1 /* 2 * Copyright (c) 1995 Terrence R. Lambert 3 * All rights reserved. 4 * 5 * Copyright (c) 1982, 1986, 1989, 1991, 1992, 1993 6 * The Regents of the University of California. All rights reserved. 7 * (c) UNIX System Laboratories, Inc. 8 * All or some portions of this file are derived from material licensed 9 * to the University of California by American Telephone and Telegraph 10 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 11 * the permission of UNIX System Laboratories, Inc. 12 * 13 * Redistribution and use in source and binary forms, with or without 14 * modification, are permitted provided that the following conditions 15 * are met: 16 * 1. Redistributions of source code must retain the above copyright 17 * notice, this list of conditions and the following disclaimer. 18 * 2. Redistributions in binary form must reproduce the above copyright 19 * notice, this list of conditions and the following disclaimer in the 20 * documentation and/or other materials provided with the distribution. 21 * 3. Neither the name of the University nor the names of its contributors 22 * may be used to endorse or promote products derived from this software 23 * without specific prior written permission. 24 * 25 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 26 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 27 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 28 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 29 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 30 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 31 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 32 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 33 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 34 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 35 * SUCH DAMAGE. 36 * 37 * @(#)init_main.c 8.9 (Berkeley) 1/21/94 38 * $FreeBSD: src/sys/kern/init_main.c,v 1.134.2.8 2003/06/06 20:21:32 tegge Exp $ 39 */ 40 41 #include "opt_init_path.h" 42 43 #include <sys/param.h> 44 #include <sys/file.h> 45 #include <sys/filedesc.h> 46 #include <sys/kernel.h> 47 #include <sys/mount.h> 48 #include <sys/sysctl.h> 49 #include <sys/proc.h> 50 #include <sys/resourcevar.h> 51 #include <sys/signalvar.h> 52 #include <sys/systm.h> 53 #include <sys/vnode.h> 54 #include <sys/sysent.h> 55 #include <sys/reboot.h> 56 #include <sys/sysproto.h> 57 #include <sys/vmmeter.h> 58 #include <sys/unistd.h> 59 #include <sys/malloc.h> 60 #include <sys/machintr.h> 61 62 #include <sys/refcount.h> 63 #include <sys/file2.h> 64 #include <sys/thread2.h> 65 #include <sys/sysref2.h> 66 #include <sys/spinlock2.h> 67 68 #include <machine/cpu.h> 69 70 #include <vm/vm.h> 71 #include <vm/vm_param.h> 72 #include <sys/lock.h> 73 #include <vm/pmap.h> 74 #include <vm/vm_map.h> 75 #include <vm/vm_extern.h> 76 #include <sys/user.h> 77 #include <sys/copyright.h> 78 79 int vfs_mountroot_devfs(void); 80 81 /* Components of the first process -- never freed. */ 82 static struct session session0; 83 static struct pgrp pgrp0; 84 static struct sigacts sigacts0; 85 static struct filedesc filedesc0; 86 static struct plimit limit0; 87 static struct vmspace vmspace0; 88 struct proc *initproc; 89 struct proc proc0; 90 struct lwp lwp0; 91 struct thread thread0; 92 struct sys_kpmap *kpmap; 93 struct