]> CyberLeo.Net >> Repos - FreeBSD/stable/8.git/blob - sys/kern/kern_sx.c
MFC r196772:
[FreeBSD/stable/8.git] / sys / kern / kern_sx.c
1 /*-
2  * Copyright (c) 2007 Attilio Rao <attilio@freebsd.org>
3  * Copyright (c) 2001 Jason Evans <jasone@freebsd.org>
4  * All rights reserved.
5  *
6  * Redistribution and use in source and binary forms, with or without
7  * modification, are permitted provided that the following conditions
8  * are met:
9  * 1. Redistributions of source code must retain the above copyright
10  *    notice(s), this list of conditions and the following disclaimer as
11  *    the first lines of this file unmodified other than the possible
12  *    addition of one or more copyright notices.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice(s), this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY
18  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
19  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
20  * DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY
21  * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
22  * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
23  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
24  * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
27  * DAMAGE.
28  */
29
30 /*
31  * Shared/exclusive locks.  This implementation attempts to ensure
32  * deterministic lock granting behavior, so that slocks and xlocks are
33  * interleaved.
34  *
35  * Priority propagation will not generally raise the priority of lock holders,
36  * so should not be relied upon in combination with sx locks.
37  */
38
39 #include "opt_ddb.h"
40 #include "opt_kdtrace.h"
41 #include "opt_no_adaptive_sx.h"
42
43 #include <sys/cdefs.h>
44 __FBSDID("$FreeBSD$");
45
46 #include <sys/param.h>
47 #include <sys/ktr.h>
48 #include <sys/linker_set.h>
49 #include <sys/lock.h>
50 #include <sys/mutex.h>
51 #include <sys/proc.h>
52 #include <sys/sleepqueue.h>
53 #include <sys/sx.h>
54 #include <sys/sysctl.h>
55 #include <sys/systm.h>
56
57 #if defined(SMP) && !defined(NO_ADAPTIVE_SX)
58 #include <machine/cpu.h>
59 #endif
60
61 #ifdef DDB
62 #include <ddb/ddb.h>
63 #endif
64
65 #if defined(SMP) && !defined(NO_ADAPTIVE_SX)
66 #define ADAPTIVE_SX
67 #endif
68
69 CTASSERT((SX_NOADAPTIVE & LO_CLASSFLAGS) == SX_NOADAPTIVE);
70
71 /* Handy macros for sleep queues. */
72 #define SQ_EXCLUSIVE_QUEUE      0
73 #define SQ_SHARED_QUEUE         1
74
75 /*
76  * Variations on DROP_GIANT()/PICKUP_GIANT() for use in this file.  We
77  * drop Giant anytime we have to sleep or if we adaptively spin.
78  */
79 #define GIANT_DECLARE                                                   \
80         int _giantcnt = 0;                                              \
81         WITNESS_SAVE_DECL(Giant)                                        \
82
83 #define GIANT_SAVE() do {                                               \
84         if (mtx_owned(&Giant)) {                                        \
85                 WITNESS_SAVE(&Giant.lock_object, Giant);                \
86                 while (mtx_owned(&Giant)) {                             \
87                         _giantcnt++;                                    \
88                         mtx_unlock(&Giant);                             \
89                 }                                                       \
90         }                                                               \
91 } while (0)
92
93 #define GIANT_RESTORE() do {                                            \
94         if (_giantcnt > 0) {                                            \
95                 mtx_assert(&Giant, MA_NOTOWNED);                        \
96                 while (_giantcnt--)                                     \
97                         mtx_lock(&Giant);                               \
98                 WITNESS_RESTORE(&Giant.lock_object, Giant);             \
99         }                                                               \
100 } while (0)
101
102 /*
103  * Returns true if an exclusive lock is recursed.  It assumes
104  * curthread currently has an exclusive lock.
105  */
106 #define sx_recurse              lock_object.lo_data
107 #define sx_recursed(sx)         ((sx)->sx_recurse != 0)
108
109 static void     assert_sx(struct lock_object *lock, int what);
110 #ifdef DDB
111 static void     db_show_sx(struct lock_object *lock);
112 #endif
113 static void     lock_sx(struct lock_object *lock, int how);
114 #ifdef KDTRACE_HOOKS
115 static int      owner_sx(struct lock_object *lock, struct thread **owner);
116 #endif
117 static int      unlock_sx(struct lock_object *lock);
118
119 struct lock_class lock_class_sx = {
120         .lc_name = "sx",
121         .lc_flags = LC_SLEEPLOCK | LC_SLEEPABLE | LC_RECURSABLE | LC_UPGRADABLE,
122         .lc_assert = assert_sx,
123 #ifdef DDB
124         .lc_ddb_show = db_show_sx,
125 #endif
126         .lc_lock = lock_sx,
127         .lc_unlock = unlock_sx,
128 #ifdef KDTRACE_HOOKS
129         .lc_owner = owner_sx,
130 #endif
131 };
132
133 #ifndef INVARIANTS
134 #define _sx_assert(sx, what, file, line)
135 #endif
136
137 #ifdef ADAPTIVE_SX
138 static u_int asx_retries = 10;
139 static u_int asx_loops = 10000;
140 SYSCTL_NODE(_debug, OID_AUTO, sx, CTLFLAG_RD, NULL, "sxlock debugging");
