1 /*- 2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD 3 * 4 * Copyright (c) 2008 Attilio Rao <attilio@FreeBSD.org> 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice(s), this list of conditions and the following disclaimer as 12 * the first lines of this file unmodified other than the possible 13 * addition of one or more copyright notices. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice(s), this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 18 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY 19 * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED 20 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE 21 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY 22 * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 23 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR 24 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER 25 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 26 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH 28 * DAMAGE. 29 */ 30 31 #include "opt_ddb.h" 32 #include "opt_hwpmc_hooks.h" 33 34 #include <sys/cdefs.h> 35 __FBSDID("$FreeBSD$"); 36 37 #include <sys/param.h> 38 #include <sys/kdb.h> 39 #include <sys/ktr.h> 40 #include <sys/lock.h> 41 #include <sys/lock_profile.h> 42 #include <sys/lockmgr.h> 43 #include <sys/lockstat.h> 44 #include <sys/mutex.h> 45 #include <sys/proc.h> 46 #include <sys/sleepqueue.h> 47 #ifdef DEBUG_LOCKS 48 #include <sys/stack.h> 49 #endif 50 #include <sys/sysctl.h> 51 #include <sys/systm.h> 52 53 #include <machine/cpu.h> 54 55 #ifdef DDB 56 #include <ddb/ddb.h> 57 #endif 58 59 #ifdef HWPMC_HOOKS 60 #include <sys/pmckern.h> 61 PMC_SOFT_DECLARE( , , lock, failed); 62 #endif 63 64 CTASSERT(LK_UNLOCKED == (LK_UNLOCKED & 65 ~(LK_ALL_WAITERS | LK_EXCLUSIVE_SPINNERS))); 66 67 #define SQ_EXCLUSIVE_QUEUE 0 68 #define SQ_SHARED_QUEUE 1 69 70 #ifndef INVARIANTS 71 #define _lockmgr_assert(lk, what, file, line) 72 #endif 73 74 #define TD_SLOCKS_INC(td) ((td)->td_lk_slocks++) 75 #define TD_SLOCKS_DEC(td) ((td)->td_lk_slocks--) 76 77 #ifndef DEBUG_LOCKS 78 #define STACK_PRINT(lk) 79 #define STACK_SAVE(lk) 80 #define STACK_ZERO(lk) 81 #else 82 #define STACK_PRINT(lk) stack_print_ddb(&(lk)->lk_stack) 83 #define STACK_SAVE(lk) stack_save(&(lk)->lk_stack) 84 #define STACK_ZERO(lk) stack_zero(&(lk)->lk_stack) 85 #endif 86 87 #define LOCK_LOG2(lk, string, arg1, arg2) \ 88 if (LOCK_LOG_TEST(&(lk)->lock_object, 0)) \ 89 CTR2(KTR_LOCK, (string), (arg1), (arg2)) 90 #define LOCK_LOG3(lk, string, arg1, arg2, arg3) \ 91 if (LOCK_LOG_TEST(&(lk)->lock_object, 0)) \ 92 CTR3(KTR_LOCK, (string), (arg1), (arg2), (arg3)) 93 94 #define GIANT_DECLARE \ 95 int _i = 0; \ 96 WITNESS_SAVE_DECL(Giant) 97 #define GIANT_RESTORE() do { \ 98 if (__predict_false(_i > 0)) { \ 99 while (_i--) \ 100 mtx_lock(&Giant); \ 101 WITNESS_RESTORE(&Giant.lock_object, Giant); \ 102 } \ 103 } while (0) 104 #define GIANT_SAVE() do { \ 105 if (__predict_false(mtx_owned(&Giant))) { \ 106 WITNESS_SAVE(&Giant.lock_object, Giant); \ 107 while (mtx_owned(&Giant)) { \ 108 _i++; \ 109 mtx_unlock(&Giant); \ 110 } \ 111 } \ 112 } while (0) 113 114 static bool __always_inline 115 LK_CAN_SHARE(uintptr_t x, int flags, bool fp) 116 { 117 118 if ((x & (LK_SHARE | LK_EXCLUSIVE_WAITERS | LK_EXCLUSIVE_SPINNERS)) == 119 LK_SHARE) 120 return (true); 121 if (fp || (!(x & LK_SHARE))) 122 return (false); 123 if ((curthread->td_lk_slocks != 0 && !(flags & LK_NODDLKTREAT)) || 124 (curthread->td_pflags & TDP_DEADLKTREAT)) 125 return (true); 126 return (false); 127 } 128 129 #define LK_TRYOP(x) \ 130 ((x) & LK_NOWAIT) 131 132 #define LK_CAN_WITNESS(x) \ 133 (((x) & LK_NOWITNESS) == 0 && !LK_TRYOP(x)) 134 #define LK_TRYWIT(x) \ 135 (LK_TRYOP(x) ? LOP_TRYLOCK : 0) 136 137 #define lockmgr_disowned(lk) \ 138 (((lk)->lk_lock & ~(LK_FLAGMASK & ~LK_SHARE)) == LK_KERNPROC) 139 140 #define lockmgr_xlocked_v(v) \ 141 (((v) & ~(LK_FLAGMASK & ~LK_SHARE)) == (uintptr_t)curthread) 142 143 #define lockmgr_xlocked(lk) lockmgr_xlocked_v((lk)->lk_lock) 144 145 static void assert_lockmgr(const struct lock_object *lock, int how); 146 #ifdef DDB 147 static void db_show_lockmgr(const struct lock_object *lock); 148 #endif 149 static void lock_lockmgr(struct lock_object *lock, uintptr_t how); 150 #ifdef KDTRACE_HOOKS 151 static int owner_lockmgr(const struct lock_object *lock, 152 struct thread **owner); 153 #endif 154 static uintptr_t unlock_lockmgr(struct lock_object *lock); 155 156 struct lock_class lock_class_lockmgr = { 157 .lc_name = "lockmgr", 158 .lc_flags = LC_RECURSABLE | LC_SLEEPABLE | LC_SLEEPLOCK | LC_UPGRADABLE, 159 .lc_assert = assert_lockmgr, 160 #ifdef DDB 161 .lc_ddb_show = db_show_lockmgr, 162 #endif 163 .lc_lock = lock_lockmgr, 164 .lc_unlock = unlock_lockmgr, 165 #ifdef KDTRACE_HOOKS 166 .lc_owner = owner_lockmgr, 167 #endif 168 }; 169 170 struct lockmgr_wait { 171 const char *iwmesg; 172 int ipri; 173 int itimo; 174 }; 175 176 static bool __always_inline lockmgr_slock_try(struct lock *lk, uintptr_t *xp, 177 int flags, bool fp); 178 static bool __always_inline lockmgr_sunlock_try(struct lock *lk, uintptr_t *xp); 179 180 static void 181 lockmgr_exit(u_int flags, struct lock_object *ilk, int wakeup_swapper) 182 { 183 struct lock_class *class; 184 185 if (flags & LK_INTERLOCK) { 186 class = LOCK_CLASS(ilk); 187 class->lc_unlock(ilk); 188 } 189 190 if (__predict_false(wakeup_swapper)) 191 kick_proc0(); 192 } 193 194 static void 195 lockmgr_note_shared_acquire(struct lock *lk, int contested, 196 uint64_t waittime, const char *file, int line, int flags) 197 { 198 199 LOCKSTAT_PROFILE_OBTAIN_RWLOCK_SUCCESS(lockmgr__acquire, lk, contested, 200 waittime, file, line, LOCKSTAT_READER); 201 LOCK_LOG_LOCK("SLOCK", &lk->lock_object, 0, 0, file, line); 202 WITNESS_LOCK(&lk->lock_object, LK_TRYWIT(flags), file, line); 203 TD_LOCKS_INC(curthread); 204 TD_SLOCKS_INC(curthread); 205 STACK_SAVE(lk); 206 } 207 208 static void 209 lockmgr_note_shared_release(struct lock *lk, const char *file, int line) 210 { 211 212 WITNESS_UNLOCK(&lk->lock_object, 0, file, line); 213 LOCK_LOG_LOCK("SUNLOCK", &lk->lock_object, 0, 