mmap tarfiles directly when mmapping is enabled, instead of copying data into separate mmapped block

Co-authored-by: Kajari Ghosh <ghoshkaj@gmail.com>

include/engine/datafacade/mmap_memory_allocator.hpp

@@ -10,6 +10,7 @@
 #include <boost/iostreams/device/mapped_file.hpp>
 
 #include <memory>
+#include <string>
 
 namespace osrm
 {
@@ -24,8 +25,7 @@ namespace datafacade
 class MMapMemoryAllocator : public ContiguousBlockAllocator
 {
   public:
-    explicit MMapMemoryAllocator(const storage::StorageConfig &config,
+    explicit MMapMemoryAllocator(const storage::StorageConfig &config);
-                                 const boost::filesystem::path &memory_file);
     ~MMapMemoryAllocator() override final;
 
     // interface to give access to the datafacades
@@ -33,8 +33,8 @@ class MMapMemoryAllocator : public ContiguousBlockAllocator
 
   private:
     storage::SharedDataIndex index;
-    util::vector_view<char> mapped_memory;
+    std::vector<boost::iostreams::mapped_file> mapped_memory_files;
-    boost::iostreams::mapped_file mapped_memory_file;
+    std::string rtree_filename;
 };
 
 } // namespace datafacade

include/engine/datafacade_provider.hpp

@@ -32,9 +32,8 @@ class ExternalProvider final : public DataFacadeProvider<AlgorithmT, FacadeT>
   public:
     using Facade = typename DataFacadeProvider<AlgorithmT, FacadeT>::Facade;
 
-    ExternalProvider(const storage::StorageConfig &config,
+    ExternalProvider(const storage::StorageConfig &config)
-                     const boost::filesystem::path &memory_file)
+        : facade_factory(std::make_shared<datafacade::MMapMemoryAllocator>(config))
-        : facade_factory(std::make_shared<datafacade::MMapMemoryAllocator>(config, memory_file))
     {
     }
 

include/storage/block.hpp

@@ -16,10 +16,15 @@ struct Block
 {
     std::uint64_t num_entries;
     std::uint64_t byte_size;
+    std::uint64_t offset;
 
-    Block() : num_entries(0), byte_size(0) {}
+    Block() : num_entries(0), byte_size(0), offset(0) {}
+    Block(std::uint64_t num_entries, std::uint64_t byte_size, std::uint64_t offset)
+        : num_entries(num_entries), byte_size(byte_size), offset(offset)
+    {
+    }
     Block(std::uint64_t num_entries, std::uint64_t byte_size)
-        : num_entries(num_entries), byte_size(byte_size)
+        : num_entries(num_entries), byte_size(byte_size), offset(0)
     {
     }
 };
@@ -29,7 +34,7 @@ using NamedBlock = std::tuple<std::string, Block>;
 template <typename T> Block make_block(uint64_t num_entries)
 {
     static_assert(sizeof(T) % alignof(T) == 0, "aligned T* can't be used as an array pointer");
-    return Block{num_entries, sizeof(T) * num_entries};
+    return Block{num_entries, sizeof(T) * num_entries, 0};
 }
 }
 }

include/storage/serialization.hpp

@@ -284,9 +284,9 @@ template <typename K, typename V> void write(io::BufferWriter &writer, const std
     }
 }
 
-inline void read(io::BufferReader &reader, DataLayout &layout) { read(reader, layout.blocks); }
+inline void read(io::BufferReader &reader, BaseDataLayout &layout) { read(reader, layout.blocks); }
 
-inline void write(io::BufferWriter &writer, const DataLayout &layout)
+inline void write(io::BufferWriter &writer, const BaseDataLayout &layout)
 {
     write(writer, layout.blocks);
 }

include/storage/shared_data_index.hpp

@@ -5,6 +5,7 @@
 
 #include <boost/function_output_iterator.hpp>
 
+#include <type_traits>
 #include <unordered_map>
 
 namespace osrm
@@ -19,8 +20,8 @@ class SharedDataIndex
   public:
     struct AllocatedRegion
     {
-        char *memory_ptr;
+        void *memory_ptr;
-        DataLayout layout;
+        std::unique_ptr<BaseDataLayout> layout;
     };
 
     SharedDataIndex() = default;
@@ -29,7 +30,7 @@ class SharedDataIndex
         // Build mapping from block name to region
         for (auto index : util::irange<std::uint32_t>(0, regions.size()))
         {
-            regions[index].layout.List("",
+            regions[index].layout->List("",
                                         boost::make_function_output_iterator([&](const auto &name) {
                                             block_to_region[name] = index;
                                         }));
@@ -40,32 +41,44 @@ class SharedDataIndex
     {
         for (const auto &region : regions)
         {
-            region.layout.List(name_prefix, out);
+            region.layout->List(name_prefix, out);
         }
     }
 
     template <typename T> auto GetBlockPtr(const std::string &name) const
     {
+#if !defined(__GNUC__) || (__GNUC__ > 4)
+        // is_tivially_copyable only exists in GCC >=5
+        static_assert(std::is_trivially_copyable<T>::value,
+                      "Block-based data must be a trivially copyable type");
+        static_assert(sizeof(T) % alignof(T) == 0, "aligned T* can't be used as an array pointer");
+#endif
         const auto &region = GetBlockRegion(name);
-        return region.layout.GetBlockPtr<T>(region.memory_ptr, name);
+        return reinterpret_cast<T *>(region.layout->GetBlockPtr(region.memory_ptr, name));
     }
 
