Fully synchronize sys/boot from FreeBSD-5.x, but add / to the module path
[dragonfly.git] / sys / kern / init_main.c
CommitLineData
984263bc
MD
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: src/sys/kern/init_main.c,v 1.134.2.8 2003/06/06 20:21:32 tegge Exp $
7ea21ed1 43 * $DragonFly: src/sys/kern/init_main.c,v 1.25 2003/09/28 03:44:01 dillon Exp $
984263bc
MD
44 */
45
46#include "opt_init_path.h"
47
48#include <sys/param.h>
49#include <sys/file.h>
50#include <sys/filedesc.h>
51#include <sys/kernel.h>
52#include <sys/mount.h>
53#include <sys/sysctl.h>
54#include <sys/proc.h>
55#include <sys/resourcevar.h>
56#include <sys/signalvar.h>
57#include <sys/systm.h>
58#include <sys/vnode.h>
59#include <sys/sysent.h>
60#include <sys/reboot.h>
61#include <sys/sysproto.h>
62#include <sys/vmmeter.h>
63#include <sys/unistd.h>
64#include <sys/malloc.h>
dadab5e9 65#include <sys/file2.h>
f8c3996b 66#include <sys/thread2.h>
984263bc
MD
67
68#include <machine/cpu.h>
69
70#include <vm/vm.h>
71#include <vm/vm_param.h>
72#include <sys/lock.h>
73#include <vm/pmap.h>
74#include <vm/vm_map.h>
75#include <sys/user.h>
76#include <sys/copyright.h>
77
78extern struct linker_set sysinit_set; /* XXX */
79
80void mi_startup(void); /* Should be elsewhere */
81
82/* Components of the first process -- never freed. */
83static struct session session0;
84static struct pgrp pgrp0;
984263bc
MD
85static struct procsig procsig0;
86static struct filedesc0 filedesc0;
87static struct plimit limit0;
88static struct vmspace vmspace0;
b7c628e4
MD
89struct proc *initproc;
90struct proc proc0;
91struct thread thread0;
984263bc
MD
92
93int cmask = CMASK;
94extern struct user *proc0paddr;
95extern int fallback_elf_brand;
96
97struct vnode *rootvp;
98int boothowto = 0; /* initialized so that it can be patched */
99SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 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 */
112struct sysinit **sysinit = (struct sysinit **)sysinit_set.ls_items;
113struct sysinit **newsysinit;
114
115/*
116 * Merge a new sysinit set into the current set, reallocating it if
117 * necessary. This can only be called after malloc is running.
118 */
119void
120sysinit_add(struct sysinit **set)
121{
122 struct sysinit **newset;
123 struct sysinit **sipp;
124 struct sysinit **xipp;
125 int count = 0;
126
127 if (newsysinit)
128 for (sipp = newsysinit; *sipp; sipp++)
129 count++;
130 else
131 for (sipp = sysinit; *sipp; sipp++)
132 count++;
133 for (sipp = set; *sipp; sipp++)
134 count++;
135 count++; /* Trailing NULL */
136 newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
137 if (newset == NULL)
138 panic("cannot malloc for sysinit");
139 xipp = newset;
140 if (newsysinit)
141 for (sipp = newsysinit; *sipp; sipp++)
142 *xipp++ = *sipp;
143 else
144 for (sipp = sysinit; *sipp; sipp++)
145 *xipp++ = *sipp;
146 for (sipp = set; *sipp; sipp++)
147 *xipp++ = *sipp;
148 *xipp = NULL;
149 if (newsysinit)
150 free(newsysinit, M_TEMP);
151 newsysinit = newset;
152}
153
154/*
155 * System startup; initialize the world, create process 0, mount root
156 * filesystem, and fork to create init and pagedaemon. Most of the
157 * hard work is done in the lower-level initialization routines including
158 * startup(), which does memory initialization and autoconfiguration.
159 *
160 * This allows simple addition of new kernel subsystems that require
161 * boot time initialization. It also allows substitution of subsystem
162 * (for instance, a scheduler, kernel profiler, or VM system) by object
163 * module. Finally, it allows for optional "kernel threads".
164 */
165void
166mi_startup(void)
167{
8a8d5d85 168 struct sysinit *sip; /* system initialization*/
17a9f566
MD
169 struct sysinit **sipp; /* system initialization*/
170 struct sysinit **xipp; /* interior loop of sort*/
171 struct sysinit *save; /* bubble*/
984263bc
MD
172
173restart:
174 /*
175 * Perform a bubble sort of the system initialization objects by
176 * their subsystem (primary key) and order (secondary key).
