kern_proc.c revision 1.51
1 /* $NetBSD: kern_proc.c,v 1.51 2002/08/28 07:16:38 gmcgarry Exp $ */ 2 3 /*- 4 * Copyright (c) 1999 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to The NetBSD Foundation 8 * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility, 9 * NASA Ames Research Center. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 3. All advertising materials mentioning features or use of this software 20 * must display the following acknowledgement: 21 * This product includes software developed by the NetBSD 22 * Foundation, Inc. and its contributors. 23 * 4. Neither the name of The NetBSD Foundation nor the names of its 24 * contributors may be used to endorse or promote products derived 25 * from this software without specific prior written permission. 26 * 27 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 28 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 29 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 30 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 31 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 32 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 33 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 34 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 35 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 36 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 37 * POSSIBILITY OF SUCH DAMAGE. 38 */ 39 40 /* 41 * Copyright (c) 1982, 1986, 1989, 1991, 1993 42 * The Regents of the University of California. All rights reserved. 43 * 44 * Redistribution and use in source and binary forms, with or without 45 * modification, are permitted provided that the following conditions 46 * are met: 47 * 1. Redistributions of source code must retain the above copyright 48 * notice, this list of conditions and the following disclaimer. 49 * 2. Redistributions in binary form must reproduce the above copyright 50 * notice, this list of conditions and the following disclaimer in the 51 * documentation and/or other materials provided with the distribution. 52 * 3. All advertising materials mentioning features or use of this software 53 * must display the following acknowledgement: 54 * This product includes software developed by the University of 55 * California, Berkeley and its contributors. 56 * 4. Neither the name of the University nor the names of its contributors 57 * may be used to endorse or promote products derived from this software 58 * without specific prior written permission. 59 * 60 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 61 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 62 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 63 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 64 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 65 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 66 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 67 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 68 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 69 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 70 * SUCH DAMAGE. 71 * 72 * @(#)kern_proc.c 8.7 (Berkeley) 2/14/95 73 */ 74 75 #include <sys/cdefs.h> 76 __KERNEL_RCSID(0, "$NetBSD: kern_proc.c,v 1.51 2002/08/28 07:16:38 gmcgarry Exp $"); 77 78 #include "opt_kstack.h" 79 80 #include <sys/param.h> 81 #include <sys/systm.h> 82 #include <sys/map.h> 83 #include <sys/kernel.h> 84 #include <sys/proc.h> 85 #include <sys/resourcevar.h> 86 #include <sys/buf.h> 87 #include <sys/acct.h> 88 #include <sys/wait.h> 89 #include <sys/file.h> 90 #include <ufs/ufs/quota.h> 91 #include <sys/uio.h> 92 #include <sys/malloc.h> 93 #include <sys/pool.h> 94 #include <sys/mbuf.h> 95 #include <sys/ioctl.h> 96 #include <sys/tty.h> 97 #include <sys/signalvar.h> 98 #include <sys/ras.h> 99 100 /* 101 * Structure associated with user cacheing. 