The Design and Implementation of the FreeBSD Operating System, Second Edition
Now available: The Design and Implementation of the FreeBSD Operating System (Second Edition)


[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]

FreeBSD/Linux Kernel Cross Reference
sys/kern/init_main.c

Version: -  FREEBSD  -  FREEBSD-13-STABLE  -  FREEBSD-13-0  -  FREEBSD-12-STABLE  -  FREEBSD-12-0  -  FREEBSD-11-STABLE  -  FREEBSD-11-0  -  FREEBSD-10-STABLE  -  FREEBSD-10-0  -  FREEBSD-9-STABLE  -  FREEBSD-9-0  -  FREEBSD-8-STABLE  -  FREEBSD-8-0  -  FREEBSD-7-STABLE  -  FREEBSD-7-0  -  FREEBSD-6-STABLE  -  FREEBSD-6-0  -  FREEBSD-5-STABLE  -  FREEBSD-5-0  -  FREEBSD-4-STABLE  -  FREEBSD-3-STABLE  -  FREEBSD22  -  l41  -  OPENBSD  -  linux-2.6  -  MK84  -  PLAN9  -  xnu-8792 
SearchContext: -  none  -  3  -  10 

    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. All advertising materials mentioning features or use of this software
   22  *    must display the following acknowledgement:
   23  *      This product includes software developed by the University of
   24  *      California, Berkeley and its contributors.
   25  * 4. Neither the name of the University nor the names of its contributors
   26  *    may be used to endorse or promote products derived from this software
   27  *    without specific prior written permission.
   28  *
   29  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
   30  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
   31  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
   32  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
   33  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
   34  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
   35  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
   36  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
   37  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
   38  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
   39  * SUCH DAMAGE.
   40  *
   41  *      @(#)init_main.c 8.9 (Berkeley) 1/21/94
   42  */
   43 
   44 #include <sys/cdefs.h>
   45 __FBSDID("$FreeBSD: releng/8.4/sys/kern/init_main.c 247875 2013-03-06 09:16:01Z avg $");
   46 
   47 #include "opt_ddb.h"
   48 #include "opt_init_path.h"
   49 
   50 #include <sys/param.h>
   51 #include <sys/kernel.h>
   52 #include <sys/exec.h>
   53 #include <sys/file.h>
   54 #include <sys/filedesc.h>
   55 #include <sys/jail.h>
   56 #include <sys/ktr.h>
   57 #include <sys/lock.h>
   58 #include <sys/mount.h>
   59 #include <sys/mutex.h>
   60 #include <sys/syscallsubr.h>
   61 #include <sys/sysctl.h>
   62 #include <sys/proc.h>
   63 #include <sys/resourcevar.h>
   64 #include <sys/systm.h>
   65 #include <sys/signalvar.h>
   66 #include <sys/vnode.h>
   67 #include <sys/sysent.h>
   68 #include <sys/reboot.h>
   69 #include <sys/sched.h>
   70 #include <sys/sx.h>
   71 #include <sys/sysproto.h>
   72 #include <sys/vmmeter.h>
   73 #include <sys/unistd.h>
   74 #include <sys/malloc.h>
   75 #include <sys/conf.h>
   76 #include <sys/cpuset.h>
   77 
   78 #include <machine/cpu.h>
   79 
   80 #include <security/audit/audit.h>
   81 #include <security/mac/mac_framework.h>
   82 
   83 #include <vm/vm.h>
   84 #include <vm/vm_param.h>
   85 #include <vm/pmap.h>
   86 #include <vm/vm_map.h>
   87 #include <sys/copyright.h>
   88 
   89 #include <ddb/ddb.h>
   90 #include <ddb/db_sym.h>
   91 
   92 void mi_startup(void);                          /* Should be elsewhere */
   93 
   94 /* Components of the first process -- never freed. */
   95 static struct session session0;
   96 static struct pgrp pgrp0;
   97 struct  proc proc0;
   98 struct  thread thread0 __aligned(16);
   99 struct  vmspace vmspace0;
  100 struct  proc *initproc;
  101 
  102 int     boothowto = 0;          /* initialized so that it can be patched */
  103 SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, "");
  104 int     bootverbose;
  105 SYSCTL_INT(_debug, OID_AUTO, bootverbose, CTLFLAG_RW, &bootverbose, 0, "");
  106 
  107 /*
  108  * This ensures that there is at least one entry so that the sysinit_set
  109  * symbol is not undefined.  A sybsystem ID of SI_SUB_DUMMY is never
  110  * executed.
