rcutorture.c 44.5 KB
Newer Older
1
/*
2
 * Read-Copy Update module-based torture test facility
3 4 5 6 7 8 9 10 11 12 13 14
 *
 * 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
15 16
 * along with this program; if not, you can access it online at
 * http://www.gnu.org/licenses/gpl-2.0.html.
17
 *
18
 * Copyright (C) IBM Corporation, 2005, 2006
19 20
 *
 * Authors: Paul E. McKenney <paulmck@us.ibm.com>
21
 *	  Josh Triplett <josh@freedesktop.org>
22 23 24 25 26 27 28 29 30 31 32 33 34 35
 *
 * See also:  Documentation/RCU/torture.txt
 */
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/kthread.h>
#include <linux/err.h>
#include <linux/spinlock.h>
#include <linux/smp.h>
#include <linux/rcupdate.h>
#include <linux/interrupt.h>
#include <linux/sched.h>
A
Arun Sharma 已提交
36
#include <linux/atomic.h>
37 38 39 40 41
#include <linux/bitops.h>
#include <linux/completion.h>
#include <linux/moduleparam.h>
#include <linux/percpu.h>
#include <linux/notifier.h>
42
#include <linux/reboot.h>
43
#include <linux/freezer.h>
44 45 46
#include <linux/cpu.h>
#include <linux/delay.h>
#include <linux/stat.h>
47
#include <linux/srcu.h>
48
#include <linux/slab.h>
49
#include <linux/trace_clock.h>
50
#include <asm/byteorder.h>
51
#include <linux/torture.h>
52 53

MODULE_LICENSE("GPL");
54
MODULE_AUTHOR("Paul E. McKenney <paulmck@us.ibm.com> and Josh Triplett <josh@freedesktop.org>");
55

56

57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88
torture_param(int, fqs_duration, 0,
	      "Duration of fqs bursts (us), 0 to disable");
torture_param(int, fqs_holdoff, 0, "Holdoff time within fqs bursts (us)");
torture_param(int, fqs_stutter, 3, "Wait time between fqs bursts (s)");
torture_param(bool, gp_exp, false, "Use expedited GP wait primitives");
torture_param(bool, gp_normal, false,
	     "Use normal (non-expedited) GP wait primitives");
torture_param(int, irqreader, 1, "Allow RCU readers from irq handlers");
torture_param(int, n_barrier_cbs, 0,
	     "# of callbacks/kthreads for barrier testing");
torture_param(int, nfakewriters, 4, "Number of RCU fake writer threads");
torture_param(int, nreaders, -1, "Number of RCU reader threads");
torture_param(int, object_debug, 0,
	     "Enable debug-object double call_rcu() testing");
torture_param(int, onoff_holdoff, 0, "Time after boot before CPU hotplugs (s)");
torture_param(int, onoff_interval, 0,
	     "Time between CPU hotplugs (s), 0=disable");
torture_param(int, shuffle_interval, 3, "Number of seconds between shuffles");
torture_param(int, shutdown_secs, 0, "Shutdown time (s), <= zero to disable.");
torture_param(int, stall_cpu, 0, "Stall duration (s), zero to disable.");
torture_param(int, stall_cpu_holdoff, 10,
	     "Time to wait before starting stall (s).");
torture_param(int, stat_interval, 60,
	     "Number of seconds between stats printk()s");
torture_param(int, stutter, 5, "Number of seconds to run/halt test");
torture_param(int, test_boost, 1, "Test RCU prio boost: 0=no, 1=maybe, 2=yes.");
torture_param(int, test_boost_duration, 4,
	     "Duration of each boost test, seconds.");
torture_param(int, test_boost_interval, 7,
	     "Interval between boost tests, seconds.");
torture_param(bool, test_no_idle_hz, true,
	     "Test support for tickless idle CPUs");
89 90
torture_param(bool, verbose, true,
	     "Enable verbose debugging printk()s");
91

92
static char *torture_type = "rcu";
93
module_param(torture_type, charp, 0444);
94
MODULE_PARM_DESC(torture_type, "Type of RCU to torture (rcu, rcu_bh, ...)");
95 96 97

static int nrealreaders;
static struct task_struct *writer_task;
98
static struct task_struct **fakewriter_tasks;
99 100
static struct task_struct **reader_tasks;
static struct task_struct *stats_task;
101
static struct task_struct *fqs_task;
102
static struct task_struct *boost_tasks[NR_CPUS];
103
static struct task_struct *stall_task;
104 105
static struct task_struct **barrier_cbs_tasks;
static struct task_struct *barrier_task;
106 107 108 109 110 111 112

#define RCU_TORTURE_PIPE_LEN 10

struct rcu_torture {
	struct rcu_head rtort_rcu;
	int rtort_pipe_count;
	struct list_head rtort_free;
113
	int rtort_mbtest;
114 115 116
};

static LIST_HEAD(rcu_torture_freelist);
117
static struct rcu_torture __rcu *rcu_torture_current;
118
static unsigned long rcu_torture_current_version;
119 120
static struct rcu_torture rcu_tortures[10 * RCU_TORTURE_PIPE_LEN];
static DEFINE_SPINLOCK(rcu_torture_lock);
121 122 123 124
static DEFINE_PER_CPU(long [RCU_TORTURE_PIPE_LEN + 1],
		      rcu_torture_count) = { 0 };
static DEFINE_PER_CPU(long [RCU_TORTURE_PIPE_LEN + 1],
		      rcu_torture_batch) = { 0 };
125
static atomic_t rcu_torture_wcount[RCU_TORTURE_PIPE_LEN + 1];
126 127 128 129 130
static atomic_t n_rcu_torture_alloc;
static atomic_t n_rcu_torture_alloc_fail;
static atomic_t n_rcu_torture_free;
static atomic_t n_rcu_torture_mberror;
static atomic_t n_rcu_torture_error;
131
static long n_rcu_torture_barrier_error;
132 133 134 135
static long n_rcu_torture_boost_ktrerror;
static long n_rcu_torture_boost_rterror;
static long n_rcu_torture_boost_failure;
static long n_rcu_torture_boosts;
136
static long n_rcu_torture_timers;
137 138
static long n_barrier_attempts;
static long n_barrier_successes;
139
static struct list_head rcu_torture_removed;
140

141 142 143 144 145 146
#if defined(MODULE) || defined(CONFIG_RCU_TORTURE_TEST_RUNNABLE)
#define RCUTORTURE_RUNNABLE_INIT 1
#else
#define RCUTORTURE_RUNNABLE_INIT 0
#endif
int rcutorture_runnable = RCUTORTURE_RUNNABLE_INIT;
147 148
module_param(rcutorture_runnable, int, 0444);
MODULE_PARM_DESC(rcutorture_runnable, "Start rcutorture at boot");
149

150
#if defined(CONFIG_RCU_BOOST) && !defined(CONFIG_HOTPLUG_CPU)
151
#define rcu_can_boost() 1
152
#else /* #if defined(CONFIG_RCU_BOOST) && !defined(CONFIG_HOTPLUG_CPU) */
153
#define rcu_can_boost() 0
154
#endif /* #else #if defined(CONFIG_RCU_BOOST) && !defined(CONFIG_HOTPLUG_CPU) */
155

156 157 158 159 160 161 162 163 164 165 166 167 168 169
#ifdef CONFIG_RCU_TRACE
static u64 notrace rcu_trace_clock_local(void)
{
	u64 ts = trace_clock_local();
	unsigned long __maybe_unused ts_rem = do_div(ts, NSEC_PER_USEC);
	return ts;
}
#else /* #ifdef CONFIG_RCU_TRACE */
static u64 notrace rcu_trace_clock_local(void)
{
	return 0ULL;
}
#endif /* #else #ifdef CONFIG_RCU_TRACE */

170 171 172
static unsigned long boost_starttime;	/* jiffies of next boost test start. */
DEFINE_MUTEX(boost_mutex);		/* protect setting boost_starttime */
					/*  and boost task create/destroy. */
173
static atomic_t barrier_cbs_count;	/* Barrier callbacks registered. */
174
static bool barrier_phase;		/* Test phase. */
175 176 177
static atomic_t barrier_cbs_invoked;	/* Barrier callbacks invoked. */
static wait_queue_head_t *barrier_cbs_wq; /* Coordinate barrier testing. */
static DECLARE_WAIT_QUEUE_HEAD(barrier_wq);
178

