Clean-up

Numerize names, remove unneeded properties and spacer.

CMakeLists.txt

@@ -518,6 +518,10 @@ set(FFmpeg_COMPONENTS
     avutil
     swscale)
 
+if (${CMAKE_SYSTEM_NAME} STREQUAL "Linux")
+    Include(FindPkgConfig REQUIRED)
+    pkg_check_modules(LIBVA libva)
+endif()
 if (NOT YUZU_USE_BUNDLED_FFMPEG)
     # Use system installed FFmpeg
     find_package(FFmpeg QUIET COMPONENTS ${FFmpeg_COMPONENTS})
@@ -540,6 +544,9 @@ endif()
 
 if (YUZU_USE_BUNDLED_FFMPEG)
     if (NOT WIN32)
+        # TODO(lat9nq): Move this to externals/ffmpeg/CMakeLists.txt (and move externals/ffmpeg to
+        # externals/ffmpeg/ffmpeg)
+
         # Build FFmpeg from externals
         message(STATUS "Using FFmpeg from externals")
 
@@ -579,20 +586,23 @@ if (YUZU_USE_BUNDLED_FFMPEG)
                 CACHE PATH "Paths to FFmpeg libraries" FORCE)
         endforeach()
 
-        set(FFmpeg_INCLUDE_DIR
-            "${FFmpeg_PREFIX};${FFmpeg_BUILD_DIR}"
-            CACHE PATH "Path to FFmpeg headers" FORCE)
+        Include(FindPkgConfig REQUIRED)
+        pkg_check_modules(LIBVA libva)
+        pkg_check_modules(CUDA cuda)
+        pkg_check_modules(FFNVCODEC ffnvcodec)
+        pkg_check_modules(VDPAU vdpau)
+
+        set(FFmpeg_HWACCEL_LIBRARIES)
+        set(FFmpeg_HWACCEL_FLAGS)
+        set(FFmpeg_HWACCEL_INCLUDE_DIRS)
+        set(FFmpeg_HWACCEL_LDFLAGS)
 
-        if (${CMAKE_SYSTEM_NAME} STREQUAL "Linux")
-            Include(FindPkgConfig REQUIRED)
-            pkg_check_modules(LIBVA libva)
-        endif()
         if(LIBVA_FOUND)
             pkg_check_modules(LIBDRM libdrm REQUIRED)
             find_package(X11 REQUIRED)
             pkg_check_modules(LIBVA-DRM libva-drm REQUIRED)
             pkg_check_modules(LIBVA-X11 libva-x11 REQUIRED)
-            set(FFmpeg_LIBVA_LIBRARIES
+            list(APPEND FFmpeg_HWACCEL_LIBRARIES
                 ${LIBDRM_LIBRARIES}
                 ${X11_LIBRARIES}
                 ${LIBVA-DRM_LIBRARIES}
@@ -602,11 +612,56 @@ if (YUZU_USE_BUNDLED_FFMPEG)
                 --enable-hwaccel=h264_vaapi
                 --enable-hwaccel=vp9_vaapi
                 --enable-libdrm)
+            list(APPEND FFmpeg_HWACCEL_INCLUDE_DIRS
+                ${LIBDRM_INCLUDE_DIRS}
+                ${X11_INCLUDE_DIRS}
+                ${LIBVA-DRM_INCLUDE_DIRS}
+                ${LIBVA-X11_INCLUDE_DIRS}
+                ${LIBVA_INCLUDE_DIRS}
+            )
             message(STATUS "VA-API found")
         else()
             set(FFmpeg_HWACCEL_FLAGS --disable-vaapi)
         endif()
 
+        if (FFNVCODEC_FOUND AND CUDA_FOUND)
+            list(APPEND FFmpeg_HWACCEL_FLAGS
+                --enable-cuvid
+                --enable-ffnvcodec
+                --enable-nvdec
+                --enable-hwaccel=h264_nvdec
+                --enable-hwaccel=vp9_nvdec
+                --extra-cflags=-I${CUDA_INCLUDE_DIRS}
+            )
+            list(APPEND FFmpeg_HWACCEL_LIBRARIES
+                ${FFNVCODEC_LIBRARIES}
+                ${CUDA_LIBRARIES}
+            )
+            list(APPEND FFmpeg_HWACCEL_INCLUDE_DIRS
+                ${FFNVCODEC_INCLUDE_DIRS}
+                ${CUDA_INCLUDE_DIRS}
+            )
+            list(APPEND FFmpeg_HWACCEL_LDFLAGS
+                ${FFNVCODEC_LDFLAGS}
+                ${CUDA_LDFLAGS}
+            )
+            message(STATUS "ffnvcodec libraries version ${FFNVCODEC_VERSION} found")
+        endif()
+
+        if (VDPAU_FOUND)
+            list(APPEND FFmpeg_HWACCEL_FLAGS
+                --enable-vdpau
+                --enable-hwaccel=h264_vdpau
+                --enable-hwaccel=vp9_vdpau
+            )
+            list(APPEND FFmpeg_HWACCEL_LIBRARIES ${VDPAU_LIBRARIES})
+            list(APPEND FFmpeg_HWACCEL_INCLUDE_DIRS ${VDPAU_INCLUDE_DIRS})
+            list(APPEND FFmpeg_HWACCEL_LDFLAGS ${VDPAU_LDFLAGS})
+            message(STATUS "vdpau libraries version ${VDPAU_VERSION} found")
+        else()
+            list(APPEND FFmpeg_HWACCEL_FLAGS --disable-vdpau)
+        endif()
+
         # `configure` parameters builds only exactly what yuzu needs from FFmpeg
         # `--disable-vdpau` is needed to avoid linking issues
         add_custom_command(
@@ -624,7 +679,6 @@ if (YUZU_USE_BUNDLED_FFMPEG)
                     --disable-network
                     --disable-postproc
                     --disable-swresample
-                    --disable-vdpau
                     --enable-decoder=h264
                     --enable-decoder=vp9
                     --cc="${CMAKE_C_COMPILER}"
@@ -653,15 +707,26 @@ if (YUZU_USE_BUNDLED_FFMPEG)
                 ${FFmpeg_BUILD_DIR}
         )
 
+        set(FFmpeg_INCLUDE_DIR
+            "${FFmpeg_PREFIX};${FFmpeg_BUILD_DIR};${FFmpeg_HWACCEL_INCLUDE_DIRS}"
+            CACHE PATH "Path to FFmpeg headers" FORCE)
+
+        set(FFmpeg_LDFLAGS
+            "${FFmpeg_HWACCEL_LDFLAGS}"
+            CACHE STRING "FFmpeg linker flags" FORCE)
+
         # ALL makes this custom target build every time
         # but it won't actually build if the DEPENDS parameter is up to date
         add_custom_target(ffmpeg-configure ALL DEPENDS ${FFmpeg_MAKEFILE})
         add_custom_target(ffmpeg-build ALL DEPENDS ${FFmpeg_BUILD_LIBRARIES} ffmpeg-configure)
         link_libraries(${FFmpeg_LIBVA_LIBRARIES})
-        set(FFmpeg_LIBRARIES ${FFmpeg_LIBVA_LIBRARIES} ${FFmpeg_BUILD_LIBRARIES}
+        set(FFmpeg_LIBRARIES ${FFmpeg_BUILD_LIBRARIES} ${FFmpeg_HWACCEL_LIBRARIES}
             CACHE PATH "Paths to FFmpeg libraries" FORCE)
         unset(FFmpeg_BUILD_LIBRARIES)
-        unset(FFmpeg_LIBVA_LIBRARIES)
+        unset(FFmpeg_HWACCEL_FLAGS)
+        unset(FFmpeg_HWACCEL_INCLUDE_DIRS)
+        unset(FFmpeg_HWACCEL_LDFLAGS)
+        unset(FFmpeg_HWACCEL_LIBRARIES)
 
         if (FFmpeg_FOUND)
             message(STATUS "Found FFmpeg version ${FFmpeg_VERSION}")
@@ -670,12 +735,13 @@ if (YUZU_USE_BUNDLED_FFMPEG)
         endif()
     else() # WIN32
         # Use yuzu FFmpeg binaries
-        set(FFmpeg_EXT_NAME "ffmpeg-4.3.1")
+        set(FFmpeg_EXT_NAME "ffmpeg-4.4")
         set(FFmpeg_PATH "${CMAKE_BINARY_DIR}/externals/${FFmpeg_EXT_NAME}")
         download_bundled_external("ffmpeg/" ${FFmpeg_EXT_NAME} "")
         set(FFmpeg_FOUND YES)
         set(FFmpeg_INCLUDE_DIR "${FFmpeg_PATH}/include" CACHE PATH "Path to FFmpeg headers" FORCE)
         set(FFmpeg_LIBRARY_DIR "${FFmpeg_PATH}/bin" CACHE PATH "Path to FFmpeg library directory" FORCE)
+        set(FFmpeg_LDFLAGS "" CACHE STRING "FFmpeg linker flags" FORCE)
         set(FFmpeg_DLL_DIR "${FFmpeg_PATH}/bin" CACHE PATH "Path to FFmpeg dll's" FORCE)
         set(FFmpeg_LIBRARIES
             ${FFmpeg_LIBRARY_DIR}/swscale.lib

src/common/CMakeLists.txt

@@ -53,6 +53,8 @@ add_library(common STATIC
     div_ceil.h
     dynamic_library.cpp
     dynamic_library.h
+    error.cpp
+    error.h
     fiber.cpp
     fiber.h
     fs/file.cpp
@@ -88,7 +90,6 @@ add_library(common STATIC
     microprofile.cpp
     microprofile.h
     microprofileui.h
-    misc.cpp
     nvidia_flags.cpp
     nvidia_flags.h
     page_table.cpp

src/common/common_funcs.h

@@ -4,9 +4,8 @@
 
 #pragma once
 
-#include <algorithm>
 #include <array>
-#include <string>
+#include <iterator>
 
 #if !defined(ARCHITECTURE_x86_64)
 #include <cstdlib> // for exit
@@ -49,16 +48,6 @@ __declspec(dllimport) void __stdcall DebugBreak(void);
 
 #endif // _MSC_VER ndef
 
-// Generic function to get last error message.
-// Call directly after the command or use the error num.
-// This function might change the error code.
-// Defined in misc.cpp.
-[[nodiscard]] std::string GetLastErrorMsg();
-
-// Like GetLastErrorMsg(), but passing an explicit error code.
-// Defined in misc.cpp.
-[[nodiscard]] std::string NativeErrorToString(int e);
-
 #define DECLARE_ENUM_FLAG_OPERATORS(type)                                                          \
     [[nodiscard]] constexpr type operator|(type a, type b) noexcept {                              \
         using T = std::underlying_type_t<type>;                                                    \
@@ -72,6 +61,14 @@ __declspec(dllimport) void __stdcall DebugBreak(void);
         using T = std::underlying_type_t<type>;                                                    \
         return static_cast<type>(static_cast<T>(a) ^ static_cast<T>(b));                           \
     }                                                                                              \
+    [[nodiscard]] constexpr type operator<<(type a, type b) noexcept {                             \
+        using T = std::underlying_type_t<type>;                                                    \
+        return static_cast<type>(static_cast<T>(a) << static_cast<T>(b));                          \
+    }                                                                                              \
+    [[nodiscard]] constexpr type operator>>(type a, type b) noexcept {                             \
+        using T = std::underlying_type_t<type>;                                                    \
+        return static_cast<type>(static_cast<T>(a) >> static_cast<T>(b));                          \
+    }                                                                                              \
     constexpr type& operator|=(type& a, type b) noexcept {                                         \
         a = a | b;                                                                                 \
         return a;                                                                                  \
@@ -84,6 +81,14 @@ __declspec(dllimport) void __stdcall DebugBreak(void);
         a = a ^ b;                                                                                 \
         return a;                                                                                  \
     }                                                                                              \
+    constexpr type& operator<<=(type& a, type b) noexcept {                                        \
+        a = a << b;                                                                                \
+        return a;                                                                                  \
+    }                                                                                              \
+    constexpr type& operator>>=(type& a, type b) noexcept {                                        \
+        a = a >> b;                                                                                \
+        return a;                                                                                  \
+    }                                                                                              \
     [[nodiscard]] constexpr type operator~(type key) noexcept {                                    \
         using T = std::underlying_type_t<type>;                                                    \
         return static_cast<type>(~static_cast<T>(key));                                            \

src/common/error.cpp

@@ -10,7 +10,9 @@
 #include <cstring>
 #endif
 
-#include "common/common_funcs.h"
+#include "common/error.h"
+
+namespace Common {
 
 std::string NativeErrorToString(int e) {
 #ifdef _WIN32
@@ -50,3 +52,5 @@ std::string GetLastErrorMsg() {
     return NativeErrorToString(errno);
 #endif
 }
+
+} // namespace Common

src/common/error.h

@@ -0,0 +1,21 @@
+// Copyright 2013 Dolphin Emulator Project / 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <string>
+
+namespace Common {
+
+// Generic function to get last error message.
+// Call directly after the command or use the error num.
+// This function might change the error code.
+// Defined in error.cpp.
+[[nodiscard]] std::string GetLastErrorMsg();
+
+// Like GetLastErrorMsg(), but passing an explicit error code.
+// Defined in error.cpp.
+[[nodiscard]] std::string NativeErrorToString(int e);
+
+} // namespace Common

src/common/fs/fs_paths.h

@@ -21,6 +21,7 @@
 #define SCREENSHOTS_DIR "screenshots"
 #define SDMC_DIR "sdmc"
 #define SHADER_DIR "shader"
+#define TAS_DIR "tas"
 
 // yuzu-specific files
 

src/common/fs/path_util.cpp

@@ -116,6 +116,7 @@ private:
         GenerateYuzuPath(YuzuPath::ScreenshotsDir, yuzu_path / SCREENSHOTS_DIR);
         GenerateYuzuPath(YuzuPath::SDMCDir, yuzu_path / SDMC_DIR);
         GenerateYuzuPath(YuzuPath::ShaderDir, yuzu_path / SHADER_DIR);
+        GenerateYuzuPath(YuzuPath::TASDir, yuzu_path / TAS_DIR);
     }
 
     ~PathManagerImpl() = default;

src/common/fs/path_util.h

@@ -23,6 +23,7 @@ enum class YuzuPath {
     ScreenshotsDir, // Where yuzu screenshots are stored.
     SDMCDir,        // Where the emulated SDMC is stored.
     ShaderDir,      // Where shaders are stored.
+    TASDir,         // Where TAS scripts are stored.
 };
 
 /**

src/common/settings.cpp

@@ -54,14 +54,13 @@ void LogSettings() {
     log_setting("Renderer_GPUAccuracyLevel", values.gpu_accuracy.GetValue());
     log_setting("Renderer_UseAsynchronousGpuEmulation",
                 values.use_asynchronous_gpu_emulation.GetValue());
-    log_setting("Renderer_UseNvdecEmulation", values.use_nvdec_emulation.GetValue());
+    log_setting("Renderer_NvdecEmulation", values.nvdec_emulation.GetValue());
     log_setting("Renderer_AccelerateASTC", values.accelerate_astc.GetValue());
     log_setting("Renderer_UseVsync", values.use_vsync.GetValue());
     log_setting("Renderer_ShaderBackend", values.shader_backend.GetValue());
     log_setting("Renderer_UseAsynchronousShaders", values.use_asynchronous_shaders.GetValue());
     log_setting("Renderer_AnisotropicFilteringLevel", values.max_anisotropy.GetValue());
     log_setting("Audio_OutputEngine", values.sink_id.GetValue());
-    log_setting("Audio_EnableAudioStretching", values.enable_audio_stretching.GetValue());
     log_setting("Audio_OutputDevice", values.audio_device_id.GetValue());
     log_setting("DataStorage_UseVirtualSd", values.use_virtual_sd.GetValue());
     log_path("DataStorage_CacheDir", Common::FS::GetYuzuPath(Common::FS::YuzuPath::CacheDir));
@@ -72,6 +71,9 @@ void LogSettings() {
     log_setting("Debugging_ProgramArgs", values.program_args.GetValue());
     log_setting("Services_BCATBackend", values.bcat_backend.GetValue());
     log_setting("Services_BCATBoxcatLocal", values.bcat_boxcat_local.GetValue());
+    log_setting("Input_EnableMotion", values.motion_enabled.GetValue());
+    log_setting("Input_EnableVibration", values.vibration_enabled.GetValue());
+    log_setting("Input_EnableRawInput", values.enable_raw_input.GetValue());
 }
 
 bool IsConfiguringGlobal() {
@@ -112,7 +114,6 @@ void RestoreGlobalState(bool is_powered_on) {
     }
 
     // Audio
-    values.enable_audio_stretching.SetGlobal(true);
     values.volume.SetGlobal(true);
 
     // Core
@@ -136,7 +137,7 @@ void RestoreGlobalState(bool is_powered_on) {
     values.use_disk_shader_cache.SetGlobal(true);
     values.gpu_accuracy.SetGlobal(true);
     values.use_asynchronous_gpu_emulation.SetGlobal(true);
-    values.use_nvdec_emulation.SetGlobal(true);
+    values.nvdec_emulation.SetGlobal(true);
     values.accelerate_astc.SetGlobal(true);
     values.use_vsync.SetGlobal(true);
     values.shader_backend.SetGlobal(true);

src/common/settings.h

@@ -16,7 +16,6 @@
 
 #include "common/common_types.h"
 #include "common/settings_input.h"
-#include "input_common/udp/client.h"
 
 namespace Settings {
 
@@ -48,6 +47,12 @@ enum class FullscreenMode : u32 {
     Exclusive = 1,
 };
 
+enum class NvdecEmulation : u32 {
+    Off = 0,
+    CPU = 1,
+    GPU = 2,
+};
+
 /** The BasicSetting class is a simple resource manager. It defines a label and default value
  * alongside the actual value of the setting for simpler and less-error prone use with frontend
  * configurations. Setting a default value and label is required, though subclasses may deviate from
@@ -409,7 +414,6 @@ struct Values {
     BasicSetting<std::string> audio_device_id{"auto", "output_device"};
     BasicSetting<std::string> sink_id{"auto", "output_engine"};
     BasicSetting<bool> audio_muted{false, "audio_muted"};
-    Setting<bool> enable_audio_stretching{true, "enable_audio_stretching"};
     RangedSetting<u8> volume{100, 0, 100, "volume"};
 
     // Core
@@ -466,7 +470,7 @@ struct Values {
     RangedSetting<GPUAccuracy> gpu_accuracy{GPUAccuracy::High, GPUAccuracy::Normal,
                                             GPUAccuracy::Extreme, "gpu_accuracy"};
     Setting<bool> use_asynchronous_gpu_emulation{true, "use_asynchronous_gpu_emulation"};
-    Setting<bool> use_nvdec_emulation{true, "use_nvdec_emulation"};
+    Setting<NvdecEmulation> nvdec_emulation{NvdecEmulation::GPU, "nvdec_emulation"};
     Setting<bool> accelerate_astc{true, "accelerate_astc"};
     Setting<bool> use_vsync{true, "use_vsync"};
     BasicRangedSetting<u16> fps_cap{1000, 1, 1000, "fps_cap"};
@@ -498,14 +502,20 @@ struct Values {
 
     Setting<bool> use_docked_mode{true, "use_docked_mode"};
 
+    BasicSetting<bool> enable_raw_input{false, "enable_raw_input"};
+
     Setting<bool> vibration_enabled{true, "vibration_enabled"};
     Setting<bool> enable_accurate_vibrations{false, "enable_accurate_vibrations"};
 
     Setting<bool> motion_enabled{true, "motion_enabled"};
     BasicSetting<std::string> motion_device{"engine:motion_emu,update_period:100,sensitivity:0.01",
                                             "motion_device"};
-    BasicSetting<std::string> udp_input_servers{InputCommon::CemuhookUDP::DEFAULT_SRV,
-                                                "udp_input_servers"};
+    BasicSetting<std::string> udp_input_servers{"127.0.0.1:26760", "udp_input_servers"};
+
+    BasicSetting<bool> pause_tas_on_load{true, "pause_tas_on_load"};
+    BasicSetting<bool> tas_enable{false, "tas_enable"};
+    BasicSetting<bool> tas_loop{false, "tas_loop"};
+    BasicSetting<bool> tas_swap_controllers{true, "tas_swap_controllers"};
 
     BasicSetting<bool> mouse_panning{false, "mouse_panning"};
     BasicRangedSetting<u8> mouse_panning_sensitivity{10, 1, 100, "mouse_panning_sensitivity"};

src/common/thread.cpp

@@ -2,7 +2,9 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "common/common_funcs.h"
+#include <string>
+
+#include "common/error.h"
 #include "common/logging/log.h"
 #include "common/thread.h"
 #ifdef __APPLE__
@@ -21,8 +23,6 @@
 #include <unistd.h>
 #endif
 
-#include <string>
-
 #ifdef __FreeBSD__
 #define cpu_set_t cpuset_t
 #endif

src/core/CMakeLists.txt

@@ -263,6 +263,8 @@ add_library(core STATIC
     hle/service/acc/acc_u0.h
     hle/service/acc/acc_u1.cpp
     hle/service/acc/acc_u1.h
+    hle/service/acc/async_context.cpp
+    hle/service/acc/async_context.h
     hle/service/acc/errors.h
     hle/service/acc/profile_manager.cpp
     hle/service/acc/profile_manager.h

src/core/core.cpp

@@ -83,6 +83,12 @@ FileSys::StorageId GetStorageIdForFrontendSlot(
     }
 }
 
+void KProcessDeleter(Kernel::KProcess* process) {
+    process->Destroy();
+}
+
+using KProcessPtr = std::unique_ptr<Kernel::KProcess, decltype(&KProcessDeleter)>;
+
 } // Anonymous namespace
 
 FileSys::VirtualFile GetGameFileFromPath(const FileSys::VirtualFilesystem& vfs,
@@ -233,8 +239,8 @@ struct System::Impl {
         }
 
         telemetry_session->AddInitialInfo(*app_loader, fs_controller, *content_provider);
-        auto main_process = Kernel::KProcess::Create(system.Kernel());
-        ASSERT(Kernel::KProcess::Initialize(main_process, system, "main",
+        main_process = KProcessPtr{Kernel::KProcess::Create(system.Kernel()), KProcessDeleter};
+        ASSERT(Kernel::KProcess::Initialize(main_process.get(), system, "main",
                                             Kernel::KProcess::ProcessType::Userland)
                    .IsSuccess());
         main_process->Open();
@@ -247,7 +253,7 @@ struct System::Impl {
                                              static_cast<u32>(load_result));
         }
         AddGlueRegistrationForProcess(*app_loader, *main_process);
-        kernel.MakeCurrentProcess(main_process);
+        kernel.MakeCurrentProcess(main_process.get());
         kernel.InitializeCores();
 
         // Initialize cheat engine
@@ -316,6 +322,8 @@ struct System::Impl {
         kernel.Shutdown();
         memory.Reset();
         applet_manager.ClearAll();
+        // TODO: The main process should be freed based on KAutoObject ref counting.
+        main_process.reset();
 
         LOG_DEBUG(Core, "Shutdown OK");
     }
@@ -374,6 +382,7 @@ struct System::Impl {
     std::unique_ptr<Tegra::GPU> gpu_core;
     std::unique_ptr<Hardware::InterruptManager> interrupt_manager;
     std::unique_ptr<Core::DeviceMemory> device_memory;
+    KProcessPtr main_process{nullptr, KProcessDeleter};
     Core::Memory::Memory memory;
     CpuManager cpu_manager;
     std::atomic_bool is_powered_on{};

src/core/cpu_manager.cpp

@@ -21,34 +21,25 @@ namespace Core {
 CpuManager::CpuManager(System& system_) : system{system_} {}
 CpuManager::~CpuManager() = default;
 
-void CpuManager::ThreadStart(CpuManager& cpu_manager, std::size_t core) {
-    cpu_manager.RunThread(core);
+void CpuManager::ThreadStart(std::stop_token stop_token, CpuManager& cpu_manager,
+                             std::size_t core) {
+    cpu_manager.RunThread(stop_token, core);
 }
 
 void CpuManager::Initialize() {
     running_mode = true;
     if (is_multicore) {
         for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
-            core_data[core].host_thread =
-                std::make_unique<std::thread>(ThreadStart, std::ref(*this), core);
+            core_data[core].host_thread = std::jthread(ThreadStart, std::ref(*this), core);
         }
     } else {
-        core_data[0].host_thread = std::make_unique<std::thread>(ThreadStart, std::ref(*this), 0);
+        core_data[0].host_thread = std::jthread(ThreadStart, std::ref(*this), 0);
     }
 }
 
 void CpuManager::Shutdown() {
     running_mode = false;
     Pause(false);
-    if (is_multicore) {
-        for (auto& data : core_data) {
-            data.host_thread->join();
-            data.host_thread.reset();
-        }
-    } else {
-        core_data[0].host_thread->join();
-        core_data[0].host_thread.reset();
-    }
 }
 
 std::function<void(void*)> CpuManager::GetGuestThreadStartFunc() {
@@ -317,7 +308,7 @@ void CpuManager::Pause(bool paused) {
     }
 }
 
-void CpuManager::RunThread(std::size_t core) {
+void CpuManager::RunThread(std::stop_token stop_token, std::size_t core) {
     /// Initialization
     system.RegisterCoreThread(core);
     std::string name;
@@ -361,6 +352,10 @@ void CpuManager::RunThread(std::size_t core) {
             return;
         }
 
+        if (stop_token.stop_requested()) {
+            break;
+        }
+
         auto current_thread = system.Kernel().CurrentScheduler()->GetCurrentThread();
         data.is_running = true;
         Common::Fiber::YieldTo(data.host_context, *current_thread->GetHostContext());

src/core/cpu_manager.h

@@ -78,9 +78,9 @@ private:
     void SingleCoreRunSuspendThread();
     void SingleCorePause(bool paused);
 
-    static void ThreadStart(CpuManager& cpu_manager, std::size_t core);
+    static void ThreadStart(std::stop_token stop_token, CpuManager& cpu_manager, std::size_t core);
 
-    void RunThread(std::size_t core);
+    void RunThread(std::stop_token stop_token, std::size_t core);
 
     struct CoreData {
         std::shared_ptr<Common::Fiber> host_context;
@@ -89,7 +89,7 @@ private:
         std::atomic<bool> is_running;
         std::atomic<bool> is_paused;
         std::atomic<bool> initialized;
-        std::unique_ptr<std::thread> host_thread;
+        std::jthread host_thread;
     };
 
     std::atomic<bool> running_mode{};

src/core/file_sys/kernel_executable.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <array>
+#include <string>
 #include <vector>
 
 #include "common/common_funcs.h"

src/core/file_sys/vfs.cpp

@@ -273,6 +273,10 @@ VirtualFile VfsDirectory::GetFile(std::string_view name) const {
     return iter == files.end() ? nullptr : *iter;
 }
 
+FileTimeStampRaw VfsDirectory::GetFileTimeStamp([[maybe_unused]] std::string_view path) const {
+    return {};
+}
+
 VirtualDir VfsDirectory::GetSubdirectory(std::string_view name) const {
     const auto& subs = GetSubdirectories();
     const auto iter = std::find_if(subs.begin(), subs.end(),

src/core/file_sys/vfs.h

@@ -199,6 +199,9 @@ public:
     // file with name.
     virtual VirtualFile GetFile(std::string_view name) const;
 
+    // Returns a struct containing the file's timestamp.
+    virtual FileTimeStampRaw GetFileTimeStamp(std::string_view path) const;
+
     // Returns a vector containing all of the subdirectories in this directory.
     virtual std::vector<VirtualDir> GetSubdirectories() const = 0;
     // Returns the directory with name matching name. Returns nullptr if directory dosen't have a

src/core/file_sys/vfs_real.cpp

@@ -13,6 +13,13 @@
 #include "common/logging/log.h"
 #include "core/file_sys/vfs_real.h"
 
+// For FileTimeStampRaw
+#include <sys/stat.h>
+
+#ifdef _MSC_VER
+#define stat _stat64
+#endif
+
 namespace FileSys {
 
 namespace FS = Common::FS;
@@ -392,6 +399,28 @@ std::vector<VirtualFile> RealVfsDirectory::GetFiles() const {
     return IterateEntries<RealVfsFile, VfsFile>();
 }
 
+FileTimeStampRaw RealVfsDirectory::GetFileTimeStamp(std::string_view path_) const {
+    const auto full_path = FS::SanitizePath(path + '/' + std::string(path_));
+    const auto fs_path = std::filesystem::path{FS::ToU8String(full_path)};
+    struct stat file_status;
+
+#ifdef _WIN32
+    const auto stat_result = _wstat64(fs_path.c_str(), &file_status);
+#else
+    const auto stat_result = stat(fs_path.c_str(), &file_status);
+#endif
+
+    if (stat_result != 0) {
+        return {};
+    }
+
+    return {
+        .created{static_cast<u64>(file_status.st_ctime)},
+        .accessed{static_cast<u64>(file_status.st_atime)},
+        .modified{static_cast<u64>(file_status.st_mtime)},
+    };
+}
+
 std::vector<VirtualDir> RealVfsDirectory::GetSubdirectories() const {
     return IterateEntries<RealVfsDirectory, VfsDirectory>();
 }

src/core/file_sys/vfs_real.h

@@ -86,6 +86,7 @@ public:
     VirtualDir CreateDirectoryRelative(std::string_view relative_path) override;
     bool DeleteSubdirectoryRecursive(std::string_view name) override;
     std::vector<VirtualFile> GetFiles() const override;
+    FileTimeStampRaw GetFileTimeStamp(std::string_view path) const override;
     std::vector<VirtualDir> GetSubdirectories() const override;
     bool IsWritable() const override;
     bool IsReadable() const override;

src/core/file_sys/vfs_types.h

@@ -6,6 +6,8 @@
 
 #include <memory>
 
+#include "common/common_types.h"
+
 namespace FileSys {
 
 class VfsDirectory;
@@ -18,4 +20,11 @@ using VirtualDir = std::shared_ptr<VfsDirectory>;
 using VirtualFile = std::shared_ptr<VfsFile>;
 using VirtualFilesystem = std::shared_ptr<VfsFilesystem>;
 
+struct FileTimeStampRaw {
+    u64 created{};
+    u64 accessed{};
+    u64 modified{};
+    u64 padding{};
+};
+
 } // namespace FileSys

src/core/hle/api_version.h

@@ -28,13 +28,20 @@ constexpr char DISPLAY_TITLE[] = "NintendoSDK Firmware for NX 12.1.0-1.0";
 
 // Atmosphere version constants.
 
