GPU: Implemented the nvmap Free ioctl.

It releases a reference to an nvmap object

src/common/file_util.cpp

@@ -809,16 +809,16 @@ IOFile::~IOFile() {
     Close();
 }
 
-IOFile::IOFile(IOFile&& other) {
+IOFile::IOFile(IOFile&& other) noexcept {
     Swap(other);
 }
 
-IOFile& IOFile::operator=(IOFile&& other) {
+IOFile& IOFile::operator=(IOFile&& other) noexcept {
     Swap(other);
     return *this;
 }
 
-void IOFile::Swap(IOFile& other) {
+void IOFile::Swap(IOFile& other) noexcept {
     std::swap(m_file, other.m_file);
     std::swap(m_good, other.m_good);
 }

src/common/file_util.h

@@ -160,10 +160,10 @@ public:
 
     ~IOFile();
 
-    IOFile(IOFile&& other);
-    IOFile& operator=(IOFile&& other);
+    IOFile(IOFile&& other) noexcept;
+    IOFile& operator=(IOFile&& other) noexcept;
 
-    void Swap(IOFile& other);
+    void Swap(IOFile& other) noexcept;
 
     bool Open(const std::string& filename, const char openmode[]);
     bool Close();
@@ -202,11 +202,15 @@ public:
         return items_written;
     }
 
-    size_t ReadBytes(void* data, size_t length) {
+    template <typename T>
+    size_t ReadBytes(T* data, size_t length) {
+        static_assert(std::is_trivially_copyable<T>(), "T must be trivially copyable");
         return ReadArray(reinterpret_cast<char*>(data), length);
     }
 
-    size_t WriteBytes(const void* data, size_t length) {
+    template <typename T>
+    size_t WriteBytes(const T* data, size_t length) {
+        static_assert(std::is_trivially_copyable<T>(), "T must be trivially copyable");
         return WriteArray(reinterpret_cast<const char*>(data), length);
     }
 

src/common/memory_util.cpp

@@ -167,8 +167,7 @@ std::string MemUsage() {
         return "MemUsage Error";
 
     if (GetProcessMemoryInfo(hProcess, &pmc, sizeof(pmc)))
-        Ret = Common::StringFromFormat(
-            "%s K", Common::ThousandSeparate(pmc.WorkingSetSize / 1024, 7).c_str());
+        Ret = fmt::format("{} K", Common::ThousandSeparate(pmc.WorkingSetSize / 1024, 7));
 
     CloseHandle(hProcess);
     return Ret;

src/common/string_util.cpp

@@ -46,76 +46,6 @@ bool AsciiToHex(const char* _szValue, u32& result) {
     return true;
 }
 
-bool CharArrayFromFormatV(char* out, int outsize, const char* format, va_list args) {
-    int writtenCount;
-
-#ifdef _MSC_VER
-    // You would think *printf are simple, right? Iterate on each character,
-    // if it's a format specifier handle it properly, etc.
-    //
-    // Nooooo. Not according to the C standard.
-    //
-    // According to the C99 standard (7.19.6.1 "The fprintf function")
-    //     The format shall be a multibyte character sequence
-    //
-    // Because some character encodings might have '%' signs in the middle of
-    // a multibyte sequence (SJIS for example only specifies that the first
-    // byte of a 2 byte sequence is "high", the second byte can be anything),
-    // printf functions have to decode the multibyte sequences and try their
-    // best to not screw up.
-    //
-    // Unfortunately, on Windows, the locale for most languages is not UTF-8
-    // as we would need. Notably, for zh_TW, Windows chooses EUC-CN as the
-    // locale, and completely fails when trying to decode UTF-8 as EUC-CN.
-    //
-    // On the other hand, the fix is simple: because we use UTF-8, no such
-    // multibyte handling is required as we can simply assume that no '%' char
-    // will be present in the middle of a multibyte sequence.
-    //
-    // This is why we lookup an ANSI (cp1252) locale here and use _vsnprintf_l.
-    static locale_t c_locale = nullptr;
-    if (!c_locale)
-        c_locale = _create_locale(LC_ALL, ".1252");
-    writtenCount = _vsnprintf_l(out, outsize, format, c_locale, args);
-#else
-    writtenCount = vsnprintf(out, outsize, format, args);
-#endif
-
-    if (writtenCount > 0 && writtenCount < outsize) {
-        out[writtenCount] = '\0';
-        return true;
-    } else {
-        out[outsize - 1] = '\0';
-        return false;
-    }
-}
-
-std::string StringFromFormat(const char* format, ...) {
-    va_list args;
-    char* buf = nullptr;
-#ifdef _WIN32
-    int required = 0;
-
-    va_start(args, format);
-    required = _vscprintf(format, args);
-    buf = new char[required + 1];
-    CharArrayFromFormatV(buf, required + 1, format, args);
-    va_end(args);
-
-    std::string temp = buf;
-    delete[] buf;
-#else
-    va_start(args, format);
-    if (vasprintf(&buf, format, args) < 0)
-        NGLOG_ERROR(Common, "Unable to allocate memory for string");
-    va_end(args);
-
-    std::string temp = buf;
-    free(buf);
-#endif
-    return temp;
-}
-
 // For Debugging. Read out an u8 array.
 std::string ArrayToString(const u8* data, size_t size, int line_len, bool spaces) {
     std::ostringstream oss;

src/common/string_util.h

@@ -4,7 +4,6 @@
 
 #pragma once
 
-#include <cstdarg>
 #include <cstddef>
 #include <iomanip>
 #include <sstream>
@@ -20,19 +19,6 @@ std::string ToLower(std::string str);
 /// Make a string uppercase
 std::string ToUpper(std::string str);
 
-std::string StringFromFormat(const char* format, ...);
-// Cheap!
-bool CharArrayFromFormatV(char* out, int outsize, const char* format, va_list args);
-
-template <size_t Count>
-inline void CharArrayFromFormat(char (&out)[Count], const char* format, ...) {
-    va_list args;
-    va_start(args, format);
-    CharArrayFromFormatV(out, Count, format, args);
-    va_end(args);
-}
-
-// Good
 std::string ArrayToString(const u8* data, size_t size, int line_len = 20, bool spaces = true);
 
 std::string StripSpaces(const std::string& s);

src/common/timer.cpp

@@ -2,7 +2,10 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include <time.h>
+#include <ctime>
+
+#include <fmt/format.h>
+
 #ifdef _WIN32
 #include <windows.h>
 // windows.h needs to be included before other windows headers
@@ -104,8 +107,8 @@ std::string Timer::GetTimeElapsedFormatted() const {
     // Hours
     u32 Hours = Minutes / 60;
 
-    std::string TmpStr = StringFromFormat("%02i:%02i:%02i:%03i", Hours, Minutes % 60, Seconds % 60,
-                                          Milliseconds % 1000);
+    std::string TmpStr = fmt::format("{:02}:{:02}:{:02}:{:03}", Hours, Minutes % 60, Seconds % 60,
+                                     Milliseconds % 1000);
     return TmpStr;
 }
 
@@ -165,11 +168,11 @@ std::string Timer::GetTimeFormatted() {
 #ifdef _WIN32
     struct timeb tp;
     (void)::ftime(&tp);
-    return StringFromFormat("%s:%03i", tmp, tp.millitm);
+    return fmt::format("{}:{:03}", tmp, tp.millitm);
 #else
     struct timeval t;
     (void)gettimeofday(&t, nullptr);
-    return StringFromFormat("%s:%03d", tmp, (int)(t.tv_usec / 1000));
+    return fmt::format("{}:{:03}", tmp, static_cast<int>(t.tv_usec / 1000));
 #endif
 }
 

src/common/vector_math.h

@@ -52,8 +52,8 @@ static inline Vec4<T> MakeVec(const T& x, const T& y, const T& z, const T& w);
 template <typename T>
 class Vec2 {
 public:
-    T x;
-    T y;
+    T x{};
+    T y{};
 
     Vec2() = default;
     Vec2(const T& _x, const T& _y) : x(_x), y(_y) {}
@@ -192,9 +192,9 @@ inline float Vec2<float>::Normalize() {
 template <typename T>
 class Vec3 {
 public:
-    T x;
-    T y;
-    T z;
+    T x{};
+    T y{};
+    T z{};
 
     Vec3() = default;
     Vec3(const T& _x, const T& _y, const T& _z) : x(_x), y(_y), z(_z) {}
@@ -392,10 +392,10 @@ typedef Vec3<float> Vec3f;
 template <typename T>
 class Vec4 {
 public:
-    T x;
-    T y;
-    T z;
-    T w;
+    T x{};
+    T y{};
+    T z{};
+    T w{};
 
     Vec4() = default;
     Vec4(const T& _x, const T& _y, const T& _z, const T& _w) : x(_x), y(_y), z(_z), w(_w) {}

src/core/CMakeLists.txt

@@ -4,6 +4,8 @@ add_library(core STATIC
     arm/unicorn/arm_unicorn.h
     core.cpp
     core.h
+    core_cpu.cpp
+    core_cpu.h
     core_timing.cpp
     core_timing.h
     file_sys/directory.h
@@ -253,6 +255,7 @@ add_library(core STATIC
     loader/nso.h
     memory.cpp
     memory.h
+    memory_hook.cpp
     memory_hook.h
     memory_setup.h
     perf_stats.cpp

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -55,7 +55,7 @@ public:
     }
 
     void InterpreterFallback(u64 pc, size_t num_instructions) override {
-        NGLOG_INFO(Core_ARM, "Unicorn fallback @ {:#X} for {} instructions (instr = {:08X})", pc,
+        NGLOG_INFO(Core_ARM, "Unicorn fallback @ 0x{:X} for {} instructions (instr = {:08X})", pc,
                    num_instructions, MemoryReadCode(pc));
 
         ARM_Interface::ThreadContext ctx;

src/core/arm/unicorn/arm_unicorn.cpp

@@ -52,8 +52,8 @@ static void InterruptHook(uc_engine* uc, u32 intNo, void* user_data) {
 static bool UnmappedMemoryHook(uc_engine* uc, uc_mem_type type, u64 addr, int size, u64 value,
                                void* user_data) {
     ARM_Interface::ThreadContext ctx{};
-    Core::CPU().SaveContext(ctx);
-    ASSERT_MSG(false, "Attempted to read from unmapped memory: {:#X}, pc={:#X}, lr={:#X}", addr,
+    Core::CurrentArmInterface().SaveContext(ctx);
+    ASSERT_MSG(false, "Attempted to read from unmapped memory: 0x{:X}, pc=0x{:X}, lr=0x{:X}", addr,
                ctx.pc, ctx.cpu_registers[30]);
     return {};
 }

src/core/core.cpp

@@ -5,10 +5,6 @@
 #include <memory>
 #include <utility>
 #include "common/logging/log.h"
-#ifdef ARCHITECTURE_x86_64
-#include "core/arm/dynarmic/arm_dynarmic.h"
-#endif
-#include "core/arm/unicorn/arm_unicorn.h"
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/gdbstub/gdbstub.h"
@@ -31,11 +27,31 @@ namespace Core {
 
 System::~System() = default;
 
+/// Runs a CPU core while the system is powered on
+static void RunCpuCore(std::shared_ptr<Cpu> cpu_state) {
+    while (Core::System().GetInstance().IsPoweredOn()) {
+        cpu_state->RunLoop(true);
+    }
+}
+
+Cpu& System::CurrentCpuCore() {
+    // If multicore is enabled, use host thread to figure out the current CPU core
+    if (Settings::values.use_multi_core) {
+        const auto& search = thread_to_cpu.find(std::this_thread::get_id());
+        ASSERT(search != thread_to_cpu.end());
+        ASSERT(search->second);
+        return *search->second;
+    }
+
+    // Otherwise, use single-threaded mode active_core variable
+    return *cpu_cores[active_core];
+}
+
 System::ResultStatus System::RunLoop(bool tight_loop) {
     status = ResultStatus::Success;
-    if (!cpu_core) {
-        return ResultStatus::ErrorNotInitialized;
-    }
+
+    // Update thread_to_cpu in case Core 0 is run from a different host thread
+    thread_to_cpu[std::this_thread::get_id()] = cpu_cores[0];
 
     if (GDBStub::IsServerEnabled()) {
         GDBStub::HandlePacket();
@@ -52,25 +68,14 @@ System::ResultStatus System::RunLoop(bool tight_loop) {
         }
     }
 
-    // If we don't have a currently active thread then don't execute instructions,
-    // instead advance to the next event and try to yield to the next thread
-    if (Kernel::GetCurrentThread() == nullptr) {
-        NGLOG_TRACE(Core_ARM, "Idling");
-        CoreTiming::Idle();
-        CoreTiming::Advance();
-        PrepareReschedule();
-    } else {
-        CoreTiming::Advance();
-        if (tight_loop) {
-            cpu_core->Run();
-        } else {
-            cpu_core->Step();
+    for (active_core = 0; active_core < NUM_CPU_CORES; ++active_core) {
+        cpu_cores[active_core]->RunLoop(tight_loop);
+        if (Settings::values.use_multi_core) {
+            // Cores 1-3 are run on other threads in this mode
+            break;
         }
     }
 
-    HW::Update();
-    Reschedule();
-
     return status;
 }
 
@@ -133,21 +138,26 @@ System::ResultStatus System::Load(EmuWindow* emu_window, const std::string& file
 }
 
 void System::PrepareReschedule() {
-    cpu_core->PrepareReschedule();
-    reschedule_pending = true;
+    CurrentCpuCore().PrepareReschedule();
 }
 
 PerfStats::Results System::GetAndResetPerfStats() {
     return perf_stats.GetAndResetStats(CoreTiming::GetGlobalTimeUs());
 }
 
-void System::Reschedule() {
-    if (!reschedule_pending) {
-        return;
-    }
+const std::shared_ptr<Kernel::Scheduler>& System::Scheduler(size_t core_index) {
+    ASSERT(core_index < NUM_CPU_CORES);
+    return cpu_cores[core_index]->Scheduler();
+}
 
-    reschedule_pending = false;
-    Core::System::GetInstance().Scheduler().Reschedule();
+ARM_Interface& System::ArmInterface(size_t core_index) {
+    ASSERT(core_index < NUM_CPU_CORES);
+    return cpu_cores[core_index]->ArmInterface();
+}
+
+Cpu& System::CpuCore(size_t core_index) {
+    ASSERT(core_index < NUM_CPU_CORES);
+    return *cpu_cores[core_index];
 }
 
 System::ResultStatus System::Init(EmuWindow* emu_window, u32 system_mode) {
@@ -157,26 +167,17 @@ System::ResultStatus System::Init(EmuWindow* emu_window, u32 system_mode) {
 
     current_process = Kernel::Process::Create("main");
 
-    if (Settings::values.use_cpu_jit) {
-#ifdef ARCHITECTURE_x86_64
-        cpu_core = std::make_shared<ARM_Dynarmic>();
-#else
-        cpu_core = std::make_shared<ARM_Unicorn>();
-        NGLOG_WARNING(Core, "CPU JIT requested, but Dynarmic not available");
-#endif
-    } else {
-        cpu_core = std::make_shared<ARM_Unicorn>();
+    cpu_barrier = std::make_shared<CpuBarrier>();
+    for (size_t index = 0; index < cpu_cores.size(); ++index) {
+        cpu_cores[index] = std::make_shared<Cpu>(cpu_barrier, index);
     }
 
     gpu_core = std::make_unique<Tegra::GPU>();
-
     telemetry_session = std::make_unique<Core::TelemetrySession>();
-
     service_manager = std::make_shared<Service::SM::ServiceManager>();
 
     HW::Init();
     Kernel::Init(system_mode);
-    scheduler = std::make_unique<Kernel::Scheduler>(cpu_core.get());
     Service::Init(service_manager);
     GDBStub::Init();
 
@@ -184,6 +185,17 @@ System::ResultStatus System::Init(EmuWindow* emu_window, u32 system_mode) {
         return ResultStatus::ErrorVideoCore;
     }
 
+    // Create threads for CPU cores 1-3, and build thread_to_cpu map
+    // CPU core 0 is run on the main thread
+    thread_to_cpu[std::this_thread::get_id()] = cpu_cores[0];
+    if (Settings::values.use_multi_core) {
+        for (size_t index = 0; index < cpu_core_threads.size(); ++index) {
+            cpu_core_threads[index] =
+                std::make_unique<std::thread>(RunCpuCore, cpu_cores[index + 1]);
+            thread_to_cpu[cpu_core_threads[index]->get_id()] = cpu_cores[index + 1];
+        }
+    }
+
     NGLOG_DEBUG(Core, "Initialized OK");
 
     // Reset counters and set time origin to current frame
@@ -207,15 +219,30 @@ void System::Shutdown() {
     VideoCore::Shutdown();
     GDBStub::Shutdown();
     Service::Shutdown();
-    scheduler.reset();
     Kernel::Shutdown();
     HW::Shutdown();
     service_manager.reset();
     telemetry_session.reset();
     gpu_core.reset();
-    cpu_core.reset();
+
+    // Close all CPU/threading state
+    cpu_barrier->NotifyEnd();
+    if (Settings::values.use_multi_core) {
+        for (auto& thread : cpu_core_threads) {
+            thread->join();
+            thread.reset();
+        }
+    }
+    thread_to_cpu.clear();
+    for (auto& cpu_core : cpu_cores) {
+        cpu_core.reset();
+    }
+    cpu_barrier.reset();
+
+    // Close core timing
     CoreTiming::Shutdown();
 
+    // Close app loader
     app_loader.reset();
 
     NGLOG_DEBUG(Core, "Shutdown OK");

src/core/core.h

@@ -4,9 +4,12 @@
 
 #pragma once
 
+#include <array>
 #include <memory>
 #include <string>
+#include <thread>
 #include "common/common_types.h"
+#include "core/core_cpu.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/scheduler.h"
 #include "core/loader/loader.h"
@@ -89,7 +92,7 @@ public:
      * @returns True if the emulated system is powered on, otherwise false.
      */
     bool IsPoweredOn() const {
-        return cpu_core != nullptr;
+        return cpu_barrier && cpu_barrier->IsAlive();
     }
 
     /**
@@ -103,24 +106,34 @@ public:
     /// Prepare the core emulation for a reschedule
     void PrepareReschedule();
 
+    /// Gets and resets core performance statistics
     PerfStats::Results GetAndResetPerfStats();
 
-    /**
-     * Gets a reference to the emulated CPU.
-     * @returns A reference to the emulated CPU.
-     */
-    ARM_Interface& CPU() {
-        return *cpu_core;
+    /// Gets an ARM interface to the CPU core that is currently running
+    ARM_Interface& CurrentArmInterface() {
+        return CurrentCpuCore().ArmInterface();
     }
 
+    /// Gets an ARM interface to the CPU core with the specified index
+    ARM_Interface& ArmInterface(size_t core_index);
+
+    /// Gets a CPU interface to the CPU core with the specified index
+    Cpu& CpuCore(size_t core_index);
+
+    /// Gets the GPU interface
     Tegra::GPU& GPU() {
         return *gpu_core;
     }
 
-    Kernel::Scheduler& Scheduler() {
-        return *scheduler;
+    /// Gets the scheduler for the CPU core that is currently running
+    Kernel::Scheduler& CurrentScheduler() {
+        return *CurrentCpuCore().Scheduler();
     }
 
+    /// Gets the scheduler for the CPU core with the specified index
+    const std::shared_ptr<Kernel::Scheduler>& Scheduler(size_t core_index);
+
+    /// Gets the current process
     Kernel::SharedPtr<Kernel::Process>& CurrentProcess() {
         return current_process;
     }
@@ -155,6 +168,9 @@ public:
     }
 
 private:
+    /// Returns the currently running CPU core
+    Cpu& CurrentCpuCore();
+
     /**
      * Initialize the emulated system.
      * @param emu_window Pointer to the host-system window used for video output and keyboard input.
@@ -163,22 +179,15 @@ private:
      */
     ResultStatus Init(EmuWindow* emu_window, u32 system_mode);
 
-    /// Reschedule the core emulation
-    void Reschedule();
-
     /// AppLoader used to load the current executing application
     std::unique_ptr<Loader::AppLoader> app_loader;
-
-    std::shared_ptr<ARM_Interface> cpu_core;
-    std::unique_ptr<Kernel::Scheduler> scheduler;
     std::unique_ptr<Tegra::GPU> gpu_core;
-
     std::shared_ptr<Tegra::DebugContext> debug_context;
-
     Kernel::SharedPtr<Kernel::Process> current_process;
-
-    /// When true, signals that a reschedule should happen
-    bool reschedule_pending{};
+    std::shared_ptr<CpuBarrier> cpu_barrier;
+    std::array<std::shared_ptr<Cpu>, NUM_CPU_CORES> cpu_cores;
+    std::array<std::unique_ptr<std::thread>, NUM_CPU_CORES - 1> cpu_core_threads;
+    size_t active_core{}; ///< Active core, only used in single thread mode
 
     /// Service manager
     std::shared_ptr<Service::SM::ServiceManager> service_manager;
@@ -190,10 +199,13 @@ private:
 
     ResultStatus status = ResultStatus::Success;
     std::string status_details = "";
+
+    /// Map of guest threads to CPU cores
+    std::map<std::thread::id, std::shared_ptr<Cpu>> thread_to_cpu;
 };
 
-inline ARM_Interface& CPU() {
-    return System::GetInstance().CPU();
+inline ARM_Interface& CurrentArmInterface() {
+    return System::GetInstance().CurrentArmInterface();
 }
 
 inline TelemetrySession& Telemetry() {

src/core/core_cpu.cpp

@@ -0,0 +1,119 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <condition_variable>
+#include <mutex>
+
+#include "common/logging/log.h"
+#ifdef ARCHITECTURE_x86_64
+#include "core/arm/dynarmic/arm_dynarmic.h"
+#endif
+#include "core/arm/unicorn/arm_unicorn.h"
+#include "core/core_cpu.h"
+#include "core/core_timing.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/scheduler.h"
+#include "core/hle/kernel/thread.h"
+#include "core/settings.h"
+
+namespace Core {
+
+void CpuBarrier::NotifyEnd() {
+    std::unique_lock<std::mutex> lock(mutex);
+    end = true;
+    condition.notify_all();
+}
+
+bool CpuBarrier::Rendezvous() {
+    if (!Settings::values.use_multi_core) {
+        // Meaningless when running in single-core mode
+        return true;
+    }
+
+    if (!end) {
+        std::unique_lock<std::mutex> lock(mutex);
+
+        --cores_waiting;
+        if (!cores_waiting) {
+            cores_waiting = NUM_CPU_CORES;
+            condition.notify_all();
+            return true;
+        }
+
+        condition.wait(lock);
+        return true;
+    }
+
+    return false;
+}
+
+Cpu::Cpu(std::shared_ptr<CpuBarrier> cpu_barrier, size_t core_index)
+    : cpu_barrier{std::move(cpu_barrier)}, core_index{core_index} {
+
+    if (Settings::values.use_cpu_jit) {
+#ifdef ARCHITECTURE_x86_64
+        arm_interface = std::make_shared<ARM_Dynarmic>();
+#else
+        cpu_core = std::make_shared<ARM_Unicorn>();
+        NGLOG_WARNING(Core, "CPU JIT requested, but Dynarmic not available");
+#endif
+    } else {
+        arm_interface = std::make_shared<ARM_Unicorn>();
+    }
+
+    scheduler = std::make_shared<Kernel::Scheduler>(arm_interface.get());
+}
+
+void Cpu::RunLoop(bool tight_loop) {
+    // Wait for all other CPU cores to complete the previous slice, such that they run in lock-step
+    if (!cpu_barrier->Rendezvous()) {
+        // If rendezvous failed, session has been killed
+        return;
+    }
+
+    // If we don't have a currently active thread then don't execute instructions,
+    // instead advance to the next event and try to yield to the next thread
+    if (Kernel::GetCurrentThread() == nullptr) {
+        NGLOG_TRACE(Core, "Core-{} idling", core_index);
+
+        if (IsMainCore()) {
+            CoreTiming::Idle();
+            CoreTiming::Advance();
+        }
+
+        PrepareReschedule();
+    } else {
+        if (IsMainCore()) {
+            CoreTiming::Advance();
+        }
+
+        if (tight_loop) {
+            arm_interface->Run();
+        } else {
+            arm_interface->Step();
+        }
+    }
+
+    Reschedule();
+}
+
+void Cpu::SingleStep() {
+    return RunLoop(false);
+}
+
+void Cpu::PrepareReschedule() {
+    arm_interface->PrepareReschedule();
+    reschedule_pending = true;
+}
+
+void Cpu::Reschedule() {
+    if (!reschedule_pending) {
+        return;
+    }
+
+    reschedule_pending = false;
+    scheduler->Reschedule();
+}
+
+} // namespace Core

src/core/core_cpu.h

@@ -0,0 +1,78 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <atomic>
+#include <condition_variable>
+#include <memory>
+#include <mutex>
+#include <string>
+#include "common/common_types.h"
+
+class ARM_Interface;
+
+namespace Kernel {
+class Scheduler;
+}
+
+namespace Core {
+
+constexpr unsigned NUM_CPU_CORES{4};
+
+class CpuBarrier {
+public:
+    bool IsAlive() const {
+        return !end;
+    }
+
+    void NotifyEnd();
+
+    bool Rendezvous();
+
+private:
+    unsigned cores_waiting{NUM_CPU_CORES};
+    std::mutex mutex;
+    std::condition_variable condition;
+    std::atomic<bool> end{};
+};
+
+class Cpu {
+public:
+    Cpu(std::shared_ptr<CpuBarrier> cpu_barrier, size_t core_index);
+
+    void RunLoop(bool tight_loop = true);
+
+    void SingleStep();
+
+    void PrepareReschedule();
+
+    ARM_Interface& ArmInterface() {
+        return *arm_interface;
+    }
+
+    const ARM_Interface& ArmInterface() const {
+        return *arm_interface;
+    }
+
+    const std::shared_ptr<Kernel::Scheduler>& Scheduler() const {
+        return scheduler;
+    }
+
+    bool IsMainCore() const {
+        return core_index == 0;
+    }
+
+private:
+    void Reschedule();
+
+    std::shared_ptr<ARM_Interface> arm_interface;
+    std::shared_ptr<CpuBarrier> cpu_barrier;
+    std::shared_ptr<Kernel::Scheduler> scheduler;
+
+    bool reschedule_pending{};
+    size_t core_index;
+};
+
+} // namespace Core

src/core/core_timing.cpp

@@ -6,6 +6,7 @@
 
 #include <algorithm>
 #include <cinttypes>
+#include <limits>
 #include <mutex>
 #include <string>
 #include <tuple>
@@ -57,7 +58,8 @@ static u64 event_fifo_id;
 // to the event_queue by the emu thread
 static Common::MPSCQueue<Event, false> ts_queue;
 
