bind/lib/dns/qpcache.c

/*
 * Copyright (C) Internet Systems Consortium, Inc. ("ISC")
 *
 * SPDX-License-Identifier: MPL-2.0
 *
 * This Source Code Form is subject to the terms of the Mozilla Public
 * License, v. 2.0. If a copy of the MPL was not distributed with this
 * file, you can obtain one at https://mozilla.org/MPL/2.0/.
 *
 * See the COPYRIGHT file distributed with this work for additional
 * information regarding copyright ownership.
 */

/*! \file */

#include <inttypes.h>
#include <stdalign.h>
#include <stdbool.h>

#include <isc/ascii.h>
#include <isc/async.h>
#include <isc/atomic.h>
#include <isc/file.h>
#include <isc/heap.h>
#include <isc/hex.h>
#include <isc/log.h>
#include <isc/loop.h>
#include <isc/mem.h>
#include <isc/mutex.h>
#include <isc/os.h>
#include <isc/queue.h>
#include <isc/random.h>
#include <isc/refcount.h>
#include <isc/result.h>
#include <isc/rwlock.h>
#include <isc/sieve.h>
#include <isc/stdio.h>
#include <isc/string.h>
#include <isc/time.h>
#include <isc/urcu.h>
#include <isc/util.h>

#include <dns/callbacks.h>
#include <dns/db.h>
#include <dns/dbiterator.h>
#include <dns/fixedname.h>
#include <dns/masterdump.h>
#include <dns/nsec.h>
#include <dns/qp.h>
#include <dns/rdata.h>
#include <dns/rdataset.h>
#include <dns/rdatasetiter.h>
#include <dns/rdataslab.h>
#include <dns/rdatastruct.h>
#include <dns/stats.h>
#include <dns/time.h>
#include <dns/view.h>

#include "db_p.h"
#include "qpcache_p.h"

#ifndef DNS_QPCACHE_LOG_STATS_LEVEL
#define DNS_QPCACHE_LOG_STATS_LEVEL 3
#endif

#define CHECK(op)                            \
	do {                                 \
		result = (op);               \
		if (result != ISC_R_SUCCESS) \
			goto failure;        \
	} while (0)

#define EXISTS(header)                                 \
	((atomic_load_acquire(&(header)->attributes) & \
	  DNS_SLABHEADERATTR_NONEXISTENT) == 0)
#define NXDOMAIN(header)                               \
	((atomic_load_acquire(&(header)->attributes) & \
	  DNS_SLABHEADERATTR_NXDOMAIN) != 0)
#define STALE(header)                                  \
	((atomic_load_acquire(&(header)->attributes) & \
	  DNS_SLABHEADERATTR_STALE) != 0)
#define STALE_WINDOW(header)                           \
	((atomic_load_acquire(&(header)->attributes) & \
	  DNS_SLABHEADERATTR_STALE_WINDOW) != 0)
#define OPTOUT(header)                                 \
	((atomic_load_acquire(&(header)->attributes) & \
	  DNS_SLABHEADERATTR_OPTOUT) != 0)
#define NEGATIVE(header)                               \
	((atomic_load_acquire(&(header)->attributes) & \
	  DNS_SLABHEADERATTR_NEGATIVE) != 0)
#define PREFETCH(header)                               \
	((atomic_load_acquire(&(header)->attributes) & \
	  DNS_SLABHEADERATTR_PREFETCH) != 0)
#define ZEROTTL(header)                                \
	((atomic_load_acquire(&(header)->attributes) & \
	  DNS_SLABHEADERATTR_ZEROTTL) != 0)
#define ANCIENT(header)                                \
	((atomic_load_acquire(&(header)->attributes) & \
	  DNS_SLABHEADERATTR_ANCIENT) != 0)
#define STATCOUNT(header)                              \
	((atomic_load_acquire(&(header)->attributes) & \
	  DNS_SLABHEADERATTR_STATCOUNT) != 0)

#define STALE_TTL(header, qpdb) \
	(NXDOMAIN(header) ? 0 : qpdb->common.serve_stale_ttl)

#define ACTIVE(header, now)            \
	(((header)->expire > (now)) || \
	 ((header)->expire == (now) && ZEROTTL(header)))

#define EXPIREDOK(iterator) \
	(((iterator)->common.options & DNS_DB_EXPIREDOK) != 0)

#define STALEOK(iterator) (((iterator)->common.options & DNS_DB_STALEOK) != 0)

#define KEEPSTALE(qpdb) ((qpdb)->common.serve_stale_ttl > 0)

/*%
 * Note that "impmagic" is not the first four bytes of the struct, so
 * ISC_MAGIC_VALID cannot be used.
 */
#define QPDB_MAGIC ISC_MAGIC('Q', 'P', 'D', '4')
#define VALID_QPDB(qpdb) \
	((qpdb) != NULL && (qpdb)->common.impmagic == QPDB_MAGIC)

#define HEADERNODE(h) ((qpcnode_t *)((h)->node))

/*
 * Allow clients with a virtual time of up to 10 seconds in the past to see
 * records that would have otherwise have expired.
 */
#define QPDB_VIRTUAL 10

/*
 * This defines the number of headers that we try to expire each time the
 * expire_ttl_headers() is run.  The number should be small enough, so the
 * TTL-based header expiration doesn't take too long, but it should be large
 * enough, so we expire enough headers if their TTL is clustered.
 */
#define DNS_QPDB_EXPIRE_TTL_COUNT 10

/*%
 * Forward declarations
 */
typedef struct qpcache qpcache_t;

/*%
 * This is the structure that is used for each node in the qp trie of
 * trees.
 */
typedef struct qpcnode qpcnode_t;
struct qpcnode {
	DBNODE_FIELDS;

	qpcache_t *qpdb;

	uint8_t			: 0;
	unsigned int delegating : 1;
	unsigned int nspace	: 2; /*%< range is 0..3 */
	unsigned int havensec	: 1;
	uint8_t			: 0;

	/*
	 * 'erefs' counts external references held by a caller: for
	 * example, it could be incremented by dns_db_findnode(),
	 * and decremented by dns_db_detachnode().
	 *
	 * 'references' counts internal references to the node object,
	 * including the one held by the QP trie so the node won't be
	 * deleted while it's quiescently stored in the database - even
	 * though 'erefs' may be zero because no external caller is
	 * using it at the time.
	 *
	 * Generally when 'erefs' is incremented or decremented,
	 * 'references' is too. When both go to zero (meaning callers
	 * and the database have both released the object) the object
	 * is freed.
	 *
	 * Whenever 'erefs' is incremented from zero, we also aquire a
	 * node use reference (see 'qpcache->references' below), and
	 * release it when 'erefs' goes back to zero. This prevents the
	 * database from being shut down until every caller has released
	 * all nodes.
	 */
	isc_refcount_t references;
	isc_refcount_t erefs;
	void *data;

	/*%
	 * NOTE: The 'dirty' flag is protected by the node lock, so
	 * this bitfield has to be separated from the one above.
	 * We don't want it to share the same qword with bits
	 * that can be accessed without the node lock.
	 */
	uint8_t	      : 0;
	uint8_t dirty : 1;
	uint8_t	      : 0;

	/*%
	 * Used for dead nodes cleaning.  This linked list is used to mark nodes
	 * which have no data any longer, but we cannot unlink at that exact
	 * moment because we did not or could not obtain a write lock on the
	 * tree.
	 */
	isc_queue_node_t deadlink;
};

/*%
 * One bucket structure will be created for each loop, and
 * nodes in the database will evenly distributed among buckets
 * to reduce contention between threads.
 */
typedef struct qpcache_bucket {
	/*%
	 * Temporary storage for stale cache nodes and dynamically
	 * deleted nodes that await being cleaned up.
	 */
	isc_queue_t deadnodes;

	/* Per-bucket lock. */
	isc_rwlock_t lock;

	/*
	 * The heap is used for TTL based expiry.  Note that qpcache->hmctx
	 * is the memory context to use for heap memory; this differs from
	 * the main database memory context, which is qpcache->common.mctx.
	 */
	isc_heap_t *heap;

	/* SIEVE-LRU cache cleaning state. */
	ISC_SIEVE(dns_slabheader_t) sieve;

	/* Padding to prevent false sharing between locks. */
	uint8_t __padding[ISC_OS_CACHELINE_SIZE -
			  (sizeof(isc_queue_t) + sizeof(isc_rwlock_t) +
			   sizeof(isc_heap_t *) +
			   sizeof(ISC_SIEVE(dns_slabheader_t))) %
				  ISC_OS_CACHELINE_SIZE];

} qpcache_bucket_t;

struct qpcache {
	/* Unlocked. */
	dns_db_t common;
	/* Locks the data in this struct */
	isc_rwlock_t lock;
	/* Locks the tree structure (prevents nodes appearing/disappearing) */
	isc_rwlock_t tree_lock;

	/*
	 * NOTE: 'references' is NOT the global reference counter for
	 * the database object handled by dns_db_attach() and _detach();
	 * that one is 'common.references'.
	 *
	 * Instead, 'references' counts the number of nodes being used by
	 * at least one external caller. (It's called 'references' to
	 * leverage the ISC_REFCOUNT_STATIC macros, but 'nodes_in_use'
	 * might be a clearer name.)
	 *
	 * One additional reference to this counter is held by the database
	 * object itself. When 'common.references' goes to zero, that
	 * reference is released. When in turn 'references' goes to zero,
	 * the database is shut down and freed.
	 */
	isc_refcount_t references;

	dns_stats_t *rrsetstats;
	isc_stats_t *cachestats;

	uint32_t maxrrperset;	 /* Maximum RRs per RRset */
	uint32_t maxtypepername; /* Maximum number of RR types per owner */

	/*
	 * The time after a failed lookup, where stale answers from cache
	 * may be used directly in a DNS response without attempting a
	 * new iterative lookup.
	 */
	uint32_t serve_stale_refresh;

	/* Locked by tree_lock. */
	dns_qp_t *tree;
	dns_qp_t *nsec;

	isc_mem_t *hmctx; /* Memory context for the heaps */

	size_t buckets_count;
	qpcache_bucket_t buckets[]; /* attribute((counted_by(buckets_count))) */
};

#ifdef DNS_DB_NODETRACE
#define qpcache_ref(ptr)   qpcache__ref(ptr, __func__, __FILE__, __LINE__)
#define qpcache_unref(ptr) qpcache__unref(ptr, __func__, __FILE__, __LINE__)
#define qpcache_attach(ptr, ptrp) \
	qpcache__attach(ptr, ptrp, __func__, __FILE__, __LINE__)
#define qpcache_detach(ptrp) qpcache__detach(ptrp, __func__, __FILE__, __LINE__)
ISC_REFCOUNT_STATIC_TRACE_DECL(qpcache);
#else
ISC_REFCOUNT_STATIC_DECL(qpcache);
#endif

/*%
 * Search Context
 */
typedef struct {
	qpcache_t *qpdb;
	unsigned int options;
	dns_qpchain_t chain;
	dns_qpiter_t iter;
	bool need_cleanup;
	qpcnode_t *zonecut;
	dns_slabheader_t *zonecut_header;
	dns_slabheader_t *zonecut_sigheader;
	isc_stdtime_t now;
} qpc_search_t;

#ifdef DNS_DB_NODETRACE
#define qpcnode_ref(ptr)   qpcnode__ref(ptr, __func__, __FILE__, __LINE__)
#define qpcnode_unref(ptr) qpcnode__unref(ptr, __func__, __FILE__, __LINE__)
#define qpcnode_attach(ptr, ptrp) \
	qpcnode__attach(ptr, ptrp, __func__, __FILE__, __LINE__)
#define qpcnode_detach(ptrp) qpcnode__detach(ptrp, __func__, __FILE__, __LINE__)
ISC_REFCOUNT_STATIC_TRACE_DECL(qpcnode);
#else
ISC_REFCOUNT_STATIC_DECL(qpcnode);
#endif

/*
 * Node methods forward declarations
 */
static void
qpcnode_attachnode(dns_dbnode_t *source, dns_dbnode_t **targetp DNS__DB_FLARG);
static void
qpcnode_detachnode(dns_dbnode_t **nodep DNS__DB_FLARG);
static void
qpcnode_locknode(dns_dbnode_t *node, isc_rwlocktype_t type);
static void
qpcnode_unlocknode(dns_dbnode_t *node, isc_rwlocktype_t type);
static void
qpcnode_deletedata(dns_dbnode_t *node, void *data);
static void
qpcnode_expiredata(dns_dbnode_t *node, void *data);

static dns_dbnode_methods_t qpcnode_methods = (dns_dbnode_methods_t){
	.attachnode = qpcnode_attachnode,
	.detachnode = qpcnode_detachnode,
	.locknode = qpcnode_locknode,
	.unlocknode = qpcnode_unlocknode,
	.deletedata = qpcnode_deletedata,
	.expiredata = qpcnode_expiredata,
};

/* QP methods */
static void
qp_attach(void *uctx, void *pval, uint32_t ival);
static void
qp_detach(void *uctx, void *pval, uint32_t ival);
static size_t
qp_makekey(dns_qpkey_t key, void *uctx, void *pval, uint32_t ival);
static void
qp_triename(void *uctx, char *buf, size_t size);

static dns_qpmethods_t qpmethods = {
	qp_attach,
	qp_detach,
	qp_makekey,
	qp_triename,
};

static void
qp_attach(void *uctx ISC_ATTR_UNUSED, void *pval,
	  uint32_t ival ISC_ATTR_UNUSED) {
	qpcnode_t *data = pval;
	qpcnode_ref(data);
}

static void
qp_detach(void *uctx ISC_ATTR_UNUSED, void *pval,
	  uint32_t ival ISC_ATTR_UNUSED) {
	qpcnode_t *data = pval;
	qpcnode_detach(&data);
}

static size_t
qp_makekey(dns_qpkey_t key, void *uctx ISC_ATTR_UNUSED, void *pval,
	   uint32_t ival ISC_ATTR_UNUSED) {
	qpcnode_t *data = pval;
	return dns_qpkey_fromname(key, &data->name, data->nspace);
}

static void
qp_triename(void *uctx ISC_ATTR_UNUSED, char *buf, size_t size) {
	snprintf(buf, size, "qpdb-lite");
}

static void
rdatasetiter_destroy(dns_rdatasetiter_t **iteratorp DNS__DB_FLARG);
static isc_result_t
rdatasetiter_first(dns_rdatasetiter_t *iterator DNS__DB_FLARG);
static isc_result_t
rdatasetiter_next(dns_rdatasetiter_t *iterator DNS__DB_FLARG);
static void
rdatasetiter_current(dns_rdatasetiter_t *iterator,
		     dns_rdataset_t *rdataset DNS__DB_FLARG);

static dns_rdatasetitermethods_t rdatasetiter_methods = {
	rdatasetiter_destroy, rdatasetiter_first, rdatasetiter_next,
	rdatasetiter_current
};

typedef struct qpc_rditer {
	dns_rdatasetiter_t common;
	dns_slabheader_t *current;
} qpc_rditer_t;

static void
dbiterator_destroy(dns_dbiterator_t **iteratorp DNS__DB_FLARG);
static isc_result_t
dbiterator_first(dns_dbiterator_t *iterator DNS__DB_FLARG);
static isc_result_t
dbiterator_last(dns_dbiterator_t *iterator DNS__DB_FLARG);
static isc_result_t
dbiterator_seek(dns_dbiterator_t *iterator,
		const dns_name_t *name DNS__DB_FLARG);
static isc_result_t
dbiterator_prev(dns_dbiterator_t *iterator DNS__DB_FLARG);
static isc_result_t
dbiterator_next(dns_dbiterator_t *iterator DNS__DB_FLARG);
static isc_result_t
dbiterator_current(dns_dbiterator_t *iterator, dns_dbnode_t **nodep,
		   dns_name_t *name DNS__DB_FLARG);
static isc_result_t
dbiterator_pause(dns_dbiterator_t *iterator);
static isc_result_t
dbiterator_origin(dns_dbiterator_t *iterator, dns_name_t *name);

static dns_dbiteratormethods_t dbiterator_methods = {
	dbiterator_destroy, dbiterator_first, dbiterator_last,
	dbiterator_seek,    dbiterator_prev,  dbiterator_next,
	dbiterator_current, dbiterator_pause, dbiterator_origin
};

