Merge branch 'core-rcu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull RCU updates from Ingo Molnar:
 "The main changes are:

   - Documentation updates, including fixes to the design-level
     requirements documentation and a fixed version of the design-level
     data-structure documentation.  These fixes include removing
     cartoons and getting rid of the html/htmlx duplication.

   - Further improvements to the new-age expedited grace periods.

   - Miscellaneous fixes.

   - Torture-test changes, including a new rcuperf module for measuring
     RCU grace-period performance and scalability, which is useful for
     the expedited-grace-period changes"

* 'core-rcu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (56 commits)
  rcutorture: Add boot-time adjustment of leaf fanout
  rcutorture: Add irqs-disabled test for call_rcu()
  rcutorture: Dump trace buffer upon shutdown
  rcutorture: Don't rebuild identical kernel
  rcutorture: Add OS-jitter capability
  documentation: Add documentation for RCU's major data structures
  rcutorture: Convert test duration to seconds early
  torture: Kill qemu, not parent process
  torture: Clarify refusal to run more than one torture test
  rcutorture: Consider FROZEN hotplug notifier transitions
  rcutorture: Remove redundant initialization to zero
  rcuperf: Do not wake up shutdown wait queue if "shutdown" is false.
  rcutorture: Add largish-system rcuperf scenario
  rcutorture: Avoid RCU CPU stall warning and RT throttling
  rcutorture: Add rcuperf holdoff boot parameter to reduce interference
  rcutorture: Make scripts analyze rcuperf trace data, if present
  rcutorture: Make rcuperf collect expedited event-trace data
  rcutorture: Print measure of batching efficiency
  rcutorture: Set rcuperf writer kthreads to real-time priority
  rcutorture: Bind rcuperf reader/writer kthreads to CPUs
  ...

