Memory manager: Fix possible softlock

Y.F.C Implement Asynchronous Fence manager and Rework Query async downloads

.ci/scripts/linux/docker.sh

@@ -22,6 +22,7 @@ cmake .. \
       -DUSE_DISCORD_PRESENCE=ON \
       -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${ENABLE_COMPATIBILITY_REPORTING:-"OFF"} \
       -DYUZU_USE_BUNDLED_FFMPEG=ON \
+      -DYUZU_ENABLE_LTO=ON \
       -GNinja
 
 ninja

CMakeLists.txt

@@ -210,7 +210,7 @@ set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${PROJECT_BINARY_DIR}/bin)
 # =======================================================================
 
 # Enforce the search mode of non-required packages for better and shorter failure messages
-find_package(Boost 1.73.0 REQUIRED context)
+find_package(Boost 1.79.0 REQUIRED context)
 find_package(enet 1.3 MODULE)
 find_package(fmt 9 REQUIRED)
 find_package(inih 52 MODULE COMPONENTS INIReader)
@@ -222,7 +222,7 @@ find_package(ZLIB 1.2 REQUIRED)
 find_package(zstd 1.5 REQUIRED)
 
 if (NOT YUZU_USE_EXTERNAL_VULKAN_HEADERS)
-    find_package(Vulkan 1.3.238 REQUIRED)
+    find_package(Vulkan 1.3.246 REQUIRED)
 endif()
 
 if (ENABLE_LIBUSB)

src/CMakeLists.txt

@@ -113,6 +113,9 @@ else()
 
         $<$<CXX_COMPILER_ID:Clang>:-Wno-braced-scalar-init>
         $<$<CXX_COMPILER_ID:Clang>:-Wno-unused-private-field>
+        $<$<CXX_COMPILER_ID:Clang>:-Werror=shadow-uncaptured-local>
+        $<$<CXX_COMPILER_ID:Clang>:-Werror=implicit-fallthrough>
+        $<$<CXX_COMPILER_ID:Clang>:-Werror=type-limits>
         $<$<CXX_COMPILER_ID:AppleClang>:-Wno-braced-scalar-init>
         $<$<CXX_COMPILER_ID:AppleClang>:-Wno-unused-private-field>
     )
@@ -126,6 +129,17 @@ else()
         add_compile_options("-stdlib=libc++")
     endif()
 
+    # GCC bugs
+    if (CMAKE_CXX_COMPILER_VERSION VERSION_GREATER_EQUAL "12" AND CMAKE_CXX_COMPILER_ID STREQUAL "GNU")
+        # These diagnostics would be great if they worked, but are just completely broken
+        # and produce bogus errors on external libraries like fmt.
+        add_compile_options(
+            -Wno-array-bounds
+            -Wno-stringop-overread
+            -Wno-stringop-overflow
+        )
+    endif()
+
     # Set file offset size to 64 bits.
     #
     # On modern Unixes, this is typically already the case. The lone exception is

src/audio_core/device/device_session.cpp

@@ -93,7 +93,7 @@ void DeviceSession::AppendBuffers(std::span<const AudioBuffer> buffers) const {
             stream->AppendBuffer(new_buffer, samples);
         } else {
             std::vector<s16> samples(buffer.size / sizeof(s16));
-            system.Memory().ReadBlockUnsafe(buffer.samples, samples.data(), buffer.size);
+            system.ApplicationMemory().ReadBlockUnsafe(buffer.samples, samples.data(), buffer.size);
             stream->AppendBuffer(new_buffer, samples);
         }
     }
@@ -102,7 +102,7 @@ void DeviceSession::AppendBuffers(std::span<const AudioBuffer> buffers) const {
 void DeviceSession::ReleaseBuffer(const AudioBuffer& buffer) const {
     if (type == Sink::StreamType::In) {
         auto samples{stream->ReleaseBuffer(buffer.size / sizeof(s16))};
-        system.Memory().WriteBlockUnsafe(buffer.samples, samples.data(), buffer.size);
+        system.ApplicationMemory().WriteBlockUnsafe(buffer.samples, samples.data(), buffer.size);
     }
 }
 
@@ -121,8 +121,7 @@ u64 DeviceSession::GetPlayedSampleCount() const {
 }
 
 std::optional<std::chrono::nanoseconds> DeviceSession::ThreadFunc() {
-    // Add 5ms of samples at a 48K sample rate.
-    played_sample_count += 48'000 * INCREMENT_TIME / 1s;
+    played_sample_count = stream->GetExpectedPlayedSampleCount();
     if (type == Sink::StreamType::Out) {
         system.AudioCore().GetAudioManager().SetEvent(Event::Type::AudioOutManager, true);
     } else {

src/audio_core/renderer/adsp/adsp.cpp

@@ -13,7 +13,7 @@
 namespace AudioCore::AudioRenderer::ADSP {
 
 ADSP::ADSP(Core::System& system_, Sink::Sink& sink_)
-    : system{system_}, memory{system.Memory()}, sink{sink_} {}
+    : system{system_}, memory{system.ApplicationMemory()}, sink{sink_} {}
 
 ADSP::~ADSP() {
     ClearCommandBuffers();

src/audio_core/renderer/adsp/audio_renderer.cpp

@@ -189,6 +189,8 @@ void AudioRenderer::ThreadFunc() {
                     max_time = std::min(command_buffer.time_limit, max_time);
                     command_list_processor.SetProcessTimeMax(max_time);
 
+                    streams[index]->WaitFreeSpace();
+
                     // Process the command list
                     {
                         MICROPROFILE_SCOPE(Audio_Renderer);

src/audio_core/renderer/adsp/audio_renderer.h

@@ -10,6 +10,7 @@
 #include "audio_core/renderer/adsp/command_buffer.h"
 #include "audio_core/renderer/adsp/command_list_processor.h"
 #include "common/common_types.h"
+#include "common/polyfill_thread.h"
 #include "common/reader_writer_queue.h"
 #include "common/thread.h"
 

src/audio_core/renderer/adsp/command_list_processor.cpp

@@ -17,7 +17,7 @@ namespace AudioCore::AudioRenderer::ADSP {
 void CommandListProcessor::Initialize(Core::System& system_, CpuAddr buffer, u64 size,
                                       Sink::SinkStream* stream_) {
     system = &system_;
-    memory = &system->Memory();
+    memory = &system->ApplicationMemory();
     stream = stream_;
     header = reinterpret_cast<CommandListHeader*>(buffer);
     commands = reinterpret_cast<u8*>(buffer + sizeof(CommandListHeader));

src/audio_core/renderer/system.cpp

@@ -127,8 +127,7 @@ Result System::Initialize(const AudioRendererParameterInternal& params,
     render_device = params.rendering_device;
     execution_mode = params.execution_mode;
 
-    core.Memory().ZeroBlock(*core.ApplicationProcess(), transfer_memory->GetSourceAddress(),
-                            transfer_memory_size);
+    core.ApplicationMemory().ZeroBlock(transfer_memory->GetSourceAddress(), transfer_memory_size);
 
     // Note: We're not actually using the transfer memory because it's a pain to code for.
     // Allocate the memory normally instead and hope the game doesn't try to read anything back

src/audio_core/renderer/system_manager.cpp

@@ -15,14 +15,9 @@ MICROPROFILE_DEFINE(Audio_RenderSystemManager, "Audio", "Render System Manager",
                     MP_RGB(60, 19, 97));
 
 namespace AudioCore::AudioRenderer {
-constexpr std::chrono::nanoseconds RENDER_TIME{5'000'000UL};
 
 SystemManager::SystemManager(Core::System& core_)
-    : core{core_}, adsp{core.AudioCore().GetADSP()}, mailbox{adsp.GetRenderMailbox()},
-      thread_event{Core::Timing::CreateEvent(
-          "AudioRendererSystemManager", [this](std::uintptr_t, s64 time, std::chrono::nanoseconds) {
-              return ThreadFunc2(time);
-          })} {}
+    : core{core_}, adsp{core.AudioCore().GetADSP()}, mailbox{adsp.GetRenderMailbox()} {}
 
 SystemManager::~SystemManager() {
     Stop();
@@ -33,8 +28,6 @@ bool SystemManager::InitializeUnsafe() {
         if (adsp.Start()) {
             active = true;
             thread = std::jthread([this](std::stop_token stop_token) { ThreadFunc(); });
-            core.CoreTiming().ScheduleLoopingEvent(std::chrono::nanoseconds(0), RENDER_TIME,
-                                                   thread_event);
         }
     }
 
@@ -45,7 +38,6 @@ void SystemManager::Stop() {
     if (!active) {
         return;
     }
-    core.CoreTiming().UnscheduleEvent(thread_event, {});
     active = false;
     update.store(true);
     update.notify_all();
@@ -111,16 +103,7 @@ void SystemManager::ThreadFunc() {
 
         adsp.Signal();
         adsp.Wait();
-
-        update.wait(false);
-        update.store(false);
     }
 }
 
-std::optional<std::chrono::nanoseconds> SystemManager::ThreadFunc2(s64 time) {
-    update.store(true);
-    update.notify_all();
-    return std::nullopt;
-}
-
 } // namespace AudioCore::AudioRenderer

src/audio_core/renderer/system_manager.h

@@ -68,11 +68,6 @@ private:
      */
     void ThreadFunc();
 
-    /**
-     * Signalling core timing thread to run ThreadFunc.
-     */
-    std::optional<std::chrono::nanoseconds> ThreadFunc2(s64 time);
-
     enum class StreamState {
         Filling,
         Steady,
@@ -95,8 +90,6 @@ private:
     ADSP::ADSP& adsp;
     /// AudioRenderer mailbox for communication
     ADSP::AudioRenderer_Mailbox* mailbox{};
-    /// Core timing event to signal main thread
-    std::shared_ptr<Core::Timing::EventType> thread_event;
     /// Atomic for main thread to wait on
     std::atomic<bool> update{};
 };

src/audio_core/sink/cubeb_sink.cpp

@@ -101,8 +101,6 @@ public:
     ~CubebSinkStream() override {
         LOG_DEBUG(Service_Audio, "Destructing cubeb stream {}", name);
 
-        Unstall();
-
         if (!ctx) {
             return;
         }
@@ -143,8 +141,6 @@ public:
      * Stop the sink stream.
      */
     void Stop() override {
-        Unstall();
-
         if (!ctx || paused) {
             return;
         }

src/audio_core/sink/sdl2_sink.cpp

@@ -88,7 +88,6 @@ public:
      * Finalize the sink stream.
      */
     void Finalize() override {
-        Unstall();
         if (device == 0) {
             return;
         }
@@ -116,7 +115,6 @@ public:
      * Stop the sink stream.
      */
     void Stop() override {
-        Unstall();
         if (device == 0 || paused) {
             return;
         }

src/audio_core/sink/sink_stream.cpp

@@ -14,6 +14,8 @@
 #include "common/fixed_point.h"
 #include "common/settings.h"
 #include "core/core.h"
+#include "core/core_timing.h"
+#include "core/core_timing_util.h"
 
 namespace AudioCore::Sink {
 
@@ -149,10 +151,6 @@ void SinkStream::ProcessAudioIn(std::span<const s16> input_buffer, std::size_t n
         return;
     }
 
-    if (queued_buffers > max_queue_size) {
-        Stall();
-    }
-
     while (frames_written < num_frames) {
         // If the playing buffer has been consumed or has no frames, we need a new one
         if (playing_buffer.consumed || playing_buffer.frames == 0) {
@@ -187,10 +185,6 @@ void SinkStream::ProcessAudioIn(std::span<const s16> input_buffer, std::size_t n
     }
 
     std::memcpy(&last_frame[0], &input_buffer[(frames_written - 1) * frame_size], frame_size_bytes);
-
-    if (queued_buffers <= max_queue_size) {
-        Unstall();
-    }
 }
 
 void SinkStream::ProcessAudioOutAndRender(std::span<s16> output_buffer, std::size_t num_frames) {
@@ -198,10 +192,15 @@ void SinkStream::ProcessAudioOutAndRender(std::span<s16> output_buffer, std::siz
     const std::size_t frame_size = num_channels;
     const std::size_t frame_size_bytes = frame_size * sizeof(s16);
     size_t frames_written{0};
+    size_t actual_frames_written{0};
 
     // If we're paused or going to shut down, we don't want to consume buffers as coretiming is
     // paused and we'll desync, so just play silence.
     if (system.IsPaused() || system.IsShuttingDown()) {
+        if (system.IsShuttingDown()) {
+            release_cv.notify_one();
+        }
+
         static constexpr std::array<s16, 6> silence{};
         for (size_t i = frames_written; i < num_frames; i++) {
             std::memcpy(&output_buffer[i * frame_size], &silence[0], frame_size_bytes);
@@ -209,20 +208,6 @@ void SinkStream::ProcessAudioOutAndRender(std::span<s16> output_buffer, std::siz
         return;
     }
 
-    // Due to many frames being queued up with nvdec (5 frames or so?), a lot of buffers also get
-    // queued up (30+) but not all at once, which causes constant stalling here, so just let the
-    // video play out without attempting to stall.
-    // Can hopefully remove this later with a more complete NVDEC implementation.
-    const auto nvdec_active{system.AudioCore().IsNVDECActive()};
-
-    // Core timing cannot be paused in single-core mode, so Stall ends up being called over and over
-    // and never recovers to a normal state, so just skip attempting to sync things on single-core.
-    if (system.IsMulticore() && !nvdec_active && queued_buffers > max_queue_size) {
-        Stall();
-    } else if (system.IsMulticore() && queued_buffers <= max_queue_size) {
-        Unstall();
-    }
-
     while (frames_written < num_frames) {
         // If the playing buffer has been consumed or has no frames, we need a new one
         if (playing_buffer.consumed || playing_buffer.frames == 0) {
@@ -237,6 +222,10 @@ void SinkStream::ProcessAudioOutAndRender(std::span<s16> output_buffer, std::siz
             }
             // Successfully dequeued a new buffer.
             queued_buffers--;
+
+            { std::unique_lock lk{release_mutex}; }
+
+            release_cv.notify_one();
         }
 
         // Get the minimum frames available between the currently playing buffer, and the
@@ -248,6 +237,7 @@ void SinkStream::ProcessAudioOutAndRender(std::span<s16> output_buffer, std::siz
                            frames_available * frame_size);
 
         frames_written += frames_available;
+        actual_frames_written += frames_available;
         playing_buffer.frames_played += frames_available;
 
         // If that's all the frames in the current buffer, add its samples and mark it as
@@ -260,26 +250,29 @@ void SinkStream::ProcessAudioOutAndRender(std::span<s16> output_buffer, std::siz
     std::memcpy(&last_frame[0], &output_buffer[(frames_written - 1) * frame_size],
                 frame_size_bytes);
 
-    if (system.IsMulticore() && queued_buffers <= max_queue_size) {
-        Unstall();
+    {
+        std::scoped_lock lk{sample_count_lock};
+        last_sample_count_update_time = system.CoreTiming().GetGlobalTimeNs();
+        min_played_sample_count = max_played_sample_count;
+        max_played_sample_count += actual_frames_written;
     }
 }
 
-void SinkStream::Stall() {
-    std::scoped_lock lk{stall_guard};
-    if (stalled_lock) {
-        return;
-    }
-    stalled_lock = system.StallApplication();
+u64 SinkStream::GetExpectedPlayedSampleCount() {
+    std::scoped_lock lk{sample_count_lock};
+    auto cur_time{system.CoreTiming().GetGlobalTimeNs()};
+    auto time_delta{cur_time - last_sample_count_update_time};
+    auto exp_played_sample_count{min_played_sample_count +
+                                 (TargetSampleRate * time_delta) / std::chrono::seconds{1}};
+
+    // Add 15ms of latency in sample reporting to allow for some leeway in scheduler timings
+    return std::min<u64>(exp_played_sample_count, max_played_sample_count) + TargetSampleCount * 3;
 }
 
-void SinkStream::Unstall() {
-    std::scoped_lock lk{stall_guard};
-    if (!stalled_lock) {
-        return;
-    }
-    system.UnstallApplication();
-    stalled_lock.unlock();
+void SinkStream::WaitFreeSpace() {
+    std::unique_lock lk{release_mutex};
+    release_cv.wait(
+        lk, [this]() { return queued_buffers < max_queue_size || system.IsShuttingDown(); });
 }
 
 } // namespace AudioCore::Sink

src/audio_core/sink/sink_stream.h

@@ -5,6 +5,7 @@
 
 #include <array>
 #include <atomic>
+#include <chrono>
 #include <memory>
 #include <mutex>
 #include <span>
@@ -14,6 +15,7 @@
 #include "common/common_types.h"
 #include "common/reader_writer_queue.h"
 #include "common/ring_buffer.h"
+#include "common/thread.h"
 
 namespace Core {
 class System;
@@ -53,9 +55,7 @@ struct SinkBuffer {
 class SinkStream {
 public:
     explicit SinkStream(Core::System& system_, StreamType type_) : system{system_}, type{type_} {}
-    virtual ~SinkStream() {
-        Unstall();
-    }
+    virtual ~SinkStream() {}
 
     /**
      * Finalize the sink stream.
@@ -201,14 +201,16 @@ public:
     void ProcessAudioOutAndRender(std::span<s16> output_buffer, std::size_t num_frames);
 
     /**
-     * Stall core processes if the audio thread falls too far behind.
+     * Get the total number of samples expected to have been played by this stream.
+     *
+     * @return The number of samples.
      */
-    void Stall();
+    u64 GetExpectedPlayedSampleCount();
 
