rtmutex.c 46.3 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0-only
Ingo Molnar's avatar
Ingo Molnar committed
2
3
4
5
6
7
8
9
10
/*
 * RT-Mutexes: simple blocking mutual exclusion locks with PI support
 *
 * started by Ingo Molnar and Thomas Gleixner.
 *
 *  Copyright (C) 2004-2006 Red Hat, Inc., Ingo Molnar <mingo@redhat.com>
 *  Copyright (C) 2005-2006 Timesys Corp., Thomas Gleixner <tglx@timesys.com>
 *  Copyright (C) 2005 Kihon Technologies Inc., Steven Rostedt
 *  Copyright (C) 2006 Esben Nielsen
11
12
13
14
15
 * Adaptive Spinlocks:
 *  Copyright (C) 2008 Novell, Inc., Gregory Haskins, Sven Dietrich,
 *				     and Peter Morreale,
 * Adaptive Spinlocks simplification:
 *  Copyright (C) 2008 Red Hat, Inc., Steven Rostedt <srostedt@redhat.com>
16
 *
17
 *  See Documentation/locking/rt-mutex-design.rst for details.
Ingo Molnar's avatar
Ingo Molnar committed
18
 */
19
20
21
#include <linux/sched.h>
#include <linux/sched/debug.h>
#include <linux/sched/deadline.h>
22
#include <linux/sched/signal.h>
23
#include <linux/sched/rt.h>
24
#include <linux/sched/wake_q.h>
25
#include <linux/ww_mutex.h>
Ingo Molnar's avatar
Ingo Molnar committed
26
27
28

#include "rtmutex_common.h"

29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
#ifndef WW_RT
# define build_ww_mutex()	(false)
# define ww_container_of(rtm)	NULL

static inline int __ww_mutex_add_waiter(struct rt_mutex_waiter *waiter,
					struct rt_mutex *lock,
					struct ww_acquire_ctx *ww_ctx)
{
	return 0;
}

static inline void __ww_mutex_check_waiters(struct rt_mutex *lock,
					    struct ww_acquire_ctx *ww_ctx)
{
}

static inline void ww_mutex_lock_acquired(struct ww_mutex *lock,
					  struct ww_acquire_ctx *ww_ctx)
{
}

static inline int __ww_mutex_check_kill(struct rt_mutex *lock,
					struct rt_mutex_waiter *waiter,
					struct ww_acquire_ctx *ww_ctx)
{
	return 0;
}

#else
# define build_ww_mutex()	(true)
# define ww_container_of(rtm)	container_of(rtm, struct ww_mutex, base)
# include "ww_mutex.h"
#endif

Ingo Molnar's avatar
Ingo Molnar committed
63
64
65
/*
 * lock->owner state tracking:
 *
66
67
 * lock->owner holds the task_struct pointer of the owner. Bit 0
 * is used to keep track of the "lock has waiters" state.
Ingo Molnar's avatar
Ingo Molnar committed
68
 *
69
70
71
72
73
74
 * owner	bit0
 * NULL		0	lock is free (fast acquire possible)
 * NULL		1	lock is free and has waiters and the top waiter
 *				is going to take the lock*
 * taskpointer	0	lock is held (fast release possible)
 * taskpointer	1	lock is held and has waiters**
Ingo Molnar's avatar
Ingo Molnar committed
75
76
 *
 * The fast atomic compare exchange based acquire and release is only
77
78
79
80
81
82
 * possible when bit 0 of lock->owner is 0.
 *
 * (*) It also can be a transitional state when grabbing the lock
 * with ->wait_lock is held. To prevent any fast path cmpxchg to the lock,
 * we need to set the bit0 before looking at the lock, and the owner may be
 * NULL in this small time, hence this can be a transitional state.
Ingo Molnar's avatar
Ingo Molnar committed
83
 *
84
85
86
87
 * (**) There is a small time when bit 0 is set but there are no
 * waiters. This can happen when grabbing the lock in the slow path.
 * To prevent a cmpxchg of the owner releasing the lock, we need to
 * set this bit before looking at the lock.
Ingo Molnar's avatar
Ingo Molnar committed
88
89
 */

90
static __always_inline void
91
rt_mutex_set_owner(struct rt_mutex_base *lock, struct task_struct *owner)
Ingo Molnar's avatar
Ingo Molnar committed
92
{
93
	unsigned long val = (unsigned long)owner;
Ingo Molnar's avatar
Ingo Molnar committed
94
95
96
97

	if (rt_mutex_has_waiters(lock))
		val |= RT_MUTEX_HAS_WAITERS;

98
	WRITE_ONCE(lock->owner, (struct task_struct *)val);
Ingo Molnar's avatar
Ingo Molnar committed
99
100
}

101
static __always_inline void clear_rt_mutex_waiters(struct rt_mutex_base *lock)
Ingo Molnar's avatar
Ingo Molnar committed
102
103
104
105
106
{
	lock->owner = (struct task_struct *)
			((unsigned long)lock->owner & ~RT_MUTEX_HAS_WAITERS);
}

107
static __always_inline void fixup_rt_mutex_waiters(struct rt_mutex_base *lock)
Ingo Molnar's avatar
Ingo Molnar committed
108
{
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
	unsigned long owner, *p = (unsigned long *) &lock->owner;

	if (rt_mutex_has_waiters(lock))
		return;

