131 files changed, 2139 insertions, 947 deletions

diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index 63dd9febf..19d16147d 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -24,6 +24,7 @@ if (MSVC)
     # /W3                 - Level 3 warnings
     # /MP                 - Multi-threaded compilation
     # /Zi                 - Output debugging information
+    # /Zm                 - Specifies the precompiled header memory allocation limit
     # /Zo                 - Enhanced debug info for optimized builds
     # /permissive-        - Enables stricter C++ standards conformance checks
     # /EHsc               - C++-only exception handling semantics
@@ -36,6 +37,7 @@ if (MSVC)
     add_compile_options(
         /MP
         /Zi
+        /Zm200
         /Zo
         /permissive-
         /EHsc
diff --git a/src/common/logging/filter.cpp b/src/common/logging/filter.cpp
index 42744c994..b898a652c 100644
--- a/src/common/logging/filter.cpp
+++ b/src/common/logging/filter.cpp
@@ -114,6 +114,7 @@ bool ParseFilterRule(Filter& instance, Iterator begin, Iterator end) {
     SUB(Service, NGCT)                                                                             \
     SUB(Service, NIFM)                                                                             \
     SUB(Service, NIM)                                                                              \
+    SUB(Service, NOTIF)                                                                            \
     SUB(Service, NPNS)                                                                             \
     SUB(Service, NS)                                                                               \
     SUB(Service, NVDRV)                                                                            \
diff --git a/src/common/logging/types.h b/src/common/logging/types.h
index 2d21fc483..9ed0c7ad6 100644
--- a/src/common/logging/types.h
+++ b/src/common/logging/types.h
@@ -82,6 +82,7 @@ enum class Class : u8 {
     Service_NGCT,      ///< The NGCT (No Good Content for Terra) service
     Service_NIFM,      ///< The NIFM (Network interface) service
     Service_NIM,       ///< The NIM service
+    Service_NOTIF,     ///< The NOTIF (Notification) service
     Service_NPNS,      ///< The NPNS service
     Service_NS,        ///< The NS services
     Service_NVDRV,     ///< The NVDRV (Nvidia driver) service
diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt
index eee8e2ccd..49bed614a 100644
--- a/src/core/CMakeLists.txt
+++ b/src/core/CMakeLists.txt
@@ -179,12 +179,15 @@ add_library(core STATIC
     hle/kernel/k_client_port.h
     hle/kernel/k_client_session.cpp
     hle/kernel/k_client_session.h
+    hle/kernel/k_code_memory.cpp
+    hle/kernel/k_code_memory.h
     hle/kernel/k_condition_variable.cpp
     hle/kernel/k_condition_variable.h
     hle/kernel/k_event.cpp
     hle/kernel/k_event.h
     hle/kernel/k_handle_table.cpp
     hle/kernel/k_handle_table.h
+    hle/kernel/k_light_condition_variable.cpp
     hle/kernel/k_light_condition_variable.h
     hle/kernel/k_light_lock.cpp
     hle/kernel/k_light_lock.h
@@ -237,6 +240,7 @@ add_library(core STATIC
     hle/kernel/k_system_control.h
     hle/kernel/k_thread.cpp
     hle/kernel/k_thread.h
+    hle/kernel/k_thread_queue.cpp
     hle/kernel/k_thread_queue.h
     hle/kernel/k_trace.h
     hle/kernel/k_transfer_memory.cpp
@@ -408,6 +412,8 @@ add_library(core STATIC
     hle/service/glue/glue.h
     hle/service/glue/glue_manager.cpp
     hle/service/glue/glue_manager.h
+    hle/service/glue/notif.cpp
+    hle/service/glue/notif.h
     hle/service/grc/grc.cpp
     hle/service/grc/grc.h
     hle/service/hid/hid.cpp
diff --git a/src/core/core.cpp b/src/core/core.cpp
index 473ab9f81..aa96f709b 100644
--- a/src/core/core.cpp
+++ b/src/core/core.cpp
@@ -521,12 +521,6 @@ const ARM_Interface& System::CurrentArmInterface() const {
     return impl->kernel.CurrentPhysicalCore().ArmInterface();
 }
 
-std::size_t System::CurrentCoreIndex() const {
-    std::size_t core = impl->kernel.GetCurrentHostThreadID();
-    ASSERT(core < Core::Hardware::NUM_CPU_CORES);
-    return core;
-}
-
 Kernel::PhysicalCore& System::CurrentPhysicalCore() {
     return impl->kernel.CurrentPhysicalCore();
 }
diff --git a/src/core/core.h b/src/core/core.h
index 645e5c241..52ff90359 100644
--- a/src/core/core.h
+++ b/src/core/core.h
@@ -208,9 +208,6 @@ public:
     /// Gets an ARM interface to the CPU core that is currently running
     [[nodiscard]] const ARM_Interface& CurrentArmInterface() const;
 
-    /// Gets the index of the currently running CPU core
-    [[nodiscard]] std::size_t CurrentCoreIndex() const;
-
     /// Gets the physical core for the CPU core that is currently running
     [[nodiscard]] Kernel::PhysicalCore& CurrentPhysicalCore();
 
diff --git a/src/core/cpu_manager.cpp b/src/core/cpu_manager.cpp
index 5d43c6e5d..cbcc54891 100644
--- a/src/core/cpu_manager.cpp
+++ b/src/core/cpu_manager.cpp
@@ -117,17 +117,18 @@ void CpuManager::MultiCoreRunGuestLoop() {
             physical_core = &kernel.CurrentPhysicalCore();
         }
         system.ExitDynarmicProfile();
-        physical_core->ArmInterface().ClearExclusiveState();
-        kernel.CurrentScheduler()->RescheduleCurrentCore();
+        {
+            Kernel::KScopedDisableDispatch dd(kernel);
+            physical_core->ArmInterface().ClearExclusiveState();
+        }
     }
 }
 
 void CpuManager::MultiCoreRunIdleThread() {
     auto& kernel = system.Kernel();
     while (true) {
-        auto& physical_core = kernel.CurrentPhysicalCore();
-        physical_core.Idle();
-        kernel.CurrentScheduler()->RescheduleCurrentCore();
+        Kernel::KScopedDisableDispatch dd(kernel);
+        kernel.CurrentPhysicalCore().Idle();
     }
 }
 
@@ -135,12 +136,12 @@ void CpuManager::MultiCoreRunSuspendThread() {
     auto& kernel = system.Kernel();
     kernel.CurrentScheduler()->OnThreadStart();
     while (true) {
-        auto core = kernel.GetCurrentHostThreadID();
+        auto core = kernel.CurrentPhysicalCoreIndex();
         auto& scheduler = *kernel.CurrentScheduler();
         Kernel::KThread* current_thread = scheduler.GetCurrentThread();
         Common::Fiber::YieldTo(current_thread->GetHostContext(), *core_data[core].host_context);
         ASSERT(scheduler.ContextSwitchPending());
-        ASSERT(core == kernel.GetCurrentHostThreadID());
+        ASSERT(core == kernel.CurrentPhysicalCoreIndex());
         scheduler.RescheduleCurrentCore();
     }
 }
@@ -346,13 +347,9 @@ void CpuManager::RunThread(std::stop_token stop_token, std::size_t core) {
             sc_sync_first_use = false;
         }
 
-        // Abort if emulation was killed before the session really starts
-        if (!system.IsPoweredOn()) {
-            return;
-        }
-
+        // Emulation was stopped
         if (stop_token.stop_requested()) {
-            break;
+            return;
         }
 
         auto current_thread = system.Kernel().CurrentScheduler()->GetCurrentThread();
diff --git a/src/core/hid/hid_core.cpp b/src/core/hid/hid_core.cpp
index 0c3eb5a62..a1c3bbb57 100644
--- a/src/core/hid/hid_core.cpp
+++ b/src/core/hid/hid_core.cpp
@@ -145,6 +145,16 @@ NpadIdType HIDCore::GetFirstNpadId() const {
     return NpadIdType::Player1;
 }
 
+NpadIdType HIDCore::GetFirstDisconnectedNpadId() const {
+    for (std::size_t player_index = 0; player_index < available_controllers; ++player_index) {
+        const auto* const controller = GetEmulatedControllerByIndex(player_index);
+        if (!controller->IsConnected()) {
+            return controller->GetNpadIdType();
+        }
+    }
+    return NpadIdType::Player1;
+}
+
 void HIDCore::EnableAllControllerConfiguration() {
     player_1->EnableConfiguration();
     player_2->EnableConfiguration();
diff --git a/src/core/hid/hid_core.h b/src/core/hid/hid_core.h
index 2fb0f7e19..837f7de49 100644
--- a/src/core/hid/hid_core.h
+++ b/src/core/hid/hid_core.h
@@ -45,6 +45,9 @@ public:
     /// Returns the first connected npad id
     NpadIdType GetFirstNpadId() const;
 
+    /// Returns the first disconnected npad id
+    NpadIdType GetFirstDisconnectedNpadId() const;
+
     /// Sets all emulated controllers into configuring mode.
     void EnableAllControllerConfiguration();
 
diff --git a/src/core/hle/kernel/init/init_slab_setup.cpp b/src/core/hle/kernel/init/init_slab_setup.cpp
index 8ff0f695d..36fc0944a 100644
--- a/src/core/hle/kernel/init/init_slab_setup.cpp
+++ b/src/core/hle/kernel/init/init_slab_setup.cpp
@@ -9,6 +9,7 @@
 #include "core/core.h"
 #include "core/hardware_properties.h"
 #include "core/hle/kernel/init/init_slab_setup.h"
+#include "core/hle/kernel/k_code_memory.h"
 #include "core/hle/kernel/k_event.h"
 #include "core/hle/kernel/k_memory_layout.h"
 #include "core/hle/kernel/k_memory_manager.h"
@@ -32,6 +33,7 @@ namespace Kernel::Init {
     HANDLER(KPort, (SLAB_COUNT(KPort)), ##__VA_ARGS__)                                             \
     HANDLER(KSharedMemory, (SLAB_COUNT(KSharedMemory)), ##__VA_ARGS__)                             \
     HANDLER(KTransferMemory, (SLAB_COUNT(KTransferMemory)), ##__VA_ARGS__)                         \
+    HANDLER(KCodeMemory, (SLAB_COUNT(KCodeMemory)), ##__VA_ARGS__)                                 \
     HANDLER(KSession, (SLAB_COUNT(KSession)), ##__VA_ARGS__)                                       \
     HANDLER(KResourceLimit, (SLAB_COUNT(KResourceLimit)), ##__VA_ARGS__)
 
diff --git a/src/core/hle/kernel/k_address_arbiter.cpp b/src/core/hle/kernel/k_address_arbiter.cpp
index 1b429bc1e..783c69858 100644
--- a/src/core/hle/kernel/k_address_arbiter.cpp
+++ b/src/core/hle/kernel/k_address_arbiter.cpp
@@ -8,6 +8,7 @@
 #include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
 #include "core/hle/kernel/k_thread.h"
+#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"
@@ -28,7 +29,7 @@ bool ReadFromUser(Core::System& system, s32* out, VAddr address) {
 
 bool DecrementIfLessThan(Core::System& system, s32* out, VAddr address, s32 value) {
     auto& monitor = system.Monitor();
-    const auto current_core = system.CurrentCoreIndex();
+    const auto current_core = system.Kernel().CurrentPhysicalCoreIndex();
 
     // TODO(bunnei): We should disable interrupts here via KScopedInterruptDisable.
     // TODO(bunnei): We should call CanAccessAtomic(..) here.
@@ -58,7 +59,7 @@ bool DecrementIfLessThan(Core::System& system, s32* out, VAddr address, s32 valu
 
 bool UpdateIfEqual(Core::System& system, s32* out, VAddr address, s32 value, s32 new_value) {
     auto& monitor = system.Monitor();
-    const auto current_core = system.CurrentCoreIndex();
+    const auto current_core = system.Kernel().CurrentPhysicalCoreIndex();
 
     // TODO(bunnei): We should disable interrupts here via KScopedInterruptDisable.
     // TODO(bunnei): We should call CanAccessAtomic(..) here.
@@ -85,6 +86,27 @@ bool UpdateIfEqual(Core::System& system, s32* out, VAddr address, s32 value, s32
     return true;
 }
 
+class ThreadQueueImplForKAddressArbiter final : public KThreadQueue {
+public:
+    explicit ThreadQueueImplForKAddressArbiter(KernelCore& kernel_, KAddressArbiter::ThreadTree* t)
+        : KThreadQueue(kernel_), m_tree(t) {}
+
+    void CancelWait(KThread* waiting_thread, ResultCode wait_result,
+                    bool cancel_timer_task) override {
+        // If the thread is waiting on an address arbiter, remove it from the tree.
+        if (waiting_thread->IsWaitingForAddressArbiter()) {
+            m_tree->erase(m_tree->iterator_to(*waiting_thread));
+            waiting_thread->ClearAddressArbiter();
+        }
+
+        // Invoke the base cancel wait handler.
+        KThreadQueue::CancelWait(waiting_thread, wait_result, cancel_timer_task);
+    }
+
+private:
+    KAddressArbiter::ThreadTree* m_tree;
+};
+
 } // namespace
 
 ResultCode KAddressArbiter::Signal(VAddr addr, s32 count) {
@@ -96,14 +118,14 @@ ResultCode KAddressArbiter::Signal(VAddr addr, s32 count) {
         auto it = thread_tree.nfind_light({addr, -1});
         while ((it != thread_tree.end()) && (count <= 0 || num_waiters < count) &&
                (it->GetAddressArbiterKey() == addr)) {
+            // End the thread's wait.
             KThread* target_thread = std::addressof(*it);
-            target_thread->SetSyncedObject(nullptr, ResultSuccess);
+            target_thread->EndWait(ResultSuccess);
 
             ASSERT(target_thread->IsWaitingForAddressArbiter());
-            target_thread->Wakeup();
+            target_thread->ClearAddressArbiter();
 
             it = thread_tree.erase(it);
-            target_thread->ClearAddressArbiter();
             ++num_waiters;
         }
     }
@@ -129,14 +151,14 @@ ResultCode KAddressArbiter::SignalAndIncrementIfEqual(VAddr addr, s32 value, s32
         auto it = thread_tree.nfind_light({addr, -1});
         while ((it != thread_tree.end()) && (count <= 0 || num_waiters < count) &&
                (it->GetAddressArbiterKey() == addr)) {
+            // End the thread's wait.
             KThread* target_thread = std::addressof(*it);
-            target_thread->SetSyncedObject(nullptr, ResultSuccess);
+            target_thread->EndWait(ResultSuccess);
 
             ASSERT(target_thread->IsWaitingForAddressArbiter());
-            target_thread->Wakeup();
+            target_thread->ClearAddressArbiter();
 
             it = thread_tree.erase(it);
-            target_thread->ClearAddressArbiter();
             ++num_waiters;
         }
     }
@@ -197,14 +219,14 @@ ResultCode KAddressArbiter::SignalAndModifyByWaitingCountIfEqual(VAddr addr, s32
 
         while ((it != thread_tree.end()) && (count <= 0 || num_waiters < count) &&
                (it->GetAddressArbiterKey() == addr)) {
+            // End the thread's wait.
             KThread* target_thread = std::addressof(*it);
-            target_thread->SetSyncedObject(nullptr, ResultSuccess);
+            target_thread->EndWait(ResultSuccess);
 
             ASSERT(target_thread->IsWaitingForAddressArbiter());
-            target_thread->Wakeup();
+            target_thread->ClearAddressArbiter();
 
             it = thread_tree.erase(it);
-            target_thread->ClearAddressArbiter();
             ++num_waiters;
         }
     }
@@ -214,6 +236,7 @@ ResultCode KAddressArbiter::SignalAndModifyByWaitingCountIfEqual(VAddr addr, s32
 ResultCode KAddressArbiter::WaitIfLessThan(VAddr addr, s32 value, bool decrement, s64 timeout) {
     // Prepare to wait.
     KThread* cur_thread = kernel.CurrentScheduler()->GetCurrentThread();
+    ThreadQueueImplForKAddressArbiter wait_queue(kernel, std::addressof(thread_tree));
 
     {
         KScopedSchedulerLockAndSleep slp{kernel, cur_thread, timeout};
@@ -224,9 +247,6 @@ ResultCode KAddressArbiter::WaitIfLessThan(VAddr addr, s32 value, bool decrement
             return ResultTerminationRequested;
         }
 
-        // Set the synced object.
-        cur_thread->SetSyncedObject(nullptr, ResultTimedOut);
-
         // Read the value from userspace.
         s32 user_value{};
         bool succeeded{};
@@ -256,31 +276,20 @@ ResultCode KAddressArbiter::WaitIfLessThan(VAddr addr, s32 value, bool decrement
         // Set the arbiter.
         cur_thread->SetAddressArbiter(&thread_tree, addr);
         thread_tree.insert(*cur_thread);
-        cur_thread->SetState(ThreadState::Waiting);
-        cur_thread->SetWaitReasonForDebugging(ThreadWaitReasonForDebugging::Arbitration);
-    }
-
-    // Cancel the timer wait.
-    kernel.TimeManager().UnscheduleTimeEvent(cur_thread);
 
-    // Remove from the address arbiter.
-    {
-        KScopedSchedulerLock sl(kernel);
-
-        if (cur_thread->IsWaitingForAddressArbiter()) {
-            thread_tree.erase(thread_tree.iterator_to(*cur_thread));
-            cur_thread->ClearAddressArbiter();
-        }
+        // Wait for the thread to finish.
+        cur_thread->BeginWait(std::addressof(wait_queue));
+        cur_thread->SetWaitReasonForDebugging(ThreadWaitReasonForDebugging::Arbitration);
     }
 
     // Get the result.
-    KSynchronizationObject* dummy{};
-    return cur_thread->GetWaitResult(&dummy);
+    return cur_thread->GetWaitResult();
 }
 
 ResultCode KAddressArbiter::WaitIfEqual(VAddr addr, s32 value, s64 timeout) {
     // Prepare to wait.
     KThread* cur_thread = kernel.CurrentScheduler()->GetCurrentThread();
+    ThreadQueueImplForKAddressArbiter wait_queue(kernel, std::addressof(thread_tree));
 
     {
         KScopedSchedulerLockAndSleep slp{kernel, cur_thread, timeout};
@@ -291,9 +300,6 @@ ResultCode KAddressArbiter::WaitIfEqual(VAddr addr, s32 value, s64 timeout) {
             return ResultTerminationRequested;
         }
 
-        // Set the synced object.
-        cur_thread->SetSyncedObject(nullptr, ResultTimedOut);
-
         // Read the value from userspace.
         s32 user_value{};
         if (!ReadFromUser(system, &user_value, addr)) {
@@ -316,26 +322,14 @@ ResultCode KAddressArbiter::WaitIfEqual(VAddr addr, s32 value, s64 timeout) {
         // Set the arbiter.
         cur_thread->SetAddressArbiter(&thread_tree, addr);
         thread_tree.insert(*cur_thread);
-        cur_thread->SetState(ThreadState::Waiting);
-        cur_thread->SetWaitReasonForDebugging(ThreadWaitReasonForDebugging::Arbitration);
-    }
-
-    // Cancel the timer wait.
-    kernel.TimeManager().UnscheduleTimeEvent(cur_thread);
 
-    // Remove from the address arbiter.
-    {
-        KScopedSchedulerLock sl(kernel);
-
-        if (cur_thread->IsWaitingForAddressArbiter()) {
-            thread_tree.erase(thread_tree.iterator_to(*cur_thread));
-            cur_thread->ClearAddressArbiter();
-        }
+        // Wait for the thread to finish.
+        cur_thread->BeginWait(std::addressof(wait_queue));
+        cur_thread->SetWaitReasonForDebugging(ThreadWaitReasonForDebugging::Arbitration);
     }
 
     // Get the result.
-    KSynchronizationObject* dummy{};
-    return cur_thread->GetWaitResult(&dummy);
+    return cur_thread->GetWaitResult();
 }
 
 } // namespace Kernel
diff --git a/src/core/hle/kernel/k_auto_object.h b/src/core/hle/kernel/k_auto_object.h
index e4fcdbc67..165b76747 100644
--- a/src/core/hle/kernel/k_auto_object.h
+++ b/src/core/hle/kernel/k_auto_object.h
@@ -170,6 +170,10 @@ public:
         }
     }
 
+    const std::string& GetName() const {
+        return name;
+    }
+
 private:
     void RegisterWithKernel();
     void UnregisterWithKernel();
diff --git a/src/core/hle/kernel/k_class_token.cpp b/src/core/hle/kernel/k_class_token.cpp
index 0be0027be..21e2fe494 100644
--- a/src/core/hle/kernel/k_class_token.cpp
+++ b/src/core/hle/kernel/k_class_token.cpp
@@ -6,6 +6,7 @@
 #include "core/hle/kernel/k_class_token.h"
 #include "core/hle/kernel/k_client_port.h"
 #include "core/hle/kernel/k_client_session.h"
+#include "core/hle/kernel/k_code_memory.h"
 #include "core/hle/kernel/k_event.h"
 #include "core/hle/kernel/k_port.h"
 #include "core/hle/kernel/k_process.h"
@@ -48,7 +49,7 @@ static_assert(ClassToken<KWritableEvent> == 0b10001001'00000000);
 static_assert(ClassToken<KTransferMemory> == 0b10010001'00000000);
 // static_assert(ClassToken<KDeviceAddressSpace> == 0b01100001'00000000);
 // static_assert(ClassToken<KSessionRequest> == 0b10100001'00000000);
-// static_assert(ClassToken<KCodeMemory> == 0b11000001'00000000);
+static_assert(ClassToken<KCodeMemory> == 0b11000001'00000000);
 
 // Ensure that the token hierarchy is correct.
 
@@ -79,7 +80,7 @@ static_assert(ClassToken<KWritableEvent> == ((0b10001001 << 8) | ClassToken<KAut
 static_assert(ClassToken<KTransferMemory> == ((0b10010001 << 8) | ClassToken<KAutoObject>));
 // static_assert(ClassToken<KDeviceAddressSpace> == ((0b01100001 << 8) | ClassToken<KAutoObject>));
 // static_assert(ClassToken<KSessionRequest> == ((0b10100001 << 8) | ClassToken<KAutoObject>));
-// static_assert(ClassToken<KCodeMemory> == ((0b11000001 << 8) | ClassToken<KAutoObject>));
+static_assert(ClassToken<KCodeMemory> == ((0b11000001 << 8) | ClassToken<KAutoObject>));
 
 // Ensure that the token hierarchy reflects the class hierarchy.
 
diff --git a/src/core/hle/kernel/k_code_memory.cpp b/src/core/hle/kernel/k_code_memory.cpp
new file mode 100644
index 000000000..d69f7ffb7
--- /dev/null
+++ b/src/core/hle/kernel/k_code_memory.cpp
@@ -0,0 +1,146 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/common_types.h"
+#include "core/device_memory.h"
+#include "core/hle/kernel/k_auto_object.h"
+#include "core/hle/kernel/k_code_memory.h"
+#include "core/hle/kernel/k_light_lock.h"
+#include "core/hle/kernel/k_memory_block.h"
+#include "core/hle/kernel/k_page_linked_list.h"
+#include "core/hle/kernel/k_page_table.h"
+#include "core/hle/kernel/k_process.h"
+#include "core/hle/kernel/slab_helpers.h"
+#include "core/hle/kernel/svc_types.h"
+#include "core/hle/result.h"
+
+namespace Kernel {
+
+KCodeMemory::KCodeMemory(KernelCore& kernel_)
+    : KAutoObjectWithSlabHeapAndContainer{kernel_}, m_lock(kernel_) {}
+
+ResultCode KCodeMemory::Initialize(Core::DeviceMemory& device_memory, VAddr addr, size_t size) {
+    // Set members.
+    m_owner = kernel.CurrentProcess();
+
+    // Get the owner page table.
+    auto& page_table = m_owner->PageTable();
+
+    // Construct the page group.
+    KMemoryInfo kBlockInfo = page_table.QueryInfo(addr);
+    m_page_group = KPageLinkedList(kBlockInfo.GetAddress(), kBlockInfo.GetNumPages());
+
+    // Lock the memory.
+    R_TRY(page_table.LockForCodeMemory(addr, size))
+
+    // Clear the memory.
+    for (const auto& block : m_page_group.Nodes()) {
+        std::memset(device_memory.GetPointer(block.GetAddress()), 0xFF, block.GetSize());
+    }
+
+    // Set remaining tracking members.
+    m_address = addr;
+    m_is_initialized = true;
+    m_is_owner_mapped = false;
+    m_is_mapped = false;
+
+    // We succeeded.
+    return ResultSuccess;
+}
+
+void KCodeMemory::Finalize() {
+    // Unlock.
+    if (!m_is_mapped && !m_is_owner_mapped) {
+        const size_t size = m_page_group.GetNumPages() * PageSize;
+        m_owner->PageTable().UnlockForCodeMemory(m_address, size);
+    }
+}
+
+ResultCode KCodeMemory::Map(VAddr address, size_t size) {
+    // Validate the size.
+    R_UNLESS(m_page_group.GetNumPages() == Common::DivideUp(size, PageSize), ResultInvalidSize);
+
+    // Lock ourselves.
+    KScopedLightLock lk(m_lock);
+
+    // Ensure we're not already mapped.
+    R_UNLESS(!m_is_mapped, ResultInvalidState);
+
+    // Map the memory.
+    R_TRY(kernel.CurrentProcess()->PageTable().MapPages(
+        address, m_page_group, KMemoryState::CodeOut, KMemoryPermission::UserReadWrite));
+
+    // Mark ourselves as mapped.
+    m_is_mapped = true;
+
+    return ResultSuccess;
+}
+
+ResultCode KCodeMemory::Unmap(VAddr address, size_t size) {
+    // Validate the size.
+    R_UNLESS(m_page_group.GetNumPages() == Common::DivideUp(size, PageSize), ResultInvalidSize);
+
+    // Lock ourselves.
+    KScopedLightLock lk(m_lock);
+
+    // Unmap the memory.
+    R_TRY(kernel.CurrentProcess()->PageTable().UnmapPages(address, m_page_group,
+                                                          KMemoryState::CodeOut));
+
+    // Mark ourselves as unmapped.
+    m_is_mapped = false;
+
+    return ResultSuccess;
+}
+
+ResultCode KCodeMemory::MapToOwner(VAddr address, size_t size, Svc::MemoryPermission perm) {
+    // Validate the size.
+    R_UNLESS(m_page_group.GetNumPages() == Common::DivideUp(size, PageSize), ResultInvalidSize);
+
+    // Lock ourselves.
+    KScopedLightLock lk(m_lock);
+
+    // Ensure we're not already mapped.
+    R_UNLESS(!m_is_owner_mapped, ResultInvalidState);
+
+    // Convert the memory permission.
+    KMemoryPermission k_perm{};
+    switch (perm) {
+    case Svc::MemoryPermission::Read:
+        k_perm = KMemoryPermission::UserRead;
+        break;
+    case Svc::MemoryPermission::ReadExecute:
+        k_perm = KMemoryPermission::UserReadExecute;
+        break;
+    default:
+        break;
+    }
+
+    // Map the memory.
+    R_TRY(
+        m_owner->PageTable().MapPages(address, m_page_group, KMemoryState::GeneratedCode, k_perm));
+
+    // Mark ourselves as mapped.
+    m_is_owner_mapped = true;
+
+    return ResultSuccess;
+}
+
+ResultCode KCodeMemory::UnmapFromOwner(VAddr address, size_t size) {
+    // Validate the size.
+    R_UNLESS(m_page_group.GetNumPages() == Common::DivideUp(size, PageSize), ResultInvalidSize);
+
+    // Lock ourselves.
+    KScopedLightLock lk(m_lock);
+
+    // Unmap the memory.
+    R_TRY(m_owner->PageTable().UnmapPages(address, m_page_group, KMemoryState::GeneratedCode));
+
+    // Mark ourselves as unmapped.
+    m_is_owner_mapped = false;
+
+    return ResultSuccess;
+}
+
+} // namespace Kernel
\ No newline at end of file
diff --git a/src/core/hle/kernel/k_code_memory.h b/src/core/hle/kernel/k_code_memory.h
new file mode 100644
index 000000000..e0ba19a53
--- /dev/null
+++ b/src/core/hle/kernel/k_code_memory.h
@@ -0,0 +1,66 @@
+// Copyright 2021 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 "core/device_memory.h"
+#include "core/hle/kernel/k_auto_object.h"
+#include "core/hle/kernel/k_light_lock.h"
+#include "core/hle/kernel/k_page_linked_list.h"
+#include "core/hle/kernel/k_process.h"
+#include "core/hle/kernel/slab_helpers.h"
+#include "core/hle/kernel/svc_types.h"
+#include "core/hle/result.h"
+
+namespace Kernel {
+
+enum class CodeMemoryOperation : u32 {
+    Map = 0,
+    MapToOwner = 1,
+    Unmap = 2,
+    UnmapFromOwner = 3,
+};
+
+class KCodeMemory final
+    : public KAutoObjectWithSlabHeapAndContainer<KCodeMemory, KAutoObjectWithList> {
+    KERNEL_AUTOOBJECT_TRAITS(KCodeMemory, KAutoObject);
+
+public:
+    explicit KCodeMemory(KernelCore& kernel_);
+
+    ResultCode Initialize(Core::DeviceMemory& device_memory, VAddr address, size_t size);
+    void Finalize();
+
+    ResultCode Map(VAddr address, size_t size);
+    ResultCode Unmap(VAddr address, size_t size);
+    ResultCode MapToOwner(VAddr address, size_t size, Svc::MemoryPermission perm);
+    ResultCode UnmapFromOwner(VAddr address, size_t size);
+
+    bool IsInitialized() const {
+        return m_is_initialized;
+    }
+    static void PostDestroy([[maybe_unused]] uintptr_t arg) {}
+
+    KProcess* GetOwner() const {
+        return m_owner;
+    }
+    VAddr GetSourceAddress() const {
+        return m_address;
+    }
+    size_t GetSize() const {
+        return m_is_initialized ? m_page_group.GetNumPages() * PageSize : 0;
+    }
+
+private:
+    KPageLinkedList m_page_group{};
+    KProcess* m_owner{};
+    VAddr m_address{};
+    KLightLock m_lock;
+    bool m_is_initialized{};
+    bool m_is_owner_mapped{};
+    bool m_is_mapped{};
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_condition_variable.cpp b/src/core/hle/kernel/k_condition_variable.cpp
index 7fa9b8cc3..aadcc297a 100644
--- a/src/core/hle/kernel/k_condition_variable.cpp
+++ b/src/core/hle/kernel/k_condition_variable.cpp
@@ -11,6 +11,7 @@
 #include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
 #include "core/hle/kernel/k_synchronization_object.h"
 #include "core/hle/kernel/k_thread.h"
+#include "core/hle/kernel/k_thread_queue.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/svc_common.h"
 #include "core/hle/kernel/svc_results.h"
@@ -33,7 +34,7 @@ bool WriteToUser(Core::System& system, VAddr address, const u32* p) {
 bool UpdateLockAtomic(Core::System& system, u32* out, VAddr address, u32 if_zero,
                       u32 new_orr_mask) {
     auto& monitor = system.Monitor();
-    const auto current_core = system.CurrentCoreIndex();
+    const auto current_core = system.Kernel().CurrentPhysicalCoreIndex();
 