-static constexpr int MAX_SLICE_LENGTH = 20000;
+constexpr int MAX_SLICE_LENGTH = 20000;
+constexpr u64 MAX_VALUE_TO_MULTIPLY = std::numeric_limits<s64>::max() / BASE_CLOCK_RATE;
 
 static s64 idled_cycles;
 
@@ -70,6 +72,54 @@ static EventType* ev_lost = nullptr;
 
 static void EmptyTimedCallback(u64 userdata, s64 cyclesLate) {}
 
+s64 usToCycles(s64 us) {
+    if (us / 1000000 > MAX_VALUE_TO_MULTIPLY) {
+        NGLOG_ERROR(Core_Timing, "Integer overflow, use max value");
+        return std::numeric_limits<s64>::max();
+    }
+    if (us > MAX_VALUE_TO_MULTIPLY) {
+        NGLOG_DEBUG(Core_Timing, "Time very big, do rounding");
+        return BASE_CLOCK_RATE * (us / 1000000);
+    }
+    return (BASE_CLOCK_RATE * us) / 1000000;
+}
+
+s64 usToCycles(u64 us) {
+    if (us / 1000000 > MAX_VALUE_TO_MULTIPLY) {
+        NGLOG_ERROR(Core_Timing, "Integer overflow, use max value");
+        return std::numeric_limits<s64>::max();
+    }
+    if (us > MAX_VALUE_TO_MULTIPLY) {
+        NGLOG_DEBUG(Core_Timing, "Time very big, do rounding");
+        return BASE_CLOCK_RATE * static_cast<s64>(us / 1000000);
+    }
+    return (BASE_CLOCK_RATE * static_cast<s64>(us)) / 1000000;
+}
+
+s64 nsToCycles(s64 ns) {
+    if (ns / 1000000000 > MAX_VALUE_TO_MULTIPLY) {
+        NGLOG_ERROR(Core_Timing, "Integer overflow, use max value");
+        return std::numeric_limits<s64>::max();
+    }
+    if (ns > MAX_VALUE_TO_MULTIPLY) {
+        NGLOG_DEBUG(Core_Timing, "Time very big, do rounding");
+        return BASE_CLOCK_RATE * (ns / 1000000000);
+    }
+    return (BASE_CLOCK_RATE * ns) / 1000000000;
+}
+
+s64 nsToCycles(u64 ns) {
+    if (ns / 1000000000 > MAX_VALUE_TO_MULTIPLY) {
+        NGLOG_ERROR(Core_Timing, "Integer overflow, use max value");
+        return std::numeric_limits<s64>::max();
+    }
+    if (ns > MAX_VALUE_TO_MULTIPLY) {
+        NGLOG_DEBUG(Core_Timing, "Time very big, do rounding");
+        return BASE_CLOCK_RATE * (static_cast<s64>(ns) / 1000000000);
+    }
+    return (BASE_CLOCK_RATE * static_cast<s64>(ns)) / 1000000000;
+}
+
 EventType* RegisterEvent(const std::string& name, TimedCallback callback) {
     // check for existing type with same name.
     // we want event type names to remain unique so that we can use them for serialization.

src/core/core_timing.h

@@ -18,15 +18,14 @@
  */
 
 #include <functional>
-#include <limits>
 #include <string>
 #include "common/common_types.h"
-#include "common/logging/log.h"
+
+namespace CoreTiming {
 
 // The below clock rate is based on Switch's clockspeed being widely known as 1.020GHz
 // The exact value used is of course unverified.
 constexpr u64 BASE_CLOCK_RATE = 1019215872; // Switch clock speed is 1020MHz un/docked
-constexpr u64 MAX_VALUE_TO_MULTIPLY = std::numeric_limits<s64>::max() / BASE_CLOCK_RATE;
 
 inline s64 msToCycles(int ms) {
     // since ms is int there is no way to overflow
@@ -49,29 +48,9 @@ inline s64 usToCycles(int us) {
     return (BASE_CLOCK_RATE * static_cast<s64>(us) / 1000000);
 }
 
-inline s64 usToCycles(s64 us) {
-    if (us / 1000000 > MAX_VALUE_TO_MULTIPLY) {
-        NGLOG_ERROR(Core_Timing, "Integer overflow, use max value");
-        return std::numeric_limits<s64>::max();
-    }
-    if (us > MAX_VALUE_TO_MULTIPLY) {
-        NGLOG_DEBUG(Core_Timing, "Time very big, do rounding");
-        return BASE_CLOCK_RATE * (us / 1000000);
-    }
-    return (BASE_CLOCK_RATE * us) / 1000000;
-}
+s64 usToCycles(s64 us);
 
-inline s64 usToCycles(u64 us) {
-    if (us / 1000000 > MAX_VALUE_TO_MULTIPLY) {
-        NGLOG_ERROR(Core_Timing, "Integer overflow, use max value");
-        return std::numeric_limits<s64>::max();
-    }
-    if (us > MAX_VALUE_TO_MULTIPLY) {
-        NGLOG_DEBUG(Core_Timing, "Time very big, do rounding");
-        return BASE_CLOCK_RATE * static_cast<s64>(us / 1000000);
-    }
-    return (BASE_CLOCK_RATE * static_cast<s64>(us)) / 1000000;
-}
+s64 usToCycles(u64 us);
 
 inline s64 nsToCycles(float ns) {
     return static_cast<s64>(BASE_CLOCK_RATE * (0.000000001f) * ns);
@@ -81,29 +60,9 @@ inline s64 nsToCycles(int ns) {
     return BASE_CLOCK_RATE * static_cast<s64>(ns) / 1000000000;
 }
 
-inline s64 nsToCycles(s64 ns) {
-    if (ns / 1000000000 > MAX_VALUE_TO_MULTIPLY) {
-        NGLOG_ERROR(Core_Timing, "Integer overflow, use max value");
-        return std::numeric_limits<s64>::max();
-    }
-    if (ns > MAX_VALUE_TO_MULTIPLY) {
-        NGLOG_DEBUG(Core_Timing, "Time very big, do rounding");
-        return BASE_CLOCK_RATE * (ns / 1000000000);
-    }
-    return (BASE_CLOCK_RATE * ns) / 1000000000;
-}
+s64 nsToCycles(s64 ns);
 
-inline s64 nsToCycles(u64 ns) {
-    if (ns / 1000000000 > MAX_VALUE_TO_MULTIPLY) {
-        NGLOG_ERROR(Core_Timing, "Integer overflow, use max value");
-        return std::numeric_limits<s64>::max();
-    }
-    if (ns > MAX_VALUE_TO_MULTIPLY) {
-        NGLOG_DEBUG(Core_Timing, "Time very big, do rounding");
-        return BASE_CLOCK_RATE * (static_cast<s64>(ns) / 1000000000);
-    }
-    return (BASE_CLOCK_RATE * static_cast<s64>(ns)) / 1000000000;
-}
+s64 nsToCycles(u64 ns);
 
 inline u64 cyclesToNs(s64 cycles) {
     return cycles * 1000000000 / BASE_CLOCK_RATE;
@@ -117,8 +76,6 @@ inline u64 cyclesToMs(s64 cycles) {
     return cycles * 1000 / BASE_CLOCK_RATE;
 }
 
-namespace CoreTiming {
-
 /**
  * CoreTiming begins at the boundary of timing slice -1. An initial call to Advance() is
  * required to end slice -1 and start slice 0 before the first cycle of code is executed.

src/core/file_sys/program_metadata.cpp

@@ -77,11 +77,11 @@ u64 ProgramMetadata::GetFilesystemPermissions() const {
 
 void ProgramMetadata::Print() const {
     NGLOG_DEBUG(Service_FS, "Magic:                  {:.4}", npdm_header.magic.data());
-    NGLOG_DEBUG(Service_FS, "Main thread priority:   {:#04X}", npdm_header.main_thread_priority);
+    NGLOG_DEBUG(Service_FS, "Main thread priority:   0x{:02X}", npdm_header.main_thread_priority);
     NGLOG_DEBUG(Service_FS, "Main thread core:       {}", npdm_header.main_thread_cpu);
-    NGLOG_DEBUG(Service_FS, "Main thread stack size: {:#X} bytes", npdm_header.main_stack_size);
+    NGLOG_DEBUG(Service_FS, "Main thread stack size: 0x{:X} bytes", npdm_header.main_stack_size);
     NGLOG_DEBUG(Service_FS, "Process category:       {}", npdm_header.process_category);
-    NGLOG_DEBUG(Service_FS, "Flags:                  {:02X}", npdm_header.flags);
+    NGLOG_DEBUG(Service_FS, "Flags:                  0x{:02X}", npdm_header.flags);
     NGLOG_DEBUG(Service_FS, " > 64-bit instructions: {}",
                 npdm_header.has_64_bit_instructions ? "YES" : "NO");
 
@@ -99,15 +99,15 @@ void ProgramMetadata::Print() const {
 
     // Begin ACID printing (potential perms, signed)
     NGLOG_DEBUG(Service_FS, "Magic:                  {:.4}", acid_header.magic.data());
-    NGLOG_DEBUG(Service_FS, "Flags:                  {:02X}", acid_header.flags);
+    NGLOG_DEBUG(Service_FS, "Flags:                  0x{:02X}", acid_header.flags);
     NGLOG_DEBUG(Service_FS, " > Is Retail:           {}", acid_header.is_retail ? "YES" : "NO");
-    NGLOG_DEBUG(Service_FS, "Title ID Min:           {:016X}", acid_header.title_id_min);
-    NGLOG_DEBUG(Service_FS, "Title ID Max:           {:016X}", acid_header.title_id_max);
-    NGLOG_DEBUG(Service_FS, "Filesystem Access:      {:016X}\n", acid_file_access.permissions);
+    NGLOG_DEBUG(Service_FS, "Title ID Min:           0x{:016X}", acid_header.title_id_min);
+    NGLOG_DEBUG(Service_FS, "Title ID Max:           0x{:016X}", acid_header.title_id_max);
+    NGLOG_DEBUG(Service_FS, "Filesystem Access:      0x{:016X}\n", acid_file_access.permissions);
 
     // Begin ACI0 printing (actual perms, unsigned)
     NGLOG_DEBUG(Service_FS, "Magic:                  {:.4}", aci_header.magic.data());
-    NGLOG_DEBUG(Service_FS, "Title ID:               {:016X}", aci_header.title_id);
-    NGLOG_DEBUG(Service_FS, "Filesystem Access:      {:016X}\n", aci_file_access.permissions);
+    NGLOG_DEBUG(Service_FS, "Title ID:               0x{:016X}", aci_header.title_id);
+    NGLOG_DEBUG(Service_FS, "Filesystem Access:      0x{:016X}\n", aci_file_access.permissions);
 }
 } // namespace FileSys

src/core/gdbstub/gdbstub.cpp

@@ -598,11 +598,11 @@ static void ReadRegister() {
     }
 
     if (id <= SP_REGISTER) {
-        LongToGdbHex(reply, Core::CPU().GetReg(static_cast<int>(id)));
+        LongToGdbHex(reply, Core::CurrentArmInterface().GetReg(static_cast<int>(id)));
     } else if (id == PC_REGISTER) {
-        LongToGdbHex(reply, Core::CPU().GetPC());
+        LongToGdbHex(reply, Core::CurrentArmInterface().GetPC());
     } else if (id == CPSR_REGISTER) {
-        IntToGdbHex(reply, Core::CPU().GetCPSR());
+        IntToGdbHex(reply, Core::CurrentArmInterface().GetCPSR());
     } else {
         return SendReply("E01");
     }
@@ -618,16 +618,16 @@ static void ReadRegisters() {
     u8* bufptr = buffer;
 
     for (int reg = 0; reg <= SP_REGISTER; reg++) {
-        LongToGdbHex(bufptr + reg * 16, Core::CPU().GetReg(reg));
+        LongToGdbHex(bufptr + reg * 16, Core::CurrentArmInterface().GetReg(reg));
     }
 
     bufptr += (32 * 16);
 
-    LongToGdbHex(bufptr, Core::CPU().GetPC());
+    LongToGdbHex(bufptr, Core::CurrentArmInterface().GetPC());
 
     bufptr += 16;
 
-    IntToGdbHex(bufptr, Core::CPU().GetCPSR());
+    IntToGdbHex(bufptr, Core::CurrentArmInterface().GetCPSR());
 
     bufptr += 8;
 
@@ -646,11 +646,11 @@ static void WriteRegister() {
     }
 
     if (id <= SP_REGISTER) {
-        Core::CPU().SetReg(id, GdbHexToLong(buffer_ptr));
+        Core::CurrentArmInterface().SetReg(id, GdbHexToLong(buffer_ptr));
     } else if (id == PC_REGISTER) {
-        Core::CPU().SetPC(GdbHexToLong(buffer_ptr));
+        Core::CurrentArmInterface().SetPC(GdbHexToLong(buffer_ptr));
     } else if (id == CPSR_REGISTER) {
-        Core::CPU().SetCPSR(GdbHexToInt(buffer_ptr));
+        Core::CurrentArmInterface().SetCPSR(GdbHexToInt(buffer_ptr));
     } else {
         return SendReply("E01");
     }
@@ -667,11 +667,11 @@ static void WriteRegisters() {
 
     for (int i = 0, reg = 0; reg <= CPSR_REGISTER; i++, reg++) {
         if (reg <= SP_REGISTER) {
-            Core::CPU().SetReg(reg, GdbHexToLong(buffer_ptr + i * 16));
+            Core::CurrentArmInterface().SetReg(reg, GdbHexToLong(buffer_ptr + i * 16));
         } else if (reg == PC_REGISTER) {
-            Core::CPU().SetPC(GdbHexToLong(buffer_ptr + i * 16));
+            Core::CurrentArmInterface().SetPC(GdbHexToLong(buffer_ptr + i * 16));
         } else if (reg == CPSR_REGISTER) {
-            Core::CPU().SetCPSR(GdbHexToInt(buffer_ptr + i * 16));
+            Core::CurrentArmInterface().SetCPSR(GdbHexToInt(buffer_ptr + i * 16));
         } else {
             UNIMPLEMENTED();
         }

src/core/hle/ipc.h

@@ -167,6 +167,7 @@ struct DomainMessageHeader {
         struct {
             union {
                 BitField<0, 8, CommandType> command;
+                BitField<8, 8, u32_le> input_object_count;
                 BitField<16, 16, u32_le> size;
             };
             u32_le object_id;

src/core/hle/ipc_helpers.h

@@ -298,6 +298,13 @@ public:
 
     template <typename T>
     Kernel::SharedPtr<T> GetCopyObject(size_t index);
+
+    template <class T>
+    std::shared_ptr<T> PopIpcInterface() {
+        ASSERT(context->Session()->IsDomain());
+        ASSERT(context->GetDomainMessageHeader()->input_object_count > 0);
+        return context->GetDomainRequestHandler<T>(Pop<u32>() - 1);
+    }
 };
 
 /// Pop ///

src/core/hle/kernel/hle_ipc.cpp

@@ -251,24 +251,26 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(Thread& thread) {
     return RESULT_SUCCESS;
 }
 
-std::vector<u8> HLERequestContext::ReadBuffer() const {
+std::vector<u8> HLERequestContext::ReadBuffer(int buffer_index) const {
     std::vector<u8> buffer;
-    const bool is_buffer_a{BufferDescriptorA().size() && BufferDescriptorA()[0].Size()};
+    const bool is_buffer_a{BufferDescriptorA().size() && BufferDescriptorA()[buffer_index].Size()};
 
     if (is_buffer_a) {
-        buffer.resize(BufferDescriptorA()[0].Size());
-        Memory::ReadBlock(BufferDescriptorA()[0].Address(), buffer.data(), buffer.size());
+        buffer.resize(BufferDescriptorA()[buffer_index].Size());
+        Memory::ReadBlock(BufferDescriptorA()[buffer_index].Address(), buffer.data(),
+                          buffer.size());
     } else {
-        buffer.resize(BufferDescriptorX()[0].Size());
-        Memory::ReadBlock(BufferDescriptorX()[0].Address(), buffer.data(), buffer.size());
+        buffer.resize(BufferDescriptorX()[buffer_index].Size());
+        Memory::ReadBlock(BufferDescriptorX()[buffer_index].Address(), buffer.data(),
+                          buffer.size());
     }
 
     return buffer;
 }
 
-size_t HLERequestContext::WriteBuffer(const void* buffer, size_t size) const {
-    const bool is_buffer_b{BufferDescriptorB().size() && BufferDescriptorB()[0].Size()};
-    const size_t buffer_size{GetWriteBufferSize()};
+size_t HLERequestContext::WriteBuffer(const void* buffer, size_t size, int buffer_index) const {
+    const bool is_buffer_b{BufferDescriptorB().size() && BufferDescriptorB()[buffer_index].Size()};
+    const size_t buffer_size{GetWriteBufferSize(buffer_index)};
     if (size > buffer_size) {
         NGLOG_CRITICAL(Core, "size ({:016X}) is greater than buffer_size ({:016X})", size,
                        buffer_size);
@@ -276,26 +278,28 @@ size_t HLERequestContext::WriteBuffer(const void* buffer, size_t size) const {
     }
 
     if (is_buffer_b) {
-        Memory::WriteBlock(BufferDescriptorB()[0].Address(), buffer, size);
+        Memory::WriteBlock(BufferDescriptorB()[buffer_index].Address(), buffer, size);
     } else {
-        Memory::WriteBlock(BufferDescriptorC()[0].Address(), buffer, size);
+        Memory::WriteBlock(BufferDescriptorC()[buffer_index].Address(), buffer, size);
     }
 
     return size;
 }
 
-size_t HLERequestContext::WriteBuffer(const std::vector<u8>& buffer) const {
+size_t HLERequestContext::WriteBuffer(const std::vector<u8>& buffer, int buffer_index) const {
     return WriteBuffer(buffer.data(), buffer.size());
 }
 
-size_t HLERequestContext::GetReadBufferSize() const {
-    const bool is_buffer_a{BufferDescriptorA().size() && BufferDescriptorA()[0].Size()};
-    return is_buffer_a ? BufferDescriptorA()[0].Size() : BufferDescriptorX()[0].Size();
+size_t HLERequestContext::GetReadBufferSize(int buffer_index) const {
+    const bool is_buffer_a{BufferDescriptorA().size() && BufferDescriptorA()[buffer_index].Size()};
+    return is_buffer_a ? BufferDescriptorA()[buffer_index].Size()
+                       : BufferDescriptorX()[buffer_index].Size();
 }
 
-size_t HLERequestContext::GetWriteBufferSize() const {
-    const bool is_buffer_b{BufferDescriptorB().size() && BufferDescriptorB()[0].Size()};
-    return is_buffer_b ? BufferDescriptorB()[0].Size() : BufferDescriptorC()[0].Size();
+size_t HLERequestContext::GetWriteBufferSize(int buffer_index) const {
+    const bool is_buffer_b{BufferDescriptorB().size() && BufferDescriptorB()[buffer_index].Size()};
+    return is_buffer_b ? BufferDescriptorB()[buffer_index].Size()
+                       : BufferDescriptorC()[buffer_index].Size();
 }
 
 std::string HLERequestContext::Description() const {

src/core/hle/kernel/hle_ipc.h

@@ -164,19 +164,19 @@ public:
     }
 
     /// Helper function to read a buffer using the appropriate buffer descriptor
-    std::vector<u8> ReadBuffer() const;
+    std::vector<u8> ReadBuffer(int buffer_index = 0) const;
 
     /// Helper function to write a buffer using the appropriate buffer descriptor
-    size_t WriteBuffer(const void* buffer, size_t size) const;
+    size_t WriteBuffer(const void* buffer, size_t size, int buffer_index = 0) const;
 
     /// Helper function to write a buffer using the appropriate buffer descriptor
-    size_t WriteBuffer(const std::vector<u8>& buffer) const;
+    size_t WriteBuffer(const std::vector<u8>& buffer, int buffer_index = 0) const;
 
     /// Helper function to get the size of the input buffer
-    size_t GetReadBufferSize() const;
+    size_t GetReadBufferSize(int buffer_index = 0) const;
 
     /// Helper function to get the size of the output buffer
-    size_t GetWriteBufferSize() const;
+    size_t GetWriteBufferSize(int buffer_index = 0) const;
 
     template <typename T>
     SharedPtr<T> GetCopyObject(size_t index) {
@@ -202,6 +202,16 @@ public:
         domain_objects.emplace_back(std::move(object));
     }
 
+    template <typename T>
+    std::shared_ptr<T> GetDomainRequestHandler(size_t index) const {
+        return std::static_pointer_cast<T>(domain_request_handlers[index]);
+    }
+
+    void SetDomainRequestHandlers(
+        const std::vector<std::shared_ptr<SessionRequestHandler>>& handlers) {
+        domain_request_handlers = handlers;
+    }
+
     /// Clears the list of objects so that no lingering objects are written accidentally to the
     /// response buffer.
     void ClearIncomingObjects() {
@@ -245,6 +255,8 @@ private:
     unsigned data_payload_offset{};
     unsigned buffer_c_offset{};
     u32_le command{};
+
+    std::vector<std::shared_ptr<SessionRequestHandler>> domain_request_handlers;
 };
 
 } // namespace Kernel

src/core/hle/kernel/object_address_table.cpp

@@ -10,12 +10,12 @@ namespace Kernel {
 ObjectAddressTable g_object_address_table;
 
 void ObjectAddressTable::Insert(VAddr addr, SharedPtr<Object> obj) {
-    ASSERT_MSG(objects.find(addr) == objects.end(), "Object already exists with addr={:#X}", addr);
+    ASSERT_MSG(objects.find(addr) == objects.end(), "Object already exists with addr=0x{:X}", addr);
     objects[addr] = obj;
 }
 
 void ObjectAddressTable::Close(VAddr addr) {
-    ASSERT_MSG(objects.find(addr) != objects.end(), "Object does not exist with addr={:#X}", addr);
+    ASSERT_MSG(objects.find(addr) != objects.end(), "Object does not exist with addr=0x{:X}", addr);
     objects.erase(addr);
 }
 

src/core/hle/kernel/process.cpp

@@ -111,7 +111,7 @@ void Process::ParseKernelCaps(const u32* kernel_caps, size_t len) {
             int major = (kernel_version >> 8) & 0xFF;
             NGLOG_INFO(Loader, "ExHeader kernel version: {}.{}", major, minor);
         } else {
-            NGLOG_ERROR(Loader, "Unhandled kernel caps descriptor: {:#010X}", descriptor);
+            NGLOG_ERROR(Loader, "Unhandled kernel caps descriptor: 0x{:08X}", descriptor);
         }
     }
 }

src/core/hle/kernel/scheduler.cpp

@@ -9,6 +9,8 @@
 
 namespace Kernel {
 
+std::mutex Scheduler::scheduler_mutex;
+
 Scheduler::Scheduler(ARM_Interface* cpu_core) : cpu_core(cpu_core) {}
 
 Scheduler::~Scheduler() {
@@ -18,6 +20,7 @@ Scheduler::~Scheduler() {
 }
 
 bool Scheduler::HaveReadyThreads() {
+    std::lock_guard<std::mutex> lock(scheduler_mutex);
     return ready_queue.get_first() != nullptr;
 }
 
@@ -90,6 +93,8 @@ void Scheduler::SwitchContext(Thread* new_thread) {
 }
 
 void Scheduler::Reschedule() {
+    std::lock_guard<std::mutex> lock(scheduler_mutex);
+
     Thread* cur = GetCurrentThread();
     Thread* next = PopNextReadyThread();
 
@@ -105,26 +110,36 @@ void Scheduler::Reschedule() {
 }
 
 void Scheduler::AddThread(SharedPtr<Thread> thread, u32 priority) {
+    std::lock_guard<std::mutex> lock(scheduler_mutex);
+
     thread_list.push_back(thread);
     ready_queue.prepare(priority);
 }
 
 void Scheduler::RemoveThread(Thread* thread) {
+    std::lock_guard<std::mutex> 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);
+
     ASSERT(thread->status == THREADSTATUS_READY);
     ready_queue.push_back(priority, thread);
 }
 
 void Scheduler::UnscheduleThread(Thread* thread, u32 priority) {
+    std::lock_guard<std::mutex> lock(scheduler_mutex);
+
     ASSERT(thread->status == THREADSTATUS_READY);
     ready_queue.remove(priority, thread);
 }
 
 void Scheduler::SetThreadPriority(Thread* thread, u32 priority) {
+    std::lock_guard<std::mutex> lock(scheduler_mutex);
+
     // If thread was ready, adjust queues
     if (thread->status == THREADSTATUS_READY)
         ready_queue.move(thread, thread->current_priority, priority);

src/core/hle/kernel/scheduler.h

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <mutex>
 #include <vector>
 #include "common/common_types.h"
 #include "common/thread_queue_list.h"
@@ -68,6 +69,8 @@ private:
     SharedPtr<Thread> current_thread = nullptr;
 