  111  */
  112 SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL);
  113 
  114 /*
  115  * The sysinit table itself.  Items are checked off as the are run.
  116  * If we want to register new sysinit types, add them to newsysinit.
  117  */
  118 SET_DECLARE(sysinit_set, struct sysinit);
  119 struct sysinit **sysinit, **sysinit_end;
  120 struct sysinit **newsysinit, **newsysinit_end;
  121 
  122 /*
  123  * 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 = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
  140         if (newset == NULL)
  141                 panic("cannot malloc for sysinit");
  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         for (sipp = set; sipp < set_end; sipp++)
  150                 *xipp++ = *sipp;
  151         if (newsysinit)
  152                 free(newsysinit, M_TEMP);
  153         newsysinit = newset;
  154         newsysinit_end = newset + count;
  155 }
  156 
  157 /*
  158  * System startup; initialize the world, create process 0, mount root
  159  * filesystem, and fork to create init and pagedaemon.  Most of the
  160  * hard work is done in the lower-level initialization routines including
  161  * startup(), which does memory initialization and autoconfiguration.
  162  *
  163  * This allows simple addition of new kernel subsystems that require
  164  * boot time initialization.  It also allows substitution of subsystem
  165  * (for instance, a scheduler, kernel profiler, or VM system) by object
  166  * module.  Finally, it allows for optional "kernel threads".
  167  */
  168 void
  169 mi_startup(void)
  170 {
  171 
  172         register struct sysinit **sipp;         /* system initialization*/
  173         register struct sysinit **xipp;         /* interior loop of sort*/
  174         register struct sysinit *save;          /* bubble*/
  175 
  176 #if defined(VERBOSE_SYSINIT)
  177         int last;
  178         int verbose;
  179 #endif
  180 
  181         if (boothowto & RB_VERBOSE)
  182                 bootverbose++;
  183 
  184         if (sysinit == NULL) {
  185                 sysinit = SET_BEGIN(sysinit_set);
  186                 sysinit_end = SET_LIMIT(sysinit_set);
  187         }
  188 
  189 restart:
  190         /*
  191          * Perform a bubble sort of the system initialization objects by
  192          * their subsystem (primary key) and order (secondary key).
  193          */
  194         for (sipp = sysinit; sipp < sysinit_end; sipp++) {
  195                 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
  196                         if ((*sipp)->subsystem < (*xipp)->subsystem ||
  197                              ((*sipp)->subsystem == (*xipp)->subsystem &&
  198                               (*sipp)->order <= (*xipp)->order))
  199                                 continue;       /* skip*/
  200                         save = *sipp;
  201                         *sipp = *xipp;
  202                         *xipp = save;
  203                 }
  204         }
  205 
  206 #if defined(VERBOSE_SYSINIT)
  207         last = SI_SUB_COPYRIGHT;
  208         verbose = 0;
  209 #if !defined(DDB)
  210         printf("VERBOSE_SYSINIT: DDB not enabled, symbol lookups disabled.\n");
  211 #endif
  212 #endif
  213 
  214         /*
  215          * Traverse the (now) ordered list of system initialization tasks.
  216          * Perform each task, and continue on to the next task.
  217          *
  218          * The last item on the list is expected to be the scheduler,
  219          * which will not return.