     // Load the value from the address.
     const auto expected = monitor.ExclusiveRead32(current_core, address);
@@ -57,6 +58,48 @@ bool UpdateLockAtomic(Core::System& system, u32* out, VAddr address, u32 if_zero
     return true;
 }
 
+class ThreadQueueImplForKConditionVariableWaitForAddress final : public KThreadQueue {
+public:
+    explicit ThreadQueueImplForKConditionVariableWaitForAddress(KernelCore& kernel_)
+        : KThreadQueue(kernel_) {}
+
+    void CancelWait(KThread* waiting_thread, ResultCode wait_result,
+                    bool cancel_timer_task) override {
+        // Remove the thread as a waiter from its owner.
+        waiting_thread->GetLockOwner()->RemoveWaiter(waiting_thread);
+
+        // Invoke the base cancel wait handler.
+        KThreadQueue::CancelWait(waiting_thread, wait_result, cancel_timer_task);
+    }
+};
+
+class ThreadQueueImplForKConditionVariableWaitConditionVariable final : public KThreadQueue {
+private:
+    KConditionVariable::ThreadTree* m_tree;
+
+public:
+    explicit ThreadQueueImplForKConditionVariableWaitConditionVariable(
+        KernelCore& kernel_, KConditionVariable::ThreadTree* t)
+        : KThreadQueue(kernel_), m_tree(t) {}
+
+    void CancelWait(KThread* waiting_thread, ResultCode wait_result,
+                    bool cancel_timer_task) override {
+        // Remove the thread as a waiter from its owner.
+        if (KThread* owner = waiting_thread->GetLockOwner(); owner != nullptr) {
+            owner->RemoveWaiter(waiting_thread);
+        }
+
+        // If the thread is waiting on a condvar, remove it from the tree.
+        if (waiting_thread->IsWaitingForConditionVariable()) {
+            m_tree->erase(m_tree->iterator_to(*waiting_thread));
+            waiting_thread->ClearConditionVariable();
+        }
+
+        // Invoke the base cancel wait handler.
+        KThreadQueue::CancelWait(waiting_thread, wait_result, cancel_timer_task);
+    }
+};
+
 } // namespace
 
 KConditionVariable::KConditionVariable(Core::System& system_)
@@ -78,84 +121,77 @@ ResultCode KConditionVariable::SignalToAddress(VAddr addr) {
 
         // Determine the next tag.
         u32 next_value{};
-        if (next_owner_thread) {
+        if (next_owner_thread != nullptr) {
             next_value = next_owner_thread->GetAddressKeyValue();
             if (num_waiters > 1) {
                 next_value |= Svc::HandleWaitMask;
             }
 
-            next_owner_thread->SetSyncedObject(nullptr, ResultSuccess);
-            next_owner_thread->Wakeup();
-        }
-
-        // Write the value to userspace.
-        if (!WriteToUser(system, addr, std::addressof(next_value))) {
-            if (next_owner_thread) {
-                next_owner_thread->SetSyncedObject(nullptr, ResultInvalidCurrentMemory);
+            // Write the value to userspace.
+            ResultCode result{ResultSuccess};
+            if (WriteToUser(system, addr, std::addressof(next_value))) [[likely]] {
+                result = ResultSuccess;
+            } else {
+                result = ResultInvalidCurrentMemory;
             }
 
-            return ResultInvalidCurrentMemory;
+            // Signal the next owner thread.
+            next_owner_thread->EndWait(result);
+            return result;
+        } else {
+            // Just write the value to userspace.
+            R_UNLESS(WriteToUser(system, addr, std::addressof(next_value)),
+                     ResultInvalidCurrentMemory);
+
+            return ResultSuccess;
         }
     }
-
-    return ResultSuccess;
 }
 
 ResultCode KConditionVariable::WaitForAddress(Handle handle, VAddr addr, u32 value) {
     KThread* cur_thread = kernel.CurrentScheduler()->GetCurrentThread();
+    ThreadQueueImplForKConditionVariableWaitForAddress wait_queue(kernel);
 
     // Wait for the address.
+    KThread* owner_thread{};
     {
-        KScopedAutoObject<KThread> owner_thread;
-        ASSERT(owner_thread.IsNull());
-        {
-            KScopedSchedulerLock sl(kernel);
-            cur_thread->SetSyncedObject(nullptr, ResultSuccess);
+        KScopedSchedulerLock sl(kernel);
 
-            // Check if the thread should terminate.
-            R_UNLESS(!cur_thread->IsTerminationRequested(), ResultTerminationRequested);
+        // Check if the thread should terminate.
+        R_UNLESS(!cur_thread->IsTerminationRequested(), ResultTerminationRequested);
 
-            {
-                // Read the tag from userspace.
-                u32 test_tag{};
-                R_UNLESS(ReadFromUser(system, std::addressof(test_tag), addr),
-                         ResultInvalidCurrentMemory);
-
-                // If the tag isn't the handle (with wait mask), we're done.
-                R_UNLESS(test_tag == (handle | Svc::HandleWaitMask), ResultSuccess);
-
-                // Get the lock owner thread.
-                owner_thread =
-                    kernel.CurrentProcess()->GetHandleTable().GetObjectWithoutPseudoHandle<KThread>(
-                        handle);
-                R_UNLESS(owner_thread.IsNotNull(), ResultInvalidHandle);
-
-                // Update the lock.
-                cur_thread->SetAddressKey(addr, value);
-                owner_thread->AddWaiter(cur_thread);
-                cur_thread->SetState(ThreadState::Waiting);
-                cur_thread->SetWaitReasonForDebugging(ThreadWaitReasonForDebugging::ConditionVar);
-                cur_thread->SetMutexWaitAddressForDebugging(addr);
-            }
-        }
-        ASSERT(owner_thread.IsNotNull());
-    }
+        // Read the tag from userspace.
+        u32 test_tag{};
+        R_UNLESS(ReadFromUser(system, std::addressof(test_tag), addr), ResultInvalidCurrentMemory);
 
-    // Remove the thread as a waiter from the lock owner.
-    {
-        KScopedSchedulerLock sl(kernel);
-        KThread* owner_thread = cur_thread->GetLockOwner();
-        if (owner_thread != nullptr) {
-            owner_thread->RemoveWaiter(cur_thread);
-        }
+        // If the tag isn't the handle (with wait mask), we're done.
+        R_SUCCEED_IF(test_tag != (handle | Svc::HandleWaitMask));
+
+        // Get the lock owner thread.
+        owner_thread = kernel.CurrentProcess()
+                           ->GetHandleTable()
+                           .GetObjectWithoutPseudoHandle<KThread>(handle)
+                           .ReleasePointerUnsafe();
+        R_UNLESS(owner_thread != nullptr, ResultInvalidHandle);
+
+        // Update the lock.
+        cur_thread->SetAddressKey(addr, value);
+        owner_thread->AddWaiter(cur_thread);
+
+        // Begin waiting.
+        cur_thread->BeginWait(std::addressof(wait_queue));
+        cur_thread->SetWaitReasonForDebugging(ThreadWaitReasonForDebugging::ConditionVar);
+        cur_thread->SetMutexWaitAddressForDebugging(addr);
     }
 
+    // Close our reference to the owner thread, now that the wait is over.
+    owner_thread->Close();
+
     // Get the wait result.
-    KSynchronizationObject* dummy{};
-    return cur_thread->GetWaitResult(std::addressof(dummy));
+    return cur_thread->GetWaitResult();
 }
 
-KThread* KConditionVariable::SignalImpl(KThread* thread) {
+void KConditionVariable::SignalImpl(KThread* thread) {
     // Check pre-conditions.
     ASSERT(kernel.GlobalSchedulerContext().IsLocked());
 
@@ -169,18 +205,16 @@ KThread* KConditionVariable::SignalImpl(KThread* thread) {
         // TODO(bunnei): We should disable interrupts here via KScopedInterruptDisable.
         // TODO(bunnei): We should call CanAccessAtomic(..) here.
         can_access = true;
-        if (can_access) {
+        if (can_access) [[likely]] {
             UpdateLockAtomic(system, std::addressof(prev_tag), address, own_tag,
                              Svc::HandleWaitMask);
         }
     }
 
-    KThread* thread_to_close = nullptr;
-    if (can_access) {
+    if (can_access) [[likely]] {
         if (prev_tag == Svc::InvalidHandle) {
             // If nobody held the lock previously, we're all good.
-            thread->SetSyncedObject(nullptr, ResultSuccess);
-            thread->Wakeup();
+            thread->EndWait(ResultSuccess);
         } else {
             // Get the previous owner.
             KThread* owner_thread = kernel.CurrentProcess()
@@ -189,33 +223,22 @@ KThread* KConditionVariable::SignalImpl(KThread* thread) {
                                             static_cast<Handle>(prev_tag & ~Svc::HandleWaitMask))
                                         .ReleasePointerUnsafe();
 
-            if (owner_thread) {
+            if (owner_thread) [[likely]] {
                 // Add the thread as a waiter on the owner.
                 owner_thread->AddWaiter(thread);
-                thread_to_close = owner_thread;
+                owner_thread->Close();
             } else {
                 // The lock was tagged with a thread that doesn't exist.
-                thread->SetSyncedObject(nullptr, ResultInvalidState);
-                thread->Wakeup();
+                thread->EndWait(ResultInvalidState);
             }
         }
     } else {
         // If the address wasn't accessible, note so.
-        thread->SetSyncedObject(nullptr, ResultInvalidCurrentMemory);
-        thread->Wakeup();
+        thread->EndWait(ResultInvalidCurrentMemory);
     }
-
-    return thread_to_close;
 }
 
 void KConditionVariable::Signal(u64 cv_key, s32 count) {
-    // Prepare for signaling.
-    constexpr int MaxThreads = 16;
-
-    KLinkedList<KThread> thread_list{kernel};
-    std::array<KThread*, MaxThreads> thread_array;
-    s32 num_to_close{};
-
     // Perform signaling.
     s32 num_waiters{};
     {
@@ -226,14 +249,7 @@ void KConditionVariable::Signal(u64 cv_key, s32 count) {
                (it->GetConditionVariableKey() == cv_key)) {
             KThread* target_thread = std::addressof(*it);
 
-            if (KThread* thread = SignalImpl(target_thread); thread != nullptr) {
-                if (num_to_close < MaxThreads) {
-                    thread_array[num_to_close++] = thread;
-                } else {
-                    thread_list.push_back(*thread);
-                }
-            }
-
+            this->SignalImpl(target_thread);
             it = thread_tree.erase(it);
             target_thread->ClearConditionVariable();
             ++num_waiters;
@@ -245,27 +261,16 @@ void KConditionVariable::Signal(u64 cv_key, s32 count) {
             WriteToUser(system, cv_key, std::addressof(has_waiter_flag));
         }
     }
-
-    // Close threads in the array.
-    for (auto i = 0; i < num_to_close; ++i) {
-        thread_array[i]->Close();
-    }
-
-    // Close threads in the list.
-    for (auto it = thread_list.begin(); it != thread_list.end(); it = thread_list.erase(it)) {
-        (*it).Close();
-    }
 }
 
 ResultCode KConditionVariable::Wait(VAddr addr, u64 key, u32 value, s64 timeout) {
     // Prepare to wait.
-    KThread* cur_thread = kernel.CurrentScheduler()->GetCurrentThread();
+    KThread* cur_thread = GetCurrentThreadPointer(kernel);
+    ThreadQueueImplForKConditionVariableWaitConditionVariable wait_queue(
+        kernel, std::addressof(thread_tree));
 
     {
-        KScopedSchedulerLockAndSleep slp{kernel, cur_thread, timeout};
-
-        // Set the synced object.
-        cur_thread->SetSyncedObject(nullptr, ResultTimedOut);
+        KScopedSchedulerLockAndSleep slp(kernel, cur_thread, timeout);
 
         // Check that the thread isn't terminating.
         if (cur_thread->IsTerminationRequested()) {
@@ -290,8 +295,7 @@ ResultCode KConditionVariable::Wait(VAddr addr, u64 key, u32 value, s64 timeout)
                 }
 
                 // Wake up the next owner.
-                next_owner_thread->SetSyncedObject(nullptr, ResultSuccess);
-                next_owner_thread->Wakeup();
+                next_owner_thread->EndWait(ResultSuccess);
             }
 
             // Write to the cv key.
@@ -308,40 +312,21 @@ ResultCode KConditionVariable::Wait(VAddr addr, u64 key, u32 value, s64 timeout)
             }
         }
 
-        // Update condition variable tracking.
-        {
-            cur_thread->SetConditionVariable(std::addressof(thread_tree), addr, key, value);
-            thread_tree.insert(*cur_thread);
-        }
+        // If timeout is zero, time out.
+        R_UNLESS(timeout != 0, ResultTimedOut);
 
-        // If the timeout is non-zero, set the thread as waiting.
-        if (timeout != 0) {
-            cur_thread->SetState(ThreadState::Waiting);
-            cur_thread->SetWaitReasonForDebugging(ThreadWaitReasonForDebugging::ConditionVar);
-            cur_thread->SetMutexWaitAddressForDebugging(addr);
-        }
-    }
-
-    // Cancel the timer wait.
-    kernel.TimeManager().UnscheduleTimeEvent(cur_thread);
-
-    // Remove from the condition variable.
-    {
-        KScopedSchedulerLock sl(kernel);
-
-        if (KThread* owner = cur_thread->GetLockOwner(); owner != nullptr) {
-            owner->RemoveWaiter(cur_thread);
-        }
+        // Update condition variable tracking.
+        cur_thread->SetConditionVariable(std::addressof(thread_tree), addr, key, value);
+        thread_tree.insert(*cur_thread);
 
-        if (cur_thread->IsWaitingForConditionVariable()) {
-            thread_tree.erase(thread_tree.iterator_to(*cur_thread));
-            cur_thread->ClearConditionVariable();
-        }
+        // Begin waiting.
+        cur_thread->BeginWait(std::addressof(wait_queue));
+        cur_thread->SetWaitReasonForDebugging(ThreadWaitReasonForDebugging::ConditionVar);
+        cur_thread->SetMutexWaitAddressForDebugging(addr);
     }
 
-    // Get the result.
-    KSynchronizationObject* dummy{};
-    return cur_thread->GetWaitResult(std::addressof(dummy));
+    // Get the wait result.
+    return cur_thread->GetWaitResult();
 }
 
 } // namespace Kernel
diff --git a/src/core/hle/kernel/k_condition_variable.h b/src/core/hle/kernel/k_condition_variable.h
index 861dbd420..5e4815d08 100644
--- a/src/core/hle/kernel/k_condition_variable.h
+++ b/src/core/hle/kernel/k_condition_variable.h
@@ -34,7 +34,7 @@ public:
     [[nodiscard]] ResultCode Wait(VAddr addr, u64 key, u32 value, s64 timeout);
 
 private:
-    [[nodiscard]] KThread* SignalImpl(KThread* thread);
+    void SignalImpl(KThread* thread);
 
     ThreadTree thread_tree;
 
diff --git a/src/core/hle/kernel/k_handle_table.cpp b/src/core/hle/kernel/k_handle_table.cpp
index e90fc0628..cf95f0852 100644
--- a/src/core/hle/kernel/k_handle_table.cpp
+++ b/src/core/hle/kernel/k_handle_table.cpp
@@ -13,6 +13,7 @@ ResultCode KHandleTable::Finalize() {
     // Get the table and clear our record of it.
     u16 saved_table_size = 0;
     {
+        KScopedDisableDispatch dd(kernel);
         KScopedSpinLock lk(m_lock);
 
         std::swap(m_table_size, saved_table_size);
@@ -43,6 +44,7 @@ bool KHandleTable::Remove(Handle handle) {
     // Find the object and free the entry.
     KAutoObject* obj = nullptr;
     {
+        KScopedDisableDispatch dd(kernel);
         KScopedSpinLock lk(m_lock);
 
         if (this->IsValidHandle(handle)) {
@@ -62,6 +64,7 @@ bool KHandleTable::Remove(Handle handle) {
 }
 
 ResultCode KHandleTable::Add(Handle* out_handle, KAutoObject* obj, u16 type) {
+    KScopedDisableDispatch dd(kernel);
     KScopedSpinLock lk(m_lock);
 
     // Never exceed our capacity.
@@ -84,6 +87,7 @@ ResultCode KHandleTable::Add(Handle* out_handle, KAutoObject* obj, u16 type) {
 }
 
 ResultCode KHandleTable::Reserve(Handle* out_handle) {
+    KScopedDisableDispatch dd(kernel);
     KScopedSpinLock lk(m_lock);
 
     // Never exceed our capacity.
@@ -94,6 +98,7 @@ ResultCode KHandleTable::Reserve(Handle* out_handle) {
 }
 
 void KHandleTable::Unreserve(Handle handle) {
+    KScopedDisableDispatch dd(kernel);
     KScopedSpinLock lk(m_lock);
 
     // Unpack the handle.
@@ -112,6 +117,7 @@ void KHandleTable::Unreserve(Handle handle) {
 }
 
 void KHandleTable::Register(Handle handle, KAutoObject* obj, u16 type) {
+    KScopedDisableDispatch dd(kernel);
     KScopedSpinLock lk(m_lock);
 
     // Unpack the handle.
diff --git a/src/core/hle/kernel/k_handle_table.h b/src/core/hle/kernel/k_handle_table.h
index 95ec905ae..4b114ec2f 100644
--- a/src/core/hle/kernel/k_handle_table.h
+++ b/src/core/hle/kernel/k_handle_table.h
@@ -68,6 +68,7 @@ public:
     template <typename T = KAutoObject>
     KScopedAutoObject<T> GetObjectWithoutPseudoHandle(Handle handle) const {
         // Lock and look up in table.
+        KScopedDisableDispatch dd(kernel);
         KScopedSpinLock lk(m_lock);
 
         if constexpr (std::is_same_v<T, KAutoObject>) {
@@ -122,6 +123,7 @@ public:
         size_t num_opened;
         {
             // Lock the table.
+            KScopedDisableDispatch dd(kernel);
             KScopedSpinLock lk(m_lock);
             for (num_opened = 0; num_opened < num_handles; num_opened++) {
                 // Get the current handle.
diff --git a/src/core/hle/kernel/k_light_condition_variable.cpp b/src/core/hle/kernel/k_light_condition_variable.cpp
new file mode 100644
index 000000000..a8001fffc
--- /dev/null
+++ b/src/core/hle/kernel/k_light_condition_variable.cpp
@@ -0,0 +1,80 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/kernel/k_light_condition_variable.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
+#include "core/hle/kernel/k_thread_queue.h"
+#include "core/hle/kernel/svc_results.h"
+
+namespace Kernel {
+
+namespace {
+
+class ThreadQueueImplForKLightConditionVariable final : public KThreadQueue {
+public:
+    ThreadQueueImplForKLightConditionVariable(KernelCore& kernel_, KThread::WaiterList* wl,
+                                              bool term)
+        : KThreadQueue(kernel_), m_wait_list(wl), m_allow_terminating_thread(term) {}
+
+    void CancelWait(KThread* waiting_thread, ResultCode wait_result,
+                    bool cancel_timer_task) override {
+        // Only process waits if we're allowed to.
+        if (ResultTerminationRequested == wait_result && m_allow_terminating_thread) {
+            return;
+        }
+
+        // Remove the thread from the waiting thread from the light condition variable.
+        m_wait_list->erase(m_wait_list->iterator_to(*waiting_thread));
+
+        // Invoke the base cancel wait handler.
+        KThreadQueue::CancelWait(waiting_thread, wait_result, cancel_timer_task);
+    }
+
+private:
+    KThread::WaiterList* m_wait_list;
+    bool m_allow_terminating_thread;
+};
+
+} // namespace
+
+void KLightConditionVariable::Wait(KLightLock* lock, s64 timeout, bool allow_terminating_thread) {
+    // Create thread queue.
+    KThread* owner = GetCurrentThreadPointer(kernel);
+
+    ThreadQueueImplForKLightConditionVariable wait_queue(kernel, std::addressof(wait_list),
+                                                         allow_terminating_thread);
+
+    // Sleep the thread.
+    {
+        KScopedSchedulerLockAndSleep lk(kernel, owner, timeout);
+
+        if (!allow_terminating_thread && owner->IsTerminationRequested()) {
+            lk.CancelSleep();
+            return;
+        }
+
+        lock->Unlock();
+
+        // Add the thread to the queue.
+        wait_list.push_back(*owner);
+
+        // Begin waiting.
+        owner->BeginWait(std::addressof(wait_queue));
+    }
+
+    // Re-acquire the lock.
+    lock->Lock();
+}
+
+void KLightConditionVariable::Broadcast() {
+    KScopedSchedulerLock lk(kernel);
+
+    // Signal all threads.
+    for (auto it = wait_list.begin(); it != wait_list.end(); it = wait_list.erase(it)) {
+        it->EndWait(ResultSuccess);
+    }
+}
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_light_condition_variable.h b/src/core/hle/kernel/k_light_condition_variable.h
index fb0ad783a..5d6d7f128 100644
--- a/src/core/hle/kernel/k_light_condition_variable.h
+++ b/src/core/hle/kernel/k_light_condition_variable.h
@@ -2,72 +2,24 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-// This file references various implementation details from Atmosphere, an open-source firmware for
-// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
-
 #pragma once
 
 #include "common/common_types.h"
-#include "core/hle/kernel/k_scheduler.h"
-#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
-#include "core/hle/kernel/time_manager.h"
+#include "core/hle/kernel/k_thread.h"
 
 namespace Kernel {
+
 class KernelCore;
+class KLightLock;
 
 class KLightConditionVariable {
 public:
     explicit KLightConditionVariable(KernelCore& kernel_) : kernel{kernel_} {}
 
-    void Wait(KLightLock* lock, s64 timeout = -1, bool allow_terminating_thread = true) {
-        WaitImpl(lock, timeout, allow_terminating_thread);
-    }
-
-    void Broadcast() {
-        KScopedSchedulerLock lk{kernel};
-
-        // Signal all threads.
-        for (auto& thread : wait_list) {
-            thread.SetState(ThreadState::Runnable);
-        }
-    }
+    void Wait(KLightLock* lock, s64 timeout = -1, bool allow_terminating_thread = true);
+    void Broadcast();
 
 private:
-    void WaitImpl(KLightLock* lock, s64 timeout, bool allow_terminating_thread) {
-        KThread* owner = GetCurrentThreadPointer(kernel);
-
-        // Sleep the thread.
-        {
-            KScopedSchedulerLockAndSleep lk{kernel, owner, timeout};
-
-            if (!allow_terminating_thread && owner->IsTerminationRequested()) {
-                lk.CancelSleep();
-                return;
-            }
-
-            lock->Unlock();
-
-            // Set the thread as waiting.
-            GetCurrentThread(kernel).SetState(ThreadState::Waiting);
-
-            // Add the thread to the queue.
-            wait_list.push_back(GetCurrentThread(kernel));
-        }
-
-        // Remove the thread from the wait list.
-        {
-            KScopedSchedulerLock sl{kernel};
-
-            wait_list.erase(wait_list.iterator_to(GetCurrentThread(kernel)));
-        }
-
-        // Cancel the task that the sleep setup.
-        kernel.TimeManager().UnscheduleTimeEvent(owner);
-
-        // Re-acquire the lock.
-        lock->Lock();
-    }
-
     KernelCore& kernel;
     KThread::WaiterList wait_list{};
 };
diff --git a/src/core/hle/kernel/k_light_lock.cpp b/src/core/hle/kernel/k_light_lock.cpp
index 0896e705f..4620342eb 100644
--- a/src/core/hle/kernel/k_light_lock.cpp
+++ b/src/core/hle/kernel/k_light_lock.cpp
@@ -5,44 +5,59 @@
 #include "core/hle/kernel/k_light_lock.h"
 #include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/k_thread.h"
+#include "core/hle/kernel/k_thread_queue.h"
 #include "core/hle/kernel/kernel.h"
 
 namespace Kernel {
 
+namespace {
+
+class ThreadQueueImplForKLightLock final : public KThreadQueue {
+public:
+    explicit ThreadQueueImplForKLightLock(KernelCore& kernel_) : KThreadQueue(kernel_) {}
+
+    void CancelWait(KThread* waiting_thread, ResultCode wait_result,
+                    bool cancel_timer_task) override {
+        // Remove the thread as a waiter from its owner.
+        if (KThread* owner = waiting_thread->GetLockOwner(); owner != nullptr) {
+            owner->RemoveWaiter(waiting_thread);
+        }
+
+        // Invoke the base cancel wait handler.
+        KThreadQueue::CancelWait(waiting_thread, wait_result, cancel_timer_task);
+    }
+};
+
+} // namespace
+
 void KLightLock::Lock() {
     const uintptr_t cur_thread = reinterpret_cast<uintptr_t>(GetCurrentThreadPointer(kernel));
-    const uintptr_t cur_thread_tag = (cur_thread | 1);
 
     while (true) {
         uintptr_t old_tag = tag.load(std::memory_order_relaxed);
 
-        while (!tag.compare_exchange_weak(old_tag, (old_tag == 0) ? cur_thread : old_tag | 1,
+        while (!tag.compare_exchange_weak(old_tag, (old_tag == 0) ? cur_thread : (old_tag | 1),
                                           std::memory_order_acquire)) {
-            if ((old_tag | 1) == cur_thread_tag) {
-                return;
-            }
         }
 
-        if ((old_tag == 0) || ((old_tag | 1) == cur_thread_tag)) {
+        if (old_tag == 0 || this->LockSlowPath(old_tag | 1, cur_thread)) {
             break;
         }
-
-        LockSlowPath(old_tag | 1, cur_thread);
     }
 }
 
 void KLightLock::Unlock() {
     const uintptr_t cur_thread = reinterpret_cast<uintptr_t>(GetCurrentThreadPointer(kernel));
+
     uintptr_t expected = cur_thread;
-    do {
-        if (expected != cur_thread) {
-            return UnlockSlowPath(cur_thread);
-        }
-    } while (!tag.compare_exchange_weak(expected, 0, std::memory_order_release));
+    if (!tag.compare_exchange_strong(expected, 0, std::memory_order_release)) {
+        this->UnlockSlowPath(cur_thread);
+    }
 }
 
