44 files changed, 2788 insertions, 145 deletions

diff --git a/src/common/CMakeLists.txt b/src/common/CMakeLists.txt
index 73bf626d4..635fb85c8 100644
--- a/src/common/CMakeLists.txt
+++ b/src/common/CMakeLists.txt
@@ -1,3 +1,6 @@
+# SPDX-FileCopyrightText: 2018 yuzu Emulator Project
+# SPDX-License-Identifier: GPL-2.0-or-later
+
 if (DEFINED ENV{AZURECIREPO})
   set(BUILD_REPOSITORY $ENV{AZURECIREPO})
 endif()
@@ -41,8 +44,10 @@ add_custom_command(OUTPUT scm_rev.cpp
 add_library(common STATIC
     algorithm.h
     alignment.h
+    announce_multiplayer_room.h
     assert.cpp
     assert.h
+    atomic_helpers.h
     atomic_ops.h
     detached_tasks.cpp
     detached_tasks.h
@@ -64,6 +69,7 @@ add_library(common STATIC
     expected.h
     fiber.cpp
     fiber.h
+    fixed_point.h
     fs/file.cpp
     fs/file.h
     fs/fs.cpp
@@ -109,6 +115,7 @@ add_library(common STATIC
     parent_of_member.h
     point.h
     quaternion.h
+    reader_writer_queue.h
     ring_buffer.h
     scm_rev.cpp
     scm_rev.h
@@ -117,6 +124,7 @@ add_library(common STATIC
     settings.h
     settings_input.cpp
     settings_input.h
+    socket_types.h
     spin_lock.cpp
     spin_lock.h
     stream.cpp
@@ -182,8 +190,9 @@ create_target_directory_groups(common)
 
 target_link_libraries(common PUBLIC ${Boost_LIBRARIES} fmt::fmt microprofile Threads::Threads)
 target_link_libraries(common PRIVATE lz4::lz4 xbyak)
-if (MSVC)
+if (TARGET zstd::zstd)
   target_link_libraries(common PRIVATE zstd::zstd)
 else()
-  target_link_libraries(common PRIVATE zstd)
+  target_link_libraries(common PRIVATE
+    $<IF:$<TARGET_EXISTS:zstd::libzstd_shared>,zstd::libzstd_shared,zstd::libzstd_static>)
 endif()
diff --git a/src/common/alignment.h b/src/common/alignment.h
index 8570c7d3c..7e897334b 100644
--- a/src/common/alignment.h
+++ b/src/common/alignment.h
@@ -1,4 +1,5 @@
-// This file is under the public domain.
+// SPDX-FileCopyrightText: 2014 Jannik Vogel <email@jannikvogel.de>
+// SPDX-License-Identifier: CC0-1.0
 
