1 /*- 2 * Copyright (C) 2001 Julian Elischer <julian@freebsd.org>. 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice(s), this list of conditions and the following disclaimer as 10 * the first lines of this file unmodified other than the possible 11 * addition of one or more copyright notices. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice(s), this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY 17 * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED 18 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE 19 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY 20 * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 21 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR 22 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER 23 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 24 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH 26 * DAMAGE. 27 */ 28 29 #include "opt_witness.h" 30 #include "opt_hwpmc_hooks.h" 31 32 #include <sys/cdefs.h> 33 __FBSDID("$FreeBSD$"); 34 35 #include <sys/param.h> 36 #include <sys/systm.h> 37 #include <sys/kernel.h> 38 #include <sys/lock.h> 39 #include <sys/mutex.h> 40 #include <sys/proc.h> 41 #include <sys/rangelock.h> 42 #include <sys/resourcevar.h> 43 #include <sys/sdt.h> 44 #include <sys/smp.h> 45 #include <sys/sched.h> 46 #include <sys/sleepqueue.h> 47 #include <sys/selinfo.h> 48 #include <sys/syscallsubr.h> 49 #include <sys/sysent.h> 50 #include <sys/turnstile.h> 51 #include <sys/ktr.h> 52 #include <sys/rwlock.h> 53 #include <sys/umtx.h> 54 #include <sys/cpuset.h> 55 #ifdef HWPMC_HOOKS 56 #include <sys/pmckern.h> 57 #endif 58 59 #include <security/audit/audit.h> 60 61 #include <vm/vm.h> 62 #include <vm/vm_extern.h> 63 #include <vm/uma.h> 64 #include <vm/vm_domain.h> 65 #include <sys/eventhandler.h> 66 67 SDT_PROVIDER_DECLARE(proc); 68 SDT_PROBE_DEFINE(proc, , , lwp__exit); 69 70 /* 71 * thread related storage. 72 */ 73 static uma_zone_t thread_zone; 74 75 TAILQ_HEAD(, thread) zombie_threads = TAILQ_HEAD_INITIALIZER(zombie_threads); 76 static struct mtx zombie_lock; 77 MTX_SYSINIT(zombie_lock, &zombie_lock, "zombie lock", MTX_SPIN); 78 79 static void thread_zombie(struct thread *); 80 static int thread_unsuspend_one(struct thread *td, struct proc *p, 81 bool boundary); 82 83 #define TID_BUFFER_SIZE 1024 84 85 struct mtx tid_lock; 86 static struct unrhdr *tid_unrhdr; 87 static lwpid_t tid_buffer[TID_BUFFER_SIZE]; 88 static int tid_head, tid_tail; 89 static MALLOC_DEFINE(M_TIDHASH, "tidhash", "thread hash"); 90 91 struct tidhashhead *tidhashtbl; 92 u_long tidhash; 93 struct rwlock tidhash_lock; 94 95 static lwpid_t 96 tid_alloc(void) 97 { 98 lwpid_t tid; 99 100 tid = alloc_unr(tid_unrhdr); 101 if (tid != -1) 102 return (tid); 103 mtx_lock(&tid_lock); 104 if (tid_head == tid_tail) { 105 mtx_unlock(&tid_lock); 106 return (-1); 107 } 108 tid = tid_buffer[tid_head]; 109 tid_head = (tid_head + 1) % TID_BUFFER_SIZE; 110 mtx_unlock(&tid_lock); 111 return (tid); 112 } 113 114 static void 115 tid_free(lwpid_t tid) 116 { 117 lwpid_t tmp_tid = -1; 118 119 mtx_lock(&tid_lock); 120 if ((tid_tail + 1) % TID_BUFFER_SIZE == tid_head) { 121 tmp_tid = tid_buffer[tid_head]; 122 tid_head = (tid_head + 1) % TID_BUFFER_SIZE; 123 } 124 tid_buffer[tid_tail] = tid; 125 tid_tail = (tid_tail + 1) % TID_BUFFER_SIZE; 126 mtx_unlock(&tid_lock); 127 if (tmp_tid != -1) 128 free_unr(tid_unrhdr, tmp_tid); 129 } 130 131 /* 132 * Prepare a thread for use. 133 */ 134 static int 135 thread_ctor(void *mem, int size, void *arg, int flags) 136 { 137 struct thread *td; 138 139 td = (struct thread *)mem; 140 td->td_state = TDS_INACTIVE; 141 td->td_oncpu = NOCPU; 142 143 td->td_tid = tid_alloc(); 144 145 /* 146 * Note that td_critnest begins life as 1 because the thread is not 147 * running and is thereby implicitly waiting to be on the receiving 148 * end of a context switch. 