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 #include <sys/mplock2.h> 68 69 #include <machine/cpu.h> 70 71 #include <vm/vm.h> 72 #include <vm/vm_param.h> 73 #include <sys/lock.h> 74 #include <vm/pmap.h> 75 #include <vm/vm_map.h> 76 #include <vm/vm_extern.h> 77 #include <sys/user.h> 78 #include <sys/copyright.h> 79 80 int vfs_mountroot_devfs(void); 81 82 /* Components of the first process -- never freed. */ 83 static struct session session0; 84 static struct pgrp pgrp0; 85 static struct sigacts sigacts0; 86 static struct filedesc filedesc0; 87 static struct plimit limit0; 88 static struct vmspace vmspace0; 89 struct proc *initproc; 90 struct proc proc0; 91 struct lwp lwp0; 92 struct thread thread0; 93 struct sys_kpmap *kpmap; 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_ULONG(_kern, OID_AUTO, usched_global_cpumask, CTLFLAG_RW, 104 &usched_global_cpumask, 0, "global user scheduler cpumask"); 105 106 /* 107 * This ensures that there is at least one entry so that the sysinit_set 108 * symbol is not undefined. A subsystem ID of SI_SPECIAL_DUMMY is never 109 * executed. 110 */ 111 SYSINIT(placeholder, SI_SPECIAL_DUMMY, SI_ORDER_ANY, NULL, NULL) 112 113 /* 114 * The sysinit table itself. Items are checked off as the are run. 115 * If we want to register new sysinit types, add them to newsysinit. 116 */ 117 SET_DECLARE(sysinit_set, struct sysinit); 118 struct sysinit **sysinit, **sysinit_end; 119 struct sysinit **newsysinit, **newsysinit_end; 120 121 122 /* 123 * Merge a new sysinit set into the current set, reallocating it if 124 * necessary. This can only be called after malloc is running. 125 */ 126 void 127 sysinit_add(struct sysinit **set, struct sysinit **set_end) 128 { 129 struct sysinit **newset; 130 struct sysinit **sipp; 131 struct sysinit **xipp; 132 int count; 133 134 count = set_end - set; 135 if (newsysinit) 136 count += newsysinit_end - newsysinit; 137 else 138 count += sysinit_end - sysinit; 139 newset = kmalloc(count * sizeof(*sipp), M_TEMP, M_WAITOK); 140 xipp = newset; 141 if (newsysinit) { 142 for (sipp = newsysinit; sipp < newsysinit_end; sipp++) 143 *xipp++ = *sipp; 144 } else { 145 for (sipp = sysinit; sipp < sysinit_end; sipp++) 146 *xipp++ = *sipp; 147 } 148 for (sipp = set; sipp < set_end; sipp++) 149 *xipp++ = *sipp; 150 if (newsysinit) 151 kfree(newsysinit, M_TEMP); 152 newsysinit = newset; 153 newsysinit_end = newset + count; 154 } 155 156 /* 157 * Callbacks from machine-dependant startup code (e.g. init386) to set 158 * up low level entities related to cpu #0's globaldata. 159 * 160 * Called from very low level boot code. 161 */ 162 void 163 mi_proc0init(struct globaldata *gd, struct user *proc0paddr) 164 { 165 lwkt_init_thread(&thread0, proc0paddr, LWKT_THREAD_STACK, 0, gd); 166 lwkt_set_comm(&thread0, "thread0"); 167 RB_INIT(&proc0.p_lwp_tree); 168 spin_init(&proc0.p_spin, "iproc_proc0"); 169 lwkt_token_init(&proc0.p_token, "iproc"); 170 proc0.p_lasttid = 0; /* +1 = next TID */ 171 lwp_rb_tree_RB_INSERT(&proc0.p_lwp_tree, &lwp0); 172 lwp0.lwp_thread = &thread0; 173 lwp0.lwp_proc = &proc0; 174 proc0.p_usched = usched_init(); 175 CPUMASK_ASSALLONES(lwp0.lwp_cpumask); 176 lwkt_token_init(&lwp0.lwp_token, "lwp_token"); 177 spin_init(&lwp0.lwp_spin, "iproc_lwp0"); 178 varsymset_init(&proc0.p_varsymset, NULL); 179 thread0.td_flags |= TDF_RUNNING; 180 thread0.td_proc = &proc0; 181 thread0.td_lwp = &lwp0; 182 thread0.td_switch = cpu_lwkt_switch; 183 lwkt_schedule_self(curthread); 184 } 185 186 /* 187 * System startup; initialize the world, create process 0, mount root 188 * filesystem, and fork to create init and pagedaemon. Most of the 189 * hard work is done in the lower-level initialization routines including 190 * startup(), which does memory initialization and autoconfiguration. 