1 /* $NetBSD: kern_proc.c,v 1.47 2002/04/12 17:02:33 christos 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 1.47 2002/04/12 17:02:33 christos Exp $"); 77 78 #include <sys/param.h> 79 #include <sys/systm.h> 80 #include <sys/map.h> 81 #include <sys/kernel.h> 82 #include <sys/proc.h> 83 #include <sys/resourcevar.h> 84 #include <sys/buf.h> 85 #include <sys/acct.h> 86 #include <sys/wait.h> 87 #include <sys/file.h> 88 #include <ufs/ufs/quota.h> 89 #include <sys/uio.h> 90 #include <sys/malloc.h> 91 #include <sys/pool.h> 92 #include <sys/mbuf.h> 93 #include <sys/ioctl.h> 94 #include <sys/tty.h> 95 #include <sys/signalvar.h> 96 97 /* 98 * Structure associated with user cacheing. 99 */ 100 struct uidinfo { 101 LIST_ENTRY(uidinfo) ui_hash; 102 uid_t ui_uid; 103 long ui_proccnt; 104 }; 105 #define UIHASH(uid) (&uihashtbl[(uid) & uihash]) 106 LIST_HEAD(uihashhead, uidinfo) *uihashtbl; 107 u_long uihash; /* size of hash table - 1 */ 108 109 /* 110 * Other process lists 111 */ 112 struct pidhashhead *pidhashtbl; 113 u_long pidhash; 114 struct pgrphashhead *pgrphashtbl; 115 u_long pgrphash; 116 117 struct proclist allproc; 118 struct proclist zombproc; /* resources have been freed */ 119 120 /* 121 * Process list locking: 122 * 123 * We have two types of locks on the proclists: read locks and write 124 * locks. Read locks can be used in interrupt context, so while we 125 * hold the write lock, we must also block clock interrupts to 126 * lock out any scheduling changes that may happen in interrupt 127 * context. 128 * 129 * The proclist lock locks the following structures: 130 * 131 * allproc 132 * zombproc 133 * pidhashtbl 134 */ 135 struct lock proclist_lock; 136 137 /* 138 * Locking of this proclist is special; it's accessed in a 139 * critical section of process exit, and thus locking it can't 140 * modify interrupt state. We use a simple spin lock for this 141 * proclist. Processes on this proclist are also on zombproc; 142 * we use the p_hash member to linkup to deadproc. 143 */ 144 struct simplelock deadproc_slock; 145 struct proclist deadproc; /* dead, but not yet undead */ 146 147 struct pool proc_pool; 148 struct pool pcred_pool; 149 struct pool plimit_pool; 150 struct pool pgrp_pool; 151 struct pool rusage_pool; 152 153 /* 154 * The process list descriptors, used during pid allocation and 155 * by sysctl. No locking on this data structure is needed since 156 * it is completely static. 157 */ 158 const struct proclist_desc proclists[] = { 159 { &allproc }, 160 { &zombproc }, 161 { NULL }, 162 }; 163 164 static void orphanpg __P((struct pgrp *)); 165 #ifdef DEBUG 166 void pgrpdump __P((void)); 167 #endif 168 169 /* 170 * Initialize global process hashing structures. 