init_main.c revision 1.146 1 /* $NetBSD: init_main.c,v 1.146 1999/04/12 00:22:08 gwr 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 8.16 (Berkeley) 5/14/95
42 */
43
44 #include "fs_nfs.h"
45 #include "opt_nfsserver.h"
46 #include "opt_sysv.h"
47
48 #include "rnd.h"
49
50 #include <sys/param.h>
51 #include <sys/filedesc.h>
52 #include <sys/file.h>
53 #include <sys/errno.h>
54 #include <sys/exec.h>
55 #include <sys/kernel.h>
56 #include <sys/mount.h>
57 #include <sys/map.h>
58 #include <sys/proc.h>
59 #include <sys/kthread.h>
60 #include <sys/resourcevar.h>
61 #include <sys/signalvar.h>
62 #include <sys/systm.h>
63 #include <sys/vnode.h>
64 #include <sys/tty.h>
65 #include <sys/conf.h>
66 #include <sys/disklabel.h>
67 #include <sys/buf.h>
68 #ifdef REAL_CLISTS
69 #include <sys/clist.h>
70 #endif
71 #include <sys/device.h>
72 #include <sys/socketvar.h>
73 #include <sys/protosw.h>
74 #include <sys/reboot.h>
75 #include <sys/user.h>
76 #ifdef SYSVSHM
77 #include <sys/shm.h>
78 #endif
79 #ifdef SYSVSEM
80 #include <sys/sem.h>
81 #endif
82 #ifdef SYSVMSG
83 #include <sys/msg.h>
84 #endif
85 #include <sys/domain.h>
86 #include <sys/mbuf.h>
87 #include <sys/namei.h>
88 #if NRND > 0
89 #include <sys/rnd.h>
90 #endif
91
92 #include <sys/syscall.h>
93 #include <sys/syscallargs.h>
94
95 #include <ufs/ufs/quota.h>
96
97 #include <machine/cpu.h>
98
99 #include <vm/vm.h>
100 #include <vm/vm_pageout.h>
101
102 #include <uvm/uvm.h>
103
104 #include <net/if.h>
105 #include <net/raw_cb.h>
106
107 char copyright[] = "\
108 Copyright (c) 1996, 1997, 1998, 1999
109 The NetBSD Foundation, Inc. All rights reserved.
110 Copyright (c) 1982, 1986, 1989, 1991, 1993
111 The Regents of the University of California. All rights reserved.
112
113 ";
114
115 /* Components of the first process -- never freed. */
116 struct session session0;
117 struct pgrp pgrp0;
118 struct proc proc0;
119 struct pcred cred0;
120 struct filedesc0 filedesc0;
121 struct plimit limit0;
122 struct vmspace vmspace0;
123 #ifndef curproc
124 struct proc *curproc = &proc0;
125 #endif
126 struct proc *initproc;
127
128 int cmask = CMASK;
129 extern struct user *proc0paddr;
130
131 struct vnode *rootvp, *swapdev_vp;
132 int boothowto;
133 struct timeval boottime;
134 struct timeval runtime;
135
136 static void check_console __P((struct proc *p));
137 static void start_init __P((void *));
138 static void start_pagedaemon __P((void *));
139 static void start_reaper __P((void *));
140 void main __P((void));
141
142 extern char sigcode[], esigcode[];
143 #ifdef SYSCALL_DEBUG
144 extern char *syscallnames[];
145 #endif
146
147 struct emul emul_netbsd = {
148 "netbsd",
149 NULL,
150 sendsig,
151 SYS_syscall,
152 SYS_MAXSYSCALL,
153 sysent,
154 #ifdef SYSCALL_DEBUG
155 syscallnames,
156 #else
157 NULL,
158 #endif
159 0,
160 copyargs,
161 setregs,
162 sigcode,
163 esigcode,
164 };
165
166 /*
167 * System startup; initialize the world, create process 0, mount root
168 * filesystem, and fork to create init and pagedaemon. Most of the
169 * hard work is done in the lower-level initialization routines including
170 * startup(), which does memory initialization and autoconfiguration.
