init_main.c revision 1.367
1 /* $NetBSD: init_main.c,v 1.367 2008/10/20 10:24:18 ad 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 1.367 2008/10/20 10:24:18 ad Exp $"); 101 102 #include "opt_ipsec.h" 103 #include "opt_ntp.h" 104 #include "opt_pipe.h" 105 #include "opt_posix.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 #ifdef P1003_1B_SEMAPHORE 180 #include <sys/ksem.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 #ifdef LKM 189 #include <sys/lkm.h> 190 #endif 191 #if NVERIEXEC > 0 192 #include <sys/verified_exec.h> 193 #endif /* NVERIEXEC > 0 */ 194 #ifdef KTRACE 195 #include <sys/ktrace.h> 196 #endif 197 #include <sys/kauth.h> 198 #ifdef WAPBL 199 #include <sys/wapbl.h> 200 #endif 201 #include <net80211/ieee80211_netbsd.h> 202 203 #include <sys/syscall.h> 204 #include <sys/syscallargs.h> 205 206 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 207 #include <sys/pax.h> 208 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 209 210 #include <ufs/ufs/quota.h> 211 212 #include <miscfs/genfs/genfs.h> 213 #include <miscfs/syncfs/syncfs.h> 214 215 #include <sys/cpu.h> 216 217 #include <uvm/uvm.h> 218 219 #if NSYSMON_TASKQ > 0 220 #include <dev/sysmon/sysmon_taskq.h> 221 #endif 222 223 #include <dev/cons.h> 224 225 #if NSYSMON_ENVSYS > 0 || NSYSMON_POWER > 0 || NSYSMON_WDOG > 0 226 #include <dev/sysmon/sysmonvar.h> 227 #endif 228 229 #include <net/if.h> 230 #include <net/raw_cb.h> 231 232 #include <secmodel/secmodel.h> 233 234 extern struct proc proc0; 235 extern struct lwp lwp0; 236 extern struct cwdinfo cwdi0; 237 extern time_t rootfstime; 238 239 #ifndef curlwp 240 struct lwp *curlwp = &lwp0; 241 #endif 242 struct proc *initproc; 243 244 struct vnode *rootvp, *swapdev_vp; 245 int boothowto; 246 int cold = 1; /* still working on startup */ 247 struct timeval boottime; /* time at system startup - will only follow settime deltas */ 248 249 int start_init_exec; /* semaphore for start_init() */ 250 251 static void check_console(struct lwp *l); 252 static void start_init(void *); 253 void main(void); 254 void ssp_init(void); 255 256 #if defined(__SSP__) || defined(__SSP_ALL__) 257 long __stack_chk_guard[8] = {0, 0, 0, 0, 0, 0, 0, 0}; 258 void __stack_chk_fail(void); 259 260 void 261 __stack_chk_fail(void) 262 { 263 panic("stack overflow detected; terminated"); 264 } 265 266 void 267 ssp_init(void) 268 { 269 int s; 270 271 #ifdef DIAGNOSTIC 272 printf("Initializing SSP:"); 273 #endif 274 /* 275 * We initialize ssp here carefully: 276 * 1. after we got some entropy 277 * 2. without calling a function 278 */ 279 size_t i; 280 long guard[__arraycount(__stack_chk_guard)]; 281 282 arc4randbytes(guard, sizeof(guard)); 283 s = splhigh(); 284 for (i = 0; i < __arraycount(guard); i++) 285 __stack_chk_guard[i] = guard[i]; 286 splx(s); 287 #ifdef DIAGNOSTIC 288 for (i = 0; i < __arraycount(guard); i++) 289 printf("%lx ", guard[i]); 290 printf("\n"); 291 #endif 292 } 293 #else 294 void 295 ssp_init(void) 296 { 297 298 } 299 #endif 300 301 void __secmodel_none(void); 302 __weak_alias(secmodel_start,__secmodel_none); 303 void 304 __secmodel_none(void) 305 { 306 return; 307 } 308 309 /* 310 * System startup; initialize the world, create process 0, mount root 311 * filesystem, and fork to create init and pagedaemon. Most of the 312 * hard work is done in the lower-level initialization routines including 313 * startup(), which does memory initialization and autoconfiguration. 