Deleted Added
full compact
init_main.c (91066) init_main.c (91140)
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
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 * $FreeBSD: head/sys/kern/init_main.c 91066 2002-02-22 13:32:01Z phk $
42 * $FreeBSD: head/sys/kern/init_main.c 91140 2002-02-23 11:12:57Z tanimura $
43 */
44
45#include "opt_init_path.h"
46
47#include <sys/param.h>
48#include <sys/kernel.h>
49#include <sys/file.h>
50#include <sys/filedesc.h>
51#include <sys/ktr.h>
52#include <sys/lock.h>
53#include <sys/mount.h>
54#include <sys/mutex.h>
55#include <sys/sysctl.h>
56#include <sys/proc.h>
57#include <sys/resourcevar.h>
58#include <sys/systm.h>
59#include <sys/signalvar.h>
60#include <sys/vnode.h>
61#include <sys/sysent.h>
62#include <sys/reboot.h>
63#include <sys/sx.h>
64#include <sys/sysproto.h>
65#include <sys/vmmeter.h>
66#include <sys/unistd.h>
67#include <sys/malloc.h>
68#include <sys/conf.h>
69
70#include <machine/cpu.h>
71
72#include <vm/vm.h>
73#include <vm/vm_param.h>
74#include <vm/pmap.h>
75#include <vm/vm_map.h>
76#include <sys/user.h>
77#include <sys/copyright.h>
78
79void mi_startup(void); /* Should be elsewhere */
80
81/* Components of the first process -- never freed. */
82static struct session session0;
83static struct pgrp pgrp0;
84struct proc proc0;
85struct thread thread0;
86static struct procsig procsig0;
87static struct filedesc0 filedesc0;
88static struct plimit limit0;
89static struct vmspace vmspace0;
90struct proc *initproc;
91
92int cmask = CMASK;
93extern int fallback_elf_brand;
94
95struct vnode *rootvp;
96int boothowto = 0; /* initialized so that it can be patched */
97SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, "");
98int bootverbose;
99SYSCTL_INT(_debug, OID_AUTO, bootverbose, CTLFLAG_RW, &bootverbose, 0, "");
100
101/*
102 * This ensures that there is at least one entry so that the sysinit_set
103 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never
104 * executed.
105 */
106SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL)
107
108/*
109 * The sysinit table itself. Items are checked off as the are run.
110 * If we want to register new sysinit types, add them to newsysinit.
111 */
112SET_DECLARE(sysinit_set, struct sysinit);
113struct sysinit **sysinit, **sysinit_end;
114struct sysinit **newsysinit, **newsysinit_end;
115
116/*
117 * Merge a new sysinit set into the current set, reallocating it if
118 * necessary. This can only be called after malloc is running.
119 */
120void
121sysinit_add(struct sysinit **set, struct sysinit **set_end)
122{
123 struct sysinit **newset;
124 struct sysinit **sipp;
125 struct sysinit **xipp;
126 int count;
127
128 count = set_end - set;
129 if (newsysinit)
130 count += newsysinit_end - newsysinit;
131 else
132 count += sysinit_end - sysinit;
133 newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
134 if (newset == NULL)
135 panic("cannot malloc for sysinit");
136 xipp = newset;
137 if (newsysinit)
138 for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
139 *xipp++ = *sipp;
140 else
141 for (sipp = sysinit; sipp < sysinit_end; sipp++)
142 *xipp++ = *sipp;
143 for (sipp = set; sipp < set_end; sipp++)
144 *xipp++ = *sipp;
145 if (newsysinit)
146 free(newsysinit, M_TEMP);
147 newsysinit = newset;
148 newsysinit_end = newset + count;
149}
150
151/*
152 * System startup; initialize the world, create process 0, mount root
153 * filesystem, and fork to create init and pagedaemon. Most of the
154 * hard work is done in the lower-level initialization routines including
155 * startup(), which does memory initialization and autoconfiguration.
156 *
157 * This allows simple addition of new kernel subsystems that require
158 * boot time initialization. It also allows substitution of subsystem
159 * (for instance, a scheduler, kernel profiler, or VM system) by object
160 * module. Finally, it allows for optional "kernel threads".
161 */
162void
163mi_startup(void)
164{
165
166 register struct sysinit **sipp; /* system initialization*/
167 register struct sysinit **xipp; /* interior loop of sort*/
168 register struct sysinit *save; /* bubble*/
169
170 if (sysinit == NULL) {
171 sysinit = SET_BEGIN(sysinit_set);
172 sysinit_end = SET_LIMIT(sysinit_set);
173 }
174
175restart:
176 /*
177 * Perform a bubble sort of the system initialization objects by
178 * their subsystem (primary key) and order (secondary key).