	/*
	 * The rbtree has no waiters enqueued, now make sure that the
	 * lock->owner still has the waiters bit set, otherwise the
	 * following can happen:
	 *
	 * CPU 0	CPU 1		CPU2
	 * l->owner=T1
	 *		rt_mutex_lock(l)
	 *		lock(l->lock)
	 *		l->owner = T1 | HAS_WAITERS;
	 *		enqueue(T2)
	 *		boost()
	 *		  unlock(l->lock)
	 *		block()
	 *
	 *				rt_mutex_lock(l)
	 *				lock(l->lock)
	 *				l->owner = T1 | HAS_WAITERS;
	 *				enqueue(T3)
	 *				boost()
	 *				  unlock(l->lock)
	 *				block()
	 *		signal(->T2)	signal(->T3)
	 *		lock(l->lock)
	 *		dequeue(T2)
	 *		deboost()
	 *		  unlock(l->lock)
	 *				lock(l->lock)
	 *				dequeue(T3)
	 *				 ==> wait list is empty
	 *				deboost()
	 *				 unlock(l->lock)
	 *		lock(l->lock)
	 *		fixup_rt_mutex_waiters()
	 *		  if (wait_list_empty(l) {
	 *		    l->owner = owner
	 *		    owner = l->owner & ~HAS_WAITERS;
	 *		      ==> l->owner = T1
	 *		  }
	 *				lock(l->lock)
	 * rt_mutex_unlock(l)		fixup_rt_mutex_waiters()
	 *				  if (wait_list_empty(l) {
	 *				    owner = l->owner & ~HAS_WAITERS;
	 * cmpxchg(l->owner, T1, NULL)
	 *  ===> Success (l->owner = NULL)
	 *
	 *				    l->owner = owner
	 *				      ==> l->owner = T1
	 *				  }
	 *
	 * With the check for the waiter bit in place T3 on CPU2 will not
	 * overwrite. All tasks fiddling with the waiters bit are
	 * serialized by l->lock, so nothing else can modify the waiters
	 * bit. If the bit is set then nothing can change l->owner either
	 * so the simple RMW is safe. The cmpxchg() will simply fail if it
	 * happens in the middle of the RMW because the waiters bit is
	 * still set.
	 */
	owner = READ_ONCE(*p);
	if (owner & RT_MUTEX_HAS_WAITERS)
		WRITE_ONCE(*p, owner & ~RT_MUTEX_HAS_WAITERS);
Ingo Molnar's avatar
Ingo Molnar committed
175
176
}

177
/*
178
179
 * We can speed up the acquire/release, if there's no debugging state to be
 * set up.
180
 */
181
#ifndef CONFIG_DEBUG_RT_MUTEXES
182
static __always_inline bool rt_mutex_cmpxchg_acquire(struct rt_mutex_base *lock,
183
184
185
						     struct task_struct *old,
						     struct task_struct *new)
{
186
	return try_cmpxchg_acquire(&lock->owner, &old, new);
187
188
}

189
static __always_inline bool rt_mutex_cmpxchg_release(struct rt_mutex_base *lock,
190
191
192
						     struct task_struct *old,
						     struct task_struct *new)
{
193
	return try_cmpxchg_release(&lock->owner, &old, new);
194
}
195
196
197
198
199
200

/*
 * Callers must hold the ->wait_lock -- which is the whole purpose as we force
 * all future threads that attempt to [Rmw] the lock to the slowpath. As such
 * relaxed semantics suffice.
 */
201
static __always_inline void mark_rt_mutex_waiters(struct rt_mutex_base *lock)
202
203
204
205
206
{
	unsigned long owner, *p = (unsigned long *) &lock->owner;

	do {
		owner = *p;
207
208
	} while (cmpxchg_relaxed(p, owner,
				 owner | RT_MUTEX_HAS_WAITERS) != owner);
209
}
210
211
212
213
214
215
216

/*
 * Safe fastpath aware unlock:
 * 1) Clear the waiters bit
 * 2) Drop lock->wait_lock
 * 3) Try to unlock the lock with cmpxchg
 */
217
static __always_inline bool unlock_rt_mutex_safe(struct rt_mutex_base *lock,
218
						 unsigned long flags)
219
220
221
222
223
	__releases(lock->wait_lock)
{
	struct task_struct *owner = rt_mutex_owner(lock);

	clear_rt_mutex_waiters(lock);
224
	raw_spin_unlock_irqrestore(&lock->wait_lock, flags);
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
	/*
	 * If a new waiter comes in between the unlock and the cmpxchg
	 * we have two situations:
	 *
	 * unlock(wait_lock);
	 *					lock(wait_lock);
	 * cmpxchg(p, owner, 0) == owner
	 *					mark_rt_mutex_waiters(lock);
	 *					acquire(lock);
	 * or:
	 *
	 * unlock(wait_lock);
	 *					lock(wait_lock);
	 *					mark_rt_mutex_waiters(lock);
	 *
	 * cmpxchg(p, owner, 0) != owner
	 *					enqueue_waiter();
	 *					unlock(wait_lock);
	 * lock(wait_lock);
	 * wake waiter();
	 * unlock(wait_lock);
	 *					lock(wait_lock);
	 *					acquire(lock);
	 */
249
	return rt_mutex_cmpxchg_release(lock, owner, NULL);
250
251
}

252
#else
253
static __always_inline bool rt_mutex_cmpxchg_acquire(struct rt_mutex_base *lock,
254
255
256
257
258
259
260
						     struct task_struct *old,
						     struct task_struct *new)
{
	return false;

}

261
static __always_inline bool rt_mutex_cmpxchg_release(struct rt_mutex_base *lock,
262
263
264
265
266
						     struct task_struct *old,
						     struct task_struct *new)
{
	return false;
}
267

268
static __always_inline void mark_rt_mutex_waiters(struct rt_mutex_base *lock)
269
270
271
272
{
	lock->owner = (struct task_struct *)
			((unsigned long)lock->owner | RT_MUTEX_HAS_WAITERS);
}
273
274
275
276

/*
 * Simple slow path only version: lock->owner is protected by lock->wait_lock.
 */
277
static __always_inline bool unlock_rt_mutex_safe(struct rt_mutex_base *lock,
278
						 unsigned long flags)
279
280
281
	__releases(lock->wait_lock)
{
	lock->owner = NULL;
282
	raw_spin_unlock_irqrestore(&lock->wait_lock, flags);
283
284
	return true;
}
285
286
#endif

287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
static __always_inline int __waiter_prio(struct task_struct *task)
{
	int prio = task->prio;

	if (!rt_prio(prio))
		return DEFAULT_PRIO;

	return prio;
}

static __always_inline void
waiter_update_prio(struct rt_mutex_waiter *waiter, struct task_struct *task)
{
	waiter->prio = __waiter_prio(task);
	waiter->deadline = task->dl.deadline;
}

304
305
306
307
/*
 * Only use with rt_mutex_waiter_{less,equal}()
 */
#define task_to_waiter(p)	\
308
	&(struct rt_mutex_waiter){ .prio = __waiter_prio(p), .deadline = (p)->dl.deadline }
309

310
311
static __always_inline int rt_mutex_waiter_less(struct rt_mutex_waiter *left,
						struct rt_mutex_waiter *right)
312
{
313
	if (left->prio < right->prio)
314
315
316
		return 1;

	/*
317
318
319
320
	 * If both waiters have dl_prio(), we check the deadlines of the
	 * associated tasks.
	 * If left waiter has a dl_prio(), and we didn't return 1 above,
	 * then right waiter has a dl_prio() too.
321
	 */
322
	if (dl_prio(left->prio))
323
		return dl_time_before(left->deadline, right->deadline);
324
325
326
327

	return 0;
}

328
329
static __always_inline int rt_mutex_waiter_equal(struct rt_mutex_waiter *left,
						 struct rt_mutex_waiter *right)
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
{
	if (left->prio != right->prio)
		return 0;

