tree_plugin.h 89.5 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
 *
 * 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
17 18
 * along with this program; if not, you can access it online at
 * http://www.gnu.org/licenses/gpl-2.0.html.
19 20 21 22 23 24 25 26
 *
 * 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>
P
Paul E. McKenney 已提交
28
#include <linux/gfp.h>
29
#include <linux/oom.h>
30
#include <linux/smpboot.h>
31
#include "../time/tick-internal.h"
32

33 34 35
#define RCU_KTHREAD_PRIO 1

#ifdef CONFIG_RCU_BOOST
36
#include "../locking/rtmutex_common.h"
37 38 39 40 41
#define RCU_BOOST_PRIO CONFIG_RCU_BOOST_PRIO
#else
#define RCU_BOOST_PRIO RCU_KTHREAD_PRIO
#endif

P
Paul E. McKenney 已提交
42 43 44
#ifdef CONFIG_RCU_NOCB_CPU
static cpumask_var_t rcu_nocb_mask; /* CPUs to have callbacks offloaded. */
static bool have_rcu_nocb_mask;	    /* Was rcu_nocb_mask allocated? */
45
static bool __read_mostly rcu_nocb_poll;    /* Offload kthread are to poll. */
P
Paul E. McKenney 已提交
46 47 48
static char __initdata nocb_buf[NR_CPUS * 5];
#endif /* #ifdef CONFIG_RCU_NOCB_CPU */

49 50 51 52 53 54 55 56
/*
 * 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
57
	pr_info("\tRCU debugfs-based tracing is enabled.\n");
58 59
#endif
#if (defined(CONFIG_64BIT) && CONFIG_RCU_FANOUT != 64) || (!defined(CONFIG_64BIT) && CONFIG_RCU_FANOUT != 32)
60
	pr_info("\tCONFIG_RCU_FANOUT set to non-default value of %d\n",
61 62 63
	       CONFIG_RCU_FANOUT);
#endif
#ifdef CONFIG_RCU_FANOUT_EXACT
64
	pr_info("\tHierarchical RCU autobalancing is disabled.\n");
65 66
#endif
#ifdef CONFIG_RCU_FAST_NO_HZ
67
	pr_info("\tRCU dyntick-idle grace-period acceleration is enabled.\n");
68 69
#endif
#ifdef CONFIG_PROVE_RCU
70
	pr_info("\tRCU lockdep checking is enabled.\n");
71 72
#endif
#ifdef CONFIG_RCU_TORTURE_TEST_RUNNABLE
73
	pr_info("\tRCU torture testing starts during boot.\n");
74
#endif
75
#if defined(CONFIG_TREE_PREEMPT_RCU) && !defined(CONFIG_RCU_CPU_STALL_VERBOSE)
76
	pr_info("\tDump stacks of tasks blocking RCU-preempt GP.\n");
77 78
#endif
#if defined(CONFIG_RCU_CPU_STALL_INFO)
79
	pr_info("\tAdditional per-CPU info printed with stalls.\n");
80 81
#endif
#if NUM_RCU_LVL_4 != 0
82
	pr_info("\tFour-level hierarchy is enabled.\n");
83
#endif
84
	if (rcu_fanout_leaf != CONFIG_RCU_FANOUT_LEAF)
85
		pr_info("\tBoot-time adjustment of leaf fanout to %d.\n", rcu_fanout_leaf);
86
	if (nr_cpu_ids != NR_CPUS)
87
		pr_info("\tRCU restricting CPUs from NR_CPUS=%d to nr_cpu_ids=%d.\n", NR_CPUS, nr_cpu_ids);
88 89
}

90 91
#ifdef CONFIG_TREE_PREEMPT_RCU

92
RCU_STATE_INITIALIZER(rcu_preempt, 'p', call_rcu);
93
static struct rcu_state *rcu_state_p = &rcu_preempt_state;
94

95 96
static int rcu_preempted_readers_exp(struct rcu_node *rnp);

97 98 99
/*
 * Tell them what RCU they are running.
 */
100
static void __init rcu_bootup_announce(void)
101
{
102
	pr_info("Preemptible hierarchical RCU implementation.\n");
103
	rcu_bootup_announce_oddness();
104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125
}

/*
 * 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);

/*
P
Paul E. McKenney 已提交
126
 * Record a preemptible-RCU quiescent state for the specified CPU.  Note
127 128 129
 * 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.
130 131 132 133
 *
 * Unlike the other rcu_*_qs() functions, callers to this function
 * must disable irqs in order to protect the assignment to
 * ->rcu_read_unlock_special.
134
 */
135
static void rcu_preempt_qs(int cpu)
136 137
{
	struct rcu_data *rdp = &per_cpu(rcu_preempt_data, cpu);
138

139
	if (rdp->passed_quiesce == 0)
140
		trace_rcu_grace_period(TPS("rcu_preempt"), rdp->gpnum, TPS("cpuqs"));
141
	rdp->passed_quiesce = 1;
142
	current->rcu_read_unlock_special &= ~RCU_READ_UNLOCK_NEED_QS;
143 144 145
}

/*
146 147 148
 * 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
149 150 151 152 153 154
 * 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.
155 156
 *
 * Caller must disable preemption.
157
 */
158
static void rcu_preempt_note_context_switch(int cpu)
159 160
{
	struct task_struct *t = current;
161
	unsigned long flags;
162 163 164
	struct rcu_data *rdp;
	struct rcu_node *rnp;

165
	if (t->rcu_read_lock_nesting > 0 &&
166 167 168
	    (t->rcu_read_unlock_special & RCU_READ_UNLOCK_BLOCKED) == 0) {

		/* Possibly blocking in an RCU read-side critical section. */
169
		rdp = per_cpu_ptr(rcu_preempt_state.rda, cpu);
170
		rnp = rdp->mynode;
P
Paul E. McKenney 已提交
171
		raw_spin_lock_irqsave(&rnp->lock, flags);
172
		smp_mb__after_unlock_lock();
173
		t->rcu_read_unlock_special |= RCU_READ_UNLOCK_BLOCKED;
174
		t->rcu_blocked_node = rnp;
175 176 177 178 179 180 181 182 183

		/*
		 * 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
184 185 186 187 188 189
		 * 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!!!
190 191 192
		 *
		 * But first, note that the current CPU must still be
		 * on line!
193
		 */
194
		WARN_ON_ONCE((rdp->grpmask & rnp->qsmaskinit) == 0);
195
		WARN_ON_ONCE(!list_empty(&t->rcu_node_entry));
196 197 198
		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;
199 200 201 202
#ifdef CONFIG_RCU_BOOST
			if (rnp->boost_tasks != NULL)
				rnp->boost_tasks = rnp->gp_tasks;
#endif /* #ifdef CONFIG_RCU_BOOST */
203 204 205 206 207
		} else {
			list_add(&t->rcu_node_entry, &rnp->blkd_tasks);
			if (rnp->qsmask & rdp->grpmask)
				rnp->gp_tasks = &t->rcu_node_entry;
		}
208 209 210 211 212
		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
213
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
214 215 216 217 218 219 220 221
	} 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);
222 223 224 225 226 227 228 229 230 231 232
	}

	/*
	 * 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.
	 */
233
	local_irq_save(flags);
234
	rcu_preempt_qs(cpu);
235
	local_irq_restore(flags);
236 237
}

238 239 240 241 242
/*
 * 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.
 */
243
static int rcu_preempt_blocked_readers_cgp(struct rcu_node *rnp)
244
{
245
	return rnp->gp_tasks != NULL;
246 247
}

248 249 250 251 252 253 254
/*
 * 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
255
static void rcu_report_unblock_qs_rnp(struct rcu_node *rnp, unsigned long flags)
256 257 258 259 260
	__releases(rnp->lock)
{
	unsigned long mask;
	struct rcu_node *rnp_p;

261
	if (rnp->qsmask != 0 || rcu_preempt_blocked_readers_cgp(rnp)) {
P
Paul E. McKenney 已提交
262
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
263 264 265 266 267 268 269 270 271 272
		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
273
		rcu_report_qs_rsp(&rcu_preempt_state, flags);
274 275 276 277 278
		return;
	}

	/* Report up the rest of the hierarchy. */
	mask = rnp->grpmask;
P
Paul E. McKenney 已提交
279 280
	raw_spin_unlock(&rnp->lock);	/* irqs remain disabled. */
	raw_spin_lock(&rnp_p->lock);	/* irqs already disabled. */
281
	smp_mb__after_unlock_lock();
P
Paul E. McKenney 已提交
282
	rcu_report_qs_rnp(mask, &rcu_preempt_state, rnp_p, flags);
283 284
}

285 286 287 288 289 290 291 292 293 294 295 296 297 298 299
/*
 * 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;
}

300 301 302 303 304
/*
 * 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.
 */
305
void rcu_read_unlock_special(struct task_struct *t)
306 307
{
	int empty;
308
	int empty_exp;
309
	int empty_exp_now;
310
	unsigned long flags;
311
	struct list_head *np;
312
#ifdef CONFIG_RCU_BOOST
313
	bool drop_boost_mutex = false;
314
#endif /* #ifdef CONFIG_RCU_BOOST */
315 316 317 318 319 320 321 322 323 324 325 326 327 328 329
	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) {
330
		rcu_preempt_qs(smp_processor_id());
331 332 333 334
		if (!t->rcu_read_unlock_special) {
			local_irq_restore(flags);
			return;
		}
335 336
	}

337 338
	/* Hardware IRQ handlers cannot block, complain if they get here. */
	if (WARN_ON_ONCE(in_irq() || in_serving_softirq())) {
339 340 341 342 343 344 345 346
		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;

347 348 349 350 351 352
		/*
		 * 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 (;;) {
353
			rnp = t->rcu_blocked_node;
P
Paul E. McKenney 已提交
354
			raw_spin_lock(&rnp->lock);  /* irqs already disabled. */
355
			smp_mb__after_unlock_lock();
356
			if (rnp == t->rcu_blocked_node)
357
				break;
P
Paul E. McKenney 已提交
358
			raw_spin_unlock(&rnp->lock); /* irqs remain disabled. */
359
		}
360
		empty = !rcu_preempt_blocked_readers_cgp(rnp);
361 362
		empty_exp = !rcu_preempted_readers_exp(rnp);
		smp_mb(); /* ensure expedited fastpath sees end of RCU c-s. */
363
		np = rcu_next_node_entry(t, rnp);
364
		list_del_init(&t->rcu_node_entry);
365
		t->rcu_blocked_node = NULL;
366
		trace_rcu_unlock_preempted_task(TPS("rcu_preempt"),
367
						rnp->gpnum, t->pid);
368 369 370 371
		if (&t->rcu_node_entry == rnp->gp_tasks)
			rnp->gp_tasks = np;
		if (&t->rcu_node_entry == rnp->exp_tasks)
			rnp->exp_tasks = np;
372 373 374
#ifdef CONFIG_RCU_BOOST
		if (&t->rcu_node_entry == rnp->boost_tasks)
			rnp->boost_tasks = np;
375 376
		/* Snapshot ->boost_mtx ownership with rcu_node lock held. */
		drop_boost_mutex = rt_mutex_owner(&rnp->boost_mtx) == t;
377
#endif /* #ifdef CONFIG_RCU_BOOST */
378 379 380 381

		/*
		 * If this was the last task on the current list, and if
		 * we aren't waiting on any CPUs, report the quiescent state.
382 383
		 * Note that rcu_report_unblock_qs_rnp() releases rnp->lock,
		 * so we must take a snapshot of the expedited state.
384
		 */
385
		empty_exp_now = !rcu_preempted_readers_exp(rnp);
386
		if (!empty && !rcu_preempt_blocked_readers_cgp(rnp)) {
387
			trace_rcu_quiescent_state_report(TPS("preempt_rcu"),
388 389 390 391 392 393
							 rnp->gpnum,
							 0, rnp->qsmask,
							 rnp->level,
							 rnp->grplo,
							 rnp->grphi,
							 !!rnp->gp_tasks);
P
Paul E. McKenney 已提交
394
			rcu_report_unblock_qs_rnp(rnp, flags);
395
		} else {
396
			raw_spin_unlock_irqrestore(&rnp->lock, flags);
397
		}
398

399 400
#ifdef CONFIG_RCU_BOOST
		/* Unboost if we were boosted. */
401 402
		if (drop_boost_mutex) {
			rt_mutex_unlock(&rnp->boost_mtx);
403 404
			complete(&rnp->boost_completion);
		}
405 406
#endif /* #ifdef CONFIG_RCU_BOOST */

407 408 409 410
		/*
		 * If this was the last task on the expedited lists,
		 * then we need to report up the rcu_node hierarchy.
		 */
411
		if (!empty_exp && empty_exp_now)
412
			rcu_report_exp_rnp(&rcu_preempt_state, rnp, true);
413 414
	} else {
		local_irq_restore(flags);
415 416 417
	}
}

418 419 420 421 422 423 424 425 426 427 428
#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;

429
	raw_spin_lock_irqsave(&rnp->lock, flags);
430 431 432 433
	if (!rcu_preempt_blocked_readers_cgp(rnp)) {
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
		return;
	}
434 435 436 437 438
	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);
439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461
}

/*
 * 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 */

462 463 464 465
#ifdef CONFIG_RCU_CPU_STALL_INFO

static void rcu_print_task_stall_begin(struct rcu_node *rnp)
{
466
	pr_err("\tTasks blocked on level-%d rcu_node (CPUs %d-%d):",
467 468 469 470 471
	       rnp->level, rnp->grplo, rnp->grphi);
}

static void rcu_print_task_stall_end(void)
{
472
	pr_cont("\n");
473 474 475 476 477 478 479 480 481 482 483 484 485 486
}

#else /* #ifdef CONFIG_RCU_CPU_STALL_INFO */

static void rcu_print_task_stall_begin(struct rcu_node *rnp)
{
}

static void rcu_print_task_stall_end(void)
{
}

#endif /* #else #ifdef CONFIG_RCU_CPU_STALL_INFO */

487 488 489 490
/*
 * Scan the current list of tasks blocked within RCU read-side critical
 * sections, printing out the tid of each.
 */
491
static int rcu_print_task_stall(struct rcu_node *rnp)
492 493
{
	struct task_struct *t;
494
	int ndetected = 0;
495

