Revert "k_page_group: synchronize"

video_core: Implement all vulkan topology

.ci/scripts/clang/docker.sh

@@ -11,6 +11,7 @@ ccache -s
 mkdir build || true && cd build
 cmake .. \
       -DCMAKE_BUILD_TYPE=Release \
+      -DCMAKE_CXX_FLAGS="-march=x86-64-v2" \
       -DCMAKE_CXX_COMPILER=/usr/lib/ccache/clang++ \
       -DCMAKE_C_COMPILER=/usr/lib/ccache/clang \
       -DCMAKE_INSTALL_PREFIX="/usr" \

.ci/scripts/linux/docker.sh

@@ -12,6 +12,7 @@ mkdir build || true && cd build
 cmake .. \
       -DBoost_USE_STATIC_LIBS=ON \
       -DCMAKE_BUILD_TYPE=Release \
+      -DCMAKE_CXX_FLAGS="-march=x86-64-v2" \
       -DCMAKE_CXX_COMPILER=/usr/lib/ccache/g++ \
       -DCMAKE_C_COMPILER=/usr/lib/ccache/gcc \
       -DCMAKE_INSTALL_PREFIX="/usr" \

.ci/templates/build-msvc.yml

@@ -9,7 +9,7 @@ parameters:
 steps:
 - script: choco install vulkan-sdk
   displayName: 'Install vulkan-sdk'
-- script: refreshenv && mkdir build && cd build && cmake -G "Visual Studio 17 2022" -A x64 -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DYUZU_TESTS=OFF -DUSE_DISCORD_PRESENCE=ON -DENABLE_QT_TRANSLATION=ON -DDISPLAY_VERSION=${{ parameters['version'] }} -DCMAKE_BUILD_TYPE=Release -DYUZU_CRASH_DUMPS=ON .. && cd ..
+- script: refreshenv && mkdir build && cd build && cmake -E env CXXFLAGS="/Gw /GA /Gr /Ob2" cmake -G "Visual Studio 17 2022" -A x64 -DCMAKE_INTERPROCEDURAL_OPTIMIZATION=ON -DCMAKE_POLICY_DEFAULT_CMP0069=NEW -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DYUZU_TESTS=OFF -DUSE_DISCORD_PRESENCE=ON -DENABLE_QT_TRANSLATION=ON -DDISPLAY_VERSION=${{ parameters['version'] }} -DCMAKE_BUILD_TYPE=Release -DYUZU_CRASH_DUMPS=ON .. && cd ..
   displayName: 'Configure CMake'
 - task: MSBuild@1
   displayName: 'Build'

CMakeLists.txt

@@ -208,7 +208,7 @@ find_package(libusb 1.0.24)
 find_package(lz4 REQUIRED)
 find_package(nlohmann_json 3.8 REQUIRED)
 find_package(Opus 1.3)
-find_package(Vulkan 1.3.213)
+find_package(Vulkan 1.3.238)
 find_package(ZLIB 1.2 REQUIRED)
 find_package(zstd 1.5 REQUIRED)
 

externals/find-modules/FindOpus.cmake

@@ -2,9 +2,7 @@
 # SPDX-License-Identifier: GPL-2.0-or-later
 
 find_package(PkgConfig QUIET)
-if (PKG_CONFIG_FOUND)
-    pkg_search_module(OPUS QUIET IMPORTED_TARGET opus)
-endif()
+pkg_search_module(OPUS QUIET IMPORTED_TARGET opus)
 
 include(FindPackageHandleStandardArgs)
 find_package_handle_standard_args(Opus

externals/find-modules/Findenet.cmake

@@ -3,9 +3,7 @@
 # SPDX-License-Identifier: GPL-3.0-or-later
 
 find_package(PkgConfig QUIET)
-if (PKG_CONFIG_FOUND)
-    pkg_search_module(ENET QUIET IMPORTED_TARGET libenet)
-endif()
+pkg_search_module(ENET QUIET IMPORTED_TARGET libenet)
 
 include(FindPackageHandleStandardArgs)
 find_package_handle_standard_args(enet

externals/find-modules/Findhttplib.cmake

@@ -9,9 +9,7 @@ if (httplib_FOUND)
     find_package_handle_standard_args(httplib CONFIG_MODE)
 else()
     find_package(PkgConfig QUIET)
-    if (PKG_CONFIG_FOUND)
-        pkg_search_module(HTTPLIB QUIET IMPORTED_TARGET cpp-httplib)
-    endif()
+    pkg_search_module(HTTPLIB QUIET IMPORTED_TARGET cpp-httplib)
     find_package_handle_standard_args(httplib
         REQUIRED_VARS HTTPLIB_INCLUDEDIR
         VERSION_VAR HTTPLIB_VERSION

externals/find-modules/Findinih.cmake

@@ -3,9 +3,7 @@
 # SPDX-License-Identifier: GPL-3.0-or-later
 
 find_package(PkgConfig QUIET)
-if (PKG_CONFIG_FOUND)
-    pkg_search_module(INIREADER QUIET IMPORTED_TARGET INIReader)
-endif()
+pkg_search_module(INIREADER QUIET IMPORTED_TARGET INIReader)
 
 include(FindPackageHandleStandardArgs)
 find_package_handle_standard_args(inih

externals/find-modules/Findlibusb.cmake

@@ -3,9 +3,7 @@
 # SPDX-License-Identifier: GPL-3.0-or-later
 
 find_package(PkgConfig QUIET)
-if (PKG_CONFIG_FOUND)
-    pkg_search_module(LIBUSB QUIET IMPORTED_TARGET libusb-1.0)
-endif()
+pkg_search_module(LIBUSB QUIET IMPORTED_TARGET libusb-1.0)
 
 include(FindPackageHandleStandardArgs)
 find_package_handle_standard_args(libusb

externals/find-modules/Findlz4.cmake

@@ -8,9 +8,7 @@ if (lz4_FOUND)
     find_package_handle_standard_args(lz4 CONFIG_MODE)
 else()
     find_package(PkgConfig QUIET)
-    if (PKG_CONFIG_FOUND)
-        pkg_search_module(LZ4 QUIET IMPORTED_TARGET liblz4)
-    endif()
+    pkg_search_module(LZ4 QUIET IMPORTED_TARGET liblz4)
     find_package_handle_standard_args(lz4
         REQUIRED_VARS LZ4_LINK_LIBRARIES
         VERSION_VAR LZ4_VERSION

externals/find-modules/Findzstd.cmake

@@ -8,9 +8,7 @@ if (zstd_FOUND)
     find_package_handle_standard_args(zstd CONFIG_MODE)
 else()
     find_package(PkgConfig QUIET)
-    if (PKG_CONFIG_FOUND)
-        pkg_search_module(ZSTD QUIET IMPORTED_TARGET libzstd)
-    endif()
+    pkg_search_module(ZSTD QUIET IMPORTED_TARGET libzstd)
     find_package_handle_standard_args(zstd
         REQUIRED_VARS ZSTD_LINK_LIBRARIES
         VERSION_VAR ZSTD_VERSION

src/audio_core/device/audio_buffers.h

@@ -91,9 +91,10 @@ public:
      * @param core_timing - The CoreTiming instance
      * @param session     - The device session
      *
-     * @return Is the buffer was released.
+     * @return If any buffer was released.
      */
-    bool ReleaseBuffers(const Core::Timing::CoreTiming& core_timing, const DeviceSession& session) {
+    bool ReleaseBuffers(const Core::Timing::CoreTiming& core_timing, const DeviceSession& session,
+                        bool force) {
         std::scoped_lock l{lock};
         bool buffer_released{false};
         while (registered_count > 0) {
@@ -103,7 +104,8 @@ public:
             }
 
             // Check with the backend if this buffer can be released yet.
-            if (!session.IsBufferConsumed(buffers[index])) {
+            // If we're shutting down, we don't care if it's been played or not.
+            if (!force && !session.IsBufferConsumed(buffers[index])) {
                 break;
             }
 

src/audio_core/device/device_session.cpp

@@ -73,6 +73,12 @@ void DeviceSession::Stop() {
     }
 }
 
+void DeviceSession::ClearBuffers() {
+    if (stream) {
+        stream->ClearQueue();
+    }
+}
+
 void DeviceSession::AppendBuffers(std::span<const AudioBuffer> buffers) const {
     for (const auto& buffer : buffers) {
         Sink::SinkBuffer new_buffer{

src/audio_core/device/device_session.h

@@ -90,6 +90,11 @@ public:
      */
     void Stop();
 
+    /**
+     * Clear out the underlying audio buffers in the backend stream.
+     */
+    void ClearBuffers();
+
     /**
      * Set this device session's volume.
      *

src/audio_core/in/audio_in_system.cpp

@@ -23,7 +23,6 @@ System::~System() {
 void System::Finalize() {
     Stop();
     session->Finalize();
-    buffer_event->Signal();
 }
 
 void System::StartSession() {
@@ -102,6 +101,10 @@ Result System::Stop() {
     if (state == State::Started) {
         session->Stop();
         session->SetVolume(0.0f);
+        session->ClearBuffers();
+        if (buffers.ReleaseBuffers(system.CoreTiming(), *session, true)) {
+            buffer_event->Signal();
+        }
         state = State::Stopped;
     }
 
@@ -138,7 +141,7 @@ void System::RegisterBuffers() {
 }
 
 void System::ReleaseBuffers() {
-    bool signal{buffers.ReleaseBuffers(system.CoreTiming(), *session)};
+    bool signal{buffers.ReleaseBuffers(system.CoreTiming(), *session, false)};
 
     if (signal) {
         // Signal if any buffer was released, or if none are registered, we need more.

src/audio_core/out/audio_out_system.cpp

@@ -24,7 +24,6 @@ System::~System() {
 void System::Finalize() {
     Stop();
     session->Finalize();
-    buffer_event->Signal();
 }
 
 std::string_view System::GetDefaultOutputDeviceName() const {
@@ -102,6 +101,10 @@ Result System::Stop() {
     if (state == State::Started) {
         session->Stop();
         session->SetVolume(0.0f);
+        session->ClearBuffers();
+        if (buffers.ReleaseBuffers(system.CoreTiming(), *session, true)) {
+            buffer_event->Signal();
+        }
         state = State::Stopped;
     }
 
@@ -138,7 +141,7 @@ void System::RegisterBuffers() {
 }
 
 void System::ReleaseBuffers() {
-    bool signal{buffers.ReleaseBuffers(system.CoreTiming(), *session)};
+    bool signal{buffers.ReleaseBuffers(system.CoreTiming(), *session, false)};
     if (signal) {
         // Signal if any buffer was released, or if none are registered, we need more.
         buffer_event->Signal();

src/common/CMakeLists.txt

@@ -78,6 +78,7 @@ add_library(common STATIC
     logging/types.h
     lz4_compression.cpp
     lz4_compression.h
+    make_unique_for_overwrite.h
     math_util.h
     memory_detect.cpp
     memory_detect.h
@@ -101,6 +102,7 @@ add_library(common STATIC
     ${CMAKE_CURRENT_BINARY_DIR}/scm_rev.cpp
     scm_rev.h
     scope_exit.h
+    scratch_buffer.h
     settings.cpp
     settings.h
     settings_input.cpp

src/common/assert.h

@@ -69,7 +69,7 @@ void assert_fail_impl();
 #define ASSERT_OR_EXECUTE(_a_, _b_)                                                                \
     do {                                                                                           \
         ASSERT(_a_);                                                                               \
-        if (!(_a_)) {                                                                              \
+        if (!(_a_)) [[unlikely]] {                                                                 \
             _b_                                                                                    \
         }                                                                                          \
     } while (0)
@@ -78,7 +78,7 @@ void assert_fail_impl();
 #define ASSERT_OR_EXECUTE_MSG(_a_, _b_, ...)                                                       \
     do {                                                                                           \
         ASSERT_MSG(_a_, __VA_ARGS__);                                                              \
-        if (!(_a_)) {                                                                              \
+        if (!(_a_)) [[unlikely]] {                                                                 \
             _b_                                                                                    \
         }                                                                                          \
     } while (0)

src/common/make_unique_for_overwrite.h

@@ -0,0 +1,25 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <memory>
+#include <type_traits>
+
+namespace Common {
+
+template <class T>
+requires(!std::is_array_v<T>) std::unique_ptr<T> make_unique_for_overwrite() {
+    return std::unique_ptr<T>(new T);
+}
+
+template <class T>
+requires std::is_unbounded_array_v<T> std::unique_ptr<T> make_unique_for_overwrite(std::size_t n) {
+    return std::unique_ptr<T>(new std::remove_extent_t<T>[n]);
+}
+
+template <class T, class... Args>
+requires std::is_bounded_array_v<T>
+void make_unique_for_overwrite(Args&&...) = delete;
+
+} // namespace Common

src/common/scratch_buffer.h

@@ -0,0 +1,95 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "common/make_unique_for_overwrite.h"
+
+namespace Common {
+
+/**
+ * ScratchBuffer class
+ * This class creates a default initialized heap allocated buffer for cases such as intermediate
+ * buffers being copied into entirely, where value initializing members during allocation or resize
+ * is redundant.
+ */
+template <typename T>
+class ScratchBuffer {
+public:
+    ScratchBuffer() = default;
+
+    explicit ScratchBuffer(size_t initial_capacity)
+        : last_requested_size{initial_capacity}, buffer_capacity{initial_capacity},
+          buffer{Common::make_unique_for_overwrite<T[]>(initial_capacity)} {}
+
+    ~ScratchBuffer() = default;
+
+    /// This will only grow the buffer's capacity if size is greater than the current capacity.
+    /// The previously held data will remain intact.
+    void resize(size_t size) {
+        if (size > buffer_capacity) {
+            auto new_buffer = Common::make_unique_for_overwrite<T[]>(size);
+            std::move(buffer.get(), buffer.get() + buffer_capacity, new_buffer.get());
+            buffer = std::move(new_buffer);
+            buffer_capacity = size;
+        }
+        last_requested_size = size;
+    }
+
+    /// This will only grow the buffer's capacity if size is greater than the current capacity.
+    /// The previously held data will be destroyed if a reallocation occurs.
+    void resize_destructive(size_t size) {
+        if (size > buffer_capacity) {
+            buffer_capacity = size;
+            buffer = Common::make_unique_for_overwrite<T[]>(buffer_capacity);
+        }
+        last_requested_size = size;
+    }
+
+    [[nodiscard]] T* data() noexcept {
+        return buffer.get();
+    }
+
+    [[nodiscard]] const T* data() const noexcept {
+        return buffer.get();
+    }
+
+    [[nodiscard]] T* begin() noexcept {
+        return data();
+    }
+
+    [[nodiscard]] const T* begin() const noexcept {
+        return data();
+    }
+
+    [[nodiscard]] T* end() noexcept {
+        return data() + last_requested_size;
+    }
+
+    [[nodiscard]] const T* end() const noexcept {
+        return data() + last_requested_size;
+    }
+
+    [[nodiscard]] T& operator[](size_t i) {
+        return buffer[i];
+    }
+
+    [[nodiscard]] const T& operator[](size_t i) const {
+        return buffer[i];
+    }
+
+    [[nodiscard]] size_t size() const noexcept {
+        return last_requested_size;
+    }
+
+    [[nodiscard]] size_t capacity() const noexcept {
+        return buffer_capacity;
+    }
+
+private:
+    size_t last_requested_size{};
+    size_t buffer_capacity{};
+    std::unique_ptr<T[]> buffer{};
+};
+
+} // namespace Common

src/common/settings.cpp

@@ -40,6 +40,7 @@ void LogSettings() {
     LOG_INFO(Config, "yuzu Configuration:");
     log_setting("Controls_UseDockedMode", values.use_docked_mode.GetValue());
     log_setting("System_RngSeed", values.rng_seed.GetValue().value_or(0));
+    log_setting("System_DeviceName", values.device_name.GetValue());
     log_setting("System_CurrentUser", values.current_user.GetValue());
     log_setting("System_LanguageIndex", values.language_index.GetValue());
     log_setting("System_RegionIndex", values.region_index.GetValue());

src/common/settings.h

@@ -458,6 +458,7 @@ struct Values {
 
     // System
     SwitchableSetting<std::optional<u32>> rng_seed{std::optional<u32>(), "rng_seed"};
+    Setting<std::string> device_name{"Yuzu", "device_name"};
     // Measured in seconds since epoch
     std::optional<s64> custom_rtc;
     // Set on game boot, reset on stop. Seconds difference between current time and `custom_rtc`

src/common/thread.h

@@ -11,6 +11,7 @@
 #include <mutex>
 #include <thread>
 #include "common/common_types.h"
+#include "common/polyfill_thread.h"
 
 namespace Common {
 
@@ -69,7 +70,7 @@ public:
     explicit Barrier(std::size_t count_) : count(count_) {}
 
     /// Blocks until all "count" threads have called Sync()
-    void Sync() {
+    bool Sync(std::stop_token token = {}) {
         std::unique_lock lk{mutex};
         const std::size_t current_generation = generation;
 
@@ -77,14 +78,16 @@ public:
             generation++;
             waiting = 0;
             condvar.notify_all();
+            return true;
         } else {
-            condvar.wait(lk,
-                         [this, current_generation] { return current_generation != generation; });
+            CondvarWait(condvar, lk, token,
+                        [this, current_generation] { return current_generation != generation; });
+            return !token.stop_requested();
         }
     }
 
 private:
-    std::condition_variable condvar;
+    std::condition_variable_any condvar;
     std::mutex mutex;
     std::size_t count;
     std::size_t waiting = 0;

src/core/CMakeLists.txt

@@ -201,6 +201,9 @@ add_library(core STATIC
     hle/kernel/k_event_info.h
     hle/kernel/k_handle_table.cpp
     hle/kernel/k_handle_table.h
+    hle/kernel/k_hardware_timer_base.h
+    hle/kernel/k_hardware_timer.cpp
+    hle/kernel/k_hardware_timer.h
     hle/kernel/k_interrupt_manager.cpp
     hle/kernel/k_interrupt_manager.h
     hle/kernel/k_light_condition_variable.cpp
@@ -268,6 +271,7 @@ add_library(core STATIC
     hle/kernel/k_thread_local_page.h
     hle/kernel/k_thread_queue.cpp
     hle/kernel/k_thread_queue.h
+    hle/kernel/k_timer_task.h
     hle/kernel/k_trace.h
     hle/kernel/k_transfer_memory.cpp
     hle/kernel/k_transfer_memory.h
@@ -290,8 +294,6 @@ add_library(core STATIC
     hle/kernel/svc_common.h
     hle/kernel/svc_types.h
     hle/kernel/svc_wrap.h
-    hle/kernel/time_manager.cpp
-    hle/kernel/time_manager.h
     hle/result.h
     hle/service/acc/acc.cpp
     hle/service/acc/acc.h

src/core/core.cpp

@@ -183,26 +183,20 @@ struct System::Impl {
         Initialize(system);
     }
 
-    SystemResultStatus Run() {
+    void Run() {
         std::unique_lock<std::mutex> lk(suspend_guard);
-        status = SystemResultStatus::Success;
 
         kernel.Suspend(false);
         core_timing.SyncPause(false);
         is_paused.store(false, std::memory_order_relaxed);
-
-        return status;
     }
 
-    SystemResultStatus Pause() {
+    void Pause() {
         std::unique_lock<std::mutex> lk(suspend_guard);
-        status = SystemResultStatus::Success;
 
         core_timing.SyncPause(true);
         kernel.Suspend(true);
         is_paused.store(true, std::memory_order_relaxed);
-
-        return status;
     }
 
     bool IsPaused() const {
@@ -389,7 +383,9 @@ struct System::Impl {
         kernel.ShutdownCores();
         cpu_manager.Shutdown();
         debugger.reset();
-        services->KillNVNFlinger();
+        if (services) {
+            services->KillNVNFlinger();
+        }
         kernel.CloseServices();
         services.reset();
         service_manager.reset();
@@ -551,12 +547,12 @@ void System::Initialize() {
     impl->Initialize(*this);
 }
 
-SystemResultStatus System::Run() {
-    return impl->Run();
+void System::Run() {
+    impl->Run();
 }
 
-SystemResultStatus System::Pause() {
-    return impl->Pause();
+void System::Pause() {
+    impl->Pause();
 }
 
 bool System::IsPaused() const {

src/core/core.h

@@ -152,13 +152,13 @@ public:
      * Run the OS and Application
      * This function will start emulation and run the relevant devices
      */
-    [[nodiscard]] SystemResultStatus Run();
+    void Run();
 
