Change hash index concurrent build

CMakeLists.txt

@@ -1,6 +1,6 @@
 cmake_minimum_required(VERSION 3.11)
 
-project(Kuzu VERSION 0.0.3.4 LANGUAGES CXX)
+project(Kuzu VERSION 0.0.3.5 LANGUAGES CXX)
 
 find_package(Threads REQUIRED)
 

src/binder/bound_statement_result.cpp

@@ -7,7 +7,7 @@ namespace binder {
 
 std::unique_ptr<BoundStatementResult> BoundStatementResult::createSingleStringColumnResult() {
     auto result = std::make_unique<BoundStatementResult>();
-    auto columnName = std::string("outputMsg");
+    auto columnName = std::string("result");
     auto value = std::make_unique<common::Value>(columnName);
     auto stringColumn = std::make_shared<LiteralExpression>(std::move(value), columnName);
     result->addColumn(stringColumn, expression_vector{stringColumn});

src/include/storage/index/hash_index_builder.h

@@ -107,17 +107,6 @@ class HashIndexBuilder : public BaseHashIndex {
     uint32_t allocatePSlots(uint32_t numSlotsToAllocate);
     uint32_t allocateAOSlot();
 
-    inline void lockSlot(SlotInfo& slotInfo) {
-        assert(slotInfo.slotType == SlotType::PRIMARY);
-        std::shared_lock sLck{pSlotSharedMutex};
-        pSlotsMutexes[slotInfo.slotId]->lock();
-    }
-    inline void unlockSlot(const SlotInfo& slotInfo) {
-        assert(slotInfo.slotType == SlotType::PRIMARY);
-        std::shared_lock sLck{pSlotSharedMutex};
-        pSlotsMutexes[slotInfo.slotId]->unlock();
-    }
-
 private:
     std::unique_ptr<FileHandle> fileHandle;
     std::unique_ptr<InMemDiskArrayBuilder<HashIndexHeader>> headerArray;
@@ -151,6 +140,10 @@ class PrimaryKeyIndexBuilder {
         }
     }
 
+    inline void lock() { mtx.lock(); }
+
+    inline void unlock() { mtx.unlock(); }
+
     inline void bulkReserve(uint32_t numEntries) {
         keyDataTypeID == common::LogicalTypeID::INT64 ?
             hashIndexBuilderForInt64->bulkReserve(numEntries) :
@@ -189,6 +182,7 @@ class PrimaryKeyIndexBuilder {
     }
 
 private:
+    std::mutex mtx;
     common::LogicalTypeID keyDataTypeID;
     std::unique_ptr<HashIndexBuilder<int64_t>> hashIndexBuilderForInt64;
     std::unique_ptr<HashIndexBuilder<common::ku_string_t>> hashIndexBuilderForString;

src/include/storage/storage_info.h

@@ -12,7 +12,8 @@ using storage_version_t = uint64_t;
 
 struct StorageVersionInfo {
     static std::unordered_map<std::string, storage_version_t> getStorageVersionInfo() {
-        return {{"0.0.3.4", 5}, {"0.0.3.3", 4}, {"0.0.3.2", 3}, {"0.0.3.1", 2}, {"0.0.3", 1}};
+        return {{"0.0.3.5", 6}, {"0.0.3.4", 5}, {"0.0.3.3", 4}, {"0.0.3.2", 3}, {"0.0.3.1", 2},
+            {"0.0.3", 1}};
     }
 
     static storage_version_t getStorageVersion();

src/storage/copier/node_copier.cpp

@@ -100,6 +100,7 @@ void NodeCopier::populatePKIndex(InMemColumnChunk* chunk, InMemOverflowFile* ove
         }
     }
     // No nulls, so we can populate the index with actual values.
+    pkIndex->lock();
     switch (chunk->getDataType().getLogicalTypeID()) {
     case LogicalTypeID::INT64: {
         appendToPKIndex<int64_t>(chunk, startOffset, numValues);
@@ -112,6 +113,7 @@ void NodeCopier::populatePKIndex(InMemColumnChunk* chunk, InMemOverflowFile* ove
         throw CopyException("Primary key must be either INT64, STRING or SERIAL.");
     }
     }
+    pkIndex->unlock();
 }
 
