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/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 struct sysreaper initreaper;
  95
  96 int cmask = CMASK;
  97 u_int cpu_mi_feature;
  98 cpumask_t usched_global_cpumask;
  99 extern  struct user *proc0paddr;
 100
 101 int     boothowto = 0;          /* initialized so that it can be patched */
 102 SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0,
 103     "Reboot flags, from console subsystem");
 104 SYSCTL_OPAQUE(_kern, OID_AUTO, usched_global_cpumask, CTLFLAG_RW,
 105     &usched_global_cpumask, sizeof(usched_global_cpumask), "LU",
 106     "global user scheduler cpumask");
 107
 108 /*
 109  * This ensures that there is at least one entry so that the sysinit_set
 110  * symbol is not undefined.  A subsystem ID of SI_SPECIAL_DUMMY is never
 111  * executed.
 112  */
 113 SYSINIT(placeholder, SI_SPECIAL_DUMMY, SI_ORDER_ANY, NULL, NULL);
 114
 115 /*
 116  * The sysinit table itself.  Items are checked off as the are run.
 117  * If we want to register new sysinit types, add them to newsysinit.
 118  */
 119 SET_DECLARE(sysinit_set, struct sysinit);
 120 struct sysinit **sysinit, **sysinit_end;
 121 struct sysinit **newsysinit, **newsysinit_end;
 122
 123
 124 /*
 125  * Merge a new sysinit set into the current set, reallocating it if
 126  * necessary.  This can only be called after malloc is running.
 127  */
 128 void
 129 sysinit_add(struct sysinit **set, struct sysinit **set_end)
 130 {
 131         struct sysinit **newset;
 132         struct sysinit **sipp;
 133         struct sysinit **xipp;
 134         int count;
 135
 136         count = set_end - set;
 137         if (newsysinit)
 138                 count += newsysinit_end - newsysinit;
 139         else
 140                 count += sysinit_end - sysinit;
 141         newset = kmalloc(count * sizeof(*sipp), M_TEMP, M_WAITOK);
 142         xipp = newset;
 143         if (newsysinit) {
 144                 for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
 145                         *xipp++ = *sipp;
 146         } else {
 147                 for (sipp = sysinit; sipp < sysinit_end; sipp++)
 148                         *xipp++ = *sipp;
 149         }
 150         for (sipp = set; sipp < set_end; sipp++)
 151                 *xipp++ = *sipp;
 152         if (newsysinit)
 153                 kfree(newsysinit, M_TEMP);
 154         newsysinit = newset;
 155         newsysinit_end = newset + count;
 156 }
 157
 158 /*
 159  * Callbacks from machine-dependant startup code (e.g. init386) to set
 160  * up low level entities related to cpu #0's globaldata.
 161  *
 162  * Called from very low level boot code.
 163  */
 164 void
 165 mi_proc0init(struct globaldata *gd, struct user *proc0paddr)
 166 {
 167         lwkt_init_thread(&thread0, proc0paddr, LWKT_THREAD_STACK, 0, gd);
 168         lwkt_set_comm(&thread0, "thread0");
 169         RB_INIT(&proc0.p_lwp_tree);
 170         spin_init(&proc0.p_spin, "iproc_proc0");
 171         lwkt_token_init(&proc0.p_token, "iproc");
 172         proc0.p_lasttid = 0;    /* +1 = next TID */
 173         lwp_rb_tree_RB_INSERT(&proc0.p_lwp_tree, &lwp0);
 174         lwp0.lwp_thread = &thread0;
 175         lwp0.lwp_proc = &proc0;
 176         proc0.p_usched = usched_init();
 177         CPUMASK_ASSALLONES(lwp0.lwp_cpumask);
 178         lwkt_token_init(&lwp0.lwp_token, "lwp_token");
 179         spin_init(&lwp0.lwp_spin, "iproc_lwp0");
 180         varsymset_init(&proc0.p_varsymset, NULL);
 181         thread0.td_flags |= TDF_RUNNING;
 182         thread0.td_proc = &proc0;
 183         thread0.td_lwp = &lwp0;
 184         thread0.td_switch = cpu_lwkt_switch;
 185         lwkt_schedule_self(curthread);
 186 }
 187
 188 /*
 189  * System startup; initialize the world, create process 0, mount root
 190  * filesystem, and fork to create init and pagedaemon.  Most of the
 191  * hard work is done in the lower-level initialization routines including
 192  * startup(), which does memory initialization and autoconfiguration.
