vulkan_device: Fix sType for VkPhysicalDeviceShaderAtomicInt64Features

Kernel Memory Updates (Part 3): Clear KMemoryManager pages & other fixes

src/common/fiber.cpp

@@ -124,7 +124,10 @@ void Fiber::YieldTo(std::weak_ptr<Fiber> weak_from, Fiber& to) {
 
     // "from" might no longer be valid if the thread was killed
     if (auto from = weak_from.lock()) {
-        ASSERT(from->impl->previous_fiber != nullptr);
+        if (from->impl->previous_fiber == nullptr) {
+            ASSERT_MSG(false, "previous_fiber is nullptr!");
+            return;
+        }
         from->impl->previous_fiber->impl->context = transfer.fctx;
         from->impl->previous_fiber->impl->guard.unlock();
         from->impl->previous_fiber.reset();

src/common/input.h

@@ -209,6 +209,13 @@ enum class ButtonNames {
     Triangle,
     Share,
     Options,
+
+    // Mouse buttons
+    ButtonMouseWheel,
+    ButtonBackward,
+    ButtonForward,
+    ButtonTask,
+    ButtonExtra,
 };
 
 // Callback data consisting of an input type and the equivalent data status

src/core/CMakeLists.txt

@@ -247,6 +247,9 @@ add_library(core STATIC
     hle/kernel/k_trace.h
     hle/kernel/k_transfer_memory.cpp
     hle/kernel/k_transfer_memory.h
+    hle/kernel/k_worker_task.h
+    hle/kernel/k_worker_task_manager.cpp
+    hle/kernel/k_worker_task_manager.h
     hle/kernel/k_writable_event.cpp
     hle/kernel/k_writable_event.h
     hle/kernel/kernel.cpp

src/core/hid/emulated_console.cpp

@@ -158,7 +158,7 @@ void EmulatedConsole::SetMotion(const Common::Input::CallbackStatus& callback) {
     auto& motion = console.motion_state;
     motion.accel = emulated.GetAcceleration();
     motion.gyro = emulated.GetGyroscope();
-    motion.rotation = emulated.GetGyroscope();
+    motion.rotation = emulated.GetRotations();
     motion.orientation = emulated.GetOrientation();
     motion.quaternion = emulated.GetQuaternion();
     motion.gyro_bias = emulated.GetGyroBias();

src/core/hid/emulated_controller.cpp

@@ -145,7 +145,7 @@ void EmulatedController::LoadDevices() {
                    motion_devices.begin(), Common::Input::CreateDevice<Common::Input::InputDevice>);
     std::transform(trigger_params.begin(), trigger_params.end(), trigger_devices.begin(),
                    Common::Input::CreateDevice<Common::Input::InputDevice>);
-    std::transform(battery_params.begin(), battery_params.begin(), battery_devices.end(),
+    std::transform(battery_params.begin(), battery_params.end(), battery_devices.begin(),
                    Common::Input::CreateDevice<Common::Input::InputDevice>);
     std::transform(output_params.begin(), output_params.end(), output_devices.begin(),
                    Common::Input::CreateDevice<Common::Input::OutputDevice>);

src/core/hle/kernel/k_memory_manager.cpp

@@ -8,12 +8,16 @@
 #include "common/assert.h"
 #include "common/common_types.h"
 #include "common/scope_exit.h"
+#include "core/core.h"
+#include "core/device_memory.h"
 #include "core/hle/kernel/k_memory_manager.h"
 #include "core/hle/kernel/k_page_linked_list.h"
 #include "core/hle/kernel/svc_results.h"
 
 namespace Kernel {
 
+KMemoryManager::KMemoryManager(Core::System& system_) : system{system_} {}
+
 std::size_t KMemoryManager::Impl::Initialize(Pool new_pool, u64 start_address, u64 end_address) {
     const auto size{end_address - start_address};
 
@@ -81,7 +85,7 @@ VAddr KMemoryManager::AllocateAndOpenContinuous(std::size_t num_pages, std::size
 }
 
 ResultCode KMemoryManager::Allocate(KPageLinkedList& page_list, std::size_t num_pages, Pool pool,
-                                    Direction dir) {
+                                    Direction dir, u32 heap_fill_value) {
     ASSERT(page_list.GetNumPages() == 0);
 
     // Early return if we're allocating no pages
@@ -139,6 +143,12 @@ ResultCode KMemoryManager::Allocate(KPageLinkedList& page_list, std::size_t num_
         }
     }
 
+    // Clear allocated memory.
+    for (const auto& it : page_list.Nodes()) {
+        std::memset(system.DeviceMemory().GetPointer(it.GetAddress()), heap_fill_value,
+                    it.GetSize());
+    }
+
     // Only succeed if we allocated as many pages as we wanted
     if (num_pages) {
         return ResultOutOfMemory;
@@ -146,11 +156,12 @@ ResultCode KMemoryManager::Allocate(KPageLinkedList& page_list, std::size_t num_
 
     // We succeeded!
     group_guard.Cancel();
+
     return ResultSuccess;
 }
 
 ResultCode KMemoryManager::Free(KPageLinkedList& page_list, std::size_t num_pages, Pool pool,
-                                Direction dir) {
+                                Direction dir, u32 heap_fill_value) {
     // Early return if we're freeing no pages
     if (!num_pages) {
         return ResultSuccess;

src/core/hle/kernel/k_memory_manager.h

@@ -12,6 +12,10 @@
 #include "core/hle/kernel/k_page_heap.h"
 #include "core/hle/result.h"
 
