gpu_thread: Improve synchronization by using CoreTiming.

core/yuzu: Remove enable_nfc setting

.travis/macos/deps.sh

@@ -1,5 +1,6 @@
 #!/bin/sh -ex
 
 brew update
-brew install dylibbundler p7zip qt5 sdl2 ccache
+brew install p7zip qt5 sdl2 ccache
 brew outdated cmake || brew upgrade cmake
+pip3 install macpack

.travis/macos/upload.sh

@@ -11,92 +11,19 @@ mkdir "$REV_NAME"
 cp build/bin/yuzu-cmd "$REV_NAME"
 cp -r build/bin/yuzu.app "$REV_NAME"
 
-# move qt libs into app bundle for deployment
-$(brew --prefix)/opt/qt5/bin/macdeployqt "${REV_NAME}/yuzu.app"
+# move libs into folder for deployment
+macpack "${REV_NAME}/yuzu.app/Contents/MacOS/yuzu" -d "../Frameworks"
+# move qt frameworks into app bundle for deployment
+$(brew --prefix)/opt/qt5/bin/macdeployqt "${REV_NAME}/yuzu.app" -executable="${REV_NAME}/yuzu.app/Contents/MacOS/yuzu"
 
-# move SDL2 libs into folder for deployment
-dylibbundler -b -x "${REV_NAME}/yuzu-cmd" -cd -d "${REV_NAME}/libs" -p "@executable_path/libs/"
-
-# Make the changes to make the yuzu app standalone (i.e. not dependent on the current brew installation).
-# To do this, the absolute references to each and every QT framework must be re-written to point to the local frameworks
-# (in the Contents/Frameworks folder).
-# The "install_name_tool" is used to do so.
-
-# Coreutils is a hack to coerce Homebrew to point to the absolute Cellar path (symlink dereferenced). i.e:
-# ls -l /usr/local/opt/qt5:: /usr/local/opt/qt5 -> ../Cellar/qt5/5.6.1-1
-# grealpath ../Cellar/qt5/5.6.1-1:: /usr/local/Cellar/qt5/5.6.1-1
-brew install coreutils || brew upgrade coreutils || true
-
-REV_NAME_ALT=$REV_NAME/
-# grealpath is located in coreutils, there is no "realpath" for OS X :(
-QT_BREWS_PATH=$(grealpath "$(brew --prefix qt5)")
-BREW_PATH=$(brew --prefix)
-QT_VERSION_NUM=5
-
-$BREW_PATH/opt/qt5/bin/macdeployqt "${REV_NAME_ALT}yuzu.app" \
-    -executable="${REV_NAME_ALT}yuzu.app/Contents/MacOS/yuzu"
-
-# These are the files that macdeployqt packed into Contents/Frameworks/ - we don't want those, so we replace them.
-declare -a macos_libs=("QtCore" "QtWidgets" "QtGui" "QtOpenGL" "QtPrintSupport")
-
-for macos_lib in "${macos_libs[@]}"
-do
-    SC_FRAMEWORK_PART=$macos_lib.framework/Versions/$QT_VERSION_NUM/$macos_lib
-    # Replace macdeployqt versions of the Frameworks with our own (from /usr/local/opt/qt5/lib/)
-    cp "$BREW_PATH/opt/qt5/lib/$SC_FRAMEWORK_PART" "${REV_NAME_ALT}yuzu.app/Contents/Frameworks/$SC_FRAMEWORK_PART"
-
-    # Replace references within the embedded Framework files with "internal" versions.
-    for macos_lib2 in "${macos_libs[@]}"
-    do
-        # Since brew references both the non-symlinked and symlink paths of QT5, it needs to be duplicated.
-        # /usr/local/Cellar/qt5/5.6.1-1/lib and /usr/local/opt/qt5/lib both resolve to the same files.
-        # So the two lines below are effectively duplicates when resolved as a path, but as strings, they aren't.
-        RM_FRAMEWORK_PART=$macos_lib2.framework/Versions/$QT_VERSION_NUM/$macos_lib2
-        install_name_tool -change \
-            $QT_BREWS_PATH/lib/$RM_FRAMEWORK_PART \
-            @executable_path/../Frameworks/$RM_FRAMEWORK_PART \
-            "${REV_NAME_ALT}yuzu.app/Contents/Frameworks/$SC_FRAMEWORK_PART"
-        install_name_tool -change \
-            "$BREW_PATH/opt/qt5/lib/$RM_FRAMEWORK_PART" \
-            @executable_path/../Frameworks/$RM_FRAMEWORK_PART \
-            "${REV_NAME_ALT}yuzu.app/Contents/Frameworks/$SC_FRAMEWORK_PART"
-    done
-done
-
-# Handles `This application failed to start because it could not find or load the Qt platform plugin "cocoa"`
-# Which manifests itself as:
-# "Exception Type: EXC_CRASH (SIGABRT) | Exception Codes: 0x0000000000000000, 0x0000000000000000 | Exception Note: EXC_CORPSE_NOTIFY"
-# There may be more dylibs needed to be fixed...
-declare -a macos_plugins=("Plugins/platforms/libqcocoa.dylib")
-
-for macos_lib in "${macos_plugins[@]}"
-do
-    install_name_tool -id @executable_path/../$macos_lib "${REV_NAME_ALT}yuzu.app/Contents/$macos_lib"
-    for macos_lib2 in "${macos_libs[@]}"
-    do
-        RM_FRAMEWORK_PART=$macos_lib2.framework/Versions/$QT_VERSION_NUM/$macos_lib2
-        install_name_tool -change \
-            $QT_BREWS_PATH/lib/$RM_FRAMEWORK_PART \
-            @executable_path/../Frameworks/$RM_FRAMEWORK_PART \
-            "${REV_NAME_ALT}yuzu.app/Contents/$macos_lib"
-        install_name_tool -change \
-            "$BREW_PATH/opt/qt5/lib/$RM_FRAMEWORK_PART" \
-            @executable_path/../Frameworks/$RM_FRAMEWORK_PART \
-            "${REV_NAME_ALT}yuzu.app/Contents/$macos_lib"
-    done
-done
-
-for macos_lib in "${macos_libs[@]}"
-do
-    # Debugging info for Travis-CI
-    otool -L "${REV_NAME_ALT}yuzu.app/Contents/Frameworks/$macos_lib.framework/Versions/$QT_VERSION_NUM/$macos_lib"
-done
+# move libs into folder for deployment
+macpack "${REV_NAME}/yuzu-cmd" -d "libs"
 
 # Make the yuzu.app application launch a debugging terminal.
 # Store away the actual binary
-mv ${REV_NAME_ALT}yuzu.app/Contents/MacOS/yuzu ${REV_NAME_ALT}yuzu.app/Contents/MacOS/yuzu-bin
+mv ${REV_NAME}/yuzu.app/Contents/MacOS/yuzu ${REV_NAME}/yuzu.app/Contents/MacOS/yuzu-bin
 
-cat > ${REV_NAME_ALT}yuzu.app/Contents/MacOS/yuzu <<EOL
+cat > ${REV_NAME}/yuzu.app/Contents/MacOS/yuzu <<EOL
 #!/usr/bin/env bash
 cd "\`dirname "\$0"\`"
 chmod +x yuzu-bin
@@ -105,6 +32,9 @@ EOL
 # Content that will serve as the launching script for yuzu (within the .app folder)
 
 # Make the launching script executable
-chmod +x ${REV_NAME_ALT}yuzu.app/Contents/MacOS/yuzu
+chmod +x ${REV_NAME}/yuzu.app/Contents/MacOS/yuzu
+
+# Verify loader instructions
+find "$REV_NAME" -exec otool -L {} \;
 
 . .travis/common/post-upload.sh

src/audio_core/stream.cpp

@@ -38,7 +38,7 @@ Stream::Stream(Core::Timing::CoreTiming& core_timing, u32 sample_rate, Format fo
       sink_stream{sink_stream}, core_timing{core_timing}, name{std::move(name_)} {
 
     release_event = core_timing.RegisterEvent(
-        name, [this](u64 userdata, int cycles_late) { ReleaseActiveBuffer(); });
+        name, [this](u64 userdata, s64 cycles_late) { ReleaseActiveBuffer(); });
 }
 
 void Stream::Play() {

src/common/CMakeLists.txt

@@ -98,6 +98,7 @@ add_library(common STATIC
     microprofile.h
     microprofileui.h
     misc.cpp
+    multi_level_queue.h
     page_table.cpp
     page_table.h
     param_package.cpp

src/common/bit_util.h

@@ -58,4 +58,43 @@ inline u64 CountLeadingZeroes64(u64 value) {
     return __builtin_clzll(value);
 }
 #endif
+
+#ifdef _MSC_VER
+inline u32 CountTrailingZeroes32(u32 value) {
+    unsigned long trailing_zero = 0;
+
+    if (_BitScanForward(&trailing_zero, value) != 0) {
+        return trailing_zero;
+    }
+
+    return 32;
+}
+
+inline u64 CountTrailingZeroes64(u64 value) {
+    unsigned long trailing_zero = 0;
+
+    if (_BitScanForward64(&trailing_zero, value) != 0) {
+        return trailing_zero;
+    }
+
+    return 64;
+}
+#else
+inline u32 CountTrailingZeroes32(u32 value) {
+    if (value == 0) {
+        return 32;
+    }
+
+    return __builtin_ctz(value);
+}
+
+inline u64 CountTrailingZeroes64(u64 value) {
+    if (value == 0) {
+        return 64;
+    }
+
+    return __builtin_ctzll(value);
+}
+#endif
+
 } // namespace Common

src/common/detached_tasks.cpp

@@ -16,22 +16,22 @@ DetachedTasks::DetachedTasks() {
 }
 
 void DetachedTasks::WaitForAllTasks() {
-    std::unique_lock<std::mutex> lock(mutex);
+    std::unique_lock lock{mutex};
     cv.wait(lock, [this]() { return count == 0; });
 }
 
 DetachedTasks::~DetachedTasks() {
-    std::unique_lock<std::mutex> lock(mutex);
+    std::unique_lock lock{mutex};
     ASSERT(count == 0);
     instance = nullptr;
 }
 
 void DetachedTasks::AddTask(std::function<void()> task) {
-    std::unique_lock<std::mutex> lock(instance->mutex);
+    std::unique_lock lock{instance->mutex};
     ++instance->count;
     std::thread([task{std::move(task)}]() {
         task();
-        std::unique_lock<std::mutex> lock(instance->mutex);
+        std::unique_lock lock{instance->mutex};
         --instance->count;
         std::notify_all_at_thread_exit(instance->cv, std::move(lock));
     })

src/common/logging/backend.cpp

@@ -46,12 +46,12 @@ public:
     }
 
     void AddBackend(std::unique_ptr<Backend> backend) {
-        std::lock_guard<std::mutex> lock(writing_mutex);
+        std::lock_guard lock{writing_mutex};
         backends.push_back(std::move(backend));
     }
 
     void RemoveBackend(std::string_view backend_name) {
-        std::lock_guard<std::mutex> lock(writing_mutex);
+        std::lock_guard lock{writing_mutex};
         const auto it =
             std::remove_if(backends.begin(), backends.end(),
                            [&backend_name](const auto& i) { return backend_name == i->GetName(); });
@@ -80,7 +80,7 @@ private:
         backend_thread = std::thread([&] {
             Entry entry;
             auto write_logs = [&](Entry& e) {
-                std::lock_guard<std::mutex> lock(writing_mutex);
+                std::lock_guard lock{writing_mutex};
                 for (const auto& backend : backends) {
                     backend->Write(e);
                 }

