Blame view

kernel/rcu/srcu.c 21.6 KB
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
1
2
3
4
5
6
7
8
9
10
11
12
13
14
  /*
   * Sleepable Read-Copy Update mechanism for mutual exclusion.
   *
   * This program is free software; you can redistribute it and/or modify
   * it under the terms of the GNU General Public License as published by
   * the Free Software Foundation; either version 2 of the License, or
   * (at your option) any later version.
   *
   * This program is distributed in the hope that it will be useful,
   * but WITHOUT ANY WARRANTY; without even the implied warranty of
   * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   * GNU General Public License for more details.
   *
   * You should have received a copy of the GNU General Public License
87de1cfdc   Paul E. McKenney   rcu: Stop trackin...
15
16
   * along with this program; if not, you can access it online at
   * http://www.gnu.org/licenses/gpl-2.0.html.
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
17
18
   *
   * Copyright (C) IBM Corporation, 2006
4e87b2d7e   Lai Jiangshan   srcu: Credit Lai ...
19
   * Copyright (C) Fujitsu, 2012
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
20
21
   *
   * Author: Paul McKenney <paulmck@us.ibm.com>
4e87b2d7e   Lai Jiangshan   srcu: Credit Lai ...
22
   *	   Lai Jiangshan <laijs@cn.fujitsu.com>
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
23
24
25
26
27
   *
   * For detailed explanation of Read-Copy Update mechanism see -
   * 		Documentation/RCU/ *.txt
   *
   */
9984de1a5   Paul Gortmaker   kernel: Map most ...
28
  #include <linux/export.h>
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
29
30
31
32
33
  #include <linux/mutex.h>
  #include <linux/percpu.h>
  #include <linux/preempt.h>
  #include <linux/rcupdate.h>
  #include <linux/sched.h>
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
34
  #include <linux/smp.h>
46fdb0937   Paul E. McKenney   rcu: Make synchro...
35
  #include <linux/delay.h>
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
36
  #include <linux/srcu.h>
3705b88db   Antti P Miettinen   rcu: Add a module...
37
  #include "rcu.h"
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
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
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
  /*
   * Initialize an rcu_batch structure to empty.
   */
  static inline void rcu_batch_init(struct rcu_batch *b)
  {
  	b->head = NULL;
  	b->tail = &b->head;
  }
  
  /*
   * Enqueue a callback onto the tail of the specified rcu_batch structure.
   */
  static inline void rcu_batch_queue(struct rcu_batch *b, struct rcu_head *head)
  {
  	*b->tail = head;
  	b->tail = &head->next;
  }
  
  /*
   * Is the specified rcu_batch structure empty?
   */
  static inline bool rcu_batch_empty(struct rcu_batch *b)
  {
  	return b->tail == &b->head;
  }
  
  /*
   * Remove the callback at the head of the specified rcu_batch structure
   * and return a pointer to it, or return NULL if the structure is empty.
   */
  static inline struct rcu_head *rcu_batch_dequeue(struct rcu_batch *b)
  {
  	struct rcu_head *head;
  
  	if (rcu_batch_empty(b))
  		return NULL;
  
  	head = b->head;
  	b->head = head->next;
  	if (b->tail == &head->next)
  		rcu_batch_init(b);
  
  	return head;
  }
  
  /*
   * Move all callbacks from the rcu_batch structure specified by "from" to
   * the structure specified by "to".
   */
  static inline void rcu_batch_move(struct rcu_batch *to, struct rcu_batch *from)
  {
  	if (!rcu_batch_empty(from)) {
  		*to->tail = from->head;
  		to->tail = from->tail;
  		rcu_batch_init(from);
  	}
  }
632ee2001   Paul E. McKenney   rcu: Introduce lo...
95
96
97
  static int init_srcu_struct_fields(struct srcu_struct *sp)
  {
  	sp->completed = 0;
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
98
99
100
101
102
103
104
  	spin_lock_init(&sp->queue_lock);
  	sp->running = false;
  	rcu_batch_init(&sp->batch_queue);
  	rcu_batch_init(&sp->batch_check0);
  	rcu_batch_init(&sp->batch_check1);
  	rcu_batch_init(&sp->batch_done);
  	INIT_DELAYED_WORK(&sp->work, process_srcu);
632ee2001   Paul E. McKenney   rcu: Introduce lo...
105
106
107
108
109
110
111
112
113
  	sp->per_cpu_ref = alloc_percpu(struct srcu_struct_array);
  	return sp->per_cpu_ref ? 0 : -ENOMEM;
  }
  
