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 94 int cmask = CMASK; 95 u_int cpu_mi_feature; 96 cpumask_t usched_global_cpumask; 97 extern struct user *proc0paddr; 98 99 int boothowto = 0; /* initialized so that it can be patched */ 100 SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, 101 "Reboot flags, from console subsystem"); 102 SYSCTL_ULONG(_kern, OID_AUTO, usched_global_cpumask, CTLFLAG_RW, 103 &usched_global_cpumask, 0, "global user scheduler cpumask"); 104 105 /* 106 * This ensures that there is at least one entry so that the sysinit_set 107 * symbol is not undefined. A subsystem ID of SI_SPECIAL_DUMMY is never 108 * executed. 109 */ 110 SYSINIT(placeholder, SI_SPECIAL_DUMMY, SI_ORDER_ANY, NULL, NULL) 111 112 /* 113 * The sysinit table itself. Items are checked off as the are run. 114 * If we want to register new sysinit types, add them to newsysinit. 115 */ 116 SET_DECLARE(sysinit_set, struct sysinit); 117 struct sysinit **sysinit, **sysinit_end; 118 struct sysinit **newsysinit, **newsysinit_end; 119 120 121 /* 122 * Merge a new sysinit set into the current set, reallocating it if 123 * necessary. This can only be called after malloc is running. 124 */ 125 void 126 sysinit_add(struct sysinit **set, struct sysinit **set_end) 127 { 128 struct sysinit **newset; 129 struct sysinit **sipp; 130 struct sysinit **xipp; 131 int count; 132 133 count = set_end - set; 134 if (newsysinit) 135 count += newsysinit_end - newsysinit; 136 else 137 count += sysinit_end - sysinit; 138 newset = kmalloc(count * sizeof(*sipp), M_TEMP, M_WAITOK); 139 xipp = newset; 140 if (newsysinit) { 141 for (sipp = newsysinit; sipp < newsysinit_end; sipp++) 142 *xipp++ = *sipp; 143 } else { 144 for (sipp = sysinit; sipp < sysinit_end; sipp++) 145 *xipp++ = *sipp; 146 } 147 for (sipp = set; sipp < set_end; sipp++) 148 *xipp++ = *sipp; 149 if (newsysinit) 150 kfree(newsysinit, M_TEMP); 151 newsysinit = newset; 152 newsysinit_end = newset + count; 153 } 154 155 /* 156 * Callbacks from machine-dependant startup code (e.g. init386) to set 157 * up low level entities related to cpu #0's globaldata. 158 * 159 * Called from very low level boot code. 160 */ 161 void 162 mi_proc0init(struct globaldata *gd, struct user *proc0paddr) 163 { 164 lwkt_init_thread(&thread0, proc0paddr, LWKT_THREAD_STACK, 0, gd); 165 lwkt_set_comm(&thread0, "thread0"); 166 RB_INIT(&proc0.p_lwp_tree); 167 spin_init(&proc0.p_spin); 168 lwkt_token_init(&proc0.p_token, "iproc"); 169 proc0.p_lasttid = 0; /* +1 = next TID */ 170 lwp_rb_tree_RB_INSERT(&proc0.p_lwp_tree, &lwp0); 171 lwp0.lwp_thread = &thread0; 172 lwp0.lwp_proc = &proc0; 173 proc0.p_usched = usched_init(); 174 lwp0.lwp_cpumask = (cpumask_t)-1; 175 lwkt_token_init(&lwp0.lwp_token, "lwp_token"); 176 spin_init(&lwp0.lwp_spin); 177 varsymset_init(&proc0.p_varsymset, NULL); 178 thread0.td_flags |= TDF_RUNNING; 179 thread0.td_proc = &proc0; 180 thread0.td_lwp = &lwp0; 181 thread0.td_switch = cpu_lwkt_switch; 182 lwkt_schedule_self(curthread); 183 } 184 185 /* 186 * System startup; initialize the world, create process 0, mount root 187 * filesystem, and fork to create init and pagedaemon. Most of the 188 * hard work is done in the lower-level initialization routines including 189 * startup(), which does memory initialization and autoconfiguration. 190 * 191 * This allows simple addition of new kernel subsystems that require 192 * boot time initialization. It also allows substitution of subsystem 193 * (for instance, a scheduler, kernel profiler, or VM system) by object 194 * module. Finally, it allows for optional "kernel threads". 