     template <typename T> auto GetBlockPtr(const std::string &name)
     {
+#if !defined(__GNUC__) || (__GNUC__ > 4)
+        // is_tivially_copyable only exists in GCC >=5
+        static_assert(std::is_trivially_copyable<T>::value,
+                      "Block-based data must be a trivially copyable type");
+        static_assert(sizeof(T) % alignof(T) == 0, "aligned T* can't be used as an array pointer");
+#endif
         const auto &region = GetBlockRegion(name);
-        return region.layout.GetBlockPtr<T>(region.memory_ptr, name);
+        return reinterpret_cast<T *>(region.layout->GetBlockPtr(region.memory_ptr, name));
     }
 
     std::size_t GetBlockEntries(const std::string &name) const
     {
         const auto &region = GetBlockRegion(name);
-        return region.layout.GetBlockEntries(name);
+        return region.layout->GetBlockEntries(name);
     }
 
     std::size_t GetBlockSize(const std::string &name) const
     {
         const auto &region = GetBlockRegion(name);
-        return region.layout.GetBlockSize(name);
+        return region.layout->GetBlockSize(name);
     }
 
   private:

include/storage/shared_datatype.hpp

@@ -20,12 +20,12 @@ namespace osrm
 namespace storage
 {
 
-class DataLayout;
+class BaseDataLayout;
 namespace serialization
 {
-inline void read(io::BufferReader &reader, DataLayout &layout);
+inline void read(io::BufferReader &reader, BaseDataLayout &layout);
 
-inline void write(io::BufferWriter &writer, const DataLayout &layout);
+inline void write(io::BufferWriter &writer, const BaseDataLayout &layout);
 } // namespace serialization
 
 namespace detail
@@ -54,44 +54,28 @@ inline std::string trimName(const std::string &name_prefix, const std::string &n
 }
 } // namespace detail
 
-class DataLayout
+class BaseDataLayout
 {
   public:
-    DataLayout() : blocks{} {}
+    virtual ~BaseDataLayout() = default;
 
     inline void SetBlock(const std::string &name, Block block) { blocks[name] = std::move(block); }
 
-    inline uint64_t GetBlockEntries(const std::string &name) const
+    inline std::uint64_t GetBlockEntries(const std::string &name) const
     {
         return GetBlock(name).num_entries;
     }
 
-    inline uint64_t GetBlockSize(const std::string &name) const { return GetBlock(name).byte_size; }
+    inline std::uint64_t GetBlockSize(const std::string &name) const
+    {
+        return GetBlock(name).byte_size;
+    }
 
     inline bool HasBlock(const std::string &name) const
     {
         return blocks.find(name) != blocks.end();
     }
 
-    inline uint64_t GetSizeOfLayout() const
-    {
-        uint64_t result = 0;
-        for (const auto &name_and_block : blocks)
-        {
-            result += GetBlockSize(name_and_block.first) + BLOCK_ALIGNMENT;
-        }
-        return result;
-    }
-
-    template <typename T> inline T *GetBlockPtr(char *shared_memory, const std::string &name) const
-    {
-        static_assert(BLOCK_ALIGNMENT % std::alignment_of<T>::value == 0,
-                      "Datatype does not fit alignment constraints.");
-
-        char *ptr = (char *)GetAlignedBlockPtr(shared_memory, name);
-        return (T *)ptr;
-    }
-
     // Depending on the name prefix this function either lists all blocks with the same prefix
     // or all entries in the sub-directory.
     // '/ch/edge' -> '/ch/edge_filter/0/blocks', '/ch/edge_filter/1/blocks'
@@ -115,10 +99,10 @@ class DataLayout
         }
     }
 
-  private:
+    virtual inline void *GetBlockPtr(void *base_ptr, const std::string &name) const = 0;
-    friend void serialization::read(io::BufferReader &reader, DataLayout &layout);
+    virtual inline std::uint64_t GetSizeOfLayout() const = 0;
-    friend void serialization::write(io::BufferWriter &writer, const DataLayout &layout);
 
+  protected:
     const Block &GetBlock(const std::string &name) const
     {
         auto iter = blocks.find(name);
@@ -130,10 +114,42 @@ class DataLayout
         return iter->second;
     }
 
