intr: Further delay MachIntrABI.finalize()
[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 $
7bd34050 43 * $DragonFly: src/sys/kern/init_main.c,v 1.87 2008/06/07 11:37:23 mneumann 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>
0b692e79 65#include <sys/machintr.h>
684a93c4 66
dadab5e9 67#include <sys/file2.h>
f8c3996b 68#include <sys/thread2.h>
e3161323 69#include <sys/sysref2.h>
8f1f6170 70#include <sys/spinlock2.h>
684a93c4 71#include <sys/mplock2.h>
984263bc
MD
72
73#include <machine/cpu.h>
74
75#include <vm/vm.h>
76#include <vm/vm_param.h>
77#include <sys/lock.h>
78#include <vm/pmap.h>
79#include <vm/vm_map.h>
ba39e2e0 80#include <vm/vm_extern.h>
984263bc
MD
81#include <sys/user.h>
82#include <sys/copyright.h>
83
cd29885a
MD
84int vfs_mountroot_devfs(void);
85
984263bc
MD
86/* Components of the first process -- never freed. */
87static struct session session0;
88static struct pgrp pgrp0;
b1b4e5a6 89static struct sigacts sigacts0;
0679adc4 90static struct filedesc filedesc0;
984263bc
MD
91static struct plimit limit0;
92static struct vmspace vmspace0;
b7c628e4
MD
93struct proc *initproc;
94struct proc proc0;
08f2f1bb 95struct lwp lwp0;
b7c628e4 96struct thread thread0;
984263bc
MD
97
98int cmask = CMASK;
d87e79b9 99u_int cpu_mi_feature;
916e604f 100cpumask_t usched_global_cpumask;
984263bc
MD
101extern struct user *proc0paddr;
102extern int fallback_elf_brand;
103
984263bc 104int boothowto = 0; /* initialized so that it can be patched */
0c52fa62
SG
105SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0,
106 "Reboot flags, from console subsystem");
916e604f
MD
107SYSCTL_ULONG(_kern, OID_AUTO, usched_global_cpumask, CTLFLAG_RW,
108 &usched_global_cpumask, 0, "global user scheduler cpumask");
984263bc
MD
109
110/*
111 * This ensures that there is at least one entry so that the sysinit_set
7bd34050 112 * symbol is not undefined. A subsystem ID of SI_SPECIAL_DUMMY is never
984263bc
MD
113 * executed.
114 */
ba39e2e0 115SYSINIT(placeholder, SI_SPECIAL_DUMMY, SI_ORDER_ANY, NULL, NULL)
984263bc
MD
116
117/*
118 * The sysinit table itself. Items are checked off as the are run.
119 * If we want to register new sysinit types, add them to newsysinit.
120 */
dc62b251
MD
121SET_DECLARE(sysinit_set, struct sysinit);
122struct sysinit **sysinit, **sysinit_end;
123struct sysinit **newsysinit, **newsysinit_end;
124
984263bc
MD
125
126/*
127 * Merge a new sysinit set into the current set, reallocating it if
128 * necessary. This can only be called after malloc is running.
129 */
130void
dc62b251 131sysinit_add(struct sysinit **set, struct sysinit **set_end)
984263bc
MD
132{
133 struct sysinit **newset;
134 struct sysinit **sipp;
135 struct sysinit **xipp;
dc62b251 136 int count;
984263bc 137
dc62b251 138 count = set_end - set;
984263bc 139 if (newsysinit)
dc62b251 140 count += newsysinit_end - newsysinit;
984263bc 141 else
dc62b251 142 count += sysinit_end - sysinit;
efda3bd0 143 newset = kmalloc(count * sizeof(*sipp), M_TEMP, M_WAITOK);
984263bc 144 xipp = newset;
dc62b251
MD
145 if (newsysinit) {
146 for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
984263bc 147 *xipp++ = *sipp;
dc62b251
MD
148 } else {
149 for (sipp = sysinit; sipp < sysinit_end; sipp++)
984263bc 150 *xipp++ = *sipp;
dc62b251
MD
151 }
152 for (sipp = set; sipp < set_end; sipp++)
984263bc 153 *xipp++ = *sipp;
984263bc 154 if (newsysinit)
efda3bd0 155 kfree(newsysinit, M_TEMP);
984263bc 156 newsysinit = newset;
dc62b251 157 newsysinit_end = newset + count;
984263bc
MD
158}
159
160/*
6bf59cd2
MD
161 * Callbacks from machine-dependant startup code (e.g. init386) to set
162 * up low level entities related to cpu #0's globaldata.
