rcutorture.c 63.5 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 15 16 17
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
 *
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 52

MODULE_LICENSE("GPL");
53
MODULE_AUTHOR("Paul E. McKenney <paulmck@us.ibm.com> and Josh Triplett <josh@freedesktop.org>");
54

55
static int nreaders = -1;	/* # reader threads, defaults to 2*ncpus */
56
static int nfakewriters = 4;	/* # fake writer threads */
57 58
static int stat_interval = 60;	/* Interval between stats, in seconds. */
				/*  Zero means "only at end of test". */
59
static bool verbose;		/* Print more debug info. */
60 61
static bool test_no_idle_hz = true;
				/* Test RCU support for tickless idle CPUs. */
62 63
static int shuffle_interval = 3; /* Interval between shuffles (in sec)*/
static int stutter = 5;		/* Start/stop testing interval (in sec) */
64
static int irqreader = 1;	/* RCU readers from irq (timers). */
65 66
static int fqs_duration;	/* Duration of bursts (us), 0 to disable. */
static int fqs_holdoff;		/* Hold time within burst (us). */
67
static int fqs_stutter = 3;	/* Wait time between bursts (s). */
68
static int n_barrier_cbs;	/* Number of callbacks to test RCU barriers. */
69
static int onoff_interval;	/* Wait time between CPU hotplugs, 0=disable. */
70
static int onoff_holdoff;	/* Seconds after boot before CPU hotplugs. */
71
static int shutdown_secs;	/* Shutdown time (s).  <=0 for no shutdown. */
72 73
static int stall_cpu;		/* CPU-stall duration (s).  0 for no stall. */
static int stall_cpu_holdoff = 10; /* Time to wait until stall (s).  */
74 75 76
static int test_boost = 1;	/* Test RCU prio boost: 0=no, 1=maybe, 2=yes. */
static int test_boost_interval = 7; /* Interval between boost tests, seconds. */
static int test_boost_duration = 4; /* Duration of each boost test, seconds. */
77
static char *torture_type = "rcu"; /* What RCU implementation to torture. */
78

79
module_param(nreaders, int, 0444);
80
MODULE_PARM_DESC(nreaders, "Number of RCU reader threads");
81
module_param(nfakewriters, int, 0444);
82
MODULE_PARM_DESC(nfakewriters, "Number of RCU fake writer threads");
83
module_param(stat_interval, int, 0644);
84
MODULE_PARM_DESC(stat_interval, "Number of seconds between stats printk()s");
85
module_param(verbose, bool, 0444);
86
MODULE_PARM_DESC(verbose, "Enable verbose debugging printk()s");
87
module_param(test_no_idle_hz, bool, 0444);
88
MODULE_PARM_DESC(test_no_idle_hz, "Test support for tickless idle CPUs");
89
module_param(shuffle_interval, int, 0444);
90
MODULE_PARM_DESC(shuffle_interval, "Number of seconds between shuffles");
91 92
module_param(stutter, int, 0444);
MODULE_PARM_DESC(stutter, "Number of seconds to run/halt test");
93 94
module_param(irqreader, int, 0444);
MODULE_PARM_DESC(irqreader, "Allow RCU readers from irq handlers");
95 96 97 98 99 100
module_param(fqs_duration, int, 0444);
MODULE_PARM_DESC(fqs_duration, "Duration of fqs bursts (us)");
module_param(fqs_holdoff, int, 0444);
MODULE_PARM_DESC(fqs_holdoff, "Holdoff time within fqs bursts (us)");
module_param(fqs_stutter, int, 0444);
MODULE_PARM_DESC(fqs_stutter, "Wait time between fqs bursts (s)");
101 102
module_param(n_barrier_cbs, int, 0444);
MODULE_PARM_DESC(n_barrier_cbs, "# of callbacks/kthreads for barrier testing");
103 104
module_param(onoff_interval, int, 0444);
MODULE_PARM_DESC(onoff_interval, "Time between CPU hotplugs (s), 0=disable");
105 106
module_param(onoff_holdoff, int, 0444);
MODULE_PARM_DESC(onoff_holdoff, "Time after boot before CPU hotplugs (s)");
107 108
module_param(shutdown_secs, int, 0444);
MODULE_PARM_DESC(shutdown_secs, "Shutdown time (s), zero to disable.");
109 110 111 112
module_param(stall_cpu, int, 0444);
MODULE_PARM_DESC(stall_cpu, "Stall duration (s), zero to disable.");
module_param(stall_cpu_holdoff, int, 0444);
MODULE_PARM_DESC(stall_cpu_holdoff, "Time to wait before starting stall (s).");
113 114 115 116 117 118
module_param(test_boost, int, 0444);
MODULE_PARM_DESC(test_boost, "Test RCU prio boost: 0=no, 1=maybe, 2=yes.");
module_param(test_boost_interval, int, 0444);
MODULE_PARM_DESC(test_boost_interval, "Interval between boost tests, seconds.");
module_param(test_boost_duration, int, 0444);
MODULE_PARM_DESC(test_boost_duration, "Duration of each boost test, seconds.");
119
module_param(torture_type, charp, 0444);
120
MODULE_PARM_DESC(torture_type, "Type of RCU to torture (rcu, rcu_bh, srcu)");
121 122

#define TORTURE_FLAG "-torture:"
123
#define PRINTK_STRING(s) \
124
	do { pr_alert("%s" TORTURE_FLAG s "\n", torture_type); } while (0)
125
#define VERBOSE_PRINTK_STRING(s) \
126
	do { if (verbose) pr_alert("%s" TORTURE_FLAG s "\n", torture_type); } while (0)
127
#define VERBOSE_PRINTK_ERRSTRING(s) \
128
	do { if (verbose) pr_alert("%s" TORTURE_FLAG "!!! " s "\n", torture_type); } while (0)
129 130 131 132 133

static char printk_buf[4096];

static int nrealreaders;
static struct task_struct *writer_task;
134
static struct task_struct **fakewriter_tasks;
135 136
static struct task_struct **reader_tasks;
static struct task_struct *stats_task;
137
static struct task_struct *shuffler_task;
138
static struct task_struct *stutter_task;
139
static struct task_struct *fqs_task;
140
static struct task_struct *boost_tasks[NR_CPUS];
141
static struct task_struct *shutdown_task;
142 143 144
#ifdef CONFIG_HOTPLUG_CPU
static struct task_struct *onoff_task;
#endif /* #ifdef CONFIG_HOTPLUG_CPU */
145
static struct task_struct *stall_task;
146 147
static struct task_struct **barrier_cbs_tasks;
static struct task_struct *barrier_task;
148 149 150 151 152 153 154

#define RCU_TORTURE_PIPE_LEN 10

struct rcu_torture {
	struct rcu_head rtort_rcu;
	int rtort_pipe_count;
	struct list_head rtort_free;
155
	int rtort_mbtest;
156 157 158
};

static LIST_HEAD(rcu_torture_freelist);
159
static struct rcu_torture __rcu *rcu_torture_current;
160
static unsigned long rcu_torture_current_version;
161 162 163 164 165 166 167
static struct rcu_torture rcu_tortures[10 * RCU_TORTURE_PIPE_LEN];
static DEFINE_SPINLOCK(rcu_torture_lock);
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 };
static atomic_t rcu_torture_wcount[RCU_TORTURE_PIPE_LEN + 1];
168 169 170 171 172
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;
173
static long n_rcu_torture_barrier_error;
174 175 176 177
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;
178
static long n_rcu_torture_timers;
179 180
static long n_offline_attempts;
static long n_offline_successes;
181 182 183
static unsigned long sum_offline;
static int min_offline = -1;
static int max_offline;
184 185
static long n_online_attempts;
static long n_online_successes;
186 187 188
static unsigned long sum_online;
static int min_online = -1;
static int max_online;
189 190
static long n_barrier_attempts;
static long n_barrier_successes;
191
static struct list_head rcu_torture_removed;
R
Rusty Russell 已提交
192
static cpumask_var_t shuffle_tmp_mask;
193

194
static int stutter_pause_test;
195

196 197 198 199 200 201
#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;
202 203
module_param(rcutorture_runnable, int, 0444);
MODULE_PARM_DESC(rcutorture_runnable, "Start rcutorture at boot");
204

205
#if defined(CONFIG_RCU_BOOST) && !defined(CONFIG_HOTPLUG_CPU)
206
#define rcu_can_boost() 1
207
#else /* #if defined(CONFIG_RCU_BOOST) && !defined(CONFIG_HOTPLUG_CPU) */
208
#define rcu_can_boost() 0
209
#endif /* #else #if defined(CONFIG_RCU_BOOST) && !defined(CONFIG_HOTPLUG_CPU) */
210

211 212 213 214 215 216 217 218 219 220 221 222 223 224
#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 */

225
static unsigned long shutdown_time;	/* jiffies to system shutdown. */
226 227 228
static unsigned long boost_starttime;	/* jiffies of next boost test start. */
DEFINE_MUTEX(boost_mutex);		/* protect setting boost_starttime */
					/*  and boost task create/destroy. */
229
static atomic_t barrier_cbs_count;	/* Barrier callbacks registered. */
230
static bool barrier_phase;		/* Test phase. */
231 232 233
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);
234

235 236 237 238 239 240
/* Mediate rmmod and system shutdown.  Concurrent rmmod & shutdown illegal! */

#define FULLSTOP_DONTSTOP 0	/* Normal operation. */
#define FULLSTOP_SHUTDOWN 1	/* System shutdown with rcutorture running. */
#define FULLSTOP_RMMOD    2	/* Normal rmmod of rcutorture. */
static int fullstop = FULLSTOP_RMMOD;
241 242 243 244
/*
 * Protect fullstop transitions and spawning of kthreads.
 */
static DEFINE_MUTEX(fullstop_mutex);
245