141 SYSCTL_INT(_debug_sx, OID_AUTO, retries, CTLFLAG_RW, &asx_retries, 0, "");
142 SYSCTL_INT(_debug_sx, OID_AUTO, loops, CTLFLAG_RW, &asx_loops, 0, "");
143 #endif
144
145 void
146 assert_sx(struct lock_object *lock, int what)
147 {
148
149         sx_assert((struct sx *)lock, what);
150 }
151
152 void
153 lock_sx(struct lock_object *lock, int how)
154 {
155         struct sx *sx;
156
157         sx = (struct sx *)lock;
158         if (how)
159                 sx_xlock(sx);
160         else
161                 sx_slock(sx);
162 }
163
164 int
165 unlock_sx(struct lock_object *lock)
166 {
167         struct sx *sx;
168
169         sx = (struct sx *)lock;
170         sx_assert(sx, SA_LOCKED | SA_NOTRECURSED);
171         if (sx_xlocked(sx)) {
172                 sx_xunlock(sx);
173                 return (1);
174         } else {
175                 sx_sunlock(sx);
176                 return (0);
177         }
178 }
179
180 #ifdef KDTRACE_HOOKS
181 int
182 owner_sx(struct lock_object *lock, struct thread **owner)
183 {
184         struct sx *sx = (struct sx *)lock;
185         uintptr_t x = sx->sx_lock;
186
187         *owner = (struct thread *)SX_OWNER(x);
188         return ((x & SX_LOCK_SHARED) != 0 ? (SX_SHARERS(x) != 0) :
189             (*owner != NULL));
190 }
191 #endif
192
193 void
194 sx_sysinit(void *arg)
195 {
196         struct sx_args *sargs = arg;
197
198         sx_init(sargs->sa_sx, sargs->sa_desc);
199 }
200
201 void
202 sx_init_flags(struct sx *sx, const char *description, int opts)
203 {
204         int flags;
205
206         MPASS((opts & ~(SX_QUIET | SX_RECURSE | SX_NOWITNESS | SX_DUPOK |
207             SX_NOPROFILE | SX_NOADAPTIVE)) == 0);
208         ASSERT_ATOMIC_LOAD_PTR(sx->sx_lock,
209             ("%s: sx_lock not aligned for %s: %p", __func__, description,
210             &sx->sx_lock));
211
212         flags = LO_SLEEPABLE | LO_UPGRADABLE;
213         if (opts & SX_DUPOK)
214                 flags |= LO_DUPOK;
215         if (opts & SX_NOPROFILE)
216                 flags |= LO_NOPROFILE;
217         if (!(opts & SX_NOWITNESS))
218                 flags |= LO_WITNESS;
219         if (opts & SX_RECURSE)
220                 flags |= LO_RECURSABLE;
221         if (opts & SX_QUIET)
222                 flags |= LO_QUIET;
223
224         flags |= opts & SX_NOADAPTIVE;
225         sx->sx_lock = SX_LOCK_UNLOCKED;
226         sx->sx_recurse = 0;
227         lock_init(&sx->lock_object, &lock_class_sx, description, NULL, flags);
228 }
229
230 void
231 sx_destroy(struct sx *sx)
232 {
233
234         KASSERT(sx->sx_lock == SX_LOCK_UNLOCKED, ("sx lock still held"));
235         KASSERT(sx->sx_recurse == 0, ("sx lock still recursed"));
236         sx->sx_lock = SX_LOCK_DESTROYED;
237         lock_destroy(&sx->lock_object);
238 }
239
240 int
241 _sx_slock(struct sx *sx, int opts, const char *file, int line)
242 {
243         int error = 0;
244
245         MPASS(curthread != NULL);
246         KASSERT(sx->sx_lock != SX_LOCK_DESTROYED,
247             ("sx_slock() of destroyed sx @ %s:%d", file, line));
248         WITNESS_CHECKORDER(&sx->lock_object, LOP_NEWORDER, file, line, NULL);
249         error = __sx_slock(sx, opts, file, line);
250         if (!error) {
251                 LOCK_LOG_LOCK("SLOCK", &sx->lock_object, 0, 0, file, line);
252                 WITNESS_LOCK(&sx->lock_object, 0, file, line);
253                 curthread->td_locks++;
254         }
255
256         return (error);
257 }
258
259 int
260 _sx_try_slock(struct sx *sx, const char *file, int line)
261 {
262         uintptr_t x;
263
264         for (;;) {
265                 x = sx->sx_lock;
266                 KASSERT(x != SX_LOCK_DESTROYED,
267                     ("sx_try_slock() of destroyed sx @ %s:%d", file, line));
268                 if (!(x & SX_LOCK_SHARED))
269                         break;
270                 if (atomic_cmpset_acq_ptr(&sx->sx_lock, x, x + SX_ONE_SHARER)) {
271                         LOCK_LOG_TRY("SLOCK", &sx->lock_object, 0, 1, file, line);
272                         WITNESS_LOCK(&sx->lock_object, LOP_TRYLOCK, file, line);
273                         curthread->td_locks++;
274                         return (1);
275                 }
276         }
277
278         LOCK_LOG_TRY("SLOCK", &sx->lock_object, 0, 0, file, line);
279         return (0);
280 }
281
282 int
283 _sx_xlock(struct sx *sx, int opts, const char *file, int line)
284 {
285         int error = 0;
286
287         MPASS(curthread != NULL);
288         KASSERT(sx->sx_lock != SX_LOCK_DESTROYED,
289             ("sx_xlock() of destroyed sx @ %s:%d", file, line));
290         WITNESS_CHECKORDER(&sx->lock_object, LOP_NEWORDER | LOP_EXCLUSIVE, file,
291             line, NULL);
292         error = __sx_xlock(sx, curthread, opts, file, line);
293         if (!error) {
294                 LOCK_LOG_LOCK("XLOCK", &sx->lock_object, 0, sx->sx_recurse,
295                     file, line);
296                 WITNESS_LOCK(&sx->lock_object, LOP_EXCLUSIVE, file, line);
297                 curthread->td_locks++;
298         }
299
300         return (error);
301 }
302
303 int
304 _sx_try_xlock(struct sx *sx, const char *file, int line)
305 {
306         int rval;
307
308         MPASS(curthread != NULL);
309         KASSERT(sx->sx_lock != SX_LOCK_DESTROYED,