179 180 181
/*
 * Allocate an element from the rcu_tortures pool.
 */
A
Adrian Bunk 已提交
182
static struct rcu_torture *
183 184 185 186
rcu_torture_alloc(void)
{
	struct list_head *p;

187
	spin_lock_bh(&rcu_torture_lock);
188 189
	if (list_empty(&rcu_torture_freelist)) {
		atomic_inc(&n_rcu_torture_alloc_fail);
190
		spin_unlock_bh(&rcu_torture_lock);
191 192 193 194 195
		return NULL;
	}
	atomic_inc(&n_rcu_torture_alloc);
	p = rcu_torture_freelist.next;
	list_del_init(p);
196
	spin_unlock_bh(&rcu_torture_lock);
197 198 199 200 201 202 203 204 205 206
	return container_of(p, struct rcu_torture, rtort_free);
}

/*
 * Free an element to the rcu_tortures pool.
 */
static void
rcu_torture_free(struct rcu_torture *p)
{
	atomic_inc(&n_rcu_torture_free);
207
	spin_lock_bh(&rcu_torture_lock);
208
	list_add_tail(&p->rtort_free, &rcu_torture_freelist);
209
	spin_unlock_bh(&rcu_torture_lock);
210 211
}

212 213 214 215 216 217 218
/*
 * Operations vector for selecting different types of tests.
 */

struct rcu_torture_ops {
	void (*init)(void);
	int (*readlock)(void);
219
	void (*read_delay)(struct torture_random_state *rrsp);
220 221
	void (*readunlock)(int idx);
	int (*completed)(void);
222
	void (*deferred_free)(struct rcu_torture *p);
223
	void (*sync)(void);
224
	void (*exp_sync)(void);
225
	void (*call)(struct rcu_head *head, void (*func)(struct rcu_head *rcu));
226
	void (*cb_barrier)(void);
227
	void (*fqs)(void);
228
	void (*stats)(char *page);
229
	int irq_capable;
230
	int can_boost;
231
	const char *name;
232
};
233 234

static struct rcu_torture_ops *cur_ops;
235 236 237 238 239

/*
 * Definitions for rcu torture testing.
 */

240
static int rcu_torture_read_lock(void) __acquires(RCU)
241 242 243 244 245
{
	rcu_read_lock();
	return 0;
}

246
static void rcu_read_delay(struct torture_random_state *rrsp)
247
{
248 249
	const unsigned long shortdelay_us = 200;
	const unsigned long longdelay_ms = 50;
250

251 252 253
	/* We want a short delay sometimes to make a reader delay the grace
	 * period, and we want a long delay occasionally to trigger
	 * force_quiescent_state. */
254

255
	if (!(torture_random(rrsp) % (nrealreaders * 2000 * longdelay_ms)))
256
		mdelay(longdelay_ms);
257
	if (!(torture_random(rrsp) % (nrealreaders * 2 * shortdelay_us)))
258
		udelay(shortdelay_us);
259
#ifdef CONFIG_PREEMPT
260 261
	if (!preempt_count() &&
	    !(torture_random(rrsp) % (nrealreaders * 20000)))
262 263
		preempt_schedule();  /* No QS if preempt_disable() in effect */
#endif
264 265
}

266
static void rcu_torture_read_unlock(int idx) __releases(RCU)
267 268 269 270 271 272 273 274 275 276 277 278 279 280 281
{
	rcu_read_unlock();
}

static int rcu_torture_completed(void)
{
	return rcu_batches_completed();
}

static void
rcu_torture_cb(struct rcu_head *p)
{
	int i;
	struct rcu_torture *rp = container_of(p, struct rcu_torture, rtort_rcu);

282
	if (torture_must_stop_irq()) {
283 284 285 286 287 288 289 290 291 292 293
		/* Test is ending, just drop callbacks on the floor. */
		/* The next initialization will pick up the pieces. */
		return;
	}
	i = rp->rtort_pipe_count;
	if (i > RCU_TORTURE_PIPE_LEN)
		i = RCU_TORTURE_PIPE_LEN;
	atomic_inc(&rcu_torture_wcount[i]);
	if (++rp->rtort_pipe_count >= RCU_TORTURE_PIPE_LEN) {
		rp->rtort_mbtest = 0;
		rcu_torture_free(rp);
294
	} else {
295
		cur_ops->deferred_free(rp);
296
	}
297 298
}

299 300 301 302 303
static int rcu_no_completed(void)
{
	return 0;
}

304 305 306 307 308
static void rcu_torture_deferred_free(struct rcu_torture *p)
{
	call_rcu(&p->rtort_rcu, rcu_torture_cb);
}

309 310 311 312 313
static void rcu_sync_torture_init(void)
{
	INIT_LIST_HEAD(&rcu_torture_removed);
}

314
static struct rcu_torture_ops rcu_ops = {
315 316 317 318 319
	.init		= rcu_sync_torture_init,
	.readlock	= rcu_torture_read_lock,
	.read_delay	= rcu_read_delay,
	.readunlock	= rcu_torture_read_unlock,
	.completed	= rcu_torture_completed,
320
	.deferred_free	= rcu_torture_deferred_free,
321
	.sync		= synchronize_rcu,
322 323 324
	.exp_sync	= synchronize_rcu_expedited,
	.call		= call_rcu,
	.cb_barrier	= rcu_barrier,
325
	.fqs		= rcu_force_quiescent_state,
326 327
	.stats		= NULL,
	.irq_capable	= 1,
328
	.can_boost	= rcu_can_boost(),
329
	.name		= "rcu"
330 331
};

332 333 334 335
/*
 * Definitions for rcu_bh torture testing.
 */

336
static int rcu_bh_torture_read_lock(void) __acquires(RCU_BH)
337 338 339 340 341
{
	rcu_read_lock_bh();
	return 0;
}

342
static void rcu_bh_torture_read_unlock(int idx) __releases(RCU_BH)
343 344 345 346 347 348 349 350 351 352 353 354 355 356 357
{
	rcu_read_unlock_bh();
}

static int rcu_bh_torture_completed(void)
{
	return rcu_batches_completed_bh();
}

static void rcu_bh_torture_deferred_free(struct rcu_torture *p)
{
	call_rcu_bh(&p->rtort_rcu, rcu_torture_cb);
}

static struct rcu_torture_ops rcu_bh_ops = {
358
	.init		= rcu_sync_torture_init,
359 360 361 362 363
	.readlock	= rcu_bh_torture_read_lock,
	.read_delay	= rcu_read_delay,  /* just reuse rcu's version. */
	.readunlock	= rcu_bh_torture_read_unlock,
	.completed	= rcu_bh_torture_completed,
	.deferred_free	= rcu_bh_torture_deferred_free,
364
	.sync		= synchronize_rcu_bh,
365
	.exp_sync	= synchronize_rcu_bh_expedited,
366
	.call		= call_rcu_bh,
367
	.cb_barrier	= rcu_barrier_bh,
368
	.fqs		= rcu_bh_force_quiescent_state,
369 370 371
	.stats		= NULL,
	.irq_capable	= 1,
	.name		= "rcu_bh"
372 373
};

374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415
/*
 * Don't even think about trying any of these in real life!!!
 * The names includes "busted", and they really means it!
 * The only purpose of these functions is to provide a buggy RCU
 * implementation to make sure that rcutorture correctly emits
 * buggy-RCU error messages.
 */
static void rcu_busted_torture_deferred_free(struct rcu_torture *p)
{
	/* This is a deliberate bug for testing purposes only! */
	rcu_torture_cb(&p->rtort_rcu);
}

static void synchronize_rcu_busted(void)
{
	/* This is a deliberate bug for testing purposes only! */
}

static void
call_rcu_busted(struct rcu_head *head, void (*func)(struct rcu_head *rcu))
{
	/* This is a deliberate bug for testing purposes only! */
	func(head);
}

static struct rcu_torture_ops rcu_busted_ops = {
	.init		= rcu_sync_torture_init,
	.readlock	= rcu_torture_read_lock,
	.read_delay	= rcu_read_delay,  /* just reuse rcu's version. */
	.readunlock	= rcu_torture_read_unlock,
	.completed	= rcu_no_completed,
	.deferred_free	= rcu_busted_torture_deferred_free,
	.sync		= synchronize_rcu_busted,
	.exp_sync	= synchronize_rcu_busted,
	.call		= call_rcu_busted,
	.cb_barrier	= NULL,
	.fqs		= NULL,
	.stats		= NULL,
	.irq_capable	= 1,
	.name		= "rcu_busted"
};

