freebsd-src/sys/kern/subr_epoch.c

/*-
 * SPDX-License-Identifier: BSD-2-Clause
 *
 * Copyright (c) 2018, Matthew Macy <mmacy@freebsd.org>
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/counter.h>
#include <sys/epoch.h>
#include <sys/gtaskqueue.h>
#include <sys/kernel.h>
#include <sys/limits.h>
#include <sys/lock.h>
#include <sys/malloc.h>
#include <sys/mutex.h>
#include <sys/pcpu.h>
#include <sys/proc.h>
#include <sys/sched.h>
#include <sys/sx.h>
#include <sys/smp.h>
#include <sys/sysctl.h>
#include <sys/turnstile.h>
#ifdef EPOCH_TRACE
#include <machine/stdarg.h>
#include <sys/stack.h>
#include <sys/tree.h>
#endif
#include <vm/vm.h>
#include <vm/vm_extern.h>
#include <vm/vm_kern.h>
#include <vm/uma.h>

#include <machine/stack.h>

#include <ck_epoch.h>

#ifdef __amd64__
#define EPOCH_ALIGN CACHE_LINE_SIZE*2
#else
#define EPOCH_ALIGN CACHE_LINE_SIZE
#endif

TAILQ_HEAD (epoch_tdlist, epoch_tracker);
typedef struct epoch_record {
	ck_epoch_record_t er_record;
	struct epoch_context er_drain_ctx;
	struct epoch *er_parent;
	volatile struct epoch_tdlist er_tdlist;
	volatile uint32_t er_gen;
	uint32_t er_cpuid;
#ifdef INVARIANTS
	/* Used to verify record ownership for non-preemptible epochs. */
	struct thread *er_td;
#endif
} __aligned(EPOCH_ALIGN)     *epoch_record_t;

struct epoch {
	struct ck_epoch e_epoch __aligned(EPOCH_ALIGN);
	epoch_record_t e_pcpu_record;
	int	e_in_use;
	int	e_flags;
	struct sx e_drain_sx;
	struct mtx e_drain_mtx;
	volatile int e_drain_count;
	const char *e_name;
};

/* arbitrary --- needs benchmarking */
#define MAX_ADAPTIVE_SPIN 100
#define MAX_EPOCHS 64

CTASSERT(sizeof(ck_epoch_entry_t) == sizeof(struct epoch_context));
SYSCTL_NODE(_kern, OID_AUTO, epoch, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
    "epoch information");
SYSCTL_NODE(_kern_epoch, OID_AUTO, stats, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
    "epoch stats");

/* Stats. */
static counter_u64_t block_count;

SYSCTL_COUNTER_U64(_kern_epoch_stats, OID_AUTO, nblocked, CTLFLAG_RW,
    &block_count, "# of times a thread was in an epoch when epoch_wait was called");
static counter_u64_t migrate_count;

SYSCTL_COUNTER_U64(_kern_epoch_stats, OID_AUTO, migrations, CTLFLAG_RW,
    &migrate_count, "# of times thread was migrated to another CPU in epoch_wait");
static counter_u64_t turnstile_count;

SYSCTL_COUNTER_U64(_kern_epoch_stats, OID_AUTO, ncontended, CTLFLAG_RW,
    &turnstile_count, "# of times a thread was blocked on a lock in an epoch during an epoch_wait");
static counter_u64_t switch_count;

SYSCTL_COUNTER_U64(_kern_epoch_stats, OID_AUTO, switches, CTLFLAG_RW,
    &switch_count, "# of times a thread voluntarily context switched in epoch_wait");
static counter_u64_t epoch_call_count;

SYSCTL_COUNTER_U64(_kern_epoch_stats, OID_AUTO, epoch_calls, CTLFLAG_RW,
    &epoch_call_count, "# of times a callback was deferred");
static counter_u64_t epoch_call_task_count;

SYSCTL_COUNTER_U64(_kern_epoch_stats, OID_AUTO, epoch_call_tasks, CTLFLAG_RW,
    &epoch_call_task_count, "# of times a callback task was run");

TAILQ_HEAD (threadlist, thread);