/*
 * Note that the QP cache database only needs a single QP iterator, because
 * unlike the QP zone database, NSEC3 records are cached in the main tree.
 *
 * If we ever implement synth-from-dnssec using NSEC3 records, we'll need
 * to have a separate tree for NSEC3 records, and to copy in the more complex
 * iterator implementation from qpzone.c.
 */
typedef struct qpc_dbit {
	dns_dbiterator_t common;
	bool paused;
	isc_rwlocktype_t tree_locked;
	isc_result_t result;
	dns_fixedname_t fixed;
	dns_name_t *name;
	dns_qpiter_t iter;
	qpcnode_t *node;
} qpc_dbit_t;

static void
qpcache__destroy(qpcache_t *qpdb);

static dns_dbmethods_t qpdb_cachemethods;

static void
cleanup_deadnodes_cb(void *arg);

/*%
 * 'init_count' is used to initialize 'newheader->count' which in turn
 * is used to determine where in the cycle rrset-order cyclic starts.
 * We don't lock this as we don't care about simultaneous updates.
 */
static atomic_uint_fast16_t init_count = 0;

/*
 * Locking
 *
 * If a routine is going to lock more than one lock in this module, then
 * the locking must be done in the following order:
 *
 *      Tree Lock
 *
 *      Node Lock       (Only one from the set may be locked at one time by
 *                       any caller)
 *
 *      Database Lock
 *
 * Failure to follow this hierarchy can result in deadlock.
 */

/*
 * Cache-eviction routines.
 */

static void
expireheader(dns_slabheader_t *header, isc_rwlocktype_t *nlocktypep,
	     isc_rwlocktype_t *tlocktypep, dns_expire_t reason DNS__DB_FLARG);

static size_t
rdataset_size(dns_slabheader_t *header) {
	if (EXISTS(header)) {
		return dns_rdataslab_size(header);
	}

	return sizeof(*header);
}

static void
expire_lru_headers(qpcache_t *qpdb, uint32_t idx, size_t requested,
		   isc_rwlocktype_t *nlocktypep,
		   isc_rwlocktype_t *tlocktypep DNS__DB_FLARG) {
	size_t expired = 0;

	do {
		dns_slabheader_t *header =
			ISC_SIEVE_NEXT(qpdb->buckets[idx].sieve, visited, link);
		if (header == NULL) {
			return;
		}

		ISC_SIEVE_UNLINK(qpdb->buckets[idx].sieve, header, link);

		expired += rdataset_size(header);

		expireheader(header, nlocktypep, tlocktypep,
			     dns_expire_lru DNS__DB_FLARG_PASS);
	} while (expired < requested);
}

static void
qpcache_miss(qpcache_t *qpdb, dns_slabheader_t *newheader,
	     isc_rwlocktype_t *nlocktypep,
	     isc_rwlocktype_t *tlocktypep DNS__DB_FLARG) {
	uint32_t idx = HEADERNODE(newheader)->locknum;

	isc_heap_insert(qpdb->buckets[idx].heap, newheader);
	newheader->heap = qpdb->buckets[idx].heap;

	if (isc_mem_isovermem(qpdb->common.mctx)) {
		/*
		 * Maximum estimated size of the data being added: The size
		 * of the rdataset, plus a new QP database node and nodename,
		 * and a possible additional NSEC node and nodename. Also add
		 * a 12k margin for a possible QP-trie chunk allocation.
		 * (It's okay to overestimate, we want to get cache memory
		 * down quickly.)
		 */

		size_t purgesize =
			2 * (sizeof(qpcnode_t) +
			     dns_name_size(&HEADERNODE(newheader)->name)) +
			rdataset_size(newheader) + QP_SAFETY_MARGIN;

		expire_lru_headers(qpdb, idx, purgesize, nlocktypep,
				   tlocktypep DNS__DB_FLARG_PASS);
	}

	ISC_SIEVE_INSERT(qpdb->buckets[idx].sieve, newheader, link);
}

static void
qpcache_hit(qpcache_t *qpdb ISC_ATTR_UNUSED, dns_slabheader_t *header) {
	/*
	 * On cache hit, we only mark the header as seen.
	 */
	ISC_SIEVE_MARK(header, visited);
}

/*
 * DB Routines
 */

static void
clean_stale_headers(dns_slabheader_t *top) {
	dns_slabheader_t *d = NULL, *down_next = NULL;

	for (d = top->down; d != NULL; d = down_next) {
		down_next = d->down;
		dns_slabheader_destroy(&d);
	}
	top->down = NULL;
}

static void
clean_cache_node(qpcache_t *qpdb, qpcnode_t *node) {
	dns_slabheader_t *current = NULL, *top_prev = NULL, *top_next = NULL;

	/*
	 * Caller must be holding the node lock.
	 */

	for (current = node->data; current != NULL; current = top_next) {
		top_next = current->next;
		clean_stale_headers(current);
		/*
		 * If current is nonexistent, ancient, or stale and
		 * we are not keeping stale, we can clean it up.
		 */
		if (!EXISTS(current) || ANCIENT(current) ||
		    (STALE(current) && !KEEPSTALE(qpdb)))
		{
			if (top_prev != NULL) {
				top_prev->next = current->next;
			} else {
				node->data = current->next;
			}
			dns_slabheader_destroy(&current);
		} else {
			top_prev = current;
		}
	}
	node->dirty = 0;
}

/*
 * tree_lock(write) must be held.
 */
static void
delete_node(qpcache_t *qpdb, qpcnode_t *node) {
	isc_result_t result = ISC_R_UNEXPECTED;

	if (isc_log_wouldlog(ISC_LOG_DEBUG(DNS_QPCACHE_LOG_STATS_LEVEL))) {
		char printname[DNS_NAME_FORMATSIZE];
		dns_name_format(&node->name, printname, sizeof(printname));
		isc_log_write(DNS_LOGCATEGORY_DATABASE, DNS_LOGMODULE_CACHE,
			      ISC_LOG_DEBUG(DNS_QPCACHE_LOG_STATS_LEVEL),
			      "delete_node(): %p %s (bucket %d)", node,
			      printname, node->locknum);
	}

	switch (node->nspace) {
	case DNS_DBNAMESPACE_NORMAL:
		if (node->havensec) {
			/*
			 * Delete the corresponding node from the auxiliary NSEC
			 * tree before deleting from the main tree.
			 */
			result = dns_qp_deletename(qpdb->nsec, &node->name,
						   DNS_DBNAMESPACE_NSEC, NULL,
						   NULL);
			if (result != ISC_R_SUCCESS) {
				isc_log_write(DNS_LOGCATEGORY_DATABASE,
					      DNS_LOGMODULE_CACHE,
					      ISC_LOG_WARNING,
					      "delete_node(): "
					      "dns_qp_deletename: %s",
					      isc_result_totext(result));
			}
		}
		result = dns_qp_deletename(qpdb->tree, &node->name,
					   node->nspace, NULL, NULL);
		break;
	case DNS_DBNAMESPACE_NSEC:
		result = dns_qp_deletename(qpdb->nsec, &node->name,
					   node->nspace, NULL, NULL);
		break;
	}
	if (result != ISC_R_SUCCESS) {
		isc_log_write(DNS_LOGCATEGORY_DATABASE, DNS_LOGMODULE_CACHE,
			      ISC_LOG_WARNING,
			      "delete_node(): "
			      "dns_qp_deletename: %s",
			      isc_result_totext(result));
	}
}

/*
 * The caller must specify its currect node and tree lock status.
 * It's okay for neither lock to be held if there are existing external
 * references to the node, but if this is the first external reference,
 * then the caller must be holding at least one lock.
 *
 * If incrementing erefs from zero, we also increment the node use counter
 * in the qpcache object.
 *
 * This function is called from qpcnode_acquire(), so that internal
 * and external references are acquired at the same time, and from
 * qpcnode_release() when we only need to increase the internal references.
 */
static void
qpcnode_erefs_increment(qpcache_t *qpdb, qpcnode_t *node,
			isc_rwlocktype_t nlocktype,
			isc_rwlocktype_t tlocktype DNS__DB_FLARG) {
	uint_fast32_t refs = isc_refcount_increment0(&node->erefs);

#if DNS_DB_NODETRACE
	fprintf(stderr, "incr:node:%s:%s:%u:%p->erefs = %" PRIuFAST32 "\n",
		func, file, line, node, refs + 1);
#endif

	if (refs > 0) {
		return;
	}

	/*
	 * this is the first external reference to the node.
	 *
	 * we need to hold the node or tree lock to avoid
	 * incrementing the reference count while also deleting
	 * the node. delete_node() is always protected by both
	 * tree and node locks being write-locked.
	 */
	INSIST(nlocktype != isc_rwlocktype_none ||
	       tlocktype != isc_rwlocktype_none);

	qpcache_ref(qpdb);
}

static void
qpcnode_acquire(qpcache_t *qpdb, qpcnode_t *node, isc_rwlocktype_t nlocktype,
		isc_rwlocktype_t tlocktype DNS__DB_FLARG) {
	qpcnode_ref(node);
	qpcnode_erefs_increment(qpdb, node, nlocktype,
				tlocktype DNS__DB_FLARG_PASS);
}

/*
 * Decrement the external references to a node. If the counter
 * goes to zero, decrement the node use counter in the qpcache object
 * as well, and return true. Otherwise return false.
 */
static bool
qpcnode_erefs_decrement(qpcache_t *qpdb, qpcnode_t *node DNS__DB_FLARG) {
	uint_fast32_t refs = isc_refcount_decrement(&node->erefs);

#if DNS_DB_NODETRACE
	fprintf(stderr, "decr:node:%s:%s:%u:%p->erefs = %" PRIuFAST32 "\n",
		func, file, line, node, refs - 1);
#endif
	if (refs > 1) {
		return false;
	}

	qpcache_unref(qpdb);
	return true;
}

/*
 * Caller must be holding a node lock, either read or write.
 *
 * Note that the lock must be held even when node references are
 * atomically modified; in that case the decrement operation itself does not
 * have to be protected, but we must avoid a race condition where multiple
 * threads are decreasing the reference to zero simultaneously and at least
 * one of them is going to free the node.
 *
 * This calls dec_erefs() to decrement the external node reference counter,
 * (and possibly the node use counter), cleans up and deletes the node
 * if necessary, then decrements the internal reference counter as well.
 */
static void
qpcnode_release(qpcache_t *qpdb, qpcnode_t *node, isc_rwlocktype_t *nlocktypep,
		isc_rwlocktype_t *tlocktypep DNS__DB_FLARG) {
	REQUIRE(*nlocktypep != isc_rwlocktype_none);

	if (!qpcnode_erefs_decrement(qpdb, node DNS__DB_FLARG_PASS)) {
		goto unref;
	}

	/* Handle easy and typical case first. */
	if (!node->dirty && node->data != NULL) {
		goto unref;
	}

	if (*nlocktypep == isc_rwlocktype_read) {
		/*
		 * The external reference count went to zero and the node
		 * is dirty or has no data, so we might want to delete it.
		 * To do that, we'll need a write lock. If we don't already
		 * have one, we have to make sure nobody else has
		 * acquired a reference in the meantime, so we increment
		 * erefs (but NOT references!), upgrade the node lock,
		 * decrement erefs again, and see if it's still zero.
		 *
		 * We can't really assume anything about the result code of
		 * erefs_increment.  If another thread acquires reference it
		 * will be larger than 0, if it doesn't it is going to be 0.
		 */
		isc_rwlock_t *nlock = &qpdb->buckets[node->locknum].lock;
		qpcnode_erefs_increment(qpdb, node, *nlocktypep,
					*tlocktypep DNS__DB_FLARG_PASS);
		NODE_FORCEUPGRADE(nlock, nlocktypep);
		if (!qpcnode_erefs_decrement(qpdb, node DNS__DB_FLARG_PASS)) {
			goto unref;
		}
	}

	if (node->dirty) {
		clean_cache_node(qpdb, node);
	}

	if (node->data != NULL) {
		goto unref;
	}

	if (*tlocktypep == isc_rwlocktype_write) {
		/*
		 * We can delete the node if we have the tree write lock.
		 */
		delete_node(qpdb, node);
	} else {
		/*
		 * If we don't have the tree lock, we will add this node to a
		 * linked list of nodes in this locking bucket which we will
		 * free later.
		 */
		qpcnode_acquire(qpdb, node, *nlocktypep,
				*tlocktypep DNS__DB_FLARG_PASS);

		isc_queue_node_init(&node->deadlink);
		if (!isc_queue_enqueue_entry(
			    &qpdb->buckets[node->locknum].deadnodes, node,
			    deadlink))
		{
			/* Queue was empty, trigger new cleaning */
			isc_loop_t *loop = isc_loop_get(node->locknum);

			qpcache_ref(qpdb);
			isc_async_run(loop, cleanup_deadnodes_cb, qpdb);
		}
	}

unref:
	qpcnode_unref(node);
}

static void
update_rrsetstats(dns_stats_t *stats, const dns_typepair_t typepair,
		  const uint_least16_t hattributes, const bool increment) {
	dns_rdatastatstype_t statattributes = 0;
	dns_rdatastatstype_t base = 0;
	dns_rdatastatstype_t type;
	dns_slabheader_t *header = &(dns_slabheader_t){
		.typepair = typepair,
		.attributes = hattributes,
	};

	if (!EXISTS(header) || !STATCOUNT(header)) {
		return;
	}

	if (NEGATIVE(header)) {
		if (NXDOMAIN(header)) {
			statattributes = DNS_RDATASTATSTYPE_ATTR_NXDOMAIN;
		} else {
			statattributes = DNS_RDATASTATSTYPE_ATTR_NXRRSET;
			base = DNS_TYPEPAIR_COVERS(header->typepair);
		}
	} else {
		base = DNS_TYPEPAIR_TYPE(header->typepair);
	}

	if (STALE(header)) {
		statattributes |= DNS_RDATASTATSTYPE_ATTR_STALE;
	}
	if (ANCIENT(header)) {
		statattributes |= DNS_RDATASTATSTYPE_ATTR_ANCIENT;
	}

	type = DNS_RDATASTATSTYPE_VALUE(base, statattributes);
	if (increment) {
		dns_rdatasetstats_increment(stats, type);
	} else {
		dns_rdatasetstats_decrement(stats, type);
	}
}

static void
mark(dns_slabheader_t *header, uint_least16_t flag) {
	uint_least16_t attributes = atomic_load_acquire(&header->attributes);
	uint_least16_t newattributes = 0;
	dns_stats_t *stats = NULL;

	qpcache_t *qpdb = HEADERNODE(header)->qpdb;
	/*
	 * If we are already ancient there is nothing to do.
	 */
	do {
		if ((attributes & flag) != 0) {
			return;
		}
		newattributes = attributes | flag;
	} while (!atomic_compare_exchange_weak_acq_rel(
		&header->attributes, &attributes, newattributes));

	/*
	 * Decrement and increment the stats counter for the appropriate
	 * RRtype.
	 */
	stats = dns_db_getrrsetstats(&qpdb->common);
	if (stats != NULL) {
		update_rrsetstats(stats, header->typepair, attributes, false);
		update_rrsetstats(stats, header->typepair, newattributes, true);
	}
}

static void
setttl(dns_slabheader_t *header, isc_stdtime_t newts) {
	isc_stdtime_t oldts = header->expire;

	header->expire = newts;

	if (header->heap == NULL || header->heap_index == 0 || newts == oldts) {
		return;
	}

	if (newts < oldts) {
		isc_heap_increased(header->heap, header->heap_index);
	} else {
		isc_heap_decreased(header->heap, header->heap_index);
	}

	if (newts == 0) {
		isc_heap_delete(header->heap, header->heap_index);
	}
}

static void
mark_ancient(dns_slabheader_t *header) {
	setttl(header, 0);
	mark(header, DNS_SLABHEADERATTR_ANCIENT);
	HEADERNODE(header)->dirty = 1;
}

/*
 * Caller must hold the node (write) lock.
 */
static void
expireheader(dns_slabheader_t *header, isc_rwlocktype_t *nlocktypep,
	     isc_rwlocktype_t *tlocktypep, dns_expire_t reason DNS__DB_FLARG) {
	mark_ancient(header);

	if (isc_refcount_current(&HEADERNODE(header)->erefs) == 0) {
		qpcache_t *qpdb = HEADERNODE(header)->qpdb;

