init_main.c revision 1.375
1 /* $NetBSD: init_main.c,v 1.375 2008/11/18 11:36:58 pooka 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 1.375 2008/11/18 11:36:58 pooka 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 percpu_init(); 339 340 /* Initialize lock caches. */ 341 mutex_obj_init(); 342 343 /* Initialize the extent manager. */ 344 extent_init(); 345 346 /* Do machine-dependent initialization. */ 347 cpu_startup(); 348 349 /* Initialize callouts, part 1. */ 350 callout_startup(); 351 352 /* 353 * Initialize the kernel authorization subsystem and start the 354 * default security model, if any. We need to do this early 355 * enough so that subsystems relying on any of the aforementioned 356 * can work properly. Since the security model may dictate the 357 * credential inheritance policy, it is needed at least before 358 * any process is created, specifically proc0. 359 */ 360 kauth_init(); 361 secmodel_start(); 362 363 /* Initialize the buffer cache */ 364 bufinit(); 365 366 /* Initialize sockets. */ 367 soinit(); 368 369 /* 370 * The following things must be done before autoconfiguration. 371 */ 372 evcnt_init(); /* initialize event counters */ 373 #if NRND > 0 374 rnd_init(); /* initialize random number generator */ 375 #endif 376 377 /* Initialize process and pgrp structures. */ 378 procinit(); 379 lwpinit(); 380 381 /* Initialize signal-related data structures. */ 382 signal_init(); 383 384 /* Initialize resource management. */ 385 resource_init(); 386 387 /* Create process 0 (the swapper). */ 388 proc0_init(); 389 390 /* Initialize the UID hash table. */ 391 uid_init(); 392 393 /* Charge root for one process. */ 394 (void)chgproccnt(0, 1); 395 396 /* Initialize timekeeping. */ 397 time_init(); 398 399 /* Initialize the run queues, turnstiles and sleep queues. */ 400 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 401 sched_rqinit(); 402 turnstile_init(); 403 sleeptab_init(&sleeptab); 404 405 /* Initialize processor-sets */ 406 psets_init(); 407 408 /* MI initialization of the boot cpu */ 409 error = mi_cpu_attach(curcpu()); 410 KASSERT(error == 0); 411 412 /* Initialize timekeeping, part 2. */ 413 time_init2(); 414 415 /* 416 * Initialize mbuf's. Do this now because we might attempt to 417 * allocate mbufs or mbuf clusters during autoconfiguration. 418 */ 419 mbinit(); 420 421 /* Initialize the sysctl subsystem. */ 422 sysctl_init(); 423 424 /* Initialize I/O statistics. */ 425 iostat_init(); 426 427 /* Initialize the log device. */ 428 loginit(); 429 430 /* Start module system. */ 431 module_init(); 432 433 /* Initialize the file systems. */ 434 #ifdef NVNODE_IMPLICIT 435 /* 436 * If maximum number of vnodes in namei vnode cache is not explicitly 437 * defined in kernel config, adjust the number such as we use roughly 438 * 10% of memory for vnodes and associated data structures in the 439 * assumed worst case. Do not provide fewer than NVNODE vnodes. 440 */ 441 usevnodes = 442 calc_cache_size(kernel_map, 10, VNODE_VA_MAXPCT) / VNODE_COST; 443 if (usevnodes > desiredvnodes) 444 desiredvnodes = usevnodes; 445 #endif 446 vfsinit(); 447 lf_init(); 448 449 /* Initialize fstrans. */ 450 fstrans_init(); 451 452 /* Initialize the file descriptor system. */ 453 fd_sys_init(); 454 455 /* Initialize cwd structures */ 456 cwd_sys_init(); 457 458 /* Initialize kqueue. */ 459 kqueue_init(); 460 461 /* Initialize message queues. */ 462 mqueue_sysinit(); 463 464 /* Initialize the system monitor subsystems. */ 465 #if NSYSMON_TASKQ > 0 466 sysmon_task_queue_preinit(); 467 #endif 468 469 #if NSYSMON_ENVSYS > 0 470 sysmon_envsys_init(); 471 #endif 472 473 #if NSYSMON_POWER > 0 474 sysmon_power_init(); 475 #endif 476 477 #if NSYSMON_WDOG > 0 478 sysmon_wdog_init(); 479 #endif 480 481 inittimecounter(); 482 ntp_init(); 483 484 /* Initialize the device switch tables. */ 485 