  #ifdef CONFIG_DEBUG_LOCK_ALLOC
  
  int __init_srcu_struct(struct srcu_struct *sp, const char *name,
  		       struct lock_class_key *key)
  {
632ee2001   Paul E. McKenney   rcu: Introduce lo...
114
115
116
  	/* Don't re-initialize a lock while it is held. */
  	debug_check_no_locks_freed((void *)sp, sizeof(*sp));
  	lockdep_init_map(&sp->dep_map, name, key, 0);
632ee2001   Paul E. McKenney   rcu: Introduce lo...
117
118
119
120
121
  	return init_srcu_struct_fields(sp);
  }
  EXPORT_SYMBOL_GPL(__init_srcu_struct);
  
  #else /* #ifdef CONFIG_DEBUG_LOCK_ALLOC */
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
122
123
124
125
126
127
128
129
  /**
   * init_srcu_struct - initialize a sleep-RCU structure
   * @sp: structure to initialize.
   *
   * Must invoke this on a given srcu_struct before passing that srcu_struct
   * to any other function.  Each srcu_struct represents a separate domain
   * of SRCU protection.
   */
e6a92013b   Alan Stern   [PATCH] SRCU: rep...
130
  int init_srcu_struct(struct srcu_struct *sp)
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
131
  {
632ee2001   Paul E. McKenney   rcu: Introduce lo...
132
  	return init_srcu_struct_fields(sp);
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
133
  }
0cd397d33   Paul E. McKenney   rcu: Add synchron...
134
  EXPORT_SYMBOL_GPL(init_srcu_struct);
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
135

632ee2001   Paul E. McKenney   rcu: Introduce lo...
136
  #endif /* #else #ifdef CONFIG_DEBUG_LOCK_ALLOC */
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
137
  /*
b52ce066c   Lai Jiangshan   rcu: Implement a ...
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
   * Returns approximate total of the readers' ->seq[] values for the
   * rank of per-CPU counters specified by idx.
   */
  static unsigned long srcu_readers_seq_idx(struct srcu_struct *sp, int idx)
  {
  	int cpu;
  	unsigned long sum = 0;
  	unsigned long t;
  
  	for_each_possible_cpu(cpu) {
  		t = ACCESS_ONCE(per_cpu_ptr(sp->per_cpu_ref, cpu)->seq[idx]);
  		sum += t;
  	}
  	return sum;
  }
  
  /*
cef50120b   Paul E. McKenney   rcu: Direct algor...
155
   * Returns approximate number of readers active on the specified rank
b52ce066c   Lai Jiangshan   rcu: Implement a ...
156
   * of the per-CPU ->c[] counters.
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
157
   */
cef50120b   Paul E. McKenney   rcu: Direct algor...
158
159
160
161
162
  static unsigned long srcu_readers_active_idx(struct srcu_struct *sp, int idx)
  {
  	int cpu;
  	unsigned long sum = 0;
  	unsigned long t;
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
163

cef50120b   Paul E. McKenney   rcu: Direct algor...
164
165
166
  	for_each_possible_cpu(cpu) {
  		t = ACCESS_ONCE(per_cpu_ptr(sp->per_cpu_ref, cpu)->c[idx]);
  		sum += t;
cef50120b   Paul E. McKenney   rcu: Direct algor...
167
  	}
b52ce066c   Lai Jiangshan   rcu: Implement a ...
168
  	return sum;
cef50120b   Paul E. McKenney   rcu: Direct algor...
169
170
171
  }
  
  /*
b52ce066c   Lai Jiangshan   rcu: Implement a ...
172
173
174
175
176
177
178
   * Return true if the number of pre-existing readers is determined to
   * be stably zero.  An example unstable zero can occur if the call
   * to srcu_readers_active_idx() misses an __srcu_read_lock() increment,
   * but due to task migration, sees the corresponding __srcu_read_unlock()
   * decrement.  This can happen because srcu_readers_active_idx() takes
   * time to sum the array, and might in fact be interrupted or preempted
   * partway through the summation.
cef50120b   Paul E. McKenney   rcu: Direct algor...
179
180
   */
  static bool srcu_readers_active_idx_check(struct srcu_struct *sp, int idx)
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
181
  {
b52ce066c   Lai Jiangshan   rcu: Implement a ...
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
  	unsigned long seq;
  
  	seq = srcu_readers_seq_idx(sp, idx);
  