179 */
180 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
181 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
182 if ((*sipp)->subsystem < (*xipp)->subsystem ||
183 ((*sipp)->subsystem == (*xipp)->subsystem &&
184 (*sipp)->order <= (*xipp)->order))
185 continue; /* skip*/
186 save = *sipp;
187 *sipp = *xipp;
188 *xipp = save;
189 }
190 }
191
192 /*
193 * Traverse the (now) ordered list of system initialization tasks.
194 * Perform each task, and continue on to the next task.
195 *
196 * The last item on the list is expected to be the scheduler,
197 * which will not return.
198 */
199 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
200
201 if ((*sipp)->subsystem == SI_SUB_DUMMY)
202 continue; /* skip dummy task(s)*/
203
204 if ((*sipp)->subsystem == SI_SUB_DONE)
205 continue;
206
207 /* Call function */
208 (*((*sipp)->func))((*sipp)->udata);
209
210 /* Check off the one we're just done */
211 (*sipp)->subsystem = SI_SUB_DONE;
212
213 /* Check if we've installed more sysinit items via KLD */
214 if (newsysinit != NULL) {
215 if (sysinit != SET_BEGIN(sysinit_set))
216 free(sysinit, M_TEMP);
217 sysinit = newsysinit;
218 sysinit_end = newsysinit_end;
219 newsysinit = NULL;
220 newsysinit_end = NULL;
221 goto restart;
222 }
223 }
224
225 panic("Shouldn't get here!");
226 /* NOTREACHED*/
227}
228
229
230/*
231 ***************************************************************************
232 ****
233 **** The following SYSINIT's belong elsewhere, but have not yet
234 **** been moved.
235 ****
236 ***************************************************************************
237 */
238static void
239print_caddr_t(void *data __unused)
240{
241 printf("%s", (char *)data);
242}
243SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
244SYSINIT(version, SI_SUB_COPYRIGHT, SI_ORDER_SECOND, print_caddr_t, version)
245
246static void
247set_boot_verbose(void *data __unused)
248{
249
250 if (boothowto & RB_VERBOSE)
251 bootverbose++;
252}
253SYSINIT(boot_verbose, SI_SUB_TUNABLES, SI_ORDER_ANY, set_boot_verbose, NULL)
254
255/*
256 ***************************************************************************
257 ****
258 **** The two following SYSINT's are proc0 specific glue code. I am not
259 **** convinced that they can not be safely combined, but their order of
260 **** operation has been maintained as the same as the original init_main.c
261 **** for right now.
262 ****
263 **** These probably belong in init_proc.c or kern_proc.c, since they
264 **** deal with proc0 (the fork template process).
265 ****
266 ***************************************************************************
267 */
268/* ARGSUSED*/
269static void
270proc0_init(void *dummy __unused)
271{
272 register struct proc *p;
273 register struct filedesc0 *fdp;
274 register unsigned i;
275 struct thread *td;
276 struct ksegrp *kg;
277 struct kse *ke;
278
279 GIANT_REQUIRED;
280 p = &proc0;
281 td = &thread0;
282
283 /*
284 * Initialize magic number.
285 */
286 p->p_magic = P_MAGIC;
287
288 /*
289 * Initialize thread, process and pgrp structures.
290 */
291 procinit();
292
293 /*
294 * Initialize sleep queue hash table
295 */
296 sleepinit();
297
298 /*
299 * additional VM structures
300 */
301 vm_init2();
302
303 /*
304 * Create process 0 (the swapper).
305 */
306 LIST_INSERT_HEAD(&allproc, p, p_list);
307 LIST_INSERT_HEAD(PIDHASH(0), p, p_hash);
43 */
44
45#include "opt_init_path.h"
46
47#include <sys/param.h>
48#include <sys/kernel.h>
49#include <sys/file.h>
50#include <sys/filedesc.h>
51#include <sys/ktr.h>
52#include <sys/lock.h>
53#include <sys/mount.h>
54#include <sys/mutex.h>
55#include <sys/sysctl.h>
56#include <sys/proc.h>
57#include <sys/resourcevar.h>
58#include <sys/systm.h>
59#include <sys/signalvar.h>
60#include <sys/vnode.h>
61#include <sys/sysent.h>
62#include <sys/reboot.h>
63#include <sys/sx.h>
64#include <sys/sysproto.h>
65#include <sys/vmmeter.h>
66#include <sys/unistd.h>
67#include <sys/malloc.h>
68#include <sys/conf.h>
69
70#include <machine/cpu.h>
71
72#include <vm/vm.h>
73#include <vm/vm_param.h>
74#include <vm/pmap.h>
75#include <vm/vm_map.h>
76#include <sys/user.h>
77#include <sys/copyright.h>
78
79void mi_startup(void); /* Should be elsewhere */
80
81/* Components of the first process -- never freed. */
82static struct session session0;
83static struct pgrp pgrp0;
84struct proc proc0;
85struct thread thread0;
86static struct procsig procsig0;
87static struct filedesc0 filedesc0;
88static struct plimit limit0;
89static struct vmspace vmspace0;
90struct proc *initproc;
91
92int cmask = CMASK;
93extern int fallback_elf_brand;
94
95struct vnode *rootvp;
96int boothowto = 0; /* initialized so that it can be patched */
97SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, "");
98int bootverbose;
99SYSCTL_INT(_debug, OID_AUTO, bootverbose, CTLFLAG_RW, &bootverbose, 0, "");
100
101/*
102 * This ensures that there is at least one entry so that the sysinit_set
103 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never
104 * executed.
