result: Check against self-assignment in ResultVal's copy assignment operator

Avoids doing work that doesn't need to be done.

.appveyor/UtilityFunctions.ps1

@@ -0,0 +1,39 @@
+# Set-up Visual Studio Command Prompt environment for PowerShell
+pushd "C:\Program Files (x86)\Microsoft Visual Studio\2017\Community\Common7\Tools\"
+cmd /c "VsDevCmd.bat -arch=x64 & set" | foreach {
+    if ($_ -match "=") {
+        $v = $_.split("="); Set-Item -Force -Path "ENV:\$($v[0])" -Value "$($v[1])"
+    }
+}
+popd
+
+function Which ($search_path, $name) {
+    ($search_path).Split(";") | Get-ChildItem -Filter $name | Select -First 1 -Exp FullName
+}
+
+function GetDeps ($search_path, $binary) {
+    ((dumpbin /dependents $binary).Where({ $_ -match "dependencies:"}, "SkipUntil") | Select-String "[^ ]*\.dll").Matches | foreach {
+        Which $search_path $_.Value
+    }
+}
+
+function RecursivelyGetDeps ($search_path, $binary) {
+    $final_deps = @()
+    $deps_to_process = GetDeps $search_path $binary
+    while ($deps_to_process.Count -gt 0) {
+        $current, $deps_to_process = $deps_to_process
+        if ($final_deps -contains $current) { continue }
+
+        # Is this a system dll file?
+        # We use the same algorithm that cmake uses to determine this.
+        if ($current -match "$([regex]::Escape($env:SystemRoot))\\sys") { continue }
+        if ($current -match "$([regex]::Escape($env:WinDir))\\sys") { continue }
+        if ($current -match "\\msvc[^\\]+dll") { continue }
+        if ($current -match "\\api-ms-win-[^\\]+dll") { continue }
+
+        $final_deps += $current
+        $new_deps = GetDeps $search_path $current
+        $deps_to_process += ($new_deps | ?{-not ($final_deps -contains $_)})
+    }
+    return $final_deps
+}

.gitmodules

@@ -18,7 +18,7 @@
     url = https://github.com/fmtlib/fmt.git
 [submodule "lz4"]
     path = externals/lz4
-    url = http://github.com/lz4/lz4.git
+    url = https://github.com/lz4/lz4.git
 [submodule "unicorn"]
     path = externals/unicorn
     url = https://github.com/yuzu-emu/unicorn

CMakeLists.txt

@@ -160,6 +160,9 @@ set_property(DIRECTORY APPEND PROPERTY
     COMPILE_DEFINITIONS $<$<CONFIG:Debug>:_DEBUG> $<$<NOT:$<CONFIG:Debug>>:NDEBUG>)
 
 
+math(EXPR EMU_ARCH_BITS ${CMAKE_SIZEOF_VOID_P}*8)
+add_definitions(-DEMU_ARCH_BITS=${EMU_ARCH_BITS})
+
 # System imported libraries
 # ======================
 

appveyor.yml

@@ -121,23 +121,16 @@ after_build:
           Get-ChildItem "$CMAKE_BINARY_DIR" -Filter "yuzu*.exe" | Copy-Item -destination $RELEASE_DIST
           Copy-Item -path "$CMAKE_SOURCE_DIR/license.txt" -destination $RELEASE_DIST
           Copy-Item -path "$CMAKE_SOURCE_DIR/README.md" -destination $RELEASE_DIST
+
           # copy all the dll dependencies to the release folder
-          # hardcoded list because we don't build static and determining the list of dlls from the binary is a pain.
-          $MingwDLLs = "Qt5Core.dll","Qt5Widgets.dll","Qt5Gui.dll","Qt5OpenGL.dll",
-                       # QT dll dependencies
-                       "libbz2-*.dll","libicudt*.dll","libicuin*.dll","libicuuc*.dll","libffi-*.dll",
-                       "libfreetype-*.dll","libglib-*.dll","libgobject-*.dll","libgraphite2.dll","libiconv-*.dll",
-                       "libharfbuzz-*.dll","libintl-*.dll","libpcre-*.dll","libpcre2-16-*.dll","libpcre16-*.dll","libpng16-*.dll",
-                       # Runtime/Other dependencies
-                       "libgcc_s_seh-*.dll","libstdc++-*.dll","libwinpthread-*.dll","SDL2.dll","zlib1.dll"
+          . "./.appveyor/UtilityFunctions.ps1"
+          $DLLSearchPath = "C:\msys64\mingw64\bin;$env:PATH"
+          $MingwDLLs = RecursivelyGetDeps $DLLSearchPath "$RELEASE_DIST\yuzu.exe"
+          $MingwDLLs += RecursivelyGetDeps $DLLSearchPath  "$RELEASE_DIST\yuzu_cmd.exe"
+          Write-Host "Detected the following dependencies:"
+          Write-Host $MingwDLLs
           foreach ($file in $MingwDLLs) {
-            Copy-Item -path "C:/msys64/mingw64/bin/$file" -force -destination "$RELEASE_DIST"
-          }
-          # the above list copies a few extra debug dlls that aren't needed (thanks globbing patterns!)
-          # so we can remove them by hardcoding another list of extra dlls to remove
-          $DebugDLLs = "libicudtd*.dll","libicuind*.dll","libicuucd*.dll"
-          foreach ($file in $DebugDLLs) {
-            Remove-Item -path "$RELEASE_DIST/$file"
+            Copy-Item -path "$file" -force -destination "$RELEASE_DIST"
           }
 
           # copy the qt windows plugin dll to platforms

externals/glad/Readme.md

@@ -1,5 +1,5 @@
 These files were generated by the [glad](https://github.com/Dav1dde/glad) OpenGL loader generator and have been checked in as-is. You can re-generate them using glad with the following command:
 
 ```
-python -m glad --profile core --out-path glad/ --api gl=3.3,gles=3.0
+python -m glad --profile core --out-path glad/ --api gl=3.3 --generator=c
 ```

externals/glad/include/KHR/khrplatform.h

@@ -26,7 +26,7 @@
 
 /* Khronos platform-specific types and definitions.
  *
- * $Revision: 23298 $ on $Date: 2013-09-30 17:07:13 -0700 (Mon, 30 Sep 2013) $
+ * $Revision: 32517 $ on $Date: 2016-03-11 02:41:19 -0800 (Fri, 11 Mar 2016) $
  *
  * Adopters may modify this file to suit their platform. Adopters are
  * encouraged to submit platform specific modifications to the Khronos
@@ -101,6 +101,8 @@
 #   define KHRONOS_APICALL __declspec(dllimport)
 #elif defined (__SYMBIAN32__)
 #   define KHRONOS_APICALL IMPORT_C
+#elif defined(__ANDROID__)
+#   define KHRONOS_APICALL __attribute__((visibility("default")))
 #else
 #   define KHRONOS_APICALL
 #endif
@@ -223,7 +225,7 @@ typedef signed   short int     khronos_int16_t;
 typedef unsigned short int     khronos_uint16_t;
 
 /*
- * Types that differ between LLP64 and LP64 architectures - in LLP64, 
+ * Types that differ between LLP64 and LP64 architectures - in LLP64,
  * pointers are 64 bits, but 'long' is still 32 bits. Win64 appears
  * to be the only LLP64 architecture in current use.
  */

hooks/pre-commit

@@ -1,6 +1,6 @@
 #!/bin/sh
 
-# Enforce citra's whitespace policy
+# Enforce yuzu's whitespace policy
 git config --local core.whitespace tab-in-indent,trailing-space
 
 paths_to_check="src/ CMakeLists.txt"

src/common/CMakeLists.txt

@@ -58,7 +58,6 @@ add_library(common STATIC
     misc.cpp
     param_package.cpp
     param_package.h
-    platform.h
     quaternion.h
     scm_rev.cpp
     scm_rev.h

src/common/file_util.h

@@ -121,7 +121,7 @@ void CopyDir(const std::string& source_path, const std::string& dest_path);
 // Set the current directory to given directory
 bool SetCurrentDir(const std::string& directory);
 
-// Returns a pointer to a string with a Citra data dir in the user's home
+// Returns a pointer to a string with a yuzu data dir in the user's home
 // directory. To be used in "multi-user" mode (that is, installed).
 const std::string& GetUserPath(const unsigned int DirIDX, const std::string& newPath = "");
 

src/common/logging/backend.cpp

@@ -37,6 +37,7 @@ namespace Log {
     SUB(Service, AM)                                                                               \
     SUB(Service, AOC)                                                                              \
     SUB(Service, APM)                                                                              \
+    SUB(Service, Fatal)                                                                            \
     SUB(Service, Friend)                                                                           \
     SUB(Service, FS)                                                                               \
     SUB(Service, HID)                                                                              \
@@ -47,6 +48,8 @@ namespace Log {
     SUB(Service, PCTL)                                                                             \
     SUB(Service, SET)                                                                              \
     SUB(Service, SM)                                                                               \
+    SUB(Service, SPL)                                                                              \
+    SUB(Service, SSL)                                                                              \
     SUB(Service, Time)                                                                             \
     SUB(Service, VI)                                                                               \
     CLS(HW)                                                                                        \

src/common/logging/log.h

@@ -54,6 +54,7 @@ enum class Class : ClassType {
     Service_AOC,       ///< The AOC (AddOn Content) service
     Service_APM,       ///< The APM (Performance) service
     Service_Audio,     ///< The Audio (Audio control) service
+    Service_Fatal,     ///< The Fatal service
     Service_Friend,    ///< The friend service
     Service_FS,        ///< The FS (Filesystem) service
     Service_HID,       ///< The HID (Human interface device) service
@@ -64,6 +65,8 @@ enum class Class : ClassType {
     Service_PCTL,      ///< The PCTL (Parental control) service
     Service_SET,       ///< The SET (Settings) service
     Service_SM,        ///< The SM (Service manager) service
+    Service_SPL,       ///< The SPL service
+    Service_SSL,       ///< The SSL service
     Service_Time,      ///< The time service
     Service_VI,        ///< The VI (Video interface) service
     HW,                ///< Low-level hardware emulation
@@ -82,7 +85,7 @@ enum class Class : ClassType {
     Loader,            ///< ROM loader
     Input,             ///< Input emulation
     Network,           ///< Network emulation
-    WebService,        ///< Interface to Citra Web Services
+    WebService,        ///< Interface to yuzu Web Services
     Count              ///< Total number of logging classes
 };
 

src/common/platform.h

@@ -1,34 +0,0 @@
-/**
- * Copyright (C) 2005-2012 Gekko Emulator
- *
- * @file    platform.h
- * @author  ShizZy <shizzy247@gmail.com>
- * @date    2012-02-11
- * @brief   Platform detection macros for portable compilation
- *
- * @section LICENSE
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of the GNU General Public License as
- * published by the Free Software Foundation; either version 2 of
- * the License, or (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful, but
- * WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- * General Public License for more details at
- * http://www.gnu.org/copyleft/gpl.html
- *
- * Official project repository can be found at:
- * http://code.google.com/p/gekko-gc-emu/
- */
-
-#pragma once
-
-////////////////////////////////////////////////////////////////////////////////////////////////////
-// Platform detection
-
-#if defined(ARCHITECTURE_x86_64) || defined(__aarch64__)
-#define EMU_ARCH_BITS 64
-#elif defined(__i386) || defined(_M_IX86) || defined(__arm__) || defined(_M_ARM)
-#define EMU_ARCH_BITS 32
-#endif

src/common/x64/cpu_detect.cpp

@@ -54,7 +54,7 @@ static CPUCaps Detect() {
     caps.num_cores = std::thread::hardware_concurrency();
 
     // Assumes the CPU supports the CPUID instruction. Those that don't would likely not support
-    // Citra at all anyway
+    // yuzu at all anyway
 
     int cpu_id[4];
     memset(caps.brand_string, 0, sizeof(caps.brand_string));

src/core/CMakeLists.txt

@@ -22,6 +22,8 @@ add_library(core STATIC
     file_sys/romfs_filesystem.h
     file_sys/savedata_factory.cpp
     file_sys/savedata_factory.h
+    file_sys/sdmc_factory.cpp
+    file_sys/sdmc_factory.h
     file_sys/storage.h
     frontend/emu_window.cpp
     frontend/emu_window.h
@@ -114,6 +116,12 @@ add_library(core STATIC
     hle/service/audio/audren_u.h
     hle/service/audio/codecctl.cpp
     hle/service/audio/codecctl.h
+    hle/service/fatal/fatal.cpp
+    hle/service/fatal/fatal.h
+    hle/service/fatal/fatal_p.cpp
+    hle/service/fatal/fatal_p.h
+    hle/service/fatal/fatal_u.cpp
+    hle/service/fatal/fatal_u.h
     hle/service/filesystem/filesystem.cpp
     hle/service/filesystem/filesystem.h
     hle/service/filesystem/fsp_srv.cpp
@@ -181,12 +189,22 @@ add_library(core STATIC
     hle/service/sm/controller.h
     hle/service/sm/sm.cpp
     hle/service/sm/sm.h
-    hle/service/sockets/bsd_u.cpp
-    hle/service/sockets/bsd_u.h
+    hle/service/sockets/bsd.cpp
+    hle/service/sockets/bsd.h
+    hle/service/sockets/nsd.cpp
+    hle/service/sockets/nsd.h
     hle/service/sockets/sfdnsres.cpp
     hle/service/sockets/sfdnsres.h
     hle/service/sockets/sockets.cpp
     hle/service/sockets/sockets.h
+    hle/service/spl/csrng.cpp
+    hle/service/spl/csrng.h
+    hle/service/spl/module.cpp
+    hle/service/spl/module.h
+    hle/service/spl/spl.cpp
+    hle/service/spl/spl.h
+    hle/service/ssl/ssl.cpp
+    hle/service/ssl/ssl.h
     hle/service/time/time.cpp
     hle/service/time/time.h
     hle/service/time/time_s.cpp

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -86,21 +86,17 @@ public:
     }
 
     void AddTicks(u64 ticks) override {
-        if (ticks > ticks_remaining) {
-            ticks_remaining = 0;
-            return;
-        }
-        ticks -= ticks_remaining;
+        CoreTiming::AddTicks(ticks - num_interpreted_instructions);
+        num_interpreted_instructions = 0;
     }
     u64 GetTicksRemaining() override {
-        return ticks_remaining;
+        return std::max(CoreTiming::GetDowncount(), 0);
     }
     u64 GetCNTPCT() override {
         return CoreTiming::GetTicks();
     }
 
     ARM_Dynarmic& parent;
-    size_t ticks_remaining = 0;
     size_t num_interpreted_instructions = 0;
     u64 tpidrro_el0 = 0;
     u64 tpidr_el0 = 0;

src/core/core.cpp

@@ -148,19 +148,15 @@ System::ResultStatus System::Init(EmuWindow* emu_window, u32 system_mode) {
 
     current_process = Kernel::Process::Create("main");
 
-    switch (Settings::values.cpu_core) {
-    case Settings::CpuCore::Unicorn:
-        cpu_core = std::make_shared<ARM_Unicorn>();
-        break;
-    case Settings::CpuCore::Dynarmic:
-    default:
+    if (Settings::values.use_cpu_jit) {
 #ifdef ARCHITECTURE_x86_64
         cpu_core = std::make_shared<ARM_Dynarmic>();
 #else
         cpu_core = std::make_shared<ARM_Unicorn>();
         LOG_WARNING(Core, "CPU JIT requested, but Dynarmic not available");
 #endif
-        break;
+    } else {
+        cpu_core = std::make_shared<ARM_Unicorn>();
     }
 
     gpu_core = std::make_unique<Tegra::GPU>();

src/core/core.h

@@ -13,6 +13,7 @@
 #include "core/memory.h"
 #include "core/perf_stats.h"
 #include "core/telemetry_session.h"
+#include "video_core/debug_utils/debug_utils.h"
 #include "video_core/gpu.h"
 
 class EmuWindow;
@@ -135,6 +136,14 @@ public:
         return *app_loader;
     }
 
+    void SetGPUDebugContext(std::shared_ptr<Tegra::DebugContext> context) {
+        debug_context = std::move(context);
+    }
+
+    std::shared_ptr<Tegra::DebugContext> GetGPUDebugContext() const {
+        return debug_context;
+    }
+
 private:
     /**
      * Initialize the emulated system.
@@ -154,6 +163,8 @@ private:
     std::unique_ptr<Kernel::Scheduler> scheduler;
     std::unique_ptr<Tegra::GPU> gpu_core;
 
+    std::shared_ptr<Tegra::DebugContext> debug_context;
+
     Kernel::SharedPtr<Kernel::Process> current_process;
 
     /// When true, signals that a reschedule should happen

src/core/file_sys/directory.h

@@ -6,34 +6,28 @@
 
 #include <array>
 #include <cstddef>
+#include "common/common_funcs.h"
 #include "common/common_types.h"
+#include "core/file_sys/filesystem.h"
 
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 // FileSys namespace
 
 namespace FileSys {
 
-// Structure of a directory entry, from http://3dbrew.org/wiki/FSDir:Read#Entry_format
-const size_t FILENAME_LENGTH = 0x20C / 2;
+// Structure of a directory entry, from
+// http://switchbrew.org/index.php?title=Filesystem_services#DirectoryEntry
+const size_t FILENAME_LENGTH = 0x300;
 struct Entry {
-    char16_t filename[FILENAME_LENGTH]; // Entry name (UTF-16, null-terminated)
-    std::array<char, 9> short_name; // 8.3 file name ('longfilename' -> 'LONGFI~1', null-terminated)
-    char unknown1;                  // unknown (observed values: 0x0A, 0x70, 0xFD)
-    std::array<char, 4>
-        extension;     // 8.3 file extension (set to spaces for directories, null-terminated)
-    char unknown2;     // unknown (always 0x01)
-    char unknown3;     // unknown (0x00 or 0x08)
-    char is_directory; // directory flag
-    char is_hidden;    // hidden flag
-    char is_archive;   // archive flag
-    char is_read_only; // read-only flag
-    u64 file_size;     // file size (for files only)
+    char filename[FILENAME_LENGTH];
+    INSERT_PADDING_BYTES(4);
+    EntryType type;
+    INSERT_PADDING_BYTES(3);
+    u64 file_size;
 };
-static_assert(sizeof(Entry) == 0x228, "Directory Entry struct isn't exactly 0x228 bytes long!");
-static_assert(offsetof(Entry, short_name) == 0x20C, "Wrong offset for short_name in Entry.");
-static_assert(offsetof(Entry, extension) == 0x216, "Wrong offset for extension in Entry.");
-static_assert(offsetof(Entry, is_archive) == 0x21E, "Wrong offset for is_archive in Entry.");
-static_assert(offsetof(Entry, file_size) == 0x220, "Wrong offset for file_size in Entry.");
+static_assert(sizeof(Entry) == 0x310, "Directory Entry struct isn't exactly 0x310 bytes long!");
+static_assert(offsetof(Entry, type) == 0x304, "Wrong offset for type in Entry.");
+static_assert(offsetof(Entry, file_size) == 0x308, "Wrong offset for file_size in Entry.");
 
 class DirectoryBackend : NonCopyable {
 public:
@@ -46,7 +40,10 @@ public:
      * @param entries Buffer to read data into
      * @return Number of entries listed
      */
-    virtual u32 Read(const u32 count, Entry* entries) = 0;
+    virtual u64 Read(const u64 count, Entry* entries) = 0;
+
+    /// Returns the number of entries still left to read.
+    virtual u64 GetEntryCount() const = 0;
 
     /**
      * Close the directory

src/core/file_sys/disk_filesystem.cpp

@@ -11,16 +11,43 @@
 
 namespace FileSys {
 
+static std::string ModeFlagsToString(Mode mode) {
+    std::string mode_str;
+    u32 mode_flags = static_cast<u32>(mode);
+
+    // Calculate the correct open mode for the file.
+    if ((mode_flags & static_cast<u32>(Mode::Read)) &&
+        (mode_flags & static_cast<u32>(Mode::Write))) {
+        if (mode_flags & static_cast<u32>(Mode::Append))
+            mode_str = "a+";
+        else
+            mode_str = "r+";
+    } else {
+        if (mode_flags & static_cast<u32>(Mode::Read))
+            mode_str = "r";
+        else if (mode_flags & static_cast<u32>(Mode::Append))
+            mode_str = "a";
+        else if (mode_flags & static_cast<u32>(Mode::Write))
+            mode_str = "w";
+    }
+
+    mode_str += "b";
+
+    return mode_str;
+}
+
 std::string Disk_FileSystem::GetName() const {
     return "Disk";
 }
 
 ResultVal<std::unique_ptr<StorageBackend>> Disk_FileSystem::OpenFile(const std::string& path,
                                                                      Mode mode) const {
-    ASSERT_MSG(mode == Mode::Read || mode == Mode::Write, "Other file modes are not supported");
+
+    // Calculate the correct open mode for the file.
+    std::string mode_str = ModeFlagsToString(mode);
 
     std::string full_path = base_directory + path;
-    auto file = std::make_shared<FileUtil::IOFile>(full_path, mode == Mode::Read ? "rb" : "wb");
+    auto file = std::make_shared<FileUtil::IOFile>(full_path, mode_str.c_str());
 
     if (!file->IsOpen()) {
         return ERROR_PATH_NOT_FOUND;
@@ -75,8 +102,15 @@ ResultCode Disk_FileSystem::CreateFile(const std::string& path, u64 size) const
     return ResultCode(-1);
 }
 
-ResultCode Disk_FileSystem::CreateDirectory(const Path& path) const {
-    LOG_WARNING(Service_FS, "(STUBBED) called");
+ResultCode Disk_FileSystem::CreateDirectory(const std::string& path) const {
+    // TODO(Subv): Perform path validation to prevent escaping the emulator sandbox.
+    std::string full_path = base_directory + path;
+
+    if (FileUtil::CreateDir(full_path)) {
+        return RESULT_SUCCESS;
+    }
+
+    LOG_CRITICAL(Service_FS, "(unreachable) Unknown error creating %s", full_path.c_str());
     // TODO(wwylele): Use correct error code
     return ResultCode(-1);
 }
@@ -88,8 +122,17 @@ ResultCode Disk_FileSystem::RenameDirectory(const Path& src_path, const Path& de
 }
 
 ResultVal<std::unique_ptr<DirectoryBackend>> Disk_FileSystem::OpenDirectory(
-    const Path& path) const {
-    return MakeResult<std::unique_ptr<DirectoryBackend>>(std::make_unique<Disk_Directory>());
+    const std::string& path) const {
+
+    std::string full_path = base_directory + path;
+
+    if (!FileUtil::IsDirectory(full_path)) {
+        // TODO(Subv): Find the correct error code for this.
+        return ResultCode(-1);
+    }
+
+    auto directory = std::make_unique<Disk_Directory>(full_path);
+    return MakeResult<std::unique_ptr<DirectoryBackend>>(std::move(directory));
 }
 
 u64 Disk_FileSystem::GetFreeSpaceSize() const {
@@ -103,8 +146,10 @@ ResultVal<FileSys::EntryType> Disk_FileSystem::GetEntryType(const std::string& p
         return ERROR_PATH_NOT_FOUND;
     }
 
-    // TODO(Subv): Find out the EntryType values
-    UNIMPLEMENTED_MSG("Unimplemented GetEntryType");
+    if (FileUtil::IsDirectory(full_path))
+        return MakeResult(EntryType::Directory);
+
+    return MakeResult(EntryType::File);
 }
 
 ResultVal<size_t> Disk_Storage::Read(const u64 offset, const size_t length, u8* buffer) const {
@@ -133,14 +178,50 @@ bool Disk_Storage::SetSize(const u64 size) const {
     return false;
 }
 
-u32 Disk_Directory::Read(const u32 count, Entry* entries) {
-    LOG_WARNING(Service_FS, "(STUBBED) called");
-    return 0;
+Disk_Directory::Disk_Directory(const std::string& path) : directory() {
+    unsigned size = FileUtil::ScanDirectoryTree(path, directory);
+    directory.size = size;
+    directory.isDirectory = true;
+    children_iterator = directory.children.begin();
 }
 
-bool Disk_Directory::Close() const {
-    LOG_WARNING(Service_FS, "(STUBBED) called");
-    return true;
+u64 Disk_Directory::Read(const u64 count, Entry* entries) {
+    u64 entries_read = 0;
+
+    while (entries_read < count && children_iterator != directory.children.cend()) {
+        const FileUtil::FSTEntry& file = *children_iterator;
+        const std::string& filename = file.virtualName;
+        Entry& entry = entries[entries_read];
+
+        LOG_TRACE(Service_FS, "File %s: size=%llu dir=%d", filename.c_str(), file.size,
+                  file.isDirectory);
+
+        // TODO(Link Mauve): use a proper conversion to UTF-16.
+        for (size_t j = 0; j < FILENAME_LENGTH; ++j) {
+            entry.filename[j] = filename[j];
+            if (!filename[j])
+                break;
+        }
+
+        if (file.isDirectory) {
+            entry.file_size = 0;
+            entry.type = EntryType::Directory;
+        } else {
+            entry.file_size = file.size;
+            entry.type = EntryType::File;
+        }
+
+        ++entries_read;
+        ++children_iterator;
+    }
+    return entries_read;
+}
+
+u64 Disk_Directory::GetEntryCount() const {
+    // We convert the children iterator into a const_iterator to allow template argument deduction
+    // in std::distance.
+    std::vector<FileUtil::FSTEntry>::const_iterator current = children_iterator;
+    return std::distance(current, directory.children.end());
 }
 
 } // namespace FileSys

src/core/file_sys/disk_filesystem.h

@@ -30,9 +30,10 @@ public:
     ResultCode DeleteDirectory(const Path& path) const override;
     ResultCode DeleteDirectoryRecursively(const Path& path) const override;
     ResultCode CreateFile(const std::string& path, u64 size) const override;
-    ResultCode CreateDirectory(const Path& path) const override;
+    ResultCode CreateDirectory(const std::string& path) const override;
     ResultCode RenameDirectory(const Path& src_path, const Path& dest_path) const override;
-    ResultVal<std::unique_ptr<DirectoryBackend>> OpenDirectory(const Path& path) const override;
+    ResultVal<std::unique_ptr<DirectoryBackend>> OpenDirectory(
+        const std::string& path) const override;
     u64 GetFreeSpaceSize() const override;
     ResultVal<EntryType> GetEntryType(const std::string& path) const override;
 
@@ -59,8 +60,26 @@ private:
 
 class Disk_Directory : public DirectoryBackend {
 public:
-    u32 Read(const u32 count, Entry* entries) override;
-    bool Close() const override;
+    Disk_Directory(const std::string& path);
+
+    ~Disk_Directory() override {
+        Close();
+    }
+
+    u64 Read(const u64 count, Entry* entries) override;
+    u64 GetEntryCount() const override;
+
+    bool Close() const override {
+        return true;
+    }
+
+protected:
+    u32 total_entries_in_directory;
+    FileUtil::FSTEntry directory;
+
+    // We need to remember the last entry we returned, so a subsequent call to Read will continue
+    // from the next one. This iterator will always point to the next unread entry.
+    std::vector<FileUtil::FSTEntry>::iterator children_iterator;
 };
 
 } // namespace FileSys

src/core/file_sys/filesystem.h

@@ -27,7 +27,7 @@ enum LowPathType : u32 {
     Wchar = 4,
 };
 
-enum EntryType : u32 {
+enum EntryType : u8 {
     Directory = 0,
     File = 1,
 };
@@ -35,6 +35,7 @@ enum EntryType : u32 {
 enum class Mode : u32 {
     Read = 1,
     Write = 2,
+    Append = 4,
 };
 
 class Path {
@@ -103,7 +104,7 @@ public:
      * @param path Path relative to the archive
      * @return Result of the operation
      */
-    virtual ResultCode CreateDirectory(const Path& path) const = 0;
+    virtual ResultCode CreateDirectory(const std::string& path) const = 0;
 
     /**
      * Delete a directory specified by its path
@@ -149,7 +150,8 @@ public:
      * @param path Path relative to the archive
      * @return Opened directory, or error code
      */
-    virtual ResultVal<std::unique_ptr<DirectoryBackend>> OpenDirectory(const Path& path) const = 0;
+    virtual ResultVal<std::unique_ptr<DirectoryBackend>> OpenDirectory(
+        const std::string& path) const = 0;
 
     /**
      * Get the free space

src/core/file_sys/romfs_filesystem.cpp

@@ -55,7 +55,7 @@ ResultCode RomFS_FileSystem::CreateFile(const std::string& path, u64 size) const
     return ResultCode(-1);
 }
 
-ResultCode RomFS_FileSystem::CreateDirectory(const Path& path) const {
+ResultCode RomFS_FileSystem::CreateDirectory(const std::string& path) const {
     LOG_CRITICAL(Service_FS, "Attempted to create a directory in an ROMFS archive (%s).",
                  GetName().c_str());
     // TODO(wwylele): Use correct error code
@@ -70,7 +70,8 @@ ResultCode RomFS_FileSystem::RenameDirectory(const Path& src_path, const Path& d
 }
 
