1 /* $NetBSD: kern_exit.c,v 1.209 2008/05/19 17:06:02 ad Exp $ */ 2 3 /*- 4 * Copyright (c) 1998, 1999, 2006, 2007, 2008 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, and by Andrew Doran. 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 * 20 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 21 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 22 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 23 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 24 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 25 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 26 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 27 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 28 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 29 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 30 * POSSIBILITY OF SUCH DAMAGE. 31 */ 32 33 /* 34 * Copyright (c) 1982, 1986, 1989, 1991, 1993 35 * The Regents of the University of California. All rights reserved. 36 * (c) UNIX System Laboratories, Inc. 37 * All or some portions of this file are derived from material licensed 38 * to the University of California by American Telephone and Telegraph 39 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 40 * the permission of UNIX System Laboratories, Inc. 41 * 42 * Redistribution and use in source and binary forms, with or without 43 * modification, are permitted provided that the following conditions 44 * are met: 45 * 1. Redistributions of source code must retain the above copyright 46 * notice, this list of conditions and the following disclaimer. 47 * 2. Redistributions in binary form must reproduce the above copyright 48 * notice, this list of conditions and the following disclaimer in the 49 * documentation and/or other materials provided with the distribution. 50 * 3. Neither the name of the University nor the names of its contributors 51 * may be used to endorse or promote products derived from this software 52 * without specific prior written permission. 53 * 54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 64 * SUCH DAMAGE. 65 * 66 * @(#)kern_exit.c 8.10 (Berkeley) 2/23/95 67 */ 68 69 #include <sys/cdefs.h> 70 __KERNEL_RCSID(0, "$NetBSD: kern_exit.c,v 1.209 2008/05/19 17:06:02 ad Exp $"); 71 72 #include "opt_ktrace.h" 73 #include "opt_perfctrs.h" 74 #include "opt_sysv.h" 75 76 #include <sys/param.h> 77 #include <sys/aio.h> 78 #include <sys/systm.h> 79 #include <sys/ioctl.h> 80 #include <sys/tty.h> 81 #include <sys/time.h> 82 #include <sys/resource.h> 83 #include <sys/kernel.h> 84 #include <sys/proc.h> 85 #include <sys/buf.h> 86 #include <sys/wait.h> 87 #include <sys/file.h> 88 #include <sys/vnode.h> 89 #include <sys/syslog.h> 90 #include <sys/malloc.h> 91 #include <sys/pool.h> 92 #include <sys/resourcevar.h> 93 #if defined(PERFCTRS) 94 #include <sys/pmc.h> 95 #endif 96 #include <sys/ptrace.h> 97 #include <sys/acct.h> 98 #include <sys/filedesc.h> 99 #include <sys/ras.h> 100 #include <sys/signalvar.h> 101 #include <sys/sched.h> 102 #include <sys/mount.h> 103 #include <sys/syscallargs.h> 104 #include <sys/kauth.h> 105 #include <sys/sleepq.h> 106 #include <sys/lockdebug.h> 107 #include <sys/ktrace.h> 108 #include <sys/cpu.h> 109 #include <sys/lwpctl.h> 110 #include <sys/atomic.h> 111 112 #include <uvm/uvm_extern.h> 113 114 #define DEBUG_EXIT 115 116 #ifdef DEBUG_EXIT 117 int debug_exit = 0; 118 #define DPRINTF(x) if (debug_exit) printf x 119 #else 120 #define DPRINTF(x) 121 #endif 122 123 static int find_stopped_child(struct proc *, pid_t, int, struct proc **, int *); 124 static void proc_free(struct proc *, struct rusage *); 125 126 /* 127 * Fill in the appropriate signal information, and signal the parent. 128 */ 129 static void 130 exit_psignal(struct proc *p, struct proc *pp, ksiginfo_t *ksi) 131 { 132 133 KSI_INIT(ksi); 134 if ((ksi->ksi_signo = P_EXITSIG(p)) == SIGCHLD) { 135 if (WIFSIGNALED(p->p_xstat)) { 136 if (WCOREDUMP(p->p_xstat)) 137 ksi->ksi_code = CLD_DUMPED; 138 else 139 ksi->ksi_code = CLD_KILLED; 140 } else { 141 ksi->ksi_code = CLD_EXITED; 142 } 143 } 144 /* 145 * We fill those in, even for non-SIGCHLD. 146 * It's safe to access p->p_cred unlocked here. 147 */ 148 ksi->ksi_pid = p->p_pid; 149 ksi->ksi_uid = kauth_cred_geteuid(p->p_cred); 150 ksi->ksi_status = p->p_xstat; 151 /* XXX: is this still valid? */ 152 ksi->ksi_utime = p->p_stats->p_ru.ru_utime.tv_sec; 153 ksi->ksi_stime = p->p_stats->p_ru.ru_stime.tv_sec; 154 } 155 156 /* 157 * exit -- 158 * Death of process. 159 */ 160 int 161 sys_exit(struct lwp *l, const struct sys_exit_args *uap, register_t *retval) 162 { 163 /* { 164 syscallarg(int) rval; 165 } */ 166 struct proc *p = l->l_proc; 167 168 /* Don't call exit1() multiple times in the same process. */ 169 mutex_enter(p->p_lock); 170 if (p->p_sflag & PS_WEXIT) { 171 mutex_exit(p->p_lock); 172 lwp_exit(l); 173 } 174 175 /* exit1() will release the mutex. */ 176 exit1(l, W_EXITCODE(SCARG(uap, rval), 0)); 177 /* NOTREACHED */ 178 return (0); 179 } 180 181 /* 182 * Exit: deallocate address space and other resources, change proc state 183 * to zombie, and unlink proc from allproc and parent's lists. Save exit 184 * status and rusage for wait(). Check for child processes and orphan them. 185 * 186 * Must be called with p->p_lock held. Does not return. 187 */ 188 void 189 exit1(struct lwp *l, int rv) 190 { 191 struct proc *p, *q, *nq; 192 struct pgrp *pgrp; 193 ksiginfo_t ksi; 194 ksiginfoq_t kq; 195 int wakeinit; 196 197 p = l->l_proc; 198 199 KASSERT(mutex_owned(p->p_lock)); 200 201 if (__predict_false(p == initproc)) 202 panic("init died (signal %d, exit %d)", 203 WTERMSIG(rv), WEXITSTATUS(rv)); 204 205 p->p_sflag |= PS_WEXIT; 206 207 /* 208 * Force all other LWPs to exit before we do. Only then can we 209 * begin to tear down the rest of the process state. 210 */ 211 if (p->p_nlwps > 1) 212 exit_lwps(l); 213 214 ksiginfo_queue_init(&kq); 215 216 /* 217 * If we have been asked to stop on exit, do so now. 218 */ 219 if (p->p_sflag & PS_STOPEXIT) { 220 KERNEL_UNLOCK_ALL(l, &l->l_biglocks); 221 sigclearall(p, &contsigmask, &kq); 222 p->p_waited = 0; 223 membar_producer(); 224 p->p_stat = SSTOP; 225 lwp_lock(l); 226 p->p_nrlwps--; 227 l->l_stat = LSSTOP; 228 mutex_exit(p->p_lock); 229 mi_switch(l); 230 KERNEL_LOCK(l->l_biglocks, l); 231 } else 232 mutex_exit(p->p_lock); 233 234 /* Destroy any lwpctl info. */ 235 if (p->p_lwpctl != NULL) 236 lwp_ctl_exit(); 237 238 /* Destroy all AIO works */ 239 aio_exit(p, p->p_aio); 240 241 /* 242 * Drain all remaining references that procfs, ptrace and others may 243 * have on the process. 