sysreaper initreaper; 94 95 int cmask = CMASK; 96 u_int cpu_mi_feature; 97 cpumask_t usched_global_cpumask; 98 extern struct user *proc0paddr; 99 100 int boothowto = 0; /* initialized so that it can be patched */ 101 SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, 102 "Reboot flags, from console subsystem"); 103 SYSCTL_OPAQUE(_kern, OID_AUTO, usched_global_cpumask, CTLFLAG_RW, 104 &usched_global_cpumask, sizeof(usched_global_cpumask), "LU", 105 "global user scheduler cpumask"); 106 107 /* 108 * This ensures that there is at least one entry so that the sysinit_set 109 * symbol is not undefined. A subsystem ID of SI_SPECIAL_DUMMY is never 110 * executed. 111 */ 112 SYSINIT(placeholder, SI_SPECIAL_DUMMY, SI_ORDER_ANY, NULL, NULL); 113 114 /* 115 * The sysinit table itself. Items are checked off as the are run. 116 * If we want to register new sysinit types, add them to newsysinit. 117 */ 118 SET_DECLARE(sysinit_set, struct sysinit); 119 struct sysinit **sysinit, **sysinit_end; 120 struct sysinit **newsysinit, **newsysinit_end; 121 122 123 /* 124 * Merge a new sysinit set into the current set, reallocating it if 125 * necessary. This can only be called after malloc is running. 126 */ 127 void 128 sysinit_add(struct sysinit **set, struct sysinit **set_end) 129 { 130 struct sysinit **newset; 131 struct sysinit **sipp; 132 struct sysinit **xipp; 133 int count; 134 135 count = set_end - set; 136 if (newsysinit) 137 count += newsysinit_end - newsysinit; 138 else 139 count += sysinit_end - sysinit; 140 newset = kmalloc(count * sizeof(*sipp), M_TEMP, M_WAITOK); 141 xipp = newset; 142 if (newsysinit) { 143 for (sipp = newsysinit; sipp < newsysinit_end; sipp++) 144 *xipp++ = *sipp; 145 } else { 146 for (sipp = sysinit; sipp < sysinit_end; sipp++) 147 *xipp++ = *sipp; 148 } 149 for (sipp = set; sipp < set_end; sipp++) 150 *xipp++ = *sipp; 151 if (newsysinit) 152 kfree(newsysinit, M_TEMP); 153 newsysinit = newset; 154 newsysinit_end = newset + count; 155 } 156 157 /* 158 * Callbacks from machine-dependant startup code (e.g. init386) to set 159 * up low level entities related to cpu #0's globaldata. 160 * 161 * Called from very low level boot code. 162 */ 163 void 164 mi_proc0init(struct globaldata *gd, struct user *proc0paddr) 165 { 166 lwkt_init_thread(&thread0, proc0paddr, LWKT_THREAD_STACK, 0, gd); 167 lwkt_set_comm(&thread0, "thread0"); 168 RB_INIT(&proc0.p_lwp_tree); 169 spin_init(&proc0.p_spin, "iproc_proc0"); 170 lwkt_token_init(&proc0.p_token, "iproc"); 171 proc0.p_lasttid = 0; /* +1 = next TID */ 172 lwp_rb_tree_RB_INSERT(&proc0.p_lwp_tree, &lwp0); 173 lwp0.lwp_thread = &thread0; 174 lwp0.lwp_proc = &proc0; 175 proc0.p_usched = usched_init(); 176 CPUMASK_ASSALLONES(lwp0.lwp_cpumask); 177 lwkt_token_init(&lwp0.lwp_token, "lwp_token"); 178 spin_init(&lwp0.lwp_spin, "iproc_lwp0"); 179 varsymset_init(&proc0.p_varsymset, NULL); 180 thread0.td_flags |= TDF_RUNNING; 181 thread0.td_proc = &proc0; 182 thread0.td_lwp = &lwp0; 183 thread0.td_switch = cpu_lwkt_switch; 184 lwkt_schedule_self(curthread); 185 } 186 187 /* 188 * System startup; initialize the world, create process 0, mount root 189 * filesystem, and fork to create init and pagedaemon. Most of the 190 * hard work is done in the lower-level initialization routines including 191 * startup(), which does memory initialization and autoconfiguration. 