rcutree_plugin.h 60.7 KB
Newer Older
1 2 3
/*
 * Read-Copy Update mechanism for mutual exclusion (tree-based version)
 * Internal non-public definitions that provide either classic
P
Paul E. McKenney 已提交
4
 * or preemptible semantics.
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
 *
 * 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
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
 *
 * Copyright Red Hat, 2009
 * Copyright IBM Corporation, 2009
 *
 * Author: Ingo Molnar <mingo@elte.hu>
 *	   Paul E. McKenney <paulmck@linux.vnet.ibm.com>
 */

27
#include <linux/delay.h>
28
#include <linux/stop_machine.h>
29

30 31 32 33 34 35 36 37
#define RCU_KTHREAD_PRIO 1

#ifdef CONFIG_RCU_BOOST
#define RCU_BOOST_PRIO CONFIG_RCU_BOOST_PRIO
#else
#define RCU_BOOST_PRIO RCU_KTHREAD_PRIO
#endif

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
/*
 * Check the RCU kernel configuration parameters and print informative
 * messages about anything out of the ordinary.  If you like #ifdef, you
 * will love this function.
 */
static void __init rcu_bootup_announce_oddness(void)
{
#ifdef CONFIG_RCU_TRACE
	printk(KERN_INFO "\tRCU debugfs-based tracing is enabled.\n");
#endif
#if (defined(CONFIG_64BIT) && CONFIG_RCU_FANOUT != 64) || (!defined(CONFIG_64BIT) && CONFIG_RCU_FANOUT != 32)
	printk(KERN_INFO "\tCONFIG_RCU_FANOUT set to non-default value of %d\n",
	       CONFIG_RCU_FANOUT);
#endif
#ifdef CONFIG_RCU_FANOUT_EXACT
	printk(KERN_INFO "\tHierarchical RCU autobalancing is disabled.\n");
#endif
#ifdef CONFIG_RCU_FAST_NO_HZ
	printk(KERN_INFO
	       "\tRCU dyntick-idle grace-period acceleration is enabled.\n");
#endif
#ifdef CONFIG_PROVE_RCU
	printk(KERN_INFO "\tRCU lockdep checking is enabled.\n");
#endif
#ifdef CONFIG_RCU_TORTURE_TEST_RUNNABLE
	printk(KERN_INFO "\tRCU torture testing starts during boot.\n");
#endif
65
#if defined(CONFIG_TREE_PREEMPT_RCU) && !defined(CONFIG_RCU_CPU_STALL_VERBOSE)
66 67 68 69 70 71 72
	printk(KERN_INFO "\tVerbose stalled-CPUs detection is disabled.\n");
#endif
#if NUM_RCU_LVL_4 != 0
	printk(KERN_INFO "\tExperimental four-level hierarchy is enabled.\n");
#endif
}

73 74
#ifdef CONFIG_TREE_PREEMPT_RCU

75
struct rcu_state rcu_preempt_state = RCU_STATE_INITIALIZER(rcu_preempt);
76
DEFINE_PER_CPU(struct rcu_data, rcu_preempt_data);
77
static struct rcu_state *rcu_state = &rcu_preempt_state;
78

79
static void rcu_read_unlock_special(struct task_struct *t);
80 81
static int rcu_preempted_readers_exp(struct rcu_node *rnp);

82 83 84
/*
 * Tell them what RCU they are running.
 */
85
static void __init rcu_bootup_announce(void)
86
{
P
Paul E. McKenney 已提交
87
	printk(KERN_INFO "Preemptible hierarchical RCU implementation.\n");
88
	rcu_bootup_announce_oddness();
89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109
}

/*
 * Return the number of RCU-preempt batches processed thus far
 * for debug and statistics.
 */
long rcu_batches_completed_preempt(void)
{
	return rcu_preempt_state.completed;
}
EXPORT_SYMBOL_GPL(rcu_batches_completed_preempt);

/*
 * Return the number of RCU batches processed thus far for debug & stats.
 */
long rcu_batches_completed(void)
{
	return rcu_batches_completed_preempt();
}
EXPORT_SYMBOL_GPL(rcu_batches_completed);

110 111 112 113 114 115 116 117 118
/*
 * Force a quiescent state for preemptible RCU.
 */
void rcu_force_quiescent_state(void)
{
	force_quiescent_state(&rcu_preempt_state, 0);
}
EXPORT_SYMBOL_GPL(rcu_force_quiescent_state);

119
/*
P
Paul E. McKenney 已提交
120
 * Record a preemptible-RCU quiescent state for the specified CPU.  Note
121 122 123
 * that this just means that the task currently running on the CPU is
 * not in a quiescent state.  There might be any number of tasks blocked
 * while in an RCU read-side critical section.
124 125 126 127
 *
 * Unlike the other rcu_*_qs() functions, callers to this function
 * must disable irqs in order to protect the assignment to
 * ->rcu_read_unlock_special.
128
 */
129
static void rcu_preempt_qs(int cpu)
130 131
{
	struct rcu_data *rdp = &per_cpu(rcu_preempt_data, cpu);
132

133
	rdp->passed_quiesce_gpnum = rdp->gpnum;
134
	barrier();
135
	if (rdp->passed_quiesce == 0)
136
		trace_rcu_grace_period("rcu_preempt", rdp->gpnum, "cpuqs");
137
	rdp->passed_quiesce = 1;
138
	current->rcu_read_unlock_special &= ~RCU_READ_UNLOCK_NEED_QS;
139 140 141
}

/*
142 143 144
 * We have entered the scheduler, and the current task might soon be
 * context-switched away from.  If this task is in an RCU read-side
 * critical section, we will no longer be able to rely on the CPU to
145 146 147 148 149 150
 * record that fact, so we enqueue the task on the blkd_tasks list.
 * The task will dequeue itself when it exits the outermost enclosing
 * RCU read-side critical section.  Therefore, the current grace period
 * cannot be permitted to complete until the blkd_tasks list entries
 * predating the current grace period drain, in other words, until
 * rnp->gp_tasks becomes NULL.
151 152
 *
 * Caller must disable preemption.
153
 */
154
static void rcu_preempt_note_context_switch(int cpu)
155 156
{
	struct task_struct *t = current;
157
	unsigned long flags;
158 159 160
	struct rcu_data *rdp;
	struct rcu_node *rnp;

161
	if (t->rcu_read_lock_nesting > 0 &&
162 163 164
	    (t->rcu_read_unlock_special & RCU_READ_UNLOCK_BLOCKED) == 0) {

		/* Possibly blocking in an RCU read-side critical section. */
165
		rdp = per_cpu_ptr(rcu_preempt_state.rda, cpu);
166
		rnp = rdp->mynode;
P
Paul E. McKenney 已提交
167
		raw_spin_lock_irqsave(&rnp->lock, flags);
168
		t->rcu_read_unlock_special |= RCU_READ_UNLOCK_BLOCKED;
169
		t->rcu_blocked_node = rnp;
170 171 172 173 174 175 176 177 178

		/*
		 * If this CPU has already checked in, then this task
		 * will hold up the next grace period rather than the
		 * current grace period.  Queue the task accordingly.
		 * If the task is queued for the current grace period
		 * (i.e., this CPU has not yet passed through a quiescent
		 * state for the current grace period), then as long
		 * as that task remains queued, the current grace period
179 180 181 182 183 184
		 * cannot end.  Note that there is some uncertainty as
		 * to exactly when the current grace period started.
		 * We take a conservative approach, which can result
		 * in unnecessarily waiting on tasks that started very
		 * slightly after the current grace period began.  C'est
		 * la vie!!!
185 186 187
		 *
		 * But first, note that the current CPU must still be
		 * on line!
188
		 */
189
		WARN_ON_ONCE((rdp->grpmask & rnp->qsmaskinit) == 0);
190
		WARN_ON_ONCE(!list_empty(&t->rcu_node_entry));
191 192 193
		if ((rnp->qsmask & rdp->grpmask) && rnp->gp_tasks != NULL) {
			list_add(&t->rcu_node_entry, rnp->gp_tasks->prev);
			rnp->gp_tasks = &t->rcu_node_entry;
194 195 196 197
#ifdef CONFIG_RCU_BOOST
			if (rnp->boost_tasks != NULL)
				rnp->boost_tasks = rnp->gp_tasks;
#endif /* #ifdef CONFIG_RCU_BOOST */
198 199 200 201 202
		} else {
			list_add(&t->rcu_node_entry, &rnp->blkd_tasks);
			if (rnp->qsmask & rdp->grpmask)
				rnp->gp_tasks = &t->rcu_node_entry;
		}
203 204 205 206 207
		trace_rcu_preempt_task(rdp->rsp->name,
				       t->pid,
				       (rnp->qsmask & rdp->grpmask)
				       ? rnp->gpnum
				       : rnp->gpnum + 1);
P
Paul E. McKenney 已提交
208
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
209 210 211 212 213 214 215 216
	} else if (t->rcu_read_lock_nesting < 0 &&
		   t->rcu_read_unlock_special) {

		/*
		 * Complete exit from RCU read-side critical section on
		 * behalf of preempted instance of __rcu_read_unlock().
		 */
		rcu_read_unlock_special(t);
217 218 219 220 221 222 223 224 225 226 227
	}

	/*
	 * Either we were not in an RCU read-side critical section to
	 * begin with, or we have now recorded that critical section
	 * globally.  Either way, we can now note a quiescent state
	 * for this CPU.  Again, if we were in an RCU read-side critical
	 * section, and if that critical section was blocking the current
	 * grace period, then the fact that the task has been enqueued
	 * means that we continue to block the current grace period.
	 */
228
	local_irq_save(flags);
229
	rcu_preempt_qs(cpu);
230
	local_irq_restore(flags);
231 232 233
}

/*
P
Paul E. McKenney 已提交
234
 * Tree-preemptible RCU implementation for rcu_read_lock().
235 236 237 238 239
 * Just increment ->rcu_read_lock_nesting, shared state will be updated
 * if we block.
 */
void __rcu_read_lock(void)
{
240
	current->rcu_read_lock_nesting++;
241 242 243 244
	barrier();  /* needed if we ever invoke rcu_read_lock in rcutree.c */
}
EXPORT_SYMBOL_GPL(__rcu_read_lock);

245 246 247 248 249
/*
 * Check for preempted RCU readers blocking the current grace period
 * for the specified rcu_node structure.  If the caller needs a reliable
 * answer, it must hold the rcu_node's ->lock.
 */
250
static int rcu_preempt_blocked_readers_cgp(struct rcu_node *rnp)
251
{
252
	return rnp->gp_tasks != NULL;
253 254
}

