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 * $DragonFly: src/sys/kern/init_main.c,v 1.2 2003/06/17 04:28:41 dillon Exp $ 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> 65 66 #include <machine/cpu.h> 67 68 #include <vm/vm.h> 69 #include <vm/vm_param.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 76 extern struct linker_set sysinit_set; /* XXX */ 77 78 void mi_startup(void); /* Should be elsewhere */ 79 80 /* Components of the first process -- never freed. */ 81 static struct session session0; 82 static struct pgrp pgrp0; 83 struct proc proc0; 84 static struct pcred cred0; 85 static struct procsig procsig0; 86 static struct filedesc0 filedesc0; 87 static struct plimit limit0; 88 static struct vmspace vmspace0; 89 struct proc *initproc; 90 91 int cmask = CMASK; 92 extern struct user *proc0paddr; 93 extern int fallback_elf_brand; 94 95 struct vnode *rootvp; 96 int boothowto = 0; /* initialized so that it can be patched */ 97 SYSCTL_INT(_debug, OID_AUTO, boothowto, CTLFLAG_RD, &boothowto, 0, ""); 98 99 /* 100 * This ensures that there is at least one entry so that the sysinit_set 101 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never 102 * executed. 103 */ 104 SYSINIT(placeholder, SI_SUB_DUMMY, SI_ORDER_ANY, NULL, NULL) 105 106 /* 107 * The sysinit table itself. Items are checked off as the are run. 108 * If we want to register new sysinit types, add them to newsysinit. 109 */ 110 struct sysinit **sysinit = (struct sysinit **)sysinit_set.ls_items; 111 struct sysinit **newsysinit; 112 113 /* 114 * Merge a new sysinit set into the current set, reallocating it if 115 * necessary. This can only be called after malloc is running. 116 */ 117 void 118 sysinit_add(struct sysinit **set) 119 { 120 struct sysinit **newset; 121 struct sysinit **sipp; 122 struct sysinit **xipp; 123 int count = 0; 124 125 if (newsysinit) 126 for (sipp = newsysinit; *sipp; sipp++) 127 count++; 128 else 129 for (sipp = sysinit; *sipp; sipp++) 130 count++; 131 for (sipp = set; *sipp; sipp++) 132 count++; 133 count++; /* Trailing NULL */ 134 newset = malloc(count * sizeof(*sipp), M_TEMP, M_NOWAIT); 135 if (newset == NULL) 136 panic("cannot malloc for sysinit"); 137 xipp = newset; 138 if (newsysinit) 139 for (sipp = newsysinit; *sipp; sipp++) 140 *xipp++ = *sipp; 141 else 142 for (sipp = sysinit; *sipp; sipp++) 143 *xipp++ = *sipp; 144 for (sipp = set; *sipp; sipp++) 145 *xipp++ = *sipp; 146 *xipp = NULL; 147 if (newsysinit) 148 free(newsysinit, M_TEMP); 149 newsysinit = newset; 150 } 151 152 /* 153 * System startup; initialize the world, create process 0, mount root 154 * filesystem, and fork to create init and pagedaemon. Most of the 155 * hard work is done in the lower-level initialization routines including 156 * startup(), which does memory initialization and autoconfiguration. 157 * 158 * This allows simple addition of new kernel subsystems that require 159 * boot time initialization. It also allows substitution of subsystem 160 * (for instance, a scheduler, kernel profiler, or VM system) by object 161 * module. Finally, it allows for optional "kernel threads". 162 */ 163 void 164 mi_startup(void) 165 { 166 167 register struct sysinit **sipp; /* system initialization*/ 168 register struct sysinit **xipp; /* interior loop of sort*/ 169 register struct sysinit *save; /* bubble*/ 170 171 restart: 172 /* 173 * Perform a bubble sort of the system initialization objects by 174 * their subsystem (primary key) and order (secondary key). 175 */ 176 for (sipp = sysinit; *sipp; sipp++) { 177 for (xipp = sipp + 1; *xipp; xipp++) { 178 if ((*sipp)->subsystem < (*xipp)->subsystem || 179 ((*sipp)->subsystem == (*xipp)->subsystem && 180 (*sipp)->order <= (*xipp)->order)) 181 continue; /* skip*/ 182 save = *sipp; 183 *sipp = *xipp; 184 *xipp = save; 185 } 186 } 187 188 /* 189 * Traverse the (now) ordered list of system initialization tasks. 