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
+}

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

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/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

@@ -49,6 +49,7 @@ namespace Log {
     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

@@ -66,6 +66,7 @@ enum class Class : ClassType {
     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
@@ -84,7 +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/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
@@ -187,8 +189,10 @@ 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
@@ -199,6 +203,8 @@ add_library(core STATIC
     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/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

@@ -35,6 +35,7 @@
 #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"
 
@@ -192,6 +193,7 @@ void Init() {
     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/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/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"
@@ -41,6 +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);
@@ -112,91 +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;
 }
 
@@ -214,7 +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;
 }
 
@@ -288,6 +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);
 }
@@ -304,17 +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;
@@ -324,21 +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]);
@@ -347,6 +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();
         }
@@ -378,17 +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;
@@ -397,20 +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]);
 
@@ -418,6 +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();
         }
@@ -433,9 +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;
@@ -443,27 +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();
         }
@@ -474,37 +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:
@@ -519,78 +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;

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,
 };
 
@@ -242,4 +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
@@ -29,8 +31,12 @@ add_library(video_core STATIC
     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 {
@@ -20,18 +24,13 @@ 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;
@@ -62,6 +61,192 @@ 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(0x200);
@@ -69,9 +254,9 @@ public:
                 struct {
                     u32 address_high;
                     u32 address_low;
-                    u32 horiz;
-                    u32 vert;
-                    u32 format;
+                    u32 width;
+                    u32 height;
+                    Tegra::RenderTargetFormat format;
                     u32 block_dimensions;
                     u32 array_mode;
                     u32 layer_stride;
@@ -84,7 +269,31 @@ public:
                     }
                 } rt[NumRenderTargets];
 
-                INSERT_PADDING_WORDS(0xDD);
+                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;
@@ -108,14 +317,7 @@ public:
 
                 INSERT_PADDING_WORDS(0x5B);
 
-                union {
-                    BitField<0, 5, u32> buffer;
-                    BitField<6, 1, u32> constant;
-                    BitField<7, 14, u32> offset;
-                    BitField<21, 6, u32> size;
-                    BitField<27, 3, u32> type;
-                    BitField<31, 1, u32> bgra;
-                } vertex_attrib_format[NumVertexAttributes];
+                VertexAttribute vertex_attrib_format[NumVertexAttributes];
 
                 INSERT_PADDING_WORDS(0xF);
 
@@ -163,13 +365,15 @@ public:
                     }
                 } code_address;
                 INSERT_PADDING_WORDS(1);
+
                 struct {
                     u32 vertex_end_gl;
                     union {
                         u32 vertex_begin_gl;
-                        BitField<0, 16, u32> topology;
+                        BitField<0, 16, PrimitiveTopology> topology;
                     };
                 } draw;
+
                 INSERT_PADDING_WORDS(0x139);
                 struct {
                     u32 query_address_high;
@@ -310,6 +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;
 
@@ -320,6 +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
@@ -358,6 +577,7 @@ private:
                   "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);

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

@@ -5,6 +5,7 @@
 #pragma once
 
 #include "common/common_types.h"
+#include "video_core/gpu.h"
 
 struct ScreenInfo;
 
@@ -14,8 +15,8 @@ class RasterizerInterface {
 public:
     virtual ~RasterizerInterface() {}
 
-    /// Draw the current batch of triangles
-    virtual void DrawTriangles() = 0;
+    /// 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;
@@ -24,14 +25,14 @@ public:
     virtual void FlushAll() = 0;
 
     /// Notify rasterizer that any caches of the specified region should be flushed to 3DS memory
-    virtual void FlushRegion(PAddr addr, u32 size) = 0;
+    virtual void FlushRegion(VAddr addr, u64 size) = 0;
 
     /// Notify rasterizer that any caches of the specified region should be invalidated
-    virtual void InvalidateRegion(PAddr addr, u32 size) = 0;
+    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(PAddr addr, u32 size) = 0;
+    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) {
@@ -49,7 +50,8 @@ public:
     }
 
     /// Attempt to use a faster method to display the framebuffer to screen
-    virtual bool AccelerateDisplay(const void* config, PAddr framebuffer_addr, u32 pixel_stride,
+    virtual bool AccelerateDisplay(const Tegra::FramebufferConfig& framebuffer,
+                                   VAddr framebuffer_addr, u32 pixel_stride,
                                    ScreenInfo& screen_info) {
         return false;
     }

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

@@ -14,11 +14,16 @@
 #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;
 
@@ -54,6 +59,8 @@ static void SetShaderUniformBlockBindings(GLuint shader) {
 }
 
 RasterizerOpenGL::RasterizerOpenGL() {
+    shader_dirty = true;
+
     has_ARB_buffer_storage = false;
     has_ARB_direct_state_access = false;
     has_ARB_separate_shader_objects = false;
@@ -106,8 +113,6 @@ RasterizerOpenGL::RasterizerOpenGL() {
         state.draw.vertex_buffer = stream_buffer->GetHandle();
 
         pipeline.Create();
-        vs_input_index_min = 0;
-        vs_input_index_max = 0;
         state.draw.program_pipeline = pipeline.handle;
         state.draw.shader_program = 0;
         state.draw.vertex_array = hw_vao.handle;
@@ -120,20 +125,14 @@ RasterizerOpenGL::RasterizerOpenGL() {
         glBufferData(GL_UNIFORM_BUFFER, sizeof(VSUniformData), nullptr, GL_STREAM_COPY);
         glBindBufferBase(GL_UNIFORM_BUFFER, 1, vs_uniform_buffer.handle);
     } else {
-        UNIMPLEMENTED();
+        UNREACHABLE();
     }
 
     accelerate_draw = AccelDraw::Disabled;
 
     glEnable(GL_BLEND);
 
-    // Sync fixed function OpenGL state
-    SyncClipEnabled();
-    SyncClipCoef();
-    SyncCullMode();
-    SyncBlendEnabled();
-    SyncBlendFuncs();
-    SyncBlendColor();
+    LOG_CRITICAL(Render_OpenGL, "Sync fixed function OpenGL state here!");
 }
 
 RasterizerOpenGL::~RasterizerOpenGL() {
@@ -144,47 +143,235 @@ RasterizerOpenGL::~RasterizerOpenGL() {
     }
 }
 
-static constexpr std::array<GLenum, 4> vs_attrib_types{
-    GL_BYTE,          // VertexAttributeFormat::BYTE
-    GL_UNSIGNED_BYTE, // VertexAttributeFormat::UBYTE
-    GL_SHORT,         // VertexAttributeFormat::SHORT
-    GL_FLOAT          // VertexAttributeFormat::FLOAT
-};
-
 void RasterizerOpenGL::AnalyzeVertexArray(bool is_indexed) {
-    UNIMPLEMENTED();
+    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);
-    UNIMPLEMENTED();
+    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);
-    UNIMPLEMENTED();
+    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);
-    UNIMPLEMENTED();
+    UNREACHABLE();
 }
 
 bool RasterizerOpenGL::AccelerateDrawBatch(bool is_indexed) {
     if (!has_ARB_separate_shader_objects) {
-        UNIMPLEMENTED();
+        UNREACHABLE();
         return false;
     }
 
     accelerate_draw = is_indexed ? AccelDraw::Indexed : AccelDraw::Arrays;
-    DrawTriangles();
+    DrawArrays();
 
     return true;
 }
 
-void RasterizerOpenGL::DrawTriangles() {
+void RasterizerOpenGL::DrawArrays() {
+    if (accelerate_draw == AccelDraw::Disabled)
+        return;
+
     MICROPROFILE_SCOPE(OpenGL_Drawing);
-    UNIMPLEMENTED();
+    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) {}
@@ -194,17 +381,17 @@ void RasterizerOpenGL::FlushAll() {
     res_cache.FlushAll();
 }
 
-void RasterizerOpenGL::FlushRegion(PAddr addr, u32 size) {
+void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
     res_cache.FlushRegion(addr, size);
 }
 
-void RasterizerOpenGL::InvalidateRegion(PAddr addr, u32 size) {
+void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
     res_cache.InvalidateRegion(addr, size, nullptr);
 }
 
-void RasterizerOpenGL::FlushAndInvalidateRegion(PAddr addr, u32 size) {
+void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
     res_cache.FlushRegion(addr, size);
     res_cache.InvalidateRegion(addr, size, nullptr);
@@ -212,58 +399,180 @@ void RasterizerOpenGL::FlushAndInvalidateRegion(PAddr addr, u32 size) {
 
 bool RasterizerOpenGL::AccelerateDisplayTransfer(const void* config) {
     MICROPROFILE_SCOPE(OpenGL_Blits);
-    UNIMPLEMENTED();
+    UNREACHABLE();
     return true;
 }
 
 bool RasterizerOpenGL::AccelerateTextureCopy(const void* config) {
-    UNIMPLEMENTED();
+    UNREACHABLE();
     return true;
 }
 
 bool RasterizerOpenGL::AccelerateFill(const void* config) {
-    UNIMPLEMENTED();
+    UNREACHABLE();
     return true;
 }
 
-bool RasterizerOpenGL::AccelerateDisplay(const void* config, PAddr framebuffer_addr,
-                                         u32 pixel_stride, ScreenInfo& screen_info) {
-    UNIMPLEMENTED();
+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() {
-    UNIMPLEMENTED();
+    // 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() {
-    UNIMPLEMENTED();
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncClipCoef() {
-    UNIMPLEMENTED();
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncCullMode() {
-    UNIMPLEMENTED();
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncDepthScale() {
-    UNIMPLEMENTED();
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncDepthOffset() {
-    UNIMPLEMENTED();
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncBlendEnabled() {
-    UNIMPLEMENTED();
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncBlendFuncs() {
-    UNIMPLEMENTED();
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncBlendColor() {
-    UNIMPLEMENTED();
+    UNREACHABLE();
 }

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -29,19 +29,25 @@ public:
     RasterizerOpenGL();
     ~RasterizerOpenGL() override;
 
-    void DrawTriangles() override;
+    void DrawArrays() override;
     void NotifyMaxwellRegisterChanged(u32 id) override;
     void FlushAll() override;
-    void FlushRegion(PAddr addr, u32 size) override;
-    void InvalidateRegion(PAddr addr, u32 size) override;
-    void FlushAndInvalidateRegion(PAddr addr, u32 size) 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 void* config, PAddr framebuffer_addr, u32 pixel_stride,
-                           ScreenInfo& screen_info) 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;
     };
@@ -81,6 +87,13 @@ public:
 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();
 
@@ -117,6 +130,12 @@ private:
 
     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;
@@ -127,7 +146,7 @@ private:
     OGLVertexArray hw_vao;
     std::array<bool, 16> hw_vao_enabled_attributes;
 
-    std::array<SamplerInfo, 3> texture_samplers;
+    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;
@@ -136,8 +155,6 @@ private:
     static constexpr size_t STREAM_BUFFER_SIZE = 4 * 1024 * 1024;
     std::unique_ptr<OGLStreamBuffer> stream_buffer;
 
-    GLint vs_input_index_min;
-    GLint vs_input_index_max;
     GLsizeiptr vs_input_size;
 
     void AnalyzeVertexArray(bool is_indexed);

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -21,9 +21,13 @@
 #include "common/microprofile.h"
 #include "common/scope_exit.h"
 #include "common/vector_math.h"
+#include "core/core.h"
 #include "core/frontend/emu_window.h"
+#include "core/hle/kernel/process.h"
+#include "core/hle/kernel/vm_manager.h"
 #include "core/memory.h"
 #include "core/settings.h"
+#include "video_core/engines/maxwell_3d.h"
 #include "video_core/renderer_opengl/gl_rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_state.h"
 #include "video_core/utils.h"
@@ -107,67 +111,28 @@ static void MortonCopyTile(u32 stride, u8* tile_buffer, u8* gl_buffer) {
 }
 
 template <bool morton_to_gl, PixelFormat format>
-static void MortonCopy(u32 stride, u32 height, u8* gl_buffer, PAddr base, PAddr start, PAddr end) {
+static void MortonCopy(u32 stride, u32 height, u8* gl_buffer, VAddr base, VAddr start, VAddr end) {
     constexpr u32 bytes_per_pixel = SurfaceParams::GetFormatBpp(format) / 8;
-    constexpr u32 tile_size = bytes_per_pixel * 64;
-
     constexpr u32 gl_bytes_per_pixel = CachedSurface::GetGLBytesPerPixel(format);
-    static_assert(gl_bytes_per_pixel >= bytes_per_pixel, "");
-    gl_buffer += gl_bytes_per_pixel - bytes_per_pixel;
 
-    const PAddr aligned_down_start = base + Common::AlignDown(start - base, tile_size);
-    const PAddr aligned_start = base + Common::AlignUp(start - base, tile_size);
-    const PAddr aligned_end = base + Common::AlignDown(end - base, tile_size);
-
-    ASSERT(!morton_to_gl || (aligned_start == start && aligned_end == end));
-
-    const u64 begin_pixel_index = (aligned_down_start - base) / bytes_per_pixel;
-    u32 x = static_cast<u32>((begin_pixel_index % (stride * 8)) / 8);
-    u32 y = static_cast<u32>((begin_pixel_index / (stride * 8)) * 8);
-
-    gl_buffer += ((height - 8 - y) * stride + x) * gl_bytes_per_pixel;
-
-    auto glbuf_next_tile = [&] {
-        x = (x + 8) % stride;
-        gl_buffer += 8 * gl_bytes_per_pixel;
-        if (!x) {
-            y += 8;
-            gl_buffer -= stride * 9 * gl_bytes_per_pixel;
-        }
-    };
-
-    u8* tile_buffer = Memory::GetPhysicalPointer(start);
-
-    if (start < aligned_start && !morton_to_gl) {
-        std::array<u8, tile_size> tmp_buf;
-        MortonCopyTile<morton_to_gl, format>(stride, &tmp_buf[0], gl_buffer);
-        std::memcpy(tile_buffer, &tmp_buf[start - aligned_down_start],
-                    std::min(aligned_start, end) - start);
-
-        tile_buffer += aligned_start - start;
-        glbuf_next_tile();
-    }
-
-    const u8* const buffer_end = tile_buffer + aligned_end - aligned_start;
-    while (tile_buffer < buffer_end) {
-        MortonCopyTile<morton_to_gl, format>(stride, tile_buffer, gl_buffer);
-        tile_buffer += tile_size;
-        glbuf_next_tile();
-    }
-
-    if (end > std::max(aligned_start, aligned_end) && !morton_to_gl) {
-        std::array<u8, tile_size> tmp_buf;
-        MortonCopyTile<morton_to_gl, format>(stride, &tmp_buf[0], gl_buffer);
-        std::memcpy(tile_buffer, &tmp_buf[0], end - aligned_end);
-    }
+    // TODO(bunnei): Assumes the default rendering GOB size of 16 (128 lines). We should check the
+    // configuration for this and perform more generic un/swizzle
+    LOG_WARNING(Render_OpenGL, "need to use correct swizzle/GOB parameters!");
+    VideoCore::MortonCopyPixels128(stride, height, bytes_per_pixel, gl_bytes_per_pixel,
+                                   Memory::GetPointer(base), gl_buffer, morton_to_gl);
 }
 
-static constexpr std::array<void (*)(u32, u32, u8*, PAddr, PAddr, PAddr), 18> morton_to_gl_fns = {
-    MortonCopy<true, PixelFormat::RGBA8>,  // 0
-    MortonCopy<true, PixelFormat::RGB8>,   // 1
-    MortonCopy<true, PixelFormat::RGB5A1>, // 2
-    MortonCopy<true, PixelFormat::RGB565>, // 3
-    MortonCopy<true, PixelFormat::RGBA4>,  // 4
+static constexpr std::array<void (*)(u32, u32, u8*, VAddr, VAddr, VAddr), 18> morton_to_gl_fns = {
+    MortonCopy<true, PixelFormat::RGBA8>,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
     nullptr,
     nullptr,
     nullptr,
@@ -176,19 +141,19 @@ static constexpr std::array<void (*)(u32, u32, u8*, PAddr, PAddr, PAddr), 18> mo
     nullptr,
     nullptr,
     nullptr,
-    nullptr,                             // 5 - 13
-    MortonCopy<true, PixelFormat::D16>,  // 14
-    nullptr,                             // 15
-    MortonCopy<true, PixelFormat::D24>,  // 16
-    MortonCopy<true, PixelFormat::D24S8> // 17
 };
 