195 */ 196 void 197 mi_startup(void) 198 { 199 struct sysinit *sip; /* system initialization*/ 200 struct sysinit **sipp; /* system initialization*/ 201 struct sysinit **xipp; /* interior loop of sort*/ 202 struct sysinit *save; /* bubble*/ 203 204 if (sysinit == NULL) { 205 sysinit = SET_BEGIN(sysinit_set); 206 #if defined(__x86_64__) && defined(_KERNEL_VIRTUAL) 207 /* 208 * XXX For whatever reason, on 64-bit vkernels 209 * the value of sysinit obtained from the 210 * linker set is wrong. 211 */ 212 if ((long)sysinit % 8 != 0) { 213 kprintf("Fixing sysinit value...\n"); 214 sysinit = (void *)((long)(intptr_t)sysinit + 4); 215 } 216 #endif 217 sysinit_end = SET_LIMIT(sysinit_set); 218 } 219 #if defined(__x86_64__) && defined(_KERNEL_VIRTUAL) 220 KKASSERT((long)sysinit % 8 == 0); 221 #endif 222 223 restart: 224 /* 225 * Perform a bubble sort of the system initialization objects by 226 * their subsystem (primary key) and order (secondary key). 227 */ 228 for (sipp = sysinit; sipp < sysinit_end; sipp++) { 229 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) { 230 if ((*sipp)->subsystem < (*xipp)->subsystem || 231 ((*sipp)->subsystem == (*xipp)->subsystem && 232 (*sipp)->order <= (*xipp)->order)) 233 continue; /* skip*/ 234 save = *sipp; 235 *sipp = *xipp; 236 *xipp = save; 237 } 238 } 239 240 /* 241 * Traverse the (now) ordered list of system initialization tasks. 242 * Perform each task, and continue on to the next task. 243 * 244 * The last item on the list is expected to be the scheduler, 245 * which will not return. 246 */ 247 for (sipp = sysinit; sipp < sysinit_end; sipp++) { 248 sip = *sipp; 249 if (sip->subsystem == SI_SPECIAL_DUMMY) 250 continue; /* skip dummy task(s)*/ 251 252 if (sip->subsystem == SI_SPECIAL_DONE) 253 continue; 254 255 /* Call function */ 256 (*(sip->func))(sip->udata); 257 258 /* Check off the one we're just done */ 259 sip->subsystem = SI_SPECIAL_DONE; 260 261 /* Check if we've installed more sysinit items via KLD */ 262 if (newsysinit != NULL) { 263 if (sysinit != SET_BEGIN(sysinit_set)) 264 kfree(sysinit, M_TEMP); 265 sysinit = newsysinit; 266 sysinit_end = newsysinit_end; 267 newsysinit = NULL; 268 newsysinit_end = NULL; 269 goto restart; 270 } 271 } 272 273 panic("Shouldn't get here!"); 274 /* NOTREACHED*/ 275 } 276 277 278 /* 279 *************************************************************************** 280 **** 281 **** The following SYSINIT's belong elsewhere, but have not yet 282 **** been moved. 283 **** 284 *************************************************************************** 285 */ 286 static void 287 print_caddr_t(void *data) 288 { 289 kprintf("%s", (char *)data); 290 } 291 SYSINIT(announce, SI_BOOT1_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright) 292 293 /* 294 * Leave the critical section that protected us from spurious interrupts 295 * so device probes work. 296 */ 297 static void 298 leavecrit(void *dummy __unused) 299 { 300 MachIntrABI.stabilize(); 301 cpu_enable_intr(); 302 MachIntrABI.cleanup(); 303 crit_exit(); 304 KKASSERT(!IN_CRITICAL_SECT(curthread)); 305 306 if (bootverbose) 307 kprintf("Leaving critical section, allowing interrupts\n"); 308 } 309 SYSINIT(leavecrit, SI_BOOT2_LEAVE_CRIT, SI_ORDER_ANY, leavecrit, NULL) 310 311 /* 312 * This is called after the threading system is up and running, 313 * including the softclock, clock interrupts, and SMP. 314 */ 315 static void 316 tsleepworks(void *dummy __unused) 317 { 318 tsleep_now_works = 1; 319 } 320 SYSINIT(tsleepworks, SI_BOOT2_FINISH_SMP, SI_ORDER_SECOND, tsleepworks, NULL) 321 322 /* 323 * This is called after devices have configured. Tell the kernel we are 324 * no longer in cold boot. 325 */ 326 static void 327 endofcoldboot(void *dummy __unused) 328 { 329 cold = 0; 330 } 331 SYSINIT(endofcoldboot, SI_SUB_ISWARM, SI_ORDER_ANY, endofcoldboot, NULL) 332 333 /* 334 *************************************************************************** 335 **** 336 **** The two following SYSINT's are proc0 specific glue code. I am not 337 **** convinced that they can not be safely combined, but their order of 338 **** operation has been maintained as the same as the original init_main.c 339 **** for right now. 340 **** 341 **** These probably belong in init_proc.c or kern_proc.c, since they 342 **** deal with proc0 (the fork template process). 