CK_STACK_CONTAINER(struct ck_epoch_entry, stack_entry,
    ck_epoch_entry_container)

static struct epoch epoch_array[MAX_EPOCHS];

DPCPU_DEFINE(struct grouptask, epoch_cb_task);
DPCPU_DEFINE(int, epoch_cb_count);

static __read_mostly int inited;
__read_mostly epoch_t global_epoch;
__read_mostly epoch_t global_epoch_preempt;

static void epoch_call_task(void *context __unused);
static 	uma_zone_t pcpu_zone_record;

static struct sx epoch_sx;

#define	EPOCH_LOCK() sx_xlock(&epoch_sx)
#define	EPOCH_UNLOCK() sx_xunlock(&epoch_sx)

static epoch_record_t
epoch_currecord(epoch_t epoch)
{

	return (zpcpu_get(epoch->e_pcpu_record));
}

#ifdef EPOCH_TRACE
struct stackentry {
	RB_ENTRY(stackentry) se_node;
	struct stack se_stack;
};

static int
stackentry_compare(struct stackentry *a, struct stackentry *b)
{

	if (a->se_stack.depth > b->se_stack.depth)
		return (1);
	if (a->se_stack.depth < b->se_stack.depth)
		return (-1);
	for (int i = 0; i < a->se_stack.depth; i++) {
		if (a->se_stack.pcs[i] > b->se_stack.pcs[i])
			return (1);
		if (a->se_stack.pcs[i] < b->se_stack.pcs[i])
			return (-1);
	}

	return (0);
}

RB_HEAD(stacktree, stackentry) epoch_stacks = RB_INITIALIZER(&epoch_stacks);
RB_GENERATE_STATIC(stacktree, stackentry, se_node, stackentry_compare);

static struct mtx epoch_stacks_lock;
MTX_SYSINIT(epochstacks, &epoch_stacks_lock, "epoch_stacks", MTX_DEF);

static bool epoch_trace_stack_print = true;
SYSCTL_BOOL(_kern_epoch, OID_AUTO, trace_stack_print, CTLFLAG_RWTUN,
    &epoch_trace_stack_print, 0, "Print stack traces on epoch reports");

static void epoch_trace_report(const char *fmt, ...) __printflike(1, 2);
static inline void
epoch_trace_report(const char *fmt, ...)
{
	va_list ap;
	struct stackentry se, *new;

	stack_save(&se.se_stack);

	/* Tree is never reduced - go lockless. */
	if (RB_FIND(stacktree, &epoch_stacks, &se) != NULL)
		return;

	new = malloc(sizeof(*new), M_STACK, M_NOWAIT);
	if (new != NULL) {
		bcopy(&se.se_stack, &new->se_stack, sizeof(struct stack));

		mtx_lock(&epoch_stacks_lock);
		new = RB_INSERT(stacktree, &epoch_stacks, new);
		mtx_unlock(&epoch_stacks_lock);
		if (new != NULL)
			free(new, M_STACK);
	}

	va_start(ap, fmt);
	(void)vprintf(fmt, ap);
	va_end(ap);
	if (epoch_trace_stack_print)
		stack_print_ddb(&se.se_stack);
}

static inline void
epoch_trace_enter(struct thread *td, epoch_t epoch, epoch_tracker_t et,
    const char *file, int line)
{
	epoch_tracker_t iet;

	SLIST_FOREACH(iet, &td->td_epochs, et_tlink) {
		if (iet->et_epoch != epoch)
			continue;
		epoch_trace_report("Recursively entering epoch %s "
		    "at %s:%d, previously entered at %s:%d\n",
		    epoch->e_name, file, line,
		    iet->et_file, iet->et_line);
	}
	et->et_epoch = epoch;
	et->et_file = file;
	et->et_line = line;
	et->et_flags = 0;
	SLIST_INSERT_HEAD(&td->td_epochs, et, et_tlink);
}

static inline void
epoch_trace_exit(struct thread *td, epoch_t epoch, epoch_tracker_t et,
    const char *file, int line)
{