+namespace Core {
+class System;
+}
+
 namespace Kernel {
 
 class KPageLinkedList;
@@ -42,7 +46,7 @@ public:
         Mask = (0xF << Shift),
     };
 
-    KMemoryManager() = default;
+    explicit KMemoryManager(Core::System& system_);
 
     constexpr std::size_t GetSize(Pool pool) const {
         return managers[static_cast<std::size_t>(pool)].GetSize();
@@ -51,10 +55,10 @@ public:
     void InitializeManager(Pool pool, u64 start_address, u64 end_address);
 
     VAddr AllocateAndOpenContinuous(size_t num_pages, size_t align_pages, u32 option);
-    ResultCode Allocate(KPageLinkedList& page_list, std::size_t num_pages, Pool pool,
-                        Direction dir = Direction::FromFront);
-    ResultCode Free(KPageLinkedList& page_list, std::size_t num_pages, Pool pool,
-                    Direction dir = Direction::FromFront);
+    ResultCode Allocate(KPageLinkedList& page_list, std::size_t num_pages, Pool pool, Direction dir,
+                        u32 heap_fill_value = 0);
+    ResultCode Free(KPageLinkedList& page_list, std::size_t num_pages, Pool pool, Direction dir,
+                    u32 heap_fill_value = 0);
 
     static constexpr std::size_t MaxManagerCount = 10;
 
@@ -129,6 +133,7 @@ private:
     };
 
 private:
+    Core::System& system;
     std::array<std::mutex, static_cast<std::size_t>(Pool::Count)> pool_locks;
     std::array<Impl, MaxManagerCount> managers;
 };

src/core/hle/kernel/k_page_table.cpp

@@ -289,8 +289,8 @@ ResultCode KPageTable::MapProcessCode(VAddr addr, std::size_t num_pages, KMemory
     }
 
     KPageLinkedList page_linked_list;
-    CASCADE_CODE(
-        system.Kernel().MemoryManager().Allocate(page_linked_list, num_pages, memory_pool));
+    CASCADE_CODE(system.Kernel().MemoryManager().Allocate(page_linked_list, num_pages, memory_pool,
+                                                          allocation_option));
     CASCADE_CODE(Operate(addr, num_pages, page_linked_list, OperationType::MapGroup));
 
     block_manager->Update(addr, num_pages, state, perm);
@@ -457,8 +457,8 @@ ResultCode KPageTable::MapPhysicalMemory(VAddr addr, std::size_t size) {
 
     KPageLinkedList page_linked_list;
 
-    CASCADE_CODE(
-        system.Kernel().MemoryManager().Allocate(page_linked_list, remaining_pages, memory_pool));
+    CASCADE_CODE(system.Kernel().MemoryManager().Allocate(page_linked_list, remaining_pages,
+                                                          memory_pool, allocation_option));
 
     // We succeeded, so commit the memory reservation.
     memory_reservation.Commit();
@@ -541,7 +541,8 @@ ResultCode KPageTable::UnmapMemory(VAddr addr, std::size_t size) {
     }
 
     const std::size_t num_pages{size / PageSize};
-    system.Kernel().MemoryManager().Free(page_linked_list, num_pages, memory_pool);
+    system.Kernel().MemoryManager().Free(page_linked_list, num_pages, memory_pool,
+                                         allocation_option);
 
     block_manager->Update(addr, num_pages, KMemoryState::Free);
 
@@ -960,7 +961,7 @@ ResultCode KPageTable::SetHeapSize(VAddr* out, std::size_t size) {
     // Allocate pages for the heap extension.
     KPageLinkedList page_linked_list;
     R_TRY(system.Kernel().MemoryManager().Allocate(page_linked_list, allocation_size / PageSize,
-                                                   memory_pool));
+                                                   memory_pool, allocation_option));
 
     // Map the pages.
     {
@@ -1027,8 +1028,8 @@ ResultVal<VAddr> KPageTable::AllocateAndMapMemory(std::size_t needed_num_pages,
         CASCADE_CODE(Operate(addr, needed_num_pages, perm, OperationType::Map, map_addr));
     } else {
         KPageLinkedList page_group;
-        CASCADE_CODE(
-            system.Kernel().MemoryManager().Allocate(page_group, needed_num_pages, memory_pool));
+        CASCADE_CODE(system.Kernel().MemoryManager().Allocate(page_group, needed_num_pages,
+                                                              memory_pool, allocation_option));
         CASCADE_CODE(Operate(addr, needed_num_pages, page_group, OperationType::MapGroup));
     }
 

src/core/hle/kernel/k_page_table.h

@@ -303,6 +303,7 @@ private:
     bool is_aslr_enabled{};
 
     KMemoryManager::Pool memory_pool{KMemoryManager::Pool::Application};
+    KMemoryManager::Direction allocation_option{KMemoryManager::Direction::FromFront};
 
     Common::PageTable page_table_impl;
 

src/core/hle/kernel/k_process.cpp

@@ -149,6 +149,10 @@ ResultCode KProcess::Initialize(KProcess* process, Core::System& system, std::st
     return ResultSuccess;
 }
 
