proc->thread stage 3.5: Add an IO_CORE flag so coda doesn't have to dig
[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 $
d16a8831 43 * $DragonFly: src/sys/kern/init_main.c,v 1.11 2003/06/23 23:36:11 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>
65
66#include <machine/cpu.h>
67
68#include <vm/vm.h>
69#include <vm/vm_param.h>
70#include <sys/lock.h>
71#include <vm/pmap.h>
72#include <vm/vm_map.h>
73#include <sys/user.h>
74#include <sys/copyright.h>
75
76extern struct linker_set sysinit_set; /* XXX */
77
78void mi_startup(void); /* Should be elsewhere */
79
80/* Components of the first process -- never freed. */
81static struct session session0;
82static struct pgrp pgrp0;
984263bc
MD
83static struct procsig procsig0;
84static struct filedesc0 filedesc0;
85static struct plimit limit0;
86static struct vmspace vmspace0;
b7c628e4
MD
87struct proc *initproc;
88struct proc proc0;
89struct thread thread0;
984263bc
MD
90
91int cmask = CMASK;
92extern struct user *proc0paddr;
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, "");
98
99/*
100 * This ensures that there is at least one entry so that the sysinit_set
101 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never
102 * executed.
103 */
104SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL)
105
106/*
107 * The sysinit table itself. Items are checked off as the are run.
108 * If we want to register new sysinit types, add them to newsysinit.
109 */
110struct sysinit **sysinit = (struct sysinit **)sysinit_set.ls_items;
111struct sysinit **newsysinit;
112
113/*
114 * Merge a new sysinit set into the current set, reallocating it if
115 * necessary. This can only be called after malloc is running.
116 */
117void
118sysinit_add(struct sysinit **set)
119{
120 struct sysinit **newset;
121 struct sysinit **sipp;
122 struct sysinit **xipp;
123 int count = 0;
124
125 if (newsysinit)
126 for (sipp = newsysinit; *sipp; sipp++)
127 count++;
128 else
129 for (sipp = sysinit; *sipp; sipp++)
130 count++;
131 for (sipp = set; *sipp; sipp++)
132 count++;
133 count++; /* Trailing NULL */
134 newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
135 if (newset == NULL)
136 panic("cannot malloc for sysinit");
137 xipp = newset;
138 if (newsysinit)
139 for (sipp = newsysinit; *sipp; sipp++)
140 *xipp++ = *sipp;
141 else
142 for (sipp = sysinit; *sipp; sipp++)
143 *xipp++ = *sipp;
144 for (sipp = set; *sipp; sipp++)
145 *xipp++ = *sipp;
146 *xipp = NULL;
147 if (newsysinit)
148 free(newsysinit, M_TEMP);
149 newsysinit = newset;
150}
151
152/*
153 * System startup; initialize the world, create process 0, mount root
154 * filesystem, and fork to create init and pagedaemon. Most of the
155 * hard work is done in the lower-level initialization routines including
156 * startup(), which does memory initialization and autoconfiguration.
157 *
158 * This allows simple addition of new kernel subsystems that require
159 * boot time initialization. It also allows substitution of subsystem
160 * (for instance, a scheduler, kernel profiler, or VM system) by object
161 * module. Finally, it allows for optional "kernel threads".
162 */
163void
164mi_startup(void)
165{
166
167 register struct sysinit **sipp; /* system initialization*/
168 register struct sysinit **xipp; /* interior loop of sort*/
169 register struct sysinit *save; /* bubble*/
170
171restart:
172 /*
173 * Perform a bubble sort of the system initialization objects by
174 * their subsystem (primary key) and order (secondary key).
175 */
176 for (sipp = sysinit; *sipp; sipp++) {
177 for (xipp = sipp + 1; *xipp; xipp++) {
178 if ((*sipp)->subsystem < (*xipp)->subsystem ||
179 ((*sipp)->subsystem == (*xipp)->subsystem &&
180 (*sipp)->order <= (*xipp)->order))
181 continue; /* skip*/
182 save = *sipp;
183 *sipp = *xipp;
184 *xipp = save;
185 }
186 }
187
188 /*
189 * Traverse the (now) ordered list of system initialization tasks.