-void KLightLock::LockSlowPath(uintptr_t _owner, uintptr_t _cur_thread) {
+bool KLightLock::LockSlowPath(uintptr_t _owner, uintptr_t _cur_thread) {
     KThread* cur_thread = reinterpret_cast<KThread*>(_cur_thread);
+    ThreadQueueImplForKLightLock wait_queue(kernel);
 
     // Pend the current thread waiting on the owner thread.
     {
@@ -50,7 +65,7 @@ void KLightLock::LockSlowPath(uintptr_t _owner, uintptr_t _cur_thread) {
 
         // Ensure we actually have locking to do.
         if (tag.load(std::memory_order_relaxed) != _owner) {
-            return;
+            return false;
         }
 
         // Add the current thread as a waiter on the owner.
@@ -58,22 +73,15 @@ void KLightLock::LockSlowPath(uintptr_t _owner, uintptr_t _cur_thread) {
         cur_thread->SetAddressKey(reinterpret_cast<uintptr_t>(std::addressof(tag)));
         owner_thread->AddWaiter(cur_thread);
 
-        // Set thread states.
-        cur_thread->SetState(ThreadState::Waiting);
+        // Begin waiting to hold the lock.
+        cur_thread->BeginWait(std::addressof(wait_queue));
 
         if (owner_thread->IsSuspended()) {
             owner_thread->ContinueIfHasKernelWaiters();
         }
     }
 
-    // We're no longer waiting on the lock owner.
-    {
-        KScopedSchedulerLock sl{kernel};
-
-        if (KThread* owner_thread = cur_thread->GetLockOwner(); owner_thread != nullptr) {
-            owner_thread->RemoveWaiter(cur_thread);
-        }
-    }
+    return true;
 }
 
 void KLightLock::UnlockSlowPath(uintptr_t _cur_thread) {
@@ -81,22 +89,20 @@ void KLightLock::UnlockSlowPath(uintptr_t _cur_thread) {
 
     // Unlock.
     {
-        KScopedSchedulerLock sl{kernel};
+        KScopedSchedulerLock sl(kernel);
 
         // Get the next owner.
-        s32 num_waiters = 0;
+        s32 num_waiters;
         KThread* next_owner = owner_thread->RemoveWaiterByKey(
             std::addressof(num_waiters), reinterpret_cast<uintptr_t>(std::addressof(tag)));
 
         // Pass the lock to the next owner.
         uintptr_t next_tag = 0;
         if (next_owner != nullptr) {
-            next_tag = reinterpret_cast<uintptr_t>(next_owner);
-            if (num_waiters > 1) {
-                next_tag |= 0x1;
-            }
+            next_tag =
+                reinterpret_cast<uintptr_t>(next_owner) | static_cast<uintptr_t>(num_waiters > 1);
 
-            next_owner->SetState(ThreadState::Runnable);
+            next_owner->EndWait(ResultSuccess);
 
             if (next_owner->IsSuspended()) {
                 next_owner->ContinueIfHasKernelWaiters();
@@ -110,7 +116,7 @@ void KLightLock::UnlockSlowPath(uintptr_t _cur_thread) {
         }
 
         // Write the new tag value.
-        tag.store(next_tag);
+        tag.store(next_tag, std::memory_order_release);
     }
 }
 
diff --git a/src/core/hle/kernel/k_light_lock.h b/src/core/hle/kernel/k_light_lock.h
index ad853661d..4163b8a85 100644
--- a/src/core/hle/kernel/k_light_lock.h
+++ b/src/core/hle/kernel/k_light_lock.h
@@ -20,7 +20,7 @@ public:
 
     void Unlock();
 
-    void LockSlowPath(uintptr_t owner, uintptr_t cur_thread);
+    bool LockSlowPath(uintptr_t owner, uintptr_t cur_thread);
 
     void UnlockSlowPath(uintptr_t cur_thread);
 
diff --git a/src/core/hle/kernel/k_memory_block.h b/src/core/hle/kernel/k_memory_block.h
index a7fdb5fb8..fd491146f 100644
--- a/src/core/hle/kernel/k_memory_block.h
+++ b/src/core/hle/kernel/k_memory_block.h
@@ -131,6 +131,26 @@ enum class KMemoryPermission : u8 {
 
     UserMask = static_cast<u8>(Svc::MemoryPermission::Read | Svc::MemoryPermission::Write |
                                Svc::MemoryPermission::Execute),
+
+    KernelShift = 3,
+
+    KernelRead = Read << KernelShift,
+    KernelWrite = Write << KernelShift,
+    KernelExecute = Execute << KernelShift,
+
+    NotMapped = (1 << (2 * KernelShift)),
+
+    KernelReadWrite = KernelRead | KernelWrite,
+    KernelReadExecute = KernelRead | KernelExecute,
+
+    UserRead = Read | KernelRead,
+    UserWrite = Write | KernelWrite,
+    UserExecute = Execute,
+
+    UserReadWrite = UserRead | UserWrite,
+    UserReadExecute = UserRead | UserExecute,
+
+    IpcLockChangeMask = NotMapped | UserReadWrite
 };
 DECLARE_ENUM_FLAG_OPERATORS(KMemoryPermission);
 
diff --git a/src/core/hle/kernel/k_page_linked_list.h b/src/core/hle/kernel/k_page_linked_list.h
index 3362fb236..0e2ae582a 100644
--- a/src/core/hle/kernel/k_page_linked_list.h
+++ b/src/core/hle/kernel/k_page_linked_list.h
@@ -27,6 +27,10 @@ public:
             return num_pages;
         }
 
+        constexpr std::size_t GetSize() const {
+            return GetNumPages() * PageSize;
+        }
+
     private:
         u64 addr{};
         std::size_t num_pages{};
diff --git a/src/core/hle/kernel/k_page_table.cpp b/src/core/hle/kernel/k_page_table.cpp
index 9bda5c5b2..99982e5a3 100644
--- a/src/core/hle/kernel/k_page_table.cpp
+++ b/src/core/hle/kernel/k_page_table.cpp
@@ -368,6 +368,33 @@ ResultCode KPageTable::UnmapProcessCodeMemory(VAddr dst_addr, VAddr src_addr, st
     return ResultSuccess;
 }
 
+ResultCode KPageTable::UnmapProcessMemory(VAddr dst_addr, std::size_t size,
+                                          KPageTable& src_page_table, VAddr src_addr) {
+    std::lock_guard lock{page_table_lock};
+
+    const std::size_t num_pages{size / PageSize};
+
+    // Check that the memory is mapped in the destination process.
+    size_t num_allocator_blocks;
+    R_TRY(CheckMemoryState(&num_allocator_blocks, dst_addr, size, KMemoryState::All,
+                           KMemoryState::SharedCode, KMemoryPermission::UserReadWrite,
+                           KMemoryPermission::UserReadWrite, KMemoryAttribute::All,
+                           KMemoryAttribute::None));
+
+    // Check that the memory is mapped in the source process.
+    R_TRY(src_page_table.CheckMemoryState(src_addr, size, KMemoryState::FlagCanMapProcess,
+                                          KMemoryState::FlagCanMapProcess, KMemoryPermission::None,
+                                          KMemoryPermission::None, KMemoryAttribute::All,
+                                          KMemoryAttribute::None));
+
+    CASCADE_CODE(Operate(dst_addr, num_pages, KMemoryPermission::None, OperationType::Unmap));
+
+    // Apply the memory block update.
+    block_manager->Update(dst_addr, num_pages, KMemoryState::Free, KMemoryPermission::None,
+                          KMemoryAttribute::None);
+
+    return ResultSuccess;
+}
 void KPageTable::MapPhysicalMemory(KPageLinkedList& page_linked_list, VAddr start, VAddr end) {
     auto node{page_linked_list.Nodes().begin()};
     PAddr map_addr{node->GetAddress()};
@@ -942,6 +969,60 @@ ResultCode KPageTable::UnlockForDeviceAddressSpace(VAddr addr, std::size_t size)
     return ResultSuccess;
 }
 
+ResultCode KPageTable::LockForCodeMemory(VAddr addr, std::size_t size) {
+    std::lock_guard lock{page_table_lock};
+
+    KMemoryPermission new_perm = KMemoryPermission::NotMapped | KMemoryPermission::KernelReadWrite;
+
+    KMemoryPermission old_perm{};
+
+    if (const ResultCode result{CheckMemoryState(
+            nullptr, &old_perm, nullptr, addr, size, KMemoryState::FlagCanCodeMemory,
+            KMemoryState::FlagCanCodeMemory, KMemoryPermission::Mask,
+            KMemoryPermission::UserReadWrite, KMemoryAttribute::All, KMemoryAttribute::None)};
+        result.IsError()) {
+        return result;
+    }
+
+    new_perm = (new_perm != KMemoryPermission::None) ? new_perm : old_perm;
+
+    block_manager->UpdateLock(
+        addr, size / PageSize,
+        [](KMemoryBlockManager::iterator block, KMemoryPermission permission) {
+            block->ShareToDevice(permission);
+        },
+        new_perm);
+
+    return ResultSuccess;
+}
+
+ResultCode KPageTable::UnlockForCodeMemory(VAddr addr, std::size_t size) {
+    std::lock_guard lock{page_table_lock};
+
+    KMemoryPermission new_perm = KMemoryPermission::UserReadWrite;
+
+    KMemoryPermission old_perm{};
+
+    if (const ResultCode result{CheckMemoryState(
+            nullptr, &old_perm, nullptr, addr, size, KMemoryState::FlagCanCodeMemory,
+            KMemoryState::FlagCanCodeMemory, KMemoryPermission::None, KMemoryPermission::None,
+            KMemoryAttribute::All, KMemoryAttribute::Locked)};
+        result.IsError()) {
+        return result;
+    }
+
+    new_perm = (new_perm != KMemoryPermission::None) ? new_perm : old_perm;
+
+    block_manager->UpdateLock(
+        addr, size / PageSize,
+        [](KMemoryBlockManager::iterator block, KMemoryPermission permission) {
+            block->UnshareToDevice(permission);
+        },
+        new_perm);
+
+    return ResultSuccess;
+}
+
 ResultCode KPageTable::InitializeMemoryLayout(VAddr start, VAddr end) {
     block_manager = std::make_unique<KMemoryBlockManager>(start, end);
 
@@ -1231,4 +1312,42 @@ ResultCode KPageTable::CheckMemoryState(KMemoryState* out_state, KMemoryPermissi
     return ResultSuccess;
 }
 
+ResultCode KPageTable::CheckMemoryState(size_t* out_blocks_needed, VAddr addr, size_t size,
+                                        KMemoryState state_mask, KMemoryState state,
+                                        KMemoryPermission perm_mask, KMemoryPermission perm,
+                                        KMemoryAttribute attr_mask, KMemoryAttribute attr) const {
+    // Get information about the first block.
+    const VAddr last_addr = addr + size - 1;
+    KMemoryBlockManager::const_iterator it{block_manager->FindIterator(addr)};
+    KMemoryInfo info = it->GetMemoryInfo();
+
+    // If the start address isn't aligned, we need a block.
+    const size_t blocks_for_start_align =
+        (Common::AlignDown(addr, PageSize) != info.GetAddress()) ? 1 : 0;
+
+    while (true) {
+        // Validate against the provided masks.
+        R_TRY(CheckMemoryState(info, state_mask, state, perm_mask, perm, attr_mask, attr));
+
+        // Break once we're done.
+        if (last_addr <= info.GetLastAddress()) {
+            break;
+        }
+
+        // Advance our iterator.
+        it++;
+        info = it->GetMemoryInfo();
+    }
+
+    // If the end address isn't aligned, we need a block.
+    const size_t blocks_for_end_align =
+        (Common::AlignUp(addr + size, PageSize) != info.GetEndAddress()) ? 1 : 0;
+
+    if (out_blocks_needed != nullptr) {
+        *out_blocks_needed = blocks_for_start_align + blocks_for_end_align;
+    }
+
+    return ResultSuccess;
+}
+
 } // namespace Kernel
diff --git a/src/core/hle/kernel/k_page_table.h b/src/core/hle/kernel/k_page_table.h
index b7ec38f06..d784aa67e 100644
--- a/src/core/hle/kernel/k_page_table.h
+++ b/src/core/hle/kernel/k_page_table.h
@@ -33,6 +33,8 @@ public:
                               KMemoryPermission perm);
     ResultCode MapProcessCodeMemory(VAddr dst_addr, VAddr src_addr, std::size_t size);
     ResultCode UnmapProcessCodeMemory(VAddr dst_addr, VAddr src_addr, std::size_t size);
+    ResultCode UnmapProcessMemory(VAddr dst_addr, std::size_t size, KPageTable& src_page_table,
+                                  VAddr src_addr);
     ResultCode MapPhysicalMemory(VAddr addr, std::size_t size);
     ResultCode UnmapPhysicalMemory(VAddr addr, std::size_t size);
     ResultCode UnmapMemory(VAddr addr, std::size_t size);
@@ -55,6 +57,8 @@ public:
                                           KMemoryPermission perm, PAddr map_addr = 0);
     ResultCode LockForDeviceAddressSpace(VAddr addr, std::size_t size);
     ResultCode UnlockForDeviceAddressSpace(VAddr addr, std::size_t size);
+    ResultCode LockForCodeMemory(VAddr addr, std::size_t size);
+    ResultCode UnlockForCodeMemory(VAddr addr, std::size_t size);
 
     Common::PageTable& PageTableImpl() {
         return page_table_impl;
@@ -115,6 +119,10 @@ private:
         return CheckMemoryState(nullptr, nullptr, nullptr, addr, size, state_mask, state, perm_mask,
                                 perm, attr_mask, attr, ignore_attr);
     }
+    ResultCode CheckMemoryState(size_t* out_blocks_needed, VAddr addr, size_t size,
+                                KMemoryState state_mask, KMemoryState state,
+                                KMemoryPermission perm_mask, KMemoryPermission perm,
+                                KMemoryAttribute attr_mask, KMemoryAttribute attr) const;
 
     std::recursive_mutex page_table_lock;
     std::unique_ptr<KMemoryBlockManager> block_manager;
diff --git a/src/core/hle/kernel/k_process.cpp b/src/core/hle/kernel/k_process.cpp
index 1aad061e1..90dda40dc 100644
--- a/src/core/hle/kernel/k_process.cpp
+++ b/src/core/hle/kernel/k_process.cpp
@@ -60,6 +60,7 @@ void SetupMainThread(Core::System& system, KProcess& owner_process, u32 priority
     thread->GetContext64().cpu_registers[0] = 0;
     thread->GetContext32().cpu_registers[1] = thread_handle;
     thread->GetContext64().cpu_registers[1] = thread_handle;
+    thread->DisableDispatch();
 
     auto& kernel = system.Kernel();
     // Threads by default are dormant, wake up the main thread so it runs when the scheduler fires
@@ -227,12 +228,15 @@ void KProcess::PinCurrentThread() {
     const s32 core_id = GetCurrentCoreId(kernel);
     KThread* cur_thread = GetCurrentThreadPointer(kernel);
 
-    // Pin it.
-    PinThread(core_id, cur_thread);
-    cur_thread->Pin();
+    // If the thread isn't terminated, pin it.
+    if (!cur_thread->IsTerminationRequested()) {
+        // Pin it.
+        PinThread(core_id, cur_thread);
+        cur_thread->Pin();
 
-    // An update is needed.
-    KScheduler::SetSchedulerUpdateNeeded(kernel);
+        // An update is needed.
+        KScheduler::SetSchedulerUpdateNeeded(kernel);
+    }
 }
 
 void KProcess::UnpinCurrentThread() {
@@ -250,6 +254,20 @@ void KProcess::UnpinCurrentThread() {
     KScheduler::SetSchedulerUpdateNeeded(kernel);
 }
 
+void KProcess::UnpinThread(KThread* thread) {
+    ASSERT(kernel.GlobalSchedulerContext().IsLocked());
+
+    // Get the thread's core id.
+    const auto core_id = thread->GetActiveCore();
+
+    // Unpin it.
+    UnpinThread(core_id, thread);
+    thread->Unpin();
+
+    // An update is needed.
+    KScheduler::SetSchedulerUpdateNeeded(kernel);
+}
+
 ResultCode KProcess::AddSharedMemory(KSharedMemory* shmem, [[maybe_unused]] VAddr address,
                                      [[maybe_unused]] size_t size) {
     // Lock ourselves, to prevent concurrent access.
diff --git a/src/core/hle/kernel/k_process.h b/src/core/hle/kernel/k_process.h
index 8a8c1fcbb..cb93c7e24 100644
--- a/src/core/hle/kernel/k_process.h
+++ b/src/core/hle/kernel/k_process.h
@@ -347,6 +347,7 @@ public:
 
     void PinCurrentThread();
     void UnpinCurrentThread();
+    void UnpinThread(KThread* thread);
 
     KLightLock& GetStateLock() {
         return state_lock;
diff --git a/src/core/hle/kernel/k_scheduler.cpp b/src/core/hle/kernel/k_scheduler.cpp
index 6a7d80d03..277201de4 100644
--- a/src/core/hle/kernel/k_scheduler.cpp
+++ b/src/core/hle/kernel/k_scheduler.cpp
@@ -240,8 +240,8 @@ void KScheduler::OnThreadPriorityChanged(KernelCore& kernel, KThread* thread, s3
 
     // If the thread is runnable, we want to change its priority in the queue.
     if (thread->GetRawState() == ThreadState::Runnable) {
-        GetPriorityQueue(kernel).ChangePriority(
-            old_priority, thread == kernel.CurrentScheduler()->GetCurrentThread(), thread);
+        GetPriorityQueue(kernel).ChangePriority(old_priority,
+                                                thread == kernel.GetCurrentEmuThread(), thread);
         IncrementScheduledCount(thread);
         SetSchedulerUpdateNeeded(kernel);
     }
@@ -360,7 +360,7 @@ void KScheduler::RotateScheduledQueue(s32 cpu_core_id, s32 priority) {
 }
 
 bool KScheduler::CanSchedule(KernelCore& kernel) {
-    return kernel.CurrentScheduler()->GetCurrentThread()->GetDisableDispatchCount() <= 1;
+    return kernel.GetCurrentEmuThread()->GetDisableDispatchCount() <= 1;
 }
 
 bool KScheduler::IsSchedulerUpdateNeeded(const KernelCore& kernel) {
@@ -376,20 +376,30 @@ void KScheduler::ClearSchedulerUpdateNeeded(KernelCore& kernel) {
 }
 
 void KScheduler::DisableScheduling(KernelCore& kernel) {
-    if (auto* scheduler = kernel.CurrentScheduler(); scheduler) {
-        ASSERT(scheduler->GetCurrentThread()->GetDisableDispatchCount() >= 0);
-        scheduler->GetCurrentThread()->DisableDispatch();
+    // If we are shutting down the kernel, none of this is relevant anymore.
+    if (kernel.IsShuttingDown()) {
+        return;
     }
+
+    ASSERT(GetCurrentThreadPointer(kernel)->GetDisableDispatchCount() >= 0);
+    GetCurrentThreadPointer(kernel)->DisableDispatch();
 }
 
 void KScheduler::EnableScheduling(KernelCore& kernel, u64 cores_needing_scheduling) {
-    if (auto* scheduler = kernel.CurrentScheduler(); scheduler) {
-        ASSERT(scheduler->GetCurrentThread()->GetDisableDispatchCount() >= 1);
-        if (scheduler->GetCurrentThread()->GetDisableDispatchCount() >= 1) {
-            scheduler->GetCurrentThread()->EnableDispatch();
-        }
+    // If we are shutting down the kernel, none of this is relevant anymore.
+    if (kernel.IsShuttingDown()) {
+        return;
+    }
+
+    auto* current_thread = GetCurrentThreadPointer(kernel);
+
+    ASSERT(current_thread->GetDisableDispatchCount() >= 1);
+
+    if (current_thread->GetDisableDispatchCount() > 1) {
+        current_thread->EnableDispatch();
+    } else {
+        RescheduleCores(kernel, cores_needing_scheduling);
     }
-    RescheduleCores(kernel, cores_needing_scheduling);
 }
 
 u64 KScheduler::UpdateHighestPriorityThreads(KernelCore& kernel) {
@@ -617,13 +627,17 @@ KScheduler::KScheduler(Core::System& system_, s32 core_id_) : system{system_}, c
     state.highest_priority_thread = nullptr;
 }
 
-KScheduler::~KScheduler() {
+void KScheduler::Finalize() {
     if (idle_thread) {
         idle_thread->Close();
         idle_thread = nullptr;
     }
 }
 
+KScheduler::~KScheduler() {
+    ASSERT(!idle_thread);
+}
+
 KThread* KScheduler::GetCurrentThread() const {
     if (auto result = current_thread.load(); result) {
         return result;
@@ -642,10 +656,12 @@ void KScheduler::RescheduleCurrentCore() {
     if (phys_core.IsInterrupted()) {
         phys_core.ClearInterrupt();
     }
+
     guard.Lock();
     if (state.needs_scheduling.load()) {
         Schedule();
     } else {
+        GetCurrentThread()->EnableDispatch();
         guard.Unlock();
     }
 }
@@ -655,26 +671,33 @@ void KScheduler::OnThreadStart() {
 }
 
 void KScheduler::Unload(KThread* thread) {
+    ASSERT(thread);
+
     LOG_TRACE(Kernel, "core {}, unload thread {}", core_id, thread ? thread->GetName() : "nullptr");
 
-    if (thread) {
-        if (thread->IsCallingSvc()) {
-            thread->ClearIsCallingSvc();
-        }
-        if (!thread->IsTerminationRequested()) {
-            prev_thread = thread;
-
-            Core::ARM_Interface& cpu_core = system.ArmInterface(core_id);
-            cpu_core.SaveContext(thread->GetContext32());
-            cpu_core.SaveContext(thread->GetContext64());
-            // Save the TPIDR_EL0 system register in case it was modified.
-            thread->SetTPIDR_EL0(cpu_core.GetTPIDR_EL0());
-            cpu_core.ClearExclusiveState();
-        } else {
-            prev_thread = nullptr;
-        }
-        thread->context_guard.Unlock();
+    if (thread->IsCallingSvc()) {
+        thread->ClearIsCallingSvc();
+    }
+
+    auto& physical_core = system.Kernel().PhysicalCore(core_id);
+    if (!physical_core.IsInitialized()) {
+        return;
+    }
+
+    Core::ARM_Interface& cpu_core = physical_core.ArmInterface();
+    cpu_core.SaveContext(thread->GetContext32());
+    cpu_core.SaveContext(thread->GetContext64());
+    // Save the TPIDR_EL0 system register in case it was modified.
+    thread->SetTPIDR_EL0(cpu_core.GetTPIDR_EL0());
+    cpu_core.ClearExclusiveState();
+
+    if (!thread->IsTerminationRequested() && thread->GetActiveCore() == core_id) {
+        prev_thread = thread;
+    } else {
+        prev_thread = nullptr;
     }
+
+    thread->context_guard.Unlock();
 }
 
 void KScheduler::Reload(KThread* thread) {
@@ -683,11 +706,6 @@ void KScheduler::Reload(KThread* thread) {
     if (thread) {
         ASSERT_MSG(thread->GetState() == ThreadState::Runnable, "Thread must be runnable.");
 
-        auto* const thread_owner_process = thread->GetOwnerProcess();
-        if (thread_owner_process != nullptr) {
-            system.Kernel().MakeCurrentProcess(thread_owner_process);
-        }
-
         Core::ARM_Interface& cpu_core = system.ArmInterface(core_id);
         cpu_core.LoadContext(thread->GetContext32());
         cpu_core.LoadContext(thread->GetContext64());
@@ -705,7 +723,7 @@ void KScheduler::SwitchContextStep2() {
 }
 
 void KScheduler::ScheduleImpl() {
-    KThread* previous_thread = current_thread.load();
+    KThread* previous_thread = GetCurrentThread();
     KThread* next_thread = state.highest_priority_thread;
 
     state.needs_scheduling = false;
@@ -717,10 +735,15 @@ void KScheduler::ScheduleImpl() {
 
     // If we're not actually switching thread, there's nothing to do.
     if (next_thread == current_thread.load()) {
+        previous_thread->EnableDispatch();
         guard.Unlock();
         return;
     }
 
+    if (next_thread->GetCurrentCore() != core_id) {
+        next_thread->SetCurrentCore(core_id);
+    }
+
     current_thread.store(next_thread);
 
     KProcess* const previous_process = system.Kernel().CurrentProcess();
@@ -731,11 +754,7 @@ void KScheduler::ScheduleImpl() {
     Unload(previous_thread);
 
     std::shared_ptr<Common::Fiber>* old_context;
-    if (previous_thread != nullptr) {
-        old_context = &previous_thread->GetHostContext();
-    } else {
-        old_context = &idle_thread->GetHostContext();
-    }
+    old_context = &previous_thread->GetHostContext();
     guard.Unlock();
 
     Common::Fiber::YieldTo(*old_context, *switch_fiber);
diff --git a/src/core/hle/kernel/k_scheduler.h b/src/core/hle/kernel/k_scheduler.h
index 7df288438..82fcd99e7 100644
--- a/src/core/hle/kernel/k_scheduler.h
+++ b/src/core/hle/kernel/k_scheduler.h
@@ -33,6 +33,8 @@ public:
     explicit KScheduler(Core::System& system_, s32 core_id_);
     ~KScheduler();
 
+    void Finalize();
+
     /// Reschedules to the next available thread (call after current thread is suspended)
     void RescheduleCurrentCore();
 
diff --git a/src/core/hle/kernel/k_scheduler_lock.h b/src/core/hle/kernel/k_scheduler_lock.h
index c571f2992..93c47f1b1 100644
--- a/src/core/hle/kernel/k_scheduler_lock.h
+++ b/src/core/hle/kernel/k_scheduler_lock.h
@@ -23,6 +23,11 @@ public:
     }
 
     void Lock() {
+        // If we are shutting down the kernel, none of this is relevant anymore.
+        if (kernel.IsShuttingDown()) {
+            return;
+        }
+
         if (IsLockedByCurrentThread()) {
             // If we already own the lock, we can just increment the count.
             ASSERT(lock_count > 0);
@@ -43,6 +48,11 @@ public:
     }
 
     void Unlock() {
+        // If we are shutting down the kernel, none of this is relevant anymore.
+        if (kernel.IsShuttingDown()) {
+            return;
+        }
+
         ASSERT(IsLockedByCurrentThread());
         ASSERT(lock_count > 0);
 
diff --git a/src/core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h b/src/core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h
index 61dc2858f..2995c492d 100644
--- a/src/core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h
+++ b/src/core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h
@@ -8,6 +8,7 @@
 #pragma once
 
 #include "common/common_types.h"
+#include "core/hle/kernel/global_scheduler_context.h"
 #include "core/hle/kernel/k_thread.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/time_manager.h"
diff --git a/src/core/hle/kernel/k_server_session.cpp b/src/core/hle/kernel/k_server_session.cpp
index 2bd53ccbd..d4e4a6b06 100644
--- a/src/core/hle/kernel/k_server_session.cpp
+++ b/src/core/hle/kernel/k_server_session.cpp
@@ -175,8 +175,7 @@ ResultCode KServerSession::CompleteSyncRequest(HLERequestContext& context) {
     {
         KScopedSchedulerLock lock(kernel);
         if (!context.IsThreadWaiting()) {
-            context.GetThread().Wakeup();
-            context.GetThread().SetSyncedObject(nullptr, result);
+            context.GetThread().EndWait(result);
         }
     }
 
diff --git a/src/core/hle/kernel/k_synchronization_object.cpp b/src/core/hle/kernel/k_synchronization_object.cpp
index f168b4f21..e4c5eb74f 100644
--- a/src/core/hle/kernel/k_synchronization_object.cpp
+++ b/src/core/hle/kernel/k_synchronization_object.cpp
@@ -8,11 +8,66 @@
 #include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
 #include "core/hle/kernel/k_synchronization_object.h"
 #include "core/hle/kernel/k_thread.h"
+#include "core/hle/kernel/k_thread_queue.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/svc_results.h"
 
 namespace Kernel {
 
+namespace {
+
+class ThreadQueueImplForKSynchronizationObjectWait final : public KThreadQueueWithoutEndWait {
+public:
+    ThreadQueueImplForKSynchronizationObjectWait(KernelCore& kernel_, KSynchronizationObject** o,
+                                                 KSynchronizationObject::ThreadListNode* n, s32 c)
+        : KThreadQueueWithoutEndWait(kernel_), m_objects(o), m_nodes(n), m_count(c) {}
+
+    void NotifyAvailable(KThread* waiting_thread, KSynchronizationObject* signaled_object,
+                         ResultCode wait_result) override {
+        // Determine the sync index, and unlink all nodes.
+        s32 sync_index = -1;
+        for (auto i = 0; i < m_count; ++i) {
+            // Check if this is the signaled object.
+            if (m_objects[i] == signaled_object && sync_index == -1) {
+                sync_index = i;
+            }
+
+            // Unlink the current node from the current object.
+            m_objects[i]->UnlinkNode(std::addressof(m_nodes[i]));
+        }
+
+        // Set the waiting thread's sync index.
+        waiting_thread->SetSyncedIndex(sync_index);
+
+        // Set the waiting thread as not cancellable.
+        waiting_thread->ClearCancellable();
+
+        // Invoke the base end wait handler.
+        KThreadQueue::EndWait(waiting_thread, wait_result);
+    }
+
+    void CancelWait(KThread* waiting_thread, ResultCode wait_result,
+                    bool cancel_timer_task) override {
+        // Remove all nodes from our list.
+        for (auto i = 0; i < m_count; ++i) {
+            m_objects[i]->UnlinkNode(std::addressof(m_nodes[i]));
+        }
+
+        // Set the waiting thread as not cancellable.
+        waiting_thread->ClearCancellable();
+
+        // Invoke the base cancel wait handler.
+        KThreadQueue::CancelWait(waiting_thread, wait_result, cancel_timer_task);
+    }
+
+private:
+    KSynchronizationObject** m_objects;
+    KSynchronizationObject::ThreadListNode* m_nodes;
+    s32 m_count;
+};
+
+} // namespace
+
 void KSynchronizationObject::Finalize() {
     this->OnFinalizeSynchronizationObject();
     KAutoObject::Finalize();
@@ -25,11 +80,19 @@ ResultCode KSynchronizationObject::Wait(KernelCore& kernel_ctx, s32* out_index,
     std::vector<ThreadListNode> thread_nodes(num_objects);
 
