2 * Copyright (c) 1995 Terrence R. Lambert
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.
13 * Redistribution and use in source and binary forms, with or without
14 * modification, are permitted provided that the following conditions
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.
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
41 * @(#)init_main.c 8.9 (Berkeley) 1/21/94
44 #include <sys/cdefs.h>
45 __FBSDID("$FreeBSD$");
48 #include "opt_init_path.h"
51 #include <sys/param.h>
52 #include <sys/kernel.h>
55 #include <sys/filedesc.h>
58 #include <sys/mount.h>
59 #include <sys/mutex.h>
60 #include <sys/syscallsubr.h>
61 #include <sys/sysctl.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>
71 #include <sys/sysproto.h>
72 #include <sys/vmmeter.h>
73 #include <sys/unistd.h>
74 #include <sys/malloc.h>
76 #include <sys/cpuset.h>
78 #include <machine/cpu.h>
80 #include <security/audit/audit.h>
81 #include <security/mac/mac_framework.h>
84 #include <vm/vm_param.h>
86 #include <vm/vm_map.h>
87 #include <sys/copyright.h>
90 #include <ddb/db_sym.h>
92 void mi_startup(void); /* Should be elsewhere */
94 /* Components of the first process -- never freed. */
95 static struct session session0
;
96 static struct pgrp pgrp0
;
98 struct thread thread0
__aligned(16);
99 struct vmspace vmspace0
;
100 struct proc
*initproc
;
102 int boothowto
= 0; /* initialized so that it can be patched */
103 SYSCTL_INT(_debug
, OID_AUTO
, boothowto
, CTLFLAG_RD
, &boothowto
, 0, "");
105 SYSCTL_INT(_debug
, OID_AUTO
, bootverbose
, CTLFLAG_RW
, &bootverbose
, 0, "");
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
112 SYSINIT(placeholder
, SI_SUB_DUMMY
, SI_ORDER_ANY
, NULL
, NULL
);
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.
118 SET_DECLARE(sysinit_set
, struct sysinit
);
119 struct sysinit
**sysinit
, **sysinit_end
;
120 struct sysinit
**newsysinit
, **newsysinit_end
;
123 * Merge a new sysinit set into the current set, reallocating it if
124 * necessary. This can only be called after malloc is running.
127 sysinit_add(struct sysinit
**set
, struct sysinit
**set_end
)
129 struct sysinit
**newset
;
130 struct sysinit
**sipp
;
131 struct sysinit
**xipp
;
134 count
= set_end
- set
;
136 count
+= newsysinit_end
- newsysinit
;
138 count
+= sysinit_end
- sysinit
;
139 newset
= malloc(count
* sizeof(*sipp
), M_TEMP
, M_NOWAIT
);
141 panic("cannot malloc for sysinit");
144 for (sipp
= newsysinit
; sipp
< newsysinit_end
; sipp
++)
147 for (sipp
= sysinit
; sipp
< sysinit_end
; sipp
++)
149 for (sipp
= set
; sipp
< set_end
; sipp
++)
152 free(newsysinit
, M_TEMP
);
154 newsysinit_end
= newset
+ count
;
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.
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".
172 register struct sysinit
**sipp
; /* system initialization*/
173 register struct sysinit
**xipp
; /* interior loop of sort*/
174 register struct sysinit
*save
; /* bubble*/
176 #if defined(VERBOSE_SYSINIT)
181 if (sysinit
== NULL
) {
182 sysinit
= SET_BEGIN(sysinit_set
);
183 sysinit_end
= SET_LIMIT(sysinit_set
);
188 * Perform a bubble sort of the system initialization objects by
189 * their subsystem (primary key) and order (secondary key).
191 for (sipp
= sysinit
; sipp
< sysinit_end
; sipp
++) {
192 for (xipp
= sipp
+ 1; xipp
< sysinit_end
; xipp
++) {
193 if ((*sipp
)->subsystem
< (*xipp
)->subsystem
||
194 ((*sipp
)->subsystem
== (*xipp
)->subsystem
&&
195 (*sipp
)->order
<= (*xipp
)->order
))
203 #if defined(VERBOSE_SYSINIT)
204 last
= SI_SUB_COPYRIGHT
;
207 printf("VERBOSE_SYSINIT: DDB not enabled, symbol lookups disabled.\n");
212 * Traverse the (now) ordered list of system initialization tasks.