343 **** 344 *************************************************************************** 345 */ 346 /* ARGSUSED*/ 347 static void 348 proc0_init(void *dummy __unused) 349 { 350 struct proc *p; 351 struct lwp *lp; 352 353 p = &proc0; 354 lp = &lwp0; 355 356 /* 357 * Initialize osrel 358 */ 359 p->p_osrel = osreldate; 360 361 /* 362 * Initialize process and pgrp structures. 363 */ 364 procinit(); 365 366 /* 367 * additional VM structures 368 */ 369 vm_init2(); 370 371 /* 372 * Create process 0 (the swapper). 373 */ 374 LIST_INSERT_HEAD(&allproc, p, p_list); 375 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash); 376 LIST_INIT(&pgrp0.pg_members); 377 lwkt_token_init(&pgrp0.pg_token, "pgrp0"); 378 refcount_init(&pgrp0.pg_refs, 1); 379 lockinit(&pgrp0.pg_lock, "pgwt0", 0, 0); 380 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist); 381 382 pgrp0.pg_session = &session0; 383 session0.s_count = 1; 384 session0.s_leader = p; 385 386 pgref(&pgrp0); 387 p->p_pgrp = &pgrp0; 388 389 p->p_sysent = &aout_sysvec; 390 391 p->p_flags = P_SYSTEM; 392 p->p_stat = SACTIVE; 393 lp->lwp_stat = LSRUN; 394 p->p_nice = NZERO; 395 p->p_rtprio.type = RTP_PRIO_NORMAL; 396 p->p_rtprio.prio = 0; 397 lp->lwp_rtprio = p->p_rtprio; 398 399 p->p_peers = NULL; 400 p->p_leader = p; 401 402 bcopy("swapper", p->p_comm, sizeof ("swapper")); 403 bcopy("swapper", thread0.td_comm, sizeof ("swapper")); 404 405 /* Create credentials. */ 406 p->p_ucred = crget(); 407 p->p_ucred->cr_ruidinfo = uifind(0); 408 p->p_ucred->cr_ngroups = 1; /* group 0 */ 409 p->p_ucred->cr_uidinfo = uifind(0); 410 thread0.td_ucred = crhold(p->p_ucred); /* bootstrap fork1() */ 411 412 /* Don't jail it */ 413 p->p_ucred->cr_prison = NULL; 414 415 /* Create sigacts. */ 416 p->p_sigacts = &sigacts0; 417 refcount_init(&p->p_sigacts->ps_refcnt, 1); 418 419 /* Initialize signal state for process 0. */ 420 siginit(p); 421 422 /* Create the file descriptor table. */ 423 fdinit_bootstrap(p, &filedesc0, cmask); 424 425 /* Create the limits structures. */ 426 plimit_init0(&limit0); 427 p->p_limit = &limit0; 428 429 /* Allocate a prototype map so we have something to fork. */ 430 pmap_pinit0(vmspace_pmap(&vmspace0)); 431 p->p_vmspace = &vmspace0; 432 lp->lwp_vmspace = p->p_vmspace; 433 sysref_init(&vmspace0.vm_sysref, &vmspace_sysref_class); 434 vm_map_init(&vmspace0.vm_map, 435 round_page(VM_MIN_USER_ADDRESS), 436 trunc_page(VM_MAX_USER_ADDRESS), 437 vmspace_pmap(&vmspace0)); 438 sysref_activate(&vmspace0.vm_sysref); 439 440 kqueue_init(&lwp0.lwp_kqueue, &filedesc0); 441 442 /* 443 * Charge root for one process. 444 */ 445 (void)chgproccnt(p->p_ucred->cr_uidinfo, 1, 0); 446 vm_init_limits(p); 447 } 448 SYSINIT(p0init, SI_BOOT2_PROC0, SI_ORDER_FIRST, proc0_init, NULL) 449 450 static int proc0_post_callback(struct proc *p, void *data __unused); 451 452 /* ARGSUSED*/ 453 static void 454 proc0_post(void *dummy __unused) 455 { 456 struct timespec ts; 457 458 /* 459 * Now we can look at the time, having had a chance to verify the 460 * time from the file system. Pretend that proc0 started now. 461 */ 462 allproc_scan(proc0_post_callback, NULL); 463 464 /* 465 * Give the ``random'' number generator a thump. 