+void KProcess::DoWorkerTaskImpl() {
+    UNIMPLEMENTED();
+}
+
 KResourceLimit* KProcess::GetResourceLimit() const {
     return resource_limit;
 }
@@ -477,7 +481,7 @@ void KProcess::Finalize() {
     }
 
     // Perform inherited finalization.
-    KAutoObjectWithSlabHeapAndContainer<KProcess, KSynchronizationObject>::Finalize();
+    KAutoObjectWithSlabHeapAndContainer<KProcess, KWorkerTask>::Finalize();
 }
 
 /**

src/core/hle/kernel/k_process.h

@@ -15,6 +15,7 @@
 #include "core/hle/kernel/k_condition_variable.h"
 #include "core/hle/kernel/k_handle_table.h"
 #include "core/hle/kernel/k_synchronization_object.h"
+#include "core/hle/kernel/k_worker_task.h"
 #include "core/hle/kernel/process_capability.h"
 #include "core/hle/kernel/slab_helpers.h"
 #include "core/hle/result.h"
@@ -62,8 +63,7 @@ enum class ProcessStatus {
     DebugBreak,
 };
 
-class KProcess final
-    : public KAutoObjectWithSlabHeapAndContainer<KProcess, KSynchronizationObject> {
+class KProcess final : public KAutoObjectWithSlabHeapAndContainer<KProcess, KWorkerTask> {
     KERNEL_AUTOOBJECT_TRAITS(KProcess, KSynchronizationObject);
 
 public:
@@ -345,6 +345,8 @@ public:
 
     bool IsSignaled() const override;
 
+    void DoWorkerTaskImpl();
+
     void PinCurrentThread(s32 core_id);
     void UnpinCurrentThread(s32 core_id);
     void UnpinThread(KThread* thread);

src/core/hle/kernel/k_scheduler.cpp

@@ -49,8 +49,6 @@ void KScheduler::RescheduleCores(KernelCore& kernel, u64 cores_pending_reschedul
         if (!must_context_switch || core != current_core) {
             auto& phys_core = kernel.PhysicalCore(core);
             phys_core.Interrupt();
-        } else {
-            must_context_switch = true;
         }
         cores_pending_reschedule &= ~(1ULL << core);
     }

src/core/hle/kernel/k_thread.cpp

@@ -30,6 +30,7 @@
 #include "core/hle/kernel/k_system_control.h"
 #include "core/hle/kernel/k_thread.h"
 #include "core/hle/kernel/k_thread_queue.h"
+#include "core/hle/kernel/k_worker_task_manager.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/svc_results.h"
 #include "core/hle/kernel/time_manager.h"
@@ -332,7 +333,7 @@ void KThread::Finalize() {
     }
 
     // Perform inherited finalization.
-    KAutoObjectWithSlabHeapAndContainer<KThread, KSynchronizationObject>::Finalize();
+    KSynchronizationObject::Finalize();
 }
 
 bool KThread::IsSignaled() const {
@@ -376,11 +377,28 @@ void KThread::StartTermination() {
 
     // Register terminated dpc flag.
     RegisterDpc(DpcFlag::Terminated);
+}
+
+void KThread::FinishTermination() {
+    // Ensure that the thread is not executing on any core.
+    if (parent != nullptr) {
+        for (std::size_t i = 0; i < static_cast<std::size_t>(Core::Hardware::NUM_CPU_CORES); ++i) {
+            KThread* core_thread{};
+            do {
+                core_thread = kernel.Scheduler(i).GetCurrentThread();
+            } while (core_thread == this);
+        }
+    }
 
     // Close the thread.
     this->Close();
 }
 
+void KThread::DoWorkerTaskImpl() {
+    // Finish the termination that was begun by Exit().
+    this->FinishTermination();
+}
+
 void KThread::Pin(s32 current_core) {
     ASSERT(kernel.GlobalSchedulerContext().IsLocked());
 
@@ -417,12 +435,7 @@ void KThread::Pin(s32 current_core) {
                                          static_cast<u32>(ThreadState::SuspendShift)));
 
         // Update our state.
-        const ThreadState old_state = thread_state;
-        thread_state = static_cast<ThreadState>(GetSuspendFlags() |
-                                                static_cast<u32>(old_state & ThreadState::Mask));
-        if (thread_state != old_state) {
-            KScheduler::OnThreadStateChanged(kernel, this, old_state);
-        }
+        UpdateState();
     }
 
     // TODO(bunnei): Update our SVC access permissions.
@@ -463,20 +476,13 @@ void KThread::Unpin() {
     }
 
     // Allow performing thread suspension (if termination hasn't been requested).
-    {
+    if (!IsTerminationRequested()) {
         // Update our allow flags.
-        if (!IsTerminationRequested()) {
-            suspend_allowed_flags |= (1 << (static_cast<u32>(SuspendType::Thread) +
-                                            static_cast<u32>(ThreadState::SuspendShift)));
-        }
+        suspend_allowed_flags |= (1 << (static_cast<u32>(SuspendType::Thread) +
+                                        static_cast<u32>(ThreadState::SuspendShift)));
 