244 */ 245 rw_enter(&p->p_reflock, RW_WRITER); 246 247 /* 248 * Bin any remaining signals and mark the process as dying so it will 249 * not be found for, e.g. signals. 250 */ 251 mutex_enter(p->p_lock); 252 sigfillset(&p->p_sigctx.ps_sigignore); 253 sigclearall(p, NULL, &kq); 254 p->p_stat = SDYING; 255 mutex_exit(p->p_lock); 256 ksiginfo_queue_drain(&kq); 257 258 DPRINTF(("exit1: %d.%d exiting.\n", p->p_pid, l->l_lid)); 259 260 #ifdef PGINPROF 261 vmsizmon(); 262 #endif 263 timers_free(p, TIMERS_ALL); 264 #if defined(__HAVE_RAS) 265 ras_purgeall(); 266 #endif 267 268 /* 269 * Close open files, release open-file table and free signal 270 * actions. This may block! 271 */ 272 fd_free(); 273 cwdfree(p->p_cwdi); 274 p->p_cwdi = NULL; 275 doexithooks(p); 276 sigactsfree(p->p_sigacts); 277 278 /* 279 * Write out accounting data. 280 */ 281 (void)acct_process(l); 282 283 #ifdef KTRACE 284 /* 285 * Release trace file. 286 */ 287 if (p->p_tracep != NULL) { 288 mutex_enter(&ktrace_lock); 289 ktrderef(p); 290 mutex_exit(&ktrace_lock); 291 } 292 #endif 293 294 /* 295 * If emulation has process exit hook, call it now. 296 * Set the exit status now so that the exit hook has 297 * an opportunity to tweak it (COMPAT_LINUX requires 298 * this for thread group emulation) 299 */ 300 p->p_xstat = rv; 301 if (p->p_emul->e_proc_exit) 302 (*p->p_emul->e_proc_exit)(p); 303 304 /* 305 * Free the VM resources we're still holding on to. 306 * We must do this from a valid thread because doing 307 * so may block. This frees vmspace, which we don't 308 * need anymore. The only remaining lwp is the one 309 * we run at this moment, nothing runs in userland 310 * anymore. 311 */ 312 uvm_proc_exit(p); 313 314 /* 315 * While we can still block, and mark the LWP as unswappable to 316 * prevent conflicts with the with the swapper. We also shouldn't 317 * be swapped out, because we are about to exit and will release 318 * memory. 319 */ 320 uvm_lwp_hold(l); 321 322 /* 323 * Stop profiling. 324 */ 325 if ((p->p_stflag & PST_PROFIL) != 0) { 326 mutex_spin_enter(&p->p_stmutex); 327 stopprofclock(p); 328 mutex_spin_exit(&p->p_stmutex); 329 } 330 331 /* 332 * If parent is waiting for us to exit or exec, P_PPWAIT is set; we 333 * wake up the parent early to avoid deadlock. We can do this once 334 * the VM resources are released. 335 */ 336 mutex_enter(proc_lock); 337 mutex_enter(p->p_lock); 338 if (p->p_sflag & PS_PPWAIT) { 339 p->p_sflag &= ~PS_PPWAIT; 340 cv_broadcast(&p->p_pptr->p_waitcv); 341 } 342 mutex_exit(p->p_lock); 343 344 if (SESS_LEADER(p)) { 345 struct vnode *vprele = NULL, *vprevoke = NULL; 346 struct session *sp = p->p_session; 347 struct tty *tp; 348 349 if (sp->s_ttyvp) { 350 /* 351 * Controlling process. 352 * Signal foreground pgrp, 353 * drain controlling terminal 354 * and revoke access to controlling terminal. 