  	/*
  	 * The following smp_mb() A pairs with the smp_mb() B located in
  	 * __srcu_read_lock().  This pairing ensures that if an
  	 * __srcu_read_lock() increments its counter after the summation
  	 * in srcu_readers_active_idx(), then the corresponding SRCU read-side
  	 * critical section will see any changes made prior to the start
  	 * of the current SRCU grace period.
  	 *
  	 * Also, if the above call to srcu_readers_seq_idx() saw the
  	 * increment of ->seq[], then the call to srcu_readers_active_idx()
  	 * must see the increment of ->c[].
  	 */
  	smp_mb(); /* A */
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
199

cef50120b   Paul E. McKenney   rcu: Direct algor...
200
201
202
203
204
  	/*
  	 * Note that srcu_readers_active_idx() can incorrectly return
  	 * zero even though there is a pre-existing reader throughout.
  	 * To see this, suppose that task A is in a very long SRCU
  	 * read-side critical section that started on CPU 0, and that
b52ce066c   Lai Jiangshan   rcu: Implement a ...
205
  	 * no other reader exists, so that the sum of the counters
cef50120b   Paul E. McKenney   rcu: Direct algor...
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
  	 * is equal to one.  Then suppose that task B starts executing
  	 * srcu_readers_active_idx(), summing up to CPU 1, and then that
  	 * task C starts reading on CPU 0, so that its increment is not
  	 * summed, but finishes reading on CPU 2, so that its decrement
  	 * -is- summed.  Then when task B completes its sum, it will
  	 * incorrectly get zero, despite the fact that task A has been
  	 * in its SRCU read-side critical section the whole time.
  	 *
  	 * We therefore do a validation step should srcu_readers_active_idx()
  	 * return zero.
  	 */
  	if (srcu_readers_active_idx(sp, idx) != 0)
  		return false;
  
  	/*
b52ce066c   Lai Jiangshan   rcu: Implement a ...
221
222
223
224
225
226
  	 * The remainder of this function is the validation step.
  	 * The following smp_mb() D pairs with the smp_mb() C in
  	 * __srcu_read_unlock().  If the __srcu_read_unlock() was seen
  	 * by srcu_readers_active_idx() above, then any destructive
  	 * operation performed after the grace period will happen after
  	 * the corresponding SRCU read-side critical section.
cef50120b   Paul E. McKenney   rcu: Direct algor...
227
  	 *
b52ce066c   Lai Jiangshan   rcu: Implement a ...
228
229
230
231
232
233
234
235
236
237
238
239
240
241
  	 * Note that there can be at most NR_CPUS worth of readers using
  	 * the old index, which is not enough to overflow even a 32-bit
  	 * integer.  (Yes, this does mean that systems having more than
  	 * a billion or so CPUs need to be 64-bit systems.)  Therefore,
  	 * the sum of the ->seq[] counters cannot possibly overflow.
  	 * Therefore, the only way that the return values of the two
  	 * calls to srcu_readers_seq_idx() can be equal is if there were
  	 * no increments of the corresponding rank of ->seq[] counts
  	 * in the interim.  But the missed-increment scenario laid out
  	 * above includes an increment of the ->seq[] counter by
  	 * the corresponding __srcu_read_lock().  Therefore, if this
  	 * scenario occurs, the return values from the two calls to
  	 * srcu_readers_seq_idx() will differ, and thus the validation
  	 * step below suffices.
cef50120b   Paul E. McKenney   rcu: Direct algor...
242
  	 */
b52ce066c   Lai Jiangshan   rcu: Implement a ...
243
244
245
  	smp_mb(); /* D */
  
  	return srcu_readers_seq_idx(sp, idx) == seq;
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
246
247
248
249
250
251
252
253
254
255
  }
  
  /**
   * srcu_readers_active - returns approximate number of readers.
   * @sp: which srcu_struct to count active readers (holding srcu_read_lock).
   *
   * Note that this is not an atomic primitive, and can therefore suffer
   * severe errors when invoked on an active srcu_struct.  That said, it
   * can be useful as an error check at cleanup time.
   */
bb695170d   Adrian Bunk   make srcu_readers...
256
  static int srcu_readers_active(struct srcu_struct *sp)
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
257
  {
dc8791750   Lai Jiangshan   rcu: Improve srcu...
258
259
260
261
262
263
264
265
  	int cpu;
  	unsigned long sum = 0;
  
  	for_each_possible_cpu(cpu) {
  		sum += ACCESS_ONCE(per_cpu_ptr(sp->per_cpu_ref, cpu)->c[0]);
  		sum += ACCESS_ONCE(per_cpu_ptr(sp->per_cpu_ref, cpu)->c[1]);
  	}
  	return sum;
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
266
267
268
269
270
271
272
273
274
275
276
  }
  