         // Update our state.
-        const ThreadState old_state = thread_state;
-        thread_state = static_cast<ThreadState>(GetSuspendFlags() |
-                                                static_cast<u32>(old_state & ThreadState::Mask));
-        if (thread_state != old_state) {
-            KScheduler::OnThreadStateChanged(kernel, this, old_state);
-        }
+        UpdateState();
     }
 
     // TODO(bunnei): Update our SVC access permissions.
@@ -689,12 +695,7 @@ void KThread::Resume(SuspendType type) {
         ~(1u << (static_cast<u32>(ThreadState::SuspendShift) + static_cast<u32>(type)));
 
     // Update our state.
-    const ThreadState old_state = thread_state;
-    thread_state = static_cast<ThreadState>(GetSuspendFlags() |
-                                            static_cast<u32>(old_state & ThreadState::Mask));
-    if (thread_state != old_state) {
-        KScheduler::OnThreadStateChanged(kernel, this, old_state);
-    }
+    this->UpdateState();
 }
 
 void KThread::WaitCancel() {
@@ -721,19 +722,22 @@ void KThread::TrySuspend() {
     ASSERT(GetNumKernelWaiters() == 0);
 
     // Perform the suspend.
-    Suspend();
+    this->UpdateState();
 }
 
-void KThread::Suspend() {
+void KThread::UpdateState() {
     ASSERT(kernel.GlobalSchedulerContext().IsLocked());
-    ASSERT(IsSuspendRequested());
 
     // Set our suspend flags in state.
     const auto old_state = thread_state;
-    thread_state = static_cast<ThreadState>(GetSuspendFlags()) | (old_state & ThreadState::Mask);
+    const auto new_state =
+        static_cast<ThreadState>(this->GetSuspendFlags()) | (old_state & ThreadState::Mask);
+    thread_state = new_state;
 
     // Note the state change in scheduler.
-    KScheduler::OnThreadStateChanged(kernel, this, old_state);
+    if (new_state != old_state) {
+        KScheduler::OnThreadStateChanged(kernel, this, old_state);
+    }
 }
 
 void KThread::Continue() {
@@ -998,13 +1002,16 @@ ResultCode KThread::Run() {
 
         // If the current thread has been asked to suspend, suspend it and retry.
         if (GetCurrentThread(kernel).IsSuspended()) {
-            GetCurrentThread(kernel).Suspend();
+            GetCurrentThread(kernel).UpdateState();
             continue;
         }
 
         // If we're not a kernel thread and we've been asked to suspend, suspend ourselves.
-        if (IsUserThread() && IsSuspended()) {
-            Suspend();
+        if (KProcess* owner = this->GetOwnerProcess(); owner != nullptr) {
+            if (IsUserThread() && IsSuspended()) {
+                this->UpdateState();
+            }
+            owner->IncrementThreadCount();
         }
 
         // Set our state and finish.
@@ -1029,11 +1036,18 @@ void KThread::Exit() {
     {
         KScopedSchedulerLock sl{kernel};
 
+        // Disallow all suspension.
+        suspend_allowed_flags = 0;
+        this->UpdateState();
+
         // Disallow all suspension.
         suspend_allowed_flags = 0;
 
         // Start termination.
         StartTermination();
+
+        // Register the thread as a work task.
+        KWorkerTaskManager::AddTask(kernel, KWorkerTaskManager::WorkerType::Exit, this);
     }
 }
 

src/core/hle/kernel/k_thread.h

@@ -19,6 +19,7 @@
 #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_worker_task.h"
 #include "core/hle/kernel/slab_helpers.h"
 #include "core/hle/kernel/svc_common.h"
 #include "core/hle/kernel/svc_types.h"
@@ -100,7 +101,7 @@ enum class ThreadWaitReasonForDebugging : u32 {
 [[nodiscard]] KThread& GetCurrentThread(KernelCore& kernel);
 [[nodiscard]] s32 GetCurrentCoreId(KernelCore& kernel);
 
-class KThread final : public KAutoObjectWithSlabHeapAndContainer<KThread, KSynchronizationObject>,
+class KThread final : public KAutoObjectWithSlabHeapAndContainer<KThread, KWorkerTask>,
                       public boost::intrusive::list_base_hook<> {
     KERNEL_AUTOOBJECT_TRAITS(KThread, KSynchronizationObject);
 
@@ -192,9 +193,9 @@ public:
 
     void TrySuspend();
 
-    void Continue();
+    void UpdateState();
 
-    void Suspend();
+    void Continue();
 
     constexpr void SetSyncedIndex(s32 index) {
         synced_index = index;
@@ -385,6 +386,8 @@ public:
 
     void OnTimer();
 
+    void DoWorkerTaskImpl();
+
     static void PostDestroy(uintptr_t arg);
 
     [[nodiscard]] static ResultCode InitializeDummyThread(KThread* thread);
@@ -679,6 +682,8 @@ private:
 
     void StartTermination();
 