     /**
-     * Unstall core processes.
+     * Waits for free space in the sample ring buffer
      */
-    void Unstall();
+    void WaitFreeSpace();
 
 protected:
     /// Core system
@@ -237,12 +239,21 @@ private:
     std::atomic<u32> queued_buffers{};
     /// The ring size for audio out buffers (usually 4, rarely 2 or 8)
     u32 max_queue_size{};
+    /// Locks access to sample count tracking info
+    std::mutex sample_count_lock;
+    /// Minimum number of total samples that have been played since the last callback
+    u64 min_played_sample_count{};
+    /// Maximum number of total samples that can be played since the last callback
+    u64 max_played_sample_count{};
+    /// The time the two above tracking variables were last written to
+    std::chrono::nanoseconds last_sample_count_update_time{};
     /// Set by the audio render/in/out system which uses this stream
     f32 system_volume{1.0f};
     /// Set via IAudioDevice service calls
     f32 device_volume{1.0f};
-    std::mutex stall_guard;
-    std::unique_lock<std::mutex> stalled_lock;
+    /// Signalled when ring buffer entries are consumed
+    std::condition_variable release_cv;
+    std::mutex release_mutex;
 };
 
 using SinkStreamPtr = std::unique_ptr<SinkStream>;

src/common/CMakeLists.txt

@@ -38,6 +38,7 @@ add_library(common STATIC
     common_precompiled_headers.h
     common_types.h
     concepts.h
+    container_hash.h
     demangle.cpp
     demangle.h
     div_ceil.h
@@ -132,6 +133,7 @@ add_library(common STATIC
     time_zone.h
     tiny_mt.h
     tree.h
+    typed_address.h
     uint128.h
     unique_function.h
     uuid.cpp
@@ -158,6 +160,8 @@ if(ARCHITECTURE_x86_64)
         PRIVATE
             x64/cpu_detect.cpp
             x64/cpu_detect.h
+            x64/cpu_wait.cpp
+            x64/cpu_wait.h
             x64/native_clock.cpp
             x64/native_clock.h
             x64/xbyak_abi.h

src/common/bounded_threadsafe_queue.h

@@ -1,158 +1,249 @@
-// SPDX-FileCopyrightText: Copyright (c) 2020 Erik Rigtorp <erik@rigtorp.se>
-// SPDX-License-Identifier: MIT
+// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
 #include <atomic>
-#include <bit>
 #include <condition_variable>
-#include <memory>
+#include <cstddef>
 #include <mutex>
 #include <new>
-#include <stop_token>
-#include <type_traits>
-#include <utility>
+
+#include "common/polyfill_thread.h"
 
 namespace Common {
 
-#if defined(__cpp_lib_hardware_interference_size)
-constexpr size_t hardware_interference_size = std::hardware_destructive_interference_size;
-#else
-constexpr size_t hardware_interference_size = 64;
-#endif
+namespace detail {
+constexpr size_t DefaultCapacity = 0x1000;
+} // namespace detail
+
+template <typename T, size_t Capacity = detail::DefaultCapacity>
+class SPSCQueue {
+    static_assert((Capacity & (Capacity - 1)) == 0, "Capacity must be a power of two.");
 
-template <typename T, size_t capacity = 0x400>
-class MPSCQueue {
 public:
-    explicit MPSCQueue() : allocator{std::allocator<Slot<T>>()} {
-        // Allocate one extra slot to prevent false sharing on the last slot
-        slots = allocator.allocate(capacity + 1);
-        // Allocators are not required to honor alignment for over-aligned types
-        // (see http://eel.is/c++draft/allocator.requirements#10) so we verify
-        // alignment here
-        if (reinterpret_cast<uintptr_t>(slots) % alignof(Slot<T>) != 0) {
-            allocator.deallocate(slots, capacity + 1);
-            throw std::bad_alloc();
-        }
-        for (size_t i = 0; i < capacity; ++i) {
-            std::construct_at(&slots[i]);
-        }
-        static_assert(std::has_single_bit(capacity), "capacity must be an integer power of 2");
-        static_assert(alignof(Slot<T>) == hardware_interference_size,
-                      "Slot must be aligned to cache line boundary to prevent false sharing");
-        static_assert(sizeof(Slot<T>) % hardware_interference_size == 0,
-                      "Slot size must be a multiple of cache line size to prevent "
-                      "false sharing between adjacent slots");
-        static_assert(sizeof(MPSCQueue) % hardware_interference_size == 0,
-                      "Queue size must be a multiple of cache line size to "
-                      "prevent false sharing between adjacent queues");
+    template <typename... Args>
+    bool TryEmplace(Args&&... args) {
+        return Emplace<PushMode::Try>(std::forward<Args>(args)...);
     }
 
-    ~MPSCQueue() noexcept {
-        for (size_t i = 0; i < capacity; ++i) {
-            std::destroy_at(&slots[i]);
-        }
-        allocator.deallocate(slots, capacity + 1);
+    template <typename... Args>
+    void EmplaceWait(Args&&... args) {
+        Emplace<PushMode::Wait>(std::forward<Args>(args)...);
     }
 
-    // The queue must be both non-copyable and non-movable
-    MPSCQueue(const MPSCQueue&) = delete;
-    MPSCQueue& operator=(const MPSCQueue&) = delete;
-
-    MPSCQueue(MPSCQueue&&) = delete;
-    MPSCQueue& operator=(MPSCQueue&&) = delete;
-
-    void Push(const T& v) noexcept {
-        static_assert(std::is_nothrow_copy_constructible_v<T>,
-                      "T must be nothrow copy constructible");
-        emplace(v);
+    bool TryPop(T& t) {
+        return Pop<PopMode::Try>(t);
     }
 
-    template <typename P, typename = std::enable_if_t<std::is_nothrow_constructible_v<T, P&&>>>
-    void Push(P&& v) noexcept {
-        emplace(std::forward<P>(v));
+    void PopWait(T& t) {
+        Pop<PopMode::Wait>(t);
     }
 
-    void Pop(T& v, std::stop_token stop) noexcept {
-        auto const tail = tail_.fetch_add(1);
-        auto& slot = slots[idx(tail)];
-        if (!slot.turn.test()) {
-            std::unique_lock lock{cv_mutex};
-            cv.wait(lock, stop, [&slot] { return slot.turn.test(); });
-        }
-        v = slot.move();
-        slot.destroy();
-        slot.turn.clear();
-        slot.turn.notify_one();
+    void PopWait(T& t, std::stop_token stop_token) {
+        Pop<PopMode::WaitWithStopToken>(t, stop_token);
+    }
+
+    T PopWait() {
+        T t;
+        Pop<PopMode::Wait>(t);
+        return t;
+    }
+
+    T PopWait(std::stop_token stop_token) {
+        T t;
+        Pop<PopMode::WaitWithStopToken>(t, stop_token);
+        return t;
     }
 
 private:
-    template <typename U = T>
-    struct Slot {
-        ~Slot() noexcept {
-            if (turn.test()) {
-                destroy();
-            }
-        }
-
-        template <typename... Args>
-        void construct(Args&&... args) noexcept {
-            static_assert(std::is_nothrow_constructible_v<U, Args&&...>,
-                          "T must be nothrow constructible with Args&&...");
-            std::construct_at(reinterpret_cast<U*>(&storage), std::forward<Args>(args)...);
-        }
-
-        void destroy() noexcept {
-            static_assert(std::is_nothrow_destructible_v<U>, "T must be nothrow destructible");
-            std::destroy_at(reinterpret_cast<U*>(&storage));
-        }
-
-        U&& move() noexcept {
-            return reinterpret_cast<U&&>(storage);
-        }
-
-        // Align to avoid false sharing between adjacent slots
-        alignas(hardware_interference_size) std::atomic_flag turn{};
-        struct aligned_store {
-            struct type {
-                alignas(U) unsigned char data[sizeof(U)];
-            };
-        };
-        typename aligned_store::type storage;
+    enum class PushMode {
+        Try,
+        Wait,
+        Count,
     };
 
+    enum class PopMode {
+        Try,
+        Wait,
+        WaitWithStopToken,
+        Count,
+    };
+
+    template <PushMode Mode, typename... Args>
+    bool Emplace(Args&&... args) {
+        const size_t write_index = m_write_index.load(std::memory_order::relaxed);
+
+        if constexpr (Mode == PushMode::Try) {
+            // Check if we have free slots to write to.
+            if ((write_index - m_read_index.load(std::memory_order::acquire)) == Capacity) {
+                return false;
+            }
+        } else if constexpr (Mode == PushMode::Wait) {
+            // Wait until we have free slots to write to.
+            std::unique_lock lock{producer_cv_mutex};
+            producer_cv.wait(lock, [this, write_index] {
+                return (write_index - m_read_index.load(std::memory_order::acquire)) < Capacity;
+            });
+        } else {
+            static_assert(Mode < PushMode::Count, "Invalid PushMode.");
+        }
+
+        // Determine the position to write to.
+        const size_t pos = write_index % Capacity;
+
+        // Emplace into the queue.
+        std::construct_at(std::addressof(m_data[pos]), std::forward<Args>(args)...);
+
+        // Increment the write index.
+        ++m_write_index;
+
+        // Notify the consumer that we have pushed into the queue.
+        std::scoped_lock lock{consumer_cv_mutex};
+        consumer_cv.notify_one();
+
+        return true;
+    }
+
+    template <PopMode Mode>
+    bool Pop(T& t, [[maybe_unused]] std::stop_token stop_token = {}) {
+        const size_t read_index = m_read_index.load(std::memory_order::relaxed);
+
+        if constexpr (Mode == PopMode::Try) {
+            // Check if the queue is empty.
+            if (read_index == m_write_index.load(std::memory_order::acquire)) {
+                return false;
+            }
+        } else if constexpr (Mode == PopMode::Wait) {
+            // Wait until the queue is not empty.
+            std::unique_lock lock{consumer_cv_mutex};
+            consumer_cv.wait(lock, [this, read_index] {
+                return read_index != m_write_index.load(std::memory_order::acquire);
+            });
+        } else if constexpr (Mode == PopMode::WaitWithStopToken) {
+            // Wait until the queue is not empty.
+            std::unique_lock lock{consumer_cv_mutex};
+            Common::CondvarWait(consumer_cv, lock, stop_token, [this, read_index] {
+                return read_index != m_write_index.load(std::memory_order::acquire);
+            });
+            if (stop_token.stop_requested()) {
+                return false;
+            }
+        } else {
+            static_assert(Mode < PopMode::Count, "Invalid PopMode.");
+        }
+
+        // Determine the position to read from.
+        const size_t pos = read_index % Capacity;
+
+        // Pop the data off the queue, moving it.
+        t = std::move(m_data[pos]);
+
+        // Increment the read index.
+        ++m_read_index;
+
+        // Notify the producer that we have popped off the queue.
+        std::scoped_lock lock{producer_cv_mutex};
+        producer_cv.notify_one();
+
+        return true;
+    }
+
+    alignas(128) std::atomic_size_t m_read_index{0};
+    alignas(128) std::atomic_size_t m_write_index{0};
+
+    std::array<T, Capacity> m_data;
+
+    std::condition_variable_any producer_cv;
+    std::mutex producer_cv_mutex;
+    std::condition_variable_any consumer_cv;
+    std::mutex consumer_cv_mutex;
+};
+
+template <typename T, size_t Capacity = detail::DefaultCapacity>
+class MPSCQueue {
+public:
     template <typename... Args>
-    void emplace(Args&&... args) noexcept {
-        static_assert(std::is_nothrow_constructible_v<T, Args&&...>,
-                      "T must be nothrow constructible with Args&&...");
-        auto const head = head_.fetch_add(1);
-        auto& slot = slots[idx(head)];
-        slot.turn.wait(true);
-        slot.construct(std::forward<Args>(args)...);
-        slot.turn.test_and_set();
-        cv.notify_one();
+    bool TryEmplace(Args&&... args) {
+        std::scoped_lock lock{write_mutex};
+        return spsc_queue.TryEmplace(std::forward<Args>(args)...);
     }
 
-    constexpr size_t idx(size_t i) const noexcept {
-        return i & mask;
+    template <typename... Args>
+    void EmplaceWait(Args&&... args) {
+        std::scoped_lock lock{write_mutex};
+        spsc_queue.EmplaceWait(std::forward<Args>(args)...);
     }
 
-    static constexpr size_t mask = capacity - 1;
+    bool TryPop(T& t) {
+        return spsc_queue.TryPop(t);
+    }
 
-    // Align to avoid false sharing between head_ and tail_
-    alignas(hardware_interference_size) std::atomic<size_t> head_{0};
-    alignas(hardware_interference_size) std::atomic<size_t> tail_{0};
+    void PopWait(T& t) {
+        spsc_queue.PopWait(t);
+    }
 
-    std::mutex cv_mutex;
-    std::condition_variable_any cv;
+    void PopWait(T& t, std::stop_token stop_token) {
+        spsc_queue.PopWait(t, stop_token);
+    }
 
-    Slot<T>* slots;
-    [[no_unique_address]] std::allocator<Slot<T>> allocator;
+    T PopWait() {
+        return spsc_queue.PopWait();
+    }
 
-    static_assert(std::is_nothrow_copy_assignable_v<T> || std::is_nothrow_move_assignable_v<T>,
-                  "T must be nothrow copy or move assignable");
+    T PopWait(std::stop_token stop_token) {
+        return spsc_queue.PopWait(stop_token);
+    }
 
-    static_assert(std::is_nothrow_destructible_v<T>, "T must be nothrow destructible");
+private:
+    SPSCQueue<T, Capacity> spsc_queue;
+    std::mutex write_mutex;
+};
+
+template <typename T, size_t Capacity = detail::DefaultCapacity>
+class MPMCQueue {
+public:
+    template <typename... Args>
+    bool TryEmplace(Args&&... args) {
+        std::scoped_lock lock{write_mutex};
+        return spsc_queue.TryEmplace(std::forward<Args>(args)...);
+    }
+
+    template <typename... Args>
+    void EmplaceWait(Args&&... args) {
+        std::scoped_lock lock{write_mutex};
+        spsc_queue.EmplaceWait(std::forward<Args>(args)...);
+    }
+
+    bool TryPop(T& t) {
+        std::scoped_lock lock{read_mutex};
+        return spsc_queue.TryPop(t);
+    }
+
+    void PopWait(T& t) {
+        std::scoped_lock lock{read_mutex};
+        spsc_queue.PopWait(t);
+    }
+
+    void PopWait(T& t, std::stop_token stop_token) {
+        std::scoped_lock lock{read_mutex};
+        spsc_queue.PopWait(t, stop_token);
+    }
+
+    T PopWait() {
+        std::scoped_lock lock{read_mutex};
+        return spsc_queue.PopWait();
+    }
+
+    T PopWait(std::stop_token stop_token) {
+        std::scoped_lock lock{read_mutex};
+        return spsc_queue.PopWait(stop_token);
+    }
+
+private:
+    SPSCQueue<T, Capacity> spsc_queue;
+    std::mutex write_mutex;
+    std::mutex read_mutex;
 };
 
 } // namespace Common

src/common/container_hash.h

@@ -0,0 +1,92 @@
+// SPDX-FileCopyrightText: 2005-2014 Daniel James
+// SPDX-FileCopyrightText: 2016 Austin Appleby
+// SPDX-License-Identifier: BSL-1.0
+
+#include <array>
+#include <climits>
+#include <cstdint>
+#include <limits>
+#include <type_traits>
+#include <vector>
+
+namespace Common {
+
+namespace detail {
+
+template <typename T>
+    requires std::is_unsigned_v<T>
+inline std::size_t HashValue(T val) {
+    const unsigned int size_t_bits = std::numeric_limits<std::size_t>::digits;
+    const unsigned int length =
+        (std::numeric_limits<T>::digits - 1) / static_cast<unsigned int>(size_t_bits);
+
+    std::size_t seed = 0;
+
+    for (unsigned int i = length * size_t_bits; i > 0; i -= size_t_bits) {
+        seed ^= static_cast<size_t>(val >> i) + (seed << 6) + (seed >> 2);
+    }
+
+    seed ^= static_cast<size_t>(val) + (seed << 6) + (seed >> 2);
+
+    return seed;
+}
+
+template <size_t Bits>
+struct HashCombineImpl {
+    template <typename T>
+    static inline T fn(T seed, T value) {
+        seed ^= value + 0x9e3779b9 + (seed << 6) + (seed >> 2);
+        return seed;
+    }
+};
+
+template <>
+struct HashCombineImpl<64> {
+    static inline std::uint64_t fn(std::uint64_t h, std::uint64_t k) {
+        const std::uint64_t m = (std::uint64_t(0xc6a4a793) << 32) + 0x5bd1e995;
+        const int r = 47;
+
+        k *= m;
+        k ^= k >> r;
+        k *= m;
+
+        h ^= k;
+        h *= m;
+
+        // Completely arbitrary number, to prevent 0's
+        // from hashing to 0.
+        h += 0xe6546b64;
+
+        return h;
+    }
+};
+
+} // namespace detail
+
+template <typename T>
+inline void HashCombine(std::size_t& seed, const T& v) {
+    seed = detail::HashCombineImpl<sizeof(std::size_t) * CHAR_BIT>::fn(seed, detail::HashValue(v));
+}
+
+template <typename It>
+inline std::size_t HashRange(It first, It last) {
+    std::size_t seed = 0;
+
+    for (; first != last; ++first) {
+        HashCombine<typename std::iterator_traits<It>::value_type>(seed, *first);
+    }
+
+    return seed;
+}
+
+template <typename T, size_t Size>
+std::size_t HashValue(const std::array<T, Size>& v) {
+    return HashRange(v.cbegin(), v.cend());
+}
+
+template <typename T, typename Allocator>
+std::size_t HashValue(const std::vector<T, Allocator>& v) {
+    return HashRange(v.cbegin(), v.cend());
+}
+
+} // namespace Common

