kernel - Use atomic op for usched_global_cpumask
[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 $
43 */
44
45#include "opt_init_path.h"
46
47#include <sys/param.h>
48#include <sys/file.h>
49#include <sys/filedesc.h>
50#include <sys/kernel.h>
51#include <sys/mount.h>
52#include <sys/sysctl.h>
53#include <sys/proc.h>
54#include <sys/resourcevar.h>
55#include <sys/signalvar.h>
56#include <sys/systm.h>
57#include <sys/vnode.h>
58#include <sys/sysent.h>
59#include <sys/reboot.h>
60#include <sys/sysproto.h>
61#include <sys/vmmeter.h>
62#include <sys/unistd.h>
63#include <sys/malloc.h>
0b692e79 64#include <sys/machintr.h>
684a93c4 65
58c2553a 66#include <sys/refcount.h>
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 101extern struct user *proc0paddr;
984263bc 102
984263bc 103int boothowto = 0; /* initialized so that it can be patched */
0c52fa62
SG
104SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0,
105 "Reboot flags, from console subsystem");
916e604f
MD
106SYSCTL_ULONG(_kern, OID_AUTO, usched_global_cpumask, CTLFLAG_RW,
107 &usched_global_cpumask, 0, "global user scheduler cpumask");
984263bc
MD
108
109/*
110 * This ensures that there is at least one entry so that the sysinit_set
7bd34050 111 * symbol is not undefined. A subsystem ID of SI_SPECIAL_DUMMY is never
984263bc
MD
112 * executed.
113 */
ba39e2e0 114SYSINIT(placeholder, SI_SPECIAL_DUMMY, SI_ORDER_ANY, NULL, NULL)
984263bc
MD
115
116/*
117 * The sysinit table itself. Items are checked off as the are run.
118 * If we want to register new sysinit types, add them to newsysinit.
119 */
dc62b251
MD
120SET_DECLARE(sysinit_set, struct sysinit);
121struct sysinit **sysinit, **sysinit_end;
122struct sysinit **newsysinit, **newsysinit_end;
123
984263bc
MD
124
125/*
126 * Merge a new sysinit set into the current set, reallocating it if
127 * necessary. This can only be called after malloc is running.
128 */
129void
dc62b251 130sysinit_add(struct sysinit **set, struct sysinit **set_end)
984263bc
MD
131{
132 struct sysinit **newset;
133 struct sysinit **sipp;
134 struct sysinit **xipp;
dc62b251 135 int count;
984263bc 136
dc62b251 137 count = set_end - set;
984263bc 138 if (newsysinit)
dc62b251 139 count += newsysinit_end - newsysinit;
984263bc 140 else
dc62b251 141 count += sysinit_end - sysinit;
efda3bd0 142 newset = kmalloc(count * sizeof(*sipp), M_TEMP, M_WAITOK);
984263bc 143 xipp = newset;
dc62b251
MD
144 if (newsysinit) {
145 for (sipp = newsysinit; sipp < newsysinit_end; sipp++)
984263bc 146 *xipp++ = *sipp;
dc62b251
MD
147 } else {
148 for (sipp = sysinit; sipp < sysinit_end; sipp++)
984263bc 149 *xipp++ = *sipp;
dc62b251
MD
150 }
151 for (sipp = set; sipp < set_end; sipp++)
984263bc 152 *xipp++ = *sipp;
984263bc 153 if (newsysinit)
efda3bd0 154 kfree(newsysinit, M_TEMP);
984263bc 155 newsysinit = newset;
dc62b251 156 newsysinit_end = newset + count;
984263bc
MD
157}
158
159/*
6bf59cd2
MD
160 * Callbacks from machine-dependant startup code (e.g. init386) to set
161 * up low level entities related to cpu #0's globaldata.
162 *
163 * Called from very low level boot code.
