sys/kern/init_main.c

   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/spinlock2.h>
  66
  67 #include <machine/cpu.h>
  68
  69 #include <vm/vm.h>
  70 #include <vm/vm_param.h>
  71 #include <sys/lock.h>
  72 #include <vm/pmap.h>
  73 #include <vm/vm_map.h>
  74 #include <vm/vm_extern.h>
  75 #include <sys/user.h>
  76 #include <sys/copyright.h>
  77
  78 int vfs_mountroot_devfs(void);
  79
  80 /* Components of the first process -- never freed. */
  81 static struct session session0;
  82 static struct pgrp pgrp0;
  83 static struct sigacts sigacts0;
  84 static struct filedesc filedesc0;
  85 static struct plimit limit0;
  86 static struct vmspace vmspace0;
  87 struct proc *initproc;
  88 struct proc proc0;
  89 struct lwp lwp0;
  90 struct thread thread0;
  91 struct sys_kpmap *kpmap;
  92 struct sysreaper initreaper;
  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_OPAQUE(_kern, OID_AUTO, usched_global_cpumask, CTLFLAG_RW,
 103     &usched_global_cpumask, sizeof(usched_global_cpumask), "LU",
 104     "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         lwp0.lwp_tid = 1;
 171         proc0.p_lasttid = lwp0.lwp_tid; /* +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         struct uidinfo *uip;
 360
 361         p = &proc0;
 362         lp = &lwp0;
 363
 364         /*
 365          * Initialize osrel
 366          */
 367         p->p_osrel = osreldate;
 368
 369         /*
 370          * Initialize process and pgrp structures.
 371          */
 372         procinit();
 373
 374         /*
 375          * additional VM structures
 376          */
 377         vm_init2();
 378
 379         /*
 380          * Create process 0 (the swapper).
 381          */
 382         procinsertinit(p);
 383         pgrpinsertinit(&pgrp0);
 384         LIST_INIT(&pgrp0.pg_members);
 385         lwkt_token_init(&pgrp0.pg_token, "pgrp0");
 386         refcount_init(&pgrp0.pg_refs, 1);
 387         lockinit(&pgrp0.pg_lock, "pgwt0", 0, 0);
 388         LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
 389
 390         pgrp0.pg_session = &session0;
 391         session0.s_count = 1;
 392         session0.s_leader = p;
 393         sessinsertinit(&session0);
 394
 395         pgref(&pgrp0);
 396         p->p_pgrp = &pgrp0;
 397
 398         p->p_sysent = &aout_sysvec;
 399
 400         p->p_flags = P_SYSTEM;
 401         p->p_stat = SACTIVE;
 402         lp->lwp_stat = LSRUN;
 403         p->p_nice = NZERO;
 404         p->p_rtprio.type = RTP_PRIO_NORMAL;
 405         p->p_rtprio.prio = 0;
 406         lp->lwp_rtprio = p->p_rtprio;
 407
 408         p->p_peers = NULL;
 409         p->p_leader = p;
 410
 411         bcopy("swapper", p->p_comm, sizeof ("swapper"));
 412         bcopy("swapper", thread0.td_comm, sizeof ("swapper"));
 413
 414         /* Create credentials. */
 415         uip = uicreate(0);      /* for cr_ruidinfo */
 416         uihold(uip);            /* for cr_uidinfo */
 417         p->p_ucred = crget();
 418         p->p_ucred->cr_ruidinfo = uip;
 419         p->p_ucred->cr_uidinfo = uip;
 420         p->p_ucred->cr_ngroups = 1;     /* group 0 */
 421         thread0.td_ucred = crhold(p->p_ucred);  /* bootstrap fork1() */
 422
 423         /* Don't jail it */
 424         p->p_ucred->cr_prison = NULL;
 425
 426         /* Create sigacts. */
 427         p->p_sigacts = &sigacts0;
 428         refcount_init(&p->p_sigacts->ps_refcnt, 1);
 429
 430         /* Initialize signal state for process 0. */
 431         siginit(p);
 432
 433         /* Create the file descriptor table. */
 434         fdinit_bootstrap(p, &filedesc0, cmask);
 435
 436         /* Create the limits structures. */
 437         plimit_init0(&limit0);
 438         p->p_limit = &limit0;
 439
 440         /* Allocate a prototype map so we have something to fork. */
 441         pmap_pinit0(vmspace_pmap(&vmspace0));
 442         p->p_vmspace = &vmspace0;
 443         lp->lwp_vmspace = p->p_vmspace;
 444         vmspace_initrefs(&vmspace0);
 445         vm_map_init(&vmspace0.vm_map,
 446                     round_page(VM_MIN_USER_ADDRESS),
 447                     trunc_page(VM_MAX_USER_ADDRESS),
 448                     vmspace_pmap(&vmspace0));
 449
 450         kqueue_init(&lwp0.lwp_kqueue, &filedesc0);
 451
 452         /*
 453          * Charge root for one process.