0, file, line); 214 TD_LOCKS_DEC(curthread); 215 TD_SLOCKS_DEC(curthread); 216 } 217 218 static void 219 lockmgr_note_exclusive_acquire(struct lock *lk, int contested, 220 uint64_t waittime, const char *file, int line, int flags) 221 { 222 223 LOCKSTAT_PROFILE_OBTAIN_RWLOCK_SUCCESS(lockmgr__acquire, lk, contested, 224 waittime, file, line, LOCKSTAT_WRITER); 225 LOCK_LOG_LOCK("XLOCK", &lk->lock_object, 0, lk->lk_recurse, file, line); 226 WITNESS_LOCK(&lk->lock_object, LOP_EXCLUSIVE | LK_TRYWIT(flags), file, 227 line); 228 TD_LOCKS_INC(curthread); 229 STACK_SAVE(lk); 230 } 231 232 static void 233 lockmgr_note_exclusive_release(struct lock *lk, const char *file, int line) 234 { 235 236 if (LK_HOLDER(lk->lk_lock) != LK_KERNPROC) { 237 WITNESS_UNLOCK(&lk->lock_object, LOP_EXCLUSIVE, file, line); 238 TD_LOCKS_DEC(curthread); 239 } 240 LOCK_LOG_LOCK("XUNLOCK", &lk->lock_object, 0, lk->lk_recurse, file, 241 line); 242 } 243 244 static __inline struct thread * 245 lockmgr_xholder(const struct lock *lk) 246 { 247 uintptr_t x; 248 249 x = lk->lk_lock; 250 return ((x & LK_SHARE) ? NULL : (struct thread *)LK_HOLDER(x)); 251 } 252 253 /* 254 * It assumes sleepq_lock held and returns with this one unheld. 255 * It also assumes the generic interlock is sane and previously checked. 256 * If LK_INTERLOCK is specified the interlock is not reacquired after the 257 * sleep. 258 */ 259 static __inline int 260 sleeplk(struct lock *lk, u_int flags, struct lock_object *ilk, 261 const char *wmesg, int pri, int timo, int queue) 262 { 263 GIANT_DECLARE; 264 struct lock_class *class; 265 int catch, error; 266 267 class = (flags & LK_INTERLOCK) ? LOCK_CLASS(ilk) : NULL; 268 catch = pri & PCATCH; 269 pri &= PRIMASK; 270 error = 0; 271 272 LOCK_LOG3(lk, "%s: %p blocking on the %s sleepqueue", __func__, lk, 273 (queue == SQ_EXCLUSIVE_QUEUE) ? "exclusive" : "shared"); 274 275 if (flags & LK_INTERLOCK) 276 class->lc_unlock(ilk); 277 if (queue == SQ_EXCLUSIVE_QUEUE && (flags & LK_SLEEPFAIL) != 0) 278 lk->lk_exslpfail++; 279 GIANT_SAVE(); 280 sleepq_add(&lk->lock_object, NULL, wmesg, SLEEPQ_LK | (catch ? 281 SLEEPQ_INTERRUPTIBLE : 0), queue); 282 if ((flags & LK_TIMELOCK) && timo) 283 sleepq_set_timeout(&lk->lock_object, timo); 284 285 /* 286 * Decisional switch for real sleeping. 287 */ 288 if ((flags & LK_TIMELOCK) && timo && catch) 289 error = sleepq_timedwait_sig(&lk->lock_object, pri); 290 else if ((flags & LK_TIMELOCK) && timo) 291 error = sleepq_timedwait(&lk->lock_object, pri); 292 else if (catch) 293 error = sleepq_wait_sig(&lk->lock_object, pri); 294 else 295 sleepq_wait(&lk->lock_object, pri); 296 GIANT_RESTORE(); 297 if ((flags & LK_SLEEPFAIL) && error == 0) 298 error = ENOLCK; 299 300 return (error); 301 } 302 303 static __inline int 304 wakeupshlk(struct lock *lk, const char *file, int line) 305 { 306 uintptr_t v, x, orig_x; 307 u_int realexslp; 308 int queue, wakeup_swapper; 309 310 wakeup_swapper = 0; 311 for (;;) { 312 x = lk->lk_lock; 313 if (lockmgr_sunlock_try(lk, &x)) 314 break; 315 316 /* 317 * We should have a sharer with waiters, so enter the hard 318 * path in order to handle wakeups correctly. 319 */ 320 sleepq_lock(&lk->lock_object); 321 orig_x = lk->lk_lock; 322 retry_sleepq: 323 x = orig_x & (LK_ALL_WAITERS | LK_EXCLUSIVE_SPINNERS); 324 v = LK_UNLOCKED; 325 326 /* 327 * If the lock has exclusive waiters, give them preference in 328 * order to avoid deadlock with shared runners up. 329 * If interruptible sleeps left the exclusive queue empty 330 * avoid a starvation for the threads sleeping on the shared 331 * queue by giving them precedence and cleaning up the 332 * exclusive waiters bit anyway. 333 * Please note that lk_exslpfail count may be lying about 334 * the real number of waiters with the LK_SLEEPFAIL flag on 335 * because they may be used in conjunction with interruptible 336 * sleeps so lk_exslpfail might be considered an 'upper limit' 337 * bound, including the edge cases. 338 */ 339 realexslp = sleepq_sleepcnt(&lk->lock_object, 340 SQ_EXCLUSIVE_QUEUE); 341 if ((x & LK_EXCLUSIVE_WAITERS) != 0 && realexslp != 0) { 342 if (lk->lk_exslpfail < realexslp) { 343 lk->lk_exslpfail = 0; 344 queue = SQ_EXCLUSIVE_QUEUE; 345 v |= (x & LK_SHARED_WAITERS); 346 } else { 347 lk->lk_exslpfail = 0; 348 LOCK_LOG2(lk, 349 "%s: %p has only LK_SLEEPFAIL sleepers", 350 __func__, lk); 351 LOCK_LOG2(lk, 352 "%s: %p waking up threads on the exclusive queue", 353 __func__, lk); 354 wakeup_swapper = 355 sleepq_broadcast(&lk->lock_object, 356 SLEEPQ_LK, 0, SQ_EXCLUSIVE_QUEUE); 357 queue = SQ_SHARED_QUEUE; 358 } 359 360 } else { 361 362 /* 363 * Exclusive waiters sleeping with LK_SLEEPFAIL on 364 * and using interruptible sleeps/timeout may have 365 * left spourious lk_exslpfail counts on, so clean 366 * it up anyway. 367 */ 368 lk->lk_exslpfail = 0; 369 queue = SQ_SHARED_QUEUE; 370 } 371 372 if (lockmgr_sunlock_try(lk, &orig_x)) { 373 sleepq_release(&lk->lock_object); 374 break; 375 } 376 377 x |= LK_SHARERS_LOCK(1); 378 if (!atomic_fcmpset_rel_ptr(&lk->lk_lock, &x, v)) { 379 orig_x = x; 380 goto retry_sleepq; 381 } 382 LOCK_LOG3(lk, "%s: %p waking up threads on the %s queue", 383 __func__, lk, queue == SQ_SHARED_QUEUE ? "shared" : 384 "exclusive"); 385 wakeup_swapper |= sleepq_broadcast(&lk->lock_object, SLEEPQ_LK, 386 0, queue); 387 sleepq_release(&lk->lock_object); 388 break; 389 } 390 391 LOCKSTAT_PROFILE_RELEASE_RWLOCK(lockmgr__release, lk, LOCKSTAT_READER); 392 return (wakeup_swapper); 393 } 394 395 static void 396 assert_lockmgr(const struct lock_object *lock, int what) 397 { 398 399 panic("lockmgr locks do not support assertions"); 400 } 401 402 static void 403 lock_lockmgr(struct lock_object *lock, uintptr_t how) 404 { 405 406 panic("lockmgr locks do not support sleep interlocking"); 407 } 408 409 static uintptr_t 410 unlock_lockmgr(struct lock_object *lock) 411 { 412 413 panic("lockmgr locks do not support sleep interlocking"); 414 } 415 416 #ifdef KDTRACE_HOOKS 417 static int 418 owner_lockmgr(const struct lock_object *lock, struct thread **owner) 419 { 420 421 panic("lockmgr locks do not support owner inquiring"); 422 } 423 #endif 424 425 void 426 lockinit(struct lock *lk, int pri, const char *wmesg, int timo, int flags) 427 { 428 int iflags; 429 430 MPASS((flags & ~LK_INIT_MASK) == 0); 431 ASSERT_ATOMIC_LOAD_PTR(lk->lk_lock, 432 ("%s: lockmgr not aligned for %s: %p", __func__, wmesg, 433 &lk->lk_lock)); 434 435 iflags = LO_SLEEPABLE | LO_UPGRADABLE; 436 if (flags & LK_CANRECURSE) 437 iflags |= LO_RECURSABLE; 438 if ((flags & LK_NODUP) == 0) 439 iflags |= LO_DUPOK; 440 if (flags & LK_NOPROFILE) 441 iflags |= LO_NOPROFILE; 442 if ((flags & LK_NOWITNESS) == 0) 443 iflags |= LO_WITNESS; 444 if (flags & LK_QUIET) 445 iflags |= LO_QUIET; 446 if (flags & LK_IS_VNODE) 447 iflags |= LO_IS_VNODE; 448 if (flags & LK_NEW) 449 iflags |= LO_NEW; 450 iflags |= flags & LK_NOSHARE; 451 452 lock_init(&lk->lock_object, &lock_class_lockmgr, wmesg, NULL, iflags); 453 lk->lk_lock = LK_UNLOCKED; 454 lk->lk_recurse = 0; 455 lk->lk_exslpfail = 0; 456 lk->lk_timo = timo; 457 lk->lk_pri = pri; 458 STACK_ZERO(lk); 459 } 460 461 /* 462 * XXX: Gross hacks to manipulate external lock flags after 463 * initialization. Used for certain vnode and buf locks. 464 */ 465 void 466 lockallowshare(struct lock *lk) 467 { 468 469 lockmgr_assert(lk, KA_XLOCKED); 470 lk->lock_object.lo_flags &= ~LK_NOSHARE; 471 } 472 473 void 474 lockdisableshare(struct lock *lk) 475 { 476 477 lockmgr_assert(lk, KA_XLOCKED); 478 lk->lock_object.lo_flags |= LK_NOSHARE; 479 } 480 481 void 482 lockallowrecurse(struct lock *lk) 483 { 484 485 lockmgr_assert(lk, KA_XLOCKED); 486 lk->lock_object.lo_flags |= LO_RECURSABLE; 487 } 488 489 void 490 lockdisablerecurse(struct lock *lk) 491 { 492 493 lockmgr_assert(lk, KA_XLOCKED); 494 lk->lock_object.lo_flags &= ~LO_RECURSABLE; 495 } 496 497 void 498 lockdestroy(struct lock *lk) 499 { 500 501 KASSERT(lk->lk_lock == LK_UNLOCKED, ("lockmgr still held")); 502 KASSERT(lk->lk_recurse == 0, ("lockmgr still recursed")); 503 KASSERT(lk->lk_exslpfail == 0, ("lockmgr still exclusive waiters")); 504 lock_destroy(&lk->lock_object); 505 } 506 507 static bool __always_inline 508 lockmgr_slock_try(struct lock *lk, uintptr_t *xp, int flags, bool fp) 509 { 510 511 /* 512 * If no other thread has an exclusive lock, or 513 * no exclusive waiter is present, bump the count of 514 * sharers. Since we have to preserve the state of 515 * waiters, if we fail to acquire the shared lock 516 * loop back and retry. 517 */ 518 *xp = lk->lk_lock; 519 while (LK_CAN_SHARE(*xp, flags, fp)) { 520 if (atomic_fcmpset_acq_ptr(&lk->lk_lock, xp, 521 *xp + LK_ONE_SHARER)) { 522 return (true); 523 } 524 } 525 return (false); 526 } 527 528 static bool __always_inline 529 lockmgr_sunlock_try(struct lock *lk, uintptr_t *xp) 530 { 531 532 for (;;) { 533 if (LK_SHARERS(*xp) > 1 || !(*xp & LK_ALL_WAITERS)) { 534 if (atomic_fcmpset_rel_ptr(&lk->lk_lock, xp, 535 *xp - LK_ONE_SHARER)) 536 return (true); 537 continue; 538 } 539 break; 540 } 541 return (false); 542 } 543 544 static __noinline int 545 lockmgr_slock_hard(struct lock *lk, u_int flags, struct lock_object *ilk, 546 const char *file, int line, struct lockmgr_wait *lwa) 547 { 548 uintptr_t tid, x; 549 int error = 0; 550 const char *iwmesg; 551 int ipri, itimo; 552 553 #ifdef KDTRACE_HOOKS 554 uint64_t sleep_time = 0; 555 #endif 556 #ifdef LOCK_PROFILING 557 uint64_t waittime = 0; 558 int contested = 0; 559 #endif 560 561 if (KERNEL_PANICKED()) 562 goto out; 563 564 tid = (uintptr_t)curthread; 565 566 if (LK_CAN_WITNESS(flags)) 567 WITNESS_CHECKORDER(&lk->lock_object, LOP_NEWORDER, 568 file, line, flags & LK_INTERLOCK ? ilk : NULL); 569 for (;;) { 570 if (lockmgr_slock_try(lk, &x, flags, false)) 571 break; 572 #ifdef HWPMC_HOOKS 573 PMC_SOFT_CALL( , , lock, failed); 574 #endif 575 lock_profile_obtain_lock_failed(&lk->lock_object, 576 &contested, &waittime); 577 578 /* 579 * If the lock is already held by curthread in 580 * exclusive way avoid a deadlock. 581 */ 582 if (LK_HOLDER(x) == tid) { 583 LOCK_LOG2(lk, 584 "%s: %p already held in exclusive mode", 585 __func__, lk); 586 error = EDEADLK; 587 break; 588 } 589 590 /* 591 * If the lock is expected to not sleep just give up 592 * and return. 593 */ 594 if (LK_TRYOP(flags)) { 595 LOCK_LOG2(lk, "%s: %p fails the try operation", 596 __func__, lk); 597 error = EBUSY; 598 break; 599 } 600 601 /* 602 * Acquire the sleepqueue chain lock because we 603 * probabilly will need to manipulate waiters flags. 604 */ 605 sleepq_lock(&lk->lock_object); 606 x = lk->lk_lock; 607 retry_sleepq: 608 609 /* 610 * if the lock can be acquired in shared mode, try 611 * again. 612 */ 613 if (LK_CAN_SHARE(x, flags, false)) { 614 sleepq_release(&lk->lock_object); 615 continue; 616 } 617 618 /* 619 * Try to set the LK_SHARED_WAITERS flag. If we fail, 620 * loop back and retry. 621 */ 622 if ((x & LK_SHARED_WAITERS) == 0) { 623 if (!atomic_fcmpset_acq_ptr(&lk->lk_lock, &x, 624 x | LK_SHARED_WAITERS)) { 625 goto retry_sleepq; 626 } 627 LOCK_LOG2(lk, "%s: %p set shared waiters flag", 628 __func__, lk); 629 } 630 631 if (lwa == NULL) { 632 iwmesg = lk->lock_object.lo_name; 633 ipri = lk->lk_pri; 634 itimo = lk->lk_timo; 635 } else { 636 iwmesg = lwa->iwmesg; 637 ipri = lwa->ipri; 638 itimo = lwa->itimo; 639 } 640 641 /* 642 * As far as we have been unable to acquire the 643 * shared lock and the shared waiters flag is set, 644 * we will sleep. 645 */ 646 #ifdef KDTRACE_HOOKS 647 sleep_time -= lockstat_nsecs(&lk->lock_object); 648 #endif 649 error = sleeplk(lk, flags, ilk, iwmesg, ipri, itimo, 650 SQ_SHARED_QUEUE); 651 #ifdef KDTRACE_HOOKS 652 sleep_time += lockstat_nsecs(&lk->lock_object); 653 #endif 654 flags &= ~LK_INTERLOCK; 655 if (error) { 656 LOCK_LOG3(lk, 657 "%s: interrupted sleep for %p with %d", 658 __func__, lk, error); 659 break; 660 } 661 LOCK_LOG2(lk, "%s: %p resuming from the sleep queue", 662 __func__, lk); 663 } 664 if (error == 0) { 665 #ifdef KDTRACE_HOOKS 666 if (sleep_time != 0) 667 LOCKSTAT_RECORD4(lockmgr__block, lk, sleep_time, 668 LOCKSTAT_READER, (x & LK_SHARE) == 0, 669 (x & LK_SHARE) == 0 ? 