use 2mb pages

Signed-off-by: Caio Oliveira <caiooliveirafarias0@gmail.com>

src/common/fs/file.cpp

@@ -7,18 +7,25 @@
 #include <vector>
 
 #include "common/assert.h"
+#include "common/bit_util.h"
 #include "common/fs/file.h"
 #include "common/fs/fs.h"
+#include "common/fs/fs_types.h"
 #ifdef __ANDROID__
 #include "common/fs/fs_android.h"
 #endif
 #include "common/logging.h"
+#include "common/literals.h"
 
 #ifdef _WIN32
 #include <io.h>
 #include <share.h>
+#include <windows.h>
 #else
 #include <unistd.h>
+#include <sys/mman.h>
+#include <sys/stat.h>
+#include <fcntl.h>
 #endif
 
 #ifdef _MSC_VER
@@ -244,42 +251,162 @@ FileType IOFile::GetType() const {
     return file_type;
 }
 
+#if defined(__unix__)
+static int PlatformMapReadOnly(IOFile& io, const char* path) {
+    io.mmap_fd = open(path, O_RDONLY);
+    if (io.mmap_fd > 0) {
+        struct stat st;
+        fstat(io.mmap_fd, &st);
+        io.mmap_size = st.st_size;
+
+        int map_flags = MAP_PRIVATE;
+#ifdef MAP_PREFAULT_READ
+        // Prefaults reads so the final resulting pagetable from this big stupid mmap()
+        // isn't comically lazily loaded, we just coalesce everything in-place for our
+        // lovely mmap flags; if we didn't prefault the reads the page table will be
+        // constructed in-place (i.e on a read-by-read basis) causing lovely soft-faults
+        // which would nuke any performance gains.
+        //
+        // This of course incurs a cost in the initial mmap(2) call, but that is fine.
+        map_flags |= MAP_PREFAULT_READ;
+#endif
+#ifdef MAP_NOSYNC
+        // This causes physical media to not be synched to our file/memory
+        // This means that if the read-only file is written to, we won't see changes
+        // or we may see changes which are just funnily scattered, in any case
+        // this presumes the files won't be changed during execution
+        //
+        // Do not ever use this on write files (if we ever support that); this will create
+        // a fun amount of fragmentation on the disk.
+        map_flags |= MAP_NOSYNC;
+#endif
+#if defined(NDEBUG) && defined(MAP_NOCORE)
+        map_flags |= MAP_NOCORE;
+#endif
+#ifdef MAP_HUGE_2MB
+        map_flags |= MAP_HUGE_2MB; //2mb pages
+#elif defined(MAP_ALIGNED)
+        // File must be big enough that it's worth to super align. We can't just super-align every
+        // file otherwise we will run out of alignments for actually important files :)
+        // System doesn't guarantee a super alignment, but if it's available it will delete
+        // about 3 layers(?) of the TLB tree for each read/write.
+        // Again the cost of faults may make this negligible gains, but hey, we gotta work
+        // what we gotta work with.
+        map_flags |= MAP_ALIGNED(21); //2^21 = 2mb use 2MB pages
+#endif
+        io.mmap_base = (u8*)mmap(nullptr, io.mmap_size, PROT_READ, map_flags, io.mmap_fd, 0);
+        if (io.mmap_base == MAP_FAILED) {
+            close(io.mmap_fd);
+            io.mmap_fd = -1;
+        } else {
+            using namespace Common::Literals;
+            // For small files it is acceptable to use a full readahead
+            // See https://github.com/torvalds/linux/blob/e80d033851b3bc94c3d254ac66660ddd0a49d72c/include/linux/pagemap.h#L1392
+            if (u64(st.st_size) >= 256_MiB) {
+                posix_madvise(io.mmap_base, io.mmap_size, POSIX_MADV_RANDOM);
+            } else {
+                posix_madvise(io.mmap_base, io.mmap_size, POSIX_MADV_SEQUENTIAL);
+            }
+        }
+    }
+    return io.mmap_fd;
+}
+static void PlatformUnmap(IOFile& io) {
+    if (io.mmap_fd != -1) {
+        munmap(io.mmap_base, io.mmap_size);
+        close(io.mmap_fd);
+        io.mmap_fd = -1;
+    }
+}
+#elif defined(__APPLE__)
+// NO IMPLEMENTATION YET
+#else
+static int PlatformMapReadOnly(IOFile& io, const wchar_t* path) {
+    io.file_handle = CreateFileW(path, GENERIC_READ, FILE_SHARE_READ, nullptr, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL | FILE_FLAG_SEQUENTIAL_SCAN, nullptr);
+    if (HANDLE(io.file_handle) != INVALID_HANDLE_VALUE) {
