thread stage 3: create independant thread structure, unembed from proc.
[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 $
263e4574 43 * $DragonFly: src/sys/kern/init_main.c,v 1.4 2003/06/18 16:30:14 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;
83struct proc proc0;
263e4574 84static struct thread thread0;
984263bc
MD
85static struct pcred cred0;
86static struct procsig procsig0;
87static struct filedesc0 filedesc0;
88static struct plimit limit0;
89static struct vmspace vmspace0;
90struct proc *initproc;
91
92int cmask = CMASK;
93extern struct user *proc0paddr;
94extern int fallback_elf_brand;
95
96struct vnode *rootvp;
97int boothowto = 0; /* initialized so that it can be patched */
98SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, "");
99
100/*
101 * This ensures that there is at least one entry so that the sysinit_set
102 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never
103 * executed.
104 */
105SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL)
106
107/*
108 * The sysinit table itself. Items are checked off as the are run.
109 * If we want to register new sysinit types, add them to newsysinit.
110 */
111struct sysinit **sysinit = (struct sysinit **)sysinit_set.ls_items;
112struct sysinit **newsysinit;
113
114/*
115 * Merge a new sysinit set into the current set, reallocating it if
116 * necessary. This can only be called after malloc is running.
117 */
118void
119sysinit_add(struct sysinit **set)
120{
121 struct sysinit **newset;
122 struct sysinit **sipp;
123 struct sysinit **xipp;
124 int count = 0;
125
126 if (newsysinit)
127 for (sipp = newsysinit; *sipp; sipp++)
128 count++;
129 else
130 for (sipp = sysinit; *sipp; sipp++)
131 count++;
132 for (sipp = set; *sipp; sipp++)
133 count++;
134 count++; /* Trailing NULL */
135 newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT);
136 if (newset == NULL)
137 panic("cannot malloc for sysinit");
138 xipp = newset;
139 if (newsysinit)
140 for (sipp = newsysinit; *sipp; sipp++)
141 *xipp++ = *sipp;
142 else
143 for (sipp = sysinit; *sipp; sipp++)
144 *xipp++ = *sipp;
145 for (sipp = set; *sipp; sipp++)
146 *xipp++ = *sipp;
147 *xipp = NULL;
148 if (newsysinit)
149 free(newsysinit, M_TEMP);
150 newsysinit = newset;
151}
152
153/*
154 * System startup; initialize the world, create process 0, mount root
155 * filesystem, and fork to create init and pagedaemon. Most of the
156 * hard work is done in the lower-level initialization routines including
157 * startup(), which does memory initialization and autoconfiguration.
158 *
159 * This allows simple addition of new kernel subsystems that require
160 * boot time initialization. It also allows substitution of subsystem
161 * (for instance, a scheduler, kernel profiler, or VM system) by object
162 * module. Finally, it allows for optional "kernel threads".
163 */
164void
165mi_startup(void)
166{
167
168 register struct sysinit **sipp; /* system initialization*/
169 register struct sysinit **xipp; /* interior loop of sort*/
170 register struct sysinit *save; /* bubble*/
171
172restart:
173 /*
174 * Perform a bubble sort of the system initialization objects by
175 * their subsystem (primary key) and order (secondary key).
176 */
177 for (sipp = sysinit; *sipp; sipp++) {
178 for (xipp = sipp + 1; *xipp; xipp++) {
179 if ((*sipp)->subsystem < (*xipp)->subsystem ||
180 ((*sipp)->subsystem == (*xipp)->subsystem &&
181 (*sipp)->order <= (*xipp)->order))
182 continue; /* skip*/
183 save = *sipp;
184 *sipp = *xipp;
185 *xipp = save;
186 }
187 }
188
189 /*
190 * Traverse the (now) ordered list of system initialization tasks.
191 * Perform each task, and continue on to the next task.
192 *
193 * The last item on the list is expected to be the scheduler,
194 * which will not return.