 193  *
 194  * This allows simple addition of new kernel subsystems that require
 195  * boot time initialization.  It also allows substitution of subsystem
 196  * (for instance, a scheduler, kernel profiler, or VM system) by object
 197  * module.  Finally, it allows for optional "kernel threads".
 198  */
 199 void
 200 mi_startup(void)
 201 {
 202         struct sysinit *sip;            /* system initialization*/
 203         struct sysinit **sipp;          /* system initialization*/
 204         struct sysinit **xipp;          /* interior loop of sort*/
 205         struct sysinit *save;           /* bubble*/
 206
 207         if (sysinit == NULL) {
 208                 sysinit = SET_BEGIN(sysinit_set);
 209 #if defined(__x86_64__) && defined(_KERNEL_VIRTUAL)
 210                 /*
 211                  * XXX For whatever reason, on 64-bit vkernels
 212                  * the value of sysinit obtained from the
 213                  * linker set is wrong.
 214                  */
 215                 if ((long)sysinit % 8 != 0) {
 216                         kprintf("Fixing sysinit value...\n");
 217                         sysinit = (void *)((long)(intptr_t)sysinit + 4);
 218                 }
 219 #endif
 220                 sysinit_end = SET_LIMIT(sysinit_set);
 221         }
 222 #if defined(__x86_64__) && defined(_KERNEL_VIRTUAL)
 223         KKASSERT((long)sysinit % 8 == 0);
 224 #endif
 225
 226 restart:
 227         /*
 228          * Perform a bubble sort of the system initialization objects by
 229          * their subsystem (primary key) and order (secondary key).
 230          */
 231         for (sipp = sysinit; sipp < sysinit_end; sipp++) {
 232                 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
 233                         if ((*sipp)->subsystem < (*xipp)->subsystem ||
 234                              ((*sipp)->subsystem == (*xipp)->subsystem &&
 235                               (*sipp)->order <= (*xipp)->order))
 236                                 continue;       /* skip*/
 237                         save = *sipp;
 238                         *sipp = *xipp;
 239                         *xipp = save;
 240                 }
 241         }
 242
 243         /*
 244          * Traverse the (now) ordered list of system initialization tasks.
 245          * Perform each task, and continue on to the next task.
 246          *
 247          * The last item on the list is expected to be the scheduler,
 248          * which will not return.
 249          */
 250         for (sipp = sysinit; sipp < sysinit_end; sipp++) {
 251                 sip = *sipp;
 252                 if (sip->subsystem == SI_SPECIAL_DUMMY)
 253                         continue;       /* skip dummy task(s)*/
 254
 255                 if (sip->subsystem == SI_SPECIAL_DONE)
 256                         continue;
 257
 258 #if 0
 259                 if (bootverbose)
 260                         kprintf("(%08x-%p)\n", sip->subsystem, sip->func);
 261 #endif
 262
 263                 /* Call function */
 264                 (*(sip->func))(sip->udata);
 265
 266                 /* Check off the one we're just done */
 267                 sip->subsystem = SI_SPECIAL_DONE;
 268
 269                 /* Check if we've installed more sysinit items via KLD */
 270                 if (newsysinit != NULL) {
 271                         if (sysinit != SET_BEGIN(sysinit_set))
 272                                 kfree(sysinit, M_TEMP);
 273                         sysinit = newsysinit;
 274                         sysinit_end = newsysinit_end;
 275                         newsysinit = NULL;
 276                         newsysinit_end = NULL;
 277                         goto restart;
 278                 }
 279         }
 280
 281         panic("Shouldn't get here!");
 282         /* NOTREACHED*/
 283 }
 284
 285
 286 /*
 287  ***************************************************************************
 288  ****
 289  **** The following SYSINIT's belong elsewhere, but have not yet
 290  **** been moved.