314 */ 315 void 316 main(void) 317 { 318 struct timeval time; 319 struct lwp *l; 320 struct proc *p; 321 int s, error; 322 #ifdef NVNODE_IMPLICIT 323 int usevnodes; 324 #endif 325 CPU_INFO_ITERATOR cii; 326 struct cpu_info *ci; 327 328 l = &lwp0; 329 #ifndef LWP0_CPU_INFO 330 l->l_cpu = curcpu(); 331 #endif 332 333 /* 334 * Attempt to find console and initialize 335 * in case of early panic or other messages. 336 */ 337 consinit(); 338 339 kernel_lock_init(); 340 once_init(); 341 342 uvm_init(); 343 344 percpu_init(); 345 346 /* Initialize lock caches. */ 347 mutex_obj_init(); 348 349 /* Initialize the extent manager. */ 350 extent_init(); 351 352 /* Do machine-dependent initialization. */ 353 cpu_startup(); 354 355 /* Initialize callouts, part 1. */ 356 callout_startup(); 357 358 /* 359 * Initialize the kernel authorization subsystem and start the 360 * default security model, if any. We need to do this early 361 * enough so that subsystems relying on any of the aforementioned 362 * can work properly. Since the security model may dictate the 363 * credential inheritance policy, it is needed at least before 364 * any process is created, specifically proc0. 365 */ 366 kauth_init(); 367 secmodel_start(); 368 369 /* Initialize the buffer cache */ 370 bufinit(); 371 372 /* Initialize sockets. */ 373 soinit(); 374 375 /* 376 * The following things must be done before autoconfiguration. 377 */ 378 evcnt_init(); /* initialize event counters */ 379 #if NRND > 0 380 rnd_init(); /* initialize random number generator */ 381 #endif 382 383 /* Initialize process and pgrp structures. */ 384 procinit(); 385 lwpinit(); 386 387 /* Initialize signal-related data structures. */ 388 signal_init(); 389 390 /* Initialize resource management. */ 391 resource_init(); 392 393 /* Create process 0 (the swapper). */ 394 proc0_init(); 395 396 /* Initialize the UID hash table. */ 397 uid_init(); 398 399 /* Charge root for one process. */ 400 (void)chgproccnt(0, 1); 401 402 /* Initialize timekeeping. */ 403 time_init(); 404 405 /* Initialize the run queues, turnstiles and sleep queues. */ 406 mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE); 407 sched_rqinit(); 408 turnstile_init(); 409 sleeptab_init(&sleeptab); 410 411 /* Initialize processor-sets */ 412 psets_init(); 413 414 /* MI initialization of the boot cpu */ 415 error = mi_cpu_attach(curcpu()); 416 KASSERT(error == 0); 417 418 /* Initialize timekeeping, part 2. */ 419 time_init2(); 420 421 /* 422 * Initialize mbuf's. Do this now because we might attempt to 423 * allocate mbufs or mbuf clusters during autoconfiguration. 