init_main.c revision 33108
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 * $Id: init_main.c,v 1.81 1998/01/30 11:32:52 phk Exp $ 43 */ 44 45#include "opt_devfs.h" 46#include "opt_diagnostic.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 65#include <machine/cpu.h> 66 67#include <vm/vm.h> 68#include <vm/vm_param.h> 69#include <vm/vm_prot.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 78extern void __main __P((void)); 79extern void main __P((void *framep)); 80 81/* Components of the first process -- never freed. */ 82static struct session session0; 83static struct pgrp pgrp0; 84struct proc proc0; 85static struct pcred cred0; 86static struct filedesc0 filedesc0; 87static struct plimit limit0; 88static struct vmspace vmspace0; 89#ifndef SMP /* per-cpu on smp */ 90struct proc *curproc = &proc0; 91#endif 92struct proc *initproc; 93 94int cmask = CMASK; 95extern struct user *proc0paddr; 96 97struct vnode *rootvp; 98int boothowto = 0; /* initialized so that it can be patched */ 99 100struct timeval boottime; 101SYSCTL_STRUCT(_kern, KERN_BOOTTIME, boottime, 102 CTLFLAG_RD, &boottime, timeval, ""); 103 104static int shutdowntimeout = 120; 105SYSCTL_INT(_kern, OID_AUTO, shutdown_timeout, 106 CTLFLAG_RW, &shutdowntimeout, 0, ""); 107 108#ifndef SMP /* per-cpu on smp */ 109struct timeval runtime; 110#endif 111 112/* 113 * Promiscuous argument pass for start_init() 114 * 115 * This is a kludge because we use a return from main() rather than a call 116 * to a new routine in locore.s to kick the kernel alive from locore.s. 117 */ 118static void *init_framep; 119 120 121#if __GNUC__ >= 2 122void __main() {} 123#endif 124 125 126/* 127 * This ensures that there is at least one entry so that the sysinit_set 128 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never 129 * executed. 130 */ 131SYSINIT(placeholder, SI_SUB_DUMMY,SI_ORDER_ANY, NULL, NULL) 132 133 134/* 135 * System startup; initialize the world, create process 0, mount root 136 * filesystem, and fork to create init and pagedaemon. Most of the 137 * hard work is done in the lower-level initialization routines including 138 * startup(), which does memory initialization and autoconfiguration. 139 * 140 * This allows simple addition of new kernel subsystems that require 141 * boot time initialization. It also allows substitution of subsystem 142 * (for instance, a scheduler, kernel profiler, or VM system) by object 143 * module. Finally, it allows for optional "kernel threads". 144 */ 145void 146main(framep) 147 void *framep; 148{ 149 150 register struct sysinit **sipp; /* system initialization*/ 151 register struct sysinit **xipp; /* interior loop of sort*/ 152 register struct sysinit *save; /* bubble*/ 153 154 /* 155 * Copy the locore.s frame pointer for proc0, this is forked into 156 * all other processes. 157 */ 158 init_framep = framep; 159 160#ifdef SMP 161 /* 162 * XXX curproc is per-cpu in SMP, and exists in freshly mapped pages, 163 * so its value must be initialized now before it is used by various 164 * initialization routines. proc0_init() isn't soon enough: 165 * among other things, configure() is called first, and may call 166 * setdumpdev(), which panics without a valid curproc. 167 */ 168 curproc = &proc0; 169#endif /* SMP */ 170 171 /* 172 * Perform a bubble sort of the system initialization objects by 173 * their subsystem (primary key) and order (secondary key). 174 * 175 * Since some things care about execution order, this is the 176 * operation which ensures continued function. 177 */ 178 for( sipp = (struct sysinit **)sysinit_set.ls_items; *sipp; sipp++) { 179 for( xipp = sipp + 1; *xipp; xipp++) { 180 if( (*sipp)->subsystem < (*xipp)->subsystem || 181 ( (*sipp)->subsystem == (*xipp)->subsystem && 182 (*sipp)->order < (*xipp)->order)) 183 continue; /* skip*/ 184 save = *sipp; 185 *sipp = *xipp; 186 *xipp = save; 187 } 188 } 189 190 /* 191 * Traverse the (now) ordered list of system initialization tasks. 192 * Perform each task, and continue on to the next task. 193 * 194 * The last item on the list is expected to be the scheduler, 195 * which will not return. 