496
	if (!rcu_preempt_blocked_readers_cgp(rnp))
497
		return 0;
498
	rcu_print_task_stall_begin(rnp);
499 500
	t = list_entry(rnp->gp_tasks,
		       struct task_struct, rcu_node_entry);
501
	list_for_each_entry_continue(t, &rnp->blkd_tasks, rcu_node_entry) {
502
		pr_cont(" P%d", t->pid);
503 504
		ndetected++;
	}
505
	rcu_print_task_stall_end();
506
	return ndetected;
507 508
}

509 510 511 512 513 514
/*
 * 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.
515 516 517
 *
 * Also, if there are blocked tasks on the list, they automatically
 * block the newly created grace period, so set up ->gp_tasks accordingly.
518 519 520
 */
static void rcu_preempt_check_blocked_tasks(struct rcu_node *rnp)
{
521
	WARN_ON_ONCE(rcu_preempt_blocked_readers_cgp(rnp));
522 523
	if (!list_empty(&rnp->blkd_tasks))
		rnp->gp_tasks = rnp->blkd_tasks.next;
524
	WARN_ON_ONCE(rnp->qsmask);
525 526
}

527 528
#ifdef CONFIG_HOTPLUG_CPU

529 530 531 532 533 534
/*
 * 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.
535 536
 * Returns true if there were tasks blocking the current RCU grace
 * period.
537
 *
538 539 540
 * Returns 1 if there was previously a task blocking the current grace
 * period on the specified rcu_node structure.
 *
541 542
 * The caller must hold rnp->lock with irqs disabled.
 */
543 544 545
static int rcu_preempt_offline_tasks(struct rcu_state *rsp,
				     struct rcu_node *rnp,
				     struct rcu_data *rdp)
546 547 548
{
	struct list_head *lp;
	struct list_head *lp_root;
549
	int retval = 0;
550
	struct rcu_node *rnp_root = rcu_get_root(rsp);
551
	struct task_struct *t;
552

553 554
	if (rnp == rnp_root) {
		WARN_ONCE(1, "Last CPU thought to be offlined?");
555
		return 0;  /* Shouldn't happen: at least one CPU online. */
556
	}
557 558 559

	/* If we are on an internal node, complain bitterly. */
	WARN_ON_ONCE(rnp != rdp->mynode);
560 561

	/*
562 563 564 565 566 567 568
	 * 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.
569
	 */
570
	if (rcu_preempt_blocked_readers_cgp(rnp) && rnp->qsmask == 0)
571 572 573
		retval |= RCU_OFL_TASKS_NORM_GP;
	if (rcu_preempted_readers_exp(rnp))
		retval |= RCU_OFL_TASKS_EXP_GP;
574 575 576 577 578
	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 */
579
		smp_mb__after_unlock_lock();
580 581 582 583 584 585 586
		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;
587 588 589 590
#ifdef CONFIG_RCU_BOOST
		if (&t->rcu_node_entry == rnp->boost_tasks)
			rnp_root->boost_tasks = rnp->boost_tasks;
#endif /* #ifdef CONFIG_RCU_BOOST */
591
		raw_spin_unlock(&rnp_root->lock); /* irqs still disabled */
592
	}
593

594 595
	rnp->gp_tasks = NULL;
	rnp->exp_tasks = NULL;
596
#ifdef CONFIG_RCU_BOOST
597
	rnp->boost_tasks = NULL;
598 599 600 601 602
	/*
	 * In case root is being boosted and leaf was not.  Make sure
	 * that we boost the tasks blocking the current grace period
	 * in this case.
	 */
603
	raw_spin_lock(&rnp_root->lock); /* irqs already disabled */
604
	smp_mb__after_unlock_lock();
605
	if (rnp_root->boost_tasks != NULL &&
606 607
	    rnp_root->boost_tasks != rnp_root->gp_tasks &&
	    rnp_root->boost_tasks != rnp_root->exp_tasks)
608 609 610 611
		rnp_root->boost_tasks = rnp_root->gp_tasks;
	raw_spin_unlock(&rnp_root->lock); /* irqs still disabled */
#endif /* #ifdef CONFIG_RCU_BOOST */

612
	return retval;
613 614
}

615 616
#endif /* #ifdef CONFIG_HOTPLUG_CPU */

617 618 619 620 621 622 623 624 625 626 627 628
/*
 * 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) {
629
		rcu_preempt_qs(cpu);
630 631
		return;
	}
632 633
	if (t->rcu_read_lock_nesting > 0 &&
	    per_cpu(rcu_preempt_data, cpu).qs_pending)
634
		t->rcu_read_unlock_special |= RCU_READ_UNLOCK_NEED_QS;
635 636
}

637 638
#ifdef CONFIG_RCU_BOOST

639 640
static void rcu_preempt_do_callbacks(void)
{
641
	rcu_do_batch(&rcu_preempt_state, this_cpu_ptr(&rcu_preempt_data));
642 643
}

644 645
#endif /* #ifdef CONFIG_RCU_BOOST */

646
/*
P
Paul E. McKenney 已提交
647
 * Queue a preemptible-RCU callback for invocation after a grace period.
648 649 650
 */
void call_rcu(struct rcu_head *head, void (*func)(struct rcu_head *rcu))
{
P
Paul E. McKenney 已提交
651
	__call_rcu(head, func, &rcu_preempt_state, -1, 0);
652 653 654
}
EXPORT_SYMBOL_GPL(call_rcu);

655 656 657 658 659
/**
 * 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
660 661 662 663 664
 * 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.
665 666 667
 *
 * See the description of synchronize_sched() for more detailed information
 * on memory ordering guarantees.
668 669 670
 */
void synchronize_rcu(void)
{
671 672 673 674
	rcu_lockdep_assert(!lock_is_held(&rcu_bh_lock_map) &&
			   !lock_is_held(&rcu_lock_map) &&
			   !lock_is_held(&rcu_sched_lock_map),
			   "Illegal synchronize_rcu() in RCU read-side critical section");
675 676
	if (!rcu_scheduler_active)
		return;
677 678 679 680
	if (rcu_expedited)
		synchronize_rcu_expedited();
	else
		wait_rcu_gp(call_rcu);
681 682 683
}
EXPORT_SYMBOL_GPL(synchronize_rcu);

684
static DECLARE_WAIT_QUEUE_HEAD(sync_rcu_preempt_exp_wq);
685
static unsigned long sync_rcu_preempt_exp_count;
686 687 688 689 690 691 692 693 694 695
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)
{
696
	return rnp->exp_tasks != NULL;
697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721
}

/*
 * 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!)
 *
722 723 724
 * Most callers will set the "wake" flag, but the task initiating the
 * expedited grace period need not wake itself.
 *
725 726
 * Caller must hold sync_rcu_preempt_exp_mutex.
 */
727 728
static void rcu_report_exp_rnp(struct rcu_state *rsp, struct rcu_node *rnp,
			       bool wake)
729 730 731 732
{
	unsigned long flags;
	unsigned long mask;

P
Paul E. McKenney 已提交
733
	raw_spin_lock_irqsave(&rnp->lock, flags);
734
	smp_mb__after_unlock_lock();
735
	for (;;) {
736 737
		if (!sync_rcu_preempt_exp_done(rnp)) {
			raw_spin_unlock_irqrestore(&rnp->lock, flags);
738
			break;
739
		}
740
		if (rnp->parent == NULL) {
741
			raw_spin_unlock_irqrestore(&rnp->lock, flags);
742 743
			if (wake) {
				smp_mb(); /* EGP done before wake_up(). */
744
				wake_up(&sync_rcu_preempt_exp_wq);
745
			}
746 747 748
			break;
		}
		mask = rnp->grpmask;
P
Paul E. McKenney 已提交
749
		raw_spin_unlock(&rnp->lock); /* irqs remain disabled */
750
		rnp = rnp->parent;
P
Paul E. McKenney 已提交
751
		raw_spin_lock(&rnp->lock); /* irqs already disabled */
752
		smp_mb__after_unlock_lock();
753 754 755 756 757 758 759 760 761
		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.
 *
762 763
 * Caller must hold sync_rcu_preempt_exp_mutex and must exclude
 * CPU hotplug operations.
764 765 766 767
 */
static void
sync_rcu_preempt_exp_init(struct rcu_state *rsp, struct rcu_node *rnp)
{
768
	unsigned long flags;
769
	int must_wait = 0;
770

771
	raw_spin_lock_irqsave(&rnp->lock, flags);
772
	smp_mb__after_unlock_lock();
773
	if (list_empty(&rnp->blkd_tasks)) {
774
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
775
	} else {
776
		rnp->exp_tasks = rnp->blkd_tasks.next;
777
		rcu_initiate_boost(rnp, flags);  /* releases rnp->lock */
778 779
		must_wait = 1;
	}
780
	if (!must_wait)
781
		rcu_report_exp_rnp(rsp, rnp, false); /* Don't wake self. */
782 783
}

784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799
/**
 * synchronize_rcu_expedited - Brute-force RCU grace period
 *
 * 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
 * the ->blkd_tasks lists and wait for this list to drain.  This consumes
 * significant time on all CPUs and is unfriendly to real-time workloads,
 * so is thus not recommended for any sort of common-case code.
 * In fact, if you are using synchronize_rcu_expedited() in a loop,
 * please restructure your code to batch your updates, and then Use a
 * single synchronize_rcu() instead.
 *
 * Note that it is illegal to call this function while holding any lock
 * that is acquired by a CPU-hotplug notifier.  And yes, it is also illegal
 * to call this function from a CPU-hotplug notifier.  Failing to observe
 * these restriction will result in deadlock.
800 801 802
 */
void synchronize_rcu_expedited(void)
{
803 804 805
	unsigned long flags;
	struct rcu_node *rnp;
	struct rcu_state *rsp = &rcu_preempt_state;
806
	unsigned long snap;
807 808 809 810 811 812
	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. */

813 814 815 816 817 818 819 820 821 822
	/*
	 * Block CPU-hotplug operations.  This means that any CPU-hotplug
	 * operation that finds an rcu_node structure with tasks in the
	 * process of being boosted will know that all tasks blocking
	 * this expedited grace period will already be in the process of
	 * being boosted.  This simplifies the process of moving tasks
	 * from leaf to root rcu_node structures.
	 */
	get_online_cpus();

823 824 825 826 827 828
	/*
	 * 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)) {
829 830 831 832 833
		if (ULONG_CMP_LT(snap,
		    ACCESS_ONCE(sync_rcu_preempt_exp_count))) {
			put_online_cpus();
			goto mb_ret; /* Others did our work for us. */
		}
834
		if (trycount++ < 10) {
835
			udelay(trycount * num_online_cpus());
836
		} else {
837
			put_online_cpus();
838
			wait_rcu_gp(call_rcu);
839 840 841
			return;
		}
	}
842 843
	if (ULONG_CMP_LT(snap, ACCESS_ONCE(sync_rcu_preempt_exp_count))) {
		put_online_cpus();
844
		goto unlock_mb_ret; /* Others did our work for us. */
845
	}
846

847
	/* force all RCU readers onto ->blkd_tasks lists. */
848 849 850 851
	synchronize_sched_expedited();

	/* Initialize ->expmask for all non-leaf rcu_node structures. */
	rcu_for_each_nonleaf_node_breadth_first(rsp, rnp) {
852
		raw_spin_lock_irqsave(&rnp->lock, flags);
853
		smp_mb__after_unlock_lock();
854
		rnp->expmask = rnp->qsmaskinit;
855
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
856 857
	}

858
	/* Snapshot current state of ->blkd_tasks lists. */
859 860 861 862 863
	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));

864
	put_online_cpus();
865

866
	/* Wait for snapshotted ->blkd_tasks lists to drain. */
867 868 869 870 871 872 873 874 875 876 877
	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. */
878 879 880
}
EXPORT_SYMBOL_GPL(synchronize_rcu_expedited);

881 882
/**
 * rcu_barrier - Wait until all in-flight call_rcu() callbacks complete.
883 884 885 886 887
 *
 * Note that this primitive does not necessarily wait for an RCU grace period
 * to complete.  For example, if there are no RCU callbacks queued anywhere
 * in the system, then rcu_barrier() is within its rights to return
 * immediately, without waiting for anything, much less an RCU grace period.
888 889 890
 */
void rcu_barrier(void)
{
891
	_rcu_barrier(&rcu_preempt_state);
892 893 894
}
EXPORT_SYMBOL_GPL(rcu_barrier);

895
/*
P
Paul E. McKenney 已提交
896
 * Initialize preemptible RCU's state structures.
897 898 899
 */
static void __init __rcu_init_preempt(void)
{
900
	rcu_init_one(&rcu_preempt_state, &rcu_preempt_data);
901 902
}

903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920
/*
 * Check for a task exiting while in a preemptible-RCU read-side
 * 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 (likely(list_empty(&current->rcu_node_entry)))
		return;
	t->rcu_read_lock_nesting = 1;
	barrier();
	t->rcu_read_unlock_special = RCU_READ_UNLOCK_BLOCKED;
	__rcu_read_unlock();
}

921 922
#else /* #ifdef CONFIG_TREE_PREEMPT_RCU */

923
static struct rcu_state *rcu_state_p = &rcu_sched_state;
924

925 926 927
/*
 * Tell them what RCU they are running.
 */
928
static void __init rcu_bootup_announce(void)
929
{
930
	pr_info("Hierarchical RCU implementation.\n");
931
	rcu_bootup_announce_oddness();
932 933 934 935 936 937 938 939 940 941 942
}

/*
 * 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);

943 944 945 946 947 948 949 950
/*
 * Because preemptible RCU does not exist, we never have to check for
 * CPUs being in quiescent states.
 */
static void rcu_preempt_note_context_switch(int cpu)
{
}

951
/*
P
Paul E. McKenney 已提交
952
 * Because preemptible RCU does not exist, there are never any preempted
953 954
 * RCU readers.
 */
955
static int rcu_preempt_blocked_readers_cgp(struct rcu_node *rnp)
956 957 958 959
{
	return 0;
}