+    friend void serialization::read(io::BufferReader &reader, BaseDataLayout &layout);
+    friend void serialization::write(io::BufferWriter &writer, const BaseDataLayout &layout);
+
+    std::map<std::string, Block> blocks;
+};
+
+class ContiguousDataLayout final : public BaseDataLayout
+{
+  public:
+    inline std::uint64_t GetSizeOfLayout() const override final
+    {
+        std::uint64_t result = 0;
+        for (const auto &name_and_block : blocks)
+        {
+            result += GetBlockSize(name_and_block.first) + BLOCK_ALIGNMENT;
+        }
+        return result;
+    }
+
+    inline void *GetBlockPtr(void *base_ptr, const std::string &name) const override final
+    {
+        // TODO: re-enable this alignment checking somehow
+        // static_assert(BLOCK_ALIGNMENT % std::alignment_of<T>::value == 0,
+        //               "Datatype does not fit alignment constraints.");
+
+        return GetAlignedBlockPtr(base_ptr, name);
+    }
+
+  private:
+    friend void serialization::read(io::BufferReader &reader, BaseDataLayout &layout);
+    friend void serialization::write(io::BufferWriter &writer, const BaseDataLayout &layout);
+
     // Fit aligned storage in buffer to 64 bytes to conform with AVX 512 types
     inline void *align(void *&ptr) const noexcept
     {
-        const auto intptr = reinterpret_cast<uintptr_t>(ptr);
+        const auto intptr = reinterpret_cast<std::uintptr_t>(ptr);
         const auto aligned = (intptr - 1u + BLOCK_ALIGNMENT) & -BLOCK_ALIGNMENT;
         return ptr = reinterpret_cast<void *>(aligned);
     }
@@ -157,7 +173,27 @@ class DataLayout
     }
 
     static constexpr std::size_t BLOCK_ALIGNMENT = 64;
-    std::map<std::string, Block> blocks;
+};
+
+class TarDataLayout final : public BaseDataLayout
+{
+  public:
+    inline std::uint64_t GetSizeOfLayout() const override final
+    {
+        std::uint64_t result = 0;
+        for (const auto &name_and_block : blocks)
+        {
+            result += GetBlockSize(name_and_block.first);
+        }
+        return result;
+    }
+
+    inline void *GetBlockPtr(void *base_ptr, const std::string &name) const override final
+    {
+        auto offset = GetBlock(name).offset;
+        const auto offset_address = reinterpret_cast<std::uintptr_t>(base_ptr) + offset;
+        return reinterpret_cast<void *>(offset_address);
+    }
 };
 
 struct SharedRegion

include/storage/storage.hpp

@@ -35,22 +35,28 @@ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include <boost/filesystem/path.hpp>
 
 #include <string>
+#include <vector>
 
 namespace osrm
 {
 namespace storage
 {
+
+void populateLayoutFromFile(const boost::filesystem::path &path, storage::BaseDataLayout &layout);
+
 class Storage
 {
   public:
     Storage(StorageConfig config);
 
     int Run(int max_wait, const std::string &name, bool only_metric);
-
-    void PopulateStaticLayout(DataLayout &layout);
-    void PopulateUpdatableLayout(DataLayout &layout);
     void PopulateStaticData(const SharedDataIndex &index);
     void PopulateUpdatableData(const SharedDataIndex &index);
+    void PopulateLayout(storage::BaseDataLayout &layout,
+                        const std::vector<std::pair<bool, boost::filesystem::path>> &files);
+    std::string PopulateLayoutWithRTree(storage::BaseDataLayout &layout);
+    std::vector<std::pair<bool, boost::filesystem::path>> GetUpdatableFiles();
+    std::vector<std::pair<bool, boost::filesystem::path>> GetStaticFiles();
 
   private:
     StorageConfig config;

include/util/mmap_file.hpp

@@ -15,14 +15,14 @@ namespace util
 
 namespace detail
 {
-template <typename T, typename RegionT>
+template <typename T, typename MmapContainerT>
-util::vector_view<T> mmapFile(const boost::filesystem::path &file, RegionT &region)
+util::vector_view<T> mmapFile(const boost::filesystem::path &file, MmapContainerT &mmap_container)
 {
     try
     {
-        region.open(file);
+        mmap_container.open(file);
-        std::size_t num_objects = region.size() / sizeof(T);
+        std::size_t num_objects = mmap_container.size() / sizeof(T);
-        auto data_ptr = region.data();
+        auto data_ptr = mmap_container.data();
         BOOST_ASSERT(reinterpret_cast<uintptr_t>(data_ptr) % alignof(T) == 0);
         return util::vector_view<T>(reinterpret_cast<T *>(data_ptr), num_objects);
     }
@@ -34,9 +34,10 @@ util::vector_view<T> mmapFile(const boost::filesystem::path &file, RegionT &regi
     }
 }
 
-template <typename T, typename RegionT>
+template <typename T, typename MmapContainerT>
-util::vector_view<T>
+util::vector_view<T> mmapFile(const boost::filesystem::path &file,
-mmapFile(const boost::filesystem::path &file, RegionT &region, const std::size_t size)
+                              MmapContainerT &mmap_container,
+                              const std::size_t size)
 {
     try
     {
@@ -45,10 +46,10 @@ mmapFile(const boost::filesystem::path &file, RegionT &region, const std::size_t
         params.path = file.string();
         params.flags = boost::iostreams::mapped_file::readwrite;
         params.new_file_size = size;
-        region.open(params);
+        mmap_container.open(params);
 
         std::size_t num_objects = size / sizeof(T);
-        auto data_ptr = region.data();
+        auto data_ptr = mmap_container.data();
         BOOST_ASSERT(reinterpret_cast<uintptr_t>(data_ptr) % alignof(T) == 0);
         return util::vector_view<T>(reinterpret_cast<T *>(data_ptr), num_objects);
     }
@@ -63,24 +64,24 @@ mmapFile(const boost::filesystem::path &file, RegionT &region, const std::size_t
 