149 */ 150 td->td_critnest = 1; 151 td->td_lend_user_pri = PRI_MAX; 152 EVENTHANDLER_INVOKE(thread_ctor, td); 153 #ifdef AUDIT 154 audit_thread_alloc(td); 155 #endif 156 umtx_thread_alloc(td); 157 return (0); 158 } 159 160 /* 161 * Reclaim a thread after use. 162 */ 163 static void 164 thread_dtor(void *mem, int size, void *arg) 165 { 166 struct thread *td; 167 168 td = (struct thread *)mem; 169 170 #ifdef INVARIANTS 171 /* Verify that this thread is in a safe state to free. */ 172 switch (td->td_state) { 173 case TDS_INHIBITED: 174 case TDS_RUNNING: 175 case TDS_CAN_RUN: 176 case TDS_RUNQ: 177 /* 178 * We must never unlink a thread that is in one of 179 * these states, because it is currently active. 180 */ 181 panic("bad state for thread unlinking"); 182 /* NOTREACHED */ 183 case TDS_INACTIVE: 184 break; 185 default: 186 panic("bad thread state"); 187 /* NOTREACHED */ 188 } 189 #endif 190 #ifdef AUDIT 191 audit_thread_free(td); 192 #endif 193 /* Free all OSD associated to this thread. */ 194 osd_thread_exit(td); 195 196 EVENTHANDLER_INVOKE(thread_dtor, td); 197 tid_free(td->td_tid); 198 } 199 200 /* 201 * Initialize type-stable parts of a thread (when newly created). 202 */ 203 static int 204 thread_init(void *mem, int size, int flags) 205 { 206 struct thread *td; 207 208 td = (struct thread *)mem; 209 210 td->td_sleepqueue = sleepq_alloc(); 211 td->td_turnstile = turnstile_alloc(); 212 td->td_rlqe = NULL; 213 EVENTHANDLER_INVOKE(thread_init, td); 214 td->td_sched = (struct td_sched *)&td[1]; 215 umtx_thread_init(td); 216 td->td_kstack = 0; 217 td->td_sel = NULL; 218 return (0); 219 } 220 221 /* 222 * Tear down type-stable parts of a thread (just before being discarded). 223 */ 224 static void 225 thread_fini(void *mem, int size) 226 { 227 struct thread *td; 228 229 td = (struct thread *)mem; 230 EVENTHANDLER_INVOKE(thread_fini, td); 231 rlqentry_free(td->td_rlqe); 232 turnstile_free(td->td_turnstile); 233 sleepq_free(td->td_sleepqueue); 234 umtx_thread_fini(td); 235 seltdfini(td); 236 } 237 238 /* 239 * For a newly created process, 240 * link up all the structures and its initial threads etc. 241 * called from: 242 * {arch}/{arch}/machdep.c {arch}_init(), init386() etc. 243 * proc_dtor() (should go away) 244 * proc_init() 245 */ 246 void 247 proc_linkup0(struct proc *p, struct thread *td) 248 { 249 TAILQ_INIT(&p->p_threads); /* all threads in proc */ 250 proc_linkup(p, td); 251 } 252 253 void 254 proc_linkup(struct proc *p, struct thread *td) 255 { 256 257 sigqueue_init(&p->p_sigqueue, p); 258 p->p_ksi = ksiginfo_alloc(1); 259 if (p->p_ksi != NULL) { 260 /* XXX p_ksi may be null if ksiginfo zone is not ready */ 261 p->p_ksi->ksi_flags = KSI_EXT | KSI_INS; 262 } 263 LIST_INIT(&p->p_mqnotifier); 264 p->p_numthreads = 0; 265 thread_link(td, p); 266 } 267 268 /* 269 * Initialize global thread allocation resources. 270 */ 271 void 272 threadinit(void) 273 { 274 275 mtx_init(&tid_lock, "TID lock", NULL, MTX_DEF); 276 277 /* 278 * pid_max cannot be greater than PID_MAX. 279 * leave one number for thread0. 280 */ 281 tid_unrhdr = new_unrhdr(PID_MAX + 2, INT_MAX, &tid_lock); 282 283 thread_zone = uma_zcreate("THREAD", sched_sizeof_thread(), 284 thread_ctor, thread_dtor, thread_init, thread_fini, 285 16 - 1, UMA_ZONE_NOFREE); 286 tidhashtbl = hashinit(maxproc / 2, M_TIDHASH, &tidhash); 287 rw_init(&tidhash_lock, "tidhash"); 288 } 289 290 /* 291 * Place an unused thread on the zombie list. 292 * Use the slpq as that must be unused by now. 293 */ 294 void 295 thread_zombie(struct thread *td) 296 { 297 mtx_lock_spin(&zombie_lock); 298 TAILQ_INSERT_HEAD(&zombie_threads, td, td_slpq); 299 mtx_unlock_spin(&zombie_lock); 300 } 301 302 /* 303 * Release a thread that has exited after cpu_throw(). 304 */ 305 void 306 thread_stash(struct thread *td) 307 { 308 atomic_subtract_rel_int(&td->td_proc->p_exitthreads, 1); 309 thread_zombie(td); 310 } 311 312 /* 313 * Reap zombie resources. 314 */ 315 void 316 thread_reap(void) 317 { 318 struct thread *td_first, *td_next; 319 320 /* 321 * Don't even bother to lock if none at this instant, 322 * we really don't care about the next instant.. 323 */ 324 if (!TAILQ_EMPTY(&zombie_threads)) { 325 mtx_lock_spin(&zombie_lock); 326 td_first = TAILQ_FIRST(&zombie_threads); 327 if (td_first) 328 TAILQ_INIT(&zombie_threads); 329 mtx_unlock_spin(&zombie_lock); 330 while (td_first) { 331 td_next = TAILQ_NEXT(td_first, td_slpq); 332 thread_cow_free(td_first); 333 thread_free(td_first); 334 td_first = td_next; 335 } 336 } 337 } 338 339 /* 340 * Allocate a thread. 