190 * Perform each task, and continue on to the next task.
191 *
192 * The last item on the list is expected to be the scheduler,
193 * which will not return.
194 */
195 for (sipp = sysinit; *sipp; sipp++) {
196
197 if ((*sipp)->subsystem == SI_SUB_DUMMY)
198 continue; /* skip dummy task(s)*/
199
200 if ((*sipp)->subsystem == SI_SUB_DONE)
201 continue;
202
203 /* Call function */
204 (*((*sipp)->func))((*sipp)->udata);
205
206 /* Check off the one we're just done */
207 (*sipp)->subsystem = SI_SUB_DONE;
208
209 /* Check if we've installed more sysinit items via KLD */
210 if (newsysinit != NULL) {
211 if (sysinit != (struct sysinit **)sysinit_set.ls_items)
212 free(sysinit, M_TEMP);
213 sysinit = newsysinit;
214 newsysinit = NULL;
215 goto restart;
216 }
217 }
218
219 panic("Shouldn't get here!");
220 /* NOTREACHED*/
221}
222
223
224/*
225 ***************************************************************************
226 ****
227 **** The following SYSINIT's belong elsewhere, but have not yet
228 **** been moved.
229 ****
230 ***************************************************************************
231 */
232static void
233print_caddr_t(void *data __unused)
234{
235 printf("%s", (char *)data);
236}
237SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
238
239
240/*
241 ***************************************************************************
242 ****
243 **** The two following SYSINT's are proc0 specific glue code. I am not
244 **** convinced that they can not be safely combined, but their order of
245 **** operation has been maintained as the same as the original init_main.c
246 **** for right now.
247 ****
248 **** These probably belong in init_proc.c or kern_proc.c, since they
249 **** deal with proc0 (the fork template process).
250 ****
251 ***************************************************************************
252 */
253/* ARGSUSED*/
254static void
255proc0_init(void *dummy __unused)
256{
257 register struct proc *p;
258 register struct filedesc0 *fdp;
259 register unsigned i;
260
261 p = &proc0;
262
263 /*
264 * Initialize process and pgrp structures.
265 */
266 procinit();
267
268 /*
269 * Initialize sleep queue hash table
270 */
271 sleepinit();
272
273 /*
274 * additional VM structures
275 */
276 vm_init2();
277
278 /*
279 * Create process 0 (the swapper).
280 */
281 LIST_INSERT_HEAD(&allproc, p, p_list);
282 p->p_pgrp = &pgrp0;
283 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
284 LIST_INIT(&pgrp0.pg_members);
285 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
286
287 pgrp0.pg_session = &session0;
288 session0.s_count = 1;
289 session0.s_leader = p;
290
291 p->p_sysent = &aout_sysvec;
292
293 p->p_flag = P_INMEM | P_SYSTEM;
294 p->p_stat = SRUN;
295 p->p_nice = NZERO;
296 p->p_rtprio.type = RTP_PRIO_NORMAL;
297 p->p_rtprio.prio = 0;
298
299 p->p_peers = 0;
300 p->p_leader = p;
301
302 bcopy("swapper", p->p_comm, sizeof ("swapper"));
303
304 /* Create credentials. */
984263bc 305 p->p_ucred = crget();
41c20dac 306 p->p_ucred->cr_ruidinfo = uifind(0);
984263bc
MD
307 p->p_ucred->cr_ngroups = 1; /* group 0 */
308 p->p_ucred->cr_uidinfo = uifind(0);
309
310 /* Don't jail it */
41c20dac 311 p->p_ucred->cr_prison = NULL;
984263bc
MD
312
313 /* Create procsig. */
314 p->p_procsig = &procsig0;
315 p->p_procsig->ps_refcnt = 1;
316
317 /* Initialize signal state for process 0. */
318 siginit(&proc0);
319
320 /* Create the file descriptor table. */
321 fdp = &filedesc0;
322 p->p_fd = &fdp->fd_fd;
323 p->p_fdtol = NULL;
324 fdp->fd_fd.fd_refcnt = 1;
325 fdp->fd_fd.fd_cmask = cmask;
326 fdp->fd_fd.fd_ofiles = fdp->fd_dfiles;
327 fdp->fd_fd.fd_ofileflags = fdp->fd_dfileflags;
328 fdp->fd_fd.fd_nfiles = NDFILE;
329
330 /* Create the limits structures. */
331 p->p_limit = &limit0;