  /**
   * cleanup_srcu_struct - deconstruct a sleep-RCU structure
   * @sp: structure to clean up.
   *
   * Must invoke this after you are finished using a given srcu_struct that
   * was initialized via init_srcu_struct(), else you leak memory.
   */
  void cleanup_srcu_struct(struct srcu_struct *sp)
  {
ab4d2986e   Lai Jiangshan   srcu: Simple clea...
277
278
  	if (WARN_ON(srcu_readers_active(sp)))
  		return; /* Leakage unless caller handles error. */
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
279
280
281
  	free_percpu(sp->per_cpu_ref);
  	sp->per_cpu_ref = NULL;
  }
0cd397d33   Paul E. McKenney   rcu: Add synchron...
282
  EXPORT_SYMBOL_GPL(cleanup_srcu_struct);
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
283

632ee2001   Paul E. McKenney   rcu: Introduce lo...
284
  /*
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
285
286
287
288
   * Counts the new reader in the appropriate per-CPU element of the
   * srcu_struct.  Must be called from process context.
   * Returns an index that must be passed to the matching srcu_read_unlock().
   */
632ee2001   Paul E. McKenney   rcu: Introduce lo...
289
  int __srcu_read_lock(struct srcu_struct *sp)
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
290
291
  {
  	int idx;
7a6b55e71   Lai Jiangshan   srcu: use ACCESS_...
292
  	idx = ACCESS_ONCE(sp->completed) & 0x1;
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
293
  	preempt_disable();
a792563bd   Paul E. McKenney   rcu: Eliminate re...
294
  	__this_cpu_inc(sp->per_cpu_ref->c[idx]);
cef50120b   Paul E. McKenney   rcu: Direct algor...
295
  	smp_mb(); /* B */  /* Avoid leaking the critical section. */
a792563bd   Paul E. McKenney   rcu: Eliminate re...
296
  	__this_cpu_inc(sp->per_cpu_ref->seq[idx]);
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
297
298
299
  	preempt_enable();
  	return idx;
  }
632ee2001   Paul E. McKenney   rcu: Introduce lo...
300
  EXPORT_SYMBOL_GPL(__srcu_read_lock);
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
301

632ee2001   Paul E. McKenney   rcu: Introduce lo...
302
  /*
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
303
304
305
306
307
   * Removes the count for the old reader from the appropriate per-CPU
   * element of the srcu_struct.  Note that this may well be a different
   * CPU than that which was incremented by the corresponding srcu_read_lock().
   * Must be called from process context.
   */
632ee2001   Paul E. McKenney   rcu: Introduce lo...
308
  void __srcu_read_unlock(struct srcu_struct *sp, int idx)
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
309
  {
cef50120b   Paul E. McKenney   rcu: Direct algor...
310
  	smp_mb(); /* C */  /* Avoid leaking the critical section. */
5a41344a3   Lai Jiangshan   srcu: Simplify __...
311
  	this_cpu_dec(sp->per_cpu_ref->c[idx]);
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
312
  }
632ee2001   Paul E. McKenney   rcu: Introduce lo...
313
  EXPORT_SYMBOL_GPL(__srcu_read_unlock);
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
314

0cd397d33   Paul E. McKenney   rcu: Add synchron...
315
  /*
c072a388d   Paul E. McKenney   rcu: demote SRCU_...
316
317
318
319
320
321
322
   * We use an adaptive strategy for synchronize_srcu() and especially for
   * synchronize_srcu_expedited().  We spin for a fixed time period
   * (defined below) to allow SRCU readers to exit their read-side critical
   * sections.  If there are still some readers after 10 microseconds,
   * we repeatedly block for 1-millisecond time periods.  This approach
   * has done well in testing, so there is no need for a config parameter.
   */
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
323
  #define SRCU_RETRY_CHECK_DELAY		5
d9792edd7   Lai Jiangshan   rcu: Use single v...
324
325
  #define SYNCHRONIZE_SRCU_TRYCOUNT	2
  #define SYNCHRONIZE_SRCU_EXP_TRYCOUNT	12
cef50120b   Paul E. McKenney   rcu: Direct algor...
326