163 *
164 * Called from very low level boot code.
165 */
166void
167mi_proc0init(struct globaldata *gd, struct user *proc0paddr)
168{
fdce8919 169 lwkt_init_thread(&thread0, proc0paddr, LWKT_THREAD_STACK, 0, gd);
6bf59cd2 170 lwkt_set_comm(&thread0, "thread0");
3e291793 171 RB_INIT(&proc0.p_lwp_tree);
8f1f6170 172 spin_init(&proc0.p_spin);
3e291793
MD
173 proc0.p_lasttid = 0; /* +1 = next TID */
174 lwp_rb_tree_RB_INSERT(&proc0.p_lwp_tree, &lwp0);
08f2f1bb
SS
175 lwp0.lwp_thread = &thread0;
176 lwp0.lwp_proc = &proc0;
6bf59cd2 177 proc0.p_usched = usched_init();
da23a592 178 lwp0.lwp_cpumask = (cpumask_t)-1;
6bf59cd2
MD
179 varsymset_init(&proc0.p_varsymset, NULL);
180 thread0.td_flags |= TDF_RUNNING;
181 thread0.td_proc = &proc0;
08f2f1bb 182 thread0.td_lwp = &lwp0;
6179d4f3 183 thread0.td_switch = cpu_lwkt_switch;
6b7dca13 184 lwkt_schedule_self(curthread);
6bf59cd2
MD
185}
186
187/*
984263bc
MD
188 * System startup; initialize the world, create process 0, mount root
189 * filesystem, and fork to create init and pagedaemon. Most of the
190 * hard work is done in the lower-level initialization routines including
191 * startup(), which does memory initialization and autoconfiguration.
192 *
193 * This allows simple addition of new kernel subsystems that require
194 * boot time initialization. It also allows substitution of subsystem
195 * (for instance, a scheduler, kernel profiler, or VM system) by object
196 * module. Finally, it allows for optional "kernel threads".
197 */
198void
199mi_startup(void)
200{
8a8d5d85 201 struct sysinit *sip; /* system initialization*/
17a9f566
MD
202 struct sysinit **sipp; /* system initialization*/
203 struct sysinit **xipp; /* interior loop of sort*/
204 struct sysinit *save; /* bubble*/
984263bc 205
dc62b251
MD
206 if (sysinit == NULL) {
207 sysinit = SET_BEGIN(sysinit_set);
ca97d438
JG
208#if defined(__amd64__) && defined(_KERNEL_VIRTUAL)
209 /*
210 * XXX For whatever reason, on 64-bit vkernels
211 * the value of sysinit obtained from the
212 * linker set is wrong.
213 */
214 if ((long)sysinit % 8 != 0) {
8608b858 215 kprintf("Fixing sysinit value...\n");
ca97d438
JG
216 sysinit = (long)sysinit + 4;
217 }
218#endif
dc62b251
MD
219 sysinit_end = SET_LIMIT(sysinit_set);
220 }
ca97d438
JG
221#if defined(__amd64__) && defined(_KERNEL_VIRTUAL)
222 KKASSERT((long)sysinit % 8 == 0);
223#endif
dc62b251 224
984263bc
MD
225restart:
226 /*
227 * Perform a bubble sort of the system initialization objects by
228 * their subsystem (primary key) and order (secondary key).
229 */
dc62b251
MD
230 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
231 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
984263bc
MD
232 if ((*sipp)->subsystem < (*xipp)->subsystem ||
233 ((*sipp)->subsystem == (*xipp)->subsystem &&
234 (*sipp)->order <= (*xipp)->order))
235 continue; /* skip*/
236 save = *sipp;
237 *sipp = *xipp;
238 *xipp = save;
239 }
240 }
241
242 /*
243 * Traverse the (now) ordered list of system initialization tasks.
244 * Perform each task, and continue on to the next task.
245 *
246 * The last item on the list is expected to be the scheduler,
247 * which will not return.