9 files changed, 916 insertions, 149 deletions

diff --git a/kernel/rcu/Makefile b/kernel/rcu/Makefile
index 032b2c015beb..18dfc485225c 100644
--- a/kernel/rcu/Makefile
+++ b/kernel/rcu/Makefile
@@ -5,6 +5,7 @@ KCOV_INSTRUMENT := n
 obj-y += update.o sync.o
 obj-$(CONFIG_SRCU) += srcu.o
 obj-$(CONFIG_RCU_TORTURE_TEST) += rcutorture.o
+obj-$(CONFIG_RCU_PERF_TEST) += rcuperf.o
 obj-$(CONFIG_TREE_RCU) += tree.o
 obj-$(CONFIG_PREEMPT_RCU) += tree.o
 obj-$(CONFIG_TREE_RCU_TRACE) += tree_trace.o
diff --git a/kernel/rcu/rcuperf.c b/kernel/rcu/rcuperf.c
new file mode 100644
index 000000000000..3cee0d8393ed
--- /dev/null
+++ b/kernel/rcu/rcuperf.c
@@ -0,0 +1,655 @@
+/*
+ * Read-Copy Update module-based performance-test facility
+ *
+ * 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, you can access it online at
+ * http://www.gnu.org/licenses/gpl-2.0.html.
+ *
+ * Copyright (C) IBM Corporation, 2015
+ *
+ * Authors: Paul E. McKenney <paulmck@us.ibm.com>
+ */
+#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>
+#include <linux/atomic.h>
+#include <linux/bitops.h>
+#include <linux/completion.h>
+#include <linux/moduleparam.h>
+#include <linux/percpu.h>
+#include <linux/notifier.h>
+#include <linux/reboot.h>
+#include <linux/freezer.h>
+#include <linux/cpu.h>
+#include <linux/delay.h>
+#include <linux/stat.h>
+#include <linux/srcu.h>
+#include <linux/slab.h>
+#include <asm/byteorder.h>
+#include <linux/torture.h>
+#include <linux/vmalloc.h>
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Paul E. McKenney <paulmck@linux.vnet.ibm.com>");
+
+#define PERF_FLAG "-perf:"
+#define PERFOUT_STRING(s) \
+	pr_alert("%s" PERF_FLAG s "\n", perf_type)
+#define VERBOSE_PERFOUT_STRING(s) \
+	do { if (verbose) pr_alert("%s" PERF_FLAG " %s\n", perf_type, s); } while (0)
+#define VERBOSE_PERFOUT_ERRSTRING(s) \
+	do { if (verbose) pr_alert("%s" PERF_FLAG "!!! %s\n", perf_type, s); } while (0)
+
+torture_param(bool, gp_exp, true, "Use expedited GP wait primitives");
+torture_param(int, holdoff, 10, "Holdoff time before test start (s)");
+torture_param(int, nreaders, -1, "Number of RCU reader threads");
+torture_param(int, nwriters, -1, "Number of RCU updater threads");
+torture_param(bool, shutdown, false, "Shutdown at end of performance tests.");
+torture_param(bool, verbose, true, "Enable verbose debugging printk()s");
+
+static char *perf_type = "rcu";
+module_param(perf_type, charp, 0444);
+MODULE_PARM_DESC(perf_type, "Type of RCU to performance-test (rcu, rcu_bh, ...)");
+
+static int nrealreaders;
+static int nrealwriters;
+static struct task_struct **writer_tasks;
+static struct task_struct **reader_tasks;
+static struct task_struct *shutdown_task;
+
+static u64 **writer_durations;
+static int *writer_n_durations;
+static atomic_t n_rcu_perf_reader_started;
+static atomic_t n_rcu_perf_writer_started;
+static atomic_t n_rcu_perf_writer_finished;
+static wait_queue_head_t shutdown_wq;
+static u64 t_rcu_perf_writer_started;
+static u64 t_rcu_perf_writer_finished;
+static unsigned long b_rcu_perf_writer_started;
+static unsigned long b_rcu_perf_writer_finished;
+
+static int rcu_perf_writer_state;
+#define RTWS_INIT		0
+#define RTWS_EXP_SYNC		1
+#define RTWS_SYNC		2
+#define RTWS_IDLE		2
+#define RTWS_STOPPING		3
+
+#define MAX_MEAS 10000
+#define MIN_MEAS 100
+
+#if defined(MODULE) || defined(CONFIG_RCU_PERF_TEST_RUNNABLE)
+#define RCUPERF_RUNNABLE_INIT 1
+#else
+#define RCUPERF_RUNNABLE_INIT 0
+#endif
+static int perf_runnable = RCUPERF_RUNNABLE_INIT;
+module_param(perf_runnable, int, 0444);
+MODULE_PARM_DESC(perf_runnable, "Start rcuperf at boot");
+
+/*
+ * Operations vector for selecting different types of tests.
+ */
+
+struct rcu_perf_ops {
+	int ptype;
+	void (*init)(void);
+	void (*cleanup)(void);
+	int (*readlock)(void);
+	void (*readunlock)(int idx);
+	unsigned long (*started)(void);
+	unsigned long (*completed)(void);
+	unsigned long (*exp_completed)(void);
+	void (*sync)(void);
+	void (*exp_sync)(void);
+	const char *name;
+};
+
+static struct rcu_perf_ops *cur_ops;
+
+/*
+ * Definitions for rcu perf testing.
+ */
+
+static int rcu_perf_read_lock(void) __acquires(RCU)
+{
+	rcu_read_lock();
+	return 0;
+}
+
+static void rcu_perf_read_unlock(int idx) __releases(RCU)
+{
+	rcu_read_unlock();
+}
+
+static unsigned long __maybe_unused rcu_no_completed(void)
+{
+	return 0;
+}
+
+static void rcu_sync_perf_init(void)
+{
+}
+
+static struct rcu_perf_ops rcu_ops = {
+	.ptype		= RCU_FLAVOR,
+	.init		= rcu_sync_perf_init,
+	.readlock	= rcu_perf_read_lock,
+	.readunlock	= rcu_perf_read_unlock,
+	.started	= rcu_batches_started,
+	.completed	= rcu_batches_completed,
+	.exp_completed	= rcu_exp_batches_completed,
+	.sync		= synchronize_rcu,
+	.exp_sync	= synchronize_rcu_expedited,
+	.name		= "rcu"
+};
+
+/*
+ * Definitions for rcu_bh perf testing.
+ */
+
+static int rcu_bh_perf_read_lock(void) __acquires(RCU_BH)
+{
+	rcu_read_lock_bh();
+	return 0;
+}
+
+static void rcu_bh_perf_read_unlock(int idx) __releases(RCU_BH)
+{
+	rcu_read_unlock_bh();
+}
+
+static struct rcu_perf_ops rcu_bh_ops = {
+	.ptype		= RCU_BH_FLAVOR,
+	.init		= rcu_sync_perf_init,
+	.readlock	= rcu_bh_perf_read_lock,
+	.readunlock	= rcu_bh_perf_read_unlock,
+	.started	= rcu_batches_started_bh,
+	.completed	= rcu_batches_completed_bh,
+	.exp_completed	= rcu_exp_batches_completed_sched,
+	.sync		= synchronize_rcu_bh,
+	.exp_sync	= synchronize_rcu_bh_expedited,
+	.name		= "rcu_bh"
+};
+
+/*
+ * Definitions for srcu perf testing.
+ */
+
+DEFINE_STATIC_SRCU(srcu_ctl_perf);
+static struct srcu_struct *srcu_ctlp = &srcu_ctl_perf;
+
+static int srcu_perf_read_lock(void) __acquires(srcu_ctlp)
+{
+	return srcu_read_lock(srcu_ctlp);
+}
+
+static void srcu_perf_read_unlock(int idx) __releases(srcu_ctlp)
+{
+	srcu_read_unlock(srcu_ctlp, idx);
+}
+
+static unsigned long srcu_perf_completed(void)
+{
+	return srcu_batches_completed(srcu_ctlp);
+}
+
+static void srcu_perf_synchronize(void)
+{
+	synchronize_srcu(srcu_ctlp);
+}
+
+static void srcu_perf_synchronize_expedited(void)
+{
+	synchronize_srcu_expedited(srcu_ctlp);
+}
+
+static struct rcu_perf_ops srcu_ops = {
+	.ptype		= SRCU_FLAVOR,
+	.init		= rcu_sync_perf_init,
+	.readlock	= srcu_perf_read_lock,
+	.readunlock	= srcu_perf_read_unlock,
+	.started	= NULL,
+	.completed	= srcu_perf_completed,
+	.exp_completed	= srcu_perf_completed,
+	.sync		= srcu_perf_synchronize,
+	.exp_sync	= srcu_perf_synchronize_expedited,
+	.name		= "srcu"
+};
+
+/*
+ * Definitions for sched perf testing.
+ */
+
+static int sched_perf_read_lock(void)
+{
+	preempt_disable();
+	return 0;
+}
+
+static void sched_perf_read_unlock(int idx)
+{
+	preempt_enable();
+}
+
+static struct rcu_perf_ops sched_ops = {
+	.ptype		= RCU_SCHED_FLAVOR,
+	.init		= rcu_sync_perf_init,
+	.readlock	= sched_perf_read_lock,
+	.readunlock	= sched_perf_read_unlock,
+	.started	= rcu_batches_started_sched,
+	.completed	= rcu_batches_completed_sched,
+	.exp_completed	= rcu_exp_batches_completed_sched,
+	.sync		= synchronize_sched,
+	.exp_sync	= synchronize_sched_expedited,
+	.name		= "sched"
+};
+
+#ifdef CONFIG_TASKS_RCU
+
+/*
+ * Definitions for RCU-tasks perf testing.
+ */
+
+static int tasks_perf_read_lock(void)
+{
+	return 0;
+}
+
+static void tasks_perf_read_unlock(int idx)
+{
+}
+
+static struct rcu_perf_ops tasks_ops = {
+	.ptype		= RCU_TASKS_FLAVOR,
+	.init		= rcu_sync_perf_init,
+	.readlock	= tasks_perf_read_lock,
+	.readunlock	= tasks_perf_read_unlock,
+	.started	= rcu_no_completed,
+	.completed	= rcu_no_completed,
+	.sync		= synchronize_rcu_tasks,
+	.exp_sync	= synchronize_rcu_tasks,
+	.name		= "tasks"
+};
+
+#define RCUPERF_TASKS_OPS &tasks_ops,
+
+static bool __maybe_unused torturing_tasks(void)
+{
+	return cur_ops == &tasks_ops;
+}
+
+#else /* #ifdef CONFIG_TASKS_RCU */
+
+#define RCUPERF_TASKS_OPS
+
+static bool __maybe_unused torturing_tasks(void)
+{
+	return false;
+}
+
+#endif /* #else #ifdef CONFIG_TASKS_RCU */
+
+/*
+ * If performance tests complete, wait for shutdown to commence.
+ */
+static void rcu_perf_wait_shutdown(void)
+{
+	cond_resched_rcu_qs();
+	if (atomic_read(&n_rcu_perf_writer_finished) < nrealwriters)
+		return;
+	while (!torture_must_stop())
+		schedule_timeout_uninterruptible(1);
+}
+
+/*
+ * RCU perf reader kthread.  Repeatedly does empty RCU read-side
+ * critical section, minimizing update-side interference.
+ */
+static int
+rcu_perf_reader(void *arg)
+{
+	unsigned long flags;
+	int idx;
+	long me = (long)arg;
+
+	VERBOSE_PERFOUT_STRING("rcu_perf_reader task started");
+	set_cpus_allowed_ptr(current, cpumask_of(me % nr_cpu_ids));
+	set_user_nice(current, MAX_NICE);
+	atomic_inc(&n_rcu_perf_reader_started);
+
+	do {
+		local_irq_save(flags);
+		idx = cur_ops->readlock();
+		cur_ops->readunlock(idx);
+		local_irq_restore(flags);
+		rcu_perf_wait_shutdown();
+	} while (!torture_must_stop());
+	torture_kthread_stopping("rcu_perf_reader");
+	return 0;
+}
+
+/*
+ * RCU perf writer kthread.  Repeatedly does a grace period.
+ */
+static int
+rcu_perf_writer(void *arg)
+{
+	int i = 0;
+	int i_max;
+	long me = (long)arg;
+	struct sched_param sp;
+	bool started = false, done = false, alldone = false;
+	u64 t;
+	u64 *wdp;
+	u64 *wdpp = writer_durations[me];
+
+	VERBOSE_PERFOUT_STRING("rcu_perf_writer task started");
+	WARN_ON(rcu_gp_is_expedited() && !rcu_gp_is_normal() && !gp_exp);
+	WARN_ON(rcu_gp_is_normal() && gp_exp);
+	WARN_ON(!wdpp);
+	set_cpus_allowed_ptr(current, cpumask_of(me % nr_cpu_ids));
+	sp.sched_priority = 1;
+	sched_setscheduler_nocheck(current, SCHED_FIFO, &sp);
+
+	if (holdoff)
+		schedule_timeout_uninterruptible(holdoff * HZ);
+
+	t = ktime_get_mono_fast_ns();
+	if (atomic_inc_return(&n_rcu_perf_writer_started) >= nrealwriters) {
+		t_rcu_perf_writer_started = t;
+		if (gp_exp) {
+			b_rcu_perf_writer_started =
+				cur_ops->exp_completed() / 2;
+		} else {
+			b_rcu_perf_writer_started =
+				cur_ops->completed();
+		}
+	}
+
+	do {
+		wdp = &wdpp[i];
+		*wdp = ktime_get_mono_fast_ns();
+		if (gp_exp) {
+			rcu_perf_writer_state = RTWS_EXP_SYNC;
+			cur_ops->exp_sync();
+		} else {
+			rcu_perf_writer_state = RTWS_SYNC;
+			cur_ops->sync();
+		}
+		rcu_perf_writer_state = RTWS_IDLE;
+		t = ktime_get_mono_fast_ns();
+		*wdp = t - *wdp;
+		i_max = i;
+		if (!started &&
+		    atomic_read(&n_rcu_perf_writer_started) >= nrealwriters)
+			started = true;
+		if (!done && i >= MIN_MEAS) {
+			done = true;
+			sp.sched_priority = 0;
+			sched_setscheduler_nocheck(current,
+						   SCHED_NORMAL, &sp);
+			pr_alert("%s" PERF_FLAG
+				 "rcu_perf_writer %ld has %d measurements\n",
+				 perf_type, me, MIN_MEAS);
+			if (atomic_inc_return(&n_rcu_perf_writer_finished) >=
+			    nrealwriters) {
+				schedule_timeout_interruptible(10);
+				rcu_ftrace_dump(DUMP_ALL);
+				PERFOUT_STRING("Test complete");
+				t_rcu_perf_writer_finished = t;
+				if (gp_exp) {
+					b_rcu_perf_writer_finished =
+						cur_ops->exp_completed() / 2;
+				} else {
+					b_rcu_perf_writer_finished =
+						cur_ops->completed();
+				}
+				if (shutdown) {
+					smp_mb(); /* Assign before wake. */
+					wake_up(&shutdown_wq);
+				}
+			}
+		}
+		if (done && !alldone &&
+		    atomic_read(&n_rcu_perf_writer_finished) >= nrealwriters)
+			alldone = true;
+		if (started && !alldone && i < MAX_MEAS - 1)
+			i++;
+		rcu_perf_wait_shutdown();
+	} while (!torture_must_stop());
+	rcu_perf_writer_state = RTWS_STOPPING;
+	writer_n_durations[me] = i_max;
+	torture_kthread_stopping("rcu_perf_writer");
+	return 0;
+}
+
+static inline void
+rcu_perf_print_module_parms(struct rcu_perf_ops *cur_ops, const char *tag)
+{
+	pr_alert("%s" PERF_FLAG
+		 "--- %s: nreaders=%d nwriters=%d verbose=%d shutdown=%d\n",
+		 perf_type, tag, nrealreaders, nrealwriters, verbose, shutdown);
+}
+
+static void
+rcu_perf_cleanup(void)
+{
+	int i;
+	int j;
+	int ngps = 0;
+	u64 *wdp;
+	u64 *wdpp;
+
+	if (torture_cleanup_begin())
+		return;
+
+	if (reader_tasks) {
+		for (i = 0; i < nrealreaders; i++)
+			torture_stop_kthread(rcu_perf_reader,
+					     reader_tasks[i]);
+		kfree(reader_tasks);
+	}
+
+	if (writer_tasks) {
+		for (i = 0; i < nrealwriters; i++) {
+			torture_stop_kthread(rcu_perf_writer,
+					     writer_tasks[i]);
+			if (!writer_n_durations)
+				continue;
+			j = writer_n_durations[i];
+			pr_alert("%s%s writer %d gps: %d\n",
+				 perf_type, PERF_FLAG, i, j);
+			ngps += j;
+		}
+		pr_alert("%s%s start: %llu end: %llu duration: %llu gps: %d batches: %ld\n",
+			 perf_type, PERF_FLAG,
+			 t_rcu_perf_writer_started, t_rcu_perf_writer_finished,
+			 t_rcu_perf_writer_finished -
+			 t_rcu_perf_writer_started,
+			 ngps,
+			 b_rcu_perf_writer_finished -
+			 b_rcu_perf_writer_started);
+		for (i = 0; i < nrealwriters; i++) {
+			if (!writer_durations)
+				break;
+			if (!writer_n_durations)
+				continue;
+			wdpp = writer_durations[i];
+			if (!wdpp)
+				continue;
+			for (j = 0; j <= writer_n_durations[i]; j++) {
+				wdp = &wdpp[j];
+				pr_alert("%s%s %4d writer-duration: %5d %llu\n",
+					perf_type, PERF_FLAG,
+					i, j, *wdp);
+				if (j % 100 == 0)
+					schedule_timeout_uninterruptible(1);
+			}
+			kfree(writer_durations[i]);
+		}
+		kfree(writer_tasks);
+		kfree(writer_durations);
+		kfree(writer_n_durations);
+	}
+
+	/* Do flavor-specific cleanup operations.  */
+	if (cur_ops->cleanup != NULL)
+		cur_ops->cleanup();
+
+	torture_cleanup_end();
+}
+
+/*
+ * Return the number if non-negative.  If -1, the number of CPUs.
+ * If less than -1, that much less than the number of CPUs, but
+ * at least one.
+ */
+static int compute_real(int n)
+{
+	int nr;
+
+	if (n >= 0) {
+		nr = n;
+	} else {
+		nr = num_online_cpus() + 1 + n;
+		if (nr <= 0)
+			nr = 1;
+	}
+	return nr;
+}
+
+/*
+ * RCU perf shutdown kthread.  Just waits to be awakened, then shuts
+ * down system.
+ */
+static int
+rcu_perf_shutdown(void *arg)
+{
+	do {
+		wait_event(shutdown_wq,
+			   atomic_read(&n_rcu_perf_writer_finished) >=
+			   nrealwriters);
+	} while (atomic_read(&n_rcu_perf_writer_finished) < nrealwriters);
+	smp_mb(); /* Wake before output. */
+	rcu_perf_cleanup();
+	kernel_power_off();
+	return -EINVAL;
+}
+
+static int __init
+rcu_perf_init(void)
+{
+	long i;
+	int firsterr = 0;
+	static struct rcu_perf_ops *perf_ops[] = {
+		&rcu_ops, &rcu_bh_ops, &srcu_ops, &sched_ops,
+		RCUPERF_TASKS_OPS
+	};
+
+	if (!torture_init_begin(perf_type, verbose, &perf_runnable))
+		return -EBUSY;
+
+	/* Process args and tell the world that the perf'er is on the job. */
+	for (i = 0; i < ARRAY_SIZE(perf_ops); i++) {
+		cur_ops = perf_ops[i];
+		if (strcmp(perf_type, cur_ops->name) == 0)
+			break;
+	}
+	if (i == ARRAY_SIZE(perf_ops)) {
+		pr_alert("rcu-perf: invalid perf type: \"%s\"\n",
+			 perf_type);
+		pr_alert("rcu-perf types:");
+		for (i = 0; i < ARRAY_SIZE(perf_ops); i++)
+			pr_alert(" %s", perf_ops[i]->name);
+		pr_alert("\n");
+		firsterr = -EINVAL;
+		goto unwind;
+	}
+	if (cur_ops->init)
+		cur_ops->init();
+
+	nrealwriters = compute_real(nwriters);
+	nrealreaders = compute_real(nreaders);
+	atomic_set(&n_rcu_perf_reader_started, 0);
+	atomic_set(&n_rcu_perf_writer_started, 0);
+	atomic_set(&n_rcu_perf_writer_finished, 0);
+	rcu_perf_print_module_parms(cur_ops, "Start of test");
+
+	/* Start up the kthreads. */
+
+	if (shutdown) {
+		init_waitqueue_head(&shutdown_wq);
+		firsterr = torture_create_kthread(rcu_perf_shutdown, NULL,
+						  shutdown_task);
+		if (firsterr)
+			goto unwind;
+		schedule_timeout_uninterruptible(1);
+	}
+	reader_tasks = kcalloc(nrealreaders, sizeof(reader_tasks[0]),
+			       GFP_KERNEL);
+	if (reader_tasks == NULL) {
+		VERBOSE_PERFOUT_ERRSTRING("out of memory");
+		firsterr = -ENOMEM;
+		goto unwind;
+	}
+	for (i = 0; i < nrealreaders; i++) {
+		firsterr = torture_create_kthread(rcu_perf_reader, (void *)i,
+						  reader_tasks[i]);
+		if (firsterr)
+			goto unwind;
+	}
+	while (atomic_read(&n_rcu_perf_reader_started) < nrealreaders)
+		schedule_timeout_uninterruptible(1);
+	writer_tasks = kcalloc(nrealwriters, sizeof(reader_tasks[0]),
+			       GFP_KERNEL);
+	writer_durations = kcalloc(nrealwriters, sizeof(*writer_durations),
+				   GFP_KERNEL);
+	writer_n_durations =
+		kcalloc(nrealwriters, sizeof(*writer_n_durations),
+			GFP_KERNEL);
+	if (!writer_tasks || !writer_durations || !writer_n_durations) {
+		VERBOSE_PERFOUT_ERRSTRING("out of memory");
+		firsterr = -ENOMEM;
+		goto unwind;
+	}
+	for (i = 0; i < nrealwriters; i++) {
+		writer_durations[i] =
+			kcalloc(MAX_MEAS, sizeof(*writer_durations[i]),
+				GFP_KERNEL);
+		if (!writer_durations[i])
+			goto unwind;
+		firsterr = torture_create_kthread(rcu_perf_writer, (void *)i,
+						  writer_tasks[i]);
+		if (firsterr)
+			goto unwind;
+	}
+	torture_init_end();
+	return 0;
+
+unwind:
+	torture_init_end();
+	rcu_perf_cleanup();
+	return firsterr;
+}
+
+module_init(rcu_perf_init);
+module_exit(rcu_perf_cleanup);
diff --git a/kernel/rcu/rcutorture.c b/kernel/rcu/rcutorture.c
index 250ea67c1615..084a28a732eb 100644
--- a/kernel/rcu/rcutorture.c
+++ b/kernel/rcu/rcutorture.c
@@ -130,8 +130,8 @@ static struct rcu_torture __rcu *rcu_torture_current;
 static unsigned long rcu_torture_current_version;
 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 DEFINE_PER_CPU(long [RCU_TORTURE_PIPE_LEN + 1], rcu_torture_count);
+static DEFINE_PER_CPU(long [RCU_TORTURE_PIPE_LEN + 1], rcu_torture_batch);
 static atomic_t rcu_torture_wcount[RCU_TORTURE_PIPE_LEN + 1];
 static atomic_t n_rcu_torture_alloc;
 static atomic_t n_rcu_torture_alloc_fail;
@@ -916,7 +916,7 @@ rcu_torture_fqs(void *arg)
 static int
 rcu_torture_writer(void *arg)
 {
-	bool can_expedite = !rcu_gp_is_expedited();
+	bool can_expedite = !rcu_gp_is_expedited() && !rcu_gp_is_normal();
 	int expediting = 0;
 	unsigned long gp_snap;
 	bool gp_cond1 = gp_cond, gp_exp1 = gp_exp, gp_normal1 = gp_normal;
@@ -932,7 +932,7 @@ rcu_torture_writer(void *arg)
 	VERBOSE_TOROUT_STRING("rcu_torture_writer task started");
 	if (!can_expedite) {
 		pr_alert("%s" TORTURE_FLAG
-			 " Grace periods expedited from boot/sysfs for %s,\n",
+			 " GP expediting controlled from boot/sysfs for %s,\n",
 			 torture_type, cur_ops->name);
 		pr_alert("%s" TORTURE_FLAG
 			 " Disabled dynamic grace-period expediting.\n",
@@ -1082,17 +1082,6 @@ rcu_torture_fakewriter(void *arg)
 	return 0;
 }
 