src/common/multi_level_queue.h

@@ -0,0 +1,337 @@
+// Copyright 2019 TuxSH
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <iterator>
+#include <list>
+#include <utility>
+
+#include "common/bit_util.h"
+#include "common/common_types.h"
+
+namespace Common {
+
+/**
+ * A MultiLevelQueue is a type of priority queue which has the following characteristics:
+ * - iteratable through each of its elements.
+ * - back can be obtained.
+ * - O(1) add, lookup (both front and back)
+ * - discrete priorities and a max of 64 priorities (limited domain)
+ * This type of priority queue is normaly used for managing threads within an scheduler
+ */
+template <typename T, std::size_t Depth>
+class MultiLevelQueue {
+public:
+    using value_type = T;
+    using reference = value_type&;
+    using const_reference = const value_type&;
+    using pointer = value_type*;
+    using const_pointer = const value_type*;
+
+    using difference_type = typename std::pointer_traits<pointer>::difference_type;
+    using size_type = std::size_t;
+
+    template <bool is_constant>
+    class iterator_impl {
+    public:
+        using iterator_category = std::bidirectional_iterator_tag;
+        using value_type = T;
+        using pointer = std::conditional_t<is_constant, T*, const T*>;
+        using reference = std::conditional_t<is_constant, const T&, T&>;
+        using difference_type = typename std::pointer_traits<pointer>::difference_type;
+
+        friend bool operator==(const iterator_impl& lhs, const iterator_impl& rhs) {
+            if (lhs.IsEnd() && rhs.IsEnd())
+                return true;
+            return std::tie(lhs.current_priority, lhs.it) == std::tie(rhs.current_priority, rhs.it);
+        }
+
+        friend bool operator!=(const iterator_impl& lhs, const iterator_impl& rhs) {
+            return !operator==(lhs, rhs);
+        }
+
+        reference operator*() const {
+            return *it;
+        }
+
+        pointer operator->() const {
+            return it.operator->();
+        }
+
+        iterator_impl& operator++() {
+            if (IsEnd()) {
+                return *this;
+            }
+
+            ++it;
+
+            if (it == GetEndItForPrio()) {
+                u64 prios = mlq.used_priorities;
+                prios &= ~((1ULL << (current_priority + 1)) - 1);
+                if (prios == 0) {
+                    current_priority = mlq.depth();
+                } else {
+                    current_priority = CountTrailingZeroes64(prios);
+                    it = GetBeginItForPrio();
+                }
+            }
+            return *this;
+        }
+
+        iterator_impl& operator--() {
+            if (IsEnd()) {
+                if (mlq.used_priorities != 0) {
+                    current_priority = 63 - CountLeadingZeroes64(mlq.used_priorities);
+                    it = GetEndItForPrio();
+                    --it;
+                }
+            } else if (it == GetBeginItForPrio()) {
+                u64 prios = mlq.used_priorities;
+                prios &= (1ULL << current_priority) - 1;
+                if (prios != 0) {
+                    current_priority = CountTrailingZeroes64(prios);
+                    it = GetEndItForPrio();
+                    --it;
+                }
+            } else {
+                --it;
+            }
+            return *this;
+        }
+
+        iterator_impl operator++(int) {
+            const iterator_impl v{*this};
+            ++(*this);
+            return v;
+        }
+
+        iterator_impl operator--(int) {
+            const iterator_impl v{*this};
+            --(*this);
+            return v;
+        }
+
+        // allow implicit const->non-const
+        iterator_impl(const iterator_impl<false>& other)
+            : mlq(other.mlq), it(other.it), current_priority(other.current_priority) {}
+
+        iterator_impl(const iterator_impl<true>& other)
+            : mlq(other.mlq), it(other.it), current_priority(other.current_priority) {}
+
+        iterator_impl& operator=(const iterator_impl<false>& other) {
+            mlq = other.mlq;
+            it = other.it;
+            current_priority = other.current_priority;
+            return *this;
+        }
+
+        friend class iterator_impl<true>;
+        iterator_impl() = default;
+
+    private:
+        friend class MultiLevelQueue;
+        using container_ref =
+            std::conditional_t<is_constant, const MultiLevelQueue&, MultiLevelQueue&>;
+        using list_iterator = std::conditional_t<is_constant, typename std::list<T>::const_iterator,
+                                                 typename std::list<T>::iterator>;
+
+        explicit iterator_impl(container_ref mlq, list_iterator it, u32 current_priority)
+            : mlq(mlq), it(it), current_priority(current_priority) {}
+        explicit iterator_impl(container_ref mlq, u32 current_priority)
+            : mlq(mlq), it(), current_priority(current_priority) {}
+
+        bool IsEnd() const {
+            return current_priority == mlq.depth();
+        }
+
+        list_iterator GetBeginItForPrio() const {
+            return mlq.levels[current_priority].begin();
+        }
+
+        list_iterator GetEndItForPrio() const {
+            return mlq.levels[current_priority].end();
+        }
+
+        container_ref mlq;
+        list_iterator it;
+        u32 current_priority;
+    };
+
+    using iterator = iterator_impl<false>;
+    using const_iterator = iterator_impl<true>;
+
+    void add(const T& element, u32 priority, bool send_back = true) {
+        if (send_back)
+            levels[priority].push_back(element);
+        else
+            levels[priority].push_front(element);
+        used_priorities |= 1ULL << priority;
+    }
+
+    void remove(const T& element, u32 priority) {
+        auto it = ListIterateTo(levels[priority], element);
+        if (it == levels[priority].end())
+            return;
+        levels[priority].erase(it);
+        if (levels[priority].empty()) {
+            used_priorities &= ~(1ULL << priority);
+        }
+    }
+
+    void adjust(const T& element, u32 old_priority, u32 new_priority, bool adjust_front = false) {
+        remove(element, old_priority);
+        add(element, new_priority, !adjust_front);
+    }
+    void adjust(const_iterator it, u32 old_priority, u32 new_priority, bool adjust_front = false) {
+        adjust(*it, old_priority, new_priority, adjust_front);
+    }
+
+    void transfer_to_front(const T& element, u32 priority, MultiLevelQueue& other) {
+        ListSplice(other.levels[priority], other.levels[priority].begin(), levels[priority],
+                   ListIterateTo(levels[priority], element));
+
+        other.used_priorities |= 1ULL << priority;
+
+        if (levels[priority].empty()) {
+            used_priorities &= ~(1ULL << priority);
+        }
+    }
+
+    void transfer_to_front(const_iterator it, u32 priority, MultiLevelQueue& other) {
+        transfer_to_front(*it, priority, other);
+    }
+
+    void transfer_to_back(const T& element, u32 priority, MultiLevelQueue& other) {
+        ListSplice(other.levels[priority], other.levels[priority].end(), levels[priority],
+                   ListIterateTo(levels[priority], element));
+
+        other.used_priorities |= 1ULL << priority;
+
+        if (levels[priority].empty()) {
+            used_priorities &= ~(1ULL << priority);
+        }
+    }
+
+    void transfer_to_back(const_iterator it, u32 priority, MultiLevelQueue& other) {
+        transfer_to_back(*it, priority, other);
+    }
+
+    void yield(u32 priority, std::size_t n = 1) {
+        ListShiftForward(levels[priority], n);
+    }
+
+    std::size_t depth() const {
+        return Depth;
+    }
+
+    std::size_t size(u32 priority) const {
+        return levels[priority].size();
+    }
+
+    std::size_t size() const {
+        u64 priorities = used_priorities;
+        std::size_t size = 0;
+        while (priorities != 0) {
+            const u64 current_priority = CountTrailingZeroes64(priorities);
+            size += levels[current_priority].size();
+            priorities &= ~(1ULL << current_priority);
+        }
+        return size;
+    }
+
+    bool empty() const {
+        return used_priorities == 0;
+    }
+
+    bool empty(u32 priority) const {
+        return (used_priorities & (1ULL << priority)) == 0;
+    }
+
+    u32 highest_priority_set(u32 max_priority = 0) const {
+        const u64 priorities =
+            max_priority == 0 ? used_priorities : (used_priorities & ~((1ULL << max_priority) - 1));
+        return priorities == 0 ? Depth : static_cast<u32>(CountTrailingZeroes64(priorities));
+    }
+
+    u32 lowest_priority_set(u32 min_priority = Depth - 1) const {
+        const u64 priorities = min_priority >= Depth - 1
+                                   ? used_priorities
+                                   : (used_priorities & ((1ULL << (min_priority + 1)) - 1));
+        return priorities == 0 ? Depth : 63 - CountLeadingZeroes64(priorities);
+    }
+
+    const_iterator cbegin(u32 max_prio = 0) const {
+        const u32 priority = highest_priority_set(max_prio);
+        return priority == Depth ? cend()
+                                 : const_iterator{*this, levels[priority].cbegin(), priority};
+    }
+    const_iterator begin(u32 max_prio = 0) const {
+        return cbegin(max_prio);
+    }
+    iterator begin(u32 max_prio = 0) {
+        const u32 priority = highest_priority_set(max_prio);
+        return priority == Depth ? end() : iterator{*this, levels[priority].begin(), priority};
+    }
+
+    const_iterator cend(u32 min_prio = Depth - 1) const {
+        return min_prio == Depth - 1 ? const_iterator{*this, Depth} : cbegin(min_prio + 1);
+    }
+    const_iterator end(u32 min_prio = Depth - 1) const {
+        return cend(min_prio);
+    }
+    iterator end(u32 min_prio = Depth - 1) {
+        return min_prio == Depth - 1 ? iterator{*this, Depth} : begin(min_prio + 1);
+    }
+
+    T& front(u32 max_priority = 0) {
+        const u32 priority = highest_priority_set(max_priority);
+        return levels[priority == Depth ? 0 : priority].front();
+    }
+    const T& front(u32 max_priority = 0) const {
+        const u32 priority = highest_priority_set(max_priority);
+        return levels[priority == Depth ? 0 : priority].front();
+    }
+
+    T back(u32 min_priority = Depth - 1) {
+        const u32 priority = lowest_priority_set(min_priority); // intended
+        return levels[priority == Depth ? 63 : priority].back();
+    }
+    const T& back(u32 min_priority = Depth - 1) const {
+        const u32 priority = lowest_priority_set(min_priority); // intended
+        return levels[priority == Depth ? 63 : priority].back();
+    }
+
+private:
+    using const_list_iterator = typename std::list<T>::const_iterator;
+
+    static void ListShiftForward(std::list<T>& list, const std::size_t shift = 1) {
+        if (shift >= list.size()) {
+            return;
+        }
+
+        const auto begin_range = list.begin();
+        const auto end_range = std::next(begin_range, shift);
+        list.splice(list.end(), list, begin_range, end_range);
+    }
+
+    static void ListSplice(std::list<T>& in_list, const_list_iterator position,
+                           std::list<T>& out_list, const_list_iterator element) {
+        in_list.splice(position, out_list, element);
+    }
+
+    static const_list_iterator ListIterateTo(const std::list<T>& list, const T& element) {
+        auto it = list.cbegin();
+        while (it != list.cend() && *it != element) {
+            ++it;
+        }
+        return it;
+    }
+
+    std::array<std::list<T>, Depth> levels;
+    u64 used_priorities = 0;
+};
+
+} // namespace Common

src/common/thread.cpp

@@ -27,18 +27,6 @@ namespace Common {
 
 #ifdef _MSC_VER
 
-void SetThreadAffinity(std::thread::native_handle_type thread, u32 mask) {
-    SetThreadAffinityMask(thread, mask);
-}
-
-void SetCurrentThreadAffinity(u32 mask) {
-    SetThreadAffinityMask(GetCurrentThread(), mask);
-}
-
-void SwitchCurrentThread() {
-    SwitchToThread();
-}
-
 // Sets the debugger-visible name of the current thread.
 // Uses undocumented (actually, it is now documented) trick.
 // http://msdn.microsoft.com/library/default.asp?url=/library/en-us/vsdebug/html/vxtsksettingthreadname.asp
@@ -70,31 +58,6 @@ void SetCurrentThreadName(const char* name) {
 
 #else // !MSVC_VER, so must be POSIX threads
 
-void SetThreadAffinity(std::thread::native_handle_type thread, u32 mask) {
-#ifdef __APPLE__
-    thread_policy_set(pthread_mach_thread_np(thread), THREAD_AFFINITY_POLICY, (integer_t*)&mask, 1);
-#elif (defined __linux__ || defined __FreeBSD__) && !(defined ANDROID)
-    cpu_set_t cpu_set;
-    CPU_ZERO(&cpu_set);
-
-    for (int i = 0; i != sizeof(mask) * 8; ++i)
-        if ((mask >> i) & 1)
-            CPU_SET(i, &cpu_set);
-
-    pthread_setaffinity_np(thread, sizeof(cpu_set), &cpu_set);
-#endif
-}
-
-void SetCurrentThreadAffinity(u32 mask) {
-    SetThreadAffinity(pthread_self(), mask);
-}
-
-#ifndef _WIN32
-void SwitchCurrentThread() {
-    usleep(1000 * 1);
-}
-#endif
-
 // MinGW with the POSIX threading model does not support pthread_setname_np
 #if !defined(_WIN32) || defined(_MSC_VER)
 void SetCurrentThreadName(const char* name) {

src/common/thread.h

@@ -9,14 +9,13 @@
 #include <cstddef>
 #include <mutex>
 #include <thread>
-#include "common/common_types.h"
 
 namespace Common {
 
 class Event {
 public:
     void Set() {
-        std::lock_guard<std::mutex> lk(mutex);
+        std::lock_guard lk{mutex};
         if (!is_set) {
             is_set = true;
             condvar.notify_one();
@@ -24,14 +23,14 @@ public:
     }
 
     void Wait() {
-        std::unique_lock<std::mutex> lk(mutex);
+        std::unique_lock lk{mutex};
         condvar.wait(lk, [&] { return is_set; });
         is_set = false;
     }
 
     template <class Clock, class Duration>
     bool WaitUntil(const std::chrono::time_point<Clock, Duration>& time) {
-        std::unique_lock<std::mutex> lk(mutex);
+        std::unique_lock lk{mutex};
         if (!condvar.wait_until(lk, time, [this] { return is_set; }))
             return false;
         is_set = false;
@@ -39,7 +38,7 @@ public:
     }
 
     void Reset() {
-        std::unique_lock<std::mutex> lk(mutex);
+        std::unique_lock lk{mutex};
         // no other action required, since wait loops on the predicate and any lingering signal will
         // get cleared on the first iteration
         is_set = false;
@@ -57,7 +56,7 @@ public:
 
     /// Blocks until all "count" threads have called Sync()
     void Sync() {
-        std::unique_lock<std::mutex> lk(mutex);
+        std::unique_lock lk{mutex};
         const std::size_t current_generation = generation;
 
         if (++waiting == count) {
@@ -78,9 +77,6 @@ private:
     std::size_t generation = 0; // Incremented once each time the barrier is used
 };
 
-void SetThreadAffinity(std::thread::native_handle_type thread, u32 mask);
-void SetCurrentThreadAffinity(u32 mask);
-void SwitchCurrentThread(); // On Linux, this is equal to sleep 1ms
 void SetCurrentThreadName(const char* name);
 
 } // namespace Common

src/common/threadsafe_queue.h

@@ -78,7 +78,7 @@ public:
 
     T PopWait() {
         if (Empty()) {
-            std::unique_lock<std::mutex> lock(cv_mutex);
+            std::unique_lock lock{cv_mutex};
             cv.wait(lock, [this]() { return !Empty(); });
         }
         T t;
@@ -137,7 +137,7 @@ public:
 
     template <typename Arg>
     void Push(Arg&& t) {
-        std::lock_guard<std::mutex> lock(write_lock);
+        std::lock_guard lock{write_lock};
         spsc_queue.Push(t);
     }
 

src/core/CMakeLists.txt

@@ -146,6 +146,8 @@ add_library(core STATIC
     hle/kernel/svc_wrap.h
     hle/kernel/thread.cpp
     hle/kernel/thread.h
+    hle/kernel/transfer_memory.cpp
+    hle/kernel/transfer_memory.h
     hle/kernel/vm_manager.cpp
     hle/kernel/vm_manager.h
     hle/kernel/wait_object.cpp

src/core/core_cpu.cpp

@@ -22,7 +22,7 @@
 namespace Core {
 
 void CpuBarrier::NotifyEnd() {
-    std::unique_lock<std::mutex> lock(mutex);
+    std::unique_lock lock{mutex};
     end = true;
     condition.notify_all();
 }
@@ -34,7 +34,7 @@ bool CpuBarrier::Rendezvous() {
     }
 
     if (!end) {
-        std::unique_lock<std::mutex> lock(mutex);
+        std::unique_lock lock{mutex};
 
         --cores_waiting;
         if (!cores_waiting) {
@@ -131,7 +131,7 @@ void Cpu::Reschedule() {
 
     reschedule_pending = false;
     // Lock the global kernel mutex when we manipulate the HLE state
-    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
+    std::lock_guard lock{HLE::g_hle_lock};
     scheduler->Reschedule();
 }
 

src/core/core_timing.cpp

@@ -186,7 +186,7 @@ void CoreTiming::Advance() {
         Event evt = std::move(event_queue.front());
         std::pop_heap(event_queue.begin(), event_queue.end(), std::greater<>());
         event_queue.pop_back();
-        evt.type->callback(evt.userdata, static_cast<int>(global_timer - evt.time));
+        evt.type->callback(evt.userdata, global_timer - evt.time);
     }
 
     is_global_timer_sane = false;

src/core/core_timing.h

@@ -15,7 +15,7 @@
 namespace Core::Timing {
 
 /// A callback that may be scheduled for a particular core timing event.
-using TimedCallback = std::function<void(u64 userdata, int cycles_late)>;
+using TimedCallback = std::function<void(u64 userdata, s64 cycles_late)>;
 
 /// Contains the characteristics of a particular event.
 struct EventType {

src/core/file_sys/cheat_engine.cpp

@@ -423,6 +423,7 @@ std::array<u8, 16> TextCheatParser::ParseSingleLineCheat(const std::string& line
     return out;
 }
 
+namespace {
 u64 MemoryReadImpl(u32 width, VAddr addr) {
     switch (width) {
     case 1:
@@ -457,6 +458,7 @@ void MemoryWriteImpl(u32 width, VAddr addr, u64 value) {
         UNREACHABLE();
     }
 }
+} // Anonymous namespace
 