+        io.mapping_handle = CreateFileMappingW(HANDLE(io.file_handle), nullptr, PAGE_READONLY, 0, 0, nullptr);
+        if (io.mapping_handle) {
+            io.mmap_base = (u8*)MapViewOfFile(HANDLE(io.mapping_handle), FILE_MAP_READ, 0, 0, 0);
+            if (io.mmap_base) {
+                _LARGE_INTEGER pvalue;
+                GetFileSizeEx(io.file_handle, &pvalue);
+                io.mmap_size = uint32_t(pvalue.QuadPart);
+            } else {
+                CloseHandle(io.mapping_handle);
+                CloseHandle(io.file_handle);
+                return -1;
+            }
+        } else {
+            CloseHandle(io.file_handle);
+            return -1;
+        }
+    }
+    return 0;
+}
+static void PlatformUnmap(IOFile& io) {
+    if(io.mapping_handle) {
+        if(io.mmap_base)
+            UnmapViewOfFile(HANDLE(io.mmap_base));
+        CloseHandle(HANDLE(io.mapping_handle));
+    }
+    if(io.file_handle != INVALID_HANDLE_VALUE)
+        CloseHandle(HANDLE(io.file_handle));
+}
+#endif
+
 void IOFile::Open(const fs::path& path, FileAccessMode mode, FileType type, FileShareFlag flag) {
     Close();
-
     file_path = path;
     file_access_mode = mode;
     file_type = type;
-
     errno = 0;
-
 #ifdef _WIN32
+    // TODO: this probably can use better logic but oh well I'm not a windowser
+    file_handle = nullptr;
+    if (type == FileType::BinaryFile && mode == FileAccessMode::Read) {
+        if (PlatformMapReadOnly(*this, path.c_str()) == -1) {
+            LOG_ERROR(Common_Filesystem, "Error mmap'ing file"); //: {}", path.c_str());
+        }
+    }
+    if (file_handle == nullptr) {
         if (flag != FileShareFlag::ShareNone) {
             file = _wfsopen(path.c_str(), AccessModeToWStr(mode, type), ToWindowsFileShareFlag(flag));
         } else {
             _wfopen_s(&file, path.c_str(), AccessModeToWStr(mode, type));
         }
+    }
 #elif __ANDROID__
     if (Android::IsContentUri(path)) {
         ASSERT_MSG(mode == FileAccessMode::Read, "Content URI file access is for read-only!");
-        const auto fd = Android::OpenContentUri(path, Android::OpenMode::Read);
+        if (PlatformMapReadOnly(*this, path.c_str()) == -1) {
+            LOG_ERROR(Common_Filesystem, "Error mmap'ing file: {}", path.c_str());
+            int const fd = Android::OpenContentUri(path, Android::OpenMode::Read);
             if (fd != -1) {
                 file = fdopen(fd, "r");
-            const auto error_num = errno;
+                if (errno != 0 && file == nullptr)
-            if (error_num != 0 && file == nullptr) {
+                    LOG_ERROR(Common_Filesystem, "Error opening file: {}, error: {}", path.c_str(), strerror(errno));
-                LOG_ERROR(Common_Filesystem, "Error opening file: {}, error: {}", path.c_str(),
-                          strerror(error_num));
-            }
             } else {
                 LOG_ERROR(Common_Filesystem, "Error opening file: {}", path.c_str());
             }
+        }
     } else {
         file = std::fopen(path.c_str(), AccessModeToStr(mode, type));
     }
+#elif defined(__HAIKU__) || defined(__managarm__) || defined(__OPENORBIS__) || defined(__APPLE__)
+    file = std::fopen(path.c_str(), AccessModeToStr(mode, type));
+#elif defined(__unix__)
+    if (type == FileType::BinaryFile && mode == FileAccessMode::Read) {
+        if (PlatformMapReadOnly(*this, path.c_str()) == -1) {
+            LOG_ERROR(Common_Filesystem, "Error mmap'ing file: {}", path.c_str());
+        }
+    }
+    if (mmap_fd == -1) {
+        file = std::fopen(path.c_str(), AccessModeToStr(mode, type)); // mmap(2) failed or simply we can't use it
+    }
 #else
+    // Some other fancy OS (ahem fucking Darwin/Mac OSX)
     file = std::fopen(path.c_str(), AccessModeToStr(mode, type));
 #endif
-
     if (!IsOpen()) {
         const auto ec = std::error_code{errno, std::generic_category()};
         LOG_ERROR(Common_Filesystem, "Failed to open the file at path={}, ec_message={}",
@@ -288,25 +415,25 @@ void IOFile::Open(const fs::path& path, FileAccessMode mode, FileType type, File
 }
 