246 247 248
/* Forward reference. */
static void rcu_torture_cleanup(void);

249
/*
250
 * Detect and respond to a system shutdown.
251 252 253 254 255
 */
static int
rcutorture_shutdown_notify(struct notifier_block *unused1,
			   unsigned long unused2, void *unused3)
{
P
Paul E. McKenney 已提交
256
	mutex_lock(&fullstop_mutex);
257
	if (fullstop == FULLSTOP_DONTSTOP)
P
Paul E. McKenney 已提交
258
		fullstop = FULLSTOP_SHUTDOWN;
259
	else
260
		pr_warn(/* but going down anyway, so... */
261
		       "Concurrent 'rmmod rcutorture' and shutdown illegal!\n");
P
Paul E. McKenney 已提交
262
	mutex_unlock(&fullstop_mutex);
263 264 265
	return NOTIFY_DONE;
}

266 267 268 269 270 271 272
/*
 * Absorb kthreads into a kernel function that won't return, so that
 * they won't ever access module text or data again.
 */
static void rcutorture_shutdown_absorb(char *title)
{
	if (ACCESS_ONCE(fullstop) == FULLSTOP_SHUTDOWN) {
273
		pr_notice(
274 275 276 277 278 279
		       "rcutorture thread %s parking due to system shutdown\n",
		       title);
		schedule_timeout_uninterruptible(MAX_SCHEDULE_TIMEOUT);
	}
}

280 281 282
/*
 * Allocate an element from the rcu_tortures pool.
 */
A
Adrian Bunk 已提交
283
static struct rcu_torture *
284 285 286 287
rcu_torture_alloc(void)
{
	struct list_head *p;

288
	spin_lock_bh(&rcu_torture_lock);
289 290
	if (list_empty(&rcu_torture_freelist)) {
		atomic_inc(&n_rcu_torture_alloc_fail);
291
		spin_unlock_bh(&rcu_torture_lock);
292 293 294 295 296
		return NULL;
	}
	atomic_inc(&n_rcu_torture_alloc);
	p = rcu_torture_freelist.next;
	list_del_init(p);
297
	spin_unlock_bh(&rcu_torture_lock);
298 299 300 301 302 303 304 305 306 307
	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);
308
	spin_lock_bh(&rcu_torture_lock);
309
	list_add_tail(&p->rtort_free, &rcu_torture_freelist);
310
	spin_unlock_bh(&rcu_torture_lock);
311 312 313 314
}

struct rcu_random_state {
	unsigned long rrs_state;
315
	long rrs_count;
316 317 318 319 320 321 322 323 324 325
};

#define RCU_RANDOM_MULT 39916801  /* prime */
#define RCU_RANDOM_ADD	479001701 /* prime */
#define RCU_RANDOM_REFRESH 10000

#define DEFINE_RCU_RANDOM(name) struct rcu_random_state name = { 0, 0 }

/*
 * Crude but fast random-number generator.  Uses a linear congruential
326
 * generator, with occasional help from cpu_clock().
327
 */
328
static unsigned long
329 330 331
rcu_random(struct rcu_random_state *rrsp)
{
	if (--rrsp->rrs_count < 0) {
332
		rrsp->rrs_state += (unsigned long)local_clock();
333 334 335 336 337 338
		rrsp->rrs_count = RCU_RANDOM_REFRESH;
	}
	rrsp->rrs_state = rrsp->rrs_state * RCU_RANDOM_MULT + RCU_RANDOM_ADD;
	return swahw32(rrsp->rrs_state);
}

339
static void
340
rcu_stutter_wait(char *title)
341
{
342
	while (stutter_pause_test || !rcutorture_runnable) {
343 344 345
		if (rcutorture_runnable)
			schedule_timeout_interruptible(1);
		else
346
			schedule_timeout_interruptible(round_jiffies_relative(HZ));
347
		rcutorture_shutdown_absorb(title);
348
	}
349 350
}

351 352 353 354 355 356 357
/*
 * Operations vector for selecting different types of tests.
 */

struct rcu_torture_ops {
	void (*init)(void);
	int (*readlock)(void);
358
	void (*read_delay)(struct rcu_random_state *rrsp);
359 360
	void (*readunlock)(int idx);
	int (*completed)(void);
361
	void (*deferred_free)(struct rcu_torture *p);
362
	void (*sync)(void);
363
	void (*call)(struct rcu_head *head, void (*func)(struct rcu_head *rcu));
364
	void (*cb_barrier)(void);
365
	void (*fqs)(void);
366
	int (*stats)(char *page);
367
	int irq_capable;
368
	int can_boost;
369 370
	char *name;
};
371 372

static struct rcu_torture_ops *cur_ops;
373 374 375 376 377

/*
 * Definitions for rcu torture testing.
 */

378
static int rcu_torture_read_lock(void) __acquires(RCU)
379 380 381 382 383
{
	rcu_read_lock();
	return 0;
}

384 385
static void rcu_read_delay(struct rcu_random_state *rrsp)
{
386 387
	const unsigned long shortdelay_us = 200;
	const unsigned long longdelay_ms = 50;
388

389 390 391
	/* 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. */
392

393 394 395 396
	if (!(rcu_random(rrsp) % (nrealreaders * 2000 * longdelay_ms)))
		mdelay(longdelay_ms);
	if (!(rcu_random(rrsp) % (nrealreaders * 2 * shortdelay_us)))
		udelay(shortdelay_us);
397 398 399 400
#ifdef CONFIG_PREEMPT
	if (!preempt_count() && !(rcu_random(rrsp) % (nrealreaders * 20000)))
		preempt_schedule();  /* No QS if preempt_disable() in effect */
#endif
401 402
}

403
static void rcu_torture_read_unlock(int idx) __releases(RCU)
404 405 406 407 408 409 410 411 412 413 414 415 416 417 418
{
	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);

419
	if (fullstop != FULLSTOP_DONTSTOP) {
420 421 422 423 424 425 426 427 428 429 430
		/* 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);
431
	} else {
432
		cur_ops->deferred_free(rp);
433
	}
434 435
}

436 437 438 439 440
static int rcu_no_completed(void)
{
	return 0;
}

441 442 443 444 445 446
static void rcu_torture_deferred_free(struct rcu_torture *p)
{
	call_rcu(&p->rtort_rcu, rcu_torture_cb);
}

static struct rcu_torture_ops rcu_ops = {
447 448 449 450 451 452 453
	.init		= NULL,
	.readlock	= rcu_torture_read_lock,
	.read_delay	= rcu_read_delay,
	.readunlock	= rcu_torture_read_unlock,
	.completed	= rcu_torture_completed,
	.deferred_free	= rcu_torture_deferred_free,
	.sync		= synchronize_rcu,
454
	.call		= call_rcu,
455
	.cb_barrier	= rcu_barrier,
456
	.fqs		= rcu_force_quiescent_state,
457
	.stats		= NULL,
458
	.irq_capable	= 1,
459
	.can_boost	= rcu_can_boost(),
460
	.name		= "rcu"
461 462
};

463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488
static void rcu_sync_torture_deferred_free(struct rcu_torture *p)
{
	int i;
	struct rcu_torture *rp;
	struct rcu_torture *rp1;

	cur_ops->sync();
	list_add(&p->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);
		}
	}
}

static void rcu_sync_torture_init(void)
{
	INIT_LIST_HEAD(&rcu_torture_removed);
}

489
static struct rcu_torture_ops rcu_sync_ops = {
490 491 492 493 494 495 496
	.init		= rcu_sync_torture_init,
	.readlock	= rcu_torture_read_lock,
	.read_delay	= rcu_read_delay,
	.readunlock	= rcu_torture_read_unlock,
	.completed	= rcu_torture_completed,
	.deferred_free	= rcu_sync_torture_deferred_free,
	.sync		= synchronize_rcu,
497
	.call		= NULL,
498
	.cb_barrier	= NULL,
499
	.fqs		= rcu_force_quiescent_state,
500 501
	.stats		= NULL,
	.irq_capable	= 1,
502
	.can_boost	= rcu_can_boost(),
503
	.name		= "rcu_sync"
504 505
};

506 507 508 509 510 511 512 513
static struct rcu_torture_ops rcu_expedited_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_sync_torture_deferred_free,
	.sync		= synchronize_rcu_expedited,
514
	.call		= NULL,
515
	.cb_barrier	= NULL,
516
	.fqs		= rcu_force_quiescent_state,
517 518
	.stats		= NULL,
	.irq_capable	= 1,
519
	.can_boost	= rcu_can_boost(),
520 521 522
	.name		= "rcu_expedited"
};

523 524 525 526
/*
 * Definitions for rcu_bh torture testing.
 */

527
static int rcu_bh_torture_read_lock(void) __acquires(RCU_BH)
528 529 530 531 532
{
	rcu_read_lock_bh();
	return 0;
}

533
static void rcu_bh_torture_read_unlock(int idx) __releases(RCU_BH)
534 535 536 537 538 539 540 541 542 543 544 545 546 547 548
{
	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 = {
549 550 551 552 553 554
	.init		= NULL,
	.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,
555
	.sync		= synchronize_rcu_bh,
556
	.call		= call_rcu_bh,
557
	.cb_barrier	= rcu_barrier_bh,
558
	.fqs		= rcu_bh_force_quiescent_state,
559 560 561
	.stats		= NULL,
	.irq_capable	= 1,
	.name		= "rcu_bh"
562 563
};

