first pass at cache

cache.h

@@ -0,0 +1,253 @@
+#ifndef CACHE_H
+#define CACHE_H
+
+#include <boost/intrusive/circular_list_algorithms.hpp>
+#include <boost/intrusive/rbtree_algorithms.hpp>
+#include <boost/optional.hpp>
+#include <list>
+#include <map>
+#include <stdexcept>
+
+//----------------------------------------------------------------
+
+namespace base {
+	// FIXME: move somewhere more useful
+	template <typename T>
+	struct deleter {
+		void operator()(T *t) {
+			delete t;
+		}
+	};
+
+	// ValueTraits needs to define value_type, key_type and a get_key()
+	// static function.  Commonly you will want value_type to be a
+	// shared_ptr, with any teardown specific stuff in the destructor.
+
+	template <typename ValueTraits>
+	class cache {
+	public:
+		typedef typename ValueTraits::value_type value_type;
+		typedef typename ValueTraits::key_type key_type;
+
+		cache(unsigned max_entries);
+		~cache();
+
+		void insert(value_type const &v);
+
+		boost::optional<value_type> get(key_type const &k);
+		void put(value_type const &k);
+
+	private:
+		void make_space();
+
+		struct value_entry {
+			// FIXME: this means the cached object must have a
+			// default constructor also, which is a shame.
+			// so we can construct the headers.
+			value_entry()
+				: ref_count_(0) {
+			}
+
+			explicit value_entry(value_type v)
+				: ref_count_(0),
+				  v_(v) {
+			}
+
+			struct {
+				value_entry *next_, *prev_;
+			} lru_;
+
+			struct {
+				value_entry *parent_, *left_, *right_;
+				int color_;
+			} lookup_;
+
+			unsigned ref_count_;
+			value_type v_;
+		};
+
+		struct value_ptr_cmp {
+			bool operator() (value_entry const *lhs, value_entry const *rhs) {
+				key_type k1 = ValueTraits::get_key(lhs->v_);
+				key_type k2 = ValueTraits::get_key(rhs->v_);
+
+				return k1 < k2;
+			}
+		};
+
+		struct key_value_ptr_cmp {
+			bool operator() (key_type const &k1, value_entry const *rhs) {
+				key_type k2 = ValueTraits::get_key(rhs->v_);
+				return k1 < k2;
+			}
+
+			bool operator() (value_entry const *lhs, key_type const &k2) {
+				key_type k1 = ValueTraits::get_key(lhs->v_);
+				return k1 < k2;
+			}
+
+		};
+
+		struct list_node_traits {
+			typedef value_entry node;
+			typedef value_entry *node_ptr;
+			typedef const value_entry *const_node_ptr;
+
+			static node_ptr get_next(const_node_ptr n) {
+				return n->lru_.next_;
+			}
+
+			static void set_next(node_ptr n, node_ptr next) {
+				n->lru_.next_ = next;
+			}
+
+			static node_ptr get_previous(const_node_ptr n) {
+				return n->lru_.prev_;
+			}
+
+			static void set_previous(node_ptr n, node_ptr prev) {
+				n->lru_.prev_ = prev;
+			}
+		};
+
+		struct rbtree_node_traits {
+			typedef value_entry node;
+			typedef value_entry *node_ptr;
+			typedef const value_entry * const_node_ptr;
+			typedef int color;
+
+			static node_ptr get_parent(const_node_ptr n) {
+				return n->lookup_.parent_;
+			}
+
+			static void set_parent(node_ptr n, node_ptr parent) {
+				n->lookup_.parent_ = parent;
+			}
+
+			static node_ptr get_left(const_node_ptr n) {