171 */ 172 void 173 procinit() 174 { 175 const struct proclist_desc *pd; 176 177 for (pd = proclists; pd->pd_list != NULL; pd++) 178 LIST_INIT(pd->pd_list); 179 180 spinlockinit(&proclist_lock, "proclk", 0); 181 182 LIST_INIT(&deadproc); 183 simple_lock_init(&deadproc_slock); 184 185 pidhashtbl = 186 hashinit(maxproc / 4, HASH_LIST, M_PROC, M_WAITOK, &pidhash); 187 pgrphashtbl = 188 hashinit(maxproc / 4, HASH_LIST, M_PROC, M_WAITOK, &pgrphash); 189 uihashtbl = 190 hashinit(maxproc / 16, HASH_LIST, M_PROC, M_WAITOK, &uihash); 191 192 pool_init(&proc_pool, sizeof(struct proc), 0, 0, 0, "procpl", 193 &pool_allocator_nointr); 194 pool_init(&pgrp_pool, sizeof(struct pgrp), 0, 0, 0, "pgrppl", 195 &pool_allocator_nointr); 196 pool_init(&pcred_pool, sizeof(struct pcred), 0, 0, 0, "pcredpl", 197 &pool_allocator_nointr); 198 pool_init(&plimit_pool, sizeof(struct plimit), 0, 0, 0, "plimitpl", 199 &pool_allocator_nointr); 200 pool_init(&rusage_pool, sizeof(struct rusage), 0, 0, 0, "rusgepl", 201 &pool_allocator_nointr); 202 } 203 204 /* 205 * Acquire a read lock on the proclist. 206 */ 207 void 208 proclist_lock_read() 209 { 210 int error; 211 212 error = spinlockmgr(&proclist_lock, LK_SHARED, NULL); 213 #ifdef DIAGNOSTIC 214 if (__predict_false(error != 0)) 215 panic("proclist_lock_read: failed to acquire lock"); 216 #endif 217 } 218 219 /* 220 * Release a read lock on the proclist. 221 */ 222 void 223 proclist_unlock_read() 224 { 225 226 (void) spinlockmgr(&proclist_lock, LK_RELEASE, NULL); 227 } 228 229 /* 230 * Acquire a write lock on the proclist. 231 */ 232 int 233 proclist_lock_write() 234 { 235 int s, error; 236 237 s = splclock(); 238 error = spinlockmgr(&proclist_lock, LK_EXCLUSIVE, NULL); 239 #ifdef DIAGNOSTIC 240 if (__predict_false(error != 0)) 241 panic("proclist_lock: failed to acquire lock"); 242 #endif 243 return (s); 244 } 245 246 /* 247 * Release a write lock on the proclist. 248 */ 249 void 250 proclist_unlock_write(s) 251 int s; 252 { 253 254 (void) spinlockmgr(&proclist_lock, LK_RELEASE, NULL); 255 splx(s); 256 } 257 258 /* 259 * Change the count associated with number of processes 260 * a given user is using. 261 */ 262 int 263 chgproccnt(uid, diff) 264 uid_t uid; 265 int diff; 266 { 267 struct uidinfo *uip; 268 struct uihashhead *uipp; 269 270 uipp = UIHASH(uid); 271 for (uip = uipp->lh_first; uip != 0; uip = uip->ui_hash.le_next) 272 if (uip->ui_uid == uid) 273 break; 274 if (uip) { 275 uip->ui_proccnt += diff; 276 if (uip->ui_proccnt > 0) 277 return (uip->ui_proccnt); 278 if (uip->ui_proccnt < 0) 279 panic("chgproccnt: procs < 0"); 280 LIST_REMOVE(uip, ui_hash); 281 FREE(uip, M_PROC); 282 return (0); 283 } 284 if (diff <= 0) { 285 if (diff == 0) 286 return(0); 287 panic("chgproccnt: lost user"); 288 } 289 MALLOC(uip, struct uidinfo *, sizeof(*uip), M_PROC, M_WAITOK); 290 LIST_INSERT_HEAD(uipp, uip, ui_hash); 291 uip->ui_uid = uid; 292 uip->ui_proccnt = diff; 293 return (diff); 294 } 295 296 /* 297 * Is p an inferior of q? 298 */ 299 int 300 inferior(p, q) 301 struct proc *p; 302 struct proc *q; 303 { 304 305 for (; p != q; p = p->p_pptr) 306 if (p->p_pid == 0) 307 return (0); 308 return (1); 309 } 310 311 /* 312 * Locate a process by number 313 */ 314 struct proc * 315 pfind(pid) 316 pid_t pid; 317 { 318 struct proc *p; 319 320 proclist_lock_read(); 321 for (p = PIDHASH(pid)->lh_first; p != 0; p = p->p_hash.le_next) 322 if (p->p_pid == pid) 323 goto out; 324 out: 325 proclist_unlock_read(); 326 return (p); 327 } 328 329 /* 330 * Locate a process group by number 331 */ 332 