192 * 193 * This allows simple addition of new kernel subsystems that require 194 * boot time initialization. It also allows substitution of subsystem 195 * (for instance, a scheduler, kernel profiler, or VM system) by object 196 * module. Finally, it allows for optional "kernel threads". 197 */ 198 void 199 mi_startup(void) 200 { 201 struct sysinit *sip; /* system initialization*/ 202 struct sysinit **sipp; /* system initialization*/ 203 struct sysinit **xipp; /* interior loop of sort*/ 204 struct sysinit *save; /* bubble*/ 205 206 if (sysinit == NULL) { 207 sysinit = SET_BEGIN(sysinit_set); 208 #if defined(__x86_64__) && defined(_KERNEL_VIRTUAL) 209 /* 210 * XXX For whatever reason, on 64-bit vkernels 211 * the value of sysinit obtained from the 212 * linker set is wrong. 213 */ 214 if ((long)sysinit % 8 != 0) { 215 kprintf("Fixing sysinit value...\n"); 216 sysinit = (void *)((long)(intptr_t)sysinit + 4); 217 } 218 #endif 219 sysinit_end = SET_LIMIT(sysinit_set); 220 } 221 #if defined(__x86_64__) && defined(_KERNEL_VIRTUAL) 222 KKASSERT((long)sysinit % 8 == 0); 223 #endif 224 225 restart: 226 /* 227 * Perform a bubble sort of the system initialization objects by 228 * their subsystem (primary key) and order (secondary key). 229 */ 230 for (sipp = sysinit; sipp < sysinit_end; sipp++) { 231 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) { 232 if ((*sipp)->subsystem < (*xipp)->subsystem || 233 ((*sipp)->subsystem == (*xipp)->subsystem && 234 (*sipp)->order <= (*xipp)->order)) 235 continue; /* skip*/ 236 save = *sipp; 237 *sipp = *xipp; 238 *xipp = save; 239 } 240 } 241 242 /* 243 * Traverse the (now) ordered list of system initialization tasks. 244 * Perform each task, and continue on to the next task. 245 * 246 * The last item on the list is expected to be the scheduler, 247 * which will not return. 248 */ 249 for (sipp = sysinit; sipp < sysinit_end; sipp++) { 250 sip = *sipp; 251 if (sip->subsystem == SI_SPECIAL_DUMMY) 252 continue; /* skip dummy task(s)*/ 253 254 if (sip->subsystem == SI_SPECIAL_DONE) 255 continue; 256 257 #if 0 258 if (bootverbose) 259 kprintf("(%08x-%p)\n", sip->subsystem, sip->func); 260 #endif 261 262 /* Call function */ 263 (*(sip->func))(sip->udata); 264 265 /* Check off the one we're just done */ 266 sip->subsystem = SI_SPECIAL_DONE; 267 268 /* Check if we've installed more sysinit items via KLD */ 269 if (newsysinit != NULL) { 270 if (sysinit != SET_BEGIN(sysinit_set)) 271 kfree(sysinit, M_TEMP); 272 sysinit = newsysinit; 273 sysinit_end = newsysinit_end; 274 newsysinit = NULL; 275 newsysinit_end = NULL; 276 goto restart; 277 } 278 } 279 280 panic("Shouldn't get here!"); 281 /* NOTREACHED*/ 282 } 283 284 285 /* 286 *************************************************************************** 287 **** 288 **** The following SYSINIT's belong elsewhere, but have not yet 289 **** been moved. 