	if (SLIST_FIRST(&td->td_epochs) != et) {
		epoch_trace_report("Exiting epoch %s in a not nested order "
		    "at %s:%d. Most recently entered %s at %s:%d\n",
		    epoch->e_name,
		    file, line,
		    SLIST_FIRST(&td->td_epochs)->et_epoch->e_name,
		    SLIST_FIRST(&td->td_epochs)->et_file,
		    SLIST_FIRST(&td->td_epochs)->et_line);
		/* This will panic if et is not anywhere on td_epochs. */
		SLIST_REMOVE(&td->td_epochs, et, epoch_tracker, et_tlink);
	} else
		SLIST_REMOVE_HEAD(&td->td_epochs, et_tlink);
	if (et->et_flags & ET_REPORT_EXIT)
		printf("Td %p exiting epoch %s at %s:%d\n", td, epoch->e_name,
		    file, line);
}

/* Used by assertions that check thread state before going to sleep. */
void
epoch_trace_list(struct thread *td)
{
	epoch_tracker_t iet;

	SLIST_FOREACH(iet, &td->td_epochs, et_tlink)
		printf("Epoch %s entered at %s:%d\n", iet->et_epoch->e_name,
		    iet->et_file, iet->et_line);
}

void
epoch_where_report(epoch_t epoch)
{
	epoch_record_t er;
	struct epoch_tracker *tdwait;

	MPASS(epoch != NULL);
	MPASS((epoch->e_flags & EPOCH_PREEMPT) != 0);
	MPASS(!THREAD_CAN_SLEEP());
	critical_enter();
	er = epoch_currecord(epoch);
	TAILQ_FOREACH(tdwait, &er->er_tdlist, et_link)
		if (tdwait->et_td == curthread)
			break;
	critical_exit();
	if (tdwait != NULL) {
		tdwait->et_flags |= ET_REPORT_EXIT;
		printf("Td %p entered epoch %s at %s:%d\n", curthread,
		    epoch->e_name, tdwait->et_file, tdwait->et_line);
	}
}
#endif /* EPOCH_TRACE */

static void
epoch_init(void *arg __unused)
{
	int cpu;

	block_count = counter_u64_alloc(M_WAITOK);
	migrate_count = counter_u64_alloc(M_WAITOK);
	turnstile_count = counter_u64_alloc(M_WAITOK);
	switch_count = counter_u64_alloc(M_WAITOK);
	epoch_call_count = counter_u64_alloc(M_WAITOK);
	epoch_call_task_count = counter_u64_alloc(M_WAITOK);

	pcpu_zone_record = uma_zcreate("epoch_record pcpu",
	    sizeof(struct epoch_record), NULL, NULL, NULL, NULL,
	    UMA_ALIGN_PTR, UMA_ZONE_PCPU);
	CPU_FOREACH(cpu) {
		GROUPTASK_INIT(DPCPU_ID_PTR(cpu, epoch_cb_task), 0,
		    epoch_call_task, NULL);
		taskqgroup_attach_cpu(qgroup_softirq,
		    DPCPU_ID_PTR(cpu, epoch_cb_task), NULL, cpu, NULL, NULL,
		    "epoch call task");
	}
#ifdef EPOCH_TRACE
	SLIST_INIT(&thread0.td_epochs);
#endif
	sx_init(&epoch_sx, "epoch-sx");
	inited = 1;
	global_epoch = epoch_alloc("Global", 0);
	global_epoch_preempt = epoch_alloc("Global preemptible", EPOCH_PREEMPT);
}
SYSINIT(epoch, SI_SUB_EPOCH, SI_ORDER_FIRST, epoch_init, NULL);

#if !defined(EARLY_AP_STARTUP)
static void
epoch_init_smp(void *dummy __unused)
{
	inited = 2;
}
SYSINIT(epoch_smp, SI_SUB_SMP + 1, SI_ORDER_FIRST, epoch_init_smp, NULL);
#endif

static void
epoch_ctor(epoch_t epoch)
{
	epoch_record_t er;
	int cpu;

	epoch->e_pcpu_record = uma_zalloc_pcpu(pcpu_zone_record, M_WAITOK);
	CPU_FOREACH(cpu) {
		er = zpcpu_get_cpu(epoch->e_pcpu_record, cpu);
		bzero(er, sizeof(*er));
		ck_epoch_register(&epoch->e_epoch, &er->er_record, NULL);
		TAILQ_INIT((struct threadlist *)(uintptr_t)&er->er_tdlist);
		er->er_cpuid = cpu;
		er->er_parent = epoch;
	}
}

static void
epoch_adjust_prio(struct thread *td, u_char prio)
{

	thread_lock(td);
	sched_prio(td, prio);
	thread_unlock(td);
}

epoch_t
epoch_alloc(const char *name, int flags)
{
	epoch_t epoch;
	int i;