416 417 418 419
/*
 * Definitions for srcu torture testing.
 */

420
DEFINE_STATIC_SRCU(srcu_ctl);
421

422
static int srcu_torture_read_lock(void) __acquires(&srcu_ctl)
423 424 425 426
{
	return srcu_read_lock(&srcu_ctl);
}

427
static void srcu_read_delay(struct torture_random_state *rrsp)
428 429 430 431 432 433 434
{
	long delay;
	const long uspertick = 1000000 / HZ;
	const long longdelay = 10;

	/* We want there to be long-running readers, but not all the time. */

435 436
	delay = torture_random(rrsp) %
		(nrealreaders * 2 * longdelay * uspertick);
437 438
	if (!delay)
		schedule_timeout_interruptible(longdelay);
439 440
	else
		rcu_read_delay(rrsp);
441 442
}

443
static void srcu_torture_read_unlock(int idx) __releases(&srcu_ctl)
444 445 446 447 448 449 450 451 452
{
	srcu_read_unlock(&srcu_ctl, idx);
}

static int srcu_torture_completed(void)
{
	return srcu_batches_completed(&srcu_ctl);
}

453 454 455 456 457
static void srcu_torture_deferred_free(struct rcu_torture *rp)
{
	call_srcu(&srcu_ctl, &rp->rtort_rcu, rcu_torture_cb);
}

458 459 460 461 462
static void srcu_torture_synchronize(void)
{
	synchronize_srcu(&srcu_ctl);
}

463 464 465 466 467 468 469 470 471 472 473
static void srcu_torture_call(struct rcu_head *head,
			      void (*func)(struct rcu_head *head))
{
	call_srcu(&srcu_ctl, head, func);
}

static void srcu_torture_barrier(void)
{
	srcu_barrier(&srcu_ctl);
}

474
static void srcu_torture_stats(char *page)
475 476 477 478
{
	int cpu;
	int idx = srcu_ctl.completed & 0x1;

479
	page += sprintf(page, "%s%s per-CPU(idx=%d):",
480 481
		       torture_type, TORTURE_FLAG, idx);
	for_each_possible_cpu(cpu) {
482
		page += sprintf(page, " %d(%lu,%lu)", cpu,
483 484 485
			       per_cpu_ptr(srcu_ctl.per_cpu_ref, cpu)->c[!idx],
			       per_cpu_ptr(srcu_ctl.per_cpu_ref, cpu)->c[idx]);
	}
486
	sprintf(page, "\n");
487 488
}

489 490 491 492 493
static void srcu_torture_synchronize_expedited(void)
{
	synchronize_srcu_expedited(&srcu_ctl);
}

494
static struct rcu_torture_ops srcu_ops = {
495
	.init		= rcu_sync_torture_init,
496 497 498 499
	.readlock	= srcu_torture_read_lock,
	.read_delay	= srcu_read_delay,
	.readunlock	= srcu_torture_read_unlock,
	.completed	= srcu_torture_completed,
500
	.deferred_free	= srcu_torture_deferred_free,
501
	.sync		= srcu_torture_synchronize,
502
	.exp_sync	= srcu_torture_synchronize_expedited,
503 504
	.call		= srcu_torture_call,
	.cb_barrier	= srcu_torture_barrier,
505 506
	.stats		= srcu_torture_stats,
	.name		= "srcu"
507 508
};

509 510 511 512 513 514 515 516 517 518 519 520 521 522 523
/*
 * Definitions for sched torture testing.
 */

static int sched_torture_read_lock(void)
{
	preempt_disable();
	return 0;
}

static void sched_torture_read_unlock(int idx)
{
	preempt_enable();
}

524 525 526 527 528
static void rcu_sched_torture_deferred_free(struct rcu_torture *p)
{
	call_rcu_sched(&p->rtort_rcu, rcu_torture_cb);
}

529
static struct rcu_torture_ops sched_ops = {
530 531 532 533
	.init		= rcu_sync_torture_init,
	.readlock	= sched_torture_read_lock,
	.read_delay	= rcu_read_delay,  /* just reuse rcu's version. */
	.readunlock	= sched_torture_read_unlock,
534
	.completed	= rcu_no_completed,
535
	.deferred_free	= rcu_sched_torture_deferred_free,
536
	.sync		= synchronize_sched,
537 538
	.exp_sync	= synchronize_sched_expedited,
	.call		= call_rcu_sched,
539
	.cb_barrier	= rcu_barrier_sched,
540
	.fqs		= rcu_sched_force_quiescent_state,
541 542 543
	.stats		= NULL,
	.irq_capable	= 1,
	.name		= "sched"
544 545
};

546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574
/*
 * RCU torture priority-boost testing.  Runs one real-time thread per
 * CPU for moderate bursts, repeatedly registering RCU callbacks and
 * spinning waiting for them to be invoked.  If a given callback takes
 * too long to be invoked, we assume that priority inversion has occurred.
 */

struct rcu_boost_inflight {
	struct rcu_head rcu;
	int inflight;
};

static void rcu_torture_boost_cb(struct rcu_head *head)
{
	struct rcu_boost_inflight *rbip =
		container_of(head, struct rcu_boost_inflight, rcu);

	smp_mb(); /* Ensure RCU-core accesses precede clearing ->inflight */
	rbip->inflight = 0;
}

static int rcu_torture_boost(void *arg)
{
	unsigned long call_rcu_time;
	unsigned long endtime;
	unsigned long oldstarttime;
	struct rcu_boost_inflight rbi = { .inflight = 0 };
	struct sched_param sp;

575
	VERBOSE_TOROUT_STRING("rcu_torture_boost started");
576 577 578 579

	/* Set real-time priority. */
	sp.sched_priority = 1;
	if (sched_setscheduler(current, SCHED_FIFO, &sp) < 0) {
580
		VERBOSE_TOROUT_STRING("rcu_torture_boost RT prio failed!");
581 582 583
		n_rcu_torture_boost_rterror++;
	}

584
	init_rcu_head_on_stack(&rbi.rcu);
585 586 587 588
	/* Each pass through the following loop does one boost-test cycle. */
	do {
		/* Wait for the next test interval. */
		oldstarttime = boost_starttime;
589
		while (ULONG_CMP_LT(jiffies, oldstarttime)) {
590
			schedule_timeout_interruptible(oldstarttime - jiffies);
591
			stutter_wait("rcu_torture_boost");
592
			if (torture_must_stop())
593 594 595 596 597 598
				goto checkwait;
		}

		/* Do one boost-test interval. */
		endtime = oldstarttime + test_boost_duration * HZ;
		call_rcu_time = jiffies;
599
		while (ULONG_CMP_LT(jiffies, endtime)) {
600 601 602 603 604 605 606
			/* If we don't have a callback in flight, post one. */
			if (!rbi.inflight) {
				smp_mb(); /* RCU core before ->inflight = 1. */
				rbi.inflight = 1;
				call_rcu(&rbi.rcu, rcu_torture_boost_cb);
				if (jiffies - call_rcu_time >
					 test_boost_duration * HZ - HZ / 2) {
607
					VERBOSE_TOROUT_STRING("rcu_torture_boost boosting failed");
608 609 610 611 612
					n_rcu_torture_boost_failure++;
				}
				call_rcu_time = jiffies;
			}
			cond_resched();
613
			stutter_wait("rcu_torture_boost");
614
			if (torture_must_stop())
615 616 617 618 619 620 621 622 623 624
				goto checkwait;
		}

		/*
		 * Set the start time of the next test interval.
		 * Yes, this is vulnerable to long delays, but such
		 * delays simply cause a false negative for the next
		 * interval.  Besides, we are running at RT priority,
		 * so delays should be relatively rare.
		 */
625 626
		while (oldstarttime == boost_starttime &&
		       !kthread_should_stop()) {
627 628 629 630 631 632 633 634 635 636 637
			if (mutex_trylock(&boost_mutex)) {
				boost_starttime = jiffies +
						  test_boost_interval * HZ;
				n_rcu_torture_boosts++;
				mutex_unlock(&boost_mutex);
				break;
			}
			schedule_timeout_uninterruptible(1);
		}