564
static struct rcu_torture_ops rcu_bh_sync_ops = {
565 566 567 568 569 570
	.init		= rcu_sync_torture_init,
	.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_sync_torture_deferred_free,
571
	.sync		= synchronize_rcu_bh,
572
	.call		= NULL,
573
	.cb_barrier	= NULL,
574
	.fqs		= rcu_bh_force_quiescent_state,
575 576 577
	.stats		= NULL,
	.irq_capable	= 1,
	.name		= "rcu_bh_sync"
578 579
};

580 581 582 583 584 585 586 587
static struct rcu_torture_ops rcu_bh_expedited_ops = {
	.init		= rcu_sync_torture_init,
	.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_sync_torture_deferred_free,
	.sync		= synchronize_rcu_bh_expedited,
588
	.call		= NULL,
589 590 591 592 593 594 595
	.cb_barrier	= NULL,
	.fqs		= rcu_bh_force_quiescent_state,
	.stats		= NULL,
	.irq_capable	= 1,
	.name		= "rcu_bh_expedited"
};

596 597 598 599
/*
 * Definitions for srcu torture testing.
 */

600
DEFINE_STATIC_SRCU(srcu_ctl);
601

602
static int srcu_torture_read_lock(void) __acquires(&srcu_ctl)
603 604 605 606 607 608 609 610 611 612 613 614 615 616 617
{
	return srcu_read_lock(&srcu_ctl);
}

static void srcu_read_delay(struct rcu_random_state *rrsp)
{
	long delay;
	const long uspertick = 1000000 / HZ;
	const long longdelay = 10;

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

	delay = rcu_random(rrsp) % (nrealreaders * 2 * longdelay * uspertick);
	if (!delay)
		schedule_timeout_interruptible(longdelay);
618 619
	else
		rcu_read_delay(rrsp);
620 621
}

622
static void srcu_torture_read_unlock(int idx) __releases(&srcu_ctl)
623 624 625 626 627 628 629 630 631
{
	srcu_read_unlock(&srcu_ctl, idx);
}

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

632 633 634 635 636
static void srcu_torture_deferred_free(struct rcu_torture *rp)
{
	call_srcu(&srcu_ctl, &rp->rtort_rcu, rcu_torture_cb);
}

637 638 639 640 641
static void srcu_torture_synchronize(void)
{
	synchronize_srcu(&srcu_ctl);
}

642 643 644 645 646 647 648 649 650 651 652
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);
}

653 654 655 656 657 658 659 660 661
static int srcu_torture_stats(char *page)
{
	int cnt = 0;
	int cpu;
	int idx = srcu_ctl.completed & 0x1;

	cnt += sprintf(&page[cnt], "%s%s per-CPU(idx=%d):",
		       torture_type, TORTURE_FLAG, idx);
	for_each_possible_cpu(cpu) {
662
		cnt += sprintf(&page[cnt], " %d(%lu,%lu)", cpu,
663 664 665 666 667 668 669 670
			       per_cpu_ptr(srcu_ctl.per_cpu_ref, cpu)->c[!idx],
			       per_cpu_ptr(srcu_ctl.per_cpu_ref, cpu)->c[idx]);
	}
	cnt += sprintf(&page[cnt], "\n");
	return cnt;
}

static struct rcu_torture_ops srcu_ops = {
671
	.init		= rcu_sync_torture_init,
672 673 674 675
	.readlock	= srcu_torture_read_lock,
	.read_delay	= srcu_read_delay,
	.readunlock	= srcu_torture_read_unlock,
	.completed	= srcu_torture_completed,
676
	.deferred_free	= srcu_torture_deferred_free,
677
	.sync		= srcu_torture_synchronize,
678 679
	.call		= srcu_torture_call,
	.cb_barrier	= srcu_torture_barrier,
680 681
	.stats		= srcu_torture_stats,
	.name		= "srcu"
682 683
};

684
static struct rcu_torture_ops srcu_sync_ops = {
685
	.init		= rcu_sync_torture_init,
686 687 688 689 690 691 692 693 694 695 696 697
	.readlock	= srcu_torture_read_lock,
	.read_delay	= srcu_read_delay,
	.readunlock	= srcu_torture_read_unlock,
	.completed	= srcu_torture_completed,
	.deferred_free	= rcu_sync_torture_deferred_free,
	.sync		= srcu_torture_synchronize,
	.call		= NULL,
	.cb_barrier	= NULL,
	.stats		= srcu_torture_stats,
	.name		= "srcu_sync"
};

698 699 700 701 702 703 704 705 706 707 708
static int srcu_torture_read_lock_raw(void) __acquires(&srcu_ctl)
{
	return srcu_read_lock_raw(&srcu_ctl);
}

static void srcu_torture_read_unlock_raw(int idx) __releases(&srcu_ctl)
{
	srcu_read_unlock_raw(&srcu_ctl, idx);
}

static struct rcu_torture_ops srcu_raw_ops = {
709
	.init		= rcu_sync_torture_init,
710 711 712 713
	.readlock	= srcu_torture_read_lock_raw,
	.read_delay	= srcu_read_delay,
	.readunlock	= srcu_torture_read_unlock_raw,
	.completed	= srcu_torture_completed,
714
	.deferred_free	= srcu_torture_deferred_free,
715
	.sync		= srcu_torture_synchronize,
716
	.call		= NULL,
717 718 719 720 721
	.cb_barrier	= NULL,
	.stats		= srcu_torture_stats,
	.name		= "srcu_raw"
};

722
static struct rcu_torture_ops srcu_raw_sync_ops = {
723
	.init		= rcu_sync_torture_init,
724 725 726 727 728 729 730 731 732 733 734 735
	.readlock	= srcu_torture_read_lock_raw,
	.read_delay	= srcu_read_delay,
	.readunlock	= srcu_torture_read_unlock_raw,
	.completed	= srcu_torture_completed,
	.deferred_free	= rcu_sync_torture_deferred_free,
	.sync		= srcu_torture_synchronize,
	.call		= NULL,
	.cb_barrier	= NULL,
	.stats		= srcu_torture_stats,
	.name		= "srcu_raw_sync"
};

736 737 738 739 740 741
static void srcu_torture_synchronize_expedited(void)
{
	synchronize_srcu_expedited(&srcu_ctl);
}

static struct rcu_torture_ops srcu_expedited_ops = {
742
	.init		= rcu_sync_torture_init,
743 744 745 746 747 748
	.readlock	= srcu_torture_read_lock,
	.read_delay	= srcu_read_delay,
	.readunlock	= srcu_torture_read_unlock,
	.completed	= srcu_torture_completed,
	.deferred_free	= rcu_sync_torture_deferred_free,
	.sync		= srcu_torture_synchronize_expedited,
749
	.call		= NULL,
750 751 752 753 754
	.cb_barrier	= NULL,
	.stats		= srcu_torture_stats,
	.name		= "srcu_expedited"
};

755 756 757 758 759 760 761 762 763 764 765 766 767 768 769
/*
 * 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();
}

770 771 772 773 774
static void rcu_sched_torture_deferred_free(struct rcu_torture *p)
{
	call_rcu_sched(&p->rtort_rcu, rcu_torture_cb);
}

775
static struct rcu_torture_ops sched_ops = {
776 777 778 779
	.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,
780
	.completed	= rcu_no_completed,
781
	.deferred_free	= rcu_sched_torture_deferred_free,
782
	.sync		= synchronize_sched,
783
	.cb_barrier	= rcu_barrier_sched,
784
	.fqs		= rcu_sched_force_quiescent_state,
785 786 787
	.stats		= NULL,
	.irq_capable	= 1,
	.name		= "sched"
788 789
};

790
static struct rcu_torture_ops sched_sync_ops = {
791 792 793 794
	.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,
795
	.completed	= rcu_no_completed,
796
	.deferred_free	= rcu_sync_torture_deferred_free,
797
	.sync		= synchronize_sched,
798
	.cb_barrier	= NULL,
799
	.fqs		= rcu_sched_force_quiescent_state,
800 801 802 803 804 805 806 807 808
	.stats		= NULL,
	.name		= "sched_sync"
};

static struct rcu_torture_ops sched_expedited_ops = {
	.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,
809
	.completed	= rcu_no_completed,
810 811 812
	.deferred_free	= rcu_sync_torture_deferred_free,
	.sync		= synchronize_sched_expedited,
	.cb_barrier	= NULL,
813
	.fqs		= rcu_sched_force_quiescent_state,
814
	.stats		= NULL,
815 816
	.irq_capable	= 1,
	.name		= "sched_expedited"
817 818
};

819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856
/*
 * 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;

	VERBOSE_PRINTK_STRING("rcu_torture_boost started");

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

857
	init_rcu_head_on_stack(&rbi.rcu);
858 859 860 861
	/* Each pass through the following loop does one boost-test cycle. */
	do {
		/* Wait for the next test interval. */
		oldstarttime = boost_starttime;
862
		while (ULONG_CMP_LT(jiffies, oldstarttime)) {
863
			schedule_timeout_interruptible(oldstarttime - jiffies);
864 865 866 867 868 869 870 871 872
			rcu_stutter_wait("rcu_torture_boost");
			if (kthread_should_stop() ||
			    fullstop != FULLSTOP_DONTSTOP)
				goto checkwait;
		}

		/* Do one boost-test interval. */
		endtime = oldstarttime + test_boost_duration * HZ;
		call_rcu_time = jiffies;
873
		while (ULONG_CMP_LT(jiffies, endtime)) {
874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899
			/* 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) {
					VERBOSE_PRINTK_STRING("rcu_torture_boost boosting failed");
					n_rcu_torture_boost_failure++;
				}
				call_rcu_time = jiffies;
			}
			cond_resched();
			rcu_stutter_wait("rcu_torture_boost");
			if (kthread_should_stop() ||
			    fullstop != FULLSTOP_DONTSTOP)
				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.
		 */
900 901
		while (oldstarttime == boost_starttime &&
		       !kthread_should_stop()) {
902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921
			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. */
checkwait:	rcu_stutter_wait("rcu_torture_boost");
	} while (!kthread_should_stop() && fullstop  == FULLSTOP_DONTSTOP);