     // Prepare for wait.
-    KThread* thread = kernel_ctx.CurrentScheduler()->GetCurrentThread();
+    KThread* thread = GetCurrentThreadPointer(kernel_ctx);
+    ThreadQueueImplForKSynchronizationObjectWait wait_queue(kernel_ctx, objects,
+                                                            thread_nodes.data(), num_objects);
 
     {
         // Setup the scheduling lock and sleep.
-        KScopedSchedulerLockAndSleep slp{kernel_ctx, thread, timeout};
+        KScopedSchedulerLockAndSleep slp(kernel_ctx, thread, timeout);
+
+        // Check if the thread should terminate.
+        if (thread->IsTerminationRequested()) {
+            slp.CancelSleep();
+            return ResultTerminationRequested;
+        }
 
         // Check if any of the objects are already signaled.
         for (auto i = 0; i < num_objects; ++i) {
@@ -48,12 +111,6 @@ ResultCode KSynchronizationObject::Wait(KernelCore& kernel_ctx, s32* out_index,
             return ResultTimedOut;
         }
 
-        // Check if the thread should terminate.
-        if (thread->IsTerminationRequested()) {
-            slp.CancelSleep();
-            return ResultTerminationRequested;
-        }
-
         // Check if waiting was canceled.
         if (thread->IsWaitCancelled()) {
             slp.CancelSleep();
@@ -66,73 +123,25 @@ ResultCode KSynchronizationObject::Wait(KernelCore& kernel_ctx, s32* out_index,
             thread_nodes[i].thread = thread;
             thread_nodes[i].next = nullptr;
 
-            if (objects[i]->thread_list_tail == nullptr) {
-                objects[i]->thread_list_head = std::addressof(thread_nodes[i]);
-            } else {
-                objects[i]->thread_list_tail->next = std::addressof(thread_nodes[i]);
-            }
-
-            objects[i]->thread_list_tail = std::addressof(thread_nodes[i]);
+            objects[i]->LinkNode(std::addressof(thread_nodes[i]));
         }
 
-        // For debugging only
-        thread->SetWaitObjectsForDebugging({objects, static_cast<std::size_t>(num_objects)});
-
-        // Mark the thread as waiting.
+        // Mark the thread as cancellable.
         thread->SetCancellable();
-        thread->SetSyncedObject(nullptr, ResultTimedOut);
-        thread->SetState(ThreadState::Waiting);
-        thread->SetWaitReasonForDebugging(ThreadWaitReasonForDebugging::Synchronization);
-    }
 
-    // The lock/sleep is done, so we should be able to get our result.
+        // Clear the thread's synced index.
+        thread->SetSyncedIndex(-1);
 
-    // Thread is no longer cancellable.
-    thread->ClearCancellable();
-
-    // For debugging only
-    thread->SetWaitObjectsForDebugging({});
+        // Wait for an object to be signaled.
+        thread->BeginWait(std::addressof(wait_queue));
+        thread->SetWaitReasonForDebugging(ThreadWaitReasonForDebugging::Synchronization);
+    }
 
-    // Cancel the timer as needed.
-    kernel_ctx.TimeManager().UnscheduleTimeEvent(thread);
+    // Set the output index.
+    *out_index = thread->GetSyncedIndex();
 
     // Get the wait result.
-    ResultCode wait_result{ResultSuccess};
-    s32 sync_index = -1;
-    {
-        KScopedSchedulerLock lock(kernel_ctx);
-        KSynchronizationObject* synced_obj;
-        wait_result = thread->GetWaitResult(std::addressof(synced_obj));
-
-        for (auto i = 0; i < num_objects; ++i) {
-            // Unlink the object from the list.
-            ThreadListNode* prev_ptr =
-                reinterpret_cast<ThreadListNode*>(std::addressof(objects[i]->thread_list_head));
-            ThreadListNode* prev_val = nullptr;
-            ThreadListNode *prev, *tail_prev;
-
-            do {
-                prev = prev_ptr;
-                prev_ptr = prev_ptr->next;
-                tail_prev = prev_val;
-                prev_val = prev_ptr;
-            } while (prev_ptr != std::addressof(thread_nodes[i]));
-
-            if (objects[i]->thread_list_tail == std::addressof(thread_nodes[i])) {
-                objects[i]->thread_list_tail = tail_prev;
-            }
-
-            prev->next = thread_nodes[i].next;
-
-            if (objects[i] == synced_obj) {
-                sync_index = i;
-            }
-        }
-    }
-
-    // Set output.
-    *out_index = sync_index;
-    return wait_result;
+    return thread->GetWaitResult();
 }
 
 KSynchronizationObject::KSynchronizationObject(KernelCore& kernel_)
@@ -141,7 +150,7 @@ KSynchronizationObject::KSynchronizationObject(KernelCore& kernel_)
 KSynchronizationObject::~KSynchronizationObject() = default;
 
 void KSynchronizationObject::NotifyAvailable(ResultCode result) {
-    KScopedSchedulerLock lock(kernel);
+    KScopedSchedulerLock sl(kernel);
 
     // If we're not signaled, we've nothing to notify.
     if (!this->IsSignaled()) {
@@ -150,11 +159,7 @@ void KSynchronizationObject::NotifyAvailable(ResultCode result) {
 
     // Iterate over each thread.
     for (auto* cur_node = thread_list_head; cur_node != nullptr; cur_node = cur_node->next) {
-        KThread* thread = cur_node->thread;
-        if (thread->GetState() == ThreadState::Waiting) {
-            thread->SetSyncedObject(this, result);
-            thread->SetState(ThreadState::Runnable);
-        }
+        cur_node->thread->NotifyAvailable(this, result);
     }
 }
 
diff --git a/src/core/hle/kernel/k_synchronization_object.h b/src/core/hle/kernel/k_synchronization_object.h
index 898e58e16..ec235437b 100644
--- a/src/core/hle/kernel/k_synchronization_object.h
+++ b/src/core/hle/kernel/k_synchronization_object.h
@@ -35,6 +35,38 @@ public:
 
     [[nodiscard]] std::vector<KThread*> GetWaitingThreadsForDebugging() const;
 
+    void LinkNode(ThreadListNode* node_) {
+        // Link the node to the list.
+        if (thread_list_tail == nullptr) {
+            thread_list_head = node_;
+        } else {
+            thread_list_tail->next = node_;
+        }
+
+        thread_list_tail = node_;
+    }
+
+    void UnlinkNode(ThreadListNode* node_) {
+        // Unlink the node from the list.
+        ThreadListNode* prev_ptr =
+            reinterpret_cast<ThreadListNode*>(std::addressof(thread_list_head));
+        ThreadListNode* prev_val = nullptr;
+        ThreadListNode *prev, *tail_prev;
+
+        do {
+            prev = prev_ptr;
+            prev_ptr = prev_ptr->next;
+            tail_prev = prev_val;
+            prev_val = prev_ptr;
+        } while (prev_ptr != node_);
+
+        if (thread_list_tail == node_) {
+            thread_list_tail = tail_prev;
+        }
+
+        prev->next = node_->next;
+    }
+
 protected:
     explicit KSynchronizationObject(KernelCore& kernel);
     ~KSynchronizationObject() override;
diff --git a/src/core/hle/kernel/k_thread.cpp b/src/core/hle/kernel/k_thread.cpp
index db65ce79a..752592e2e 100644
--- a/src/core/hle/kernel/k_thread.cpp
+++ b/src/core/hle/kernel/k_thread.cpp
@@ -13,6 +13,9 @@
 #include "common/common_types.h"
 #include "common/fiber.h"
 #include "common/logging/log.h"
+#include "common/scope_exit.h"
+#include "common/settings.h"
+#include "common/thread_queue_list.h"
 #include "core/core.h"
 #include "core/cpu_manager.h"
 #include "core/hardware_properties.h"
@@ -56,6 +59,34 @@ static void ResetThreadContext64(Core::ARM_Interface::ThreadContext64& context,
 
 namespace Kernel {
 
+namespace {
+
+class ThreadQueueImplForKThreadSleep final : public KThreadQueueWithoutEndWait {
+public:
+    explicit ThreadQueueImplForKThreadSleep(KernelCore& kernel_)
+        : KThreadQueueWithoutEndWait(kernel_) {}
+};
+
+class ThreadQueueImplForKThreadSetProperty final : public KThreadQueue {
+public:
+    explicit ThreadQueueImplForKThreadSetProperty(KernelCore& kernel_, KThread::WaiterList* wl)
+        : KThreadQueue(kernel_), m_wait_list(wl) {}
+
+    void CancelWait(KThread* waiting_thread, ResultCode wait_result,
+                    bool cancel_timer_task) override {
+        // Remove the thread from the wait list.
+        m_wait_list->erase(m_wait_list->iterator_to(*waiting_thread));
+
+        // Invoke the base cancel wait handler.
+        KThreadQueue::CancelWait(waiting_thread, wait_result, cancel_timer_task);
+    }
+
+private:
+    KThread::WaiterList* m_wait_list;
+};
+
+} // namespace
+
 KThread::KThread(KernelCore& kernel_)
     : KAutoObjectWithSlabHeapAndContainer{kernel_}, activity_pause_lock{kernel_} {}
 KThread::~KThread() = default;
@@ -82,6 +113,8 @@ ResultCode KThread::Initialize(KThreadFunction func, uintptr_t arg, VAddr user_s
         [[fallthrough]];
     case ThreadType::HighPriority:
         [[fallthrough]];
+    case ThreadType::Dummy:
+        [[fallthrough]];
     case ThreadType::User:
         ASSERT(((owner == nullptr) ||
                 (owner->GetCoreMask() | (1ULL << virt_core)) == owner->GetCoreMask()));
@@ -127,11 +160,8 @@ ResultCode KThread::Initialize(KThreadFunction func, uintptr_t arg, VAddr user_s
     priority = prio;
     base_priority = prio;
 
-    // Set sync object and waiting lock to null.
-    synced_object = nullptr;
-
     // Initialize sleeping queue.
-    sleeping_queue = nullptr;
+    wait_queue = nullptr;
 
     // Set suspend flags.
     suspend_request_flags = 0;
@@ -184,7 +214,7 @@ ResultCode KThread::Initialize(KThreadFunction func, uintptr_t arg, VAddr user_s
     // Setup the stack parameters.
     StackParameters& sp = GetStackParameters();
     sp.cur_thread = this;
-    sp.disable_count = 1;
+    sp.disable_count = 0;
     SetInExceptionHandler();
 
     // Set thread ID.
@@ -211,15 +241,16 @@ ResultCode KThread::InitializeThread(KThread* thread, KThreadFunction func, uint
     // Initialize the thread.
     R_TRY(thread->Initialize(func, arg, user_stack_top, prio, core, owner, type));
 
-    // Initialize host context.
+    // Initialize emulation parameters.
     thread->host_context =
         std::make_shared<Common::Fiber>(std::move(init_func), init_func_parameter);
+    thread->is_single_core = !Settings::values.use_multi_core.GetValue();
 
     return ResultSuccess;
 }
 
 ResultCode KThread::InitializeDummyThread(KThread* thread) {
-    return thread->Initialize({}, {}, {}, DefaultThreadPriority, 3, {}, ThreadType::Main);
+    return thread->Initialize({}, {}, {}, DefaultThreadPriority, 3, {}, ThreadType::Dummy);
 }
 
 ResultCode KThread::InitializeIdleThread(Core::System& system, KThread* thread, s32 virt_core) {
@@ -273,11 +304,14 @@ void KThread::Finalize() {
 
         auto it = waiter_list.begin();
         while (it != waiter_list.end()) {
-            // The thread shouldn't be a kernel waiter.
+            // Clear the lock owner
             it->SetLockOwner(nullptr);
-            it->SetSyncedObject(nullptr, ResultInvalidState);
-            it->Wakeup();
+
+            // Erase the waiter from our list.
             it = waiter_list.erase(it);
+
+            // Cancel the thread's wait.
+            it->CancelWait(ResultInvalidState, true);
         }
     }
 
@@ -294,15 +328,12 @@ bool KThread::IsSignaled() const {
     return signaled;
 }
 
-void KThread::Wakeup() {
-    KScopedSchedulerLock sl{kernel};
+void KThread::OnTimer() {
+    ASSERT(kernel.GlobalSchedulerContext().IsLocked());
 
+    // If we're waiting, cancel the wait.
     if (GetState() == ThreadState::Waiting) {
-        if (sleeping_queue != nullptr) {
-            sleeping_queue->WakeupThread(this);
-        } else {
-            SetState(ThreadState::Runnable);
-        }
+        wait_queue->CancelWait(this, ResultTimedOut, false);
     }
 }
 
@@ -327,7 +358,7 @@ void KThread::StartTermination() {
 
     // Signal.
     signaled = true;
-    NotifyAvailable();
+    KSynchronizationObject::NotifyAvailable();
 
     // Clear previous thread in KScheduler.
     KScheduler::ClearPreviousThread(kernel, this);
@@ -475,30 +506,32 @@ ResultCode KThread::GetPhysicalCoreMask(s32* out_ideal_core, u64* out_affinity_m
     return ResultSuccess;
 }
 
-ResultCode KThread::SetCoreMask(s32 cpu_core_id, u64 v_affinity_mask) {
+ResultCode KThread::SetCoreMask(s32 core_id_, u64 v_affinity_mask) {
     ASSERT(parent != nullptr);
     ASSERT(v_affinity_mask != 0);
-    KScopedLightLock lk{activity_pause_lock};
+    KScopedLightLock lk(activity_pause_lock);
 
     // Set the core mask.
     u64 p_affinity_mask = 0;
     {
-        KScopedSchedulerLock sl{kernel};
+        KScopedSchedulerLock sl(kernel);
         ASSERT(num_core_migration_disables >= 0);
 
-        // If the core id is no-update magic, preserve the ideal core id.
-        if (cpu_core_id == Svc::IdealCoreNoUpdate) {
-            cpu_core_id = virtual_ideal_core_id;
-            R_UNLESS(((1ULL << cpu_core_id) & v_affinity_mask) != 0, ResultInvalidCombination);
+        // If we're updating, set our ideal virtual core.
+        if (core_id_ != Svc::IdealCoreNoUpdate) {
+            virtual_ideal_core_id = core_id_;
+        } else {
+            // Preserve our ideal core id.
+            core_id_ = virtual_ideal_core_id;
+            R_UNLESS(((1ULL << core_id_) & v_affinity_mask) != 0, ResultInvalidCombination);
         }
 
-        // Set the virtual core/affinity mask.
-        virtual_ideal_core_id = cpu_core_id;
+        // Set our affinity mask.
         virtual_affinity_mask = v_affinity_mask;
 
         // Translate the virtual core to a physical core.
-        if (cpu_core_id >= 0) {
-            cpu_core_id = Core::Hardware::VirtualToPhysicalCoreMap[cpu_core_id];
+        if (core_id_ >= 0) {
+            core_id_ = Core::Hardware::VirtualToPhysicalCoreMap[core_id_];
         }
 
         // Translate the virtual affinity mask to a physical one.
@@ -513,7 +546,7 @@ ResultCode KThread::SetCoreMask(s32 cpu_core_id, u64 v_affinity_mask) {
             const KAffinityMask old_mask = physical_affinity_mask;
 
             // Set our new ideals.
-            physical_ideal_core_id = cpu_core_id;
+            physical_ideal_core_id = core_id_;
             physical_affinity_mask.SetAffinityMask(p_affinity_mask);
 
             if (physical_affinity_mask.GetAffinityMask() != old_mask.GetAffinityMask()) {
@@ -531,18 +564,18 @@ ResultCode KThread::SetCoreMask(s32 cpu_core_id, u64 v_affinity_mask) {
             }
         } else {
             // Otherwise, we edit the original affinity for restoration later.
-            original_physical_ideal_core_id = cpu_core_id;
+            original_physical_ideal_core_id = core_id_;
             original_physical_affinity_mask.SetAffinityMask(p_affinity_mask);
         }
     }
 
     // Update the pinned waiter list.
+    ThreadQueueImplForKThreadSetProperty wait_queue_(kernel, std::addressof(pinned_waiter_list));
     {
         bool retry_update{};
-        bool thread_is_pinned{};
         do {
             // Lock the scheduler.
-            KScopedSchedulerLock sl{kernel};
+            KScopedSchedulerLock sl(kernel);
 
             // Don't do any further management if our termination has been requested.
             R_SUCCEED_IF(IsTerminationRequested());
@@ -570,12 +603,9 @@ ResultCode KThread::SetCoreMask(s32 cpu_core_id, u64 v_affinity_mask) {
                     R_UNLESS(!GetCurrentThread(kernel).IsTerminationRequested(),
                              ResultTerminationRequested);
 
-                    // Note that the thread was pinned.
-                    thread_is_pinned = true;
-
                     // Wait until the thread isn't pinned any more.
                     pinned_waiter_list.push_back(GetCurrentThread(kernel));
-                    GetCurrentThread(kernel).SetState(ThreadState::Waiting);
+                    GetCurrentThread(kernel).BeginWait(std::addressof(wait_queue_));
                 } else {
                     // If the thread isn't pinned, release the scheduler lock and retry until it's
                     // not current.
@@ -583,16 +613,6 @@ ResultCode KThread::SetCoreMask(s32 cpu_core_id, u64 v_affinity_mask) {
                 }
             }
         } while (retry_update);
-
-        // If the thread was pinned, it no longer is, and we should remove the current thread from
-        // our waiter list.
-        if (thread_is_pinned) {
-            // Lock the scheduler.
-            KScopedSchedulerLock sl{kernel};
-
-            // Remove from the list.
-            pinned_waiter_list.erase(pinned_waiter_list.iterator_to(GetCurrentThread(kernel)));
-        }
     }
 
     return ResultSuccess;
@@ -641,15 +661,9 @@ void KThread::WaitCancel() {
     KScopedSchedulerLock sl{kernel};
 
     // Check if we're waiting and cancellable.
-    if (GetState() == ThreadState::Waiting && cancellable) {
-        if (sleeping_queue != nullptr) {
-            sleeping_queue->WakeupThread(this);
-            wait_cancelled = true;
-        } else {
-            SetSyncedObject(nullptr, ResultCancelled);
-            SetState(ThreadState::Runnable);
-            wait_cancelled = false;
-        }
+    if (this->GetState() == ThreadState::Waiting && cancellable) {
+        wait_cancelled = false;
+        wait_queue->CancelWait(this, ResultCancelled, true);
     } else {
         // Otherwise, note that we cancelled a wait.
         wait_cancelled = true;
@@ -700,60 +714,59 @@ ResultCode KThread::SetActivity(Svc::ThreadActivity activity) {
     // Set the activity.
     {
         // Lock the scheduler.
-        KScopedSchedulerLock sl{kernel};
+        KScopedSchedulerLock sl(kernel);
 
         // Verify our state.
-        const auto cur_state = GetState();
+        const auto cur_state = this->GetState();
         R_UNLESS((cur_state == ThreadState::Waiting || cur_state == ThreadState::Runnable),
                  ResultInvalidState);
 
         // Either pause or resume.
         if (activity == Svc::ThreadActivity::Paused) {
             // Verify that we're not suspended.
-            R_UNLESS(!IsSuspendRequested(SuspendType::Thread), ResultInvalidState);
+            R_UNLESS(!this->IsSuspendRequested(SuspendType::Thread), ResultInvalidState);
 
             // Suspend.
-            RequestSuspend(SuspendType::Thread);
+            this->RequestSuspend(SuspendType::Thread);
         } else {
             ASSERT(activity == Svc::ThreadActivity::Runnable);
 
             // Verify that we're suspended.
-            R_UNLESS(IsSuspendRequested(SuspendType::Thread), ResultInvalidState);
+            R_UNLESS(this->IsSuspendRequested(SuspendType::Thread), ResultInvalidState);
 
             // Resume.
-            Resume(SuspendType::Thread);
+            this->Resume(SuspendType::Thread);
         }
     }
 
     // If the thread is now paused, update the pinned waiter list.
     if (activity == Svc::ThreadActivity::Paused) {
-        bool thread_is_pinned{};
-        bool thread_is_current{};
+        ThreadQueueImplForKThreadSetProperty wait_queue_(kernel,
+                                                         std::addressof(pinned_waiter_list));
+
+        bool thread_is_current;
         do {
             // Lock the scheduler.
-            KScopedSchedulerLock sl{kernel};
+            KScopedSchedulerLock sl(kernel);
 
             // Don't do any further management if our termination has been requested.
-            R_SUCCEED_IF(IsTerminationRequested());
+            R_SUCCEED_IF(this->IsTerminationRequested());
+
+            // By default, treat the thread as not current.
+            thread_is_current = false;
 
             // Check whether the thread is pinned.
-            if (GetStackParameters().is_pinned) {
+            if (this->GetStackParameters().is_pinned) {
                 // Verify that the current thread isn't terminating.
                 R_UNLESS(!GetCurrentThread(kernel).IsTerminationRequested(),
                          ResultTerminationRequested);
 
-                // Note that the thread was pinned and not current.
-                thread_is_pinned = true;
-                thread_is_current = false;
-
                 // Wait until the thread isn't pinned any more.
                 pinned_waiter_list.push_back(GetCurrentThread(kernel));
-                GetCurrentThread(kernel).SetState(ThreadState::Waiting);
+                GetCurrentThread(kernel).BeginWait(std::addressof(wait_queue_));
             } else {
                 // Check if the thread is currently running.
                 // If it is, we'll need to retry.
-                thread_is_current = false;
-
                 for (auto i = 0; i < static_cast<s32>(Core::Hardware::NUM_CPU_CORES); ++i) {
                     if (kernel.Scheduler(i).GetCurrentThread() == this) {
                         thread_is_current = true;
@@ -762,16 +775,6 @@ ResultCode KThread::SetActivity(Svc::ThreadActivity activity) {
                 }
             }
         } while (thread_is_current);
-
-        // If the thread was pinned, it no longer is, and we should remove the current thread from
-        // our waiter list.
-        if (thread_is_pinned) {
-            // Lock the scheduler.
-            KScopedSchedulerLock sl{kernel};
-
-            // Remove from the list.
-            pinned_waiter_list.erase(pinned_waiter_list.iterator_to(GetCurrentThread(kernel)));
-        }
     }
 
     return ResultSuccess;
@@ -966,6 +969,9 @@ ResultCode KThread::Run() {
 
         // Set our state and finish.
         SetState(ThreadState::Runnable);
+
+        DisableDispatch();
+
         return ResultSuccess;
     }
 }
@@ -996,27 +1002,61 @@ ResultCode KThread::Sleep(s64 timeout) {
     ASSERT(this == GetCurrentThreadPointer(kernel));
     ASSERT(timeout > 0);
 
+    ThreadQueueImplForKThreadSleep wait_queue_(kernel);
     {
         // Setup the scheduling lock and sleep.
-        KScopedSchedulerLockAndSleep slp{kernel, this, timeout};
+        KScopedSchedulerLockAndSleep slp(kernel, this, timeout);
 
         // Check if the thread should terminate.
-        if (IsTerminationRequested()) {
+        if (this->IsTerminationRequested()) {
             slp.CancelSleep();
             return ResultTerminationRequested;
         }
 
-        // Mark the thread as waiting.
-        SetState(ThreadState::Waiting);
+        // Wait for the sleep to end.
+        this->BeginWait(std::addressof(wait_queue_));
         SetWaitReasonForDebugging(ThreadWaitReasonForDebugging::Sleep);
     }
 
-    // The lock/sleep is done.
+    return ResultSuccess;
+}
 
-    // Cancel the timer.
-    kernel.TimeManager().UnscheduleTimeEvent(this);
+void KThread::BeginWait(KThreadQueue* queue) {
+    // Set our state as waiting.
+    SetState(ThreadState::Waiting);
 
-    return ResultSuccess;
+    // Set our wait queue.
+    wait_queue = queue;
+}
+
+void KThread::NotifyAvailable(KSynchronizationObject* signaled_object, ResultCode wait_result_) {
+    // Lock the scheduler.
+    KScopedSchedulerLock sl(kernel);
+
+    // If we're waiting, notify our queue that we're available.
+    if (GetState() == ThreadState::Waiting) {
+        wait_queue->NotifyAvailable(this, signaled_object, wait_result_);
+    }
+}
+
+void KThread::EndWait(ResultCode wait_result_) {
+    // Lock the scheduler.
+    KScopedSchedulerLock sl(kernel);
+
+    // If we're waiting, notify our queue that we're available.
+    if (GetState() == ThreadState::Waiting) {
+        wait_queue->EndWait(this, wait_result_);
+    }
+}
+
+void KThread::CancelWait(ResultCode wait_result_, bool cancel_timer_task) {
+    // Lock the scheduler.
+    KScopedSchedulerLock sl(kernel);
+
+    // If we're waiting, notify our queue that we're available.
+    if (GetState() == ThreadState::Waiting) {
+        wait_queue->CancelWait(this, wait_result_, cancel_timer_task);
+    }
 }
 
 void KThread::SetState(ThreadState state) {
@@ -1050,4 +1090,26 @@ s32 GetCurrentCoreId(KernelCore& kernel) {
     return GetCurrentThread(kernel).GetCurrentCore();
 }
 
+KScopedDisableDispatch::~KScopedDisableDispatch() {
+    // If we are shutting down the kernel, none of this is relevant anymore.
+    if (kernel.IsShuttingDown()) {
+        return;
+    }
+
+    // Skip the reschedule if single-core, as dispatch tracking is disabled here.
+    if (!Settings::values.use_multi_core.GetValue()) {
+        return;
+    }
+
+    if (GetCurrentThread(kernel).GetDisableDispatchCount() <= 1) {
+        auto scheduler = kernel.CurrentScheduler();
+
+        if (scheduler) {
+            scheduler->RescheduleCurrentCore();
+        }
+    } else {
+        GetCurrentThread(kernel).EnableDispatch();
+    }
+}
+
 } // namespace Kernel
diff --git a/src/core/hle/kernel/k_thread.h b/src/core/hle/kernel/k_thread.h
index c77f44ad4..c8a08bd71 100644
--- a/src/core/hle/kernel/k_thread.h
+++ b/src/core/hle/kernel/k_thread.h
@@ -48,6 +48,7 @@ enum class ThreadType : u32 {
     Kernel = 1,
     HighPriority = 2,
     User = 3,
+    Dummy = 100, // Special thread type for emulation purposes only
 };
 DECLARE_ENUM_FLAG_OPERATORS(ThreadType);
 
@@ -161,8 +162,6 @@ public:
         }
     }
 
-    void Wakeup();
-
     void SetBasePriority(s32 value);
 
     [[nodiscard]] ResultCode Run();
@@ -197,13 +196,19 @@ public:
 
     void Suspend();
 
-    void SetSyncedObject(KSynchronizationObject* obj, ResultCode wait_res) {
-        synced_object = obj;
+    constexpr void SetSyncedIndex(s32 index) {
+        synced_index = index;
+    }
+
+    [[nodiscard]] constexpr s32 GetSyncedIndex() const {
+        return synced_index;
+    }
+
+    constexpr void SetWaitResult(ResultCode wait_res) {
         wait_result = wait_res;
     }
 