-static 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);
-}
-
 /*
  * RCU torture reader from timer handler.  Dereferences rcu_torture_current,
  * incrementing the corresponding element of the pipeline array.  The
@@ -1142,7 +1131,7 @@ static void rcu_torture_timer(unsigned long unused)
 	if (pipe_count > 1) {
 		do_trace_rcu_torture_read(cur_ops->name, &p->rtort_rcu, ts,
 					  started, completed);
-		rcutorture_trace_dump();
+		rcu_ftrace_dump(DUMP_ALL);
 	}
 	__this_cpu_inc(rcu_torture_count[pipe_count]);
 	completed = completed - started;
@@ -1215,7 +1204,7 @@ rcu_torture_reader(void *arg)
 		if (pipe_count > 1) {
 			do_trace_rcu_torture_read(cur_ops->name, &p->rtort_rcu,
 						  ts, started, completed);
-			rcutorture_trace_dump();
+			rcu_ftrace_dump(DUMP_ALL);
 		}
 		__this_cpu_inc(rcu_torture_count[pipe_count]);
 		completed = completed - started;
@@ -1333,7 +1322,7 @@ rcu_torture_stats_print(void)
 			 rcu_torture_writer_state,
 			 gpnum, completed, flags);
 		show_rcu_gp_kthreads();
-		rcutorture_trace_dump();
+		rcu_ftrace_dump(DUMP_ALL);
 	}
 	rtcv_snap = rcu_torture_current_version;
 }
@@ -1489,7 +1478,9 @@ static int rcu_torture_barrier_cbs(void *arg)
 		 * The above smp_load_acquire() ensures barrier_phase load
 		 * is ordered before the folloiwng ->call().
 		 */
+		local_irq_disable(); /* Just to test no-irq call_rcu(). */
 		cur_ops->call(&rcu, rcu_torture_barrier_cbf);
+		local_irq_enable();
 		if (atomic_dec_and_test(&barrier_cbs_count))
 			wake_up(&barrier_wq);
 	} while (!torture_must_stop());
@@ -1596,7 +1587,7 @@ static int rcutorture_cpu_notify(struct notifier_block *self,
 {
 	long cpu = (long)hcpu;
 