src/common/intrusive_list.h

@@ -0,0 +1,631 @@
+// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "common/common_funcs.h"
+#include "common/parent_of_member.h"
+
+namespace Common {
+
+// Forward declare implementation class for Node.
+namespace impl {
+
+class IntrusiveListImpl;
+
+}
+
+class IntrusiveListNode {
+    YUZU_NON_COPYABLE(IntrusiveListNode);
+
+private:
+    friend class impl::IntrusiveListImpl;
+
+    IntrusiveListNode* m_prev;
+    IntrusiveListNode* m_next;
+
+public:
+    constexpr IntrusiveListNode() : m_prev(this), m_next(this) {}
+
+    constexpr bool IsLinked() const {
+        return m_next != this;
+    }
+
+private:
+    constexpr void LinkPrev(IntrusiveListNode* node) {
+        // We can't link an already linked node.
+        ASSERT(!node->IsLinked());
+        this->SplicePrev(node, node);
+    }
+
+    constexpr void SplicePrev(IntrusiveListNode* first, IntrusiveListNode* last) {
+        // Splice a range into the list.
+        auto last_prev = last->m_prev;
+        first->m_prev = m_prev;
+        last_prev->m_next = this;
+        m_prev->m_next = first;
+        m_prev = last_prev;
+    }
+
+    constexpr void LinkNext(IntrusiveListNode* node) {
+        // We can't link an already linked node.
+        ASSERT(!node->IsLinked());
+        return this->SpliceNext(node, node);
+    }
+
+    constexpr void SpliceNext(IntrusiveListNode* first, IntrusiveListNode* last) {
+        // Splice a range into the list.
+        auto last_prev = last->m_prev;
+        first->m_prev = this;
+        last_prev->m_next = m_next;
+        m_next->m_prev = last_prev;
+        m_next = first;
+    }
+
+    constexpr void Unlink() {
+        this->Unlink(m_next);
+    }
+
+    constexpr void Unlink(IntrusiveListNode* last) {
+        // Unlink a node from a next node.
+        auto last_prev = last->m_prev;
+        m_prev->m_next = last;
+        last->m_prev = m_prev;
+        last_prev->m_next = this;
+        m_prev = last_prev;
+    }
+
+    constexpr IntrusiveListNode* GetPrev() {
+        return m_prev;
+    }
+
+    constexpr const IntrusiveListNode* GetPrev() const {
+        return m_prev;
+    }
+
+    constexpr IntrusiveListNode* GetNext() {
+        return m_next;
+    }
+
+    constexpr const IntrusiveListNode* GetNext() const {
+        return m_next;
+    }
+};
+// DEPRECATED: static_assert(std::is_literal_type<IntrusiveListNode>::value);
+
+namespace impl {
+
+class IntrusiveListImpl {
+    YUZU_NON_COPYABLE(IntrusiveListImpl);
+
+private:
+    IntrusiveListNode m_root_node;
+
+public:
+    template <bool Const>
+    class Iterator;
+
+    using value_type = IntrusiveListNode;
+    using size_type = size_t;
+    using difference_type = ptrdiff_t;
+    using pointer = value_type*;
+    using const_pointer = const value_type*;
+    using reference = value_type&;
+    using const_reference = const value_type&;
+    using iterator = Iterator<false>;
+    using const_iterator = Iterator<true>;
+    using reverse_iterator = std::reverse_iterator<iterator>;
+    using const_reverse_iterator = std::reverse_iterator<const_iterator>;
+
+    template <bool Const>
+    class Iterator {
+    public:
+        using iterator_category = std::bidirectional_iterator_tag;
+        using value_type = typename IntrusiveListImpl::value_type;
+        using difference_type = typename IntrusiveListImpl::difference_type;
+        using pointer =
+            std::conditional_t<Const, IntrusiveListImpl::const_pointer, IntrusiveListImpl::pointer>;
+        using reference = std::conditional_t<Const, IntrusiveListImpl::const_reference,
+                                             IntrusiveListImpl::reference>;
+
+    private:
+        pointer m_node;
+
+    public:
+        constexpr explicit Iterator(pointer n) : m_node(n) {}
+
+        constexpr bool operator==(const Iterator& rhs) const {
+            return m_node == rhs.m_node;
+        }
+
+        constexpr pointer operator->() const {
+            return m_node;
+        }
+
+        constexpr reference operator*() const {
+            return *m_node;
+        }
+
+        constexpr Iterator& operator++() {
+            m_node = m_node->m_next;
+            return *this;
+        }
+
+        constexpr Iterator& operator--() {
+            m_node = m_node->m_prev;
+            return *this;
+        }
+
+        constexpr Iterator operator++(int) {
+            const Iterator it{*this};
+            ++(*this);
+            return it;
+        }
+
+        constexpr Iterator operator--(int) {
+            const Iterator it{*this};
+            --(*this);
+            return it;
+        }
+
+        constexpr operator Iterator<true>() const {
+            return Iterator<true>(m_node);
+        }
+
+        constexpr Iterator<false> GetNonConstIterator() const {
+            return Iterator<false>(const_cast<IntrusiveListImpl::pointer>(m_node));
+        }
+    };
+
+public:
+    constexpr IntrusiveListImpl() : m_root_node() {}
+
+    // Iterator accessors.
+    constexpr iterator begin() {
+        return iterator(m_root_node.GetNext());
+    }
+
+    constexpr const_iterator begin() const {
+        return const_iterator(m_root_node.GetNext());
+    }
+
+    constexpr iterator end() {
+        return iterator(std::addressof(m_root_node));
+    }
+
+    constexpr const_iterator end() const {
+        return const_iterator(std::addressof(m_root_node));
+    }
+
+    constexpr iterator iterator_to(reference v) {
+        // Only allow iterator_to for values in lists.
+        ASSERT(v.IsLinked());
+        return iterator(std::addressof(v));
+    }
+
+    constexpr const_iterator iterator_to(const_reference v) const {
+        // Only allow iterator_to for values in lists.
+        ASSERT(v.IsLinked());
+        return const_iterator(std::addressof(v));
+    }
+
+    // Content management.
+    constexpr bool empty() const {
+        return !m_root_node.IsLinked();
+    }
+
+    constexpr size_type size() const {
+        return static_cast<size_type>(std::distance(this->begin(), this->end()));
+    }
+
+    constexpr reference back() {
+        return *m_root_node.GetPrev();
+    }
+
+    constexpr const_reference back() const {
+        return *m_root_node.GetPrev();
+    }
+
+    constexpr reference front() {
+        return *m_root_node.GetNext();
+    }
+
+    constexpr const_reference front() const {
+        return *m_root_node.GetNext();
+    }
+
+    constexpr void push_back(reference node) {
+        m_root_node.LinkPrev(std::addressof(node));
+    }
+
+    constexpr void push_front(reference node) {
+        m_root_node.LinkNext(std::addressof(node));
+    }
+
+    constexpr void pop_back() {
+        m_root_node.GetPrev()->Unlink();
+    }
+
+    constexpr void pop_front() {
+        m_root_node.GetNext()->Unlink();
+    }
+
+    constexpr iterator insert(const_iterator pos, reference node) {
+        pos.GetNonConstIterator()->LinkPrev(std::addressof(node));
+        return iterator(std::addressof(node));
+    }
+
+    constexpr void splice(const_iterator pos, IntrusiveListImpl& o) {
+        splice_impl(pos, o.begin(), o.end());
+    }
+
+    constexpr void splice(const_iterator pos, IntrusiveListImpl& o, const_iterator first) {
+        const_iterator last(first);
+        std::advance(last, 1);
+        splice_impl(pos, first, last);
+    }
+
+    constexpr void splice(const_iterator pos, IntrusiveListImpl& o, const_iterator first,
+                          const_iterator last) {
+        splice_impl(pos, first, last);
+    }
+
+    constexpr iterator erase(const_iterator pos) {
+        if (pos == this->end()) {
+            return this->end();
+        }
+        iterator it(pos.GetNonConstIterator());
+        (it++)->Unlink();
+        return it;
+    }
+
+    constexpr void clear() {
+        while (!this->empty()) {
+            this->pop_front();
+        }
+    }
+
+private:
+    constexpr void splice_impl(const_iterator _pos, const_iterator _first, const_iterator _last) {
+        if (_first == _last) {
+            return;
+        }
+        iterator pos(_pos.GetNonConstIterator());
+        iterator first(_first.GetNonConstIterator());
+        iterator last(_last.GetNonConstIterator());
+        first->Unlink(std::addressof(*last));
+        pos->SplicePrev(std::addressof(*first), std::addressof(*first));
+    }
+};
+
+} // namespace impl
+
+template <class T, class Traits>
+class IntrusiveList {
+    YUZU_NON_COPYABLE(IntrusiveList);
+
+private:
+    impl::IntrusiveListImpl m_impl;
+
+public:
+    template <bool Const>
+    class Iterator;
+
+    using value_type = T;
+    using size_type = size_t;
+    using difference_type = ptrdiff_t;
+    using pointer = value_type*;
+    using const_pointer = const value_type*;
+    using reference = value_type&;
+    using const_reference = const value_type&;
+    using iterator = Iterator<false>;
+    using const_iterator = Iterator<true>;
+    using reverse_iterator = std::reverse_iterator<iterator>;
+    using const_reverse_iterator = std::reverse_iterator<const_iterator>;
+
+    template <bool Const>
+    class Iterator {
+    public:
+        friend class Common::IntrusiveList<T, Traits>;
+
+        using ImplIterator =
+            std::conditional_t<Const, Common::impl::IntrusiveListImpl::const_iterator,
+                               Common::impl::IntrusiveListImpl::iterator>;
+
+        using iterator_category = std::bidirectional_iterator_tag;
+        using value_type = typename IntrusiveList::value_type;
+        using difference_type = typename IntrusiveList::difference_type;
+        using pointer =
+            std::conditional_t<Const, IntrusiveList::const_pointer, IntrusiveList::pointer>;
+        using reference =
+            std::conditional_t<Const, IntrusiveList::const_reference, IntrusiveList::reference>;
+
+    private:
+        ImplIterator m_iterator;
+
+    private:
+        constexpr explicit Iterator(ImplIterator it) : m_iterator(it) {}
+
+        constexpr ImplIterator GetImplIterator() const {
+            return m_iterator;
+        }
+
+    public:
+        constexpr bool operator==(const Iterator& rhs) const {
+            return m_iterator == rhs.m_iterator;
+        }
+
+        constexpr pointer operator->() const {
+            return std::addressof(Traits::GetParent(*m_iterator));
+        }
+
+        constexpr reference operator*() const {
+            return Traits::GetParent(*m_iterator);
+        }
+
+        constexpr Iterator& operator++() {
+            ++m_iterator;
+            return *this;
+        }
+
+        constexpr Iterator& operator--() {
+            --m_iterator;
+            return *this;
+        }
+
+        constexpr Iterator operator++(int) {
+            const Iterator it{*this};
+            ++m_iterator;
+            return it;
+        }
+
+        constexpr Iterator operator--(int) {
+            const Iterator it{*this};
+            --m_iterator;
+            return it;
+        }
+
+        constexpr operator Iterator<true>() const {
+            return Iterator<true>(m_iterator);
+        }
+    };
+
+private:
+    static constexpr IntrusiveListNode& GetNode(reference ref) {
+        return Traits::GetNode(ref);
+    }
+
+    static constexpr IntrusiveListNode const& GetNode(const_reference ref) {
+        return Traits::GetNode(ref);
+    }
+
+    static constexpr reference GetParent(IntrusiveListNode& node) {
+        return Traits::GetParent(node);
+    }
+
+    static constexpr const_reference GetParent(IntrusiveListNode const& node) {
+        return Traits::GetParent(node);
+    }
+
+public:
+    constexpr IntrusiveList() : m_impl() {}
+
+    // Iterator accessors.
+    constexpr iterator begin() {
+        return iterator(m_impl.begin());
+    }
+
+    constexpr const_iterator begin() const {
+        return const_iterator(m_impl.begin());
+    }
+
+    constexpr iterator end() {
+        return iterator(m_impl.end());
+    }
+
+    constexpr const_iterator end() const {
+        return const_iterator(m_impl.end());
+    }
+
+    constexpr const_iterator cbegin() const {
+        return this->begin();
+    }
+
+    constexpr const_iterator cend() const {
+        return this->end();
+    }
+
+    constexpr reverse_iterator rbegin() {
+        return reverse_iterator(this->end());
+    }
+
+    constexpr const_reverse_iterator rbegin() const {
+        return const_reverse_iterator(this->end());
+    }
+
+    constexpr reverse_iterator rend() {
+        return reverse_iterator(this->begin());
+    }
+
+    constexpr const_reverse_iterator rend() const {
+        return const_reverse_iterator(this->begin());
+    }
+
+    constexpr const_reverse_iterator crbegin() const {
+        return this->rbegin();
+    }
+
+    constexpr const_reverse_iterator crend() const {
+        return this->rend();
+    }
+
+    constexpr iterator iterator_to(reference v) {
+        return iterator(m_impl.iterator_to(GetNode(v)));
+    }
+
+    constexpr const_iterator iterator_to(const_reference v) const {
+        return const_iterator(m_impl.iterator_to(GetNode(v)));
+    }
+
+    // Content management.
+    constexpr bool empty() const {
+        return m_impl.empty();
+    }
+
+    constexpr size_type size() const {
+        return m_impl.size();
+    }
+
+    constexpr reference back() {
+        return GetParent(m_impl.back());
+    }
+
+    constexpr const_reference back() const {
+        return GetParent(m_impl.back());
+    }
+
+    constexpr reference front() {
+        return GetParent(m_impl.front());
+    }
+
+    constexpr const_reference front() const {
+        return GetParent(m_impl.front());
+    }
+
+    constexpr void push_back(reference ref) {
+        m_impl.push_back(GetNode(ref));
+    }
+
+    constexpr void push_front(reference ref) {
+        m_impl.push_front(GetNode(ref));
+    }
+
+    constexpr void pop_back() {
+        m_impl.pop_back();
+    }
+
+    constexpr void pop_front() {
+        m_impl.pop_front();
+    }
+
+    constexpr iterator insert(const_iterator pos, reference ref) {
+        return iterator(m_impl.insert(pos.GetImplIterator(), GetNode(ref)));
+    }
+
+    constexpr void splice(const_iterator pos, IntrusiveList& o) {
+        m_impl.splice(pos.GetImplIterator(), o.m_impl);
+    }
+
+    constexpr void splice(const_iterator pos, IntrusiveList& o, const_iterator first) {
+        m_impl.splice(pos.GetImplIterator(), o.m_impl, first.GetImplIterator());
+    }
+
+    constexpr void splice(const_iterator pos, IntrusiveList& o, const_iterator first,
+                          const_iterator last) {
+        m_impl.splice(pos.GetImplIterator(), o.m_impl, first.GetImplIterator(),
+                      last.GetImplIterator());
+    }
+
+    constexpr iterator erase(const_iterator pos) {
+        return iterator(m_impl.erase(pos.GetImplIterator()));
+    }
+
+    constexpr void clear() {
+        m_impl.clear();
+    }
+};
+
+template <auto T, class Derived = Common::impl::GetParentType<T>>
+class IntrusiveListMemberTraits;
+
+template <class Parent, IntrusiveListNode Parent::*Member, class Derived>
+class IntrusiveListMemberTraits<Member, Derived> {
+public:
+    using ListType = IntrusiveList<Derived, IntrusiveListMemberTraits>;
+
+private:
+    friend class IntrusiveList<Derived, IntrusiveListMemberTraits>;
+
+    static constexpr IntrusiveListNode& GetNode(Derived& parent) {
+        return parent.*Member;
+    }
+
+    static constexpr IntrusiveListNode const& GetNode(Derived const& parent) {
+        return parent.*Member;
+    }
+
+    static Derived& GetParent(IntrusiveListNode& node) {
+        return Common::GetParentReference<Member, Derived>(std::addressof(node));
+    }
+
+    static Derived const& GetParent(IntrusiveListNode const& node) {
+        return Common::GetParentReference<Member, Derived>(std::addressof(node));
+    }
+};
+
+template <auto T, class Derived = Common::impl::GetParentType<T>>
+class IntrusiveListMemberTraitsByNonConstexprOffsetOf;
+
+template <class Parent, IntrusiveListNode Parent::*Member, class Derived>
+class IntrusiveListMemberTraitsByNonConstexprOffsetOf<Member, Derived> {
+public:
+    using ListType = IntrusiveList<Derived, IntrusiveListMemberTraitsByNonConstexprOffsetOf>;
+
+private:
+    friend class IntrusiveList<Derived, IntrusiveListMemberTraitsByNonConstexprOffsetOf>;
+
+    static constexpr IntrusiveListNode& GetNode(Derived& parent) {
+        return parent.*Member;
+    }
+
+    static constexpr IntrusiveListNode const& GetNode(Derived const& parent) {
+        return parent.*Member;
+    }
+
+    static Derived& GetParent(IntrusiveListNode& node) {
+        return *reinterpret_cast<Derived*>(reinterpret_cast<char*>(std::addressof(node)) -
+                                           GetOffset());
+    }
+
+    static Derived const& GetParent(IntrusiveListNode const& node) {
+        return *reinterpret_cast<const Derived*>(
+            reinterpret_cast<const char*>(std::addressof(node)) - GetOffset());
+    }
+
+    static uintptr_t GetOffset() {
+        return reinterpret_cast<uintptr_t>(std::addressof(reinterpret_cast<Derived*>(0)->*Member));
+    }
+};
+
+template <class Derived>
+class IntrusiveListBaseNode : public IntrusiveListNode {};
+
+template <class Derived>
+class IntrusiveListBaseTraits {
+public:
+    using ListType = IntrusiveList<Derived, IntrusiveListBaseTraits>;
+
+private:
+    friend class IntrusiveList<Derived, IntrusiveListBaseTraits>;
+
+    static constexpr IntrusiveListNode& GetNode(Derived& parent) {
+        return static_cast<IntrusiveListNode&>(
+            static_cast<IntrusiveListBaseNode<Derived>&>(parent));
+    }
+
+    static constexpr IntrusiveListNode const& GetNode(Derived const& parent) {
+        return static_cast<const IntrusiveListNode&>(
+            static_cast<const IntrusiveListBaseNode<Derived>&>(parent));
+    }
+
+    static constexpr Derived& GetParent(IntrusiveListNode& node) {
+        return static_cast<Derived&>(static_cast<IntrusiveListBaseNode<Derived>&>(node));
+    }
+
+    static constexpr Derived const& GetParent(IntrusiveListNode const& node) {
+        return static_cast<const Derived&>(
+            static_cast<const IntrusiveListBaseNode<Derived>&>(node));
+    }
+};
+
+} // namespace Common

src/common/intrusive_red_black_tree.h

@@ -96,10 +96,6 @@ public:
             return m_node == rhs.m_node;
         }
 
-        constexpr bool operator!=(const Iterator& rhs) const {
-            return !(*this == rhs);
-        }
-
         constexpr pointer operator->() const {
             return m_node;
         }
@@ -324,10 +320,6 @@ public:
             return m_impl == rhs.m_impl;
         }
 