	/*
	 * If both waiters have dl_prio(), we check the deadlines of the
	 * associated tasks.
	 * If left waiter has a dl_prio(), and we didn't return 0 above,
	 * then right waiter has a dl_prio() too.
	 */
	if (dl_prio(left->prio))
		return left->deadline == right->deadline;

	return 1;
}

346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
static inline bool rt_mutex_steal(struct rt_mutex_waiter *waiter,
				  struct rt_mutex_waiter *top_waiter)
{
	if (rt_mutex_waiter_less(waiter, top_waiter))
		return true;

#ifdef RT_MUTEX_BUILD_SPINLOCKS
	/*
	 * Note that RT tasks are excluded from same priority (lateral)
	 * steals to prevent the introduction of an unbounded latency.
	 */
	if (rt_prio(waiter->prio) || dl_prio(waiter->prio))
		return false;

	return rt_mutex_waiter_equal(waiter, top_waiter);
#else
	return false;
#endif
}

366
367
368
#define __node_2_waiter(node) \
	rb_entry((node), struct rt_mutex_waiter, tree_entry)

369
static __always_inline bool __waiter_less(struct rb_node *a, const struct rb_node *b)
370
{
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
	struct rt_mutex_waiter *aw = __node_2_waiter(a);
	struct rt_mutex_waiter *bw = __node_2_waiter(b);

	if (rt_mutex_waiter_less(aw, bw))
		return 1;

	if (!build_ww_mutex())
		return 0;

	if (rt_mutex_waiter_less(bw, aw))
		return 0;

	/* NOTE: relies on waiter->ww_ctx being set before insertion */
	if (aw->ww_ctx) {
		if (!bw->ww_ctx)
			return 1;

		return (signed long)(aw->ww_ctx->stamp -
				     bw->ww_ctx->stamp) < 0;
	}

	return 0;
393
394
}

395
static __always_inline void
396
rt_mutex_enqueue(struct rt_mutex_base *lock, struct rt_mutex_waiter *waiter)
397
{
398
	rb_add_cached(&waiter->tree_entry, &lock->waiters, __waiter_less);
399
400
}

401
static __always_inline void
402
rt_mutex_dequeue(struct rt_mutex_base *lock, struct rt_mutex_waiter *waiter)
403
404
405
406
{
	if (RB_EMPTY_NODE(&waiter->tree_entry))
		return;

407
	rb_erase_cached(&waiter->tree_entry, &lock->waiters);
408
409
410
	RB_CLEAR_NODE(&waiter->tree_entry);
}

411
412
413
#define __node_2_pi_waiter(node) \
	rb_entry((node), struct rt_mutex_waiter, pi_tree_entry)

414
415
static __always_inline bool
__pi_waiter_less(struct rb_node *a, const struct rb_node *b)
416
417
418
419
{
	return rt_mutex_waiter_less(__node_2_pi_waiter(a), __node_2_pi_waiter(b));
}

420
static __always_inline void
421
422
rt_mutex_enqueue_pi(struct task_struct *task, struct rt_mutex_waiter *waiter)
{
423
	rb_add_cached(&waiter->pi_tree_entry, &task->pi_waiters, __pi_waiter_less);
424
425
}

426
static __always_inline void
427
428
429
430
431
rt_mutex_dequeue_pi(struct task_struct *task, struct rt_mutex_waiter *waiter)
{
	if (RB_EMPTY_NODE(&waiter->pi_tree_entry))
		return;

432
	rb_erase_cached(&waiter->pi_tree_entry, &task->pi_waiters);
433
434
435
	RB_CLEAR_NODE(&waiter->pi_tree_entry);
}

436
static __always_inline void rt_mutex_adjust_prio(struct task_struct *p)
437
{
438
	struct task_struct *pi_task = NULL;
439

440
	lockdep_assert_held(&p->pi_lock);
441

442
443
	if (task_has_pi_waiters(p))
		pi_task = task_top_pi_waiter(p)->task;
444

445
	rt_mutex_setprio(p, pi_task);
Ingo Molnar's avatar
Ingo Molnar committed
446
447
}

448
449
450
451
/* RT mutex specific wake_q wrappers */
static __always_inline void rt_mutex_wake_q_add(struct rt_wake_q_head *wqh,
						struct rt_mutex_waiter *w)
{
452
453
454
455
456
457
458
459
	if (IS_ENABLED(CONFIG_PREEMPT_RT) && w->wake_state != TASK_NORMAL) {
		if (IS_ENABLED(CONFIG_PROVE_LOCKING))
			WARN_ON_ONCE(wqh->rtlock_task);
		get_task_struct(w->task);
		wqh->rtlock_task = w->task;
	} else {
		wake_q_add(&wqh->head, w->task);
	}
460
461
462
463
}

static __always_inline void rt_mutex_wake_up_q(struct rt_wake_q_head *wqh)
{
464
465
466
467
468
469
470
471
	if (IS_ENABLED(CONFIG_PREEMPT_RT) && wqh->rtlock_task) {
		wake_up_state(wqh->rtlock_task, TASK_RTLOCK_WAIT);
		put_task_struct(wqh->rtlock_task);
		wqh->rtlock_task = NULL;
	}

	if (!wake_q_empty(&wqh->head))
		wake_up_q(&wqh->head);
472
473
474
475
476

	/* Pairs with preempt_disable() in mark_wakeup_next_waiter() */
	preempt_enable();
}

477
478
479
480
481
482
483
484
485
486
487
488
489
/*
 * Deadlock detection is conditional:
 *
 * If CONFIG_DEBUG_RT_MUTEXES=n, deadlock detection is only conducted
 * if the detect argument is == RT_MUTEX_FULL_CHAINWALK.
 *
 * If CONFIG_DEBUG_RT_MUTEXES=y, deadlock detection is always
 * conducted independent of the detect argument.
 *
 * If the waiter argument is NULL this indicates the deboost path and
 * deadlock detection is disabled independent of the detect argument
 * and the config settings.
 */
490
491
492
static __always_inline bool
rt_mutex_cond_detect_deadlock(struct rt_mutex_waiter *waiter,
			      enum rtmutex_chainwalk chwalk)
493
{
494
	if (IS_ENABLED(CONFIG_DEBUG_RT_MUTEXES))
495
496
		return waiter != NULL;
	return chwalk == RT_MUTEX_FULL_CHAINWALK;
497
498
}