 ResultVal<std::unique_ptr<DirectoryBackend>> RomFS_FileSystem::OpenDirectory(
-    const Path& path) const {
+    const std::string& path) const {
+    LOG_WARNING(Service_FS, "Opening Directory in a ROMFS archive");
     return MakeResult<std::unique_ptr<DirectoryBackend>>(std::make_unique<ROMFSDirectory>());
 }
 

src/core/file_sys/romfs_filesystem.h

@@ -36,9 +36,10 @@ public:
     ResultCode DeleteDirectory(const Path& path) const override;
     ResultCode DeleteDirectoryRecursively(const Path& path) const override;
     ResultCode CreateFile(const std::string& path, u64 size) const override;
-    ResultCode CreateDirectory(const Path& path) const override;
+    ResultCode CreateDirectory(const std::string& path) const override;
     ResultCode RenameDirectory(const Path& src_path, const Path& dest_path) const override;
-    ResultVal<std::unique_ptr<DirectoryBackend>> OpenDirectory(const Path& path) const override;
+    ResultVal<std::unique_ptr<DirectoryBackend>> OpenDirectory(
+        const std::string& path) const override;
     u64 GetFreeSpaceSize() const override;
     ResultVal<EntryType> GetEntryType(const std::string& path) const override;
 
@@ -70,7 +71,10 @@ private:
 
 class ROMFSDirectory : public DirectoryBackend {
 public:
-    u32 Read(const u32 count, Entry* entries) override {
+    u64 Read(const u64 count, Entry* entries) override {
+        return 0;
+    }
+    u64 GetEntryCount() const override {
         return 0;
     }
     bool Close() const override {

src/core/file_sys/sdmc_factory.cpp

@@ -0,0 +1,40 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cinttypes>
+#include <memory>
+#include "common/common_types.h"
+#include "common/logging/log.h"
+#include "common/string_util.h"
+#include "core/core.h"
+#include "core/file_sys/disk_filesystem.h"
+#include "core/file_sys/sdmc_factory.h"
+
+namespace FileSys {
+
+SDMC_Factory::SDMC_Factory(std::string sd_directory) : sd_directory(std::move(sd_directory)) {}
+
+ResultVal<std::unique_ptr<FileSystemBackend>> SDMC_Factory::Open(const Path& path) {
+    // Create the SD Card directory if it doesn't already exist.
+    if (!FileUtil::IsDirectory(sd_directory)) {
+        FileUtil::CreateFullPath(sd_directory);
+    }
+
+    auto archive = std::make_unique<Disk_FileSystem>(sd_directory);
+    return MakeResult<std::unique_ptr<FileSystemBackend>>(std::move(archive));
+}
+
+ResultCode SDMC_Factory::Format(const Path& path) {
+    LOG_ERROR(Service_FS, "Unimplemented Format archive %s", GetName().c_str());
+    // TODO(Subv): Find the right error code for this
+    return ResultCode(-1);
+}
+
+ResultVal<ArchiveFormatInfo> SDMC_Factory::GetFormatInfo(const Path& path) const {
+    LOG_ERROR(Service_FS, "Unimplemented GetFormatInfo archive %s", GetName().c_str());
+    // TODO(bunnei): Find the right error code for this
+    return ResultCode(-1);
+}
+
+} // namespace FileSys

src/core/file_sys/sdmc_factory.h

@@ -0,0 +1,31 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include <string>
+#include "common/common_types.h"
+#include "core/file_sys/filesystem.h"
+#include "core/hle/result.h"
+
+namespace FileSys {
+
+/// File system interface to the SDCard archive
+class SDMC_Factory final : public FileSystemFactory {
+public:
+    explicit SDMC_Factory(std::string sd_directory);
+
+    std::string GetName() const override {
+        return "SDMC_Factory";
+    }
+    ResultVal<std::unique_ptr<FileSystemBackend>> Open(const Path& path) override;
+    ResultCode Format(const Path& path) override;
+    ResultVal<ArchiveFormatInfo> GetFormatInfo(const Path& path) const override;
+
+private:
+    std::string sd_directory;
+};
+
+} // namespace FileSys

src/core/hle/result.h

@@ -200,6 +200,9 @@ public:
     }
 
     ResultVal& operator=(const ResultVal& o) {
+        if (this == &o) {
+            return *this;
+        }
         if (!empty()) {
             if (!o.empty()) {
                 object = o.object;

src/core/hle/service/am/am.cpp

@@ -12,6 +12,7 @@
 #include "core/hle/service/apm/apm.h"
 #include "core/hle/service/filesystem/filesystem.h"
 #include "core/hle/service/nvflinger/nvflinger.h"
+#include "core/settings.h"
 
 namespace Service {
 namespace AM {
@@ -241,17 +242,20 @@ void ICommonStateGetter::GetCurrentFocusState(Kernel::HLERequestContext& ctx) {
 }
 
 void ICommonStateGetter::GetOperationMode(Kernel::HLERequestContext& ctx) {
+    const bool use_docked_mode{Settings::values.use_docked_mode};
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
-    rb.Push(static_cast<u8>(OperationMode::Handheld));
+    rb.Push(static_cast<u8>(use_docked_mode ? OperationMode::Docked : OperationMode::Handheld));
 
     LOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
 void ICommonStateGetter::GetPerformanceMode(Kernel::HLERequestContext& ctx) {
+    const bool use_docked_mode{Settings::values.use_docked_mode};
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
-    rb.Push(static_cast<u32>(APM::PerformanceMode::Handheld));
+    rb.Push(static_cast<u32>(use_docked_mode ? APM::PerformanceMode::Docked
+                                             : APM::PerformanceMode::Handheld));
 
     LOG_WARNING(Service_AM, "(STUBBED) called");
 }

src/core/hle/service/audio/audren_u.cpp

@@ -151,12 +151,52 @@ private:
     Kernel::SharedPtr<Kernel::Event> system_event;
 };
 
+class IAudioDevice final : public ServiceFramework<IAudioDevice> {
+public:
+    IAudioDevice() : ServiceFramework("IAudioDevice") {
+        static const FunctionInfo functions[] = {
+            {0x0, &IAudioDevice::ListAudioDeviceName, "ListAudioDeviceName"},
+            {0x1, &IAudioDevice::SetAudioDeviceOutputVolume, "SetAudioDeviceOutputVolume"}};
+        RegisterHandlers(functions);
+
+        buffer_event =
+            Kernel::Event::Create(Kernel::ResetType::OneShot, "IAudioOutBufferReleasedEvent");
+    }
+
+private:
+    void ListAudioDeviceName(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        IPC::RequestParser rp{ctx};
+
+        const std::string audio_interface = "AudioInterface";
+        ctx.WriteBuffer(audio_interface.c_str(), audio_interface.size());
+
+        IPC::ResponseBuilder rb = rp.MakeBuilder(3, 0, 0);
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(1);
+    }
+
+    void SetAudioDeviceOutputVolume(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_Audio, "(STUBBED) called");
+
+        IPC::RequestParser rp{ctx};
+        f32 volume = static_cast<f32>(rp.Pop<u32>());
+
+        auto file_buffer = ctx.ReadBuffer();
+        auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0');
+
+        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        rb.Push(RESULT_SUCCESS);
+    }
+
+    Kernel::SharedPtr<Kernel::Event> buffer_event;
+};
+
 AudRenU::AudRenU() : ServiceFramework("audren:u") {
     static const FunctionInfo functions[] = {
         {0, &AudRenU::OpenAudioRenderer, "OpenAudioRenderer"},
         {1, &AudRenU::GetAudioRendererWorkBufferSize, "GetAudioRendererWorkBufferSize"},
-        {2, &AudRenU::GetAudioRenderersProcessMasterVolume, "GetAudioRenderersProcessMasterVolume"},
-        {3, nullptr, "SetAudioRenderersProcessMasterVolume"},
+        {2, &AudRenU::GetAudioDevice, "GetAudioDevice"},
     };
     RegisterHandlers(functions);
 }
@@ -179,12 +219,13 @@ void AudRenU::GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_Audio, "(STUBBED) called");
 }
 
-void AudRenU::GetAudioRenderersProcessMasterVolume(Kernel::HLERequestContext& ctx) {
-    IPC::ResponseBuilder rb{ctx, 3};
+void AudRenU::GetAudioDevice(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
 
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u32>(100);
-    LOG_WARNING(Service_Audio, "(STUBBED) called");
+    rb.PushIpcInterface<Audio::IAudioDevice>();
+
+    LOG_DEBUG(Service_Audio, "called");
 }
 
 } // namespace Audio

src/core/hle/service/audio/audren_u.h

@@ -21,7 +21,7 @@ public:
 private:
     void OpenAudioRenderer(Kernel::HLERequestContext& ctx);
     void GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx);
-    void GetAudioRenderersProcessMasterVolume(Kernel::HLERequestContext& ctx);
+    void GetAudioDevice(Kernel::HLERequestContext& ctx);
 };
 
 } // namespace Audio

src/core/hle/service/fatal/fatal.cpp

@@ -0,0 +1,38 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/logging/log.h"
+#include "core/hle/ipc_helpers.h"
+#include "core/hle/service/fatal/fatal.h"
+#include "core/hle/service/fatal/fatal_p.h"
+#include "core/hle/service/fatal/fatal_u.h"
+
+namespace Service {
+namespace Fatal {
+
+Module::Interface::Interface(std::shared_ptr<Module> module, const char* name)
+    : ServiceFramework(name), module(std::move(module)) {}
+
+void Module::Interface::FatalSimple(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp(ctx);
+    u32 error_code = rp.Pop<u32>();
+    LOG_WARNING(Service_Fatal, "(STUBBED) called, error_code=0x%X", error_code);
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+}
+
+void Module::Interface::TransitionToFatalError(Kernel::HLERequestContext& ctx) {
+    LOG_WARNING(Service_Fatal, "(STUBBED) called");
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+}
+
+void InstallInterfaces(SM::ServiceManager& service_manager) {
+    auto module = std::make_shared<Module>();
+    std::make_shared<Fatal_P>(module)->InstallAsService(service_manager);
+    std::make_shared<Fatal_U>(module)->InstallAsService(service_manager);
+}
+
+} // namespace Fatal
+} // namespace Service

src/core/hle/service/fatal/fatal.h

@@ -0,0 +1,29 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/service.h"
+
+namespace Service {
+namespace Fatal {
+
+class Module final {
+public:
+    class Interface : public ServiceFramework<Interface> {
+    public:
+        Interface(std::shared_ptr<Module> module, const char* name);
+
+        void FatalSimple(Kernel::HLERequestContext& ctx);
+        void TransitionToFatalError(Kernel::HLERequestContext& ctx);
+
+    protected:
+        std::shared_ptr<Module> module;
+    };
+};
+
+void InstallInterfaces(SM::ServiceManager& service_manager);
+
+} // namespace Fatal
+} // namespace Service

src/core/hle/service/fatal/fatal_p.cpp

@@ -0,0 +1,14 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/service/fatal/fatal_p.h"
+
+namespace Service {
+namespace Fatal {
+
+Fatal_P::Fatal_P(std::shared_ptr<Module> module)
+    : Module::Interface(std::move(module), "fatal:p") {}
+
+} // namespace Fatal
+} // namespace Service

src/core/hle/service/fatal/fatal_p.h

@@ -0,0 +1,18 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/fatal/fatal.h"
+
+namespace Service {
+namespace Fatal {
+
+class Fatal_P final : public Module::Interface {
+public:
+    explicit Fatal_P(std::shared_ptr<Module> module);
+};
+
+} // namespace Fatal
+} // namespace Service

src/core/hle/service/fatal/fatal_u.cpp

@@ -0,0 +1,19 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/service/fatal/fatal_u.h"
+
+namespace Service {
+namespace Fatal {
+
+Fatal_U::Fatal_U(std::shared_ptr<Module> module) : Module::Interface(std::move(module), "fatal:u") {
+    static const FunctionInfo functions[] = {
+        {1, &Fatal_U::FatalSimple, "FatalSimple"},
+        {2, &Fatal_U::TransitionToFatalError, "TransitionToFatalError"},
+    };
+    RegisterHandlers(functions);
+}
+
+} // namespace Fatal
+} // namespace Service

src/core/hle/service/fatal/fatal_u.h

@@ -0,0 +1,18 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/fatal/fatal.h"
+
+namespace Service {
+namespace Fatal {
+
+class Fatal_U final : public Module::Interface {
+public:
+    explicit Fatal_U(std::shared_ptr<Module> module);
+};
+
+} // namespace Fatal
+} // namespace Service

src/core/hle/service/filesystem/filesystem.cpp

@@ -6,6 +6,7 @@
 #include "common/file_util.h"
 #include "core/file_sys/filesystem.h"
 #include "core/file_sys/savedata_factory.h"
+#include "core/file_sys/sdmc_factory.h"
 #include "core/hle/service/filesystem/filesystem.h"
 #include "core/hle/service/filesystem/fsp_srv.h"
 
@@ -60,9 +61,13 @@ void RegisterFileSystems() {
     filesystem_map.clear();
 
     std::string nand_directory = FileUtil::GetUserPath(D_NAND_IDX);
+    std::string sd_directory = FileUtil::GetUserPath(D_SDMC_IDX);
 
     auto savedata = std::make_unique<FileSys::SaveData_Factory>(std::move(nand_directory));
     RegisterFileSystem(std::move(savedata), Type::SaveData);
+
+    auto sdcard = std::make_unique<FileSys::SDMC_Factory>(std::move(sd_directory));
+    RegisterFileSystem(std::move(sdcard), Type::SDMC);
 }
 
 void InstallInterfaces(SM::ServiceManager& service_manager) {

src/core/hle/service/filesystem/filesystem.h

@@ -26,6 +26,7 @@ namespace FileSystem {
 enum class Type {
     RomFS = 1,
     SaveData = 2,
+    SDMC = 3,
 };
 
 /**

src/core/hle/service/filesystem/fsp_srv.cpp

@@ -5,6 +5,7 @@
 #include <cinttypes>
 #include "common/logging/log.h"
 #include "core/core.h"
+#include "core/file_sys/directory.h"
 #include "core/file_sys/filesystem.h"
 #include "core/file_sys/storage.h"
 #include "core/hle/ipc_helpers.h"
@@ -151,14 +152,66 @@ private:
     }
 };
 
+class IDirectory final : public ServiceFramework<IDirectory> {
+public:
+    explicit IDirectory(std::unique_ptr<FileSys::DirectoryBackend>&& backend)
+        : ServiceFramework("IDirectory"), backend(std::move(backend)) {
+        static const FunctionInfo functions[] = {
+            {0, &IDirectory::Read, "Read"},
+            {1, &IDirectory::GetEntryCount, "GetEntryCount"},
+        };
+        RegisterHandlers(functions);
+    }
+
+private:
+    std::unique_ptr<FileSys::DirectoryBackend> backend;
+
+    void Read(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const u64 unk = rp.Pop<u64>();
+
+        LOG_DEBUG(Service_FS, "called, unk=0x%llx", unk);
+
+        // Calculate how many entries we can fit in the output buffer
+        u64 count_entries = ctx.GetWriteBufferSize() / sizeof(FileSys::Entry);
+
+        // Read the data from the Directory backend
+        std::vector<FileSys::Entry> entries(count_entries);
+        u64 read_entries = backend->Read(count_entries, entries.data());
+
+        // Convert the data into a byte array
+        std::vector<u8> output(entries.size() * sizeof(FileSys::Entry));
+        std::memcpy(output.data(), entries.data(), output.size());
+
+        // Write the data to memory
+        ctx.WriteBuffer(output);
+
+        IPC::ResponseBuilder rb{ctx, 4};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push(read_entries);
+    }
+
+    void GetEntryCount(Kernel::HLERequestContext& ctx) {
+        LOG_DEBUG(Service_FS, "called");
+
+        u64 count = backend->GetEntryCount();
+
+        IPC::ResponseBuilder rb{ctx, 4};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push(count);
+    }
+};
+
 class IFileSystem final : public ServiceFramework<IFileSystem> {
 public:
     explicit IFileSystem(std::unique_ptr<FileSys::FileSystemBackend>&& backend)
         : ServiceFramework("IFileSystem"), backend(std::move(backend)) {
         static const FunctionInfo functions[] = {
             {0, &IFileSystem::CreateFile, "CreateFile"},
+            {2, &IFileSystem::CreateDirectory, "CreateDirectory"},
             {7, &IFileSystem::GetEntryType, "GetEntryType"},
             {8, &IFileSystem::OpenFile, "OpenFile"},
+            {9, &IFileSystem::OpenDirectory, "OpenDirectory"},
             {10, &IFileSystem::Commit, "Commit"},
         };
         RegisterHandlers(functions);
@@ -182,6 +235,20 @@ public:
         rb.Push(backend->CreateFile(name, size));
     }
 
+    void CreateDirectory(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+
+        auto file_buffer = ctx.ReadBuffer();
+        auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0');
+
+        std::string name(file_buffer.begin(), end);
+
+        LOG_DEBUG(Service_FS, "called directory %s", name.c_str());
+
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(backend->CreateDirectory(name));
+    }
+
     void OpenFile(Kernel::HLERequestContext& ctx) {
         IPC::RequestParser rp{ctx};
 
@@ -208,6 +275,33 @@ public:
         rb.PushIpcInterface<IFile>(std::move(file));
     }
 
+    void OpenDirectory(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+
+        auto file_buffer = ctx.ReadBuffer();
+        auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0');
+
+        std::string name(file_buffer.begin(), end);
+
+        // TODO(Subv): Implement this filter.
+        u32 filter_flags = rp.Pop<u32>();
+
+        LOG_DEBUG(Service_FS, "called directory %s filter %u", name.c_str(), filter_flags);
+
+        auto result = backend->OpenDirectory(name);
+        if (result.Failed()) {
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(result.Code());
+            return;
+        }
+
+        auto directory = std::move(result.Unwrap());
+
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<IDirectory>(std::move(directory));
+    }
+
     void GetEntryType(Kernel::HLERequestContext& ctx) {
         IPC::RequestParser rp{ctx};
 
@@ -274,10 +368,14 @@ void FSP_SRV::Initalize(Kernel::HLERequestContext& ctx) {
 }
 
 void FSP_SRV::MountSdCard(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_FS, "(STUBBED) called");
+    LOG_DEBUG(Service_FS, "called");
 
-    IPC::ResponseBuilder rb{ctx, 2};
+    FileSys::Path unused;
+    auto filesystem = OpenFileSystem(Type::SDMC, unused).Unwrap();
+
+    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
+    rb.PushIpcInterface<IFileSystem>(std::move(filesystem));
 }
 
 void FSP_SRV::CreateSaveData(Kernel::HLERequestContext& ctx) {

src/core/hle/service/hid/hid.cpp

@@ -193,7 +193,7 @@ public:
             {121, &Hid::GetNpadJoyHoldType, "GetNpadJoyHoldType"},
             {122, &Hid::SetNpadJoyAssignmentModeSingleByDefault,
              "SetNpadJoyAssignmentModeSingleByDefault"},
-            {124, nullptr, "SetNpadJoyAssignmentModeDual"},
+            {124, &Hid::SetNpadJoyAssignmentModeDual, "SetNpadJoyAssignmentModeDual"},
             {128, &Hid::SetNpadHandheldActivationMode, "SetNpadHandheldActivationMode"},
             {200, &Hid::GetVibrationDeviceInfo, "GetVibrationDeviceInfo"},
             {201, &Hid::SendVibrationValue, "SendVibrationValue"},
@@ -315,6 +315,12 @@ private:
         LOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
+    void SetNpadJoyAssignmentModeDual(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_WARNING(Service_HID, "(STUBBED) called");
+    }
+
     void SetNpadHandheldActivationMode(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);

src/core/hle/service/ns/pl_u.cpp

@@ -33,6 +33,7 @@ enum class LoadState : u32 {
 
 PL_U::PL_U() : ServiceFramework("pl:u") {
     static const FunctionInfo functions[] = {
+        {0, &PL_U::RequestLoad, "RequestLoad"},
         {1, &PL_U::GetLoadState, "GetLoadState"},
         {2, &PL_U::GetSize, "GetSize"},
         {3, &PL_U::GetSharedMemoryAddressOffset, "GetSharedMemoryAddressOffset"},
@@ -54,6 +55,15 @@ PL_U::PL_U() : ServiceFramework("pl:u") {
     }
 }
 
+void PL_U::RequestLoad(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    const u32 shared_font_type{rp.Pop<u32>()};
+
+    LOG_DEBUG(Service_NS, "called, shared_font_type=%d", shared_font_type);
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+}
+
 void PL_U::GetLoadState(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     const u32 font_id{rp.Pop<u32>()};

src/core/hle/service/ns/pl_u.h

@@ -17,6 +17,7 @@ public:
     ~PL_U() = default;
 
 private:
+    void RequestLoad(Kernel::HLERequestContext& ctx);
     void GetLoadState(Kernel::HLERequestContext& ctx);
     void GetSize(Kernel::HLERequestContext& ctx);
     void GetSharedMemoryAddressOffset(Kernel::HLERequestContext& ctx);

src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp

@@ -26,14 +26,13 @@ void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u3
                 "Drawing from address %lx offset %08X Width %u Height %u Stride %u Format %u", addr,
                 offset, width, height, stride, format);
 
-    using PixelFormat = RendererBase::FramebufferInfo::PixelFormat;
-    using Flags = NVFlinger::BufferQueue::BufferTransformFlags;
-    const bool flip_vertical = static_cast<u32>(transform) & static_cast<u32>(Flags::FlipV);
-    const RendererBase::FramebufferInfo framebuffer_info{
-        addr, offset, width, height, stride, static_cast<PixelFormat>(format), flip_vertical};
+    using PixelFormat = Tegra::FramebufferConfig::PixelFormat;
+    const Tegra::FramebufferConfig framebuffer{
+        addr, offset, width, height, stride, static_cast<PixelFormat>(format), transform};
 
     Core::System::GetInstance().perf_stats.EndGameFrame();
-    VideoCore::g_renderer->SwapBuffers(framebuffer_info);
+
+    VideoCore::g_renderer->SwapBuffers(framebuffer);
 }
 
 } // namespace Devices

src/core/hle/service/nvflinger/buffer_queue.h

@@ -47,6 +47,8 @@ public:
     ~BufferQueue() = default;
 
     enum class BufferTransformFlags : u32 {
+        /// No transform flags are set
+        Unset = 0x00,
         /// Flip source image horizontally (around the vertical axis)
         FlipH = 0x01,
         /// Flip source image vertically (around the horizontal axis)

src/core/hle/service/service.cpp

@@ -20,6 +20,7 @@
 #include "core/hle/service/aoc/aoc_u.h"
 #include "core/hle/service/apm/apm.h"
 #include "core/hle/service/audio/audio.h"
+#include "core/hle/service/fatal/fatal.h"
 #include "core/hle/service/filesystem/filesystem.h"
 #include "core/hle/service/friend/friend.h"
 #include "core/hle/service/hid/hid.h"
@@ -33,6 +34,8 @@
 #include "core/hle/service/sm/controller.h"
 #include "core/hle/service/sm/sm.h"
 #include "core/hle/service/sockets/sockets.h"
+#include "core/hle/service/spl/module.h"
+#include "core/hle/service/ssl/ssl.h"
 #include "core/hle/service/time/time.h"
 #include "core/hle/service/vi/vi.h"
 
@@ -179,6 +182,7 @@ void Init() {
     AOC::InstallInterfaces(*SM::g_service_manager);
     APM::InstallInterfaces(*SM::g_service_manager);
     Audio::InstallInterfaces(*SM::g_service_manager);
+    Fatal::InstallInterfaces(*SM::g_service_manager);
     FileSystem::InstallInterfaces(*SM::g_service_manager);
     Friend::InstallInterfaces(*SM::g_service_manager);
     HID::InstallInterfaces(*SM::g_service_manager);
@@ -188,6 +192,8 @@ void Init() {
     Nvidia::InstallInterfaces(*SM::g_service_manager);
     PCTL::InstallInterfaces(*SM::g_service_manager);
     Sockets::InstallInterfaces(*SM::g_service_manager);
+    SPL::InstallInterfaces(*SM::g_service_manager);
+    SSL::InstallInterfaces(*SM::g_service_manager);
     Time::InstallInterfaces(*SM::g_service_manager);
     VI::InstallInterfaces(*SM::g_service_manager, nv_flinger);
     Set::InstallInterfaces(*SM::g_service_manager);

src/core/hle/service/sockets/bsd.cpp

@@ -3,12 +3,12 @@
 // Refer to the license.txt file included.
 
 #include "core/hle/ipc_helpers.h"
-#include "core/hle/service/sockets/bsd_u.h"
+#include "core/hle/service/sockets/bsd.h"
 
 namespace Service {
 namespace Sockets {
 
-void BSD_U::RegisterClient(Kernel::HLERequestContext& ctx) {
+void BSD::RegisterClient(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 3};
@@ -17,7 +17,7 @@ void BSD_U::RegisterClient(Kernel::HLERequestContext& ctx) {
     rb.Push<u32>(0); // bsd errno
 }
 
-void BSD_U::StartMonitoring(Kernel::HLERequestContext& ctx) {
+void BSD::StartMonitoring(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 3};
@@ -26,7 +26,7 @@ void BSD_U::StartMonitoring(Kernel::HLERequestContext& ctx) {
     rb.Push<u32>(0); // bsd errno
 }
 
-void BSD_U::Socket(Kernel::HLERequestContext& ctx) {
+void BSD::Socket(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
 
     u32 domain = rp.Pop<u32>();
@@ -44,7 +44,7 @@ void BSD_U::Socket(Kernel::HLERequestContext& ctx) {
     rb.Push<u32>(0); // bsd errno
 }
 
-void BSD_U::Connect(Kernel::HLERequestContext& ctx) {
+void BSD::Connect(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 4};
@@ -54,7 +54,7 @@ void BSD_U::Connect(Kernel::HLERequestContext& ctx) {
     rb.Push<u32>(0); // bsd errno
 }
 
-void BSD_U::SendTo(Kernel::HLERequestContext& ctx) {
+void BSD::SendTo(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 4};
@@ -64,7 +64,7 @@ void BSD_U::SendTo(Kernel::HLERequestContext& ctx) {
     rb.Push<u32>(0); // bsd errno
 }
 
-void BSD_U::Close(Kernel::HLERequestContext& ctx) {
+void BSD::Close(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 4};
@@ -74,13 +74,15 @@ void BSD_U::Close(Kernel::HLERequestContext& ctx) {
     rb.Push<u32>(0); // bsd errno
 }
 
-BSD_U::BSD_U() : ServiceFramework("bsd:u") {
-    static const FunctionInfo functions[] = {{0, &BSD_U::RegisterClient, "RegisterClient"},
-                                             {1, &BSD_U::StartMonitoring, "StartMonitoring"},
-                                             {2, &BSD_U::Socket, "Socket"},
-                                             {11, &BSD_U::SendTo, "SendTo"},
-                                             {14, &BSD_U::Connect, "Connect"},
-                                             {26, &BSD_U::Close, "Close"}};
+BSD::BSD(const char* name) : ServiceFramework(name) {
+    static const FunctionInfo functions[] = {
+        {0, &BSD::RegisterClient, "RegisterClient"},
+        {1, &BSD::StartMonitoring, "StartMonitoring"},
+        {2, &BSD::Socket, "Socket"},
+        {11, &BSD::SendTo, "SendTo"},
+        {14, &BSD::Connect, "Connect"},
+        {26, &BSD::Close, "Close"},
+    };
     RegisterHandlers(functions);
 }
 

src/core/hle/service/sockets/bsd.h

@@ -10,10 +10,10 @@
 namespace Service {
 namespace Sockets {
 
-class BSD_U final : public ServiceFramework<BSD_U> {
+class BSD final : public ServiceFramework<BSD> {
 public:
-    BSD_U();
-    ~BSD_U() = default;
+    explicit BSD(const char* name);
+    ~BSD() = default;
 
 private:
     void RegisterClient(Kernel::HLERequestContext& ctx);

src/core/hle/service/sockets/nsd.cpp

@@ -0,0 +1,34 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/service/sockets/nsd.h"
+
+namespace Service {
+namespace Sockets {
+
+NSD::NSD(const char* name) : ServiceFramework(name) {
+    static const FunctionInfo functions[] = {
+        {10, nullptr, "GetSettingName"},
+        {11, nullptr, "GetEnvironmentIdentifier"},
+        {12, nullptr, "GetDeviceId"},
+        {13, nullptr, "DeleteSettings"},
+        {14, nullptr, "ImportSettings"},
+        {20, nullptr, "Resolve"},
+        {21, nullptr, "ResolveEx"},
+        {30, nullptr, "GetNasServiceSetting"},
+        {31, nullptr, "GetNasServiceSettingEx"},
+        {40, nullptr, "GetNasRequestFqdn"},
+        {41, nullptr, "GetNasRequestFqdnEx"},
+        {42, nullptr, "GetNasApiFqdn"},
+        {43, nullptr, "GetNasApiFqdnEx"},
+        {50, nullptr, "GetCurrentSetting"},
+        {60, nullptr, "ReadSaveDataFromFsForTest"},
+        {61, nullptr, "WriteSaveDataToFsForTest"},
+        {62, nullptr, "DeleteSaveDataOfFsForTest"},
+    };
+    RegisterHandlers(functions);
+}
+
+} // namespace Sockets
+} // namespace Service

src/core/hle/service/sockets/nsd.h

@@ -0,0 +1,20 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/kernel/hle_ipc.h"
+#include "core/hle/service/service.h"
+
+namespace Service {
+namespace Sockets {
+
+class NSD final : public ServiceFramework<NSD> {
+public:
+    explicit NSD(const char* name);
+    ~NSD() = default;
+};
+
+} // namespace Sockets
+} // namespace Service

src/core/hle/service/sockets/sfdnsres.cpp

@@ -19,16 +19,18 @@ void SFDNSRES::GetAddrInfo(Kernel::HLERequestContext& ctx) {
 }
 