		/*
		 * If no one else is using the node, we can clean it up now.
		 * We first need to gain a new reference to the node to meet a
		 * requirement of qpcnode_release().
		 */
		qpcnode_acquire(qpdb, HEADERNODE(header), *nlocktypep,
				*tlocktypep DNS__DB_FLARG_PASS);
		qpcnode_release(qpdb, HEADERNODE(header), nlocktypep,
				tlocktypep DNS__DB_FLARG_PASS);

		if (qpdb->cachestats == NULL) {
			return;
		}

		switch (reason) {
		case dns_expire_ttl:
			isc_stats_increment(qpdb->cachestats,
					    dns_cachestatscounter_deletettl);
			break;
		case dns_expire_lru:
			isc_stats_increment(qpdb->cachestats,
					    dns_cachestatscounter_deletelru);
			break;
		default:
			break;
		}
	}
}

static void
update_cachestats(qpcache_t *qpdb, isc_result_t result) {
	if (qpdb->cachestats == NULL) {
		return;
	}

	switch (result) {
	case DNS_R_COVERINGNSEC:
		isc_stats_increment(qpdb->cachestats,
				    dns_cachestatscounter_coveringnsec);
		FALLTHROUGH;
	case ISC_R_SUCCESS:
	case DNS_R_CNAME:
	case DNS_R_DNAME:
	case DNS_R_DELEGATION:
	case DNS_R_NCACHENXDOMAIN:
	case DNS_R_NCACHENXRRSET:
		isc_stats_increment(qpdb->cachestats,
				    dns_cachestatscounter_hits);
		break;
	default:
		isc_stats_increment(qpdb->cachestats,
				    dns_cachestatscounter_misses);
	}
}

static void
bindrdataset(qpcache_t *qpdb, qpcnode_t *node, dns_slabheader_t *header,
	     isc_stdtime_t now, isc_rwlocktype_t nlocktype,
	     isc_rwlocktype_t tlocktype,
	     dns_rdataset_t *rdataset DNS__DB_FLARG) {
	bool stale = STALE(header);
	bool ancient = ANCIENT(header);

	/*
	 * Caller must be holding the node reader lock.
	 * XXXJT: technically, we need a writer lock, since we'll increment
	 * the header count below.  However, since the actual counter value
	 * doesn't matter, we prioritize performance here.  (We may want to
	 * use atomic increment when available).
	 */

	if (rdataset == NULL) {
		return;
	}

	qpcnode_acquire(qpdb, node, nlocktype, tlocktype DNS__DB_FLARG_PASS);

	INSIST(rdataset->methods == NULL); /* We must be disassociated. */

	/*
	 * Mark header stale or ancient if the RRset is no longer active.
	 */
	if (!ACTIVE(header, now)) {
		dns_ttl_t stale_ttl = header->expire + STALE_TTL(header, qpdb);
		/*
		 * If this data is in the stale window keep it and if
		 * DNS_DBFIND_STALEOK is not set we tell the caller to
		 * skip this record.  We skip the records with ZEROTTL
		 * (these records should not be cached anyway).
		 */

		if (!ZEROTTL(header) && KEEPSTALE(qpdb) && stale_ttl > now) {
			stale = true;
		} else {
			/*
			 * We are not keeping stale, or it is outside the
			 * stale window. Mark ancient, i.e. ready for cleanup.
			 */
			ancient = true;
		}
	}

	rdataset->methods = &dns_rdataslab_rdatasetmethods;
	rdataset->rdclass = qpdb->common.rdclass;
	rdataset->type = DNS_TYPEPAIR_TYPE(header->typepair);
	rdataset->covers = DNS_TYPEPAIR_COVERS(header->typepair);
	rdataset->ttl = !ZEROTTL(header) ? header->expire - now : 0;
	rdataset->trust = header->trust;
	rdataset->resign = 0;

	if (NEGATIVE(header)) {
		rdataset->attributes.negative = true;
	}
	if (NXDOMAIN(header)) {
		rdataset->attributes.nxdomain = true;
	}
	if (OPTOUT(header)) {
		rdataset->attributes.optout = true;
	}
	if (PREFETCH(header)) {
		rdataset->attributes.prefetch = true;
	}

	if (stale && !ancient) {
		dns_ttl_t stale_ttl = header->expire + STALE_TTL(header, qpdb);
		if (stale_ttl > now) {
			rdataset->ttl = stale_ttl - now;
		} else {
			rdataset->ttl = 0;
		}
		if (STALE_WINDOW(header)) {
			rdataset->attributes.stale_window = true;
		}
		rdataset->attributes.stale = true;
		rdataset->expire = header->expire;
	} else if (!ACTIVE(header, now)) {
		rdataset->attributes.ancient = true;
		rdataset->ttl = 0;
	}

	rdataset->count = atomic_fetch_add_relaxed(&header->count, 1);

	rdataset->slab.db = (dns_db_t *)qpdb;
	rdataset->slab.node = (dns_dbnode_t *)node;
	rdataset->slab.raw = dns_slabheader_raw(header);
	rdataset->slab.iter_pos = NULL;
	rdataset->slab.iter_count = 0;

	/*
	 * Add noqname proof.
	 */
	rdataset->slab.noqname = header->noqname;
	if (header->noqname != NULL) {
		rdataset->attributes.noqname = true;
	}
	rdataset->slab.closest = header->closest;
	if (header->closest != NULL) {
		rdataset->attributes.closest = true;
	}
}

static void
bindrdatasets(qpcache_t *qpdb, qpcnode_t *qpnode, dns_slabheader_t *found,
	      dns_slabheader_t *foundsig, isc_stdtime_t now,
	      isc_rwlocktype_t nlocktype, isc_rwlocktype_t tlocktype,
	      dns_rdataset_t *rdataset,
	      dns_rdataset_t *sigrdataset DNS__DB_FLARG) {
	bindrdataset(qpdb, qpnode, found, now, nlocktype, tlocktype,
		     rdataset DNS__DB_FLARG_PASS);
	qpcache_hit(qpdb, found);
	if (!NEGATIVE(found) && foundsig != NULL) {
		bindrdataset(qpdb, qpnode, foundsig, now, nlocktype, tlocktype,
			     sigrdataset DNS__DB_FLARG_PASS);
		qpcache_hit(qpdb, foundsig);
	}
}

static isc_result_t
setup_delegation(qpc_search_t *search, dns_dbnode_t **nodep,
		 dns_rdataset_t *rdataset, dns_rdataset_t *sigrdataset,
		 isc_rwlocktype_t tlocktype DNS__DB_FLARG) {
	dns_typepair_t typepair;
	qpcnode_t *node = NULL;

	REQUIRE(search != NULL);
	REQUIRE(search->zonecut != NULL);
	REQUIRE(search->zonecut_header != NULL);

	/*
	 * The caller MUST NOT be holding any node locks.
	 */

	node = search->zonecut;
	typepair = search->zonecut_header->typepair;

	if (nodep != NULL) {
		/*
		 * Note that we don't have to increment the node's reference
		 * count here because we're going to use the reference we
		 * already have in the search block.
		 */
		*nodep = (dns_dbnode_t *)node;
		search->need_cleanup = false;
	}
	if (rdataset != NULL) {
		isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
		isc_rwlock_t *nlock =
			&search->qpdb->buckets[node->locknum].lock;
		NODE_RDLOCK(nlock, &nlocktype);
		bindrdatasets(search->qpdb, node, search->zonecut_header,
			      search->zonecut_sigheader, search->now, nlocktype,
			      tlocktype, rdataset,
			      sigrdataset DNS__DB_FLARG_PASS);
		NODE_UNLOCK(nlock, &nlocktype);
	}

	if (typepair == dns_rdatatype_dname) {
		return DNS_R_DNAME;
	}
	return DNS_R_DELEGATION;
}

static bool
check_stale_header(dns_slabheader_t *header, qpc_search_t *search,
		   dns_slabheader_t **header_prev) {
	if (ACTIVE(header, search->now)) {
		*header_prev = header;
		return false;
	}

	isc_stdtime_t stale = header->expire + STALE_TTL(header, search->qpdb);
	/*
	 * If this data is in the stale window keep it and if
	 * DNS_DBFIND_STALEOK is not set we tell the caller to
	 * skip this record.  We skip the records with ZEROTTL
	 * (these records should not be cached anyway).
	 */

	DNS_SLABHEADER_CLRATTR(header, DNS_SLABHEADERATTR_STALE_WINDOW);
	if (!ZEROTTL(header) && KEEPSTALE(search->qpdb) && stale > search->now)
	{
		mark(header, DNS_SLABHEADERATTR_STALE);
		*header_prev = header;
		/*
		 * If DNS_DBFIND_STALESTART is set then it means we
		 * failed to resolve the name during recursion, in
		 * this case we mark the time in which the refresh
		 * failed.
		 */
		if ((search->options & DNS_DBFIND_STALESTART) != 0) {
			atomic_store_release(&header->last_refresh_fail_ts,
					     search->now);
		} else if ((search->options & DNS_DBFIND_STALEENABLED) != 0 &&
			   search->now <
				   (atomic_load_acquire(
					    &header->last_refresh_fail_ts) +
				    search->qpdb->serve_stale_refresh))
		{
			/*
			 * If we are within interval between last
			 * refresh failure time + 'stale-refresh-time',
			 * then don't skip this stale entry but use it
			 * instead.
			 */
			DNS_SLABHEADER_SETATTR(header,
					       DNS_SLABHEADERATTR_STALE_WINDOW);
			return false;
		} else if ((search->options & DNS_DBFIND_STALETIMEOUT) != 0) {
			/*
			 * We want stale RRset due to timeout, so we
			 * don't skip it.
			 */
			return false;
		}
		return (search->options & DNS_DBFIND_STALEOK) == 0;
	}

	*header_prev = header;
	return true;
}

/*
 * Return true if we've found headers for both 'type' and RRSIG('type'),
 * or (optionally, if 'negtype' is nonzero) if we've found a single
 * negative header covering either 'negtype' or ANY.
 */
static bool
related_headers(dns_slabheader_t *header, dns_typepair_t typepair,
		dns_typepair_t sigpair, dns_typepair_t negpair,
		dns_slabheader_t **foundp, dns_slabheader_t **foundsigp,
		bool *matchp) {
	if (!EXISTS(header) || ANCIENT(header)) {
		return false;
	}

	if (header->typepair == typepair) {
		*foundp = header;
		SET_IF_NOT_NULL(matchp, true);
		if (*foundsigp != NULL) {
			return true;
		}
	} else if (header->typepair == sigpair) {
		*foundsigp = header;
		SET_IF_NOT_NULL(matchp, true);
		if (*foundp != NULL) {
			return true;
		}
	} else if (negpair != 0 && (header->typepair == RDATATYPE_NCACHEANY ||
				    header->typepair == negpair))
	{
		*foundp = header;
		*foundsigp = NULL;
		SET_IF_NOT_NULL(matchp, true);
		return true;
	}

	return false;
}

/*
 * Return true if we've found headers for both 'type' and RRSIG('type').
 */
static bool
both_headers(dns_slabheader_t *header, dns_rdatatype_t type,
	     dns_slabheader_t **foundp, dns_slabheader_t **foundsigp) {
	dns_typepair_t typepair = DNS_TYPEPAIR_VALUE(type, 0);
	dns_typepair_t sigpair = DNS_SIGTYPE(type);

	return related_headers(header, typepair, sigpair, 0, foundp, foundsigp,
			       NULL);
}

static isc_result_t
check_zonecut(qpcnode_t *node, void *arg DNS__DB_FLARG) {
	qpc_search_t *search = arg;
	dns_slabheader_t *header = NULL;
	dns_slabheader_t *header_prev = NULL, *header_next = NULL;
	dns_slabheader_t *dname_header = NULL, *sigdname_header = NULL;
	isc_result_t result;
	isc_rwlock_t *nlock = NULL;
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;

	REQUIRE(search->zonecut == NULL);

	nlock = &search->qpdb->buckets[node->locknum].lock;
	NODE_RDLOCK(nlock, &nlocktype);

	/*
	 * Look for a DNAME or RRSIG DNAME rdataset.
	 */
	for (header = node->data; header != NULL; header = header_next) {
		header_next = header->next;
		if (check_stale_header(header, search, &header_prev)) {
			continue;
		}

		if (both_headers(header, dns_rdatatype_dname, &dname_header,
				 &sigdname_header))
		{
			break;
		}
	}

	if (dname_header != NULL &&
	    (!DNS_TRUST_PENDING(dname_header->trust) ||
	     (search->options & DNS_DBFIND_PENDINGOK) != 0))
	{
		/*
		 * We increment the reference count on node to ensure that
		 * search->zonecut_header will still be valid later.
		 */
		qpcnode_acquire(search->qpdb, node, nlocktype,
				isc_rwlocktype_none DNS__DB_FLARG_PASS);
		search->zonecut = node;
		search->zonecut_header = dname_header;
		search->zonecut_sigheader = sigdname_header;
		search->need_cleanup = true;
		result = DNS_R_PARTIALMATCH;
	} else {
		result = DNS_R_CONTINUE;
	}

	NODE_UNLOCK(nlock, &nlocktype);

	return result;
}

static isc_result_t
find_deepest_zonecut(qpc_search_t *search, qpcnode_t *node,
		     dns_dbnode_t **nodep, dns_name_t *foundname,
		     dns_rdataset_t *rdataset,
		     dns_rdataset_t *sigrdataset DNS__DB_FLARG) {
	isc_result_t result = ISC_R_NOTFOUND;
	qpcache_t *qpdb = NULL;

	/*
	 * Caller must be holding the tree lock.
	 */

	qpdb = search->qpdb;

	for (int i = dns_qpchain_length(&search->chain) - 1; i >= 0; i--) {
		dns_slabheader_t *header = NULL;
		dns_slabheader_t *header_prev = NULL, *header_next = NULL;
		dns_slabheader_t *found = NULL, *foundsig = NULL;
		isc_rwlock_t *nlock = NULL;
		isc_rwlocktype_t nlocktype = isc_rwlocktype_none;

		dns_qpchain_node(&search->chain, i, NULL, (void **)&node, NULL);
		nlock = &qpdb->buckets[node->locknum].lock;

		NODE_RDLOCK(nlock, &nlocktype);

		/*
		 * Look for NS and RRSIG NS rdatasets.
		 */
		for (header = node->data; header != NULL; header = header_next)
		{
			header_next = header->next;
			if (check_stale_header(header, search, &header_prev)) {
				continue;
			}

			if (both_headers(header, dns_rdatatype_ns, &found,
					 &foundsig))
			{
				break;
			}
		}

		if (found != NULL) {
			/*
			 * If we have to set foundname, we do it before
			 * anything else.
			 */
			if (foundname != NULL) {
				dns_name_copy(&node->name, foundname);
			}
			result = DNS_R_DELEGATION;
			if (nodep != NULL) {
				qpcnode_acquire(
					search->qpdb, node, nlocktype,
					isc_rwlocktype_none DNS__DB_FLARG_PASS);
				*nodep = (dns_dbnode_t *)node;
			}
			bindrdatasets(search->qpdb, node, found, foundsig,
				      search->now, nlocktype,
				      isc_rwlocktype_none, rdataset,
				      sigrdataset DNS__DB_FLARG_PASS);
		}

		NODE_UNLOCK(nlock, &nlocktype);

		if (found != NULL) {
			break;
		}
	}

	return result;
}

/*
 * Look for a potentially covering NSEC in the cache where `name`
 * is known not to exist.  This uses the auxiliary NSEC tree to find
 * the potential NSEC owner. If found, we update 'foundname', 'nodep',
 * 'rdataset' and 'sigrdataset', and return DNS_R_COVERINGNSEC.
 * Otherwise, return ISC_R_NOTFOUND.
 */
static isc_result_t
find_coveringnsec(qpc_search_t *search, const dns_name_t *name,
		  dns_dbnode_t **nodep, dns_name_t *foundname,
		  dns_rdataset_t *rdataset,
		  dns_rdataset_t *sigrdataset DNS__DB_FLARG) {
	dns_fixedname_t fpredecessor, fixed;
	dns_name_t *predecessor = NULL, *fname = NULL;
	qpcnode_t *node = NULL;
	dns_qpiter_t iter;
	isc_result_t result;
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
	isc_rwlock_t *nlock = NULL;
	dns_slabheader_t *found = NULL, *foundsig = NULL;
	dns_slabheader_t *header = NULL;
	dns_slabheader_t *header_next = NULL, *header_prev = NULL;

	/*
	 * Look for the node in the auxilary tree.
	 */
	result = dns_qp_lookup(search->qpdb->nsec, name, DNS_DBNAMESPACE_NSEC,
			       NULL, &iter, NULL, (void **)&node, NULL);
	/*
	 * When DNS_R_PARTIALMATCH or ISC_R_NOTFOUND is returned from
	 * dns_qp_lookup there is potentially a covering NSEC present
	 * in the cache so we need to search for it.  Otherwise we are
	 * done here.
	 */
	if (result != DNS_R_PARTIALMATCH && result != ISC_R_NOTFOUND) {
		return ISC_R_NOTFOUND;
	}

	fname = dns_fixedname_initname(&fixed);
	predecessor = dns_fixedname_initname(&fpredecessor);

	/*
	 * Extract predecessor from iterator.
	 */
	result = dns_qpiter_current(&iter, predecessor, NULL, NULL);
	if (result != ISC_R_SUCCESS) {
		return ISC_R_NOTFOUND;
	}