341 */ 342 struct thread * 343 thread_alloc(int pages) 344 { 345 struct thread *td; 346 347 thread_reap(); /* check if any zombies to get */ 348 349 td = (struct thread *)uma_zalloc(thread_zone, M_WAITOK); 350 KASSERT(td->td_kstack == 0, ("thread_alloc got thread with kstack")); 351 if (!vm_thread_new(td, pages)) { 352 uma_zfree(thread_zone, td); 353 return (NULL); 354 } 355 cpu_thread_alloc(td); 356 vm_domain_policy_init(&td->td_vm_dom_policy); 357 return (td); 358 } 359 360 int 361 thread_alloc_stack(struct thread *td, int pages) 362 { 363 364 KASSERT(td->td_kstack == 0, 365 ("thread_alloc_stack called on a thread with kstack")); 366 if (!vm_thread_new(td, pages)) 367 return (0); 368 cpu_thread_alloc(td); 369 return (1); 370 } 371 372 /* 373 * Deallocate a thread. 374 */ 375 void 376 thread_free(struct thread *td) 377 { 378 379 lock_profile_thread_exit(td); 380 if (td->td_cpuset) 381 cpuset_rel(td->td_cpuset); 382 td->td_cpuset = NULL; 383 cpu_thread_free(td); 384 if (td->td_kstack != 0) 385 vm_thread_dispose(td); 386 vm_domain_policy_cleanup(&td->td_vm_dom_policy); 387 uma_zfree(thread_zone, td); 388 } 389 390 void 391 thread_cow_get_proc(struct thread *newtd, struct proc *p) 392 { 393 394 PROC_LOCK_ASSERT(p, MA_OWNED); 395 newtd->td_ucred = crhold(p->p_ucred); 396 newtd->td_limit = lim_hold(p->p_limit); 397 newtd->td_cowgen = p->p_cowgen; 398 } 399 400 void 401 thread_cow_get(struct thread *newtd, struct thread *td) 402 { 403 404 newtd->td_ucred = crhold(td->td_ucred); 405 newtd->td_limit = lim_hold(td->td_limit); 406 newtd->td_cowgen = td->td_cowgen; 407 } 408 409 void 410 thread_cow_free(struct thread *td) 411 { 412 413 if (td->td_ucred != NULL) 414 crfree(td->td_ucred); 415 if (td->td_limit != NULL) 416 lim_free(td->td_limit); 417 } 418 419 void 420 thread_cow_update(struct thread *td) 421 { 422 struct proc *p; 423 struct ucred *oldcred; 424 struct plimit *oldlimit; 425 426 p = td->td_proc; 427 oldcred = NULL; 428 oldlimit = NULL; 429 PROC_LOCK(p); 430 if (td->td_ucred != p->p_ucred) { 431 oldcred = td->td_ucred; 432 td->td_ucred = crhold(p->p_ucred); 433 } 434 if (td->td_limit != p->p_limit) { 435 oldlimit = td->td_limit; 436 td->td_limit = lim_hold(p->p_limit); 437 } 438 td->td_cowgen = p->p_cowgen; 439 PROC_UNLOCK(p); 440 if (oldcred != NULL) 441 crfree(oldcred); 442 if (oldlimit != NULL) 443 lim_free(oldlimit); 444 } 445 446 /* 447 * Discard the current thread and exit from its context. 448 * Always called with scheduler locked. 449 * 450 * Because we can't free a thread while we're operating under its context, 451 * push the current thread into our CPU's deadthread holder. This means 452 * we needn't worry about someone else grabbing our context before we 453 * do a cpu_throw(). 454 */ 455 void 456 thread_exit(void) 457 { 458 uint64_t runtime, new_switchtime; 459 struct thread *td; 460 struct thread *td2; 461 struct proc *p; 462 int wakeup_swapper; 463 464 td = curthread; 465 p = td->td_proc; 466 467 PROC_SLOCK_ASSERT(p, MA_OWNED); 468 mtx_assert(&Giant, MA_NOTOWNED); 469 470 PROC_LOCK_ASSERT(p, MA_OWNED); 471 KASSERT(p != NULL, ("thread exiting without a process")); 472 CTR3(KTR_PROC, "thread_exit: thread %p (pid %ld, %s)", td, 473 (long)p->p_pid, td->td_name); 474 KASSERT(TAILQ_EMPTY(&td->td_sigqueue.sq_list), ("signal pending")); 475 476 #ifdef AUDIT 477 AUDIT_SYSCALL_EXIT(0, td); 478 #endif 479 /* 480 * drop FPU & debug register state storage, or any other 481 * architecture specific resources that 482 * would not be on a new untouched process. 483 */ 484 cpu_thread_exit(td); /* XXXSMP */ 485 486 /* 487 * The last thread is left attached to the process 488 * So that the whole bundle gets recycled. Skip 489 * all this stuff if we never had threads. 490 * EXIT clears all sign of other threads when 491 * it goes to single threading, so the last thread always 492 * takes the short path. 493 */ 494 if (p->p_flag & P_HADTHREADS) { 495 if (p->p_numthreads > 1) { 496 atomic_add_int(&td->td_proc->p_exitthreads, 1); 497 thread_unlink(td); 498 td2 = FIRST_THREAD_IN_PROC(p); 499 sched_exit_thread(td2, td); 500 501 /* 502 * The test below is NOT true if we are the 503 * sole exiting thread. P_STOPPED_SINGLE is unset 504 * in exit1() after it is the only survivor. 