-	switch (action) {
+	switch (action & ~CPU_TASKS_FROZEN) {
 	case CPU_ONLINE:
 	case CPU_DOWN_FAILED:
 		(void)rcutorture_booster_init(cpu);
diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 9a535a86e732..c7f1bc4f817c 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -102,6 +102,8 @@ struct rcu_state sname##_state = { \
 	.barrier_mutex = __MUTEX_INITIALIZER(sname##_state.barrier_mutex), \
 	.name = RCU_STATE_NAME(sname), \
 	.abbr = sabbr, \
+	.exp_mutex = __MUTEX_INITIALIZER(sname##_state.exp_mutex), \
+	.exp_wake_mutex = __MUTEX_INITIALIZER(sname##_state.exp_wake_mutex), \
 }
 
 RCU_STATE_INITIALIZER(rcu_sched, 's', call_rcu_sched);
@@ -370,6 +372,21 @@ void rcu_all_qs(void)
 		rcu_momentary_dyntick_idle();
 		local_irq_restore(flags);
 	}
+	if (unlikely(raw_cpu_read(rcu_sched_data.cpu_no_qs.b.exp))) {
+		/*
+		 * Yes, we just checked a per-CPU variable with preemption
+		 * enabled, so we might be migrated to some other CPU at
+		 * this point.  That is OK because in that case, the
+		 * migration will supply the needed quiescent state.
+		 * We might end up needlessly disabling preemption and
+		 * invoking rcu_sched_qs() on the destination CPU, but
+		 * the probability and cost are both quite low, so this
+		 * should not be a problem in practice.
+		 */
+		preempt_disable();
+		rcu_sched_qs();
+		preempt_enable();
+	}
 	this_cpu_inc(rcu_qs_ctr);
 	barrier(); /* Avoid RCU read-side critical sections leaking up. */
 }
@@ -385,9 +402,11 @@ module_param(qlowmark, long, 0444);
 
 static ulong jiffies_till_first_fqs = ULONG_MAX;
 static ulong jiffies_till_next_fqs = ULONG_MAX;
+static bool rcu_kick_kthreads;
 
 module_param(jiffies_till_first_fqs, ulong, 0644);
 module_param(jiffies_till_next_fqs, ulong, 0644);
+module_param(rcu_kick_kthreads, bool, 0644);
 
 /*
  * How long the grace period must be before we start recruiting
@@ -460,6 +479,28 @@ unsigned long rcu_batches_completed_bh(void)
 EXPORT_SYMBOL_GPL(rcu_batches_completed_bh);
 
 /*
+ * Return the number of RCU expedited batches completed thus far for
+ * debug & stats.  Odd numbers mean that a batch is in progress, even
+ * numbers mean idle.  The value returned will thus be roughly double
+ * the cumulative batches since boot.
+ */
+unsigned long rcu_exp_batches_completed(void)
+{
+	return rcu_state_p->expedited_sequence;
+}
+EXPORT_SYMBOL_GPL(rcu_exp_batches_completed);
+
+/*
+ * Return the number of RCU-sched expedited batches completed thus far
+ * for debug & stats.  Similar to rcu_exp_batches_completed().
+ */
+unsigned long rcu_exp_batches_completed_sched(void)
+{
+	return rcu_sched_state.expedited_sequence;
+}
+EXPORT_SYMBOL_GPL(rcu_exp_batches_completed_sched);
+
+/*
  * Force a quiescent state.
  */
 void rcu_force_quiescent_state(void)
@@ -637,7 +678,7 @@ static void rcu_eqs_enter_common(long long oldval, bool user)
 			idle_task(smp_processor_id());
 
 		trace_rcu_dyntick(TPS("Error on entry: not idle task"), oldval, 0);
-		ftrace_dump(DUMP_ORIG);
+		rcu_ftrace_dump(DUMP_ORIG);
 		WARN_ONCE(1, "Current pid: %d comm: %s / Idle pid: %d comm: %s",
 			  current->pid, current->comm,
 			  idle->pid, idle->comm); /* must be idle task! */
@@ -799,7 +840,7 @@ static void rcu_eqs_exit_common(long long oldval, int user)
 
 		trace_rcu_dyntick(TPS("Error on exit: not idle task"),
 				  oldval, rdtp->dynticks_nesting);
-		ftrace_dump(DUMP_ORIG);
+		rcu_ftrace_dump(DUMP_ORIG);
 		WARN_ONCE(1, "Current pid: %d comm: %s / Idle pid: %d comm: %s",
 			  current->pid, current->comm,
 			  idle->pid, idle->comm); /* must be idle task! */
@@ -1224,8 +1265,10 @@ static void rcu_check_gp_kthread_starvation(struct rcu_state *rsp)
 		       rsp->gp_flags,
 		       gp_state_getname(rsp->gp_state), rsp->gp_state,
 		       rsp->gp_kthread ? rsp->gp_kthread->state : ~0);
-		if (rsp->gp_kthread)
+		if (rsp->gp_kthread) {
 			sched_show_task(rsp->gp_kthread);
+			wake_up_process(rsp->gp_kthread);
+		}
 	}
 }
 