499
static __always_inline struct rt_mutex_base *task_blocked_on_lock(struct task_struct *p)
500
501
502
503
{
	return p->pi_blocked_on ? p->pi_blocked_on->lock : NULL;
}

Ingo Molnar's avatar
Ingo Molnar committed
504
505
506
/*
 * Adjust the priority chain. Also used for deadlock detection.
 * Decreases task's usage by one - may thus free the task.
507
 *
508
509
 * @task:	the task owning the mutex (owner) for which a chain walk is
 *		probably needed
510
 * @chwalk:	do we have to carry out deadlock detection?
511
512
513
514
515
516
 * @orig_lock:	the mutex (can be NULL if we are walking the chain to recheck
 *		things for a task that has just got its priority adjusted, and
 *		is waiting on a mutex)
 * @next_lock:	the mutex on which the owner of @orig_lock was blocked before
 *		we dropped its pi_lock. Is never dereferenced, only used for
 *		comparison to detect lock chain changes.
517
 * @orig_waiter: rt_mutex_waiter struct for the task that has just donated
518
519
520
521
 *		its priority to the mutex owner (can be NULL in the case
 *		depicted above or if the top waiter is gone away and we are
 *		actually deboosting the owner)
 * @top_task:	the current top waiter
522
 *
Ingo Molnar's avatar
Ingo Molnar committed
523
 * Returns 0 or -EDEADLK.
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
 *
 * Chain walk basics and protection scope
 *
 * [R] refcount on task
 * [P] task->pi_lock held
 * [L] rtmutex->wait_lock held
 *
 * Step	Description				Protected by
 *	function arguments:
 *	@task					[R]
 *	@orig_lock if != NULL			@top_task is blocked on it
 *	@next_lock				Unprotected. Cannot be
 *						dereferenced. Only used for
 *						comparison.
 *	@orig_waiter if != NULL			@top_task is blocked on it
 *	@top_task				current, or in case of proxy
 *						locking protected by calling
 *						code
 *	again:
 *	  loop_sanity_check();
 *	retry:
 * [1]	  lock(task->pi_lock);			[R] acquire [P]
 * [2]	  waiter = task->pi_blocked_on;		[P]
 * [3]	  check_exit_conditions_1();		[P]
 * [4]	  lock = waiter->lock;			[P]
 * [5]	  if (!try_lock(lock->wait_lock)) {	[P] try to acquire [L]
 *	    unlock(task->pi_lock);		release [P]
 *	    goto retry;
 *	  }
 * [6]	  check_exit_conditions_2();		[P] + [L]
 * [7]	  requeue_lock_waiter(lock, waiter);	[P] + [L]
 * [8]	  unlock(task->pi_lock);		release [P]
 *	  put_task_struct(task);		release [R]
 * [9]	  check_exit_conditions_3();		[L]
 * [10]	  task = owner(lock);			[L]
 *	  get_task_struct(task);		[L] acquire [R]
 *	  lock(task->pi_lock);			[L] acquire [P]
 * [11]	  requeue_pi_waiter(tsk, waiters(lock));[P] + [L]
 * [12]	  check_exit_conditions_4();		[P] + [L]
 * [13]	  unlock(task->pi_lock);		release [P]
 *	  unlock(lock->wait_lock);		release [L]
 *	  goto again;
Ingo Molnar's avatar
Ingo Molnar committed
566
 */
567
568
static int __sched rt_mutex_adjust_prio_chain(struct task_struct *task,
					      enum rtmutex_chainwalk chwalk,
569
570
					      struct rt_mutex_base *orig_lock,
					      struct rt_mutex_base *next_lock,
571
572
					      struct rt_mutex_waiter *orig_waiter,
					      struct task_struct *top_task)
Ingo Molnar's avatar
Ingo Molnar committed
573
574
{
	struct rt_mutex_waiter *waiter, *top_waiter = orig_waiter;
575
	struct rt_mutex_waiter *prerequeue_top_waiter;
576
	int ret = 0, depth = 0;
577
	struct rt_mutex_base *lock;
578
	bool detect_deadlock;
579
	bool requeue = true;
Ingo Molnar's avatar
Ingo Molnar committed
580

581
	detect_deadlock = rt_mutex_cond_detect_deadlock(orig_waiter, chwalk);
Ingo Molnar's avatar
Ingo Molnar committed
582
583
584
585
586
587
588
589

	/*
	 * The (de)boosting is a step by step approach with a lot of
	 * pitfalls. We want this to be preemptible and we want hold a
	 * maximum of two locks per step. So we have to check
	 * carefully whether things change under us.
	 */
 again:
590
591
592
	/*
	 * We limit the lock chain length for each invocation.
	 */
Ingo Molnar's avatar
Ingo Molnar committed
593
594
595
596
597
598
599
600
601
602
603
	if (++depth > max_lock_depth) {
		static int prev_max;

		/*
		 * Print this only once. If the admin changes the limit,
		 * print a new message when reaching the limit again.
		 */
		if (prev_max != max_lock_depth) {
			prev_max = max_lock_depth;
			printk(KERN_WARNING "Maximum lock depth %d reached "
			       "task: %s (%d)\n", max_lock_depth,
604
			       top_task->comm, task_pid_nr(top_task));
Ingo Molnar's avatar
Ingo Molnar committed
605
606
607
		}
		put_task_struct(task);

608
		return -EDEADLK;
Ingo Molnar's avatar
Ingo Molnar committed
609
	}
610
611
612
613
614
615
616

	/*
	 * We are fully preemptible here and only hold the refcount on
	 * @task. So everything can have changed under us since the
	 * caller or our own code below (goto retry/again) dropped all
	 * locks.
	 */
Ingo Molnar's avatar
Ingo Molnar committed
617
618
 retry:
	/*
619
	 * [1] Task cannot go away as we did a get_task() before !
Ingo Molnar's avatar
Ingo Molnar committed
620
	 */
621
	raw_spin_lock_irq(&task->pi_lock);
Ingo Molnar's avatar
Ingo Molnar committed
622

623
624
625
	/*
	 * [2] Get the waiter on which @task is blocked on.
	 */
Ingo Molnar's avatar
Ingo Molnar committed
626
	waiter = task->pi_blocked_on;
627
628
629
630
631