191 * 192 * This allows simple addition of new kernel subsystems that require 193 * boot time initialization. It also allows substitution of subsystem 194 * (for instance, a scheduler, kernel profiler, or VM system) by object 195 * module. Finally, it allows for optional "kernel threads". 196 */ 197 void 198 mi_startup(void) 199 { 200 struct sysinit *sip; /* system initialization*/ 201 struct sysinit **sipp; /* system initialization*/ 202 struct sysinit **xipp; /* interior loop of sort*/ 203 struct sysinit *save; /* bubble*/ 204 205 if (sysinit == NULL) { 206 sysinit = SET_BEGIN(sysinit_set); 207 #if defined(__x86_64__) && defined(_KERNEL_VIRTUAL) 208 /* 209 * XXX For whatever reason, on 64-bit vkernels 210 * the value of sysinit obtained from the 211 * linker set is wrong. 212 */ 213 if ((long)sysinit % 8 != 0) { 214 kprintf("Fixing sysinit value...\n"); 215 sysinit = (void *)((long)(intptr_t)sysinit + 4); 216 } 217 #endif 218 sysinit_end = SET_LIMIT(sysinit_set); 219 } 220 #if defined(__x86_64__) && defined(_KERNEL_VIRTUAL) 221 KKASSERT((long)sysinit % 8 == 0); 222 #endif 223 224 restart: 225 /* 226 * Perform a bubble sort of the system initialization objects by 227 * their subsystem (primary key) and order (secondary key). 228 */ 229 for (sipp = sysinit; sipp < sysinit_end; sipp++) { 230 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) { 231 if ((*sipp)->subsystem < (*xipp)->subsystem || 232 ((*sipp)->subsystem == (*xipp)->subsystem && 233 (*sipp)->order <= (*xipp)->order)) 234 continue; /* skip*/ 235 save = *sipp; 236 *sipp = *xipp; 237 *xipp = save; 238 } 239 } 240 241 /* 242 * Traverse the (now) ordered list of system initialization tasks. 243 * Perform each task, and continue on to the next task. 244 * 245 * The last item on the list is expected to be the scheduler, 246 * which will not return. 247 */ 248 for (sipp = sysinit; sipp < sysinit_end; sipp++) { 249 sip = *sipp; 250 if (sip->subsystem == SI_SPECIAL_DUMMY) 251 continue; /* skip dummy task(s)*/ 252 253 if (sip->subsystem == SI_SPECIAL_DONE) 254 continue; 255 256 #if 0 257 if (bootverbose) 258 kprintf("(%08x-%p)\n", sip->subsystem, sip->func); 259 #endif 260 261 /* Call function */ 262 (*(sip->func))(sip->udata); 263 264 /* Check off the one we're just done */ 265 sip->subsystem = SI_SPECIAL_DONE; 266 267 /* Check if we've installed more sysinit items via KLD */ 268 if (newsysinit != NULL) { 269 if (sysinit != SET_BEGIN(sysinit_set)) 270 kfree(sysinit, M_TEMP); 271 sysinit = newsysinit; 272 sysinit_end = newsysinit_end; 273 newsysinit = NULL; 274 newsysinit_end = NULL; 275 goto restart; 276 } 277 } 278 279 panic("Shouldn't get here!"); 280 /* NOTREACHED*/ 281 } 282 283 284 /* 285 *************************************************************************** 286 **** 287 **** The following SYSINIT's belong elsewhere, but have not yet 288 **** been moved. 