412 KASSERT(error == 0); 413 414 /* Initialize timekeeping, part 2. */ 415 time_init2(); 416 417 /* 418 * Initialize mbuf's. Do this now because we might attempt to 419 * allocate mbufs or mbuf clusters during autoconfiguration. 420 */ 421 mbinit(); 422 423 /* Initialize the sysctl subsystem. */ 424 sysctl_init(); 425 426 /* Initialize I/O statistics. */ 427 iostat_init(); 428 429 /* Initialize the log device. */ 430 loginit(); 431 432 /* Start module system. */ 433 module_init(); 434 435 /* Initialize the file systems. */ 436 #ifdef NVNODE_IMPLICIT 437 /* 438 * If maximum number of vnodes in namei vnode cache is not explicitly 439 * defined in kernel config, adjust the number such as we use roughly 440 * 10% of memory for vnodes and associated data structures in the 441 * assumed worst case. Do not provide fewer than NVNODE vnodes. 442 */ 443 usevnodes = 444 calc_cache_size(kernel_map, 10, VNODE_VA_MAXPCT) / VNODE_COST; 445 if (usevnodes > desiredvnodes) 446 desiredvnodes = usevnodes; 447 #endif 448 vfsinit(); 449 lf_init(); 450 451 /* Initialize fstrans. */ 452 fstrans_init(); 453 454 /* Initialize the file descriptor system. */ 455 fd_sys_init(); 456 457 /* Initialize cwd structures */ 458 cwd_sys_init(); 459 460 /* Initialize kqueue. */ 461 kqueue_init(); 462 463 /* Initialize message queues. */ 464 mqueue_sysinit(); 465 466 /* Initialize the system monitor subsystems. */ 467 #if NSYSMON_TASKQ > 0 468 sysmon_task_queue_preinit(); 469 #endif 470 471 #if NSYSMON_ENVSYS > 0 472 sysmon_envsys_init(); 473 #endif 474 475 #if NSYSMON_POWER > 0 476 sysmon_power_init(); 477 #endif 478 479 #if NSYSMON_WDOG > 0 480 sysmon_wdog_init(); 481 #endif 482 483 inittimecounter(); 484 ntp_init(); 485 486 /* Initialize the device switch tables. */ 487 devsw_init(); 488 489 /* Initialize tty subsystem. */ 490 tty_init(); 491 ttyldisc_init(); 492 493 /* Initialize the buffer cache, part 2. */ 494 bufinit2(); 495 496 /* Initialize the disk wedge subsystem. */ 497 dkwedge_init(); 498 499 /* Initialize interfaces. */ 500 ifinit1(); 501 502 /* Configure the system hardware. This will enable interrupts. */ 503 configure(); 504 505 ubc_init(); /* must be after autoconfig */ 506 507 #ifdef SYSVSHM 508 /* Initialize System V style shared memory. */ 509 shminit(); 510 #endif 511 512 #ifdef SYSVSEM 513 /* Initialize System V style semaphores. */ 514 seminit(); 515 #endif 516 517 #ifdef SYSVMSG 518 /* Initialize System V style message queues. */ 519 msginit(); 520 #endif 521 522 #if NVERIEXEC > 0 523 /* 524 * Initialise the Veriexec subsystem. 525 */ 526 veriexec_init(); 527 #endif /* NVERIEXEC > 0 */ 528 529 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 530 pax_init(); 531 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 532 533 #ifdef FAST_IPSEC 534 /* Attach network crypto subsystem */ 535 ipsec_attach(); 536 #endif 537 538 /* 539 * Initialize protocols. Block reception of incoming packets 540 * until everything is ready. 