 SFDNSRES::SFDNSRES() : ServiceFramework("sfdnsres") {
-    static const FunctionInfo functions[] = {{0, nullptr, "SetDnsAddressesPrivate"},
-                                             {1, nullptr, "GetDnsAddressPrivate"},
-                                             {2, nullptr, "GetHostByName"},
-                                             {3, nullptr, "GetHostByAddr"},
-                                             {4, nullptr, "GetHostStringError"},
-                                             {5, nullptr, "GetGaiStringError"},
-                                             {6, &SFDNSRES::GetAddrInfo, "GetAddrInfo"},
-                                             {7, nullptr, "GetNameInfo"},
-                                             {8, nullptr, "RequestCancelHandle"},
-                                             {9, nullptr, "CancelSocketCall"}};
+    static const FunctionInfo functions[] = {
+        {0, nullptr, "SetDnsAddressesPrivate"},
+        {1, nullptr, "GetDnsAddressPrivate"},
+        {2, nullptr, "GetHostByName"},
+        {3, nullptr, "GetHostByAddr"},
+        {4, nullptr, "GetHostStringError"},
+        {5, nullptr, "GetGaiStringError"},
+        {6, &SFDNSRES::GetAddrInfo, "GetAddrInfo"},
+        {7, nullptr, "GetNameInfo"},
+        {8, nullptr, "RequestCancelHandle"},
+        {9, nullptr, "CancelSocketCall"},
+    };
     RegisterHandlers(functions);
 }
 

src/core/hle/service/sockets/sfdnsres.h

@@ -12,7 +12,7 @@ namespace Sockets {
 
 class SFDNSRES final : public ServiceFramework<SFDNSRES> {
 public:
-    SFDNSRES();
+    explicit SFDNSRES();
     ~SFDNSRES() = default;
 
 private:

src/core/hle/service/sockets/sockets.cpp

@@ -2,7 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "core/hle/service/sockets/bsd_u.h"
+#include "core/hle/service/sockets/bsd.h"
+#include "core/hle/service/sockets/nsd.h"
 #include "core/hle/service/sockets/sfdnsres.h"
 #include "core/hle/service/sockets/sockets.h"
 
@@ -10,7 +11,10 @@ namespace Service {
 namespace Sockets {
 
 void InstallInterfaces(SM::ServiceManager& service_manager) {
-    std::make_shared<BSD_U>()->InstallAsService(service_manager);
+    std::make_shared<BSD>("bsd:s")->InstallAsService(service_manager);
+    std::make_shared<BSD>("bsd:u")->InstallAsService(service_manager);
+    std::make_shared<NSD>("nsd:a")->InstallAsService(service_manager);
+    std::make_shared<NSD>("nsd:u")->InstallAsService(service_manager);
     std::make_shared<SFDNSRES>()->InstallAsService(service_manager);
 }
 

src/core/hle/service/spl/csrng.cpp

@@ -0,0 +1,18 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/service/spl/csrng.h"
+
+namespace Service {
+namespace SPL {
+
+CSRNG::CSRNG(std::shared_ptr<Module> module) : Module::Interface(std::move(module), "csrng") {
+    static const FunctionInfo functions[] = {
+        {0, &CSRNG::GetRandomBytes, "GetRandomBytes"},
+    };
+    RegisterHandlers(functions);
+}
+
+} // namespace SPL
+} // namespace Service

src/core/hle/service/spl/csrng.h

@@ -0,0 +1,18 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/spl/module.h"
+
+namespace Service {
+namespace SPL {
+
+class CSRNG final : public Module::Interface {
+public:
+    explicit CSRNG(std::shared_ptr<Module> module);
+};
+
+} // namespace SPL
+} // namespace Service

src/core/hle/service/spl/module.cpp

@@ -0,0 +1,42 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <cstdlib>
+#include <vector>
+#include "common/logging/log.h"
+#include "core/hle/ipc_helpers.h"
+#include "core/hle/service/spl/csrng.h"
+#include "core/hle/service/spl/module.h"
+#include "core/hle/service/spl/spl.h"
+
+namespace Service {
+namespace SPL {
+
+Module::Interface::Interface(std::shared_ptr<Module> module, const char* name)
+    : ServiceFramework(name), module(std::move(module)) {}
+
+void Module::Interface::GetRandomBytes(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+
+    size_t size = ctx.GetWriteBufferSize();
+
+    std::vector<u8> data(size);
+    std::generate(data.begin(), data.end(), std::rand);
+
+    ctx.WriteBuffer(data);
+
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+    LOG_DEBUG(Service_SPL, "called");
+}
+
+void InstallInterfaces(SM::ServiceManager& service_manager) {
+    auto module = std::make_shared<Module>();
+    std::make_shared<CSRNG>(module)->InstallAsService(service_manager);
+    std::make_shared<SPL>(module)->InstallAsService(service_manager);
+}
+
+} // namespace SPL
+} // namespace Service

src/core/hle/service/spl/module.h

@@ -0,0 +1,29 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/service.h"
+
+namespace Service {
+namespace SPL {
+
+class Module final {
+public:
+    class Interface : public ServiceFramework<Interface> {
+    public:
+        Interface(std::shared_ptr<Module> module, const char* name);
+
+        void GetRandomBytes(Kernel::HLERequestContext& ctx);
+
+    protected:
+        std::shared_ptr<Module> module;
+    };
+};
+
+/// Registers all SPL services with the specified service manager.
+void InstallInterfaces(SM::ServiceManager& service_manager);
+
+} // namespace SPL
+} // namespace Service

src/core/hle/service/spl/spl.cpp

@@ -0,0 +1,41 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/service/spl/spl.h"
+
+namespace Service {
+namespace SPL {
+
+SPL::SPL(std::shared_ptr<Module> module) : Module::Interface(std::move(module), "spl:") {
+    static const FunctionInfo functions[] = {
+        {0, nullptr, "GetConfig"},
+        {1, nullptr, "UserExpMod"},
+        {2, nullptr, "GenerateAesKek"},
+        {3, nullptr, "LoadAesKey"},
+        {4, nullptr, "GenerateAesKey"},
+        {5, nullptr, "SetConfig"},
+        {7, &SPL::GetRandomBytes, "GetRandomBytes"},
+        {9, nullptr, "LoadSecureExpModKey"},
+        {10, nullptr, "SecureExpMod"},
+        {11, nullptr, "IsDevelopment"},
+        {12, nullptr, "GenerateSpecificAesKey"},
+        {13, nullptr, "DecryptPrivk"},
+        {14, nullptr, "DecryptAesKey"},
+        {15, nullptr, "DecryptAesCtr"},
+        {16, nullptr, "ComputeCmac"},
+        {17, nullptr, "LoadRsaOaepKey"},
+        {18, nullptr, "UnwrapRsaOaepWrappedTitleKey"},
+        {19, nullptr, "LoadTitleKey"},
+        {20, nullptr, "UnwrapAesWrappedTitleKey"},
+        {21, nullptr, "LockAesEngine"},
+        {22, nullptr, "UnlockAesEngine"},
+        {23, nullptr, "GetSplWaitEvent"},
+        {24, nullptr, "SetSharedData"},
+        {25, nullptr, "GetSharedData"},
+    };
+    RegisterHandlers(functions);
+}
+
+} // namespace SPL
+} // namespace Service

src/core/hle/service/spl/spl.h

@@ -0,0 +1,18 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/spl/module.h"
+
+namespace Service {
+namespace SPL {
+
+class SPL final : public Module::Interface {
+public:
+    explicit SPL(std::shared_ptr<Module> module);
+};
+
+} // namespace SPL
+} // namespace Service

src/core/hle/service/ssl/ssl.cpp

@@ -0,0 +1,17 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/service/ssl/ssl.h"
+
+namespace Service {
+namespace SSL {
+
+SSL::SSL() : ServiceFramework("ssl") {}
+
+void InstallInterfaces(SM::ServiceManager& service_manager) {
+    std::make_shared<SSL>()->InstallAsService(service_manager);
+}
+
+} // namespace SSL
+} // namespace Service

src/core/hle/service/ssl/ssl.h

@@ -0,0 +1,22 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/service.h"
+
+namespace Service {
+namespace SSL {
+
+class SSL final : public ServiceFramework<SSL> {
+public:
+    explicit SSL();
+    ~SSL() = default;
+};
+
+/// Registers all SSL services with the specified service manager.
+void InstallInterfaces(SM::ServiceManager& service_manager);
+
+} // namespace SSL
+} // namespace Service

src/core/hle/service/vi/vi.cpp

@@ -4,10 +4,13 @@
 
 #include <algorithm>
 #include <array>
+#include <memory>
+#include <boost/optional.hpp>
 #include "common/alignment.h"
 #include "common/scope_exit.h"
 #include "core/core_timing.h"
 #include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/event.h"
 #include "core/hle/service/nvdrv/nvdrv.h"
 #include "core/hle/service/nvflinger/buffer_queue.h"
 #include "core/hle/service/vi/vi.h"
@@ -646,144 +649,152 @@ private:
     std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
 };
 
-void IApplicationDisplayService::GetRelayService(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_VI, "(STUBBED) called");
+class IApplicationDisplayService final : public ServiceFramework<IApplicationDisplayService> {
+public:
+    IApplicationDisplayService(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
+    ~IApplicationDisplayService() = default;
 
-    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
-    rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<IHOSBinderDriver>(nv_flinger);
-}
+private:
+    void GetRelayService(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_VI, "(STUBBED) called");
 
-void IApplicationDisplayService::GetSystemDisplayService(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_VI, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<IHOSBinderDriver>(nv_flinger);
+    }
 
-    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
-    rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<ISystemDisplayService>();
-}
+    void GetSystemDisplayService(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_VI, "(STUBBED) called");
 
-void IApplicationDisplayService::GetManagerDisplayService(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_VI, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<ISystemDisplayService>();
+    }
 
-    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
-    rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<IManagerDisplayService>(nv_flinger);
-}
+    void GetManagerDisplayService(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_VI, "(STUBBED) called");
 
-void IApplicationDisplayService::GetIndirectDisplayTransactionService(
-    Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_VI, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<IManagerDisplayService>(nv_flinger);
+    }
 
-    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
-    rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<IHOSBinderDriver>(nv_flinger);
-}
+    void GetIndirectDisplayTransactionService(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_VI, "(STUBBED) called");
 
-void IApplicationDisplayService::OpenDisplay(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_VI, "(STUBBED) called");
-    IPC::RequestParser rp{ctx};
-    auto name_buf = rp.PopRaw<std::array<u8, 0x40>>();
-    auto end = std::find(name_buf.begin(), name_buf.end(), '\0');
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<IHOSBinderDriver>(nv_flinger);
+    }
 
-    std::string name(name_buf.begin(), end);
+    void OpenDisplay(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_VI, "(STUBBED) called");
+        IPC::RequestParser rp{ctx};
+        auto name_buf = rp.PopRaw<std::array<u8, 0x40>>();
+        auto end = std::find(name_buf.begin(), name_buf.end(), '\0');
 
-    ASSERT_MSG(name == "Default", "Non-default displays aren't supported yet");
+        std::string name(name_buf.begin(), end);
 
-    IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
-    rb.Push(RESULT_SUCCESS);
-    rb.Push<u64>(nv_flinger->OpenDisplay(name));
-}
+        ASSERT_MSG(name == "Default", "Non-default displays aren't supported yet");
 
-void IApplicationDisplayService::CloseDisplay(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_VI, "(STUBBED) called");
-    IPC::RequestParser rp{ctx};
-    u64 display_id = rp.Pop<u64>();
+        IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u64>(nv_flinger->OpenDisplay(name));
+    }
 
-    IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
-    rb.Push(RESULT_SUCCESS);
-}
+    void CloseDisplay(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_VI, "(STUBBED) called");
+        IPC::RequestParser rp{ctx};
+        u64 display_id = rp.Pop<u64>();
 
-void IApplicationDisplayService::OpenLayer(Kernel::HLERequestContext& ctx) {
-    LOG_DEBUG(Service_VI, "called");
-    IPC::RequestParser rp{ctx};
-    auto name_buf = rp.PopRaw<std::array<u8, 0x40>>();
-    auto end = std::find(name_buf.begin(), name_buf.end(), '\0');
+        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        rb.Push(RESULT_SUCCESS);
+    }
 
-    std::string display_name(name_buf.begin(), end);
+    void SetLayerScalingMode(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_VI, "(STUBBED) called");
+        IPC::RequestParser rp{ctx};
+        u32 scaling_mode = rp.Pop<u32>();
+        u64 unknown = rp.Pop<u64>();
 
-    u64 layer_id = rp.Pop<u64>();
-    u64 aruid = rp.Pop<u64>();
+        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        rb.Push(RESULT_SUCCESS);
+    }
 
-    u64 display_id = nv_flinger->OpenDisplay(display_name);
-    u32 buffer_queue_id = nv_flinger->GetBufferQueueId(display_id, layer_id);
+    void ListDisplays(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        DisplayInfo display_info;
+        ctx.WriteBuffer(&display_info, sizeof(DisplayInfo));
+        IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u64>(1);
+        LOG_WARNING(Service_VI, "(STUBBED) called");
+    }
 
-    NativeWindow native_window{buffer_queue_id};
-    IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
-    rb.Push(RESULT_SUCCESS);
-    rb.Push<u64>(ctx.WriteBuffer(native_window.Serialize()));
-}
+    void OpenLayer(Kernel::HLERequestContext& ctx) {
+        LOG_DEBUG(Service_VI, "called");
+        IPC::RequestParser rp{ctx};
+        auto name_buf = rp.PopRaw<std::array<u8, 0x40>>();
+        auto end = std::find(name_buf.begin(), name_buf.end(), '\0');
 
-void IApplicationDisplayService::CreateStrayLayer(Kernel::HLERequestContext& ctx) {
-    LOG_DEBUG(Service_VI, "called");
+        std::string display_name(name_buf.begin(), end);
 
-    IPC::RequestParser rp{ctx};
-    u32 flags = rp.Pop<u32>();
-    rp.Pop<u32>(); // padding
-    u64 display_id = rp.Pop<u64>();
+        u64 layer_id = rp.Pop<u64>();
+        u64 aruid = rp.Pop<u64>();
 
-    // TODO(Subv): What's the difference between a Stray and a Managed layer?
+        u64 display_id = nv_flinger->OpenDisplay(display_name);
+        u32 buffer_queue_id = nv_flinger->GetBufferQueueId(display_id, layer_id);
 
-    u64 layer_id = nv_flinger->CreateLayer(display_id);
-    u32 buffer_queue_id = nv_flinger->GetBufferQueueId(display_id, layer_id);
+        NativeWindow native_window{buffer_queue_id};
+        IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u64>(ctx.WriteBuffer(native_window.Serialize()));
+    }
 
-    NativeWindow native_window{buffer_queue_id};
-    IPC::ResponseBuilder rb = rp.MakeBuilder(6, 0, 0);
-    rb.Push(RESULT_SUCCESS);
-    rb.Push(layer_id);
-    rb.Push<u64>(ctx.WriteBuffer(native_window.Serialize()));
-}
+    void CreateStrayLayer(Kernel::HLERequestContext& ctx) {
+        LOG_DEBUG(Service_VI, "called");
 
-void IApplicationDisplayService::DestroyStrayLayer(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_VI, "(STUBBED) called");
+        IPC::RequestParser rp{ctx};
+        u32 flags = rp.Pop<u32>();
+        rp.Pop<u32>(); // padding
+        u64 display_id = rp.Pop<u64>();
 
-    IPC::RequestParser rp{ctx};
-    u64 layer_id = rp.Pop<u64>();
+        // TODO(Subv): What's the difference between a Stray and a Managed layer?
 
-    IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
-    rb.Push(RESULT_SUCCESS);
-}
+        u64 layer_id = nv_flinger->CreateLayer(display_id);
+        u32 buffer_queue_id = nv_flinger->GetBufferQueueId(display_id, layer_id);
 
-void IApplicationDisplayService::SetLayerScalingMode(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_VI, "(STUBBED) called");
-    IPC::RequestParser rp{ctx};
-    u32 scaling_mode = rp.Pop<u32>();
-    u64 unknown = rp.Pop<u64>();
+        NativeWindow native_window{buffer_queue_id};
+        IPC::ResponseBuilder rb = rp.MakeBuilder(6, 0, 0);
+        rb.Push(RESULT_SUCCESS);
+        rb.Push(layer_id);
+        rb.Push<u64>(ctx.WriteBuffer(native_window.Serialize()));
+    }
 
-    IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
-    rb.Push(RESULT_SUCCESS);
-}
+    void DestroyStrayLayer(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_VI, "(STUBBED) called");
 
-void IApplicationDisplayService::ListDisplays(Kernel::HLERequestContext& ctx) {
-    IPC::RequestParser rp{ctx};
-    DisplayInfo display_info;
-    ctx.WriteBuffer(&display_info, sizeof(DisplayInfo));
-    IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
-    rb.Push(RESULT_SUCCESS);
-    rb.Push<u64>(1);
-    LOG_WARNING(Service_VI, "(STUBBED) called");
-}
+        IPC::RequestParser rp{ctx};
+        u64 layer_id = rp.Pop<u64>();
 
-void IApplicationDisplayService::GetDisplayVsyncEvent(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_VI, "(STUBBED) called");
-    IPC::RequestParser rp{ctx};
-    u64 display_id = rp.Pop<u64>();
+        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
+        rb.Push(RESULT_SUCCESS);
+    }
 
-    auto vsync_event = nv_flinger->GetVsyncEvent(display_id);
+    void GetDisplayVsyncEvent(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_VI, "(STUBBED) called");
+        IPC::RequestParser rp{ctx};
+        u64 display_id = rp.Pop<u64>();
 
-    IPC::ResponseBuilder rb = rp.MakeBuilder(2, 1, 0);
-    rb.Push(RESULT_SUCCESS);
-    rb.PushCopyObjects(vsync_event);
-}
+        auto vsync_event = nv_flinger->GetVsyncEvent(display_id);
+
+        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 1, 0);
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(vsync_event);
+    }
+
+    std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
+};
 
 IApplicationDisplayService::IApplicationDisplayService(
     std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
@@ -806,11 +817,24 @@ IApplicationDisplayService::IApplicationDisplayService(
     RegisterHandlers(functions);
 }
 
+Module::Interface::Interface(std::shared_ptr<Module> module, const char* name,
+                             std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
+    : ServiceFramework(name), module(std::move(module)), nv_flinger(std::move(nv_flinger)) {}
+
+void Module::Interface::GetDisplayService(Kernel::HLERequestContext& ctx) {
+    LOG_WARNING(Service_VI, "(STUBBED) called");
+
+    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushIpcInterface<IApplicationDisplayService>(nv_flinger);
+}
+
 void InstallInterfaces(SM::ServiceManager& service_manager,
                        std::shared_ptr<NVFlinger::NVFlinger> nv_flinger) {
-    std::make_shared<VI_M>(nv_flinger)->InstallAsService(service_manager);
-    std::make_shared<VI_S>(nv_flinger)->InstallAsService(service_manager);
-    std::make_shared<VI_U>(nv_flinger)->InstallAsService(service_manager);
+    auto module = std::make_shared<Module>();
+    std::make_shared<VI_M>(module, nv_flinger)->InstallAsService(service_manager);
+    std::make_shared<VI_S>(module, nv_flinger)->InstallAsService(service_manager);
+    std::make_shared<VI_U>(module, nv_flinger)->InstallAsService(service_manager);
 }
 
 } // namespace VI

src/core/hle/service/vi/vi.h

@@ -4,9 +4,6 @@
 
 #pragma once
 
-#include <memory>
-#include <boost/optional.hpp>
-#include "core/hle/kernel/event.h"
 #include "core/hle/service/nvflinger/nvflinger.h"
 #include "core/hle/service/service.h"
 
@@ -17,26 +14,19 @@ struct EventType;
 namespace Service {
 namespace VI {
 
-class IApplicationDisplayService final : public ServiceFramework<IApplicationDisplayService> {
+class Module final {
 public:
-    IApplicationDisplayService(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
-    ~IApplicationDisplayService() = default;
+    class Interface : public ServiceFramework<Interface> {
+    public:
+        Interface(std::shared_ptr<Module> module, const char* name,
+                  std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
 
-private:
-    void GetRelayService(Kernel::HLERequestContext& ctx);
-    void GetSystemDisplayService(Kernel::HLERequestContext& ctx);
-    void GetManagerDisplayService(Kernel::HLERequestContext& ctx);
-    void GetIndirectDisplayTransactionService(Kernel::HLERequestContext& ctx);
-    void OpenDisplay(Kernel::HLERequestContext& ctx);
-    void CloseDisplay(Kernel::HLERequestContext& ctx);
-    void SetLayerScalingMode(Kernel::HLERequestContext& ctx);
-    void ListDisplays(Kernel::HLERequestContext& ctx);
-    void OpenLayer(Kernel::HLERequestContext& ctx);
-    void CreateStrayLayer(Kernel::HLERequestContext& ctx);
-    void DestroyStrayLayer(Kernel::HLERequestContext& ctx);
-    void GetDisplayVsyncEvent(Kernel::HLERequestContext& ctx);
+        void GetDisplayService(Kernel::HLERequestContext& ctx);
 
-    std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
+    protected:
+        std::shared_ptr<Module> module;
+        std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
+    };
 };
 
 /// Registers all VI services with the specified service manager.

src/core/hle/service/vi/vi_m.cpp

@@ -2,24 +2,13 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "common/logging/log.h"
-#include "core/hle/ipc_helpers.h"
-#include "core/hle/service/vi/vi.h"
 #include "core/hle/service/vi/vi_m.h"
 
 namespace Service {
 namespace VI {
 
-void VI_M::GetDisplayService(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_VI, "(STUBBED) called");
-
-    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
-    rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<IApplicationDisplayService>(nv_flinger);
-}
-
-VI_M::VI_M(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
-    : ServiceFramework("vi:m"), nv_flinger(std::move(nv_flinger)) {
+VI_M::VI_M(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
+    : Module::Interface(std::move(module), "vi:m", std::move(nv_flinger)) {
     static const FunctionInfo functions[] = {
         {2, &VI_M::GetDisplayService, "GetDisplayService"},
         {3, nullptr, "GetDisplayServiceWithProxyNameExchange"},

src/core/hle/service/vi/vi_m.h

@@ -4,25 +4,14 @@
 
 #pragma once
 
-#include <memory>
-#include "core/hle/service/service.h"
+#include "core/hle/service/vi/vi.h"
 
 namespace Service {
-namespace NVFlinger {
-class NVFlinger;
-}
-
 namespace VI {
 
-class VI_M final : public ServiceFramework<VI_M> {
+class VI_M final : public Module::Interface {
 public:
-    VI_M(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
-    ~VI_M() = default;
-
-private:
-    void GetDisplayService(Kernel::HLERequestContext& ctx);
-
-    std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
+    explicit VI_M(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
 };
 
 } // namespace VI

src/core/hle/service/vi/vi_s.cpp

@@ -2,24 +2,13 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "common/logging/log.h"
-#include "core/hle/ipc_helpers.h"
-#include "core/hle/service/vi/vi.h"
 #include "core/hle/service/vi/vi_s.h"
 
 namespace Service {
 namespace VI {
 
-void VI_S::GetDisplayService(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_VI, "(STUBBED) called");
-
-    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
-    rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<IApplicationDisplayService>(nv_flinger);
-}
-
-VI_S::VI_S(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
-    : ServiceFramework("vi:s"), nv_flinger(std::move(nv_flinger)) {
+VI_S::VI_S(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
+    : Module::Interface(std::move(module), "vi:s", std::move(nv_flinger)) {
     static const FunctionInfo functions[] = {
         {1, &VI_S::GetDisplayService, "GetDisplayService"},
         {3, nullptr, "GetDisplayServiceWithProxyNameExchange"},

src/core/hle/service/vi/vi_s.h

@@ -4,25 +4,14 @@
 
 #pragma once
 
-#include <memory>
-#include "core/hle/service/service.h"
+#include "core/hle/service/vi/vi.h"
 
 namespace Service {
-namespace NVFlinger {
-class NVFlinger;
-}
-
 namespace VI {
 
-class VI_S final : public ServiceFramework<VI_S> {
+class VI_S final : public Module::Interface {
 public:
-    VI_S(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
-    ~VI_S() = default;
-
-private:
-    void GetDisplayService(Kernel::HLERequestContext& ctx);
-
-    std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
+    explicit VI_S(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
 };
 
 } // namespace VI

src/core/hle/service/vi/vi_u.cpp

@@ -2,24 +2,13 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "common/logging/log.h"
-#include "core/hle/ipc_helpers.h"
-#include "core/hle/service/vi/vi.h"
 #include "core/hle/service/vi/vi_u.h"
 
 namespace Service {
 namespace VI {
 
-void VI_U::GetDisplayService(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_VI, "(STUBBED) called");
-
-    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
-    rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<IApplicationDisplayService>(nv_flinger);
-}
-
-VI_U::VI_U(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
-    : ServiceFramework("vi:u"), nv_flinger(std::move(nv_flinger)) {
+VI_U::VI_U(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger)
+    : Module::Interface(std::move(module), "vi:u", std::move(nv_flinger)) {
     static const FunctionInfo functions[] = {
         {0, &VI_U::GetDisplayService, "GetDisplayService"},
         {3, nullptr, "GetDisplayServiceWithProxyNameExchange"},

src/core/hle/service/vi/vi_u.h

@@ -4,25 +4,14 @@
 
 #pragma once
 
-#include <memory>
-#include "core/hle/service/service.h"
+#include "core/hle/service/vi/vi.h"
 
 namespace Service {
-namespace NVFlinger {
-class NVFlinger;
-}
-
 namespace VI {
 
-class VI_U final : public ServiceFramework<VI_U> {
+class VI_U final : public Module::Interface {
 public:
-    VI_U(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
-    ~VI_U() = default;
-
-private:
-    void GetDisplayService(Kernel::HLERequestContext& ctx);
-
-    std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
+    explicit VI_U(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger);
 };
 
 } // namespace VI

src/core/memory.cpp

@@ -15,6 +15,7 @@
 #include "core/core.h"
 #include "core/hle/kernel/memory.h"
 #include "core/hle/kernel/process.h"
+#include "core/hle/lock.h"
 #include "core/memory.h"
 #include "core/memory_setup.h"
 #include "video_core/renderer_base.h"
@@ -23,7 +24,6 @@
 namespace Memory {
 
 static std::array<u8, Memory::VRAM_SIZE> vram;
-static std::array<u8, Memory::N3DS_EXTRA_RAM_SIZE> n3ds_extra_ram;
 
 static PageTable* current_page_table = nullptr;
 
@@ -42,6 +42,9 @@ static void MapPages(PageTable& page_table, VAddr base, u64 size, u8* memory, Pa
     LOG_DEBUG(HW_Memory, "Mapping %p onto %016" PRIX64 "-%016" PRIX64, memory, base * PAGE_SIZE,
               (base + size) * PAGE_SIZE);
 
+    RasterizerFlushVirtualRegion(base << PAGE_BITS, size * PAGE_SIZE,
+                                 FlushMode::FlushAndInvalidate);
+
     VAddr end = base + size;
     while (base != end) {
         ASSERT_MSG(base < PAGE_TABLE_NUM_ENTRIES, "out of range mapping at %016" PRIX64, base);
@@ -113,91 +116,120 @@ static std::set<MemoryHookPointer> GetSpecialHandlers(VAddr vaddr, u64 size) {
     return GetSpecialHandlers(page_table, vaddr, size);
 }
 
