libgit2 · vmg · Apr 22, 2013 · Apr 1, 2013 · Apr 1, 2013 · Apr 3, 2013
diff --git a/include/git2/common.h b/include/git2/common.h
@@ -131,8 +131,9 @@ enum {
 	GIT_OPT_SET_MWINDOW_MAPPED_LIMIT,
 	GIT_OPT_GET_SEARCH_PATH,
 	GIT_OPT_SET_SEARCH_PATH,
-	GIT_OPT_GET_ODB_CACHE_SIZE,
-	GIT_OPT_SET_ODB_CACHE_SIZE,
+	GIT_OPT_SET_CACHE_OBJECT_LIMIT,
+	GIT_OPT_SET_CACHE_MAX_SIZE,
+	GIT_OPT_ENABLE_CACHING
 };
 
 /**
@@ -169,15 +170,6 @@ enum {
  *		- `level` must be GIT_CONFIG_LEVEL_SYSTEM, GIT_CONFIG_LEVEL_GLOBAL,
  *		  or GIT_CONFIG_LEVEL_XDG.
  *
- *	opts(GIT_OPT_GET_ODB_CACHE_SIZE):
- *		Get the size of the libgit2 odb cache.
- *
- *	opts(GIT_OPT_SET_ODB_CACHE_SIZE):
- *		Set the size of the of the libgit2 odb cache. This needs
- *		to be done before git_repository_open is called, since
- *		git_repository_open initializes the odb layer. Defaults
- *		to 128.
- *
  * @param option Option key
  * @param ... value to set the option
  * @return 0 on success, <0 on failure

diff --git a/src/blob.c b/src/blob.c
@@ -18,32 +18,34 @@
 const void *git_blob_rawcontent(const git_blob *blob)
 {
 	assert(blob);
-	return blob->odb_object->raw.data;
+	return git_odb_object_data(blob->odb_object);
 }
 
 git_off_t git_blob_rawsize(const git_blob *blob)
 {
 	assert(blob);
-	return (git_off_t)blob->odb_object->raw.len;
+	return (git_off_t)git_odb_object_size(blob->odb_object);
 }
 
 int git_blob__getbuf(git_buf *buffer, git_blob *blob)
 {
 	return git_buf_set(
-		buffer, blob->odb_object->raw.data, blob->odb_object->raw.len);
+		buffer,
+		git_odb_object_data(blob->odb_object),
+		git_odb_object_size(blob->odb_object));
 }
 
-void git_blob__free(git_blob *blob)
+void git_blob__free(void *blob)
 {
-	git_odb_object_free(blob->odb_object);
+	git_odb_object_free(((git_blob *)blob)->odb_object);
 	git__free(blob);
 }
 
-int git_blob__parse(git_blob *blob, git_odb_object *odb_obj)
+int git_blob__parse(void *blob, git_odb_object *odb_obj)
 {
 	assert(blob);
 	git_cached_obj_incref((git_cached_obj *)odb_obj);
-	blob->odb_object = odb_obj;
+	((git_blob *)blob)->odb_object = odb_obj;
 	return 0;
 }
 
@@ -315,8 +317,8 @@ int git_blob_is_binary(git_blob *blob)
 
 	assert(blob);
 
-	content.ptr = blob->odb_object->raw.data;
-	content.size = min(blob->odb_object->raw.len, 4000);
+	content.ptr = blob->odb_object->buffer;
+	content.size = min(blob->odb_object->cached.size, 4000);
 
 	return git_buf_text_is_binary(&content);
 }
diff --git a/src/blob.h b/src/blob.h
@@ -17,8 +17,8 @@ struct git_blob {
 	git_odb_object *odb_object;
 };
 
-void git_blob__free(git_blob *blob);
-int git_blob__parse(git_blob *blob, git_odb_object *obj);
+void git_blob__free(void *blob);
+int git_blob__parse(void *blob, git_odb_object *obj);
 int git_blob__getbuf(git_buf *buffer, git_blob *blob);
 
 #endif
diff --git a/src/cache.c b/src/cache.c
@@ -11,100 +11,251 @@
 #include "thread-utils.h"
 #include "util.h"
 #include "cache.h"
+#include "odb.h"
+#include "object.h"
 #include "git2/oid.h"
 