 template <typename T>
 util::vector_view<const T> mmapFile(const boost::filesystem::path &file,
-                                    boost::iostreams::mapped_file_source &region)
+                                    boost::iostreams::mapped_file_source &mmap_container)
 {
-    return detail::mmapFile<const T>(file, region);
+    return detail::mmapFile<const T>(file, mmap_container);
 }
 
 template <typename T>
 util::vector_view<T> mmapFile(const boost::filesystem::path &file,
-                              boost::iostreams::mapped_file &region)
+                              boost::iostreams::mapped_file &mmap_container)
 {
-    return detail::mmapFile<T>(file, region);
+    return detail::mmapFile<T>(file, mmap_container);
 }
 
 template <typename T>
 util::vector_view<T> mmapFile(const boost::filesystem::path &file,
-                              boost::iostreams::mapped_file &region,
+                              boost::iostreams::mapped_file &mmap_container,
                               std::size_t size)
 {
-    return detail::mmapFile<T>(file, region, size);
+    return detail::mmapFile<T>(file, mmap_container, size);
 }
 }
 }

src/engine/datafacade/mmap_memory_allocator.cpp

@@ -1,5 +1,6 @@
 #include "engine/datafacade/mmap_memory_allocator.hpp"
 
+#include "storage/block.hpp"
 #include "storage/io.hpp"
 #include "storage/serialization.hpp"
 #include "storage/storage.hpp"
@@ -7,7 +8,7 @@
 #include "util/log.hpp"
 #include "util/mmap_file.hpp"
 
-#include "boost/assert.hpp"
+#include <boost/assert.hpp>
 
 namespace osrm
 {
@@ -16,46 +17,50 @@ namespace engine
 namespace datafacade
 {
 
-MMapMemoryAllocator::MMapMemoryAllocator(const storage::StorageConfig &config,
+MMapMemoryAllocator::MMapMemoryAllocator(const storage::StorageConfig &config)
-                                         const boost::filesystem::path &memory_file)
 {
     storage::Storage storage(config);
+    std::vector<storage::SharedDataIndex::AllocatedRegion> allocated_regions;
 
-    if (!boost::filesystem::exists(memory_file))
     {
-        storage::DataLayout initial_layout;
+        std::unique_ptr<storage::BaseDataLayout> fake_layout =
-        storage.PopulateStaticLayout(initial_layout);
+            std::make_unique<storage::TarDataLayout>();
-        storage.PopulateUpdatableLayout(initial_layout);
 
-        auto data_size = initial_layout.GetSizeOfLayout();
+        // Convert the boost::filesystem::path object into a plain string
+        // that's stored as a member of this allocator object
+        rtree_filename = storage.PopulateLayoutWithRTree(*fake_layout);
 
-        storage::io::BufferWriter writer;
+        // Now, we add one more AllocatedRegion, with it's start address as the start
-        storage::serialization::write(writer, initial_layout);
+        // of the rtree_filename string we've saved.  In the fake_layout, we've
-        auto encoded_layout = writer.GetBuffer();
+        // stated that the data is at offset 0, which is where the string starts
-
+        // at it's own memory address.
-        auto total_size = data_size + encoded_layout.size();
+        // The syntax &(rtree_filename[0]) gets the memory address of the first char.
-
+        // We can't use the convenient `.data()` or `.c_str()` methods, because
-        mapped_memory = util::mmapFile<char>(memory_file, mapped_memory_file, total_size);
+        // prior to C++17 (which we're not using), those return a `const char *`,
-
+        // which isn't compatible with the `char *` that AllocatedRegion expects
-        std::copy(encoded_layout.begin(), encoded_layout.end(), mapped_memory.data());
+        // for it's memory_ptr
-
+        allocated_regions.push_back({&(rtree_filename[0]), std::move(fake_layout)});
-        index = storage::SharedDataIndex(
-            {{mapped_memory.data() + encoded_layout.size(), std::move(initial_layout)}});
-
-        storage.PopulateStaticData(index);
-        storage.PopulateUpdatableData(index);
     }
-    else
+
+    auto files = storage.GetStaticFiles();
+    auto updatable_files = storage.GetUpdatableFiles();
+    files.insert(files.end(), updatable_files.begin(), updatable_files.end());
+
+    for (const auto &file : files)
     {
-        mapped_memory = util::mmapFile<char>(memory_file, mapped_memory_file);
+        if (boost::filesystem::exists(file.second))
-
+        {
-        storage::DataLayout layout;
+            std::unique_ptr<storage::BaseDataLayout> layout =
-        storage::io::BufferReader reader(mapped_memory.data(), mapped_memory.size());
+                std::make_unique<storage::TarDataLayout>();
-        storage::serialization::read(reader, layout);
+            boost::iostreams::mapped_file mapped_memory_file;
-        auto layout_size = reader.GetPosition();
+            util::mmapFile<char>(file.second, mapped_memory_file);
-
+            mapped_memory_files.push_back(std::move(mapped_memory_file));
-        index = storage::SharedDataIndex({{mapped_memory.data() + layout_size, std::move(layout)}});
+            storage::populateLayoutFromFile(file.second, *layout);
+            allocated_regions.push_back({mapped_memory_file.data(), std::move(layout)});
         }
+    }
+
+    index = storage::SharedDataIndex{std::move(allocated_regions)};
 }
 
