init_main.c revision 1.116 1 /* $NetBSD: init_main.c,v 1.116 1998/02/14 00:37:30 thorpej Exp $ */
2
3 /*
4 * Copyright (c) 1995 Christopher G. Demetriou. All rights reserved.
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 */
43
44 #include "opt_uvm.h"
45
46 #include "rnd.h"
47
48 #include <sys/param.h>
49 #include <sys/filedesc.h>
50 #include <sys/errno.h>
51 #include <sys/exec.h>
52 #include <sys/kernel.h>
53 #include <sys/mount.h>
54 #include <sys/map.h>
55 #include <sys/proc.h>
56 #include <sys/resourcevar.h>
57 #include <sys/signalvar.h>
58 #include <sys/systm.h>
59 #include <sys/vnode.h>
60 #include <sys/tty.h>
61 #include <sys/conf.h>
62 #include <sys/disklabel.h>
63 #include <sys/buf.h>
64 #ifdef REAL_CLISTS
65 #include <sys/clist.h>
66 #endif
67 #include <sys/device.h>
68 #include <sys/protosw.h>
69 #include <sys/reboot.h>
70 #include <sys/user.h>
71 #ifdef SYSVSHM
72 #include <sys/shm.h>
73 #endif
74 #ifdef SYSVSEM
75 #include <sys/sem.h>
76 #endif
77 #ifdef SYSVMSG
78 #include <sys/msg.h>
79 #endif
80 #include <sys/domain.h>
81 #include <sys/mbuf.h>
82 #include <sys/namei.h>
83 #if NRND > 0
84 #include <sys/rnd.h>
85 #endif
86
87 #include <sys/syscall.h>
88 #include <sys/syscallargs.h>
89
90 #include <ufs/ufs/quota.h>
91
92 #include <machine/cpu.h>
93
94 #include <vm/vm.h>
95 #include <vm/vm_pageout.h>
96
97 #if defined(UVM)
98 #include <uvm/uvm.h>
99 #endif
100
101 #include <net/if.h>
102 #include <net/raw_cb.h>
103
104 char copyright[] = "\
105 Copyright (c) 1996, 1997, 1998
106 The NetBSD Foundation, Inc. All rights reserved.
107 Copyright (c) 1982, 1986, 1989, 1991, 1993
108 The Regents of the University of California. All rights reserved.
109
110 ";
111
112 /* Components of the first process -- never freed. */
113 struct session session0;
114 struct pgrp pgrp0;
115 struct proc proc0;
116 struct pcred cred0;
117 struct filedesc0 filedesc0;
118 struct plimit limit0;
119 struct vmspace vmspace0;
120 struct proc *curproc = &proc0;
121 struct proc *initproc;
122
123 int cmask = CMASK;
124 extern struct user *proc0paddr;
125
126 struct vnode *rootvp, *swapdev_vp;
127 int boothowto;
128 struct timeval boottime;
129 struct timeval runtime;
130
131 static void check_console __P((struct proc *p));
132 static void start_init __P((struct proc *));
133 static void start_pagedaemon __P((struct proc *));
134 void main __P((void *));
135
136 extern char sigcode[], esigcode[];
137 #ifdef SYSCALL_DEBUG
138 extern char *syscallnames[];
139 #endif
140
141 struct emul emul_netbsd = {
142 "netbsd",
143 NULL,
144 sendsig,
145 SYS_syscall,
146 SYS_MAXSYSCALL,
147 sysent,
148 #ifdef SYSCALL_DEBUG
149 syscallnames,
150 #else
151 NULL,
152 #endif
153 0,
154 copyargs,
155 setregs,
156 sigcode,
157 esigcode,
158 };
159
160 /*
161 * System startup; initialize the world, create process 0, mount root
162 * filesystem, and fork to create init and pagedaemon. Most of the
163 * hard work is done in the lower-level initialization routines including
164 * startup(), which does memory initialization and autoconfiguration.