18108ebfe   Lai Jiangshan   rcu: Improve SRCU...
327
  /*
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
328
   * @@@ Wait until all pre-existing readers complete.  Such readers
18108ebfe   Lai Jiangshan   rcu: Improve SRCU...
329
   * will have used the index specified by "idx".
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
330
331
   * the caller should ensures the ->completed is not changed while checking
   * and idx = (->completed & 1) ^ 1
18108ebfe   Lai Jiangshan   rcu: Improve SRCU...
332
   */
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
333
  static bool try_check_zero(struct srcu_struct *sp, int idx, int trycount)
cef50120b   Paul E. McKenney   rcu: Direct algor...
334
  {
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
335
336
337
338
339
340
  	for (;;) {
  		if (srcu_readers_active_idx_check(sp, idx))
  			return true;
  		if (--trycount <= 0)
  			return false;
  		udelay(SRCU_RETRY_CHECK_DELAY);
cef50120b   Paul E. McKenney   rcu: Direct algor...
341
  	}
cef50120b   Paul E. McKenney   rcu: Direct algor...
342
  }
c072a388d   Paul E. McKenney   rcu: demote SRCU_...
343

931ea9d1a   Lai Jiangshan   rcu: Implement pe...
344
345
346
347
348
  /*
   * Increment the ->completed counter so that future SRCU readers will
   * use the other rank of the ->c[] and ->seq[] arrays.  This allows
   * us to wait for pre-existing readers in a starvation-free manner.
   */
18108ebfe   Lai Jiangshan   rcu: Improve SRCU...
349
  static void srcu_flip(struct srcu_struct *sp)
944ce9af4   Lai Jiangshan   rcu: Flip ->compl...
350
  {
18108ebfe   Lai Jiangshan   rcu: Improve SRCU...
351
  	sp->completed++;
944ce9af4   Lai Jiangshan   rcu: Flip ->compl...
352
353
354
  }
  
  /*
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
355
356
   * Enqueue an SRCU callback on the specified srcu_struct structure,
   * initiating grace-period processing if it is not already running.
bc72d962d   Paul E. McKenney   rcu: Improve SRCU...
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
   *
   * Note that all CPUs must agree that the grace period extended beyond
   * all pre-existing SRCU read-side critical section.  On systems with
   * more than one CPU, this means that when "func()" is invoked, each CPU
   * is guaranteed to have executed a full memory barrier since the end of
   * its last corresponding SRCU read-side critical section whose beginning
   * preceded the call to call_rcu().  It also means that each CPU executing
   * an SRCU read-side critical section that continues beyond the start of
   * "func()" must have executed a memory barrier after the call_rcu()
   * but before the beginning of that SRCU read-side critical section.
   * Note that these guarantees include CPUs that are offline, idle, or
   * executing in user mode, as well as CPUs that are executing in the kernel.
   *
   * Furthermore, if CPU A invoked call_rcu() and CPU B invoked the
   * resulting SRCU callback function "func()", then both CPU A and CPU
   * B are guaranteed to execute a full memory barrier during the time
   * interval between the call to call_rcu() and the invocation of "func()".
   * This guarantee applies even if CPU A and CPU B are the same CPU (but
   * again only if the system has more than one CPU).
   *
   * Of course, these guarantees apply only for invocations of call_srcu(),
   * srcu_read_lock(), and srcu_read_unlock() that are all passed the same
   * srcu_struct structure.
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
380
381
382
383
384
385
386
387
388
389
390
391
   */
  void call_srcu(struct srcu_struct *sp, struct rcu_head *head,
  		void (*func)(struct rcu_head *head))
  {
  	unsigned long flags;
  
  	head->next = NULL;
  	head->func = func;
  	spin_lock_irqsave(&sp->queue_lock, flags);
  	rcu_batch_queue(&sp->batch_queue, head);
  	if (!sp->running) {
  		sp->running = true;
ae1670339   Shaibal Dutta   rcu: Move SRCU gr...
392
  		queue_delayed_work(system_power_efficient_wq, &sp->work, 0);
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
393
394
395
396
  	}
  	spin_unlock_irqrestore(&sp->queue_lock, flags);
  }
  EXPORT_SYMBOL_GPL(call_srcu);
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
397
398
399
400
  static void srcu_advance_batches(struct srcu_struct *sp, int trycount);
  static void srcu_reschedule(struct srcu_struct *sp);
  