 MMapMemoryAllocator::~MMapMemoryAllocator() {}

src/engine/datafacade/process_memory_allocator.cpp

@@ -15,14 +15,20 @@ ProcessMemoryAllocator::ProcessMemoryAllocator(const storage::StorageConfig &con
     storage::Storage storage(config);
 
     // Calculate the layout/size of the memory block
-    storage::DataLayout layout;
+    auto static_files = storage.GetStaticFiles();
-    storage.PopulateStaticLayout(layout);
+    auto updatable_files = storage.GetUpdatableFiles();
-    storage.PopulateUpdatableLayout(layout);
+    std::unique_ptr<storage::BaseDataLayout> layout =
+        std::make_unique<storage::ContiguousDataLayout>();
+    storage.PopulateLayoutWithRTree(*layout);
+    storage.PopulateLayout(*layout, static_files);
+    storage.PopulateLayout(*layout, updatable_files);
 
     // Allocate the memory block, then load data from files into it
-    internal_memory = std::make_unique<char[]>(layout.GetSizeOfLayout());
+    internal_memory = std::make_unique<char[]>(layout->GetSizeOfLayout());
 
-    index = storage::SharedDataIndex({{internal_memory.get(), std::move(layout)}});
+    std::vector<storage::SharedDataIndex::AllocatedRegion> regions;
+    regions.push_back({internal_memory.get(), std::move(layout)});
+    index = {std::move(regions)};
 
     storage.PopulateStaticData(index);
     storage.PopulateUpdatableData(index);

src/engine/datafacade/shared_memory_allocator.cpp

@@ -25,8 +25,9 @@ SharedMemoryAllocator::SharedMemoryAllocator(
         auto mem = storage::makeSharedMemory(shm_key);
 
         storage::io::BufferReader reader(reinterpret_cast<char *>(mem->Ptr()), mem->Size());
-        storage::DataLayout layout;
+        std::unique_ptr<storage::BaseDataLayout> layout =
-        storage::serialization::read(reader, layout);
+            std::make_unique<storage::ContiguousDataLayout>();
+        storage::serialization::read(reader, *layout);
         auto layout_size = reader.GetPosition();
 
         regions.push_back({reinterpret_cast<char *>(mem->Ptr()) + layout_size, std::move(layout)});

src/storage/storage.cpp

@@ -44,24 +44,6 @@ namespace
 {
 using Monitor = SharedMonitor<SharedRegionRegister>;
 
-void readBlocks(const boost::filesystem::path &path, DataLayout &layout)
-{
-    tar::FileReader reader(path, tar::FileReader::VerifyFingerprint);
-
-    std::vector<tar::FileReader::FileEntry> entries;
-    reader.List(std::back_inserter(entries));
-
-    for (const auto &entry : entries)
-    {
-        const auto name_end = entry.name.rfind(".meta");
-        if (name_end == std::string::npos)
-        {
-            auto number_of_elements = reader.ReadElementCount64(entry.name);
-            layout.SetBlock(entry.name, Block{number_of_elements, entry.size});
-        }
-    }
-}
-
 struct RegionHandle
 {
     std::unique_ptr<SharedMemory> memory;
@@ -69,7 +51,8 @@ struct RegionHandle
     std::uint16_t shm_key;
 };
 
-auto setupRegion(SharedRegionRegister &shared_register, const DataLayout &layout)
+RegionHandle setupRegion(SharedRegionRegister &shared_register,
+                         const storage::BaseDataLayout &layout)
 {
     // This is safe because we have an exclusive lock for all osrm-datastore processes.
     auto shm_key = shared_register.ReserveKey();
@@ -184,6 +167,24 @@ bool swapData(Monitor &monitor,
 }
 }
 
+void populateLayoutFromFile(const boost::filesystem::path &path, storage::BaseDataLayout &layout)
+{
+    tar::FileReader reader(path, tar::FileReader::VerifyFingerprint);
+
+    std::vector<tar::FileReader::FileEntry> entries;
+    reader.List(std::back_inserter(entries));
+
+    for (const auto &entry : entries)
+    {
+        const auto name_end = entry.name.rfind(".meta");
+        if (name_end == std::string::npos)
+        {
+            auto number_of_elements = reader.ReadElementCount64(entry.name);
+            layout.SetBlock(entry.name, Block{number_of_elements, entry.size, entry.offset});
+        }
+    }
+}
+
 Storage::Storage(StorageConfig config_) : config(std::move(config_)) {}
 
 int Storage::Run(int max_wait, const std::string &dataset_name, bool only_metric)
@@ -243,29 +244,35 @@ int Storage::Run(int max_wait, const std::string &dataset_name, bool only_metric
         auto static_region = shared_register.GetRegion(region_id);
         auto static_memory = makeSharedMemory(static_region.shm_key);
 