	/* Clean up and exit. */
	VERBOSE_PRINTK_STRING("rcu_torture_boost task stopping");
	rcutorture_shutdown_absorb("rcu_torture_boost");
	while (!kthread_should_stop() || rbi.inflight)
		schedule_timeout_uninterruptible(1);
	smp_mb(); /* order accesses to ->inflight before stack-frame death. */
922
	destroy_rcu_head_on_stack(&rbi.rcu);
923 924 925
	return 0;
}

926 927 928 929 930 931 932 933 934 935 936 937 938 939
/*
 * 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;

	VERBOSE_PRINTK_STRING("rcu_torture_fqs task started");
	do {
		fqs_resume_time = jiffies + fqs_stutter * HZ;
940 941
		while (ULONG_CMP_LT(jiffies, fqs_resume_time) &&
		       !kthread_should_stop()) {
942 943 944
			schedule_timeout_interruptible(1);
		}
		fqs_burst_remaining = fqs_duration;
945 946
		while (fqs_burst_remaining > 0 &&
		       !kthread_should_stop()) {
947 948 949 950 951 952 953 954 955 956 957 958 959
			cur_ops->fqs();
			udelay(fqs_holdoff);
			fqs_burst_remaining -= fqs_holdoff;
		}
		rcu_stutter_wait("rcu_torture_fqs");
	} while (!kthread_should_stop() && fullstop == FULLSTOP_DONTSTOP);
	VERBOSE_PRINTK_STRING("rcu_torture_fqs task stopping");
	rcutorture_shutdown_absorb("rcu_torture_fqs");
	while (!kthread_should_stop())
		schedule_timeout_uninterruptible(1);
	return 0;
}

960 961 962 963 964 965 966 967 968 969 970 971 972 973 974
/*
 * 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)
{
	int i;
	long oldbatch = rcu_batches_completed();
	struct rcu_torture *rp;
	struct rcu_torture *old_rp;
	static DEFINE_RCU_RANDOM(rand);

	VERBOSE_PRINTK_STRING("rcu_torture_writer task started");
975 976
	set_user_nice(current, 19);

977 978
	do {
		schedule_timeout_uninterruptible(1);
979 980
		rp = rcu_torture_alloc();
		if (rp == NULL)
981 982 983
			continue;
		rp->rtort_pipe_count = 0;
		udelay(rcu_random(&rand) & 0x3ff);
984 985
		old_rp = rcu_dereference_check(rcu_torture_current,
					       current == writer_task);
986
		rp->rtort_mbtest = 1;
987
		rcu_assign_pointer(rcu_torture_current, rp);
988
		smp_wmb(); /* Mods to old_rp must follow rcu_assign_pointer() */
989
		if (old_rp) {
990 991 992 993 994
			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++;
995
			cur_ops->deferred_free(old_rp);
996
		}
997
		rcutorture_record_progress(++rcu_torture_current_version);
998
		oldbatch = cur_ops->completed();
999 1000
		rcu_stutter_wait("rcu_torture_writer");
	} while (!kthread_should_stop() && fullstop == FULLSTOP_DONTSTOP);
1001
	VERBOSE_PRINTK_STRING("rcu_torture_writer task stopping");
1002 1003
	rcutorture_shutdown_absorb("rcu_torture_writer");
	while (!kthread_should_stop())
1004 1005 1006 1007
		schedule_timeout_uninterruptible(1);
	return 0;
}

1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022
/*
 * RCU torture fake writer kthread.  Repeatedly calls sync, with a random
 * delay between calls.
 */
static int
rcu_torture_fakewriter(void *arg)
{
	DEFINE_RCU_RANDOM(rand);

	VERBOSE_PRINTK_STRING("rcu_torture_fakewriter task started");
	set_user_nice(current, 19);

	do {
		schedule_timeout_uninterruptible(1 + rcu_random(&rand)%10);
		udelay(rcu_random(&rand) & 0x3ff);
1023 1024 1025 1026 1027
		if (cur_ops->cb_barrier != NULL &&
		    rcu_random(&rand) % (nfakewriters * 8) == 0)
			cur_ops->cb_barrier();
		else
			cur_ops->sync();
1028 1029
		rcu_stutter_wait("rcu_torture_fakewriter");
	} while (!kthread_should_stop() && fullstop == FULLSTOP_DONTSTOP);
1030 1031

	VERBOSE_PRINTK_STRING("rcu_torture_fakewriter task stopping");
1032 1033
	rcutorture_shutdown_absorb("rcu_torture_fakewriter");
	while (!kthread_should_stop())
1034 1035 1036 1037
		schedule_timeout_uninterruptible(1);
	return 0;
}

1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048
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);
}

1049 1050 1051 1052 1053 1054 1055 1056 1057 1058
/*
 * 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;
1059
	int completed_end;
1060 1061 1062 1063
	static DEFINE_RCU_RANDOM(rand);
	static DEFINE_SPINLOCK(rand_lock);
	struct rcu_torture *p;
	int pipe_count;
1064
	unsigned long long ts;
1065 1066 1067

	idx = cur_ops->readlock();
	completed = cur_ops->completed();
1068
	ts = rcu_trace_clock_local();
1069 1070 1071 1072
	p = rcu_dereference_check(rcu_torture_current,
				  rcu_read_lock_bh_held() ||
				  rcu_read_lock_sched_held() ||
				  srcu_read_lock_held(&srcu_ctl));
1073 1074 1075 1076 1077 1078 1079 1080
	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);
1081
	cur_ops->read_delay(&rand);
1082 1083 1084 1085 1086 1087 1088 1089
	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;
	}
1090 1091 1092 1093
	completed_end = cur_ops->completed();
	if (pipe_count > 1) {
		do_trace_rcu_torture_read(cur_ops->name, &p->rtort_rcu, ts,
					  completed, completed_end);
1094
		rcutorture_trace_dump();
1095
	}
R
Rusty Russell 已提交
1096
	__this_cpu_inc(rcu_torture_count[pipe_count]);
1097
	completed = completed_end - completed;
1098 1099 1100 1101
	if (completed > RCU_TORTURE_PIPE_LEN) {
		/* Should not happen, but... */
		completed = RCU_TORTURE_PIPE_LEN;
	}
R
Rusty Russell 已提交
1102
	__this_cpu_inc(rcu_torture_batch[completed]);
1103 1104 1105 1106
	preempt_enable();
	cur_ops->readunlock(idx);
}

1107 1108 1109 1110 1111 1112 1113 1114 1115 1116
/*
 * 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;
1117
	int completed_end;
1118
	int idx;
1119 1120 1121
	DEFINE_RCU_RANDOM(rand);
	struct rcu_torture *p;
	int pipe_count;
1122
	struct timer_list t;
1123
	unsigned long long ts;
1124 1125

	VERBOSE_PRINTK_STRING("rcu_torture_reader task started");
1126
	set_user_nice(current, 19);
1127
	if (irqreader && cur_ops->irq_capable)
1128
		setup_timer_on_stack(&t, rcu_torture_timer, 0);
1129

1130
	do {
1131
		if (irqreader && cur_ops->irq_capable) {
1132
			if (!timer_pending(&t))
1133
				mod_timer(&t, jiffies + 1);
1134
		}
1135 1136
		idx = cur_ops->readlock();
		completed = cur_ops->completed();
1137
		ts = rcu_trace_clock_local();
1138 1139 1140 1141
		p = rcu_dereference_check(rcu_torture_current,
					  rcu_read_lock_bh_held() ||
					  rcu_read_lock_sched_held() ||
					  srcu_read_lock_held(&srcu_ctl));
1142 1143
		if (p == NULL) {
			/* Wait for rcu_torture_writer to get underway */
1144
			cur_ops->readunlock(idx);
1145 1146 1147
			schedule_timeout_interruptible(HZ);
			continue;
		}
1148 1149
		if (p->rtort_mbtest == 0)
			atomic_inc(&n_rcu_torture_mberror);
1150
		cur_ops->read_delay(&rand);
1151 1152 1153 1154 1155 1156
		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;
		}
1157 1158 1159 1160
		completed_end = cur_ops->completed();
		if (pipe_count > 1) {
			do_trace_rcu_torture_read(cur_ops->name, &p->rtort_rcu,
						  ts, completed, completed_end);
1161
			rcutorture_trace_dump();
1162
		}
R
Rusty Russell 已提交
1163
		__this_cpu_inc(rcu_torture_count[pipe_count]);
1164
		completed = completed_end - completed;
1165 1166 1167 1168
		if (completed > RCU_TORTURE_PIPE_LEN) {
			/* Should not happen, but... */
			completed = RCU_TORTURE_PIPE_LEN;
		}
R
Rusty Russell 已提交
1169
		__this_cpu_inc(rcu_torture_batch[completed]);
1170
		preempt_enable();
1171
		cur_ops->readunlock(idx);
1172
		schedule();
1173 1174
		rcu_stutter_wait("rcu_torture_reader");
	} while (!kthread_should_stop() && fullstop == FULLSTOP_DONTSTOP);
1175
	VERBOSE_PRINTK_STRING("rcu_torture_reader task stopping");
1176
	rcutorture_shutdown_absorb("rcu_torture_reader");
1177
	if (irqreader && cur_ops->irq_capable)
1178
		del_timer_sync(&t);
1179
	while (!kthread_should_stop())
1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195
		schedule_timeout_uninterruptible(1);
	return 0;
}