-template <typename T>
-boost::optional<T> ReadSpecial(VAddr addr);
+/**
+ * Gets a pointer to the exact memory at the virtual address (i.e. not page aligned)
+ * using a VMA from the current process
+ */
+static u8* GetPointerFromVMA(const Kernel::Process& process, VAddr vaddr) {
+    u8* direct_pointer = nullptr;
+
+    auto& vm_manager = process.vm_manager;
+
+    auto it = vm_manager.FindVMA(vaddr);
+    ASSERT(it != vm_manager.vma_map.end());
+
+    auto& vma = it->second;
+    switch (vma.type) {
+    case Kernel::VMAType::AllocatedMemoryBlock:
+        direct_pointer = vma.backing_block->data() + vma.offset;
+        break;
+    case Kernel::VMAType::BackingMemory:
+        direct_pointer = vma.backing_memory;
+        break;
+    case Kernel::VMAType::Free:
+        return nullptr;
+    default:
+        UNREACHABLE();
+    }
+
+    return direct_pointer + (vaddr - vma.base);
+}
+
+/**
+ * Gets a pointer to the exact memory at the virtual address (i.e. not page aligned)
+ * using a VMA from the current process.
+ */
+static u8* GetPointerFromVMA(VAddr vaddr) {
+    return GetPointerFromVMA(*Core::CurrentProcess(), vaddr);
+}
 
 template <typename T>
 T Read(const VAddr vaddr) {
-    if ((vaddr >> PAGE_BITS) >= PAGE_TABLE_NUM_ENTRIES) {
-        LOG_ERROR(HW_Memory, "Read%lu after page table @ 0x%016" PRIX64, sizeof(T) * 8, vaddr);
-        return 0;
-    }
-
-    const PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
-    switch (type) {
-    case PageType::Unmapped:
-        LOG_ERROR(HW_Memory, "unmapped Read%zu @ 0x%016" PRIX64, sizeof(T) * 8, vaddr);
-        return 0;
-    case PageType::Special: {
-        if (auto result = ReadSpecial<T>(vaddr))
-            return *result;
-        [[fallthrough]];
-    }
-    case PageType::Memory: {
-        const u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
-        ASSERT_MSG(page_pointer, "Mapped memory page without a pointer @ %016" PRIX64, vaddr);
-
+    const u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
+    if (page_pointer) {
+        // NOTE: Avoid adding any extra logic to this fast-path block
         T value;
         std::memcpy(&value, &page_pointer[vaddr & PAGE_MASK], sizeof(T));
         return value;
     }
+
+    // The memory access might do an MMIO or cached access, so we have to lock the HLE kernel state
+    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
+
+    PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
+    switch (type) {
+    case PageType::Unmapped:
+        LOG_ERROR(HW_Memory, "unmapped Read%lu @ 0x%08X", sizeof(T) * 8, vaddr);
+        return 0;
+    case PageType::Memory:
+        ASSERT_MSG(false, "Mapped memory page without a pointer @ %08X", vaddr);
+        break;
+    case PageType::RasterizerCachedMemory: {
+        RasterizerFlushVirtualRegion(vaddr, sizeof(T), FlushMode::Flush);
+
+        T value;
+        std::memcpy(&value, GetPointerFromVMA(vaddr), sizeof(T));
+        return value;
+    }
+    default:
+        UNREACHABLE();
     }
-    UNREACHABLE();
-    return 0;
 }
 
-template <typename T>
-bool WriteSpecial(VAddr addr, const T data);
-
 template <typename T>
 void Write(const VAddr vaddr, const T data) {
-    if ((vaddr >> PAGE_BITS) >= PAGE_TABLE_NUM_ENTRIES) {
-        LOG_ERROR(HW_Memory, "Write%lu after page table 0x%08X @ 0x%016" PRIX64, sizeof(data) * 8,
-                  (u32)data, vaddr);
-        return;
-    }
-
-    const PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
-    switch (type) {
-    case PageType::Unmapped:
-        LOG_ERROR(HW_Memory, "unmapped Write%zu 0x%08X @ 0x%016" PRIX64, sizeof(data) * 8,
-                  static_cast<u32>(data), vaddr);
-        return;
-    case PageType::Special: {
-        if (WriteSpecial<T>(vaddr, data))
-            return;
-        [[fallthrough]];
-    }
-    case PageType::Memory: {
-        u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
-        ASSERT_MSG(page_pointer, "Mapped memory page without a pointer @ %016" PRIX64, vaddr);
+    u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
+    if (page_pointer) {
+        // NOTE: Avoid adding any extra logic to this fast-path block
         std::memcpy(&page_pointer[vaddr & PAGE_MASK], &data, sizeof(T));
         return;
     }
+
+    // The memory access might do an MMIO or cached access, so we have to lock the HLE kernel state
+    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
+
+    PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
+    switch (type) {
+    case PageType::Unmapped:
+        LOG_ERROR(HW_Memory, "unmapped Write%lu 0x%08X @ 0x%08X", sizeof(data) * 8, (u32)data,
+                  vaddr);
+        return;
+    case PageType::Memory:
+        ASSERT_MSG(false, "Mapped memory page without a pointer @ %08X", vaddr);
+        break;
+    case PageType::RasterizerCachedMemory: {
+        RasterizerFlushVirtualRegion(vaddr, sizeof(T), FlushMode::Invalidate);
+        std::memcpy(GetPointerFromVMA(vaddr), &data, sizeof(T));
+        break;
+    }
+    default:
+        UNREACHABLE();
     }
-    UNREACHABLE();
 }
 
 bool IsValidVirtualAddress(const Kernel::Process& process, const VAddr vaddr) {
     auto& page_table = process.vm_manager.page_table;
 
-    if ((vaddr >> PAGE_BITS) >= PAGE_TABLE_NUM_ENTRIES)
+    const u8* page_pointer = page_table.pointers[vaddr >> PAGE_BITS];
+    if (page_pointer)
+        return true;
+
+    if (page_table.attributes[vaddr >> PAGE_BITS] == PageType::RasterizerCachedMemory)
+        return true;
+
+    if (page_table.attributes[vaddr >> PAGE_BITS] != PageType::Special)
         return false;
 
-    const PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
-    switch (type) {
-    case PageType::Unmapped:
-        return false;
-    case PageType::Memory:
-        return true;
-    case PageType::Special: {
-        for (auto handler : GetSpecialHandlers(page_table, vaddr, 1))
-            if (auto result = handler->IsValidAddress(vaddr))
-                return *result;
-        return current_page_table->pointers[vaddr >> PAGE_BITS] != nullptr;
-    }
-    }
-    UNREACHABLE();
     return false;
 }
 
@@ -215,7 +247,11 @@ u8* GetPointer(const VAddr vaddr) {
         return page_pointer + (vaddr & PAGE_MASK);
     }
 
-    LOG_ERROR(HW_Memory, "unknown GetPointer @ 0x%016" PRIx64, vaddr);
+    if (current_page_table->attributes[vaddr >> PAGE_BITS] == PageType::RasterizerCachedMemory) {
+        return GetPointerFromVMA(vaddr);
+    }
+
+    LOG_ERROR(HW_Memory, "unknown GetPointer @ 0x%08x", vaddr);
     return nullptr;
 }
 
@@ -244,7 +280,6 @@ u8* GetPhysicalPointer(PAddr address) {
         {IO_AREA_PADDR, IO_AREA_SIZE},
         {DSP_RAM_PADDR, DSP_RAM_SIZE},
         {FCRAM_PADDR, FCRAM_N3DS_SIZE},
-        {N3DS_EXTRA_RAM_PADDR, N3DS_EXTRA_RAM_SIZE},
     };
 
     const auto area =
@@ -283,9 +318,6 @@ u8* GetPhysicalPointer(PAddr address) {
         }
         ASSERT_MSG(target_pointer != nullptr, "Invalid FCRAM address");
         break;
-    case N3DS_EXTRA_RAM_PADDR:
-        target_pointer = n3ds_extra_ram.data() + offset_into_region;
-        break;
     default:
         UNREACHABLE();
     }
@@ -293,6 +325,95 @@ u8* GetPhysicalPointer(PAddr address) {
     return target_pointer;
 }
 
+void RasterizerMarkRegionCached(VAddr start, u64 size, bool cached) {
+    if (start == 0) {
+        return;
+    }
+
+    u64 num_pages = ((start + size - 1) >> PAGE_BITS) - (start >> PAGE_BITS) + 1;
+    VAddr vaddr = start;
+
+    for (unsigned i = 0; i < num_pages; ++i, vaddr += PAGE_SIZE) {
+        PageType& page_type = current_page_table->attributes[vaddr >> PAGE_BITS];
+
+        if (cached) {
+            // Switch page type to cached if now cached
+            switch (page_type) {
+            case PageType::Unmapped:
+                // It is not necessary for a process to have this region mapped into its address
+                // space, for example, a system module need not have a VRAM mapping.
+                break;
+            case PageType::Memory:
+                page_type = PageType::RasterizerCachedMemory;
+                current_page_table->pointers[vaddr >> PAGE_BITS] = nullptr;
+                break;
+            default:
+                UNREACHABLE();
+            }
+        } else {
+            // Switch page type to uncached if now uncached
+            switch (page_type) {
+            case PageType::Unmapped:
+                // It is not necessary for a process to have this region mapped into its address
+                // space, for example, a system module need not have a VRAM mapping.
+                break;
+            case PageType::RasterizerCachedMemory: {
+                u8* pointer = GetPointerFromVMA(vaddr & ~PAGE_MASK);
+                if (pointer == nullptr) {
+                    // It's possible that this function has been called while updating the pagetable
+                    // after unmapping a VMA. In that case the underlying VMA will no longer exist,
+                    // and we should just leave the pagetable entry blank.
+                    page_type = PageType::Unmapped;
+                } else {
+                    page_type = PageType::Memory;
+                    current_page_table->pointers[vaddr >> PAGE_BITS] = pointer;
+                }
+                break;
+            }
+            default:
+                UNREACHABLE();
+            }
+        }
+    }
+}
+
+void RasterizerFlushVirtualRegion(VAddr start, u64 size, FlushMode mode) {
+    // Since pages are unmapped on shutdown after video core is shutdown, the renderer may be
+    // null here
+    if (VideoCore::g_renderer == nullptr) {
+        return;
+    }
+
+    VAddr end = start + size;
+
+    auto CheckRegion = [&](VAddr region_start, VAddr region_end) {
+        if (start >= region_end || end <= region_start) {
+            // No overlap with region
+            return;
+        }
+
+        VAddr overlap_start = std::max(start, region_start);
+        VAddr overlap_end = std::min(end, region_end);
+        u64 overlap_size = overlap_end - overlap_start;
+
+        auto* rasterizer = VideoCore::g_renderer->Rasterizer();
+        switch (mode) {
+        case FlushMode::Flush:
+            rasterizer->FlushRegion(overlap_start, overlap_size);
+            break;
+        case FlushMode::Invalidate:
+            rasterizer->InvalidateRegion(overlap_start, overlap_size);
+            break;
+        case FlushMode::FlushAndInvalidate:
+            rasterizer->FlushAndInvalidateRegion(overlap_start, overlap_size);
+            break;
+        }
+    };
+
+    CheckRegion(PROCESS_IMAGE_VADDR, PROCESS_IMAGE_VADDR_END);
+    CheckRegion(HEAP_VADDR, HEAP_VADDR_END);
+}
+
 u8 Read8(const VAddr addr) {
     return Read<u8>(addr);
 }
@@ -309,17 +430,6 @@ u64 Read64(const VAddr addr) {
     return Read<u64_le>(addr);
 }
 
-static bool ReadSpecialBlock(const Kernel::Process& process, const VAddr src_addr,
-                             void* dest_buffer, const size_t size) {
-    auto& page_table = process.vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, src_addr, size)) {
-        if (handler->ReadBlock(src_addr, dest_buffer, size)) {
-            return true;
-        }
-    }
-    return false;
-}
-
 void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_buffer,
                const size_t size) {
     auto& page_table = process.vm_manager.page_table;
@@ -329,21 +439,16 @@ void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_
     size_t page_offset = src_addr & PAGE_MASK;
 
     while (remaining_size > 0) {
-        const size_t copy_amount = std::min<size_t>(PAGE_SIZE - page_offset, remaining_size);
+        const size_t copy_amount =
+            std::min(static_cast<size_t>(PAGE_SIZE) - page_offset, remaining_size);
         const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
 
         switch (page_table.attributes[page_index]) {
-        case PageType::Unmapped:
-            LOG_ERROR(HW_Memory,
-                      "unmapped ReadBlock @ 0x%016" PRIX64 " (start address = 0x%" PRIx64
-                      ", size = %zu)",
+        case PageType::Unmapped: {
+            LOG_ERROR(HW_Memory, "unmapped ReadBlock @ 0x%08X (start address = 0x%08X, size = %zu)",
                       current_vaddr, src_addr, size);
             std::memset(dest_buffer, 0, copy_amount);
             break;
-        case PageType::Special: {
-            if (ReadSpecialBlock(process, current_vaddr, dest_buffer, copy_amount))
-                break;
-            [[fallthrough]];
         }
         case PageType::Memory: {
             DEBUG_ASSERT(page_table.pointers[page_index]);
@@ -352,6 +457,12 @@ void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_
             std::memcpy(dest_buffer, src_ptr, copy_amount);
             break;
         }
+        case PageType::RasterizerCachedMemory: {
+            RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount),
+                                         FlushMode::Flush);
+            std::memcpy(dest_buffer, GetPointerFromVMA(process, current_vaddr), copy_amount);
+            break;
+        }
         default:
             UNREACHABLE();
         }
@@ -383,17 +494,6 @@ void Write64(const VAddr addr, const u64 data) {
     Write<u64_le>(addr, data);
 }
 
-static bool WriteSpecialBlock(const Kernel::Process& process, const VAddr dest_addr,
-                              const void* src_buffer, const size_t size) {
-    auto& page_table = process.vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, dest_addr, size)) {
-        if (handler->WriteBlock(dest_addr, src_buffer, size)) {
-            return true;
-        }
-    }
-    return false;
-}
-
 void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const void* src_buffer,
                 const size_t size) {
     auto& page_table = process.vm_manager.page_table;
@@ -402,20 +502,17 @@ void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const voi
     size_t page_offset = dest_addr & PAGE_MASK;
 
     while (remaining_size > 0) {
-        const size_t copy_amount = std::min<size_t>(PAGE_SIZE - page_offset, remaining_size);
+        const size_t copy_amount =
+            std::min(static_cast<size_t>(PAGE_SIZE) - page_offset, remaining_size);
         const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
 
         switch (page_table.attributes[page_index]) {
-        case PageType::Unmapped:
+        case PageType::Unmapped: {
             LOG_ERROR(HW_Memory,
-                      "unmapped WriteBlock @ 0x%016" PRIX64 " (start address = 0x%016" PRIX64
-                      ", size = %zu)",
+                      "unmapped WriteBlock @ 0x%08X (start address = 0x%08X, size = %zu)",
                       current_vaddr, dest_addr, size);
             break;
-        case PageType::Special:
-            if (WriteSpecialBlock(process, current_vaddr, src_buffer, copy_amount))
-                break;
-            [[fallthrough]];
+        }
         case PageType::Memory: {
             DEBUG_ASSERT(page_table.pointers[page_index]);
 
@@ -423,6 +520,12 @@ void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const voi
             std::memcpy(dest_ptr, src_buffer, copy_amount);
             break;
         }
+        case PageType::RasterizerCachedMemory: {
+            RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount),
+                                         FlushMode::Invalidate);
+            std::memcpy(GetPointerFromVMA(process, current_vaddr), src_buffer, copy_amount);
+            break;
+        }
         default:
             UNREACHABLE();
         }
@@ -438,9 +541,8 @@ void WriteBlock(const VAddr dest_addr, const void* src_buffer, const size_t size
     WriteBlock(*Core::CurrentProcess(), dest_addr, src_buffer, size);
 }
 
-void ZeroBlock(const VAddr dest_addr, const size_t size) {
-    const auto& process = *Core::CurrentProcess();
-
+void ZeroBlock(const Kernel::Process& process, const VAddr dest_addr, const size_t size) {
+    auto& page_table = process.vm_manager.page_table;
     size_t remaining_size = size;
     size_t page_index = dest_addr >> PAGE_BITS;
     size_t page_offset = dest_addr & PAGE_MASK;
@@ -448,27 +550,29 @@ void ZeroBlock(const VAddr dest_addr, const size_t size) {
     static const std::array<u8, PAGE_SIZE> zeros = {};
 
     while (remaining_size > 0) {
-        const size_t copy_amount = std::min<size_t>(PAGE_SIZE - page_offset, remaining_size);
+        const size_t copy_amount =
+            std::min(static_cast<size_t>(PAGE_SIZE) - page_offset, remaining_size);
         const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
 
-        switch (current_page_table->attributes[page_index]) {
-        case PageType::Unmapped:
-            LOG_ERROR(HW_Memory,
-                      "unmapped ZeroBlock @ 0x%016" PRIX64 " (start address = 0x%016" PRIX64
-                      ", size = %zu)",
+        switch (page_table.attributes[page_index]) {
+        case PageType::Unmapped: {
+            LOG_ERROR(HW_Memory, "unmapped ZeroBlock @ 0x%08X (start address = 0x%08X, size = %zu)",
                       current_vaddr, dest_addr, size);
             break;
-        case PageType::Special:
-            if (WriteSpecialBlock(process, current_vaddr, zeros.data(), copy_amount))
-                break;
-            [[fallthrough]];
+        }
         case PageType::Memory: {
-            DEBUG_ASSERT(current_page_table->pointers[page_index]);
+            DEBUG_ASSERT(page_table.pointers[page_index]);
 
-            u8* dest_ptr = current_page_table->pointers[page_index] + page_offset;
+            u8* dest_ptr = page_table.pointers[page_index] + page_offset;
             std::memset(dest_ptr, 0, copy_amount);
             break;
         }
+        case PageType::RasterizerCachedMemory: {
+            RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount),
+                                         FlushMode::Invalidate);
+            std::memset(GetPointerFromVMA(process, current_vaddr), 0, copy_amount);
+            break;
+        }
         default:
             UNREACHABLE();
         }
@@ -479,37 +583,34 @@ void ZeroBlock(const VAddr dest_addr, const size_t size) {
     }
 }
 
-void CopyBlock(VAddr dest_addr, VAddr src_addr, const size_t size) {
-    const auto& process = *Core::CurrentProcess();
-
+void CopyBlock(const Kernel::Process& process, VAddr dest_addr, VAddr src_addr, const size_t size) {
+    auto& page_table = process.vm_manager.page_table;
     size_t remaining_size = size;
     size_t page_index = src_addr >> PAGE_BITS;
     size_t page_offset = src_addr & PAGE_MASK;
 
     while (remaining_size > 0) {
-        const size_t copy_amount = std::min<size_t>(PAGE_SIZE - page_offset, remaining_size);
+        const size_t copy_amount =
+            std::min(static_cast<size_t>(PAGE_SIZE) - page_offset, remaining_size);
         const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
 
-        switch (current_page_table->attributes[page_index]) {
-        case PageType::Unmapped:
-            LOG_ERROR(HW_Memory,
-                      "unmapped CopyBlock @ 0x%016" PRIX64 " (start address = 0x%016" PRIX64
-                      ", size = %zu)",
+        switch (page_table.attributes[page_index]) {
+        case PageType::Unmapped: {
+            LOG_ERROR(HW_Memory, "unmapped CopyBlock @ 0x%08X (start address = 0x%08X, size = %zu)",
                       current_vaddr, src_addr, size);
-            ZeroBlock(dest_addr, copy_amount);
+            ZeroBlock(process, dest_addr, copy_amount);
             break;
-        case PageType::Special: {
-            std::vector<u8> buffer(copy_amount);
-            if (ReadSpecialBlock(process, current_vaddr, buffer.data(), buffer.size())) {
-                WriteBlock(dest_addr, buffer.data(), buffer.size());
-                break;
-            }
-            [[fallthrough]];
         }
         case PageType::Memory: {
-            DEBUG_ASSERT(current_page_table->pointers[page_index]);
-            const u8* src_ptr = current_page_table->pointers[page_index] + page_offset;
-            WriteBlock(dest_addr, src_ptr, copy_amount);
+            DEBUG_ASSERT(page_table.pointers[page_index]);
+            const u8* src_ptr = page_table.pointers[page_index] + page_offset;
+            WriteBlock(process, dest_addr, src_ptr, copy_amount);
+            break;
+        }
+        case PageType::RasterizerCachedMemory: {
+            RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount),
+                                         FlushMode::Flush);
+            WriteBlock(process, dest_addr, GetPointerFromVMA(process, current_vaddr), copy_amount);
             break;
         }
         default:
@@ -524,78 +625,6 @@ void CopyBlock(VAddr dest_addr, VAddr src_addr, const size_t size) {
     }
 }
 
-template <>
-boost::optional<u8> ReadSpecial<u8>(VAddr addr) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u8)))
-        if (auto result = handler->Read8(addr))
-            return *result;
-    return {};
-}
-
-template <>
-boost::optional<u16> ReadSpecial<u16>(VAddr addr) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u16)))
-        if (auto result = handler->Read16(addr))
-            return *result;
-    return {};
-}
-
-template <>
-boost::optional<u32> ReadSpecial<u32>(VAddr addr) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u32)))
-        if (auto result = handler->Read32(addr))
-            return *result;
-    return {};
-}
-
-template <>
-boost::optional<u64> ReadSpecial<u64>(VAddr addr) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u64)))
-        if (auto result = handler->Read64(addr))
-            return *result;
-    return {};
-}
-
-template <>
-bool WriteSpecial<u8>(VAddr addr, const u8 data) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u8)))
-        if (handler->Write8(addr, data))
-            return true;
-    return false;
-}
-
-template <>
-bool WriteSpecial<u16>(VAddr addr, const u16 data) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u16)))
-        if (handler->Write16(addr, data))
-            return true;
-    return false;
-}
-
-template <>
-bool WriteSpecial<u32>(VAddr addr, const u32 data) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u32)))
-        if (handler->Write32(addr, data))
-            return true;
-    return false;
-}
-
-template <>
-bool WriteSpecial<u64>(VAddr addr, const u64 data) {
-    const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table;
-    for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u64)))
-        if (handler->Write64(addr, data))
-            return true;
-    return false;
-}
-
 boost::optional<PAddr> TryVirtualToPhysicalAddress(const VAddr addr) {
     if (addr == 0) {
         return 0;
@@ -609,8 +638,6 @@ boost::optional<PAddr> TryVirtualToPhysicalAddress(const VAddr addr) {
         return addr - DSP_RAM_VADDR + DSP_RAM_PADDR;
     } else if (addr >= IO_AREA_VADDR && addr < IO_AREA_VADDR_END) {
         return addr - IO_AREA_VADDR + IO_AREA_PADDR;
-    } else if (addr >= N3DS_EXTRA_RAM_VADDR && addr < N3DS_EXTRA_RAM_VADDR_END) {
-        return addr - N3DS_EXTRA_RAM_VADDR + N3DS_EXTRA_RAM_PADDR;
     }
 
     return boost::none;
@@ -637,8 +664,6 @@ boost::optional<VAddr> PhysicalToVirtualAddress(const PAddr addr) {
         return addr - DSP_RAM_PADDR + DSP_RAM_VADDR;
     } else if (addr >= IO_AREA_PADDR && addr < IO_AREA_PADDR_END) {
         return addr - IO_AREA_PADDR + IO_AREA_VADDR;
-    } else if (addr >= N3DS_EXTRA_RAM_PADDR && addr < N3DS_EXTRA_RAM_PADDR_END) {
-        return addr - N3DS_EXTRA_RAM_PADDR + N3DS_EXTRA_RAM_VADDR;
     }
 
     return boost::none;

src/core/memory.h

@@ -36,7 +36,10 @@ enum class PageType : u8 {
     Unmapped,
     /// Page is mapped to regular memory. This is the only type you can get pointers to.
     Memory,
-    /// Page is mapped to a memory hook, which intercepts read and write requests.
+    /// Page is mapped to regular memory, but also needs to check for rasterizer cache flushing and
+    /// invalidation
+    RasterizerCachedMemory,
+    /// Page is mapped to a I/O region. Writing and reading to this page is handled by functions.
     Special,
 };
 
@@ -98,12 +101,6 @@ enum : PAddr {
     VRAM_SIZE = 0x00600000, ///< VRAM size (6MB)
     VRAM_PADDR_END = VRAM_PADDR + VRAM_SIZE,
 
-    /// New 3DS additional memory. Supposedly faster than regular FCRAM. Part of it can be used by
-    /// applications and system modules if mapped via the ExHeader.
-    N3DS_EXTRA_RAM_PADDR = 0x1F000000,
-    N3DS_EXTRA_RAM_SIZE = 0x00400000, ///< New 3DS additional memory size (4MB)
-    N3DS_EXTRA_RAM_PADDR_END = N3DS_EXTRA_RAM_PADDR + N3DS_EXTRA_RAM_SIZE,
-
     /// DSP memory
     DSP_RAM_PADDR = 0x1FF00000,
     DSP_RAM_SIZE = 0x00080000, ///< DSP memory size (512KB)
@@ -119,7 +116,6 @@ enum : PAddr {
     FCRAM_SIZE = 0x08000000,      ///< FCRAM size on the Old 3DS (128MB)
     FCRAM_N3DS_SIZE = 0x10000000, ///< FCRAM size on the New 3DS (256MB)
     FCRAM_PADDR_END = FCRAM_PADDR + FCRAM_SIZE,
-    FCRAM_N3DS_PADDR_END = FCRAM_PADDR + FCRAM_N3DS_SIZE,
 };
 
 /// Virtual user-space memory regions
@@ -135,10 +131,6 @@ enum : VAddr {
     LINEAR_HEAP_SIZE = 0x08000000,
     LINEAR_HEAP_VADDR_END = LINEAR_HEAP_VADDR + LINEAR_HEAP_SIZE,
 
-    /// Maps 1:1 to New 3DS additional memory
-    N3DS_EXTRA_RAM_VADDR = 0x1E800000,
-    N3DS_EXTRA_RAM_VADDR_END = N3DS_EXTRA_RAM_VADDR + N3DS_EXTRA_RAM_SIZE,
-
     /// Maps 1:1 to the IO register area.
     IO_AREA_VADDR = 0x1EC00000,
     IO_AREA_VADDR_END = IO_AREA_VADDR + IO_AREA_SIZE,
@@ -253,4 +245,24 @@ boost::optional<VAddr> PhysicalToVirtualAddress(PAddr addr);
  */
 u8* GetPhysicalPointer(PAddr address);
 
+enum class FlushMode {
+    /// Write back modified surfaces to RAM
+    Flush,
+    /// Remove region from the cache
+    Invalidate,
+    /// Write back modified surfaces to RAM, and also remove them from the cache
+    FlushAndInvalidate,
+};
+
+/**
+ * Mark each page touching the region as cached.
+ */
+void RasterizerMarkRegionCached(VAddr start, u64 size, bool cached);
+
+/**
+ * Flushes and invalidates any externally cached rasterizer resources touching the given virtual
+ * address region.
+ */
+void RasterizerFlushVirtualRegion(VAddr start, u64 size, FlushMode mode);
+
 } // namespace Memory

src/core/settings.h

@@ -105,12 +105,10 @@ static const std::array<const char*, NumAnalogs> mapping = {{
 }};
 } // namespace NativeAnalog
 
-enum class CpuCore {
-    Unicorn,
-    Dynarmic,
-};
-
 struct Values {
+    // System
+    bool use_docked_mode;
+
     // Controls
     std::array<std::string, NativeButton::NumButtons> buttons;
     std::array<std::string, NativeAnalog::NumAnalogs> analogs;
@@ -118,7 +116,7 @@ struct Values {
     std::string touch_device;
 
     // Core
-    CpuCore cpu_core;
+    bool use_cpu_jit;
 
     // Data Storage
     bool use_virtual_sd;

src/core/telemetry_session.cpp

@@ -154,12 +154,13 @@ TelemetrySession::TelemetrySession() {
 #endif
 
     // Log user configuration information
-    AddField(Telemetry::FieldType::UserConfig, "Core_CpuCore",
-             static_cast<int>(Settings::values.cpu_core));
+    AddField(Telemetry::FieldType::UserConfig, "Core_UseCpuJit", Settings::values.use_cpu_jit);
     AddField(Telemetry::FieldType::UserConfig, "Renderer_ResolutionFactor",
              Settings::values.resolution_factor);
     AddField(Telemetry::FieldType::UserConfig, "Renderer_ToggleFramelimit",
              Settings::values.toggle_framelimit);
+    AddField(Telemetry::FieldType::UserConfig, "System_UseDockedMode",
+             Settings::values.use_docked_mode);
 }
 
