Merge c227320b4e into b3298465cf

Updates CMakeLists to use Qt 5.10.0 instead of Qt 5.7

.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
+}

CMakeLists.txt

@@ -278,7 +278,7 @@ endif()
 if (ENABLE_QT)
     if (YUZU_USE_BUNDLED_QT)
         if (MSVC14 AND ARCHITECTURE_x86_64)
-            set(QT_VER qt-5.7-msvc2015_64)
+            set(QT_VER qt-5.10.0-msvc2015_64)
         else()
             message(FATAL_ERROR "No bundled Qt binaries for your toolchain. Disable YUZU_USE_BUNDLED_QT and provide your own.")
         endif()

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

dist/icons/icons.qrc

@@ -1,5 +0,0 @@
-<RCC>
-  <qresource prefix="icons">
-    <file>yuzu.png</file>
-  </qresource>
-</RCC>

dist/qt_themes/default/default.qrc

@@ -0,0 +1,11 @@
+<RCC>
+    <qresource prefix="icons/default">
+        <file alias="index.theme">icons/index.theme</file>
+      
+        <file alias="16x16/checked.png">icons/16x16/checked.png</file>
+
+        <file alias="16x16/failed.png">icons/16x16/failed.png</file>
+      
+        <file alias="256x256/yuzu.png">icons/256x256/yuzu.png</file>
+    </qresource>
+</RCC>

dist/qt_themes/default/icons/index.theme

@@ -0,0 +1,10 @@
+[Icon Theme]
+Name=default
+Comment=default theme
+Directories=16x16,256x256
+ 
+[16x16]
+Size=16
+ 
+[256x256]
+Size=256

dist/qt_themes/qdarkstyle/icons/index.theme

@@ -0,0 +1,11 @@
+[Icon Theme]
+Name=qdarkstyle
+Comment=dark theme
+Inherits=default
+Directories=16x16,256x256
+ 
+[16x16]
+Size=16
+ 
+[256x256]
+Size=256

dist/qt_themes/qdarkstyle/style.qrc

@@ -0,0 +1,49 @@
+<RCC>
+  <qresource prefix="icons/qdarkstyle">
+    <file alias="index.theme">icons/index.theme</file>
+  </qresource>
+  <qresource prefix="qss_icons">
+    <file>rc/up_arrow_disabled.png</file>
+    <file>rc/Hmovetoolbar.png</file>
+    <file>rc/stylesheet-branch-end.png</file>
+    <file>rc/branch_closed-on.png</file>
+    <file>rc/stylesheet-vline.png</file>
+    <file>rc/branch_closed.png</file>
+    <file>rc/branch_open-on.png</file>
+    <file>rc/transparent.png</file>
+    <file>rc/right_arrow_disabled.png</file>
+    <file>rc/sizegrip.png</file>
+    <file>rc/close.png</file>
+    <file>rc/close-hover.png</file>
+    <file>rc/close-pressed.png</file>
+    <file>rc/down_arrow.png</file>
+    <file>rc/Vmovetoolbar.png</file>
+    <file>rc/left_arrow.png</file>
+    <file>rc/stylesheet-branch-more.png</file>
+    <file>rc/up_arrow.png</file>
+    <file>rc/right_arrow.png</file>
+    <file>rc/left_arrow_disabled.png</file>
+    <file>rc/Hsepartoolbar.png</file>
+    <file>rc/branch_open.png</file>
+    <file>rc/Vsepartoolbar.png</file>
+    <file>rc/down_arrow_disabled.png</file>
+    <file>rc/undock.png</file>
+    <file>rc/checkbox_checked_disabled.png</file>
+    <file>rc/checkbox_checked_focus.png</file>
+    <file>rc/checkbox_checked.png</file>
+    <file>rc/checkbox_indeterminate.png</file>
+    <file>rc/checkbox_indeterminate_focus.png</file>
+    <file>rc/checkbox_unchecked_disabled.png</file>
+    <file>rc/checkbox_unchecked_focus.png</file>
+    <file>rc/checkbox_unchecked.png</file>
+    <file>rc/radio_checked_disabled.png</file>
+    <file>rc/radio_checked_focus.png</file>
+    <file>rc/radio_checked.png</file>
+    <file>rc/radio_unchecked_disabled.png</file>
+    <file>rc/radio_unchecked_focus.png</file>
+    <file>rc/radio_unchecked.png</file>
+  </qresource>
+  <qresource prefix="qdarkstyle">
+      <file>style.qss</file>
+  </qresource>
+</RCC>