960 961 962
#ifdef CONFIG_HOTPLUG_CPU

/* Because preemptible RCU does not exist, no quieting of tasks. */
P
Paul E. McKenney 已提交
963
static void rcu_report_unblock_qs_rnp(struct rcu_node *rnp, unsigned long flags)
964
	__releases(rnp->lock)
965
{
P
Paul E. McKenney 已提交
966
	raw_spin_unlock_irqrestore(&rnp->lock, flags);
967 968 969 970
}

#endif /* #ifdef CONFIG_HOTPLUG_CPU */

971
/*
P
Paul E. McKenney 已提交
972
 * Because preemptible RCU does not exist, we never have to check for
973 974 975 976 977 978
 * tasks blocked within RCU read-side critical sections.
 */
static void rcu_print_detail_task_stall(struct rcu_state *rsp)
{
}

979
/*
P
Paul E. McKenney 已提交
980
 * Because preemptible RCU does not exist, we never have to check for
981 982
 * tasks blocked within RCU read-side critical sections.
 */
983
static int rcu_print_task_stall(struct rcu_node *rnp)
984
{
985
	return 0;
986 987
}

988
/*
P
Paul E. McKenney 已提交
989
 * Because there is no preemptible RCU, there can be no readers blocked,
990 991
 * so there is no need to check for blocked tasks.  So check only for
 * bogus qsmask values.
992 993 994
 */
static void rcu_preempt_check_blocked_tasks(struct rcu_node *rnp)
{
995
	WARN_ON_ONCE(rnp->qsmask);
996 997
}

998 999
#ifdef CONFIG_HOTPLUG_CPU

1000
/*
P
Paul E. McKenney 已提交
1001
 * Because preemptible RCU does not exist, it never needs to migrate
1002 1003 1004
 * tasks that were blocked within RCU read-side critical sections, and
 * such non-existent tasks cannot possibly have been blocking the current
 * grace period.
1005
 */
1006 1007 1008
static int rcu_preempt_offline_tasks(struct rcu_state *rsp,
				     struct rcu_node *rnp,
				     struct rcu_data *rdp)
1009
{
1010
	return 0;
1011 1012
}

1013 1014
#endif /* #ifdef CONFIG_HOTPLUG_CPU */

1015
/*
P
Paul E. McKenney 已提交
1016
 * Because preemptible RCU does not exist, it never has any callbacks
1017 1018
 * to check.
 */
1019
static void rcu_preempt_check_callbacks(int cpu)
1020 1021 1022
{
}

1023 1024
/*
 * Wait for an rcu-preempt grace period, but make it happen quickly.
P
Paul E. McKenney 已提交
1025
 * But because preemptible RCU does not exist, map to rcu-sched.
1026 1027 1028 1029 1030 1031 1032
 */
void synchronize_rcu_expedited(void)
{
	synchronize_sched_expedited();
}
EXPORT_SYMBOL_GPL(synchronize_rcu_expedited);

1033 1034 1035
#ifdef CONFIG_HOTPLUG_CPU

/*
P
Paul E. McKenney 已提交
1036
 * Because preemptible RCU does not exist, there is never any need to
1037 1038 1039
 * report on tasks preempted in RCU read-side critical sections during
 * expedited RCU grace periods.
 */
1040 1041
static void rcu_report_exp_rnp(struct rcu_state *rsp, struct rcu_node *rnp,
			       bool wake)
1042 1043 1044 1045 1046
{
}

#endif /* #ifdef CONFIG_HOTPLUG_CPU */

1047
/*
P
Paul E. McKenney 已提交
1048
 * Because preemptible RCU does not exist, rcu_barrier() is just
1049 1050 1051 1052 1053 1054 1055 1056
 * another name for rcu_barrier_sched().
 */
void rcu_barrier(void)
{
	rcu_barrier_sched();
}
EXPORT_SYMBOL_GPL(rcu_barrier);

1057
/*
P
Paul E. McKenney 已提交
1058
 * Because preemptible RCU does not exist, it need not be initialized.
1059 1060 1061 1062 1063
 */
static void __init __rcu_init_preempt(void)
{
}

1064 1065 1066 1067 1068 1069 1070 1071
/*
 * Because preemptible RCU does not exist, tasks cannot possibly exit
 * while in preemptible RCU read-side critical sections.
 */
void exit_rcu(void)
{
}

1072
#endif /* #else #ifdef CONFIG_TREE_PREEMPT_RCU */
1073

1074 1075
#ifdef CONFIG_RCU_BOOST

1076
#include "../locking/rtmutex_common.h"
1077

1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090
#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 &&
1091
		 ULONG_CMP_LT(jiffies, rnp->boost_time))
1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104
		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 */

T
Thomas Gleixner 已提交
1105 1106 1107 1108 1109 1110 1111 1112 1113 1114
static void rcu_wake_cond(struct task_struct *t, int status)
{
	/*
	 * If the thread is yielding, only wake it when this
	 * is invoked from idle
	 */
	if (status != RCU_KTHREAD_YIELDING || is_idle_task(current))
		wake_up_process(t);
}

1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132
/*
 * 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 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);
1133
	smp_mb__after_unlock_lock();
1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149

	/*
	 * 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.
	 */
1150
	if (rnp->exp_tasks != NULL) {
1151
		tb = rnp->exp_tasks;
1152 1153
		rnp->n_exp_boosts++;
	} else {
1154
		tb = rnp->boost_tasks;
1155 1156 1157
		rnp->n_normal_boosts++;
	}
	rnp->n_tasks_boosted++;
1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175

	/*
	 * 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);
1176
	rt_mutex_init_proxy_locked(&rnp->boost_mtx, t);
1177
	init_completion(&rnp->boost_completion);
1178
	raw_spin_unlock_irqrestore(&rnp->lock, flags);
1179 1180 1181
	/* Lock only for side effect: boosts task t's priority. */
	rt_mutex_lock(&rnp->boost_mtx);
	rt_mutex_unlock(&rnp->boost_mtx);  /* Then keep lockdep happy. */
1182

1183
	/* Wait for boostee to be done w/boost_mtx before reinitializing. */
1184
	wait_for_completion(&rnp->boost_completion);
1185

1186 1187
	return ACCESS_ONCE(rnp->exp_tasks) != NULL ||
	       ACCESS_ONCE(rnp->boost_tasks) != NULL;
1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199
}

/*
 * 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;

1200
	trace_rcu_utilization(TPS("Start boost kthread@init"));
1201
	for (;;) {
1202
		rnp->boost_kthread_status = RCU_KTHREAD_WAITING;
1203
		trace_rcu_utilization(TPS("End boost kthread@rcu_wait"));
1204
		rcu_wait(rnp->boost_tasks || rnp->exp_tasks);
1205
		trace_rcu_utilization(TPS("Start boost kthread@rcu_wait"));
1206
		rnp->boost_kthread_status = RCU_KTHREAD_RUNNING;
1207 1208 1209 1210 1211 1212
		more2boost = rcu_boost(rnp);
		if (more2boost)
			spincnt++;
		else
			spincnt = 0;
		if (spincnt > 10) {
T
Thomas Gleixner 已提交
1213
			rnp->boost_kthread_status = RCU_KTHREAD_YIELDING;
1214
			trace_rcu_utilization(TPS("End boost kthread@rcu_yield"));
T
Thomas Gleixner 已提交
1215
			schedule_timeout_interruptible(2);
1216
			trace_rcu_utilization(TPS("Start boost kthread@rcu_yield"));
1217 1218 1219
			spincnt = 0;
		}
	}
1220
	/* NOTREACHED */
1221
	trace_rcu_utilization(TPS("End boost kthread@notreached"));
1222 1223 1224 1225 1226 1227 1228 1229 1230
	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.
 *
1231 1232 1233
 * The caller must hold rnp->lock, which this function releases.
 * The ->boost_kthread_task is immortal, so we don't need to worry
 * about it going away.
1234
 */
1235
static void rcu_initiate_boost(struct rcu_node *rnp, unsigned long flags)
1236
	__releases(rnp->lock)
1237 1238 1239
{
	struct task_struct *t;

1240 1241
	if (!rcu_preempt_blocked_readers_cgp(rnp) && rnp->exp_tasks == NULL) {
		rnp->n_balk_exp_gp_tasks++;
1242
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
1243
		return;
1244
	}
1245 1246 1247 1248 1249 1250 1251
	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;
1252
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
1253
		t = rnp->boost_kthread_task;
T
Thomas Gleixner 已提交
1254 1255
		if (t)
			rcu_wake_cond(t, rnp->boost_kthread_status);
1256
	} else {
1257
		rcu_initiate_boost_trace(rnp);
1258 1259
		raw_spin_unlock_irqrestore(&rnp->lock, flags);
	}
1260 1261
}

1262 1263 1264 1265 1266 1267 1268 1269 1270
/*
 * 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);
1271
	if (__this_cpu_read(rcu_cpu_kthread_task) != NULL &&
T
Thomas Gleixner 已提交
1272 1273 1274 1275
	    current != __this_cpu_read(rcu_cpu_kthread_task)) {
		rcu_wake_cond(__this_cpu_read(rcu_cpu_kthread_task),
			      __this_cpu_read(rcu_cpu_kthread_status));
	}
1276 1277 1278
	local_irq_restore(flags);
}

1279 1280 1281 1282 1283 1284
/*
 * Is the current CPU running the RCU-callbacks kthread?
 * Caller must have preemption disabled.
 */
static bool rcu_is_callbacks_kthread(void)
{
1285
	return __this_cpu_read(rcu_cpu_kthread_task) == current;
1286 1287
}

1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302
#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.
 */
1303
static int rcu_spawn_one_boost_kthread(struct rcu_state *rsp,
T
Thomas Gleixner 已提交
1304
						 struct rcu_node *rnp)
1305
{
T
Thomas Gleixner 已提交
1306
	int rnp_index = rnp - &rsp->node[0];
1307 1308 1309 1310 1311 1312
	unsigned long flags;
	struct sched_param sp;
	struct task_struct *t;

	if (&rcu_preempt_state != rsp)
		return 0;
T
Thomas Gleixner 已提交
1313 1314 1315 1316

	if (!rcu_scheduler_fully_active || rnp->qsmaskinit == 0)
		return 0;

1317
	rsp->boost = 1;
1318 1319 1320
	if (rnp->boost_kthread_task != NULL)
		return 0;
	t = kthread_create(rcu_boost_kthread, (void *)rnp,
1321
			   "rcub/%d", rnp_index);
1322 1323 1324
	if (IS_ERR(t))
		return PTR_ERR(t);
	raw_spin_lock_irqsave(&rnp->lock, flags);
1325
	smp_mb__after_unlock_lock();
1326 1327
	rnp->boost_kthread_task = t;
	raw_spin_unlock_irqrestore(&rnp->lock, flags);
1328
	sp.sched_priority = RCU_BOOST_PRIO;
1329
	sched_setscheduler_nocheck(t, SCHED_FIFO, &sp);
1330
	wake_up_process(t); /* get to TASK_INTERRUPTIBLE quickly. */
1331 1332 1333
	return 0;
}

1334 1335
static void rcu_kthread_do_work(void)
{
1336 1337
	rcu_do_batch(&rcu_sched_state, this_cpu_ptr(&rcu_sched_data));
	rcu_do_batch(&rcu_bh_state, this_cpu_ptr(&rcu_bh_data));
1338 1339 1340
	rcu_preempt_do_callbacks();
}

1341
static void rcu_cpu_kthread_setup(unsigned int cpu)
1342 1343 1344
{
	struct sched_param sp;

1345 1346
	sp.sched_priority = RCU_KTHREAD_PRIO;
	sched_setscheduler_nocheck(current, SCHED_FIFO, &sp);
1347 1348
}

1349
static void rcu_cpu_kthread_park(unsigned int cpu)
1350
{
1351
	per_cpu(rcu_cpu_kthread_status, cpu) = RCU_KTHREAD_OFFCPU;
1352 1353
}

1354
static int rcu_cpu_kthread_should_run(unsigned int cpu)
1355
{
1356
	return __this_cpu_read(rcu_cpu_has_work);
1357 1358 1359 1360
}

/*
 * Per-CPU kernel thread that invokes RCU callbacks.  This replaces the
1361 1362
 * RCU softirq used in flavors and configurations of RCU that do not
 * support RCU priority boosting.
1363
 */
1364
static void rcu_cpu_kthread(unsigned int cpu)
1365
{
1366 1367
	unsigned int *statusp = this_cpu_ptr(&rcu_cpu_kthread_status);
	char work, *workp = this_cpu_ptr(&rcu_cpu_has_work);
1368
	int spincnt;
1369

1370
	for (spincnt = 0; spincnt < 10; spincnt++) {
1371
		trace_rcu_utilization(TPS("Start CPU kthread@rcu_wait"));
1372 1373
		local_bh_disable();
		*statusp = RCU_KTHREAD_RUNNING;
1374 1375
		this_cpu_inc(rcu_cpu_kthread_loops);
		local_irq_disable();
1376 1377
		work = *workp;
		*workp = 0;
1378
		local_irq_enable();
1379 1380 1381
		if (work)
			rcu_kthread_do_work();
		local_bh_enable();
1382
		if (*workp == 0) {
1383
			trace_rcu_utilization(TPS("End CPU kthread@rcu_wait"));
1384 1385
			*statusp = RCU_KTHREAD_WAITING;
			return;
1386 1387
		}
	}
1388
	*statusp = RCU_KTHREAD_YIELDING;
1389
	trace_rcu_utilization(TPS("Start CPU kthread@rcu_yield"));
1390
	schedule_timeout_interruptible(2);
1391
	trace_rcu_utilization(TPS("End CPU kthread@rcu_yield"));
1392
	*statusp = RCU_KTHREAD_WAITING;
1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403
}

/*
 * 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.
 */
T
Thomas Gleixner 已提交
1404
static void rcu_boost_kthread_setaffinity(struct rcu_node *rnp, int outgoingcpu)
1405
{
T
Thomas Gleixner 已提交
1406 1407
	struct task_struct *t = rnp->boost_kthread_task;
	unsigned long mask = rnp->qsmaskinit;
1408 1409 1410
	cpumask_var_t cm;
	int cpu;