196 */ 197 for( sipp = (struct sysinit **)sysinit_set.ls_items; *sipp; sipp++) { 198 199 if( (*sipp)->subsystem == SI_SUB_DUMMY) 200 continue; /* skip dummy task(s)*/ 201 202 switch( (*sipp)->type) { 203 case SI_TYPE_DEFAULT: 204 /* no special processing*/ 205 (*((*sipp)->func))( (*sipp)->udata); 206 break; 207 208 case SI_TYPE_KTHREAD: 209#if !defined(SMP) 210 /* kernel thread*/ 211 if (fork1(&proc0, RFMEM|RFFDG|RFPROC)) 212 panic("fork kernel thread"); 213 cpu_set_fork_handler(pfind(proc0.p_retval[0]), 214 (*sipp)->func, (*sipp)->udata); 215 break; 216#endif 217 218 case SI_TYPE_KPROCESS: 219 if (fork1(&proc0, RFFDG|RFPROC)) 220 panic("fork kernel process"); 221 cpu_set_fork_handler(pfind(proc0.p_retval[0]), 222 (*sipp)->func, (*sipp)->udata); 223 break; 224 225 default: 226 panic( "init_main: unrecognized init type"); 227 } 228 } 229 230 panic("Shouldn't get here!"); 231 /* NOTREACHED*/ 232} 233 234 235/* 236 * Start a kernel process. This is called after a fork() call in 237 * main() in the file kern/init_main.c. 238 * 239 * This function is used to start "internal" daemons. 240 */ 241/* ARGSUSED*/ 242void 243kproc_start(udata) 244 void *udata; 245{ 246 struct kproc_desc *kp = udata; 247 struct proc *p = curproc; 248 249#ifdef DIAGNOSTIC 250 printf("Start pid=%d <%s>\n",p->p_pid, kp->arg0); 251#endif 252 253 /* save a global descriptor, if desired*/ 254 if( kp->global_procpp != NULL) 255 *kp->global_procpp = p; 256 257 /* this is a non-swapped system process*/ 258 p->p_flag |= P_INMEM | P_SYSTEM; 259 260 /* set up arg0 for 'ps', et al*/ 261 strcpy( p->p_comm, kp->arg0); 262 263 /* call the processes' main()...*/ 264 (*kp->func)(); 265 266 /* NOTREACHED */ 267 panic("kproc_start: %s", kp->arg0); 268} 269 270 271/* 272 *************************************************************************** 273 **** 274 **** The following SYSINIT's belong elsewhere, but have not yet 275 **** been moved. 276 **** 277 *************************************************************************** 278 */ 279#ifdef OMIT 280/* 281 * Handled by vfs_mountroot (bad idea) at this time... should be 282 * done the same as 4.4Lite2. 283 */ 284SYSINIT(swapinit, SI_SUB_SWAP, SI_ORDER_FIRST, swapinit, NULL) 285#endif /* OMIT*/ 286 287static void print_caddr_t __P((void *data)); 288static void 289print_caddr_t(data) 290 void *data; 291{ 292 printf("%s", (char *)data); 293} 294SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright) 295 296 297/* 298 *************************************************************************** 299 **** 300 **** The two following SYSINT's are proc0 specific glue code. I am not 301 **** convinced that they can not be safely combined, but their order of 302 **** operation has been maintained as the same as the original init_main.c 303 **** for right now. 304 **** 305 **** These probably belong in init_proc.c or kern_proc.c, since they 306 **** deal with proc0 (the fork template process). 307 **** 308 *************************************************************************** 309 */ 310/* ARGSUSED*/ 311static void proc0_init __P((void *dummy)); 312static void 313proc0_init(dummy) 314 void *dummy; 315{ 316 register struct proc *p; 317 register struct filedesc0 *fdp; 318 register unsigned i; 319 320 /* 321 * Initialize the current process pointer (curproc) before 322 * any possible traps/probes to simplify trap processing. 323 */ 324 p = &proc0; 325 curproc = p; /* XXX redundant*/ 326 327 /* 328 * Initialize process and pgrp structures. 329 */ 330 procinit(); 331 332 /* 333 * Initialize sleep queue hash table 334 */ 335 sleepinit(); 336 337 /* 338 * additional VM structures 339 */ 340 vm_init2(); 341 342 /* 343 * Create process 0 (the swapper). 