+    void FinishTermination();
+
     [[nodiscard]] ResultCode Initialize(KThreadFunction func, uintptr_t arg, VAddr user_stack_top,
                                         s32 prio, s32 virt_core, KProcess* owner, ThreadType type);
 

src/core/hle/kernel/k_worker_task.h

@@ -0,0 +1,18 @@
+// Copyright 2022 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/kernel/k_synchronization_object.h"
+
+namespace Kernel {
+
+class KWorkerTask : public KSynchronizationObject {
+public:
+    explicit KWorkerTask(KernelCore& kernel_);
+
+    void DoWorkerTask();
+};
+
+} // namespace Kernel

src/core/hle/kernel/k_worker_task_manager.cpp

@@ -0,0 +1,42 @@
+// Copyright 2022 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/assert.h"
+#include "core/hle/kernel/k_process.h"
+#include "core/hle/kernel/k_thread.h"
+#include "core/hle/kernel/k_worker_task.h"
+#include "core/hle/kernel/k_worker_task_manager.h"
+#include "core/hle/kernel/kernel.h"
+
+namespace Kernel {
+
+KWorkerTask::KWorkerTask(KernelCore& kernel_) : KSynchronizationObject{kernel_} {}
+
+void KWorkerTask::DoWorkerTask() {
+    if (auto* const thread = this->DynamicCast<KThread*>(); thread != nullptr) {
+        return thread->DoWorkerTaskImpl();
+    } else {
+        auto* const process = this->DynamicCast<KProcess*>();
+        ASSERT(process != nullptr);
+
+        return process->DoWorkerTaskImpl();
+    }
+}
+
+KWorkerTaskManager::KWorkerTaskManager() : m_waiting_thread(1, "yuzu:KWorkerTaskManager") {}
+
+void KWorkerTaskManager::AddTask(KernelCore& kernel, WorkerType type, KWorkerTask* task) {
+    ASSERT(type <= WorkerType::Count);
+    kernel.WorkerTaskManager().AddTask(kernel, task);
+}
+
+void KWorkerTaskManager::AddTask(KernelCore& kernel, KWorkerTask* task) {
+    KScopedSchedulerLock sl(kernel);
+    m_waiting_thread.QueueWork([task]() {
+        // Do the task.
+        task->DoWorkerTask();
+    });
+}
+
+} // namespace Kernel

src/core/hle/kernel/k_worker_task_manager.h

@@ -0,0 +1,33 @@
+// Copyright 2022 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_types.h"
+#include "common/thread_worker.h"
+
+namespace Kernel {
+
+class KernelCore;
+class KWorkerTask;
+
+class KWorkerTaskManager final {
+public:
+    enum class WorkerType : u32 {
+        Exit,
+        Count,
+    };
+
+    KWorkerTaskManager();
+
+    static void AddTask(KernelCore& kernel_, WorkerType type, KWorkerTask* task);
+
+private:
+    void AddTask(KernelCore& kernel, KWorkerTask* task);
+
+private:
+    Common::ThreadWorker m_waiting_thread;
+};
+
+} // namespace Kernel

src/core/hle/kernel/kernel.cpp

@@ -37,6 +37,7 @@
 #include "core/hle/kernel/k_shared_memory.h"
 #include "core/hle/kernel/k_slab_heap.h"
 #include "core/hle/kernel/k_thread.h"
+#include "core/hle/kernel/k_worker_task_manager.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/physical_core.h"
 #include "core/hle/kernel/service_thread.h"
@@ -51,7 +52,8 @@ namespace Kernel {
 
 struct KernelCore::Impl {
     explicit Impl(Core::System& system_, KernelCore& kernel_)
-        : time_manager{system_}, object_list_container{kernel_}, system{system_} {}
+        : time_manager{system_}, object_list_container{kernel_},
+          service_threads_manager{1, "yuzu:ServiceThreadsManager"}, system{system_} {}
 
     void SetMulticore(bool is_multi) {
         is_multicore = is_multi;
@@ -629,7 +631,7 @@ struct KernelCore::Impl {
         const auto application_pool = memory_layout.GetKernelApplicationPoolRegionPhysicalExtents();
 
         // Initialize memory managers
-        memory_manager = std::make_unique<KMemoryManager>();
+        memory_manager = std::make_unique<KMemoryManager>(system);
         memory_manager->InitializeManager(KMemoryManager::Pool::Application,
                                           application_pool.GetAddress(),
                                           application_pool.GetEndAddress());
@@ -707,24 +709,22 @@ struct KernelCore::Impl {
     std::weak_ptr<Kernel::ServiceThread> CreateServiceThread(KernelCore& kernel,
                                                              const std::string& name) {
         auto service_thread = std::make_shared<Kernel::ServiceThread>(kernel, 1, name);
-        {
-            std::lock_guard lk(service_threads_lock);
-            service_threads.emplace(service_thread);
-        }
+
+        service_threads_manager.QueueWork(
+            [this, service_thread]() { service_threads.emplace(service_thread); });
+
         return service_thread;
     }
 
     void ReleaseServiceThread(std::weak_ptr<Kernel::ServiceThread> service_thread) {
-        auto strong_ptr = service_thread.lock();
-        {
-            std::lock_guard lk(service_threads_lock);
-            service_threads.erase(strong_ptr);
+        if (auto strong_ptr = service_thread.lock()) {
+            service_threads_manager.QueueWork(
+                [this, strong_ptr{std::move(strong_ptr)}]() { service_threads.erase(strong_ptr); });
         }
     }
 
     void ClearServiceThreads() {
-        std::lock_guard lk(service_threads_lock);
-        service_threads.clear();
+        service_threads_manager.QueueWork([this]() { service_threads.clear(); });
     }
 
     std::mutex server_ports_lock;
@@ -732,7 +732,6 @@ struct KernelCore::Impl {
     std::mutex registered_objects_lock;
     std::mutex registered_in_use_objects_lock;
     std::mutex dummy_thread_lock;
-    std::mutex service_threads_lock;
 
     std::atomic<u32> next_object_id{0};
     std::atomic<u64> next_kernel_process_id{KProcess::InitialKIPIDMin};
@@ -783,6 +782,7 @@ struct KernelCore::Impl {
 