165 */
166 void
167 main(framep)
168 void *framep; /* XXX should go away */
169 {
170 struct proc *p, *p2;
171 struct pdevinit *pdev;
172 int i, s, error;
173 extern struct pdevinit pdevinit[];
174 extern void roundrobin __P((void *));
175 extern void schedcpu __P((void *));
176 extern void disk_init __P((void));
177 #if defined(NFSSERVER) || defined(NFS)
178 extern void nfs_init __P((void));
179 #endif
180
181 /*
182 * Initialize the current process pointer (curproc) before
183 * any possible traps/probes to simplify trap processing.
184 */
185 p = &proc0;
186 curproc = p;
187 /*
188 * Attempt to find console and initialize
189 * in case of early panic or other messages.
190 */
191 consinit();
192 printf(copyright);
193
194 #if defined(UVM)
195 uvm_init();
196 #else
197 vm_mem_init();
198 kmeminit();
199 #if defined(MACHINE_NEW_NONCONTIG)
200 vm_page_physrehash();
201 #endif
202 #endif /* UVM */
203 disk_init(); /* must come before autoconfiguration */
204 tty_init(); /* initialise tty list */
205 #if NRND > 0
206 rnd_init();
207 #endif
208 config_init(); /* init autoconfiguration data structures */
209 cpu_startup();
210
211 /*
212 * Initialize process and pgrp structures.
213 */
214 procinit();
215
216 /*
217 * Create process 0 (the swapper).
218 */
219 LIST_INSERT_HEAD(&allproc, p, p_list);
220 p->p_pgrp = &pgrp0;
221 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
222 LIST_INIT(&pgrp0.pg_members);
223 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
224
225 pgrp0.pg_session = &session0;
226 session0.s_count = 1;
227 session0.s_sid = p->p_pid;
228 session0.s_leader = p;
229
230 p->p_flag = P_INMEM | P_SYSTEM;
231 p->p_stat = SRUN;
232 p->p_nice = NZERO;
233 p->p_emul = &emul_netbsd;
234 bcopy("swapper", p->p_comm, sizeof ("swapper"));
235
236 /* Create credentials. */
237 cred0.p_refcnt = 1;
238 p->p_cred = &cred0;
239 p->p_ucred = crget();
240 p->p_ucred->cr_ngroups = 1; /* group 0 */
241
242 /* Create the file descriptor table. */
243 p->p_fd = &filedesc0.fd_fd;
244 fdinit1(&filedesc0);
245
246 /* Create the limits structures. */
247 p->p_limit = &limit0;
248 for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++)
249 limit0.pl_rlimit[i].rlim_cur =
250 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
251 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur = NOFILE;
252 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur = MAXUPRC;
253 #if defined(UVM)
254 i = ptoa(uvmexp.free);
255 #else
256 i = ptoa(cnt.v_free_count);
257 #endif
258 limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i;
259 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
260 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
261 limit0.p_refcnt = 1;
262
263 /* Allocate a prototype map so we have something to fork. */
264 #if defined(UVM)
265 p->p_vmspace = uvmspace_alloc(round_page(VM_MIN_ADDRESS),
266 trunc_page(VM_MAX_ADDRESS), TRUE);
267 #else
268 p->p_vmspace = vmspace_alloc(round_page(VM_MIN_ADDRESS),
269 trunc_page(VM_MAX_ADDRESS), TRUE);
270 #endif
271
272 p->p_addr = proc0paddr; /* XXX */
273
274 /*
275 * We continue to place resource usage info and signal
276 * actions in the user struct so they're pageable.
277 */
278 p->p_stats = &p->p_addr->u_stats;
279 p->p_sigacts = &p->p_addr->u_sigacts;
280
281 /*
282 * Charge root for one process.