-static constexpr std::array<void (*)(u32, u32, u8*, PAddr, PAddr, PAddr), 18> gl_to_morton_fns = {
-    MortonCopy<false, PixelFormat::RGBA8>,  // 0
-    MortonCopy<false, PixelFormat::RGB8>,   // 1
-    MortonCopy<false, PixelFormat::RGB5A1>, // 2
-    MortonCopy<false, PixelFormat::RGB565>, // 3
-    MortonCopy<false, PixelFormat::RGBA4>,  // 4
+static constexpr std::array<void (*)(u32, u32, u8*, VAddr, VAddr, VAddr), 18> gl_to_morton_fns = {
+    MortonCopy<false, PixelFormat::RGBA8>,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
     nullptr,
     nullptr,
     nullptr,
@@ -197,11 +162,6 @@ static constexpr std::array<void (*)(u32, u32, u8*, PAddr, PAddr, PAddr), 18> gl
     nullptr,
     nullptr,
     nullptr,
-    nullptr,                              // 5 - 13
-    MortonCopy<false, PixelFormat::D16>,  // 14
-    nullptr,                              // 15
-    MortonCopy<false, PixelFormat::D24>,  // 16
-    MortonCopy<false, PixelFormat::D24S8> // 17
 };
 
 // Allocate an uninitialized texture of appropriate size and format for the surface
@@ -290,17 +250,17 @@ static bool BlitTextures(GLuint src_tex, const MathUtil::Rectangle<u32>& src_rec
 
 static bool FillSurface(const Surface& surface, const u8* fill_data,
                         const MathUtil::Rectangle<u32>& fill_rect, GLuint draw_fb_handle) {
-    UNIMPLEMENTED();
-    return true;
+    UNREACHABLE();
+    return {};
 }
 
 SurfaceParams SurfaceParams::FromInterval(SurfaceInterval interval) const {
     SurfaceParams params = *this;
     const u32 tiled_size = is_tiled ? 8 : 1;
     const u64 stride_tiled_bytes = BytesInPixels(stride * tiled_size);
-    PAddr aligned_start =
+    VAddr aligned_start =
         addr + Common::AlignDown(boost::icl::first(interval) - addr, stride_tiled_bytes);
-    PAddr aligned_end =
+    VAddr aligned_end =
         addr + Common::AlignUp(boost::icl::last_next(interval) - addr, stride_tiled_bytes);
 
     if (aligned_end - aligned_start > stride_tiled_bytes) {
@@ -527,10 +487,10 @@ void RasterizerCacheOpenGL::CopySurface(const Surface& src_surface, const Surfac
 }
 
 MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64, 192));
-void CachedSurface::LoadGLBuffer(PAddr load_start, PAddr load_end) {
+void CachedSurface::LoadGLBuffer(VAddr load_start, VAddr load_end) {
     ASSERT(type != SurfaceType::Fill);
 
-    const u8* const texture_src_data = Memory::GetPhysicalPointer(addr);
+    u8* const texture_src_data = Memory::GetPointer(addr);
     if (texture_src_data == nullptr)
         return;
 
@@ -539,35 +499,30 @@ void CachedSurface::LoadGLBuffer(PAddr load_start, PAddr load_end) {
         gl_buffer.reset(new u8[gl_buffer_size]);
     }
 
-    // TODO: Should probably be done in ::Memory:: and check for other regions too
-    if (load_start < Memory::VRAM_VADDR_END && load_end > Memory::VRAM_VADDR_END)
-        load_end = Memory::VRAM_VADDR_END;
-
-    if (load_start < Memory::VRAM_VADDR && load_end > Memory::VRAM_VADDR)
-        load_start = Memory::VRAM_VADDR;
-
     MICROPROFILE_SCOPE(OpenGL_SurfaceLoad);
 
     ASSERT(load_start >= addr && load_end <= end);
-    const u32 start_offset = load_start - addr;
+    const u64 start_offset = load_start - addr;
 
     if (!is_tiled) {
         ASSERT(type == SurfaceType::Color);
-        std::memcpy(&gl_buffer[start_offset], texture_src_data + start_offset,
-                    load_end - load_start);
+        const u32 bytes_per_pixel{GetFormatBpp() >> 3};
+
+        // TODO(bunnei): Assumes the default rendering GOB size of 16 (128 lines). We should check
+        // the configuration for this and perform more generic un/swizzle
+        LOG_WARNING(Render_OpenGL, "need to use correct swizzle/GOB parameters!");
+        VideoCore::MortonCopyPixels128(width, height, bytes_per_pixel, 4,
+                                       texture_src_data + start_offset, &gl_buffer[start_offset],
+                                       true);
     } else {
-        if (type == SurfaceType::Texture) {
-            UNIMPLEMENTED();
-        } else {
-            morton_to_gl_fns[static_cast<size_t>(pixel_format)](stride, height, &gl_buffer[0], addr,
-                                                                load_start, load_end);
-        }
+        morton_to_gl_fns[static_cast<size_t>(pixel_format)](stride, height, &gl_buffer[0], addr,
+                                                            load_start, load_end);
     }
 }
 
 MICROPROFILE_DEFINE(OpenGL_SurfaceFlush, "OpenGL", "Surface Flush", MP_RGB(128, 192, 64));
-void CachedSurface::FlushGLBuffer(PAddr flush_start, PAddr flush_end) {
-    u8* const dst_buffer = Memory::GetPhysicalPointer(addr);
+void CachedSurface::FlushGLBuffer(VAddr flush_start, VAddr flush_end) {
+    u8* const dst_buffer = Memory::GetPointer(addr);
     if (dst_buffer == nullptr)
         return;
 
@@ -1102,18 +1057,106 @@ SurfaceRect_Tuple RasterizerCacheOpenGL::GetSurfaceSubRect(const SurfaceParams&
 }
 
 Surface RasterizerCacheOpenGL::GetTextureSurface(const void* config) {
-    UNIMPLEMENTED();
+    UNREACHABLE();
     return {};
 }
 
 SurfaceSurfaceRect_Tuple RasterizerCacheOpenGL::GetFramebufferSurfaces(
-    bool using_color_fb, bool using_depth_fb, const MathUtil::Rectangle<s32>& viewport_rect) {
-    UNIMPLEMENTED();
-    return {};
+    bool using_color_fb, bool using_depth_fb, const MathUtil::Rectangle<s32>& viewport) {
+    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    const auto& memory_manager = Core::System().GetInstance().GPU().memory_manager;
+    const auto& config = regs.rt[0];
+
+    // TODO(bunnei): This is hard corded to use just the first render buffer
+    LOG_WARNING(Render_OpenGL, "hard-coded for render target 0!");
+
+    // update resolution_scale_factor and reset cache if changed
+    // TODO (bunnei): This code was ported as-is from Citra, and is technically not thread-safe. We
+    // need to fix this before making the renderer multi-threaded.
+    static u16 resolution_scale_factor = GetResolutionScaleFactor();
+    if (resolution_scale_factor != GetResolutionScaleFactor()) {
+        resolution_scale_factor = GetResolutionScaleFactor();
+        FlushAll();
+        while (!surface_cache.empty())
+            UnregisterSurface(*surface_cache.begin()->second.begin());
+    }
+
+    MathUtil::Rectangle<u32> viewport_clamped{
+        static_cast<u32>(MathUtil::Clamp(viewport.left, 0, static_cast<s32>(config.width))),
+        static_cast<u32>(MathUtil::Clamp(viewport.top, 0, static_cast<s32>(config.height))),
+        static_cast<u32>(MathUtil::Clamp(viewport.right, 0, static_cast<s32>(config.width))),
+        static_cast<u32>(MathUtil::Clamp(viewport.bottom, 0, static_cast<s32>(config.height)))};
+
+    // get color and depth surfaces
+    SurfaceParams color_params;
+    color_params.is_tiled = true;
+    color_params.res_scale = resolution_scale_factor;
+    color_params.width = config.width;
+    color_params.height = config.height;
+    SurfaceParams depth_params = color_params;
+
+    color_params.addr = memory_manager->PhysicalToVirtualAddress(config.Address());
+    color_params.pixel_format = SurfaceParams::PixelFormatFromRenderTargetFormat(config.format);
+    color_params.UpdateParams();
+
+    ASSERT_MSG(!using_depth_fb, "depth buffer is unimplemented");
+    // depth_params.addr = config.GetDepthBufferPhysicalAddress();
+    // depth_params.pixel_format = SurfaceParams::PixelFormatFromDepthFormat(config.depth_format);
+    // depth_params.UpdateParams();
+
+    auto color_vp_interval = color_params.GetSubRectInterval(viewport_clamped);
+    auto depth_vp_interval = depth_params.GetSubRectInterval(viewport_clamped);
+
+    // Make sure that framebuffers don't overlap if both color and depth are being used
+    if (using_color_fb && using_depth_fb &&
+        boost::icl::length(color_vp_interval & depth_vp_interval)) {
+        LOG_CRITICAL(Render_OpenGL, "Color and depth framebuffer memory regions overlap; "
+                                    "overlapping framebuffers not supported!");
+        using_depth_fb = false;
+    }
+
+    MathUtil::Rectangle<u32> color_rect{};
+    Surface color_surface = nullptr;
+    if (using_color_fb)
+        std::tie(color_surface, color_rect) =
+            GetSurfaceSubRect(color_params, ScaleMatch::Exact, false);
+
+    MathUtil::Rectangle<u32> depth_rect{};
+    Surface depth_surface = nullptr;
+    if (using_depth_fb)
+        std::tie(depth_surface, depth_rect) =
+            GetSurfaceSubRect(depth_params, ScaleMatch::Exact, false);
+
+    MathUtil::Rectangle<u32> fb_rect{};
+    if (color_surface != nullptr && depth_surface != nullptr) {
+        fb_rect = color_rect;
+        // Color and Depth surfaces must have the same dimensions and offsets
+        if (color_rect.bottom != depth_rect.bottom || color_rect.top != depth_rect.top ||
+            color_rect.left != depth_rect.left || color_rect.right != depth_rect.right) {
+            color_surface = GetSurface(color_params, ScaleMatch::Exact, false);
+            depth_surface = GetSurface(depth_params, ScaleMatch::Exact, false);
+            fb_rect = color_surface->GetScaledRect();
+        }
+    } else if (color_surface != nullptr) {
+        fb_rect = color_rect;
+    } else if (depth_surface != nullptr) {
+        fb_rect = depth_rect;
+    }
+
+    if (color_surface != nullptr) {
+        ValidateSurface(color_surface, boost::icl::first(color_vp_interval),
+                        boost::icl::length(color_vp_interval));
+    }
+    if (depth_surface != nullptr) {
+        ValidateSurface(depth_surface, boost::icl::first(depth_vp_interval),
+                        boost::icl::length(depth_vp_interval));
+    }
+
+    return std::make_tuple(color_surface, depth_surface, fb_rect);
 }
 
 Surface RasterizerCacheOpenGL::GetFillSurface(const void* config) {
-    UNIMPLEMENTED();
+    UNREACHABLE();
     return {};
 }
 
@@ -1167,7 +1210,7 @@ void RasterizerCacheOpenGL::DuplicateSurface(const Surface& src_surface,
     }
 }
 
-void RasterizerCacheOpenGL::ValidateSurface(const Surface& surface, PAddr addr, u64 size) {
+void RasterizerCacheOpenGL::ValidateSurface(const Surface& surface, VAddr addr, u64 size) {
     if (size == 0)
         return;
 
@@ -1227,7 +1270,7 @@ void RasterizerCacheOpenGL::ValidateSurface(const Surface& surface, PAddr addr,
     }
 }
 
-void RasterizerCacheOpenGL::FlushRegion(PAddr addr, u64 size, Surface flush_surface) {
+void RasterizerCacheOpenGL::FlushRegion(VAddr addr, u64 size, Surface flush_surface) {
     if (size == 0)
         return;
 
@@ -1260,10 +1303,10 @@ void RasterizerCacheOpenGL::FlushRegion(PAddr addr, u64 size, Surface flush_surf
 }
 
 void RasterizerCacheOpenGL::FlushAll() {
-    FlushRegion(0, 0xFFFFFFFF);
+    FlushRegion(0, Kernel::VMManager::MAX_ADDRESS);
 }
 
-void RasterizerCacheOpenGL::InvalidateRegion(PAddr addr, u64 size, const Surface& region_owner) {
+void RasterizerCacheOpenGL::InvalidateRegion(VAddr addr, u64 size, const Surface& region_owner) {
     if (size == 0)
         return;
 
@@ -1356,6 +1399,34 @@ void RasterizerCacheOpenGL::UnregisterSurface(const Surface& surface) {
     surface_cache.subtract({surface->GetInterval(), SurfaceSet{surface}});
 }
 
-void RasterizerCacheOpenGL::UpdatePagesCachedCount(PAddr addr, u64 size, int delta) {
-    UNIMPLEMENTED();
+void RasterizerCacheOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
+    const u64 num_pages =
+        ((addr + size - 1) >> Memory::PAGE_BITS) - (addr >> Memory::PAGE_BITS) + 1;
+    const u64 page_start = addr >> Memory::PAGE_BITS;
+    const u64 page_end = page_start + num_pages;
+
+    // Interval maps will erase segments if count reaches 0, so if delta is negative we have to
+    // subtract after iterating
+    const auto pages_interval = PageMap::interval_type::right_open(page_start, page_end);
+    if (delta > 0)
+        cached_pages.add({pages_interval, delta});
+
+    for (const auto& pair : RangeFromInterval(cached_pages, pages_interval)) {
+        const auto interval = pair.first & pages_interval;
+        const int count = pair.second;
+
+        const VAddr interval_start_addr = boost::icl::first(interval) << Memory::PAGE_BITS;
+        const VAddr interval_end_addr = boost::icl::last_next(interval) << Memory::PAGE_BITS;
+        const u64 interval_size = interval_end_addr - interval_start_addr;
+
+        if (delta > 0 && count == delta)
+            Memory::RasterizerMarkRegionCached(interval_start_addr, interval_size, true);
+        else if (delta < 0 && count == -delta)
+            Memory::RasterizerMarkRegionCached(interval_start_addr, interval_size, false);
+        else
+            ASSERT(count >= 0);
+    }
+
+    if (delta < 0)
+        cached_pages.add({pages_interval, delta});
 }

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -22,15 +22,16 @@
 #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<PAddr>;
-using SurfaceMap = boost::icl::interval_map<PAddr, Surface>;
-using SurfaceCache = boost::icl::interval_map<PAddr, SurfaceSet>;
+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>() &&
@@ -40,7 +41,7 @@ static_assert(std::is_same<SurfaceRegions::interval_type, SurfaceCache::interval
 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<u32, int>;
+using PageMap = boost::icl::interval_map<u64, int>;
 
 enum class ScaleMatch {
     Exact,   // only accept same res scale
@@ -115,6 +116,24 @@ struct SurfaceParams {
         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);
@@ -211,8 +230,8 @@ struct SurfaceParams {
     MathUtil::Rectangle<u32> GetSubRect(const SurfaceParams& sub_surface) const;
     MathUtil::Rectangle<u32> GetScaledSubRect(const SurfaceParams& sub_surface) const;
 
-    PAddr addr = 0;
-    PAddr end = 0;
+    VAddr addr = 0;
+    VAddr end = 0;
     u64 size = 0;
 
     u32 width = 0;
@@ -257,9 +276,9 @@ struct CachedSurface : SurfaceParams {
     std::unique_ptr<u8[]> gl_buffer;
     size_t gl_buffer_size = 0;
 
-    // Read/Write data in 3DS memory to/from gl_buffer
-    void LoadGLBuffer(PAddr load_start, PAddr load_end);
-    void FlushGLBuffer(PAddr flush_start, PAddr flush_end);
+    // 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,
@@ -298,7 +317,7 @@ public:
 
     /// 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_rect);
+                                                    const MathUtil::Rectangle<s32>& viewport);
 
     /// Get a surface that matches the fill config
     Surface GetFillSurface(const void* config);
@@ -307,10 +326,10 @@ public:
     SurfaceRect_Tuple GetTexCopySurface(const SurfaceParams& params);
 
     /// Write any cached resources overlapping the region back to memory (if dirty)
-    void FlushRegion(PAddr addr, u64 size, Surface flush_surface = nullptr);
+    void FlushRegion(VAddr addr, u64 size, Surface flush_surface = nullptr);
 
     /// Mark region as being invalidated by region_owner (nullptr if 3DS memory)
-    void InvalidateRegion(PAddr addr, u64 size, const Surface& region_owner);
+    void InvalidateRegion(VAddr addr, u64 size, const Surface& region_owner);
 
     /// Flush all cached resources tracked by this cache manager
     void FlushAll();
@@ -319,7 +338,7 @@ 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, PAddr addr, u64 size);
+    void ValidateSurface(const Surface& surface, VAddr addr, u64 size);
 
     /// Create a new surface
     Surface CreateSurface(const SurfaceParams& params);
@@ -331,7 +350,7 @@ private:
     void UnregisterSurface(const Surface& surface);
 