290 **** 291 *************************************************************************** 292 */ 293 static void 294 print_caddr_t(void *data) 295 { 296 kprintf("%s", (char *)data); 297 } 298 SYSINIT(announce, SI_BOOT1_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright); 299 300 /* 301 * Leave the critical section that protected us from spurious interrupts 302 * so device probes work. 303 */ 304 static void 305 leavecrit(void *dummy __unused) 306 { 307 MachIntrABI.stabilize(); 308 cpu_enable_intr(); 309 MachIntrABI.cleanup(); 310 crit_exit(); 311 KKASSERT(!IN_CRITICAL_SECT(curthread)); 312 313 if (bootverbose) 314 kprintf("Leaving critical section, allowing interrupts\n"); 315 } 316 SYSINIT(leavecrit, SI_BOOT2_LEAVE_CRIT, SI_ORDER_ANY, leavecrit, NULL); 317 318 /* 319 * This is called after the threading system is up and running, 320 * including the softclock, clock interrupts, and SMP. 321 */ 322 static void 323 tsleepworks(void *dummy __unused) 324 { 325 tsleep_now_works = 1; 326 } 327 SYSINIT(tsleepworks, SI_BOOT2_FINISH_SMP, SI_ORDER_SECOND, tsleepworks, NULL); 328 329 /* 330 * This is called after devices have configured. Tell the kernel we are 331 * no longer in cold boot. 332 */ 333 static void 334 endofcoldboot(void *dummy __unused) 335 { 336 cold = 0; 337 } 338 SYSINIT(endofcoldboot, SI_SUB_ISWARM, SI_ORDER_ANY, endofcoldboot, NULL); 339 340 /* 341 *************************************************************************** 342 **** 343 **** The two following SYSINT's are proc0 specific glue code. I am not 344 **** convinced that they can not be safely combined, but their order of 345 **** operation has been maintained as the same as the original init_main.c 346 **** for right now. 347 **** 348 **** These probably belong in init_proc.c or kern_proc.c, since they 349 **** deal with proc0 (the fork template process). 350 **** 351 *************************************************************************** 352 */ 353 /* ARGSUSED*/ 354 static void 355 proc0_init(void *dummy __unused) 356 { 357 struct proc *p; 358 struct lwp *lp; 359 360 p = &proc0; 361 lp = &lwp0; 362 363 /* 364 * Initialize osrel 365 */ 366 p->p_osrel = osreldate; 367 368 /* 369 * Initialize process and pgrp structures. 370 */ 371 procinit(); 372 373 /* 374 * additional VM structures 375 */ 376 vm_init2(); 377 378 /* 379 * Create process 0 (the swapper). 380 */ 381 procinsertinit(p); 382 pgrpinsertinit(&pgrp0); 383 LIST_INIT(&pgrp0.pg_members); 384 lwkt_token_init(&pgrp0.pg_token, "pgrp0"); 385 refcount_init(&pgrp0.pg_refs, 1); 386 lockinit(&pgrp0.pg_lock, "pgwt0", 0, 0); 387 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist); 388 389 pgrp0.pg_session = &session0; 390 session0.s_count = 1; 391 session0.s_leader = p; 392 sessinsertinit(&session0); 393 394 pgref(&pgrp0); 395 p->p_pgrp = &pgrp0; 396 397 p->p_sysent = &aout_sysvec; 398 399 p->p_flags = P_SYSTEM; 400 p->p_stat = SACTIVE; 401 lp->lwp_stat = LSRUN; 402 p->p_nice = NZERO; 403 p->p_rtprio.type = RTP_PRIO_NORMAL; 404 p->p_rtprio.prio = 0; 405 lp->lwp_rtprio = p->p_rtprio; 406 407 p->p_peers = NULL; 408 p->p_leader = p; 409 410 bcopy("swapper", p->p_comm, sizeof ("swapper")); 411 bcopy("swapper", thread0.td_comm, sizeof ("swapper")); 412 413 /* Create credentials. */ 414 p->p_ucred = crget(); 415 p->p_ucred->cr_ruidinfo = uifind(0); 416 p->p_ucred->cr_ngroups = 1; /* group 0 */ 417 p->p_ucred->cr_uidinfo = uifind(0); 418 thread0.td_ucred = crhold(p->p_ucred); /* bootstrap fork1() */ 419 420 /* Don't jail it */ 421 p->p_ucred->cr_prison = NULL; 422 423 /* Create sigacts. */ 424 p->p_sigacts = &sigacts0; 425 refcount_init(&p->p_sigacts->ps_refcnt, 