-constexpr u8 ATMOSPHERE_RELEASE_VERSION_MAJOR = 0;
-constexpr u8 ATMOSPHERE_RELEASE_VERSION_MINOR = 19;
-constexpr u8 ATMOSPHERE_RELEASE_VERSION_MICRO = 5;
+constexpr u8 ATMOSPHERE_RELEASE_VERSION_MAJOR = 1;
+constexpr u8 ATMOSPHERE_RELEASE_VERSION_MINOR = 0;
+constexpr u8 ATMOSPHERE_RELEASE_VERSION_MICRO = 0;
+
+constexpr u32 AtmosphereTargetFirmwareWithRevision(u8 major, u8 minor, u8 micro, u8 rev) {
+    return u32{major} << 24 | u32{minor} << 16 | u32{micro} << 8 | u32{rev};
+}
+
+constexpr u32 AtmosphereTargetFirmware(u8 major, u8 minor, u8 micro) {
+    return AtmosphereTargetFirmwareWithRevision(major, minor, micro, 0);
+}
 
 constexpr u32 GetTargetFirmware() {
-    return u32{HOS_VERSION_MAJOR} << 24 | u32{HOS_VERSION_MINOR} << 16 |
-           u32{HOS_VERSION_MICRO} << 8 | 0U;
+    return AtmosphereTargetFirmware(HOS_VERSION_MAJOR, HOS_VERSION_MINOR, HOS_VERSION_MICRO);
 }
 
 } // namespace HLE::ApiVersion

src/core/hle/kernel/kernel.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <array>
+#include <functional>
 #include <memory>
 #include <string>
 #include <unordered_map>

src/core/hle/service/acc/acc.cpp

@@ -23,6 +23,7 @@
 #include "core/hle/service/acc/acc_su.h"
 #include "core/hle/service/acc/acc_u0.h"
 #include "core/hle/service/acc/acc_u1.h"
+#include "core/hle/service/acc/async_context.h"
 #include "core/hle/service/acc/errors.h"
 #include "core/hle/service/acc/profile_manager.h"
 #include "core/hle/service/glue/arp.h"
@@ -454,22 +455,6 @@ public:
         : IProfileCommon{system_, "IProfileEditor", true, user_id_, profile_manager_} {}
 };
 
-class IAsyncContext final : public ServiceFramework<IAsyncContext> {
-public:
-    explicit IAsyncContext(Core::System& system_) : ServiceFramework{system_, "IAsyncContext"} {
-        // clang-format off
-        static const FunctionInfo functions[] = {
-            {0, nullptr, "GetSystemEvent"},
-            {1, nullptr, "Cancel"},
-            {2, nullptr, "HasDone"},
-            {3, nullptr, "GetResult"},
-        };
-        // clang-format on
-
-        RegisterHandlers(functions);
-    }
-};
-
 class ISessionObject final : public ServiceFramework<ISessionObject> {
 public:
     explicit ISessionObject(Core::System& system_, Common::UUID)
@@ -504,16 +489,44 @@ public:
     }
 };
 
+class EnsureTokenIdCacheAsyncInterface final : public IAsyncContext {
+public:
+    explicit EnsureTokenIdCacheAsyncInterface(Core::System& system_) : IAsyncContext{system_} {
+        MarkComplete();
+    }
+    ~EnsureTokenIdCacheAsyncInterface() = default;
+
+    void LoadIdTokenCache(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_ACC, "(STUBBED) called");
+
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(ResultSuccess);
+    }
+
+protected:
+    bool IsComplete() const override {
+        return true;
+    }
+
+    void Cancel() override {}
+
+    ResultCode GetResult() const override {
+        return ResultSuccess;
+    }
+};
+
 class IManagerForApplication final : public ServiceFramework<IManagerForApplication> {
 public:
     explicit IManagerForApplication(Core::System& system_, Common::UUID user_id_)
-        : ServiceFramework{system_, "IManagerForApplication"}, user_id{user_id_} {
+        : ServiceFramework{system_, "IManagerForApplication"},
+          ensure_token_id{std::make_shared<EnsureTokenIdCacheAsyncInterface>(system)},
+          user_id{user_id_} {
         // clang-format off
         static const FunctionInfo functions[] = {
             {0, &IManagerForApplication::CheckAvailability, "CheckAvailability"},
             {1, &IManagerForApplication::GetAccountId, "GetAccountId"},
-            {2, nullptr, "EnsureIdTokenCacheAsync"},
-            {3, nullptr, "LoadIdTokenCache"},
+            {2, &IManagerForApplication::EnsureIdTokenCacheAsync, "EnsureIdTokenCacheAsync"},
+            {3, &IManagerForApplication::LoadIdTokenCache, "LoadIdTokenCache"},
             {130, &IManagerForApplication::GetNintendoAccountUserResourceCacheForApplication, "GetNintendoAccountUserResourceCacheForApplication"},
             {150, nullptr, "CreateAuthorizationRequest"},
             {160, &IManagerForApplication::StoreOpenContext, "StoreOpenContext"},
@@ -540,6 +553,20 @@ private:
         rb.PushRaw<u64>(user_id.GetNintendoID());
     }
 
+    void EnsureIdTokenCacheAsync(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_ACC, "(STUBBED) called");
+
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(ResultSuccess);
+        rb.PushIpcInterface(ensure_token_id);
+    }
+
+    void LoadIdTokenCache(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_ACC, "(STUBBED) called");
+
+        ensure_token_id->LoadIdTokenCache(ctx);
+    }
+
     void GetNintendoAccountUserResourceCacheForApplication(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_ACC, "(STUBBED) called");
 
@@ -562,6 +589,7 @@ private:
         rb.Push(ResultSuccess);
     }
 
+    std::shared_ptr<EnsureTokenIdCacheAsyncInterface> ensure_token_id{};
     Common::UUID user_id{Common::INVALID_UUID};
 };
 

src/core/hle/service/acc/async_context.cpp

@@ -0,0 +1,68 @@
+// Copyright 2021 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/core.h"
+#include "core/hle/ipc_helpers.h"
+#include "core/hle/service/acc/async_context.h"
+
+namespace Service::Account {
+IAsyncContext::IAsyncContext(Core::System& system_)
+    : ServiceFramework{system_, "IAsyncContext"}, compeletion_event{system_.Kernel()} {
+
+    Kernel::KAutoObject::Create(std::addressof(compeletion_event));
+    compeletion_event.Initialize("IAsyncContext:CompletionEvent");
+
+    // clang-format off
+    static const FunctionInfo functions[] = {
+        {0, &IAsyncContext::GetSystemEvent, "GetSystemEvent"},
+        {1, &IAsyncContext::Cancel, "Cancel"},
+        {2, &IAsyncContext::HasDone, "HasDone"},
+        {3, &IAsyncContext::GetResult, "GetResult"},
+    };
+    // clang-format on
+
+    RegisterHandlers(functions);
+}
+
+void IAsyncContext::GetSystemEvent(Kernel::HLERequestContext& ctx) {
+    LOG_DEBUG(Service_ACC, "called");
+
+    IPC::ResponseBuilder rb{ctx, 2, 1};
+    rb.Push(ResultSuccess);
+    rb.PushCopyObjects(compeletion_event.GetReadableEvent());
+}
+
+void IAsyncContext::Cancel(Kernel::HLERequestContext& ctx) {
+    LOG_DEBUG(Service_ACC, "called");
+
+    Cancel();
+    MarkComplete();
+
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(ResultSuccess);
+}
+
+void IAsyncContext::HasDone(Kernel::HLERequestContext& ctx) {
+    LOG_DEBUG(Service_ACC, "called");
+
+    is_complete.store(IsComplete());
+
+    IPC::ResponseBuilder rb{ctx, 3};
+    rb.Push(ResultSuccess);
+    rb.Push(is_complete.load());
+}
+
+void IAsyncContext::GetResult(Kernel::HLERequestContext& ctx) {
+    LOG_DEBUG(Service_ACC, "called");
+
+    IPC::ResponseBuilder rb{ctx, 3};
+    rb.Push(GetResult());
+}
+
+void IAsyncContext::MarkComplete() {
+    is_complete.store(true);
+    compeletion_event.GetWritableEvent().Signal();
+}
+
+} // namespace Service::Account

src/core/hle/service/acc/async_context.h

@@ -0,0 +1,37 @@
+// Copyright 2021 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <atomic>
+#include "core/hle/kernel/k_event.h"
+#include "core/hle/service/service.h"
+
+namespace Core {
+class System;
+}
+
+namespace Service::Account {
+
+class IAsyncContext : public ServiceFramework<IAsyncContext> {
+public:
+    explicit IAsyncContext(Core::System& system_);
+
+    void GetSystemEvent(Kernel::HLERequestContext& ctx);
+    void Cancel(Kernel::HLERequestContext& ctx);
+    void HasDone(Kernel::HLERequestContext& ctx);
+    void GetResult(Kernel::HLERequestContext& ctx);
+
+protected:
+    virtual bool IsComplete() const = 0;
+    virtual void Cancel() = 0;
+    virtual ResultCode GetResult() const = 0;
+
+    void MarkComplete();
+
+    std::atomic<bool> is_complete{false};
+    Kernel::KEvent compeletion_event;
+};
+
+} // namespace Service::Account

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

@@ -1270,7 +1270,8 @@ void ILibraryAppletCreator::CreateHandleStorage(Kernel::HLERequestContext& ctx)
 IApplicationFunctions::IApplicationFunctions(Core::System& system_)
     : ServiceFramework{system_, "IApplicationFunctions"}, gpu_error_detected_event{system.Kernel()},
       friend_invitation_storage_channel_event{system.Kernel()},
-      health_warning_disappeared_system_event{system.Kernel()} {
+      notification_storage_channel_event{system.Kernel()}, health_warning_disappeared_system_event{
+                                                               system.Kernel()} {
     // clang-format off
     static const FunctionInfo functions[] = {
         {1, &IApplicationFunctions::PopLaunchParameter, "PopLaunchParameter"},
@@ -1322,7 +1323,7 @@ IApplicationFunctions::IApplicationFunctions(Core::System& system_)
         {131, nullptr, "SetDelayTimeToAbortOnGpuError"},
         {140, &IApplicationFunctions::GetFriendInvitationStorageChannelEvent, "GetFriendInvitationStorageChannelEvent"},
         {141, &IApplicationFunctions::TryPopFromFriendInvitationStorageChannel, "TryPopFromFriendInvitationStorageChannel"},
-        {150, nullptr, "GetNotificationStorageChannelEvent"},
+        {150, &IApplicationFunctions::GetNotificationStorageChannelEvent, "GetNotificationStorageChannelEvent"},
         {151, nullptr, "TryPopFromNotificationStorageChannel"},
         {160, &IApplicationFunctions::GetHealthWarningDisappearedSystemEvent, "GetHealthWarningDisappearedSystemEvent"},
         {170, nullptr, "SetHdcpAuthenticationActivated"},
@@ -1340,11 +1341,14 @@ IApplicationFunctions::IApplicationFunctions(Core::System& system_)
 
     Kernel::KAutoObject::Create(std::addressof(gpu_error_detected_event));
     Kernel::KAutoObject::Create(std::addressof(friend_invitation_storage_channel_event));
+    Kernel::KAutoObject::Create(std::addressof(notification_storage_channel_event));
     Kernel::KAutoObject::Create(std::addressof(health_warning_disappeared_system_event));
 
     gpu_error_detected_event.Initialize("IApplicationFunctions:GpuErrorDetectedSystemEvent");
     friend_invitation_storage_channel_event.Initialize(
         "IApplicationFunctions:FriendInvitationStorageChannelEvent");
+    notification_storage_channel_event.Initialize(
+        "IApplicationFunctions:NotificationStorageChannelEvent");
     health_warning_disappeared_system_event.Initialize(
         "IApplicationFunctions:HealthWarningDisappearedSystemEvent");
 }
@@ -1762,6 +1766,14 @@ void IApplicationFunctions::TryPopFromFriendInvitationStorageChannel(
     rb.Push(ERR_NO_DATA_IN_CHANNEL);
 }
 
+void IApplicationFunctions::GetNotificationStorageChannelEvent(Kernel::HLERequestContext& ctx) {
+    LOG_DEBUG(Service_AM, "called");
+
+    IPC::ResponseBuilder rb{ctx, 2, 1};
+    rb.Push(ResultSuccess);
+    rb.PushCopyObjects(notification_storage_channel_event.GetReadableEvent());
+}
+
 void IApplicationFunctions::GetHealthWarningDisappearedSystemEvent(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_AM, "called");
 

src/core/hle/service/am/am.h

@@ -295,6 +295,7 @@ private:
     void GetGpuErrorDetectedSystemEvent(Kernel::HLERequestContext& ctx);
     void GetFriendInvitationStorageChannelEvent(Kernel::HLERequestContext& ctx);
     void TryPopFromFriendInvitationStorageChannel(Kernel::HLERequestContext& ctx);
+    void GetNotificationStorageChannelEvent(Kernel::HLERequestContext& ctx);
     void GetHealthWarningDisappearedSystemEvent(Kernel::HLERequestContext& ctx);
 
     bool launch_popped_application_specific = false;
@@ -302,6 +303,7 @@ private:
     s32 previous_program_index{-1};
     Kernel::KEvent gpu_error_detected_event;
     Kernel::KEvent friend_invitation_storage_channel_event;
+    Kernel::KEvent notification_storage_channel_event;
     Kernel::KEvent health_warning_disappeared_system_event;
 };
 

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

@@ -97,14 +97,24 @@ ResultCode VfsDirectoryServiceWrapper::DeleteFile(const std::string& path_) cons
 
 ResultCode VfsDirectoryServiceWrapper::CreateDirectory(const std::string& path_) const {
     std::string path(Common::FS::SanitizePath(path_));
-    auto dir = GetDirectoryRelativeWrapped(backing, Common::FS::GetParentPath(path));
-    if (dir == nullptr || Common::FS::GetFilename(Common::FS::GetParentPath(path)).empty()) {
-        dir = backing;
-    }
-    auto new_dir = dir->CreateSubdirectory(Common::FS::GetFilename(path));
-    if (new_dir == nullptr) {
-        // TODO(DarkLordZach): Find a better error code for this
-        return ResultUnknown;
+
+    // NOTE: This is inaccurate behavior. CreateDirectory is not recursive.
+    // CreateDirectory should return PathNotFound if the parent directory does not exist.
+    // This is here temporarily in order to have UMM "work" in the meantime.
+    // TODO (Morph): Remove this when a hardware test verifies the correct behavior.
+    const auto components = Common::FS::SplitPathComponents(path);
+    std::string relative_path;
+    for (const auto& component : components) {
+        // Skip empty path components
+        if (component.empty()) {
+            continue;
+        }
+        relative_path = Common::FS::SanitizePath(relative_path + '/' + component);
+        auto new_dir = backing->CreateSubdirectory(relative_path);
+        if (new_dir == nullptr) {
+            // TODO(DarkLordZach): Find a better error code for this
+            return ResultUnknown;
+        }
     }
     return ResultSuccess;
 }
@@ -251,6 +261,18 @@ ResultVal<FileSys::EntryType> VfsDirectoryServiceWrapper::GetEntryType(
     return FileSys::ERROR_PATH_NOT_FOUND;
 }
 
+ResultVal<FileSys::FileTimeStampRaw> VfsDirectoryServiceWrapper::GetFileTimeStampRaw(
+    const std::string& path) const {
+    auto dir = GetDirectoryRelativeWrapped(backing, Common::FS::GetParentPath(path));
+    if (dir == nullptr) {
+        return FileSys::ERROR_PATH_NOT_FOUND;
+    }
+    if (GetEntryType(path).Failed()) {
+        return FileSys::ERROR_PATH_NOT_FOUND;
+    }
+    return MakeResult(dir->GetFileTimeStamp(Common::FS::GetFilename(path)));
+}
+
 FileSystemController::FileSystemController(Core::System& system_) : system{system_} {}
 
 FileSystemController::~FileSystemController() = default;

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

@@ -240,6 +240,12 @@ public:
      */
     ResultVal<FileSys::EntryType> GetEntryType(const std::string& path) const;
 
+    /**
+     * Get the timestamp of the specified path
+     * @return The timestamp of the specified path or error code
+     */
+    ResultVal<FileSys::FileTimeStampRaw> GetFileTimeStampRaw(const std::string& path) const;
+
 private:
     FileSys::VirtualDir backing;
 };

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

@@ -326,7 +326,7 @@ public:
             {11, &IFileSystem::GetFreeSpaceSize, "GetFreeSpaceSize"},
             {12, &IFileSystem::GetTotalSpaceSize, "GetTotalSpaceSize"},
             {13, &IFileSystem::CleanDirectoryRecursively, "CleanDirectoryRecursively"},
-            {14, nullptr, "GetFileTimeStampRaw"},
+            {14, &IFileSystem::GetFileTimeStampRaw, "GetFileTimeStampRaw"},
             {15, nullptr, "QueryEntry"},
         };
         RegisterHandlers(functions);
@@ -501,6 +501,24 @@ public:
         rb.Push(size.get_total_size());
     }
 
+    void GetFileTimeStampRaw(Kernel::HLERequestContext& ctx) {
+        const auto file_buffer = ctx.ReadBuffer();
+        const std::string name = Common::StringFromBuffer(file_buffer);
+
+        LOG_WARNING(Service_FS, "(Partial Implementation) called. file={}", name);
+
+        auto result = backend.GetFileTimeStampRaw(name);
+        if (result.Failed()) {
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(result.Code());
+            return;
+        }
+
+        IPC::ResponseBuilder rb{ctx, 10};
+        rb.Push(ResultSuccess);
+        rb.PushRaw(*result);
+    }
+
 private:
     VfsDirectoryServiceWrapper backend;
     SizeGetter size;

src/core/hle/service/hid/controllers/touchscreen.h

@@ -15,6 +15,20 @@
 namespace Service::HID {
 class Controller_Touchscreen final : public ControllerBase {
 public:
+    enum class TouchScreenModeForNx : u8 {
+        UseSystemSetting,
+        Finger,
+        Heat2,
+    };
+
+    struct TouchScreenConfigurationForNx {
+        TouchScreenModeForNx mode;
+        INSERT_PADDING_BYTES_NOINIT(0x7);
+        INSERT_PADDING_BYTES_NOINIT(0xF); // Reserved
+    };
+    static_assert(sizeof(TouchScreenConfigurationForNx) == 0x17,
+                  "TouchScreenConfigurationForNx is an invalid size");
+
     explicit Controller_Touchscreen(Core::System& system_);
     ~Controller_Touchscreen() override;
 

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

@@ -331,7 +331,7 @@ Hid::Hid(Core::System& system_)
         {529, nullptr, "SetDisallowedPalmaConnection"},
         {1000, &Hid::SetNpadCommunicationMode, "SetNpadCommunicationMode"},
         {1001, &Hid::GetNpadCommunicationMode, "GetNpadCommunicationMode"},
-        {1002, nullptr, "SetTouchScreenConfiguration"},
+        {1002, &Hid::SetTouchScreenConfiguration, "SetTouchScreenConfiguration"},
         {1003, nullptr, "IsFirmwareUpdateNeededForNotification"},
         {2000, nullptr, "ActivateDigitizer"},
     };
@@ -1631,6 +1631,18 @@ void Hid::GetNpadCommunicationMode(Kernel::HLERequestContext& ctx) {
                     .GetNpadCommunicationMode());
 }
 
+void Hid::SetTouchScreenConfiguration(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    const auto touchscreen_mode{rp.PopRaw<Controller_Touchscreen::TouchScreenConfigurationForNx>()};
+    const auto applet_resource_user_id{rp.Pop<u64>()};
+
+    LOG_WARNING(Service_HID, "(STUBBED) called, touchscreen_mode={}, applet_resource_user_id={}",
+                touchscreen_mode.mode, applet_resource_user_id);
+
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(ResultSuccess);
+}
+
 class HidDbg final : public ServiceFramework<HidDbg> {
 public:
     explicit HidDbg(Core::System& system_) : ServiceFramework{system_, "hid:dbg"} {

src/core/hle/service/hid/hid.h

@@ -159,6 +159,7 @@ private:
     void SetPalmaBoostMode(Kernel::HLERequestContext& ctx);
     void SetNpadCommunicationMode(Kernel::HLERequestContext& ctx);
     void GetNpadCommunicationMode(Kernel::HLERequestContext& ctx);
+    void SetTouchScreenConfiguration(Kernel::HLERequestContext& ctx);
 
     enum class VibrationDeviceType : u32 {
         Unknown = 0,

src/core/hle/service/ngct/ngct.cpp

@@ -15,7 +15,7 @@ public:
     explicit IService(Core::System& system_) : ServiceFramework{system_, "ngct:u"} {
         // clang-format off
         static const FunctionInfo functions[] = {
-            {0, nullptr, "Match"},
+            {0, &IService::Match, "Match"},
             {1, &IService::Filter, "Filter"},
         };
         // clang-format on
@@ -24,6 +24,19 @@ public:
     }
 
 private:
+    void Match(Kernel::HLERequestContext& ctx) {
+        const auto buffer = ctx.ReadBuffer();
+        const auto text = Common::StringFromFixedZeroTerminatedBuffer(
+            reinterpret_cast<const char*>(buffer.data()), buffer.size());
+
+        LOG_WARNING(Service_NGCT, "(STUBBED) called, text={}", text);
+
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(ResultSuccess);
+        // Return false since we don't censor anything
+        rb.Push(false);
+    }
+
     void Filter(Kernel::HLERequestContext& ctx) {
         const auto buffer = ctx.ReadBuffer();
         const auto text = Common::StringFromFixedZeroTerminatedBuffer(

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

@@ -42,15 +42,14 @@ void nvdisp_disp0::OnClose(DeviceFD fd) {}
 void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u32 height,
                         u32 stride, NVFlinger::BufferQueue::BufferTransformFlags transform,
                         const Common::Rectangle<int>& crop_rect) {
-    VAddr addr = nvmap_dev->GetObjectAddress(buffer_handle);
+    const VAddr addr = nvmap_dev->GetObjectAddress(buffer_handle);
     LOG_TRACE(Service,
               "Drawing from address {:X} offset {:08X} Width {} Height {} Stride {} Format {}",
               addr, offset, width, height, stride, format);
 
-    using PixelFormat = Tegra::FramebufferConfig::PixelFormat;
-    const Tegra::FramebufferConfig framebuffer{
-        addr,      offset,   width, height, stride, static_cast<PixelFormat>(format),
-        transform, crop_rect};
+    const auto pixel_format = static_cast<Tegra::FramebufferConfig::PixelFormat>(format);
+    const Tegra::FramebufferConfig framebuffer{addr,   offset,       width,     height,
+                                               stride, pixel_format, transform, crop_rect};
 
     system.GetPerfStats().EndSystemFrame();
     system.GPU().SwapBuffers(&framebuffer);

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

@@ -9,17 +9,20 @@
 #include "core/core.h"
 #include "core/hle/kernel/k_writable_event.h"
 #include "core/hle/kernel/kernel.h"
+#include "core/hle/service/kernel_helpers.h"
 #include "core/hle/service/nvflinger/buffer_queue.h"
 
 namespace Service::NVFlinger {
 
-BufferQueue::BufferQueue(Kernel::KernelCore& kernel, u32 id_, u64 layer_id_)
-    : id(id_), layer_id(layer_id_), buffer_wait_event{kernel} {
-    Kernel::KAutoObject::Create(std::addressof(buffer_wait_event));
-    buffer_wait_event.Initialize("BufferQueue:WaitEvent");
+BufferQueue::BufferQueue(Kernel::KernelCore& kernel, u32 id_, u64 layer_id_,
+                         KernelHelpers::ServiceContext& service_context_)
+    : id(id_), layer_id(layer_id_), service_context{service_context_} {
+    buffer_wait_event = service_context.CreateEvent("BufferQueue:WaitEvent");
 }
 