     /// Increase/decrease the number of surface in pages touching the specified region
-    void UpdatePagesCachedCount(PAddr addr, u64 size, int delta);
+    void UpdatePagesCachedCount(VAddr addr, u64 size, int delta);
 
     SurfaceCache surface_cache;
     PageMap cached_pages;

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -26,7 +26,7 @@ public:
           sanitize_mul(sanitize_mul), emit_cb(emit_cb), setemit_cb(setemit_cb) {}
 
     std::string Decompile() {
-        UNIMPLEMENTED();
+        UNREACHABLE();
         return {};
     }
 

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -8,12 +8,12 @@
 namespace GLShader {
 
 std::string GenerateVertexShader(const MaxwellVSConfig& config) {
-    UNIMPLEMENTED();
+    UNREACHABLE();
     return {};
 }
 
 std::string GenerateFragmentShader(const MaxwellFSConfig& config) {
-    UNIMPLEMENTED();
+    UNREACHABLE();
     return {};
 }
 

src/video_core/renderer_opengl/gl_shader_util.cpp

@@ -38,8 +38,8 @@ GLuint LoadProgram(const char* vertex_shader, const char* geometry_shader,
             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]);
+                LOG_CRITICAL(Render_OpenGL, "Error compiling vertex shader:\n%s",
+                             &vertex_shader_error[0]);
             }
         }
     }
@@ -62,8 +62,8 @@ GLuint LoadProgram(const char* vertex_shader, const char* geometry_shader,
             if (result == GL_TRUE) {
                 LOG_DEBUG(Render_OpenGL, "%s", &geometry_shader_error[0]);
             } else {
-                LOG_ERROR(Render_OpenGL, "Error compiling geometry shader:\n%s",
-                          &geometry_shader_error[0]);
+                LOG_CRITICAL(Render_OpenGL, "Error compiling geometry shader:\n%s",
+                             &geometry_shader_error[0]);
             }
         }
     }
@@ -86,8 +86,8 @@ GLuint LoadProgram(const char* vertex_shader, const char* geometry_shader,
             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]);
+                LOG_CRITICAL(Render_OpenGL, "Error compiling fragment shader:\n%s",
+                             &fragment_shader_error[0]);
             }
         }
     }
@@ -128,20 +128,20 @@ GLuint LoadProgram(const char* vertex_shader, const char* geometry_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) {
         if (vertex_shader) {
-            LOG_ERROR(Render_OpenGL, "Vertex shader:\n%s", vertex_shader);
+            LOG_CRITICAL(Render_OpenGL, "Vertex shader:\n%s", vertex_shader);
         }
         if (geometry_shader) {
-            LOG_ERROR(Render_OpenGL, "Geometry shader:\n%s", geometry_shader);
+            LOG_CRITICAL(Render_OpenGL, "Geometry shader:\n%s", geometry_shader);
         }
         if (fragment_shader) {
-            LOG_ERROR(Render_OpenGL, "Fragment shader:\n%s", fragment_shader);
+            LOG_CRITICAL(Render_OpenGL, "Fragment shader:\n%s", fragment_shader);
         }
     }
     ASSERT_MSG(result == GL_TRUE, "Shader not linked");

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

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

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -20,6 +20,7 @@
 #include "core/settings.h"
 #include "core/tracer/recorder.h"
 #include "video_core/renderer_opengl/renderer_opengl.h"
+#include "video_core/utils.h"
 #include "video_core/video_core.h"
 
 static const char vertex_shader[] = R"(
@@ -98,197 +99,88 @@ RendererOpenGL::RendererOpenGL() = default;
 RendererOpenGL::~RendererOpenGL() = default;
 
 /// Swap buffers (render frame)
-void RendererOpenGL::SwapBuffers(boost::optional<const FramebufferInfo&> framebuffer_info) {
+void RendererOpenGL::SwapBuffers(boost::optional<const Tegra::FramebufferConfig&> framebuffer) {
+    Core::System::GetInstance().perf_stats.EndSystemFrame();
+
     // Maintain the rasterizer's state as a priority
     OpenGLState prev_state = OpenGLState::GetCurState();
     state.Apply();
 
-    if (framebuffer_info != boost::none) {
-        // If framebuffer_info is provided, reload it from memory to a texture
-        if (screen_info.texture.width != (GLsizei)framebuffer_info->width ||
-            screen_info.texture.height != (GLsizei)framebuffer_info->height ||
-            screen_info.texture.pixel_format != framebuffer_info->pixel_format) {
+    if (framebuffer != boost::none) {
+        // If framebuffer is provided, reload it from memory to a texture
+        if (screen_info.texture.width != (GLsizei)framebuffer->width ||
+            screen_info.texture.height != (GLsizei)framebuffer->height ||
+            screen_info.texture.pixel_format != framebuffer->pixel_format) {
             // Reallocate texture if the framebuffer size has changed.
             // This is expected to not happen very often and hence should not be a
             // performance problem.
-            ConfigureFramebufferTexture(screen_info.texture, *framebuffer_info);
+            ConfigureFramebufferTexture(screen_info.texture, *framebuffer);
         }
-        LoadFBToScreenInfo(*framebuffer_info, screen_info);
+
+        // Load the framebuffer from memory, draw it to the screen, and swap buffers
+        LoadFBToScreenInfo(*framebuffer, screen_info);
+        DrawScreen();
+        render_window->SwapBuffers();
     }
 
-    DrawScreens();
-
-    Core::System::GetInstance().perf_stats.EndSystemFrame();
-
-    // Swap buffers
     render_window->PollEvents();
-    render_window->SwapBuffers();
 
     Core::System::GetInstance().frame_limiter.DoFrameLimiting(CoreTiming::GetGlobalTimeUs());
     Core::System::GetInstance().perf_stats.BeginSystemFrame();
 
+    // Restore the rasterizer state
     prev_state.Apply();
     RefreshRasterizerSetting();
 }
 
-static inline u32 MortonInterleave128(u32 x, u32 y) {
-    // 128x128 Z-Order coordinate from 2D coordinates
-    static constexpr u32 xlut[] = {
-        0x0000, 0x0001, 0x0002, 0x0003, 0x0008, 0x0009, 0x000a, 0x000b, 0x0040, 0x0041, 0x0042,
-        0x0043, 0x0048, 0x0049, 0x004a, 0x004b, 0x0800, 0x0801, 0x0802, 0x0803, 0x0808, 0x0809,
-        0x080a, 0x080b, 0x0840, 0x0841, 0x0842, 0x0843, 0x0848, 0x0849, 0x084a, 0x084b, 0x1000,
-        0x1001, 0x1002, 0x1003, 0x1008, 0x1009, 0x100a, 0x100b, 0x1040, 0x1041, 0x1042, 0x1043,
-        0x1048, 0x1049, 0x104a, 0x104b, 0x1800, 0x1801, 0x1802, 0x1803, 0x1808, 0x1809, 0x180a,
-        0x180b, 0x1840, 0x1841, 0x1842, 0x1843, 0x1848, 0x1849, 0x184a, 0x184b, 0x2000, 0x2001,
-        0x2002, 0x2003, 0x2008, 0x2009, 0x200a, 0x200b, 0x2040, 0x2041, 0x2042, 0x2043, 0x2048,
-        0x2049, 0x204a, 0x204b, 0x2800, 0x2801, 0x2802, 0x2803, 0x2808, 0x2809, 0x280a, 0x280b,
-        0x2840, 0x2841, 0x2842, 0x2843, 0x2848, 0x2849, 0x284a, 0x284b, 0x3000, 0x3001, 0x3002,
-        0x3003, 0x3008, 0x3009, 0x300a, 0x300b, 0x3040, 0x3041, 0x3042, 0x3043, 0x3048, 0x3049,
-        0x304a, 0x304b, 0x3800, 0x3801, 0x3802, 0x3803, 0x3808, 0x3809, 0x380a, 0x380b, 0x3840,
-        0x3841, 0x3842, 0x3843, 0x3848, 0x3849, 0x384a, 0x384b, 0x0000, 0x0001, 0x0002, 0x0003,
-        0x0008, 0x0009, 0x000a, 0x000b, 0x0040, 0x0041, 0x0042, 0x0043, 0x0048, 0x0049, 0x004a,
-        0x004b, 0x0800, 0x0801, 0x0802, 0x0803, 0x0808, 0x0809, 0x080a, 0x080b, 0x0840, 0x0841,
-        0x0842, 0x0843, 0x0848, 0x0849, 0x084a, 0x084b, 0x1000, 0x1001, 0x1002, 0x1003, 0x1008,
-        0x1009, 0x100a, 0x100b, 0x1040, 0x1041, 0x1042, 0x1043, 0x1048, 0x1049, 0x104a, 0x104b,
-        0x1800, 0x1801, 0x1802, 0x1803, 0x1808, 0x1809, 0x180a, 0x180b, 0x1840, 0x1841, 0x1842,
-        0x1843, 0x1848, 0x1849, 0x184a, 0x184b, 0x2000, 0x2001, 0x2002, 0x2003, 0x2008, 0x2009,
-        0x200a, 0x200b, 0x2040, 0x2041, 0x2042, 0x2043, 0x2048, 0x2049, 0x204a, 0x204b, 0x2800,
-        0x2801, 0x2802, 0x2803, 0x2808, 0x2809, 0x280a, 0x280b, 0x2840, 0x2841, 0x2842, 0x2843,
-        0x2848, 0x2849, 0x284a, 0x284b, 0x3000, 0x3001, 0x3002, 0x3003, 0x3008, 0x3009, 0x300a,
-        0x300b, 0x3040, 0x3041, 0x3042, 0x3043, 0x3048, 0x3049, 0x304a, 0x304b, 0x3800, 0x3801,
-        0x3802, 0x3803, 0x3808, 0x3809, 0x380a, 0x380b, 0x3840, 0x3841, 0x3842, 0x3843, 0x3848,
-        0x3849, 0x384a, 0x384b, 0x0000, 0x0001, 0x0002, 0x0003, 0x0008, 0x0009, 0x000a, 0x000b,
-        0x0040, 0x0041, 0x0042, 0x0043, 0x0048, 0x0049, 0x004a, 0x004b, 0x0800, 0x0801, 0x0802,
-        0x0803, 0x0808, 0x0809, 0x080a, 0x080b, 0x0840, 0x0841, 0x0842, 0x0843, 0x0848, 0x0849,
-        0x084a, 0x084b, 0x1000, 0x1001, 0x1002, 0x1003, 0x1008, 0x1009, 0x100a, 0x100b, 0x1040,
-        0x1041, 0x1042, 0x1043, 0x1048, 0x1049, 0x104a, 0x104b, 0x1800, 0x1801, 0x1802, 0x1803,
-        0x1808, 0x1809, 0x180a, 0x180b, 0x1840, 0x1841, 0x1842, 0x1843, 0x1848, 0x1849, 0x184a,
-        0x184b, 0x2000, 0x2001, 0x2002, 0x2003, 0x2008, 0x2009, 0x200a, 0x200b, 0x2040, 0x2041,
-        0x2042, 0x2043, 0x2048, 0x2049, 0x204a, 0x204b, 0x2800, 0x2801, 0x2802, 0x2803, 0x2808,
-        0x2809, 0x280a, 0x280b, 0x2840, 0x2841, 0x2842, 0x2843, 0x2848, 0x2849, 0x284a, 0x284b,
-        0x3000, 0x3001, 0x3002, 0x3003, 0x3008, 0x3009, 0x300a, 0x300b, 0x3040, 0x3041, 0x3042,
-        0x3043, 0x3048, 0x3049, 0x304a, 0x304b, 0x3800, 0x3801, 0x3802, 0x3803, 0x3808, 0x3809,
-        0x380a, 0x380b, 0x3840, 0x3841, 0x3842, 0x3843, 0x3848, 0x3849, 0x384a, 0x384b,
-    };
-    static constexpr u32 ylut[] = {
-        0x0000, 0x0004, 0x0010, 0x0014, 0x0020, 0x0024, 0x0030, 0x0034, 0x0080, 0x0084, 0x0090,
-        0x0094, 0x00a0, 0x00a4, 0x00b0, 0x00b4, 0x0100, 0x0104, 0x0110, 0x0114, 0x0120, 0x0124,
-        0x0130, 0x0134, 0x0180, 0x0184, 0x0190, 0x0194, 0x01a0, 0x01a4, 0x01b0, 0x01b4, 0x0200,
-        0x0204, 0x0210, 0x0214, 0x0220, 0x0224, 0x0230, 0x0234, 0x0280, 0x0284, 0x0290, 0x0294,
-        0x02a0, 0x02a4, 0x02b0, 0x02b4, 0x0300, 0x0304, 0x0310, 0x0314, 0x0320, 0x0324, 0x0330,
-        0x0334, 0x0380, 0x0384, 0x0390, 0x0394, 0x03a0, 0x03a4, 0x03b0, 0x03b4, 0x0400, 0x0404,
-        0x0410, 0x0414, 0x0420, 0x0424, 0x0430, 0x0434, 0x0480, 0x0484, 0x0490, 0x0494, 0x04a0,
-        0x04a4, 0x04b0, 0x04b4, 0x0500, 0x0504, 0x0510, 0x0514, 0x0520, 0x0524, 0x0530, 0x0534,
-        0x0580, 0x0584, 0x0590, 0x0594, 0x05a0, 0x05a4, 0x05b0, 0x05b4, 0x0600, 0x0604, 0x0610,
-        0x0614, 0x0620, 0x0624, 0x0630, 0x0634, 0x0680, 0x0684, 0x0690, 0x0694, 0x06a0, 0x06a4,
-        0x06b0, 0x06b4, 0x0700, 0x0704, 0x0710, 0x0714, 0x0720, 0x0724, 0x0730, 0x0734, 0x0780,
-        0x0784, 0x0790, 0x0794, 0x07a0, 0x07a4, 0x07b0, 0x07b4, 0x0000, 0x0004, 0x0010, 0x0014,
-        0x0020, 0x0024, 0x0030, 0x0034, 0x0080, 0x0084, 0x0090, 0x0094, 0x00a0, 0x00a4, 0x00b0,
-        0x00b4, 0x0100, 0x0104, 0x0110, 0x0114, 0x0120, 0x0124, 0x0130, 0x0134, 0x0180, 0x0184,
-        0x0190, 0x0194, 0x01a0, 0x01a4, 0x01b0, 0x01b4, 0x0200, 0x0204, 0x0210, 0x0214, 0x0220,
-        0x0224, 0x0230, 0x0234, 0x0280, 0x0284, 0x0290, 0x0294, 0x02a0, 0x02a4, 0x02b0, 0x02b4,
-        0x0300, 0x0304, 0x0310, 0x0314, 0x0320, 0x0324, 0x0330, 0x0334, 0x0380, 0x0384, 0x0390,
-        0x0394, 0x03a0, 0x03a4, 0x03b0, 0x03b4, 0x0400, 0x0404, 0x0410, 0x0414, 0x0420, 0x0424,
-        0x0430, 0x0434, 0x0480, 0x0484, 0x0490, 0x0494, 0x04a0, 0x04a4, 0x04b0, 0x04b4, 0x0500,
-        0x0504, 0x0510, 0x0514, 0x0520, 0x0524, 0x0530, 0x0534, 0x0580, 0x0584, 0x0590, 0x0594,
-        0x05a0, 0x05a4, 0x05b0, 0x05b4, 0x0600, 0x0604, 0x0610, 0x0614, 0x0620, 0x0624, 0x0630,
-        0x0634, 0x0680, 0x0684, 0x0690, 0x0694, 0x06a0, 0x06a4, 0x06b0, 0x06b4, 0x0700, 0x0704,
-        0x0710, 0x0714, 0x0720, 0x0724, 0x0730, 0x0734, 0x0780, 0x0784, 0x0790, 0x0794, 0x07a0,
-        0x07a4, 0x07b0, 0x07b4, 0x0000, 0x0004, 0x0010, 0x0014, 0x0020, 0x0024, 0x0030, 0x0034,
-        0x0080, 0x0084, 0x0090, 0x0094, 0x00a0, 0x00a4, 0x00b0, 0x00b4, 0x0100, 0x0104, 0x0110,
-        0x0114, 0x0120, 0x0124, 0x0130, 0x0134, 0x0180, 0x0184, 0x0190, 0x0194, 0x01a0, 0x01a4,
-        0x01b0, 0x01b4, 0x0200, 0x0204, 0x0210, 0x0214, 0x0220, 0x0224, 0x0230, 0x0234, 0x0280,
-        0x0284, 0x0290, 0x0294, 0x02a0, 0x02a4, 0x02b0, 0x02b4, 0x0300, 0x0304, 0x0310, 0x0314,
-        0x0320, 0x0324, 0x0330, 0x0334, 0x0380, 0x0384, 0x0390, 0x0394, 0x03a0, 0x03a4, 0x03b0,
-        0x03b4, 0x0400, 0x0404, 0x0410, 0x0414, 0x0420, 0x0424, 0x0430, 0x0434, 0x0480, 0x0484,
-        0x0490, 0x0494, 0x04a0, 0x04a4, 0x04b0, 0x04b4, 0x0500, 0x0504, 0x0510, 0x0514, 0x0520,
-        0x0524, 0x0530, 0x0534, 0x0580, 0x0584, 0x0590, 0x0594, 0x05a0, 0x05a4, 0x05b0, 0x05b4,
-        0x0600, 0x0604, 0x0610, 0x0614, 0x0620, 0x0624, 0x0630, 0x0634, 0x0680, 0x0684, 0x0690,
-        0x0694, 0x06a0, 0x06a4, 0x06b0, 0x06b4, 0x0700, 0x0704, 0x0710, 0x0714, 0x0720, 0x0724,
-        0x0730, 0x0734, 0x0780, 0x0784, 0x0790, 0x0794, 0x07a0, 0x07a4, 0x07b0, 0x07b4,
-    };
-    return xlut[x % 128] + ylut[y % 128];
-}
-
-static inline u32 GetMortonOffset128(u32 x, u32 y, u32 bytes_per_pixel) {
-    // Calculates the offset of the position of the pixel in Morton order
-    // Framebuffer images are split into 128x128 tiles.
-
-    const unsigned int block_height = 128;
-    const unsigned int coarse_x = x & ~127;
-
-    u32 i = MortonInterleave128(x, y);
-
-    const unsigned int offset = coarse_x * block_height;
-
-    return (i + offset) * bytes_per_pixel;
-}
-
-static void MortonCopyPixels128(u32 width, u32 height, u32 bytes_per_pixel, u32 gl_bytes_per_pixel,
-                                u8* morton_data, u8* gl_data, bool morton_to_gl) {
-    u8* data_ptrs[2];
-    for (unsigned y = 0; y < height; ++y) {
-        for (unsigned x = 0; x < width; ++x) {
-            const u32 coarse_y = y & ~127;
-            u32 morton_offset =
-                GetMortonOffset128(x, y, bytes_per_pixel) + coarse_y * width * bytes_per_pixel;
-            u32 gl_pixel_index = (x + (height - 1 - y) * width) * gl_bytes_per_pixel;
-
-            data_ptrs[morton_to_gl] = morton_data + morton_offset;
-            data_ptrs[!morton_to_gl] = &gl_data[gl_pixel_index];
-
-            memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel);
-        }
-    }
-}
-
 /**
  * Loads framebuffer from emulated memory into the active OpenGL texture.
  */
-void RendererOpenGL::LoadFBToScreenInfo(const FramebufferInfo& framebuffer_info,
+void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuffer,
                                         ScreenInfo& screen_info) {
-    const u32 bpp{FramebufferInfo::BytesPerPixel(framebuffer_info.pixel_format)};
-    const u32 size_in_bytes{framebuffer_info.stride * framebuffer_info.height * bpp};
+    const u32 bytes_per_pixel{Tegra::FramebufferConfig::BytesPerPixel(framebuffer.pixel_format)};
+    const u64 size_in_bytes{framebuffer.stride * framebuffer.height * bytes_per_pixel};
+    const VAddr framebuffer_addr{framebuffer.address + framebuffer.offset};
 