 void IOFile::Close() {
-    if (!IsOpen()) {
+#if defined(__APPLE__)
-        return;
+    // NO IMPLEMENTATION YET
-    }
+#else
-
+    PlatformUnmap(*this);
+#endif
+    if (file) {
         errno = 0;
-
         const auto close_result = std::fclose(file) == 0;
-
         if (!close_result) {
             const auto ec = std::error_code{errno, std::generic_category()};
             LOG_ERROR(Common_Filesystem, "Failed to close the file at path={}, ec_message={}",
                 PathToUTF8String(file_path), ec.message());
         }
-
         file = nullptr;
     }
+}
 
 bool IOFile::IsOpen() const {
-    return file != nullptr;
+    return file != nullptr || IsMappedFile();
 }
 
 std::string IOFile::ReadString(size_t length) const {
@@ -323,80 +450,65 @@ size_t IOFile::WriteString(std::span<const char> string) const {
 }
 
 bool IOFile::Flush() const {
-    if (!IsOpen()) {
+    ASSERT(!IsMappedFile());
-        return false;
+    if (file) {
-    }
-
         errno = 0;
-
+        auto const flush_result = std::fflush(file) == 0;
-#ifdef _WIN32
-    const auto flush_result = std::fflush(file) == 0;
-#else
-    const auto flush_result = std::fflush(file) == 0;
-#endif
-
         if (!flush_result) {
             const auto ec = std::error_code{errno, std::generic_category()};
             LOG_ERROR(Common_Filesystem, "Failed to flush the file at path={}, ec_message={}",
                 PathToUTF8String(file_path), ec.message());
         }
-
         return flush_result;
     }
-
-bool IOFile::Commit() const {
-    if (!IsOpen()) {
     return false;
 }
 
+bool IOFile::Commit() const {
+    ASSERT(!IsMappedFile());
+    if (file) {
         errno = 0;
-
 #ifdef _WIN32
         const auto commit_result = std::fflush(file) == 0 && _commit(fileno(file)) == 0;
 #else
         const auto commit_result = std::fflush(file) == 0 && fsync(fileno(file)) == 0;
 #endif
-
         if (!commit_result) {
             const auto ec = std::error_code{errno, std::generic_category()};
             LOG_ERROR(Common_Filesystem, "Failed to commit the file at path={}, ec_message={}",
                 PathToUTF8String(file_path), ec.message());
         }
-
         return commit_result;
     }
-
-bool IOFile::SetSize(u64 size) const {
-    if (!IsOpen()) {
     return false;
 }
 
+bool IOFile::SetSize(u64 size) const {
+    ASSERT(!IsMappedFile());
+    if (file) {
         errno = 0;
-
 #ifdef _WIN32
-    const auto set_size_result = _chsize_s(fileno(file), static_cast<s64>(size)) == 0;
+        const auto set_size_result = _chsize_s(fileno(file), s64(size)) == 0;
 #else
-    const auto set_size_result = ftruncate(fileno(file), static_cast<s64>(size)) == 0;
+        const auto set_size_result = ftruncate(fileno(file), s64(size)) == 0;
 #endif
-
         if (!set_size_result) {
             const auto ec = std::error_code{errno, std::generic_category()};
             LOG_ERROR(Common_Filesystem, "Failed to resize the file at path={}, size={}, ec_message={}",
                 PathToUTF8String(file_path), size, ec.message());
         }
-
         return set_size_result;
     }
+    return false;
+}
 
 u64 IOFile::GetSize() const {
-    if (!IsOpen()) {
+    if (IsMappedFile())
-        return 0;
+        return mmap_size;
-    }
+    if (file) {
-
+        // Flush any unwritten buffered data into the file prior to retrieving the file mmap_size.
-    // Flush any unwritten buffered data into the file prior to retrieving the file size.
         std::fflush(file);
-
+#if __ANDROID__
-#ifdef __ANDROID__
         u64 file_size = 0;
         if (Android::IsContentUri(file_path)) {
             file_size = Android::GetSize(file_path);
@@ -406,54 +518,64 @@ u64 IOFile::GetSize() const {
             file_size = fs::file_size(file_path, ec);
 