 #pragma once
 
diff --git a/src/common/announce_multiplayer_room.h b/src/common/announce_multiplayer_room.h
new file mode 100644
index 000000000..cb004e0eb
--- /dev/null
+++ b/src/common/announce_multiplayer_room.h
@@ -0,0 +1,139 @@
+// SPDX-FileCopyrightText: Copyright 2017 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <functional>
+#include <string>
+#include <vector>
+#include "common/common_types.h"
+#include "common/socket_types.h"
+#include "web_service/web_result.h"
+
+namespace AnnounceMultiplayerRoom {
+
+struct GameInfo {
+    std::string name{""};
+    u64 id{0};
+};
+
+struct Member {
+    std::string username;
+    std::string nickname;
+    std::string display_name;
+    std::string avatar_url;
+    Network::IPv4Address fake_ip;
+    GameInfo game;
+};
+
+struct RoomInformation {
+    std::string name;          ///< Name of the server
+    std::string description;   ///< Server description
+    u32 member_slots;          ///< Maximum number of members in this room
+    u16 port;                  ///< The port of this room
+    GameInfo preferred_game;   ///< Game to advertise that you want to play
+    std::string host_username; ///< Forum username of the host
+    bool enable_yuzu_mods;     ///< Allow yuzu Moderators to moderate on this room
+};
+
+struct Room {
+    RoomInformation information;
+
+    std::string id;
+    std::string verify_uid; ///< UID used for verification
+    std::string ip;
+    u32 net_version;
+    bool has_password;
+
+    std::vector<Member> members;
+};
+using RoomList = std::vector<Room>;
+
+/**
+ * A AnnounceMultiplayerRoom interface class. A backend to submit/get to/from a web service should
+ * implement this interface.
+ */
+class Backend {
+public:
+    virtual ~Backend() = default;
+
+    /**
+     * Sets the Information that gets used for the announce
+     * @param uid The Id of the room
+     * @param name The name of the room
+     * @param description The room description
+     * @param port The port of the room
+     * @param net_version The version of the libNetwork that gets used
+     * @param has_password True if the room is passowrd protected
+     * @param preferred_game The preferred game of the room
+     * @param preferred_game_id The title id of the preferred game
+     */
+    virtual void SetRoomInformation(const std::string& name, const std::string& description,
+                                    const u16 port, const u32 max_player, const u32 net_version,
+                                    const bool has_password, const GameInfo& preferred_game) = 0;
+    /**
+     * Adds a player information to the data that gets announced
+     * @param member The player to add
+     */
+    virtual void AddPlayer(const Member& member) = 0;
+
+    /**
+     * Updates the data in the announce service. Re-register the room when required.
+     * @result The result of the update attempt
+     */
+    virtual WebService::WebResult Update() = 0;
+
+    /**
+     * Registers the data in the announce service
+     * @result The result of the register attempt. When the result code is Success, A global Guid of
+     * the room which may be used for verification will be in the result's returned_data.
+     */
+    virtual WebService::WebResult Register() = 0;
+
+    /**
+     * Empties the stored players
+     */
+    virtual void ClearPlayers() = 0;
+
+    /**
+     * Get the room information from the announce service
+     * @result A list of all rooms the announce service has
+     */
+    virtual RoomList GetRoomList() = 0;
+
+    /**
+     * Sends a delete message to the announce service
+     */
+    virtual void Delete() = 0;
+};
+
+/**
+ * Empty implementation of AnnounceMultiplayerRoom interface that drops all data. Used when a
+ * functional backend implementation is not available.
+ */
+class NullBackend : public Backend {
+public:
+    ~NullBackend() = default;
+    void SetRoomInformation(const std::string& /*name*/, const std::string& /*description*/,
+                            const u16 /*port*/, const u32 /*max_player*/, const u32 /*net_version*/,
+                            const bool /*has_password*/,
+                            const GameInfo& /*preferred_game*/) override {}
+    void AddPlayer(const Member& /*member*/) override {}
+    WebService::WebResult Update() override {
+        return WebService::WebResult{WebService::WebResult::Code::NoWebservice,
+                                     "WebService is missing", ""};
+    }
+    WebService::WebResult Register() override {
+        return WebService::WebResult{WebService::WebResult::Code::NoWebservice,
+                                     "WebService is missing", ""};
+    }
+    void ClearPlayers() override {}
+    RoomList GetRoomList() override {
+        return RoomList{};
+    }
+
+    void Delete() override {}
+};
+
+} // namespace AnnounceMultiplayerRoom
diff --git a/src/common/atomic_helpers.h b/src/common/atomic_helpers.h
new file mode 100644
index 000000000..bef5015c1
--- /dev/null
+++ b/src/common/atomic_helpers.h
@@ -0,0 +1,775 @@
+// SPDX-FileCopyrightText: 2013-2016 Cameron Desrochers
+// SPDX-FileCopyrightText: 2015 Jeff Preshing
+// SPDX-License-Identifier: BSD-2-Clause AND Zlib
+
+// Distributed under the simplified BSD license (see the license file that
+// should have come with this header).
+// Uses Jeff Preshing's semaphore implementation (under the terms of its
+// separate zlib license, embedded below).
+
+#pragma once
+
+// Provides portable (VC++2010+, Intel ICC 13, GCC 4.7+, and anything C++11 compliant)
+// implementation of low-level memory barriers, plus a few semi-portable utility macros (for
+// inlining and alignment). Also has a basic atomic type (limited to hardware-supported atomics with
+// no memory ordering guarantees). Uses the AE_* prefix for macros (historical reasons), and the
+// "moodycamel" namespace for symbols.
+
+#include <cassert>
+#include <cerrno>
+#include <cstdint>
+#include <ctime>
+#include <type_traits>
+
+// Platform detection
+#if defined(__INTEL_COMPILER)
+#define AE_ICC
+#elif defined(_MSC_VER)
+#define AE_VCPP
+#elif defined(__GNUC__)
+#define AE_GCC
+#endif
+
+#if defined(_M_IA64) || defined(__ia64__)
+#define AE_ARCH_IA64
+#elif defined(_WIN64) || defined(__amd64__) || defined(_M_X64) || defined(__x86_64__)
+#define AE_ARCH_X64
+#elif defined(_M_IX86) || defined(__i386__)
+#define AE_ARCH_X86
+#elif defined(_M_PPC) || defined(__powerpc__)
+#define AE_ARCH_PPC
+#else
+#define AE_ARCH_UNKNOWN
+#endif
+
+// AE_UNUSED
+#define AE_UNUSED(x) ((void)x)
+
+// AE_NO_TSAN/AE_TSAN_ANNOTATE_*
+#if defined(__has_feature)
+#if __has_feature(thread_sanitizer)
+#if __cplusplus >= 201703L // inline variables require C++17
+namespace Common {
+inline int ae_tsan_global;
+}
+#define AE_TSAN_ANNOTATE_RELEASE()                                                                 \
+    AnnotateHappensBefore(__FILE__, __LINE__, (void*)(&::moodycamel::ae_tsan_global))
+#define AE_TSAN_ANNOTATE_ACQUIRE()                                                                 \
+    AnnotateHappensAfter(__FILE__, __LINE__, (void*)(&::moodycamel::ae_tsan_global))
+extern "C" void AnnotateHappensBefore(const char*, int, void*);
+extern "C" void AnnotateHappensAfter(const char*, int, void*);
+#else // when we can't work with tsan, attempt to disable its warnings
+#define AE_NO_TSAN __attribute__((no_sanitize("thread")))
+#endif
+#endif
+#endif
+#ifndef AE_NO_TSAN
+#define AE_NO_TSAN
+#endif
+#ifndef AE_TSAN_ANNOTATE_RELEASE
+#define AE_TSAN_ANNOTATE_RELEASE()
+#define AE_TSAN_ANNOTATE_ACQUIRE()
+#endif
+
+// AE_FORCEINLINE
+#if defined(AE_VCPP) || defined(AE_ICC)
+#define AE_FORCEINLINE __forceinline
+#elif defined(AE_GCC)
+//#define AE_FORCEINLINE __attribute__((always_inline))
+#define AE_FORCEINLINE inline
+#else
+#define AE_FORCEINLINE inline
+#endif
+
+// AE_ALIGN
+#if defined(AE_VCPP) || defined(AE_ICC)
+#define AE_ALIGN(x) __declspec(align(x))
+#elif defined(AE_GCC)
+#define AE_ALIGN(x) __attribute__((aligned(x)))
+#else
+// Assume GCC compliant syntax...
+#define AE_ALIGN(x) __attribute__((aligned(x)))
+#endif
+
+// Portable atomic fences implemented below:
+
+namespace Common {
+
+enum memory_order {
+    memory_order_relaxed,
+    memory_order_acquire,
+    memory_order_release,
+    memory_order_acq_rel,
+    memory_order_seq_cst,
+
+    // memory_order_sync: Forces a full sync:
+    // #LoadLoad, #LoadStore, #StoreStore, and most significantly, #StoreLoad
+    memory_order_sync = memory_order_seq_cst
+};
+
+} // namespace Common
+
+#if (defined(AE_VCPP) && (_MSC_VER < 1700 || defined(__cplusplus_cli))) ||                         \
+    (defined(AE_ICC) && __INTEL_COMPILER < 1600)
+// VS2010 and ICC13 don't support std::atomic_*_fence, implement our own fences
+
+#include <intrin.h>
+
+#if defined(AE_ARCH_X64) || defined(AE_ARCH_X86)
+#define AeFullSync _mm_mfence
+#define AeLiteSync _mm_mfence
+#elif defined(AE_ARCH_IA64)
+#define AeFullSync __mf
+#define AeLiteSync __mf
+#elif defined(AE_ARCH_PPC)
+#include <ppcintrinsics.h>
+#define AeFullSync __sync
+#define AeLiteSync __lwsync
+#endif
+
+#ifdef AE_VCPP
+#pragma warning(push)
+#pragma warning(disable : 4365) // Disable erroneous 'conversion from long to unsigned int,
+                                // signed/unsigned mismatch' error when using `assert`
+#ifdef __cplusplus_cli
+#pragma managed(push, off)
+#endif
+#endif
+
+namespace Common {
+
+AE_FORCEINLINE void compiler_fence(memory_order order) AE_NO_TSAN {
+    switch (order) {
+    case memory_order_relaxed:
+        break;
+    case memory_order_acquire:
+        _ReadBarrier();
+        break;
+    case memory_order_release:
+        _WriteBarrier();
+        break;
+    case memory_order_acq_rel:
+        _ReadWriteBarrier();
+        break;
+    case memory_order_seq_cst:
+        _ReadWriteBarrier();
+        break;
+    default:
+        assert(false);
+    }
+}
+
+// x86/x64 have a strong memory model -- all loads and stores have
+// acquire and release semantics automatically (so only need compiler
+// barriers for those).
+#if defined(AE_ARCH_X86) || defined(AE_ARCH_X64)
+AE_FORCEINLINE void fence(memory_order order) AE_NO_TSAN {
+    switch (order) {
+    case memory_order_relaxed:
+        break;
+    case memory_order_acquire:
+        _ReadBarrier();
+        break;
+    case memory_order_release:
+        _WriteBarrier();
+        break;
+    case memory_order_acq_rel:
+        _ReadWriteBarrier();
+        break;
+    case memory_order_seq_cst:
+        _ReadWriteBarrier();
+        AeFullSync();
+        _ReadWriteBarrier();
+        break;
+    default:
+        assert(false);
+    }
+}
+#else
+AE_FORCEINLINE void fence(memory_order order) AE_NO_TSAN {
+    // Non-specialized arch, use heavier memory barriers everywhere just in case :-(
+    switch (order) {
+    case memory_order_relaxed:
+        break;
+    case memory_order_acquire:
+        _ReadBarrier();
+        AeLiteSync();
+        _ReadBarrier();
+        break;
+    case memory_order_release:
+        _WriteBarrier();
+        AeLiteSync();
+        _WriteBarrier();
+        break;
+    case memory_order_acq_rel:
+        _ReadWriteBarrier();
+        AeLiteSync();
+        _ReadWriteBarrier();
+        break;
+    case memory_order_seq_cst:
+        _ReadWriteBarrier();
+        AeFullSync();
+        _ReadWriteBarrier();
+        break;
+    default:
+        assert(false);
+    }
+}
+#endif
+} // namespace Common
+#else
+// Use standard library of atomics
+#include <atomic>
+
+namespace Common {
+
+AE_FORCEINLINE void compiler_fence(memory_order order) AE_NO_TSAN {
+    switch (order) {
+    case memory_order_relaxed:
+        break;
+    case memory_order_acquire:
+        std::atomic_signal_fence(std::memory_order_acquire);
+        break;
+    case memory_order_release:
+        std::atomic_signal_fence(std::memory_order_release);
+        break;
+    case memory_order_acq_rel:
+        std::atomic_signal_fence(std::memory_order_acq_rel);
+        break;
+    case memory_order_seq_cst:
+        std::atomic_signal_fence(std::memory_order_seq_cst);
+        break;
+    default:
+        assert(false);
+    }
+}
+
+AE_FORCEINLINE void fence(memory_order order) AE_NO_TSAN {
+    switch (order) {
+    case memory_order_relaxed:
+        break;
+    case memory_order_acquire:
+        AE_TSAN_ANNOTATE_ACQUIRE();
+        std::atomic_thread_fence(std::memory_order_acquire);
+        break;
+    case memory_order_release:
+        AE_TSAN_ANNOTATE_RELEASE();
+        std::atomic_thread_fence(std::memory_order_release);
+        break;
+    case memory_order_acq_rel:
+        AE_TSAN_ANNOTATE_ACQUIRE();
+        AE_TSAN_ANNOTATE_RELEASE();
+        std::atomic_thread_fence(std::memory_order_acq_rel);
+        break;
+    case memory_order_seq_cst:
+        AE_TSAN_ANNOTATE_ACQUIRE();
+        AE_TSAN_ANNOTATE_RELEASE();
+        std::atomic_thread_fence(std::memory_order_seq_cst);
+        break;
+    default:
+        assert(false);
+    }
+}
+
+} // namespace Common
+
+#endif
+
+#if !defined(AE_VCPP) || (_MSC_VER >= 1700 && !defined(__cplusplus_cli))
+#define AE_USE_STD_ATOMIC_FOR_WEAK_ATOMIC
+#endif
+
+#ifdef AE_USE_STD_ATOMIC_FOR_WEAK_ATOMIC
+#include <atomic>
+#endif
+#include <utility>
+
+// WARNING: *NOT* A REPLACEMENT FOR std::atomic. READ CAREFULLY:
+// Provides basic support for atomic variables -- no memory ordering guarantees are provided.
+// The guarantee of atomicity is only made for types that already have atomic load and store
+// guarantees at the hardware level -- on most platforms this generally means aligned pointers and
+// integers (only).
+namespace Common {
+template <typename T>
+class weak_atomic {
+public:
+    AE_NO_TSAN weak_atomic() : value() {}
+#ifdef AE_VCPP
+#pragma warning(push)
+#pragma warning(disable : 4100) // Get rid of (erroneous) 'unreferenced formal parameter' warning
+#endif
+    template <typename U>
+    AE_NO_TSAN weak_atomic(U&& x) : value(std::forward<U>(x)) {}
+#ifdef __cplusplus_cli
+    // Work around bug with universal reference/nullptr combination that only appears when /clr is
+    // on
+    AE_NO_TSAN weak_atomic(nullptr_t) : value(nullptr) {}
+#endif
+    AE_NO_TSAN weak_atomic(weak_atomic const& other) : value(other.load()) {}
+    AE_NO_TSAN weak_atomic(weak_atomic&& other) : value(std::move(other.load())) {}
+#ifdef AE_VCPP
+#pragma warning(pop)
+#endif
+
+    AE_FORCEINLINE operator T() const AE_NO_TSAN {
+        return load();
+    }
+
+#ifndef AE_USE_STD_ATOMIC_FOR_WEAK_ATOMIC
+    template <typename U>
+    AE_FORCEINLINE weak_atomic const& operator=(U&& x) AE_NO_TSAN {
+        value = std::forward<U>(x);
+        return *this;
+    }
+    AE_FORCEINLINE weak_atomic const& operator=(weak_atomic const& other) AE_NO_TSAN {
+        value = other.value;
+        return *this;
+    }
+
+    AE_FORCEINLINE T load() const AE_NO_TSAN {
+        return value;
+    }
+
+    AE_FORCEINLINE T fetch_add_acquire(T increment) AE_NO_TSAN {
+#if defined(AE_ARCH_X64) || defined(AE_ARCH_X86)
+        if (sizeof(T) == 4)
+            return _InterlockedExchangeAdd((long volatile*)&value, (long)increment);
+#if defined(_M_AMD64)
+        else if (sizeof(T) == 8)
+            return _InterlockedExchangeAdd64((long long volatile*)&value, (long long)increment);
+#endif
+#else
+#error Unsupported platform
+#endif
+        assert(false && "T must be either a 32 or 64 bit type");
+        return value;
+    }
+
+    AE_FORCEINLINE T fetch_add_release(T increment) AE_NO_TSAN {
+#if defined(AE_ARCH_X64) || defined(AE_ARCH_X86)
+        if (sizeof(T) == 4)
+            return _InterlockedExchangeAdd((long volatile*)&value, (long)increment);
+#if defined(_M_AMD64)
+        else if (sizeof(T) == 8)
+            return _InterlockedExchangeAdd64((long long volatile*)&value, (long long)increment);
+#endif
+#else
+#error Unsupported platform
+#endif
+        assert(false && "T must be either a 32 or 64 bit type");
+        return value;
+    }
+#else
+    template <typename U>
+    AE_FORCEINLINE weak_atomic const& operator=(U&& x) AE_NO_TSAN {
+        value.store(std::forward<U>(x), std::memory_order_relaxed);
+        return *this;
+    }
+
+    AE_FORCEINLINE weak_atomic const& operator=(weak_atomic const& other) AE_NO_TSAN {
+        value.store(other.value.load(std::memory_order_relaxed), std::memory_order_relaxed);
+        return *this;
+    }
+
+    AE_FORCEINLINE T load() const AE_NO_TSAN {
+        return value.load(std::memory_order_relaxed);
+    }
+
+    AE_FORCEINLINE T fetch_add_acquire(T increment) AE_NO_TSAN {
+        return value.fetch_add(increment, std::memory_order_acquire);
+    }
+
+    AE_FORCEINLINE T fetch_add_release(T increment) AE_NO_TSAN {
+        return value.fetch_add(increment, std::memory_order_release);
+    }
+#endif
+
+private:
+#ifndef AE_USE_STD_ATOMIC_FOR_WEAK_ATOMIC
+    // No std::atomic support, but still need to circumvent compiler optimizations.
+    // `volatile` will make memory access slow, but is guaranteed to be reliable.
+    volatile T value;
+#else
+    std::atomic<T> value;
+#endif
+};
+
+} // namespace Common
+
+// Portable single-producer, single-consumer semaphore below:
+
+#if defined(_WIN32)
+// Avoid including windows.h in a header; we only need a handful of
+// items, so we'll redeclare them here (this is relatively safe since
+// the API generally has to remain stable between Windows versions).
+// I know this is an ugly hack but it still beats polluting the global
+// namespace with thousands of generic names or adding a .cpp for nothing.
+extern "C" {
+struct _SECURITY_ATTRIBUTES;
+__declspec(dllimport) void* __stdcall CreateSemaphoreW(_SECURITY_ATTRIBUTES* lpSemaphoreAttributes,
+                                                       long lInitialCount, long lMaximumCount,
+                                                       const wchar_t* lpName);
+__declspec(dllimport) int __stdcall CloseHandle(void* hObject);
+__declspec(dllimport) unsigned long __stdcall WaitForSingleObject(void* hHandle,
+                                                                  unsigned long dwMilliseconds);
+__declspec(dllimport) int __stdcall ReleaseSemaphore(void* hSemaphore, long lReleaseCount,
+                                                     long* lpPreviousCount);
+}
+#elif defined(__MACH__)
+#include <mach/mach.h>
+#elif defined(__unix__)
+#include <semaphore.h>
+#elif defined(FREERTOS)
+#include <FreeRTOS.h>
+#include <semphr.h>
+#include <task.h>
+#endif
+
+namespace Common {
+// Code in the spsc_sema namespace below is an adaptation of Jeff Preshing's
+// portable + lightweight semaphore implementations, originally from
+// https://github.com/preshing/cpp11-on-multicore/blob/master/common/sema.h
+// LICENSE:
+// Copyright (c) 2015 Jeff Preshing
+//
+// This software is provided 'as-is', without any express or implied
+// warranty. In no event will the authors be held liable for any damages
+// arising from the use of this software.
+//
+// Permission is granted to anyone to use this software for any purpose,
+// including commercial applications, and to alter it and redistribute it
+// freely, subject to the following restrictions:
+//
+// 1. The origin of this software must not be misrepresented; you must not
+//    claim that you wrote the original software. If you use this software
+//    in a product, an acknowledgement in the product documentation would be
+//    appreciated but is not required.
+// 2. Altered source versions must be plainly marked as such, and must not be
+//    misrepresented as being the original software.
+// 3. This notice may not be removed or altered from any source distribution.
+namespace spsc_sema {
+#if defined(_WIN32)
+class Semaphore {
+private:
+    void* m_hSema;
+
+    Semaphore(const Semaphore& other);
+    Semaphore& operator=(const Semaphore& other);
+
+public:
+    AE_NO_TSAN Semaphore(int initialCount = 0) : m_hSema() {
+        assert(initialCount >= 0);
+        const long maxLong = 0x7fffffff;
+        m_hSema = CreateSemaphoreW(nullptr, initialCount, maxLong, nullptr);
+        assert(m_hSema);
+    }
+
+    AE_NO_TSAN ~Semaphore() {
+        CloseHandle(m_hSema);
+    }
+
+    bool wait() AE_NO_TSAN {
+        const unsigned long infinite = 0xffffffff;
+        return WaitForSingleObject(m_hSema, infinite) == 0;
+    }
+
+    bool try_wait() AE_NO_TSAN {
+        return WaitForSingleObject(m_hSema, 0) == 0;
+    }
+
+    bool timed_wait(std::uint64_t usecs) AE_NO_TSAN {
+        return WaitForSingleObject(m_hSema, (unsigned long)(usecs / 1000)) == 0;
+    }
+
+    void signal(int count = 1) AE_NO_TSAN {
+        while (!ReleaseSemaphore(m_hSema, count, nullptr))
+            ;
+    }
+};
+#elif defined(__MACH__)
+//---------------------------------------------------------
+// Semaphore (Apple iOS and OSX)
+// Can't use POSIX semaphores due to
+// http://lists.apple.com/archives/darwin-kernel/2009/Apr/msg00010.html
+//---------------------------------------------------------
+class Semaphore {
+private:
+    semaphore_t m_sema;
+
+    Semaphore(const Semaphore& other);
+    Semaphore& operator=(const Semaphore& other);
+
+public:
+    AE_NO_TSAN Semaphore(int initialCount = 0) : m_sema() {
+        assert(initialCount >= 0);
+        kern_return_t rc =
+            semaphore_create(mach_task_self(), &m_sema, SYNC_POLICY_FIFO, initialCount);
+        assert(rc == KERN_SUCCESS);
+        AE_UNUSED(rc);
+    }
+
+    AE_NO_TSAN ~Semaphore() {
+        semaphore_destroy(mach_task_self(), m_sema);
+    }
+
+    bool wait() AE_NO_TSAN {
+        return semaphore_wait(m_sema) == KERN_SUCCESS;
+    }
+
+    bool try_wait() AE_NO_TSAN {
+        return timed_wait(0);
+    }
+
+    bool timed_wait(std::uint64_t timeout_usecs) AE_NO_TSAN {
+        mach_timespec_t ts;
+        ts.tv_sec = static_cast<unsigned int>(timeout_usecs / 1000000);
+        ts.tv_nsec = static_cast<int>((timeout_usecs % 1000000) * 1000);
+
+        // added in OSX 10.10:
+        // https://developer.apple.com/library/prerelease/mac/documentation/General/Reference/APIDiffsMacOSX10_10SeedDiff/modules/Darwin.html
+        kern_return_t rc = semaphore_timedwait(m_sema, ts);
+        return rc == KERN_SUCCESS;
+    }
+
+    void signal() AE_NO_TSAN {
+        while (semaphore_signal(m_sema) != KERN_SUCCESS)
+            ;
+    }
+
+    void signal(int count) AE_NO_TSAN {
+        while (count-- > 0) {
+            while (semaphore_signal(m_sema) != KERN_SUCCESS)
+                ;
+        }
+    }
+};
+#elif defined(__unix__)
+//---------------------------------------------------------
+// Semaphore (POSIX, Linux)
+//---------------------------------------------------------
+class Semaphore {
+private:
+    sem_t m_sema;
+
+    Semaphore(const Semaphore& other);
+    Semaphore& operator=(const Semaphore& other);
+
+public:
+    AE_NO_TSAN Semaphore(int initialCount = 0) : m_sema() {
+        assert(initialCount >= 0);
+        int rc = sem_init(&m_sema, 0, static_cast<unsigned int>(initialCount));
+        assert(rc == 0);
+        AE_UNUSED(rc);
+    }
+
+    AE_NO_TSAN ~Semaphore() {
+        sem_destroy(&m_sema);
+    }
+
+    bool wait() AE_NO_TSAN {
+        // http://stackoverflow.com/questions/2013181/gdb-causes-sem-wait-to-fail-with-eintr-error
+        int rc;
+        do {
+            rc = sem_wait(&m_sema);
+        } while (rc == -1 && errno == EINTR);
+        return rc == 0;
+    }
+
+    bool try_wait() AE_NO_TSAN {
+        int rc;
+        do {
+            rc = sem_trywait(&m_sema);
+        } while (rc == -1 && errno == EINTR);
+        return rc == 0;
+    }
+
+    bool timed_wait(std::uint64_t usecs) AE_NO_TSAN {
+        struct timespec ts;
+        const int usecs_in_1_sec = 1000000;
+        const int nsecs_in_1_sec = 1000000000;
+        clock_gettime(CLOCK_REALTIME, &ts);
+        ts.tv_sec += static_cast<time_t>(usecs / usecs_in_1_sec);
+        ts.tv_nsec += static_cast<long>(usecs % usecs_in_1_sec) * 1000;
+        // sem_timedwait bombs if you have more than 1e9 in tv_nsec
+        // so we have to clean things up before passing it in
+        if (ts.tv_nsec >= nsecs_in_1_sec) {
+            ts.tv_nsec -= nsecs_in_1_sec;
+            ++ts.tv_sec;
+        }
+
+        int rc;
+        do {
+            rc = sem_timedwait(&m_sema, &ts);
+        } while (rc == -1 && errno == EINTR);
+        return rc == 0;
+    }
+
+    void signal() AE_NO_TSAN {
+        while (sem_post(&m_sema) == -1)
+            ;
+    }
+
+    void signal(int count) AE_NO_TSAN {
+        while (count-- > 0) {
+            while (sem_post(&m_sema) == -1)
+                ;
+        }
+    }
+};
+#elif defined(FREERTOS)
+//---------------------------------------------------------
+// Semaphore (FreeRTOS)
+//---------------------------------------------------------
+class Semaphore {
+private:
+    SemaphoreHandle_t m_sema;
+
+    Semaphore(const Semaphore& other);
+    Semaphore& operator=(const Semaphore& other);
+
+public:
+    AE_NO_TSAN Semaphore(int initialCount = 0) : m_sema() {
+        assert(initialCount >= 0);
+        m_sema = xSemaphoreCreateCounting(static_cast<UBaseType_t>(~0ull),
+                                          static_cast<UBaseType_t>(initialCount));
+        assert(m_sema);
+    }
+
+    AE_NO_TSAN ~Semaphore() {
+        vSemaphoreDelete(m_sema);
+    }
+
+    bool wait() AE_NO_TSAN {
+        return xSemaphoreTake(m_sema, portMAX_DELAY) == pdTRUE;
+    }
+
+    bool try_wait() AE_NO_TSAN {
+        // Note: In an ISR context, if this causes a task to unblock,
+        // the caller won't know about it
+        if (xPortIsInsideInterrupt())
+            return xSemaphoreTakeFromISR(m_sema, NULL) == pdTRUE;
+        return xSemaphoreTake(m_sema, 0) == pdTRUE;
+    }
+
+    bool timed_wait(std::uint64_t usecs) AE_NO_TSAN {
+        std::uint64_t msecs = usecs / 1000;
+        TickType_t ticks = static_cast<TickType_t>(msecs / portTICK_PERIOD_MS);
+        if (ticks == 0)
+            return try_wait();
+        return xSemaphoreTake(m_sema, ticks) == pdTRUE;
+    }
+
+    void signal() AE_NO_TSAN {
+        // Note: In an ISR context, if this causes a task to unblock,
+        // the caller won't know about it
+        BaseType_t rc;
+        if (xPortIsInsideInterrupt())
+            rc = xSemaphoreGiveFromISR(m_sema, NULL);
+        else
+            rc = xSemaphoreGive(m_sema);
+        assert(rc == pdTRUE);
+        AE_UNUSED(rc);
+    }
+
+    void signal(int count) AE_NO_TSAN {
+        while (count-- > 0)
+            signal();
+    }
+};
+#else
+#error Unsupported platform! (No semaphore wrapper available)
+#endif
+
+//---------------------------------------------------------
+// LightweightSemaphore
+//---------------------------------------------------------
+class LightweightSemaphore {
+public:
+    typedef std::make_signed<std::size_t>::type ssize_t;
+
+private:
+    weak_atomic<ssize_t> m_count;
+    Semaphore m_sema;
+
+    bool waitWithPartialSpinning(std::int64_t timeout_usecs = -1) AE_NO_TSAN {
+        ssize_t oldCount;
+        // Is there a better way to set the initial spin count?
+        // If we lower it to 1000, testBenaphore becomes 15x slower on my Core i7-5930K Windows PC,
+        // as threads start hitting the kernel semaphore.
+        int spin = 1024;
+        while (--spin >= 0) {
+            if (m_count.load() > 0) {
+                m_count.fetch_add_acquire(-1);
+                return true;
+            }
+            compiler_fence(memory_order_acquire); // Prevent the compiler from collapsing the loop.
+        }
+        oldCount = m_count.fetch_add_acquire(-1);
+        if (oldCount > 0)
+            return true;
+        if (timeout_usecs < 0) {
+            if (m_sema.wait())
+                return true;
+        }
+        if (timeout_usecs > 0 && m_sema.timed_wait(static_cast<uint64_t>(timeout_usecs)))
+            return true;
+        // At this point, we've timed out waiting for the semaphore, but the
+        // count is still decremented indicating we may still be waiting on
+        // it. So we have to re-adjust the count, but only if the semaphore
+        // wasn't signaled enough times for us too since then. If it was, we
+        // need to release the semaphore too.
+        while (true) {
+            oldCount = m_count.fetch_add_release(1);
+            if (oldCount < 0)
+                return false; // successfully restored things to the way they were
+            // Oh, the producer thread just signaled the semaphore after all. Try again:
+            oldCount = m_count.fetch_add_acquire(-1);
+            if (oldCount > 0 && m_sema.try_wait())
+                return true;
+        }
+    }
+
+public:
+    AE_NO_TSAN LightweightSemaphore(ssize_t initialCount = 0) : m_count(initialCount), m_sema() {
+        assert(initialCount >= 0);
+    }
+
+    bool tryWait() AE_NO_TSAN {
+        if (m_count.load() > 0) {
+            m_count.fetch_add_acquire(-1);
+            return true;
+        }
+        return false;
+    }
+
+    bool wait() AE_NO_TSAN {
+        return tryWait() || waitWithPartialSpinning();
+    }
+
+    bool wait(std::int64_t timeout_usecs) AE_NO_TSAN {
+        return tryWait() || waitWithPartialSpinning(timeout_usecs);
+    }
+
+    void signal(ssize_t count = 1) AE_NO_TSAN {
+        assert(count >= 0);
+        ssize_t oldCount = m_count.fetch_add_release(count);
+        assert(oldCount >= -1);
+        if (oldCount < 0) {
+            m_sema.signal(1);
+        }
+    }
+
+    std::size_t availableApprox() const AE_NO_TSAN {
+        ssize_t count = m_count.load();
+        return count > 0 ? static_cast<std::size_t>(count) : 0;
+    }
+};
+} // namespace spsc_sema
+} // namespace Common
+
+#if defined(AE_VCPP) && (_MSC_VER < 1700 || defined(__cplusplus_cli))
+#pragma warning(pop)
+#ifdef __cplusplus_cli
+#pragma managed(pop)
+#endif
+#endif
diff --git a/src/common/bit_field.h b/src/common/bit_field.h
index 16d805694..7e1df62b1 100644
--- a/src/common/bit_field.h
+++ b/src/common/bit_field.h
@@ -146,7 +146,16 @@ public:
     }
 