164 */
165void
166mi_proc0init(struct globaldata *gd, struct user *proc0paddr)
167{
fdce8919 168 lwkt_init_thread(&thread0, proc0paddr, LWKT_THREAD_STACK, 0, gd);
6bf59cd2 169 lwkt_set_comm(&thread0, "thread0");
3e291793 170 RB_INIT(&proc0.p_lwp_tree);
8f1f6170 171 spin_init(&proc0.p_spin);
5686ec5a 172 lwkt_token_init(&proc0.p_token, "iproc");
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;
e2b148c6 179 lwkt_token_init(&lwp0.lwp_token, "lwp_token");
94f98873 180 spin_init(&lwp0.lwp_spin);
6bf59cd2
MD
181 varsymset_init(&proc0.p_varsymset, NULL);
182 thread0.td_flags |= TDF_RUNNING;
183 thread0.td_proc = &proc0;
08f2f1bb 184 thread0.td_lwp = &lwp0;
6179d4f3 185 thread0.td_switch = cpu_lwkt_switch;
6b7dca13 186 lwkt_schedule_self(curthread);
6bf59cd2
MD
187}
188
189/*
984263bc
MD
190 * System startup; initialize the world, create process 0, mount root
191 * filesystem, and fork to create init and pagedaemon. Most of the
192 * hard work is done in the lower-level initialization routines including
193 * startup(), which does memory initialization and autoconfiguration.
194 *
195 * This allows simple addition of new kernel subsystems that require
196 * boot time initialization. It also allows substitution of subsystem
197 * (for instance, a scheduler, kernel profiler, or VM system) by object
198 * module. Finally, it allows for optional "kernel threads".
199 */
200void
201mi_startup(void)
202{
8a8d5d85 203 struct sysinit *sip; /* system initialization*/
17a9f566
MD
204 struct sysinit **sipp; /* system initialization*/
205 struct sysinit **xipp; /* interior loop of sort*/
206 struct sysinit *save; /* bubble*/
984263bc 207
dc62b251
MD
208 if (sysinit == NULL) {
209 sysinit = SET_BEGIN(sysinit_set);
ca97d438
JG
210#if defined(__amd64__) && defined(_KERNEL_VIRTUAL)
211 /*
212 * XXX For whatever reason, on 64-bit vkernels
213 * the value of sysinit obtained from the
214 * linker set is wrong.
215 */
216 if ((long)sysinit % 8 != 0) {
8608b858 217 kprintf("Fixing sysinit value...\n");
b12defdc 218 sysinit = (void *)((long)(intptr_t)sysinit + 4);
ca97d438
JG
219 }
220#endif
dc62b251
MD
221 sysinit_end = SET_LIMIT(sysinit_set);
222 }
ca97d438
JG
223#if defined(__amd64__) && defined(_KERNEL_VIRTUAL)
224 KKASSERT((long)sysinit % 8 == 0);
225#endif
dc62b251 226
984263bc
MD
227restart:
228 /*
229 * Perform a bubble sort of the system initialization objects by
230 * their subsystem (primary key) and order (secondary key).
231 */
dc62b251
MD
232 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
233 for (xipp = sipp + 1; xipp < sysinit_end; xipp++) {
984263bc
MD
234 if ((*sipp)->subsystem < (*xipp)->subsystem ||
235 ((*sipp)->subsystem == (*xipp)->subsystem &&
236 (*sipp)->order <= (*xipp)->order))
237 continue; /* skip*/
238 save = *sipp;
239 *sipp = *xipp;
240 *xipp = save;
241 }
242 }
243
244 /*
245 * Traverse the (now) ordered list of system initialization tasks.
246 * Perform each task, and continue on to the next task.
247 *
248 * The last item on the list is expected to be the scheduler,
249 * which will not return.
250 */
dc62b251
MD
251 for (sipp = sysinit; sipp < sysinit_end; sipp++) {
252 sip = *sipp;
ba39e2e0 253 if (sip->subsystem == SI_SPECIAL_DUMMY)
984263bc
MD
254 continue; /* skip dummy task(s)*/
255
ba39e2e0 256 if (sip->subsystem == SI_SPECIAL_DONE)
984263bc
MD
257 continue;
258
259 /* Call function */
8a8d5d85 260 (*(sip->func))(sip->udata);
984263bc
MD
261
262 /* Check off the one we're just done */
ba39e2e0 263 sip->subsystem = SI_SPECIAL_DONE;
984263bc
MD
264
265 /* Check if we've installed more sysinit items via KLD */
266 if (newsysinit != NULL) {
dc62b251 267 if (sysinit != SET_BEGIN(sysinit_set))
efda3bd0 268 kfree(sysinit, M_TEMP);
984263bc 269 sysinit = newsysinit;
dc62b251 270 sysinit_end = newsysinit_end;
984263bc 271 newsysinit = NULL;
dc62b251 272 newsysinit_end = NULL;
984263bc
MD
273 goto restart;
274 }
275 }
276
277 panic("Shouldn't get here!");
278 /* NOTREACHED*/
279}
280
281
282/*
283 ***************************************************************************
284 ****
285 **** The following SYSINIT's belong elsewhere, but have not yet
286 **** been moved.