190 * Perform each task, and continue on to the next task. 191 * 192 * The last item on the list is expected to be the scheduler, 193 * which will not return. 194 */ 195 for (sipp = sysinit; *sipp; sipp++) { 196 197 if ((*sipp)->subsystem == SI_SUB_DUMMY) 198 continue; /* skip dummy task(s)*/ 199 200 if ((*sipp)->subsystem == SI_SUB_DONE) 201 continue; 202 203 /* Call function */ 204 (*((*sipp)->func))((*sipp)->udata); 205 206 /* Check off the one we're just done */ 207 (*sipp)->subsystem = SI_SUB_DONE; 208 209 /* Check if we've installed more sysinit items via KLD */ 210 if (newsysinit != NULL) { 211 if (sysinit != (struct sysinit **)sysinit_set.ls_items) 212 free(sysinit, M_TEMP); 213 sysinit = newsysinit; 214 newsysinit = NULL; 215 goto restart; 216 } 217 } 218 219 panic("Shouldn't get here!"); 220 /* NOTREACHED*/ 221 } 222 223 224 /* 225 *************************************************************************** 226 **** 227 **** The following SYSINIT's belong elsewhere, but have not yet 228 **** been moved. 229 **** 230 *************************************************************************** 231 */ 232 static void 233 print_caddr_t(void *data __unused) 234 { 235 printf("%s", (char *)data); 236 } 237 SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright) 238 239 240 /* 241 *************************************************************************** 242 **** 243 **** The two following SYSINT's are proc0 specific glue code. I am not 244 **** convinced that they can not be safely combined, but their order of 245 **** operation has been maintained as the same as the original init_main.c 246 **** for right now. 247 **** 248 **** These probably belong in init_proc.c or kern_proc.c, since they 249 **** deal with proc0 (the fork template process). 250 **** 251 *************************************************************************** 252 */ 253 /* ARGSUSED*/ 254 static void 255 proc0_init(void *dummy __unused) 256 { 257 register struct proc *p; 258 register struct filedesc0 *fdp; 259 register unsigned i; 260 261 p = &proc0; 262 263 /* 264 * Initialize process and pgrp structures. 265 */ 266 procinit(); 267 268 /* 269 * Initialize sleep queue hash table 270 */ 271 sleepinit(); 272 273 /* 274 * additional VM structures 275 */ 276 vm_init2(); 277 278 /* 279 * Create process 0 (the swapper). 280 */ 281 LIST_INSERT_HEAD(&allproc, p, p_list); 282 p->p_pgrp = &pgrp0; 283 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash); 284 LIST_INIT(&pgrp0.pg_members); 285 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist); 286 287 pgrp0.pg_session = &session0; 288 session0.s_count = 1; 289 session0.s_leader = p; 290 291 p->p_sysent = &aout_sysvec; 292 293 p->p_flag = P_INMEM | P_SYSTEM; 294 p->p_stat = SRUN; 295 p->p_nice = NZERO; 296 p->p_rtprio.type = RTP_PRIO_NORMAL; 297 p->p_rtprio.prio = 0; 298 299 p->p_peers = 0; 300 p->p_leader = p; 301 302 bcopy("swapper", p->p_comm, sizeof ("swapper")); 303 304 /* Create credentials. */ 305 cred0.p_refcnt = 1; 306 cred0.p_uidinfo = uifind(0); 307 p->p_cred = &cred0; 308 p->p_ucred = crget(); 309 p->p_ucred->cr_ngroups = 1; /* group 0 */ 310 p->p_ucred->cr_uidinfo = uifind(0); 311 312 /* Don't jail it */ 313 p->p_prison = 0; 314 315 /* Create procsig. */ 316 p->p_procsig = &procsig0; 317 p->p_procsig->ps_refcnt = 1; 318 319 /* Initialize signal state for process 0. */ 320 siginit(&proc0); 321 322 /* Create the file descriptor table. */ 323 fdp = &filedesc0; 324 p->p_fd = &fdp->fd_fd; 325 p->p_fdtol = NULL; 326 fdp->fd_fd.fd_refcnt = 1; 327 fdp->fd_fd.fd_cmask = cmask; 328 fdp->fd_fd.fd_ofiles = fdp->fd_dfiles; 329 