 454          */
 455         (void)chgproccnt(p->p_ucred->cr_uidinfo, 1, 0);
 456         vm_init_limits(p);
 457 }
 458 SYSINIT(p0init, SI_BOOT2_PROC0, SI_ORDER_FIRST, proc0_init, NULL);
 459
 460 static int proc0_post_callback(struct proc *p, void *data __unused);
 461
 462 /* ARGSUSED*/
 463 static void
 464 proc0_post(void *dummy __unused)
 465 {
 466         struct timespec ts;
 467
 468         /*
 469          * Now we can look at the time, having had a chance to verify the
 470          * time from the file system.  Pretend that proc0 started now.
 471          */
 472         allproc_scan(proc0_post_callback, NULL, 0);
 473
 474         /*
 475          * Give the ``random'' number generator a thump.
 476          * XXX: Does read_random() contain enough bits to be used here ?
 477          */
 478         nanotime(&ts);
 479         skrandom(ts.tv_sec ^ ts.tv_nsec);
 480 }
 481
 482 static int
 483 proc0_post_callback(struct proc *p, void *data __unused)
 484 {
 485         microtime(&p->p_start);
 486         return(0);
 487 }
 488
 489 SYSINIT(p0post, SI_SUB_PROC0_POST, SI_ORDER_FIRST, proc0_post, NULL);
 490
 491 /*
 492  ***************************************************************************
 493  ****
 494  **** The following SYSINIT's and glue code should be moved to the
 495  **** respective files on a per subsystem basis.
 496  ****
 497  ***************************************************************************
 498  */
 499
 500
 501 /*
 502  ***************************************************************************
 503  ****
 504  **** The following code probably belongs in another file, like
 505  **** kern/init_init.c.
 506  ****
 507  ***************************************************************************
 508  */
 509
 510 /*
 511  * List of paths to try when searching for "init".
 512  */
 513 static char init_path[MAXPATHLEN] =
 514 #ifdef  INIT_PATH
 515     __XSTRING(INIT_PATH);
 516 #else
 517     "/sbin/init:/sbin/oinit:/sbin/init.bak";
 518 #endif
 519 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, "");
 520
 521 /*
 522  * Shutdown timeout of init(8).
 523  * Unused within kernel, but used to control init(8), hence do not remove.
 524  */
 525 #ifndef INIT_SHUTDOWN_TIMEOUT
 526 #define INIT_SHUTDOWN_TIMEOUT 120
 527 #endif
 528 static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT;
 529 SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout,
 530         CTLFLAG_RW, &init_shutdown_timeout, 0, "Shutdown timeout of init(8). "
 531         "Unused within kernel, but used to control init(8)");
 532
 533 /*
 534  * Start the initial user process; try exec'ing each pathname in init_path.
 535  * The program is invoked with one argument containing the boot flags.