 TelemetrySession::~TelemetrySession() {

src/video_core/CMakeLists.txt

@@ -1,6 +1,8 @@
 add_library(video_core STATIC
     command_processor.cpp
     command_processor.h
+    debug_utils/debug_utils.cpp
+    debug_utils/debug_utils.h
     engines/fermi_2d.cpp
     engines/fermi_2d.h
     engines/maxwell_3d.cpp
@@ -11,15 +13,30 @@ add_library(video_core STATIC
     gpu.h
     memory_manager.cpp
     memory_manager.h
+    rasterizer_interface.h
     renderer_base.cpp
     renderer_base.h
+    renderer_opengl/gl_rasterizer.cpp
+    renderer_opengl/gl_rasterizer.h
+    renderer_opengl/gl_rasterizer_cache.cpp
+    renderer_opengl/gl_rasterizer_cache.h
     renderer_opengl/gl_resource_manager.h
+    renderer_opengl/gl_shader_decompiler.cpp
+    renderer_opengl/gl_shader_decompiler.h
+    renderer_opengl/gl_shader_gen.cpp
+    renderer_opengl/gl_shader_gen.h
     renderer_opengl/gl_shader_util.cpp
     renderer_opengl/gl_shader_util.h
     renderer_opengl/gl_state.cpp
     renderer_opengl/gl_state.h
+    renderer_opengl/gl_stream_buffer.cpp
+    renderer_opengl/gl_stream_buffer.h
+    renderer_opengl/maxwell_to_gl.h
     renderer_opengl/renderer_opengl.cpp
     renderer_opengl/renderer_opengl.h
+    textures/decoders.cpp
+    textures/decoders.h
+    textures/texture.h
     utils.h
     video_core.cpp
     video_core.h

src/video_core/debug_utils/debug_utils.cpp

@@ -0,0 +1,64 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <condition_variable>
+#include <cstdint>
+#include <cstring>
+#include <fstream>
+#include <map>
+#include <mutex>
+#include <string>
+
+#include "common/assert.h"
+#include "common/bit_field.h"
+#include "common/color.h"
+#include "common/common_types.h"
+#include "common/file_util.h"
+#include "common/logging/log.h"
+#include "common/math_util.h"
+#include "common/vector_math.h"
+#include "video_core/debug_utils/debug_utils.h"
+
+namespace Tegra {
+
+void DebugContext::DoOnEvent(Event event, void* data) {
+    {
+        std::unique_lock<std::mutex> lock(breakpoint_mutex);
+
+        // TODO(Subv): Commit the rasterizer's caches so framebuffers, render targets, etc. will
+        // show on debug widgets
+
+        // TODO: Should stop the CPU thread here once we multithread emulation.
+
+        active_breakpoint = event;
+        at_breakpoint = true;
+
+        // Tell all observers that we hit a breakpoint
+        for (auto& breakpoint_observer : breakpoint_observers) {
+            breakpoint_observer->OnMaxwellBreakPointHit(event, data);
+        }
+
+        // Wait until another thread tells us to Resume()
+        resume_from_breakpoint.wait(lock, [&] { return !at_breakpoint; });
+    }
+}
+
+void DebugContext::Resume() {
+    {
+        std::lock_guard<std::mutex> lock(breakpoint_mutex);
+
+        // Tell all observers that we are about to resume
+        for (auto& breakpoint_observer : breakpoint_observers) {
+            breakpoint_observer->OnMaxwellResume();
+        }
+
+        // Resume the waiting thread (i.e. OnEvent())
+        at_breakpoint = false;
+    }
+
+    resume_from_breakpoint.notify_one();
+}
+
+} // namespace Tegra

src/video_core/debug_utils/debug_utils.h

@@ -0,0 +1,163 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <algorithm>
+#include <array>
+#include <condition_variable>
+#include <iterator>
+#include <list>
+#include <map>
+#include <memory>
+#include <mutex>
+#include <string>
+#include <utility>
+#include <vector>
+#include "common/common_types.h"
+#include "common/vector_math.h"
+
+namespace Tegra {
+
+class DebugContext {
+public:
+    enum class Event {
+        FirstEvent = 0,
+
+        MaxwellCommandLoaded = FirstEvent,
+        MaxwellCommandProcessed,
+        IncomingPrimitiveBatch,
+        FinishedPrimitiveBatch,
+
+        NumEvents
+    };
+
+    /**
+     * Inherit from this class to be notified of events registered to some debug context.
+     * Most importantly this is used for our debugger GUI.
+     *
+     * To implement event handling, override the OnMaxwellBreakPointHit and OnMaxwellResume methods.
+     * @warning All BreakPointObservers need to be on the same thread to guarantee thread-safe state
+     * access
+     * @todo Evaluate an alternative interface, in which there is only one managing observer and
+     * multiple child observers running (by design) on the same thread.
+     */
+    class BreakPointObserver {
+    public:
+        /// Constructs the object such that it observes events of the given DebugContext.
+        BreakPointObserver(std::shared_ptr<DebugContext> debug_context)
+            : context_weak(debug_context) {
+            std::unique_lock<std::mutex> lock(debug_context->breakpoint_mutex);
+            debug_context->breakpoint_observers.push_back(this);
+        }
+
+        virtual ~BreakPointObserver() {
+            auto context = context_weak.lock();
+            if (context) {
+                std::unique_lock<std::mutex> lock(context->breakpoint_mutex);
+                context->breakpoint_observers.remove(this);
+
+                // If we are the last observer to be destroyed, tell the debugger context that
+                // it is free to continue. In particular, this is required for a proper yuzu
+                // shutdown, when the emulation thread is waiting at a breakpoint.
+                if (context->breakpoint_observers.empty())
+                    context->Resume();
+            }
+        }
+
+        /**
+         * Action to perform when a breakpoint was reached.
+         * @param event Type of event which triggered the breakpoint
+         * @param data Optional data pointer (if unused, this is a nullptr)
+         * @note This function will perform nothing unless it is overridden in the child class.
+         */
+        virtual void OnMaxwellBreakPointHit(Event event, void* data) {}
+
+        /**
+         * Action to perform when emulation is resumed from a breakpoint.
+         * @note This function will perform nothing unless it is overridden in the child class.
+         */
+        virtual void OnMaxwellResume() {}
+
+    protected:
+        /**
+         * Weak context pointer. This need not be valid, so when requesting a shared_ptr via
+         * context_weak.lock(), always compare the result against nullptr.
+         */
+        std::weak_ptr<DebugContext> context_weak;
+    };
+
+    /**
+     * Simple structure defining a breakpoint state
+     */
+    struct BreakPoint {
+        bool enabled = false;
+    };
+
+    /**
+     * Static constructor used to create a shared_ptr of a DebugContext.
+     */
+    static std::shared_ptr<DebugContext> Construct() {
+        return std::shared_ptr<DebugContext>(new DebugContext);
+    }
+
+    /**
+     * Used by the emulation core when a given event has happened. If a breakpoint has been set
+     * for this event, OnEvent calls the event handlers of the registered breakpoint observers.
+     * The current thread then is halted until Resume() is called from another thread (or until
+     * emulation is stopped).
+     * @param event Event which has happened
+     * @param data Optional data pointer (pass nullptr if unused). Needs to remain valid until
+     * Resume() is called.
+     */
+    void OnEvent(Event event, void* data) {
+        // This check is left in the header to allow the compiler to inline it.
+        if (!breakpoints[(int)event].enabled)
+            return;
+        // For the rest of event handling, call a separate function.
+        DoOnEvent(event, data);
+    }
+
+    void DoOnEvent(Event event, void* data);
+
+    /**
+     * Resume from the current breakpoint.
+     * @warning Calling this from the same thread that OnEvent was called in will cause a deadlock.
+     * Calling from any other thread is safe.
+     */
+    void Resume();
+
+    /**
+     * Delete all set breakpoints and resume emulation.
+     */
+    void ClearBreakpoints() {
+        for (auto& bp : breakpoints) {
+            bp.enabled = false;
+        }
+        Resume();
+    }
+
+    // TODO: Evaluate if access to these members should be hidden behind a public interface.
+    std::array<BreakPoint, (int)Event::NumEvents> breakpoints;
+    Event active_breakpoint;
+    bool at_breakpoint = false;
+
+private:
+    /**
+     * Private default constructor to make sure people always construct this through Construct()
+     * instead.
+     */
+    DebugContext() = default;
+
+    /// Mutex protecting current breakpoint state and the observer list.
+    std::mutex breakpoint_mutex;
+
+    /// Used by OnEvent to wait for resumption.
+    std::condition_variable resume_from_breakpoint;
+
+    /// List of registered observers
+    std::list<BreakPointObserver*> breakpoint_observers;
+};
+
+} // namespace Tegra

src/video_core/engines/maxwell_3d.cpp

@@ -2,8 +2,16 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <cinttypes>
 #include "common/assert.h"
+#include "core/core.h"
+#include "video_core/debug_utils/debug_utils.h"
 #include "video_core/engines/maxwell_3d.h"
+#include "video_core/rasterizer_interface.h"
+#include "video_core/renderer_base.h"
+#include "video_core/textures/decoders.h"
+#include "video_core/textures/texture.h"
+#include "video_core/video_core.h"
 
 namespace Tegra {
 namespace Engines {
@@ -46,6 +54,8 @@ void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
     ASSERT_MSG(method < Regs::NUM_REGS,
                "Invalid Maxwell3D register, increase the size of the Regs structure");
 
+    auto debug_context = Core::System::GetInstance().GetGPUDebugContext();
+
     // It is an error to write to a register other than the current macro's ARG register before it
     // has finished execution.
     if (executing_macro != 0) {
@@ -72,6 +82,10 @@ void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
         return;
     }
 
+    if (debug_context) {
+        debug_context->OnEvent(Tegra::DebugContext::Event::MaxwellCommandLoaded, nullptr);
+    }
+
     regs.reg_array[method] = value;
 
 #define MAXWELL3D_REG_INDEX(field_name) (offsetof(Regs, field_name) / sizeof(u32))
@@ -137,6 +151,10 @@ void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
     }
 
 #undef MAXWELL3D_REG_INDEX
+
+    if (debug_context) {
+        debug_context->OnEvent(Tegra::DebugContext::Event::MaxwellCommandProcessed, nullptr);
+    }
 }
 
 void Maxwell3D::ProcessQueryGet() {
@@ -159,7 +177,20 @@ void Maxwell3D::ProcessQueryGet() {
 }
 
 void Maxwell3D::DrawArrays() {
-    LOG_WARNING(HW_GPU, "Game requested a DrawArrays, ignoring");
+    LOG_DEBUG(HW_GPU, "called, topology=%d, count=%d", regs.draw.topology.Value(),
+              regs.vertex_buffer.count);
+
+    auto debug_context = Core::System::GetInstance().GetGPUDebugContext();
+
+    if (debug_context) {
+        debug_context->OnEvent(Tegra::DebugContext::Event::IncomingPrimitiveBatch, nullptr);
+    }
+
+    if (debug_context) {
+        debug_context->OnEvent(Tegra::DebugContext::Event::FinishedPrimitiveBatch, nullptr);
+    }
+
+    VideoCore::g_renderer->Rasterizer()->AccelerateDrawBatch(false /*is_indexed*/);
 }
 
 void Maxwell3D::BindTextureInfoBuffer(const std::vector<u32>& parameters) {
@@ -270,5 +301,90 @@ void Maxwell3D::ProcessCBData(u32 value) {
     regs.const_buffer.cb_pos = regs.const_buffer.cb_pos + 4;
 }
 
+Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const {
+    GPUVAddr tic_base_address = regs.tic.TICAddress();
+
+    GPUVAddr tic_address_gpu = tic_base_address + tic_index * sizeof(Texture::TICEntry);
+    VAddr tic_address_cpu = memory_manager.PhysicalToVirtualAddress(tic_address_gpu);
+
+    Texture::TICEntry tic_entry;
+    Memory::ReadBlock(tic_address_cpu, &tic_entry, sizeof(Texture::TICEntry));
+
+    ASSERT_MSG(tic_entry.header_version == Texture::TICHeaderVersion::BlockLinear,
+               "TIC versions other than BlockLinear are unimplemented");
+
+    ASSERT_MSG(tic_entry.texture_type == Texture::TextureType::Texture2D,
+               "Texture types other than Texture2D are unimplemented");
+
+    auto r_type = tic_entry.r_type.Value();
+    auto g_type = tic_entry.g_type.Value();
+    auto b_type = tic_entry.b_type.Value();
+    auto a_type = tic_entry.a_type.Value();
+
+    // TODO(Subv): Different data types for separate components are not supported
+    ASSERT(r_type == g_type && r_type == b_type && r_type == a_type);
+
+    return tic_entry;
+}
+
+Texture::TSCEntry Maxwell3D::GetTSCEntry(u32 tsc_index) const {
+    GPUVAddr tsc_base_address = regs.tsc.TSCAddress();
+
+    GPUVAddr tsc_address_gpu = tsc_base_address + tsc_index * sizeof(Texture::TSCEntry);
+    VAddr tsc_address_cpu = memory_manager.PhysicalToVirtualAddress(tsc_address_gpu);
+
+    Texture::TSCEntry tsc_entry;
+    Memory::ReadBlock(tsc_address_cpu, &tsc_entry, sizeof(Texture::TSCEntry));
+    return tsc_entry;
+}
+
+std::vector<Texture::FullTextureInfo> Maxwell3D::GetStageTextures(Regs::ShaderStage stage) const {
+    std::vector<Texture::FullTextureInfo> textures;
+
+    auto& fragment_shader = state.shader_stages[static_cast<size_t>(stage)];
+    auto& tex_info_buffer = fragment_shader.const_buffers[regs.tex_cb_index];
+    ASSERT(tex_info_buffer.enabled && tex_info_buffer.address != 0);
+
+    GPUVAddr tic_base_address = regs.tic.TICAddress();
+
+    GPUVAddr tex_info_buffer_end = tex_info_buffer.address + tex_info_buffer.size;
+
+    // Offset into the texture constbuffer where the texture info begins.
+    static constexpr size_t TextureInfoOffset = 0x20;
+
+    for (GPUVAddr current_texture = tex_info_buffer.address + TextureInfoOffset;
+         current_texture < tex_info_buffer_end; current_texture += sizeof(Texture::TextureHandle)) {
+
+        Texture::TextureHandle tex_handle{
+            Memory::Read32(memory_manager.PhysicalToVirtualAddress(current_texture))};
+
+        Texture::FullTextureInfo tex_info{};
+        // TODO(Subv): Use the shader to determine which textures are actually accessed.
+        tex_info.index = (current_texture - tex_info_buffer.address - TextureInfoOffset) /
+                         sizeof(Texture::TextureHandle);
+
+        // Load the TIC data.
+        if (tex_handle.tic_id != 0) {
+            tex_info.enabled = true;
+
+            auto tic_entry = GetTICEntry(tex_handle.tic_id);
+            // TODO(Subv): Workaround for BitField's move constructor being deleted.
+            std::memcpy(&tex_info.tic, &tic_entry, sizeof(tic_entry));
+        }
+
+        // Load the TSC data
+        if (tex_handle.tsc_id != 0) {
+            auto tsc_entry = GetTSCEntry(tex_handle.tsc_id);
+            // TODO(Subv): Workaround for BitField's move constructor being deleted.
+            std::memcpy(&tex_info.tsc, &tsc_entry, sizeof(tsc_entry));
+        }
+
+        if (tex_info.enabled)
+            textures.push_back(tex_info);
+    }
+
+    return textures;
+}
+
 } // namespace Engines
 } // namespace Tegra

src/video_core/engines/maxwell_3d.h

@@ -7,10 +7,14 @@
 #include <array>
 #include <unordered_map>
 #include <vector>
+#include "common/assert.h"
 #include "common/bit_field.h"
 #include "common/common_funcs.h"
 #include "common/common_types.h"
+#include "common/math_util.h"
+#include "video_core/gpu.h"
 #include "video_core/memory_manager.h"
+#include "video_core/textures/texture.h"
 
 namespace Tegra {
 namespace Engines {
@@ -20,19 +24,16 @@ public:
     explicit Maxwell3D(MemoryManager& memory_manager);
     ~Maxwell3D() = default;
 
-    /// Write the value to the register identified by method.
-    void WriteReg(u32 method, u32 value, u32 remaining_params);
-
-    /// Uploads the code for a GPU macro program associated with the specified entry.
-    void SubmitMacroCode(u32 entry, std::vector<u32> code);
-
     /// Register structure of the Maxwell3D engine.
     /// TODO(Subv): This structure will need to be made bigger as more registers are discovered.
     struct Regs {
         static constexpr size_t NUM_REGS = 0xE36;
 
+        static constexpr size_t NumRenderTargets = 8;
+        static constexpr size_t NumViewports = 16;
         static constexpr size_t NumCBData = 16;
         static constexpr size_t NumVertexArrays = 32;
+        static constexpr size_t NumVertexAttributes = 32;
         static constexpr size_t MaxShaderProgram = 6;
         static constexpr size_t MaxShaderStage = 5;
         // Maximum number of const buffers per shader stage.
@@ -60,9 +61,273 @@ public:
             Fragment = 4,
         };
 
+        struct VertexAttribute {
+            enum class Size : u32 {
+                Size_32_32_32_32 = 0x01,
+                Size_32_32_32 = 0x02,
+                Size_16_16_16_16 = 0x03,
+                Size_32_32 = 0x04,
+                Size_16_16_16 = 0x05,
+                Size_8_8_8_8 = 0x0a,
+                Size_16_16 = 0x0f,
+                Size_32 = 0x12,
+                Size_8_8_8 = 0x13,
+                Size_8_8 = 0x18,
+                Size_16 = 0x1b,
+                Size_8 = 0x1d,
+                Size_10_10_10_2 = 0x30,
+                Size_11_11_10 = 0x31,
+            };
+
+            enum class Type : u32 {
+                SignedNorm = 1,
+                UnsignedNorm = 2,
+                SignedInt = 3,
+                UnsignedInt = 4,
+                UnsignedScaled = 5,
+                SignedScaled = 6,
+                Float = 7,
+            };
+
+            union {
+                BitField<0, 5, u32> buffer;
+                BitField<6, 1, u32> constant;
+                BitField<7, 14, u32> offset;
+                BitField<21, 6, Size> size;
+                BitField<27, 3, Type> type;
+                BitField<31, 1, u32> bgra;
+            };
+
+            u32 ComponentCount() const {
+                switch (size) {
+                case Size::Size_32_32_32_32:
+                    return 4;
+                case Size::Size_32_32_32:
+                    return 3;
+                case Size::Size_16_16_16_16:
+                    return 4;
+                case Size::Size_32_32:
+                    return 2;
+                case Size::Size_16_16_16:
+                    return 3;
+                case Size::Size_8_8_8_8:
+                    return 4;
+                case Size::Size_16_16:
+                    return 2;
+                case Size::Size_32:
+                    return 1;
+                case Size::Size_8_8_8:
+                    return 3;
+                case Size::Size_8_8:
+                    return 2;
+                case Size::Size_16:
+                    return 1;
+                case Size::Size_8:
+                    return 1;
+                case Size::Size_10_10_10_2:
+                    return 4;
+                case Size::Size_11_11_10:
+                    return 3;
+                default:
+                    UNREACHABLE();
+                }
+            }
+
+            u32 SizeInBytes() const {
+                switch (size) {
+                case Size::Size_32_32_32_32:
+                    return 16;
+                case Size::Size_32_32_32:
+                    return 12;
+                case Size::Size_16_16_16_16:
+                    return 8;
+                case Size::Size_32_32:
+                    return 8;
+                case Size::Size_16_16_16:
+                    return 6;
+                case Size::Size_8_8_8_8:
+                    return 4;
+                case Size::Size_16_16:
+                    return 4;
+                case Size::Size_32:
+                    return 4;
+                case Size::Size_8_8_8:
+                    return 3;
+                case Size::Size_8_8:
+                    return 2;
+                case Size::Size_16:
+                    return 2;
+                case Size::Size_8:
+                    return 1;
+                case Size::Size_10_10_10_2:
+                    return 4;
+                case Size::Size_11_11_10:
+                    return 4;
+                default:
+                    UNREACHABLE();
+                }
+            }
+
+            std::string SizeString() const {
+                switch (size) {
+                case Size::Size_32_32_32_32:
+                    return "32_32_32_32";
+                case Size::Size_32_32_32:
+                    return "32_32_32";
+                case Size::Size_16_16_16_16:
+                    return "16_16_16_16";
+                case Size::Size_32_32:
+                    return "32_32";
+                case Size::Size_16_16_16:
+                    return "16_16_16";
+                case Size::Size_8_8_8_8:
+                    return "8_8_8_8";
+                case Size::Size_16_16:
+                    return "16_16";
+                case Size::Size_32:
+                    return "32";
+                case Size::Size_8_8_8:
+                    return "8_8_8";
+                case Size::Size_8_8:
+                    return "8_8";
+                case Size::Size_16:
+                    return "16";
+                case Size::Size_8:
+                    return "8";
+                case Size::Size_10_10_10_2:
+                    return "10_10_10_2";
+                case Size::Size_11_11_10:
+                    return "11_11_10";
+                }
+                UNREACHABLE();
+                return {};
+            }
+
+            std::string TypeString() const {
+                switch (type) {
+                case Type::SignedNorm:
+                    return "SNORM";
+                case Type::UnsignedNorm:
+                    return "UNORM";
+                case Type::SignedInt:
+                    return "SINT";
+                case Type::UnsignedInt:
+                    return "UINT";
+                case Type::UnsignedScaled:
+                    return "USCALED";
+                case Type::SignedScaled:
+                    return "SSCALED";
+                case Type::Float:
+                    return "FLOAT";
+                }
+                UNREACHABLE();
+                return {};
+            }
+
+            bool IsNormalized() const {
+                return (type == Type::SignedNorm) || (type == Type::UnsignedNorm);
+            }
+        };
+
+        enum class PrimitiveTopology : u32 {
+            Points = 0x0,
+            Lines = 0x1,
+            LineLoop = 0x2,
+            LineStrip = 0x3,
+            Triangles = 0x4,
+            TriangleStrip = 0x5,
+            TriangleFan = 0x6,
+            Quads = 0x7,
+            QuadStrip = 0x8,
+            Polygon = 0x9,
+            LinesAdjacency = 0xa,
+            LineStripAdjacency = 0xb,
+            TrianglesAdjacency = 0xc,
+            TriangleStripAdjacency = 0xd,
+            Patches = 0xe,
+        };
+
         union {
             struct {
-                INSERT_PADDING_WORDS(0x557);
+                INSERT_PADDING_WORDS(0x200);
+
+                struct {
+                    u32 address_high;
+                    u32 address_low;
+                    u32 width;
+                    u32 height;
+                    Tegra::RenderTargetFormat format;
+                    u32 block_dimensions;
+                    u32 array_mode;
+                    u32 layer_stride;
+                    u32 base_layer;
+                    INSERT_PADDING_WORDS(7);
+
+                    GPUVAddr Address() const {
+                        return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) |
+                                                     address_low);
+                    }
+                } rt[NumRenderTargets];
+
+                INSERT_PADDING_WORDS(0x80);
+
+                struct {
+                    union {
+                        BitField<0, 16, u32> x;
+                        BitField<16, 16, u32> width;
+                    };
+                    union {
+                        BitField<0, 16, u32> y;
+                        BitField<16, 16, u32> height;
+                    };
+                    float depth_range_near;
+                    float depth_range_far;
+
+                    MathUtil::Rectangle<s32> GetRect() const {
+                        return {
+                            static_cast<s32>(x),          // left
+                            static_cast<s32>(y + height), // top
+                            static_cast<s32>(x + width),  // right
+                            static_cast<s32>(y)           // bottom
+                        };
+                    };
+                } viewport[NumViewports];
+
+                INSERT_PADDING_WORDS(0x1D);
+
+                struct {
+                    u32 first;
+                    u32 count;
+                } vertex_buffer;
+
+                INSERT_PADDING_WORDS(0x99);
+
+                struct {
+                    u32 address_high;
+                    u32 address_low;
+                    u32 format;
+                    u32 block_dimensions;
+                    u32 layer_stride;
+
+                    GPUVAddr Address() const {
+                        return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) |
+                                                     address_low);
+                    }
+                } zeta;
+
+                INSERT_PADDING_WORDS(0x5B);
+
+                VertexAttribute vertex_attrib_format[NumVertexAttributes];
+
+                INSERT_PADDING_WORDS(0xF);
+
+                struct {
+                    union {
+                        BitField<0, 4, u32> count;
+                    };
+                } rt_control;
+
+                INSERT_PADDING_WORDS(0xCF);
 
                 struct {
                     u32 tsc_address_high;
@@ -100,10 +365,15 @@ public:
                     }
                 } code_address;
                 INSERT_PADDING_WORDS(1);
+
                 struct {
                     u32 vertex_end_gl;
-                    u32 vertex_begin_gl;
+                    union {
+                        u32 vertex_begin_gl;
+                        BitField<0, 16, PrimitiveTopology> topology;
+                    };
                 } draw;
+
                 INSERT_PADDING_WORDS(0x139);
                 struct {
                     u32 query_address_high;
@@ -244,6 +514,15 @@ public:
 
     State state{};
 
+    /// Write the value to the register identified by method.
+    void WriteReg(u32 method, u32 value, u32 remaining_params);
+
+    /// Uploads the code for a GPU macro program associated with the specified entry.
+    void SubmitMacroCode(u32 entry, std::vector<u32> code);
+
+    /// Returns a list of enabled textures for the specified shader stage.
+    std::vector<Texture::FullTextureInfo> GetStageTextures(Regs::ShaderStage stage) const;
+
 private:
     MemoryManager& memory_manager;
 
@@ -254,6 +533,12 @@ private:
     /// Parameters that have been submitted to the macro call so far.
     std::vector<u32> macro_params;
 
+    /// Retrieves information about a specific TIC entry from the TIC buffer.
+    Texture::TICEntry GetTICEntry(u32 tic_index) const;
+
+    /// Retrieves information about a specific TSC entry from the TSC buffer.
+    Texture::TSCEntry GetTSCEntry(u32 tsc_index) const;
+
     /**
      * Call a macro on this engine.
      * @param method Method to call
@@ -291,6 +576,12 @@ private:
     static_assert(offsetof(Maxwell3D::Regs, field_name) == position * 4,                           \
                   "Field " #field_name " has invalid position")
 
+ASSERT_REG_POSITION(rt, 0x200);
+ASSERT_REG_POSITION(viewport, 0x300);
+ASSERT_REG_POSITION(vertex_buffer, 0x35D);
+ASSERT_REG_POSITION(zeta, 0x3F8);
+ASSERT_REG_POSITION(vertex_attrib_format[0], 0x458);
+ASSERT_REG_POSITION(rt_control, 0x487);
 ASSERT_REG_POSITION(tsc, 0x557);
 ASSERT_REG_POSITION(tic, 0x55D);
 ASSERT_REG_POSITION(code_address, 0x582);

src/video_core/gpu.cpp

@@ -18,4 +18,8 @@ GPU::GPU() {
 
 GPU::~GPU() = default;
 
+const Tegra::Engines::Maxwell3D& GPU::Get3DEngine() const {
+    return *maxwell_3d;
+}
+
 } // namespace Tegra

src/video_core/gpu.h

@@ -8,10 +8,49 @@
 #include <unordered_map>
 #include <vector>
 #include "common/common_types.h"
+#include "core/hle/service/nvflinger/buffer_queue.h"
 #include "video_core/memory_manager.h"
 
 namespace Tegra {
 
+enum class RenderTargetFormat : u32 {
+    NONE = 0x0,
+    RGBA8_UNORM = 0xD5,
+};
+
+class DebugContext;
+
+/**
+ * Struct describing framebuffer configuration
+ */
+struct FramebufferConfig {
+    enum class PixelFormat : u32 {
+        ABGR8 = 1,
+    };
+
+    /**
+     * Returns the number of bytes per pixel.
+     */
+    static u32 BytesPerPixel(PixelFormat format) {
+        switch (format) {
+        case PixelFormat::ABGR8:
+            return 4;
+        }
+
+        UNREACHABLE();
+    }
+
+    VAddr address;
+    u32 offset;
+    u32 width;
+    u32 height;
+    u32 stride;
+    PixelFormat pixel_format;
+
+    using TransformFlags = Service::NVFlinger::BufferQueue::BufferTransformFlags;
+    TransformFlags transform_flags;
+};
+
 namespace Engines {
 class Fermi2D;
 class Maxwell3D;
@@ -34,8 +73,15 @@ public:
     /// Processes a command list stored at the specified address in GPU memory.
     void ProcessCommandList(GPUVAddr address, u32 size);
 
+    /// Returns a reference to the Maxwell3D GPU engine.
+    const Engines::Maxwell3D& Get3DEngine() const;
+
     std::unique_ptr<MemoryManager> memory_manager;
 
+    Engines::Maxwell3D& Maxwell3D() {
+        return *maxwell_3d;
+    }
+
 private:
     static constexpr u32 InvalidGraphMacroEntry = 0xFFFFFFFF;
 

src/video_core/rasterizer_interface.h

@@ -0,0 +1,63 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_types.h"
+#include "video_core/gpu.h"
+
+struct ScreenInfo;
+
+namespace VideoCore {
+
+class RasterizerInterface {
+public:
+    virtual ~RasterizerInterface() {}
+
+    /// Draw the current batch of vertex arrays
+    virtual void DrawArrays() = 0;
+
+    /// Notify rasterizer that the specified Maxwell register has been changed
+    virtual void NotifyMaxwellRegisterChanged(u32 id) = 0;
+
+    /// Notify rasterizer that all caches should be flushed to 3DS memory
+    virtual void FlushAll() = 0;
+
+    /// Notify rasterizer that any caches of the specified region should be flushed to 3DS memory
+    virtual void FlushRegion(VAddr addr, u64 size) = 0;
+
+    /// Notify rasterizer that any caches of the specified region should be invalidated
+    virtual void InvalidateRegion(VAddr addr, u64 size) = 0;
+
+    /// Notify rasterizer that any caches of the specified region should be flushed to 3DS memory
+    /// and invalidated
+    virtual void FlushAndInvalidateRegion(VAddr addr, u64 size) = 0;
+
+    /// Attempt to use a faster method to perform a display transfer with is_texture_copy = 0
+    virtual bool AccelerateDisplayTransfer(const void* config) {
+        return false;
+    }
+
+    /// Attempt to use a faster method to perform a display transfer with is_texture_copy = 1
+    virtual bool AccelerateTextureCopy(const void* config) {
+        return false;
+    }
+
+    /// Attempt to use a faster method to fill a region
+    virtual bool AccelerateFill(const void* config) {
+        return false;
+    }
+
+    /// Attempt to use a faster method to display the framebuffer to screen
+    virtual bool AccelerateDisplay(const Tegra::FramebufferConfig& framebuffer,
+                                   VAddr framebuffer_addr, u32 pixel_stride,
+                                   ScreenInfo& screen_info) {
+        return false;
+    }
+
+    virtual bool AccelerateDrawBatch(bool is_indexed) {
+        return false;
+    }
+};
+} // namespace VideoCore

src/video_core/renderer_base.cpp

@@ -5,6 +5,11 @@
 #include <atomic>
 #include <memory>
 #include "video_core/renderer_base.h"
+#include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/video_core.h"
 
