lruhash insert and lookup.

git-svn-id: file:///svn/unbound/trunk@176 be551aaa-1e26-0410-a405-d3ace91eadb9

doc/Changelog

@@ -1,3 +1,6 @@
+14 March 2007: Wouter
+	- hash table insert (and subroutines) and lookup implemented.
+
 13 March 2007: Wouter
 	- lock_unprotect in checklocks.
 	- util/storage/lruhash.h for LRU hash table structure.

util/storage/lruhash.c

@@ -45,6 +45,9 @@
 
 /* ------ local helper functions ------------- */
 
+/** init the hash bins for the table. */
+static void bin_init(struct lruhash_bin* array, size_t size);
+
 /** delete the hash bin and entries inside it */
 static void bin_delete(struct lruhash* table, struct lruhash_bin* bin);
 
@@ -58,25 +61,83 @@ static void bin_delete(struct lruhash* table, struct lruhash_bin* bin);
 static struct lruhash_entry* bin_find_entry(struct lruhash* table, 
 	struct lruhash_bin* bin, hashvalue_t hash, void* key);
 
-/** 
+/**
- * Try to make howmuch space available for a new entry into the table.
+ * Remove entry from bin overflow chain.
- * works by deleting old entries.
+ * You must have locked the bin.
- * Assumes that the lock on the hashtable is being held by caller.
+ * @param bin: hash bin to look into.
- * @param table: hash table.
+ * @param entry: entry ptr that needs removal.
- * @param howmuch: will try to make max-used at least this amount.
  */
-static void reclaim_space(struct lruhash* table, size_t howmuch);
+static void bin_overflow_remove(struct lruhash_bin* bin, 
+	struct lruhash_entry* entry);
+
+/**
+ * Split hash bin into two new ones. Based on increased size_mask.
+ * Caller must hold hash table lock.
+ * At the end the routine acquires all hashbin locks (in the old array).
+ * This makes it wait for other threads to finish with the bins.
+ * So the bins are ready to be deleted after this function.
+ * @param table: hash table with function pointers.
+ * @param newa: new increased array.
+ * @param newmask: new lookup mask.
+ */
+static void bin_split(struct lruhash* table, struct lruhash_bin* newa, 
+	int newmask);
+
+/** 
+ * Try to make space available by deleting old entries.
+ * Assumes that the lock on the hashtable is being held by caller.
+ * Caller must not hold bin locks.
+ * @param table: hash table.
+ * @param list: list of entries that are to be deleted later.
+ *	Entries have been removed from the hash table and writelock is held.
+ */
+static void reclaim_space(struct lruhash* table, struct lruhash_entry** list);
 
 /**
  * Grow the table lookup array. Becomes twice as large.
- * Caller must hold the hash table lock.
+ * Caller must hold the hash table lock. Must not hold any bin locks.
+ * Tries to grow, on malloc failure, nothing happened.
  * @param table: hash table.
- * @return false on error (malloc).
  */
-static int table_grow(struct lruhash* table);
+static void table_grow(struct lruhash* table);
+
+/**
+ * Touch entry, so it becomes the most recently used in the LRU list.
+ * Caller must hold hash table lock. The entry must be inserted already.
+ * @param table: hash table.
+ * @param entry: entry to make first in LRU.
+ */
+static void lru_touch(struct lruhash* table, struct lruhash_entry* entry);
+
+/**
+ * Put entry at front of lru. entry must be unlinked from lru.
+ * Caller must hold hash table lock.
+ * @param table: hash table with lru head and tail.
+ * @param entry: entry to make most recently used.
+ */
+static void lru_front(struct lruhash* table, struct lruhash_entry* entry);
+
+/**
+ * Remove entry from lru list.
+ * Caller must hold hash table lock.
+ * @param table: hash table with lru head and tail.
+ * @param entry: entry to remove from lru.
+ */
+static void lru_remove(struct lruhash* table, struct lruhash_entry* entry);
 