0 : LK_SHARERS(x)); 670 #endif 671 #ifdef LOCK_PROFILING 672 lockmgr_note_shared_acquire(lk, contested, waittime, 673 file, line, flags); 674 #else 675 lockmgr_note_shared_acquire(lk, 0, 0, file, line, 676 flags); 677 #endif 678 } 679 680 out: 681 lockmgr_exit(flags, ilk, 0); 682 return (error); 683 } 684 685 static __noinline int 686 lockmgr_xlock_hard(struct lock *lk, u_int flags, struct lock_object *ilk, 687 const char *file, int line, struct lockmgr_wait *lwa) 688 { 689 struct lock_class *class; 690 uintptr_t tid, x, v; 691 int error = 0; 692 const char *iwmesg; 693 int ipri, itimo; 694 695 #ifdef KDTRACE_HOOKS 696 uint64_t sleep_time = 0; 697 #endif 698 #ifdef LOCK_PROFILING 699 uint64_t waittime = 0; 700 int contested = 0; 701 #endif 702 703 if (KERNEL_PANICKED()) 704 goto out; 705 706 tid = (uintptr_t)curthread; 707 708 if (LK_CAN_WITNESS(flags)) 709 WITNESS_CHECKORDER(&lk->lock_object, LOP_NEWORDER | 710 LOP_EXCLUSIVE, file, line, flags & LK_INTERLOCK ? 711 ilk : NULL); 712 713 /* 714 * If curthread already holds the lock and this one is 715 * allowed to recurse, simply recurse on it. 716 */ 717 if (lockmgr_xlocked(lk)) { 718 if ((flags & LK_CANRECURSE) == 0 && 719 (lk->lock_object.lo_flags & LO_RECURSABLE) == 0) { 720 /* 721 * If the lock is expected to not panic just 722 * give up and return. 723 */ 724 if (LK_TRYOP(flags)) { 725 LOCK_LOG2(lk, 726 "%s: %p fails the try operation", 727 __func__, lk); 728 error = EBUSY; 729 goto out; 730 } 731 if (flags & LK_INTERLOCK) { 732 class = LOCK_CLASS(ilk); 733 class->lc_unlock(ilk); 734 } 735 STACK_PRINT(lk); 736 panic("%s: recursing on non recursive lockmgr %p " 737 "@ %s:%d\n", __func__, lk, file, line); 738 } 739 lk->lk_recurse++; 740 LOCK_LOG2(lk, "%s: %p recursing", __func__, lk); 741 LOCK_LOG_LOCK("XLOCK", &lk->lock_object, 0, 742 lk->lk_recurse, file, line); 743 WITNESS_LOCK(&lk->lock_object, LOP_EXCLUSIVE | 744 LK_TRYWIT(flags), file, line); 745 TD_LOCKS_INC(curthread); 746 goto out; 747 } 748 749 for (;;) { 750 if (lk->lk_lock == LK_UNLOCKED && 751 atomic_cmpset_acq_ptr(&lk->lk_lock, LK_UNLOCKED, tid)) 752 break; 753 #ifdef HWPMC_HOOKS 754 PMC_SOFT_CALL( , , lock, failed); 755 #endif 756 lock_profile_obtain_lock_failed(&lk->lock_object, 757 &contested, &waittime); 758 759 /* 760 * If the lock is expected to not sleep just give up 761 * and return. 762 */ 763 if (LK_TRYOP(flags)) { 764 LOCK_LOG2(lk, "%s: %p fails the try operation", 765 __func__, lk); 766 error = EBUSY; 767 break; 768 } 769 770 /* 771 * Acquire the sleepqueue chain lock because we 772 * probabilly will need to manipulate waiters flags. 773 */ 774 sleepq_lock(&lk->lock_object); 775 x = lk->lk_lock; 776 retry_sleepq: 777 778 /* 779 * if the lock has been released while we spun on 780 * the sleepqueue chain lock just try again. 781 */ 782 if (x == LK_UNLOCKED) { 783 sleepq_release(&lk->lock_object); 784 continue; 785 } 786 787 /* 788 * The lock can be in the state where there is a 789 * pending queue of waiters, but still no owner. 790 * This happens when the lock is contested and an 791 * owner is going to claim the lock. 792 * If curthread is the one successfully acquiring it 793 * claim lock ownership and return, preserving waiters 794 * flags. 795 */ 796 v = x & (LK_ALL_WAITERS | LK_EXCLUSIVE_SPINNERS); 797 if ((x & ~v) == LK_UNLOCKED) { 798 v &= ~LK_EXCLUSIVE_SPINNERS; 799 if (atomic_fcmpset_acq_ptr(&lk->lk_lock, &x, 800 tid | v)) { 801 sleepq_release(&lk->lock_object); 802 LOCK_LOG2(lk, 803 "%s: %p claimed by a new writer", 804 __func__, lk); 805 break; 806 } 807 goto retry_sleepq; 808 } 809 810 /* 811 * Try to set the LK_EXCLUSIVE_WAITERS flag. If we 812 * fail, loop back and retry. 813 */ 814 if ((x & LK_EXCLUSIVE_WAITERS) == 0) { 815 if (!atomic_fcmpset_ptr(&lk->lk_lock, &x, 816 x | LK_EXCLUSIVE_WAITERS)) { 817 goto retry_sleepq; 818 } 819 LOCK_LOG2(lk, "%s: %p set excl waiters flag", 820 __func__, lk); 821 } 822 823 if (lwa == NULL) { 824 iwmesg = lk->lock_object.lo_name; 825 ipri = lk->lk_pri; 826 itimo = lk->lk_timo; 827 } else { 828 iwmesg = lwa->iwmesg; 829 ipri = lwa->ipri; 830 itimo = lwa->itimo; 831 } 832 833 /* 834 * As far as we have been unable to acquire the 835 * exclusive lock and the exclusive waiters flag 836 * is set, we will sleep. 837 */ 838 #ifdef KDTRACE_HOOKS 839 sleep_time -= lockstat_nsecs(&lk->lock_object); 840 #endif 841 error = sleeplk(lk, flags, ilk, iwmesg, ipri, itimo, 842 SQ_EXCLUSIVE_QUEUE); 843 #ifdef KDTRACE_HOOKS 844 sleep_time += lockstat_nsecs(&lk->lock_object); 845 #endif 846 flags &= ~LK_INTERLOCK; 847 if (error) { 848 LOCK_LOG3(lk, 849 "%s: interrupted sleep for %p with %d", 850 __func__, lk, error); 851 break; 852 } 853 LOCK_LOG2(lk, "%s: %p resuming from the sleep queue", 854 __func__, lk); 855 } 856 if (error == 0) { 857 #ifdef KDTRACE_HOOKS 858 if (sleep_time != 0) 859 LOCKSTAT_RECORD4(lockmgr__block, lk, sleep_time, 860 LOCKSTAT_WRITER, (x & LK_SHARE) == 0, 861 (x & LK_SHARE) == 0 ? 0 : LK_SHARERS(x)); 862 #endif 863 #ifdef LOCK_PROFILING 864 lockmgr_note_exclusive_acquire(lk, contested, waittime, 865 file, line, flags); 866 #else 867 lockmgr_note_exclusive_acquire(lk, 0, 0, file, line, 868 flags); 869 #endif 870 } 871 872 out: 873 lockmgr_exit(flags, ilk, 0); 874 return (error); 875 } 876 877 static __noinline int 878 lockmgr_upgrade(struct lock *lk, u_int flags, struct lock_object *ilk, 879 const char *file, int line, struct lockmgr_wait *lwa) 880 { 881 uintptr_t tid, x, v; 882 int error = 0; 883 int wakeup_swapper = 0; 884 int op; 885 886 if (KERNEL_PANICKED()) 887 goto out; 888 889 tid = (uintptr_t)curthread; 890 891 _lockmgr_assert(lk, KA_SLOCKED, file, line); 892 v = lk->lk_lock; 893 x = v & LK_ALL_WAITERS; 894 v &= LK_EXCLUSIVE_SPINNERS; 895 896 /* 897 * Try to switch from one shared lock to an exclusive one. 898 * We need to preserve waiters flags during the operation. 899 */ 900 if (atomic_cmpset_ptr(&lk->lk_lock, LK_SHARERS_LOCK(1) | x | v, 901 tid | x)) { 902 LOCK_LOG_LOCK("XUPGRADE", &lk->lock_object, 0, 0, file, 903 line); 904 WITNESS_UPGRADE(&lk->lock_object, LOP_EXCLUSIVE | 905 LK_TRYWIT(flags), file, line); 906 LOCKSTAT_RECORD0(lockmgr__upgrade, lk); 907 TD_SLOCKS_DEC(curthread); 908 goto out; 909 } 910 911 op = flags & LK_TYPE_MASK; 912 913 /* 914 * In LK_TRYUPGRADE mode, do not drop the lock, 915 * returning EBUSY instead. 