		/* Go do the stutter. */
638
checkwait:	stutter_wait("rcu_torture_boost");
639
	} while (!torture_must_stop());
640 641

	/* Clean up and exit. */
642 643
	while (!kthread_should_stop() || rbi.inflight) {
		torture_shutdown_absorb("rcu_torture_boost");
644
		schedule_timeout_uninterruptible(1);
645
	}
646
	smp_mb(); /* order accesses to ->inflight before stack-frame death. */
647
	destroy_rcu_head_on_stack(&rbi.rcu);
648
	torture_kthread_stopping("rcu_torture_boost");
649 650 651
	return 0;
}

652 653 654 655 656 657 658 659 660 661 662
/*
 * RCU torture force-quiescent-state kthread.  Repeatedly induces
 * bursts of calls to force_quiescent_state(), increasing the probability
 * of occurrence of some important types of race conditions.
 */
static int
rcu_torture_fqs(void *arg)
{
	unsigned long fqs_resume_time;
	int fqs_burst_remaining;

663
	VERBOSE_TOROUT_STRING("rcu_torture_fqs task started");
664 665
	do {
		fqs_resume_time = jiffies + fqs_stutter * HZ;
666 667
		while (ULONG_CMP_LT(jiffies, fqs_resume_time) &&
		       !kthread_should_stop()) {
668 669 670
			schedule_timeout_interruptible(1);
		}
		fqs_burst_remaining = fqs_duration;
671 672
		while (fqs_burst_remaining > 0 &&
		       !kthread_should_stop()) {
673 674 675 676
			cur_ops->fqs();
			udelay(fqs_holdoff);
			fqs_burst_remaining -= fqs_holdoff;
		}
677
		stutter_wait("rcu_torture_fqs");
678
	} while (!torture_must_stop());
679
	torture_kthread_stopping("rcu_torture_fqs");
680 681 682
	return 0;
}

683 684 685 686 687 688 689 690
/*
 * RCU torture writer kthread.  Repeatedly substitutes a new structure
 * for that pointed to by rcu_torture_current, freeing the old structure
 * after a series of grace periods (the "pipeline").
 */
static int
rcu_torture_writer(void *arg)
{
691
	bool exp;
692 693
	int i;
	struct rcu_torture *rp;
694
	struct rcu_torture *rp1;
695
	struct rcu_torture *old_rp;
696
	static DEFINE_TORTURE_RANDOM(rand);
697

698
	VERBOSE_TOROUT_STRING("rcu_torture_writer task started");
699
	set_user_nice(current, MAX_NICE);
700

701 702
	do {
		schedule_timeout_uninterruptible(1);
703 704
		rp = rcu_torture_alloc();
		if (rp == NULL)
705 706
			continue;
		rp->rtort_pipe_count = 0;
707
		udelay(torture_random(&rand) & 0x3ff);
708 709
		old_rp = rcu_dereference_check(rcu_torture_current,
					       current == writer_task);
710
		rp->rtort_mbtest = 1;
711
		rcu_assign_pointer(rcu_torture_current, rp);
712
		smp_wmb(); /* Mods to old_rp must follow rcu_assign_pointer() */
713
		if (old_rp) {
714 715 716 717 718
			i = old_rp->rtort_pipe_count;
			if (i > RCU_TORTURE_PIPE_LEN)
				i = RCU_TORTURE_PIPE_LEN;
			atomic_inc(&rcu_torture_wcount[i]);
			old_rp->rtort_pipe_count++;
719
			if (gp_normal == gp_exp)
720
				exp = !!(torture_random(&rand) & 0x80);
721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743
			else
				exp = gp_exp;
			if (!exp) {
				cur_ops->deferred_free(old_rp);
			} else {
				cur_ops->exp_sync();
				list_add(&old_rp->rtort_free,
					 &rcu_torture_removed);
				list_for_each_entry_safe(rp, rp1,
							 &rcu_torture_removed,
							 rtort_free) {
					i = rp->rtort_pipe_count;
					if (i > RCU_TORTURE_PIPE_LEN)
						i = RCU_TORTURE_PIPE_LEN;
					atomic_inc(&rcu_torture_wcount[i]);
					if (++rp->rtort_pipe_count >=
					    RCU_TORTURE_PIPE_LEN) {
						rp->rtort_mbtest = 0;
						list_del(&rp->rtort_free);
						rcu_torture_free(rp);
					}
				 }
			}
744
		}
745
		rcutorture_record_progress(++rcu_torture_current_version);
746
		stutter_wait("rcu_torture_writer");
747
	} while (!torture_must_stop());
748
	torture_kthread_stopping("rcu_torture_writer");
749 750 751
	return 0;
}

752 753 754 755 756 757 758
/*
 * RCU torture fake writer kthread.  Repeatedly calls sync, with a random
 * delay between calls.
 */
static int
rcu_torture_fakewriter(void *arg)
{
759
	DEFINE_TORTURE_RANDOM(rand);
760

761
	VERBOSE_TOROUT_STRING("rcu_torture_fakewriter task started");
762
	set_user_nice(current, MAX_NICE);
763 764

	do {
765 766
		schedule_timeout_uninterruptible(1 + torture_random(&rand)%10);
		udelay(torture_random(&rand) & 0x3ff);
767
		if (cur_ops->cb_barrier != NULL &&
768
		    torture_random(&rand) % (nfakewriters * 8) == 0) {
769
			cur_ops->cb_barrier();
770
		} else if (gp_normal == gp_exp) {
771
			if (torture_random(&rand) & 0x80)
772 773 774 775
				cur_ops->sync();
			else
				cur_ops->exp_sync();
		} else if (gp_normal) {
776
			cur_ops->sync();
777 778 779
		} else {
			cur_ops->exp_sync();
		}
780
		stutter_wait("rcu_torture_fakewriter");
781
	} while (!torture_must_stop());
782

783
	torture_kthread_stopping("rcu_torture_fakewriter");
784 785 786
	return 0;
}

787 788 789 790 791 792 793 794 795 796 797
void rcutorture_trace_dump(void)
{
	static atomic_t beenhere = ATOMIC_INIT(0);

	if (atomic_read(&beenhere))
		return;
	if (atomic_xchg(&beenhere, 1) != 0)
		return;
	ftrace_dump(DUMP_ALL);
}

798 799 800 801 802 803 804 805 806 807
/*
 * RCU torture reader from timer handler.  Dereferences rcu_torture_current,
 * incrementing the corresponding element of the pipeline array.  The
 * counter in the element should never be greater than 1, otherwise, the
 * RCU implementation is broken.
 */
static void rcu_torture_timer(unsigned long unused)
{
	int idx;
	int completed;
808
	int completed_end;
809
	static DEFINE_TORTURE_RANDOM(rand);
810 811 812
	static DEFINE_SPINLOCK(rand_lock);
	struct rcu_torture *p;
	int pipe_count;
813
	unsigned long long ts;
814 815 816

	idx = cur_ops->readlock();
	completed = cur_ops->completed();
817
	ts = rcu_trace_clock_local();
818 819 820 821
	p = rcu_dereference_check(rcu_torture_current,
				  rcu_read_lock_bh_held() ||
				  rcu_read_lock_sched_held() ||
				  srcu_read_lock_held(&srcu_ctl));
822 823 824 825 826 827 828 829
	if (p == NULL) {
		/* Leave because rcu_torture_writer is not yet underway */
		cur_ops->readunlock(idx);
		return;
	}
	if (p->rtort_mbtest == 0)
		atomic_inc(&n_rcu_torture_mberror);
	spin_lock(&rand_lock);
830
	cur_ops->read_delay(&rand);
831 832 833 834 835 836 837 838
	n_rcu_torture_timers++;
	spin_unlock(&rand_lock);
	preempt_disable();
	pipe_count = p->rtort_pipe_count;
	if (pipe_count > RCU_TORTURE_PIPE_LEN) {
		/* Should not happen, but... */
		pipe_count = RCU_TORTURE_PIPE_LEN;
	}
839 840 841 842
	completed_end = cur_ops->completed();
	if (pipe_count > 1) {
		do_trace_rcu_torture_read(cur_ops->name, &p->rtort_rcu, ts,
					  completed, completed_end);
843
		rcutorture_trace_dump();
844
	}
R
Rusty Russell 已提交
845
	__this_cpu_inc(rcu_torture_count[pipe_count]);
846
	completed = completed_end - completed;
847 848 849 850
	if (completed > RCU_TORTURE_PIPE_LEN) {
		/* Should not happen, but... */
		completed = RCU_TORTURE_PIPE_LEN;
	}
R
Rusty Russell 已提交
851
	__this_cpu_inc(rcu_torture_batch[completed]);
852 853 854 855
	preempt_enable();
	cur_ops->readunlock(idx);
}