             if (ec) {
-            LOG_ERROR(Common_Filesystem,
+                LOG_ERROR(Common_Filesystem, "Failed to retrieve the file mmap_size of path={}, ec_message={}",
-                      "Failed to retrieve the file size of path={}, ec_message={}",
                     PathToUTF8String(file_path), ec.message());
                 return 0;
             }
         }
 #else
         std::error_code ec;
-
+        auto const file_size = fs::file_size(file_path, ec);
-    const auto file_size = fs::file_size(file_path, ec);
-
         if (ec) {
-        LOG_ERROR(Common_Filesystem, "Failed to retrieve the file size of path={}, ec_message={}",
+            LOG_ERROR(Common_Filesystem, "Failed to retrieve the file mmap_size of path={}, ec_message={}",
                 PathToUTF8String(file_path), ec.message());
             return 0;
         }
 #endif
-
         return file_size;
     }
-
-bool IOFile::Seek(s64 offset, SeekOrigin origin) const {
-    if (!IsOpen()) {
-        return false;
-    }
-
-    errno = 0;
-
-    const auto seek_result = fseeko(file, offset, ToSeekOrigin(origin)) == 0;
-
-    if (!seek_result) {
-        const auto ec = std::error_code{errno, std::generic_category()};
-        LOG_ERROR(Common_Filesystem,
-                  "Failed to seek the file at path={}, offset={}, origin={}, ec_message={}",
-                  PathToUTF8String(file_path), offset, origin, ec.message());
-    }
-
-    return seek_result;
-}
-
-s64 IOFile::Tell() const {
-    if (!IsOpen()) {
     return 0;
 }
 
+bool IOFile::Seek(s64 offset, SeekOrigin origin) const {
+    if (IsMappedFile()) {
+        // fuck you to whoever made this method const
+        switch (origin) {
+        case SeekOrigin::SetOrigin:
+            mmap_offset = s64(offset);
+            break;
+        case SeekOrigin::CurrentPosition:
+            mmap_offset += s64(offset);
+            break;
+        case SeekOrigin::End:
+            mmap_offset = s64(mmap_size) + s64(offset);
+            break;
+        }
+        return true;
+    }
+    if (file) {
         errno = 0;
+        const auto seek_result = fseeko(file, offset, ToSeekOrigin(origin)) == 0;
+        if (!seek_result) {
+            const auto ec = std::error_code{errno, std::generic_category()};
+            LOG_ERROR(Common_Filesystem, "Failed to seek the file at path={}, offset={}, origin={}, ec_message={}",
+                PathToUTF8String(file_path), offset, origin, ec.message());
+        }
+        return seek_result;
+    }
+    return false;
+}
 
+s64 IOFile::Tell() const {
+    if (IsMappedFile()) {
+        errno = 0;
+        return s64(mmap_offset);
+    }
+    if (file) {
+        errno = 0;
         return ftello(file);
     }
+    return 0;
+}
 
 } // namespace Common::FS

src/common/fs/file.h

@@ -7,6 +7,7 @@
 #pragma once
 
 #include <cstdio>
+#include <cstring>
 #include <filesystem>
 #include <span>
 #include <type_traits>
@@ -186,19 +187,6 @@ public:
             FileType type = FileType::BinaryFile,
             FileShareFlag flag = FileShareFlag::ShareReadOnly);
 
-// #ifdef _WIN32
-//     template <typename Path>
-//     void Open(const Path& path, FileAccessMode mode, FileType type = FileType::BinaryFile,
-//               FileShareFlag flag = FileShareFlag::ShareReadOnly) {
-//         using ValueType = typename Path::value_type;
-//         if constexpr (IsChar<ValueType>) {
-//             Open(ToU8String(path), mode, type, flag);
-//         } else {
-//             Open(std::filesystem::path{path}, mode, type, flag);
-//         }
-//     }
-// #endif
-
     /// Closes the file if it is opened.
     void Close();
 