-BufferQueue::~BufferQueue() = default;
+BufferQueue::~BufferQueue() {
+    service_context.CloseEvent(buffer_wait_event);
+}
 
 void BufferQueue::SetPreallocatedBuffer(u32 slot, const IGBPBuffer& igbp_buffer) {
     ASSERT(slot < buffer_slots);
@@ -41,7 +44,7 @@ void BufferQueue::SetPreallocatedBuffer(u32 slot, const IGBPBuffer& igbp_buffer)
         .multi_fence = {},
     };
 
-    buffer_wait_event.GetWritableEvent().Signal();
+    buffer_wait_event->GetWritableEvent().Signal();
 }
 
 std::optional<std::pair<u32, Service::Nvidia::MultiFence*>> BufferQueue::DequeueBuffer(u32 width,
@@ -119,7 +122,7 @@ void BufferQueue::CancelBuffer(u32 slot, const Service::Nvidia::MultiFence& mult
     }
     free_buffers_condition.notify_one();
 
-    buffer_wait_event.GetWritableEvent().Signal();
+    buffer_wait_event->GetWritableEvent().Signal();
 }
 
 std::optional<std::reference_wrapper<const BufferQueue::Buffer>> BufferQueue::AcquireBuffer() {
@@ -154,7 +157,7 @@ void BufferQueue::ReleaseBuffer(u32 slot) {
     }
     free_buffers_condition.notify_one();
 
-    buffer_wait_event.GetWritableEvent().Signal();
+    buffer_wait_event->GetWritableEvent().Signal();
 }
 
 void BufferQueue::Connect() {
@@ -169,7 +172,7 @@ void BufferQueue::Disconnect() {
         std::unique_lock lock{queue_sequence_mutex};
         queue_sequence.clear();
     }
-    buffer_wait_event.GetWritableEvent().Signal();
+    buffer_wait_event->GetWritableEvent().Signal();
     is_connect = false;
     free_buffers_condition.notify_one();
 }
@@ -189,11 +192,11 @@ u32 BufferQueue::Query(QueryType type) {
 }
 
 Kernel::KWritableEvent& BufferQueue::GetWritableBufferWaitEvent() {
-    return buffer_wait_event.GetWritableEvent();
+    return buffer_wait_event->GetWritableEvent();
 }
 
 Kernel::KReadableEvent& BufferQueue::GetBufferWaitEvent() {
-    return buffer_wait_event.GetReadableEvent();
+    return buffer_wait_event->GetReadableEvent();
 }
 
 } // namespace Service::NVFlinger

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

@@ -24,6 +24,10 @@ class KReadableEvent;
 class KWritableEvent;
 } // namespace Kernel
 
+namespace Service::KernelHelpers {
+class ServiceContext;
+} // namespace Service::KernelHelpers
+
 namespace Service::NVFlinger {
 
 constexpr u32 buffer_slots = 0x40;
@@ -38,7 +42,9 @@ struct IGBPBuffer {
     u32_le index;
     INSERT_PADDING_WORDS(3);
     u32_le gpu_buffer_id;
-    INSERT_PADDING_WORDS(17);
+    INSERT_PADDING_WORDS(6);
+    u32_le external_format;
+    INSERT_PADDING_WORDS(10);
     u32_le nvmap_handle;
     u32_le offset;
     INSERT_PADDING_WORDS(60);
@@ -54,7 +60,8 @@ public:
         NativeWindowFormat = 2,
     };
 
-    explicit BufferQueue(Kernel::KernelCore& kernel, u32 id_, u64 layer_id_);
+    explicit BufferQueue(Kernel::KernelCore& kernel, u32 id_, u64 layer_id_,
+                         KernelHelpers::ServiceContext& service_context_);
     ~BufferQueue();
 
     enum class BufferTransformFlags : u32 {
@@ -130,12 +137,14 @@ private:
     std::list<u32> free_buffers;
     std::array<Buffer, buffer_slots> buffers;
     std::list<u32> queue_sequence;
-    Kernel::KEvent buffer_wait_event;
+    Kernel::KEvent* buffer_wait_event{};
 
     std::mutex free_buffers_mutex;
     std::condition_variable free_buffers_condition;
 
     std::mutex queue_sequence_mutex;
+
+    KernelHelpers::ServiceContext& service_context;
 };
 
 } // namespace Service::NVFlinger

src/core/hle/service/nvflinger/nvflinger.cpp

@@ -61,12 +61,13 @@ void NVFlinger::SplitVSync() {
     }
 }
 
-NVFlinger::NVFlinger(Core::System& system_) : system(system_) {
-    displays.emplace_back(0, "Default", system);
-    displays.emplace_back(1, "External", system);
-    displays.emplace_back(2, "Edid", system);
-    displays.emplace_back(3, "Internal", system);
-    displays.emplace_back(4, "Null", system);
+NVFlinger::NVFlinger(Core::System& system_)
+    : system(system_), service_context(system_, "nvflinger") {
+    displays.emplace_back(0, "Default", service_context, system);
+    displays.emplace_back(1, "External", service_context, system);
+    displays.emplace_back(2, "Edid", service_context, system);
+    displays.emplace_back(3, "Internal", service_context, system);
+    displays.emplace_back(4, "Null", service_context, system);
     guard = std::make_shared<std::mutex>();
 
     // Schedule the screen composition events
@@ -146,7 +147,7 @@ std::optional<u64> NVFlinger::CreateLayer(u64 display_id) {
 void NVFlinger::CreateLayerAtId(VI::Display& display, u64 layer_id) {
     const u32 buffer_queue_id = next_buffer_queue_id++;
     buffer_queues.emplace_back(
-        std::make_unique<BufferQueue>(system.Kernel(), buffer_queue_id, layer_id));
+        std::make_unique<BufferQueue>(system.Kernel(), buffer_queue_id, layer_id, service_context));
     display.CreateLayer(layer_id, *buffer_queues.back());
 }
 
@@ -297,7 +298,7 @@ void NVFlinger::Compose() {
         auto nvdisp = nvdrv->GetDevice<Nvidia::Devices::nvdisp_disp0>("/dev/nvdisp_disp0");
         ASSERT(nvdisp);
 
-        nvdisp->flip(igbp_buffer.gpu_buffer_id, igbp_buffer.offset, igbp_buffer.format,
+        nvdisp->flip(igbp_buffer.gpu_buffer_id, igbp_buffer.offset, igbp_buffer.external_format,
                      igbp_buffer.width, igbp_buffer.height, igbp_buffer.stride,
                      buffer->get().transform, buffer->get().crop_rect);
 

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

@@ -15,6 +15,7 @@
 #include <vector>
 
 #include "common/common_types.h"
+#include "core/hle/service/kernel_helpers.h"
 
 namespace Common {
 class Event;
@@ -135,6 +136,8 @@ private:
     std::unique_ptr<std::thread> vsync_thread;
     std::unique_ptr<Common::Event> wait_event;
     std::atomic<bool> is_running{};
+
+    KernelHelpers::ServiceContext service_context;
 };
 
 } // namespace Service::NVFlinger

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

@@ -12,18 +12,21 @@
 #include "core/hle/kernel/k_event.h"
 #include "core/hle/kernel/k_readable_event.h"
 #include "core/hle/kernel/k_writable_event.h"
+#include "core/hle/service/kernel_helpers.h"
 #include "core/hle/service/vi/display/vi_display.h"
 #include "core/hle/service/vi/layer/vi_layer.h"
 
 namespace Service::VI {
 
-Display::Display(u64 id, std::string name_, Core::System& system)
-    : display_id{id}, name{std::move(name_)}, vsync_event{system.Kernel()} {
-    Kernel::KAutoObject::Create(std::addressof(vsync_event));
-    vsync_event.Initialize(fmt::format("Display VSync Event {}", id));
+Display::Display(u64 id, std::string name_, KernelHelpers::ServiceContext& service_context_,
+                 Core::System& system_)
+    : display_id{id}, name{std::move(name_)}, service_context{service_context_} {
+    vsync_event = service_context.CreateEvent(fmt::format("Display VSync Event {}", id));
 }
 
-Display::~Display() = default;
+Display::~Display() {
+    service_context.CloseEvent(vsync_event);
+}
 
 Layer& Display::GetLayer(std::size_t index) {
     return *layers.at(index);
@@ -34,11 +37,11 @@ const Layer& Display::GetLayer(std::size_t index) const {
 }
 
 Kernel::KReadableEvent& Display::GetVSyncEvent() {
-    return vsync_event.GetReadableEvent();
+    return vsync_event->GetReadableEvent();
 }
 
 void Display::SignalVSyncEvent() {
-    vsync_event.GetWritableEvent().Signal();
+    vsync_event->GetWritableEvent().Signal();
 }
 
 void Display::CreateLayer(u64 layer_id, NVFlinger::BufferQueue& buffer_queue) {

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

@@ -18,6 +18,9 @@ class KEvent;
 namespace Service::NVFlinger {
 class BufferQueue;
 }
+namespace Service::KernelHelpers {
+class ServiceContext;
+} // namespace Service::KernelHelpers
 
 namespace Service::VI {
 
@@ -31,10 +34,13 @@ class Display {
 public:
     /// Constructs a display with a given unique ID and name.
     ///
-    /// @param id   The unique ID for this display.
+    /// @param id The unique ID for this display.
+    /// @param service_context_ The ServiceContext for the owning service.
     /// @param name_ The name for this display.
+    /// @param system_ The global system instance.
     ///
-    Display(u64 id, std::string name_, Core::System& system);
+    Display(u64 id, std::string name_, KernelHelpers::ServiceContext& service_context_,
+            Core::System& system_);
     ~Display();
 
     /// Gets the unique ID assigned to this display.
@@ -98,9 +104,10 @@ public:
 private:
     u64 display_id;
     std::string name;
+    KernelHelpers::ServiceContext& service_context;
 
     std::vector<std::shared_ptr<Layer>> layers;
-    Kernel::KEvent vsync_event;
+    Kernel::KEvent* vsync_event{};
 };
 
 } // namespace Service::VI

src/core/network/network.cpp

@@ -7,7 +7,8 @@
 #include <limits>
 #include <utility>
 #include <vector>
-#include "common/common_funcs.h"
+
+#include "common/error.h"
 
 #ifdef _WIN32
 #include <winsock2.h>
@@ -223,7 +224,7 @@ Errno GetAndLogLastError() {
     if (err == Errno::AGAIN) {
         return err;
     }
-    LOG_ERROR(Network, "Socket operation error: {}", NativeErrorToString(e));
+    LOG_ERROR(Network, "Socket operation error: {}", Common::NativeErrorToString(e));
     return err;
 }
 

src/core/telemetry_session.cpp

@@ -72,6 +72,18 @@ static const char* TranslateGPUAccuracyLevel(Settings::GPUAccuracy backend) {
     return "Unknown";
 }
 
+static const char* TranslateNvdecEmulation(Settings::NvdecEmulation backend) {
+    switch (backend) {
+    case Settings::NvdecEmulation::Off:
+        return "Off";
+    case Settings::NvdecEmulation::CPU:
+        return "CPU";
+    case Settings::NvdecEmulation::GPU:
+        return "GPU";
+    }
+    return "Unknown";
+}
+
 u64 GetTelemetryId() {
     u64 telemetry_id{};
     const auto filename = Common::FS::GetYuzuPath(Common::FS::YuzuPath::ConfigDir) / "telemetry_id";
@@ -214,8 +226,6 @@ void TelemetrySession::AddInitialInfo(Loader::AppLoader& app_loader,
     // Log user configuration information
     constexpr auto field_type = Telemetry::FieldType::UserConfig;
     AddField(field_type, "Audio_SinkId", Settings::values.sink_id.GetValue());
-    AddField(field_type, "Audio_EnableAudioStretching",
-             Settings::values.enable_audio_stretching.GetValue());
     AddField(field_type, "Core_UseMultiCore", Settings::values.use_multi_core.GetValue());
     AddField(field_type, "Renderer_Backend",
              TranslateRenderer(Settings::values.renderer_backend.GetValue()));
@@ -229,8 +239,8 @@ void TelemetrySession::AddInitialInfo(Loader::AppLoader& app_loader,
              TranslateGPUAccuracyLevel(Settings::values.gpu_accuracy.GetValue()));
     AddField(field_type, "Renderer_UseAsynchronousGpuEmulation",
              Settings::values.use_asynchronous_gpu_emulation.GetValue());
-    AddField(field_type, "Renderer_UseNvdecEmulation",
-             Settings::values.use_nvdec_emulation.GetValue());
+    AddField(field_type, "Renderer_NvdecEmulation",
+             TranslateNvdecEmulation(Settings::values.nvdec_emulation.GetValue()));
     AddField(field_type, "Renderer_AccelerateASTC", Settings::values.accelerate_astc.GetValue());
     AddField(field_type, "Renderer_UseVsync", Settings::values.use_vsync.GetValue());
     AddField(field_type, "Renderer_ShaderBackend",

src/input_common/CMakeLists.txt

@@ -21,6 +21,10 @@ add_library(input_common STATIC
     mouse/mouse_poller.h
     sdl/sdl.cpp
     sdl/sdl.h
+    tas/tas_input.cpp
+    tas/tas_input.h
+    tas/tas_poller.cpp
+    tas/tas_poller.h
     udp/client.cpp
     udp/client.h
     udp/protocol.cpp

src/input_common/main.cpp

@@ -5,6 +5,7 @@
 #include <memory>
 #include <thread>
 #include "common/param_package.h"
+#include "common/settings.h"
 #include "input_common/analog_from_button.h"
 #include "input_common/gcadapter/gc_adapter.h"
 #include "input_common/gcadapter/gc_poller.h"
@@ -13,6 +14,8 @@
 #include "input_common/motion_from_button.h"
 #include "input_common/mouse/mouse_input.h"
 #include "input_common/mouse/mouse_poller.h"
+#include "input_common/tas/tas_input.h"
+#include "input_common/tas/tas_poller.h"
 #include "input_common/touch_from_button.h"
 #include "input_common/udp/client.h"
 #include "input_common/udp/udp.h"
@@ -60,6 +63,12 @@ struct InputSubsystem::Impl {
         Input::RegisterFactory<Input::MotionDevice>("mouse", mousemotion);
         mousetouch = std::make_shared<MouseTouchFactory>(mouse);
         Input::RegisterFactory<Input::TouchDevice>("mouse", mousetouch);
+
+        tas = std::make_shared<TasInput::Tas>();
+        tasbuttons = std::make_shared<TasButtonFactory>(tas);
+        Input::RegisterFactory<Input::ButtonDevice>("tas", tasbuttons);
+        tasanalog = std::make_shared<TasAnalogFactory>(tas);
+        Input::RegisterFactory<Input::AnalogDevice>("tas", tasanalog);
     }
 
     void Shutdown() {
@@ -94,6 +103,12 @@ struct InputSubsystem::Impl {
         mouseanalog.reset();
         mousemotion.reset();
         mousetouch.reset();
+
+        Input::UnregisterFactory<Input::ButtonDevice>("tas");
+        Input::UnregisterFactory<Input::AnalogDevice>("tas");
+
+        tasbuttons.reset();
+        tasanalog.reset();
     }
 
     [[nodiscard]] std::vector<Common::ParamPackage> GetInputDevices() const {
@@ -101,6 +116,10 @@ struct InputSubsystem::Impl {
             Common::ParamPackage{{"display", "Any"}, {"class", "any"}},
             Common::ParamPackage{{"display", "Keyboard/Mouse"}, {"class", "keyboard"}},
         };
+        if (Settings::values.tas_enable) {
+            devices.emplace_back(
+                Common::ParamPackage{{"display", "TAS Controller"}, {"class", "tas"}});
+        }
 #ifdef HAVE_SDL2
         auto sdl_devices = sdl->GetInputDevices();
         devices.insert(devices.end(), sdl_devices.begin(), sdl_devices.end());
@@ -120,6 +139,9 @@ struct InputSubsystem::Impl {
         if (params.Get("class", "") == "gcpad") {
             return gcadapter->GetAnalogMappingForDevice(params);
         }
+        if (params.Get("class", "") == "tas") {
+            return tas->GetAnalogMappingForDevice(params);
+        }
 #ifdef HAVE_SDL2
         if (params.Get("class", "") == "sdl") {
             return sdl->GetAnalogMappingForDevice(params);
@@ -136,6 +158,9 @@ struct InputSubsystem::Impl {
         if (params.Get("class", "") == "gcpad") {
             return gcadapter->GetButtonMappingForDevice(params);
         }
+        if (params.Get("class", "") == "tas") {
+            return tas->GetButtonMappingForDevice(params);
+        }
 #ifdef HAVE_SDL2
         if (params.Get("class", "") == "sdl") {
             return sdl->GetButtonMappingForDevice(params);
@@ -174,9 +199,12 @@ struct InputSubsystem::Impl {
     std::shared_ptr<MouseAnalogFactory> mouseanalog;
     std::shared_ptr<MouseMotionFactory> mousemotion;
     std::shared_ptr<MouseTouchFactory> mousetouch;
+    std::shared_ptr<TasButtonFactory> tasbuttons;
+    std::shared_ptr<TasAnalogFactory> tasanalog;
     std::shared_ptr<CemuhookUDP::Client> udp;
     std::shared_ptr<GCAdapter::Adapter> gcadapter;
     std::shared_ptr<MouseInput::Mouse> mouse;
+    std::shared_ptr<TasInput::Tas> tas;
 };
 
 InputSubsystem::InputSubsystem() : impl{std::make_unique<Impl>()} {}
@@ -207,6 +235,14 @@ const MouseInput::Mouse* InputSubsystem::GetMouse() const {
     return impl->mouse.get();
 }
 
+TasInput::Tas* InputSubsystem::GetTas() {
+    return impl->tas.get();
+}
+
+const TasInput::Tas* InputSubsystem::GetTas() const {
+    return impl->tas.get();
+}
+
 std::vector<Common::ParamPackage> InputSubsystem::GetInputDevices() const {
     return impl->GetInputDevices();
 }
@@ -287,6 +323,22 @@ const MouseTouchFactory* InputSubsystem::GetMouseTouch() const {
     return impl->mousetouch.get();
 }
 
+TasButtonFactory* InputSubsystem::GetTasButtons() {
+    return impl->tasbuttons.get();
+}
+
+const TasButtonFactory* InputSubsystem::GetTasButtons() const {
+    return impl->tasbuttons.get();
+}
+
+TasAnalogFactory* InputSubsystem::GetTasAnalogs() {
+    return impl->tasanalog.get();
+}
+
+const TasAnalogFactory* InputSubsystem::GetTasAnalogs() const {
+    return impl->tasanalog.get();
+}
+
 void InputSubsystem::ReloadInputDevices() {
     if (!impl->udp) {
         return;

src/input_common/main.h

@@ -29,6 +29,10 @@ namespace MouseInput {
 class Mouse;
 }
 
+namespace TasInput {
+class Tas;
+}
+
 namespace InputCommon {
 namespace Polling {
 
@@ -64,6 +68,8 @@ class MouseButtonFactory;
 class MouseAnalogFactory;
 class MouseMotionFactory;
 class MouseTouchFactory;
+class TasButtonFactory;
+class TasAnalogFactory;
 class Keyboard;
 
 /**
@@ -103,6 +109,11 @@ public:
     /// Retrieves the underlying mouse device.
     [[nodiscard]] const MouseInput::Mouse* GetMouse() const;
 
+    /// Retrieves the underlying tas device.
+    [[nodiscard]] TasInput::Tas* GetTas();
+
+    /// Retrieves the underlying tas device.
+    [[nodiscard]] const TasInput::Tas* GetTas() const;
     /**
      * Returns all available input devices that this Factory can create a new device with.
      * Each returned ParamPackage should have a `display` field used for display, a class field for
@@ -144,30 +155,42 @@ public:
     /// Retrieves the underlying udp touch handler.
     [[nodiscard]] const UDPTouchFactory* GetUDPTouch() const;
 
-    /// Retrieves the underlying GameCube button handler.
+    /// Retrieves the underlying mouse button handler.
     [[nodiscard]] MouseButtonFactory* GetMouseButtons();
 
-    /// Retrieves the underlying GameCube button handler.
+    /// Retrieves the underlying mouse button handler.
     [[nodiscard]] const MouseButtonFactory* GetMouseButtons() const;
 
-    /// Retrieves the underlying udp touch handler.
+    /// Retrieves the underlying mouse analog handler.
     [[nodiscard]] MouseAnalogFactory* GetMouseAnalogs();
 
-    /// Retrieves the underlying udp touch handler.
+    /// Retrieves the underlying mouse analog handler.
     [[nodiscard]] const MouseAnalogFactory* GetMouseAnalogs() const;
 
-    /// Retrieves the underlying udp motion handler.
+    /// Retrieves the underlying mouse motion handler.
     [[nodiscard]] MouseMotionFactory* GetMouseMotions();
 
-    /// Retrieves the underlying udp motion handler.
+    /// Retrieves the underlying mouse motion handler.
     [[nodiscard]] const MouseMotionFactory* GetMouseMotions() const;
 
-    /// Retrieves the underlying udp touch handler.
+    /// Retrieves the underlying mouse touch handler.
     [[nodiscard]] MouseTouchFactory* GetMouseTouch();
 
-    /// Retrieves the underlying udp touch handler.
+    /// Retrieves the underlying mouse touch handler.
     [[nodiscard]] const MouseTouchFactory* GetMouseTouch() const;
 
+    /// Retrieves the underlying tas button handler.
+    [[nodiscard]] TasButtonFactory* GetTasButtons();
+
+    /// Retrieves the underlying tas button handler.
+    [[nodiscard]] const TasButtonFactory* GetTasButtons() const;
+
+    /// Retrieves the underlying tas analogs handler.
+    [[nodiscard]] TasAnalogFactory* GetTasAnalogs();
+
+    /// Retrieves the underlying tas analogs handler.
+    [[nodiscard]] const TasAnalogFactory* GetTasAnalogs() const;
+
     /// Reloads the input devices
     void ReloadInputDevices();
 

src/input_common/sdl/sdl_impl.cpp

@@ -21,7 +21,7 @@
 #include "common/logging/log.h"
 #include "common/math_util.h"
 #include "common/param_package.h"
-#include "common/settings_input.h"
+#include "common/settings.h"
 #include "common/threadsafe_queue.h"
 #include "core/frontend/input.h"
 #include "input_common/motion_input.h"
@@ -889,8 +889,10 @@ SDLState::SDLState() {
     RegisterFactory<VibrationDevice>("sdl", vibration_factory);
     RegisterFactory<MotionDevice>("sdl", motion_factory);
 
-    // Disable raw input. When enabled this setting causes SDL to die when a web applet opens
-    SDL_SetHint(SDL_HINT_JOYSTICK_RAWINPUT, "0");
+    if (!Settings::values.enable_raw_input) {
+        // Disable raw input. When enabled this setting causes SDL to die when a web applet opens
+        SDL_SetHint(SDL_HINT_JOYSTICK_RAWINPUT, "0");
+    }
 
     // Enable HIDAPI rumble. This prevents SDL from disabling motion on PS4 and PS5 controllers
     SDL_SetHint(SDL_HINT_JOYSTICK_HIDAPI_PS4_RUMBLE, "1");
@@ -898,10 +900,10 @@ SDLState::SDLState() {
 
     // Tell SDL2 to use the hidapi driver. This will allow joycons to be detected as a
     // GameController and not a generic one
-    SDL_SetHint("SDL_JOYSTICK_HIDAPI_JOY_CONS", "1");
+    SDL_SetHint(SDL_HINT_JOYSTICK_HIDAPI_JOY_CONS, "1");
 
     // Turn off Pro controller home led
-    SDL_SetHint("SDL_JOYSTICK_HIDAPI_SWITCH_HOME_LED", "0");
+    SDL_SetHint(SDL_HINT_JOYSTICK_HIDAPI_SWITCH_HOME_LED, "0");
 