1); 426 427 /* Initialize signal state for process 0. */ 428 siginit(p); 429 430 /* Create the file descriptor table. */ 431 fdinit_bootstrap(p, &filedesc0, cmask); 432 433 /* Create the limits structures. */ 434 plimit_init0(&limit0); 435 p->p_limit = &limit0; 436 437 /* Allocate a prototype map so we have something to fork. */ 438 pmap_pinit0(vmspace_pmap(&vmspace0)); 439 p->p_vmspace = &vmspace0; 440 lp->lwp_vmspace = p->p_vmspace; 441 vmspace_initrefs(&vmspace0); 442 vm_map_init(&vmspace0.vm_map, 443 round_page(VM_MIN_USER_ADDRESS), 444 trunc_page(VM_MAX_USER_ADDRESS), 445 vmspace_pmap(&vmspace0)); 446 447 kqueue_init(&lwp0.lwp_kqueue, &filedesc0); 448 449 /* 450 * Charge root for one process. 451 */ 452 (void)chgproccnt(p->p_ucred->cr_uidinfo, 1, 0); 453 vm_init_limits(p); 454 } 455 SYSINIT(p0init, SI_BOOT2_PROC0, SI_ORDER_FIRST, proc0_init, NULL); 456 457 static int proc0_post_callback(struct proc *p, void *data __unused); 458 459 /* ARGSUSED*/ 460 static void 461 proc0_post(void *dummy __unused) 462 { 463 struct timespec ts; 464 465 /* 466 * Now we can look at the time, having had a chance to verify the 467 * time from the file system. Pretend that proc0 started now. 468 */ 469 allproc_scan(proc0_post_callback, NULL); 470 471 /* 472 * Give the ``random'' number generator a thump. 473 * XXX: Does read_random() contain enough bits to be used here ? 474 */ 475 nanotime(&ts); 476 skrandom(ts.tv_sec ^ ts.tv_nsec); 477 } 478 479 static int 480 proc0_post_callback(struct proc *p, void *data __unused) 481 { 482 microtime(&p->p_start); 483 return(0); 484 } 485 486 SYSINIT(p0post, SI_SUB_PROC0_POST, SI_ORDER_FIRST, proc0_post, NULL); 487 488 /* 489 *************************************************************************** 490 **** 491 **** The following SYSINIT's and glue code should be moved to the 492 **** respective files on a per subsystem basis. 493 **** 494 *************************************************************************** 495 */ 496 497 498 /* 499 *************************************************************************** 500 **** 501 **** The following code probably belongs in another file, like 502 **** kern/init_init.c. 503 **** 504 *************************************************************************** 505 */ 506 507 /* 508 * List of paths to try when searching for "init". 509 */ 510 static char init_path[MAXPATHLEN] = 511 #ifdef INIT_PATH 512 __XSTRING(INIT_PATH); 513 #else 514 "/sbin/init:/sbin/oinit:/sbin/init.bak"; 515 #endif 516 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, ""); 517 518 /* 519 * Shutdown timeout of init(8). 520 * Unused within kernel, but used to control init(8), hence do not remove. 521 */ 522 #ifndef INIT_SHUTDOWN_TIMEOUT 523 #define INIT_SHUTDOWN_TIMEOUT 120 524 #endif 525 static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT; 526 SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout, 527 CTLFLAG_RW, &init_shutdown_timeout, 0, "Shutdown timeout of init(8). " 528 "Unused within kernel, but used to control init(8)"); 529 530 /* 531 * Start the initial user process; try exec'ing each pathname in init_path. 532 * The program is invoked with one argument containing the boot flags. 