T
Thomas Gleixner 已提交
1411
	if (!t)
1412
		return;
T
Thomas Gleixner 已提交
1413
	if (!zalloc_cpumask_var(&cm, GFP_KERNEL))
1414 1415 1416 1417 1418 1419 1420 1421 1422 1423
		return;
	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);
	}
T
Thomas Gleixner 已提交
1424
	set_cpus_allowed_ptr(t, cm);
1425 1426 1427
	free_cpumask_var(cm);
}

1428 1429 1430 1431 1432 1433 1434 1435
static struct smp_hotplug_thread rcu_cpu_thread_spec = {
	.store			= &rcu_cpu_kthread_task,
	.thread_should_run	= rcu_cpu_kthread_should_run,
	.thread_fn		= rcu_cpu_kthread,
	.thread_comm		= "rcuc/%u",
	.setup			= rcu_cpu_kthread_setup,
	.park			= rcu_cpu_kthread_park,
};
1436 1437 1438 1439 1440 1441 1442

/*
 * Spawn all kthreads -- called as soon as the scheduler is running.
 */
static int __init rcu_spawn_kthreads(void)
{
	struct rcu_node *rnp;
T
Thomas Gleixner 已提交
1443
	int cpu;
1444

1445
	rcu_scheduler_fully_active = 1;
1446
	for_each_possible_cpu(cpu)
1447
		per_cpu(rcu_cpu_has_work, cpu) = 0;
1448
	BUG_ON(smpboot_register_percpu_thread(&rcu_cpu_thread_spec));
1449 1450
	rnp = rcu_get_root(rcu_state_p);
	(void)rcu_spawn_one_boost_kthread(rcu_state_p, rnp);
1451
	if (NUM_RCU_NODES > 1) {
1452 1453
		rcu_for_each_leaf_node(rcu_state_p, rnp)
			(void)rcu_spawn_one_boost_kthread(rcu_state_p, rnp);
1454 1455 1456 1457 1458
	}
	return 0;
}
early_initcall(rcu_spawn_kthreads);

1459
static void rcu_prepare_kthreads(int cpu)
1460
{
1461
	struct rcu_data *rdp = per_cpu_ptr(rcu_state_p->rda, cpu);
1462 1463 1464
	struct rcu_node *rnp = rdp->mynode;

	/* Fire up the incoming CPU's kthread and leaf rcu_node kthread. */
1465
	if (rcu_scheduler_fully_active)
1466
		(void)rcu_spawn_one_boost_kthread(rcu_state_p, rnp);
1467 1468
}

1469 1470
#else /* #ifdef CONFIG_RCU_BOOST */

1471
static void rcu_initiate_boost(struct rcu_node *rnp, unsigned long flags)
1472
	__releases(rnp->lock)
1473
{
1474
	raw_spin_unlock_irqrestore(&rnp->lock, flags);
1475 1476
}

1477
static void invoke_rcu_callbacks_kthread(void)
1478
{
1479
	WARN_ON_ONCE(1);
1480 1481
}

1482 1483 1484 1485 1486
static bool rcu_is_callbacks_kthread(void)
{
	return false;
}

1487 1488 1489 1490
static void rcu_preempt_boost_start_gp(struct rcu_node *rnp)
{
}

T
Thomas Gleixner 已提交
1491
static void rcu_boost_kthread_setaffinity(struct rcu_node *rnp, int outgoingcpu)
1492 1493 1494
{
}

1495 1496 1497 1498 1499 1500 1501
static int __init rcu_scheduler_really_started(void)
{
	rcu_scheduler_fully_active = 1;
	return 0;
}
early_initcall(rcu_scheduler_really_started);

1502
static void rcu_prepare_kthreads(int cpu)
1503 1504 1505
{
}

1506 1507
#endif /* #else #ifdef CONFIG_RCU_BOOST */

1508 1509 1510 1511 1512 1513 1514 1515
#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.
 *
1516 1517
 * Because we not have RCU_FAST_NO_HZ, just check whether this CPU needs
 * any flavor of RCU.
1518
 */
1519
#ifndef CONFIG_RCU_NOCB_CPU_ALL
1520
int rcu_needs_cpu(int cpu, unsigned long *delta_jiffies)
1521
{
1522
	*delta_jiffies = ULONG_MAX;
1523
	return rcu_cpu_has_callbacks(cpu, NULL);
1524
}
1525
#endif /* #ifndef CONFIG_RCU_NOCB_CPU_ALL */
1526 1527 1528 1529 1530 1531 1532 1533 1534

/*
 * Because we do not have RCU_FAST_NO_HZ, don't bother cleaning up
 * after it.
 */
static void rcu_cleanup_after_idle(int cpu)
{
}

1535
/*
1536
 * Do the idle-entry grace-period work, which, because CONFIG_RCU_FAST_NO_HZ=n,
1537 1538 1539 1540 1541 1542
 * is nothing.
 */
static void rcu_prepare_for_idle(int cpu)
{
}

1543 1544 1545 1546 1547 1548 1549 1550
/*
 * Don't bother keeping a running count of the number of RCU callbacks
 * posted because CONFIG_RCU_FAST_NO_HZ=n.
 */
static void rcu_idle_count_callbacks_posted(void)
{
}

1551 1552
#else /* #if !defined(CONFIG_RCU_FAST_NO_HZ) */

1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567
/*
 * This code is invoked when a CPU goes idle, at which point we want
 * to have the CPU do everything required for RCU so that it can enter
 * the energy-efficient dyntick-idle mode.  This is handled by a
 * state machine implemented by rcu_prepare_for_idle() below.
 *
 * The following three proprocessor symbols control this state machine:
 *
 * RCU_IDLE_GP_DELAY gives the number of jiffies that a CPU is permitted
 *	to sleep in dyntick-idle mode with RCU callbacks pending.  This
 *	is sized to be roughly one RCU grace period.  Those energy-efficiency
 *	benchmarkers who might otherwise be tempted to set this to a large
 *	number, be warned: Setting RCU_IDLE_GP_DELAY too high can hang your
 *	system.  And if you are -that- concerned about energy efficiency,
 *	just power the system down and be done with it!
1568 1569 1570
 * RCU_IDLE_LAZY_GP_DELAY gives the number of jiffies that a CPU is
 *	permitted to sleep in dyntick-idle mode with only lazy RCU
 *	callbacks pending.  Setting this too high can OOM your system.
1571 1572 1573 1574 1575
 *
 * The values below work well in practice.  If future workloads require
 * adjustment, they can be converted into kernel config parameters, though
 * making the state machine smarter might be a better option.
 */
1576
#define RCU_IDLE_GP_DELAY 4		/* Roughly one grace period. */
1577
#define RCU_IDLE_LAZY_GP_DELAY (6 * HZ)	/* Roughly six seconds. */
1578

1579 1580 1581 1582
static int rcu_idle_gp_delay = RCU_IDLE_GP_DELAY;
module_param(rcu_idle_gp_delay, int, 0644);
static int rcu_idle_lazy_gp_delay = RCU_IDLE_LAZY_GP_DELAY;
module_param(rcu_idle_lazy_gp_delay, int, 0644);
1583

1584
extern int tick_nohz_active;
1585 1586

/*
1587 1588 1589
 * Try to advance callbacks for all flavors of RCU on the current CPU, but
 * only if it has been awhile since the last time we did so.  Afterwards,
 * if there are any callbacks ready for immediate invocation, return true.
1590
 */
1591
static bool __maybe_unused rcu_try_advance_all_cbs(void)
1592
{
1593 1594
	bool cbs_ready = false;
	struct rcu_data *rdp;
1595
	struct rcu_dynticks *rdtp = this_cpu_ptr(&rcu_dynticks);
1596 1597
	struct rcu_node *rnp;
	struct rcu_state *rsp;
1598

1599 1600 1601 1602 1603
	/* Exit early if we advanced recently. */
	if (jiffies == rdtp->last_advance_all)
		return 0;
	rdtp->last_advance_all = jiffies;

1604 1605 1606
	for_each_rcu_flavor(rsp) {
		rdp = this_cpu_ptr(rsp->rda);
		rnp = rdp->mynode;
1607

1608 1609 1610 1611 1612 1613 1614
		/*
		 * Don't bother checking unless a grace period has
		 * completed since we last checked and there are
		 * callbacks not yet ready to invoke.
		 */
		if (rdp->completed != rnp->completed &&
		    rdp->nxttail[RCU_DONE_TAIL] != rdp->nxttail[RCU_NEXT_TAIL])
1615
			note_gp_changes(rsp, rdp);
1616

1617 1618 1619 1620
		if (cpu_has_callbacks_ready_to_invoke(rdp))
			cbs_ready = true;
	}
	return cbs_ready;
1621 1622
}

1623
/*
1624 1625 1626 1627
 * Allow the CPU to enter dyntick-idle mode unless it has callbacks ready
 * to invoke.  If the CPU has callbacks, try to advance them.  Tell the
 * caller to set the timeout based on whether or not there are non-lazy
 * callbacks.
1628
 *
1629
 * The caller must have disabled interrupts.
1630
 */
1631
#ifndef CONFIG_RCU_NOCB_CPU_ALL
1632
int rcu_needs_cpu(int cpu, unsigned long *dj)
1633 1634 1635
{
	struct rcu_dynticks *rdtp = &per_cpu(rcu_dynticks, cpu);

1636 1637 1638
	/* Snapshot to detect later posting of non-lazy callback. */
	rdtp->nonlazy_posted_snap = rdtp->nonlazy_posted;

1639
	/* If no callbacks, RCU doesn't need the CPU. */
1640 1641
	if (!rcu_cpu_has_callbacks(cpu, &rdtp->all_lazy)) {
		*dj = ULONG_MAX;
1642 1643
		return 0;
	}
1644 1645 1646 1647 1648

	/* Attempt to advance callbacks. */
	if (rcu_try_advance_all_cbs()) {
		/* Some ready to invoke, so initiate later invocation. */
		invoke_rcu_core();
1649 1650
		return 1;
	}
1651 1652 1653
	rdtp->last_accelerate = jiffies;

	/* Request timer delay depending on laziness, and round. */
1654
	if (!rdtp->all_lazy) {
1655 1656
		*dj = round_up(rcu_idle_gp_delay + jiffies,
			       rcu_idle_gp_delay) - jiffies;
1657
	} else {
1658
		*dj = round_jiffies(rcu_idle_lazy_gp_delay + jiffies) - jiffies;
1659
	}
1660 1661
	return 0;
}
1662
#endif /* #ifndef CONFIG_RCU_NOCB_CPU_ALL */
1663

1664
/*
1665 1666 1667 1668 1669 1670
 * Prepare a CPU for idle from an RCU perspective.  The first major task
 * is to sense whether nohz mode has been enabled or disabled via sysfs.
 * The second major task is to check to see if a non-lazy callback has
 * arrived at a CPU that previously had only lazy callbacks.  The third
 * major task is to accelerate (that is, assign grace-period numbers to)
 * any recently arrived callbacks.
1671 1672
 *
 * The caller must have disabled interrupts.
1673
 */
1674
static void rcu_prepare_for_idle(int cpu)
1675
{
1676
#ifndef CONFIG_RCU_NOCB_CPU_ALL
1677
	bool needwake;
1678
	struct rcu_data *rdp;
1679
	struct rcu_dynticks *rdtp = &per_cpu(rcu_dynticks, cpu);
1680 1681
	struct rcu_node *rnp;
	struct rcu_state *rsp;
1682 1683 1684
	int tne;

	/* Handle nohz enablement switches conservatively. */
1685
	tne = ACCESS_ONCE(tick_nohz_active);
1686
	if (tne != rdtp->tick_nohz_enabled_snap) {
1687
		if (rcu_cpu_has_callbacks(cpu, NULL))
1688 1689 1690 1691 1692 1693
			invoke_rcu_core(); /* force nohz to see update. */
		rdtp->tick_nohz_enabled_snap = tne;
		return;
	}
	if (!tne)
		return;
1694

1695
	/* If this is a no-CBs CPU, no callbacks, just return. */
1696
	if (rcu_is_nocb_cpu(cpu))
1697 1698
		return;

1699
	/*
1700 1701 1702
	 * If a non-lazy callback arrived at a CPU having only lazy
	 * callbacks, invoke RCU core for the side-effect of recalculating
	 * idle duration on re-entry to idle.
1703
	 */
1704 1705
	if (rdtp->all_lazy &&
	    rdtp->nonlazy_posted != rdtp->nonlazy_posted_snap) {
1706 1707
		rdtp->all_lazy = false;
		rdtp->nonlazy_posted_snap = rdtp->nonlazy_posted;
1708
		invoke_rcu_core();
1709 1710 1711
		return;
	}

1712
	/*
1713 1714
	 * If we have not yet accelerated this jiffy, accelerate all
	 * callbacks on this CPU.
1715
	 */
1716
	if (rdtp->last_accelerate == jiffies)
1717
		return;
1718 1719 1720 1721 1722 1723 1724
	rdtp->last_accelerate = jiffies;
	for_each_rcu_flavor(rsp) {
		rdp = per_cpu_ptr(rsp->rda, cpu);
		if (!*rdp->nxttail[RCU_DONE_TAIL])
			continue;
		rnp = rdp->mynode;
		raw_spin_lock(&rnp->lock); /* irqs already disabled. */
1725
		smp_mb__after_unlock_lock();
1726
		needwake = rcu_accelerate_cbs(rsp, rnp, rdp);
1727
		raw_spin_unlock(&rnp->lock); /* irqs remain disabled. */
1728 1729
		if (needwake)
			rcu_gp_kthread_wake(rsp);
1730
	}
1731
#endif /* #ifndef CONFIG_RCU_NOCB_CPU_ALL */
1732
}
1733

1734 1735 1736 1737 1738 1739 1740
/*
 * Clean up for exit from idle.  Attempt to advance callbacks based on
 * any grace periods that elapsed while the CPU was idle, and if any
 * callbacks are now ready to invoke, initiate invocation.
 */
static void rcu_cleanup_after_idle(int cpu)
{
1741
#ifndef CONFIG_RCU_NOCB_CPU_ALL
1742
	if (rcu_is_nocb_cpu(cpu))
1743
		return;
1744 1745
	if (rcu_try_advance_all_cbs())
		invoke_rcu_core();
1746
#endif /* #ifndef CONFIG_RCU_NOCB_CPU_ALL */
1747 1748
}