     // If the frontend is going to manage the event loop, then we don't start one here
     start_thread = SDL_WasInit(SDL_INIT_JOYSTICK) == 0;

src/input_common/tas/tas_input.cpp

@@ -0,0 +1,455 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2+
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include <regex>
+
+#include "common/fs/file.h"
+#include "common/fs/fs_types.h"
+#include "common/fs/path_util.h"
+#include "common/logging/log.h"
+#include "common/settings.h"
+#include "input_common/tas/tas_input.h"
+
+namespace TasInput {
+
+// Supported keywords and buttons from a TAS file
+constexpr std::array<std::pair<std::string_view, TasButton>, 20> text_to_tas_button = {
+    std::pair{"KEY_A", TasButton::BUTTON_A},
+    {"KEY_B", TasButton::BUTTON_B},
+    {"KEY_X", TasButton::BUTTON_X},
+    {"KEY_Y", TasButton::BUTTON_Y},
+    {"KEY_LSTICK", TasButton::STICK_L},
+    {"KEY_RSTICK", TasButton::STICK_R},
+    {"KEY_L", TasButton::TRIGGER_L},
+    {"KEY_R", TasButton::TRIGGER_R},
+    {"KEY_PLUS", TasButton::BUTTON_PLUS},
+    {"KEY_MINUS", TasButton::BUTTON_MINUS},
+    {"KEY_DLEFT", TasButton::BUTTON_LEFT},
+    {"KEY_DUP", TasButton::BUTTON_UP},
+    {"KEY_DRIGHT", TasButton::BUTTON_RIGHT},
+    {"KEY_DDOWN", TasButton::BUTTON_DOWN},
+    {"KEY_SL", TasButton::BUTTON_SL},
+    {"KEY_SR", TasButton::BUTTON_SR},
+    {"KEY_CAPTURE", TasButton::BUTTON_CAPTURE},
+    {"KEY_HOME", TasButton::BUTTON_HOME},
+    {"KEY_ZL", TasButton::TRIGGER_ZL},
+    {"KEY_ZR", TasButton::TRIGGER_ZR},
+};
+
+Tas::Tas() {
+    if (!Settings::values.tas_enable) {
+        needs_reset = true;
+        return;
+    }
+    LoadTasFiles();
+}
+
+Tas::~Tas() {
+    Stop();
+};
+
+void Tas::LoadTasFiles() {
+    script_length = 0;
+    for (size_t i = 0; i < commands.size(); i++) {
+        LoadTasFile(i);
+        if (commands[i].size() > script_length) {
+            script_length = commands[i].size();
+        }
+    }
+}
+
+void Tas::LoadTasFile(size_t player_index) {
+    if (!commands[player_index].empty()) {
+        commands[player_index].clear();
+    }
+    std::string file =
+        Common::FS::ReadStringFromFile(Common::FS::GetYuzuPath(Common::FS::YuzuPath::TASDir) /
+                                           fmt::format("script0-{}.txt", player_index + 1),
+                                       Common::FS::FileType::BinaryFile);
+    std::stringstream command_line(file);
+    std::string line;
+    int frame_no = 0;
+    while (std::getline(command_line, line, '\n')) {
+        if (line.empty()) {
+            continue;
+        }
+        LOG_DEBUG(Input, "Loading line: {}", line);
+        std::smatch m;
+
+        std::stringstream linestream(line);
+        std::string segment;
+        std::vector<std::string> seglist;
+
+        while (std::getline(linestream, segment, ' ')) {
+            seglist.push_back(segment);
+        }
+
+        if (seglist.size() < 4) {
+            continue;
+        }
+
+        while (frame_no < std::stoi(seglist.at(0))) {
+            commands[player_index].push_back({});
+            frame_no++;
+        }
+
+        TASCommand command = {
+            .buttons = ReadCommandButtons(seglist.at(1)),
+            .l_axis = ReadCommandAxis(seglist.at(2)),
+            .r_axis = ReadCommandAxis(seglist.at(3)),
+        };
+        commands[player_index].push_back(command);
+        frame_no++;
+    }
+    LOG_INFO(Input, "TAS file loaded! {} frames", frame_no);
+}
+
+void Tas::WriteTasFile(std::u8string file_name) {
+    std::string output_text;
+    for (size_t frame = 0; frame < record_commands.size(); frame++) {
+        if (!output_text.empty()) {
+            output_text += "\n";
+        }
+        const TASCommand& line = record_commands[frame];
+        output_text += std::to_string(frame) + " " + WriteCommandButtons(line.buttons) + " " +
+                       WriteCommandAxis(line.l_axis) + " " + WriteCommandAxis(line.r_axis);
+    }
+    const auto bytes_written = Common::FS::WriteStringToFile(
+        Common::FS::GetYuzuPath(Common::FS::YuzuPath::TASDir) / file_name,
+        Common::FS::FileType::TextFile, output_text);
+    if (bytes_written == output_text.size()) {
+        LOG_INFO(Input, "TAS file written to file!");
+    } else {
+        LOG_ERROR(Input, "Writing the TAS-file has failed! {} / {} bytes written", bytes_written,
+                  output_text.size());
+    }
+}
+
+std::pair<float, float> Tas::FlipAxisY(std::pair<float, float> old) {
+    auto [x, y] = old;
+    return {x, -y};
+}
+
+void Tas::RecordInput(u32 buttons, const std::array<std::pair<float, float>, 2>& axes) {
+    last_input = {buttons, FlipAxisY(axes[0]), FlipAxisY(axes[1])};
+}
+
+std::tuple<TasState, size_t, size_t> Tas::GetStatus() const {
+    TasState state;
+    if (is_recording) {
+        return {TasState::Recording, 0, record_commands.size()};
+    }
+
+    if (is_running) {
+        state = TasState::Running;
+    } else {
+        state = TasState::Stopped;
+    }
+
+    return {state, current_command, script_length};
+}
+
+std::string Tas::DebugButtons(u32 buttons) const {
+    return fmt::format("{{ {} }}", TasInput::Tas::ButtonsToString(buttons));
+}
+
+std::string Tas::DebugJoystick(float x, float y) const {
+    return fmt::format("[ {} , {} ]", std::to_string(x), std::to_string(y));
+}
+
+std::string Tas::DebugInput(const TasData& data) const {
+    return fmt::format("{{ {} , {} , {} }}", DebugButtons(data.buttons),
+                       DebugJoystick(data.axis[0], data.axis[1]),
+                       DebugJoystick(data.axis[2], data.axis[3]));
+}
+
+std::string Tas::DebugInputs(const std::array<TasData, PLAYER_NUMBER>& arr) const {
+    std::string returns = "[ ";
+    for (size_t i = 0; i < arr.size(); i++) {
+        returns += DebugInput(arr[i]);
+        if (i != arr.size() - 1) {
+            returns += " , ";
+        }
+    }
+    return returns + "]";
+}
+
+std::string Tas::ButtonsToString(u32 button) const {
+    std::string returns;
+    for (auto [text_button, tas_button] : text_to_tas_button) {
+        if ((button & static_cast<u32>(tas_button)) != 0)
+            returns += fmt::format(", {}", text_button.substr(4));
+    }
+    return returns.empty() ? "" : returns.substr(2);
+}
+
+void Tas::UpdateThread() {
+    if (!Settings::values.tas_enable) {
+        if (is_running) {
+            Stop();
+        }
+        return;
+    }
+
+    if (is_recording) {
+        record_commands.push_back(last_input);
+    }
+    if (needs_reset) {
+        current_command = 0;
+        needs_reset = false;
+        LoadTasFiles();
+        LOG_DEBUG(Input, "tas_reset done");
+    }
+
+    if (!is_running) {
+        tas_data.fill({});
+        return;
+    }
+    if (current_command < script_length) {
+        LOG_DEBUG(Input, "Playing TAS {}/{}", current_command, script_length);
+        size_t frame = current_command++;
+        for (size_t i = 0; i < commands.size(); i++) {
+            if (frame < commands[i].size()) {
+                TASCommand command = commands[i][frame];
+                tas_data[i].buttons = command.buttons;
+                auto [l_axis_x, l_axis_y] = command.l_axis;
+                tas_data[i].axis[0] = l_axis_x;
+                tas_data[i].axis[1] = l_axis_y;
+                auto [r_axis_x, r_axis_y] = command.r_axis;
+                tas_data[i].axis[2] = r_axis_x;
+                tas_data[i].axis[3] = r_axis_y;
+            } else {
+                tas_data[i] = {};
+            }
+        }
+    } else {
+        is_running = Settings::values.tas_loop.GetValue();
+        current_command = 0;
+        tas_data.fill({});
+        if (!is_running) {
+            SwapToStoredController();
+        }
+    }
+    LOG_DEBUG(Input, "TAS inputs: {}", DebugInputs(tas_data));
+}
+
+TasAnalog Tas::ReadCommandAxis(const std::string& line) const {
+    std::stringstream linestream(line);
+    std::string segment;
+    std::vector<std::string> seglist;
+
+    while (std::getline(linestream, segment, ';')) {
+        seglist.push_back(segment);
+    }
+
+    const float x = std::stof(seglist.at(0)) / 32767.0f;
+    const float y = std::stof(seglist.at(1)) / 32767.0f;
+
+    return {x, y};
+}
+
+u32 Tas::ReadCommandButtons(const std::string& data) const {
+    std::stringstream button_text(data);
+    std::string line;
+    u32 buttons = 0;
+    while (std::getline(button_text, line, ';')) {
+        for (auto [text, tas_button] : text_to_tas_button) {
+            if (text == line) {
+                buttons |= static_cast<u32>(tas_button);
+                break;
+            }
+        }
+    }
+    return buttons;
+}
+
+std::string Tas::WriteCommandAxis(TasAnalog data) const {
+    auto [x, y] = data;
+    std::string line;
+    line += std::to_string(static_cast<int>(x * 32767));
+    line += ";";
+    line += std::to_string(static_cast<int>(y * 32767));
+    return line;
+}
+
+std::string Tas::WriteCommandButtons(u32 data) const {
+    if (data == 0) {
+        return "NONE";
+    }
+
+    std::string line;
+    u32 index = 0;
+    while (data > 0) {
+        if ((data & 1) == 1) {
+            for (auto [text, tas_button] : text_to_tas_button) {
+                if (tas_button == static_cast<TasButton>(1 << index)) {
+                    if (line.size() > 0) {
+                        line += ";";
+                    }
+                    line += text;
+                    break;
+                }
+            }
+        }
+        index++;
+        data >>= 1;
+    }
+    return line;
+}
+
+void Tas::StartStop() {
+    if (!Settings::values.tas_enable) {
+        return;
+    }
+    if (is_running) {
+        Stop();
+    } else {
+        is_running = true;
+        SwapToTasController();
+    }
+}
+
+void Tas::Stop() {
+    is_running = false;
+    SwapToStoredController();
+}
+
+void Tas::SwapToTasController() {
+    if (!Settings::values.tas_swap_controllers) {
+        return;
+    }
+    auto& players = Settings::values.players.GetValue();
+    for (std::size_t index = 0; index < players.size(); index++) {
+        auto& player = players[index];
+        player_mappings[index] = player;
+
+        // Only swap active controllers
+        if (!player.connected) {
+            continue;
+        }
+
+        Common::ParamPackage tas_param;
+        tas_param.Set("pad", static_cast<u8>(index));
+        auto button_mapping = GetButtonMappingForDevice(tas_param);
+        auto analog_mapping = GetAnalogMappingForDevice(tas_param);
+        auto& buttons = player.buttons;
+        auto& analogs = player.analogs;
+
+        for (std::size_t i = 0; i < buttons.size(); ++i) {
+            buttons[i] = button_mapping[static_cast<Settings::NativeButton::Values>(i)].Serialize();
+        }
+        for (std::size_t i = 0; i < analogs.size(); ++i) {
+            analogs[i] = analog_mapping[static_cast<Settings::NativeAnalog::Values>(i)].Serialize();
+        }
+    }
+    is_old_input_saved = true;
+    Settings::values.is_device_reload_pending.store(true);
+}
+
+void Tas::SwapToStoredController() {
+    if (!is_old_input_saved) {
+        return;
+    }
+    auto& players = Settings::values.players.GetValue();
+    for (std::size_t index = 0; index < players.size(); index++) {
+        players[index] = player_mappings[index];
+    }
+    is_old_input_saved = false;
+    Settings::values.is_device_reload_pending.store(true);
+}
+
+void Tas::Reset() {
+    if (!Settings::values.tas_enable) {
+        return;
+    }
+    needs_reset = true;
+}
+
+bool Tas::Record() {
+    if (!Settings::values.tas_enable) {
+        return true;
+    }
+    is_recording = !is_recording;
+    return is_recording;
+}
+
+void Tas::SaveRecording(bool overwrite_file) {
+    if (is_recording) {
+        return;
+    }
+    if (record_commands.empty()) {
+        return;
+    }
+    WriteTasFile(u8"record.txt");
+    if (overwrite_file) {
+        WriteTasFile(u8"script0-1.txt");
+    }
+    needs_reset = true;
+    record_commands.clear();
+}
+
+InputCommon::ButtonMapping Tas::GetButtonMappingForDevice(
+    const Common::ParamPackage& params) const {
+    // This list is missing ZL/ZR since those are not considered buttons.
+    // We will add those afterwards
+    // This list also excludes any button that can't be really mapped
+    static constexpr std::array<std::pair<Settings::NativeButton::Values, TasButton>, 20>
+        switch_to_tas_button = {
+            std::pair{Settings::NativeButton::A, TasButton::BUTTON_A},
+            {Settings::NativeButton::B, TasButton::BUTTON_B},
+            {Settings::NativeButton::X, TasButton::BUTTON_X},
+            {Settings::NativeButton::Y, TasButton::BUTTON_Y},
+            {Settings::NativeButton::LStick, TasButton::STICK_L},
+            {Settings::NativeButton::RStick, TasButton::STICK_R},
+            {Settings::NativeButton::L, TasButton::TRIGGER_L},
+            {Settings::NativeButton::R, TasButton::TRIGGER_R},
+            {Settings::NativeButton::Plus, TasButton::BUTTON_PLUS},
+            {Settings::NativeButton::Minus, TasButton::BUTTON_MINUS},
+            {Settings::NativeButton::DLeft, TasButton::BUTTON_LEFT},
+            {Settings::NativeButton::DUp, TasButton::BUTTON_UP},
+            {Settings::NativeButton::DRight, TasButton::BUTTON_RIGHT},
+            {Settings::NativeButton::DDown, TasButton::BUTTON_DOWN},
+            {Settings::NativeButton::SL, TasButton::BUTTON_SL},
+            {Settings::NativeButton::SR, TasButton::BUTTON_SR},
+            {Settings::NativeButton::Screenshot, TasButton::BUTTON_CAPTURE},
+            {Settings::NativeButton::Home, TasButton::BUTTON_HOME},
+            {Settings::NativeButton::ZL, TasButton::TRIGGER_ZL},
+            {Settings::NativeButton::ZR, TasButton::TRIGGER_ZR},
+        };
+
+    InputCommon::ButtonMapping mapping{};
+    for (const auto& [switch_button, tas_button] : switch_to_tas_button) {
+        Common::ParamPackage button_params({{"engine", "tas"}});
+        button_params.Set("pad", params.Get("pad", 0));
+        button_params.Set("button", static_cast<int>(tas_button));
+        mapping.insert_or_assign(switch_button, std::move(button_params));
+    }
+
+    return mapping;
+}
+
+InputCommon::AnalogMapping Tas::GetAnalogMappingForDevice(
+    const Common::ParamPackage& params) const {
+
+    InputCommon::AnalogMapping mapping = {};
+    Common::ParamPackage left_analog_params;
+    left_analog_params.Set("engine", "tas");
+    left_analog_params.Set("pad", params.Get("pad", 0));
+    left_analog_params.Set("axis_x", static_cast<int>(TasAxes::StickX));
+    left_analog_params.Set("axis_y", static_cast<int>(TasAxes::StickY));
+    mapping.insert_or_assign(Settings::NativeAnalog::LStick, std::move(left_analog_params));
+    Common::ParamPackage right_analog_params;
+    right_analog_params.Set("engine", "tas");
+    right_analog_params.Set("pad", params.Get("pad", 0));
+    right_analog_params.Set("axis_x", static_cast<int>(TasAxes::SubstickX));
+    right_analog_params.Set("axis_y", static_cast<int>(TasAxes::SubstickY));
+    mapping.insert_or_assign(Settings::NativeAnalog::RStick, std::move(right_analog_params));
+    return mapping;
+}
+
+const TasData& Tas::GetTasState(std::size_t pad) const {
+    return tas_data[pad];
+}
+} // namespace TasInput

src/input_common/tas/tas_input.h

@@ -0,0 +1,237 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+
+#include "common/common_types.h"
+#include "common/settings_input.h"
+#include "core/frontend/input.h"
+#include "input_common/main.h"
+
+/*
+To play back TAS scripts on Yuzu, select the folder with scripts in the configuration menu below
+Tools -> Configure TAS. The file itself has normal text format and has to be called script0-1.txt
+for controller 1, script0-2.txt for controller 2 and so forth (with max. 8 players).
+
+A script file has the same format as TAS-nx uses, so final files will look like this:
+
+1 KEY_B 0;0 0;0
+6 KEY_ZL 0;0 0;0
+41 KEY_ZL;KEY_Y 0;0 0;0
+43 KEY_X;KEY_A 32767;0 0;0
+44 KEY_A 32767;0 0;0
+45 KEY_A 32767;0 0;0
+46 KEY_A 32767;0 0;0
+47 KEY_A 32767;0 0;0
+
+After placing the file at the correct location, it can be read into Yuzu with the (default) hotkey
+CTRL+F6 (refresh). In the bottom left corner, it will display the amount of frames the script file
+has. Playback can be started or stopped using CTRL+F5.
+
+However, for playback to actually work, the correct input device has to be selected: In the Controls
+menu, select TAS from the device list for the controller that the script should be played on.
+
+Recording a new script file is really simple: Just make sure that the proper device (not TAS) is
+connected on P1, and press CTRL+F7 to start recording. When done, just press the same keystroke
+again (CTRL+F7). The new script will be saved at the location previously selected, as the filename
+record.txt.
+
+For debugging purposes, the common controller debugger can be used (View -> Debugging -> Controller
+P1).
+*/
+
+namespace TasInput {
+
+constexpr size_t PLAYER_NUMBER = 8;
+
+using TasAnalog = std::pair<float, float>;
+
+enum class TasState {
+    Running,
+    Recording,
+    Stopped,
+};
+
+enum class TasButton : u32 {
+    BUTTON_A = 1U << 0,
+    BUTTON_B = 1U << 1,
+    BUTTON_X = 1U << 2,
+    BUTTON_Y = 1U << 3,
+    STICK_L = 1U << 4,
+    STICK_R = 1U << 5,
+    TRIGGER_L = 1U << 6,
+    TRIGGER_R = 1U << 7,
+    TRIGGER_ZL = 1U << 8,
+    TRIGGER_ZR = 1U << 9,
+    BUTTON_PLUS = 1U << 10,
+    BUTTON_MINUS = 1U << 11,
+    BUTTON_LEFT = 1U << 12,
+    BUTTON_UP = 1U << 13,
+    BUTTON_RIGHT = 1U << 14,
+    BUTTON_DOWN = 1U << 15,
+    BUTTON_SL = 1U << 16,
+    BUTTON_SR = 1U << 17,
+    BUTTON_HOME = 1U << 18,
+    BUTTON_CAPTURE = 1U << 19,
+};
+
+enum class TasAxes : u8 {
+    StickX,
+    StickY,
+    SubstickX,
+    SubstickY,
+    Undefined,
+};
+
+struct TasData {
+    u32 buttons{};
+    std::array<float, 4> axis{};
+};
+
+class Tas {
+public:
+    Tas();
+    ~Tas();
+
+    // Changes the input status that will be stored in each frame
+    void RecordInput(u32 buttons, const std::array<std::pair<float, float>, 2>& axes);
+
+    // Main loop that records or executes input
+    void UpdateThread();
+
+    //  Sets the flag to start or stop the TAS command excecution and swaps controllers profiles
+    void StartStop();
+
+    //  Stop the TAS and reverts any controller profile
+    void Stop();
+
+    // Sets the flag to reload the file and start from the begining in the next update
+    void Reset();
+
+    /**
+     * Sets the flag to enable or disable recording of inputs
+     * @return Returns true if the current recording status is enabled
+     */
+    bool Record();
+
+    // Saves contents of record_commands on a file if overwrite is enabled player 1 will be
+    // overwritten with the recorded commands
+    void SaveRecording(bool overwrite_file);
+
+    /**
+     * Returns the current status values of TAS playback/recording
+     * @return Tuple of
+     * TasState indicating the current state out of Running, Recording or Stopped ;
+     * Current playback progress or amount of frames (so far) for Recording ;
+     * Total length of script file currently loaded or amount of frames (so far) for Recording
+     */
+    std::tuple<TasState, size_t, size_t> GetStatus() const;
+
+    // Retuns an array of the default button mappings
+    InputCommon::ButtonMapping GetButtonMappingForDevice(const Common::ParamPackage& params) const;
+
+    // Retuns an array of the default analog mappings
+    InputCommon::AnalogMapping GetAnalogMappingForDevice(const Common::ParamPackage& params) const;
+    [[nodiscard]] const TasData& GetTasState(std::size_t pad) const;
+
+private:
+    struct TASCommand {
+        u32 buttons{};
+        TasAnalog l_axis{};
+        TasAnalog r_axis{};
+    };
+
+    // Loads TAS files from all players
+    void LoadTasFiles();
+
+    // Loads TAS file from the specified player
+    void LoadTasFile(size_t player_index);
+
+    // Writes a TAS file from the recorded commands
+    void WriteTasFile(std::u8string file_name);
+
+    /**
+     * Parses a string containing the axis values with the following format "x;y"
+     * X and Y have a range from -32767 to 32767
+     * @return Returns a TAS analog object with axis values with range from -1.0 to 1.0
+     */
+    TasAnalog ReadCommandAxis(const std::string& line) const;
+
+    /**
+     * Parses a string containing the button values with the following format "a;b;c;d..."
+     * Each button is represented by it's text format specified in text_to_tas_button array
+     * @return Returns a u32 with each bit representing the status of a button
+     */
+    u32 ReadCommandButtons(const std::string& line) const;
+
+    /**
+     * Converts an u32 containing the button status into the text equivalent
+     * @return Returns a string with the name of the buttons to be written to the file
+     */
+    std::string WriteCommandButtons(u32 data) const;
+
+    /**
+     * Converts an TAS analog object containing the axis status into the text equivalent
+     * @return Returns a string with the value of the axis to be written to the file
+     */
+    std::string WriteCommandAxis(TasAnalog data) const;
+
+    // Inverts the Y axis polarity
+    std::pair<float, float> FlipAxisY(std::pair<float, float> old);
+
+    /**
+     * Converts an u32 containing the button status into the text equivalent
+     * @return Returns a string with the name of the buttons to be printed on console
+     */
+    std::string DebugButtons(u32 buttons) const;
+
+    /**
+     * Converts an TAS analog object containing the axis status into the text equivalent
+     * @return Returns a string with the value of the axis to be printed on console
+     */
+    std::string DebugJoystick(float x, float y) const;
+
+    /**
+     * Converts the given TAS status into the text equivalent
+     * @return Returns a string with the value of the TAS status to be printed on console
+     */
+    std::string DebugInput(const TasData& data) const;
+
+    /**
+     * Converts the given TAS status of multiple players into the text equivalent
+     * @return Returns a string with the value of the status of all TAS players to be printed on
+     * console
+     */
+    std::string DebugInputs(const std::array<TasData, PLAYER_NUMBER>& arr) const;
+
+    /**
+     * Converts an u32 containing the button status into the text equivalent
+     * @return Returns a string with the name of the buttons
+     */
+    std::string ButtonsToString(u32 button) const;
+
+    // Stores current controller configuration and sets a TAS controller for every active controller
+    // to the current config
+    void SwapToTasController();
+
+    // Sets the stored controller configuration to the current config
+    void SwapToStoredController();
+
+    size_t script_length{0};
+    std::array<TasData, PLAYER_NUMBER> tas_data;
+    bool is_old_input_saved{false};
+    bool is_recording{false};
+    bool is_running{false};
+    bool needs_reset{false};
+    std::array<std::vector<TASCommand>, PLAYER_NUMBER> commands{};
+    std::vector<TASCommand> record_commands{};
+    size_t current_command{0};
+    TASCommand last_input{}; // only used for recording
+
+    // Old settings for swapping controllers
+    std::array<Settings::PlayerInput, 10> player_mappings;
+};
+} // namespace TasInput

src/input_common/tas/tas_poller.cpp

@@ -0,0 +1,101 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <mutex>
+#include <utility>
+
+#include "common/settings.h"
+#include "common/threadsafe_queue.h"
+#include "input_common/tas/tas_input.h"
+#include "input_common/tas/tas_poller.h"
+
+namespace InputCommon {
+
+class TasButton final : public Input::ButtonDevice {
+public:
+    explicit TasButton(u32 button_, u32 pad_, const TasInput::Tas* tas_input_)
+        : button(button_), pad(pad_), tas_input(tas_input_) {}
+
+    bool GetStatus() const override {
+        return (tas_input->GetTasState(pad).buttons & button) != 0;
+    }
+
+private:
+    const u32 button;
+    const u32 pad;
+    const TasInput::Tas* tas_input;
+};
+
+TasButtonFactory::TasButtonFactory(std::shared_ptr<TasInput::Tas> tas_input_)
+    : tas_input(std::move(tas_input_)) {}
+
+std::unique_ptr<Input::ButtonDevice> TasButtonFactory::Create(const Common::ParamPackage& params) {
+    const auto button_id = params.Get("button", 0);
+    const auto pad = params.Get("pad", 0);
+
+    return std::make_unique<TasButton>(button_id, pad, tas_input.get());
+}
+
+class TasAnalog final : public Input::AnalogDevice {
+public:
+    explicit TasAnalog(u32 pad_, u32 axis_x_, u32 axis_y_, const TasInput::Tas* tas_input_)
+        : pad(pad_), axis_x(axis_x_), axis_y(axis_y_), tas_input(tas_input_) {}
+
+    float GetAxis(u32 axis) const {
+        std::lock_guard lock{mutex};
+        return tas_input->GetTasState(pad).axis.at(axis);
+    }
+
+    std::pair<float, float> GetAnalog(u32 analog_axis_x, u32 analog_axis_y) const {
+        float x = GetAxis(analog_axis_x);
+        float y = GetAxis(analog_axis_y);
+
+        // Make sure the coordinates are in the unit circle,
+        // otherwise normalize it.
+        float r = x * x + y * y;
+        if (r > 1.0f) {
+            r = std::sqrt(r);
+            x /= r;
+            y /= r;
+        }
+
+        return {x, y};
+    }
+
+    std::tuple<float, float> GetStatus() const override {
+        return GetAnalog(axis_x, axis_y);
+    }
+
+    Input::AnalogProperties GetAnalogProperties() const override {
+        return {0.0f, 1.0f, 0.5f};
+    }
+
+private:
+    const u32 pad;
+    const u32 axis_x;
+    const u32 axis_y;
+    const TasInput::Tas* tas_input;
+    mutable std::mutex mutex;
+};
+
+/// An analog device factory that creates analog devices from GC Adapter
+TasAnalogFactory::TasAnalogFactory(std::shared_ptr<TasInput::Tas> tas_input_)
+    : tas_input(std::move(tas_input_)) {}
+
+/**
+ * Creates analog device from joystick axes
+ * @param params contains parameters for creating the device:
+ *     - "port": the nth gcpad on the adapter
+ *     - "axis_x": the index of the axis to be bind as x-axis
+ *     - "axis_y": the index of the axis to be bind as y-axis
+ */
+std::unique_ptr<Input::AnalogDevice> TasAnalogFactory::Create(const Common::ParamPackage& params) {
+    const auto pad = static_cast<u32>(params.Get("pad", 0));
+    const auto axis_x = static_cast<u32>(params.Get("axis_x", 0));
+    const auto axis_y = static_cast<u32>(params.Get("axis_y", 1));
+
+    return std::make_unique<TasAnalog>(pad, axis_x, axis_y, tas_input.get());
+}
+
+} // namespace InputCommon

src/input_common/tas/tas_poller.h

@@ -0,0 +1,43 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include "core/frontend/input.h"
+#include "input_common/tas/tas_input.h"
+
+namespace InputCommon {
+
+/**
+ * A button device factory representing a tas bot. It receives tas events and forward them
+ * to all button devices it created.
+ */
+class TasButtonFactory final : public Input::Factory<Input::ButtonDevice> {
+public:
+    explicit TasButtonFactory(std::shared_ptr<TasInput::Tas> tas_input_);
+
+    /**
+     * Creates a button device from a button press
+     * @param params contains parameters for creating the device:
+     *     - "code": the code of the key to bind with the button
+     */
+    std::unique_ptr<Input::ButtonDevice> Create(const Common::ParamPackage& params) override;
+
+private:
+    std::shared_ptr<TasInput::Tas> tas_input;
+};
+
+/// An analog device factory that creates analog devices from tas
+class TasAnalogFactory final : public Input::Factory<Input::AnalogDevice> {
+public:
+    explicit TasAnalogFactory(std::shared_ptr<TasInput::Tas> tas_input_);
+
+    std::unique_ptr<Input::AnalogDevice> Create(const Common::ParamPackage& params) override;
+
+private:
+    std::shared_ptr<TasInput::Tas> tas_input;
+};
+
+} // namespace InputCommon

src/input_common/udp/client.h

@@ -21,8 +21,6 @@
 
 namespace InputCommon::CemuhookUDP {
 
-constexpr char DEFAULT_SRV[] = "127.0.0.1:26760";
-
 class Socket;
 