255 256 257 258 259 260 261
/*
 * Record a quiescent state for all tasks that were previously queued
 * on the specified rcu_node structure and that were blocking the current
 * RCU grace period.  The caller must hold the specified rnp->lock with
 * irqs disabled, and this lock is released upon return, but irqs remain
 * disabled.
 */
P
Paul E. McKenney 已提交
262
static void rcu_report_unblock_qs_rnp(struct rcu_node *rnp, unsigned long flags)
263 264 265 266 267
	__releases(rnp->lock)
{
	unsigned long mask;
	struct rcu_node *rnp_p;

268
	if (rnp->qsmask != 0 || rcu_preempt_blocked_readers_cgp(rnp)) {
P
Paul E. McKenney 已提交
269
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
270 271 272 273 274 275 276 277 278 279
		return;  /* Still need more quiescent states! */
	}

	rnp_p = rnp->parent;
	if (rnp_p == NULL) {
		/*
		 * Either there is only one rcu_node in the tree,
		 * or tasks were kicked up to root rcu_node due to
		 * CPUs going offline.
		 */
P
Paul E. McKenney 已提交
280
		rcu_report_qs_rsp(&rcu_preempt_state, flags);
281 282 283 284 285
		return;
	}

	/* Report up the rest of the hierarchy. */
	mask = rnp->grpmask;
P
Paul E. McKenney 已提交
286 287
	raw_spin_unlock(&rnp->lock);	/* irqs remain disabled. */
	raw_spin_lock(&rnp_p->lock);	/* irqs already disabled. */
P
Paul E. McKenney 已提交
288
	rcu_report_qs_rnp(mask, &rcu_preempt_state, rnp_p, flags);
289 290
}

291 292 293 294 295 296 297 298 299 300 301 302 303 304 305
/*
 * Advance a ->blkd_tasks-list pointer to the next entry, instead
 * returning NULL if at the end of the list.
 */
static struct list_head *rcu_next_node_entry(struct task_struct *t,
					     struct rcu_node *rnp)
{
	struct list_head *np;

	np = t->rcu_node_entry.next;
	if (np == &rnp->blkd_tasks)
		np = NULL;
	return np;
}

306 307 308 309 310
/*
 * Handle special cases during rcu_read_unlock(), such as needing to
 * notify RCU core processing or task having blocked during the RCU
 * read-side critical section.
 */
311
static noinline void rcu_read_unlock_special(struct task_struct *t)
312 313
{
	int empty;
314
	int empty_exp;
315
	int empty_exp_now;
316
	unsigned long flags;
317
	struct list_head *np;
318 319 320
#ifdef CONFIG_RCU_BOOST
	struct rt_mutex *rbmp = NULL;
#endif /* #ifdef CONFIG_RCU_BOOST */
321 322 323 324 325 326 327 328 329 330 331 332 333 334 335
	struct rcu_node *rnp;
	int special;

	/* NMI handlers cannot block and cannot safely manipulate state. */
	if (in_nmi())
		return;

	local_irq_save(flags);

	/*
	 * If RCU core is waiting for this CPU to exit critical section,
	 * let it know that we have done so.
	 */
	special = t->rcu_read_unlock_special;
	if (special & RCU_READ_UNLOCK_NEED_QS) {
336
		rcu_preempt_qs(smp_processor_id());
337 338 339
	}

	/* Hardware IRQ handlers cannot block. */
340
	if (in_irq() || in_serving_softirq()) {
341 342 343 344 345 346 347 348
		local_irq_restore(flags);
		return;
	}

	/* Clean up if blocked during RCU read-side critical section. */
	if (special & RCU_READ_UNLOCK_BLOCKED) {
		t->rcu_read_unlock_special &= ~RCU_READ_UNLOCK_BLOCKED;

349 350 351 352 353 354
		/*
		 * Remove this task from the list it blocked on.  The
		 * task can migrate while we acquire the lock, but at
		 * most one time.  So at most two passes through loop.
		 */
		for (;;) {
355
			rnp = t->rcu_blocked_node;
P
Paul E. McKenney 已提交
356
			raw_spin_lock(&rnp->lock);  /* irqs already disabled. */
357
			if (rnp == t->rcu_blocked_node)
358
				break;
P
Paul E. McKenney 已提交
359
			raw_spin_unlock(&rnp->lock); /* irqs remain disabled. */
360
		}
361
		empty = !rcu_preempt_blocked_readers_cgp(rnp);
362 363
		empty_exp = !rcu_preempted_readers_exp(rnp);
		smp_mb(); /* ensure expedited fastpath sees end of RCU c-s. */
364
		np = rcu_next_node_entry(t, rnp);
365
		list_del_init(&t->rcu_node_entry);
366
		t->rcu_blocked_node = NULL;
367 368
		trace_rcu_unlock_preempted_task("rcu_preempt",
						rnp->gpnum, t->pid);
369 370 371 372
		if (&t->rcu_node_entry == rnp->gp_tasks)
			rnp->gp_tasks = np;
		if (&t->rcu_node_entry == rnp->exp_tasks)
			rnp->exp_tasks = np;
373 374 375
#ifdef CONFIG_RCU_BOOST
		if (&t->rcu_node_entry == rnp->boost_tasks)
			rnp->boost_tasks = np;
376 377 378 379
		/* Snapshot/clear ->rcu_boost_mutex with rcu_node lock held. */
		if (t->rcu_boost_mutex) {
			rbmp = t->rcu_boost_mutex;
			t->rcu_boost_mutex = NULL;
380
		}
381
#endif /* #ifdef CONFIG_RCU_BOOST */
382 383 384 385

		/*
		 * If this was the last task on the current list, and if
		 * we aren't waiting on any CPUs, report the quiescent state.
386 387
		 * Note that rcu_report_unblock_qs_rnp() releases rnp->lock,
		 * so we must take a snapshot of the expedited state.
388
		 */
389
		empty_exp_now = !rcu_preempted_readers_exp(rnp);
390 391 392 393 394 395 396 397
		if (!empty && !rcu_preempt_blocked_readers_cgp(rnp)) {
			trace_rcu_quiescent_state_report("preempt_rcu",
							 rnp->gpnum,
							 0, rnp->qsmask,
							 rnp->level,
							 rnp->grplo,
							 rnp->grphi,
							 !!rnp->gp_tasks);
P
Paul E. McKenney 已提交
398
			rcu_report_unblock_qs_rnp(rnp, flags);
399 400
		} else
			raw_spin_unlock_irqrestore(&rnp->lock, flags);
401

402 403
#ifdef CONFIG_RCU_BOOST
		/* Unboost if we were boosted. */
404 405
		if (rbmp)
			rt_mutex_unlock(rbmp);
406 407
#endif /* #ifdef CONFIG_RCU_BOOST */

408 409 410 411
		/*
		 * If this was the last task on the expedited lists,
		 * then we need to report up the rcu_node hierarchy.
		 */
412
		if (!empty_exp && empty_exp_now)
413
			rcu_report_exp_rnp(&rcu_preempt_state, rnp, true);
414 415
	} else {
		local_irq_restore(flags);
416 417 418 419
	}
}

/*
P
Paul E. McKenney 已提交
420
 * Tree-preemptible RCU implementation for rcu_read_unlock().
421 422 423 424 425 426 427 428 429
 * Decrement ->rcu_read_lock_nesting.  If the result is zero (outermost
 * rcu_read_unlock()) and ->rcu_read_unlock_special is non-zero, then
 * invoke rcu_read_unlock_special() to clean up after a context switch
 * in an RCU read-side critical section and other special cases.
 */
void __rcu_read_unlock(void)
{
	struct task_struct *t = current;

430 431 432
	if (t->rcu_read_lock_nesting != 1)
		--t->rcu_read_lock_nesting;
	else {
433
		barrier();  /* critical section before exit code. */
434 435
		t->rcu_read_lock_nesting = INT_MIN;
		barrier();  /* assign before ->rcu_read_unlock_special load */
436 437
		if (unlikely(ACCESS_ONCE(t->rcu_read_unlock_special)))
			rcu_read_unlock_special(t);
438 439
		barrier();  /* ->rcu_read_unlock_special load before assign */
		t->rcu_read_lock_nesting = 0;
440
	}
441
#ifdef CONFIG_PROVE_LOCKING
442 443 444 445 446
	{
		int rrln = ACCESS_ONCE(t->rcu_read_lock_nesting);

		WARN_ON_ONCE(rrln < 0 && rrln > INT_MIN / 2);
	}
447
#endif /* #ifdef CONFIG_PROVE_LOCKING */
448 449 450
}
EXPORT_SYMBOL_GPL(__rcu_read_unlock);

451 452 453 454 455 456 457 458 459 460 461
#ifdef CONFIG_RCU_CPU_STALL_VERBOSE

/*
 * Dump detailed information for all tasks blocking the current RCU
 * grace period on the specified rcu_node structure.
 */
static void rcu_print_detail_task_stall_rnp(struct rcu_node *rnp)
{
	unsigned long flags;
	struct task_struct *t;

462
	if (!rcu_preempt_blocked_readers_cgp(rnp))
463 464 465 466 467 468 469
		return;
	raw_spin_lock_irqsave(&rnp->lock, flags);
	t = list_entry(rnp->gp_tasks,
		       struct task_struct, rcu_node_entry);
	list_for_each_entry_continue(t, &rnp->blkd_tasks, rcu_node_entry)
		sched_show_task(t);
	raw_spin_unlock_irqrestore(&rnp->lock, flags);
470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492
}

/*
 * Dump detailed information for all tasks blocking the current RCU
 * grace period.
 */
static void rcu_print_detail_task_stall(struct rcu_state *rsp)
{
	struct rcu_node *rnp = rcu_get_root(rsp);

	rcu_print_detail_task_stall_rnp(rnp);
	rcu_for_each_leaf_node(rsp, rnp)
		rcu_print_detail_task_stall_rnp(rnp);
}

#else /* #ifdef CONFIG_RCU_CPU_STALL_VERBOSE */

static void rcu_print_detail_task_stall(struct rcu_state *rsp)
{
}

#endif /* #else #ifdef CONFIG_RCU_CPU_STALL_VERBOSE */

493 494 495 496
/*
 * Scan the current list of tasks blocked within RCU read-side critical
 * sections, printing out the tid of each.
 */
497
static int rcu_print_task_stall(struct rcu_node *rnp)
498 499
{
	struct task_struct *t;
500
	int ndetected = 0;
501

502
	if (!rcu_preempt_blocked_readers_cgp(rnp))
503
		return 0;
504 505
	t = list_entry(rnp->gp_tasks,
		       struct task_struct, rcu_node_entry);
506
	list_for_each_entry_continue(t, &rnp->blkd_tasks, rcu_node_entry) {
507
		printk(" P%d", t->pid);
508 509 510
		ndetected++;
	}
	return ndetected;
511 512
}