248 */
dc62b251
MD
249 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
250 sip = *sipp;
ba39e2e0 251 if (sip->subsystem == SI_SPECIAL_DUMMY)
984263bc
MD
252 continue; /* skip dummy task(s)*/
253
ba39e2e0 254 if (sip->subsystem == SI_SPECIAL_DONE)
984263bc
MD
255 continue;
256
257 /* Call function */
8a8d5d85 258 (*(sip->func))(sip->udata);
984263bc
MD
259
260 /* Check off the one we're just done */
ba39e2e0 261 sip->subsystem = SI_SPECIAL_DONE;
984263bc
MD
262
263 /* Check if we've installed more sysinit items via KLD */
264 if (newsysinit != NULL) {
dc62b251 265 if (sysinit != SET_BEGIN(sysinit_set))
efda3bd0 266 kfree(sysinit, M_TEMP);
984263bc 267 sysinit = newsysinit;
dc62b251 268 sysinit_end = newsysinit_end;
984263bc 269 newsysinit = NULL;
dc62b251 270 newsysinit_end = NULL;
984263bc
MD
271 goto restart;
272 }
273 }
274
275 panic("Shouldn't get here!");
276 /* NOTREACHED*/
277}
278
279
280/*
281 ***************************************************************************
282 ****
283 **** The following SYSINIT's belong elsewhere, but have not yet
284 **** been moved.
285 ****
286 ***************************************************************************
287 */
288static void
289print_caddr_t(void *data __unused)
290{
6ea70f76 291 kprintf("%s", (char *)data);
984263bc 292}
ba39e2e0 293SYSINIT(announce, SI_BOOT1_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
984263bc 294
f8c3996b
MD
295/*
296 * Leave the critical section that protected us from spurious interrupts
297 * so device probes work.
298 */
299static void
300leavecrit(void *dummy __unused)
301{
7bf5fa56 302 MachIntrABI.stabilize();
0b692e79
MD
303 cpu_enable_intr();
304 MachIntrABI.cleanup();
f8c3996b
MD
305 crit_exit();
306 KKASSERT(!IN_CRITICAL_SECT(curthread));
dbcd0c9b 307
f8c3996b 308 if (bootverbose)
6ea70f76 309 kprintf("Leaving critical section, allowing interrupts\n");
f8c3996b 310}
ba39e2e0 311SYSINIT(leavecrit, SI_BOOT2_LEAVE_CRIT, SI_ORDER_ANY, leavecrit, NULL)
984263bc
MD
312
313/*
dbcd0c9b
MD
314 * This is called after the threading system is up and running,
315 * including the softclock, clock interrupts, and SMP.
316 */
317static void
318tsleepworks(void *dummy __unused)
319{
320 tsleep_now_works = 1;
321}
322SYSINIT(tsleepworks, SI_BOOT2_FINISH_SMP, SI_ORDER_SECOND, tsleepworks, NULL)
323
324/*
325 * This is called after devices have configured. Tell the kernel we are
326 * no longer in cold boot.
327 */
328static void
329endofcoldboot(void *dummy __unused)
330{
331 cold = 0;
332}
333SYSINIT(endofcoldboot, SI_SUB_ISWARM, SI_ORDER_ANY, endofcoldboot, NULL)
334
335/*
984263bc
MD
336 ***************************************************************************
337 ****
338 **** The two following SYSINT's are proc0 specific glue code. I am not
339 **** convinced that they can not be safely combined, but their order of
340 **** operation has been maintained as the same as the original init_main.c
341 **** for right now.
342 ****
343 **** These probably belong in init_proc.c or kern_proc.c, since they
344 **** deal with proc0 (the fork template process).
345 ****
346 ***************************************************************************
347 */
348/* ARGSUSED*/
349static void
350proc0_init(void *dummy __unused)
351{
0679adc4 352 struct proc *p;
81eea9f4 353 struct lwp *lp;
984263bc
MD
354
355 p = &proc0;
08f2f1bb 356 lp = &lwp0;
984263bc
MD
357
358 /*
359 * Initialize process and pgrp structures.
360 */
361 procinit();
362
363 /*
984263bc
MD
364 * additional VM structures
365 */
366 vm_init2();
367
368 /*
369 * Create process 0 (the swapper).
