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,
-                                 const boost::filesystem::path &memory_file);
+    explicit MMapMemoryAllocator(const storage::StorageConfig &config);
     ~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;
-    boost::iostreams::mapped_file mapped_memory_file;
+    std::vector<boost::iostreams::mapped_file> mapped_memory_files;
+    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,
-                     const boost::filesystem::path &memory_file)
-        : facade_factory(std::make_shared<datafacade::MMapMemoryAllocator>(config, memory_file))
+    ExternalProvider(const storage::StorageConfig &config)
+        : facade_factory(std::make_shared<datafacade::MMapMemoryAllocator>(config))
     {
     }
 

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;
-        DataLayout layout;
+        void *memory_ptr;
+        std::unique_ptr<BaseDataLayout> layout;
     };
 
     SharedDataIndex() = default;
@@ -29,10 +30,10 @@ class SharedDataIndex
         // Build mapping from block name to region
         for (auto index : util::irange<std::uint32_t>(0, regions.size()))
         {
-            regions[index].layout.List("",
-                                       boost::make_function_output_iterator([&](const auto &name) {
-                                           block_to_region[name] = index;
-                                       }));
+            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:
-    friend void serialization::read(io::BufferReader &reader, DataLayout &layout);
-    friend void serialization::write(io::BufferWriter &writer, const DataLayout &layout);
+    virtual inline void *GetBlockPtr(void *base_ptr, const std::string &name) const = 0;
+    virtual inline std::uint64_t GetSizeOfLayout() const = 0;
 
+  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>
-util::vector_view<T> mmapFile(const boost::filesystem::path &file, RegionT &region)
+template <typename T, typename MmapContainerT>
+util::vector_view<T> mmapFile(const boost::filesystem::path &file, MmapContainerT &mmap_container)
 {
     try
     {
-        region.open(file);
-        std::size_t num_objects = region.size() / sizeof(T);
-        auto data_ptr = region.data();
+        mmap_container.open(file);
+        std::size_t num_objects = mmap_container.size() / sizeof(T);
+        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>
-util::vector_view<T>
-mmapFile(const boost::filesystem::path &file, RegionT &region, const std::size_t size)
+template <typename T, typename MmapContainerT>
+util::vector_view<T> mmapFile(const boost::filesystem::path &file,
+                              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);
 }
 }
 }