/*
 * Create an RCU-torture statistics message in the specified buffer.
 */
static int
rcu_torture_printk(char *page)
{
	int cnt = 0;
	int cpu;
	int i;
	long pipesummary[RCU_TORTURE_PIPE_LEN + 1] = { 0 };
	long batchsummary[RCU_TORTURE_PIPE_LEN + 1] = { 0 };

1196
	for_each_possible_cpu(cpu) {
1197 1198 1199 1200 1201 1202 1203 1204 1205
		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;
	}
1206
	cnt += sprintf(&page[cnt], "%s%s ", torture_type, TORTURE_FLAG);
1207
	cnt += sprintf(&page[cnt],
1208
		       "rtc: %p ver: %lu tfle: %d rta: %d rtaf: %d rtf: %d ",
1209 1210 1211 1212 1213
		       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),
1214 1215
		       atomic_read(&n_rcu_torture_free));
	cnt += sprintf(&page[cnt], "rtmbe: %d rtbke: %ld rtbre: %ld ",
1216
		       atomic_read(&n_rcu_torture_mberror),
1217
		       n_rcu_torture_boost_ktrerror,
1218 1219
		       n_rcu_torture_boost_rterror);
	cnt += sprintf(&page[cnt], "rtbf: %ld rtb: %ld nt: %ld ",
1220 1221
		       n_rcu_torture_boost_failure,
		       n_rcu_torture_boosts,
1222
		       n_rcu_torture_timers);
1223 1224 1225 1226 1227 1228 1229
	cnt += sprintf(&page[cnt],
		       "onoff: %ld/%ld:%ld/%ld %d,%d:%d,%d %lu:%lu (HZ=%d) ",
		       n_online_successes, n_online_attempts,
		       n_offline_successes, n_offline_attempts,
		       min_online, max_online,
		       min_offline, max_offline,
		       sum_online, sum_offline, HZ);
1230
	cnt += sprintf(&page[cnt], "barrier: %ld/%ld:%ld",
1231 1232 1233 1234
		       n_barrier_successes,
		       n_barrier_attempts,
		       n_rcu_torture_barrier_error);
	cnt += sprintf(&page[cnt], "\n%s%s ", torture_type, TORTURE_FLAG);
1235
	if (atomic_read(&n_rcu_torture_mberror) != 0 ||
1236
	    n_rcu_torture_barrier_error != 0 ||
1237 1238
	    n_rcu_torture_boost_ktrerror != 0 ||
	    n_rcu_torture_boost_rterror != 0 ||
1239 1240
	    n_rcu_torture_boost_failure != 0 ||
	    i > 1) {
1241
		cnt += sprintf(&page[cnt], "!!! ");
1242
		atomic_inc(&n_rcu_torture_error);
1243
		WARN_ON_ONCE(1);
1244
	}
1245 1246 1247
	cnt += sprintf(&page[cnt], "Reader Pipe: ");
	for (i = 0; i < RCU_TORTURE_PIPE_LEN + 1; i++)
		cnt += sprintf(&page[cnt], " %ld", pipesummary[i]);
1248
	cnt += sprintf(&page[cnt], "\n%s%s ", torture_type, TORTURE_FLAG);
1249
	cnt += sprintf(&page[cnt], "Reader Batch: ");
1250
	for (i = 0; i < RCU_TORTURE_PIPE_LEN + 1; i++)
1251
		cnt += sprintf(&page[cnt], " %ld", batchsummary[i]);
1252
	cnt += sprintf(&page[cnt], "\n%s%s ", torture_type, TORTURE_FLAG);
1253 1254 1255 1256 1257 1258
	cnt += sprintf(&page[cnt], "Free-Block Circulation: ");
	for (i = 0; i < RCU_TORTURE_PIPE_LEN + 1; i++) {
		cnt += sprintf(&page[cnt], " %d",
			       atomic_read(&rcu_torture_wcount[i]));
	}
	cnt += sprintf(&page[cnt], "\n");
1259
	if (cur_ops->stats)
1260
		cnt += cur_ops->stats(&page[cnt]);
1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277
	return cnt;
}

/*
 * 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)
{
	int cnt;

	cnt = rcu_torture_printk(printk_buf);
1278
	pr_alert("%s", printk_buf);
1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294
}

/*
 * Periodically prints torture statistics, if periodic statistics printing
 * was specified via the stat_interval module parameter.
 *
 * No need to worry about fullstop here, since this one doesn't reference
 * volatile state or register callbacks.
 */
static int
rcu_torture_stats(void *arg)
{
	VERBOSE_PRINTK_STRING("rcu_torture_stats task started");
	do {
		schedule_timeout_interruptible(stat_interval * HZ);
		rcu_torture_stats_print();
1295 1296
		rcutorture_shutdown_absorb("rcu_torture_stats");
	} while (!kthread_should_stop());
1297 1298 1299 1300
	VERBOSE_PRINTK_STRING("rcu_torture_stats task stopping");
	return 0;
}

1301 1302 1303 1304 1305
static int rcu_idle_cpu;	/* Force all torture tasks off this CPU */

/* Shuffle tasks such that we allow @rcu_idle_cpu to become idle. A special case
 * is when @rcu_idle_cpu = -1, when we allow the tasks to run on all CPUs.
 */
1306
static void rcu_torture_shuffle_tasks(void)
1307 1308 1309
{
	int i;

R
Rusty Russell 已提交
1310
	cpumask_setall(shuffle_tmp_mask);
1311
	get_online_cpus();
1312 1313

	/* No point in shuffling if there is only one online CPU (ex: UP) */
1314 1315 1316 1317
	if (num_online_cpus() == 1) {
		put_online_cpus();
		return;
	}
1318 1319

	if (rcu_idle_cpu != -1)
R
Rusty Russell 已提交
1320
		cpumask_clear_cpu(rcu_idle_cpu, shuffle_tmp_mask);
1321

R
Rusty Russell 已提交
1322
	set_cpus_allowed_ptr(current, shuffle_tmp_mask);
1323

1324
	if (reader_tasks) {
1325 1326
		for (i = 0; i < nrealreaders; i++)
			if (reader_tasks[i])
1327
				set_cpus_allowed_ptr(reader_tasks[i],
R
Rusty Russell 已提交
1328
						     shuffle_tmp_mask);
1329
	}
1330
	if (fakewriter_tasks) {
1331 1332
		for (i = 0; i < nfakewriters; i++)
			if (fakewriter_tasks[i])
1333
				set_cpus_allowed_ptr(fakewriter_tasks[i],
R
Rusty Russell 已提交
1334
						     shuffle_tmp_mask);
1335
	}
1336
	if (writer_task)
R
Rusty Russell 已提交
1337
		set_cpus_allowed_ptr(writer_task, shuffle_tmp_mask);
1338
	if (stats_task)
R
Rusty Russell 已提交
1339
		set_cpus_allowed_ptr(stats_task, shuffle_tmp_mask);
1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358
	if (stutter_task)
		set_cpus_allowed_ptr(stutter_task, shuffle_tmp_mask);
	if (fqs_task)
		set_cpus_allowed_ptr(fqs_task, shuffle_tmp_mask);
	if (shutdown_task)
		set_cpus_allowed_ptr(shutdown_task, shuffle_tmp_mask);
#ifdef CONFIG_HOTPLUG_CPU
	if (onoff_task)
		set_cpus_allowed_ptr(onoff_task, shuffle_tmp_mask);
#endif /* #ifdef CONFIG_HOTPLUG_CPU */
	if (stall_task)
		set_cpus_allowed_ptr(stall_task, shuffle_tmp_mask);
	if (barrier_cbs_tasks)
		for (i = 0; i < n_barrier_cbs; i++)
			if (barrier_cbs_tasks[i])
				set_cpus_allowed_ptr(barrier_cbs_tasks[i],
						     shuffle_tmp_mask);
	if (barrier_task)
		set_cpus_allowed_ptr(barrier_task, shuffle_tmp_mask);
1359 1360 1361 1362 1363 1364

	if (rcu_idle_cpu == -1)
		rcu_idle_cpu = num_online_cpus() - 1;
	else
		rcu_idle_cpu--;

1365
	put_online_cpus();
1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378
}

/* Shuffle tasks across CPUs, with the intent of allowing each CPU in the
 * system to become idle at a time and cut off its timer ticks. This is meant
 * to test the support for such tickless idle CPU in RCU.
 */
static int
rcu_torture_shuffle(void *arg)
{
	VERBOSE_PRINTK_STRING("rcu_torture_shuffle task started");
	do {
		schedule_timeout_interruptible(shuffle_interval * HZ);
		rcu_torture_shuffle_tasks();
1379 1380
		rcutorture_shutdown_absorb("rcu_torture_shuffle");
	} while (!kthread_should_stop());
1381 1382 1383 1384
	VERBOSE_PRINTK_STRING("rcu_torture_shuffle task stopping");
	return 0;
}

1385 1386 1387 1388 1389 1390 1391 1392 1393 1394
/* Cause the rcutorture test to "stutter", starting and stopping all
 * threads periodically.
 */
static int
rcu_torture_stutter(void *arg)
{
	VERBOSE_PRINTK_STRING("rcu_torture_stutter task started");
	do {
		schedule_timeout_interruptible(stutter * HZ);
		stutter_pause_test = 1;
1395
		if (!kthread_should_stop())
1396 1397
			schedule_timeout_interruptible(stutter * HZ);
		stutter_pause_test = 0;
1398 1399
		rcutorture_shutdown_absorb("rcu_torture_stutter");
	} while (!kthread_should_stop());
1400 1401 1402 1403
	VERBOSE_PRINTK_STRING("rcu_torture_stutter task stopping");
	return 0;
}