 291  ****
 292  ***************************************************************************
 293  */
 294 static void
 295 print_caddr_t(void *data)
 296 {
 297         kprintf("%s", (char *)data);
 298 }
 299 SYSINIT(announce, SI_BOOT1_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright);
 300
 301 /*
 302  * Leave the critical section that protected us from spurious interrupts
 303  * so device probes work.
 304  */
 305 static void
 306 leavecrit(void *dummy __unused)
 307 {
 308         MachIntrABI.stabilize();
 309         cpu_enable_intr();
 310         MachIntrABI.cleanup();
 311         crit_exit();
 312         KKASSERT(!IN_CRITICAL_SECT(curthread));
 313
 314         if (bootverbose)
 315                 kprintf("Leaving critical section, allowing interrupts\n");
 316 }
 317 SYSINIT(leavecrit, SI_BOOT2_LEAVE_CRIT, SI_ORDER_ANY, leavecrit, NULL);
 318
 319 /*
 320  * This is called after the threading system is up and running,
 321  * including the softclock, clock interrupts, and SMP.
 322  */
 323 static void
 324 tsleepworks(void *dummy __unused)
 325 {
 326         tsleep_now_works = 1;
 327 }
 328 SYSINIT(tsleepworks, SI_BOOT2_FINISH_SMP, SI_ORDER_SECOND, tsleepworks, NULL);
 329
 330 /*
 331  * This is called after devices have configured.  Tell the kernel we are
 332  * no longer in cold boot.
 333  */
 334 static void
 335 endofcoldboot(void *dummy __unused)
 336 {
 337         cold = 0;
 338 }
 339 SYSINIT(endofcoldboot, SI_SUB_ISWARM, SI_ORDER_ANY, endofcoldboot, NULL);
 340
 341 /*
 342  ***************************************************************************
 343  ****
 344  **** The two following SYSINT's are proc0 specific glue code.  I am not
 345  **** convinced that they can not be safely combined, but their order of
 346  **** operation has been maintained as the same as the original init_main.c
 347  **** for right now.
 348  ****
 349  **** These probably belong in init_proc.c or kern_proc.c, since they
 350  **** deal with proc0 (the fork template process).
 351  ****
 352  ***************************************************************************
 353  */
 354 /* ARGSUSED*/
 355 static void
 356 proc0_init(void *dummy __unused)
 357 {
 358         struct proc *p;
 359         struct lwp *lp;
 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         p->p_ucred = crget();
 416         p->p_ucred->cr_ruidinfo = uifind(0);
 417         p->p_ucred->cr_ngroups = 1;     /* group 0 */
 418         p->p_ucred->cr_uidinfo = uifind(0);
 419         thread0.td_ucred = crhold(p->p_ucred);  /* bootstrap fork1() */
 420
 421         /* Don't jail it */
 422         p->p_ucred->cr_prison = NULL;
 423
 424         /* Create sigacts. */
 425         p->p_sigacts = &sigacts0;
 426         refcount_init(&p->p_sigacts->ps_refcnt, 1);
 427
 428         /* Initialize signal state for process 0. */
 429         siginit(p);
 430
 431         /* Create the file descriptor table. */
 432         fdinit_bootstrap(p, &filedesc0, cmask);
 433
 434         /* Create the limits structures. */
 435         plimit_init0(&limit0);
 436         p->p_limit = &limit0;
 437
 438         /* Allocate a prototype map so we have something to fork. */
 439         pmap_pinit0(vmspace_pmap(&vmspace0));
 440         p->p_vmspace = &vmspace0;
 441         lp->lwp_vmspace = p->p_vmspace;
 442         vmspace_initrefs(&vmspace0);
 443         vm_map_init(&vmspace0.vm_map,
 444                     round_page(VM_MIN_USER_ADDRESS),
 445                     trunc_page(VM_MAX_USER_ADDRESS),
 446                     vmspace_pmap(&vmspace0));
 447
 448         kqueue_init(&lwp0.lwp_kqueue, &filedesc0);
 449
 450         /*
 451          * Charge root for one process.