     /**
      * Pause the OS and Application
      * This function will pause emulation and stop the relevant devices
      */
-    [[nodiscard]] SystemResultStatus Pause();
+    void Pause();
 
     /// Check if the core is currently paused.
     [[nodiscard]] bool IsPaused() const;

src/core/cpu_manager.cpp

@@ -20,23 +20,20 @@ namespace Core {
 CpuManager::CpuManager(System& system_) : system{system_} {}
 CpuManager::~CpuManager() = default;
 
-void CpuManager::ThreadStart(std::stop_token stop_token, CpuManager& cpu_manager,
-                             std::size_t core) {
-    cpu_manager.RunThread(core);
-}
-
 void CpuManager::Initialize() {
     num_cores = is_multicore ? Core::Hardware::NUM_CPU_CORES : 1;
     gpu_barrier = std::make_unique<Common::Barrier>(num_cores + 1);
 
     for (std::size_t core = 0; core < num_cores; core++) {
-        core_data[core].host_thread = std::jthread(ThreadStart, std::ref(*this), core);
+        core_data[core].host_thread =
+            std::jthread([this, core](std::stop_token token) { RunThread(token, core); });
     }
 }
 
 void CpuManager::Shutdown() {
     for (std::size_t core = 0; core < num_cores; core++) {
         if (core_data[core].host_thread.joinable()) {
+            core_data[core].host_thread.request_stop();
             core_data[core].host_thread.join();
         }
     }
@@ -184,7 +181,7 @@ void CpuManager::ShutdownThread() {
     UNREACHABLE();
 }
 
-void CpuManager::RunThread(std::size_t core) {
+void CpuManager::RunThread(std::stop_token token, std::size_t core) {
     /// Initialization
     system.RegisterCoreThread(core);
     std::string name;
@@ -206,7 +203,9 @@ void CpuManager::RunThread(std::size_t core) {
     });
 
     // Running
-    gpu_barrier->Sync();
+    if (!gpu_barrier->Sync(token)) {
+        return;
+    }
 
     if (!is_async_gpu && !is_multicore) {
         system.GPU().ObtainContext();

src/core/cpu_manager.h

@@ -81,12 +81,10 @@ private:
     void SingleCoreRunGuestThread();
     void SingleCoreRunIdleThread();
 
-    static void ThreadStart(std::stop_token stop_token, CpuManager& cpu_manager, std::size_t core);
-
     void GuestActivate();
     void HandleInterrupt();
     void ShutdownThread();
-    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;

src/core/frontend/emu_window.h

@@ -131,6 +131,10 @@ public:
         return active_config;
     }
 
+    bool StrictContextRequired() const {
+        return strict_context_required;
+    }
+
     /**
      * Requests the internal configuration to be replaced by the specified argument at some point in
      * the future.
@@ -207,6 +211,8 @@ protected:
 
     WindowSystemInfo window_info;
 
+    bool strict_context_required = false;
+
 private:
     /**
      * Handler called when the minimal client area was requested to be changed via SetConfig.

src/core/hid/emulated_controller.cpp

@@ -145,6 +145,7 @@ void EmulatedController::LoadDevices() {
     output_params[3].Set("output", true);
 
     LoadTASParams();
+    LoadVirtualGamepadParams();
 
     std::ranges::transform(button_params, button_devices.begin(), Common::Input::CreateInputDevice);
     std::ranges::transform(stick_params, stick_devices.begin(), Common::Input::CreateInputDevice);
@@ -163,6 +164,12 @@ void EmulatedController::LoadDevices() {
                            Common::Input::CreateInputDevice);
     std::ranges::transform(tas_stick_params, tas_stick_devices.begin(),
                            Common::Input::CreateInputDevice);
+
+    // Initialize virtual gamepad devices
+    std::ranges::transform(virtual_button_params, virtual_button_devices.begin(),
+                           Common::Input::CreateInputDevice);
+    std::ranges::transform(virtual_stick_params, virtual_stick_devices.begin(),
+                           Common::Input::CreateInputDevice);
 }
 
 void EmulatedController::LoadTASParams() {
@@ -203,6 +210,53 @@ void EmulatedController::LoadTASParams() {
     tas_stick_params[Settings::NativeAnalog::LStick].Set("axis_y", 1);
     tas_stick_params[Settings::NativeAnalog::RStick].Set("axis_x", 2);
     tas_stick_params[Settings::NativeAnalog::RStick].Set("axis_y", 3);
+
+    // set to optimal stick to avoid sanitizing the stick and tweaking the coordinates
+    // making sure they play back in the game as originally written down in the script file
+    tas_stick_params[Settings::NativeAnalog::LStick].Set("deadzone", 0.0f);
+    tas_stick_params[Settings::NativeAnalog::LStick].Set("range", 1.0f);
+    tas_stick_params[Settings::NativeAnalog::RStick].Set("deadzone", 0.0f);
+    tas_stick_params[Settings::NativeAnalog::RStick].Set("range", 1.0f);
+}
+
+void EmulatedController::LoadVirtualGamepadParams() {
+    const auto player_index = NpadIdTypeToIndex(npad_id_type);
+    Common::ParamPackage common_params{};
+    common_params.Set("engine", "virtual_gamepad");
+    common_params.Set("port", static_cast<int>(player_index));
+    for (auto& param : virtual_button_params) {
+        param = common_params;
+    }
+    for (auto& param : virtual_stick_params) {
+        param = common_params;
+    }
+
+    // TODO(german77): Replace this with an input profile or something better
+    virtual_button_params[Settings::NativeButton::A].Set("button", 0);
+    virtual_button_params[Settings::NativeButton::B].Set("button", 1);
+    virtual_button_params[Settings::NativeButton::X].Set("button", 2);
+    virtual_button_params[Settings::NativeButton::Y].Set("button", 3);
+    virtual_button_params[Settings::NativeButton::LStick].Set("button", 4);
+    virtual_button_params[Settings::NativeButton::RStick].Set("button", 5);
+    virtual_button_params[Settings::NativeButton::L].Set("button", 6);
+    virtual_button_params[Settings::NativeButton::R].Set("button", 7);
+    virtual_button_params[Settings::NativeButton::ZL].Set("button", 8);
+    virtual_button_params[Settings::NativeButton::ZR].Set("button", 9);
+    virtual_button_params[Settings::NativeButton::Plus].Set("button", 10);
+    virtual_button_params[Settings::NativeButton::Minus].Set("button", 11);
+    virtual_button_params[Settings::NativeButton::DLeft].Set("button", 12);
+    virtual_button_params[Settings::NativeButton::DUp].Set("button", 13);
+    virtual_button_params[Settings::NativeButton::DRight].Set("button", 14);
+    virtual_button_params[Settings::NativeButton::DDown].Set("button", 15);
+    virtual_button_params[Settings::NativeButton::SL].Set("button", 16);
+    virtual_button_params[Settings::NativeButton::SR].Set("button", 17);
+    virtual_button_params[Settings::NativeButton::Home].Set("button", 18);
+    virtual_button_params[Settings::NativeButton::Screenshot].Set("button", 19);
+
+    virtual_stick_params[Settings::NativeAnalog::LStick].Set("axis_x", 0);
+    virtual_stick_params[Settings::NativeAnalog::LStick].Set("axis_y", 1);
+    virtual_stick_params[Settings::NativeAnalog::RStick].Set("axis_x", 2);
+    virtual_stick_params[Settings::NativeAnalog::RStick].Set("axis_y", 3);
 }
 
 void EmulatedController::ReloadInput() {
@@ -322,6 +376,35 @@ void EmulatedController::ReloadInput() {
                 },
         });
     }
+
+    // Use a common UUID for Virtual Gamepad
+    static constexpr Common::UUID VIRTUAL_UUID = Common::UUID{
+        {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x7, 0xFF, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}};
+
+    // Register virtual devices. No need to force update
+    for (std::size_t index = 0; index < virtual_button_devices.size(); ++index) {
+        if (!virtual_button_devices[index]) {
+            continue;
+        }
+        virtual_button_devices[index]->SetCallback({
+            .on_change =
+                [this, index](const Common::Input::CallbackStatus& callback) {
+                    SetButton(callback, index, VIRTUAL_UUID);
+                },
+        });
+    }
+
+    for (std::size_t index = 0; index < virtual_stick_devices.size(); ++index) {
+        if (!virtual_stick_devices[index]) {
+            continue;
+        }
+        virtual_stick_devices[index]->SetCallback({
+            .on_change =
+                [this, index](const Common::Input::CallbackStatus& callback) {
+                    SetStick(callback, index, VIRTUAL_UUID);
+                },
+        });
+    }
 }
 
 void EmulatedController::UnloadInput() {
@@ -349,6 +432,12 @@ void EmulatedController::UnloadInput() {
     for (auto& stick : tas_stick_devices) {
         stick.reset();
     }
+    for (auto& button : virtual_button_devices) {
+        button.reset();
+    }
+    for (auto& stick : virtual_stick_devices) {
+        stick.reset();
+    }
     camera_devices.reset();
     nfc_devices.reset();
 }

src/core/hid/emulated_controller.h

@@ -385,6 +385,9 @@ private:
     /// Set the params for TAS devices
     void LoadTASParams();
 
+    /// Set the params for virtual pad devices
+    void LoadVirtualGamepadParams();
+
     /**
      * @param use_temporary_value If true tmp_npad_type will be used
      * @return true if the controller style is fullkey
@@ -500,6 +503,12 @@ private:
     ButtonDevices tas_button_devices;
     StickDevices tas_stick_devices;
 
+    // Virtual gamepad related variables
+    ButtonParams virtual_button_params;
+    StickParams virtual_stick_params;
+    ButtonDevices virtual_button_devices;
+    StickDevices virtual_stick_devices;
+
     mutable std::mutex mutex;
     mutable std::mutex callback_mutex;
     std::unordered_map<int, ControllerUpdateCallback> callback_list;

src/core/hle/kernel/hle_ipc.cpp

@@ -167,6 +167,9 @@ void HLERequestContext::ParseCommandBuffer(const KHandleTable& handle_table, u32
         }
         if (incoming) {
             // Populate the object lists with the data in the IPC request.
+            incoming_copy_handles.reserve(handle_descriptor_header->num_handles_to_copy);
+            incoming_move_handles.reserve(handle_descriptor_header->num_handles_to_move);
+
             for (u32 handle = 0; handle < handle_descriptor_header->num_handles_to_copy; ++handle) {
                 incoming_copy_handles.push_back(rp.Pop<Handle>());
             }
@@ -181,6 +184,11 @@ void HLERequestContext::ParseCommandBuffer(const KHandleTable& handle_table, u32
         }
     }
 
+    buffer_x_desciptors.reserve(command_header->num_buf_x_descriptors);
+    buffer_a_desciptors.reserve(command_header->num_buf_a_descriptors);
+    buffer_b_desciptors.reserve(command_header->num_buf_b_descriptors);
+    buffer_w_desciptors.reserve(command_header->num_buf_w_descriptors);
+
     for (u32 i = 0; i < command_header->num_buf_x_descriptors; ++i) {
         buffer_x_desciptors.push_back(rp.PopRaw<IPC::BufferDescriptorX>());
     }
@@ -318,25 +326,23 @@ Result HLERequestContext::WriteToOutgoingCommandBuffer(KThread& requesting_threa
 }
 
 std::vector<u8> HLERequestContext::ReadBuffer(std::size_t buffer_index) const {
-    std::vector<u8> buffer{};
     const bool is_buffer_a{BufferDescriptorA().size() > buffer_index &&
                            BufferDescriptorA()[buffer_index].Size()};
-
     if (is_buffer_a) {
         ASSERT_OR_EXECUTE_MSG(
-            BufferDescriptorA().size() > buffer_index, { return buffer; },
+            BufferDescriptorA().size() > buffer_index, { return {}; },
             "BufferDescriptorA invalid buffer_index {}", buffer_index);
-        buffer.resize(BufferDescriptorA()[buffer_index].Size());
+        std::vector<u8> buffer(BufferDescriptorA()[buffer_index].Size());
         memory.ReadBlock(BufferDescriptorA()[buffer_index].Address(), buffer.data(), buffer.size());
+        return buffer;
     } else {
         ASSERT_OR_EXECUTE_MSG(
-            BufferDescriptorX().size() > buffer_index, { return buffer; },
+            BufferDescriptorX().size() > buffer_index, { return {}; },
             "BufferDescriptorX invalid buffer_index {}", buffer_index);
-        buffer.resize(BufferDescriptorX()[buffer_index].Size());
+        std::vector<u8> buffer(BufferDescriptorX()[buffer_index].Size());
         memory.ReadBlock(BufferDescriptorX()[buffer_index].Address(), buffer.data(), buffer.size());
+        return buffer;
     }
-
-    return buffer;
 }
 
 std::size_t HLERequestContext::WriteBuffer(const void* buffer, std::size_t size,

src/core/hle/kernel/k_address_arbiter.cpp

@@ -10,7 +10,6 @@
 #include "core/hle/kernel/k_thread_queue.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/svc_results.h"
-#include "core/hle/kernel/time_manager.h"
 #include "core/memory.h"
 
 namespace Kernel {

src/core/hle/kernel/k_hardware_timer.cpp

@@ -0,0 +1,74 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "core/core.h"
+#include "core/core_timing.h"
+#include "core/hle/kernel/k_hardware_timer.h"
+#include "core/hle/kernel/k_scheduler.h"
+
+namespace Kernel {
+
+void KHardwareTimer::Initialize() {
+    // Create the timing callback to register with CoreTiming.
+    m_event_type = Core::Timing::CreateEvent(
+        "KHardwareTimer::Callback", [](std::uintptr_t timer_handle, s64, std::chrono::nanoseconds) {
+            reinterpret_cast<KHardwareTimer*>(timer_handle)->DoTask();
+            return std::nullopt;
+        });
+}
+
+void KHardwareTimer::Finalize() {
+    this->DisableInterrupt();
+    m_event_type.reset();
+}
+
+void KHardwareTimer::DoTask() {
+    // Handle the interrupt.
+    {
+        KScopedSchedulerLock slk{m_kernel};
+        KScopedSpinLock lk(this->GetLock());
+
+        //! Ignore this event if needed.
+        if (!this->GetInterruptEnabled()) {
+            return;
+        }
+
+        // Disable the timer interrupt while we handle this.
+        this->DisableInterrupt();
+
+        if (const s64 next_time = this->DoInterruptTaskImpl(GetTick());
+            0 < next_time && next_time <= m_wakeup_time) {
+            // We have a next time, so we should set the time to interrupt and turn the interrupt
+            // on.
+            this->EnableInterrupt(next_time);
+        }
+    }
+
+    // Clear the timer interrupt.
+    // Kernel::GetInterruptManager().ClearInterrupt(KInterruptName_NonSecurePhysicalTimer,
+    //                                              GetCurrentCoreId());
+}
+
+void KHardwareTimer::EnableInterrupt(s64 wakeup_time) {
+    this->DisableInterrupt();
+
+    m_wakeup_time = wakeup_time;
+    m_kernel.System().CoreTiming().ScheduleEvent(std::chrono::nanoseconds{m_wakeup_time},
+                                                 m_event_type, reinterpret_cast<uintptr_t>(this),
+                                                 true);
+}
+
+void KHardwareTimer::DisableInterrupt() {
+    m_kernel.System().CoreTiming().UnscheduleEvent(m_event_type, reinterpret_cast<uintptr_t>(this));
+    m_wakeup_time = std::numeric_limits<s64>::max();
+}
+
+s64 KHardwareTimer::GetTick() const {
+    return m_kernel.System().CoreTiming().GetGlobalTimeNs().count();
+}
+
+bool KHardwareTimer::GetInterruptEnabled() {
+    return m_wakeup_time != std::numeric_limits<s64>::max();
+}
+
+} // namespace Kernel

src/core/hle/kernel/k_hardware_timer.h

@@ -0,0 +1,54 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "core/hle/kernel/k_hardware_timer_base.h"
+
+namespace Core::Timing {
+struct EventType;
+} // namespace Core::Timing
+
+namespace Kernel {
+
+class KHardwareTimer : /* public KInterruptTask, */ public KHardwareTimerBase {
+public:
+    explicit KHardwareTimer(KernelCore& kernel) : KHardwareTimerBase{kernel} {}
+
+    // Public API.
+    void Initialize();
+    void Finalize();
+
+    s64 GetCount() const {
+        return GetTick();
+    }
+
+    void RegisterTask(KTimerTask* task, s64 time_from_now) {
+        this->RegisterAbsoluteTask(task, GetTick() + time_from_now);
+    }
+
+    void RegisterAbsoluteTask(KTimerTask* task, s64 task_time) {
+        KScopedDisableDispatch dd{m_kernel};
+        KScopedSpinLock lk{this->GetLock()};
+
+        if (this->RegisterAbsoluteTaskImpl(task, task_time)) {
+            if (task_time <= m_wakeup_time) {
+                this->EnableInterrupt(task_time);
+            }
+        }
+    }
+
+private:
+    void EnableInterrupt(s64 wakeup_time);
+    void DisableInterrupt();
+    bool GetInterruptEnabled();
+    s64 GetTick() const;
+    void DoTask();
+
+private:
+    // Absolute time in nanoseconds
+    s64 m_wakeup_time{std::numeric_limits<s64>::max()};
+    std::shared_ptr<Core::Timing::EventType> m_event_type{};
+};
+
+} // namespace Kernel

src/core/hle/kernel/k_hardware_timer_base.h

@@ -0,0 +1,92 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "core/hle/kernel/k_spin_lock.h"
+#include "core/hle/kernel/k_thread.h"
+#include "core/hle/kernel/k_timer_task.h"
+
+namespace Kernel {
+
+class KHardwareTimerBase {
+public:
+    explicit KHardwareTimerBase(KernelCore& kernel) : m_kernel{kernel} {}
+
+    void CancelTask(KTimerTask* task) {
+        KScopedDisableDispatch dd{m_kernel};
+        KScopedSpinLock lk{m_lock};
+
+        if (const s64 task_time = task->GetTime(); task_time > 0) {
+            this->RemoveTaskFromTree(task);
+        }
+    }
+
+protected:
+    KSpinLock& GetLock() {
+        return m_lock;
+    }
+
+    s64 DoInterruptTaskImpl(s64 cur_time) {
+        // We want to handle all tasks, returning the next time that a task is scheduled.
+        while (true) {
+            // Get the next task. If there isn't one, return 0.
+            KTimerTask* task = m_next_task;
+            if (task == nullptr) {
+                return 0;
+            }
+
+            // If the task needs to be done in the future, do it in the future and not now.
+            if (const s64 task_time = task->GetTime(); task_time > cur_time) {
+                return task_time;
+            }
+
+            // Remove the task from the tree of tasks, and update our next task.
+            this->RemoveTaskFromTree(task);
+
+            // Handle the task.
+            task->OnTimer();
+        }
+    }
+
+    bool RegisterAbsoluteTaskImpl(KTimerTask* task, s64 task_time) {
+        ASSERT(task_time > 0);
+
+        // Set the task's time, and insert it into our tree.
+        task->SetTime(task_time);
+        m_task_tree.insert(*task);
+
+        // Update our next task if relevant.
+        if (m_next_task != nullptr && m_next_task->GetTime() <= task_time) {
+            return false;
+        }
+        m_next_task = task;
+        return true;
+    }
+
+private:
+    void RemoveTaskFromTree(KTimerTask* task) {
+        // Erase from the tree.
+        auto it = m_task_tree.erase(m_task_tree.iterator_to(*task));
+
+        // Clear the task's scheduled time.
+        task->SetTime(0);
+
+        // Update our next task if relevant.
+        if (m_next_task == task) {
+            m_next_task = (it != m_task_tree.end()) ? std::addressof(*it) : nullptr;
+        }
+    }
+
+protected:
+    KernelCore& m_kernel;
+
+private:
+    using TimerTaskTree = Common::IntrusiveRedBlackTreeBaseTraits<KTimerTask>::TreeType<KTimerTask>;
+
+    KSpinLock m_lock{};
+    TimerTaskTree m_task_tree{};
+    KTimerTask* m_next_task{};
+};
+
+} // namespace Kernel

src/core/hle/kernel/k_process.cpp

@@ -285,6 +285,17 @@ void KProcess::UnregisterThread(KThread* thread) {
     thread_list.remove(thread);
 }
 