  /*
0cd397d33   Paul E. McKenney   rcu: Add synchron...
401
   * Helper function for synchronize_srcu() and synchronize_srcu_expedited().
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
402
   */
d9792edd7   Lai Jiangshan   rcu: Use single v...
403
  static void __synchronize_srcu(struct srcu_struct *sp, int trycount)
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
404
  {
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
405
406
407
  	struct rcu_synchronize rcu;
  	struct rcu_head *head = &rcu.head;
  	bool done = false;
18108ebfe   Lai Jiangshan   rcu: Improve SRCU...
408

fe15d706c   Paul E. McKenney   rcu: Add lockdep-...
409
410
411
412
413
  	rcu_lockdep_assert(!lock_is_held(&sp->dep_map) &&
  			   !lock_is_held(&rcu_bh_lock_map) &&
  			   !lock_is_held(&rcu_lock_map) &&
  			   !lock_is_held(&rcu_sched_lock_map),
  			   "Illegal synchronize_srcu() in same-type SRCU (or RCU) read-side critical section");
6e6f1b307   Lai Jiangshan   srcu: Add might_s...
414
  	might_sleep();
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
  	init_completion(&rcu.completion);
  
  	head->next = NULL;
  	head->func = wakeme_after_rcu;
  	spin_lock_irq(&sp->queue_lock);
  	if (!sp->running) {
  		/* steal the processing owner */
  		sp->running = true;
  		rcu_batch_queue(&sp->batch_check0, head);
  		spin_unlock_irq(&sp->queue_lock);
  
  		srcu_advance_batches(sp, trycount);
  		if (!rcu_batch_empty(&sp->batch_done)) {
  			BUG_ON(sp->batch_done.head != head);
  			rcu_batch_dequeue(&sp->batch_done);
  			done = true;
  		}
  		/* give the processing owner to work_struct */
  		srcu_reschedule(sp);
  	} else {
  		rcu_batch_queue(&sp->batch_queue, head);
  		spin_unlock_irq(&sp->queue_lock);
  	}
944ce9af4   Lai Jiangshan   rcu: Flip ->compl...
438

931ea9d1a   Lai Jiangshan   rcu: Implement pe...
439
440
  	if (!done)
  		wait_for_completion(&rcu.completion);
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
441
442
443
  }
  
  /**
0cd397d33   Paul E. McKenney   rcu: Add synchron...
444
445
446
   * synchronize_srcu - wait for prior SRCU read-side critical-section completion
   * @sp: srcu_struct with which to synchronize.
   *
34a64b6bb   Lai Jiangshan   srcu: Update sync...
447
448
449
450
451
452
   * Wait for the count to drain to zero of both indexes. To avoid the
   * possible starvation of synchronize_srcu(), it waits for the count of
   * the index=((->completed & 1) ^ 1) to drain to zero at first,
   * and then flip the completed and wait for the count of the other index.
   *
   * Can block; must be called from process context.
0cd397d33   Paul E. McKenney   rcu: Add synchron...
453
454
455
456
   *
   * Note that it is illegal to call synchronize_srcu() from the corresponding
   * SRCU read-side critical section; doing so will result in deadlock.
   * However, it is perfectly legal to call synchronize_srcu() on one
bc72d962d   Paul E. McKenney   rcu: Improve SRCU...
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
   * srcu_struct from some other srcu_struct's read-side critical section,
   * as long as the resulting graph of srcu_structs is acyclic.
   *
   * There are memory-ordering constraints implied by synchronize_srcu().
   * On systems with more than one CPU, when synchronize_srcu() returns,
   * each CPU is guaranteed to have executed a full memory barrier since
   * the end of its last corresponding SRCU-sched read-side critical section
   * whose beginning preceded the call to synchronize_srcu().  In addition,
   * each CPU having an SRCU read-side critical section that extends beyond
   * the return from synchronize_srcu() is guaranteed to have executed a
   * full memory barrier after the beginning of synchronize_srcu() and before
   * the beginning of that SRCU read-side critical section.  Note that these
   * guarantees include CPUs that are offline, idle, or executing in user mode,
   * as well as CPUs that are executing in the kernel.
   *
   * Furthermore, if CPU A invoked synchronize_srcu(), which returned
   * to its caller on CPU B, then both CPU A and CPU B are guaranteed
   * to have executed a full memory barrier during the execution of
   * synchronize_srcu().  This guarantee applies even if CPU A and CPU B
   * are the same CPU, but again only if the system has more than one CPU.
   *
   * Of course, these memory-ordering guarantees apply only when
   * synchronize_srcu(), srcu_read_lock(), and srcu_read_unlock() are
   * passed the same srcu_struct structure.
0cd397d33   Paul E. McKenney   rcu: Add synchron...
481
482
483
   */
  void synchronize_srcu(struct srcu_struct *sp)
  {
5afff48bd   Paul E. McKenney   rcu: Update from ...
484
  	__synchronize_srcu(sp, rcu_gp_is_expedited()
3705b88db   Antti P Miettinen   rcu: Add a module...
485
486
  			   ? SYNCHRONIZE_SRCU_EXP_TRYCOUNT
  			   : SYNCHRONIZE_SRCU_TRYCOUNT);
0cd397d33   Paul E. McKenney   rcu: Add synchron...
487
488
489
490
  }
  EXPORT_SYMBOL_GPL(synchronize_srcu);
  