 452          */
 453         (void)chgproccnt(p->p_ucred->cr_uidinfo, 1, 0);
 454         vm_init_limits(p);
 455 }
 456 SYSINIT(p0init, SI_BOOT2_PROC0, SI_ORDER_FIRST, proc0_init, NULL);
 457
 458 static int proc0_post_callback(struct proc *p, void *data __unused);
 459
 460 /* ARGSUSED*/
 461 static void
 462 proc0_post(void *dummy __unused)
 463 {
 464         struct timespec ts;
 465
 466         /*
 467          * Now we can look at the time, having had a chance to verify the
 468          * time from the file system.  Pretend that proc0 started now.
 469          */
 470         allproc_scan(proc0_post_callback, NULL);
 471
 472         /*
 473          * Give the ``random'' number generator a thump.
 474          * XXX: Does read_random() contain enough bits to be used here ?
 475          */
 476         nanotime(&ts);
 477         skrandom(ts.tv_sec ^ ts.tv_nsec);
 478 }
 479
 480 static int
 481 proc0_post_callback(struct proc *p, void *data __unused)
 482 {
 483         microtime(&p->p_start);
 484         return(0);
 485 }
 486
 487 SYSINIT(p0post, SI_SUB_PROC0_POST, SI_ORDER_FIRST, proc0_post, NULL);
 488
 489 /*
 490  ***************************************************************************
 491  ****
 492  **** The following SYSINIT's and glue code should be moved to the
 493  **** respective files on a per subsystem basis.
 494  ****
 495  ***************************************************************************
 496  */
 497
 498
 499 /*
 500  ***************************************************************************
 501  ****
 502  **** The following code probably belongs in another file, like
 503  **** kern/init_init.c.
 504  ****
 505  ***************************************************************************
 506  */
 507
 508 /*
 509  * List of paths to try when searching for "init".
 510  */
 511 static char init_path[MAXPATHLEN] =
 512 #ifdef  INIT_PATH
 513     __XSTRING(INIT_PATH);
 514 #else
 515     "/sbin/init:/sbin/oinit:/sbin/init.bak";
 516 #endif
 517 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, "");
 518
 519 /*
 520  * Shutdown timeout of init(8).
 521  * Unused within kernel, but used to control init(8), hence do not remove.
 522  */
 523 #ifndef INIT_SHUTDOWN_TIMEOUT
 524 #define INIT_SHUTDOWN_TIMEOUT 120
 525 #endif
 526 static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT;
 527 SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout,
 528         CTLFLAG_RW, &init_shutdown_timeout, 0, "Shutdown timeout of init(8). "
 529         "Unused within kernel, but used to control init(8)");
 530
 531 /*
 532  * Start the initial user process; try exec'ing each pathname in init_path.
 533  * The program is invoked with one argument containing the boot flags.
 534  */
 535 static void
 536 start_init(void *dummy, struct trapframe *frame)
 537 {
 538         vm_offset_t addr;
 539         struct execve_args args;
 540         int options, error;
 541         char *var, *path, *next, *s;
 542         char *ucp, **uap, *arg0, *arg1;
 543         struct proc *p;
 544         struct lwp *lp;
 545         struct mount *mp;
 546         struct vnode *vp;
 547         char *env;
 548
 549         /*
 550          * This is passed in by the bootloader
 551          */
 552         env = kgetenv("kernelname");
 553         if (env != NULL)
 554                 strlcpy(kernelname, env, sizeof(kernelname));
 555
 556         /*
 557          * The MP lock is not held on entry.  We release it before
 558          * returning to userland.