856 857 858 859 860 861 862 863 864 865
/*
 * RCU torture reader kthread.  Repeatedly dereferences rcu_torture_current,
 * incrementing the corresponding element of the pipeline array.  The
 * counter in the element should never be greater than 1, otherwise, the
 * RCU implementation is broken.
 */
static int
rcu_torture_reader(void *arg)
{
	int completed;
866
	int completed_end;
867
	int idx;
868
	DEFINE_TORTURE_RANDOM(rand);
869 870
	struct rcu_torture *p;
	int pipe_count;
871
	struct timer_list t;
872
	unsigned long long ts;
873

874
	VERBOSE_TOROUT_STRING("rcu_torture_reader task started");
875
	set_user_nice(current, MAX_NICE);
876
	if (irqreader && cur_ops->irq_capable)
877
		setup_timer_on_stack(&t, rcu_torture_timer, 0);
878

879
	do {
880
		if (irqreader && cur_ops->irq_capable) {
881
			if (!timer_pending(&t))
882
				mod_timer(&t, jiffies + 1);
883
		}
884 885
		idx = cur_ops->readlock();
		completed = cur_ops->completed();
886
		ts = rcu_trace_clock_local();
887 888 889 890
		p = rcu_dereference_check(rcu_torture_current,
					  rcu_read_lock_bh_held() ||
					  rcu_read_lock_sched_held() ||
					  srcu_read_lock_held(&srcu_ctl));
891 892
		if (p == NULL) {
			/* Wait for rcu_torture_writer to get underway */
893
			cur_ops->readunlock(idx);
894 895 896
			schedule_timeout_interruptible(HZ);
			continue;
		}
897 898
		if (p->rtort_mbtest == 0)
			atomic_inc(&n_rcu_torture_mberror);
899
		cur_ops->read_delay(&rand);
900 901 902 903 904 905
		preempt_disable();
		pipe_count = p->rtort_pipe_count;
		if (pipe_count > RCU_TORTURE_PIPE_LEN) {
			/* Should not happen, but... */
			pipe_count = RCU_TORTURE_PIPE_LEN;
		}
906 907 908 909
		completed_end = cur_ops->completed();
		if (pipe_count > 1) {
			do_trace_rcu_torture_read(cur_ops->name, &p->rtort_rcu,
						  ts, completed, completed_end);
910
			rcutorture_trace_dump();
911
		}
R
Rusty Russell 已提交
912
		__this_cpu_inc(rcu_torture_count[pipe_count]);
913
		completed = completed_end - completed;
914 915 916 917
		if (completed > RCU_TORTURE_PIPE_LEN) {
			/* Should not happen, but... */
			completed = RCU_TORTURE_PIPE_LEN;
		}
R
Rusty Russell 已提交
918
		__this_cpu_inc(rcu_torture_batch[completed]);
919
		preempt_enable();
920
		cur_ops->readunlock(idx);
921
		schedule();
922
		stutter_wait("rcu_torture_reader");
923
	} while (!torture_must_stop());
924
	if (irqreader && cur_ops->irq_capable)
925
		del_timer_sync(&t);
926
	torture_kthread_stopping("rcu_torture_reader");
927 928 929 930 931 932
	return 0;
}

/*
 * Create an RCU-torture statistics message in the specified buffer.
 */
933
static void
934 935 936 937 938 939 940
rcu_torture_printk(char *page)
{
	int cpu;
	int i;
	long pipesummary[RCU_TORTURE_PIPE_LEN + 1] = { 0 };
	long batchsummary[RCU_TORTURE_PIPE_LEN + 1] = { 0 };

941
	for_each_possible_cpu(cpu) {
942 943 944 945 946 947 948 949 950
		for (i = 0; i < RCU_TORTURE_PIPE_LEN + 1; i++) {
			pipesummary[i] += per_cpu(rcu_torture_count, cpu)[i];
			batchsummary[i] += per_cpu(rcu_torture_batch, cpu)[i];
		}
	}
	for (i = RCU_TORTURE_PIPE_LEN - 1; i >= 0; i--) {
		if (pipesummary[i] != 0)
			break;
	}
951 952
	page += sprintf(page, "%s%s ", torture_type, TORTURE_FLAG);
	page += sprintf(page,
953
		       "rtc: %p ver: %lu tfle: %d rta: %d rtaf: %d rtf: %d ",
954 955 956 957 958
		       rcu_torture_current,
		       rcu_torture_current_version,
		       list_empty(&rcu_torture_freelist),
		       atomic_read(&n_rcu_torture_alloc),
		       atomic_read(&n_rcu_torture_alloc_fail),
959
		       atomic_read(&n_rcu_torture_free));
960
	page += sprintf(page, "rtmbe: %d rtbke: %ld rtbre: %ld ",
961
		       atomic_read(&n_rcu_torture_mberror),
962
		       n_rcu_torture_boost_ktrerror,
963
		       n_rcu_torture_boost_rterror);
964
	page += sprintf(page, "rtbf: %ld rtb: %ld nt: %ld ",
965 966
		       n_rcu_torture_boost_failure,
		       n_rcu_torture_boosts,
967
		       n_rcu_torture_timers);
968
	page = torture_onoff_stats(page);
969
	page += sprintf(page, "barrier: %ld/%ld:%ld",
970 971 972
		       n_barrier_successes,
		       n_barrier_attempts,
		       n_rcu_torture_barrier_error);
973
	page += sprintf(page, "\n%s%s ", torture_type, TORTURE_FLAG);
974
	if (atomic_read(&n_rcu_torture_mberror) != 0 ||
975
	    n_rcu_torture_barrier_error != 0 ||
976 977
	    n_rcu_torture_boost_ktrerror != 0 ||
	    n_rcu_torture_boost_rterror != 0 ||
978 979
	    n_rcu_torture_boost_failure != 0 ||
	    i > 1) {
980
		page += sprintf(page, "!!! ");
981
		atomic_inc(&n_rcu_torture_error);
982
		WARN_ON_ONCE(1);
983
	}
984
	page += sprintf(page, "Reader Pipe: ");
985
	for (i = 0; i < RCU_TORTURE_PIPE_LEN + 1; i++)
986 987 988
		page += sprintf(page, " %ld", pipesummary[i]);
	page += sprintf(page, "\n%s%s ", torture_type, TORTURE_FLAG);
	page += sprintf(page, "Reader Batch: ");
989
	for (i = 0; i < RCU_TORTURE_PIPE_LEN + 1; i++)
990 991 992
		page += sprintf(page, " %ld", batchsummary[i]);
	page += sprintf(page, "\n%s%s ", torture_type, TORTURE_FLAG);
	page += sprintf(page, "Free-Block Circulation: ");
993
	for (i = 0; i < RCU_TORTURE_PIPE_LEN + 1; i++) {
994
		page += sprintf(page, " %d",
995 996
			       atomic_read(&rcu_torture_wcount[i]));
	}
997
	page += sprintf(page, "\n");
998
	if (cur_ops->stats)
999
		cur_ops->stats(page);
1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012
}

/*
 * Print torture statistics.  Caller must ensure that there is only
 * one call to this function at a given time!!!  This is normally
 * accomplished by relying on the module system to only have one copy
 * of the module loaded, and then by giving the rcu_torture_stats
 * kthread full control (or the init/cleanup functions when rcu_torture_stats
 * thread is not running).
 */
static void
rcu_torture_stats_print(void)
{
1013 1014
	int size = nr_cpu_ids * 200 + 8192;
	char *buf;
1015