-    [[nodiscard]] ResultCode GetWaitResult(KSynchronizationObject** out) const {
-        *out = synced_object;
+    [[nodiscard]] constexpr ResultCode GetWaitResult() const {
         return wait_result;
     }
 
@@ -374,6 +379,8 @@ public:
 
     [[nodiscard]] bool IsSignaled() const override;
 
+    void OnTimer();
+
     static void PostDestroy(uintptr_t arg);
 
     [[nodiscard]] static ResultCode InitializeDummyThread(KThread* thread);
@@ -446,20 +453,39 @@ public:
         return per_core_priority_queue_entry[core];
     }
 
-    void SetSleepingQueue(KThreadQueue* q) {
-        sleeping_queue = q;
+    [[nodiscard]] bool IsKernelThread() const {
+        return GetActiveCore() == 3;
+    }
+
+    [[nodiscard]] bool IsDispatchTrackingDisabled() const {
+        return is_single_core || IsKernelThread();
     }
 
     [[nodiscard]] s32 GetDisableDispatchCount() const {
+        if (IsDispatchTrackingDisabled()) {
+            // TODO(bunnei): Until kernel threads are emulated, we cannot enable/disable dispatch.
+            return 1;
+        }
+
         return this->GetStackParameters().disable_count;
     }
 
     void DisableDispatch() {
+        if (IsDispatchTrackingDisabled()) {
+            // TODO(bunnei): Until kernel threads are emulated, we cannot enable/disable dispatch.
+            return;
+        }
+
         ASSERT(GetCurrentThread(kernel).GetDisableDispatchCount() >= 0);
         this->GetStackParameters().disable_count++;
     }
 
     void EnableDispatch() {
+        if (IsDispatchTrackingDisabled()) {
+            // TODO(bunnei): Until kernel threads are emulated, we cannot enable/disable dispatch.
+            return;
+        }
+
         ASSERT(GetCurrentThread(kernel).GetDisableDispatchCount() > 0);
         this->GetStackParameters().disable_count--;
     }
@@ -573,6 +599,15 @@ public:
         address_key_value = val;
     }
 
+    void ClearWaitQueue() {
+        wait_queue = nullptr;
+    }
+
+    void BeginWait(KThreadQueue* queue);
+    void NotifyAvailable(KSynchronizationObject* signaled_object, ResultCode wait_result_);
+    void EndWait(ResultCode wait_result_);
+    void CancelWait(ResultCode wait_result_, bool cancel_timer_task);
+
     [[nodiscard]] bool HasWaiters() const {
         return !waiter_list.empty();
     }
@@ -667,7 +702,6 @@ private:
     KAffinityMask physical_affinity_mask{};
     u64 thread_id{};
     std::atomic<s64> cpu_time{};
-    KSynchronizationObject* synced_object{};
     VAddr address_key{};
     KProcess* parent{};
     VAddr kernel_stack_top{};
@@ -677,13 +711,14 @@ private:
     s64 schedule_count{};
     s64 last_scheduled_tick{};
     std::array<QueueEntry, Core::Hardware::NUM_CPU_CORES> per_core_priority_queue_entry{};
-    KThreadQueue* sleeping_queue{};
+    KThreadQueue* wait_queue{};
     WaiterList waiter_list{};
     WaiterList pinned_waiter_list{};
     KThread* lock_owner{};
     u32 address_key_value{};
     u32 suspend_request_flags{};
     u32 suspend_allowed_flags{};
+    s32 synced_index{};
     ResultCode wait_result{ResultSuccess};
     s32 base_priority{};
     s32 physical_ideal_core_id{};
@@ -708,6 +743,7 @@ private:
 
     // For emulation
     std::shared_ptr<Common::Fiber> host_context{};
+    bool is_single_core{};
 
     // For debugging
     std::vector<KSynchronizationObject*> wait_objects_for_debugging;
@@ -752,4 +788,20 @@ public:
     }
 };
 
+class KScopedDisableDispatch {
+public:
+    [[nodiscard]] explicit KScopedDisableDispatch(KernelCore& kernel_) : kernel{kernel_} {
+        // If we are shutting down the kernel, none of this is relevant anymore.
+        if (kernel.IsShuttingDown()) {
+            return;
+        }
+        GetCurrentThread(kernel).DisableDispatch();
+    }
+
+    ~KScopedDisableDispatch();
+
+private:
+    KernelCore& kernel;
+};
+
 } // namespace Kernel
diff --git a/src/core/hle/kernel/k_thread_queue.cpp b/src/core/hle/kernel/k_thread_queue.cpp
new file mode 100644
index 000000000..46f27172b
--- /dev/null
+++ b/src/core/hle/kernel/k_thread_queue.cpp
@@ -0,0 +1,51 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/kernel/k_thread_queue.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/time_manager.h"
+
+namespace Kernel {
+
+void KThreadQueue::NotifyAvailable([[maybe_unused]] KThread* waiting_thread,
+                                   [[maybe_unused]] KSynchronizationObject* signaled_object,
+                                   [[maybe_unused]] ResultCode wait_result) {}
+
+void KThreadQueue::EndWait(KThread* waiting_thread, ResultCode wait_result) {
+    // Set the thread's wait result.
+    waiting_thread->SetWaitResult(wait_result);
+
+    // Set the thread as runnable.
+    waiting_thread->SetState(ThreadState::Runnable);
+
+    // Clear the thread's wait queue.
+    waiting_thread->ClearWaitQueue();
+
+    // Cancel the thread task.
+    kernel.TimeManager().UnscheduleTimeEvent(waiting_thread);
+}
+
+void KThreadQueue::CancelWait(KThread* waiting_thread, ResultCode wait_result,
+                              bool cancel_timer_task) {
+    // Set the thread's wait result.
+    waiting_thread->SetWaitResult(wait_result);
+
+    // Set the thread as runnable.
+    waiting_thread->SetState(ThreadState::Runnable);
+
+    // Clear the thread's wait queue.
+    waiting_thread->ClearWaitQueue();
+
+    // Cancel the thread task.
+    if (cancel_timer_task) {
+        kernel.TimeManager().UnscheduleTimeEvent(waiting_thread);
+    }
+}
+
+void KThreadQueueWithoutEndWait::EndWait([[maybe_unused]] KThread* waiting_thread,
+                                         [[maybe_unused]] ResultCode wait_result) {}
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_thread_queue.h b/src/core/hle/kernel/k_thread_queue.h
index 35d471dc5..ccb718e49 100644
--- a/src/core/hle/kernel/k_thread_queue.h
+++ b/src/core/hle/kernel/k_thread_queue.h
@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/k_thread.h"
 
 namespace Kernel {
@@ -11,71 +12,24 @@ namespace Kernel {
 class KThreadQueue {
 public:
     explicit KThreadQueue(KernelCore& kernel_) : kernel{kernel_} {}
+    virtual ~KThreadQueue() = default;
 
-    bool IsEmpty() const {
-        return wait_list.empty();
-    }
-
-    KThread::WaiterList::iterator begin() {
-        return wait_list.begin();
-    }
-    KThread::WaiterList::iterator end() {
-        return wait_list.end();
-    }
-
-    bool SleepThread(KThread* t) {
-        KScopedSchedulerLock sl{kernel};
-
-        // If the thread needs terminating, don't enqueue it.
-        if (t->IsTerminationRequested()) {
-            return false;
-        }
-
-        // Set the thread's queue and mark it as waiting.
-        t->SetSleepingQueue(this);
-        t->SetState(ThreadState::Waiting);
-
-        // Add the thread to the queue.
-        wait_list.push_back(*t);
-
-        return true;
-    }
-
-    void WakeupThread(KThread* t) {
-        KScopedSchedulerLock sl{kernel};
-
-        // Remove the thread from the queue.
-        wait_list.erase(wait_list.iterator_to(*t));
-
-        // Mark the thread as no longer sleeping.
-        t->SetState(ThreadState::Runnable);
-        t->SetSleepingQueue(nullptr);
-    }
-
-    KThread* WakeupFrontThread() {
-        KScopedSchedulerLock sl{kernel};
-
-        if (wait_list.empty()) {
-            return nullptr;
-        } else {
-            // Remove the thread from the queue.
-            auto it = wait_list.begin();
-            KThread* thread = std::addressof(*it);
-            wait_list.erase(it);
-
-            ASSERT(thread->GetState() == ThreadState::Waiting);
-
-            // Mark the thread as no longer sleeping.
-            thread->SetState(ThreadState::Runnable);
-            thread->SetSleepingQueue(nullptr);
-
-            return thread;
-        }
-    }
+    virtual void NotifyAvailable(KThread* waiting_thread, KSynchronizationObject* signaled_object,
+                                 ResultCode wait_result);
+    virtual void EndWait(KThread* waiting_thread, ResultCode wait_result);
+    virtual void CancelWait(KThread* waiting_thread, ResultCode wait_result,
+                            bool cancel_timer_task);
 
 private:
     KernelCore& kernel;
     KThread::WaiterList wait_list{};
 };
 
+class KThreadQueueWithoutEndWait : public KThreadQueue {
+public:
+    explicit KThreadQueueWithoutEndWait(KernelCore& kernel_) : KThreadQueue(kernel_) {}
+
+    void EndWait(KThread* waiting_thread, ResultCode wait_result) override final;
+};
+
 } // namespace Kernel
diff --git a/src/core/hle/kernel/kernel.cpp b/src/core/hle/kernel/kernel.cpp
index 45e86a677..2e4e4cb1c 100644
--- a/src/core/hle/kernel/kernel.cpp
+++ b/src/core/hle/kernel/kernel.cpp
@@ -14,6 +14,7 @@
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "common/microprofile.h"
+#include "common/scope_exit.h"
 #include "common/thread.h"
 #include "common/thread_worker.h"
 #include "core/arm/arm_interface.h"
@@ -83,12 +84,16 @@ struct KernelCore::Impl {
     }
 
     void InitializeCores() {
-        for (auto& core : cores) {
-            core.Initialize(current_process->Is64BitProcess());
+        for (u32 core_id = 0; core_id < Core::Hardware::NUM_CPU_CORES; core_id++) {
+            cores[core_id].Initialize(current_process->Is64BitProcess());
+            system.Memory().SetCurrentPageTable(*current_process, core_id);
         }
     }
 
     void Shutdown() {
+        is_shutting_down.store(true, std::memory_order_relaxed);
+        SCOPE_EXIT({ is_shutting_down.store(false, std::memory_order_relaxed); });
+
         process_list.clear();
 
         // Close all open server ports.
@@ -123,15 +128,6 @@ struct KernelCore::Impl {
         next_user_process_id = KProcess::ProcessIDMin;
         next_thread_id = 1;
 
-        for (u32 core_id = 0; core_id < Core::Hardware::NUM_CPU_CORES; core_id++) {
-            if (suspend_threads[core_id]) {
-                suspend_threads[core_id]->Close();
-                suspend_threads[core_id] = nullptr;
-            }
-
-            schedulers[core_id].reset();
-        }
-
         cores.clear();
 
         global_handle_table->Finalize();
@@ -159,6 +155,16 @@ struct KernelCore::Impl {
         CleanupObject(time_shared_mem);
         CleanupObject(system_resource_limit);
 
+        for (u32 core_id = 0; core_id < Core::Hardware::NUM_CPU_CORES; core_id++) {
+            if (suspend_threads[core_id]) {
+                suspend_threads[core_id]->Close();
+                suspend_threads[core_id] = nullptr;
+            }
+
+            schedulers[core_id]->Finalize();
+            schedulers[core_id].reset();
+        }
+
         // Next host thead ID to use, 0-3 IDs represent core threads, >3 represent others
         next_host_thread_id = Core::Hardware::NUM_CPU_CORES;
 
@@ -245,13 +251,11 @@ struct KernelCore::Impl {
                     KScopedSchedulerLock lock(kernel);
                     global_scheduler_context->PreemptThreads();
                 }
-                const auto time_interval = std::chrono::nanoseconds{
-                    Core::Timing::msToCycles(std::chrono::milliseconds(10))};
+                const auto time_interval = std::chrono::nanoseconds{std::chrono::milliseconds(10)};
                 system.CoreTiming().ScheduleEvent(time_interval, preemption_event);
             });
 
-        const auto time_interval =
-            std::chrono::nanoseconds{Core::Timing::msToCycles(std::chrono::milliseconds(10))};
+        const auto time_interval = std::chrono::nanoseconds{std::chrono::milliseconds(10)};
         system.CoreTiming().ScheduleEvent(time_interval, preemption_event);
     }
 
@@ -267,14 +271,6 @@ struct KernelCore::Impl {
 
     void MakeCurrentProcess(KProcess* process) {
         current_process = process;
-        if (process == nullptr) {
-            return;
-        }
-
-        const u32 core_id = GetCurrentHostThreadID();
-        if (core_id < Core::Hardware::NUM_CPU_CORES) {
-            system.Memory().SetCurrentPageTable(*process, core_id);
-        }
     }
 
     static inline thread_local u32 host_thread_id = UINT32_MAX;
@@ -344,7 +340,16 @@ struct KernelCore::Impl {
         is_phantom_mode_for_singlecore = value;
     }
 
+    bool IsShuttingDown() const {
+        return is_shutting_down.load(std::memory_order_relaxed);
+    }
+
     KThread* GetCurrentEmuThread() {
+        // If we are shutting down the kernel, none of this is relevant anymore.
+        if (IsShuttingDown()) {
+            return {};
+        }
+
         const auto thread_id = GetCurrentHostThreadID();
         if (thread_id >= Core::Hardware::NUM_CPU_CORES) {
             return GetHostDummyThread();
@@ -760,6 +765,7 @@ struct KernelCore::Impl {
     std::vector<std::unique_ptr<KThread>> dummy_threads;
 
     bool is_multicore{};
+    std::atomic_bool is_shutting_down{};
     bool is_phantom_mode_for_singlecore{};
     u32 single_core_thread_id{};
 
@@ -845,16 +851,20 @@ const Kernel::PhysicalCore& KernelCore::PhysicalCore(std::size_t id) const {
     return impl->cores[id];
 }
 
+size_t KernelCore::CurrentPhysicalCoreIndex() const {
+    const u32 core_id = impl->GetCurrentHostThreadID();
+    if (core_id >= Core::Hardware::NUM_CPU_CORES) {
+        return Core::Hardware::NUM_CPU_CORES - 1;
+    }
+    return core_id;
+}
+
 Kernel::PhysicalCore& KernelCore::CurrentPhysicalCore() {
-    u32 core_id = impl->GetCurrentHostThreadID();
-    ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
-    return impl->cores[core_id];
+    return impl->cores[CurrentPhysicalCoreIndex()];
 }
 
 const Kernel::PhysicalCore& KernelCore::CurrentPhysicalCore() const {
-    u32 core_id = impl->GetCurrentHostThreadID();
-    ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
-    return impl->cores[core_id];
+    return impl->cores[CurrentPhysicalCoreIndex()];
 }
 
 Kernel::KScheduler* KernelCore::CurrentScheduler() {
@@ -1057,6 +1067,9 @@ void KernelCore::Suspend(bool in_suspention) {
             impl->suspend_threads[core_id]->SetState(state);
             impl->suspend_threads[core_id]->SetWaitReasonForDebugging(
                 ThreadWaitReasonForDebugging::Suspended);
+            if (!should_suspend) {
+                impl->suspend_threads[core_id]->DisableDispatch();
+            }
         }
     }
 }
@@ -1065,19 +1078,21 @@ bool KernelCore::IsMulticore() const {
     return impl->is_multicore;
 }
 
+bool KernelCore::IsShuttingDown() const {
+    return impl->IsShuttingDown();
+}
+
 void KernelCore::ExceptionalExit() {
     exception_exited = true;
     Suspend(true);
 }
 
 void KernelCore::EnterSVCProfile() {
-    std::size_t core = impl->GetCurrentHostThreadID();
-    impl->svc_ticks[core] = MicroProfileEnter(MICROPROFILE_TOKEN(Kernel_SVC));
+    impl->svc_ticks[CurrentPhysicalCoreIndex()] = MicroProfileEnter(MICROPROFILE_TOKEN(Kernel_SVC));
 }
 
 void KernelCore::ExitSVCProfile() {
-    std::size_t core = impl->GetCurrentHostThreadID();
-    MicroProfileLeave(MICROPROFILE_TOKEN(Kernel_SVC), impl->svc_ticks[core]);
+    MicroProfileLeave(MICROPROFILE_TOKEN(Kernel_SVC), impl->svc_ticks[CurrentPhysicalCoreIndex()]);
 }
 
 std::weak_ptr<Kernel::ServiceThread> KernelCore::CreateServiceThread(const std::string& name) {
diff --git a/src/core/hle/kernel/kernel.h b/src/core/hle/kernel/kernel.h
index d2ceae950..b9b423908 100644
--- a/src/core/hle/kernel/kernel.h
+++ b/src/core/hle/kernel/kernel.h
@@ -53,6 +53,7 @@ class KSharedMemoryInfo;
 class KThread;
 class KTransferMemory;
 class KWritableEvent;
+class KCodeMemory;
 class PhysicalCore;
 class ServiceThread;
 class Synchronization;
@@ -148,6 +149,9 @@ public:
     /// Gets the an instance of the respective physical CPU core.
     const Kernel::PhysicalCore& PhysicalCore(std::size_t id) const;
 
+    /// Gets the current physical core index for the running host thread.
+    std::size_t CurrentPhysicalCoreIndex() const;
+
     /// Gets the sole instance of the Scheduler at the current running core.
     Kernel::KScheduler* CurrentScheduler();
 
@@ -271,6 +275,8 @@ public:
 
     bool IsMulticore() const;
 
+    bool IsShuttingDown() const;
+
     void EnterSVCProfile();
 
     void ExitSVCProfile();
@@ -326,6 +332,8 @@ public:
             return slab_heap_container->transfer_memory;
         } else if constexpr (std::is_same_v<T, KWritableEvent>) {
             return slab_heap_container->writeable_event;
+        } else if constexpr (std::is_same_v<T, KCodeMemory>) {
+            return slab_heap_container->code_memory;
         }
     }
 
@@ -377,6 +385,7 @@ private:
         KSlabHeap<KThread> thread;
         KSlabHeap<KTransferMemory> transfer_memory;
         KSlabHeap<KWritableEvent> writeable_event;
+        KSlabHeap<KCodeMemory> code_memory;
     };
 
     std::unique_ptr<SlabHeapContainer> slab_heap_container;
diff --git a/src/core/hle/kernel/service_thread.cpp b/src/core/hle/kernel/service_thread.cpp
index 6721b6276..03f3dec10 100644
--- a/src/core/hle/kernel/service_thread.cpp
+++ b/src/core/hle/kernel/service_thread.cpp
@@ -25,24 +25,27 @@ public:
     void QueueSyncRequest(KSession& session, std::shared_ptr<HLERequestContext>&& context);
 
 private:
-    std::vector<std::thread> threads;
+    std::vector<std::jthread> threads;
     std::queue<std::function<void()>> requests;
     std::mutex queue_mutex;
-    std::condition_variable condition;
+    std::condition_variable_any condition;
     const std::string service_name;
-    bool stop{};
 };
 
 ServiceThread::Impl::Impl(KernelCore& kernel, std::size_t num_threads, const std::string& name)
     : service_name{name} {
-    for (std::size_t i = 0; i < num_threads; ++i)
-        threads.emplace_back([this, &kernel] {
+    for (std::size_t i = 0; i < num_threads; ++i) {
+        threads.emplace_back([this, &kernel](std::stop_token stop_token) {
             Common::SetCurrentThreadName(std::string{"yuzu:HleService:" + service_name}.c_str());
 
             // Wait for first request before trying to acquire a render context
             {
                 std::unique_lock lock{queue_mutex};
-                condition.wait(lock, [this] { return stop || !requests.empty(); });
+                condition.wait(lock, stop_token, [this] { return !requests.empty(); });
+            }
+
+            if (stop_token.stop_requested()) {
+                return;
             }
 
             kernel.RegisterHostThread();
@@ -52,10 +55,16 @@ ServiceThread::Impl::Impl(KernelCore& kernel, std::size_t num_threads, const std
 
                 {
                     std::unique_lock lock{queue_mutex};
-                    condition.wait(lock, [this] { return stop || !requests.empty(); });
-                    if (stop || requests.empty()) {
+                    condition.wait(lock, stop_token, [this] { return !requests.empty(); });
+
+                    if (stop_token.stop_requested()) {
                         return;
                     }
+
+                    if (requests.empty()) {
+                        continue;
+                    }
+
                     task = std::move(requests.front());
                     requests.pop();
                 }
@@ -63,6 +72,7 @@ ServiceThread::Impl::Impl(KernelCore& kernel, std::size_t num_threads, const std
                 task();
             }
         });
+    }
 }
 
 void ServiceThread::Impl::QueueSyncRequest(KSession& session,
@@ -88,12 +98,9 @@ void ServiceThread::Impl::QueueSyncRequest(KSession& session,
 }
 
 ServiceThread::Impl::~Impl() {
-    {
-        std::unique_lock lock{queue_mutex};
-        stop = true;
-    }
     condition.notify_all();
-    for (std::thread& thread : threads) {
+    for (auto& thread : threads) {
+        thread.request_stop();
         thread.join();
     }
 }
diff --git a/src/core/hle/kernel/svc.cpp b/src/core/hle/kernel/svc.cpp
index f0cd8471e..a9f7438ea 100644
--- a/src/core/hle/kernel/svc.cpp
+++ b/src/core/hle/kernel/svc.cpp
@@ -18,6 +18,7 @@
 #include "core/core_timing.h"
 #include "core/hle/kernel/k_client_port.h"
 #include "core/hle/kernel/k_client_session.h"
+#include "core/hle/kernel/k_code_memory.h"
 #include "core/hle/kernel/k_event.h"
 #include "core/hle/kernel/k_handle_table.h"
 #include "core/hle/kernel/k_memory_block.h"
@@ -31,6 +32,7 @@
 #include "core/hle/kernel/k_shared_memory.h"
 #include "core/hle/kernel/k_synchronization_object.h"
 #include "core/hle/kernel/k_thread.h"
+#include "core/hle/kernel/k_thread_queue.h"
 #include "core/hle/kernel/k_transfer_memory.h"
 #include "core/hle/kernel/k_writable_event.h"
 #include "core/hle/kernel/kernel.h"
@@ -307,26 +309,29 @@ static ResultCode ConnectToNamedPort32(Core::System& system, Handle* out_handle,
 
 /// Makes a blocking IPC call to an OS service.
 static ResultCode SendSyncRequest(Core::System& system, Handle handle) {
-
     auto& kernel = system.Kernel();
 
+    // Create the wait queue.
+    KThreadQueue wait_queue(kernel);
+
+    // Get the client session from its handle.
+    KScopedAutoObject session =
+        kernel.CurrentProcess()->GetHandleTable().GetObject<KClientSession>(handle);
+    R_UNLESS(session.IsNotNull(), ResultInvalidHandle);
+
+    LOG_TRACE(Kernel_SVC, "called handle=0x{:08X}({})", handle, session->GetName());
+
     auto thread = kernel.CurrentScheduler()->GetCurrentThread();
     {
         KScopedSchedulerLock lock(kernel);
-        thread->SetState(ThreadState::Waiting);
-        thread->SetWaitReasonForDebugging(ThreadWaitReasonForDebugging::IPC);
-
-        {
-            KScopedAutoObject session =
-                kernel.CurrentProcess()->GetHandleTable().GetObject<KClientSession>(handle);
-            R_UNLESS(session.IsNotNull(), ResultInvalidHandle);
-            LOG_TRACE(Kernel_SVC, "called handle=0x{:08X}({})", handle, session->GetName());
-            session->SendSyncRequest(thread, system.Memory(), system.CoreTiming());
-        }
+
+        // This is a synchronous request, so we should wait for our request to complete.
+        GetCurrentThread(kernel).BeginWait(std::addressof(wait_queue));
+        GetCurrentThread(kernel).SetWaitReasonForDebugging(ThreadWaitReasonForDebugging::IPC);
+        session->SendSyncRequest(&GetCurrentThread(kernel), system.Memory(), system.CoreTiming());
     }
 
-    KSynchronizationObject* dummy{};
-    return thread->GetWaitResult(std::addressof(dummy));
+    return thread->GetWaitResult();
 }
 
 static ResultCode SendSyncRequest32(Core::System& system, Handle handle) {
@@ -873,7 +878,7 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, Handle
             const u64 thread_ticks = current_thread->GetCpuTime();
 
             out_ticks = thread_ticks + (core_timing.GetCPUTicks() - prev_ctx_ticks);
-        } else if (same_thread && info_sub_id == system.CurrentCoreIndex()) {
+        } else if (same_thread && info_sub_id == system.Kernel().CurrentPhysicalCoreIndex()) {
             out_ticks = core_timing.GetCPUTicks() - prev_ctx_ticks;
         }
 
@@ -887,7 +892,8 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, Handle
             return ResultInvalidHandle;
         }
 
-        if (info_sub_id != 0xFFFFFFFFFFFFFFFF && info_sub_id != system.CurrentCoreIndex()) {
+        if (info_sub_id != 0xFFFFFFFFFFFFFFFF &&
+            info_sub_id != system.Kernel().CurrentPhysicalCoreIndex()) {
             LOG_ERROR(Kernel_SVC, "Core is not the current core, got {}", info_sub_id);
             return ResultInvalidCombination;
         }
@@ -1197,6 +1203,22 @@ constexpr bool IsValidProcessMemoryPermission(Svc::MemoryPermission perm) {
     }
 }
 
+constexpr bool IsValidMapCodeMemoryPermission(Svc::MemoryPermission perm) {
+    return perm == Svc::MemoryPermission::ReadWrite;
+}
+
+constexpr bool IsValidMapToOwnerCodeMemoryPermission(Svc::MemoryPermission perm) {
+    return perm == Svc::MemoryPermission::Read || perm == Svc::MemoryPermission::ReadExecute;
+}
+
+constexpr bool IsValidUnmapCodeMemoryPermission(Svc::MemoryPermission perm) {
+    return perm == Svc::MemoryPermission::None;
+}
+
+constexpr bool IsValidUnmapFromOwnerCodeMemoryPermission(Svc::MemoryPermission perm) {
+    return perm == Svc::MemoryPermission::None;
+}
+
 } // Anonymous namespace
 
 static ResultCode MapSharedMemory(Core::System& system, Handle shmem_handle, VAddr address,
@@ -1306,6 +1328,195 @@ static ResultCode SetProcessMemoryPermission(Core::System& system, Handle proces
     return page_table.SetProcessMemoryPermission(address, size, ConvertToKMemoryPermission(perm));
 }
 