 559          */
 560         get_mplock();
 561         p = curproc;
 562
 563         lp = ONLY_LWP_IN_PROC(p);
 564
 565         /* Get the vnode for '/'.  Set p->p_fd->fd_cdir to reference it. */
 566         mp = mountlist_boot_getfirst();
 567         if (VFS_ROOT(mp, &vp))
 568                 panic("cannot find root vnode");
 569         if (mp->mnt_ncmountpt.ncp == NULL) {
 570                 cache_allocroot(&mp->mnt_ncmountpt, mp, vp);
 571                 cache_unlock(&mp->mnt_ncmountpt);       /* leave ref intact */
 572         }
 573         p->p_fd->fd_cdir = vp;
 574         vref(p->p_fd->fd_cdir);
 575         p->p_fd->fd_rdir = vp;
 576         vref(p->p_fd->fd_rdir);
 577         vfs_cache_setroot(vp, cache_hold(&mp->mnt_ncmountpt));
 578         vn_unlock(vp);                  /* leave ref intact */
 579         cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_ncdir);
 580         cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_nrdir);
 581
 582         kprintf("Mounting devfs\n");
 583         vfs_mountroot_devfs();
 584
 585         /*
 586          * Need just enough stack to hold the faked-up "execve()" arguments.
 587          */
 588         addr = trunc_page(USRSTACK - PAGE_SIZE);
 589         error = vm_map_find(&p->p_vmspace->vm_map, NULL, NULL,
 590                             0, &addr, PAGE_SIZE,
 591                             PAGE_SIZE, FALSE,
 592                             VM_MAPTYPE_NORMAL, VM_SUBSYS_INIT,
 593                             VM_PROT_ALL, VM_PROT_ALL, 0);
 594         if (error)
 595                 panic("init: couldn't allocate argument space");
 596         p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
 597         p->p_vmspace->vm_ssize = 1;
 598
 599         if ((var = kgetenv("init_path")) != NULL) {
 600                 strncpy(init_path, var, sizeof init_path);
 601                 init_path[sizeof init_path - 1] = 0;
 602         }
 603
 604         for (path = init_path; *path != '\0'; path = next) {
 605                 while (*path == ':')
 606                         path++;
 607                 if (*path == '\0')
 608                         break;
 609                 for (next = path; *next != '\0' && *next != ':'; next++)
 610                         /* nothing */ ;
 611                 if (bootverbose)
 612                         kprintf("start_init: trying %.*s\n", (int)(next - path),
 613                             path);
 614
 615                 /*
 616                  * Move out the boot flag argument.
 617                  */
 618                 options = 0;
 619                 ucp = (char *)USRSTACK;
 620                 (void)subyte(--ucp, 0);         /* trailing zero */
 621                 if (boothowto & RB_SINGLE) {
 622                         (void)subyte(--ucp, 's');
 623                         options = 1;
 624                 }
 625 #ifdef notyet
 626                 if (boothowto & RB_FASTBOOT) {
 627                         (void)subyte(--ucp, 'f');
 628                         options = 1;
 629                 }
 630 #endif
 631
 632 #ifdef BOOTCDROM
 633                 (void)subyte(--ucp, 'C');
 634                 options = 1;
 635 #endif
 636                 if (options == 0)
 637                         (void)subyte(--ucp, '-');
 638                 (void)subyte(--ucp, '-');               /* leading hyphen */
 639                 arg1 = ucp;
 640
 641                 /*
 642                  * Move out the file name (also arg 0).
 643                  */
 644                 (void)subyte(--ucp, 0);
 645                 for (s = next - 1; s >= path; s--)
 646                         (void)subyte(--ucp, *s);
 647                 arg0 = ucp;
 648
 649                 /*
 650                  * Move out the arg pointers.