	/*
	 * [3] check_exit_conditions_1() protected by task->pi_lock.
	 */

Ingo Molnar's avatar
Ingo Molnar committed
632
633
634
635
636
	/*
	 * Check whether the end of the boosting chain has been
	 * reached or the state of the chain has changed while we
	 * dropped the locks.
	 */
637
	if (!waiter)
Ingo Molnar's avatar
Ingo Molnar committed
638
639
		goto out_unlock_pi;

640
641
	/*
	 * Check the orig_waiter state. After we dropped the locks,
642
	 * the previous owner of the lock might have released the lock.
643
	 */
644
	if (orig_waiter && !rt_mutex_owner(orig_lock))
645
646
		goto out_unlock_pi;

647
648
649
650
651
652
653
654
655
656
657
658
	/*
	 * We dropped all locks after taking a refcount on @task, so
	 * the task might have moved on in the lock chain or even left
	 * the chain completely and blocks now on an unrelated lock or
	 * on @orig_lock.
	 *
	 * We stored the lock on which @task was blocked in @next_lock,
	 * so we can detect the chain change.
	 */
	if (next_lock != waiter->lock)
		goto out_unlock_pi;

659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
	/*
	 * There could be 'spurious' loops in the lock graph due to ww_mutex,
	 * consider:
	 *
	 *   P1: A, ww_A, ww_B
	 *   P2: ww_B, ww_A
	 *   P3: A
	 *
	 * P3 should not return -EDEADLK because it gets trapped in the cycle
	 * created by P1 and P2 (which will resolve -- and runs into
	 * max_lock_depth above). Therefore disable detect_deadlock such that
	 * the below termination condition can trigger once all relevant tasks
	 * are boosted.
	 *
	 * Even when we start with ww_mutex we can disable deadlock detection,
	 * since we would supress a ww_mutex induced deadlock at [6] anyway.
	 * Supressing it here however is not sufficient since we might still
	 * hit [6] due to adjustment driven iteration.
	 *
	 * NOTE: if someone were to create a deadlock between 2 ww_classes we'd
	 * utterly fail to report it; lockdep should.
	 */
	if (IS_ENABLED(CONFIG_PREEMPT_RT) && waiter->ww_ctx && detect_deadlock)
		detect_deadlock = false;

684
685
686
687
688
	/*
	 * Drop out, when the task has no waiters. Note,
	 * top_waiter can be NULL, when we are in the deboosting
	 * mode!
	 */
689
690
691
692
693
	if (top_waiter) {
		if (!task_has_pi_waiters(task))
			goto out_unlock_pi;
		/*
		 * If deadlock detection is off, we stop here if we
694
695
696
		 * are not the top pi waiter of the task. If deadlock
		 * detection is enabled we continue, but stop the
		 * requeueing in the chain walk.
697
		 */
698
699
700
701
702
703
		if (top_waiter != task_top_pi_waiter(task)) {
			if (!detect_deadlock)
				goto out_unlock_pi;
			else
				requeue = false;
		}
704
	}
Ingo Molnar's avatar
Ingo Molnar committed
705
706

	/*
707
708
709
710
711
	 * If the waiter priority is the same as the task priority
	 * then there is no further priority adjustment necessary.  If
	 * deadlock detection is off, we stop the chain walk. If its
	 * enabled we continue, but stop the requeueing in the chain
	 * walk.
Ingo Molnar's avatar
Ingo Molnar committed
712
	 */
713
	if (rt_mutex_waiter_equal(waiter, task_to_waiter(task))) {
714
715
716
717
718
		if (!detect_deadlock)
			goto out_unlock_pi;
		else
			requeue = false;
	}
Ingo Molnar's avatar
Ingo Molnar committed
719

720
721
722
	/*
	 * [4] Get the next lock
	 */
Ingo Molnar's avatar
Ingo Molnar committed
723
	lock = waiter->lock;
724
725
726
727
728
	/*
	 * [5] We need to trylock here as we are holding task->pi_lock,
	 * which is the reverse lock order versus the other rtmutex
	 * operations.
	 */
729
	if (!raw_spin_trylock(&lock->wait_lock)) {
730
		raw_spin_unlock_irq(&task->pi_lock);
Ingo Molnar's avatar
Ingo Molnar committed
731
732
733
734
		cpu_relax();
		goto retry;
	}

735
	/*
736
737
738
	 * [6] check_exit_conditions_2() protected by task->pi_lock and
	 * lock->wait_lock.
	 *
739
740
741
742
743
	 * Deadlock detection. If the lock is the same as the original
	 * lock which caused us to walk the lock chain or if the
	 * current lock is owned by the task which initiated the chain
	 * walk, we detected a deadlock.
	 */
744
	if (lock == orig_lock || rt_mutex_owner(lock) == top_task) {
745
		ret = -EDEADLK;
746
747
748
749
750
751
752
753
754
755
756
757
758
759

		/*
		 * When the deadlock is due to ww_mutex; also see above. Don't
		 * report the deadlock and instead let the ww_mutex wound/die
		 * logic pick which of the contending threads gets -EDEADLK.
		 *
		 * NOTE: assumes the cycle only contains a single ww_class; any
		 * other configuration and we fail to report; also, see
		 * lockdep.
		 */
		if (IS_ENABLED(CONFIG_PREEMPT_RT) && orig_waiter->ww_ctx)
			ret = 0;

		raw_spin_unlock(&lock->wait_lock);
Ingo Molnar's avatar
Ingo Molnar committed
760
761
762
		goto out_unlock_pi;
	}

763
764
765
766
767
768
769
770
771
772
	/*
	 * If we just follow the lock chain for deadlock detection, no
	 * need to do all the requeue operations. To avoid a truckload
	 * of conditionals around the various places below, just do the
	 * minimum chain walk checks.
	 */
	if (!requeue) {
		/*
		 * No requeue[7] here. Just release @task [8]
		 */
773
		raw_spin_unlock(&task->pi_lock);
774
775
776
777
778
779
780
		put_task_struct(task);

		/*
		 * [9] check_exit_conditions_3 protected by lock->wait_lock.
		 * If there is no owner of the lock, end of chain.
		 */
		if (!rt_mutex_owner(lock)) {
781
			raw_spin_unlock_irq(&lock->wait_lock);
782
783
784
785
			return 0;
		}

		/* [10] Grab the next task, i.e. owner of @lock */
786
		task = get_task_struct(rt_mutex_owner(lock));
787
		raw_spin_lock(&task->pi_lock);
788
789
790
791
792
793
794
795
796
797
798
799
800
801