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

@@ -42,6 +42,7 @@ namespace Log {
     SUB(Service, FS)                                                                               \
     SUB(Service, HID)                                                                              \
     SUB(Service, LM)                                                                               \
+    SUB(Service, NFP)                                                                              \
     SUB(Service, NIFM)                                                                             \
     SUB(Service, NS)                                                                               \
     SUB(Service, NVDRV)                                                                            \

src/common/logging/log.h

@@ -59,6 +59,7 @@ enum class Class : ClassType {
     Service_FS,        ///< The FS (Filesystem) service
     Service_HID,       ///< The HID (Human interface device) service
     Service_LM,        ///< The LM (Logger) service
+    Service_NFP,       ///< The NFP service
     Service_NIFM,      ///< The NIFM (Network interface) service
     Service_NS,        ///< The NS services
     Service_NVDRV,     ///< The NVDRV (Nvidia driver) service
@@ -85,7 +86,7 @@ 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

@@ -142,6 +142,10 @@ add_library(core STATIC
     hle/service/nifm/nifm_s.h
     hle/service/nifm/nifm_u.cpp
     hle/service/nifm/nifm_u.h
+    hle/service/nfp/nfp.cpp
+    hle/service/nfp/nfp.h
+    hle/service/nfp/nfp_user.cpp
+    hle/service/nfp/nfp_user.h
     hle/service/ns/ns.cpp
     hle/service/ns/ns.h
     hle/service/ns/pl_u.cpp

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/hle/kernel/svc.cpp

@@ -756,8 +756,16 @@ static ResultCode CreateTransferMemory(Handle* handle, VAddr addr, u64 size, u32
     return RESULT_SUCCESS;
 }
 
-static ResultCode SetThreadCoreMask(u64, u64, u64) {
-    LOG_WARNING(Kernel_SVC, "(STUBBED) called");
+static ResultCode GetThreadCoreMask(Handle handle, u32* mask, u64* unknown) {
+    LOG_WARNING(Kernel_SVC, "(STUBBED) called, handle=0x%08X", handle);
+    *mask = 0x0;
+    *unknown = 0xf;
+    return RESULT_SUCCESS;
+}
+
+static ResultCode SetThreadCoreMask(Handle handle, u32 mask, u64 unknown) {
+    LOG_WARNING(Kernel_SVC, "(STUBBED) called, handle=0x%08X, mask=0x%08X, unknown=0x%lx", handle,
+                mask, unknown);
     return RESULT_SUCCESS;
 }
 
@@ -809,7 +817,7 @@ static const FunctionDef SVC_Table[] = {
     {0x0B, SvcWrap<SleepThread>, "SleepThread"},
     {0x0C, SvcWrap<GetThreadPriority>, "GetThreadPriority"},
     {0x0D, SvcWrap<SetThreadPriority>, "SetThreadPriority"},
-    {0x0E, nullptr, "GetThreadCoreMask"},
+    {0x0E, SvcWrap<GetThreadCoreMask>, "GetThreadCoreMask"},
     {0x0F, SvcWrap<SetThreadCoreMask>, "SetThreadCoreMask"},
     {0x10, SvcWrap<GetCurrentProcessorNumber>, "GetCurrentProcessorNumber"},
     {0x11, nullptr, "SignalEvent"},

src/core/hle/kernel/svc_wrap.h

@@ -70,6 +70,21 @@ void SvcWrap() {
     FuncReturn(retval);
 }
 
+template <ResultCode func(u32, u32, u64)>
+void SvcWrap() {
+    FuncReturn(func((u32)(PARAM(0) & 0xFFFFFFFF), (u32)(PARAM(1) & 0xFFFFFFFF), PARAM(2)).raw);
+}
+
+template <ResultCode func(u32, u32*, u64*)>
+void SvcWrap() {
+    u32 param_1 = 0;
+    u64 param_2 = 0;
+    ResultCode retval = func((u32)(PARAM(2) & 0xFFFFFFFF), &param_1, &param_2);
+    Core::CPU().SetReg(1, param_1);
+    Core::CPU().SetReg(2, param_2);
+    FuncReturn(retval.raw);
+}
+
 template <ResultCode func(u64, u64, u32, u32)>
 void SvcWrap() {
     FuncReturn(