102 */ 103 struct uidinfo { 104 LIST_ENTRY(uidinfo) ui_hash; 105 uid_t ui_uid; 106 long ui_proccnt; 107 }; 108 #define UIHASH(uid) (&uihashtbl[(uid) & uihash]) 109 LIST_HEAD(uihashhead, uidinfo) *uihashtbl; 110 u_long uihash; /* size of hash table - 1 */ 111 112 /* 113 * Other process lists 114 */ 115 struct pidhashhead *pidhashtbl; 116 u_long pidhash; 117 struct pgrphashhead *pgrphashtbl; 118 u_long pgrphash; 119 120 struct proclist allproc; 121 struct proclist zombproc; /* resources have been freed */ 122 123 /* 124 * Process list locking: 125 * 126 * We have two types of locks on the proclists: read locks and write 127 * locks. Read locks can be used in interrupt context, so while we 128 * hold the write lock, we must also block clock interrupts to 129 * lock out any scheduling changes that may happen in interrupt 130 * context. 131 * 132 * The proclist lock locks the following structures: 133 * 134 * allproc 135 * zombproc 136 * pidhashtbl 137 */ 138 struct lock proclist_lock; 139 140 /* 141 * Locking of this proclist is special; it's accessed in a 142 * critical section of process exit, and thus locking it can't 143 * modify interrupt state. We use a simple spin lock for this 144 * proclist. Processes on this proclist are also on zombproc; 145 * we use the p_hash member to linkup to deadproc. 146 */ 147 struct simplelock deadproc_slock; 148 struct proclist deadproc; /* dead, but not yet undead */ 149 150 struct pool proc_pool; 151 struct pool pcred_pool; 152 struct pool plimit_pool; 153 struct pool pgrp_pool; 154 struct pool rusage_pool; 155 struct pool ras_pool; 156 157 /* 158 * The process list descriptors, used during pid allocation and 159 * by sysctl. No locking on this data structure is needed since 160 * it is completely static. 161 */ 162 const struct proclist_desc proclists[] = { 163 { &allproc }, 164 { &zombproc }, 165 { NULL }, 166 }; 167 168 static void orphanpg __P((struct pgrp *)); 169 #ifdef DEBUG 170 void pgrpdump __P((void)); 171 #endif 172 173 /* 174 * Initialize global process hashing structures. 175 */ 176 void 177 procinit() 178 { 179 const struct proclist_desc *pd; 180 181 for (pd = proclists; pd->pd_list != NULL; pd++) 182 LIST_INIT(pd->pd_list); 183 184 spinlockinit(&proclist_lock, "proclk", 0); 185 186 LIST_INIT(&deadproc); 187 simple_lock_init(&deadproc_slock); 188 189 pidhashtbl = 190 hashinit(maxproc / 4, HASH_LIST, M_PROC, M_WAITOK, &pidhash); 191 pgrphashtbl = 192 hashinit(maxproc / 4, HASH_LIST, M_PROC, M_WAITOK, &pgrphash); 193 uihashtbl = 194 hashinit(maxproc / 16, HASH_LIST, M_PROC, M_WAITOK, &uihash); 195 196 pool_init(&proc_pool, sizeof(struct proc), 0, 0, 0, "procpl", 197 &pool_allocator_nointr); 198 pool_init(&pgrp_pool, sizeof(struct pgrp), 0, 0, 0, "pgrppl", 199 &pool_allocator_nointr); 200 pool_init(&pcred_pool, sizeof(struct pcred), 0, 0, 0, "pcredpl", 201 &pool_allocator_nointr); 202 pool_init(&plimit_pool, sizeof(struct plimit), 0, 0, 0, "plimitpl", 203 &pool_allocator_nointr); 204 pool_init(&rusage_pool, sizeof(struct rusage), 0, 0, 0, "rusgepl", 205 &pool_allocator_nointr); 206 pool_init(&ras_pool, sizeof(struct ras), 0, 0, 0, "raspl", 207 &pool_allocator_nointr); 208 } 209 210 /* 211 * Acquire a read lock on the proclist. 212 */ 213 void 214 proclist_lock_read() 215 { 216 int error; 217 218 error = spinlockmgr(&proclist_lock, LK_SHARED, NULL); 219 #ifdef DIAGNOSTIC 220 if (__predict_false(error != 0)) 221 panic("proclist_lock_read: failed to acquire lock"); 222 #endif 223 } 224 225 /* 226 * Release a read lock on the proclist. 