177 */
178 for (sipp = sysinit; *sipp; sipp++) {
179 for (xipp = sipp + 1; *xipp; xipp++) {
180 if ((*sipp)->subsystem < (*xipp)->subsystem ||
181 ((*sipp)->subsystem == (*xipp)->subsystem &&
182 (*sipp)->order <= (*xipp)->order))
183 continue; /* skip*/
184 save = *sipp;
185 *sipp = *xipp;
186 *xipp = save;
187 }
188 }
189
190 /*
191 * Traverse the (now) ordered list of system initialization tasks.
192 * Perform each task, and continue on to the next task.
193 *
194 * The last item on the list is expected to be the scheduler,
195 * which will not return.
196 */
8a8d5d85
MD
197 for (sipp = sysinit; (sip = *sipp) != NULL; sipp++) {
198 if (sip->subsystem == SI_SUB_DUMMY)
984263bc
MD
199 continue; /* skip dummy task(s)*/
200
8a8d5d85 201 if (sip->subsystem == SI_SUB_DONE)
984263bc
MD
202 continue;
203
204 /* Call function */
8a8d5d85 205 (*(sip->func))(sip->udata);
984263bc
MD
206
207 /* Check off the one we're just done */
8a8d5d85 208 sip->subsystem = SI_SUB_DONE;
984263bc
MD
209
210 /* Check if we've installed more sysinit items via KLD */
211 if (newsysinit != NULL) {
212 if (sysinit != (struct sysinit **)sysinit_set.ls_items)
213 free(sysinit, M_TEMP);
214 sysinit = newsysinit;
215 newsysinit = NULL;
216 goto restart;
217 }
218 }
219
220 panic("Shouldn't get here!");
221 /* NOTREACHED*/
222}
223
224
225/*
226 ***************************************************************************
227 ****
228 **** The following SYSINIT's belong elsewhere, but have not yet
229 **** been moved.
230 ****
231 ***************************************************************************
232 */
233static void
234print_caddr_t(void *data __unused)
235{
236 printf("%s", (char *)data);
237}
238SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
239
f8c3996b
MD
240/*
241 * Leave the critical section that protected us from spurious interrupts
242 * so device probes work.
243 */
244static void
245leavecrit(void *dummy __unused)
246{
247 crit_exit();
248 KKASSERT(!IN_CRITICAL_SECT(curthread));
249 if (bootverbose)
250 printf("Leaving critical section, allowing interrupts\n");
251}
252SYSINIT(leavecrit, SI_SUB_LEAVE_CRIT, SI_ORDER_ANY, leavecrit, NULL)
984263bc
MD
253
254/*
255 ***************************************************************************
256 ****
257 **** The two following SYSINT's are proc0 specific glue code. I am not
258 **** convinced that they can not be safely combined, but their order of
259 **** operation has been maintained as the same as the original init_main.c
260 **** for right now.
261 ****
262 **** These probably belong in init_proc.c or kern_proc.c, since they
263 **** deal with proc0 (the fork template process).
264 ****
265 ***************************************************************************
266 */
267/* ARGSUSED*/
268static void
269proc0_init(void *dummy __unused)
270{
1fd87d54
RG
271 struct proc *p;
272 struct filedesc0 *fdp;
273 unsigned i;
984263bc
MD
274
275 p = &proc0;
276
277 /*
278 * Initialize process and pgrp structures.
279 */
280 procinit();
281
282 /*
283 * Initialize sleep queue hash table
284 */
285 sleepinit();
286
287 /*
288 * additional VM structures
289 */
290 vm_init2();
291
292 /*
293 * Create process 0 (the swapper).