     constexpr void Assign(const T& value) {
+#ifdef _MSC_VER
         storage = static_cast<StorageType>((storage & ~mask) | FormatValue(value));
+#else
+        // Explicitly reload with memcpy to avoid compiler aliasing quirks
+        // regarding optimization: GCC/Clang clobber chained stores to
+        // different bitfields in the same struct with the last value.
+        StorageTypeWithEndian storage_;
+        std::memcpy(&storage_, &storage, sizeof(storage_));
+        storage = static_cast<StorageType>((storage_ & ~mask) | FormatValue(value));
+#endif
     }
 
     [[nodiscard]] constexpr T Value() const {
diff --git a/src/common/common_funcs.h b/src/common/common_funcs.h
index adc31c758..e1e2a90fc 100644
--- a/src/common/common_funcs.h
+++ b/src/common/common_funcs.h
@@ -18,14 +18,16 @@
 /// Helper macros to insert unused bytes or words to properly align structs. These values will be
 /// zero-initialized.
 #define INSERT_PADDING_BYTES(num_bytes)                                                            \
-    std::array<u8, num_bytes> CONCAT2(pad, __LINE__) {}
+    [[maybe_unused]] std::array<u8, num_bytes> CONCAT2(pad, __LINE__) {}
 #define INSERT_PADDING_WORDS(num_words)                                                            \
-    std::array<u32, num_words> CONCAT2(pad, __LINE__) {}
+    [[maybe_unused]] std::array<u32, num_words> CONCAT2(pad, __LINE__) {}
 
 /// These are similar to the INSERT_PADDING_* macros but do not zero-initialize the contents.
 /// This keeps the structure trivial to construct.
-#define INSERT_PADDING_BYTES_NOINIT(num_bytes) std::array<u8, num_bytes> CONCAT2(pad, __LINE__)
-#define INSERT_PADDING_WORDS_NOINIT(num_words) std::array<u32, num_words> CONCAT2(pad, __LINE__)
+#define INSERT_PADDING_BYTES_NOINIT(num_bytes)                                                     \
+    [[maybe_unused]] std::array<u8, num_bytes> CONCAT2(pad, __LINE__)
+#define INSERT_PADDING_WORDS_NOINIT(num_words)                                                     \
+    [[maybe_unused]] std::array<u32, num_words> CONCAT2(pad, __LINE__)
 