1016 1017 1018 1019 1020 1021 1022 1023
	buf = kmalloc(size, GFP_KERNEL);
	if (!buf) {
		pr_err("rcu-torture: Out of memory, need: %d", size);
		return;
	}
	rcu_torture_printk(buf);
	pr_alert("%s", buf);
	kfree(buf);
1024 1025 1026 1027 1028 1029 1030 1031 1032
}

/*
 * Periodically prints torture statistics, if periodic statistics printing
 * was specified via the stat_interval module parameter.
 */
static int
rcu_torture_stats(void *arg)
{
1033
	VERBOSE_TOROUT_STRING("rcu_torture_stats task started");
1034 1035 1036
	do {
		schedule_timeout_interruptible(stat_interval * HZ);
		rcu_torture_stats_print();
1037
		torture_shutdown_absorb("rcu_torture_stats");
1038
	} while (!torture_must_stop());
1039
	torture_kthread_stopping("rcu_torture_stats");
1040 1041 1042
	return 0;
}

1043
static inline void
1044
rcu_torture_print_module_parms(struct rcu_torture_ops *cur_ops, const char *tag)
1045
{
1046 1047 1048 1049 1050 1051 1052
	pr_alert("%s" TORTURE_FLAG
		 "--- %s: nreaders=%d nfakewriters=%d "
		 "stat_interval=%d verbose=%d test_no_idle_hz=%d "
		 "shuffle_interval=%d stutter=%d irqreader=%d "
		 "fqs_duration=%d fqs_holdoff=%d fqs_stutter=%d "
		 "test_boost=%d/%d test_boost_interval=%d "
		 "test_boost_duration=%d shutdown_secs=%d "
1053 1054
		 "stall_cpu=%d stall_cpu_holdoff=%d "
		 "n_barrier_cbs=%d "
1055 1056 1057 1058 1059 1060
		 "onoff_interval=%d onoff_holdoff=%d\n",
		 torture_type, tag, nrealreaders, nfakewriters,
		 stat_interval, verbose, test_no_idle_hz, shuffle_interval,
		 stutter, irqreader, fqs_duration, fqs_holdoff, fqs_stutter,
		 test_boost, cur_ops->can_boost,
		 test_boost_interval, test_boost_duration, shutdown_secs,
1061 1062
		 stall_cpu, stall_cpu_holdoff,
		 n_barrier_cbs,
1063
		 onoff_interval, onoff_holdoff);
1064 1065
}

1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077
static void rcutorture_booster_cleanup(int cpu)
{
	struct task_struct *t;

	if (boost_tasks[cpu] == NULL)
		return;
	mutex_lock(&boost_mutex);
	t = boost_tasks[cpu];
	boost_tasks[cpu] = NULL;
	mutex_unlock(&boost_mutex);

	/* This must be outside of the mutex, otherwise deadlock! */
1078
	torture_stop_kthread(rcu_torture_boost, t);
1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089
}

static int rcutorture_booster_init(int cpu)
{
	int retval;

	if (boost_tasks[cpu] != NULL)
		return 0;  /* Already created, nothing more to do. */

	/* Don't allow time recalculation while creating a new task. */
	mutex_lock(&boost_mutex);
1090
	VERBOSE_TOROUT_STRING("Creating rcu_torture_boost task");
E
Eric Dumazet 已提交
1091 1092 1093
	boost_tasks[cpu] = kthread_create_on_node(rcu_torture_boost, NULL,
						  cpu_to_node(cpu),
						  "rcu_torture_boost");
1094 1095
	if (IS_ERR(boost_tasks[cpu])) {
		retval = PTR_ERR(boost_tasks[cpu]);
1096
		VERBOSE_TOROUT_STRING("rcu_torture_boost task create failed");
1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107
		n_rcu_torture_boost_ktrerror++;
		boost_tasks[cpu] = NULL;
		mutex_unlock(&boost_mutex);
		return retval;
	}
	kthread_bind(boost_tasks[cpu], cpu);
	wake_up_process(boost_tasks[cpu]);
	mutex_unlock(&boost_mutex);
	return 0;
}

1108 1109 1110 1111
/*
 * CPU-stall kthread.  It waits as specified by stall_cpu_holdoff, then
 * induces a CPU stall for the time specified by stall_cpu.
 */
1112
static int rcu_torture_stall(void *args)
1113 1114 1115
{
	unsigned long stop_at;

1116
	VERBOSE_TOROUT_STRING("rcu_torture_stall task started");
1117
	if (stall_cpu_holdoff > 0) {
1118
		VERBOSE_TOROUT_STRING("rcu_torture_stall begin holdoff");
1119
		schedule_timeout_interruptible(stall_cpu_holdoff * HZ);
1120
		VERBOSE_TOROUT_STRING("rcu_torture_stall end holdoff");
1121 1122 1123 1124
	}
	if (!kthread_should_stop()) {
		stop_at = get_seconds() + stall_cpu;
		/* RCU CPU stall is expected behavior in following code. */
1125
		pr_alert("rcu_torture_stall start.\n");
1126 1127 1128 1129 1130 1131
		rcu_read_lock();
		preempt_disable();
		while (ULONG_CMP_LT(get_seconds(), stop_at))
			continue;  /* Induce RCU CPU stall warning. */
		preempt_enable();
		rcu_read_unlock();
1132
		pr_alert("rcu_torture_stall end.\n");
1133
	}
1134
	torture_shutdown_absorb("rcu_torture_stall");
1135 1136 1137 1138 1139 1140 1141 1142 1143 1144
	while (!kthread_should_stop())
		schedule_timeout_interruptible(10 * HZ);
	return 0;
}

/* Spawn CPU-stall kthread, if stall_cpu specified. */
static int __init rcu_torture_stall_init(void)
{
	if (stall_cpu <= 0)
		return 0;
1145
	return torture_create_kthread(rcu_torture_stall, NULL, stall_task);
1146 1147
}

1148 1149 1150 1151 1152 1153 1154 1155 1156 1157
/* Callback function for RCU barrier testing. */
void rcu_torture_barrier_cbf(struct rcu_head *rcu)
{
	atomic_inc(&barrier_cbs_invoked);
}

/* kthread function to register callbacks used to test RCU barriers. */
static int rcu_torture_barrier_cbs(void *arg)
{
	long myid = (long)arg;
1158
	bool lastphase = 0;
1159
	bool newphase;
1160 1161 1162
	struct rcu_head rcu;

	init_rcu_head_on_stack(&rcu);
1163
	VERBOSE_TOROUT_STRING("rcu_torture_barrier_cbs task started");
1164
	set_user_nice(current, MAX_NICE);
1165 1166
	do {
		wait_event(barrier_cbs_wq[myid],
1167 1168
			   (newphase =
			    ACCESS_ONCE(barrier_phase)) != lastphase ||
1169
			   torture_must_stop());
1170
		lastphase = newphase;
1171
		smp_mb(); /* ensure barrier_phase load before ->call(). */
1172
		if (torture_must_stop())
1173 1174 1175 1176
			break;
		cur_ops->call(&rcu, rcu_torture_barrier_cbf);
		if (atomic_dec_and_test(&barrier_cbs_count))
			wake_up(&barrier_wq);
1177
	} while (!torture_must_stop());
1178 1179
	cur_ops->cb_barrier();
	destroy_rcu_head_on_stack(&rcu);
1180
	torture_kthread_stopping("rcu_torture_barrier_cbs");
1181 1182 1183 1184 1185 1186 1187 1188
	return 0;
}

/* kthread function to drive and coordinate RCU barrier testing. */
static int rcu_torture_barrier(void *arg)
{
	int i;