105 */
106SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL)
107
108/*
109 * The sysinit table itself. Items are checked off as the are run.
110 * If we want to register new sysinit types, add them to newsysinit.
111 */
112SET_DECLARE(sysinit_set, struct sysinit);
113struct sysinit **sysinit, **sysinit_end;
114struct sysinit **newsysinit, **newsysinit_end;
115
116/*
117 * Merge a new sysinit set into the current set, reallocating it if
118 * necessary. This can only be called after malloc is running.
119 */
120void
121sysinit_add(struct sysinit **set, struct sysinit **set_end)
122{
123 struct sysinit **newset;
124 struct sysinit **sipp;
125 struct sysinit **xipp;
126 int count;
127
128 count = set_end - set;
129 if (newsysinit)
130 count += newsysinit_end - newsysinit;
131 else
132 count += sysinit_end - sysinit;
133 newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
134 if (newset == NULL)
135 panic("cannot malloc for sysinit");
136 xipp = newset;
137 if (newsysinit)
138 for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
139 *xipp++ = *sipp;
140 else
141 for (sipp = sysinit; sipp < sysinit_end; sipp++)
142 *xipp++ = *sipp;
143 for (sipp = set; sipp < set_end; sipp++)
144 *xipp++ = *sipp;
145 if (newsysinit)
146 free(newsysinit, M_TEMP);
147 newsysinit = newset;
148 newsysinit_end = newset + count;
149}
150
151/*
152 * System startup; initialize the world, create process 0, mount root
153 * filesystem, and fork to create init and pagedaemon. Most of the
154 * hard work is done in the lower-level initialization routines including
155 * startup(), which does memory initialization and autoconfiguration.
156 *
157 * This allows simple addition of new kernel subsystems that require
158 * boot time initialization. It also allows substitution of subsystem
159 * (for instance, a scheduler, kernel profiler, or VM system) by object
160 * module. Finally, it allows for optional "kernel threads".
161 */
162void
163mi_startup(void)
164{
165
166 register struct sysinit **sipp; /* system initialization*/
167 register struct sysinit **xipp; /* interior loop of sort*/
168 register struct sysinit *save; /* bubble*/
169
170 if (sysinit == NULL) {
171 sysinit = SET_BEGIN(sysinit_set);
172 sysinit_end = SET_LIMIT(sysinit_set);
173 }
174
175restart:
176 /*
177 * Perform a bubble sort of the system initialization objects by
178 * their subsystem (primary key) and order (secondary key).
179 */
180 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
181 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
182 if ((*sipp)->subsystem < (*xipp)->subsystem ||
183 ((*sipp)->subsystem == (*xipp)->subsystem &&
184 (*sipp)->order <= (*xipp)->order))
185 continue; /* skip*/
186 save = *sipp;
187 *sipp = *xipp;
188 *xipp = save;
189 }
190 }
191
192 /*
193 * Traverse the (now) ordered list of system initialization tasks.
194 * Perform each task, and continue on to the next task.
195 *
196 * The last item on the list is expected to be the scheduler,
197 * which will not return.
198 */
199 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
200
201 if ((*sipp)->subsystem == SI_SUB_DUMMY)
202 continue; /* skip dummy task(s)*/
203
204 if ((*sipp)->subsystem == SI_SUB_DONE)
205 continue;
206
207 /* Call function */
208 (*((*sipp)->func))((*sipp)->udata);
209
210 /* Check off the one we're just done */
211 (*sipp)->subsystem = SI_SUB_DONE;
212
213 /* Check if we've installed more sysinit items via KLD */
214 if (newsysinit != NULL) {
215 if (sysinit != SET_BEGIN(sysinit_set))
216 free(sysinit, M_TEMP);
217 sysinit = newsysinit;
218 sysinit_end = newsysinit_end;
219 newsysinit = NULL;
220 newsysinit_end = NULL;
221 goto restart;
222 }
223 }
224
225 panic("Shouldn't get here!");
226 /* NOTREACHED*/
227}
228
229
230/*
231 ***************************************************************************
232 ****
233 **** The following SYSINIT's belong elsewhere, but have not yet
234 **** been moved.