@@ -1249,6 +1292,25 @@ static void rcu_dump_cpu_stacks(struct rcu_state *rsp)
 	}
 }
 
+/*
+ * If too much time has passed in the current grace period, and if
+ * so configured, go kick the relevant kthreads.
+ */
+static void rcu_stall_kick_kthreads(struct rcu_state *rsp)
+{
+	unsigned long j;
+
+	if (!rcu_kick_kthreads)
+		return;
+	j = READ_ONCE(rsp->jiffies_kick_kthreads);
+	if (time_after(jiffies, j) && rsp->gp_kthread) {
+		WARN_ONCE(1, "Kicking %s grace-period kthread\n", rsp->name);
+		rcu_ftrace_dump(DUMP_ALL);
+		wake_up_process(rsp->gp_kthread);
+		WRITE_ONCE(rsp->jiffies_kick_kthreads, j + HZ);
+	}
+}
+
 static void print_other_cpu_stall(struct rcu_state *rsp, unsigned long gpnum)
 {
 	int cpu;
@@ -1260,6 +1322,11 @@ static void print_other_cpu_stall(struct rcu_state *rsp, unsigned long gpnum)
 	struct rcu_node *rnp = rcu_get_root(rsp);
 	long totqlen = 0;
 
+	/* Kick and suppress, if so configured. */
+	rcu_stall_kick_kthreads(rsp);
+	if (rcu_cpu_stall_suppress)
+		return;
+
 	/* Only let one CPU complain about others per time interval. */
 
 	raw_spin_lock_irqsave_rcu_node(rnp, flags);
@@ -1333,6 +1400,11 @@ static void print_cpu_stall(struct rcu_state *rsp)
 	struct rcu_node *rnp = rcu_get_root(rsp);
 	long totqlen = 0;
 
+	/* Kick and suppress, if so configured. */
+	rcu_stall_kick_kthreads(rsp);
+	if (rcu_cpu_stall_suppress)
+		return;
+
 	/*
 	 * OK, time to rat on ourselves...
 	 * See Documentation/RCU/stallwarn.txt for info on how to debug
@@ -1377,8 +1449,10 @@ static void check_cpu_stall(struct rcu_state *rsp, struct rcu_data *rdp)
 	unsigned long js;
 	struct rcu_node *rnp;
 
-	if (rcu_cpu_stall_suppress || !rcu_gp_in_progress(rsp))
+	if ((rcu_cpu_stall_suppress && !rcu_kick_kthreads) ||
+	    !rcu_gp_in_progress(rsp))
 		return;
+	rcu_stall_kick_kthreads(rsp);
 	j = jiffies;
 
 	/*
@@ -2117,8 +2191,11 @@ static int __noreturn rcu_gp_kthread(void *arg)
 		}
 		ret = 0;
 		for (;;) {
-			if (!ret)
+			if (!ret) {
 				rsp->jiffies_force_qs = jiffies + j;
+				WRITE_ONCE(rsp->jiffies_kick_kthreads,
+					   jiffies + 3 * j);
+			}
 			trace_rcu_grace_period(rsp->name,
 					       READ_ONCE(rsp->gpnum),
 					       TPS("fqswait"));
@@ -2144,6 +2221,15 @@ static int __noreturn rcu_gp_kthread(void *arg)
 						       TPS("fqsend"));
 				cond_resched_rcu_qs();
 				WRITE_ONCE(rsp->gp_activity, jiffies);
+				ret = 0; /* Force full wait till next FQS. */
+				j = jiffies_till_next_fqs;
+				if (j > HZ) {
+					j = HZ;
+					jiffies_till_next_fqs = HZ;
+				} else if (j < 1) {
+					j = 1;
+					jiffies_till_next_fqs = 1;
+				}
 			} else {
 				/* Deal with stray signal. */
 				cond_resched_rcu_qs();
@@ -2152,14 +2238,12 @@ static int __noreturn rcu_gp_kthread(void *arg)
 				trace_rcu_grace_period(rsp->name,
 						       READ_ONCE(rsp->gpnum),
 						       TPS("fqswaitsig"));
-			}
-			j = jiffies_till_next_fqs;
-			if (j > HZ) {
-				j = HZ;
-				jiffies_till_next_fqs = HZ;
-			} else if (j < 1) {
-				j = 1;
-				jiffies_till_next_fqs = 1;
+				ret = 1; /* Keep old FQS timing. */
+				j = jiffies;
+				if (time_after(jiffies, rsp->jiffies_force_qs))
+					j = 1;
+				else
+					j = rsp->jiffies_force_qs - j;
 			}
 		}
 
@@ -3376,8 +3460,12 @@ static void rcu_exp_gp_seq_end(struct rcu_state *rsp)
 }
 static unsigned long rcu_exp_gp_seq_snap(struct rcu_state *rsp)
 {
+	unsigned long s;
+
 	smp_mb(); /* Caller's modifications seen first by other CPUs. */
-	return rcu_seq_snap(&rsp->expedited_sequence);
+	s = rcu_seq_snap(&rsp->expedited_sequence);
+	trace_rcu_exp_grace_period(rsp->name, s, TPS("snap"));
+	return s;
 }
 static bool rcu_exp_gp_seq_done(struct rcu_state *rsp, unsigned long s)
 {
@@ -3469,7 +3557,7 @@ static void __maybe_unused sync_exp_reset_tree(struct rcu_state *rsp)
  * for the current expedited grace period.  Works only for preemptible
  * RCU -- other RCU implementation use other means.
  *
- * Caller must hold the root rcu_node's exp_funnel_mutex.
+ * Caller must hold the rcu_state's exp_mutex.
  */
 static int sync_rcu_preempt_exp_done(struct rcu_node *rnp)
 {
@@ -3485,8 +3573,8 @@ static int sync_rcu_preempt_exp_done(struct rcu_node *rnp)
  * recursively up the tree.  (Calm down, calm down, we do the recursion
  * iteratively!)
  *
- * Caller must hold the root rcu_node's exp_funnel_mutex and the
- * specified rcu_node structure's ->lock.
+ * Caller must hold the rcu_state's exp_mutex and the specified rcu_node
+ * structure's ->lock.
  */
 static void __rcu_report_exp_rnp(struct rcu_state *rsp, struct rcu_node *rnp,
 				 bool wake, unsigned long flags)
@@ -3523,7 +3611,7 @@ static void __rcu_report_exp_rnp(struct rcu_state *rsp, struct rcu_node *rnp,
  * Report expedited quiescent state for specified node.  This is a
  * lock-acquisition wrapper function for __rcu_report_exp_rnp().
  *
- * Caller must hold the root rcu_node's exp_funnel_mutex.
+ * Caller must hold the rcu_state's exp_mutex.
  */
 static void __maybe_unused rcu_report_exp_rnp(struct rcu_state *rsp,
 					      struct rcu_node *rnp, bool wake)
@@ -3536,8 +3624,8 @@ static void __maybe_unused rcu_report_exp_rnp(struct rcu_state *rsp,
 
 /*
  * Report expedited quiescent state for multiple CPUs, all covered by the
- * specified leaf rcu_node structure.  Caller must hold the root
- * rcu_node's exp_funnel_mutex.
+ * specified leaf rcu_node structure.  Caller must hold the rcu_state's
+ * exp_mutex.
  */
 static void rcu_report_exp_cpu_mult(struct rcu_state *rsp, struct rcu_node *rnp,
 				    unsigned long mask, bool wake)
@@ -3555,7 +3643,6 @@ static void rcu_report_exp_cpu_mult(struct rcu_state *rsp, struct rcu_node *rnp,
 
 /*
  * Report expedited quiescent state for specified rcu_data (CPU).
- * Caller must hold the root rcu_node's exp_funnel_mutex.
  */
 static void rcu_report_exp_rdp(struct rcu_state *rsp, struct rcu_data *rdp,
 			       bool wake)
@@ -3564,15 +3651,11 @@ static void rcu_report_exp_rdp(struct rcu_state *rsp, struct rcu_data *rdp,
 }
 
 /* Common code for synchronize_{rcu,sched}_expedited() work-done checking. */
-static bool sync_exp_work_done(struct rcu_state *rsp, struct rcu_node *rnp,
-			       struct rcu_data *rdp,
-			       atomic_long_t *stat, unsigned long s)
+static bool sync_exp_work_done(struct rcu_state *rsp, atomic_long_t *stat,
+			       unsigned long s)
 {
 	if (rcu_exp_gp_seq_done(rsp, s)) {
-		if (rnp)
-			mutex_unlock(&rnp->exp_funnel_mutex);
-		else if (rdp)
-			mutex_unlock(&rdp->exp_funnel_mutex);
+		trace_rcu_exp_grace_period(rsp->name, s, TPS("done"));
 		/* Ensure test happens before caller kfree(). */
 		smp_mb__before_atomic(); /* ^^^ */
 		atomic_long_inc(stat);
@@ -3582,59 +3665,65 @@ static bool sync_exp_work_done(struct rcu_state *rsp, struct rcu_node *rnp,
 }
 