devsw_init(); 486 487 /* Initialize tty subsystem. */ 488 tty_init(); 489 ttyldisc_init(); 490 491 /* Initialize the buffer cache, part 2. */ 492 bufinit2(); 493 494 /* Initialize the disk wedge subsystem. */ 495 dkwedge_init(); 496 497 /* Initialize interfaces. */ 498 ifinit1(); 499 500 /* Configure the system hardware. This will enable interrupts. */ 501 configure(); 502 503 ubc_init(); /* must be after autoconfig */ 504 505 #ifdef SYSVSHM 506 /* Initialize System V style shared memory. */ 507 shminit(); 508 #endif 509 510 #ifdef SYSVSEM 511 /* Initialize System V style semaphores. */ 512 seminit(); 513 #endif 514 515 #ifdef SYSVMSG 516 /* Initialize System V style message queues. */ 517 msginit(); 518 #endif 519 520 #if NVERIEXEC > 0 521 /* 522 * Initialise the Veriexec subsystem. 523 */ 524 veriexec_init(); 525 #endif /* NVERIEXEC > 0 */ 526 527 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 528 pax_init(); 529 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 530 531 #ifdef FAST_IPSEC 532 /* Attach network crypto subsystem */ 533 ipsec_attach(); 534 #endif 535 536 /* 537 * Initialize protocols. Block reception of incoming packets 538 * until everything is ready. 539 */ 540 s = splnet(); 541 ifinit(); 542 domaininit(); 543 if_attachdomain(); 544 splx(s); 545 546 #ifdef GPROF 547 /* Initialize kernel profiling. */ 548 kmstartup(); 549 #endif 550 551 /* Initialize system accounting. */ 552 acct_init(); 553 554 #ifndef PIPE_SOCKETPAIR 555 /* Initialize pipes. */ 556 pipe_init(); 557 #endif 558 559 #ifdef KTRACE 560 /* Initialize ktrace. */ 561 ktrinit(); 562 #endif 563 564 /* Initialize the UUID system calls. */ 565 uuid_init(); 566 567 #ifdef WAPBL 568 /* Initialize write-ahead physical block logging. */ 569 wapbl_init(); 570 #endif 571 572 /* 573 * Create process 1 (init(8)). We do this now, as Unix has 574 * historically had init be process 1, and changing this would 575 * probably upset a lot of people. 576 * 577 * Note that process 1 won't immediately exec init(8), but will 578 * wait for us to inform it that the root file system has been 579 * mounted. 580 */ 581 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 582 panic("fork init"); 583 584 /* 585 * Load any remaining builtin modules, and hand back temporary 586 * storage to the VM system. 587 */ 588 module_init_class(MODULE_CLASS_ANY); 589 590 /* 591 * Finalize configuration now that all real devices have been 592 * found. This needs to be done before the root device is 593 * selected, since finalization may create the root device. 594 */ 595 config_finalize(); 596 597 /* 598 * Now that autoconfiguration has completed, we can determine 599 * the root and dump devices. 600 */ 601 cpu_rootconf(); 602 cpu_dumpconf(); 603 604 /* Mount the root file system. */ 605 do { 606 domountroothook(); 607 if ((error = vfs_mountroot())) { 608 printf("cannot mount root, error = %d\n", error); 609 boothowto |= RB_ASKNAME; 610 setroot(root_device, 611 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 612 } 613 } while (error != 0); 614 mountroothook_destroy(); 615 616 /* 617 * Initialise the time-of-day clock, passing the time recorded 618 * in the root filesystem (if any) for use by systems that 619 * don't have a non-volatile time-of-day device. 620 */ 621 inittodr(rootfstime); 622 623 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 624 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 625 626 /* 627 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 628 * reference it. 629 */ 630 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 631 if (error) 632 panic("cannot find root vnode, error=%d", error); 633 cwdi0.cwdi_cdir = rootvnode; 634 VREF(cwdi0.cwdi_cdir); 635 VOP_UNLOCK(rootvnode, 0); 636 cwdi0.cwdi_rdir = NULL; 637 638 /* 639 * Now that root is mounted, we can fixup initproc's CWD 640 * info. All other