355 */ 356 tp = sp->s_ttyp; 357 mutex_spin_enter(&tty_lock); 358 if (tp->t_session == sp) { 359 /* we can't guarantee the revoke will do this */ 360 pgrp = tp->t_pgrp; 361 tp->t_pgrp = NULL; 362 tp->t_session = NULL; 363 mutex_spin_exit(&tty_lock); 364 if (pgrp != NULL) { 365 pgsignal(pgrp, SIGHUP, 1); 366 } 367 mutex_exit(proc_lock); 368 (void) ttywait(tp); 369 mutex_enter(proc_lock); 370 371 /* The tty could have been revoked. */ 372 vprevoke = sp->s_ttyvp; 373 } else 374 mutex_spin_exit(&tty_lock); 375 vprele = sp->s_ttyvp; 376 sp->s_ttyvp = NULL; 377 /* 378 * s_ttyp is not zero'd; we use this to indicate 379 * that the session once had a controlling terminal. 380 * (for logging and informational purposes) 381 */ 382 } 383 sp->s_leader = NULL; 384 385 if (vprevoke != NULL || vprele != NULL) { 386 if (vprevoke != NULL) { 387 SESSRELE(sp); 388 mutex_exit(proc_lock); 389 VOP_REVOKE(vprevoke, REVOKEALL); 390 } else 391 mutex_exit(proc_lock); 392 if (vprele != NULL) 393 vrele(vprele); 394 mutex_enter(proc_lock); 395 } 396 } 397 fixjobc(p, p->p_pgrp, 0); 398 399 /* 400 * Finalize the last LWP's specificdata, as well as the 401 * specificdata for the proc itself. 402 */ 403 lwp_finispecific(l); 404 proc_finispecific(p); 405 406 /* 407 * Notify interested parties of our demise. 408 */ 409 KNOTE(&p->p_klist, NOTE_EXIT); 410 411 412 413 #if PERFCTRS 414 /* 415 * Save final PMC information in parent process & clean up. 416 */ 417 if (PMC_ENABLED(p)) { 418 pmc_save_context(p); 419 pmc_accumulate(p->p_pptr, p); 420 pmc_process_exit(p); 421 } 422 #endif 423 424 /* 425 * Reset p_opptr pointer of all former children which got 426 * traced by another process and were reparented. We reset 427 * it to NULL here; the trace detach code then reparents 428 * the child to initproc. We only check allproc list, since 429 * eventual former children on zombproc list won't reference 430 * p_opptr anymore. 431 */ 432 if (__predict_false(p->p_slflag & PSL_CHTRACED)) { 433 PROCLIST_FOREACH(q, &allproc) { 434 if ((q->p_flag & PK_MARKER) != 0) 435 continue; 436 if (q->p_opptr == p) 437 q->p_opptr = NULL; 438 } 439 } 440 441 /* 442 * Give orphaned children to init(8). 443 */ 444 q = LIST_FIRST(&p->p_children); 445 wakeinit = (q != NULL); 446 for (; q != NULL; q = nq) { 447 nq = LIST_NEXT(q, p_sibling); 448 449 /* 450 * Traced processes are killed since their existence 451 * means someone is screwing up. Since we reset the 452 * trace flags, the logic in sys_wait4() would not be 453 * triggered to reparent the process to its 454 * original parent, so we must do this here. 455 */ 456 if (q->p_slflag & PSL_TRACED) { 457 mutex_enter(p->p_lock); 458 q->p_slflag &= ~(PSL_TRACED|PSL_FSTRACE|PSL_SYSCALL); 459 mutex_exit(p->p_lock); 460 if (q->p_opptr != q->p_pptr) { 461 struct proc *t = q->p_opptr; 462 proc_reparent(q, t ? t : initproc); 463 q->p_opptr = NULL; 464 } else 465 proc_reparent(q, initproc); 466 killproc(q, "orphaned traced process"); 467 } else 468 proc_reparent(q, initproc); 469 } 470 471 /* 472 * Move proc from allproc to zombproc, it's now nearly ready to be 473 * collected by parent. 474 */ 475 LIST_REMOVE(l, l_list); 476 LIST_REMOVE(p, p_list); 477 LIST_INSERT_HEAD(&zombproc, p, p_list); 478 479 /* 480 * Mark the process as dead. We must do this before we signal 481 * the parent. 