	MPASS(name != NULL);

	if (__predict_false(!inited))
		panic("%s called too early in boot", __func__);

	EPOCH_LOCK();

	/*
	 * Find a free index in the epoch array. If no free index is
	 * found, try to use the index after the last one.
	 */
	for (i = 0;; i++) {
		/*
		 * If too many epochs are currently allocated,
		 * return NULL.
		 */
		if (i == MAX_EPOCHS) {
			epoch = NULL;
			goto done;
		}
		if (epoch_array[i].e_in_use == 0)
			break;
	}

	epoch = epoch_array + i;
	ck_epoch_init(&epoch->e_epoch);
	epoch_ctor(epoch);
	epoch->e_flags = flags;
	epoch->e_name = name;
	sx_init(&epoch->e_drain_sx, "epoch-drain-sx");
	mtx_init(&epoch->e_drain_mtx, "epoch-drain-mtx", NULL, MTX_DEF);

	/*
	 * Set e_in_use last, because when this field is set the
	 * epoch_call_task() function will start scanning this epoch
	 * structure.
	 */
	atomic_store_rel_int(&epoch->e_in_use, 1);
done:
	EPOCH_UNLOCK();
	return (epoch);
}

void
epoch_free(epoch_t epoch)
{
#ifdef INVARIANTS
	int cpu;
#endif

	EPOCH_LOCK();

	MPASS(epoch->e_in_use != 0);

	epoch_drain_callbacks(epoch);

	atomic_store_rel_int(&epoch->e_in_use, 0);
	/*
	 * Make sure the epoch_call_task() function see e_in_use equal
	 * to zero, by calling epoch_wait() on the global_epoch:
	 */
	epoch_wait(global_epoch);
#ifdef INVARIANTS
	CPU_FOREACH(cpu) {
		epoch_record_t er;

		er = zpcpu_get_cpu(epoch->e_pcpu_record, cpu);

		/*
		 * Sanity check: none of the records should be in use anymore.
		 * We drained callbacks above and freeing the pcpu records is
		 * imminent.
		 */
		MPASS(er->er_td == NULL);
		MPASS(TAILQ_EMPTY(&er->er_tdlist));
	}
#endif
	uma_zfree_pcpu(pcpu_zone_record, epoch->e_pcpu_record);
	mtx_destroy(&epoch->e_drain_mtx);
	sx_destroy(&epoch->e_drain_sx);
	memset(epoch, 0, sizeof(*epoch));

	EPOCH_UNLOCK();
}

#define INIT_CHECK(epoch)					\
	do {							\
		if (__predict_false((epoch) == NULL))		\
			return;					\
	} while (0)

void
_epoch_enter_preempt(epoch_t epoch, epoch_tracker_t et EPOCH_FILE_LINE)
{
	struct epoch_record *er;
	struct thread *td;

	MPASS(cold || epoch != NULL);
	td = curthread;
	MPASS(kstack_contains(td, (vm_offset_t)et, sizeof(*et)));

	INIT_CHECK(epoch);
	MPASS(epoch->e_flags & EPOCH_PREEMPT);

#ifdef EPOCH_TRACE
	epoch_trace_enter(td, epoch, et, file, line);
#endif
	et->et_td = td;
	THREAD_NO_SLEEPING();
	critical_enter();
	sched_pin();
	et->et_old_priority = td->td_priority;
	er = epoch_currecord(epoch);
	/* Record-level tracking is reserved for non-preemptible epochs. */
	MPASS(er->er_td == NULL);
	TAILQ_INSERT_TAIL(&er->er_tdlist, et, et_link);
	ck_epoch_begin(&er->er_record, &et->et_section);
	critical_exit();
}

void
epoch_enter(epoch_t epoch)
{
	epoch_record_t er;