294 */
295 LIST_INSERT_HEAD(&allproc, p, p_list);
296 p->p_pgrp = &pgrp0;
297 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
298 LIST_INIT(&pgrp0.pg_members);
299 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
300
301 pgrp0.pg_session = &session0;
302 session0.s_count = 1;
303 session0.s_leader = p;
304
305 p->p_sysent = &aout_sysvec;
306
a2a5ad0d 307 p->p_flag = P_INMEM | P_SYSTEM | P_CP_RELEASED;
984263bc
MD
308 p->p_stat = SRUN;
309 p->p_nice = NZERO;
310 p->p_rtprio.type = RTP_PRIO_NORMAL;
311 p->p_rtprio.prio = 0;
312
313 p->p_peers = 0;
314 p->p_leader = p;
315
316 bcopy("swapper", p->p_comm, sizeof ("swapper"));
317
318 /* Create credentials. */
984263bc 319 p->p_ucred = crget();
41c20dac 320 p->p_ucred->cr_ruidinfo = uifind(0);
984263bc
MD
321 p->p_ucred->cr_ngroups = 1; /* group 0 */
322 p->p_ucred->cr_uidinfo = uifind(0);
323
324 /* Don't jail it */
41c20dac 325 p->p_ucred->cr_prison = NULL;
984263bc
MD
326
327 /* Create procsig. */
328 p->p_procsig = &procsig0;
329 p->p_procsig->ps_refcnt = 1;
330
331 /* Initialize signal state for process 0. */
332 siginit(&proc0);
333
334 /* Create the file descriptor table. */
335 fdp = &filedesc0;
336 p->p_fd = &fdp->fd_fd;
337 p->p_fdtol = NULL;
338 fdp->fd_fd.fd_refcnt = 1;
339 fdp->fd_fd.fd_cmask = cmask;
340 fdp->fd_fd.fd_ofiles = fdp->fd_dfiles;
341 fdp->fd_fd.fd_ofileflags = fdp->fd_dfileflags;
342 fdp->fd_fd.fd_nfiles = NDFILE;
343
344 /* Create the limits structures. */
345 p->p_limit = &limit0;
346 for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++)
347 limit0.pl_rlimit[i].rlim_cur =
348 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
349 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur =
350 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
351 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur =
352 limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
12e4aaff 353 i = ptoa(vmstats.v_free_count);
984263bc
MD
354 limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i;
355 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
356 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
357 limit0.p_cpulimit = RLIM_INFINITY;
358 limit0.p_refcnt = 1;
359
360 /* Allocate a prototype map so we have something to fork. */
361 pmap_pinit0(vmspace_pmap(&vmspace0));
362 p->p_vmspace = &vmspace0;
363 vmspace0.vm_refcnt = 1;
364 vm_map_init(&vmspace0.vm_map, round_page(VM_MIN_ADDRESS),
365 trunc_page(VM_MAXUSER_ADDRESS));
366 vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0);
984263bc
MD
367
368 /*
369 * We continue to place resource usage info and signal
370 * actions in the user struct so they're pageable.
371 */
372 p->p_stats = &p->p_addr->u_stats;
373 p->p_sigacts = &p->p_addr->u_sigacts;
374
375 /*
376 * Charge root for one process.
377 */
41c20dac 378 (void)chgproccnt(p->p_ucred->cr_uidinfo, 1, 0);
984263bc 379
984263bc
MD
380}
381SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL)
382
383/* ARGSUSED*/
384static void
385proc0_post(void *dummy __unused)
386{
387 struct timespec ts;
388 struct proc *p;
389
390 /*
391 * Now we can look at the time, having had a chance to verify the
392 * time from the file system. Pretend that proc0 started now.
393 */
f62004ad 394 FOREACH_PROC_IN_SYSTEM(p) {
984263bc 395 microtime(&p->p_stats->p_start);
984263bc 396 }
984263bc
MD
397
398 /*
399 * Give the ``random'' number generator a thump.
400 * XXX: Does read_random() contain enough bits to be used here ?
401 */
402 nanotime(&ts);
403 srandom(ts.tv_sec ^ ts.tv_nsec);
404}
405SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL)
406
407/*
408 ***************************************************************************
409 ****
410 **** The following SYSINIT's and glue code should be moved to the
411 **** respective files on a per subsystem basis.
412 ****
413 ***************************************************************************
414 */
415
416
417/*
418 ***************************************************************************
419 ****
420 **** The following code probably belongs in another file, like
421 **** kern/init_init.c.
422 ****
423 ***************************************************************************
424 */
425
426/*
427 * List of paths to try when searching for "init".
428 */
429static char init_path[MAXPATHLEN] =
430#ifdef INIT_PATH
431 __XSTRING(INIT_PATH);
432#else
433 "/sbin/init:/sbin/oinit:/sbin/init.bak:/stand/sysinstall";
434#endif
435SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, "");
436
437/*
438 * Start the initial user process; try exec'ing each pathname in init_path.
439 * The program is invoked with one argument containing the boot flags.
8a8d5d85
MD
440 *
441 * The MP lock is held on entry.
984263bc
MD
442 */
443static void
444start_init(void *dummy)
445{
446 vm_offset_t addr;
447 struct execve_args args;
448 int options, error;
449 char *var, *path, *next, *s;
450 char *ucp, **uap, *arg0, *arg1;
451 struct proc *p;
452
453 p = curproc;
454
455 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */
456 if (VFS_ROOT(TAILQ_FIRST(&mountlist), &rootvnode))
457 panic("cannot find root vnode");
458 p->p_fd->fd_cdir = rootvnode;
459 VREF(p->p_fd->fd_cdir);
460 p->p_fd->fd_rdir = rootvnode;
461 VREF(p->p_fd->fd_rdir);
7ea21ed1 462 vfs_cache_setroot(rootvnode);
dadab5e9 463 VOP_UNLOCK(rootvnode, 0, curthread);
984263bc
MD
464
465 /*
466 * Need just enough stack to hold the faked-up "execve()" arguments.