  220          */
  221         for (sipp = sysinit; sipp < sysinit_end; sipp++) {
  222 
  223                 if ((*sipp)->subsystem == SI_SUB_DUMMY)
  224                         continue;       /* skip dummy task(s)*/
  225 
  226                 if ((*sipp)->subsystem == SI_SUB_DONE)
  227                         continue;
  228 
  229 #if defined(VERBOSE_SYSINIT)
  230                 if ((*sipp)->subsystem > last) {
  231                         verbose = 1;
  232                         last = (*sipp)->subsystem;
  233                         printf("subsystem %x\n", last);
  234                 }
  235                 if (verbose) {
  236 #if defined(DDB)
  237                         const char *name;
  238                         c_db_sym_t sym;
  239                         db_expr_t  offset;
  240 
  241                         sym = db_search_symbol((vm_offset_t)(*sipp)->func,
  242                             DB_STGY_PROC, &offset);
  243                         db_symbol_values(sym, &name, NULL);
  244                         if (name != NULL)
  245                                 printf("   %s(%p)... ", name, (*sipp)->udata);
  246                         else
  247 #endif
  248                                 printf("   %p(%p)... ", (*sipp)->func,
  249                                     (*sipp)->udata);
  250                 }
  251 #endif
  252 
  253                 /* Call function */
  254                 (*((*sipp)->func))((*sipp)->udata);
  255 
  256 #if defined(VERBOSE_SYSINIT)
  257                 if (verbose)
  258                         printf("done.\n");
  259 #endif
  260 
  261                 /* Check off the one we're just done */
  262                 (*sipp)->subsystem = SI_SUB_DONE;
  263 
  264                 /* Check if we've installed more sysinit items via KLD */
  265                 if (newsysinit != NULL) {
  266                         if (sysinit != SET_BEGIN(sysinit_set))
  267                                 free(sysinit, M_TEMP);
  268                         sysinit = newsysinit;
  269                         sysinit_end = newsysinit_end;
  270                         newsysinit = NULL;
  271                         newsysinit_end = NULL;
  272                         goto restart;
  273                 }
  274         }
  275 
  276         panic("Shouldn't get here!");
  277         /* NOTREACHED*/
  278 }
  279 
  280 
  281 /*
  282  ***************************************************************************
  283  ****
  284  **** The following SYSINIT's belong elsewhere, but have not yet
  285  **** been moved.
  286  ****
  287  ***************************************************************************
  288  */
  289 static void
  290 print_caddr_t(void *data)
  291 {
  292         printf("%s", (char *)data);
  293 }
  294 
  295 static void
  296 print_version(void *data __unused)
  297 {
  298         int len;
  299 
  300         /* Strip a trailing newline from version. */
  301         len = strlen(version);
  302         while (len > 0 && version[len - 1] == '\n')
  303                 len--;
  304         printf("%.*s %s\n", len, version, machine);
  305         printf("%s\n", compiler_version);
  306 }
  307 
  308 SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t,
  309     copyright);
  310 SYSINIT(trademark, SI_SUB_COPYRIGHT, SI_ORDER_SECOND, print_caddr_t,
  311     trademark);
  312 SYSINIT(version, SI_SUB_COPYRIGHT, SI_ORDER_THIRD, print_version, NULL);
  313 
  314 #ifdef WITNESS
  315 static char wit_warn[] =
  316      "WARNING: WITNESS option enabled, expect reduced performance.\n";
  317 SYSINIT(witwarn, SI_SUB_COPYRIGHT, SI_ORDER_THIRD + 1,
  318    print_caddr_t, wit_warn);
  319 SYSINIT(witwarn2, SI_SUB_RUN_SCHEDULER, SI_ORDER_THIRD + 1,
  320    print_caddr_t, wit_warn);
  321 #endif
  322 
  323 #ifdef DIAGNOSTIC
  324 static char diag_warn[] =
  325      "WARNING: DIAGNOSTIC option enabled, expect reduced performance.\n";
  326 SYSINIT(diagwarn, SI_SUB_COPYRIGHT, SI_ORDER_THIRD + 2,
  327     print_caddr_t, diag_warn);
  328 SYSINIT(diagwarn2, SI_SUB_RUN_SCHEDULER, SI_ORDER_THIRD + 2,
  329     print_caddr_t, diag_warn);
  330 #endif
  331 
  332 static int
  333 null_fetch_syscall_args(struct thread *td __unused,