 /* ------ end local helper functions --------- */
 
+static void
+bin_init(struct lruhash_bin* array, size_t size)
+{
+	size_t i;
+	for(i=0; i<size; i++) {
+		lock_quick_init(&array[i].lock);
+		lock_protect(&array[i].lock, &array[i], 
+			sizeof(struct lruhash_bin));
+	}
+}
+
 struct lruhash* lruhash_create(size_t start_size, size_t maxmem,
         lruhash_sizefunc_t sizefunc, lruhash_compfunc_t compfunc,
 	lruhash_delkeyfunc_t delkeyfunc, lruhash_deldatafunc_t deldatafunc,
@@ -84,7 +145,6 @@ struct lruhash* lruhash_create(size_t start_size, size_t maxmem,
 {
 	struct lruhash* table = (struct lruhash*)calloc(1, 
 		sizeof(struct lruhash));
-	size_t i;
 	if(!table)
 		return NULL;
 	lock_quick_init(&table->lock);
@@ -94,7 +154,7 @@ struct lruhash* lruhash_create(size_t start_size, size_t maxmem,
 	table->deldatafunc = deldatafunc;
 	table->cb_arg = arg;
 	table->size = start_size;
-	table->size_mask = start_size-1;
+	table->size_mask = (int)(start_size-1);
 	table->lru_start = NULL;
 	table->lru_end = NULL;
 	table->num = 0;
@@ -106,9 +166,7 @@ struct lruhash* lruhash_create(size_t start_size, size_t maxmem,
 		free(table);
 		return NULL;
 	}
-	for(i=0; i<table->size; i++) {
+	bin_init(table->array, table->size);
-		lock_quick_init(&table->array[i].lock);
-	}
 	lock_protect(&table->lock, table, sizeof(*table));
 	lock_protect(&table->lock, table->array, 
 		table->size*sizeof(struct lruhash_bin));
@@ -131,6 +189,34 @@ static void bin_delete(struct lruhash* table, struct lruhash_bin* bin)
 	}
 }
 
+static void 
+bin_split(struct lruhash* table, struct lruhash_bin* newa, 
+	int newmask)
+{
+	size_t i;
+	struct lruhash_entry *p, *np;
+	int newbin;
+	/* move entries to new table. Notice that since hash x is mapped to
+	 * bin x & mask, and new mask uses one more bit, so all entries in
+	 * one bin will go into the old bin or bin | newbit */
+	/* newbit = newmask - table->size_mask; */
+	/* so, really, this task could also be threaded, per bin. */
+	/* LRU list is not changed */
+	for(i=0; i<table->size; i++)
+	{
+		lock_quick_lock(&table->array[i].lock);
+		p = table->array[i].overflow_list;
+		while(p) {
+			np = p->overflow_next;
+			/* link into correct new bin */
+			newbin = p->hash & newmask;
+			p->overflow_next = newa[newbin].overflow_list;
+			newa[newbin].overflow_list = p;
+			p=np;
+		}
+	}
+}
+
 void lruhash_delete(struct lruhash* table)
 {
 	size_t i;
@@ -145,35 +231,146 @@ void lruhash_delete(struct lruhash* table)
 }
 
 static void 
-reclaim_space(struct lruhash* table, size_t howmuch)
+bin_overflow_remove(struct lruhash_bin* bin, struct lruhash_entry* entry)
 {
+	struct lruhash_entry* p = bin->overflow_list;
+	struct lruhash_entry** prevp = &bin->overflow_list;
+	while(p) {
+		if(p == entry) {
+			*prevp = p->overflow_next;
+			return;
+		}
+		prevp = &p->overflow_next;
+		p = p->overflow_next;
+	}
+}
+
+static void 
+reclaim_space(struct lruhash* table, struct lruhash_entry** list)
+{
+	struct lruhash_entry* d;
+	struct lruhash_bin* bin;
+	log_assert(table);
+	/* does not delete MRU entry, so table will not be empty. */
+	while(table->num > 1 && table->space_used > table->space_max) {
+		/* notice that since we hold the hashtable lock, nobody
+		   can change the lru chain. So it cannot be deleted underneath
+		   us. We still need the hashbin and entry write lock to make 
+		   sure we flush all users away from the entry. 
+		   which is unlikely, since it is LRU, if someone got a rdlock
+		   it would be moved to front, but to be sure. */
+		d = table->lru_end;
+		/* specialised, delete from end of double linked list,
+		   and we know num>1, so there is a previous lru entry. */
+		log_assert(d && d->lru_prev);
+		table->lru_end = d->lru_prev;
+		d->lru_prev->lru_next = NULL;
+		/* schedule entry for deletion */
+		bin = &table->array[d->hash & table->size_mask];
+		table->num --;
+		lock_quick_lock(&bin->lock);
+		bin_overflow_remove(bin, d);
+		d->overflow_next = *list;
+		*list = d;
+		lock_rw_wrlock(&d->lock);
+		table->space_used -= table->sizefunc(d->key, d->data);
+		lock_quick_unlock(&bin->lock);
+	}
 }
 