310             ("sx_try_xlock() of destroyed sx @ %s:%d", file, line));
311
312         if (sx_xlocked(sx) &&
313             (sx->lock_object.lo_flags & LO_RECURSABLE) != 0) {
314                 sx->sx_recurse++;
315                 atomic_set_ptr(&sx->sx_lock, SX_LOCK_RECURSED);
316                 rval = 1;
317         } else
318                 rval = atomic_cmpset_acq_ptr(&sx->sx_lock, SX_LOCK_UNLOCKED,
319                     (uintptr_t)curthread);
320         LOCK_LOG_TRY("XLOCK", &sx->lock_object, 0, rval, file, line);
321         if (rval) {
322                 WITNESS_LOCK(&sx->lock_object, LOP_EXCLUSIVE | LOP_TRYLOCK,
323                     file, line);
324                 curthread->td_locks++;
325         }
326
327         return (rval);
328 }
329
330 void
331 _sx_sunlock(struct sx *sx, const char *file, int line)
332 {
333
334         MPASS(curthread != NULL);
335         KASSERT(sx->sx_lock != SX_LOCK_DESTROYED,
336             ("sx_sunlock() of destroyed sx @ %s:%d", file, line));
337         _sx_assert(sx, SA_SLOCKED, file, line);
338         curthread->td_locks--;
339         WITNESS_UNLOCK(&sx->lock_object, 0, file, line);
340         LOCK_LOG_LOCK("SUNLOCK", &sx->lock_object, 0, 0, file, line);
341         __sx_sunlock(sx, file, line);
342         LOCKSTAT_PROFILE_RELEASE_LOCK(LS_SX_SUNLOCK_RELEASE, sx);
343 }
344
345 void
346 _sx_xunlock(struct sx *sx, const char *file, int line)
347 {
348
349         MPASS(curthread != NULL);
350         KASSERT(sx->sx_lock != SX_LOCK_DESTROYED,
351             ("sx_xunlock() of destroyed sx @ %s:%d", file, line));
352         _sx_assert(sx, SA_XLOCKED, file, line);
353         curthread->td_locks--;
354         WITNESS_UNLOCK(&sx->lock_object, LOP_EXCLUSIVE, file, line);
355         LOCK_LOG_LOCK("XUNLOCK", &sx->lock_object, 0, sx->sx_recurse, file,
356             line);
357         if (!sx_recursed(sx))
358                 LOCKSTAT_PROFILE_RELEASE_LOCK(LS_SX_XUNLOCK_RELEASE, sx);
359         __sx_xunlock(sx, curthread, file, line);
360 }
361
362 /*
363  * Try to do a non-blocking upgrade from a shared lock to an exclusive lock.
364  * This will only succeed if this thread holds a single shared lock.
365  * Return 1 if if the upgrade succeed, 0 otherwise.
366  */
367 int
368 _sx_try_upgrade(struct sx *sx, const char *file, int line)
369 {
370         uintptr_t x;
371         int success;
372
373         KASSERT(sx->sx_lock != SX_LOCK_DESTROYED,
374             ("sx_try_upgrade() of destroyed sx @ %s:%d", file, line));
375         _sx_assert(sx, SA_SLOCKED, file, line);
376
377         /*
378          * Try to switch from one shared lock to an exclusive lock.  We need
379          * to maintain the SX_LOCK_EXCLUSIVE_WAITERS flag if set so that
380          * we will wake up the exclusive waiters when we drop the lock.
381          */
382         x = sx->sx_lock & SX_LOCK_EXCLUSIVE_WAITERS;
383         success = atomic_cmpset_ptr(&sx->sx_lock, SX_SHARERS_LOCK(1) | x,
384             (uintptr_t)curthread | x);
385         LOCK_LOG_TRY("XUPGRADE", &sx->lock_object, 0, success, file, line);
386         if (success) {
387                 WITNESS_UPGRADE(&sx->lock_object, LOP_EXCLUSIVE | LOP_TRYLOCK,
388                     file, line);
389                 LOCKSTAT_RECORD0(LS_SX_TRYUPGRADE_UPGRADE, sx);
390         }
391         return (success);
392 }
393
394 /*
395  * Downgrade an unrecursed exclusive lock into a single shared lock.
396  */
397 void
398 _sx_downgrade(struct sx *sx, const char *file, int line)
399 {
400         uintptr_t x;
401         int wakeup_swapper;
402
403         KASSERT(sx->sx_lock != SX_LOCK_DESTROYED,
404             ("sx_downgrade() of destroyed sx @ %s:%d", file, line));
405         _sx_assert(sx, SA_XLOCKED | SA_NOTRECURSED, file, line);
406 #ifndef INVARIANTS
407         if (sx_recursed(sx))
408                 panic("downgrade of a recursed lock");
409 #endif
410
411         WITNESS_DOWNGRADE(&sx->lock_object, 0, file, line);
412
413         /*
414          * Try to switch from an exclusive lock with no shared waiters
415          * to one sharer with no shared waiters.  If there are
416          * exclusive waiters, we don't need to lock the sleep queue so
417          * long as we preserve the flag.  We do one quick try and if
418          * that fails we grab the sleepq lock to keep the flags from
419          * changing and do it the slow way.
420          *
421          * We have to lock the sleep queue if there are shared waiters
422          * so we can wake them up.
423          */
424         x = sx->sx_lock;
425         if (!(x & SX_LOCK_SHARED_WAITERS) &&
426             atomic_cmpset_rel_ptr(&sx->sx_lock, x, SX_SHARERS_LOCK(1) |
427             (x & SX_LOCK_EXCLUSIVE_WAITERS))) {
428                 LOCK_LOG_LOCK("XDOWNGRADE", &sx->lock_object, 0, 0, file, line);
429                 return;
430         }
431
432         /*
433          * Lock the sleep queue so we can read the waiters bits
434          * without any races and wakeup any shared waiters.
435          */
436         sleepq_lock(&sx->lock_object);
437
438         /*
439          * Preserve SX_LOCK_EXCLUSIVE_WAITERS while downgraded to a single
440          * shared lock.  If there are any shared waiters, wake them up.