1404
static inline void
1405
rcu_torture_print_module_parms(struct rcu_torture_ops *cur_ops, char *tag)
1406
{
1407 1408 1409 1410 1411 1412 1413
	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 "
1414 1415
		 "stall_cpu=%d stall_cpu_holdoff=%d "
		 "n_barrier_cbs=%d "
1416 1417 1418 1419 1420 1421
		 "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,
1422 1423
		 stall_cpu, stall_cpu_holdoff,
		 n_barrier_cbs,
1424
		 onoff_interval, onoff_holdoff);
1425 1426
}

1427
static struct notifier_block rcutorture_shutdown_nb = {
1428 1429 1430
	.notifier_call = rcutorture_shutdown_notify,
};

1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444
static void rcutorture_booster_cleanup(int cpu)
{
	struct task_struct *t;

	if (boost_tasks[cpu] == NULL)
		return;
	mutex_lock(&boost_mutex);
	VERBOSE_PRINTK_STRING("Stopping rcu_torture_boost task");
	t = boost_tasks[cpu];
	boost_tasks[cpu] = NULL;
	mutex_unlock(&boost_mutex);

	/* This must be outside of the mutex, otherwise deadlock! */
	kthread_stop(t);
1445
	boost_tasks[cpu] = NULL;
1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457
}

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);
	VERBOSE_PRINTK_STRING("Creating rcu_torture_boost task");
E
Eric Dumazet 已提交
1458 1459 1460
	boost_tasks[cpu] = kthread_create_on_node(rcu_torture_boost, NULL,
						  cpu_to_node(cpu),
						  "rcu_torture_boost");
1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474
	if (IS_ERR(boost_tasks[cpu])) {
		retval = PTR_ERR(boost_tasks[cpu]);
		VERBOSE_PRINTK_STRING("rcu_torture_boost task create failed");
		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;
}

1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490
/*
 * Cause the rcutorture test to shutdown the system after the test has
 * run for the time specified by the shutdown_secs module parameter.
 */
static int
rcu_torture_shutdown(void *arg)
{
	long delta;
	unsigned long jiffies_snap;

	VERBOSE_PRINTK_STRING("rcu_torture_shutdown task started");
	jiffies_snap = ACCESS_ONCE(jiffies);
	while (ULONG_CMP_LT(jiffies_snap, shutdown_time) &&
	       !kthread_should_stop()) {
		delta = shutdown_time - jiffies_snap;
		if (verbose)
1491 1492 1493
			pr_alert("%s" TORTURE_FLAG
				 "rcu_torture_shutdown task: %lu jiffies remaining\n",
				 torture_type, delta);
1494 1495 1496
		schedule_timeout_interruptible(delta);
		jiffies_snap = ACCESS_ONCE(jiffies);
	}
1497
	if (kthread_should_stop()) {
1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510
		VERBOSE_PRINTK_STRING("rcu_torture_shutdown task stopping");
		return 0;
	}

	/* OK, shut down the system. */

	VERBOSE_PRINTK_STRING("rcu_torture_shutdown task shutting down system");
	shutdown_task = NULL;	/* Avoid self-kill deadlock. */
	rcu_torture_cleanup();	/* Get the success/failure message. */
	kernel_power_off();	/* Shut down the system. */
	return 0;
}

1511 1512 1513 1514 1515 1516
#ifdef CONFIG_HOTPLUG_CPU

/*
 * Execute random CPU-hotplug operations at the interval specified
 * by the onoff_interval.
 */
1517
static int __cpuinit
1518 1519 1520
rcu_torture_onoff(void *arg)
{
	int cpu;
1521
	unsigned long delta;
1522 1523
	int maxcpu = -1;
	DEFINE_RCU_RANDOM(rand);
1524
	int ret;
1525
	unsigned long starttime;
1526 1527 1528 1529 1530

	VERBOSE_PRINTK_STRING("rcu_torture_onoff task started");
	for_each_online_cpu(cpu)
		maxcpu = cpu;
	WARN_ON(maxcpu < 0);
1531 1532 1533 1534 1535
	if (onoff_holdoff > 0) {
		VERBOSE_PRINTK_STRING("rcu_torture_onoff begin holdoff");
		schedule_timeout_interruptible(onoff_holdoff * HZ);
		VERBOSE_PRINTK_STRING("rcu_torture_onoff end holdoff");
	}
1536 1537
	while (!kthread_should_stop()) {
		cpu = (rcu_random(&rand) >> 4) % (maxcpu + 1);
1538
		if (cpu_online(cpu) && cpu_is_hotpluggable(cpu)) {
1539
			if (verbose)
1540 1541 1542
				pr_alert("%s" TORTURE_FLAG
					 "rcu_torture_onoff task: offlining %d\n",
					 torture_type, cpu);
1543
			starttime = jiffies;
1544
			n_offline_attempts++;
1545 1546 1547 1548 1549 1550 1551
			ret = cpu_down(cpu);
			if (ret) {
				if (verbose)
					pr_alert("%s" TORTURE_FLAG
						 "rcu_torture_onoff task: offline %d failed: errno %d\n",
						 torture_type, cpu, ret);
			} else {
1552
				if (verbose)
1553 1554 1555
					pr_alert("%s" TORTURE_FLAG
						 "rcu_torture_onoff task: offlined %d\n",
						 torture_type, cpu);
1556
				n_offline_successes++;
1557 1558 1559 1560 1561 1562 1563 1564 1565 1566
				delta = jiffies - starttime;
				sum_offline += delta;
				if (min_offline < 0) {
					min_offline = delta;
					max_offline = delta;
				}
				if (min_offline > delta)
					min_offline = delta;
				if (max_offline < delta)
					max_offline = delta;
1567
			}
1568
		} else if (cpu_is_hotpluggable(cpu)) {
1569
			if (verbose)
1570 1571 1572
				pr_alert("%s" TORTURE_FLAG
					 "rcu_torture_onoff task: onlining %d\n",
					 torture_type, cpu);
1573
			starttime = jiffies;
1574 1575 1576
			n_online_attempts++;
			if (cpu_up(cpu) == 0) {
				if (verbose)
1577 1578 1579
					pr_alert("%s" TORTURE_FLAG
						 "rcu_torture_onoff task: onlined %d\n",
						 torture_type, cpu);
1580
				n_online_successes++;
1581 1582 1583 1584 1585 1586 1587 1588 1589 1590
				delta = jiffies - starttime;
				sum_online += delta;
				if (min_online < 0) {
					min_online = delta;
					max_online = delta;
				}
				if (min_online > delta)
					min_online = delta;
				if (max_online < delta)
					max_online = delta;
1591 1592 1593 1594 1595 1596 1597 1598
			}
		}
		schedule_timeout_interruptible(onoff_interval * HZ);
	}
	VERBOSE_PRINTK_STRING("rcu_torture_onoff task stopping");
	return 0;
}

1599
static int __cpuinit
1600 1601
rcu_torture_onoff_init(void)
{
1602 1603
	int ret;

1604 1605 1606 1607
	if (onoff_interval <= 0)
		return 0;
	onoff_task = kthread_run(rcu_torture_onoff, NULL, "rcu_torture_onoff");
	if (IS_ERR(onoff_task)) {
1608
		ret = PTR_ERR(onoff_task);
1609
		onoff_task = NULL;
1610
		return ret;
1611 1612 1613 1614 1615 1616 1617 1618 1619 1620
	}
	return 0;
}

static void rcu_torture_onoff_cleanup(void)
{
	if (onoff_task == NULL)
		return;
	VERBOSE_PRINTK_STRING("Stopping rcu_torture_onoff task");
	kthread_stop(onoff_task);
1621
	onoff_task = NULL;
1622 1623 1624 1625
}

#else /* #ifdef CONFIG_HOTPLUG_CPU */

1626
static int
1627 1628
rcu_torture_onoff_init(void)
{
1629
	return 0;
1630 1631 1632 1633 1634 1635 1636 1637
}

static void rcu_torture_onoff_cleanup(void)
{
}

#endif /* #else #ifdef CONFIG_HOTPLUG_CPU */

1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654
/*
 * CPU-stall kthread.  It waits as specified by stall_cpu_holdoff, then
 * induces a CPU stall for the time specified by stall_cpu.
 */
static int __cpuinit rcu_torture_stall(void *args)
{
	unsigned long stop_at;

	VERBOSE_PRINTK_STRING("rcu_torture_stall task started");
	if (stall_cpu_holdoff > 0) {
		VERBOSE_PRINTK_STRING("rcu_torture_stall begin holdoff");
		schedule_timeout_interruptible(stall_cpu_holdoff * HZ);
		VERBOSE_PRINTK_STRING("rcu_torture_stall end holdoff");
	}
	if (!kthread_should_stop()) {
		stop_at = get_seconds() + stall_cpu;
		/* RCU CPU stall is expected behavior in following code. */
1655
		pr_alert("rcu_torture_stall start.\n");
1656 1657 1658 1659 1660 1661
		rcu_read_lock();
		preempt_disable();
		while (ULONG_CMP_LT(get_seconds(), stop_at))
			continue;  /* Induce RCU CPU stall warning. */
		preempt_enable();
		rcu_read_unlock();
1662
		pr_alert("rcu_torture_stall end.\n");
1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692
	}
	rcutorture_shutdown_absorb("rcu_torture_stall");
	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)
{
	int ret;

	if (stall_cpu <= 0)
		return 0;
	stall_task = kthread_run(rcu_torture_stall, NULL, "rcu_torture_stall");
	if (IS_ERR(stall_task)) {
		ret = PTR_ERR(stall_task);
		stall_task = NULL;
		return ret;
	}
	return 0;
}