  /**
236fefafe   Paul E. McKenney   rcu: Call out dan...
491
   * synchronize_srcu_expedited - Brute-force SRCU grace period
0cd397d33   Paul E. McKenney   rcu: Add synchron...
492
493
   * @sp: srcu_struct with which to synchronize.
   *
cef50120b   Paul E. McKenney   rcu: Direct algor...
494
495
   * Wait for an SRCU grace period to elapse, but be more aggressive about
   * spinning rather than blocking when waiting.
0cd397d33   Paul E. McKenney   rcu: Add synchron...
496
   *
bc72d962d   Paul E. McKenney   rcu: Improve SRCU...
497
498
   * Note that synchronize_srcu_expedited() has the same deadlock and
   * memory-ordering properties as does synchronize_srcu().
0cd397d33   Paul E. McKenney   rcu: Add synchron...
499
500
501
   */
  void synchronize_srcu_expedited(struct srcu_struct *sp)
  {
d9792edd7   Lai Jiangshan   rcu: Use single v...
502
  	__synchronize_srcu(sp, SYNCHRONIZE_SRCU_EXP_TRYCOUNT);
0cd397d33   Paul E. McKenney   rcu: Add synchron...
503
504
505
506
  }
  EXPORT_SYMBOL_GPL(synchronize_srcu_expedited);
  
  /**
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
507
   * srcu_barrier - Wait until all in-flight call_srcu() callbacks complete.
4461212aa   Paul E. McKenney   rcu: Fix srcu_bar...
508
   * @sp: srcu_struct on which to wait for in-flight callbacks.
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
509
510
511
512
513
514
515
516
   */
  void srcu_barrier(struct srcu_struct *sp)
  {
  	synchronize_srcu(sp);
  }
  EXPORT_SYMBOL_GPL(srcu_barrier);
  
  /**
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
517
518
519
520
521
522
   * srcu_batches_completed - return batches completed.
   * @sp: srcu_struct on which to report batch completion.
   *
   * Report the number of batches, correlated with, but not necessarily
   * precisely the same as, the number of grace periods that have elapsed.
   */
a5c198f4f   Paul E. McKenney   rcu: Expand SRCU ...
523
  unsigned long srcu_batches_completed(struct srcu_struct *sp)
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
524
525
526
  {
  	return sp->completed;
  }
621934ee7   Paul E. McKenney   [PATCH] srcu-3: R...
527
  EXPORT_SYMBOL_GPL(srcu_batches_completed);
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
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
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
  
  #define SRCU_CALLBACK_BATCH	10
  #define SRCU_INTERVAL		1
  
  /*
   * Move any new SRCU callbacks to the first stage of the SRCU grace
   * period pipeline.
   */
  static void srcu_collect_new(struct srcu_struct *sp)
  {
  	if (!rcu_batch_empty(&sp->batch_queue)) {
  		spin_lock_irq(&sp->queue_lock);
  		rcu_batch_move(&sp->batch_check0, &sp->batch_queue);
  		spin_unlock_irq(&sp->queue_lock);
  	}
  }
  
  /*
   * Core SRCU state machine.  Advance callbacks from ->batch_check0 to
   * ->batch_check1 and then to ->batch_done as readers drain.
   */
  static void srcu_advance_batches(struct srcu_struct *sp, int trycount)
  {
  	int idx = 1 ^ (sp->completed & 1);
  