 namespace Response {

src/shader_recompiler/backend/glsl/emit_glsl_warp.cpp

@@ -11,6 +11,8 @@
 
 namespace Shader::Backend::GLSL {
 namespace {
+constexpr char THREAD_ID[]{"gl_SubGroupInvocationARB"};
+
 void SetInBoundsFlag(EmitContext& ctx, IR::Inst& inst) {
     IR::Inst* const in_bounds{inst.GetAssociatedPseudoOperation(IR::Opcode::GetInBoundsFromOp)};
     if (!in_bounds) {
@@ -43,84 +45,100 @@ void UseShuffleNv(EmitContext& ctx, IR::Inst& inst, std::string_view shfl_op,
     ctx.AddU32("{}={}({},{},{},shfl_in_bounds);", inst, shfl_op, value, index, width);
     SetInBoundsFlag(ctx, inst);
 }
+
+std::string_view BallotIndex(EmitContext& ctx) {
+    if (!ctx.profile.warp_size_potentially_larger_than_guest) {
+        return ".x";
+    }
+    return "[gl_SubGroupInvocationARB>>5]";
+}
+
+std::string GetMask(EmitContext& ctx, std::string_view mask) {
+    const auto ballot_index{BallotIndex(ctx)};
+    return fmt::format("uint(uvec2({}){})", mask, ballot_index);
+}
 } // Anonymous namespace
 
 void EmitLaneId(EmitContext& ctx, IR::Inst& inst) {
-    ctx.AddU32("{}=gl_SubGroupInvocationARB&31u;", inst);
+    ctx.AddU32("{}={}&31u;", inst, THREAD_ID);
 }
 
 void EmitVoteAll(EmitContext& ctx, IR::Inst& inst, std::string_view pred) {
     if (!ctx.profile.warp_size_potentially_larger_than_guest) {
         ctx.AddU1("{}=allInvocationsEqualARB({});", inst, pred);
-    } else {
-        const auto active_mask{fmt::format("uvec2(ballotARB(true))[gl_SubGroupInvocationARB]")};
-        const auto ballot{fmt::format("uvec2(ballotARB({}))[gl_SubGroupInvocationARB]", pred)};
-        ctx.AddU1("{}=({}&{})=={};", inst, ballot, active_mask, active_mask);
+        return;
     }
+    const auto ballot_index{BallotIndex(ctx)};
+    const auto active_mask{fmt::format("uvec2(ballotARB(true)){}", ballot_index)};
+    const auto ballot{fmt::format("uvec2(ballotARB({})){}", pred, ballot_index)};
+    ctx.AddU1("{}=({}&{})=={};", inst, ballot, active_mask, active_mask);
 }
 
 void EmitVoteAny(EmitContext& ctx, IR::Inst& inst, std::string_view pred) {
     if (!ctx.profile.warp_size_potentially_larger_than_guest) {
         ctx.AddU1("{}=anyInvocationARB({});", inst, pred);
-    } else {
-        const auto active_mask{fmt::format("uvec2(ballotARB(true))[gl_SubGroupInvocationARB]")};
-        const auto ballot{fmt::format("uvec2(ballotARB({}))[gl_SubGroupInvocationARB]", pred)};
-        ctx.AddU1("{}=({}&{})!=0u;", inst, ballot, active_mask, active_mask);
+        return;
     }
+    const auto ballot_index{BallotIndex(ctx)};
+    const auto active_mask{fmt::format("uvec2(ballotARB(true)){}", ballot_index)};
+    const auto ballot{fmt::format("uvec2(ballotARB({})){}", pred, ballot_index)};
+    ctx.AddU1("{}=({}&{})!=0u;", inst, ballot, active_mask, active_mask);
 }
 
 void EmitVoteEqual(EmitContext& ctx, IR::Inst& inst, std::string_view pred) {
     if (!ctx.profile.warp_size_potentially_larger_than_guest) {
         ctx.AddU1("{}=allInvocationsEqualARB({});", inst, pred);
-    } else {
-        const auto active_mask{fmt::format("uvec2(ballotARB(true))[gl_SubGroupInvocationARB]")};
-        const auto ballot{fmt::format("uvec2(ballotARB({}))[gl_SubGroupInvocationARB]", pred)};
-        const auto value{fmt::format("({}^{})", ballot, active_mask)};
-        ctx.AddU1("{}=({}==0)||({}=={});", inst, value, value, active_mask);
+        return;
     }
+    const auto ballot_index{BallotIndex(ctx)};
+    const auto active_mask{fmt::format("uvec2(ballotARB(true)){}", ballot_index)};
+    const auto ballot{fmt::format("uvec2(ballotARB({})){}", pred, ballot_index)};
+    const auto value{fmt::format("({}^{})", ballot, active_mask)};
+    ctx.AddU1("{}=({}==0)||({}=={});", inst, value, value, active_mask);
 }
 
 void EmitSubgroupBallot(EmitContext& ctx, IR::Inst& inst, std::string_view pred) {
-    if (!ctx.profile.warp_size_potentially_larger_than_guest) {
-        ctx.AddU32("{}=uvec2(ballotARB({})).x;", inst, pred);
-    } else {
-        ctx.AddU32("{}=uvec2(ballotARB({}))[gl_SubGroupInvocationARB];", inst, pred);
-    }
+    const auto ballot_index{BallotIndex(ctx)};
+    ctx.AddU32("{}=uvec2(ballotARB({})){};", inst, pred, ballot_index);
 }
 
 void EmitSubgroupEqMask(EmitContext& ctx, IR::Inst& inst) {
-    ctx.AddU32("{}=uint(gl_SubGroupEqMaskARB.x);", inst);
+    ctx.AddU32("{}={};", inst, GetMask(ctx, "gl_SubGroupEqMaskARB"));
 }
 
 void EmitSubgroupLtMask(EmitContext& ctx, IR::Inst& inst) {
-    ctx.AddU32("{}=uint(gl_SubGroupLtMaskARB.x);", inst);
+    ctx.AddU32("{}={};", inst, GetMask(ctx, "gl_SubGroupLtMaskARB"));
 }
 
 void EmitSubgroupLeMask(EmitContext& ctx, IR::Inst& inst) {
-    ctx.AddU32("{}=uint(gl_SubGroupLeMaskARB.x);", inst);
+    ctx.AddU32("{}={};", inst, GetMask(ctx, "gl_SubGroupLeMaskARB"));
 }
 
 void EmitSubgroupGtMask(EmitContext& ctx, IR::Inst& inst) {
-    ctx.AddU32("{}=uint(gl_SubGroupGtMaskARB.x);", inst);
+    ctx.AddU32("{}={};", inst, GetMask(ctx, "gl_SubGroupGtMaskARB"));
 }
 
 void EmitSubgroupGeMask(EmitContext& ctx, IR::Inst& inst) {
-    ctx.AddU32("{}=uint(gl_SubGroupGeMaskARB.x);", inst);
+    ctx.AddU32("{}={};", inst, GetMask(ctx, "gl_SubGroupGeMaskARB"));
 }
 
 void EmitShuffleIndex(EmitContext& ctx, IR::Inst& inst, std::string_view value,
-                      std::string_view index, std::string_view clamp,
-                      std::string_view segmentation_mask) {
+                      std::string_view index, std::string_view clamp, std::string_view seg_mask) {
     if (ctx.profile.support_gl_warp_intrinsics) {
-        UseShuffleNv(ctx, inst, "shuffleNV", value, index, clamp, segmentation_mask);
+        UseShuffleNv(ctx, inst, "shuffleNV", value, index, clamp, seg_mask);
         return;
     }
-    const auto not_seg_mask{fmt::format("(~{})", segmentation_mask)};
-    const auto thread_id{"gl_SubGroupInvocationARB"};
-    const auto min_thread_id{ComputeMinThreadId(thread_id, segmentation_mask)};
-    const auto max_thread_id{ComputeMaxThreadId(min_thread_id, clamp, not_seg_mask)};
+    const bool big_warp{ctx.profile.warp_size_potentially_larger_than_guest};
+    const auto is_upper_partition{"int(gl_SubGroupInvocationARB)>=32"};
+    const auto upper_index{fmt::format("{}?{}+32:{}", is_upper_partition, index, index)};
+    const auto upper_clamp{fmt::format("{}?{}+32:{}", is_upper_partition, clamp, clamp)};
 
-    const auto lhs{fmt::format("({}&{})", index, not_seg_mask)};
+    const auto not_seg_mask{fmt::format("(~{})", seg_mask)};
+    const auto min_thread_id{ComputeMinThreadId(THREAD_ID, seg_mask)};
+    const auto max_thread_id{
+        ComputeMaxThreadId(min_thread_id, big_warp ? upper_clamp : clamp, not_seg_mask)};
+
+    const auto lhs{fmt::format("({}&{})", big_warp ? upper_index : index, not_seg_mask)};
     const auto src_thread_id{fmt::format("({})|({})", lhs, min_thread_id)};
     ctx.Add("shfl_in_bounds=int({})<=int({});", src_thread_id, max_thread_id);
     SetInBoundsFlag(ctx, inst);
@@ -128,29 +146,34 @@ void EmitShuffleIndex(EmitContext& ctx, IR::Inst& inst, std::string_view value,
 }
 
 void EmitShuffleUp(EmitContext& ctx, IR::Inst& inst, std::string_view value, std::string_view index,
-                   std::string_view clamp, std::string_view segmentation_mask) {
+                   std::string_view clamp, std::string_view seg_mask) {
     if (ctx.profile.support_gl_warp_intrinsics) {
-        UseShuffleNv(ctx, inst, "shuffleUpNV", value, index, clamp, segmentation_mask);
+        UseShuffleNv(ctx, inst, "shuffleUpNV", value, index, clamp, seg_mask);
         return;
     }
-    const auto thread_id{"gl_SubGroupInvocationARB"};
-    const auto max_thread_id{GetMaxThreadId(thread_id, clamp, segmentation_mask)};
-    const auto src_thread_id{fmt::format("({}-{})", thread_id, index)};
+    const bool big_warp{ctx.profile.warp_size_potentially_larger_than_guest};
+    const auto is_upper_partition{"int(gl_SubGroupInvocationARB)>=32"};
+    const auto upper_clamp{fmt::format("{}?{}+32:{}", is_upper_partition, clamp, clamp)};
+
+    const auto max_thread_id{GetMaxThreadId(THREAD_ID, big_warp ? upper_clamp : clamp, seg_mask)};
+    const auto src_thread_id{fmt::format("({}-{})", THREAD_ID, index)};
     ctx.Add("shfl_in_bounds=int({})>=int({});", src_thread_id, max_thread_id);
     SetInBoundsFlag(ctx, inst);
     ctx.AddU32("{}=shfl_in_bounds?readInvocationARB({},{}):{};", inst, value, src_thread_id, value);
 }
 
 void EmitShuffleDown(EmitContext& ctx, IR::Inst& inst, std::string_view value,
-                     std::string_view index, std::string_view clamp,
-                     std::string_view segmentation_mask) {
+                     std::string_view index, std::string_view clamp, std::string_view seg_mask) {
     if (ctx.profile.support_gl_warp_intrinsics) {
-        UseShuffleNv(ctx, inst, "shuffleDownNV", value, index, clamp, segmentation_mask);
+        UseShuffleNv(ctx, inst, "shuffleDownNV", value, index, clamp, seg_mask);
         return;
     }
-    const auto thread_id{"gl_SubGroupInvocationARB"};
-    const auto max_thread_id{GetMaxThreadId(thread_id, clamp, segmentation_mask)};
-    const auto src_thread_id{fmt::format("({}+{})", thread_id, index)};
+    const bool big_warp{ctx.profile.warp_size_potentially_larger_than_guest};
+    const auto is_upper_partition{"int(gl_SubGroupInvocationARB)>=32"};
+    const auto upper_clamp{fmt::format("{}?{}+32:{}", is_upper_partition, clamp, clamp)};
+
+    const auto max_thread_id{GetMaxThreadId(THREAD_ID, big_warp ? upper_clamp : clamp, seg_mask)};
+    const auto src_thread_id{fmt::format("({}+{})", THREAD_ID, index)};
     ctx.Add("shfl_in_bounds=int({})<=int({});", src_thread_id, max_thread_id);
     SetInBoundsFlag(ctx, inst);
     ctx.AddU32("{}=shfl_in_bounds?readInvocationARB({},{}):{};", inst, value, src_thread_id, value);
@@ -158,14 +181,17 @@ void EmitShuffleDown(EmitContext& ctx, IR::Inst& inst, std::string_view value,
 
 void EmitShuffleButterfly(EmitContext& ctx, IR::Inst& inst, std::string_view value,
                           std::string_view index, std::string_view clamp,
-                          std::string_view segmentation_mask) {
+                          std::string_view seg_mask) {
     if (ctx.profile.support_gl_warp_intrinsics) {
-        UseShuffleNv(ctx, inst, "shuffleXorNV", value, index, clamp, segmentation_mask);
+        UseShuffleNv(ctx, inst, "shuffleXorNV", value, index, clamp, seg_mask);
         return;
     }
-    const auto thread_id{"gl_SubGroupInvocationARB"};
-    const auto max_thread_id{GetMaxThreadId(thread_id, clamp, segmentation_mask)};
-    const auto src_thread_id{fmt::format("({}^{})", thread_id, index)};
+    const bool big_warp{ctx.profile.warp_size_potentially_larger_than_guest};
+    const auto is_upper_partition{"int(gl_SubGroupInvocationARB)>=32"};
+    const auto upper_clamp{fmt::format("{}?{}+32:{}", is_upper_partition, clamp, clamp)};
+
+    const auto max_thread_id{GetMaxThreadId(THREAD_ID, big_warp ? upper_clamp : clamp, seg_mask)};
+    const auto src_thread_id{fmt::format("({}^{})", THREAD_ID, index)};
     ctx.Add("shfl_in_bounds=int({})<=int({});", src_thread_id, max_thread_id);
     SetInBoundsFlag(ctx, inst);
     ctx.AddU32("{}=shfl_in_bounds?readInvocationARB({},{}):{};", inst, value, src_thread_id, value);

src/shader_recompiler/backend/spirv/emit_context.cpp

@@ -15,6 +15,8 @@
 
 namespace Shader::Backend::SPIRV {
 namespace {
+constexpr size_t NUM_FIXEDFNCTEXTURE = 10;
+
 enum class Operation {
     Increment,
     Decrement,
@@ -427,6 +429,16 @@ Id DescType(EmitContext& ctx, Id sampled_type, Id pointer_type, u32 count) {
         return pointer_type;
     }
 }
+
+size_t FindNextUnusedLocation(const std::bitset<IR::NUM_GENERICS>& used_locations,
+                              size_t start_offset) {
+    for (size_t location = start_offset; location < used_locations.size(); ++location) {
+        if (!used_locations.test(location)) {
+            return location;
+        }
+    }
+    throw RuntimeError("Unable to get an unused location for legacy attribute");
+}
 } // Anonymous namespace
 
 void VectorTypes::Define(Sirit::Module& sirit_ctx, Id base_type, std::string_view name) {
@@ -1227,6 +1239,7 @@ void EmitContext::DefineInputs(const IR::Program& program) {
         loads[IR::Attribute::TessellationEvaluationPointV]) {
         tess_coord = DefineInput(*this, F32[3], false, spv::BuiltIn::TessCoord);
     }
+    std::bitset<IR::NUM_GENERICS> used_locations{};
     for (size_t index = 0; index < IR::NUM_GENERICS; ++index) {
         const AttributeType input_type{runtime_info.generic_input_types[index]};
         if (!runtime_info.previous_stage_stores.Generic(index)) {
@@ -1238,6 +1251,7 @@ void EmitContext::DefineInputs(const IR::Program& program) {
         if (input_type == AttributeType::Disabled) {
             continue;
         }
+        used_locations.set(index);
         const Id type{GetAttributeType(*this, input_type)};
         const Id id{DefineInput(*this, type, true)};
         Decorate(id, spv::Decoration::Location, static_cast<u32>(index));
@@ -1263,6 +1277,26 @@ void EmitContext::DefineInputs(const IR::Program& program) {
             break;
         }
     }
+    size_t previous_unused_location = 0;
+    if (loads.AnyComponent(IR::Attribute::ColorFrontDiffuseR)) {
+        const size_t location = FindNextUnusedLocation(used_locations, previous_unused_location);
+        previous_unused_location = location;
+        used_locations.set(location);
+        const Id id{DefineInput(*this, F32[4], true)};
+        Decorate(id, spv::Decoration::Location, location);
+        input_front_color = id;
+    }
+    for (size_t index = 0; index < NUM_FIXEDFNCTEXTURE; ++index) {
+        if (loads.AnyComponent(IR::Attribute::FixedFncTexture0S + index * 4)) {
+            const size_t location =
+                FindNextUnusedLocation(used_locations, previous_unused_location);
+            previous_unused_location = location;
+            used_locations.set(location);
+            const Id id{DefineInput(*this, F32[4], true)};
+            Decorate(id, spv::Decoration::Location, location);
+            input_fixed_fnc_textures[index] = id;
+        }
+    }
     if (stage == Stage::TessellationEval) {
         for (size_t index = 0; index < info.uses_patches.size(); ++index) {
             if (!info.uses_patches[index]) {
@@ -1313,9 +1347,31 @@ void EmitContext::DefineOutputs(const IR::Program& program) {
         viewport_mask = DefineOutput(*this, TypeArray(U32[1], Const(1u)), std::nullopt,
                                      spv::BuiltIn::ViewportMaskNV);
     }
+    std::bitset<IR::NUM_GENERICS> used_locations{};
     for (size_t index = 0; index < IR::NUM_GENERICS; ++index) {
         if (info.stores.Generic(index)) {
             DefineGenericOutput(*this, index, invocations);
+            used_locations.set(index);
+        }
+    }
+    size_t previous_unused_location = 0;
+    if (info.stores.AnyComponent(IR::Attribute::ColorFrontDiffuseR)) {
+        const size_t location = FindNextUnusedLocation(used_locations, previous_unused_location);
+        previous_unused_location = location;
+        used_locations.set(location);
+        const Id id{DefineOutput(*this, F32[4], invocations)};
+        Decorate(id, spv::Decoration::Location, static_cast<u32>(location));
+        output_front_color = id;
+    }
+    for (size_t index = 0; index < NUM_FIXEDFNCTEXTURE; ++index) {
+        if (info.stores.AnyComponent(IR::Attribute::FixedFncTexture0S + index * 4)) {
+            const size_t location =
+                FindNextUnusedLocation(used_locations, previous_unused_location);
+            previous_unused_location = location;
+            used_locations.set(location);
+            const Id id{DefineOutput(*this, F32[4], invocations)};
+            Decorate(id, spv::Decoration::Location, location);
+            output_fixed_fnc_textures[index] = id;
         }
     }
     switch (stage) {

src/shader_recompiler/backend/spirv/emit_context.h

@@ -268,10 +268,14 @@ public:
     Id write_global_func_u32x4{};
 
     Id input_position{};
+    Id input_front_color{};
+    std::array<Id, 10> input_fixed_fnc_textures{};
     std::array<Id, 32> input_generics{};
 
     Id output_point_size{};
     Id output_position{};
+    Id output_front_color{};
+    std::array<Id, 10> output_fixed_fnc_textures{};
     std::array<std::array<GenericElementInfo, 4>, 32> output_generics{};
 
     Id output_tess_level_outer{};

src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp

@@ -43,6 +43,25 @@ Id AttrPointer(EmitContext& ctx, Id pointer_type, Id vertex, Id base, Args&&...
     }
 }
 
+bool IsFixedFncTexture(IR::Attribute attribute) {
+    return attribute >= IR::Attribute::FixedFncTexture0S &&
+           attribute <= IR::Attribute::FixedFncTexture9Q;
+}
+
+u32 FixedFncTextureAttributeIndex(IR::Attribute attribute) {
+    if (!IsFixedFncTexture(attribute)) {
+        throw InvalidArgument("Attribute {} is not a FixedFncTexture", attribute);
+    }
+    return (static_cast<u32>(attribute) - static_cast<u32>(IR::Attribute::FixedFncTexture0S)) / 4u;
+}
+
+u32 FixedFncTextureAttributeElement(IR::Attribute attribute) {
+    if (!IsFixedFncTexture(attribute)) {
+        throw InvalidArgument("Attribute {} is not a FixedFncTexture", attribute);
+    }
+    return static_cast<u32>(attribute) % 4u;
+}
+
 template <typename... Args>
 Id OutputAccessChain(EmitContext& ctx, Id result_type, Id base, Args&&... args) {
     if (ctx.stage == Stage::TessellationControl) {
@@ -74,6 +93,13 @@ std::optional<OutAttr> OutputAttrPointer(EmitContext& ctx, IR::Attribute attr) {
             return OutputAccessChain(ctx, ctx.output_f32, info.id, index_id);
         }
     }
+    if (IsFixedFncTexture(attr)) {
+        const u32 index{FixedFncTextureAttributeIndex(attr)};
+        const u32 element{FixedFncTextureAttributeElement(attr)};
+        const Id element_id{ctx.Const(element)};
+        return OutputAccessChain(ctx, ctx.output_f32, ctx.output_fixed_fnc_textures[index],
+                                 element_id);
+    }
     switch (attr) {
     case IR::Attribute::PointSize:
         return ctx.output_point_size;
@@ -85,6 +111,14 @@ std::optional<OutAttr> OutputAttrPointer(EmitContext& ctx, IR::Attribute attr) {
         const Id element_id{ctx.Const(element)};
         return OutputAccessChain(ctx, ctx.output_f32, ctx.output_position, element_id);
     }
+    case IR::Attribute::ColorFrontDiffuseR:
+    case IR::Attribute::ColorFrontDiffuseG:
+    case IR::Attribute::ColorFrontDiffuseB:
+    case IR::Attribute::ColorFrontDiffuseA: {
+        const u32 element{static_cast<u32>(attr) % 4};
+        const Id element_id{ctx.Const(element)};
+        return OutputAccessChain(ctx, ctx.output_f32, ctx.output_front_color, element_id);
+    }
     case IR::Attribute::ClipDistance0:
     case IR::Attribute::ClipDistance1:
     case IR::Attribute::ClipDistance2:
@@ -307,6 +341,12 @@ Id EmitGetAttribute(EmitContext& ctx, IR::Attribute attr, Id vertex) {
         const Id value{ctx.OpLoad(type->id, pointer)};
         return type->needs_cast ? ctx.OpBitcast(ctx.F32[1], value) : value;
     }
+    if (IsFixedFncTexture(attr)) {
+        const u32 index{FixedFncTextureAttributeIndex(attr)};
+        const Id attr_id{ctx.input_fixed_fnc_textures[index]};
+        const Id attr_ptr{AttrPointer(ctx, ctx.input_f32, vertex, attr_id, ctx.Const(element))};
+        return ctx.OpLoad(ctx.F32[1], attr_ptr);
+    }
     switch (attr) {
     case IR::Attribute::PrimitiveId:
         return ctx.OpBitcast(ctx.F32[1], ctx.OpLoad(ctx.U32[1], ctx.primitive_id));
@@ -316,6 +356,13 @@ Id EmitGetAttribute(EmitContext& ctx, IR::Attribute attr, Id vertex) {
     case IR::Attribute::PositionW:
         return ctx.OpLoad(ctx.F32[1], AttrPointer(ctx, ctx.input_f32, vertex, ctx.input_position,
                                                   ctx.Const(element)));
+    case IR::Attribute::ColorFrontDiffuseR:
+    case IR::Attribute::ColorFrontDiffuseG:
+    case IR::Attribute::ColorFrontDiffuseB:
+    case IR::Attribute::ColorFrontDiffuseA: {
+        return ctx.OpLoad(ctx.F32[1], AttrPointer(ctx, ctx.input_f32, vertex, ctx.input_front_color,
+                                                  ctx.Const(element)));
+    }
     case IR::Attribute::InstanceId:
         if (ctx.profile.support_vertex_instance_id) {
             return ctx.OpBitcast(ctx.F32[1], ctx.OpLoad(ctx.U32[1], ctx.instance_id));

src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp

@@ -7,8 +7,13 @@
 
 namespace Shader::Backend::SPIRV {
 namespace {
+Id GetThreadId(EmitContext& ctx) {
+    return ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id);
+}
+
 Id WarpExtract(EmitContext& ctx, Id value) {
-    const Id local_index{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+    const Id thread_id{GetThreadId(ctx)};
+    const Id local_index{ctx.OpShiftRightArithmetic(ctx.U32[1], thread_id, ctx.Const(5U))};
     return ctx.OpVectorExtractDynamic(ctx.U32[1], value, local_index);
 }
 
@@ -48,10 +53,17 @@ Id SelectValue(EmitContext& ctx, Id in_range, Id value, Id src_thread_id) {
     return ctx.OpSelect(ctx.U32[1], in_range,
                         ctx.OpSubgroupReadInvocationKHR(ctx.U32[1], value, src_thread_id), value);
 }
+
+Id GetUpperClamp(EmitContext& ctx, Id invocation_id, Id clamp) {
+    const Id thirty_two{ctx.Const(32u)};
+    const Id is_upper_partition{ctx.OpSGreaterThanEqual(ctx.U1, invocation_id, thirty_two)};
+    const Id upper_clamp{ctx.OpIAdd(ctx.U32[1], thirty_two, clamp)};
+    return ctx.OpSelect(ctx.U32[1], is_upper_partition, upper_clamp, clamp);
+}
 } // Anonymous namespace
 
 Id EmitLaneId(EmitContext& ctx) {
-    const Id id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+    const Id id{GetThreadId(ctx)};
     if (!ctx.profile.warp_size_potentially_larger_than_guest) {
         return id;
     }
@@ -123,7 +135,15 @@ Id EmitSubgroupGeMask(EmitContext& ctx) {
 Id EmitShuffleIndex(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
                     Id segmentation_mask) {
     const Id not_seg_mask{ctx.OpNot(ctx.U32[1], segmentation_mask)};
-    const Id thread_id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+    const Id thread_id{GetThreadId(ctx)};
+    if (ctx.profile.warp_size_potentially_larger_than_guest) {
+        const Id thirty_two{ctx.Const(32u)};
+        const Id is_upper_partition{ctx.OpSGreaterThanEqual(ctx.U1, thread_id, thirty_two)};
+        const Id upper_index{ctx.OpIAdd(ctx.U32[1], thirty_two, index)};
+        const Id upper_clamp{ctx.OpIAdd(ctx.U32[1], thirty_two, clamp)};
+        index = ctx.OpSelect(ctx.U32[1], is_upper_partition, upper_index, index);
+        clamp = ctx.OpSelect(ctx.U32[1], is_upper_partition, upper_clamp, clamp);
+    }
     const Id min_thread_id{ComputeMinThreadId(ctx, thread_id, segmentation_mask)};
     const Id max_thread_id{ComputeMaxThreadId(ctx, min_thread_id, clamp, not_seg_mask)};
 
@@ -137,7 +157,10 @@ Id EmitShuffleIndex(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id cla
 
 Id EmitShuffleUp(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
                  Id segmentation_mask) {
-    const Id thread_id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+    const Id thread_id{GetThreadId(ctx)};
+    if (ctx.profile.warp_size_potentially_larger_than_guest) {
+        clamp = GetUpperClamp(ctx, thread_id, clamp);
+    }
     const Id max_thread_id{GetMaxThreadId(ctx, thread_id, clamp, segmentation_mask)};
     const Id src_thread_id{ctx.OpISub(ctx.U32[1], thread_id, index)};
     const Id in_range{ctx.OpSGreaterThanEqual(ctx.U1, src_thread_id, max_thread_id)};
@@ -148,7 +171,10 @@ Id EmitShuffleUp(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
 