-    MortonCopyPixels128(framebuffer_info.width, framebuffer_info.height, bpp, 4,
-                        Memory::GetPointer(framebuffer_info.address), gl_framebuffer_data.data(),
-                        true);
-
-    LOG_TRACE(Render_OpenGL, "0x%08x bytes from 0x%llx(%dx%d), fmt %x", size_in_bytes,
-              framebuffer_info.address, framebuffer_info.width, framebuffer_info.height,
-              (int)framebuffer_info.pixel_format);
+    // Framebuffer orientation handling
+    framebuffer_transform_flags = framebuffer.transform_flags;
 
     // Ensure no bad interactions with GL_UNPACK_ALIGNMENT, which by default
     // only allows rows to have a memory alignement of 4.
-    ASSERT(framebuffer_info.stride % 4 == 0);
+    ASSERT(framebuffer.stride % 4 == 0);
 
-    framebuffer_flip_vertical = framebuffer_info.flip_vertical;
+    if (!Rasterizer()->AccelerateDisplay(framebuffer, framebuffer_addr, framebuffer.stride,
+                                         screen_info)) {
+        // Reset the screen info's display texture to its own permanent texture
+        screen_info.display_texture = screen_info.texture.resource.handle;
+        screen_info.display_texcoords = MathUtil::Rectangle<float>(0.f, 0.f, 1.f, 1.f);
 
-    // Reset the screen info's display texture to its own permanent texture
-    screen_info.display_texture = screen_info.texture.resource.handle;
-    screen_info.display_texcoords = MathUtil::Rectangle<float>(0.f, 0.f, 1.f, 1.f);
+        Rasterizer()->FlushRegion(framebuffer_addr, size_in_bytes);
 
-    // Memory::RasterizerFlushRegion(framebuffer_info.address, size_in_bytes);
+        VideoCore::MortonCopyPixels128(framebuffer.width, framebuffer.height, bytes_per_pixel, 4,
+                                       Memory::GetPointer(framebuffer_addr),
+                                       gl_framebuffer_data.data(), true);
 
-    state.texture_units[0].texture_2d = screen_info.texture.resource.handle;
-    state.Apply();
+        state.texture_units[0].texture_2d = screen_info.texture.resource.handle;
+        state.Apply();
 
-    glActiveTexture(GL_TEXTURE0);
-    glPixelStorei(GL_UNPACK_ROW_LENGTH, (GLint)framebuffer_info.stride);
+        glActiveTexture(GL_TEXTURE0);
+        glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(framebuffer.stride));
 
-    // Update existing texture
-    // TODO: Test what happens on hardware when you change the framebuffer dimensions so that
-    //       they differ from the LCD resolution.
-    // TODO: Applications could theoretically crash Citra here by specifying too large
-    //       framebuffer sizes. We should make sure that this cannot happen.
-    glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, framebuffer_info.width, framebuffer_info.height,
-                    screen_info.texture.gl_format, screen_info.texture.gl_type,
-                    gl_framebuffer_data.data());
+        // Update existing texture
+        // TODO: Test what happens on hardware when you change the framebuffer dimensions so that
+        //       they differ from the LCD resolution.
+        // TODO: Applications could theoretically crash yuzu here by specifying too large
+        //       framebuffer sizes. We should make sure that this cannot happen.
+        glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, framebuffer.width, framebuffer.height,
+                        screen_info.texture.gl_format, screen_info.texture.gl_type,
+                        gl_framebuffer_data.data());
 
-    glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
+        glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
 
-    state.texture_units[0].texture_2d = 0;
-    state.Apply();
+        state.texture_units[0].texture_2d = 0;
+        state.Apply();
+    }
 }
 
 /**
@@ -372,14 +264,14 @@ void RendererOpenGL::InitOpenGLObjects() {
 }
 
 void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,
-                                                 const FramebufferInfo& framebuffer_info) {
+                                                 const Tegra::FramebufferConfig& framebuffer) {
 
-    texture.width = framebuffer_info.width;
-    texture.height = framebuffer_info.height;
+    texture.width = framebuffer.width;
+    texture.height = framebuffer.height;
 
     GLint internal_format;
-    switch (framebuffer_info.pixel_format) {
-    case FramebufferInfo::PixelFormat::ABGR8:
+    switch (framebuffer.pixel_format) {
+    case Tegra::FramebufferConfig::PixelFormat::ABGR8:
         // Use RGBA8 and swap in the fragment shader
         internal_format = GL_RGBA;
         texture.gl_format = GL_RGBA;
@@ -387,7 +279,7 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,
         gl_framebuffer_data.resize(texture.width * texture.height * 4);
         break;
     default:
-        UNIMPLEMENTED();
+        UNREACHABLE();
     }
 
     state.texture_units[0].texture_2d = texture.resource.handle;
@@ -401,11 +293,22 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,
     state.Apply();
 }
 
-void RendererOpenGL::DrawSingleScreen(const ScreenInfo& screen_info, float x, float y, float w,
-                                      float h) {
+void RendererOpenGL::DrawScreenTriangles(const ScreenInfo& screen_info, float x, float y, float w,
+                                         float h) {
     const auto& texcoords = screen_info.display_texcoords;
-    const auto& left = framebuffer_flip_vertical ? texcoords.right : texcoords.left;
-    const auto& right = framebuffer_flip_vertical ? texcoords.left : texcoords.right;
+    auto left = texcoords.left;
+    auto right = texcoords.right;
+    if (framebuffer_transform_flags != Tegra::FramebufferConfig::TransformFlags::Unset)
+        if (framebuffer_transform_flags == Tegra::FramebufferConfig::TransformFlags::FlipV) {
+            // Flip the framebuffer vertically
+            left = texcoords.right;
+            right = texcoords.left;
+        } else {
+            // Other transformations are unsupported
+            LOG_CRITICAL(Render_OpenGL, "Unsupported framebuffer_transform_flags=%d",
+                         framebuffer_transform_flags);
+            UNIMPLEMENTED();
+        }
 
     std::array<ScreenRectVertex, 4> vertices = {{
         ScreenRectVertex(x, y, texcoords.top, right),
@@ -427,7 +330,7 @@ void RendererOpenGL::DrawSingleScreen(const ScreenInfo& screen_info, float x, fl
 /**
  * Draws the emulated screens to the emulator window.
  */
-void RendererOpenGL::DrawScreens() {
+void RendererOpenGL::DrawScreen() {
     const auto& layout = render_window->GetFramebufferLayout();
     const auto& screen = layout.screen;
 
@@ -443,8 +346,8 @@ void RendererOpenGL::DrawScreens() {
     glActiveTexture(GL_TEXTURE0);
     glUniform1i(uniform_color_texture, 0);
 
-    DrawSingleScreen(screen_info, (float)screen.left, (float)screen.top, (float)screen.GetWidth(),
-                     (float)screen.GetHeight());
+    DrawScreenTriangles(screen_info, (float)screen.left, (float)screen.top,
+                        (float)screen.GetWidth(), (float)screen.GetHeight());
 
     m_current_frame++;
 }

src/video_core/renderer_opengl/renderer_opengl.h

@@ -21,7 +21,7 @@ struct TextureInfo {
     GLsizei height;
     GLenum gl_format;
     GLenum gl_type;
-    RendererBase::FramebufferInfo::PixelFormat pixel_format;
+    Tegra::FramebufferConfig::PixelFormat pixel_format;
 };
 
 /// Structure used for storing information about the display target for each 3DS screen
@@ -37,7 +37,7 @@ public:
     ~RendererOpenGL() override;
 
     /// Swap buffers (render frame)
-    void SwapBuffers(boost::optional<const FramebufferInfo&> framebuffer_info) override;
+    void SwapBuffers(boost::optional<const Tegra::FramebufferConfig&> framebuffer) override;
 
     /**
      * Set the emulator window to use for renderer
@@ -53,13 +53,14 @@ public:
 
 private:
     void InitOpenGLObjects();
-    void ConfigureFramebufferTexture(TextureInfo& texture, const FramebufferInfo& framebuffer_info);
-    void DrawScreens();
-    void DrawSingleScreen(const ScreenInfo& screen_info, float x, float y, float w, float h);
+    void ConfigureFramebufferTexture(TextureInfo& texture,
+                                     const Tegra::FramebufferConfig& framebuffer);
+    void DrawScreen();
+    void DrawScreenTriangles(const ScreenInfo& screen_info, float x, float y, float w, float h);
     void UpdateFramerate();
 
     // Loads framebuffer from emulated memory into the display information structure
-    void LoadFBToScreenInfo(const FramebufferInfo& framebuffer_info, ScreenInfo& screen_info);
+    void LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuffer, ScreenInfo& screen_info);
     // Fills active OpenGL texture with the given RGBA color.
     void LoadColorToActiveGLTexture(u8 color_r, u8 color_g, u8 color_b, u8 color_a,
                                     const TextureInfo& texture);
@@ -87,6 +88,6 @@ private:
     GLuint attrib_position;
     GLuint attrib_tex_coord;
 
-    /// Flips the framebuffer vertically when true
-    bool framebuffer_flip_vertical;
+    /// Used for transforming the framebuffer orientation
+    Tegra::FramebufferConfig::TransformFlags framebuffer_transform_flags;
 };

src/video_core/textures/decoders.cpp

@@ -0,0 +1,105 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include "common/assert.h"
+#include "video_core/textures/decoders.h"
+#include "video_core/textures/texture.h"
+
+namespace Tegra {
+namespace Texture {
+
+/**
+ * Calculates the offset of an (x, y) position within a swizzled texture.
+ * Taken from the Tegra X1 TRM.
+ */
+static u32 GetSwizzleOffset(u32 x, u32 y, u32 image_width, u32 bytes_per_pixel, u32 block_height) {
+    u32 image_width_in_gobs = image_width * bytes_per_pixel / 64;
+    u32 GOB_address = 0 + (y / (8 * block_height)) * 512 * block_height * image_width_in_gobs +
+                      (x * bytes_per_pixel / 64) * 512 * block_height +
+                      (y % (8 * block_height) / 8) * 512;
+    x *= bytes_per_pixel;
+    u32 address = GOB_address + ((x % 64) / 32) * 256 + ((y % 8) / 2) * 64 + ((x % 32) / 16) * 32 +
+                  (y % 2) * 16 + (x % 16);
+
+    return address;
+}
+
+static void CopySwizzledData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel,
+                             u8* swizzled_data, u8* unswizzled_data, bool unswizzle,
+                             u32 block_height) {
+    u8* data_ptrs[2];
+    for (unsigned y = 0; y < height; ++y) {
+        for (unsigned x = 0; x < width; ++x) {
+            u32 swizzle_offset = GetSwizzleOffset(x, y, width, bytes_per_pixel, block_height);
+            u32 pixel_index = (x + y * width) * out_bytes_per_pixel;
+
+            data_ptrs[unswizzle] = swizzled_data + swizzle_offset;
+            data_ptrs[!unswizzle] = &unswizzled_data[pixel_index];
+
+            std::memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel);
+        }
+    }
+}
+
+u32 BytesPerPixel(TextureFormat format) {
+    switch (format) {
+    case TextureFormat::DXT1:
+        // In this case a 'pixel' actually refers to a 4x4 tile.
+        return 8;
+    case TextureFormat::A8R8G8B8:
+        return 4;
+    default:
+        UNIMPLEMENTED_MSG("Format not implemented");
+        break;
+    }
+}
+
+std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width, u32 height) {
+    u8* data = Memory::GetPointer(address);
+    u32 bytes_per_pixel = BytesPerPixel(format);
+
+    static constexpr u32 DefaultBlockHeight = 16;
+
+    std::vector<u8> unswizzled_data(width * height * bytes_per_pixel);
+
+    switch (format) {
+    case TextureFormat::DXT1:
+        // In the DXT1 format, each 4x4 tile is swizzled instead of just individual pixel values.
+        CopySwizzledData(width / 4, height / 4, bytes_per_pixel, bytes_per_pixel, data,
+                         unswizzled_data.data(), true, DefaultBlockHeight);
+        break;
+    case TextureFormat::A8R8G8B8:
+        CopySwizzledData(width, height, bytes_per_pixel, bytes_per_pixel, data,
+                         unswizzled_data.data(), true, DefaultBlockHeight);
+        break;
+    default:
+        UNIMPLEMENTED_MSG("Format not implemented");
+        break;
+    }
+
+    return unswizzled_data;
+}
+
+std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat format, u32 width,
+                              u32 height) {
+    std::vector<u8> rgba_data;
+
+    // TODO(Subv): Implement.
+    switch (format) {
+    case TextureFormat::DXT1:
+    case TextureFormat::A8R8G8B8:
+        // TODO(Subv): For the time being just forward the same data without any decoding.
+        rgba_data = texture_data;
+        break;
+    default:
+        UNIMPLEMENTED_MSG("Format not implemented");
+        break;
+    }
+
+    return rgba_data;
+}
+
+} // namespace Texture
+} // namespace Tegra

src/video_core/textures/decoders.h

@@ -0,0 +1,26 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <vector>
+#include "common/common_types.h"
+#include "video_core/textures/texture.h"
+
+namespace Tegra {
+namespace Texture {
+
+/**
+ * Unswizzles a swizzled texture without changing its format.
+ */
+std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width, u32 height);
+
+/**
+ * Decodes an unswizzled texture into a A8R8G8B8 texture.
+ */
+std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat format, u32 width,
+                              u32 height);
+
+} // namespace Texture
+} // namespace Tegra