 CheatEngine::CheatEngine(Core::System& system, std::vector<CheatList> cheats_,
                          const std::string& build_id, VAddr code_region_start,

src/core/frontend/emu_window.cpp

@@ -30,7 +30,7 @@ private:
         explicit Device(std::weak_ptr<TouchState>&& touch_state) : touch_state(touch_state) {}
         std::tuple<float, float, bool> GetStatus() const override {
             if (auto state = touch_state.lock()) {
-                std::lock_guard<std::mutex> guard(state->mutex);
+                std::lock_guard guard{state->mutex};
                 return std::make_tuple(state->touch_x, state->touch_y, state->touch_pressed);
             }
             return std::make_tuple(0.0f, 0.0f, false);
@@ -81,7 +81,7 @@ void EmuWindow::TouchPressed(unsigned framebuffer_x, unsigned framebuffer_y) {
     if (!IsWithinTouchscreen(framebuffer_layout, framebuffer_x, framebuffer_y))
         return;
 
-    std::lock_guard<std::mutex> guard(touch_state->mutex);
+    std::lock_guard guard{touch_state->mutex};
     touch_state->touch_x = static_cast<float>(framebuffer_x - framebuffer_layout.screen.left) /
                            (framebuffer_layout.screen.right - framebuffer_layout.screen.left);
     touch_state->touch_y = static_cast<float>(framebuffer_y - framebuffer_layout.screen.top) /
@@ -91,7 +91,7 @@ void EmuWindow::TouchPressed(unsigned framebuffer_x, unsigned framebuffer_y) {
 }
 
 void EmuWindow::TouchReleased() {
-    std::lock_guard<std::mutex> guard(touch_state->mutex);
+    std::lock_guard guard{touch_state->mutex};
     touch_state->touch_pressed = false;
     touch_state->touch_x = 0;
     touch_state->touch_y = 0;

src/core/hle/kernel/address_arbiter.cpp

@@ -26,7 +26,7 @@ void WakeThreads(const std::vector<SharedPtr<Thread>>& waiting_threads, s32 num_
     // them all.
     std::size_t last = waiting_threads.size();
     if (num_to_wake > 0) {
-        last = num_to_wake;
+        last = std::min(last, static_cast<std::size_t>(num_to_wake));
     }
 
     // Signal the waiting threads.
@@ -90,9 +90,9 @@ ResultCode AddressArbiter::ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr a
     // Determine the modified value depending on the waiting count.
     s32 updated_value;
     if (waiting_threads.empty()) {
-        updated_value = value - 1;
-    } else if (num_to_wake <= 0 || waiting_threads.size() <= static_cast<u32>(num_to_wake)) {
         updated_value = value + 1;
+    } else if (num_to_wake <= 0 || waiting_threads.size() <= static_cast<u32>(num_to_wake)) {
+        updated_value = value - 1;
     } else {
         updated_value = value;
     }

src/core/hle/kernel/kernel.cpp

@@ -29,12 +29,12 @@ namespace Kernel {
  * @param thread_handle The handle of the thread that's been awoken
  * @param cycles_late The number of CPU cycles that have passed since the desired wakeup time
  */
-static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] int cycles_late) {
+static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_late) {
     const auto proper_handle = static_cast<Handle>(thread_handle);
     const auto& system = Core::System::GetInstance();
 
     // Lock the global kernel mutex when we enter the kernel HLE.
-    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
+    std::lock_guard lock{HLE::g_hle_lock};
 
     SharedPtr<Thread> thread =
         system.Kernel().RetrieveThreadFromWakeupCallbackHandleTable(proper_handle);
@@ -62,7 +62,8 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] int cycles_
 
     if (thread->GetMutexWaitAddress() != 0 || thread->GetCondVarWaitAddress() != 0 ||
         thread->GetWaitHandle() != 0) {
-        ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex);
+        ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex ||
+               thread->GetStatus() == ThreadStatus::WaitCondVar);
         thread->SetMutexWaitAddress(0);
         thread->SetCondVarWaitAddress(0);
         thread->SetWaitHandle(0);

src/core/hle/kernel/kernel.h

@@ -8,9 +8,6 @@
 #include <unordered_map>
 #include "core/hle/kernel/object.h"
 
-template <typename T>
-class ResultVal;
-
 namespace Core {
 class System;
 }

src/core/hle/kernel/object.cpp

@@ -23,6 +23,7 @@ bool Object::IsWaitable() const {
     case HandleType::Unknown:
     case HandleType::WritableEvent:
     case HandleType::SharedMemory:
+    case HandleType::TransferMemory:
     case HandleType::AddressArbiter:
     case HandleType::ResourceLimit:
     case HandleType::ClientPort:

src/core/hle/kernel/object.h

@@ -22,6 +22,7 @@ enum class HandleType : u32 {
     WritableEvent,
     ReadableEvent,
     SharedMemory,
+    TransferMemory,
     Thread,
     Process,
     AddressArbiter,

src/core/hle/kernel/process.cpp

@@ -5,6 +5,7 @@
 #include <algorithm>
 #include <memory>
 #include <random>
+#include "common/alignment.h"
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "core/core.h"
@@ -75,6 +76,10 @@ SharedPtr<ResourceLimit> Process::GetResourceLimit() const {
     return resource_limit;
 }
 
+u64 Process::GetTotalPhysicalMemoryUsed() const {
+    return vm_manager.GetCurrentHeapSize() + main_thread_stack_size + code_memory_size;
+}
+
 ResultCode Process::ClearSignalState() {
     if (status == ProcessStatus::Exited) {
         LOG_ERROR(Kernel, "called on a terminated process instance.");
@@ -107,14 +112,17 @@ ResultCode Process::LoadFromMetadata(const FileSys::ProgramMetadata& metadata) {
     return handle_table.SetSize(capabilities.GetHandleTableSize());
 }
 
-void Process::Run(VAddr entry_point, s32 main_thread_priority, u32 stack_size) {
+void Process::Run(VAddr entry_point, s32 main_thread_priority, u64 stack_size) {
+    // The kernel always ensures that the given stack size is page aligned.
+    main_thread_stack_size = Common::AlignUp(stack_size, Memory::PAGE_SIZE);
+
     // Allocate and map the main thread stack
     // TODO(bunnei): This is heap area that should be allocated by the kernel and not mapped as part
     // of the user address space.
+    const VAddr mapping_address = vm_manager.GetTLSIORegionEndAddress() - main_thread_stack_size;
     vm_manager
-        .MapMemoryBlock(vm_manager.GetTLSIORegionEndAddress() - stack_size,
-                        std::make_shared<std::vector<u8>>(stack_size, 0), 0, stack_size,
-                        MemoryState::Stack)
+        .MapMemoryBlock(mapping_address, std::make_shared<std::vector<u8>>(main_thread_stack_size),
+                        0, main_thread_stack_size, MemoryState::Stack)
         .Unwrap();
 
     vm_manager.LogLayout();
@@ -224,6 +232,8 @@ void Process::LoadModule(CodeSet module_, VAddr base_addr) {
     MapSegment(module_.RODataSegment(), VMAPermission::Read, MemoryState::CodeData);
     MapSegment(module_.DataSegment(), VMAPermission::ReadWrite, MemoryState::CodeData);
 
+    code_memory_size += module_.memory->size();
+
     // Clear instruction cache in CPU JIT
     system.InvalidateCpuInstructionCaches();
 }

src/core/hle/kernel/process.h

@@ -35,14 +35,6 @@ class Thread;
 
 struct CodeSet;
 
-struct AddressMapping {
-    // Address and size must be page-aligned
-    VAddr address;
-    u64 size;
-    bool read_only;
-    bool unk_flag;
-};
-
 enum class MemoryRegion : u16 {
     APPLICATION = 1,
     SYSTEM = 2,
@@ -194,6 +186,9 @@ public:
         return random_entropy.at(index);
     }
 
+    /// Retrieves the total physical memory used by this process in bytes.
+    u64 GetTotalPhysicalMemoryUsed() const;
+
     /// Clears the signaled state of the process if and only if it's signaled.
     ///
     /// @pre The process must not be already terminated. If this is called on a
@@ -218,7 +213,7 @@ public:
     /**
      * Applies address space changes and launches the process main thread.
      */
-    void Run(VAddr entry_point, s32 main_thread_priority, u32 stack_size);
+    void Run(VAddr entry_point, s32 main_thread_priority, u64 stack_size);
 
     /**
      * Prepares a process for termination by stopping all of its threads
@@ -255,6 +250,12 @@ private:
     /// Memory manager for this process.
     Kernel::VMManager vm_manager;
 
+    /// Size of the main thread's stack in bytes.
+    u64 main_thread_stack_size = 0;
+
+    /// Size of the loaded code memory in bytes.
+    u64 code_memory_size = 0;
+
     /// Current status of the process
     ProcessStatus status;
 

src/core/hle/kernel/scheduler.cpp

@@ -29,8 +29,8 @@ Scheduler::~Scheduler() {
 }
 
 bool Scheduler::HaveReadyThreads() const {
-    std::lock_guard<std::mutex> lock(scheduler_mutex);
-    return ready_queue.get_first() != nullptr;
+    std::lock_guard lock{scheduler_mutex};
+    return !ready_queue.empty();
 }
 
 Thread* Scheduler::GetCurrentThread() const {
@@ -46,22 +46,27 @@ Thread* Scheduler::PopNextReadyThread() {
     Thread* thread = GetCurrentThread();
 
     if (thread && thread->GetStatus() == ThreadStatus::Running) {
+        if (ready_queue.empty()) {
+            return thread;
+        }
         // We have to do better than the current thread.
         // This call returns null when that's not possible.
-        next = ready_queue.pop_first_better(thread->GetPriority());
-        if (!next) {
-            // Otherwise just keep going with the current thread
+        next = ready_queue.front();
+        if (next == nullptr || next->GetPriority() >= thread->GetPriority()) {
             next = thread;
         }
     } else {
-        next = ready_queue.pop_first();
+        if (ready_queue.empty()) {
+            return nullptr;
+        }
+        next = ready_queue.front();
     }
 
     return next;
 }
 
 void Scheduler::SwitchContext(Thread* new_thread) {
-    Thread* const previous_thread = GetCurrentThread();
+    Thread* previous_thread = GetCurrentThread();
     Process* const previous_process = system.Kernel().CurrentProcess();
 
     UpdateLastContextSwitchTime(previous_thread, previous_process);
@@ -75,7 +80,7 @@ void Scheduler::SwitchContext(Thread* new_thread) {
         if (previous_thread->GetStatus() == ThreadStatus::Running) {
             // This is only the case when a reschedule is triggered without the current thread
             // yielding execution (i.e. an event triggered, system core time-sliced, etc)
-            ready_queue.push_front(previous_thread->GetPriority(), previous_thread);
+            ready_queue.add(previous_thread, previous_thread->GetPriority(), false);
             previous_thread->SetStatus(ThreadStatus::Ready);
         }
     }
@@ -90,7 +95,7 @@ void Scheduler::SwitchContext(Thread* new_thread) {
 
         current_thread = new_thread;
 
-        ready_queue.remove(new_thread->GetPriority(), new_thread);
+        ready_queue.remove(new_thread, new_thread->GetPriority());
         new_thread->SetStatus(ThreadStatus::Running);
 
         auto* const thread_owner_process = current_thread->GetOwnerProcess();
@@ -127,7 +132,7 @@ void Scheduler::UpdateLastContextSwitchTime(Thread* thread, Process* process) {
 }
 
 void Scheduler::Reschedule() {
-    std::lock_guard<std::mutex> lock(scheduler_mutex);
+    std::lock_guard lock{scheduler_mutex};
 
     Thread* cur = GetCurrentThread();
     Thread* next = PopNextReadyThread();
@@ -143,51 +148,54 @@ void Scheduler::Reschedule() {
     SwitchContext(next);
 }
 
-void Scheduler::AddThread(SharedPtr<Thread> thread, u32 priority) {
-    std::lock_guard<std::mutex> lock(scheduler_mutex);
+void Scheduler::AddThread(SharedPtr<Thread> thread) {
+    std::lock_guard lock{scheduler_mutex};
 
     thread_list.push_back(std::move(thread));
-    ready_queue.prepare(priority);
 }
 
 void Scheduler::RemoveThread(Thread* thread) {
-    std::lock_guard<std::mutex> lock(scheduler_mutex);
+    std::lock_guard lock{scheduler_mutex};
 
     thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread),
                       thread_list.end());
 }
 
 void Scheduler::ScheduleThread(Thread* thread, u32 priority) {
-    std::lock_guard<std::mutex> lock(scheduler_mutex);
+    std::lock_guard lock{scheduler_mutex};
 
     ASSERT(thread->GetStatus() == ThreadStatus::Ready);
-    ready_queue.push_back(priority, thread);
+    ready_queue.add(thread, priority);
 }
 
 void Scheduler::UnscheduleThread(Thread* thread, u32 priority) {
-    std::lock_guard<std::mutex> lock(scheduler_mutex);
+    std::lock_guard lock{scheduler_mutex};
 
     ASSERT(thread->GetStatus() == ThreadStatus::Ready);
-    ready_queue.remove(priority, thread);
+    ready_queue.remove(thread, priority);
 }
 
 void Scheduler::SetThreadPriority(Thread* thread, u32 priority) {
-    std::lock_guard<std::mutex> lock(scheduler_mutex);
+    std::lock_guard lock{scheduler_mutex};
+    if (thread->GetPriority() == priority) {
+        return;
+    }
 
     // If thread was ready, adjust queues
     if (thread->GetStatus() == ThreadStatus::Ready)
-        ready_queue.move(thread, thread->GetPriority(), priority);
-    else
-        ready_queue.prepare(priority);
+        ready_queue.adjust(thread, thread->GetPriority(), priority);
 }
 
 Thread* Scheduler::GetNextSuggestedThread(u32 core, u32 maximum_priority) const {
-    std::lock_guard<std::mutex> lock(scheduler_mutex);
+    std::lock_guard lock{scheduler_mutex};
 
     const u32 mask = 1U << core;
-    return ready_queue.get_first_filter([mask, maximum_priority](Thread const* thread) {
-        return (thread->GetAffinityMask() & mask) != 0 && thread->GetPriority() < maximum_priority;
-    });
+    for (auto* thread : ready_queue) {
+        if ((thread->GetAffinityMask() & mask) != 0 && thread->GetPriority() < maximum_priority) {
+            return thread;
+        }
+    }
+    return nullptr;
 }
 
 void Scheduler::YieldWithoutLoadBalancing(Thread* thread) {

src/core/hle/kernel/scheduler.h

@@ -7,7 +7,7 @@
 #include <mutex>
 #include <vector>
 #include "common/common_types.h"
-#include "common/thread_queue_list.h"
+#include "common/multi_level_queue.h"
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/thread.h"
 
@@ -38,7 +38,7 @@ public:
     u64 GetLastContextSwitchTicks() const;
 
     /// Adds a new thread to the scheduler
-    void AddThread(SharedPtr<Thread> thread, u32 priority);
+    void AddThread(SharedPtr<Thread> thread);
 
     /// Removes a thread from the scheduler
     void RemoveThread(Thread* thread);
@@ -156,7 +156,7 @@ private:
     std::vector<SharedPtr<Thread>> thread_list;
 