505 */ 506 if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE) { 507 if (p->p_numthreads == p->p_suspcount) { 508 thread_lock(p->p_singlethread); 509 wakeup_swapper = thread_unsuspend_one( 510 p->p_singlethread, p, false); 511 thread_unlock(p->p_singlethread); 512 if (wakeup_swapper) 513 kick_proc0(); 514 } 515 } 516 517 PCPU_SET(deadthread, td); 518 } else { 519 /* 520 * The last thread is exiting.. but not through exit() 521 */ 522 panic ("thread_exit: Last thread exiting on its own"); 523 } 524 } 525 #ifdef HWPMC_HOOKS 526 /* 527 * If this thread is part of a process that is being tracked by hwpmc(4), 528 * inform the module of the thread's impending exit. 529 */ 530 if (PMC_PROC_IS_USING_PMCS(td->td_proc)) 531 PMC_SWITCH_CONTEXT(td, PMC_FN_CSW_OUT); 532 #endif 533 PROC_UNLOCK(p); 534 PROC_STATLOCK(p); 535 thread_lock(td); 536 PROC_SUNLOCK(p); 537 538 /* Do the same timestamp bookkeeping that mi_switch() would do. */ 539 new_switchtime = cpu_ticks(); 540 runtime = new_switchtime - PCPU_GET(switchtime); 541 td->td_runtime += runtime; 542 td->td_incruntime += runtime; 543 PCPU_SET(switchtime, new_switchtime); 544 PCPU_SET(switchticks, ticks); 545 PCPU_INC(cnt.v_swtch); 546 547 /* Save our resource usage in our process. */ 548 td->td_ru.ru_nvcsw++; 549 ruxagg(p, td); 550 rucollect(&p->p_ru, &td->td_ru); 551 PROC_STATUNLOCK(p); 552 553 td->td_state = TDS_INACTIVE; 554 #ifdef WITNESS 555 witness_thread_exit(td); 556 #endif 557 CTR1(KTR_PROC, "thread_exit: cpu_throw() thread %p", td); 558 sched_throw(td); 559 panic("I'm a teapot!"); 560 /* NOTREACHED */ 561 } 562 563 /* 564 * Do any thread specific cleanups that may be needed in wait() 565 * called with Giant, proc and schedlock not held. 566 */ 567 void 568 thread_wait(struct proc *p) 569 { 570 struct thread *td; 571 572 mtx_assert(&Giant, MA_NOTOWNED); 573 KASSERT(p->p_numthreads == 1, ("multiple threads in thread_wait()")); 574 KASSERT(p->p_exitthreads == 0, ("p_exitthreads leaking")); 575 td = FIRST_THREAD_IN_PROC(p); 576 /* Lock the last thread so we spin until it exits cpu_throw(). */ 577 thread_lock(td); 578 thread_unlock(td); 579 lock_profile_thread_exit(td); 580 cpuset_rel(td->td_cpuset); 581 td->td_cpuset = NULL; 582 cpu_thread_clean(td); 583 thread_cow_free(td); 584 thread_reap(); /* check for zombie threads etc. */ 585 } 586 587 /* 588 * Link a thread to a process. 589 * set up anything that needs to be initialized for it to 590 * be used by the process. 591 */ 592 void 593 thread_link(struct thread *td, struct proc *p) 594 { 595 596 /* 597 * XXX This can't be enabled because it's called for proc0 before 598 * its lock has been created. 599 * PROC_LOCK_ASSERT(p, MA_OWNED); 600 */ 601 td->td_state = TDS_INACTIVE; 602 td->td_proc = p; 603 td->td_flags = TDF_INMEM; 604 605 LIST_INIT(&td->td_contested); 606 LIST_INIT(&td->td_lprof[0]); 607 LIST_INIT(&td->td_lprof[1]); 608 sigqueue_init(&td->td_sigqueue, p); 609 callout_init(&td->td_slpcallout, 1); 610 TAILQ_INSERT_TAIL(&p->p_threads, td, td_plist); 611 p->p_numthreads++; 612 } 613 614 /* 615 * Called from: 616 * thread_exit() 617 */ 618 void 619 thread_unlink(struct thread *td) 620 { 621 struct proc *p = td->td_proc; 622 623 PROC_LOCK_ASSERT(p, MA_OWNED); 624 TAILQ_REMOVE(&p->p_threads, td, td_plist); 625 p->p_numthreads--; 626 /* could clear a few other things here */ 627 /* Must NOT clear links to proc! */ 628 } 629 630 static int 631 calc_remaining(struct proc *p, int mode) 632 { 633 int remaining; 634 635 PROC_LOCK_ASSERT(p, MA_OWNED); 636 PROC_SLOCK_ASSERT(p, MA_OWNED); 637 if (mode == SINGLE_EXIT) 638 remaining = p->p_numthreads; 639 else if (mode == SINGLE_BOUNDARY) 640 remaining = p->p_numthreads - p->p_boundary_count; 641 else if (mode == SINGLE_NO_EXIT || mode == SINGLE_ALLPROC) 642 remaining = p->p_numthreads - p->p_suspcount; 643 else 644 panic("calc_remaining: wrong mode %d", mode); 645 return (remaining); 646 } 647 648 static int 649 remain_for_mode(int mode) 650 { 651 652 return (mode == SINGLE_ALLPROC ? 