287 ****
288 ***************************************************************************
289 */
290static void
b0dfaa52 291print_caddr_t(void *data)
984263bc 292{
6ea70f76 293 kprintf("%s", (char *)data);
984263bc 294}
ba39e2e0 295SYSINIT(announce, SI_BOOT1_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright)
984263bc 296
f8c3996b
MD
297/*
298 * Leave the critical section that protected us from spurious interrupts
299 * so device probes work.
300 */
301static void
302leavecrit(void *dummy __unused)
303{
7bf5fa56 304 MachIntrABI.stabilize();
0b692e79
MD
305 cpu_enable_intr();
306 MachIntrABI.cleanup();
f8c3996b
MD
307 crit_exit();
308 KKASSERT(!IN_CRITICAL_SECT(curthread));
dbcd0c9b 309
f8c3996b 310 if (bootverbose)
6ea70f76 311 kprintf("Leaving critical section, allowing interrupts\n");
f8c3996b 312}
ba39e2e0 313SYSINIT(leavecrit, SI_BOOT2_LEAVE_CRIT, SI_ORDER_ANY, leavecrit, NULL)
984263bc
MD
314
315/*
dbcd0c9b
MD
316 * This is called after the threading system is up and running,
317 * including the softclock, clock interrupts, and SMP.
318 */
319static void
320tsleepworks(void *dummy __unused)
321{
322 tsleep_now_works = 1;
323}
324SYSINIT(tsleepworks, SI_BOOT2_FINISH_SMP, SI_ORDER_SECOND, tsleepworks, NULL)
325
326/*
327 * This is called after devices have configured. Tell the kernel we are
328 * no longer in cold boot.
329 */
330static void
331endofcoldboot(void *dummy __unused)
332{
333 cold = 0;
334}
335SYSINIT(endofcoldboot, SI_SUB_ISWARM, SI_ORDER_ANY, endofcoldboot, NULL)
336
337/*
984263bc
MD
338 ***************************************************************************
339 ****
340 **** The two following SYSINT's are proc0 specific glue code. I am not
341 **** convinced that they can not be safely combined, but their order of
342 **** operation has been maintained as the same as the original init_main.c
343 **** for right now.
344 ****
345 **** These probably belong in init_proc.c or kern_proc.c, since they
346 **** deal with proc0 (the fork template process).
347 ****
348 ***************************************************************************
349 */
350/* ARGSUSED*/
351static void
352proc0_init(void *dummy __unused)
353{
0679adc4 354 struct proc *p;
81eea9f4 355 struct lwp *lp;
984263bc
MD
356
357 p = &proc0;
08f2f1bb 358 lp = &lwp0;
984263bc
MD
359
360 /*
315b8b8b
JM
361 * Initialize osrel
362 */
363 p->p_osrel = osreldate;
364
365 /*
984263bc
MD
366 * Initialize process and pgrp structures.
367 */
368 procinit();
369
370 /*
984263bc
MD
371 * additional VM structures
372 */
373 vm_init2();
374
375 /*
376 * Create process 0 (the swapper).