src/video_core/textures/texture.h

@@ -0,0 +1,137 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/bit_field.h"
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "video_core/memory_manager.h"
+
+namespace Tegra {
+namespace Texture {
+
+enum class TextureFormat : u32 {
+    A8R8G8B8 = 8,
+    DXT1 = 0x24,
+};
+
+enum class TextureType : u32 {
+    Texture1D = 0,
+    Texture2D = 1,
+    Texture3D = 2,
+    TextureCubemap = 3,
+    Texture1DArray = 4,
+    Texture2DArray = 5,
+    Texture1DBuffer = 6,
+    Texture2DNoMipmap = 7,
+    TextureCubeArray = 8,
+};
+
+enum class TICHeaderVersion : u32 {
+    OneDBuffer = 0,
+    PitchColorKey = 1,
+    Pitch = 2,
+    BlockLinear = 3,
+    BlockLinearColorKey = 4,
+};
+
+union TextureHandle {
+    u32 raw;
+    BitField<0, 20, u32> tic_id;
+    BitField<20, 12, u32> tsc_id;
+};
+static_assert(sizeof(TextureHandle) == 4, "TextureHandle has wrong size");
+
+struct TICEntry {
+    union {
+        u32 raw;
+        BitField<0, 7, TextureFormat> format;
+        BitField<7, 3, u32> r_type;
+        BitField<10, 3, u32> g_type;
+        BitField<13, 3, u32> b_type;
+        BitField<16, 3, u32> a_type;
+    };
+    u32 address_low;
+    union {
+        BitField<0, 16, u32> address_high;
+        BitField<21, 3, TICHeaderVersion> header_version;
+    };
+    INSERT_PADDING_BYTES(4);
+    union {
+        BitField<0, 16, u32> width_minus_1;
+        BitField<23, 4, TextureType> texture_type;
+    };
+    u16 height_minus_1;
+    INSERT_PADDING_BYTES(10);
+
+    GPUVAddr Address() const {
+        return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) | address_low);
+    }
+
+    u32 Width() const {
+        return width_minus_1 + 1;
+    }
+
+    u32 Height() const {
+        return height_minus_1 + 1;
+    }
+};
+static_assert(sizeof(TICEntry) == 0x20, "TICEntry has wrong size");
+
+enum class WrapMode : u32 {
+    Wrap = 0,
+    Mirror = 1,
+    ClampToEdge = 2,
+    Border = 3,
+    ClampOGL = 4,
+    MirrorOnceClampToEdge = 5,
+    MirrorOnceBorder = 6,
+    MirrorOnceClampOGL = 7,
+};
+
+enum class TextureFilter : u32 {
+    Nearest = 1,
+    Linear = 2,
+};
+
+enum class TextureMipmapFilter : u32 {
+    None = 1,
+    Nearest = 2,
+    Linear = 3,
+};
+
+struct TSCEntry {
+    union {
+        BitField<0, 3, WrapMode> wrap_u;
+        BitField<3, 3, WrapMode> wrap_v;
+        BitField<6, 3, WrapMode> wrap_p;
+        BitField<9, 1, u32> depth_compare_enabled;
+        BitField<10, 3, u32> depth_compare_func;
+    };
+    union {
+        BitField<0, 2, TextureFilter> mag_filter;
+        BitField<4, 2, TextureFilter> min_filter;
+        BitField<6, 2, TextureMipmapFilter> mip_filter;
+    };
+    INSERT_PADDING_BYTES(8);
+    u32 border_color_r;
+    u32 border_color_g;
+    u32 border_color_b;
+    u32 border_color_a;
+};
+static_assert(sizeof(TSCEntry) == 0x20, "TSCEntry has wrong size");
+
+struct FullTextureInfo {
+    u32 index;
+    TICEntry tic;
+    TSCEntry tsc;
+    bool enabled;
+};
+
+/// Returns the number of bytes per pixel of the input texture format.
+u32 BytesPerPixel(TextureFormat format);
+
+} // namespace Texture
+} // namespace Tegra

src/video_core/utils.h

@@ -49,4 +49,116 @@ static inline u32 GetMortonOffset(u32 x, u32 y, u32 bytes_per_pixel) {
     return (i + offset) * bytes_per_pixel;
 }
 
+static inline u32 MortonInterleave128(u32 x, u32 y) {
+    // 128x128 Z-Order coordinate from 2D coordinates
+    static constexpr u32 xlut[] = {
+        0x0000, 0x0001, 0x0002, 0x0003, 0x0008, 0x0009, 0x000a, 0x000b, 0x0040, 0x0041, 0x0042,
+        0x0043, 0x0048, 0x0049, 0x004a, 0x004b, 0x0800, 0x0801, 0x0802, 0x0803, 0x0808, 0x0809,
+        0x080a, 0x080b, 0x0840, 0x0841, 0x0842, 0x0843, 0x0848, 0x0849, 0x084a, 0x084b, 0x1000,
+        0x1001, 0x1002, 0x1003, 0x1008, 0x1009, 0x100a, 0x100b, 0x1040, 0x1041, 0x1042, 0x1043,
+        0x1048, 0x1049, 0x104a, 0x104b, 0x1800, 0x1801, 0x1802, 0x1803, 0x1808, 0x1809, 0x180a,
+        0x180b, 0x1840, 0x1841, 0x1842, 0x1843, 0x1848, 0x1849, 0x184a, 0x184b, 0x2000, 0x2001,
+        0x2002, 0x2003, 0x2008, 0x2009, 0x200a, 0x200b, 0x2040, 0x2041, 0x2042, 0x2043, 0x2048,
+        0x2049, 0x204a, 0x204b, 0x2800, 0x2801, 0x2802, 0x2803, 0x2808, 0x2809, 0x280a, 0x280b,
+        0x2840, 0x2841, 0x2842, 0x2843, 0x2848, 0x2849, 0x284a, 0x284b, 0x3000, 0x3001, 0x3002,
+        0x3003, 0x3008, 0x3009, 0x300a, 0x300b, 0x3040, 0x3041, 0x3042, 0x3043, 0x3048, 0x3049,
+        0x304a, 0x304b, 0x3800, 0x3801, 0x3802, 0x3803, 0x3808, 0x3809, 0x380a, 0x380b, 0x3840,
+        0x3841, 0x3842, 0x3843, 0x3848, 0x3849, 0x384a, 0x384b, 0x0000, 0x0001, 0x0002, 0x0003,
+        0x0008, 0x0009, 0x000a, 0x000b, 0x0040, 0x0041, 0x0042, 0x0043, 0x0048, 0x0049, 0x004a,
+        0x004b, 0x0800, 0x0801, 0x0802, 0x0803, 0x0808, 0x0809, 0x080a, 0x080b, 0x0840, 0x0841,
+        0x0842, 0x0843, 0x0848, 0x0849, 0x084a, 0x084b, 0x1000, 0x1001, 0x1002, 0x1003, 0x1008,
+        0x1009, 0x100a, 0x100b, 0x1040, 0x1041, 0x1042, 0x1043, 0x1048, 0x1049, 0x104a, 0x104b,
+        0x1800, 0x1801, 0x1802, 0x1803, 0x1808, 0x1809, 0x180a, 0x180b, 0x1840, 0x1841, 0x1842,
+        0x1843, 0x1848, 0x1849, 0x184a, 0x184b, 0x2000, 0x2001, 0x2002, 0x2003, 0x2008, 0x2009,
+        0x200a, 0x200b, 0x2040, 0x2041, 0x2042, 0x2043, 0x2048, 0x2049, 0x204a, 0x204b, 0x2800,
+        0x2801, 0x2802, 0x2803, 0x2808, 0x2809, 0x280a, 0x280b, 0x2840, 0x2841, 0x2842, 0x2843,
+        0x2848, 0x2849, 0x284a, 0x284b, 0x3000, 0x3001, 0x3002, 0x3003, 0x3008, 0x3009, 0x300a,
+        0x300b, 0x3040, 0x3041, 0x3042, 0x3043, 0x3048, 0x3049, 0x304a, 0x304b, 0x3800, 0x3801,
+        0x3802, 0x3803, 0x3808, 0x3809, 0x380a, 0x380b, 0x3840, 0x3841, 0x3842, 0x3843, 0x3848,
+        0x3849, 0x384a, 0x384b, 0x0000, 0x0001, 0x0002, 0x0003, 0x0008, 0x0009, 0x000a, 0x000b,
+        0x0040, 0x0041, 0x0042, 0x0043, 0x0048, 0x0049, 0x004a, 0x004b, 0x0800, 0x0801, 0x0802,
+        0x0803, 0x0808, 0x0809, 0x080a, 0x080b, 0x0840, 0x0841, 0x0842, 0x0843, 0x0848, 0x0849,
+        0x084a, 0x084b, 0x1000, 0x1001, 0x1002, 0x1003, 0x1008, 0x1009, 0x100a, 0x100b, 0x1040,
+        0x1041, 0x1042, 0x1043, 0x1048, 0x1049, 0x104a, 0x104b, 0x1800, 0x1801, 0x1802, 0x1803,
+        0x1808, 0x1809, 0x180a, 0x180b, 0x1840, 0x1841, 0x1842, 0x1843, 0x1848, 0x1849, 0x184a,
+        0x184b, 0x2000, 0x2001, 0x2002, 0x2003, 0x2008, 0x2009, 0x200a, 0x200b, 0x2040, 0x2041,
+        0x2042, 0x2043, 0x2048, 0x2049, 0x204a, 0x204b, 0x2800, 0x2801, 0x2802, 0x2803, 0x2808,
+        0x2809, 0x280a, 0x280b, 0x2840, 0x2841, 0x2842, 0x2843, 0x2848, 0x2849, 0x284a, 0x284b,
+        0x3000, 0x3001, 0x3002, 0x3003, 0x3008, 0x3009, 0x300a, 0x300b, 0x3040, 0x3041, 0x3042,
+        0x3043, 0x3048, 0x3049, 0x304a, 0x304b, 0x3800, 0x3801, 0x3802, 0x3803, 0x3808, 0x3809,
+        0x380a, 0x380b, 0x3840, 0x3841, 0x3842, 0x3843, 0x3848, 0x3849, 0x384a, 0x384b,
+    };
+    static constexpr u32 ylut[] = {
+        0x0000, 0x0004, 0x0010, 0x0014, 0x0020, 0x0024, 0x0030, 0x0034, 0x0080, 0x0084, 0x0090,
+        0x0094, 0x00a0, 0x00a4, 0x00b0, 0x00b4, 0x0100, 0x0104, 0x0110, 0x0114, 0x0120, 0x0124,
+        0x0130, 0x0134, 0x0180, 0x0184, 0x0190, 0x0194, 0x01a0, 0x01a4, 0x01b0, 0x01b4, 0x0200,
+        0x0204, 0x0210, 0x0214, 0x0220, 0x0224, 0x0230, 0x0234, 0x0280, 0x0284, 0x0290, 0x0294,
+        0x02a0, 0x02a4, 0x02b0, 0x02b4, 0x0300, 0x0304, 0x0310, 0x0314, 0x0320, 0x0324, 0x0330,
+        0x0334, 0x0380, 0x0384, 0x0390, 0x0394, 0x03a0, 0x03a4, 0x03b0, 0x03b4, 0x0400, 0x0404,
+        0x0410, 0x0414, 0x0420, 0x0424, 0x0430, 0x0434, 0x0480, 0x0484, 0x0490, 0x0494, 0x04a0,
+        0x04a4, 0x04b0, 0x04b4, 0x0500, 0x0504, 0x0510, 0x0514, 0x0520, 0x0524, 0x0530, 0x0534,
+        0x0580, 0x0584, 0x0590, 0x0594, 0x05a0, 0x05a4, 0x05b0, 0x05b4, 0x0600, 0x0604, 0x0610,
+        0x0614, 0x0620, 0x0624, 0x0630, 0x0634, 0x0680, 0x0684, 0x0690, 0x0694, 0x06a0, 0x06a4,
+        0x06b0, 0x06b4, 0x0700, 0x0704, 0x0710, 0x0714, 0x0720, 0x0724, 0x0730, 0x0734, 0x0780,
+        0x0784, 0x0790, 0x0794, 0x07a0, 0x07a4, 0x07b0, 0x07b4, 0x0000, 0x0004, 0x0010, 0x0014,
+        0x0020, 0x0024, 0x0030, 0x0034, 0x0080, 0x0084, 0x0090, 0x0094, 0x00a0, 0x00a4, 0x00b0,
+        0x00b4, 0x0100, 0x0104, 0x0110, 0x0114, 0x0120, 0x0124, 0x0130, 0x0134, 0x0180, 0x0184,
+        0x0190, 0x0194, 0x01a0, 0x01a4, 0x01b0, 0x01b4, 0x0200, 0x0204, 0x0210, 0x0214, 0x0220,
+        0x0224, 0x0230, 0x0234, 0x0280, 0x0284, 0x0290, 0x0294, 0x02a0, 0x02a4, 0x02b0, 0x02b4,
+        0x0300, 0x0304, 0x0310, 0x0314, 0x0320, 0x0324, 0x0330, 0x0334, 0x0380, 0x0384, 0x0390,
+        0x0394, 0x03a0, 0x03a4, 0x03b0, 0x03b4, 0x0400, 0x0404, 0x0410, 0x0414, 0x0420, 0x0424,
+        0x0430, 0x0434, 0x0480, 0x0484, 0x0490, 0x0494, 0x04a0, 0x04a4, 0x04b0, 0x04b4, 0x0500,
+        0x0504, 0x0510, 0x0514, 0x0520, 0x0524, 0x0530, 0x0534, 0x0580, 0x0584, 0x0590, 0x0594,
+        0x05a0, 0x05a4, 0x05b0, 0x05b4, 0x0600, 0x0604, 0x0610, 0x0614, 0x0620, 0x0624, 0x0630,
+        0x0634, 0x0680, 0x0684, 0x0690, 0x0694, 0x06a0, 0x06a4, 0x06b0, 0x06b4, 0x0700, 0x0704,
+        0x0710, 0x0714, 0x0720, 0x0724, 0x0730, 0x0734, 0x0780, 0x0784, 0x0790, 0x0794, 0x07a0,
+        0x07a4, 0x07b0, 0x07b4, 0x0000, 0x0004, 0x0010, 0x0014, 0x0020, 0x0024, 0x0030, 0x0034,
+        0x0080, 0x0084, 0x0090, 0x0094, 0x00a0, 0x00a4, 0x00b0, 0x00b4, 0x0100, 0x0104, 0x0110,
+        0x0114, 0x0120, 0x0124, 0x0130, 0x0134, 0x0180, 0x0184, 0x0190, 0x0194, 0x01a0, 0x01a4,
+        0x01b0, 0x01b4, 0x0200, 0x0204, 0x0210, 0x0214, 0x0220, 0x0224, 0x0230, 0x0234, 0x0280,
+        0x0284, 0x0290, 0x0294, 0x02a0, 0x02a4, 0x02b0, 0x02b4, 0x0300, 0x0304, 0x0310, 0x0314,
+        0x0320, 0x0324, 0x0330, 0x0334, 0x0380, 0x0384, 0x0390, 0x0394, 0x03a0, 0x03a4, 0x03b0,
+        0x03b4, 0x0400, 0x0404, 0x0410, 0x0414, 0x0420, 0x0424, 0x0430, 0x0434, 0x0480, 0x0484,
+        0x0490, 0x0494, 0x04a0, 0x04a4, 0x04b0, 0x04b4, 0x0500, 0x0504, 0x0510, 0x0514, 0x0520,
+        0x0524, 0x0530, 0x0534, 0x0580, 0x0584, 0x0590, 0x0594, 0x05a0, 0x05a4, 0x05b0, 0x05b4,
+        0x0600, 0x0604, 0x0610, 0x0614, 0x0620, 0x0624, 0x0630, 0x0634, 0x0680, 0x0684, 0x0690,
+        0x0694, 0x06a0, 0x06a4, 0x06b0, 0x06b4, 0x0700, 0x0704, 0x0710, 0x0714, 0x0720, 0x0724,
+        0x0730, 0x0734, 0x0780, 0x0784, 0x0790, 0x0794, 0x07a0, 0x07a4, 0x07b0, 0x07b4,
+    };
+    return xlut[x % 128] + ylut[y % 128];
+}
+
+static inline u32 GetMortonOffset128(u32 x, u32 y, u32 bytes_per_pixel) {
+    // Calculates the offset of the position of the pixel in Morton order
+    // Framebuffer images are split into 128x128 tiles.
+
+    const unsigned int block_height = 128;
+    const unsigned int coarse_x = x & ~127;
+
+    u32 i = MortonInterleave128(x, y);
+
+    const unsigned int offset = coarse_x * block_height;
+
+    return (i + offset) * bytes_per_pixel;
+}
+
+static inline void MortonCopyPixels128(u32 width, u32 height, u32 bytes_per_pixel,
+                                       u32 gl_bytes_per_pixel, u8* morton_data, u8* gl_data,
+                                       bool morton_to_gl) {
+    u8* data_ptrs[2];
+    for (unsigned y = 0; y < height; ++y) {
+        for (unsigned x = 0; x < width; ++x) {
+            const u32 coarse_y = y & ~127;
+            u32 morton_offset =
+                GetMortonOffset128(x, y, bytes_per_pixel) + coarse_y * width * bytes_per_pixel;
+            u32 gl_pixel_index = (x + (height - 1 - y) * width) * gl_bytes_per_pixel;
+
+            data_ptrs[morton_to_gl] = morton_data + morton_offset;
+            data_ptrs[!morton_to_gl] = &gl_data[gl_pixel_index];
+
+            memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel);
+        }
+    }
+}
+
 } // namespace VideoCore