482 */ 483 p->p_stat = SDEAD; 484 485 /* Put in front of parent's sibling list for parent to collect it */ 486 q = p->p_pptr; 487 q->p_nstopchild++; 488 if (LIST_FIRST(&q->p_children) != p) { 489 /* Put child where it can be found quickly */ 490 LIST_REMOVE(p, p_sibling); 491 LIST_INSERT_HEAD(&q->p_children, p, p_sibling); 492 } 493 494 /* 495 * Notify parent that we're gone. If parent has the P_NOCLDWAIT 496 * flag set, notify init instead (and hope it will handle 497 * this situation). 498 */ 499 mutex_enter(q->p_lock); 500 if (q->p_flag & (PK_NOCLDWAIT|PK_CLDSIGIGN)) { 501 proc_reparent(p, initproc); 502 wakeinit = 1; 503 504 /* 505 * If this was the last child of our parent, notify 506 * parent, so in case he was wait(2)ing, he will 507 * continue. 508 */ 509 if (LIST_FIRST(&q->p_children) == NULL) 510 cv_broadcast(&q->p_waitcv); 511 } 512 mutex_exit(q->p_lock); 513 514 /* Reload parent pointer, since p may have been reparented above */ 515 q = p->p_pptr; 516 517 if ((p->p_slflag & PSL_FSTRACE) == 0 && p->p_exitsig != 0) { 518 exit_psignal(p, q, &ksi); 519 kpsignal(q, &ksi, NULL); 520 } 521 522 /* Calculate the final rusage info. */ 523 calcru(p, &p->p_stats->p_ru.ru_utime, &p->p_stats->p_ru.ru_stime, 524 NULL, NULL); 525 526 if (wakeinit) 527 cv_broadcast(&initproc->p_waitcv); 528 529 callout_destroy(&l->l_timeout_ch); 530 531 /* 532 * Remaining lwp resources will be freed in lwp_exit2() once we've 533 * switch to idle context; at that point, we will be marked as a 534 * full blown zombie. 535 */ 536 mutex_enter(p->p_lock); 537 lwp_drainrefs(l); 538 lwp_lock(l); 539 l->l_prflag &= ~LPR_DETACHED; 540 l->l_stat = LSZOMB; 541 lwp_unlock(l); 542 KASSERT(curlwp == l); 543 KASSERT(p->p_nrlwps == 1); 544 KASSERT(p->p_nlwps == 1); 545 p->p_stat = SZOMB; 546 p->p_nrlwps--; 547 p->p_nzlwps++; 548 p->p_ndlwps = 0; 549 mutex_exit(p->p_lock); 550 551 /* 552 * Signal the parent to collect us, and drop the proclist lock. 553 * Drop debugger/procfs lock; no new references can be gained. 554 */ 555 cv_broadcast(&p->p_pptr->p_waitcv); 556 mutex_exit(proc_lock); 557 rw_exit(&p->p_reflock); 558 559 /* Verify that we hold no locks other than the kernel lock. */ 560 LOCKDEBUG_BARRIER(&kernel_lock, 0); 561 562 /* 563 * NOTE: WE ARE NO LONGER ALLOWED TO SLEEP! 564 */ 565 566 /* 567 * Give machine-dependent code a chance to free any MD LWP 568 * resources. This must be done before uvm_lwp_exit(), in 569 * case these resources are in the PCB. 570 */ 571 #ifndef __NO_CPU_LWP_FREE 572 cpu_lwp_free(l, 1); 573 #endif 574 pmap_deactivate(l); 575 576 /* This process no longer needs to hold the kernel lock. */ 577 #ifdef notyet 578 /* XXXSMP hold in lwp_userret() */ 579 KERNEL_UNLOCK_LAST(l); 580 #else 581 KERNEL_UNLOCK_ALL(l, NULL); 582 #endif 583 584 lwp_exit_switchaway(l); 585 } 586 587 void 588 exit_lwps(struct lwp *l) 589 { 590 struct proc *p; 591 struct lwp *l2; 592 int error; 593 lwpid_t waited; 594 int nlocks; 595 596 KERNEL_UNLOCK_ALL(l, &nlocks); 597 598 p = l->l_proc; 599 KASSERT(mutex_owned(p->p_lock)); 600 601 retry: 602 /* 603 * Interrupt LWPs in interruptable sleep, unsuspend suspended 604 * LWPs and then wait for everyone else to finish. 