-void RendererBase::RefreshRasterizerSetting() {}
+void RendererBase::RefreshRasterizerSetting() {
+    if (rasterizer == nullptr) {
+        rasterizer = std::make_unique<RasterizerOpenGL>();
+    }
+}

src/video_core/renderer_base.h

@@ -8,6 +8,8 @@
 #include <boost/optional.hpp>
 #include "common/assert.h"
 #include "common/common_types.h"
+#include "video_core/gpu.h"
+#include "video_core/rasterizer_interface.h"
 
 class EmuWindow;
 
@@ -16,40 +18,10 @@ public:
     /// Used to reference a framebuffer
     enum kFramebuffer { kFramebuffer_VirtualXFB = 0, kFramebuffer_EFB, kFramebuffer_Texture };
 
-    /**
-     * Struct describing framebuffer metadata
-     * TODO(bunnei): This struct belongs in the GPU code, but we don't have a good place for it yet.
-     */
-    struct FramebufferInfo {
-        enum class PixelFormat : u32 {
-            ABGR8 = 1,
-        };
-
-        /**
-         * Returns the number of bytes per pixel.
-         */
-        static u32 BytesPerPixel(PixelFormat format) {
-            switch (format) {
-            case PixelFormat::ABGR8:
-                return 4;
-            }
-
-            UNREACHABLE();
-        }
-
-        VAddr address;
-        u32 offset;
-        u32 width;
-        u32 height;
-        u32 stride;
-        PixelFormat pixel_format;
-        bool flip_vertical;
-    };
-
     virtual ~RendererBase() {}
 
     /// Swap buffers (render frame)
-    virtual void SwapBuffers(boost::optional<const FramebufferInfo&> framebuffer_info) = 0;
+    virtual void SwapBuffers(boost::optional<const Tegra::FramebufferConfig&> framebuffer) = 0;
 
     /**
      * Set the emulator window to use for renderer
@@ -74,12 +46,16 @@ public:
         return m_current_frame;
     }
 
+    VideoCore::RasterizerInterface* Rasterizer() const {
+        return rasterizer.get();
+    }
+
     void RefreshRasterizerSetting();
 
 protected:
+    std::unique_ptr<VideoCore::RasterizerInterface> rasterizer;
     f32 m_current_fps = 0.0f; ///< Current framerate, should be set by the renderer
     int m_current_frame = 0;  ///< Current frame, should be set by the renderer
 
 private:
-    bool opengl_rasterizer_active = false;
 };

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -0,0 +1,578 @@
+// Copyright 2015 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <memory>
+#include <string>
+#include <tuple>
+#include <utility>
+#include <glad/glad.h>
+#include "common/alignment.h"
+#include "common/assert.h"
+#include "common/logging/log.h"
+#include "common/math_util.h"
+#include "common/microprofile.h"
+#include "common/scope_exit.h"
+#include "common/vector_math.h"
+#include "core/core.h"
+#include "core/hle/kernel/process.h"
+#include "core/settings.h"
+#include "video_core/engines/maxwell_3d.h"
+#include "video_core/renderer_opengl/gl_rasterizer.h"
+#include "video_core/renderer_opengl/gl_shader_gen.h"
+#include "video_core/renderer_opengl/maxwell_to_gl.h"
+#include "video_core/renderer_opengl/renderer_opengl.h"
+
+using Maxwell = Tegra::Engines::Maxwell3D::Regs;
+using PixelFormat = SurfaceParams::PixelFormat;
+using SurfaceType = SurfaceParams::SurfaceType;
+
+MICROPROFILE_DEFINE(OpenGL_VAO, "OpenGL", "Vertex Array Setup", MP_RGB(128, 128, 192));
+MICROPROFILE_DEFINE(OpenGL_VS, "OpenGL", "Vertex Shader Setup", MP_RGB(128, 128, 192));
+MICROPROFILE_DEFINE(OpenGL_FS, "OpenGL", "Fragment Shader Setup", MP_RGB(128, 128, 192));
+MICROPROFILE_DEFINE(OpenGL_Drawing, "OpenGL", "Drawing", MP_RGB(128, 128, 192));
+MICROPROFILE_DEFINE(OpenGL_Blits, "OpenGL", "Blits", MP_RGB(100, 100, 255));
+MICROPROFILE_DEFINE(OpenGL_CacheManagement, "OpenGL", "Cache Mgmt", MP_RGB(100, 255, 100));
+
+enum class UniformBindings : GLuint { Common, VS, FS };
+
+static void SetShaderUniformBlockBinding(GLuint shader, const char* name, UniformBindings binding,
+                                         size_t expected_size) {
+    GLuint ub_index = glGetUniformBlockIndex(shader, name);
+    if (ub_index != GL_INVALID_INDEX) {
+        GLint ub_size = 0;
+        glGetActiveUniformBlockiv(shader, ub_index, GL_UNIFORM_BLOCK_DATA_SIZE, &ub_size);
+        ASSERT_MSG(ub_size == expected_size,
+                   "Uniform block size did not match! Got %d, expected %zu",
+                   static_cast<int>(ub_size), expected_size);
+        glUniformBlockBinding(shader, ub_index, static_cast<GLuint>(binding));
+    }
+}
+
+static void SetShaderUniformBlockBindings(GLuint shader) {
+    SetShaderUniformBlockBinding(shader, "shader_data", UniformBindings::Common,
+                                 sizeof(RasterizerOpenGL::UniformData));
+    SetShaderUniformBlockBinding(shader, "vs_config", UniformBindings::VS,
+                                 sizeof(RasterizerOpenGL::VSUniformData));
+    SetShaderUniformBlockBinding(shader, "fs_config", UniformBindings::FS,
+                                 sizeof(RasterizerOpenGL::FSUniformData));
+}
+
+RasterizerOpenGL::RasterizerOpenGL() {
+    shader_dirty = true;
+
+    has_ARB_buffer_storage = false;
+    has_ARB_direct_state_access = false;
+    has_ARB_separate_shader_objects = false;
+    has_ARB_vertex_attrib_binding = false;
+
+    GLint ext_num;
+    glGetIntegerv(GL_NUM_EXTENSIONS, &ext_num);
+    for (GLint i = 0; i < ext_num; i++) {
+        std::string extension{reinterpret_cast<const char*>(glGetStringi(GL_EXTENSIONS, i))};
+
+        if (extension == "GL_ARB_buffer_storage") {
+            has_ARB_buffer_storage = true;
+        } else if (extension == "GL_ARB_direct_state_access") {
+            has_ARB_direct_state_access = true;
+        } else if (extension == "GL_ARB_separate_shader_objects") {
+            has_ARB_separate_shader_objects = true;
+        } else if (extension == "GL_ARB_vertex_attrib_binding") {
+            has_ARB_vertex_attrib_binding = true;
+        }
+    }
+
+    // Clipping plane 0 is always enabled for PICA fixed clip plane z <= 0
+    state.clip_distance[0] = true;
+
+    // Generate VBO, VAO and UBO
+    vertex_buffer = OGLStreamBuffer::MakeBuffer(GLAD_GL_ARB_buffer_storage, GL_ARRAY_BUFFER);
+    vertex_buffer->Create(VERTEX_BUFFER_SIZE, VERTEX_BUFFER_SIZE / 2);
+    sw_vao.Create();
+    uniform_buffer.Create();
+
+    state.draw.vertex_array = sw_vao.handle;
+    state.draw.vertex_buffer = vertex_buffer->GetHandle();
+    state.draw.uniform_buffer = uniform_buffer.handle;
+    state.Apply();
+
+    glBufferData(GL_UNIFORM_BUFFER, sizeof(UniformData), nullptr, GL_STATIC_DRAW);
+    glBindBufferBase(GL_UNIFORM_BUFFER, 0, uniform_buffer.handle);
+
+    uniform_block_data.dirty = true;
+
+    // Create render framebuffer
+    framebuffer.Create();
+
+    if (has_ARB_separate_shader_objects) {
+        hw_vao.Create();
+        hw_vao_enabled_attributes.fill(false);
+
+        stream_buffer = OGLStreamBuffer::MakeBuffer(has_ARB_buffer_storage, GL_ARRAY_BUFFER);
+        stream_buffer->Create(STREAM_BUFFER_SIZE, STREAM_BUFFER_SIZE / 2);
+        state.draw.vertex_buffer = stream_buffer->GetHandle();
+
+        pipeline.Create();
+        state.draw.program_pipeline = pipeline.handle;
+        state.draw.shader_program = 0;
+        state.draw.vertex_array = hw_vao.handle;
+        state.Apply();
+
+        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, stream_buffer->GetHandle());
+
+        vs_uniform_buffer.Create();
+        glBindBuffer(GL_UNIFORM_BUFFER, vs_uniform_buffer.handle);
+        glBufferData(GL_UNIFORM_BUFFER, sizeof(VSUniformData), nullptr, GL_STREAM_COPY);
+        glBindBufferBase(GL_UNIFORM_BUFFER, 1, vs_uniform_buffer.handle);
+    } else {
+        UNREACHABLE();
+    }
+
+    accelerate_draw = AccelDraw::Disabled;
+
+    glEnable(GL_BLEND);
+
+    LOG_CRITICAL(Render_OpenGL, "Sync fixed function OpenGL state here!");
+}
+
+RasterizerOpenGL::~RasterizerOpenGL() {
+    if (stream_buffer != nullptr) {
+        state.draw.vertex_buffer = stream_buffer->GetHandle();
+        state.Apply();
+        stream_buffer->Release();
+    }
+}
+
+void RasterizerOpenGL::AnalyzeVertexArray(bool is_indexed) {
+    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+
+    if (is_indexed) {
+        UNREACHABLE();
+    }
+
+    // TODO(bunnei): Add support for 1+ vertex arrays
+    vs_input_size = regs.vertex_buffer.count * regs.vertex_array[0].stride;
+}
+
+void RasterizerOpenGL::SetupVertexArray(u8* array_ptr, GLintptr buffer_offset) {
+    MICROPROFILE_SCOPE(OpenGL_VAO);
+    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const auto& memory_manager = Core::System().GetInstance().GPU().memory_manager;
+
+    state.draw.vertex_array = hw_vao.handle;
+    state.draw.vertex_buffer = stream_buffer->GetHandle();
+    state.Apply();
+
+    // TODO(bunnei): Add support for 1+ vertex arrays
+    const auto& vertex_array{regs.vertex_array[0]};
+    ASSERT_MSG(vertex_array.enable, "vertex array 0 is disabled?");
+    ASSERT_MSG(!vertex_array.divisor, "vertex array 0 divisor is unimplemented!");
+    for (unsigned index = 1; index < Maxwell::NumVertexArrays; ++index) {
+        ASSERT_MSG(!regs.vertex_array[index].enable, "vertex array %d is unimplemented!", index);
+    }
+
+    // Use the vertex array as-is, assumes that the data is formatted correctly for OpenGL.
+    // Enables the first 16 vertex attributes always, as we don't know which ones are actually used
+    // until shader time. Note, Tegra technically supports 32, but we're cappinig this to 16 for now
+    // to avoid OpenGL errors.
+    for (unsigned index = 0; index < 16; ++index) {
+        auto& attrib = regs.vertex_attrib_format[index];
+        glVertexAttribPointer(index, attrib.ComponentCount(), MaxwellToGL::VertexType(attrib),
+                              attrib.IsNormalized() ? GL_TRUE : GL_FALSE, vertex_array.stride,
+                              reinterpret_cast<GLvoid*>(buffer_offset + attrib.offset));
+        glEnableVertexAttribArray(index);
+        hw_vao_enabled_attributes[index] = true;
+    }
+
+    // Copy vertex array data
+    const u32 data_size{vertex_array.stride * regs.vertex_buffer.count};
+    const VAddr data_addr{memory_manager->PhysicalToVirtualAddress(vertex_array.StartAddress())};
+    res_cache.FlushRegion(data_addr, data_size, nullptr);
+    Memory::ReadBlock(data_addr, array_ptr, data_size);
+
+    array_ptr += data_size;
+    buffer_offset += data_size;
+}
+
+void RasterizerOpenGL::SetupVertexShader(VSUniformData* ub_ptr, GLintptr buffer_offset) {
+    MICROPROFILE_SCOPE(OpenGL_VS);
+    LOG_CRITICAL(Render_OpenGL, "Emulated shaders are not supported! Using a passthrough shader.");
+    glUseProgramStages(pipeline.handle, GL_VERTEX_SHADER_BIT, current_shader->shader.handle);
+}
+
+void RasterizerOpenGL::SetupFragmentShader(FSUniformData* ub_ptr, GLintptr buffer_offset) {
+    MICROPROFILE_SCOPE(OpenGL_FS);
+    UNREACHABLE();
+}
+
+bool RasterizerOpenGL::AccelerateDrawBatch(bool is_indexed) {
+    if (!has_ARB_separate_shader_objects) {
+        UNREACHABLE();
+        return false;
+    }
+
+    accelerate_draw = is_indexed ? AccelDraw::Indexed : AccelDraw::Arrays;
+    DrawArrays();
+
+    return true;
+}
+
+void RasterizerOpenGL::DrawArrays() {
+    if (accelerate_draw == AccelDraw::Disabled)
+        return;
+
+    MICROPROFILE_SCOPE(OpenGL_Drawing);
+    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+
+    // TODO(bunnei): Implement these
+    const bool has_stencil = false;
+    const bool using_color_fb = true;
+    const bool using_depth_fb = false;
+    const MathUtil::Rectangle<s32> viewport_rect{regs.viewport[0].GetRect()};
+
+    const bool write_color_fb =
+        state.color_mask.red_enabled == GL_TRUE || state.color_mask.green_enabled == GL_TRUE ||
+        state.color_mask.blue_enabled == GL_TRUE || state.color_mask.alpha_enabled == GL_TRUE;
+
+    const bool write_depth_fb =
+        (state.depth.test_enabled && state.depth.write_mask == GL_TRUE) ||
+        (has_stencil && state.stencil.test_enabled && state.stencil.write_mask != 0);
+
+    Surface color_surface;
+    Surface depth_surface;
+    MathUtil::Rectangle<u32> surfaces_rect;
+    std::tie(color_surface, depth_surface, surfaces_rect) =
+        res_cache.GetFramebufferSurfaces(using_color_fb, using_depth_fb, viewport_rect);
+
+    const u16 res_scale = color_surface != nullptr
+                              ? color_surface->res_scale
+                              : (depth_surface == nullptr ? 1u : depth_surface->res_scale);
+
+    MathUtil::Rectangle<u32> draw_rect{
+        static_cast<u32>(MathUtil::Clamp<s32>(static_cast<s32>(surfaces_rect.left) +
+                                                  viewport_rect.left * res_scale,
+                                              surfaces_rect.left, surfaces_rect.right)), // Left
+        static_cast<u32>(MathUtil::Clamp<s32>(static_cast<s32>(surfaces_rect.bottom) +
+                                                  viewport_rect.top * res_scale,
+                                              surfaces_rect.bottom, surfaces_rect.top)), // Top
+        static_cast<u32>(MathUtil::Clamp<s32>(static_cast<s32>(surfaces_rect.left) +
+                                                  viewport_rect.right * res_scale,
+                                              surfaces_rect.left, surfaces_rect.right)), // Right
+        static_cast<u32>(MathUtil::Clamp<s32>(static_cast<s32>(surfaces_rect.bottom) +
+                                                  viewport_rect.bottom * res_scale,
+                                              surfaces_rect.bottom, surfaces_rect.top))}; // Bottom
+
+    // Bind the framebuffer surfaces
+    BindFramebufferSurfaces(color_surface, depth_surface, has_stencil);
+
+    // Sync the viewport
+    SyncViewport(surfaces_rect, res_scale);
+
+    // TODO(bunnei): Sync framebuffer_scale uniform here
+    // TODO(bunnei): Sync scissorbox uniform(s) here
+    // TODO(bunnei): Sync and bind the texture surfaces
+
+    // Sync and bind the shader
+    if (shader_dirty) {
+        SetShader();
+        shader_dirty = false;
+    }
+
+    // Sync the uniform data
+    if (uniform_block_data.dirty) {
+        glBufferSubData(GL_UNIFORM_BUFFER, 0, sizeof(UniformData), &uniform_block_data.data);
+        uniform_block_data.dirty = false;
+    }
+
+    // Viewport can have negative offsets or larger dimensions than our framebuffer sub-rect. Enable
+    // scissor test to prevent drawing outside of the framebuffer region
+    state.scissor.enabled = true;
+    state.scissor.x = draw_rect.left;
+    state.scissor.y = draw_rect.bottom;
+    state.scissor.width = draw_rect.GetWidth();
+    state.scissor.height = draw_rect.GetHeight();
+    state.Apply();
+
+    // Draw the vertex batch
+    const bool is_indexed = accelerate_draw == AccelDraw::Indexed;
+    AnalyzeVertexArray(is_indexed);
+    state.draw.vertex_buffer = stream_buffer->GetHandle();
+    state.Apply();
+
+    size_t buffer_size = static_cast<size_t>(vs_input_size);
+    if (is_indexed) {
+        UNREACHABLE();
+    }
+    buffer_size += sizeof(VSUniformData);
+
+    size_t ptr_pos = 0;
+    u8* buffer_ptr;
+    GLintptr buffer_offset;
+    std::tie(buffer_ptr, buffer_offset) =
+        stream_buffer->Map(static_cast<GLsizeiptr>(buffer_size), 4);
+
+    SetupVertexArray(buffer_ptr, buffer_offset);
+    ptr_pos += vs_input_size;
+
+    GLintptr index_buffer_offset = 0;
+    if (is_indexed) {
+        UNREACHABLE();
+    }
+
+    SetupVertexShader(reinterpret_cast<VSUniformData*>(&buffer_ptr[ptr_pos]),
+                      buffer_offset + static_cast<GLintptr>(ptr_pos));
+    const GLintptr vs_ubo_offset = buffer_offset + static_cast<GLintptr>(ptr_pos);
+    ptr_pos += sizeof(VSUniformData);
+
+    stream_buffer->Unmap();
+
+    const auto copy_buffer = [&](GLuint handle, GLintptr offset, GLsizeiptr size) {
+        if (has_ARB_direct_state_access) {
+            glCopyNamedBufferSubData(stream_buffer->GetHandle(), handle, offset, 0, size);
+        } else {
+            glBindBuffer(GL_COPY_WRITE_BUFFER, handle);
+            glCopyBufferSubData(GL_ARRAY_BUFFER, GL_COPY_WRITE_BUFFER, offset, 0, size);
+        }
+    };
+
+    copy_buffer(vs_uniform_buffer.handle, vs_ubo_offset, sizeof(VSUniformData));
+
+    glUseProgramStages(pipeline.handle, GL_FRAGMENT_SHADER_BIT, current_shader->shader.handle);
+
+    if (is_indexed) {
+        UNREACHABLE();
+    } else {
+        glDrawArrays(MaxwellToGL::PrimitiveTopology(regs.draw.topology), 0,
+                     regs.vertex_buffer.count);
+    }
+
+    // Disable scissor test
+    state.scissor.enabled = false;
+
+    accelerate_draw = AccelDraw::Disabled;
+
+    // Unbind textures for potential future use as framebuffer attachments
+    for (auto& texture_unit : state.texture_units) {
+        texture_unit.texture_2d = 0;
+    }
+    state.Apply();
+
+    // Mark framebuffer surfaces as dirty
+    MathUtil::Rectangle<u32> draw_rect_unscaled{
+        draw_rect.left / res_scale, draw_rect.top / res_scale, draw_rect.right / res_scale,
+        draw_rect.bottom / res_scale};
+
+    if (color_surface != nullptr && write_color_fb) {
+        auto interval = color_surface->GetSubRectInterval(draw_rect_unscaled);
+        res_cache.InvalidateRegion(boost::icl::first(interval), boost::icl::length(interval),
+                                   color_surface);
+    }
+    if (depth_surface != nullptr && write_depth_fb) {
+        auto interval = depth_surface->GetSubRectInterval(draw_rect_unscaled);
+        res_cache.InvalidateRegion(boost::icl::first(interval), boost::icl::length(interval),
+                                   depth_surface);
+    }
+}
+
+void RasterizerOpenGL::NotifyMaxwellRegisterChanged(u32 id) {}
+
+void RasterizerOpenGL::FlushAll() {
+    MICROPROFILE_SCOPE(OpenGL_CacheManagement);
+    res_cache.FlushAll();
+}
+
+void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {
+    MICROPROFILE_SCOPE(OpenGL_CacheManagement);
+    res_cache.FlushRegion(addr, size);
+}
+
+void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
+    MICROPROFILE_SCOPE(OpenGL_CacheManagement);
+    res_cache.InvalidateRegion(addr, size, nullptr);
+}
+
+void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) {
+    MICROPROFILE_SCOPE(OpenGL_CacheManagement);
+    res_cache.FlushRegion(addr, size);
+    res_cache.InvalidateRegion(addr, size, nullptr);
+}
+
+bool RasterizerOpenGL::AccelerateDisplayTransfer(const void* config) {
+    MICROPROFILE_SCOPE(OpenGL_Blits);
+    UNREACHABLE();
+    return true;
+}
+
+bool RasterizerOpenGL::AccelerateTextureCopy(const void* config) {
+    UNREACHABLE();
+    return true;
+}
+
+bool RasterizerOpenGL::AccelerateFill(const void* config) {
+    UNREACHABLE();
+    return true;
+}
+
+bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& framebuffer,
+                                         VAddr framebuffer_addr, u32 pixel_stride,
+                                         ScreenInfo& screen_info) {
+    if (framebuffer_addr == 0) {
+        return false;
+    }
+    MICROPROFILE_SCOPE(OpenGL_CacheManagement);
+
+    SurfaceParams src_params;
+    src_params.addr = framebuffer_addr;
+    src_params.width = std::min(framebuffer.width, pixel_stride);
+    src_params.height = framebuffer.height;
+    src_params.stride = pixel_stride;
+    src_params.is_tiled = false;
+    src_params.pixel_format =
+        SurfaceParams::PixelFormatFromGPUPixelFormat(framebuffer.pixel_format);
+    src_params.UpdateParams();
+
+    MathUtil::Rectangle<u32> src_rect;
+    Surface src_surface;
+    std::tie(src_surface, src_rect) =
+        res_cache.GetSurfaceSubRect(src_params, ScaleMatch::Ignore, true);
+
+    if (src_surface == nullptr) {
+        return false;
+    }
+
+    u32 scaled_width = src_surface->GetScaledWidth();
+    u32 scaled_height = src_surface->GetScaledHeight();
+
+    screen_info.display_texcoords = MathUtil::Rectangle<float>(
+        (float)src_rect.bottom / (float)scaled_height, (float)src_rect.left / (float)scaled_width,
+        (float)src_rect.top / (float)scaled_height, (float)src_rect.right / (float)scaled_width);
+
+    screen_info.display_texture = src_surface->texture.handle;
+
+    return true;
+}
+
+void RasterizerOpenGL::SetShader() {
+    // TODO(bunnei): The below sets up a static test shader for passing untransformed vertices to
+    // OpenGL for rendering. This should be removed/replaced when we start emulating Maxwell
+    // shaders.
+
+    static constexpr char vertex_shader[] = R"(
+#version 150 core
+
+in vec2 vert_position;
+in vec2 vert_tex_coord;
+out vec2 frag_tex_coord;
+
+void main() {
+    // Multiply input position by the rotscale part of the matrix and then manually translate by
+    // the last column. This is equivalent to using a full 3x3 matrix and expanding the vector
+    // to `vec3(vert_position.xy, 1.0)`
+    gl_Position = vec4(mat2(mat3x2(0.0015625f, 0.0, 0.0, -0.0027778, -1.0, 1.0)) * vert_position + mat3x2(0.0015625f, 0.0, 0.0, -0.0027778, -1.0, 1.0)[2], 0.0, 1.0);
+    frag_tex_coord = vert_tex_coord;
+}
+)";
+
+    static constexpr char fragment_shader[] = R"(
+#version 150 core
+
+in vec2 frag_tex_coord;
+out vec4 color;
+
+uniform sampler2D color_texture;
+
+void main() {
+    color = vec4(1.0, 0.0, 1.0, 0.0);
+}
+)";
+
+    if (current_shader) {
+        return;
+    }
+
+    LOG_CRITICAL(Render_OpenGL, "Emulated shaders are not supported! Using a passthrough shader.");
+
+    current_shader = &test_shader;
+    if (has_ARB_separate_shader_objects) {
+        test_shader.shader.Create(vertex_shader, nullptr, fragment_shader, {}, true);
+        glActiveShaderProgram(pipeline.handle, test_shader.shader.handle);
+    } else {
+        UNREACHABLE();
+    }
+
+    state.draw.shader_program = test_shader.shader.handle;
+    state.Apply();
+
+    if (has_ARB_separate_shader_objects) {
+        state.draw.shader_program = 0;
+        state.Apply();
+    }
+}
+
+void RasterizerOpenGL::BindFramebufferSurfaces(const Surface& color_surface,
+                                               const Surface& depth_surface, bool has_stencil) {
+    state.draw.draw_framebuffer = framebuffer.handle;
+    state.Apply();
+
+    glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D,
+                           color_surface != nullptr ? color_surface->texture.handle : 0, 0);
+    if (depth_surface != nullptr) {
+        if (has_stencil) {
+            // attach both depth and stencil
+            glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
+                                   depth_surface->texture.handle, 0);
+        } else {
+            // attach depth
+            glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D,
+                                   depth_surface->texture.handle, 0);
+            // clear stencil attachment
+            glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
+        }
+    } else {
+        // clear both depth and stencil attachment
+        glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
+                               0);
+    }
+}
+
+void RasterizerOpenGL::SyncViewport(const MathUtil::Rectangle<u32>& surfaces_rect, u16 res_scale) {
+    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const MathUtil::Rectangle<s32> viewport_rect{regs.viewport[0].GetRect()};
+
+    state.viewport.x = static_cast<GLint>(surfaces_rect.left) + viewport_rect.left * res_scale;
+    state.viewport.y = static_cast<GLint>(surfaces_rect.bottom) + viewport_rect.bottom * res_scale;
+    state.viewport.width = static_cast<GLsizei>(viewport_rect.GetWidth() * res_scale);
+    state.viewport.height = static_cast<GLsizei>(viewport_rect.GetHeight() * res_scale);
+}
+
+void RasterizerOpenGL::SyncClipEnabled() {
+    UNREACHABLE();
+}
+
+void RasterizerOpenGL::SyncClipCoef() {
+    UNREACHABLE();
+}
+
+void RasterizerOpenGL::SyncCullMode() {
+    UNREACHABLE();
+}
+
+void RasterizerOpenGL::SyncDepthScale() {
+    UNREACHABLE();
+}
+
+void RasterizerOpenGL::SyncDepthOffset() {
+    UNREACHABLE();
+}
+
+void RasterizerOpenGL::SyncBlendEnabled() {
+    UNREACHABLE();
+}
+
+void RasterizerOpenGL::SyncBlendFuncs() {
+    UNREACHABLE();
+}
+
+void RasterizerOpenGL::SyncBlendColor() {
+    UNREACHABLE();
+}