	MPASS(cold || epoch != NULL);
	INIT_CHECK(epoch);
	critical_enter();
	er = epoch_currecord(epoch);
#ifdef INVARIANTS
	if (er->er_record.active == 0) {
		MPASS(er->er_td == NULL);
		er->er_td = curthread;
	} else {
		/* We've recursed, just make sure our accounting isn't wrong. */
		MPASS(er->er_td == curthread);
	}
#endif
	ck_epoch_begin(&er->er_record, NULL);
}

void
_epoch_exit_preempt(epoch_t epoch, epoch_tracker_t et EPOCH_FILE_LINE)
{
	struct epoch_record *er;
	struct thread *td;

	INIT_CHECK(epoch);
	td = curthread;
	critical_enter();
	sched_unpin();
	THREAD_SLEEPING_OK();
	er = epoch_currecord(epoch);
	MPASS(epoch->e_flags & EPOCH_PREEMPT);
	MPASS(et != NULL);
	MPASS(et->et_td == td);
#ifdef INVARIANTS
	et->et_td = (void*)0xDEADBEEF;
	/* Record-level tracking is reserved for non-preemptible epochs. */
	MPASS(er->er_td == NULL);
#endif
	ck_epoch_end(&er->er_record, &et->et_section);
	TAILQ_REMOVE(&er->er_tdlist, et, et_link);
	er->er_gen++;
	if (__predict_false(et->et_old_priority != td->td_priority))
		epoch_adjust_prio(td, et->et_old_priority);
	critical_exit();
#ifdef EPOCH_TRACE
	epoch_trace_exit(td, epoch, et, file, line);
#endif
}

void
epoch_exit(epoch_t epoch)
{
	epoch_record_t er;

	INIT_CHECK(epoch);
	er = epoch_currecord(epoch);
	ck_epoch_end(&er->er_record, NULL);
#ifdef INVARIANTS
	MPASS(er->er_td == curthread);
	if (er->er_record.active == 0)
		er->er_td = NULL;
#endif
	critical_exit();
}

/*
 * epoch_block_handler_preempt() is a callback from the CK code when another
 * thread is currently in an epoch section.
 */
static void
epoch_block_handler_preempt(struct ck_epoch *global __unused,
    ck_epoch_record_t *cr, void *arg __unused)
{
	epoch_record_t record;
	struct thread *td, *owner, *curwaittd;
	struct epoch_tracker *tdwait;
	struct turnstile *ts;
	struct lock_object *lock;
	int spincount, gen;
	int locksheld __unused;

	record = __containerof(cr, struct epoch_record, er_record);
	td = curthread;
	locksheld = td->td_locks;
	spincount = 0;
	counter_u64_add(block_count, 1);
	/*
	 * We lost a race and there's no longer any threads
	 * on the CPU in an epoch section.
	 */
	if (TAILQ_EMPTY(&record->er_tdlist))
		return;