     // Threads used for services
     std::unordered_set<std::shared_ptr<Kernel::ServiceThread>> service_threads;
+    Common::ThreadWorker service_threads_manager;
 
     std::array<KThread*, Core::Hardware::NUM_CPU_CORES> suspend_threads;
     std::array<Core::CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES> interrupts{};
@@ -798,6 +798,8 @@ struct KernelCore::Impl {
 
     std::array<u64, Core::Hardware::NUM_CPU_CORES> svc_ticks{};
 
+    KWorkerTaskManager worker_task_manager;
+
     // System context
     Core::System& system;
 };
@@ -1138,6 +1140,14 @@ const Init::KSlabResourceCounts& KernelCore::SlabResourceCounts() const {
     return impl->slab_resource_counts;
 }
 
+KWorkerTaskManager& KernelCore::WorkerTaskManager() {
+    return impl->worker_task_manager;
+}
+
+const KWorkerTaskManager& KernelCore::WorkerTaskManager() const {
+    return impl->worker_task_manager;
+}
+
 bool KernelCore::IsPhantomModeForSingleCore() const {
     return impl->IsPhantomModeForSingleCore();
 }

src/core/hle/kernel/kernel.h

@@ -52,6 +52,7 @@ class KSharedMemory;
 class KSharedMemoryInfo;
 class KThread;
 class KTransferMemory;
+class KWorkerTaskManager;
 class KWritableEvent;
 class KCodeMemory;
 class PhysicalCore;
@@ -343,6 +344,12 @@ public:
     /// Gets the current slab resource counts.
     const Init::KSlabResourceCounts& SlabResourceCounts() const;
 
+    /// Gets the current worker task manager, used for dispatching KThread/KProcess tasks.
+    KWorkerTaskManager& WorkerTaskManager();
+
+    /// Gets the current worker task manager, used for dispatching KThread/KProcess tasks.
+    const KWorkerTaskManager& WorkerTaskManager() const;
+
 private:
     friend class KProcess;
     friend class KThread;

src/input_common/drivers/mouse.cpp

@@ -16,6 +16,7 @@ constexpr int mouse_axis_x = 0;
 constexpr int mouse_axis_y = 1;
 constexpr int wheel_axis_x = 2;
 constexpr int wheel_axis_y = 3;
+constexpr int motion_wheel_y = 4;
 constexpr int touch_axis_x = 10;
 constexpr int touch_axis_y = 11;
 constexpr PadIdentifier identifier = {
@@ -30,8 +31,9 @@ Mouse::Mouse(std::string input_engine_) : InputEngine(std::move(input_engine_))
     PreSetAxis(identifier, mouse_axis_y);
     PreSetAxis(identifier, wheel_axis_x);
     PreSetAxis(identifier, wheel_axis_y);
+    PreSetAxis(identifier, motion_wheel_y);
     PreSetAxis(identifier, touch_axis_x);
-    PreSetAxis(identifier, touch_axis_x);
+    PreSetAxis(identifier, touch_axis_y);
     update_thread = std::jthread([this](std::stop_token stop_token) { UpdateThread(stop_token); });
 }
 
@@ -48,6 +50,8 @@ void Mouse::UpdateThread(std::stop_token stop_token) {
             SetAxis(identifier, mouse_axis_y, -last_mouse_change.y * sensitivity);
         }
 
+        SetAxis(identifier, motion_wheel_y, 0.0f);
+
         if (mouse_panning_timout++ > 20) {
             StopPanning();
         }
@@ -136,6 +140,7 @@ void Mouse::MouseWheelChange(int x, int y) {
     wheel_position.y += y;
     SetAxis(identifier, wheel_axis_x, static_cast<f32>(wheel_position.x));
     SetAxis(identifier, wheel_axis_y, static_cast<f32>(wheel_position.y));
+    SetAxis(identifier, motion_wheel_y, static_cast<f32>(y) / 100.0f);
 }
 
 void Mouse::ReleaseAllButtons() {
@@ -171,13 +176,39 @@ AnalogMapping Mouse::GetAnalogMappingForDevice(
     return mapping;
 }
 