-        constexpr bool operator!=(const Iterator& rhs) const {
-            return !(*this == rhs);
-        }
-
         constexpr pointer operator->() const {
             return Traits::GetParent(std::addressof(*m_impl));
         }

src/common/logging/backend.cpp

@@ -28,7 +28,7 @@
 #ifdef _WIN32
 #include "common/string_util.h"
 #endif
-#include "common/threadsafe_queue.h"
+#include "common/bounded_threadsafe_queue.h"
 
 namespace Common::Log {
 
@@ -204,11 +204,11 @@ public:
 
     void PushEntry(Class log_class, Level log_level, const char* filename, unsigned int line_num,
                    const char* function, std::string&& message) {
-        if (!filter.CheckMessage(log_class, log_level))
+        if (!filter.CheckMessage(log_class, log_level)) {
             return;
-        const Entry& entry =
-            CreateEntry(log_class, log_level, filename, line_num, function, std::move(message));
-        message_queue.Push(entry);
+        }
+        message_queue.EmplaceWait(
+            CreateEntry(log_class, log_level, filename, line_num, function, std::move(message)));
     }
 
 private:
@@ -225,7 +225,7 @@ private:
                 ForEachBackend([&entry](Backend& backend) { backend.Write(entry); });
             };
             while (!stop_token.stop_requested()) {
-                entry = message_queue.PopWait(stop_token);
+                message_queue.PopWait(entry, stop_token);
                 if (entry.filename != nullptr) {
                     write_logs();
                 }
@@ -233,7 +233,7 @@ private:
             // Drain the logging queue. Only writes out up to MAX_LOGS_TO_WRITE to prevent a
             // case where a system is repeatedly spamming logs even on close.
             int max_logs_to_write = filter.IsDebug() ? INT_MAX : 100;
-            while (max_logs_to_write-- && message_queue.Pop(entry)) {
+            while (max_logs_to_write-- && message_queue.TryPop(entry)) {
                 write_logs();
             }
         });
@@ -273,7 +273,7 @@ private:
     ColorConsoleBackend color_console_backend{};
     FileBackend file_backend;
 
-    MPSCQueue<Entry, true> message_queue{};
+    MPSCQueue<Entry> message_queue{};
     std::chrono::steady_clock::time_point time_origin{std::chrono::steady_clock::now()};
     std::jthread backend_thread;
 };

src/common/range_map.h

@@ -38,12 +38,12 @@ public:
         Map(address, address_end, null_value);
     }
 
-    [[nodiscard]] size_t GetContinousSizeFrom(KeyTBase address) const {
+    [[nodiscard]] size_t GetContinuousSizeFrom(KeyTBase address) const {
         const KeyT new_address = static_cast<KeyT>(address);
         if (new_address < 0) {
             return 0;
         }
-        return ContinousSizeInternal(new_address);
+        return ContinuousSizeInternal(new_address);
     }
 
     [[nodiscard]] ValueT GetValueAt(KeyT address) const {
@@ -59,7 +59,7 @@ private:
     using IteratorType = typename MapType::iterator;
     using ConstIteratorType = typename MapType::const_iterator;
 
-    size_t ContinousSizeInternal(KeyT address) const {
+    size_t ContinuousSizeInternal(KeyT address) const {
         const auto it = GetFirstElementBeforeOrOn(address);
         if (it == container.end() || it->second == null_value) {
             return 0;

src/common/settings.cpp

@@ -45,6 +45,7 @@ void LogSettings() {
     log_setting("System_LanguageIndex", values.language_index.GetValue());
     log_setting("System_RegionIndex", values.region_index.GetValue());
     log_setting("System_TimeZoneIndex", values.time_zone_index.GetValue());
+    log_setting("System_UnsafeMemoryLayout", values.use_unsafe_extended_memory_layout.GetValue());
     log_setting("Core_UseMultiCore", values.use_multi_core.GetValue());
     log_setting("CPU_Accuracy", values.cpu_accuracy.GetValue());
     log_setting("Renderer_UseResolutionScaling", values.resolution_setup.GetValue());
@@ -191,7 +192,7 @@ void RestoreGlobalState(bool is_powered_on) {
 
     // Core
     values.use_multi_core.SetGlobal(true);
-    values.use_extended_memory_layout.SetGlobal(true);
+    values.use_unsafe_extended_memory_layout.SetGlobal(true);
 
     // CPU
     values.cpu_accuracy.SetGlobal(true);
@@ -205,6 +206,7 @@ void RestoreGlobalState(bool is_powered_on) {
     // Renderer
     values.fsr_sharpening_slider.SetGlobal(true);
     values.renderer_backend.SetGlobal(true);
+    values.async_presentation.SetGlobal(true);
     values.renderer_force_max_clock.SetGlobal(true);
     values.vulkan_device.SetGlobal(true);
     values.fullscreen_mode.SetGlobal(true);

src/common/settings.h

@@ -388,7 +388,8 @@ struct Values {
 
     // Core
     SwitchableSetting<bool> use_multi_core{true, "use_multi_core"};
-    SwitchableSetting<bool> use_extended_memory_layout{false, "use_extended_memory_layout"};
+    SwitchableSetting<bool> use_unsafe_extended_memory_layout{false,
+                                                              "use_unsafe_extended_memory_layout"};
 
     // Cpu
     SwitchableSetting<CPUAccuracy, true> cpu_accuracy{CPUAccuracy::Auto, CPUAccuracy::Auto,
@@ -422,6 +423,7 @@ struct Values {
     // Renderer
     SwitchableSetting<RendererBackend, true> renderer_backend{
         RendererBackend::Vulkan, RendererBackend::OpenGL, RendererBackend::Null, "backend"};
+    SwitchableSetting<bool> async_presentation{false, "async_presentation"};
     SwitchableSetting<bool> renderer_force_max_clock{false, "force_max_clock"};
     Setting<bool> renderer_debug{false, "debug"};
     Setting<bool> renderer_shader_feedback{false, "shader_feedback"};

src/common/string_util.cpp

@@ -125,18 +125,18 @@ std::string ReplaceAll(std::string result, const std::string& src, const std::st
     return result;
 }
 
-std::string UTF16ToUTF8(const std::u16string& input) {
+std::string UTF16ToUTF8(std::u16string_view input) {
     std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t> convert;
-    return convert.to_bytes(input);
+    return convert.to_bytes(input.data(), input.data() + input.size());
 }
 
-std::u16string UTF8ToUTF16(const std::string& input) {
+std::u16string UTF8ToUTF16(std::string_view input) {
     std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t> convert;
-    return convert.from_bytes(input);
+    return convert.from_bytes(input.data(), input.data() + input.size());
 }
 
 #ifdef _WIN32
-static std::wstring CPToUTF16(u32 code_page, const std::string& input) {
+static std::wstring CPToUTF16(u32 code_page, std::string_view input) {
     const auto size =
         MultiByteToWideChar(code_page, 0, input.data(), static_cast<int>(input.size()), nullptr, 0);
 
@@ -154,7 +154,7 @@ static std::wstring CPToUTF16(u32 code_page, const std::string& input) {
     return output;
 }
 
-std::string UTF16ToUTF8(const std::wstring& input) {
+std::string UTF16ToUTF8(std::wstring_view input) {
     const auto size = WideCharToMultiByte(CP_UTF8, 0, input.data(), static_cast<int>(input.size()),
                                           nullptr, 0, nullptr, nullptr);
     if (size == 0) {
@@ -172,7 +172,7 @@ std::string UTF16ToUTF8(const std::wstring& input) {
     return output;
 }
 
-std::wstring UTF8ToUTF16W(const std::string& input) {
+std::wstring UTF8ToUTF16W(std::string_view input) {
     return CPToUTF16(CP_UTF8, input);
 }
 

src/common/string_util.h

@@ -36,12 +36,12 @@ bool SplitPath(const std::string& full_path, std::string* _pPath, std::string* _
 [[nodiscard]] std::string ReplaceAll(std::string result, const std::string& src,
                                      const std::string& dest);
 
-[[nodiscard]] std::string UTF16ToUTF8(const std::u16string& input);
-[[nodiscard]] std::u16string UTF8ToUTF16(const std::string& input);
+[[nodiscard]] std::string UTF16ToUTF8(std::u16string_view input);
+[[nodiscard]] std::u16string UTF8ToUTF16(std::string_view input);
 
 #ifdef _WIN32
-[[nodiscard]] std::string UTF16ToUTF8(const std::wstring& input);
-[[nodiscard]] std::wstring UTF8ToUTF16W(const std::string& str);
+[[nodiscard]] std::string UTF16ToUTF8(std::wstring_view input);
+[[nodiscard]] std::wstring UTF8ToUTF16W(std::string_view str);
 
 #endif
 

src/common/telemetry.cpp

@@ -97,6 +97,7 @@ void AppendCPUInfo(FieldCollection& fc) {
     add_field("CPU_Extension_x64_PCLMULQDQ", caps.pclmulqdq);
     add_field("CPU_Extension_x64_POPCNT", caps.popcnt);
     add_field("CPU_Extension_x64_SHA", caps.sha);
+    add_field("CPU_Extension_x64_WAITPKG", caps.waitpkg);
 #else
     fc.AddField(FieldType::UserSystem, "CPU_Model", "Other");
 #endif

src/common/typed_address.h

@@ -0,0 +1,315 @@
+// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <compare>
+#include <type_traits>
+#include <fmt/format.h>
+
+#include "common/common_types.h"
+
+namespace Common {
+
+template <bool Virtual, typename T>
+class TypedAddress {
+public:
+    // Constructors.
+    constexpr inline TypedAddress() : m_address(0) {}
+    constexpr inline TypedAddress(uint64_t a) : m_address(a) {}
+
+    template <typename U>
+    constexpr inline explicit TypedAddress(const U* ptr)
+        : m_address(reinterpret_cast<uint64_t>(ptr)) {}
+
+    // Copy constructor.
+    constexpr inline TypedAddress(const TypedAddress& rhs) = default;
+
+    // Assignment operator.
+    constexpr inline TypedAddress& operator=(const TypedAddress& rhs) = default;
+
+    // Arithmetic operators.
+    template <typename I>
+    constexpr inline TypedAddress operator+(I rhs) const {
+        static_assert(std::is_integral_v<I>);
+        return m_address + rhs;
+    }
+
+    constexpr inline TypedAddress operator+(TypedAddress rhs) const {
+        return m_address + rhs.m_address;
+    }
+
+    constexpr inline TypedAddress operator++() {
+        return ++m_address;
+    }
+
+    constexpr inline TypedAddress operator++(int) {
+        return m_address++;
+    }
+
+    template <typename I>
+    constexpr inline TypedAddress operator-(I rhs) const {
+        static_assert(std::is_integral_v<I>);
+        return m_address - rhs;
+    }
+
+    constexpr inline ptrdiff_t operator-(TypedAddress rhs) const {
+        return m_address - rhs.m_address;
+    }
+
+    constexpr inline TypedAddress operator--() {
+        return --m_address;
+    }
+
+    constexpr inline TypedAddress operator--(int) {
+        return m_address--;
+    }
+
+    template <typename I>
+    constexpr inline TypedAddress operator+=(I rhs) {
+        static_assert(std::is_integral_v<I>);
+        m_address += rhs;
+        return *this;
+    }
+
+    template <typename I>
+    constexpr inline TypedAddress operator-=(I rhs) {
+        static_assert(std::is_integral_v<I>);
+        m_address -= rhs;
+        return *this;
+    }
+
+    // Logical operators.
+    constexpr inline uint64_t operator&(uint64_t mask) const {
+        return m_address & mask;
+    }
+
+    constexpr inline uint64_t operator|(uint64_t mask) const {
+        return m_address | mask;
+    }
+
+    template <typename I>
+    constexpr inline TypedAddress operator|=(I rhs) {
+        static_assert(std::is_integral_v<I>);
+        m_address |= rhs;
+        return *this;
+    }
+
+    constexpr inline uint64_t operator<<(int shift) const {
+        return m_address << shift;
+    }
+
+    constexpr inline uint64_t operator>>(int shift) const {
+        return m_address >> shift;
+    }
+
+    template <typename U>
+    constexpr inline size_t operator/(U size) const {
+        return m_address / size;
+    }
+
+    constexpr explicit operator bool() const {
+        return m_address != 0;
+    }
+
+    // constexpr inline uint64_t operator%(U align) const { return m_address % align; }
+
+    // Comparison operators.
+    constexpr bool operator==(const TypedAddress&) const = default;
+    constexpr auto operator<=>(const TypedAddress&) const = default;
+
+    // For convenience, also define comparison operators versus uint64_t.
+    constexpr inline bool operator==(uint64_t rhs) const {
+        return m_address == rhs;
+    }
+
+    // Allow getting the address explicitly, for use in accessors.
+    constexpr inline uint64_t GetValue() const {
+        return m_address;
+    }
+
+private:
+    uint64_t m_address{};
+};
+
+struct PhysicalAddressTag {};
+struct VirtualAddressTag {};
+struct ProcessAddressTag {};
+
+using PhysicalAddress = TypedAddress<false, PhysicalAddressTag>;
+using VirtualAddress = TypedAddress<true, VirtualAddressTag>;
+using ProcessAddress = TypedAddress<true, ProcessAddressTag>;
+
+// Define accessors.
+template <typename T>
+concept IsTypedAddress = std::same_as<T, PhysicalAddress> || std::same_as<T, VirtualAddress> ||
+                         std::same_as<T, ProcessAddress>;
+
+template <typename T>
+constexpr inline T Null = [] {
+    if constexpr (std::is_same<T, uint64_t>::value) {
+        return 0;
+    } else {
+        static_assert(std::is_same<T, PhysicalAddress>::value ||
+                      std::is_same<T, VirtualAddress>::value ||
+                      std::is_same<T, ProcessAddress>::value);
+        return T(0);
+    }
+}();
+
+// Basic type validations.
+static_assert(sizeof(PhysicalAddress) == sizeof(uint64_t));
+static_assert(sizeof(VirtualAddress) == sizeof(uint64_t));
+static_assert(sizeof(ProcessAddress) == sizeof(uint64_t));
+
+static_assert(std::is_trivially_copyable_v<PhysicalAddress>);
+static_assert(std::is_trivially_copyable_v<VirtualAddress>);
+static_assert(std::is_trivially_copyable_v<ProcessAddress>);
+
+static_assert(std::is_trivially_copy_constructible_v<PhysicalAddress>);
+static_assert(std::is_trivially_copy_constructible_v<VirtualAddress>);
+static_assert(std::is_trivially_copy_constructible_v<ProcessAddress>);
+
+static_assert(std::is_trivially_move_constructible_v<PhysicalAddress>);
+static_assert(std::is_trivially_move_constructible_v<VirtualAddress>);
+static_assert(std::is_trivially_move_constructible_v<ProcessAddress>);
+
+static_assert(std::is_trivially_copy_assignable_v<PhysicalAddress>);
+static_assert(std::is_trivially_copy_assignable_v<VirtualAddress>);
+static_assert(std::is_trivially_copy_assignable_v<ProcessAddress>);
+
+static_assert(std::is_trivially_move_assignable_v<PhysicalAddress>);
+static_assert(std::is_trivially_move_assignable_v<VirtualAddress>);
+static_assert(std::is_trivially_move_assignable_v<ProcessAddress>);
+
+static_assert(std::is_trivially_destructible_v<PhysicalAddress>);
+static_assert(std::is_trivially_destructible_v<VirtualAddress>);
+static_assert(std::is_trivially_destructible_v<ProcessAddress>);
+
+static_assert(Null<uint64_t> == 0);
+static_assert(Null<PhysicalAddress> == Null<uint64_t>);
+static_assert(Null<VirtualAddress> == Null<uint64_t>);
+static_assert(Null<ProcessAddress> == Null<uint64_t>);
+
+// Constructor/assignment validations.
+static_assert([] {
+    const PhysicalAddress a(5);
+    PhysicalAddress b(a);
+    return b;
+}() == PhysicalAddress(5));
+static_assert([] {
+    const PhysicalAddress a(5);
+    PhysicalAddress b(10);
+    b = a;
+    return b;
+}() == PhysicalAddress(5));
+
+// Arithmetic validations.
+static_assert(PhysicalAddress(10) + 5 == PhysicalAddress(15));
+static_assert(PhysicalAddress(10) - 5 == PhysicalAddress(5));
+static_assert([] {
+    PhysicalAddress v(10);
+    v += 5;
+    return v;
+}() == PhysicalAddress(15));
+static_assert([] {
+    PhysicalAddress v(10);
+    v -= 5;
+    return v;
+}() == PhysicalAddress(5));
+static_assert(PhysicalAddress(10)++ == PhysicalAddress(10));
+static_assert(++PhysicalAddress(10) == PhysicalAddress(11));
+static_assert(PhysicalAddress(10)-- == PhysicalAddress(10));
+static_assert(--PhysicalAddress(10) == PhysicalAddress(9));
+
+// Logical validations.
+static_assert((PhysicalAddress(0b11111111) >> 1) == 0b01111111);
+static_assert((PhysicalAddress(0b10101010) >> 1) == 0b01010101);
+static_assert((PhysicalAddress(0b11111111) << 1) == 0b111111110);
+static_assert((PhysicalAddress(0b01010101) << 1) == 0b10101010);
+static_assert((PhysicalAddress(0b11111111) & 0b01010101) == 0b01010101);
+static_assert((PhysicalAddress(0b11111111) & 0b10101010) == 0b10101010);
+static_assert((PhysicalAddress(0b01010101) & 0b10101010) == 0b00000000);
+static_assert((PhysicalAddress(0b00000000) | 0b01010101) == 0b01010101);
+static_assert((PhysicalAddress(0b11111111) | 0b01010101) == 0b11111111);
+static_assert((PhysicalAddress(0b10101010) | 0b01010101) == 0b11111111);
+
+// Comparisons.
+static_assert(PhysicalAddress(0) == PhysicalAddress(0));
+static_assert(PhysicalAddress(0) != PhysicalAddress(1));
+static_assert(PhysicalAddress(0) < PhysicalAddress(1));
+static_assert(PhysicalAddress(0) <= PhysicalAddress(1));
+static_assert(PhysicalAddress(1) > PhysicalAddress(0));
+static_assert(PhysicalAddress(1) >= PhysicalAddress(0));
+
+static_assert(!(PhysicalAddress(0) == PhysicalAddress(1)));
+static_assert(!(PhysicalAddress(0) != PhysicalAddress(0)));
+static_assert(!(PhysicalAddress(1) < PhysicalAddress(0)));
+static_assert(!(PhysicalAddress(1) <= PhysicalAddress(0)));
+static_assert(!(PhysicalAddress(0) > PhysicalAddress(1)));
+static_assert(!(PhysicalAddress(0) >= PhysicalAddress(1)));
+
+} // namespace Common
+
+template <bool Virtual, typename T>
+constexpr inline uint64_t GetInteger(Common::TypedAddress<Virtual, T> address) {
+    return address.GetValue();
+}
+
+template <>
+struct fmt::formatter<Common::PhysicalAddress> {
+    constexpr auto parse(fmt::format_parse_context& ctx) {
+        return ctx.begin();
+    }
+    template <typename FormatContext>
+    auto format(const Common::PhysicalAddress& addr, FormatContext& ctx) {
+        return fmt::format_to(ctx.out(), "{:#x}", static_cast<u64>(addr.GetValue()));
+    }
+};
+
+template <>
+struct fmt::formatter<Common::ProcessAddress> {
+    constexpr auto parse(fmt::format_parse_context& ctx) {
+        return ctx.begin();
+    }
+    template <typename FormatContext>
+    auto format(const Common::ProcessAddress& addr, FormatContext& ctx) {
+        return fmt::format_to(ctx.out(), "{:#x}", static_cast<u64>(addr.GetValue()));
+    }
+};
+
+template <>
+struct fmt::formatter<Common::VirtualAddress> {
+    constexpr auto parse(fmt::format_parse_context& ctx) {
+        return ctx.begin();
+    }
+    template <typename FormatContext>
+    auto format(const Common::VirtualAddress& addr, FormatContext& ctx) {
+        return fmt::format_to(ctx.out(), "{:#x}", static_cast<u64>(addr.GetValue()));
+    }
+};
+
+namespace std {
+
+template <>
+struct hash<Common::PhysicalAddress> {
+    size_t operator()(const Common::PhysicalAddress& k) const noexcept {
+        return k.GetValue();
+    }
+};
+
+template <>
+struct hash<Common::ProcessAddress> {
+    size_t operator()(const Common::ProcessAddress& k) const noexcept {
+        return k.GetValue();
+    }
+};
+
+template <>
+struct hash<Common::VirtualAddress> {
+    size_t operator()(const Common::VirtualAddress& k) const noexcept {
+        return k.GetValue();
+    }
+};
+
+} // namespace std

