From af710d3c65ca82f370980b3a603e8f4943bcd6a2 Mon Sep 17 00:00:00 2001
From: Vitaliy Filippov <vitalif@yourcmc.ru>
Date: Fri, 11 Jul 2025 02:18:52 +0300
Subject: [PATCH] Use emhash hashmap (2x faster)

---
 .gitmodules                        | 3 +++
 emhash                             | 1 +
 src/blockstore/blockstore_heap.cpp | 4 ++--
 src/blockstore/blockstore_heap.h   | 7 ++++---
 4 files changed, 10 insertions(+), 5 deletions(-)
 create mode 160000 emhash

diff --git a/.gitmodules b/.gitmodules
index c6636c20..22ccad4e 100644
--- a/.gitmodules
+++ b/.gitmodules
@@ -4,3 +4,6 @@
 [submodule "json11"]
 	path = json11
 	url = ../json11.git
+[submodule "emhash"]
+	path = emhash
+	url = ../emhash.git
diff --git a/emhash b/emhash
new file mode 160000
index 00000000..b7ff3147
--- /dev/null
+++ b/emhash
@@ -0,0 +1 @@
+Subproject commit b7ff3147a5b206d6fccd52cd3d63cdbfc4acf8fd
diff --git a/src/blockstore/blockstore_heap.cpp b/src/blockstore/blockstore_heap.cpp
index 702a8b63..471309fd 100644
--- a/src/blockstore/blockstore_heap.cpp
+++ b/src/blockstore/blockstore_heap.cpp
@@ -541,7 +541,7 @@ uint64_t blockstore_heap_t::load_blocks(uint64_t disk_offset, uint64_t size, uin
                 next_lsn = wr->lsn;
             }
         }
-        // btree_map<ui64, ui32> anyway stores std::pair<ui64, ui32>'s of 16 bytes size
+        // maps anyway store std::pair<ui64, ui32>'s of 16 bytes size
         // so we can store block_offset in it too
         block_index[get_pg_id(obj->inode, obj->stripe)][obj->inode][obj->stripe] = (uint8_t*)obj - buf + disk_offset;
         entries_loaded += wr_i;
@@ -818,7 +818,7 @@ void blockstore_heap_t::reshard(pool_id_t pool, uint32_t pg_count, uint32_t pg_s
         return;
     }
     uint64_t pool_id = (uint64_t)pool;
-    std::unordered_map<uint64_t, std::unordered_map<inode_t, std::unordered_map<uint64_t, uint64_t>>> new_shards;
+    emhash7::HashMap<uint64_t, emhash7::HashMap<inode_t, emhash5::HashMap<uint64_t, uint64_t>>> new_shards;
     auto sh_it = block_index.begin();
     while (sh_it != block_index.end())
     {
diff --git a/src/blockstore/blockstore_heap.h b/src/blockstore/blockstore_heap.h
index 705f299c..140734a4 100644
--- a/src/blockstore/blockstore_heap.h
+++ b/src/blockstore/blockstore_heap.h
@@ -11,7 +11,8 @@
 #include <vector>
 
 #include "../client/object_id.h"
-//#include "../../cpp-btree/btree_map.h"
+#include "../../emhash/hash_table5.hpp"
+#include "../../emhash/hash_table7.hpp"
 #include "blockstore_disk.h"
 #include "multilist.h"
 
@@ -137,9 +138,9 @@ class blockstore_heap_t
     uint32_t target_block_free_space = 800;
 
     uint64_t next_lsn = 0;
-    std::unordered_map<pool_id_t, pool_shard_settings_t> pool_shard_settings;
+    emhash7::HashMap<pool_id_t, pool_shard_settings_t> pool_shard_settings;
     // PG => inode => stripe => block number
-    std::unordered_map<uint64_t, std::unordered_map<inode_t, std::unordered_map<uint64_t, uint64_t>>> block_index;
+    emhash7::HashMap<uint64_t, emhash7::HashMap<inode_t, emhash5::HashMap<uint64_t, uint64_t>>> block_index;
     std::vector<heap_block_info_t> block_info;
     allocator_t *data_alloc = NULL;
     multilist_index_t *meta_alloc = NULL;