     /// Lists only ready thread ids.
-    Common::ThreadQueueList<Thread*, THREADPRIO_LOWEST + 1> ready_queue;
+    Common::MultiLevelQueue<Thread*, THREADPRIO_LOWEST + 1> ready_queue;
 
     SharedPtr<Thread> current_thread = nullptr;
 

src/core/hle/kernel/svc.cpp

@@ -32,6 +32,7 @@
 #include "core/hle/kernel/svc.h"
 #include "core/hle/kernel/svc_wrap.h"
 #include "core/hle/kernel/thread.h"
+#include "core/hle/kernel/transfer_memory.h"
 #include "core/hle/kernel/writable_event.h"
 #include "core/hle/lock.h"
 #include "core/hle/result.h"
@@ -708,7 +709,7 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
         HeapRegionBaseAddr = 4,
         HeapRegionSize = 5,
         TotalMemoryUsage = 6,
-        TotalHeapUsage = 7,
+        TotalPhysicalMemoryUsed = 7,
         IsCurrentProcessBeingDebugged = 8,
         RegisterResourceLimit = 9,
         IdleTickCount = 10,
@@ -744,7 +745,7 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
     case GetInfoType::NewMapRegionBaseAddr:
     case GetInfoType::NewMapRegionSize:
     case GetInfoType::TotalMemoryUsage:
-    case GetInfoType::TotalHeapUsage:
+    case GetInfoType::TotalPhysicalMemoryUsed:
     case GetInfoType::IsVirtualAddressMemoryEnabled:
     case GetInfoType::PersonalMmHeapUsage:
     case GetInfoType::TitleId:
@@ -804,8 +805,8 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
             *result = process->VMManager().GetTotalMemoryUsage();
             return RESULT_SUCCESS;
 
-        case GetInfoType::TotalHeapUsage:
-            *result = process->VMManager().GetCurrentHeapSize();
+        case GetInfoType::TotalPhysicalMemoryUsed:
+            *result = process->GetTotalPhysicalMemoryUsed();
             return RESULT_SUCCESS;
 
         case GetInfoType::IsVirtualAddressMemoryEnabled:
@@ -1352,7 +1353,7 @@ static ResultCode WaitProcessWideKeyAtomic(VAddr mutex_addr, VAddr condition_var
     current_thread->SetCondVarWaitAddress(condition_variable_addr);
     current_thread->SetMutexWaitAddress(mutex_addr);
     current_thread->SetWaitHandle(thread_handle);
-    current_thread->SetStatus(ThreadStatus::WaitMutex);
+    current_thread->SetStatus(ThreadStatus::WaitCondVar);
     current_thread->InvalidateWakeupCallback();
 
     current_thread->WakeAfterDelay(nano_seconds);
@@ -1396,10 +1397,10 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
     // them all.
     std::size_t last = waiting_threads.size();
     if (target != -1)
-        last = target;
+        last = std::min(waiting_threads.size(), static_cast<std::size_t>(target));
 
     // If there are no threads waiting on this condition variable, just exit
-    if (last > waiting_threads.size())
+    if (last == 0)
         return RESULT_SUCCESS;
 
     for (std::size_t index = 0; index < last; ++index) {
@@ -1407,6 +1408,9 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
 
         ASSERT(thread->GetCondVarWaitAddress() == condition_variable_addr);
 
+        // liberate Cond Var Thread.
+        thread->SetCondVarWaitAddress(0);
+
         std::size_t current_core = Core::System::GetInstance().CurrentCoreIndex();
 
         auto& monitor = Core::System::GetInstance().Monitor();
@@ -1425,10 +1429,9 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
             }
         } while (!monitor.ExclusiveWrite32(current_core, thread->GetMutexWaitAddress(),
                                            thread->GetWaitHandle()));
-
         if (mutex_val == 0) {
             // We were able to acquire the mutex, resume this thread.
-            ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex);
+            ASSERT(thread->GetStatus() == ThreadStatus::WaitCondVar);
             thread->ResumeFromWait();
 
             auto* const lock_owner = thread->GetLockOwner();
@@ -1438,8 +1441,8 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
 
             thread->SetLockOwner(nullptr);
             thread->SetMutexWaitAddress(0);
-            thread->SetCondVarWaitAddress(0);
             thread->SetWaitHandle(0);
+            Core::System::GetInstance().CpuCore(thread->GetProcessorID()).PrepareReschedule();
         } else {
             // Atomically signal that the mutex now has a waiting thread.
             do {
@@ -1458,12 +1461,11 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
             const auto& handle_table = Core::CurrentProcess()->GetHandleTable();
             auto owner = handle_table.Get<Thread>(owner_handle);
             ASSERT(owner);
-            ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex);
+            ASSERT(thread->GetStatus() == ThreadStatus::WaitCondVar);
             thread->InvalidateWakeupCallback();
+            thread->SetStatus(ThreadStatus::WaitMutex);
 
             owner->AddMutexWaiter(thread);
-
-            Core::System::GetInstance().CpuCore(thread->GetProcessorID()).PrepareReschedule();
         }
     }
 
@@ -1583,14 +1585,121 @@ static ResultCode CreateTransferMemory(Handle* handle, VAddr addr, u64 size, u32
     }
 
     auto& kernel = Core::System::GetInstance().Kernel();
-    auto process = kernel.CurrentProcess();
-    auto& handle_table = process->GetHandleTable();
-    const auto shared_mem_handle = SharedMemory::Create(kernel, process, size, perms, perms, addr);
+    auto transfer_mem_handle = TransferMemory::Create(kernel, addr, size, perms);
 
-    CASCADE_RESULT(*handle, handle_table.Create(shared_mem_handle));
+    auto& handle_table = kernel.CurrentProcess()->GetHandleTable();
+    const auto result = handle_table.Create(std::move(transfer_mem_handle));
+    if (result.Failed()) {
+        return result.Code();
+    }
+
+    *handle = *result;
     return RESULT_SUCCESS;
 }
 
+static ResultCode MapTransferMemory(Handle handle, VAddr address, u64 size, u32 permission_raw) {
+    LOG_DEBUG(Kernel_SVC,
+              "called. handle=0x{:08X}, address=0x{:016X}, size=0x{:016X}, permissions=0x{:08X}",
+              handle, address, size, permission_raw);
+
+    if (!Common::Is4KBAligned(address)) {
+        LOG_ERROR(Kernel_SVC, "Transfer memory addresses must be 4KB aligned (size=0x{:016X}).",
+                  address);
+        return ERR_INVALID_ADDRESS;
+    }
+
+    if (size == 0 || !Common::Is4KBAligned(size)) {
+        LOG_ERROR(Kernel_SVC,
+                  "Transfer memory sizes must be 4KB aligned and not be zero (size=0x{:016X}).",
+                  size);
+        return ERR_INVALID_SIZE;
+    }
+
+    if (!IsValidAddressRange(address, size)) {
+        LOG_ERROR(Kernel_SVC,
+                  "Given address and size overflows the 64-bit range (address=0x{:016X}, "
+                  "size=0x{:016X}).",
+                  address, size);
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+
+    const auto permissions = static_cast<MemoryPermission>(permission_raw);
+    if (permissions != MemoryPermission::None && permissions != MemoryPermission::Read &&
+        permissions != MemoryPermission::ReadWrite) {
+        LOG_ERROR(Kernel_SVC, "Invalid transfer memory permissions given (permissions=0x{:08X}).",
+                  permission_raw);
+        return ERR_INVALID_STATE;
+    }
+
+    const auto& kernel = Core::System::GetInstance().Kernel();
+    const auto* const current_process = kernel.CurrentProcess();
+    const auto& handle_table = current_process->GetHandleTable();
+
+    auto transfer_memory = handle_table.Get<TransferMemory>(handle);
+    if (!transfer_memory) {
+        LOG_ERROR(Kernel_SVC, "Nonexistent transfer memory handle given (handle=0x{:08X}).",
+                  handle);
+        return ERR_INVALID_HANDLE;
+    }
+
+    if (!current_process->VMManager().IsWithinASLRRegion(address, size)) {
+        LOG_ERROR(Kernel_SVC,
+                  "Given address and size don't fully fit within the ASLR region "
+                  "(address=0x{:016X}, size=0x{:016X}).",
+                  address, size);
+        return ERR_INVALID_MEMORY_RANGE;
+    }
+
+    return transfer_memory->MapMemory(address, size, permissions);
+}
+
+static ResultCode UnmapTransferMemory(Handle handle, VAddr address, u64 size) {
+    LOG_DEBUG(Kernel_SVC, "called. handle=0x{:08X}, address=0x{:016X}, size=0x{:016X}", handle,
+              address, size);
+
+    if (!Common::Is4KBAligned(address)) {
+        LOG_ERROR(Kernel_SVC, "Transfer memory addresses must be 4KB aligned (size=0x{:016X}).",
+                  address);
+        return ERR_INVALID_ADDRESS;
+    }
+
+    if (size == 0 || !Common::Is4KBAligned(size)) {
+        LOG_ERROR(Kernel_SVC,
+                  "Transfer memory sizes must be 4KB aligned and not be zero (size=0x{:016X}).",
+                  size);
+        return ERR_INVALID_SIZE;
+    }
+
+    if (!IsValidAddressRange(address, size)) {
+        LOG_ERROR(Kernel_SVC,
+                  "Given address and size overflows the 64-bit range (address=0x{:016X}, "
+                  "size=0x{:016X}).",
+                  address, size);
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+
+    const auto& kernel = Core::System::GetInstance().Kernel();
+    const auto* const current_process = kernel.CurrentProcess();
+    const auto& handle_table = current_process->GetHandleTable();
+
+    auto transfer_memory = handle_table.Get<TransferMemory>(handle);
+    if (!transfer_memory) {
+        LOG_ERROR(Kernel_SVC, "Nonexistent transfer memory handle given (handle=0x{:08X}).",
+                  handle);
+        return ERR_INVALID_HANDLE;
+    }
+
+    if (!current_process->VMManager().IsWithinASLRRegion(address, size)) {
+        LOG_ERROR(Kernel_SVC,
+                  "Given address and size don't fully fit within the ASLR region "
+                  "(address=0x{:016X}, size=0x{:016X}).",
+                  address, size);
+        return ERR_INVALID_MEMORY_RANGE;
+    }
+
+    return transfer_memory->UnmapMemory(address, size);
+}
+
 static ResultCode GetThreadCoreMask(Handle thread_handle, u32* core, u64* mask) {
     LOG_TRACE(Kernel_SVC, "called, handle=0x{:08X}", thread_handle);
 
@@ -1966,8 +2075,8 @@ static const FunctionDef SVC_Table[] = {
     {0x4E, nullptr, "ReadWriteRegister"},
     {0x4F, nullptr, "SetProcessActivity"},
     {0x50, SvcWrap<CreateSharedMemory>, "CreateSharedMemory"},
-    {0x51, nullptr, "MapTransferMemory"},
-    {0x52, nullptr, "UnmapTransferMemory"},
+    {0x51, SvcWrap<MapTransferMemory>, "MapTransferMemory"},
+    {0x52, SvcWrap<UnmapTransferMemory>, "UnmapTransferMemory"},
     {0x53, nullptr, "CreateInterruptEvent"},
     {0x54, nullptr, "QueryPhysicalAddress"},
     {0x55, nullptr, "QueryIoMapping"},
@@ -2029,7 +2138,7 @@ void CallSVC(u32 immediate) {
     MICROPROFILE_SCOPE(Kernel_SVC);
 
     // Lock the global kernel mutex when we enter the kernel HLE.
-    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
+    std::lock_guard lock{HLE::g_hle_lock};
 
     const FunctionDef* info = GetSVCInfo(immediate);
     if (info) {

src/core/hle/kernel/thread.cpp

@@ -105,6 +105,7 @@ void Thread::ResumeFromWait() {
     case ThreadStatus::WaitSleep:
     case ThreadStatus::WaitIPC:
     case ThreadStatus::WaitMutex:
+    case ThreadStatus::WaitCondVar:
     case ThreadStatus::WaitArb:
         break;
 
@@ -198,7 +199,7 @@ ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name
     thread->callback_handle = kernel.ThreadWakeupCallbackHandleTable().Create(thread).Unwrap();
     thread->owner_process = &owner_process;
     thread->scheduler = &system.Scheduler(processor_id);
-    thread->scheduler->AddThread(thread, priority);
+    thread->scheduler->AddThread(thread);
     thread->tls_address = thread->owner_process->MarkNextAvailableTLSSlotAsUsed(*thread);
 
     // TODO(peachum): move to ScheduleThread() when scheduler is added so selected core is used
@@ -351,7 +352,7 @@ void Thread::ChangeScheduler() {
     if (*new_processor_id != processor_id) {
         // Remove thread from previous core's scheduler
         scheduler->RemoveThread(this);
-        next_scheduler.AddThread(this, current_priority);
+        next_scheduler.AddThread(this);
     }
 
     processor_id = *new_processor_id;

src/core/hle/kernel/thread.h

@@ -51,7 +51,8 @@ enum class ThreadStatus {
     WaitIPC,      ///< Waiting for the reply from an IPC request
     WaitSynchAny, ///< Waiting due to WaitSynch1 or WaitSynchN with wait_all = false
     WaitSynchAll, ///< Waiting due to WaitSynchronizationN with wait_all = true
-    WaitMutex,    ///< Waiting due to an ArbitrateLock/WaitProcessWideKey svc
+    WaitMutex,    ///< Waiting due to an ArbitrateLock svc
+    WaitCondVar,  ///< Waiting due to an WaitProcessWideKey svc
     WaitArb,      ///< Waiting due to a SignalToAddress/WaitForAddress svc
     Dormant,      ///< Created but not yet made ready
     Dead          ///< Run to completion, or forcefully terminated

src/core/hle/kernel/transfer_memory.cpp

@@ -0,0 +1,73 @@
+// Copyright 2019 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/kernel/errors.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/process.h"
+#include "core/hle/kernel/shared_memory.h"
+#include "core/hle/kernel/transfer_memory.h"
+#include "core/hle/result.h"
+
+namespace Kernel {
+
+TransferMemory::TransferMemory(KernelCore& kernel) : Object{kernel} {}
+TransferMemory::~TransferMemory() = default;
+
+SharedPtr<TransferMemory> TransferMemory::Create(KernelCore& kernel, VAddr base_address,
+                                                 size_t size, MemoryPermission permissions) {
+    SharedPtr<TransferMemory> transfer_memory{new TransferMemory(kernel)};
+
+    transfer_memory->base_address = base_address;
+    transfer_memory->memory_size = size;
+    transfer_memory->owner_permissions = permissions;
+    transfer_memory->owner_process = kernel.CurrentProcess();
+
+    return transfer_memory;
+}
+
+ResultCode TransferMemory::MapMemory(VAddr address, size_t size, MemoryPermission permissions) {
+    if (memory_size != size) {
+        return ERR_INVALID_SIZE;
+    }
+
+    if (owner_permissions != permissions) {
+        return ERR_INVALID_STATE;
+    }
+
+    if (is_mapped) {
+        return ERR_INVALID_STATE;
+    }
+
+    const auto map_state = owner_permissions == MemoryPermission::None
+                               ? MemoryState::TransferMemoryIsolated
+                               : MemoryState::TransferMemory;
+    auto& vm_manager = owner_process->VMManager();
+    const auto map_result = vm_manager.MapMemoryBlock(
+        address, std::make_shared<std::vector<u8>>(size), 0, size, map_state);
+
+    if (map_result.Failed()) {
+        return map_result.Code();
+    }
+
+    is_mapped = true;
+    return RESULT_SUCCESS;
+}
+
+ResultCode TransferMemory::UnmapMemory(VAddr address, size_t size) {
+    if (memory_size != size) {
+        return ERR_INVALID_SIZE;
+    }
+
+    auto& vm_manager = owner_process->VMManager();
+    const auto result = vm_manager.UnmapRange(address, size);
+
+    if (result.IsError()) {
+        return result;
+    }
+
+    is_mapped = false;
+    return RESULT_SUCCESS;
+}
+
+} // namespace Kernel