916 */ 917 if (op == LK_TRYUPGRADE) { 918 LOCK_LOG2(lk, "%s: %p failed the nowait upgrade", 919 __func__, lk); 920 error = EBUSY; 921 goto out; 922 } 923 924 /* 925 * We have been unable to succeed in upgrading, so just 926 * give up the shared lock. 927 */ 928 lockmgr_note_shared_release(lk, file, line); 929 wakeup_swapper |= wakeupshlk(lk, file, line); 930 error = lockmgr_xlock_hard(lk, flags, ilk, file, line, lwa); 931 flags &= ~LK_INTERLOCK; 932 out: 933 lockmgr_exit(flags, ilk, wakeup_swapper); 934 return (error); 935 } 936 937 int 938 lockmgr_lock_flags(struct lock *lk, u_int flags, struct lock_object *ilk, 939 const char *file, int line) 940 { 941 struct lock_class *class; 942 uintptr_t x, tid; 943 u_int op; 944 bool locked; 945 946 if (KERNEL_PANICKED()) 947 return (0); 948 949 op = flags & LK_TYPE_MASK; 950 locked = false; 951 switch (op) { 952 case LK_SHARED: 953 if (LK_CAN_WITNESS(flags)) 954 WITNESS_CHECKORDER(&lk->lock_object, LOP_NEWORDER, 955 file, line, flags & LK_INTERLOCK ? ilk : NULL); 956 if (__predict_false(lk->lock_object.lo_flags & LK_NOSHARE)) 957 break; 958 if (lockmgr_slock_try(lk, &x, flags, true)) { 959 lockmgr_note_shared_acquire(lk, 0, 0, 960 file, line, flags); 961 locked = true; 962 } else { 963 return (lockmgr_slock_hard(lk, flags, ilk, file, line, 964 NULL)); 965 } 966 break; 967 case LK_EXCLUSIVE: 968 if (LK_CAN_WITNESS(flags)) 969 WITNESS_CHECKORDER(&lk->lock_object, LOP_NEWORDER | 970 LOP_EXCLUSIVE, file, line, flags & LK_INTERLOCK ? 971 ilk : NULL); 972 tid = (uintptr_t)curthread; 973 if (lk->lk_lock == LK_UNLOCKED && 974 atomic_cmpset_acq_ptr(&lk->lk_lock, LK_UNLOCKED, tid)) { 975 lockmgr_note_exclusive_acquire(lk, 0, 0, file, line, 976 flags); 977 locked = true; 978 } else { 979 return (lockmgr_xlock_hard(lk, flags, ilk, file, line, 980 NULL)); 981 } 982 break; 983 case LK_UPGRADE: 984 case LK_TRYUPGRADE: 985 return (lockmgr_upgrade(lk, flags, ilk, file, line, NULL)); 986 default: 987 break; 988 } 989 if (__predict_true(locked)) { 990 if (__predict_false(flags & LK_INTERLOCK)) { 991 class = LOCK_CLASS(ilk); 992 class->lc_unlock(ilk); 993 } 994 return (0); 995 } else { 996 return (__lockmgr_args(lk, flags, ilk, LK_WMESG_DEFAULT, 997 LK_PRIO_DEFAULT, LK_TIMO_DEFAULT, file, line)); 998 } 999 } 1000 1001 static __noinline int 1002 lockmgr_sunlock_hard(struct lock *lk, uintptr_t x, u_int flags, struct lock_object *ilk, 1003 const char *file, int line) 1004 1005 { 1006 int wakeup_swapper = 0; 1007 1008 if (KERNEL_PANICKED()) 1009 goto out; 1010 1011 wakeup_swapper = wakeupshlk(lk, file, line); 1012 1013 out: 1014 lockmgr_exit(flags, ilk, wakeup_swapper); 1015 return (0); 1016 } 1017 1018 static __noinline int 1019 lockmgr_xunlock_hard(struct lock *lk, uintptr_t x, u_int flags, struct lock_object *ilk, 1020 const char *file, int line) 1021 { 1022 uintptr_t tid, v; 1023 int wakeup_swapper = 0; 1024 u_int realexslp; 1025 int queue; 1026 1027 if (KERNEL_PANICKED()) 1028 goto out; 1029 1030 tid = (uintptr_t)curthread; 1031 1032 /* 1033 * As first option, treact the lock as if it has not 1034 * any waiter. 1035 * Fix-up the tid var if the lock has been disowned. 1036 */ 1037 if (LK_HOLDER(x) == LK_KERNPROC) 1038 tid = LK_KERNPROC; 1039 1040 /* 1041 * The lock is held in exclusive mode. 1042 * If the lock is recursed also, then unrecurse it. 1043 */ 1044 if (lockmgr_xlocked_v(x) && lockmgr_recursed(lk)) { 1045 LOCK_LOG2(lk, "%s: %p unrecursing", __func__, lk); 1046 lk->lk_recurse--; 1047 goto out; 1048 } 1049 if (tid != LK_KERNPROC) 1050 LOCKSTAT_PROFILE_RELEASE_RWLOCK(lockmgr__release, lk, 1051 LOCKSTAT_WRITER); 1052 1053 if (x == tid && atomic_cmpset_rel_ptr(&lk->lk_lock, tid, LK_UNLOCKED)) 1054 goto out; 1055 1056 sleepq_lock(&lk->lock_object); 1057 x = lk->lk_lock; 1058 v = LK_UNLOCKED; 1059 1060 /* 1061 * If the lock has exclusive waiters, give them 1062 * preference in order to avoid deadlock with 1063 * shared runners up. 1064 * If interruptible sleeps left the exclusive queue 1065 * empty avoid a starvation for the threads sleeping 1066 * on the shared queue by giving them precedence 1067 * and cleaning up the exclusive waiters bit anyway. 1068 * Please note that lk_exslpfail count may be lying 1069 * about the real number of waiters with the 1070 * LK_SLEEPFAIL flag on because they may be used in 1071 * conjunction with interruptible sleeps so 1072 * lk_exslpfail might be considered an 'upper limit' 1073 * bound, including the edge cases. 1074 */ 1075 MPASS((x & LK_EXCLUSIVE_SPINNERS) == 0); 1076 realexslp = sleepq_sleepcnt(&lk->lock_object, SQ_EXCLUSIVE_QUEUE); 1077 if ((x & LK_EXCLUSIVE_WAITERS) != 0 && realexslp != 0) { 1078 if (lk->lk_exslpfail < realexslp) { 1079 lk->lk_exslpfail = 0; 1080 queue = SQ_EXCLUSIVE_QUEUE; 1081 v |= (x & LK_SHARED_WAITERS); 1082 } else { 1083 lk->lk_exslpfail = 0; 1084 LOCK_LOG2(lk, 1085 "%s: %p has only LK_SLEEPFAIL sleepers", 1086 __func__, lk); 1087 LOCK_LOG2(lk, 1088 "%s: %p waking up threads on the exclusive queue", 1089 __func__, lk); 1090 wakeup_swapper = sleepq_broadcast(&lk->lock_object, 1091 SLEEPQ_LK, 0, SQ_EXCLUSIVE_QUEUE); 1092 queue = SQ_SHARED_QUEUE; 1093 } 1094 } else { 1095 1096 /* 1097 * Exclusive waiters sleeping with LK_SLEEPFAIL 1098 * on and using interruptible sleeps/timeout 1099 * may have left spourious lk_exslpfail counts 1100 * on, so clean it up anyway. 1101 */ 1102 lk->lk_exslpfail = 0; 1103 queue = SQ_SHARED_QUEUE; 1104 } 1105 1106 LOCK_LOG3(lk, "%s: %p waking up threads on the %s queue", 1107 __func__, lk, queue == SQ_SHARED_QUEUE ? "shared" : 1108 "exclusive"); 1109 atomic_store_rel_ptr(&lk->lk_lock, v); 1110 wakeup_swapper |= sleepq_broadcast(&lk->lock_object, SLEEPQ_LK, 0, queue); 1111 sleepq_release(&lk->lock_object); 1112 1113 out: 1114 lockmgr_exit(flags, ilk, wakeup_swapper); 1115 return (0); 1116 } 1117 1118 /* 1119 * Lightweight entry points for common operations. 