370 */
371 LIST_INSERT_HEAD(&allproc, p, p_list);
372 p->p_pgrp = &pgrp0;
373 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
374 LIST_INIT(&pgrp0.pg_members);
375 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
376
377 pgrp0.pg_session = &session0;
378 session0.s_count = 1;
379 session0.s_leader = p;
380
381 p->p_sysent = &aout_sysvec;
382
344ad853 383 p->p_flag = P_SYSTEM;
164b8401
SS
384 p->p_stat = SACTIVE;
385 lp->lwp_stat = LSRUN;
984263bc
MD
386 p->p_nice = NZERO;
387 p->p_rtprio.type = RTP_PRIO_NORMAL;
388 p->p_rtprio.prio = 0;
08f2f1bb 389 lp->lwp_rtprio = p->p_rtprio;
984263bc
MD
390
391 p->p_peers = 0;
392 p->p_leader = p;
393
394 bcopy("swapper", p->p_comm, sizeof ("swapper"));
81eea9f4 395 bcopy("swapper", thread0.td_comm, sizeof ("swapper"));
984263bc
MD
396
397 /* Create credentials. */
984263bc 398 p->p_ucred = crget();
41c20dac 399 p->p_ucred->cr_ruidinfo = uifind(0);
984263bc
MD
400 p->p_ucred->cr_ngroups = 1; /* group 0 */
401 p->p_ucred->cr_uidinfo = uifind(0);
9910d07b 402 thread0.td_ucred = crhold(p->p_ucred); /* bootstrap fork1() */
984263bc
MD
403
404 /* Don't jail it */
41c20dac 405 p->p_ucred->cr_prison = NULL;
984263bc 406
b1b4e5a6
SS
407 /* Create sigacts. */
408 p->p_sigacts = &sigacts0;
409 p->p_sigacts->ps_refcnt = 1;
984263bc
MD
410
411 /* Initialize signal state for process 0. */
c0b8a06d 412 siginit(p);
984263bc
MD
413
414 /* Create the file descriptor table. */
228b401d 415 fdinit_bootstrap(p, &filedesc0, cmask);
984263bc
MD
416
417 /* Create the limits structures. */
c0b8a06d 418 plimit_init0(&limit0);
984263bc 419 p->p_limit = &limit0;
984263bc
MD
420
421 /* Allocate a prototype map so we have something to fork. */
422 pmap_pinit0(vmspace_pmap(&vmspace0));
423 p->p_vmspace = &vmspace0;
287ebb09 424 lp->lwp_vmspace = p->p_vmspace;
e3161323 425 sysref_init(&vmspace0.vm_sysref, &vmspace_sysref_class);
e4846942
MD
426 vm_map_init(&vmspace0.vm_map,
427 round_page(VM_MIN_USER_ADDRESS),
428 trunc_page(VM_MAX_USER_ADDRESS),
429 vmspace_pmap(&vmspace0));
e3161323 430 sysref_activate(&vmspace0.vm_sysref);
984263bc 431
a591f597
MD
432 kqueue_init(&lwp0.lwp_kqueue, &filedesc0);
433
984263bc 434 /*
984263bc
MD
435 * Charge root for one process.
436 */
41c20dac 437 (void)chgproccnt(p->p_ucred->cr_uidinfo, 1, 0);
ba39e2e0 438 vm_init_limits(p);
984263bc 439}
ba39e2e0 440SYSINIT(p0init, SI_BOOT2_PROC0, SI_ORDER_FIRST, proc0_init, NULL)
984263bc 441
8fa76237
MD
442static int proc0_post_callback(struct proc *p, void *data __unused);
443
984263bc
MD
444/* ARGSUSED*/
445static void
446proc0_post(void *dummy __unused)
447{
448 struct timespec ts;
984263bc
MD
449
450 /*
451 * Now we can look at the time, having had a chance to verify the
452 * time from the file system. Pretend that proc0 started now.
453 */
8fa76237 454 allproc_scan(proc0_post_callback, NULL);
984263bc
MD
455
456 /*
457 * Give the ``random'' number generator a thump.
458 * XXX: Does read_random() contain enough bits to be used here ?
459 */
460 nanotime(&ts);
cddfb7bb 461 skrandom(ts.tv_sec ^ ts.tv_nsec);
984263bc 462}
8fa76237
MD
463
464static int
465proc0_post_callback(struct proc *p, void *data __unused)
466{
467 microtime(&p->p_start);
468 return(0);
469}
470
ba39e2e0 471SYSINIT(p0post, SI_SUB_PROC0_POST, SI_ORDER_FIRST, proc0_post, NULL)
984263bc
MD
472
473/*
474 ***************************************************************************
475 ****
476 **** The following SYSINIT's and glue code should be moved to the
477 **** respective files on a per subsystem basis.