 static struct lruhash_entry* 
 bin_find_entry(struct lruhash* table, 
 	struct lruhash_bin* bin, hashvalue_t hash, void* key)
 {
+	struct lruhash_entry* p = bin->overflow_list;
+	while(p) {
+		if(p->hash == hash && table->compfunc(p->key, key) == 0)
+			return p;
+		p = p->overflow_next;
+	}
 	return NULL;
 }
 
-static int table_grow(struct lruhash* table)
+static void 
+table_grow(struct lruhash* table)
 {
-	return 0;
+	struct lruhash_bin* newa;
+	int newmask;
+	size_t i;
+	if(table->size_mask == (int)(((size_t)-1)>>1)) {
+		log_err("hash array malloc: size_t too small");
+		return;
+	}
+	/* try to allocate new array, if not fail */
+	newa = calloc(table->size*2, sizeof(struct lruhash_bin));
+	if(!newa) {
+		log_err("hash grow: malloc failed");
+		/* continue with smaller array. Though its slower. */
+		return;
+	}
+	bin_init(newa, table->size*2);
+	newmask = (table->size_mask << 1) | 1;
+	bin_split(table, newa, newmask);
+	/* delete the old bins */
+	lock_unprotect(&table->lock, table->array);
+	for(i=0; i<table->size; i++) {
+		lock_quick_unlock(&table->array[i].lock);
+		lock_quick_destroy(&table->array[i].lock);
+	}
+	free(table->array);
+	
+	table->size *= 2;
+	table->size_mask = newmask;
+	table->array = newa;
+	lock_protect(&table->lock, table->array, 
+		table->size*sizeof(struct lruhash_bin));
+	return;
+}
+
+static void 
+lru_front(struct lruhash* table, struct lruhash_entry* entry)
+{
+	entry->lru_prev = NULL;
+	entry->lru_next = table->lru_start;
+	if(!table->lru_start)
+		table->lru_end = entry;
+	else	table->lru_start->lru_prev = entry;
+	table->lru_start = entry;
+}
+
+static void 
+lru_remove(struct lruhash* table, struct lruhash_entry* entry)
+{
+	if(entry->lru_prev)
+		entry->lru_prev->lru_next = entry->lru_next;
+	else	table->lru_start = entry->lru_next;
+	if(entry->lru_next)
+		entry->lru_next->lru_prev = entry->lru_prev;
+	else	table->lru_end = entry->lru_prev;
+}
+
+static void 
+lru_touch(struct lruhash* table, struct lruhash_entry* entry)
+{
+	log_assert(table && entry);
+	if(entry == table->lru_start)
+		return; /* nothing to do */
+	/* remove from current lru position */
+	lru_remove(table, entry);
+	/* add at front */
+	lru_front(table, entry);
 }
 
 void lruhash_insert(struct lruhash* table, hashvalue_t hash,
         struct lruhash_entry* entry, void* data)
 {
 	struct lruhash_bin* bin;
-	struct lruhash_entry* found;
+	struct lruhash_entry* found, *reclaimlist=NULL;
 	size_t need_size;
-	lock_quick_lock(&table->lock);
-	/* see if table memory exceeded and needs clipping. */
 	need_size = table->sizefunc(entry->key, data);
-	if(table->space_used + need_size > table->space_max)
-		reclaim_space(table, need_size);
 
 	/* find bin */
+	lock_quick_lock(&table->lock);
 	bin = &table->array[hash & table->size_mask];
 	lock_quick_lock(&bin->lock);
 