	/*
	 * Lookup the predecessor in the main tree.
	 */
	node = NULL;
	result = dns_qp_getname(search->qpdb->tree, predecessor,
				DNS_DBNAMESPACE_NORMAL, (void **)&node, NULL);
	if (result != ISC_R_SUCCESS) {
		return result;
	}
	dns_name_copy(&node->name, fname);

	nlock = &search->qpdb->buckets[node->locknum].lock;
	NODE_RDLOCK(nlock, &nlocktype);
	for (header = node->data; header != NULL; header = header_next) {
		header_next = header->next;
		if (check_stale_header(header, search, &header_prev)) {
			continue;
		}

		if (DNS_TYPEPAIR_TYPE(header->typepair) == 0) {
			continue;
		}

		if (both_headers(header, dns_rdatatype_nsec, &found, &foundsig))
		{
			break;
		}
	}
	if (found != NULL) {
		if (nodep != NULL) {
			qpcnode_acquire(search->qpdb, node, nlocktype,
					isc_rwlocktype_none DNS__DB_FLARG_PASS);
			*nodep = (dns_dbnode_t *)node;
		}
		bindrdatasets(search->qpdb, node, found, foundsig, search->now,
			      nlocktype, isc_rwlocktype_none, rdataset,
			      sigrdataset DNS__DB_FLARG_PASS);
		dns_name_copy(fname, foundname);

		result = DNS_R_COVERINGNSEC;
	} else {
		result = ISC_R_NOTFOUND;
	}
	NODE_UNLOCK(nlock, &nlocktype);
	return result;
}

#define MISSING_ANSWER(found, options)                     \
	((found) == NULL ||                                \
	 (DNS_TRUST_ADDITIONAL((found)->trust) &&          \
	  (((options) & DNS_DBFIND_ADDITIONALOK) == 0)) || \
	 ((found)->trust == dns_trust_glue &&              \
	  (((options) & DNS_DBFIND_GLUEOK) == 0)) ||       \
	 (DNS_TRUST_PENDING((found)->trust) &&             \
	  (((options) & DNS_DBFIND_PENDINGOK) == 0)))

static void
qpc_search_init(qpc_search_t *search, qpcache_t *db, unsigned int options,
		isc_stdtime_t now) {
	/*
	 * qpc_search_t contains two structures with large buffers (dns_qpiter_t
	 * and dns_qpchain_t). Those two structures will be initialized later by
	 * dns_qp_lookup anyway.
	 * To avoid the overhead of zero initialization, we avoid designated
	 * initializers and initialize all "small" fields manually.
	 */
	search->qpdb = (qpcache_t *)db;
	search->options = options;
	/*
	 * qpch->in - Init by dns_qp_lookup
	 * qpiter - Init by dns_qp_lookup
	 */
	search->need_cleanup = false;
	search->now = now ? now : isc_stdtime_now();
	search->zonecut = NULL;
	search->zonecut_header = NULL;
	search->zonecut_sigheader = NULL;
}

static isc_result_t
qpcache_find(dns_db_t *db, const dns_name_t *name, dns_dbversion_t *version,
	     dns_rdatatype_t type, unsigned int options, isc_stdtime_t __now,
	     dns_dbnode_t **nodep, dns_name_t *foundname,
	     dns_rdataset_t *rdataset,
	     dns_rdataset_t *sigrdataset DNS__DB_FLARG) {
	qpcnode_t *node = NULL;
	isc_result_t result;
	bool cname_ok = true;
	bool found_noqname = false;
	bool all_negative = true;
	bool empty_node;
	isc_rwlock_t *nlock = NULL;
	isc_rwlocktype_t tlocktype = isc_rwlocktype_none;
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
	dns_slabheader_t *header = NULL;
	dns_slabheader_t *header_prev = NULL, *header_next = NULL;
	dns_slabheader_t *found = NULL, *nsheader = NULL;
	dns_slabheader_t *foundsig = NULL, *nssig = NULL, *cnamesig = NULL;
	dns_slabheader_t *nsecheader = NULL, *nsecsig = NULL;
	dns_typepair_t typepair, sigpair, negpair;

	qpc_search_t search;
	qpc_search_init(&search, (qpcache_t *)db, options, __now);

	REQUIRE(VALID_QPDB((qpcache_t *)db));
	REQUIRE(version == NULL);

	TREE_RDLOCK(&search.qpdb->tree_lock, &tlocktype);

	/*
	 * Search down from the root of the tree.
	 */
	result = dns_qp_lookup(search.qpdb->tree, name, DNS_DBNAMESPACE_NORMAL,
			       NULL, NULL, &search.chain, (void **)&node, NULL);
	if (result != ISC_R_NOTFOUND && foundname != NULL) {
		dns_name_copy(&node->name, foundname);
	}

	/*
	 * Check the QP chain to see if there's a node above us with a
	 * active DNAME or NS rdatasets.
	 *
	 * We're only interested in nodes above QNAME, so if the result
	 * was success, then we skip the last item in the chain.
	 */
	unsigned int len = dns_qpchain_length(&search.chain);
	if (result == ISC_R_SUCCESS) {
		len--;
	}

	for (unsigned int i = 0; i < len; i++) {
		isc_result_t zcresult;
		qpcnode_t *encloser = NULL;

		dns_qpchain_node(&search.chain, i, NULL, (void **)&encloser,
				 NULL);

		zcresult = check_zonecut(encloser,
					 (void *)&search DNS__DB_FLARG_PASS);
		if (zcresult != DNS_R_CONTINUE) {
			result = DNS_R_PARTIALMATCH;
			search.chain.len = i - 1;
			node = encloser;
			if (foundname != NULL) {
				dns_name_copy(&node->name, foundname);
			}
			break;
		}
	}

	if (result == DNS_R_PARTIALMATCH) {
		/*
		 * If we discovered a covering DNAME skip looking for a covering
		 * NSEC.
		 */
		if ((search.options & DNS_DBFIND_COVERINGNSEC) != 0 &&
		    (search.zonecut_header == NULL ||
		     search.zonecut_header->typepair != dns_rdatatype_dname))
		{
			result = find_coveringnsec(
				&search, name, nodep, foundname, rdataset,
				sigrdataset DNS__DB_FLARG_PASS);
			if (result == DNS_R_COVERINGNSEC) {
				goto tree_exit;
			}
		}
		if (search.zonecut != NULL) {
			result = setup_delegation(&search, nodep, rdataset,
						  sigrdataset,
						  tlocktype DNS__DB_FLARG_PASS);
			goto tree_exit;
		} else {
		find_ns:
			result = find_deepest_zonecut(
				&search, node, nodep, foundname, rdataset,
				sigrdataset DNS__DB_FLARG_PASS);
			goto tree_exit;
		}
	} else if (result != ISC_R_SUCCESS) {
		goto tree_exit;
	}

	/*
	 * Certain DNSSEC types are not subject to CNAME matching
	 * (RFC4035, section 2.5 and RFC3007).
	 *
	 * We don't check for RRSIG, because we don't store RRSIG records
	 * directly.
	 */
	if (type == dns_rdatatype_key || type == dns_rdatatype_nsec) {
		cname_ok = false;
	}

	/*
	 * We now go looking for rdata...
	 */

	nlock = &search.qpdb->buckets[node->locknum].lock;
	NODE_RDLOCK(nlock, &nlocktype);

	/*
	 * These pointers need to be reset here in case we did
	 * 'goto find_ns' from somewhere below.
	 */
	found = NULL;
	foundsig = NULL;
	typepair = DNS_TYPEPAIR_VALUE(type, 0);
	sigpair = DNS_SIGTYPE(type);
	negpair = DNS_TYPEPAIR_VALUE(0, type);
	nsheader = NULL;
	nsecheader = NULL;
	nssig = NULL;
	nsecsig = NULL;
	cnamesig = NULL;
	empty_node = true;
	header_prev = NULL;
	for (header = node->data; header != NULL; header = header_next) {
		header_next = header->next;
		if (check_stale_header(header, &search, &header_prev)) {
			continue;
		}

		if (!EXISTS(header) || ANCIENT(header)) {
			continue;
		}

		/*
		 * We now know that there is at least one active
		 * non-stale rdataset at this node.
		 */
		empty_node = false;

		if (header->noqname != NULL &&
		    header->trust == dns_trust_secure)
		{
			found_noqname = true;
		}

		if (!NEGATIVE(header)) {
			all_negative = false;
		}

		bool match = false;
		if (related_headers(header, typepair, sigpair, negpair, &found,
				    &foundsig, &match) &&
		    !MISSING_ANSWER(found, options))
		{
			/*
			 * We can't exit early until we have an answer with
			 * sufficient trust level, see MISSING_ANSWER() macro
			 * for details, because we might need NS or NSEC
			 * records.
			 */

			break;
		}

		if (match) {
			/* We found something, continue with next header */
			continue;
		}

		switch (header->typepair) {
		case dns_rdatatype_cname:
			if (!cname_ok) {
				break;
			}

			found = header;
			if (cnamesig != NULL) {
				/* We already have CNAME signature */
				foundsig = cnamesig;
			} else {
				/* Look for CNAME signature instead */
				sigpair = DNS_SIGTYPE(dns_rdatatype_cname);
				foundsig = NULL;
			}
			break;
		case DNS_SIGTYPE(dns_rdatatype_cname):
			if (!cname_ok) {
				break;
			}

			cnamesig = header;
			break;
		case dns_rdatatype_ns:
			/* Remember the NS rdataset */
			nsheader = header;
			break;
		case DNS_SIGTYPE(dns_rdatatype_ns):
			/* ...and its signature */
			nssig = header;
			break;

		case dns_rdatatype_nsec:
			nsecheader = header;
			break;
		case DNS_SIGTYPE(dns_rdatatype_nsec):
			nsecsig = header;
			break;

		default:
			if (typepair == dns_rdatatype_any &&
			    DNS_TYPEPAIR_TYPE(header->typepair) != 0)
			{
				/* QTYPE==ANY, so any anwers will do */
				found = header;
				break;
			}
		}
	}

	if (empty_node) {
		/*
		 * We have an exact match for the name, but there are no
		 * extant rdatasets.  That means that this node doesn't
		 * meaningfully exist, and that we really have a partial match.
		 */
		NODE_UNLOCK(nlock, &nlocktype);
		if ((search.options & DNS_DBFIND_COVERINGNSEC) != 0) {
			result = find_coveringnsec(
				&search, name, nodep, foundname, rdataset,
				sigrdataset DNS__DB_FLARG_PASS);
			if (result == DNS_R_COVERINGNSEC) {
				goto tree_exit;
			}
		}
		goto find_ns;
	}

	/*
	 * If we didn't find what we were looking for...
	 */
	if (MISSING_ANSWER(found, options)) {
		/*
		 * Return covering NODATA NSEC record.
		 */
		if ((search.options & DNS_DBFIND_COVERINGNSEC) != 0 &&
		    nsecheader != NULL)
		{
			if (nodep != NULL) {
				qpcnode_acquire(search.qpdb, node, nlocktype,
						tlocktype DNS__DB_FLARG_PASS);
				*nodep = (dns_dbnode_t *)node;
			}
			bindrdatasets(search.qpdb, node, nsecheader, nsecsig,
				      search.now, nlocktype, tlocktype,
				      rdataset, sigrdataset DNS__DB_FLARG_PASS);
			result = DNS_R_COVERINGNSEC;
			goto node_exit;
		}

		/*
		 * This name was from a wild card.  Look for a covering NSEC.
		 */
		if (found == NULL && (found_noqname || all_negative) &&
		    (search.options & DNS_DBFIND_COVERINGNSEC) != 0)
		{
			NODE_UNLOCK(nlock, &nlocktype);
			result = find_coveringnsec(
				&search, name, nodep, foundname, rdataset,
				sigrdataset DNS__DB_FLARG_PASS);
			if (result == DNS_R_COVERINGNSEC) {
				goto tree_exit;
			}
			goto find_ns;
		}

		/*
		 * If there is an NS rdataset at this node, then this is the
		 * deepest zone cut.
		 */
		if (nsheader != NULL) {
			if (nodep != NULL) {
				qpcnode_acquire(search.qpdb, node, nlocktype,
						tlocktype DNS__DB_FLARG_PASS);
				*nodep = (dns_dbnode_t *)node;
			}
			bindrdatasets(search.qpdb, node, nsheader, nssig,
				      search.now, nlocktype, tlocktype,
				      rdataset, sigrdataset DNS__DB_FLARG_PASS);
			result = DNS_R_DELEGATION;
			goto node_exit;
		}

		/*
		 * Go find the deepest zone cut.
		 */
		NODE_UNLOCK(nlock, &nlocktype);
		goto find_ns;
	}

	/*
	 * We found what we were looking for, or we found a CNAME.
	 */

	if (nodep != NULL) {
		qpcnode_acquire(search.qpdb, node, nlocktype,
				tlocktype DNS__DB_FLARG_PASS);
		*nodep = (dns_dbnode_t *)node;
	}

	if (NEGATIVE(found)) {
		/*
		 * We found a negative cache entry.
		 */
		if (NXDOMAIN(found)) {
			result = DNS_R_NCACHENXDOMAIN;
		} else {
			result = DNS_R_NCACHENXRRSET;
		}
	} else if (typepair != found->typepair &&
		   typepair != dns_rdatatype_any &&
		   found->typepair == dns_rdatatype_cname)
	{
		/*
		 * We weren't doing an ANY query and we found a CNAME instead
		 * of the type we were looking for, so we need to indicate
		 * that result to the caller.
		 */
		result = DNS_R_CNAME;
	} else {
		/*
		 * An ordinary successful query!
		 */
		result = ISC_R_SUCCESS;
	}

	if (typepair != dns_rdatatype_any || result == DNS_R_NCACHENXDOMAIN ||
	    result == DNS_R_NCACHENXRRSET)
	{
		bindrdatasets(search.qpdb, node, found, foundsig, search.now,
			      nlocktype, tlocktype, rdataset,
			      sigrdataset DNS__DB_FLARG_PASS);
	}

node_exit:
	NODE_UNLOCK(nlock, &nlocktype);

tree_exit:
	TREE_UNLOCK(&search.qpdb->tree_lock, &tlocktype);

	/*
	 * If we found a zonecut but aren't going to use it, we have to
	 * let go of it.
	 */
	if (search.need_cleanup) {
		node = search.zonecut;
		INSIST(node != NULL);
		nlock = &search.qpdb->buckets[node->locknum].lock;

		NODE_RDLOCK(nlock, &nlocktype);
		qpcnode_release(search.qpdb, node, &nlocktype,
				&tlocktype DNS__DB_FLARG_PASS);
		NODE_UNLOCK(nlock, &nlocktype);
		INSIST(tlocktype == isc_rwlocktype_none);
	}

	update_cachestats(search.qpdb, result);
	return result;
}

static isc_result_t
seek_ns_headers(qpc_search_t *search, qpcnode_t *node, dns_dbnode_t **nodep,
		dns_rdataset_t *rdataset, dns_rdataset_t *sigrdataset,
		dns_name_t *foundname, dns_name_t *dcname,
		isc_rwlocktype_t *tlocktype) {
	dns_slabheader_t *header = NULL;
	dns_slabheader_t *header_prev = NULL, *header_next = NULL;
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
	isc_rwlock_t *nlock = &search->qpdb->buckets[node->locknum].lock;
	dns_slabheader_t *found = NULL, *foundsig = NULL;

	NODE_RDLOCK(nlock, &nlocktype);

	for (header = node->data; header != NULL; header = header_next) {
		header_next = header->next;
		bool ns = (header->typepair == dns_rdatatype_ns ||
			   header->typepair == DNS_SIGTYPE(dns_rdatatype_ns));
		if (check_stale_header(header, search, &header_prev)) {
			if (ns) {
				/*
				 * We found a cached NS, but was either
				 * ancient or it was stale and serve-stale
				 * is disabled, so this node can't be used
				 * as a zone cut we know about. Instead we
				 * bail out and call find_deepest_zonecut()
				 * below.
				 */
				break;
			}
			continue;
		}

		if (both_headers(header, dns_rdatatype_ns, &found, &foundsig)) {
			break;
		}
	}

	if (found == NULL) {
		isc_result_t result;

		/*
		 * No active NS records found. Call find_deepest_zonecut()
		 * to look for them in nodes above this one.
		 */
		NODE_UNLOCK(nlock, &nlocktype);
		result = find_deepest_zonecut(search, node, nodep, foundname,
					      rdataset,
					      sigrdataset DNS__DB_FLARG_PASS);
		if (dcname != NULL) {
			dns_name_copy(foundname, dcname);
		}
		return result;
	}

	if (nodep != NULL) {
		qpcnode_acquire(search->qpdb, node, nlocktype,
				*tlocktype DNS__DB_FLARG_PASS);
		*nodep = (dns_dbnode_t *)node;
	}

	bindrdatasets(search->qpdb, node, found, foundsig, search->now,
		      nlocktype, *tlocktype, rdataset,
		      sigrdataset DNS__DB_FLARG_PASS);

