From 33168caf638660f90fd5d607cdb4b2267a30c707 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Luk=C3=A1=C5=A1=20Ondr=C3=A1=C4=8Dek?=
 <lukas.ondracek@nic.cz>
Date: Thu, 17 Jul 2025 01:35:46 +0200
Subject: [PATCH] lib/cache: decrease LMDB size by KRU size

---
 lib/cache/api.c | 11 ++++++++++-
 lib/cache/top.c | 33 +++++++++++++++++++++++++--------
 lib/cache/top.h |  3 +++
 3 files changed, 38 insertions(+), 9 deletions(-)

diff --git a/lib/cache/api.c b/lib/cache/api.c
index 737d38857..1677240be 100644
--- a/lib/cache/api.c
+++ b/lib/cache/api.c
@@ -116,6 +116,15 @@ int kr_cache_open(struct kr_cache *cache, const struct kr_cdb_api *api, struct k
 	if (!api)
 		api = kr_cdb_lmdb();
 	cache->api = api;
+
+	size_t orig_maxsize = opts->maxsize;
+	if (opts->maxsize) {
+		const size_t top_size = kr_cache_top_get_size(opts->maxsize);
+		if (!kr_fails_assert(top_size < opts->maxsize)) {
+			opts->maxsize -= top_size;
+		}
+	}
+
 	int ret = cache->api->open(&cache->db, &cache->stats, opts, mm);
 	if (ret == 0) {
 		ret = assert_right_version(cache);
@@ -161,7 +170,7 @@ int kr_cache_open(struct kr_cache *cache, const struct kr_cdb_api *api, struct k
 		ret = kr_error(errno);
 	}
 	if (ret == 0) {
-		ret = kr_cache_top_init(&cache->top, top_path, maxsize);
+		ret = kr_cache_top_init(&cache->top, top_path, orig_maxsize);
 		free(top_path);
 	}
 	if (ret != 0) {
diff --git a/lib/cache/top.c b/lib/cache/top.c
index 37cdd363b..a26e5d5b1 100644
--- a/lib/cache/top.c
+++ b/lib/cache/top.c
@@ -17,9 +17,10 @@
 #define FILE_FORMAT_VERSION 1  // fail if different
 
 
-#define KRU_CAPACITY(cache_size) (cache_size / 64)   // KRU size is approx. (8 * capacity) B
-	// average entry size seems to be 70-100 B
-	// -> KRU size is around 12.5 % of cache size, so 112.5 % of the set value is required
+#define KRU_CAPACITY(cache_size) (cache_size / 128)   // KRU size is approx. (8 * capacity) B
+	// average entry size seems to be 100-200 B,
+	// make KRU capacity between cache_size/128 and cache_size/64 (power of two)
+	// -> KRU size: between cache_size/16 and cache_size/8 (cache data size is the rest)
 #define TICK_SEC        1
 #define NORMAL_SIZE  (150 + KR_CACHE_SIZE_OVERHEAD)  // B; normal size of cache entry
 	// used as baseline for the following
@@ -93,13 +94,29 @@ static inline bool first_access(struct kr_cache_top_context *ctx, kru_hash_t has
 }
 
 
-int kr_cache_top_init(struct kr_cache_top *top, char *mmap_file, size_t cache_size) {
+static inline void get_size_capacity(size_t cache_size, size_t *top_size, size_t *capacity_log)
+{
+	*top_size = 0;
+	*capacity_log = 0;
+
+	const size_t capacity = KRU_CAPACITY(cache_size);
+	for (size_t c = capacity - 1; c > 0; c >>= 1) (*capacity_log)++;
+
+	*top_size = offsetof(struct top_data, kru) + KRU.get_size(*capacity_log);
+}
+
+int kr_cache_top_get_size(size_t cache_size)
+{
+	size_t top_size, capacity_log;
+	get_size_capacity(cache_size, &top_size, &capacity_log);
+	return top_size;
+}
+
+int kr_cache_top_init(struct kr_cache_top *top, char *mmap_file, size_t cache_size)
+{
 	size_t size = 0, capacity_log = 0;
 	if (cache_size > 0) {
-		const size_t capacity = KRU_CAPACITY(cache_size);
-		for (size_t c = capacity - 1; c > 0; c >>= 1) capacity_log++;
-
-		size = offsetof(struct top_data, kru) + KRU.get_size(capacity_log);
+		get_size_capacity(cache_size, &size, &capacity_log);
 	} // else use existing file settings
 
 	VERBOSE_LOG("INIT, cache size %d, KRU capacity_log %d\n", cache_size, capacity_log);
diff --git a/lib/cache/top.h b/lib/cache/top.h
index 8cdc17e91..c6eb920cb 100644
--- a/lib/cache/top.h
+++ b/lib/cache/top.h
@@ -33,6 +33,9 @@ static inline kru_price_t kr_cache_top_entry_price(struct kr_cache_top *top, siz
 	return top->data->base_price_norm / size;
 }
 
+KR_EXPORT
+int kr_cache_top_get_size(size_t cache_size);
+
 KR_EXPORT
 int kr_cache_top_init(struct kr_cache_top *top, char *mmap_file, size_t cache_size);
 
-- 
2.47.2