283 */
284 (void)chgproccnt(0, 1);
285
286 rqinit();
287
288 /* Configure virtual memory system, set vm rlimits. */
289 #if defined(UVM)
290 uvm_init_limits(p);
291 #else
292 vm_init_limits(p);
293 #endif
294
295 /* Initialize the file systems. */
296 #if defined(NFSSERVER) || defined(NFS)
297 nfs_init(); /* initialize server/shared data */
298 #endif
299 vfsinit();
300
301 /* Start real time and statistics clocks. */
302 initclocks();
303
304 /* Initialize mbuf's. */
305 mbinit();
306
307 #ifdef REAL_CLISTS
308 /* Initialize clists. */
309 clist_init();
310 #endif
311
312 #ifdef SYSVSHM
313 /* Initialize System V style shared memory. */
314 shminit();
315 #endif
316
317 #ifdef SYSVSEM
318 /* Initialize System V style semaphores. */
319 seminit();
320 #endif
321
322 #ifdef SYSVMSG
323 /* Initialize System V style message queues. */
324 msginit();
325 #endif
326
327 /* Attach pseudo-devices. */
328 for (pdev = pdevinit; pdev->pdev_attach != NULL; pdev++)
329 (*pdev->pdev_attach)(pdev->pdev_count);
330
331 /*
332 * Initialize protocols. Block reception of incoming packets
333 * until everything is ready.
334 */
335 s = splimp();
336 ifinit();
337 domaininit();
338 splx(s);
339
340 #ifdef GPROF
341 /* Initialize kernel profiling. */
342 kmstartup();
343 #endif
344
345 /* Kick off timeout driven events by calling first time. */
346 roundrobin(NULL);
347 schedcpu(NULL);
348
349 /* Determine the root and dump devices. */
350 cpu_rootconf();
351 cpu_dumpconf();
352
353 /* Mount the root file system. */
354 do {
355 domountroothook();
356 if ((error = vfs_mountroot())) {
357 printf("cannot mount root, error = %d\n", error);
358 boothowto |= RB_ASKNAME;
359 setroot(root_device,
360 (rootdev != NODEV) ? DISKPART(rootdev) : 0, NULL);
361 }
362 } while (error != 0);
363 mountroothook_destroy();
364
365 mountlist.cqh_first->mnt_flag |= MNT_ROOTFS;
366 mountlist.cqh_first->mnt_op->vfs_refcount++;
367
368 /*
369 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to
370 * reference it.
371 */
372 if (VFS_ROOT(mountlist.cqh_first, &rootvnode))
373 panic("cannot find root vnode");
374 filedesc0.fd_fd.fd_cdir = rootvnode;
375 VREF(filedesc0.fd_fd.fd_cdir);
376 VOP_UNLOCK(rootvnode);
377 filedesc0.fd_fd.fd_rdir = NULL;
378 #if defined(UVM)
379 uvm_swap_init();
380 #else
381 swapinit();
382 #endif
383
384 /*
385 * Now can look at time, having had a chance to verify the time
386 * from the file system. Reset p->p_rtime as it may have been
387 * munched in mi_switch() after the time got set.
388 */
389 p->p_stats->p_start = runtime = mono_time = boottime = time;
390 p->p_rtime.tv_sec = p->p_rtime.tv_usec = 0;
391
392 /* Initialize signal state for process 0. */
393 siginit(p);
394
395 /* Create process 1 (init(8)). */
396 if (fork1(p, 0, NULL, &p2))
397 panic("fork init");
398 cpu_set_kpc(p2, start_init);
399
400 /* Create process 2 (the pageout daemon). */
401 if (fork1(p, 0, NULL, &p2))
402 panic("fork pager");
403 cpu_set_kpc(p2, start_pagedaemon);
404
405 /* The scheduler is an infinite loop. */
406 #if defined(UVM)
407 uvm_scheduler();
408 #else
409 scheduler();
410 #endif
411 /* NOTREACHED */
412 }
413
414 static void
415 check_console(p)
416 struct proc *p;
417 {
418 struct nameidata nd;
419 int error;
420
421 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console", p);
422 error = namei(&nd);
423 if (error == 0)
424 vrele(nd.ni_vp);
425 else if (error == ENOENT)
426 printf("warning: no /dev/console\n");
427 else
428 printf("warning: lookup /dev/console: error %d\n", error);
429 }
430
431 /*
432 * List of paths to try when searching for "init".
433 */
434 static char *initpaths[] = {
435 "/sbin/init",
436 "/sbin/oinit",
437 "/sbin/init.bak",
438 NULL,
439 };
440
441 /*
442 * Start the initial user process; try exec'ing each pathname in "initpaths".
443 * The program is invoked with one argument containing the boot flags.