213 * Perform each task, and continue on to the next task.
215 * The last item on the list is expected to be the scheduler,
216 * which will not return.
218 for (sipp
= sysinit
; sipp
< sysinit_end
; sipp
++) {
220 if ((*sipp
)->subsystem
== SI_SUB_DUMMY
)
221 continue; /* skip dummy task(s)*/
223 if ((*sipp
)->subsystem
== SI_SUB_DONE
)
226 #if defined(VERBOSE_SYSINIT)
227 if ((*sipp
)->subsystem
> last
) {
229 last
= (*sipp
)->subsystem
;
230 printf("subsystem %x\n", last
);
238 sym
= db_search_symbol((vm_offset_t
)(*sipp
)->func
,
239 DB_STGY_PROC
, &offset
);
240 db_symbol_values(sym
, &name
, NULL
);
242 printf(" %s(%p)... ", name
, (*sipp
)->udata
);
245 printf(" %p(%p)... ", (*sipp
)->func
,
251 (*((*sipp
)->func
))((*sipp
)->udata
);
253 #if defined(VERBOSE_SYSINIT)
258 /* Check off the one we're just done */
259 (*sipp
)->subsystem
= SI_SUB_DONE
;
261 /* Check if we've installed more sysinit items via KLD */
262 if (newsysinit
!= NULL
) {
263 if (sysinit
!= SET_BEGIN(sysinit_set
))
264 free(sysinit
, M_TEMP
);
265 sysinit
= newsysinit
;
266 sysinit_end
= newsysinit_end
;
268 newsysinit_end
= NULL
;
273 panic("Shouldn't get here!");
279 ***************************************************************************
281 **** The following SYSINIT's belong elsewhere, but have not yet
284 ***************************************************************************
287 print_caddr_t(void *data __unused
)
289 printf("%s", (char *)data
);
291 SYSINIT(announce
, SI_SUB_COPYRIGHT
, SI_ORDER_FIRST
, print_caddr_t
,
293 SYSINIT(trademark
, SI_SUB_COPYRIGHT
, SI_ORDER_SECOND
, print_caddr_t
,
295 SYSINIT(version
, SI_SUB_COPYRIGHT
, SI_ORDER_THIRD
, print_caddr_t
, version
);
298 static char wit_warn
[] =
299 "WARNING: WITNESS option enabled, expect reduced performance.\n";
300 SYSINIT(witwarn
, SI_SUB_COPYRIGHT
, SI_ORDER_THIRD
+ 1,
301 print_caddr_t
, wit_warn
);
302 SYSINIT(witwarn2
, SI_SUB_RUN_SCHEDULER
, SI_ORDER_THIRD
+ 1,
303 print_caddr_t
, wit_warn
);
307 static char diag_warn
[] =
308 "WARNING: DIAGNOSTIC option enabled, expect reduced performance.\n";
309 SYSINIT(diagwarn
, SI_SUB_COPYRIGHT
, SI_ORDER_THIRD
+ 2,
310 print_caddr_t
, diag_warn
);
311 SYSINIT(diagwarn2
, SI_SUB_RUN_SCHEDULER
, SI_ORDER_THIRD
+ 2,
312 print_caddr_t
, diag_warn
);
316 set_boot_verbose(void *data __unused
)
319 if (boothowto
& RB_VERBOSE
)
322 SYSINIT(boot_verbose
, SI_SUB_TUNABLES
, SI_ORDER_ANY
, set_boot_verbose
, NULL
);
324 struct sysentvec null_sysvec
= {
354 ***************************************************************************
356 **** The two following SYSINIT's are proc0 specific glue code. I am not
357 **** convinced that they can not be safely combined, but their order of
358 **** operation has been maintained as the same as the original init_main.c
361 **** These probably belong in init_proc.c or kern_proc.c, since they
362 **** deal with proc0 (the fork template process).
364 ***************************************************************************
368 proc0_init(void *dummy __unused
)
379 * Initialize magic number and osrel.
381 p
->p_magic
= P_MAGIC
;
382 p
->p_osrel
= osreldate
;
385 * Initialize thread and process structures.