541 */ 542 s = splnet(); 543 ifinit(); 544 domaininit(); 545 if_attachdomain(); 546 splx(s); 547 548 #ifdef GPROF 549 /* Initialize kernel profiling. */ 550 kmstartup(); 551 #endif 552 553 /* Initialize system accounting. */ 554 acct_init(); 555 556 #ifndef PIPE_SOCKETPAIR 557 /* Initialize pipes. */ 558 pipe_init(); 559 #endif 560 561 #ifdef KTRACE 562 /* Initialize ktrace. */ 563 ktrinit(); 564 #endif 565 566 /* Initialize the UUID system calls. */ 567 uuid_init(); 568 569 #ifdef WAPBL 570 /* Initialize write-ahead physical block logging. */ 571 wapbl_init(); 572 #endif 573 574 /* 575 * Create process 1 (init(8)). We do this now, as Unix has 576 * historically had init be process 1, and changing this would 577 * probably upset a lot of people. 578 * 579 * Note that process 1 won't immediately exec init(8), but will 580 * wait for us to inform it that the root file system has been 581 * mounted. 582 */ 583 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 584 panic("fork init"); 585 586 /* 587 * Load any remaining builtin modules, and hand back temporary 588 * storage to the VM system. 589 */ 590 module_init_class(MODULE_CLASS_ANY); 591 592 /* 593 * Finalize configuration now that all real devices have been 594 * found. This needs to be done before the root device is 595 * selected, since finalization may create the root device. 596 */ 597 config_finalize(); 598 599 /* 600 * Now that autoconfiguration has completed, we can determine 601 * the root and dump devices. 602 */ 603 cpu_rootconf(); 604 cpu_dumpconf(); 605 606 /* Mount the root file system. */ 607 do { 608 domountroothook(); 609 if ((error = vfs_mountroot())) { 610 printf("cannot mount root, error = %d\n", error); 611 boothowto |= RB_ASKNAME; 612 setroot(root_device, 613 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 614 } 615 } while (error != 0); 616 mountroothook_destroy(); 617 618 /* 619 * Initialise the time-of-day clock, passing the time recorded 620 * in the root filesystem (if any) for use by systems that 621 * don't have a non-volatile time-of-day device. 622 */ 623 inittodr(rootfstime); 624 625 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 626 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 627 628 /* 629 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 630 * reference it. 631 */ 632 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 633 if (error) 634 panic("cannot find root vnode, error=%d", error); 635 cwdi0.cwdi_cdir = rootvnode; 636 VREF(cwdi0.cwdi_cdir); 637 VOP_UNLOCK(rootvnode, 0); 638 cwdi0.cwdi_rdir = NULL; 639 640 /* 641 * Now that root is mounted, we can fixup initproc's CWD 642 * info. All other processes are kthreads, which merely 643 * share proc0's CWD info. 644 */ 645 initproc->p_cwdi->cwdi_cdir = rootvnode; 646 VREF(initproc->p_cwdi->cwdi_cdir); 647 initproc->p_cwdi->cwdi_rdir = NULL; 648 649 /* 650 * Now can look at time, having had a chance to verify the time 651 * from the file system. Reset l->l_rtime as it may have been 652 * munched in mi_switch() after the time got set. 653 */ 654 getmicrotime(&time); 655 boottime = time; 656 mutex_enter(proc_lock); 657 LIST_FOREACH(p, &allproc, p_list) { 658 KASSERT((p->p_flag & PK_MARKER) == 0); 659 mutex_enter(p->p_lock); 660 p->p_stats->p_start = time; 661 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 662 lwp_lock(l); 663 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 664 lwp_unlock(l); 665 } 666 mutex_exit(p->p_lock); 667 } 668 mutex_exit(proc_lock); 669 binuptime(&curlwp->l_stime); 670 671 for (CPU_INFO_FOREACH(cii, ci)) { 672 ci->ci_schedstate.spc_lastmod = time_second; 673 } 674 675 /* Create the pageout daemon kernel thread. */ 676 uvm_swap_init(); 677 