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 * $FreeBSD$ 43 */ 44 45 #include "opt_rlimit.h" 46 #include "opt_devfs.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 64 #include <machine/cpu.h> 65 66 #include <vm/vm.h> 67 #include <vm/vm_param.h> 68 #include <vm/vm_prot.h> 69 #include <sys/lock.h> 70 #include <vm/pmap.h> 71 #include <vm/vm_map.h> 72 #include <sys/user.h> 73 74 extern struct linker_set sysinit_set; /* XXX */ 75 76 extern void __main __P((void)); 77 extern void main __P((void *framep)); 78 79 /* Components of the first process -- never freed. */ 80 static struct session session0; 81 static struct pgrp pgrp0; 82 struct proc proc0; 83 static struct pcred cred0; 84 static struct filedesc0 filedesc0; 85 static struct plimit limit0; 86 static struct vmspace vmspace0; 87 struct proc *curproc = &proc0; 88 struct proc *initproc; 89 90 int cmask = CMASK; 91 extern struct user *proc0paddr; 92 93 struct vnode *rootvp; 94 int boothowto; 95 96 struct timeval boottime; 97 SYSCTL_STRUCT(_kern, KERN_BOOTTIME, boottime, 98 CTLFLAG_RW, &boottime, timeval, ""); 99 100 struct timeval runtime; 101 102 /* 103 * Promiscuous argument pass for start_init() 104 * 105 * This is a kludge because we use a return from main() rather than a call 106 * to a new routine in locore.s to kick the kernel alive from locore.s. 107 */ 108 static void *init_framep; 109 110 111 #if __GNUC__ >= 2 112 void __main() {} 113 #endif 114 115 116 /* 117 * This ensures that there is at least one entry so that the sysinit_set 118 * symbol is not undefined. A sybsystem ID of SI_SUB_DUMMY is never 119 * executed. 120 */ 121 SYSINIT(placeholder, SI_SUB_DUMMY,SI_ORDER_ANY, NULL, NULL) 122 123 124 /* 125 * System startup; initialize the world, create process 0, mount root 126 * filesystem, and fork to create init and pagedaemon. Most of the 127 * hard work is done in the lower-level initialization routines including 128 * startup(), which does memory initialization and autoconfiguration. 129 * 130 * This allows simple addition of new kernel subsystems that require 131 * boot time initialization. It also allows substitution of subsystem 132 * (for instance, a scheduler, kernel profiler, or VM system) by object 133 * module. Finally, it allows for optional "kernel threads", like an LFS 134 * cleaner. 135 */ 136 void 137 main(framep) 138 void *framep; 139 { 140 141 register struct sysinit **sipp; /* system initialization*/ 142 register struct sysinit **xipp; /* interior loop of sort*/ 143 register struct sysinit *save; /* bubble*/ 144 int rval[2]; /* SI_TYPE_KTHREAD support*/ 145 146 /* 147 * Save the locore.s frame pointer for start_init(). 148 */ 149 init_framep = framep; 150 151 /* 152 * Perform a bubble sort of the system initialization objects by 153 * their subsystem (primary key) and order (secondary key). 154 * 155 * Since some things care about execution order, this is the 156 * operation which ensures continued function. 157 */ 158 for( sipp = (struct sysinit **)sysinit_set.ls_items; *sipp; sipp++) { 159 for( xipp = sipp + 1; *xipp; xipp++) { 160 if( (*sipp)->subsystem < (*xipp)->subsystem || 161 ( (*sipp)->subsystem == (*xipp)->subsystem && 162 (*sipp)->order < (*xipp)->order)) 163 continue; /* skip*/ 164 save = *sipp; 165 *sipp = *xipp; 166 *xipp = save; 167 } 168 } 169 170 /* 171 * Traverse the (now) ordered list of system initialization tasks. 