     ARM_Interface* cpu_core;
+
+    static std::mutex scheduler_mutex;
 };
 
 } // namespace Kernel

src/core/hle/kernel/server_session.cpp

@@ -61,6 +61,9 @@ void ServerSession::Acquire(Thread* thread) {
 ResultCode ServerSession::HandleDomainSyncRequest(Kernel::HLERequestContext& context) {
     auto& domain_message_header = context.GetDomainMessageHeader();
     if (domain_message_header) {
+        // Set domain handlers in HLE context, used for domain objects (IPC interfaces) as inputs
+        context.SetDomainRequestHandlers(domain_request_handlers);
+
         // If there is a DomainMessageHeader, then this is CommandType "Request"
         const u32 object_id{context.GetDomainMessageHeader()->object_id};
         switch (domain_message_header->command) {
@@ -68,7 +71,7 @@ ResultCode ServerSession::HandleDomainSyncRequest(Kernel::HLERequestContext& con
             return domain_request_handlers[object_id - 1]->HandleSyncRequest(context);
 
         case IPC::DomainMessageHeader::CommandType::CloseVirtualHandle: {
-            NGLOG_DEBUG(IPC, "CloseVirtualHandle, object_id={:#010X}", object_id);
+            NGLOG_DEBUG(IPC, "CloseVirtualHandle, object_id=0x{:08X}", object_id);
 
             domain_request_handlers[object_id - 1] = nullptr;
 

src/core/hle/kernel/shared_memory.cpp

@@ -107,7 +107,7 @@ ResultCode SharedMemory::Map(Process* target_process, VAddr address, MemoryPermi
 
     // Error out if the requested permissions don't match what the creator process allows.
     if (static_cast<u32>(permissions) & ~static_cast<u32>(own_other_permissions)) {
-        NGLOG_ERROR(Kernel, "cannot map id={}, address={:#X} name={}, permissions don't match",
+        NGLOG_ERROR(Kernel, "cannot map id={}, address=0x{:X} name={}, permissions don't match",
                     GetObjectId(), address, name);
         return ERR_INVALID_COMBINATION;
     }
@@ -115,7 +115,7 @@ ResultCode SharedMemory::Map(Process* target_process, VAddr address, MemoryPermi
     // Error out if the provided permissions are not compatible with what the creator process needs.
     if (other_permissions != MemoryPermission::DontCare &&
         static_cast<u32>(this->permissions) & ~static_cast<u32>(other_permissions)) {
-        NGLOG_ERROR(Kernel, "cannot map id={}, address={:#X} name={}, permissions don't match",
+        NGLOG_ERROR(Kernel, "cannot map id={}, address=0x{:X} name={}, permissions don't match",
                     GetObjectId(), address, name);
         return ERR_WRONG_PERMISSION;
     }
@@ -133,7 +133,7 @@ ResultCode SharedMemory::Map(Process* target_process, VAddr address, MemoryPermi
     if (result.Failed()) {
         NGLOG_ERROR(
             Kernel,
-            "cannot map id={}, target_address={:#X} name={}, error mapping to virtual memory",
+            "cannot map id={}, target_address=0x{:X} name={}, error mapping to virtual memory",
             GetObjectId(), target_address, name);
         return result.Code();
     }

src/core/hle/kernel/svc.cpp

@@ -31,7 +31,7 @@ namespace Kernel {
 
 /// Set the process heap to a given Size. It can both extend and shrink the heap.
 static ResultCode SetHeapSize(VAddr* heap_addr, u64 heap_size) {
-    NGLOG_TRACE(Kernel_SVC, "called, heap_size={:#X}", heap_size);
+    NGLOG_TRACE(Kernel_SVC, "called, heap_size=0x{:X}", heap_size);
     auto& process = *Core::CurrentProcess();
     CASCADE_RESULT(*heap_addr,
                    process.HeapAllocate(Memory::HEAP_VADDR, heap_size, VMAPermission::ReadWrite));
@@ -39,20 +39,20 @@ static ResultCode SetHeapSize(VAddr* heap_addr, u64 heap_size) {
 }
 
 static ResultCode SetMemoryAttribute(VAddr addr, u64 size, u32 state0, u32 state1) {
-    NGLOG_WARNING(Kernel_SVC, "(STUBBED) called, addr={:#X}", addr);
+    NGLOG_WARNING(Kernel_SVC, "(STUBBED) called, addr=0x{:X}", addr);
     return RESULT_SUCCESS;
 }
 
 /// Maps a memory range into a different range.
 static ResultCode MapMemory(VAddr dst_addr, VAddr src_addr, u64 size) {
-    NGLOG_TRACE(Kernel_SVC, "called, dst_addr={:#X}, src_addr={:#X}, size={:#X}", dst_addr,
+    NGLOG_TRACE(Kernel_SVC, "called, dst_addr=0x{:X}, src_addr=0x{:X}, size=0x{:X}", dst_addr,
                 src_addr, size);
     return Core::CurrentProcess()->MirrorMemory(dst_addr, src_addr, size);
 }
 
 /// Unmaps a region that was previously mapped with svcMapMemory
 static ResultCode UnmapMemory(VAddr dst_addr, VAddr src_addr, u64 size) {
-    NGLOG_TRACE(Kernel_SVC, "called, dst_addr={:#X}, src_addr={:#X}, size={:#X}", dst_addr,
+    NGLOG_TRACE(Kernel_SVC, "called, dst_addr=0x{:X}, src_addr=0x{:X}, size=0x{:X}", dst_addr,
                 src_addr, size);
     return Core::CurrentProcess()->UnmapMemory(dst_addr, src_addr, size);
 }
@@ -90,11 +90,11 @@ static ResultCode ConnectToNamedPort(Handle* out_handle, VAddr port_name_address
 static ResultCode SendSyncRequest(Handle handle) {
     SharedPtr<ClientSession> session = g_handle_table.Get<ClientSession>(handle);
     if (!session) {
-        NGLOG_ERROR(Kernel_SVC, "called with invalid handle={:#010X}", handle);
+        NGLOG_ERROR(Kernel_SVC, "called with invalid handle=0x{:08X}", handle);
         return ERR_INVALID_HANDLE;
     }
 
-    NGLOG_TRACE(Kernel_SVC, "called handle={:#010X}({})", handle, session->GetName());
+    NGLOG_TRACE(Kernel_SVC, "called handle=0x{:08X}({})", handle, session->GetName());
 
     Core::System::GetInstance().PrepareReschedule();
 
@@ -105,7 +105,7 @@ static ResultCode SendSyncRequest(Handle handle) {
 
 /// Get the ID for the specified thread.
 static ResultCode GetThreadId(u32* thread_id, Handle thread_handle) {
-    NGLOG_TRACE(Kernel_SVC, "called thread={:#010X}", thread_handle);
+    NGLOG_TRACE(Kernel_SVC, "called thread=0x{:08X}", thread_handle);
 
     const SharedPtr<Thread> thread = g_handle_table.Get<Thread>(thread_handle);
     if (!thread) {
@@ -118,7 +118,7 @@ static ResultCode GetThreadId(u32* thread_id, Handle thread_handle) {
 
 /// Get the ID of the specified process
 static ResultCode GetProcessId(u32* process_id, Handle process_handle) {
-    NGLOG_TRACE(Kernel_SVC, "called process={:#010X}", process_handle);
+    NGLOG_TRACE(Kernel_SVC, "called process=0x{:08X}", process_handle);
 
     const SharedPtr<Process> process = g_handle_table.Get<Process>(process_handle);
     if (!process) {
@@ -178,7 +178,7 @@ static ResultCode WaitSynchronization1(
 /// Wait for the given handles to synchronize, timeout after the specified nanoseconds
 static ResultCode WaitSynchronization(Handle* index, VAddr handles_address, u64 handle_count,
                                       s64 nano_seconds) {
-    NGLOG_TRACE(Kernel_SVC, "called handles_address={:#X}, handle_count={}, nano_seconds={}",
+    NGLOG_TRACE(Kernel_SVC, "called handles_address=0x{:X}, handle_count={}, nano_seconds={}",
                 handles_address, handle_count, nano_seconds);
 
     if (!Memory::IsValidVirtualAddress(handles_address))
@@ -239,7 +239,7 @@ static ResultCode WaitSynchronization(Handle* index, VAddr handles_address, u64
 
 /// Resumes a thread waiting on WaitSynchronization
 static ResultCode CancelSynchronization(Handle thread_handle) {
-    NGLOG_TRACE(Kernel_SVC, "called thread={:#X}", thread_handle);
+    NGLOG_TRACE(Kernel_SVC, "called thread=0x{:X}", thread_handle);
 
     const SharedPtr<Thread> thread = g_handle_table.Get<Thread>(thread_handle);
     if (!thread) {
@@ -257,8 +257,8 @@ static ResultCode CancelSynchronization(Handle thread_handle) {
 static ResultCode ArbitrateLock(Handle holding_thread_handle, VAddr mutex_addr,
                                 Handle requesting_thread_handle) {
     NGLOG_TRACE(Kernel_SVC,
-                "called holding_thread_handle={:#010X}, mutex_addr={:#X}, "
-                "requesting_current_thread_handle={:#010X}",
+                "called holding_thread_handle=0x{:08X}, mutex_addr=0x{:X}, "
+                "requesting_current_thread_handle=0x{:08X}",
                 holding_thread_handle, mutex_addr, requesting_thread_handle);
 
     return Mutex::TryAcquire(mutex_addr, holding_thread_handle, requesting_thread_handle);
@@ -266,7 +266,7 @@ static ResultCode ArbitrateLock(Handle holding_thread_handle, VAddr mutex_addr,
 
 /// Unlock a mutex
 static ResultCode ArbitrateUnlock(VAddr mutex_addr) {
-    NGLOG_TRACE(Kernel_SVC, "called mutex_addr={:#X}", mutex_addr);
+    NGLOG_TRACE(Kernel_SVC, "called mutex_addr=0x{:X}", mutex_addr);
 
     return Mutex::Release(mutex_addr);
 }
@@ -286,7 +286,7 @@ static void OutputDebugString(VAddr address, s32 len) {
 
 /// Gets system/memory information for the current process
 static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id) {
-    NGLOG_TRACE(Kernel_SVC, "called info_id={:#X}, info_sub_id={:#X}, handle={:#010X}", info_id,
+    NGLOG_TRACE(Kernel_SVC, "called info_id=0x{:X}, info_sub_id=0x{:X}, handle=0x{:08X}", info_id,
                 info_sub_id, handle);
 
     auto& vm_manager = Core::CurrentProcess()->vm_manager;
@@ -355,14 +355,14 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
 
 /// Sets the thread activity
 static ResultCode SetThreadActivity(Handle handle, u32 unknown) {
-    NGLOG_WARNING(Kernel_SVC, "(STUBBED) called, handle={:#010X}, unknown={:#010X}", handle,
+    NGLOG_WARNING(Kernel_SVC, "(STUBBED) called, handle=0x{:08X}, unknown=0x{:08X}", handle,
                   unknown);
     return RESULT_SUCCESS;
 }
 
 /// Gets the thread context
 static ResultCode GetThreadContext(Handle handle, VAddr addr) {
-    NGLOG_WARNING(Kernel_SVC, "(STUBBED) called, handle={:#010X}, addr={:#X}", handle, addr);
+    NGLOG_WARNING(Kernel_SVC, "(STUBBED) called, handle=0x{:08X}, addr=0x{:X}", handle, addr);
     return RESULT_SUCCESS;
 }
 
@@ -401,15 +401,16 @@ static ResultCode SetThreadPriority(Handle handle, u32 priority) {
 
 /// Get which CPU core is executing the current thread
 static u32 GetCurrentProcessorNumber() {
-    NGLOG_WARNING(Kernel_SVC, "(STUBBED) called, defaulting to processor 0");
-    return 0;
+    NGLOG_TRACE(Kernel_SVC, "called");
+    return GetCurrentThread()->processor_id;
 }
 
 static ResultCode MapSharedMemory(Handle shared_memory_handle, VAddr addr, u64 size,
                                   u32 permissions) {
-    NGLOG_TRACE(Kernel_SVC,
-                "called, shared_memory_handle={:#X}, addr={:#X}, size={:#X}, permissions={:#010X}",
-                shared_memory_handle, addr, size, permissions);
+    NGLOG_TRACE(
+        Kernel_SVC,
+        "called, shared_memory_handle=0x{:X}, addr=0x{:X}, size=0x{:X}, permissions=0x{:08X}",
+        shared_memory_handle, addr, size, permissions);
 
     SharedPtr<SharedMemory> shared_memory = g_handle_table.Get<SharedMemory>(shared_memory_handle);
     if (!shared_memory) {
@@ -429,14 +430,14 @@ static ResultCode MapSharedMemory(Handle shared_memory_handle, VAddr addr, u64 s
         return shared_memory->Map(Core::CurrentProcess().get(), addr, permissions_type,
                                   MemoryPermission::DontCare);
     default:
-        NGLOG_ERROR(Kernel_SVC, "unknown permissions={:#010X}", permissions);
+        NGLOG_ERROR(Kernel_SVC, "unknown permissions=0x{:08X}", permissions);
     }
 
     return RESULT_SUCCESS;
 }
 
 static ResultCode UnmapSharedMemory(Handle shared_memory_handle, VAddr addr, u64 size) {
-    NGLOG_WARNING(Kernel_SVC, "called, shared_memory_handle={:#010X}, addr={:#X}, size={:#X}",
+    NGLOG_WARNING(Kernel_SVC, "called, shared_memory_handle=0x{:08X}, addr=0x{:X}, size=0x{:X}",
                   shared_memory_handle, addr, size);
 
     SharedPtr<SharedMemory> shared_memory = g_handle_table.Get<SharedMemory>(shared_memory_handle);
@@ -465,7 +466,7 @@ static ResultCode QueryProcessMemory(MemoryInfo* memory_info, PageInfo* /*page_i
         memory_info->type = static_cast<u32>(vma->second.meminfo_state);
     }
 
-    NGLOG_TRACE(Kernel_SVC, "called process={:#010X} addr={:X}", process_handle, addr);
+    NGLOG_TRACE(Kernel_SVC, "called process=0x{:08X} addr={:X}", process_handle, addr);
     return RESULT_SUCCESS;
 }
 
@@ -484,22 +485,28 @@ static void ExitProcess() {
 
     Core::CurrentProcess()->status = ProcessStatus::Exited;
 
-    // Stop all the process threads that are currently waiting for objects.
-    auto& thread_list = Core::System::GetInstance().Scheduler().GetThreadList();
-    for (auto& thread : thread_list) {
-        if (thread->owner_process != Core::CurrentProcess())
-            continue;
+    auto stop_threads = [](const std::vector<SharedPtr<Thread>>& thread_list) {
+        for (auto& thread : thread_list) {
+            if (thread->owner_process != Core::CurrentProcess())
+                continue;
 
-        if (thread == GetCurrentThread())
-            continue;
+            if (thread == GetCurrentThread())
+                continue;
 
-        // TODO(Subv): When are the other running/ready threads terminated?
-        ASSERT_MSG(thread->status == THREADSTATUS_WAIT_SYNCH_ANY ||
-                       thread->status == THREADSTATUS_WAIT_SYNCH_ALL,
-                   "Exiting processes with non-waiting threads is currently unimplemented");
+            // TODO(Subv): When are the other running/ready threads terminated?
+            ASSERT_MSG(thread->status == THREADSTATUS_WAIT_SYNCH_ANY ||
+                           thread->status == THREADSTATUS_WAIT_SYNCH_ALL,
+                       "Exiting processes with non-waiting threads is currently unimplemented");
 
-        thread->Stop();
-    }
+            thread->Stop();
+        }
+    };
+
+    auto& system = Core::System::GetInstance();
+    stop_threads(system.Scheduler(0)->GetThreadList());
+    stop_threads(system.Scheduler(1)->GetThreadList());
+    stop_threads(system.Scheduler(2)->GetThreadList());
+    stop_threads(system.Scheduler(3)->GetThreadList());
 
     // Kill the current thread
     GetCurrentThread()->Stop();
@@ -510,7 +517,7 @@ static void ExitProcess() {
 /// Creates a new thread
 static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, VAddr stack_top,
                                u32 priority, s32 processor_id) {
-    std::string name = Common::StringFromFormat("unknown-%llx", entry_point);
+    std::string name = fmt::format("unknown-{:X}", entry_point);
 
     if (priority > THREADPRIO_LOWEST) {
         return ERR_OUT_OF_RANGE;
@@ -529,14 +536,9 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V
 
     switch (processor_id) {
     case THREADPROCESSORID_0:
-        break;
     case THREADPROCESSORID_1:
     case THREADPROCESSORID_2:
     case THREADPROCESSORID_3:
-        // TODO(bunnei): Implement support for other processor IDs
-        NGLOG_ERROR(Kernel_SVC,
-                    "Newly created thread must run in another thread ({}), unimplemented.",
-                    processor_id);
         break;
     default:
         ASSERT_MSG(false, "Unsupported thread processor ID: {}", processor_id);
@@ -552,8 +554,8 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V
     Core::System::GetInstance().PrepareReschedule();
 
     NGLOG_TRACE(Kernel_SVC,
-                "called entrypoint={:#010X} ({}), arg={:#010X}, stacktop={:#010X}, "
-                "threadpriority={:#010X}, processorid={:#010X} : created handle={:#010X}",
+                "called entrypoint=0x{:08X} ({}), arg=0x{:08X}, stacktop=0x{:08X}, "
+                "threadpriority=0x{:08X}, processorid=0x{:08X} : created handle=0x{:08X}",
                 entry_point, name, arg, stack_top, priority, processor_id, *out_handle);
 
     return RESULT_SUCCESS;
@@ -561,7 +563,7 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V
 
 /// Starts the thread for the provided handle
 static ResultCode StartThread(Handle thread_handle) {
-    NGLOG_TRACE(Kernel_SVC, "called thread={:#010X}", thread_handle);
+    NGLOG_TRACE(Kernel_SVC, "called thread=0x{:08X}", thread_handle);
 
     const SharedPtr<Thread> thread = g_handle_table.Get<Thread>(thread_handle);
     if (!thread) {
@@ -575,7 +577,7 @@ static ResultCode StartThread(Handle thread_handle) {
 
 /// Called when a thread exits
 static void ExitThread() {
-    NGLOG_TRACE(Kernel_SVC, "called, pc={:#010X}", Core::CPU().GetPC());
+    NGLOG_TRACE(Kernel_SVC, "called, pc=0x{:08X}", Core::CurrentArmInterface().GetPC());
 
     ExitCurrentThread();
     Core::System::GetInstance().PrepareReschedule();
@@ -587,7 +589,7 @@ static void SleepThread(s64 nanoseconds) {
 
     // Don't attempt to yield execution if there are no available threads to run,
     // this way we avoid a useless reschedule to the idle thread.
-    if (nanoseconds == 0 && !Core::System::GetInstance().Scheduler().HaveReadyThreads())
+    if (nanoseconds == 0 && !Core::System::GetInstance().CurrentScheduler().HaveReadyThreads())
         return;
 
     // Sleep current thread and check for next thread to schedule
@@ -604,7 +606,7 @@ static ResultCode WaitProcessWideKeyAtomic(VAddr mutex_addr, VAddr condition_var
                                            Handle thread_handle, s64 nano_seconds) {
     NGLOG_TRACE(
         Kernel_SVC,
-        "called mutex_addr={:X}, condition_variable_addr={:X}, thread_handle={:#010X}, timeout={}",
+        "called mutex_addr={:X}, condition_variable_addr={:X}, thread_handle=0x{:08X}, timeout={}",
         mutex_addr, condition_variable_addr, thread_handle, nano_seconds);
 
     SharedPtr<Thread> thread = g_handle_table.Get<Thread>(thread_handle);
@@ -623,63 +625,70 @@ static ResultCode WaitProcessWideKeyAtomic(VAddr mutex_addr, VAddr condition_var
 
     // Note: Deliberately don't attempt to inherit the lock owner's priority.
 
-    Core::System::GetInstance().PrepareReschedule();
+    Core::System::GetInstance().CpuCore(current_thread->processor_id).PrepareReschedule();
     return RESULT_SUCCESS;
 }
 
 /// Signal process wide key
 static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target) {
-    NGLOG_TRACE(Kernel_SVC, "called, condition_variable_addr={:#X}, target={:#010X}",
+    NGLOG_TRACE(Kernel_SVC, "called, condition_variable_addr=0x{:X}, target=0x{:08X}",
                 condition_variable_addr, target);
 
     u32 processed = 0;
-    auto& thread_list = Core::System::GetInstance().Scheduler().GetThreadList();
 
-    for (auto& thread : thread_list) {
-        if (thread->condvar_wait_address != condition_variable_addr)
-            continue;
+    auto signal_process_wide_key = [&](size_t core_index) {
+        const auto& scheduler = Core::System::GetInstance().Scheduler(core_index);
+        for (auto& thread : scheduler->GetThreadList()) {
+            if (thread->condvar_wait_address != condition_variable_addr)
+                continue;
 
-        // Only process up to 'target' threads, unless 'target' is -1, in which case process
-        // them all.
-        if (target != -1 && processed >= target)
-            break;
+            // Only process up to 'target' threads, unless 'target' is -1, in which case process
+            // them all.
+            if (target != -1 && processed >= target)
+                break;
 
-        // If the mutex is not yet acquired, acquire it.
-        u32 mutex_val = Memory::Read32(thread->mutex_wait_address);
+            // If the mutex is not yet acquired, acquire it.
+            u32 mutex_val = Memory::Read32(thread->mutex_wait_address);
 
-        if (mutex_val == 0) {
-            // We were able to acquire the mutex, resume this thread.
-            Memory::Write32(thread->mutex_wait_address, thread->wait_handle);
-            ASSERT(thread->status == THREADSTATUS_WAIT_MUTEX);
-            thread->ResumeFromWait();
+            if (mutex_val == 0) {
+                // We were able to acquire the mutex, resume this thread.
+                Memory::Write32(thread->mutex_wait_address, thread->wait_handle);
+                ASSERT(thread->status == THREADSTATUS_WAIT_MUTEX);
+                thread->ResumeFromWait();
 
-            auto lock_owner = thread->lock_owner;
-            if (lock_owner)
-                lock_owner->RemoveMutexWaiter(thread);
+                auto lock_owner = thread->lock_owner;
+                if (lock_owner)
+                    lock_owner->RemoveMutexWaiter(thread);
 
-            thread->lock_owner = nullptr;
-            thread->mutex_wait_address = 0;
-            thread->condvar_wait_address = 0;
-            thread->wait_handle = 0;
-        } else {
-            // Couldn't acquire the mutex, block the thread.
-            Handle owner_handle = static_cast<Handle>(mutex_val & Mutex::MutexOwnerMask);
-            auto owner = g_handle_table.Get<Thread>(owner_handle);
-            ASSERT(owner);
-            ASSERT(thread->status != THREADSTATUS_RUNNING);
-            thread->status = THREADSTATUS_WAIT_MUTEX;
-            thread->wakeup_callback = nullptr;
+                thread->lock_owner = nullptr;
+                thread->mutex_wait_address = 0;
+                thread->condvar_wait_address = 0;
+                thread->wait_handle = 0;
+            } else {
+                // Couldn't acquire the mutex, block the thread.
+                Handle owner_handle = static_cast<Handle>(mutex_val & Mutex::MutexOwnerMask);
+                auto owner = g_handle_table.Get<Thread>(owner_handle);
+                ASSERT(owner);
+                ASSERT(thread->status != THREADSTATUS_RUNNING);
+                thread->status = THREADSTATUS_WAIT_MUTEX;
+                thread->wakeup_callback = nullptr;
 
-            // Signal that the mutex now has a waiting thread.
-            Memory::Write32(thread->mutex_wait_address, mutex_val | Mutex::MutexHasWaitersFlag);
+                // Signal that the mutex now has a waiting thread.
+                Memory::Write32(thread->mutex_wait_address, mutex_val | Mutex::MutexHasWaitersFlag);
 
-            owner->AddMutexWaiter(thread);
+                owner->AddMutexWaiter(thread);
 
-            Core::System::GetInstance().PrepareReschedule();
+                Core::System::GetInstance().CpuCore(thread->processor_id).PrepareReschedule();
+            }
+
+            ++processed;
         }
+    };
 
-        ++processed;
-    }
+    signal_process_wide_key(0);
+    signal_process_wide_key(1);
+    signal_process_wide_key(2);
+    signal_process_wide_key(3);
 
     return RESULT_SUCCESS;
 }
@@ -696,13 +705,13 @@ static u64 GetSystemTick() {
 
 /// Close a handle
 static ResultCode CloseHandle(Handle handle) {
-    NGLOG_TRACE(Kernel_SVC, "Closing handle {:#010X}", handle);
+    NGLOG_TRACE(Kernel_SVC, "Closing handle 0x{:08X}", handle);
     return g_handle_table.Close(handle);
 }
 
 /// Reset an event
 static ResultCode ResetSignal(Handle handle) {
-    NGLOG_WARNING(Kernel_SVC, "(STUBBED) called handle {:#010X}", handle);
+    NGLOG_WARNING(Kernel_SVC, "(STUBBED) called handle 0x{:08X}", handle);
     auto event = g_handle_table.Get<Event>(handle);
     ASSERT(event != nullptr);
     event->Clear();
@@ -711,28 +720,43 @@ static ResultCode ResetSignal(Handle handle) {
 
 /// Creates a TransferMemory object
 static ResultCode CreateTransferMemory(Handle* handle, VAddr addr, u64 size, u32 permissions) {
-    NGLOG_WARNING(Kernel_SVC, "(STUBBED) called addr={:#X}, size={:#X}, perms={:010X}", addr, size,
-                  permissions);
+    NGLOG_WARNING(Kernel_SVC, "(STUBBED) called addr=0x{:X}, size=0x{:X}, perms=0x{:08X}", addr,
+                  size, permissions);
     *handle = 0;
     return RESULT_SUCCESS;
 }
 