+Common::Input::ButtonNames Mouse::GetUIButtonName(const Common::ParamPackage& params) const {
+    const auto button = static_cast<MouseButton>(params.Get("button", 0));
+    switch (button) {
+    case MouseButton::Left:
+        return Common::Input::ButtonNames::ButtonLeft;
+    case MouseButton::Right:
+        return Common::Input::ButtonNames::ButtonRight;
+    case MouseButton::Wheel:
+        return Common::Input::ButtonNames::ButtonMouseWheel;
+    case MouseButton::Backward:
+        return Common::Input::ButtonNames::ButtonBackward;
+    case MouseButton::Forward:
+        return Common::Input::ButtonNames::ButtonForward;
+    case MouseButton::Task:
+        return Common::Input::ButtonNames::ButtonTask;
+    case MouseButton::Extra:
+        return Common::Input::ButtonNames::ButtonExtra;
+    case MouseButton::Undefined:
+    default:
+        return Common::Input::ButtonNames::Undefined;
+    }
+}
+
 Common::Input::ButtonNames Mouse::GetUIName(const Common::ParamPackage& params) const {
     if (params.Has("button")) {
-        return Common::Input::ButtonNames::Value;
+        return GetUIButtonName(params);
     }
     if (params.Has("axis")) {
         return Common::Input::ButtonNames::Value;
     }
+    if (params.Has("axis_x") && params.Has("axis_y") && params.Has("axis_z")) {
+        return Common::Input::ButtonNames::Engine;
+    }
 
     return Common::Input::ButtonNames::Invalid;
 }

src/input_common/drivers/mouse.h

@@ -69,6 +69,8 @@ private:
     void UpdateThread(std::stop_token stop_token);
     void StopPanning();
 
+    Common::Input::ButtonNames GetUIButtonName(const Common::ParamPackage& params) const;
+
     Common::Vec2<int> mouse_origin;
     Common::Vec2<int> last_mouse_position;
     Common::Vec2<float> last_mouse_change;

src/input_common/drivers/sdl_driver.cpp

@@ -198,9 +198,9 @@ public:
         if (sdl_controller) {
             switch (SDL_GameControllerGetType(sdl_controller.get())) {
             case SDL_CONTROLLER_TYPE_XBOX360:
-                return "XBox 360 Controller";
+                return "Xbox 360 Controller";
             case SDL_CONTROLLER_TYPE_XBOXONE:
-                return "XBox One Controller";
+                return "Xbox One Controller";
             case SDL_CONTROLLER_TYPE_PS3:
                 return "DualShock 3 Controller";
             case SDL_CONTROLLER_TYPE_PS4:

src/input_common/input_engine.cpp

@@ -173,7 +173,7 @@ void InputEngine::ResetButtonState() {
             SetButton(controller.first, button.first, false);
         }
         for (const auto& button : controller.second.hat_buttons) {
-            SetHatButton(controller.first, button.first, false);
+            SetHatButton(controller.first, button.first, 0);
         }
     }
 }

src/input_common/input_mapping.cpp

@@ -143,6 +143,19 @@ void MappingFactory::RegisterMotion(const MappingData& data) {
     }
     new_input.Set("port", static_cast<int>(data.pad.port));
     new_input.Set("pad", static_cast<int>(data.pad.pad));
+
+    // If engine is mouse map the mouse position as 3 axis motion
+    if (data.engine == "mouse") {
+        new_input.Set("axis_x", 1);
+        new_input.Set("invert_x", "-");
+        new_input.Set("axis_y", 0);
+        new_input.Set("axis_z", 4);
+        new_input.Set("range", 1.0f);
+        new_input.Set("deadzone", 0.0f);
+        input_queue.Push(new_input);
+        return;
+    }
+
     switch (data.type) {
     case EngineInputType::Button:
     case EngineInputType::HatButton:

src/video_core/host_shaders/astc_decoder.comp

@@ -155,9 +155,6 @@ uint SwizzleOffset(uvec2 pos) {
 // Replicates low num_bits such that [(to_bit - 1):(to_bit - 1 - from_bit)]
 // is the same as [(num_bits - 1):0] and repeats all the way down.
 uint Replicate(uint val, uint num_bits, uint to_bit) {
-    if (num_bits == 0 || to_bit == 0) {
-        return 0;
-    }
     const uint v = val & uint((1 << num_bits) - 1);
     uint res = v;
     uint reslen = num_bits;
@@ -187,42 +184,57 @@ uint ReplicateBitTo9(uint value) {
     return REPLICATE_1_BIT_TO_9_TABLE[value];
 }
 
-uint FastReplicateTo8(uint value, uint num_bits) {
-    switch (num_bits) {
-    case 1:
-        return REPLICATE_1_BIT_TO_8_TABLE[value];
-    case 2:
-        return REPLICATE_2_BIT_TO_8_TABLE[value];
-    case 3:
-        return REPLICATE_3_BIT_TO_8_TABLE[value];
-    case 4:
-        return REPLICATE_4_BIT_TO_8_TABLE[value];
-    case 5:
-        return REPLICATE_5_BIT_TO_8_TABLE[value];
-    case 6:
-        return REPLICATE_6_BIT_TO_8_TABLE[value];
-    case 7:
-        return REPLICATE_7_BIT_TO_8_TABLE[value];
-    case 8:
+uint FastReplicate(uint value, uint num_bits, uint to_bit) {
+    if (num_bits == 0) {
+        return 0;
+    }
+    if (num_bits == to_bit) {
         return value;
     }
-    return Replicate(value, num_bits, 8);
+    if (to_bit == 6) {
+        switch (num_bits) {
+        case 1:
+            return REPLICATE_1_BIT_TO_6_TABLE[value];
+        case 2:
+            return REPLICATE_2_BIT_TO_6_TABLE[value];
+        case 3:
+            return REPLICATE_3_BIT_TO_6_TABLE[value];
+        case 4:
+            return REPLICATE_4_BIT_TO_6_TABLE[value];
+        case 5:
+            return REPLICATE_5_BIT_TO_6_TABLE[value];
+        default:
+            break;
+        }
+    } else { /* if (to_bit == 8) */
+        switch (num_bits) {
+        case 1:
+            return REPLICATE_1_BIT_TO_8_TABLE[value];
+        case 2:
+            return REPLICATE_2_BIT_TO_8_TABLE[value];
+        case 3:
+            return REPLICATE_3_BIT_TO_8_TABLE[value];
+        case 4:
+            return REPLICATE_4_BIT_TO_8_TABLE[value];
+        case 5:
+            return REPLICATE_5_BIT_TO_8_TABLE[value];
+        case 6:
+            return REPLICATE_6_BIT_TO_8_TABLE[value];
+        case 7:
+            return REPLICATE_7_BIT_TO_8_TABLE[value];
+        default:
+            break;
+        }
+    }
+    return Replicate(value, num_bits, to_bit);
+}
+
+uint FastReplicateTo8(uint value, uint num_bits) {
+    return FastReplicate(value, num_bits, 8);
 }
 