344 */ 345 LIST_INSERT_HEAD(&allproc, p, p_list); 346 p->p_pgrp = &pgrp0; 347 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash); 348 LIST_INIT(&pgrp0.pg_members); 349 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist); 350 351 pgrp0.pg_session = &session0; 352 session0.s_count = 1; 353 session0.s_leader = p; 354 355 p->p_sysent = &aout_sysvec; 356 357 p->p_flag = P_INMEM | P_SYSTEM; 358 p->p_stat = SRUN; 359 p->p_nice = NZERO; 360 p->p_rtprio.type = RTP_PRIO_NORMAL; 361 p->p_rtprio.prio = 0; 362 363/* 364 * Link for kernel based threads 365 */ 366 p->p_peers = 0; 367 p->p_leader = p; 368 369 bcopy("swapper", p->p_comm, sizeof ("swapper")); 370 371 /* Create credentials. */ 372 cred0.p_refcnt = 1; 373 p->p_cred = &cred0; 374 p->p_ucred = crget(); 375 p->p_ucred->cr_ngroups = 1; /* group 0 */ 376 377 /* Create the file descriptor table. */ 378 fdp = &filedesc0; 379 p->p_fd = &fdp->fd_fd; 380 fdp->fd_fd.fd_refcnt = 1; 381 fdp->fd_fd.fd_cmask = cmask; 382 fdp->fd_fd.fd_ofiles = fdp->fd_dfiles; 383 fdp->fd_fd.fd_ofileflags = fdp->fd_dfileflags; 384 fdp->fd_fd.fd_nfiles = NDFILE; 385 386 /* Create the limits structures. */ 387 p->p_limit = &limit0; 388 for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++) 389 limit0.pl_rlimit[i].rlim_cur = 390 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY; 391 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur = 392 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles; 393 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur = 394 limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc; 395 i = ptoa(cnt.v_free_count); 396 limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i; 397 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i; 398 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3; 399 limit0.p_refcnt = 1; 400 401 /* Allocate a prototype map so we have something to fork. */ 402 p->p_vmspace = &vmspace0; 403 vmspace0.vm_refcnt = 1; 404 pmap_pinit0(&vmspace0.vm_pmap); 405 vm_map_init(&vmspace0.vm_map, round_page(VM_MIN_ADDRESS), 406 trunc_page(VM_MAXUSER_ADDRESS)); 407 vmspace0.vm_map.pmap = &vmspace0.vm_pmap; 408 p->p_addr = proc0paddr; /* XXX */ 409 410#define INCOMPAT_LITES2 411#ifdef INCOMPAT_LITES2 412 /* 413 * proc0 needs to have a coherent frame base in it's stack. 414 */ 415 cpu_set_init_frame(p, init_framep); /* XXX! */ 416#endif /* INCOMPAT_LITES2*/ 417 418 /* 419 * We continue to place resource usage info and signal 420 * actions in the user struct so they're pageable. 421 */ 422 p->p_stats = &p->p_addr->u_stats; 423 p->p_sigacts = &p->p_addr->u_sigacts; 424 425 /* 426 * Charge root for one process. 427 */ 428 (void)chgproccnt(0, 1); 429 430 /* 431 * Initialize the procfs flags (to 0, of course) 432 */ 433 p->p_stops = p->p_stype = p->p_step = 0; 434 435} 436SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL) 437 438/* ARGSUSED*/ 439static void proc0_post __P((void *dummy)); 440static void 441proc0_post(dummy) 442 void *dummy; 443{ 444 struct timeval tv; 445 446 /* 447 * Now can look at time, having had a chance to verify the time 448 * from the file system. Reset p->p_rtime as it may have been 449 * munched in mi_switch() after the time got set. 450 */ 451 gettime(&boottime); 452 proc0.p_stats->p_start = runtime = mono_time = boottime; 453 proc0.p_rtime.tv_sec = proc0.p_rtime.tv_usec = 0; 454 455 /* 456 * Give the ``random'' number generator a thump. 457 */ 458 microtime(&tv); 459 srandom(tv.tv_sec ^ tv.tv_usec); 460 461 /* Initialize signal state for process 0. */ 462 siginit(&proc0); 463} 464SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL) 465 466 467 468 469/* 470 *************************************************************************** 471 **** 472 **** The following SYSINIT's and glue code should be moved to the 473 **** respective files on a per subsystem basis. 