src/video_core/video_core.cpp

@@ -26,7 +26,7 @@ bool Init(EmuWindow* emu_window) {
     if (g_renderer->Init()) {
         LOG_DEBUG(Render, "initialized OK");
     } else {
-        LOG_ERROR(Render, "initialization failed !");
+        LOG_CRITICAL(Render, "initialization failed !");
         return false;
     }
     return true;

src/video_core/video_core.h

@@ -15,6 +15,8 @@ class RendererBase;
 
 namespace VideoCore {
 
+enum class Renderer { Software, OpenGL };
+
 extern std::unique_ptr<RendererBase> g_renderer; ///< Renderer plugin
 extern EmuWindow* g_emu_window;                  ///< Emu window
 

src/yuzu/CMakeLists.txt

@@ -23,6 +23,13 @@ add_executable(yuzu
     configuration/configure_input.h
     configuration/configure_system.cpp
     configuration/configure_system.h
+    debugger/graphics/graphics_breakpoint_observer.cpp
+    debugger/graphics/graphics_breakpoint_observer.h
+    debugger/graphics/graphics_breakpoints.cpp
+    debugger/graphics/graphics_breakpoints.h
+    debugger/graphics/graphics_breakpoints_p.h
+    debugger/graphics/graphics_surface.cpp
+    debugger/graphics/graphics_surface.h
     debugger/profiler.cpp
     debugger/profiler.h
     debugger/registers.cpp

src/yuzu/configuration/config.cpp

@@ -77,8 +77,7 @@ void Config::ReadValues() {
     qt_config->endGroup();
 
     qt_config->beginGroup("Core");
-    Settings::values.cpu_core =
-        static_cast<Settings::CpuCore>(qt_config->value("cpu_core", 1).toInt());
+    Settings::values.use_cpu_jit = qt_config->value("use_cpu_jit", true).toBool();
     qt_config->endGroup();
 
     qt_config->beginGroup("Renderer");
@@ -94,6 +93,10 @@ void Config::ReadValues() {
     Settings::values.use_virtual_sd = qt_config->value("use_virtual_sd", true).toBool();
     qt_config->endGroup();
 
+    qt_config->beginGroup("System");
+    Settings::values.use_docked_mode = qt_config->value("use_docked_mode", true).toBool();
+    qt_config->endGroup();
+
     qt_config->beginGroup("Miscellaneous");
     Settings::values.log_filter = qt_config->value("log_filter", "*:Info").toString().toStdString();
     qt_config->endGroup();
@@ -171,7 +174,7 @@ void Config::SaveValues() {
     qt_config->endGroup();
 
     qt_config->beginGroup("Core");
-    qt_config->setValue("cpu_core", static_cast<int>(Settings::values.cpu_core));
+    qt_config->setValue("use_cpu_jit", Settings::values.use_cpu_jit);
     qt_config->endGroup();
 
     qt_config->beginGroup("Renderer");
@@ -188,6 +191,10 @@ void Config::SaveValues() {
     qt_config->setValue("use_virtual_sd", Settings::values.use_virtual_sd);
     qt_config->endGroup();
 
+    qt_config->beginGroup("System");
+    qt_config->setValue("use_docked_mode", Settings::values.use_docked_mode);
+    qt_config->endGroup();
+
     qt_config->beginGroup("Miscellaneous");
     qt_config->setValue("log_filter", QString::fromStdString(Settings::values.log_filter));
     qt_config->endGroup();

src/yuzu/configuration/configure_general.cpp

@@ -15,7 +15,8 @@ ConfigureGeneral::ConfigureGeneral(QWidget* parent)
 
     this->setConfiguration();
 
-    ui->cpu_core_combobox->setEnabled(!Core::System::GetInstance().IsPoweredOn());
+    ui->use_cpu_jit->setEnabled(!Core::System::GetInstance().IsPoweredOn());
+    ui->use_docked_mode->setEnabled(!Core::System::GetInstance().IsPoweredOn());
 }
 
 ConfigureGeneral::~ConfigureGeneral() {}
@@ -23,13 +24,14 @@ ConfigureGeneral::~ConfigureGeneral() {}
 void ConfigureGeneral::setConfiguration() {
     ui->toggle_deepscan->setChecked(UISettings::values.gamedir_deepscan);
     ui->toggle_check_exit->setChecked(UISettings::values.confirm_before_closing);
-    ui->cpu_core_combobox->setCurrentIndex(static_cast<int>(Settings::values.cpu_core));
+    ui->use_cpu_jit->setChecked(Settings::values.use_cpu_jit);
+    ui->use_docked_mode->setChecked(Settings::values.use_docked_mode);
 }
 
 void ConfigureGeneral::applyConfiguration() {
     UISettings::values.gamedir_deepscan = ui->toggle_deepscan->isChecked();
     UISettings::values.confirm_before_closing = ui->toggle_check_exit->isChecked();
-    Settings::values.cpu_core =
-        static_cast<Settings::CpuCore>(ui->cpu_core_combobox->currentIndex());
+    Settings::values.use_cpu_jit = ui->use_cpu_jit->isChecked();
+    Settings::values.use_docked_mode = ui->use_docked_mode->isChecked();
     Settings::Apply();
 }

src/yuzu/configuration/configure_general.ui

@@ -13,17 +13,17 @@
   <property name="windowTitle">
    <string>Form</string>
   </property>
-  <layout class="QHBoxLayout" name="horizontalLayout">
+  <layout class="QHBoxLayout" name="HorizontalLayout">
    <item>
-    <layout class="QVBoxLayout" name="verticalLayout">
+    <layout class="QVBoxLayout" name="VerticalLayout">
      <item>
-      <widget class="QGroupBox" name="groupBox">
+      <widget class="QGroupBox" name="GeneralGroupBox">
        <property name="title">
         <string>General</string>
        </property>
-       <layout class="QHBoxLayout" name="horizontalLayout_3">
+       <layout class="QHBoxLayout" name="GeneralHorizontalLayout">
         <item>
-         <layout class="QVBoxLayout" name="verticalLayout_2">
+         <layout class="QVBoxLayout" name="GeneralVerticalLayout">
           <item>
            <widget class="QCheckBox" name="toggle_deepscan">
             <property name="text">
@@ -44,40 +44,53 @@
       </widget>
      </item>
      <item>
-       <widget class="QGroupBox" name="groupBox_2">
-         <property name="title">
-           <string>CPU Core</string>
-         </property>
-         <layout class="QHBoxLayout" name="horizontalLayout_7">
-           <item>
-             <layout class="QVBoxLayout" name="verticalLayout_5">
-               <item>
-                 <widget class="QComboBox" name="cpu_core_combobox">
-                  <item>
-                   <property name="text">
-                    <string>Unicorn</string>
-                   </property>
-                  </item>
-                  <item>
-                   <property name="text">
-                    <string>Dynarmic</string>
-                   </property>
-                  </item>
-                 </widget>
-               </item>
-             </layout>
-           </item>
+      <widget class="QGroupBox" name="PerformanceGroupBox">
+       <property name="title">
+        <string>Performance</string>
+       </property>
+       <layout class="QHBoxLayout" name="PerformanceHorizontalLayout">
+        <item>
+         <layout class="QVBoxLayout" name="PerformanceVerticalLayout">
+          <item>
+           <widget class="QCheckBox" name="use_cpu_jit">
+            <property name="text">
+             <string>Enable CPU JIT</string>
+            </property>
+           </widget>
+          </item>
          </layout>
-       </widget>
+        </item>
+       </layout>
+      </widget>
      </item>
      <item>
-      <widget class="QGroupBox" name="groupBox_3">
+      <widget class="QGroupBox" name="EmulationGroupBox">
+       <property name="title">
+        <string>Emulation</string>
+       </property>
+        <layout class="QHBoxLayout" name="EmulationHorizontalLayout">
+          <item>
+            <layout class="QVBoxLayout" name="EmulationVerticalLayout">
+              <item>
+                <widget class="QCheckBox" name="use_docked_mode">
+                  <property name="text">
+                    <string>Enable docked mode</string>
+                  </property>
+                </widget>
+              </item>
+            </layout>
+          </item>
+        </layout>
+      </widget>
+     </item>
+     <item>
+      <widget class="QGroupBox" name="HotKeysGroupBox">
        <property name="title">
         <string>Hotkeys</string>
        </property>
-       <layout class="QHBoxLayout" name="horizontalLayout_4">
+       <layout class="QHBoxLayout" name="HotKeysHorizontalLayout">
         <item>
-         <layout class="QVBoxLayout" name="verticalLayout_4">
+         <layout class="QVBoxLayout" name="HotKeysVerticalLayout">
           <item>
            <widget class="GHotkeysDialog" name="widget" native="true"/>
           </item>

src/yuzu/debugger/graphics/graphics_breakpoint_observer.cpp

@@ -0,0 +1,27 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <QMetaType>
+#include "yuzu/debugger/graphics/graphics_breakpoint_observer.h"
+
+BreakPointObserverDock::BreakPointObserverDock(std::shared_ptr<Tegra::DebugContext> debug_context,
+                                               const QString& title, QWidget* parent)
+    : QDockWidget(title, parent), BreakPointObserver(debug_context) {
+    qRegisterMetaType<Tegra::DebugContext::Event>("Tegra::DebugContext::Event");
+
+    connect(this, SIGNAL(Resumed()), this, SLOT(OnResumed()));
+
+    // NOTE: This signal is emitted from a non-GUI thread, but connect() takes
+    //       care of delaying its handling to the GUI thread.
+    connect(this, SIGNAL(BreakPointHit(Tegra::DebugContext::Event, void*)), this,
+            SLOT(OnBreakPointHit(Tegra::DebugContext::Event, void*)), Qt::BlockingQueuedConnection);
+}
+
+void BreakPointObserverDock::OnMaxwellBreakPointHit(Tegra::DebugContext::Event event, void* data) {
+    emit BreakPointHit(event, data);
+}
+
+void BreakPointObserverDock::OnMaxwellResume() {
+    emit Resumed();
+}

src/yuzu/debugger/graphics/graphics_breakpoint_observer.h

@@ -0,0 +1,33 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <QDockWidget>
+#include "video_core/debug_utils/debug_utils.h"
+
+/**
+ * Utility class which forwards calls to OnMaxwellBreakPointHit and OnMaxwellResume to public slots.
+ * This is because the Maxwell breakpoint callbacks are called from a non-GUI thread, while
+ * the widget usually wants to perform reactions in the GUI thread.
+ */
+class BreakPointObserverDock : public QDockWidget,
+                               protected Tegra::DebugContext::BreakPointObserver {
+    Q_OBJECT
+
+public:
+    BreakPointObserverDock(std::shared_ptr<Tegra::DebugContext> debug_context, const QString& title,
+                           QWidget* parent = nullptr);
+
+    void OnMaxwellBreakPointHit(Tegra::DebugContext::Event event, void* data) override;
+    void OnMaxwellResume() override;
+
+private slots:
+    virtual void OnBreakPointHit(Tegra::DebugContext::Event event, void* data) = 0;
+    virtual void OnResumed() = 0;
+
+signals:
+    void Resumed();
+    void BreakPointHit(Tegra::DebugContext::Event event, void* data);
+};

src/yuzu/debugger/graphics/graphics_breakpoints.cpp

@@ -0,0 +1,212 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <QLabel>
+#include <QMetaType>
+#include <QPushButton>
+#include <QTreeView>
+#include <QVBoxLayout>
+#include "common/assert.h"
+#include "yuzu/debugger/graphics/graphics_breakpoints.h"
+#include "yuzu/debugger/graphics/graphics_breakpoints_p.h"
+
+BreakPointModel::BreakPointModel(std::shared_ptr<Tegra::DebugContext> debug_context,
+                                 QObject* parent)
+    : QAbstractListModel(parent), context_weak(debug_context),
+      at_breakpoint(debug_context->at_breakpoint),
+      active_breakpoint(debug_context->active_breakpoint) {}
+
+int BreakPointModel::columnCount(const QModelIndex& parent) const {
+    return 1;
+}
+
+int BreakPointModel::rowCount(const QModelIndex& parent) const {
+    return static_cast<int>(Tegra::DebugContext::Event::NumEvents);
+}
+
+QVariant BreakPointModel::data(const QModelIndex& index, int role) const {
+    const auto event = static_cast<Tegra::DebugContext::Event>(index.row());
+
+    switch (role) {
+    case Qt::DisplayRole: {
+        if (index.column() == 0) {
+            static const std::map<Tegra::DebugContext::Event, QString> map = {
+                {Tegra::DebugContext::Event::MaxwellCommandLoaded, tr("Maxwell command loaded")},
+                {Tegra::DebugContext::Event::MaxwellCommandProcessed,
+                 tr("Maxwell command processed")},
+                {Tegra::DebugContext::Event::IncomingPrimitiveBatch,
+                 tr("Incoming primitive batch")},
+                {Tegra::DebugContext::Event::FinishedPrimitiveBatch,
+                 tr("Finished primitive batch")},
+            };
+
+            DEBUG_ASSERT(map.size() == static_cast<size_t>(Tegra::DebugContext::Event::NumEvents));
+            return (map.find(event) != map.end()) ? map.at(event) : QString();
+        }
+
+        break;
+    }
+
+    case Qt::CheckStateRole: {
+        if (index.column() == 0)
+            return data(index, Role_IsEnabled).toBool() ? Qt::Checked : Qt::Unchecked;
+        break;
+    }
+
+    case Qt::BackgroundRole: {
+        if (at_breakpoint && index.row() == static_cast<int>(active_breakpoint)) {
+            return QBrush(QColor(0xE0, 0xE0, 0x10));
+        }
+        break;
+    }
+
+    case Role_IsEnabled: {
+        auto context = context_weak.lock();
+        return context && context->breakpoints[(int)event].enabled;
+    }
+
+    default:
+        break;
+    }
+    return QVariant();
+}
+
+Qt::ItemFlags BreakPointModel::flags(const QModelIndex& index) const {
+    if (!index.isValid())
+        return 0;
+
+    Qt::ItemFlags flags = Qt::ItemIsEnabled;
+    if (index.column() == 0)
+        flags |= Qt::ItemIsUserCheckable;
+    return flags;
+}
+
+bool BreakPointModel::setData(const QModelIndex& index, const QVariant& value, int role) {
+    const auto event = static_cast<Tegra::DebugContext::Event>(index.row());
+
+    switch (role) {
+    case Qt::CheckStateRole: {
+        if (index.column() != 0)
+            return false;
+
+        auto context = context_weak.lock();
+        if (!context)
+            return false;
+
+        context->breakpoints[(int)event].enabled = value == Qt::Checked;
+        QModelIndex changed_index = createIndex(index.row(), 0);
+        emit dataChanged(changed_index, changed_index);
+        return true;
+    }
+    }
+
+    return false;
+}
+
+void BreakPointModel::OnBreakPointHit(Tegra::DebugContext::Event event) {
+    auto context = context_weak.lock();
+    if (!context)
+        return;
+
+    active_breakpoint = context->active_breakpoint;
+    at_breakpoint = context->at_breakpoint;
+    emit dataChanged(createIndex(static_cast<int>(event), 0),
+                     createIndex(static_cast<int>(event), 0));
+}
+
+void BreakPointModel::OnResumed() {
+    auto context = context_weak.lock();
+    if (!context)
+        return;
+
+    at_breakpoint = context->at_breakpoint;
+    emit dataChanged(createIndex(static_cast<int>(active_breakpoint), 0),
+                     createIndex(static_cast<int>(active_breakpoint), 0));
+    active_breakpoint = context->active_breakpoint;
+}
+
+GraphicsBreakPointsWidget::GraphicsBreakPointsWidget(
+    std::shared_ptr<Tegra::DebugContext> debug_context, QWidget* parent)
+    : QDockWidget(tr("Maxwell Breakpoints"), parent), Tegra::DebugContext::BreakPointObserver(
+                                                          debug_context) {
+    setObjectName("TegraBreakPointsWidget");
+
+    status_text = new QLabel(tr("Emulation running"));
+    resume_button = new QPushButton(tr("Resume"));
+    resume_button->setEnabled(false);
+
+    breakpoint_model = new BreakPointModel(debug_context, this);
+    breakpoint_list = new QTreeView;
+    breakpoint_list->setRootIsDecorated(false);
+    breakpoint_list->setHeaderHidden(true);
+    breakpoint_list->setModel(breakpoint_model);
+
+    qRegisterMetaType<Tegra::DebugContext::Event>("Tegra::DebugContext::Event");
+
+    connect(breakpoint_list, SIGNAL(doubleClicked(const QModelIndex&)), this,
+            SLOT(OnItemDoubleClicked(const QModelIndex&)));
+
+    connect(resume_button, SIGNAL(clicked()), this, SLOT(OnResumeRequested()));
+
+    connect(this, SIGNAL(BreakPointHit(Tegra::DebugContext::Event, void*)), this,
+            SLOT(OnBreakPointHit(Tegra::DebugContext::Event, void*)), Qt::BlockingQueuedConnection);
+    connect(this, SIGNAL(Resumed()), this, SLOT(OnResumed()));
+
+    connect(this, SIGNAL(BreakPointHit(Tegra::DebugContext::Event, void*)), breakpoint_model,
+            SLOT(OnBreakPointHit(Tegra::DebugContext::Event)), Qt::BlockingQueuedConnection);
+    connect(this, SIGNAL(Resumed()), breakpoint_model, SLOT(OnResumed()));
+
+    connect(this, SIGNAL(BreakPointsChanged(const QModelIndex&, const QModelIndex&)),
+            breakpoint_model, SIGNAL(dataChanged(const QModelIndex&, const QModelIndex&)));
+
+    QWidget* main_widget = new QWidget;
+    auto main_layout = new QVBoxLayout;
+    {
+        auto sub_layout = new QHBoxLayout;
+        sub_layout->addWidget(status_text);
+        sub_layout->addWidget(resume_button);
+        main_layout->addLayout(sub_layout);
+    }
+    main_layout->addWidget(breakpoint_list);
+    main_widget->setLayout(main_layout);
+
+    setWidget(main_widget);
+}
+
+void GraphicsBreakPointsWidget::OnMaxwellBreakPointHit(Event event, void* data) {
+    // Process in GUI thread
+    emit BreakPointHit(event, data);
+}
+
+void GraphicsBreakPointsWidget::OnBreakPointHit(Tegra::DebugContext::Event event, void* data) {
+    status_text->setText(tr("Emulation halted at breakpoint"));
+    resume_button->setEnabled(true);
+}
+
+void GraphicsBreakPointsWidget::OnMaxwellResume() {
+    // Process in GUI thread
+    emit Resumed();
+}
+
+void GraphicsBreakPointsWidget::OnResumed() {
+    status_text->setText(tr("Emulation running"));
+    resume_button->setEnabled(false);
+}
+
+void GraphicsBreakPointsWidget::OnResumeRequested() {
+    if (auto context = context_weak.lock())
+        context->Resume();
+}
+
+void GraphicsBreakPointsWidget::OnItemDoubleClicked(const QModelIndex& index) {
+    if (!index.isValid())
+        return;
+
+    QModelIndex check_index = breakpoint_list->model()->index(index.row(), 0);
+    QVariant enabled = breakpoint_list->model()->data(check_index, Qt::CheckStateRole);
+    QVariant new_state = Qt::Unchecked;
+    if (enabled == Qt::Unchecked)
+        new_state = Qt::Checked;
+    breakpoint_list->model()->setData(check_index, new_state, Qt::CheckStateRole);
+}