195 */
196 for (sipp = sysinit; *sipp; sipp++) {
197
198 if ((*sipp)->subsystem == SI_SUB_DUMMY)
199 continue; /* skip dummy task(s)*/
200
201 if ((*sipp)->subsystem == SI_SUB_DONE)
202 continue;
203
204 /* Call function */
205 (*((*sipp)->func))((*sipp)->udata);
206
207 /* Check off the one we're just done */
208 (*sipp)->subsystem = SI_SUB_DONE;
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
240
241/*
242 ***************************************************************************
243 ****
244 **** The two following SYSINT's are proc0 specific glue code. I am not
245 **** convinced that they can not be safely combined, but their order of
246 **** operation has been maintained as the same as the original init_main.c
247 **** for right now.
248 ****
249 **** These probably belong in init_proc.c or kern_proc.c, since they
250 **** deal with proc0 (the fork template process).
251 ****
252 ***************************************************************************
253 */
254/* ARGSUSED*/
255static void
256proc0_init(void *dummy __unused)
257{
258 register struct proc *p;
263e4574 259 register struct thread *td;
984263bc
MD
260 register struct filedesc0 *fdp;
261 register unsigned i;
262
263 p = &proc0;
263e4574 264 td = &thread0;
984263bc
MD
265
266 /*
267 * Initialize process and pgrp structures.
268 */
269 procinit();
270
271 /*
272 * Initialize sleep queue hash table
273 */
274 sleepinit();
275
276 /*
277 * additional VM structures
278 */
279 vm_init2();
280
281 /*
282 * Create process 0 (the swapper).
283 */
284 LIST_INSERT_HEAD(&allproc, p, p_list);
285 p->p_pgrp = &pgrp0;
286 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
287 LIST_INIT(&pgrp0.pg_members);
288 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
289
290 pgrp0.pg_session = &session0;
291 session0.s_count = 1;
292 session0.s_leader = p;
293
294 p->p_sysent = &aout_sysvec;
295
296 p->p_flag = P_INMEM | P_SYSTEM;
297 p->p_stat = SRUN;
298 p->p_nice = NZERO;
299 p->p_rtprio.type = RTP_PRIO_NORMAL;
300 p->p_rtprio.prio = 0;
301
302 p->p_peers = 0;
303 p->p_leader = p;
304
305 bcopy("swapper", p->p_comm, sizeof ("swapper"));
306
307 /* Create credentials. */
308 cred0.p_refcnt = 1;
309 cred0.p_uidinfo = uifind(0);
310 p->p_cred = &cred0;
311 p->p_ucred = crget();
312 p->p_ucred->cr_ngroups = 1; /* group 0 */
313 p->p_ucred->cr_uidinfo = uifind(0);
314
315 /* Don't jail it */
316 p->p_prison = 0;
317
318 /* Create procsig. */
319 p->p_procsig = &procsig0;
320 p->p_procsig->ps_refcnt = 1;
321
322 /* Initialize signal state for process 0. */
323 siginit(&proc0);
324
325 /* Create the file descriptor table. */
326 fdp = &filedesc0;
327 p->p_fd = &fdp->fd_fd;
328 p->p_fdtol = NULL;
329 fdp->fd_fd.fd_refcnt = 1;
330 fdp->fd_fd.fd_cmask = cmask;
331 fdp->fd_fd.fd_ofiles = fdp->fd_dfiles;
332 fdp->fd_fd.fd_ofileflags = fdp->fd_dfileflags;
333 fdp->fd_fd.fd_nfiles = NDFILE;
334
335 /* Create the limits structures. */
336 p->p_limit = &limit0;
337 for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++)
338 limit0.pl_rlimit[i].rlim_cur =
339 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
340 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur =
341 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
342 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur =
343 limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
344 i = ptoa(cnt.v_free_count);
345 limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i;
346 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
347 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
348 limit0.p_cpulimit = RLIM_INFINITY;
349 limit0.p_refcnt = 1;
350
351 /* Allocate a prototype map so we have something to fork. */
352 pmap_pinit0(vmspace_pmap(&vmspace0));
353 p->p_vmspace = &vmspace0;
354 vmspace0.vm_refcnt = 1;
355 vm_map_init(&vmspace0.vm_map, round_page(VM_MIN_ADDRESS),
356 trunc_page(VM_MAXUSER_ADDRESS));
357 vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0);
358 p->p_addr = proc0paddr; /* XXX */
263e4574
MD
359 p->p_thread = td;
360 td->td_proc = p;
984263bc
MD
361
362 /*
363 * We continue to place resource usage info and signal
364 * actions in the user struct so they're pageable.