  	/*
  	 * Because readers might be delayed for an extended period after
  	 * fetching ->completed for their index, at any point in time there
  	 * might well be readers using both idx=0 and idx=1.  We therefore
  	 * need to wait for readers to clear from both index values before
  	 * invoking a callback.
  	 */
  
  	if (rcu_batch_empty(&sp->batch_check0) &&
  	    rcu_batch_empty(&sp->batch_check1))
  		return; /* no callbacks need to be advanced */
  
  	if (!try_check_zero(sp, idx, trycount))
  		return; /* failed to advance, will try after SRCU_INTERVAL */
  
  	/*
  	 * The callbacks in ->batch_check1 have already done with their
  	 * first zero check and flip back when they were enqueued on
  	 * ->batch_check0 in a previous invocation of srcu_advance_batches().
  	 * (Presumably try_check_zero() returned false during that
  	 * invocation, leaving the callbacks stranded on ->batch_check1.)
  	 * They are therefore ready to invoke, so move them to ->batch_done.
  	 */
  	rcu_batch_move(&sp->batch_done, &sp->batch_check1);
  
  	if (rcu_batch_empty(&sp->batch_check0))
  		return; /* no callbacks need to be advanced */
  	srcu_flip(sp);
  
  	/*
  	 * The callbacks in ->batch_check0 just finished their
  	 * first check zero and flip, so move them to ->batch_check1
  	 * for future checking on the other idx.
  	 */
  	rcu_batch_move(&sp->batch_check1, &sp->batch_check0);
  
  	/*
  	 * SRCU read-side critical sections are normally short, so check
  	 * at least twice in quick succession after a flip.
  	 */
  	trycount = trycount < 2 ? 2 : trycount;
  	if (!try_check_zero(sp, idx^1, trycount))
  		return; /* failed to advance, will try after SRCU_INTERVAL */
  
  	/*
  	 * The callbacks in ->batch_check1 have now waited for all
  	 * pre-existing readers using both idx values.  They are therefore
  	 * ready to invoke, so move them to ->batch_done.
  	 */
  	rcu_batch_move(&sp->batch_done, &sp->batch_check1);
  }
  
  /*
   * Invoke a limited number of SRCU callbacks that have passed through
   * their grace period.  If there are more to do, SRCU will reschedule
   * the workqueue.
   */
  static void srcu_invoke_callbacks(struct srcu_struct *sp)
  {
  	int i;
  	struct rcu_head *head;
  
  	for (i = 0; i < SRCU_CALLBACK_BATCH; i++) {
  		head = rcu_batch_dequeue(&sp->batch_done);
  		if (!head)
  			break;
  		local_bh_disable();
  		head->func(head);
  		local_bh_enable();
  	}
  }
  
  /*
   * Finished one round of SRCU grace period.  Start another if there are
   * more SRCU callbacks queued, otherwise put SRCU into not-running state.
   */
  static void srcu_reschedule(struct srcu_struct *sp)
  {
  	bool pending = true;
  
  	if (rcu_batch_empty(&sp->batch_done) &&
  	    rcu_batch_empty(&sp->batch_check1) &&
  	    rcu_batch_empty(&sp->batch_check0) &&
  	    rcu_batch_empty(&sp->batch_queue)) {
  		spin_lock_irq(&sp->queue_lock);
  		if (rcu_batch_empty(&sp->batch_done) &&
  		    rcu_batch_empty(&sp->batch_check1) &&
  		    rcu_batch_empty(&sp->batch_check0) &&
  		    rcu_batch_empty(&sp->batch_queue)) {
  			sp->running = false;
  			pending = false;
  		}
  		spin_unlock_irq(&sp->queue_lock);
  	}
  
  	if (pending)
ae1670339   Shaibal Dutta   rcu: Move SRCU gr...
649
650
  		queue_delayed_work(system_power_efficient_wq,
  				   &sp->work, SRCU_INTERVAL);
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
651
652
653
654
655
  }
  
  /*
   * This is the work-queue function that handles SRCU grace periods.
   */
f2ebfbc99   Lai Jiangshan   srcu: Export proc...
656
  void process_srcu(struct work_struct *work)
931ea9d1a   Lai Jiangshan   rcu: Implement pe...
657
658
659
660
661
662
663
664
665
666
  {
  	struct srcu_struct *sp;
  
  	sp = container_of(work, struct srcu_struct, work.work);
  
  	srcu_collect_new(sp);
  	srcu_advance_batches(sp, 1);
  	srcu_invoke_callbacks(sp);
  	srcu_reschedule(sp);
  }
f2ebfbc99   Lai Jiangshan   srcu: Export proc...
667
  EXPORT_SYMBOL_GPL(process_srcu);