172 * Perform each task, and continue on to the next task. 173 * 174 * The last item on the list is expected to be the scheduler, 175 * which will not return. 176 */ 177 for( sipp = (struct sysinit **)sysinit_set.ls_items; *sipp; sipp++) { 178 if( (*sipp)->subsystem == SI_SUB_DUMMY) 179 continue; /* skip dummy task(s)*/ 180 181 switch( (*sipp)->type) { 182 case SI_TYPE_DEFAULT: 183 /* no special processing*/ 184 (*((*sipp)->func))( (*sipp)->udata); 185 break; 186 187 case SI_TYPE_KTHREAD: 188 /* kernel thread*/ 189 if (fork(&proc0, NULL, rval)) 190 panic("fork kernel process"); 191 if (rval[1]) { 192 (*((*sipp)->func))( (*sipp)->udata); 193 /* 194 * The call to start "init" returns 195 * here after the scheduler has been 196 * started, and returns to the caller 197 * in i386/i386/locore.s. This is a 198 * necessary part of initialization 199 * and is rather non-obvious. 200 * 201 * No other "kernel threads" should 202 * return here. Call panic() instead. 203 */ 204 return; 205 } 206 break; 207 208 default: 209 panic( "init_main: unrecognized init type"); 210 } 211 } 212 213 /* NOTREACHED*/ 214 } 215 216 217 /* 218 * Start a kernel process. This is called after a fork() call in 219 * main() in the file kern/init_main.c. 220 * 221 * This function is used to start "internal" daemons. 222 */ 223 /* ARGSUSED*/ 224 void 225 kproc_start(udata) 226 void *udata; 227 { 228 struct kproc_desc *kp = udata; 229 struct proc *p = curproc; 230 231 /* save a global descriptor, if desired*/ 232 if( kp->global_procpp != NULL) 233 *kp->global_procpp = p; 234 235 /* this is a non-swapped system process*/ 236 p->p_flag |= P_INMEM | P_SYSTEM; 237 238 /* set up arg0 for 'ps', et al*/ 239 strcpy( p->p_comm, kp->arg0); 240 241 /* call the processes' main()...*/ 242 (*kp->func)(); 243 244 /* NOTREACHED */ 245 panic("kproc_start: %s", kp->arg0); 246 } 247 248 249 /* 250 *************************************************************************** 251 **** 252 **** The following SYSINIT's belong elsewhere, but have not yet 253 **** been moved. 254 **** 255 *************************************************************************** 256 */ 257 #ifdef OMIT 258 /* 259 * Handled by vfs_mountroot (bad idea) at this time... should be 260 * done the same as 4.4Lite2. 261 */ 262 SYSINIT(swapinit, SI_SUB_SWAP, SI_ORDER_FIRST, swapinit, NULL) 263 #endif /* OMIT*/ 264 265 /* 266 * Should get its own file... 267 */ 268 #ifdef HPFPLIB 269 char copyright[] = 270 "Copyright (c) 1982, 1986, 1989, 1991, 1993\n\tThe Regents of the University of California.\nCopyright (c) 1992 Hewlett-Packard Company\nCopyright (c) 1992 Motorola Inc.\nAll rights reserved.\n\n"; 271 #else 272 char copyright[] = 273 "Copyright (c) 1992-1996 FreeBSD Inc.\n" 274 #ifdef PC98 275 "Copyright (c) 1994-1996 FreeBSD(98) porting team.\n" 276 "Copyright (c) 1982, 1986, 1989, 1991, 1993\n\tThe Regents of the University of California.\n" 277 "Copyright (c) 1992 A.Kojima F.Ukai M.Ishii (KMC).\n" 278 "\tAll rights reserved.\n\n"; 279 #else 280 "Copyright (c) 1982, 1986, 1989, 1991, 1993\n\tThe Regents of the University of California. All rights reserved.