src/core/hle/kernel/transfer_memory.h

@@ -0,0 +1,91 @@
+// Copyright 2019 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/kernel/object.h"
+
+union ResultCode;
+
+namespace Kernel {
+
+class KernelCore;
+class Process;
+
+enum class MemoryPermission : u32;
+
+/// Defines the interface for transfer memory objects.
+///
+/// Transfer memory is typically used for the purpose of
+/// transferring memory between separate process instances,
+/// thus the name.
+///
+class TransferMemory final : public Object {
+public:
+    static constexpr HandleType HANDLE_TYPE = HandleType::TransferMemory;
+
+    static SharedPtr<TransferMemory> Create(KernelCore& kernel, VAddr base_address, size_t size,
+                                            MemoryPermission permissions);
+
+    TransferMemory(const TransferMemory&) = delete;
+    TransferMemory& operator=(const TransferMemory&) = delete;
+
+    TransferMemory(TransferMemory&&) = delete;
+    TransferMemory& operator=(TransferMemory&&) = delete;
+
+    std::string GetTypeName() const override {
+        return "TransferMemory";
+    }
+
+    std::string GetName() const override {
+        return GetTypeName();
+    }
+
+    HandleType GetHandleType() const override {
+        return HANDLE_TYPE;
+    }
+
+    /// Attempts to map transfer memory with the given range and memory permissions.
+    ///
+    /// @param address     The base address to being mapping memory at.
+    /// @param size        The size of the memory to map, in bytes.
+    /// @param permissions The memory permissions to check against when mapping memory.
+    ///
+    /// @pre The given address, size, and memory permissions must all match
+    ///      the same values that were given when creating the transfer memory
+    ///      instance.
+    ///
+    ResultCode MapMemory(VAddr address, size_t size, MemoryPermission permissions);
+
+    /// Unmaps the transfer memory with the given range
+    ///
+    /// @param address The base address to begin unmapping memory at.
+    /// @param size    The size of the memory to unmap, in bytes.
+    ///
+    /// @pre The given address and size must be the same as the ones used
+    ///      to create the transfer memory instance.
+    ///
+    ResultCode UnmapMemory(VAddr address, size_t size);
+
+private:
+    explicit TransferMemory(KernelCore& kernel);
+    ~TransferMemory() override;
+
+    /// The base address for the memory managed by this instance.
+    VAddr base_address = 0;
+
+    /// Size of the memory, in bytes, that this instance manages.
+    size_t memory_size = 0;
+
+    /// The memory permissions that are applied to this instance.
+    MemoryPermission owner_permissions{};
+
+    /// The process that this transfer memory instance was created under.
+    Process* owner_process = nullptr;
+
+    /// Whether or not this transfer memory instance has mapped memory.
+    bool is_mapped = false;
+};
+
+} // namespace Kernel

src/core/hle/service/fatal/fatal.cpp

@@ -25,21 +25,34 @@ Module::Interface::Interface(std::shared_ptr<Module> module, const char* name)
 Module::Interface::~Interface() = default;
 
 struct FatalInfo {
-    std::array<u64_le, 31> registers{}; // TODO(ogniK): See if this actually is registers or
-                                        // not(find a game which has non zero valeus)
-    u64_le unk0{};
-    u64_le unk1{};
-    u64_le unk2{};
-    u64_le unk3{};
-    u64_le unk4{};
-    u64_le unk5{};
-    u64_le unk6{};
+    enum class Architecture : s32 {
+        AArch64,
+        AArch32,
+    };
+
+    const char* ArchAsString() const {
+        return arch == Architecture::AArch64 ? "AArch64" : "AArch32";
+    }
+
+    std::array<u64_le, 31> registers{};
+    u64_le sp{};
+    u64_le pc{};
+    u64_le pstate{};
+    u64_le afsr0{};
+    u64_le afsr1{};
+    u64_le esr{};
+    u64_le far{};
 
     std::array<u64_le, 32> backtrace{};
-    u64_le unk7{};
-    u64_le unk8{};
+    u64_le program_entry_point{};
+
+    // Bit flags that indicate which registers have been set with values
+    // for this context. The service itself uses these to determine which
+    // registers to specifically print out.
+    u64_le set_flags{};
+
     u32_le backtrace_size{};
-    u32_le unk9{};
+    Architecture arch{};
     u32_le unk10{}; // TODO(ogniK): Is this even used or is it just padding?
 };
 static_assert(sizeof(FatalInfo) == 0x250, "FatalInfo is an invalid size");
@@ -52,36 +65,36 @@ enum class FatalType : u32 {
 
 static void GenerateErrorReport(ResultCode error_code, const FatalInfo& info) {
     const auto title_id = Core::CurrentProcess()->GetTitleID();
-    std::string crash_report =
-        fmt::format("Yuzu {}-{} crash report\n"
-                    "Title ID:                        {:016x}\n"
-                    "Result:                          0x{:X} ({:04}-{:04d})\n"
-                    "\n",
-                    Common::g_scm_branch, Common::g_scm_desc, title_id, error_code.raw,
-                    2000 + static_cast<u32>(error_code.module.Value()),
-                    static_cast<u32>(error_code.description.Value()), info.unk8, info.unk7);
+    std::string crash_report = fmt::format(
+        "Yuzu {}-{} crash report\n"
+        "Title ID:                        {:016x}\n"
+        "Result:                          0x{:X} ({:04}-{:04d})\n"
+        "Set flags:                       0x{:16X}\n"
+        "Program entry point:             0x{:16X}\n"
+        "\n",
+        Common::g_scm_branch, Common::g_scm_desc, title_id, error_code.raw,
+        2000 + static_cast<u32>(error_code.module.Value()),
+        static_cast<u32>(error_code.description.Value()), info.set_flags, info.program_entry_point);
     if (info.backtrace_size != 0x0) {
         crash_report += "Registers:\n";
-        // TODO(ogniK): This is just a guess, find a game which actually has non zero values
         for (size_t i = 0; i < info.registers.size(); i++) {
             crash_report +=
                 fmt::format("    X[{:02d}]:                       {:016x}\n", i, info.registers[i]);
         }
-        crash_report += fmt::format("    Unknown 0:                   {:016x}\n", info.unk0);
-        crash_report += fmt::format("    Unknown 1:                   {:016x}\n", info.unk1);
-        crash_report += fmt::format("    Unknown 2:                   {:016x}\n", info.unk2);
-        crash_report += fmt::format("    Unknown 3:                   {:016x}\n", info.unk3);
-        crash_report += fmt::format("    Unknown 4:                   {:016x}\n", info.unk4);
-        crash_report += fmt::format("    Unknown 5:                   {:016x}\n", info.unk5);
-        crash_report += fmt::format("    Unknown 6:                   {:016x}\n", info.unk6);
+        crash_report += fmt::format("    SP:                          {:016x}\n", info.sp);
+        crash_report += fmt::format("    PC:                          {:016x}\n", info.pc);
+        crash_report += fmt::format("    PSTATE:                      {:016x}\n", info.pstate);
+        crash_report += fmt::format("    AFSR0:                       {:016x}\n", info.afsr0);
+        crash_report += fmt::format("    AFSR1:                       {:016x}\n", info.afsr1);
+        crash_report += fmt::format("    ESR:                         {:016x}\n", info.esr);
+        crash_report += fmt::format("    FAR:                         {:016x}\n", info.far);
         crash_report += "\nBacktrace:\n";
         for (size_t i = 0; i < info.backtrace_size; i++) {
             crash_report +=
                 fmt::format("    Backtrace[{:02d}]:               {:016x}\n", i, info.backtrace[i]);
         }
-        crash_report += fmt::format("\nUnknown 7:                       0x{:016x}\n", info.unk7);
-        crash_report += fmt::format("Unknown 8:                       0x{:016x}\n", info.unk8);
-        crash_report += fmt::format("Unknown 9:                       0x{:016x}\n", info.unk9);
+
+        crash_report += fmt::format("Architecture:                    {}\n", info.ArchAsString());
         crash_report += fmt::format("Unknown 10:                      0x{:016x}\n", info.unk10);
     }
 
@@ -125,13 +138,13 @@ static void ThrowFatalError(ResultCode error_code, FatalType fatal_type, const F
     case FatalType::ErrorReport:
         GenerateErrorReport(error_code, info);
         break;
-    };
+    }
 }
 
 void Module::Interface::ThrowFatal(Kernel::HLERequestContext& ctx) {
     LOG_ERROR(Service_Fatal, "called");
     IPC::RequestParser rp{ctx};
-    auto error_code = rp.Pop<ResultCode>();
+    const auto error_code = rp.Pop<ResultCode>();
 
     ThrowFatalError(error_code, FatalType::ErrorScreen, {});
     IPC::ResponseBuilder rb{ctx, 2};
@@ -141,8 +154,8 @@ void Module::Interface::ThrowFatal(Kernel::HLERequestContext& ctx) {
 void Module::Interface::ThrowFatalWithPolicy(Kernel::HLERequestContext& ctx) {
     LOG_ERROR(Service_Fatal, "called");
     IPC::RequestParser rp(ctx);
-    auto error_code = rp.Pop<ResultCode>();
-    auto fatal_type = rp.PopEnum<FatalType>();
+    const auto error_code = rp.Pop<ResultCode>();
+    const auto fatal_type = rp.PopEnum<FatalType>();
 
     ThrowFatalError(error_code, fatal_type, {}); // No info is passed with ThrowFatalWithPolicy
     IPC::ResponseBuilder rb{ctx, 2};
@@ -152,9 +165,9 @@ void Module::Interface::ThrowFatalWithPolicy(Kernel::HLERequestContext& ctx) {
 void Module::Interface::ThrowFatalWithCpuContext(Kernel::HLERequestContext& ctx) {
     LOG_ERROR(Service_Fatal, "called");
     IPC::RequestParser rp(ctx);
-    auto error_code = rp.Pop<ResultCode>();
-    auto fatal_type = rp.PopEnum<FatalType>();
-    auto fatal_info = ctx.ReadBuffer();
+    const auto error_code = rp.Pop<ResultCode>();
+    const auto fatal_type = rp.PopEnum<FatalType>();
+    const auto fatal_info = ctx.ReadBuffer();
     FatalInfo info{};
 
     ASSERT_MSG(fatal_info.size() == sizeof(FatalInfo), "Invalid fatal info buffer size!");

src/core/hle/service/hid/hid.cpp

@@ -36,9 +36,9 @@ namespace Service::HID {
 
 // Updating period for each HID device.
 // TODO(ogniK): Find actual polling rate of hid
-constexpr u64 pad_update_ticks = Core::Timing::BASE_CLOCK_RATE / 66;
-constexpr u64 accelerometer_update_ticks = Core::Timing::BASE_CLOCK_RATE / 100;
-constexpr u64 gyroscope_update_ticks = Core::Timing::BASE_CLOCK_RATE / 100;
+constexpr s64 pad_update_ticks = static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / 66);
+constexpr s64 accelerometer_update_ticks = static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / 100);
+constexpr s64 gyroscope_update_ticks = static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / 100);
 constexpr std::size_t SHARED_MEMORY_SIZE = 0x40000;
 
 IAppletResource::IAppletResource() : ServiceFramework("IAppletResource") {
@@ -75,7 +75,7 @@ IAppletResource::IAppletResource() : ServiceFramework("IAppletResource") {
     // Register update callbacks
     auto& core_timing = Core::System::GetInstance().CoreTiming();
     pad_update_event =
-        core_timing.RegisterEvent("HID::UpdatePadCallback", [this](u64 userdata, int cycles_late) {
+        core_timing.RegisterEvent("HID::UpdatePadCallback", [this](u64 userdata, s64 cycles_late) {
             UpdateControllers(userdata, cycles_late);
         });
 
@@ -106,7 +106,7 @@ void IAppletResource::GetSharedMemoryHandle(Kernel::HLERequestContext& ctx) {
     rb.PushCopyObjects(shared_mem);
 }
 
-void IAppletResource::UpdateControllers(u64 userdata, int cycles_late) {
+void IAppletResource::UpdateControllers(u64 userdata, s64 cycles_late) {
     auto& core_timing = Core::System::GetInstance().CoreTiming();
 
     const bool should_reload = Settings::values.is_device_reload_pending.exchange(false);

src/core/hle/service/hid/hid.h

@@ -65,7 +65,7 @@ private:
     }
 
     void GetSharedMemoryHandle(Kernel::HLERequestContext& ctx);
-    void UpdateControllers(u64 userdata, int cycles_late);
+    void UpdateControllers(u64 userdata, s64 cycles_late);
 
     Kernel::SharedPtr<Kernel::SharedMemory> shared_mem;
 

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

@@ -150,7 +150,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
-        rb.PushRaw<u8>(Settings::values.enable_nfc);
+        rb.PushRaw<u8>(true);
     }
 
     void GetStateOld(Kernel::HLERequestContext& ctx) {

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

@@ -335,7 +335,7 @@ void Module::Interface::CreateUserInterface(Kernel::HLERequestContext& ctx) {
 }
 
 bool Module::Interface::LoadAmiibo(const std::vector<u8>& buffer) {
-    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
+    std::lock_guard lock{HLE::g_hle_lock};
     if (buffer.size() < sizeof(AmiiboFile)) {
         return false;
     }

src/core/hle/service/nvflinger/nvflinger.cpp

@@ -26,7 +26,7 @@
 namespace Service::NVFlinger {
 
 constexpr std::size_t SCREEN_REFRESH_RATE = 60;
-constexpr u64 frame_ticks = static_cast<u64>(Core::Timing::BASE_CLOCK_RATE / SCREEN_REFRESH_RATE);
+constexpr s64 frame_ticks = static_cast<s64>(Core::Timing::BASE_CLOCK_RATE / SCREEN_REFRESH_RATE);
 
 NVFlinger::NVFlinger(Core::Timing::CoreTiming& core_timing) : core_timing{core_timing} {
     displays.emplace_back(0, "Default");
@@ -37,7 +37,7 @@ NVFlinger::NVFlinger(Core::Timing::CoreTiming& core_timing) : core_timing{core_t
 