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 678 NULL, NULL, "pgdaemon")) 679 panic("fork pagedaemon"); 680 681 /* Create the filesystem syncer kernel thread. */ 682 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 683 NULL, NULL, "ioflush")) 684 panic("fork syncer"); 685 686 /* Create the aiodone daemon kernel thread. */ 687 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 688 uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE)) 689 panic("fork aiodoned"); 690 691 vmem_rehash_start(); 692 693 /* Initialize exec structures */ 694 exec_init(1); 695 696 /* 697 * Okay, now we can let init(8) exec! It's off to userland! 698 */ 699 mutex_enter(proc_lock); 700 start_init_exec = 1; 701 cv_broadcast(&lbolt); 702 mutex_exit(proc_lock); 703 704 /* The scheduler is an infinite loop. */ 705 uvm_scheduler(); 706 /* NOTREACHED */ 707 } 708 709 static void 710 check_console(struct lwp *l) 711 { 712 struct nameidata nd; 713 int error; 714 715 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console"); 716 error = namei(&nd); 717 if (error == 0) 718 vrele(nd.ni_vp); 719 else if (error == ENOENT) 720 printf("warning: no /dev/console\n"); 721 else 722 printf("warning: lookup /dev/console: error %d\n", error); 723 } 724 725 /* 726 * List of paths to try when searching for "init". 727 */ 728 static const char * const initpaths[] = { 729 "/sbin/init", 730 "/sbin/oinit", 731 "/sbin/init.bak", 732 NULL, 733 }; 734 735 /* 736 * Start the initial user process; try exec'ing each pathname in "initpaths". 737 * The program is invoked with one argument containing the boot flags. 738 */ 739 static void 740 start_init(void *arg) 741 { 742 struct lwp *l = arg; 743 struct proc *p = l->l_proc; 744 vaddr_t addr; 745 struct sys_execve_args /* { 746 syscallarg(const char *) path; 747 syscallarg(char * const *) argp; 748 syscallarg(char * const *) envp; 749 } */ args; 750 int options, i, error; 751 register_t retval[2]; 752 char flags[4], *flagsp; 753 const char *path, *slash; 754 char *ucp, **uap, *arg0, *arg1 = NULL; 755 char ipath[129]; 756 int ipx, len; 757 758 /* 759 * Now in process 1. 760 */ 761 strncpy(p->p_comm, "init", MAXCOMLEN); 762 763 /* 764 * Wait for main() to tell us that it's safe to exec. 765 */ 766 mutex_enter(proc_lock); 767 while (start_init_exec == 0) 768 cv_wait(&lbolt, proc_lock); 769 mutex_exit(proc_lock); 770 771 /* 772 * This is not the right way to do this. We really should 773 * hand-craft a descriptor onto /dev/console to hand to init, 774 * but that's a _lot_ more work, and the benefit from this easy 775 * hack makes up for the "good is the enemy of the best" effect. 776 */ 777 check_console(l); 778 779 /* 780 * Need just enough stack to hold the faked-up "execve()" arguments. 781 */ 782 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 783 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 784 NULL, UVM_UNKNOWN_OFFSET, 0, 785 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 786 UVM_ADV_NORMAL, 787 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 788 panic("init: couldn't allocate argument space"); 789 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 790 791 ipx = 0; 792 while (1) { 793 if (boothowto & RB_ASKNAME) { 794 printf("init path"); 795 if (initpaths[ipx]) 796 printf(" (default %s)", initpaths[ipx]); 797 printf(": "); 798 