1120 * 1121 * Functionality is similar to sx locks, in that none of the additional lockmgr 1122 * features are supported. To be clear, these are NOT supported: 1123 * 1. shared locking disablement 1124 * 2. returning with an error after sleep 1125 * 3. unlocking the interlock 1126 * 1127 * If in doubt, use lockmgr_lock_flags. 1128 */ 1129 int 1130 lockmgr_slock(struct lock *lk, u_int flags, const char *file, int line) 1131 { 1132 uintptr_t x; 1133 1134 MPASS((flags & LK_TYPE_MASK) == LK_SHARED); 1135 MPASS((flags & LK_INTERLOCK) == 0); 1136 MPASS((lk->lock_object.lo_flags & LK_NOSHARE) == 0); 1137 1138 if (LK_CAN_WITNESS(flags)) 1139 WITNESS_CHECKORDER(&lk->lock_object, LOP_NEWORDER, 1140 file, line, NULL); 1141 if (__predict_true(lockmgr_slock_try(lk, &x, flags, true))) { 1142 lockmgr_note_shared_acquire(lk, 0, 0, file, line, flags); 1143 return (0); 1144 } 1145 1146 return (lockmgr_slock_hard(lk, flags, NULL, file, line, NULL)); 1147 } 1148 1149 int 1150 lockmgr_xlock(struct lock *lk, u_int flags, const char *file, int line) 1151 { 1152 uintptr_t tid; 1153 1154 MPASS((flags & LK_TYPE_MASK) == LK_EXCLUSIVE); 1155 MPASS((flags & LK_INTERLOCK) == 0); 1156 1157 if (LK_CAN_WITNESS(flags)) 1158 WITNESS_CHECKORDER(&lk->lock_object, LOP_NEWORDER | 1159 LOP_EXCLUSIVE, file, line, NULL); 1160 tid = (uintptr_t)curthread; 1161 if (atomic_cmpset_acq_ptr(&lk->lk_lock, LK_UNLOCKED, tid)) { 1162 lockmgr_note_exclusive_acquire(lk, 0, 0, file, line, 1163 flags); 1164 return (0); 1165 } 1166 1167 return (lockmgr_xlock_hard(lk, flags, NULL, file, line, NULL)); 1168 } 1169 1170 int 1171 lockmgr_unlock(struct lock *lk) 1172 { 1173 uintptr_t x, tid; 1174 const char *file; 1175 int line; 1176 1177 file = __FILE__; 1178 line = __LINE__; 1179 1180 _lockmgr_assert(lk, KA_LOCKED, file, line); 1181 x = lk->lk_lock; 1182 if (__predict_true(x & LK_SHARE) != 0) { 1183 lockmgr_note_shared_release(lk, file, line); 1184 if (lockmgr_sunlock_try(lk, &x)) { 1185 LOCKSTAT_PROFILE_RELEASE_RWLOCK(lockmgr__release, lk, LOCKSTAT_READER); 1186 } else { 1187 return (lockmgr_sunlock_hard(lk, x, LK_RELEASE, NULL, file, line)); 1188 } 1189 } else { 1190 tid = (uintptr_t)curthread; 1191 lockmgr_note_exclusive_release(lk, file, line); 1192 if (!lockmgr_recursed(lk) && 1193 atomic_cmpset_rel_ptr(&lk->lk_lock, tid, LK_UNLOCKED)) { 1194 LOCKSTAT_PROFILE_RELEASE_RWLOCK(lockmgr__release, lk, LOCKSTAT_WRITER); 1195 } else { 1196 return (lockmgr_xunlock_hard(lk, x, LK_RELEASE, NULL, file, line)); 1197 } 1198 } 1199 return (0); 1200 } 1201 1202 int 1203 __lockmgr_args(struct lock *lk, u_int flags, struct lock_object *ilk, 1204 const char *wmesg, int pri, int timo, const char *file, int line) 1205 { 1206 GIANT_DECLARE; 1207 struct lockmgr_wait lwa; 1208 struct lock_class *class; 1209 const char *iwmesg; 1210 uintptr_t tid, v, x; 1211 u_int op, realexslp; 1212 int error, ipri, itimo, queue, wakeup_swapper; 1213 #ifdef LOCK_PROFILING 1214 uint64_t waittime = 0; 1215 int contested = 0; 1216 #endif 1217 1218 if (KERNEL_PANICKED()) 1219 return (0); 1220 1221 error = 0; 1222 tid = (uintptr_t)curthread; 1223 op = (flags & LK_TYPE_MASK); 1224 iwmesg = (wmesg == LK_WMESG_DEFAULT) ? lk->lock_object.lo_name : wmesg; 1225 ipri = (pri == LK_PRIO_DEFAULT) ? lk->lk_pri : pri; 1226 itimo = (timo == LK_TIMO_DEFAULT) ? lk->lk_timo : timo; 1227 1228 lwa.iwmesg = iwmesg; 1229 lwa.ipri = ipri; 1230 lwa.itimo = itimo; 1231 1232 MPASS((flags & ~LK_TOTAL_MASK) == 0); 1233 KASSERT((op & (op - 1)) == 0, 1234 ("%s: Invalid requested operation @ %s:%d", __func__, file, line)); 1235 KASSERT((flags & (LK_NOWAIT | LK_SLEEPFAIL)) == 0 || 1236 (op != LK_DOWNGRADE && op != LK_RELEASE), 1237 ("%s: Invalid flags in regard of the operation desired @ %s:%d", 1238 __func__, file, line)); 1239 KASSERT((flags & LK_INTERLOCK) == 0 || ilk != NULL, 1240 ("%s: LK_INTERLOCK passed without valid interlock @ %s:%d", 1241 __func__, file, line)); 1242 KASSERT(kdb_active != 0 || !TD_IS_IDLETHREAD(curthread), 1243 ("%s: idle thread %p on lockmgr %s @ %s:%d", __func__, curthread, 1244 lk->lock_object.lo_name, file, line)); 1245 1246 class = (flags & LK_INTERLOCK) ? LOCK_CLASS(ilk) : NULL; 1247 1248 if (lk->lock_object.lo_flags & LK_NOSHARE) { 1249 switch (op) { 1250 case LK_SHARED: 1251 op = LK_EXCLUSIVE; 1252 break; 1253 case LK_UPGRADE: 1254 case LK_TRYUPGRADE: 1255 case LK_DOWNGRADE: 1256 _lockmgr_assert(lk, KA_XLOCKED | KA_NOTRECURSED, 1257 file, line); 1258 if (flags & LK_INTERLOCK) 1259 class->lc_unlock(ilk); 1260 return (0); 1261 } 1262 } 1263 1264 wakeup_swapper = 0; 1265 switch (op) { 1266 case LK_SHARED: 1267 return (lockmgr_slock_hard(lk, flags, ilk, file, line, &lwa)); 1268 break; 1269 case LK_UPGRADE: 1270 case LK_TRYUPGRADE: 1271 return (lockmgr_upgrade(lk, flags, ilk, file, line, &lwa)); 1272 break; 1273 case LK_EXCLUSIVE: 1274 return (lockmgr_xlock_hard(lk, flags, ilk, file, line, &lwa)); 1275 break; 1276 case LK_DOWNGRADE: 1277 _lockmgr_assert(lk, KA_XLOCKED, file, line); 1278 WITNESS_DOWNGRADE(&lk->lock_object, 0, file, line); 1279 1280 /* 1281 * Panic if the lock is recursed. 1282 */ 1283 if (lockmgr_xlocked(lk) && lockmgr_recursed(lk)) { 1284 if (flags & LK_INTERLOCK) 1285 class->lc_unlock(ilk); 1286 panic("%s: downgrade a recursed lockmgr %s @ %s:%d\n", 1287 __func__, iwmesg, file, line); 1288 } 1289 TD_SLOCKS_INC(curthread); 1290 1291 /* 1292 * In order to preserve waiters flags, just spin. 1293 */ 1294 for (;;) { 1295 x = lk->lk_lock; 1296 MPASS((x & LK_EXCLUSIVE_SPINNERS) == 0); 1297 x &= LK_ALL_WAITERS; 1298 if (atomic_cmpset_rel_ptr(&lk->lk_lock, tid | x, 1299 LK_SHARERS_LOCK(1) | x)) 1300 break; 1301 cpu_spinwait(); 1302 } 1303 LOCK_LOG_LOCK("XDOWNGRADE", &lk->lock_object, 0, 0, file, line); 1304 LOCKSTAT_RECORD0(lockmgr__downgrade, lk); 1305 break; 1306 case LK_RELEASE: 1307 _lockmgr_assert(lk, KA_LOCKED, file, line); 1308 x = lk->lk_lock; 1309 1310 if (__predict_true(x & LK_SHARE) != 0) { 1311 lockmgr_note_shared_release(lk, file, line); 1312 return (lockmgr_sunlock_hard(lk, x, flags, ilk, file, line)); 1313 } else { 1314 lockmgr_note_exclusive_release(lk, file, line); 1315 return (lockmgr_xunlock_hard(lk, x, flags, ilk, file, line)); 1316 } 1317 break; 1318 case LK_DRAIN: 1319 if (LK_CAN_WITNESS(flags)) 1320 WITNESS_CHECKORDER(&lk->lock_object, LOP_NEWORDER | 1321 LOP_EXCLUSIVE, file, line, flags & LK_INTERLOCK ? 