src/common/x64/cpu_detect.cpp

@@ -144,6 +144,7 @@ static CPUCaps Detect() {
             caps.bmi2 = Common::Bit<8>(cpu_id[1]);
             caps.sha = Common::Bit<29>(cpu_id[1]);
 
+            caps.waitpkg = Common::Bit<5>(cpu_id[2]);
             caps.gfni = Common::Bit<8>(cpu_id[2]);
 
             __cpuidex(cpu_id, 0x00000007, 0x00000001);

src/common/x64/cpu_detect.h

@@ -67,6 +67,7 @@ struct CPUCaps {
     bool pclmulqdq : 1;
     bool popcnt : 1;
     bool sha : 1;
+    bool waitpkg : 1;
 };
 
 /**

src/common/x64/cpu_wait.cpp

@@ -0,0 +1,69 @@
+// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <thread>
+
+#ifdef _MSC_VER
+#include <intrin.h>
+#endif
+
+#include "common/x64/cpu_detect.h"
+#include "common/x64/cpu_wait.h"
+
+namespace Common::X64 {
+
+#ifdef _MSC_VER
+__forceinline static u64 FencedRDTSC() {
+    _mm_lfence();
+    _ReadWriteBarrier();
+    const u64 result = __rdtsc();
+    _mm_lfence();
+    _ReadWriteBarrier();
+    return result;
+}
+
+__forceinline static void TPAUSE() {
+    // 100,000 cycles is a reasonable amount of time to wait to save on CPU resources.
+    // For reference:
+    // At 1 GHz, 100K cycles is 100us
+    // At 2 GHz, 100K cycles is 50us
+    // At 4 GHz, 100K cycles is 25us
+    static constexpr auto PauseCycles = 100'000;
+    _tpause(0, FencedRDTSC() + PauseCycles);
+}
+#else
+static u64 FencedRDTSC() {
+    u64 eax;
+    u64 edx;
+    asm volatile("lfence\n\t"
+                 "rdtsc\n\t"
+                 "lfence\n\t"
+                 : "=a"(eax), "=d"(edx));
+    return (edx << 32) | eax;
+}
+
+static void TPAUSE() {
+    // 100,000 cycles is a reasonable amount of time to wait to save on CPU resources.
+    // For reference:
+    // At 1 GHz, 100K cycles is 100us
+    // At 2 GHz, 100K cycles is 50us
+    // At 4 GHz, 100K cycles is 25us
+    static constexpr auto PauseCycles = 100'000;
+    const auto tsc = FencedRDTSC() + PauseCycles;
+    const auto eax = static_cast<u32>(tsc & 0xFFFFFFFF);
+    const auto edx = static_cast<u32>(tsc >> 32);
+    asm volatile("tpause %0" : : "r"(0), "d"(edx), "a"(eax));
+}
+#endif
+
+void MicroSleep() {
+    static const bool has_waitpkg = GetCPUCaps().waitpkg;
+
+    if (has_waitpkg) {
+        TPAUSE();
+    } else {
+        std::this_thread::yield();
+    }
+}
+
+} // namespace Common::X64

src/common/x64/cpu_wait.h

@@ -0,0 +1,10 @@
+// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+namespace Common::X64 {
+
+void MicroSleep();
+
+} // namespace Common::X64

src/common/x64/native_clock.cpp

@@ -27,16 +27,13 @@ __forceinline static u64 FencedRDTSC() {
 }
 #else
 static u64 FencedRDTSC() {
-    u64 result;
+    u64 eax;
+    u64 edx;
     asm volatile("lfence\n\t"
                  "rdtsc\n\t"
-                 "shl $32, %%rdx\n\t"
-                 "or %%rdx, %0\n\t"
-                 "lfence"
-                 : "=a"(result)
-                 :
-                 : "rdx", "memory", "cc");
-    return result;
+                 "lfence\n\t"
+                 : "=a"(eax), "=d"(edx));
+    return (edx << 32) | eax;
 }
 #endif
 

src/common/zstd_compression.cpp

@@ -33,7 +33,7 @@ std::vector<u8> CompressDataZSTDDefault(const u8* source, std::size_t source_siz
 
 std::vector<u8> DecompressDataZSTD(std::span<const u8> compressed) {
     const std::size_t decompressed_size =
-        ZSTD_getDecompressedSize(compressed.data(), compressed.size());
+        ZSTD_getFrameContentSize(compressed.data(), compressed.size());
     std::vector<u8> decompressed(decompressed_size);
 
     const std::size_t uncompressed_result_size = ZSTD_decompress(

src/core/CMakeLists.txt

@@ -278,6 +278,7 @@ add_library(core STATIC
     hle/kernel/k_trace.h
     hle/kernel/k_transfer_memory.cpp
     hle/kernel/k_transfer_memory.h
+    hle/kernel/k_typed_address.h
     hle/kernel/k_worker_task.h
     hle/kernel/k_worker_task_manager.cpp
     hle/kernel/k_worker_task_manager.h
@@ -569,10 +570,10 @@ add_library(core STATIC
     hle/service/nfp/nfp.h
     hle/service/nfp/nfp_device.cpp
     hle/service/nfp/nfp_device.h
+    hle/service/nfp/nfp_interface.cpp
+    hle/service/nfp/nfp_interface.h
     hle/service/nfp/nfp_result.h
     hle/service/nfp/nfp_types.h
-    hle/service/nfp/nfp_user.cpp
-    hle/service/nfp/nfp_user.h
     hle/service/ngct/ngct.cpp
     hle/service/ngct/ngct.h
     hle/service/nifm/nifm.cpp

src/core/arm/arm_interface.cpp

@@ -44,7 +44,7 @@ void ARM_Interface::SymbolicateBacktrace(Core::System& system, std::vector<Backt
     std::map<std::string, Symbols::Symbols> symbols;
     for (const auto& module : modules) {
         symbols.insert_or_assign(
-            module.second, Symbols::GetSymbols(module.first, system.Memory(),
+            module.second, Symbols::GetSymbols(module.first, system.ApplicationMemory(),
                                                system.ApplicationProcess()->Is64BitProcess()));
     }
 
@@ -168,21 +168,21 @@ void ARM_Interface::LoadWatchpointArray(const WatchpointArray& wp) {
 }
 
 const Kernel::DebugWatchpoint* ARM_Interface::MatchingWatchpoint(
-    VAddr addr, u64 size, Kernel::DebugWatchpointType access_type) const {
+    u64 addr, u64 size, Kernel::DebugWatchpointType access_type) const {
     if (!watchpoints) {
         return nullptr;
     }
 
-    const VAddr start_address{addr};
-    const VAddr end_address{addr + size};
+    const u64 start_address{addr};
+    const u64 end_address{addr + size};
 
     for (size_t i = 0; i < Core::Hardware::NUM_WATCHPOINTS; i++) {
         const auto& watch{(*watchpoints)[i]};
 
-        if (end_address <= watch.start_address) {
+        if (end_address <= GetInteger(watch.start_address)) {
             continue;
         }
-        if (start_address >= watch.end_address) {
+        if (start_address >= GetInteger(watch.end_address)) {
             continue;
         }
         if ((access_type & watch.type) == Kernel::DebugWatchpointType::None) {

src/core/arm/arm_interface.h

@@ -78,7 +78,7 @@ public:
      * @param addr Start address of the cache range to clear
      * @param size Size of the cache range to clear, starting at addr
      */
-    virtual void InvalidateCacheRange(VAddr addr, std::size_t size) = 0;
+    virtual void InvalidateCacheRange(u64 addr, std::size_t size) = 0;
 
     /**
      * Notifies CPU emulation that the current page table has changed.
@@ -149,9 +149,9 @@ public:
      */
     virtual void SetPSTATE(u32 pstate) = 0;
 
-    virtual VAddr GetTlsAddress() const = 0;
+    virtual u64 GetTlsAddress() const = 0;
 
-    virtual void SetTlsAddress(VAddr address) = 0;
+    virtual void SetTlsAddress(u64 address) = 0;
 
     /**
      * Gets the value within the TPIDR_EL0 (read/write software thread ID) register.
@@ -214,7 +214,7 @@ protected:
 
     static void SymbolicateBacktrace(Core::System& system, std::vector<BacktraceEntry>& out);
     const Kernel::DebugWatchpoint* MatchingWatchpoint(
-        VAddr addr, u64 size, Kernel::DebugWatchpointType access_type) const;
+        u64 addr, u64 size, Kernel::DebugWatchpointType access_type) const;
 
     virtual Dynarmic::HaltReason RunJit() = 0;
     virtual Dynarmic::HaltReason StepJit() = 0;

src/core/arm/dynarmic/arm_dynarmic_32.cpp

@@ -5,7 +5,6 @@
 #include <memory>
 #include <dynarmic/interface/A32/a32.h>
 #include <dynarmic/interface/A32/config.h>
-#include <dynarmic/interface/A32/context.h>
 #include "common/assert.h"
 #include "common/literals.h"
 #include "common/logging/log.h"
@@ -28,8 +27,8 @@ using namespace Common::Literals;
 class DynarmicCallbacks32 : public Dynarmic::A32::UserCallbacks {
 public:
     explicit DynarmicCallbacks32(ARM_Dynarmic_32& parent_)
-        : parent{parent_},
-          memory(parent.system.Memory()), debugger_enabled{parent.system.DebuggerEnabled()},
+        : parent{parent_}, memory(parent.system.ApplicationMemory()),
+          debugger_enabled{parent.system.DebuggerEnabled()},
           check_memory_access{debugger_enabled ||
                               !Settings::values.cpuopt_ignore_memory_aborts.GetValue()} {}
 
@@ -155,7 +154,7 @@ public:
         return std::max<s64>(parent.system.CoreTiming().GetDowncount(), 0);
     }
 
-    bool CheckMemoryAccess(VAddr addr, u64 size, Kernel::DebugWatchpointType type) {
+    bool CheckMemoryAccess(u64 addr, u64 size, Kernel::DebugWatchpointType type) {
         if (!check_memory_access) {
             return true;
         }
@@ -397,7 +396,7 @@ u64 ARM_Dynarmic_32::GetTlsAddress() const {
     return cp15->uro;
 }
 
-void ARM_Dynarmic_32::SetTlsAddress(VAddr address) {
+void ARM_Dynarmic_32::SetTlsAddress(u64 address) {
     cp15->uro = static_cast<u32>(address);
 }
 
@@ -410,21 +409,19 @@ void ARM_Dynarmic_32::SetTPIDR_EL0(u64 value) {
 }
 
 void ARM_Dynarmic_32::SaveContext(ThreadContext32& ctx) {
-    Dynarmic::A32::Context context;
-    jit.load()->SaveContext(context);
-    ctx.cpu_registers = context.Regs();
-    ctx.extension_registers = context.ExtRegs();
-    ctx.cpsr = context.Cpsr();
-    ctx.fpscr = context.Fpscr();
+    Dynarmic::A32::Jit* j = jit.load();
+    ctx.cpu_registers = j->Regs();
+    ctx.extension_registers = j->ExtRegs();
+    ctx.cpsr = j->Cpsr();
+    ctx.fpscr = j->Fpscr();
 }
 
 void ARM_Dynarmic_32::LoadContext(const ThreadContext32& ctx) {
-    Dynarmic::A32::Context context;
-    context.Regs() = ctx.cpu_registers;
-    context.ExtRegs() = ctx.extension_registers;
-    context.SetCpsr(ctx.cpsr);
-    context.SetFpscr(ctx.fpscr);
-    jit.load()->LoadContext(context);
+    Dynarmic::A32::Jit* j = jit.load();
+    j->Regs() = ctx.cpu_registers;
+    j->ExtRegs() = ctx.extension_registers;
+    j->SetCpsr(ctx.cpsr);
+    j->SetFpscr(ctx.fpscr);
 }
 
 void ARM_Dynarmic_32::SignalInterrupt() {
@@ -439,7 +436,7 @@ void ARM_Dynarmic_32::ClearInstructionCache() {
     jit.load()->ClearCache();
 }
 
-void ARM_Dynarmic_32::InvalidateCacheRange(VAddr addr, std::size_t size) {
+void ARM_Dynarmic_32::InvalidateCacheRange(u64 addr, std::size_t size) {
     jit.load()->InvalidateCacheRange(static_cast<u32>(addr), size);
 }
 
@@ -468,7 +465,7 @@ void ARM_Dynarmic_32::PageTableChanged(Common::PageTable& page_table,
 std::vector<ARM_Interface::BacktraceEntry> ARM_Dynarmic_32::GetBacktrace(Core::System& system,
                                                                          u64 fp, u64 lr, u64 pc) {
     std::vector<BacktraceEntry> out;
-    auto& memory = system.Memory();
+    auto& memory = system.ApplicationMemory();
 
     out.push_back({"", 0, pc, 0, ""});
 

src/core/arm/dynarmic/arm_dynarmic_32.h

@@ -41,8 +41,8 @@ public:
     void SetVectorReg(int index, u128 value) override;
     u32 GetPSTATE() const override;
     void SetPSTATE(u32 pstate) override;
-    VAddr GetTlsAddress() const override;
-    void SetTlsAddress(VAddr address) override;
+    u64 GetTlsAddress() const override;
+    void SetTlsAddress(u64 address) override;
     void SetTPIDR_EL0(u64 value) override;
     u64 GetTPIDR_EL0() const override;
 
@@ -60,7 +60,7 @@ public:
     void ClearExclusiveState() override;
 
     void ClearInstructionCache() override;
-    void InvalidateCacheRange(VAddr addr, std::size_t size) override;
+    void InvalidateCacheRange(u64 addr, std::size_t size) override;
     void PageTableChanged(Common::PageTable& new_page_table,
                           std::size_t new_address_space_size_in_bits) override;
 

src/core/arm/dynarmic/arm_dynarmic_64.cpp

@@ -28,8 +28,8 @@ using namespace Common::Literals;
 class DynarmicCallbacks64 : public Dynarmic::A64::UserCallbacks {
 public:
     explicit DynarmicCallbacks64(ARM_Dynarmic_64& parent_)
-        : parent{parent_},
-          memory(parent.system.Memory()), debugger_enabled{parent.system.DebuggerEnabled()},
+        : parent{parent_}, memory(parent.system.ApplicationMemory()),
+          debugger_enabled{parent.system.DebuggerEnabled()},
           check_memory_access{debugger_enabled ||
                               !Settings::values.cpuopt_ignore_memory_aborts.GetValue()} {}
 
@@ -117,7 +117,7 @@ public:
     }
 
     void InstructionCacheOperationRaised(Dynarmic::A64::InstructionCacheOperation op,
-                                         VAddr value) override {
+                                         u64 value) override {
         switch (op) {
         case Dynarmic::A64::InstructionCacheOperation::InvalidateByVAToPoU: {
             static constexpr u64 ICACHE_LINE_SIZE = 64;
@@ -199,7 +199,7 @@ public:
         return parent.system.CoreTiming().GetClockTicks();
     }
 
-    bool CheckMemoryAccess(VAddr addr, u64 size, Kernel::DebugWatchpointType type) {
+    bool CheckMemoryAccess(u64 addr, u64 size, Kernel::DebugWatchpointType type) {
         if (!check_memory_access) {
             return true;
         }
@@ -452,7 +452,7 @@ u64 ARM_Dynarmic_64::GetTlsAddress() const {
     return cb->tpidrro_el0;
 }
 
-void ARM_Dynarmic_64::SetTlsAddress(VAddr address) {
+void ARM_Dynarmic_64::SetTlsAddress(u64 address) {
     cb->tpidrro_el0 = address;
 }
 