289 **** 290 *************************************************************************** 291 */ 292 static void 293 print_caddr_t(void *data) 294 { 295 kprintf("%s", (char *)data); 296 } 297 SYSINIT(announce, SI_BOOT1_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright) 298 299 /* 300 * Leave the critical section that protected us from spurious interrupts 301 * so device probes work. 302 */ 303 static void 304 leavecrit(void *dummy __unused) 305 { 306 MachIntrABI.stabilize(); 307 cpu_enable_intr(); 308 MachIntrABI.cleanup(); 309 crit_exit(); 310 KKASSERT(!IN_CRITICAL_SECT(curthread)); 311 312 if (bootverbose) 313 kprintf("Leaving critical section, allowing interrupts\n"); 314 } 315 SYSINIT(leavecrit, SI_BOOT2_LEAVE_CRIT, SI_ORDER_ANY, leavecrit, NULL) 316 317 /* 318 * This is called after the threading system is up and running, 319 * including the softclock, clock interrupts, and SMP. 320 */ 321 static void 322 tsleepworks(void *dummy __unused) 323 { 324 tsleep_now_works = 1; 325 } 326 SYSINIT(tsleepworks, SI_BOOT2_FINISH_SMP, SI_ORDER_SECOND, tsleepworks, NULL) 327 328 /* 329 * This is called after devices have configured. Tell the kernel we are 330 * no longer in cold boot. 331 */ 332 static void 333 endofcoldboot(void *dummy __unused) 334 { 335 cold = 0; 336 } 337 SYSINIT(endofcoldboot, SI_SUB_ISWARM, SI_ORDER_ANY, endofcoldboot, NULL) 338 339 /* 340 *************************************************************************** 341 **** 342 **** The two following SYSINT's are proc0 specific glue code. I am not 343 **** convinced that they can not be safely combined, but their order of 344 **** operation has been maintained as the same as the original init_main.c 345 **** for right now. 346 **** 347 **** These probably belong in init_proc.c or kern_proc.c, since they 348 **** deal with proc0 (the fork template process). 349 **** 350 *************************************************************************** 351 */ 352 /* ARGSUSED*/ 353 static void 354 proc0_init(void *dummy __unused) 355 { 356 struct proc *p; 357 struct lwp *lp; 358 359 p = &proc0; 360 lp = &lwp0; 361 362 /* 363 * Initialize osrel 364 */ 365 p->p_osrel = osreldate; 366 367 /* 368 * Initialize process and pgrp structures. 369 */ 370 procinit(); 371 372 /* 373 * additional VM structures 374 */ 375 vm_init2(); 376 377 /* 378 * Create process 0 (the swapper). 379 */ 380 procinsertinit(p); 381 pgrpinsertinit(&pgrp0); 382 LIST_INIT(&pgrp0.pg_members); 383 lwkt_token_init(&pgrp0.pg_token, "pgrp0"); 384 refcount_init(&pgrp0.pg_refs, 1); 385 lockinit(&pgrp0.pg_lock, "pgwt0", 0, 0); 386 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist); 387 388 pgrp0.pg_session = &session0; 389 session0.s_count = 1; 390 session0.s_leader = p; 391 sessinsertinit(&session0); 392 393 pgref(&pgrp0); 394 p->p_pgrp = &pgrp0; 395 396 p->p_sysent = &aout_sysvec; 397 398 p->p_flags = P_SYSTEM; 399 p->p_stat = SACTIVE; 400 lp->lwp_stat = LSRUN; 401 p->p_nice = NZERO; 402 p->p_rtprio.type = RTP_PRIO_NORMAL; 403 p->p_rtprio.prio = 0; 404 lp->lwp_rtprio = p->p_rtprio; 405 406 p->p_peers = NULL; 407 p->p_leader = p; 408 409 bcopy("swapper", p->p_comm, sizeof ("swapper")); 410 bcopy("swapper", thread0.td_comm, sizeof ("swapper")); 411 412 /* Create credentials. */ 413 p->p_ucred = crget(); 414 p->p_ucred->cr_ruidinfo = uifind(0); 415 p->p_ucred->cr_ngroups = 1; /* group 