441          */
442         wakeup_swapper = 0;
443         x = sx->sx_lock;
444         atomic_store_rel_ptr(&sx->sx_lock, SX_SHARERS_LOCK(1) |
445             (x & SX_LOCK_EXCLUSIVE_WAITERS));
446         if (x & SX_LOCK_SHARED_WAITERS)
447                 wakeup_swapper = sleepq_broadcast(&sx->lock_object, SLEEPQ_SX,
448                     0, SQ_SHARED_QUEUE);
449         sleepq_release(&sx->lock_object);
450
451         LOCK_LOG_LOCK("XDOWNGRADE", &sx->lock_object, 0, 0, file, line);
452         LOCKSTAT_RECORD0(LS_SX_DOWNGRADE_DOWNGRADE, sx);
453
454         if (wakeup_swapper)
455                 kick_proc0();
456 }
457
458 /*
459  * This function represents the so-called 'hard case' for sx_xlock
460  * operation.  All 'easy case' failures are redirected to this.  Note
461  * that ideally this would be a static function, but it needs to be
462  * accessible from at least sx.h.
463  */
464 int
465 _sx_xlock_hard(struct sx *sx, uintptr_t tid, int opts, const char *file,
466     int line)
467 {
468         GIANT_DECLARE;
469 #ifdef ADAPTIVE_SX
470         volatile struct thread *owner;
471         u_int i, spintries = 0;
472 #endif
473         uintptr_t x;
474 #ifdef LOCK_PROFILING
475         uint64_t waittime = 0;
476         int contested = 0;
477 #endif
478         int error = 0;
479 #ifdef  KDTRACE_HOOKS
480         uint64_t spin_cnt = 0;
481         uint64_t sleep_cnt = 0;
482         int64_t sleep_time = 0;
483 #endif
484
485         /* If we already hold an exclusive lock, then recurse. */
486         if (sx_xlocked(sx)) {
487                 KASSERT((sx->lock_object.lo_flags & LO_RECURSABLE) != 0,
488             ("_sx_xlock_hard: recursed on non-recursive sx %s @ %s:%d\n",
489                     sx->lock_object.lo_name, file, line));
490                 sx->sx_recurse++;
491                 atomic_set_ptr(&sx->sx_lock, SX_LOCK_RECURSED);
492                 if (LOCK_LOG_TEST(&sx->lock_object, 0))
493                         CTR2(KTR_LOCK, "%s: %p recursing", __func__, sx);
494                 return (0);
495         }
496
497         if (LOCK_LOG_TEST(&sx->lock_object, 0))
498                 CTR5(KTR_LOCK, "%s: %s contested (lock=%p) at %s:%d", __func__,
499                     sx->lock_object.lo_name, (void *)sx->sx_lock, file, line);
500
501         while (!atomic_cmpset_acq_ptr(&sx->sx_lock, SX_LOCK_UNLOCKED, tid)) {
502 #ifdef KDTRACE_HOOKS
503                 spin_cnt++;
504 #endif
505                 lock_profile_obtain_lock_failed(&sx->lock_object, &contested,
506                     &waittime);
507 #ifdef ADAPTIVE_SX
508                 /*
509                  * If the lock is write locked and the owner is
510                  * running on another CPU, spin until the owner stops
511                  * running or the state of the lock changes.
512                  */
513                 x = sx->sx_lock;
514                 if ((sx->lock_object.lo_flags & SX_NOADAPTIVE) != 0) {
515                         if ((x & SX_LOCK_SHARED) == 0) {
516                                 x = SX_OWNER(x);
517                                 owner = (struct thread *)x;
518                                 if (TD_IS_RUNNING(owner)) {
519                                         if (LOCK_LOG_TEST(&sx->lock_object, 0))
520                                                 CTR3(KTR_LOCK,
521                                             "%s: spinning on %p held by %p",
522                                                     __func__, sx, owner);
523                                         GIANT_SAVE();
524                                         while (SX_OWNER(sx->sx_lock) == x &&
525                                             TD_IS_RUNNING(owner)) {
526                                                 cpu_spinwait();
527 #ifdef KDTRACE_HOOKS
528                                                 spin_cnt++;
529 #endif
530                                         }
531                                         continue;
532                                 }
533                         } else if (SX_SHARERS(x) && spintries < asx_retries) {
534                                 GIANT_SAVE();
535                                 spintries++;
536                                 for (i = 0; i < asx_loops; i++) {
537                                         if (LOCK_LOG_TEST(&sx->lock_object, 0))
538                                                 CTR4(KTR_LOCK,
539                                     "%s: shared spinning on %p with %u and %u",
540                                                     __func__, sx, spintries, i);
541                                         x = sx->sx_lock;
542                                         if ((x & SX_LOCK_SHARED) == 0 ||
543                                             SX_SHARERS(x) == 0)
544                                                 break;
545                                         cpu_spinwait();
546 #ifdef KDTRACE_HOOKS
547                                         spin_cnt++;
548 #endif
549                                 }
550                                 if (i != asx_loops)
551                                         continue;
552                         }
553                 }
554 #endif
555
556                 sleepq_lock(&sx->lock_object);
557                 x = sx->sx_lock;
558
559                 /*
560                  * If the lock was released while spinning on the
561                  * sleep queue chain lock, try again.
562                  */
563                 if (x == SX_LOCK_UNLOCKED) {
564                         sleepq_release(&sx->lock_object);
565                         continue;
566                 }
567
568 #ifdef ADAPTIVE_SX
569                 /*
570                  * The current lock owner might have started executing
571                  * on another CPU (or the lock could have changed
572                  * owners) while we were waiting on the sleep queue
573                  * chain lock.  If so, drop the sleep queue lock and try
574                  * again.
575                  */
576                 if (!(x & SX_LOCK_SHARED) &&
577                     (sx->lock_object.lo_flags & SX_NOADAPTIVE) == 0) {
578                         owner = (struct thread *)SX_OWNER(x);
579                         if (TD_IS_RUNNING(owner)) {
580                                 sleepq_release(&sx->lock_object);
581                                 continue;
582                         }
583                 }
584 #endif
585
586                 /*
587                  * If an exclusive lock was released with both shared
588                  * and exclusive waiters and a shared waiter hasn't
589                  * woken up and acquired the lock yet, sx_lock will be
590                  * set to SX_LOCK_UNLOCKED | SX_LOCK_EXCLUSIVE_WAITERS.