struct pgrp * 333 pgfind(pgid) 334 pid_t pgid; 335 { 336 struct pgrp *pgrp; 337 338 for (pgrp = PGRPHASH(pgid)->lh_first; pgrp != 0; pgrp = pgrp->pg_hash.le_next) 339 if (pgrp->pg_id == pgid) 340 return (pgrp); 341 return (NULL); 342 } 343 344 /* 345 * Move p to a new or existing process group (and session) 346 */ 347 int 348 enterpgrp(p, pgid, mksess) 349 struct proc *p; 350 pid_t pgid; 351 int mksess; 352 { 353 struct pgrp *pgrp = pgfind(pgid); 354 355 #ifdef DIAGNOSTIC 356 if (__predict_false(pgrp != NULL && mksess)) /* firewalls */ 357 panic("enterpgrp: setsid into non-empty pgrp"); 358 if (__predict_false(SESS_LEADER(p))) 359 panic("enterpgrp: session leader attempted setpgrp"); 360 #endif 361 if (pgrp == NULL) { 362 pid_t savepid = p->p_pid; 363 struct proc *np; 364 /* 365 * new process group 366 */ 367 #ifdef DIAGNOSTIC 368 if (__predict_false(p->p_pid != pgid)) 369 panic("enterpgrp: new pgrp and pid != pgid"); 370 #endif 371 pgrp = pool_get(&pgrp_pool, PR_WAITOK); 372 if ((np = pfind(savepid)) == NULL || np != p) 373 return (ESRCH); 374 if (mksess) { 375 struct session *sess; 376 377 /* 378 * new session 379 */ 380 MALLOC(sess, struct session *, sizeof(struct session), 381 M_SESSION, M_WAITOK); 382 sess->s_sid = p->p_pid; 383 sess->s_leader = p; 384 sess->s_count = 1; 385 sess->s_ttyvp = NULL; 386 sess->s_ttyp = NULL; 387 memcpy(sess->s_login, p->p_session->s_login, 388 sizeof(sess->s_login)); 389 p->p_flag &= ~P_CONTROLT; 390 pgrp->pg_session = sess; 391 #ifdef DIAGNOSTIC 392 if (__predict_false(p != curproc)) 393 panic("enterpgrp: mksession and p != curproc"); 394 #endif 395 } else { 396 SESSHOLD(p->p_session); 397 pgrp->pg_session = p->p_session; 398 } 399 pgrp->pg_id = pgid; 400 LIST_INIT(&pgrp->pg_members); 401 LIST_INSERT_HEAD(PGRPHASH(pgid), pgrp, pg_hash); 402 pgrp->pg_jobc = 0; 403 } else if (pgrp == p->p_pgrp) 404 return (0); 405 406 /* 407 * Adjust eligibility of affected pgrps to participate in job control. 408 * Increment eligibility counts before decrementing, otherwise we 409 * could reach 0 spuriously during the first call. 410 */ 411 fixjobc(p, pgrp, 1); 412 fixjobc(p, p->p_pgrp, 0); 413 414 LIST_REMOVE(p, p_pglist); 415 if (p->p_pgrp->pg_members.lh_first == 0) 416 pgdelete(p->p_pgrp); 417 p->p_pgrp = pgrp; 418 LIST_INSERT_HEAD(&pgrp->pg_members, p, p_pglist); 419 return (0); 420 } 421 422 /* 423 * remove process from process group 424 */ 425 int 426 leavepgrp(p) 427 struct proc *p; 428 { 429 430 LIST_REMOVE(p, p_pglist); 431 if (p->p_pgrp->pg_members.lh_first == 0) 432 pgdelete(p->p_pgrp); 433 p->p_pgrp = 0; 434 return (0); 435 } 436 437 /* 438 * delete a process group 439 */ 440 void 441 pgdelete(pgrp) 442 struct pgrp *pgrp; 443 { 444 445 /* Remove reference (if any) from tty to this process group */ 446 if (pgrp->pg_session->s_ttyp != NULL && 447 pgrp->pg_session->s_ttyp->t_pgrp == pgrp) 448 pgrp->pg_session->s_ttyp->t_pgrp = NULL; 449 LIST_REMOVE(pgrp, pg_hash); 450 SESSRELE(pgrp->pg_session); 451 pool_put(&pgrp_pool, pgrp); 452 } 453 454 /* 455 * Adjust pgrp jobc counters when specified process changes process group. 