0 */ 416 p->p_ucred->cr_uidinfo = uifind(0); 417 thread0.td_ucred = crhold(p->p_ucred); /* bootstrap fork1() */ 418 419 /* Don't jail it */ 420 p->p_ucred->cr_prison = NULL; 421 422 /* Create sigacts. */ 423 p->p_sigacts = &sigacts0; 424 refcount_init(&p->p_sigacts->ps_refcnt, 1); 425 426 /* Initialize signal state for process 0. */ 427 siginit(p); 428 429 /* Create the file descriptor table. */ 430 fdinit_bootstrap(p, &filedesc0, cmask); 431 432 /* Create the limits structures. */ 433 plimit_init0(&limit0); 434 p->p_limit = &limit0; 435 436 /* Allocate a prototype map so we have something to fork. */ 437 pmap_pinit0(vmspace_pmap(&vmspace0)); 438 p->p_vmspace = &vmspace0; 439 lp->lwp_vmspace = p->p_vmspace; 440 vmspace_initrefs(&vmspace0); 441 vm_map_init(&vmspace0.vm_map, 442 round_page(VM_MIN_USER_ADDRESS), 443 trunc_page(VM_MAX_USER_ADDRESS), 444 vmspace_pmap(&vmspace0)); 445 446 kqueue_init(&lwp0.lwp_kqueue, &filedesc0); 447 448 /* 449 * Charge root for one process. 450 */ 451 (void)chgproccnt(p->p_ucred->cr_uidinfo, 1, 0); 452 vm_init_limits(p); 453 } 454 SYSINIT(p0init, SI_BOOT2_PROC0, SI_ORDER_FIRST, proc0_init, NULL) 455 456 static int proc0_post_callback(struct proc *p, void *data __unused); 457 458 /* ARGSUSED*/ 459 static void 460 proc0_post(void *dummy __unused) 461 { 462 struct timespec ts; 463 464 /* 465 * Now we can look at the time, having had a chance to verify the 466 * time from the file system. Pretend that proc0 started now. 467 */ 468 allproc_scan(proc0_post_callback, NULL); 469 470 /* 471 * Give the ``random'' number generator a thump. 472 * XXX: Does read_random() contain enough bits to be used here ? 473 */ 474 nanotime(&ts); 475 skrandom(ts.tv_sec ^ ts.tv_nsec); 476 } 477 478 static int 479 proc0_post_callback(struct proc *p, void *data __unused) 480 { 481 microtime(&p->p_start); 482 return(0); 483 } 484 485 SYSINIT(p0post, SI_SUB_PROC0_POST, SI_ORDER_FIRST, proc0_post, NULL) 486 487 /* 488 *************************************************************************** 489 **** 490 **** The following SYSINIT's and glue code should be moved to the 491 **** respective files on a per subsystem basis. 492 **** 493 *************************************************************************** 494 */ 495 496 497 /* 498 *************************************************************************** 499 **** 500 **** The following code probably belongs in another file, like 501 **** kern/init_init.c. 502 **** 503 *************************************************************************** 504 */ 505 506 /* 507 * List of paths to try when searching for "init". 508 */ 509 static char init_path[MAXPATHLEN] = 510 #ifdef INIT_PATH 511 __XSTRING(INIT_PATH); 512 #else 513 "/sbin/init:/sbin/oinit:/sbin/init.bak"; 514 #endif 515 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, ""); 516 517 /* 518 * Shutdown timeout of init(8). 519 * Unused within kernel, but used to control init(8), hence do not remove. 520 */ 521 #ifndef INIT_SHUTDOWN_TIMEOUT 522 #define INIT_SHUTDOWN_TIMEOUT 120 523 #endif 524 static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT; 525 SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout, 526 CTLFLAG_RW, &init_shutdown_timeout, 0, "Shutdown timeout of init(8). " 527 "Unused within kernel, but used to control init(8)"); 528 529 /* 530 * Start the initial user process; try exec'ing each pathname in init_path. 531 * The program is invoked with one argument containing the boot flags. 