-static ResultCode GetThreadCoreMask(Handle handle, u32* mask, u64* unknown) {
-    NGLOG_WARNING(Kernel_SVC, "(STUBBED) called, handle={:010X}", handle);
-    *mask = 0x0;
-    *unknown = 0xf;
+static ResultCode GetThreadCoreMask(Handle thread_handle, u32* core, u64* mask) {
+    NGLOG_TRACE(Kernel_SVC, "called, handle=0x{:08X}", thread_handle);
+
+    const SharedPtr<Thread> thread = g_handle_table.Get<Thread>(thread_handle);
+    if (!thread) {
+        return ERR_INVALID_HANDLE;
+    }
+
+    *core = thread->ideal_core;
+    *mask = thread->affinity_mask;
+
     return RESULT_SUCCESS;
 }
 
-static ResultCode SetThreadCoreMask(Handle handle, u32 mask, u64 unknown) {
-    NGLOG_WARNING(Kernel_SVC, "(STUBBED) called, handle={:#010X}, mask={:#010X}, unknown={:#X}",
-                  handle, mask, unknown);
+static ResultCode SetThreadCoreMask(Handle thread_handle, u32 core, u64 mask) {
+    NGLOG_TRACE(Kernel_SVC, "called, handle=0x{:08X}, mask=0x{:08X}, core=0x{:X}", thread_handle,
+                mask, core);
+
+    const SharedPtr<Thread> thread = g_handle_table.Get<Thread>(thread_handle);
+    if (!thread) {
+        return ERR_INVALID_HANDLE;
+    }
+
+    thread->ChangeCore(core, mask);
+
     return RESULT_SUCCESS;
 }
 
 static ResultCode CreateSharedMemory(Handle* handle, u64 size, u32 local_permissions,
                                      u32 remote_permissions) {
-    NGLOG_TRACE(Kernel_SVC, "called, size={:#X}, localPerms={:#010X}, remotePerms={:#010X}", size,
+    NGLOG_TRACE(Kernel_SVC, "called, size=0x{:X}, localPerms=0x{:08X}, remotePerms=0x{:08X}", size,
                 local_permissions, remote_permissions);
     auto sharedMemHandle =
         SharedMemory::Create(g_handle_table.Get<Process>(KernelHandle::CurrentProcess), size,
@@ -744,7 +768,7 @@ static ResultCode CreateSharedMemory(Handle* handle, u64 size, u32 local_permiss
 }
 
 static ResultCode ClearEvent(Handle handle) {
-    NGLOG_TRACE(Kernel_SVC, "called, event={:010X}", handle);
+    NGLOG_TRACE(Kernel_SVC, "called, event=0x{:08X}", handle);
 
     SharedPtr<Event> evt = g_handle_table.Get<Event>(handle);
     if (evt == nullptr)
@@ -896,7 +920,7 @@ static const FunctionDef SVC_Table[] = {
 
 static const FunctionDef* GetSVCInfo(u32 func_num) {
     if (func_num >= std::size(SVC_Table)) {
-        NGLOG_ERROR(Kernel_SVC, "Unknown svc={:#04X}", func_num);
+        NGLOG_ERROR(Kernel_SVC, "Unknown svc=0x{:02X}", func_num);
         return nullptr;
     }
     return &SVC_Table[func_num];
@@ -918,7 +942,7 @@ void CallSVC(u32 immediate) {
             NGLOG_CRITICAL(Kernel_SVC, "Unimplemented SVC function {}(..)", info->name);
         }
     } else {
-        NGLOG_CRITICAL(Kernel_SVC, "Unknown SVC function {:#X}", immediate);
+        NGLOG_CRITICAL(Kernel_SVC, "Unknown SVC function 0x{:X}", immediate);
     }
 }
 

src/core/hle/kernel/svc_wrap.h

@@ -13,14 +13,14 @@
 
 namespace Kernel {
 
-#define PARAM(n) Core::CPU().GetReg(n)
+#define PARAM(n) Core::CurrentArmInterface().GetReg(n)
 
 /**
  * HLE a function return from the current ARM userland process
  * @param res Result to return
  */
 static inline void FuncReturn(u64 res) {
-    Core::CPU().SetReg(0, res);
+    Core::CurrentArmInterface().SetReg(0, res);
 }
 
 ////////////////////////////////////////////////////////////////////////////////////////////////////
@@ -45,7 +45,7 @@ template <ResultCode func(u32*, u32)>
 void SvcWrap() {
     u32 param_1 = 0;
     u32 retval = func(&param_1, (u32)PARAM(1)).raw;
-    Core::CPU().SetReg(1, param_1);
+    Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
 
@@ -53,7 +53,7 @@ template <ResultCode func(u32*, u64)>
 void SvcWrap() {
     u32 param_1 = 0;
     u32 retval = func(&param_1, PARAM(1)).raw;
-    Core::CPU().SetReg(1, param_1);
+    Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
 
@@ -66,7 +66,7 @@ template <ResultCode func(u64*, u64)>
 void SvcWrap() {
     u64 param_1 = 0;
     u32 retval = func(&param_1, PARAM(1)).raw;
-    Core::CPU().SetReg(1, param_1);
+    Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
 
@@ -85,8 +85,8 @@ void SvcWrap() {
     u32 param_1 = 0;
     u64 param_2 = 0;
     ResultCode retval = func((u32)(PARAM(2) & 0xFFFFFFFF), &param_1, &param_2);
-    Core::CPU().SetReg(1, param_1);
-    Core::CPU().SetReg(2, param_2);
+    Core::CurrentArmInterface().SetReg(1, param_1);
+    Core::CurrentArmInterface().SetReg(2, param_2);
     FuncReturn(retval.raw);
 }
 
@@ -120,7 +120,7 @@ template <ResultCode func(u32*, u64, u64, s64)>
 void SvcWrap() {
     u32 param_1 = 0;
     ResultCode retval = func(&param_1, PARAM(1), (u32)(PARAM(2) & 0xFFFFFFFF), (s64)PARAM(3));
-    Core::CPU().SetReg(1, param_1);
+    Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval.raw);
 }
 
@@ -133,7 +133,7 @@ template <ResultCode func(u64*, u64, u64, u64)>
 void SvcWrap() {
     u64 param_1 = 0;
     u32 retval = func(&param_1, PARAM(1), PARAM(2), PARAM(3)).raw;
-    Core::CPU().SetReg(1, param_1);
+    Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
 
@@ -143,7 +143,7 @@ void SvcWrap() {
     u32 retval =
         func(&param_1, PARAM(1), PARAM(2), PARAM(3), (u32)PARAM(4), (s32)(PARAM(5) & 0xFFFFFFFF))
             .raw;
-    Core::CPU().SetReg(1, param_1);
+    Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
 
@@ -166,7 +166,7 @@ template <ResultCode func(u32*, u64, u64, u32)>
 void SvcWrap() {
     u32 param_1 = 0;
     u32 retval = func(&param_1, PARAM(1), PARAM(2), (u32)(PARAM(3) & 0xFFFFFFFF)).raw;
-    Core::CPU().SetReg(1, param_1);
+    Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
 
@@ -175,7 +175,7 @@ void SvcWrap() {
     u32 param_1 = 0;
     u32 retval =
         func(&param_1, PARAM(1), (u32)(PARAM(2) & 0xFFFFFFFF), (u32)(PARAM(3) & 0xFFFFFFFF)).raw;
-    Core::CPU().SetReg(1, param_1);
+    Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
 

src/core/hle/kernel/thread.cpp

@@ -64,7 +64,7 @@ void Thread::Stop() {
     // Clean up thread from ready queue
     // This is only needed when the thread is termintated forcefully (SVC TerminateProcess)
     if (status == THREADSTATUS_READY) {
-        Core::System::GetInstance().Scheduler().UnscheduleThread(this, current_priority);
+        scheduler->UnscheduleThread(this, current_priority);
     }
 
     status = THREADSTATUS_DEAD;
@@ -92,7 +92,7 @@ void WaitCurrentThread_Sleep() {
 void ExitCurrentThread() {
     Thread* thread = GetCurrentThread();
     thread->Stop();
-    Core::System::GetInstance().Scheduler().RemoveThread(thread);
+    Core::System::GetInstance().CurrentScheduler().RemoveThread(thread);
 }
 
 /**
@@ -146,13 +146,26 @@ void Thread::WakeAfterDelay(s64 nanoseconds) {
     if (nanoseconds == -1)
         return;
 
-    CoreTiming::ScheduleEvent(nsToCycles(nanoseconds), ThreadWakeupEventType, callback_handle);
+    CoreTiming::ScheduleEvent(CoreTiming::nsToCycles(nanoseconds), ThreadWakeupEventType,
+                              callback_handle);
 }
 
 void Thread::CancelWakeupTimer() {
     CoreTiming::UnscheduleEvent(ThreadWakeupEventType, callback_handle);
 }
 
+static boost::optional<s32> GetNextProcessorId(u64 mask) {
+    for (s32 index = 0; index < Core::NUM_CPU_CORES; ++index) {
+        if (mask & (1ULL << index)) {
+            if (!Core::System().GetInstance().Scheduler(index)->GetCurrentThread()) {
+                // Core is enabled and not running any threads, use this one
+                return index;
+            }
+        }
+    }
+    return {};
+}
+
 void Thread::ResumeFromWait() {
     ASSERT_MSG(wait_objects.empty(), "Thread is waking up while waiting for objects");
 
@@ -187,8 +200,37 @@ void Thread::ResumeFromWait() {
     wakeup_callback = nullptr;
 
     status = THREADSTATUS_READY;
-    Core::System::GetInstance().Scheduler().ScheduleThread(this, current_priority);
-    Core::System::GetInstance().PrepareReschedule();
+
+    boost::optional<s32> new_processor_id = GetNextProcessorId(affinity_mask);
+    if (!new_processor_id) {
+        new_processor_id = processor_id;
+    }
+    if (ideal_core != -1 &&
+        Core::System().GetInstance().Scheduler(ideal_core)->GetCurrentThread() == nullptr) {
+        new_processor_id = ideal_core;
+    }
+
+    ASSERT(*new_processor_id < 4);
+
+    // Add thread to new core's scheduler
+    auto& next_scheduler = Core::System().GetInstance().Scheduler(*new_processor_id);
+
+    if (*new_processor_id != processor_id) {
+        // Remove thread from previous core's scheduler
+        scheduler->RemoveThread(this);
+        next_scheduler->AddThread(this, current_priority);
+    }
+
+    processor_id = *new_processor_id;
+
+    // If the thread was ready, unschedule from the previous core and schedule on the new core
+    scheduler->UnscheduleThread(this, current_priority);
+    next_scheduler->ScheduleThread(this, current_priority);
+
+    // Change thread's scheduler
+    scheduler = next_scheduler;
+
+    Core::System::GetInstance().CpuCore(processor_id).PrepareReschedule();
 }
 
 /**
@@ -258,8 +300,6 @@ ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point,
 
     SharedPtr<Thread> thread(new Thread);
 
-    Core::System::GetInstance().Scheduler().AddThread(thread, priority);
-
     thread->thread_id = NewThreadId();
     thread->status = THREADSTATUS_DORMANT;
     thread->entry_point = entry_point;
@@ -267,6 +307,8 @@ ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point,
     thread->nominal_priority = thread->current_priority = priority;
     thread->last_running_ticks = CoreTiming::GetTicks();
     thread->processor_id = processor_id;
+    thread->ideal_core = processor_id;
+    thread->affinity_mask = 1ULL << processor_id;
     thread->wait_objects.clear();
     thread->mutex_wait_address = 0;
     thread->condvar_wait_address = 0;
@@ -274,6 +316,8 @@ ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point,
     thread->name = std::move(name);
     thread->callback_handle = wakeup_callback_handle_table.Create(thread).Unwrap();
     thread->owner_process = owner_process;
+    thread->scheduler = Core::System().GetInstance().Scheduler(processor_id);
+    thread->scheduler->AddThread(thread, priority);
 
     // Find the next available TLS index, and mark it as used
     auto& tls_slots = owner_process->tls_slots;
@@ -336,7 +380,7 @@ void Thread::SetPriority(u32 priority) {
 }
 
 void Thread::BoostPriority(u32 priority) {
-    Core::System::GetInstance().Scheduler().SetThreadPriority(this, priority);
+    scheduler->SetThreadPriority(this, priority);
     current_priority = priority;
 }
 
@@ -405,7 +449,7 @@ void Thread::UpdatePriority() {
     if (new_priority == current_priority)
         return;
 
-    Core::System::GetInstance().Scheduler().SetThreadPriority(this, new_priority);
+    scheduler->SetThreadPriority(this, new_priority);
 
     current_priority = new_priority;
 
@@ -414,13 +458,54 @@ void Thread::UpdatePriority() {
         lock_owner->UpdatePriority();
 }
 
+void Thread::ChangeCore(u32 core, u64 mask) {
+    ideal_core = core;
+    mask = mask;
+
+    if (status != THREADSTATUS_READY) {
+        return;
+    }
+
+    boost::optional<s32> new_processor_id{GetNextProcessorId(mask)};
+
+    if (!new_processor_id) {
+        new_processor_id = processor_id;
+    }
+    if (ideal_core != -1 &&
+        Core::System().GetInstance().Scheduler(ideal_core)->GetCurrentThread() == nullptr) {
+        new_processor_id = ideal_core;
+    }
+
+    ASSERT(new_processor_id < 4);
+
+    // Add thread to new core's scheduler
+    auto& next_scheduler = Core::System().GetInstance().Scheduler(*new_processor_id);
+
+    if (*new_processor_id != processor_id) {
+        // Remove thread from previous core's scheduler
+        scheduler->RemoveThread(this);
+        next_scheduler->AddThread(this, current_priority);
+    }
+
+    processor_id = *new_processor_id;
+
+    // If the thread was ready, unschedule from the previous core and schedule on the new core
+    scheduler->UnscheduleThread(this, current_priority);
+    next_scheduler->ScheduleThread(this, current_priority);
+
+    // Change thread's scheduler
+    scheduler = next_scheduler;
+
+    Core::System::GetInstance().CpuCore(processor_id).PrepareReschedule();
+}
+
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 
 /**
  * Gets the current thread
  */
 Thread* GetCurrentThread() {
-    return Core::System::GetInstance().Scheduler().GetCurrentThread();
+    return Core::System::GetInstance().CurrentScheduler().GetCurrentThread();
 }
 
 void ThreadingInit() {

src/core/hle/kernel/thread.h

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <memory>
 #include <string>
 #include <unordered_map>
 #include <vector>
@@ -56,6 +57,7 @@ enum class ThreadWakeupReason {
 namespace Kernel {
 
 class Process;
+class Scheduler;
 
 class Thread final : public WaitObject {
 public:
@@ -118,6 +120,9 @@ public:
     /// Recalculates the current priority taking into account priority inheritance.
     void UpdatePriority();
 
+    /// Changes the core that the thread is running or scheduled to run on.
+    void ChangeCore(u32 core, u64 mask);
+
     /**
      * Gets the thread's thread ID
      * @return The thread's ID
@@ -240,6 +245,11 @@ public:
     // available. In case of a timeout, the object will be nullptr.
     std::function<WakeupCallback> wakeup_callback;
 
+    std::shared_ptr<Scheduler> scheduler;
+
+    u32 ideal_core{0xFFFFFFFF};
+    u64 affinity_mask{0x1};
+
 private:
     Thread();
     ~Thread() override;

src/core/hle/kernel/timer.cpp

@@ -57,7 +57,8 @@ void Timer::Set(s64 initial, s64 interval) {
         // Immediately invoke the callback
         Signal(0);
     } else {
-        CoreTiming::ScheduleEvent(nsToCycles(initial), timer_callback_event_type, callback_handle);
+        CoreTiming::ScheduleEvent(CoreTiming::nsToCycles(initial), timer_callback_event_type,
+                                  callback_handle);
     }
 }
 
@@ -86,7 +87,7 @@ void Timer::Signal(int cycles_late) {
 
     if (interval_delay != 0) {
         // Reschedule the timer with the interval delay
-        CoreTiming::ScheduleEvent(nsToCycles(interval_delay) - cycles_late,
+        CoreTiming::ScheduleEvent(CoreTiming::nsToCycles(interval_delay) - cycles_late,
                                   timer_callback_event_type, callback_handle);
     }
 }

src/core/hle/kernel/vm_manager.cpp

@@ -104,8 +104,15 @@ ResultVal<VMManager::VMAHandle> VMManager::MapMemoryBlock(VAddr target,
     VirtualMemoryArea& final_vma = vma_handle->second;
     ASSERT(final_vma.size == size);
 
-    Core::CPU().MapBackingMemory(target, size, block->data() + offset,
-                                 VMAPermission::ReadWriteExecute);
+    auto& system = Core::System::GetInstance();
+    system.ArmInterface(0).MapBackingMemory(target, size, block->data() + offset,
+                                            VMAPermission::ReadWriteExecute);
+    system.ArmInterface(1).MapBackingMemory(target, size, block->data() + offset,
+                                            VMAPermission::ReadWriteExecute);
+    system.ArmInterface(2).MapBackingMemory(target, size, block->data() + offset,
+                                            VMAPermission::ReadWriteExecute);
+    system.ArmInterface(3).MapBackingMemory(target, size, block->data() + offset,
+                                            VMAPermission::ReadWriteExecute);
 
     final_vma.type = VMAType::AllocatedMemoryBlock;
     final_vma.permissions = VMAPermission::ReadWrite;
@@ -126,7 +133,11 @@ ResultVal<VMManager::VMAHandle> VMManager::MapBackingMemory(VAddr target, u8* me
     VirtualMemoryArea& final_vma = vma_handle->second;
     ASSERT(final_vma.size == size);
 
-    Core::CPU().MapBackingMemory(target, size, memory, VMAPermission::ReadWriteExecute);
+    auto& system = Core::System::GetInstance();
+    system.ArmInterface(0).MapBackingMemory(target, size, memory, VMAPermission::ReadWriteExecute);
+    system.ArmInterface(1).MapBackingMemory(target, size, memory, VMAPermission::ReadWriteExecute);
+    system.ArmInterface(2).MapBackingMemory(target, size, memory, VMAPermission::ReadWriteExecute);
+    system.ArmInterface(3).MapBackingMemory(target, size, memory, VMAPermission::ReadWriteExecute);
 
     final_vma.type = VMAType::BackingMemory;
     final_vma.permissions = VMAPermission::ReadWrite;
@@ -184,7 +195,11 @@ ResultCode VMManager::UnmapRange(VAddr target, u64 size) {
 
     ASSERT(FindVMA(target)->second.size >= size);
 
-    Core::CPU().UnmapMemory(target, size);
+    auto& system = Core::System::GetInstance();
+    system.ArmInterface(0).UnmapMemory(target, size);
+    system.ArmInterface(1).UnmapMemory(target, size);
+    system.ArmInterface(2).UnmapMemory(target, size);
+    system.ArmInterface(3).UnmapMemory(target, size);
 
     return RESULT_SUCCESS;
 }
@@ -243,8 +258,8 @@ VMManager::VMAIter VMManager::StripIterConstness(const VMAHandle& iter) {
 }
 
 ResultVal<VMManager::VMAIter> VMManager::CarveVMA(VAddr base, u64 size) {
-    ASSERT_MSG((size & Memory::PAGE_MASK) == 0, "non-page aligned size: {:#018X}", size);
-    ASSERT_MSG((base & Memory::PAGE_MASK) == 0, "non-page aligned base: {:#018X}", base);
+    ASSERT_MSG((size & Memory::PAGE_MASK) == 0, "non-page aligned size: 0x{:016X}", size);
+    ASSERT_MSG((base & Memory::PAGE_MASK) == 0, "non-page aligned base: 0x{:016X}", base);
 
     VMAIter vma_handle = StripIterConstness(FindVMA(base));
     if (vma_handle == vma_map.end()) {
@@ -279,8 +294,8 @@ ResultVal<VMManager::VMAIter> VMManager::CarveVMA(VAddr base, u64 size) {
 }
 
 ResultVal<VMManager::VMAIter> VMManager::CarveVMARange(VAddr target, u64 size) {
-    ASSERT_MSG((size & Memory::PAGE_MASK) == 0, "non-page aligned size: {:#018X}", size);
-    ASSERT_MSG((target & Memory::PAGE_MASK) == 0, "non-page aligned base: {:#018X}", target);
+    ASSERT_MSG((size & Memory::PAGE_MASK) == 0, "non-page aligned size: 0x{:016X}", size);
+    ASSERT_MSG((target & Memory::PAGE_MASK) == 0, "non-page aligned base: 0x{:016X}", target);
 
     VAddr target_end = target + size;
     ASSERT(target_end >= target);

src/core/hle/service/am/am.cpp

@@ -103,7 +103,7 @@ ISelfController::ISelfController(std::shared_ptr<NVFlinger::NVFlinger> nvflinger
         {40, &ISelfController::CreateManagedDisplayLayer, "CreateManagedDisplayLayer"},
         {41, nullptr, "IsSystemBufferSharingEnabled"},
         {42, nullptr, "GetSystemSharedLayerHandle"},
-        {50, nullptr, "SetHandlesRequestToDisplay"},
+        {50, &ISelfController::SetHandlesRequestToDisplay, "SetHandlesRequestToDisplay"},
         {51, nullptr, "ApproveToDisplay"},
         {60, nullptr, "OverrideAutoSleepTimeAndDimmingTime"},
         {61, nullptr, "SetMediaPlaybackState"},
@@ -229,6 +229,13 @@ void ISelfController::CreateManagedDisplayLayer(Kernel::HLERequestContext& ctx)
     NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
+void ISelfController::SetHandlesRequestToDisplay(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
+}
+
 ICommonStateGetter::ICommonStateGetter() : ServiceFramework("ICommonStateGetter") {
     static const FunctionInfo functions[] = {
         {0, &ICommonStateGetter::GetEventHandle, "GetEventHandle"},
@@ -445,7 +452,8 @@ IApplicationFunctions::IApplicationFunctions() : ServiceFramework("IApplicationF
         {10, nullptr, "CreateApplicationAndPushAndRequestToStart"},
         {11, nullptr, "CreateApplicationAndPushAndRequestToStartForQuest"},
         {12, nullptr, "CreateApplicationAndRequestToStart"},
-        {13, nullptr, "CreateApplicationAndRequestToStartForQuest"},
+        {13, &IApplicationFunctions::CreateApplicationAndRequestToStartForQuest,
+         "CreateApplicationAndRequestToStartForQuest"},
         {20, &IApplicationFunctions::EnsureSaveData, "EnsureSaveData"},
         {21, &IApplicationFunctions::GetDesiredLanguage, "GetDesiredLanguage"},
         {22, &IApplicationFunctions::SetTerminateResult, "SetTerminateResult"},
@@ -502,6 +510,13 @@ void IApplicationFunctions::PopLaunchParameter(Kernel::HLERequestContext& ctx) {
     NGLOG_DEBUG(Service_AM, "called");
 }
 
+void IApplicationFunctions::CreateApplicationAndRequestToStartForQuest(
+    Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
+}
+
 void IApplicationFunctions::EnsureSaveData(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     u128 uid = rp.PopRaw<u128>();
@@ -534,7 +549,7 @@ void IApplicationFunctions::SetTerminateResult(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 
-    NGLOG_WARNING(Service_AM, "(STUBBED) called, result={:#010}", result);
+    NGLOG_WARNING(Service_AM, "(STUBBED) called, result=0x{:08X}", result);
 }
 
 void IApplicationFunctions::GetDesiredLanguage(Kernel::HLERequestContext& ctx) {
@@ -572,4 +587,64 @@ void InstallInterfaces(SM::ServiceManager& service_manager,
     std::make_shared<AppletOE>(nvflinger)->InstallAsService(service_manager);
 }
 
+IHomeMenuFunctions::IHomeMenuFunctions() : ServiceFramework("IHomeMenuFunctions") {
+    static const FunctionInfo functions[] = {
+        {10, &IHomeMenuFunctions::RequestToGetForeground, "RequestToGetForeground"},
+        {11, nullptr, "LockForeground"},
+        {12, nullptr, "UnlockForeground"},
+        {20, nullptr, "PopFromGeneralChannel"},
+        {21, nullptr, "GetPopFromGeneralChannelEvent"},
+        {30, nullptr, "GetHomeButtonWriterLockAccessor"},
+        {31, nullptr, "GetWriterLockAccessorEx"},
+    };
+    RegisterHandlers(functions);
+}
+
+void IHomeMenuFunctions::RequestToGetForeground(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
+}
+
+IGlobalStateController::IGlobalStateController() : ServiceFramework("IGlobalStateController") {
+    static const FunctionInfo functions[] = {
+        {0, nullptr, "RequestToEnterSleep"},
+        {1, nullptr, "EnterSleep"},
+        {2, nullptr, "StartSleepSequence"},
+        {3, nullptr, "StartShutdownSequence"},
+        {4, nullptr, "StartRebootSequence"},
+        {10, nullptr, "LoadAndApplyIdlePolicySettings"},
+        {11, nullptr, "NotifyCecSettingsChanged"},
+        {12, nullptr, "SetDefaultHomeButtonLongPressTime"},
+        {13, nullptr, "UpdateDefaultDisplayResolution"},
+        {14, nullptr, "ShouldSleepOnBoot"},
+        {15, nullptr, "GetHdcpAuthenticationFailedEvent"},
+    };
+    RegisterHandlers(functions);
+}
+
+IApplicationCreator::IApplicationCreator() : ServiceFramework("IApplicationCreator") {
+    static const FunctionInfo functions[] = {
+        {0, nullptr, "CreateApplication"},
+        {1, nullptr, "PopLaunchRequestedApplication"},
+        {10, nullptr, "CreateSystemApplication"},
+        {100, nullptr, "PopFloatingApplicationForDevelopment"},
+    };
+    RegisterHandlers(functions);
+}
+
+IProcessWindingController::IProcessWindingController()
+    : ServiceFramework("IProcessWindingController") {
+    static const FunctionInfo functions[] = {
+        {0, nullptr, "GetLaunchReason"},
+        {11, nullptr, "OpenCallingLibraryApplet"},
+        {21, nullptr, "PushContext"},
+        {22, nullptr, "PopContext"},
+        {23, nullptr, "CancelWindingReservation"},
+        {30, nullptr, "WindAndDoReserved"},
+        {40, nullptr, "ReserveToStartAndWaitAndUnwindThis"},
+        {41, nullptr, "ReserveToStartAndWait"},
+    };
+    RegisterHandlers(functions);
+}
 } // namespace Service::AM

src/core/hle/service/am/am.h

@@ -70,6 +70,7 @@ private:
     void GetLibraryAppletLaunchableEvent(Kernel::HLERequestContext& ctx);
     void CreateManagedDisplayLayer(Kernel::HLERequestContext& ctx);
     void SetScreenShotPermission(Kernel::HLERequestContext& ctx);
+    void SetHandlesRequestToDisplay(Kernel::HLERequestContext& ctx);
 