513 514 515 516 517 518 519 520 521 522
/*
 * Suppress preemptible RCU's CPU stall warnings by pushing the
 * time of the next stall-warning message comfortably far into the
 * future.
 */
static void rcu_preempt_stall_reset(void)
{
	rcu_preempt_state.jiffies_stall = jiffies + ULONG_MAX / 2;
}

523 524 525 526 527 528
/*
 * Check that the list of blocked tasks for the newly completed grace
 * period is in fact empty.  It is a serious bug to complete a grace
 * period that still has RCU readers blocked!  This function must be
 * invoked -before- updating this rnp's ->gpnum, and the rnp's ->lock
 * must be held by the caller.
529 530 531
 *
 * Also, if there are blocked tasks on the list, they automatically
 * block the newly created grace period, so set up ->gp_tasks accordingly.
532 533 534
 */
static void rcu_preempt_check_blocked_tasks(struct rcu_node *rnp)
{
535
	WARN_ON_ONCE(rcu_preempt_blocked_readers_cgp(rnp));
536 537
	if (!list_empty(&rnp->blkd_tasks))
		rnp->gp_tasks = rnp->blkd_tasks.next;
538
	WARN_ON_ONCE(rnp->qsmask);
539 540
}

541 542
#ifdef CONFIG_HOTPLUG_CPU

543 544 545 546 547 548
/*
 * Handle tasklist migration for case in which all CPUs covered by the
 * specified rcu_node have gone offline.  Move them up to the root
 * rcu_node.  The reason for not just moving them to the immediate
 * parent is to remove the need for rcu_read_unlock_special() to
 * make more than two attempts to acquire the target rcu_node's lock.
549 550
 * Returns true if there were tasks blocking the current RCU grace
 * period.
551
 *
552 553 554
 * Returns 1 if there was previously a task blocking the current grace
 * period on the specified rcu_node structure.
 *
555 556
 * The caller must hold rnp->lock with irqs disabled.
 */
557 558 559
static int rcu_preempt_offline_tasks(struct rcu_state *rsp,
				     struct rcu_node *rnp,
				     struct rcu_data *rdp)
560 561 562
{
	struct list_head *lp;
	struct list_head *lp_root;
563
	int retval = 0;
564
	struct rcu_node *rnp_root = rcu_get_root(rsp);
565
	struct task_struct *t;
566

567 568
	if (rnp == rnp_root) {
		WARN_ONCE(1, "Last CPU thought to be offlined?");
569
		return 0;  /* Shouldn't happen: at least one CPU online. */
570
	}
571 572 573

	/* If we are on an internal node, complain bitterly. */
	WARN_ON_ONCE(rnp != rdp->mynode);
574 575

	/*
576 577 578 579 580 581 582
	 * Move tasks up to root rcu_node.  Don't try to get fancy for
	 * this corner-case operation -- just put this node's tasks
	 * at the head of the root node's list, and update the root node's
	 * ->gp_tasks and ->exp_tasks pointers to those of this node's,
	 * if non-NULL.  This might result in waiting for more tasks than
	 * absolutely necessary, but this is a good performance/complexity
	 * tradeoff.
583
	 */
584
	if (rcu_preempt_blocked_readers_cgp(rnp))
585 586 587
		retval |= RCU_OFL_TASKS_NORM_GP;
	if (rcu_preempted_readers_exp(rnp))
		retval |= RCU_OFL_TASKS_EXP_GP;
588 589 590 591 592 593 594 595 596 597 598 599
	lp = &rnp->blkd_tasks;
	lp_root = &rnp_root->blkd_tasks;
	while (!list_empty(lp)) {
		t = list_entry(lp->next, typeof(*t), rcu_node_entry);
		raw_spin_lock(&rnp_root->lock); /* irqs already disabled */
		list_del(&t->rcu_node_entry);
		t->rcu_blocked_node = rnp_root;
		list_add(&t->rcu_node_entry, lp_root);
		if (&t->rcu_node_entry == rnp->gp_tasks)
			rnp_root->gp_tasks = rnp->gp_tasks;
		if (&t->rcu_node_entry == rnp->exp_tasks)
			rnp_root->exp_tasks = rnp->exp_tasks;
600 601 602 603
#ifdef CONFIG_RCU_BOOST
		if (&t->rcu_node_entry == rnp->boost_tasks)
			rnp_root->boost_tasks = rnp->boost_tasks;
#endif /* #ifdef CONFIG_RCU_BOOST */
604
		raw_spin_unlock(&rnp_root->lock); /* irqs still disabled */
605
	}
606 607 608 609 610 611 612 613 614 615

#ifdef CONFIG_RCU_BOOST
	/* In case root is being boosted and leaf is not. */
	raw_spin_lock(&rnp_root->lock); /* irqs already disabled */
	if (rnp_root->boost_tasks != NULL &&
	    rnp_root->boost_tasks != rnp_root->gp_tasks)
		rnp_root->boost_tasks = rnp_root->gp_tasks;
	raw_spin_unlock(&rnp_root->lock); /* irqs still disabled */
#endif /* #ifdef CONFIG_RCU_BOOST */

616 617
	rnp->gp_tasks = NULL;
	rnp->exp_tasks = NULL;
618
	return retval;
619 620
}

621
/*
P
Paul E. McKenney 已提交
622
 * Do CPU-offline processing for preemptible RCU.
623 624 625 626 627 628 629 630
 */
static void rcu_preempt_offline_cpu(int cpu)
{
	__rcu_offline_cpu(cpu, &rcu_preempt_state);
}

#endif /* #ifdef CONFIG_HOTPLUG_CPU */

631 632 633 634 635 636 637 638 639 640 641 642
/*
 * Check for a quiescent state from the current CPU.  When a task blocks,
 * the task is recorded in the corresponding CPU's rcu_node structure,
 * which is checked elsewhere.
 *
 * Caller must disable hard irqs.
 */
static void rcu_preempt_check_callbacks(int cpu)
{
	struct task_struct *t = current;

	if (t->rcu_read_lock_nesting == 0) {
643
		rcu_preempt_qs(cpu);
644 645
		return;
	}
646 647
	if (t->rcu_read_lock_nesting > 0 &&
	    per_cpu(rcu_preempt_data, cpu).qs_pending)
648
		t->rcu_read_unlock_special |= RCU_READ_UNLOCK_NEED_QS;
649 650 651
}

/*
P
Paul E. McKenney 已提交
652
 * Process callbacks for preemptible RCU.
653 654 655 656 657 658 659
 */
static void rcu_preempt_process_callbacks(void)
{
	__rcu_process_callbacks(&rcu_preempt_state,
				&__get_cpu_var(rcu_preempt_data));
}

660 661
#ifdef CONFIG_RCU_BOOST

662 663 664 665 666
static void rcu_preempt_do_callbacks(void)
{
	rcu_do_batch(&rcu_preempt_state, &__get_cpu_var(rcu_preempt_data));
}

667 668
#endif /* #ifdef CONFIG_RCU_BOOST */

669
/*
P
Paul E. McKenney 已提交
670
 * Queue a preemptible-RCU callback for invocation after a grace period.
671 672 673 674 675 676 677
 */
void call_rcu(struct rcu_head *head, void (*func)(struct rcu_head *rcu))
{
	__call_rcu(head, func, &rcu_preempt_state);
}
EXPORT_SYMBOL_GPL(call_rcu);

678 679 680 681 682
/**
 * synchronize_rcu - wait until a grace period has elapsed.
 *
 * Control will return to the caller some time after a full grace
 * period has elapsed, in other words after all currently executing RCU
683 684 685 686 687
 * read-side critical sections have completed.  Note, however, that
 * upon return from synchronize_rcu(), the caller might well be executing
 * concurrently with new RCU read-side critical sections that began while
 * synchronize_rcu() was waiting.  RCU read-side critical sections are
 * delimited by rcu_read_lock() and rcu_read_unlock(), and may be nested.
688 689 690 691 692
 */
void synchronize_rcu(void)
{
	if (!rcu_scheduler_active)
		return;
693
	wait_rcu_gp(call_rcu);
694 695 696
}
EXPORT_SYMBOL_GPL(synchronize_rcu);

697 698 699 700 701 702 703 704 705 706 707 708
static DECLARE_WAIT_QUEUE_HEAD(sync_rcu_preempt_exp_wq);
static long sync_rcu_preempt_exp_count;
static DEFINE_MUTEX(sync_rcu_preempt_exp_mutex);

/*
 * Return non-zero if there are any tasks in RCU read-side critical
 * sections blocking the current preemptible-RCU expedited grace period.
 * If there is no preemptible-RCU expedited grace period currently in
 * progress, returns zero unconditionally.
 */
static int rcu_preempted_readers_exp(struct rcu_node *rnp)
{
709
	return rnp->exp_tasks != NULL;
710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734
}

/*
 * return non-zero if there is no RCU expedited grace period in progress
 * for the specified rcu_node structure, in other words, if all CPUs and
 * tasks covered by the specified rcu_node structure have done their bit
 * for the current expedited grace period.  Works only for preemptible
 * RCU -- other RCU implementation use other means.
 *
 * Caller must hold sync_rcu_preempt_exp_mutex.
 */
static int sync_rcu_preempt_exp_done(struct rcu_node *rnp)
{
	return !rcu_preempted_readers_exp(rnp) &&
	       ACCESS_ONCE(rnp->expmask) == 0;
}

/*
 * Report the exit from RCU read-side critical section for the last task
 * that queued itself during or before the current expedited preemptible-RCU
 * grace period.  This event is reported either to the rcu_node structure on
 * which the task was queued or to one of that rcu_node structure's ancestors,
 * recursively up the tree.  (Calm down, calm down, we do the recursion
 * iteratively!)
 *
735 736 737
 * Most callers will set the "wake" flag, but the task initiating the
 * expedited grace period need not wake itself.
 *
738 739
 * Caller must hold sync_rcu_preempt_exp_mutex.
 */
740 741
static void rcu_report_exp_rnp(struct rcu_state *rsp, struct rcu_node *rnp,
			       bool wake)
742 743 744 745
{
	unsigned long flags;
	unsigned long mask;