235 ****
236 ***************************************************************************
237 */
238static void
239print_caddr_t(void *data __unused)
240{
241 printf("%s", (char *)data);
242}
243SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
244SYSINIT(version, SI_SUB_COPYRIGHT, SI_ORDER_SECOND, print_caddr_t, version)
245
246static void
247set_boot_verbose(void *data __unused)
248{
249
250 if (boothowto & RB_VERBOSE)
251 bootverbose++;
252}
253SYSINIT(boot_verbose, SI_SUB_TUNABLES, SI_ORDER_ANY, set_boot_verbose, NULL)
254
255/*
256 ***************************************************************************
257 ****
258 **** The two following SYSINT's are proc0 specific glue code. I am not
259 **** convinced that they can not be safely combined, but their order of
260 **** operation has been maintained as the same as the original init_main.c
261 **** for right now.
262 ****
263 **** These probably belong in init_proc.c or kern_proc.c, since they
264 **** deal with proc0 (the fork template process).
265 ****
266 ***************************************************************************
267 */
268/* ARGSUSED*/
269static void
270proc0_init(void *dummy __unused)
271{
272 register struct proc *p;
273 register struct filedesc0 *fdp;
274 register unsigned i;
275 struct thread *td;
276 struct ksegrp *kg;
277 struct kse *ke;
278
279 GIANT_REQUIRED;
280 p = &proc0;
281 td = &thread0;
282
283 /*
284 * Initialize magic number.
285 */
286 p->p_magic = P_MAGIC;
287
288 /*
289 * Initialize thread, process and pgrp structures.
290 */
291 procinit();
292
293 /*
294 * Initialize sleep queue hash table
295 */
296 sleepinit();
297
298 /*
299 * additional VM structures
300 */
301 vm_init2();
302
303 /*
304 * Create process 0 (the swapper).
305 */
306 LIST_INSERT_HEAD(&allproc, p, p_list);
307 LIST_INSERT_HEAD(PIDHASH(0), p, p_hash);
308 mtx_init(&pgrp0.pg_mtx, "process group", MTX_DEF);
308 p->p_pgrp = &pgrp0;
309 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
310 LIST_INIT(&pgrp0.pg_members);
311 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
312
313 pgrp0.pg_session = &session0;
309 p->p_pgrp = &pgrp0;
310 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
311 LIST_INIT(&pgrp0.pg_members);
312 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
313
314 pgrp0.pg_session = &session0;
315 mtx_init(&session0.s_mtx, "session", MTX_DEF);
314 session0.s_count = 1;
315 session0.s_leader = p;
316
317#ifdef __ELF__
318 p->p_sysent = &elf_freebsd_sysvec;
319#else
320 p->p_sysent = &aout_sysvec;
321#endif
322
323 ke = &proc0.p_kse; /* XXXKSE */
324 kg = &proc0.p_ksegrp; /* XXXKSE */
325 p->p_flag = P_SYSTEM;
326 p->p_sflag = PS_INMEM;
327 p->p_stat = SRUN;
328 p->p_ksegrp.kg_nice = NZERO;
329 kg->kg_pri_class = PRI_TIMESHARE;
330 kg->kg_user_pri = PUSER;
331 td->td_priority = PVM;
332 td->td_base_pri = PUSER;
333
334 p->p_peers = 0;
335 p->p_leader = p;
336
337 bcopy("swapper", p->p_comm, sizeof ("swapper"));
338
339 callout_init(&p->p_itcallout, 0);
340 callout_init(&td->td_slpcallout, 1);
341
342 /* Create credentials. */
343 p->p_ucred = crget();
344 p->p_ucred->cr_ngroups = 1; /* group 0 */
345 p->p_ucred->cr_uidinfo = uifind(0);
346 p->p_ucred->cr_ruidinfo = uifind(0);
347 p->p_ucred->cr_prison = NULL; /* Don't jail it. */
348 td->td_ucred = crhold(p->p_ucred);
349
350 /* Create procsig. */
351 p->p_procsig = &procsig0;
352 p->p_procsig->ps_refcnt = 1;
353
354 /* Initialize signal state for process 0. */
355 siginit(&proc0);
356
357 /* Create the file descriptor table. */
358 fdp = &filedesc0;
359 p->p_fd = &fdp->fd_fd;
360 mtx_init(&fdp->fd_fd.fd_mtx, "struct filedesc", MTX_DEF);
361 fdp->fd_fd.fd_refcnt = 1;
362 fdp->fd_fd.fd_cmask = cmask;
363 fdp->fd_fd.fd_ofiles = fdp->fd_dfiles;
364 fdp->fd_fd.fd_ofileflags = fdp->fd_dfileflags;
365 fdp->fd_fd.fd_nfiles = NDFILE;
366
367 /* Create the limits structures. */
368 p->p_limit = &limit0;
369 for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++)
370 limit0.pl_rlimit[i].rlim_cur =
371 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
372 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur =
373 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
374 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur =
375 limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
376 i = ptoa(cnt.v_free_count);
377 limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i;
378 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
379 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
380 limit0.p_cpulimit = RLIM_INFINITY;
381 limit0.p_refcnt = 1;
382
383 /* Allocate a prototype map so we have something to fork. */
384 pmap_pinit0(vmspace_pmap(&vmspace0));
385 p->p_vmspace = &vmspace0;
386 vmspace0.vm_refcnt = 1;
387 vm_map_init(&vmspace0.vm_map, round_page(VM_MIN_ADDRESS),
388 trunc_page(VM_MAXUSER_ADDRESS));
389 vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0);
390
391 /*
392 * We continue to place resource usage info and signal
393 * actions in the user struct so they're pageable.