591                  * If we see that value, try to acquire it once.  Note
592                  * that we have to preserve SX_LOCK_EXCLUSIVE_WAITERS
593                  * as there are other exclusive waiters still.  If we
594                  * fail, restart the loop.
595                  */
596                 if (x == (SX_LOCK_UNLOCKED | SX_LOCK_EXCLUSIVE_WAITERS)) {
597                         if (atomic_cmpset_acq_ptr(&sx->sx_lock,
598                             SX_LOCK_UNLOCKED | SX_LOCK_EXCLUSIVE_WAITERS,
599                             tid | SX_LOCK_EXCLUSIVE_WAITERS)) {
600                                 sleepq_release(&sx->lock_object);
601                                 CTR2(KTR_LOCK, "%s: %p claimed by new writer",
602                                     __func__, sx);
603                                 break;
604                         }
605                         sleepq_release(&sx->lock_object);
606                         continue;
607                 }
608
609                 /*
610                  * Try to set the SX_LOCK_EXCLUSIVE_WAITERS.  If we fail,
611                  * than loop back and retry.
612                  */
613                 if (!(x & SX_LOCK_EXCLUSIVE_WAITERS)) {
614                         if (!atomic_cmpset_ptr(&sx->sx_lock, x,
615                             x | SX_LOCK_EXCLUSIVE_WAITERS)) {
616                                 sleepq_release(&sx->lock_object);
617                                 continue;
618                         }
619                         if (LOCK_LOG_TEST(&sx->lock_object, 0))
620                                 CTR2(KTR_LOCK, "%s: %p set excl waiters flag",
621                                     __func__, sx);
622                 }
623
624                 /*
625                  * Since we have been unable to acquire the exclusive
626                  * lock and the exclusive waiters flag is set, we have
627                  * to sleep.
628                  */
629                 if (LOCK_LOG_TEST(&sx->lock_object, 0))
630                         CTR2(KTR_LOCK, "%s: %p blocking on sleep queue",
631                             __func__, sx);
632
633 #ifdef KDTRACE_HOOKS
634                 sleep_time -= lockstat_nsecs();
635 #endif
636                 GIANT_SAVE();
637                 sleepq_add(&sx->lock_object, NULL, sx->lock_object.lo_name,
638                     SLEEPQ_SX | ((opts & SX_INTERRUPTIBLE) ?
639                     SLEEPQ_INTERRUPTIBLE : 0), SQ_EXCLUSIVE_QUEUE);
640                 if (!(opts & SX_INTERRUPTIBLE))
641                         sleepq_wait(&sx->lock_object, 0);
642                 else
643                         error = sleepq_wait_sig(&sx->lock_object, 0);
644 #ifdef KDTRACE_HOOKS
645                 sleep_time += lockstat_nsecs();
646                 sleep_cnt++;
647 #endif
648                 if (error) {
649                         if (LOCK_LOG_TEST(&sx->lock_object, 0))
650                                 CTR2(KTR_LOCK,
651                         "%s: interruptible sleep by %p suspended by signal",
652                                     __func__, sx);
653                         break;
654                 }
655                 if (LOCK_LOG_TEST(&sx->lock_object, 0))
656                         CTR2(KTR_LOCK, "%s: %p resuming from sleep queue",
657                             __func__, sx);
658         }
659
660         GIANT_RESTORE();
661         if (!error)
662                 LOCKSTAT_PROFILE_OBTAIN_LOCK_SUCCESS(LS_SX_XLOCK_ACQUIRE, sx,
663                     contested, waittime, file, line);
664 #ifdef KDTRACE_HOOKS
665         if (sleep_time)
666                 LOCKSTAT_RECORD1(LS_SX_XLOCK_BLOCK, sx, sleep_time);
667         if (spin_cnt > sleep_cnt)
668                 LOCKSTAT_RECORD1(LS_SX_XLOCK_SPIN, sx, (spin_cnt - sleep_cnt));
669 #endif
670         return (error);
671 }
672
673 /*
674  * This function represents the so-called 'hard case' for sx_xunlock
675  * operation.  All 'easy case' failures are redirected to this.  Note
676  * that ideally this would be a static function, but it needs to be
677  * accessible from at least sx.h.
678  */
679 void
680 _sx_xunlock_hard(struct sx *sx, uintptr_t tid, const char *file, int line)
681 {
682         uintptr_t x;
683         int queue, wakeup_swapper;
684
685         MPASS(!(sx->sx_lock & SX_LOCK_SHARED));
686
687         /* If the lock is recursed, then unrecurse one level. */
688         if (sx_xlocked(sx) && sx_recursed(sx)) {
689                 if ((--sx->sx_recurse) == 0)
690                         atomic_clear_ptr(&sx->sx_lock, SX_LOCK_RECURSED);
691                 if (LOCK_LOG_TEST(&sx->lock_object, 0))
692                         CTR2(KTR_LOCK, "%s: %p unrecursing", __func__, sx);
693                 return;
694         }
695         MPASS(sx->sx_lock & (SX_LOCK_SHARED_WAITERS |
696             SX_LOCK_EXCLUSIVE_WAITERS));
697         if (LOCK_LOG_TEST(&sx->lock_object, 0))
698                 CTR2(KTR_LOCK, "%s: %p contested", __func__, sx);
699
700         sleepq_lock(&sx->lock_object);
701         x = SX_LOCK_UNLOCKED;
702
703         /*
704          * The wake up algorithm here is quite simple and probably not
705          * ideal.  It gives precedence to shared waiters if they are
706          * present.  For this condition, we have to preserve the
707          * state of the exclusive waiters flag.