 #ifndef _MSC_VER
 
diff --git a/src/common/detached_tasks.cpp b/src/common/detached_tasks.cpp
index ec31d0b88..da64848da 100644
--- a/src/common/detached_tasks.cpp
+++ b/src/common/detached_tasks.cpp
@@ -1,6 +1,5 @@
-// Copyright 2018 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2018 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #include <thread>
 #include "common/assert.h"
diff --git a/src/common/detached_tasks.h b/src/common/detached_tasks.h
index 5dd8fc27b..416a2d7f3 100644
--- a/src/common/detached_tasks.h
+++ b/src/common/detached_tasks.h
@@ -1,6 +1,5 @@
-// Copyright 2018 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2018 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
diff --git a/src/common/error.cpp b/src/common/error.cpp
index d4455e310..ddb03bd45 100644
--- a/src/common/error.cpp
+++ b/src/common/error.cpp
@@ -1,6 +1,6 @@
-// Copyright 2013 Dolphin Emulator Project / 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2013 Dolphin Emulator Project
+// SPDX-FileCopyrightText: 2014 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #include <cstddef>
 #ifdef _WIN32
diff --git a/src/common/error.h b/src/common/error.h
index e084d4b0f..62a3bd835 100644
--- a/src/common/error.h
+++ b/src/common/error.h
@@ -1,6 +1,6 @@
-// Copyright 2013 Dolphin Emulator Project / 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2013 Dolphin Emulator Project
+// SPDX-FileCopyrightText: 2014 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
diff --git a/src/common/fixed_point.h b/src/common/fixed_point.h
new file mode 100644
index 000000000..4a0f72cc9
--- /dev/null
+++ b/src/common/fixed_point.h
@@ -0,0 +1,706 @@
+// SPDX-FileCopyrightText: 2015 Evan Teran
+// SPDX-License-Identifier: MIT
+
+// From: https://github.com/eteran/cpp-utilities/blob/master/fixed/include/cpp-utilities/fixed.h
+// See also: http://stackoverflow.com/questions/79677/whats-the-best-way-to-do-fixed-point-math
+
+#ifndef FIXED_H_
+#define FIXED_H_
+
+#if __cplusplus >= 201402L
+#define CONSTEXPR14 constexpr
+#else
+#define CONSTEXPR14
+#endif
+
+#include <cstddef> // for size_t
+#include <cstdint>
+#include <exception>
+#include <ostream>
+#include <type_traits>
+
+namespace Common {
+
+template <size_t I, size_t F>
+class FixedPoint;
+
+namespace detail {
+
+// helper templates to make magic with types :)
+// these allow us to determine resonable types from
+// a desired size, they also let us infer the next largest type
+// from a type which is nice for the division op
+template <size_t T>
+struct type_from_size {
+    using value_type = void;
+    using unsigned_type = void;
+    using signed_type = void;
+    static constexpr bool is_specialized = false;
+};
+
+#if defined(__GNUC__) && defined(__x86_64__) && !defined(__STRICT_ANSI__)
+template <>
+struct type_from_size<128> {
+    static constexpr bool is_specialized = true;
+    static constexpr size_t size = 128;
+
+    using value_type = __int128;
+    using unsigned_type = unsigned __int128;
+    using signed_type = __int128;
+    using next_size = type_from_size<256>;
+};
+#endif
+
+template <>
+struct type_from_size<64> {
+    static constexpr bool is_specialized = true;
+    static constexpr size_t size = 64;
+
+    using value_type = int64_t;
+    using unsigned_type = std::make_unsigned<value_type>::type;
+    using signed_type = std::make_signed<value_type>::type;
+    using next_size = type_from_size<128>;
+};
+
+template <>
+struct type_from_size<32> {
+    static constexpr bool is_specialized = true;
+    static constexpr size_t size = 32;
+
+    using value_type = int32_t;
+    using unsigned_type = std::make_unsigned<value_type>::type;
+    using signed_type = std::make_signed<value_type>::type;
+    using next_size = type_from_size<64>;
+};
+
+template <>
+struct type_from_size<16> {
+    static constexpr bool is_specialized = true;
+    static constexpr size_t size = 16;
+
+    using value_type = int16_t;
+    using unsigned_type = std::make_unsigned<value_type>::type;
+    using signed_type = std::make_signed<value_type>::type;
+    using next_size = type_from_size<32>;
+};
+
+template <>
+struct type_from_size<8> {
+    static constexpr bool is_specialized = true;
+    static constexpr size_t size = 8;
+
+    using value_type = int8_t;
+    using unsigned_type = std::make_unsigned<value_type>::type;
+    using signed_type = std::make_signed<value_type>::type;
+    using next_size = type_from_size<16>;
+};
+
+// this is to assist in adding support for non-native base
+// types (for adding big-int support), this should be fine
+// unless your bit-int class doesn't nicely support casting
+template <class B, class N>
+constexpr B next_to_base(N rhs) {
+    return static_cast<B>(rhs);
+}
+
+struct divide_by_zero : std::exception {};
+
+template <size_t I, size_t F>
+CONSTEXPR14 FixedPoint<I, F> divide(
+    FixedPoint<I, F> numerator, FixedPoint<I, F> denominator, FixedPoint<I, F>& remainder,
+    typename std::enable_if<type_from_size<I + F>::next_size::is_specialized>::type* = nullptr) {
+
+    using next_type = typename FixedPoint<I, F>::next_type;
+    using base_type = typename FixedPoint<I, F>::base_type;
+    constexpr size_t fractional_bits = FixedPoint<I, F>::fractional_bits;
+
+    next_type t(numerator.to_raw());
+    t <<= fractional_bits;
+
+    FixedPoint<I, F> quotient;
+
+    quotient = FixedPoint<I, F>::from_base(next_to_base<base_type>(t / denominator.to_raw()));
+    remainder = FixedPoint<I, F>::from_base(next_to_base<base_type>(t % denominator.to_raw()));
+
+    return quotient;
+}
+
+template <size_t I, size_t F>
+CONSTEXPR14 FixedPoint<I, F> divide(
+    FixedPoint<I, F> numerator, FixedPoint<I, F> denominator, FixedPoint<I, F>& remainder,
+    typename std::enable_if<!type_from_size<I + F>::next_size::is_specialized>::type* = nullptr) {
+
+    using unsigned_type = typename FixedPoint<I, F>::unsigned_type;
+
+    constexpr int bits = FixedPoint<I, F>::total_bits;
+
+    if (denominator == 0) {
+        throw divide_by_zero();
+    } else {
+
+        int sign = 0;
+
+        FixedPoint<I, F> quotient;
+
+        if (numerator < 0) {
+            sign ^= 1;
+            numerator = -numerator;
+        }
+
+        if (denominator < 0) {
+            sign ^= 1;
+            denominator = -denominator;
+        }
+
+        unsigned_type n = numerator.to_raw();
+        unsigned_type d = denominator.to_raw();
+        unsigned_type x = 1;
+        unsigned_type answer = 0;
+
+        // egyptian division algorithm
+        while ((n >= d) && (((d >> (bits - 1)) & 1) == 0)) {
+            x <<= 1;
+            d <<= 1;
+        }
+
+        while (x != 0) {
+            if (n >= d) {
+                n -= d;
+                answer += x;
+            }
+
+            x >>= 1;
+            d >>= 1;
+        }
+
+        unsigned_type l1 = n;
+        unsigned_type l2 = denominator.to_raw();
+
+        // calculate the lower bits (needs to be unsigned)
+        while (l1 >> (bits - F) > 0) {
+            l1 >>= 1;
+            l2 >>= 1;
+        }
+        const unsigned_type lo = (l1 << F) / l2;
+
+        quotient = FixedPoint<I, F>::from_base((answer << F) | lo);
+        remainder = n;
+
+        if (sign) {
+            quotient = -quotient;
+        }
+
+        return quotient;
+    }
+}
+
+// this is the usual implementation of multiplication
+template <size_t I, size_t F>
+CONSTEXPR14 FixedPoint<I, F> multiply(
+    FixedPoint<I, F> lhs, FixedPoint<I, F> rhs,
+    typename std::enable_if<type_from_size<I + F>::next_size::is_specialized>::type* = nullptr) {
+
+    using next_type = typename FixedPoint<I, F>::next_type;
+    using base_type = typename FixedPoint<I, F>::base_type;
+
+    constexpr size_t fractional_bits = FixedPoint<I, F>::fractional_bits;
+
+    next_type t(static_cast<next_type>(lhs.to_raw()) * static_cast<next_type>(rhs.to_raw()));
+    t >>= fractional_bits;
+
+    return FixedPoint<I, F>::from_base(next_to_base<base_type>(t));
+}
+
+// this is the fall back version we use when we don't have a next size
+// it is slightly slower, but is more robust since it doesn't
+// require and upgraded type
+template <size_t I, size_t F>
+CONSTEXPR14 FixedPoint<I, F> multiply(
+    FixedPoint<I, F> lhs, FixedPoint<I, F> rhs,
+    typename std::enable_if<!type_from_size<I + F>::next_size::is_specialized>::type* = nullptr) {
+
+    using base_type = typename FixedPoint<I, F>::base_type;
+
+    constexpr size_t fractional_bits = FixedPoint<I, F>::fractional_bits;
+    constexpr base_type integer_mask = FixedPoint<I, F>::integer_mask;
+    constexpr base_type fractional_mask = FixedPoint<I, F>::fractional_mask;
+
+    // more costly but doesn't need a larger type
+    const base_type a_hi = (lhs.to_raw() & integer_mask) >> fractional_bits;
+    const base_type b_hi = (rhs.to_raw() & integer_mask) >> fractional_bits;
+    const base_type a_lo = (lhs.to_raw() & fractional_mask);
+    const base_type b_lo = (rhs.to_raw() & fractional_mask);
+
+    const base_type x1 = a_hi * b_hi;
+    const base_type x2 = a_hi * b_lo;
+    const base_type x3 = a_lo * b_hi;
+    const base_type x4 = a_lo * b_lo;
+
+    return FixedPoint<I, F>::from_base((x1 << fractional_bits) + (x3 + x2) +
+                                       (x4 >> fractional_bits));
+}
+} // namespace detail
+
+template <size_t I, size_t F>
+class FixedPoint {
+    static_assert(detail::type_from_size<I + F>::is_specialized, "invalid combination of sizes");
+
+public:
+    static constexpr size_t fractional_bits = F;
+    static constexpr size_t integer_bits = I;
+    static constexpr size_t total_bits = I + F;
+
+    using base_type_info = detail::type_from_size<total_bits>;
+
+    using base_type = typename base_type_info::value_type;
+    using next_type = typename base_type_info::next_size::value_type;
+    using unsigned_type = typename base_type_info::unsigned_type;
+
+public:
+#ifdef __GNUC__
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Woverflow"
+#endif
+    static constexpr base_type fractional_mask =
+        ~(static_cast<unsigned_type>(~base_type(0)) << fractional_bits);
+    static constexpr base_type integer_mask = ~fractional_mask;
+#ifdef __GNUC__
+#pragma GCC diagnostic pop
+#endif
+
+public:
+    static constexpr base_type one = base_type(1) << fractional_bits;
+
+public: // constructors
+    FixedPoint() = default;
+    FixedPoint(const FixedPoint&) = default;
+    FixedPoint(FixedPoint&&) = default;
+    FixedPoint& operator=(const FixedPoint&) = default;
+
+    template <class Number>
+    constexpr FixedPoint(
+        Number n, typename std::enable_if<std::is_arithmetic<Number>::value>::type* = nullptr)
+        : data_(static_cast<base_type>(n * one)) {}
+
+public: // conversion
+    template <size_t I2, size_t F2>
+    CONSTEXPR14 explicit FixedPoint(FixedPoint<I2, F2> other) {
+        static_assert(I2 <= I && F2 <= F, "Scaling conversion can only upgrade types");
+        using T = FixedPoint<I2, F2>;
+
+        const base_type fractional = (other.data_ & T::fractional_mask);
+        const base_type integer = (other.data_ & T::integer_mask) >> T::fractional_bits;
+        data_ =
+            (integer << fractional_bits) | (fractional << (fractional_bits - T::fractional_bits));
+    }
+
+private:
+    // this makes it simpler to create a FixedPoint point object from
+    // a native type without scaling
+    // use "FixedPoint::from_base" in order to perform this.
+    struct NoScale {};
+
+    constexpr FixedPoint(base_type n, const NoScale&) : data_(n) {}
+
+public:
+    static constexpr FixedPoint from_base(base_type n) {
+        return FixedPoint(n, NoScale());
+    }
+
+public: // comparison operators
+    constexpr bool operator==(FixedPoint rhs) const {
+        return data_ == rhs.data_;
+    }
+
+    constexpr bool operator!=(FixedPoint rhs) const {
+        return data_ != rhs.data_;
+    }
+
+    constexpr bool operator<(FixedPoint rhs) const {
+        return data_ < rhs.data_;
+    }
+
+    constexpr bool operator>(FixedPoint rhs) const {
+        return data_ > rhs.data_;
+    }
+
+    constexpr bool operator<=(FixedPoint rhs) const {
+        return data_ <= rhs.data_;
+    }
+
+    constexpr bool operator>=(FixedPoint rhs) const {
+        return data_ >= rhs.data_;
+    }
+
+public: // unary operators
+    constexpr bool operator!() const {
+        return !data_;
+    }
+
+    constexpr FixedPoint operator~() const {
+        // NOTE(eteran): this will often appear to "just negate" the value
+        // that is not an error, it is because -x == (~x+1)
+        // and that "+1" is adding an infinitesimally small fraction to the
+        // complimented value
+        return FixedPoint::from_base(~data_);
+    }
+
+    constexpr FixedPoint operator-() const {
+        return FixedPoint::from_base(-data_);
+    }
+
+    constexpr FixedPoint operator+() const {
+        return FixedPoint::from_base(+data_);
+    }
+
+    CONSTEXPR14 FixedPoint& operator++() {
+        data_ += one;
+        return *this;
+    }
+
+    CONSTEXPR14 FixedPoint& operator--() {
+        data_ -= one;
+        return *this;
+    }
+
+    CONSTEXPR14 FixedPoint operator++(int) {
+        FixedPoint tmp(*this);
+        data_ += one;
+        return tmp;
+    }
+
+    CONSTEXPR14 FixedPoint operator--(int) {
+        FixedPoint tmp(*this);
+        data_ -= one;
+        return tmp;
+    }
+
+public: // basic math operators
+    CONSTEXPR14 FixedPoint& operator+=(FixedPoint n) {
+        data_ += n.data_;
+        return *this;
+    }
+
+    CONSTEXPR14 FixedPoint& operator-=(FixedPoint n) {
+        data_ -= n.data_;
+        return *this;
+    }
+
+    CONSTEXPR14 FixedPoint& operator*=(FixedPoint n) {
+        return assign(detail::multiply(*this, n));
+    }
+
+    CONSTEXPR14 FixedPoint& operator/=(FixedPoint n) {
+        FixedPoint temp;
+        return assign(detail::divide(*this, n, temp));
+    }
+
+private:
+    CONSTEXPR14 FixedPoint& assign(FixedPoint rhs) {
+        data_ = rhs.data_;
+        return *this;
+    }
+
+public: // binary math operators, effects underlying bit pattern since these
+        // don't really typically make sense for non-integer values
+    CONSTEXPR14 FixedPoint& operator&=(FixedPoint n) {
+        data_ &= n.data_;
+        return *this;
+    }
+
+    CONSTEXPR14 FixedPoint& operator|=(FixedPoint n) {
+        data_ |= n.data_;
+        return *this;
+    }
+
+    CONSTEXPR14 FixedPoint& operator^=(FixedPoint n) {
+        data_ ^= n.data_;
+        return *this;
+    }
+
+    template <class Integer,
+              class = typename std::enable_if<std::is_integral<Integer>::value>::type>
+    CONSTEXPR14 FixedPoint& operator>>=(Integer n) {
+        data_ >>= n;
+        return *this;
+    }
+
+    template <class Integer,
+              class = typename std::enable_if<std::is_integral<Integer>::value>::type>
+    CONSTEXPR14 FixedPoint& operator<<=(Integer n) {
+        data_ <<= n;
+        return *this;
+    }
+
+public: // conversion to basic types
+    constexpr void round_up() {
+        data_ += (data_ & fractional_mask) >> 1;
+    }
+
+    constexpr int to_int() {
+        round_up();
+        return static_cast<int>((data_ & integer_mask) >> fractional_bits);
+    }
+
+    constexpr unsigned int to_uint() const {
+        round_up();
+        return static_cast<unsigned int>((data_ & integer_mask) >> fractional_bits);
+    }
+
+    constexpr int64_t to_long() {
+        round_up();
+        return static_cast<int64_t>((data_ & integer_mask) >> fractional_bits);
+    }
+
+    constexpr int to_int_floor() const {
+        return static_cast<int>((data_ & integer_mask) >> fractional_bits);
+    }
+
+    constexpr int64_t to_long_floor() {
+        return static_cast<int64_t>((data_ & integer_mask) >> fractional_bits);
+    }
+
+    constexpr unsigned int to_uint_floor() const {
+        return static_cast<unsigned int>((data_ & integer_mask) >> fractional_bits);
+    }
+
+    constexpr float to_float() const {
+        return static_cast<float>(data_) / FixedPoint::one;
+    }
+
+    constexpr double to_double() const {
+        return static_cast<double>(data_) / FixedPoint::one;
+    }
+
+    constexpr base_type to_raw() const {
+        return data_;
+    }
+
+    constexpr void clear_int() {
+        data_ &= fractional_mask;
+    }
+
+    constexpr base_type get_frac() const {
+        return data_ & fractional_mask;
+    }
+
+public:
+    CONSTEXPR14 void swap(FixedPoint& rhs) {
+        using std::swap;
+        swap(data_, rhs.data_);
+    }
+
+public:
+    base_type data_;
+};
+
+// if we have the same fractional portion, but differing integer portions, we trivially upgrade the
+// smaller type
+template <size_t I1, size_t I2, size_t F>
+CONSTEXPR14 typename std::conditional<I1 >= I2, FixedPoint<I1, F>, FixedPoint<I2, F>>::type
+operator+(FixedPoint<I1, F> lhs, FixedPoint<I2, F> rhs) {
+
+    using T = typename std::conditional<I1 >= I2, FixedPoint<I1, F>, FixedPoint<I2, F>>::type;
+
+    const T l = T::from_base(lhs.to_raw());
+    const T r = T::from_base(rhs.to_raw());
+    return l + r;
+}
+
+template <size_t I1, size_t I2, size_t F>
+CONSTEXPR14 typename std::conditional<I1 >= I2, FixedPoint<I1, F>, FixedPoint<I2, F>>::type
+operator-(FixedPoint<I1, F> lhs, FixedPoint<I2, F> rhs) {
+
+    using T = typename std::conditional<I1 >= I2, FixedPoint<I1, F>, FixedPoint<I2, F>>::type;
+
+    const T l = T::from_base(lhs.to_raw());
+    const T r = T::from_base(rhs.to_raw());
+    return l - r;
+}
+
+template <size_t I1, size_t I2, size_t F>
+CONSTEXPR14 typename std::conditional<I1 >= I2, FixedPoint<I1, F>, FixedPoint<I2, F>>::type
+operator*(FixedPoint<I1, F> lhs, FixedPoint<I2, F> rhs) {
+
+    using T = typename std::conditional<I1 >= I2, FixedPoint<I1, F>, FixedPoint<I2, F>>::type;
+
+    const T l = T::from_base(lhs.to_raw());
+    const T r = T::from_base(rhs.to_raw());
+    return l * r;
+}
+
+template <size_t I1, size_t I2, size_t F>
+CONSTEXPR14 typename std::conditional<I1 >= I2, FixedPoint<I1, F>, FixedPoint<I2, F>>::type
+operator/(FixedPoint<I1, F> lhs, FixedPoint<I2, F> rhs) {
+
+    using T = typename std::conditional<I1 >= I2, FixedPoint<I1, F>, FixedPoint<I2, F>>::type;
+
+    const T l = T::from_base(lhs.to_raw());
+    const T r = T::from_base(rhs.to_raw());
+    return l / r;
+}
+
+template <size_t I, size_t F>
+std::ostream& operator<<(std::ostream& os, FixedPoint<I, F> f) {
+    os << f.to_double();
+    return os;
+}
+
+// basic math operators
+template <size_t I, size_t F>
+CONSTEXPR14 FixedPoint<I, F> operator+(FixedPoint<I, F> lhs, FixedPoint<I, F> rhs) {
+    lhs += rhs;
+    return lhs;
+}
+template <size_t I, size_t F>
+CONSTEXPR14 FixedPoint<I, F> operator-(FixedPoint<I, F> lhs, FixedPoint<I, F> rhs) {
+    lhs -= rhs;
+    return lhs;
+}
+template <size_t I, size_t F>
+CONSTEXPR14 FixedPoint<I, F> operator*(FixedPoint<I, F> lhs, FixedPoint<I, F> rhs) {
+    lhs *= rhs;
+    return lhs;
+}
+template <size_t I, size_t F>
+CONSTEXPR14 FixedPoint<I, F> operator/(FixedPoint<I, F> lhs, FixedPoint<I, F> rhs) {
+    lhs /= rhs;
+    return lhs;
+}
+
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+CONSTEXPR14 FixedPoint<I, F> operator+(FixedPoint<I, F> lhs, Number rhs) {
+    lhs += FixedPoint<I, F>(rhs);
+    return lhs;
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+CONSTEXPR14 FixedPoint<I, F> operator-(FixedPoint<I, F> lhs, Number rhs) {
+    lhs -= FixedPoint<I, F>(rhs);
+    return lhs;
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+CONSTEXPR14 FixedPoint<I, F> operator*(FixedPoint<I, F> lhs, Number rhs) {
+    lhs *= FixedPoint<I, F>(rhs);
+    return lhs;
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+CONSTEXPR14 FixedPoint<I, F> operator/(FixedPoint<I, F> lhs, Number rhs) {
+    lhs /= FixedPoint<I, F>(rhs);
+    return lhs;
+}
+
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+CONSTEXPR14 FixedPoint<I, F> operator+(Number lhs, FixedPoint<I, F> rhs) {
+    FixedPoint<I, F> tmp(lhs);
+    tmp += rhs;
+    return tmp;
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+CONSTEXPR14 FixedPoint<I, F> operator-(Number lhs, FixedPoint<I, F> rhs) {
+    FixedPoint<I, F> tmp(lhs);
+    tmp -= rhs;
+    return tmp;
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+CONSTEXPR14 FixedPoint<I, F> operator*(Number lhs, FixedPoint<I, F> rhs) {
+    FixedPoint<I, F> tmp(lhs);
+    tmp *= rhs;
+    return tmp;
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+CONSTEXPR14 FixedPoint<I, F> operator/(Number lhs, FixedPoint<I, F> rhs) {
+    FixedPoint<I, F> tmp(lhs);
+    tmp /= rhs;
+    return tmp;
+}
+
+// shift operators
+template <size_t I, size_t F, class Integer,
+          class = typename std::enable_if<std::is_integral<Integer>::value>::type>
+CONSTEXPR14 FixedPoint<I, F> operator<<(FixedPoint<I, F> lhs, Integer rhs) {
+    lhs <<= rhs;
+    return lhs;
+}
+template <size_t I, size_t F, class Integer,
+          class = typename std::enable_if<std::is_integral<Integer>::value>::type>
+CONSTEXPR14 FixedPoint<I, F> operator>>(FixedPoint<I, F> lhs, Integer rhs) {
+    lhs >>= rhs;
+    return lhs;
+}
+
+// comparison operators
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+constexpr bool operator>(FixedPoint<I, F> lhs, Number rhs) {
+    return lhs > FixedPoint<I, F>(rhs);
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+constexpr bool operator<(FixedPoint<I, F> lhs, Number rhs) {
+    return lhs < FixedPoint<I, F>(rhs);
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+constexpr bool operator>=(FixedPoint<I, F> lhs, Number rhs) {
+    return lhs >= FixedPoint<I, F>(rhs);
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+constexpr bool operator<=(FixedPoint<I, F> lhs, Number rhs) {
+    return lhs <= FixedPoint<I, F>(rhs);
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+constexpr bool operator==(FixedPoint<I, F> lhs, Number rhs) {
+    return lhs == FixedPoint<I, F>(rhs);
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+constexpr bool operator!=(FixedPoint<I, F> lhs, Number rhs) {
+    return lhs != FixedPoint<I, F>(rhs);
+}
+
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+constexpr bool operator>(Number lhs, FixedPoint<I, F> rhs) {
+    return FixedPoint<I, F>(lhs) > rhs;
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+constexpr bool operator<(Number lhs, FixedPoint<I, F> rhs) {
+    return FixedPoint<I, F>(lhs) < rhs;
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+constexpr bool operator>=(Number lhs, FixedPoint<I, F> rhs) {
+    return FixedPoint<I, F>(lhs) >= rhs;
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+constexpr bool operator<=(Number lhs, FixedPoint<I, F> rhs) {
+    return FixedPoint<I, F>(lhs) <= rhs;
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+constexpr bool operator==(Number lhs, FixedPoint<I, F> rhs) {
+    return FixedPoint<I, F>(lhs) == rhs;
+}
+template <size_t I, size_t F, class Number,
+          class = typename std::enable_if<std::is_arithmetic<Number>::value>::type>
+constexpr bool operator!=(Number lhs, FixedPoint<I, F> rhs) {
+    return FixedPoint<I, F>(lhs) != rhs;
+}
+
+} // namespace Common
+
+#undef CONSTEXPR14
+
+#endif
diff --git a/src/common/hash.h b/src/common/hash.h
index 298930702..b6f3e6d6f 100644
--- a/src/common/hash.h
+++ b/src/common/hash.h
@@ -1,6 +1,5 @@
-// Copyright 2015 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2015 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
diff --git a/src/common/input.h b/src/common/input.h
index bb42aaacc..213aa2384 100644
--- a/src/common/input.h
+++ b/src/common/input.h
@@ -1,6 +1,5 @@
-// Copyright 2017 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2017 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
@@ -28,7 +27,7 @@ enum class InputType {
     Color,
     Vibration,
     Nfc,
-    Ir,
+    IrSensor,
 };
 
 // Internal battery charge level
@@ -53,6 +52,15 @@ enum class PollingMode {
     IR,
 };
 
+enum class CameraFormat {
+    Size320x240,
+    Size160x120,
+    Size80x60,
+    Size40x30,
+    Size20x15,
+    None,
+};
+
 // Vibration reply from the controller
 enum class VibrationError {
     None,
@@ -68,6 +76,13 @@ enum class PollingError {
     Unknown,
 };
 
+// Ir camera reply from the controller
+enum class CameraError {
+    None,
+    NotSupported,
+    Unknown,
+};
+
 // Hint for amplification curve to be used
 enum class VibrationAmplificationType {
     Linear,
@@ -176,6 +191,12 @@ struct LedStatus {
     bool led_4{};
 };
 
+// Raw data fom camera
+struct CameraStatus {
+    CameraFormat format{CameraFormat::None};
+    std::vector<u8> data{};
+};
+
 // List of buttons to be passed to Qt that can be translated
 enum class ButtonNames {
     Undefined,
@@ -233,6 +254,7 @@ struct CallbackStatus {
     BodyColorStatus color_status{};
     BatteryStatus battery_status{};
     VibrationStatus vibration_status{};
+    CameraStatus camera_status{};
 };
 