605 */ 606 LIST_FOREACH(l2, &p->p_lwps, l_sibling) { 607 if (l2 == l) 608 continue; 609 lwp_lock(l2); 610 l2->l_flag |= LW_WEXIT; 611 if ((l2->l_stat == LSSLEEP && (l2->l_flag & LW_SINTR)) || 612 l2->l_stat == LSSUSPENDED || l2->l_stat == LSSTOP) { 613 /* setrunnable() will release the lock. */ 614 setrunnable(l2); 615 DPRINTF(("exit_lwps: Made %d.%d runnable\n", 616 p->p_pid, l2->l_lid)); 617 continue; 618 } 619 lwp_unlock(l2); 620 } 621 while (p->p_nlwps > 1) { 622 DPRINTF(("exit_lwps: waiting for %d LWPs (%d zombies)\n", 623 p->p_nlwps, p->p_nzlwps)); 624 error = lwp_wait1(l, 0, &waited, LWPWAIT_EXITCONTROL); 625 if (p->p_nlwps == 1) 626 break; 627 if (error == EDEADLK) { 628 /* 629 * LWPs can get suspended/slept behind us. 630 * (eg. sa_setwoken) 631 * kick them again and retry. 632 */ 633 goto retry; 634 } 635 if (error) 636 panic("exit_lwps: lwp_wait1 failed with error %d", 637 error); 638 DPRINTF(("exit_lwps: Got LWP %d from lwp_wait1()\n", waited)); 639 } 640 641 KERNEL_LOCK(nlocks, l); 642 KASSERT(p->p_nlwps == 1); 643 } 644 645 int 646 do_sys_wait(struct lwp *l, int *pid, int *status, int options, 647 struct rusage *ru, int *was_zombie) 648 { 649 struct proc *child; 650 int error; 651 652 mutex_enter(proc_lock); 653 error = find_stopped_child(l->l_proc, *pid, options, &child, status); 654 655 if (child == NULL) { 656 mutex_exit(proc_lock); 657 *pid = 0; 658 return error; 659 } 660 661 *pid = child->p_pid; 662 663 if (child->p_stat == SZOMB) { 664 /* proc_free() will release the proc_lock. */ 665 *was_zombie = 1; 666 if (options & WNOWAIT) 667 mutex_exit(proc_lock); 668 else { 669 proc_free(child, ru); 670 } 671 } else { 672 /* Child state must have been SSTOP. */ 673 *was_zombie = 0; 674 mutex_exit(proc_lock); 675 *status = W_STOPCODE(*status); 676 } 677 678 return 0; 679 } 680 681 int 682 sys_wait4(struct lwp *l, const struct sys_wait4_args *uap, register_t *retval) 683 { 684 /* { 685 syscallarg(int) pid; 686 syscallarg(int *) status; 687 syscallarg(int) options; 688 syscallarg(struct rusage *) rusage; 689 } */ 690 int status, error; 691 int was_zombie; 692 struct rusage ru; 693 int pid = SCARG(uap, pid); 694 695 error = do_sys_wait(l, &pid, &status, SCARG(uap, options), 696 SCARG(uap, rusage) != NULL ? &ru : NULL, &was_zombie); 697 698 retval[0] = pid; 699 if (pid == 0) 700 return error; 701 702 if (SCARG(uap, rusage)) 703 error = copyout(&ru, SCARG(uap, rusage), sizeof(ru)); 704 705 if (error == 0 && SCARG(uap, status)) 706 error = copyout(&status, SCARG(uap, status), sizeof(status)); 707 708 return error; 709 } 710 711 /* 712 * Scan list of child processes for a child process that has stopped or 713 * exited. Used by sys_wait4 and 'compat' equivalents. 714 * 715 * Must be called with the proc_lock held, and may release while waiting. 716 */ 717 static int 718 find_stopped_child(struct proc *parent, pid_t pid, int options, 719 struct proc **child_p, int *status_p) 720 { 721 struct proc *child, *dead; 722 int error; 723 724 KASSERT(mutex_owned(proc_lock)); 725 726 if (options & ~(WUNTRACED|WNOHANG|WALTSIG|WALLSIG) 727 && !(options & WOPTSCHECKED)) { 728 *child_p = NULL; 729 return EINVAL; 730 } 731 732 if (pid == 0 && !