 651                  */
 652                 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
 653                 (void)suword((caddr_t)--uap, (long)0);  /* terminator */
 654                 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
 655                 (void)suword((caddr_t)--uap, (long)(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                         rel_mplock();
 678                         lp->lwp_proc->p_usched->acquire_curproc(lp);
 679                         return;
 680                 }
 681                 if (error != ENOENT)
 682                         kprintf("exec %.*s: error %d\n", (int)(next - path),
 683                             path, error);
 684         }
 685         kprintf("init: not found in path %s\n", init_path);
 686         panic("no init");
 687 }
 688
 689 /*
 690  * Like kthread_create(), but runs in it's own address space.
 691  * We do this early to reserve pid 1.
 692  *
 693  * Note special case - do not make it runnable yet.  Other work
 694  * in progress will change this more.
 695  */
 696 static void
 697 create_init(const void *udata __unused)
 698 {
 699         int error;
 700         struct lwp *lp;
 701
 702         crit_enter();
 703         error = fork1(&lwp0, RFFDG | RFPROC, &initproc);
 704         if (error)
 705                 panic("cannot fork init: %d", error);
 706         initproc->p_flags |= P_SYSTEM;
 707         reaper_init(initproc, &initreaper);
 708         lp = ONLY_LWP_IN_PROC(initproc);
 709         cpu_set_fork_handler(lp, start_init, NULL);
 710         crit_exit();
 711 }
 712 SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL);
 713
 714 /*
 715  * Make it runnable now.
 716  */
 717 static void
 718 kick_init(const void *udata __unused)
 719 {
 720         start_forked_proc(&lwp0, initproc);
 721 }
 722 SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL);
 723
 724 static void
 725 kpmap_init(const void *udata __unused)
 726 {
 727         kpmap = kmalloc(roundup2(sizeof(*kpmap), PAGE_SIZE),
 728                         M_TEMP, M_ZERO | M_WAITOK);
 729
 730         kpmap->header[0].type = UKPTYPE_VERSION;
 731         kpmap->header[0].offset = offsetof(struct sys_kpmap, version);
 732         kpmap->header[1].type = KPTYPE_UPTICKS;
 733         kpmap->header[1].offset = offsetof(struct sys_kpmap, upticks);
 734         kpmap->header[2].type = KPTYPE_TS_UPTIME;
 735         kpmap->header[2].offset = offsetof(struct sys_kpmap, ts_uptime);
 736         kpmap->header[3].type = KPTYPE_TS_REALTIME;
 737         kpmap->header[3].offset = offsetof(struct sys_kpmap, ts_realtime);
 738         kpmap->header[4].type = KPTYPE_TSC_FREQ;
 739         kpmap->header[4].offset = offsetof(struct sys_kpmap, tsc_freq);
 740         kpmap->header[5].type = KPTYPE_TICK_FREQ;
 741         kpmap->header[5].offset = offsetof(struct sys_kpmap, tick_freq);
 742         kpmap->version = KPMAP_VERSION;
 743 }
 744 SYSINIT(kpmapinit, SI_BOOT1_POST, SI_ORDER_FIRST, kpmap_init, NULL);
 745
 746 /*
 747  * Machine independant globaldata initialization
 748  *
 749  * WARNING!  Called from early boot, 'mycpu' may not work yet.
 750  */
 751 void
 752 mi_gdinit(struct globaldata *gd, int cpuid)
 753 {
 754         TAILQ_INIT(&gd->gd_systimerq);
 755         gd->gd_sysid_alloc = cpuid;     /* prime low bits for cpu lookup */
 756         gd->gd_cpuid = cpuid;
 757         CPUMASK_ASSBIT(gd->gd_cpumask, cpuid);
 758         gd->gd_cpumask_simple = CPUMASK_SIMPLE(cpuid);
 759         gd->gd_cpumask_offset = (uintptr_t)CPUMASK_ADDR(*(cpumask_t *)0, cpuid);
 760         lwkt_gdinit(gd);
 761         vm_map_entry_reserve_cpu_init(gd);
 762         sleep_gdinit(gd);
 763         slab_gdinit(gd);
 764         ATOMIC_CPUMASK_ORBIT(usched_global_cpumask, cpuid);
 765 }
 766