466 * XXX: Does read_random() contain enough bits to be used here ? 467 */ 468 nanotime(&ts); 469 skrandom(ts.tv_sec ^ ts.tv_nsec); 470 } 471 472 static int 473 proc0_post_callback(struct proc *p, void *data __unused) 474 { 475 microtime(&p->p_start); 476 return(0); 477 } 478 479 SYSINIT(p0post, SI_SUB_PROC0_POST, SI_ORDER_FIRST, proc0_post, NULL) 480 481 /* 482 *************************************************************************** 483 **** 484 **** The following SYSINIT's and glue code should be moved to the 485 **** respective files on a per subsystem basis. 486 **** 487 *************************************************************************** 488 */ 489 490 491 /* 492 *************************************************************************** 493 **** 494 **** The following code probably belongs in another file, like 495 **** kern/init_init.c. 496 **** 497 *************************************************************************** 498 */ 499 500 /* 501 * List of paths to try when searching for "init". 502 */ 503 static char init_path[MAXPATHLEN] = 504 #ifdef INIT_PATH 505 __XSTRING(INIT_PATH); 506 #else 507 "/sbin/init:/sbin/oinit:/sbin/init.bak"; 508 #endif 509 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, ""); 510 511 /* 512 * Shutdown timeout of init(8). 513 * Unused within kernel, but used to control init(8), hence do not remove. 514 */ 515 #ifndef INIT_SHUTDOWN_TIMEOUT 516 #define INIT_SHUTDOWN_TIMEOUT 120 517 #endif 518 static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT; 519 SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout, 520 CTLFLAG_RW, &init_shutdown_timeout, 0, "Shutdown timeout of init(8). " 521 "Unused within kernel, but used to control init(8)"); 522 523 /* 524 * Start the initial user process; try exec'ing each pathname in init_path. 525 * The program is invoked with one argument containing the boot flags. 526 */ 527 static void 528 start_init(void *dummy, struct trapframe *frame) 529 { 530 vm_offset_t addr; 531 struct execve_args args; 532 int options, error; 533 char *var, *path, *next, *s; 534 char *ucp, **uap, *arg0, *arg1; 535 struct proc *p; 536 struct lwp *lp; 537 struct mount *mp; 538 struct vnode *vp; 539 char *env; 540 541 /* 542 * This is passed in by the bootloader 543 */ 544 env = kgetenv("kernelname"); 545 if (env != NULL) 546 strlcpy(kernelname, env, sizeof(kernelname)); 547 548 /* 549 * The MP lock is not held on entry. We release it before 550 * returning to userland. 551 */ 552 get_mplock(); 553 p = curproc; 554 555 lp = ONLY_LWP_IN_PROC(p); 556 557 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */ 558 mp = mountlist_boot_getfirst(); 559 if (VFS_ROOT(mp, &vp)) 560 panic("cannot find root vnode"); 561 if (mp->mnt_ncmountpt.ncp == NULL) { 562 cache_allocroot(&mp->mnt_ncmountpt, mp, vp); 563 cache_unlock(&mp->mnt_ncmountpt); /* leave ref intact */ 564 } 565 p->p_fd->fd_cdir = vp; 566 vref(p->p_fd->fd_cdir); 567 p->p_fd->fd_rdir = vp; 568 vref(p->p_fd->fd_rdir); 569 vfs_cache_setroot(vp, cache_hold(&mp->mnt_ncmountpt)); 570 vn_unlock(vp); /* leave ref intact */ 571 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_ncdir); 572 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_nrdir); 573 574 kprintf("Mounting devfs\n"); 575 vfs_mountroot_devfs(); 576 577 /* 578 * Need just enough stack to hold the faked-up "execve()" arguments. 