708          */
709         if (sx->sx_lock & SX_LOCK_SHARED_WAITERS) {
710                 queue = SQ_SHARED_QUEUE;
711                 x |= (sx->sx_lock & SX_LOCK_EXCLUSIVE_WAITERS);
712         } else
713                 queue = SQ_EXCLUSIVE_QUEUE;
714
715         /* Wake up all the waiters for the specific queue. */
716         if (LOCK_LOG_TEST(&sx->lock_object, 0))
717                 CTR3(KTR_LOCK, "%s: %p waking up all threads on %s queue",
718                     __func__, sx, queue == SQ_SHARED_QUEUE ? "shared" :
719                     "exclusive");
720         atomic_store_rel_ptr(&sx->sx_lock, x);
721         wakeup_swapper = sleepq_broadcast(&sx->lock_object, SLEEPQ_SX, 0,
722             queue);
723         sleepq_release(&sx->lock_object);
724         if (wakeup_swapper)
725                 kick_proc0();
726 }
727
728 /*
729  * This function represents the so-called 'hard case' for sx_slock
730  * operation.  All 'easy case' failures are redirected to this.  Note
731  * that ideally this would be a static function, but it needs to be
732  * accessible from at least sx.h.
733  */
734 int
735 _sx_slock_hard(struct sx *sx, int opts, const char *file, int line)
736 {
737         GIANT_DECLARE;
738 #ifdef ADAPTIVE_SX
739         volatile struct thread *owner;
740 #endif
741 #ifdef LOCK_PROFILING
742         uint64_t waittime = 0;
743         int contested = 0;
744 #endif
745         uintptr_t x;
746         int error = 0;
747 #ifdef KDTRACE_HOOKS
748         uint64_t spin_cnt = 0;
749         uint64_t sleep_cnt = 0;
750         int64_t sleep_time = 0;
751 #endif
752
753         /*
754          * As with rwlocks, we don't make any attempt to try to block
755          * shared locks once there is an exclusive waiter.
756          */
757         for (;;) {
758 #ifdef KDTRACE_HOOKS
759                 spin_cnt++;
760 #endif
761                 x = sx->sx_lock;
762
763                 /*
764                  * If no other thread has an exclusive lock then try to bump up
765                  * the count of sharers.  Since we have to preserve the state
766                  * of SX_LOCK_EXCLUSIVE_WAITERS, if we fail to acquire the
767                  * shared lock loop back and retry.
768                  */
769                 if (x & SX_LOCK_SHARED) {
770                         MPASS(!(x & SX_LOCK_SHARED_WAITERS));
771                         if (atomic_cmpset_acq_ptr(&sx->sx_lock, x,
772                             x + SX_ONE_SHARER)) {
773                                 if (LOCK_LOG_TEST(&sx->lock_object, 0))
774                                         CTR4(KTR_LOCK,
775                                             "%s: %p succeed %p -> %p", __func__,
776                                             sx, (void *)x,
777                                             (void *)(x + SX_ONE_SHARER));
778                                 break;
779                         }
780                         continue;
781                 }
782                 lock_profile_obtain_lock_failed(&sx->lock_object, &contested,
783                     &waittime);
784
785 #ifdef ADAPTIVE_SX
786                 /*
787                  * If the owner is running on another CPU, spin until
788                  * the owner stops running or the state of the lock
789                  * changes.
790                  */
791                 if ((sx->lock_object.lo_flags & SX_NOADAPTIVE) == 0) {
792                         x = SX_OWNER(x);
793                         owner = (struct thread *)x;
794                         if (TD_IS_RUNNING(owner)) {
795                                 if (LOCK_LOG_TEST(&sx->lock_object, 0))
796                                         CTR3(KTR_LOCK,
797                                             "%s: spinning on %p held by %p",
798                                             __func__, sx, owner);
799                                 GIANT_SAVE();
800                                 while (SX_OWNER(sx->sx_lock) == x &&
801                                     TD_IS_RUNNING(owner)) {
802 #ifdef KDTRACE_HOOKS
803                                         spin_cnt++;
804 #endif
805                                         cpu_spinwait();
806                                 }
807                                 continue;
808                         }
809                 }
810 #endif
811
812                 /*
813                  * Some other thread already has an exclusive lock, so
814                  * start the process of blocking.
815                  */
816                 sleepq_lock(&sx->lock_object);
817                 x = sx->sx_lock;
818
819                 /*
820                  * The lock could have been released while we spun.
821                  * In this case loop back and retry.
822                  */
823                 if (x & SX_LOCK_SHARED) {
824                         sleepq_release(&sx->lock_object);
825                         continue;
826                 }
827
828 #ifdef ADAPTIVE_SX
829                 /*
830                  * If the owner is running on another CPU, spin until
831                  * the owner stops running or the state of the lock
832                  * changes.
833                  */
834                 if (!(x & SX_LOCK_SHARED) &&
835                     (sx->lock_object.lo_flags & SX_NOADAPTIVE) == 0) {
836                         owner = (struct thread *)SX_OWNER(x);
837                         if (TD_IS_RUNNING(owner)) {
838                                 sleepq_release(&sx->lock_object);
839                                 continue;
840                         }
841                 }
842 #endif
843
844                 /*
845                  * Try to set the SX_LOCK_SHARED_WAITERS flag.  If we
846                  * fail to set it drop the sleep queue lock and loop
847                  * back.
848                  */
849                 if (!(x & SX_LOCK_SHARED_WAITERS)) {
850                         if (!atomic_cmpset_ptr(&sx->sx_lock, x,
851                             x | SX_LOCK_SHARED_WAITERS)) {
852                                 sleepq_release(&sx->lock_object);
853                                 continue;
854                         }
855                         if (LOCK_LOG_TEST(&sx->lock_object, 0))
856                                 CTR2(KTR_LOCK, "%s: %p set shared waiters flag",
857                                     __func__, sx);
858                 }
859
860                 /*
861                  * Since we have been unable to acquire the shared lock,
862                  * we have to sleep.