-        DataLayout static_layout;
+        std::unique_ptr<storage::BaseDataLayout> static_layout =
+            std::make_unique<storage::ContiguousDataLayout>();
         io::BufferReader reader(reinterpret_cast<char *>(static_memory->Ptr()),
                                 static_memory->Size());
-        serialization::read(reader, static_layout);
+        serialization::read(reader, *static_layout);
         auto layout_size = reader.GetPosition();
         auto *data_ptr = reinterpret_cast<char *>(static_memory->Ptr()) + layout_size;
 
-        regions.push_back({data_ptr, static_layout});
+        regions.push_back({data_ptr, std::move(static_layout)});
         readonly_handles.push_back({std::move(static_memory), data_ptr, static_region.shm_key});
     }
     else
     {
-        DataLayout static_layout;
+        std::unique_ptr<storage::BaseDataLayout> static_layout =
-        PopulateStaticLayout(static_layout);
+            std::make_unique<storage::ContiguousDataLayout>();
-        auto static_handle = setupRegion(shared_register, static_layout);
+        Storage::PopulateLayoutWithRTree(*static_layout);
-        regions.push_back({static_handle.data_ptr, static_layout});
+        std::vector<std::pair<bool, boost::filesystem::path>> files = Storage::GetStaticFiles();
+        Storage::PopulateLayout(*static_layout, files);
+        auto static_handle = setupRegion(shared_register, *static_layout);
+        regions.push_back({static_handle.data_ptr, std::move(static_layout)});
         handles[dataset_name + "/static"] = std::move(static_handle);
     }
 
-    DataLayout updatable_layout;
+    std::unique_ptr<storage::BaseDataLayout> updatable_layout =
-    PopulateUpdatableLayout(updatable_layout);
+        std::make_unique<storage::ContiguousDataLayout>();
-    auto updatable_handle = setupRegion(shared_register, updatable_layout);
+    std::vector<std::pair<bool, boost::filesystem::path>> files = Storage::GetUpdatableFiles();
-    regions.push_back({updatable_handle.data_ptr, updatable_layout});
+    Storage::PopulateLayout(*updatable_layout, files);
+    auto updatable_handle = setupRegion(shared_register, *updatable_layout);
+    regions.push_back({updatable_handle.data_ptr, std::move(updatable_layout)});
     handles[dataset_name + "/updatable"] = std::move(updatable_handle);
 
     SharedDataIndex index{std::move(regions)};
@@ -281,24 +288,12 @@ int Storage::Run(int max_wait, const std::string &dataset_name, bool only_metric
     return EXIT_SUCCESS;
 }
 
-/**
+std::vector<std::pair<bool, boost::filesystem::path>> Storage::GetStaticFiles()
- * This function examines all our data files and figures out how much
- * memory needs to be allocated, and the position of each data structure
- * in that big block.  It updates the fields in the DataLayout parameter.
- */
-void Storage::PopulateStaticLayout(DataLayout &static_layout)
 {
-    {
-        auto absolute_file_index_path =
-            boost::filesystem::absolute(config.GetPath(".osrm.fileIndex"));
-
-        static_layout.SetBlock("/common/rtree/file_index_path",
-                               make_block<char>(absolute_file_index_path.string().length() + 1));
-    }
-
     constexpr bool REQUIRED = true;
     constexpr bool OPTIONAL = false;
-    std::vector<std::pair<bool, boost::filesystem::path>> tar_files = {
+
+    std::vector<std::pair<bool, boost::filesystem::path>> files = {
         {OPTIONAL, config.GetPath(".osrm.cells")},
         {OPTIONAL, config.GetPath(".osrm.partition")},
         {REQUIRED, config.GetPath(".osrm.icd")},
@@ -310,53 +305,73 @@ void Storage::PopulateStaticLayout(DataLayout &static_layout)
         {REQUIRED, config.GetPath(".osrm.maneuver_overrides")},
         {REQUIRED, config.GetPath(".osrm.edges")},
         {REQUIRED, config.GetPath(".osrm.names")},
-        {REQUIRED, config.GetPath(".osrm.ramIndex")},
+        {REQUIRED, config.GetPath(".osrm.ramIndex")}};
-    };
 
-    for (const auto &file : tar_files)
+    for (const auto &file : files)
     {
-        if (boost::filesystem::exists(file.second))
+        if (file.first == REQUIRED && !boost::filesystem::exists(file.second))
         {
-            readBlocks(file.second, static_layout);
+            throw util::exception("Could not find required filed: " + std::get<1>(file).string());
-        }
-        else
-        {
-            if (file.first == REQUIRED)
-            {
-                throw util::exception("Could not find required filed: " +
-                                      std::get<1>(file).string());
-            }
         }
     }
+
+    return files;
 }
 