377 */
378 LIST_INSERT_HEAD(&allproc, p, p_list);
984263bc
MD
379 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
380 LIST_INIT(&pgrp0.pg_members);
58c2553a
MD
381 lwkt_token_init(&pgrp0.pg_token, "pgrp0");
382 refcount_init(&pgrp0.pg_refs, 1);
383 lockinit(&pgrp0.pg_lock, "pgwt0", 0, 0);
984263bc
MD
384 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
385
386 pgrp0.pg_session = &session0;
387 session0.s_count = 1;
388 session0.s_leader = p;
389
58c2553a
MD
390 pgref(&pgrp0);
391 p->p_pgrp = &pgrp0;
392
984263bc
MD
393 p->p_sysent = &aout_sysvec;
394
4643740a 395 p->p_flags = P_SYSTEM;
164b8401
SS
396 p->p_stat = SACTIVE;
397 lp->lwp_stat = LSRUN;
984263bc
MD
398 p->p_nice = NZERO;
399 p->p_rtprio.type = RTP_PRIO_NORMAL;
400 p->p_rtprio.prio = 0;
08f2f1bb 401 lp->lwp_rtprio = p->p_rtprio;
984263bc 402
d8061892 403 p->p_peers = NULL;
984263bc
MD
404 p->p_leader = p;
405
406 bcopy("swapper", p->p_comm, sizeof ("swapper"));
81eea9f4 407 bcopy("swapper", thread0.td_comm, sizeof ("swapper"));
984263bc
MD
408
409 /* Create credentials. */
984263bc 410 p->p_ucred = crget();
41c20dac 411 p->p_ucred->cr_ruidinfo = uifind(0);
984263bc
MD
412 p->p_ucred->cr_ngroups = 1; /* group 0 */
413 p->p_ucred->cr_uidinfo = uifind(0);
9910d07b 414 thread0.td_ucred = crhold(p->p_ucred); /* bootstrap fork1() */
984263bc
MD
415
416 /* Don't jail it */
41c20dac 417 p->p_ucred->cr_prison = NULL;
984263bc 418
b1b4e5a6
SS
419 /* Create sigacts. */
420 p->p_sigacts = &sigacts0;
6fa9e71a 421 refcount_init(&p->p_sigacts->ps_refcnt, 1);
984263bc
MD
422
423 /* Initialize signal state for process 0. */
c0b8a06d 424 siginit(p);
984263bc
MD
425
426 /* Create the file descriptor table. */
228b401d 427 fdinit_bootstrap(p, &filedesc0, cmask);
984263bc
MD
428
429 /* Create the limits structures. */
c0b8a06d 430 plimit_init0(&limit0);
984263bc 431 p->p_limit = &limit0;
984263bc
MD
432
433 /* Allocate a prototype map so we have something to fork. */
434 pmap_pinit0(vmspace_pmap(&vmspace0));
435 p->p_vmspace = &vmspace0;
287ebb09 436 lp->lwp_vmspace = p->p_vmspace;
e3161323 437 sysref_init(&vmspace0.vm_sysref, &vmspace_sysref_class);
e4846942
MD
438 vm_map_init(&vmspace0.vm_map,
439 round_page(VM_MIN_USER_ADDRESS),
440 trunc_page(VM_MAX_USER_ADDRESS),
441 vmspace_pmap(&vmspace0));
e3161323 442 sysref_activate(&vmspace0.vm_sysref);
984263bc 443
a591f597
MD
444 kqueue_init(&lwp0.lwp_kqueue, &filedesc0);
445
984263bc 446 /*
984263bc
MD
447 * Charge root for one process.
448 */
41c20dac 449 (void)chgproccnt(p->p_ucred->cr_uidinfo, 1, 0);
ba39e2e0 450 vm_init_limits(p);
984263bc 451}
ba39e2e0 452SYSINIT(p0init, SI_BOOT2_PROC0, SI_ORDER_FIRST, proc0_init, NULL)
984263bc 453
8fa76237
MD
454static int proc0_post_callback(struct proc *p, void *data __unused);
455
984263bc
MD
456/* ARGSUSED*/
457static void
458proc0_post(void *dummy __unused)
459{
460 struct timespec ts;
984263bc
MD
461
462 /*
463 * Now we can look at the time, having had a chance to verify the
464 * time from the file system. Pretend that proc0 started now.
465 */
8fa76237 466 allproc_scan(proc0_post_callback, NULL);
984263bc
MD
467
468 /*
469 * Give the ``random'' number generator a thump.
470 * XXX: Does read_random() contain enough bits to be used here ?
471 */
472 nanotime(&ts);
cddfb7bb 473 skrandom(ts.tv_sec ^ ts.tv_nsec);
984263bc 474}
8fa76237
MD
475
476static int
477proc0_post_callback(struct proc *p, void *data __unused)
478{
479 microtime(&p->p_start);
480 return(0);
481}
482
ba39e2e0 483SYSINIT(p0post, SI_SUB_PROC0_POST, SI_ORDER_FIRST, proc0_post, NULL)
984263bc
MD
484
485/*
486 ***************************************************************************
487 ****
488 **** The following SYSINIT's and glue code should be moved to the
489 **** respective files on a per subsystem basis.
490 ****
491 ***************************************************************************
492 */
493
494
495/*
496 ***************************************************************************
497 ****
498 **** The following code probably belongs in another file, like
499 **** kern/init_init.c.