 Id EmitShuffleDown(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
                    Id segmentation_mask) {
-    const Id thread_id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+    const Id thread_id{GetThreadId(ctx)};
+    if (ctx.profile.warp_size_potentially_larger_than_guest) {
+        clamp = GetUpperClamp(ctx, thread_id, clamp);
+    }
     const Id max_thread_id{GetMaxThreadId(ctx, thread_id, clamp, segmentation_mask)};
     const Id src_thread_id{ctx.OpIAdd(ctx.U32[1], thread_id, index)};
     const Id in_range{ctx.OpSLessThanEqual(ctx.U1, src_thread_id, max_thread_id)};
@@ -159,7 +185,10 @@ Id EmitShuffleDown(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clam
 
 Id EmitShuffleButterfly(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
                         Id segmentation_mask) {
-    const Id thread_id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+    const Id thread_id{GetThreadId(ctx)};
+    if (ctx.profile.warp_size_potentially_larger_than_guest) {
+        clamp = GetUpperClamp(ctx, thread_id, clamp);
+    }
     const Id max_thread_id{GetMaxThreadId(ctx, thread_id, clamp, segmentation_mask)};
     const Id src_thread_id{ctx.OpBitwiseXor(ctx.U32[1], thread_id, index)};
     const Id in_range{ctx.OpSLessThanEqual(ctx.U1, src_thread_id, max_thread_id)};

src/video_core/CMakeLists.txt

@@ -231,6 +231,7 @@ endif()
 
 target_include_directories(video_core PRIVATE ${FFmpeg_INCLUDE_DIR})
 target_link_libraries(video_core PRIVATE ${FFmpeg_LIBRARIES})
+target_link_options(video_core PRIVATE ${FFmpeg_LDFLAGS})
 
 add_dependencies(video_core host_shaders)
 target_include_directories(video_core PRIVATE ${HOST_SHADERS_INCLUDE})

src/video_core/command_classes/codecs/codec.cpp

@@ -5,6 +5,7 @@
 #include <fstream>
 #include <vector>
 #include "common/assert.h"
+#include "common/settings.h"
 #include "video_core/command_classes/codecs/codec.h"
 #include "video_core/command_classes/codecs/h264.h"
 #include "video_core/command_classes/codecs/vp9.h"
@@ -16,44 +17,28 @@ extern "C" {
 }
 
 namespace Tegra {
-#if defined(LIBVA_FOUND)
-// Hardware acceleration code from FFmpeg/doc/examples/hw_decode.c originally under MIT license
 namespace {
-constexpr std::array<const char*, 2> VAAPI_DRIVERS = {
-    "i915",
-    "amdgpu",
-};
+constexpr AVPixelFormat PREFERRED_GPU_FMT = AV_PIX_FMT_NV12;
+constexpr AVPixelFormat PREFERRED_CPU_FMT = AV_PIX_FMT_YUV420P;
 
-AVPixelFormat GetHwFormat(AVCodecContext*, const AVPixelFormat* pix_fmts) {
+void AVPacketDeleter(AVPacket* ptr) {
+    av_packet_free(&ptr);
+}
+
+using AVPacketPtr = std::unique_ptr<AVPacket, decltype(&AVPacketDeleter)>;
+
+AVPixelFormat GetGpuFormat(AVCodecContext* av_codec_ctx, const AVPixelFormat* pix_fmts) {
     for (const AVPixelFormat* p = pix_fmts; *p != AV_PIX_FMT_NONE; ++p) {
-        if (*p == AV_PIX_FMT_VAAPI) {
-            return AV_PIX_FMT_VAAPI;
+        if (*p == av_codec_ctx->pix_fmt) {
+            return av_codec_ctx->pix_fmt;
         }
     }
     LOG_INFO(Service_NVDRV, "Could not find compatible GPU AV format, falling back to CPU");
-    return *pix_fmts;
-}
-
-bool CreateVaapiHwdevice(AVBufferRef** av_hw_device) {
-    AVDictionary* hwdevice_options = nullptr;
-    av_dict_set(&hwdevice_options, "connection_type", "drm", 0);
-    for (const auto& driver : VAAPI_DRIVERS) {
-        av_dict_set(&hwdevice_options, "kernel_driver", driver, 0);
-        const int hwdevice_error = av_hwdevice_ctx_create(av_hw_device, AV_HWDEVICE_TYPE_VAAPI,
-                                                          nullptr, hwdevice_options, 0);
-        if (hwdevice_error >= 0) {
-            LOG_INFO(Service_NVDRV, "Using VA-API with {}", driver);
-            av_dict_free(&hwdevice_options);
-            return true;
-        }
-        LOG_DEBUG(Service_NVDRV, "VA-API av_hwdevice_ctx_create failed {}", hwdevice_error);
-    }
-    LOG_DEBUG(Service_NVDRV, "VA-API av_hwdevice_ctx_create failed for all drivers");
-    av_dict_free(&hwdevice_options);
-    return false;
+    av_buffer_unref(&av_codec_ctx->hw_device_ctx);
+    av_codec_ctx->pix_fmt = PREFERRED_CPU_FMT;
+    return PREFERRED_CPU_FMT;
 }
 } // namespace
-#endif
 
 void AVFrameDeleter(AVFrame* ptr) {
     av_frame_free(&ptr);
@@ -68,56 +53,110 @@ Codec::~Codec() {
         return;
     }
     // Free libav memory
-    avcodec_send_packet(av_codec_ctx, nullptr);
-    AVFrame* av_frame = av_frame_alloc();
-    avcodec_receive_frame(av_codec_ctx, av_frame);
-    avcodec_flush_buffers(av_codec_ctx);
-    av_frame_free(&av_frame);
-    avcodec_close(av_codec_ctx);
-    av_buffer_unref(&av_hw_device);
+    avcodec_free_context(&av_codec_ctx);
+    av_buffer_unref(&av_gpu_decoder);
 }
 
-void Codec::InitializeHwdec() {
-    // Prioritize integrated GPU to mitigate bandwidth bottlenecks
+bool Codec::CreateGpuAvDevice() {
 #if defined(LIBVA_FOUND)
-    if (CreateVaapiHwdevice(&av_hw_device)) {
-        const auto hw_device_ctx = av_buffer_ref(av_hw_device);
-        ASSERT_MSG(hw_device_ctx, "av_buffer_ref failed");
-        av_codec_ctx->hw_device_ctx = hw_device_ctx;
-        av_codec_ctx->get_format = GetHwFormat;
+    static constexpr std::array<const char*, 3> VAAPI_DRIVERS = {
+        "i915",
+        "iHD",
+        "amdgpu",
+    };
+    AVDictionary* hwdevice_options = nullptr;
+    av_dict_set(&hwdevice_options, "connection_type", "drm", 0);
+    for (const auto& driver : VAAPI_DRIVERS) {
+        av_dict_set(&hwdevice_options, "kernel_driver", driver, 0);
+        const int hwdevice_error = av_hwdevice_ctx_create(&av_gpu_decoder, AV_HWDEVICE_TYPE_VAAPI,
+                                                          nullptr, hwdevice_options, 0);
+        if (hwdevice_error >= 0) {
+            LOG_INFO(Service_NVDRV, "Using VA-API with {}", driver);
+            av_dict_free(&hwdevice_options);
+            av_codec_ctx->pix_fmt = AV_PIX_FMT_VAAPI;
+            return true;
+        }
+        LOG_DEBUG(Service_NVDRV, "VA-API av_hwdevice_ctx_create failed {}", hwdevice_error);
+    }
+    LOG_DEBUG(Service_NVDRV, "VA-API av_hwdevice_ctx_create failed for all drivers");
+    av_dict_free(&hwdevice_options);
+#endif
+    static constexpr auto HW_CONFIG_METHOD = AV_CODEC_HW_CONFIG_METHOD_HW_DEVICE_CTX;
+    static constexpr std::array GPU_DECODER_TYPES{
+        AV_HWDEVICE_TYPE_CUDA,
+#ifdef _WIN32
+        AV_HWDEVICE_TYPE_D3D11VA,
+#else
+        AV_HWDEVICE_TYPE_VDPAU,
+#endif
+    };
+    for (const auto& type : GPU_DECODER_TYPES) {
+        const int hwdevice_res = av_hwdevice_ctx_create(&av_gpu_decoder, type, nullptr, nullptr, 0);
+        if (hwdevice_res < 0) {
+            LOG_DEBUG(Service_NVDRV, "{} av_hwdevice_ctx_create failed {}",
+                      av_hwdevice_get_type_name(type), hwdevice_res);
+            continue;
+        }
+        for (int i = 0;; i++) {
+            const AVCodecHWConfig* config = avcodec_get_hw_config(av_codec, i);
+            if (!config) {
+                LOG_DEBUG(Service_NVDRV, "{} decoder does not support device type {}.",
+                          av_codec->name, av_hwdevice_get_type_name(type));
+                break;
+            }
+            if (config->methods & HW_CONFIG_METHOD && config->device_type == type) {
+                av_codec_ctx->pix_fmt = config->pix_fmt;
+                LOG_INFO(Service_NVDRV, "Using {} GPU decoder", av_hwdevice_get_type_name(type));
+                return true;
+            }
+        }
+    }
+    return false;
+}
+
+void Codec::InitializeAvCodecContext() {
+    av_codec_ctx = avcodec_alloc_context3(av_codec);
+    av_opt_set(av_codec_ctx->priv_data, "tune", "zerolatency", 0);
+}
+
+void Codec::InitializeGpuDecoder() {
+    if (!CreateGpuAvDevice()) {
+        av_buffer_unref(&av_gpu_decoder);
         return;
     }
-#endif
-    // TODO more GPU accelerated decoders
+    auto* hw_device_ctx = av_buffer_ref(av_gpu_decoder);
+    ASSERT_MSG(hw_device_ctx, "av_buffer_ref failed");
+    av_codec_ctx->hw_device_ctx = hw_device_ctx;
+    av_codec_ctx->get_format = GetGpuFormat;
 }
 
 void Codec::Initialize() {
-    AVCodecID codec;
-    switch (current_codec) {
-    case NvdecCommon::VideoCodec::H264:
-        codec = AV_CODEC_ID_H264;
-        break;
-    case NvdecCommon::VideoCodec::Vp9:
-        codec = AV_CODEC_ID_VP9;
-        break;
-    default:
-        UNIMPLEMENTED_MSG("Unknown codec {}", current_codec);
+    const AVCodecID codec = [&] {
+        switch (current_codec) {
+        case NvdecCommon::VideoCodec::H264:
+            return AV_CODEC_ID_H264;
+        case NvdecCommon::VideoCodec::Vp9:
+            return AV_CODEC_ID_VP9;
+        default:
+            UNIMPLEMENTED_MSG("Unknown codec {}", current_codec);
+            return AV_CODEC_ID_NONE;
+        }
+    }();
+    av_codec = avcodec_find_decoder(codec);
+
+    InitializeAvCodecContext();
+    if (Settings::values.nvdec_emulation.GetValue() == Settings::NvdecEmulation::GPU) {
+        InitializeGpuDecoder();
+    }
+    if (const int res = avcodec_open2(av_codec_ctx, av_codec, nullptr); res < 0) {
+        LOG_ERROR(Service_NVDRV, "avcodec_open2() Failed with result {}", res);
+        avcodec_free_context(&av_codec_ctx);
+        av_buffer_unref(&av_gpu_decoder);
         return;
     }
-    av_codec = avcodec_find_decoder(codec);
-    av_codec_ctx = avcodec_alloc_context3(av_codec);
-    av_opt_set(av_codec_ctx->priv_data, "tune", "zerolatency", 0);
-    InitializeHwdec();
     if (!av_codec_ctx->hw_device_ctx) {
         LOG_INFO(Service_NVDRV, "Using FFmpeg software decoding");
     }
-    const auto av_error = avcodec_open2(av_codec_ctx, av_codec, nullptr);
-    if (av_error < 0) {
-        LOG_ERROR(Service_NVDRV, "avcodec_open2() Failed.");
-        avcodec_close(av_codec_ctx);
-        av_buffer_unref(&av_hw_device);
-        return;
-    }
     initialized = true;
 }
 
@@ -133,6 +172,9 @@ void Codec::Decode() {
     if (is_first_frame) {
         Initialize();
     }
+    if (!initialized) {
+        return;
+    }
     bool vp9_hidden_frame = false;
     std::vector<u8> frame_data;
     if (current_codec == NvdecCommon::VideoCodec::H264) {
@@ -141,50 +183,48 @@ void Codec::Decode() {
         frame_data = vp9_decoder->ComposeFrameHeader(state);
         vp9_hidden_frame = vp9_decoder->WasFrameHidden();
     }
-    AVPacket packet{};
-    av_init_packet(&packet);
-    packet.data = frame_data.data();
-    packet.size = static_cast<s32>(frame_data.size());
-    if (const int ret = avcodec_send_packet(av_codec_ctx, &packet); ret) {
-        LOG_DEBUG(Service_NVDRV, "avcodec_send_packet error {}", ret);
+    AVPacketPtr packet{av_packet_alloc(), AVPacketDeleter};
+    if (!packet) {
+        LOG_ERROR(Service_NVDRV, "av_packet_alloc failed");
+        return;
+    }
+    packet->data = frame_data.data();
+    packet->size = static_cast<s32>(frame_data.size());
+    if (const int res = avcodec_send_packet(av_codec_ctx, packet.get()); res != 0) {
+        LOG_DEBUG(Service_NVDRV, "avcodec_send_packet error {}", res);
         return;
     }
     // Only receive/store visible frames
     if (vp9_hidden_frame) {
         return;
     }
-    AVFrame* hw_frame = av_frame_alloc();
-    AVFrame* sw_frame = hw_frame;
-    ASSERT_MSG(hw_frame, "av_frame_alloc hw_frame failed");
-    if (const int ret = avcodec_receive_frame(av_codec_ctx, hw_frame); ret) {
+    AVFramePtr initial_frame{av_frame_alloc(), AVFrameDeleter};
+    AVFramePtr final_frame{nullptr, AVFrameDeleter};
+    ASSERT_MSG(initial_frame, "av_frame_alloc initial_frame failed");
+    if (const int ret = avcodec_receive_frame(av_codec_ctx, initial_frame.get()); ret) {
         LOG_DEBUG(Service_NVDRV, "avcodec_receive_frame error {}", ret);
-        av_frame_free(&hw_frame);
         return;
     }
-    if (!hw_frame->width || !hw_frame->height) {
+    if (initial_frame->width == 0 || initial_frame->height == 0) {
         LOG_WARNING(Service_NVDRV, "Zero width or height in frame");
-        av_frame_free(&hw_frame);
         return;
     }
-#if defined(LIBVA_FOUND)
-    // Hardware acceleration code from FFmpeg/doc/examples/hw_decode.c under MIT license
-    if (hw_frame->format == AV_PIX_FMT_VAAPI) {
-        sw_frame = av_frame_alloc();
-        ASSERT_MSG(sw_frame, "av_frame_alloc sw_frame failed");
+    if (av_codec_ctx->hw_device_ctx) {
+        final_frame = AVFramePtr{av_frame_alloc(), AVFrameDeleter};
+        ASSERT_MSG(final_frame, "av_frame_alloc final_frame failed");
         // Can't use AV_PIX_FMT_YUV420P and share code with software decoding in vic.cpp
         // because Intel drivers crash unless using AV_PIX_FMT_NV12
-        sw_frame->format = AV_PIX_FMT_NV12;
-        const int transfer_data_ret = av_hwframe_transfer_data(sw_frame, hw_frame, 0);
-        ASSERT_MSG(!transfer_data_ret, "av_hwframe_transfer_data error {}", transfer_data_ret);
-        av_frame_free(&hw_frame);
+        final_frame->format = PREFERRED_GPU_FMT;
+        const int ret = av_hwframe_transfer_data(final_frame.get(), initial_frame.get(), 0);
+        ASSERT_MSG(!ret, "av_hwframe_transfer_data error {}", ret);
+    } else {
+        final_frame = std::move(initial_frame);
     }
-#endif
-    if (sw_frame->format != AV_PIX_FMT_YUV420P && sw_frame->format != AV_PIX_FMT_NV12) {
-        UNIMPLEMENTED_MSG("Unexpected video format from host graphics: {}", sw_frame->format);
-        av_frame_free(&sw_frame);
+    if (final_frame->format != PREFERRED_CPU_FMT && final_frame->format != PREFERRED_GPU_FMT) {
+        UNIMPLEMENTED_MSG("Unexpected video format: {}", final_frame->format);
         return;
     }
-    av_frames.push(AVFramePtr{sw_frame, AVFrameDeleter});
+    av_frames.push(std::move(final_frame));
     if (av_frames.size() > 10) {
         LOG_TRACE(Service_NVDRV, "av_frames.push overflow dropped frame");
         av_frames.pop();

src/video_core/command_classes/codecs/codec.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <memory>
+#include <string_view>
 #include <queue>
 #include "common/common_types.h"
 #include "video_core/command_classes/nvdec_common.h"
@@ -50,18 +51,23 @@ public:
 
     /// Returns the value of current_codec
     [[nodiscard]] NvdecCommon::VideoCodec GetCurrentCodec() const;
+
     /// Return name of the current codec
     [[nodiscard]] std::string_view GetCurrentCodecName() const;
 
 private:
-    void InitializeHwdec();
+    void InitializeAvCodecContext();
+
+    void InitializeGpuDecoder();
+
+    bool CreateGpuAvDevice();
 
     bool initialized{};
     NvdecCommon::VideoCodec current_codec{NvdecCommon::VideoCodec::None};
 
     AVCodec* av_codec{nullptr};
-    AVBufferRef* av_hw_device{nullptr};
     AVCodecContext* av_codec_ctx{nullptr};
+    AVBufferRef* av_gpu_decoder{nullptr};
 
     GPU& gpu;
     const NvdecCommon::NvdecRegisters& state;

src/video_core/command_classes/codecs/h264.cpp

@@ -95,7 +95,8 @@ const std::vector<u8>& H264::ComposeFrameHeader(const NvdecCommon::NvdecRegister
     const s32 pic_height = context.h264_parameter_set.frame_height_in_map_units /
                            (context.h264_parameter_set.frame_mbs_only_flag ? 1 : 2);
 
-    writer.WriteUe(16);
+    // TODO (ameerj): Where do we get this number, it seems to be particular for each stream
+    writer.WriteUe(6); // Max number of reference frames
     writer.WriteBit(false);
     writer.WriteUe(context.h264_parameter_set.pic_width_in_mbs - 1);
     writer.WriteUe(pic_height - 1);

src/video_core/engines/maxwell_3d.h

@@ -475,10 +475,10 @@ public:
 
                 // These values are used by Nouveau and some games.
                 AddGL = 0x8006,
-                SubtractGL = 0x8007,
-                ReverseSubtractGL = 0x8008,
-                MinGL = 0x800a,
-                MaxGL = 0x800b
+                MinGL = 0x8007,
+                MaxGL = 0x8008,
+                SubtractGL = 0x800a,
+                ReverseSubtractGL = 0x800b
             };
 
             enum class Factor : u32 {

src/video_core/memory_manager.cpp

@@ -2,6 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <algorithm>
+
 #include "common/alignment.h"
 #include "common/assert.h"
 #include "common/logging/log.h"

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -293,6 +293,8 @@ void ShaderCache::LoadDiskResources(u64 title_id, std::stop_token stop_loading,
     }};
     LoadPipelines(stop_loading, shader_cache_filename, CACHE_VERSION, load_compute, load_graphics);
 
+    LOG_INFO(Render_OpenGL, "Total Pipeline Count: {}", state.total);
+
     std::unique_lock lock{state.mutex};
     callback(VideoCore::LoadCallbackStage::Build, 0, state.total);
     state.has_loaded = true;

src/video_core/renderer_vulkan/vk_blit_screen.cpp

@@ -358,7 +358,7 @@ void VKBlitScreen::CreateDescriptorPool() {
 void VKBlitScreen::CreateRenderPass() {
     const VkAttachmentDescription color_attachment{
         .flags = 0,
-        .format = swapchain.GetImageFormat(),
+        .format = swapchain.GetImageViewFormat(),
         .samples = VK_SAMPLE_COUNT_1_BIT,
         .loadOp = VK_ATTACHMENT_LOAD_OP_CLEAR,
         .storeOp = VK_ATTACHMENT_STORE_OP_STORE,

src/video_core/renderer_vulkan/vk_descriptor_pool.cpp

@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <algorithm>
 #include <mutex>
 #include <span>
 #include <vector>
@@ -18,7 +19,6 @@ namespace Vulkan {
 // Prefer small grow rates to avoid saturating the descriptor pool with barely used pipelines
 constexpr size_t SETS_GROW_RATE = 16;
 constexpr s32 SCORE_THRESHOLD = 3;
-constexpr u32 SETS_PER_POOL = 64;
 
 struct DescriptorBank {
     DescriptorBankInfo info;
@@ -58,11 +58,12 @@ static DescriptorBankInfo MakeBankInfo(std::span<const Shader::Info> infos) {
 static void AllocatePool(const Device& device, DescriptorBank& bank) {
     std::array<VkDescriptorPoolSize, 6> pool_sizes;
     size_t pool_cursor{};
+    const u32 sets_per_pool = device.GetSetsPerPool();
     const auto add = [&](VkDescriptorType type, u32 count) {
         if (count > 0) {
             pool_sizes[pool_cursor++] = {
                 .type = type,
-                .descriptorCount = count * SETS_PER_POOL,
+                .descriptorCount = count * sets_per_pool,
             };
         }
     };
@@ -77,7 +78,7 @@ static void AllocatePool(const Device& device, DescriptorBank& bank) {
         .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_POOL_CREATE_INFO,
         .pNext = nullptr,
         .flags = VK_DESCRIPTOR_POOL_CREATE_FREE_DESCRIPTOR_SET_BIT,
-        .maxSets = SETS_PER_POOL,
+        .maxSets = sets_per_pool,
         .poolSizeCount = static_cast<u32>(pool_cursor),
         .pPoolSizes = std::data(pool_sizes),
     }));

src/video_core/renderer_vulkan/vk_pipeline_cache.cpp

@@ -447,6 +447,8 @@ void PipelineCache::LoadDiskResources(u64 title_id, std::stop_token stop_loading
     VideoCommon::LoadPipelines(stop_loading, pipeline_cache_filename, CACHE_VERSION, load_compute,
                                load_graphics);
 
+    LOG_INFO(Render_Vulkan, "Total Pipeline Count: {}", state.total);
+
     std::unique_lock lock{state.mutex};
     callback(VideoCore::LoadCallbackStage::Build, 0, state.total);
     state.has_loaded = true;

src/video_core/renderer_vulkan/vk_rasterizer.cpp

@@ -228,9 +228,7 @@ void RasterizerVulkan::Clear() {
     };
 
     const u32 color_attachment = regs.clear_buffers.RT;
-    const auto attachment_aspect_mask = framebuffer->ImageRanges()[color_attachment].aspectMask;
-    const bool is_color_rt = (attachment_aspect_mask & VK_IMAGE_ASPECT_COLOR_BIT) != 0;
-    if (use_color && is_color_rt) {
+    if (use_color && framebuffer->HasAspectColorBit(color_attachment)) {
         VkClearValue clear_value;
         std::memcpy(clear_value.color.float32, regs.clear_color, sizeof(regs.clear_color));
 
@@ -248,12 +246,15 @@ void RasterizerVulkan::Clear() {
         return;
     }
     VkImageAspectFlags aspect_flags = 0;
-    if (use_depth) {
+    if (use_depth && framebuffer->HasAspectDepthBit()) {
         aspect_flags |= VK_IMAGE_ASPECT_DEPTH_BIT;
     }
-    if (use_stencil) {
+    if (use_stencil && framebuffer->HasAspectStencilBit()) {
         aspect_flags |= VK_IMAGE_ASPECT_STENCIL_BIT;
     }
+    if (aspect_flags == 0) {
+        return;
+    }
     scheduler.Record([clear_depth = regs.clear_depth, clear_stencil = regs.clear_stencil,
                       clear_rect, aspect_flags](vk::CommandBuffer cmdbuf) {
         VkClearAttachment attachment;
@@ -764,12 +765,7 @@ void RasterizerVulkan::UpdateStencilOp(Tegra::Engines::Maxwell3D::Regs& regs) {
     const Maxwell::StencilOp zpass = regs.stencil_front_op_zpass;
     const Maxwell::ComparisonOp compare = regs.stencil_front_func_func;
     if (regs.stencil_two_side_enable) {
-        scheduler.Record([fail, zfail, zpass, compare](vk::CommandBuffer cmdbuf) {
-            cmdbuf.SetStencilOpEXT(VK_STENCIL_FACE_FRONT_AND_BACK, MaxwellToVK::StencilOp(fail),
-                                   MaxwellToVK::StencilOp(zpass), MaxwellToVK::StencilOp(zfail),
-                                   MaxwellToVK::ComparisonOp(compare));
-        });
-    } else {
+        // Separate stencil op per face
         const Maxwell::StencilOp back_fail = regs.stencil_back_op_fail;
         const Maxwell::StencilOp back_zfail = regs.stencil_back_op_zfail;
         const Maxwell::StencilOp back_zpass = regs.stencil_back_op_zpass;
@@ -784,6 +780,13 @@ void RasterizerVulkan::UpdateStencilOp(Tegra::Engines::Maxwell3D::Regs& regs) {
                                    MaxwellToVK::StencilOp(back_zfail),
                                    MaxwellToVK::ComparisonOp(back_compare));
         });
+    } else {
+        // Front face defines the stencil op of both faces
+        scheduler.Record([fail, zfail, zpass, compare](vk::CommandBuffer cmdbuf) {
+            cmdbuf.SetStencilOpEXT(VK_STENCIL_FACE_FRONT_AND_BACK, MaxwellToVK::StencilOp(fail),
+                                   MaxwellToVK::StencilOp(zpass), MaxwellToVK::StencilOp(zfail),
+                                   MaxwellToVK::ComparisonOp(compare));
+        });
     }
 }
 

src/video_core/renderer_vulkan/vk_state_tracker.h

@@ -110,10 +110,6 @@ public:
         return Exchange(Dirty::DepthTestEnable, false);
     }
 