387 procinit(); /* set up proc zone */
388 threadinit(); /* set up UMA zones */
391 * Initialise scheduler resources.
392 * Add scheduler specific parts to proc, thread as needed.
394 schedinit(); /* scheduler gets its house in order */
396 * Initialize sleep queue hash table
401 * additional VM structures
406 * Create process 0 (the swapper).
408 LIST_INSERT_HEAD(&allproc
, p
, p_list
);
409 LIST_INSERT_HEAD(PIDHASH(0), p
, p_hash
);
410 mtx_init(&pgrp0
.pg_mtx
, "process group", NULL
, MTX_DEF
| MTX_DUPOK
);
412 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0
, pg_hash
);
413 LIST_INIT(&pgrp0
.pg_members
);
414 LIST_INSERT_HEAD(&pgrp0
.pg_members
, p
, p_pglist
);
416 pgrp0
.pg_session
= &session0
;
417 mtx_init(&session0
.s_mtx
, "session", NULL
, MTX_DEF
);
418 refcount_init(&session0
.s_count
, 1);
419 session0
.s_leader
= p
;
421 p
->p_sysent
= &null_sysvec
;
422 p
->p_flag
= P_SYSTEM
| P_INMEM
;
423 p
->p_state
= PRS_NORMAL
;
424 knlist_init(&p
->p_klist
, &p
->p_mtx
, NULL
, NULL
, NULL
);
425 STAILQ_INIT(&p
->p_ktr
);
427 td
->td_tid
= PID_MAX
+ 1;
428 td
->td_state
= TDS_RUNNING
;
429 td
->td_pri_class
= PRI_TIMESHARE
;
430 td
->td_user_pri
= PUSER
;
431 td
->td_base_user_pri
= PUSER
;
432 td
->td_priority
= PVM
;
433 td
->td_base_pri
= PUSER
;
435 td
->td_flags
= TDF_INMEM
|TDP_KTHREAD
;
436 td
->td_cpuset
= cpuset_thread0();
441 strncpy(p
->p_comm
, "kernel", sizeof (p
->p_comm
));
442 strncpy(td
->td_name
, "swapper", sizeof (td
->td_name
));
444 callout_init(&p
->p_itcallout
, CALLOUT_MPSAFE
);
445 callout_init_mtx(&p
->p_limco
, &p
->p_mtx
, 0);
446 callout_init(&td
->td_slpcallout
, CALLOUT_MPSAFE
);
448 /* Create credentials. */
449 p
->p_ucred
= crget();
450 p
->p_ucred
->cr_ngroups
= 1; /* group 0 */
451 p
->p_ucred
->cr_uidinfo
= uifind(0);
452 p
->p_ucred
->cr_ruidinfo
= uifind(0);
453 p
->p_ucred
->cr_prison
= NULL
; /* Don't jail it. */
455 audit_cred_kproc0(p
->p_ucred
);
458 mac_proc_create_swapper(p
->p_ucred
);
460 td
->td_ucred
= crhold(p
->p_ucred
);
462 /* Create sigacts. */
463 p
->p_sigacts
= sigacts_alloc();
465 /* Initialize signal state for process 0. */
468 /* Create the file descriptor table. */
469 p
->p_fd
= fdinit(NULL
);
472 /* Create the limits structures. */
473 p
->p_limit
= lim_alloc();
474 for (i
= 0; i
< RLIM_NLIMITS
; i
++)
475 p
->p_limit
->pl_rlimit
[i
].rlim_cur
=
476 p
->p_limit
->pl_rlimit
[i
].rlim_max
= RLIM_INFINITY
;
477 p
->p_limit
->pl_rlimit
[RLIMIT_NOFILE
].rlim_cur
=
478 p
->p_limit
->pl_rlimit
[RLIMIT_NOFILE
].rlim_max
= maxfiles
;
479 p
->p_limit
->pl_rlimit
[RLIMIT_NPROC
].rlim_cur
=
480 p
->p_limit
->pl_rlimit
[RLIMIT_NPROC
].rlim_max
= maxproc
;
481 i
= ptoa(cnt
.v_free_count
);
482 p
->p_limit
->pl_rlimit
[RLIMIT_RSS
].rlim_max
= i
;
483 p
->p_limit
->pl_rlimit
[RLIMIT_MEMLOCK
].rlim_max
= i
;
484 p
->p_limit
->pl_rlimit
[RLIMIT_MEMLOCK
].rlim_cur
= i
/ 3;
485 p
->p_cpulimit
= RLIM_INFINITY
;
487 p
->p_stats
= pstats_alloc();
489 /* Allocate a prototype map so we have something to fork. */