	NODE_UNLOCK(nlock, &nlocktype);

	return ISC_R_SUCCESS;
}

static isc_result_t
qpcache_findzonecut(dns_db_t *db, const dns_name_t *name, unsigned int options,
		    isc_stdtime_t __now, dns_dbnode_t **nodep,
		    dns_name_t *foundname, dns_name_t *dcname,
		    dns_rdataset_t *rdataset,
		    dns_rdataset_t *sigrdataset DNS__DB_FLARG) {
	qpcnode_t *node = NULL;
	isc_result_t result;
	isc_rwlocktype_t tlocktype = isc_rwlocktype_none;
	qpc_search_t search = (qpc_search_t){
		.qpdb = (qpcache_t *)db,
		.options = options,
		.now = __now ? __now : isc_stdtime_now(),
	};
	unsigned int len = 0;

	REQUIRE(VALID_QPDB((qpcache_t *)db));

	TREE_RDLOCK(&search.qpdb->tree_lock, &tlocktype);

	/*
	 * Search down from the root of the tree.
	 */
	result = dns_qp_lookup(search.qpdb->tree, name, DNS_DBNAMESPACE_NORMAL,
			       NULL, NULL, &search.chain, (void **)&node, NULL);

	switch (result) {
	case ISC_R_SUCCESS:
		if ((options & DNS_DBFIND_NOEXACT) == 0) {
			if (dcname != NULL) {
				dns_name_copy(&node->name, dcname);
			}
			dns_name_copy(&node->name, foundname);
			result = seek_ns_headers(&search, node, nodep, rdataset,
						 sigrdataset, foundname, dcname,
						 &tlocktype);
			break;
		}

		len = dns_qpchain_length(&search.chain);
		if (len < 2) {
			result = ISC_R_NOTFOUND;
			break;
		}

		FALLTHROUGH;
	case DNS_R_PARTIALMATCH:
		if (dcname != NULL) {
			dns_name_copy(&node->name, dcname);
		}

		if (result == ISC_R_SUCCESS) {
			/* Fell through from the previous case */
			INSIST(len >= 2);

			node = NULL;
			dns_qpchain_node(&search.chain, len - 2, NULL,
					 (void **)&node, NULL);
			search.chain.len = len - 1;
		}

		result = find_deepest_zonecut(&search, node, nodep, foundname,
					      rdataset,
					      sigrdataset DNS__DB_FLARG_PASS);
		break;
	default:
		break;
	}

	TREE_UNLOCK(&search.qpdb->tree_lock, &tlocktype);

	INSIST(!search.need_cleanup);

	if (result == DNS_R_DELEGATION) {
		result = ISC_R_SUCCESS;
	}

	return result;
}

static isc_result_t
qpcache_findrdataset(dns_db_t *db, dns_dbnode_t *node, dns_dbversion_t *version,
		     dns_rdatatype_t type, dns_rdatatype_t covers,
		     isc_stdtime_t __now, dns_rdataset_t *rdataset,
		     dns_rdataset_t *sigrdataset DNS__DB_FLARG) {
	qpcache_t *qpdb = (qpcache_t *)db;
	qpcnode_t *qpnode = (qpcnode_t *)node;
	dns_slabheader_t *header = NULL;
	dns_slabheader_t *header_prev = NULL, *header_next = NULL;
	dns_slabheader_t *found = NULL, *foundsig = NULL;
	dns_typepair_t typepair, sigpair, negpair;
	isc_result_t result = ISC_R_SUCCESS;
	isc_rwlock_t *nlock = NULL;
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
	qpc_search_t search = (qpc_search_t){
		.qpdb = (qpcache_t *)db,
		.now = __now ? __now : isc_stdtime_now(),
	};

	REQUIRE(VALID_QPDB(qpdb));
	REQUIRE(version == NULL);
	REQUIRE(type != dns_rdatatype_any);

	if (type == dns_rdatatype_none && covers == dns_rdatatype_none) {
		return ISC_R_NOTFOUND;
	}

	nlock = &qpdb->buckets[qpnode->locknum].lock;
	NODE_RDLOCK(nlock, &nlocktype);

	typepair = DNS_TYPEPAIR_VALUE(type, covers);
	negpair = DNS_TYPEPAIR_VALUE(0, type);
	sigpair = (covers == 0) ? DNS_SIGTYPE(type) : 0;

	for (header = qpnode->data; header != NULL; header = header_next) {
		header_next = header->next;

		if (check_stale_header(header, &search, &header_prev)) {
			continue;
		}

		if (related_headers(header, typepair, sigpair, negpair, &found,
				    &foundsig, NULL))
		{
			break;
		}
	}
	if (found != NULL) {
		bindrdatasets(qpdb, qpnode, found, foundsig, search.now,
			      nlocktype, isc_rwlocktype_none, rdataset,
			      sigrdataset DNS__DB_FLARG_PASS);
	}

	NODE_UNLOCK(nlock, &nlocktype);

	if (found == NULL) {
		return ISC_R_NOTFOUND;
	}

	if (NEGATIVE(found)) {
		/*
		 * We found a negative cache entry.
		 */
		if (NXDOMAIN(found)) {
			result = DNS_R_NCACHENXDOMAIN;
		} else {
			result = DNS_R_NCACHENXRRSET;
		}
	}

	update_cachestats(qpdb, result);

	return result;
}

static isc_result_t
setcachestats(dns_db_t *db, isc_stats_t *stats) {
	qpcache_t *qpdb = (qpcache_t *)db;

	REQUIRE(VALID_QPDB(qpdb));
	REQUIRE(stats != NULL);

	isc_stats_attach(stats, &qpdb->cachestats);
	return ISC_R_SUCCESS;
}

static dns_stats_t *
getrrsetstats(dns_db_t *db) {
	qpcache_t *qpdb = (qpcache_t *)db;

	REQUIRE(VALID_QPDB(qpdb));

	return qpdb->rrsetstats;
}

static isc_result_t
setservestalettl(dns_db_t *db, dns_ttl_t ttl) {
	qpcache_t *qpdb = (qpcache_t *)db;

	REQUIRE(VALID_QPDB(qpdb));

	/* currently no bounds checking.  0 means disable. */
	qpdb->common.serve_stale_ttl = ttl;
	return ISC_R_SUCCESS;
}

static isc_result_t
getservestalettl(dns_db_t *db, dns_ttl_t *ttl) {
	qpcache_t *qpdb = (qpcache_t *)db;

	REQUIRE(VALID_QPDB(qpdb));

	*ttl = qpdb->common.serve_stale_ttl;
	return ISC_R_SUCCESS;
}

static isc_result_t
setservestalerefresh(dns_db_t *db, uint32_t interval) {
	qpcache_t *qpdb = (qpcache_t *)db;

	REQUIRE(VALID_QPDB(qpdb));

	/* currently no bounds checking.  0 means disable. */
	qpdb->serve_stale_refresh = interval;
	return ISC_R_SUCCESS;
}

static isc_result_t
getservestalerefresh(dns_db_t *db, uint32_t *interval) {
	qpcache_t *qpdb = (qpcache_t *)db;

	REQUIRE(VALID_QPDB(qpdb));

	*interval = qpdb->serve_stale_refresh;
	return ISC_R_SUCCESS;
}

static void
qpcnode_expiredata(dns_dbnode_t *node, void *data) {
	qpcnode_t *qpnode = (qpcnode_t *)node;
	qpcache_t *qpdb = (qpcache_t *)qpnode->qpdb;

	dns_slabheader_t *header = data;
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
	isc_rwlocktype_t tlocktype = isc_rwlocktype_none;

	isc_rwlock_t *nlock = &qpdb->buckets[qpnode->locknum].lock;
	NODE_WRLOCK(nlock, &nlocktype);
	expireheader(header, &nlocktype, &tlocktype,
		     dns_expire_flush DNS__DB_FILELINE);
	NODE_UNLOCK(nlock, &nlocktype);
	INSIST(tlocktype == isc_rwlocktype_none);
}

/*%
 * These functions allow the heap code to rank the priority of each
 * element.  It returns true if v1 happens "sooner" than v2.
 */
static bool
ttl_sooner(void *v1, void *v2) {
	dns_slabheader_t *h1 = v1;
	dns_slabheader_t *h2 = v2;

	return h1->expire < h2->expire;
}

/*%
 * This function sets the heap index into the header.
 */
static void
set_index(void *what, unsigned int idx) {
	dns_slabheader_t *h = what;

	h->heap_index = idx;
}

static void
qpcache__destroy(qpcache_t *qpdb) {
	unsigned int i;
	char buf[DNS_NAME_FORMATSIZE];
	dns_qp_t **treep = NULL;

	for (;;) {
		/*
		 * pick the next tree to (start to) destroy
		 */
		treep = &qpdb->tree;
		if (*treep == NULL) {
			treep = &qpdb->nsec;
			if (*treep == NULL) {
				break;
			}
		}

		dns_qp_destroy(treep);
		INSIST(*treep == NULL);
	}

	if (dns_name_dynamic(&qpdb->common.origin)) {
		dns_name_format(&qpdb->common.origin, buf, sizeof(buf));
	} else {
		strlcpy(buf, "<UNKNOWN>", sizeof(buf));
	}
	isc_log_write(DNS_LOGCATEGORY_DATABASE, DNS_LOGMODULE_CACHE,
		      ISC_LOG_DEBUG(DNS_QPCACHE_LOG_STATS_LEVEL), "done %s(%s)",
		      __func__, buf);

	if (dns_name_dynamic(&qpdb->common.origin)) {
		dns_name_free(&qpdb->common.origin, qpdb->common.mctx);
	}
	for (i = 0; i < qpdb->buckets_count; i++) {
		NODE_DESTROYLOCK(&qpdb->buckets[i].lock);

		INSIST(ISC_SIEVE_EMPTY(qpdb->buckets[i].sieve));

		INSIST(isc_queue_empty(&qpdb->buckets[i].deadnodes));
		isc_queue_destroy(&qpdb->buckets[i].deadnodes);

		isc_heap_destroy(&qpdb->buckets[i].heap);
	}

	if (qpdb->rrsetstats != NULL) {
		dns_stats_detach(&qpdb->rrsetstats);
	}
	if (qpdb->cachestats != NULL) {
		isc_stats_detach(&qpdb->cachestats);
	}

	TREE_DESTROYLOCK(&qpdb->tree_lock);
	isc_refcount_destroy(&qpdb->references);
	isc_refcount_destroy(&qpdb->common.references);

	isc_rwlock_destroy(&qpdb->lock);
	qpdb->common.magic = 0;
	qpdb->common.impmagic = 0;
	isc_mem_detach(&qpdb->hmctx);

	isc_mem_putanddetach(&qpdb->common.mctx, qpdb,
			     sizeof(*qpdb) + qpdb->buckets_count *
						     sizeof(qpdb->buckets[0]));
}

static void
qpcache_destroy(dns_db_t *arg) {
	qpcache_t *qpdb = (qpcache_t *)arg;

	qpcache_detach(&qpdb);
}

/*%
 * Clean up dead nodes.  These are nodes which have no references, and
 * have no data.  They are dead but we could not or chose not to delete
 * them when we deleted all the data at that node because we did not want
 * to wait for the tree write lock.
 */
static void
cleanup_deadnodes(qpcache_t *qpdb, uint16_t locknum) {
	isc_rwlocktype_t tlocktype = isc_rwlocktype_none;
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
	isc_rwlock_t *nlock = &qpdb->buckets[locknum].lock;
	qpcnode_t *qpnode = NULL, *qpnext = NULL;
	isc_queue_t deadnodes;

	INSIST(locknum < qpdb->buckets_count);

	isc_queue_init(&deadnodes);

	TREE_WRLOCK(&qpdb->tree_lock, &tlocktype);
	NODE_WRLOCK(nlock, &nlocktype);

	isc_queue_splice(&deadnodes, &qpdb->buckets[locknum].deadnodes);
	isc_queue_for_each_entry_safe(&deadnodes, qpnode, qpnext, deadlink) {
		qpcnode_release(qpdb, qpnode, &nlocktype,
				&tlocktype DNS__DB_FILELINE);
	}

	NODE_UNLOCK(nlock, &nlocktype);
	TREE_UNLOCK(&qpdb->tree_lock, &tlocktype);
}

static void
cleanup_deadnodes_cb(void *arg) {
	qpcache_t *qpdb = arg;
	uint16_t locknum = isc_tid();

	cleanup_deadnodes(qpdb, locknum);
	qpcache_unref(qpdb);
}
/*
 * This function is assumed to be called when a node is newly referenced
 * and can be in the deadnode list.  In that case the node will be references
 * and cleanup_deadnodes() will remove it from the list when the cleaning
 * happens.
 * Note: while a new reference is gained in multiple places, there are only very
 * few cases where the node can be in the deadnode list (only empty nodes can
 * have been added to the list).
 */
static void
reactivate_node(qpcache_t *qpdb, qpcnode_t *node,
		isc_rwlocktype_t tlocktype ISC_ATTR_UNUSED DNS__DB_FLARG) {
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
	isc_rwlock_t *nlock = &qpdb->buckets[node->locknum].lock;

	NODE_RDLOCK(nlock, &nlocktype);
	qpcnode_acquire(qpdb, node, nlocktype, tlocktype DNS__DB_FLARG_PASS);
	NODE_UNLOCK(nlock, &nlocktype);
}

static qpcnode_t *
new_qpcnode(qpcache_t *qpdb, const dns_name_t *name, dns_namespace_t nspace) {
	qpcnode_t *newdata = isc_mem_get(qpdb->common.mctx, sizeof(*newdata));
	*newdata = (qpcnode_t){
		.methods = &qpcnode_methods,
		.qpdb = qpdb,
		.name = DNS_NAME_INITEMPTY,
		.nspace = nspace,
		.references = ISC_REFCOUNT_INITIALIZER(1),
		.locknum = isc_random_uniform(qpdb->buckets_count),
	};

	isc_mem_attach(qpdb->common.mctx, &newdata->mctx);
	dns_name_dup(name, newdata->mctx, &newdata->name);

#ifdef DNS_DB_NODETRACE
	fprintf(stderr, "new_qpcnode:%s:%s:%d:%p->references = 1\n", __func__,
		__FILE__, __LINE__ + 1, name);
#endif
	return newdata;
}

static isc_result_t
qpcache_findnode(dns_db_t *db, const dns_name_t *name, bool create,
		 dns_dbnode_t **nodep DNS__DB_FLARG) {
	qpcache_t *qpdb = (qpcache_t *)db;
	qpcnode_t *node = NULL;
	isc_result_t result;
	isc_rwlocktype_t tlocktype = isc_rwlocktype_none;
	dns_namespace_t nspace = DNS_DBNAMESPACE_NORMAL;

	TREE_RDLOCK(&qpdb->tree_lock, &tlocktype);
	result = dns_qp_getname(qpdb->tree, name, nspace, (void **)&node, NULL);
	if (result != ISC_R_SUCCESS) {
		if (!create) {
			goto unlock;
		}
		/*
		 * Try to upgrade the lock and if that fails unlock then relock.
		 */
		TREE_FORCEUPGRADE(&qpdb->tree_lock, &tlocktype);
		result = dns_qp_getname(qpdb->tree, name, nspace,
					(void **)&node, NULL);
		if (result != ISC_R_SUCCESS) {
			node = new_qpcnode(qpdb, name, nspace);
			result = dns_qp_insert(qpdb->tree, node, 0);
			INSIST(result == ISC_R_SUCCESS);
			qpcnode_unref(node);
		}
	}

	reactivate_node(qpdb, node, tlocktype DNS__DB_FLARG_PASS);

	*nodep = (dns_dbnode_t *)node;
unlock:
	TREE_UNLOCK(&qpdb->tree_lock, &tlocktype);

	return result;
}

static isc_result_t
qpcache_createiterator(dns_db_t *db, unsigned int options ISC_ATTR_UNUSED,
		       dns_dbiterator_t **iteratorp) {
	qpcache_t *qpdb = (qpcache_t *)db;
	qpc_dbit_t *qpdbiter = NULL;

	REQUIRE(VALID_QPDB(qpdb));

	qpdbiter = isc_mem_get(qpdb->common.mctx, sizeof(*qpdbiter));
	*qpdbiter = (qpc_dbit_t){
		.common.methods = &dbiterator_methods,
		.common.magic = DNS_DBITERATOR_MAGIC,
		.paused = true,
	};

	qpdbiter->name = dns_fixedname_initname(&qpdbiter->fixed);
	dns_db_attach(db, &qpdbiter->common.db);
	dns_qpiter_init(qpdb->tree, &qpdbiter->iter);