processes are kthreads, which merely 641 * share proc0's CWD info. 642 */ 643 initproc->p_cwdi->cwdi_cdir = rootvnode; 644 VREF(initproc->p_cwdi->cwdi_cdir); 645 initproc->p_cwdi->cwdi_rdir = NULL; 646 647 /* 648 * Now can look at time, having had a chance to verify the time 649 * from the file system. Reset l->l_rtime as it may have been 650 * munched in mi_switch() after the time got set. 651 */ 652 getmicrotime(&time); 653 boottime = time; 654 mutex_enter(proc_lock); 655 LIST_FOREACH(p, &allproc, p_list) { 656 KASSERT((p->p_flag & PK_MARKER) == 0); 657 mutex_enter(p->p_lock); 658 p->p_stats->p_start = time; 659 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 660 lwp_lock(l); 661 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 662 lwp_unlock(l); 663 } 664 mutex_exit(p->p_lock); 665 } 666 mutex_exit(proc_lock); 667 binuptime(&curlwp->l_stime); 668 669 for (CPU_INFO_FOREACH(cii, ci)) { 670 ci->ci_schedstate.spc_lastmod = time_second; 671 } 672 673 /* Create the pageout daemon kernel thread. */ 674 uvm_swap_init(); 675 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 676 NULL, NULL, "pgdaemon")) 677 panic("fork pagedaemon"); 678 679 /* Create the filesystem syncer kernel thread. */ 680 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 681 NULL, NULL, "ioflush")) 682 panic("fork syncer"); 683 684 /* Create the aiodone daemon kernel thread. */ 685 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 686 uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE)) 687 panic("fork aiodoned"); 688 689 vmem_rehash_start(); 690 691 /* Initialize exec structures */ 692 exec_init(1); 693 694 /* 695 * Okay, now we can let init(8) exec! It's off to userland! 696 */ 697 mutex_enter(proc_lock); 698 start_init_exec = 1; 699 cv_broadcast(&lbolt); 700 mutex_exit(proc_lock); 701 702 /* The scheduler is an infinite loop. */ 703 uvm_scheduler(); 704 /* NOTREACHED */ 705 } 706 707 static void 708 check_console(struct lwp *l) 709 { 710 struct nameidata nd; 711 int error; 712 713 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console"); 714 error = namei(&nd); 715 if (error == 0) 716 vrele(nd.ni_vp); 717 else if (error == ENOENT) 718 printf("warning: no /dev/console\n"); 719 else 720 printf("warning: lookup /dev/console: error %d\n", error); 721 } 722 723 /* 724 * List of paths to try when searching for "init". 725 */ 726 static const char * const initpaths[] = { 727 "/sbin/init", 728 "/sbin/oinit", 729 "/sbin/init.bak", 730 NULL, 731 }; 732 733 /* 734 * Start the initial user process; try exec'ing each pathname in "initpaths". 735 * The program is invoked with one argument containing the boot flags. 736 */ 737 static void 738 start_init(void *arg) 739 { 740 struct lwp *l = arg; 741 struct proc *p = l->l_proc; 742 vaddr_t addr; 743 struct sys_execve_args /* { 744 syscallarg(const char *) path; 745 syscallarg(char * const *) argp; 746 syscallarg(char * const *) envp; 747 } */ args; 748 int options, i, error; 749 register_t retval[2]; 750 char flags[4], *flagsp; 751 const char *path, *slash; 752 char *ucp, **uap, *arg0, *arg1 = NULL; 753 char ipath[129]; 754 int ipx, len; 755 756 /* 757 * Now in process 1. 758 */ 759 strncpy(p->p_comm, "init", MAXCOMLEN); 760 761 /* 762 * Wait for main() to tell us that it's safe to exec. 763 */ 764 mutex_enter(proc_lock); 765 while (start_init_exec == 0) 766 cv_wait(&lbolt, proc_lock); 767 mutex_exit(proc_lock); 768 769 /* 770 * This is not the right way to do this. We really should 771 * hand-craft a descriptor onto /dev/console to hand to init, 772 * but that's a _lot_ more work, and the benefit from this easy 773 * hack makes up for the "good is the enemy of the best" effect. 774 */ 775 check_console(l); 776 777 /* 778 * Need just enough stack to hold the faked-up "execve()" arguments. 