(options & WOPTSCHECKED)) 733 pid = -parent->p_pgid; 734 735 for (;;) { 736 error = ECHILD; 737 dead = NULL; 738 739 LIST_FOREACH(child, &parent->p_children, p_sibling) { 740 if (pid >= 0) { 741 if (child->p_pid != pid) { 742 child = p_find(pid, PFIND_ZOMBIE | 743 PFIND_LOCKED); 744 if (child == NULL || 745 child->p_pptr != parent) { 746 child = NULL; 747 break; 748 } 749 } 750 } else if (pid != WAIT_ANY && child->p_pgid != -pid) { 751 /* Child not in correct pgrp */ 752 continue; 753 } 754 755 /* 756 * Wait for processes with p_exitsig != SIGCHLD 757 * processes only if WALTSIG is set; wait for 758 * processes with p_exitsig == SIGCHLD only 759 * if WALTSIG is clear. 760 */ 761 if (((options & WALLSIG) == 0) && 762 (options & WALTSIG ? child->p_exitsig == SIGCHLD 763 : P_EXITSIG(child) != SIGCHLD)){ 764 if (child->p_pid == pid) { 765 child = NULL; 766 break; 767 } 768 continue; 769 } 770 771 error = 0; 772 if ((options & WNOZOMBIE) == 0) { 773 if (child->p_stat == SZOMB) 774 break; 775 if (child->p_stat == SDEAD) { 776 /* 777 * We may occasionally arrive here 778 * after receiving a signal, but 779 * immediatley before the child 780 * process is zombified. The wait 781 * will be short, so avoid returning 782 * to userspace. 783 */ 784 dead = child; 785 } 786 } 787 788 if (child->p_stat == SSTOP && 789 child->p_waited == 0 && 790 (child->p_slflag & PSL_TRACED || 791 options & WUNTRACED)) { 792 if ((options & WNOWAIT) == 0) { 793 child->p_waited = 1; 794 parent->p_nstopchild--; 795 } 796 break; 797 } 798 if (parent->p_nstopchild == 0 || child->p_pid == pid) { 799 child = NULL; 800 break; 801 } 802 } 803 804 if (child != NULL || error != 0 || 805 ((options & WNOHANG) != 0 && dead == NULL)) { 806 if (child != NULL) { 807 *status_p = child->p_xstat; 808 } 809 *child_p = child; 810 return error; 811 } 812 813 /* 814 * Wait for another child process to stop. 815 */ 816 error = cv_wait_sig(&parent->p_waitcv, proc_lock); 817 818 if (error != 0) { 819 *child_p = NULL; 820 return error; 821 } 822 } 823 } 824 825 /* 826 * Free a process after parent has taken all the state info. Must be called 827 * with the proclist lock held, and will release before returning. 828 * 829 * *ru is returned to the caller, and must be freed by the caller. 830 */ 831 static void 832 proc_free(struct proc *p, struct rusage *ru) 833 { 834 struct proc *parent; 835 struct lwp *l; 836 ksiginfo_t ksi; 837 kauth_cred_t cred1, cred2; 838 uid_t uid; 839 840 KASSERT(mutex_owned(proc_lock)); 841 KASSERT(p->p_nlwps == 1); 842 KASSERT(p->p_nzlwps == 1); 843 KASSERT(p->p_nrlwps == 0); 844 KASSERT(p->p_stat == SZOMB); 845 846 /* 847 * If we got the child via ptrace(2) or procfs, and 848 * the parent is different (meaning the process was 849 * attached, rather than run as a child), then we need 850 * to give it back to the old parent, and send the 851 * parent the exit signal. The rest of the cleanup 852 * will be done when the old parent waits on the child. 