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
     Kernel::SharedPtr<Kernel::Event> launchable_event;
@@ -113,6 +114,7 @@ public:
 
 private:
     void PopLaunchParameter(Kernel::HLERequestContext& ctx);
+    void CreateApplicationAndRequestToStartForQuest(Kernel::HLERequestContext& ctx);
     void EnsureSaveData(Kernel::HLERequestContext& ctx);
     void SetTerminateResult(Kernel::HLERequestContext& ctx);
     void GetDesiredLanguage(Kernel::HLERequestContext& ctx);
@@ -121,6 +123,29 @@ private:
     void NotifyRunning(Kernel::HLERequestContext& ctx);
 };
 
+class IHomeMenuFunctions final : public ServiceFramework<IHomeMenuFunctions> {
+public:
+    IHomeMenuFunctions();
+
+private:
+    void RequestToGetForeground(Kernel::HLERequestContext& ctx);
+};
+
+class IGlobalStateController final : public ServiceFramework<IGlobalStateController> {
+public:
+    IGlobalStateController();
+};
+
+class IApplicationCreator final : public ServiceFramework<IApplicationCreator> {
+public:
+    IApplicationCreator();
+};
+
+class IProcessWindingController final : public ServiceFramework<IProcessWindingController> {
+public:
+    IProcessWindingController();
+};
+
 /// Registers all AM services with the specified service manager.
 void InstallInterfaces(SM::ServiceManager& service_manager,
                        std::shared_ptr<NVFlinger::NVFlinger> nvflinger);

src/core/hle/service/am/applet_ae.cpp

@@ -20,7 +20,7 @@ public:
             {2, &ILibraryAppletProxy::GetWindowController, "GetWindowController"},
             {3, &ILibraryAppletProxy::GetAudioController, "GetAudioController"},
             {4, &ILibraryAppletProxy::GetDisplayController, "GetDisplayController"},
-            {10, nullptr, "GetProcessWindingController"},
+            {10, &ILibraryAppletProxy::GetProcessWindingController, "GetProcessWindingController"},
             {11, &ILibraryAppletProxy::GetLibraryAppletCreator, "GetLibraryAppletCreator"},
             {20, &ILibraryAppletProxy::GetApplicationFunctions, "GetApplicationFunctions"},
             {1000, &ILibraryAppletProxy::GetDebugFunctions, "GetDebugFunctions"},
@@ -28,6 +28,93 @@ public:
         RegisterHandlers(functions);
     }
 
+private:
+    void GetCommonStateGetter(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<ICommonStateGetter>();
+        NGLOG_DEBUG(Service_AM, "called");
+    }
+
+    void GetSelfController(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<ISelfController>(nvflinger);
+        NGLOG_DEBUG(Service_AM, "called");
+    }
+
+    void GetWindowController(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<IWindowController>();
+        NGLOG_DEBUG(Service_AM, "called");
+    }
+
+    void GetAudioController(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<IAudioController>();
+        NGLOG_DEBUG(Service_AM, "called");
+    }
+
+    void GetDisplayController(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<IDisplayController>();
+        NGLOG_DEBUG(Service_AM, "called");
+    }
+
+    void GetProcessWindingController(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<IProcessWindingController>();
+        NGLOG_DEBUG(Service_AM, "called");
+    }
+
+    void GetDebugFunctions(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<IDebugFunctions>();
+        NGLOG_DEBUG(Service_AM, "called");
+    }
+
+    void GetLibraryAppletCreator(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<ILibraryAppletCreator>();
+        NGLOG_DEBUG(Service_AM, "called");
+    }
+
+    void GetApplicationFunctions(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<IApplicationFunctions>();
+        NGLOG_DEBUG(Service_AM, "called");
+    }
+
+    std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
+};
+
+class ISystemAppletProxy final : public ServiceFramework<ISystemAppletProxy> {
+public:
+    explicit ISystemAppletProxy(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
+        : ServiceFramework("ISystemAppletProxy"), nvflinger(std::move(nvflinger)) {
+        static const FunctionInfo functions[] = {
+            {0, &ISystemAppletProxy::GetCommonStateGetter, "GetCommonStateGetter"},
+            {1, &ISystemAppletProxy::GetSelfController, "GetSelfController"},
+            {2, &ISystemAppletProxy::GetWindowController, "GetWindowController"},
+            {3, &ISystemAppletProxy::GetAudioController, "GetAudioController"},
+            {4, &ISystemAppletProxy::GetDisplayController, "GetDisplayController"},
+            {10, nullptr, "GetProcessWindingController"},
+            {11, &ISystemAppletProxy::GetLibraryAppletCreator, "GetLibraryAppletCreator"},
+            {20, &ISystemAppletProxy::GetHomeMenuFunctions, "GetHomeMenuFunctions"},
+            {21, &ISystemAppletProxy::GetGlobalStateController, "GetGlobalStateController"},
+            {22, &ISystemAppletProxy::GetApplicationCreator, "GetApplicationCreator"},
+            {1000, &ISystemAppletProxy::GetDebugFunctions, "GetDebugFunctions"},
+        };
+        RegisterHandlers(functions);
+    }
+
 private:
     void GetCommonStateGetter(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
@@ -78,16 +165,43 @@ private:
         NGLOG_DEBUG(Service_AM, "called");
     }
 
-    void GetApplicationFunctions(Kernel::HLERequestContext& ctx) {
+    void GetHomeMenuFunctions(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(RESULT_SUCCESS);
-        rb.PushIpcInterface<IApplicationFunctions>();
+        rb.PushIpcInterface<IHomeMenuFunctions>();
         NGLOG_DEBUG(Service_AM, "called");
     }
 
+    void GetGlobalStateController(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<IGlobalStateController>();
+        NGLOG_DEBUG(Service_AM, "called");
+    }
+
+    void GetApplicationCreator(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<IApplicationCreator>();
+        NGLOG_DEBUG(Service_AM, "called");
+    }
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;
 };
 
+void AppletAE::OpenSystemAppletProxy(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushIpcInterface<ISystemAppletProxy>(nvflinger);
+    NGLOG_DEBUG(Service_AM, "called");
+}
+
+void AppletAE::OpenLibraryAppletProxy(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushIpcInterface<ILibraryAppletProxy>(nvflinger);
+    NGLOG_DEBUG(Service_AM, "called");
+}
+
 void AppletAE::OpenLibraryAppletProxyOld(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(RESULT_SUCCESS);
@@ -98,9 +212,9 @@ void AppletAE::OpenLibraryAppletProxyOld(Kernel::HLERequestContext& ctx) {
 AppletAE::AppletAE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger)
     : ServiceFramework("appletAE"), nvflinger(std::move(nvflinger)) {
     static const FunctionInfo functions[] = {
-        {100, nullptr, "OpenSystemAppletProxy"},
+        {100, &AppletAE::OpenSystemAppletProxy, "OpenSystemAppletProxy"},
         {200, &AppletAE::OpenLibraryAppletProxyOld, "OpenLibraryAppletProxyOld"},
-        {201, nullptr, "OpenLibraryAppletProxy"},
+        {201, &AppletAE::OpenLibraryAppletProxy, "OpenLibraryAppletProxy"},
         {300, nullptr, "OpenOverlayAppletProxy"},
         {350, nullptr, "OpenSystemApplicationProxy"},
         {400, nullptr, "CreateSelfLibraryAppletCreatorForDevelop"},

src/core/hle/service/am/applet_ae.h

@@ -21,6 +21,8 @@ public:
     ~AppletAE() = default;
 
 private:
+    void OpenSystemAppletProxy(Kernel::HLERequestContext& ctx);
+    void OpenLibraryAppletProxy(Kernel::HLERequestContext& ctx);
     void OpenLibraryAppletProxyOld(Kernel::HLERequestContext& ctx);
 
     std::shared_ptr<NVFlinger::NVFlinger> nvflinger;

src/core/hle/service/audio/audout_u.cpp

@@ -18,7 +18,7 @@ constexpr u32 sample_rate{48000};
 /// to more audio channels (probably when Docked I guess)
 constexpr u32 audio_channels{2};
 /// TODO(st4rk): find a proper value for the audio_ticks
-constexpr u64 audio_ticks{static_cast<u64>(BASE_CLOCK_RATE / 500)};
+constexpr u64 audio_ticks{static_cast<u64>(CoreTiming::BASE_CLOCK_RATE / 500)};
 
 class IAudioOut final : public ServiceFramework<IAudioOut> {
 public:

src/core/hle/service/audio/audren_u.cpp

@@ -12,7 +12,7 @@
 namespace Service::Audio {
 
 /// TODO(bunnei): Find a proper value for the audio_ticks
-constexpr u64 audio_ticks{static_cast<u64>(BASE_CLOCK_RATE / 200)};
+constexpr u64 audio_ticks{static_cast<u64>(CoreTiming::BASE_CLOCK_RATE / 200)};
 
 class IAudioRenderer final : public ServiceFramework<IAudioRenderer> {
 public:

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

@@ -16,7 +16,7 @@ Module::Interface::Interface(std::shared_ptr<Module> module, const char* name)
 void Module::Interface::FatalSimple(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp(ctx);
     u32 error_code = rp.Pop<u32>();
-    NGLOG_WARNING(Service_Fatal, "(STUBBED) called, error_code={:#X}", error_code);
+    NGLOG_WARNING(Service_Fatal, "(STUBBED) called, error_code=0x{:X}", error_code);
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(RESULT_SUCCESS);
 }

src/core/hle/service/filesystem/filesystem.cpp

@@ -25,7 +25,7 @@ ResultCode RegisterFileSystem(std::unique_ptr<FileSys::FileSystemFactory>&& fact
     ASSERT_MSG(inserted, "Tried to register more than one system with same id code");
 
     auto& filesystem = result.first->second;
-    NGLOG_DEBUG(Service_FS, "Registered file system {} with id code {:#010X}",
+    NGLOG_DEBUG(Service_FS, "Registered file system {} with id code 0x{:08X}",
                 filesystem->GetName(), static_cast<u32>(type));
     return RESULT_SUCCESS;
 }

src/core/hle/service/filesystem/fsp_srv.cpp

@@ -35,7 +35,7 @@ private:
         const s64 offset = rp.Pop<s64>();
         const s64 length = rp.Pop<s64>();
 
-        NGLOG_DEBUG(Service_FS, "called, offset={:#X}, length={}", offset, length);
+        NGLOG_DEBUG(Service_FS, "called, offset=0x{:X}, length={}", offset, length);
 
         // Error checking
         if (length < 0) {
@@ -87,7 +87,7 @@ private:
         const s64 offset = rp.Pop<s64>();
         const s64 length = rp.Pop<s64>();
 
-        NGLOG_DEBUG(Service_FS, "called, offset={:#X}, length={}", offset, length);
+        NGLOG_DEBUG(Service_FS, "called, offset=0x{:X}, length={}", offset, length);
 
         // Error checking
         if (length < 0) {
@@ -124,7 +124,7 @@ private:
         const s64 offset = rp.Pop<s64>();
         const s64 length = rp.Pop<s64>();
 
-        NGLOG_DEBUG(Service_FS, "called, offset={:#X}, length={}", offset, length);
+        NGLOG_DEBUG(Service_FS, "called, offset=0x{:X}, length={}", offset, length);
 
         // Error checking
         if (length < 0) {
@@ -197,7 +197,7 @@ private:
         IPC::RequestParser rp{ctx};
         const u64 unk = rp.Pop<u64>();
 
-        NGLOG_DEBUG(Service_FS, "called, unk={:#X}", unk);
+        NGLOG_DEBUG(Service_FS, "called, unk=0x{:X}", unk);
 
         // Calculate how many entries we can fit in the output buffer
         u64 count_entries = ctx.GetWriteBufferSize() / sizeof(FileSys::Entry);
@@ -265,7 +265,7 @@ public:
         u64 mode = rp.Pop<u64>();
         u32 size = rp.Pop<u32>();
 
-        NGLOG_DEBUG(Service_FS, "called file {} mode {:#X} size {:#010X}", name, mode, size);
+        NGLOG_DEBUG(Service_FS, "called file {} mode 0x{:X} size 0x{:08X}", name, mode, size);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(backend->CreateFile(name, size));

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

@@ -18,9 +18,9 @@ namespace Service::HID {
 
 // Updating period for each HID device.
 // TODO(shinyquagsire23): These need better values.
-constexpr u64 pad_update_ticks = BASE_CLOCK_RATE / 10000;
-constexpr u64 accelerometer_update_ticks = BASE_CLOCK_RATE / 10000;
-constexpr u64 gyroscope_update_ticks = BASE_CLOCK_RATE / 10000;
+constexpr u64 pad_update_ticks = CoreTiming::BASE_CLOCK_RATE / 10000;
+constexpr u64 accelerometer_update_ticks = CoreTiming::BASE_CLOCK_RATE / 10000;
+constexpr u64 gyroscope_update_ticks = CoreTiming::BASE_CLOCK_RATE / 10000;
 
 class IAppletResource final : public ServiceFramework<IAppletResource> {
 public:
@@ -60,7 +60,10 @@ private:
         std::transform(Settings::values.buttons.begin() + Settings::NativeButton::BUTTON_HID_BEGIN,
                        Settings::values.buttons.begin() + Settings::NativeButton::BUTTON_HID_END,
                        buttons.begin(), Input::CreateDevice<Input::ButtonDevice>);
-        // TODO(shinyquagsire23): sticks, gyro, touch, mouse, keyboard
+        std::transform(Settings::values.analogs.begin() + Settings::NativeAnalog::STICK_HID_BEGIN,
+                       Settings::values.analogs.begin() + Settings::NativeAnalog::STICK_HID_END,
+                       sticks.begin(), Input::CreateDevice<Input::AnalogDevice>);
+        // TODO(shinyquagsire23): gyro, touch, mouse, keyboard
     }
 
     void UpdatePadCallback(u64 userdata, int cycles_late) {
@@ -79,61 +82,70 @@ private:
         controller_header.left_color_body = JOYCON_BODY_NEON_BLUE;
         controller_header.left_color_buttons = JOYCON_BUTTONS_NEON_BLUE;
 
-        for (int index = 0; index < HID_NUM_LAYOUTS; index++) {
-            ControllerLayout& layout = mem.controllers[Controller_Handheld].layouts[index];
-            layout.header.num_entries = HID_NUM_ENTRIES;
-            layout.header.max_entry_index = HID_NUM_ENTRIES - 1;
+        for (size_t controller = 0; controller < mem.controllers.size(); controller++) {
+            for (int index = 0; index < HID_NUM_LAYOUTS; index++) {
+                ControllerLayout& layout = mem.controllers[controller].layouts[index];
+                layout.header.num_entries = HID_NUM_ENTRIES;
+                layout.header.max_entry_index = HID_NUM_ENTRIES - 1;
 
-            // HID shared memory stores the state of the past 17 samples in a circlular buffer,
-            // each with a timestamp in number of samples since boot.
-            layout.header.timestamp_ticks = CoreTiming::GetTicks();
-            layout.header.latest_entry = (layout.header.latest_entry + 1) % HID_NUM_ENTRIES;
+                // HID shared memory stores the state of the past 17 samples in a circlular buffer,
+                // each with a timestamp in number of samples since boot.
+                layout.header.timestamp_ticks = CoreTiming::GetTicks();
+                layout.header.latest_entry = (layout.header.latest_entry + 1) % HID_NUM_ENTRIES;
 
-            ControllerInputEntry& entry = layout.entries[layout.header.latest_entry];
-            entry.connection_state = ConnectionState_Connected | ConnectionState_Wired;
-            entry.timestamp++;
-            entry.timestamp_2++; // TODO(shinyquagsire23): Is this always identical to timestamp?
+                ControllerInputEntry& entry = layout.entries[layout.header.latest_entry];
+                entry.connection_state = ConnectionState_Connected | ConnectionState_Wired;
+                entry.timestamp++;
+                // TODO(shinyquagsire23): Is this always identical to timestamp?
+                entry.timestamp_2++;
 
-            // TODO(shinyquagsire23): Set up some LUTs for each layout mapping in the future?
-            // For now everything is just the default handheld layout, but split Joy-Con will
-            // rotate the face buttons and directions for certain layouts.
-            ControllerPadState& state = entry.buttons;
-            using namespace Settings::NativeButton;
-            state.a.Assign(buttons[A - BUTTON_HID_BEGIN]->GetStatus());
-            state.b.Assign(buttons[B - BUTTON_HID_BEGIN]->GetStatus());
-            state.x.Assign(buttons[X - BUTTON_HID_BEGIN]->GetStatus());
-            state.y.Assign(buttons[Y - BUTTON_HID_BEGIN]->GetStatus());
-            state.lstick.Assign(buttons[LStick - BUTTON_HID_BEGIN]->GetStatus());
-            state.rstick.Assign(buttons[RStick - BUTTON_HID_BEGIN]->GetStatus());
-            state.l.Assign(buttons[L - BUTTON_HID_BEGIN]->GetStatus());
-            state.r.Assign(buttons[R - BUTTON_HID_BEGIN]->GetStatus());
-            state.zl.Assign(buttons[ZL - BUTTON_HID_BEGIN]->GetStatus());
-            state.zr.Assign(buttons[ZR - BUTTON_HID_BEGIN]->GetStatus());
-            state.plus.Assign(buttons[Plus - BUTTON_HID_BEGIN]->GetStatus());
-            state.minus.Assign(buttons[Minus - BUTTON_HID_BEGIN]->GetStatus());
+                // TODO(shinyquagsire23): More than just handheld input
+                if (controller != Controller_Handheld)
+                    continue;
 
-            state.dleft.Assign(buttons[DLeft - BUTTON_HID_BEGIN]->GetStatus());
-            state.dup.Assign(buttons[DUp - BUTTON_HID_BEGIN]->GetStatus());
-            state.dright.Assign(buttons[DRight - BUTTON_HID_BEGIN]->GetStatus());
-            state.ddown.Assign(buttons[DDown - BUTTON_HID_BEGIN]->GetStatus());
+                // TODO(shinyquagsire23): Set up some LUTs for each layout mapping in the future?
+                // For now everything is just the default handheld layout, but split Joy-Con will
+                // rotate the face buttons and directions for certain layouts.
+                ControllerPadState& state = entry.buttons;
+                using namespace Settings::NativeButton;
+                state.a.Assign(buttons[A - BUTTON_HID_BEGIN]->GetStatus());
+                state.b.Assign(buttons[B - BUTTON_HID_BEGIN]->GetStatus());
+                state.x.Assign(buttons[X - BUTTON_HID_BEGIN]->GetStatus());
+                state.y.Assign(buttons[Y - BUTTON_HID_BEGIN]->GetStatus());
+                state.lstick.Assign(buttons[LStick - BUTTON_HID_BEGIN]->GetStatus());
+                state.rstick.Assign(buttons[RStick - BUTTON_HID_BEGIN]->GetStatus());
+                state.l.Assign(buttons[L - BUTTON_HID_BEGIN]->GetStatus());
+                state.r.Assign(buttons[R - BUTTON_HID_BEGIN]->GetStatus());
+                state.zl.Assign(buttons[ZL - BUTTON_HID_BEGIN]->GetStatus());
+                state.zr.Assign(buttons[ZR - BUTTON_HID_BEGIN]->GetStatus());
+                state.plus.Assign(buttons[Plus - BUTTON_HID_BEGIN]->GetStatus());
+                state.minus.Assign(buttons[Minus - BUTTON_HID_BEGIN]->GetStatus());
 
-            state.lstick_left.Assign(buttons[LStick_Left - BUTTON_HID_BEGIN]->GetStatus());
-            state.lstick_up.Assign(buttons[LStick_Up - BUTTON_HID_BEGIN]->GetStatus());
-            state.lstick_right.Assign(buttons[LStick_Right - BUTTON_HID_BEGIN]->GetStatus());
-            state.lstick_down.Assign(buttons[LStick_Down - BUTTON_HID_BEGIN]->GetStatus());
+                state.dleft.Assign(buttons[DLeft - BUTTON_HID_BEGIN]->GetStatus());
+                state.dup.Assign(buttons[DUp - BUTTON_HID_BEGIN]->GetStatus());
+                state.dright.Assign(buttons[DRight - BUTTON_HID_BEGIN]->GetStatus());
+                state.ddown.Assign(buttons[DDown - BUTTON_HID_BEGIN]->GetStatus());
 
-            state.rstick_left.Assign(buttons[RStick_Left - BUTTON_HID_BEGIN]->GetStatus());
-            state.rstick_up.Assign(buttons[RStick_Up - BUTTON_HID_BEGIN]->GetStatus());
-            state.rstick_right.Assign(buttons[RStick_Right - BUTTON_HID_BEGIN]->GetStatus());
-            state.rstick_down.Assign(buttons[RStick_Down - BUTTON_HID_BEGIN]->GetStatus());
+                state.lstick_left.Assign(buttons[LStick_Left - BUTTON_HID_BEGIN]->GetStatus());
+                state.lstick_up.Assign(buttons[LStick_Up - BUTTON_HID_BEGIN]->GetStatus());
+                state.lstick_right.Assign(buttons[LStick_Right - BUTTON_HID_BEGIN]->GetStatus());
+                state.lstick_down.Assign(buttons[LStick_Down - BUTTON_HID_BEGIN]->GetStatus());
 
-            state.sl.Assign(buttons[SL - BUTTON_HID_BEGIN]->GetStatus());
-            state.sr.Assign(buttons[SR - BUTTON_HID_BEGIN]->GetStatus());
+                state.rstick_left.Assign(buttons[RStick_Left - BUTTON_HID_BEGIN]->GetStatus());
+                state.rstick_up.Assign(buttons[RStick_Up - BUTTON_HID_BEGIN]->GetStatus());
+                state.rstick_right.Assign(buttons[RStick_Right - BUTTON_HID_BEGIN]->GetStatus());
+                state.rstick_down.Assign(buttons[RStick_Down - BUTTON_HID_BEGIN]->GetStatus());
 
-            // TODO(shinyquagsire23): Analog stick vals
+                state.sl.Assign(buttons[SL - BUTTON_HID_BEGIN]->GetStatus());
+                state.sr.Assign(buttons[SR - BUTTON_HID_BEGIN]->GetStatus());
 
-            // TODO(shinyquagsire23): Update pad info proper, (circular buffers, timestamps,
-            // layouts)
+                const auto [stick_l_x_f, stick_l_y_f] = sticks[Joystick_Left]->GetStatus();
+                const auto [stick_r_x_f, stick_r_y_f] = sticks[Joystick_Right]->GetStatus();
+                entry.joystick_left_x = static_cast<s32>(stick_l_x_f * HID_JOYSTICK_MAX);
+                entry.joystick_left_y = static_cast<s32>(stick_l_y_f * HID_JOYSTICK_MAX);
+                entry.joystick_right_x = static_cast<s32>(stick_r_x_f * HID_JOYSTICK_MAX);
+                entry.joystick_right_y = static_cast<s32>(stick_r_y_f * HID_JOYSTICK_MAX);
+            }
         }
 
         // TODO(bunnei): Properly implement the touch screen, the below will just write empty data
@@ -151,6 +163,71 @@ private:
         touchscreen.entries[curr_entry].header.timestamp = sample_counter;
         touchscreen.entries[curr_entry].header.num_touches = 0;
 