P
Paul E. McKenney 已提交
746
	raw_spin_lock_irqsave(&rnp->lock, flags);
747
	for (;;) {
748 749
		if (!sync_rcu_preempt_exp_done(rnp)) {
			raw_spin_unlock_irqrestore(&rnp->lock, flags);
750
			break;
751
		}
752
		if (rnp->parent == NULL) {
753
			raw_spin_unlock_irqrestore(&rnp->lock, flags);
754 755
			if (wake)
				wake_up(&sync_rcu_preempt_exp_wq);
756 757 758
			break;
		}
		mask = rnp->grpmask;
P
Paul E. McKenney 已提交
759
		raw_spin_unlock(&rnp->lock); /* irqs remain disabled */
760
		rnp = rnp->parent;
P
Paul E. McKenney 已提交
761
		raw_spin_lock(&rnp->lock); /* irqs already disabled */
762 763 764 765 766 767 768 769 770 771 772 773 774 775
		rnp->expmask &= ~mask;
	}
}

/*
 * Snapshot the tasks blocking the newly started preemptible-RCU expedited
 * grace period for the specified rcu_node structure.  If there are no such
 * tasks, report it up the rcu_node hierarchy.
 *
 * Caller must hold sync_rcu_preempt_exp_mutex and rsp->onofflock.
 */
static void
sync_rcu_preempt_exp_init(struct rcu_state *rsp, struct rcu_node *rnp)
{
776
	unsigned long flags;
777
	int must_wait = 0;
778

779 780 781 782
	raw_spin_lock_irqsave(&rnp->lock, flags);
	if (list_empty(&rnp->blkd_tasks))
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
	else {
783
		rnp->exp_tasks = rnp->blkd_tasks.next;
784
		rcu_initiate_boost(rnp, flags);  /* releases rnp->lock */
785 786
		must_wait = 1;
	}
787
	if (!must_wait)
788
		rcu_report_exp_rnp(rsp, rnp, false); /* Don't wake self. */
789 790
}

791
/*
792 793
 * Wait for an rcu-preempt grace period, but expedite it.  The basic idea
 * is to invoke synchronize_sched_expedited() to push all the tasks to
794
 * the ->blkd_tasks lists and wait for this list to drain.
795 796 797
 */
void synchronize_rcu_expedited(void)
{
798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825
	unsigned long flags;
	struct rcu_node *rnp;
	struct rcu_state *rsp = &rcu_preempt_state;
	long snap;
	int trycount = 0;

	smp_mb(); /* Caller's modifications seen first by other CPUs. */
	snap = ACCESS_ONCE(sync_rcu_preempt_exp_count) + 1;
	smp_mb(); /* Above access cannot bleed into critical section. */

	/*
	 * Acquire lock, falling back to synchronize_rcu() if too many
	 * lock-acquisition failures.  Of course, if someone does the
	 * expedited grace period for us, just leave.
	 */
	while (!mutex_trylock(&sync_rcu_preempt_exp_mutex)) {
		if (trycount++ < 10)
			udelay(trycount * num_online_cpus());
		else {
			synchronize_rcu();
			return;
		}
		if ((ACCESS_ONCE(sync_rcu_preempt_exp_count) - snap) > 0)
			goto mb_ret; /* Others did our work for us. */
	}
	if ((ACCESS_ONCE(sync_rcu_preempt_exp_count) - snap) > 0)
		goto unlock_mb_ret; /* Others did our work for us. */

826
	/* force all RCU readers onto ->blkd_tasks lists. */
827 828
	synchronize_sched_expedited();

P
Paul E. McKenney 已提交
829
	raw_spin_lock_irqsave(&rsp->onofflock, flags);
830 831 832

	/* Initialize ->expmask for all non-leaf rcu_node structures. */
	rcu_for_each_nonleaf_node_breadth_first(rsp, rnp) {
P
Paul E. McKenney 已提交
833
		raw_spin_lock(&rnp->lock); /* irqs already disabled. */
834
		rnp->expmask = rnp->qsmaskinit;
P
Paul E. McKenney 已提交
835
		raw_spin_unlock(&rnp->lock); /* irqs remain disabled. */
836 837
	}

838
	/* Snapshot current state of ->blkd_tasks lists. */
839 840 841 842 843
	rcu_for_each_leaf_node(rsp, rnp)
		sync_rcu_preempt_exp_init(rsp, rnp);
	if (NUM_RCU_NODES > 1)
		sync_rcu_preempt_exp_init(rsp, rcu_get_root(rsp));

P
Paul E. McKenney 已提交
844
	raw_spin_unlock_irqrestore(&rsp->onofflock, flags);
845

846
	/* Wait for snapshotted ->blkd_tasks lists to drain. */
847 848 849 850 851 852 853 854 855 856 857
	rnp = rcu_get_root(rsp);
	wait_event(sync_rcu_preempt_exp_wq,
		   sync_rcu_preempt_exp_done(rnp));

	/* Clean up and exit. */
	smp_mb(); /* ensure expedited GP seen before counter increment. */
	ACCESS_ONCE(sync_rcu_preempt_exp_count)++;
unlock_mb_ret:
	mutex_unlock(&sync_rcu_preempt_exp_mutex);
mb_ret:
	smp_mb(); /* ensure subsequent action seen after grace period. */
858 859 860
}
EXPORT_SYMBOL_GPL(synchronize_rcu_expedited);

861
/*
P
Paul E. McKenney 已提交
862
 * Check to see if there is any immediate preemptible-RCU-related work
863 864 865 866 867 868 869 870 871
 * to be done.
 */
static int rcu_preempt_pending(int cpu)
{
	return __rcu_pending(&rcu_preempt_state,
			     &per_cpu(rcu_preempt_data, cpu));
}

/*
P
Paul E. McKenney 已提交
872
 * Does preemptible RCU need the CPU to stay out of dynticks mode?
873 874 875 876 877 878
 */
static int rcu_preempt_needs_cpu(int cpu)
{
	return !!per_cpu(rcu_preempt_data, cpu).nxtlist;
}

879 880 881 882 883 884 885 886 887
/**
 * rcu_barrier - Wait until all in-flight call_rcu() callbacks complete.
 */
void rcu_barrier(void)
{
	_rcu_barrier(&rcu_preempt_state, call_rcu);
}
EXPORT_SYMBOL_GPL(rcu_barrier);

888
/*
P
Paul E. McKenney 已提交
889
 * Initialize preemptible RCU's per-CPU data.
890 891 892 893 894 895
 */
static void __cpuinit rcu_preempt_init_percpu_data(int cpu)
{
	rcu_init_percpu_data(cpu, &rcu_preempt_state, 1);
}

896
/*
P
Paul E. McKenney 已提交
897
 * Move preemptible RCU's callbacks from dying CPU to other online CPU.
898
 */
899
static void rcu_preempt_send_cbs_to_online(void)
900
{
901
	rcu_send_cbs_to_online(&rcu_preempt_state);
902 903
}

904
/*
P
Paul E. McKenney 已提交
905
 * Initialize preemptible RCU's state structures.
906 907 908
 */
static void __init __rcu_init_preempt(void)
{
909
	rcu_init_one(&rcu_preempt_state, &rcu_preempt_data);
910 911
}

912
/*
P
Paul E. McKenney 已提交
913
 * Check for a task exiting while in a preemptible-RCU read-side
914 915 916 917 918 919 920 921 922 923 924
 * critical section, clean up if so.  No need to issue warnings,
 * as debug_check_no_locks_held() already does this if lockdep
 * is enabled.
 */
void exit_rcu(void)
{
	struct task_struct *t = current;

	if (t->rcu_read_lock_nesting == 0)
		return;
	t->rcu_read_lock_nesting = 1;
925
	__rcu_read_unlock();
926 927 928 929
}

#else /* #ifdef CONFIG_TREE_PREEMPT_RCU */

930 931
static struct rcu_state *rcu_state = &rcu_sched_state;

932 933 934
/*
 * Tell them what RCU they are running.
 */
935
static void __init rcu_bootup_announce(void)
936 937
{
	printk(KERN_INFO "Hierarchical RCU implementation.\n");
938
	rcu_bootup_announce_oddness();
939 940 941 942 943 944 945 946 947 948 949
}

/*
 * Return the number of RCU batches processed thus far for debug & stats.
 */
long rcu_batches_completed(void)
{
	return rcu_batches_completed_sched();
}
EXPORT_SYMBOL_GPL(rcu_batches_completed);

950 951 952 953 954 955 956 957 958 959
/*
 * Force a quiescent state for RCU, which, because there is no preemptible
 * RCU, becomes the same as rcu-sched.
 */
void rcu_force_quiescent_state(void)
{
	rcu_sched_force_quiescent_state();
}
EXPORT_SYMBOL_GPL(rcu_force_quiescent_state);

960
/*
P
Paul E. McKenney 已提交
961
 * Because preemptible RCU does not exist, we never have to check for
962 963
 * CPUs being in quiescent states.
 */
964
static void rcu_preempt_note_context_switch(int cpu)
965 966 967
{
}

968
/*
P
Paul E. McKenney 已提交
969
 * Because preemptible RCU does not exist, there are never any preempted
970 971
 * RCU readers.
 */
972
static int rcu_preempt_blocked_readers_cgp(struct rcu_node *rnp)
973 974 975 976
{
	return 0;
}

977 978 979
#ifdef CONFIG_HOTPLUG_CPU

/* Because preemptible RCU does not exist, no quieting of tasks. */
P
Paul E. McKenney 已提交
980
static void rcu_report_unblock_qs_rnp(struct rcu_node *rnp, unsigned long flags)
981
{
P
Paul E. McKenney 已提交
982
	raw_spin_unlock_irqrestore(&rnp->lock, flags);
983 984 985 986
}

#endif /* #ifdef CONFIG_HOTPLUG_CPU */

987
/*
P
Paul E. McKenney 已提交
988
 * Because preemptible RCU does not exist, we never have to check for
989 990 991 992 993 994
 * tasks blocked within RCU read-side critical sections.
 */
static void rcu_print_detail_task_stall(struct rcu_state *rsp)
{
}

995
/*
P
Paul E. McKenney 已提交
996
 * Because preemptible RCU does not exist, we never have to check for
997 998
 * tasks blocked within RCU read-side critical sections.
 */
999
static int rcu_print_task_stall(struct rcu_node *rnp)
1000
{
1001
	return 0;
1002 1003
}

1004 1005 1006 1007 1008 1009 1010 1011
/*
 * Because preemptible RCU does not exist, there is no need to suppress
 * its CPU stall warnings.
 */
static void rcu_preempt_stall_reset(void)
{
}

1012
/*
P
Paul E. McKenney 已提交
1013
 * Because there is no preemptible RCU, there can be no readers blocked,
1014 1015
 * so there is no need to check for blocked tasks.  So check only for
 * bogus qsmask values.
1016 1017 1018
 */
static void rcu_preempt_check_blocked_tasks(struct rcu_node *rnp)
{
1019
	WARN_ON_ONCE(rnp->qsmask);
1020 1021
}