1189
	VERBOSE_TOROUT_STRING("rcu_torture_barrier task starting");
1190 1191 1192
	do {
		atomic_set(&barrier_cbs_invoked, 0);
		atomic_set(&barrier_cbs_count, n_barrier_cbs);
1193 1194
		smp_mb(); /* Ensure barrier_phase after prior assignments. */
		barrier_phase = !barrier_phase;
1195 1196 1197 1198
		for (i = 0; i < n_barrier_cbs; i++)
			wake_up(&barrier_cbs_wq[i]);
		wait_event(barrier_wq,
			   atomic_read(&barrier_cbs_count) == 0 ||
1199 1200
			   torture_must_stop());
		if (torture_must_stop())
1201 1202
			break;
		n_barrier_attempts++;
1203
		cur_ops->cb_barrier(); /* Implies smp_mb() for wait_event(). */
1204 1205 1206 1207 1208 1209
		if (atomic_read(&barrier_cbs_invoked) != n_barrier_cbs) {
			n_rcu_torture_barrier_error++;
			WARN_ON_ONCE(1);
		}
		n_barrier_successes++;
		schedule_timeout_interruptible(HZ / 10);
1210
	} while (!torture_must_stop());
1211
	torture_kthread_stopping("rcu_torture_barrier");
1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223
	return 0;
}

/* Initialize RCU barrier testing. */
static int rcu_torture_barrier_init(void)
{
	int i;
	int ret;

	if (n_barrier_cbs == 0)
		return 0;
	if (cur_ops->call == NULL || cur_ops->cb_barrier == NULL) {
1224 1225 1226 1227 1228 1229
		pr_alert("%s" TORTURE_FLAG
			 " Call or barrier ops missing for %s,\n",
			 torture_type, cur_ops->name);
		pr_alert("%s" TORTURE_FLAG
			 " RCU barrier testing omitted from run.\n",
			 torture_type);
1230 1231 1232 1233 1234 1235 1236 1237 1238 1239
		return 0;
	}
	atomic_set(&barrier_cbs_count, 0);
	atomic_set(&barrier_cbs_invoked, 0);
	barrier_cbs_tasks =
		kzalloc(n_barrier_cbs * sizeof(barrier_cbs_tasks[0]),
			GFP_KERNEL);
	barrier_cbs_wq =
		kzalloc(n_barrier_cbs * sizeof(barrier_cbs_wq[0]),
			GFP_KERNEL);
1240
	if (barrier_cbs_tasks == NULL || !barrier_cbs_wq)
1241 1242 1243
		return -ENOMEM;
	for (i = 0; i < n_barrier_cbs; i++) {
		init_waitqueue_head(&barrier_cbs_wq[i]);
1244 1245 1246 1247
		ret = torture_create_kthread(rcu_torture_barrier_cbs,
					     (void *)(long)i,
					     barrier_cbs_tasks[i]);
		if (ret)
1248 1249
			return ret;
	}
1250
	return torture_create_kthread(rcu_torture_barrier, NULL, barrier_task);
1251 1252 1253 1254 1255 1256 1257
}

/* Clean up after RCU barrier testing. */
static void rcu_torture_barrier_cleanup(void)
{
	int i;

1258
	torture_stop_kthread(rcu_torture_barrier, barrier_task);
1259
	if (barrier_cbs_tasks != NULL) {
1260 1261 1262
		for (i = 0; i < n_barrier_cbs; i++)
			torture_stop_kthread(rcu_torture_barrier_cbs,
					     barrier_cbs_tasks[i]);
1263 1264 1265 1266 1267 1268 1269 1270 1271
		kfree(barrier_cbs_tasks);
		barrier_cbs_tasks = NULL;
	}
	if (barrier_cbs_wq != NULL) {
		kfree(barrier_cbs_wq);
		barrier_cbs_wq = NULL;
	}
}

1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294
static int rcutorture_cpu_notify(struct notifier_block *self,
				 unsigned long action, void *hcpu)
{
	long cpu = (long)hcpu;

	switch (action) {
	case CPU_ONLINE:
	case CPU_DOWN_FAILED:
		(void)rcutorture_booster_init(cpu);
		break;
	case CPU_DOWN_PREPARE:
		rcutorture_booster_cleanup(cpu);
		break;
	default:
		break;
	}
	return NOTIFY_OK;
}

static struct notifier_block rcutorture_cpu_nb = {
	.notifier_call = rcutorture_cpu_notify,
};

1295 1296 1297 1298 1299
static void
rcu_torture_cleanup(void)
{
	int i;

1300
	rcutorture_record_test_transition();
1301
	if (torture_cleanup()) {
1302 1303 1304 1305
		if (cur_ops->cb_barrier != NULL)
			cur_ops->cb_barrier();
		return;
	}
1306

1307
	rcu_torture_barrier_cleanup();
1308 1309
	torture_stop_kthread(rcu_torture_stall, stall_task);
	torture_stop_kthread(rcu_torture_writer, writer_task);
1310

1311
	if (reader_tasks) {
1312 1313 1314
		for (i = 0; i < nrealreaders; i++)
			torture_stop_kthread(rcu_torture_reader,
					     reader_tasks[i]);
1315 1316 1317 1318
		kfree(reader_tasks);
	}
	rcu_torture_current = NULL;

1319
	if (fakewriter_tasks) {
1320
		for (i = 0; i < nfakewriters; i++) {
1321 1322
			torture_stop_kthread(rcu_torture_fakewriter,
					     fakewriter_tasks[i]);
1323 1324 1325 1326 1327
		}
		kfree(fakewriter_tasks);
		fakewriter_tasks = NULL;
	}

1328 1329
	torture_stop_kthread(rcu_torture_stats, stats_task);
	torture_stop_kthread(rcu_torture_fqs, fqs_task);
1330 1331 1332 1333 1334 1335
	if ((test_boost == 1 && cur_ops->can_boost) ||
	    test_boost == 2) {
		unregister_cpu_notifier(&rcutorture_cpu_nb);
		for_each_possible_cpu(i)
			rcutorture_booster_cleanup(i);
	}
1336

1337
	/* Wait for all RCU callbacks to fire.  */
1338 1339 1340

	if (cur_ops->cb_barrier != NULL)
		cur_ops->cb_barrier();
1341 1342

	rcu_torture_stats_print();  /* -After- the stats thread is stopped! */
1343

1344
	if (atomic_read(&n_rcu_torture_error) || n_rcu_torture_barrier_error)
1345
		rcu_torture_print_module_parms(cur_ops, "End of test: FAILURE");
1346
	else if (torture_onoff_failures())
1347 1348
		rcu_torture_print_module_parms(cur_ops,
					       "End of test: RCU_HOTPLUG");
1349
	else
1350
		rcu_torture_print_module_parms(cur_ops, "End of test: SUCCESS");
1351 1352
}

1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408
#ifdef CONFIG_DEBUG_OBJECTS_RCU_HEAD
static void rcu_torture_leak_cb(struct rcu_head *rhp)
{
}

static void rcu_torture_err_cb(struct rcu_head *rhp)
{
	/*
	 * This -might- happen due to race conditions, but is unlikely.
	 * The scenario that leads to this happening is that the
	 * first of the pair of duplicate callbacks is queued,
	 * someone else starts a grace period that includes that
	 * callback, then the second of the pair must wait for the
	 * next grace period.  Unlikely, but can happen.  If it
	 * does happen, the debug-objects subsystem won't have splatted.
	 */
	pr_alert("rcutorture: duplicated callback was invoked.\n");
}
#endif /* #ifdef CONFIG_DEBUG_OBJECTS_RCU_HEAD */

/*
 * Verify that double-free causes debug-objects to complain, but only
 * if CONFIG_DEBUG_OBJECTS_RCU_HEAD=y.  Otherwise, say that the test
 * cannot be carried out.
 */
static void rcu_test_debug_objects(void)
{
#ifdef CONFIG_DEBUG_OBJECTS_RCU_HEAD
	struct rcu_head rh1;
	struct rcu_head rh2;

	init_rcu_head_on_stack(&rh1);
	init_rcu_head_on_stack(&rh2);
	pr_alert("rcutorture: WARN: Duplicate call_rcu() test starting.\n");

	/* Try to queue the rh2 pair of callbacks for the same grace period. */
	preempt_disable(); /* Prevent preemption from interrupting test. */
	rcu_read_lock(); /* Make it impossible to finish a grace period. */
	call_rcu(&rh1, rcu_torture_leak_cb); /* Start grace period. */
	local_irq_disable(); /* Make it harder to start a new grace period. */
	call_rcu(&rh2, rcu_torture_leak_cb);
	call_rcu(&rh2, rcu_torture_err_cb); /* Duplicate callback. */
	local_irq_enable();
	rcu_read_unlock();
	preempt_enable();