478 ****
479 ***************************************************************************
480 */
481
482
483/*
484 ***************************************************************************
485 ****
486 **** The following code probably belongs in another file, like
487 **** kern/init_init.c.
488 ****
489 ***************************************************************************
490 */
491
492/*
493 * List of paths to try when searching for "init".
494 */
495static char init_path[MAXPATHLEN] =
496#ifdef INIT_PATH
497 __XSTRING(INIT_PATH);
498#else
22628b14 499 "/sbin/init:/sbin/oinit:/sbin/init.bak";
984263bc
MD
500#endif
501SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, "");
502
503/*
504 * Start the initial user process; try exec'ing each pathname in init_path.
505 * The program is invoked with one argument containing the boot flags.
506 */
507static void
91bd9c1e 508start_init(void *dummy, struct trapframe *frame)
984263bc
MD
509{
510 vm_offset_t addr;
511 struct execve_args args;
512 int options, error;
513 char *var, *path, *next, *s;
514 char *ucp, **uap, *arg0, *arg1;
515 struct proc *p;
553ea3c8 516 struct lwp *lp;
21739618
MD
517 struct mount *mp;
518 struct vnode *vp;
fc350ba1
MD
519 char *env;
520
521 /*
522 * This is passed in by the bootloader
523 */
524 env = kgetenv("kernelname");
525 if (env != NULL)
526 strlcpy(kernelname, env, sizeof(kernelname));
984263bc 527
2b0bd8aa
MD
528 /*
529 * The MP lock is not held on entry. We release it before
530 * returning to userland.
531 */
532 get_mplock();
984263bc
MD
533 p = curproc;
534
08f2f1bb 535 lp = ONLY_LWP_IN_PROC(p);
553ea3c8 536
984263bc 537 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */
861905fb 538 mp = mountlist_boot_getfirst();
21739618 539 if (VFS_ROOT(mp, &vp))
984263bc 540 panic("cannot find root vnode");
28623bf9
MD
541 if (mp->mnt_ncmountpt.ncp == NULL) {
542 cache_allocroot(&mp->mnt_ncmountpt, mp, vp);
543 cache_unlock(&mp->mnt_ncmountpt); /* leave ref intact */
8c361dda 544 }
21739618 545 p->p_fd->fd_cdir = vp;
597aea93 546 vref(p->p_fd->fd_cdir);
21739618 547 p->p_fd->fd_rdir = vp;
597aea93 548 vref(p->p_fd->fd_rdir);
28623bf9 549 vfs_cache_setroot(vp, cache_hold(&mp->mnt_ncmountpt));
a11aaa81 550 vn_unlock(vp); /* leave ref intact */
28623bf9
MD
551 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_ncdir);
552 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_nrdir);
984263bc 553
cd29885a
MD
554 kprintf("Mounting devfs\n");
555 vfs_mountroot_devfs();
556
984263bc
MD
557 /*
558 * Need just enough stack to hold the faked-up "execve()" arguments.
559 */
560 addr = trunc_page(USRSTACK - PAGE_SIZE);
9388fcaa
MD
561 error = vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr,
562 PAGE_SIZE, PAGE_SIZE,
563 FALSE, VM_MAPTYPE_NORMAL,
1b874851
MD
564 VM_PROT_ALL, VM_PROT_ALL,
565 0);
566 if (error)
984263bc
MD
567 panic("init: couldn't allocate argument space");
568 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
569 p->p_vmspace->vm_ssize = 1;
570
bc01a404 571 if ((var = kgetenv("init_path")) != NULL) {
984263bc
MD
572 strncpy(init_path, var, sizeof init_path);
573 init_path[sizeof init_path - 1] = 0;
574 }
bc01a404 575 if ((var = kgetenv("kern.fallback_elf_brand")) != NULL)
984263bc
MD
576 fallback_elf_brand = strtol(var, NULL, 0);
577
578 for (path = init_path; *path != '\0'; path = next) {
579 while (*path == ':')
580 path++;
581 if (*path == '\0')
582 break;
583 for (next = path; *next != '\0' && *next != ':'; next++)
584 /* nothing */ ;
585 if (bootverbose)
6ea70f76 586 kprintf("start_init: trying %.*s\n", (int)(next - path),
984263bc
MD
587 path);
588
589 /*
590 * Move out the boot flag argument.