1022 1023
#ifdef CONFIG_HOTPLUG_CPU

1024
/*
P
Paul E. McKenney 已提交
1025
 * Because preemptible RCU does not exist, it never needs to migrate
1026 1027 1028
 * tasks that were blocked within RCU read-side critical sections, and
 * such non-existent tasks cannot possibly have been blocking the current
 * grace period.
1029
 */
1030 1031 1032
static int rcu_preempt_offline_tasks(struct rcu_state *rsp,
				     struct rcu_node *rnp,
				     struct rcu_data *rdp)
1033
{
1034
	return 0;
1035 1036
}

1037
/*
P
Paul E. McKenney 已提交
1038
 * Because preemptible RCU does not exist, it never needs CPU-offline
1039 1040 1041 1042 1043 1044 1045 1046
 * processing.
 */
static void rcu_preempt_offline_cpu(int cpu)
{
}

#endif /* #ifdef CONFIG_HOTPLUG_CPU */

1047
/*
P
Paul E. McKenney 已提交
1048
 * Because preemptible RCU does not exist, it never has any callbacks
1049 1050
 * to check.
 */
1051
static void rcu_preempt_check_callbacks(int cpu)
1052 1053 1054 1055
{
}

/*
P
Paul E. McKenney 已提交
1056
 * Because preemptible RCU does not exist, it never has any callbacks
1057 1058
 * to process.
 */
1059
static void rcu_preempt_process_callbacks(void)
1060 1061 1062
{
}

1063 1064
/*
 * Wait for an rcu-preempt grace period, but make it happen quickly.
P
Paul E. McKenney 已提交
1065
 * But because preemptible RCU does not exist, map to rcu-sched.
1066 1067 1068 1069 1070 1071 1072
 */
void synchronize_rcu_expedited(void)
{
	synchronize_sched_expedited();
}
EXPORT_SYMBOL_GPL(synchronize_rcu_expedited);

1073 1074 1075
#ifdef CONFIG_HOTPLUG_CPU

/*
P
Paul E. McKenney 已提交
1076
 * Because preemptible RCU does not exist, there is never any need to
1077 1078 1079
 * report on tasks preempted in RCU read-side critical sections during
 * expedited RCU grace periods.
 */
1080 1081
static void rcu_report_exp_rnp(struct rcu_state *rsp, struct rcu_node *rnp,
			       bool wake)
1082 1083 1084 1085 1086
{
}

#endif /* #ifdef CONFIG_HOTPLUG_CPU */

1087
/*
P
Paul E. McKenney 已提交
1088
 * Because preemptible RCU does not exist, it never has any work to do.
1089 1090 1091 1092 1093 1094 1095
 */
static int rcu_preempt_pending(int cpu)
{
	return 0;
}

/*
P
Paul E. McKenney 已提交
1096
 * Because preemptible RCU does not exist, it never needs any CPU.
1097 1098 1099 1100 1101 1102
 */
static int rcu_preempt_needs_cpu(int cpu)
{
	return 0;
}

1103
/*
P
Paul E. McKenney 已提交
1104
 * Because preemptible RCU does not exist, rcu_barrier() is just
1105 1106 1107 1108 1109 1110 1111 1112
 * another name for rcu_barrier_sched().
 */
void rcu_barrier(void)
{
	rcu_barrier_sched();
}
EXPORT_SYMBOL_GPL(rcu_barrier);

1113
/*
P
Paul E. McKenney 已提交
1114
 * Because preemptible RCU does not exist, there is no per-CPU
1115 1116 1117 1118 1119 1120
 * data to initialize.
 */
static void __cpuinit rcu_preempt_init_percpu_data(int cpu)
{
}

1121
/*
P
Paul E. McKenney 已提交
1122
 * Because there is no preemptible RCU, there are no callbacks to move.
1123
 */
1124
static void rcu_preempt_send_cbs_to_online(void)
1125 1126 1127
{
}

1128
/*
P
Paul E. McKenney 已提交
1129
 * Because preemptible RCU does not exist, it need not be initialized.
1130 1131 1132 1133 1134
 */
static void __init __rcu_init_preempt(void)
{
}

1135
#endif /* #else #ifdef CONFIG_TREE_PREEMPT_RCU */
1136

1137 1138 1139 1140
#ifdef CONFIG_RCU_BOOST

#include "rtmutex_common.h"

1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153
#ifdef CONFIG_RCU_TRACE

static void rcu_initiate_boost_trace(struct rcu_node *rnp)
{
	if (list_empty(&rnp->blkd_tasks))
		rnp->n_balk_blkd_tasks++;
	else if (rnp->exp_tasks == NULL && rnp->gp_tasks == NULL)
		rnp->n_balk_exp_gp_tasks++;
	else if (rnp->gp_tasks != NULL && rnp->boost_tasks != NULL)
		rnp->n_balk_boost_tasks++;
	else if (rnp->gp_tasks != NULL && rnp->qsmask != 0)
		rnp->n_balk_notblocked++;
	else if (rnp->gp_tasks != NULL &&
1154
		 ULONG_CMP_LT(jiffies, rnp->boost_time))
1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167
		rnp->n_balk_notyet++;
	else
		rnp->n_balk_nos++;
}

#else /* #ifdef CONFIG_RCU_TRACE */

static void rcu_initiate_boost_trace(struct rcu_node *rnp)
{
}

#endif /* #else #ifdef CONFIG_RCU_TRACE */

1168 1169
static struct lock_class_key rcu_boost_class;

1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204
/*
 * Carry out RCU priority boosting on the task indicated by ->exp_tasks
 * or ->boost_tasks, advancing the pointer to the next task in the
 * ->blkd_tasks list.
 *
 * Note that irqs must be enabled: boosting the task can block.
 * Returns 1 if there are more tasks needing to be boosted.
 */
static int rcu_boost(struct rcu_node *rnp)
{
	unsigned long flags;
	struct rt_mutex mtx;
	struct task_struct *t;
	struct list_head *tb;

	if (rnp->exp_tasks == NULL && rnp->boost_tasks == NULL)
		return 0;  /* Nothing left to boost. */

	raw_spin_lock_irqsave(&rnp->lock, flags);

	/*
	 * Recheck under the lock: all tasks in need of boosting
	 * might exit their RCU read-side critical sections on their own.
	 */
	if (rnp->exp_tasks == NULL && rnp->boost_tasks == NULL) {
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
		return 0;
	}

	/*
	 * Preferentially boost tasks blocking expedited grace periods.
	 * This cannot starve the normal grace periods because a second
	 * expedited grace period must boost all blocked tasks, including
	 * those blocking the pre-existing normal grace period.
	 */
1205
	if (rnp->exp_tasks != NULL) {
1206
		tb = rnp->exp_tasks;
1207 1208
		rnp->n_exp_boosts++;
	} else {
1209
		tb = rnp->boost_tasks;
1210 1211 1212
		rnp->n_normal_boosts++;
	}
	rnp->n_tasks_boosted++;
1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231

	/*
	 * We boost task t by manufacturing an rt_mutex that appears to
	 * be held by task t.  We leave a pointer to that rt_mutex where
	 * task t can find it, and task t will release the mutex when it
	 * exits its outermost RCU read-side critical section.  Then
	 * simply acquiring this artificial rt_mutex will boost task
	 * t's priority.  (Thanks to tglx for suggesting this approach!)
	 *
	 * Note that task t must acquire rnp->lock to remove itself from
	 * the ->blkd_tasks list, which it will do from exit() if from
	 * nowhere else.  We therefore are guaranteed that task t will
	 * stay around at least until we drop rnp->lock.  Note that
	 * rnp->lock also resolves races between our priority boosting
	 * and task t's exiting its outermost RCU read-side critical
	 * section.
	 */
	t = container_of(tb, struct task_struct, rcu_node_entry);
	rt_mutex_init_proxy_locked(&mtx, t);
1232 1233 1234
	/* Avoid lockdep false positives.  This rt_mutex is its own thing. */
	lockdep_set_class_and_name(&mtx.wait_lock, &rcu_boost_class,
				   "rcu_boost_mutex");
1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250
	t->rcu_boost_mutex = &mtx;
	raw_spin_unlock_irqrestore(&rnp->lock, flags);
	rt_mutex_lock(&mtx);  /* Side effect: boosts task t's priority. */
	rt_mutex_unlock(&mtx);  /* Keep lockdep happy. */

	return rnp->exp_tasks != NULL || rnp->boost_tasks != NULL;
}

/*
 * Timer handler to initiate waking up of boost kthreads that
 * have yielded the CPU due to excessive numbers of tasks to
 * boost.  We wake up the per-rcu_node kthread, which in turn
 * will wake up the booster kthread.
 */
static void rcu_boost_kthread_timer(unsigned long arg)
{
1251
	invoke_rcu_node_kthread((struct rcu_node *)arg);
1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263
}

/*
 * Priority-boosting kthread.  One per leaf rcu_node and one for the
 * root rcu_node.
 */
static int rcu_boost_kthread(void *arg)
{
	struct rcu_node *rnp = (struct rcu_node *)arg;
	int spincnt = 0;
	int more2boost;

1264
	trace_rcu_utilization("Start boost kthread@init");
1265
	for (;;) {
1266
		rnp->boost_kthread_status = RCU_KTHREAD_WAITING;
1267
		trace_rcu_utilization("End boost kthread@rcu_wait");
1268
		rcu_wait(rnp->boost_tasks || rnp->exp_tasks);
1269
		trace_rcu_utilization("Start boost kthread@rcu_wait");
1270
		rnp->boost_kthread_status = RCU_KTHREAD_RUNNING;
1271 1272 1273 1274 1275 1276
		more2boost = rcu_boost(rnp);
		if (more2boost)
			spincnt++;
		else
			spincnt = 0;
		if (spincnt > 10) {
1277
			trace_rcu_utilization("End boost kthread@rcu_yield");
1278
			rcu_yield(rcu_boost_kthread_timer, (unsigned long)rnp);
1279
			trace_rcu_utilization("Start boost kthread@rcu_yield");
1280 1281 1282
			spincnt = 0;
		}
	}
1283
	/* NOTREACHED */
1284
	trace_rcu_utilization("End boost kthread@notreached");
1285 1286 1287 1288 1289 1290 1291 1292 1293
	return 0;
}

/*
 * Check to see if it is time to start boosting RCU readers that are
 * blocking the current grace period, and, if so, tell the per-rcu_node
 * kthread to start boosting them.  If there is an expedited grace
 * period in progress, it is always time to boost.
 *
1294 1295 1296
 * The caller must hold rnp->lock, which this function releases,
 * but irqs remain disabled.  The ->boost_kthread_task is immortal,
 * so we don't need to worry about it going away.
1297
 */
1298
static void rcu_initiate_boost(struct rcu_node *rnp, unsigned long flags)
1299 1300 1301
{
	struct task_struct *t;