456 * We count the number of processes in each process group that "qualify" 457 * the group for terminal job control (those with a parent in a different 458 * process group of the same session). If that count reaches zero, the 459 * process group becomes orphaned. Check both the specified process' 460 * process group and that of its children. 461 * entering == 0 => p is leaving specified group. 462 * entering == 1 => p is entering specified group. 463 */ 464 void 465 fixjobc(p, pgrp, entering) 466 struct proc *p; 467 struct pgrp *pgrp; 468 int entering; 469 { 470 struct pgrp *hispgrp; 471 struct session *mysession = pgrp->pg_session; 472 473 /* 474 * Check p's parent to see whether p qualifies its own process 475 * group; if so, adjust count for p's process group. 476 */ 477 if ((hispgrp = p->p_pptr->p_pgrp) != pgrp && 478 hispgrp->pg_session == mysession) { 479 if (entering) 480 pgrp->pg_jobc++; 481 else if (--pgrp->pg_jobc == 0) 482 orphanpg(pgrp); 483 } 484 485 /* 486 * Check this process' children to see whether they qualify 487 * their process groups; if so, adjust counts for children's 488 * process groups. 489 */ 490 for (p = p->p_children.lh_first; p != 0; p = p->p_sibling.le_next) { 491 if ((hispgrp = p->p_pgrp) != pgrp && 492 hispgrp->pg_session == mysession && 493 P_ZOMBIE(p) == 0) { 494 if (entering) 495 hispgrp->pg_jobc++; 496 else if (--hispgrp->pg_jobc == 0) 497 orphanpg(hispgrp); 498 } 499 } 500 } 501 502 /* 503 * A process group has become orphaned; 504 * if there are any stopped processes in the group, 505 * hang-up all process in that group. 506 */ 507 static void 508 orphanpg(pg) 509 struct pgrp *pg; 510 { 511 struct proc *p; 512 513 for (p = pg->pg_members.lh_first; p != 0; p = p->p_pglist.le_next) { 514 if (p->p_stat == SSTOP) { 515 for (p = pg->pg_members.lh_first; p != 0; 516 p = p->p_pglist.le_next) { 517 psignal(p, SIGHUP); 518 psignal(p, SIGCONT); 519 } 520 return; 521 } 522 } 523 } 524 525 /* mark process as suid/sgid, reset some values do defaults */ 526 void 527 p_sugid(p) 528 struct proc *p; 529 { 530 struct plimit *newlim; 531 532 p->p_flag |= P_SUGID; 533 /* reset what needs to be reset in plimit */ 534 if (p->p_limit->pl_corename != defcorename) { 535 if (p->p_limit->p_refcnt > 1 && 536 (p->p_limit->p_lflags & PL_SHAREMOD) == 0) { 537 newlim = limcopy(p->p_limit); 538 limfree(p->p_limit); 539 p->p_limit = newlim; 540 } else { 541 free(p->p_limit->pl_corename, M_TEMP); 542 } 543 p->p_limit->pl_corename = defcorename; 544 } 545 } 546 547 548 #ifdef DEBUG 549 void 550 pgrpdump() 551 { 552 struct pgrp *pgrp; 553 struct proc *p; 554 int i; 555 556 for (i = 0; i <= pgrphash; i++) { 557 if ((pgrp = pgrphashtbl[i].lh_first) != NULL) { 558 printf("\tindx %d\n", i); 559 for (; pgrp != 0; pgrp = pgrp->pg_hash.le_next) { 560 printf("\tpgrp %p, pgid %d, sess %p, sesscnt %d, mem %p\n", 561 pgrp, pgrp->pg_id, pgrp->pg_session, 562 pgrp->pg_session->s_count, 563 pgrp->pg_members.lh_first); 564 for (p = pgrp->pg_members.lh_first; p != 0; 565 p = p->p_pglist.le_next) { 566 printf("\t\tpid %d addr %p pgrp %p\n", 567 p->p_pid, p, p->p_pgrp); 568 } 569 } 570 } 571 } 572 } 573 #endif /* DEBUG */ 574