 /*
- * Funnel-lock acquisition for expedited grace periods.  Returns a
- * pointer to the root rcu_node structure, or NULL if some other
- * task did the expedited grace period for us.
+ * Funnel-lock acquisition for expedited grace periods.  Returns true
+ * if some other task completed an expedited grace period that this task
+ * can piggy-back on, and with no mutex held.  Otherwise, returns false
+ * with the mutex held, indicating that the caller must actually do the
+ * expedited grace period.
  */
-static struct rcu_node *exp_funnel_lock(struct rcu_state *rsp, unsigned long s)
+static bool exp_funnel_lock(struct rcu_state *rsp, unsigned long s)
 {
 	struct rcu_data *rdp = per_cpu_ptr(rsp->rda, raw_smp_processor_id());
-	struct rcu_node *rnp0;
-	struct rcu_node *rnp1 = NULL;
+	struct rcu_node *rnp = rdp->mynode;
+	struct rcu_node *rnp_root = rcu_get_root(rsp);
+
+	/* Low-contention fastpath. */
+	if (ULONG_CMP_LT(READ_ONCE(rnp->exp_seq_rq), s) &&
+	    (rnp == rnp_root ||
+	     ULONG_CMP_LT(READ_ONCE(rnp_root->exp_seq_rq), s)) &&
+	    !mutex_is_locked(&rsp->exp_mutex) &&
+	    mutex_trylock(&rsp->exp_mutex))
+		goto fastpath;
 
 	/*
-	 * First try directly acquiring the root lock in order to reduce
-	 * latency in the common case where expedited grace periods are
-	 * rare.  We check mutex_is_locked() to avoid pathological levels of
-	 * memory contention on ->exp_funnel_mutex in the heavy-load case.
+	 * Each pass through the following loop works its way up
+	 * the rcu_node tree, returning if others have done the work or
+	 * otherwise falls through to acquire rsp->exp_mutex.  The mapping
+	 * from CPU to rcu_node structure can be inexact, as it is just
+	 * promoting locality and is not strictly needed for correctness.
 	 */
-	rnp0 = rcu_get_root(rsp);
-	if (!mutex_is_locked(&rnp0->exp_funnel_mutex)) {
-		if (mutex_trylock(&rnp0->exp_funnel_mutex)) {
-			if (sync_exp_work_done(rsp, rnp0, NULL,
-					       &rdp->expedited_workdone0, s))
-				return NULL;
-			return rnp0;
+	for (; rnp != NULL; rnp = rnp->parent) {
+		if (sync_exp_work_done(rsp, &rdp->exp_workdone1, s))
+			return true;
+
+		/* Work not done, either wait here or go up. */
+		spin_lock(&rnp->exp_lock);
+		if (ULONG_CMP_GE(rnp->exp_seq_rq, s)) {
+
+			/* Someone else doing GP, so wait for them. */
+			spin_unlock(&rnp->exp_lock);
+			trace_rcu_exp_funnel_lock(rsp->name, rnp->level,
+						  rnp->grplo, rnp->grphi,
+						  TPS("wait"));
+			wait_event(rnp->exp_wq[(s >> 1) & 0x3],
+				   sync_exp_work_done(rsp,
+						      &rdp->exp_workdone2, s));
+			return true;
 		}
+		rnp->exp_seq_rq = s; /* Followers can wait on us. */
+		spin_unlock(&rnp->exp_lock);
+		trace_rcu_exp_funnel_lock(rsp->name, rnp->level, rnp->grplo,
+					  rnp->grphi, TPS("nxtlvl"));
 	}
-
-	/*
-	 * Each pass through the following loop works its way
-	 * up the rcu_node tree, returning if others have done the
-	 * work or otherwise falls through holding the root rnp's
-	 * ->exp_funnel_mutex.  The mapping from CPU to rcu_node structure
-	 * can be inexact, as it is just promoting locality and is not
-	 * strictly needed for correctness.
-	 */
-	if (sync_exp_work_done(rsp, NULL, NULL, &rdp->expedited_workdone1, s))
-		return NULL;
-	mutex_lock(&rdp->exp_funnel_mutex);
-	rnp0 = rdp->mynode;
-	for (; rnp0 != NULL; rnp0 = rnp0->parent) {
-		if (sync_exp_work_done(rsp, rnp1, rdp,
-				       &rdp->expedited_workdone2, s))
-			return NULL;
-		mutex_lock(&rnp0->exp_funnel_mutex);
-		if (rnp1)
-			mutex_unlock(&rnp1->exp_funnel_mutex);
-		else
-			mutex_unlock(&rdp->exp_funnel_mutex);
-		rnp1 = rnp0;
+	mutex_lock(&rsp->exp_mutex);
+fastpath:
+	if (sync_exp_work_done(rsp, &rdp->exp_workdone3, s)) {
+		mutex_unlock(&rsp->exp_mutex);
+		return true;
 	}
-	if (sync_exp_work_done(rsp, rnp1, rdp,
-			       &rdp->expedited_workdone3, s))
-		return NULL;
-	return rnp1;
+	rcu_exp_gp_seq_start(rsp);
+	trace_rcu_exp_grace_period(rsp->name, s, TPS("start"));
+	return false;
 }
 
 /* Invoked on each online non-idle CPU for expedited quiescent state. */
@@ -3649,6 +3738,11 @@ static void sync_sched_exp_handler(void *data)
 	if (!(READ_ONCE(rnp->expmask) & rdp->grpmask) ||
 	    __this_cpu_read(rcu_sched_data.cpu_no_qs.b.exp))
 		return;
+	if (rcu_is_cpu_rrupt_from_idle()) {
+		rcu_report_exp_rdp(&rcu_sched_state,
+				   this_cpu_ptr(&rcu_sched_data), true);
+		return;
+	}
 	__this_cpu_write(rcu_sched_data.cpu_no_qs.b.exp, true);
 	resched_cpu(smp_processor_id());
 }
@@ -3773,7 +3867,7 @@ static void synchronize_sched_expedited_wait(struct rcu_state *rsp)
 		       rsp->name);
 		ndetected = 0;
 		rcu_for_each_leaf_node(rsp, rnp) {
-			ndetected = rcu_print_task_exp_stall(rnp);
+			ndetected += rcu_print_task_exp_stall(rnp);
 			mask = 1;
 			for (cpu = rnp->grplo; cpu <= rnp->grphi; cpu++, mask <<= 1) {
 				struct rcu_data *rdp;
@@ -3783,7 +3877,7 @@ static void synchronize_sched_expedited_wait(struct rcu_state *rsp)
 				ndetected++;
 				rdp = per_cpu_ptr(rsp->rda, cpu);
 				pr_cont(" %d-%c%c%c", cpu,
-					"O."[cpu_online(cpu)],
+					"O."[!!cpu_online(cpu)],
 					"o."[!!(rdp->grpmask & rnp->expmaskinit)],
 					"N."[!!(rdp->grpmask & rnp->expmaskinitnext)]);
 			}
@@ -3792,7 +3886,7 @@ static void synchronize_sched_expedited_wait(struct rcu_state *rsp)
 		pr_cont(" } %lu jiffies s: %lu root: %#lx/%c\n",
 			jiffies - jiffies_start, rsp->expedited_sequence,
 			rnp_root->expmask, ".T"[!!rnp_root->exp_tasks]);
-		if (!ndetected) {
+		if (ndetected) {
 			pr_err("blocking rcu_node structures:");
 			rcu_for_each_node_breadth_first(rsp, rnp) {
 				if (rnp == rnp_root)
@@ -3818,6 +3912,41 @@ static void synchronize_sched_expedited_wait(struct rcu_state *rsp)
 	}
 }
 