532 */ 533 static void 534 start_init(void *dummy, struct trapframe *frame) 535 { 536 vm_offset_t addr; 537 struct execve_args args; 538 int options, error; 539 char *var, *path, *next, *s; 540 char *ucp, **uap, *arg0, *arg1; 541 struct proc *p; 542 struct lwp *lp; 543 struct mount *mp; 544 struct vnode *vp; 545 char *env; 546 547 /* 548 * This is passed in by the bootloader 549 */ 550 env = kgetenv("kernelname"); 551 if (env != NULL) 552 strlcpy(kernelname, env, sizeof(kernelname)); 553 554 /* 555 * The MP lock is not held on entry. We release it before 556 * returning to userland. 557 */ 558 get_mplock(); 559 p = curproc; 560 561 lp = ONLY_LWP_IN_PROC(p); 562 563 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */ 564 mp = mountlist_boot_getfirst(); 565 if (VFS_ROOT(mp, &vp)) 566 panic("cannot find root vnode"); 567 if (mp->mnt_ncmountpt.ncp == NULL) { 568 cache_allocroot(&mp->mnt_ncmountpt, mp, vp); 569 cache_unlock(&mp->mnt_ncmountpt); /* leave ref intact */ 570 } 571 p->p_fd->fd_cdir = vp; 572 vref(p->p_fd->fd_cdir); 573 p->p_fd->fd_rdir = vp; 574 vref(p->p_fd->fd_rdir); 575 vfs_cache_setroot(vp, cache_hold(&mp->mnt_ncmountpt)); 576 vn_unlock(vp); /* leave ref intact */ 577 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_ncdir); 578 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_nrdir); 579 580 kprintf("Mounting devfs\n"); 581 vfs_mountroot_devfs(); 582 583 /* 584 * Need just enough stack to hold the faked-up "execve()" arguments. 585 */ 586 addr = trunc_page(USRSTACK - PAGE_SIZE); 587 error = vm_map_find(&p->p_vmspace->vm_map, NULL, NULL, 588 0, &addr, PAGE_SIZE, 589 PAGE_SIZE, 590 FALSE, VM_MAPTYPE_NORMAL, 591 VM_PROT_ALL, VM_PROT_ALL, 0); 592 if (error) 593 panic("init: couldn't allocate argument space"); 594 p->p_vmspace->vm_maxsaddr = (caddr_t)addr; 595 p->p_vmspace->vm_ssize = 1; 596 597 if ((var = kgetenv("init_path")) != NULL) { 598 strncpy(init_path, var, sizeof init_path); 599 init_path[sizeof init_path - 1] = 0; 600 } 601 602 for (path = init_path; *path != '\0'; path = next) { 603 while (*path == ':') 604 path++; 605 if (*path == '\0') 606 break; 607 for (next = path; *next != '\0' && *next != ':'; next++) 608 /* nothing */ ; 609 if (bootverbose) 610 kprintf("start_init: trying %.*s\n", (int)(next - path), 611 path); 612 613 /* 614 * Move out the boot flag argument. 615 */ 616 options = 0; 617 ucp = (char *)USRSTACK; 618 (void)subyte(--ucp, 0); /* trailing zero */ 619 if (boothowto & RB_SINGLE) { 620 (void)subyte(--ucp, 's'); 621 options = 1; 622 } 623 #ifdef notyet 624 if (boothowto & RB_FASTBOOT) { 625 (void)subyte(--ucp, 'f'); 626 options = 1; 627 } 628 #endif 629 630 #ifdef BOOTCDROM 631 (void)subyte(--ucp, 'C'); 632 options = 1; 633 #endif 634 if (options == 0) 635 (void)subyte(--ucp, '-'); 636 (void)subyte(--ucp, '-'); /* leading hyphen */ 637 arg1 = ucp; 638 639 /* 640 * Move out the file name (also arg 0). 641 */ 642 (void)subyte(--ucp, 0); 643 for (s = next - 1; s >= path; s--) 644 (void)subyte(--ucp, *s); 645 arg0 = ucp; 646 647 /* 648 * Move out the arg pointers. 