	*iteratorp = (dns_dbiterator_t *)qpdbiter;
	return ISC_R_SUCCESS;
}

static isc_result_t
qpcache_allrdatasets(dns_db_t *db, dns_dbnode_t *node, dns_dbversion_t *version,
		     unsigned int options, isc_stdtime_t __now,
		     dns_rdatasetiter_t **iteratorp DNS__DB_FLARG) {
	qpcache_t *qpdb = (qpcache_t *)db;
	qpcnode_t *qpnode = (qpcnode_t *)node;
	qpc_rditer_t *iterator = NULL;

	REQUIRE(VALID_QPDB(qpdb));
	REQUIRE(version == NULL);

	iterator = isc_mem_get(qpdb->common.mctx, sizeof(*iterator));
	*iterator = (qpc_rditer_t){
		.common.magic = DNS_RDATASETITER_MAGIC,
		.common.methods = &rdatasetiter_methods,
		.common.db = db,
		.common.node = node,
		.common.options = options,
		.common.now = __now ? __now : isc_stdtime_now(),
	};

	qpcnode_acquire(qpdb, qpnode, isc_rwlocktype_none,
			isc_rwlocktype_none DNS__DB_FLARG_PASS);

	*iteratorp = (dns_rdatasetiter_t *)iterator;

	return ISC_R_SUCCESS;
}

static bool
overmaxtype(qpcache_t *qpdb, uint32_t ntypes) {
	if (qpdb->maxtypepername == 0) {
		return false;
	}

	return ntypes >= qpdb->maxtypepername;
}

static bool
prio_header(dns_slabheader_t *header) {
	if (NEGATIVE(header) &&
	    prio_type(DNS_TYPEPAIR_COVERS(header->typepair)))
	{
		return true;
	}

	return prio_type(header->typepair);
}

static void
qpcnode_attachnode(dns_dbnode_t *source, dns_dbnode_t **targetp DNS__DB_FLARG) {
	REQUIRE(targetp != NULL && *targetp == NULL);

	qpcnode_t *node = (qpcnode_t *)source;
	qpcache_t *qpdb = (qpcache_t *)node->qpdb;

	qpcnode_acquire(qpdb, node, isc_rwlocktype_none,
			isc_rwlocktype_none DNS__DB_FLARG_PASS);

	*targetp = source;
}

static void
qpcnode_detachnode(dns_dbnode_t **nodep DNS__DB_FLARG) {
	qpcnode_t *node = NULL;
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
	isc_rwlocktype_t tlocktype = isc_rwlocktype_none;
	isc_rwlock_t *nlock = NULL;

	REQUIRE(nodep != NULL && *nodep != NULL);

	node = (qpcnode_t *)(*nodep);
	qpcache_t *qpdb = (qpcache_t *)node->qpdb;
	*nodep = NULL;
	nlock = &qpdb->buckets[node->locknum].lock;

	REQUIRE(VALID_QPDB(qpdb));

	/*
	 * We can't destroy qpcache while holding a nodelock, so we need to
	 * reference it before acquiring the lock and release it afterward.
	 * Additionally, we must ensure that we don't destroy the database while
	 * the NODE_LOCK is locked.
	 */
	qpcache_ref(qpdb);

	rcu_read_lock();
	NODE_RDLOCK(nlock, &nlocktype);
	qpcnode_release(qpdb, node, &nlocktype, &tlocktype DNS__DB_FLARG_PASS);
	NODE_UNLOCK(nlock, &nlocktype);
	rcu_read_unlock();

	qpcache_detach(&qpdb);
}

static isc_result_t
add(qpcache_t *qpdb, qpcnode_t *qpnode,
    const dns_name_t *nodename ISC_ATTR_UNUSED, dns_slabheader_t *newheader,
    unsigned int options, dns_rdataset_t *addedrdataset, isc_stdtime_t now,
    isc_rwlocktype_t nlocktype, isc_rwlocktype_t tlocktype DNS__DB_FLARG) {
	dns_slabheader_t *topheader = NULL, *topheader_prev = NULL;
	dns_slabheader_t *header = NULL, *sigheader = NULL;
	dns_slabheader_t *prioheader = NULL, *expireheader = NULL;
	dns_typepair_t negpair = 0;
	dns_trust_t trust;
	uint32_t ntypes = 0;

	if ((options & DNS_DBADD_FORCE) != 0) {
		trust = dns_trust_ultimate;
	} else {
		trust = newheader->trust;
	}

	if (EXISTS(newheader)) {
		dns_rdatatype_t rdtype = DNS_TYPEPAIR_TYPE(newheader->typepair);
		dns_rdatatype_t covers =
			DNS_TYPEPAIR_COVERS(newheader->typepair);
		dns_typepair_t sigpair = DNS_SIGTYPE(covers);
		if (NEGATIVE(newheader)) {
			/*
			 * We're adding a negative cache entry.
			 */
			if (covers == dns_rdatatype_any) {
				/*
				 * If we're adding an negative cache entry
				 * which covers all types (NXDOMAIN,
				 * NODATA(QTYPE=ANY)),
				 *
				 * We make all other data ancient so that the
				 * only rdataset that can be found at this
				 * node is the negative cache entry.
				 */
				for (topheader = qpnode->data;
				     topheader != NULL;
				     topheader = topheader->next)
				{
					mark_ancient(topheader);
				}
				goto find_header;
			}
			/*
			 * Otherwise look for any RRSIGs of the given
			 * type so they can be marked ancient later.
			 */
			for (topheader = qpnode->data; topheader != NULL;
			     topheader = topheader->next)
			{
				if (topheader->typepair == sigpair) {
					sigheader = topheader;
					break;
				}
			}
			negpair = DNS_TYPEPAIR_VALUE(covers, 0);
		} else {
			/*
			 * We're adding something that isn't a
			 * negative cache entry.  Look for an extant
			 * non-ancient NXDOMAIN/NODATA(QTYPE=ANY) negative
			 * cache entry.  If we're adding an RRSIG, also
			 * check for an extant non-ancient NODATA ncache
			 * entry which covers the same type as the RRSIG.
			 */
			for (topheader = qpnode->data; topheader != NULL;
			     topheader = topheader->next)
			{
				if ((topheader->typepair ==
				     RDATATYPE_NCACHEANY) ||
				    (newheader->typepair == sigpair &&
				     topheader->typepair ==
					     DNS_TYPEPAIR_VALUE(0, covers)))
				{
					break;
				}
			}
			if (topheader != NULL && EXISTS(topheader) &&
			    ACTIVE(topheader, now))
			{
				/*
				 * Found one.
				 */
				if (trust < topheader->trust) {
					/*
					 * The NXDOMAIN/NODATA(QTYPE=ANY)
					 * is more trusted.
					 */
					dns_slabheader_destroy(&newheader);
					if (addedrdataset != NULL) {
						bindrdataset(
							qpdb, qpnode, topheader,
							now, nlocktype,
							tlocktype,
							addedrdataset
								DNS__DB_FLARG_PASS);
					}
					return DNS_R_UNCHANGED;
				}
				/*
				 * The new rdataset is better.  Expire the
				 * ncache entry.
				 */
				mark_ancient(topheader);
				topheader = NULL;
				goto find_header;
			}
			negpair = DNS_TYPEPAIR_VALUE(0, rdtype);
		}
	}

	for (topheader = qpnode->data; topheader != NULL;
	     topheader = topheader->next)
	{
		if (ACTIVE(topheader, now)) {
			++ntypes;
			expireheader = topheader;
		}
		if (prio_header(topheader)) {
			prioheader = topheader;
		}

		if (topheader->typepair == newheader->typepair ||
		    topheader->typepair == negpair)
		{
			break;
		}
		topheader_prev = topheader;
	}

find_header:
	/*
	 * If header isn't NULL, we've found the right type.
	 */
	header = topheader;
	if (header != NULL) {
		/*
		 * Deleting an already non-existent rdataset has no effect.
		 */
		if (!EXISTS(header) && !EXISTS(newheader)) {
			dns_slabheader_destroy(&newheader);
			return DNS_R_UNCHANGED;
		}

		/*
		 * Trying to add an rdataset with lower trust to a cache
		 * DB has no effect, provided that the cache data isn't
		 * stale. If the cache data is stale, new lower trust
		 * data will supersede it below. Unclear what the best
		 * policy is here.
		 */
		if (trust < header->trust &&
		    (ACTIVE(header, now) || !EXISTS(header)))
		{
			dns_slabheader_destroy(&newheader);
			if (addedrdataset != NULL) {
				bindrdataset(qpdb, qpnode, header, now,
					     nlocktype, tlocktype,
					     addedrdataset DNS__DB_FLARG_PASS);
			}
			return DNS_R_UNCHANGED;
		}

		/*
		 * Don't replace existing NS, A and AAAA RRsets in the
		 * cache if they already exist. This prevents named
		 * being locked to old servers. Don't lower trust of
		 * existing record if the update is forced. Nothing
		 * special to be done w.r.t stale data; it gets replaced
		 * normally further down.
		 */
		if (ACTIVE(header, now) &&
		    header->typepair == dns_rdatatype_ns && EXISTS(header) &&
		    EXISTS(newheader) && header->trust >= newheader->trust &&
		    dns_rdataslab_equalx(header, newheader,
					 qpdb->common.rdclass,
					 DNS_TYPEPAIR_TYPE(header->typepair)))
		{
			/*
			 * Honour the new ttl if it is less than the
			 * older one.
			 */
			if (header->expire > newheader->expire) {
				setttl(header, newheader->expire);
			}

			qpcache_hit(qpdb, header);

			if (header->noqname == NULL &&
			    newheader->noqname != NULL)
			{
				header->noqname = newheader->noqname;
				newheader->noqname = NULL;
			}
			if (header->closest == NULL &&
			    newheader->closest != NULL)
			{
				header->closest = newheader->closest;
				newheader->closest = NULL;
			}
			dns_slabheader_destroy(&newheader);
			if (addedrdataset != NULL) {
				bindrdataset(qpdb, qpnode, header, now,
					     nlocktype, tlocktype,
					     addedrdataset DNS__DB_FLARG_PASS);
			}
			return ISC_R_SUCCESS;
		}

		/*
		 * If we will be replacing a NS RRset force its TTL
		 * to be no more than the current NS RRset's TTL.  This
		 * ensures the delegations that are withdrawn are honoured.
		 */
		if (ACTIVE(header, now) &&
		    header->typepair == dns_rdatatype_ns && EXISTS(header) &&
		    EXISTS(newheader) && header->trust <= newheader->trust)
		{
			if (newheader->expire > header->expire) {
				newheader->expire = header->expire;
			}
		}
		if (ACTIVE(header, now) &&
		    (options & DNS_DBADD_PREFETCH) == 0 &&
		    (header->typepair == dns_rdatatype_a ||
		     header->typepair == dns_rdatatype_aaaa ||
		     header->typepair == dns_rdatatype_ds ||
		     header->typepair == DNS_SIGTYPE(dns_rdatatype_ds)) &&
		    EXISTS(header) && EXISTS(newheader) &&
		    header->trust >= newheader->trust &&
		    dns_rdataslab_equal(header, newheader))
		{
			/*
			 * Honour the new ttl if it is less than the
			 * older one.
			 */
			if (header->expire > newheader->expire) {
				setttl(header, newheader->expire);
			}

			qpcache_hit(qpdb, header);

			if (header->noqname == NULL &&
			    newheader->noqname != NULL)
			{
				header->noqname = newheader->noqname;
				newheader->noqname = NULL;
			}
			if (header->closest == NULL &&
			    newheader->closest != NULL)
			{
				header->closest = newheader->closest;
				newheader->closest = NULL;
			}
			dns_slabheader_destroy(&newheader);
			if (addedrdataset != NULL) {
				bindrdataset(qpdb, qpnode, header, now,
					     nlocktype, tlocktype,
					     addedrdataset DNS__DB_FLARG_PASS);
			}
			return ISC_R_SUCCESS;
		}

		qpcache_miss(qpdb, newheader, &nlocktype,
			     &tlocktype DNS__DB_FLARG_PASS);

		if (topheader_prev != NULL) {
			topheader_prev->next = newheader;
		} else {
			qpnode->data = newheader;
		}
		newheader->next = topheader->next;
		newheader->down = topheader;
		topheader->next = newheader;
		mark_ancient(header);
		if (sigheader != NULL) {
			mark_ancient(sigheader);
		}
	} else if (!EXISTS(newheader)) {
		/*
		 * The type already doesn't exist; no point trying
		 * to delete it.
		 */
		dns_slabheader_destroy(&newheader);
		return DNS_R_UNCHANGED;
	} else {
		/* No rdatasets of the given type exist at the node. */
		INSIST(newheader->down == NULL);

		qpcache_miss(qpdb, newheader, &nlocktype,
			     &tlocktype DNS__DB_FLARG_PASS);
		if (prio_header(newheader)) {
			/* This is a priority type, prepend it */
			newheader->next = qpnode->data;
			qpnode->data = newheader;
		} else if (prioheader != NULL) {
			/* Append after the priority headers */
			newheader->next = prioheader->next;
			prioheader->next = newheader;
		} else {
			/* There were no priority headers */
			newheader->next = qpnode->data;
			qpnode->data = newheader;
		}

		if (overmaxtype(qpdb, ntypes)) {
			if (expireheader == NULL) {
				expireheader = newheader;
			}
			if (NEGATIVE(newheader) && !prio_header(newheader)) {
				/*
				 * Add the new non-priority negative
				 * header to the database only
				 * temporarily.
				 */
				expireheader = newheader;
			}

			mark_ancient(expireheader);
			/*
			 * FIXME: In theory, we should mark the RRSIG
			 * and the header at the same time, but there is
			 * no direct link between those two headers, so
			 * we would have to check the whole list again.
			 */
		}
	}

	if (addedrdataset != NULL) {
		bindrdataset(qpdb, qpnode, newheader, now, nlocktype, tlocktype,
			     addedrdataset DNS__DB_FLARG_PASS);
	}

	return ISC_R_SUCCESS;
}

static isc_result_t
addnoqname(isc_mem_t *mctx, dns_slabheader_t *newheader, uint32_t maxrrperset,
	   dns_rdataset_t *rdataset) {
	isc_result_t result;
	dns_slabheader_proof_t *noqname = NULL;
	dns_name_t name = DNS_NAME_INITEMPTY;
	dns_rdataset_t neg = DNS_RDATASET_INIT, negsig = DNS_RDATASET_INIT;
	isc_region_t r1, r2;

	result = dns_rdataset_getnoqname(rdataset, &name, &neg, &negsig);
	RUNTIME_CHECK(result == ISC_R_SUCCESS);

	result = dns_rdataslab_fromrdataset(&neg, mctx, &r1, maxrrperset);
	if (result != ISC_R_SUCCESS) {
		goto cleanup;
	}

	result = dns_rdataslab_fromrdataset(&negsig, mctx, &r2, maxrrperset);
	if (result != ISC_R_SUCCESS) {
		goto cleanup;
	}

	noqname = isc_mem_get(mctx, sizeof(*noqname));
	*noqname = (dns_slabheader_proof_t){
		.neg = dns_slabheader_raw((dns_slabheader_t *)r1.base),
		.negsig = dns_slabheader_raw((dns_slabheader_t *)r2.base),
		.type = neg.type,
		.name = DNS_NAME_INITEMPTY,
	};
	dns_name_dup(&name, mctx, &noqname->name);
	newheader->noqname = noqname;

cleanup:
	dns_rdataset_disassociate(&neg);
	dns_rdataset_disassociate(&negsig);

	return result;
}

static isc_result_t
addclosest(isc_mem_t *mctx, dns_slabheader_t *newheader, uint32_t maxrrperset,
	   dns_rdataset_t *rdataset) {
	isc_result_t result;
	dns_slabheader_proof_t *closest = NULL;
	dns_name_t name = DNS_NAME_INITEMPTY;
	dns_rdataset_t neg = DNS_RDATASET_INIT, negsig = DNS_RDATASET_INIT;
	isc_region_t r1, r2;

	result = dns_rdataset_getclosest(rdataset, &name, &neg, &negsig);
	RUNTIME_CHECK(result == ISC_R_SUCCESS);

	result = dns_rdataslab_fromrdataset(&neg, mctx, &r1, maxrrperset);
	if (result != ISC_R_SUCCESS) {
		goto cleanup;
	}

	result = dns_rdataslab_fromrdataset(&negsig, mctx, &r2, maxrrperset);
	if (result != ISC_R_SUCCESS) {
		goto cleanup;
	}

	closest = isc_mem_get(mctx, sizeof(*closest));
	*closest = (dns_slabheader_proof_t){
		.neg = dns_slabheader_raw((dns_slabheader_t *)r1.base),
		.negsig = dns_slabheader_raw((dns_slabheader_t *)r2.base),
		.name = DNS_NAME_INITEMPTY,
		.type = neg.type,
	};
	dns_name_dup(&name, mctx, &closest->name);
	newheader->closest = closest;

cleanup:
	dns_rdataset_disassociate(&neg);
	dns_rdataset_disassociate(&negsig);
	return result;
}

static void
expire_ttl_headers(qpcache_t *qpdb, unsigned int locknum,
		   isc_rwlocktype_t *nlocktypep, isc_rwlocktype_t *tlocktypep,
		   isc_stdtime_t now DNS__DB_FLARG);

static isc_result_t
qpcache_addrdataset(dns_db_t *db, dns_dbnode_t *node, dns_dbversion_t *version,
		    isc_stdtime_t __now, dns_rdataset_t *rdataset,
		    unsigned int options,
		    dns_rdataset_t *addedrdataset DNS__DB_FLARG) {
	qpcache_t *qpdb = (qpcache_t *)db;
	qpcnode_t *qpnode = (qpcnode_t *)node;
	isc_region_t region;
	dns_slabheader_t *newheader = NULL;
	isc_result_t result;
	bool delegating = false;
	bool newnsec = false;
	isc_rwlocktype_t tlocktype = isc_rwlocktype_none;
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
	isc_rwlock_t *nlock = NULL;
	dns_fixedname_t fixed;
	dns_name_t *name = NULL;
	isc_stdtime_t now = __now ? __now : isc_stdtime_now();