1749
/*
1750 1751 1752 1753 1754 1755
 * Keep a running count of the number of non-lazy callbacks posted
 * on this CPU.  This running counter (which is never decremented) allows
 * rcu_prepare_for_idle() to detect when something out of the idle loop
 * posts a callback, even if an equal number of callbacks are invoked.
 * Of course, callbacks should only be posted from within a trace event
 * designed to be called from idle or from within RCU_NONIDLE().
1756 1757 1758
 */
static void rcu_idle_count_callbacks_posted(void)
{
1759
	__this_cpu_add(rcu_dynticks.nonlazy_posted, 1);
1760 1761
}

1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790
/*
 * Data for flushing lazy RCU callbacks at OOM time.
 */
static atomic_t oom_callback_count;
static DECLARE_WAIT_QUEUE_HEAD(oom_callback_wq);

/*
 * RCU OOM callback -- decrement the outstanding count and deliver the
 * wake-up if we are the last one.
 */
static void rcu_oom_callback(struct rcu_head *rhp)
{
	if (atomic_dec_and_test(&oom_callback_count))
		wake_up(&oom_callback_wq);
}

/*
 * Post an rcu_oom_notify callback on the current CPU if it has at
 * least one lazy callback.  This will unnecessarily post callbacks
 * to CPUs that already have a non-lazy callback at the end of their
 * callback list, but this is an infrequent operation, so accept some
 * extra overhead to keep things simple.
 */
static void rcu_oom_notify_cpu(void *unused)
{
	struct rcu_state *rsp;
	struct rcu_data *rdp;

	for_each_rcu_flavor(rsp) {
1791
		rdp = raw_cpu_ptr(rsp->rda);
1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812
		if (rdp->qlen_lazy != 0) {
			atomic_inc(&oom_callback_count);
			rsp->call(&rdp->oom_head, rcu_oom_callback);
		}
	}
}

/*
 * If low on memory, ensure that each CPU has a non-lazy callback.
 * This will wake up CPUs that have only lazy callbacks, in turn
 * ensuring that they free up the corresponding memory in a timely manner.
 * Because an uncertain amount of memory will be freed in some uncertain
 * timeframe, we do not claim to have freed anything.
 */
static int rcu_oom_notify(struct notifier_block *self,
			  unsigned long notused, void *nfreed)
{
	int cpu;

	/* Wait for callbacks from earlier instance to complete. */
	wait_event(oom_callback_wq, atomic_read(&oom_callback_count) == 0);
1813
	smp_mb(); /* Ensure callback reuse happens after callback invocation. */
1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844

	/*
	 * Prevent premature wakeup: ensure that all increments happen
	 * before there is a chance of the counter reaching zero.
	 */
	atomic_set(&oom_callback_count, 1);

	get_online_cpus();
	for_each_online_cpu(cpu) {
		smp_call_function_single(cpu, rcu_oom_notify_cpu, NULL, 1);
		cond_resched();
	}
	put_online_cpus();

	/* Unconditionally decrement: no need to wake ourselves up. */
	atomic_dec(&oom_callback_count);

	return NOTIFY_OK;
}

static struct notifier_block rcu_oom_nb = {
	.notifier_call = rcu_oom_notify
};

static int __init rcu_register_oom_notifier(void)
{
	register_oom_notifier(&rcu_oom_nb);
	return 0;
}
early_initcall(rcu_register_oom_notifier);

1845
#endif /* #else #if !defined(CONFIG_RCU_FAST_NO_HZ) */
1846 1847 1848 1849 1850 1851 1852

#ifdef CONFIG_RCU_CPU_STALL_INFO

#ifdef CONFIG_RCU_FAST_NO_HZ

static void print_cpu_stall_fast_no_hz(char *cp, int cpu)
{
1853
	struct rcu_dynticks *rdtp = &per_cpu(rcu_dynticks, cpu);
1854
	unsigned long nlpd = rdtp->nonlazy_posted - rdtp->nonlazy_posted_snap;
1855

1856 1857 1858 1859 1860
	sprintf(cp, "last_accelerate: %04lx/%04lx, nonlazy_posted: %ld, %c%c",
		rdtp->last_accelerate & 0xffff, jiffies & 0xffff,
		ulong2long(nlpd),
		rdtp->all_lazy ? 'L' : '.',
		rdtp->tick_nohz_enabled_snap ? '.' : 'D');
1861 1862 1863 1864 1865 1866
}

#else /* #ifdef CONFIG_RCU_FAST_NO_HZ */

static void print_cpu_stall_fast_no_hz(char *cp, int cpu)
{
1867
	*cp = '\0';
1868 1869 1870 1871 1872 1873 1874
}

#endif /* #else #ifdef CONFIG_RCU_FAST_NO_HZ */

/* Initiate the stall-info list. */
static void print_cpu_stall_info_begin(void)
{
1875
	pr_cont("\n");
1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905
}

/*
 * Print out diagnostic information for the specified stalled CPU.
 *
 * If the specified CPU is aware of the current RCU grace period
 * (flavor specified by rsp), then print the number of scheduling
 * clock interrupts the CPU has taken during the time that it has
 * been aware.  Otherwise, print the number of RCU grace periods
 * that this CPU is ignorant of, for example, "1" if the CPU was
 * aware of the previous grace period.
 *
 * Also print out idle and (if CONFIG_RCU_FAST_NO_HZ) idle-entry info.
 */
static void print_cpu_stall_info(struct rcu_state *rsp, int cpu)
{
	char fast_no_hz[72];
	struct rcu_data *rdp = per_cpu_ptr(rsp->rda, cpu);
	struct rcu_dynticks *rdtp = rdp->dynticks;
	char *ticks_title;
	unsigned long ticks_value;

	if (rsp->gpnum == rdp->gpnum) {
		ticks_title = "ticks this GP";
		ticks_value = rdp->ticks_this_gp;
	} else {
		ticks_title = "GPs behind";
		ticks_value = rsp->gpnum - rdp->gpnum;
	}
	print_cpu_stall_fast_no_hz(fast_no_hz, cpu);
1906
	pr_err("\t%d: (%lu %s) idle=%03x/%llx/%d softirq=%u/%u %s\n",
1907 1908 1909
	       cpu, ticks_value, ticks_title,
	       atomic_read(&rdtp->dynticks) & 0xfff,
	       rdtp->dynticks_nesting, rdtp->dynticks_nmi_nesting,
1910
	       rdp->softirq_snap, kstat_softirqs_cpu(RCU_SOFTIRQ, cpu),
1911 1912 1913 1914 1915 1916
	       fast_no_hz);
}

/* Terminate the stall-info list. */
static void print_cpu_stall_info_end(void)
{
1917
	pr_err("\t");
1918 1919 1920 1921 1922 1923
}

/* Zero ->ticks_this_gp for all flavors of RCU. */
static void zero_cpu_stall_ticks(struct rcu_data *rdp)
{
	rdp->ticks_this_gp = 0;
1924
	rdp->softirq_snap = kstat_softirqs_cpu(RCU_SOFTIRQ, smp_processor_id());
1925 1926 1927 1928 1929
}

/* Increment ->ticks_this_gp for all flavors of RCU. */
static void increment_cpu_stall_ticks(void)
{
1930 1931 1932
	struct rcu_state *rsp;

	for_each_rcu_flavor(rsp)
1933
		raw_cpu_inc(rsp->rda->ticks_this_gp);
1934 1935 1936 1937 1938 1939
}

#else /* #ifdef CONFIG_RCU_CPU_STALL_INFO */

static void print_cpu_stall_info_begin(void)
{
1940
	pr_cont(" {");
1941 1942 1943 1944
}

static void print_cpu_stall_info(struct rcu_state *rsp, int cpu)
{
1945
	pr_cont(" %d", cpu);
1946 1947 1948 1949
}

static void print_cpu_stall_info_end(void)
{
1950
	pr_cont("} ");
1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961
}

static void zero_cpu_stall_ticks(struct rcu_data *rdp)
{
}

static void increment_cpu_stall_ticks(void)
{
}

#endif /* #else #ifdef CONFIG_RCU_CPU_STALL_INFO */
P
Paul E. McKenney 已提交
1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995

#ifdef CONFIG_RCU_NOCB_CPU

/*
 * Offload callback processing from the boot-time-specified set of CPUs
 * specified by rcu_nocb_mask.  For each CPU in the set, there is a
 * kthread created that pulls the callbacks from the corresponding CPU,
 * waits for a grace period to elapse, and invokes the callbacks.
 * The no-CBs CPUs do a wake_up() on their kthread when they insert
 * a callback into any empty list, unless the rcu_nocb_poll boot parameter
 * has been specified, in which case each kthread actively polls its
 * CPU.  (Which isn't so great for energy efficiency, but which does
 * reduce RCU's overhead on that CPU.)
 *
 * This is intended to be used in conjunction with Frederic Weisbecker's
 * adaptive-idle work, which would seriously reduce OS jitter on CPUs
 * running CPU-bound user-mode computations.
 *
 * Offloading of callback processing could also in theory be used as
 * an energy-efficiency measure because CPUs with no RCU callbacks
 * queued are more aggressive about entering dyntick-idle mode.
 */


/* Parse the boot-time rcu_nocb_mask CPU list from the kernel parameters. */
static int __init rcu_nocb_setup(char *str)
{
	alloc_bootmem_cpumask_var(&rcu_nocb_mask);
	have_rcu_nocb_mask = true;
	cpulist_parse(str, rcu_nocb_mask);
	return 1;
}
__setup("rcu_nocbs=", rcu_nocb_setup);

1996 1997 1998 1999 2000 2001 2002
static int __init parse_rcu_nocb_poll(char *arg)
{
	rcu_nocb_poll = 1;
	return 0;
}
early_param("rcu_nocb_poll", parse_rcu_nocb_poll);

2003
/*
2004 2005
 * Wake up any no-CBs CPUs' kthreads that were waiting on the just-ended
 * grace period.
2006
 */
2007
static void rcu_nocb_gp_cleanup(struct rcu_state *rsp, struct rcu_node *rnp)
2008
{
2009
	wake_up_all(&rnp->nocb_gp_wq[rnp->completed & 0x1]);
2010 2011 2012
}

/*
2013
 * Set the root rcu_node structure's ->need_future_gp field
2014 2015 2016 2017 2018
 * based on the sum of those of all rcu_node structures.  This does
 * double-count the root rcu_node structure's requests, but this
 * is necessary to handle the possibility of a rcu_nocb_kthread()
 * having awakened during the time that the rcu_node structures
 * were being updated for the end of the previous grace period.
2019
 */
2020 2021
static void rcu_nocb_gp_set(struct rcu_node *rnp, int nrq)
{
2022
	rnp->need_future_gp[(rnp->completed + 1) & 0x1] += nrq;
2023 2024 2025
}

static void rcu_init_one_nocb(struct rcu_node *rnp)
2026
{
2027 2028
	init_waitqueue_head(&rnp->nocb_gp_wq[0]);
	init_waitqueue_head(&rnp->nocb_gp_wq[1]);
2029 2030
}

2031
#ifndef CONFIG_RCU_NOCB_CPU_ALL
L
Liu Ping Fan 已提交
2032
/* Is the specified CPU a no-CBs CPU? */
2033
bool rcu_is_nocb_cpu(int cpu)
P
Paul E. McKenney 已提交
2034 2035 2036 2037 2038
{
	if (have_rcu_nocb_mask)
		return cpumask_test_cpu(cpu, rcu_nocb_mask);
	return false;
}
2039
#endif /* #ifndef CONFIG_RCU_NOCB_CPU_ALL */
P
Paul E. McKenney 已提交
2040

2041 2042 2043 2044 2045 2046 2047 2048 2049
/*
 * Kick the leader kthread for this NOCB group.
 */
static void wake_nocb_leader(struct rcu_data *rdp, bool force)
{
	struct rcu_data *rdp_leader = rdp->nocb_leader;

	if (!ACCESS_ONCE(rdp_leader->nocb_kthread))
		return;
2050
	if (ACCESS_ONCE(rdp_leader->nocb_leader_sleep) || force) {
2051
		/* Prior xchg orders against prior callback enqueue. */
2052
		ACCESS_ONCE(rdp_leader->nocb_leader_sleep) = false;
2053 2054 2055 2056
		wake_up(&rdp_leader->nocb_wq);
	}
}

P
Paul E. McKenney 已提交
2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067
/*
 * Enqueue the specified string of rcu_head structures onto the specified
 * CPU's no-CBs lists.  The CPU is specified by rdp, the head of the
 * string by rhp, and the tail of the string by rhtp.  The non-lazy/lazy
 * counts are supplied by rhcount and rhcount_lazy.
 *
 * If warranted, also wake up the kthread servicing this CPUs queues.
 */
static void __call_rcu_nocb_enqueue(struct rcu_data *rdp,
				    struct rcu_head *rhp,
				    struct rcu_head **rhtp,
2068 2069
				    int rhcount, int rhcount_lazy,
				    unsigned long flags)
P
Paul E. McKenney 已提交
2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082
{
	int len;
	struct rcu_head **old_rhpp;
	struct task_struct *t;

	/* Enqueue the callback on the nocb list and update counts. */
	old_rhpp = xchg(&rdp->nocb_tail, rhtp);
	ACCESS_ONCE(*old_rhpp) = rhp;
	atomic_long_add(rhcount, &rdp->nocb_q_count);
	atomic_long_add(rhcount_lazy, &rdp->nocb_q_count_lazy);