533 */ 534 static void 535 start_init(void *dummy, struct trapframe *frame) 536 { 537 vm_offset_t addr; 538 struct execve_args args; 539 int options, error; 540 char *var, *path, *next, *s; 541 char *ucp, **uap, *arg0, *arg1; 542 struct proc *p; 543 struct lwp *lp; 544 struct mount *mp; 545 struct vnode *vp; 546 char *env; 547 548 /* 549 * This is passed in by the bootloader 550 */ 551 env = kgetenv("kernelname"); 552 if (env != NULL) 553 strlcpy(kernelname, env, sizeof(kernelname)); 554 555 p = curproc; 556 557 lp = ONLY_LWP_IN_PROC(p); 558 559 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */ 560 mp = mountlist_boot_getfirst(); 561 if (VFS_ROOT(mp, &vp)) 562 panic("cannot find root vnode"); 563 if (mp->mnt_ncmountpt.ncp == NULL) { 564 cache_allocroot(&mp->mnt_ncmountpt, mp, vp); 565 cache_unlock(&mp->mnt_ncmountpt); /* leave ref intact */ 566 } 567 p->p_fd->fd_cdir = vp; 568 vref(p->p_fd->fd_cdir); 569 p->p_fd->fd_rdir = vp; 570 vref(p->p_fd->fd_rdir); 571 vfs_cache_setroot(vp, cache_hold(&mp->mnt_ncmountpt)); 572 vn_unlock(vp); /* leave ref intact */ 573 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_ncdir); 574 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_nrdir); 575 576 kprintf("Mounting devfs\n"); 577 vfs_mountroot_devfs(); 578 579 /* 580 * Need just enough stack to hold the faked-up "execve()" arguments. 581 */ 582 addr = trunc_page(USRSTACK - PAGE_SIZE); 583 error = vm_map_find(&p->p_vmspace->vm_map, NULL, NULL, 584 0, &addr, PAGE_SIZE, 585 PAGE_SIZE, FALSE, 586 VM_MAPTYPE_NORMAL, VM_SUBSYS_INIT, 587 VM_PROT_ALL, VM_PROT_ALL, 0); 588 if (error) 589 panic("init: couldn't allocate argument space"); 590 p->p_vmspace->vm_maxsaddr = (caddr_t)addr; 591 p->p_vmspace->vm_ssize = 1; 592 593 if ((var = kgetenv("init_path")) != NULL) { 594 strncpy(init_path, var, sizeof init_path); 595 init_path[sizeof init_path - 1] = 0; 596 } 597 598 for (path = init_path; *path != '\0'; path = next) { 599 while (*path == ':') 600 path++; 601 if (*path == '\0') 602 break; 603 for (next = path; *next != '\0' && *next != ':'; next++) 604 /* nothing */ ; 605 if (bootverbose) 606 kprintf("start_init: trying %.*s\n", (int)(next - path), 607 path); 608 609 /* 610 * Move out the boot flag argument. 611 */ 612 options = 0; 613 ucp = (char *)USRSTACK; 614 (void)subyte(--ucp, 0); /* trailing zero */ 615 if (boothowto & RB_SINGLE) { 616 (void)subyte(--ucp, 's'); 617 options = 1; 618 } 619 #ifdef notyet 620 if (boothowto & RB_FASTBOOT) { 621 (void)subyte(--ucp, 'f'); 622 options = 1; 623 } 624 #endif 625 626 #ifdef BOOTCDROM 627 (void)subyte(--ucp, 'C'); 628 options = 1; 629 #endif 630 if (options == 0) 631 (void)subyte(--ucp, '-'); 632 (void)subyte(--ucp, '-'); /* leading hyphen */ 633 arg1 = ucp; 634 635 /* 636 * Move out the file name (also arg 0). 637 */ 638 (void)subyte(--ucp, 0); 639 for (s = next - 1; s >= path; s--) 640 (void)subyte(--ucp, *s); 641 arg0 = ucp; 642 643 /* 644 * Move out the arg pointers. 645 */ 646 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1)); 647 648 /* terminator */ 649 suword64((uint64_t *)(caddr_t)--uap, (uint64_t)0); 650 651 suword64((uint64_t *)(caddr_t)--uap, (uint64_t)(intptr_t)arg1); 652 suword64((uint64_t *)(caddr_t)--uap, (uint64_t)(intptr_t)arg0); 653 654 /* 655 * Point at the arguments. 656 */ 657 args.fname = arg0; 658 args.argv = uap; 659 args.envv = NULL; 660 661 /* 662 * Now try to exec the program. If can't for any reason 663 * other than it doesn't exist, complain. 