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

@@ -59,12 +59,12 @@ private:
         AudioRendererResponseData response_data{};
 
         response_data.section_0_size =
-            response_data.state_entries.size() * sizeof(AudioRendererStateEntry);
-        response_data.section_1_size = response_data.section_1.size();
-        response_data.section_2_size = response_data.section_2.size();
-        response_data.section_3_size = response_data.section_3.size();
-        response_data.section_4_size = response_data.section_4.size();
-        response_data.section_5_size = response_data.section_5.size();
+            static_cast<u32>(response_data.state_entries.size() * sizeof(AudioRendererStateEntry));
+        response_data.section_1_size = static_cast<u32>(response_data.section_1.size());
+        response_data.section_2_size = static_cast<u32>(response_data.section_2.size());
+        response_data.section_3_size = static_cast<u32>(response_data.section_3.size());
+        response_data.section_4_size = static_cast<u32>(response_data.section_4.size());
+        response_data.section_5_size = static_cast<u32>(response_data.section_5.size());
         response_data.total_size = sizeof(AudioRendererResponseData);
 
         for (unsigned i = 0; i < response_data.state_entries.size(); i++) {
@@ -151,12 +151,80 @@ 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"},
+            {0x2, nullptr, "GetAudioDeviceOutputVolume"},
+            {0x3, nullptr, "GetActiveAudioDeviceName"},
+            {0x4, &IAudioDevice::QueryAudioDeviceSystemEvent, "QueryAudioDeviceSystemEvent"},
+            {0x5, &IAudioDevice::GetActiveChannelCount, "GetActiveChannelCount"},
+            {0x6, nullptr, "ListAudioDeviceNameAuto"},
+            {0x7, nullptr, "SetAudioDeviceOutputVolumeAuto"},
+            {0x8, nullptr, "GetAudioDeviceOutputVolumeAuto"},
+            {0x10, nullptr, "GetActiveAudioDeviceNameAuto"},
+            {0x11, nullptr, "QueryAudioDeviceInputEvent"},
+            {0x12, nullptr, "QueryAudioDeviceOutputEvent"}};
+        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);
+    }
+
+    void QueryAudioDeviceSystemEvent(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_Audio, "(STUBBED) called");
+
+        buffer_event->Signal();
+
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(buffer_event);
+    }
+
+    void GetActiveChannelCount(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_Audio, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(1);
+    }
+
+    Kernel::SharedPtr<Kernel::Event> buffer_event;
+
+}; // namespace Audio
+
 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 +247,13 @@ 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/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/nfp/nfp.cpp

@@ -0,0 +1,28 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/logging/log.h"
+#include "core/hle/ipc_helpers.h"
+#include "core/hle/service/nfp/nfp.h"
+#include "core/hle/service/nfp/nfp_user.h"
+
+namespace Service {
+namespace NFP {
+
+Module::Interface::Interface(std::shared_ptr<Module> module, const char* name)
+    : ServiceFramework(name), module(std::move(module)) {}
+
+void Module::Interface::Unknown(Kernel::HLERequestContext& ctx) {
+    LOG_WARNING(Service_NFP, "(STUBBED) called");
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+}
+
+void InstallInterfaces(SM::ServiceManager& service_manager) {
+    auto module = std::make_shared<Module>();
+    std::make_shared<NFP_User>(module)->InstallAsService(service_manager);
+}
+
+} // namespace NFP
+} // namespace Service

src/core/hle/service/nfp/nfp.h

@@ -0,0 +1,28 @@
+// 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 NFP {
+
+class Module final {
+public:
+    class Interface : public ServiceFramework<Interface> {
+    public:
+        Interface(std::shared_ptr<Module> module, const char* name);
+
+        void Unknown(Kernel::HLERequestContext& ctx);
+
+    protected:
+        std::shared_ptr<Module> module;
+    };
+};
+
+void InstallInterfaces(SM::ServiceManager& service_manager);
+
+} // namespace NFP
+} // namespace Service