490 pmap_pinit0(vmspace_pmap(&vmspace0
));
491 p
->p_vmspace
= &vmspace0
;
492 vmspace0
.vm_refcnt
= 1;
493 vm_map_init(&vmspace0
.vm_map
, p
->p_sysent
->sv_minuser
,
494 p
->p_sysent
->sv_maxuser
);
495 vmspace0
.vm_map
.pmap
= vmspace_pmap(&vmspace0
);
498 * call the init and ctor for the new thread and proc
499 * we wait to do this until all other structures
502 EVENTHANDLER_INVOKE(process_init
, p
);
503 EVENTHANDLER_INVOKE(thread_init
, td
);
504 EVENTHANDLER_INVOKE(process_ctor
, p
);
505 EVENTHANDLER_INVOKE(thread_ctor
, td
);
508 * Charge root for one process.
510 (void)chgproccnt(p
->p_ucred
->cr_ruidinfo
, 1, 0);
512 SYSINIT(p0init
, SI_SUB_INTRINSIC
, SI_ORDER_FIRST
, proc0_init
, NULL
);
516 proc0_post(void *dummy __unused
)
524 * Now we can look at the time, having had a chance to verify the
525 * time from the filesystem. Pretend that proc0 started now.
527 sx_slock(&allproc_lock
);
528 FOREACH_PROC_IN_SYSTEM(p
) {
529 microuptime(&p
->p_stats
->p_start
);
531 rufetch(p
, &ru
); /* Clears thread stats */
533 p
->p_rux
.rux_runtime
= 0;
534 p
->p_rux
.rux_uticks
= 0;
535 p
->p_rux
.rux_sticks
= 0;
536 p
->p_rux
.rux_iticks
= 0;
537 FOREACH_THREAD_IN_PROC(p
, td
) {
541 sx_sunlock(&allproc_lock
);
542 PCPU_SET(switchtime
, cpu_ticks());
543 PCPU_SET(switchticks
, ticks
);
546 * Give the ``random'' number generator a thump.
549 srandom(ts
.tv_sec
^ ts
.tv_nsec
);
551 SYSINIT(p0post
, SI_SUB_INTRINSIC_POST
, SI_ORDER_FIRST
, proc0_post
, NULL
);
554 ***************************************************************************
556 **** The following SYSINIT's and glue code should be moved to the
557 **** respective files on a per subsystem basis.
559 ***************************************************************************
564 ***************************************************************************
566 **** The following code probably belongs in another file, like
567 **** kern/init_init.c.
569 ***************************************************************************
573 * List of paths to try when searching for "init".
575 static char init_path
[MAXPATHLEN
] =
577 __XSTRING(INIT_PATH
);
579 "/sbin/init:/sbin/oinit:/sbin/init.bak:/rescue/init:/stand/sysinstall";
581 SYSCTL_STRING(_kern
, OID_AUTO
, init_path
, CTLFLAG_RD
, init_path
, 0,
582 "Path used to search the init process");
585 * Shutdown timeout of init(8).
586 * Unused within kernel, but used to control init(8), hence do not remove.
588 #ifndef INIT_SHUTDOWN_TIMEOUT
589 #define INIT_SHUTDOWN_TIMEOUT 120
591 static int init_shutdown_timeout
= INIT_SHUTDOWN_TIMEOUT
;
592 SYSCTL_INT(_kern
, OID_AUTO
, init_shutdown_timeout
,
593 CTLFLAG_RW
, &init_shutdown_timeout
, 0, "");
596 * Start the initial user process; try exec'ing each pathname in init_path.
597 * The program is invoked with one argument containing the boot flags.
600 start_init(void *dummy
)
603 struct execve_args args
;
605 char *var
, *path
, *next
, *s
;
606 char *ucp
, **uap
, *arg0
, *arg1
;
620 * Need just enough stack to hold the faked-up "execve()" arguments.