+u64 KProcess::GetFreeThreadCount() const {
+    if (resource_limit != nullptr) {
+        const auto current_value =
+            resource_limit->GetCurrentValue(LimitableResource::ThreadCountMax);
+        const auto limit_value = resource_limit->GetLimitValue(LimitableResource::ThreadCountMax);
+        return limit_value - current_value;
+    } else {
+        return 0;
+    }
+}
+
 Result KProcess::Reset() {
     // Lock the process and the scheduler.
     KScopedLightLock lk(state_lock);

src/core/hle/kernel/k_process.h

@@ -304,6 +304,9 @@ public:
     /// from this process' thread list.
     void UnregisterThread(KThread* thread);
 
+    /// Retrieves the number of available threads for this process.
+    u64 GetFreeThreadCount() const;
+
     /// Clears the signaled state of the process if and only if it's signaled.
     ///
     /// @pre The process must not be already terminated. If this is called on a

src/core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h

@@ -5,9 +5,9 @@
 
 #include "common/common_types.h"
 #include "core/hle/kernel/global_scheduler_context.h"
+#include "core/hle/kernel/k_hardware_timer.h"
 #include "core/hle/kernel/k_thread.h"
 #include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/time_manager.h"
 
 namespace Kernel {
 
@@ -22,7 +22,7 @@ public:
     ~KScopedSchedulerLockAndSleep() {
         // Register the sleep.
         if (timeout_tick > 0) {
-            kernel.TimeManager().ScheduleTimeEvent(thread, timeout_tick);
+            kernel.HardwareTimer().RegisterTask(thread, timeout_tick);
         }
 
         // Unlock the scheduler.

src/core/hle/kernel/k_shared_memory.cpp

@@ -6,6 +6,7 @@
 #include "core/hle/kernel/k_page_table.h"
 #include "core/hle/kernel/k_scoped_resource_reservation.h"
 #include "core/hle/kernel/k_shared_memory.h"
+#include "core/hle/kernel/k_system_resource.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/svc_results.h"
 
@@ -18,19 +19,19 @@ KSharedMemory::~KSharedMemory() {
 }
 
 Result KSharedMemory::Initialize(Core::DeviceMemory& device_memory_, KProcess* owner_process_,
-                                 KPageGroup&& page_list_, Svc::MemoryPermission owner_permission_,
-                                 Svc::MemoryPermission user_permission_, PAddr physical_address_,
-                                 std::size_t size_, std::string name_) {
+                                 Svc::MemoryPermission owner_permission_,
+                                 Svc::MemoryPermission user_permission_, std::size_t size_,
+                                 std::string name_) {
     // Set members.
     owner_process = owner_process_;
     device_memory = &device_memory_;
-    page_list = std::move(page_list_);
     owner_permission = owner_permission_;
     user_permission = user_permission_;
-    physical_address = physical_address_;
-    size = size_;
+    size = Common::AlignUp(size_, PageSize);
     name = std::move(name_);
 
+    const size_t num_pages = Common::DivideUp(size, PageSize);
+
     // Get the resource limit.
     KResourceLimit* reslimit = kernel.GetSystemResourceLimit();
 
@@ -39,6 +40,17 @@ Result KSharedMemory::Initialize(Core::DeviceMemory& device_memory_, KProcess* o
                                                   size_);
     R_UNLESS(memory_reservation.Succeeded(), ResultLimitReached);
 
+    // Allocate the memory.
+
+    //! HACK: Open continuous mapping from sysmodule pool.
+    auto option = KMemoryManager::EncodeOption(KMemoryManager::Pool::Secure,
+                                               KMemoryManager::Direction::FromBack);
+    physical_address = kernel.MemoryManager().AllocateAndOpenContinuous(num_pages, 1, option);
+    R_UNLESS(physical_address != 0, ResultOutOfMemory);
+
+    //! Insert the result into our page group.
+    page_group.emplace(physical_address, num_pages);
+
     // Commit our reservation.
     memory_reservation.Commit();
 
@@ -50,12 +62,23 @@ Result KSharedMemory::Initialize(Core::DeviceMemory& device_memory_, KProcess* o
     is_initialized = true;
 
     // Clear all pages in the memory.
-    std::memset(device_memory_.GetPointer<void>(physical_address_), 0, size_);
+    for (const auto& block : page_group->Nodes()) {
+        std::memset(device_memory_.GetPointer<void>(block.GetAddress()), 0, block.GetSize());
+    }
 
     return ResultSuccess;
 }
 
 void KSharedMemory::Finalize() {
+    // Close and finalize the page group.
+    // page_group->Close();
+    // page_group->Finalize();
+
+    //! HACK: Manually close.
+    for (const auto& block : page_group->Nodes()) {
+        kernel.MemoryManager().Close(block.GetAddress(), block.GetNumPages());
+    }
+
     // Release the memory reservation.
     resource_limit->Release(LimitableResource::PhysicalMemoryMax, size);
     resource_limit->Close();
@@ -65,32 +88,28 @@ void KSharedMemory::Finalize() {
 }
 
 Result KSharedMemory::Map(KProcess& target_process, VAddr address, std::size_t map_size,
-                          Svc::MemoryPermission permissions) {
-    const u64 page_count{(map_size + PageSize - 1) / PageSize};
+                          Svc::MemoryPermission map_perm) {
+    // Validate the size.
+    R_UNLESS(size == map_size, ResultInvalidSize);
 
-    if (page_list.GetNumPages() != page_count) {
-        UNIMPLEMENTED_MSG("Page count does not match");
-    }
-
-    const Svc::MemoryPermission expected =
+    // Validate the permission.
+    const Svc::MemoryPermission test_perm =
         &target_process == owner_process ? owner_permission : user_permission;
-
-    if (permissions != expected) {
-        UNIMPLEMENTED_MSG("Permission does not match");
+    if (test_perm == Svc::MemoryPermission::DontCare) {
+        ASSERT(map_perm == Svc::MemoryPermission::Read || map_perm == Svc::MemoryPermission::Write);
+    } else {
+        R_UNLESS(map_perm == test_perm, ResultInvalidNewMemoryPermission);
     }
 
-    return target_process.PageTable().MapPages(address, page_list, KMemoryState::Shared,
-                                               ConvertToKMemoryPermission(permissions));
+    return target_process.PageTable().MapPages(address, *page_group, KMemoryState::Shared,
+                                               ConvertToKMemoryPermission(map_perm));
 }
 
 Result KSharedMemory::Unmap(KProcess& target_process, VAddr address, std::size_t unmap_size) {
-    const u64 page_count{(unmap_size + PageSize - 1) / PageSize};
+    // Validate the size.
+    R_UNLESS(size == unmap_size, ResultInvalidSize);
 
-    if (page_list.GetNumPages() != page_count) {
-        UNIMPLEMENTED_MSG("Page count does not match");
-    }
-
-    return target_process.PageTable().UnmapPages(address, page_list, KMemoryState::Shared);
+    return target_process.PageTable().UnmapPages(address, *page_group, KMemoryState::Shared);
 }
 
 } // namespace Kernel

src/core/hle/kernel/k_shared_memory.h

@@ -3,6 +3,7 @@
 
 #pragma once
 
+#include <optional>
 #include <string>
 
 #include "common/common_types.h"
@@ -26,9 +27,8 @@ public:
     ~KSharedMemory() override;
 
     Result Initialize(Core::DeviceMemory& device_memory_, KProcess* owner_process_,
-                      KPageGroup&& page_list_, Svc::MemoryPermission owner_permission_,
-                      Svc::MemoryPermission user_permission_, PAddr physical_address_,
-                      std::size_t size_, std::string name_);
+                      Svc::MemoryPermission owner_permission_,
+                      Svc::MemoryPermission user_permission_, std::size_t size_, std::string name_);
 
     /**
      * Maps a shared memory block to an address in the target process' address space
@@ -76,7 +76,7 @@ public:
 private:
     Core::DeviceMemory* device_memory{};
     KProcess* owner_process{};
-    KPageGroup page_list;
+    std::optional<KPageGroup> page_group{};
     Svc::MemoryPermission owner_permission{};
     Svc::MemoryPermission user_permission{};
     PAddr physical_address{};

src/core/hle/kernel/k_thread.h

@@ -22,6 +22,7 @@
 #include "core/hle/kernel/k_light_lock.h"
 #include "core/hle/kernel/k_spin_lock.h"
 #include "core/hle/kernel/k_synchronization_object.h"
+#include "core/hle/kernel/k_timer_task.h"
 #include "core/hle/kernel/k_worker_task.h"
 #include "core/hle/kernel/slab_helpers.h"
 #include "core/hle/kernel/svc_common.h"
@@ -112,7 +113,8 @@ void SetCurrentThread(KernelCore& kernel, KThread* thread);
 [[nodiscard]] s32 GetCurrentCoreId(KernelCore& kernel);
 
 class KThread final : public KAutoObjectWithSlabHeapAndContainer<KThread, KWorkerTask>,
-                      public boost::intrusive::list_base_hook<> {
+                      public boost::intrusive::list_base_hook<>,
+                      public KTimerTask {
     KERNEL_AUTOOBJECT_TRAITS(KThread, KSynchronizationObject);
 
 private:
@@ -840,4 +842,8 @@ private:
     KernelCore& kernel;
 };
 
+inline void KTimerTask::OnTimer() {
+    static_cast<KThread*>(this)->OnTimer();
+}
+
 } // namespace Kernel

src/core/hle/kernel/k_thread_queue.cpp

@@ -1,9 +1,9 @@
 // SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
+#include "core/hle/kernel/k_hardware_timer.h"
 #include "core/hle/kernel/k_thread_queue.h"
 #include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/time_manager.h"
 
 namespace Kernel {
 
@@ -22,7 +22,7 @@ void KThreadQueue::EndWait(KThread* waiting_thread, Result wait_result) {
     waiting_thread->ClearWaitQueue();
 
     // Cancel the thread task.
-    kernel.TimeManager().UnscheduleTimeEvent(waiting_thread);
+    kernel.HardwareTimer().CancelTask(waiting_thread);
 }
 
 void KThreadQueue::CancelWait(KThread* waiting_thread, Result wait_result, bool cancel_timer_task) {
@@ -37,7 +37,7 @@ void KThreadQueue::CancelWait(KThread* waiting_thread, Result wait_result, bool
 
     // Cancel the thread task.
     if (cancel_timer_task) {
-        kernel.TimeManager().UnscheduleTimeEvent(waiting_thread);
+        kernel.HardwareTimer().CancelTask(waiting_thread);
     }
 }
 

src/core/hle/kernel/k_timer_task.h

@@ -0,0 +1,40 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "common/intrusive_red_black_tree.h"
+
+namespace Kernel {
+
+class KTimerTask : public Common::IntrusiveRedBlackTreeBaseNode<KTimerTask> {
+public:
+    static constexpr int Compare(const KTimerTask& lhs, const KTimerTask& rhs) {
+        if (lhs.GetTime() < rhs.GetTime()) {
+            return -1;
+        } else {
+            return 1;
+        }
+    }
+
+    constexpr explicit KTimerTask() = default;
+
+    constexpr void SetTime(s64 t) {
+        m_time = t;
+    }
+
+    constexpr s64 GetTime() const {
+        return m_time;
+    }
+
+    // NOTE: This is virtual in Nintendo's kernel. Prior to 13.0.0, KWaitObject was also a
+    // TimerTask; this is no longer the case. Since this is now KThread exclusive, we have
+    // devirtualized (see inline declaration for this inside k_thread.h).
+    void OnTimer();
+
+private:
+    // Absolute time in nanoseconds
+    s64 m_time{};
+};
+
+} // namespace Kernel

src/core/hle/kernel/kernel.cpp

@@ -26,6 +26,7 @@
 #include "core/hle/kernel/k_client_port.h"
 #include "core/hle/kernel/k_dynamic_resource_manager.h"
 #include "core/hle/kernel/k_handle_table.h"
+#include "core/hle/kernel/k_hardware_timer.h"
 #include "core/hle/kernel/k_memory_layout.h"
 #include "core/hle/kernel/k_memory_manager.h"
 #include "core/hle/kernel/k_page_buffer.h"
@@ -39,7 +40,6 @@
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/physical_core.h"
 #include "core/hle/kernel/service_thread.h"
-#include "core/hle/kernel/time_manager.h"
 #include "core/hle/result.h"
 #include "core/hle/service/sm/sm.h"
 #include "core/memory.h"
@@ -55,7 +55,7 @@ struct KernelCore::Impl {
     static constexpr size_t ReservedDynamicPageCount = 64;
 
     explicit Impl(Core::System& system_, KernelCore& kernel_)
-        : time_manager{system_}, service_threads_manager{1, "ServiceThreadsManager"},
+        : service_threads_manager{1, "ServiceThreadsManager"},
           service_thread_barrier{2}, system{system_} {}
 
     void SetMulticore(bool is_multi) {
@@ -63,6 +63,9 @@ struct KernelCore::Impl {
     }
 
     void Initialize(KernelCore& kernel) {
+        hardware_timer = std::make_unique<Kernel::KHardwareTimer>(kernel);
+        hardware_timer->Initialize();
+
         global_object_list_container = std::make_unique<KAutoObjectWithListContainer>(kernel);
         global_scheduler_context = std::make_unique<Kernel::GlobalSchedulerContext>(kernel);
         global_handle_table = std::make_unique<Kernel::KHandleTable>(kernel);
@@ -91,6 +94,7 @@ struct KernelCore::Impl {
                                        pt_heap_region.GetSize());
         }
 
+        InitializeHackSharedMemory();
         RegisterHostThread(nullptr);
 
         default_service_thread = &CreateServiceThread(kernel, "DefaultServiceThread");
@@ -104,12 +108,16 @@ struct KernelCore::Impl {
     }
 
     void CloseCurrentProcess() {
-        (*current_process).Finalize();
-        // current_process->Close();
-        // TODO: The current process should be destroyed based on accurate ref counting after
+        KProcess* old_process = current_process.exchange(nullptr);
+        if (old_process == nullptr) {
+            return;
+        }
+
+        // old_process->Close();
+        // TODO: The process should be destroyed based on accurate ref counting after
         // calling Close(). Adding a manual Destroy() call instead to avoid a memory leak.
-        (*current_process).Destroy();
-        current_process = nullptr;
+        old_process->Finalize();
+        old_process->Destroy();
     }
 
     void Shutdown() {
@@ -189,6 +197,9 @@ struct KernelCore::Impl {
         // Ensure that the object list container is finalized and properly shutdown.
         global_object_list_container->Finalize();
         global_object_list_container.reset();
+
+        hardware_timer->Finalize();
+        hardware_timer.reset();
     }
 
     void CloseServices() {
@@ -716,14 +727,14 @@ struct KernelCore::Impl {
     }
 
     void InitializeMemoryLayout() {
-        const auto system_pool = memory_layout->GetKernelSystemPoolRegionPhysicalExtents();
-
         // Initialize the memory manager.
         memory_manager = std::make_unique<KMemoryManager>(system);
         const auto& management_region = memory_layout->GetPoolManagementRegion();
         ASSERT(management_region.GetEndAddress() != 0);
         memory_manager->Initialize(management_region.GetAddress(), management_region.GetSize());
+    }
 
+    void InitializeHackSharedMemory() {
         // Setup memory regions for emulated processes
         // TODO(bunnei): These should not be hardcoded regions initialized within the kernel
         constexpr std::size_t hid_size{0x40000};
@@ -732,39 +743,23 @@ struct KernelCore::Impl {
         constexpr std::size_t time_size{0x1000};
         constexpr std::size_t hidbus_size{0x1000};
 
-        const PAddr hid_phys_addr{system_pool.GetAddress()};
-        const PAddr font_phys_addr{system_pool.GetAddress() + hid_size};
-        const PAddr irs_phys_addr{system_pool.GetAddress() + hid_size + font_size};
-        const PAddr time_phys_addr{system_pool.GetAddress() + hid_size + font_size + irs_size};
-        const PAddr hidbus_phys_addr{system_pool.GetAddress() + hid_size + font_size + irs_size +
-                                     time_size};
-
         hid_shared_mem = KSharedMemory::Create(system.Kernel());
         font_shared_mem = KSharedMemory::Create(system.Kernel());
         irs_shared_mem = KSharedMemory::Create(system.Kernel());
         time_shared_mem = KSharedMemory::Create(system.Kernel());
         hidbus_shared_mem = KSharedMemory::Create(system.Kernel());
 
-        hid_shared_mem->Initialize(system.DeviceMemory(), nullptr,
-                                   {hid_phys_addr, hid_size / PageSize},
-                                   Svc::MemoryPermission::None, Svc::MemoryPermission::Read,
-                                   hid_phys_addr, hid_size, "HID:SharedMemory");
-        font_shared_mem->Initialize(system.DeviceMemory(), nullptr,
-                                    {font_phys_addr, font_size / PageSize},
-                                    Svc::MemoryPermission::None, Svc::MemoryPermission::Read,
-                                    font_phys_addr, font_size, "Font:SharedMemory");
-        irs_shared_mem->Initialize(system.DeviceMemory(), nullptr,
-                                   {irs_phys_addr, irs_size / PageSize},
-                                   Svc::MemoryPermission::None, Svc::MemoryPermission::Read,
-                                   irs_phys_addr, irs_size, "IRS:SharedMemory");
-        time_shared_mem->Initialize(system.DeviceMemory(), nullptr,
-                                    {time_phys_addr, time_size / PageSize},
-                                    Svc::MemoryPermission::None, Svc::MemoryPermission::Read,
-                                    time_phys_addr, time_size, "Time:SharedMemory");
-        hidbus_shared_mem->Initialize(system.DeviceMemory(), nullptr,
-                                      {hidbus_phys_addr, hidbus_size / PageSize},
-                                      Svc::MemoryPermission::None, Svc::MemoryPermission::Read,
-                                      hidbus_phys_addr, hidbus_size, "HidBus:SharedMemory");
+        hid_shared_mem->Initialize(system.DeviceMemory(), nullptr, Svc::MemoryPermission::None,
+                                   Svc::MemoryPermission::Read, hid_size, "HID:SharedMemory");
+        font_shared_mem->Initialize(system.DeviceMemory(), nullptr, Svc::MemoryPermission::None,
+                                    Svc::MemoryPermission::Read, font_size, "Font:SharedMemory");
+        irs_shared_mem->Initialize(system.DeviceMemory(), nullptr, Svc::MemoryPermission::None,
+                                   Svc::MemoryPermission::Read, irs_size, "IRS:SharedMemory");
+        time_shared_mem->Initialize(system.DeviceMemory(), nullptr, Svc::MemoryPermission::None,
+                                    Svc::MemoryPermission::Read, time_size, "Time:SharedMemory");
+        hidbus_shared_mem->Initialize(system.DeviceMemory(), nullptr, Svc::MemoryPermission::None,
+                                      Svc::MemoryPermission::Read, hidbus_size,
+                                      "HidBus:SharedMemory");
     }
 
     KClientPort* CreateNamedServicePort(std::string name) {
@@ -828,7 +823,7 @@ struct KernelCore::Impl {
     std::vector<KProcess*> process_list;
     std::atomic<KProcess*> current_process{};
     std::unique_ptr<Kernel::GlobalSchedulerContext> global_scheduler_context;
-    Kernel::TimeManager time_manager;
+    std::unique_ptr<Kernel::KHardwareTimer> hardware_timer;
 
     Init::KSlabResourceCounts slab_resource_counts{};
     KResourceLimit* system_resource_limit{};
@@ -1015,12 +1010,8 @@ Kernel::KScheduler* KernelCore::CurrentScheduler() {
     return impl->schedulers[core_id].get();
 }
 
-Kernel::TimeManager& KernelCore::TimeManager() {
-    return impl->time_manager;
-}
-
-const Kernel::TimeManager& KernelCore::TimeManager() const {
-    return impl->time_manager;
+Kernel::KHardwareTimer& KernelCore::HardwareTimer() {
+    return *impl->hardware_timer;
 }
 
 Core::ExclusiveMonitor& KernelCore::GetExclusiveMonitor() {

src/core/hle/kernel/kernel.h

@@ -39,6 +39,7 @@ class KDynamicPageManager;
 class KEvent;
 class KEventInfo;
 class KHandleTable;
+class KHardwareTimer;
 class KLinkedListNode;
 class KMemoryLayout;
 class KMemoryManager;
@@ -63,7 +64,6 @@ class KCodeMemory;
 class PhysicalCore;
 class ServiceThread;
 class Synchronization;
-class TimeManager;
 
 using ServiceInterfaceFactory =
     std::function<KClientPort&(Service::SM::ServiceManager&, Core::System&)>;
@@ -175,11 +175,8 @@ public:
     /// Gets the an instance of the current physical CPU core.
     const Kernel::PhysicalCore& CurrentPhysicalCore() const;
 