 // Triggered once every input change
@@ -281,6 +303,10 @@ public:
     virtual PollingError SetPollingMode([[maybe_unused]] PollingMode polling_mode) {
         return PollingError::NotSupported;
     }
+
+    virtual CameraError SetCameraFormat([[maybe_unused]] CameraFormat camera_format) {
+        return CameraError::NotSupported;
+    }
 };
 
 /// An abstract class template for a factory that can create input devices.
diff --git a/src/common/logging/backend.cpp b/src/common/logging/backend.cpp
index b3793106d..8ce1c2fd1 100644
--- a/src/common/logging/backend.cpp
+++ b/src/common/logging/backend.cpp
@@ -1,6 +1,5 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2014 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #include <atomic>
 #include <chrono>
diff --git a/src/common/logging/backend.h b/src/common/logging/backend.h
index a0e80fe3c..12e5e2498 100644
--- a/src/common/logging/backend.h
+++ b/src/common/logging/backend.h
@@ -1,6 +1,5 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2014 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
diff --git a/src/common/logging/filter.cpp b/src/common/logging/filter.cpp
index 4acbff649..a959acb74 100644
--- a/src/common/logging/filter.cpp
+++ b/src/common/logging/filter.cpp
@@ -1,6 +1,5 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2014 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #include <algorithm>
 #include "common/logging/filter.h"
@@ -128,7 +127,7 @@ bool ParseFilterRule(Filter& instance, Iterator begin, Iterator end) {
     SUB(Service, PM)                                                                               \
     SUB(Service, PREPO)                                                                            \
     SUB(Service, PSC)                                                                              \
-    SUB(Service, PSM)                                                                              \
+    SUB(Service, PTM)                                                                              \
     SUB(Service, SET)                                                                              \
     SUB(Service, SM)                                                                               \
     SUB(Service, SPL)                                                                              \
diff --git a/src/common/logging/filter.h b/src/common/logging/filter.h
index 29419f051..54d172cc0 100644
--- a/src/common/logging/filter.h
+++ b/src/common/logging/filter.h
@@ -1,6 +1,5 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2014 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
diff --git a/src/common/logging/log.h b/src/common/logging/log.h
index 0c80d01ee..c00c01a9e 100644
--- a/src/common/logging/log.h
+++ b/src/common/logging/log.h
@@ -1,6 +1,5 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2014 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
diff --git a/src/common/logging/text_formatter.cpp b/src/common/logging/text_formatter.cpp
index b2cad58d8..09398ea64 100644
--- a/src/common/logging/text_formatter.cpp
+++ b/src/common/logging/text_formatter.cpp
@@ -1,6 +1,5 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2014 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #include <array>
 #include <cstdio>
diff --git a/src/common/logging/text_formatter.h b/src/common/logging/text_formatter.h
index 92c0bf0c5..0d0ec4370 100644
--- a/src/common/logging/text_formatter.h
+++ b/src/common/logging/text_formatter.h
@@ -1,6 +1,5 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2014 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
diff --git a/src/common/logging/types.h b/src/common/logging/types.h
index cabb4db8e..595c15ada 100644
--- a/src/common/logging/types.h
+++ b/src/common/logging/types.h
@@ -95,7 +95,7 @@ enum class Class : u8 {
     Service_PM,        ///< The PM service
     Service_PREPO,     ///< The PREPO (Play report) service
     Service_PSC,       ///< The PSC service
-    Service_PSM,       ///< The PSM service
+    Service_PTM,       ///< The PTM service
     Service_SET,       ///< The SET (Settings) service
     Service_SM,        ///< The SM (Service manager) service
     Service_SPL,       ///< The SPL service
diff --git a/src/common/microprofile.cpp b/src/common/microprofile.cpp
index ee25dd37f..e6657c82f 100644
--- a/src/common/microprofile.cpp
+++ b/src/common/microprofile.cpp
@@ -1,6 +1,5 @@
-// Copyright 2015 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2015 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 // Includes the MicroProfile implementation in this file for compilation
 #define MICROPROFILE_IMPL 1
diff --git a/src/common/microprofile.h b/src/common/microprofile.h
index 54e7f3cc4..56ef0a2dc 100644
--- a/src/common/microprofile.h
+++ b/src/common/microprofile.h
@@ -1,6 +1,5 @@
-// Copyright 2015 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2015 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
@@ -23,12 +22,3 @@ typedef void* HANDLE;
 #include <microprofile.h>
 