0 : 1); 653 } 654 655 static int 656 weed_inhib(int mode, struct thread *td2, struct proc *p) 657 { 658 int wakeup_swapper; 659 660 PROC_LOCK_ASSERT(p, MA_OWNED); 661 PROC_SLOCK_ASSERT(p, MA_OWNED); 662 THREAD_LOCK_ASSERT(td2, MA_OWNED); 663 664 wakeup_swapper = 0; 665 switch (mode) { 666 case SINGLE_EXIT: 667 if (TD_IS_SUSPENDED(td2)) 668 wakeup_swapper |= thread_unsuspend_one(td2, p, true); 669 if (TD_ON_SLEEPQ(td2) && (td2->td_flags & TDF_SINTR) != 0) 670 wakeup_swapper |= sleepq_abort(td2, EINTR); 671 break; 672 case SINGLE_BOUNDARY: 673 if (TD_IS_SUSPENDED(td2) && (td2->td_flags & TDF_BOUNDARY) == 0) 674 wakeup_swapper |= thread_unsuspend_one(td2, p, false); 675 if (TD_ON_SLEEPQ(td2) && (td2->td_flags & TDF_SINTR) != 0) 676 wakeup_swapper |= sleepq_abort(td2, ERESTART); 677 break; 678 case SINGLE_NO_EXIT: 679 if (TD_IS_SUSPENDED(td2) && (td2->td_flags & TDF_BOUNDARY) == 0) 680 wakeup_swapper |= thread_unsuspend_one(td2, p, false); 681 if (TD_ON_SLEEPQ(td2) && (td2->td_flags & TDF_SINTR) != 0) 682 wakeup_swapper |= sleepq_abort(td2, ERESTART); 683 break; 684 case SINGLE_ALLPROC: 685 /* 686 * ALLPROC suspend tries to avoid spurious EINTR for 687 * threads sleeping interruptable, by suspending the 688 * thread directly, similarly to sig_suspend_threads(). 689 * Since such sleep is not performed at the user 690 * boundary, TDF_BOUNDARY flag is not set, and TDF_ALLPROCSUSP 691 * is used to avoid immediate un-suspend. 692 */ 693 if (TD_IS_SUSPENDED(td2) && (td2->td_flags & (TDF_BOUNDARY | 694 TDF_ALLPROCSUSP)) == 0) 695 wakeup_swapper |= thread_unsuspend_one(td2, p, false); 696 if (TD_ON_SLEEPQ(td2) && (td2->td_flags & TDF_SINTR) != 0) { 697 if ((td2->td_flags & TDF_SBDRY) == 0) { 698 thread_suspend_one(td2); 699 td2->td_flags |= TDF_ALLPROCSUSP; 700 } else { 701 wakeup_swapper |= sleepq_abort(td2, ERESTART); 702 } 703 } 704 break; 705 } 706 return (wakeup_swapper); 707 } 708 709 /* 710 * Enforce single-threading. 711 * 712 * Returns 1 if the caller must abort (another thread is waiting to 713 * exit the process or similar). Process is locked! 714 * Returns 0 when you are successfully the only thread running. 715 * A process has successfully single threaded in the suspend mode when 716 * There are no threads in user mode. Threads in the kernel must be 717 * allowed to continue until they get to the user boundary. They may even 718 * copy out their return values and data before suspending. They may however be 719 * accelerated in reaching the user boundary as we will wake up 720 * any sleeping threads that are interruptable. (PCATCH). 721 */ 722 int 723 thread_single(struct proc *p, int mode) 724 { 725 struct thread *td; 726 struct thread *td2; 727 int remaining, wakeup_swapper; 728 729 td = curthread; 730 KASSERT(mode == SINGLE_EXIT || mode == SINGLE_BOUNDARY || 731 mode == SINGLE_ALLPROC || mode == SINGLE_NO_EXIT, 732 ("invalid mode %d", mode)); 733 /* 734 * If allowing non-ALLPROC singlethreading for non-curproc 735 * callers, calc_remaining() and remain_for_mode() should be 736 * adjusted to also account for td->td_proc != p. For now 737 * this is not implemented because it is not used. 738 */ 739 KASSERT((mode == SINGLE_ALLPROC && td->td_proc != p) || 740 (mode != SINGLE_ALLPROC && td->td_proc == p), 741 ("mode %d proc %p curproc %p", mode, p, td->td_proc)); 742 mtx_assert(&Giant, MA_NOTOWNED); 743 PROC_LOCK_ASSERT(p, MA_OWNED); 744 745 if ((p->p_flag & P_HADTHREADS) == 0 && mode != SINGLE_ALLPROC) 746 return (0); 747 748 /* Is someone already single threading? */ 749 if (p->p_singlethread != NULL && p->p_singlethread != td) 750 return (1); 751 752 if (mode == SINGLE_EXIT) { 753 p->p_flag |= P_SINGLE_EXIT; 754 p->p_flag &= ~P_SINGLE_BOUNDARY; 755 } else { 756 p->p_flag &= ~P_SINGLE_EXIT; 757 if (mode == SINGLE_BOUNDARY) 758 p->p_flag |= P_SINGLE_BOUNDARY; 759 else 760 p->p_flag &= ~P_SINGLE_BOUNDARY; 761 } 762 if (mode == SINGLE_ALLPROC) 763 p->p_flag |= P_TOTAL_STOP; 764 p->p_flag |= P_STOPPED_SINGLE; 765 PROC_SLOCK(p); 766 p->p_singlethread = td; 767 remaining = calc_remaining(p, mode); 768 while (remaining != remain_for_mode(mode)) { 769 if (P_SHOULDSTOP(p) != P_STOPPED_SINGLE) 770 goto stopme; 771 wakeup_swapper = 0; 772 FOREACH_THREAD_IN_PROC(p, td2) { 773 if (td2 == td) 774 continue; 775 thread_lock(td2); 776 td2->td_flags |= TDF_ASTPENDING | TDF_NEEDSUSPCHK; 777 if (TD_IS_INHIBITED(td2)) { 778 wakeup_swapper |= weed_inhib(mode, td2, p); 779 #ifdef SMP 780 } else if (TD_IS_RUNNING(td2) && td != td2) { 781 forward_signal(td2); 782 #endif 783 } 784 thread_unlock(td2); 785 } 786 if (wakeup_swapper) 787 kick_proc0(); 788 remaining = calc_remaining(p, mode); 789 790 /* 791 * Maybe we suspended some threads.. was it enough? 