	if (record->er_cpuid != curcpu) {
		/*
		 * If the head of the list is running, we can wait for it
		 * to remove itself from the list and thus save us the
		 * overhead of a migration
		 */
		gen = record->er_gen;
		thread_unlock(td);
		/*
		 * We can't actually check if the waiting thread is running
		 * so we simply poll for it to exit before giving up and
		 * migrating.
		 */
		do {
			cpu_spinwait();
		} while (!TAILQ_EMPTY(&record->er_tdlist) &&
				 gen == record->er_gen &&
				 spincount++ < MAX_ADAPTIVE_SPIN);
		thread_lock(td);
		/*
		 * If the generation has changed we can poll again
		 * otherwise we need to migrate.
		 */
		if (gen != record->er_gen)
			return;
		/*
		 * Being on the same CPU as that of the record on which
		 * we need to wait allows us access to the thread
		 * list associated with that CPU. We can then examine the
		 * oldest thread in the queue and wait on its turnstile
		 * until it resumes and so on until a grace period
		 * elapses.
		 *
		 */
		counter_u64_add(migrate_count, 1);
		sched_bind(td, record->er_cpuid);
		/*
		 * At this point we need to return to the ck code
		 * to scan to see if a grace period has elapsed.
		 * We can't move on to check the thread list, because
		 * in the meantime new threads may have arrived that
		 * in fact belong to a different epoch.
		 */
		return;
	}
	/*
	 * Try to find a thread in an epoch section on this CPU
	 * waiting on a turnstile. Otherwise find the lowest
	 * priority thread (highest prio value) and drop our priority
	 * to match to allow it to run.
	 */
	TAILQ_FOREACH(tdwait, &record->er_tdlist, et_link) {
		/*
		 * Propagate our priority to any other waiters to prevent us
		 * from starving them. They will have their original priority
		 * restore on exit from epoch_wait().
		 */
		curwaittd = tdwait->et_td;
		if (!TD_IS_INHIBITED(curwaittd) && curwaittd->td_priority > td->td_priority) {
			critical_enter();
			thread_unlock(td);
			thread_lock(curwaittd);
			sched_prio(curwaittd, td->td_priority);
			thread_unlock(curwaittd);
			thread_lock(td);
			critical_exit();
		}
		if (TD_IS_INHIBITED(curwaittd) && TD_ON_LOCK(curwaittd) &&
		    ((ts = curwaittd->td_blocked) != NULL)) {
			/*
			 * We unlock td to allow turnstile_wait to reacquire
			 * the thread lock. Before unlocking it we enter a
			 * critical section to prevent preemption after we
			 * reenable interrupts by dropping the thread lock in
			 * order to prevent curwaittd from getting to run.
			 */
			critical_enter();
			thread_unlock(td);

			if (turnstile_lock(ts, &lock, &owner)) {
				if (ts == curwaittd->td_blocked) {
					MPASS(TD_IS_INHIBITED(curwaittd) &&
					    TD_ON_LOCK(curwaittd));
					critical_exit();
					turnstile_wait(ts, owner,
					    curwaittd->td_tsqueue);
					counter_u64_add(turnstile_count, 1);
					thread_lock(td);
					return;
				}
				turnstile_unlock(ts, lock);
			}
			thread_lock(td);
			critical_exit();
			KASSERT(td->td_locks == locksheld,
			    ("%d extra locks held", td->td_locks - locksheld));
		}
	}
	/*
	 * We didn't find any threads actually blocked on a lock
	 * so we have nothing to do except context switch away.
	 */
	counter_u64_add(switch_count, 1);
	mi_switch(SW_VOL | SWT_RELINQUISH);
	/*
	 * It is important the thread lock is dropped while yielding
	 * to allow other threads to acquire the lock pointed to by
	 * TDQ_LOCKPTR(td). Currently mi_switch() will unlock the
	 * thread lock before returning. Else a deadlock like
	 * situation might happen.
	 */
	thread_lock(td);
}

void
epoch_wait_preempt(epoch_t epoch)
{
	struct thread *td;
	int was_bound;
	int old_cpu;
	int old_pinned;
	u_char old_prio;
	int locks __unused;

	MPASS(cold || epoch != NULL);
	INIT_CHECK(epoch);
	td = curthread;
#ifdef INVARIANTS
	locks = curthread->td_locks;
	MPASS(epoch->e_flags & EPOCH_PREEMPT);
	if ((epoch->e_flags & EPOCH_LOCKED) == 0)
		WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
		    "epoch_wait() can be long running");
	KASSERT(!in_epoch(epoch), ("epoch_wait_preempt() called in the middle "
	    "of an epoch section of the same epoch"));
#endif
	DROP_GIANT();
	thread_lock(td);

	old_cpu = PCPU_GET(cpuid);
	old_pinned = td->td_pinned;
	old_prio = td->td_priority;
	was_bound = sched_is_bound(td);
	sched_unbind(td);
	td->td_pinned = 0;
	sched_bind(td, old_cpu);

	ck_epoch_synchronize_wait(&epoch->e_epoch, epoch_block_handler_preempt,
	    NULL);