-int git_cache_init(git_cache *cache, size_t size, git_cached_obj_freeptr free_ptr)
+GIT__USE_OIDMAP
+
+bool git_cache__enabled = true;
+size_t git_cache__max_storage = (4 * 1024 * 1024);
+
+static size_t git_cache__max_object_size[8] = {
+	0,     /* GIT_OBJ__EXT1 */
+	4096,  /* GIT_OBJ_COMMIT */
+	4096,  /* GIT_OBJ_TREE */
+	0,     /* GIT_OBJ_BLOB */
+	4096,  /* GIT_OBJ_TAG */
+	0,     /* GIT_OBJ__EXT2 */
+	0,     /* GIT_OBJ_OFS_DELTA */
+	0      /* GIT_OBJ_REF_DELTA */
+};
+
+int git_cache_set_max_object_size(git_otype type, size_t size)
 {
-	if (size < 8)
-		size = 8;
-	size = git__size_t_powerof2(size);
+	if (type < 0 || (size_t)type >= ARRAY_SIZE(git_cache__max_object_size)) {
+		giterr_set(GITERR_INVALID, "type out of range");
+		return -1;
+	}
+
+	git_cache__max_object_size[type] = size;
+	return 0;
+}
 
-	cache->size_mask = size - 1;
-	cache->lru_count = 0;
-	cache->free_obj = free_ptr;
+void git_cache_dump_stats(git_cache *cache)
+{
+	git_cached_obj *object;
 
-	git_mutex_init(&cache->lock);
+	if (kh_size(cache->map) == 0)
+		return;
 
-	cache->nodes = git__malloc(size * sizeof(git_cached_obj *));
-	GITERR_CHECK_ALLOC(cache->nodes);
+	printf("Cache %p: %d items cached, %d bytes\n",
+		cache, kh_size(cache->map), (int)cache->used_memory);
 
-	memset(cache->nodes, 0x0, size * sizeof(git_cached_obj *));
+	kh_foreach_value(cache->map, object, {
+		char oid_str[9];
+		printf(" %s%c %s (%d)\n",
+			git_object_type2string(object->type),
+			object->flags == GIT_CACHE_STORE_PARSED ? '*' : ' ',
+			git_oid_tostr(oid_str, sizeof(oid_str), &object->oid),
+			(int)object->size
+		);
+	});
+}
+
+int git_cache_init(git_cache *cache)
+{
+	cache->used_memory = 0;
+	cache->map = git_oidmap_alloc();
+	git_mutex_init(&cache->lock);
 	return 0;
 }
 
+/* called with lock */
+static void clear_cache(git_cache *cache)
+{
+	git_cached_obj *evict = NULL;
+
+	kh_foreach_value(cache->map, evict, {
+		git_cached_obj_decref(evict);
+	});
+
+	kh_clear(oid, cache->map);
+	cache->used_memory = 0;
+}
+
+void git_cache_clear(git_cache *cache)
+{
+	if (git_mutex_lock(&cache->lock) < 0)
+		return;
+
+	clear_cache(cache);
+
+	git_mutex_unlock(&cache->lock);
+}
+
 void git_cache_free(git_cache *cache)
 {
-	size_t i;
+	git_cache_clear(cache);
+
+	git_oidmap_free(cache->map);
+	git_mutex_free(&cache->lock);
+}
+
+/* Called with lock */
+static void cache_evict_entries(git_cache *cache)
+{
+	uint32_t seed = rand();
+	size_t evict_count = 8;
 
-	for (i = 0; i < (cache->size_mask + 1); ++i) {
-		if (cache->nodes[i] != NULL)
-			git_cached_obj_decref(cache->nodes[i], cache->free_obj);
+	/* do not infinite loop if there's not enough entries to evict  */
+	if (evict_count > kh_size(cache->map)) {
+		clear_cache(cache);
+		return;
 	}
 
-	git_mutex_free(&cache->lock);
-	git__free(cache->nodes);
+	while (evict_count > 0) {
+		khiter_t pos = seed++ % kh_end(cache->map);
+
+		if (kh_exist(cache->map, pos)) {
+			git_cached_obj *evict = kh_val(cache->map, pos);
+
+			evict_count--;
+			cache->used_memory -= evict->size;
+			git_cached_obj_decref(evict);
+
+			kh_del(oid, cache->map, pos);
+		}
+	}
 }
 
-void *git_cache_get(git_cache *cache, const git_oid *oid)
+static bool cache_should_store(git_otype object_type, size_t object_size)
 {
-	uint32_t hash;
-	git_cached_obj *node = NULL, *result = NULL;
+	size_t max_size = git_cache__max_object_size[object_type];
+	return git_cache__enabled && object_size < max_size;
+}
 
-	memcpy(&hash, oid->id, sizeof(hash));
+static void *cache_get(git_cache *cache, const git_oid *oid, unsigned int flags)
+{
+	khiter_t pos;
+	git_cached_obj *entry = NULL;
 
-	if (git_mutex_lock(&cache->lock)) {
-		giterr_set(GITERR_THREAD, "unable to lock cache mutex");
+	if (!git_cache__enabled || git_mutex_lock(&cache->lock) < 0)
 		return NULL;
-	}
 
-	{
-		node = cache->nodes[hash & cache->size_mask];
+	pos = kh_get(oid, cache->map, oid);
+	if (pos != kh_end(cache->map)) {
+		entry = kh_val(cache->map, pos);
 