-    /// Gets the an instance of the TimeManager Interface.
-    Kernel::TimeManager& TimeManager();
-
-    /// Gets the an instance of the TimeManager Interface.
-    const Kernel::TimeManager& TimeManager() const;
+    /// Gets the an instance of the hardware timer.
+    Kernel::KHardwareTimer& HardwareTimer();
 
     /// Stops execution of 'id' core, in order to reschedule a new thread.
     void PrepareReschedule(std::size_t id);

src/core/hle/kernel/svc.cpp

@@ -784,63 +784,29 @@ static Result GetInfo(Core::System& system, u64* result, u64 info_id, Handle han
     LOG_TRACE(Kernel_SVC, "called info_id=0x{:X}, info_sub_id=0x{:X}, handle=0x{:08X}", info_id,
               info_sub_id, handle);
 
-    enum class GetInfoType : u64 {
-        // 1.0.0+
-        AllowedCPUCoreMask = 0,
-        AllowedThreadPriorityMask = 1,
-        MapRegionBaseAddr = 2,
-        MapRegionSize = 3,
-        HeapRegionBaseAddr = 4,
-        HeapRegionSize = 5,
-        TotalPhysicalMemoryAvailable = 6,
-        TotalPhysicalMemoryUsed = 7,
-        IsCurrentProcessBeingDebugged = 8,
-        RegisterResourceLimit = 9,
-        IdleTickCount = 10,
-        RandomEntropy = 11,
-        ThreadTickCount = 0xF0000002,
-        // 2.0.0+
-        ASLRRegionBaseAddr = 12,
-        ASLRRegionSize = 13,
-        StackRegionBaseAddr = 14,
-        StackRegionSize = 15,
-        // 3.0.0+
-        SystemResourceSize = 16,
-        SystemResourceUsage = 17,
-        TitleId = 18,
-        // 4.0.0+
-        PrivilegedProcessId = 19,
-        // 5.0.0+
-        UserExceptionContextAddr = 20,
-        // 6.0.0+
-        TotalPhysicalMemoryAvailableWithoutSystemResource = 21,
-        TotalPhysicalMemoryUsedWithoutSystemResource = 22,
-
-        // Homebrew only
-        MesosphereCurrentProcess = 65001,
-    };
-
-    const auto info_id_type = static_cast<GetInfoType>(info_id);
+    const auto info_id_type = static_cast<InfoType>(info_id);
 
     switch (info_id_type) {
-    case GetInfoType::AllowedCPUCoreMask:
-    case GetInfoType::AllowedThreadPriorityMask:
-    case GetInfoType::MapRegionBaseAddr:
-    case GetInfoType::MapRegionSize:
-    case GetInfoType::HeapRegionBaseAddr:
-    case GetInfoType::HeapRegionSize:
-    case GetInfoType::ASLRRegionBaseAddr:
-    case GetInfoType::ASLRRegionSize:
-    case GetInfoType::StackRegionBaseAddr:
-    case GetInfoType::StackRegionSize:
-    case GetInfoType::TotalPhysicalMemoryAvailable:
-    case GetInfoType::TotalPhysicalMemoryUsed:
-    case GetInfoType::SystemResourceSize:
-    case GetInfoType::SystemResourceUsage:
-    case GetInfoType::TitleId:
-    case GetInfoType::UserExceptionContextAddr:
-    case GetInfoType::TotalPhysicalMemoryAvailableWithoutSystemResource:
-    case GetInfoType::TotalPhysicalMemoryUsedWithoutSystemResource: {
+    case InfoType::CoreMask:
+    case InfoType::PriorityMask:
+    case InfoType::AliasRegionAddress:
+    case InfoType::AliasRegionSize:
+    case InfoType::HeapRegionAddress:
+    case InfoType::HeapRegionSize:
+    case InfoType::AslrRegionAddress:
+    case InfoType::AslrRegionSize:
+    case InfoType::StackRegionAddress:
+    case InfoType::StackRegionSize:
+    case InfoType::TotalMemorySize:
+    case InfoType::UsedMemorySize:
+    case InfoType::SystemResourceSizeTotal:
+    case InfoType::SystemResourceSizeUsed:
+    case InfoType::ProgramId:
+    case InfoType::UserExceptionContextAddress:
+    case InfoType::TotalNonSystemMemorySize:
+    case InfoType::UsedNonSystemMemorySize:
+    case InfoType::IsApplication:
+    case InfoType::FreeThreadCount: {
         if (info_sub_id != 0) {
             LOG_ERROR(Kernel_SVC, "Info sub id is non zero! info_id={}, info_sub_id={}", info_id,
                       info_sub_id);
@@ -856,79 +822,83 @@ static Result GetInfo(Core::System& system, u64* result, u64 info_id, Handle han
         }
 
         switch (info_id_type) {
-        case GetInfoType::AllowedCPUCoreMask:
+        case InfoType::CoreMask:
             *result = process->GetCoreMask();
             return ResultSuccess;
 
-        case GetInfoType::AllowedThreadPriorityMask:
+        case InfoType::PriorityMask:
             *result = process->GetPriorityMask();
             return ResultSuccess;
 
-        case GetInfoType::MapRegionBaseAddr:
+        case InfoType::AliasRegionAddress:
             *result = process->PageTable().GetAliasRegionStart();
             return ResultSuccess;
 
-        case GetInfoType::MapRegionSize:
+        case InfoType::AliasRegionSize:
             *result = process->PageTable().GetAliasRegionSize();
             return ResultSuccess;
 
-        case GetInfoType::HeapRegionBaseAddr:
+        case InfoType::HeapRegionAddress:
             *result = process->PageTable().GetHeapRegionStart();
             return ResultSuccess;
 
-        case GetInfoType::HeapRegionSize:
+        case InfoType::HeapRegionSize:
             *result = process->PageTable().GetHeapRegionSize();
             return ResultSuccess;
 
-        case GetInfoType::ASLRRegionBaseAddr:
+        case InfoType::AslrRegionAddress:
             *result = process->PageTable().GetAliasCodeRegionStart();
             return ResultSuccess;
 
-        case GetInfoType::ASLRRegionSize:
+        case InfoType::AslrRegionSize:
             *result = process->PageTable().GetAliasCodeRegionSize();
             return ResultSuccess;
 
-        case GetInfoType::StackRegionBaseAddr:
+        case InfoType::StackRegionAddress:
             *result = process->PageTable().GetStackRegionStart();
             return ResultSuccess;
 
-        case GetInfoType::StackRegionSize:
+        case InfoType::StackRegionSize:
             *result = process->PageTable().GetStackRegionSize();
             return ResultSuccess;
 
-        case GetInfoType::TotalPhysicalMemoryAvailable:
+        case InfoType::TotalMemorySize:
             *result = process->GetTotalPhysicalMemoryAvailable();
             return ResultSuccess;
 
-        case GetInfoType::TotalPhysicalMemoryUsed:
+        case InfoType::UsedMemorySize:
             *result = process->GetTotalPhysicalMemoryUsed();
             return ResultSuccess;
 
-        case GetInfoType::SystemResourceSize:
+        case InfoType::SystemResourceSizeTotal:
             *result = process->GetSystemResourceSize();
             return ResultSuccess;
 
-        case GetInfoType::SystemResourceUsage:
+        case InfoType::SystemResourceSizeUsed:
             LOG_WARNING(Kernel_SVC, "(STUBBED) Attempted to query system resource usage");
             *result = process->GetSystemResourceUsage();
             return ResultSuccess;
 
-        case GetInfoType::TitleId:
+        case InfoType::ProgramId:
             *result = process->GetProgramID();
             return ResultSuccess;
 
-        case GetInfoType::UserExceptionContextAddr:
+        case InfoType::UserExceptionContextAddress:
             *result = process->GetProcessLocalRegionAddress();
             return ResultSuccess;
 
-        case GetInfoType::TotalPhysicalMemoryAvailableWithoutSystemResource:
+        case InfoType::TotalNonSystemMemorySize:
             *result = process->GetTotalPhysicalMemoryAvailableWithoutSystemResource();
             return ResultSuccess;
 
-        case GetInfoType::TotalPhysicalMemoryUsedWithoutSystemResource:
+        case InfoType::UsedNonSystemMemorySize:
             *result = process->GetTotalPhysicalMemoryUsedWithoutSystemResource();
             return ResultSuccess;
 
+        case InfoType::FreeThreadCount:
+            *result = process->GetFreeThreadCount();
+            return ResultSuccess;
+
         default:
             break;
         }
@@ -937,11 +907,11 @@ static Result GetInfo(Core::System& system, u64* result, u64 info_id, Handle han
         return ResultInvalidEnumValue;
     }
 
-    case GetInfoType::IsCurrentProcessBeingDebugged:
+    case InfoType::DebuggerAttached:
         *result = 0;
         return ResultSuccess;
 
-    case GetInfoType::RegisterResourceLimit: {
+    case InfoType::ResourceLimit: {
         if (handle != 0) {
             LOG_ERROR(Kernel, "Handle is non zero! handle={:08X}", handle);
             return ResultInvalidHandle;
@@ -969,7 +939,7 @@ static Result GetInfo(Core::System& system, u64* result, u64 info_id, Handle han
         return ResultSuccess;
     }
 
-    case GetInfoType::RandomEntropy:
+    case InfoType::RandomEntropy:
         if (handle != 0) {
             LOG_ERROR(Kernel_SVC, "Process Handle is non zero, expected 0 result but got {:016X}",
                       handle);
@@ -985,13 +955,13 @@ static Result GetInfo(Core::System& system, u64* result, u64 info_id, Handle han
         *result = system.Kernel().CurrentProcess()->GetRandomEntropy(info_sub_id);
         return ResultSuccess;
 
-    case GetInfoType::PrivilegedProcessId:
+    case InfoType::InitialProcessIdRange:
         LOG_WARNING(Kernel_SVC,
                     "(STUBBED) Attempted to query privileged process id bounds, returned 0");
         *result = 0;
         return ResultSuccess;
 
-    case GetInfoType::ThreadTickCount: {
+    case InfoType::ThreadTickCount: {
         constexpr u64 num_cpus = 4;
         if (info_sub_id != 0xFFFFFFFFFFFFFFFF && info_sub_id >= num_cpus) {
             LOG_ERROR(Kernel_SVC, "Core count is out of range, expected {} but got {}", num_cpus,
@@ -1026,7 +996,7 @@ static Result GetInfo(Core::System& system, u64* result, u64 info_id, Handle han
         *result = out_ticks;
         return ResultSuccess;
     }
-    case GetInfoType::IdleTickCount: {
+    case InfoType::IdleTickCount: {
         // Verify the input handle is invalid.
         R_UNLESS(handle == InvalidHandle, ResultInvalidHandle);
 
@@ -1040,7 +1010,7 @@ static Result GetInfo(Core::System& system, u64* result, u64 info_id, Handle han
         *result = system.Kernel().CurrentScheduler()->GetIdleThread()->GetCpuTime();
         return ResultSuccess;
     }
-    case GetInfoType::MesosphereCurrentProcess: {
+    case InfoType::MesosphereCurrentProcess: {
         // Verify the input handle is invalid.
         R_UNLESS(handle == InvalidHandle, ResultInvalidHandle);
 

src/core/hle/kernel/time_manager.cpp

@@ -1,44 +0,0 @@
-// SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
-// SPDX-License-Identifier: GPL-2.0-or-later
-
-#include "common/assert.h"
-#include "core/core.h"
-#include "core/core_timing.h"
-#include "core/hle/kernel/k_scheduler.h"
-#include "core/hle/kernel/k_thread.h"
-#include "core/hle/kernel/time_manager.h"
-
-namespace Kernel {
-
-TimeManager::TimeManager(Core::System& system_) : system{system_} {
-    time_manager_event_type = Core::Timing::CreateEvent(
-        "Kernel::TimeManagerCallback",
-        [this](std::uintptr_t thread_handle, s64 time,
-               std::chrono::nanoseconds) -> std::optional<std::chrono::nanoseconds> {
-            KThread* thread = reinterpret_cast<KThread*>(thread_handle);
-            {
-                KScopedSchedulerLock sl(system.Kernel());
-                thread->OnTimer();
-            }
-            return std::nullopt;
-        });
-}
-
-void TimeManager::ScheduleTimeEvent(KThread* thread, s64 nanoseconds) {
-    std::scoped_lock lock{mutex};
-    if (nanoseconds > 0) {
-        ASSERT(thread);
-        ASSERT(thread->GetState() != ThreadState::Runnable);
-        system.CoreTiming().ScheduleEvent(std::chrono::nanoseconds{nanoseconds},
-                                          time_manager_event_type,
-                                          reinterpret_cast<uintptr_t>(thread));
-    }
-}
-
-void TimeManager::UnscheduleTimeEvent(KThread* thread) {
-    std::scoped_lock lock{mutex};
-    system.CoreTiming().UnscheduleEvent(time_manager_event_type,
-                                        reinterpret_cast<uintptr_t>(thread));
-}
-
-} // namespace Kernel

src/core/hle/kernel/time_manager.h

@@ -1,41 +0,0 @@
-// SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
-// SPDX-License-Identifier: GPL-2.0-or-later
-
-#pragma once
-
-#include <memory>
-#include <mutex>
-
-namespace Core {
-class System;
-} // namespace Core
-
-namespace Core::Timing {
-struct EventType;
-} // namespace Core::Timing
-
-namespace Kernel {
-
-class KThread;
-
-/**
- * The `TimeManager` takes care of scheduling time events on threads and executes their TimeUp
- * method when the event is triggered.
- */
-class TimeManager {
-public:
-    explicit TimeManager(Core::System& system);
-
-    /// Schedule a time event on `timetask` thread that will expire in 'nanoseconds'
-    void ScheduleTimeEvent(KThread* time_task, s64 nanoseconds);
-
-    /// Unschedule an existing time event
-    void UnscheduleTimeEvent(KThread* thread);
-
-private:
-    Core::System& system;
-    std::shared_ptr<Core::Timing::EventType> time_manager_event_type;
-    std::mutex mutex;
-};
-
-} // namespace Kernel

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

@@ -203,8 +203,9 @@ private:
 };
 
 AudInU::AudInU(Core::System& system_)
-    : ServiceFramework{system_, "audin:u"}, service_context{system_, "AudInU"},
-      impl{std::make_unique<AudioCore::AudioIn::Manager>(system_)} {
+    : ServiceFramework{system_, "audin:u", ServiceThreadType::CreateNew},
+      service_context{system_, "AudInU"}, impl{std::make_unique<AudioCore::AudioIn::Manager>(
+                                              system_)} {
     // clang-format off
     static const FunctionInfo functions[] = {
         {0, &AudInU::ListAudioIns, "ListAudioIns"},

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

@@ -26,8 +26,9 @@ public:
     explicit IAudioOut(Core::System& system_, AudioCore::AudioOut::Manager& manager,
                        size_t session_id, const std::string& device_name,
                        const AudioOutParameter& in_params, u32 handle, u64 applet_resource_user_id)
-        : ServiceFramework{system_, "IAudioOut"}, service_context{system_, "IAudioOut"},
-          event{service_context.CreateEvent("AudioOutEvent")},
+        : ServiceFramework{system_, "IAudioOut", ServiceThreadType::CreateNew},
+          service_context{system_, "IAudioOut"}, event{service_context.CreateEvent(
+                                                     "AudioOutEvent")},
           impl{std::make_shared<AudioCore::AudioOut::Out>(system_, manager, event, session_id)} {
 
         // clang-format off
@@ -220,8 +221,9 @@ private:
 };
 
 AudOutU::AudOutU(Core::System& system_)
-    : ServiceFramework{system_, "audout:u"}, service_context{system_, "AudOutU"},
-      impl{std::make_unique<AudioCore::AudioOut::Manager>(system_)} {
+    : ServiceFramework{system_, "audout:u", ServiceThreadType::CreateNew},
+      service_context{system_, "AudOutU"}, impl{std::make_unique<AudioCore::AudioOut::Manager>(
+                                               system_)} {
     // clang-format off
     static const FunctionInfo functions[] = {
         {0, &AudOutU::ListAudioOuts, "ListAudioOuts"},

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

@@ -35,9 +35,10 @@ public:
                             AudioCore::AudioRendererParameterInternal& params,
                             Kernel::KTransferMemory* transfer_memory, u64 transfer_memory_size,
                             u32 process_handle, u64 applet_resource_user_id, s32 session_id)
-        : ServiceFramework{system_, "IAudioRenderer"}, service_context{system_, "IAudioRenderer"},
-          rendered_event{service_context.CreateEvent("IAudioRendererEvent")}, manager{manager_},
-          impl{std::make_unique<Renderer>(system_, manager, rendered_event)} {
+        : ServiceFramework{system_, "IAudioRenderer", ServiceThreadType::CreateNew},
+          service_context{system_, "IAudioRenderer"}, rendered_event{service_context.CreateEvent(
+                                                          "IAudioRendererEvent")},
+          manager{manager_}, impl{std::make_unique<Renderer>(system_, manager, rendered_event)} {
         // clang-format off
         static const FunctionInfo functions[] = {
             {0, &IAudioRenderer::GetSampleRate, "GetSampleRate"},
@@ -242,8 +243,10 @@ class IAudioDevice final : public ServiceFramework<IAudioDevice> {
 public:
     explicit IAudioDevice(Core::System& system_, u64 applet_resource_user_id, u32 revision,
                           u32 device_num)
-        : ServiceFramework{system_, "IAudioDevice"}, service_context{system_, "IAudioDevice"},
-          impl{std::make_unique<AudioDevice>(system_, applet_resource_user_id, revision)},
+        : ServiceFramework{system_, "IAudioDevice", ServiceThreadType::CreateNew},
+          service_context{system_, "IAudioDevice"}, impl{std::make_unique<AudioDevice>(
+                                                        system_, applet_resource_user_id,
+                                                        revision)},
           event{service_context.CreateEvent(fmt::format("IAudioDeviceEvent-{}", device_num))} {
         static const FunctionInfo functions[] = {
             {0, &IAudioDevice::ListAudioDeviceName, "ListAudioDeviceName"},
@@ -418,7 +421,7 @@ private:
 };
 
 AudRenU::AudRenU(Core::System& system_)
-    : ServiceFramework{system_, "audren:u"},
+    : ServiceFramework{system_, "audren:u", ServiceThreadType::CreateNew},
       service_context{system_, "audren:u"}, impl{std::make_unique<Manager>(system_)} {
     // clang-format off
     static const FunctionInfo functions[] = {

src/core/hle/service/nfc/nfc_user.cpp

@@ -97,7 +97,7 @@ void IUser::IsNfcEnabled(Kernel::HLERequestContext& ctx) {
 }
 
 void IUser::ListDevices(Kernel::HLERequestContext& ctx) {
-    LOG_INFO(Service_NFC, "called");
+    LOG_DEBUG(Service_NFC, "called");
 
     if (state == State::NonInitialized) {
         IPC::ResponseBuilder rb{ctx, 2};

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

@@ -83,7 +83,7 @@ void IUser::Finalize(Kernel::HLERequestContext& ctx) {
 }
 
 void IUser::ListDevices(Kernel::HLERequestContext& ctx) {
-    LOG_INFO(Service_NFP, "called");
+    LOG_DEBUG(Service_NFP, "called");
 
     if (state == State::NonInitialized) {
         IPC::ResponseBuilder rb{ctx, 2};

src/core/hle/service/set/set.cpp

@@ -191,6 +191,13 @@ void SET::GetKeyCodeMap2(Kernel::HLERequestContext& ctx) {
     GetKeyCodeMapImpl(ctx);
 }
 
+void SET::GetDeviceNickName(Kernel::HLERequestContext& ctx) {
+    LOG_DEBUG(Service_SET, "called");
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(ResultSuccess);
+    ctx.WriteBuffer(Settings::values.device_name.GetValue());
+}
+
 SET::SET(Core::System& system_) : ServiceFramework{system_, "set"} {
     // clang-format off
     static const FunctionInfo functions[] = {
@@ -205,7 +212,7 @@ SET::SET(Core::System& system_) : ServiceFramework{system_, "set"} {
         {8, &SET::GetQuestFlag, "GetQuestFlag"},
         {9, &SET::GetKeyCodeMap2, "GetKeyCodeMap2"},
         {10, nullptr, "GetFirmwareVersionForDebug"},
-        {11, nullptr, "GetDeviceNickName"},
+        {11, &SET::GetDeviceNickName, "GetDeviceNickName"},
     };
     // clang-format on
 

src/core/hle/service/set/set.h

@@ -50,6 +50,7 @@ private:
     void GetRegionCode(Kernel::HLERequestContext& ctx);
     void GetKeyCodeMap(Kernel::HLERequestContext& ctx);
     void GetKeyCodeMap2(Kernel::HLERequestContext& ctx);
+    void GetDeviceNickName(Kernel::HLERequestContext& ctx);
 };
 