	/* restore CPU binding, if any */
	if (was_bound != 0) {
		sched_bind(td, old_cpu);
	} else {
		/* get thread back to initial CPU, if any */
		if (old_pinned != 0)
			sched_bind(td, old_cpu);
		sched_unbind(td);
	}
	/* restore pinned after bind */
	td->td_pinned = old_pinned;

	/* restore thread priority */
	sched_prio(td, old_prio);
	thread_unlock(td);
	PICKUP_GIANT();
	KASSERT(td->td_locks == locks,
	    ("%d residual locks held", td->td_locks - locks));
}

static void
epoch_block_handler(struct ck_epoch *g __unused, ck_epoch_record_t *c __unused,
    void *arg __unused)
{
	cpu_spinwait();
}

void
epoch_wait(epoch_t epoch)
{

	MPASS(cold || epoch != NULL);
	INIT_CHECK(epoch);
	MPASS(epoch->e_flags == 0);
	critical_enter();
	ck_epoch_synchronize_wait(&epoch->e_epoch, epoch_block_handler, NULL);
	critical_exit();
}

void
epoch_call(epoch_t epoch, epoch_callback_t callback, epoch_context_t ctx)
{
	epoch_record_t er;
	ck_epoch_entry_t *cb;

	cb = (void *)ctx;

	MPASS(callback);
	/* too early in boot to have epoch set up */
	if (__predict_false(epoch == NULL))
		goto boottime;
#if !defined(EARLY_AP_STARTUP)
	if (__predict_false(inited < 2))
		goto boottime;
#endif

	critical_enter();
	*DPCPU_PTR(epoch_cb_count) += 1;
	er = epoch_currecord(epoch);
	ck_epoch_call(&er->er_record, cb, (ck_epoch_cb_t *)callback);
	critical_exit();
	return;
boottime:
	callback(ctx);
}

static void
epoch_call_task(void *arg __unused)
{
	ck_stack_entry_t *cursor, *head, *next;
	ck_epoch_record_t *record;
	epoch_record_t er;
	epoch_t epoch;
	ck_stack_t cb_stack;
	int i, npending, total;

	ck_stack_init(&cb_stack);
	critical_enter();
	epoch_enter(global_epoch);
	for (total = i = 0; i != MAX_EPOCHS; i++) {
		epoch = epoch_array + i;
		if (__predict_false(
		    atomic_load_acq_int(&epoch->e_in_use) == 0))
			continue;
		er = epoch_currecord(epoch);
		record = &er->er_record;
		if ((npending = record->n_pending) == 0)
			continue;
		ck_epoch_poll_deferred(record, &cb_stack);
		total += npending - record->n_pending;
	}
	epoch_exit(global_epoch);
	*DPCPU_PTR(epoch_cb_count) -= total;
	critical_exit();

	counter_u64_add(epoch_call_count, total);
	counter_u64_add(epoch_call_task_count, 1);

	head = ck_stack_batch_pop_npsc(&cb_stack);
	for (cursor = head; cursor != NULL; cursor = next) {
		struct ck_epoch_entry *entry =
		    ck_epoch_entry_container(cursor);

		next = CK_STACK_NEXT(cursor);
		entry->function(entry);
	}
}

static int
in_epoch_verbose_preempt(epoch_t epoch, int dump_onfail)
{
	epoch_record_t er;
	struct epoch_tracker *tdwait;
	struct thread *td;

	MPASS(epoch != NULL);
	MPASS((epoch->e_flags & EPOCH_PREEMPT) != 0);
	td = curthread;
	if (THREAD_CAN_SLEEP())
		return (0);
	critical_enter();
	er = epoch_currecord(epoch);
	TAILQ_FOREACH(tdwait, &er->er_tdlist, et_link)
		if (tdwait->et_td == td) {
			critical_exit();
			return (1);
		}
#ifdef INVARIANTS
	if (dump_onfail) {
		MPASS(td->td_pinned);
		printf("cpu: %d id: %d\n", curcpu, td->td_tid);
		TAILQ_FOREACH(tdwait, &er->er_tdlist, et_link)
			printf("td_tid: %d ", tdwait->et_td->td_tid);
		printf("\n");
	}
#endif
	critical_exit();
	return (0);
}