1322 ilk : NULL); 1323 1324 /* 1325 * Trying to drain a lock we already own will result in a 1326 * deadlock. 1327 */ 1328 if (lockmgr_xlocked(lk)) { 1329 if (flags & LK_INTERLOCK) 1330 class->lc_unlock(ilk); 1331 panic("%s: draining %s with the lock held @ %s:%d\n", 1332 __func__, iwmesg, file, line); 1333 } 1334 1335 for (;;) { 1336 if (lk->lk_lock == LK_UNLOCKED && 1337 atomic_cmpset_acq_ptr(&lk->lk_lock, LK_UNLOCKED, tid)) 1338 break; 1339 1340 #ifdef HWPMC_HOOKS 1341 PMC_SOFT_CALL( , , lock, failed); 1342 #endif 1343 lock_profile_obtain_lock_failed(&lk->lock_object, 1344 &contested, &waittime); 1345 1346 /* 1347 * If the lock is expected to not sleep just give up 1348 * and return. 1349 */ 1350 if (LK_TRYOP(flags)) { 1351 LOCK_LOG2(lk, "%s: %p fails the try operation", 1352 __func__, lk); 1353 error = EBUSY; 1354 break; 1355 } 1356 1357 /* 1358 * Acquire the sleepqueue chain lock because we 1359 * probabilly will need to manipulate waiters flags. 1360 */ 1361 sleepq_lock(&lk->lock_object); 1362 x = lk->lk_lock; 1363 1364 /* 1365 * if the lock has been released while we spun on 1366 * the sleepqueue chain lock just try again. 1367 */ 1368 if (x == LK_UNLOCKED) { 1369 sleepq_release(&lk->lock_object); 1370 continue; 1371 } 1372 1373 v = x & (LK_ALL_WAITERS | LK_EXCLUSIVE_SPINNERS); 1374 if ((x & ~v) == LK_UNLOCKED) { 1375 v = (x & ~LK_EXCLUSIVE_SPINNERS); 1376 1377 /* 1378 * If interruptible sleeps left the exclusive 1379 * queue empty avoid a starvation for the 1380 * threads sleeping on the shared queue by 1381 * giving them precedence and cleaning up the 1382 * exclusive waiters bit anyway. 1383 * Please note that lk_exslpfail count may be 1384 * lying about the real number of waiters with 1385 * the LK_SLEEPFAIL flag on because they may 1386 * be used in conjunction with interruptible 1387 * sleeps so lk_exslpfail might be considered 1388 * an 'upper limit' bound, including the edge 1389 * cases. 1390 */ 1391 if (v & LK_EXCLUSIVE_WAITERS) { 1392 queue = SQ_EXCLUSIVE_QUEUE; 1393 v &= ~LK_EXCLUSIVE_WAITERS; 1394 } else { 1395 1396 /* 1397 * Exclusive waiters sleeping with 1398 * LK_SLEEPFAIL on and using 1399 * interruptible sleeps/timeout may 1400 * have left spourious lk_exslpfail 1401 * counts on, so clean it up anyway. 1402 */ 1403 MPASS(v & LK_SHARED_WAITERS); 1404 lk->lk_exslpfail = 0; 1405 queue = SQ_SHARED_QUEUE; 1406 v &= ~LK_SHARED_WAITERS; 1407 } 1408 if (queue == SQ_EXCLUSIVE_QUEUE) { 1409 realexslp = 1410 sleepq_sleepcnt(&lk->lock_object, 1411 SQ_EXCLUSIVE_QUEUE); 1412 if (lk->lk_exslpfail >= realexslp) { 1413 lk->lk_exslpfail = 0; 1414 queue = SQ_SHARED_QUEUE; 1415 v &= ~LK_SHARED_WAITERS; 1416 if (realexslp != 0) { 1417 LOCK_LOG2(lk, 1418 "%s: %p has only LK_SLEEPFAIL sleepers", 1419 __func__, lk); 1420 LOCK_LOG2(lk, 1421 "%s: %p waking up threads on the exclusive queue", 1422 __func__, lk); 1423 wakeup_swapper = 1424 sleepq_broadcast( 1425 &lk->lock_object, 1426 SLEEPQ_LK, 0, 1427 SQ_EXCLUSIVE_QUEUE); 1428 } 1429 } else 1430 lk->lk_exslpfail = 0; 1431 } 1432 if (!atomic_cmpset_ptr(&lk->lk_lock, x, v)) { 1433 sleepq_release(&lk->lock_object); 1434 continue; 1435 } 1436 LOCK_LOG3(lk, 1437 "%s: %p waking up all threads on the %s queue", 1438 __func__, lk, queue == SQ_SHARED_QUEUE ? 1439 "shared" : "exclusive"); 1440 wakeup_swapper |= sleepq_broadcast( 1441 &lk->lock_object, SLEEPQ_LK, 0, queue); 1442 1443 /* 1444 * If shared waiters have been woken up we need 1445 * to wait for one of them to acquire the lock 1446 * before to set the exclusive waiters in 1447 * order to avoid a deadlock. 1448 */ 1449 if (queue == SQ_SHARED_QUEUE) { 1450 for (v = lk->lk_lock; 1451 (v & LK_SHARE) && !LK_SHARERS(v); 1452 v = lk->lk_lock) 1453 cpu_spinwait(); 1454 } 1455 } 1456 1457 /* 1458 * Try to set the LK_EXCLUSIVE_WAITERS flag. If we 1459 * fail, loop back and retry. 1460 */ 1461 if ((x & LK_EXCLUSIVE_WAITERS) == 0) { 1462 if (!atomic_cmpset_ptr(&lk->lk_lock, x, 1463 x | LK_EXCLUSIVE_WAITERS)) { 1464 sleepq_release(&lk->lock_object); 1465 continue; 1466 } 1467 LOCK_LOG2(lk, "%s: %p set drain waiters flag", 1468 __func__, lk); 1469 } 1470 1471 /* 1472 * As far as we have been unable to acquire the 1473 * exclusive lock and the exclusive waiters flag 1474 * is set, we will sleep. 1475 */ 1476 if (flags & LK_INTERLOCK) { 1477 class->lc_unlock(ilk); 1478 flags &= ~LK_INTERLOCK; 1479 } 1480 GIANT_SAVE(); 1481 sleepq_add(&lk->lock_object, NULL, iwmesg, SLEEPQ_LK, 1482 SQ_EXCLUSIVE_QUEUE); 1483 sleepq_wait(&lk->lock_object, ipri & PRIMASK); 1484 GIANT_RESTORE(); 1485 LOCK_LOG2(lk, "%s: %p resuming from the sleep queue", 1486 __func__, lk); 1487 } 1488 1489 if (error == 0) { 1490 lock_profile_obtain_lock_success(&lk->lock_object, 1491 contested, waittime, file, line); 1492 LOCK_LOG_LOCK("DRAIN", &lk->lock_object, 0, 1493 lk->lk_recurse, file, line); 1494 WITNESS_LOCK(&lk->lock_object, LOP_EXCLUSIVE | 1495 LK_TRYWIT(flags), file, line); 1496 TD_LOCKS_INC(curthread); 1497 STACK_SAVE(lk); 1498 } 1499 break; 1500 default: 1501 if (flags & LK_INTERLOCK) 1502 class->lc_unlock(ilk); 1503 panic("%s: unknown lockmgr request 0x%x\n", __func__, op); 1504 } 1505 1506 if (flags & LK_INTERLOCK) 1507 class->lc_unlock(ilk); 1508 if (wakeup_swapper) 1509 kick_proc0(); 1510 1511 return (error); 1512 } 1513 1514 void 1515 _lockmgr_disown(struct lock *lk, const char *file, int line) 1516 { 1517 uintptr_t tid, x; 1518 1519 if (SCHEDULER_STOPPED()) 1520 return; 1521 1522 tid = (uintptr_t)curthread; 1523 _lockmgr_assert(lk, KA_XLOCKED, file, line); 1524 1525 /* 1526 * Panic if the lock is recursed. 1527 */ 1528 if (lockmgr_xlocked(lk) && lockmgr_recursed(lk)) 1529 panic("%s: disown a recursed lockmgr @ %s:%d\n", 1530 __func__, file, line); 1531 1532 /* 1533 * If the owner is already LK_KERNPROC just skip the whole operation. 