 536  */
 537 static void
 538 start_init(void *dummy, struct trapframe *frame)
 539 {
 540         vm_offset_t addr;
 541         struct execve_args args;
 542         int options, error;
 543         char *var, *path, *next, *s;
 544         char *ucp, **uap, *arg0, *arg1;
 545         struct proc *p;
 546         struct lwp *lp;
 547         struct mount *mp;
 548         struct vnode *vp;
 549         char *env;
 550
 551         /*
 552          * This is passed in by the bootloader
 553          */
 554         env = kgetenv("kernelname");
 555         if (env != NULL)
 556                 strlcpy(kernelname, env, sizeof(kernelname));
 557
 558         p = curproc;
 559
 560         lp = ONLY_LWP_IN_PROC(p);
 561
 562         /* Get the vnode for '/'.  Set p->p_fd->fd_cdir to reference it. */
 563         mp = mountlist_boot_getfirst();
 564         if (VFS_ROOT(mp, &vp))
 565                 panic("cannot find root vnode");
 566         if (mp->mnt_ncmountpt.ncp == NULL) {
 567                 cache_allocroot(&mp->mnt_ncmountpt, mp, vp);
 568                 cache_unlock(&mp->mnt_ncmountpt);       /* leave ref intact */
 569         }
 570         p->p_fd->fd_cdir = vp;
 571         vref(p->p_fd->fd_cdir);
 572         p->p_fd->fd_rdir = vp;
 573         vref(p->p_fd->fd_rdir);
 574         vfs_cache_setroot(vp, cache_hold(&mp->mnt_ncmountpt));
 575         vn_unlock(vp);                  /* leave ref intact */
 576         cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_ncdir);
 577         cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_nrdir);
 578
 579         kprintf("Mounting devfs\n");
 580         vfs_mountroot_devfs();
 581
 582         /*
 583          * Need just enough stack to hold the faked-up "execve()" arguments.
 584          */
 585         addr = trunc_page(USRSTACK - PAGE_SIZE);
 586         error = vm_map_find(&p->p_vmspace->vm_map, NULL, NULL,
 587                             0, &addr, PAGE_SIZE,
 588                             PAGE_SIZE, FALSE,
 589                             VM_MAPTYPE_NORMAL, VM_SUBSYS_INIT,
 590                             VM_PROT_ALL, VM_PROT_ALL, 0);
 591         if (error)
 592                 panic("init: couldn't allocate argument space");
 593         p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
 594         p->p_vmspace->vm_ssize = 1;
 595
 596         if ((var = kgetenv("init_path")) != NULL) {
 597                 strncpy(init_path, var, sizeof init_path);
 598                 init_path[sizeof init_path - 1] = 0;
 599         }
 600
 601         for (path = init_path; *path != '\0'; path = next) {
 602                 while (*path == ':')
 603                         path++;
 604                 if (*path == '\0')
 605                         break;
 606                 for (next = path; *next != '\0' && *next != ':'; next++)
 607                         /* nothing */ ;
 608                 if (bootverbose)
 609                         kprintf("start_init: trying %.*s\n", (int)(next - path),
 610                             path);
 611
 612                 /*
 613                  * Move out the boot flag argument.
 614                  */
 615                 options = 0;
 616                 ucp = (char *)USRSTACK;
 617                 (void)subyte(--ucp, 0);         /* trailing zero */
 618                 if (boothowto & RB_SINGLE) {
 619                         (void)subyte(--ucp, 's');
 620                         options = 1;
 621                 }
 622 #ifdef notyet
 623                 if (boothowto & RB_FASTBOOT) {
 624                         (void)subyte(--ucp, 'f');
 625                         options = 1;
 626                 }
 627 #endif
 628
 629 #ifdef BOOTCDROM
 630                 (void)subyte(--ucp, 'C');
 631                 options = 1;
 632 #endif
 633                 if (options == 0)
 634                         (void)subyte(--ucp, '-');
 635                 (void)subyte(--ucp, '-');               /* leading hyphen */
 636                 arg1 = ucp;
 637
 638                 /*
 639                  * Move out the file name (also arg 0).
 640                  */
 641                 (void)subyte(--ucp, 0);
 642                 for (s = next - 1; s >= path; s--)
 643                         (void)subyte(--ucp, *s);
 644                 arg0 = ucp;
 645
 646                 /*
 647                  * Move out the arg pointers.
 648                  */
 649                 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
 650
 651                 /* terminator */
 652                 suword64((uint64_t *)(caddr_t)--uap, (uint64_t)0);
 653
 654                 suword64((uint64_t *)(caddr_t)--uap, (uint64_t)(intptr_t)arg1);
 655                 suword64((uint64_t *)(caddr_t)--uap, (uint64_t)(intptr_t)arg0);
 656
 657                 /*
 658                  * Point at the arguments.
 659                  */
 660                 args.fname = arg0;
 661                 args.argv = uap;
 662                 args.envv = NULL;
 663
 664                 /*
 665                  * Now try to exec the program.  If can't for any reason
 666                  * other than it doesn't exist, complain.
 667                  *
 668                  * Otherwise, return via fork_trampoline() all the way
 669                  * to user mode as init!
 670                  *
 671                  * WARNING!  We may have been moved to another cpu after
 672                  * acquiring the current user process designation.  The
 673                  * MP lock will migrate with us though so we still have to
 674                  * release it.