	/* If we are not being polled and there is a kthread, awaken it ... */
	t = ACCESS_ONCE(rdp->nocb_kthread);
2083
	if (rcu_nocb_poll || !t) {
2084 2085
		trace_rcu_nocb_wake(rdp->rsp->name, rdp->cpu,
				    TPS("WakeNotPoll"));
P
Paul E. McKenney 已提交
2086
		return;
2087
	}
P
Paul E. McKenney 已提交
2088 2089
	len = atomic_long_read(&rdp->nocb_q_count);
	if (old_rhpp == &rdp->nocb_head) {
2090
		if (!irqs_disabled_flags(flags)) {
2091 2092
			/* ... if queue was empty ... */
			wake_nocb_leader(rdp, false);
2093 2094 2095 2096 2097 2098 2099
			trace_rcu_nocb_wake(rdp->rsp->name, rdp->cpu,
					    TPS("WakeEmpty"));
		} else {
			rdp->nocb_defer_wakeup = true;
			trace_rcu_nocb_wake(rdp->rsp->name, rdp->cpu,
					    TPS("WakeEmptyIsDeferred"));
		}
P
Paul E. McKenney 已提交
2100 2101
		rdp->qlen_last_fqs_check = 0;
	} else if (len > rdp->qlen_last_fqs_check + qhimark) {
2102 2103
		/* ... or if many callbacks queued. */
		wake_nocb_leader(rdp, true);
P
Paul E. McKenney 已提交
2104
		rdp->qlen_last_fqs_check = LONG_MAX / 2;
2105 2106 2107
		trace_rcu_nocb_wake(rdp->rsp->name, rdp->cpu, TPS("WakeOvf"));
	} else {
		trace_rcu_nocb_wake(rdp->rsp->name, rdp->cpu, TPS("WakeNot"));
P
Paul E. McKenney 已提交
2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121
	}
	return;
}

/*
 * This is a helper for __call_rcu(), which invokes this when the normal
 * callback queue is inoperable.  If this is not a no-CBs CPU, this
 * function returns failure back to __call_rcu(), which can complain
 * appropriately.
 *
 * Otherwise, this function queues the callback where the corresponding
 * "rcuo" kthread can find it.
 */
static bool __call_rcu_nocb(struct rcu_data *rdp, struct rcu_head *rhp,
2122
			    bool lazy, unsigned long flags)
P
Paul E. McKenney 已提交
2123 2124
{

2125
	if (!rcu_is_nocb_cpu(rdp->cpu))
2126
		return false;
2127
	__call_rcu_nocb_enqueue(rdp, rhp, &rhp->next, 1, lazy, flags);
2128 2129 2130
	if (__is_kfree_rcu_offset((unsigned long)rhp->func))
		trace_rcu_kfree_callback(rdp->rsp->name, rhp,
					 (unsigned long)rhp->func,
2131 2132
					 -atomic_long_read(&rdp->nocb_q_count_lazy),
					 -atomic_long_read(&rdp->nocb_q_count));
2133 2134
	else
		trace_rcu_callback(rdp->rsp->name, rhp,
2135 2136
				   -atomic_long_read(&rdp->nocb_q_count_lazy),
				   -atomic_long_read(&rdp->nocb_q_count));
2137
	return true;
P
Paul E. McKenney 已提交
2138 2139 2140 2141 2142 2143 2144
}

/*
 * Adopt orphaned callbacks on a no-CBs CPU, or return 0 if this is
 * not a no-CBs CPU.
 */
static bool __maybe_unused rcu_nocb_adopt_orphan_cbs(struct rcu_state *rsp,
2145 2146
						     struct rcu_data *rdp,
						     unsigned long flags)
P
Paul E. McKenney 已提交
2147 2148 2149 2150 2151
{
	long ql = rsp->qlen;
	long qll = rsp->qlen_lazy;

	/* If this is not a no-CBs CPU, tell the caller to do it the old way. */
2152
	if (!rcu_is_nocb_cpu(smp_processor_id()))
2153
		return false;
P
Paul E. McKenney 已提交
2154 2155 2156 2157 2158 2159
	rsp->qlen = 0;
	rsp->qlen_lazy = 0;

	/* First, enqueue the donelist, if any.  This preserves CB ordering. */
	if (rsp->orphan_donelist != NULL) {
		__call_rcu_nocb_enqueue(rdp, rsp->orphan_donelist,
2160
					rsp->orphan_donetail, ql, qll, flags);
P
Paul E. McKenney 已提交
2161 2162 2163 2164 2165 2166
		ql = qll = 0;
		rsp->orphan_donelist = NULL;
		rsp->orphan_donetail = &rsp->orphan_donelist;
	}
	if (rsp->orphan_nxtlist != NULL) {
		__call_rcu_nocb_enqueue(rdp, rsp->orphan_nxtlist,
2167
					rsp->orphan_nxttail, ql, qll, flags);
P
Paul E. McKenney 已提交
2168 2169 2170 2171
		ql = qll = 0;
		rsp->orphan_nxtlist = NULL;
		rsp->orphan_nxttail = &rsp->orphan_nxtlist;
	}
2172
	return true;
P
Paul E. McKenney 已提交
2173 2174 2175
}

/*
2176 2177
 * If necessary, kick off a new grace period, and either way wait
 * for a subsequent grace period to complete.
P
Paul E. McKenney 已提交
2178
 */
2179
static void rcu_nocb_wait_gp(struct rcu_data *rdp)
P
Paul E. McKenney 已提交
2180
{
2181
	unsigned long c;
2182
	bool d;
2183
	unsigned long flags;
2184
	bool needwake;
2185 2186 2187
	struct rcu_node *rnp = rdp->mynode;

	raw_spin_lock_irqsave(&rnp->lock, flags);
2188
	smp_mb__after_unlock_lock();
2189
	needwake = rcu_start_future_gp(rnp, rdp, &c);
2190
	raw_spin_unlock_irqrestore(&rnp->lock, flags);
2191 2192
	if (needwake)
		rcu_gp_kthread_wake(rdp->rsp);
P
Paul E. McKenney 已提交
2193 2194

	/*
2195 2196
	 * Wait for the grace period.  Do so interruptibly to avoid messing
	 * up the load average.
P
Paul E. McKenney 已提交
2197
	 */
2198
	trace_rcu_future_gp(rnp, rdp, c, TPS("StartWait"));
2199
	for (;;) {
2200 2201 2202 2203
		wait_event_interruptible(
			rnp->nocb_gp_wq[c & 0x1],
			(d = ULONG_CMP_GE(ACCESS_ONCE(rnp->completed), c)));
		if (likely(d))
2204
			break;
2205
		flush_signals(current);
2206
		trace_rcu_future_gp(rnp, rdp, c, TPS("ResumeWait"));
2207
	}
2208
	trace_rcu_future_gp(rnp, rdp, c, TPS("EndWait"));
2209
	smp_mb(); /* Ensure that CB invocation happens after GP end. */
P
Paul E. McKenney 已提交
2210 2211
}

2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228
/*
 * Leaders come here to wait for additional callbacks to show up.
 * This function does not return until callbacks appear.
 */
static void nocb_leader_wait(struct rcu_data *my_rdp)
{
	bool firsttime = true;
	bool gotcbs;
	struct rcu_data *rdp;
	struct rcu_head **tail;

wait_again:

	/* Wait for callbacks to appear. */
	if (!rcu_nocb_poll) {
		trace_rcu_nocb_wake(my_rdp->rsp->name, my_rdp->cpu, "Sleep");
		wait_event_interruptible(my_rdp->nocb_wq,
2229
				!ACCESS_ONCE(my_rdp->nocb_leader_sleep));
2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267
		/* Memory barrier handled by smp_mb() calls below and repoll. */
	} else if (firsttime) {
		firsttime = false; /* Don't drown trace log with "Poll"! */
		trace_rcu_nocb_wake(my_rdp->rsp->name, my_rdp->cpu, "Poll");
	}

	/*
	 * Each pass through the following loop checks a follower for CBs.
	 * We are our own first follower.  Any CBs found are moved to
	 * nocb_gp_head, where they await a grace period.
	 */
	gotcbs = false;
	for (rdp = my_rdp; rdp; rdp = rdp->nocb_next_follower) {
		rdp->nocb_gp_head = ACCESS_ONCE(rdp->nocb_head);
		if (!rdp->nocb_gp_head)
			continue;  /* No CBs here, try next follower. */

		/* Move callbacks to wait-for-GP list, which is empty. */
		ACCESS_ONCE(rdp->nocb_head) = NULL;
		rdp->nocb_gp_tail = xchg(&rdp->nocb_tail, &rdp->nocb_head);
		rdp->nocb_gp_count = atomic_long_xchg(&rdp->nocb_q_count, 0);
		rdp->nocb_gp_count_lazy =
			atomic_long_xchg(&rdp->nocb_q_count_lazy, 0);
		gotcbs = true;
	}

	/*
	 * If there were no callbacks, sleep a bit, rescan after a
	 * memory barrier, and go retry.
	 */
	if (unlikely(!gotcbs)) {
		if (!rcu_nocb_poll)
			trace_rcu_nocb_wake(my_rdp->rsp->name, my_rdp->cpu,
					    "WokeEmpty");
		flush_signals(current);
		schedule_timeout_interruptible(1);

		/* Rescan in case we were a victim of memory ordering. */
2268 2269
		my_rdp->nocb_leader_sleep = true;
		smp_mb();  /* Ensure _sleep true before scan. */
2270 2271 2272
		for (rdp = my_rdp; rdp; rdp = rdp->nocb_next_follower)
			if (ACCESS_ONCE(rdp->nocb_head)) {
				/* Found CB, so short-circuit next wait. */
2273
				my_rdp->nocb_leader_sleep = false;
2274 2275 2276 2277 2278 2279 2280 2281 2282
				break;
			}
		goto wait_again;
	}

	/* Wait for one grace period. */
	rcu_nocb_wait_gp(my_rdp);

	/*
2283 2284
	 * We left ->nocb_leader_sleep unset to reduce cache thrashing.
	 * We set it now, but recheck for new callbacks while
2285 2286
	 * traversing our follower list.
	 */
2287 2288
	my_rdp->nocb_leader_sleep = true;
	smp_mb(); /* Ensure _sleep true before scan of ->nocb_head. */
2289 2290 2291 2292

	/* Each pass through the following loop wakes a follower, if needed. */
	for (rdp = my_rdp; rdp; rdp = rdp->nocb_next_follower) {
		if (ACCESS_ONCE(rdp->nocb_head))
2293
			my_rdp->nocb_leader_sleep = false;/* No need to sleep.*/
2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347
		if (!rdp->nocb_gp_head)
			continue; /* No CBs, so no need to wake follower. */

		/* Append callbacks to follower's "done" list. */
		tail = xchg(&rdp->nocb_follower_tail, rdp->nocb_gp_tail);
		*tail = rdp->nocb_gp_head;
		atomic_long_add(rdp->nocb_gp_count, &rdp->nocb_follower_count);
		atomic_long_add(rdp->nocb_gp_count_lazy,
				&rdp->nocb_follower_count_lazy);
		if (rdp != my_rdp && tail == &rdp->nocb_follower_head) {
			/*
			 * List was empty, wake up the follower.
			 * Memory barriers supplied by atomic_long_add().
			 */
			wake_up(&rdp->nocb_wq);
		}
	}

	/* If we (the leader) don't have CBs, go wait some more. */
	if (!my_rdp->nocb_follower_head)
		goto wait_again;
}

/*
 * Followers come here to wait for additional callbacks to show up.
 * This function does not return until callbacks appear.
 */
static void nocb_follower_wait(struct rcu_data *rdp)
{
	bool firsttime = true;

	for (;;) {
		if (!rcu_nocb_poll) {
			trace_rcu_nocb_wake(rdp->rsp->name, rdp->cpu,
					    "FollowerSleep");
			wait_event_interruptible(rdp->nocb_wq,
						 ACCESS_ONCE(rdp->nocb_follower_head));
		} else if (firsttime) {
			/* Don't drown trace log with "Poll"! */
			firsttime = false;
			trace_rcu_nocb_wake(rdp->rsp->name, rdp->cpu, "Poll");
		}
		if (smp_load_acquire(&rdp->nocb_follower_head)) {
			/* ^^^ Ensure CB invocation follows _head test. */
			return;
		}
		if (!rcu_nocb_poll)
			trace_rcu_nocb_wake(rdp->rsp->name, rdp->cpu,
					    "WokeEmpty");
		flush_signals(current);
		schedule_timeout_interruptible(1);
	}
}

P
Paul E. McKenney 已提交
2348 2349
/*
 * Per-rcu_data kthread, but only for no-CBs CPUs.  Each kthread invokes
2350 2351 2352
 * callbacks queued by the corresponding no-CBs CPU, however, there is
 * an optional leader-follower relationship so that the grace-period
 * kthreads don't have to do quite so many wakeups.
P
Paul E. McKenney 已提交
2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363
 */
static int rcu_nocb_kthread(void *arg)
{
	int c, cl;
	struct rcu_head *list;
	struct rcu_head *next;
	struct rcu_head **tail;
	struct rcu_data *rdp = arg;

	/* Each pass through this loop invokes one batch of callbacks */
	for (;;) {
2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379
		/* Wait for callbacks. */
		if (rdp->nocb_leader == rdp)
			nocb_leader_wait(rdp);
		else
			nocb_follower_wait(rdp);

		/* Pull the ready-to-invoke callbacks onto local list. */
		list = ACCESS_ONCE(rdp->nocb_follower_head);
		BUG_ON(!list);
		trace_rcu_nocb_wake(rdp->rsp->name, rdp->cpu, "WokeNonEmpty");
		ACCESS_ONCE(rdp->nocb_follower_head) = NULL;
		tail = xchg(&rdp->nocb_follower_tail, &rdp->nocb_follower_head);
		c = atomic_long_xchg(&rdp->nocb_follower_count, 0);
		cl = atomic_long_xchg(&rdp->nocb_follower_count_lazy, 0);
		rdp->nocb_p_count += c;
		rdp->nocb_p_count_lazy += cl;
P
Paul E. McKenney 已提交
2380 2381 2382 2383 2384 2385 2386 2387

		/* Each pass through the following loop invokes a callback. */
		trace_rcu_batch_start(rdp->rsp->name, cl, c, -1);
		c = cl = 0;
		while (list) {
			next = list->next;
			/* Wait for enqueuing to complete, if needed. */
			while (next == NULL && &list->next != tail) {
2388 2389
				trace_rcu_nocb_wake(rdp->rsp->name, rdp->cpu,
						    TPS("WaitQueue"));
P
Paul E. McKenney 已提交
2390
				schedule_timeout_interruptible(1);
2391 2392
				trace_rcu_nocb_wake(rdp->rsp->name, rdp->cpu,
						    TPS("WokeQueue"));
P
Paul E. McKenney 已提交
2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405
				next = list->next;
			}
			debug_rcu_head_unqueue(list);
			local_bh_disable();
			if (__rcu_reclaim(rdp->rsp->name, list))
				cl++;
			c++;
			local_bh_enable();
			list = next;
		}
		trace_rcu_batch_end(rdp->rsp->name, c, !!list, 0, 0, 1);
		ACCESS_ONCE(rdp->nocb_p_count) -= c;
		ACCESS_ONCE(rdp->nocb_p_count_lazy) -= cl;
2406
		rdp->n_nocbs_invoked += c;
P
Paul E. McKenney 已提交
2407 2408 2409 2410
	}
	return 0;
}