fdp->fd_fd.fd_ofileflags = fdp->fd_dfileflags; 330 fdp->fd_fd.fd_nfiles = NDFILE; 331 332 /* Create the limits structures. */ 333 p->p_limit = &limit0; 334 for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++) 335 limit0.pl_rlimit[i].rlim_cur = 336 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY; 337 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur = 338 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles; 339 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur = 340 limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc; 341 i = ptoa(cnt.v_free_count); 342 limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i; 343 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i; 344 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3; 345 limit0.p_cpulimit = RLIM_INFINITY; 346 limit0.p_refcnt = 1; 347 348 /* Allocate a prototype map so we have something to fork. */ 349 pmap_pinit0(vmspace_pmap(&vmspace0)); 350 p->p_vmspace = &vmspace0; 351 vmspace0.vm_refcnt = 1; 352 vm_map_init(&vmspace0.vm_map, round_page(VM_MIN_ADDRESS), 353 trunc_page(VM_MAXUSER_ADDRESS)); 354 vmspace0.vm_map.pmap = vmspace_pmap(&vmspace0); 355 p->p_addr = proc0paddr; /* XXX */ 356 357 /* 358 * We continue to place resource usage info and signal 359 * actions in the user struct so they're pageable. 360 */ 361 p->p_stats = &p->p_addr->u_stats; 362 p->p_sigacts = &p->p_addr->u_sigacts; 363 364 /* 365 * Charge root for one process. 366 */ 367 (void)chgproccnt(cred0.p_uidinfo, 1, 0); 368 369 /* 370 * Initialize the current process pointer (curproc) before 371 * any possible traps/probes to simplify trap processing. 372 */ 373 SET_CURPROC(p); 374 375 } 376 SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL) 377 378 /* ARGSUSED*/ 379 static void 380 proc0_post(void *dummy __unused) 381 { 382 struct timespec ts; 383 struct proc *p; 384 385 /* 386 * Now we can look at the time, having had a chance to verify the 387 * time from the file system. Pretend that proc0 started now. 388 */ 389 LIST_FOREACH(p, &allproc, p_list) { 390 microtime(&p->p_stats->p_start); 391 p->p_runtime = 0; 392 } 393 microuptime(&switchtime); 394 switchticks = ticks; 395 396 /* 397 * Give the ``random'' number generator a thump. 398 * XXX: Does read_random() contain enough bits to be used here ? 399 */ 400 nanotime(&ts); 401 srandom(ts.tv_sec ^ ts.tv_nsec); 402 } 403 SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL) 404 405 /* 406 *************************************************************************** 407 **** 408 **** The following SYSINIT's and glue code should be moved to the 409 **** respective files on a per subsystem basis. 410 **** 411 *************************************************************************** 412 */ 413 414 415 /* 416 *************************************************************************** 417 **** 418 **** The following code probably belongs in another file, like 419 **** kern/init_init.c. 420 **** 421 *************************************************************************** 422 */ 423 424 /* 425 * List of paths to try when searching for "init". 426 */ 427 static char init_path[MAXPATHLEN] = 428 #ifdef INIT_PATH 429 __XSTRING(INIT_PATH); 430 #else 431 "/sbin/init:/sbin/oinit:/sbin/init.bak:/stand/sysinstall"; 432 #endif 433 SYSCTL_STRING(_kern, OID_AUTO, init_path, CTLFLAG_RD, init_path, 0, ""); 434 435 /* 436 * Start the initial user process; try exec'ing each pathname in init_path. 437 * The program is invoked with one argument containing the boot flags. 