 #define MP_RGB(r, g, b) ((r) << 16 | (g) << 8 | (b) << 0)
-
-// On OS X, some Mach header included by MicroProfile defines these as macros, conflicting with
-// identifiers we use.
-#ifdef PAGE_SIZE
-#undef PAGE_SIZE
-#endif
-#ifdef PAGE_MASK
-#undef PAGE_MASK
-#endif
diff --git a/src/common/microprofileui.h b/src/common/microprofileui.h
index 41abe6b75..39ed18ffa 100644
--- a/src/common/microprofileui.h
+++ b/src/common/microprofileui.h
@@ -1,6 +1,5 @@
-// Copyright 2015 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2015 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
diff --git a/src/common/param_package.cpp b/src/common/param_package.cpp
index 462502e34..629babb81 100644
--- a/src/common/param_package.cpp
+++ b/src/common/param_package.cpp
@@ -1,6 +1,5 @@
-// Copyright 2017 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2017 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #include <array>
 #include <stdexcept>
diff --git a/src/common/param_package.h b/src/common/param_package.h
index c13e45479..d7c13cb1f 100644
--- a/src/common/param_package.h
+++ b/src/common/param_package.h
@@ -1,6 +1,5 @@
-// Copyright 2017 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2017 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
diff --git a/src/common/quaternion.h b/src/common/quaternion.h
index 4d0871eb4..5bb5f2af0 100644
--- a/src/common/quaternion.h
+++ b/src/common/quaternion.h
@@ -1,6 +1,5 @@
-// Copyright 2016 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2016 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
diff --git a/src/common/reader_writer_queue.h b/src/common/reader_writer_queue.h
new file mode 100644
index 000000000..60c41a8cb
--- /dev/null
+++ b/src/common/reader_writer_queue.h
@@ -0,0 +1,940 @@
+// SPDX-FileCopyrightText: 2013-2020 Cameron Desrochers
+// SPDX-License-Identifier: BSD-2-Clause
+
+#pragma once
+
+#include <cassert>
+#include <cstdint>
+#include <cstdlib> // For malloc/free/abort & size_t
+#include <memory>
+#include <new>
+#include <stdexcept>
+#include <type_traits>
+#include <utility>
+
+#include "common/atomic_helpers.h"
+
+#if __cplusplus > 199711L || _MSC_VER >= 1700 // C++11 or VS2012
+#include <chrono>
+#endif
+
+// A lock-free queue for a single-consumer, single-producer architecture.
+// The queue is also wait-free in the common path (except if more memory
+// needs to be allocated, in which case malloc is called).
+// Allocates memory sparingly, and only once if the original maximum size
+// estimate is never exceeded.
+// Tested on x86/x64 processors, but semantics should be correct for all
+// architectures (given the right implementations in atomicops.h), provided
+// that aligned integer and pointer accesses are naturally atomic.
+// Note that there should only be one consumer thread and producer thread;
+// Switching roles of the threads, or using multiple consecutive threads for
+// one role, is not safe unless properly synchronized.
+// Using the queue exclusively from one thread is fine, though a bit silly.
+
+#ifndef MOODYCAMEL_CACHE_LINE_SIZE
+#define MOODYCAMEL_CACHE_LINE_SIZE 64
+#endif
+
+#ifndef MOODYCAMEL_EXCEPTIONS_ENABLED
+#if (defined(_MSC_VER) && defined(_CPPUNWIND)) || (defined(__GNUC__) && defined(__EXCEPTIONS)) ||  \
+    (!defined(_MSC_VER) && !defined(__GNUC__))
+#define MOODYCAMEL_EXCEPTIONS_ENABLED
+#endif
+#endif
+
+#ifndef MOODYCAMEL_HAS_EMPLACE
+#if !defined(_MSC_VER) ||                                                                          \
+    _MSC_VER >= 1800 // variadic templates: either a non-MS compiler or VS >= 2013
+#define MOODYCAMEL_HAS_EMPLACE 1
+#endif
+#endif
+
+#ifndef MOODYCAMEL_MAYBE_ALIGN_TO_CACHELINE
+#if defined(__APPLE__) && defined(__MACH__) && __cplusplus >= 201703L
+// This is required to find out what deployment target we are using
+#include <CoreFoundation/CoreFoundation.h>
+#if !defined(MAC_OS_X_VERSION_MIN_REQUIRED) ||                                                     \
+    MAC_OS_X_VERSION_MIN_REQUIRED < MAC_OS_X_VERSION_10_14
+// C++17 new(size_t, align_val_t) is not backwards-compatible with older versions of macOS, so we
+// can't support over-alignment in this case
+#define MOODYCAMEL_MAYBE_ALIGN_TO_CACHELINE
+#endif
+#endif
+#endif
+
+#ifndef MOODYCAMEL_MAYBE_ALIGN_TO_CACHELINE
+#define MOODYCAMEL_MAYBE_ALIGN_TO_CACHELINE AE_ALIGN(MOODYCAMEL_CACHE_LINE_SIZE)
+#endif
+
+#ifdef AE_VCPP
+#pragma warning(push)
+#pragma warning(disable : 4324) // structure was padded due to __declspec(align())
+#pragma warning(disable : 4820) // padding was added
+#pragma warning(disable : 4127) // conditional expression is constant
+#endif
+
+namespace Common {
+
+template <typename T, size_t MAX_BLOCK_SIZE = 512>
+class MOODYCAMEL_MAYBE_ALIGN_TO_CACHELINE ReaderWriterQueue {
+    // Design: Based on a queue-of-queues. The low-level queues are just
+    // circular buffers with front and tail indices indicating where the
+    // next element to dequeue is and where the next element can be enqueued,
+    // respectively. Each low-level queue is called a "block". Each block
+    // wastes exactly one element's worth of space to keep the design simple
+    // (if front == tail then the queue is empty, and can't be full).
+    // The high-level queue is a circular linked list of blocks; again there
+    // is a front and tail, but this time they are pointers to the blocks.
+    // The front block is where the next element to be dequeued is, provided
+    // the block is not empty. The back block is where elements are to be
+    // enqueued, provided the block is not full.
+    // The producer thread owns all the tail indices/pointers. The consumer
+    // thread owns all the front indices/pointers. Both threads read each
+    // other's variables, but only the owning thread updates them. E.g. After
+    // the consumer reads the producer's tail, the tail may change before the
+    // consumer is done dequeuing an object, but the consumer knows the tail
+    // will never go backwards, only forwards.
+    // If there is no room to enqueue an object, an additional block (of
+    // equal size to the last block) is added. Blocks are never removed.
+
+public:
+    typedef T value_type;
+
+    // Constructs a queue that can hold at least `size` elements without further
+    // allocations. If more than MAX_BLOCK_SIZE elements are requested,
+    // then several blocks of MAX_BLOCK_SIZE each are reserved (including
+    // at least one extra buffer block).
+    AE_NO_TSAN explicit ReaderWriterQueue(size_t size = 15)
+#ifndef NDEBUG
+        : enqueuing(false), dequeuing(false)
+#endif
+    {
+        assert(MAX_BLOCK_SIZE == ceilToPow2(MAX_BLOCK_SIZE) &&
+               "MAX_BLOCK_SIZE must be a power of 2");
+        assert(MAX_BLOCK_SIZE >= 2 && "MAX_BLOCK_SIZE must be at least 2");
+
+        Block* firstBlock = nullptr;
+
+        largestBlockSize =
+            ceilToPow2(size + 1); // We need a spare slot to fit size elements in the block
+        if (largestBlockSize > MAX_BLOCK_SIZE * 2) {
+            // We need a spare block in case the producer is writing to a different block the
+            // consumer is reading from, and wants to enqueue the maximum number of elements. We
+            // also need a spare element in each block to avoid the ambiguity between front == tail
+            // meaning "empty" and "full". So the effective number of slots that are guaranteed to
+            // be usable at any time is the block size - 1 times the number of blocks - 1. Solving
+            // for size and applying a ceiling to the division gives us (after simplifying):
+            size_t initialBlockCount = (size + MAX_BLOCK_SIZE * 2 - 3) / (MAX_BLOCK_SIZE - 1);
+            largestBlockSize = MAX_BLOCK_SIZE;
+            Block* lastBlock = nullptr;
+            for (size_t i = 0; i != initialBlockCount; ++i) {
+                auto block = make_block(largestBlockSize);
+                if (block == nullptr) {
+#ifdef MOODYCAMEL_EXCEPTIONS_ENABLED
+                    throw std::bad_alloc();
+#else
+                    abort();
+#endif
+                }
+                if (firstBlock == nullptr) {
+                    firstBlock = block;
+                } else {
+                    lastBlock->next = block;
+                }
+                lastBlock = block;
+                block->next = firstBlock;
+            }
+        } else {
+            firstBlock = make_block(largestBlockSize);
+            if (firstBlock == nullptr) {
+#ifdef MOODYCAMEL_EXCEPTIONS_ENABLED
+                throw std::bad_alloc();
+#else
+                abort();
+#endif
+            }
+            firstBlock->next = firstBlock;
+        }
+        frontBlock = firstBlock;
+        tailBlock = firstBlock;
+
+        // Make sure the reader/writer threads will have the initialized memory setup above:
+        fence(memory_order_sync);
+    }
+
+    // Note: The queue should not be accessed concurrently while it's
+    // being moved. It's up to the user to synchronize this.
+    AE_NO_TSAN ReaderWriterQueue(ReaderWriterQueue&& other)
+        : frontBlock(other.frontBlock.load()), tailBlock(other.tailBlock.load()),
+          largestBlockSize(other.largestBlockSize)
+#ifndef NDEBUG
+          ,
+          enqueuing(false), dequeuing(false)
+#endif
+    {
+        other.largestBlockSize = 32;
+        Block* b = other.make_block(other.largestBlockSize);
+        if (b == nullptr) {
+#ifdef MOODYCAMEL_EXCEPTIONS_ENABLED
+            throw std::bad_alloc();
+#else
+            abort();
+#endif
+        }
+        b->next = b;
+        other.frontBlock = b;
+        other.tailBlock = b;
+    }
+
+    // Note: The queue should not be accessed concurrently while it's
+    // being moved. It's up to the user to synchronize this.
+    ReaderWriterQueue& operator=(ReaderWriterQueue&& other) AE_NO_TSAN {
+        Block* b = frontBlock.load();
+        frontBlock = other.frontBlock.load();
+        other.frontBlock = b;
+        b = tailBlock.load();
+        tailBlock = other.tailBlock.load();
+        other.tailBlock = b;
+        std::swap(largestBlockSize, other.largestBlockSize);
+        return *this;
+    }
+
+    // Note: The queue should not be accessed concurrently while it's
+    // being deleted. It's up to the user to synchronize this.
+    AE_NO_TSAN ~ReaderWriterQueue() {
+        // Make sure we get the latest version of all variables from other CPUs:
+        fence(memory_order_sync);
+
+        // Destroy any remaining objects in queue and free memory
+        Block* frontBlock_ = frontBlock;
+        Block* block = frontBlock_;
+        do {
+            Block* nextBlock = block->next;
+            size_t blockFront = block->front;
+            size_t blockTail = block->tail;
+
+            for (size_t i = blockFront; i != blockTail; i = (i + 1) & block->sizeMask) {
+                auto element = reinterpret_cast<T*>(block->data + i * sizeof(T));
+                element->~T();
+                (void)element;
+            }
+
+            auto rawBlock = block->rawThis;
+            block->~Block();
+            std::free(rawBlock);
+            block = nextBlock;
+        } while (block != frontBlock_);
+    }
+
+    // Enqueues a copy of element if there is room in the queue.
+    // Returns true if the element was enqueued, false otherwise.
+    // Does not allocate memory.
+    AE_FORCEINLINE bool try_enqueue(T const& element) AE_NO_TSAN {
+        return inner_enqueue<CannotAlloc>(element);
+    }
+
+    // Enqueues a moved copy of element if there is room in the queue.
+    // Returns true if the element was enqueued, false otherwise.
+    // Does not allocate memory.
+    AE_FORCEINLINE bool try_enqueue(T&& element) AE_NO_TSAN {
+        return inner_enqueue<CannotAlloc>(std::forward<T>(element));
+    }
+
+#if MOODYCAMEL_HAS_EMPLACE
+    // Like try_enqueue() but with emplace semantics (i.e. construct-in-place).
+    template <typename... Args>
+    AE_FORCEINLINE bool try_emplace(Args&&... args) AE_NO_TSAN {
+        return inner_enqueue<CannotAlloc>(std::forward<Args>(args)...);
+    }
+#endif
+
+    // Enqueues a copy of element on the queue.
+    // Allocates an additional block of memory if needed.
+    // Only fails (returns false) if memory allocation fails.
+    AE_FORCEINLINE bool enqueue(T const& element) AE_NO_TSAN {
+        return inner_enqueue<CanAlloc>(element);
+    }
+
+    // Enqueues a moved copy of element on the queue.
+    // Allocates an additional block of memory if needed.
+    // Only fails (returns false) if memory allocation fails.
+    AE_FORCEINLINE bool enqueue(T&& element) AE_NO_TSAN {
+        return inner_enqueue<CanAlloc>(std::forward<T>(element));
+    }
+
+#if MOODYCAMEL_HAS_EMPLACE
+    // Like enqueue() but with emplace semantics (i.e. construct-in-place).
+    template <typename... Args>
+    AE_FORCEINLINE bool emplace(Args&&... args) AE_NO_TSAN {
+        return inner_enqueue<CanAlloc>(std::forward<Args>(args)...);
+    }
+#endif
+
+    // Attempts to dequeue an element; if the queue is empty,
+    // returns false instead. If the queue has at least one element,
+    // moves front to result using operator=, then returns true.
+    template <typename U>
+    bool try_dequeue(U& result) AE_NO_TSAN {
+#ifndef NDEBUG
+        ReentrantGuard guard(this->dequeuing);
+#endif
+
+        // High-level pseudocode:
+        // Remember where the tail block is
+        // If the front block has an element in it, dequeue it
+        // Else
+        //     If front block was the tail block when we entered the function, return false
+        //     Else advance to next block and dequeue the item there
+
+        // Note that we have to use the value of the tail block from before we check if the front
+        // block is full or not, in case the front block is empty and then, before we check if the
+        // tail block is at the front block or not, the producer fills up the front block *and
+        // moves on*, which would make us skip a filled block. Seems unlikely, but was consistently
+        // reproducible in practice.
+        // In order to avoid overhead in the common case, though, we do a double-checked pattern
+        // where we have the fast path if the front block is not empty, then read the tail block,
+        // then re-read the front block and check if it's not empty again, then check if the tail
+        // block has advanced.
+
+        Block* frontBlock_ = frontBlock.load();
+        size_t blockTail = frontBlock_->localTail;
+        size_t blockFront = frontBlock_->front.load();
+
+        if (blockFront != blockTail ||
+            blockFront != (frontBlock_->localTail = frontBlock_->tail.load())) {
+            fence(memory_order_acquire);
+
+        non_empty_front_block:
+            // Front block not empty, dequeue from here
+            auto element = reinterpret_cast<T*>(frontBlock_->data + blockFront * sizeof(T));
+            result = std::move(*element);
+            element->~T();
+
+            blockFront = (blockFront + 1) & frontBlock_->sizeMask;
+
+            fence(memory_order_release);
+            frontBlock_->front = blockFront;
+        } else if (frontBlock_ != tailBlock.load()) {
+            fence(memory_order_acquire);
+
+            frontBlock_ = frontBlock.load();
+            blockTail = frontBlock_->localTail = frontBlock_->tail.load();
+            blockFront = frontBlock_->front.load();
+            fence(memory_order_acquire);
+
+            if (blockFront != blockTail) {
+                // Oh look, the front block isn't empty after all
+                goto non_empty_front_block;
+            }
+
+            // Front block is empty but there's another block ahead, advance to it
+            Block* nextBlock = frontBlock_->next;
+            // Don't need an acquire fence here since next can only ever be set on the tailBlock,
+            // and we're not the tailBlock, and we did an acquire earlier after reading tailBlock
+            // which ensures next is up-to-date on this CPU in case we recently were at tailBlock.
+
+            size_t nextBlockFront = nextBlock->front.load();
+            size_t nextBlockTail = nextBlock->localTail = nextBlock->tail.load();
+            fence(memory_order_acquire);
+
+            // Since the tailBlock is only ever advanced after being written to,
+            // we know there's for sure an element to dequeue on it
+            assert(nextBlockFront != nextBlockTail);
+            AE_UNUSED(nextBlockTail);
+
+            // We're done with this block, let the producer use it if it needs
+            fence(memory_order_release); // Expose possibly pending changes to frontBlock->front
+                                         // from last dequeue
+            frontBlock = frontBlock_ = nextBlock;
+
+            compiler_fence(memory_order_release); // Not strictly needed
+
+            auto element = reinterpret_cast<T*>(frontBlock_->data + nextBlockFront * sizeof(T));
+
+            result = std::move(*element);
+            element->~T();
+
+            nextBlockFront = (nextBlockFront + 1) & frontBlock_->sizeMask;
+
+            fence(memory_order_release);
+            frontBlock_->front = nextBlockFront;
+        } else {
+            // No elements in current block and no other block to advance to
+            return false;
+        }
+
+        return true;
+    }
+
+    // Returns a pointer to the front element in the queue (the one that
+    // would be removed next by a call to `try_dequeue` or `pop`). If the
+    // queue appears empty at the time the method is called, nullptr is
+    // returned instead.
+    // Must be called only from the consumer thread.
+    T* peek() const AE_NO_TSAN {
+#ifndef NDEBUG
+        ReentrantGuard guard(this->dequeuing);
+#endif
+        // See try_dequeue() for reasoning
+
+        Block* frontBlock_ = frontBlock.load();
+        size_t blockTail = frontBlock_->localTail;
+        size_t blockFront = frontBlock_->front.load();
+
+        if (blockFront != blockTail ||
+            blockFront != (frontBlock_->localTail = frontBlock_->tail.load())) {
+            fence(memory_order_acquire);
+        non_empty_front_block:
+            return reinterpret_cast<T*>(frontBlock_->data + blockFront * sizeof(T));
+        } else if (frontBlock_ != tailBlock.load()) {
+            fence(memory_order_acquire);
+            frontBlock_ = frontBlock.load();
+            blockTail = frontBlock_->localTail = frontBlock_->tail.load();
+            blockFront = frontBlock_->front.load();
+            fence(memory_order_acquire);
+
+            if (blockFront != blockTail) {
+                goto non_empty_front_block;
+            }
+
+            Block* nextBlock = frontBlock_->next;
+
+            size_t nextBlockFront = nextBlock->front.load();
+            fence(memory_order_acquire);
+
+            assert(nextBlockFront != nextBlock->tail.load());
+            return reinterpret_cast<T*>(nextBlock->data + nextBlockFront * sizeof(T));
+        }
+
+        return nullptr;
+    }
+
+    // Removes the front element from the queue, if any, without returning it.
+    // Returns true on success, or false if the queue appeared empty at the time
+    // `pop` was called.
+    bool pop() AE_NO_TSAN {
+#ifndef NDEBUG
+        ReentrantGuard guard(this->dequeuing);
+#endif
+        // See try_dequeue() for reasoning
+
+        Block* frontBlock_ = frontBlock.load();
+        size_t blockTail = frontBlock_->localTail;
+        size_t blockFront = frontBlock_->front.load();
+
+        if (blockFront != blockTail ||
+            blockFront != (frontBlock_->localTail = frontBlock_->tail.load())) {
+            fence(memory_order_acquire);
+
+        non_empty_front_block:
+            auto element = reinterpret_cast<T*>(frontBlock_->data + blockFront * sizeof(T));
+            element->~T();
+
+            blockFront = (blockFront + 1) & frontBlock_->sizeMask;
+
+            fence(memory_order_release);
+            frontBlock_->front = blockFront;
+        } else if (frontBlock_ != tailBlock.load()) {
+            fence(memory_order_acquire);
+            frontBlock_ = frontBlock.load();
+            blockTail = frontBlock_->localTail = frontBlock_->tail.load();
+            blockFront = frontBlock_->front.load();
+            fence(memory_order_acquire);
+
+            if (blockFront != blockTail) {
+                goto non_empty_front_block;
+            }
+
+            // Front block is empty but there's another block ahead, advance to it
+            Block* nextBlock = frontBlock_->next;
+
+            size_t nextBlockFront = nextBlock->front.load();
+            size_t nextBlockTail = nextBlock->localTail = nextBlock->tail.load();
+            fence(memory_order_acquire);
+
+            assert(nextBlockFront != nextBlockTail);
+            AE_UNUSED(nextBlockTail);
+
+            fence(memory_order_release);
+            frontBlock = frontBlock_ = nextBlock;
+
+            compiler_fence(memory_order_release);
+
+            auto element = reinterpret_cast<T*>(frontBlock_->data + nextBlockFront * sizeof(T));
+            element->~T();
+
+            nextBlockFront = (nextBlockFront + 1) & frontBlock_->sizeMask;
+
+            fence(memory_order_release);
+            frontBlock_->front = nextBlockFront;
+        } else {
+            // No elements in current block and no other block to advance to
+            return false;
+        }
+
+        return true;
+    }
+
+    // Returns the approximate number of items currently in the queue.
+    // Safe to call from both the producer and consumer threads.
+    inline size_t size_approx() const AE_NO_TSAN {
+        size_t result = 0;
+        Block* frontBlock_ = frontBlock.load();
+        Block* block = frontBlock_;
+        do {
+            fence(memory_order_acquire);
+            size_t blockFront = block->front.load();
+            size_t blockTail = block->tail.load();
+            result += (blockTail - blockFront) & block->sizeMask;
+            block = block->next.load();
+        } while (block != frontBlock_);
+        return result;
+    }
+
+    // Returns the total number of items that could be enqueued without incurring
+    // an allocation when this queue is empty.
+    // Safe to call from both the producer and consumer threads.
+    //
+    // NOTE: The actual capacity during usage may be different depending on the consumer.
+    //       If the consumer is removing elements concurrently, the producer cannot add to
+    //       the block the consumer is removing from until it's completely empty, except in
+    //       the case where the producer was writing to the same block the consumer was
+    //       reading from the whole time.
+    inline size_t max_capacity() const {
+        size_t result = 0;
+        Block* frontBlock_ = frontBlock.load();
+        Block* block = frontBlock_;
+        do {
+            fence(memory_order_acquire);
+            result += block->sizeMask;
+            block = block->next.load();
+        } while (block != frontBlock_);
+        return result;
+    }
+
+private:
+    enum AllocationMode { CanAlloc, CannotAlloc };
+
+#if MOODYCAMEL_HAS_EMPLACE
+    template <AllocationMode canAlloc, typename... Args>
+    bool inner_enqueue(Args&&... args) AE_NO_TSAN
+#else
+    template <AllocationMode canAlloc, typename U>
+    bool inner_enqueue(U&& element) AE_NO_TSAN
+#endif
+    {
+#ifndef NDEBUG
+        ReentrantGuard guard(this->enqueuing);
+#endif
+
+        // High-level pseudocode (assuming we're allowed to alloc a new block):
+        // If room in tail block, add to tail
+        // Else check next block
+        //     If next block is not the head block, enqueue on next block
+        //     Else create a new block and enqueue there
+        //     Advance tail to the block we just enqueued to
+
+        Block* tailBlock_ = tailBlock.load();
+        size_t blockFront = tailBlock_->localFront;
+        size_t blockTail = tailBlock_->tail.load();
+
+        size_t nextBlockTail = (blockTail + 1) & tailBlock_->sizeMask;
+        if (nextBlockTail != blockFront ||
+            nextBlockTail != (tailBlock_->localFront = tailBlock_->front.load())) {
+            fence(memory_order_acquire);
+            // This block has room for at least one more element
+            char* location = tailBlock_->data + blockTail * sizeof(T);
+#if MOODYCAMEL_HAS_EMPLACE
+            new (location) T(std::forward<Args>(args)...);
+#else
+            new (location) T(std::forward<U>(element));
+#endif
+
+            fence(memory_order_release);
+            tailBlock_->tail = nextBlockTail;
+        } else {
+            fence(memory_order_acquire);
+            if (tailBlock_->next.load() != frontBlock) {
+                // Note that the reason we can't advance to the frontBlock and start adding new
+                // entries there is because if we did, then dequeue would stay in that block,
+                // eventually reading the new values, instead of advancing to the next full block
+                // (whose values were enqueued first and so should be consumed first).
+
+                fence(memory_order_acquire); // Ensure we get latest writes if we got the latest
+                                             // frontBlock
+
+                // tailBlock is full, but there's a free block ahead, use it
+                Block* tailBlockNext = tailBlock_->next.load();
+                size_t nextBlockFront = tailBlockNext->localFront = tailBlockNext->front.load();
+                nextBlockTail = tailBlockNext->tail.load();
+                fence(memory_order_acquire);
+
+                // This block must be empty since it's not the head block and we
+                // go through the blocks in a circle
+                assert(nextBlockFront == nextBlockTail);
+                tailBlockNext->localFront = nextBlockFront;
+
+                char* location = tailBlockNext->data + nextBlockTail * sizeof(T);
+#if MOODYCAMEL_HAS_EMPLACE
+                new (location) T(std::forward<Args>(args)...);
+#else
+                new (location) T(std::forward<U>(element));
+#endif
+
+                tailBlockNext->tail = (nextBlockTail + 1) & tailBlockNext->sizeMask;
+
+                fence(memory_order_release);
+                tailBlock = tailBlockNext;
+            } else if (canAlloc == CanAlloc) {
+                // tailBlock is full and there's no free block ahead; create a new block
+                auto newBlockSize =
+                    largestBlockSize >= MAX_BLOCK_SIZE ? largestBlockSize : largestBlockSize * 2;
+                auto newBlock = make_block(newBlockSize);
+                if (newBlock == nullptr) {
+                    // Could not allocate a block!
+                    return false;
+                }
+                largestBlockSize = newBlockSize;
+
+#if MOODYCAMEL_HAS_EMPLACE
+                new (newBlock->data) T(std::forward<Args>(args)...);
+#else
+                new (newBlock->data) T(std::forward<U>(element));
+#endif
+                assert(newBlock->front == 0);
+                newBlock->tail = newBlock->localTail = 1;
+
+                newBlock->next = tailBlock_->next.load();
+                tailBlock_->next = newBlock;
+
+                // Might be possible for the dequeue thread to see the new tailBlock->next
+                // *without* seeing the new tailBlock value, but this is OK since it can't
+                // advance to the next block until tailBlock is set anyway (because the only
+                // case where it could try to read the next is if it's already at the tailBlock,
+                // and it won't advance past tailBlock in any circumstance).
+
+                fence(memory_order_release);
+                tailBlock = newBlock;
+            } else if (canAlloc == CannotAlloc) {
+                // Would have had to allocate a new block to enqueue, but not allowed
+                return false;
+            } else {
+                assert(false && "Should be unreachable code");
+                return false;
+            }
+        }
+
+        return true;
+    }
+
+    // Disable copying
+    ReaderWriterQueue(ReaderWriterQueue const&) {}
+
+    // Disable assignment
+    ReaderWriterQueue& operator=(ReaderWriterQueue const&) {}
+
+    AE_FORCEINLINE static size_t ceilToPow2(size_t x) {
+        // From http://graphics.stanford.edu/~seander/bithacks.html#RoundUpPowerOf2
+        --x;
+        x |= x >> 1;
+        x |= x >> 2;
+        x |= x >> 4;
+        for (size_t i = 1; i < sizeof(size_t); i <<= 1) {
+            x |= x >> (i << 3);
+        }
+        ++x;
+        return x;
+    }
+
+    template <typename U>
+    static AE_FORCEINLINE char* align_for(char* ptr) AE_NO_TSAN {
+        const std::size_t alignment = std::alignment_of<U>::value;
+        return ptr + (alignment - (reinterpret_cast<std::uintptr_t>(ptr) % alignment)) % alignment;
+    }
+
+private:
+#ifndef NDEBUG
+    struct ReentrantGuard {
+        AE_NO_TSAN ReentrantGuard(weak_atomic<bool>& _inSection) : inSection(_inSection) {
+            assert(!inSection &&
+                   "Concurrent (or re-entrant) enqueue or dequeue operation detected (only one "
+                   "thread at a time may hold the producer or consumer role)");
+            inSection = true;
+        }
+
+        AE_NO_TSAN ~ReentrantGuard() {
+            inSection = false;
+        }
+
+    private:
+        ReentrantGuard& operator=(ReentrantGuard const&);
+
+    private:
+        weak_atomic<bool>& inSection;
+    };
+#endif
+
+    struct Block {
+        // Avoid false-sharing by putting highly contended variables on their own cache lines
+        weak_atomic<size_t> front; // (Atomic) Elements are read from here
+        size_t localTail;          // An uncontended shadow copy of tail, owned by the consumer
+
+        char cachelineFiller0[MOODYCAMEL_CACHE_LINE_SIZE - sizeof(weak_atomic<size_t>) -
+                              sizeof(size_t)];
+        weak_atomic<size_t> tail; // (Atomic) Elements are enqueued here
+        size_t localFront;
+
+        char cachelineFiller1[MOODYCAMEL_CACHE_LINE_SIZE - sizeof(weak_atomic<size_t>) -
+                              sizeof(size_t)]; // next isn't very contended, but we don't want it on
+                                               // the same cache line as tail (which is)
+        weak_atomic<Block*> next;              // (Atomic)
+
+        char* data; // Contents (on heap) are aligned to T's alignment
+
+        const size_t sizeMask;
+
+        // size must be a power of two (and greater than 0)
+        AE_NO_TSAN Block(size_t const& _size, char* _rawThis, char* _data)
+            : front(0UL), localTail(0), tail(0UL), localFront(0), next(nullptr), data(_data),
+              sizeMask(_size - 1), rawThis(_rawThis) {}
+
+    private:
+        // C4512 - Assignment operator could not be generated
+        Block& operator=(Block const&);
+
+    public:
+        char* rawThis;
+    };
+
+    static Block* make_block(size_t capacity) AE_NO_TSAN {
+        // Allocate enough memory for the block itself, as well as all the elements it will contain
+        auto size = sizeof(Block) + std::alignment_of<Block>::value - 1;
+        size += sizeof(T) * capacity + std::alignment_of<T>::value - 1;
+        auto newBlockRaw = static_cast<char*>(std::malloc(size));
+        if (newBlockRaw == nullptr) {
+            return nullptr;
+        }
+
+        auto newBlockAligned = align_for<Block>(newBlockRaw);
+        auto newBlockData = align_for<T>(newBlockAligned + sizeof(Block));
+        return new (newBlockAligned) Block(capacity, newBlockRaw, newBlockData);
+    }
+
+private:
+    weak_atomic<Block*> frontBlock; // (Atomic) Elements are dequeued from this block
+
+    char cachelineFiller[MOODYCAMEL_CACHE_LINE_SIZE - sizeof(weak_atomic<Block*>)];
+    weak_atomic<Block*> tailBlock; // (Atomic) Elements are enqueued to this block
+
+    size_t largestBlockSize;
+
+#ifndef NDEBUG
+    weak_atomic<bool> enqueuing;
+    mutable weak_atomic<bool> dequeuing;
+#endif
+};
+
+// Like ReaderWriterQueue, but also providees blocking operations
+template <typename T, size_t MAX_BLOCK_SIZE = 512>
+class BlockingReaderWriterQueue {
+private:
+    typedef ::Common::ReaderWriterQueue<T, MAX_BLOCK_SIZE> ReaderWriterQueue;
+
+public:
+    explicit BlockingReaderWriterQueue(size_t size = 15) AE_NO_TSAN
+        : inner(size),
+          sema(new spsc_sema::LightweightSemaphore()) {}
+
+    BlockingReaderWriterQueue(BlockingReaderWriterQueue&& other) AE_NO_TSAN
+        : inner(std::move(other.inner)),
+          sema(std::move(other.sema)) {}
+
+    BlockingReaderWriterQueue& operator=(BlockingReaderWriterQueue&& other) AE_NO_TSAN {
+        std::swap(sema, other.sema);
+        std::swap(inner, other.inner);
+        return *this;
+    }
+
+    // Enqueues a copy of element if there is room in the queue.
+    // Returns true if the element was enqueued, false otherwise.
+    // Does not allocate memory.
+    AE_FORCEINLINE bool try_enqueue(T const& element) AE_NO_TSAN {
+        if (inner.try_enqueue(element)) {
+            sema->signal();
+            return true;
+        }
+        return false;
+    }
+
+    // Enqueues a moved copy of element if there is room in the queue.
+    // Returns true if the element was enqueued, false otherwise.
+    // Does not allocate memory.
+    AE_FORCEINLINE bool try_enqueue(T&& element) AE_NO_TSAN {
+        if (inner.try_enqueue(std::forward<T>(element))) {
+            sema->signal();
+            return true;
+        }
+        return false;
+    }
+
+#if MOODYCAMEL_HAS_EMPLACE
+    // Like try_enqueue() but with emplace semantics (i.e. construct-in-place).
+    template <typename... Args>
+    AE_FORCEINLINE bool try_emplace(Args&&... args) AE_NO_TSAN {
+        if (inner.try_emplace(std::forward<Args>(args)...)) {
+            sema->signal();
+            return true;
+        }
+        return false;
+    }
+#endif
+
+    // Enqueues a copy of element on the queue.
+    // Allocates an additional block of memory if needed.
+    // Only fails (returns false) if memory allocation fails.
+    AE_FORCEINLINE bool enqueue(T const& element) AE_NO_TSAN {
+        if (inner.enqueue(element)) {
+            sema->signal();
+            return true;
+        }
+        return false;
+    }
+
+    // Enqueues a moved copy of element on the queue.
+    // Allocates an additional block of memory if needed.
+    // Only fails (returns false) if memory allocation fails.
+    AE_FORCEINLINE bool enqueue(T&& element) AE_NO_TSAN {
+        if (inner.enqueue(std::forward<T>(element))) {
+            sema->signal();
+            return true;
+        }
+        return false;
+    }
+
+#if MOODYCAMEL_HAS_EMPLACE
+    // Like enqueue() but with emplace semantics (i.e. construct-in-place).
+    template <typename... Args>
+    AE_FORCEINLINE bool emplace(Args&&... args) AE_NO_TSAN {
+        if (inner.emplace(std::forward<Args>(args)...)) {
+            sema->signal();
+            return true;
+        }
+        return false;
+    }
+#endif
+
+    // Attempts to dequeue an element; if the queue is empty,
+    // returns false instead. If the queue has at least one element,
+    // moves front to result using operator=, then returns true.
+    template <typename U>
+    bool try_dequeue(U& result) AE_NO_TSAN {
+        if (sema->tryWait()) {
+            bool success = inner.try_dequeue(result);
+            assert(success);
+            AE_UNUSED(success);
+            return true;
+        }
+        return false;
+    }
+
+    // Attempts to dequeue an element; if the queue is empty,
+    // waits until an element is available, then dequeues it.
+    template <typename U>
+    void wait_dequeue(U& result) AE_NO_TSAN {
+        while (!sema->wait())
+            ;
+        bool success = inner.try_dequeue(result);
+        AE_UNUSED(result);
+        assert(success);
+        AE_UNUSED(success);
+    }
+
+    // Attempts to dequeue an element; if the queue is empty,
+    // waits until an element is available up to the specified timeout,
+    // then dequeues it and returns true, or returns false if the timeout
+    // expires before an element can be dequeued.
+    // Using a negative timeout indicates an indefinite timeout,
+    // and is thus functionally equivalent to calling wait_dequeue.
+    template <typename U>
+    bool wait_dequeue_timed(U& result, std::int64_t timeout_usecs) AE_NO_TSAN {
+        if (!sema->wait(timeout_usecs)) {
+            return false;
+        }
+        bool success = inner.try_dequeue(result);
+        AE_UNUSED(result);
+        assert(success);
+        AE_UNUSED(success);
+        return true;
+    }
+
+#if __cplusplus > 199711L || _MSC_VER >= 1700
+    // Attempts to dequeue an element; if the queue is empty,
+    // waits until an element is available up to the specified timeout,
+    // then dequeues it and returns true, or returns false if the timeout
+    // expires before an element can be dequeued.
+    // Using a negative timeout indicates an indefinite timeout,
+    // and is thus functionally equivalent to calling wait_dequeue.
+    template <typename U, typename Rep, typename Period>
+    inline bool wait_dequeue_timed(U& result,
+                                   std::chrono::duration<Rep, Period> const& timeout) AE_NO_TSAN {
+        return wait_dequeue_timed(
+            result, std::chrono::duration_cast<std::chrono::microseconds>(timeout).count());
+    }
+#endif
+
+    // Returns a pointer to the front element in the queue (the one that
+    // would be removed next by a call to `try_dequeue` or `pop`). If the
+    // queue appears empty at the time the method is called, nullptr is
+    // returned instead.
+    // Must be called only from the consumer thread.
+    AE_FORCEINLINE T* peek() const AE_NO_TSAN {
+        return inner.peek();
+    }
+
+    // Removes the front element from the queue, if any, without returning it.
+    // Returns true on success, or false if the queue appeared empty at the time
+    // `pop` was called.
+    AE_FORCEINLINE bool pop() AE_NO_TSAN {
+        if (sema->tryWait()) {
+            bool result = inner.pop();
+            assert(result);
+            AE_UNUSED(result);
+            return true;
+        }
+        return false;
+    }
+
+    // Returns the approximate number of items currently in the queue.
+    // Safe to call from both the producer and consumer threads.
+    AE_FORCEINLINE size_t size_approx() const AE_NO_TSAN {
+        return sema->availableApprox();
+    }
+
+    // Returns the total number of items that could be enqueued without incurring
+    // an allocation when this queue is empty.
+    // Safe to call from both the producer and consumer threads.
+    //
+    // NOTE: The actual capacity during usage may be different depending on the consumer.
+    //       If the consumer is removing elements concurrently, the producer cannot add to
+    //       the block the consumer is removing from until it's completely empty, except in
+    //       the case where the producer was writing to the same block the consumer was
+    //       reading from the whole time.
+    AE_FORCEINLINE size_t max_capacity() const {
+        return inner.max_capacity();
+    }
+
+private:
+    // Disable copying & assignment
+    BlockingReaderWriterQueue(BlockingReaderWriterQueue const&) {}
+    BlockingReaderWriterQueue& operator=(BlockingReaderWriterQueue const&) {}
+
+private:
+    ReaderWriterQueue inner;
+    std::unique_ptr<spsc_sema::LightweightSemaphore> sema;
+};
+
+} // namespace Common
+
+#ifdef AE_VCPP
+#pragma warning(pop)
+#endif
diff --git a/src/common/scm_rev.cpp.in b/src/common/scm_rev.cpp.in
index cc88994c6..f0c124d69 100644
--- a/src/common/scm_rev.cpp.in
+++ b/src/common/scm_rev.cpp.in
@@ -1,6 +1,5 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2014 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #include "common/scm_rev.h"
 