2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422
/* Is a deferred wakeup of rcu_nocb_kthread() required? */
static bool rcu_nocb_need_deferred_wakeup(struct rcu_data *rdp)
{
	return ACCESS_ONCE(rdp->nocb_defer_wakeup);
}

/* Do a deferred wakeup of rcu_nocb_kthread(). */
static void do_nocb_deferred_wakeup(struct rcu_data *rdp)
{
	if (!rcu_nocb_need_deferred_wakeup(rdp))
		return;
	ACCESS_ONCE(rdp->nocb_defer_wakeup) = false;
2423
	wake_nocb_leader(rdp, false);
2424 2425 2426
	trace_rcu_nocb_wake(rdp->rsp->name, rdp->cpu, TPS("DeferredWakeEmpty"));
}

2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442
void __init rcu_init_nohz(void)
{
	int cpu;
	bool need_rcu_nocb_mask = true;
	struct rcu_state *rsp;

#ifdef CONFIG_RCU_NOCB_CPU_NONE
	need_rcu_nocb_mask = false;
#endif /* #ifndef CONFIG_RCU_NOCB_CPU_NONE */

#if defined(CONFIG_NO_HZ_FULL)
	if (tick_nohz_full_running && cpumask_weight(tick_nohz_full_mask))
		need_rcu_nocb_mask = true;
#endif /* #if defined(CONFIG_NO_HZ_FULL) */

	if (!have_rcu_nocb_mask && need_rcu_nocb_mask) {
2443 2444 2445 2446
		if (!zalloc_cpumask_var(&rcu_nocb_mask, GFP_KERNEL)) {
			pr_info("rcu_nocb_mask allocation failed, callback offloading disabled.\n");
			return;
		}
2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490
		have_rcu_nocb_mask = true;
	}
	if (!have_rcu_nocb_mask)
		return;

#ifdef CONFIG_RCU_NOCB_CPU_ZERO
	pr_info("\tOffload RCU callbacks from CPU 0\n");
	cpumask_set_cpu(0, rcu_nocb_mask);
#endif /* #ifdef CONFIG_RCU_NOCB_CPU_ZERO */
#ifdef CONFIG_RCU_NOCB_CPU_ALL
	pr_info("\tOffload RCU callbacks from all CPUs\n");
	cpumask_copy(rcu_nocb_mask, cpu_possible_mask);
#endif /* #ifdef CONFIG_RCU_NOCB_CPU_ALL */
#if defined(CONFIG_NO_HZ_FULL)
	if (tick_nohz_full_running)
		cpumask_or(rcu_nocb_mask, rcu_nocb_mask, tick_nohz_full_mask);
#endif /* #if defined(CONFIG_NO_HZ_FULL) */

	if (!cpumask_subset(rcu_nocb_mask, cpu_possible_mask)) {
		pr_info("\tNote: kernel parameter 'rcu_nocbs=' contains nonexistent CPUs.\n");
		cpumask_and(rcu_nocb_mask, cpu_possible_mask,
			    rcu_nocb_mask);
	}
	cpulist_scnprintf(nocb_buf, sizeof(nocb_buf), rcu_nocb_mask);
	pr_info("\tOffload RCU callbacks from CPUs: %s.\n", nocb_buf);
	if (rcu_nocb_poll)
		pr_info("\tPoll for callbacks from no-CBs CPUs.\n");

	for_each_rcu_flavor(rsp) {
		for_each_cpu(cpu, rcu_nocb_mask) {
			struct rcu_data *rdp = per_cpu_ptr(rsp->rda, cpu);

			/*
			 * If there are early callbacks, they will need
			 * to be moved to the nocb lists.
			 */
			WARN_ON_ONCE(rdp->nxttail[RCU_NEXT_TAIL] !=
				     &rdp->nxtlist &&
				     rdp->nxttail[RCU_NEXT_TAIL] != NULL);
			init_nocb_callback_list(rdp);
		}
	}
}

P
Paul E. McKenney 已提交
2491 2492 2493 2494 2495
/* Initialize per-rcu_data variables for no-CBs CPUs. */
static void __init rcu_boot_init_nocb_percpu_data(struct rcu_data *rdp)
{
	rdp->nocb_tail = &rdp->nocb_head;
	init_waitqueue_head(&rdp->nocb_wq);
2496
	rdp->nocb_follower_tail = &rdp->nocb_follower_head;
P
Paul E. McKenney 已提交
2497 2498
}

2499 2500 2501 2502 2503 2504 2505 2506
/* How many follower CPU IDs per leader?  Default of -1 for sqrt(nr_cpu_ids). */
static int rcu_nocb_leader_stride = -1;
module_param(rcu_nocb_leader_stride, int, 0444);

/*
 * Create a kthread for each RCU flavor for each no-CBs CPU.
 * Also initialize leader-follower relationships.
 */
P
Paul E. McKenney 已提交
2507 2508 2509
static void __init rcu_spawn_nocb_kthreads(struct rcu_state *rsp)
{
	int cpu;
2510 2511
	int ls = rcu_nocb_leader_stride;
	int nl = 0;  /* Next leader. */
P
Paul E. McKenney 已提交
2512
	struct rcu_data *rdp;
2513 2514
	struct rcu_data *rdp_leader = NULL;  /* Suppress misguided gcc warn. */
	struct rcu_data *rdp_prev = NULL;
P
Paul E. McKenney 已提交
2515 2516 2517 2518
	struct task_struct *t;

	if (rcu_nocb_mask == NULL)
		return;
2519 2520 2521 2522 2523 2524 2525 2526 2527
	if (ls == -1) {
		ls = int_sqrt(nr_cpu_ids);
		rcu_nocb_leader_stride = ls;
	}

	/*
	 * Each pass through this loop sets up one rcu_data structure and
	 * spawns one rcu_nocb_kthread().
	 */
P
Paul E. McKenney 已提交
2528 2529
	for_each_cpu(cpu, rcu_nocb_mask) {
		rdp = per_cpu_ptr(rsp->rda, cpu);
2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542
		if (rdp->cpu >= nl) {
			/* New leader, set up for followers & next leader. */
			nl = DIV_ROUND_UP(rdp->cpu + 1, ls) * ls;
			rdp->nocb_leader = rdp;
			rdp_leader = rdp;
		} else {
			/* Another follower, link to previous leader. */
			rdp->nocb_leader = rdp_leader;
			rdp_prev->nocb_next_follower = rdp;
		}
		rdp_prev = rdp;

		/* Spawn the kthread for this CPU. */
2543 2544
		t = kthread_run(rcu_nocb_kthread, rdp,
				"rcuo%c/%d", rsp->abbr, cpu);
P
Paul E. McKenney 已提交
2545 2546 2547 2548 2549 2550
		BUG_ON(IS_ERR(t));
		ACCESS_ONCE(rdp->nocb_kthread) = t;
	}
}

/* Prevent __call_rcu() from enqueuing callbacks on no-CBs CPUs */
2551
static bool init_nocb_callback_list(struct rcu_data *rdp)
P
Paul E. McKenney 已提交
2552 2553 2554
{
	if (rcu_nocb_mask == NULL ||
	    !cpumask_test_cpu(rdp->cpu, rcu_nocb_mask))
2555
		return false;
P
Paul E. McKenney 已提交
2556
	rdp->nxttail[RCU_NEXT_TAIL] = NULL;
2557
	return true;
P
Paul E. McKenney 已提交
2558 2559
}

2560 2561
#else /* #ifdef CONFIG_RCU_NOCB_CPU */

2562
static void rcu_nocb_gp_cleanup(struct rcu_state *rsp, struct rcu_node *rnp)
P
Paul E. McKenney 已提交
2563 2564 2565
{
}

2566 2567 2568 2569 2570 2571 2572
static void rcu_nocb_gp_set(struct rcu_node *rnp, int nrq)
{
}

static void rcu_init_one_nocb(struct rcu_node *rnp)
{
}
P
Paul E. McKenney 已提交
2573 2574

static bool __call_rcu_nocb(struct rcu_data *rdp, struct rcu_head *rhp,
2575
			    bool lazy, unsigned long flags)
P
Paul E. McKenney 已提交
2576
{
2577
	return false;
P
Paul E. McKenney 已提交
2578 2579 2580
}

static bool __maybe_unused rcu_nocb_adopt_orphan_cbs(struct rcu_state *rsp,
2581 2582
						     struct rcu_data *rdp,
						     unsigned long flags)
P
Paul E. McKenney 已提交
2583
{
2584
	return false;
P
Paul E. McKenney 已提交
2585 2586 2587 2588 2589 2590
}

static void __init rcu_boot_init_nocb_percpu_data(struct rcu_data *rdp)
{
}

2591 2592 2593 2594 2595 2596 2597 2598 2599
static bool rcu_nocb_need_deferred_wakeup(struct rcu_data *rdp)
{
	return false;
}

static void do_nocb_deferred_wakeup(struct rcu_data *rdp)
{
}

P
Paul E. McKenney 已提交
2600 2601 2602 2603
static void __init rcu_spawn_nocb_kthreads(struct rcu_state *rsp)
{
}

2604
static bool init_nocb_callback_list(struct rcu_data *rdp)
P
Paul E. McKenney 已提交
2605
{
2606
	return false;
P
Paul E. McKenney 已提交
2607 2608 2609
}

#endif /* #else #ifdef CONFIG_RCU_NOCB_CPU */
2610 2611 2612 2613 2614 2615 2616 2617 2618 2619

/*
 * An adaptive-ticks CPU can potentially execute in kernel mode for an
 * arbitrarily long period of time with the scheduling-clock tick turned
 * off.  RCU will be paying attention to this CPU because it is in the
 * kernel, but the CPU cannot be guaranteed to be executing the RCU state
 * machine because the scheduling-clock tick has been disabled.  Therefore,
 * if an adaptive-ticks CPU is failing to respond to the current grace
 * period and has not be idle from an RCU perspective, kick it.
 */
2620
static void __maybe_unused rcu_kick_nohz_cpu(int cpu)
2621 2622 2623 2624 2625 2626
{
#ifdef CONFIG_NO_HZ_FULL
	if (tick_nohz_full_cpu(cpu))
		smp_send_reschedule(cpu);
#endif /* #ifdef CONFIG_NO_HZ_FULL */
}
2627 2628 2629 2630


#ifdef CONFIG_NO_HZ_FULL_SYSIDLE

2631 2632 2633 2634 2635
/*
 * Define RCU flavor that holds sysidle state.  This needs to be the
 * most active flavor of RCU.
 */
#ifdef CONFIG_PREEMPT_RCU
2636
static struct rcu_state *rcu_sysidle_state = &rcu_preempt_state;
2637
#else /* #ifdef CONFIG_PREEMPT_RCU */
2638
static struct rcu_state *rcu_sysidle_state = &rcu_sched_state;
2639 2640
#endif /* #else #ifdef CONFIG_PREEMPT_RCU */

2641
static int full_sysidle_state;		/* Current system-idle state. */
2642 2643 2644 2645 2646 2647
#define RCU_SYSIDLE_NOT		0	/* Some CPU is not idle. */
#define RCU_SYSIDLE_SHORT	1	/* All CPUs idle for brief period. */
#define RCU_SYSIDLE_LONG	2	/* All CPUs idle for long enough. */
#define RCU_SYSIDLE_FULL	3	/* All CPUs idle, ready for sysidle. */
#define RCU_SYSIDLE_FULL_NOTED	4	/* Actually entered sysidle state. */

2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677
/*
 * Invoked to note exit from irq or task transition to idle.  Note that
 * usermode execution does -not- count as idle here!  After all, we want
 * to detect full-system idle states, not RCU quiescent states and grace
 * periods.  The caller must have disabled interrupts.
 */
static void rcu_sysidle_enter(struct rcu_dynticks *rdtp, int irq)
{
	unsigned long j;

	/* Adjust nesting, check for fully idle. */
	if (irq) {
		rdtp->dynticks_idle_nesting--;
		WARN_ON_ONCE(rdtp->dynticks_idle_nesting < 0);
		if (rdtp->dynticks_idle_nesting != 0)
			return;  /* Still not fully idle. */
	} else {
		if ((rdtp->dynticks_idle_nesting & DYNTICK_TASK_NEST_MASK) ==
		    DYNTICK_TASK_NEST_VALUE) {
			rdtp->dynticks_idle_nesting = 0;
		} else {
			rdtp->dynticks_idle_nesting -= DYNTICK_TASK_NEST_VALUE;
			WARN_ON_ONCE(rdtp->dynticks_idle_nesting < 0);
			return;  /* Still not fully idle. */
		}
	}

	/* Record start of fully idle period. */
	j = jiffies;
	ACCESS_ONCE(rdtp->dynticks_idle_jiffies) = j;
2678
	smp_mb__before_atomic();
2679
	atomic_inc(&rdtp->dynticks_idle);
2680
	smp_mb__after_atomic();
2681 2682 2683
	WARN_ON_ONCE(atomic_read(&rdtp->dynticks_idle) & 0x1);
}

2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715
/*
 * Unconditionally force exit from full system-idle state.  This is
 * invoked when a normal CPU exits idle, but must be called separately
 * for the timekeeping CPU (tick_do_timer_cpu).  The reason for this
 * is that the timekeeping CPU is permitted to take scheduling-clock
 * interrupts while the system is in system-idle state, and of course
 * rcu_sysidle_exit() has no way of distinguishing a scheduling-clock
 * interrupt from any other type of interrupt.
 */
void rcu_sysidle_force_exit(void)
{
	int oldstate = ACCESS_ONCE(full_sysidle_state);
	int newoldstate;