394 */
395 p->p_stats = &p->p_uarea->u_stats;
396 p->p_sigacts = &p->p_uarea->u_sigacts;
397
398 /*
399 * Charge root for one process.
400 */
401 (void)chgproccnt(p->p_ucred->cr_ruidinfo, 1, 0);
402}
403SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL)
404
405/* ARGSUSED*/
406static void
407proc0_post(void *dummy __unused)
408{
409 struct timespec ts;
410 struct proc *p;
411
412 /*
413 * Now we can look at the time, having had a chance to verify the
414 * time from the file system. Pretend that proc0 started now.
415 */
416 sx_slock(&allproc_lock);
417 LIST_FOREACH(p, &allproc, p_list) {
418 microtime(&p->p_stats->p_start);
419 p->p_runtime.sec = 0;
420 p->p_runtime.frac = 0;
421 }
422 sx_sunlock(&allproc_lock);
423 microuptime(PCPU_PTR(switchtime));
424 PCPU_SET(switchticks, ticks);
425
426 /*
427 * Give the ``random'' number generator a thump.
428 */
429 nanotime(&ts);
430 srandom(ts.tv_sec ^ ts.tv_nsec);
431}
432SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL)
433
434/*
435 ***************************************************************************
436 ****
437 **** The following SYSINIT's and glue code should be moved to the
438 **** respective files on a per subsystem basis.
439 ****
440 ***************************************************************************
441 */
442
443
444/*
445 ***************************************************************************
446 ****
447 **** The following code probably belongs in another file, like
448 **** kern/init_init.c.
449 ****
450 ***************************************************************************
451 */
452
453/*
454 * List of paths to try when searching for "init".
455 */
456static char init_path[MAXPATHLEN] =
457#ifdef INIT_PATH
458 __XSTRING(INIT_PATH);
459#else
460 "/sbin/init:/sbin/oinit:/sbin/init.bak:/stand/sysinstall";
461#endif
462SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0,
463 "Path used to search the init process");
464
465/*
466 * Start the initial user process; try exec'ing each pathname in init_path.
467 * The program is invoked with one argument containing the boot flags.
468 */
469static void
470start_init(void *dummy)
471{
472 vm_offset_t addr;
473 struct execve_args args;
474 int options, error;
475 char *var, *path, *next, *s;
476 char *ucp, **uap, *arg0, *arg1;
477 struct thread *td;
478 struct proc *p;
479 int init_does_devfs = 0;
480
481 mtx_lock(&Giant);
482
483 GIANT_REQUIRED;
484
485 td = curthread;
486 p = td->td_proc;
487
488 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */
489 if (VFS_ROOT(TAILQ_FIRST(&mountlist), &rootvnode))
490 panic("cannot find root vnode");
491 FILEDESC_LOCK(p->p_fd);
492 p->p_fd->fd_cdir = rootvnode;
493 VREF(p->p_fd->fd_cdir);
494 p->p_fd->fd_rdir = rootvnode;
495 VREF(p->p_fd->fd_rdir);
496 FILEDESC_UNLOCK(p->p_fd);
497 VOP_UNLOCK(rootvnode, 0, td);
498
499 if (devfs_present) {
500 /*
501 * For disk based systems, we probably cannot do this yet
502 * since the fs will be read-only. But a NFS root
503 * might be ok. It is worth a shot.
504 */
505 error = vn_mkdir("/dev", 0700, UIO_SYSSPACE, td);
506 if (error == EEXIST)
507 error = 0;
508 if (error == 0)
509 error = vfs_mount(td, "devfs", "/dev", 0, 0);
510 if (error != 0)
511 init_does_devfs = 1;
512 }
513
514 /*
515 * Need just enough stack to hold the faked-up "execve()" arguments.
516 */
517 addr = trunc_page(USRSTACK - PAGE_SIZE);
518 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
519 FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0)
520 panic("init: couldn't allocate argument space");
521 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
522 p->p_vmspace->vm_ssize = 1;
523
524 if ((var = getenv("init_path")) != NULL) {
525 strncpy(init_path, var, sizeof init_path);
526 init_path[sizeof init_path - 1] = 0;
527 }
528 if ((var = getenv("kern.fallback_elf_brand")) != NULL)
529 fallback_elf_brand = strtol(var, NULL, 0);
530
531 for (path = init_path; *path != '\0'; path = next) {
532 while (*path == ':')
533 path++;
534 if (*path == '\0')
535 break;
536 for (next = path; *next != '\0' && *next != ':'; next++)
537 /* nothing */ ;
538 if (bootverbose)
539 printf("start_init: trying %.*s\n", (int)(next - path),
540 path);
541
542 /*
543 * Move out the boot flag argument.