579 */ 580 addr = trunc_page(USRSTACK - PAGE_SIZE); 581 error = vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, 582 PAGE_SIZE, PAGE_SIZE, 583 FALSE, VM_MAPTYPE_NORMAL, 584 VM_PROT_ALL, VM_PROT_ALL, 585 0); 586 if (error) 587 panic("init: couldn't allocate argument space"); 588 p->p_vmspace->vm_maxsaddr = (caddr_t)addr; 589 p->p_vmspace->vm_ssize = 1; 590 591 if ((var = kgetenv("init_path")) != NULL) { 592 strncpy(init_path, var, sizeof init_path); 593 init_path[sizeof init_path - 1] = 0; 594 } 595 596 for (path = init_path; *path != '\0'; path = next) { 597 while (*path == ':') 598 path++; 599 if (*path == '\0') 600 break; 601 for (next = path; *next != '\0' && *next != ':'; next++) 602 /* nothing */ ; 603 if (bootverbose) 604 kprintf("start_init: trying %.*s\n", (int)(next - path), 605 path); 606 607 /* 608 * Move out the boot flag argument. 609 */ 610 options = 0; 611 ucp = (char *)USRSTACK; 612 (void)subyte(--ucp, 0); /* trailing zero */ 613 if (boothowto & RB_SINGLE) { 614 (void)subyte(--ucp, 's'); 615 options = 1; 616 } 617 #ifdef notyet 618 if (boothowto & RB_FASTBOOT) { 619 (void)subyte(--ucp, 'f'); 620 options = 1; 621 } 622 #endif 623 624 #ifdef BOOTCDROM 625 (void)subyte(--ucp, 'C'); 626 options = 1; 627 #endif 628 if (options == 0) 629 (void)subyte(--ucp, '-'); 630 (void)subyte(--ucp, '-'); /* leading hyphen */ 631 arg1 = ucp; 632 633 /* 634 * Move out the file name (also arg 0). 635 */ 636 (void)subyte(--ucp, 0); 637 for (s = next - 1; s >= path; s--) 638 (void)subyte(--ucp, *s); 639 arg0 = ucp; 640 641 /* 642 * Move out the arg pointers. 643 */ 644 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1)); 645 (void)suword((caddr_t)--uap, (long)0); /* terminator */ 646 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1); 647 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0); 648 649 /* 650 * Point at the arguments. 651 */ 652 args.fname = arg0; 653 args.argv = uap; 654 args.envv = NULL; 655 656 /* 657 * Now try to exec the program. If can't for any reason 658 * other than it doesn't exist, complain. 659 * 660 * Otherwise, return via fork_trampoline() all the way 661 * to user mode as init! 662 * 663 * WARNING! We may have been moved to another cpu after 664 * acquiring the current user process designation. The 665 * MP lock will migrate with us though so we still have to 666 * release it. 667 */ 668 if ((error = sys_execve(&args)) == 0) { 669 rel_mplock(); 670 lp->lwp_proc->p_usched->acquire_curproc(lp); 671 return; 672 } 673 if (error != ENOENT) 674 kprintf("exec %.*s: error %d\n", (int)(next - path), 675 path, error); 676 } 677 kprintf("init: not found in path %s\n", init_path); 678 panic("no init"); 679 } 680 681 /* 682 * Like kthread_create(), but runs in it's own address space. 683 * We do this early to reserve pid 1. 684 * 685 * Note special case - do not make it runnable yet. Other work 686 * in progress will change this more. 687 */ 688 static void 689 create_init(const void *udata __unused) 690 { 691 int error; 692 struct lwp *lp; 693 694 crit_enter(); 695 error = fork1(&lwp0, RFFDG | RFPROC, &initproc); 696 if (error) 697 panic("cannot fork init: %d", error); 698 initproc->p_flags |= P_SYSTEM; 699 lp = ONLY_LWP_IN_PROC(initproc); 700 cpu_set_fork_handler(lp, start_init, NULL); 701 crit_exit(); 702 } 703 SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL) 704 705 /* 706 * Make it runnable now. 707 */ 708 static void 709 kick_init(const void *udata __unused) 710 { 711 start_forked_proc(&lwp0, initproc); 712 } 713 SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL) 714 715 /* 716 * Machine independant globaldata initialization 717 * 718 * WARNING! Called from early boot, 'mycpu' may not work yet. 719 */ 720 void 721 mi_gdinit(struct globaldata *gd, int cpuid) 722 { 723 TAILQ_INIT(&gd->gd_systimerq); 724 gd->gd_sysid_alloc = cpuid; /* prime low bits for cpu lookup */ 725 gd->gd_cpuid = cpuid; 726 gd->gd_cpumask = CPUMASK(cpuid); 727 lwkt_gdinit(gd); 728 vm_map_entry_reserve_cpu_init(gd); 729 sleep_gdinit(gd); 730 atomic_set_cpumask(&usched_global_cpumask, CPUMASK(cpuid)); 731 } 732 733