-		if (node != NULL && git_oid_cmp(&node->oid, oid) == 0) {
-			git_cached_obj_incref(node);
-			result = node;
+		if (flags && entry->flags != flags) {
+			entry = NULL;
+		} else {
+			git_cached_obj_incref(entry);
 		}
 	}
+
 	git_mutex_unlock(&cache->lock);
 
-	return result;
+	return entry;
 }
 
-void *git_cache_try_store(git_cache *cache, void *_entry)
+static void *cache_store(git_cache *cache, git_cached_obj *entry)
 {
-	git_cached_obj *entry = _entry;
-	uint32_t hash;
+	khiter_t pos;
 
-	memcpy(&hash, &entry->oid, sizeof(uint32_t));
+	git_cached_obj_incref(entry);
 
-	if (git_mutex_lock(&cache->lock)) {
-		giterr_set(GITERR_THREAD, "unable to lock cache mutex");
-		return NULL;
-	}
+	if (!cache_should_store(entry->type, entry->size))
+		return entry;
 
-	{
-		git_cached_obj *node = cache->nodes[hash & cache->size_mask];
+	if (git_mutex_lock(&cache->lock) < 0)
+		return entry;
 
-		/* increase the refcount on this object, because
-		 * the cache now owns it */
-		git_cached_obj_incref(entry);
+	if (cache->used_memory > git_cache__max_storage)
+		cache_evict_entries(cache);
 
-		if (node == NULL) {
-			cache->nodes[hash & cache->size_mask] = entry;
-		} else if (git_oid_cmp(&node->oid, &entry->oid) == 0) {
-			git_cached_obj_decref(entry, cache->free_obj);
-			entry = node;
-		} else {
-			git_cached_obj_decref(node, cache->free_obj);
-			cache->nodes[hash & cache->size_mask] = entry;
+	pos = kh_get(oid, cache->map, &entry->oid);
+
+	/* not found */
+	if (pos == kh_end(cache->map)) {
+		int rval;
+
+		pos = kh_put(oid, cache->map, &entry->oid, &rval);
+		if (rval >= 0) {
+			kh_key(cache->map, pos) = &entry->oid;
+			kh_val(cache->map, pos) = entry;
+			git_cached_obj_incref(entry);
+			cache->used_memory += entry->size;
 		}
+	}
+	/* found */
+	else {
+		git_cached_obj *stored_entry = kh_val(cache->map, pos);
 
-		/* increase the refcount again, because we are
-		 * returning it to the user */
-		git_cached_obj_incref(entry);
+		if (stored_entry->flags == entry->flags) {
+			git_cached_obj_decref(entry);
+			git_cached_obj_incref(stored_entry);
+			entry = stored_entry;
+		} else if (stored_entry->flags == GIT_CACHE_STORE_RAW &&
+			entry->flags == GIT_CACHE_STORE_PARSED) {
+			git_cached_obj_decref(stored_entry);
+			git_cached_obj_incref(entry);
 
+			kh_key(cache->map, pos) = &entry->oid;
+			kh_val(cache->map, pos) = entry;
+		} else {
+			/* NO OP */
+		}
 	}
-	git_mutex_unlock(&cache->lock);
 
+	git_mutex_unlock(&cache->lock);
 	return entry;
 }
+
+void *git_cache_store_raw(git_cache *cache, git_odb_object *entry)
+{
+	entry->cached.flags = GIT_CACHE_STORE_RAW;
+	return cache_store(cache, (git_cached_obj *)entry);
+}
+
+void *git_cache_store_parsed(git_cache *cache, git_object *entry)
+{
+	entry->cached.flags = GIT_CACHE_STORE_PARSED;
+	return cache_store(cache, (git_cached_obj *)entry);
+}
+
+git_odb_object *git_cache_get_raw(git_cache *cache, const git_oid *oid)
+{
+	return cache_get(cache, oid, GIT_CACHE_STORE_RAW);
+}
+
+git_object *git_cache_get_parsed(git_cache *cache, const git_oid *oid)
+{
+	return cache_get(cache, oid, GIT_CACHE_STORE_PARSED);
+}
+
+void *git_cache_get_any(git_cache *cache, const git_oid *oid)
+{
+	return cache_get(cache, oid, GIT_CACHE_STORE_ANY);
+}
+
+void git_cached_obj_decref(void *_obj)
+{
+	git_cached_obj *obj = _obj;
+
+	if (git_atomic_dec(&obj->refcount) == 0) {
+		switch (obj->flags) {
+		case GIT_CACHE_STORE_RAW:
+			git_odb_object__free(_obj);
+			break;
+
+		case GIT_CACHE_STORE_PARSED:
+			git_object__free(_obj);
+			break;
+
+		default:
+			git__free(_obj);
+			break;
+		}
+	}
+}