src/core/hle/service/nfp/nfp_user.cpp

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

src/core/hle/service/nfp/nfp_user.h

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

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/service.cpp

@@ -25,6 +25,7 @@
 #include "core/hle/service/friend/friend.h"
 #include "core/hle/service/hid/hid.h"
 #include "core/hle/service/lm/lm.h"
+#include "core/hle/service/nfp/nfp.h"
 #include "core/hle/service/nifm/nifm.h"
 #include "core/hle/service/ns/ns.h"
 #include "core/hle/service/nvdrv/nvdrv.h"
@@ -187,6 +188,7 @@ void Init() {
     Friend::InstallInterfaces(*SM::g_service_manager);
     HID::InstallInterfaces(*SM::g_service_manager);
     LM::InstallInterfaces(*SM::g_service_manager);
+    NFP::InstallInterfaces(*SM::g_service_manager);
     NIFM::InstallInterfaces(*SM::g_service_manager);
     NS::InstallInterfaces(*SM::g_service_manager);
     Nvidia::InstallInterfaces(*SM::g_service_manager);

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"
@@ -115,91 +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;
 }
 
@@ -217,7 +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;
 }
 
@@ -291,6 +325,58 @@ 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
@@ -344,17 +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;
@@ -364,21 +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]);
@@ -387,6 +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();
         }
@@ -418,17 +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;
@@ -437,20 +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]);
 
@@ -458,6 +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();
         }
@@ -473,9 +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;
@@ -483,27 +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();
         }
@@ -514,37 +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:
@@ -559,78 +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

@@ -254,6 +254,11 @@ enum class FlushMode {
     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.

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

@@ -11,7 +11,10 @@
 #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 {
@@ -21,12 +24,6 @@ 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 {
@@ -64,88 +61,173 @@ public:
             Fragment = 4,
         };
 
-        enum class VertexSize : 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,
-        };
+        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,
+            };
 
-        static std::string VertexSizeToString(VertexSize vertex_size) {
-            switch (vertex_size) {
-            case VertexSize::Size_32_32_32_32:
-                return "32_32_32_32";
-            case VertexSize::Size_32_32_32:
-                return "32_32_32";
-            case VertexSize::Size_16_16_16_16:
-                return "16_16_16_16";
-            case VertexSize::Size_32_32:
-                return "32_32";
-            case VertexSize::Size_16_16_16:
-                return "16_16_16";
-            case VertexSize::Size_8_8_8_8:
-                return "8_8_8_8";
-            case VertexSize::Size_16_16:
-                return "16_16";
-            case VertexSize::Size_32:
-                return "32";
-            case VertexSize::Size_8_8_8:
-                return "8_8_8";
-            case VertexSize::Size_8_8:
-                return "8_8";
-            case VertexSize::Size_16:
-                return "16";
-            case VertexSize::Size_8:
-                return "8";
-            case VertexSize::Size_10_10_10_2:
-                return "10_10_10_2";
-            case VertexSize::Size_11_11_10:
-                return "11_11_10";
+            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();
+                }
             }
-            UNIMPLEMENTED();
-            return {};
-        }
 
-        enum class VertexType : u32 {
-            SignedNorm = 1,
-            UnsignedNorm = 2,
-            SignedInt = 3,
-            UnsignedInt = 4,
-            UnsignedScaled = 5,
-            SignedScaled = 6,
-            Float = 7,
-        };
-
-        static std::string VertexTypeToString(VertexType vertex_type) {
-            switch (vertex_type) {
-            case VertexType::SignedNorm:
-                return "SignedNorm";
-            case VertexType::UnsignedNorm:
-                return "UnsignedNorm";
-            case VertexType::SignedInt:
-                return "SignedInt";
-            case VertexType::UnsignedInt:
-                return "UnsignedInt";
-            case VertexType::UnsignedScaled:
-                return "UnsignedScaled";
-            case VertexType::SignedScaled:
-                return "SignedScaled";
-            case VertexType::Float:
-                return "Float";
+            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();
+                }
             }
-            UNIMPLEMENTED();
-            return {};
-        }
+
+            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,
@@ -172,9 +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;
@@ -200,6 +282,15 @@ public:
                     };
                     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);
@@ -226,14 +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, VertexSize> size;
-                    BitField<27, 3, VertexType> type;
-                    BitField<31, 1, u32> bgra;
-                } vertex_attrib_format[NumVertexAttributes];
+                VertexAttribute vertex_attrib_format[NumVertexAttributes];
 