863                  */
864                 if (LOCK_LOG_TEST(&sx->lock_object, 0))
865                         CTR2(KTR_LOCK, "%s: %p blocking on sleep queue",
866                             __func__, sx);
867
868 #ifdef KDTRACE_HOOKS
869                 sleep_time -= lockstat_nsecs();
870 #endif
871                 GIANT_SAVE();
872                 sleepq_add(&sx->lock_object, NULL, sx->lock_object.lo_name,
873                     SLEEPQ_SX | ((opts & SX_INTERRUPTIBLE) ?
874                     SLEEPQ_INTERRUPTIBLE : 0), SQ_SHARED_QUEUE);
875                 if (!(opts & SX_INTERRUPTIBLE))
876                         sleepq_wait(&sx->lock_object, 0);
877                 else
878                         error = sleepq_wait_sig(&sx->lock_object, 0);
879 #ifdef KDTRACE_HOOKS
880                 sleep_time += lockstat_nsecs();
881                 sleep_cnt++;
882 #endif
883                 if (error) {
884                         if (LOCK_LOG_TEST(&sx->lock_object, 0))
885                                 CTR2(KTR_LOCK,
886                         "%s: interruptible sleep by %p suspended by signal",
887                                     __func__, sx);
888                         break;
889                 }
890                 if (LOCK_LOG_TEST(&sx->lock_object, 0))
891                         CTR2(KTR_LOCK, "%s: %p resuming from sleep queue",
892                             __func__, sx);
893         }
894         if (error == 0)
895                 LOCKSTAT_PROFILE_OBTAIN_LOCK_SUCCESS(LS_SX_SLOCK_ACQUIRE, sx,
896                     contested, waittime, file, line);
897 #ifdef KDTRACE_HOOKS
898         if (sleep_time)
899                 LOCKSTAT_RECORD1(LS_SX_XLOCK_BLOCK, sx, sleep_time);
900         if (spin_cnt > sleep_cnt)
901                 LOCKSTAT_RECORD1(LS_SX_XLOCK_SPIN, sx, (spin_cnt - sleep_cnt));
902 #endif
903         GIANT_RESTORE();
904         return (error);
905 }
906
907 /*
908  * This function represents the so-called 'hard case' for sx_sunlock
909  * operation.  All 'easy case' failures are redirected to this.  Note
910  * that ideally this would be a static function, but it needs to be
911  * accessible from at least sx.h.
912  */
913 void
914 _sx_sunlock_hard(struct sx *sx, const char *file, int line)
915 {
916         uintptr_t x;
917         int wakeup_swapper;
918
919         for (;;) {
920                 x = sx->sx_lock;
921
922                 /*
923                  * We should never have sharers while at least one thread
924                  * holds a shared lock.
925                  */
926                 KASSERT(!(x & SX_LOCK_SHARED_WAITERS),
927                     ("%s: waiting sharers", __func__));
928
929                 /*
930                  * See if there is more than one shared lock held.  If
931                  * so, just drop one and return.
932                  */
933                 if (SX_SHARERS(x) > 1) {
934                         if (atomic_cmpset_ptr(&sx->sx_lock, x,
935                             x - SX_ONE_SHARER)) {
936                                 if (LOCK_LOG_TEST(&sx->lock_object, 0))
937                                         CTR4(KTR_LOCK,
938                                             "%s: %p succeeded %p -> %p",
939                                             __func__, sx, (void *)x,
940                                             (void *)(x - SX_ONE_SHARER));
941                                 break;
942                         }
943                         continue;
944                 }
945
946                 /*
947                  * If there aren't any waiters for an exclusive lock,
948                  * then try to drop it quickly.
949                  */
950                 if (!(x & SX_LOCK_EXCLUSIVE_WAITERS)) {
951                         MPASS(x == SX_SHARERS_LOCK(1));
952                         if (atomic_cmpset_ptr(&sx->sx_lock, SX_SHARERS_LOCK(1),
953                             SX_LOCK_UNLOCKED)) {
954                                 if (LOCK_LOG_TEST(&sx->lock_object, 0))
955                                         CTR2(KTR_LOCK, "%s: %p last succeeded",
956                                             __func__, sx);
957                                 break;
958                         }
959                         continue;
960                 }
961
962                 /*
963                  * At this point, there should just be one sharer with
964                  * exclusive waiters.
965                  */
966                 MPASS(x == (SX_SHARERS_LOCK(1) | SX_LOCK_EXCLUSIVE_WAITERS));
967
968                 sleepq_lock(&sx->lock_object);
969
970                 /*
971                  * Wake up semantic here is quite simple:
972                  * Just wake up all the exclusive waiters.
973                  * Note that the state of the lock could have changed,
974                  * so if it fails loop back and retry.
975                  */
976                 if (!atomic_cmpset_ptr(&sx->sx_lock,
977                     SX_SHARERS_LOCK(1) | SX_LOCK_EXCLUSIVE_WAITERS,
978                     SX_LOCK_UNLOCKED)) {
979                         sleepq_release(&sx->lock_object);
980                         continue;
981                 }
982                 if (LOCK_LOG_TEST(&sx->lock_object, 0))
983                         CTR2(KTR_LOCK, "%s: %p waking up all thread on"
984                             "exclusive queue", __func__, sx);
985                 wakeup_swapper = sleepq_broadcast(&sx->lock_object, SLEEPQ_SX,
986                     0, SQ_EXCLUSIVE_QUEUE);
987                 sleepq_release(&sx->lock_object);
988                 if (wakeup_swapper)
989                         kick_proc0();
990                 break;
991         }
992 }
993
994 #ifdef INVARIANT_SUPPORT
995 #ifndef INVARIANTS
996 #undef  _sx_assert
997 #endif
998
999 /*
1000  * In the non-WITNESS case, sx_assert() can only detect that at least
1001  * *some* thread owns an slock, but it cannot guarantee that *this*
1002  * thread owns an slock.
1003  */
1004 void
1005 _sx_assert(struct sx *sx, int what, const char *file, int line)
1006 {
1007 #ifndef WITNESS
1008         int slocked = 0;
1009 #endif
1010
1011         if (panicstr != NULL)
1012                 return;
1013         switch (what) {
1014         case SA_SLOCKED:
1015         case SA_SLOCKED | SA_NOTRECURSED:
1016         case SA_SLOCKED | SA_RECURSED:
1017 #ifndef WITNESS
1018                 slocked = 1;
1019                 /* FALLTHROUGH */
1020 #endif
1021         case SA_LOCKED:
1022         case SA_LOCKED | SA_NOTRECURSED:
1023         case SA_LOCKED | SA_RECURSED:
1024 #ifdef WITNESS
1025                 witness_assert(&sx->lock_object, what, file, line);
1026 #else
1027                 /*
1028                  * If some other thread has an exclusive lock or we
1029                  * have one and are asserting a shared lock, fail.