171 */
172 void
173 main()
174 {
175 struct proc *p;
176 struct pdevinit *pdev;
177 int i, s, error;
178 extern struct pdevinit pdevinit[];
179 extern void roundrobin __P((void *));
180 extern void schedcpu __P((void *));
181 extern void disk_init __P((void));
182 #if defined(NFSSERVER) || defined(NFS)
183 extern void nfs_init __P((void));
184 #endif
185
186 /*
187 * Initialize the current process pointer (curproc) before
188 * any possible traps/probes to simplify trap processing.
189 */
190 p = &proc0;
191 curproc = p;
192 /*
193 * Attempt to find console and initialize
194 * in case of early panic or other messages.
195 */
196 consinit();
197 printf("%s", copyright);
198
199 uvm_init();
200
201 /* Do machine-dependent initialization. */
202 cpu_startup();
203
204 /*
205 * Initialize mbuf's. Do this now because we might attempt to
206 * allocate mbufs or mbuf clusters during autoconfiguration.
207 */
208 mbinit();
209
210 /* Initialize sockets. */
211 soinit();
212
213 disk_init(); /* must come before autoconfiguration */
214 tty_init(); /* initialise tty list */
215 #if NRND > 0
216 rnd_init();
217 #endif
218 config_init(); /* init autoconfiguration data structures */
219 configure(); /* ...and configure the hardware */
220
221 /*
222 * Initialize process and pgrp structures.
223 */
224 procinit();
225
226 /*
227 * Create process 0 (the swapper).
228 */
229 LIST_INSERT_HEAD(&allproc, p, p_list);
230 p->p_pgrp = &pgrp0;
231 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash);
232 LIST_INIT(&pgrp0.pg_members);
233 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist);
234
235 pgrp0.pg_session = &session0;
236 session0.s_count = 1;
237 session0.s_sid = p->p_pid;
238 session0.s_leader = p;
239
240 /*
241 * Set P_NOCLDWAIT so that kernel threads are reparented to
242 * init(8) when they exit. init(8) can easily wait them out
243 * for us.
244 */
245 p->p_flag = P_INMEM | P_SYSTEM | P_NOCLDWAIT;
246 p->p_stat = SRUN;
247 p->p_nice = NZERO;
248 p->p_emul = &emul_netbsd;
249 strncpy(p->p_comm, "swapper", MAXCOMLEN);
250
251 /* Create credentials. */
252 cred0.p_refcnt = 1;
253 p->p_cred = &cred0;
254 p->p_ucred = crget();
255 p->p_ucred->cr_ngroups = 1; /* group 0 */
256
257 /* Create the file descriptor table. */
258 finit();
259 p->p_fd = &filedesc0.fd_fd;
260 fdinit1(&filedesc0);
261
262 /* Create the limits structures. */
263 p->p_limit = &limit0;
264 for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++)
265 limit0.pl_rlimit[i].rlim_cur =
266 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
267
268 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles;
269 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur =
270 maxfiles < NOFILE ? maxfiles : NOFILE;
271
272 limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
273 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur =
274 maxproc < MAXUPRC ? maxproc : MAXUPRC;
275
276 i = ptoa(uvmexp.free);
277 limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i;
278 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i;
279 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3;
280 limit0.p_refcnt = 1;
281
282 /*
283 * Initialize proc0's vmspace, which uses the kernel pmap.
284 * All kernel processes (which never have user space mappings)
285 * share proc0's vmspace, and thus, the kernel pmap.
286 */
287 uvmspace_init(&vmspace0, pmap_kernel(), round_page(VM_MIN_ADDRESS),
288 trunc_page(VM_MAX_ADDRESS), TRUE);
289 p->p_vmspace = &vmspace0;
290
291 p->p_addr = proc0paddr; /* XXX */
292
293 /*
294 * We continue to place resource usage info and signal
295 * actions in the user struct so they're pageable.
296 */
297 p->p_stats = &p->p_addr->u_stats;
298 p->p_sigacts = &p->p_addr->u_sigacts;
299
300 /*
301 * Charge root for one process.