500 ****
501 ***************************************************************************
502 */
503
504/*
505 * List of paths to try when searching for "init".
506 */
507static char init_path[MAXPATHLEN] =
508#ifdef INIT_PATH
509 __XSTRING(INIT_PATH);
510#else
22628b14 511 "/sbin/init:/sbin/oinit:/sbin/init.bak";
984263bc
MD
512#endif
513SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, "");
514
515/*
516 * Start the initial user process; try exec'ing each pathname in init_path.
517 * The program is invoked with one argument containing the boot flags.
518 */
519static void
91bd9c1e 520start_init(void *dummy, struct trapframe *frame)
984263bc
MD
521{
522 vm_offset_t addr;
523 struct execve_args args;
524 int options, error;
525 char *var, *path, *next, *s;
526 char *ucp, **uap, *arg0, *arg1;
527 struct proc *p;
553ea3c8 528 struct lwp *lp;
21739618
MD
529 struct mount *mp;
530 struct vnode *vp;
fc350ba1
MD
531 char *env;
532
533 /*
534 * This is passed in by the bootloader
535 */
536 env = kgetenv("kernelname");
537 if (env != NULL)
538 strlcpy(kernelname, env, sizeof(kernelname));
984263bc 539
2b0bd8aa
MD
540 /*
541 * The MP lock is not held on entry. We release it before
542 * returning to userland.
543 */
544 get_mplock();
984263bc
MD
545 p = curproc;
546
08f2f1bb 547 lp = ONLY_LWP_IN_PROC(p);
553ea3c8 548
984263bc 549 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */
861905fb 550 mp = mountlist_boot_getfirst();
21739618 551 if (VFS_ROOT(mp, &vp))
984263bc 552 panic("cannot find root vnode");
28623bf9
MD
553 if (mp->mnt_ncmountpt.ncp == NULL) {
554 cache_allocroot(&mp->mnt_ncmountpt, mp, vp);
555 cache_unlock(&mp->mnt_ncmountpt); /* leave ref intact */
8c361dda 556 }
21739618 557 p->p_fd->fd_cdir = vp;
597aea93 558 vref(p->p_fd->fd_cdir);
21739618 559 p->p_fd->fd_rdir = vp;
597aea93 560 vref(p->p_fd->fd_rdir);
28623bf9 561 vfs_cache_setroot(vp, cache_hold(&mp->mnt_ncmountpt));
a11aaa81 562 vn_unlock(vp); /* leave ref intact */
28623bf9
MD
563 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_ncdir);
564 cache_copy(&mp->mnt_ncmountpt, &p->p_fd->fd_nrdir);
984263bc 565
cd29885a
MD
566 kprintf("Mounting devfs\n");
567 vfs_mountroot_devfs();
568
984263bc
MD
569 /*
570 * Need just enough stack to hold the faked-up "execve()" arguments.
571 */
572 addr = trunc_page(USRSTACK - PAGE_SIZE);
9388fcaa
MD
573 error = vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr,
574 PAGE_SIZE, PAGE_SIZE,
575 FALSE, VM_MAPTYPE_NORMAL,
1b874851
MD
576 VM_PROT_ALL, VM_PROT_ALL,
577 0);
578 if (error)
984263bc
MD
579 panic("init: couldn't allocate argument space");
580 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
581 p->p_vmspace->vm_ssize = 1;
582
bc01a404 583 if ((var = kgetenv("init_path")) != NULL) {
984263bc
MD
584 strncpy(init_path, var, sizeof init_path);
585 init_path[sizeof init_path - 1] = 0;
586 }
315b8b8b 587
984263bc
MD
588 for (path = init_path; *path != '\0'; path = next) {
589 while (*path == ':')
590 path++;
591 if (*path == '\0')
592 break;
593 for (next = path; *next != '\0' && *next != ':'; next++)
594 /* nothing */ ;
595 if (bootverbose)
6ea70f76 596 kprintf("start_init: trying %.*s\n", (int)(next - path),
984263bc
MD
597 path);
598
599 /*
600 * Move out the boot flag argument.