+        // TODO(shinyquagsire23): Properly implement mouse
+        Mouse& mouse = mem.mouse;
+        const u64 last_mouse_entry = mouse.header.latest_entry;
+        const u64 curr_mouse_entry = (mouse.header.latest_entry + 1) % mouse.entries.size();
+        const u64 mouse_sample_counter = mouse.entries[last_mouse_entry].timestamp + 1;
+        mouse.header.timestamp_ticks = timestamp;
+        mouse.header.num_entries = mouse.entries.size();
+        mouse.header.max_entry_index = mouse.entries.size();
+        mouse.header.latest_entry = curr_mouse_entry;
+
+        mouse.entries[curr_mouse_entry].timestamp = mouse_sample_counter;
+        mouse.entries[curr_mouse_entry].timestamp_2 = mouse_sample_counter;
+
+        // TODO(shinyquagsire23): Properly implement keyboard
+        Keyboard& keyboard = mem.keyboard;
+        const u64 last_keyboard_entry = keyboard.header.latest_entry;
+        const u64 curr_keyboard_entry =
+            (keyboard.header.latest_entry + 1) % keyboard.entries.size();
+        const u64 keyboard_sample_counter = keyboard.entries[last_keyboard_entry].timestamp + 1;
+        keyboard.header.timestamp_ticks = timestamp;
+        keyboard.header.num_entries = keyboard.entries.size();
+        keyboard.header.latest_entry = last_keyboard_entry;
+        keyboard.header.max_entry_index = keyboard.entries.size();
+
+        keyboard.entries[curr_keyboard_entry].timestamp = keyboard_sample_counter;
+        keyboard.entries[curr_keyboard_entry].timestamp_2 = keyboard_sample_counter;
+
+        // TODO(shinyquagsire23): Figure out what any of these are
+        for (size_t i = 0; i < mem.unk_input_1.size(); i++) {
+            UnkInput1& input = mem.unk_input_1[i];
+            const u64 last_input_entry = input.header.latest_entry;
+            const u64 curr_input_entry = (input.header.latest_entry + 1) % input.entries.size();
+            const u64 input_sample_counter = input.entries[last_input_entry].timestamp + 1;
+
+            input.header.timestamp_ticks = timestamp;
+            input.header.num_entries = input.entries.size();
+            input.header.latest_entry = last_input_entry;
+            input.header.max_entry_index = input.entries.size();
+
+            input.entries[curr_input_entry].timestamp = input_sample_counter;
+            input.entries[curr_input_entry].timestamp_2 = input_sample_counter;
+        }
+
+        for (size_t i = 0; i < mem.unk_input_2.size(); i++) {
+            UnkInput2& input = mem.unk_input_2[i];
+
+            input.header.timestamp_ticks = timestamp;
+            input.header.num_entries = 17;
+            input.header.latest_entry = 0;
+            input.header.max_entry_index = 0;
+        }
+
+        UnkInput3& input = mem.unk_input_3;
+        const u64 last_input_entry = input.header.latest_entry;
+        const u64 curr_input_entry = (input.header.latest_entry + 1) % input.entries.size();
+        const u64 input_sample_counter = input.entries[last_input_entry].timestamp + 1;
+
+        input.header.timestamp_ticks = timestamp;
+        input.header.num_entries = input.entries.size();
+        input.header.latest_entry = last_input_entry;
+        input.header.max_entry_index = input.entries.size();
+
+        input.entries[curr_input_entry].timestamp = input_sample_counter;
+        input.entries[curr_input_entry].timestamp_2 = input_sample_counter;
+
         // TODO(shinyquagsire23): Signal events
 
         std::memcpy(shared_mem->GetPointer(), &mem, sizeof(SharedMemory));
@@ -169,6 +246,7 @@ private:
     std::atomic<bool> is_device_reload_pending{true};
     std::array<std::unique_ptr<Input::ButtonDevice>, Settings::NativeButton::NUM_BUTTONS_HID>
         buttons;
+    std::array<std::unique_ptr<Input::AnalogDevice>, Settings::NativeAnalog::NUM_STICKS_HID> sticks;
 };
 
 class IActiveVibrationDeviceList final : public ServiceFramework<IActiveVibrationDeviceList> {

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

@@ -48,6 +48,11 @@ enum ControllerConnectionState {
     ConnectionState_Wired = 1 << 1,
 };
 
+enum ControllerJoystick {
+    Joystick_Left = 0,
+    Joystick_Right = 1,
+};
+
 enum ControllerID {
     Controller_Player1 = 0,
     Controller_Player2 = 1,
@@ -63,6 +68,34 @@ enum ControllerID {
 
 // End enums and output structs
 
+// Begin UnkInput3
+
+struct UnkInput3Header {
+    u64 timestamp_ticks;
+    u64 num_entries;
+    u64 latest_entry;
+    u64 max_entry_index;
+};
+static_assert(sizeof(UnkInput3Header) == 0x20, "HID UnkInput3 header structure has incorrect size");
+
+struct UnkInput3Entry {
+    u64 timestamp;
+    u64 timestamp_2;
+    u64 unk_8;
+    u64 unk_10;
+    u64 unk_18;
+};
+static_assert(sizeof(UnkInput3Entry) == 0x28, "HID UnkInput3 entry structure has incorrect size");
+
+struct UnkInput3 {
+    UnkInput3Header header;
+    std::array<UnkInput3Entry, 17> entries;
+    std::array<u8, 0x138> padding;
+};
+static_assert(sizeof(UnkInput3) == 0x400, "HID UnkInput3 structure has incorrect size");
+
+// End UnkInput3
+
 // Begin TouchScreen
 
 struct TouchScreenHeader {
@@ -204,6 +237,52 @@ static_assert(sizeof(Keyboard) == 0x400, "HID keyboard structure has incorrect s
 
 // End Keyboard
 
+// Begin UnkInput1
+
+struct UnkInput1Header {
+    u64 timestamp_ticks;
+    u64 num_entries;
+    u64 latest_entry;
+    u64 max_entry_index;
+};
+static_assert(sizeof(UnkInput1Header) == 0x20, "HID UnkInput1 header structure has incorrect size");
+
+struct UnkInput1Entry {
+    u64 timestamp;
+    u64 timestamp_2;
+    u64 unk_8;
+    u64 unk_10;
+    u64 unk_18;
+};
+static_assert(sizeof(UnkInput1Entry) == 0x28, "HID UnkInput1 entry structure has incorrect size");
+
+struct UnkInput1 {
+    UnkInput1Header header;
+    std::array<UnkInput1Entry, 17> entries;
+    std::array<u8, 0x138> padding;
+};
+static_assert(sizeof(UnkInput1) == 0x400, "HID UnkInput1 structure has incorrect size");
+
+// End UnkInput1
+
+// Begin UnkInput2
+
+struct UnkInput2Header {
+    u64 timestamp_ticks;
+    u64 num_entries;
+    u64 latest_entry;
+    u64 max_entry_index;
+};
+static_assert(sizeof(UnkInput2Header) == 0x20, "HID UnkInput2 header structure has incorrect size");
+
+struct UnkInput2 {
+    UnkInput2Header header;
+    std::array<u8, 0x1E0> padding;
+};
+static_assert(sizeof(UnkInput2) == 0x200, "HID UnkInput2 structure has incorrect size");
+
+// End UnkInput2
+
 // Begin Controller
 
 struct ControllerMAC {
@@ -283,10 +362,10 @@ struct ControllerInputEntry {
     u64 timestamp;
     u64 timestamp_2;
     ControllerPadState buttons;
-    u32 joystick_left_x;
-    u32 joystick_left_y;
-    u32 joystick_right_x;
-    u32 joystick_right_y;
+    s32 joystick_left_x;
+    s32 joystick_left_y;
+    s32 joystick_right_x;
+    s32 joystick_right_y;
     u64 connection_state;
 };
 static_assert(sizeof(ControllerInputEntry) == 0x30,
@@ -312,17 +391,12 @@ static_assert(sizeof(Controller) == 0x5000, "HID controller structure has incorr
 // End Controller
 
 struct SharedMemory {
-    std::array<u8, 0x400> header;
+    UnkInput3 unk_input_3;
     TouchScreen touchscreen;
     Mouse mouse;
     Keyboard keyboard;
-    std::array<u8, 0x400> unk_section_1;
-    std::array<u8, 0x400> unk_section_2;
-    std::array<u8, 0x400> unk_section_3;
-    std::array<u8, 0x400> unk_section_4;
-    std::array<u8, 0x200> unk_section_5;
-    std::array<u8, 0x200> unk_section_6;
-    std::array<u8, 0x200> unk_section_7;
+    std::array<UnkInput1, 4> unk_input_1;
+    std::array<UnkInput2, 3> unk_input_2;
     std::array<u8, 0x800> unk_section_8;
     std::array<u8, 0x4000> controller_serials;
     std::array<Controller, 10> controllers;

src/core/hle/service/ns/pl_u.cpp

@@ -37,7 +37,7 @@ PL_U::PL_U() : ServiceFramework("pl:u") {
         {2, &PL_U::GetSize, "GetSize"},
         {3, &PL_U::GetSharedMemoryAddressOffset, "GetSharedMemoryAddressOffset"},
         {4, &PL_U::GetSharedMemoryNativeHandle, "GetSharedMemoryNativeHandle"},
-        {5, nullptr, "GetSharedFontInOrderOfPriority"},
+        {5, &PL_U::GetSharedFontInOrderOfPriority, "GetSharedFontInOrderOfPriority"},
     };
     RegisterHandlers(functions);
 
@@ -116,4 +116,29 @@ void PL_U::GetSharedMemoryNativeHandle(Kernel::HLERequestContext& ctx) {
     rb.PushCopyObjects(shared_font_mem);
 }
 
+void PL_U::GetSharedFontInOrderOfPriority(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    const u64 language_code{rp.Pop<u64>()}; // TODO(ogniK): Find out what this is used for
+    NGLOG_DEBUG(Service_NS, "called, language_code=%lx", language_code);
+    IPC::ResponseBuilder rb{ctx, 4};
+    std::vector<u32> font_codes;
+    std::vector<u32> font_offsets;
+    std::vector<u32> font_sizes;
+
+    // TODO(ogniK): Have actual priority order
+    for (size_t i = 0; i < SHARED_FONT_REGIONS.size(); i++) {
+        font_codes.push_back(static_cast<u32>(i));
+        font_offsets.push_back(SHARED_FONT_REGIONS[i].offset);
+        font_sizes.push_back(SHARED_FONT_REGIONS[i].size);
+    }
+
+    ctx.WriteBuffer(font_codes.data(), font_codes.size(), 0);
+    ctx.WriteBuffer(font_offsets.data(), font_offsets.size(), 1);
+    ctx.WriteBuffer(font_sizes.data(), font_sizes.size(), 2);
+
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<u8>(static_cast<u8>(LoadState::Done)); // Fonts Loaded
+    rb.Push<u32>(static_cast<u32>(font_codes.size()));
+}
+
 } // namespace Service::NS

src/core/hle/service/ns/pl_u.h

@@ -21,6 +21,7 @@ private:
     void GetSize(Kernel::HLERequestContext& ctx);
     void GetSharedMemoryAddressOffset(Kernel::HLERequestContext& ctx);
     void GetSharedMemoryNativeHandle(Kernel::HLERequestContext& ctx);
+    void GetSharedFontInOrderOfPriority(Kernel::HLERequestContext& ctx);
 
     /// Handle to shared memory region designated for a shared font
     Kernel::SharedPtr<Kernel::SharedMemory> shared_font_mem;

src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp

@@ -12,7 +12,7 @@
 namespace Service::Nvidia::Devices {
 
 u32 nvhost_as_gpu::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
-    NGLOG_DEBUG(Service_NVDRV, "called, command={:#010X}, input_size={:#X}, output_size={:#X}",
+    NGLOG_DEBUG(Service_NVDRV, "called, command=0x{:08X}, input_size=0x{:X}, output_size=0x{:X}",
                 command.raw, input.size(), output.size());
 
     switch (static_cast<IoctlCommand>(command.raw)) {
@@ -26,6 +26,8 @@ u32 nvhost_as_gpu::ioctl(Ioctl command, const std::vector<u8>& input, std::vecto
         return BindChannel(input, output);
     case IoctlCommand::IocGetVaRegionsCommand:
         return GetVARegions(input, output);
+    case IoctlCommand::IocUnmapBufferCommand:
+        return UnmapBuffer(input, output);
     }
 
     if (static_cast<IoctlCommand>(command.cmd.Value()) == IoctlCommand::IocRemapCommand)
@@ -38,8 +40,7 @@ u32 nvhost_as_gpu::ioctl(Ioctl command, const std::vector<u8>& input, std::vecto
 u32 nvhost_as_gpu::InitalizeEx(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlInitalizeEx params{};
     std::memcpy(&params, input.data(), input.size());
-    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, big_page_size={:#X}", params.big_page_size);
-    std::memcpy(output.data(), &params, output.size());
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, big_page_size=0x{:X}", params.big_page_size);
     return 0;
 }
 
@@ -126,6 +127,37 @@ u32 nvhost_as_gpu::MapBufferEx(const std::vector<u8>& input, std::vector<u8>& ou
         params.offset = gpu.memory_manager->MapBufferEx(object->addr, object->size);
     }
 
+    // Create a new mapping entry for this operation.
+    ASSERT_MSG(buffer_mappings.find(params.offset) == buffer_mappings.end(),
+               "Offset is already mapped");
+
+    BufferMapping mapping{};
+    mapping.nvmap_handle = params.nvmap_handle;
+    mapping.offset = params.offset;
+    mapping.size = object->size;
+
+    buffer_mappings[params.offset] = mapping;
+
+    std::memcpy(output.data(), &params, output.size());
+    return 0;
+}
+
+u32 nvhost_as_gpu::UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& output) {
+    IoctlUnmapBuffer params{};
+    std::memcpy(&params, input.data(), input.size());
+
+    NGLOG_DEBUG(Service_NVDRV, "called, offset=0x{:X}", params.offset);
+
+    auto& gpu = Core::System::GetInstance().GPU();
+
+    auto itr = buffer_mappings.find(params.offset);
+
+    ASSERT_MSG(itr != buffer_mappings.end(), "Tried to unmap invalid mapping");
+
+    params.offset = gpu.memory_manager->UnmapBuffer(params.offset, itr->second.size);
+
+    buffer_mappings.erase(itr->second.offset);
+
     std::memcpy(output.data(), &params, output.size());
     return 0;
 }
@@ -135,7 +167,6 @@ u32 nvhost_as_gpu::BindChannel(const std::vector<u8>& input, std::vector<u8>& ou
     std::memcpy(&params, input.data(), input.size());
     NGLOG_DEBUG(Service_NVDRV, "called, fd={:X}", params.fd);
     channel = params.fd;
-    std::memcpy(output.data(), &params, output.size());
     return 0;
 }
 

src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <memory>
+#include <unordered_map>
 #include <utility>
 #include <vector>
 #include "common/common_types.h"
@@ -30,6 +31,7 @@ private:
         IocMapBufferExCommand = 0xC0284106,
         IocBindChannelCommand = 0x40044101,
         IocGetVaRegionsCommand = 0xC0404108,
+        IocUnmapBufferCommand = 0xC0084105,
     };
 
     struct IoctlInitalizeEx {
@@ -76,6 +78,11 @@ private:
     };
     static_assert(sizeof(IoctlMapBufferEx) == 40, "IoctlMapBufferEx is incorrect size");
 
+    struct IoctlUnmapBuffer {
+        u64_le offset;
+    };
+    static_assert(sizeof(IoctlUnmapBuffer) == 8, "IoctlUnmapBuffer is incorrect size");
+
     struct IoctlBindChannel {
         u32_le fd;
     };
@@ -98,12 +105,22 @@ private:
     static_assert(sizeof(IoctlGetVaRegions) == 16 + sizeof(IoctlVaRegion) * 2,
                   "IoctlGetVaRegions is incorrect size");
 
+    struct BufferMapping {
+        u64 offset;
+        u64 size;
+        u32 nvmap_handle;
+    };
+
+    /// Map containing the nvmap object mappings in GPU memory.
+    std::unordered_map<u64, BufferMapping> buffer_mappings;
+
     u32 channel{};
 
     u32 InitalizeEx(const std::vector<u8>& input, std::vector<u8>& output);
     u32 AllocateSpace(const std::vector<u8>& input, std::vector<u8>& output);
     u32 Remap(const std::vector<u8>& input, std::vector<u8>& output);
     u32 MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& output);
     u32 BindChannel(const std::vector<u8>& input, std::vector<u8>& output);
     u32 GetVARegions(const std::vector<u8>& input, std::vector<u8>& output);
 

src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp

@@ -9,7 +9,7 @@
 namespace Service::Nvidia::Devices {
 
 u32 nvhost_ctrl::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
-    NGLOG_DEBUG(Service_NVDRV, "called, command={:#010X}, input_size={:#X}, output_size={:#X}",
+    NGLOG_DEBUG(Service_NVDRV, "called, command=0x{:08X}, input_size=0x{:X}, output_size=0x{:X}",
                 command.raw, input.size(), output.size());
 
     switch (static_cast<IoctlCommand>(command.raw)) {

src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp

@@ -10,7 +10,7 @@
 namespace Service::Nvidia::Devices {
 
 u32 nvhost_ctrl_gpu::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
-    NGLOG_DEBUG(Service_NVDRV, "called, command={:#010X}, input_size={:#X}, output_size={:#X}",
+    NGLOG_DEBUG(Service_NVDRV, "called, command=0x{:08X}, input_size=0x{:X}, output_size=0x{:X}",
                 command.raw, input.size(), output.size());
 
     switch (static_cast<IoctlCommand>(command.raw)) {
@@ -77,9 +77,13 @@ u32 nvhost_ctrl_gpu::GetCharacteristics(const std::vector<u8>& input, std::vecto
 u32 nvhost_ctrl_gpu::GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output) {
     IoctlGpuGetTpcMasksArgs params{};
     std::memcpy(&params, input.data(), input.size());
-    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, mask={:#X}, mask_buf_addr={:#X}",
-                  params.mask_buf_size, params.mask_buf_addr);
-    params.unk = 0xcafe; // TODO(ogniK): Needs to be non 0, what does this actually do?
+    NGLOG_INFO(Service_NVDRV, "called, mask=0x{:X}, mask_buf_addr=0x{:X}", params.mask_buf_size,
+               params.mask_buf_addr);
+    // TODO(ogniK): Confirm value on hardware
+    if (params.mask_buf_size)
+        params.tpc_mask_size = 4 * 1; // 4 * num_gpc
+    else
+        params.tpc_mask_size = 0;
     std::memcpy(output.data(), &params, sizeof(params));
     return 0;
 }

src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h

@@ -86,7 +86,7 @@ private:
         /// [in]  pointer to TPC mask buffer. It will receive one 32-bit TPC mask per GPC or 0 if
         /// GPC is not enabled or not present. This parameter is ignored if mask_buf_size is 0.
         u64_le mask_buf_addr;
-        u64_le unk; // Nintendo add this?
+        u64_le tpc_mask_size; // Nintendo add this?
     };
     static_assert(sizeof(IoctlGpuGetTpcMasksArgs) == 24,
                   "IoctlGpuGetTpcMasksArgs is incorrect size");

src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp

@@ -12,7 +12,7 @@
 namespace Service::Nvidia::Devices {
 
 u32 nvhost_gpu::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) {
-    NGLOG_DEBUG(Service_NVDRV, "called, command={:#010X}, input_size={:#X}, output_size={:#X}",
+    NGLOG_DEBUG(Service_NVDRV, "called, command=0x{:08X}, input_size=0x{:X}, output_size=0x{:X}",
                 command.raw, input.size(), output.size());
 
     switch (static_cast<IoctlCommand>(command.raw)) {
@@ -49,7 +49,6 @@ u32 nvhost_gpu::SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output
     std::memcpy(&params, input.data(), input.size());
     NGLOG_DEBUG(Service_NVDRV, "called, fd={}", params.nvmap_fd);
     nvmap_fd = params.nvmap_fd;
-    std::memcpy(output.data(), &params, output.size());
     return 0;
 }
 
@@ -58,7 +57,6 @@ u32 nvhost_gpu::SetClientData(const std::vector<u8>& input, std::vector<u8>& out
     IoctlClientData params{};
     std::memcpy(&params, input.data(), input.size());
     user_data = params.data;
-    std::memcpy(output.data(), &params, output.size());
     return 0;
 }
 
@@ -91,7 +89,6 @@ u32 nvhost_gpu::SetErrorNotifier(const std::vector<u8>& input, std::vector<u8>&
 u32 nvhost_gpu::SetChannelPriority(const std::vector<u8>& input, std::vector<u8>& output) {
     std::memcpy(&channel_priority, input.data(), input.size());
     NGLOG_DEBUG(Service_NVDRV, "(STUBBED) called, priority={:X}", channel_priority);
-    std::memcpy(output.data(), &channel_priority, output.size());
     return 0;
 }
 

src/core/hle/service/nvdrv/devices/nvmap.cpp

@@ -30,6 +30,8 @@ u32 nvmap::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& o
         return IocFromId(input, output);
     case IoctlCommand::Param:
         return IocParam(input, output);
+    case IoctlCommand::Free:
+        return IocFree(input, output);
     }
 
     UNIMPLEMENTED_MSG("Unimplemented ioctl");
@@ -45,11 +47,12 @@ u32 nvmap::IocCreate(const std::vector<u8>& input, std::vector<u8>& output) {
     object->id = next_id++;
     object->size = params.size;
     object->status = Object::Status::Created;
+    object->refcount = 1;
 
     u32 handle = next_handle++;
     handles[handle] = std::move(object);
 
-    NGLOG_DEBUG(Service_NVDRV, "size={:#010X}", params.size);
+    NGLOG_DEBUG(Service_NVDRV, "size=0x{:08X}", params.size);
 
     params.handle = handle;
 
@@ -101,6 +104,8 @@ u32 nvmap::IocFromId(const std::vector<u8>& input, std::vector<u8>& output) {
                             [&](const auto& entry) { return entry.second->id == params.id; });
     ASSERT(itr != handles.end());
 
+    itr->second->refcount++;
+
     // Return the existing handle instead of creating a new one.
     params.handle = itr->first;
 
@@ -142,4 +147,34 @@ u32 nvmap::IocParam(const std::vector<u8>& input, std::vector<u8>& output) {
     return 0;
 }
 
+u32 nvmap::IocFree(const std::vector<u8>& input, std::vector<u8>& output) {
+    enum FreeFlags {
+        Freed = 0,
+        NotFreedYet = 1,
+    };
+
+    IocFreeParams params;
+    std::memcpy(&params, input.data(), sizeof(params));
+
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called");
+
+    auto itr = handles.find(params.handle);
+    ASSERT(itr != handles.end());
+
+    itr->second->refcount--;
+
+    params.refcount = itr->second->refcount;
+    params.size = itr->second->size;
+
+    if (itr->second->refcount == 0)
+        params.flags = Freed;
+    else
+        params.flags = NotFreedYet;
+
+    handles.erase(params.handle);
+
+    std::memcpy(output.data(), &params, sizeof(params));
+    return 0;
+}
+
 } // namespace Service::Nvidia::Devices

src/core/hle/service/nvdrv/devices/nvmap.h

@@ -34,6 +34,7 @@ public:
         u8 kind;
         VAddr addr;
         Status status;
+        u32 refcount;
     };
 
     std::shared_ptr<Object> GetObject(u32 handle) const {
@@ -59,7 +60,8 @@ private:
         FromId = 0xC0080103,
         Alloc = 0xC0200104,
         Param = 0xC00C0109,
-        GetId = 0xC008010E
+        GetId = 0xC008010E,
+        Free = 0xC0180105,
     };
 
     struct IocCreateParams {
@@ -102,11 +104,21 @@ private:
         u32_le value;
     };
 
+    struct IocFreeParams {
+        u32_le handle;
+        INSERT_PADDING_BYTES(4);
+        u64_le refcount;
+        u32_le size;
+        u32_le flags;
+    };
+    static_assert(sizeof(IocFreeParams) == 24, "IocFreeParams has wrong size");
+
     u32 IocCreate(const std::vector<u8>& input, std::vector<u8>& output);
     u32 IocAlloc(const std::vector<u8>& input, std::vector<u8>& output);
     u32 IocGetId(const std::vector<u8>& input, std::vector<u8>& output);
     u32 IocFromId(const std::vector<u8>& input, std::vector<u8>& output);
     u32 IocParam(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 IocFree(const std::vector<u8>& input, std::vector<u8>& output);
 };
 
 } // namespace Service::Nvidia::Devices

src/core/hle/service/nvdrv/interface.cpp

@@ -75,7 +75,7 @@ void NVDRV::SetClientPID(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     pid = rp.Pop<u64>();
 
-    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, pid={:#X}", pid);
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, pid=0x{:X}", pid);
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
     rb.Push<u32>(0);

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

@@ -19,7 +19,7 @@
 namespace Service::NVFlinger {
 
 constexpr size_t SCREEN_REFRESH_RATE = 60;
-constexpr u64 frame_ticks = static_cast<u64>(BASE_CLOCK_RATE / SCREEN_REFRESH_RATE);
+constexpr u64 frame_ticks = static_cast<u64>(CoreTiming::BASE_CLOCK_RATE / SCREEN_REFRESH_RATE);
 
 NVFlinger::NVFlinger() {
     // Add the different displays to the list of displays.

src/core/hle/service/service.cpp

@@ -58,10 +58,9 @@ static std::string MakeFunctionString(const char* name, const char* port_name,
     // Number of params == bits 0-5 + bits 6-11
     int num_params = (cmd_buff[0] & 0x3F) + ((cmd_buff[0] >> 6) & 0x3F);
 
-    std::string function_string =
-        Common::StringFromFormat("function '%s': port=%s", name, port_name);
+    std::string function_string = fmt::format("function '{}': port={}", name, port_name);
     for (int i = 1; i <= num_params; ++i) {
-        function_string += Common::StringFromFormat(", cmd_buff[%i]=0x%X", i, cmd_buff[i]);
+        function_string += fmt::format(", cmd_buff[{}]=0x{:X}", i, cmd_buff[i]);
     }
     return function_string;
 }
@@ -114,10 +113,10 @@ void ServiceFrameworkBase::ReportUnimplementedFunction(Kernel::HLERequestContext
     std::string function_name = info == nullptr ? fmt::format("{}", ctx.GetCommand()) : info->name;
 
     fmt::memory_buffer buf;
-    fmt::format_to(buf, "function '{}': port='{}' cmd_buf={{[0]={:#x}", function_name, service_name,
-                   cmd_buf[0]);
+    fmt::format_to(buf, "function '{}': port='{}' cmd_buf={{[0]=0x{:X}", function_name,
+                   service_name, cmd_buf[0]);
     for (int i = 1; i <= 8; ++i) {
-        fmt::format_to(buf, ", [{}]={:#x}", i, cmd_buf[i]);
+        fmt::format_to(buf, ", [{}]=0x{:X}", i, cmd_buf[i]);
     }
     buf.push_back('}');
 

src/core/hle/service/sm/sm.cpp

@@ -102,7 +102,7 @@ void SM::GetService(Kernel::HLERequestContext& ctx) {
     if (client_port.Failed()) {
         IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
         rb.Push(client_port.Code());
-        NGLOG_ERROR(Service_SM, "called service={} -> error {:#010X}", name,
+        NGLOG_ERROR(Service_SM, "called service={} -> error 0x{:08X}", name,
                     client_port.Code().raw);
         if (name.length() == 0)
             return; // LibNX Fix

src/core/hle/service/time/time.cpp

@@ -59,7 +59,8 @@ public:
 private:
     void GetCurrentTimePoint(Kernel::HLERequestContext& ctx) {
         NGLOG_DEBUG(Service_Time, "called");
-        SteadyClockTimePoint steady_clock_time_point{cyclesToMs(CoreTiming::GetTicks()) / 1000};
+        SteadyClockTimePoint steady_clock_time_point{
+            CoreTiming::cyclesToMs(CoreTiming::GetTicks()) / 1000};
         IPC::ResponseBuilder rb{ctx, (sizeof(SteadyClockTimePoint) / 4) + 2};
         rb.Push(RESULT_SUCCESS);
         rb.PushRaw(steady_clock_time_point);
@@ -110,7 +111,7 @@ private:
         IPC::RequestParser rp{ctx};
         u64 posix_time = rp.Pop<u64>();
 