diff --git a/src/common/scm_rev.h b/src/common/scm_rev.h
index 563015ec9..88404316a 100644
--- a/src/common/scm_rev.h
+++ b/src/common/scm_rev.h
@@ -1,6 +1,5 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2014 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
diff --git a/src/common/scope_exit.h b/src/common/scope_exit.h
index 35dac3a8f..e9c789c88 100644
--- a/src/common/scope_exit.h
+++ b/src/common/scope_exit.h
@@ -1,6 +1,5 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2014 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
diff --git a/src/common/settings.cpp b/src/common/settings.cpp
index 751549583..7282a45d3 100644
--- a/src/common/settings.cpp
+++ b/src/common/settings.cpp
@@ -62,7 +62,8 @@ void LogSettings() {
     log_setting("Renderer_UseAsynchronousShaders", values.use_asynchronous_shaders.GetValue());
     log_setting("Renderer_AnisotropicFilteringLevel", values.max_anisotropy.GetValue());
     log_setting("Audio_OutputEngine", values.sink_id.GetValue());
-    log_setting("Audio_OutputDevice", values.audio_device_id.GetValue());
+    log_setting("Audio_OutputDevice", values.audio_output_device_id.GetValue());
+    log_setting("Audio_InputDevice", values.audio_input_device_id.GetValue());
     log_setting("DataStorage_UseVirtualSd", values.use_virtual_sd.GetValue());
     log_path("DataStorage_CacheDir", Common::FS::GetYuzuPath(Common::FS::YuzuPath::CacheDir));
     log_path("DataStorage_ConfigDir", Common::FS::GetYuzuPath(Common::FS::YuzuPath::ConfigDir));
@@ -104,7 +105,7 @@ float Volume() {
     if (values.audio_muted) {
         return 0.0f;
     }
-    return values.volume.GetValue() / 100.0f;
+    return values.volume.GetValue() / static_cast<f32>(values.volume.GetDefault());
 }
 
 void UpdateRescalingInfo() {
@@ -185,7 +186,6 @@ void RestoreGlobalState(bool is_powered_on) {
     values.max_anisotropy.SetGlobal(true);
     values.use_speed_limit.SetGlobal(true);
     values.speed_limit.SetGlobal(true);
-    values.fps_cap.SetGlobal(true);
     values.use_disk_shader_cache.SetGlobal(true);
     values.gpu_accuracy.SetGlobal(true);
     values.use_asynchronous_gpu_emulation.SetGlobal(true);
diff --git a/src/common/settings.h b/src/common/settings.h
index 3583a2e70..14ed9b237 100644
--- a/src/common/settings.h
+++ b/src/common/settings.h
@@ -106,7 +106,7 @@ struct ResolutionScalingInfo {
  * configurations. Specifying a default value and label is required. A minimum and maximum range can
  * be specified for sanitization.
  */
-template <typename Type>
+template <typename Type, bool ranged = false>
 class Setting {
 protected:
     Setting() = default;
@@ -126,8 +126,8 @@ public:
      * @param default_val Intial value of the setting, and default value of the setting
      * @param name Label for the setting
      */
-    explicit Setting(const Type& default_val, const std::string& name)
-        : value{default_val}, default_value{default_val}, ranged{false}, label{name} {}
+    explicit Setting(const Type& default_val, const std::string& name) requires(!ranged)
+        : value{default_val}, default_value{default_val}, label{name} {}
     virtual ~Setting() = default;
 
     /**
@@ -139,9 +139,9 @@ public:
      * @param name Label for the setting
      */
     explicit Setting(const Type& default_val, const Type& min_val, const Type& max_val,
-                     const std::string& name)
-        : value{default_val}, default_value{default_val}, maximum{max_val}, minimum{min_val},
-          ranged{true}, label{name} {}
+                     const std::string& name) requires(ranged)
+        : value{default_val},
+          default_value{default_val}, maximum{max_val}, minimum{min_val}, label{name} {}
 
     /**
      *  Returns a reference to the setting's value.
@@ -158,7 +158,7 @@ public:
      * @param val The desired value
      */
     virtual void SetValue(const Type& val) {
-        Type temp{(ranged) ? std::clamp(val, minimum, maximum) : val};
+        Type temp{ranged ? std::clamp(val, minimum, maximum) : val};
         std::swap(value, temp);
     }
 
@@ -188,7 +188,7 @@ public:
      * @returns A reference to the setting
      */
     virtual const Type& operator=(const Type& val) {
-        Type temp{(ranged) ? std::clamp(val, minimum, maximum) : val};
+        Type temp{ranged ? std::clamp(val, minimum, maximum) : val};
         std::swap(value, temp);
         return value;
     }
@@ -207,7 +207,6 @@ protected:
     const Type default_value{}; ///< The default value
     const Type maximum{};       ///< Maximum allowed value of the setting
     const Type minimum{};       ///< Minimum allowed value of the setting
-    const bool ranged;          ///< The setting has sanitization ranges
     const std::string label{};  ///< The setting's label
 };
 
@@ -219,8 +218,8 @@ protected:
  *
  * By default, the global setting is used.
  */
-template <typename Type>
-class SwitchableSetting : virtual public Setting<Type> {
+template <typename Type, bool ranged = false>
+class SwitchableSetting : virtual public Setting<Type, ranged> {
 public:
     /**
      * Sets a default value, label, and setting value.
@@ -228,7 +227,7 @@ public:
      * @param default_val Intial value of the setting, and default value of the setting
      * @param name Label for the setting
      */
-    explicit SwitchableSetting(const Type& default_val, const std::string& name)
+    explicit SwitchableSetting(const Type& default_val, const std::string& name) requires(!ranged)
         : Setting<Type>{default_val, name} {}
     virtual ~SwitchableSetting() = default;
 