444 */
445 static void
446 start_init(p)
447 struct proc *p;
448 {
449 vm_offset_t addr;
450 struct sys_execve_args /* {
451 syscallarg(const char *) path;
452 syscallarg(char * const *) argp;
453 syscallarg(char * const *) envp;
454 } */ args;
455 int options, i, error;
456 register_t retval[2];
457 char flags[4], *flagsp;
458 char **pathp, *path, *ucp, **uap, *arg0, *arg1 = NULL;
459
460 /*
461 * Now in process 1.
462 */
463 initproc = p;
464
465 /*
466 * This is not the right way to do this. We really should
467 * hand-craft a descriptor onto /dev/console to hand to init,
468 * but that's a _lot_ more work, and the benefit from this easy
469 * hack makes up for the "good is the enemy of the best" effect.
470 */
471 check_console(p);
472
473 /*
474 * Need just enough stack to hold the faked-up "execve()" arguments.
475 */
476 addr = USRSTACK - PAGE_SIZE;
477 #if defined(UVM)
478 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE,
479 NULL, UVM_UNKNOWN_OFFSET,
480 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY,
481 UVM_ADV_NORMAL,
482 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW))
483 != KERN_SUCCESS)
484 panic("init: couldn't allocate argument space");
485 #else
486 if (vm_allocate(&p->p_vmspace->vm_map, &addr, (vm_size_t)PAGE_SIZE,
487 FALSE) != 0)
488 panic("init: couldn't allocate argument space");
489 #endif
490 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
491
492 for (pathp = &initpaths[0]; (path = *pathp) != NULL; pathp++) {
493 ucp = (char *)(addr + PAGE_SIZE);
494
495 /*
496 * Construct the boot flag argument.
497 */
498 flagsp = flags;
499 *flagsp++ = '-';
500 options = 0;
501
502 if (boothowto & RB_SINGLE) {
503 *flagsp++ = 's';
504 options = 1;
505 }
506 #ifdef notyet
507 if (boothowto & RB_FASTBOOT) {
508 *flagsp++ = 'f';
509 options = 1;
510 }
511 #endif
512
513 /*
514 * Move out the flags (arg 1), if necessary.
515 */
516 if (options != 0) {
517 *flagsp++ = '\0';
518 i = flagsp - flags;
519 #ifdef DEBUG
520 printf("init: copying out flags `%s' %d\n", flags, i);
521 #endif
522 (void)copyout((caddr_t)flags, (caddr_t)(ucp -= i), i);
523 arg1 = ucp;
524 }
525
526 /*
527 * Move out the file name (also arg 0).
528 */
529 i = strlen(path) + 1;
530 #ifdef DEBUG
531 printf("init: copying out path `%s' %d\n", path, i);
532 #endif
533 (void)copyout((caddr_t)path, (caddr_t)(ucp -= i), i);
534 arg0 = ucp;
535
536 /*
537 * Move out the arg pointers.
538 */
539 uap = (char **)((long)ucp & ~ALIGNBYTES);
540 (void)suword((caddr_t)--uap, 0); /* terminator */
541 if (options != 0)
542 (void)suword((caddr_t)--uap, (long)arg1);
543 (void)suword((caddr_t)--uap, (long)arg0);
544
545 /*
546 * Point at the arguments.
547 */
548 SCARG(&args, path) = arg0;
549 SCARG(&args, argp) = uap;
550 SCARG(&args, envp) = NULL;
551
552 /*
553 * Now try to exec the program. If can't for any reason
554 * other than it doesn't exist, complain.
555 */
556 error = sys_execve(p, &args, retval);
557 if (error == 0 || error == EJUSTRETURN)
558 return;
559 if (error != ENOENT)
560 printf("exec %s: error %d\n", path, error);
561 }
562 printf("init: not found\n");
563 panic("no init");
564 }
565
566 static void
567 start_pagedaemon(p)
568 struct proc *p;
569 {
570
571 /*
572 * Now in process 2.
573 */
574 p->p_flag |= P_INMEM | P_SYSTEM; /* XXX */
575 bcopy("pagedaemon", curproc->p_comm, sizeof ("pagedaemon"));
576 #if defined(UVM)
577 uvm_pageout();
578 #else
579 vm_pageout();
580 #endif
581 /* NOTREACHED */
582 }
583