BlobRoute : Comparison and Hash can be made much cheaper without stri…

…ng formatting. (#66)

* Comparison and Hash can be made much cheaper without string formatting.

* Remove magic numbers

* Lift shard_mask

* Uplift make_new_shard_id

* Use boost::hash_value

src/lib/blob_route.hpp

@@ -0,0 +1,41 @@
+#include <functional>
+
+#include <boost/functional/hash.hpp>
+
+#include "homeobject/common.hpp"
+
+namespace homeobject {
+
+///
+// A Key used in the IndexService (BTree). The inclusion of Shard allows BlobRoutes
+// to appear in a different Index should the Blob (Shard) be moved between Pgs.
+struct BlobRoute {
+    shard_id shard;
+    blob_id blob;
+    auto operator<=>(BlobRoute const&) const = default;
+};
+
+} // namespace homeobject
+
+namespace fmt {
+template <>
+struct formatter< homeobject::BlobRoute > {
+    template < typename ParseContext >
+    constexpr auto parse(ParseContext& ctx) {
+        return ctx.begin();
+    }
+
+    template < typename FormatContext >
+    auto format(homeobject::BlobRoute const& r, FormatContext& ctx) {
+        return format_to(ctx.out(), "{:04x}:{:012x}:{:016x}", (r.shard >> homeobject::shard_width),
+                         (r.shard & homeobject::shard_mask), r.blob);
+    }
+};
+} // namespace fmt
+
+template <>
+struct std::hash< homeobject::BlobRoute > {
+    std::size_t operator()(homeobject::BlobRoute const& r) const noexcept {
+        return boost::hash_value< homeobject::blob_id >(std::make_pair(r.shard, r.blob));
+    }
+};

src/lib/homeobject_impl.hpp

@@ -13,6 +13,12 @@ class ReplicationService;
 
 namespace homeobject {
 
+constexpr size_t pg_width = sizeof(pg_id) * 8;
+constexpr size_t shard_width = (sizeof(shard_id) * 8) - pg_width;
+constexpr size_t shard_mask = std::numeric_limits< homeobject::shard_id >::max() >> pg_width;
+
+inline shard_id make_new_shard_id(pg_id pg, shard_id next_shard) { return ((uint64_t)pg << shard_width) | next_shard; }
+
 inline bool operator<(ShardInfo const& lhs, ShardInfo const& rhs) { return lhs.id < rhs.id; }
 inline bool operator==(ShardInfo const& lhs, ShardInfo const& rhs) { return lhs.id == rhs.id; }
 

src/lib/homestore/homeobject.hpp

@@ -27,7 +27,6 @@ class HSHomeObject : public HomeObjectImpl {
 
 private:
     shard_id generate_new_shard_id(pg_id pg);
-    shard_id make_new_shard_id(pg_id pg, uint64_t sequence_num) const;
     uint64_t get_sequence_num_from_shard_id(uint64_t shard_id);
     std::string serialize_shard(const Shard& shard) const;
     Shard deserialize_shard(const std::string& shard_info_str) const;

src/lib/homestore/shard_manager.cpp

@@ -7,11 +7,9 @@
 
 namespace homeobject {
 
-static constexpr uint32_t SEQUENCE_BIT_NUM_IN_SHARD{48};
-
 uint64_t ShardManager::max_shard_size() { return Gi; }
 
-uint64_t ShardManager::max_shard_num_in_pg() { return ((uint64_t)0x01) << SEQUENCE_BIT_NUM_IN_SHARD; }
+uint64_t ShardManager::max_shard_num_in_pg() { return ((uint64_t)0x01) << shard_width; }
 
 shard_id HSHomeObject::generate_new_shard_id(pg_id pg) {
     std::scoped_lock lock_guard(_pg_lock);
@@ -23,10 +21,6 @@ shard_id HSHomeObject::generate_new_shard_id(pg_id pg) {
     return make_new_shard_id(pg, new_sequence_num);
 }
 
