sys/kern/kern_lock.c

   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(lockmgr_read_value(lk))
 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(lockmgr_read_value(lk)) != 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 = lockmgr_read_value(lk);
 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 = lockmgr_read_value(lk);
 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 = lockmgr_read_value(lk);
 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 = lockmgr_read_value(lk);
 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 = lockmgr_read_value(lk);
 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 = lockmgr_read_value(lk);
 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 = lockmgr_read_value(lk);
 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 (lockmgr_read_value(lk) == 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 = lockmgr_read_value(lk);
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 = lockmgr_read_value(lk);
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 = lockmgr_read_value(lk);
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 = lockmgr_read_value(lk);
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 = lockmgr_read_value(lk);
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 = lockmgr_read_value(lk);
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 = lockmgr_read_value(lk);
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