1030                  * Also, if no one has a lock at all, fail.
1031                  */
1032                 if (sx->sx_lock == SX_LOCK_UNLOCKED ||
1033                     (!(sx->sx_lock & SX_LOCK_SHARED) && (slocked ||
1034                     sx_xholder(sx) != curthread)))
1035                         panic("Lock %s not %slocked @ %s:%d\n",
1036                             sx->lock_object.lo_name, slocked ? "share " : "",
1037                             file, line);
1038
1039                 if (!(sx->sx_lock & SX_LOCK_SHARED)) {
1040                         if (sx_recursed(sx)) {
1041                                 if (what & SA_NOTRECURSED)
1042                                         panic("Lock %s recursed @ %s:%d\n",
1043                                             sx->lock_object.lo_name, file,
1044                                             line);
1045                         } else if (what & SA_RECURSED)
1046                                 panic("Lock %s not recursed @ %s:%d\n",
1047                                     sx->lock_object.lo_name, file, line);
1048                 }
1049 #endif
1050                 break;
1051         case SA_XLOCKED:
1052         case SA_XLOCKED | SA_NOTRECURSED:
1053         case SA_XLOCKED | SA_RECURSED:
1054                 if (sx_xholder(sx) != curthread)
1055                         panic("Lock %s not exclusively locked @ %s:%d\n",
1056                             sx->lock_object.lo_name, file, line);
1057                 if (sx_recursed(sx)) {
1058                         if (what & SA_NOTRECURSED)
1059                                 panic("Lock %s recursed @ %s:%d\n",
1060                                     sx->lock_object.lo_name, file, line);
1061                 } else if (what & SA_RECURSED)
1062                         panic("Lock %s not recursed @ %s:%d\n",
1063                             sx->lock_object.lo_name, file, line);
1064                 break;
1065         case SA_UNLOCKED:
1066 #ifdef WITNESS
1067                 witness_assert(&sx->lock_object, what, file, line);
1068 #else
1069                 /*
1070                  * If we hold an exclusve lock fail.  We can't
1071                  * reliably check to see if we hold a shared lock or
1072                  * not.
1073                  */
1074                 if (sx_xholder(sx) == curthread)
1075                         panic("Lock %s exclusively locked @ %s:%d\n",
1076                             sx->lock_object.lo_name, file, line);
1077 #endif
1078                 break;
1079         default:
1080                 panic("Unknown sx lock assertion: %d @ %s:%d", what, file,
1081                     line);
1082         }
1083 }
1084 #endif  /* INVARIANT_SUPPORT */
1085
1086 #ifdef DDB
1087 static void
1088 db_show_sx(struct lock_object *lock)
1089 {
1090         struct thread *td;
1091         struct sx *sx;
1092
1093         sx = (struct sx *)lock;
1094
1095         db_printf(" state: ");
1096         if (sx->sx_lock == SX_LOCK_UNLOCKED)
1097                 db_printf("UNLOCKED\n");
1098         else if (sx->sx_lock == SX_LOCK_DESTROYED) {
1099                 db_printf("DESTROYED\n");
1100                 return;
1101         } else if (sx->sx_lock & SX_LOCK_SHARED)
1102                 db_printf("SLOCK: %ju\n", (uintmax_t)SX_SHARERS(sx->sx_lock));
1103         else {
1104                 td = sx_xholder(sx);
1105                 db_printf("XLOCK: %p (tid %d, pid %d, \"%s\")\n", td,
1106                     td->td_tid, td->td_proc->p_pid, td->td_name);
1107                 if (sx_recursed(sx))
1108                         db_printf(" recursed: %d\n", sx->sx_recurse);
1109         }
1110
1111         db_printf(" waiters: ");
1112         switch(sx->sx_lock &
1113             (SX_LOCK_SHARED_WAITERS | SX_LOCK_EXCLUSIVE_WAITERS)) {
1114         case SX_LOCK_SHARED_WAITERS:
1115                 db_printf("shared\n");
1116                 break;
1117         case SX_LOCK_EXCLUSIVE_WAITERS:
1118                 db_printf("exclusive\n");
1119                 break;
1120         case SX_LOCK_SHARED_WAITERS | SX_LOCK_EXCLUSIVE_WAITERS:
1121                 db_printf("exclusive and shared\n");
1122                 break;
1123         default:
1124                 db_printf("none\n");
1125         }
1126 }
1127
1128 /*
1129  * Check to see if a thread that is blocked on a sleep queue is actually
1130  * blocked on an sx lock.  If so, output some details and return true.
1131  * If the lock has an exclusive owner, return that in *ownerp.
1132  */
1133 int
1134 sx_chain(struct thread *td, struct thread **ownerp)
1135 {
1136         struct sx *sx;
1137
1138         /*
1139          * Check to see if this thread is blocked on an sx lock.
1140          * First, we check the lock class.  If that is ok, then we
1141          * compare the lock name against the wait message.
1142          */
1143         sx = td->td_wchan;
1144         if (LOCK_CLASS(&sx->lock_object) != &lock_class_sx ||
1145             sx->lock_object.lo_name != td->td_wmesg)
1146                 return (0);
1147
1148         /* We think we have an sx lock, so output some details. */
1149         db_printf("blocked on sx \"%s\" ", td->td_wmesg);
1150         *ownerp = sx_xholder(sx);
1151         if (sx->sx_lock & SX_LOCK_SHARED)
1152                 db_printf("SLOCK (count %ju)\n",
1153                     (uintmax_t)SX_SHARERS(sx->sx_lock));
1154         else
1155                 db_printf("XLOCK\n");
1156         return (1);
1157 }
1158 #endif