424 */ 425 mbinit(); 426 427 /* Initialize the sysctl subsystem. */ 428 sysctl_init(); 429 430 /* Initialize I/O statistics. */ 431 iostat_init(); 432 433 /* Initialize the log device. */ 434 loginit(); 435 436 /* Start module system. */ 437 mutex_init(&ksyms_lock, MUTEX_DEFAULT, IPL_NONE); 438 module_init(); 439 440 /* Initialize the file systems. */ 441 #ifdef NVNODE_IMPLICIT 442 /* 443 * If maximum number of vnodes in namei vnode cache is not explicitly 444 * defined in kernel config, adjust the number such as we use roughly 445 * 10% of memory for vnodes and associated data structures in the 446 * assumed worst case. Do not provide fewer than NVNODE vnodes. 447 */ 448 usevnodes = 449 calc_cache_size(kernel_map, 10, VNODE_VA_MAXPCT) / VNODE_COST; 450 if (usevnodes > desiredvnodes) 451 desiredvnodes = usevnodes; 452 #endif 453 vfsinit(); 454 lf_init(); 455 456 /* Initialize fstrans. */ 457 fstrans_init(); 458 459 /* Initialize the file descriptor system. */ 460 fd_sys_init(); 461 462 /* Initialize kqueue. */ 463 kqueue_init(); 464 465 /* Initialize asynchronous I/O. */ 466 aio_sysinit(); 467 468 /* Initialize message queues. */ 469 mqueue_sysinit(); 470 471 /* Initialize the system monitor subsystems. */ 472 #if NSYSMON_TASKQ > 0 473 sysmon_task_queue_preinit(); 474 #endif 475 476 #if NSYSMON_ENVSYS > 0 477 sysmon_envsys_init(); 478 #endif 479 480 #if NSYSMON_POWER > 0 481 sysmon_power_init(); 482 #endif 483 484 #if NSYSMON_WDOG > 0 485 sysmon_wdog_init(); 486 #endif 487 488 inittimecounter(); 489 ntp_init(); 490 491 /* Initialize the device switch tables. */ 492 devsw_init(); 493 494 /* Initialize tty subsystem. */ 495 tty_init(); 496 ttyldisc_init(); 497 498 /* Initialize the buffer cache, part 2. */ 499 bufinit2(); 500 501 /* Initialize the disk wedge subsystem. */ 502 dkwedge_init(); 503 504 /* Initialize interfaces. */ 505 ifinit1(); 506 507 /* Configure the system hardware. This will enable interrupts. */ 508 configure(); 509 510 ubc_init(); /* must be after autoconfig */ 511 512 #ifdef SYSVSHM 513 /* Initialize System V style shared memory. */ 514 shminit(); 515 #endif 516 517 #ifdef SYSVSEM 518 /* Initialize System V style semaphores. */ 519 seminit(); 520 #endif 521 522 #ifdef SYSVMSG 523 /* Initialize System V style message queues. */ 524 msginit(); 525 #endif 526 527 #ifdef P1003_1B_SEMAPHORE 528 /* Initialize posix semaphores */ 529 ksem_init(); 530 #endif 531 532 #if NVERIEXEC > 0 533 /* 534 * Initialise the Veriexec subsystem. 535 */ 536 veriexec_init(); 537 #endif /* NVERIEXEC > 0 */ 538 539 #if defined(PAX_MPROTECT) || defined(PAX_SEGVGUARD) || defined(PAX_ASLR) 540 pax_init(); 541 #endif /* PAX_MPROTECT || PAX_SEGVGUARD || PAX_ASLR */ 542 543 #ifdef FAST_IPSEC 544 /* Attach network crypto subsystem */ 545 ipsec_attach(); 546 #endif 547 548 /* 549 * Initialize protocols. Block reception of incoming packets 550 * until everything is ready. 551 */ 552 s = splnet(); 553 ifinit(); 554 domaininit(); 555 if_attachdomain(); 556 splx(s); 557 558 #ifdef GPROF 559 /* Initialize kernel profiling. */ 560 kmstartup(); 561 #endif 562 563 /* Initialize system accounting. */ 564 acct_init(); 565 566 #ifndef PIPE_SOCKETPAIR 567 /* Initialize pipes. */ 568 pipe_init(); 569 #endif 570 571 #ifdef KTRACE 572 /* Initialize ktrace. */ 573 ktrinit(); 574 #endif 575 576 /* Initialize the UUID system calls. */ 577 uuid_init(); 578 579 #ifdef WAPBL 580 /* Initialize write-ahead physical block logging. */ 581 wapbl_init(); 582 #endif 583 584 /* 585 * Create process 1 (init(8)). We do this now, as Unix has 586 * historically had init be process 1, and changing this would 587 * probably upset a lot of people. 