332 for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++)
333 limit0.pl_rlimit[i].rlim_cur =
334 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
335 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur =
336 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
337 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur =
338 limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
339 i = ptoa(cnt.v_free_count);
340 limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i;
341 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
342 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
343 limit0.p_cpulimit = RLIM_INFINITY;
344 limit0.p_refcnt = 1;
345
346 /* Allocate a prototype map so we have something to fork. */
347 pmap_pinit0(vmspace_pmap(&vmspace0));
348 p->p_vmspace = &vmspace0;
349 vmspace0.vm_refcnt = 1;
350 vm_map_init(&vmspace0.vm_map, round_page(VM_MIN_ADDRESS),
351 trunc_page(VM_MAXUSER_ADDRESS));
352 vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0);
984263bc
MD
353
354 /*
355 * We continue to place resource usage info and signal
356 * actions in the user struct so they're pageable.
357 */
358 p->p_stats = &p->p_addr->u_stats;
359 p->p_sigacts = &p->p_addr->u_sigacts;
360
361 /*
362 * Charge root for one process.
363 */
41c20dac 364 (void)chgproccnt(p->p_ucred->cr_uidinfo, 1, 0);
984263bc 365
984263bc
MD
366}
367SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL)
368
369/* ARGSUSED*/
370static void
371proc0_post(void *dummy __unused)
372{
373 struct timespec ts;
374 struct proc *p;
375
376 /*
377 * Now we can look at the time, having had a chance to verify the
378 * time from the file system. Pretend that proc0 started now.
379 */
380 LIST_FOREACH(p, &allproc, p_list) {
381 microtime(&p->p_stats->p_start);
984263bc 382 }
984263bc
MD
383
384 /*
385 * Give the ``random'' number generator a thump.
386 * XXX: Does read_random() contain enough bits to be used here ?
387 */
388 nanotime(&ts);
389 srandom(ts.tv_sec ^ ts.tv_nsec);
390}
391SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL)
392
393/*
394 ***************************************************************************
395 ****
396 **** The following SYSINIT's and glue code should be moved to the
397 **** respective files on a per subsystem basis.
398 ****
399 ***************************************************************************
400 */
401
402
403/*
404 ***************************************************************************
405 ****
406 **** The following code probably belongs in another file, like
407 **** kern/init_init.c.
408 ****
409 ***************************************************************************
410 */
411
412/*
413 * List of paths to try when searching for "init".
414 */
415static char init_path[MAXPATHLEN] =
416#ifdef INIT_PATH
417 __XSTRING(INIT_PATH);
418#else
419 "/sbin/init:/sbin/oinit:/sbin/init.bak:/stand/sysinstall";
420#endif
421SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, "");
422
423/*
424 * Start the initial user process; try exec'ing each pathname in init_path.
425 * The program is invoked with one argument containing the boot flags.
426 */
427static void
428start_init(void *dummy)
429{
430 vm_offset_t addr;
431 struct execve_args args;
432 int options, error;
433 char *var, *path, *next, *s;
434 char *ucp, **uap, *arg0, *arg1;
435 struct proc *p;
436
437 p = curproc;
438
439 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */
440 if (VFS_ROOT(TAILQ_FIRST(&mountlist), &rootvnode))
441 panic("cannot find root vnode");
442 p->p_fd->fd_cdir = rootvnode;
443 VREF(p->p_fd->fd_cdir);
444 p->p_fd->fd_rdir = rootvnode;
445 VREF(p->p_fd->fd_rdir);
446 VOP_UNLOCK(rootvnode, 0, p);
447
448 /*
449 * Need just enough stack to hold the faked-up "execve()" arguments.