     // Schedule the screen composition events
     composition_event =
-        core_timing.RegisterEvent("ScreenComposition", [this](u64 userdata, int cycles_late) {
+        core_timing.RegisterEvent("ScreenComposition", [this](u64 userdata, s64 cycles_late) {
             Compose();
             this->core_timing.ScheduleEvent(frame_ticks - cycles_late, composition_event);
         });

src/core/perf_stats.cpp

@@ -18,13 +18,13 @@ using std::chrono::microseconds;
 namespace Core {
 
 void PerfStats::BeginSystemFrame() {
-    std::lock_guard<std::mutex> lock(object_mutex);
+    std::lock_guard lock{object_mutex};
 
     frame_begin = Clock::now();
 }
 
 void PerfStats::EndSystemFrame() {
-    std::lock_guard<std::mutex> lock(object_mutex);
+    std::lock_guard lock{object_mutex};
 
     auto frame_end = Clock::now();
     accumulated_frametime += frame_end - frame_begin;
@@ -35,13 +35,13 @@ void PerfStats::EndSystemFrame() {
 }
 
 void PerfStats::EndGameFrame() {
-    std::lock_guard<std::mutex> lock(object_mutex);
+    std::lock_guard lock{object_mutex};
 
     game_frames += 1;
 }
 
 PerfStatsResults PerfStats::GetAndResetStats(microseconds current_system_time_us) {
-    std::lock_guard<std::mutex> lock(object_mutex);
+    std::lock_guard lock{object_mutex};
 
     const auto now = Clock::now();
     // Walltime elapsed since stats were reset
@@ -67,7 +67,7 @@ PerfStatsResults PerfStats::GetAndResetStats(microseconds current_system_time_us
 }
 
 double PerfStats::GetLastFrameTimeScale() {
-    std::lock_guard<std::mutex> lock(object_mutex);
+    std::lock_guard lock{object_mutex};
 
     constexpr double FRAME_LENGTH = 1.0 / 60;
     return duration_cast<DoubleSecs>(previous_frame_length).count() / FRAME_LENGTH;

src/core/settings.cpp

@@ -82,7 +82,6 @@ void LogSetting(const std::string& name, const T& value) {
 void LogSettings() {
     LOG_INFO(Config, "yuzu Configuration:");
     LogSetting("System_UseDockedMode", Settings::values.use_docked_mode);
-    LogSetting("System_EnableNfc", Settings::values.enable_nfc);
     LogSetting("System_RngSeed", Settings::values.rng_seed.value_or(0));
     LogSetting("System_CurrentUser", Settings::values.current_user);
     LogSetting("System_LanguageIndex", Settings::values.language_index);

src/core/settings.h

@@ -349,7 +349,6 @@ struct TouchscreenInput {
 struct Values {
     // System
     bool use_docked_mode;
-    bool enable_nfc;
     std::optional<u32> rng_seed;
     // Measured in seconds since epoch
     std::optional<std::chrono::seconds> custom_rtc;

src/input_common/keyboard.cpp

@@ -36,18 +36,18 @@ struct KeyButtonPair {
 class KeyButtonList {
 public:
     void AddKeyButton(int key_code, KeyButton* key_button) {
-        std::lock_guard<std::mutex> guard(mutex);
+        std::lock_guard guard{mutex};
         list.push_back(KeyButtonPair{key_code, key_button});
     }
 
     void RemoveKeyButton(const KeyButton* key_button) {
-        std::lock_guard<std::mutex> guard(mutex);
+        std::lock_guard guard{mutex};
         list.remove_if(
             [key_button](const KeyButtonPair& pair) { return pair.key_button == key_button; });
     }
 
     void ChangeKeyStatus(int key_code, bool pressed) {
-        std::lock_guard<std::mutex> guard(mutex);
+        std::lock_guard guard{mutex};
         for (const KeyButtonPair& pair : list) {
             if (pair.key_code == key_code)
                 pair.key_button->status.store(pressed);
@@ -55,7 +55,7 @@ public:
     }
 
     void ChangeAllKeyStatus(bool pressed) {
-        std::lock_guard<std::mutex> guard(mutex);
+        std::lock_guard guard{mutex};
         for (const KeyButtonPair& pair : list) {
             pair.key_button->status.store(pressed);
         }

src/input_common/motion_emu.cpp

@@ -39,7 +39,7 @@ public:
     void Tilt(int x, int y) {
         auto mouse_move = Common::MakeVec(x, y) - mouse_origin;
         if (is_tilting) {
-            std::lock_guard<std::mutex> guard(tilt_mutex);
+            std::lock_guard guard{tilt_mutex};
             if (mouse_move.x == 0 && mouse_move.y == 0) {
                 tilt_angle = 0;
             } else {
@@ -51,13 +51,13 @@ public:
     }
 
     void EndTilt() {
-        std::lock_guard<std::mutex> guard(tilt_mutex);
+        std::lock_guard guard{tilt_mutex};
         tilt_angle = 0;
         is_tilting = false;
     }
 
     std::tuple<Common::Vec3<float>, Common::Vec3<float>> GetStatus() {
-        std::lock_guard<std::mutex> guard(status_mutex);
+        std::lock_guard guard{status_mutex};
         return status;
     }
 
@@ -93,7 +93,7 @@ private:
             old_q = q;
 
             {
-                std::lock_guard<std::mutex> guard(tilt_mutex);
+                std::lock_guard guard{tilt_mutex};
 
                 // Find the quaternion describing current 3DS tilting
                 q = Common::MakeQuaternion(
@@ -115,7 +115,7 @@ private:
 
             // Update the sensor state
             {
-                std::lock_guard<std::mutex> guard(status_mutex);
+                std::lock_guard guard{status_mutex};
                 status = std::make_tuple(gravity, angular_rate);
             }
         }

src/input_common/sdl/sdl_impl.cpp

@@ -55,22 +55,22 @@ public:
         : guid{std::move(guid_)}, port{port_}, sdl_joystick{joystick, deleter} {}
 
     void SetButton(int button, bool value) {
-        std::lock_guard<std::mutex> lock(mutex);
+        std::lock_guard lock{mutex};
         state.buttons[button] = value;
     }
 
     bool GetButton(int button) const {
-        std::lock_guard<std::mutex> lock(mutex);
+        std::lock_guard lock{mutex};
         return state.buttons.at(button);
     }
 
     void SetAxis(int axis, Sint16 value) {
-        std::lock_guard<std::mutex> lock(mutex);
+        std::lock_guard lock{mutex};
         state.axes[axis] = value;
     }
 
     float GetAxis(int axis) const {
-        std::lock_guard<std::mutex> lock(mutex);
+        std::lock_guard lock{mutex};
         return state.axes.at(axis) / 32767.0f;
     }
 
@@ -92,12 +92,12 @@ public:
     }
 
     void SetHat(int hat, Uint8 direction) {
-        std::lock_guard<std::mutex> lock(mutex);
+        std::lock_guard lock{mutex};
         state.hats[hat] = direction;
     }
 
     bool GetHatDirection(int hat, Uint8 direction) const {
-        std::lock_guard<std::mutex> lock(mutex);
+        std::lock_guard lock{mutex};
         return (state.hats.at(hat) & direction) != 0;
     }
     /**
@@ -140,7 +140,7 @@ private:
  * Get the nth joystick with the corresponding GUID
  */
 std::shared_ptr<SDLJoystick> SDLState::GetSDLJoystickByGUID(const std::string& guid, int port) {
-    std::lock_guard<std::mutex> lock(joystick_map_mutex);
+    std::lock_guard lock{joystick_map_mutex};
     const auto it = joystick_map.find(guid);
     if (it != joystick_map.end()) {
         while (it->second.size() <= port) {
@@ -161,7 +161,8 @@ std::shared_ptr<SDLJoystick> SDLState::GetSDLJoystickByGUID(const std::string& g
 std::shared_ptr<SDLJoystick> SDLState::GetSDLJoystickBySDLID(SDL_JoystickID sdl_id) {
     auto sdl_joystick = SDL_JoystickFromInstanceID(sdl_id);
     const std::string guid = GetGUID(sdl_joystick);
-    std::lock_guard<std::mutex> lock(joystick_map_mutex);
+
+    std::lock_guard lock{joystick_map_mutex};
     auto map_it = joystick_map.find(guid);
     if (map_it != joystick_map.end()) {
         auto vec_it = std::find_if(map_it->second.begin(), map_it->second.end(),
@@ -198,8 +199,9 @@ void SDLState::InitJoystick(int joystick_index) {
         LOG_ERROR(Input, "failed to open joystick {}", joystick_index);
         return;
     }
-    std::string guid = GetGUID(sdl_joystick);
-    std::lock_guard<std::mutex> lock(joystick_map_mutex);
+    const std::string guid = GetGUID(sdl_joystick);
+
+    std::lock_guard lock{joystick_map_mutex};
     if (joystick_map.find(guid) == joystick_map.end()) {
         auto joystick = std::make_shared<SDLJoystick>(guid, 0, sdl_joystick);
         joystick_map[guid].emplace_back(std::move(joystick));
@@ -221,7 +223,7 @@ void SDLState::CloseJoystick(SDL_Joystick* sdl_joystick) {
     std::string guid = GetGUID(sdl_joystick);
     std::shared_ptr<SDLJoystick> joystick;
     {
-        std::lock_guard<std::mutex> lock(joystick_map_mutex);
+        std::lock_guard lock{joystick_map_mutex};
         // This call to guid is safe since the joystick is guaranteed to be in the map
         auto& joystick_guid_list = joystick_map[guid];
         const auto joystick_it =
@@ -274,7 +276,7 @@ void SDLState::HandleGameControllerEvent(const SDL_Event& event) {
 }
 
 void SDLState::CloseJoysticks() {
-    std::lock_guard<std::mutex> lock(joystick_map_mutex);
+    std::lock_guard lock{joystick_map_mutex};
     joystick_map.clear();
 }
 

src/tests/CMakeLists.txt

@@ -1,5 +1,7 @@
 add_executable(tests
     common/bit_field.cpp
+    common/bit_utils.cpp
+    common/multi_level_queue.cpp
     common/param_package.cpp
     common/ring_buffer.cpp
     core/arm/arm_test_common.cpp

src/tests/common/bit_utils.cpp

@@ -0,0 +1,23 @@
+// Copyright 2017 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <catch2/catch.hpp>
+#include <math.h>
+#include "common/bit_util.h"
+
+namespace Common {
+
+TEST_CASE("BitUtils::CountTrailingZeroes", "[common]") {
+    REQUIRE(Common::CountTrailingZeroes32(0) == 32);
+    REQUIRE(Common::CountTrailingZeroes64(0) == 64);
+    REQUIRE(Common::CountTrailingZeroes32(9) == 0);
+    REQUIRE(Common::CountTrailingZeroes32(8) == 3);
+    REQUIRE(Common::CountTrailingZeroes32(0x801000) == 12);
+    REQUIRE(Common::CountTrailingZeroes64(9) == 0);
+    REQUIRE(Common::CountTrailingZeroes64(8) == 3);
+    REQUIRE(Common::CountTrailingZeroes64(0x801000) == 12);
+    REQUIRE(Common::CountTrailingZeroes64(0x801000000000UL) == 36);
+}
+
+} // namespace Common

src/tests/common/multi_level_queue.cpp

@@ -0,0 +1,55 @@
+// Copyright 2019 Yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <catch2/catch.hpp>
+#include <math.h>
+#include "common/common_types.h"
+#include "common/multi_level_queue.h"
+
+namespace Common {
+
+TEST_CASE("MultiLevelQueue", "[common]") {
+    std::array<f32, 8> values = {0.0, 5.0, 1.0, 9.0, 8.0, 2.0, 6.0, 7.0};
+    Common::MultiLevelQueue<f32, 64> mlq;
+    REQUIRE(mlq.empty());
+    mlq.add(values[2], 2);
+    mlq.add(values[7], 7);
+    mlq.add(values[3], 3);
+    mlq.add(values[4], 4);
+    mlq.add(values[0], 0);
+    mlq.add(values[5], 5);
+    mlq.add(values[6], 6);
+    mlq.add(values[1], 1);
+    u32 index = 0;
+    bool all_set = true;
+    for (auto& f : mlq) {
+        all_set &= (f == values[index]);
+        index++;
+    }
+    REQUIRE(all_set);
+    REQUIRE(!mlq.empty());
+    f32 v = 8.0;
+    mlq.add(v, 2);
+    v = -7.0;
+    mlq.add(v, 2, false);
+    REQUIRE(mlq.front(2) == -7.0);
+    mlq.yield(2);
+    REQUIRE(mlq.front(2) == values[2]);
+    REQUIRE(mlq.back(2) == -7.0);
+    REQUIRE(mlq.empty(8));
+    v = 10.0;
+    mlq.add(v, 8);
+    mlq.adjust(v, 8, 9);
+    REQUIRE(mlq.front(9) == v);
+    REQUIRE(mlq.empty(8));
+    REQUIRE(!mlq.empty(9));
+    mlq.adjust(values[0], 0, 9);
+    REQUIRE(mlq.highest_priority_set() == 1);
+    REQUIRE(mlq.lowest_priority_set() == 9);
+    mlq.remove(values[1], 1);
+    REQUIRE(mlq.highest_priority_set() == 2);
+    REQUIRE(mlq.empty(1));
+}
+
+} // namespace Common

src/video_core/debug_utils/debug_utils.cpp

@@ -10,7 +10,7 @@ namespace Tegra {
 
 void DebugContext::DoOnEvent(Event event, void* data) {
     {
-        std::unique_lock<std::mutex> lock(breakpoint_mutex);
+        std::unique_lock lock{breakpoint_mutex};
 
         // TODO(Subv): Commit the rasterizer's caches so framebuffers, render targets, etc. will
         // show on debug widgets
@@ -32,7 +32,7 @@ void DebugContext::DoOnEvent(Event event, void* data) {
 
 void DebugContext::Resume() {
     {
-        std::lock_guard<std::mutex> lock(breakpoint_mutex);
+        std::lock_guard lock{breakpoint_mutex};
 
         // Tell all observers that we are about to resume
         for (auto& breakpoint_observer : breakpoint_observers) {

src/video_core/debug_utils/debug_utils.h

@@ -40,7 +40,7 @@ public:
         /// Constructs the object such that it observes events of the given DebugContext.
         explicit BreakPointObserver(std::shared_ptr<DebugContext> debug_context)
             : context_weak(debug_context) {
-            std::unique_lock<std::mutex> lock(debug_context->breakpoint_mutex);
+            std::unique_lock lock{debug_context->breakpoint_mutex};
             debug_context->breakpoint_observers.push_back(this);
         }
 