-    bool TouchDepthBoundsEnable() {
-        return Exchange(Dirty::DepthBoundsEnable, false);
-    }
-
     bool TouchDepthWriteEnable() {
         return Exchange(Dirty::DepthWriteEnable, false);
     }

src/video_core/renderer_vulkan/vk_swapchain.cpp

@@ -20,16 +20,15 @@ namespace Vulkan {
 
 namespace {
 
-VkSurfaceFormatKHR ChooseSwapSurfaceFormat(vk::Span<VkSurfaceFormatKHR> formats, bool srgb) {
+VkSurfaceFormatKHR ChooseSwapSurfaceFormat(vk::Span<VkSurfaceFormatKHR> formats) {
     if (formats.size() == 1 && formats[0].format == VK_FORMAT_UNDEFINED) {
         VkSurfaceFormatKHR format;
         format.format = VK_FORMAT_B8G8R8A8_UNORM;
         format.colorSpace = VK_COLOR_SPACE_SRGB_NONLINEAR_KHR;
         return format;
     }
-    const auto& found = std::find_if(formats.begin(), formats.end(), [srgb](const auto& format) {
-        const auto request_format = srgb ? VK_FORMAT_B8G8R8A8_SRGB : VK_FORMAT_B8G8R8A8_UNORM;
-        return format.format == request_format &&
+    const auto& found = std::find_if(formats.begin(), formats.end(), [](const auto& format) {
+        return format.format == VK_FORMAT_B8G8R8A8_UNORM &&
                format.colorSpace == VK_COLOR_SPACE_SRGB_NONLINEAR_KHR;
     });
     return found != formats.end() ? *found : formats[0];
@@ -143,7 +142,7 @@ void VKSwapchain::CreateSwapchain(const VkSurfaceCapabilitiesKHR& capabilities,
     const auto formats{physical_device.GetSurfaceFormatsKHR(surface)};
     const auto present_modes{physical_device.GetSurfacePresentModesKHR(surface)};
 
-    const VkSurfaceFormatKHR surface_format{ChooseSwapSurfaceFormat(formats, srgb)};
+    const VkSurfaceFormatKHR surface_format{ChooseSwapSurfaceFormat(formats)};
     const VkPresentModeKHR present_mode{ChooseSwapPresentMode(present_modes)};
 
     u32 requested_image_count{capabilities.minImageCount + 1};
@@ -178,6 +177,17 @@ void VKSwapchain::CreateSwapchain(const VkSurfaceCapabilitiesKHR& capabilities,
         swapchain_ci.queueFamilyIndexCount = static_cast<u32>(queue_indices.size());
         swapchain_ci.pQueueFamilyIndices = queue_indices.data();
     }
+    static constexpr std::array view_formats{VK_FORMAT_B8G8R8A8_UNORM, VK_FORMAT_B8G8R8A8_SRGB};
+    VkImageFormatListCreateInfo format_list{
+        .sType = VK_STRUCTURE_TYPE_IMAGE_FORMAT_LIST_CREATE_INFO_KHR,
+        .pNext = nullptr,
+        .viewFormatCount = static_cast<u32>(view_formats.size()),
+        .pViewFormats = view_formats.data(),
+    };
+    if (device.IsKhrSwapchainMutableFormatEnabled()) {
+        format_list.pNext = std::exchange(swapchain_ci.pNext, &format_list);
+        swapchain_ci.flags |= VK_SWAPCHAIN_CREATE_MUTABLE_FORMAT_BIT_KHR;
+    }
     // Request the size again to reduce the possibility of a TOCTOU race condition.
     const auto updated_capabilities = physical_device.GetSurfaceCapabilitiesKHR(surface);
     swapchain_ci.imageExtent = ChooseSwapExtent(updated_capabilities, width, height);
@@ -189,7 +199,7 @@ void VKSwapchain::CreateSwapchain(const VkSurfaceCapabilitiesKHR& capabilities,
 
     images = swapchain.GetImages();
     image_count = static_cast<u32>(images.size());
-    image_format = surface_format.format;
+    image_view_format = srgb ? VK_FORMAT_B8G8R8A8_SRGB : VK_FORMAT_B8G8R8A8_UNORM;
 }
 
 void VKSwapchain::CreateSemaphores() {
@@ -205,7 +215,7 @@ void VKSwapchain::CreateImageViews() {
         .flags = 0,
         .image = {},
         .viewType = VK_IMAGE_VIEW_TYPE_2D,
-        .format = image_format,
+        .format = image_view_format,
         .components =
             {
                 .r = VK_COMPONENT_SWIZZLE_IDENTITY,

src/video_core/renderer_vulkan/vk_swapchain.h

@@ -68,8 +68,8 @@ public:
         return *image_views[index];
     }
 
-    VkFormat GetImageFormat() const {
-        return image_format;
+    VkFormat GetImageViewFormat() const {
+        return image_view_format;
     }
 
     VkSemaphore CurrentPresentSemaphore() const {
@@ -100,7 +100,7 @@ private:
     u32 image_index{};
     u32 frame_index{};
 
-    VkFormat image_format{};
+    VkFormat image_view_format{};
     VkExtent2D extent{};
 
     bool current_srgb{};

src/video_core/renderer_vulkan/vk_texture_cache.cpp

@@ -1186,9 +1186,12 @@ Framebuffer::Framebuffer(TextureCacheRuntime& runtime, std::span<ImageView*, NUM
         renderpass_key.depth_format = depth_buffer->format;
         num_layers = std::max(num_layers, depth_buffer->range.extent.layers);
         images[num_images] = depth_buffer->ImageHandle();
-        image_ranges[num_images] = MakeSubresourceRange(depth_buffer);
+        const VkImageSubresourceRange subresource_range = MakeSubresourceRange(depth_buffer);
+        image_ranges[num_images] = subresource_range;
         samples = depth_buffer->Samples();
         ++num_images;
+        has_depth = (subresource_range.aspectMask & VK_IMAGE_ASPECT_DEPTH_BIT) != 0;
+        has_stencil = (subresource_range.aspectMask & VK_IMAGE_ASPECT_STENCIL_BIT) != 0;
     } else {
         renderpass_key.depth_format = PixelFormat::Invalid;
     }

src/video_core/renderer_vulkan/vk_texture_cache.h

@@ -232,6 +232,18 @@ public:
         return image_ranges;
     }
 
+    [[nodiscard]] bool HasAspectColorBit(size_t index) const noexcept {
+        return (image_ranges.at(index).aspectMask & VK_IMAGE_ASPECT_COLOR_BIT) != 0;
+    }
+
+    [[nodiscard]] bool HasAspectDepthBit() const noexcept {
+        return has_depth;
+    }
+
+    [[nodiscard]] bool HasAspectStencilBit() const noexcept {
+        return has_stencil;
+    }
+
 private:
     vk::Framebuffer framebuffer;
     VkRenderPass renderpass{};
@@ -241,6 +253,8 @@ private:
     u32 num_images = 0;
     std::array<VkImage, 9> images{};
     std::array<VkImageSubresourceRange, 9> image_ranges{};
+    bool has_depth{};
+    bool has_stencil{};
 };
 
 struct TextureCacheParams {

src/video_core/shader_environment.cpp

@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <algorithm>
 #include <filesystem>
 #include <fstream>
 #include <memory>

src/video_core/texture_cache/slot_vector.h

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <algorithm>
 #include <array>
 #include <bit>
 #include <concepts>

src/video_core/video_core.cpp

@@ -37,7 +37,8 @@ std::unique_ptr<VideoCore::RendererBase> CreateRenderer(
 namespace VideoCore {
 
 std::unique_ptr<Tegra::GPU> CreateGPU(Core::Frontend::EmuWindow& emu_window, Core::System& system) {
-    const bool use_nvdec = Settings::values.use_nvdec_emulation.GetValue();
+    const auto nvdec_value = Settings::values.nvdec_emulation.GetValue();
+    const bool use_nvdec = nvdec_value != Settings::NvdecEmulation::Off;
     const bool use_async = Settings::values.use_asynchronous_gpu_emulation.GetValue();
     auto gpu = std::make_unique<Tegra::GPU>(system, use_async, use_nvdec);
     auto context = emu_window.CreateSharedContext();

src/video_core/vulkan_common/vulkan_debug_callback.cpp

@@ -16,6 +16,7 @@ VkBool32 Callback(VkDebugUtilsMessageSeverityFlagBitsEXT severity,
     switch (static_cast<u32>(data->messageIdNumber)) {
     case 0x682a878au: // VUID-vkCmdBindVertexBuffers2EXT-pBuffers-parameter
     case 0x99fb7dfdu: // UNASSIGNED-RequiredParameter (vkCmdBindVertexBuffers2EXT pBuffers[0])
+    case 0xe8616bf2u: // Bound VkDescriptorSet 0x0[] was destroyed. Likely push_descriptor related
         return VK_FALSE;
     default:
         break;

src/video_core/vulkan_common/vulkan_device.cpp

@@ -368,8 +368,9 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
     };
     SetNext(next, demote);
 
+    VkPhysicalDeviceFloat16Int8FeaturesKHR float16_int8;
     if (is_int8_supported || is_float16_supported) {
-        VkPhysicalDeviceFloat16Int8FeaturesKHR float16_int8{
+        float16_int8 = {
             .sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_FLOAT16_INT8_FEATURES_KHR,
             .pNext = nullptr,
             .shaderFloat16 = is_float16_supported,
@@ -587,6 +588,26 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
             ext_extended_dynamic_state = false;
         }
     }
+
+    sets_per_pool = 64;
+    if (driver_id == VK_DRIVER_ID_AMD_PROPRIETARY || driver_id == VK_DRIVER_ID_AMD_OPEN_SOURCE) {
+        // AMD drivers need a higher amount of Sets per Pool in certain circunstances like in XC2.
+        sets_per_pool = 96;
+    }
+
+    const bool is_amd = driver_id == VK_DRIVER_ID_AMD_PROPRIETARY ||
+                        driver_id == VK_DRIVER_ID_MESA_RADV ||
+                        driver_id == VK_DRIVER_ID_AMD_OPEN_SOURCE;
+    if (ext_sampler_filter_minmax && is_amd) {
+        // Disable ext_sampler_filter_minmax on AMD GCN4 and lower as it is broken.
+        if (!is_float16_supported) {
+            LOG_WARNING(
+                Render_Vulkan,
+                "Blacklisting AMD GCN4 and lower for VK_EXT_SAMPLER_FILTER_MINMAX_EXTENSION_NAME");
+            ext_sampler_filter_minmax = false;
+        }
+    }
+
     if (ext_vertex_input_dynamic_state && driver_id == VK_DRIVER_ID_INTEL_PROPRIETARY_WINDOWS) {
         LOG_WARNING(Render_Vulkan, "Blacklisting Intel for VK_EXT_vertex_input_dynamic_state");
         ext_vertex_input_dynamic_state = false;
@@ -839,6 +860,8 @@ std::vector<const char*> Device::LoadExtensions(bool requires_surface) {
     bool has_khr_shader_float16_int8{};
     bool has_khr_workgroup_memory_explicit_layout{};
     bool has_khr_pipeline_executable_properties{};
+    bool has_khr_image_format_list{};
+    bool has_khr_swapchain_mutable_format{};
     bool has_ext_subgroup_size_control{};
     bool has_ext_transform_feedback{};
     bool has_ext_custom_border_color{};
@@ -888,6 +911,9 @@ std::vector<const char*> Device::LoadExtensions(bool requires_surface) {
         test(has_ext_shader_atomic_int64, VK_KHR_SHADER_ATOMIC_INT64_EXTENSION_NAME, false);
         test(has_khr_workgroup_memory_explicit_layout,
              VK_KHR_WORKGROUP_MEMORY_EXPLICIT_LAYOUT_EXTENSION_NAME, false);
+        test(has_khr_image_format_list, VK_KHR_IMAGE_FORMAT_LIST_EXTENSION_NAME, false);
+        test(has_khr_swapchain_mutable_format, VK_KHR_SWAPCHAIN_MUTABLE_FORMAT_EXTENSION_NAME,
+             false);
         test(has_ext_line_rasterization, VK_EXT_LINE_RASTERIZATION_EXTENSION_NAME, false);
         if (Settings::values.enable_nsight_aftermath) {
             test(nv_device_diagnostics_config, VK_NV_DEVICE_DIAGNOSTICS_CONFIG_EXTENSION_NAME,
@@ -1066,6 +1092,11 @@ std::vector<const char*> Device::LoadExtensions(bool requires_surface) {
             khr_pipeline_executable_properties = true;
         }
     }
+    if (has_khr_image_format_list && has_khr_swapchain_mutable_format) {
+        extensions.push_back(VK_KHR_IMAGE_FORMAT_LIST_EXTENSION_NAME);
+        extensions.push_back(VK_KHR_SWAPCHAIN_MUTABLE_FORMAT_EXTENSION_NAME);
+        khr_swapchain_mutable_format = true;
+    }
     if (khr_push_descriptor) {
         VkPhysicalDevicePushDescriptorPropertiesKHR push_descriptor;
         push_descriptor.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PUSH_DESCRIPTOR_PROPERTIES_KHR;

src/video_core/vulkan_common/vulkan_device.h

@@ -224,6 +224,11 @@ public:
         return khr_pipeline_executable_properties;
     }
 
+    /// Returns true if VK_KHR_swapchain_mutable_format is enabled.
+    bool IsKhrSwapchainMutableFormatEnabled() const {
+        return khr_swapchain_mutable_format;
+    }
+
     /// Returns true if the device supports VK_KHR_workgroup_memory_explicit_layout.
     bool IsKhrWorkgroupMemoryExplicitLayoutSupported() const {
         return khr_workgroup_memory_explicit_layout;
@@ -318,6 +323,10 @@ public:
         return device_access_memory;
     }
 
+    u32 GetSetsPerPool() const {
+        return sets_per_pool;
+    }
+
 private:
     /// Checks if the physical device is suitable.
     void CheckSuitability(bool requires_swapchain) const;
@@ -371,6 +380,7 @@ private:
     VkShaderStageFlags guest_warp_stages{};     ///< Stages where the guest warp size can be forced.
     u64 device_access_memory{};                 ///< Total size of device local memory in bytes.
     u32 max_push_descriptors{};                 ///< Maximum number of push descriptors
+    u32 sets_per_pool{};                        ///< Sets per Description Pool
     bool is_optimal_astc_supported{};           ///< Support for native ASTC.
     bool is_float16_supported{};                ///< Support for float16 arithmetic.
     bool is_int8_supported{};                   ///< Support for int8 arithmetic.
@@ -390,6 +400,7 @@ private:
     bool khr_workgroup_memory_explicit_layout{}; ///< Support for explicit workgroup layouts.
     bool khr_push_descriptor{};                  ///< Support for VK_KHR_push_descritor.
     bool khr_pipeline_executable_properties{};   ///< Support for executable properties.
+    bool khr_swapchain_mutable_format{};         ///< Support for VK_KHR_swapchain_mutable_format.
     bool ext_index_type_uint8{};                 ///< Support for VK_EXT_index_type_uint8.
     bool ext_sampler_filter_minmax{};            ///< Support for VK_EXT_sampler_filter_minmax.
     bool ext_depth_range_unrestricted{};         ///< Support for VK_EXT_depth_range_unrestricted.

src/yuzu/CMakeLists.txt

@@ -108,6 +108,9 @@ add_executable(yuzu
     configuration/configure_system.cpp
     configuration/configure_system.h
     configuration/configure_system.ui
+    configuration/configure_tas.cpp
+    configuration/configure_tas.h
+    configuration/configure_tas.ui
     configuration/configure_touch_from_button.cpp
     configuration/configure_touch_from_button.h
     configuration/configure_touch_from_button.ui

src/yuzu/bootmanager.cpp

@@ -36,6 +36,7 @@
 #include "input_common/keyboard.h"
 #include "input_common/main.h"
 #include "input_common/mouse/mouse_input.h"
+#include "input_common/tas/tas_input.h"
 #include "video_core/renderer_base.h"
 #include "video_core/video_core.h"
 #include "yuzu/bootmanager.h"
@@ -312,6 +313,7 @@ GRenderWindow::~GRenderWindow() {
 }
 
 void GRenderWindow::OnFrameDisplayed() {
+    input_subsystem->GetTas()->UpdateThread();
     if (!first_frame) {
         first_frame = true;
         emit FirstFrameDisplayed();

src/yuzu/configuration/config.cpp

@@ -221,7 +221,7 @@ const std::array<int, Settings::NativeKeyboard::NumKeyboardMods> Config::default
 // This must be in alphabetical order according to action name as it must have the same order as
 // UISetting::values.shortcuts, which is alphabetically ordered.
 // clang-format off
-const std::array<UISettings::Shortcut, 18> Config::default_hotkeys{{
+const std::array<UISettings::Shortcut, 21> Config::default_hotkeys{{
     {QStringLiteral("Capture Screenshot"),       QStringLiteral("Main Window"), {QStringLiteral("Ctrl+P"), Qt::WidgetWithChildrenShortcut}},
     {QStringLiteral("Change Docked Mode"),       QStringLiteral("Main Window"), {QStringLiteral("F10"), Qt::ApplicationShortcut}},
     {QStringLiteral("Continue/Pause Emulation"), QStringLiteral("Main Window"), {QStringLiteral("F4"), Qt::WindowShortcut}},
@@ -235,6 +235,9 @@ const std::array<UISettings::Shortcut, 18> Config::default_hotkeys{{
     {QStringLiteral("Mute Audio"),               QStringLiteral("Main Window"), {QStringLiteral("Ctrl+M"), Qt::WindowShortcut}},
     {QStringLiteral("Restart Emulation"),        QStringLiteral("Main Window"), {QStringLiteral("F6"), Qt::WindowShortcut}},
     {QStringLiteral("Stop Emulation"),           QStringLiteral("Main Window"), {QStringLiteral("F5"), Qt::WindowShortcut}},
+    {QStringLiteral("TAS Start/Stop"),           QStringLiteral("Main Window"), {QStringLiteral("Ctrl+F5"), Qt::ApplicationShortcut}},
+    {QStringLiteral("TAS Reset"),                QStringLiteral("Main Window"), {QStringLiteral("Ctrl+F6"), Qt::ApplicationShortcut}},
+    {QStringLiteral("TAS Record"),               QStringLiteral("Main Window"), {QStringLiteral("Ctrl+F7"), Qt::ApplicationShortcut}},
     {QStringLiteral("Toggle Filter Bar"),        QStringLiteral("Main Window"), {QStringLiteral("Ctrl+F"), Qt::WindowShortcut}},
     {QStringLiteral("Toggle Framerate Limit"),   QStringLiteral("Main Window"), {QStringLiteral("Ctrl+U"), Qt::ApplicationShortcut}},
     {QStringLiteral("Toggle Mouse Panning"),     QStringLiteral("Main Window"), {QStringLiteral("Ctrl+F9"), Qt::ApplicationShortcut}},
@@ -542,7 +545,6 @@ void Config::ReadAudioValues() {
         ReadBasicSetting(Settings::values.audio_device_id);
         ReadBasicSetting(Settings::values.sink_id);
     }
-    ReadGlobalSetting(Settings::values.enable_audio_stretching);
     ReadGlobalSetting(Settings::values.volume);
 
     qt_config->endGroup();
@@ -560,10 +562,16 @@ void Config::ReadControlValues() {
     ReadTouchscreenValues();
     ReadMotionTouchValues();
 
+    ReadBasicSetting(Settings::values.enable_raw_input);
     ReadBasicSetting(Settings::values.emulate_analog_keyboard);
     Settings::values.mouse_panning = false;
     ReadBasicSetting(Settings::values.mouse_panning_sensitivity);
 
+    ReadBasicSetting(Settings::values.tas_enable);
+    ReadBasicSetting(Settings::values.tas_loop);
+    ReadBasicSetting(Settings::values.tas_swap_controllers);
+    ReadBasicSetting(Settings::values.pause_tas_on_load);
+
     ReadGlobalSetting(Settings::values.use_docked_mode);
 
     // Disable docked mode if handheld is selected
@@ -661,6 +669,13 @@ void Config::ReadDataStorageValues() {
                     QString::fromStdString(FS::GetYuzuPathString(FS::YuzuPath::DumpDir)))
             .toString()
             .toStdString());
+    FS::SetYuzuPath(FS::YuzuPath::TASDir,
+                    qt_config
+                        ->value(QStringLiteral("tas_directory"),
+                                QString::fromStdString(FS::GetYuzuPathString(FS::YuzuPath::TASDir)))
+                        .toString()
+                        .toStdString());
+
     ReadBasicSetting(Settings::values.gamecard_inserted);
     ReadBasicSetting(Settings::values.gamecard_current_game);
     ReadBasicSetting(Settings::values.gamecard_path);
@@ -812,7 +827,7 @@ void Config::ReadRendererValues() {
     ReadGlobalSetting(Settings::values.use_disk_shader_cache);
     ReadGlobalSetting(Settings::values.gpu_accuracy);
     ReadGlobalSetting(Settings::values.use_asynchronous_gpu_emulation);
-    ReadGlobalSetting(Settings::values.use_nvdec_emulation);
+    ReadGlobalSetting(Settings::values.nvdec_emulation);
     ReadGlobalSetting(Settings::values.accelerate_astc);
     ReadGlobalSetting(Settings::values.use_vsync);
     ReadGlobalSetting(Settings::values.shader_backend);
@@ -1163,7 +1178,6 @@ void Config::SaveAudioValues() {
         WriteBasicSetting(Settings::values.sink_id);
         WriteBasicSetting(Settings::values.audio_device_id);
     }
-    WriteGlobalSetting(Settings::values.enable_audio_stretching);
     WriteGlobalSetting(Settings::values.volume);
 
     qt_config->endGroup();
@@ -1184,10 +1198,16 @@ void Config::SaveControlValues() {
     WriteGlobalSetting(Settings::values.vibration_enabled);
     WriteGlobalSetting(Settings::values.enable_accurate_vibrations);
     WriteGlobalSetting(Settings::values.motion_enabled);
+    WriteBasicSetting(Settings::values.enable_raw_input);
     WriteBasicSetting(Settings::values.keyboard_enabled);
     WriteBasicSetting(Settings::values.emulate_analog_keyboard);
     WriteBasicSetting(Settings::values.mouse_panning_sensitivity);
 
+    WriteBasicSetting(Settings::values.tas_enable);
+    WriteBasicSetting(Settings::values.tas_loop);
+    WriteBasicSetting(Settings::values.tas_swap_controllers);
+    WriteBasicSetting(Settings::values.pause_tas_on_load);
+
     qt_config->endGroup();
 }
 
@@ -1215,6 +1235,10 @@ void Config::SaveDataStorageValues() {
     WriteSetting(QStringLiteral("dump_directory"),
                  QString::fromStdString(FS::GetYuzuPathString(FS::YuzuPath::DumpDir)),
                  QString::fromStdString(FS::GetYuzuPathString(FS::YuzuPath::DumpDir)));
+    WriteSetting(QStringLiteral("tas_directory"),
+                 QString::fromStdString(FS::GetYuzuPathString(FS::YuzuPath::TASDir)),
+                 QString::fromStdString(FS::GetYuzuPathString(FS::YuzuPath::TASDir)));
+
     WriteBasicSetting(Settings::values.gamecard_inserted);
     WriteBasicSetting(Settings::values.gamecard_current_game);
     WriteBasicSetting(Settings::values.gamecard_path);
@@ -1349,7 +1373,10 @@ void Config::SaveRendererValues() {
                  static_cast<u32>(Settings::values.gpu_accuracy.GetDefault()),
                  Settings::values.gpu_accuracy.UsingGlobal());
     WriteGlobalSetting(Settings::values.use_asynchronous_gpu_emulation);
-    WriteGlobalSetting(Settings::values.use_nvdec_emulation);
+    WriteSetting(QString::fromStdString(Settings::values.nvdec_emulation.GetLabel()),
+                 static_cast<u32>(Settings::values.nvdec_emulation.GetValue(global)),
+                 static_cast<u32>(Settings::values.nvdec_emulation.GetDefault()),
+                 Settings::values.nvdec_emulation.UsingGlobal());
     WriteGlobalSetting(Settings::values.accelerate_astc);
     WriteGlobalSetting(Settings::values.use_vsync);
     WriteSetting(QString::fromStdString(Settings::values.shader_backend.GetLabel()),

src/yuzu/configuration/config.h

@@ -42,7 +42,7 @@ public:
         default_mouse_buttons;
     static const std::array<int, Settings::NativeKeyboard::NumKeyboardKeys> default_keyboard_keys;
     static const std::array<int, Settings::NativeKeyboard::NumKeyboardMods> default_keyboard_mods;
-    static const std::array<UISettings::Shortcut, 18> default_hotkeys;
+    static const std::array<UISettings::Shortcut, 21> default_hotkeys;
 
 private:
     void Initialize(const std::string& config_name);
@@ -182,5 +182,6 @@ private:
 Q_DECLARE_METATYPE(Settings::CPUAccuracy);
 Q_DECLARE_METATYPE(Settings::GPUAccuracy);
 Q_DECLARE_METATYPE(Settings::FullscreenMode);
+Q_DECLARE_METATYPE(Settings::NvdecEmulation);
 Q_DECLARE_METATYPE(Settings::RendererBackend);
 Q_DECLARE_METATYPE(Settings::ShaderBackend);

src/yuzu/configuration/configure_audio.cpp

@@ -50,8 +50,6 @@ void ConfigureAudio::SetConfiguration() {
     const auto volume_value = static_cast<int>(Settings::values.volume.GetValue());
     ui->volume_slider->setValue(volume_value);
 
-    ui->toggle_audio_stretching->setChecked(Settings::values.enable_audio_stretching.GetValue());
-
     if (!Settings::IsConfiguringGlobal()) {
         if (Settings::values.volume.UsingGlobal()) {
             ui->volume_combo_box->setCurrentIndex(0);
@@ -100,8 +98,6 @@ void ConfigureAudio::SetVolumeIndicatorText(int percentage) {
 }
 
 void ConfigureAudio::ApplyConfiguration() {
-    ConfigurationShared::ApplyPerGameSetting(&Settings::values.enable_audio_stretching,
-                                             ui->toggle_audio_stretching, enable_audio_stretching);
 
     if (Settings::IsConfiguringGlobal()) {
         Settings::values.sink_id =
@@ -162,15 +158,10 @@ void ConfigureAudio::RetranslateUI() {
 void ConfigureAudio::SetupPerGameUI() {
     if (Settings::IsConfiguringGlobal()) {
         ui->volume_slider->setEnabled(Settings::values.volume.UsingGlobal());
-        ui->toggle_audio_stretching->setEnabled(
-            Settings::values.enable_audio_stretching.UsingGlobal());
 
         return;
     }
 
-    ConfigurationShared::SetColoredTristate(ui->toggle_audio_stretching,
-                                            Settings::values.enable_audio_stretching,
-                                            enable_audio_stretching);
     connect(ui->volume_combo_box, qOverload<int>(&QComboBox::activated), this, [this](int index) {
         ui->volume_slider->setEnabled(index == 1);
         ConfigurationShared::SetHighlight(ui->volume_layout, index == 1);

src/yuzu/configuration/configure_audio.h

@@ -41,6 +41,4 @@ private:
     void SetupPerGameUI();
 
     std::unique_ptr<Ui::ConfigureAudio> ui;
-
-    ConfigurationShared::CheckState enable_audio_stretching;
 };

src/yuzu/configuration/configure_audio.ui

@@ -31,16 +31,6 @@
         </item>
        </layout>
       </item>
-      <item>
-       <widget class="QCheckBox" name="toggle_audio_stretching">
-        <property name="toolTip">
-         <string>This post-processing effect adjusts audio speed to match emulation speed and helps prevent audio stutter. This however increases audio latency.</string>
-        </property>
-        <property name="text">
-         <string>Enable audio stretching</string>
-        </property>
-       </widget>
-      </item>
       <item>
        <layout class="QHBoxLayout" name="_2">
         <item>

src/yuzu/configuration/configure_graphics.cpp

@@ -88,24 +88,30 @@ void ConfigureGraphics::SetConfiguration() {
     ui->api_widget->setEnabled(runtime_lock);
     ui->use_asynchronous_gpu_emulation->setEnabled(runtime_lock);
     ui->use_disk_shader_cache->setEnabled(runtime_lock);
-    ui->use_nvdec_emulation->setEnabled(runtime_lock);
+    ui->nvdec_emulation_widget->setEnabled(runtime_lock);
     ui->accelerate_astc->setEnabled(runtime_lock);
     ui->use_disk_shader_cache->setChecked(Settings::values.use_disk_shader_cache.GetValue());
     ui->use_asynchronous_gpu_emulation->setChecked(
         Settings::values.use_asynchronous_gpu_emulation.GetValue());
-    ui->use_nvdec_emulation->setChecked(Settings::values.use_nvdec_emulation.GetValue());
     ui->accelerate_astc->setChecked(Settings::values.accelerate_astc.GetValue());
 
     if (Settings::IsConfiguringGlobal()) {
         ui->api->setCurrentIndex(static_cast<int>(Settings::values.renderer_backend.GetValue()));
         ui->fullscreen_mode_combobox->setCurrentIndex(
             static_cast<int>(Settings::values.fullscreen_mode.GetValue()));
+        ui->nvdec_emulation->setCurrentIndex(
+            static_cast<int>(Settings::values.nvdec_emulation.GetValue()));
         ui->aspect_ratio_combobox->setCurrentIndex(Settings::values.aspect_ratio.GetValue());
     } else {
         ConfigurationShared::SetPerGameSetting(ui->api, &Settings::values.renderer_backend);
         ConfigurationShared::SetHighlight(ui->api_widget,
                                           !Settings::values.renderer_backend.UsingGlobal());
 