		/*
		 * No requeue [11] here. We just do deadlock detection.
		 *
		 * [12] Store whether owner is blocked
		 * itself. Decision is made after dropping the locks
		 */
		next_lock = task_blocked_on_lock(task);
		/*
		 * Get the top waiter for the next iteration
		 */
		top_waiter = rt_mutex_top_waiter(lock);

		/* [13] Drop locks */
802
803
		raw_spin_unlock(&task->pi_lock);
		raw_spin_unlock_irq(&lock->wait_lock);
804
805
806
807
808
809
810

		/* If owner is not blocked, end of chain. */
		if (!next_lock)
			goto out_put_task;
		goto again;
	}

811
812
813
814
815
816
	/*
	 * Store the current top waiter before doing the requeue
	 * operation on @lock. We need it for the boost/deboost
	 * decision below.
	 */
	prerequeue_top_waiter = rt_mutex_top_waiter(lock);
Ingo Molnar's avatar
Ingo Molnar committed
817

818
	/* [7] Requeue the waiter in the lock waiter tree. */
819
	rt_mutex_dequeue(lock, waiter);
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836

	/*
	 * Update the waiter prio fields now that we're dequeued.
	 *
	 * These values can have changed through either:
	 *
	 *   sys_sched_set_scheduler() / sys_sched_setattr()
	 *
	 * or
	 *
	 *   DL CBS enforcement advancing the effective deadline.
	 *
	 * Even though pi_waiters also uses these fields, and that tree is only
	 * updated in [11], we can do this here, since we hold [L], which
	 * serializes all pi_waiters access and rb_erase() does not care about
	 * the values of the node being removed.
	 */
837
	waiter_update_prio(waiter, task);
838

839
	rt_mutex_enqueue(lock, waiter);
Ingo Molnar's avatar
Ingo Molnar committed
840

841
	/* [8] Release the task */
842
	raw_spin_unlock(&task->pi_lock);
843
844
	put_task_struct(task);

845
	/*
846
847
	 * [9] check_exit_conditions_3 protected by lock->wait_lock.
	 *
848
849
850
851
	 * We must abort the chain walk if there is no lock owner even
	 * in the dead lock detection case, as we have nothing to
	 * follow here. This is the end of the chain we are walking.
	 */
852
853
	if (!rt_mutex_owner(lock)) {
		/*
854
855
856
		 * If the requeue [7] above changed the top waiter,
		 * then we need to wake the new top waiter up to try
		 * to get the lock.
857
		 */
858
		if (prerequeue_top_waiter != rt_mutex_top_waiter(lock))
859
			wake_up_state(waiter->task, waiter->wake_state);
860
		raw_spin_unlock_irq(&lock->wait_lock);
861
		return 0;
862
	}
Ingo Molnar's avatar
Ingo Molnar committed
863

864
	/* [10] Grab the next task, i.e. the owner of @lock */
865
	task = get_task_struct(rt_mutex_owner(lock));
866
	raw_spin_lock(&task->pi_lock);
Ingo Molnar's avatar
Ingo Molnar committed
867

868
	/* [11] requeue the pi waiters if necessary */
Ingo Molnar's avatar
Ingo Molnar committed
869
	if (waiter == rt_mutex_top_waiter(lock)) {
870
871
872
		/*
		 * The waiter became the new top (highest priority)
		 * waiter on the lock. Replace the previous top waiter
873
		 * in the owner tasks pi waiters tree with this waiter
874
875
876
		 * and adjust the priority of the owner.
		 */
		rt_mutex_dequeue_pi(task, prerequeue_top_waiter);
877
		rt_mutex_enqueue_pi(task, waiter);
878
		rt_mutex_adjust_prio(task);
Ingo Molnar's avatar
Ingo Molnar committed
879

880
881
882
	} else if (prerequeue_top_waiter == waiter) {
		/*
		 * The waiter was the top waiter on the lock, but is
Ingo Molnar's avatar
Ingo Molnar committed
883
		 * no longer the top priority waiter. Replace waiter in
884
		 * the owner tasks pi waiters tree with the new top
885
886
887
888
889
890
		 * (highest priority) waiter and adjust the priority
		 * of the owner.
		 * The new top waiter is stored in @waiter so that
		 * @waiter == @top_waiter evaluates to true below and
		 * we continue to deboost the rest of the chain.
		 */
891
		rt_mutex_dequeue_pi(task, waiter);
Ingo Molnar's avatar
Ingo Molnar committed
892
		waiter = rt_mutex_top_waiter(lock);
893
		rt_mutex_enqueue_pi(task, waiter);
894
		rt_mutex_adjust_prio(task);
895
896
897
898
899
	} else {
		/*
		 * Nothing changed. No need to do any priority
		 * adjustment.
		 */
Ingo Molnar's avatar
Ingo Molnar committed
900
901
	}

902
	/*
903
904
905
906
	 * [12] check_exit_conditions_4() protected by task->pi_lock
	 * and lock->wait_lock. The actual decisions are made after we
	 * dropped the locks.
	 *
907
908
909
910
911
912
	 * Check whether the task which owns the current lock is pi
	 * blocked itself. If yes we store a pointer to the lock for
	 * the lock chain change detection above. After we dropped
	 * task->pi_lock next_lock cannot be dereferenced anymore.
	 */
	next_lock = task_blocked_on_lock(task);
913
914
915
916
	/*
	 * Store the top waiter of @lock for the end of chain walk
	 * decision below.
	 */
Ingo Molnar's avatar
Ingo Molnar committed
917
	top_waiter = rt_mutex_top_waiter(lock);
918
919

	/* [13] Drop the locks */
920
921
	raw_spin_unlock(&task->pi_lock);
	raw_spin_unlock_irq(&lock->wait_lock);
Ingo Molnar's avatar
Ingo Molnar committed
922

923
	/*
924
925
926
	 * Make the actual exit decisions [12], based on the stored
	 * values.
	 *
927
928
929
930
931
932
	 * We reached the end of the lock chain. Stop right here. No
	 * point to go back just to figure that out.
	 */
	if (!next_lock)
		goto out_put_task;

933
934
935
936
937
	/*
	 * If the current waiter is not the top waiter on the lock,
	 * then we can stop the chain walk here if we are not in full
	 * deadlock detection mode.
	 */
Ingo Molnar's avatar
Ingo Molnar committed
938
939
940
941
942
943
	if (!detect_deadlock && waiter != top_waiter)
		goto out_put_task;

	goto again;

 out_unlock_pi:
944
	raw_spin_unlock_irq(&task->pi_lock);
Ingo Molnar's avatar
Ingo Molnar committed
945
946
 out_put_task:
	put_task_struct(task);
947