302 */
303 (void)chgproccnt(0, 1);
304
305 rqinit();
306
307 /* Configure virtual memory system, set vm rlimits. */
308 uvm_init_limits(p);
309
310 /* Initialize the file systems. */
311 #if defined(NFSSERVER) || defined(NFS)
312 nfs_init(); /* initialize server/shared data */
313 #endif
314 vfsinit();
315
316 /* Start real time and statistics clocks. */
317 initclocks();
318
319 #ifdef REAL_CLISTS
320 /* Initialize clists. */
321 clist_init();
322 #endif
323
324 #ifdef SYSVSHM
325 /* Initialize System V style shared memory. */
326 shminit();
327 #endif
328
329 #ifdef SYSVSEM
330 /* Initialize System V style semaphores. */
331 seminit();
332 #endif
333
334 #ifdef SYSVMSG
335 /* Initialize System V style message queues. */
336 msginit();
337 #endif
338
339 /* Attach pseudo-devices. */
340 for (pdev = pdevinit; pdev->pdev_attach != NULL; pdev++)
341 (*pdev->pdev_attach)(pdev->pdev_count);
342
343 /*
344 * Initialize protocols. Block reception of incoming packets
345 * until everything is ready.
346 */
347 s = splimp();
348 ifinit();
349 domaininit();
350 splx(s);
351
352 #ifdef GPROF
353 /* Initialize kernel profiling. */
354 kmstartup();
355 #endif
356
357 /* Kick off timeout driven events by calling first time. */
358 roundrobin(NULL);
359 schedcpu(NULL);
360
361 /* Determine the root and dump devices. */
362 cpu_rootconf();
363 cpu_dumpconf();
364
365 /* Mount the root file system. */
366 do {
367 domountroothook();
368 if ((error = vfs_mountroot())) {
369 printf("cannot mount root, error = %d\n", error);
370 boothowto |= RB_ASKNAME;
371 setroot(root_device,
372 (rootdev != NODEV) ? DISKPART(rootdev) : 0, NULL);
373 }
374 } while (error != 0);
375 mountroothook_destroy();
376
377 mountlist.cqh_first->mnt_flag |= MNT_ROOTFS;
378 mountlist.cqh_first->mnt_op->vfs_refcount++;
379
380 /*
381 * Get the vnode for '/'. Set filedesc0.fd_fd.fd_cdir to
382 * reference it.
383 */
384 if (VFS_ROOT(mountlist.cqh_first, &rootvnode))
385 panic("cannot find root vnode");
386 filedesc0.fd_fd.fd_cdir = rootvnode;
387 VREF(filedesc0.fd_fd.fd_cdir);
388 VOP_UNLOCK(rootvnode, 0);
389 filedesc0.fd_fd.fd_rdir = NULL;
390 uvm_swap_init();
391
392 /*
393 * Now can look at time, having had a chance to verify the time
394 * from the file system. Reset p->p_rtime as it may have been
395 * munched in mi_switch() after the time got set.
396 */
397 p->p_stats->p_start = runtime = mono_time = boottime = time;
398 p->p_rtime.tv_sec = p->p_rtime.tv_usec = 0;
399
400 /* Initialize signal state for process 0. */
401 siginit(p);
402
403 /* Create process 1 (init(8)). */
404 if (fork1(p, 0, NULL, &initproc))
405 panic("fork init");
406 cpu_set_kpc(initproc, start_init, initproc);
407
408 /* Create process 2, the pageout daemon kernel thread. */
409 if (kthread_create(start_pagedaemon, NULL, NULL, "pagedaemon"))
410 panic("fork pagedaemon");
411
412 /* Create process 3, the process reaper kernel thread. */
413 if (kthread_create(start_reaper, NULL, NULL, "reaper"))
414 panic("fork reaper");
415
416 /* Create any other deferred kernel threads. */
417 kthread_run_deferred_queue();
418
419 /* The scheduler is an infinite loop. */
420 uvm_scheduler();
421 /* NOTREACHED */
422 }
423
424 static void
425 check_console(p)
426 struct proc *p;
427 {
428 struct nameidata nd;
429 int error;
430
431 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, "/dev/console", p);
432 error = namei(&nd);
433 if (error == 0)
434 vrele(nd.ni_vp);
435 else if (error == ENOENT)
436 printf("warning: no /dev/console\n");
437 else
438 printf("warning: lookup /dev/console: error %d\n", error);
439 }
440
441 /*
442 * List of paths to try when searching for "init".
443 */
444 static char *initpaths[] = {
445 "/sbin/init",
446 "/sbin/oinit",
447 "/sbin/init.bak",
448 NULL,
449 };
450
451 /*
452 * Start the initial user process; try exec'ing each pathname in "initpaths".
453 * The program is invoked with one argument containing the boot flags.