	REQUIRE(VALID_QPDB(qpdb));
	REQUIRE(version == NULL);

	result = dns_rdataslab_fromrdataset(rdataset, qpnode->mctx, &region,
					    qpdb->maxrrperset);
	if (result != ISC_R_SUCCESS) {
		if (result == DNS_R_TOOMANYRECORDS) {
			dns__db_logtoomanyrecords((dns_db_t *)qpdb,
						  &qpnode->name, rdataset->type,
						  "adding", qpdb->maxrrperset);
		}
		return result;
	}

	name = dns_fixedname_initname(&fixed);
	dns_name_copy(&qpnode->name, name);
	dns_rdataset_getownercase(rdataset, name);

	newheader = (dns_slabheader_t *)region.base;
	dns_slabheader_reset(newheader, node);

	/*
	 * By default, dns_rdataslab_fromrdataset() sets newheader->ttl
	 * to the rdataset TTL. In the case of the cache, that's wrong;
	 * we need it to be set to the expire time instead.
	 */
	setttl(newheader, rdataset->ttl + now);
	if (rdataset->ttl == 0U) {
		DNS_SLABHEADER_SETATTR(newheader, DNS_SLABHEADERATTR_ZEROTTL);
	}

	atomic_init(&newheader->count,
		    atomic_fetch_add_relaxed(&init_count, 1));
	if (rdataset->attributes.prefetch) {
		DNS_SLABHEADER_SETATTR(newheader, DNS_SLABHEADERATTR_PREFETCH);
	}
	if (rdataset->attributes.negative) {
		DNS_SLABHEADER_SETATTR(newheader, DNS_SLABHEADERATTR_NEGATIVE);
	}
	if (rdataset->attributes.nxdomain) {
		DNS_SLABHEADER_SETATTR(newheader, DNS_SLABHEADERATTR_NXDOMAIN);
	}
	if (rdataset->attributes.optout) {
		DNS_SLABHEADER_SETATTR(newheader, DNS_SLABHEADERATTR_OPTOUT);
	}
	if (rdataset->attributes.noqname) {
		result = addnoqname(qpnode->mctx, newheader, qpdb->maxrrperset,
				    rdataset);
		if (result != ISC_R_SUCCESS) {
			dns_slabheader_destroy(&newheader);
			return result;
		}
	}
	if (rdataset->attributes.closest) {
		result = addclosest(qpnode->mctx, newheader, qpdb->maxrrperset,
				    rdataset);
		if (result != ISC_R_SUCCESS) {
			dns_slabheader_destroy(&newheader);
			return result;
		}
	}

	nlock = &qpdb->buckets[qpnode->locknum].lock;

	/*
	 * If we're adding a delegation type (which would be an NS or DNAME
	 * for a zone, but only DNAME counts for a cache), we need to set
	 * the callback bit on the node.
	 */
	if (rdataset->type == dns_rdatatype_dname) {
		delegating = true;
	}

	/*
	 * Add to the auxiliary NSEC tree if we're adding an NSEC record.
	 */
	if (rdataset->type == dns_rdatatype_nsec) {
		NODE_RDLOCK(nlock, &nlocktype);
		if (!qpnode->havensec) {
			newnsec = true;
		}
		NODE_UNLOCK(nlock, &nlocktype);
	}

	/*
	 * If we're adding a delegation type or adding to the auxiliary
	 * NSEC tree, hold an exclusive lock on the tree.
	 */
	if (delegating || newnsec) {
		TREE_WRLOCK(&qpdb->tree_lock, &tlocktype);
	}

	NODE_WRLOCK(nlock, &nlocktype);

	if (qpdb->rrsetstats != NULL) {
		DNS_SLABHEADER_SETATTR(newheader, DNS_SLABHEADERATTR_STATCOUNT);
		update_rrsetstats(qpdb->rrsetstats, newheader->typepair,
				  atomic_load_acquire(&newheader->attributes),
				  true);
	}

	expire_ttl_headers(qpdb, qpnode->locknum, &nlocktype, &tlocktype,
			   now DNS__DB_FLARG_PASS);

	if (newnsec && !qpnode->havensec) {
		qpcnode_t *nsecnode = NULL;

		result = dns_qp_getname(qpdb->nsec, name, DNS_DBNAMESPACE_NSEC,
					(void **)&nsecnode, NULL);
		if (result != ISC_R_SUCCESS) {
			INSIST(nsecnode == NULL);
			nsecnode = new_qpcnode(qpdb, name,
					       DNS_DBNAMESPACE_NSEC);
			result = dns_qp_insert(qpdb->nsec, nsecnode, 0);
			INSIST(result == ISC_R_SUCCESS);
			qpcnode_detach(&nsecnode);
		}
		qpnode->havensec = true;
	}

	result = add(qpdb, qpnode, name, newheader, options, addedrdataset, now,
		     nlocktype, tlocktype DNS__DB_FLARG_PASS);

	if (result == ISC_R_SUCCESS && delegating) {
		qpnode->delegating = 1;
	}

	NODE_UNLOCK(nlock, &nlocktype);

	if (tlocktype != isc_rwlocktype_none) {
		TREE_UNLOCK(&qpdb->tree_lock, &tlocktype);
	}

	INSIST(tlocktype == isc_rwlocktype_none);

	return result;
}

static isc_result_t
qpcache_deleterdataset(dns_db_t *db, dns_dbnode_t *node,
		       dns_dbversion_t *version, dns_rdatatype_t type,
		       dns_rdatatype_t covers DNS__DB_FLARG) {
	qpcache_t *qpdb = (qpcache_t *)db;
	qpcnode_t *qpnode = (qpcnode_t *)node;
	isc_result_t result;
	dns_slabheader_t *newheader = NULL;
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
	isc_rwlock_t *nlock = NULL;

	REQUIRE(VALID_QPDB(qpdb));
	REQUIRE(version == NULL);

	if (type == dns_rdatatype_any) {
		return ISC_R_NOTIMPLEMENTED;
	}
	if (type == dns_rdatatype_rrsig && covers == 0) {
		return ISC_R_NOTIMPLEMENTED;
	}

	newheader = dns_slabheader_new(db, node);
	newheader->typepair = DNS_TYPEPAIR_VALUE(type, covers);
	setttl(newheader, 0);
	atomic_init(&newheader->attributes, DNS_SLABHEADERATTR_NONEXISTENT);

	nlock = &qpdb->buckets[qpnode->locknum].lock;
	NODE_WRLOCK(nlock, &nlocktype);
	result = add(qpdb, qpnode, NULL, newheader, DNS_DBADD_FORCE, NULL, 0,
		     nlocktype, isc_rwlocktype_none DNS__DB_FLARG_PASS);
	NODE_UNLOCK(nlock, &nlocktype);

	return result;
}

static unsigned int
nodecount(dns_db_t *db, dns_dbtree_t tree) {
	qpcache_t *qpdb = (qpcache_t *)db;
	dns_qp_memusage_t mu;
	isc_rwlocktype_t tlocktype = isc_rwlocktype_none;

	REQUIRE(VALID_QPDB(qpdb));

	TREE_RDLOCK(&qpdb->tree_lock, &tlocktype);
	switch (tree) {
	case dns_dbtree_main:
		mu = dns_qp_memusage(qpdb->tree);
		break;
	case dns_dbtree_nsec:
		mu = dns_qp_memusage(qpdb->nsec);
		break;
	default:
		UNREACHABLE();
	}
	TREE_UNLOCK(&qpdb->tree_lock, &tlocktype);

	return mu.leaves;
}

static void
qpcnode_locknode(dns_dbnode_t *node, isc_rwlocktype_t type) {
	qpcnode_t *qpnode = (qpcnode_t *)node;
	qpcache_t *qpdb = qpnode->qpdb;

	RWLOCK(&qpdb->buckets[qpnode->locknum].lock, type);
}

static void
qpcnode_unlocknode(dns_dbnode_t *node, isc_rwlocktype_t type) {
	qpcnode_t *qpnode = (qpcnode_t *)node;
	qpcache_t *qpdb = qpnode->qpdb;

	RWUNLOCK(&qpdb->buckets[qpnode->locknum].lock, type);
}

isc_result_t
dns__qpcache_create(isc_mem_t *mctx, const dns_name_t *origin,
		    dns_dbtype_t type, dns_rdataclass_t rdclass,
		    unsigned int argc, char *argv[],
		    void *driverarg ISC_ATTR_UNUSED, dns_db_t **dbp) {
	qpcache_t *qpdb = NULL;
	isc_mem_t *hmctx = mctx;
	isc_loop_t *loop = isc_loop();
	int i;
	size_t nloops = isc_loopmgr_nloops();

	/* This database implementation only supports cache semantics */
	REQUIRE(type == dns_dbtype_cache);
	REQUIRE(loop != NULL);

	qpdb = isc_mem_get(mctx,
			   sizeof(*qpdb) + nloops * sizeof(qpdb->buckets[0]));
	*qpdb = (qpcache_t){
		.common.methods = &qpdb_cachemethods,
		.common.origin = DNS_NAME_INITEMPTY,
		.common.rdclass = rdclass,
		.common.attributes = DNS_DBATTR_CACHE,
		.common.references = 1,
		.references = 1,
		.buckets_count = nloops,
	};

	/*
	 * If argv[0] exists, it points to a memory context to use for heap
	 */
	if (argc != 0) {
		hmctx = (isc_mem_t *)argv[0];
	}

	isc_rwlock_init(&qpdb->lock);
	TREE_INITLOCK(&qpdb->tree_lock);

	qpdb->buckets_count = isc_loopmgr_nloops();

	dns_rdatasetstats_create(mctx, &qpdb->rrsetstats);
	for (i = 0; i < (int)qpdb->buckets_count; i++) {
		ISC_SIEVE_INIT(qpdb->buckets[i].sieve);

		qpdb->buckets[i].heap = NULL;
		isc_heap_create(hmctx, ttl_sooner, set_index, 0,
				&qpdb->buckets[i].heap);

		isc_queue_init(&qpdb->buckets[i].deadnodes);

		NODE_INITLOCK(&qpdb->buckets[i].lock);
	}

	/*
	 * Attach to the mctx.  The database will persist so long as there
	 * are references to it, and attaching to the mctx ensures that our
	 * mctx won't disappear out from under us.
	 */
	isc_mem_attach(mctx, &qpdb->common.mctx);
	isc_mem_attach(hmctx, &qpdb->hmctx);

	/*
	 * Make a copy of the origin name.
	 */
	dns_name_dup(origin, mctx, &qpdb->common.origin);

	/*
	 * Make the qp tries.
	 */
	dns_qp_create(mctx, &qpmethods, qpdb, &qpdb->tree);
	dns_qp_create(mctx, &qpmethods, qpdb, &qpdb->nsec);

	qpdb->common.magic = DNS_DB_MAGIC;
	qpdb->common.impmagic = QPDB_MAGIC;

	*dbp = (dns_db_t *)qpdb;

	return ISC_R_SUCCESS;
}

/*
 * Rdataset Iterator Methods
 */

static void
rdatasetiter_destroy(dns_rdatasetiter_t **iteratorp DNS__DB_FLARG) {
	qpc_rditer_t *iterator = NULL;

	iterator = (qpc_rditer_t *)(*iteratorp);

	dns__db_detachnode(&iterator->common.node DNS__DB_FLARG_PASS);
	isc_mem_put(iterator->common.db->mctx, iterator, sizeof(*iterator));

	*iteratorp = NULL;
}

static bool
iterator_active(qpcache_t *qpdb, qpc_rditer_t *iterator,
		dns_slabheader_t *header) {
	dns_ttl_t stale_ttl = header->expire + STALE_TTL(header, qpdb);

	/*
	 * Is this a "this rdataset doesn't exist" record?
	 */
	if (!EXISTS(header)) {
		return false;
	}

	/*
	 * If this header is still active then return it.
	 */
	if (ACTIVE(header, iterator->common.now)) {
		return true;
	}

	/*
	 * If we are not returning stale records or the rdataset is
	 * too old don't return it.
	 */
	if (!STALEOK(iterator) || (iterator->common.now > stale_ttl)) {
		return false;
	}
	return true;
}

static isc_result_t
rdatasetiter_first(dns_rdatasetiter_t *it DNS__DB_FLARG) {
	qpc_rditer_t *iterator = (qpc_rditer_t *)it;
	qpcache_t *qpdb = (qpcache_t *)(iterator->common.db);
	qpcnode_t *qpnode = (qpcnode_t *)iterator->common.node;
	dns_slabheader_t *header = NULL;
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
	isc_rwlock_t *nlock = &qpdb->buckets[qpnode->locknum].lock;

	NODE_RDLOCK(nlock, &nlocktype);

	for (header = qpnode->data; header != NULL; header = header->next) {
		if ((EXPIREDOK(iterator) && EXISTS(header)) ||
		    iterator_active(qpdb, iterator, header))
		{
			break;
		}
	}

	NODE_UNLOCK(nlock, &nlocktype);

	iterator->current = header;

	if (header == NULL) {
		return ISC_R_NOMORE;
	}

	return ISC_R_SUCCESS;
}

static isc_result_t
rdatasetiter_next(dns_rdatasetiter_t *it DNS__DB_FLARG) {
	qpc_rditer_t *iterator = (qpc_rditer_t *)it;
	qpcache_t *qpdb = (qpcache_t *)(iterator->common.db);
	qpcnode_t *qpnode = (qpcnode_t *)iterator->common.node;
	dns_slabheader_t *header = NULL;
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
	isc_rwlock_t *nlock = &qpdb->buckets[qpnode->locknum].lock;

	header = iterator->current;
	if (header == NULL) {
		return ISC_R_NOMORE;
	}

	NODE_RDLOCK(nlock, &nlocktype);

	for (header = header->next; header != NULL; header = header->next) {
		if ((EXPIREDOK(iterator) && EXISTS(header)) ||
		    iterator_active(qpdb, iterator, header))
		{
			break;
		}
	}

	NODE_UNLOCK(nlock, &nlocktype);

	iterator->current = header;

	if (header == NULL) {
		return ISC_R_NOMORE;
	}

	return ISC_R_SUCCESS;
}

static void
rdatasetiter_current(dns_rdatasetiter_t *it,
		     dns_rdataset_t *rdataset DNS__DB_FLARG) {
	qpc_rditer_t *iterator = (qpc_rditer_t *)it;
	qpcache_t *qpdb = (qpcache_t *)(iterator->common.db);
	qpcnode_t *qpnode = (qpcnode_t *)iterator->common.node;
	dns_slabheader_t *header = NULL;
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
	isc_rwlock_t *nlock = &qpdb->buckets[qpnode->locknum].lock;

	header = iterator->current;
	REQUIRE(header != NULL);

	NODE_RDLOCK(nlock, &nlocktype);

	bindrdataset(qpdb, qpnode, header, iterator->common.now, nlocktype,
		     isc_rwlocktype_none, rdataset DNS__DB_FLARG_PASS);