 } // namespace Service::Set

src/core/hle/service/set/set_sys.cpp

@@ -3,6 +3,7 @@
 
 #include "common/assert.h"
 #include "common/logging/log.h"
+#include "common/settings.h"
 #include "core/file_sys/errors.h"
 #include "core/file_sys/system_archive/system_version.h"
 #include "core/hle/ipc_helpers.h"
@@ -176,6 +177,13 @@ void SET_SYS::GetSettingsItemValue(Kernel::HLERequestContext& ctx) {
     rb.Push(response);
 }
 
+void SET_SYS::GetDeviceNickName(Kernel::HLERequestContext& ctx) {
+    LOG_DEBUG(Service_SET, "called");
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(ResultSuccess);
+    ctx.WriteBuffer(::Settings::values.device_name.GetValue());
+}
+
 SET_SYS::SET_SYS(Core::System& system_) : ServiceFramework{system_, "set:sys"} {
     // clang-format off
     static const FunctionInfo functions[] = {
@@ -253,7 +261,7 @@ SET_SYS::SET_SYS(Core::System& system_) : ServiceFramework{system_, "set:sys"} {
         {74, nullptr, "SetWirelessLanEnableFlag"},
         {75, nullptr, "GetInitialLaunchSettings"},
         {76, nullptr, "SetInitialLaunchSettings"},
-        {77, nullptr, "GetDeviceNickName"},
+        {77, &SET_SYS::GetDeviceNickName, "GetDeviceNickName"},
         {78, nullptr, "SetDeviceNickName"},
         {79, nullptr, "GetProductModel"},
         {80, nullptr, "GetLdnChannel"},

src/core/hle/service/set/set_sys.h

@@ -29,6 +29,7 @@ private:
     void GetFirmwareVersion2(Kernel::HLERequestContext& ctx);
     void GetColorSetId(Kernel::HLERequestContext& ctx);
     void SetColorSetId(Kernel::HLERequestContext& ctx);
+    void GetDeviceNickName(Kernel::HLERequestContext& ctx);
 
     ColorSet color_set = ColorSet::BasicWhite;
 };

src/core/hle/service/time/clock_types.h

@@ -49,6 +49,7 @@ struct SteadyClockContext {
 static_assert(sizeof(SteadyClockContext) == 0x18, "SteadyClockContext is incorrect size");
 static_assert(std::is_trivially_copyable_v<SteadyClockContext>,
               "SteadyClockContext must be trivially copyable");
+using StandardSteadyClockTimePointType = SteadyClockContext;
 
 struct SystemClockContext {
     s64 offset;

src/core/hle/service/time/time_sharedmemory.cpp

@@ -26,23 +26,24 @@ void SharedMemory::SetupStandardSteadyClock(const Common::UUID& clock_source_id,
     const Clock::SteadyClockContext context{
         static_cast<u64>(current_time_point.nanoseconds - ticks_time_span.nanoseconds),
         clock_source_id};
-    shared_memory_format.standard_steady_clock_timepoint.StoreData(
-        system.Kernel().GetTimeSharedMem().GetPointer(), context);
+    StoreToLockFreeAtomicType(&GetFormat()->standard_steady_clock_timepoint, context);
 }
 
 void SharedMemory::UpdateLocalSystemClockContext(const Clock::SystemClockContext& context) {
-    shared_memory_format.standard_local_system_clock_context.StoreData(
-        system.Kernel().GetTimeSharedMem().GetPointer(), context);
+    StoreToLockFreeAtomicType(&GetFormat()->standard_local_system_clock_context, context);
 }
 
 void SharedMemory::UpdateNetworkSystemClockContext(const Clock::SystemClockContext& context) {
-    shared_memory_format.standard_network_system_clock_context.StoreData(
-        system.Kernel().GetTimeSharedMem().GetPointer(), context);
+    StoreToLockFreeAtomicType(&GetFormat()->standard_network_system_clock_context, context);
 }
 
 void SharedMemory::SetAutomaticCorrectionEnabled(bool is_enabled) {
-    shared_memory_format.standard_user_system_clock_automatic_correction.StoreData(
-        system.Kernel().GetTimeSharedMem().GetPointer(), is_enabled);
+    StoreToLockFreeAtomicType(
+        &GetFormat()->is_standard_user_system_clock_automatic_correction_enabled, is_enabled);
+}
+
+SharedMemory::Format* SharedMemory::GetFormat() {
+    return reinterpret_cast<SharedMemory::Format*>(system.Kernel().GetTimeSharedMem().GetPointer());
 }
 
 } // namespace Service::Time

src/core/hle/service/time/time_sharedmemory.h

@@ -10,45 +10,68 @@
 
 namespace Service::Time {
 
+// Note: this type is not safe for concurrent writes.
+template <typename T>
+struct LockFreeAtomicType {
+    u32 counter_;
+    std::array<T, 2> value_;
+};
+
+template <typename T>
+static inline void StoreToLockFreeAtomicType(LockFreeAtomicType<T>* p, const T& value) {
+    // Get the current counter.
+    auto counter = p->counter_;
+
+    // Increment the counter.
+    ++counter;
+
+    // Store the updated value.
+    p->value_[counter % 2] = value;
+
+    // Fence memory.
+    std::atomic_thread_fence(std::memory_order_release);
+
+    // Set the updated counter.
+    p->counter_ = counter;
+}
+
+template <typename T>
+static inline T LoadFromLockFreeAtomicType(const LockFreeAtomicType<T>* p) {
+    while (true) {
+        // Get the counter.
+        auto counter = p->counter_;
+
+        // Get the value.
+        auto value = p->value_[counter % 2];
+
+        // Fence memory.
+        std::atomic_thread_fence(std::memory_order_acquire);
+
+        // Check that the counter matches.
+        if (counter == p->counter_) {
+            return value;
+        }
+    }
+}
+
 class SharedMemory final {
 public:
     explicit SharedMemory(Core::System& system_);
     ~SharedMemory();
 
-    // TODO(ogniK): We have to properly simulate memory barriers, how are we going to do this?
-    template <typename T, std::size_t Offset>
-    struct MemoryBarrier {
-        static_assert(std::is_trivially_copyable_v<T>, "T must be trivially copyable");
-        u32_le read_attempt{};
-        std::array<T, 2> data{};
-
-        // These are not actually memory barriers at the moment as we don't have multicore and all
-        // HLE is mutexed. This will need to properly be implemented when we start updating the time
-        // points on threads. As of right now, we'll be updated both values synchronously and just
-        // incrementing the read_attempt to indicate that we waited.
-        void StoreData(u8* shared_memory, T data_to_store) {
-            std::memcpy(this, shared_memory + Offset, sizeof(*this));
-            read_attempt++;
-            data[read_attempt & 1] = data_to_store;
-            std::memcpy(shared_memory + Offset, this, sizeof(*this));
-        }
-
-        // For reading we're just going to read the last stored value. If there was no value stored
-        // it will just end up reading an empty value as intended.
-        T ReadData(u8* shared_memory) {
-            std::memcpy(this, shared_memory + Offset, sizeof(*this));
-            return data[(read_attempt - 1) & 1];
-        }
-    };
-
     // Shared memory format
     struct Format {
-        MemoryBarrier<Clock::SteadyClockContext, 0x0> standard_steady_clock_timepoint;
-        MemoryBarrier<Clock::SystemClockContext, 0x38> standard_local_system_clock_context;
-        MemoryBarrier<Clock::SystemClockContext, 0x80> standard_network_system_clock_context;
-        MemoryBarrier<bool, 0xc8> standard_user_system_clock_automatic_correction;
-        u32_le format_version;
+        LockFreeAtomicType<Clock::StandardSteadyClockTimePointType> standard_steady_clock_timepoint;
+        LockFreeAtomicType<Clock::SystemClockContext> standard_local_system_clock_context;
+        LockFreeAtomicType<Clock::SystemClockContext> standard_network_system_clock_context;
+        LockFreeAtomicType<bool> is_standard_user_system_clock_automatic_correction_enabled;
+        u32 format_version;
     };
+    static_assert(offsetof(Format, standard_steady_clock_timepoint) == 0x0);
+    static_assert(offsetof(Format, standard_local_system_clock_context) == 0x38);
+    static_assert(offsetof(Format, standard_network_system_clock_context) == 0x80);
+    static_assert(offsetof(Format, is_standard_user_system_clock_automatic_correction_enabled) ==
+                  0xc8);
     static_assert(sizeof(Format) == 0xd8, "Format is an invalid size");
 
     void SetupStandardSteadyClock(const Common::UUID& clock_source_id,
@@ -56,10 +79,10 @@ public:
     void UpdateLocalSystemClockContext(const Clock::SystemClockContext& context);
     void UpdateNetworkSystemClockContext(const Clock::SystemClockContext& context);
     void SetAutomaticCorrectionEnabled(bool is_enabled);
+    Format* GetFormat();
 
 private:
     Core::System& system;
-    Format shared_memory_format{};
 };
 
 } // namespace Service::Time

src/input_common/CMakeLists.txt

@@ -20,6 +20,8 @@ add_library(input_common STATIC
     drivers/udp_client.h
     drivers/virtual_amiibo.cpp
     drivers/virtual_amiibo.h
+    drivers/virtual_gamepad.cpp
+    drivers/virtual_gamepad.h
     helpers/stick_from_buttons.cpp
     helpers/stick_from_buttons.h
     helpers/touch_from_buttons.cpp

src/input_common/drivers/camera.cpp

@@ -17,7 +17,7 @@ Camera::Camera(std::string input_engine_) : InputEngine(std::move(input_engine_)
     PreSetController(identifier);
 }
 
-void Camera::SetCameraData(std::size_t width, std::size_t height, std::vector<u32> data) {
+void Camera::SetCameraData(std::size_t width, std::size_t height, std::span<const u32> data) {
     const std::size_t desired_width = getImageWidth();
     const std::size_t desired_height = getImageHeight();
     status.data.resize(desired_width * desired_height);

src/input_common/drivers/camera.h

@@ -3,6 +3,8 @@
 
 #pragma once
 
+#include <span>
+
 #include "input_common/input_engine.h"
 
 namespace InputCommon {
@@ -15,7 +17,7 @@ class Camera final : public InputEngine {
 public:
     explicit Camera(std::string input_engine_);
 
-    void SetCameraData(std::size_t width, std::size_t height, std::vector<u32> data);
+    void SetCameraData(std::size_t width, std::size_t height, std::span<const u32> data);
 
     std::size_t getImageWidth() const;
     std::size_t getImageHeight() const;
@@ -23,6 +25,7 @@ public:
     Common::Input::CameraError SetCameraFormat(const PadIdentifier& identifier_,
                                                Common::Input::CameraFormat camera_format) override;
 
+private:
     Common::Input::CameraStatus status{};
 };
 

src/input_common/drivers/sdl_driver.cpp

@@ -16,6 +16,8 @@ Common::UUID GetGUID(SDL_Joystick* joystick) {
     const SDL_JoystickGUID guid = SDL_JoystickGetGUID(joystick);
     std::array<u8, 16> data{};
     std::memcpy(data.data(), guid.data, sizeof(data));
+    // Clear controller name crc
+    std::memset(data.data() + 2, 0, sizeof(u16));
     return Common::UUID{data};
 }
 } // Anonymous namespace

src/input_common/drivers/virtual_gamepad.cpp

@@ -0,0 +1,78 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "input_common/drivers/virtual_gamepad.h"
+
+namespace InputCommon {
+constexpr std::size_t PlayerIndexCount = 10;
+
+VirtualGamepad::VirtualGamepad(std::string input_engine_) : InputEngine(std::move(input_engine_)) {
+    for (std::size_t i = 0; i < PlayerIndexCount; i++) {
+        PreSetController(GetIdentifier(i));
+    }
+}
+
+void VirtualGamepad::SetButtonState(std::size_t player_index, int button_id, bool value) {
+    if (player_index > PlayerIndexCount) {
+        return;
+    }
+    const auto identifier = GetIdentifier(player_index);
+    SetButton(identifier, button_id, value);
+}
+
+void VirtualGamepad::SetButtonState(std::size_t player_index, VirtualButton button_id, bool value) {
+    SetButtonState(player_index, static_cast<int>(button_id), value);
+}
+
+void VirtualGamepad::SetStickPosition(std::size_t player_index, int axis_id, float x_value,
+                                      float y_value) {
+    if (player_index > PlayerIndexCount) {
+        return;
+    }
+    const auto identifier = GetIdentifier(player_index);
+    SetAxis(identifier, axis_id * 2, x_value);
+    SetAxis(identifier, (axis_id * 2) + 1, y_value);
+}
+
+void VirtualGamepad::SetStickPosition(std::size_t player_index, VirtualStick axis_id, float x_value,
+                                      float y_value) {
+    SetStickPosition(player_index, static_cast<int>(axis_id), x_value, y_value);
+}
+
+void VirtualGamepad::ResetControllers() {
+    for (std::size_t i = 0; i < PlayerIndexCount; i++) {
+        SetStickPosition(i, VirtualStick::Left, 0.0f, 0.0f);
+        SetStickPosition(i, VirtualStick::Right, 0.0f, 0.0f);
+
+        SetButtonState(i, VirtualButton::ButtonA, false);
+        SetButtonState(i, VirtualButton::ButtonB, false);
+        SetButtonState(i, VirtualButton::ButtonX, false);
+        SetButtonState(i, VirtualButton::ButtonY, false);
+        SetButtonState(i, VirtualButton::StickL, false);
+        SetButtonState(i, VirtualButton::StickR, false);
+        SetButtonState(i, VirtualButton::TriggerL, false);
+        SetButtonState(i, VirtualButton::TriggerR, false);
+        SetButtonState(i, VirtualButton::TriggerZL, false);
+        SetButtonState(i, VirtualButton::TriggerZR, false);
+        SetButtonState(i, VirtualButton::ButtonPlus, false);
+        SetButtonState(i, VirtualButton::ButtonMinus, false);
+        SetButtonState(i, VirtualButton::ButtonLeft, false);
+        SetButtonState(i, VirtualButton::ButtonUp, false);
+        SetButtonState(i, VirtualButton::ButtonRight, false);
+        SetButtonState(i, VirtualButton::ButtonDown, false);
+        SetButtonState(i, VirtualButton::ButtonSL, false);
+        SetButtonState(i, VirtualButton::ButtonSR, false);
+        SetButtonState(i, VirtualButton::ButtonHome, false);
+        SetButtonState(i, VirtualButton::ButtonCapture, false);
+    }
+}
+
+PadIdentifier VirtualGamepad::GetIdentifier(std::size_t player_index) const {
+    return {
+        .guid = Common::UUID{},
+        .port = player_index,
+        .pad = 0,
+    };
+}
+
+} // namespace InputCommon

src/input_common/drivers/virtual_gamepad.h

@@ -0,0 +1,73 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "input_common/input_engine.h"
+
+namespace InputCommon {
+
+/**
+ * A virtual controller that is always assigned to the game input
+ */
+class VirtualGamepad final : public InputEngine {
+public:
+    enum class VirtualButton {
+        ButtonA,
+        ButtonB,
+        ButtonX,
+        ButtonY,
+        StickL,
+        StickR,
+        TriggerL,
+        TriggerR,
+        TriggerZL,
+        TriggerZR,
+        ButtonPlus,
+        ButtonMinus,
+        ButtonLeft,
+        ButtonUp,
+        ButtonRight,
+        ButtonDown,
+        ButtonSL,
+        ButtonSR,
+        ButtonHome,
+        ButtonCapture,
+    };
+
+    enum class VirtualStick {
+        Left = 0,
+        Right = 1,
+    };
+
+    explicit VirtualGamepad(std::string input_engine_);
+
+    /**
+     * Sets the status of all buttons bound with the key to pressed
+     * @param player_index the player number that will take this action
+     * @param button_id the id of the button
+     * @param value indicates if the button is pressed or not
+     */
+    void SetButtonState(std::size_t player_index, int button_id, bool value);
+    void SetButtonState(std::size_t player_index, VirtualButton button_id, bool value);
+
+    /**
+     * Sets the status of all buttons bound with the key to released
+     * @param player_index the player number that will take this action
+     * @param axis_id the id of the axis to move
+     * @param x_value the position of the stick in the x axis
+     * @param y_value the position of the stick in the y axis
+     */
+    void SetStickPosition(std::size_t player_index, int axis_id, float x_value, float y_value);
+    void SetStickPosition(std::size_t player_index, VirtualStick axis_id, float x_value,
+                          float y_value);
+
+    /// Restores all inputs into the neutral position
+    void ResetControllers();
+
+private:
+    /// Returns the correct identifier corresponding to the player index
+    PadIdentifier GetIdentifier(std::size_t player_index) const;
+};
+
+} // namespace InputCommon

src/input_common/input_mapping.cpp

@@ -200,12 +200,6 @@ bool MappingFactory::IsDriverValid(const MappingData& data) const {
         return false;
     }
     // The following drivers don't need to be mapped
-    if (data.engine == "tas") {
-        return false;
-    }
-    if (data.engine == "touch") {
-        return false;
-    }
     if (data.engine == "touch_from_button") {
         return false;
     }

src/input_common/main.cpp

@@ -12,6 +12,7 @@
 #include "input_common/drivers/touch_screen.h"
 #include "input_common/drivers/udp_client.h"
 #include "input_common/drivers/virtual_amiibo.h"
+#include "input_common/drivers/virtual_gamepad.h"
 #include "input_common/helpers/stick_from_buttons.h"
 #include "input_common/helpers/touch_from_buttons.h"
 #include "input_common/input_engine.h"
@@ -25,73 +26,33 @@
 namespace InputCommon {
 
 struct InputSubsystem::Impl {
-    void Initialize() {
-        mapping_factory = std::make_shared<MappingFactory>();
+    template <typename Engine>
+    void RegisterEngine(std::string name, std::shared_ptr<Engine>& engine) {
         MappingCallback mapping_callback{[this](const MappingData& data) { RegisterInput(data); }};
 
-        keyboard = std::make_shared<Keyboard>("keyboard");
-        keyboard->SetMappingCallback(mapping_callback);
-        keyboard_factory = std::make_shared<InputFactory>(keyboard);
-        keyboard_output_factory = std::make_shared<OutputFactory>(keyboard);
-        Common::Input::RegisterInputFactory(keyboard->GetEngineName(), keyboard_factory);
-        Common::Input::RegisterOutputFactory(keyboard->GetEngineName(), keyboard_output_factory);
+        engine = std::make_shared<Engine>(name);
+        engine->SetMappingCallback(mapping_callback);
 
-        mouse = std::make_shared<Mouse>("mouse");
-        mouse->SetMappingCallback(mapping_callback);
-        mouse_factory = std::make_shared<InputFactory>(mouse);
-        mouse_output_factory = std::make_shared<OutputFactory>(mouse);
-        Common::Input::RegisterInputFactory(mouse->GetEngineName(), mouse_factory);
-        Common::Input::RegisterOutputFactory(mouse->GetEngineName(), mouse_output_factory);
+        std::shared_ptr<InputFactory> input_factory = std::make_shared<InputFactory>(engine);
+        std::shared_ptr<OutputFactory> output_factory = std::make_shared<OutputFactory>(engine);
+        Common::Input::RegisterInputFactory(engine->GetEngineName(), std::move(input_factory));
+        Common::Input::RegisterOutputFactory(engine->GetEngineName(), std::move(output_factory));
+    }
 