Ingo Molnar's avatar
Ingo Molnar committed
948
949
950
951
952
953
	return ret;
}

/*
 * Try to take an rt-mutex
 *
954
 * Must be called with lock->wait_lock held and interrupts disabled
955
 *
956
957
 * @lock:   The lock to be acquired.
 * @task:   The task which wants to acquire the lock
958
 * @waiter: The waiter that is queued to the lock's wait tree if the
959
 *	    callsite called task_blocked_on_lock(), otherwise NULL
Ingo Molnar's avatar
Ingo Molnar committed
960
 */
961
static int __sched
962
try_to_take_rt_mutex(struct rt_mutex_base *lock, struct task_struct *task,
963
		     struct rt_mutex_waiter *waiter)
Ingo Molnar's avatar
Ingo Molnar committed
964
{
965
966
	lockdep_assert_held(&lock->wait_lock);

Ingo Molnar's avatar
Ingo Molnar committed
967
	/*
968
969
970
971
	 * Before testing whether we can acquire @lock, we set the
	 * RT_MUTEX_HAS_WAITERS bit in @lock->owner. This forces all
	 * other tasks which try to modify @lock into the slow path
	 * and they serialize on @lock->wait_lock.
Ingo Molnar's avatar
Ingo Molnar committed
972
	 *
973
974
	 * The RT_MUTEX_HAS_WAITERS bit can have a transitional state
	 * as explained at the top of this file if and only if:
Ingo Molnar's avatar
Ingo Molnar committed
975
	 *
976
977
978
979
980
981
982
	 * - There is a lock owner. The caller must fixup the
	 *   transient state if it does a trylock or leaves the lock
	 *   function due to a signal or timeout.
	 *
	 * - @task acquires the lock and there are no other
	 *   waiters. This is undone in rt_mutex_set_owner(@task) at
	 *   the end of this function.
Ingo Molnar's avatar
Ingo Molnar committed
983
984
985
	 */
	mark_rt_mutex_waiters(lock);

986
987
988
	/*
	 * If @lock has an owner, give up.
	 */
989
	if (rt_mutex_owner(lock))
Ingo Molnar's avatar
Ingo Molnar committed
990
991
		return 0;

992
	/*
993
	 * If @waiter != NULL, @task has already enqueued the waiter
994
	 * into @lock waiter tree. If @waiter == NULL then this is a
995
	 * trylock attempt.
996
	 */
997
	if (waiter) {
998
		struct rt_mutex_waiter *top_waiter = rt_mutex_top_waiter(lock);
999

1000
		/*
1001
1002
		 * If waiter is the highest priority waiter of @lock,
		 * or allowed to steal it, take it over.
1003
		 */
1004
1005
1006
1007
1008
1009
1010
1011
1012
		if (waiter == top_waiter || rt_mutex_steal(waiter, top_waiter)) {
			/*
			 * We can acquire the lock. Remove the waiter from the
			 * lock waiters tree.
			 */
			rt_mutex_dequeue(lock, waiter);
		} else {
			return 0;
		}
1013
	} else {
1014
		/*
1015
1016
1017
1018
1019
1020
		 * If the lock has waiters already we check whether @task is
		 * eligible to take over the lock.
		 *
		 * If there are no other waiters, @task can acquire
		 * the lock.  @task->pi_blocked_on is NULL, so it does
		 * not need to be dequeued.
1021
1022
		 */
		if (rt_mutex_has_waiters(lock)) {
1023
1024
1025
			/* Check whether the trylock can steal it. */
			if (!rt_mutex_steal(task_to_waiter(task),
					    rt_mutex_top_waiter(lock)))
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
				return 0;

			/*
			 * The current top waiter stays enqueued. We
			 * don't have to change anything in the lock
			 * waiters order.
			 */
		} else {
			/*
			 * No waiters. Take the lock without the
			 * pi_lock dance.@task->pi_blocked_on is NULL
			 * and we have no waiters to enqueue in @task
1038
			 * pi waiters tree.
1039
1040
			 */
			goto takeit;
1041
1042
1043
		}
	}

1044
1045
1046
1047
1048
1049
	/*
	 * Clear @task->pi_blocked_on. Requires protection by
	 * @task->pi_lock. Redundant operation for the @waiter == NULL
	 * case, but conditionals are more expensive than a redundant
	 * store.
	 */
1050
	raw_spin_lock(&task->pi_lock);
1051
1052
1053
1054
	task->pi_blocked_on = NULL;
	/*
	 * Finish the lock acquisition. @task is the new owner. If
	 * other waiters exist we have to insert the highest priority
1055
	 * waiter into @task->pi_waiters tree.
1056
1057
1058
	 */
	if (rt_mutex_has_waiters(lock))
		rt_mutex_enqueue_pi(task, rt_mutex_top_waiter(lock));
1059
	raw_spin_unlock(&task->pi_lock);
1060
1061
1062
1063
1064
1065

takeit:
	/*
	 * This either preserves the RT_MUTEX_HAS_WAITERS bit if there
	 * are still waiters or clears it.
	 */
1066
	rt_mutex_set_owner(lock, task);
Ingo Molnar's avatar
Ingo Molnar committed
1067
1068
1069
1070
1071
1072
1073
1074
1075

	return 1;
}

/*
 * Task blocks on lock.
 *
 * Prepare waiter and propagate pi chain
 *
1076
 * This must be called with lock->wait_lock held and interrupts disabled
Ingo Molnar's avatar
Ingo Molnar committed
1077
 */
1078
static int __sched task_blocks_on_rt_mutex(struct rt_mutex_base *lock,
1079
1080
					   struct rt_mutex_waiter *waiter,
					   struct task_struct *task,
1081
					   struct ww_acquire_ctx *ww_ctx,
1082
					   enum rtmutex_chainwalk chwalk)
Ingo Molnar's avatar
Ingo Molnar committed
1083
{
1084
	struct task_struct *owner = rt_mutex_owner(lock);
Ingo Molnar's avatar
Ingo Molnar committed
1085
	struct rt_mutex_waiter *top_waiter = waiter;
1086
	struct rt_mutex_base *next_lock;
1087
	int chain_walk = 0, res;
Ingo Molnar's avatar
Ingo Molnar committed
1088

1089
1090
	lockdep_assert_held(&lock->wait_lock);