  334     struct syscall_args *sa __unused)
  335 {
  336 
  337         panic("null_fetch_syscall_args");
  338 }
  339 
  340 static void
  341 null_set_syscall_retval(struct thread *td __unused, int error __unused)
  342 {
  343 
  344         panic("null_set_syscall_retval");
  345 }
  346 
  347 struct sysentvec null_sysvec = {
  348         .sv_size        = 0,
  349         .sv_table       = NULL,
  350         .sv_mask        = 0,
  351         .sv_sigsize     = 0,
  352         .sv_sigtbl      = NULL,
  353         .sv_errsize     = 0,
  354         .sv_errtbl      = NULL,
  355         .sv_transtrap   = NULL,
  356         .sv_fixup       = NULL,
  357         .sv_sendsig     = NULL,
  358         .sv_sigcode     = NULL,
  359         .sv_szsigcode   = NULL,
  360         .sv_prepsyscall = NULL,
  361         .sv_name        = "null",
  362         .sv_coredump    = NULL,
  363         .sv_imgact_try  = NULL,
  364         .sv_minsigstksz = 0,
  365         .sv_pagesize    = PAGE_SIZE,
  366         .sv_minuser     = VM_MIN_ADDRESS,
  367         .sv_maxuser     = VM_MAXUSER_ADDRESS,
  368         .sv_usrstack    = USRSTACK,
  369         .sv_psstrings   = PS_STRINGS,
  370         .sv_stackprot   = VM_PROT_ALL,
  371         .sv_copyout_strings     = NULL,
  372         .sv_setregs     = NULL,
  373         .sv_fixlimit    = NULL,
  374         .sv_maxssiz     = NULL,
  375         .sv_flags       = 0,
  376         .sv_set_syscall_retval = null_set_syscall_retval,
  377         .sv_fetch_syscall_args = null_fetch_syscall_args,
  378         .sv_syscallnames = NULL,
  379         .sv_schedtail   = NULL,
  380 };
  381 
  382 /*
  383  ***************************************************************************
  384  ****
  385  **** The two following SYSINIT's are proc0 specific glue code.  I am not
  386  **** convinced that they can not be safely combined, but their order of
  387  **** operation has been maintained as the same as the original init_main.c
  388  **** for right now.
  389  ****
  390  **** These probably belong in init_proc.c or kern_proc.c, since they
  391  **** deal with proc0 (the fork template process).
  392  ****
  393  ***************************************************************************
  394  */
  395 /* ARGSUSED*/
  396 static void
  397 proc0_init(void *dummy __unused)
  398 {
  399         struct proc *p;
  400         unsigned i;
  401         struct thread *td;
  402 
  403         GIANT_REQUIRED;
  404         p = &proc0;
  405         td = &thread0;
  406         
  407         /*
  408          * Initialize magic number and osrel.
  409          */
  410         p->p_magic = P_MAGIC;
  411         p->p_osrel = osreldate;
  412 
  413         /*
  414          * Initialize thread and process structures.
  415          */
  416         procinit();     /* set up proc zone */
  417         threadinit();   /* set up UMA zones */
  418 
  419         /*
  420          * Initialise scheduler resources.
  421          * Add scheduler specific parts to proc, thread as needed.
  422          */
  423         schedinit();    /* scheduler gets its house in order */
  424         /*
  425          * Initialize sleep queue hash table
  426          */
  427         sleepinit();
  428 
  429         /*
  430          * additional VM structures
  431          */
  432         vm_init2();
  433 
  434         /*
  435          * Create process 0 (the swapper).
  436          */
  437         LIST_INSERT_HEAD(&allproc, p, p_list);
  438         LIST_INSERT_HEAD(PIDHASH(0), p, p_hash);
  439         mtx_init(&pgrp0.pg_mtx, "process group", NULL, MTX_DEF | MTX_DUPOK);
  440         p->p_pgrp = &pgrp0;
  441         LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
  442         LIST_INIT(&pgrp0.pg_members);
  443         LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
  444 
  445         pgrp0.pg_session = &session0;
  446         mtx_init(&session0.s_mtx, "session", NULL, MTX_DEF);
  447         refcount_init(&session0.s_count, 1);
  448         session0.s_leader = p;
  449 
  450         p->p_sysent = &null_sysvec;
  451         p->p_flag = P_SYSTEM | P_INMEM;
  452         p->p_state = PRS_NORMAL;
  453         knlist_init_mtx(&p->p_klist, &p->p_mtx);
  454         STAILQ_INIT(&p->p_ktr);