@@ -241,8 +240,8 @@ public:
      * @param name Label for the setting
      */
     explicit SwitchableSetting(const Type& default_val, const Type& min_val, const Type& max_val,
-                               const std::string& name)
-        : Setting<Type>{default_val, min_val, max_val, name} {}
+                               const std::string& name) requires(ranged)
+        : Setting<Type, true>{default_val, min_val, max_val, name} {}
 
     /**
      * Tells this setting to represent either the global or custom setting when other member
@@ -290,7 +289,7 @@ public:
      * @param val The new value
      */
     void SetValue(const Type& val) override {
-        Type temp{(this->ranged) ? std::clamp(val, this->minimum, this->maximum) : val};
+        Type temp{ranged ? std::clamp(val, this->minimum, this->maximum) : val};
         if (use_global) {
             std::swap(this->value, temp);
         } else {
@@ -306,7 +305,7 @@ public:
      * @returns A reference to the current setting value
      */
     const Type& operator=(const Type& val) override {
-        Type temp{(this->ranged) ? std::clamp(val, this->minimum, this->maximum) : val};
+        Type temp{ranged ? std::clamp(val, this->minimum, this->maximum) : val};
         if (use_global) {
             std::swap(this->value, temp);
             return this->value;
@@ -371,18 +370,20 @@ struct TouchFromButtonMap {
 
 struct Values {
     // Audio
-    Setting<std::string> audio_device_id{"auto", "output_device"};
     Setting<std::string> sink_id{"auto", "output_engine"};
+    Setting<std::string> audio_output_device_id{"auto", "output_device"};
+    Setting<std::string> audio_input_device_id{"auto", "input_device"};
     Setting<bool> audio_muted{false, "audio_muted"};
-    SwitchableSetting<u8> volume{100, 0, 100, "volume"};
+    SwitchableSetting<u8, true> volume{100, 0, 200, "volume"};
+    Setting<bool> dump_audio_commands{false, "dump_audio_commands"};
 
     // Core
     SwitchableSetting<bool> use_multi_core{true, "use_multi_core"};
     SwitchableSetting<bool> use_extended_memory_layout{false, "use_extended_memory_layout"};
 
     // Cpu
-    SwitchableSetting<CPUAccuracy> cpu_accuracy{CPUAccuracy::Auto, CPUAccuracy::Auto,
-                                                CPUAccuracy::Paranoid, "cpu_accuracy"};
+    SwitchableSetting<CPUAccuracy, true> cpu_accuracy{CPUAccuracy::Auto, CPUAccuracy::Auto,
+                                                      CPUAccuracy::Paranoid, "cpu_accuracy"};
     // TODO: remove cpu_accuracy_first_time, migration setting added 8 July 2021
     Setting<bool> cpu_accuracy_first_time{true, "cpu_accuracy_first_time"};
     Setting<bool> cpu_debug_mode{false, "cpu_debug_mode"};
@@ -409,7 +410,7 @@ struct Values {
         true, "cpuopt_unsafe_ignore_global_monitor"};
 
     // Renderer
-    SwitchableSetting<RendererBackend> renderer_backend{
+    SwitchableSetting<RendererBackend, true> renderer_backend{
         RendererBackend::Vulkan, RendererBackend::OpenGL, RendererBackend::Vulkan, "backend"};
     Setting<bool> renderer_debug{false, "debug"};
     Setting<bool> renderer_shader_feedback{false, "shader_feedback"};
@@ -423,28 +424,26 @@ struct Values {
     SwitchableSetting<AntiAliasing> anti_aliasing{AntiAliasing::None, "anti_aliasing"};
     // *nix platforms may have issues with the borderless windowed fullscreen mode.
     // Default to exclusive fullscreen on these platforms for now.
-    SwitchableSetting<FullscreenMode> fullscreen_mode{
+    SwitchableSetting<FullscreenMode, true> fullscreen_mode{
 #ifdef _WIN32
         FullscreenMode::Borderless,
 #else
         FullscreenMode::Exclusive,
 #endif
         FullscreenMode::Borderless, FullscreenMode::Exclusive, "fullscreen_mode"};
-    SwitchableSetting<int> aspect_ratio{0, 0, 3, "aspect_ratio"};
-    SwitchableSetting<int> max_anisotropy{0, 0, 5, "max_anisotropy"};
+    SwitchableSetting<int, true> aspect_ratio{0, 0, 3, "aspect_ratio"};
+    SwitchableSetting<int, true> max_anisotropy{0, 0, 5, "max_anisotropy"};
     SwitchableSetting<bool> use_speed_limit{true, "use_speed_limit"};
-    SwitchableSetting<u16> speed_limit{100, 0, 9999, "speed_limit"};
+    SwitchableSetting<u16, true> speed_limit{100, 0, 9999, "speed_limit"};
     SwitchableSetting<bool> use_disk_shader_cache{true, "use_disk_shader_cache"};
-    SwitchableSetting<GPUAccuracy> gpu_accuracy{GPUAccuracy::High, GPUAccuracy::Normal,
-                                                GPUAccuracy::Extreme, "gpu_accuracy"};
+    SwitchableSetting<GPUAccuracy, true> gpu_accuracy{GPUAccuracy::High, GPUAccuracy::Normal,
+                                                      GPUAccuracy::Extreme, "gpu_accuracy"};
     SwitchableSetting<bool> use_asynchronous_gpu_emulation{true, "use_asynchronous_gpu_emulation"};
     SwitchableSetting<NvdecEmulation> nvdec_emulation{NvdecEmulation::GPU, "nvdec_emulation"};
     SwitchableSetting<bool> accelerate_astc{true, "accelerate_astc"};
     SwitchableSetting<bool> use_vsync{true, "use_vsync"};
-    SwitchableSetting<u16> fps_cap{1000, 1, 1000, "fps_cap"};
-    Setting<bool> disable_fps_limit{false, "disable_fps_limit"};
-    SwitchableSetting<ShaderBackend> shader_backend{ShaderBackend::GLASM, ShaderBackend::GLSL,
-                                                    ShaderBackend::SPIRV, "shader_backend"};
+    SwitchableSetting<ShaderBackend, true> shader_backend{ShaderBackend::GLASM, ShaderBackend::GLSL,
+                                                          ShaderBackend::SPIRV, "shader_backend"};
     SwitchableSetting<bool> use_asynchronous_shaders{false, "use_asynchronous_shaders"};
     SwitchableSetting<bool> use_fast_gpu_time{true, "use_fast_gpu_time"};
 
@@ -460,10 +459,10 @@ struct Values {
     s64 custom_rtc_differential;
 
     Setting<s32> current_user{0, "current_user"};
-    SwitchableSetting<s32> language_index{1, 0, 17, "language_index"};
-    SwitchableSetting<s32> region_index{1, 0, 6, "region_index"};
-    SwitchableSetting<s32> time_zone_index{0, 0, 45, "time_zone_index"};
-    SwitchableSetting<s32> sound_index{1, 0, 2, "sound_index"};
+    SwitchableSetting<s32, true> language_index{1, 0, 17, "language_index"};
+    SwitchableSetting<s32, true> region_index{1, 0, 6, "region_index"};
+    SwitchableSetting<s32, true> time_zone_index{0, 0, 45, "time_zone_index"};
+    SwitchableSetting<s32, true> sound_index{1, 0, 2, "sound_index"};
 
     // Controls
     InputSetting<std::array<PlayerInput, 10>> players;
@@ -485,7 +484,7 @@ struct Values {
     Setting<bool> tas_loop{false, "tas_loop"};
 
     Setting<bool> mouse_panning{false, "mouse_panning"};
-    Setting<u8> mouse_panning_sensitivity{10, 1, 100, "mouse_panning_sensitivity"};
+    Setting<u8, true> mouse_panning_sensitivity{10, 1, 100, "mouse_panning_sensitivity"};
     Setting<bool> mouse_enabled{false, "mouse_enabled"};
 
     Setting<bool> emulate_analog_keyboard{false, "emulate_analog_keyboard"};
@@ -504,6 +503,9 @@ struct Values {
     Setting<bool> enable_ring_controller{true, "enable_ring_controller"};
     RingconRaw ringcon_analogs;
 
+    Setting<bool> enable_ir_sensor{false, "enable_ir_sensor"};
+    Setting<std::string> ir_sensor_device{"auto", "ir_sensor_device"};
+
     // Data Storage
     Setting<bool> use_virtual_sd{true, "use_virtual_sd"};
     Setting<bool> gamecard_inserted{false, "gamecard_inserted"};
diff --git a/src/common/socket_types.h b/src/common/socket_types.h
new file mode 100644
index 000000000..0a801a443
--- /dev/null
+++ b/src/common/socket_types.h
@@ -0,0 +1,51 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace Network {
+
+/// Address families
+enum class Domain : u8 {
+    INET, ///< Address family for IPv4
+};
+
+/// Socket types
+enum class Type {
+    STREAM,
+    DGRAM,
+    RAW,
+    SEQPACKET,
+};
+
+/// Protocol values for sockets
+enum class Protocol : u8 {
+    ICMP,
+    TCP,
+    UDP,
+};
+
+/// Shutdown mode
+enum class ShutdownHow {
+    RD,
+    WR,
+    RDWR,
+};
+
+/// Array of IPv4 address
+using IPv4Address = std::array<u8, 4>;
+
+/// Cross-platform sockaddr structure
+struct SockAddrIn {
+    Domain family;
+    IPv4Address ip;
+    u16 portno;
+};
+
+constexpr u32 FLAG_MSG_PEEK = 0x2;
+constexpr u32 FLAG_MSG_DONTWAIT = 0x80;
+constexpr u32 FLAG_O_NONBLOCK = 0x800;
+
+} // namespace Network
diff --git a/src/common/telemetry.cpp b/src/common/telemetry.cpp
index 67261c55b..d26394359 100644
--- a/src/common/telemetry.cpp
+++ b/src/common/telemetry.cpp
@@ -1,6 +1,5 @@
-// Copyright 2017 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2017 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #include <algorithm>
 #include <cstring>
diff --git a/src/common/telemetry.h b/src/common/telemetry.h
index f9a824a7d..ba633d5a5 100644
--- a/src/common/telemetry.h
+++ b/src/common/telemetry.h
@@ -1,6 +1,5 @@
-// Copyright 2017 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2017 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
diff --git a/src/common/threadsafe_queue.h b/src/common/threadsafe_queue.h
index f7ae9d8c2..053798e79 100644
--- a/src/common/threadsafe_queue.h
+++ b/src/common/threadsafe_queue.h
@@ -39,7 +39,7 @@ public:
     template <typename Arg>
     void Push(Arg&& t) {
         // create the element, add it to the queue
-        write_ptr->current = std::forward<Arg>(t);
+        write_ptr->current = std::move(t);
         // set the next pointer to a new element ptr
         // then advance the write pointer
         ElementPtr* new_ptr = new ElementPtr();
diff --git a/src/common/uint128.h b/src/common/uint128.h
index 199d0f55e..f450a6db9 100644
--- a/src/common/uint128.h
+++ b/src/common/uint128.h
@@ -12,7 +12,6 @@
 #pragma intrinsic(_udiv128)
 #else
 #include <cstring>
-#include <x86intrin.h>
 #endif
 
 #include "common/common_types.h"
@@ -31,17 +30,12 @@ namespace Common {
     return _udiv128(r[1], r[0], d, &remainder);
 #endif
 #else
-#ifdef __SIZEOF_INT128__
-    const auto product = static_cast<unsigned __int128>(a) * static_cast<unsigned __int128>(b);
-    return static_cast<u64>(product / d);
-#else
     const u64 diva = a / d;
     const u64 moda = a % d;
     const u64 divb = b / d;
     const u64 modb = b % d;
     return diva * b + moda * divb + moda * modb / d;
 #endif
-#endif
 }
 
 // This function multiplies 2 u64 values and produces a u128 value;
diff --git a/src/common/wall_clock.cpp b/src/common/wall_clock.cpp
index b4fb3a59f..ae07f2811 100644
--- a/src/common/wall_clock.cpp
+++ b/src/common/wall_clock.cpp
@@ -67,7 +67,7 @@ std::unique_ptr<WallClock> CreateBestMatchingClock(u64 emulated_cpu_frequency,
     const auto& caps = GetCPUCaps();
     u64 rtsc_frequency = 0;
     if (caps.invariant_tsc) {
-        rtsc_frequency = EstimateRDTSCFrequency();
+        rtsc_frequency = caps.tsc_frequency ? caps.tsc_frequency : EstimateRDTSCFrequency();
     }
 
     // Fallback to StandardWallClock if the hardware TSC does not have the precision greater than:
diff --git a/src/common/x64/cpu_detect.cpp b/src/common/x64/cpu_detect.cpp
index 322aa1f08..1a27532d4 100644
--- a/src/common/x64/cpu_detect.cpp
+++ b/src/common/x64/cpu_detect.cpp
@@ -161,6 +161,22 @@ static CPUCaps Detect() {
         caps.invariant_tsc = Common::Bit<8>(cpu_id[3]);
     }
 
+    if (max_std_fn >= 0x15) {
+        __cpuid(cpu_id, 0x15);
+        caps.tsc_crystal_ratio_denominator = cpu_id[0];
+        caps.tsc_crystal_ratio_numerator = cpu_id[1];
+        caps.crystal_frequency = cpu_id[2];
+        // Some CPU models might not return a crystal frequency.
+        // The CPU model can be detected to use the values from turbostat
+        // https://github.com/torvalds/linux/blob/master/tools/power/x86/turbostat/turbostat.c#L5569
+        // but it's easier to just estimate the TSC tick rate for these cases.
+        if (caps.tsc_crystal_ratio_denominator) {
+            caps.tsc_frequency = static_cast<u64>(caps.crystal_frequency) *
+                                 caps.tsc_crystal_ratio_numerator /
+                                 caps.tsc_crystal_ratio_denominator;
+        }
+    }
+
     if (max_std_fn >= 0x16) {
         __cpuid(cpu_id, 0x16);
         caps.base_frequency = cpu_id[0];
diff --git a/src/common/x64/cpu_detect.h b/src/common/x64/cpu_detect.h
index 9bdc9dbfa..6830f3795 100644
--- a/src/common/x64/cpu_detect.h
+++ b/src/common/x64/cpu_detect.h
@@ -30,6 +30,11 @@ struct CPUCaps {
     u32 max_frequency;
     u32 bus_frequency;
 
+    u32 tsc_crystal_ratio_denominator;
+    u32 tsc_crystal_ratio_numerator;
+    u32 crystal_frequency;
+    u64 tsc_frequency; // Derived from the above three values
+
     bool sse : 1;
     bool sse2 : 1;
     bool sse3 : 1;
diff --git a/src/common/x64/native_clock.cpp b/src/common/x64/native_clock.cpp
index 6aaa8cdf9..8b08332ab 100644
--- a/src/common/x64/native_clock.cpp
+++ b/src/common/x64/native_clock.cpp
@@ -75,8 +75,8 @@ NativeClock::NativeClock(u64 emulated_cpu_frequency_, u64 emulated_clock_frequen
 }
 
 u64 NativeClock::GetRTSC() {
-    TimePoint current_time_point{};
     TimePoint new_time_point{};
+    TimePoint current_time_point{};
 
     current_time_point.pack = Common::AtomicLoad128(time_point.pack.data());
     do {
diff --git a/src/common/x64/xbyak_abi.h b/src/common/x64/xbyak_abi.h
index 87b3d63a4..67e6e63c8 100644
--- a/src/common/x64/xbyak_abi.h
+++ b/src/common/x64/xbyak_abi.h
@@ -1,6 +1,5 @@
-// Copyright 2016 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2016 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once
 
diff --git a/src/common/x64/xbyak_util.h b/src/common/x64/xbyak_util.h
index 44d2558f1..250e5cddb 100644
--- a/src/common/x64/xbyak_util.h
+++ b/src/common/x64/xbyak_util.h
@@ -1,6 +1,5 @@
-// Copyright 2016 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
+// SPDX-FileCopyrightText: 2016 Citra Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
 
 #pragma once