454 */
455 static void
456 start_init(arg)
457 void *arg;
458 {
459 struct proc *p = arg;
460 vaddr_t addr;
461 struct sys_execve_args /* {
462 syscallarg(const char *) path;
463 syscallarg(char * const *) argp;
464 syscallarg(char * const *) envp;
465 } */ args;
466 int options, i, error;
467 register_t retval[2];
468 char flags[4], *flagsp;
469 char **pathp, *path, *slash, *ucp, **uap, *arg0, *arg1 = NULL;
470
471 /*
472 * Now in process 1.
473 */
474 strncpy(p->p_comm, "init", MAXCOMLEN);
475
476 /*
477 * This is not the right way to do this. We really should
478 * hand-craft a descriptor onto /dev/console to hand to init,
479 * but that's a _lot_ more work, and the benefit from this easy
480 * hack makes up for the "good is the enemy of the best" effect.
481 */
482 check_console(p);
483
484 /*
485 * Need just enough stack to hold the faked-up "execve()" arguments.
486 */
487 addr = USRSTACK - PAGE_SIZE;
488 if (uvm_map(&p->p_vmspace->vm_map, &addr, PAGE_SIZE,
489 NULL, UVM_UNKNOWN_OFFSET,
490 UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY,
491 UVM_ADV_NORMAL,
492 UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW))
493 != KERN_SUCCESS)
494 panic("init: couldn't allocate argument space");
495 p->p_vmspace->vm_maxsaddr = (caddr_t)addr;
496
497 for (pathp = &initpaths[0]; (path = *pathp) != NULL; pathp++) {
498 ucp = (char *)(addr + PAGE_SIZE);
499
500 /*
501 * Construct the boot flag argument.
502 */
503 flagsp = flags;
504 *flagsp++ = '-';
505 options = 0;
506
507 if (boothowto & RB_SINGLE) {
508 *flagsp++ = 's';
509 options = 1;
510 }
511 #ifdef notyet
512 if (boothowto & RB_FASTBOOT) {
513 *flagsp++ = 'f';
514 options = 1;
515 }
516 #endif
517
518 /*
519 * Move out the flags (arg 1), if necessary.
520 */
521 if (options != 0) {
522 *flagsp++ = '\0';
523 i = flagsp - flags;
524 #ifdef DEBUG
525 printf("init: copying out flags `%s' %d\n", flags, i);
526 #endif
527 (void)copyout((caddr_t)flags, (caddr_t)(ucp -= i), i);
528 arg1 = ucp;
529 }
530
531 /*
532 * Move out the file name (also arg 0).
533 */
534 i = strlen(path) + 1;
535 #ifdef DEBUG
536 printf("init: copying out path `%s' %d\n", path, i);
537 #endif
538 (void)copyout((caddr_t)path, (caddr_t)(ucp -= i), i);
539 arg0 = ucp;
540
541 /*
542 * Move out the arg pointers.
543 */
544 uap = (char **)((long)ucp & ~ALIGNBYTES);
545 (void)suword((caddr_t)--uap, 0); /* terminator */
546 if (options != 0)
547 (void)suword((caddr_t)--uap, (long)arg1);
548 slash = strrchr(path, '/');
549 if (slash)
550 (void)suword((caddr_t)--uap,
551 (long)arg0 + (slash + 1 - path));
552 else
553 (void)suword((caddr_t)--uap, (long)arg0);
554
555 /*
556 * Point at the arguments.
557 */
558 SCARG(&args, path) = arg0;
559 SCARG(&args, argp) = uap;
560 SCARG(&args, envp) = NULL;
561
562 /*
563 * Now try to exec the program. If can't for any reason
564 * other than it doesn't exist, complain.
565 */
566 error = sys_execve(p, &args, retval);
567 if (error == 0 || error == EJUSTRETURN)
568 return;
569 if (error != ENOENT)
570 printf("exec %s: error %d\n", path, error);
571 }
572 printf("init: not found\n");
573 panic("no init");
574 }
575
576 /* ARGSUSED */
577 static void
578 start_pagedaemon(arg)
579 void *arg;
580 {
581
582 uvm_pageout();
583 /* NOTREACHED */
584 }
585
586 /* ARGSUSED */
587 static void
588 start_reaper(arg)
589 void *arg;
590 {
591
592 reaper();
593 /* NOTREACHED */
594 }
595