227 */ 228 void 229 proclist_unlock_read() 230 { 231 232 (void) spinlockmgr(&proclist_lock, LK_RELEASE, NULL); 233 } 234 235 /* 236 * Acquire a write lock on the proclist. 237 */ 238 int 239 proclist_lock_write() 240 { 241 int s, error; 242 243 s = splclock(); 244 error = spinlockmgr(&proclist_lock, LK_EXCLUSIVE, NULL); 245 #ifdef DIAGNOSTIC 246 if (__predict_false(error != 0)) 247 panic("proclist_lock: failed to acquire lock"); 248 #endif 249 return (s); 250 } 251 252 /* 253 * Release a write lock on the proclist. 254 */ 255 void 256 proclist_unlock_write(s) 257 int s; 258 { 259 260 (void) spinlockmgr(&proclist_lock, LK_RELEASE, NULL); 261 splx(s); 262 } 263 264 /* 265 * Change the count associated with number of processes 266 * a given user is using. 267 */ 268 int 269 chgproccnt(uid, diff) 270 uid_t uid; 271 int diff; 272 { 273 struct uidinfo *uip; 274 struct uihashhead *uipp; 275 276 uipp = UIHASH(uid); 277 for (uip = uipp->lh_first; uip != 0; uip = uip->ui_hash.le_next) 278 if (uip->ui_uid == uid) 279 break; 280 if (uip) { 281 uip->ui_proccnt += diff; 282 if (uip->ui_proccnt > 0) 283 return (uip->ui_proccnt); 284 if (uip->ui_proccnt < 0) 285 panic("chgproccnt: procs < 0"); 286 LIST_REMOVE(uip, ui_hash); 287 FREE(uip, M_PROC); 288 return (0); 289 } 290 if (diff <= 0) { 291 if (diff == 0) 292 return(0); 293 panic("chgproccnt: lost user"); 294 } 295 MALLOC(uip, struct uidinfo *, sizeof(*uip), M_PROC, M_WAITOK); 296 LIST_INSERT_HEAD(uipp, uip, ui_hash); 297 uip->ui_uid = uid; 298 uip->ui_proccnt = diff; 299 return (diff); 300 } 301 302 /* 303 * Is p an inferior of q? 304 */ 305 int 306 inferior(p, q) 307 struct proc *p; 308 struct proc *q; 309 { 310 311 for (; p != q; p = p->p_pptr) 312 if (p->p_pid == 0) 313 return (0); 314 return (1); 315 } 316 317 /* 318 * Locate a process by number 319 */ 320 struct proc * 321 pfind(pid) 322 pid_t pid; 323 { 324 struct proc *p; 325 326 proclist_lock_read(); 327 for (p = PIDHASH(pid)->lh_first; p != 0; p = p->p_hash.le_next) 328 if (p->p_pid == pid) 329 goto out; 330 out: 331 proclist_unlock_read(); 332 return (p); 333 } 334 335 /* 336 * Locate a process group by number 337 */ 338 struct pgrp * 339 pgfind(pgid) 340 pid_t pgid; 341 { 342 struct pgrp *pgrp; 343 344 for (pgrp = PGRPHASH(pgid)->lh_first; pgrp != NULL; 345 pgrp = pgrp->pg_hash.le_next) 346 if (pgrp->pg_id == pgid) 347 return (pgrp); 348 return (NULL); 349 } 350 351 /* 352 * Move p to a new or existing process group (and session) 353 */ 354 int 355 enterpgrp(p, pgid, mksess) 356 struct proc *p; 357 pid_t pgid; 358 int mksess; 359 { 360 struct pgrp *pgrp = pgfind(pgid); 361 362 #ifdef DIAGNOSTIC 363 if (__predict_false(pgrp != NULL && mksess)) /* firewalls */ 364 panic("enterpgrp: setsid into non-empty pgrp"); 365 if (__predict_false(SESS_LEADER(p))) 366 panic("enterpgrp: session leader attempted setpgrp"); 367 #endif 368 if (pgrp == NULL) { 369 pid_t savepid = p->p_pid; 370 struct proc *np; 371 /* 372 * new process group 373 */ 374 #ifdef DIAGNOSTIC 375 if (__predict_false(p->p_pid != pgid)) 376 panic("enterpgrp: new pgrp and pid != pgid"); 377 #endif 378 pgrp = pool_get(&pgrp_pool, PR_WAITOK); 379 if ((np = pfind(savepid)) == NULL || np != p) { 380 pool_put(&pgrp_pool, pgrp); 381 return (ESRCH); 382 } 383 if (mksess) { 384 struct session *sess; 385 386 /* 387 * new session 388 */ 389 MALLOC(sess, struct session *, sizeof(struct session), 390 M_SESSION, M_WAITOK); 391 if ((np = pfind(savepid)) == NULL || np != p) { 392 FREE(sess, M_SESSION); 393 pool_put(&pgrp_pool, pgrp); 394 return (ESRCH); 395 } 396 sess->s_sid = p->p_pid; 397 sess->s_leader = p; 398 sess->s_count = 1; 399 sess->s_ttyvp = NULL; 400 sess->s_ttyp = NULL; 401 memcpy(sess->s_login, p->p_session->s_login, 402 sizeof(sess->s_login)); 403 p->p_flag &= ~P_CONTROLT; 404 pgrp->pg_session = sess; 405 #ifdef DIAGNOSTIC 406 if (__predict_false(p != curproc)) 407 panic("enterpgrp: mksession and p != curproc"); 408 #endif 409 } else { 410 SESSHOLD(p->p_session); 411 pgrp->pg_session = p->p_session; 412 } 413 pgrp->pg_id = pgid; 414 LIST_INIT(&pgrp->pg_members); 415 LIST_INSERT_HEAD(PGRPHASH(pgid), pgrp, pg_hash); 416 pgrp->pg_jobc = 0; 417 } else if (pgrp == p->p_pgrp) 418 return (0); 419 420 /* 421 * Adjust eligibility of affected pgrps to participate in job control. 422 * Increment eligibility counts before decrementing, otherwise we 423 * could reach 0 spuriously during the first call. 424 */ 425 fixjobc(p, pgrp, 1); 426 fixjobc(p, p->p_pgrp, 0); 427 428 LIST_REMOVE(p, p_pglist); 429 if (p->p_pgrp->pg_members.lh_first == 0) 430 pgdelete(p->p_pgrp); 431 p->p_pgrp = pgrp; 432 LIST_INSERT_HEAD(&pgrp->pg_members, p, p_pglist); 433 return (0); 434 } 435 436 /* 437 * remove process from process group 438 */ 439 int 440 leavepgrp(p) 441 struct proc *p; 442 { 443 444 LIST_REMOVE(p, p_pglist); 445 if (p->p_pgrp->pg_members.lh_first == 0) 446 pgdelete(p->p_pgrp); 447 p->p_pgrp = 0; 448 return (0); 449 } 450 451 /* 452 * delete a process group 453 */ 454 void 455 pgdelete(pgrp) 456 struct pgrp *pgrp; 457 { 458 459 /* Remove reference (if any) from tty to this process group */ 460 if (pgrp->pg_session->s_ttyp != NULL && 461 pgrp->pg_session->s_ttyp->t_pgrp == pgrp) 462 pgrp->pg_session->s_ttyp->t_pgrp = NULL; 463 LIST_REMOVE(pgrp, pg_hash); 464 SESSRELE(pgrp->pg_session); 465 pool_put(&pgrp_pool, pgrp); 466 } 467 468 /* 469 * Adjust pgrp jobc counters when specified process changes process group. 470 * We count the number of processes in each process group that "qualify" 471 * the group for terminal job control (those with a parent in a different 472 * process group of the same session). If that count reaches zero, the 473 * process group becomes orphaned. Check both the specified process' 474 * process group and that of its children. 475 * entering == 0 => p is leaving specified group. 476 * entering == 1 => p is entering specified group. 477 */ 478 void 479 fixjobc(p, pgrp, entering) 480 struct proc *p; 481 struct pgrp *pgrp; 482 int entering; 483 { 484 struct pgrp *hispgrp; 485 struct session *mysession = pgrp->pg_session; 486 487 /* 488 * Check p's parent to see whether p qualifies its own process 489 * group; if so, adjust count for p's process group. 490 */ 491 if ((hispgrp = p->p_pptr->p_pgrp) != pgrp && 492 hispgrp->pg_session == mysession) { 493 if (entering) 494 pgrp->pg_jobc++; 495 else if (--pgrp->pg_jobc == 0) 496 orphanpg(pgrp); 497 } 498 499 /* 500 * Check this process' children to see whether they qualify 501 * their process groups; if so, adjust counts for children's 502 * process groups. 503 */ 504 for (p = p->p_children.lh_first; p != 0; p = p->p_sibling.le_next) { 505 if ((hispgrp = p->p_pgrp) != pgrp && 506 hispgrp->pg_session == mysession && 507 P_ZOMBIE(p) == 0) { 508 if (entering) 509 hispgrp->pg_jobc++; 510 else if (--hispgrp->pg_jobc == 0) 511 orphanpg(hispgrp); 512 } 513 } 514 } 515 516 /* 517 * A process group has become orphaned; 518 * if there are any stopped processes in the group, 519 * hang-up all process in that group. 