 template<>

src/storage/copier/node_copy_executor.cpp

@@ -27,7 +27,6 @@ static column_id_t getPKColumnID(
 }
 
 void NodeCopyExecutor::populateColumns(processor::ExecutionContext* executionContext) {
-    logger->info("Populating properties");
     auto primaryKey = reinterpret_cast<NodeTableSchema*>(tableSchema)->getPrimaryKey();
     std::unique_ptr<PrimaryKeyIndexBuilder> pkIndex;
     if (primaryKey.dataType.getLogicalTypeID() != common::LogicalTypeID::SERIAL) {
@@ -80,7 +79,6 @@ void NodeCopyExecutor::populateColumns(processor::ExecutionContext* executionCon
     for (auto& task : tasks) {
         taskScheduler.scheduleTaskAndWaitOrError(task, executionContext);
     }
-    logger->info("Done populating properties, constructing the pk index.");
 }
 
 } // namespace storage

src/storage/index/hash_index_builder.cpp

@@ -62,12 +62,10 @@ bool HashIndexBuilder<T>::appendInternal(const uint8_t* key, offset_t value) {
     SlotInfo pSlotInfo{getPrimarySlotIdForKey(*indexHeader, key), SlotType::PRIMARY};
     auto currentSlotInfo = pSlotInfo;
     Slot<T>* currentSlot = nullptr;
-    lockSlot(pSlotInfo);
     while (currentSlotInfo.slotType == SlotType::PRIMARY || currentSlotInfo.slotId != 0) {
         currentSlot = getSlot(currentSlotInfo);
         if (lookupOrExistsInSlotWithoutLock<false /* exists */>(currentSlot, key)) {
             // Key already exists. No append is allowed.
-            unlockSlot(pSlotInfo);
             return false;
         }
         if (currentSlot->header.numEntries < HashIndexConstants::SLOT_CAPACITY) {
@@ -78,7 +76,6 @@ bool HashIndexBuilder<T>::appendInternal(const uint8_t* key, offset_t value) {
     }
     assert(currentSlot);
     insertToSlotWithoutLock(currentSlot, key, value);
-    unlockSlot(pSlotInfo);
     numEntries.fetch_add(1);
     return true;
 }
@@ -101,20 +98,14 @@ bool HashIndexBuilder<T>::lookupInternalWithoutLock(const uint8_t* key, offset_t
 
 template<typename T>
 uint32_t HashIndexBuilder<T>::allocatePSlots(uint32_t numSlotsToAllocate) {
-    std::unique_lock xLock{pSlotSharedMutex};
     auto oldNumSlots = pSlots->getNumElements();
     auto newNumSlots = oldNumSlots + numSlotsToAllocate;
     pSlots->resize(newNumSlots, true /* setToZero */);
-    pSlotsMutexes.resize(newNumSlots);
-    for (auto i = oldNumSlots; i < newNumSlots; i++) {
-        pSlotsMutexes[i] = std::make_unique<std::mutex>();
-    }
     return oldNumSlots;
 }
 
 template<typename T>
 uint32_t HashIndexBuilder<T>::allocateAOSlot() {
-    std::unique_lock xLock{oSlotsSharedMutex};
     auto oldNumSlots = oSlots->getNumElements();
     auto newNumSlots = oldNumSlots + 1;
     oSlots->resize(newNumSlots, true /* setToZero */);
@@ -124,10 +115,8 @@ uint32_t HashIndexBuilder<T>::allocateAOSlot() {
 template<typename T>
 Slot<T>* HashIndexBuilder<T>::getSlot(const SlotInfo& slotInfo) {
     if (slotInfo.slotType == SlotType::PRIMARY) {
-        std::shared_lock sLck{pSlotSharedMutex};
         return &pSlots->operator[](slotInfo.slotId);
     } else {
-        std::shared_lock sLck{oSlotsSharedMutex};
         return &oSlots->operator[](slotInfo.slotId);
     }
 }