-        touch_screen = std::make_shared<TouchScreen>("touch");
-        touch_screen_factory = std::make_shared<InputFactory>(touch_screen);
-        Common::Input::RegisterInputFactory(touch_screen->GetEngineName(), touch_screen_factory);
-
-        gcadapter = std::make_shared<GCAdapter>("gcpad");
-        gcadapter->SetMappingCallback(mapping_callback);
-        gcadapter_input_factory = std::make_shared<InputFactory>(gcadapter);
-        gcadapter_output_factory = std::make_shared<OutputFactory>(gcadapter);
-        Common::Input::RegisterInputFactory(gcadapter->GetEngineName(), gcadapter_input_factory);
-        Common::Input::RegisterOutputFactory(gcadapter->GetEngineName(), gcadapter_output_factory);
-
-        udp_client = std::make_shared<CemuhookUDP::UDPClient>("cemuhookudp");
-        udp_client->SetMappingCallback(mapping_callback);
-        udp_client_input_factory = std::make_shared<InputFactory>(udp_client);
-        udp_client_output_factory = std::make_shared<OutputFactory>(udp_client);
-        Common::Input::RegisterInputFactory(udp_client->GetEngineName(), udp_client_input_factory);
-        Common::Input::RegisterOutputFactory(udp_client->GetEngineName(),
-                                             udp_client_output_factory);
-
-        tas_input = std::make_shared<TasInput::Tas>("tas");
-        tas_input->SetMappingCallback(mapping_callback);
-        tas_input_factory = std::make_shared<InputFactory>(tas_input);
-        tas_output_factory = std::make_shared<OutputFactory>(tas_input);
-        Common::Input::RegisterInputFactory(tas_input->GetEngineName(), tas_input_factory);
-        Common::Input::RegisterOutputFactory(tas_input->GetEngineName(), tas_output_factory);
-
-        camera = std::make_shared<Camera>("camera");
-        camera->SetMappingCallback(mapping_callback);
-        camera_input_factory = std::make_shared<InputFactory>(camera);
-        camera_output_factory = std::make_shared<OutputFactory>(camera);
-        Common::Input::RegisterInputFactory(camera->GetEngineName(), camera_input_factory);
-        Common::Input::RegisterOutputFactory(camera->GetEngineName(), camera_output_factory);
-
-        virtual_amiibo = std::make_shared<VirtualAmiibo>("virtual_amiibo");
-        virtual_amiibo->SetMappingCallback(mapping_callback);
-        virtual_amiibo_input_factory = std::make_shared<InputFactory>(virtual_amiibo);
-        virtual_amiibo_output_factory = std::make_shared<OutputFactory>(virtual_amiibo);
-        Common::Input::RegisterInputFactory(virtual_amiibo->GetEngineName(),
-                                            virtual_amiibo_input_factory);
-        Common::Input::RegisterOutputFactory(virtual_amiibo->GetEngineName(),
-                                             virtual_amiibo_output_factory);
+    void Initialize() {
+        mapping_factory = std::make_shared<MappingFactory>();
 
+        RegisterEngine("keyboard", keyboard);
+        RegisterEngine("mouse", mouse);
+        RegisterEngine("touch", touch_screen);
+        RegisterEngine("gcpad", gcadapter);
+        RegisterEngine("cemuhookudp", udp_client);
+        RegisterEngine("tas", tas_input);
+        RegisterEngine("camera", camera);
+        RegisterEngine("virtual_amiibo", virtual_amiibo);
+        RegisterEngine("virtual_gamepad", virtual_gamepad);
 #ifdef HAVE_SDL2
-        sdl = std::make_shared<SDLDriver>("sdl");
-        sdl->SetMappingCallback(mapping_callback);
-        sdl_input_factory = std::make_shared<InputFactory>(sdl);
-        sdl_output_factory = std::make_shared<OutputFactory>(sdl);
-        Common::Input::RegisterInputFactory(sdl->GetEngineName(), sdl_input_factory);
-        Common::Input::RegisterOutputFactory(sdl->GetEngineName(), sdl_output_factory);
+        RegisterEngine("sdl", sdl);
 #endif
 
         Common::Input::RegisterInputFactory("touch_from_button",
@@ -100,42 +61,25 @@ struct InputSubsystem::Impl {
                                             std::make_shared<StickFromButton>());
     }
 
+    template <typename Engine>
+    void UnregisterEngine(std::shared_ptr<Engine>& engine) {
+        Common::Input::UnregisterInputFactory(engine->GetEngineName());
+        Common::Input::UnregisterOutputFactory(engine->GetEngineName());
+        engine.reset();
+    }
+
     void Shutdown() {
-        Common::Input::UnregisterInputFactory(keyboard->GetEngineName());
-        Common::Input::UnregisterOutputFactory(keyboard->GetEngineName());
-        keyboard.reset();
-
-        Common::Input::UnregisterInputFactory(mouse->GetEngineName());
-        Common::Input::UnregisterOutputFactory(mouse->GetEngineName());
-        mouse.reset();
-
-        Common::Input::UnregisterInputFactory(touch_screen->GetEngineName());
-        touch_screen.reset();
-
-        Common::Input::UnregisterInputFactory(gcadapter->GetEngineName());
-        Common::Input::UnregisterOutputFactory(gcadapter->GetEngineName());
-        gcadapter.reset();
-
-        Common::Input::UnregisterInputFactory(udp_client->GetEngineName());
-        Common::Input::UnregisterOutputFactory(udp_client->GetEngineName());
-        udp_client.reset();
-
-        Common::Input::UnregisterInputFactory(tas_input->GetEngineName());
-        Common::Input::UnregisterOutputFactory(tas_input->GetEngineName());
-        tas_input.reset();
-
-        Common::Input::UnregisterInputFactory(camera->GetEngineName());
-        Common::Input::UnregisterOutputFactory(camera->GetEngineName());
-        camera.reset();
-
-        Common::Input::UnregisterInputFactory(virtual_amiibo->GetEngineName());
-        Common::Input::UnregisterOutputFactory(virtual_amiibo->GetEngineName());
-        virtual_amiibo.reset();
-
+        UnregisterEngine(keyboard);
+        UnregisterEngine(mouse);
+        UnregisterEngine(touch_screen);
+        UnregisterEngine(gcadapter);
+        UnregisterEngine(udp_client);
+        UnregisterEngine(tas_input);
+        UnregisterEngine(camera);
+        UnregisterEngine(virtual_amiibo);
+        UnregisterEngine(virtual_gamepad);
 #ifdef HAVE_SDL2
-        Common::Input::UnregisterInputFactory(sdl->GetEngineName());
-        Common::Input::UnregisterOutputFactory(sdl->GetEngineName());
-        sdl.reset();
+        UnregisterEngine(sdl);
 #endif
 
         Common::Input::UnregisterInputFactory("touch_from_button");
@@ -163,117 +107,86 @@ struct InputSubsystem::Impl {
         return devices;
     }
 
-    [[nodiscard]] AnalogMapping GetAnalogMappingForDevice(
+    [[nodiscard]] std::shared_ptr<InputEngine> GetInputEngine(
         const Common::ParamPackage& params) const {
         if (!params.Has("engine") || params.Get("engine", "") == "any") {
-            return {};
+            return nullptr;
         }
         const std::string engine = params.Get("engine", "");
+        if (engine == keyboard->GetEngineName()) {
+            return keyboard;
+        }
         if (engine == mouse->GetEngineName()) {
-            return mouse->GetAnalogMappingForDevice(params);
+            return mouse;
         }
         if (engine == gcadapter->GetEngineName()) {
-            return gcadapter->GetAnalogMappingForDevice(params);
+            return gcadapter;
         }
         if (engine == udp_client->GetEngineName()) {
-            return udp_client->GetAnalogMappingForDevice(params);
-        }
-        if (engine == tas_input->GetEngineName()) {
-            return tas_input->GetAnalogMappingForDevice(params);
+            return udp_client;
         }
 #ifdef HAVE_SDL2
         if (engine == sdl->GetEngineName()) {
-            return sdl->GetAnalogMappingForDevice(params);
+            return sdl;
         }
 #endif
-        return {};
+        return nullptr;
+    }
+
+    [[nodiscard]] AnalogMapping GetAnalogMappingForDevice(
+        const Common::ParamPackage& params) const {
+        const auto input_engine = GetInputEngine(params);
+
+        if (input_engine == nullptr) {
+            return {};
+        }
+
+        return input_engine->GetAnalogMappingForDevice(params);
     }
 
     [[nodiscard]] ButtonMapping GetButtonMappingForDevice(
         const Common::ParamPackage& params) const {
-        if (!params.Has("engine") || params.Get("engine", "") == "any") {
+        const auto input_engine = GetInputEngine(params);
+
+        if (input_engine == nullptr) {
             return {};
         }
-        const std::string engine = params.Get("engine", "");
-        if (engine == gcadapter->GetEngineName()) {
-            return gcadapter->GetButtonMappingForDevice(params);
-        }
-        if (engine == udp_client->GetEngineName()) {
-            return udp_client->GetButtonMappingForDevice(params);
-        }
-        if (engine == tas_input->GetEngineName()) {
-            return tas_input->GetButtonMappingForDevice(params);
-        }
-#ifdef HAVE_SDL2
-        if (engine == sdl->GetEngineName()) {
-            return sdl->GetButtonMappingForDevice(params);
-        }
-#endif
-        return {};
+
+        return input_engine->GetButtonMappingForDevice(params);
     }
 
     [[nodiscard]] MotionMapping GetMotionMappingForDevice(
         const Common::ParamPackage& params) const {
-        if (!params.Has("engine") || params.Get("engine", "") == "any") {
+        const auto input_engine = GetInputEngine(params);
+
+        if (input_engine == nullptr) {
             return {};
         }
-        const std::string engine = params.Get("engine", "");
-        if (engine == udp_client->GetEngineName()) {
-            return udp_client->GetMotionMappingForDevice(params);
-        }
-#ifdef HAVE_SDL2
-        if (engine == sdl->GetEngineName()) {
-            return sdl->GetMotionMappingForDevice(params);
-        }
-#endif
-        return {};
+
+        return input_engine->GetMotionMappingForDevice(params);
     }
 
     Common::Input::ButtonNames GetButtonName(const Common::ParamPackage& params) const {
         if (!params.Has("engine") || params.Get("engine", "") == "any") {
             return Common::Input::ButtonNames::Undefined;
         }
-        const std::string engine = params.Get("engine", "");
-        if (engine == mouse->GetEngineName()) {
-            return mouse->GetUIName(params);
+        const auto input_engine = GetInputEngine(params);
+
+        if (input_engine == nullptr) {
+            return Common::Input::ButtonNames::Invalid;
         }
-        if (engine == gcadapter->GetEngineName()) {
-            return gcadapter->GetUIName(params);
-        }
-        if (engine == udp_client->GetEngineName()) {
-            return udp_client->GetUIName(params);
-        }
-        if (engine == tas_input->GetEngineName()) {
-            return tas_input->GetUIName(params);
-        }
-#ifdef HAVE_SDL2
-        if (engine == sdl->GetEngineName()) {
-            return sdl->GetUIName(params);
-        }
-#endif
-        return Common::Input::ButtonNames::Invalid;
+
+        return input_engine->GetUIName(params);
     }
 
     bool IsStickInverted(const Common::ParamPackage& params) {
-        const std::string engine = params.Get("engine", "");
-        if (engine == mouse->GetEngineName()) {
-            return mouse->IsStickInverted(params);
+        const auto input_engine = GetInputEngine(params);
+
+        if (input_engine == nullptr) {
+            return false;
         }
-        if (engine == gcadapter->GetEngineName()) {
-            return gcadapter->IsStickInverted(params);
-        }
-        if (engine == udp_client->GetEngineName()) {
-            return udp_client->IsStickInverted(params);
-        }
-        if (engine == tas_input->GetEngineName()) {
-            return tas_input->IsStickInverted(params);
-        }
-#ifdef HAVE_SDL2
-        if (engine == sdl->GetEngineName()) {
-            return sdl->IsStickInverted(params);
-        }
-#endif
-        return false;
+
+        return input_engine->IsStickInverted(params);
     }
 
     bool IsController(const Common::ParamPackage& params) {
@@ -290,6 +203,9 @@ struct InputSubsystem::Impl {
         if (engine == tas_input->GetEngineName()) {
             return true;
         }
+        if (engine == virtual_gamepad->GetEngineName()) {
+            return true;
+        }
 #ifdef HAVE_SDL2
         if (engine == sdl->GetEngineName()) {
             return true;
@@ -338,28 +254,10 @@ struct InputSubsystem::Impl {
     std::shared_ptr<CemuhookUDP::UDPClient> udp_client;
     std::shared_ptr<Camera> camera;
     std::shared_ptr<VirtualAmiibo> virtual_amiibo;
-
-    std::shared_ptr<InputFactory> keyboard_factory;
-    std::shared_ptr<InputFactory> mouse_factory;
-    std::shared_ptr<InputFactory> gcadapter_input_factory;
-    std::shared_ptr<InputFactory> touch_screen_factory;
-    std::shared_ptr<InputFactory> udp_client_input_factory;
-    std::shared_ptr<InputFactory> tas_input_factory;
-    std::shared_ptr<InputFactory> camera_input_factory;
-    std::shared_ptr<InputFactory> virtual_amiibo_input_factory;
-
-    std::shared_ptr<OutputFactory> keyboard_output_factory;
-    std::shared_ptr<OutputFactory> mouse_output_factory;
-    std::shared_ptr<OutputFactory> gcadapter_output_factory;
-    std::shared_ptr<OutputFactory> udp_client_output_factory;
-    std::shared_ptr<OutputFactory> tas_output_factory;
-    std::shared_ptr<OutputFactory> camera_output_factory;
-    std::shared_ptr<OutputFactory> virtual_amiibo_output_factory;
+    std::shared_ptr<VirtualGamepad> virtual_gamepad;
 
 #ifdef HAVE_SDL2
     std::shared_ptr<SDLDriver> sdl;
-    std::shared_ptr<InputFactory> sdl_input_factory;
-    std::shared_ptr<OutputFactory> sdl_output_factory;
 #endif
 };
 
@@ -423,6 +321,14 @@ const VirtualAmiibo* InputSubsystem::GetVirtualAmiibo() const {
     return impl->virtual_amiibo.get();
 }
 
+VirtualGamepad* InputSubsystem::GetVirtualGamepad() {
+    return impl->virtual_gamepad.get();
+}
+
+const VirtualGamepad* InputSubsystem::GetVirtualGamepad() const {
+    return impl->virtual_gamepad.get();
+}
+
 std::vector<Common::ParamPackage> InputSubsystem::GetInputDevices() const {
     return impl->GetInputDevices();
 }

src/input_common/main.h

@@ -34,6 +34,7 @@ class Keyboard;
 class Mouse;
 class TouchScreen;
 class VirtualAmiibo;
+class VirtualGamepad;
 struct MappingData;
 } // namespace InputCommon
 
@@ -108,6 +109,12 @@ public:
     /// Retrieves the underlying virtual amiibo input device.
     [[nodiscard]] const VirtualAmiibo* GetVirtualAmiibo() const;
 
+    /// Retrieves the underlying virtual gamepad input device.
+    [[nodiscard]] VirtualGamepad* GetVirtualGamepad();
+
+    /// Retrieves the underlying virtual gamepad input device.
+    [[nodiscard]] const VirtualGamepad* GetVirtualGamepad() 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 `engine` field

src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp

@@ -461,7 +461,7 @@ void EmitSetSampleMask(EmitContext& ctx, Id value) {
 }
 
 void EmitSetFragDepth(EmitContext& ctx, Id value) {
-    if (!ctx.runtime_info.convert_depth_mode) {
+    if (!ctx.runtime_info.convert_depth_mode || ctx.profile.support_native_ndc) {
         ctx.OpStore(ctx.frag_depth, value);
         return;
     }

src/shader_recompiler/backend/spirv/emit_spirv_special.cpp

@@ -116,7 +116,8 @@ void EmitPrologue(EmitContext& ctx) {
 }
 
 void EmitEpilogue(EmitContext& ctx) {
-    if (ctx.stage == Stage::VertexB && ctx.runtime_info.convert_depth_mode) {
+    if (ctx.stage == Stage::VertexB && ctx.runtime_info.convert_depth_mode &&
+        !ctx.profile.support_native_ndc) {
         ConvertDepthMode(ctx);
     }
     if (ctx.stage == Stage::Fragment) {
@@ -125,7 +126,7 @@ void EmitEpilogue(EmitContext& ctx) {
 }
 
 void EmitEmitVertex(EmitContext& ctx, const IR::Value& stream) {
-    if (ctx.runtime_info.convert_depth_mode) {
+    if (ctx.runtime_info.convert_depth_mode && !ctx.profile.support_native_ndc) {
         ConvertDepthMode(ctx);
     }
     if (stream.IsImmediate()) {

src/shader_recompiler/backend/spirv/spirv_emit_context.cpp

@@ -1345,8 +1345,10 @@ void EmitContext::DefineInputs(const IR::Program& program) {
     if (info.uses_fswzadd || info.uses_subgroup_invocation_id || info.uses_subgroup_shuffles ||
         (profile.warp_size_potentially_larger_than_guest &&
          (info.uses_subgroup_vote || info.uses_subgroup_mask))) {
+        AddCapability(spv::Capability::GroupNonUniform);
         subgroup_local_invocation_id =
             DefineInput(*this, U32[1], false, spv::BuiltIn::SubgroupLocalInvocationId);
+        Decorate(subgroup_local_invocation_id, spv::Decoration::Flat);
     }
     if (info.uses_fswzadd) {
         const Id f32_one{Const(1.0f)};

src/shader_recompiler/profile.h

@@ -35,6 +35,7 @@ struct Profile {
     bool support_int64_atomics{};
     bool support_derivative_control{};
     bool support_geometry_shader_passthrough{};
+    bool support_native_ndc{};
     bool support_gl_nv_gpu_shader_5{};
     bool support_gl_amd_gpu_shader_half_float{};
     bool support_gl_texture_shadow_lod{};

src/tests/CMakeLists.txt

@@ -8,6 +8,7 @@ add_executable(tests
     common/host_memory.cpp
     common/param_package.cpp
     common/ring_buffer.cpp
+    common/scratch_buffer.cpp
     common/unique_function.cpp
     core/core_timing.cpp
     core/internal_network/network.cpp