@@ -48,7 +48,7 @@ public:
             auto context = context_weak.lock();
             if (context) {
                 {
-                    std::unique_lock<std::mutex> lock(context->breakpoint_mutex);
+                    std::unique_lock lock{context->breakpoint_mutex};
                     context->breakpoint_observers.remove(this);
                 }
 

src/video_core/gpu.cpp

@@ -286,9 +286,10 @@ void GPU::ProcessSemaphoreTriggerMethod() {
         // TODO(Kmather73): Generate a real GPU timestamp and write it here instead of
         // CoreTiming
         block.timestamp = Core::System::GetInstance().CoreTiming().GetTicks();
-        memory_manager->WriteBlock(regs.smaphore_address.SmaphoreAddress(), &block, sizeof(block));
+        memory_manager->WriteBlock(regs.semaphore_address.SemaphoreAddress(), &block,
+                                   sizeof(block));
     } else {
-        const u32 word{memory_manager->Read<u32>(regs.smaphore_address.SmaphoreAddress())};
+        const u32 word{memory_manager->Read<u32>(regs.semaphore_address.SemaphoreAddress())};
         if ((op == GpuSemaphoreOperation::AcquireEqual && word == regs.semaphore_sequence) ||
             (op == GpuSemaphoreOperation::AcquireGequal &&
              static_cast<s32>(word - regs.semaphore_sequence) > 0) ||
@@ -315,11 +316,11 @@ void GPU::ProcessSemaphoreTriggerMethod() {
 }
 
 void GPU::ProcessSemaphoreRelease() {
-    memory_manager->Write<u32>(regs.smaphore_address.SmaphoreAddress(), regs.semaphore_release);
+    memory_manager->Write<u32>(regs.semaphore_address.SemaphoreAddress(), regs.semaphore_release);
 }
 
 void GPU::ProcessSemaphoreAcquire() {
-    const u32 word = memory_manager->Read<u32>(regs.smaphore_address.SmaphoreAddress());
+    const u32 word = memory_manager->Read<u32>(regs.semaphore_address.SemaphoreAddress());
     const auto value = regs.semaphore_acquire;
     if (word != value) {
         regs.acquire_active = true;

src/video_core/gpu.h

@@ -177,11 +177,11 @@ public:
                     u32 address_high;
                     u32 address_low;
 
-                    GPUVAddr SmaphoreAddress() const {
+                    GPUVAddr SemaphoreAddress() const {
                         return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) |
                                                      address_low);
                     }
-                } smaphore_address;
+                } semaphore_address;
 
                 u32 semaphore_sequence;
                 u32 semaphore_trigger;
@@ -263,7 +263,7 @@ private:
     static_assert(offsetof(GPU::Regs, field_name) == position * 4,                                 \
                   "Field " #field_name " has invalid position")
 
-ASSERT_REG_POSITION(smaphore_address, 0x4);
+ASSERT_REG_POSITION(semaphore_address, 0x4);
 ASSERT_REG_POSITION(semaphore_sequence, 0x6);
 ASSERT_REG_POSITION(semaphore_trigger, 0x7);
 ASSERT_REG_POSITION(reference_count, 0x14);

src/video_core/gpu_asynch.cpp

@@ -9,7 +9,7 @@
 namespace VideoCommon {
 
 GPUAsynch::GPUAsynch(Core::System& system, VideoCore::RendererBase& renderer)
-    : Tegra::GPU(system, renderer), gpu_thread{renderer, *dma_pusher} {}
+    : Tegra::GPU(system, renderer), gpu_thread{system, renderer, *dma_pusher} {}
 
 GPUAsynch::~GPUAsynch() = default;
 

src/video_core/gpu_thread.cpp

@@ -4,6 +4,9 @@
 
 #include "common/assert.h"
 #include "common/microprofile.h"
+#include "core/core.h"
+#include "core/core_timing.h"
+#include "core/core_timing_util.h"
 #include "core/frontend/scope_acquire_window_context.h"
 #include "video_core/dma_pusher.h"
 #include "video_core/gpu.h"
@@ -36,7 +39,6 @@ static void RunThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_p
                 dma_pusher.Push(std::move(submit_list->entries));
                 dma_pusher.DispatchCalls();
             } else if (const auto data = std::get_if<SwapBuffersCommand>(&next.data)) {
-                state.DecrementFramesCounter();
                 renderer.SwapBuffers(std::move(data->framebuffer));
             } else if (const auto data = std::get_if<FlushRegionCommand>(&next.data)) {
                 renderer.Rasterizer().FlushRegion(data->addr, data->size);
@@ -47,13 +49,18 @@ static void RunThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_p
             } else {
                 UNREACHABLE();
             }
+            state.signaled_fence = next.fence;
+            state.TrySynchronize();
         }
     }
 }
 
-ThreadManager::ThreadManager(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_pusher)
-    : renderer{renderer}, dma_pusher{dma_pusher}, thread{RunThread, std::ref(renderer),
-                                                         std::ref(dma_pusher), std::ref(state)} {}
+ThreadManager::ThreadManager(Core::System& system, VideoCore::RendererBase& renderer,
+                             Tegra::DmaPusher& dma_pusher)
+    : system{system}, thread{RunThread, std::ref(renderer), std::ref(dma_pusher), std::ref(state)} {
+    synchronization_event = system.CoreTiming().RegisterEvent(
+        "GPUThreadSynch", [this](u64 fence, int) { state.WaitForSynchronization(fence); });
+}
 
 ThreadManager::~ThreadManager() {
     // Notify GPU thread that a shutdown is pending
@@ -62,14 +69,14 @@ ThreadManager::~ThreadManager() {
 }
 
 void ThreadManager::SubmitList(Tegra::CommandList&& entries) {
-    PushCommand(SubmitListCommand(std::move(entries)));
+    const u64 fence{PushCommand(SubmitListCommand(std::move(entries)))};
+    const s64 synchronization_ticks{Core::Timing::usToCycles(9000)};
+    system.CoreTiming().ScheduleEvent(synchronization_ticks, synchronization_event, fence);
 }
 
 void ThreadManager::SwapBuffers(
     std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) {
-    state.IncrementFramesCounter();
     PushCommand(SwapBuffersCommand(std::move(framebuffer)));
-    state.WaitForFrames();
 }
 
 void ThreadManager::FlushRegion(CacheAddr addr, u64 size) {
@@ -79,7 +86,7 @@ void ThreadManager::FlushRegion(CacheAddr addr, u64 size) {
 void ThreadManager::InvalidateRegion(CacheAddr addr, u64 size) {
     if (state.queue.Empty()) {
         // It's quicker to invalidate a single region on the CPU if the queue is already empty
-        renderer.Rasterizer().InvalidateRegion(addr, size);
+        system.Renderer().Rasterizer().InvalidateRegion(addr, size);
     } else {
         PushCommand(InvalidateRegionCommand(addr, size));
     }
@@ -90,9 +97,25 @@ void ThreadManager::FlushAndInvalidateRegion(CacheAddr addr, u64 size) {
     InvalidateRegion(addr, size);
 }
 
-void ThreadManager::PushCommand(CommandData&& command_data) {
-    state.queue.Push(CommandDataContainer(std::move(command_data)));
+u64 ThreadManager::PushCommand(CommandData&& command_data) {
+    const u64 fence{++state.last_fence};
+    state.queue.Push(CommandDataContainer(std::move(command_data), fence));
     state.SignalCommands();
+    return fence;
+}
+
+MICROPROFILE_DEFINE(GPU_wait, "GPU", "Wait for the GPU", MP_RGB(128, 128, 192));
+void SynchState::WaitForSynchronization(u64 fence) {
+    if (signaled_fence >= fence) {
+        return;
+    }
+
+    // Wait for the GPU to be idle (all commands to be executed)
+    {
+        MICROPROFILE_SCOPE(GPU_wait);
+        std::unique_lock<std::mutex> lock{synchronization_mutex};
+        synchronization_condition.wait(lock, [this, fence] { return signaled_fence >= fence; });
+    }
 }
 
 } // namespace VideoCommon::GPUThread

src/video_core/gpu_thread.h

@@ -4,10 +4,8 @@
 
 #pragma once
 
-#include <array>
 #include <atomic>
 #include <condition_variable>
-#include <memory>
 #include <mutex>
 #include <optional>
 #include <thread>
@@ -21,9 +19,12 @@ struct FramebufferConfig;
 class DmaPusher;
 } // namespace Tegra
 
-namespace VideoCore {
-class RendererBase;
-} // namespace VideoCore
+namespace Core {
+class System;
+namespace Timing {
+struct EventType;
+} // namespace Timing
+} // namespace Core
 
 namespace VideoCommon::GPUThread {
 
@@ -77,81 +78,68 @@ using CommandData =
 struct CommandDataContainer {
     CommandDataContainer() = default;
 
-    CommandDataContainer(CommandData&& data) : data{std::move(data)} {}
+    CommandDataContainer(CommandData&& data, u64 next_fence)
+        : data{std::move(data)}, fence{next_fence} {}
 
     CommandDataContainer& operator=(const CommandDataContainer& t) {
         data = std::move(t.data);
+        fence = t.fence;
         return *this;
     }
 
     CommandData data;
+    u64 fence{};
 };
 
 /// Struct used to synchronize the GPU thread
 struct SynchState final {
     std::atomic_bool is_running{true};
     std::atomic_int queued_frame_count{};
-    std::mutex frames_mutex;
+    std::mutex synchronization_mutex;
     std::mutex commands_mutex;
     std::condition_variable commands_condition;
-    std::condition_variable frames_condition;
+    std::condition_variable synchronization_condition;
 
-    void IncrementFramesCounter() {
-        std::lock_guard<std::mutex> lock{frames_mutex};
-        ++queued_frame_count;
+    /// Returns true if the gap in GPU commands is small enough that we can consider the CPU and GPU
+    /// synchronized. This is entirely empirical.
+    bool IsSynchronized() const {
+        constexpr std::size_t max_queue_gap{5};
+        return queue.Size() <= max_queue_gap;
     }
 
-    void DecrementFramesCounter() {
-        {
-            std::lock_guard<std::mutex> lock{frames_mutex};
-            --queued_frame_count;
-
-            if (queued_frame_count) {
-                return;
-            }
-        }
-        frames_condition.notify_one();
-    }
-
-    void WaitForFrames() {
-        {
-            std::lock_guard<std::mutex> lock{frames_mutex};
-            if (!queued_frame_count) {
-                return;
-            }
-        }
-
-        // Wait for the GPU to be idle (all commands to be executed)
-        {
-            std::unique_lock<std::mutex> lock{frames_mutex};
-            frames_condition.wait(lock, [this] { return !queued_frame_count; });
+    void TrySynchronize() {
+        if (IsSynchronized()) {
+            std::lock_guard<std::mutex> lock{synchronization_mutex};
+            synchronization_condition.notify_one();
         }
     }
 
+    void WaitForSynchronization(u64 fence);
+
     void SignalCommands() {
-        {
-            std::unique_lock<std::mutex> lock{commands_mutex};
-            if (queue.Empty()) {
-                return;
-            }
+        if (queue.Empty()) {
+            return;
         }
 
         commands_condition.notify_one();
     }
 
     void WaitForCommands() {
-        std::unique_lock<std::mutex> lock{commands_mutex};
+        std::unique_lock lock{commands_mutex};
         commands_condition.wait(lock, [this] { return !queue.Empty(); });
     }
 
     using CommandQueue = Common::SPSCQueue<CommandDataContainer>;
     CommandQueue queue;
+    u64 last_fence{};
+    std::atomic<u64> signaled_fence{};
 };
 
 /// Class used to manage the GPU thread
 class ThreadManager final {
 public:
-    explicit ThreadManager(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_pusher);
+    explicit ThreadManager(Core::System& system, VideoCore::RendererBase& renderer,
+                           Tegra::DmaPusher& dma_pusher);
     ~ThreadManager();
 
     /// Push GPU command entries to be processed
@@ -172,12 +160,12 @@ public:
 
 private:
     /// Pushes a command to be executed by the GPU thread
-    void PushCommand(CommandData&& command_data);
+    u64 PushCommand(CommandData&& command_data);
 
 private:
     SynchState state;
-    VideoCore::RendererBase& renderer;
-    Tegra::DmaPusher& dma_pusher;
+    Core::System& system;
+    Core::Timing::EventType* synchronization_event{};
     std::thread thread;
     std::thread::id thread_id;
 };

src/video_core/rasterizer_cache.h

@@ -71,8 +71,8 @@ private:
     bool is_registered{};      ///< Whether the object is currently registered with the cache
     bool is_dirty{};           ///< Whether the object is dirty (out of sync with guest memory)
     u64 last_modified_ticks{}; ///< When the object was last modified, used for in-order flushing
-    CacheAddr cache_addr{};    ///< Cache address memory, unique from emulated virtual address space
     const u8* host_ptr{};      ///< Pointer to the memory backing this cached region
+    CacheAddr cache_addr{};    ///< Cache address memory, unique from emulated virtual address space
 };
 
 template <class T>
@@ -84,7 +84,7 @@ public:
 
     /// Write any cached resources overlapping the specified region back to memory
     void FlushRegion(CacheAddr addr, std::size_t size) {
-        std::lock_guard<std::recursive_mutex> lock{mutex};
+        std::lock_guard lock{mutex};
 
         const auto& objects{GetSortedObjectsFromRegion(addr, size)};
         for (auto& object : objects) {
@@ -94,7 +94,7 @@ public:
 
     /// Mark the specified region as being invalidated
     void InvalidateRegion(CacheAddr addr, u64 size) {
-        std::lock_guard<std::recursive_mutex> lock{mutex};
+        std::lock_guard lock{mutex};
 
         const auto& objects{GetSortedObjectsFromRegion(addr, size)};
         for (auto& object : objects) {
@@ -108,7 +108,7 @@ public:
 
     /// Invalidates everything in the cache
     void InvalidateAll() {
-        std::lock_guard<std::recursive_mutex> lock{mutex};
+        std::lock_guard lock{mutex};
 
         while (interval_cache.begin() != interval_cache.end()) {
             Unregister(*interval_cache.begin()->second.begin());
@@ -133,7 +133,7 @@ protected:
 
     /// Register an object into the cache
     virtual void Register(const T& object) {
-        std::lock_guard<std::recursive_mutex> lock{mutex};
+        std::lock_guard lock{mutex};
 
         object->SetIsRegistered(true);
         interval_cache.add({GetInterval(object), ObjectSet{object}});
@@ -143,7 +143,7 @@ protected:
 
     /// Unregisters an object from the cache
     virtual void Unregister(const T& object) {
-        std::lock_guard<std::recursive_mutex> lock{mutex};
+        std::lock_guard lock{mutex};
 
         object->SetIsRegistered(false);
         rasterizer.UpdatePagesCachedCount(object->GetCpuAddr(), object->GetSizeInBytes(), -1);
@@ -153,14 +153,14 @@ protected:
 
     /// Returns a ticks counter used for tracking when cached objects were last modified
     u64 GetModifiedTicks() {
-        std::lock_guard<std::recursive_mutex> lock{mutex};
+        std::lock_guard lock{mutex};
 
         return ++modified_ticks;
     }
 
     /// Flushes the specified object, updating appropriate cache state as needed
     void FlushObject(const T& object) {
-        std::lock_guard<std::recursive_mutex> lock{mutex};
+        std::lock_guard lock{mutex};
 
         if (!object->IsDirty()) {
             return;

src/video_core/renderer_opengl/gl_buffer_cache.cpp

@@ -15,8 +15,8 @@ namespace OpenGL {
 
 CachedBufferEntry::CachedBufferEntry(VAddr cpu_addr, std::size_t size, GLintptr offset,
                                      std::size_t alignment, u8* host_ptr)
-    : cpu_addr{cpu_addr}, size{size}, offset{offset}, alignment{alignment}, RasterizerCacheObject{
-                                                                                host_ptr} {}
+    : RasterizerCacheObject{host_ptr}, cpu_addr{cpu_addr}, size{size}, offset{offset},
+      alignment{alignment} {}
 