\n\n"; 281 #endif 282 #endif 283 static void print_caddr_t __P((void *data)); 284 static void 285 print_caddr_t(data) 286 void *data; 287 { 288 printf("%s", (char *)data); 289 } 290 SYSINIT(announce, SI_SUB_COPYRIGHT, SI_ORDER_FIRST, print_caddr_t, copyright) 291 292 293 /* 294 *************************************************************************** 295 **** 296 **** The two following SYSINT's are proc0 specific glue code. I am not 297 **** convinced that they can not be safely combined, but their order of 298 **** operation has been maintained as the same as the original init_main.c 299 **** for right now. 300 **** 301 **** These probably belong in init_proc.c or kern_proc.c, since they 302 **** deal with proc0 (the fork template process). 303 **** 304 *************************************************************************** 305 */ 306 /* ARGSUSED*/ 307 static void proc0_init __P((void *dummy)); 308 static void 309 proc0_init(dummy) 310 void *dummy; 311 { 312 register struct proc *p; 313 register struct filedesc0 *fdp; 314 register unsigned i; 315 316 /* 317 * Initialize the current process pointer (curproc) before 318 * any possible traps/probes to simplify trap processing. 319 */ 320 p = &proc0; 321 curproc = p; /* XXX redundant*/ 322 323 /* 324 * Initialize process and pgrp structures. 325 */ 326 procinit(); 327 328 /* 329 * Initialize sleep queue hash table 330 */ 331 sleepinit(); 332 333 /* 334 * Create process 0 (the swapper). 335 */ 336 LIST_INSERT_HEAD(&allproc, p, p_list); 337 p->p_pgrp = &pgrp0; 338 LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash); 339 LIST_INIT(&pgrp0.pg_members); 340 LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist); 341 342 pgrp0.pg_session = &session0; 343 session0.s_count = 1; 344 session0.s_leader = p; 345 346 p->p_sysent = &aout_sysvec; 347 348 p->p_flag = P_INMEM | P_SYSTEM; 349 p->p_stat = SRUN; 350 p->p_nice = NZERO; 351 p->p_rtprio.type = RTP_PRIO_NORMAL; 352 p->p_rtprio.prio = 0; 353 354 bcopy("swapper", p->p_comm, sizeof ("swapper")); 355 356 /* Create credentials. */ 357 cred0.p_refcnt = 1; 358 p->p_cred = &cred0; 359 p->p_ucred = crget(); 360 p->p_ucred->cr_ngroups = 1; /* group 0 */ 361 362 /* Create the file descriptor table. */ 363 fdp = &filedesc0; 364 p->p_fd = &fdp->fd_fd; 365 fdp->fd_fd.fd_refcnt = 1; 366 fdp->fd_fd.fd_cmask = cmask; 367 fdp->fd_fd.fd_ofiles = fdp->fd_dfiles; 368 fdp->fd_fd.fd_ofileflags = fdp->fd_dfileflags; 369 fdp->fd_fd.fd_nfiles = NDFILE; 370 371 /* Create the limits structures. */ 372 p->p_limit = &limit0; 373 for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++) 374 limit0.pl_rlimit[i].rlim_cur = 375 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY; 376 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur = 377 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_max = maxfiles; 378 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur = 379 limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc; 380 i = ptoa(cnt.v_free_count); 381 limit0.pl_rlimit[RLIMIT_RSS].rlim_max = i; 382 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_max = i; 383 limit0.pl_rlimit[RLIMIT_MEMLOCK].rlim_cur = i / 3; 384 limit0.p_refcnt = 1; 385 386 /* Allocate a prototype map so we have something to fork. */ 387 p->p_vmspace = &vmspace0; 388 vmspace0.vm_refcnt = 1; 389 pmap_pinit(&vmspace0.vm_pmap); 390 vm_map_init(&vmspace0.vm_map, round_page(VM_MIN_ADDRESS), 391 trunc_page(VM_MAXUSER_ADDRESS), TRUE); 392 vmspace0.vm_map.pmap = &vmspace0.vm_pmap; 393 p->p_addr = proc0paddr; /* XXX */ 394 395 #define INCOMPAT_LITES2 396 #ifdef INCOMPAT_LITES2 397 /* 398 * proc0 needs to have a coherent frame base, too. 399 * This probably makes the identical call for the init proc 400 * that happens later unnecessary since it should inherit 401 * it during the fork. 402 */ 403 cpu_set_init_frame(p, init_framep); /* XXX! */ 404 #endif /* INCOMPAT_LITES2*/ 405 406 /* 407 * We continue to place resource usage info and signal 408 * actions in the user struct so they're pageable. 