588 * 589 * Note that process 1 won't immediately exec init(8), but will 590 * wait for us to inform it that the root file system has been 591 * mounted. 592 */ 593 if (fork1(l, 0, SIGCHLD, NULL, 0, start_init, NULL, NULL, &initproc)) 594 panic("fork init"); 595 596 /* 597 * Load any remaining builtin modules, and hand back temporary 598 * storage to the VM system. 599 */ 600 module_init_class(MODULE_CLASS_ANY); 601 602 /* 603 * Finalize configuration now that all real devices have been 604 * found. This needs to be done before the root device is 605 * selected, since finalization may create the root device. 606 */ 607 config_finalize(); 608 609 /* 610 * Now that autoconfiguration has completed, we can determine 611 * the root and dump devices. 612 */ 613 cpu_rootconf(); 614 cpu_dumpconf(); 615 616 /* Mount the root file system. */ 617 do { 618 domountroothook(); 619 if ((error = vfs_mountroot())) { 620 printf("cannot mount root, error = %d\n", error); 621 boothowto |= RB_ASKNAME; 622 setroot(root_device, 623 (rootdev != NODEV) ? DISKPART(rootdev) : 0); 624 } 625 } while (error != 0); 626 mountroothook_destroy(); 627 628 /* 629 * Initialise the time-of-day clock, passing the time recorded 630 * in the root filesystem (if any) for use by systems that 631 * don't have a non-volatile time-of-day device. 632 */ 633 inittodr(rootfstime); 634 635 CIRCLEQ_FIRST(&mountlist)->mnt_flag |= MNT_ROOTFS; 636 CIRCLEQ_FIRST(&mountlist)->mnt_op->vfs_refcount++; 637 638 /* 639 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to 640 * reference it. 641 */ 642 error = VFS_ROOT(CIRCLEQ_FIRST(&mountlist), &rootvnode); 643 if (error) 644 panic("cannot find root vnode, error=%d", error); 645 cwdi0.cwdi_cdir = rootvnode; 646 VREF(cwdi0.cwdi_cdir); 647 VOP_UNLOCK(rootvnode, 0); 648 cwdi0.cwdi_rdir = NULL; 649 650 /* 651 * Now that root is mounted, we can fixup initproc's CWD 652 * info. All other processes are kthreads, which merely 653 * share proc0's CWD info. 654 */ 655 initproc->p_cwdi->cwdi_cdir = rootvnode; 656 VREF(initproc->p_cwdi->cwdi_cdir); 657 initproc->p_cwdi->cwdi_rdir = NULL; 658 659 /* 660 * Now can look at time, having had a chance to verify the time 661 * from the file system. Reset l->l_rtime as it may have been 662 * munched in mi_switch() after the time got set. 663 */ 664 getmicrotime(&time); 665 boottime = time; 666 mutex_enter(proc_lock); 667 LIST_FOREACH(p, &allproc, p_list) { 668 KASSERT((p->p_flag & PK_MARKER) == 0); 669 mutex_enter(p->p_lock); 670 p->p_stats->p_start = time; 671 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 672 lwp_lock(l); 673 memset(&l->l_rtime, 0, sizeof(l->l_rtime)); 674 lwp_unlock(l); 675 } 676 mutex_exit(p->p_lock); 677 } 678 mutex_exit(proc_lock); 679 binuptime(&curlwp->l_stime); 680 681 for (CPU_INFO_FOREACH(cii, ci)) { 682 ci->ci_schedstate.spc_lastmod = time_second; 683 } 684 685 /* Create the pageout daemon kernel thread. */ 686 uvm_swap_init(); 687 if (kthread_create(PRI_PGDAEMON, KTHREAD_MPSAFE, NULL, uvm_pageout, 688 NULL, NULL, "pgdaemon")) 689 panic("fork pagedaemon"); 690 691 /* Create the filesystem syncer kernel thread. */ 692 if (kthread_create(PRI_IOFLUSH, KTHREAD_MPSAFE, NULL, sched_sync, 693 NULL, NULL, "ioflush")) 694 panic("fork syncer"); 695 696 /* Create the aiodone daemon kernel thread. */ 697 if (workqueue_create(&uvm.aiodone_queue, "aiodoned", 698 uvm_aiodone_worker, NULL, PRI_VM, IPL_NONE, WQ_MPSAFE)) 699 panic("fork aiodoned"); 700 701 vmem_rehash_start(); 702 703 /* Initialize exec structures */ 704 exec_init(1); 705 706 /* 707 * Okay, now we can let init(8) exec! It's off to userland! 708 */ 709 mutex_enter(proc_lock); 710 start_init_exec = 1; 711 cv_broadcast(&lbolt); 712 mutex_exit(proc_lock); 713 714 /* The scheduler is an infinite loop. */ 715 uvm_scheduler(); 716 /* NOTREACHED */ 717 } 718 719 static void 720 check_console(struct lwp *l) 721 { 722 struct nameidata nd; 723 int error; 724 725 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console"); 726 error = namei(&nd); 727 if (error == 0) 728 vrele(nd.ni_vp); 729 else if (error == ENOENT) 730 printf("warning: no /dev/console\n"); 731 else 732 printf("warning: lookup /dev/console: error %d\n", error); 733 } 734 735 /* 736 * List of paths to try when searching for "init". 737 */ 738 static const char * const initpaths[] = { 739 "/sbin/init", 740 "/sbin/oinit", 741 "/sbin/init.bak", 742 NULL, 743 }; 744 745 /* 746 * Start the initial user process; try exec'ing each pathname in "initpaths". 747 * The program is invoked with one argument containing the boot flags. 748 */ 749 static void 750 start_init(void *arg) 751 { 752 struct lwp *l = arg; 753 struct proc *p = l->l_proc; 754 vaddr_t addr; 755 struct sys_execve_args /* { 756 syscallarg(const char *) path; 757 syscallarg(char * const *) argp; 758 syscallarg(char * const *) envp; 759 } */ args; 760 int options, i, error; 761 register_t retval[2]; 762 char flags[4], *flagsp; 763 const char *path, *slash; 764 char *ucp, **uap, *arg0, *arg1 = NULL; 765 char ipath[129]; 766 int ipx, len; 767 768 /* 769 * Now in process 1. 770 */ 771 strncpy(p->p_comm, "init", MAXCOMLEN); 772 773 /* 774 * Wait for main() to tell us that it's safe to exec. 775 */ 776 mutex_enter(proc_lock); 777 while (start_init_exec == 0) 778 cv_wait(&lbolt, proc_lock); 779 mutex_exit(proc_lock); 780 781 /* 782 * This is not the right way to do this. We really should 783 * hand-craft a descriptor onto /dev/console to hand to init, 784 * but that's a _lot_ more work, and the benefit from this easy 785 * hack makes up for the "good is the enemy of the best" effect. 786 */ 787 check_console(l); 788 789 /* 790 * Need just enough stack to hold the faked-up "execve()" arguments. 