853 */ 854 if ((p->p_slflag & PSL_TRACED) != 0) { 855 parent = p->p_pptr; 856 if (p->p_opptr != parent){ 857 mutex_enter(p->p_lock); 858 p->p_slflag &= ~(PSL_TRACED|PSL_FSTRACE|PSL_SYSCALL); 859 mutex_exit(p->p_lock); 860 parent = p->p_opptr; 861 if (parent == NULL) 862 parent = initproc; 863 proc_reparent(p, parent); 864 p->p_opptr = NULL; 865 if (p->p_exitsig != 0) { 866 exit_psignal(p, parent, &ksi); 867 kpsignal(parent, &ksi, NULL); 868 } 869 cv_broadcast(&parent->p_waitcv); 870 mutex_exit(proc_lock); 871 return; 872 } 873 } 874 875 /* 876 * Finally finished with old proc entry. Unlink it from its process 877 * group. 878 */ 879 leavepgrp(p); 880 881 parent = p->p_pptr; 882 sched_proc_exit(parent, p); 883 884 /* 885 * Add child times of exiting process onto its own times. 886 * This cannot be done any earlier else it might get done twice. 887 */ 888 l = LIST_FIRST(&p->p_lwps); 889 p->p_stats->p_ru.ru_nvcsw += (l->l_ncsw - l->l_nivcsw); 890 p->p_stats->p_ru.ru_nivcsw += l->l_nivcsw; 891 ruadd(&p->p_stats->p_ru, &l->l_ru); 892 ruadd(&p->p_stats->p_ru, &p->p_stats->p_cru); 893 ruadd(&parent->p_stats->p_cru, &p->p_stats->p_ru); 894 if (ru != NULL) 895 *ru = p->p_stats->p_ru; 896 p->p_xstat = 0; 897 898 /* 899 * At this point we are going to start freeing the final resources. 900 * If anyone tries to access the proc structure after here they will 901 * get a shock - bits are missing. Attempt to make it hard! We 902 * don't bother with any further locking past this point. 903 */ 904 p->p_stat = SIDL; /* not even a zombie any more */ 905 LIST_REMOVE(p, p_list); /* off zombproc */ 906 parent = p->p_pptr; 907 p->p_pptr->p_nstopchild--; 908 LIST_REMOVE(p, p_sibling); 909 910 /* 911 * Let pid be reallocated. 912 */ 913 proc_free_pid(p); 914 mutex_exit(proc_lock); 915 916 /* 917 * Delay release until after lwp_free. 918 */ 919 cred2 = l->l_cred; 920 921 /* 922 * Free the last LWP's resources. 923 * 924 * lwp_free ensures the LWP is no longer running on another CPU. 925 */ 926 lwp_free(l, false, true); 927 928 /* 929 * Now no one except us can reach the process p. 930 */ 931 932 /* 933 * Decrement the count of procs running with this uid. 934 */ 935 cred1 = p->p_cred; 936 uid = kauth_cred_getuid(cred1); 937 (void)chgproccnt(uid, -1); 938 939 /* 940 * Release substructures. 941 */ 942 943 limfree(p->p_limit); 944 pstatsfree(p->p_stats); 945 kauth_cred_free(cred1); 946 kauth_cred_free(cred2); 947 948 /* 949 * Release reference to text vnode 950 */ 951 if (p->p_textvp) 952 vrele(p->p_textvp); 953 954 mutex_destroy(&p->p_auxlock); 955 mutex_obj_free(p->p_lock); 956 mutex_destroy(&p->p_stmutex); 957 cv_destroy(&p->p_waitcv); 958 cv_destroy(&p->p_lwpcv); 959 rw_destroy(&p->p_reflock); 960 961 proc_free_mem(p); 962 } 963 964 /* 965 * make process 'parent' the new parent of process 'child'. 966 * 967 * Must be called with proc_lock held. 968 */ 969 void 970 proc_reparent(struct proc *child, struct proc *parent) 971 { 972 973 KASSERT(mutex_owned(proc_lock)); 974 975 if (child->p_pptr == parent) 976 return; 977 978 if (child->p_stat == SZOMB || 979 (child->p_stat == SSTOP && !child->p_waited)) { 980 child->p_pptr->p_nstopchild--; 981 parent->p_nstopchild++; 982 } 983 if (parent == initproc) 984 child->p_exitsig = SIGCHLD; 985 986 LIST_REMOVE(child, p_sibling); 987 LIST_INSERT_HEAD(&parent->p_children, child, p_sibling); 988 child->p_pptr = parent; 989 } 990