/* Clean up after the CPU-stall kthread, if one was spawned. */
static void rcu_torture_stall_cleanup(void)
{
	if (stall_task == NULL)
		return;
	VERBOSE_PRINTK_STRING("Stopping rcu_torture_stall_task.");
	kthread_stop(stall_task);
1693
	stall_task = NULL;
1694 1695
}

1696 1697 1698 1699 1700 1701 1702 1703 1704 1705
/* 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;
1706
	bool lastphase = 0;
1707 1708 1709 1710 1711 1712 1713
	struct rcu_head rcu;

	init_rcu_head_on_stack(&rcu);
	VERBOSE_PRINTK_STRING("rcu_torture_barrier_cbs task started");
	set_user_nice(current, 19);
	do {
		wait_event(barrier_cbs_wq[myid],
1714
			   barrier_phase != lastphase ||
1715 1716
			   kthread_should_stop() ||
			   fullstop != FULLSTOP_DONTSTOP);
1717 1718
		lastphase = barrier_phase;
		smp_mb(); /* ensure barrier_phase load before ->call(). */
1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742
		if (kthread_should_stop() || fullstop != FULLSTOP_DONTSTOP)
			break;
		cur_ops->call(&rcu, rcu_torture_barrier_cbf);
		if (atomic_dec_and_test(&barrier_cbs_count))
			wake_up(&barrier_wq);
	} while (!kthread_should_stop() && fullstop == FULLSTOP_DONTSTOP);
	VERBOSE_PRINTK_STRING("rcu_torture_barrier_cbs task stopping");
	rcutorture_shutdown_absorb("rcu_torture_barrier_cbs");
	while (!kthread_should_stop())
		schedule_timeout_interruptible(1);
	cur_ops->cb_barrier();
	destroy_rcu_head_on_stack(&rcu);
	return 0;
}

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

	VERBOSE_PRINTK_STRING("rcu_torture_barrier task starting");
	do {
		atomic_set(&barrier_cbs_invoked, 0);
		atomic_set(&barrier_cbs_count, n_barrier_cbs);
1743 1744
		smp_mb(); /* Ensure barrier_phase after prior assignments. */
		barrier_phase = !barrier_phase;
1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762
		for (i = 0; i < n_barrier_cbs; i++)
			wake_up(&barrier_cbs_wq[i]);
		wait_event(barrier_wq,
			   atomic_read(&barrier_cbs_count) == 0 ||
			   kthread_should_stop() ||
			   fullstop != FULLSTOP_DONTSTOP);
		if (kthread_should_stop() || fullstop != FULLSTOP_DONTSTOP)
			break;
		n_barrier_attempts++;
		cur_ops->cb_barrier();
		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);
	} while (!kthread_should_stop() && fullstop == FULLSTOP_DONTSTOP);
	VERBOSE_PRINTK_STRING("rcu_torture_barrier task stopping");
1763
	rcutorture_shutdown_absorb("rcu_torture_barrier");
1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777
	while (!kthread_should_stop())
		schedule_timeout_interruptible(1);
	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) {
1778 1779 1780 1781 1782 1783
		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);
1784 1785 1786 1787 1788 1789 1790 1791 1792 1793
		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);
1794
	if (barrier_cbs_tasks == NULL || !barrier_cbs_wq)
1795 1796 1797 1798
		return -ENOMEM;
	for (i = 0; i < n_barrier_cbs; i++) {
		init_waitqueue_head(&barrier_cbs_wq[i]);
		barrier_cbs_tasks[i] = kthread_run(rcu_torture_barrier_cbs,
1799
						   (void *)(long)i,
1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 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
						   "rcu_torture_barrier_cbs");
		if (IS_ERR(barrier_cbs_tasks[i])) {
			ret = PTR_ERR(barrier_cbs_tasks[i]);
			VERBOSE_PRINTK_ERRSTRING("Failed to create rcu_torture_barrier_cbs");
			barrier_cbs_tasks[i] = NULL;
			return ret;
		}
	}
	barrier_task = kthread_run(rcu_torture_barrier, NULL,
				   "rcu_torture_barrier");
	if (IS_ERR(barrier_task)) {
		ret = PTR_ERR(barrier_task);
		VERBOSE_PRINTK_ERRSTRING("Failed to create rcu_torture_barrier");
		barrier_task = NULL;
	}
	return 0;
}

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

	if (barrier_task != NULL) {
		VERBOSE_PRINTK_STRING("Stopping rcu_torture_barrier task");
		kthread_stop(barrier_task);
		barrier_task = NULL;
	}
	if (barrier_cbs_tasks != NULL) {
		for (i = 0; i < n_barrier_cbs; i++) {
			if (barrier_cbs_tasks[i] != NULL) {
				VERBOSE_PRINTK_STRING("Stopping rcu_torture_barrier_cbs task");
				kthread_stop(barrier_cbs_tasks[i]);
				barrier_cbs_tasks[i] = NULL;
			}
		}
		kfree(barrier_cbs_tasks);
		barrier_cbs_tasks = NULL;
	}
	if (barrier_cbs_wq != NULL) {
		kfree(barrier_cbs_wq);
		barrier_cbs_wq = NULL;
	}
}

1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867
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,
};

1868 1869 1870 1871 1872
static void
rcu_torture_cleanup(void)
{
	int i;

1873
	mutex_lock(&fullstop_mutex);
1874
	rcutorture_record_test_transition();
1875
	if (fullstop == FULLSTOP_SHUTDOWN) {
1876
		pr_warn(/* but going down anyway, so... */
1877
		       "Concurrent 'rmmod rcutorture' and shutdown illegal!\n");
1878
		mutex_unlock(&fullstop_mutex);
1879
		schedule_timeout_uninterruptible(10);
1880 1881 1882 1883
		if (cur_ops->cb_barrier != NULL)
			cur_ops->cb_barrier();
		return;
	}
1884
	fullstop = FULLSTOP_RMMOD;
1885
	mutex_unlock(&fullstop_mutex);
1886
	unregister_reboot_notifier(&rcutorture_shutdown_nb);
1887
	rcu_torture_barrier_cleanup();
1888
	rcu_torture_stall_cleanup();
1889 1890 1891 1892 1893
	if (stutter_task) {
		VERBOSE_PRINTK_STRING("Stopping rcu_torture_stutter task");
		kthread_stop(stutter_task);
	}
	stutter_task = NULL;
1894
	if (shuffler_task) {
1895 1896
		VERBOSE_PRINTK_STRING("Stopping rcu_torture_shuffle task");
		kthread_stop(shuffler_task);
R
Rusty Russell 已提交
1897
		free_cpumask_var(shuffle_tmp_mask);
1898 1899 1900
	}
	shuffler_task = NULL;

1901
	if (writer_task) {
1902 1903 1904 1905 1906
		VERBOSE_PRINTK_STRING("Stopping rcu_torture_writer task");
		kthread_stop(writer_task);
	}
	writer_task = NULL;

1907
	if (reader_tasks) {
1908
		for (i = 0; i < nrealreaders; i++) {
1909
			if (reader_tasks[i]) {
1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920
				VERBOSE_PRINTK_STRING(
					"Stopping rcu_torture_reader task");
				kthread_stop(reader_tasks[i]);
			}
			reader_tasks[i] = NULL;
		}
		kfree(reader_tasks);
		reader_tasks = NULL;
	}
	rcu_torture_current = NULL;

1921
	if (fakewriter_tasks) {
1922
		for (i = 0; i < nfakewriters; i++) {
1923
			if (fakewriter_tasks[i]) {
1924 1925 1926 1927 1928 1929 1930 1931 1932 1933
				VERBOSE_PRINTK_STRING(
					"Stopping rcu_torture_fakewriter task");
				kthread_stop(fakewriter_tasks[i]);
			}
			fakewriter_tasks[i] = NULL;
		}
		kfree(fakewriter_tasks);
		fakewriter_tasks = NULL;
	}

1934
	if (stats_task) {
1935 1936 1937 1938 1939
		VERBOSE_PRINTK_STRING("Stopping rcu_torture_stats task");
		kthread_stop(stats_task);
	}
	stats_task = NULL;

1940 1941 1942 1943 1944
	if (fqs_task) {
		VERBOSE_PRINTK_STRING("Stopping rcu_torture_fqs task");
		kthread_stop(fqs_task);
	}
	fqs_task = NULL;
1945 1946 1947 1948 1949 1950
	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);
	}
1951 1952 1953 1954
	if (shutdown_task != NULL) {
		VERBOSE_PRINTK_STRING("Stopping rcu_torture_shutdown task");
		kthread_stop(shutdown_task);
	}
1955
	shutdown_task = NULL;
1956
	rcu_torture_onoff_cleanup();
1957

1958
	/* Wait for all RCU callbacks to fire.  */
1959 1960 1961

	if (cur_ops->cb_barrier != NULL)
		cur_ops->cb_barrier();
1962 1963

	rcu_torture_stats_print();  /* -After- the stats thread is stopped! */
1964

1965
	if (atomic_read(&n_rcu_torture_error) || n_rcu_torture_barrier_error)
1966
		rcu_torture_print_module_parms(cur_ops, "End of test: FAILURE");
1967 1968 1969 1970
	else if (n_online_successes != n_online_attempts ||
		 n_offline_successes != n_offline_attempts)
		rcu_torture_print_module_parms(cur_ops,
					       "End of test: RCU_HOTPLUG");
1971
	else
1972
		rcu_torture_print_module_parms(cur_ops, "End of test: SUCCESS");
1973 1974
}