792 */ 793 if (remaining == remain_for_mode(mode)) 794 break; 795 796 stopme: 797 /* 798 * Wake us up when everyone else has suspended. 799 * In the mean time we suspend as well. 800 */ 801 thread_suspend_switch(td, p); 802 remaining = calc_remaining(p, mode); 803 } 804 if (mode == SINGLE_EXIT) { 805 /* 806 * Convert the process to an unthreaded process. The 807 * SINGLE_EXIT is called by exit1() or execve(), in 808 * both cases other threads must be retired. 809 */ 810 KASSERT(p->p_numthreads == 1, ("Unthreading with >1 threads")); 811 p->p_singlethread = NULL; 812 p->p_flag &= ~(P_STOPPED_SINGLE | P_SINGLE_EXIT | P_HADTHREADS); 813 814 /* 815 * Wait for any remaining threads to exit cpu_throw(). 816 */ 817 while (p->p_exitthreads != 0) { 818 PROC_SUNLOCK(p); 819 PROC_UNLOCK(p); 820 sched_relinquish(td); 821 PROC_LOCK(p); 822 PROC_SLOCK(p); 823 } 824 } else if (mode == SINGLE_BOUNDARY) { 825 /* 826 * Wait until all suspended threads are removed from 827 * the processors. The thread_suspend_check() 828 * increments p_boundary_count while it is still 829 * running, which makes it possible for the execve() 830 * to destroy vmspace while our other threads are 831 * still using the address space. 832 * 833 * We lock the thread, which is only allowed to 834 * succeed after context switch code finished using 835 * the address space. 836 */ 837 FOREACH_THREAD_IN_PROC(p, td2) { 838 if (td2 == td) 839 continue; 840 thread_lock(td2); 841 KASSERT((td2->td_flags & TDF_BOUNDARY) != 0, 842 ("td %p not on boundary", td2)); 843 KASSERT(TD_IS_SUSPENDED(td2), 844 ("td %p is not suspended", td2)); 845 thread_unlock(td2); 846 } 847 } 848 PROC_SUNLOCK(p); 849 return (0); 850 } 851 852 bool 853 thread_suspend_check_needed(void) 854 { 855 struct proc *p; 856 struct thread *td; 857 858 td = curthread; 859 p = td->td_proc; 860 PROC_LOCK_ASSERT(p, MA_OWNED); 861 return (P_SHOULDSTOP(p) || ((p->p_flag & P_TRACED) != 0 && 862 (td->td_dbgflags & TDB_SUSPEND) != 0)); 863 } 864 865 /* 866 * Called in from locations that can safely check to see 867 * whether we have to suspend or at least throttle for a 868 * single-thread event (e.g. fork). 869 * 870 * Such locations include userret(). 871 * If the "return_instead" argument is non zero, the thread must be able to 872 * accept 0 (caller may continue), or 1 (caller must abort) as a result. 873 * 874 * The 'return_instead' argument tells the function if it may do a 875 * thread_exit() or suspend, or whether the caller must abort and back 876 * out instead. 877 * 878 * If the thread that set the single_threading request has set the 879 * P_SINGLE_EXIT bit in the process flags then this call will never return 880 * if 'return_instead' is false, but will exit. 881 * 882 * P_SINGLE_EXIT | return_instead == 0| return_instead != 0 883 *---------------+--------------------+--------------------- 884 * 0 | returns 0 | returns 0 or 1 885 * | when ST ends | immediately 886 *---------------+--------------------+--------------------- 887 * 1 | thread exits | returns 1 888 * | | immediately 889 * 0 = thread_exit() or suspension ok, 890 * other = return error instead of stopping the thread. 891 * 892 * While a full suspension is under effect, even a single threading 893 * thread would be suspended if it made this call (but it shouldn't). 