                 INSERT_PADDING_WORDS(0xF);
 
@@ -430,6 +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;
 
@@ -440,6 +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

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

@@ -13,6 +13,13 @@
 
 namespace Tegra {
 
+enum class RenderTargetFormat : u32 {
+    NONE = 0x0,
+    RGBA8_UNORM = 0xD5,
+};
+
+class DebugContext;
+
 /**
  * Struct describing framebuffer configuration
  */
@@ -66,6 +73,9 @@ 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() {

src/video_core/rasterizer_interface.h

@@ -15,8 +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;

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;
 
@@ -120,14 +125,14 @@ RasterizerOpenGL::RasterizerOpenGL() {
         glBufferData(GL_UNIFORM_BUFFER, sizeof(VSUniformData), nullptr, GL_STREAM_COPY);
         glBindBufferBase(GL_UNIFORM_BUFFER, 1, vs_uniform_buffer.handle);
     } else {
-        ASSERT_MSG(false, "Unimplemented");
+        UNREACHABLE();
     }
 
     accelerate_draw = AccelDraw::Disabled;
 
     glEnable(GL_BLEND);
 
-    LOG_WARNING(HW_GPU, "Sync fixed function OpenGL state here when ready");
+    LOG_CRITICAL(Render_OpenGL, "Sync fixed function OpenGL state here!");
 }
 
 RasterizerOpenGL::~RasterizerOpenGL() {
@@ -138,47 +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);
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 bool RasterizerOpenGL::AccelerateDrawBatch(bool is_indexed) {
     if (!has_ARB_separate_shader_objects) {
-        ASSERT_MSG(false, "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) {}
@@ -206,17 +399,17 @@ void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) {
 
 bool RasterizerOpenGL::AccelerateDisplayTransfer(const void* config) {
     MICROPROFILE_SCOPE(OpenGL_Blits);
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
     return true;
 }
 
 bool RasterizerOpenGL::AccelerateTextureCopy(const void* config) {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
     return true;
 }
 
 bool RasterizerOpenGL::AccelerateFill(const void* config) {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
     return true;
 }
 
@@ -297,14 +490,14 @@ void main() {
         return;
     }
 
-    LOG_ERROR(HW_GPU, "Emulated shaders are not supported! Using a passthrough shader.");
+    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 {
-        ASSERT_MSG(false, "Unimplemented");
+        UNREACHABLE();
     }
 
     state.draw.shader_program = test_shader.shader.handle;
@@ -316,34 +509,70 @@ void main() {
     }
 }
 
+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() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncClipCoef() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncCullMode() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncDepthScale() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncDepthOffset() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncBlendEnabled() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncBlendFuncs() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }
 
 void RasterizerOpenGL::SyncBlendColor() {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
 }

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -29,7 +29,7 @@ public:
     RasterizerOpenGL();
     ~RasterizerOpenGL() override;
 
-    void DrawTriangles() override;
+    void DrawArrays() override;
     void NotifyMaxwellRegisterChanged(u32 id) override;
     void FlushAll() override;
     void FlushRegion(VAddr addr, u64 size) override;
@@ -87,6 +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();
 
@@ -139,7 +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;

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -21,10 +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"
@@ -110,65 +113,26 @@ 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, 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 VAddr aligned_down_start = base + Common::AlignDown(start - base, tile_size);
-    const VAddr aligned_start = base + Common::AlignUp(start - base, tile_size);
-    const VAddr 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::GetPointer(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*, VAddr, VAddr, VAddr), 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
+    MortonCopy<true, PixelFormat::RGBA8>,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
     nullptr,
     nullptr,
     nullptr,
@@ -177,19 +141,19 @@ static constexpr std::array<void (*)(u32, u32, u8*, VAddr, VAddr, VAddr), 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*, VAddr, VAddr, VAddr), 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
+    MortonCopy<false, PixelFormat::RGBA8>,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
     nullptr,
     nullptr,
     nullptr,
@@ -198,11 +162,6 @@ static constexpr std::array<void (*)(u32, u32, u8*, VAddr, VAddr, VAddr), 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
@@ -291,8 +250,8 @@ 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) {
-    ASSERT_MSG(false, "Unimplemented");
-    return true;
+    UNREACHABLE();
+    return {};
 }
 