791 */ 792 addr = (vaddr_t)STACK_ALLOC(USRSTACK, PAGE_SIZE); 793 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE, 794 NULL, UVM_UNKNOWN_OFFSET, 0, 795 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY, 796 UVM_ADV_NORMAL, 797 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW)) != 0) 798 panic("init: couldn't allocate argument space"); 799 p->p_vmspace->vm_maxsaddr = (void *)STACK_MAX(addr, PAGE_SIZE); 800 801 ipx = 0; 802 while (1) { 803 if (boothowto & RB_ASKNAME) { 804 printf("init path"); 805 if (initpaths[ipx]) 806 printf(" (default %s)", initpaths[ipx]); 807 printf(": "); 808 len = cngetsn(ipath, sizeof(ipath)-1); 809 if (len == 0) { 810 if (initpaths[ipx]) 811 path = initpaths[ipx++]; 812 else 813 continue; 814 } else { 815 ipath[len] = '\0'; 816 path = ipath; 817 } 818 } else { 819 if ((path = initpaths[ipx++]) == NULL) 820 break; 821 } 822 823 ucp = (char *)USRSTACK; 824 825 /* 826 * Construct the boot flag argument. 827 */ 828 flagsp = flags; 829 *flagsp++ = '-'; 830 options = 0; 831 832 if (boothowto & RB_SINGLE) { 833 *flagsp++ = 's'; 834 options = 1; 835 } 836 #ifdef notyet 837 if (boothowto & RB_FASTBOOT) { 838 *flagsp++ = 'f'; 839 options = 1; 840 } 841 #endif 842 843 /* 844 * Move out the flags (arg 1), if necessary. 845 */ 846 if (options != 0) { 847 *flagsp++ = '\0'; 848 i = flagsp - flags; 849 #ifdef DEBUG 850 printf("init: copying out flags `%s' %d\n", flags, i); 851 #endif 852 arg1 = STACK_ALLOC(ucp, i); 853 ucp = STACK_MAX(arg1, i); 854 (void)copyout((void *)flags, arg1, i); 855 } 856 857 /* 858 * Move out the file name (also arg 0). 859 */ 860 i = strlen(path) + 1; 861 #ifdef DEBUG 862 printf("init: copying out path `%s' %d\n", path, i); 863 #else 864 if (boothowto & RB_ASKNAME || path != initpaths[0]) 865 printf("init: trying %s\n", path); 866 #endif 867 arg0 = STACK_ALLOC(ucp, i); 868 ucp = STACK_MAX(arg0, i); 869 (void)copyout(path, arg0, i); 870 871 /* 872 * Move out the arg pointers. 873 */ 874 ucp = (void *)STACK_ALIGN(ucp, ALIGNBYTES); 875 uap = (char **)STACK_ALLOC(ucp, sizeof(char *) * 3); 876 SCARG(&args, path) = arg0; 877 SCARG(&args, argp) = uap; 878 SCARG(&args, envp) = NULL; 879 slash = strrchr(path, '/'); 880 if (slash) 881 (void)suword((void *)uap++, 882 (long)arg0 + (slash + 1 - path)); 883 else 884 (void)suword((void *)uap++, (long)arg0); 885 if (options != 0) 886 (void)suword((void *)uap++, (long)arg1); 887 (void)suword((void *)uap++, 0); /* terminator */ 888 889 /* 890 * Now try to exec the program. If can't for any reason 891 * other than it doesn't exist, complain. 892 */ 893 error = sys_execve(l, &args, retval); 894 if (error == 0 || error == EJUSTRETURN) { 895 KERNEL_UNLOCK_LAST(l); 896 return; 897 } 898 printf("exec %s: error %d\n", path, error); 899 } 900 printf("init: not found\n"); 901 panic("no init"); 902 } 903 904 /* 905 * calculate cache size from physmem and vm_map size. 906 */ 907 vaddr_t 908 calc_cache_size(struct vm_map *map, int pct, int va_pct) 909 { 910 paddr_t t; 911 912 /* XXX should consider competing cache if any */ 913 /* XXX should consider submaps */ 914 t = (uintmax_t)physmem * pct / 100 * PAGE_SIZE; 915 if (map != NULL) { 916 vsize_t vsize; 917 918 vsize = vm_map_max(map) - vm_map_min(map); 919 vsize = (uintmax_t)vsize * va_pct / 100; 920 if (t > vsize) { 921 t = vsize; 922 } 923 } 924 return t; 925 } 926
Indexes created Mon Sep 22 05:09:51 GMT 2025