+static ResultCode MapProcessMemory(Core::System& system, VAddr dst_address, Handle process_handle,
+                                   VAddr src_address, u64 size) {
+    LOG_TRACE(Kernel_SVC,
+              "called, dst_address=0x{:X}, process_handle=0x{:X}, src_address=0x{:X}, size=0x{:X}",
+              dst_address, process_handle, src_address, size);
+
+    // Validate the address/size.
+    R_UNLESS(Common::IsAligned(dst_address, PageSize), ResultInvalidAddress);
+    R_UNLESS(Common::IsAligned(src_address, PageSize), ResultInvalidAddress);
+    R_UNLESS(Common::IsAligned(size, PageSize), ResultInvalidSize);
+    R_UNLESS(size > 0, ResultInvalidSize);
+    R_UNLESS((dst_address < dst_address + size), ResultInvalidCurrentMemory);
+    R_UNLESS((src_address < src_address + size), ResultInvalidCurrentMemory);
+
+    // Get the processes.
+    KProcess* dst_process = system.CurrentProcess();
+    KScopedAutoObject src_process =
+        dst_process->GetHandleTable().GetObjectWithoutPseudoHandle<KProcess>(process_handle);
+    R_UNLESS(src_process.IsNotNull(), ResultInvalidHandle);
+
+    // Get the page tables.
+    auto& dst_pt = dst_process->PageTable();
+    auto& src_pt = src_process->PageTable();
+
+    // Validate that the mapping is in range.
+    R_UNLESS(src_pt.Contains(src_address, size), ResultInvalidCurrentMemory);
+    R_UNLESS(dst_pt.CanContain(dst_address, size, KMemoryState::SharedCode),
+             ResultInvalidMemoryRegion);
+
+    // Create a new page group.
+    KMemoryInfo kBlockInfo = dst_pt.QueryInfo(dst_address);
+    KPageLinkedList pg(kBlockInfo.GetAddress(), kBlockInfo.GetNumPages());
+
+    // Map the group.
+    R_TRY(dst_pt.MapPages(dst_address, pg, KMemoryState::SharedCode,
+                          KMemoryPermission::UserReadWrite));
+
+    return ResultSuccess;
+}
+
+static ResultCode UnmapProcessMemory(Core::System& system, VAddr dst_address, Handle process_handle,
+                                     VAddr src_address, u64 size) {
+    LOG_TRACE(Kernel_SVC,
+              "called, dst_address=0x{:X}, process_handle=0x{:X}, src_address=0x{:X}, size=0x{:X}",
+              dst_address, process_handle, src_address, size);
+
+    // Validate the address/size.
+    R_UNLESS(Common::IsAligned(dst_address, PageSize), ResultInvalidAddress);
+    R_UNLESS(Common::IsAligned(src_address, PageSize), ResultInvalidAddress);
+    R_UNLESS(Common::IsAligned(size, PageSize), ResultInvalidSize);
+    R_UNLESS(size > 0, ResultInvalidSize);
+    R_UNLESS((dst_address < dst_address + size), ResultInvalidCurrentMemory);
+    R_UNLESS((src_address < src_address + size), ResultInvalidCurrentMemory);
+
+    // Get the processes.
+    KProcess* dst_process = system.CurrentProcess();
+    KScopedAutoObject src_process =
+        dst_process->GetHandleTable().GetObjectWithoutPseudoHandle<KProcess>(process_handle);
+    R_UNLESS(src_process.IsNotNull(), ResultInvalidHandle);
+
+    // Get the page tables.
+    auto& dst_pt = dst_process->PageTable();
+    auto& src_pt = src_process->PageTable();
+
+    // Validate that the mapping is in range.
+    R_UNLESS(src_pt.Contains(src_address, size), ResultInvalidCurrentMemory);
+    R_UNLESS(dst_pt.CanContain(dst_address, size, KMemoryState::SharedCode),
+             ResultInvalidMemoryRegion);
+
+    // Unmap the memory.
+    R_TRY(dst_pt.UnmapProcessMemory(dst_address, size, src_pt, src_address));
+
+    return ResultSuccess;
+}
+
+static ResultCode CreateCodeMemory(Core::System& system, Handle* out, VAddr address, size_t size) {
+    LOG_TRACE(Kernel_SVC, "called, handle_out=0x{:X}, address=0x{:X}, size=0x{:X}",
+              static_cast<void*>(out), address, size);
+    // Get kernel instance.
+    auto& kernel = system.Kernel();
+
+    // Validate address / size.
+    R_UNLESS(Common::IsAligned(address, PageSize), ResultInvalidAddress);
+    R_UNLESS(Common::IsAligned(size, PageSize), ResultInvalidSize);
+    R_UNLESS(size > 0, ResultInvalidSize);
+    R_UNLESS((address < address + size), ResultInvalidCurrentMemory);
+
+    // Create the code memory.
+
+    KCodeMemory* code_mem = KCodeMemory::Create(kernel);
+    R_UNLESS(code_mem != nullptr, ResultOutOfResource);
+
+    // Verify that the region is in range.
+    R_UNLESS(system.CurrentProcess()->PageTable().Contains(address, size),
+             ResultInvalidCurrentMemory);
+
+    // Initialize the code memory.
+    R_TRY(code_mem->Initialize(system.DeviceMemory(), address, size));
+
+    // Register the code memory.
+    KCodeMemory::Register(kernel, code_mem);
+
+    // Add the code memory to the handle table.
+    R_TRY(system.CurrentProcess()->GetHandleTable().Add(out, code_mem));
+
+    code_mem->Close();
+
+    return ResultSuccess;
+}
+
+static ResultCode ControlCodeMemory(Core::System& system, Handle code_memory_handle, u32 operation,
+                                    VAddr address, size_t size, Svc::MemoryPermission perm) {
+
+    LOG_TRACE(Kernel_SVC,
+              "called, code_memory_handle=0x{:X}, operation=0x{:X}, address=0x{:X}, size=0x{:X}, "
+              "permission=0x{:X}",
+              code_memory_handle, operation, address, size, perm);
+
+    // Validate the address / size.
+    R_UNLESS(Common::IsAligned(address, PageSize), ResultInvalidAddress);
+    R_UNLESS(Common::IsAligned(size, PageSize), ResultInvalidSize);
+    R_UNLESS(size > 0, ResultInvalidSize);
+    R_UNLESS((address < address + size), ResultInvalidCurrentMemory);
+
+    // Get the code memory from its handle.
+    KScopedAutoObject code_mem =
+        system.CurrentProcess()->GetHandleTable().GetObject<KCodeMemory>(code_memory_handle);
+    R_UNLESS(code_mem.IsNotNull(), ResultInvalidHandle);
+
+    // NOTE: Here, Atmosphere extends the SVC to allow code memory operations on one's own process.
+    // This enables homebrew usage of these SVCs for JIT.
+
+    // Perform the operation.
+    switch (static_cast<CodeMemoryOperation>(operation)) {
+    case CodeMemoryOperation::Map: {
+        // Check that the region is in range.
+        R_UNLESS(
+            system.CurrentProcess()->PageTable().CanContain(address, size, KMemoryState::CodeOut),
+            ResultInvalidMemoryRegion);
+
+        // Check the memory permission.
+        R_UNLESS(IsValidMapCodeMemoryPermission(perm), ResultInvalidNewMemoryPermission);
+
+        // Map the memory.
+        R_TRY(code_mem->Map(address, size));
+    } break;
+    case CodeMemoryOperation::Unmap: {
+        // Check that the region is in range.
+        R_UNLESS(
+            system.CurrentProcess()->PageTable().CanContain(address, size, KMemoryState::CodeOut),
+            ResultInvalidMemoryRegion);
+
+        // Check the memory permission.
+        R_UNLESS(IsValidUnmapCodeMemoryPermission(perm), ResultInvalidNewMemoryPermission);
+
+        // Unmap the memory.
+        R_TRY(code_mem->Unmap(address, size));
+    } break;
+    case CodeMemoryOperation::MapToOwner: {
+        // Check that the region is in range.
+        R_UNLESS(code_mem->GetOwner()->PageTable().CanContain(address, size,
+                                                              KMemoryState::GeneratedCode),
+                 ResultInvalidMemoryRegion);
+
+        // Check the memory permission.
+        R_UNLESS(IsValidMapToOwnerCodeMemoryPermission(perm), ResultInvalidNewMemoryPermission);
+
+        // Map the memory to its owner.
+        R_TRY(code_mem->MapToOwner(address, size, perm));
+    } break;
+    case CodeMemoryOperation::UnmapFromOwner: {
+        // Check that the region is in range.
+        R_UNLESS(code_mem->GetOwner()->PageTable().CanContain(address, size,
+                                                              KMemoryState::GeneratedCode),
+                 ResultInvalidMemoryRegion);
+
+        // Check the memory permission.
+        R_UNLESS(IsValidUnmapFromOwnerCodeMemoryPermission(perm), ResultInvalidNewMemoryPermission);
+
+        // Unmap the memory from its owner.
+        R_TRY(code_mem->UnmapFromOwner(address, size));
+    } break;
+    default:
+        return ResultInvalidEnumValue;
+    }
+
+    return ResultSuccess;
+}
+
 static ResultCode QueryProcessMemory(Core::System& system, VAddr memory_info_address,
                                      VAddr page_info_address, Handle process_handle,
                                      VAddr address) {
@@ -2600,8 +2811,8 @@ static const FunctionDef SVC_Table_64[] = {
     {0x48, nullptr, "MapPhysicalMemoryUnsafe"},
     {0x49, nullptr, "UnmapPhysicalMemoryUnsafe"},
     {0x4A, nullptr, "SetUnsafeLimit"},
-    {0x4B, nullptr, "CreateCodeMemory"},
-    {0x4C, nullptr, "ControlCodeMemory"},
+    {0x4B, SvcWrap64<CreateCodeMemory>, "CreateCodeMemory"},
+    {0x4C, SvcWrap64<ControlCodeMemory>, "ControlCodeMemory"},
     {0x4D, nullptr, "SleepSystem"},
     {0x4E, nullptr, "ReadWriteRegister"},
     {0x4F, nullptr, "SetProcessActivity"},
@@ -2641,8 +2852,8 @@ static const FunctionDef SVC_Table_64[] = {
     {0x71, nullptr, "ManageNamedPort"},
     {0x72, nullptr, "ConnectToPort"},
     {0x73, SvcWrap64<SetProcessMemoryPermission>, "SetProcessMemoryPermission"},
-    {0x74, nullptr, "MapProcessMemory"},
-    {0x75, nullptr, "UnmapProcessMemory"},
+    {0x74, SvcWrap64<MapProcessMemory>, "MapProcessMemory"},
+    {0x75, SvcWrap64<UnmapProcessMemory>, "UnmapProcessMemory"},
     {0x76, SvcWrap64<QueryProcessMemory>, "QueryProcessMemory"},
     {0x77, SvcWrap64<MapProcessCodeMemory>, "MapProcessCodeMemory"},
     {0x78, SvcWrap64<UnmapProcessCodeMemory>, "UnmapProcessCodeMemory"},
diff --git a/src/core/hle/kernel/svc_wrap.h b/src/core/hle/kernel/svc_wrap.h
index 6e62e656f..86255fe6d 100644
--- a/src/core/hle/kernel/svc_wrap.h
+++ b/src/core/hle/kernel/svc_wrap.h
@@ -73,6 +73,23 @@ void SvcWrap64(Core::System& system) {
                            .raw);
 }
 
+// Used by MapProcessMemory and UnmapProcessMemory
+template <ResultCode func(Core::System&, u64, u32, u64, u64)>
+void SvcWrap64(Core::System& system) {
+    FuncReturn(system, func(system, Param(system, 0), static_cast<u32>(Param(system, 1)),
+                            Param(system, 2), Param(system, 3))
+                           .raw);
+}
+
+// Used by ControlCodeMemory
+template <ResultCode func(Core::System&, Handle, u32, u64, u64, Svc::MemoryPermission)>
+void SvcWrap64(Core::System& system) {
+    FuncReturn(system, func(system, static_cast<Handle>(Param(system, 0)),
+                            static_cast<u32>(Param(system, 1)), Param(system, 2), Param(system, 3),
+                            static_cast<Svc::MemoryPermission>(Param(system, 4)))
+                           .raw);
+}
+
 template <ResultCode func(Core::System&, u32*)>
 void SvcWrap64(Core::System& system) {
     u32 param = 0;
@@ -301,6 +318,16 @@ void SvcWrap64(Core::System& system) {
     FuncReturn(system, retval);
 }
 
+// Used by CreateCodeMemory
+template <ResultCode func(Core::System&, Handle*, u64, u64)>
+void SvcWrap64(Core::System& system) {
+    u32 param_1 = 0;
+    const u32 retval = func(system, &param_1, Param(system, 1), Param(system, 2)).raw;
+
+    system.CurrentArmInterface().SetReg(1, param_1);
+    FuncReturn(system, retval);
+}
+
 template <ResultCode func(Core::System&, Handle*, u64, u32, u32)>
 void SvcWrap64(Core::System& system) {
     u32 param_1 = 0;
diff --git a/src/core/hle/kernel/time_manager.cpp b/src/core/hle/kernel/time_manager.cpp
index 8cd7279a3..aa985d820 100644
--- a/src/core/hle/kernel/time_manager.cpp
+++ b/src/core/hle/kernel/time_manager.cpp
@@ -5,6 +5,7 @@
 #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"
 
@@ -15,7 +16,10 @@ TimeManager::TimeManager(Core::System& system_) : system{system_} {
         Core::Timing::CreateEvent("Kernel::TimeManagerCallback",
                                   [this](std::uintptr_t thread_handle, std::chrono::nanoseconds) {
                                       KThread* thread = reinterpret_cast<KThread*>(thread_handle);
-                                      thread->Wakeup();
+                                      {
+                                          KScopedSchedulerLock sl(system.Kernel());
+                                          thread->OnTimer();
+                                      }
                                   });
 }
 
diff --git a/src/core/hle/service/glue/glue.cpp b/src/core/hle/service/glue/glue.cpp
index a08dc9758..b24d469cf 100644
--- a/src/core/hle/service/glue/glue.cpp
+++ b/src/core/hle/service/glue/glue.cpp
@@ -8,6 +8,7 @@
 #include "core/hle/service/glue/bgtc.h"
 #include "core/hle/service/glue/ectx.h"
 #include "core/hle/service/glue/glue.h"
+#include "core/hle/service/glue/notif.h"
 
 namespace Service::Glue {
 
@@ -24,6 +25,9 @@ void InstallInterfaces(Core::System& system) {
 
     // Error Context
     std::make_shared<ECTX_AW>(system)->InstallAsService(system.ServiceManager());
+
+    // Notification Services for application
+    std::make_shared<NOTIF_A>(system)->InstallAsService(system.ServiceManager());
 }
 
 } // namespace Service::Glue
diff --git a/src/core/hle/service/glue/notif.cpp b/src/core/hle/service/glue/notif.cpp
new file mode 100644
index 000000000..c559ec9df
--- /dev/null
+++ b/src/core/hle/service/glue/notif.cpp
@@ -0,0 +1,44 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/ipc_helpers.h"
+#include "core/hle/service/glue/notif.h"
+
+namespace Service::Glue {
+
+NOTIF_A::NOTIF_A(Core::System& system_) : ServiceFramework{system_, "notif:a"} {
+    // clang-format off
+    static const FunctionInfo functions[] = {
+        {500, nullptr, "RegisterAlarmSetting"},
+        {510, nullptr, "UpdateAlarmSetting"},
+        {520, &NOTIF_A::ListAlarmSettings, "ListAlarmSettings"},
+        {530, nullptr, "LoadApplicationParameter"},
+        {540, nullptr, "DeleteAlarmSetting"},
+        {1000, &NOTIF_A::Initialize, "Initialize"},
+    };
+    // clang-format on
+
+    RegisterHandlers(functions);
+}
+
+NOTIF_A::~NOTIF_A() = default;
+
+void NOTIF_A::ListAlarmSettings(Kernel::HLERequestContext& ctx) {
+    // Returns an array of AlarmSetting
+    constexpr s32 alarm_count = 0;
+
+    LOG_WARNING(Service_NOTIF, "(STUBBED) called");
+
+    IPC::ResponseBuilder rb{ctx, 3};
+    rb.Push(ResultSuccess);
+    rb.Push(alarm_count);
+}
+
+void NOTIF_A::Initialize(Kernel::HLERequestContext& ctx) {
+    LOG_WARNING(Service_NOTIF, "(STUBBED) called");
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(ResultSuccess);
+}
+
+} // namespace Service::Glue
diff --git a/src/core/hle/service/glue/notif.h b/src/core/hle/service/glue/notif.h
new file mode 100644
index 000000000..6ecf2015c
--- /dev/null
+++ b/src/core/hle/service/glue/notif.h
@@ -0,0 +1,25 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/service.h"
+
+namespace Core {
+class System;
+}
+
+namespace Service::Glue {
+
+class NOTIF_A final : public ServiceFramework<NOTIF_A> {
+public:
+    explicit NOTIF_A(Core::System& system_);
+    ~NOTIF_A() override;
+
+private:
+    void ListAlarmSettings(Kernel::HLERequestContext& ctx);
+    void Initialize(Kernel::HLERequestContext& ctx);
+};
+
+} // namespace Service::Glue
diff --git a/src/core/hle/service/hid/controllers/npad.cpp b/src/core/hle/service/hid/controllers/npad.cpp
index ae56f10cf..2705e9dcb 100644
--- a/src/core/hle/service/hid/controllers/npad.cpp
+++ b/src/core/hle/service/hid/controllers/npad.cpp
@@ -110,7 +110,7 @@ void Controller_NPad::ControllerUpdate(Core::HID::ControllerTriggerType type,
         UpdateControllerAt(npad_type, npad_id, is_connected);
         break;
     case Core::HID::ControllerTriggerType::Battery: {
-        if (!controller.is_connected) {
+        if (!controller.device->IsConnected()) {
             return;
         }
         auto& shared_memory = controller.shared_memory_entry;
@@ -150,7 +150,6 @@ void Controller_NPad::InitNewlyAddedController(Core::HID::NpadIdType npad_id) {
         shared_memory.system_properties.is_vertical.Assign(1);
         shared_memory.system_properties.use_plus.Assign(1);
         shared_memory.system_properties.use_minus.Assign(1);
-        shared_memory.assignment_mode = NpadJoyAssignmentMode::Single;
         shared_memory.applet_footer.type = AppletFooterUiType::SwitchProController;
         break;
     case Core::HID::NpadStyleIndex::Handheld:
@@ -166,21 +165,30 @@ void Controller_NPad::InitNewlyAddedController(Core::HID::NpadIdType npad_id) {
         break;
     case Core::HID::NpadStyleIndex::JoyconDual:
         shared_memory.style_tag.joycon_dual.Assign(1);
-        shared_memory.device_type.joycon_left.Assign(1);
-        shared_memory.device_type.joycon_right.Assign(1);
-        shared_memory.system_properties.is_vertical.Assign(1);
-        shared_memory.system_properties.use_plus.Assign(1);
-        shared_memory.system_properties.use_minus.Assign(1);
+        if (controller.is_dual_left_connected) {
+            shared_memory.device_type.joycon_left.Assign(1);
+            shared_memory.system_properties.use_minus.Assign(1);
+        }
+        if (controller.is_dual_right_connected) {
+            shared_memory.device_type.joycon_right.Assign(1);
+            shared_memory.system_properties.use_plus.Assign(1);
+        }
         shared_memory.system_properties.use_directional_buttons.Assign(1);
+        shared_memory.system_properties.is_vertical.Assign(1);
         shared_memory.assignment_mode = NpadJoyAssignmentMode::Dual;
-        shared_memory.applet_footer.type = AppletFooterUiType::JoyDual;
+        if (controller.is_dual_left_connected && controller.is_dual_right_connected) {
+            shared_memory.applet_footer.type = AppletFooterUiType::JoyDual;
+        } else if (controller.is_dual_left_connected) {
+            shared_memory.applet_footer.type = AppletFooterUiType::JoyDualLeftOnly;
+        } else {
+            shared_memory.applet_footer.type = AppletFooterUiType::JoyDualRightOnly;
+        }
         break;
     case Core::HID::NpadStyleIndex::JoyconLeft:
         shared_memory.style_tag.joycon_left.Assign(1);
         shared_memory.device_type.joycon_left.Assign(1);
         shared_memory.system_properties.is_horizontal.Assign(1);
         shared_memory.system_properties.use_minus.Assign(1);
-        shared_memory.assignment_mode = NpadJoyAssignmentMode::Single;
         shared_memory.applet_footer.type = AppletFooterUiType::JoyLeftHorizontal;
         break;
     case Core::HID::NpadStyleIndex::JoyconRight:
@@ -188,7 +196,6 @@ void Controller_NPad::InitNewlyAddedController(Core::HID::NpadIdType npad_id) {
         shared_memory.device_type.joycon_right.Assign(1);
         shared_memory.system_properties.is_horizontal.Assign(1);
         shared_memory.system_properties.use_plus.Assign(1);
-        shared_memory.assignment_mode = NpadJoyAssignmentMode::Single;
         shared_memory.applet_footer.type = AppletFooterUiType::JoyRightHorizontal;
         break;
     case Core::HID::NpadStyleIndex::GameCube:
@@ -200,7 +207,6 @@ void Controller_NPad::InitNewlyAddedController(Core::HID::NpadIdType npad_id) {
     case Core::HID::NpadStyleIndex::Pokeball:
         shared_memory.style_tag.palma.Assign(1);
         shared_memory.device_type.palma.Assign(1);
-        shared_memory.assignment_mode = NpadJoyAssignmentMode::Single;
         break;
     case Core::HID::NpadStyleIndex::NES:
         shared_memory.style_tag.lark.Assign(1);
@@ -443,11 +449,15 @@ void Controller_NPad::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8*
         case Core::HID::NpadStyleIndex::JoyconDual:
             pad_state.connection_status.raw = 0;
             pad_state.connection_status.is_connected.Assign(1);
-            pad_state.connection_status.is_left_connected.Assign(1);
-            pad_state.connection_status.is_right_connected.Assign(1);
+            if (controller.is_dual_left_connected) {
+                pad_state.connection_status.is_left_connected.Assign(1);
+                libnx_state.connection_status.is_left_connected.Assign(1);
+            }
+            if (controller.is_dual_right_connected) {
+                pad_state.connection_status.is_right_connected.Assign(1);
+                libnx_state.connection_status.is_right_connected.Assign(1);
+            }
 
-            libnx_state.connection_status.is_left_connected.Assign(1);
-            libnx_state.connection_status.is_right_connected.Assign(1);
             pad_state.sampling_number =
                 npad.joy_dual_lifo.ReadCurrentEntry().state.sampling_number + 1;
             npad.joy_dual_lifo.WriteNextEntry(pad_state);
@@ -687,7 +697,7 @@ Controller_NPad::NpadCommunicationMode Controller_NPad::GetNpadCommunicationMode
     return communication_mode;
 }
 
-void Controller_NPad::SetNpadMode(Core::HID::NpadIdType npad_id,
+void Controller_NPad::SetNpadMode(Core::HID::NpadIdType npad_id, NpadJoyDeviceType npad_device_type,
                                   NpadJoyAssignmentMode assignment_mode) {
     if (!IsNpadIdValid(npad_id)) {
         LOG_ERROR(Service_HID, "Invalid NpadIdType npad_id:{}", npad_id);
@@ -698,6 +708,62 @@ void Controller_NPad::SetNpadMode(Core::HID::NpadIdType npad_id,
     if (controller.shared_memory_entry.assignment_mode != assignment_mode) {
         controller.shared_memory_entry.assignment_mode = assignment_mode;
     }
+
+    if (!controller.device->IsConnected()) {
+        return;
+    }
+
+    if (assignment_mode == NpadJoyAssignmentMode::Dual) {
+        if (controller.device->GetNpadStyleIndex() == Core::HID::NpadStyleIndex::JoyconLeft) {
+            DisconnectNpad(npad_id);
+            controller.is_dual_left_connected = true;
+            controller.is_dual_right_connected = false;
+            UpdateControllerAt(Core::HID::NpadStyleIndex::JoyconDual, npad_id, true);
+            return;
+        }
+        if (controller.device->GetNpadStyleIndex() == Core::HID::NpadStyleIndex::JoyconRight) {
+            DisconnectNpad(npad_id);
+            controller.is_dual_left_connected = false;
+            controller.is_dual_right_connected = true;
+            UpdateControllerAt(Core::HID::NpadStyleIndex::JoyconDual, npad_id, true);
+            return;
+        }
+        return;
+    }
+
+    // This is for NpadJoyAssignmentMode::Single
+
+    // Only JoyconDual get affected by this function
+    if (controller.device->GetNpadStyleIndex() != Core::HID::NpadStyleIndex::JoyconDual) {
+        return;
+    }
+
+    if (controller.is_dual_left_connected && !controller.is_dual_right_connected) {
+        DisconnectNpad(npad_id);
+        UpdateControllerAt(Core::HID::NpadStyleIndex::JoyconLeft, npad_id, true);
+        return;
+    }
+    if (!controller.is_dual_left_connected && controller.is_dual_right_connected) {
+        DisconnectNpad(npad_id);
+        UpdateControllerAt(Core::HID::NpadStyleIndex::JoyconRight, npad_id, true);
+        return;
+    }
+
+    // We have two controllers connected to the same npad_id we need to split them
+    const auto npad_id_2 = hid_core.GetFirstDisconnectedNpadId();
+    auto& controller_2 = GetControllerFromNpadIdType(npad_id_2);
+    DisconnectNpad(npad_id);
+    if (npad_device_type == NpadJoyDeviceType::Left) {
+        UpdateControllerAt(Core::HID::NpadStyleIndex::JoyconLeft, npad_id, true);
+        controller_2.is_dual_left_connected = false;
+        controller_2.is_dual_right_connected = true;
+        UpdateControllerAt(Core::HID::NpadStyleIndex::JoyconDual, npad_id_2, true);
+    } else {
+        UpdateControllerAt(Core::HID::NpadStyleIndex::JoyconRight, npad_id, true);
+        controller_2.is_dual_left_connected = true;
+        controller_2.is_dual_right_connected = false;
+        UpdateControllerAt(Core::HID::NpadStyleIndex::JoyconDual, npad_id_2, true);
+    }
 }
 
 bool Controller_NPad::VibrateControllerAtIndex(Core::HID::NpadIdType npad_id,
@@ -907,6 +973,7 @@ void Controller_NPad::DisconnectNpad(Core::HID::NpadIdType npad_id) {
     }
 
     auto& shared_memory_entry = controller.shared_memory_entry;
+    // Don't reset shared_memory_entry.assignment_mode this value is persistent
     shared_memory_entry.style_tag.raw = Core::HID::NpadStyleSet::None; // Zero out
     shared_memory_entry.device_type.raw = 0;
     shared_memory_entry.system_properties.raw = 0;
@@ -923,9 +990,10 @@ void Controller_NPad::DisconnectNpad(Core::HID::NpadIdType npad_id) {
         .left = {},
         .right = {},
     };
-    shared_memory_entry.assignment_mode = NpadJoyAssignmentMode::Dual;
     shared_memory_entry.applet_footer.type = AppletFooterUiType::None;
 
+    controller.is_dual_left_connected = true;
+    controller.is_dual_right_connected = true;
     controller.is_connected = false;
     controller.device->Disconnect();
     SignalStyleSetChangedEvent(npad_id);
@@ -1022,19 +1090,70 @@ void Controller_NPad::MergeSingleJoyAsDualJoy(Core::HID::NpadIdType npad_id_1,
                   npad_id_2);
         return;
     }
-    auto& controller_1 = GetControllerFromNpadIdType(npad_id_1).device;
-    auto& controller_2 = GetControllerFromNpadIdType(npad_id_2).device;
+    auto& controller_1 = GetControllerFromNpadIdType(npad_id_1);
+    auto& controller_2 = GetControllerFromNpadIdType(npad_id_2);
+    const auto controller_style_1 = controller_1.device->GetNpadStyleIndex();
+    const auto controller_style_2 = controller_2.device->GetNpadStyleIndex();
+    bool merge_controllers = false;
 
     // If the controllers at both npad indices form a pair of left and right joycons, merge them.
     // Otherwise, do nothing.
-    if ((controller_1->GetNpadStyleIndex() == Core::HID::NpadStyleIndex::JoyconLeft &&
-         controller_2->GetNpadStyleIndex() == Core::HID::NpadStyleIndex::JoyconRight) ||
-        (controller_2->GetNpadStyleIndex() == Core::HID::NpadStyleIndex::JoyconLeft &&
-         controller_1->GetNpadStyleIndex() == Core::HID::NpadStyleIndex::JoyconRight)) {
+    if (controller_style_1 == Core::HID::NpadStyleIndex::JoyconLeft &&
+        controller_style_2 == Core::HID::NpadStyleIndex::JoyconRight) {
+        merge_controllers = true;
+    }
+    if (controller_style_2 == Core::HID::NpadStyleIndex::JoyconLeft &&
+        controller_style_1 == Core::HID::NpadStyleIndex::JoyconRight) {
+        merge_controllers = true;
+    }
+    if (controller_style_1 == Core::HID::NpadStyleIndex::JoyconDual &&
+        controller_style_2 == Core::HID::NpadStyleIndex::JoyconRight &&
+        controller_1.is_dual_left_connected && !controller_1.is_dual_right_connected) {
+        merge_controllers = true;
+    }
+    if (controller_style_1 == Core::HID::NpadStyleIndex::JoyconDual &&
+        controller_style_2 == Core::HID::NpadStyleIndex::JoyconLeft &&
+        !controller_1.is_dual_left_connected && controller_1.is_dual_right_connected) {
+        merge_controllers = true;
+    }
+    if (controller_style_2 == Core::HID::NpadStyleIndex::JoyconDual &&
+        controller_style_1 == Core::HID::NpadStyleIndex::JoyconRight &&
+        controller_2.is_dual_left_connected && !controller_2.is_dual_right_connected) {
+        merge_controllers = true;
+    }
+    if (controller_style_2 == Core::HID::NpadStyleIndex::JoyconDual &&
+        controller_style_1 == Core::HID::NpadStyleIndex::JoyconLeft &&
+        !controller_2.is_dual_left_connected && controller_2.is_dual_right_connected) {
+        merge_controllers = true;
+    }
+    if (controller_style_1 == Core::HID::NpadStyleIndex::JoyconDual &&
+        controller_style_2 == Core::HID::NpadStyleIndex::JoyconDual &&
+        controller_1.is_dual_left_connected && !controller_1.is_dual_right_connected &&
+        !controller_2.is_dual_left_connected && controller_2.is_dual_right_connected) {
+        merge_controllers = true;
+    }
+    if (controller_style_1 == Core::HID::NpadStyleIndex::JoyconDual &&
+        controller_style_2 == Core::HID::NpadStyleIndex::JoyconDual &&
+        !controller_1.is_dual_left_connected && controller_1.is_dual_right_connected &&
+        controller_2.is_dual_left_connected && !controller_2.is_dual_right_connected) {
+        merge_controllers = true;
+    }
+
+    if (merge_controllers) {
         // Disconnect the joycon at the second id and connect the dual joycon at the first index.
         DisconnectNpad(npad_id_2);
+        controller_1.is_dual_left_connected = true;
+        controller_1.is_dual_right_connected = true;
         AddNewControllerAt(Core::HID::NpadStyleIndex::JoyconDual, npad_id_1);
+        return;
     }
+    LOG_WARNING(Service_HID,
+                "Controllers can't be merged npad_id_1:{}, npad_id_2:{}, type_1:{}, type_2:{}, "
+                "dual_1(left/right):{}/{}, dual_2(left/right):{}/{}",
+                npad_id_1, npad_id_2, controller_1.device->GetNpadStyleIndex(),
+                controller_2.device->GetNpadStyleIndex(), controller_1.is_dual_left_connected,
+                controller_1.is_dual_right_connected, controller_2.is_dual_left_connected,
+                controller_2.is_dual_right_connected);
 }
 
 void Controller_NPad::StartLRAssignmentMode() {
diff --git a/src/core/hle/service/hid/controllers/npad.h b/src/core/hle/service/hid/controllers/npad.h
index de5fa5a64..63281cb35 100644
--- a/src/core/hle/service/hid/controllers/npad.h
+++ b/src/core/hle/service/hid/controllers/npad.h
@@ -113,7 +113,8 @@ public:
     void SetNpadCommunicationMode(NpadCommunicationMode communication_mode_);
     NpadCommunicationMode GetNpadCommunicationMode() const;
 