+        ConfigurationShared::SetPerGameSetting(ui->nvdec_emulation,
+                                               &Settings::values.nvdec_emulation);
+        ConfigurationShared::SetHighlight(ui->nvdec_emulation_widget,
+                                          !Settings::values.nvdec_emulation.UsingGlobal());
+
         ConfigurationShared::SetPerGameSetting(ui->fullscreen_mode_combobox,
                                                &Settings::values.fullscreen_mode);
         ConfigurationShared::SetHighlight(ui->fullscreen_mode_label,
@@ -137,8 +143,6 @@ void ConfigureGraphics::ApplyConfiguration() {
     ConfigurationShared::ApplyPerGameSetting(&Settings::values.use_asynchronous_gpu_emulation,
                                              ui->use_asynchronous_gpu_emulation,
                                              use_asynchronous_gpu_emulation);
-    ConfigurationShared::ApplyPerGameSetting(&Settings::values.use_nvdec_emulation,
-                                             ui->use_nvdec_emulation, use_nvdec_emulation);
     ConfigurationShared::ApplyPerGameSetting(&Settings::values.accelerate_astc, ui->accelerate_astc,
                                              accelerate_astc);
 
@@ -147,6 +151,9 @@ void ConfigureGraphics::ApplyConfiguration() {
         if (Settings::values.renderer_backend.UsingGlobal()) {
             Settings::values.renderer_backend.SetValue(GetCurrentGraphicsBackend());
         }
+        if (Settings::values.nvdec_emulation.UsingGlobal()) {
+            Settings::values.nvdec_emulation.SetValue(GetCurrentNvdecEmulation());
+        }
         if (Settings::values.shader_backend.UsingGlobal()) {
             Settings::values.shader_backend.SetValue(shader_backend);
         }
@@ -180,6 +187,13 @@ void ConfigureGraphics::ApplyConfiguration() {
             }
         }
 
+        if (ui->nvdec_emulation->currentIndex() == ConfigurationShared::USE_GLOBAL_INDEX) {
+            Settings::values.nvdec_emulation.SetGlobal(true);
+        } else {
+            Settings::values.nvdec_emulation.SetGlobal(false);
+            Settings::values.nvdec_emulation.SetValue(GetCurrentNvdecEmulation());
+        }
+
         if (ui->bg_combobox->currentIndex() == ConfigurationShared::USE_GLOBAL_INDEX) {
             Settings::values.bg_red.SetGlobal(true);
             Settings::values.bg_green.SetGlobal(true);
@@ -278,6 +292,20 @@ Settings::RendererBackend ConfigureGraphics::GetCurrentGraphicsBackend() const {
                                                   ConfigurationShared::USE_GLOBAL_OFFSET);
 }
 
+Settings::NvdecEmulation ConfigureGraphics::GetCurrentNvdecEmulation() const {
+    if (Settings::IsConfiguringGlobal()) {
+        return static_cast<Settings::NvdecEmulation>(ui->nvdec_emulation->currentIndex());
+    }
+
+    if (ui->nvdec_emulation->currentIndex() == ConfigurationShared::USE_GLOBAL_INDEX) {
+        Settings::values.nvdec_emulation.SetGlobal(true);
+        return Settings::values.nvdec_emulation.GetValue();
+    }
+    Settings::values.nvdec_emulation.SetGlobal(false);
+    return static_cast<Settings::NvdecEmulation>(ui->nvdec_emulation->currentIndex() -
+                                                 ConfigurationShared::USE_GLOBAL_OFFSET);
+}
+
 void ConfigureGraphics::SetupPerGameUI() {
     if (Settings::IsConfiguringGlobal()) {
         ui->api->setEnabled(Settings::values.renderer_backend.UsingGlobal());
@@ -286,7 +314,7 @@ void ConfigureGraphics::SetupPerGameUI() {
         ui->aspect_ratio_combobox->setEnabled(Settings::values.aspect_ratio.UsingGlobal());
         ui->use_asynchronous_gpu_emulation->setEnabled(
             Settings::values.use_asynchronous_gpu_emulation.UsingGlobal());
-        ui->use_nvdec_emulation->setEnabled(Settings::values.use_nvdec_emulation.UsingGlobal());
+        ui->nvdec_emulation->setEnabled(Settings::values.nvdec_emulation.UsingGlobal());
         ui->accelerate_astc->setEnabled(Settings::values.accelerate_astc.UsingGlobal());
         ui->use_disk_shader_cache->setEnabled(Settings::values.use_disk_shader_cache.UsingGlobal());
         ui->bg_button->setEnabled(Settings::values.bg_red.UsingGlobal());
@@ -301,8 +329,6 @@ void ConfigureGraphics::SetupPerGameUI() {
 
     ConfigurationShared::SetColoredTristate(
         ui->use_disk_shader_cache, Settings::values.use_disk_shader_cache, use_disk_shader_cache);
-    ConfigurationShared::SetColoredTristate(
-        ui->use_nvdec_emulation, Settings::values.use_nvdec_emulation, use_nvdec_emulation);
     ConfigurationShared::SetColoredTristate(ui->accelerate_astc, Settings::values.accelerate_astc,
                                             accelerate_astc);
     ConfigurationShared::SetColoredTristate(ui->use_asynchronous_gpu_emulation,
@@ -316,4 +342,6 @@ void ConfigureGraphics::SetupPerGameUI() {
         static_cast<int>(Settings::values.fullscreen_mode.GetValue(true)));
     ConfigurationShared::InsertGlobalItem(
         ui->api, static_cast<int>(Settings::values.renderer_backend.GetValue(true)));
+    ConfigurationShared::InsertGlobalItem(
+        ui->nvdec_emulation, static_cast<int>(Settings::values.nvdec_emulation.GetValue(true)));
 }

src/yuzu/configuration/configure_graphics.h

@@ -43,6 +43,7 @@ private:
     void SetupPerGameUI();
 
     Settings::RendererBackend GetCurrentGraphicsBackend() const;
+    Settings::NvdecEmulation GetCurrentNvdecEmulation() const;
 
     std::unique_ptr<Ui::ConfigureGraphics> ui;
     QColor bg_color;

src/yuzu/configuration/configure_graphics.ui

@@ -156,7 +156,7 @@
         <item>
          <widget class="QCheckBox" name="use_disk_shader_cache">
           <property name="text">
-           <string>Use disk shader cache</string>
+           <string>Use disk pipeline cache</string>
           </property>
          </widget>
         </item>
@@ -167,13 +167,6 @@
           </property>
          </widget>
         </item>
-        <item>
-         <widget class="QCheckBox" name="use_nvdec_emulation">
-          <property name="text">
-           <string>Use NVDEC emulation</string>
-          </property>
-         </widget>
-        </item>
         <item>
           <widget class="QCheckBox" name="accelerate_astc">
             <property name="text">
@@ -181,6 +174,50 @@
             </property>
           </widget>
         </item>
+        <item>
+         <widget class="QWidget" name="nvdec_emulation_widget" native="true">
+          <layout class="QHBoxLayout" name="nvdec_emulation_layout">
+           <property name="leftMargin">
+            <number>0</number>
+           </property>
+           <property name="topMargin">
+            <number>0</number>
+           </property>
+           <property name="rightMargin">
+            <number>0</number>
+           </property>
+           <property name="bottomMargin">
+            <number>0</number>
+           </property>
+           <item>
+            <widget class="QLabel" name="nvdec_emulation_label">
+             <property name="text">
+              <string>NVDEC emulation:</string>
+             </property>
+            </widget>
+           </item>
+           <item>
+            <widget class="QComboBox" name="nvdec_emulation">
+             <item>
+              <property name="text">
+               <string>Disabled</string>
+              </property>
+             </item>
+             <item>
+              <property name="text">
+               <string>CPU Decoding</string>
+              </property>
+             </item>
+             <item>
+              <property name="text">
+               <string>GPU Decoding</string>
+              </property>
+             </item>
+            </widget>
+           </item>
+          </layout>
+         </widget>
+        </item>
         <item>
          <widget class="QWidget" name="fullscreen_mode_layout" native="true">
           <layout class="QHBoxLayout" name="horizontalLayout_1">

src/yuzu/configuration/configure_graphics_advanced.ui

@@ -82,7 +82,7 @@
            <string>Enables asynchronous shader compilation, which may reduce shader stutter. This feature is experimental.</string>
           </property>
           <property name="text">
-           <string>Use asynchronous shader building (hack)</string>
+           <string>Use asynchronous shader building (Hack)</string>
           </property>
          </widget>
         </item>
@@ -92,7 +92,7 @@
             <string>Enables Fast GPU Time. This option will force most games to run at their highest native resolution.</string>
           </property>
           <property name="text">
-           <string>Use Fast GPU Time (hack)</string>
+           <string>Use Fast GPU Time (Hack)</string>
           </property>
          </widget>
         </item>

src/yuzu/configuration/configure_input_advanced.cpp

@@ -126,6 +126,7 @@ void ConfigureInputAdvanced::ApplyConfiguration() {
     Settings::values.mouse_panning_sensitivity =
         static_cast<float>(ui->mouse_panning_sensitivity->value());
     Settings::values.touchscreen.enabled = ui->touchscreen_enabled->isChecked();
+    Settings::values.enable_raw_input = ui->enable_raw_input->isChecked();
 }
 
 void ConfigureInputAdvanced::LoadConfiguration() {
@@ -155,6 +156,7 @@ void ConfigureInputAdvanced::LoadConfiguration() {
     ui->mouse_panning->setChecked(Settings::values.mouse_panning.GetValue());
     ui->mouse_panning_sensitivity->setValue(Settings::values.mouse_panning_sensitivity.GetValue());
     ui->touchscreen_enabled->setChecked(Settings::values.touchscreen.enabled);
+    ui->enable_raw_input->setChecked(Settings::values.enable_raw_input.GetValue());
 
     UpdateUIEnabled();
 }

src/yuzu/configuration/configure_input_advanced.ui

@@ -2672,6 +2672,22 @@
                  </property>
                 </widget>
                </item>
+               <item row="9" column="0">
+                 <widget class="QCheckBox" name="enable_raw_input">
+                   <property name="toolTip">
+                     <string>Requires restarting yuzu</string>
+                   </property>
+                   <property name="minimumSize">
+                     <size>
+                       <width>0</width>
+                       <height>23</height>
+                     </size>
+                   </property>
+                   <property name="text">
+                     <string>Enable XInput 8 player support (disables web applet)</string>
+                   </property>
+                 </widget>
+               </item>
               </layout>
              </widget>
             </item>

src/yuzu/configuration/configure_input_player.cpp

@@ -124,6 +124,19 @@ QString ButtonToText(const Common::ParamPackage& param) {
         return GetKeyName(param.Get("code", 0));
     }
 
+    if (param.Get("engine", "") == "tas") {
+        if (param.Has("axis")) {
+            const QString axis_str = QString::fromStdString(param.Get("axis", ""));
+
+            return QObject::tr("TAS Axis %1").arg(axis_str);
+        }
+        if (param.Has("button")) {
+            const QString button_str = QString::number(int(std::log2(param.Get("button", 0))));
+            return QObject::tr("TAS Btn %1").arg(button_str);
+        }
+        return GetKeyName(param.Get("code", 0));
+    }
+
     if (param.Get("engine", "") == "cemuhookudp") {
         if (param.Has("pad_index")) {
             const QString motion_str = QString::fromStdString(param.Get("pad_index", ""));
@@ -187,7 +200,8 @@ QString AnalogToText(const Common::ParamPackage& param, const std::string& dir)
     const QString axis_y_str = QString::fromStdString(param.Get("axis_y", ""));
     const bool invert_x = param.Get("invert_x", "+") == "-";
     const bool invert_y = param.Get("invert_y", "+") == "-";
-    if (engine_str == "sdl" || engine_str == "gcpad" || engine_str == "mouse") {
+    if (engine_str == "sdl" || engine_str == "gcpad" || engine_str == "mouse" ||
+        engine_str == "tas") {
         if (dir == "modifier") {
             return QObject::tr("[unused]");
         }
@@ -926,9 +940,9 @@ void ConfigureInputPlayer::UpdateUI() {
 
         int slider_value;
         auto& param = analogs_param[analog_id];
-        const bool is_controller = param.Get("engine", "") == "sdl" ||
-                                   param.Get("engine", "") == "gcpad" ||
-                                   param.Get("engine", "") == "mouse";
+        const bool is_controller =
+            param.Get("engine", "") == "sdl" || param.Get("engine", "") == "gcpad" ||
+            param.Get("engine", "") == "mouse" || param.Get("engine", "") == "tas";
 
         if (is_controller) {
             if (!param.Has("deadzone")) {
@@ -1045,8 +1059,12 @@ int ConfigureInputPlayer::GetIndexFromControllerType(Settings::ControllerType ty
 void ConfigureInputPlayer::UpdateInputDevices() {
     input_devices = input_subsystem->GetInputDevices();
     ui->comboDevices->clear();
-    for (auto device : input_devices) {
-        ui->comboDevices->addItem(QString::fromStdString(device.Get("display", "Unknown")), {});
+    for (auto& device : input_devices) {
+        const std::string display = device.Get("display", "Unknown");
+        ui->comboDevices->addItem(QString::fromStdString(display), {});
+        if (display == "TAS") {
+            device.Set("pad", static_cast<u8>(player_index));
+        }
     }
 }
 

src/yuzu/configuration/configure_input_player_widget.cpp

@@ -175,7 +175,7 @@ void PlayerControlPreview::ResetInputs() {
 }
 
 void PlayerControlPreview::UpdateInput() {
-    if (!is_enabled && !mapping_active) {
+    if (!is_enabled && !mapping_active && !Settings::values.tas_enable) {
         return;
     }
     bool input_changed = false;
@@ -222,6 +222,19 @@ void PlayerControlPreview::UpdateInput() {
 
     if (input_changed) {
         update();
+        if (controller_callback.input != nullptr) {
+            ControllerInput input{
+                .axis_values = {std::pair<float, float>{
+                                    axis_values[Settings::NativeAnalog::LStick].value.x(),
+                                    axis_values[Settings::NativeAnalog::LStick].value.y()},
+                                std::pair<float, float>{
+                                    axis_values[Settings::NativeAnalog::RStick].value.x(),
+                                    axis_values[Settings::NativeAnalog::RStick].value.y()}},
+                .button_values = button_values,
+                .changed = true,
+            };
+            controller_callback.input(std::move(input));
+        }
     }
 
     if (mapping_active) {
@@ -229,6 +242,10 @@ void PlayerControlPreview::UpdateInput() {
     }
 }
 
+void PlayerControlPreview::SetCallBack(ControllerCallback callback_) {
+    controller_callback = std::move(callback_);
+}
+
 void PlayerControlPreview::paintEvent(QPaintEvent* event) {
     QFrame::paintEvent(event);
     QPainter p(this);

src/yuzu/configuration/configure_input_player_widget.h

@@ -9,6 +9,7 @@
 #include <QPointer>
 #include "common/settings.h"
 #include "core/frontend/input.h"
+#include "yuzu/debugger/controller.h"
 
 class QLabel;
 
@@ -33,6 +34,7 @@ public:
     void BeginMappingAnalog(std::size_t button_id);
     void EndMapping();
     void UpdateInput();
+    void SetCallBack(ControllerCallback callback_);
 
 protected:
     void paintEvent(QPaintEvent* event) override;
@@ -181,6 +183,7 @@ private:
     using StickArray =
         std::array<std::unique_ptr<Input::AnalogDevice>, Settings::NativeAnalog::NUM_STICKS_HID>;
 
+    ControllerCallback controller_callback;
     bool is_enabled{};
     bool mapping_active{};
     int blink_counter{};

src/yuzu/configuration/configure_tas.cpp

@@ -0,0 +1,84 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <QFileDialog>
+#include <QMessageBox>
+#include "common/fs/fs.h"
+#include "common/fs/path_util.h"
+#include "common/settings.h"
+#include "ui_configure_tas.h"
+#include "yuzu/configuration/configure_tas.h"
+#include "yuzu/uisettings.h"
+
+ConfigureTasDialog::ConfigureTasDialog(QWidget* parent)
+    : QDialog(parent), ui(std::make_unique<Ui::ConfigureTas>()) {
+
+    ui->setupUi(this);
+
+    setFocusPolicy(Qt::ClickFocus);
+    setWindowTitle(tr("TAS Configuration"));
+
+    connect(ui->tas_path_button, &QToolButton::pressed, this,
+            [this] { SetDirectory(DirectoryTarget::TAS, ui->tas_path_edit); });
+
+    LoadConfiguration();
+}
+
+ConfigureTasDialog::~ConfigureTasDialog() = default;
+
+void ConfigureTasDialog::LoadConfiguration() {
+    ui->tas_path_edit->setText(
+        QString::fromStdString(Common::FS::GetYuzuPathString(Common::FS::YuzuPath::TASDir)));
+    ui->tas_enable->setChecked(Settings::values.tas_enable.GetValue());
+    ui->tas_control_swap->setChecked(Settings::values.tas_swap_controllers.GetValue());
+    ui->tas_loop_script->setChecked(Settings::values.tas_loop.GetValue());
+    ui->tas_pause_on_load->setChecked(Settings::values.pause_tas_on_load.GetValue());
+}
+
+void ConfigureTasDialog::ApplyConfiguration() {
+    Common::FS::SetYuzuPath(Common::FS::YuzuPath::TASDir, ui->tas_path_edit->text().toStdString());
+    Settings::values.tas_enable.SetValue(ui->tas_enable->isChecked());
+    Settings::values.tas_swap_controllers.SetValue(ui->tas_control_swap->isChecked());
+    Settings::values.tas_loop.SetValue(ui->tas_loop_script->isChecked());
+    Settings::values.pause_tas_on_load.SetValue(ui->tas_pause_on_load->isChecked());
+}
+
+void ConfigureTasDialog::SetDirectory(DirectoryTarget target, QLineEdit* edit) {
+    QString caption;
+
+    switch (target) {
+    case DirectoryTarget::TAS:
+        caption = tr("Select TAS Load Directory...");
+        break;
+    }
+
+    QString str = QFileDialog::getExistingDirectory(this, caption, edit->text());
+
+    if (str.isEmpty()) {
+        return;
+    }
+
+    if (str.back() != QChar::fromLatin1('/')) {
+        str.append(QChar::fromLatin1('/'));
+    }
+
+    edit->setText(str);
+}
+
+void ConfigureTasDialog::changeEvent(QEvent* event) {
+    if (event->type() == QEvent::LanguageChange) {
+        RetranslateUI();
+    }
+
+    QDialog::changeEvent(event);
+}
+
+void ConfigureTasDialog::RetranslateUI() {
+    ui->retranslateUi(this);
+}
+
+void ConfigureTasDialog::HandleApplyButtonClicked() {
+    UISettings::values.configuration_applied = true;
+    ApplyConfiguration();
+}

src/yuzu/configuration/configure_tas.h

@@ -0,0 +1,38 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <QDialog>
+
+namespace Ui {
+class ConfigureTas;
+}
+
+class ConfigureTasDialog : public QDialog {
+    Q_OBJECT
+
+public:
+    explicit ConfigureTasDialog(QWidget* parent);
+    ~ConfigureTasDialog() override;
+
+    /// Save all button configurations to settings file
+    void ApplyConfiguration();
+
+private:
+    enum class DirectoryTarget {
+        TAS,
+    };
+
+    void LoadConfiguration();
+
+    void SetDirectory(DirectoryTarget target, QLineEdit* edit);
+
+    void changeEvent(QEvent* event) override;
+    void RetranslateUI();
+
+    void HandleApplyButtonClicked();
+
+    std::unique_ptr<Ui::ConfigureTas> ui;
+};

src/yuzu/configuration/configure_tas.ui

@@ -0,0 +1,153 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<ui version="4.0">
+  <class>ConfigureTas</class>
+  <widget class="QDialog" name="ConfigureTas">
+    <layout class="QVBoxLayout" name="verticalLayout_1">
+      <item>
+        <layout class="QHBoxLayout" name="horizontalLayout_1">
+          <item>
+            <widget class="QGroupBox" name="groupBox_1">
+              <property name="title">
+                <string>TAS</string>
+              </property>
+              <layout class="QGridLayout" name="gridLayout_1">
+                <item row="0" column="0" colspan="4">
+                  <widget class="QLabel" name="label_1">
+                    <property name="text">
+                      <string>Reads controller input from scripts in the same format as TAS-nx scripts.&lt;br/&gt;For a more detailed explanation please consult the FAQ on the yuzu website.</string>
+                    </property>
+                  </widget>
+                </item>
+                <item row="1" column="0" colspan="4">
+                  <widget class="QLabel" name="label_2">
+                    <property name="text">
+                      <string>To check which hotkeys control the playback/recording, please refer to the Hotkey settings (General -> Hotkeys).</string>
+                    </property>
+                    <property name="wordWrap">
+                      <bool>true</bool>
+                    </property>
+                  </widget>
+                </item>
+                <item row="2" column="0" colspan="4">
+                  <widget class="QLabel" name="label_3">
+                    <property name="text">
+                      <string>WARNING: This is an experimental feature.&lt;br/&gt;It will not play back scripts frame perfectly with the current, imperfect syncing method.</string>
+                    </property>
+                    <property name="wordWrap">
+                      <bool>true</bool>
+                    </property>
+                  </widget>
+                </item>
+              </layout>
+            </widget>
+          </item>
+        </layout>
+      </item>
+      <item>
+        <layout class="QHBoxLayout" name="horizontalLayout_2">
+          <item>
+            <widget class="QGroupBox" name="groupBox_2">
+              <property name="title">
+                <string>Settings</string>
+              </property>
+              <layout class="QGridLayout" name="gridLayout_2">
+                <item row="0" column="0" colspan="4">
+                  <widget class="QCheckBox" name="tas_enable">
+                    <property name="text">
+                      <string>Enable TAS features</string>
+                    </property>
+                  </widget>
+                </item>
+                <item row="1" column="0" colspan="4">
+                  <widget class="QCheckBox" name="tas_control_swap">
+                    <property name="text">
+                      <string>Automatic controller profile swapping</string>
+                    </property>
+                  </widget>
+                </item>
+                <item row="2" column="0" colspan="4">
+                  <widget class="QCheckBox" name="tas_loop_script">
+                    <property name="text">
+                      <string>Loop script</string>
+                    </property>
+                  </widget>
+                </item>
+                <item row="3" column="0" colspan="4">
+                  <widget class="QCheckBox" name="tas_pause_on_load">
+                    <property name="enabled">
+                      <bool>false</bool>
+                    </property>
+                    <property name="text">
+                      <string>Pause execution during loads</string>
+                    </property>
+                  </widget>
+                </item>
+              </layout>
+            </widget>
+          </item>
+        </layout>
+      </item>
+      <item>
+        <layout class="QHBoxLayout" name="horizontalLayout_3">
+          <item>
+            <widget class="QGroupBox" name="groupBox_3">
+              <property name="title">
+                <string>Script Directory</string>
+              </property>
+              <layout class="QGridLayout" name="gridLayout_3">
+                <item row="0" column="0">
+                  <widget class="QLabel" name="label_4">
+                    <property name="text">
+                      <string>Path</string>
+                    </property>
+                  </widget>
+                </item>
+                <item row="0" column="3">
+                  <widget class="QToolButton" name="tas_path_button">
+                    <property name="text">
+                      <string>...</string>
+                    </property>
+                  </widget>
+                </item>
+                <item row="0" column="2">
+                  <widget class="QLineEdit" name="tas_path_edit"/>
+                </item>
+              </layout>
+            </widget>
+          </item>
+        </layout>
+      </item>
+      <item>
+        <widget class="QDialogButtonBox" name="buttonBox">
+          <property name="sizePolicy">
+            <sizepolicy hsizetype="Preferred" vsizetype="Preferred">
+              <horstretch>0</horstretch>
+              <verstretch>0</verstretch>
+            </sizepolicy>
+          </property>
+          <property name="orientation">
+            <enum>Qt::Horizontal</enum>
+          </property>
+          <property name="standardButtons">
+            <set>QDialogButtonBox::Cancel|QDialogButtonBox::Ok</set>
+          </property>
+        </widget>
+      </item>
+    </layout>
+  </widget>
+  <resources/>
+  <connections>
+    <connection>
+      <sender>buttonBox</sender>
+      <signal>accepted()</signal>
+      <receiver>ConfigureTas</receiver>
+      <slot>accept()</slot>
+    </connection>
+    <connection>
+      <sender>buttonBox</sender>
+      <signal>rejected()</signal>
+      <receiver>ConfigureTas</receiver>
+      <slot>reject()</slot>
+    </connection>
+  </connections>
+</ui>

src/yuzu/configuration/configure_vibration.cpp

@@ -99,7 +99,7 @@ void ConfigureVibration::SetVibrationDevices(std::size_t player_index) {
         const auto guid = param.Get("guid", "");
         const auto port = param.Get("port", "");
 
-        if (engine.empty() || engine == "keyboard" || engine == "mouse") {
+        if (engine.empty() || engine == "keyboard" || engine == "mouse" || engine == "tas") {
             continue;
         }