src/tests/common/scratch_buffer.cpp

@@ -0,0 +1,200 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <algorithm>
+#include <array>
+#include <cstring>
+#include <span>
+#include <catch2/catch.hpp>
+#include "common/common_types.h"
+#include "common/scratch_buffer.h"
+
+namespace Common {
+
+TEST_CASE("ScratchBuffer: Basic Test", "[common]") {
+    ScratchBuffer<u8> buf;
+
+    REQUIRE(buf.size() == 0U);
+    REQUIRE(buf.capacity() == 0U);
+
+    std::array<u8, 10> payload;
+    payload.fill(66);
+
+    buf.resize(payload.size());
+    REQUIRE(buf.size() == payload.size());
+    REQUIRE(buf.capacity() == payload.size());
+
+    std::memcpy(buf.data(), payload.data(), payload.size());
+    for (size_t i = 0; i < payload.size(); ++i) {
+        REQUIRE(buf[i] == payload[i]);
+    }
+}
+
+TEST_CASE("ScratchBuffer: resize_destructive Grow", "[common]") {
+    std::array<u8, 10> payload;
+    payload.fill(66);
+
+    ScratchBuffer<u8> buf(payload.size());
+    REQUIRE(buf.size() == payload.size());
+    REQUIRE(buf.capacity() == payload.size());
+
+    // Increasing the size should reallocate the buffer
+    buf.resize_destructive(payload.size() * 2);
+    REQUIRE(buf.size() == payload.size() * 2);
+    REQUIRE(buf.capacity() == payload.size() * 2);
+
+    // Since the buffer is not value initialized, reading its data will be garbage
+}
+
+TEST_CASE("ScratchBuffer: resize_destructive Shrink", "[common]") {
+    std::array<u8, 10> payload;
+    payload.fill(66);
+
+    ScratchBuffer<u8> buf(payload.size());
+    REQUIRE(buf.size() == payload.size());
+    REQUIRE(buf.capacity() == payload.size());
+
+    std::memcpy(buf.data(), payload.data(), payload.size());
+    for (size_t i = 0; i < payload.size(); ++i) {
+        REQUIRE(buf[i] == payload[i]);
+    }
+
+    // Decreasing the size should not cause a buffer reallocation
+    // This can be tested by ensuring the buffer capacity and data has not changed,
+    buf.resize_destructive(1U);
+    REQUIRE(buf.size() == 1U);
+    REQUIRE(buf.capacity() == payload.size());
+
+    for (size_t i = 0; i < payload.size(); ++i) {
+        REQUIRE(buf[i] == payload[i]);
+    }
+}
+
+TEST_CASE("ScratchBuffer: resize Grow u8", "[common]") {
+    std::array<u8, 10> payload;
+    payload.fill(66);
+
+    ScratchBuffer<u8> buf(payload.size());
+    REQUIRE(buf.size() == payload.size());
+    REQUIRE(buf.capacity() == payload.size());
+
+    std::memcpy(buf.data(), payload.data(), payload.size());
+    for (size_t i = 0; i < payload.size(); ++i) {
+        REQUIRE(buf[i] == payload[i]);
+    }
+
+    // Increasing the size should reallocate the buffer
+    buf.resize(payload.size() * 2);
+    REQUIRE(buf.size() == payload.size() * 2);
+    REQUIRE(buf.capacity() == payload.size() * 2);
+
+    // resize() keeps the previous data intact
+    for (size_t i = 0; i < payload.size(); ++i) {
+        REQUIRE(buf[i] == payload[i]);
+    }
+}
+
+TEST_CASE("ScratchBuffer: resize Grow u64", "[common]") {
+    std::array<u64, 10> payload;
+    payload.fill(6666);
+
+    ScratchBuffer<u64> buf(payload.size());
+    REQUIRE(buf.size() == payload.size());
+    REQUIRE(buf.capacity() == payload.size());
+
+    std::memcpy(buf.data(), payload.data(), payload.size() * sizeof(u64));
+    for (size_t i = 0; i < payload.size(); ++i) {
+        REQUIRE(buf[i] == payload[i]);
+    }
+
+    // Increasing the size should reallocate the buffer
+    buf.resize(payload.size() * 2);
+    REQUIRE(buf.size() == payload.size() * 2);
+    REQUIRE(buf.capacity() == payload.size() * 2);
+
+    // resize() keeps the previous data intact
+    for (size_t i = 0; i < payload.size(); ++i) {
+        REQUIRE(buf[i] == payload[i]);
+    }
+}
+
+TEST_CASE("ScratchBuffer: resize Shrink", "[common]") {
+    std::array<u8, 10> payload;
+    payload.fill(66);
+
+    ScratchBuffer<u8> buf(payload.size());
+    REQUIRE(buf.size() == payload.size());
+    REQUIRE(buf.capacity() == payload.size());
+
+    std::memcpy(buf.data(), payload.data(), payload.size());
+    for (size_t i = 0; i < payload.size(); ++i) {
+        REQUIRE(buf[i] == payload[i]);
+    }
+
+    // Decreasing the size should not cause a buffer reallocation
+    // This can be tested by ensuring the buffer capacity and data has not changed,
+    buf.resize(1U);
+    REQUIRE(buf.size() == 1U);
+    REQUIRE(buf.capacity() == payload.size());
+
+    for (size_t i = 0; i < payload.size(); ++i) {
+        REQUIRE(buf[i] == payload[i]);
+    }
+}
+
+TEST_CASE("ScratchBuffer: Span Size", "[common]") {
+    std::array<u8, 10> payload;
+    payload.fill(66);
+
+    ScratchBuffer<u8> buf(payload.size());
+    REQUIRE(buf.size() == payload.size());
+    REQUIRE(buf.capacity() == payload.size());
+
+    std::memcpy(buf.data(), payload.data(), payload.size());
+    for (size_t i = 0; i < payload.size(); ++i) {
+        REQUIRE(buf[i] == payload[i]);
+    }
+
+    buf.resize(3U);
+    REQUIRE(buf.size() == 3U);
+    REQUIRE(buf.capacity() == payload.size());
+
+    const auto buf_span = std::span<u8>(buf);
+    // The span size is the last requested size of the buffer, not its capacity
+    REQUIRE(buf_span.size() == buf.size());
+
+    for (size_t i = 0; i < buf_span.size(); ++i) {
+        REQUIRE(buf_span[i] == buf[i]);
+        REQUIRE(buf_span[i] == payload[i]);
+    }
+}
+
+TEST_CASE("ScratchBuffer: Span Writes", "[common]") {
+    std::array<u8, 10> payload;
+    payload.fill(66);
+
+    ScratchBuffer<u8> buf(payload.size());
+    REQUIRE(buf.size() == payload.size());
+    REQUIRE(buf.capacity() == payload.size());
+
+    std::memcpy(buf.data(), payload.data(), payload.size());
+    for (size_t i = 0; i < payload.size(); ++i) {
+        REQUIRE(buf[i] == payload[i]);
+    }
+
+    buf.resize(3U);
+    REQUIRE(buf.size() == 3U);
+    REQUIRE(buf.capacity() == payload.size());
+
+    const auto buf_span = std::span<u8>(buf);
+    REQUIRE(buf_span.size() == buf.size());
+
+    for (size_t i = 0; i < buf_span.size(); ++i) {
+        const auto new_value = static_cast<u8>(i + 1U);
+        // Writes to a span of the scratch buffer will propogate to the buffer itself
+        buf_span[i] = new_value;
+        REQUIRE(buf[i] == new_value);
+    }
+}
+
+} // namespace Common

src/video_core/buffer_cache/buffer_cache.h

@@ -20,6 +20,7 @@
 #include "common/lru_cache.h"
 #include "common/microprofile.h"
 #include "common/polyfill_ranges.h"
+#include "common/scratch_buffer.h"
 #include "common/settings.h"
 #include "core/memory.h"
 #include "video_core/buffer_cache/buffer_base.h"
@@ -422,8 +423,7 @@ private:
     IntervalSet common_ranges;
     std::deque<IntervalSet> committed_ranges;
 
-    size_t immediate_buffer_capacity = 0;
-    std::unique_ptr<u8[]> immediate_buffer_alloc;
+    Common::ScratchBuffer<u8> immediate_buffer_alloc;
 
     struct LRUItemParams {
         using ObjectType = BufferId;
@@ -1927,11 +1927,8 @@ std::span<const u8> BufferCache<P>::ImmediateBufferWithData(VAddr cpu_addr, size
 
 template <class P>
 std::span<u8> BufferCache<P>::ImmediateBuffer(size_t wanted_capacity) {
-    if (wanted_capacity > immediate_buffer_capacity) {
-        immediate_buffer_capacity = wanted_capacity;
-        immediate_buffer_alloc = std::make_unique<u8[]>(wanted_capacity);
-    }
-    return std::span<u8>(immediate_buffer_alloc.get(), wanted_capacity);
+    immediate_buffer_alloc.resize_destructive(wanted_capacity);
+    return std::span<u8>(immediate_buffer_alloc.data(), wanted_capacity);
 }
 
 template <class P>

src/video_core/dma_pusher.cpp

@@ -56,7 +56,7 @@ bool DmaPusher::Step() {
 
     if (command_list.prefetch_command_list.size()) {
         // Prefetched command list from nvdrv, used for things like synchronization
-        command_headers = std::move(command_list.prefetch_command_list);
+        ProcessCommands(command_list.prefetch_command_list);
         dma_pushbuffer.pop();
     } else {
         const CommandListHeader command_list_header{
@@ -74,7 +74,7 @@ bool DmaPusher::Step() {
         }
 
         // Push buffer non-empty, read a word
-        command_headers.resize(command_list_header.size);
+        command_headers.resize_destructive(command_list_header.size);
         if (Settings::IsGPULevelHigh()) {
             memory_manager.ReadBlock(dma_get, command_headers.data(),
                                      command_list_header.size * sizeof(u32));
@@ -82,16 +82,21 @@ bool DmaPusher::Step() {
             memory_manager.ReadBlockUnsafe(dma_get, command_headers.data(),
                                            command_list_header.size * sizeof(u32));
         }
+        ProcessCommands(command_headers);
     }
-    for (std::size_t index = 0; index < command_headers.size();) {
-        const CommandHeader& command_header = command_headers[index];
+
+    return true;
+}
+
+void DmaPusher::ProcessCommands(std::span<const CommandHeader> commands) {
+    for (std::size_t index = 0; index < commands.size();) {
+        const CommandHeader& command_header = commands[index];
 
         if (dma_state.method_count) {
             // Data word of methods command
             if (dma_state.non_incrementing) {
                 const u32 max_write = static_cast<u32>(
-                    std::min<std::size_t>(index + dma_state.method_count, command_headers.size()) -
-                    index);
+                    std::min<std::size_t>(index + dma_state.method_count, commands.size()) - index);
                 CallMultiMethod(&command_header.argument, max_write);
                 dma_state.method_count -= max_write;
                 dma_state.is_last_call = true;
@@ -142,8 +147,6 @@ bool DmaPusher::Step() {
         }
         index++;
     }
-
-    return true;
 }
 
 void DmaPusher::SetState(const CommandHeader& command_header) {

src/video_core/dma_pusher.h

@@ -4,11 +4,13 @@
 #pragma once
 
 #include <array>
+#include <span>
 #include <vector>
 #include <queue>
 
 #include "common/bit_field.h"
 #include "common/common_types.h"
+#include "common/scratch_buffer.h"
 #include "video_core/engines/engine_interface.h"
 #include "video_core/engines/puller.h"
 
@@ -136,13 +138,15 @@ private:
     static constexpr u32 non_puller_methods = 0x40;
     static constexpr u32 max_subchannels = 8;
     bool Step();
+    void ProcessCommands(std::span<const CommandHeader> commands);
 
     void SetState(const CommandHeader& command_header);
 
     void CallMethod(u32 argument) const;
     void CallMultiMethod(const u32* base_start, u32 num_methods) const;
 
-    std::vector<CommandHeader> command_headers; ///< Buffer for list of commands fetched at once
+    Common::ScratchBuffer<CommandHeader>
+        command_headers; ///< Buffer for list of commands fetched at once
 
     std::queue<CommandList> dma_pushbuffer; ///< Queue of command lists to be processed
     std::size_t dma_pushbuffer_subindex{};  ///< Index within a command list within the pushbuffer
@@ -159,7 +163,7 @@ private:
     DmaState dma_state{};
     bool dma_increment_once{};
 
-    bool ib_enable{true}; ///< IB mode enabled
+    const bool ib_enable{true}; ///< IB mode enabled
 
     std::array<Engines::EngineInterface*, max_subchannels> subchannels{};
 

src/video_core/engines/draw_manager.cpp

@@ -46,21 +46,26 @@ void DrawManager::ProcessMethodCall(u32 method, u32 argument) {
         SetInlineIndexBuffer(regs.inline_index_4x8.index2);
         SetInlineIndexBuffer(regs.inline_index_4x8.index3);
         break;
-    case MAXWELL3D_REG_INDEX(topology_override):
-        use_topology_override = true;
+    case MAXWELL3D_REG_INDEX(vertex_array_instance_first):
+    case MAXWELL3D_REG_INDEX(vertex_array_instance_subsequent): {
+        LOG_WARNING(HW_GPU, "(STUBBED) called");
         break;
+    }
     default:
         break;
     }
 }
 
 void DrawManager::Clear(u32 layer_count) {
-    maxwell3d->rasterizer->Clear(layer_count);
+    if (maxwell3d->ShouldExecute()) {
+        maxwell3d->rasterizer->Clear(layer_count);
+    }
 }
 
 void DrawManager::DrawDeferred() {
-    if (draw_state.draw_mode != DrawMode::Instance || draw_state.instance_count == 0)
+    if (draw_state.draw_mode != DrawMode::Instance || draw_state.instance_count == 0) {
         return;
+    }
     DrawEnd(draw_state.instance_count + 1, true);
     draw_state.instance_count = 0;
 }
@@ -115,8 +120,9 @@ void DrawManager::DrawEnd(u32 instance_count, bool force_draw) {
     const auto& regs{maxwell3d->regs};
     switch (draw_state.draw_mode) {
     case DrawMode::Instance:
-        if (!force_draw)
+        if (!force_draw) {
             break;
+        }
         [[fallthrough]];
     case DrawMode::General:
         draw_state.base_instance = regs.global_base_instance_index;
@@ -156,25 +162,28 @@ void DrawManager::DrawIndexSmall(u32 argument) {
     ProcessDraw(true, 1);
 }
 
-void DrawManager::ProcessTopologyOverride() {
-    if (!use_topology_override)
-        return;
-
+void DrawManager::UpdateTopology() {
     const auto& regs{maxwell3d->regs};
-    switch (regs.topology_override) {
-    case PrimitiveTopologyOverride::None:
+    switch (regs.primitive_topology_control) {
+    case PrimitiveTopologyControl::UseInBeginMethods:
         break;
-    case PrimitiveTopologyOverride::Points:
-        draw_state.topology = PrimitiveTopology::Points;
-        break;
-    case PrimitiveTopologyOverride::Lines:
-        draw_state.topology = PrimitiveTopology::Lines;
-        break;
-    case PrimitiveTopologyOverride::LineStrip:
-        draw_state.topology = PrimitiveTopology::LineStrip;
-        break;
-    default:
-        draw_state.topology = static_cast<PrimitiveTopology>(regs.topology_override);
+    case PrimitiveTopologyControl::UseSeparateState:
+        switch (regs.topology_override) {
+        case PrimitiveTopologyOverride::None:
+            break;
+        case PrimitiveTopologyOverride::Points:
+            draw_state.topology = PrimitiveTopology::Points;
+            break;
+        case PrimitiveTopologyOverride::Lines:
+            draw_state.topology = PrimitiveTopology::Lines;
+            break;
+        case PrimitiveTopologyOverride::LineStrip:
+            draw_state.topology = PrimitiveTopology::LineStrip;
+            break;
+        default:
+            draw_state.topology = static_cast<PrimitiveTopology>(regs.topology_override);
+            break;
+        }
         break;
     }
 }
@@ -183,9 +192,10 @@ void DrawManager::ProcessDraw(bool draw_indexed, u32 instance_count) {
     LOG_TRACE(HW_GPU, "called, topology={}, count={}", draw_state.topology,
               draw_indexed ? draw_state.index_buffer.count : draw_state.vertex_buffer.count);
 
-    ProcessTopologyOverride();
+    UpdateTopology();
 
-    if (maxwell3d->ShouldExecute())
+    if (maxwell3d->ShouldExecute()) {
         maxwell3d->rasterizer->Draw(draw_indexed, instance_count);
+    }
 }
 } // namespace Tegra::Engines

src/video_core/engines/draw_manager.h

@@ -10,6 +10,7 @@ class RasterizerInterface;
 }
 
 namespace Tegra::Engines {
+using PrimitiveTopologyControl = Maxwell3D::Regs::PrimitiveTopologyControl;
 using PrimitiveTopology = Maxwell3D::Regs::PrimitiveTopology;
 using PrimitiveTopologyOverride = Maxwell3D::Regs::PrimitiveTopologyOverride;
 using IndexBuffer = Maxwell3D::Regs::IndexBuffer;
@@ -58,12 +59,11 @@ private:
 
     void DrawIndexSmall(u32 argument);
 
-    void ProcessTopologyOverride();
+    void UpdateTopology();
 
     void ProcessDraw(bool draw_indexed, u32 instance_count);
 
     Maxwell3D* maxwell3d{};
     State draw_state{};
-    bool use_topology_override{};
 };
 } // namespace Tegra::Engines

src/video_core/engines/engine_upload.cpp

@@ -24,7 +24,7 @@ void State::BindRasterizer(VideoCore::RasterizerInterface* rasterizer_) {
 void State::ProcessExec(const bool is_linear_) {
     write_offset = 0;
     copy_size = regs.line_length_in * regs.line_count;
-    inner_buffer.resize(copy_size);
+    inner_buffer.resize_destructive(copy_size);
     is_linear = is_linear_;
 }
 
@@ -70,7 +70,7 @@ void State::ProcessData(std::span<const u8> read_buffer) {
         const std::size_t dst_size = Tegra::Texture::CalculateSize(
             true, bytes_per_pixel, width, regs.dest.height, regs.dest.depth,
             regs.dest.BlockHeight(), regs.dest.BlockDepth());
-        tmp_buffer.resize(dst_size);
+        tmp_buffer.resize_destructive(dst_size);
         memory_manager.ReadBlock(address, tmp_buffer.data(), dst_size);
         Tegra::Texture::SwizzleSubrect(tmp_buffer, read_buffer, bytes_per_pixel, width,
                                        regs.dest.height, regs.dest.depth, x_offset, regs.dest.y,

src/video_core/engines/engine_upload.h

@@ -4,9 +4,10 @@
 #pragma once
 
 #include <span>
-#include <vector>
+
 #include "common/bit_field.h"
 #include "common/common_types.h"
+#include "common/scratch_buffer.h"
 
 namespace Tegra {
 class MemoryManager;
@@ -73,8 +74,8 @@ private:
 
     u32 write_offset = 0;
     u32 copy_size = 0;
-    std::vector<u8> inner_buffer;
-    std::vector<u8> tmp_buffer;
+    Common::ScratchBuffer<u8> inner_buffer;
+    Common::ScratchBuffer<u8> tmp_buffer;
     bool is_linear = false;
     Registers& regs;
     MemoryManager& memory_manager;

src/video_core/engines/maxwell_dma.cpp

@@ -184,12 +184,8 @@ void MaxwellDMA::CopyBlockLinearToPitch() {
     const size_t src_size =
         CalculateSize(true, bytes_per_pixel, width, height, depth, block_height, block_depth);
 
-    if (read_buffer.size() < src_size) {
-        read_buffer.resize(src_size);
-    }
-    if (write_buffer.size() < dst_size) {
-        write_buffer.resize(dst_size);
-    }
+    read_buffer.resize_destructive(src_size);
+    write_buffer.resize_destructive(dst_size);
 
     memory_manager.ReadBlock(regs.offset_in, read_buffer.data(), src_size);
     memory_manager.ReadBlock(regs.offset_out, write_buffer.data(), dst_size);
@@ -235,12 +231,8 @@ void MaxwellDMA::CopyPitchToBlockLinear() {
         CalculateSize(true, bytes_per_pixel, width, height, depth, block_height, block_depth);
     const size_t src_size = static_cast<size_t>(regs.pitch_in) * regs.line_count;
 
-    if (read_buffer.size() < src_size) {
-        read_buffer.resize(src_size);
-    }
-    if (write_buffer.size() < dst_size) {
-        write_buffer.resize(dst_size);
-    }
+    read_buffer.resize_destructive(src_size);
+    write_buffer.resize_destructive(dst_size);
 
     memory_manager.ReadBlock(regs.offset_in, read_buffer.data(), src_size);
     if (Settings::IsGPULevelExtreme()) {
@@ -269,12 +261,8 @@ void MaxwellDMA::FastCopyBlockLinearToPitch() {
     pos_x = pos_x % x_in_gob;
     pos_y = pos_y % 8;
 
-    if (read_buffer.size() < src_size) {
-        read_buffer.resize(src_size);
-    }
-    if (write_buffer.size() < dst_size) {
-        write_buffer.resize(dst_size);
-    }
+    read_buffer.resize_destructive(src_size);
+    write_buffer.resize_destructive(dst_size);
 
     if (Settings::IsGPULevelExtreme()) {
         memory_manager.ReadBlock(regs.offset_in + offset, read_buffer.data(), src_size);
@@ -333,14 +321,10 @@ void MaxwellDMA::CopyBlockLinearToBlockLinear() {
     const u32 pitch = x_elements * bytes_per_pixel;
     const size_t mid_buffer_size = pitch * regs.line_count;
 
-    if (read_buffer.size() < src_size) {
-        read_buffer.resize(src_size);
-    }
-    if (write_buffer.size() < dst_size) {
-        write_buffer.resize(dst_size);
-    }
+    read_buffer.resize_destructive(src_size);
+    write_buffer.resize_destructive(dst_size);
 
-    intermediate_buffer.resize(mid_buffer_size);
+    intermediate_buffer.resize_destructive(mid_buffer_size);
 
     memory_manager.ReadBlock(regs.offset_in, read_buffer.data(), src_size);
     memory_manager.ReadBlock(regs.offset_out, write_buffer.data(), dst_size);

src/video_core/engines/maxwell_dma.h

@@ -6,8 +6,10 @@
 #include <array>
 #include <cstddef>
 #include <vector>
+
 #include "common/bit_field.h"
 #include "common/common_types.h"
+#include "common/scratch_buffer.h"
 #include "video_core/engines/engine_interface.h"
 
 namespace Core {
@@ -234,9 +236,9 @@ private:
     MemoryManager& memory_manager;
     VideoCore::RasterizerInterface* rasterizer = nullptr;
 