1302 1303
	if (!rcu_preempt_blocked_readers_cgp(rnp) && rnp->exp_tasks == NULL) {
		rnp->n_balk_exp_gp_tasks++;
1304
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
1305
		return;
1306
	}
1307 1308 1309 1310 1311 1312 1313
	if (rnp->exp_tasks != NULL ||
	    (rnp->gp_tasks != NULL &&
	     rnp->boost_tasks == NULL &&
	     rnp->qsmask == 0 &&
	     ULONG_CMP_GE(jiffies, rnp->boost_time))) {
		if (rnp->exp_tasks == NULL)
			rnp->boost_tasks = rnp->gp_tasks;
1314
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
1315 1316 1317
		t = rnp->boost_kthread_task;
		if (t != NULL)
			wake_up_process(t);
1318
	} else {
1319
		rcu_initiate_boost_trace(rnp);
1320 1321
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
	}
1322 1323
}

1324 1325 1326 1327 1328 1329 1330 1331 1332
/*
 * Wake up the per-CPU kthread to invoke RCU callbacks.
 */
static void invoke_rcu_callbacks_kthread(void)
{
	unsigned long flags;

	local_irq_save(flags);
	__this_cpu_write(rcu_cpu_has_work, 1);
1333 1334 1335
	if (__this_cpu_read(rcu_cpu_kthread_task) != NULL &&
	    current != __this_cpu_read(rcu_cpu_kthread_task))
		wake_up_process(__this_cpu_read(rcu_cpu_kthread_task));
1336 1337 1338
	local_irq_restore(flags);
}

1339 1340 1341 1342 1343
/*
 * Set the affinity of the boost kthread.  The CPU-hotplug locks are
 * held, so no one should be messing with the existence of the boost
 * kthread.
 */
1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378
static void rcu_boost_kthread_setaffinity(struct rcu_node *rnp,
					  cpumask_var_t cm)
{
	struct task_struct *t;

	t = rnp->boost_kthread_task;
	if (t != NULL)
		set_cpus_allowed_ptr(rnp->boost_kthread_task, cm);
}

#define RCU_BOOST_DELAY_JIFFIES DIV_ROUND_UP(CONFIG_RCU_BOOST_DELAY * HZ, 1000)

/*
 * Do priority-boost accounting for the start of a new grace period.
 */
static void rcu_preempt_boost_start_gp(struct rcu_node *rnp)
{
	rnp->boost_time = jiffies + RCU_BOOST_DELAY_JIFFIES;
}

/*
 * Create an RCU-boost kthread for the specified node if one does not
 * already exist.  We only create this kthread for preemptible RCU.
 * Returns zero if all is well, a negated errno otherwise.
 */
static int __cpuinit rcu_spawn_one_boost_kthread(struct rcu_state *rsp,
						 struct rcu_node *rnp,
						 int rnp_index)
{
	unsigned long flags;
	struct sched_param sp;
	struct task_struct *t;

	if (&rcu_preempt_state != rsp)
		return 0;
1379
	rsp->boost = 1;
1380 1381 1382
	if (rnp->boost_kthread_task != NULL)
		return 0;
	t = kthread_create(rcu_boost_kthread, (void *)rnp,
1383
			   "rcub/%d", rnp_index);
1384 1385 1386 1387 1388
	if (IS_ERR(t))
		return PTR_ERR(t);
	raw_spin_lock_irqsave(&rnp->lock, flags);
	rnp->boost_kthread_task = t;
	raw_spin_unlock_irqrestore(&rnp->lock, flags);
1389
	sp.sched_priority = RCU_BOOST_PRIO;
1390
	sched_setscheduler_nocheck(t, SCHED_FIFO, &sp);
1391
	wake_up_process(t); /* get to TASK_INTERRUPTIBLE quickly. */
1392 1393 1394
	return 0;
}

1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483
#ifdef CONFIG_HOTPLUG_CPU

/*
 * Stop the RCU's per-CPU kthread when its CPU goes offline,.
 */
static void rcu_stop_cpu_kthread(int cpu)
{
	struct task_struct *t;

	/* Stop the CPU's kthread. */
	t = per_cpu(rcu_cpu_kthread_task, cpu);
	if (t != NULL) {
		per_cpu(rcu_cpu_kthread_task, cpu) = NULL;
		kthread_stop(t);
	}
}

#endif /* #ifdef CONFIG_HOTPLUG_CPU */

static void rcu_kthread_do_work(void)
{
	rcu_do_batch(&rcu_sched_state, &__get_cpu_var(rcu_sched_data));
	rcu_do_batch(&rcu_bh_state, &__get_cpu_var(rcu_bh_data));
	rcu_preempt_do_callbacks();
}

/*
 * Wake up the specified per-rcu_node-structure kthread.
 * Because the per-rcu_node kthreads are immortal, we don't need
 * to do anything to keep them alive.
 */
static void invoke_rcu_node_kthread(struct rcu_node *rnp)
{
	struct task_struct *t;

	t = rnp->node_kthread_task;
	if (t != NULL)
		wake_up_process(t);
}

/*
 * Set the specified CPU's kthread to run RT or not, as specified by
 * the to_rt argument.  The CPU-hotplug locks are held, so the task
 * is not going away.
 */
static void rcu_cpu_kthread_setrt(int cpu, int to_rt)
{
	int policy;
	struct sched_param sp;
	struct task_struct *t;

	t = per_cpu(rcu_cpu_kthread_task, cpu);
	if (t == NULL)
		return;
	if (to_rt) {
		policy = SCHED_FIFO;
		sp.sched_priority = RCU_KTHREAD_PRIO;
	} else {
		policy = SCHED_NORMAL;
		sp.sched_priority = 0;
	}
	sched_setscheduler_nocheck(t, policy, &sp);
}

/*
 * Timer handler to initiate the waking up of per-CPU kthreads that
 * have yielded the CPU due to excess numbers of RCU callbacks.
 * We wake up the per-rcu_node kthread, which in turn will wake up
 * the booster kthread.
 */
static void rcu_cpu_kthread_timer(unsigned long arg)
{
	struct rcu_data *rdp = per_cpu_ptr(rcu_state->rda, arg);
	struct rcu_node *rnp = rdp->mynode;

	atomic_or(rdp->grpmask, &rnp->wakemask);
	invoke_rcu_node_kthread(rnp);
}

/*
 * Drop to non-real-time priority and yield, but only after posting a
 * timer that will cause us to regain our real-time priority if we
 * remain preempted.  Either way, we restore our real-time priority
 * before returning.
 */
static void rcu_yield(void (*f)(unsigned long), unsigned long arg)
{
	struct sched_param sp;
	struct timer_list yield_timer;
1484
	int prio = current->rt_priority;
1485 1486 1487 1488 1489 1490 1491

	setup_timer_on_stack(&yield_timer, f, arg);
	mod_timer(&yield_timer, jiffies + 2);
	sp.sched_priority = 0;
	sched_setscheduler_nocheck(current, SCHED_NORMAL, &sp);
	set_user_nice(current, 19);
	schedule();
1492 1493
	set_user_nice(current, 0);
	sp.sched_priority = prio;
1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530
	sched_setscheduler_nocheck(current, SCHED_FIFO, &sp);
	del_timer(&yield_timer);
}

/*
 * Handle cases where the rcu_cpu_kthread() ends up on the wrong CPU.
 * This can happen while the corresponding CPU is either coming online
 * or going offline.  We cannot wait until the CPU is fully online
 * before starting the kthread, because the various notifier functions
 * can wait for RCU grace periods.  So we park rcu_cpu_kthread() until
 * the corresponding CPU is online.
 *
 * Return 1 if the kthread needs to stop, 0 otherwise.
 *
 * Caller must disable bh.  This function can momentarily enable it.
 */
static int rcu_cpu_kthread_should_stop(int cpu)
{
	while (cpu_is_offline(cpu) ||
	       !cpumask_equal(&current->cpus_allowed, cpumask_of(cpu)) ||
	       smp_processor_id() != cpu) {
		if (kthread_should_stop())
			return 1;
		per_cpu(rcu_cpu_kthread_status, cpu) = RCU_KTHREAD_OFFCPU;
		per_cpu(rcu_cpu_kthread_cpu, cpu) = raw_smp_processor_id();
		local_bh_enable();
		schedule_timeout_uninterruptible(1);
		if (!cpumask_equal(&current->cpus_allowed, cpumask_of(cpu)))
			set_cpus_allowed_ptr(current, cpumask_of(cpu));
		local_bh_disable();
	}
	per_cpu(rcu_cpu_kthread_cpu, cpu) = cpu;
	return 0;
}

/*
 * Per-CPU kernel thread that invokes RCU callbacks.  This replaces the
1531 1532
 * RCU softirq used in flavors and configurations of RCU that do not
 * support RCU priority boosting.
1533 1534 1535 1536 1537 1538 1539 1540 1541 1542
 */
static int rcu_cpu_kthread(void *arg)
{
	int cpu = (int)(long)arg;
	unsigned long flags;
	int spincnt = 0;
	unsigned int *statusp = &per_cpu(rcu_cpu_kthread_status, cpu);
	char work;
	char *workp = &per_cpu(rcu_cpu_has_work, cpu);

1543
	trace_rcu_utilization("Start CPU kthread@init");
1544 1545
	for (;;) {
		*statusp = RCU_KTHREAD_WAITING;
1546
		trace_rcu_utilization("End CPU kthread@rcu_wait");
1547
		rcu_wait(*workp != 0 || kthread_should_stop());
1548
		trace_rcu_utilization("Start CPU kthread@rcu_wait");
1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568
		local_bh_disable();
		if (rcu_cpu_kthread_should_stop(cpu)) {
			local_bh_enable();
			break;
		}
		*statusp = RCU_KTHREAD_RUNNING;
		per_cpu(rcu_cpu_kthread_loops, cpu)++;
		local_irq_save(flags);
		work = *workp;
		*workp = 0;
		local_irq_restore(flags);
		if (work)
			rcu_kthread_do_work();
		local_bh_enable();
		if (*workp != 0)
			spincnt++;
		else
			spincnt = 0;
		if (spincnt > 10) {
			*statusp = RCU_KTHREAD_YIELDING;
1569
			trace_rcu_utilization("End CPU kthread@rcu_yield");
1570
			rcu_yield(rcu_cpu_kthread_timer, (unsigned long)cpu);
1571
			trace_rcu_utilization("Start CPU kthread@rcu_yield");
1572 1573 1574 1575
			spincnt = 0;
		}
	}
	*statusp = RCU_KTHREAD_STOPPED;
1576
	trace_rcu_utilization("End CPU kthread@term");
1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605
	return 0;
}