1534 */ 1535 if (LK_HOLDER(lk->lk_lock) != tid) 1536 return; 1537 lock_profile_release_lock(&lk->lock_object); 1538 LOCKSTAT_RECORD1(lockmgr__disown, lk, LOCKSTAT_WRITER); 1539 LOCK_LOG_LOCK("XDISOWN", &lk->lock_object, 0, 0, file, line); 1540 WITNESS_UNLOCK(&lk->lock_object, LOP_EXCLUSIVE, file, line); 1541 TD_LOCKS_DEC(curthread); 1542 STACK_SAVE(lk); 1543 1544 /* 1545 * In order to preserve waiters flags, just spin. 1546 */ 1547 for (;;) { 1548 x = lk->lk_lock; 1549 MPASS((x & LK_EXCLUSIVE_SPINNERS) == 0); 1550 x &= LK_ALL_WAITERS; 1551 if (atomic_cmpset_rel_ptr(&lk->lk_lock, tid | x, 1552 LK_KERNPROC | x)) 1553 return; 1554 cpu_spinwait(); 1555 } 1556 } 1557 1558 void 1559 lockmgr_printinfo(const struct lock *lk) 1560 { 1561 struct thread *td; 1562 uintptr_t x; 1563 1564 if (lk->lk_lock == LK_UNLOCKED) 1565 printf("lock type %s: UNLOCKED\n", lk->lock_object.lo_name); 1566 else if (lk->lk_lock & LK_SHARE) 1567 printf("lock type %s: SHARED (count %ju)\n", 1568 lk->lock_object.lo_name, 1569 (uintmax_t)LK_SHARERS(lk->lk_lock)); 1570 else { 1571 td = lockmgr_xholder(lk); 1572 if (td == (struct thread *)LK_KERNPROC) 1573 printf("lock type %s: EXCL by KERNPROC\n", 1574 lk->lock_object.lo_name); 1575 else 1576 printf("lock type %s: EXCL by thread %p " 1577 "(pid %d, %s, tid %d)\n", lk->lock_object.lo_name, 1578 td, td->td_proc->p_pid, td->td_proc->p_comm, 1579 td->td_tid); 1580 } 1581 1582 x = lk->lk_lock; 1583 if (x & LK_EXCLUSIVE_WAITERS) 1584 printf(" with exclusive waiters pending\n"); 1585 if (x & LK_SHARED_WAITERS) 1586 printf(" with shared waiters pending\n"); 1587 if (x & LK_EXCLUSIVE_SPINNERS) 1588 printf(" with exclusive spinners pending\n"); 1589 1590 STACK_PRINT(lk); 1591 } 1592 1593 int 1594 lockstatus(const struct lock *lk) 1595 { 1596 uintptr_t v, x; 1597 int ret; 1598 1599 ret = LK_SHARED; 1600 x = lk->lk_lock; 1601 v = LK_HOLDER(x); 1602 1603 if ((x & LK_SHARE) == 0) { 1604 if (v == (uintptr_t)curthread || v == LK_KERNPROC) 1605 ret = LK_EXCLUSIVE; 1606 else 1607 ret = LK_EXCLOTHER; 1608 } else if (x == LK_UNLOCKED) 1609 ret = 0; 1610 1611 return (ret); 1612 } 1613 1614 #ifdef INVARIANT_SUPPORT 1615 1616 FEATURE(invariant_support, 1617 "Support for modules compiled with INVARIANTS option"); 1618 1619 #ifndef INVARIANTS 1620 #undef _lockmgr_assert 1621 #endif 1622 1623 void 1624 _lockmgr_assert(const struct lock *lk, int what, const char *file, int line) 1625 { 1626 int slocked = 0; 1627 1628 if (KERNEL_PANICKED()) 1629 return; 1630 switch (what) { 1631 case KA_SLOCKED: 1632 case KA_SLOCKED | KA_NOTRECURSED: 1633 case KA_SLOCKED | KA_RECURSED: 1634 slocked = 1; 1635 case KA_LOCKED: 1636 case KA_LOCKED | KA_NOTRECURSED: 1637 case KA_LOCKED | KA_RECURSED: 1638 #ifdef WITNESS 1639 1640 /* 1641 * We cannot trust WITNESS if the lock is held in exclusive 1642 * mode and a call to lockmgr_disown() happened. 1643 * Workaround this skipping the check if the lock is held in 1644 * exclusive mode even for the KA_LOCKED case. 1645 */ 1646 if (slocked || (lk->lk_lock & LK_SHARE)) { 1647 witness_assert(&lk->lock_object, what, file, line); 1648 break; 1649 } 1650 #endif 1651 if (lk->lk_lock == LK_UNLOCKED || 1652 ((lk->lk_lock & LK_SHARE) == 0 && (slocked || 1653 (!lockmgr_xlocked(lk) && !lockmgr_disowned(lk))))) 1654 panic("Lock %s not %slocked @ %s:%d\n", 1655 lk->lock_object.lo_name, slocked ? "share" : "", 1656 file, line); 1657 1658 if ((lk->lk_lock & LK_SHARE) == 0) { 1659 if (lockmgr_recursed(lk)) { 1660 if (what & KA_NOTRECURSED) 1661 panic("Lock %s recursed @ %s:%d\n", 1662 lk->lock_object.lo_name, file, 1663 line); 1664 } else if (what & KA_RECURSED) 1665 panic("Lock %s not recursed @ %s:%d\n", 1666 lk->lock_object.lo_name, file, line); 1667 } 1668 break; 1669 case KA_XLOCKED: 1670 case KA_XLOCKED | KA_NOTRECURSED: 1671 case KA_XLOCKED | KA_RECURSED: 1672 if (!lockmgr_xlocked(lk) && !lockmgr_disowned(lk)) 1673 panic("Lock %s not exclusively locked @ %s:%d\n", 1674 lk->lock_object.lo_name, file, line); 1675 if (lockmgr_recursed(lk)) { 1676 if (what & KA_NOTRECURSED) 1677 panic("Lock %s recursed @ %s:%d\n", 1678 lk->lock_object.lo_name, file, line); 1679 } else if (what & KA_RECURSED) 1680 panic("Lock %s not recursed @ %s:%d\n", 1681 lk->lock_object.lo_name, file, line); 1682 break; 1683 case KA_UNLOCKED: 1684 if (lockmgr_xlocked(lk) || lockmgr_disowned(lk)) 1685 panic("Lock %s exclusively locked @ %s:%d\n", 1686 lk->lock_object.lo_name, file, line); 1687 break; 1688 default: 1689 panic("Unknown lockmgr assertion: %d @ %s:%d\n", what, file, 1690 line); 1691 } 1692 } 1693 #endif 1694 1695 #ifdef DDB 1696 int 1697 lockmgr_chain(struct thread *td, struct thread **ownerp) 1698 { 1699 const struct lock *lk; 1700 1701 lk = td->td_wchan; 1702 1703 if (LOCK_CLASS(&lk->lock_object) != &lock_class_lockmgr) 1704 return (0); 1705 db_printf("blocked on lockmgr %s", lk->lock_object.lo_name); 1706 if (lk->lk_lock & LK_SHARE) 1707 db_printf("SHARED (count %ju)\n", 1708 (uintmax_t)LK_SHARERS(lk->lk_lock)); 1709 else 1710 db_printf("EXCL\n"); 1711 *ownerp = lockmgr_xholder(lk); 1712 1713 return (1); 1714 } 1715 1716 static void 1717 db_show_lockmgr(const struct lock_object *lock) 1718 { 1719 struct thread *td; 1720 const struct lock *lk; 1721 1722 lk = (const struct lock *)lock; 1723 1724 db_printf(" state: "); 1725 if (lk->lk_lock == LK_UNLOCKED) 1726 db_printf("UNLOCKED\n"); 1727 else if (lk->lk_lock & LK_SHARE) 1728 db_printf("SLOCK: %ju\n", (uintmax_t)LK_SHARERS(lk->lk_lock)); 1729 else { 1730 td = lockmgr_xholder(lk); 1731 if (td == (struct thread *)LK_KERNPROC) 1732 db_printf("XLOCK: LK_KERNPROC\n"); 1733 else 1734 db_printf("XLOCK: %p (tid %d, pid %d, \"%s\")\n", td, 1735 td->td_tid, td->td_proc->p_pid, 1736 td->td_proc->p_comm); 1737 if (lockmgr_recursed(lk)) 1738 db_printf(" recursed: %d\n", lk->lk_recurse); 1739 } 1740 db_printf(" waiters: "); 1741 switch (lk->lk_lock & LK_ALL_WAITERS) { 1742 case LK_SHARED_WAITERS: 1743 db_printf("shared\n"); 1744 break; 1745 case LK_EXCLUSIVE_WAITERS: 1746 db_printf("exclusive\n"); 1747 break; 1748 case LK_ALL_WAITERS: 1749 db_printf("shared and exclusive\n"); 1750 break; 1751 default: 1752 db_printf("none\n"); 1753 } 1754 db_printf(" spinners: "); 1755 if (lk->lk_lock & LK_EXCLUSIVE_SPINNERS) 1756 db_printf("exclusive\n"); 1757 else 1758 db_printf("none\n"); 1759 } 1760 #endif 1761