@@ -182,31 +379,55 @@ void lruhash_insert(struct lruhash* table, hashvalue_t hash,
 		/* if not: add to bin */
 		entry->overflow_next = bin->overflow_list;
 		bin->overflow_list = entry;
+		lru_front(table, entry);
 		table->num++;
 		table->space_used += need_size;
 	} else {
 		/* if so: update data */
 		table->space_used += need_size -
 			(*table->sizefunc)(found->key, found->data);
-		entry->data = NULL;
 		(*table->delkeyfunc)(entry->key, table->cb_arg);
 		(*table->deldatafunc)(found->data, table->cb_arg);
 		found->data = data;
+		lru_touch(table, found);
 	}
 	lock_quick_unlock(&bin->lock);
-
+	if(table->space_used > table->space_max)
-	/* see if table lookup must be grown up */
+		reclaim_space(table, &reclaimlist);
-	if(table->num == table->size) {
+	if(table->num >= table->size)
-		if(!table_grow(table))
+		table_grow(table);
-			log_err("hash grow: malloc failed");
-			/* continue with smaller array. Slower. */
-	}
 	lock_quick_unlock(&table->lock);
+
+	/* finish reclaim if any (outside of critical region) */
+	while(reclaimlist) {
+		struct lruhash_entry* n = reclaimlist->overflow_next;
+		lock_rw_unlock(&reclaimlist->lock);
+		(*table->delkeyfunc)(reclaimlist->key, table->cb_arg);
+		(*table->deldatafunc)(reclaimlist->data, table->cb_arg);
+		reclaimlist = n;
+	}
 }
 
-struct lruhash_entry* lruhash_lookup(struct lruhash* table, void* key, int wr)
+struct lruhash_entry* lruhash_lookup(struct lruhash* table, hashvalue_t hash,
+	void* key, int wr)
 {
-	return NULL;
+	struct lruhash_entry* entry = NULL;
+	struct lruhash_bin* bin;
+
+	lock_quick_lock(&table->lock);
+	bin = &table->array[hash & table->size_mask];
+	lock_quick_lock(&bin->lock);
+	if((entry=bin_find_entry(table, bin, hash, key)))
+		lru_touch(table, entry);
+	lock_quick_unlock(&table->lock);
+
+	if(entry) {
+		if(wr)
+			lock_rw_wrlock(&entry->lock);
+		else	lock_rw_rdlock(&entry->lock);
+	}
+	lock_quick_unlock(&bin->lock);
+	return entry;
 }
 
 void lruhash_remove(struct lruhash* table, void* key)

util/storage/lruhash.h

@@ -194,13 +194,13 @@ struct lruhash_entry {
 	 * You need to delete it to change hash or key.
 	 */
 	lock_rw_t lock;
-	/** next entry in overflow chain */
+	/** next entry in overflow chain. Covered by hashlock and binlock. */
 	struct lruhash_entry* overflow_next;
-	/** next entry in lru chain */
+	/** next entry in lru chain. covered by hashlock. */
 	struct lruhash_entry* lru_next;
-	/** prev entry in lru chain */
+	/** prev entry in lru chain. covered by hashlock. */
 	struct lruhash_entry* lru_prev;
-	/** hash value of the key */
+	/** hash value of the key. It may not change, until entry deleted. */
 	hashvalue_t hash;
 	/** key */
 	void* key;
@@ -255,6 +255,7 @@ void lruhash_insert(struct lruhash* table, hashvalue_t hash,
  * At the end of the function you hold a (read/write)lock on the entry.
  * The LRU is updated for the entry (if found).
  * @param table: hash table.
+ * @param hash: hash of key.
  * @param key: what to look for, compared against entries in overflow chain.
  *    the hash value must be set, and must work with compare function.
  * @param wr: set to true if you desire a writelock on the entry.
@@ -262,7 +263,8 @@ void lruhash_insert(struct lruhash* table, hashvalue_t hash,
  * @return: pointer to the entry or NULL. The entry is locked.
  *    The user must unlock the entry when done.
  */
-struct lruhash_entry* lruhash_lookup(struct lruhash* table, void* key, int wr);
+struct lruhash_entry* lruhash_lookup(struct lruhash* table, hashvalue_t hash, 
+	void* key, int wr);
 
 /**
  * Remove entry from hashtable. Does nothing if not found in hashtable.