649 */ 650 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1)); 651 (void)suword((caddr_t)--uap, (long)0); /* terminator */ 652 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1); 653 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0); 654 655 /* 656 * Point at the arguments. 657 */ 658 args.fname = arg0; 659 args.argv = uap; 660 args.envv = NULL; 661 662 /* 663 * Now try to exec the program. If can't for any reason 664 * other than it doesn't exist, complain. 665 * 666 * Otherwise, return via fork_trampoline() all the way 667 * to user mode as init! 668 * 669 * WARNING! We may have been moved to another cpu after 670 * acquiring the current user process designation. The 671 * MP lock will migrate with us though so we still have to 672 * release it. 673 */ 674 if ((error = sys_execve(&args)) == 0) { 675 rel_mplock(); 676 lp->lwp_proc->p_usched->acquire_curproc(lp); 677 return; 678 } 679 if (error != ENOENT) 680 kprintf("exec %.*s: error %d\n", (int)(next - path), 681 path, error); 682 } 683 kprintf("init: not found in path %s\n", init_path); 684 panic("no init"); 685 } 686 687 /* 688 * Like kthread_create(), but runs in it's own address space. 689 * We do this early to reserve pid 1. 690 * 691 * Note special case - do not make it runnable yet. Other work 692 * in progress will change this more. 693 */ 694 static void 695 create_init(const void *udata __unused) 696 { 697 int error; 698 struct lwp *lp; 699 700 crit_enter(); 701 error = fork1(&lwp0, RFFDG | RFPROC, &initproc); 702 if (error) 703 panic("cannot fork init: %d", error); 704 initproc->p_flags |= P_SYSTEM; 705 lp = ONLY_LWP_IN_PROC(initproc); 706 cpu_set_fork_handler(lp, start_init, NULL); 707 crit_exit(); 708 } 709 SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL) 710 711 /* 712 * Make it runnable now. 713 */ 714 static void 715 kick_init(const void *udata __unused) 716 { 717 start_forked_proc(&lwp0, initproc); 718 } 719 SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL) 720 721 static void 722 kpmap_init(const void *udata __unused) 723 { 724 kpmap = kmalloc(roundup2(sizeof(*kpmap), PAGE_SIZE), 725 M_TEMP, M_ZERO | M_WAITOK); 726 727 kpmap->header[0].type = UKPTYPE_VERSION; 728 kpmap->header[0].offset = offsetof(struct sys_kpmap, version); 729 kpmap->header[1].type = KPTYPE_UPTICKS; 730 kpmap->header[1].offset = offsetof(struct sys_kpmap, upticks); 731 kpmap->header[2].type = KPTYPE_TS_UPTIME; 732 kpmap->header[2].offset = offsetof(struct sys_kpmap, ts_uptime); 733 kpmap->header[3].type = KPTYPE_TS_REALTIME; 734 kpmap->header[3].offset = offsetof(struct sys_kpmap, ts_realtime); 735 kpmap->header[4].type = KPTYPE_TSC_FREQ; 736 kpmap->header[4].offset = offsetof(struct sys_kpmap, tsc_freq); 737 kpmap->header[5].type = KPTYPE_TICK_FREQ; 738 kpmap->header[5].offset = offsetof(struct sys_kpmap, tick_freq); 739 kpmap->version = KPMAP_VERSION; 740 } 741 SYSINIT(kpmapinit, SI_BOOT1_POST, SI_ORDER_FIRST, kpmap_init, NULL) 742 743 /* 744 * Machine independant globaldata initialization 745 * 746 * WARNING! Called from early boot, 'mycpu' may not work yet. 747 */ 748 void 749 mi_gdinit(struct globaldata *gd, int cpuid) 750 { 751 TAILQ_INIT(&gd->gd_systimerq); 752 gd->gd_sysid_alloc = cpuid; /* prime low bits for cpu lookup */ 753 gd->gd_cpuid = cpuid; 754 CPUMASK_ASSBIT(gd->gd_cpumask, cpuid); 755 lwkt_gdinit(gd); 756 vm_map_entry_reserve_cpu_init(gd); 757 sleep_gdinit(gd); 758 ATOMIC_CPUMASK_ORBIT(usched_global_cpumask, cpuid); 759 } 760 761