544 */
545 options = 0;
546 ucp = (char *)USRSTACK;
547 (void)subyte(--ucp, 0); /* trailing zero */
548 if (boothowto & RB_SINGLE) {
549 (void)subyte(--ucp, 's');
550 options = 1;
551 }
552#ifdef notyet
553 if (boothowto & RB_FASTBOOT) {
554 (void)subyte(--ucp, 'f');
555 options = 1;
556 }
557#endif
558
559#ifdef BOOTCDROM
560 (void)subyte(--ucp, 'C');
561 options = 1;
562#endif
563 if (init_does_devfs) {
564 (void)subyte(--ucp, 'd');
565 options = 1;
566 }
567
568 if (options == 0)
569 (void)subyte(--ucp, '-');
570 (void)subyte(--ucp, '-'); /* leading hyphen */
571 arg1 = ucp;
572
573 /*
574 * Move out the file name (also arg 0).
575 */
576 (void)subyte(--ucp, 0);
577 for (s = next - 1; s >= path; s--)
578 (void)subyte(--ucp, *s);
579 arg0 = ucp;
580
581 /*
582 * Move out the arg pointers.
583 */
584 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
585 (void)suword((caddr_t)--uap, (long)0); /* terminator */
586 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
587 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
588
589 /*
590 * Point at the arguments.
591 */
592 args.fname = arg0;
593 args.argv = uap;
594 args.envv = NULL;
595
596 /*
597 * Now try to exec the program. If can't for any reason
598 * other than it doesn't exist, complain.
599 *
600 * Otherwise, return via fork_trampoline() all the way
601 * to user mode as init!
602 */
603 if ((error = execve(td, &args)) == 0) {
604 mtx_unlock(&Giant);
605 return;
606 }
607 if (error != ENOENT)
608 printf("exec %.*s: error %d\n", (int)(next - path),
609 path, error);
610 }
611 printf("init: not found in path %s\n", init_path);
612 panic("no init");
613}
614
615/*
616 * Like kthread_create(), but runs in it's own address space.
617 * We do this early to reserve pid 1.
618 *
619 * Note special case - do not make it runnable yet. Other work
620 * in progress will change this more.
621 */
622static void
623create_init(const void *udata __unused)
624{
625 int error;
626
627 error = fork1(&thread0, RFFDG | RFPROC | RFSTOPPED, &initproc);
628 if (error)
629 panic("cannot fork init: %d\n", error);
630 PROC_LOCK(initproc);
631 initproc->p_flag |= P_SYSTEM;
632 PROC_UNLOCK(initproc);
633 mtx_lock_spin(&sched_lock);
634 initproc->p_sflag |= PS_INMEM;
635 mtx_unlock_spin(&sched_lock);
636 cpu_set_fork_handler(FIRST_THREAD_IN_PROC(initproc), start_init, NULL);
637}
638SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
639
640/*
641 * Make it runnable now.
642 */
643static void
644kick_init(const void *udata __unused)
645{
646 struct thread *td;
647
648 td = FIRST_THREAD_IN_PROC(initproc);
649 mtx_lock_spin(&sched_lock);
650 initproc->p_stat = SRUN;
651 setrunqueue(FIRST_THREAD_IN_PROC(initproc)); /* XXXKSE */
652 mtx_unlock_spin(&sched_lock);
653}
654SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)
316 session0.s_count = 1;
317 session0.s_leader = p;
318
319#ifdef __ELF__
320 p->p_sysent = &elf_freebsd_sysvec;
321#else
322 p->p_sysent = &aout_sysvec;
323#endif
324
325 ke = &proc0.p_kse; /* XXXKSE */
326 kg = &proc0.p_ksegrp; /* XXXKSE */
327 p->p_flag = P_SYSTEM;
328 p->p_sflag = PS_INMEM;
329 p->p_stat = SRUN;
330 p->p_ksegrp.kg_nice = NZERO;
331 kg->kg_pri_class = PRI_TIMESHARE;
332 kg->kg_user_pri = PUSER;
333 td->td_priority = PVM;
334 td->td_base_pri = PUSER;
335
336 p->p_peers = 0;
337 p->p_leader = p;
338
339 bcopy("swapper", p->p_comm, sizeof ("swapper"));
340
341 callout_init(&p->p_itcallout, 0);
342 callout_init(&td->td_slpcallout, 1);
343
344 /* Create credentials. */
345 p->p_ucred = crget();
346 p->p_ucred->cr_ngroups = 1; /* group 0 */
347 p->p_ucred->cr_uidinfo = uifind(0);
348 p->p_ucred->cr_ruidinfo = uifind(0);
349 p->p_ucred->cr_prison = NULL; /* Don't jail it. */
350 td->td_ucred = crhold(p->p_ucred);
351
352 /* Create procsig. */
353 p->p_procsig = &procsig0;
354 p->p_procsig->ps_refcnt = 1;
355
356 /* Initialize signal state for process 0. */
357 siginit(&proc0);
358
359 /* Create the file descriptor table. */
360 fdp = &filedesc0;
361 p->p_fd = &fdp->fd_fd;
362 mtx_init(&fdp->fd_fd.fd_mtx, "struct filedesc", MTX_DEF);
363 fdp->fd_fd.fd_refcnt = 1;
364 fdp->fd_fd.fd_cmask = cmask;
365 fdp->fd_fd.fd_ofiles = fdp->fd_dfiles;
366 fdp->fd_fd.fd_ofileflags = fdp->fd_dfileflags;