@@ -500,7 +500,7 @@ void ARM_Dynarmic_64::ClearInstructionCache() {
     jit.load()->ClearCache();
 }
 
-void ARM_Dynarmic_64::InvalidateCacheRange(VAddr addr, std::size_t size) {
+void ARM_Dynarmic_64::InvalidateCacheRange(u64 addr, std::size_t size) {
     jit.load()->InvalidateCacheRange(addr, size);
 }
 
@@ -529,7 +529,7 @@ void ARM_Dynarmic_64::PageTableChanged(Common::PageTable& page_table,
 std::vector<ARM_Interface::BacktraceEntry> ARM_Dynarmic_64::GetBacktrace(Core::System& system,
                                                                          u64 fp, u64 lr, u64 pc) {
     std::vector<BacktraceEntry> out;
-    auto& memory = system.Memory();
+    auto& memory = system.ApplicationMemory();
 
     out.push_back({"", 0, pc, 0, ""});
 

src/core/arm/dynarmic/arm_dynarmic_64.h

@@ -38,8 +38,8 @@ public:
     void SetVectorReg(int index, u128 value) override;
     u32 GetPSTATE() const override;
     void SetPSTATE(u32 pstate) override;
-    VAddr GetTlsAddress() const override;
-    void SetTlsAddress(VAddr address) override;
+    u64 GetTlsAddress() const override;
+    void SetTlsAddress(u64 address) override;
     void SetTPIDR_EL0(u64 value) override;
     u64 GetTPIDR_EL0() const override;
 
@@ -53,7 +53,7 @@ public:
     void ClearExclusiveState() override;
 
     void ClearInstructionCache() override;
-    void InvalidateCacheRange(VAddr addr, std::size_t size) override;
+    void InvalidateCacheRange(u64 addr, std::size_t size) override;
     void PageTableChanged(Common::PageTable& new_page_table,
                           std::size_t new_address_space_size_in_bits) override;
 

src/core/core.cpp

@@ -137,7 +137,7 @@ struct System::Impl {
         device_memory = std::make_unique<Core::DeviceMemory>();
 
         is_multicore = Settings::values.use_multi_core.GetValue();
-        extended_memory_layout = Settings::values.use_extended_memory_layout.GetValue();
+        extended_memory_layout = Settings::values.use_unsafe_extended_memory_layout.GetValue();
 
         core_timing.SetMulticore(is_multicore);
         core_timing.Initialize([&system]() { system.RegisterHostThread(); });
@@ -169,7 +169,7 @@ struct System::Impl {
     void ReinitializeIfNecessary(System& system) {
         const bool must_reinitialize =
             is_multicore != Settings::values.use_multi_core.GetValue() ||
-            extended_memory_layout != Settings::values.use_extended_memory_layout.GetValue();
+            extended_memory_layout != Settings::values.use_unsafe_extended_memory_layout.GetValue();
 
         if (!must_reinitialize) {
             return;
@@ -178,7 +178,7 @@ struct System::Impl {
         LOG_DEBUG(Kernel, "Re-initializing");
 
         is_multicore = Settings::values.use_multi_core.GetValue();
-        extended_memory_layout = Settings::values.use_extended_memory_layout.GetValue();
+        extended_memory_layout = Settings::values.use_unsafe_extended_memory_layout.GetValue();
 
         Initialize(system);
     }
@@ -293,6 +293,8 @@ struct System::Impl {
         ASSERT(Kernel::KProcess::Initialize(main_process, system, "main",
                                             Kernel::KProcess::ProcessType::Userland, resource_limit)
                    .IsSuccess());
+        Kernel::KProcess::Register(system.Kernel(), main_process);
+        kernel.MakeApplicationProcess(main_process);
         const auto [load_result, load_parameters] = app_loader->Load(*main_process, system);
         if (load_result != Loader::ResultStatus::Success) {
             LOG_CRITICAL(Core, "Failed to load ROM (Error {})!", load_result);
@@ -302,7 +304,6 @@ struct System::Impl {
                 static_cast<u32>(SystemResultStatus::ErrorLoader) + static_cast<u32>(load_result));
         }
         AddGlueRegistrationForProcess(*app_loader, *main_process);
-        kernel.MakeApplicationProcess(main_process);
         kernel.InitializeCores();
 
         // Initialize cheat engine
@@ -564,7 +565,7 @@ void System::InvalidateCpuInstructionCaches() {
     impl->kernel.InvalidateAllInstructionCaches();
 }
 
-void System::InvalidateCpuInstructionCacheRange(VAddr addr, std::size_t size) {
+void System::InvalidateCpuInstructionCacheRange(u64 addr, std::size_t size) {
     impl->kernel.InvalidateCpuInstructionCacheRange(addr, size);
 }
 
@@ -681,11 +682,11 @@ const ExclusiveMonitor& System::Monitor() const {
     return impl->kernel.GetExclusiveMonitor();
 }
 
-Memory::Memory& System::Memory() {
+Memory::Memory& System::ApplicationMemory() {
     return impl->memory;
 }
 
-const Core::Memory::Memory& System::Memory() const {
+const Core::Memory::Memory& System::ApplicationMemory() const {
     return impl->memory;
 }
 
@@ -794,7 +795,7 @@ FileSys::VirtualFilesystem System::GetFilesystem() const {
 }
 
 void System::RegisterCheatList(const std::vector<Memory::CheatEntry>& list,
-                               const std::array<u8, 32>& build_id, VAddr main_region_begin,
+                               const std::array<u8, 32>& build_id, u64 main_region_begin,
                                u64 main_region_size) {
     impl->cheat_engine = std::make_unique<Memory::CheatEngine>(*this, list, build_id);
     impl->cheat_engine->SetMainMemoryParameters(main_region_begin, main_region_size);

src/core/core.h

@@ -172,7 +172,7 @@ public:
      */
     void InvalidateCpuInstructionCaches();
 
-    void InvalidateCpuInstructionCacheRange(VAddr addr, std::size_t size);
+    void InvalidateCpuInstructionCacheRange(u64 addr, std::size_t size);
 
     /// Shutdown the main emulated process.
     void ShutdownMainProcess();
@@ -256,10 +256,10 @@ public:
     [[nodiscard]] const ExclusiveMonitor& Monitor() const;
 
     /// Gets a mutable reference to the system memory instance.
-    [[nodiscard]] Core::Memory::Memory& Memory();
+    [[nodiscard]] Core::Memory::Memory& ApplicationMemory();
 
     /// Gets a constant reference to the system memory instance.
-    [[nodiscard]] const Core::Memory::Memory& Memory() const;
+    [[nodiscard]] const Core::Memory::Memory& ApplicationMemory() const;
 
     /// Gets a mutable reference to the GPU interface
     [[nodiscard]] Tegra::GPU& GPU();
@@ -353,7 +353,7 @@ public:
     [[nodiscard]] FileSys::VirtualFilesystem GetFilesystem() const;
 
     void RegisterCheatList(const std::vector<Memory::CheatEntry>& list,
-                           const std::array<u8, 0x20>& build_id, VAddr main_region_begin,
+                           const std::array<u8, 0x20>& build_id, u64 main_region_begin,
                            u64 main_region_size);
 
     void SetAppletFrontendSet(Service::AM::Applets::AppletFrontendSet&& set);

src/core/core_timing.cpp

@@ -10,6 +10,10 @@
 #include "common/windows/timer_resolution.h"
 #endif
 
+#ifdef ARCHITECTURE_x86_64
+#include "common/x64/cpu_wait.h"
+#endif
+
 #include "common/microprofile.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
@@ -269,7 +273,11 @@ void CoreTiming::ThreadLoop() {
                         if (wait_time >= timer_resolution_ns) {
                             Common::Windows::SleepForOneTick();
                         } else {
+#ifdef ARCHITECTURE_x86_64
+                            Common::X64::MicroSleep();
+#else
                             std::this_thread::yield();
+#endif
                         }
                     }
 

src/core/debugger/gdbstub.cpp

@@ -118,14 +118,14 @@ void GDBStub::Watchpoint(Kernel::KThread* thread, const Kernel::DebugWatchpoint&
 
     switch (watch.type) {
     case Kernel::DebugWatchpointType::Read:
-        SendReply(fmt::format("{}rwatch:{:x};", status, watch.start_address));
+        SendReply(fmt::format("{}rwatch:{:x};", status, GetInteger(watch.start_address)));
         break;
     case Kernel::DebugWatchpointType::Write:
-        SendReply(fmt::format("{}watch:{:x};", status, watch.start_address));
+        SendReply(fmt::format("{}watch:{:x};", status, GetInteger(watch.start_address)));
         break;
     case Kernel::DebugWatchpointType::ReadOrWrite:
     default:
-        SendReply(fmt::format("{}awatch:{:x};", status, watch.start_address));
+        SendReply(fmt::format("{}awatch:{:x};", status, GetInteger(watch.start_address)));
         break;
     }
 }
@@ -261,9 +261,9 @@ void GDBStub::ExecuteCommand(std::string_view packet, std::vector<DebuggerAction
         const size_t addr{static_cast<size_t>(strtoll(command.data(), nullptr, 16))};
         const size_t size{static_cast<size_t>(strtoll(command.data() + sep, nullptr, 16))};
 
-        if (system.Memory().IsValidVirtualAddressRange(addr, size)) {
+        if (system.ApplicationMemory().IsValidVirtualAddressRange(addr, size)) {
             std::vector<u8> mem(size);
-            system.Memory().ReadBlock(addr, mem.data(), size);
+            system.ApplicationMemory().ReadBlock(addr, mem.data(), size);
 
             SendReply(Common::HexToString(mem));
         } else {
@@ -281,8 +281,8 @@ void GDBStub::ExecuteCommand(std::string_view packet, std::vector<DebuggerAction
         const auto mem_substr{std::string_view(command).substr(mem_sep)};
         const auto mem{Common::HexStringToVector(mem_substr, false)};
 
-        if (system.Memory().IsValidVirtualAddressRange(addr, size)) {
-            system.Memory().WriteBlock(addr, mem.data(), size);
+        if (system.ApplicationMemory().IsValidVirtualAddressRange(addr, size)) {
+            system.ApplicationMemory().WriteBlock(addr, mem.data(), size);
             system.InvalidateCpuInstructionCacheRange(addr, size);
             SendReply(GDB_STUB_REPLY_OK);
         } else {
@@ -325,7 +325,7 @@ void GDBStub::HandleBreakpointInsert(std::string_view command) {
     const size_t addr{static_cast<size_t>(strtoll(command.data() + addr_sep, nullptr, 16))};
     const size_t size{static_cast<size_t>(strtoll(command.data() + size_sep, nullptr, 16))};
 
-    if (!system.Memory().IsValidVirtualAddressRange(addr, size)) {
+    if (!system.ApplicationMemory().IsValidVirtualAddressRange(addr, size)) {
         SendReply(GDB_STUB_REPLY_ERR);
         return;
     }
@@ -334,22 +334,22 @@ void GDBStub::HandleBreakpointInsert(std::string_view command) {
 
     switch (type) {
     case BreakpointType::Software:
-        replaced_instructions[addr] = system.Memory().Read32(addr);
-        system.Memory().Write32(addr, arch->BreakpointInstruction());
+        replaced_instructions[addr] = system.ApplicationMemory().Read32(addr);
+        system.ApplicationMemory().Write32(addr, arch->BreakpointInstruction());
         system.InvalidateCpuInstructionCacheRange(addr, sizeof(u32));
         success = true;
         break;
     case BreakpointType::WriteWatch:
-        success = system.ApplicationProcess()->InsertWatchpoint(system, addr, size,
+        success = system.ApplicationProcess()->InsertWatchpoint(addr, size,
                                                                 Kernel::DebugWatchpointType::Write);
         break;
     case BreakpointType::ReadWatch:
-        success = system.ApplicationProcess()->InsertWatchpoint(system, addr, size,
+        success = system.ApplicationProcess()->InsertWatchpoint(addr, size,
                                                                 Kernel::DebugWatchpointType::Read);
         break;
     case BreakpointType::AccessWatch:
         success = system.ApplicationProcess()->InsertWatchpoint(
-            system, addr, size, Kernel::DebugWatchpointType::ReadOrWrite);
+            addr, size, Kernel::DebugWatchpointType::ReadOrWrite);
         break;
     case BreakpointType::Hardware:
     default:
@@ -372,7 +372,7 @@ void GDBStub::HandleBreakpointRemove(std::string_view command) {
     const size_t addr{static_cast<size_t>(strtoll(command.data() + addr_sep, nullptr, 16))};
     const size_t size{static_cast<size_t>(strtoll(command.data() + size_sep, nullptr, 16))};
 
-    if (!system.Memory().IsValidVirtualAddressRange(addr, size)) {
+    if (!system.ApplicationMemory().IsValidVirtualAddressRange(addr, size)) {
         SendReply(GDB_STUB_REPLY_ERR);
         return;
     }
@@ -383,7 +383,7 @@ void GDBStub::HandleBreakpointRemove(std::string_view command) {
     case BreakpointType::Software: {
         const auto orig_insn{replaced_instructions.find(addr)};
         if (orig_insn != replaced_instructions.end()) {
-            system.Memory().Write32(addr, orig_insn->second);
+            system.ApplicationMemory().Write32(addr, orig_insn->second);
             system.InvalidateCpuInstructionCacheRange(addr, sizeof(u32));
             replaced_instructions.erase(addr);
             success = true;
@@ -391,16 +391,16 @@ void GDBStub::HandleBreakpointRemove(std::string_view command) {
         break;
     }
     case BreakpointType::WriteWatch:
-        success = system.ApplicationProcess()->RemoveWatchpoint(system, addr, size,
+        success = system.ApplicationProcess()->RemoveWatchpoint(addr, size,
                                                                 Kernel::DebugWatchpointType::Write);
         break;
     case BreakpointType::ReadWatch:
-        success = system.ApplicationProcess()->RemoveWatchpoint(system, addr, size,
+        success = system.ApplicationProcess()->RemoveWatchpoint(addr, size,
                                                                 Kernel::DebugWatchpointType::Read);
         break;
     case BreakpointType::AccessWatch:
         success = system.ApplicationProcess()->RemoveWatchpoint(
-            system, addr, size, Kernel::DebugWatchpointType::ReadOrWrite);
+            addr, size, Kernel::DebugWatchpointType::ReadOrWrite);
         break;
     case BreakpointType::Hardware:
     default:
@@ -483,9 +483,9 @@ static std::optional<std::string> GetNameFromThreadType64(Core::Memory::Memory&
 static std::optional<std::string> GetThreadName(Core::System& system,
                                                 const Kernel::KThread* thread) {
     if (system.ApplicationProcess()->Is64BitProcess()) {
-        return GetNameFromThreadType64(system.Memory(), thread);
+        return GetNameFromThreadType64(system.ApplicationMemory(), thread);
     } else {
-        return GetNameFromThreadType32(system.Memory(), thread);
+        return GetNameFromThreadType32(system.ApplicationMemory(), thread);
     }
 }
 
@@ -554,8 +554,9 @@ void GDBStub::HandleQuery(std::string_view command) {
         if (main != modules.end()) {
             SendReply(fmt::format("TextSeg={:x}", main->first));
         } else {
-            SendReply(fmt::format("TextSeg={:x}",
-                                  system.ApplicationProcess()->PageTable().GetCodeRegionStart()));
+            SendReply(fmt::format(
+                "TextSeg={:x}",
+                GetInteger(system.ApplicationProcess()->PageTable().GetCodeRegionStart())));
         }
     } else if (command.starts_with("Xfer:libraries:read::")) {
         Loader::AppLoader::Modules modules;
@@ -757,17 +758,20 @@ void GDBStub::HandleRcmd(const std::vector<u8>& command) {
         reply = fmt::format("Process:     {:#x} ({})\n"
                             "Program Id:  {:#018x}\n",
                             process->GetProcessId(), process->GetName(), process->GetProgramId());
-        reply +=
-            fmt::format("Layout:\n"
-                        "  Alias: {:#012x} - {:#012x}\n"
-                        "  Heap:  {:#012x} - {:#012x}\n"
-                        "  Aslr:  {:#012x} - {:#012x}\n"
-                        "  Stack: {:#012x} - {:#012x}\n"
-                        "Modules:\n",
-                        page_table.GetAliasRegionStart(), page_table.GetAliasRegionEnd(),
-                        page_table.GetHeapRegionStart(), page_table.GetHeapRegionEnd(),
-                        page_table.GetAliasCodeRegionStart(), page_table.GetAliasCodeRegionEnd(),
-                        page_table.GetStackRegionStart(), page_table.GetStackRegionEnd());
+        reply += fmt::format("Layout:\n"
+                             "  Alias: {:#012x} - {:#012x}\n"
+                             "  Heap:  {:#012x} - {:#012x}\n"
+                             "  Aslr:  {:#012x} - {:#012x}\n"
+                             "  Stack: {:#012x} - {:#012x}\n"
+                             "Modules:\n",
+                             GetInteger(page_table.GetAliasRegionStart()),
+                             GetInteger(page_table.GetAliasRegionEnd()),
+                             GetInteger(page_table.GetHeapRegionStart()),
+                             GetInteger(page_table.GetHeapRegionEnd()),
+                             GetInteger(page_table.GetAliasCodeRegionStart()),
+                             GetInteger(page_table.GetAliasCodeRegionEnd()),
+                             GetInteger(page_table.GetStackRegionStart()),
+                             GetInteger(page_table.GetStackRegionEnd()));
 
         for (const auto& [vaddr, name] : modules) {
             reply += fmt::format("  {:#012x} - {:#012x} {}\n", vaddr,

src/core/device_memory.h

@@ -3,8 +3,8 @@
 
 #pragma once
 
-#include "common/common_types.h"
 #include "common/host_memory.h"
+#include "common/typed_address.h"
 
 namespace Core {
 
@@ -25,20 +25,22 @@ public:
     DeviceMemory(const DeviceMemory&) = delete;
 
     template <typename T>
-    PAddr GetPhysicalAddr(const T* ptr) const {
+    Common::PhysicalAddress GetPhysicalAddr(const T* ptr) const {
         return (reinterpret_cast<uintptr_t>(ptr) -
                 reinterpret_cast<uintptr_t>(buffer.BackingBasePointer())) +
                DramMemoryMap::Base;
     }
 
     template <typename T>
-    T* GetPointer(PAddr addr) {
-        return reinterpret_cast<T*>(buffer.BackingBasePointer() + (addr - DramMemoryMap::Base));
+    T* GetPointer(Common::PhysicalAddress addr) {
+        return reinterpret_cast<T*>(buffer.BackingBasePointer() +
+                                    (GetInteger(addr) - DramMemoryMap::Base));
     }
 
     template <typename T>
-    const T* GetPointer(PAddr addr) const {
-        return reinterpret_cast<T*>(buffer.BackingBasePointer() + (addr - DramMemoryMap::Base));
+    const T* GetPointer(Common::PhysicalAddress addr) const {
+        return reinterpret_cast<T*>(buffer.BackingBasePointer() +
+                                    (GetInteger(addr) - DramMemoryMap::Base));
     }
 