-    void SetNpadMode(Core::HID::NpadIdType npad_id, NpadJoyAssignmentMode assignment_mode);
+    void SetNpadMode(Core::HID::NpadIdType npad_id, NpadJoyDeviceType npad_device_type,
+                     NpadJoyAssignmentMode assignment_mode);
 
     bool VibrateControllerAtIndex(Core::HID::NpadIdType npad_id, std::size_t device_index,
                                   const Core::HID::VibrationValue& vibration_value);
@@ -464,7 +465,10 @@ private:
         std::array<VibrationData, 2> vibration{};
         bool unintended_home_button_input_protection{};
         bool is_connected{};
-        Core::HID::NpadStyleIndex npad_type{Core::HID::NpadStyleIndex::None};
+
+        // Dual joycons can have only one side connected
+        bool is_dual_left_connected{true};
+        bool is_dual_right_connected{true};
 
         // Motion parameters
         bool sixaxis_at_rest{true};
diff --git a/src/core/hle/service/hid/hid.cpp b/src/core/hle/service/hid/hid.cpp
index b36689552..7163e1a4e 100644
--- a/src/core/hle/service/hid/hid.cpp
+++ b/src/core/hle/service/hid/hid.cpp
@@ -293,8 +293,8 @@ Hid::Hid(Core::System& system_)
         {132, &Hid::EnableUnintendedHomeButtonInputProtection, "EnableUnintendedHomeButtonInputProtection"},
         {133, nullptr, "SetNpadJoyAssignmentModeSingleWithDestination"},
         {134, &Hid::SetNpadAnalogStickUseCenterClamp, "SetNpadAnalogStickUseCenterClamp"},
-        {135, nullptr, "SetNpadCaptureButtonAssignment"},
-        {136, nullptr, "ClearNpadCaptureButtonAssignment"},
+        {135, &Hid::SetNpadCaptureButtonAssignment, "SetNpadCaptureButtonAssignment"},
+        {136, &Hid::ClearNpadCaptureButtonAssignment, "ClearNpadCaptureButtonAssignment"},
         {200, &Hid::GetVibrationDeviceInfo, "GetVibrationDeviceInfo"},
         {201, &Hid::SendVibrationValue, "SendVibrationValue"},
         {202, &Hid::GetActualVibrationValue, "GetActualVibrationValue"},
@@ -975,35 +975,35 @@ void Hid::SetNpadJoyAssignmentModeSingleByDefault(Kernel::HLERequestContext& ctx
     const auto parameters{rp.PopRaw<Parameters>()};
 
     applet_resource->GetController<Controller_NPad>(HidController::NPad)
-        .SetNpadMode(parameters.npad_id, Controller_NPad::NpadJoyAssignmentMode::Single);
+        .SetNpadMode(parameters.npad_id, Controller_NPad::NpadJoyDeviceType::Left,
+                     Controller_NPad::NpadJoyAssignmentMode::Single);
 
-    LOG_WARNING(Service_HID, "(STUBBED) called, npad_id={}, applet_resource_user_id={}",
-                parameters.npad_id, parameters.applet_resource_user_id);
+    LOG_INFO(Service_HID, "called, npad_id={}, applet_resource_user_id={}", parameters.npad_id,
+             parameters.applet_resource_user_id);
 
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(ResultSuccess);
 }
 
 void Hid::SetNpadJoyAssignmentModeSingle(Kernel::HLERequestContext& ctx) {
-    // TODO: Check the differences between this and SetNpadJoyAssignmentModeSingleByDefault
     IPC::RequestParser rp{ctx};
     struct Parameters {
         Core::HID::NpadIdType npad_id;
         INSERT_PADDING_WORDS_NOINIT(1);
         u64 applet_resource_user_id;
-        u64 npad_joy_device_type;
+        Controller_NPad::NpadJoyDeviceType npad_joy_device_type;
     };
     static_assert(sizeof(Parameters) == 0x18, "Parameters has incorrect size.");
 
     const auto parameters{rp.PopRaw<Parameters>()};
 
     applet_resource->GetController<Controller_NPad>(HidController::NPad)
-        .SetNpadMode(parameters.npad_id, Controller_NPad::NpadJoyAssignmentMode::Single);
+        .SetNpadMode(parameters.npad_id, parameters.npad_joy_device_type,
+                     Controller_NPad::NpadJoyAssignmentMode::Single);
 
-    LOG_WARNING(Service_HID,
-                "(STUBBED) called, npad_id={}, applet_resource_user_id={}, npad_joy_device_type={}",
-                parameters.npad_id, parameters.applet_resource_user_id,
-                parameters.npad_joy_device_type);
+    LOG_INFO(Service_HID, "called, npad_id={}, applet_resource_user_id={}, npad_joy_device_type={}",
+             parameters.npad_id, parameters.applet_resource_user_id,
+             parameters.npad_joy_device_type);
 
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(ResultSuccess);
@@ -1021,10 +1021,10 @@ void Hid::SetNpadJoyAssignmentModeDual(Kernel::HLERequestContext& ctx) {
     const auto parameters{rp.PopRaw<Parameters>()};
 
     applet_resource->GetController<Controller_NPad>(HidController::NPad)
-        .SetNpadMode(parameters.npad_id, Controller_NPad::NpadJoyAssignmentMode::Dual);
+        .SetNpadMode(parameters.npad_id, {}, Controller_NPad::NpadJoyAssignmentMode::Dual);
 
-    LOG_WARNING(Service_HID, "(STUBBED) called, npad_id={}, applet_resource_user_id={}",
-                parameters.npad_id, parameters.applet_resource_user_id);
+    LOG_INFO(Service_HID, "called, npad_id={}, applet_resource_user_id={}", parameters.npad_id,
+             parameters.applet_resource_user_id);
 
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(ResultSuccess);
@@ -1186,6 +1186,37 @@ void Hid::SetNpadAnalogStickUseCenterClamp(Kernel::HLERequestContext& ctx) {
     rb.Push(ResultSuccess);
 }
 
+void Hid::SetNpadCaptureButtonAssignment(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    struct Parameters {
+        Core::HID::NpadStyleSet npad_styleset;
+        INSERT_PADDING_WORDS_NOINIT(1);
+        u64 applet_resource_user_id;
+        Core::HID::NpadButton button;
+    };
+    static_assert(sizeof(Parameters) == 0x18, "Parameters has incorrect size.");
+
+    const auto parameters{rp.PopRaw<Parameters>()};
+
+    LOG_WARNING(Service_HID,
+                "(STUBBED) called, npad_styleset={}, applet_resource_user_id={}, button={}",
+                parameters.npad_styleset, parameters.applet_resource_user_id, parameters.button);
+
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(ResultSuccess);
+}
+
+void Hid::ClearNpadCaptureButtonAssignment(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    const auto applet_resource_user_id{rp.Pop<u64>()};
+
+    LOG_WARNING(Service_HID, "(STUBBED) called, applet_resource_user_id={}",
+                applet_resource_user_id);
+
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(ResultSuccess);
+}
+
 void Hid::GetVibrationDeviceInfo(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     const auto vibration_device_handle{rp.PopRaw<Core::HID::VibrationDeviceHandle>()};
diff --git a/src/core/hle/service/hid/hid.h b/src/core/hle/service/hid/hid.h
index ab0084118..d290df161 100644
--- a/src/core/hle/service/hid/hid.h
+++ b/src/core/hle/service/hid/hid.h
@@ -136,6 +136,8 @@ private:
     void IsUnintendedHomeButtonInputProtectionEnabled(Kernel::HLERequestContext& ctx);
     void EnableUnintendedHomeButtonInputProtection(Kernel::HLERequestContext& ctx);
     void SetNpadAnalogStickUseCenterClamp(Kernel::HLERequestContext& ctx);
+    void SetNpadCaptureButtonAssignment(Kernel::HLERequestContext& ctx);
+    void ClearNpadCaptureButtonAssignment(Kernel::HLERequestContext& ctx);
     void GetVibrationDeviceInfo(Kernel::HLERequestContext& ctx);
     void SendVibrationValue(Kernel::HLERequestContext& ctx);
     void GetActualVibrationValue(Kernel::HLERequestContext& ctx);
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp
index 0d7d4ad03..8314d1ec2 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp
@@ -20,8 +20,12 @@ NvResult nvhost_nvdec::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>&
     switch (command.group) {
     case 0x0:
         switch (command.cmd) {
-        case 0x1:
-            return Submit(input, output);
+        case 0x1: {
+            if (!fd_to_id.contains(fd)) {
+                fd_to_id[fd] = next_id++;
+            }
+            return Submit(fd, input, output);
+        }
         case 0x2:
             return GetSyncpoint(input, output);
         case 0x3:
@@ -66,7 +70,10 @@ void nvhost_nvdec::OnOpen(DeviceFD fd) {}
 
 void nvhost_nvdec::OnClose(DeviceFD fd) {
     LOG_INFO(Service_NVDRV, "NVDEC video stream ended");
-    system.GPU().ClearCdmaInstance();
+    const auto iter = fd_to_id.find(fd);
+    if (iter != fd_to_id.end()) {
+        system.GPU().ClearCdmaInstance(iter->second);
+    }
 }
 
 } // namespace Service::Nvidia::Devices
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h
index 523d96e3a..a507c4d0a 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h
+++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h
@@ -24,6 +24,9 @@ public:
 
     void OnOpen(DeviceFD fd) override;
     void OnClose(DeviceFD fd) override;
+
+private:
+    u32 next_id{};
 };
 
 } // namespace Service::Nvidia::Devices
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.cpp b/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.cpp
index e61261f98..8a05f0668 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.cpp
@@ -59,7 +59,8 @@ NvResult nvhost_nvdec_common::SetNVMAPfd(const std::vector<u8>& input) {
     return NvResult::Success;
 }
 
-NvResult nvhost_nvdec_common::Submit(const std::vector<u8>& input, std::vector<u8>& output) {
+NvResult nvhost_nvdec_common::Submit(DeviceFD fd, const std::vector<u8>& input,
+                                     std::vector<u8>& output) {
     IoctlSubmit params{};
     std::memcpy(&params, input.data(), sizeof(IoctlSubmit));
     LOG_DEBUG(Service_NVDRV, "called NVDEC Submit, cmd_buffer_count={}", params.cmd_buffer_count);
@@ -93,7 +94,7 @@ NvResult nvhost_nvdec_common::Submit(const std::vector<u8>& input, std::vector<u
         Tegra::ChCommandHeaderList cmdlist(cmd_buffer.word_count);
         system.Memory().ReadBlock(object->addr + cmd_buffer.offset, cmdlist.data(),
                                   cmdlist.size() * sizeof(u32));
-        gpu.PushCommandBuffer(cmdlist);
+        gpu.PushCommandBuffer(fd_to_id[fd], cmdlist);
     }
     std::memcpy(output.data(), &params, sizeof(IoctlSubmit));
     // Some games expect command_buffers to be written back
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h b/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h
index 351625c17..e28c54df6 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h
+++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h
@@ -104,13 +104,14 @@ protected:
 
     /// Ioctl command implementations
     NvResult SetNVMAPfd(const std::vector<u8>& input);
-    NvResult Submit(const std::vector<u8>& input, std::vector<u8>& output);
+    NvResult Submit(DeviceFD fd, const std::vector<u8>& input, std::vector<u8>& output);
     NvResult GetSyncpoint(const std::vector<u8>& input, std::vector<u8>& output);
     NvResult GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output);
     NvResult MapBuffer(const std::vector<u8>& input, std::vector<u8>& output);
     NvResult UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& output);
     NvResult SetSubmitTimeout(const std::vector<u8>& input, std::vector<u8>& output);
 
+    std::unordered_map<DeviceFD, u32> fd_to_id{};
     s32_le nvmap_fd{};
     u32_le submit_timeout{};
     std::shared_ptr<nvmap> nvmap_dev;
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp b/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp
index eac4dd530..76b39806f 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp
@@ -21,7 +21,10 @@ NvResult nvhost_vic::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& i
     case 0x0:
         switch (command.cmd) {
         case 0x1:
-            return Submit(input, output);
+            if (!fd_to_id.contains(fd)) {
+                fd_to_id[fd] = next_id++;
+            }
+            return Submit(fd, input, output);
         case 0x2:
             return GetSyncpoint(input, output);
         case 0x3:
@@ -65,7 +68,10 @@ NvResult nvhost_vic::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& i
 void nvhost_vic::OnOpen(DeviceFD fd) {}
 
 void nvhost_vic::OnClose(DeviceFD fd) {
-    system.GPU().ClearCdmaInstance();
+    const auto iter = fd_to_id.find(fd);
+    if (iter != fd_to_id.end()) {
+        system.GPU().ClearCdmaInstance(iter->second);
+    }
 }
 
 } // namespace Service::Nvidia::Devices
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_vic.h b/src/core/hle/service/nvdrv/devices/nvhost_vic.h
index 6d7fda9d1..c9732c037 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_vic.h
+++ b/src/core/hle/service/nvdrv/devices/nvhost_vic.h
@@ -23,5 +23,8 @@ public:
 
     void OnOpen(DeviceFD fd) override;
     void OnClose(DeviceFD fd) override;
+
+private:
+    u32 next_id{};
 };
 } // namespace Service::Nvidia::Devices
diff --git a/src/shader_recompiler/CMakeLists.txt b/src/shader_recompiler/CMakeLists.txt
index bc3df80c8..4c76ce1ea 100644
--- a/src/shader_recompiler/CMakeLists.txt
+++ b/src/shader_recompiler/CMakeLists.txt
@@ -1,7 +1,5 @@
 add_library(shader_recompiler STATIC
     backend/bindings.h
-    backend/glasm/emit_context.cpp
-    backend/glasm/emit_context.h
     backend/glasm/emit_glasm.cpp
     backend/glasm/emit_glasm.h
     backend/glasm/emit_glasm_barriers.cpp
@@ -22,10 +20,10 @@ add_library(shader_recompiler STATIC
     backend/glasm/emit_glasm_special.cpp
     backend/glasm/emit_glasm_undefined.cpp
     backend/glasm/emit_glasm_warp.cpp
+    backend/glasm/glasm_emit_context.cpp
+    backend/glasm/glasm_emit_context.h
     backend/glasm/reg_alloc.cpp
     backend/glasm/reg_alloc.h
-    backend/glsl/emit_context.cpp
-    backend/glsl/emit_context.h
     backend/glsl/emit_glsl.cpp
     backend/glsl/emit_glsl.h
     backend/glsl/emit_glsl_atomic.cpp
@@ -47,10 +45,10 @@ add_library(shader_recompiler STATIC
     backend/glsl/emit_glsl_special.cpp
     backend/glsl/emit_glsl_undefined.cpp
     backend/glsl/emit_glsl_warp.cpp
+    backend/glsl/glsl_emit_context.cpp
+    backend/glsl/glsl_emit_context.h
     backend/glsl/var_alloc.cpp
     backend/glsl/var_alloc.h
-    backend/spirv/emit_context.cpp
-    backend/spirv/emit_context.h
     backend/spirv/emit_spirv.cpp
     backend/spirv/emit_spirv.h
     backend/spirv/emit_spirv_atomic.cpp
@@ -72,6 +70,8 @@ add_library(shader_recompiler STATIC
     backend/spirv/emit_spirv_special.cpp
     backend/spirv/emit_spirv_undefined.cpp
     backend/spirv/emit_spirv_warp.cpp
+    backend/spirv/spirv_emit_context.cpp
+    backend/spirv/spirv_emit_context.h
     environment.h
     exception.h
     frontend/ir/abstract_syntax_list.h
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm.cpp b/src/shader_recompiler/backend/glasm/emit_glasm.cpp
index 004658546..42eff443f 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm.cpp
@@ -9,9 +9,9 @@
 #include "common/div_ceil.h"
 #include "common/settings.h"
 #include "shader_recompiler/backend/bindings.h"
-#include "shader_recompiler/backend/glasm/emit_context.h"
 #include "shader_recompiler/backend/glasm/emit_glasm.h"
 #include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/ir_emitter.h"
 #include "shader_recompiler/frontend/ir/program.h"
 #include "shader_recompiler/profile.h"
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_barriers.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_barriers.cpp
index e69de29bb..c0b97683e 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_barriers.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_barriers.cpp
@@ -0,0 +1,22 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
+
+namespace Shader::Backend::GLASM {
+
+void EmitBarrier(EmitContext& ctx) {
+    ctx.Add("BAR;");
+}
+
+void EmitWorkgroupMemoryBarrier(EmitContext& ctx) {
+    ctx.Add("MEMBAR.CTA;");
+}
+
+void EmitDeviceMemoryBarrier(EmitContext& ctx) {
+    ctx.Add("MEMBAR;");
+}
+
+} // namespace Shader::Backend::GLASM
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_bitwise_conversion.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_bitwise_conversion.cpp
index 9201ccd39..3bfcbbe65 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_bitwise_conversion.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_bitwise_conversion.cpp
@@ -2,8 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "shader_recompiler/backend/glasm/emit_context.h"
 #include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 namespace Shader::Backend::GLASM {
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_composite.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_composite.cpp
index bff0b7c1c..babbe6654 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_composite.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_composite.cpp
@@ -2,8 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "shader_recompiler/backend/glasm/emit_context.h"
 #include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 namespace Shader::Backend::GLASM {
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_context_get_set.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_context_get_set.cpp
index 02c9dc6d7..081b2c8e0 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_context_get_set.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_context_get_set.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glasm/emit_context.h"
 #include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 #include "shader_recompiler/profile.h"
 #include "shader_recompiler/shader_info.h"
@@ -335,6 +335,35 @@ void EmitSetFragDepth(EmitContext& ctx, ScalarF32 value) {
     ctx.Add("MOV.F result.depth.z,{};", value);
 }
 
+void EmitWorkgroupId(EmitContext& ctx, IR::Inst& inst) {
+    ctx.Add("MOV.S {},invocation.groupid;", inst);
+}
+
+void EmitLocalInvocationId(EmitContext& ctx, IR::Inst& inst) {
+    ctx.Add("MOV.S {},invocation.localid;", inst);
+}
+
+void EmitInvocationId(EmitContext& ctx, IR::Inst& inst) {
+    ctx.Add("MOV.S {}.x,primitive_invocation.x;", inst);
+}
+
+void EmitSampleId(EmitContext& ctx, IR::Inst& inst) {
+    ctx.Add("MOV.S {}.x,fragment.sampleid.x;", inst);
+}
+
+void EmitIsHelperInvocation(EmitContext& ctx, IR::Inst& inst) {
+    ctx.Add("MOV.S {}.x,fragment.helperthread.x;", inst);
+}
+
+void EmitYDirection(EmitContext& ctx, IR::Inst& inst) {
+    ctx.uses_y_direction = true;
+    ctx.Add("MOV.F {}.x,y_direction[0].w;", inst);
+}
+
+void EmitResolutionDownFactor(EmitContext& ctx, IR::Inst& inst) {
+    ctx.Add("MOV.F {}.x,scaling[0].z;", inst);
+}
+
 void EmitLoadLocal(EmitContext& ctx, IR::Inst& inst, ScalarU32 word_offset) {
     ctx.Add("MOV.U {},lmem[{}].x;", inst, word_offset);
 }
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_control_flow.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_control_flow.cpp
index e69de29bb..8a14fc8d9 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_control_flow.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_control_flow.cpp
@@ -0,0 +1,18 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
+
+namespace Shader::Backend::GLASM {
+
+void EmitJoin(EmitContext&) {
+    throw NotImplementedException("Join shouldn't be emitted");
+}
+
+void EmitDemoteToHelperInvocation(EmitContext& ctx) {
+    ctx.Add("KIL TR.x;");
+}
+
+} // namespace Shader::Backend::GLASM
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_convert.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_convert.cpp
index ccdf1cbc8..4cff70fe4 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_convert.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_convert.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glasm/emit_context.h"
 #include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/modifiers.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_floating_point.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_floating_point.cpp
index 4ed58619d..356640471 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_floating_point.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_floating_point.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glasm/emit_context.h"
 #include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/modifiers.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_image.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_image.cpp
index d325d31c7..237a5af3f 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_image.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_image.cpp
@@ -4,8 +4,8 @@
 
 #include <utility>
 
-#include "shader_recompiler/backend/glasm/emit_context.h"
 #include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/modifiers.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_integer.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_integer.cpp
index 8aa494a4d..f698b8b9b 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_integer.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_integer.cpp
@@ -2,8 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "shader_recompiler/backend/glasm/emit_context.h"
 #include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 namespace Shader::Backend::GLASM {
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_logical.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_logical.cpp
index e69de29bb..eed7bfec2 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_logical.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_logical.cpp
@@ -0,0 +1,26 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
+
+namespace Shader::Backend::GLASM {
+
+void EmitLogicalOr(EmitContext& ctx, IR::Inst& inst, ScalarS32 a, ScalarS32 b) {
+    ctx.Add("OR.S {},{},{};", inst, a, b);
+}
+
+void EmitLogicalAnd(EmitContext& ctx, IR::Inst& inst, ScalarS32 a, ScalarS32 b) {
+    ctx.Add("AND.S {},{},{};", inst, a, b);
+}
+
+void EmitLogicalXor(EmitContext& ctx, IR::Inst& inst, ScalarS32 a, ScalarS32 b) {
+    ctx.Add("XOR.S {},{},{};", inst, a, b);
+}
+
+void EmitLogicalNot(EmitContext& ctx, IR::Inst& inst, ScalarS32 value) {
+    ctx.Add("SEQ.S {},{},0;", inst, value);
+}
+
+} // namespace Shader::Backend::GLASM
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_memory.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_memory.cpp
index af9fac7c1..f135b67f5 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_memory.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_memory.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glasm/emit_context.h"
 #include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/program.h"
 #include "shader_recompiler/frontend/ir/value.h"
 #include "shader_recompiler/runtime_info.h"
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_not_implemented.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_not_implemented.cpp
index 681aeda8d..86287ee3f 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_not_implemented.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_not_implemented.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glasm/emit_context.h"
 #include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/program.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
@@ -17,110 +17,6 @@ namespace Shader::Backend::GLASM {
 
 #define NotImplemented() throw NotImplementedException("GLASM instruction {}", __LINE__)
 
-static void DefinePhi(EmitContext& ctx, IR::Inst& phi) {
-    switch (phi.Type()) {
-    case IR::Type::U1:
-    case IR::Type::U32:
-    case IR::Type::F32:
-        ctx.reg_alloc.Define(phi);
-        break;
-    case IR::Type::U64:
-    case IR::Type::F64:
-        ctx.reg_alloc.LongDefine(phi);
-        break;
-    default:
-        throw NotImplementedException("Phi node type {}", phi.Type());
-    }
-}
-
-void EmitPhi(EmitContext& ctx, IR::Inst& phi) {
-    const size_t num_args{phi.NumArgs()};
-    for (size_t i = 0; i < num_args; ++i) {
-        ctx.reg_alloc.Consume(phi.Arg(i));
-    }
-    if (!phi.Definition<Id>().is_valid) {
-        // The phi node wasn't forward defined
-        DefinePhi(ctx, phi);
-    }
-}
-
-void EmitVoid(EmitContext&) {}
-
-void EmitReference(EmitContext& ctx, const IR::Value& value) {
-    ctx.reg_alloc.Consume(value);
-}
-
-void EmitPhiMove(EmitContext& ctx, const IR::Value& phi_value, const IR::Value& value) {
-    IR::Inst& phi{RegAlloc::AliasInst(*phi_value.Inst())};
-    if (!phi.Definition<Id>().is_valid) {
-        // The phi node wasn't forward defined
-        DefinePhi(ctx, phi);
-    }
-    const Register phi_reg{ctx.reg_alloc.Consume(IR::Value{&phi})};
-    const Value eval_value{ctx.reg_alloc.Consume(value)};
-
-    if (phi_reg == eval_value) {
-        return;
-    }
-    switch (phi.Flags<IR::Type>()) {
-    case IR::Type::U1:
-    case IR::Type::U32:
-    case IR::Type::F32:
-        ctx.Add("MOV.S {}.x,{};", phi_reg, ScalarS32{eval_value});
-        break;
-    case IR::Type::U64:
-    case IR::Type::F64:
-        ctx.Add("MOV.U64 {}.x,{};", phi_reg, ScalarRegister{eval_value});
-        break;
-    default:
-        throw NotImplementedException("Phi node type {}", phi.Type());
-    }
-}
-
-void EmitJoin(EmitContext& ctx) {
-    NotImplemented();
-}
-
-void EmitDemoteToHelperInvocation(EmitContext& ctx) {
-    ctx.Add("KIL TR.x;");
-}
-
-void EmitBarrier(EmitContext& ctx) {
-    ctx.Add("BAR;");
-}
-
-void EmitWorkgroupMemoryBarrier(EmitContext& ctx) {
-    ctx.Add("MEMBAR.CTA;");
-}
-
-void EmitDeviceMemoryBarrier(EmitContext& ctx) {
-    ctx.Add("MEMBAR;");
-}
-
-void EmitPrologue(EmitContext& ctx) {
-    // TODO
-}
-
-void EmitEpilogue(EmitContext& ctx) {
-    // TODO
-}
-
-void EmitEmitVertex(EmitContext& ctx, ScalarS32 stream) {
-    if (stream.type == Type::U32 && stream.imm_u32 == 0) {
-        ctx.Add("EMIT;");
-    } else {
-        ctx.Add("EMITS {};", stream);
-    }
-}
-
-void EmitEndPrimitive(EmitContext& ctx, const IR::Value& stream) {
-    if (!stream.IsImmediate()) {
-        LOG_WARNING(Shader_GLASM, "Stream is not immediate");
-    }
-    ctx.reg_alloc.Consume(stream);
-    ctx.Add("ENDPRIM;");
-}
-
 void EmitGetRegister(EmitContext& ctx) {
     NotImplemented();
 }
@@ -185,55 +81,6 @@ void EmitSetOFlag(EmitContext& ctx) {
     NotImplemented();
 }
 
-void EmitWorkgroupId(EmitContext& ctx, IR::Inst& inst) {
-    ctx.Add("MOV.S {},invocation.groupid;", inst);
-}
-
-void EmitLocalInvocationId(EmitContext& ctx, IR::Inst& inst) {
-    ctx.Add("MOV.S {},invocation.localid;", inst);
-}
-
-void EmitInvocationId(EmitContext& ctx, IR::Inst& inst) {
-    ctx.Add("MOV.S {}.x,primitive_invocation.x;", inst);
-}
-
-void EmitSampleId(EmitContext& ctx, IR::Inst& inst) {
-    ctx.Add("MOV.S {}.x,fragment.sampleid.x;", inst);
-}
-
-void EmitIsHelperInvocation(EmitContext& ctx, IR::Inst& inst) {
-    ctx.Add("MOV.S {}.x,fragment.helperthread.x;", inst);
-}
-
-void EmitYDirection(EmitContext& ctx, IR::Inst& inst) {
-    ctx.uses_y_direction = true;
-    ctx.Add("MOV.F {}.x,y_direction[0].w;", inst);
-}
-
-void EmitResolutionDownFactor(EmitContext& ctx, IR::Inst& inst) {
-    ctx.Add("MOV.F {}.x,scaling[0].z;", inst);
-}
-
-void EmitUndefU1(EmitContext& ctx, IR::Inst& inst) {
-    ctx.Add("MOV.S {}.x,0;", inst);
-}
-
-void EmitUndefU8(EmitContext& ctx, IR::Inst& inst) {
-    ctx.Add("MOV.S {}.x,0;", inst);
-}
-
-void EmitUndefU16(EmitContext& ctx, IR::Inst& inst) {
-    ctx.Add("MOV.S {}.x,0;", inst);
-}
-
-void EmitUndefU32(EmitContext& ctx, IR::Inst& inst) {
-    ctx.Add("MOV.S {}.x,0;", inst);
-}
-
-void EmitUndefU64(EmitContext& ctx, IR::Inst& inst) {
-    ctx.LongAdd("MOV.S64 {}.x,0;", inst);
-}
-
 void EmitGetZeroFromOp(EmitContext& ctx) {
     NotImplemented();
 }
@@ -258,20 +105,4 @@ void EmitGetInBoundsFromOp(EmitContext& ctx) {
     NotImplemented();
 }
 