-    std::vector<u8> read_buffer;
-    std::vector<u8> write_buffer;
-    std::vector<u8> intermediate_buffer;
+    Common::ScratchBuffer<u8> read_buffer;
+    Common::ScratchBuffer<u8> write_buffer;
+    Common::ScratchBuffer<u8> intermediate_buffer;
 
     static constexpr std::size_t NUM_REGS = 0x800;
     struct Regs {

src/video_core/gpu.cpp

@@ -223,8 +223,6 @@ struct GPU::Impl {
     /// core timing events.
     void Start() {
         gpu_thread.StartThread(*renderer, renderer->Context(), *scheduler);
-        cpu_context = renderer->GetRenderWindow().CreateSharedContext();
-        cpu_context->MakeCurrent();
     }
 
     void NotifyShutdown() {
@@ -235,6 +233,9 @@ struct GPU::Impl {
 
     /// Obtain the CPU Context
     void ObtainContext() {
+        if (!cpu_context) {
+            cpu_context = renderer->GetRenderWindow().CreateSharedContext();
+        }
         cpu_context->MakeCurrent();
     }
 

src/video_core/host1x/vic.cpp

@@ -155,7 +155,7 @@ void Vic::WriteRGBFrame(const AVFrame* frame, const VicConfig& config) {
         // swizzle pitch linear to block linear
         const u32 block_height = static_cast<u32>(config.block_linear_height_log2);
         const auto size = Texture::CalculateSize(true, 4, width, height, 1, block_height, 0);
-        luma_buffer.resize(size);
+        luma_buffer.resize_destructive(size);
         std::span<const u8> frame_buff(converted_frame_buf_addr, 4 * width * height);
         Texture::SwizzleSubrect(luma_buffer, frame_buff, 4, width, height, 1, 0, 0, width, height,
                                 block_height, 0, width * 4);
@@ -181,8 +181,8 @@ void Vic::WriteYUVFrame(const AVFrame* frame, const VicConfig& config) {
 
     const auto stride = static_cast<size_t>(frame->linesize[0]);
 
-    luma_buffer.resize(aligned_width * surface_height);
-    chroma_buffer.resize(aligned_width * surface_height / 2);
+    luma_buffer.resize_destructive(aligned_width * surface_height);
+    chroma_buffer.resize_destructive(aligned_width * surface_height / 2);
 
     // Populate luma buffer
     const u8* luma_src = frame->data[0];

src/video_core/host1x/vic.h

@@ -4,8 +4,9 @@
 #pragma once
 
 #include <memory>
-#include <vector>
+
 #include "common/common_types.h"
+#include "common/scratch_buffer.h"
 
 struct SwsContext;
 
@@ -49,8 +50,8 @@ private:
     /// size does not change during a stream
     using AVMallocPtr = std::unique_ptr<u8, decltype(&av_free)>;
     AVMallocPtr converted_frame_buffer;
-    std::vector<u8> luma_buffer;
-    std::vector<u8> chroma_buffer;
+    Common::ScratchBuffer<u8> luma_buffer;
+    Common::ScratchBuffer<u8> chroma_buffer;
 
     GPUVAddr config_struct_address{};
     GPUVAddr output_surface_luma_address{};

src/video_core/renderer_opengl/gl_device.cpp

@@ -112,7 +112,7 @@ bool IsASTCSupported() {
 }
 } // Anonymous namespace
 
-Device::Device() {
+Device::Device(Core::Frontend::EmuWindow& emu_window) {
     if (!GLAD_GL_VERSION_4_6) {
         LOG_ERROR(Render_OpenGL, "OpenGL 4.6 is not available");
         throw std::runtime_error{"Insufficient version"};
@@ -126,9 +126,9 @@ Device::Device() {
     const bool is_intel = vendor_name == "Intel";
 
 #ifdef __unix__
-    const bool is_linux = true;
+    constexpr bool is_linux = true;
 #else
-    const bool is_linux = false;
+    constexpr bool is_linux = false;
 #endif
 
     bool disable_fast_buffer_sub_data = false;
@@ -193,9 +193,11 @@ Device::Device() {
         }
     }
 
+    strict_context_required = emu_window.StrictContextRequired();
     // Blocks AMD and Intel OpenGL drivers on Windows from using asynchronous shader compilation.
+    // Blocks EGL on Wayland from using asynchronous shader compilation.
     use_asynchronous_shaders = Settings::values.use_asynchronous_shaders.GetValue() &&
-                               !(is_amd || (is_intel && !is_linux));
+                               !(is_amd || (is_intel && !is_linux)) && !strict_context_required;
     use_driver_cache = is_nvidia;
 
     LOG_INFO(Render_OpenGL, "Renderer_VariableAOFFI: {}", has_variable_aoffi);

src/video_core/renderer_opengl/gl_device.h

@@ -5,6 +5,7 @@
 
 #include <cstddef>
 #include "common/common_types.h"
+#include "core/frontend/emu_window.h"
 #include "shader_recompiler/stage.h"
 
 namespace Settings {
@@ -15,7 +16,7 @@ namespace OpenGL {
 
 class Device {
 public:
-    explicit Device();
+    explicit Device(Core::Frontend::EmuWindow& emu_window);
 
     [[nodiscard]] std::string GetVendorName() const;
 
@@ -173,6 +174,10 @@ public:
         return can_report_memory;
     }
 
+    bool StrictContextRequired() const {
+        return strict_context_required;
+    }
+
 private:
     static bool TestVariableAoffi();
     static bool TestPreciseBug();
@@ -216,6 +221,7 @@ private:
     bool has_cbuf_ftou_bug{};
     bool has_bool_ref_bug{};
     bool can_report_memory{};
+    bool strict_context_required{};
 
     std::string vendor_name;
 };

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -138,9 +138,6 @@ void RasterizerOpenGL::LoadDiskResources(u64 title_id, std::stop_token stop_load
 
 void RasterizerOpenGL::Clear(u32 layer_count) {
     MICROPROFILE_SCOPE(OpenGL_Clears);
-    if (!maxwell3d->ShouldExecute()) {
-        return;
-    }
 
     const auto& regs = maxwell3d->regs;
     bool use_color{};

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -174,6 +174,7 @@ ShaderCache::ShaderCache(RasterizerOpenGL& rasterizer_, Core::Frontend::EmuWindo
       texture_cache{texture_cache_}, buffer_cache{buffer_cache_}, program_manager{program_manager_},
       state_tracker{state_tracker_}, shader_notify{shader_notify_},
       use_asynchronous_shaders{device.UseAsynchronousShaders()},
+      strict_context_required{device.StrictContextRequired()},
       profile{
           .supported_spirv = 0x00010000,
 
@@ -203,6 +204,7 @@ ShaderCache::ShaderCache(RasterizerOpenGL& rasterizer_, Core::Frontend::EmuWindo
           .support_int64_atomics = false,
           .support_derivative_control = device.HasDerivativeControl(),
           .support_geometry_shader_passthrough = device.HasGeometryShaderPassthrough(),
+          .support_native_ndc = true,
           .support_gl_nv_gpu_shader_5 = device.HasNvGpuShader5(),
           .support_gl_amd_gpu_shader_half_float = device.HasAmdShaderHalfFloat(),
           .support_gl_texture_shadow_lod = device.HasTextureShadowLod(),
@@ -255,9 +257,14 @@ void ShaderCache::LoadDiskResources(u64 title_id, std::stop_token stop_loading,
     }
     shader_cache_filename = base_dir / "opengl.bin";
 
-    if (!workers) {
+    if (!workers && !strict_context_required) {
         workers = CreateWorkers();
     }
+    std::optional<Context> strict_context;
+    if (strict_context_required) {
+        strict_context.emplace(emu_window);
+    }
+
     struct {
         std::mutex mutex;
         size_t total{};
@@ -265,44 +272,49 @@ void ShaderCache::LoadDiskResources(u64 title_id, std::stop_token stop_loading,
         bool has_loaded{};
     } state;
 
+    const auto queue_work{[&](Common::UniqueFunction<void, Context*>&& work) {
+        if (strict_context_required) {
+            work(&strict_context.value());
+        } else {
+            workers->QueueWork(std::move(work));
+        }
+    }};
     const auto load_compute{[&](std::ifstream& file, FileEnvironment env) {
         ComputePipelineKey key;
         file.read(reinterpret_cast<char*>(&key), sizeof(key));
-        workers->QueueWork(
-            [this, key, env = std::move(env), &state, &callback](Context* ctx) mutable {
-                ctx->pools.ReleaseContents();
-                auto pipeline{CreateComputePipeline(ctx->pools, key, env)};
-                std::scoped_lock lock{state.mutex};
-                if (pipeline) {
-                    compute_cache.emplace(key, std::move(pipeline));
-                }
-                ++state.built;
-                if (state.has_loaded) {
-                    callback(VideoCore::LoadCallbackStage::Build, state.built, state.total);
-                }
-            });
+        queue_work([this, key, env = std::move(env), &state, &callback](Context* ctx) mutable {
+            ctx->pools.ReleaseContents();
+            auto pipeline{CreateComputePipeline(ctx->pools, key, env)};
+            std::scoped_lock lock{state.mutex};
+            if (pipeline) {
+                compute_cache.emplace(key, std::move(pipeline));
+            }
+            ++state.built;
+            if (state.has_loaded) {
+                callback(VideoCore::LoadCallbackStage::Build, state.built, state.total);
+            }
+        });
         ++state.total;
     }};
     const auto load_graphics{[&](std::ifstream& file, std::vector<FileEnvironment> envs) {
         GraphicsPipelineKey key;
         file.read(reinterpret_cast<char*>(&key), sizeof(key));
-        workers->QueueWork(
-            [this, key, envs = std::move(envs), &state, &callback](Context* ctx) mutable {
-                boost::container::static_vector<Shader::Environment*, 5> env_ptrs;
-                for (auto& env : envs) {
-                    env_ptrs.push_back(&env);
-                }
-                ctx->pools.ReleaseContents();
-                auto pipeline{CreateGraphicsPipeline(ctx->pools, key, MakeSpan(env_ptrs), false)};
-                std::scoped_lock lock{state.mutex};
-                if (pipeline) {
-                    graphics_cache.emplace(key, std::move(pipeline));
-                }
-                ++state.built;
-                if (state.has_loaded) {
-                    callback(VideoCore::LoadCallbackStage::Build, state.built, state.total);
-                }
-            });
+        queue_work([this, key, envs = std::move(envs), &state, &callback](Context* ctx) mutable {
+            boost::container::static_vector<Shader::Environment*, 5> env_ptrs;
+            for (auto& env : envs) {
+                env_ptrs.push_back(&env);
+            }
+            ctx->pools.ReleaseContents();
+            auto pipeline{CreateGraphicsPipeline(ctx->pools, key, MakeSpan(env_ptrs), false)};
+            std::scoped_lock lock{state.mutex};
+            if (pipeline) {
+                graphics_cache.emplace(key, std::move(pipeline));
+            }
+            ++state.built;
+            if (state.has_loaded) {
+                callback(VideoCore::LoadCallbackStage::Build, state.built, state.total);
+            }
+        });
         ++state.total;
     }};
     LoadPipelines(stop_loading, shader_cache_filename, CACHE_VERSION, load_compute, load_graphics);
@@ -314,6 +326,9 @@ void ShaderCache::LoadDiskResources(u64 title_id, std::stop_token stop_loading,
     state.has_loaded = true;
     lock.unlock();
 
+    if (strict_context_required) {
+        return;
+    }
     workers->WaitForRequests(stop_loading);
     if (!use_asynchronous_shaders) {
         workers.reset();

src/video_core/renderer_opengl/gl_shader_cache.h

@@ -69,6 +69,7 @@ private:
     StateTracker& state_tracker;
     VideoCore::ShaderNotify& shader_notify;
     const bool use_asynchronous_shaders;
+    const bool strict_context_required;
 
     GraphicsPipelineKey graphics_key{};
     GraphicsPipeline* current_pipeline{};

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -140,8 +140,8 @@ RendererOpenGL::RendererOpenGL(Core::TelemetrySession& telemetry_session_,
                                Core::Memory::Memory& cpu_memory_, Tegra::GPU& gpu_,
                                std::unique_ptr<Core::Frontend::GraphicsContext> context_)
     : RendererBase{emu_window_, std::move(context_)}, telemetry_session{telemetry_session_},
-      emu_window{emu_window_}, cpu_memory{cpu_memory_}, gpu{gpu_}, state_tracker{},
-      program_manager{device},
+      emu_window{emu_window_}, cpu_memory{cpu_memory_}, gpu{gpu_}, device{emu_window_},
+      state_tracker{}, program_manager{device},
       rasterizer(emu_window, gpu, cpu_memory, device, screen_info, program_manager, state_tracker) {
     if (Settings::values.renderer_debug && GLAD_GL_KHR_debug) {
         glEnable(GL_DEBUG_OUTPUT);

src/video_core/renderer_vulkan/renderer_vulkan.cpp

@@ -139,23 +139,25 @@ void RendererVulkan::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
     RenderScreenshot(*framebuffer, use_accelerated);
 
     bool has_been_recreated = false;
-    const auto recreate_swapchain = [&] {
+    const auto recreate_swapchain = [&](u32 width, u32 height) {
         if (!has_been_recreated) {
             has_been_recreated = true;
             scheduler.Finish();
         }
-        const Layout::FramebufferLayout layout = render_window.GetFramebufferLayout();
-        swapchain.Create(layout.width, layout.height, is_srgb);
+        swapchain.Create(width, height, is_srgb);
     };
-    if (swapchain.NeedsRecreation(is_srgb)) {
-        recreate_swapchain();
+
+    const Layout::FramebufferLayout layout = render_window.GetFramebufferLayout();
+    if (swapchain.NeedsRecreation(is_srgb) || swapchain.GetWidth() != layout.width ||
+        swapchain.GetHeight() != layout.height) {
+        recreate_swapchain(layout.width, layout.height);
     }
     bool is_outdated;
     do {
         swapchain.AcquireNextImage();
         is_outdated = swapchain.IsOutDated();
         if (is_outdated) {
-            recreate_swapchain();
+            recreate_swapchain(layout.width, layout.height);
         }
     } while (is_outdated);
     if (has_been_recreated) {

src/video_core/renderer_vulkan/vk_buffer_cache.cpp

@@ -461,6 +461,9 @@ void BufferCacheRuntime::BindQuadIndexBuffer(PrimitiveTopology topology, u32 fir
 
 void BufferCacheRuntime::BindVertexBuffer(u32 index, VkBuffer buffer, u32 offset, u32 size,
                                           u32 stride) {
+    if (index >= device.GetMaxVertexInputBindings()) {
+        return;
+    }
     if (device.IsExtExtendedDynamicStateSupported()) {
         scheduler.Record([index, buffer, offset, size, stride](vk::CommandBuffer cmdbuf) {
             const VkDeviceSize vk_offset = buffer != VK_NULL_HANDLE ? offset : 0;

src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp

@@ -529,7 +529,9 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
     static_vector<VkVertexInputBindingDivisorDescriptionEXT, 32> vertex_binding_divisors;
     static_vector<VkVertexInputAttributeDescription, 32> vertex_attributes;
     if (key.state.dynamic_vertex_input) {
-        for (size_t index = 0; index < key.state.attributes.size(); ++index) {
+        const size_t num_vertex_arrays = std::min(
+            key.state.attributes.size(), static_cast<size_t>(device.GetMaxVertexInputBindings()));
+        for (size_t index = 0; index < num_vertex_arrays; ++index) {
             const u32 type = key.state.DynamicAttributeType(index);
             if (!stage_infos[0].loads.Generic(index) || type == 0) {
                 continue;
@@ -551,7 +553,9 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
             });
         }
     } else {
-        for (size_t index = 0; index < Maxwell::NumVertexArrays; ++index) {
+        const size_t num_vertex_arrays = std::min(
+            Maxwell::NumVertexArrays, static_cast<size_t>(device.GetMaxVertexInputBindings()));
+        for (size_t index = 0; index < num_vertex_arrays; ++index) {
             const bool instanced = key.state.binding_divisors[index] != 0;
             const auto rate =
                 instanced ? VK_VERTEX_INPUT_RATE_INSTANCE : VK_VERTEX_INPUT_RATE_VERTEX;
@@ -580,6 +584,8 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
             });
         }
     }
+    ASSERT(vertex_attributes.size() <= device.GetMaxVertexInputAttributes());
+
     VkPipelineVertexInputStateCreateInfo vertex_input_ci{
         .sType = VK_STRUCTURE_TYPE_PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO,
         .pNext = nullptr,
@@ -634,23 +640,33 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
     };
     std::array<VkViewportSwizzleNV, Maxwell::NumViewports> swizzles;
     std::ranges::transform(key.state.viewport_swizzles, swizzles.begin(), UnpackViewportSwizzle);
-    const VkPipelineViewportSwizzleStateCreateInfoNV swizzle_ci{
+    VkPipelineViewportSwizzleStateCreateInfoNV swizzle_ci{
         .sType = VK_STRUCTURE_TYPE_PIPELINE_VIEWPORT_SWIZZLE_STATE_CREATE_INFO_NV,
         .pNext = nullptr,
         .flags = 0,
         .viewportCount = Maxwell::NumViewports,
         .pViewportSwizzles = swizzles.data(),
     };
-    const VkPipelineViewportStateCreateInfo viewport_ci{
+    VkPipelineViewportDepthClipControlCreateInfoEXT ndc_info{
+        .sType = VK_STRUCTURE_TYPE_PIPELINE_VIEWPORT_DEPTH_CLIP_CONTROL_CREATE_INFO_EXT,
+        .pNext = nullptr,
+        .negativeOneToOne = key.state.ndc_minus_one_to_one.Value() != 0 ? VK_TRUE : VK_FALSE,
+    };
+    VkPipelineViewportStateCreateInfo viewport_ci{
         .sType = VK_STRUCTURE_TYPE_PIPELINE_VIEWPORT_STATE_CREATE_INFO,
-        .pNext = device.IsNvViewportSwizzleSupported() ? &swizzle_ci : nullptr,
+        .pNext = nullptr,
         .flags = 0,
         .viewportCount = Maxwell::NumViewports,
         .pViewports = nullptr,
         .scissorCount = Maxwell::NumViewports,
         .pScissors = nullptr,
     };
-
+    if (device.IsNvViewportSwizzleSupported()) {
+        swizzle_ci.pNext = std::exchange(viewport_ci.pNext, &swizzle_ci);
+    }
+    if (device.IsExtDepthClipControlSupported()) {
+        ndc_info.pNext = std::exchange(viewport_ci.pNext, &ndc_info);
+    }
     VkPipelineRasterizationStateCreateInfo rasterization_ci{
         .sType = VK_STRUCTURE_TYPE_PIPELINE_RASTERIZATION_STATE_CREATE_INFO,
         .pNext = nullptr,

src/video_core/renderer_vulkan/vk_pipeline_cache.cpp

@@ -321,6 +321,7 @@ PipelineCache::PipelineCache(RasterizerVulkan& rasterizer_, const Device& device
         .support_int64_atomics = device.IsExtShaderAtomicInt64Supported(),
         .support_derivative_control = true,
         .support_geometry_shader_passthrough = device.IsNvGeometryShaderPassthroughSupported(),
+        .support_native_ndc = device.IsExtDepthClipControlSupported(),
 
         .warp_size_potentially_larger_than_guest = device.IsWarpSizePotentiallyBiggerThanGuest(),
 
@@ -341,6 +342,15 @@ PipelineCache::PipelineCache(RasterizerVulkan& rasterizer_, const Device& device
         .support_snorm_render_buffer = true,
         .support_viewport_index_layer = device.IsExtShaderViewportIndexLayerSupported(),
     };
+
+    if (device.GetMaxVertexInputAttributes() < Maxwell::NumVertexAttributes) {
+        LOG_WARNING(Render_Vulkan, "maxVertexInputAttributes is too low: {} < {}",
+                    device.GetMaxVertexInputAttributes(), Maxwell::NumVertexAttributes);
+    }
+    if (device.GetMaxVertexInputBindings() < Maxwell::NumVertexArrays) {
+        LOG_WARNING(Render_Vulkan, "maxVertexInputBindings is too low: {} < {}",
+                    device.GetMaxVertexInputBindings(), Maxwell::NumVertexArrays);
+    }
 }
 
 PipelineCache::~PipelineCache() = default;

src/video_core/renderer_vulkan/vk_rasterizer.cpp

@@ -220,9 +220,6 @@ void RasterizerVulkan::Draw(bool is_indexed, u32 instance_count) {
 void RasterizerVulkan::Clear(u32 layer_count) {
     MICROPROFILE_SCOPE(Vulkan_Clearing);
 
-    if (!maxwell3d->ShouldExecute()) {
-        return;
-    }
     FlushWork();
 
     query_cache.UpdateCounters();
@@ -665,8 +662,7 @@ void RasterizerVulkan::BeginTransformFeedback() {
         return;
     }
     UNIMPLEMENTED_IF(regs.IsShaderConfigEnabled(Maxwell::ShaderType::TessellationInit) ||
-                     regs.IsShaderConfigEnabled(Maxwell::ShaderType::Tessellation) ||
-                     regs.IsShaderConfigEnabled(Maxwell::ShaderType::Geometry));
+                     regs.IsShaderConfigEnabled(Maxwell::ShaderType::Tessellation));
     scheduler.Record(
         [](vk::CommandBuffer cmdbuf) { cmdbuf.BeginTransformFeedbackEXT(0, 0, nullptr, nullptr); });
 }