-        NGLOG_WARNING(Service_Time, "(STUBBED) called, posix_time={:#018X}", posix_time);
+        NGLOG_WARNING(Service_Time, "(STUBBED) called, posix_time=0x{:016X}", posix_time);
 
         CalendarTime calendar_time{2018, 1, 1, 0, 0, 0};
         CalendarAdditionalInfo additional_info{};

src/core/hle/service/vi/vi.cpp

@@ -640,7 +640,7 @@ private:
         bool visibility = rp.Pop<bool>();
         IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
         rb.Push(RESULT_SUCCESS);
-        NGLOG_WARNING(Service_VI, "(STUBBED) called, layer_id={:#010X}, visibility={}", layer_id,
+        NGLOG_WARNING(Service_VI, "(STUBBED) called, layer_id=0x{:08X}, visibility={}", layer_id,
                       visibility);
     }
 };
@@ -762,7 +762,7 @@ private:
         bool visibility = rp.Pop<bool>();
         IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
         rb.Push(RESULT_SUCCESS);
-        NGLOG_WARNING(Service_VI, "(STUBBED) called, layer_id={:#X}, visibility={}", layer_id,
+        NGLOG_WARNING(Service_VI, "(STUBBED) called, layer_id=0x{:X}, visibility={}", layer_id,
                       visibility);
     }
 

src/core/hle/shared_page.cpp

@@ -56,13 +56,14 @@ static void UpdateTimeCallback(u64 userdata, int cycles_late) {
 
     date_time.date_time = GetSystemTime();
     date_time.update_tick = CoreTiming::GetTicks();
-    date_time.tick_to_second_coefficient = BASE_CLOCK_RATE;
+    date_time.tick_to_second_coefficient = CoreTiming::BASE_CLOCK_RATE;
     date_time.tick_offset = 0;
 
     ++shared_page.date_time_counter;
 
     // system time is updated hourly
-    CoreTiming::ScheduleEvent(msToCycles(60 * 60 * 1000) - cycles_late, update_time_event);
+    CoreTiming::ScheduleEvent(CoreTiming::msToCycles(60 * 60 * 1000) - cycles_late,
+                              update_time_event);
 }
 
 void Init() {

src/core/hw/hw.cpp

@@ -33,7 +33,7 @@ inline void Read(T& var, const u32 addr) {
         LCD::Read(var, addr);
         break;
     default:
-        NGLOG_ERROR(HW_Memory, "Unknown Read{} @ {:#010X}", sizeof(var) * 8, addr);
+        NGLOG_ERROR(HW_Memory, "Unknown Read{} @ 0x{:08X}", sizeof(var) * 8, addr);
         break;
     }
 }
@@ -62,7 +62,7 @@ inline void Write(u32 addr, const T data) {
         LCD::Write(addr, data);
         break;
     default:
-        NGLOG_ERROR(HW_Memory, "Unknown Write{} {:#010X} @ {:#010X}", sizeof(data) * 8, data, addr);
+        NGLOG_ERROR(HW_Memory, "Unknown Write{} 0x{:08X} @ 0x{:08X}", sizeof(data) * 8, data, addr);
         break;
     }
 }

src/core/hw/lcd.cpp

@@ -20,7 +20,7 @@ inline void Read(T& var, const u32 raw_addr) {
 
     // Reads other than u32 are untested, so I'd rather have them abort than silently fail
     if (index >= 0x400 || !std::is_same<T, u32>::value) {
-        NGLOG_ERROR(HW_LCD, "Unknown Read{} @ {:#010X}", sizeof(var) * 8, addr);
+        NGLOG_ERROR(HW_LCD, "Unknown Read{} @ 0x{:08X}", sizeof(var) * 8, addr);
         return;
     }
 
@@ -34,7 +34,7 @@ inline void Write(u32 addr, const T data) {
 
     // Writes other than u32 are untested, so I'd rather have them abort than silently fail
     if (index >= 0x400 || !std::is_same<T, u32>::value) {
-        NGLOG_ERROR(HW_LCD, "Unknown Write{} {:#010X} @ {:#010X}", sizeof(data) * 8, data, addr);
+        NGLOG_ERROR(HW_LCD, "Unknown Write{} 0x{:08X} @ 0x{:08X}", sizeof(data) * 8, data, addr);
         return;
     }
 

src/core/loader/deconstructed_rom_directory.cpp

@@ -132,7 +132,7 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(
         const VAddr load_addr = next_load_addr;
         next_load_addr = AppLoader_NSO::LoadModule(path, load_addr);
         if (next_load_addr) {
-            NGLOG_DEBUG(Loader, "loaded module {} @ {:#X}", module, load_addr);
+            NGLOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", module, load_addr);
         } else {
             next_load_addr = load_addr;
         }
@@ -176,8 +176,8 @@ ResultStatus AppLoader_DeconstructedRomDirectory::ReadRomFS(
     offset = 0;
     size = romfs_file->GetSize();
 
-    NGLOG_DEBUG(Loader, "RomFS offset:           {:#018X}", offset);
-    NGLOG_DEBUG(Loader, "RomFS size:             {:#018X}", size);
+    NGLOG_DEBUG(Loader, "RomFS offset:           0x{:016X}", offset);
+    NGLOG_DEBUG(Loader, "RomFS size:             0x{:016X}", size);
 
     // Reset read pointer
     file.Seek(0, SEEK_SET);

src/core/loader/nso.cpp

@@ -158,7 +158,7 @@ ResultStatus AppLoader_NSO::Load(Kernel::SharedPtr<Kernel::Process>& process) {
 
     // Load module
     LoadModule(filepath, Memory::PROCESS_IMAGE_VADDR);
-    NGLOG_DEBUG(Loader, "loaded module {} @ {:#X}", filepath, Memory::PROCESS_IMAGE_VADDR);
+    NGLOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", filepath, Memory::PROCESS_IMAGE_VADDR);
 
     process->svc_access_mask.set();
     process->address_mappings = default_address_mappings;

src/core/memory.cpp

@@ -28,8 +28,13 @@ static PageTable* current_page_table = nullptr;
 
 void SetCurrentPageTable(PageTable* page_table) {
     current_page_table = page_table;
-    if (Core::System::GetInstance().IsPoweredOn()) {
-        Core::CPU().PageTableChanged();
+
+    auto& system = Core::System::GetInstance();
+    if (system.IsPoweredOn()) {
+        system.ArmInterface(0).PageTableChanged();
+        system.ArmInterface(1).PageTableChanged();
+        system.ArmInterface(2).PageTableChanged();
+        system.ArmInterface(3).PageTableChanged();
     }
 }
 
@@ -168,7 +173,7 @@ T Read(const VAddr vaddr) {
     PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
     switch (type) {
     case PageType::Unmapped:
-        NGLOG_ERROR(HW_Memory, "Unmapped Read{} @ {:#010X}", sizeof(T) * 8, vaddr);
+        NGLOG_ERROR(HW_Memory, "Unmapped Read{} @ 0x{:08X}", sizeof(T) * 8, vaddr);
         return 0;
     case PageType::Memory:
         ASSERT_MSG(false, "Mapped memory page without a pointer @ {:016X}", vaddr);
@@ -200,8 +205,8 @@ void Write(const VAddr vaddr, const T data) {
     PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
     switch (type) {
     case PageType::Unmapped:
-        NGLOG_ERROR(HW_Memory, "Unmapped Write{} {:#010X} @ {:#018X}", sizeof(data) * 8, (u32)data,
-                    vaddr);
+        NGLOG_ERROR(HW_Memory, "Unmapped Write{} 0x{:08X} @ 0x{:016X}", sizeof(data) * 8,
+                    static_cast<u32>(data), vaddr);
         return;
     case PageType::Memory:
         ASSERT_MSG(false, "Mapped memory page without a pointer @ {:016X}", vaddr);
@@ -250,7 +255,7 @@ u8* GetPointer(const VAddr vaddr) {
         return GetPointerFromVMA(vaddr);
     }
 
-    NGLOG_ERROR(HW_Memory, "Unknown GetPointer @ {:#018X}", vaddr);
+    NGLOG_ERROR(HW_Memory, "Unknown GetPointer @ 0x{:016X}", vaddr);
     return nullptr;
 }
 
@@ -287,12 +292,12 @@ u8* GetPhysicalPointer(PAddr address) {
         });
 
     if (area == std::end(memory_areas)) {
-        NGLOG_ERROR(HW_Memory, "Unknown GetPhysicalPointer @ {:#018X}", address);
+        NGLOG_ERROR(HW_Memory, "Unknown GetPhysicalPointer @ 0x{:016X}", address);
         return nullptr;
     }
 
     if (area->paddr_base == IO_AREA_PADDR) {
-        NGLOG_ERROR(HW_Memory, "MMIO mappings are not supported yet. phys_addr={:018X}", address);
+        NGLOG_ERROR(HW_Memory, "MMIO mappings are not supported yet. phys_addr={:016X}", address);
         return nullptr;
     }
 
@@ -476,7 +481,7 @@ void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_
         switch (page_table.attributes[page_index]) {
         case PageType::Unmapped: {
             NGLOG_ERROR(HW_Memory,
-                        "Unmapped ReadBlock @ {:#018X} (start address = {:#018X}, size = {})",
+                        "Unmapped ReadBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
                         current_vaddr, src_addr, size);
             std::memset(dest_buffer, 0, copy_amount);
             break;
@@ -540,7 +545,7 @@ void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const voi
         switch (page_table.attributes[page_index]) {
         case PageType::Unmapped: {
             NGLOG_ERROR(HW_Memory,
-                        "Unmapped WriteBlock @ {:#018X} (start address = {:#018X}, size = {})",
+                        "Unmapped WriteBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
                         current_vaddr, dest_addr, size);
             break;
         }
@@ -588,7 +593,7 @@ void ZeroBlock(const Kernel::Process& process, const VAddr dest_addr, const size
         switch (page_table.attributes[page_index]) {
         case PageType::Unmapped: {
             NGLOG_ERROR(HW_Memory,
-                        "Unmapped ZeroBlock @ {:#018X} (start address = {#:018X}, size = {})",
+                        "Unmapped ZeroBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
                         current_vaddr, dest_addr, size);
             break;
         }
@@ -629,7 +634,7 @@ void CopyBlock(const Kernel::Process& process, VAddr dest_addr, VAddr src_addr,
         switch (page_table.attributes[page_index]) {
         case PageType::Unmapped: {
             NGLOG_ERROR(HW_Memory,
-                        "Unmapped CopyBlock @ {:#018X} (start address = {:#018X}, size = {})",
+                        "Unmapped CopyBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
                         current_vaddr, src_addr, size);
             ZeroBlock(process, dest_addr, copy_amount);
             break;
@@ -683,7 +688,7 @@ boost::optional<PAddr> TryVirtualToPhysicalAddress(const VAddr addr) {
 PAddr VirtualToPhysicalAddress(const VAddr addr) {
     auto paddr = TryVirtualToPhysicalAddress(addr);
     if (!paddr) {
-        NGLOG_ERROR(HW_Memory, "Unknown virtual address @ {:#018X}", addr);
+        NGLOG_ERROR(HW_Memory, "Unknown virtual address @ 0x{:016X}", addr);
         // To help with debugging, set bit on address so that it's obviously invalid.
         return addr | 0x80000000;
     }

src/core/memory_hook.cpp

@@ -0,0 +1,11 @@
+// Copyright 2018 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/memory_hook.h"
+
+namespace Memory {
+
+MemoryHook::~MemoryHook() = default;
+
+} // namespace Memory

src/core/memory_hook.h

@@ -23,7 +23,7 @@ namespace Memory {
  */
 class MemoryHook {
 public:
-    virtual ~MemoryHook() = default;
+    virtual ~MemoryHook();
 
     virtual boost::optional<bool> IsValidAddress(VAddr addr) = 0;
 

src/core/settings.h

@@ -99,6 +99,10 @@ enum Values {
     NumAnalogs,
 };
 
+constexpr int STICK_HID_BEGIN = LStick;
+constexpr int STICK_HID_END = NumAnalogs;
+constexpr int NUM_STICKS_HID = NumAnalogs;
+
 static const std::array<const char*, NumAnalogs> mapping = {{
     "lstick",
     "rstick",
@@ -117,6 +121,7 @@ struct Values {
 
     // Core
     bool use_cpu_jit;
+    bool use_multi_core;
 
     // Data Storage
     bool use_virtual_sd;

src/core/telemetry_session.cpp

@@ -155,6 +155,8 @@ TelemetrySession::TelemetrySession() {
 
     // Log user configuration information
     AddField(Telemetry::FieldType::UserConfig, "Core_UseCpuJit", Settings::values.use_cpu_jit);
+    AddField(Telemetry::FieldType::UserConfig, "Core_UseMultiCore",
+             Settings::values.use_multi_core);
     AddField(Telemetry::FieldType::UserConfig, "Renderer_ResolutionFactor",
              Settings::values.resolution_factor);
     AddField(Telemetry::FieldType::UserConfig, "Renderer_ToggleFramelimit",

src/video_core/engines/fermi_2d.cpp

@@ -59,12 +59,12 @@ void Fermi2D::HandleSurfaceCopy() {
         // If the input is tiled and the output is linear, deswizzle the input and copy it over.
         Texture::CopySwizzledData(regs.src.width, regs.src.height, src_bytes_per_pixel,
                                   dst_bytes_per_pixel, src_buffer, dst_buffer, true,
-                                  regs.src.block_height);
+                                  regs.src.BlockHeight());
     } else {
         // If the input is linear and the output is tiled, swizzle the input and copy it over.
         Texture::CopySwizzledData(regs.src.width, regs.src.height, src_bytes_per_pixel,
                                   dst_bytes_per_pixel, dst_buffer, src_buffer, false,
-                                  regs.dst.block_height);
+                                  regs.dst.BlockHeight());
     }
 }
 

src/video_core/engines/fermi_2d.h

@@ -49,6 +49,11 @@ public:
                 return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) |
                                              address_low);
             }
+
+            u32 BlockHeight() const {
+                // The block height is stored in log2 format.
+                return 1 << block_height;
+            }
         };
         static_assert(sizeof(Surface) == 0x28, "Surface has incorrect size");
 

src/video_core/engines/maxwell_3d.cpp

@@ -208,6 +208,16 @@ void Maxwell3D::DrawArrays() {
 
     const bool is_indexed{regs.index_array.count && !regs.vertex_buffer.count};
     VideoCore::g_renderer->Rasterizer()->AccelerateDrawBatch(is_indexed);
+
+    // TODO(bunnei): Below, we reset vertex count so that we can use these registers to determine if
+    // the game is trying to draw indexed or direct mode. This needs to be verified on HW still -
+    // it's possible that it is incorrect and that there is some other register used to specify the
+    // drawing mode.
+    if (is_indexed) {
+        regs.index_array.count = 0;
+    } else {
+        regs.vertex_buffer.count = 0;
+    }
 }
 
 void Maxwell3D::ProcessCBBind(Regs::ShaderStage stage) {

src/video_core/engines/shader_bytecode.h

@@ -25,6 +25,13 @@ struct Register {
     /// Register 255 is special cased to always be 0
     static constexpr size_t ZeroIndex = 255;
 
+    enum class Size : u64 {
+        Byte = 0,
+        Short = 1,
+        Word = 2,
+        Long = 3,
+    };
+
     constexpr Register() = default;
 
     constexpr Register(u64 value) : value(value) {}
@@ -236,6 +243,15 @@ union Instruction {
         BitField<56, 1, u64> neg_imm;
     } fset;
 
+    union {
+        BitField<10, 2, Register::Size> size;
+        BitField<13, 1, u64> is_signed;
+        BitField<41, 2, u64> selector;
+        BitField<45, 1, u64> negate_a;
+        BitField<49, 1, u64> abs_a;
+        BitField<50, 1, u64> saturate_a;
+    } conversion;
+
     BitField<61, 1, u64> is_b_imm;
     BitField<60, 1, u64> is_b_gpr;
     BitField<59, 1, u64> is_c_gpr;
@@ -290,10 +306,11 @@ public:
         MOV_C,
         MOV_R,
         MOV_IMM,
-        MOV32I,
+        MOV32_IMM,
         SHR_C,
         SHR_R,
         SHR_IMM,
+        FMNMX,
         FSETP_C, // Set Predicate
         FSETP_R,
         FSETP_IMM,
@@ -314,6 +331,7 @@ public:
         FloatSet,
         FloatSetPredicate,
         IntegerSetPredicate,
+        Conversion,
         Unknown,
     };
 
@@ -435,20 +453,21 @@ private:
             INST("0100110010110---", Id::F2I_C, Type::Arithmetic, "F2I_C"),
             INST("0101110010110---", Id::F2I_R, Type::Arithmetic, "F2I_R"),
             INST("0011100-10110---", Id::F2I_IMM, Type::Arithmetic, "F2I_IMM"),
-            INST("0100110010111---", Id::I2F_C, Type::Arithmetic, "I2F_C"),
-            INST("0101110010111---", Id::I2F_R, Type::Arithmetic, "I2F_R"),
-            INST("0011100-10111---", Id::I2F_IMM, Type::Arithmetic, "I2F_IMM"),
-            INST("0100110011100---", Id::I2I_C, Type::Arithmetic, "I2I_C"),
-            INST("0101110011100---", Id::I2I_R, Type::Arithmetic, "I2I_R"),
-            INST("01110001-1000---", Id::I2I_IMM, Type::Arithmetic, "I2I_IMM"),
             INST("000001----------", Id::LOP32I, Type::Arithmetic, "LOP32I"),
             INST("0100110010011---", Id::MOV_C, Type::Arithmetic, "MOV_C"),
             INST("0101110010011---", Id::MOV_R, Type::Arithmetic, "MOV_R"),
             INST("0011100-10011---", Id::MOV_IMM, Type::Arithmetic, "MOV_IMM"),
-            INST("000000010000----", Id::MOV32I, Type::Arithmetic, "MOV32I"),
+            INST("000000010000----", Id::MOV32_IMM, Type::Arithmetic, "MOV32_IMM"),
             INST("0100110000101---", Id::SHR_C, Type::Arithmetic, "SHR_C"),
             INST("0101110000101---", Id::SHR_R, Type::Arithmetic, "SHR_R"),
             INST("0011100-00101---", Id::SHR_IMM, Type::Arithmetic, "SHR_IMM"),
+            INST("0101110001100---", Id::FMNMX, Type::Arithmetic, "FMNMX"),
+            INST("0100110011100---", Id::I2I_C, Type::Conversion, "I2I_C"),
+            INST("0101110011100---", Id::I2I_R, Type::Conversion, "I2I_R"),
+            INST("01110001-1000---", Id::I2I_IMM, Type::Conversion, "I2I_IMM"),
+            INST("0100110010111---", Id::I2F_C, Type::Conversion, "I2F_C"),
+            INST("0101110010111---", Id::I2F_R, Type::Conversion, "I2F_R"),
+            INST("0011100-10111---", Id::I2F_IMM, Type::Conversion, "I2F_IMM"),
             INST("01011000--------", Id::FSET_R, Type::FloatSet, "FSET_R"),
             INST("0100100---------", Id::FSET_C, Type::FloatSet, "FSET_C"),
             INST("0011000---------", Id::FSET_IMM, Type::FloatSet, "FSET_IMM"),

src/video_core/memory_manager.cpp

@@ -58,6 +58,25 @@ GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, GPUVAddr gpu_addr, u64 size)
     return gpu_addr;
 }
 
+GPUVAddr MemoryManager::UnmapBuffer(GPUVAddr gpu_addr, u64 size) {
+    ASSERT((gpu_addr & PAGE_MASK) == 0);
+
+    for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
+        ASSERT(PageSlot(gpu_addr + offset) != static_cast<u64>(PageStatus::Allocated) &&
+               PageSlot(gpu_addr + offset) != static_cast<u64>(PageStatus::Unmapped));
+        PageSlot(gpu_addr + offset) = static_cast<u64>(PageStatus::Unmapped);
+    }
+
+    // Delete the region mappings that are contained within the unmapped region
+    mapped_regions.erase(std::remove_if(mapped_regions.begin(), mapped_regions.end(),
+                                        [&](const MappedRegion& region) {
+                                            return region.gpu_addr <= gpu_addr &&
+                                                   region.gpu_addr + region.size < gpu_addr + size;
+                                        }),
+                         mapped_regions.end());
+    return gpu_addr;
+}
+
 boost::optional<GPUVAddr> MemoryManager::FindFreeBlock(u64 size, u64 align) {
     GPUVAddr gpu_addr = 0;
     u64 free_space = 0;

src/video_core/memory_manager.h

@@ -25,6 +25,7 @@ public:
     GPUVAddr AllocateSpace(GPUVAddr gpu_addr, u64 size, u64 align);
     GPUVAddr MapBufferEx(VAddr cpu_addr, u64 size);
     GPUVAddr MapBufferEx(VAddr cpu_addr, GPUVAddr gpu_addr, u64 size);
+    GPUVAddr UnmapBuffer(GPUVAddr gpu_addr, u64 size);
     boost::optional<VAddr> GpuToCpuAddress(GPUVAddr gpu_addr);
     std::vector<GPUVAddr> CpuToGpuAddress(VAddr cpu_addr) const;
 

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -252,7 +252,7 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
             break;
         }
         default:
-            NGLOG_CRITICAL(HW_GPU, "Unimplemented shader index={}, enable={}, offset={:#010X}",
+            NGLOG_CRITICAL(HW_GPU, "Unimplemented shader index={}, enable={}, offset=0x{:08X}",
                            index, shader_config.enable.Value(), shader_config.offset);
             UNREACHABLE();
         }

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -153,85 +153,61 @@ private:
  */
 class GLSLRegister {
 public:
-    GLSLRegister(size_t index, ShaderWriter& shader)
-        : index{index}, shader{shader}, float_str{"freg_" + std::to_string(index)},
-          integer_str{"ireg_" + std::to_string(index)} {}
+    enum class Type {
+        Float,
+        Integer,
+        UnsignedInteger,
+    };
 
-    /// Returns a GLSL string representing the current state of the register
-    const std::string& GetActiveString() {
-        declr_type.insert(active_type);
+    GLSLRegister(size_t index, ShaderWriter& shader) : index{index}, shader{shader} {}
 
-        switch (active_type) {
+    /// Gets the GLSL type string for a register
+    static std::string GetTypeString(Type type) {
+        switch (type) {
         case Type::Float:
-            return float_str;
+            return "float";
         case Type::Integer:
-            return integer_str;
+            return "int";
+        case Type::UnsignedInteger:
+            return "uint";
         }
 
         UNREACHABLE();
-        return float_str;
+        return {};
     }
 
-    /// Returns a GLSL string representing the register as a float
-    const std::string& GetFloatString() const {
-        ASSERT(IsFloatUsed());
-        return float_str;
+    /// Gets the GLSL register prefix string, used for declarations and referencing
+    static std::string GetPrefixString(Type type) {
+        return "reg_" + GetTypeString(type) + '_';
     }
 
-    /// Returns a GLSL string representing the register as an integer
-    const std::string& GetIntegerString() const {
-        ASSERT(IsIntegerUsed());
-        return integer_str;
+    /// Returns a GLSL string representing the current state of the register
+    const std::string GetActiveString() {
+        declr_type.insert(active_type);
+        return GetPrefixString(active_type) + std::to_string(index);
     }
 
-    /// Convert the current register state from float to integer
-    void FloatToInteger() {
-        ASSERT(active_type == Type::Float);
-
-        const std::string src = GetActiveString();
-        active_type = Type::Integer;
-        const std::string dest = GetActiveString();
-
-        shader.AddLine(dest + " = floatBitsToInt(" + src + ");");
-    }
-
-    /// Convert the current register state from integer to float
-    void IntegerToFloat() {
-        ASSERT(active_type == Type::Integer);
-
-        const std::string src = GetActiveString();
-        active_type = Type::Float;
-        const std::string dest = GetActiveString();
-
-        shader.AddLine(dest + " = intBitsToFloat(" + src + ");");
-    }
-
-    /// Returns true if the register was ever used as a float, used for register declarations
-    bool IsFloatUsed() const {
-        return declr_type.find(Type::Float) != declr_type.end();
-    }
-
-    /// Returns true if the register was ever used as an integer, used for register declarations
-    bool IsIntegerUsed() const {
-        return declr_type.find(Type::Integer) != declr_type.end();
-    }
-
-    /// Returns true if the active type is float
+    /// Returns true if the active type is a float
     bool IsFloat() const {
         return active_type == Type::Float;
     }
 
-    /// Returns true if the active type is integer
+    /// Returns true if the active type is an integer
     bool IsInteger() const {
         return active_type == Type::Integer;
     }
 