src/yuzu/debugger/graphics/graphics_breakpoints.h

@@ -0,0 +1,46 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include <QDockWidget>
+#include "video_core/debug_utils/debug_utils.h"
+
+class QLabel;
+class QPushButton;
+class QTreeView;
+
+class BreakPointModel;
+
+class GraphicsBreakPointsWidget : public QDockWidget, Tegra::DebugContext::BreakPointObserver {
+    Q_OBJECT
+
+    using Event = Tegra::DebugContext::Event;
+
+public:
+    explicit GraphicsBreakPointsWidget(std::shared_ptr<Tegra::DebugContext> debug_context,
+                                       QWidget* parent = nullptr);
+
+    void OnMaxwellBreakPointHit(Tegra::DebugContext::Event event, void* data) override;
+    void OnMaxwellResume() override;
+
+public slots:
+    void OnBreakPointHit(Tegra::DebugContext::Event event, void* data);
+    void OnItemDoubleClicked(const QModelIndex&);
+    void OnResumeRequested();
+    void OnResumed();
+
+signals:
+    void Resumed();
+    void BreakPointHit(Tegra::DebugContext::Event event, void* data);
+    void BreakPointsChanged(const QModelIndex& topLeft, const QModelIndex& bottomRight);
+
+private:
+    QLabel* status_text;
+    QPushButton* resume_button;
+
+    BreakPointModel* breakpoint_model;
+    QTreeView* breakpoint_list;
+};

src/yuzu/debugger/graphics/graphics_breakpoints_p.h

@@ -0,0 +1,36 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include <QAbstractListModel>
+#include "video_core/debug_utils/debug_utils.h"
+
+class BreakPointModel : public QAbstractListModel {
+    Q_OBJECT
+
+public:
+    enum {
+        Role_IsEnabled = Qt::UserRole,
+    };
+
+    BreakPointModel(std::shared_ptr<Tegra::DebugContext> context, QObject* parent);
+
+    int columnCount(const QModelIndex& parent = QModelIndex()) const override;
+    int rowCount(const QModelIndex& parent = QModelIndex()) const override;
+    QVariant data(const QModelIndex& index, int role = Qt::DisplayRole) const override;
+    Qt::ItemFlags flags(const QModelIndex& index) const override;
+
+    bool setData(const QModelIndex& index, const QVariant& value, int role = Qt::EditRole) override;
+
+public slots:
+    void OnBreakPointHit(Tegra::DebugContext::Event event);
+    void OnResumed();
+
+private:
+    std::weak_ptr<Tegra::DebugContext> context_weak;
+    bool at_breakpoint;
+    Tegra::DebugContext::Event active_breakpoint;
+};

src/yuzu/debugger/graphics/graphics_surface.cpp

@@ -0,0 +1,451 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <QBoxLayout>
+#include <QComboBox>
+#include <QDebug>
+#include <QFileDialog>
+#include <QLabel>
+#include <QMouseEvent>
+#include <QPushButton>
+#include <QScrollArea>
+#include <QSpinBox>
+#include "core/core.h"
+#include "video_core/engines/maxwell_3d.h"
+#include "video_core/gpu.h"
+#include "video_core/textures/decoders.h"
+#include "video_core/textures/texture.h"
+#include "video_core/utils.h"
+#include "yuzu/debugger/graphics/graphics_surface.h"
+#include "yuzu/util/spinbox.h"
+
+static Tegra::Texture::TextureFormat ConvertToTextureFormat(
+    Tegra::RenderTargetFormat render_target_format) {
+    switch (render_target_format) {
+    case Tegra::RenderTargetFormat::RGBA8_UNORM:
+        return Tegra::Texture::TextureFormat::A8R8G8B8;
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented RT format");
+    }
+}
+
+SurfacePicture::SurfacePicture(QWidget* parent, GraphicsSurfaceWidget* surface_widget_)
+    : QLabel(parent), surface_widget(surface_widget_) {}
+SurfacePicture::~SurfacePicture() {}
+
+void SurfacePicture::mousePressEvent(QMouseEvent* event) {
+    // Only do something while the left mouse button is held down
+    if (!(event->buttons() & Qt::LeftButton))
+        return;
+
+    if (pixmap() == nullptr)
+        return;
+
+    if (surface_widget)
+        surface_widget->Pick(event->x() * pixmap()->width() / width(),
+                             event->y() * pixmap()->height() / height());
+}
+
+void SurfacePicture::mouseMoveEvent(QMouseEvent* event) {
+    // We also want to handle the event if the user moves the mouse while holding down the LMB
+    mousePressEvent(event);
+}
+
+GraphicsSurfaceWidget::GraphicsSurfaceWidget(std::shared_ptr<Tegra::DebugContext> debug_context,
+                                             QWidget* parent)
+    : BreakPointObserverDock(debug_context, tr("Maxwell Surface Viewer"), parent),
+      surface_source(Source::RenderTarget0) {
+    setObjectName("MaxwellSurface");
+
+    surface_source_list = new QComboBox;
+    surface_source_list->addItem(tr("Render Target 0"));
+    surface_source_list->addItem(tr("Render Target 1"));
+    surface_source_list->addItem(tr("Render Target 2"));
+    surface_source_list->addItem(tr("Render Target 3"));
+    surface_source_list->addItem(tr("Render Target 4"));
+    surface_source_list->addItem(tr("Render Target 5"));
+    surface_source_list->addItem(tr("Render Target 6"));
+    surface_source_list->addItem(tr("Render Target 7"));
+    surface_source_list->addItem(tr("Z Buffer"));
+    surface_source_list->addItem(tr("Custom"));
+    surface_source_list->setCurrentIndex(static_cast<int>(surface_source));
+
+    surface_address_control = new CSpinBox;
+    surface_address_control->SetBase(16);
+    surface_address_control->SetRange(0, 0x7FFFFFFFFFFFFFFF);
+    surface_address_control->SetPrefix("0x");
+
+    unsigned max_dimension = 16384; // TODO: Find actual maximum
+
+    surface_width_control = new QSpinBox;
+    surface_width_control->setRange(0, max_dimension);
+
+    surface_height_control = new QSpinBox;
+    surface_height_control->setRange(0, max_dimension);
+
+    surface_picker_x_control = new QSpinBox;
+    surface_picker_x_control->setRange(0, max_dimension - 1);
+
+    surface_picker_y_control = new QSpinBox;
+    surface_picker_y_control->setRange(0, max_dimension - 1);
+
+    surface_format_control = new QComboBox;
+
+    // Color formats sorted by Maxwell texture format index
+    surface_format_control->addItem(tr("None"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("A8R8G8B8"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("Unknown"));
+    surface_format_control->addItem(tr("DXT1"));
+    surface_format_control->addItem(tr("DXT23"));
+    surface_format_control->addItem(tr("DXT45"));
+    surface_format_control->addItem(tr("DXN1"));
+    surface_format_control->addItem(tr("DXN2"));
+
+    surface_info_label = new QLabel();
+    surface_info_label->setWordWrap(true);
+
+    surface_picture_label = new SurfacePicture(0, this);
+    surface_picture_label->setSizePolicy(QSizePolicy::Fixed, QSizePolicy::Fixed);
+    surface_picture_label->setAlignment(Qt::AlignLeft | Qt::AlignTop);
+    surface_picture_label->setScaledContents(false);
+
+    auto scroll_area = new QScrollArea();
+    scroll_area->setBackgroundRole(QPalette::Dark);
+    scroll_area->setWidgetResizable(false);
+    scroll_area->setWidget(surface_picture_label);
+
+    save_surface = new QPushButton(QIcon::fromTheme("document-save"), tr("Save"));
+
+    // Connections
+    connect(this, SIGNAL(Update()), this, SLOT(OnUpdate()));
+    connect(surface_source_list, SIGNAL(currentIndexChanged(int)), this,
+            SLOT(OnSurfaceSourceChanged(int)));
+    connect(surface_address_control, SIGNAL(ValueChanged(qint64)), this,
+            SLOT(OnSurfaceAddressChanged(qint64)));
+    connect(surface_width_control, SIGNAL(valueChanged(int)), this,
+            SLOT(OnSurfaceWidthChanged(int)));
+    connect(surface_height_control, SIGNAL(valueChanged(int)), this,
+            SLOT(OnSurfaceHeightChanged(int)));
+    connect(surface_format_control, SIGNAL(currentIndexChanged(int)), this,
+            SLOT(OnSurfaceFormatChanged(int)));
+    connect(surface_picker_x_control, SIGNAL(valueChanged(int)), this,
+            SLOT(OnSurfacePickerXChanged(int)));
+    connect(surface_picker_y_control, SIGNAL(valueChanged(int)), this,
+            SLOT(OnSurfacePickerYChanged(int)));
+    connect(save_surface, SIGNAL(clicked()), this, SLOT(SaveSurface()));
+
+    auto main_widget = new QWidget;
+    auto main_layout = new QVBoxLayout;
+    {
+        auto sub_layout = new QHBoxLayout;
+        sub_layout->addWidget(new QLabel(tr("Source:")));
+        sub_layout->addWidget(surface_source_list);
+        main_layout->addLayout(sub_layout);
+    }
+    {
+        auto sub_layout = new QHBoxLayout;
+        sub_layout->addWidget(new QLabel(tr("GPU Address:")));
+        sub_layout->addWidget(surface_address_control);
+        main_layout->addLayout(sub_layout);
+    }
+    {
+        auto sub_layout = new QHBoxLayout;
+        sub_layout->addWidget(new QLabel(tr("Width:")));
+        sub_layout->addWidget(surface_width_control);
+        main_layout->addLayout(sub_layout);
+    }
+    {
+        auto sub_layout = new QHBoxLayout;
+        sub_layout->addWidget(new QLabel(tr("Height:")));
+        sub_layout->addWidget(surface_height_control);
+        main_layout->addLayout(sub_layout);
+    }
+    {
+        auto sub_layout = new QHBoxLayout;
+        sub_layout->addWidget(new QLabel(tr("Format:")));
+        sub_layout->addWidget(surface_format_control);
+        main_layout->addLayout(sub_layout);
+    }
+    main_layout->addWidget(scroll_area);
+
+    auto info_layout = new QHBoxLayout;
+    {
+        auto xy_layout = new QVBoxLayout;
+        {
+            {
+                auto sub_layout = new QHBoxLayout;
+                sub_layout->addWidget(new QLabel(tr("X:")));
+                sub_layout->addWidget(surface_picker_x_control);
+                xy_layout->addLayout(sub_layout);
+            }
+            {
+                auto sub_layout = new QHBoxLayout;
+                sub_layout->addWidget(new QLabel(tr("Y:")));
+                sub_layout->addWidget(surface_picker_y_control);
+                xy_layout->addLayout(sub_layout);
+            }
+        }
+        info_layout->addLayout(xy_layout);
+        surface_info_label->setSizePolicy(QSizePolicy::Expanding, QSizePolicy::Minimum);
+        info_layout->addWidget(surface_info_label);
+    }
+    main_layout->addLayout(info_layout);
+
+    main_layout->addWidget(save_surface);
+    main_widget->setLayout(main_layout);
+    setWidget(main_widget);
+
+    // Load current data - TODO: Make sure this works when emulation is not running
+    if (debug_context && debug_context->at_breakpoint) {
+        emit Update();
+        widget()->setEnabled(debug_context->at_breakpoint);
+    } else {
+        widget()->setEnabled(false);
+    }
+}
+
+void GraphicsSurfaceWidget::OnBreakPointHit(Tegra::DebugContext::Event event, void* data) {
+    emit Update();
+    widget()->setEnabled(true);
+}
+
+void GraphicsSurfaceWidget::OnResumed() {
+    widget()->setEnabled(false);
+}
+
+void GraphicsSurfaceWidget::OnSurfaceSourceChanged(int new_value) {
+    surface_source = static_cast<Source>(new_value);
+    emit Update();
+}
+
+void GraphicsSurfaceWidget::OnSurfaceAddressChanged(qint64 new_value) {
+    if (surface_address != new_value) {
+        surface_address = static_cast<Tegra::GPUVAddr>(new_value);
+
+        surface_source_list->setCurrentIndex(static_cast<int>(Source::Custom));
+        emit Update();
+    }
+}
+
+void GraphicsSurfaceWidget::OnSurfaceWidthChanged(int new_value) {
+    if (surface_width != static_cast<unsigned>(new_value)) {
+        surface_width = static_cast<unsigned>(new_value);
+
+        surface_source_list->setCurrentIndex(static_cast<int>(Source::Custom));
+        emit Update();
+    }
+}
+
+void GraphicsSurfaceWidget::OnSurfaceHeightChanged(int new_value) {
+    if (surface_height != static_cast<unsigned>(new_value)) {
+        surface_height = static_cast<unsigned>(new_value);
+
+        surface_source_list->setCurrentIndex(static_cast<int>(Source::Custom));
+        emit Update();
+    }
+}
+
+void GraphicsSurfaceWidget::OnSurfaceFormatChanged(int new_value) {
+    if (surface_format != static_cast<Tegra::Texture::TextureFormat>(new_value)) {
+        surface_format = static_cast<Tegra::Texture::TextureFormat>(new_value);
+
+        surface_source_list->setCurrentIndex(static_cast<int>(Source::Custom));
+        emit Update();
+    }
+}
+
+void GraphicsSurfaceWidget::OnSurfacePickerXChanged(int new_value) {
+    if (surface_picker_x != new_value) {
+        surface_picker_x = new_value;
+        Pick(surface_picker_x, surface_picker_y);
+    }
+}
+
+void GraphicsSurfaceWidget::OnSurfacePickerYChanged(int new_value) {
+    if (surface_picker_y != new_value) {
+        surface_picker_y = new_value;
+        Pick(surface_picker_x, surface_picker_y);
+    }
+}
+
+void GraphicsSurfaceWidget::Pick(int x, int y) {
+    surface_picker_x_control->setValue(x);
+    surface_picker_y_control->setValue(y);
+
+    if (x < 0 || x >= static_cast<int>(surface_width) || y < 0 ||
+        y >= static_cast<int>(surface_height)) {
+        surface_info_label->setText(tr("Pixel out of bounds"));
+        surface_info_label->setAlignment(Qt::AlignLeft | Qt::AlignVCenter);
+        return;
+    }
+
+    surface_info_label->setText(QString("Raw: <Unimplemented>\n(%1)").arg("<Unimplemented>"));
+    surface_info_label->setAlignment(Qt::AlignLeft | Qt::AlignVCenter);
+}
+
+void GraphicsSurfaceWidget::OnUpdate() {
+    auto& gpu = Core::System::GetInstance().GPU();
+
+    QPixmap pixmap;
+
+    switch (surface_source) {
+    case Source::RenderTarget0:
+    case Source::RenderTarget1:
+    case Source::RenderTarget2:
+    case Source::RenderTarget3:
+    case Source::RenderTarget4:
+    case Source::RenderTarget5:
+    case Source::RenderTarget6:
+    case Source::RenderTarget7: {
+        // TODO: Store a reference to the registers in the debug context instead of accessing them
+        // directly...
+
+        auto& registers = gpu.Get3DEngine().regs;
+        auto& rt = registers.rt[static_cast<size_t>(surface_source) -
+                                static_cast<size_t>(Source::RenderTarget0)];
+
+        surface_address = rt.Address();
+        surface_width = rt.width;
+        surface_height = rt.height;
+        if (rt.format != Tegra::RenderTargetFormat::NONE) {
+            surface_format = ConvertToTextureFormat(rt.format);
+        }
+
+        break;
+    }
+
+    case Source::Custom: {
+        // Keep user-specified values
+        break;
+    }
+
+    default:
+        qDebug() << "Unknown surface source " << static_cast<int>(surface_source);
+        break;
+    }
+
+    surface_address_control->SetValue(surface_address);
+    surface_width_control->setValue(surface_width);
+    surface_height_control->setValue(surface_height);
+    surface_format_control->setCurrentIndex(static_cast<int>(surface_format));
+
+    if (surface_address == 0) {
+        surface_picture_label->hide();
+        surface_info_label->setText(tr("(invalid surface address)"));
+        surface_info_label->setAlignment(Qt::AlignCenter);
+        surface_picker_x_control->setEnabled(false);
+        surface_picker_y_control->setEnabled(false);
+        save_surface->setEnabled(false);
+        return;
+    }
+
+    // TODO: Implement a good way to visualize alpha components!
+
+    QImage decoded_image(surface_width, surface_height, QImage::Format_ARGB32);
+    VAddr address = gpu.memory_manager->PhysicalToVirtualAddress(surface_address);
+
+    auto unswizzled_data =
+        Tegra::Texture::UnswizzleTexture(address, surface_format, surface_width, surface_height);
+
+    auto texture_data = Tegra::Texture::DecodeTexture(unswizzled_data, surface_format,
+                                                      surface_width, surface_height);
+
+    surface_picture_label->show();
+
+    for (unsigned int y = 0; y < surface_height; ++y) {
+        for (unsigned int x = 0; x < surface_width; ++x) {
+            Math::Vec4<u8> color;
+            color[0] = texture_data[x + y * surface_width + 0];
+            color[1] = texture_data[x + y * surface_width + 1];
+            color[2] = texture_data[x + y * surface_width + 2];
+            color[3] = texture_data[x + y * surface_width + 3];
+            decoded_image.setPixel(x, y, qRgba(color.r(), color.g(), color.b(), color.a()));
+        }
+    }
+
+    pixmap = QPixmap::fromImage(decoded_image);
+    surface_picture_label->setPixmap(pixmap);
+    surface_picture_label->resize(pixmap.size());
+
+    // Update the info with pixel data
+    surface_picker_x_control->setEnabled(true);
+    surface_picker_y_control->setEnabled(true);
+    Pick(surface_picker_x, surface_picker_y);
+
+    // Enable saving the converted pixmap to file
+    save_surface->setEnabled(true);
+}
+
+void GraphicsSurfaceWidget::SaveSurface() {
+    QString png_filter = tr("Portable Network Graphic (*.png)");
+    QString bin_filter = tr("Binary data (*.bin)");
+
+    QString selectedFilter;
+    QString filename = QFileDialog::getSaveFileName(
+        this, tr("Save Surface"),
+        QString("texture-0x%1.png").arg(QString::number(surface_address, 16)),
+        QString("%1;;%2").arg(png_filter, bin_filter), &selectedFilter);
+
+    if (filename.isEmpty()) {
+        // If the user canceled the dialog, don't save anything.
+        return;
+    }
+
+    if (selectedFilter == png_filter) {
+        const QPixmap* pixmap = surface_picture_label->pixmap();
+        ASSERT_MSG(pixmap != nullptr, "No pixmap set");
+
+        QFile file(filename);
+        file.open(QIODevice::WriteOnly);
+        if (pixmap)
+            pixmap->save(&file, "PNG");
+    } else if (selectedFilter == bin_filter) {
+        auto& gpu = Core::System::GetInstance().GPU();
+        VAddr address = gpu.memory_manager->PhysicalToVirtualAddress(surface_address);
+
+        const u8* buffer = Memory::GetPointer(address);
+        ASSERT_MSG(buffer != nullptr, "Memory not accessible");
+
+        QFile file(filename);
+        file.open(QIODevice::WriteOnly);
+        int size = surface_width * surface_height * Tegra::Texture::BytesPerPixel(surface_format);
+        QByteArray data(reinterpret_cast<const char*>(buffer), size);
+        file.write(data);
+    } else {
+        UNREACHABLE_MSG("Unhandled filter selected");
+    }
+}