 OGLBufferCache::OGLBufferCache(RasterizerOpenGL& rasterizer, std::size_t size)
     : RasterizerCache{rasterizer}, stream_buffer(size, true) {}

src/video_core/renderer_opengl/gl_global_cache.cpp

@@ -15,7 +15,7 @@
 namespace OpenGL {
 
 CachedGlobalRegion::CachedGlobalRegion(VAddr cpu_addr, u32 size, u8* host_ptr)
-    : cpu_addr{cpu_addr}, size{size}, RasterizerCacheObject{host_ptr} {
+    : RasterizerCacheObject{host_ptr}, cpu_addr{cpu_addr}, size{size} {
     buffer.Create();
     // Bind and unbind the buffer so it gets allocated by the driver
     glBindBuffer(GL_SHADER_STORAGE_BUFFER, buffer.handle);

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -100,11 +100,9 @@ struct FramebufferCacheKey {
     }
 };
 
-RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, Core::System& system,
-                                   ScreenInfo& info)
-    : res_cache{*this}, shader_cache{*this, system}, global_cache{*this},
-      emu_window{window}, system{system}, screen_info{info},
-      buffer_cache(*this, STREAM_BUFFER_SIZE) {
+RasterizerOpenGL::RasterizerOpenGL(Core::System& system, ScreenInfo& info)
+    : res_cache{*this}, shader_cache{*this, system}, global_cache{*this}, system{system},
+      screen_info{info}, buffer_cache(*this, STREAM_BUFFER_SIZE) {
     // Create sampler objects
     for (std::size_t i = 0; i < texture_samplers.size(); ++i) {
         texture_samplers[i].Create();

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -50,8 +50,7 @@ struct FramebufferCacheKey;
 
 class RasterizerOpenGL : public VideoCore::RasterizerInterface {
 public:
-    explicit RasterizerOpenGL(Core::Frontend::EmuWindow& window, Core::System& system,
-                              ScreenInfo& info);
+    explicit RasterizerOpenGL(Core::System& system, ScreenInfo& info);
     ~RasterizerOpenGL() override;
 
     void DrawArrays() override;
@@ -214,7 +213,6 @@ private:
     ShaderCacheOpenGL shader_cache;
     GlobalRegionCacheOpenGL global_cache;
 
-    Core::Frontend::EmuWindow& emu_window;
     Core::System& system;
 
     ScreenInfo& screen_info;

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -562,8 +562,8 @@ void RasterizerCacheOpenGL::CopySurface(const Surface& src_surface, const Surfac
 }
 
 CachedSurface::CachedSurface(const SurfaceParams& params)
-    : params{params}, gl_target{SurfaceTargetToGL(params.target)},
-      cached_size_in_bytes{params.size_in_bytes}, RasterizerCacheObject{params.host_ptr} {
+    : RasterizerCacheObject{params.host_ptr}, params{params},
+      gl_target{SurfaceTargetToGL(params.target)}, cached_size_in_bytes{params.size_in_bytes} {
 
     const auto optional_cpu_addr{
         Core::System::GetInstance().GPU().MemoryManager().GpuToCpuAddress(params.gpu_addr)};

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -538,12 +538,12 @@ private:
         return nullptr;
     }
 
-    void Register(const Surface& object) {
+    void Register(const Surface& object) override {
         RasterizerCache<Surface>::Register(object);
     }
 
     /// Unregisters an object from the cache
-    void Unregister(const Surface& object) {
+    void Unregister(const Surface& object) override {
         if (object->IsReinterpreted()) {
             auto interval = GetReinterpretInterval(object);
             reinterpreted_surfaces.erase(interval);

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -215,9 +215,9 @@ CachedShader::CachedShader(VAddr cpu_addr, u64 unique_identifier,
                            Maxwell::ShaderProgram program_type, ShaderDiskCacheOpenGL& disk_cache,
                            const PrecompiledPrograms& precompiled_programs,
                            ProgramCode&& program_code, ProgramCode&& program_code_b, u8* host_ptr)
-    : host_ptr{host_ptr}, cpu_addr{cpu_addr}, unique_identifier{unique_identifier},
-      program_type{program_type}, disk_cache{disk_cache},
-      precompiled_programs{precompiled_programs}, RasterizerCacheObject{host_ptr} {
+    : RasterizerCacheObject{host_ptr}, host_ptr{host_ptr}, cpu_addr{cpu_addr},
+      unique_identifier{unique_identifier}, program_type{program_type}, disk_cache{disk_cache},
+      precompiled_programs{precompiled_programs} {
 
     const std::size_t code_size = CalculateProgramSize(program_code);
     const std::size_t code_size_b =
@@ -245,9 +245,9 @@ CachedShader::CachedShader(VAddr cpu_addr, u64 unique_identifier,
                            Maxwell::ShaderProgram program_type, ShaderDiskCacheOpenGL& disk_cache,
                            const PrecompiledPrograms& precompiled_programs,
                            GLShader::ProgramResult result, u8* host_ptr)
-    : cpu_addr{cpu_addr}, unique_identifier{unique_identifier}, program_type{program_type},
-      disk_cache{disk_cache}, precompiled_programs{precompiled_programs}, RasterizerCacheObject{
-                                                                              host_ptr} {
+    : RasterizerCacheObject{host_ptr}, cpu_addr{cpu_addr}, unique_identifier{unique_identifier},
+      program_type{program_type}, disk_cache{disk_cache}, precompiled_programs{
+                                                              precompiled_programs} {
 
     code = std::move(result.first);
     entries = result.second;

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -266,7 +266,7 @@ void RendererOpenGL::CreateRasterizer() {
     }
     // Initialize sRGB Usage
     OpenGLState::ClearsRGBUsed();
-    rasterizer = std::make_unique<RasterizerOpenGL>(render_window, system, screen_info);
+    rasterizer = std::make_unique<RasterizerOpenGL>(system, screen_info);
 }
 
 void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,

src/video_core/renderer_vulkan/vk_buffer_cache.cpp

@@ -19,8 +19,8 @@ namespace Vulkan {
 
 CachedBufferEntry::CachedBufferEntry(VAddr cpu_addr, std::size_t size, u64 offset,
                                      std::size_t alignment, u8* host_ptr)
-    : cpu_addr{cpu_addr}, size{size}, offset{offset}, alignment{alignment}, RasterizerCacheObject{
-                                                                                host_ptr} {}
+    : RasterizerCacheObject{host_ptr}, cpu_addr{cpu_addr}, size{size}, offset{offset},
+      alignment{alignment} {}
 
 VKBufferCache::VKBufferCache(Tegra::MemoryManager& tegra_memory_manager,
                              VideoCore::RasterizerInterface& rasterizer, const VKDevice& device,

src/video_core/renderer_vulkan/vk_resource_manager.cpp

@@ -21,7 +21,7 @@ public:
     CommandBufferPool(const VKDevice& device)
         : VKFencedPool(COMMAND_BUFFER_POOL_SIZE), device{device} {}
 
-    void Allocate(std::size_t begin, std::size_t end) {
+    void Allocate(std::size_t begin, std::size_t end) override {
         const auto dev = device.GetLogical();
         const auto& dld = device.GetDispatchLoader();
         const u32 graphics_family = device.GetGraphicsFamily();

src/video_core/renderer_vulkan/vk_resource_manager.h

@@ -97,7 +97,7 @@ private:
 class VKFenceWatch final : public VKResource {
 public:
     explicit VKFenceWatch();
-    ~VKFenceWatch();
+    ~VKFenceWatch() override;
 
     /// Waits for the fence to be released.
     void Wait();

src/web_service/web_backend.cpp

@@ -24,7 +24,7 @@ constexpr u32 TIMEOUT_SECONDS = 30;
 struct Client::Impl {
     Impl(std::string host, std::string username, std::string token)
         : host{std::move(host)}, username{std::move(username)}, token{std::move(token)} {
-        std::lock_guard<std::mutex> lock(jwt_cache.mutex);
+        std::lock_guard lock{jwt_cache.mutex};
         if (this->username == jwt_cache.username && this->token == jwt_cache.token) {
             jwt = jwt_cache.jwt;
         }
@@ -151,7 +151,7 @@ struct Client::Impl {
         if (result.result_code != Common::WebResult::Code::Success) {
             LOG_ERROR(WebService, "UpdateJWT failed");
         } else {
-            std::lock_guard<std::mutex> lock(jwt_cache.mutex);
+            std::lock_guard lock{jwt_cache.mutex};
             jwt_cache.username = username;
             jwt_cache.token = token;
             jwt_cache.jwt = jwt = result.returned_data;

src/yuzu/applets/profile_select.cpp

@@ -163,6 +163,6 @@ void QtProfileSelector::SelectProfile(
 
 void QtProfileSelector::MainWindowFinishedSelection(std::optional<Service::Account::UUID> uuid) {
     // Acquire the HLE mutex
-    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
+    std::lock_guard lock{HLE::g_hle_lock};
     callback(uuid);
 }

src/yuzu/applets/software_keyboard.cpp

@@ -141,12 +141,12 @@ void QtSoftwareKeyboard::SendTextCheckDialog(std::u16string error_message,
 
 void QtSoftwareKeyboard::MainWindowFinishedText(std::optional<std::u16string> text) {
     // Acquire the HLE mutex
-    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
+    std::lock_guard lock{HLE::g_hle_lock};
     text_output(text);
 }
 
 void QtSoftwareKeyboard::MainWindowFinishedCheckDialog() {
     // Acquire the HLE mutex
-    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
+    std::lock_guard lock{HLE::g_hle_lock};
     finished_check();
 }

src/yuzu/applets/web_browser.cpp

@@ -104,12 +104,12 @@ void QtWebBrowser::OpenPage(std::string_view url, std::function<void()> unpack_r
 
 void QtWebBrowser::MainWindowUnpackRomFS() {
     // Acquire the HLE mutex
-    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
+    std::lock_guard lock{HLE::g_hle_lock};
     unpack_romfs_callback();
 }
 
 void QtWebBrowser::MainWindowFinishedBrowsing() {
     // Acquire the HLE mutex
-    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
+    std::lock_guard lock{HLE::g_hle_lock};
     finished_callback();
 }

src/yuzu/bootmanager.cpp

@@ -67,7 +67,7 @@ void EmuThread::run() {
 
             was_active = false;
         } else {
-            std::unique_lock<std::mutex> lock(running_mutex);
+            std::unique_lock lock{running_mutex};
             running_cv.wait(lock, [this] { return IsRunning() || exec_step || stop_run; });
         }
     }

src/yuzu/bootmanager.h

@@ -53,7 +53,7 @@ public:
      * @note This function is thread-safe
      */
     void SetRunning(bool running) {
-        std::unique_lock<std::mutex> lock(running_mutex);
+        std::unique_lock lock{running_mutex};
         this->running = running;
         lock.unlock();
         running_cv.notify_all();

src/yuzu/configuration/config.cpp

@@ -413,7 +413,6 @@ void Config::ReadValues() {
 
     qt_config->beginGroup("System");
     Settings::values.use_docked_mode = ReadSetting("use_docked_mode", false).toBool();
-    Settings::values.enable_nfc = ReadSetting("enable_nfc", true).toBool();
 
     Settings::values.current_user =
         std::clamp<int>(ReadSetting("current_user", 0).toInt(), 0, Service::Account::MAX_USERS - 1);
@@ -675,7 +674,6 @@ void Config::SaveValues() {
 
     qt_config->beginGroup("System");
     WriteSetting("use_docked_mode", Settings::values.use_docked_mode, false);
-    WriteSetting("enable_nfc", Settings::values.enable_nfc, true);
     WriteSetting("current_user", Settings::values.current_user, 0);
     WriteSetting("language_index", Settings::values.language_index, 1);
 

src/yuzu/configuration/configure_general.cpp

@@ -33,7 +33,6 @@ void ConfigureGeneral::setConfiguration() {
     ui->toggle_user_on_boot->setChecked(UISettings::values.select_user_on_boot);
     ui->theme_combobox->setCurrentIndex(ui->theme_combobox->findData(UISettings::values.theme));
     ui->use_cpu_jit->setChecked(Settings::values.use_cpu_jit);
-    ui->enable_nfc->setChecked(Settings::values.enable_nfc);
 }
 
 void ConfigureGeneral::PopulateHotkeyList(const HotkeyRegistry& registry) {
@@ -48,5 +47,4 @@ void ConfigureGeneral::applyConfiguration() {
         ui->theme_combobox->itemData(ui->theme_combobox->currentIndex()).toString();
 
     Settings::values.use_cpu_jit = ui->use_cpu_jit->isChecked();
-    Settings::values.enable_nfc = ui->enable_nfc->isChecked();
 }

src/yuzu/configuration/configure_general.ui

@@ -70,26 +70,6 @@
        </layout>
       </widget>
      </item>
-     <item>
-      <widget class="QGroupBox" name="EmulationGroupBox">
-       <property name="title">
-        <string>Emulation</string>
-       </property>
-       <layout class="QHBoxLayout" name="EmulationHorizontalLayout">
-        <item>
-         <layout class="QVBoxLayout" name="EmulationVerticalLayout">
-          <item>
-           <widget class="QCheckBox" name="enable_nfc">
-            <property name="text">
-             <string>Enable NFC</string>
-            </property>
-           </widget>
-          </item>
-         </layout>
-        </item>
-       </layout>
-      </widget>
-     </item>
      <item>
       <widget class="QGroupBox" name="theme_group_box">
        <property name="title">

src/yuzu/debugger/wait_tree.cpp

@@ -234,6 +234,9 @@ QString WaitTreeThread::GetText() const {
     case Kernel::ThreadStatus::WaitMutex:
         status = tr("waiting for mutex");
         break;
+    case Kernel::ThreadStatus::WaitCondVar:
+        status = tr("waiting for condition variable");
+        break;
     case Kernel::ThreadStatus::WaitArb:
         status = tr("waiting for address arbiter");
         break;
@@ -269,6 +272,7 @@ QColor WaitTreeThread::GetColor() const {
     case Kernel::ThreadStatus::WaitSynchAll:
     case Kernel::ThreadStatus::WaitSynchAny:
     case Kernel::ThreadStatus::WaitMutex:
+    case Kernel::ThreadStatus::WaitCondVar:
     case Kernel::ThreadStatus::WaitArb:
         return QColor(Qt::GlobalColor::red);
     case Kernel::ThreadStatus::Dormant:

src/yuzu_cmd/config.cpp

@@ -319,7 +319,6 @@ void Config::ReadValues() {
 
     // System
     Settings::values.use_docked_mode = sdl2_config->GetBoolean("System", "use_docked_mode", false);
-    Settings::values.enable_nfc = sdl2_config->GetBoolean("System", "enable_nfc", true);
     const auto size = sdl2_config->GetInteger("System", "users_size", 0);
 
     Settings::values.current_user = std::clamp<int>(