365 */
366 p->p_stats = &p->p_addr->u_stats;
367 p->p_sigacts = &p->p_addr->u_sigacts;
368
369 /*
370 * Charge root for one process.
371 */
372 (void)chgproccnt(cred0.p_uidinfo, 1, 0);
373
374 /*
375 * Initialize the current process pointer (curproc) before
376 * any possible traps/probes to simplify trap processing.
377 */
263e4574 378 SET_CURTHREAD(p->p_thread);
984263bc
MD
379
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 */
394 LIST_FOREACH(p, &allproc, p_list) {
395 microtime(&p->p_stats->p_start);
396 p->p_runtime = 0;
397 }
398 microuptime(&switchtime);
399 switchticks = ticks;
400
401 /*
402 * Give the ``random'' number generator a thump.
403 * XXX: Does read_random() contain enough bits to be used here ?
404 */
405 nanotime(&ts);
406 srandom(ts.tv_sec ^ ts.tv_nsec);
407}
408SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL)
409
410/*
411 ***************************************************************************
412 ****
413 **** The following SYSINIT's and glue code should be moved to the
414 **** respective files on a per subsystem basis.
415 ****
416 ***************************************************************************
417 */
418
419
420/*
421 ***************************************************************************
422 ****
423 **** The following code probably belongs in another file, like
424 **** kern/init_init.c.
425 ****
426 ***************************************************************************
427 */
428
429/*
430 * List of paths to try when searching for "init".
431 */
432static char init_path[MAXPATHLEN] =
433#ifdef INIT_PATH
434 __XSTRING(INIT_PATH);
435#else
436 "/sbin/init:/sbin/oinit:/sbin/init.bak:/stand/sysinstall";
437#endif
438SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, "");
439
440/*
441 * Start the initial user process; try exec'ing each pathname in init_path.
442 * The program is invoked with one argument containing the boot flags.
443 */
444static void
445start_init(void *dummy)
446{
447 vm_offset_t addr;
448 struct execve_args args;
449 int options, error;
450 char *var, *path, *next, *s;
451 char *ucp, **uap, *arg0, *arg1;
452 struct proc *p;
453
454 p = curproc;
455
456 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */
457 if (VFS_ROOT(TAILQ_FIRST(&mountlist), &rootvnode))
458 panic("cannot find root vnode");
459 p->p_fd->fd_cdir = rootvnode;
460 VREF(p->p_fd->fd_cdir);
461 p->p_fd->fd_rdir = rootvnode;
462 VREF(p->p_fd->fd_rdir);
463 VOP_UNLOCK(rootvnode, 0, p);
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!
548 */
549 if ((error = execve(p, &args)) == 0)
550 return;
551 if (error != ENOENT)
552 printf("exec %.*s: error %d\n", (int)(next - path),
553 path, error);
554 }
555 printf("init: not found in path %s\n", init_path);
556 panic("no init");
557}
558
559/*
560 * Like kthread_create(), but runs in it's own address space.
561 * We do this early to reserve pid 1.
562 *
563 * Note special case - do not make it runnable yet. Other work
564 * in progress will change this more.
565 */
566static void
567create_init(const void *udata __unused)
568{
569 int error;
570 int s;
571
572 s = splhigh();
573 error = fork1(&proc0, RFFDG | RFPROC, &initproc);
574 if (error)
575 panic("cannot fork init: %d\n", error);
576 initproc->p_flag |= P_INMEM | P_SYSTEM;
577 cpu_set_fork_handler(initproc, start_init, NULL);
578 remrunqueue(initproc);
579 splx(s);
580}
581SYSINIT(init,SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
582
583/*
584 * Make it runnable now.
585 */
586static void
587kick_init(const void *udata __unused)
588{
589 setrunqueue(initproc);
590}
591SYSINIT(kickinit,SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)