Merge pull request #2895 from kuzudb/inmem-file-fix

Share a single fileinfo for the hash index string overflow file
kuzudb · Feb 15, 2024 · 11b6964 · 11b6964
2 parents 7fc4519 + 00bc280
commit 11b6964
Show file tree

Hide file tree

Showing 3 changed files with 12 additions and 18 deletions.
diff --git a/src/include/storage/storage_structure/in_mem_file.h b/src/include/storage/storage_structure/in_mem_file.h
@@ -18,8 +18,8 @@ namespace storage {
 // InMemFile holds a collection of in-memory page in the memory.
 class InMemFile {
 public:
-    explicit InMemFile(std::string filePath, common::VirtualFileSystem* vfs,
-        std::atomic<common::page_idx_t>& pageCounter);
+    explicit InMemFile(
+        std::shared_ptr<common::FileInfo> fileInfo, std::atomic<common::page_idx_t>& pageCounter);
 
     void addANewPage();
 
@@ -45,8 +45,7 @@ class InMemFile {
 private:
     static const uint64_t END_OF_PAGE =
         common::BufferPoolConstants::PAGE_4KB_SIZE - sizeof(common::page_idx_t);
-    std::string filePath;
-    std::unique_ptr<common::FileInfo> fileInfo;
+    std::shared_ptr<common::FileInfo> fileInfo;
     std::unordered_map<common::page_idx_t, std::unique_ptr<InMemPage>> pages;
     PageCursor nextPosToAppend;
     std::atomic<common::page_idx_t>& pageCounter;

diff --git a/src/storage/index/hash_index_builder.cpp b/src/storage/index/hash_index_builder.cpp
@@ -1,5 +1,7 @@
 #include "storage/index/hash_index_builder.h"
 
+#include <fcntl.h>
+
 #include <cstring>
 
 #include "common/type_utils.h"
@@ -263,10 +265,12 @@ PrimaryKeyIndexBuilder::PrimaryKeyIndexBuilder(
         keyDataTypeID,
         [&]<IndexHashable T>(T) {
             if constexpr (std::is_same_v<T, ku_string_t>) {
+                auto overflowFileInfo = std::shared_ptr(vfs->openFile(
+                    StorageUtils::getOverflowFileName(fileHandle->getFileInfo()->path),
+                    O_CREAT | O_WRONLY));
                 for (auto i = 0u; i < NUM_HASH_INDEXES; i++) {
-                    auto overflowFile = std::make_unique<InMemFile>(
-                        StorageUtils::getOverflowFileName(fileHandle->getFileInfo()->path), vfs,
-                        overflowPageCounter);
+                    auto overflowFile =
+                        std::make_unique<InMemFile>(overflowFileInfo, overflowPageCounter);
                     hashIndexBuilders.push_back(
                         std::make_unique<HashIndexBuilder<std::string_view, ku_string_t>>(
                             fileHandle, std::move(overflowFile), i, keyDataType));

diff --git a/src/storage/storage_structure/in_mem_file.cpp b/src/storage/storage_structure/in_mem_file.cpp
@@ -1,11 +1,8 @@
 #include "storage/storage_structure/in_mem_file.h"
 
-#include <fcntl.h>
-
 #include "common/constants.h"
 #include "common/exception/copy.h"
 #include "common/exception/message.h"
-#include "common/file_system/virtual_file_system.h"
 #include "common/type_utils.h"
 #include "common/types/ku_string.h"
 #include "common/types/types.h"
@@ -15,11 +12,8 @@ using namespace kuzu::common;
 namespace kuzu {
 namespace storage {
 
-InMemFile::InMemFile(
-    std::string filePath, common::VirtualFileSystem* vfs, std::atomic<page_idx_t>& pageCounter)
-    : filePath{std::move(filePath)}, nextPosToAppend(0, 0), pageCounter(pageCounter) {
-    fileInfo = vfs->openFile(this->filePath, O_CREAT | O_WRONLY);
-}
+InMemFile::InMemFile(std::shared_ptr<FileInfo> fileInfo, std::atomic<page_idx_t>& pageCounter)
+    : fileInfo{fileInfo}, nextPosToAppend(0, 0), pageCounter(pageCounter) {}
 
 void InMemFile::addANewPage() {
     page_idx_t newPageIdx = pageCounter.fetch_add(1);
@@ -122,9 +116,6 @@ bool InMemFile::equals(std::string_view keyToLookup, const ku_string_t& keyInEnt
 }
 
 void InMemFile::flush() {
-    if (filePath.empty()) {
-        throw CopyException("InMemPages: Empty filename");
-    }
     for (auto& [pageIndex, page] : pages) {
         // actual page index is stored inside of the InMemPage and does not correspond to the index
         // in the vector