367 fdp->fd_fd.fd_nfiles = NDFILE;
368
369 /* Create the limits structures. */
370 p->p_limit = &limit0;
371 for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++)
372 limit0.pl_rlimit[i].rlim_cur =
373 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
374 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur =
375 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
376 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur =
377 limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
378 i = ptoa(cnt.v_free_count);
379 limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i;
380 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
381 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
382 limit0.p_cpulimit = RLIM_INFINITY;
383 limit0.p_refcnt = 1;
384
385 /* Allocate a prototype map so we have something to fork. */
386 pmap_pinit0(vmspace_pmap(&vmspace0));
387 p->p_vmspace = &vmspace0;
388 vmspace0.vm_refcnt = 1;
389 vm_map_init(&vmspace0.vm_map, round_page(VM_MIN_ADDRESS),
390 trunc_page(VM_MAXUSER_ADDRESS));
391 vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0);
392
393 /*
394 * We continue to place resource usage info and signal
395 * actions in the user struct so they're pageable.
396 */
397 p->p_stats = &p->p_uarea->u_stats;
398 p->p_sigacts = &p->p_uarea->u_sigacts;
399
400 /*
401 * Charge root for one process.
402 */
403 (void)chgproccnt(p->p_ucred->cr_ruidinfo, 1, 0);
404}
405SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL)
406
407/* ARGSUSED*/
408static void
409proc0_post(void *dummy __unused)
410{
411 struct timespec ts;
412 struct proc *p;
413
414 /*
415 * Now we can look at the time, having had a chance to verify the
416 * time from the file system. Pretend that proc0 started now.
417 */
418 sx_slock(&allproc_lock);
419 LIST_FOREACH(p, &allproc, p_list) {
420 microtime(&p->p_stats->p_start);
421 p->p_runtime.sec = 0;
422 p->p_runtime.frac = 0;
423 }
424 sx_sunlock(&allproc_lock);
425 microuptime(PCPU_PTR(switchtime));
426 PCPU_SET(switchticks, ticks);
427
428 /*
429 * Give the ``random'' number generator a thump.
430 */
431 nanotime(&ts);
432 srandom(ts.tv_sec ^ ts.tv_nsec);
433}
434SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL)
435
436/*
437 ***************************************************************************
438 ****
439 **** The following SYSINIT's and glue code should be moved to the
440 **** respective files on a per subsystem basis.
441 ****
442 ***************************************************************************
443 */
444
445
446/*
447 ***************************************************************************
448 ****
449 **** The following code probably belongs in another file, like
450 **** kern/init_init.c.
451 ****
452 ***************************************************************************
453 */
454
455/*
456 * List of paths to try when searching for "init".
457 */
458static char init_path[MAXPATHLEN] =
459#ifdef INIT_PATH
460 __XSTRING(INIT_PATH);
461#else
462 "/sbin/init:/sbin/oinit:/sbin/init.bak:/stand/sysinstall";
463#endif
464SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0,
465 "Path used to search the init process");
466
467/*
468 * Start the initial user process; try exec'ing each pathname in init_path.
469 * The program is invoked with one argument containing the boot flags.
470 */
471static void
472start_init(void *dummy)
473{
474 vm_offset_t addr;
475 struct execve_args args;
476 int options, error;
477 char *var, *path, *next, *s;
478 char *ucp, **uap, *arg0, *arg1;
479 struct thread *td;
480 struct proc *p;
481 int init_does_devfs = 0;
482
483 mtx_lock(&Giant);
484
485 GIANT_REQUIRED;
486
487 td = curthread;
488 p = td->td_proc;
489
490 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */
491 if (VFS_ROOT(TAILQ_FIRST(&mountlist), &rootvnode))
492 panic("cannot find root vnode");
493 FILEDESC_LOCK(p->p_fd);
494 p->p_fd->fd_cdir = rootvnode;
495 VREF(p->p_fd->fd_cdir);
496 p->p_fd->fd_rdir = rootvnode;
497 VREF(p->p_fd->fd_rdir);
498 FILEDESC_UNLOCK(p->p_fd);
499 VOP_UNLOCK(rootvnode, 0, td);
500
501 if (devfs_present) {
502 /*
503 * For disk based systems, we probably cannot do this yet
504 * since the fs will be read-only. But a NFS root
505 * might be ok. It is worth a shot.