 675                  */
 676                 if ((error = sys_execve(&args)) == 0) {
 677                         lp->lwp_proc->p_usched->acquire_curproc(lp);
 678                         return;
 679                 }
 680                 if (error != ENOENT)
 681                         kprintf("exec %.*s: error %d\n", (int)(next - path),
 682                             path, error);
 683         }
 684         kprintf("init: not found in path %s\n", init_path);
 685         panic("no init");
 686 }
 687
 688 /*
 689  * Like kthread_create(), but runs in it's own address space.
 690  * We do this early to reserve pid 1.
 691  *
 692  * Note special case - do not make it runnable yet.  Other work
 693  * in progress will change this more.
 694  */
 695 static void
 696 create_init(const void *udata __unused)
 697 {
 698         int error;
 699         struct lwp *lp;
 700
 701         crit_enter();
 702         error = fork1(&lwp0, RFFDG | RFPROC, &initproc);
 703         if (error)
 704                 panic("cannot fork init: %d", error);
 705         initproc->p_flags |= P_SYSTEM;
 706         reaper_init(initproc, &initreaper);
 707         lp = ONLY_LWP_IN_PROC(initproc);
 708         cpu_set_fork_handler(lp, start_init, NULL);
 709         crit_exit();
 710 }
 711 SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL);
 712
 713 /*
 714  * Make it runnable now.
 715  */
 716 static void
 717 kick_init(const void *udata __unused)
 718 {
 719         start_forked_proc(&lwp0, initproc);
 720 }
 721 SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL);
 722
 723 static void
 724 kpmap_init(const void *udata __unused)
 725 {
 726         kpmap = kmalloc(roundup2(sizeof(*kpmap), PAGE_SIZE),
 727                         M_TEMP, M_ZERO | M_WAITOK);
 728
 729         kpmap->header[0].type = UKPTYPE_VERSION;
 730         kpmap->header[0].offset = offsetof(struct sys_kpmap, version);
 731         kpmap->header[1].type = KPTYPE_UPTICKS;
 732         kpmap->header[1].offset = offsetof(struct sys_kpmap, upticks);
 733         kpmap->header[2].type = KPTYPE_TS_UPTIME;
 734         kpmap->header[2].offset = offsetof(struct sys_kpmap, ts_uptime);
 735         kpmap->header[3].type = KPTYPE_TS_REALTIME;
 736         kpmap->header[3].offset = offsetof(struct sys_kpmap, ts_realtime);
 737         kpmap->header[4].type = KPTYPE_TSC_FREQ;
 738         kpmap->header[4].offset = offsetof(struct sys_kpmap, tsc_freq);
 739         kpmap->header[5].type = KPTYPE_TICK_FREQ;
 740         kpmap->header[5].offset = offsetof(struct sys_kpmap, tick_freq);
 741         kpmap->version = KPMAP_VERSION;
 742 }
 743 SYSINIT(kpmapinit, SI_BOOT1_POST, SI_ORDER_FIRST, kpmap_init, NULL);
 744
 745 /*
 746  * Machine independant globaldata initialization
 747  *
 748  * WARNING!  Called from early boot, 'mycpu' may not work yet.
 749  */
 750 void
 751 mi_gdinit(struct globaldata *gd, int cpuid)
 752 {
 753         TAILQ_INIT(&gd->gd_systimerq);
 754         gd->gd_sysid_alloc = cpuid;     /* prime low bits for cpu lookup */
 755         gd->gd_cpuid = cpuid;
 756         CPUMASK_ASSBIT(gd->gd_cpumask, cpuid);
 757         gd->gd_cpumask_simple = CPUMASK_SIMPLE(cpuid);
 758         gd->gd_cpumask_offset = (uintptr_t)CPUMASK_ADDR(*(cpumask_t *)0, cpuid);
 759         lwkt_gdinit(gd);
 760         vm_map_entry_reserve_cpu_init(gd);
 761         if (gd->gd_cpuid == 0)
 762                 sleep_early_gdinit(gd);
 763         else
 764                 sleep_gdinit(gd);
 765         slab_gdinit(gd);
 766         ATOMIC_CPUMASK_ORBIT(usched_global_cpumask, cpuid);
 767         gd->gd_vmstats = vmstats;
 768 }
 769