-private:
-    enum class Type {
-        Float,
-        Integer,
-    };
+    /// Returns the index of the register
+    size_t GetIndex() const {
+        return index;
+    }
 
+    /// Returns a set of the declared types of the register
+    const std::set<Type>& DeclaredTypes() const {
+        return declr_type;
+    }
+
+private:
     const size_t index;
     const std::string float_str;
     const std::string integer_str;
@@ -254,18 +230,35 @@ public:
         BuildRegisterList();
     }
 
-    /// Generates code representing a temporary (GPR) register.
-    std::string GetRegister(const Register& reg, unsigned elem = 0) {
-        if (reg == Register::ZeroIndex) {
-            return "0";
-        }
-
-        return regs[reg.GetSwizzledIndex(elem)].GetActiveString();
+    /**
+     * Gets a register as an float.
+     * @param reg The register to get.
+     * @param elem The element to use for the operation.
+     * @returns GLSL string corresponding to the register as a float.
+     */
+    std::string GetRegisterAsFloat(const Register& reg, unsigned elem = 0) {
+        ASSERT(regs[reg].IsFloat());
+        return GetRegister(reg, elem);
     }
 
     /**
-     * Writes code that does a register assignment to float value operation. Should only be used
-     * with shader instructions that deal with floats.
+     * Gets a register as an integer.
+     * @param reg The register to get.
+     * @param elem The element to use for the operation.
+     * @param is_signed Whether to get the register as a signed (or unsigned) integer.
+     * @returns GLSL string corresponding to the register as an integer.
+     */
+    std::string GetRegisterAsInteger(const Register& reg, unsigned elem = 0,
+                                     bool is_signed = true) {
+        const std::string func = GetGLSLConversionFunc(
+            GLSLRegister::Type::Float,
+            is_signed ? GLSLRegister::Type::Integer : GLSLRegister::Type::UnsignedInteger);
+
+        return func + '(' + GetRegister(reg, elem) + ')';
+    }
+
+    /**
+     * Writes code that does a register assignment to float value operation.
      * @param reg The destination register to use.
      * @param elem The element to use for the operation.
      * @param value The code representing the value to assign.
@@ -277,21 +270,28 @@ public:
     void SetRegisterToFloat(const Register& reg, u64 elem, const std::string& value,
                             u64 dest_num_components, u64 value_num_components, bool is_abs = false,
                             u64 dest_elem = 0) {
-        ASSERT(regs[reg].IsFloat());
+        SetRegister(reg, elem, value, dest_num_components, value_num_components, is_abs, dest_elem);
+    }
 
-        std::string dest = GetRegister(reg, dest_elem);
-        if (dest_num_components > 1) {
-            dest += GetSwizzle(elem);
-        }
+    /**
+     * Writes code that does a register assignment to integer value operation.
+     * @param reg The destination register to use.
+     * @param elem The element to use for the operation.
+     * @param value The code representing the value to assign.
+     * @param dest_num_components Number of components in the destination.
+     * @param value_num_components Number of components in the value.
+     * @param is_abs Optional, when True, applies absolute value to output.
+     * @param dest_elem Optional, the destination element to use for the operation.
+     */
+    void SetRegisterToInteger(const Register& reg, bool is_signed, u64 elem,
+                              const std::string& value, u64 dest_num_components,
+                              u64 value_num_components, bool is_abs = false, u64 dest_elem = 0) {
+        const std::string func = GetGLSLConversionFunc(
+            is_signed ? GLSLRegister::Type::Integer : GLSLRegister::Type::UnsignedInteger,
+            GLSLRegister::Type::Float);
 
-        std::string src = '(' + value + ')';
-        if (value_num_components > 1) {
-            src += GetSwizzle(elem);
-        }
-
-        src = is_abs ? "abs(" + src + ')' : src;
-
-        shader.AddLine(dest + " = " + src + ';');
+        SetRegister(reg, elem, func + '(' + value + ')', dest_num_components, value_num_components,
+                    is_abs, dest_elem);
     }
 
     /**
@@ -302,7 +302,7 @@ public:
      * @param attribute The input attibute to use as the source value.
      */
     void SetRegisterToInputAttibute(const Register& reg, u64 elem, Attribute::Index attribute) {
-        std::string dest = GetRegister(reg);
+        std::string dest = GetRegisterAsFloat(reg);
         std::string src = GetInputAttribute(attribute) + GetSwizzle(elem);
 
         if (regs[reg].IsFloat()) {
@@ -323,7 +323,7 @@ public:
      */
     void SetOutputAttributeToRegister(Attribute::Index attribute, u64 elem, const Register& reg) {
         std::string dest = GetOutputAttribute(attribute) + GetSwizzle(elem);
-        std::string src = GetRegister(reg);
+        std::string src = GetRegisterAsFloat(reg);
         ASSERT_MSG(regs[reg].IsFloat(), "Output attributes must be set to a float");
         shader.AddLine(dest + " = " + src + ';');
     }
@@ -347,11 +347,10 @@ public:
     /// Add declarations for registers
     void GenerateDeclarations() {
         for (const auto& reg : regs) {
-            if (reg.IsFloatUsed()) {
-                declarations.AddLine("float " + reg.GetFloatString() + " = 0.0;");
-            }
-            if (reg.IsIntegerUsed()) {
-                declarations.AddLine("int " + reg.GetIntegerString() + " = 0;");
+            for (const auto& type : reg.DeclaredTypes()) {
+                declarations.AddLine(GLSLRegister::GetTypeString(type) + ' ' +
+                                     GLSLRegister::GetPrefixString(type) +
+                                     std::to_string(reg.GetIndex()) + " = 0;");
             }
         }
         declarations.AddNewLine();
@@ -395,6 +394,51 @@ public:
     }
 
 private:
+    /// Build GLSL conversion function, e.g. floatBitsToInt, intBitsToFloat, etc.
+    const std::string GetGLSLConversionFunc(GLSLRegister::Type src, GLSLRegister::Type dest) const {
+        const std::string src_type = GLSLRegister::GetTypeString(src);
+        std::string dest_type = GLSLRegister::GetTypeString(dest);
+        dest_type[0] = toupper(dest_type[0]);
+        return src_type + "BitsTo" + dest_type;
+    }
+
+    /// Generates code representing a temporary (GPR) register.
+    std::string GetRegister(const Register& reg, unsigned elem) {
+        if (reg == Register::ZeroIndex) {
+            return "0";
+        }
+
+        return regs[reg.GetSwizzledIndex(elem)].GetActiveString();
+    }
+
+    /**
+     * Writes code that does a register assignment to value operation.
+     * @param reg The destination register to use.
+     * @param elem The element to use for the operation.
+     * @param value The code representing the value to assign.
+     * @param dest_num_components Number of components in the destination.
+     * @param value_num_components Number of components in the value.
+     * @param is_abs Optional, when True, applies absolute value to output.
+     * @param dest_elem Optional, the destination element to use for the operation.
+     */
+    void SetRegister(const Register& reg, u64 elem, const std::string& value,
+                     u64 dest_num_components, u64 value_num_components, bool is_abs,
+                     u64 dest_elem) {
+        std::string dest = GetRegister(reg, dest_elem);
+        if (dest_num_components > 1) {
+            dest += GetSwizzle(elem);
+        }
+
+        std::string src = '(' + value + ')';
+        if (value_num_components > 1) {
+            src += GetSwizzle(elem);
+        }
+
+        src = is_abs ? "abs(" + src + ')' : src;
+
+        shader.AddLine(dest + " = " + src + ';');
+    }
+
     /// Build the GLSL register list.
     void BuildRegisterList() {
         for (size_t index = 0; index < Register::NumRegisters; ++index) {
@@ -598,7 +642,7 @@ private:
         switch (opcode->GetType()) {
         case OpCode::Type::Arithmetic: {
             std::string op_a = instr.alu.negate_a ? "-" : "";
-            op_a += regs.GetRegister(instr.gpr8);
+            op_a += regs.GetRegisterAsFloat(instr.gpr8);
             if (instr.alu.abs_a) {
                 op_a = "abs(" + op_a + ')';
             }
@@ -609,7 +653,7 @@ private:
                 op_b += GetImmediate19(instr);
             } else {
                 if (instr.is_b_gpr) {
-                    op_b += regs.GetRegister(instr.gpr20);
+                    op_b += regs.GetRegisterAsFloat(instr.gpr20);
                 } else {
                     op_b += regs.GetUniform(instr.uniform, instr.gpr0);
                 }
@@ -620,6 +664,17 @@ private:
             }
 
             switch (opcode->GetId()) {
+            case OpCode::Id::MOV_C:
+            case OpCode::Id::MOV_R: {
+                regs.SetRegisterToFloat(instr.gpr0, 0, op_b, 1, 1);
+                break;
+            }
+
+            case OpCode::Id::MOV32_IMM: {
+                // mov32i doesn't have abs or neg bits.
+                regs.SetRegisterToFloat(instr.gpr0, 0, GetImmediate32(instr), 1, 1);
+                break;
+            }
             case OpCode::Id::FMUL_C:
             case OpCode::Id::FMUL_R:
             case OpCode::Id::FMUL_IMM: {
@@ -629,8 +684,8 @@ private:
             case OpCode::Id::FMUL32_IMM: {
                 // fmul32i doesn't have abs or neg bits.
                 regs.SetRegisterToFloat(
-                    instr.gpr0, 0, regs.GetRegister(instr.gpr8) + " * " + GetImmediate32(instr), 1,
-                    1);
+                    instr.gpr0, 0,
+                    regs.GetRegisterAsFloat(instr.gpr8) + " * " + GetImmediate32(instr), 1, 1);
                 break;
             }
             case OpCode::Id::FADD_C:
@@ -687,29 +742,29 @@ private:
             break;
         }
         case OpCode::Type::Ffma: {
-            std::string op_a = regs.GetRegister(instr.gpr8);
+            std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
             std::string op_b = instr.ffma.negate_b ? "-" : "";
             std::string op_c = instr.ffma.negate_c ? "-" : "";
 
             switch (opcode->GetId()) {
             case OpCode::Id::FFMA_CR: {
                 op_b += regs.GetUniform(instr.uniform, instr.gpr0);
-                op_c += regs.GetRegister(instr.gpr39);
+                op_c += regs.GetRegisterAsFloat(instr.gpr39);
                 break;
             }
             case OpCode::Id::FFMA_RR: {
-                op_b += regs.GetRegister(instr.gpr20);
-                op_c += regs.GetRegister(instr.gpr39);
+                op_b += regs.GetRegisterAsFloat(instr.gpr20);
+                op_c += regs.GetRegisterAsFloat(instr.gpr39);
                 break;
             }
             case OpCode::Id::FFMA_RC: {
-                op_b += regs.GetRegister(instr.gpr39);
+                op_b += regs.GetRegisterAsFloat(instr.gpr39);
                 op_c += regs.GetUniform(instr.uniform, instr.gpr0);
                 break;
             }
             case OpCode::Id::FFMA_IMM: {
                 op_b += GetImmediate19(instr);
-                op_c += regs.GetRegister(instr.gpr39);
+                op_c += regs.GetRegisterAsFloat(instr.gpr39);
                 break;
             }
             default: {
@@ -721,6 +776,32 @@ private:
             regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " * " + op_b + " + " + op_c, 1, 1);
             break;
         }
+        case OpCode::Type::Conversion: {
+            ASSERT_MSG(instr.conversion.size == Register::Size::Word, "Unimplemented");
+            ASSERT_MSG(!instr.conversion.selector, "Unimplemented");
+            ASSERT_MSG(!instr.conversion.negate_a, "Unimplemented");
+            ASSERT_MSG(!instr.conversion.saturate_a, "Unimplemented");
+
+            switch (opcode->GetId()) {
+            case OpCode::Id::I2I_R:
+            case OpCode::Id::I2F_R: {
+                std::string op_a =
+                    regs.GetRegisterAsInteger(instr.gpr20, 0, instr.conversion.is_signed);
+
+                if (instr.conversion.abs_a) {
+                    op_a = "abs(" + op_a + ')';
+                }
+
+                regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_signed, 0, op_a, 1, 1);
+                break;
+            }
+            default: {
+                NGLOG_CRITICAL(HW_GPU, "Unhandled conversion instruction: {}", opcode->GetName());
+                UNREACHABLE();
+            }
+            }
+            break;
+        }
         case OpCode::Type::Memory: {
             const Attribute::Index attribute = instr.attribute.fmt20.index;
 
@@ -739,8 +820,8 @@ private:
             }
             case OpCode::Id::TEXS: {
                 ASSERT_MSG(instr.attribute.fmt20.size == 4, "untested");
-                const std::string op_a = regs.GetRegister(instr.gpr8);
-                const std::string op_b = regs.GetRegister(instr.gpr20);
+                const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
+                const std::string op_b = regs.GetRegisterAsFloat(instr.gpr20);
                 const std::string sampler = GetSampler(instr.sampler);
                 const std::string coord = "vec2 coords = vec2(" + op_a + ", " + op_b + ");";
                 // Add an extra scope and declare the texture coords inside to prevent overwriting
@@ -765,7 +846,7 @@ private:
         }
         case OpCode::Type::FloatSetPredicate: {
             std::string op_a = instr.fsetp.neg_a ? "-" : "";
-            op_a += regs.GetRegister(instr.gpr8);
+            op_a += regs.GetRegisterAsFloat(instr.gpr8);
 
             if (instr.fsetp.abs_a) {
                 op_a = "abs(" + op_a + ')';
@@ -781,7 +862,7 @@ private:
                 op_b += '(' + GetImmediate19(instr) + ')';
             } else {
                 if (instr.is_b_gpr) {
-                    op_b += regs.GetRegister(instr.gpr20);
+                    op_b += regs.GetRegisterAsFloat(instr.gpr20);
                 } else {
                     op_b += regs.GetUniform(instr.uniform, instr.gpr0);
                 }
@@ -807,6 +888,9 @@ private:
             case PredCondition::Equal:
                 SetPredicate(instr.fsetp.pred3, '(' + op_a + ") == (" + op_b + ')');
                 break;
+            case PredCondition::LessEqual:
+                SetPredicate(instr.fsetp.pred3, '(' + op_a + ") <= (" + op_b + ')');
+                break;
             default:
                 NGLOG_CRITICAL(HW_GPU, "Unhandled predicate condition: {} (a: {}, b: {})",
                                static_cast<unsigned>(instr.fsetp.cond.Value()), op_a, op_b);
@@ -816,7 +900,7 @@ private:
         }
         case OpCode::Type::FloatSet: {
             std::string op_a = instr.fset.neg_a ? "-" : "";
-            op_a += regs.GetRegister(instr.gpr8);
+            op_a += regs.GetRegisterAsFloat(instr.gpr8);
 
             if (instr.fset.abs_a) {
                 op_a = "abs(" + op_a + ')';
@@ -832,7 +916,7 @@ private:
                     op_b += imm;
             } else {
                 if (instr.is_b_gpr) {
-                    op_b += regs.GetRegister(instr.gpr20);
+                    op_b += regs.GetRegisterAsFloat(instr.gpr20);
                 } else {
                     op_b += regs.GetUniform(instr.uniform, instr.gpr0);
                 }
@@ -858,6 +942,10 @@ private:
                 regs.SetRegisterToFloat(instr.gpr0, 0,
                                         "((" + op_a + ") == (" + op_b + ")) ? 1.0 : 0", 1, 1);
                 break;
+            case PredCondition::LessEqual:
+                regs.SetRegisterToFloat(instr.gpr0, 0,
+                                        "((" + op_a + ") <= (" + op_b + ")) ? 1.0 : 0", 1, 1);
+                break;
             case PredCondition::GreaterThan:
                 regs.SetRegisterToFloat(instr.gpr0, 0,
                                         "((" + op_a + ") > (" + op_b + ")) ? 1.0 : 0", 1, 1);
@@ -877,10 +965,10 @@ private:
 
                 // Final color output is currently hardcoded to GPR0-3 for fragment shaders
                 if (stage == Maxwell3D::Regs::ShaderStage::Fragment) {
-                    shader.AddLine("color.r = " + regs.GetRegister(0) + ';');
-                    shader.AddLine("color.g = " + regs.GetRegister(1) + ';');
-                    shader.AddLine("color.b = " + regs.GetRegister(2) + ';');
-                    shader.AddLine("color.a = " + regs.GetRegister(3) + ';');
+                    shader.AddLine("color.r = " + regs.GetRegisterAsFloat(0) + ';');
+                    shader.AddLine("color.g = " + regs.GetRegisterAsFloat(1) + ';');
+                    shader.AddLine("color.b = " + regs.GetRegisterAsFloat(2) + ';');
+                    shader.AddLine("color.a = " + regs.GetRegisterAsFloat(3) + ';');
                 }
 
                 shader.AddLine("return true;");

src/video_core/renderer_opengl/maxwell_to_gl.h

@@ -36,6 +36,18 @@ inline GLenum VertexType(Maxwell::VertexAttribute attrib) {
         return {};
     }
 
+    case Maxwell::VertexAttribute::Type::SignedNorm: {
+
+        switch (attrib.size) {
+        case Maxwell::VertexAttribute::Size::Size_8_8_8_8:
+            return GL_BYTE;
+        }
+
+        NGLOG_CRITICAL(Render_OpenGL, "Unimplemented vertex size={}", attrib.SizeString());
+        UNREACHABLE();
+        return {};
+    }
+
     case Maxwell::VertexAttribute::Type::Float:
         return GL_FLOAT;
     }

src/yuzu/bootmanager.cpp

@@ -4,6 +4,8 @@
 #include <QScreen>
 #include <QWindow>
 
+#include <fmt/format.h>
+
 #include "common/microprofile.h"
 #include "common/scm_rev.h"
 #include "common/string_util.h"
@@ -102,8 +104,8 @@ private:
 GRenderWindow::GRenderWindow(QWidget* parent, EmuThread* emu_thread)
     : QWidget(parent), child(nullptr), emu_thread(emu_thread) {
 
-    std::string window_title = Common::StringFromFormat("yuzu %s| %s-%s", Common::g_build_name,
-                                                        Common::g_scm_branch, Common::g_scm_desc);
+    std::string window_title = fmt::format("yuzu {} | {}-{}", Common::g_build_name,
+                                           Common::g_scm_branch, Common::g_scm_desc);
     setWindowTitle(QString::fromStdString(window_title));
 
     InputCommon::Init();

src/yuzu/configuration/config.cpp

@@ -78,6 +78,7 @@ void Config::ReadValues() {
 
     qt_config->beginGroup("Core");
     Settings::values.use_cpu_jit = qt_config->value("use_cpu_jit", true).toBool();
+    Settings::values.use_multi_core = qt_config->value("use_multi_core", false).toBool();
     qt_config->endGroup();
 
     qt_config->beginGroup("Renderer");
@@ -177,6 +178,7 @@ void Config::SaveValues() {
 
     qt_config->beginGroup("Core");
     qt_config->setValue("use_cpu_jit", Settings::values.use_cpu_jit);
+    qt_config->setValue("use_multi_core", Settings::values.use_multi_core);
     qt_config->endGroup();
 
     qt_config->beginGroup("Renderer");

src/yuzu/configuration/configure_general.cpp

@@ -20,6 +20,7 @@ ConfigureGeneral::ConfigureGeneral(QWidget* parent)
     this->setConfiguration();
 
     ui->use_cpu_jit->setEnabled(!Core::System::GetInstance().IsPoweredOn());
+    ui->use_multi_core->setEnabled(!Core::System::GetInstance().IsPoweredOn());
     ui->use_docked_mode->setEnabled(!Core::System::GetInstance().IsPoweredOn());
 }
 
@@ -30,6 +31,7 @@ void ConfigureGeneral::setConfiguration() {
     ui->toggle_check_exit->setChecked(UISettings::values.confirm_before_closing);
     ui->theme_combobox->setCurrentIndex(ui->theme_combobox->findData(UISettings::values.theme));
     ui->use_cpu_jit->setChecked(Settings::values.use_cpu_jit);
+    ui->use_multi_core->setChecked(Settings::values.use_multi_core);
     ui->use_docked_mode->setChecked(Settings::values.use_docked_mode);
 }
 
@@ -40,6 +42,7 @@ void ConfigureGeneral::applyConfiguration() {
         ui->theme_combobox->itemData(ui->theme_combobox->currentIndex()).toString();
 
     Settings::values.use_cpu_jit = ui->use_cpu_jit->isChecked();
+    Settings::values.use_multi_core = ui->use_multi_core->isChecked();
     Settings::values.use_docked_mode = ui->use_docked_mode->isChecked();
     Settings::Apply();
 }

src/yuzu/configuration/configure_general.ui

@@ -58,6 +58,13 @@
             </property>
            </widget>
           </item>
+          <item>
+           <widget class="QCheckBox" name="use_multi_core">
+            <property name="text">
+             <string>Enable multi-core</string>
+            </property>
+           </widget>
+          </item>
          </layout>
         </item>
        </layout>

src/yuzu/debugger/registers.cpp

@@ -63,7 +63,7 @@ void RegistersWidget::OnDebugModeEntered() {
 
     for (int i = 0; i < core_registers->childCount(); ++i)
         core_registers->child(i)->setText(
-            1, QString("0x%1").arg(Core::CPU().GetReg(i), 8, 16, QLatin1Char('0')));
+            1, QString("0x%1").arg(Core::CurrentArmInterface().GetReg(i), 8, 16, QLatin1Char('0')));
 
     UpdateCPSRValues();
 }
@@ -122,7 +122,7 @@ void RegistersWidget::CreateCPSRChildren() {
 }
 
 void RegistersWidget::UpdateCPSRValues() {
-    const u32 cpsr_val = Core::CPU().GetCPSR();
+    const u32 cpsr_val = Core::CurrentArmInterface().GetCPSR();
 
     cpsr->setText(1, QString("0x%1").arg(cpsr_val, 8, 16, QLatin1Char('0')));
     cpsr->child(0)->setText(

src/yuzu/debugger/wait_tree.cpp

@@ -51,13 +51,21 @@ std::size_t WaitTreeItem::Row() const {
 }
 
 std::vector<std::unique_ptr<WaitTreeThread>> WaitTreeItem::MakeThreadItemList() {
-    const auto& threads = Core::System::GetInstance().Scheduler().GetThreadList();
     std::vector<std::unique_ptr<WaitTreeThread>> item_list;
-    item_list.reserve(threads.size());
-    for (std::size_t i = 0; i < threads.size(); ++i) {
-        item_list.push_back(std::make_unique<WaitTreeThread>(*threads[i]));
-        item_list.back()->row = i;
-    }
+    std::size_t row = 0;
+    auto add_threads = [&](const std::vector<Kernel::SharedPtr<Kernel::Thread>>& threads) {
+        for (std::size_t i = 0; i < threads.size(); ++i) {
+            item_list.push_back(std::make_unique<WaitTreeThread>(*threads[i]));
+            item_list.back()->row = row;
+            ++row;
+        }
+    };
+
+    add_threads(Core::System::GetInstance().Scheduler(0)->GetThreadList());
+    add_threads(Core::System::GetInstance().Scheduler(1)->GetThreadList());
+    add_threads(Core::System::GetInstance().Scheduler(2)->GetThreadList());
+    add_threads(Core::System::GetInstance().Scheduler(3)->GetThreadList());
+
     return item_list;
 }
 
@@ -240,6 +248,9 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeThread::GetChildren() const {
     }
 
     list.push_back(std::make_unique<WaitTreeText>(tr("processor = %1").arg(processor)));
+    list.push_back(std::make_unique<WaitTreeText>(tr("ideal core = %1").arg(thread.ideal_core)));
+    list.push_back(
+        std::make_unique<WaitTreeText>(tr("affinity mask = %1").arg(thread.affinity_mask)));
     list.push_back(std::make_unique<WaitTreeText>(tr("thread id = %1").arg(thread.GetThreadId())));
     list.push_back(std::make_unique<WaitTreeText>(tr("priority = %1(current) / %2(normal)")
                                                       .arg(thread.current_priority)

src/yuzu_cmd/config.cpp

@@ -91,6 +91,7 @@ void Config::ReadValues() {
 
     // Core
     Settings::values.use_cpu_jit = sdl2_config->GetBoolean("Core", "use_cpu_jit", true);
+    Settings::values.use_multi_core = sdl2_config->GetBoolean("Core", "use_multi_core", false);
 
     // Renderer
     Settings::values.resolution_factor =

src/yuzu_cmd/default_ini.h

@@ -80,6 +80,10 @@ touch_device=
 # 0: Interpreter (slow), 1 (default): JIT (fast)
 use_cpu_jit =
 
+# Whether to use multi-core for CPU emulation
+# 0 (default): Disabled, 1: Enabled
+use_multi_core=
+
 [Renderer]
 # Whether to use software or hardware rendering.
 # 0: Software, 1 (default): Hardware

src/yuzu_cmd/emu_window/emu_window_sdl2.cpp

@@ -7,6 +7,7 @@
 #include <string>
 #define SDL_MAIN_HANDLED
 #include <SDL.h>
+#include <fmt/format.h>
 #include <glad/glad.h>
 #include "common/logging/log.h"
 #include "common/scm_rev.h"
@@ -97,8 +98,8 @@ EmuWindow_SDL2::EmuWindow_SDL2(bool fullscreen) {
     SDL_GL_SetAttribute(SDL_GL_BLUE_SIZE, 8);
     SDL_GL_SetAttribute(SDL_GL_ALPHA_SIZE, 0);
 
-    std::string window_title = Common::StringFromFormat("yuzu %s| %s-%s ", Common::g_build_name,
-                                                        Common::g_scm_branch, Common::g_scm_desc);
+    std::string window_title = fmt::format("yuzu {} | {}-{}", Common::g_build_name,
+                                           Common::g_scm_branch, Common::g_scm_desc);
     render_window =
         SDL_CreateWindow(window_title.c_str(),
                          SDL_WINDOWPOS_UNDEFINED, // x position