409 */ 410 p->p_stats = &p->p_addr->u_stats; 411 p->p_sigacts = &p->p_addr->u_sigacts; 412 413 /* 414 * Charge root for one process. 415 */ 416 (void)chgproccnt(0, 1); 417 } 418 SYSINIT(p0init, SI_SUB_INTRINSIC, SI_ORDER_FIRST, proc0_init, NULL) 419 420 /* ARGSUSED*/ 421 static void proc0_post __P((void *dummy)); 422 static void 423 proc0_post(dummy) 424 void *dummy; 425 { 426 struct timeval tv; 427 428 /* 429 * Now can look at time, having had a chance to verify the time 430 * from the file system. Reset p->p_rtime as it may have been 431 * munched in mi_switch() after the time got set. 432 */ 433 proc0.p_stats->p_start = runtime = mono_time = boottime = time; 434 proc0.p_rtime.tv_sec = proc0.p_rtime.tv_usec = 0; 435 436 /* 437 * Give the ``random'' number generator a thump. 438 */ 439 microtime(&tv); 440 srandom(tv.tv_sec ^ tv.tv_usec); 441 442 /* Initialize signal state for process 0. */ 443 siginit(&proc0); 444 } 445 SYSINIT(p0post, SI_SUB_INTRINSIC_POST, SI_ORDER_FIRST, proc0_post, NULL) 446 447 448 449 450 /* 451 *************************************************************************** 452 **** 453 **** The following SYSINIT's and glue code should be moved to the 454 **** respective files on a per subsystem basis. 455 **** 456 *************************************************************************** 457 */ 458 /* ARGSUSED*/ 459 static void sched_setup __P((void *dummy)); 460 static void 461 sched_setup(dummy) 462 void *dummy; 463 { 464 /* Kick off timeout driven events by calling first time. */ 465 roundrobin(NULL); 466 schedcpu(NULL); 467 } 468 SYSINIT(sched_setup, SI_SUB_KICK_SCHEDULER, SI_ORDER_FIRST, sched_setup, NULL) 469 470 /* ARGSUSED*/ 471 static void xxx_vfs_mountroot __P((void *fsnamep)); 472 static void 473 xxx_vfs_mountroot(fsnamep) 474 void *fsnamep; 475 { 476 /* Mount the root file system. */ 477 if (vfs_mountrootfs(*((char **) fsnamep))) 478 panic("cannot mount root"); 479 } 480 SYSINIT(mountroot, SI_SUB_ROOT, SI_ORDER_FIRST, xxx_vfs_mountroot, &mountrootfsname) 481 482 /* ARGSUSED*/ 483 static void xxx_vfs_root_fdtab __P((void *dummy)); 484 static void 485 xxx_vfs_root_fdtab(dummy) 486 void *dummy; 487 { 488 register struct filedesc0 *fdp = &filedesc0; 489 490 /* Get the vnode for '/'. Set fdp->fd_fd.fd_cdir to reference it. */ 491 if (VFS_ROOT(mountlist.cqh_first, &rootvnode)) 492 panic("cannot find root vnode"); 493 fdp->fd_fd.fd_cdir = rootvnode; 494 VREF(fdp->fd_fd.fd_cdir); 495 VOP_UNLOCK(rootvnode, 0, &proc0); 496 fdp->fd_fd.fd_rdir = NULL; 497 } 498 SYSINIT(retrofit, SI_SUB_ROOT_FDTAB, SI_ORDER_FIRST, xxx_vfs_root_fdtab, NULL) 499 500 501 /* 502 *************************************************************************** 503 **** 504 **** The following code probably belongs in another file, like 505 **** kern/init_init.c. It is here for two reasons only: 506 **** 507 **** 1) This code returns to startup the system; this is 508 **** abnormal for a kernel thread. 