     Common::HostMemory buffer;

src/core/frontend/applets/applet.h

@@ -0,0 +1,14 @@
+// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+namespace Core::Frontend {
+
+class Applet {
+public:
+    virtual ~Applet() = default;
+    virtual void Close() const = 0;
+};
+
+} // namespace Core::Frontend

src/core/frontend/applets/cabinet.cpp

@@ -10,6 +10,8 @@ namespace Core::Frontend {
 
 CabinetApplet::~CabinetApplet() = default;
 
+void DefaultCabinetApplet::Close() const {}
+
 void DefaultCabinetApplet::ShowCabinetApplet(
     const CabinetCallback& callback, const CabinetParameters& parameters,
     std::shared_ptr<Service::NFP::NfpDevice> nfp_device) const {

src/core/frontend/applets/cabinet.h

@@ -4,6 +4,7 @@
 #pragma once
 
 #include <functional>
+#include "core/frontend/applets/applet.h"
 #include "core/hle/service/nfp/nfp_types.h"
 
 namespace Service::NFP {
@@ -20,7 +21,7 @@ struct CabinetParameters {
 
 using CabinetCallback = std::function<void(bool, const std::string&)>;
 
-class CabinetApplet {
+class CabinetApplet : public Applet {
 public:
     virtual ~CabinetApplet();
     virtual void ShowCabinetApplet(const CabinetCallback& callback,
@@ -30,6 +31,7 @@ public:
 
 class DefaultCabinetApplet final : public CabinetApplet {
 public:
+    void Close() const override;
     void ShowCabinetApplet(const CabinetCallback& callback, const CabinetParameters& parameters,
                            std::shared_ptr<Service::NFP::NfpDevice> nfp_device) const override;
 };

src/core/frontend/applets/controller.cpp

@@ -16,6 +16,8 @@ DefaultControllerApplet::DefaultControllerApplet(HID::HIDCore& hid_core_) : hid_
 
 DefaultControllerApplet::~DefaultControllerApplet() = default;
 
+void DefaultControllerApplet::Close() const {}
+
 void DefaultControllerApplet::ReconfigureControllers(ReconfigureCallback callback,
                                                      const ControllerParameters& parameters) const {
     LOG_INFO(Service_HID, "called, deducing the best configuration based on the given parameters!");
@@ -69,7 +71,7 @@ void DefaultControllerApplet::ReconfigureControllers(ReconfigureCallback callbac
         }
     }
 
-    callback();
+    callback(true);
 }
 
 } // namespace Core::Frontend

src/core/frontend/applets/controller.h

@@ -7,6 +7,7 @@
 #include <vector>
 
 #include "common/common_types.h"
+#include "core/frontend/applets/applet.h"
 
 namespace Core::HID {
 class HIDCore;
@@ -34,9 +35,9 @@ struct ControllerParameters {
     bool allow_gamecube_controller{};
 };
 
-class ControllerApplet {
+class ControllerApplet : public Applet {
 public:
-    using ReconfigureCallback = std::function<void()>;
+    using ReconfigureCallback = std::function<void(bool)>;
 
     virtual ~ControllerApplet();
 
@@ -49,6 +50,7 @@ public:
     explicit DefaultControllerApplet(HID::HIDCore& hid_core_);
     ~DefaultControllerApplet() override;
 
+    void Close() const override;
     void ReconfigureControllers(ReconfigureCallback callback,
                                 const ControllerParameters& parameters) const override;
 

src/core/frontend/applets/error.cpp

@@ -8,6 +8,8 @@ namespace Core::Frontend {
 
 ErrorApplet::~ErrorApplet() = default;
 
+void DefaultErrorApplet::Close() const {}
+
 void DefaultErrorApplet::ShowError(Result error, FinishedCallback finished) const {
     LOG_CRITICAL(Service_Fatal, "Application requested error display: {:04}-{:04} (raw={:08X})",
                  error.module.Value(), error.description.Value(), error.raw);

src/core/frontend/applets/error.h

@@ -6,11 +6,12 @@
 #include <chrono>
 #include <functional>
 
+#include "core/frontend/applets/applet.h"
 #include "core/hle/result.h"
 
 namespace Core::Frontend {
 
-class ErrorApplet {
+class ErrorApplet : public Applet {
 public:
     using FinishedCallback = std::function<void()>;
 
@@ -28,6 +29,7 @@ public:
 
 class DefaultErrorApplet final : public ErrorApplet {
 public:
+    void Close() const override;
     void ShowError(Result error, FinishedCallback finished) const override;
     void ShowErrorWithTimestamp(Result error, std::chrono::seconds time,
                                 FinishedCallback finished) const override;

src/core/frontend/applets/general_frontend.cpp

@@ -10,6 +10,8 @@ ParentalControlsApplet::~ParentalControlsApplet() = default;
 
 DefaultParentalControlsApplet::~DefaultParentalControlsApplet() = default;
 
+void DefaultParentalControlsApplet::Close() const {}
+
 void DefaultParentalControlsApplet::VerifyPIN(std::function<void(bool)> finished,
                                               bool suspend_future_verification_temporarily) {
     LOG_INFO(Service_AM,
@@ -39,6 +41,8 @@ PhotoViewerApplet::~PhotoViewerApplet() = default;
 
 DefaultPhotoViewerApplet::~DefaultPhotoViewerApplet() = default;
 
+void DefaultPhotoViewerApplet::Close() const {}
+
 void DefaultPhotoViewerApplet::ShowPhotosForApplication(u64 title_id,
                                                         std::function<void()> finished) const {
     LOG_INFO(Service_AM,

src/core/frontend/applets/general_frontend.h

@@ -6,9 +6,11 @@
 #include <functional>
 #include "common/common_types.h"
 
+#include "core/frontend/applets/applet.h"
+
 namespace Core::Frontend {
 
-class ParentalControlsApplet {
+class ParentalControlsApplet : public Applet {
 public:
     virtual ~ParentalControlsApplet();
 
@@ -33,6 +35,7 @@ class DefaultParentalControlsApplet final : public ParentalControlsApplet {
 public:
     ~DefaultParentalControlsApplet() override;
 
+    void Close() const override;
     void VerifyPIN(std::function<void(bool)> finished,
                    bool suspend_future_verification_temporarily) override;
     void VerifyPINForSettings(std::function<void(bool)> finished) override;
@@ -40,7 +43,7 @@ public:
     void ChangePIN(std::function<void()> finished) override;
 };
 
-class PhotoViewerApplet {
+class PhotoViewerApplet : public Applet {
 public:
     virtual ~PhotoViewerApplet();
 
@@ -52,6 +55,7 @@ class DefaultPhotoViewerApplet final : public PhotoViewerApplet {
 public:
     ~DefaultPhotoViewerApplet() override;
 
+    void Close() const override;
     void ShowPhotosForApplication(u64 title_id, std::function<void()> finished) const override;
     void ShowAllPhotos(std::function<void()> finished) const override;
 };

src/core/frontend/applets/mii_edit.cpp

@@ -8,6 +8,8 @@ namespace Core::Frontend {
 
 MiiEditApplet::~MiiEditApplet() = default;
 
+void DefaultMiiEditApplet::Close() const {}
+
 void DefaultMiiEditApplet::ShowMiiEdit(const MiiEditCallback& callback) const {
     LOG_WARNING(Service_AM, "(STUBBED) called");
 

src/core/frontend/applets/mii_edit.h

@@ -5,9 +5,11 @@
 
 #include <functional>
 
+#include "core/frontend/applets/applet.h"
+
 namespace Core::Frontend {
 
-class MiiEditApplet {
+class MiiEditApplet : public Applet {
 public:
     using MiiEditCallback = std::function<void()>;
 
@@ -18,6 +20,7 @@ public:
 
 class DefaultMiiEditApplet final : public MiiEditApplet {
 public:
+    void Close() const override;
     void ShowMiiEdit(const MiiEditCallback& callback) const override;
 };
 

src/core/frontend/applets/profile_select.cpp

@@ -9,7 +9,10 @@ namespace Core::Frontend {
 
 ProfileSelectApplet::~ProfileSelectApplet() = default;
 
-void DefaultProfileSelectApplet::SelectProfile(SelectProfileCallback callback) const {
+void DefaultProfileSelectApplet::Close() const {}
+
+void DefaultProfileSelectApplet::SelectProfile(SelectProfileCallback callback,
+                                               const ProfileSelectParameters& parameters) const {
     Service::Account::ProfileManager manager;
     callback(manager.GetUser(Settings::values.current_user.GetValue()).value_or(Common::UUID{}));
     LOG_INFO(Service_ACC, "called, selecting current user instead of prompting...");

src/core/frontend/applets/profile_select.h

@@ -5,22 +5,35 @@
 
 #include <functional>
 #include <optional>
+
 #include "common/uuid.h"
+#include "core/frontend/applets/applet.h"
+#include "core/hle/service/am/applets/applet_profile_select.h"
 
 namespace Core::Frontend {
 
-class ProfileSelectApplet {
+struct ProfileSelectParameters {
+    Service::AM::Applets::UiMode mode;
+    std::array<Common::UUID, 8> invalid_uid_list;
+    Service::AM::Applets::UiSettingsDisplayOptions display_options;
+    Service::AM::Applets::UserSelectionPurpose purpose;
+};
+
+class ProfileSelectApplet : public Applet {
 public:
     using SelectProfileCallback = std::function<void(std::optional<Common::UUID>)>;
 
     virtual ~ProfileSelectApplet();
 
-    virtual void SelectProfile(SelectProfileCallback callback) const = 0;
+    virtual void SelectProfile(SelectProfileCallback callback,
+                               const ProfileSelectParameters& parameters) const = 0;
 };
 
 class DefaultProfileSelectApplet final : public ProfileSelectApplet {
 public:
-    void SelectProfile(SelectProfileCallback callback) const override;
+    void Close() const override;
+    void SelectProfile(SelectProfileCallback callback,
+                       const ProfileSelectParameters& parameters) const override;
 };
 
 } // namespace Core::Frontend

src/core/frontend/applets/software_keyboard.cpp

@@ -13,6 +13,8 @@ SoftwareKeyboardApplet::~SoftwareKeyboardApplet() = default;
 
 DefaultSoftwareKeyboardApplet::~DefaultSoftwareKeyboardApplet() = default;
 
+void DefaultSoftwareKeyboardApplet::Close() const {}
+
 void DefaultSoftwareKeyboardApplet::InitializeKeyboard(
     bool is_inline, KeyboardInitializeParameters initialize_parameters,
     SubmitNormalCallback submit_normal_callback_, SubmitInlineCallback submit_inline_callback_) {

src/core/frontend/applets/software_keyboard.h

@@ -7,6 +7,7 @@
 
 #include "common/common_types.h"
 
+#include "core/frontend/applets/applet.h"
 #include "core/hle/service/am/applets/applet_software_keyboard_types.h"
 
 namespace Core::Frontend {
@@ -52,7 +53,7 @@ struct InlineTextParameters {
     s32 cursor_position;
 };
 
-class SoftwareKeyboardApplet {
+class SoftwareKeyboardApplet : public Applet {
 public:
     using SubmitInlineCallback =
         std::function<void(Service::AM::Applets::SwkbdReplyType, std::u16string, s32)>;
@@ -84,6 +85,8 @@ class DefaultSoftwareKeyboardApplet final : public SoftwareKeyboardApplet {
 public:
     ~DefaultSoftwareKeyboardApplet() override;
 
+    void Close() const override;
+
     void InitializeKeyboard(bool is_inline, KeyboardInitializeParameters initialize_parameters,
                             SubmitNormalCallback submit_normal_callback_,
                             SubmitInlineCallback submit_inline_callback_) override;

src/core/frontend/applets/web_browser.cpp

@@ -10,6 +10,8 @@ WebBrowserApplet::~WebBrowserApplet() = default;
 
 DefaultWebBrowserApplet::~DefaultWebBrowserApplet() = default;
 
+void DefaultWebBrowserApplet::Close() const {}
+
 void DefaultWebBrowserApplet::OpenLocalWebPage(const std::string& local_url,
                                                ExtractROMFSCallback extract_romfs_callback,
                                                OpenWebPageCallback callback) const {

src/core/frontend/applets/web_browser.h

@@ -5,11 +5,12 @@
 
 #include <functional>
 
+#include "core/frontend/applets/applet.h"
 #include "core/hle/service/am/applets/applet_web_browser_types.h"
 
 namespace Core::Frontend {
 
-class WebBrowserApplet {
+class WebBrowserApplet : public Applet {
 public:
     using ExtractROMFSCallback = std::function<void()>;
     using OpenWebPageCallback =
@@ -29,6 +30,8 @@ class DefaultWebBrowserApplet final : public WebBrowserApplet {
 public:
     ~DefaultWebBrowserApplet() override;
 
+    void Close() const override;
+
     void OpenLocalWebPage(const std::string& local_url, ExtractROMFSCallback extract_romfs_callback,
                           OpenWebPageCallback callback) const override;
 

src/core/hid/emulated_controller.cpp

@@ -280,6 +280,10 @@ void EmulatedController::LoadVirtualGamepadParams() {
     virtual_stick_params[Settings::NativeAnalog::LStick].Set("axis_y", 1);
     virtual_stick_params[Settings::NativeAnalog::RStick].Set("axis_x", 2);
     virtual_stick_params[Settings::NativeAnalog::RStick].Set("axis_y", 3);
+    virtual_stick_params[Settings::NativeAnalog::LStick].Set("deadzone", 0.0f);
+    virtual_stick_params[Settings::NativeAnalog::LStick].Set("range", 1.0f);
+    virtual_stick_params[Settings::NativeAnalog::RStick].Set("deadzone", 0.0f);
+    virtual_stick_params[Settings::NativeAnalog::RStick].Set("range", 1.0f);
 }
 
 void EmulatedController::ReloadInput() {

src/core/hle/kernel/board/nintendo/nx/k_memory_layout.cpp

@@ -76,22 +76,24 @@ void SetupDevicePhysicalMemoryRegions(KMemoryLayout& memory_layout) {
 
 void SetupDramPhysicalMemoryRegions(KMemoryLayout& memory_layout) {
     const size_t intended_memory_size = KSystemControl::Init::GetIntendedMemorySize();
-    const PAddr physical_memory_base_address =
+    const KPhysicalAddress physical_memory_base_address =
         KSystemControl::Init::GetKernelPhysicalBaseAddress(DramPhysicalAddress);
 
     // Insert blocks into the tree.
     ASSERT(memory_layout.GetPhysicalMemoryRegionTree().Insert(
-        physical_memory_base_address, intended_memory_size, KMemoryRegionType_Dram));
+        GetInteger(physical_memory_base_address), intended_memory_size, KMemoryRegionType_Dram));
     ASSERT(memory_layout.GetPhysicalMemoryRegionTree().Insert(
-        physical_memory_base_address, ReservedEarlyDramSize, KMemoryRegionType_DramReservedEarly));
+        GetInteger(physical_memory_base_address), ReservedEarlyDramSize,
+        KMemoryRegionType_DramReservedEarly));
 
     // Insert the KTrace block at the end of Dram, if KTrace is enabled.
     static_assert(!IsKTraceEnabled || KTraceBufferSize > 0);
     if constexpr (IsKTraceEnabled) {
-        const PAddr ktrace_buffer_phys_addr =
+        const KPhysicalAddress ktrace_buffer_phys_addr =
             physical_memory_base_address + intended_memory_size - KTraceBufferSize;
         ASSERT(memory_layout.GetPhysicalMemoryRegionTree().Insert(
-            ktrace_buffer_phys_addr, KTraceBufferSize, KMemoryRegionType_KernelTraceBuffer));
+            GetInteger(ktrace_buffer_phys_addr), KTraceBufferSize,
+            KMemoryRegionType_KernelTraceBuffer));
     }
 }
 

src/core/hle/kernel/board/nintendo/nx/k_memory_layout.h

@@ -3,10 +3,10 @@
 
 #pragma once
 
-#include "common/common_types.h"
+#include "core/hle/kernel/k_typed_address.h"
 
 namespace Kernel {
 
-constexpr inline PAddr MainMemoryAddress = 0x80000000;
+constexpr inline KPhysicalAddress MainMemoryAddress = 0x80000000;
 
 } // namespace Kernel

src/core/hle/kernel/board/nintendo/nx/k_system_control.cpp

@@ -35,12 +35,13 @@ namespace {
 using namespace Common::Literals;
 
 u32 GetMemorySizeForInit() {
-    return Settings::values.use_extended_memory_layout ? Smc::MemorySize_6GB : Smc::MemorySize_4GB;
+    return Settings::values.use_unsafe_extended_memory_layout ? Smc::MemorySize_8GB
+                                                              : Smc::MemorySize_4GB;
 }
 
 Smc::MemoryArrangement GetMemoryArrangeForInit() {
-    return Settings::values.use_extended_memory_layout ? Smc::MemoryArrangement_6GB
-                                                       : Smc::MemoryArrangement_4GB;
+    return Settings::values.use_unsafe_extended_memory_layout ? Smc::MemoryArrangement_8GB
+                                                              : Smc::MemoryArrangement_4GB;
 }
 } // namespace
 
@@ -61,7 +62,7 @@ size_t KSystemControl::Init::GetIntendedMemorySize() {
     }
 }
 