	NODE_UNLOCK(nlock, &nlocktype);
}

/*
 * Database Iterator Methods
 */

static void
reference_iter_node(qpc_dbit_t *qpdbiter DNS__DB_FLARG) {
	qpcache_t *qpdb = (qpcache_t *)qpdbiter->common.db;
	qpcnode_t *node = qpdbiter->node;

	if (node == NULL) {
		return;
	}

	INSIST(qpdbiter->tree_locked != isc_rwlocktype_none);
	reactivate_node(qpdb, node, qpdbiter->tree_locked DNS__DB_FLARG_PASS);
}

static void
dereference_iter_node(qpc_dbit_t *qpdbiter DNS__DB_FLARG) {
	qpcache_t *qpdb = (qpcache_t *)qpdbiter->common.db;
	qpcnode_t *node = qpdbiter->node;
	isc_rwlock_t *nlock = NULL;
	isc_rwlocktype_t nlocktype = isc_rwlocktype_none;
	isc_rwlocktype_t tlocktype = qpdbiter->tree_locked;

	if (node == NULL) {
		return;
	}

	REQUIRE(tlocktype != isc_rwlocktype_write);

	nlock = &qpdb->buckets[node->locknum].lock;
	NODE_RDLOCK(nlock, &nlocktype);
	qpcnode_release(qpdb, node, &nlocktype,
			&qpdbiter->tree_locked DNS__DB_FLARG_PASS);
	NODE_UNLOCK(nlock, &nlocktype);

	INSIST(qpdbiter->tree_locked == tlocktype);

	qpdbiter->node = NULL;
}

static void
resume_iteration(qpc_dbit_t *qpdbiter, bool continuing) {
	qpcache_t *qpdb = (qpcache_t *)qpdbiter->common.db;

	REQUIRE(qpdbiter->paused);
	REQUIRE(qpdbiter->tree_locked == isc_rwlocktype_none);

	TREE_RDLOCK(&qpdb->tree_lock, &qpdbiter->tree_locked);

	/*
	 * If we're being called from dbiterator_next or _prev,
	 * then we may need to reinitialize the iterator to the current
	 * name. The tree could have changed while it was unlocked,
	 * would make the iterator traversal inconsistent.
	 *
	 * As long as the iterator is holding a reference to
	 * qpdbiter->node, the node won't be removed from the tree,
	 * so the lookup should always succeed.
	 */
	if (continuing && qpdbiter->node != NULL) {
		isc_result_t result;
		result = dns_qp_lookup(qpdb->tree, qpdbiter->name,
				       DNS_DBNAMESPACE_NORMAL, NULL,
				       &qpdbiter->iter, NULL, NULL, NULL);
		INSIST(result == ISC_R_SUCCESS);
	}

	qpdbiter->paused = false;
}

static void
dbiterator_destroy(dns_dbiterator_t **iteratorp DNS__DB_FLARG) {
	qpc_dbit_t *qpdbiter = (qpc_dbit_t *)(*iteratorp);
	qpcache_t *qpdb = (qpcache_t *)qpdbiter->common.db;
	dns_db_t *db = NULL;

	if (qpdbiter->tree_locked == isc_rwlocktype_read) {
		TREE_UNLOCK(&qpdb->tree_lock, &qpdbiter->tree_locked);
	}
	INSIST(qpdbiter->tree_locked == isc_rwlocktype_none);

	dereference_iter_node(qpdbiter DNS__DB_FLARG_PASS);

	dns_db_attach(qpdbiter->common.db, &db);
	dns_db_detach(&qpdbiter->common.db);

	isc_mem_put(db->mctx, qpdbiter, sizeof(*qpdbiter));
	dns_db_detach(&db);

	*iteratorp = NULL;
}

static isc_result_t
dbiterator_first(dns_dbiterator_t *iterator DNS__DB_FLARG) {
	isc_result_t result;
	qpc_dbit_t *qpdbiter = (qpc_dbit_t *)iterator;
	qpcache_t *qpdb = (qpcache_t *)iterator->db;

	if (qpdbiter->result != ISC_R_SUCCESS &&
	    qpdbiter->result != ISC_R_NOTFOUND &&
	    qpdbiter->result != DNS_R_PARTIALMATCH &&
	    qpdbiter->result != ISC_R_NOMORE)
	{
		return qpdbiter->result;
	}

	if (qpdbiter->paused) {
		resume_iteration(qpdbiter, false);
	}

	dereference_iter_node(qpdbiter DNS__DB_FLARG_PASS);

	dns_qpiter_init(qpdb->tree, &qpdbiter->iter);
	result = dns_qpiter_next(&qpdbiter->iter, NULL,
				 (void **)&qpdbiter->node, NULL);

	if (result == ISC_R_SUCCESS) {
		dns_name_copy(&qpdbiter->node->name, qpdbiter->name);
		reference_iter_node(qpdbiter DNS__DB_FLARG_PASS);
	} else {
		INSIST(result == ISC_R_NOMORE); /* The tree is empty. */
		qpdbiter->node = NULL;
	}

	qpdbiter->result = result;

	if (result != ISC_R_SUCCESS) {
		ENSURE(!qpdbiter->paused);
	}

	return result;
}

static isc_result_t
dbiterator_last(dns_dbiterator_t *iterator DNS__DB_FLARG) {
	isc_result_t result;
	qpc_dbit_t *qpdbiter = (qpc_dbit_t *)iterator;
	qpcache_t *qpdb = (qpcache_t *)iterator->db;

	if (qpdbiter->result != ISC_R_SUCCESS &&
	    qpdbiter->result != ISC_R_NOTFOUND &&
	    qpdbiter->result != DNS_R_PARTIALMATCH &&
	    qpdbiter->result != ISC_R_NOMORE)
	{
		return qpdbiter->result;
	}

	if (qpdbiter->paused) {
		resume_iteration(qpdbiter, false);
	}

	dereference_iter_node(qpdbiter DNS__DB_FLARG_PASS);

	dns_qpiter_init(qpdb->tree, &qpdbiter->iter);
	result = dns_qpiter_prev(&qpdbiter->iter, NULL,
				 (void **)&qpdbiter->node, NULL);

	if (result == ISC_R_SUCCESS) {
		dns_name_copy(&qpdbiter->node->name, qpdbiter->name);
		reference_iter_node(qpdbiter DNS__DB_FLARG_PASS);
	} else {
		INSIST(result == ISC_R_NOMORE); /* The tree is empty. */
		qpdbiter->node = NULL;
	}

	qpdbiter->result = result;
	return result;
}

static isc_result_t
dbiterator_seek(dns_dbiterator_t *iterator,
		const dns_name_t *name DNS__DB_FLARG) {
	isc_result_t result;
	qpc_dbit_t *qpdbiter = (qpc_dbit_t *)iterator;
	qpcache_t *qpdb = (qpcache_t *)iterator->db;

	if (qpdbiter->result != ISC_R_SUCCESS &&
	    qpdbiter->result != ISC_R_NOTFOUND &&
	    qpdbiter->result != DNS_R_PARTIALMATCH &&
	    qpdbiter->result != ISC_R_NOMORE)
	{
		return qpdbiter->result;
	}

	if (qpdbiter->paused) {
		resume_iteration(qpdbiter, false);
	}

	dereference_iter_node(qpdbiter DNS__DB_FLARG_PASS);

	result = dns_qp_lookup(qpdb->tree, name, DNS_DBNAMESPACE_NORMAL, NULL,
			       &qpdbiter->iter, NULL, (void **)&qpdbiter->node,
			       NULL);

	if (result == ISC_R_SUCCESS || result == DNS_R_PARTIALMATCH) {
		dns_name_copy(&qpdbiter->node->name, qpdbiter->name);
		reference_iter_node(qpdbiter DNS__DB_FLARG_PASS);
	} else {
		qpdbiter->node = NULL;
	}

	qpdbiter->result = (result == DNS_R_PARTIALMATCH) ? ISC_R_SUCCESS
							  : result;
	return result;
}

static isc_result_t
dbiterator_prev(dns_dbiterator_t *iterator DNS__DB_FLARG) {
	isc_result_t result;
	qpc_dbit_t *qpdbiter = (qpc_dbit_t *)iterator;

	REQUIRE(qpdbiter->node != NULL);

	if (qpdbiter->result != ISC_R_SUCCESS) {
		return qpdbiter->result;
	}

	if (qpdbiter->paused) {
		resume_iteration(qpdbiter, true);
	}

	dereference_iter_node(qpdbiter DNS__DB_FLARG_PASS);

	result = dns_qpiter_prev(&qpdbiter->iter, NULL,
				 (void **)&qpdbiter->node, NULL);

	if (result == ISC_R_SUCCESS) {
		dns_name_copy(&qpdbiter->node->name, qpdbiter->name);
		reference_iter_node(qpdbiter DNS__DB_FLARG_PASS);
	} else {
		INSIST(result == ISC_R_NOMORE);
		qpdbiter->node = NULL;
	}

	qpdbiter->result = result;
	return result;
}

static isc_result_t
dbiterator_next(dns_dbiterator_t *iterator DNS__DB_FLARG) {
	isc_result_t result;
	qpc_dbit_t *qpdbiter = (qpc_dbit_t *)iterator;

	REQUIRE(qpdbiter->node != NULL);

	if (qpdbiter->result != ISC_R_SUCCESS) {
		return qpdbiter->result;
	}

	if (qpdbiter->paused) {
		resume_iteration(qpdbiter, true);
	}

	dereference_iter_node(qpdbiter DNS__DB_FLARG_PASS);

	result = dns_qpiter_next(&qpdbiter->iter, NULL,
				 (void **)&qpdbiter->node, NULL);

	if (result == ISC_R_SUCCESS) {
		dns_name_copy(&qpdbiter->node->name, qpdbiter->name);
		reference_iter_node(qpdbiter DNS__DB_FLARG_PASS);
	} else {
		INSIST(result == ISC_R_NOMORE);
		qpdbiter->node = NULL;
	}

	qpdbiter->result = result;
	return result;
}

static isc_result_t
dbiterator_current(dns_dbiterator_t *iterator, dns_dbnode_t **nodep,
		   dns_name_t *name DNS__DB_FLARG) {
	qpcache_t *qpdb = (qpcache_t *)iterator->db;
	qpc_dbit_t *qpdbiter = (qpc_dbit_t *)iterator;
	qpcnode_t *node = qpdbiter->node;

	REQUIRE(qpdbiter->result == ISC_R_SUCCESS);
	REQUIRE(node != NULL);

	if (qpdbiter->paused) {
		resume_iteration(qpdbiter, false);
	}

	if (name != NULL) {
		dns_name_copy(&node->name, name);
	}

	qpcnode_acquire(qpdb, node, isc_rwlocktype_none,
			qpdbiter->tree_locked DNS__DB_FLARG_PASS);

	*nodep = (dns_dbnode_t *)qpdbiter->node;
	return ISC_R_SUCCESS;
}

static isc_result_t
dbiterator_pause(dns_dbiterator_t *iterator) {
	qpcache_t *qpdb = (qpcache_t *)iterator->db;
	qpc_dbit_t *qpdbiter = (qpc_dbit_t *)iterator;

	if (qpdbiter->result != ISC_R_SUCCESS &&
	    qpdbiter->result != ISC_R_NOTFOUND &&
	    qpdbiter->result != DNS_R_PARTIALMATCH &&
	    qpdbiter->result != ISC_R_NOMORE)
	{
		return qpdbiter->result;
	}

	if (qpdbiter->paused) {
		return ISC_R_SUCCESS;
	}

	qpdbiter->paused = true;

	if (qpdbiter->tree_locked == isc_rwlocktype_read) {
		TREE_UNLOCK(&qpdb->tree_lock, &qpdbiter->tree_locked);
	}
	INSIST(qpdbiter->tree_locked == isc_rwlocktype_none);

	return ISC_R_SUCCESS;
}

static isc_result_t
dbiterator_origin(dns_dbiterator_t *iterator, dns_name_t *name) {
	qpc_dbit_t *qpdbiter = (qpc_dbit_t *)iterator;

	if (qpdbiter->result != ISC_R_SUCCESS) {
		return qpdbiter->result;
	}

	dns_name_copy(dns_rootname, name);
	return ISC_R_SUCCESS;
}

static void
qpcnode_deletedata(dns_dbnode_t *node ISC_ATTR_UNUSED, void *data) {
	dns_slabheader_t *header = data;
	qpcache_t *qpdb = HEADERNODE(header)->qpdb;

	int idx = HEADERNODE(header)->locknum;

	if (header->heap != NULL && header->heap_index != 0) {
		isc_heap_delete(header->heap, header->heap_index);
	}

	update_rrsetstats(qpdb->rrsetstats, header->typepair,
			  atomic_load_acquire(&header->attributes), false);

	if (ISC_LINK_LINKED(header, link)) {
		ISC_SIEVE_UNLINK(qpdb->buckets[idx].sieve, header, link);
	}

	if (header->noqname != NULL) {
		dns_slabheader_freeproof(qpdb->common.mctx, &header->noqname);
	}
	if (header->closest != NULL) {
		dns_slabheader_freeproof(qpdb->common.mctx, &header->closest);
	}
}

/*
 * Caller must be holding the node write lock.
 */
static void
expire_ttl_headers(qpcache_t *qpdb, unsigned int locknum,
		   isc_rwlocktype_t *nlocktypep, isc_rwlocktype_t *tlocktypep,
		   isc_stdtime_t now DNS__DB_FLARG) {
	isc_heap_t *heap = qpdb->buckets[locknum].heap;

	for (size_t i = 0; i < DNS_QPDB_EXPIRE_TTL_COUNT; i++) {
		dns_slabheader_t *header = isc_heap_element(heap, 1);

		if (header == NULL) {
			/* No headers left on this TTL heap; exit cleaning */
			return;
		}

		dns_ttl_t ttl = header->expire + STALE_TTL(header, qpdb);

		if (ttl >= now - QPDB_VIRTUAL) {
			/*
			 * The header at the top of this TTL heap is not yet
			 * eligible for expiry, so none of the other headers on
			 * the same heap can be eligible for expiry, either;
			 * exit cleaning.
			 */
			return;
		}

		expireheader(header, nlocktypep, tlocktypep,
			     dns_expire_ttl DNS__DB_FLARG_PASS);
	}
}

static void
setmaxrrperset(dns_db_t *db, uint32_t value) {
	qpcache_t *qpdb = (qpcache_t *)db;

	REQUIRE(VALID_QPDB(qpdb));

	qpdb->maxrrperset = value;
}

static void
setmaxtypepername(dns_db_t *db, uint32_t value) {
	qpcache_t *qpdb = (qpcache_t *)db;

	REQUIRE(VALID_QPDB(qpdb));

	qpdb->maxtypepername = value;
}

static dns_dbmethods_t qpdb_cachemethods = {
	.destroy = qpcache_destroy,
	.findnode = qpcache_findnode,
	.find = qpcache_find,
	.findzonecut = qpcache_findzonecut,
	.createiterator = qpcache_createiterator,
	.findrdataset = qpcache_findrdataset,
	.allrdatasets = qpcache_allrdatasets,
	.addrdataset = qpcache_addrdataset,
	.deleterdataset = qpcache_deleterdataset,
	.nodecount = nodecount,
	.getrrsetstats = getrrsetstats,
	.setcachestats = setcachestats,
	.setservestalettl = setservestalettl,
	.getservestalettl = getservestalettl,
	.setservestalerefresh = setservestalerefresh,
	.getservestalerefresh = getservestalerefresh,
	.setmaxrrperset = setmaxrrperset,
	.setmaxtypepername = setmaxtypepername,
};

static void
qpcnode_destroy(qpcnode_t *data) {
	dns_slabheader_t *current = NULL, *next = NULL;

	for (current = data->data; current != NULL; current = next) {
		dns_slabheader_t *down = current->down, *down_next = NULL;

		next = current->next;

		for (down = current->down; down != NULL; down = down_next) {
			down_next = down->down;
			dns_slabheader_destroy(&down);
		}

		dns_slabheader_destroy(&current);
	}

	dns_name_free(&data->name, data->mctx);
	isc_mem_putanddetach(&data->mctx, data, sizeof(qpcnode_t));
}

#ifdef DNS_DB_NODETRACE
ISC_REFCOUNT_STATIC_TRACE_IMPL(qpcnode, qpcnode_destroy);
#else
ISC_REFCOUNT_STATIC_IMPL(qpcnode, qpcnode_destroy);
#endif

#ifdef DNS_DB_NODETRACE
ISC_REFCOUNT_STATIC_TRACE_IMPL(qpcache, qpcache__destroy);
#else
ISC_REFCOUNT_STATIC_IMPL(qpcache, qpcache__destroy);
#endif