601 */
602 options = 0;
603 ucp = (char *)USRSTACK;
604 (void)subyte(--ucp, 0); /* trailing zero */
605 if (boothowto & RB_SINGLE) {
606 (void)subyte(--ucp, 's');
607 options = 1;
608 }
609#ifdef notyet
610 if (boothowto & RB_FASTBOOT) {
611 (void)subyte(--ucp, 'f');
612 options = 1;
613 }
614#endif
615
616#ifdef BOOTCDROM
617 (void)subyte(--ucp, 'C');
618 options = 1;
619#endif
620 if (options == 0)
621 (void)subyte(--ucp, '-');
622 (void)subyte(--ucp, '-'); /* leading hyphen */
623 arg1 = ucp;
624
625 /*
626 * Move out the file name (also arg 0).
627 */
628 (void)subyte(--ucp, 0);
629 for (s = next - 1; s >= path; s--)
630 (void)subyte(--ucp, *s);
631 arg0 = ucp;
632
633 /*
634 * Move out the arg pointers.
635 */
636 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1));
637 (void)suword((caddr_t)--uap, (long)0); /* terminator */
638 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1);
639 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0);
640
641 /*
642 * Point at the arguments.
643 */
644 args.fname = arg0;
645 args.argv = uap;
646 args.envv = NULL;
647
648 /*
649 * Now try to exec the program. If can't for any reason
650 * other than it doesn't exist, complain.
651 *
652 * Otherwise, return via fork_trampoline() all the way
653 * to user mode as init!
a2a5ad0d
MD
654 *
655 * WARNING! We may have been moved to another cpu after
0a3f9b47
MD
656 * acquiring the current user process designation. The
657 * MP lock will migrate with us though so we still have to
658 * release it.
984263bc 659 */
753fd850 660 if ((error = sys_execve(&args)) == 0) {
8a8d5d85 661 rel_mplock();
52eedfb5 662 lp->lwp_proc->p_usched->acquire_curproc(lp);
984263bc 663 return;
8a8d5d85 664 }
984263bc 665 if (error != ENOENT)
6ea70f76 666 kprintf("exec %.*s: error %d\n", (int)(next - path),
984263bc
MD
667 path, error);
668 }
6ea70f76 669 kprintf("init: not found in path %s\n", init_path);
984263bc
MD
670 panic("no init");
671}
672
673/*
674 * Like kthread_create(), but runs in it's own address space.
675 * We do this early to reserve pid 1.
676 *
677 * Note special case - do not make it runnable yet. Other work
678 * in progress will change this more.
679 */
680static void
681create_init(const void *udata __unused)
682{
683 int error;
bb3cd951 684 struct lwp *lp;
984263bc 685
e43a034f 686 crit_enter();
08f2f1bb 687 error = fork1(&lwp0, RFFDG | RFPROC, &initproc);
984263bc 688 if (error)
fc92d4aa 689 panic("cannot fork init: %d", error);
4643740a 690 initproc->p_flags |= P_SYSTEM;
08f2f1bb 691 lp = ONLY_LWP_IN_PROC(initproc);
bb3cd951 692 cpu_set_fork_handler(lp, start_init, NULL);
e43a034f 693 crit_exit();
984263bc 694}
ba39e2e0 695SYSINIT(init, SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL)
984263bc
MD
696
697/*
698 * Make it runnable now.
699 */
700static void
701kick_init(const void *udata __unused)
702{
08f2f1bb 703 start_forked_proc(&lwp0, initproc);
984263bc 704}
ba39e2e0 705SYSINIT(kickinit, SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL)
8ad65e08
MD
706
707/*
708 * Machine independant globaldata initialization
73e4f7b9
MD
709 *
710 * WARNING! Called from early boot, 'mycpu' may not work yet.
8ad65e08
MD
711 */
712void
d0e06f83 713mi_gdinit(struct globaldata *gd, int cpuid)
8ad65e08 714{
88c4d2f6 715 TAILQ_INIT(&gd->gd_systimerq);
e3161323 716 gd->gd_sysid_alloc = cpuid; /* prime low bits for cpu lookup */
d0e06f83 717 gd->gd_cpuid = cpuid;
da23a592 718 gd->gd_cpumask = CPUMASK(cpuid);
8ad65e08 719 lwkt_gdinit(gd);
41a01a4d 720 vm_map_entry_reserve_cpu_init(gd);
fc17ad60 721 sleep_gdinit(gd);
cb31dff3 722 atomic_set_cpumask(&usched_global_cpumask, CPUMASK(cpuid));
8ad65e08
MD
723}
724