-PAddr KSystemControl::Init::GetKernelPhysicalBaseAddress(u64 base_address) {
+KPhysicalAddress KSystemControl::Init::GetKernelPhysicalBaseAddress(KPhysicalAddress base_address) {
     const size_t real_dram_size = KSystemControl::Init::GetRealMemorySize();
     const size_t intended_dram_size = KSystemControl::Init::GetIntendedMemorySize();
     if (intended_dram_size * 2 < real_dram_size) {
@@ -91,7 +92,8 @@ std::size_t KSystemControl::Init::GetApplicationPoolSize() {
         case Smc::MemoryArrangement_6GBForAppletDev:
             return 3285_MiB;
         case Smc::MemoryArrangement_8GB:
-            return 4916_MiB;
+            // Real kernel sets this to 4916_MiB. We are not debugging applets.
+            return 6547_MiB;
         }
     }();
 
@@ -115,7 +117,8 @@ size_t KSystemControl::Init::GetAppletPoolSize() {
         case Smc::MemoryArrangement_6GBForAppletDev:
             return 2193_MiB;
         case Smc::MemoryArrangement_8GB:
-            return 2193_MiB;
+            //! Real kernel sets this to 2193_MiB. We are not debugging applets.
+            return 562_MiB;
         }
     }();
 

src/core/hle/kernel/board/nintendo/nx/k_system_control.h

@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include "common/common_types.h"
+#include "core/hle/kernel/k_typed_address.h"
 
 namespace Kernel::Board::Nintendo::Nx {
 
@@ -18,7 +18,7 @@ public:
         // Initialization.
         static std::size_t GetRealMemorySize();
         static std::size_t GetIntendedMemorySize();
-        static PAddr GetKernelPhysicalBaseAddress(u64 base_address);
+        static KPhysicalAddress GetKernelPhysicalBaseAddress(KPhysicalAddress base_address);
         static bool ShouldIncreaseThreadResourceLimit();
         static std::size_t GetApplicationPoolSize();
         static std::size_t GetAppletPoolSize();

src/core/hle/kernel/code_set.h

@@ -5,7 +5,7 @@
 
 #include <cstddef>
 
-#include "common/common_types.h"
+#include "core/hle/kernel/k_typed_address.h"
 #include "core/hle/kernel/physical_memory.h"
 
 namespace Kernel {
@@ -36,7 +36,7 @@ struct CodeSet final {
         std::size_t offset = 0;
 
         /// The address to map this segment to.
-        VAddr addr = 0;
+        KProcessAddress addr = 0;
 
         /// The size of this segment in bytes.
         u32 size = 0;
@@ -82,7 +82,7 @@ struct CodeSet final {
     std::array<Segment, 3> segments;
 
     /// The entry point address for this code set.
-    VAddr entrypoint = 0;
+    KProcessAddress entrypoint = 0;
 };
 
 } // namespace Kernel

src/core/hle/kernel/init/init_slab_setup.cpp

@@ -4,7 +4,6 @@
 #include "common/alignment.h"
 #include "common/assert.h"
 #include "common/common_funcs.h"
-#include "common/common_types.h"
 #include "core/core.h"
 #include "core/device_memory.h"
 #include "core/hardware_properties.h"
@@ -30,6 +29,7 @@
 #include "core/hle/kernel/k_thread.h"
 #include "core/hle/kernel/k_thread_local_page.h"
 #include "core/hle/kernel/k_transfer_memory.h"
+#include "core/hle/kernel/k_typed_address.h"
 
 namespace Kernel::Init {
 
@@ -104,17 +104,18 @@ static_assert(KernelPageBufferAdditionalSize ==
 
 /// Helper function to translate from the slab virtual address to the reserved location in physical
 /// memory.
-static PAddr TranslateSlabAddrToPhysical(KMemoryLayout& memory_layout, VAddr slab_addr) {
-    slab_addr -= memory_layout.GetSlabRegionAddress();
-    return slab_addr + Core::DramMemoryMap::SlabHeapBase;
+static KPhysicalAddress TranslateSlabAddrToPhysical(KMemoryLayout& memory_layout,
+                                                    KVirtualAddress slab_addr) {
+    slab_addr -= GetInteger(memory_layout.GetSlabRegionAddress());
+    return GetInteger(slab_addr) + Core::DramMemoryMap::SlabHeapBase;
 }
 
 template <typename T>
-VAddr InitializeSlabHeap(Core::System& system, KMemoryLayout& memory_layout, VAddr address,
-                         size_t num_objects) {
+KVirtualAddress InitializeSlabHeap(Core::System& system, KMemoryLayout& memory_layout,
+                                   KVirtualAddress address, size_t num_objects) {
 
     const size_t size = Common::AlignUp(sizeof(T) * num_objects, alignof(void*));
-    VAddr start = Common::AlignUp(address, alignof(T));
+    KVirtualAddress start = Common::AlignUp(GetInteger(address), alignof(T));
 
     // This should use the virtual memory address passed in, but currently, we do not setup the
     // kernel virtual memory layout. Instead, we simply map these at a region of physical memory
@@ -195,7 +196,7 @@ void InitializeSlabHeaps(Core::System& system, KMemoryLayout& memory_layout) {
     auto& kernel = system.Kernel();
 
     // Get the start of the slab region, since that's where we'll be working.
-    VAddr address = memory_layout.GetSlabRegionAddress();
+    KVirtualAddress address = memory_layout.GetSlabRegionAddress();
 
     // Initialize slab type array to be in sorted order.
     std::array<KSlabType, KSlabType_Count> slab_types;
@@ -228,7 +229,7 @@ void InitializeSlabHeaps(Core::System& system, KMemoryLayout& memory_layout) {
     }
 
     // Track the gaps, so that we can free them to the unused slab tree.
-    VAddr gap_start = address;
+    KVirtualAddress gap_start = address;
     size_t gap_size = 0;
 
     for (size_t i = 0; i < slab_gaps.size(); i++) {
@@ -280,7 +281,7 @@ void KPageBufferSlabHeap::Initialize(Core::System& system) {
     // Allocate memory for the slab.
     constexpr auto AllocateOption = KMemoryManager::EncodeOption(
         KMemoryManager::Pool::System, KMemoryManager::Direction::FromFront);
-    const PAddr slab_address =
+    const KPhysicalAddress slab_address =
         kernel.MemoryManager().AllocateAndOpenContinuous(num_pages, 1, AllocateOption);
     ASSERT(slab_address != 0);
 

src/core/hle/kernel/initial_process.h

@@ -14,7 +14,7 @@ using namespace Common::Literals;
 
 constexpr std::size_t InitialProcessBinarySizeMax = 12_MiB;
 
-static inline PAddr GetInitialProcessBinaryPhysicalAddress() {
+static inline KPhysicalAddress GetInitialProcessBinaryPhysicalAddress() {
     return Kernel::Board::Nintendo::Nx::KSystemControl::Init::GetKernelPhysicalBaseAddress(
         MainMemoryAddress);
 }

src/core/hle/kernel/k_address_arbiter.cpp

@@ -8,6 +8,7 @@
 #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/k_typed_address.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/svc_results.h"
 #include "core/memory.h"
@@ -20,12 +21,12 @@ KAddressArbiter::~KAddressArbiter() = default;
 
 namespace {
 
-bool ReadFromUser(Core::System& system, s32* out, VAddr address) {
-    *out = system.Memory().Read32(address);
+bool ReadFromUser(KernelCore& kernel, s32* out, KProcessAddress address) {
+    *out = GetCurrentMemory(kernel).Read32(GetInteger(address));
     return true;
 }
 
-bool DecrementIfLessThan(Core::System& system, s32* out, VAddr address, s32 value) {
+bool DecrementIfLessThan(Core::System& system, s32* out, KProcessAddress address, s32 value) {
     auto& monitor = system.Monitor();
     const auto current_core = system.Kernel().CurrentPhysicalCoreIndex();
 
@@ -34,22 +35,30 @@ bool DecrementIfLessThan(Core::System& system, s32* out, VAddr address, s32 valu
 
     // TODO(bunnei): We should call CanAccessAtomic(..) here.
 
-    // Load the value from the address.
-    const s32 current_value = static_cast<s32>(monitor.ExclusiveRead32(current_core, address));
+    s32 current_value{};
 
-    // Compare it to the desired one.
-    if (current_value < value) {
-        // If less than, we want to try to decrement.
-        const s32 decrement_value = current_value - 1;
+    while (true) {
+        // Load the value from the address.
+        current_value =
+            static_cast<s32>(monitor.ExclusiveRead32(current_core, GetInteger(address)));
+
+        // Compare it to the desired one.
+        if (current_value < value) {
+            // If less than, we want to try to decrement.
+            const s32 decrement_value = current_value - 1;
+
+            // Decrement and try to store.
+            if (monitor.ExclusiveWrite32(current_core, GetInteger(address),
+                                         static_cast<u32>(decrement_value))) {
+                break;
+            }
 
-        // Decrement and try to store.
-        if (!monitor.ExclusiveWrite32(current_core, address, static_cast<u32>(decrement_value))) {
             // If we failed to store, try again.
-            DecrementIfLessThan(system, out, address, value);
+        } else {
+            // Otherwise, clear our exclusive hold and finish
+            monitor.ClearExclusive(current_core);
+            break;
         }
-    } else {
-        // Otherwise, clear our exclusive hold and finish
-        monitor.ClearExclusive(current_core);
     }
 
     // We're done.
@@ -57,7 +66,8 @@ bool DecrementIfLessThan(Core::System& system, s32* out, VAddr address, s32 valu
     return true;
 }
 
-bool UpdateIfEqual(Core::System& system, s32* out, VAddr address, s32 value, s32 new_value) {
+bool UpdateIfEqual(Core::System& system, s32* out, KProcessAddress address, s32 value,
+                   s32 new_value) {
     auto& monitor = system.Monitor();
     const auto current_core = system.Kernel().CurrentPhysicalCoreIndex();
 
@@ -66,21 +76,29 @@ bool UpdateIfEqual(Core::System& system, s32* out, VAddr address, s32 value, s32
 
     // TODO(bunnei): We should call CanAccessAtomic(..) here.
 
+    s32 current_value{};
+
     // Load the value from the address.
-    const s32 current_value = static_cast<s32>(monitor.ExclusiveRead32(current_core, address));
+    while (true) {
+        current_value =
+            static_cast<s32>(monitor.ExclusiveRead32(current_core, GetInteger(address)));
 
-    // Compare it to the desired one.
-    if (current_value == value) {
-        // If equal, we want to try to write the new value.
+        // Compare it to the desired one.
+        if (current_value == value) {
+            // If equal, we want to try to write the new value.
+
+            // Try to store.
+            if (monitor.ExclusiveWrite32(current_core, GetInteger(address),
+                                         static_cast<u32>(new_value))) {
+                break;
+            }
 
-        // Try to store.
-        if (!monitor.ExclusiveWrite32(current_core, address, static_cast<u32>(new_value))) {
             // If we failed to store, try again.
-            UpdateIfEqual(system, out, address, value, new_value);
+        } else {
+            // Otherwise, clear our exclusive hold and finish.
+            monitor.ClearExclusive(current_core);
+            break;
         }
-    } else {
-        // Otherwise, clear our exclusive hold and finish.
-        monitor.ClearExclusive(current_core);
     }
 
     // We're done.
@@ -110,7 +128,7 @@ private:
 
 } // namespace
 
-Result KAddressArbiter::Signal(VAddr addr, s32 count) {
+Result KAddressArbiter::Signal(uint64_t addr, s32 count) {
     // Perform signaling.
     s32 num_waiters{};
     {
@@ -133,7 +151,7 @@ Result KAddressArbiter::Signal(VAddr addr, s32 count) {
     R_SUCCEED();
 }
 
-Result KAddressArbiter::SignalAndIncrementIfEqual(VAddr addr, s32 value, s32 count) {
+Result KAddressArbiter::SignalAndIncrementIfEqual(uint64_t addr, s32 value, s32 count) {
     // Perform signaling.
     s32 num_waiters{};
     {
@@ -162,7 +180,7 @@ Result KAddressArbiter::SignalAndIncrementIfEqual(VAddr addr, s32 value, s32 cou
     R_SUCCEED();
 }
 
-Result KAddressArbiter::SignalAndModifyByWaitingCountIfEqual(VAddr addr, s32 value, s32 count) {
+Result KAddressArbiter::SignalAndModifyByWaitingCountIfEqual(uint64_t addr, s32 value, s32 count) {
     // Perform signaling.
     s32 num_waiters{};
     {
@@ -203,7 +221,7 @@ Result KAddressArbiter::SignalAndModifyByWaitingCountIfEqual(VAddr addr, s32 val
         if (value != new_value) {
             succeeded = UpdateIfEqual(m_system, std::addressof(user_value), addr, value, new_value);
         } else {
-            succeeded = ReadFromUser(m_system, std::addressof(user_value), addr);
+            succeeded = ReadFromUser(m_kernel, std::addressof(user_value), addr);
         }
 
         R_UNLESS(succeeded, ResultInvalidCurrentMemory);
@@ -225,7 +243,7 @@ Result KAddressArbiter::SignalAndModifyByWaitingCountIfEqual(VAddr addr, s32 val
     R_SUCCEED();
 }
 
-Result KAddressArbiter::WaitIfLessThan(VAddr addr, s32 value, bool decrement, s64 timeout) {
+Result KAddressArbiter::WaitIfLessThan(uint64_t addr, s32 value, bool decrement, s64 timeout) {
     // Prepare to wait.
     KThread* cur_thread = GetCurrentThreadPointer(m_kernel);
     KHardwareTimer* timer{};
@@ -246,7 +264,7 @@ Result KAddressArbiter::WaitIfLessThan(VAddr addr, s32 value, bool decrement, s6
         if (decrement) {
             succeeded = DecrementIfLessThan(m_system, std::addressof(user_value), addr, value);
         } else {
-            succeeded = ReadFromUser(m_system, std::addressof(user_value), addr);
+            succeeded = ReadFromUser(m_kernel, std::addressof(user_value), addr);
         }
 
         if (!succeeded) {
@@ -280,7 +298,7 @@ Result KAddressArbiter::WaitIfLessThan(VAddr addr, s32 value, bool decrement, s6
     return cur_thread->GetWaitResult();
 }
 
-Result KAddressArbiter::WaitIfEqual(VAddr addr, s32 value, s64 timeout) {
+Result KAddressArbiter::WaitIfEqual(uint64_t addr, s32 value, s64 timeout) {
     // Prepare to wait.
     KThread* cur_thread = GetCurrentThreadPointer(m_kernel);
     KHardwareTimer* timer{};
@@ -297,7 +315,7 @@ Result KAddressArbiter::WaitIfEqual(VAddr addr, s32 value, s64 timeout) {
 
         // Read the value from userspace.
         s32 user_value{};
-        if (!ReadFromUser(m_system, std::addressof(user_value), addr)) {
+        if (!ReadFromUser(m_kernel, std::addressof(user_value), addr)) {
             slp.CancelSleep();
             R_THROW(ResultInvalidCurrentMemory);
         }

src/core/hle/kernel/k_address_arbiter.h

@@ -25,7 +25,7 @@ public:
     explicit KAddressArbiter(Core::System& system);
     ~KAddressArbiter();
 
-    Result SignalToAddress(VAddr addr, Svc::SignalType type, s32 value, s32 count) {
+    Result SignalToAddress(uint64_t addr, Svc::SignalType type, s32 value, s32 count) {
         switch (type) {
         case Svc::SignalType::Signal:
             R_RETURN(this->Signal(addr, count));
@@ -38,7 +38,7 @@ public:
         }
     }
 
-    Result WaitForAddress(VAddr addr, Svc::ArbitrationType type, s32 value, s64 timeout) {
+    Result WaitForAddress(uint64_t addr, Svc::ArbitrationType type, s32 value, s64 timeout) {
         switch (type) {
         case Svc::ArbitrationType::WaitIfLessThan:
             R_RETURN(WaitIfLessThan(addr, value, false, timeout));
@@ -52,11 +52,11 @@ public:
     }
 
 private:
-    Result Signal(VAddr addr, s32 count);
-    Result SignalAndIncrementIfEqual(VAddr addr, s32 value, s32 count);
-    Result SignalAndModifyByWaitingCountIfEqual(VAddr addr, s32 value, s32 count);
-    Result WaitIfLessThan(VAddr addr, s32 value, bool decrement, s64 timeout);
-    Result WaitIfEqual(VAddr addr, s32 value, s64 timeout);
+    Result Signal(uint64_t addr, s32 count);
+    Result SignalAndIncrementIfEqual(uint64_t addr, s32 value, s32 count);
+    Result SignalAndModifyByWaitingCountIfEqual(uint64_t addr, s32 value, s32 count);
+    Result WaitIfLessThan(uint64_t addr, s32 value, bool decrement, s64 timeout);
+    Result WaitIfEqual(uint64_t addr, s32 value, s64 timeout);
 
 private:
     ThreadTree m_tree;

src/core/hle/kernel/k_auto_object.h

@@ -182,8 +182,8 @@ public:
     explicit KAutoObjectWithList(KernelCore& kernel) : KAutoObject(kernel) {}
 
     static int Compare(const KAutoObjectWithList& lhs, const KAutoObjectWithList& rhs) {
-        const u64 lid = lhs.GetId();
-        const u64 rid = rhs.GetId();
+        const uintptr_t lid = reinterpret_cast<uintptr_t>(std::addressof(lhs));
+        const uintptr_t rid = reinterpret_cast<uintptr_t>(std::addressof(rhs));
 
         if (lid < rid) {
             return -1;

src/core/hle/kernel/k_client_session.cpp

@@ -29,7 +29,8 @@ Result KClientSession::SendSyncRequest() {
     SCOPE_EXIT({ request->Close(); });
 
     // Initialize the request.
-    request->Initialize(nullptr, GetCurrentThread(m_kernel).GetTlsAddress(), MessageBufferSize);
+    request->Initialize(nullptr, GetInteger(GetCurrentThread(m_kernel).GetTlsAddress()),
+                        MessageBufferSize);
 
     // Send the request.
     R_RETURN(m_parent->GetServerSession().OnRequest(request));