-void Storage::PopulateUpdatableLayout(DataLayout &updatable_layout)
+std::vector<std::pair<bool, boost::filesystem::path>> Storage::GetUpdatableFiles()
 {
     constexpr bool REQUIRED = true;
     constexpr bool OPTIONAL = false;
-    std::vector<std::pair<bool, boost::filesystem::path>> tar_files = {
+
+    std::vector<std::pair<bool, boost::filesystem::path>> files = {
         {OPTIONAL, config.GetPath(".osrm.mldgr")},
         {OPTIONAL, config.GetPath(".osrm.cell_metrics")},
         {OPTIONAL, config.GetPath(".osrm.hsgr")},
         {REQUIRED, config.GetPath(".osrm.datasource_names")},
         {REQUIRED, config.GetPath(".osrm.geometry")},
         {REQUIRED, config.GetPath(".osrm.turn_weight_penalties")},
-        {REQUIRED, config.GetPath(".osrm.turn_duration_penalties")},
+        {REQUIRED, config.GetPath(".osrm.turn_duration_penalties")}};
-    };
 
-    for (const auto &file : tar_files)
+    for (const auto &file : files)
+    {
+        if (file.first == REQUIRED && !boost::filesystem::exists(file.second))
+        {
+            throw util::exception("Could not find required filed: " + std::get<1>(file).string());
+        }
+    }
+
+    return files;
+}
+
+std::string Storage::PopulateLayoutWithRTree(storage::BaseDataLayout &layout)
+{
+    // Figure out the path to the rtree file (it's not a tar file)
+    auto absolute_file_index_path = boost::filesystem::absolute(config.GetPath(".osrm.fileIndex"));
+
+    // Convert the boost::filesystem::path object into a plain string
+    // that can then be stored as a member of an allocator object
+    auto rtree_filename = absolute_file_index_path.string();
+
+    // Here, we hardcode the special file_index_path block name.
+    // The important bit here is that the "offset" is set to zero
+    layout.SetBlock("/common/rtree/file_index_path", make_block<char>(rtree_filename.length() + 1));
+
+    return rtree_filename;
+}
+
+/**
+ * This function examines all our data files and figures out how much
+ * memory needs to be allocated, and the position of each data structure
+ * in that big block.  It updates the fields in the layout parameter.
+ */
+void Storage::PopulateLayout(storage::BaseDataLayout &layout,
+                             const std::vector<std::pair<bool, boost::filesystem::path>> &files)
+{
+    for (const auto &file : files)
     {
         if (boost::filesystem::exists(file.second))
         {
-            readBlocks(file.second, updatable_layout);
+            populateLayoutFromFile(file.second, layout);
-        }
-        else
-        {
-            if (file.first == REQUIRED)
-            {
-                throw util::exception("Could not find required filed: " +
-                                      std::get<1>(file).string());
-            }
         }
     }
 }

src/tools/store.cpp

@@ -52,14 +52,14 @@ void listRegions(bool show_blocks)
             auto memory = makeSharedMemory(region.shm_key);
             io::BufferReader reader(reinterpret_cast<char *>(memory->Ptr()), memory->Size());
 
-            DataLayout layout;
+            std::unique_ptr<BaseDataLayout> layout = std::make_unique<ContiguousDataLayout>();
-            serialization::read(reader, layout);
+            serialization::read(reader, *layout);
 
             std::vector<std::string> block_names;
-            layout.List("", std::back_inserter(block_names));
+            layout->List("", std::back_inserter(block_names));
             for (auto &name : block_names)
             {
-                osrm::util::Log() << "  " << name << " " << layout.GetBlockSize(name);
+                osrm::util::Log() << "  " << name << " " << layout->GetBlockSize(name);
             }
         }
     }

unit_tests/storage/data_layout.cpp

@@ -15,86 +15,89 @@ using namespace osrm::storage;
 
 BOOST_AUTO_TEST_CASE(layout_write_test)
 {
-    DataLayout layout;
+    std::unique_ptr<BaseDataLayout> layout = std::make_unique<ContiguousDataLayout>();
 
     Block block_1{20, 8 * 20};
     Block block_2{1, 4 * 1};
     Block block_3{100, static_cast<std::uint64_t>(std::ceil(100 / 64.))};
 
-    layout.SetBlock("block1", block_1);
+    layout->SetBlock("block1", block_1);
-    layout.SetBlock("block2", block_2);
+    layout->SetBlock("block2", block_2);
-    layout.SetBlock("block3", block_3);
+    layout->SetBlock("block3", block_3);
 
     // Canary and alignment change layout size
-    BOOST_CHECK_GT(layout.GetSizeOfLayout(),
+    BOOST_CHECK_GT(layout->GetSizeOfLayout(),
                    block_1.byte_size + block_2.byte_size + block_3.byte_size);
 
-    BOOST_CHECK_EQUAL(layout.GetBlockSize("block1"), block_1.byte_size);
+    BOOST_CHECK_EQUAL(layout->GetBlockSize("block1"), block_1.byte_size);
-    BOOST_CHECK_EQUAL(layout.GetBlockSize("block2"), block_2.byte_size);
+    BOOST_CHECK_EQUAL(layout->GetBlockSize("block2"), block_2.byte_size);
-    BOOST_CHECK_EQUAL(layout.GetBlockSize("block3"), block_3.byte_size);
+    BOOST_CHECK_EQUAL(layout->GetBlockSize("block3"), block_3.byte_size);
 
-    std::vector<char> buffer(layout.GetSizeOfLayout());
+    std::vector<char> buffer(layout->GetSizeOfLayout());
     auto smallest_addr = buffer.data();
     auto biggest_addr = buffer.data() + buffer.size();
 