  455         p->p_nice = NZERO;
  456         td->td_tid = PID_MAX + 1;
  457         td->td_state = TDS_RUNNING;
  458         td->td_pri_class = PRI_TIMESHARE;
  459         td->td_user_pri = PUSER;
  460         td->td_base_user_pri = PUSER;
  461         td->td_priority = PVM;
  462         td->td_base_pri = PVM;
  463         td->td_oncpu = 0;
  464         td->td_flags = TDF_INMEM|TDP_KTHREAD;
  465         td->td_cpuset = cpuset_thread0();
  466         prison0.pr_cpuset = cpuset_ref(td->td_cpuset);
  467         p->p_peers = 0;
  468         p->p_leader = p;
  469 
  470 
  471         strncpy(p->p_comm, "kernel", sizeof (p->p_comm));
  472         strncpy(td->td_name, "swapper", sizeof (td->td_name));
  473 
  474         callout_init(&p->p_itcallout, CALLOUT_MPSAFE);
  475         callout_init_mtx(&p->p_limco, &p->p_mtx, 0);
  476         callout_init(&td->td_slpcallout, CALLOUT_MPSAFE);
  477 
  478         /* Create credentials. */
  479         p->p_ucred = crget();
  480         p->p_ucred->cr_ngroups = 1;     /* group 0 */
  481         p->p_ucred->cr_uidinfo = uifind(0);
  482         p->p_ucred->cr_ruidinfo = uifind(0);
  483         p->p_ucred->cr_prison = &prison0;
  484 #ifdef AUDIT
  485         audit_cred_kproc0(p->p_ucred);
  486 #endif
  487 #ifdef MAC
  488         mac_cred_create_swapper(p->p_ucred);
  489 #endif
  490         td->td_ucred = crhold(p->p_ucred);
  491 
  492         /* Create sigacts. */
  493         p->p_sigacts = sigacts_alloc();
  494 
  495         /* Initialize signal state for process 0. */
  496         siginit(&proc0);
  497 
  498         /* Create the file descriptor table. */
  499         p->p_fd = fdinit(NULL);
  500         p->p_fdtol = NULL;
  501 
  502         /* Create the limits structures. */
  503         p->p_limit = lim_alloc();
  504         for (i = 0; i < RLIM_NLIMITS; i++)
  505                 p->p_limit->pl_rlimit[i].rlim_cur =
  506                     p->p_limit->pl_rlimit[i].rlim_max = RLIM_INFINITY;
  507         p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_cur =
  508             p->p_limit->pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
  509         p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_cur =
  510             p->p_limit->pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
  511         i = ptoa(cnt.v_free_count);
  512         p->p_limit->pl_rlimit[RLIMIT_RSS].rlim_max = i;
  513         p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
  514         p->p_limit->pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
  515         p->p_cpulimit = RLIM_INFINITY;
  516 
  517         p->p_stats = pstats_alloc();
  518 
  519         /* Allocate a prototype map so we have something to fork. */
  520         pmap_pinit0(vmspace_pmap(&vmspace0));
  521         p->p_vmspace = &vmspace0;
  522         vmspace0.vm_refcnt = 1;
  523 
  524         /*
  525          * proc0 is not expected to enter usermode, so there is no special
  526          * handling for sv_minuser here, like is done for exec_new_vmspace().
  527          */
  528         vm_map_init(&vmspace0.vm_map, p->p_sysent->sv_minuser,
  529             p->p_sysent->sv_maxuser);
  530         vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0);
  531 
  532         /*-
  533          * call the init and ctor for the new thread and proc
  534          * we wait to do this until all other structures
  535          * are fairly sane.
  536          */
  537         EVENTHANDLER_INVOKE(process_init, p);
  538         EVENTHANDLER_INVOKE(thread_init, td);
  539         EVENTHANDLER_INVOKE(process_ctor, p);
  540         EVENTHANDLER_INVOKE(thread_ctor, td);
  541 
  542         /*
  543          * Charge root for one process.
  544          */
  545         (void)chgproccnt(p->p_ucred->cr_ruidinfo, 1, 0);
  546 }
  547 SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL);
  548 
  549 /* ARGSUSED*/
  550 static void
  551 proc0_post(void *dummy __unused)
  552 {
  553         struct timespec ts;
  554         struct proc *p;
  555         struct rusage ru;
  556         struct thread *td;
  557 
  558         /*
  559          * Now we can look at the time, having had a chance to verify the
  560          * time from the filesystem.  Pretend that proc0 started now.