+				return n->lookup_.left_;
+			}
+
+			static void set_left(node_ptr n, node_ptr left) {
+				n->lookup_.left_ = left;
+			}
+
+			static node_ptr get_right(const_node_ptr n) {
+				return n->lookup_.right_;
+			}
+
+			static void set_right(node_ptr n, node_ptr right) {
+				n->lookup_.right_ = right;
+			}
+
+			static int get_color(const_node_ptr n) {
+				return n->lookup_.color_;
+			}
+
+			static void set_color(node_ptr n, color c) {
+				n->lookup_.color_ = c;
+			}
+
+			static color red() {
+				return 0;
+			}
+
+			static color black() {
+				return 1;
+			}
+		};
+
+		typedef boost::intrusive::circular_list_algorithms<list_node_traits> lru_algo;
+		typedef boost::intrusive::rbtree_algorithms<rbtree_node_traits> lookup_algo;
+
+		unsigned max_entries_;
+		unsigned current_entries_;
+
+		value_entry lru_header_;
+		value_entry lookup_header_;
+	};
+
+	template <typename ValueTraits>
+	cache<ValueTraits>::cache(unsigned max_entries)
+		: max_entries_(max_entries),
+		  current_entries_(0) {
+		lru_algo::init_header(&lru_header_);
+		lookup_algo::init_header(&lookup_header_);
+	}
+
+	template <typename ValueTraits>
+	cache<ValueTraits>::~cache() {
+		deleter<value_entry> d;
+		lookup_algo::clear_and_dispose(&lookup_header_, d);
+	}
+
+	template <typename ValueTraits>
+	void
+	cache<ValueTraits>::insert(value_type const &v) {
+		make_space();
+
+		value_entry *node = new value_entry(v);
+		try {
+			lru_algo::link_after(&lru_header_, node);
+			try {
+				value_ptr_cmp cmp;
+				lookup_algo::insert_equal(&lookup_header_, &lookup_header_, node, cmp);
+				current_entries_++;
+
+			} catch (...) {
+				lru_algo::unlink(node);
+				throw;
+			}
+
+		} catch (...) {
+			delete node;
+			throw;
+		}
+	}
+
+	template <typename ValueTraits>
+	boost::optional<typename ValueTraits::value_type>
+	cache<ValueTraits>::get(key_type const &k) {
+		key_value_ptr_cmp cmp;
+		value_entry *node = lookup_algo::find(&lookup_header_, k, cmp);
+		if (node == &lookup_header_)
+			return boost::optional<value_type>();
+
+		if (!node->ref_count_++)
+			lru_algo::unlink(node);
+		return boost::optional<value_type>(node->v_);
+	}
+
+	template <typename ValueTraits>
+	void
+	cache<ValueTraits>::put(value_type const &v) {
+		// FIXME: the lookup will go once we use a proper hook
+		key_value_ptr_cmp cmp;
+		key_type k = ValueTraits::get_key(v);
+		value_entry *node = lookup_algo::find(&lookup_header_, k, cmp);
+		if (node == &lookup_header_)
+			throw std::runtime_error("invalid put");
+
+		if (!--node->ref_count_)
+			lru_algo::link_after(&lru_header_, node);
+	}
+
+	template <typename ValueTraits>
+	void
+	cache<ValueTraits>::make_space() {
+		if (current_entries_ == max_entries_) {
+			value_entry *node = lru_header_.lru_.prev_;
+			if (node == &lru_header_)
+				throw std::runtime_error("cache full");
+
+			lru_algo::unlink(node);
+			lookup_algo::unlink(node);
+			delete node;
+			current_entries_--;
+		}
+	}
+}
+
+//----------------------------------------------------------------
+
+#endif