894 * This call should only be made from places where 895 * thread_exit() would be safe as that may be the outcome unless 896 * return_instead is set. 897 */ 898 int 899 thread_suspend_check(int return_instead) 900 { 901 struct thread *td; 902 struct proc *p; 903 int wakeup_swapper; 904 905 td = curthread; 906 p = td->td_proc; 907 mtx_assert(&Giant, MA_NOTOWNED); 908 PROC_LOCK_ASSERT(p, MA_OWNED); 909 while (thread_suspend_check_needed()) { 910 if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE) { 911 KASSERT(p->p_singlethread != NULL, 912 ("singlethread not set")); 913 /* 914 * The only suspension in action is a 915 * single-threading. Single threader need not stop. 916 * XXX Should be safe to access unlocked 917 * as it can only be set to be true by us. 918 */ 919 if (p->p_singlethread == td) 920 return (0); /* Exempt from stopping. */ 921 } 922 if ((p->p_flag & P_SINGLE_EXIT) && return_instead) 923 return (EINTR); 924 925 /* Should we goto user boundary if we didn't come from there? */ 926 if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE && 927 (p->p_flag & P_SINGLE_BOUNDARY) && return_instead) 928 return (ERESTART); 929 930 /* 931 * Ignore suspend requests if they are deferred. 932 */ 933 if ((td->td_flags & TDF_SBDRY) != 0) { 934 KASSERT(return_instead, 935 ("TDF_SBDRY set for unsafe thread_suspend_check")); 936 return (0); 937 } 938 939 /* 940 * If the process is waiting for us to exit, 941 * this thread should just suicide. 942 * Assumes that P_SINGLE_EXIT implies P_STOPPED_SINGLE. 943 */ 944 if ((p->p_flag & P_SINGLE_EXIT) && (p->p_singlethread != td)) { 945 PROC_UNLOCK(p); 946 947 /* 948 * Allow Linux emulation layer to do some work 949 * before thread suicide. 950 */ 951 if (__predict_false(p->p_sysent->sv_thread_detach != NULL)) 952 (p->p_sysent->sv_thread_detach)(td); 953 kern_thr_exit(td); 954 panic("stopped thread did not exit"); 955 } 956 957 PROC_SLOCK(p); 958 thread_stopped(p); 959 if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE) { 960 if (p->p_numthreads == p->p_suspcount + 1) { 961 thread_lock(p->p_singlethread); 962 wakeup_swapper = thread_unsuspend_one( 963 p->p_singlethread, p, false); 964 thread_unlock(p->p_singlethread); 965 if (wakeup_swapper) 966 kick_proc0(); 967 } 968 } 969 PROC_UNLOCK(p); 970 thread_lock(td); 971 /* 972 * When a thread suspends, it just 973 * gets taken off all queues. 974 */ 975 thread_suspend_one(td); 976 if (return_instead == 0) { 977 p->p_boundary_count++; 978 td->td_flags |= TDF_BOUNDARY; 979 } 980 PROC_SUNLOCK(p); 981 mi_switch(SW_INVOL | SWT_SUSPEND, NULL); 982 thread_unlock(td); 983 PROC_LOCK(p); 984 } 985 return (0); 986 } 987 988 void 989 thread_suspend_switch(struct thread *td, struct proc *p) 990 { 991 992 KASSERT(!TD_IS_SUSPENDED(td), ("already suspended")); 993 PROC_LOCK_ASSERT(p, MA_OWNED); 994 PROC_SLOCK_ASSERT(p, MA_OWNED); 995 /* 996 * We implement thread_suspend_one in stages here to avoid 997 * dropping the proc lock while the thread lock is owned. 998 */ 999 if (p == td->td_proc) { 1000 thread_stopped(p); 1001 p->p_suspcount++; 1002 } 1003 PROC_UNLOCK(p); 1004 thread_lock(td); 1005 td->td_flags &= ~TDF_NEEDSUSPCHK; 1006 TD_SET_SUSPENDED(td); 1007 sched_sleep(td, 0); 1008 PROC_SUNLOCK(p); 1009 DROP_GIANT(); 1010 mi_switch(SW_VOL | SWT_SUSPEND, NULL); 1011 thread_unlock(td); 1012 PICKUP_GIANT(); 1013 PROC_LOCK(p); 1014 PROC_SLOCK(p); 1015 } 1016 1017 void 1018 thread_suspend_one(struct thread *td) 1019 { 1020 struct proc *p; 1021 1022 p = td->td_proc; 1023 PROC_SLOCK_ASSERT(p, MA_OWNED); 1024 THREAD_LOCK_ASSERT(td, MA_OWNED); 1025 KASSERT(!TD_IS_SUSPENDED(td), ("already suspended")); 1026 p->p_suspcount++; 1027 td->td_flags &= ~TDF_NEEDSUSPCHK; 1028 TD_SET_SUSPENDED(td); 1029 sched_sleep(td, 0); 1030 } 1031 1032 static int 1033 thread_unsuspend_one(struct thread *td, struct proc *p, bool boundary) 1034 { 1035 1036 THREAD_LOCK_ASSERT(td, MA_OWNED); 1037 KASSERT(TD_IS_SUSPENDED(td), ("Thread not suspended")); 1038 TD_CLR_SUSPENDED(td); 1039 td->td_flags &= ~TDF_ALLPROCSUSP; 1040 if (td->td_proc == p) { 1041 PROC_SLOCK_ASSERT(p, MA_OWNED); 1042 p->p_suspcount--; 1043 if (boundary && (td->td_flags & TDF_BOUNDARY) != 0) { 1044 td->td_flags &= ~TDF_BOUNDARY; 1045 p->p_boundary_count--; 1046 } 1047 } 1048 return (setrunnable(td)); 1049 } 1050 1051 /* 1052 * Allow all threads blocked by single threading to continue running. 