/*
 * Spawn a per-CPU kthread, setting up affinity and priority.
 * Because the CPU hotplug lock is held, no other CPU will be attempting
 * to manipulate rcu_cpu_kthread_task.  There might be another CPU
 * attempting to access it during boot, but the locking in kthread_bind()
 * will enforce sufficient ordering.
 *
 * Please note that we cannot simply refuse to wake up the per-CPU
 * kthread because kthreads are created in TASK_UNINTERRUPTIBLE state,
 * which can result in softlockup complaints if the task ends up being
 * idle for more than a couple of minutes.
 *
 * However, please note also that we cannot bind the per-CPU kthread to its
 * CPU until that CPU is fully online.  We also cannot wait until the
 * CPU is fully online before we create its per-CPU kthread, as this would
 * deadlock the system when CPU notifiers tried waiting for grace
 * periods.  So we bind the per-CPU kthread to its CPU only if the CPU
 * is online.  If its CPU is not yet fully online, then the code in
 * rcu_cpu_kthread() will wait until it is fully online, and then do
 * the binding.
 */
static int __cpuinit rcu_spawn_one_cpu_kthread(int cpu)
{
	struct sched_param sp;
	struct task_struct *t;

1606
	if (!rcu_scheduler_fully_active ||
1607 1608
	    per_cpu(rcu_cpu_kthread_task, cpu) != NULL)
		return 0;
E
Eric Dumazet 已提交
1609 1610 1611
	t = kthread_create_on_node(rcu_cpu_kthread,
				   (void *)(long)cpu,
				   cpu_to_node(cpu),
1612
				   "rcuc/%d", cpu);
1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715
	if (IS_ERR(t))
		return PTR_ERR(t);
	if (cpu_online(cpu))
		kthread_bind(t, cpu);
	per_cpu(rcu_cpu_kthread_cpu, cpu) = cpu;
	WARN_ON_ONCE(per_cpu(rcu_cpu_kthread_task, cpu) != NULL);
	sp.sched_priority = RCU_KTHREAD_PRIO;
	sched_setscheduler_nocheck(t, SCHED_FIFO, &sp);
	per_cpu(rcu_cpu_kthread_task, cpu) = t;
	wake_up_process(t); /* Get to TASK_INTERRUPTIBLE quickly. */
	return 0;
}

/*
 * Per-rcu_node kthread, which is in charge of waking up the per-CPU
 * kthreads when needed.  We ignore requests to wake up kthreads
 * for offline CPUs, which is OK because force_quiescent_state()
 * takes care of this case.
 */
static int rcu_node_kthread(void *arg)
{
	int cpu;
	unsigned long flags;
	unsigned long mask;
	struct rcu_node *rnp = (struct rcu_node *)arg;
	struct sched_param sp;
	struct task_struct *t;

	for (;;) {
		rnp->node_kthread_status = RCU_KTHREAD_WAITING;
		rcu_wait(atomic_read(&rnp->wakemask) != 0);
		rnp->node_kthread_status = RCU_KTHREAD_RUNNING;
		raw_spin_lock_irqsave(&rnp->lock, flags);
		mask = atomic_xchg(&rnp->wakemask, 0);
		rcu_initiate_boost(rnp, flags); /* releases rnp->lock. */
		for (cpu = rnp->grplo; cpu <= rnp->grphi; cpu++, mask >>= 1) {
			if ((mask & 0x1) == 0)
				continue;
			preempt_disable();
			t = per_cpu(rcu_cpu_kthread_task, cpu);
			if (!cpu_online(cpu) || t == NULL) {
				preempt_enable();
				continue;
			}
			per_cpu(rcu_cpu_has_work, cpu) = 1;
			sp.sched_priority = RCU_KTHREAD_PRIO;
			sched_setscheduler_nocheck(t, SCHED_FIFO, &sp);
			preempt_enable();
		}
	}
	/* NOTREACHED */
	rnp->node_kthread_status = RCU_KTHREAD_STOPPED;
	return 0;
}

/*
 * Set the per-rcu_node kthread's affinity to cover all CPUs that are
 * served by the rcu_node in question.  The CPU hotplug lock is still
 * held, so the value of rnp->qsmaskinit will be stable.
 *
 * We don't include outgoingcpu in the affinity set, use -1 if there is
 * no outgoing CPU.  If there are no CPUs left in the affinity set,
 * this function allows the kthread to execute on any CPU.
 */
static void rcu_node_kthread_setaffinity(struct rcu_node *rnp, int outgoingcpu)
{
	cpumask_var_t cm;
	int cpu;
	unsigned long mask = rnp->qsmaskinit;

	if (rnp->node_kthread_task == NULL)
		return;
	if (!alloc_cpumask_var(&cm, GFP_KERNEL))
		return;
	cpumask_clear(cm);
	for (cpu = rnp->grplo; cpu <= rnp->grphi; cpu++, mask >>= 1)
		if ((mask & 0x1) && cpu != outgoingcpu)
			cpumask_set_cpu(cpu, cm);
	if (cpumask_weight(cm) == 0) {
		cpumask_setall(cm);
		for (cpu = rnp->grplo; cpu <= rnp->grphi; cpu++)
			cpumask_clear_cpu(cpu, cm);
		WARN_ON_ONCE(cpumask_weight(cm) == 0);
	}
	set_cpus_allowed_ptr(rnp->node_kthread_task, cm);
	rcu_boost_kthread_setaffinity(rnp, cm);
	free_cpumask_var(cm);
}

/*
 * Spawn a per-rcu_node kthread, setting priority and affinity.
 * Called during boot before online/offline can happen, or, if
 * during runtime, with the main CPU-hotplug locks held.  So only
 * one of these can be executing at a time.
 */
static int __cpuinit rcu_spawn_one_node_kthread(struct rcu_state *rsp,
						struct rcu_node *rnp)
{
	unsigned long flags;
	int rnp_index = rnp - &rsp->node[0];
	struct sched_param sp;
	struct task_struct *t;

1716
	if (!rcu_scheduler_fully_active ||
1717 1718 1719 1720
	    rnp->qsmaskinit == 0)
		return 0;
	if (rnp->node_kthread_task == NULL) {
		t = kthread_create(rcu_node_kthread, (void *)rnp,
1721
				   "rcun/%d", rnp_index);
1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741
		if (IS_ERR(t))
			return PTR_ERR(t);
		raw_spin_lock_irqsave(&rnp->lock, flags);
		rnp->node_kthread_task = t;
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
		sp.sched_priority = 99;
		sched_setscheduler_nocheck(t, SCHED_FIFO, &sp);
		wake_up_process(t); /* get to TASK_INTERRUPTIBLE quickly. */
	}
	return rcu_spawn_one_boost_kthread(rsp, rnp, rnp_index);
}

/*
 * Spawn all kthreads -- called as soon as the scheduler is running.
 */
static int __init rcu_spawn_kthreads(void)
{
	int cpu;
	struct rcu_node *rnp;

1742
	rcu_scheduler_fully_active = 1;
1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763
	for_each_possible_cpu(cpu) {
		per_cpu(rcu_cpu_has_work, cpu) = 0;
		if (cpu_online(cpu))
			(void)rcu_spawn_one_cpu_kthread(cpu);
	}
	rnp = rcu_get_root(rcu_state);
	(void)rcu_spawn_one_node_kthread(rcu_state, rnp);
	if (NUM_RCU_NODES > 1) {
		rcu_for_each_leaf_node(rcu_state, rnp)
			(void)rcu_spawn_one_node_kthread(rcu_state, rnp);
	}
	return 0;
}
early_initcall(rcu_spawn_kthreads);

static void __cpuinit rcu_prepare_kthreads(int cpu)
{
	struct rcu_data *rdp = per_cpu_ptr(rcu_state->rda, cpu);
	struct rcu_node *rnp = rdp->mynode;

	/* Fire up the incoming CPU's kthread and leaf rcu_node kthread. */
1764
	if (rcu_scheduler_fully_active) {
1765 1766 1767 1768 1769 1770
		(void)rcu_spawn_one_cpu_kthread(cpu);
		if (rnp->node_kthread_task == NULL)
			(void)rcu_spawn_one_node_kthread(rcu_state, rnp);
	}
}

1771 1772
#else /* #ifdef CONFIG_RCU_BOOST */

1773
static void rcu_initiate_boost(struct rcu_node *rnp, unsigned long flags)
1774
{
1775
	raw_spin_unlock_irqrestore(&rnp->lock, flags);
1776 1777
}

1778
static void invoke_rcu_callbacks_kthread(void)
1779
{
1780
	WARN_ON_ONCE(1);
1781 1782 1783 1784 1785 1786
}

static void rcu_preempt_boost_start_gp(struct rcu_node *rnp)
{
}

1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802
#ifdef CONFIG_HOTPLUG_CPU

static void rcu_stop_cpu_kthread(int cpu)
{
}

#endif /* #ifdef CONFIG_HOTPLUG_CPU */

static void rcu_node_kthread_setaffinity(struct rcu_node *rnp, int outgoingcpu)
{
}

static void rcu_cpu_kthread_setrt(int cpu, int to_rt)
{
}

1803 1804 1805 1806 1807 1808 1809
static int __init rcu_scheduler_really_started(void)
{
	rcu_scheduler_fully_active = 1;
	return 0;
}
early_initcall(rcu_scheduler_really_started);

1810 1811 1812 1813
static void __cpuinit rcu_prepare_kthreads(int cpu)
{
}

1814 1815
#endif /* #else #ifdef CONFIG_RCU_BOOST */

1816 1817 1818 1819 1820 1821 1822 1823 1824 1825
#ifndef CONFIG_SMP

void synchronize_sched_expedited(void)
{
	cond_resched();
}
EXPORT_SYMBOL_GPL(synchronize_sched_expedited);

#else /* #ifndef CONFIG_SMP */

1826 1827
static atomic_t sync_sched_expedited_started = ATOMIC_INIT(0);
static atomic_t sync_sched_expedited_done = ATOMIC_INIT(0);
1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854

static int synchronize_sched_expedited_cpu_stop(void *data)
{
	/*
	 * There must be a full memory barrier on each affected CPU
	 * between the time that try_stop_cpus() is called and the
	 * time that it returns.
	 *
	 * In the current initial implementation of cpu_stop, the
	 * above condition is already met when the control reaches
	 * this point and the following smp_mb() is not strictly
	 * necessary.  Do smp_mb() anyway for documentation and
	 * robustness against future implementation changes.
	 */
	smp_mb(); /* See above comment block. */
	return 0;
}