1975
static int __init
1976 1977 1978 1979 1980
rcu_torture_init(void)
{
	int i;
	int cpu;
	int firsterr = 0;
1981
	int retval;
1982
	static struct rcu_torture_ops *torture_ops[] =
1983
		{ &rcu_ops, &rcu_sync_ops, &rcu_expedited_ops,
1984
		  &rcu_bh_ops, &rcu_bh_sync_ops, &rcu_bh_expedited_ops,
1985 1986
		  &srcu_ops, &srcu_sync_ops, &srcu_expedited_ops,
		  &srcu_raw_ops, &srcu_raw_sync_ops,
1987
		  &sched_ops, &sched_sync_ops, &sched_expedited_ops, };
1988

1989 1990
	mutex_lock(&fullstop_mutex);

1991
	/* Process args and tell the world that the torturer is on the job. */
1992
	for (i = 0; i < ARRAY_SIZE(torture_ops); i++) {
1993
		cur_ops = torture_ops[i];
1994
		if (strcmp(torture_type, cur_ops->name) == 0)
1995 1996
			break;
	}
1997
	if (i == ARRAY_SIZE(torture_ops)) {
1998 1999 2000
		pr_alert("rcu-torture: invalid torture type: \"%s\"\n",
			 torture_type);
		pr_alert("rcu-torture types:");
2001
		for (i = 0; i < ARRAY_SIZE(torture_ops); i++)
2002 2003
			pr_alert(" %s", torture_ops[i]->name);
		pr_alert("\n");
2004
		mutex_unlock(&fullstop_mutex);
2005
		return -EINVAL;
2006
	}
2007
	if (cur_ops->fqs == NULL && fqs_duration != 0) {
2008
		pr_alert("rcu-torture: ->fqs NULL and non-zero fqs_duration, fqs disabled.\n");
2009 2010
		fqs_duration = 0;
	}
2011
	if (cur_ops->init)
2012 2013
		cur_ops->init(); /* no "goto unwind" prior to this point!!! */

2014 2015 2016 2017
	if (nreaders >= 0)
		nrealreaders = nreaders;
	else
		nrealreaders = 2 * num_online_cpus();
2018
	rcu_torture_print_module_parms(cur_ops, "Start of test");
2019
	fullstop = FULLSTOP_DONTSTOP;
2020 2021 2022 2023

	/* Set up the freelist. */

	INIT_LIST_HEAD(&rcu_torture_freelist);
2024
	for (i = 0; i < ARRAY_SIZE(rcu_tortures); i++) {
2025
		rcu_tortures[i].rtort_mbtest = 0;
2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036
		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);
2037 2038
	atomic_set(&n_rcu_torture_mberror, 0);
	atomic_set(&n_rcu_torture_error, 0);
2039
	n_rcu_torture_barrier_error = 0;
2040 2041 2042 2043
	n_rcu_torture_boost_ktrerror = 0;
	n_rcu_torture_boost_rterror = 0;
	n_rcu_torture_boost_failure = 0;
	n_rcu_torture_boosts = 0;
2044 2045
	for (i = 0; i < RCU_TORTURE_PIPE_LEN + 1; i++)
		atomic_set(&rcu_torture_wcount[i], 0);
2046
	for_each_possible_cpu(cpu) {
2047 2048 2049 2050 2051 2052 2053 2054 2055
		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. */

	VERBOSE_PRINTK_STRING("Creating rcu_torture_writer task");
2056 2057
	writer_task = kthread_create(rcu_torture_writer, NULL,
				     "rcu_torture_writer");
2058 2059 2060 2061 2062 2063
	if (IS_ERR(writer_task)) {
		firsterr = PTR_ERR(writer_task);
		VERBOSE_PRINTK_ERRSTRING("Failed to create writer");
		writer_task = NULL;
		goto unwind;
	}
2064
	wake_up_process(writer_task);
2065
	fakewriter_tasks = kzalloc(nfakewriters * sizeof(fakewriter_tasks[0]),
2066
				   GFP_KERNEL);
2067 2068 2069 2070 2071 2072 2073 2074
	if (fakewriter_tasks == NULL) {
		VERBOSE_PRINTK_ERRSTRING("out of memory");
		firsterr = -ENOMEM;
		goto unwind;
	}
	for (i = 0; i < nfakewriters; i++) {
		VERBOSE_PRINTK_STRING("Creating rcu_torture_fakewriter task");
		fakewriter_tasks[i] = kthread_run(rcu_torture_fakewriter, NULL,
2075
						  "rcu_torture_fakewriter");
2076 2077 2078 2079 2080 2081 2082
		if (IS_ERR(fakewriter_tasks[i])) {
			firsterr = PTR_ERR(fakewriter_tasks[i]);
			VERBOSE_PRINTK_ERRSTRING("Failed to create fakewriter");
			fakewriter_tasks[i] = NULL;
			goto unwind;
		}
	}
2083
	reader_tasks = kzalloc(nrealreaders * sizeof(reader_tasks[0]),
2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111
			       GFP_KERNEL);
	if (reader_tasks == NULL) {
		VERBOSE_PRINTK_ERRSTRING("out of memory");
		firsterr = -ENOMEM;
		goto unwind;
	}
	for (i = 0; i < nrealreaders; i++) {
		VERBOSE_PRINTK_STRING("Creating rcu_torture_reader task");
		reader_tasks[i] = kthread_run(rcu_torture_reader, NULL,
					      "rcu_torture_reader");
		if (IS_ERR(reader_tasks[i])) {
			firsterr = PTR_ERR(reader_tasks[i]);
			VERBOSE_PRINTK_ERRSTRING("Failed to create reader");
			reader_tasks[i] = NULL;
			goto unwind;
		}
	}
	if (stat_interval > 0) {
		VERBOSE_PRINTK_STRING("Creating rcu_torture_stats task");
		stats_task = kthread_run(rcu_torture_stats, NULL,
					"rcu_torture_stats");
		if (IS_ERR(stats_task)) {
			firsterr = PTR_ERR(stats_task);
			VERBOSE_PRINTK_ERRSTRING("Failed to create stats");
			stats_task = NULL;
			goto unwind;
		}
	}
2112 2113
	if (test_no_idle_hz) {
		rcu_idle_cpu = num_online_cpus() - 1;
R
Rusty Russell 已提交
2114 2115 2116 2117 2118 2119 2120

		if (!alloc_cpumask_var(&shuffle_tmp_mask, GFP_KERNEL)) {
			firsterr = -ENOMEM;
			VERBOSE_PRINTK_ERRSTRING("Failed to alloc mask");
			goto unwind;
		}

2121 2122 2123 2124
		/* Create the shuffler thread */
		shuffler_task = kthread_run(rcu_torture_shuffle, NULL,
					  "rcu_torture_shuffle");
		if (IS_ERR(shuffler_task)) {
R
Rusty Russell 已提交
2125
			free_cpumask_var(shuffle_tmp_mask);
2126 2127 2128 2129 2130 2131
			firsterr = PTR_ERR(shuffler_task);
			VERBOSE_PRINTK_ERRSTRING("Failed to create shuffler");
			shuffler_task = NULL;
			goto unwind;
		}
	}
2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144
	if (stutter < 0)
		stutter = 0;
	if (stutter) {
		/* Create the stutter thread */
		stutter_task = kthread_run(rcu_torture_stutter, NULL,
					  "rcu_torture_stutter");
		if (IS_ERR(stutter_task)) {
			firsterr = PTR_ERR(stutter_task);
			VERBOSE_PRINTK_ERRSTRING("Failed to create stutter");
			stutter_task = NULL;
			goto unwind;
		}
	}
2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157
	if (fqs_duration < 0)
		fqs_duration = 0;
	if (fqs_duration) {
		/* Create the stutter thread */
		fqs_task = kthread_run(rcu_torture_fqs, NULL,
				       "rcu_torture_fqs");
		if (IS_ERR(fqs_task)) {
			firsterr = PTR_ERR(fqs_task);
			VERBOSE_PRINTK_ERRSTRING("Failed to create fqs");
			fqs_task = NULL;
			goto unwind;
		}
	}
2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176
	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. */
			retval = rcutorture_booster_init(i);
			if (retval < 0) {
				firsterr = retval;
				goto unwind;
			}
		}
	}
2177 2178
	if (shutdown_secs > 0) {
		shutdown_time = jiffies + shutdown_secs * HZ;
2179 2180
		shutdown_task = kthread_create(rcu_torture_shutdown, NULL,
					       "rcu_torture_shutdown");
2181 2182 2183 2184 2185 2186
		if (IS_ERR(shutdown_task)) {
			firsterr = PTR_ERR(shutdown_task);
			VERBOSE_PRINTK_ERRSTRING("Failed to create shutdown");
			shutdown_task = NULL;
			goto unwind;
		}
2187
		wake_up_process(shutdown_task);
2188
	}
2189 2190 2191 2192 2193
	i = rcu_torture_onoff_init();
	if (i != 0) {
		firsterr = i;
		goto unwind;
	}
2194
	register_reboot_notifier(&rcutorture_shutdown_nb);
2195 2196 2197 2198 2199
	i = rcu_torture_stall_init();
	if (i != 0) {
		firsterr = i;
		goto unwind;
	}
2200 2201 2202 2203 2204
	retval = rcu_torture_barrier_init();
	if (retval != 0) {
		firsterr = retval;
		goto unwind;
	}
2205
	rcutorture_record_test_transition();
2206
	mutex_unlock(&fullstop_mutex);
2207 2208 2209
	return 0;

unwind:
2210
	mutex_unlock(&fullstop_mutex);
2211 2212 2213 2214 2215 2216
	rcu_torture_cleanup();
	return firsterr;
}

module_init(rcu_torture_init);
module_exit(rcu_torture_cleanup);