450 */
451 addr = trunc_page(USRSTACK - PAGE_SIZE);
452 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE,
453 FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0)
454 panic("init: couldn't allocate argument space");
455 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
456 p->p_vmspace->vm_ssize = 1;
457
458 if ((var = getenv("init_path")) != NULL) {
459 strncpy(init_path, var, sizeof init_path);
460 init_path[sizeof init_path - 1] = 0;
461 }
462 if ((var = getenv("kern.fallback_elf_brand")) != NULL)
463 fallback_elf_brand = strtol(var, NULL, 0);
464
465 for (path = init_path; *path != '\0'; path = next) {
466 while (*path == ':')
467 path++;
468 if (*path == '\0')
469 break;
470 for (next = path; *next != '\0' && *next != ':'; next++)
471 /* nothing */ ;
472 if (bootverbose)
473 printf("start_init: trying %.*s\n", (int)(next - path),
474 path);
475
476 /*
477 * Move out the boot flag argument.
478 */
479 options = 0;
480 ucp = (char *)USRSTACK;
481 (void)subyte(--ucp, 0); /* trailing zero */
482 if (boothowto & RB_SINGLE) {
483 (void)subyte(--ucp, 's');
484 options = 1;
485 }
486#ifdef notyet
487 if (boothowto & RB_FASTBOOT) {
488 (void)subyte(--ucp, 'f');
489 options = 1;
490 }
491#endif
492
493#ifdef BOOTCDROM
494 (void)subyte(--ucp, 'C');
495 options = 1;
496#endif
497 if (options == 0)
498 (void)subyte(--ucp, '-');
499 (void)subyte(--ucp, '-'); /* leading hyphen */
500 arg1 = ucp;
501
502 /*
503 * Move out the file name (also arg 0).
504 */
505 (void)subyte(--ucp, 0);
506 for (s = next - 1; s >= path; s--)
507 (void)subyte(--ucp, *s);
508 arg0 = ucp;
509
510 /*
511 * Move out the arg pointers.
512 */
513 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
514 (void)suword((caddr_t)--uap, (long)0); /* terminator */
515 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
516 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
517
518 /*
519 * Point at the arguments.
520 */
521 args.fname = arg0;
522 args.argv = uap;
523 args.envv = NULL;
524
525 /*
526 * Now try to exec the program. If can't for any reason
527 * other than it doesn't exist, complain.
528 *
529 * Otherwise, return via fork_trampoline() all the way
530 * to user mode as init!
531 */
41c20dac 532 if ((error = execve(&args)) == 0)
984263bc
MD
533 return;
534 if (error != ENOENT)
535 printf("exec %.*s: error %d\n", (int)(next - path),
536 path, error);
537 }
538 printf("init: not found in path %s\n", init_path);
539 panic("no init");
540}
541
542/*
543 * Like kthread_create(), but runs in it's own address space.
544 * We do this early to reserve pid 1.
545 *
546 * Note special case - do not make it runnable yet. Other work
547 * in progress will change this more.
548 */
549static void
550create_init(const void *udata __unused)
551{
552 int error;
553 int s;
554
555 s = splhigh();
556 error = fork1(&proc0, RFFDG | RFPROC, &initproc);
557 if (error)
558 panic("cannot fork init: %d\n", error);
559 initproc->p_flag |= P_INMEM | P_SYSTEM;
560 cpu_set_fork_handler(initproc, start_init, NULL);
984263bc
MD
561 splx(s);
562}
563SYSINIT(init,SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
564
565/*
566 * Make it runnable now.
567 */
568static void
569kick_init(const void *udata __unused)
570{
7d0bac62 571 start_forked_proc(&proc0, initproc);
984263bc
MD
572}
573SYSINIT(kickinit,SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)
8ad65e08
MD
574
575/*
576 * Machine independant globaldata initialization
577 */
578void
579mi_gdinit(struct globaldata *gd, int cpu)
580{
f1d1c3fa 581 gd->gd_cpu = cpu;
8ad65e08
MD
582 lwkt_gdinit(gd);
583}
584
585