+/*
+ * Wait for the current expedited grace period to complete, and then
+ * wake up everyone who piggybacked on the just-completed expedited
+ * grace period.  Also update all the ->exp_seq_rq counters as needed
+ * in order to avoid counter-wrap problems.
+ */
+static void rcu_exp_wait_wake(struct rcu_state *rsp, unsigned long s)
+{
+	struct rcu_node *rnp;
+
+	synchronize_sched_expedited_wait(rsp);
+	rcu_exp_gp_seq_end(rsp);
+	trace_rcu_exp_grace_period(rsp->name, s, TPS("end"));
+
+	/*
+	 * Switch over to wakeup mode, allowing the next GP, but -only- the
+	 * next GP, to proceed.
+	 */
+	mutex_lock(&rsp->exp_wake_mutex);
+	mutex_unlock(&rsp->exp_mutex);
+
+	rcu_for_each_node_breadth_first(rsp, rnp) {
+		if (ULONG_CMP_LT(READ_ONCE(rnp->exp_seq_rq), s)) {
+			spin_lock(&rnp->exp_lock);
+			/* Recheck, avoid hang in case someone just arrived. */
+			if (ULONG_CMP_LT(rnp->exp_seq_rq, s))
+				rnp->exp_seq_rq = s;
+			spin_unlock(&rnp->exp_lock);
+		}
+		wake_up_all(&rnp->exp_wq[(rsp->expedited_sequence >> 1) & 0x3]);
+	}
+	trace_rcu_exp_grace_period(rsp->name, s, TPS("endwake"));
+	mutex_unlock(&rsp->exp_wake_mutex);
+}
+
 /**
  * synchronize_sched_expedited - Brute-force RCU-sched grace period
  *
@@ -3837,7 +3966,6 @@ static void synchronize_sched_expedited_wait(struct rcu_state *rsp)
 void synchronize_sched_expedited(void)
 {
 	unsigned long s;
-	struct rcu_node *rnp;
 	struct rcu_state *rsp = &rcu_sched_state;
 
 	/* If only one CPU, this is automatically a grace period. */
@@ -3852,17 +3980,14 @@ void synchronize_sched_expedited(void)
 
 	/* Take a snapshot of the sequence number.  */
 	s = rcu_exp_gp_seq_snap(rsp);
-
-	rnp = exp_funnel_lock(rsp, s);
-	if (rnp == NULL)
+	if (exp_funnel_lock(rsp, s))
 		return;  /* Someone else did our work for us. */
 
-	rcu_exp_gp_seq_start(rsp);
+	/* Initialize the rcu_node tree in preparation for the wait. */
 	sync_rcu_exp_select_cpus(rsp, sync_sched_exp_handler);
-	synchronize_sched_expedited_wait(rsp);
 
-	rcu_exp_gp_seq_end(rsp);
-	mutex_unlock(&rnp->exp_funnel_mutex);
+	/* Wait and clean up, including waking everyone. */
+	rcu_exp_wait_wake(rsp, s);
 }
 EXPORT_SYMBOL_GPL(synchronize_sched_expedited);
 
@@ -4162,7 +4287,6 @@ rcu_boot_init_percpu_data(int cpu, struct rcu_state *rsp)
 	WARN_ON_ONCE(atomic_read(&rdp->dynticks->dynticks) != 1);
 	rdp->cpu = cpu;
 	rdp->rsp = rsp;
-	mutex_init(&rdp->exp_funnel_mutex);
 	rcu_boot_init_nocb_percpu_data(rdp);
 	raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 }
@@ -4420,10 +4544,8 @@ static void __init rcu_init_one(struct rcu_state *rsp)
 {
 	static const char * const buf[] = RCU_NODE_NAME_INIT;
 	static const char * const fqs[] = RCU_FQS_NAME_INIT;
-	static const char * const exp[] = RCU_EXP_NAME_INIT;
 	static struct lock_class_key rcu_node_class[RCU_NUM_LVLS];
 	static struct lock_class_key rcu_fqs_class[RCU_NUM_LVLS];
-	static struct lock_class_key rcu_exp_class[RCU_NUM_LVLS];
 	static u8 fl_mask = 0x1;
 
 	int levelcnt[RCU_NUM_LVLS];		/* # nodes in each level. */
@@ -4482,9 +4604,11 @@ static void __init rcu_init_one(struct rcu_state *rsp)
 			rnp->level = i;
 			INIT_LIST_HEAD(&rnp->blkd_tasks);
 			rcu_init_one_nocb(rnp);
-			mutex_init(&rnp->exp_funnel_mutex);
-			lockdep_set_class_and_name(&rnp->exp_funnel_mutex,
-						   &rcu_exp_class[i], exp[i]);
+			init_waitqueue_head(&rnp->exp_wq[0]);
+			init_waitqueue_head(&rnp->exp_wq[1]);
+			init_waitqueue_head(&rnp->exp_wq[2]);
+			init_waitqueue_head(&rnp->exp_wq[3]);
+			spin_lock_init(&rnp->exp_lock);
 		}
 	}
 
diff --git a/kernel/rcu/tree.h b/kernel/rcu/tree.h
index df668c0f9e64..e3959f5e6ddf 100644
--- a/kernel/rcu/tree.h
+++ b/kernel/rcu/tree.h
@@ -70,7 +70,6 @@
 #  define NUM_RCU_LVL_INIT    { NUM_RCU_LVL_0 }
 #  define RCU_NODE_NAME_INIT  { "rcu_node_0" }
 #  define RCU_FQS_NAME_INIT   { "rcu_node_fqs_0" }
-#  define RCU_EXP_NAME_INIT   { "rcu_node_exp_0" }
 #elif NR_CPUS <= RCU_FANOUT_2
 #  define RCU_NUM_LVLS	      2
 #  define NUM_RCU_LVL_0	      1
@@ -79,7 +78,6 @@
 #  define NUM_RCU_LVL_INIT    { NUM_RCU_LVL_0, NUM_RCU_LVL_1 }
 #  define RCU_NODE_NAME_INIT  { "rcu_node_0", "rcu_node_1" }
 #  define RCU_FQS_NAME_INIT   { "rcu_node_fqs_0", "rcu_node_fqs_1" }
-#  define RCU_EXP_NAME_INIT   { "rcu_node_exp_0", "rcu_node_exp_1" }
 #elif NR_CPUS <= RCU_FANOUT_3
 #  define RCU_NUM_LVLS	      3
 #  define NUM_RCU_LVL_0	      1
@@ -89,7 +87,6 @@
 #  define NUM_RCU_LVL_INIT    { NUM_RCU_LVL_0, NUM_RCU_LVL_1, NUM_RCU_LVL_2 }
 #  define RCU_NODE_NAME_INIT  { "rcu_node_0", "rcu_node_1", "rcu_node_2" }
 #  define RCU_FQS_NAME_INIT   { "rcu_node_fqs_0", "rcu_node_fqs_1", "rcu_node_fqs_2" }
-#  define RCU_EXP_NAME_INIT   { "rcu_node_exp_0", "rcu_node_exp_1", "rcu_node_exp_2" }
 #elif NR_CPUS <= RCU_FANOUT_4
 #  define RCU_NUM_LVLS	      4
 #  define NUM_RCU_LVL_0	      1
@@ -100,7 +97,6 @@
 #  define NUM_RCU_LVL_INIT    { NUM_RCU_LVL_0, NUM_RCU_LVL_1, NUM_RCU_LVL_2, NUM_RCU_LVL_3 }
 #  define RCU_NODE_NAME_INIT  { "rcu_node_0", "rcu_node_1", "rcu_node_2", "rcu_node_3" }
 #  define RCU_FQS_NAME_INIT   { "rcu_node_fqs_0", "rcu_node_fqs_1", "rcu_node_fqs_2", "rcu_node_fqs_3" }
-#  define RCU_EXP_NAME_INIT   { "rcu_node_exp_0", "rcu_node_exp_1", "rcu_node_exp_2", "rcu_node_exp_3" }
 #else
 # error "CONFIG_RCU_FANOUT insufficient for NR_CPUS"
 #endif /* #if (NR_CPUS) <= RCU_FANOUT_1 */
@@ -252,7 +248,9 @@ struct rcu_node {
 				/* Counts of upcoming no-CB GP requests. */
 	raw_spinlock_t fqslock ____cacheline_internodealigned_in_smp;
 
-	struct mutex exp_funnel_mutex ____cacheline_internodealigned_in_smp;
+	spinlock_t exp_lock ____cacheline_internodealigned_in_smp;
+	unsigned long exp_seq_rq;
+	wait_queue_head_t exp_wq[4];
 } ____cacheline_internodealigned_in_smp;
 
 /*
@@ -387,11 +385,9 @@ struct rcu_data {
 #ifdef CONFIG_RCU_FAST_NO_HZ
 	struct rcu_head oom_head;
 #endif /* #ifdef CONFIG_RCU_FAST_NO_HZ */
-	struct mutex exp_funnel_mutex;
-	atomic_long_t expedited_workdone0;	/* # done by others #0. */
-	atomic_long_t expedited_workdone1;	/* # done by others #1. */
-	atomic_long_t expedited_workdone2;	/* # done by others #2. */
-	atomic_long_t expedited_workdone3;	/* # done by others #3. */
+	atomic_long_t exp_workdone1;	/* # done by others #1. */
+	atomic_long_t exp_workdone2;	/* # done by others #2. */
+	atomic_long_t exp_workdone3;	/* # done by others #3. */
 