 SurfaceParams SurfaceParams::FromInterval(SurfaceInterval interval) const {
@@ -531,7 +490,7 @@ MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64
 void CachedSurface::LoadGLBuffer(VAddr load_start, VAddr load_end) {
     ASSERT(type != SurfaceType::Fill);
 
-    u8* texture_src_data = Memory::GetPointer(addr);
+    u8* const texture_src_data = Memory::GetPointer(addr);
     if (texture_src_data == nullptr)
         return;
 
@@ -548,11 +507,16 @@ void CachedSurface::LoadGLBuffer(VAddr load_start, VAddr load_end) {
     if (!is_tiled) {
         ASSERT(type == SurfaceType::Color);
         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 {
-        ASSERT_MSG(false, "Unimplemented");
+        morton_to_gl_fns[static_cast<size_t>(pixel_format)](stride, height, &gl_buffer[0], addr,
+                                                            load_start, load_end);
     }
 }
 
@@ -1093,18 +1057,106 @@ SurfaceRect_Tuple RasterizerCacheOpenGL::GetSurfaceSubRect(const SurfaceParams&
 }
 
 Surface RasterizerCacheOpenGL::GetTextureSurface(const void* config) {
-    ASSERT_MSG(false, "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) {
-    ASSERT_MSG(false, "Unimplemented");
+    UNREACHABLE();
     return {};
 }
 
@@ -1348,5 +1400,33 @@ void RasterizerCacheOpenGL::UnregisterSurface(const Surface& surface) {
 }
 
 void RasterizerCacheOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
-    // ASSERT_MSG(false, "Unimplemented");
+    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

@@ -41,7 +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
@@ -116,6 +116,15 @@ 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:
@@ -308,7 +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);

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

@@ -100,6 +100,8 @@ RendererOpenGL::~RendererOpenGL() = default;
 
 /// Swap buffers (render frame)
 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();
@@ -114,20 +116,19 @@ void RendererOpenGL::SwapBuffers(boost::optional<const Tegra::FramebufferConfig&
             // performance problem.
             ConfigureFramebufferTexture(screen_info.texture, *framebuffer);
         }
+
+        // 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();
 }
@@ -141,11 +142,6 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
     const u64 size_in_bytes{framebuffer.stride * framebuffer.height * bytes_per_pixel};
     const VAddr framebuffer_addr{framebuffer.address + framebuffer.offset};
 
-    // TODO(bunnei): The framebuffer region should only be invalidated if it is written to, not
-    // every frame. When we find the right place for this, the below line can be removed.
-    Memory::RasterizerFlushVirtualRegion(framebuffer_addr, size_in_bytes,
-                                         Memory::FlushMode::Invalidate);
-
     // Framebuffer orientation handling
     framebuffer_transform_flags = framebuffer.transform_flags;
 
@@ -283,7 +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;
@@ -297,8 +293,8 @@ 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;
     auto left = texcoords.left;
     auto right = texcoords.right;
@@ -309,7 +305,7 @@ void RendererOpenGL::DrawSingleScreen(const ScreenInfo& screen_info, float x, fl
             right = texcoords.left;
         } else {
             // Other transformations are unsupported
-            LOG_CRITICAL(HW_GPU, "unsupported framebuffer_transform_flags=%d",
+            LOG_CRITICAL(Render_OpenGL, "Unsupported framebuffer_transform_flags=%d",
                          framebuffer_transform_flags);
             UNIMPLEMENTED();
         }
@@ -334,7 +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;
 
@@ -350,8 +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

@@ -55,8 +55,8 @@ private:
     void InitOpenGLObjects();
     void ConfigureFramebufferTexture(TextureInfo& texture,
                                      const Tegra::FramebufferConfig& framebuffer);
-    void DrawScreens();
-    void DrawSingleScreen(const ScreenInfo& screen_info, float x, float y, float w, float h);
+    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

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