len = cngetsn(ipath, sizeof(ipath)-1); 799 if (len == 4 && strcmp(ipath, "halt") == 0) { 800 cpu_reboot(RB_HALT, NULL); 801 } else if (len == 6 && strcmp(ipath, "reboot") == 0) { 802 cpu_reboot(0, NULL); 803 #if defined(DDB) 804 } else if (len == 3 && strcmp(ipath, "ddb") == 0) { 805 console_debugger(); 806 continue; 807 #endif 808 } else if (len > 0 && ipath[0] == '/') { 809 ipath[len] = '\0'; 810 path = ipath; 811 } else if (len == 0 && initpaths[ipx] != NULL) { 812 path = initpaths[ipx++]; 813 } else { 814 printf("use absolute path, "); 815 #if defined(DDB) 816 printf("\"ddb\", "); 817 #endif 818 printf("\"halt\", or \"reboot\"\n"); 819 continue; 820 } 821 } else { 822 if ((path = initpaths[ipx++]) == NULL) { 823 ipx = 0; 824 boothowto |= RB_ASKNAME; 825 continue; 826 } 827 } 828 829 ucp = (char *)USRSTACK; 830 831 /* 832 * Construct the boot flag argument. 833 */ 834 flagsp = flags; 835 *flagsp++ = '-'; 836 options = 0; 837 838 if (boothowto & RB_SINGLE) { 839 *flagsp++ = 's'; 840 options = 1; 841 } 842 #ifdef notyet 843 if (boothowto & RB_FASTBOOT) { 844 *flagsp++ = 'f'; 845 options = 1; 846 } 847 #endif 848 849 /* 850 * Move out the flags (arg 1), if necessary. 851 */ 852 if (options != 0) { 853 *flagsp++ = '\0'; 854 i = flagsp - flags; 855 #ifdef DEBUG 856 printf("init: copying out flags `%s' %d\n", flags, i); 857 #endif 858 arg1 = STACK_ALLOC(ucp, i); 859 ucp = STACK_MAX(arg1, i); 860 (void)copyout((void *)flags, arg1, i); 861 } 862 863 /* 864 * Move out the file name (also arg 0). 865 */ 866 i = strlen(path) + 1; 867 #ifdef DEBUG 868 printf("init: copying out path `%s' %d\n", path, i); 869 #else 870 if (boothowto & RB_ASKNAME || path != initpaths[0]) 871 printf("init: trying %s\n", path); 872 #endif 873 arg0 = STACK_ALLOC(ucp, i); 874 ucp = STACK_MAX(arg0, i); 875 (void)copyout(path, arg0, i); 876 877 /* 878 * Move out the arg pointers. 879 */ 880 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 881 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 882 SCARG(&args, path) = arg0; 883 SCARG(&args, argp) = uap; 884 SCARG(&args, envp) = NULL; 885 slash = strrchr(path, '/'); 886 if (slash) 887 (void)suword((void *)uap++, 888 (long)arg0 + (slash + 1 - path)); 889 else 890 (void)suword((void *)uap++, (long)arg0); 891 if (options != 0) 892 (void)suword((void *)uap++, (long)arg1); 893 (void)suword((void *)uap++, 0); /* terminator */ 894 895 /* 896 * Now try to exec the program. If can't for any reason 897 * other than it doesn't exist, complain. 898 */ 899 error = sys_execve(l, &args, retval); 900 if (error == 0 || error == EJUSTRETURN) { 901 KERNEL_UNLOCK_LAST(l); 902 return; 903 } 904 printf("exec %s: error %d\n", path, error); 905 } 906 printf("init: not found\n"); 907 panic("no init"); 908 } 909 910 /* 911 * calculate cache size from physmem and vm_map size. 912 */ 913 vaddr_t 914 calc_cache_size(struct vm_map *map, int pct, int va_pct) 915 { 916 paddr_t t; 917 918 /* XXX should consider competing cache if any */ 919 /* XXX should consider submaps */ 920 t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; 921 if (map != NULL) { 922 vsize_t vsize; 923 924 vsize = vm_map_max(map) - vm_map_min(map); 925 vsize = (uintmax_t)vsize * va_pct / 100; 926 if (t > vsize) { 927 t = vsize; 928 } 929 } 930 return t; 931 } 932
Indexes created Tue Sep 23 07:09:52 GMT 2025