  561          */
  562         sx_slock(&allproc_lock);
  563         FOREACH_PROC_IN_SYSTEM(p) {
  564                 microuptime(&p->p_stats->p_start);
  565                 PROC_SLOCK(p);
  566                 rufetch(p, &ru);        /* Clears thread stats */
  567                 PROC_SUNLOCK(p);
  568                 p->p_rux.rux_runtime = 0;
  569                 p->p_rux.rux_uticks = 0;
  570                 p->p_rux.rux_sticks = 0;
  571                 p->p_rux.rux_iticks = 0;
  572                 FOREACH_THREAD_IN_PROC(p, td) {
  573                         td->td_runtime = 0;
  574                 }
  575         }
  576         sx_sunlock(&allproc_lock);
  577         PCPU_SET(switchtime, cpu_ticks());
  578         PCPU_SET(switchticks, ticks);
  579 
  580         /*
  581          * Give the ``random'' number generator a thump.
  582          */
  583         nanotime(&ts);
  584         srandom(ts.tv_sec ^ ts.tv_nsec);
  585 }
  586 SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL);
  587 
  588 static void
  589 random_init(void *dummy __unused)
  590 {
  591 
  592         /*
  593          * After CPU has been started we have some randomness on most
  594          * platforms via get_cyclecount().  For platforms that don't
  595          * we will reseed random(9) in proc0_post() as well.
  596          */
  597         srandom(get_cyclecount());
  598 }
  599 SYSINIT(random, SI_SUB_RANDOM, SI_ORDER_FIRST, random_init, NULL);
  600 
  601 /*
  602  ***************************************************************************
  603  ****
  604  **** The following SYSINIT's and glue code should be moved to the
  605  **** respective files on a per subsystem basis.
  606  ****
  607  ***************************************************************************
  608  */
  609 
  610 
  611 /*
  612  ***************************************************************************
  613  ****
  614  **** The following code probably belongs in another file, like
  615  **** kern/init_init.c.
  616  ****
  617  ***************************************************************************
  618  */
  619 
  620 /*
  621  * List of paths to try when searching for "init".
  622  */
  623 static char init_path[MAXPATHLEN] =
  624 #ifdef  INIT_PATH
  625     __XSTRING(INIT_PATH);
  626 #else
  627     "/sbin/init:/sbin/oinit:/sbin/init.bak:/rescue/init:/stand/sysinstall";
  628 #endif
  629 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0,
  630         "Path used to search the init process");
  631 
  632 /*
  633  * Shutdown timeout of init(8).
  634  * Unused within kernel, but used to control init(8), hence do not remove.
  635  */
  636 #ifndef INIT_SHUTDOWN_TIMEOUT
  637 #define INIT_SHUTDOWN_TIMEOUT 120
  638 #endif
  639 static int init_shutdown_timeout = INIT_SHUTDOWN_TIMEOUT;
  640 SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout,
  641         CTLFLAG_RW, &init_shutdown_timeout, 0, "");
  642 
  643 /*
  644  * Start the initial user process; try exec'ing each pathname in init_path.
  645  * The program is invoked with one argument containing the boot flags.
  646  */
  647 static void
  648 start_init(void *dummy)
  649 {
  650         vm_offset_t addr;
  651         struct execve_args args;
  652         int options, error;
  653         char *var, *path, *next, *s;
  654         char *ucp, **uap, *arg0, *arg1;
  655         struct thread *td;
  656         struct proc *p;
  657 
  658         mtx_lock(&Giant);
  659 
  660         GIANT_REQUIRED;
  661 
  662         td = curthread;
  663         p = td->td_proc;
  664 
  665         vfs_mountroot();
  666 
  667         /*
  668          * Need just enough stack to hold the faked-up "execve()" arguments.
  669          */
  670         addr = p->p_sysent->sv_usrstack - PAGE_SIZE;
  671         if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
  672                         FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0)
  673                 panic("init: couldn't allocate argument space");
  674         p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
  675         p->p_vmspace->vm_ssize = 1;
  676 
  677         if ((var = getenv("init_path")) != NULL) {
  678                 strlcpy(init_path, var, sizeof(init_path));
  679                 freeenv(var);
  680         }
  681         
  682         for (path = init_path; *path != '\0'; path = next) {
  683                 while (*path == ':')
  684                         path++;
  685                 if (*path == '\0')
  686                         break;
  687                 for (next = path; *next != '\0' && *next != ':'; next++)
  688                         /* nothing */ ;
  689                 if (bootverbose)
  690                         printf("start_init: trying %.*s\n", (int)(next - path),
  691                             path);
  692                         
  693                 /*
  694                  * Move out the boot flag argument.