src/yuzu/debugger/graphics/graphics_surface.h

@@ -0,0 +1,97 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <QLabel>
+#include <QPushButton>
+#include "video_core/memory_manager.h"
+#include "video_core/textures/texture.h"
+#include "yuzu/debugger/graphics/graphics_breakpoint_observer.h"
+
+class QComboBox;
+class QSpinBox;
+class CSpinBox;
+
+class GraphicsSurfaceWidget;
+
+class SurfacePicture : public QLabel {
+    Q_OBJECT
+
+public:
+    explicit SurfacePicture(QWidget* parent = nullptr,
+                            GraphicsSurfaceWidget* surface_widget = nullptr);
+    ~SurfacePicture();
+
+protected slots:
+    virtual void mouseMoveEvent(QMouseEvent* event);
+    virtual void mousePressEvent(QMouseEvent* event);
+
+private:
+    GraphicsSurfaceWidget* surface_widget;
+};
+
+class GraphicsSurfaceWidget : public BreakPointObserverDock {
+    Q_OBJECT
+
+    using Event = Tegra::DebugContext::Event;
+
+    enum class Source {
+        RenderTarget0 = 0,
+        RenderTarget1 = 1,
+        RenderTarget2 = 2,
+        RenderTarget3 = 3,
+        RenderTarget4 = 4,
+        RenderTarget5 = 5,
+        RenderTarget6 = 6,
+        RenderTarget7 = 7,
+        ZBuffer = 8,
+        Custom = 9,
+    };
+
+public:
+    explicit GraphicsSurfaceWidget(std::shared_ptr<Tegra::DebugContext> debug_context,
+                                   QWidget* parent = nullptr);
+    void Pick(int x, int y);
+
+public slots:
+    void OnSurfaceSourceChanged(int new_value);
+    void OnSurfaceAddressChanged(qint64 new_value);
+    void OnSurfaceWidthChanged(int new_value);
+    void OnSurfaceHeightChanged(int new_value);
+    void OnSurfaceFormatChanged(int new_value);
+    void OnSurfacePickerXChanged(int new_value);
+    void OnSurfacePickerYChanged(int new_value);
+    void OnUpdate();
+
+private slots:
+    void OnBreakPointHit(Tegra::DebugContext::Event event, void* data) override;
+    void OnResumed() override;
+
+    void SaveSurface();
+
+signals:
+    void Update();
+
+private:
+    QComboBox* surface_source_list;
+    CSpinBox* surface_address_control;
+    QSpinBox* surface_width_control;
+    QSpinBox* surface_height_control;
+    QComboBox* surface_format_control;
+
+    SurfacePicture* surface_picture_label;
+    QSpinBox* surface_picker_x_control;
+    QSpinBox* surface_picker_y_control;
+    QLabel* surface_info_label;
+    QPushButton* save_surface;
+
+    Source surface_source;
+    Tegra::GPUVAddr surface_address;
+    unsigned surface_width;
+    unsigned surface_height;
+    Tegra::Texture::TextureFormat surface_format;
+    int surface_picker_x = 0;
+    int surface_picker_y = 0;
+};

src/yuzu/main.cpp

@@ -25,10 +25,13 @@
 #include "core/gdbstub/gdbstub.h"
 #include "core/loader/loader.h"
 #include "core/settings.h"
+#include "video_core/debug_utils/debug_utils.h"
 #include "yuzu/about_dialog.h"
 #include "yuzu/bootmanager.h"
 #include "yuzu/configuration/config.h"
 #include "yuzu/configuration/configure_dialog.h"
+#include "yuzu/debugger/graphics/graphics_breakpoints.h"
+#include "yuzu/debugger/graphics/graphics_surface.h"
 #include "yuzu/debugger/profiler.h"
 #include "yuzu/debugger/registers.h"
 #include "yuzu/debugger/wait_tree.h"
@@ -68,6 +71,9 @@ static void ShowCalloutMessage(const QString& message, CalloutFlag flag) {
 void GMainWindow::ShowCallouts() {}
 
 GMainWindow::GMainWindow() : config(new Config()), emu_thread(nullptr) {
+
+    debug_context = Tegra::DebugContext::Construct();
+
     setAcceptDrops(true);
     ui.setupUi(this);
     statusBar()->hide();
@@ -160,6 +166,16 @@ void GMainWindow::InitializeDebugWidgets() {
     connect(this, &GMainWindow::EmulationStopping, registersWidget,
             &RegistersWidget::OnEmulationStopping);
 
+    graphicsBreakpointsWidget = new GraphicsBreakPointsWidget(debug_context, this);
+    addDockWidget(Qt::RightDockWidgetArea, graphicsBreakpointsWidget);
+    graphicsBreakpointsWidget->hide();
+    debug_menu->addAction(graphicsBreakpointsWidget->toggleViewAction());
+
+    graphicsSurfaceWidget = new GraphicsSurfaceWidget(debug_context, this);
+    addDockWidget(Qt::RightDockWidgetArea, graphicsSurfaceWidget);
+    graphicsSurfaceWidget->hide();
+    debug_menu->addAction(graphicsSurfaceWidget->toggleViewAction());
+
     waitTreeWidget = new WaitTreeWidget(this);
     addDockWidget(Qt::LeftDockWidgetArea, waitTreeWidget);
     waitTreeWidget->hide();
@@ -324,6 +340,8 @@ bool GMainWindow::LoadROM(const QString& filename) {
 
     Core::System& system{Core::System::GetInstance()};
 
+    system.SetGPUDebugContext(debug_context);
+
     const Core::System::ResultStatus result{system.Load(render_window, filename.toStdString())};
 
     Core::Telemetry().AddField(Telemetry::FieldType::App, "Frontend", "Qt");

src/yuzu/main.h

@@ -2,8 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#ifndef _CITRA_QT_MAIN_HXX_
-#define _CITRA_QT_MAIN_HXX_
+#pragma once
 
 #include <memory>
 #include <QMainWindow>
@@ -15,17 +14,18 @@ class Config;
 class EmuThread;
 class GameList;
 class GImageInfo;
-class GPUCommandStreamWidget;
-class GPUCommandListWidget;
 class GraphicsBreakPointsWidget;
-class GraphicsTracingWidget;
-class GraphicsVertexShaderWidget;
+class GraphicsSurfaceWidget;
 class GRenderWindow;
 class MicroProfileDialog;
 class ProfilerWidget;
 class RegistersWidget;
 class WaitTreeWidget;
 
+namespace Tegra {
+class DebugContext;
+}
+
 class GMainWindow : public QMainWindow {
     Q_OBJECT
 
@@ -138,6 +138,8 @@ private:
 
     Ui::MainWindow ui;
 
+    std::shared_ptr<Tegra::DebugContext> debug_context;
+
     GRenderWindow* render_window;
     GameList* game_list;
 
@@ -150,7 +152,7 @@ private:
 
     std::unique_ptr<Config> config;
 
-    // Whether emulation is currently running in Citra.
+    // Whether emulation is currently running in yuzu.
     bool emulation_running = false;
     std::unique_ptr<EmuThread> emu_thread;
 
@@ -158,6 +160,8 @@ private:
     ProfilerWidget* profilerWidget;
     MicroProfileDialog* microProfileDialog;
     RegistersWidget* registersWidget;
+    GraphicsBreakPointsWidget* graphicsBreakpointsWidget;
+    GraphicsSurfaceWidget* graphicsSurfaceWidget;
     WaitTreeWidget* waitTreeWidget;
 
     QAction* actions_recent_files[max_recent_files_item];
@@ -167,5 +171,3 @@ protected:
     void dragEnterEvent(QDragEnterEvent* event) override;
     void dragMoveEvent(QDragMoveEvent* event) override;
 };
-
-#endif // _CITRA_QT_MAIN_HXX_

src/yuzu_cmd/config.cpp

@@ -90,8 +90,7 @@ void Config::ReadValues() {
         sdl2_config->Get("Controls", "touch_device", "engine:emu_window");
 
     // Core
-    Settings::values.cpu_core =
-        static_cast<Settings::CpuCore>(sdl2_config->GetInteger("Core", "cpu_core", 1));
+    Settings::values.use_cpu_jit = sdl2_config->GetBoolean("Core", "use_cpu_jit", true);
 
     // Renderer
     Settings::values.resolution_factor =
@@ -107,6 +106,9 @@ void Config::ReadValues() {
     Settings::values.use_virtual_sd =
         sdl2_config->GetBoolean("Data Storage", "use_virtual_sd", true);
 
+    // System
+    Settings::values.use_docked_mode = sdl2_config->GetBoolean("System", "use_docked_mode", true);
+
     // Miscellaneous
     Settings::values.log_filter = sdl2_config->Get("Miscellaneous", "log_filter", "*:Trace");
 

src/yuzu_cmd/default_ini.h

@@ -76,9 +76,9 @@ motion_device=
 touch_device=
 
 [Core]
-# Which CPU core to use for CPU emulation
-# 0: Unicorn (slow), 1 (default): Dynarmic (faster)
-cpu_core =
+# Whether to use the Just-In-Time (JIT) compiler for CPU emulation
+# 0: Interpreter (slow), 1 (default): JIT (fast)
+use_cpu_jit =
 
 [Renderer]
 # Whether to use software or hardware rendering.
@@ -154,9 +154,9 @@ output_device =
 use_virtual_sd =
 
 [System]
-# The system model that Citra will try to emulate
-# 0: Old 3DS (default), 1: New 3DS
-is_new_3ds =
+# Whether the system is docked
+# 1 (default): Yes, 0: No
+use_docked_mode =
 
 # The system region that Citra will use during emulation
 # -1: Auto-select (default), 0: Japan, 1: USA, 2: Europe, 3: Australia, 4: China, 5: Korea, 6: Taiwan

src/yuzu_cmd/yuzu.rc

@@ -6,7 +6,7 @@
 
 // Icon with lowest ID value placed first to ensure application icon
 // remains consistent on all systems.
-CITRA_ICON              ICON                    "../../dist/yuzu.ico"
+YUZU_ICON               ICON                    "../../dist/yuzu.ico"
 
 
 /////////////////////////////////////////////////////////////////////////////