     {
-        auto block_1_ptr = layout.GetBlockPtr<std::uint64_t>(buffer.data(), "block1");
+        auto block_1_ptr =
-        auto block_2_ptr = layout.GetBlockPtr<std::uint32_t>(buffer.data(), "block2");
+            reinterpret_cast<std::uint64_t *>(layout->GetBlockPtr(buffer.data(), "block1"));
-        auto block_3_ptr = layout.GetBlockPtr<std::uint64_t>(buffer.data(), "block3");
+        auto block_2_ptr =
+            reinterpret_cast<std::uint32_t *>(layout->GetBlockPtr(buffer.data(), "block2"));
+        auto block_3_ptr =
+            reinterpret_cast<std::uint64_t *>(layout->GetBlockPtr(buffer.data(), "block3"));
 
-        BOOST_CHECK_LT(reinterpret_cast<std::size_t>(smallest_addr),
+        BOOST_CHECK_LE(reinterpret_cast<std::size_t>(smallest_addr),
                        reinterpret_cast<std::size_t>(block_1_ptr));
         BOOST_CHECK_GT(
             reinterpret_cast<std::size_t>(biggest_addr),
-            reinterpret_cast<std::size_t>(block_1_ptr + layout.GetBlockEntries("block1")));
+            reinterpret_cast<std::size_t>(block_1_ptr + layout->GetBlockEntries("block1")));
 
         BOOST_CHECK_LT(reinterpret_cast<std::size_t>(smallest_addr),
                        reinterpret_cast<std::size_t>(block_2_ptr));
         BOOST_CHECK_GT(
             reinterpret_cast<std::size_t>(biggest_addr),
-            reinterpret_cast<std::size_t>(block_2_ptr + layout.GetBlockEntries("block2")));
+            reinterpret_cast<std::size_t>(block_2_ptr + layout->GetBlockEntries("block2")));
 
         BOOST_CHECK_LT(reinterpret_cast<std::size_t>(smallest_addr),
                        reinterpret_cast<std::size_t>(block_3_ptr));
         BOOST_CHECK_GT(reinterpret_cast<std::size_t>(biggest_addr),
                        reinterpret_cast<std::size_t>(
                            block_3_ptr + static_cast<std::size_t>(
-                                             std::ceil(layout.GetBlockEntries("block3") / 64))));
+                                             std::ceil(layout->GetBlockEntries("block3") / 64))));
     }
 }
 
 BOOST_AUTO_TEST_CASE(layout_list_test)
 {
-    DataLayout layout;
+    std::unique_ptr<BaseDataLayout> layout = std::make_unique<ContiguousDataLayout>();
 
     Block block_1{20, 8 * 20};
     Block block_2{1, 4 * 1};
     Block block_3{100, static_cast<std::uint64_t>(std::ceil(100 / 64.))};
 
-    layout.SetBlock("/ch/edge_filter/block1", block_1);
+    layout->SetBlock("/ch/edge_filter/block1", block_1);
-    layout.SetBlock("/ch/edge_filter/block2", block_2);
+    layout->SetBlock("/ch/edge_filter/block2", block_2);
-    layout.SetBlock("/ch/edge_filter/block3", block_3);
+    layout->SetBlock("/ch/edge_filter/block3", block_3);
-    layout.SetBlock("/mld/metrics/0/durations", block_2);
+    layout->SetBlock("/mld/metrics/0/durations", block_2);
-    layout.SetBlock("/mld/metrics/0/weights", block_3);
+    layout->SetBlock("/mld/metrics/0/weights", block_3);
-    layout.SetBlock("/mld/metrics/1/durations", block_2);
+    layout->SetBlock("/mld/metrics/1/durations", block_2);
-    layout.SetBlock("/mld/metrics/1/weights", block_3);
+    layout->SetBlock("/mld/metrics/1/weights", block_3);
 
     std::vector<std::string> results_1;
     std::vector<std::string> results_2;
     std::vector<std::string> results_3;
-    layout.List("/ch/edge_filter", std::back_inserter(results_1));
+    layout->List("/ch/edge_filter", std::back_inserter(results_1));
-    layout.List("/ch/edge_filter/", std::back_inserter(results_2));
+    layout->List("/ch/edge_filter/", std::back_inserter(results_2));
-    layout.List("/ch/", std::back_inserter(results_3));
+    layout->List("/ch/", std::back_inserter(results_3));
 
     std::vector<std::string> results_4;
     std::vector<std::string> results_5;
     std::vector<std::string> results_6;
-    layout.List("/mld/metrics", std::back_inserter(results_4));
+    layout->List("/mld/metrics", std::back_inserter(results_4));
-    layout.List("/mld/metrics/", std::back_inserter(results_5));
+    layout->List("/mld/metrics/", std::back_inserter(results_5));
-    layout.List("/mld/", std::back_inserter(results_6));
+    layout->List("/mld/", std::back_inserter(results_6));
 
     std::vector<std::string> results_7;
-    layout.List("", std::back_inserter(results_7));
+    layout->List("", std::back_inserter(results_7));
     BOOST_CHECK_EQUAL(results_7.size(), 7);
 
     CHECK_EQUAL_RANGE(