  695                  */
  696                 options = 0;
  697                 ucp = (char *)p->p_sysent->sv_usrstack;
  698                 (void)subyte(--ucp, 0);         /* trailing zero */
  699                 if (boothowto & RB_SINGLE) {
  700                         (void)subyte(--ucp, 's');
  701                         options = 1;
  702                 }
  703 #ifdef notyet
  704                 if (boothowto & RB_FASTBOOT) {
  705                         (void)subyte(--ucp, 'f');
  706                         options = 1;
  707                 }
  708 #endif
  709 
  710 #ifdef BOOTCDROM
  711                 (void)subyte(--ucp, 'C');
  712                 options = 1;
  713 #endif
  714 
  715                 if (options == 0)
  716                         (void)subyte(--ucp, '-');
  717                 (void)subyte(--ucp, '-');               /* leading hyphen */
  718                 arg1 = ucp;
  719 
  720                 /*
  721                  * Move out the file name (also arg 0).
  722                  */
  723                 (void)subyte(--ucp, 0);
  724                 for (s = next - 1; s >= path; s--)
  725                         (void)subyte(--ucp, *s);
  726                 arg0 = ucp;
  727 
  728                 /*
  729                  * Move out the arg pointers.
  730                  */
  731                 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
  732                 (void)suword((caddr_t)--uap, (long)0);  /* terminator */
  733                 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
  734                 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
  735 
  736                 /*
  737                  * Point at the arguments.
  738                  */
  739                 args.fname = arg0;
  740                 args.argv = uap;
  741                 args.envv = NULL;
  742 
  743                 /*
  744                  * Now try to exec the program.  If can't for any reason
  745                  * other than it doesn't exist, complain.
  746                  *
  747                  * Otherwise, return via fork_trampoline() all the way
  748                  * to user mode as init!
  749                  */
  750                 if ((error = execve(td, &args)) == 0) {
  751                         mtx_unlock(&Giant);
  752                         return;
  753                 }
  754                 if (error != ENOENT)
  755                         printf("exec %.*s: error %d\n", (int)(next - path), 
  756                             path, error);
  757         }
  758         printf("init: not found in path %s\n", init_path);
  759         panic("no init");
  760 }
  761 
  762 /*
  763  * Like kproc_create(), but runs in it's own address space.
  764  * We do this early to reserve pid 1.
  765  *
  766  * Note special case - do not make it runnable yet.  Other work
  767  * in progress will change this more.
  768  */
  769 static void
  770 create_init(const void *udata __unused)
  771 {
  772         struct ucred *newcred, *oldcred;
  773         int error;
  774 
  775         error = fork1(&thread0, RFFDG | RFPROC | RFSTOPPED, 0, &initproc);
  776         if (error)
  777                 panic("cannot fork init: %d\n", error);
  778         KASSERT(initproc->p_pid == 1, ("create_init: initproc->p_pid != 1"));
  779         /* divorce init's credentials from the kernel's */
  780         newcred = crget();
  781         PROC_LOCK(initproc);
  782         initproc->p_flag |= P_SYSTEM | P_INMEM;
  783         oldcred = initproc->p_ucred;
  784         crcopy(newcred, oldcred);
  785 #ifdef MAC
  786         mac_cred_create_init(newcred);
  787 #endif
  788 #ifdef AUDIT
  789         audit_cred_proc1(newcred);
  790 #endif
  791         initproc->p_ucred = newcred;
  792         PROC_UNLOCK(initproc);
  793         crfree(oldcred);
  794         cred_update_thread(FIRST_THREAD_IN_PROC(initproc));
  795         cpu_set_fork_handler(FIRST_THREAD_IN_PROC(initproc), start_init, NULL);
  796 }
  797 SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL);
  798 
  799 /*
  800  * Make it runnable now.
  801  */
  802 static void
  803 kick_init(const void *udata __unused)
  804 {
  805         struct thread *td;
  806 
  807         td = FIRST_THREAD_IN_PROC(initproc);
  808         thread_lock(td);
  809         TD_SET_CAN_RUN(td);
  810         sched_add(td, SRQ_BORING);
  811         thread_unlock(td);
  812 }
  813 SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL);

Cache object: a51e4a4fba0a1d147b78d658ddd17cc2


[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]


This page is part of the FreeBSD/Linux Linux Kernel Cross-Reference, and was automatically generated using a modified version of the LXR engine.