438 */ 439 static void 440 start_init(void *dummy) 441 { 442 vm_offset_t addr; 443 struct execve_args args; 444 int options, error; 445 char *var, *path, *next, *s; 446 char *ucp, **uap, *arg0, *arg1; 447 struct proc *p; 448 449 p = curproc; 450 451 /* Get the vnode for '/'. Set p->p_fd->fd_cdir to reference it. */ 452 if (VFS_ROOT(TAILQ_FIRST(&mountlist), &rootvnode)) 453 panic("cannot find root vnode"); 454 p->p_fd->fd_cdir = rootvnode; 455 VREF(p->p_fd->fd_cdir); 456 p->p_fd->fd_rdir = rootvnode; 457 VREF(p->p_fd->fd_rdir); 458 VOP_UNLOCK(rootvnode, 0, p); 459 460 /* 461 * Need just enough stack to hold the faked-up "execve()" arguments. 462 */ 463 addr = trunc_page(USRSTACK - PAGE_SIZE); 464 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE, 465 FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0) 466 panic("init: couldn't allocate argument space"); 467 p->p_vmspace->vm_maxsaddr = (caddr_t)addr; 468 p->p_vmspace->vm_ssize = 1; 469 470 if ((var = getenv("init_path")) != NULL) { 471 strncpy(init_path, var, sizeof init_path); 472 init_path[sizeof init_path - 1] = 0; 473 } 474 if ((var = getenv("kern.fallback_elf_brand")) != NULL) 475 fallback_elf_brand = strtol(var, NULL, 0); 476 477 for (path = init_path; *path != '\0'; path = next) { 478 while (*path == ':') 479 path++; 480 if (*path == '\0') 481 break; 482 for (next = path; *next != '\0' && *next != ':'; next++) 483 /* nothing */ ; 484 if (bootverbose) 485 printf("start_init: trying %.*s\n", (int)(next - path), 486 path); 487 488 /* 489 * Move out the boot flag argument. 490 */ 491 options = 0; 492 ucp = (char *)USRSTACK; 493 (void)subyte(--ucp, 0); /* trailing zero */ 494 if (boothowto & RB_SINGLE) { 495 (void)subyte(--ucp, 's'); 496 options = 1; 497 } 498 #ifdef notyet 499 if (boothowto & RB_FASTBOOT) { 500 (void)subyte(--ucp, 'f'); 501 options = 1; 502 } 503 #endif 504 505 #ifdef BOOTCDROM 506 (void)subyte(--ucp, 'C'); 507 options = 1; 508 #endif 509 if (options == 0) 510 (void)subyte(--ucp, '-'); 511 (void)subyte(--ucp, '-'); /* leading hyphen */ 512 arg1 = ucp; 513 514 /* 515 * Move out the file name (also arg 0). 516 */ 517 (void)subyte(--ucp, 0); 518 for (s = next - 1; s >= path; s--) 519 (void)subyte(--ucp, *s); 520 arg0 = ucp; 521 522 /* 523 * Move out the arg pointers. 524 */ 525 uap = (char **)((intptr_t)ucp & ~(sizeof(intptr_t)-1)); 526 (void)suword((caddr_t)--uap, (long)0); /* terminator */ 527 (void)suword((caddr_t)--uap, (long)(intptr_t)arg1); 528 (void)suword((caddr_t)--uap, (long)(intptr_t)arg0); 529 530 /* 531 * Point at the arguments. 532 */ 533 args.fname = arg0; 534 args.argv = uap; 535 args.envv = NULL; 536 537 /* 538 * Now try to exec the program. If can't for any reason 539 * other than it doesn't exist, complain. 540 * 541 * Otherwise, return via fork_trampoline() all the way 542 * to user mode as init! 543 */ 544 if ((error = execve(p, &args)) == 0) 545 return; 546 if (error != ENOENT) 547 printf("exec %.*s: error %d\n", (int)(next - path), 548 path, error); 549 } 550 printf("init: not found in path %s\n", init_path); 551 panic("no init"); 552 } 553 554 /* 555 * Like kthread_create(), but runs in it's own address space. 556 * We do this early to reserve pid 1. 557 * 558 * Note special case - do not make it runnable yet. Other work 559 * in progress will change this more. 560 */ 561 static void 562 create_init(const void *udata __unused) 563 { 564 int error; 565 int s; 566 567 s = splhigh(); 568 error = fork1(&proc0, RFFDG | RFPROC, &initproc); 569 if (error) 570 panic("cannot fork init: %d\n", error); 571 initproc->p_flag |= P_INMEM | P_SYSTEM; 572 cpu_set_fork_handler(initproc, start_init, NULL); 573 remrunqueue(initproc); 574 splx(s); 575 } 576 SYSINIT(init,SI_SUB_CREATE_INIT, SI_ORDER_FIRST, create_init, NULL) 577 578 /* 579 * Make it runnable now. 580 */ 581 static void 582 kick_init(const void *udata __unused) 583 { 584 setrunqueue(initproc); 585 } 586 SYSINIT(kickinit,SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kick_init, NULL) 587