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -0,0 +1,179 @@
+// Copyright 2015 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <cstddef>
+#include <cstring>
+#include <memory>
+#include <unordered_map>
+#include <vector>
+#include <glad/glad.h>
+#include "common/bit_field.h"
+#include "common/common_types.h"
+#include "common/hash.h"
+#include "common/vector_math.h"
+#include "video_core/rasterizer_interface.h"
+#include "video_core/renderer_opengl/gl_rasterizer_cache.h"
+#include "video_core/renderer_opengl/gl_resource_manager.h"
+#include "video_core/renderer_opengl/gl_shader_gen.h"
+#include "video_core/renderer_opengl/gl_state.h"
+#include "video_core/renderer_opengl/gl_stream_buffer.h"
+
+struct ScreenInfo;
+
+class RasterizerOpenGL : public VideoCore::RasterizerInterface {
+public:
+    RasterizerOpenGL();
+    ~RasterizerOpenGL() override;
+
+    void DrawArrays() override;
+    void NotifyMaxwellRegisterChanged(u32 id) override;
+    void FlushAll() override;
+    void FlushRegion(VAddr addr, u64 size) override;
+    void InvalidateRegion(VAddr addr, u64 size) override;
+    void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
+    bool AccelerateDisplayTransfer(const void* config) override;
+    bool AccelerateTextureCopy(const void* config) override;
+    bool AccelerateFill(const void* config) override;
+    bool AccelerateDisplay(const Tegra::FramebufferConfig& framebuffer, VAddr framebuffer_addr,
+                           u32 pixel_stride, ScreenInfo& screen_info) override;
+    bool AccelerateDrawBatch(bool is_indexed) override;
+
+    /// OpenGL shader generated for a given Maxwell register state
+    struct MaxwellShader {
+        /// OpenGL shader resource
+        OGLShader shader;
+    };
+
+    struct VertexShader {
+        OGLShader shader;
+    };
+
+    struct FragmentShader {
+        OGLShader shader;
+    };
+
+    /// Uniform structure for the Uniform Buffer Object, all vectors must be 16-byte aligned
+    // NOTE: Always keep a vec4 at the end. The GL spec is not clear wether the alignment at
+    //       the end of a uniform block is included in UNIFORM_BLOCK_DATA_SIZE or not.
+    //       Not following that rule will cause problems on some AMD drivers.
+    struct UniformData {};
+
+    // static_assert(
+    //    sizeof(UniformData) == 0x460,
+    //    "The size of the UniformData structure has changed, update the structure in the shader");
+    static_assert(sizeof(UniformData) < 16384,
+                  "UniformData structure must be less than 16kb as per the OpenGL spec");
+
+    struct VSUniformData {};
+    // static_assert(
+    //    sizeof(VSUniformData) == 1856,
+    //    "The size of the VSUniformData structure has changed, update the structure in the
+    //    shader");
+    static_assert(sizeof(VSUniformData) < 16384,
+                  "VSUniformData structure must be less than 16kb as per the OpenGL spec");
+
+    struct FSUniformData {};
+    // static_assert(
+    //    sizeof(FSUniformData) == 1856,
+    //    "The size of the FSUniformData structure has changed, update the structure in the
+    //    shader");
+    static_assert(sizeof(FSUniformData) < 16384,
+                  "FSUniformData structure must be less than 16kb as per the OpenGL spec");
+
+private:
+    struct SamplerInfo {};
+
+    /// Binds the framebuffer color and depth surface
+    void BindFramebufferSurfaces(const Surface& color_surface, const Surface& depth_surface,
+                                 bool has_stencil);
+
+    /// Syncs the viewport to match the guest state
+    void SyncViewport(const MathUtil::Rectangle<u32>& surfaces_rect, u16 res_scale);
+
+    /// Syncs the clip enabled status to match the guest state
+    void SyncClipEnabled();
+
+    /// Syncs the clip coefficients to match the guest state
+    void SyncClipCoef();
+
+    /// Sets the OpenGL shader in accordance with the current guest state
+    void SetShader();
+
+    /// Syncs the cull mode to match the guest state
+    void SyncCullMode();
+
+    /// Syncs the depth scale to match the guest state
+    void SyncDepthScale();
+
+    /// Syncs the depth offset to match the guest state
+    void SyncDepthOffset();
+
+    /// Syncs the blend enabled status to match the guest state
+    void SyncBlendEnabled();
+
+    /// Syncs the blend functions to match the guest state
+    void SyncBlendFuncs();
+
+    /// Syncs the blend color to match the guest state
+    void SyncBlendColor();
+
+    bool has_ARB_buffer_storage;
+    bool has_ARB_direct_state_access;
+    bool has_ARB_separate_shader_objects;
+    bool has_ARB_vertex_attrib_binding;
+
+    OpenGLState state;
+
+    RasterizerCacheOpenGL res_cache;
+
+    /// Shader used for test renderering - to be removed once we have emulated shaders
+    MaxwellShader test_shader{};
+
+    const MaxwellShader* current_shader{};
+    bool shader_dirty{};
+
+    struct {
+        UniformData data;
+        bool dirty;
+    } uniform_block_data = {};
+
+    OGLPipeline pipeline;
+    OGLVertexArray sw_vao;
+    OGLVertexArray hw_vao;
+    std::array<bool, 16> hw_vao_enabled_attributes;
+
+    std::array<SamplerInfo, 32> texture_samplers;
+    static constexpr size_t VERTEX_BUFFER_SIZE = 128 * 1024 * 1024;
+    std::unique_ptr<OGLStreamBuffer> vertex_buffer;
+    OGLBuffer uniform_buffer;
+    OGLFramebuffer framebuffer;
+
+    static constexpr size_t STREAM_BUFFER_SIZE = 4 * 1024 * 1024;
+    std::unique_ptr<OGLStreamBuffer> stream_buffer;
+
+    GLsizeiptr vs_input_size;
+
+    void AnalyzeVertexArray(bool is_indexed);
+    void SetupVertexArray(u8* array_ptr, GLintptr buffer_offset);
+
+    OGLBuffer vs_uniform_buffer;
+    std::unordered_map<GLShader::MaxwellVSConfig, VertexShader*> vs_shader_map;
+    std::unordered_map<std::string, VertexShader> vs_shader_cache;
+    OGLShader vs_default_shader;
+
+    void SetupVertexShader(VSUniformData* ub_ptr, GLintptr buffer_offset);
+
+    OGLBuffer fs_uniform_buffer;
+    std::unordered_map<GLShader::MaxwellFSConfig, FragmentShader*> fs_shader_map;
+    std::unordered_map<std::string, FragmentShader> fs_shader_cache;
+    OGLShader fs_default_shader;
+
+    void SetupFragmentShader(FSUniformData* ub_ptr, GLintptr buffer_offset);
+
+    enum class AccelDraw { Disabled, Arrays, Indexed };
+    AccelDraw accelerate_draw;
+};

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -0,0 +1,369 @@
+// Copyright 2015 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <memory>
+#include <set>
+#include <tuple>
+#ifdef __GNUC__
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wunused-local-typedefs"
+#endif
+#include <boost/icl/interval_map.hpp>
+#include <boost/icl/interval_set.hpp>
+#ifdef __GNUC__
+#pragma GCC diagnostic pop
+#endif
+#include <glad/glad.h>
+#include "common/assert.h"
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "common/math_util.h"
+#include "video_core/gpu.h"
+#include "video_core/renderer_opengl/gl_resource_manager.h"
+
+struct CachedSurface;
+using Surface = std::shared_ptr<CachedSurface>;
+using SurfaceSet = std::set<Surface>;
+
+using SurfaceRegions = boost::icl::interval_set<VAddr>;
+using SurfaceMap = boost::icl::interval_map<VAddr, Surface>;
+using SurfaceCache = boost::icl::interval_map<VAddr, SurfaceSet>;
+
+using SurfaceInterval = SurfaceCache::interval_type;
+static_assert(std::is_same<SurfaceRegions::interval_type, SurfaceCache::interval_type>() &&
+                  std::is_same<SurfaceMap::interval_type, SurfaceCache::interval_type>(),
+              "incorrect interval types");
+
+using SurfaceRect_Tuple = std::tuple<Surface, MathUtil::Rectangle<u32>>;
+using SurfaceSurfaceRect_Tuple = std::tuple<Surface, Surface, MathUtil::Rectangle<u32>>;
+
+using PageMap = boost::icl::interval_map<u64, int>;
+
+enum class ScaleMatch {
+    Exact,   // only accept same res scale
+    Upscale, // only allow higher scale than params
+    Ignore   // accept every scaled res
+};
+
+struct SurfaceParams {
+    enum class PixelFormat {
+        // First 5 formats are shared between textures and color buffers
+        RGBA8 = 0,
+        RGB8 = 1,
+        RGB5A1 = 2,
+        RGB565 = 3,
+        RGBA4 = 4,
+
+        // Texture-only formats
+        IA8 = 5,
+        RG8 = 6,
+        I8 = 7,
+        A8 = 8,
+        IA4 = 9,
+        I4 = 10,
+        A4 = 11,
+        ETC1 = 12,
+        ETC1A4 = 13,
+
+        // Depth buffer-only formats
+        D16 = 14,
+        // gap
+        D24 = 16,
+        D24S8 = 17,
+
+        Invalid = 255,
+    };
+
+    enum class SurfaceType {
+        Color = 0,
+        Texture = 1,
+        Depth = 2,
+        DepthStencil = 3,
+        Fill = 4,
+        Invalid = 5
+    };
+
+    static constexpr unsigned int GetFormatBpp(PixelFormat format) {
+        constexpr std::array<unsigned int, 18> bpp_table = {
+            32, // RGBA8
+            24, // RGB8
+            16, // RGB5A1
+            16, // RGB565
+            16, // RGBA4
+            16, // IA8
+            16, // RG8
+            8,  // I8
+            8,  // A8
+            8,  // IA4
+            4,  // I4
+            4,  // A4
+            4,  // ETC1
+            8,  // ETC1A4
+            16, // D16
+            0,
+            24, // D24
+            32, // D24S8
+        };
+
+        assert(static_cast<size_t>(format) < bpp_table.size());
+        return bpp_table[static_cast<size_t>(format)];
+    }
+    unsigned int GetFormatBpp() const {
+        return GetFormatBpp(pixel_format);
+    }
+
+    static PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format) {
+        switch (format) {
+        case Tegra::RenderTargetFormat::RGBA8_UNORM:
+            return PixelFormat::RGBA8;
+        default:
+            UNREACHABLE();
+        }
+    }
+
+    static PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat format) {
+        switch (format) {
+        case Tegra::FramebufferConfig::PixelFormat::ABGR8:
+            return PixelFormat::RGBA8;
+        default:
+            UNREACHABLE();
+        }
+    }
+
+    static bool CheckFormatsBlittable(PixelFormat pixel_format_a, PixelFormat pixel_format_b) {
+        SurfaceType a_type = GetFormatType(pixel_format_a);
+        SurfaceType b_type = GetFormatType(pixel_format_b);
+
+        if ((a_type == SurfaceType::Color || a_type == SurfaceType::Texture) &&
+            (b_type == SurfaceType::Color || b_type == SurfaceType::Texture)) {
+            return true;
+        }
+
+        if (a_type == SurfaceType::Depth && b_type == SurfaceType::Depth) {
+            return true;
+        }
+
+        if (a_type == SurfaceType::DepthStencil && b_type == SurfaceType::DepthStencil) {
+            return true;
+        }
+
+        return false;
+    }
+
+    static constexpr SurfaceType GetFormatType(PixelFormat pixel_format) {
+        if ((unsigned int)pixel_format < 5) {
+            return SurfaceType::Color;
+        }
+
+        if ((unsigned int)pixel_format < 14) {
+            return SurfaceType::Texture;
+        }
+
+        if (pixel_format == PixelFormat::D16 || pixel_format == PixelFormat::D24) {
+            return SurfaceType::Depth;
+        }
+
+        if (pixel_format == PixelFormat::D24S8) {
+            return SurfaceType::DepthStencil;
+        }
+
+        return SurfaceType::Invalid;
+    }
+
+    /// Update the params "size", "end" and "type" from the already set "addr", "width", "height"
+    /// and "pixel_format"
+    void UpdateParams() {
+        if (stride == 0) {
+            stride = width;
+        }
+        type = GetFormatType(pixel_format);
+        size = !is_tiled ? BytesInPixels(stride * (height - 1) + width)
+                         : BytesInPixels(stride * 8 * (height / 8 - 1) + width * 8);
+        end = addr + size;
+    }
+
+    SurfaceInterval GetInterval() const {
+        return SurfaceInterval::right_open(addr, end);
+    }
+
+    // Returns the outer rectangle containing "interval"
+    SurfaceParams FromInterval(SurfaceInterval interval) const;
+
+    SurfaceInterval GetSubRectInterval(MathUtil::Rectangle<u32> unscaled_rect) const;
+
+    // Returns the region of the biggest valid rectange within interval
+    SurfaceInterval GetCopyableInterval(const Surface& src_surface) const;
+
+    u32 GetScaledWidth() const {
+        return width * res_scale;
+    }
+
+    u32 GetScaledHeight() const {
+        return height * res_scale;
+    }
+
+    MathUtil::Rectangle<u32> GetRect() const {
+        return {0, height, width, 0};
+    }
+
+    MathUtil::Rectangle<u32> GetScaledRect() const {
+        return {0, GetScaledHeight(), GetScaledWidth(), 0};
+    }
+
+    u64 PixelsInBytes(u64 size) const {
+        return size * CHAR_BIT / GetFormatBpp(pixel_format);
+    }
+
+    u64 BytesInPixels(u64 pixels) const {
+        return pixels * GetFormatBpp(pixel_format) / CHAR_BIT;
+    }
+
+    bool ExactMatch(const SurfaceParams& other_surface) const;
+    bool CanSubRect(const SurfaceParams& sub_surface) const;
+    bool CanExpand(const SurfaceParams& expanded_surface) const;
+    bool CanTexCopy(const SurfaceParams& texcopy_params) const;
+
+    MathUtil::Rectangle<u32> GetSubRect(const SurfaceParams& sub_surface) const;
+    MathUtil::Rectangle<u32> GetScaledSubRect(const SurfaceParams& sub_surface) const;
+
+    VAddr addr = 0;
+    VAddr end = 0;
+    u64 size = 0;
+
+    u32 width = 0;
+    u32 height = 0;
+    u32 stride = 0;
+    u16 res_scale = 1;
+
+    bool is_tiled = false;
+    PixelFormat pixel_format = PixelFormat::Invalid;
+    SurfaceType type = SurfaceType::Invalid;
+};
+
+struct CachedSurface : SurfaceParams {
+    bool CanFill(const SurfaceParams& dest_surface, SurfaceInterval fill_interval) const;
+    bool CanCopy(const SurfaceParams& dest_surface, SurfaceInterval copy_interval) const;
+
+    bool IsRegionValid(SurfaceInterval interval) const {
+        return (invalid_regions.find(interval) == invalid_regions.end());
+    }
+
+    bool IsSurfaceFullyInvalid() const {
+        return (invalid_regions & GetInterval()) == SurfaceRegions(GetInterval());
+    }
+
+    bool registered = false;
+    SurfaceRegions invalid_regions;
+
+    u64 fill_size = 0; /// Number of bytes to read from fill_data
+    std::array<u8, 4> fill_data;
+
+    OGLTexture texture;
+
+    static constexpr unsigned int GetGLBytesPerPixel(PixelFormat format) {
+        // OpenGL needs 4 bpp alignment for D24 since using GL_UNSIGNED_INT as type
+        return format == PixelFormat::Invalid
+                   ? 0
+                   : (format == PixelFormat::D24 || GetFormatType(format) == SurfaceType::Texture)
+                         ? 4
+                         : SurfaceParams::GetFormatBpp(format) / 8;
+    }
+
+    std::unique_ptr<u8[]> gl_buffer;
+    size_t gl_buffer_size = 0;
+
+    // Read/Write data in Switch memory to/from gl_buffer
+    void LoadGLBuffer(VAddr load_start, VAddr load_end);
+    void FlushGLBuffer(VAddr flush_start, VAddr flush_end);
+
+    // Upload/Download data in gl_buffer in/to this surface's texture
+    void UploadGLTexture(const MathUtil::Rectangle<u32>& rect, GLuint read_fb_handle,
+                         GLuint draw_fb_handle);
+    void DownloadGLTexture(const MathUtil::Rectangle<u32>& rect, GLuint read_fb_handle,
+                           GLuint draw_fb_handle);
+};
+
+class RasterizerCacheOpenGL : NonCopyable {
+public:
+    RasterizerCacheOpenGL();
+    ~RasterizerCacheOpenGL();
+
+    /// Blit one surface's texture to another
+    bool BlitSurfaces(const Surface& src_surface, const MathUtil::Rectangle<u32>& src_rect,
+                      const Surface& dst_surface, const MathUtil::Rectangle<u32>& dst_rect);
+
+    void ConvertD24S8toABGR(GLuint src_tex, const MathUtil::Rectangle<u32>& src_rect,
+                            GLuint dst_tex, const MathUtil::Rectangle<u32>& dst_rect);
+
+    /// Copy one surface's region to another
+    void CopySurface(const Surface& src_surface, const Surface& dst_surface,
+                     SurfaceInterval copy_interval);
+
+    /// Load a texture from 3DS memory to OpenGL and cache it (if not already cached)
+    Surface GetSurface(const SurfaceParams& params, ScaleMatch match_res_scale,
+                       bool load_if_create);
+
+    /// Attempt to find a subrect (resolution scaled) of a surface, otherwise loads a texture from
+    /// 3DS memory to OpenGL and caches it (if not already cached)
+    SurfaceRect_Tuple GetSurfaceSubRect(const SurfaceParams& params, ScaleMatch match_res_scale,
+                                        bool load_if_create);
+
+    /// Get a surface based on the texture configuration
+    Surface GetTextureSurface(const void* config);
+
+    /// Get the color and depth surfaces based on the framebuffer configuration
+    SurfaceSurfaceRect_Tuple GetFramebufferSurfaces(bool using_color_fb, bool using_depth_fb,
+                                                    const MathUtil::Rectangle<s32>& viewport);
+
+    /// Get a surface that matches the fill config
+    Surface GetFillSurface(const void* config);
+
+    /// Get a surface that matches a "texture copy" display transfer config
+    SurfaceRect_Tuple GetTexCopySurface(const SurfaceParams& params);
+
+    /// Write any cached resources overlapping the region back to memory (if dirty)
+    void FlushRegion(VAddr addr, u64 size, Surface flush_surface = nullptr);
+
+    /// Mark region as being invalidated by region_owner (nullptr if 3DS memory)
+    void InvalidateRegion(VAddr addr, u64 size, const Surface& region_owner);
+
+    /// Flush all cached resources tracked by this cache manager
+    void FlushAll();
+
+private:
+    void DuplicateSurface(const Surface& src_surface, const Surface& dest_surface);
+
+    /// Update surface's texture for given region when necessary
+    void ValidateSurface(const Surface& surface, VAddr addr, u64 size);
+
+    /// Create a new surface
+    Surface CreateSurface(const SurfaceParams& params);
+
+    /// Register surface into the cache
+    void RegisterSurface(const Surface& surface);
+
+    /// Remove surface from the cache
+    void UnregisterSurface(const Surface& surface);
+
+    /// Increase/decrease the number of surface in pages touching the specified region
+    void UpdatePagesCachedCount(VAddr addr, u64 size, int delta);
+
+    SurfaceCache surface_cache;
+    PageMap cached_pages;
+    SurfaceMap dirty_regions;
+    SurfaceSet remove_surfaces;
+
+    OGLFramebuffer read_framebuffer;
+    OGLFramebuffer draw_framebuffer;
+
+    OGLVertexArray attributeless_vao;
+    OGLBuffer d24s8_abgr_buffer;
+    GLsizeiptr d24s8_abgr_buffer_size;
+    OGLShader d24s8_abgr_shader;
+    GLint d24s8_abgr_tbo_size_u_id;
+    GLint d24s8_abgr_viewport_u_id;
+};

src/video_core/renderer_opengl/gl_resource_manager.h

@@ -36,7 +36,7 @@ public:
         if (handle == 0)
             return;
         glDeleteTextures(1, &handle);
-        OpenGLState::ResetTexture(handle);
+        OpenGLState::GetCurState().ResetTexture(handle).Apply();
         handle = 0;
     }
 
@@ -69,7 +69,7 @@ public:
         if (handle == 0)
             return;
         glDeleteSamplers(1, &handle);
-        OpenGLState::ResetSampler(handle);
+        OpenGLState::GetCurState().ResetSampler(handle).Apply();
         handle = 0;
     }
 
@@ -91,10 +91,13 @@ public:
     }
 
     /// Creates a new internal OpenGL resource and stores the handle
-    void Create(const char* vert_shader, const char* frag_shader) {
+    void Create(const char* vert_shader, const char* geo_shader, const char* frag_shader,
+                const std::vector<const char*>& feedback_vars = {},
+                bool separable_program = false) {
         if (handle != 0)
             return;
-        handle = GLShader::LoadProgram(vert_shader, frag_shader);
+        handle = GLShader::LoadProgram(vert_shader, geo_shader, frag_shader, feedback_vars,
+                                       separable_program);
     }
 
     /// Deletes the internal OpenGL resource
@@ -102,7 +105,40 @@ public:
         if (handle == 0)
             return;
         glDeleteProgram(handle);
-        OpenGLState::ResetProgram(handle);
+        OpenGLState::GetCurState().ResetProgram(handle).Apply();
+        handle = 0;
+    }
+
+    GLuint handle = 0;
+};
+
+class OGLPipeline : private NonCopyable {
+public:
+    OGLPipeline() = default;
+    OGLPipeline(OGLPipeline&& o) {
+        handle = std::exchange<GLuint>(o.handle, 0);
+    }
+    ~OGLPipeline() {
+        Release();
+    }
+    OGLPipeline& operator=(OGLPipeline&& o) {
+        handle = std::exchange<GLuint>(o.handle, 0);
+        return *this;
+    }
+
+    /// Creates a new internal OpenGL resource and stores the handle
+    void Create() {
+        if (handle != 0)
+            return;
+        glGenProgramPipelines(1, &handle);
+    }
+
+    /// Deletes the internal OpenGL resource
+    void Release() {
+        if (handle == 0)
+            return;
+        glDeleteProgramPipelines(1, &handle);
+        OpenGLState::GetCurState().ResetPipeline(handle).Apply();
         handle = 0;
     }
 
@@ -135,13 +171,46 @@ public:
         if (handle == 0)
             return;
         glDeleteBuffers(1, &handle);
-        OpenGLState::ResetBuffer(handle);
+        OpenGLState::GetCurState().ResetBuffer(handle).Apply();
         handle = 0;
     }
 
     GLuint handle = 0;
 };
 
+class OGLSync : private NonCopyable {
+public:
+    OGLSync() = default;
+
+    OGLSync(OGLSync&& o) : handle(std::exchange(o.handle, nullptr)) {}
+
+    ~OGLSync() {
+        Release();
+    }
+    OGLSync& operator=(OGLSync&& o) {
+        Release();
+        handle = std::exchange(o.handle, nullptr);
+        return *this;
+    }
+
+    /// Creates a new internal OpenGL resource and stores the handle
+    void Create() {
+        if (handle != 0)
+            return;
+        handle = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
+    }
+
+    /// Deletes the internal OpenGL resource
+    void Release() {
+        if (handle == 0)
+            return;
+        glDeleteSync(handle);
+        handle = 0;
+    }
+
+    GLsync handle = 0;
+};
+
 class OGLVertexArray : private NonCopyable {
 public:
     OGLVertexArray() = default;
@@ -168,7 +237,7 @@ public:
         if (handle == 0)
             return;
         glDeleteVertexArrays(1, &handle);
-        OpenGLState::ResetVertexArray(handle);
+        OpenGLState::GetCurState().ResetVertexArray(handle).Apply();
         handle = 0;
     }
 
@@ -201,7 +270,7 @@ public:
         if (handle == 0)
             return;
         glDeleteFramebuffers(1, &handle);
-        OpenGLState::ResetFramebuffer(handle);
+        OpenGLState::GetCurState().ResetFramebuffer(handle).Apply();
         handle = 0;
     }
 

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -0,0 +1,58 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <string>
+#include <queue>
+#include "common/assert.h"
+#include "common/common_types.h"
+#include "video_core/renderer_opengl/gl_shader_decompiler.h"
+
+namespace Maxwell3D {
+namespace Shader {
+namespace Decompiler {
+
+constexpr u32 PROGRAM_END = MAX_PROGRAM_CODE_LENGTH;
+
+class Impl {
+public:
+    Impl(const std::array<u32, MAX_PROGRAM_CODE_LENGTH>& program_code,
+         const std::array<u32, MAX_SWIZZLE_DATA_LENGTH>& swizzle_data, u32 main_offset,
+         const std::function<std::string(u32)>& inputreg_getter,
+         const std::function<std::string(u32)>& outputreg_getter, bool sanitize_mul,
+         const std::string& emit_cb, const std::string& setemit_cb)
+        : program_code(program_code), swizzle_data(swizzle_data), main_offset(main_offset),
+          inputreg_getter(inputreg_getter), outputreg_getter(outputreg_getter),
+          sanitize_mul(sanitize_mul), emit_cb(emit_cb), setemit_cb(setemit_cb) {}
+
+    std::string Decompile() {
+        UNREACHABLE();
+        return {};
+    }
+
+private:
+    const std::array<u32, MAX_PROGRAM_CODE_LENGTH>& program_code;
+    const std::array<u32, MAX_SWIZZLE_DATA_LENGTH>& swizzle_data;
+    u32 main_offset;
+    const std::function<std::string(u32)>& inputreg_getter;
+    const std::function<std::string(u32)>& outputreg_getter;
+    bool sanitize_mul;
+    const std::string& emit_cb;
+    const std::string& setemit_cb;
+};
+
+std::string DecompileProgram(const std::array<u32, MAX_PROGRAM_CODE_LENGTH>& program_code,
+                             const std::array<u32, MAX_SWIZZLE_DATA_LENGTH>& swizzle_data,
+                             u32 main_offset,
+                             const std::function<std::string(u32)>& inputreg_getter,
+                             const std::function<std::string(u32)>& outputreg_getter,
+                             bool sanitize_mul, const std::string& emit_cb,
+                             const std::string& setemit_cb) {
+    Impl impl(program_code, swizzle_data, main_offset, inputreg_getter, outputreg_getter,
+              sanitize_mul, emit_cb, setemit_cb);
+    return impl.Decompile();
+}
+
+} // namespace Decompiler
+} // namespace Shader
+} // namespace Maxwell3D

src/video_core/renderer_opengl/gl_shader_decompiler.h

@@ -0,0 +1,27 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <array>
+#include <functional>
+#include <string>
+#include "common/common_types.h"
+
+namespace Maxwell3D {
+namespace Shader {
+namespace Decompiler {
+
+constexpr size_t MAX_PROGRAM_CODE_LENGTH{0x100000};
+constexpr size_t MAX_SWIZZLE_DATA_LENGTH{0x100000};
+
+std::string DecompileProgram(const std::array<u32, MAX_PROGRAM_CODE_LENGTH>& program_code,
+                             const std::array<u32, MAX_SWIZZLE_DATA_LENGTH>& swizzle_data,
+                             u32 main_offset,
+                             const std::function<std::string(u32)>& inputreg_getter,
+                             const std::function<std::string(u32)>& outputreg_getter,
+                             bool sanitize_mul, const std::string& emit_cb = "",
+                             const std::string& setemit_cb = "");
+
+} // namespace Decompiler
+} // namespace Shader
+} // namespace Maxwell3D

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -0,0 +1,20 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/assert.h"
+#include "video_core/renderer_opengl/gl_shader_gen.h"
+
+namespace GLShader {
+
+std::string GenerateVertexShader(const MaxwellVSConfig& config) {
+    UNREACHABLE();
+    return {};
+}
+
+std::string GenerateFragmentShader(const MaxwellFSConfig& config) {
+    UNREACHABLE();
+    return {};
+}
+
+} // namespace GLShader

src/video_core/renderer_opengl/gl_shader_gen.h

@@ -0,0 +1,66 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <cstring>
+#include <string>
+#include <type_traits>
+#include "common/hash.h"
+
+namespace GLShader {
+
+enum Attributes {
+    ATTRIBUTE_POSITION,
+    ATTRIBUTE_COLOR,
+    ATTRIBUTE_TEXCOORD0,
+    ATTRIBUTE_TEXCOORD1,
+    ATTRIBUTE_TEXCOORD2,
+    ATTRIBUTE_TEXCOORD0_W,
+    ATTRIBUTE_NORMQUAT,
+    ATTRIBUTE_VIEW,
+};
+
+struct MaxwellShaderConfigCommon {
+    explicit MaxwellShaderConfigCommon(){};
+};
+
+struct MaxwellVSConfig : MaxwellShaderConfigCommon {
+    explicit MaxwellVSConfig() : MaxwellShaderConfigCommon() {}
+
+    bool operator==(const MaxwellVSConfig& o) const {
+        return std::memcmp(this, &o, sizeof(MaxwellVSConfig)) == 0;
+    };
+};
+
+struct MaxwellFSConfig : MaxwellShaderConfigCommon {
+    explicit MaxwellFSConfig() : MaxwellShaderConfigCommon() {}
+
+    bool operator==(const MaxwellFSConfig& o) const {
+        return std::memcmp(this, &o, sizeof(MaxwellFSConfig)) == 0;
+    };
+};
+
+std::string GenerateVertexShader(const MaxwellVSConfig& config);
+std::string GenerateFragmentShader(const MaxwellFSConfig& config);
+
+} // namespace GLShader
+
+namespace std {
+
+template <>
+struct hash<GLShader::MaxwellVSConfig> {
+    size_t operator()(const GLShader::MaxwellVSConfig& k) const {
+        return Common::ComputeHash64(&k, sizeof(GLShader::MaxwellVSConfig));
+    }
+};
+
+template <>
+struct hash<GLShader::MaxwellFSConfig> {
+    size_t operator()(const GLShader::MaxwellFSConfig& k) const {
+        return Common::ComputeHash64(&k, sizeof(GLShader::MaxwellFSConfig));
+    }
+};
+
+} // namespace std

src/video_core/renderer_opengl/gl_shader_util.cpp

@@ -10,53 +10,85 @@
 
 namespace GLShader {
 
-GLuint LoadProgram(const char* vertex_shader, const char* fragment_shader) {
-
+GLuint LoadProgram(const char* vertex_shader, const char* geometry_shader,
+                   const char* fragment_shader, const std::vector<const char*>& feedback_vars,
+                   bool separable_program) {
     // Create the shaders
-    GLuint vertex_shader_id = glCreateShader(GL_VERTEX_SHADER);
-    GLuint fragment_shader_id = glCreateShader(GL_FRAGMENT_SHADER);
+    GLuint vertex_shader_id = vertex_shader ? glCreateShader(GL_VERTEX_SHADER) : 0;
+    GLuint geometry_shader_id = geometry_shader ? glCreateShader(GL_GEOMETRY_SHADER) : 0;
+    GLuint fragment_shader_id = fragment_shader ? glCreateShader(GL_FRAGMENT_SHADER) : 0;
 