@@ -228,8 +216,7 @@ public:
     [[nodiscard]] size_t Read(T& data) const {
         if constexpr (IsContiguousContainer<T>) {
             using ContiguousType = typename T::value_type;
-            static_assert(std::is_trivially_copyable_v<ContiguousType>,
+            static_assert(std::is_trivially_copyable_v<ContiguousType>, "Data type must be trivially copyable.");
-                          "Data type must be trivially copyable.");
             return ReadSpan<ContiguousType>(data);
         } else {
             return ReadObject(data) ? 1 : 0;
@@ -254,8 +241,7 @@ public:
     [[nodiscard]] size_t Write(const T& data) const {
         if constexpr (IsContiguousContainer<T>) {
             using ContiguousType = typename T::value_type;
-            static_assert(std::is_trivially_copyable_v<ContiguousType>,
+            static_assert(std::is_trivially_copyable_v<ContiguousType>, "Data type must be trivially copyable.");
-                          "Data type must be trivially copyable.");
             return WriteSpan<ContiguousType>(data);
         } else {
             static_assert(std::is_trivially_copyable_v<T>, "Data type must be trivially copyable.");
@@ -282,10 +268,11 @@ public:
     template <typename T>
         requires std::is_trivially_copyable_v<T>
     [[nodiscard]] size_t ReadSpan(std::span<T> data) const {
-        if (!IsOpen()) {
+        if (IsMappedFile()) {
-            return 0;
+            std::memcpy(data.data(), mmap_base + mmap_offset, sizeof(T) * data.size());
+            return data.size();
         }
-        return std::fread(data.data(), sizeof(T), data.size(), file);
+        return IsOpen() ? std::fread(data.data(), sizeof(T), data.size(), file) : 0;
     }
 
     /**
@@ -306,12 +293,11 @@ public:
     template <typename T>
     [[nodiscard]] size_t WriteSpan(std::span<const T> data) const {
         static_assert(std::is_trivially_copyable_v<T>, "Data type must be trivially copyable.");
-
+        if (IsMappedFile()) {
-        if (!IsOpen()) {
+            std::memcpy(mmap_base + mmap_offset, data.data(), sizeof(T) * data.size());
-            return 0;
+            return data.size();
         }
-
+        return IsOpen() ? std::fwrite(data.data(), sizeof(T), data.size(), file) : 0;
-        return std::fwrite(data.data(), sizeof(T), data.size(), file);
     }
 
     /**
@@ -334,12 +320,15 @@ public:
     [[nodiscard]] bool ReadObject(T& object) const {
         static_assert(std::is_trivially_copyable_v<T>, "Data type must be trivially copyable.");
         static_assert(!std::is_pointer_v<T>, "T must not be a pointer to an object.");
-
+        if (IsMappedFile()) {
-        if (!IsOpen()) {
+            std::memcpy(&object, mmap_base + mmap_offset, sizeof(T));
-            return false;
+#ifdef _WIN32
+            return bool(sizeof(T));
+#else
+            return sizeof(T);
+#endif
         }
-
+        return IsOpen() ? std::fread(&object, sizeof(T), 1, file) == 1 : false;
-        return std::fread(&object, sizeof(T), 1, file) == 1;
     }
 
     /**
@@ -361,12 +350,15 @@ public:
     [[nodiscard]] bool WriteObject(const T& object) const {
         static_assert(std::is_trivially_copyable_v<T>, "Data type must be trivially copyable.");
         static_assert(!std::is_pointer_v<T>, "T must not be a pointer to an object.");
-
+        if (IsMappedFile()) {
-        if (!IsOpen()) {
+            std::memcpy(mmap_base + mmap_offset, &object, sizeof(T));
-            return false;
+#ifdef _WIN32
+            return sizeof(T) != 0;
+#else
+            return sizeof(T);
+#endif
         }
-
+        return IsOpen() ? std::fwrite(&object, sizeof(T), 1, file) == 1 : false;
-        return std::fwrite(&object, sizeof(T), 1, file) == 1;
     }
 
     /**
@@ -449,12 +441,28 @@ public:
     */
     [[nodiscard]] s64 Tell() const;
 
-private:
+#ifdef _WIN32
+    inline bool IsMappedFile() const noexcept { return mapping_handle != nullptr; }
+#else // POSIX
+    inline bool IsMappedFile() const noexcept { return mmap_fd != -1; }
+#endif
+
     std::filesystem::path file_path;
     FileAccessMode file_access_mode{};
     FileType file_type{};
-
     std::FILE* file = nullptr;
+    // Any decent system should have mmap() for files
+    // Systems with artifical mmap() limitations should simply change the logic within file.cpp
+    // and reduce the threshold for which the mmap() is set to
+#ifdef _WIN32
+    void *mapping_handle = nullptr;
+    void *file_handle = nullptr;
+#else // POSIX
+    int mmap_fd = -1;
+#endif
+    u8* mmap_base = nullptr;
+    size_t mmap_size = 0;
+    mutable s64 mmap_offset = 0; // fuck you
 };
 
 } // namespace Common::FS