1053 */ 1054 void 1055 thread_unsuspend(struct proc *p) 1056 { 1057 struct thread *td; 1058 int wakeup_swapper; 1059 1060 PROC_LOCK_ASSERT(p, MA_OWNED); 1061 PROC_SLOCK_ASSERT(p, MA_OWNED); 1062 wakeup_swapper = 0; 1063 if (!P_SHOULDSTOP(p)) { 1064 FOREACH_THREAD_IN_PROC(p, td) { 1065 thread_lock(td); 1066 if (TD_IS_SUSPENDED(td)) { 1067 wakeup_swapper |= thread_unsuspend_one(td, p, 1068 true); 1069 } 1070 thread_unlock(td); 1071 } 1072 } else if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE && 1073 p->p_numthreads == p->p_suspcount) { 1074 /* 1075 * Stopping everything also did the job for the single 1076 * threading request. Now we've downgraded to single-threaded, 1077 * let it continue. 1078 */ 1079 if (p->p_singlethread->td_proc == p) { 1080 thread_lock(p->p_singlethread); 1081 wakeup_swapper = thread_unsuspend_one( 1082 p->p_singlethread, p, false); 1083 thread_unlock(p->p_singlethread); 1084 } 1085 } 1086 if (wakeup_swapper) 1087 kick_proc0(); 1088 } 1089 1090 /* 1091 * End the single threading mode.. 1092 */ 1093 void 1094 thread_single_end(struct proc *p, int mode) 1095 { 1096 struct thread *td; 1097 int wakeup_swapper; 1098 1099 KASSERT(mode == SINGLE_EXIT || mode == SINGLE_BOUNDARY || 1100 mode == SINGLE_ALLPROC || mode == SINGLE_NO_EXIT, 1101 ("invalid mode %d", mode)); 1102 PROC_LOCK_ASSERT(p, MA_OWNED); 1103 KASSERT((mode == SINGLE_ALLPROC && (p->p_flag & P_TOTAL_STOP) != 0) || 1104 (mode != SINGLE_ALLPROC && (p->p_flag & P_TOTAL_STOP) == 0), 1105 ("mode %d does not match P_TOTAL_STOP", mode)); 1106 KASSERT(mode == SINGLE_ALLPROC || p->p_singlethread == curthread, 1107 ("thread_single_end from other thread %p %p", 1108 curthread, p->p_singlethread)); 1109 KASSERT(mode != SINGLE_BOUNDARY || 1110 (p->p_flag & P_SINGLE_BOUNDARY) != 0, 1111 ("mis-matched SINGLE_BOUNDARY flags %x", p->p_flag)); 1112 p->p_flag &= ~(P_STOPPED_SINGLE | P_SINGLE_EXIT | P_SINGLE_BOUNDARY | 1113 P_TOTAL_STOP); 1114 PROC_SLOCK(p); 1115 p->p_singlethread = NULL; 1116 wakeup_swapper = 0; 1117 /* 1118 * If there are other threads they may now run, 1119 * unless of course there is a blanket 'stop order' 1120 * on the process. The single threader must be allowed 1121 * to continue however as this is a bad place to stop. 1122 */ 1123 if (p->p_numthreads != remain_for_mode(mode) && !P_SHOULDSTOP(p)) { 1124 FOREACH_THREAD_IN_PROC(p, td) { 1125 thread_lock(td); 1126 if (TD_IS_SUSPENDED(td)) { 1127 wakeup_swapper |= thread_unsuspend_one(td, p, 1128 mode == SINGLE_BOUNDARY); 1129 } 1130 thread_unlock(td); 1131 } 1132 } 1133 KASSERT(mode != SINGLE_BOUNDARY || p->p_boundary_count == 0, 1134 ("inconsistent boundary count %d", p->p_boundary_count)); 1135 PROC_SUNLOCK(p); 1136 if (wakeup_swapper) 1137 kick_proc0(); 1138 } 1139 1140 struct thread * 1141 thread_find(struct proc *p, lwpid_t tid) 1142 { 1143 struct thread *td; 1144 1145 PROC_LOCK_ASSERT(p, MA_OWNED); 1146 FOREACH_THREAD_IN_PROC(p, td) { 1147 if (td->td_tid == tid) 1148 break; 1149 } 1150 return (td); 1151 } 1152 1153 /* Locate a thread by number; return with proc lock held. */ 1154 struct thread * 1155 tdfind(lwpid_t tid, pid_t pid) 1156 { 1157 #define RUN_THRESH 16 1158 struct thread *td; 1159 int run = 0; 1160 1161 rw_rlock(&tidhash_lock); 1162 LIST_FOREACH(td, TIDHASH(tid), td_hash) { 1163 if (td->td_tid == tid) { 1164 if (pid != -1 && td->td_proc->p_pid != pid) { 1165 td = NULL; 1166 break; 1167 } 1168 PROC_LOCK(td->td_proc); 1169 if (td->td_proc->p_state == PRS_NEW) { 1170 PROC_UNLOCK(td->td_proc); 1171 td = NULL; 1172 break; 1173 } 1174 if (run > RUN_THRESH) { 1175 if (rw_try_upgrade(&tidhash_lock)) { 1176 LIST_REMOVE(td, td_hash); 1177 LIST_INSERT_HEAD(TIDHASH(td->td_tid), 1178 td, td_hash); 1179 rw_wunlock(&tidhash_lock); 1180 return (td); 1181 } 1182 } 1183 break; 1184 } 1185 run++; 1186 } 1187 rw_runlock(&tidhash_lock); 1188 return (td); 1189 } 1190 1191 void 1192 tidhash_add(struct thread *td) 1193 { 1194 rw_wlock(&tidhash_lock); 1195 LIST_INSERT_HEAD(TIDHASH(td->td_tid), td, td_hash); 1196 rw_wunlock(&tidhash_lock); 1197 } 1198 1199 void 1200 tidhash_remove(struct thread *td) 1201 { 1202 rw_wlock(&tidhash_lock); 1203 LIST_REMOVE(td, td_hash); 1204 rw_wunlock(&tidhash_lock); 1205 } 1206