474 **** 475 *************************************************************************** 476 */ 477 478/* ARGSUSED */ 479static void root_conf __P((void *dummy)); 480static void 481root_conf(dummy) 482 void *dummy; 483{ 484 cpu_rootconf(); 485} 486SYSINIT(root_conf, SI_SUB_ROOT_CONF, SI_ORDER_FIRST, root_conf, NULL) 487 488/* ARGSUSED*/ 489static void xxx_vfs_root_fdtab __P((void *dummy)); 490static void 491xxx_vfs_root_fdtab(dummy) 492 void *dummy; 493{ 494 register struct filedesc0 *fdp = &filedesc0; 495 496 /* Get the vnode for '/'. Set fdp->fd_fd.fd_cdir to reference it. */ 497 if (VFS_ROOT(mountlist.cqh_first, &rootvnode)) 498 panic("cannot find root vnode"); 499 fdp->fd_fd.fd_cdir = rootvnode; 500 VREF(fdp->fd_fd.fd_cdir); 501 VOP_UNLOCK(rootvnode, 0, &proc0); 502 fdp->fd_fd.fd_rdir = NULL; 503} 504SYSINIT(retrofit, SI_SUB_ROOT_FDTAB, SI_ORDER_FIRST, xxx_vfs_root_fdtab, NULL) 505 506 507/* 508 *************************************************************************** 509 **** 510 **** The following code probably belongs in another file, like 511 **** kern/init_init.c. It is here for two reasons only: 512 **** 513 **** 1) This code returns to startup the system; this is 514 **** abnormal for a kernel thread. 515 **** 2) This code promiscuously uses init_frame 516 **** 517 *************************************************************************** 518 */ 519 520static void kthread_init __P((void *dummy)); 521SYSINIT_KP(init,SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kthread_init, NULL) 522 523 524extern void prepare_usermode __P((void)); 525static void start_init __P((struct proc *p)); 526 527/* ARGSUSED*/ 528static void 529kthread_init(dummy) 530 void *dummy; 531{ 532 /* Create process 1 (init(8)). */ 533 start_init(curproc); 534 535 prepare_usermode(); 536 537 /* 538 * This returns to the fork trampoline, then to user mode. 539 */ 540 return; 541} 542 543 544/* 545 * List of paths to try when searching for "init". 546 */ 547static char *initpaths[] = { 548 "/sbin/init", 549 "/sbin/oinit", 550 "/sbin/init.bak", 551 "/stand/sysinstall", 552 NULL, 553}; 554 555/* 556 * Start the initial user process; try exec'ing each pathname in "initpaths". 557 * The program is invoked with one argument containing the boot flags. 558 */ 559static void 560start_init(p) 561 struct proc *p; 562{ 563 vm_offset_t addr; 564 struct execve_args args; 565 int options, i, error; 566 char **pathp, *path, *ucp, **uap, *arg0, *arg1; 567 568 initproc = p; 569 570 /* 571 * Need just enough stack to hold the faked-up "execve()" arguments. 572 */ 573 addr = trunc_page(VM_MAXUSER_ADDRESS - PAGE_SIZE); 574 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE, FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0) 575 panic("init: couldn't allocate argument space"); 576 p->p_vmspace->vm_maxsaddr = (caddr_t)addr; 577 p->p_vmspace->vm_ssize = 1; 578 579 for (pathp = &initpaths[0]; (path = *pathp) != NULL; pathp++) { 580 /* 581 * Move out the boot flag argument. 582 */ 583 options = 0; 584 ucp = (char *)USRSTACK; 585 (void)subyte(--ucp, 0); /* trailing zero */ 586 if (boothowto & RB_SINGLE) { 587 (void)subyte(--ucp, 's'); 588 options = 1; 589 } 590#ifdef notyet 591 if (boothowto & RB_FASTBOOT) { 592 (void)subyte(--ucp, 'f'); 593 options = 1; 594 } 595#endif 596 597#ifdef BOOTCDROM 598 (void)subyte(--ucp, 'C'); 599 options = 1; 600#endif 601 602#if defined(DEVFS) && defined(DEVFS_ROOT) 603 (void)subyte(--ucp, 'd'); 604 options = 1; 605#endif 606 if (options == 0) 607 (void)subyte(--ucp, '-'); 608 (void)subyte(--ucp, '-'); /* leading hyphen */ 609 arg1 = ucp; 610 611 /* 612 * Move out the file name (also arg 0). 613 */ 614 for (i = strlen(path) + 1; i >= 0; i--) 615 (void)subyte(--ucp, path[i]); 616 arg0 = ucp; 617 618 /* 619 * Move out the arg pointers. 620 */ 621 uap = (char **)((int)ucp & ~(NBPW-1)); 622 (void)suword((caddr_t)--uap, 0); /* terminator */ 623 (void)suword((caddr_t)--uap, (int)arg1); 624 (void)suword((caddr_t)--uap, (int)arg0); 625 626 /* 627 * Point at the arguments. 628 */ 629 args.fname = arg0; 630 args.argv = uap; 631 args.envv = NULL; 632 633 /* 634 * Now try to exec the program. If can't for any reason 635 * other than it doesn't exist, complain. 636 * 637 * Otherwise return to main() which returns to btext 638 * which completes the system startup. 639 */ 640 if ((error = execve(p, &args)) == 0) 641 return; 642 if (error != ENOENT) 643 printf("exec %s: error %d\n", path, error); 644 } 645 printf("init: not found\n"); 646 panic("no init"); 647} 648