     GLint result = GL_FALSE;
     int info_log_length;
 
-    // Compile Vertex Shader
-    LOG_DEBUG(Render_OpenGL, "Compiling vertex shader...");
+    if (vertex_shader) {
+        // Compile Vertex Shader
+        LOG_DEBUG(Render_OpenGL, "Compiling vertex shader...");
 
-    glShaderSource(vertex_shader_id, 1, &vertex_shader, nullptr);
-    glCompileShader(vertex_shader_id);
+        glShaderSource(vertex_shader_id, 1, &vertex_shader, nullptr);
+        glCompileShader(vertex_shader_id);
 
-    // Check Vertex Shader
-    glGetShaderiv(vertex_shader_id, GL_COMPILE_STATUS, &result);
-    glGetShaderiv(vertex_shader_id, GL_INFO_LOG_LENGTH, &info_log_length);
+        // Check Vertex Shader
+        glGetShaderiv(vertex_shader_id, GL_COMPILE_STATUS, &result);
+        glGetShaderiv(vertex_shader_id, GL_INFO_LOG_LENGTH, &info_log_length);
 
-    if (info_log_length > 1) {
-        std::vector<char> vertex_shader_error(info_log_length);
-        glGetShaderInfoLog(vertex_shader_id, info_log_length, nullptr, &vertex_shader_error[0]);
-        if (result == GL_TRUE) {
-            LOG_DEBUG(Render_OpenGL, "%s", &vertex_shader_error[0]);
-        } else {
-            LOG_ERROR(Render_OpenGL, "Error compiling vertex shader:\n%s", &vertex_shader_error[0]);
+        if (info_log_length > 1) {
+            std::vector<char> vertex_shader_error(info_log_length);
+            glGetShaderInfoLog(vertex_shader_id, info_log_length, nullptr, &vertex_shader_error[0]);
+            if (result == GL_TRUE) {
+                LOG_DEBUG(Render_OpenGL, "%s", &vertex_shader_error[0]);
+            } else {
+                LOG_CRITICAL(Render_OpenGL, "Error compiling vertex shader:\n%s",
+                             &vertex_shader_error[0]);
+            }
         }
     }
 
-    // Compile Fragment Shader
-    LOG_DEBUG(Render_OpenGL, "Compiling fragment shader...");
+    if (geometry_shader) {
+        // Compile Geometry Shader
+        LOG_DEBUG(Render_OpenGL, "Compiling geometry shader...");
 
-    glShaderSource(fragment_shader_id, 1, &fragment_shader, nullptr);
-    glCompileShader(fragment_shader_id);
+        glShaderSource(geometry_shader_id, 1, &geometry_shader, nullptr);
+        glCompileShader(geometry_shader_id);
 
-    // Check Fragment Shader
-    glGetShaderiv(fragment_shader_id, GL_COMPILE_STATUS, &result);
-    glGetShaderiv(fragment_shader_id, GL_INFO_LOG_LENGTH, &info_log_length);
+        // Check Geometry Shader
+        glGetShaderiv(geometry_shader_id, GL_COMPILE_STATUS, &result);
+        glGetShaderiv(geometry_shader_id, GL_INFO_LOG_LENGTH, &info_log_length);
 
-    if (info_log_length > 1) {
-        std::vector<char> fragment_shader_error(info_log_length);
-        glGetShaderInfoLog(fragment_shader_id, info_log_length, nullptr, &fragment_shader_error[0]);
-        if (result == GL_TRUE) {
-            LOG_DEBUG(Render_OpenGL, "%s", &fragment_shader_error[0]);
-        } else {
-            LOG_ERROR(Render_OpenGL, "Error compiling fragment shader:\n%s",
-                      &fragment_shader_error[0]);
+        if (info_log_length > 1) {
+            std::vector<char> geometry_shader_error(info_log_length);
+            glGetShaderInfoLog(geometry_shader_id, info_log_length, nullptr,
+                               &geometry_shader_error[0]);
+            if (result == GL_TRUE) {
+                LOG_DEBUG(Render_OpenGL, "%s", &geometry_shader_error[0]);
+            } else {
+                LOG_CRITICAL(Render_OpenGL, "Error compiling geometry shader:\n%s",
+                             &geometry_shader_error[0]);
+            }
+        }
+    }
+
+    if (fragment_shader) {
+        // Compile Fragment Shader
+        LOG_DEBUG(Render_OpenGL, "Compiling fragment shader...");
+
+        glShaderSource(fragment_shader_id, 1, &fragment_shader, nullptr);
+        glCompileShader(fragment_shader_id);
+
+        // Check Fragment Shader
+        glGetShaderiv(fragment_shader_id, GL_COMPILE_STATUS, &result);
+        glGetShaderiv(fragment_shader_id, GL_INFO_LOG_LENGTH, &info_log_length);
+
+        if (info_log_length > 1) {
+            std::vector<char> fragment_shader_error(info_log_length);
+            glGetShaderInfoLog(fragment_shader_id, info_log_length, nullptr,
+                               &fragment_shader_error[0]);
+            if (result == GL_TRUE) {
+                LOG_DEBUG(Render_OpenGL, "%s", &fragment_shader_error[0]);
+            } else {
+                LOG_CRITICAL(Render_OpenGL, "Error compiling fragment shader:\n%s",
+                             &fragment_shader_error[0]);
+            }
         }
     }
 
@@ -64,8 +96,25 @@ GLuint LoadProgram(const char* vertex_shader, const char* fragment_shader) {
     LOG_DEBUG(Render_OpenGL, "Linking program...");
 
     GLuint program_id = glCreateProgram();
-    glAttachShader(program_id, vertex_shader_id);
-    glAttachShader(program_id, fragment_shader_id);
+    if (vertex_shader) {
+        glAttachShader(program_id, vertex_shader_id);
+    }
+    if (geometry_shader) {
+        glAttachShader(program_id, geometry_shader_id);
+    }
+    if (fragment_shader) {
+        glAttachShader(program_id, fragment_shader_id);
+    }
+
+    if (!feedback_vars.empty()) {
+        auto varyings = feedback_vars;
+        glTransformFeedbackVaryings(program_id, static_cast<GLsizei>(feedback_vars.size()),
+                                    &varyings[0], GL_INTERLEAVED_ATTRIBS);
+    }
+
+    if (separable_program) {
+        glProgramParameteri(program_id, GL_PROGRAM_SEPARABLE, GL_TRUE);
+    }
 
     glLinkProgram(program_id);
 
@@ -79,19 +128,36 @@ GLuint LoadProgram(const char* vertex_shader, const char* fragment_shader) {
         if (result == GL_TRUE) {
             LOG_DEBUG(Render_OpenGL, "%s", &program_error[0]);
         } else {
-            LOG_ERROR(Render_OpenGL, "Error linking shader:\n%s", &program_error[0]);
+            LOG_CRITICAL(Render_OpenGL, "Error linking shader:\n%s", &program_error[0]);
         }
     }
 
     // If the program linking failed at least one of the shaders was probably bad
     if (result == GL_FALSE) {
-        LOG_ERROR(Render_OpenGL, "Vertex shader:\n%s", vertex_shader);
-        LOG_ERROR(Render_OpenGL, "Fragment shader:\n%s", fragment_shader);
+        if (vertex_shader) {
+            LOG_CRITICAL(Render_OpenGL, "Vertex shader:\n%s", vertex_shader);
+        }
+        if (geometry_shader) {
+            LOG_CRITICAL(Render_OpenGL, "Geometry shader:\n%s", geometry_shader);
+        }
+        if (fragment_shader) {
+            LOG_CRITICAL(Render_OpenGL, "Fragment shader:\n%s", fragment_shader);
+        }
     }
     ASSERT_MSG(result == GL_TRUE, "Shader not linked");
 
-    glDeleteShader(vertex_shader_id);
-    glDeleteShader(fragment_shader_id);
+    if (vertex_shader) {
+        glDetachShader(program_id, vertex_shader_id);
+        glDeleteShader(vertex_shader_id);
+    }
+    if (geometry_shader) {
+        glDetachShader(program_id, geometry_shader_id);
+        glDeleteShader(geometry_shader_id);
+    }
+    if (fragment_shader) {
+        glDetachShader(program_id, fragment_shader_id);
+        glDeleteShader(fragment_shader_id);
+    }
 
     return program_id;
 }

src/video_core/renderer_opengl/gl_shader_util.h

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <vector>
 #include <glad/glad.h>
 
 namespace GLShader {
@@ -11,9 +12,12 @@ namespace GLShader {
 /**
  * Utility function to create and compile an OpenGL GLSL shader program (vertex + fragment shader)
  * @param vertex_shader String of the GLSL vertex shader program
+ * @param geometry_shader String of the GLSL geometry shader program
  * @param fragment_shader String of the GLSL fragment shader program
  * @returns Handle of the newly created OpenGL shader object
  */
-GLuint LoadProgram(const char* vertex_shader, const char* fragment_shader);
+GLuint LoadProgram(const char* vertex_shader, const char* geometry_shader,
+                   const char* fragment_shader, const std::vector<const char*>& feedback_vars = {},
+                   bool separable_program = false);
 
 } // namespace GLShader

src/video_core/renderer_opengl/gl_state.cpp

@@ -33,7 +33,7 @@ OpenGLState::OpenGLState() {
     stencil.action_depth_pass = GL_KEEP;
     stencil.action_stencil_fail = GL_KEEP;
 
-    blend.enabled = false;
+    blend.enabled = true;
     blend.rgb_equation = GL_FUNC_ADD;
     blend.a_equation = GL_FUNC_ADD;
     blend.src_rgb_func = GL_ONE;
@@ -68,6 +68,18 @@ OpenGLState::OpenGLState() {
     draw.vertex_buffer = 0;
     draw.uniform_buffer = 0;
     draw.shader_program = 0;
+    draw.program_pipeline = 0;
+
+    scissor.enabled = false;
+    scissor.x = 0;
+    scissor.y = 0;
+    scissor.width = 0;
+    scissor.height = 0;
+
+    viewport.x = 0;
+    viewport.y = 0;
+    viewport.width = 0;
+    viewport.height = 0;
 
     clip_distance = {};
 }
@@ -148,9 +160,6 @@ void OpenGLState::Apply() const {
     if (blend.enabled != cur_state.blend.enabled) {
         if (blend.enabled) {
             glEnable(GL_BLEND);
-
-            cur_state.logic_op = GL_COPY;
-            glLogicOp(cur_state.logic_op);
             glDisable(GL_COLOR_LOGIC_OP);
         } else {
             glDisable(GL_BLEND);
@@ -196,7 +205,7 @@ void OpenGLState::Apply() const {
     // Lighting LUTs
     if (lighting_lut.texture_buffer != cur_state.lighting_lut.texture_buffer) {
         glActiveTexture(TextureUnits::LightingLUT.Enum());
-        glBindTexture(GL_TEXTURE_BUFFER, cur_state.lighting_lut.texture_buffer);
+        glBindTexture(GL_TEXTURE_BUFFER, lighting_lut.texture_buffer);
     }
 
     // Fog LUT
@@ -263,6 +272,31 @@ void OpenGLState::Apply() const {
         glUseProgram(draw.shader_program);
     }
 
+    // Program pipeline
+    if (draw.program_pipeline != cur_state.draw.program_pipeline) {
+        glBindProgramPipeline(draw.program_pipeline);
+    }
+
+    // Scissor test
+    if (scissor.enabled != cur_state.scissor.enabled) {
+        if (scissor.enabled) {
+            glEnable(GL_SCISSOR_TEST);
+        } else {
+            glDisable(GL_SCISSOR_TEST);
+        }
+    }
+
+    if (scissor.x != cur_state.scissor.x || scissor.y != cur_state.scissor.y ||
+        scissor.width != cur_state.scissor.width || scissor.height != cur_state.scissor.height) {
+        glScissor(scissor.x, scissor.y, scissor.width, scissor.height);
+    }
+
+    if (viewport.x != cur_state.viewport.x || viewport.y != cur_state.viewport.y ||
+        viewport.width != cur_state.viewport.width ||
+        viewport.height != cur_state.viewport.height) {
+        glViewport(viewport.x, viewport.y, viewport.width, viewport.height);
+    }
+
     // Clip distance
     for (size_t i = 0; i < clip_distance.size(); ++i) {
         if (clip_distance[i] != cur_state.clip_distance[i]) {
@@ -277,62 +311,75 @@ void OpenGLState::Apply() const {
     cur_state = *this;
 }
 
-void OpenGLState::ResetTexture(GLuint handle) {
-    for (auto& unit : cur_state.texture_units) {
+OpenGLState& OpenGLState::ResetTexture(GLuint handle) {
+    for (auto& unit : texture_units) {
         if (unit.texture_2d == handle) {
             unit.texture_2d = 0;
         }
     }
-    if (cur_state.lighting_lut.texture_buffer == handle)
-        cur_state.lighting_lut.texture_buffer = 0;
-    if (cur_state.fog_lut.texture_buffer == handle)
-        cur_state.fog_lut.texture_buffer = 0;
-    if (cur_state.proctex_noise_lut.texture_buffer == handle)
-        cur_state.proctex_noise_lut.texture_buffer = 0;
-    if (cur_state.proctex_color_map.texture_buffer == handle)
-        cur_state.proctex_color_map.texture_buffer = 0;
-    if (cur_state.proctex_alpha_map.texture_buffer == handle)
-        cur_state.proctex_alpha_map.texture_buffer = 0;
-    if (cur_state.proctex_lut.texture_buffer == handle)
-        cur_state.proctex_lut.texture_buffer = 0;
-    if (cur_state.proctex_diff_lut.texture_buffer == handle)
-        cur_state.proctex_diff_lut.texture_buffer = 0;
+    if (lighting_lut.texture_buffer == handle)
+        lighting_lut.texture_buffer = 0;
+    if (fog_lut.texture_buffer == handle)
+        fog_lut.texture_buffer = 0;
+    if (proctex_noise_lut.texture_buffer == handle)
+        proctex_noise_lut.texture_buffer = 0;
+    if (proctex_color_map.texture_buffer == handle)
+        proctex_color_map.texture_buffer = 0;
+    if (proctex_alpha_map.texture_buffer == handle)
+        proctex_alpha_map.texture_buffer = 0;
+    if (proctex_lut.texture_buffer == handle)
+        proctex_lut.texture_buffer = 0;
+    if (proctex_diff_lut.texture_buffer == handle)
+        proctex_diff_lut.texture_buffer = 0;
+    return *this;
 }
 
-void OpenGLState::ResetSampler(GLuint handle) {
-    for (auto& unit : cur_state.texture_units) {
+OpenGLState& OpenGLState::ResetSampler(GLuint handle) {
+    for (auto& unit : texture_units) {
         if (unit.sampler == handle) {
             unit.sampler = 0;
         }
     }
+    return *this;
 }
 
-void OpenGLState::ResetProgram(GLuint handle) {
-    if (cur_state.draw.shader_program == handle) {
-        cur_state.draw.shader_program = 0;
+OpenGLState& OpenGLState::ResetProgram(GLuint handle) {
+    if (draw.shader_program == handle) {
+        draw.shader_program = 0;
     }
+    return *this;
 }
 
-void OpenGLState::ResetBuffer(GLuint handle) {
-    if (cur_state.draw.vertex_buffer == handle) {
-        cur_state.draw.vertex_buffer = 0;
-    }
-    if (cur_state.draw.uniform_buffer == handle) {
-        cur_state.draw.uniform_buffer = 0;
+OpenGLState& OpenGLState::ResetPipeline(GLuint handle) {
+    if (draw.program_pipeline == handle) {
+        draw.program_pipeline = 0;
     }
+    return *this;
 }
 
-void OpenGLState::ResetVertexArray(GLuint handle) {
-    if (cur_state.draw.vertex_array == handle) {
-        cur_state.draw.vertex_array = 0;
+OpenGLState& OpenGLState::ResetBuffer(GLuint handle) {
+    if (draw.vertex_buffer == handle) {
+        draw.vertex_buffer = 0;
     }
+    if (draw.uniform_buffer == handle) {
+        draw.uniform_buffer = 0;
+    }
+    return *this;
 }
 
-void OpenGLState::ResetFramebuffer(GLuint handle) {
-    if (cur_state.draw.read_framebuffer == handle) {
-        cur_state.draw.read_framebuffer = 0;
-    }
-    if (cur_state.draw.draw_framebuffer == handle) {
-        cur_state.draw.draw_framebuffer = 0;
+OpenGLState& OpenGLState::ResetVertexArray(GLuint handle) {
+    if (draw.vertex_array == handle) {
+        draw.vertex_array = 0;
     }
+    return *this;
+}
+
+OpenGLState& OpenGLState::ResetFramebuffer(GLuint handle) {
+    if (draw.read_framebuffer == handle) {
+        draw.read_framebuffer = 0;
+    }
+    if (draw.draw_framebuffer == handle) {
+        draw.draw_framebuffer = 0;
+    }
+    return *this;
 }

src/video_core/renderer_opengl/gl_state.h

@@ -85,7 +85,7 @@ public:
     struct {
         GLuint texture_2d; // GL_TEXTURE_BINDING_2D
         GLuint sampler;    // GL_SAMPLER_BINDING
-    } texture_units[3];
+    } texture_units[32];
 
     struct {
         GLuint texture_buffer; // GL_TEXTURE_BINDING_BUFFER
@@ -122,27 +122,44 @@ public:
         GLuint vertex_buffer;    // GL_ARRAY_BUFFER_BINDING
         GLuint uniform_buffer;   // GL_UNIFORM_BUFFER_BINDING
         GLuint shader_program;   // GL_CURRENT_PROGRAM
+        GLuint program_pipeline; // GL_PROGRAM_PIPELINE_BINDING
     } draw;
 
+    struct {
+        bool enabled; // GL_SCISSOR_TEST
+        GLint x;
+        GLint y;
+        GLsizei width;
+        GLsizei height;
+    } scissor;
+
+    struct {
+        GLint x;
+        GLint y;
+        GLsizei width;
+        GLsizei height;
+    } viewport;
+
     std::array<bool, 2> clip_distance; // GL_CLIP_DISTANCE
 
     OpenGLState();
 
     /// Get the currently active OpenGL state
-    static const OpenGLState& GetCurState() {
+    static OpenGLState GetCurState() {
         return cur_state;
     }
 
     /// Apply this state as the current OpenGL state
     void Apply() const;
 
-    /// Resets and unbinds any references to the given resource in the current OpenGL state
-    static void ResetTexture(GLuint handle);
-    static void ResetSampler(GLuint handle);
-    static void ResetProgram(GLuint handle);
-    static void ResetBuffer(GLuint handle);
-    static void ResetVertexArray(GLuint handle);
-    static void ResetFramebuffer(GLuint handle);
+    /// Resets any references to the given resource
+    OpenGLState& ResetTexture(GLuint handle);
+    OpenGLState& ResetSampler(GLuint handle);
+    OpenGLState& ResetProgram(GLuint handle);
+    OpenGLState& ResetPipeline(GLuint handle);
+    OpenGLState& ResetBuffer(GLuint handle);
+    OpenGLState& ResetVertexArray(GLuint handle);
+    OpenGLState& ResetFramebuffer(GLuint handle);
 
 private:
     static OpenGLState cur_state;

src/video_core/renderer_opengl/gl_stream_buffer.cpp

@@ -0,0 +1,182 @@
+// Copyright 2018 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <deque>
+#include <vector>
+#include "common/alignment.h"
+#include "common/assert.h"
+#include "video_core/renderer_opengl/gl_state.h"
+#include "video_core/renderer_opengl/gl_stream_buffer.h"
+
+class OrphanBuffer : public OGLStreamBuffer {
+public:
+    explicit OrphanBuffer(GLenum target) : OGLStreamBuffer(target) {}
+    ~OrphanBuffer() override;
+
+private:
+    void Create(size_t size, size_t sync_subdivide) override;
+    void Release() override;
+
+    std::pair<u8*, GLintptr> Map(size_t size, size_t alignment) override;
+    void Unmap() override;
+
+    std::vector<u8> data;
+};
+
+class StorageBuffer : public OGLStreamBuffer {
+public:
+    explicit StorageBuffer(GLenum target) : OGLStreamBuffer(target) {}
+    ~StorageBuffer() override;
+
+private:
+    void Create(size_t size, size_t sync_subdivide) override;
+    void Release() override;
+
+    std::pair<u8*, GLintptr> Map(size_t size, size_t alignment) override;
+    void Unmap() override;
+
+    struct Fence {
+        OGLSync sync;
+        size_t offset;
+    };
+    std::deque<Fence> head;
+    std::deque<Fence> tail;
+
+    u8* mapped_ptr;
+};
+
+OGLStreamBuffer::OGLStreamBuffer(GLenum target) {
+    gl_target = target;
+}
+
+GLuint OGLStreamBuffer::GetHandle() const {
+    return gl_buffer.handle;
+}
+
+std::unique_ptr<OGLStreamBuffer> OGLStreamBuffer::MakeBuffer(bool storage_buffer, GLenum target) {
+    if (storage_buffer) {
+        return std::make_unique<StorageBuffer>(target);
+    }
+    return std::make_unique<OrphanBuffer>(target);
+}
+
+OrphanBuffer::~OrphanBuffer() {
+    Release();
+}
+
+void OrphanBuffer::Create(size_t size, size_t /*sync_subdivide*/) {
+    buffer_pos = 0;
+    buffer_size = size;
+    data.resize(buffer_size);
+
+    if (gl_buffer.handle == 0) {
+        gl_buffer.Create();
+        glBindBuffer(gl_target, gl_buffer.handle);
+    }
+
+    glBufferData(gl_target, static_cast<GLsizeiptr>(buffer_size), nullptr, GL_STREAM_DRAW);
+}
+
+void OrphanBuffer::Release() {
+    gl_buffer.Release();
+}
+
+std::pair<u8*, GLintptr> OrphanBuffer::Map(size_t size, size_t alignment) {
+    buffer_pos = Common::AlignUp(buffer_pos, alignment);
+
+    if (buffer_pos + size > buffer_size) {
+        Create(std::max(buffer_size, size), 0);
+    }
+
+    mapped_size = size;
+    return std::make_pair(&data[buffer_pos], static_cast<GLintptr>(buffer_pos));
+}
+
+void OrphanBuffer::Unmap() {
+    glBufferSubData(gl_target, static_cast<GLintptr>(buffer_pos),
+                    static_cast<GLsizeiptr>(mapped_size), &data[buffer_pos]);
+    buffer_pos += mapped_size;
+}
+
+StorageBuffer::~StorageBuffer() {
+    Release();
+}
+
+void StorageBuffer::Create(size_t size, size_t sync_subdivide) {
+    if (gl_buffer.handle != 0)
+        return;
+
+    buffer_pos = 0;
+    buffer_size = size;
+    buffer_sync_subdivide = std::max<size_t>(sync_subdivide, 1);
+
+    gl_buffer.Create();
+    glBindBuffer(gl_target, gl_buffer.handle);
+
+    glBufferStorage(gl_target, static_cast<GLsizeiptr>(buffer_size), nullptr,
+                    GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT);
+    mapped_ptr = reinterpret_cast<u8*>(
+        glMapBufferRange(gl_target, 0, static_cast<GLsizeiptr>(buffer_size),
+                         GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | GL_MAP_FLUSH_EXPLICIT_BIT));
+}
+
+void StorageBuffer::Release() {
+    if (gl_buffer.handle == 0)
+        return;
+
+    glUnmapBuffer(gl_target);
+
+    gl_buffer.Release();
+    head.clear();
+    tail.clear();
+}
+
+std::pair<u8*, GLintptr> StorageBuffer::Map(size_t size, size_t alignment) {
+    ASSERT(size <= buffer_size);
+
+    OGLSync sync;
+
+    buffer_pos = Common::AlignUp(buffer_pos, alignment);
+    size_t effective_offset = Common::AlignDown(buffer_pos, buffer_sync_subdivide);
+
+    if (!head.empty() &&
+        (effective_offset > head.back().offset || buffer_pos + size > buffer_size)) {
+        ASSERT(head.back().sync.handle == 0);
+        head.back().sync.Create();
+    }
+
+    if (buffer_pos + size > buffer_size) {
+        if (!tail.empty()) {
+            std::swap(sync, tail.back().sync);
+            tail.clear();
+        }
+        std::swap(tail, head);
+        buffer_pos = 0;
+        effective_offset = 0;
+    }
+
+    while (!tail.empty() && buffer_pos + size > tail.front().offset) {
+        std::swap(sync, tail.front().sync);
+        tail.pop_front();
+    }
+
+    if (sync.handle != 0) {
+        glClientWaitSync(sync.handle, GL_SYNC_FLUSH_COMMANDS_BIT, GL_TIMEOUT_IGNORED);
+        sync.Release();
+    }
+
+    if (head.empty() || effective_offset > head.back().offset) {
+        head.emplace_back();
+        head.back().offset = effective_offset;
+    }
+
+    mapped_size = size;
+    return std::make_pair(&mapped_ptr[buffer_pos], static_cast<GLintptr>(buffer_pos));
+}
+
+void StorageBuffer::Unmap() {
+    glFlushMappedBufferRange(gl_target, static_cast<GLintptr>(buffer_pos),
+                             static_cast<GLsizeiptr>(mapped_size));
+    buffer_pos += mapped_size;
+}

src/video_core/renderer_opengl/gl_stream_buffer.h

@@ -0,0 +1,34 @@
+// Copyright 2018 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <memory>
+#include <glad/glad.h>
+#include "common/common_types.h"
+#include "video_core/renderer_opengl/gl_resource_manager.h"
+
+class OGLStreamBuffer : private NonCopyable {
+public:
+    explicit OGLStreamBuffer(GLenum target);
+    virtual ~OGLStreamBuffer() = default;
+
+public:
+    static std::unique_ptr<OGLStreamBuffer> MakeBuffer(bool storage_buffer, GLenum target);
+
+    virtual void Create(size_t size, size_t sync_subdivide) = 0;
+    virtual void Release() {}
+
+    GLuint GetHandle() const;
+
+    virtual std::pair<u8*, GLintptr> Map(size_t size, size_t alignment) = 0;
+    virtual void Unmap() = 0;
+
+protected:
+    OGLBuffer gl_buffer;
+    GLenum gl_target;
+
+    size_t buffer_pos = 0;
+    size_t buffer_size = 0;
+    size_t buffer_sync_subdivide = 0;
+    size_t mapped_size = 0;
+};

src/video_core/renderer_opengl/maxwell_to_gl.h

@@ -0,0 +1,50 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <glad/glad.h>
+#include "common/common_types.h"
+#include "common/logging/log.h"
+#include "video_core/engines/maxwell_3d.h"
+
+namespace MaxwellToGL {
+
+using Maxwell = Tegra::Engines::Maxwell3D::Regs;
+
+inline GLenum VertexType(Maxwell::VertexAttribute attrib) {
+    switch (attrib.type) {
+    case Maxwell::VertexAttribute::Type::UnsignedNorm: {
+
+        switch (attrib.size) {
+        case Maxwell::VertexAttribute::Size::Size_8_8_8_8:
+            return GL_UNSIGNED_BYTE;
+        }
+
+        LOG_CRITICAL(Render_OpenGL, "Unimplemented vertex size=%s", attrib.SizeString().c_str());
+        UNREACHABLE();
+        return {};
+    }
+
+    case Maxwell::VertexAttribute::Type::Float:
+        return GL_FLOAT;
+    }
+
+    LOG_CRITICAL(Render_OpenGL, "Unimplemented vertex type=%s", attrib.TypeString().c_str());
+    UNREACHABLE();
+    return {};
+}
+
+inline GLenum PrimitiveTopology(Maxwell::PrimitiveTopology topology) {
+    switch (topology) {
+    case Maxwell::PrimitiveTopology::TriangleStrip:
+        return GL_TRIANGLE_STRIP;
+    }
+    LOG_CRITICAL(Render_OpenGL, "Unimplemented primitive topology=%d", topology);
+    UNREACHABLE();
+    return {};
+}
+
+} // namespace MaxwellToGL