664 * 665 * Otherwise, return via fork_trampoline() all the way 666 * to user mode as init! 667 * 668 * WARNING! We may have been moved to another cpu after 669 * acquiring the current user process designation. The 670 * MP lock will migrate with us though so we still have to 671 * release it. 672 */ 673 if ((error = sys_execve(&args)) == 0) { 674 lp->lwp_proc->p_usched->acquire_curproc(lp); 675 return; 676 } 677 if (error != ENOENT) 678 kprintf("exec %.*s: error %d\n", (int)(next - path), 679 path, error); 680 } 681 kprintf("init: not found in path %s\n", init_path); 682 panic("no init"); 683 } 684 685 /* 686 * Like kthread_create(), but runs in it's own address space. 687 * We do this early to reserve pid 1. 688 * 689 * Note special case - do not make it runnable yet. Other work 690 * in progress will change this more. 691 */ 692 static void 693 create_init(const void *udata __unused) 694 { 695 int error; 696 struct lwp *lp; 697 698 crit_enter(); 699 error = fork1(&lwp0, RFFDG | RFPROC, &initproc); 700 if (error) 701 panic("cannot fork init: %d", error); 702 initproc->p_flags |= P_SYSTEM; 703 reaper_init(initproc, &initreaper); 704 lp = ONLY_LWP_IN_PROC(initproc); 705 cpu_set_fork_handler(lp, start_init, NULL); 706 crit_exit(); 707 } 708 SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL); 709 710 /* 711 * Make it runnable now. 712 */ 713 static void 714 kick_init(const void *udata __unused) 715 { 716 start_forked_proc(&lwp0, initproc); 717 } 718 SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL); 719 720 static void 721 kpmap_init(const void *udata __unused) 722 { 723 kpmap = kmalloc(roundup2(sizeof(*kpmap), PAGE_SIZE), 724 M_TEMP, M_ZERO | M_WAITOK); 725 726 kpmap->header[0].type = UKPTYPE_VERSION; 727 kpmap->header[0].offset = offsetof(struct sys_kpmap, version); 728 kpmap->header[1].type = KPTYPE_UPTICKS; 729 kpmap->header[1].offset = offsetof(struct sys_kpmap, upticks); 730 kpmap->header[2].type = KPTYPE_TS_UPTIME; 731 kpmap->header[2].offset = offsetof(struct sys_kpmap, ts_uptime); 732 kpmap->header[3].type = KPTYPE_TS_REALTIME; 733 kpmap->header[3].offset = offsetof(struct sys_kpmap, ts_realtime); 734 kpmap->header[4].type = KPTYPE_TSC_FREQ; 735 kpmap->header[4].offset = offsetof(struct sys_kpmap, tsc_freq); 736 kpmap->header[5].type = KPTYPE_TICK_FREQ; 737 kpmap->header[5].offset = offsetof(struct sys_kpmap, tick_freq); 738 kpmap->version = KPMAP_VERSION; 739 } 740 SYSINIT(kpmapinit, SI_BOOT1_POST, SI_ORDER_FIRST, kpmap_init, NULL); 741 742 /* 743 * Machine independant globaldata initialization 744 * 745 * WARNING! Called from early boot, 'mycpu' may not work yet. 746 */ 747 void 748 mi_gdinit(struct globaldata *gd, int cpuid) 749 { 750 TAILQ_INIT(&gd->gd_systimerq); 751 gd->gd_sysid_alloc = cpuid; /* prime low bits for cpu lookup */ 752 gd->gd_cpuid = cpuid; 753 CPUMASK_ASSBIT(gd->gd_cpumask, cpuid); 754 gd->gd_cpumask_simple = CPUMASK_SIMPLE(cpuid); 755 gd->gd_cpumask_offset = (uintptr_t)CPUMASK_ADDR(*(cpumask_t *)0, cpuid); 756 lwkt_gdinit(gd); 757 vm_map_entry_reserve_cpu_init(gd); 758 sleep_gdinit(gd); 759 slab_gdinit(gd); 760 ATOMIC_CPUMASK_ORBIT(usched_global_cpumask, cpuid); 761 gd->gd_vmstats = vmstats; 762 } 763 764