-shard_id HSHomeObject::make_new_shard_id(pg_id pg, uint64_t sequence_num) const {
-    return ((uint64_t)pg << SEQUENCE_BIT_NUM_IN_SHARD) | sequence_num;
-}
-
 uint64_t HSHomeObject::get_sequence_num_from_shard_id(uint64_t shard_id) {
     return shard_id & (max_shard_num_in_pg() - 1);
 }
@@ -99,7 +93,8 @@ ShardManager::Result< ShardInfo > HSHomeObject::_create_shard(pg_id pg_owner, ui
     replica_set->write(sisl::blob(buf->data_begin(), needed_size), sisl::blob(), value, static_cast< void* >(&p));
     auto info = std::move(sf).get();
     if (!bool(info)) {
-        LOGWARN("create new shard [{}] on pg [{}] is failed with error:{}", new_shard_id, pg_owner, info.error());
+        LOGWARN("create new shard [{}] on pg [{}] is failed with error:{}", new_shard_id & shard_mask, pg_owner,
+                info.error());
     }
     header->~ReplicationMessageHeader();
     return info;

src/lib/memory/homeobject.hpp

@@ -7,44 +7,9 @@
 #include "mocks/repl_service.h"
 
 #include "lib/homeobject_impl.hpp"
+#include "lib/blob_route.hpp"
 
 namespace homeobject {
-struct BlobRoute {
-    shard_id shard;
-    blob_id blob;
-};
-} // namespace homeobject
-
-namespace fmt {
-template <>
-struct formatter< homeobject::BlobRoute > {
-    template < typename ParseContext >
-    constexpr auto parse(ParseContext& ctx) {
-        return ctx.begin();
-    }
-
-    template < typename FormatContext >
-    auto format(const homeobject::BlobRoute& r, FormatContext& ctx) {
-        return format_to(ctx.out(), "{}:{:012x}", r.shard, r.blob);
-    }
-};
-} // namespace fmt
-
-template <>
-struct std::hash< homeobject::BlobRoute > {
-    std::size_t operator()(homeobject::BlobRoute const& r) const noexcept {
-        return std::hash< std::string >()(fmt::format("{}", r));
-    }
-};
-
-namespace homeobject {
-
-inline bool operator<(BlobRoute const& lhs, BlobRoute const& rhs) {
-    return fmt::format("{}", lhs) < fmt::format("{}", rhs);
-}
-inline bool operator==(BlobRoute const& lhs, BlobRoute const& rhs) {
-    return fmt::format("{}", lhs) == fmt::format("{}", rhs);
-}
 
 ///
 // Used to TombStone Blob's in the Index to defer for GC.
@@ -55,13 +20,11 @@ struct BlobExt {
     Blob* _blob;
 
     explicit operator bool() const { return _state == BlobState::ALIVE; }
+    bool operator==(const BlobExt& rhs) const { return _blob == rhs._blob; }
 };
-inline bool operator==(BlobExt const& lhs, BlobExt const& rhs) { return lhs._blob == rhs._blob; }
-
-using btree = folly::ConcurrentHashMap< BlobRoute, BlobExt >;
 
 struct ShardIndex {
-    btree _btree;
+    folly::ConcurrentHashMap< BlobRoute, BlobExt > _btree;
     std::atomic< blob_id > _shard_seq_num{0ull};
     ~ShardIndex();
 };

src/lib/memory/shard_manager.cpp

@@ -15,9 +15,9 @@ ShardManager::Result< ShardInfo > MemoryHomeObject::_create_shard(pg_id pg_owner
         if (_pg_map.end() == pg_it) return folly::makeUnexpected(ShardError::UNKNOWN_PG);
 
         auto& s_list = pg_it->second.shards;
-        info.id = (((uint64_t)pg_owner) << 48) + s_list.size();
+        info.id = make_new_shard_id(pg_owner, s_list.size());
         auto iter = s_list.emplace(s_list.end(), Shard(info));
-        LOGDEBUG("Creating Shard [{}]: in Pg [{}] of Size [{}b]", info.id, pg_owner, size_bytes);
+        LOGDEBUG("Creating Shard [{}]: in Pg [{}] of Size [{}b]", info.id & shard_mask, pg_owner, size_bytes);
         auto [_, s_happened] = _shard_map.emplace(info.id, iter);
         RELEASE_ASSERT(s_happened, "Duplicate Shard insertion!");
     }