520 */ 521 static void 522 orphanpg(pg) 523 struct pgrp *pg; 524 { 525 struct proc *p; 526 527 for (p = pg->pg_members.lh_first; p != 0; p = p->p_pglist.le_next) { 528 if (p->p_stat == SSTOP) { 529 for (p = pg->pg_members.lh_first; p != 0; 530 p = p->p_pglist.le_next) { 531 psignal(p, SIGHUP); 532 psignal(p, SIGCONT); 533 } 534 return; 535 } 536 } 537 } 538 539 /* mark process as suid/sgid, reset some values do defaults */ 540 void 541 p_sugid(p) 542 struct proc *p; 543 { 544 struct plimit *newlim; 545 546 p->p_flag |= P_SUGID; 547 /* reset what needs to be reset in plimit */ 548 if (p->p_limit->pl_corename != defcorename) { 549 if (p->p_limit->p_refcnt > 1 && 550 (p->p_limit->p_lflags & PL_SHAREMOD) == 0) { 551 newlim = limcopy(p->p_limit); 552 limfree(p->p_limit); 553 p->p_limit = newlim; 554 } 555 free(p->p_limit->pl_corename, M_TEMP); 556 p->p_limit->pl_corename = defcorename; 557 } 558 } 559 560 #ifdef DEBUG 561 void 562 pgrpdump() 563 { 564 struct pgrp *pgrp; 565 struct proc *p; 566 int i; 567 568 for (i = 0; i <= pgrphash; i++) { 569 if ((pgrp = pgrphashtbl[i].lh_first) != NULL) { 570 printf("\tindx %d\n", i); 571 for (; pgrp != 0; pgrp = pgrp->pg_hash.le_next) { 572 printf("\tpgrp %p, pgid %d, sess %p, " 573 "sesscnt %d, mem %p\n", 574 pgrp, pgrp->pg_id, pgrp->pg_session, 575 pgrp->pg_session->s_count, 576 pgrp->pg_members.lh_first); 577 for (p = pgrp->pg_members.lh_first; p != 0; 578 p = p->p_pglist.le_next) { 579 printf("\t\tpid %d addr %p pgrp %p\n", 580 p->p_pid, p, p->p_pgrp); 581 } 582 } 583 } 584 } 585 } 586 #endif /* DEBUG */ 587 588 #ifdef KSTACK_CHECK_MAGIC 589 #include <sys/user.h> 590 591 #define KSTACK_MAGIC 0xdeadbeaf 592 593 /* XXX should be per process basis? */ 594 int kstackleftmin = KSTACK_SIZE; 595 int kstackleftthres = KSTACK_SIZE / 8; /* warn if remaining stack is 596 less than this */ 597 598 void 599 kstack_setup_magic(const struct proc *p) 600 { 601 u_int32_t *ip; 602 u_int32_t const *end; 603 604 KASSERT(p != 0); 605 KASSERT(p != &proc0); 606 607 /* 608 * fill all the stack with magic number 609 * so that later modification on it can be detected. 610 */ 611 ip = (u_int32_t *)KSTACK_LOWEST_ADDR(p); 612 end = (u_int32_t *)((caddr_t)KSTACK_LOWEST_ADDR(p) + KSTACK_SIZE); 613 for (; ip < end; ip++) { 614 *ip = KSTACK_MAGIC; 615 } 616 } 617 618 void 619 kstack_check_magic(const struct proc *p) 620 { 621 u_int32_t const *ip, *end; 622 int stackleft; 623 624 KASSERT(p != 0); 625 626 /* don't check proc0 */ /*XXX*/ 627 if (p == &proc0) 628 return; 629 630 #ifdef __MACHINE_STACK_GROWS_UP 631 /* stack grows upwards (eg. hppa) */ 632 ip = (u_int32_t *)((caddr_t)KSTACK_LOWEST_ADDR(p) + KSTACK_SIZE); 633 end = (u_int32_t *)KSTACK_LOWEST_ADDR(p); 634 for (ip--; ip >= end; ip--) 635 if (*ip != KSTACK_MAGIC) 636 break; 637 638 stackleft = (caddr_t)KSTACK_LOWEST_ADDR(p) + KSTACK_SIZE - (caddr_t)ip; 639 #else /* __MACHINE_STACK_GROWS_UP */ 640 /* stack grows downwards (eg. i386) */ 641 ip = (u_int32_t *)KSTACK_LOWEST_ADDR(p); 642 end = (u_int32_t *)((caddr_t)KSTACK_LOWEST_ADDR(p) + KSTACK_SIZE); 643 for (; ip < end; ip++) 644 if (*ip != KSTACK_MAGIC) 645 break; 646 647 stackleft = (caddr_t)ip - KSTACK_LOWEST_ADDR(p); 648 #endif /* __MACHINE_STACK_GROWS_UP */ 649 650 if (kstackleftmin > stackleft) { 651 kstackleftmin = stackleft; 652 if (stackleft < kstackleftthres) 653 printf("warning: kernel stack left %d bytes(pid %u)\n", 654 stackleft, p->p_pid); 655 } 656 657 if (stackleft <= 0) { 658 panic("magic on the top of kernel stack changed for pid %u: " 659 "maybe kernel stack overflow\n", p->p_pid); 660 } 661 } 662 #endif /* KSTACK_CHECK_MAGIC */ 663
Indexes created Mon Sep 22 05:09:51 GMT 2025