779 */ 780 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 781 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 782 NULL, UVM_UNKNOWN_OFFSET, 0, 783 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 784 UVM_ADV_NORMAL, 785 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 786 panic("init: couldn't allocate argument space"); 787 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 788 789 ipx = 0; 790 while (1) { 791 if (boothowto & RB_ASKNAME) { 792 printf("init path"); 793 if (initpaths[ipx]) 794 printf(" (default %s)", initpaths[ipx]); 795 printf(": "); 796 len = cngetsn(ipath, sizeof(ipath)-1); 797 if (len == 4 && strcmp(ipath, "halt") == 0) { 798 cpu_reboot(RB_HALT, NULL); 799 } else if (len == 6 && strcmp(ipath, "reboot") == 0) { 800 cpu_reboot(0, NULL); 801 #if defined(DDB) 802 } else if (len == 3 && strcmp(ipath, "ddb") == 0) { 803 console_debugger(); 804 continue; 805 #endif 806 } else if (len > 0 && ipath[0] == '/') { 807 ipath[len] = '\0'; 808 path = ipath; 809 } else if (len == 0 && initpaths[ipx] != NULL) { 810 path = initpaths[ipx++]; 811 } else { 812 printf("use absolute path, "); 813 #if defined(DDB) 814 printf("\"ddb\", "); 815 #endif 816 printf("\"halt\", or \"reboot\"\n"); 817 continue; 818 } 819 } else { 820 if ((path = initpaths[ipx++]) == NULL) { 821 ipx = 0; 822 boothowto |= RB_ASKNAME; 823 continue; 824 } 825 } 826 827 ucp = (char *)USRSTACK; 828 829 /* 830 * Construct the boot flag argument. 831 */ 832 flagsp = flags; 833 *flagsp++ = '-'; 834 options = 0; 835 836 if (boothowto & RB_SINGLE) { 837 *flagsp++ = 's'; 838 options = 1; 839 } 840 #ifdef notyet 841 if (boothowto & RB_FASTBOOT) { 842 *flagsp++ = 'f'; 843 options = 1; 844 } 845 #endif 846 847 /* 848 * Move out the flags (arg 1), if necessary. 849 */ 850 if (options != 0) { 851 *flagsp++ = '\0'; 852 i = flagsp - flags; 853 #ifdef DEBUG 854 printf("init: copying out flags `%s' %d\n", flags, i); 855 #endif 856 arg1 = STACK_ALLOC(ucp, i); 857 ucp = STACK_MAX(arg1, i); 858 (void)copyout((void *)flags, arg1, i); 859 } 860 861 /* 862 * Move out the file name (also arg 0). 863 */ 864 i = strlen(path) + 1; 865 #ifdef DEBUG 866 printf("init: copying out path `%s' %d\n", path, i); 867 #else 868 if (boothowto & RB_ASKNAME || path != initpaths[0]) 869 printf("init: trying %s\n", path); 870 #endif 871 arg0 = STACK_ALLOC(ucp, i); 872 ucp = STACK_MAX(arg0, i); 873 (void)copyout(path, arg0, i); 874 875 /* 876 * Move out the arg pointers. 877 */ 878 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 879 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 880 SCARG(&args, path) = arg0; 881 SCARG(&args, argp) = uap; 882 SCARG(&args, envp) = NULL; 883 slash = strrchr(path, '/'); 884 if (slash) 885 (void)suword((void *)uap++, 886 (long)arg0 + (slash + 1 - path)); 887 else 888 (void)suword((void *)uap++, (long)arg0); 889 if (options != 0) 890 (void)suword((void *)uap++, (long)arg1); 891 (void)suword((void *)uap++, 0); /* terminator */ 892 893 /* 894 * Now try to exec the program. If can't for any reason 895 * other than it doesn't exist, complain. 896 */ 897 error = sys_execve(l, &args, retval); 898 if (error == 0 || error == EJUSTRETURN) { 899 KERNEL_UNLOCK_LAST(l); 900 return; 901 } 902 printf("exec %s: error %d\n", path, error); 903 } 904 printf("init: not found\n"); 905 panic("no init"); 906 } 907 908 /* 909 * calculate cache size from physmem and vm_map size. 910 */ 911 vaddr_t 912 calc_cache_size(struct vm_map *map, int pct, int va_pct) 913 { 914 paddr_t t; 915 916 /* XXX should consider competing cache if any */ 917 /* XXX should consider submaps */ 918 t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; 919 if (map != NULL) { 920 vsize_t vsize; 921 922 vsize = vm_map_max(map) - vm_map_min(map); 923 vsize = (uintmax_t)vsize * va_pct / 100; 924 if (t > vsize) { 925 t = vsize; 926 } 927 } 928 return t; 929 } 930
Indexes created Sun Sep 21 20:09:37 GMT 2025