	/* Wait for them all to get done so we can safely return. */
	rcu_barrier();
	pr_alert("rcutorture: WARN: Duplicate call_rcu() test complete.\n");
	destroy_rcu_head_on_stack(&rh1);
	destroy_rcu_head_on_stack(&rh2);
#else /* #ifdef CONFIG_DEBUG_OBJECTS_RCU_HEAD */
	pr_alert("rcutorture: !CONFIG_DEBUG_OBJECTS_RCU_HEAD, not testing duplicate call_rcu()\n");
#endif /* #else #ifdef CONFIG_DEBUG_OBJECTS_RCU_HEAD */
}

1409
static int __init
1410 1411 1412 1413 1414
rcu_torture_init(void)
{
	int i;
	int cpu;
	int firsterr = 0;
1415
	static struct rcu_torture_ops *torture_ops[] = {
1416
		&rcu_ops, &rcu_bh_ops, &rcu_busted_ops, &srcu_ops, &sched_ops,
1417
	};
1418

1419
	torture_init_begin(torture_type, verbose, &rcutorture_runnable);
1420

1421
	/* Process args and tell the world that the torturer is on the job. */
1422
	for (i = 0; i < ARRAY_SIZE(torture_ops); i++) {
1423
		cur_ops = torture_ops[i];
1424
		if (strcmp(torture_type, cur_ops->name) == 0)
1425 1426
			break;
	}
1427
	if (i == ARRAY_SIZE(torture_ops)) {
1428 1429 1430
		pr_alert("rcu-torture: invalid torture type: \"%s\"\n",
			 torture_type);
		pr_alert("rcu-torture types:");
1431
		for (i = 0; i < ARRAY_SIZE(torture_ops); i++)
1432 1433
			pr_alert(" %s", torture_ops[i]->name);
		pr_alert("\n");
1434
		torture_init_end();
1435
		return -EINVAL;
1436
	}
1437
	if (cur_ops->fqs == NULL && fqs_duration != 0) {
1438
		pr_alert("rcu-torture: ->fqs NULL and non-zero fqs_duration, fqs disabled.\n");
1439 1440
		fqs_duration = 0;
	}
1441
	if (cur_ops->init)
1442 1443
		cur_ops->init(); /* no "goto unwind" prior to this point!!! */

1444 1445 1446 1447
	if (nreaders >= 0)
		nrealreaders = nreaders;
	else
		nrealreaders = 2 * num_online_cpus();
1448
	rcu_torture_print_module_parms(cur_ops, "Start of test");
1449 1450 1451 1452

	/* Set up the freelist. */

	INIT_LIST_HEAD(&rcu_torture_freelist);
1453
	for (i = 0; i < ARRAY_SIZE(rcu_tortures); i++) {
1454
		rcu_tortures[i].rtort_mbtest = 0;
1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465
		list_add_tail(&rcu_tortures[i].rtort_free,
			      &rcu_torture_freelist);
	}

	/* Initialize the statistics so that each run gets its own numbers. */

	rcu_torture_current = NULL;
	rcu_torture_current_version = 0;
	atomic_set(&n_rcu_torture_alloc, 0);
	atomic_set(&n_rcu_torture_alloc_fail, 0);
	atomic_set(&n_rcu_torture_free, 0);
1466 1467
	atomic_set(&n_rcu_torture_mberror, 0);
	atomic_set(&n_rcu_torture_error, 0);
1468
	n_rcu_torture_barrier_error = 0;
1469 1470 1471 1472
	n_rcu_torture_boost_ktrerror = 0;
	n_rcu_torture_boost_rterror = 0;
	n_rcu_torture_boost_failure = 0;
	n_rcu_torture_boosts = 0;
1473 1474
	for (i = 0; i < RCU_TORTURE_PIPE_LEN + 1; i++)
		atomic_set(&rcu_torture_wcount[i], 0);
1475
	for_each_possible_cpu(cpu) {
1476 1477 1478 1479 1480 1481 1482 1483
		for (i = 0; i < RCU_TORTURE_PIPE_LEN + 1; i++) {
			per_cpu(rcu_torture_count, cpu)[i] = 0;
			per_cpu(rcu_torture_batch, cpu)[i] = 0;
		}
	}

	/* Start up the kthreads. */

1484 1485 1486
	firsterr = torture_create_kthread(rcu_torture_writer, NULL,
					  writer_task);
	if (firsterr)
1487
		goto unwind;
1488
	fakewriter_tasks = kzalloc(nfakewriters * sizeof(fakewriter_tasks[0]),
1489
				   GFP_KERNEL);
1490
	if (fakewriter_tasks == NULL) {
1491
		VERBOSE_TOROUT_ERRSTRING("out of memory");
1492 1493 1494 1495
		firsterr = -ENOMEM;
		goto unwind;
	}
	for (i = 0; i < nfakewriters; i++) {
1496 1497 1498
		firsterr = torture_create_kthread(rcu_torture_fakewriter,
						  NULL, fakewriter_tasks[i]);
		if (firsterr)
1499 1500
			goto unwind;
	}
1501
	reader_tasks = kzalloc(nrealreaders * sizeof(reader_tasks[0]),
1502 1503
			       GFP_KERNEL);
	if (reader_tasks == NULL) {
1504
		VERBOSE_TOROUT_ERRSTRING("out of memory");
1505 1506 1507 1508
		firsterr = -ENOMEM;
		goto unwind;
	}
	for (i = 0; i < nrealreaders; i++) {
1509 1510 1511
		firsterr = torture_create_kthread(rcu_torture_reader, NULL,
						  reader_tasks[i]);
		if (firsterr)
1512 1513 1514
			goto unwind;
	}
	if (stat_interval > 0) {
1515 1516 1517
		firsterr = torture_create_kthread(rcu_torture_stats, NULL,
						  stats_task);
		if (firsterr)
1518 1519
			goto unwind;
	}
1520
	if (test_no_idle_hz) {
1521 1522
		firsterr = torture_shuffle_init(shuffle_interval * HZ);
		if (firsterr)
1523 1524
			goto unwind;
	}
1525 1526 1527
	if (stutter < 0)
		stutter = 0;
	if (stutter) {
1528 1529
		firsterr = torture_stutter_init(stutter * HZ);
		if (firsterr)
1530 1531
			goto unwind;
	}
1532 1533 1534
	if (fqs_duration < 0)
		fqs_duration = 0;
	if (fqs_duration) {
1535
		/* Create the fqs thread */
1536 1537
		torture_create_kthread(rcu_torture_fqs, NULL, fqs_task);
		if (firsterr)
1538 1539
			goto unwind;
	}
1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551
	if (test_boost_interval < 1)
		test_boost_interval = 1;
	if (test_boost_duration < 2)
		test_boost_duration = 2;
	if ((test_boost == 1 && cur_ops->can_boost) ||
	    test_boost == 2) {

		boost_starttime = jiffies + test_boost_interval * HZ;
		register_cpu_notifier(&rcutorture_cpu_nb);
		for_each_possible_cpu(i) {
			if (cpu_is_offline(i))
				continue;  /* Heuristic: CPU can go offline. */
1552 1553
			firsterr = rcutorture_booster_init(i);
			if (firsterr)
1554 1555 1556
				goto unwind;
		}
	}
1557 1558
	firsterr = torture_shutdown_init(shutdown_secs, rcu_torture_cleanup);
	if (firsterr)
1559
		goto unwind;
1560 1561
	firsterr = torture_onoff_init(onoff_holdoff * HZ, onoff_interval * HZ);
	if (firsterr)
1562
		goto unwind;
1563 1564
	firsterr = rcu_torture_stall_init();
	if (firsterr)
1565
		goto unwind;
1566 1567
	firsterr = rcu_torture_barrier_init();
	if (firsterr)
1568
		goto unwind;
1569 1570
	if (object_debug)
		rcu_test_debug_objects();
1571
	rcutorture_record_test_transition();
1572
	torture_init_end();
1573 1574 1575
	return 0;

unwind:
1576
	torture_init_end();
1577 1578 1579 1580 1581 1582
	rcu_torture_cleanup();
	return firsterr;
}

module_init(rcu_torture_init);
module_exit(rcu_torture_cleanup);