unit-tests/Makefile.in

@@ -1,6 +1,7 @@
 TEST_SOURCE=\
 	unit-tests/block_t.cc \
 	unit-tests/btree_t.cc \
+	unit-tests/cache_t.cc \
 	unit-tests/endian_t.cc \
 	unit-tests/space_map_t.cc \
 	unit-tests/space_map_disk_t.cc \
@@ -19,6 +20,9 @@ unit-tests/block_t: unit-tests/block_t.o
 unit-tests/btree_t: unit-tests/btree_t.o $(OBJECTS)
 	g++ $(CPPFLAGS) -o $@ $+ $(LIBS)
 
+unit-tests/cache_t: unit-tests/cache_t.o $(OBJECTS)
+	g++ $(CPPFLAGS) -o $@ $+ $(LIBS)
+
 unit-tests/space_map_t: unit-tests/space_map_t.o $(OBJECTS)
 	g++ $(CPPFLAGS) -o $@ $+ $(LIBS)
 

unit-tests/cache_t.cc

@@ -0,0 +1,144 @@
+#include "cache.h"
+
+#define BOOST_TEST_MODULE CacheTests
+#include <boost/test/included/unit_test.hpp>
+
+#include <boost/shared_ptr.hpp>
+
+using namespace boost;
+using namespace base;
+using namespace std;
+
+//----------------------------------------------------------------
+
+namespace {
+	struct Thing {
+		Thing()
+			: key_(0),
+			  desc_("default constructed") {
+		}
+
+		Thing(unsigned n)
+			: key_(n),
+			  desc_("foo bar") {
+		}
+
+		unsigned key_;
+		string desc_;
+	};
+
+	struct ThingTraits {
+		typedef unsigned key_type;
+		typedef Thing value_type;
+
+		static key_type get_key(value_type const &t) {
+			return t.key_;
+		}
+	};
+
+	struct SharedThingTraits {
+		typedef unsigned key_type;
+		typedef shared_ptr<Thing> value_type;
+
+		static key_type get_key(value_type const &p) {
+			return p->key_;
+		}
+	};
+}
+
+//----------------------------------------------------------------
+
+BOOST_AUTO_TEST_CASE(cache_creation)
+{
+	cache<ThingTraits> c(16);
+}
+
+BOOST_AUTO_TEST_CASE(cache_caches)
+{
+	unsigned const COUNT = 16;
+	cache<ThingTraits> c(COUNT);
+
+	for (unsigned i = 0; i < COUNT; i++)
+		c.insert(Thing(i));
+
+	for (unsigned i = 0; i < COUNT; i++)
+		BOOST_ASSERT(c.get(i));
+}
+
+BOOST_AUTO_TEST_CASE(cache_drops_elements)
+{
+	unsigned const COUNT = 1024;
+	unsigned const CACHE_SIZE = 16;
+	cache<ThingTraits> c(CACHE_SIZE);
+
+	for (unsigned i = 0; i < COUNT; i++)
+		c.insert(Thing(i));
+
+	for (unsigned i = 0; i < COUNT - CACHE_SIZE; i++)
+		BOOST_ASSERT(!c.get(i));
+
+	for (unsigned i = COUNT - CACHE_SIZE; i < COUNT; i++)
+		BOOST_ASSERT(c.get(i));
+}
+
+BOOST_AUTO_TEST_CASE(held_entries_count_towards_the_cache_limit)
+{
+	unsigned const CACHE_SIZE = 16;
+	cache<ThingTraits> c(CACHE_SIZE);
+
+	unsigned i;
+	for (i = 0; i < CACHE_SIZE; i++) {
+		c.insert(Thing(i));
+		c.get(i);
+	}
+
+	BOOST_CHECK_THROW(c.insert(Thing(i)), runtime_error);
+}
+
+BOOST_AUTO_TEST_CASE(put_works)
+{
+	unsigned const CACHE_SIZE = 16;
+	cache<ThingTraits> c(CACHE_SIZE);
+
+	unsigned i;
+	for (i = 0; i < CACHE_SIZE; i++) {
+		c.insert(Thing(i));
+		c.get(i);
+		c.put(Thing(i));
+	}
+
+	// should succeed
+	c.insert(Thing(i));
+}
+
+BOOST_AUTO_TEST_CASE(multiple_gets_works)
+{
+	unsigned const CACHE_SIZE = 16;
+	cache<ThingTraits> c(CACHE_SIZE);
+
+	unsigned i;
+	for (i = 0; i < CACHE_SIZE; i++) {
+		c.insert(Thing(i));
+		c.get(i);
+		c.get(i);
+		c.put(Thing(i));
+	}
+
+	BOOST_CHECK_THROW(c.insert(Thing(i)), runtime_error);
+}
+
+BOOST_AUTO_TEST_CASE(shared_ptr_cache_works)
+{
+	unsigned const CACHE_SIZE = 16;
+	cache<SharedThingTraits> c(CACHE_SIZE);
+
+	for (unsigned i = 0; i < CACHE_SIZE; i++) {
+		c.insert(shared_ptr<Thing>(new Thing(i)));
+		optional<shared_ptr<Thing> > maybe_ptr = c.get(i);
+
+		BOOST_ASSERT(maybe_ptr);
+		BOOST_ASSERT((*maybe_ptr)->key_ == i);
+	}
+}
+
+//----------------------------------------------------------------