	/*
	 * Each pass through the following loop attempts to exit full
	 * system-idle state.  If contention proves to be a problem,
	 * a trylock-based contention tree could be used here.
	 */
	while (oldstate > RCU_SYSIDLE_SHORT) {
		newoldstate = cmpxchg(&full_sysidle_state,
				      oldstate, RCU_SYSIDLE_NOT);
		if (oldstate == newoldstate &&
		    oldstate == RCU_SYSIDLE_FULL_NOTED) {
			rcu_kick_nohz_cpu(tick_do_timer_cpu);
			return; /* We cleared it, done! */
		}
		oldstate = newoldstate;
	}
	smp_mb(); /* Order initial oldstate fetch vs. later non-idle work. */
}

2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744
/*
 * Invoked to note entry to irq or task transition from idle.  Note that
 * usermode execution does -not- count as idle here!  The caller must
 * have disabled interrupts.
 */
static void rcu_sysidle_exit(struct rcu_dynticks *rdtp, int irq)
{
	/* Adjust nesting, check for already non-idle. */
	if (irq) {
		rdtp->dynticks_idle_nesting++;
		WARN_ON_ONCE(rdtp->dynticks_idle_nesting <= 0);
		if (rdtp->dynticks_idle_nesting != 1)
			return; /* Already non-idle. */
	} else {
		/*
		 * Allow for irq misnesting.  Yes, it really is possible
		 * to enter an irq handler then never leave it, and maybe
		 * also vice versa.  Handle both possibilities.
		 */
		if (rdtp->dynticks_idle_nesting & DYNTICK_TASK_NEST_MASK) {
			rdtp->dynticks_idle_nesting += DYNTICK_TASK_NEST_VALUE;
			WARN_ON_ONCE(rdtp->dynticks_idle_nesting <= 0);
			return; /* Already non-idle. */
		} else {
			rdtp->dynticks_idle_nesting = DYNTICK_TASK_EXIT_IDLE;
		}
	}

	/* Record end of idle period. */
2745
	smp_mb__before_atomic();
2746
	atomic_inc(&rdtp->dynticks_idle);
2747
	smp_mb__after_atomic();
2748
	WARN_ON_ONCE(!(atomic_read(&rdtp->dynticks_idle) & 0x1));
2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784

	/*
	 * If we are the timekeeping CPU, we are permitted to be non-idle
	 * during a system-idle state.  This must be the case, because
	 * the timekeeping CPU has to take scheduling-clock interrupts
	 * during the time that the system is transitioning to full
	 * system-idle state.  This means that the timekeeping CPU must
	 * invoke rcu_sysidle_force_exit() directly if it does anything
	 * more than take a scheduling-clock interrupt.
	 */
	if (smp_processor_id() == tick_do_timer_cpu)
		return;

	/* Update system-idle state: We are clearly no longer fully idle! */
	rcu_sysidle_force_exit();
}

/*
 * Check to see if the current CPU is idle.  Note that usermode execution
 * does not count as idle.  The caller must have disabled interrupts.
 */
static void rcu_sysidle_check_cpu(struct rcu_data *rdp, bool *isidle,
				  unsigned long *maxj)
{
	int cur;
	unsigned long j;
	struct rcu_dynticks *rdtp = rdp->dynticks;

	/*
	 * If some other CPU has already reported non-idle, if this is
	 * not the flavor of RCU that tracks sysidle state, or if this
	 * is an offline or the timekeeping CPU, nothing to do.
	 */
	if (!*isidle || rdp->rsp != rcu_sysidle_state ||
	    cpu_is_offline(rdp->cpu) || rdp->cpu == tick_do_timer_cpu)
		return;
2785 2786
	if (rcu_gp_in_progress(rdp->rsp))
		WARN_ON_ONCE(smp_processor_id() != tick_do_timer_cpu);
2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874

	/* Pick up current idle and NMI-nesting counter and check. */
	cur = atomic_read(&rdtp->dynticks_idle);
	if (cur & 0x1) {
		*isidle = false; /* We are not idle! */
		return;
	}
	smp_mb(); /* Read counters before timestamps. */

	/* Pick up timestamps. */
	j = ACCESS_ONCE(rdtp->dynticks_idle_jiffies);
	/* If this CPU entered idle more recently, update maxj timestamp. */
	if (ULONG_CMP_LT(*maxj, j))
		*maxj = j;
}

/*
 * Is this the flavor of RCU that is handling full-system idle?
 */
static bool is_sysidle_rcu_state(struct rcu_state *rsp)
{
	return rsp == rcu_sysidle_state;
}

/*
 * Return a delay in jiffies based on the number of CPUs, rcu_node
 * leaf fanout, and jiffies tick rate.  The idea is to allow larger
 * systems more time to transition to full-idle state in order to
 * avoid the cache thrashing that otherwise occur on the state variable.
 * Really small systems (less than a couple of tens of CPUs) should
 * instead use a single global atomically incremented counter, and later
 * versions of this will automatically reconfigure themselves accordingly.
 */
static unsigned long rcu_sysidle_delay(void)
{
	if (nr_cpu_ids <= CONFIG_NO_HZ_FULL_SYSIDLE_SMALL)
		return 0;
	return DIV_ROUND_UP(nr_cpu_ids * HZ, rcu_fanout_leaf * 1000);
}

/*
 * Advance the full-system-idle state.  This is invoked when all of
 * the non-timekeeping CPUs are idle.
 */
static void rcu_sysidle(unsigned long j)
{
	/* Check the current state. */
	switch (ACCESS_ONCE(full_sysidle_state)) {
	case RCU_SYSIDLE_NOT:

		/* First time all are idle, so note a short idle period. */
		ACCESS_ONCE(full_sysidle_state) = RCU_SYSIDLE_SHORT;
		break;

	case RCU_SYSIDLE_SHORT:

		/*
		 * Idle for a bit, time to advance to next state?
		 * cmpxchg failure means race with non-idle, let them win.
		 */
		if (ULONG_CMP_GE(jiffies, j + rcu_sysidle_delay()))
			(void)cmpxchg(&full_sysidle_state,
				      RCU_SYSIDLE_SHORT, RCU_SYSIDLE_LONG);
		break;

	case RCU_SYSIDLE_LONG:

		/*
		 * Do an additional check pass before advancing to full.
		 * cmpxchg failure means race with non-idle, let them win.
		 */
		if (ULONG_CMP_GE(jiffies, j + rcu_sysidle_delay()))
			(void)cmpxchg(&full_sysidle_state,
				      RCU_SYSIDLE_LONG, RCU_SYSIDLE_FULL);
		break;

	default:
		break;
	}
}

/*
 * Found a non-idle non-timekeeping CPU, so kick the system-idle state
 * back to the beginning.
 */
static void rcu_sysidle_cancel(void)
{
	smp_mb();
2875 2876
	if (full_sysidle_state > RCU_SYSIDLE_SHORT)
		ACCESS_ONCE(full_sysidle_state) = RCU_SYSIDLE_NOT;
2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991
}

/*
 * Update the sysidle state based on the results of a force-quiescent-state
 * scan of the CPUs' dyntick-idle state.
 */
static void rcu_sysidle_report(struct rcu_state *rsp, int isidle,
			       unsigned long maxj, bool gpkt)
{
	if (rsp != rcu_sysidle_state)
		return;  /* Wrong flavor, ignore. */
	if (gpkt && nr_cpu_ids <= CONFIG_NO_HZ_FULL_SYSIDLE_SMALL)
		return;  /* Running state machine from timekeeping CPU. */
	if (isidle)
		rcu_sysidle(maxj);    /* More idle! */
	else
		rcu_sysidle_cancel(); /* Idle is over. */
}

/*
 * Wrapper for rcu_sysidle_report() when called from the grace-period
 * kthread's context.
 */
static void rcu_sysidle_report_gp(struct rcu_state *rsp, int isidle,
				  unsigned long maxj)
{
	rcu_sysidle_report(rsp, isidle, maxj, true);
}

/* Callback and function for forcing an RCU grace period. */
struct rcu_sysidle_head {
	struct rcu_head rh;
	int inuse;
};

static void rcu_sysidle_cb(struct rcu_head *rhp)
{
	struct rcu_sysidle_head *rshp;

	/*
	 * The following memory barrier is needed to replace the
	 * memory barriers that would normally be in the memory
	 * allocator.
	 */
	smp_mb();  /* grace period precedes setting inuse. */

	rshp = container_of(rhp, struct rcu_sysidle_head, rh);
	ACCESS_ONCE(rshp->inuse) = 0;
}

/*
 * Check to see if the system is fully idle, other than the timekeeping CPU.
 * The caller must have disabled interrupts.
 */
bool rcu_sys_is_idle(void)
{
	static struct rcu_sysidle_head rsh;
	int rss = ACCESS_ONCE(full_sysidle_state);

	if (WARN_ON_ONCE(smp_processor_id() != tick_do_timer_cpu))
		return false;

	/* Handle small-system case by doing a full scan of CPUs. */
	if (nr_cpu_ids <= CONFIG_NO_HZ_FULL_SYSIDLE_SMALL) {
		int oldrss = rss - 1;

		/*
		 * One pass to advance to each state up to _FULL.
		 * Give up if any pass fails to advance the state.
		 */
		while (rss < RCU_SYSIDLE_FULL && oldrss < rss) {
			int cpu;
			bool isidle = true;
			unsigned long maxj = jiffies - ULONG_MAX / 4;
			struct rcu_data *rdp;

			/* Scan all the CPUs looking for nonidle CPUs. */
			for_each_possible_cpu(cpu) {
				rdp = per_cpu_ptr(rcu_sysidle_state->rda, cpu);
				rcu_sysidle_check_cpu(rdp, &isidle, &maxj);
				if (!isidle)
					break;
			}
			rcu_sysidle_report(rcu_sysidle_state,
					   isidle, maxj, false);
			oldrss = rss;
			rss = ACCESS_ONCE(full_sysidle_state);
		}
	}

	/* If this is the first observation of an idle period, record it. */
	if (rss == RCU_SYSIDLE_FULL) {
		rss = cmpxchg(&full_sysidle_state,
			      RCU_SYSIDLE_FULL, RCU_SYSIDLE_FULL_NOTED);
		return rss == RCU_SYSIDLE_FULL;
	}

	smp_mb(); /* ensure rss load happens before later caller actions. */

	/* If already fully idle, tell the caller (in case of races). */
	if (rss == RCU_SYSIDLE_FULL_NOTED)
		return true;

	/*
	 * If we aren't there yet, and a grace period is not in flight,
	 * initiate a grace period.  Either way, tell the caller that
	 * we are not there yet.  We use an xchg() rather than an assignment
	 * to make up for the memory barriers that would otherwise be
	 * provided by the memory allocator.
	 */
	if (nr_cpu_ids > CONFIG_NO_HZ_FULL_SYSIDLE_SMALL &&
	    !rcu_gp_in_progress(rcu_sysidle_state) &&
	    !rsh.inuse && xchg(&rsh.inuse, 1) == 0)
		call_rcu(&rsh.rh, rcu_sysidle_cb);
	return false;
2992 2993
}

2994 2995 2996 2997 2998 2999 3000 3001 3002 3003
/*
 * Initialize dynticks sysidle state for CPUs coming online.
 */
static void rcu_sysidle_init_percpu_data(struct rcu_dynticks *rdtp)
{
	rdtp->dynticks_idle_nesting = DYNTICK_TASK_NEST_VALUE;
}

#else /* #ifdef CONFIG_NO_HZ_FULL_SYSIDLE */

3004 3005 3006 3007 3008 3009 3010 3011
static void rcu_sysidle_enter(struct rcu_dynticks *rdtp, int irq)
{
}

static void rcu_sysidle_exit(struct rcu_dynticks *rdtp, int irq)
{
}

3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026
static void rcu_sysidle_check_cpu(struct rcu_data *rdp, bool *isidle,
				  unsigned long *maxj)
{
}

static bool is_sysidle_rcu_state(struct rcu_state *rsp)
{
	return false;
}

static void rcu_sysidle_report_gp(struct rcu_state *rsp, int isidle,
				  unsigned long maxj)
{
}

3027 3028 3029 3030 3031
static void rcu_sysidle_init_percpu_data(struct rcu_dynticks *rdtp)
{
}

#endif /* #else #ifdef CONFIG_NO_HZ_FULL_SYSIDLE */
3032 3033 3034 3035 3036 3037 3038 3039

/*
 * Is this CPU a NO_HZ_FULL CPU that should ignore RCU so that the
 * grace-period kthread will do force_quiescent_state() processing?
 * The idea is to avoid waking up RCU core processing on such a
 * CPU unless the grace period has extended for too long.
 *
 * This code relies on the fact that all NO_HZ_FULL CPUs are also
3040
 * CONFIG_RCU_NOCB_CPU CPUs.
3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051
 */
static bool rcu_nohz_full_cpu(struct rcu_state *rsp)
{
#ifdef CONFIG_NO_HZ_FULL
	if (tick_nohz_full_cpu(smp_processor_id()) &&
	    (!rcu_gp_in_progress(rsp) ||
	     ULONG_CMP_LT(jiffies, ACCESS_ONCE(rsp->gp_start) + HZ)))
		return 1;
#endif /* #ifdef CONFIG_NO_HZ_FULL */
	return 0;
}
3052 3053 3054 3055 3056 3057 3058

/*
 * Bind the grace-period kthread for the sysidle flavor of RCU to the
 * timekeeping CPU.
 */
static void rcu_bind_gp_kthread(void)
{
3059
	int __maybe_unused cpu;
3060

3061
	if (!tick_nohz_full_enabled())
3062
		return;
3063 3064 3065
#ifdef CONFIG_NO_HZ_FULL_SYSIDLE
	cpu = tick_do_timer_cpu;
	if (cpu >= 0 && cpu < nr_cpu_ids && raw_smp_processor_id() != cpu)
3066
		set_cpus_allowed_ptr(current, cpumask_of(cpu));
3067 3068 3069 3070
#else /* #ifdef CONFIG_NO_HZ_FULL_SYSIDLE */
	if (!is_housekeeping_cpu(raw_smp_processor_id()))
		housekeeping_affine(current);
#endif /* #else #ifdef CONFIG_NO_HZ_FULL_SYSIDLE */
3071
}