591 */
592 options = 0;
593 ucp = (char *)USRSTACK;
594 (void)subyte(--ucp, 0); /* trailing zero */
595 if (boothowto & RB_SINGLE) {
596 (void)subyte(--ucp, 's');
597 options = 1;
598 }
599#ifdef notyet
600 if (boothowto & RB_FASTBOOT) {
601 (void)subyte(--ucp, 'f');
602 options = 1;
603 }
604#endif
605
606#ifdef BOOTCDROM
607 (void)subyte(--ucp, 'C');
608 options = 1;
609#endif
610 if (options == 0)
611 (void)subyte(--ucp, '-');
612 (void)subyte(--ucp, '-'); /* leading hyphen */
613 arg1 = ucp;
614
615 /*
616 * Move out the file name (also arg 0).
617 */
618 (void)subyte(--ucp, 0);
619 for (s = next - 1; s >= path; s--)
620 (void)subyte(--ucp, *s);
621 arg0 = ucp;
622
623 /*
624 * Move out the arg pointers.
625 */
626 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
627 (void)suword((caddr_t)--uap, (long)0); /* terminator */
628 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
629 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
630
631 /*
632 * Point at the arguments.
633 */
634 args.fname = arg0;
635 args.argv = uap;
636 args.envv = NULL;
637
638 /*
639 * Now try to exec the program. If can't for any reason
640 * other than it doesn't exist, complain.
641 *
642 * Otherwise, return via fork_trampoline() all the way
643 * to user mode as init!
a2a5ad0d
MD
644 *
645 * WARNING! We may have been moved to another cpu after
0a3f9b47
MD
646 * acquiring the current user process designation. The
647 * MP lock will migrate with us though so we still have to
648 * release it.
984263bc 649 */
753fd850 650 if ((error = sys_execve(&args)) == 0) {
8a8d5d85 651 rel_mplock();
52eedfb5 652 lp->lwp_proc->p_usched->acquire_curproc(lp);
984263bc 653 return;
8a8d5d85 654 }
984263bc 655 if (error != ENOENT)
6ea70f76 656 kprintf("exec %.*s: error %d\n", (int)(next - path),
984263bc
MD
657 path, error);
658 }
6ea70f76 659 kprintf("init: not found in path %s\n", init_path);
984263bc
MD
660 panic("no init");
661}
662
663/*
664 * Like kthread_create(), but runs in it's own address space.
665 * We do this early to reserve pid 1.
666 *
667 * Note special case - do not make it runnable yet. Other work
668 * in progress will change this more.
669 */
670static void
671create_init(const void *udata __unused)
672{
673 int error;
bb3cd951 674 struct lwp *lp;
984263bc 675
e43a034f 676 crit_enter();
08f2f1bb 677 error = fork1(&lwp0, RFFDG | RFPROC, &initproc);
984263bc 678 if (error)
fc92d4aa 679 panic("cannot fork init: %d", error);
344ad853 680 initproc->p_flag |= P_SYSTEM;
08f2f1bb 681 lp = ONLY_LWP_IN_PROC(initproc);
bb3cd951 682 cpu_set_fork_handler(lp, start_init, NULL);
e43a034f 683 crit_exit();
984263bc 684}
ba39e2e0 685SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
984263bc
MD
686
687/*
688 * Make it runnable now.
689 */
690static void
691kick_init(const void *udata __unused)
692{
08f2f1bb 693 start_forked_proc(&lwp0, initproc);
984263bc 694}
ba39e2e0 695SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)
8ad65e08
MD
696
697/*
698 * Machine independant globaldata initialization
73e4f7b9
MD
699 *
700 * WARNING! Called from early boot, 'mycpu' may not work yet.
8ad65e08
MD
701 */
702void
d0e06f83 703mi_gdinit(struct globaldata *gd, int cpuid)
8ad65e08 704{
88c4d2f6 705 TAILQ_INIT(&gd->gd_systimerq);
e3161323 706 gd->gd_sysid_alloc = cpuid; /* prime low bits for cpu lookup */
d0e06f83 707 gd->gd_cpuid = cpuid;
da23a592 708 gd->gd_cpumask = CPUMASK(cpuid);
8ad65e08 709 lwkt_gdinit(gd);
41a01a4d 710 vm_map_entry_reserve_cpu_init(gd);
fc17ad60 711 sleep_gdinit(gd);
916e604f 712 usched_global_cpumask |= CPUMASK(cpuid);
8ad65e08
MD
713}
714