509 **** 2) This code promiscuously uses init_frame 510 **** 511 *************************************************************************** 512 */ 513 514 static void kthread_init __P((void *dummy)); 515 SYSINIT_KT(init,SI_SUB_KTHREAD_INIT, SI_ORDER_FIRST, kthread_init, NULL) 516 517 518 static void start_init __P((struct proc *p, void *framep)); 519 520 /* ARGSUSED*/ 521 static void 522 kthread_init(dummy) 523 void *dummy; 524 { 525 526 /* Create process 1 (init(8)). */ 527 start_init(curproc, init_framep); 528 529 /* 530 * This is the only kernel thread allowed to return yo the 531 * caller!!! 532 */ 533 return; 534 } 535 536 537 /* 538 * List of paths to try when searching for "init". 539 */ 540 static char *initpaths[] = { 541 "/sbin/init", 542 "/sbin/oinit", 543 "/sbin/init.bak", 544 "/stand/sysinstall", 545 NULL, 546 }; 547 548 /* 549 * Start the initial user process; try exec'ing each pathname in "initpaths". 550 * The program is invoked with one argument containing the boot flags. 551 */ 552 static void 553 start_init(p, framep) 554 struct proc *p; 555 void *framep; 556 { 557 vm_offset_t addr; 558 struct execve_args args; 559 int options, i, retval[2], error; 560 char **pathp, *path, *ucp, **uap, *arg0, *arg1; 561 562 initproc = p; 563 564 /* 565 * We need to set the system call frame as if we were entered through 566 * a syscall() so that when we call execve() below, it will be able 567 * to set the entry point (see setregs) when it tries to exec. The 568 * startup code in "locore.s" has allocated space for the frame and 569 * passed a pointer to that space as main's argument. 570 */ 571 cpu_set_init_frame(p, framep); 572 573 /* 574 * Need just enough stack to hold the faked-up "execve()" arguments. 575 */ 576 addr = trunc_page(VM_MAXUSER_ADDRESS - PAGE_SIZE); 577 if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE, FALSE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0) 578 panic("init: couldn't allocate argument space"); 579 p->p_vmspace->vm_maxsaddr = (caddr_t)addr; 580 p->p_vmspace->vm_ssize = 1; 581 582 for (pathp = &initpaths[0]; (path = *pathp) != NULL; pathp++) { 583 /* 584 * Move out the boot flag argument. 585 */ 586 options = 0; 587 ucp = (char *)USRSTACK; 588 (void)subyte(--ucp, 0); /* trailing zero */ 589 if (boothowto & RB_SINGLE) { 590 (void)subyte(--ucp, 's'); 591 options = 1; 592 } 593 #ifdef notyet 594 if (boothowto & RB_FASTBOOT) { 595 (void)subyte(--ucp, 'f'); 596 options = 1; 597 } 598 #endif 599 600 #ifdef BOOTCDROM 601 (void)subyte(--ucp, 'C'); 602 options = 1; 603 #endif 604 605 #if defined(DEVFS) && defined(DEVFS_ROOT) 606 (void)subyte(--ucp, 'd'); 607 options = 1; 608 #endif 609 if (options == 0) 610 (void)subyte(--ucp, '-'); 611 (void)subyte(--ucp, '-'); /* leading hyphen */ 612 arg1 = ucp; 613 614 /* 615 * Move out the file name (also arg 0). 616 */ 617 for (i = strlen(path) + 1; i >= 0; i--) 618 (void)subyte(--ucp, path[i]); 619 arg0 = ucp; 620 621 /* 622 * Move out the arg pointers. 623 */ 624 uap = (char **)((int)ucp & ~(NBPW-1)); 625 (void)suword((caddr_t)--uap, 0); /* terminator */ 626 (void)suword((caddr_t)--uap, (int)arg1); 627 (void)suword((caddr_t)--uap, (int)arg0); 628 629 /* 630 * Point at the arguments. 631 */ 632 args.fname = arg0; 633 args.argv = uap; 634 args.envv = NULL; 635 636 /* 637 * Now try to exec the program. If can't for any reason 638 * other than it doesn't exist, complain. 639 * 640 * Otherwise return to main() which returns to btext 641 * which completes the system startup. 642 */ 643 if ((error = execve(p, &args, &retval[0])) == 0) 644 return; 645 if (error != ENOENT) 646 printf("exec %s: error %d\n", path, error); 647 } 648 printf("init: not found\n"); 649 panic("no init"); 650 } 651