1091
1092
1093
1094
1095
1096
1097
1098
1099
	/*
	 * Early deadlock detection. We really don't want the task to
	 * enqueue on itself just to untangle the mess later. It's not
	 * only an optimization. We drop the locks, so another waiter
	 * can come in before the chain walk detects the deadlock. So
	 * the other will detect the deadlock and return -EDEADLOCK,
	 * which is wrong, as the other waiter is not in a deadlock
	 * situation.
	 */
1100
	if (owner == task)
1101
1102
		return -EDEADLK;

1103
	raw_spin_lock(&task->pi_lock);
1104
	waiter->task = task;
Ingo Molnar's avatar
Ingo Molnar committed
1105
	waiter->lock = lock;
1106
	waiter_update_prio(waiter, task);
Ingo Molnar's avatar
Ingo Molnar committed
1107
1108
1109
1110

	/* Get the top priority waiter on the lock */
	if (rt_mutex_has_waiters(lock))
		top_waiter = rt_mutex_top_waiter(lock);
1111
	rt_mutex_enqueue(lock, waiter);
Ingo Molnar's avatar
Ingo Molnar committed
1112

1113
	task->pi_blocked_on = waiter;
Ingo Molnar's avatar
Ingo Molnar committed
1114

1115
	raw_spin_unlock(&task->pi_lock);
Ingo Molnar's avatar
Ingo Molnar committed
1116

1117
1118
1119
1120
1121
1122
	if (build_ww_mutex() && ww_ctx) {
		struct rt_mutex *rtm;

		/* Check whether the waiter should back out immediately */
		rtm = container_of(lock, struct rt_mutex, rtmutex);
		res = __ww_mutex_add_waiter(waiter, rtm, ww_ctx);
1123
1124
1125
1126
1127
		if (res) {
			raw_spin_lock(&task->pi_lock);
			rt_mutex_dequeue(lock, waiter);
			task->pi_blocked_on = NULL;
			raw_spin_unlock(&task->pi_lock);
1128
			return res;
1129
		}
1130
1131
	}

1132
1133
1134
	if (!owner)
		return 0;

1135
	raw_spin_lock(&owner->pi_lock);
Ingo Molnar's avatar
Ingo Molnar committed
1136
	if (waiter == rt_mutex_top_waiter(lock)) {
1137
1138
		rt_mutex_dequeue_pi(owner, top_waiter);
		rt_mutex_enqueue_pi(owner, waiter);
Ingo Molnar's avatar
Ingo Molnar committed
1139

1140
		rt_mutex_adjust_prio(owner);
1141
1142
		if (owner->pi_blocked_on)
			chain_walk = 1;
1143
	} else if (rt_mutex_cond_detect_deadlock(waiter, chwalk)) {
1144
		chain_walk = 1;
1145
	}
1146

1147
1148
1149
	/* Store the lock on which owner is blocked or NULL */
	next_lock = task_blocked_on_lock(owner);

1150
	raw_spin_unlock(&owner->pi_lock);
1151
1152
1153
1154
1155
1156
	/*
	 * Even if full deadlock detection is on, if the owner is not
	 * blocked itself, we can avoid finding this out in the chain
	 * walk.
	 */
	if (!chain_walk || !next_lock)
Ingo Molnar's avatar
Ingo Molnar committed
1157
1158
		return 0;

1159
1160
1161
1162
1163
1164
1165
	/*
	 * The owner can't disappear while holding a lock,
	 * so the owner struct is protected by wait_lock.
	 * Gets dropped in rt_mutex_adjust_prio_chain()!
	 */
	get_task_struct(owner);

1166
	raw_spin_unlock_irq(&lock->wait_lock);
Ingo Molnar's avatar
Ingo Molnar committed
1167

1168
	res = rt_mutex_adjust_prio_chain(owner, chwalk, lock,
1169
					 next_lock, waiter, task);
Ingo Molnar's avatar
Ingo Molnar committed
1170

1171
	raw_spin_lock_irq(&lock->wait_lock);
Ingo Molnar's avatar
Ingo Molnar committed
1172
1173
1174
1175
1176

	return res;
}

/*
1177
 * Remove the top waiter from the current tasks pi waiter tree and
1178
 * queue it up.
Ingo Molnar's avatar
Ingo Molnar committed
1179
 *
1180
 * Called with lock->wait_lock held and interrupts disabled.
Ingo Molnar's avatar
Ingo Molnar committed
1181
 */
1182
static void __sched mark_wakeup_next_waiter(struct rt_wake_q_head *wqh,
1183
					    struct rt_mutex_base *lock)
Ingo Molnar's avatar
Ingo Molnar committed
1184
1185
1186
{
	struct rt_mutex_waiter *waiter;

1187
	raw_spin_lock(&current->pi_lock);
Ingo Molnar's avatar
Ingo Molnar committed
1188
1189
1190
1191

	waiter = rt_mutex_top_waiter(lock);

	/*
1192
1193
1194
1195
1196
	 * Remove it from current->pi_waiters and deboost.
	 *
	 * We must in fact deboost here in order to ensure we call
	 * rt_mutex_setprio() to update p->pi_top_task before the
	 * task unblocks.
Ingo Molnar's avatar
Ingo Molnar committed
1197
	 */
1198
	rt_mutex_dequeue_pi(current, waiter);
1199
	rt_mutex_adjust_prio(current);
Ingo Molnar's avatar
Ingo Molnar committed
1200

1201
1202
1203
1204
1205
1206
1207
1208
1209
	/*
	 * As we are waking up the top waiter, and the waiter stays
	 * queued on the lock until it gets the lock, this lock
	 * obviously has waiters. Just set the bit here and this has
	 * the added benefit of forcing all new tasks into the
	 * slow path making sure no task of lower priority than
	 * the top waiter can steal this lock.
	 */
	lock->owner = (void *) RT_MUTEX_HAS_WAITERS;
Ingo Molnar's avatar
Ingo Molnar committed
1210

1211
1212
1213
1214
1215
1216
1217
1218
	/*
	 * We deboosted before waking the top waiter task such that we don't
	 * run two tasks with the 'same' priority (and ensure the
	 * p->pi_top_task pointer points to a blocked task). This however can
	 * lead to priority inversion if we would get preempted after the
	 * deboost but before waking our donor task, hence the preempt_disable()
	 * before unlock.
	 *
1219
	 * Pairs with preempt_enable() in rt_mutex_wake_up_q();
1220
1221
	 */
	preempt_disable();
1222
	rt_mutex_wake_q_add(wqh, waiter);
1223
	raw_spin_unlock(&current->pi_lock);
Ingo Molnar's avatar
Ingo Molnar committed
1224
1225
}

1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340