/*
 * Wait for an rcu-sched grace period to elapse, but use "big hammer"
 * approach to force grace period to end quickly.  This consumes
 * significant time on all CPUs, and is thus not recommended for
 * any sort of common-case code.
 *
 * Note that it is illegal to call this function while holding any
 * lock that is acquired by a CPU-hotplug notifier.  Failing to
 * observe this restriction will result in deadlock.
1855
 *
1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875
 * This implementation can be thought of as an application of ticket
 * locking to RCU, with sync_sched_expedited_started and
 * sync_sched_expedited_done taking on the roles of the halves
 * of the ticket-lock word.  Each task atomically increments
 * sync_sched_expedited_started upon entry, snapshotting the old value,
 * then attempts to stop all the CPUs.  If this succeeds, then each
 * CPU will have executed a context switch, resulting in an RCU-sched
 * grace period.  We are then done, so we use atomic_cmpxchg() to
 * update sync_sched_expedited_done to match our snapshot -- but
 * only if someone else has not already advanced past our snapshot.
 *
 * On the other hand, if try_stop_cpus() fails, we check the value
 * of sync_sched_expedited_done.  If it has advanced past our
 * initial snapshot, then someone else must have forced a grace period
 * some time after we took our snapshot.  In this case, our work is
 * done for us, and we can simply return.  Otherwise, we try again,
 * but keep our initial snapshot for purposes of checking for someone
 * doing our work for us.
 *
 * If we fail too many times in a row, we fall back to synchronize_sched().
1876 1877 1878
 */
void synchronize_sched_expedited(void)
{
1879
	int firstsnap, s, snap, trycount = 0;
1880

1881 1882
	/* Note that atomic_inc_return() implies full memory barrier. */
	firstsnap = snap = atomic_inc_return(&sync_sched_expedited_started);
1883
	get_online_cpus();
1884 1885 1886 1887 1888

	/*
	 * Each pass through the following loop attempts to force a
	 * context switch on each CPU.
	 */
1889 1890 1891 1892
	while (try_stop_cpus(cpu_online_mask,
			     synchronize_sched_expedited_cpu_stop,
			     NULL) == -EAGAIN) {
		put_online_cpus();
1893 1894

		/* No joy, try again later.  Or just synchronize_sched(). */
1895 1896 1897 1898 1899 1900
		if (trycount++ < 10)
			udelay(trycount * num_online_cpus());
		else {
			synchronize_sched();
			return;
		}
1901 1902 1903 1904

		/* Check to see if someone else did our work for us. */
		s = atomic_read(&sync_sched_expedited_done);
		if (UINT_CMP_GE((unsigned)s, (unsigned)firstsnap)) {
1905 1906 1907
			smp_mb(); /* ensure test happens before caller kfree */
			return;
		}
1908 1909 1910 1911 1912 1913 1914 1915 1916

		/*
		 * Refetching sync_sched_expedited_started allows later
		 * callers to piggyback on our grace period.  We subtract
		 * 1 to get the same token that the last incrementer got.
		 * We retry after they started, so our grace period works
		 * for them, and they started after our first try, so their
		 * grace period works for us.
		 */
1917
		get_online_cpus();
1918
		snap = atomic_read(&sync_sched_expedited_started);
1919
		smp_mb(); /* ensure read is before try_stop_cpus(). */
1920
	}
1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935

	/*
	 * Everyone up to our most recent fetch is covered by our grace
	 * period.  Update the counter, but only if our work is still
	 * relevant -- which it won't be if someone who started later
	 * than we did beat us to the punch.
	 */
	do {
		s = atomic_read(&sync_sched_expedited_done);
		if (UINT_CMP_GE((unsigned)s, (unsigned)snap)) {
			smp_mb(); /* ensure test happens before caller kfree */
			break;
		}
	} while (atomic_cmpxchg(&sync_sched_expedited_done, s, snap) != s);

1936 1937 1938 1939 1940 1941
	put_online_cpus();
}
EXPORT_SYMBOL_GPL(synchronize_sched_expedited);

#endif /* #else #ifndef CONFIG_SMP */

1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955
#if !defined(CONFIG_RCU_FAST_NO_HZ)

/*
 * Check to see if any future RCU-related work will need to be done
 * by the current CPU, even if none need be done immediately, returning
 * 1 if so.  This function is part of the RCU implementation; it is -not-
 * an exported member of the RCU API.
 *
 * Because we have preemptible RCU, just check whether this CPU needs
 * any flavor of RCU.  Do not chew up lots of CPU cycles with preemption
 * disabled in a most-likely vain attempt to cause RCU not to need this CPU.
 */
int rcu_needs_cpu(int cpu)
{
1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966
	return rcu_cpu_has_callbacks(cpu);
}

/*
 * Do the idle-entry grace-period work, which, because CONFIG_RCU_FAST_NO_HZ=y,
 * is nothing.
 */
static void rcu_prepare_for_idle(int cpu)
{
}

1967 1968 1969
#else /* #if !defined(CONFIG_RCU_FAST_NO_HZ) */

#define RCU_NEEDS_CPU_FLUSHES 5
1970
static DEFINE_PER_CPU(int, rcu_dyntick_drain);
1971
static DEFINE_PER_CPU(unsigned long, rcu_dyntick_holdoff);
1972 1973

/*
1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
 * Allow the CPU to enter dyntick-idle mode if either: (1) There are no
 * callbacks on this CPU, (2) this CPU has not yet attempted to enter
 * dyntick-idle mode, or (3) this CPU is in the process of attempting to
 * enter dyntick-idle mode.  Otherwise, if we have recently tried and failed
 * to enter dyntick-idle mode, we refuse to try to enter it.  After all,
 * it is better to incur scheduling-clock interrupts than to spin
 * continuously for the same time duration!
 */
int rcu_needs_cpu(int cpu)
{
	/* If no callbacks, RCU doesn't need the CPU. */
	if (!rcu_cpu_has_callbacks(cpu))
		return 0;
	/* Otherwise, RCU needs the CPU only if it recently tried and failed. */
	return per_cpu(rcu_dyntick_holdoff, cpu) == jiffies;
}

/*
 * Check to see if any RCU-related work can be done by the current CPU,
 * and if so, schedule a softirq to get it done.  This function is part
 * of the RCU implementation; it is -not- an exported member of the RCU API.
1995
 *
1996 1997 1998 1999 2000 2001
 * The idea is for the current CPU to clear out all work required by the
 * RCU core for the current grace period, so that this CPU can be permitted
 * to enter dyntick-idle mode.  In some cases, it will need to be awakened
 * at the end of the grace period by whatever CPU ends the grace period.
 * This allows CPUs to go dyntick-idle more quickly, and to reduce the
 * number of wakeups by a modest integer factor.
2002 2003 2004
 *
 * Because it is not legal to invoke rcu_process_callbacks() with irqs
 * disabled, we do one pass of force_quiescent_state(), then do a
2005
 * invoke_rcu_core() to cause rcu_process_callbacks() to be invoked
2006
 * later.  The per-cpu rcu_dyntick_drain variable controls the sequencing.
2007 2008
 *
 * The caller must have disabled interrupts.
2009
 */
2010
static void rcu_prepare_for_idle(int cpu)
2011
{
2012 2013 2014
	unsigned long flags;

	local_irq_save(flags);
2015

2016
	/*
2017 2018
	 * If there are no callbacks on this CPU, enter dyntick-idle mode.
	 * Also reset state to avoid prejudicing later attempts.
2019
	 */
2020 2021
	if (!rcu_cpu_has_callbacks(cpu)) {
		per_cpu(rcu_dyntick_holdoff, cpu) = jiffies - 1;
2022
		per_cpu(rcu_dyntick_drain, cpu) = 0;
2023
		local_irq_restore(flags);
2024
		trace_rcu_prep_idle("No callbacks");
2025
		return;
2026
	}
2027 2028 2029 2030 2031

	/*
	 * If in holdoff mode, just return.  We will presumably have
	 * refrained from disabling the scheduling-clock tick.
	 */
2032
	if (per_cpu(rcu_dyntick_holdoff, cpu) == jiffies) {
2033
		local_irq_restore(flags);
2034
		trace_rcu_prep_idle("In holdoff");
2035
		return;
2036
	}
2037 2038 2039 2040 2041 2042 2043

	/* Check and update the rcu_dyntick_drain sequencing. */
	if (per_cpu(rcu_dyntick_drain, cpu) <= 0) {
		/* First time through, initialize the counter. */
		per_cpu(rcu_dyntick_drain, cpu) = RCU_NEEDS_CPU_FLUSHES;
	} else if (--per_cpu(rcu_dyntick_drain, cpu) <= 0) {
		/* We have hit the limit, so time to give up. */
2044
		per_cpu(rcu_dyntick_holdoff, cpu) = jiffies;
2045
		local_irq_restore(flags);
2046
		trace_rcu_prep_idle("Begin holdoff");
2047 2048
		invoke_rcu_core();  /* Force the CPU out of dyntick-idle. */
		return;
2049 2050
	}

2051 2052 2053 2054 2055 2056
	/*
	 * Do one step of pushing the remaining RCU callbacks through
	 * the RCU core state machine.
	 */
#ifdef CONFIG_TREE_PREEMPT_RCU
	if (per_cpu(rcu_preempt_data, cpu).nxtlist) {
2057
		local_irq_restore(flags);
2058 2059
		rcu_preempt_qs(cpu);
		force_quiescent_state(&rcu_preempt_state, 0);
2060
		local_irq_save(flags);
2061 2062
	}
#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
2063
	if (per_cpu(rcu_sched_data, cpu).nxtlist) {
2064
		local_irq_restore(flags);
2065 2066
		rcu_sched_qs(cpu);
		force_quiescent_state(&rcu_sched_state, 0);
2067
		local_irq_save(flags);
2068 2069
	}
	if (per_cpu(rcu_bh_data, cpu).nxtlist) {
2070
		local_irq_restore(flags);
2071 2072
		rcu_bh_qs(cpu);
		force_quiescent_state(&rcu_bh_state, 0);
2073
		local_irq_save(flags);
2074 2075
	}

2076 2077 2078 2079
	/*
	 * If RCU callbacks are still pending, RCU still needs this CPU.
	 * So try forcing the callbacks through the grace period.
	 */
2080
	if (rcu_cpu_has_callbacks(cpu)) {
2081
		local_irq_restore(flags);
2082
		trace_rcu_prep_idle("More callbacks");
2083
		invoke_rcu_core();
2084 2085
	} else {
		local_irq_restore(flags);
2086
		trace_rcu_prep_idle("Callbacks drained");
2087
	}
2088 2089 2090
}

#endif /* #else #if !defined(CONFIG_RCU_FAST_NO_HZ) */