 uint FastReplicateTo6(uint value, uint num_bits) {
-    switch (num_bits) {
-    case 1:
-        return REPLICATE_1_BIT_TO_6_TABLE[value];
-    case 2:
-        return REPLICATE_2_BIT_TO_6_TABLE[value];
-    case 3:
-        return REPLICATE_3_BIT_TO_6_TABLE[value];
-    case 4:
-        return REPLICATE_4_BIT_TO_6_TABLE[value];
-    case 5:
-        return REPLICATE_5_BIT_TO_6_TABLE[value];
-    }
-    return Replicate(value, num_bits, 6);
+    return FastReplicate(value, num_bits, 6);
 }
 
 uint Div3Floor(uint v) {

src/video_core/vulkan_common/vulkan_device.cpp

@@ -1038,7 +1038,7 @@ std::vector<const char*> Device::LoadExtensions(bool requires_surface) {
     }
     if (has_ext_shader_atomic_int64) {
         VkPhysicalDeviceShaderAtomicInt64Features atomic_int64;
-        atomic_int64.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_EXTENDED_DYNAMIC_STATE_FEATURES_EXT;
+        atomic_int64.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SHADER_ATOMIC_INT64_FEATURES;
         atomic_int64.pNext = nullptr;
         features.pNext = &atomic_int64;
         physical.GetFeatures2KHR(features);

src/yuzu/configuration/config.cpp

@@ -743,7 +743,10 @@ void Config::ReadUIValues() {
     qt_config->beginGroup(QStringLiteral("UI"));
 
     UISettings::values.theme =
-        ReadSetting(QStringLiteral("theme"), QString::fromUtf8(UISettings::themes[0].second))
+        ReadSetting(
+            QStringLiteral("theme"),
+            QString::fromUtf8(
+                UISettings::themes[static_cast<size_t>(UISettings::Theme::DarkColorful)].second))
             .toString();
     ReadBasicSetting(UISettings::values.enable_discord_presence);
     ReadBasicSetting(UISettings::values.select_user_on_boot);
@@ -1270,8 +1273,10 @@ void Config::SaveSystemValues() {
 void Config::SaveUIValues() {
     qt_config->beginGroup(QStringLiteral("UI"));
 
-    WriteSetting(QStringLiteral("theme"), UISettings::values.theme,
-                 QString::fromUtf8(UISettings::themes[0].second));
+    WriteSetting(
+        QStringLiteral("theme"), UISettings::values.theme,
+        QString::fromUtf8(
+            UISettings::themes[static_cast<size_t>(UISettings::Theme::DarkColorful)].second));
     WriteBasicSetting(UISettings::values.enable_discord_presence);
     WriteBasicSetting(UISettings::values.select_user_on_boot);
 

src/yuzu/configuration/configure_input_player.cpp

@@ -102,6 +102,16 @@ QString GetButtonName(Common::Input::ButtonNames button_name) {
         return QObject::tr("Share");
     case Common::Input::ButtonNames::Options:
         return QObject::tr("Options");
+    case Common::Input::ButtonNames::ButtonMouseWheel:
+        return QObject::tr("Wheel", "Indicates the mouse wheel");
+    case Common::Input::ButtonNames::ButtonBackward:
+        return QObject::tr("Backward");
+    case Common::Input::ButtonNames::ButtonForward:
+        return QObject::tr("Forward");
+    case Common::Input::ButtonNames::ButtonTask:
+        return QObject::tr("Task");
+    case Common::Input::ButtonNames::ButtonExtra:
+        return QObject::tr("Extra");
     default:
         return QObject::tr("[undefined]");
     }

src/yuzu/uisettings.h

@@ -29,6 +29,15 @@ struct Shortcut {
     ContextualShortcut shortcut;
 };
 
+enum class Theme {
+    Default,
+    DefaultColorful,
+    Dark,
+    DarkColorful,
+    MidnightBlue,
+    MidnightBlueColorful,
+};
+
 using Themes = std::array<std::pair<const char*, const char*>, 6>;
 extern const Themes themes;