622 addr
= p
->p_sysent
->sv_usrstack
- PAGE_SIZE
;
623 if (vm_map_find(&p
->p_vmspace
->vm_map
, NULL
, 0, &addr
, PAGE_SIZE
,
624 FALSE
, VM_PROT_ALL
, VM_PROT_ALL
, 0) != 0)
625 panic("init: couldn't allocate argument space");
626 p
->p_vmspace
->vm_maxsaddr
= (caddr_t
)addr
;
627 p
->p_vmspace
->vm_ssize
= 1;
629 if ((var
= getenv("init_path")) != NULL
) {
630 strlcpy(init_path
, var
, sizeof(init_path
));
634 for (path
= init_path
; *path
!= '\0'; path
= next
) {
639 for (next
= path
; *next
!= '\0' && *next
!= ':'; next
++)
642 printf("start_init: trying %.*s\n", (int)(next
- path
),
646 * Move out the boot flag argument.
649 ucp
= (char *)p
->p_sysent
->sv_usrstack
;
650 (void)subyte(--ucp
, 0); /* trailing zero */
651 if (boothowto
& RB_SINGLE
) {
652 (void)subyte(--ucp
, 's');
656 if (boothowto
& RB_FASTBOOT
) {
657 (void)subyte(--ucp
, 'f');
663 (void)subyte(--ucp
, 'C');
668 (void)subyte(--ucp
, '-');
669 (void)subyte(--ucp
, '-'); /* leading hyphen */
673 * Move out the file name (also arg 0).
675 (void)subyte(--ucp
, 0);
676 for (s
= next
- 1; s
>= path
; s
--)
677 (void)subyte(--ucp
, *s
);
681 * Move out the arg pointers.
683 uap
= (char **)((intptr_t)ucp
& ~(sizeof(intptr_t)-1));
684 (void)suword((caddr_t
)--uap
, (long)0); /* terminator */
685 (void)suword((caddr_t
)--uap
, (long)(intptr_t)arg1
);
686 (void)suword((caddr_t
)--uap
, (long)(intptr_t)arg0
);
689 * Point at the arguments.
696 * Now try to exec the program. If can't for any reason
697 * other than it doesn't exist, complain.
699 * Otherwise, return via fork_trampoline() all the way
700 * to user mode as init!
702 if ((error
= execve(td
, &args
)) == 0) {
707 printf("exec %.*s: error %d\n", (int)(next
- path
),
710 printf("init: not found in path %s\n", init_path
);
715 * Like kproc_create(), but runs in it's own address space.
716 * We do this early to reserve pid 1.
718 * Note special case - do not make it runnable yet. Other work
719 * in progress will change this more.
722 create_init(const void *udata __unused
)
724 struct ucred
*newcred
, *oldcred
;
727 error
= fork1(&thread0
, RFFDG
| RFPROC
| RFSTOPPED
, 0, &initproc
);
729 panic("cannot fork init: %d\n", error
);
730 KASSERT(initproc
->p_pid
== 1, ("create_init: initproc->p_pid != 1"));
731 /* divorce init's credentials from the kernel's */
734 initproc
->p_flag
|= P_SYSTEM
| P_INMEM
;
735 oldcred
= initproc
->p_ucred
;
736 crcopy(newcred
, oldcred
);
738 mac_proc_create_init(newcred
);
741 audit_cred_proc1(newcred
);
743 initproc
->p_ucred
= newcred
;
744 PROC_UNLOCK(initproc
);
746 cred_update_thread(FIRST_THREAD_IN_PROC(initproc
));
747 cpu_set_fork_handler(FIRST_THREAD_IN_PROC(initproc
), start_init
, NULL
);
749 SYSINIT(init
, SI_SUB_CREATE_INIT
, SI_ORDER_FIRST
, create_init
, NULL
);
752 * Make it runnable now.
755 kick_init(const void *udata __unused
)
759 td
= FIRST_THREAD_IN_PROC(initproc
);
762 sched_add(td
, SRQ_BORING
);
765 SYSINIT(kickinit
, SI_SUB_KTHREAD_INIT
, SI_ORDER_FIRST
, kick_init
, NULL
);