 	/* 7) Callback offloading. */
 #ifdef CONFIG_RCU_NOCB_CPU
@@ -505,6 +501,8 @@ struct rcu_state {
 						/*  _rcu_barrier(). */
 	/* End of fields guarded by barrier_mutex. */
 
+	struct mutex exp_mutex;			/* Serialize expedited GP. */
+	struct mutex exp_wake_mutex;		/* Serialize wakeup. */
 	unsigned long expedited_sequence;	/* Take a ticket. */
 	atomic_long_t expedited_normal;		/* # fallbacks to normal. */
 	atomic_t expedited_need_qs;		/* # CPUs left to check in. */
@@ -513,6 +511,8 @@ struct rcu_state {
 
 	unsigned long jiffies_force_qs;		/* Time at which to invoke */
 						/*  force_quiescent_state(). */
+	unsigned long jiffies_kick_kthreads;	/* Time at which to kick */
+						/*  kthreads, if configured. */
 	unsigned long n_force_qs;		/* Number of calls to */
 						/*  force_quiescent_state(). */
 	unsigned long n_force_qs_lh;		/* ~Number of calls leaving */
diff --git a/kernel/rcu/tree_plugin.h b/kernel/rcu/tree_plugin.h
index efdf7b61ce12..ff1cd4e1188d 100644
--- a/kernel/rcu/tree_plugin.h
+++ b/kernel/rcu/tree_plugin.h
@@ -722,18 +722,22 @@ static void sync_rcu_exp_handler(void *info)
  * synchronize_rcu_expedited - Brute-force RCU grace period
  *
  * Wait for an RCU-preempt grace period, but expedite it.  The basic
- * idea is to invoke synchronize_sched_expedited() to push all the tasks to
- * the ->blkd_tasks lists and wait for this list to drain.  This consumes
- * significant time on all CPUs and is unfriendly to real-time workloads,
- * so is thus not recommended for any sort of common-case code.
- * In fact, if you are using synchronize_rcu_expedited() in a loop,
- * please restructure your code to batch your updates, and then Use a
- * single synchronize_rcu() instead.
+ * idea is to IPI all non-idle non-nohz online CPUs.  The IPI handler
+ * checks whether the CPU is in an RCU-preempt critical section, and
+ * if so, it sets a flag that causes the outermost rcu_read_unlock()
+ * to report the quiescent state.  On the other hand, if the CPU is
+ * not in an RCU read-side critical section, the IPI handler reports
+ * the quiescent state immediately.
+ *
+ * Although this is a greate improvement over previous expedited
+ * implementations, it is still unfriendly to real-time workloads, so is
+ * thus not recommended for any sort of common-case code.  In fact, if
+ * you are using synchronize_rcu_expedited() in a loop, please restructure
+ * your code to batch your updates, and then Use a single synchronize_rcu()
+ * instead.
  */
 void synchronize_rcu_expedited(void)
 {
-	struct rcu_node *rnp;
-	struct rcu_node *rnp_unlock;
 	struct rcu_state *rsp = rcu_state_p;
 	unsigned long s;
 
@@ -744,23 +748,14 @@ void synchronize_rcu_expedited(void)
 	}
 
 	s = rcu_exp_gp_seq_snap(rsp);
-
-	rnp_unlock = exp_funnel_lock(rsp, s);
-	if (rnp_unlock == NULL)
+	if (exp_funnel_lock(rsp, s))
 		return;  /* Someone else did our work for us. */
 
-	rcu_exp_gp_seq_start(rsp);
-
 	/* Initialize the rcu_node tree in preparation for the wait. */
 	sync_rcu_exp_select_cpus(rsp, sync_rcu_exp_handler);
 
-	/* Wait for snapshotted ->blkd_tasks lists to drain. */
-	rnp = rcu_get_root(rsp);
-	synchronize_sched_expedited_wait(rsp);
-
-	/* Clean up and exit. */
-	rcu_exp_gp_seq_end(rsp);
-	mutex_unlock(&rnp_unlock->exp_funnel_mutex);
+	/* Wait for ->blkd_tasks lists to drain, then wake everyone up. */
+	rcu_exp_wait_wake(rsp, s);
 }
 EXPORT_SYMBOL_GPL(synchronize_rcu_expedited);
 
diff --git a/kernel/rcu/tree_trace.c b/kernel/rcu/tree_trace.c
index 1088e64f01ad..86782f9a4604 100644
--- a/kernel/rcu/tree_trace.c
+++ b/kernel/rcu/tree_trace.c
@@ -185,17 +185,16 @@ static int show_rcuexp(struct seq_file *m, void *v)
 	int cpu;
 	struct rcu_state *rsp = (struct rcu_state *)m->private;
 	struct rcu_data *rdp;
-	unsigned long s0 = 0, s1 = 0, s2 = 0, s3 = 0;
+	unsigned long s1 = 0, s2 = 0, s3 = 0;
 
 	for_each_possible_cpu(cpu) {
 		rdp = per_cpu_ptr(rsp->rda, cpu);
-		s0 += atomic_long_read(&rdp->expedited_workdone0);
-		s1 += atomic_long_read(&rdp->expedited_workdone1);
-		s2 += atomic_long_read(&rdp->expedited_workdone2);
-		s3 += atomic_long_read(&rdp->expedited_workdone3);
+		s1 += atomic_long_read(&rdp->exp_workdone1);
+		s2 += atomic_long_read(&rdp->exp_workdone2);
+		s3 += atomic_long_read(&rdp->exp_workdone3);
 	}
-	seq_printf(m, "s=%lu wd0=%lu wd1=%lu wd2=%lu wd3=%lu n=%lu enq=%d sc=%lu\n",
-		   rsp->expedited_sequence, s0, s1, s2, s3,
+	seq_printf(m, "s=%lu wd1=%lu wd2=%lu wd3=%lu n=%lu enq=%d sc=%lu\n",
+		   rsp->expedited_sequence, s1, s2, s3,
 		   atomic_long_read(&rsp->expedited_normal),
 		   atomic_read(&rsp->expedited_need_qs),
 		   rsp->expedited_sequence / 2);
diff --git a/kernel/rcu/update.c b/kernel/rcu/update.c
index ca828b41c938..3ccdc8eebc5a 100644
--- a/kernel/rcu/update.c
+++ b/kernel/rcu/update.c
@@ -67,7 +67,7 @@ static int rcu_normal_after_boot;
 module_param(rcu_normal_after_boot, int, 0);
 #endif /* #ifndef CONFIG_TINY_RCU */
 
-#if defined(CONFIG_DEBUG_LOCK_ALLOC) && defined(CONFIG_PREEMPT_COUNT)
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
 /**
  * rcu_read_lock_sched_held() - might we be in RCU-sched read-side critical section?
  *
@@ -111,7 +111,7 @@ int rcu_read_lock_sched_held(void)
 		return 0;
 	if (debug_locks)
 		lockdep_opinion = lock_is_held(&rcu_sched_lock_map);
-	return lockdep_opinion || preempt_count() != 0 || irqs_disabled();
+	return lockdep_opinion || !preemptible();
 }
 EXPORT_SYMBOL(rcu_read_lock_sched_held);
 #endif
diff --git a/kernel/torture.c b/kernel/torture.c
index 44aa462d033f..fa0bdeee17ac 100644
--- a/kernel/torture.c
+++ b/kernel/torture.c
@@ -451,6 +451,7 @@ static int torture_shutdown(void *arg)
 		torture_shutdown_hook();
 	else
 		VERBOSE_TOROUT_STRING("No torture_shutdown_hook(), skipping.");
+	ftrace_dump(DUMP_ALL);
 	kernel_power_off();	/* Shut down the system. */
 	return 0;
 }
@@ -602,8 +603,9 @@ bool torture_init_begin(char *ttype, bool v, int *runnable)
 {
 	mutex_lock(&fullstop_mutex);
 	if (torture_type != NULL) {
-		pr_alert("torture_init_begin: refusing %s init: %s running",
+		pr_alert("torture_init_begin: Refusing %s init: %s running.\n",
 			 ttype, torture_type);
+		pr_alert("torture_init_begin: One torture test at a time!\n");
 		mutex_unlock(&fullstop_mutex);
 		return false;
 	}