#ifdef INVARIANTS
static void
epoch_assert_nocpu(epoch_t epoch, struct thread *td)
{
	epoch_record_t er;
	int cpu;
	bool crit;

	crit = td->td_critnest > 0;

	/* Check for a critical section mishap. */
	CPU_FOREACH(cpu) {
		er = zpcpu_get_cpu(epoch->e_pcpu_record, cpu);
		KASSERT(er->er_td != td,
		    ("%s critical section in epoch '%s', from cpu %d",
		    (crit ? "exited" : "re-entered"), epoch->e_name, cpu));
	}
}
#else
#define	epoch_assert_nocpu(e, td) do {} while (0)
#endif

int
in_epoch_verbose(epoch_t epoch, int dump_onfail)
{
	epoch_record_t er;
	struct thread *td;

	if (__predict_false((epoch) == NULL))
		return (0);
	if ((epoch->e_flags & EPOCH_PREEMPT) != 0)
		return (in_epoch_verbose_preempt(epoch, dump_onfail));

	/*
	 * The thread being in a critical section is a necessary
	 * condition to be correctly inside a non-preemptible epoch,
	 * so it's definitely not in this epoch.
	 */
	td = curthread;
	if (td->td_critnest == 0) {
		epoch_assert_nocpu(epoch, td);
		return (0);
	}

	/*
	 * The current cpu is in a critical section, so the epoch record will be
	 * stable for the rest of this function.  Knowing that the record is not
	 * active is sufficient for knowing whether we're in this epoch or not,
	 * since it's a pcpu record.
	 */
	er = epoch_currecord(epoch);
	if (er->er_record.active == 0) {
		epoch_assert_nocpu(epoch, td);
		return (0);
	}

	MPASS(er->er_td == td);
	return (1);
}

int
in_epoch(epoch_t epoch)
{
	return (in_epoch_verbose(epoch, 0));
}

static void
epoch_drain_cb(struct epoch_context *ctx)
{
	struct epoch *epoch =
	    __containerof(ctx, struct epoch_record, er_drain_ctx)->er_parent;

	if (atomic_fetchadd_int(&epoch->e_drain_count, -1) == 1) {
		mtx_lock(&epoch->e_drain_mtx);
		wakeup(epoch);
		mtx_unlock(&epoch->e_drain_mtx);
	}
}

void
epoch_drain_callbacks(epoch_t epoch)
{
	epoch_record_t er;
	struct thread *td;
	int was_bound;
	int old_pinned;
	int old_cpu;
	int cpu;

	WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
	    "epoch_drain_callbacks() may sleep!");

	/* too early in boot to have epoch set up */
	if (__predict_false(epoch == NULL))
		return;
#if !defined(EARLY_AP_STARTUP)
	if (__predict_false(inited < 2))
		return;
#endif
	DROP_GIANT();

	sx_xlock(&epoch->e_drain_sx);
	mtx_lock(&epoch->e_drain_mtx);

	td = curthread;
	thread_lock(td);
	old_cpu = PCPU_GET(cpuid);
	old_pinned = td->td_pinned;
	was_bound = sched_is_bound(td);
	sched_unbind(td);
	td->td_pinned = 0;

	CPU_FOREACH(cpu)
		epoch->e_drain_count++;
	CPU_FOREACH(cpu) {
		er = zpcpu_get_cpu(epoch->e_pcpu_record, cpu);
		sched_bind(td, cpu);
		epoch_call(epoch, &epoch_drain_cb, &er->er_drain_ctx);
	}

	/* restore CPU binding, if any */
	if (was_bound != 0) {
		sched_bind(td, old_cpu);
	} else {
		/* get thread back to initial CPU, if any */
		if (old_pinned != 0)
			sched_bind(td, old_cpu);
		sched_unbind(td);
	}
	/* restore pinned after bind */
	td->td_pinned = old_pinned;

	thread_unlock(td);

	while (epoch->e_drain_count != 0)
		msleep(epoch, &epoch->e_drain_mtx, PZERO, "EDRAIN", 0);

	mtx_unlock(&epoch->e_drain_mtx);
	sx_xunlock(&epoch->e_drain_sx);

	PICKUP_GIANT();
}