467 */
468 addr = trunc_page(USRSTACK - PAGE_SIZE);
469 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
470 FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0)
471 panic("init: couldn't allocate argument space");
472 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
473 p->p_vmspace->vm_ssize = 1;
474
475 if ((var = getenv("init_path")) != NULL) {
476 strncpy(init_path, var, sizeof init_path);
477 init_path[sizeof init_path - 1] = 0;
478 }
479 if ((var = getenv("kern.fallback_elf_brand")) != NULL)
480 fallback_elf_brand = strtol(var, NULL, 0);
481
482 for (path = init_path; *path != '\0'; path = next) {
483 while (*path == ':')
484 path++;
485 if (*path == '\0')
486 break;
487 for (next = path; *next != '\0' && *next != ':'; next++)
488 /* nothing */ ;
489 if (bootverbose)
490 printf("start_init: trying %.*s\n", (int)(next - path),
491 path);
492
493 /*
494 * Move out the boot flag argument.
495 */
496 options = 0;
497 ucp = (char *)USRSTACK;
498 (void)subyte(--ucp, 0); /* trailing zero */
499 if (boothowto & RB_SINGLE) {
500 (void)subyte(--ucp, 's');
501 options = 1;
502 }
503#ifdef notyet
504 if (boothowto & RB_FASTBOOT) {
505 (void)subyte(--ucp, 'f');
506 options = 1;
507 }
508#endif
509
510#ifdef BOOTCDROM
511 (void)subyte(--ucp, 'C');
512 options = 1;
513#endif
514 if (options == 0)
515 (void)subyte(--ucp, '-');
516 (void)subyte(--ucp, '-'); /* leading hyphen */
517 arg1 = ucp;
518
519 /*
520 * Move out the file name (also arg 0).
521 */
522 (void)subyte(--ucp, 0);
523 for (s = next - 1; s >= path; s--)
524 (void)subyte(--ucp, *s);
525 arg0 = ucp;
526
527 /*
528 * Move out the arg pointers.
529 */
530 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
531 (void)suword((caddr_t)--uap, (long)0); /* terminator */
532 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
533 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
534
535 /*
536 * Point at the arguments.
537 */
538 args.fname = arg0;
539 args.argv = uap;
540 args.envv = NULL;
541
542 /*
543 * Now try to exec the program. If can't for any reason
544 * other than it doesn't exist, complain.
545 *
546 * Otherwise, return via fork_trampoline() all the way
547 * to user mode as init!
a2a5ad0d
MD
548 *
549 * WARNING! We may have been moved to another cpu after
550 * acquiring P_CURPROC. The MP lock will migrate with us
551 * though so we still have to release it.
984263bc 552 */
8a8d5d85 553 if ((error = execve(&args)) == 0) {
a2a5ad0d 554 acquire_curproc(p);
8a8d5d85 555 rel_mplock();
984263bc 556 return;
8a8d5d85 557 }
984263bc
MD
558 if (error != ENOENT)
559 printf("exec %.*s: error %d\n", (int)(next - path),
560 path, error);
561 }
562 printf("init: not found in path %s\n", init_path);
563 panic("no init");
564}
565
566/*
567 * Like kthread_create(), but runs in it's own address space.
568 * We do this early to reserve pid 1.
569 *
570 * Note special case - do not make it runnable yet. Other work
571 * in progress will change this more.
572 */
573static void
574create_init(const void *udata __unused)
575{
576 int error;
577 int s;
578
579 s = splhigh();
580 error = fork1(&proc0, RFFDG | RFPROC, &initproc);
581 if (error)
582 panic("cannot fork init: %d\n", error);
583 initproc->p_flag |= P_INMEM | P_SYSTEM;
584 cpu_set_fork_handler(initproc, start_init, NULL);
984263bc
MD
585 splx(s);
586}
587SYSINIT(init,SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
588
589/*
590 * Make it runnable now.
591 */
592static void
593kick_init(const void *udata __unused)
594{
7d0bac62 595 start_forked_proc(&proc0, initproc);
984263bc
MD
596}
597SYSINIT(kickinit,SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)
8ad65e08
MD
598
599/*
600 * Machine independant globaldata initialization
73e4f7b9
MD
601 *
602 * WARNING! Called from early boot, 'mycpu' may not work yet.
8ad65e08
MD
603 */
604void
d0e06f83 605mi_gdinit(struct globaldata *gd, int cpuid)
8ad65e08 606{
85100692 607 TAILQ_INIT(&gd->gd_tdfreeq); /* for pmap_{new,dispose}_thread() */
d0e06f83 608 gd->gd_cpuid = cpuid;
8ad65e08
MD
609 lwkt_gdinit(gd);
610}
611
612