506 */
507 error = vn_mkdir("/dev", 0700, UIO_SYSSPACE, td);
508 if (error == EEXIST)
509 error = 0;
510 if (error == 0)
511 error = vfs_mount(td, "devfs", "/dev", 0, 0);
512 if (error != 0)
513 init_does_devfs = 1;
514 }
515
516 /*
517 * Need just enough stack to hold the faked-up "execve()" arguments.
518 */
519 addr = trunc_page(USRSTACK - PAGE_SIZE);
520 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
521 FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0)
522 panic("init: couldn't allocate argument space");
523 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
524 p->p_vmspace->vm_ssize = 1;
525
526 if ((var = getenv("init_path")) != NULL) {
527 strncpy(init_path, var, sizeof init_path);
528 init_path[sizeof init_path - 1] = 0;
529 }
530 if ((var = getenv("kern.fallback_elf_brand")) != NULL)
531 fallback_elf_brand = strtol(var, NULL, 0);
532
533 for (path = init_path; *path != '\0'; path = next) {
534 while (*path == ':')
535 path++;
536 if (*path == '\0')
537 break;
538 for (next = path; *next != '\0' && *next != ':'; next++)
539 /* nothing */ ;
540 if (bootverbose)
541 printf("start_init: trying %.*s\n", (int)(next - path),
542 path);
543
544 /*
545 * Move out the boot flag argument.
546 */
547 options = 0;
548 ucp = (char *)USRSTACK;
549 (void)subyte(--ucp, 0); /* trailing zero */
550 if (boothowto & RB_SINGLE) {
551 (void)subyte(--ucp, 's');
552 options = 1;
553 }
554#ifdef notyet
555 if (boothowto & RB_FASTBOOT) {
556 (void)subyte(--ucp, 'f');
557 options = 1;
558 }
559#endif
560
561#ifdef BOOTCDROM
562 (void)subyte(--ucp, 'C');
563 options = 1;
564#endif
565 if (init_does_devfs) {
566 (void)subyte(--ucp, 'd');
567 options = 1;
568 }
569
570 if (options == 0)
571 (void)subyte(--ucp, '-');
572 (void)subyte(--ucp, '-'); /* leading hyphen */
573 arg1 = ucp;
574
575 /*
576 * Move out the file name (also arg 0).
577 */
578 (void)subyte(--ucp, 0);
579 for (s = next - 1; s >= path; s--)
580 (void)subyte(--ucp, *s);
581 arg0 = ucp;
582
583 /*
584 * Move out the arg pointers.
585 */
586 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
587 (void)suword((caddr_t)--uap, (long)0); /* terminator */
588 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
589 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
590
591 /*
592 * Point at the arguments.
593 */
594 args.fname = arg0;
595 args.argv = uap;
596 args.envv = NULL;
597
598 /*
599 * Now try to exec the program. If can't for any reason
600 * other than it doesn't exist, complain.
601 *
602 * Otherwise, return via fork_trampoline() all the way
603 * to user mode as init!
604 */
605 if ((error = execve(td, &args)) == 0) {
606 mtx_unlock(&Giant);
607 return;
608 }
609 if (error != ENOENT)
610 printf("exec %.*s: error %d\n", (int)(next - path),
611 path, error);
612 }
613 printf("init: not found in path %s\n", init_path);
614 panic("no init");
615}
616
617/*
618 * Like kthread_create(), but runs in it's own address space.
619 * We do this early to reserve pid 1.
620 *
621 * Note special case - do not make it runnable yet. Other work
622 * in progress will change this more.
623 */
624static void
625create_init(const void *udata __unused)
626{
627 int error;
628
629 error = fork1(&thread0, RFFDG | RFPROC | RFSTOPPED, &initproc);
630 if (error)
631 panic("cannot fork init: %d\n", error);
632 PROC_LOCK(initproc);
633 initproc->p_flag |= P_SYSTEM;
634 PROC_UNLOCK(initproc);
635 mtx_lock_spin(&sched_lock);
636 initproc->p_sflag |= PS_INMEM;
637 mtx_unlock_spin(&sched_lock);
638 cpu_set_fork_handler(FIRST_THREAD_IN_PROC(initproc), start_init, NULL);
639}
640SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
641
642/*
643 * Make it runnable now.
644 */
645static void
646kick_init(const void *udata __unused)
647{
648 struct thread *td;
649
650 td = FIRST_THREAD_IN_PROC(initproc);
651 mtx_lock_spin(&sched_lock);
652 initproc->p_stat = SRUN;
653 setrunqueue(FIRST_THREAD_IN_PROC(initproc)); /* XXXKSE */
654 mtx_unlock_spin(&sched_lock);
655}
656SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)