-void EmitLogicalOr(EmitContext& ctx, IR::Inst& inst, ScalarS32 a, ScalarS32 b) {
-    ctx.Add("OR.S {},{},{};", inst, a, b);
-}
-
-void EmitLogicalAnd(EmitContext& ctx, IR::Inst& inst, ScalarS32 a, ScalarS32 b) {
-    ctx.Add("AND.S {},{},{};", inst, a, b);
-}
-
-void EmitLogicalXor(EmitContext& ctx, IR::Inst& inst, ScalarS32 a, ScalarS32 b) {
-    ctx.Add("XOR.S {},{},{};", inst, a, b);
-}
-
-void EmitLogicalNot(EmitContext& ctx, IR::Inst& inst, ScalarS32 value) {
-    ctx.Add("SEQ.S {},{},0;", inst, value);
-}
-
 } // namespace Shader::Backend::GLASM
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_select.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_select.cpp
index 68fff613c..dc441c56d 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_select.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_select.cpp
@@ -3,8 +3,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "shader_recompiler/backend/glasm/emit_context.h"
 #include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 namespace Shader::Backend::GLASM {
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_shared_memory.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_shared_memory.cpp
index c1498f449..39e1c6c3a 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_shared_memory.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_shared_memory.cpp
@@ -3,8 +3,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "shader_recompiler/backend/glasm/emit_context.h"
 #include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 namespace Shader::Backend::GLASM {
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_special.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_special.cpp
index e69de29bb..e7a5fb13a 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_special.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_special.cpp
@@ -0,0 +1,95 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
+#include "shader_recompiler/frontend/ir/value.h"
+
+namespace Shader::Backend::GLASM {
+
+static void DefinePhi(EmitContext& ctx, IR::Inst& phi) {
+    switch (phi.Type()) {
+    case IR::Type::U1:
+    case IR::Type::U32:
+    case IR::Type::F32:
+        ctx.reg_alloc.Define(phi);
+        break;
+    case IR::Type::U64:
+    case IR::Type::F64:
+        ctx.reg_alloc.LongDefine(phi);
+        break;
+    default:
+        throw NotImplementedException("Phi node type {}", phi.Type());
+    }
+}
+
+void EmitPhi(EmitContext& ctx, IR::Inst& phi) {
+    const size_t num_args{phi.NumArgs()};
+    for (size_t i = 0; i < num_args; ++i) {
+        ctx.reg_alloc.Consume(phi.Arg(i));
+    }
+    if (!phi.Definition<Id>().is_valid) {
+        // The phi node wasn't forward defined
+        DefinePhi(ctx, phi);
+    }
+}
+
+void EmitVoid(EmitContext&) {}
+
+void EmitReference(EmitContext& ctx, const IR::Value& value) {
+    ctx.reg_alloc.Consume(value);
+}
+
+void EmitPhiMove(EmitContext& ctx, const IR::Value& phi_value, const IR::Value& value) {
+    IR::Inst& phi{RegAlloc::AliasInst(*phi_value.Inst())};
+    if (!phi.Definition<Id>().is_valid) {
+        // The phi node wasn't forward defined
+        DefinePhi(ctx, phi);
+    }
+    const Register phi_reg{ctx.reg_alloc.Consume(IR::Value{&phi})};
+    const Value eval_value{ctx.reg_alloc.Consume(value)};
+
+    if (phi_reg == eval_value) {
+        return;
+    }
+    switch (phi.Flags<IR::Type>()) {
+    case IR::Type::U1:
+    case IR::Type::U32:
+    case IR::Type::F32:
+        ctx.Add("MOV.S {}.x,{};", phi_reg, ScalarS32{eval_value});
+        break;
+    case IR::Type::U64:
+    case IR::Type::F64:
+        ctx.Add("MOV.U64 {}.x,{};", phi_reg, ScalarRegister{eval_value});
+        break;
+    default:
+        throw NotImplementedException("Phi node type {}", phi.Type());
+    }
+}
+
+void EmitPrologue(EmitContext&) {
+    // TODO
+}
+
+void EmitEpilogue(EmitContext&) {
+    // TODO
+}
+
+void EmitEmitVertex(EmitContext& ctx, ScalarS32 stream) {
+    if (stream.type == Type::U32 && stream.imm_u32 == 0) {
+        ctx.Add("EMIT;");
+    } else {
+        ctx.Add("EMITS {};", stream);
+    }
+}
+
+void EmitEndPrimitive(EmitContext& ctx, const IR::Value& stream) {
+    if (!stream.IsImmediate()) {
+        LOG_WARNING(Shader_GLASM, "Stream is not immediate");
+    }
+    ctx.reg_alloc.Consume(stream);
+    ctx.Add("ENDPRIM;");
+}
+
+} // namespace Shader::Backend::GLASM
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_undefined.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_undefined.cpp
index e69de29bb..875e9d991 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_undefined.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_undefined.cpp
@@ -0,0 +1,30 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
+
+namespace Shader::Backend::GLASM {
+
+void EmitUndefU1(EmitContext& ctx, IR::Inst& inst) {
+    ctx.Add("MOV.S {}.x,0;", inst);
+}
+
+void EmitUndefU8(EmitContext& ctx, IR::Inst& inst) {
+    ctx.Add("MOV.S {}.x,0;", inst);
+}
+
+void EmitUndefU16(EmitContext& ctx, IR::Inst& inst) {
+    ctx.Add("MOV.S {}.x,0;", inst);
+}
+
+void EmitUndefU32(EmitContext& ctx, IR::Inst& inst) {
+    ctx.Add("MOV.S {}.x,0;", inst);
+}
+
+void EmitUndefU64(EmitContext& ctx, IR::Inst& inst) {
+    ctx.LongAdd("MOV.S64 {}.x,0;", inst);
+}
+
+} // namespace Shader::Backend::GLASM
diff --git a/src/shader_recompiler/backend/glasm/emit_glasm_warp.cpp b/src/shader_recompiler/backend/glasm/emit_glasm_warp.cpp
index 544d475b4..32e0dd923 100644
--- a/src/shader_recompiler/backend/glasm/emit_glasm_warp.cpp
+++ b/src/shader_recompiler/backend/glasm/emit_glasm_warp.cpp
@@ -2,8 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "shader_recompiler/backend/glasm/emit_context.h"
 #include "shader_recompiler/backend/glasm/emit_glasm_instructions.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 #include "shader_recompiler/profile.h"
 
diff --git a/src/shader_recompiler/backend/glasm/emit_context.cpp b/src/shader_recompiler/backend/glasm/glasm_emit_context.cpp
index 8fd459dfe..0401953f7 100644
--- a/src/shader_recompiler/backend/glasm/emit_context.cpp
+++ b/src/shader_recompiler/backend/glasm/glasm_emit_context.cpp
@@ -5,8 +5,8 @@
 #include <string_view>
 
 #include "shader_recompiler/backend/bindings.h"
-#include "shader_recompiler/backend/glasm/emit_context.h"
 #include "shader_recompiler/backend/glasm/emit_glasm.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/program.h"
 #include "shader_recompiler/profile.h"
 #include "shader_recompiler/runtime_info.h"
diff --git a/src/shader_recompiler/backend/glasm/emit_context.h b/src/shader_recompiler/backend/glasm/glasm_emit_context.h
index 8433e5c00..8433e5c00 100644
--- a/src/shader_recompiler/backend/glasm/emit_context.h
+++ b/src/shader_recompiler/backend/glasm/glasm_emit_context.h
diff --git a/src/shader_recompiler/backend/glasm/reg_alloc.cpp b/src/shader_recompiler/backend/glasm/reg_alloc.cpp
index 4c046db6e..201e428c1 100644
--- a/src/shader_recompiler/backend/glasm/reg_alloc.cpp
+++ b/src/shader_recompiler/backend/glasm/reg_alloc.cpp
@@ -6,7 +6,7 @@
 
 #include <fmt/format.h>
 
-#include "shader_recompiler/backend/glasm/emit_context.h"
+#include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/backend/glasm/reg_alloc.h"
 #include "shader_recompiler/exception.h"
 #include "shader_recompiler/frontend/ir/value.h"
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl.cpp b/src/shader_recompiler/backend/glsl/emit_glsl.cpp
index 8a430d573..78b2eeaa2 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl.cpp
@@ -9,9 +9,9 @@
 
 #include "common/div_ceil.h"
 #include "common/settings.h"
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/ir_emitter.h"
 
 namespace Shader::Backend::GLSL {
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_atomic.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_atomic.cpp
index 772acc5a4..dc377b053 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_atomic.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_atomic.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 namespace Shader::Backend::GLSL {
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_barriers.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_barriers.cpp
index e1d1b558e..8a9faa394 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_barriers.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_barriers.cpp
@@ -2,8 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 namespace Shader::Backend::GLSL {
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_bitwise_conversion.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_bitwise_conversion.cpp
index 3c1714e89..0f2668d9e 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_bitwise_conversion.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_bitwise_conversion.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 namespace Shader::Backend::GLSL {
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_composite.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_composite.cpp
index 49a66e3ec..98cc57e58 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_composite.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_composite.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 namespace Shader::Backend::GLSL {
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_context_get_set.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_context_get_set.cpp
index 4c26f3829..1920047f4 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_context_get_set.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_context_get_set.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 #include "shader_recompiler/profile.h"
 #include "shader_recompiler/runtime_info.h"
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_control_flow.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_control_flow.cpp
index 53f8896be..c86465e8b 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_control_flow.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_control_flow.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/exception.h"
 
 namespace Shader::Backend::GLSL {
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_convert.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_convert.cpp
index eeae6562c..ce6ea1bb7 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_convert.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_convert.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 namespace Shader::Backend::GLSL {
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_floating_point.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_floating_point.cpp
index d423bfb1b..b765a251b 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_floating_point.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_floating_point.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/modifiers.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_image.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_image.cpp
index 2f78d0267..fae2e397a 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_image.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_image.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/modifiers.h"
 #include "shader_recompiler/frontend/ir/value.h"
 #include "shader_recompiler/profile.h"
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_integer.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_integer.cpp
index 88c1d4c5e..44060df33 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_integer.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_integer.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 namespace Shader::Backend::GLSL {
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_logical.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_logical.cpp
index 338ff4bd6..742fec9cf 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_logical.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_logical.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 namespace Shader::Backend::GLSL {
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_memory.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_memory.cpp
index e3957491f..9fd41b4fd 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_memory.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_memory.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 #include "shader_recompiler/profile.h"
 
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_not_implemented.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_not_implemented.cpp
index f420fe388..4ebdfb3bc 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_not_implemented.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_not_implemented.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 #ifdef _MSC_VER
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_select.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_select.cpp
index 49fba9073..b1e486e5f 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_select.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_select.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 namespace Shader::Backend::GLSL {
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_shared_memory.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_shared_memory.cpp
index 518b78f06..74ae345e5 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_shared_memory.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_shared_memory.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 
 namespace Shader::Backend::GLSL {
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_special.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_special.cpp
index 67f9dad68..b8ddafe48 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_special.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_special.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/program.h"
 #include "shader_recompiler/frontend/ir/value.h"
 #include "shader_recompiler/profile.h"
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_undefined.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_undefined.cpp
index 15bf02dd6..cace1db85 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_undefined.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_undefined.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 
 namespace Shader::Backend::GLSL {
 
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_warp.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_warp.cpp
index cd285e2c8..6e01979b4 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_warp.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_warp.cpp
@@ -4,8 +4,8 @@
 
 #include <string_view>
 
-#include "shader_recompiler/backend/glsl/emit_context.h"
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 #include "shader_recompiler/profile.h"
 
diff --git a/src/shader_recompiler/backend/glsl/emit_context.cpp b/src/shader_recompiler/backend/glsl/glsl_emit_context.cpp
index 97bd59302..1de017e76 100644
--- a/src/shader_recompiler/backend/glsl/emit_context.cpp
+++ b/src/shader_recompiler/backend/glsl/glsl_emit_context.cpp
@@ -3,7 +3,7 @@
 // Refer to the license.txt file included.
 
 #include "shader_recompiler/backend/bindings.h"
-#include "shader_recompiler/backend/glsl/emit_context.h"
+#include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/program.h"
 #include "shader_recompiler/profile.h"
 #include "shader_recompiler/runtime_info.h"
diff --git a/src/shader_recompiler/backend/glsl/emit_context.h b/src/shader_recompiler/backend/glsl/glsl_emit_context.h
index d9b639d29..d9b639d29 100644
--- a/src/shader_recompiler/backend/glsl/emit_context.h
+++ b/src/shader_recompiler/backend/glsl/glsl_emit_context.h
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv.cpp b/src/shader_recompiler/backend/spirv/emit_spirv.cpp
index d7a86e270..6ce7ed12a 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv.cpp
@@ -11,6 +11,7 @@
 #include "common/settings.h"
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 #include "shader_recompiler/frontend/ir/basic_block.h"
 #include "shader_recompiler/frontend/ir/program.h"
 
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv.h b/src/shader_recompiler/backend/spirv/emit_spirv.h
index 4b25534ce..b412957c7 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv.h
+++ b/src/shader_recompiler/backend/spirv/emit_spirv.h
@@ -6,13 +6,11 @@
 
 #include <vector>
 
-#include <sirit/sirit.h>
-
 #include "common/common_types.h"
 #include "shader_recompiler/backend/bindings.h"
-#include "shader_recompiler/backend/spirv/emit_context.h"
 #include "shader_recompiler/frontend/ir/program.h"
 #include "shader_recompiler/profile.h"
+#include "shader_recompiler/runtime_info.h"
 
 namespace Shader::Backend::SPIRV {
 
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_atomic.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_atomic.cpp
index 9af8bb9e1..0d37b405c 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_atomic.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_atomic.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 
 namespace Shader::Backend::SPIRV {
 namespace {
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_barriers.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_barriers.cpp
index e0b52a001..9ce95a41b 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_barriers.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_barriers.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 #include "shader_recompiler/frontend/ir/modifiers.h"
 
 namespace Shader::Backend::SPIRV {
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_bitwise_conversion.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_bitwise_conversion.cpp
index bb11f4f4e..02d1e63f7 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_bitwise_conversion.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_bitwise_conversion.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 
 namespace Shader::Backend::SPIRV {
 
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_composite.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_composite.cpp
index 10ff4ecab..5c3e1ee2b 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_composite.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_composite.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 #include "shader_recompiler/frontend/ir/modifiers.h"
 
 namespace Shader::Backend::SPIRV {
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp
index bac683ae1..ad84966b5 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp
@@ -7,6 +7,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 
 namespace Shader::Backend::SPIRV {
 namespace {
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_control_flow.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_control_flow.cpp
index d33486f28..1eca3aa85 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_control_flow.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_control_flow.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 
 namespace Shader::Backend::SPIRV {
 
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_convert.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_convert.cpp
index fd42b7a16..832de2452 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_convert.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_convert.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 
 namespace Shader::Backend::SPIRV {
 namespace {
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_floating_point.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_floating_point.cpp
index 61cf25f9c..0cdc46495 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_floating_point.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_floating_point.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 #include "shader_recompiler/frontend/ir/modifiers.h"
 
 namespace Shader::Backend::SPIRV {
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_image.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_image.cpp
index 4d168a96d..d18d5f1d5 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_image.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_image.cpp
@@ -6,6 +6,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 #include "shader_recompiler/frontend/ir/modifiers.h"
 
 namespace Shader::Backend::SPIRV {
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_image_atomic.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_image_atomic.cpp
index d7f1a365a..a96190bc6 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_image_atomic.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_image_atomic.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 #include "shader_recompiler/frontend/ir/modifiers.h"
 
 namespace Shader::Backend::SPIRV {
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_integer.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_integer.cpp
index 50277eec3..44521f539 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_integer.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_integer.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 
 namespace Shader::Backend::SPIRV {
 namespace {
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_logical.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_logical.cpp
index b9a9500fc..47745f7ee 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_logical.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_logical.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 
 namespace Shader::Backend::SPIRV {
 
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_memory.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_memory.cpp
index 679ee2684..175f4be19 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_memory.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_memory.cpp
@@ -6,6 +6,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 
 namespace Shader::Backend::SPIRV {
 namespace {
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_select.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_select.cpp
index c5b4f4720..48caf1ffc 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_select.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_select.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 
 namespace Shader::Backend::SPIRV {
 
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_shared_memory.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_shared_memory.cpp
index 9a79fc7a2..330c9052c 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_shared_memory.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_shared_memory.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 
 namespace Shader::Backend::SPIRV {
 namespace {
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_special.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_special.cpp
index 9e7eb3cb1..d96a17583 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_special.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_special.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 
 namespace Shader::Backend::SPIRV {
 namespace {
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_undefined.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_undefined.cpp
index c9f469e90..b5766fc52 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_undefined.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_undefined.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 
 namespace Shader::Backend::SPIRV {
 
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp
index cef52c56e..7034228bf 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp
@@ -4,6 +4,7 @@
 
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
 #include "shader_recompiler/backend/spirv/emit_spirv_instructions.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 
 namespace Shader::Backend::SPIRV {
 namespace {
diff --git a/src/shader_recompiler/backend/spirv/emit_context.cpp b/src/shader_recompiler/backend/spirv/spirv_emit_context.cpp
index 723455462..4b6f792bf 100644
--- a/src/shader_recompiler/backend/spirv/emit_context.cpp
+++ b/src/shader_recompiler/backend/spirv/spirv_emit_context.cpp
@@ -13,8 +13,8 @@
 
 #include "common/common_types.h"
 #include "common/div_ceil.h"
-#include "shader_recompiler/backend/spirv/emit_context.h"
 #include "shader_recompiler/backend/spirv/emit_spirv.h"
+#include "shader_recompiler/backend/spirv/spirv_emit_context.h"
 
 namespace Shader::Backend::SPIRV {
 namespace {
diff --git a/src/shader_recompiler/backend/spirv/emit_context.h b/src/shader_recompiler/backend/spirv/spirv_emit_context.h
index 63f8185d9..63f8185d9 100644
--- a/src/shader_recompiler/backend/spirv/emit_context.h
+++ b/src/shader_recompiler/backend/spirv/spirv_emit_context.h
diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp
index ab7c21a49..8788f5148 100644
--- a/src/video_core/gpu.cpp
+++ b/src/video_core/gpu.cpp
@@ -185,16 +185,6 @@ struct GPU::Impl {
         return *dma_pusher;
     }
 
-    /// Returns a reference to the GPU CDMA pusher.
-    [[nodiscard]] Tegra::CDmaPusher& CDmaPusher() {
-        return *cdma_pusher;
-    }
-
-    /// Returns a const reference to the GPU CDMA pusher.
-    [[nodiscard]] const Tegra::CDmaPusher& CDmaPusher() const {
-        return *cdma_pusher;
-    }
-
     /// Returns a reference to the underlying renderer.
     [[nodiscard]] VideoCore::RendererBase& Renderer() {
         return *renderer;
@@ -338,25 +328,27 @@ struct GPU::Impl {
     }
 
     /// Push GPU command buffer entries to be processed
-    void PushCommandBuffer(Tegra::ChCommandHeaderList& entries) {
+    void PushCommandBuffer(u32 id, Tegra::ChCommandHeaderList& entries) {
         if (!use_nvdec) {
             return;
         }
 
-        if (!cdma_pusher) {
-            cdma_pusher = std::make_unique<Tegra::CDmaPusher>(gpu);
+        if (!cdma_pushers.contains(id)) {
+            cdma_pushers.insert_or_assign(id, std::make_unique<Tegra::CDmaPusher>(gpu));
         }
 
         // SubmitCommandBuffer would make the nvdec operations async, this is not currently working
         // TODO(ameerj): RE proper async nvdec operation
         // gpu_thread.SubmitCommandBuffer(std::move(entries));
-
-        cdma_pusher->ProcessEntries(std::move(entries));
+        cdma_pushers[id]->ProcessEntries(std::move(entries));
     }
 
     /// Frees the CDMAPusher instance to free up resources
-    void ClearCdmaInstance() {
-        cdma_pusher.reset();
+    void ClearCdmaInstance(u32 id) {
+        const auto iter = cdma_pushers.find(id);
+        if (iter != cdma_pushers.end()) {
+            cdma_pushers.erase(iter);
+        }
     }
 
     /// Swap buffers (render frame)
@@ -659,7 +651,7 @@ struct GPU::Impl {
     Core::System& system;
     std::unique_ptr<Tegra::MemoryManager> memory_manager;
     std::unique_ptr<Tegra::DmaPusher> dma_pusher;
-    std::unique_ptr<Tegra::CDmaPusher> cdma_pusher;
+    std::map<u32, std::unique_ptr<Tegra::CDmaPusher>> cdma_pushers;
     std::unique_ptr<VideoCore::RendererBase> renderer;
     VideoCore::RasterizerInterface* rasterizer = nullptr;
     const bool use_nvdec;
@@ -811,14 +803,6 @@ const Tegra::DmaPusher& GPU::DmaPusher() const {
     return impl->DmaPusher();
 }
 
-Tegra::CDmaPusher& GPU::CDmaPusher() {
-    return impl->CDmaPusher();
-}
-
-const Tegra::CDmaPusher& GPU::CDmaPusher() const {
-    return impl->CDmaPusher();
-}
-
 VideoCore::RendererBase& GPU::Renderer() {
     return impl->Renderer();
 }
@@ -887,12 +871,12 @@ void GPU::PushGPUEntries(Tegra::CommandList&& entries) {
     impl->PushGPUEntries(std::move(entries));
 }
 
-void GPU::PushCommandBuffer(Tegra::ChCommandHeaderList& entries) {
-    impl->PushCommandBuffer(entries);
+void GPU::PushCommandBuffer(u32 id, Tegra::ChCommandHeaderList& entries) {
+    impl->PushCommandBuffer(id, entries);
 }
 
-void GPU::ClearCdmaInstance() {
-    impl->ClearCdmaInstance();
+void GPU::ClearCdmaInstance(u32 id) {
+    impl->ClearCdmaInstance(id);
 }
 
 void GPU::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h
index c89a5d693..500411176 100644
--- a/src/video_core/gpu.h
+++ b/src/video_core/gpu.h
@@ -242,10 +242,10 @@ public:
     void PushGPUEntries(Tegra::CommandList&& entries);
 
     /// Push GPU command buffer entries to be processed
-    void PushCommandBuffer(Tegra::ChCommandHeaderList& entries);
+    void PushCommandBuffer(u32 id, Tegra::ChCommandHeaderList& entries);
 
     /// Frees the CDMAPusher instance to free up resources
-    void ClearCdmaInstance();
+    void ClearCdmaInstance(u32 id);
 
     /// Swap buffers (render frame)
     void SwapBuffers(const Tegra::FramebufferConfig* framebuffer);
diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
index 31adada56..e38cfbc6c 100644
--- a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
+++ b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
@@ -162,7 +162,7 @@ struct FormatTuple {
     {VK_FORMAT_UNDEFINED},                                     // R16_SINT
     {VK_FORMAT_R16G16_UNORM, Attachable | Storage},            // R16G16_UNORM
     {VK_FORMAT_R16G16_SFLOAT, Attachable | Storage},           // R16G16_FLOAT
-    {VK_FORMAT_UNDEFINED},                                     // R16G16_UINT
+    {VK_FORMAT_R16G16_UINT, Attachable | Storage},             // R16G16_UINT
     {VK_FORMAT_R16G16_SINT, Attachable | Storage},             // R16G16_SINT
     {VK_FORMAT_R16G16_SNORM, Attachable | Storage},            // R16G16_SNORM
     {VK_FORMAT_UNDEFINED},                                     // R32G32B32_FLOAT
diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h
index 565b99254..e195b1e98 100644
--- a/src/video_core/texture_cache/texture_cache.h
+++ b/src/video_core/texture_cache/texture_cache.h
@@ -1122,7 +1122,7 @@ typename TextureCache<P>::BlitImages TextureCache<P>::GetBlitImages(
             break;
         }
         if (can_be_depth_blit) {
-            const ImageBase* const dst_image = src_id ? &slot_images[src_id] : nullptr;
+            const ImageBase* const dst_image = dst_id ? &slot_images[dst_id] : nullptr;
             DeduceBlitImages(dst_info, src_info, dst_image, src_image);
             if (GetFormatType(dst_info.format) != GetFormatType(src_info.format)) {
                 continue;
@@ -1135,8 +1135,11 @@ typename TextureCache<P>::BlitImages TextureCache<P>::GetBlitImages(
             dst_id = InsertImage(dst_info, dst_addr, RelaxedOptions{});
         }
     } while (has_deleted_images);
-    if (GetFormatType(dst_info.format) != SurfaceType::ColorTexture) {
-        // Make sure the images are depth and/or stencil textures.
+    const ImageBase& src_image = slot_images[src_id];
+    const ImageBase& dst_image = slot_images[dst_id];
+    if (GetFormatType(dst_info.format) != GetFormatType(dst_image.info.format) ||
+        GetFormatType(src_info.format) != GetFormatType(src_image.info.format)) {
+        // Make sure the images match the expected format.
         do {
             has_deleted_images = false;
             src_id = FindOrInsertImage(src_info, src_addr, RelaxedOptions{});
diff --git a/src/video_core/vulkan_common/vulkan_device.cpp b/src/video_core/vulkan_common/vulkan_device.cpp
index 70c52aaac..7bf5b6578 100644
--- a/src/video_core/vulkan_common/vulkan_device.cpp
+++ b/src/video_core/vulkan_common/vulkan_device.cpp
@@ -130,6 +130,7 @@ std::unordered_map<VkFormat, VkFormatProperties> GetFormatProperties(vk::Physica
         VK_FORMAT_R16G16_UNORM,
         VK_FORMAT_R16G16_SNORM,
         VK_FORMAT_R16G16_SFLOAT,
+        VK_FORMAT_R16G16_UINT,
         VK_FORMAT_R16G16_SINT,
         VK_FORMAT_R16_UNORM,
         VK_FORMAT_R16_SNORM,
diff --git a/src/yuzu/debugger/profiler.cpp b/src/yuzu/debugger/profiler.cpp
index a8b254199..33110685a 100644
--- a/src/yuzu/debugger/profiler.cpp
+++ b/src/yuzu/debugger/profiler.cpp
@@ -163,7 +163,7 @@ void MicroProfileWidget::mouseReleaseEvent(QMouseEvent* ev) {
 }
 
 void MicroProfileWidget::wheelEvent(QWheelEvent* ev) {
-    const auto wheel_position = ev->pos();
+    const auto wheel_position = ev->position().toPoint();
     MicroProfileMousePosition(wheel_position.x() / x_scale, wheel_position.y() / y_scale,
                               ev->angleDelta().y() / 120);
     ev->accept();
diff --git a/src/yuzu/main.cpp b/src/yuzu/main.cpp
index 5a9dec8f3..cc84ea11c 100644
--- a/src/yuzu/main.cpp
+++ b/src/yuzu/main.cpp
@@ -1082,14 +1082,15 @@ void GMainWindow::OnAppFocusStateChanged(Qt::ApplicationState state) {
         state != Qt::ApplicationActive) {
         LOG_DEBUG(Frontend, "ApplicationState unusual flag: {} ", state);
     }
-    if (ui->action_Pause->isEnabled() &&
-        (state & (Qt::ApplicationHidden | Qt::ApplicationInactive))) {
-        auto_paused = true;
-        OnPauseGame();
-    } else if (emulation_running && !emu_thread->IsRunning() && auto_paused &&
-               state == Qt::ApplicationActive) {
-        auto_paused = false;
-        OnStartGame();
+    if (emulation_running) {
+        if (emu_thread->IsRunning() &&
+            (state & (Qt::ApplicationHidden | Qt::ApplicationInactive))) {
+            auto_paused = true;
+            OnPauseGame();
+        } else if (!emu_thread->IsRunning() && auto_paused && state == Qt::ApplicationActive) {
+            auto_paused = false;
+            OnStartGame();
+        }
     }
 }