Merge branch 'nsp' into nsp-bktr-merge-fix

Eliminated duplicate code and adds support for Rev1+ carts

.travis.yml

@@ -20,6 +20,7 @@ matrix:
       install: "./.travis/linux/deps.sh"
       script: "./.travis/linux/build.sh"
       after_success: "./.travis/linux/upload.sh"
+      cache: ccache
     - os: osx
       env: NAME="macos build"
       sudo: false
@@ -27,6 +28,7 @@ matrix:
       install: "./.travis/macos/deps.sh"
       script: "./.travis/macos/build.sh"
       after_success: "./.travis/macos/upload.sh"
+      cache: ccache
 
 deploy:
   provider: releases
@@ -42,7 +44,3 @@ notifications:
   webhooks:
     urls:
       - https://api.yuzu-emu.org/code/travis/notify
-
-cache:
-  directories:
-    - $HOME/.ccache

.travis/common/travis-ci.env

@@ -0,0 +1,12 @@
+# List of environment variables to be shared with Docker containers
+CI
+TRAVIS
+CONTINUOUS_INTEGRATION
+TRAVIS_BRANCH
+TRAVIS_BUILD_ID
+TRAVIS_BUILD_NUMBER
+TRAVIS_COMMIT
+TRAVIS_JOB_ID
+TRAVIS_JOB_NUMBER
+TRAVIS_REPO_SLUG
+TRAVIS_TAG

.travis/linux/build.sh

@@ -1,3 +1,4 @@
 #!/bin/bash -ex
 
-docker run -e CCACHE_DIR=/ccache -v $HOME/.ccache:/ccache -v $(pwd):/yuzu ubuntu:18.04 /bin/bash /yuzu/.travis/linux/docker.sh
+mkdir -p "$HOME/.ccache"
+docker run --env-file .travis/common/travis-ci.env -v $(pwd):/yuzu -v "$HOME/.ccache":/root/.ccache ubuntu:18.04 /bin/bash /yuzu/.travis/linux/docker.sh

.travis/linux/docker.sh

@@ -5,14 +5,8 @@ apt-get install --no-install-recommends -y build-essential git libqt5opengl5-dev
 
 cd /yuzu
 
-export PATH=/usr/lib/ccache:$PATH
-ln -sf /usr/bin/ccache /usr/lib/ccache/cc
-ln -sf /usr/bin/ccache /usr/lib/ccache/c++
 mkdir build && cd build
-ccache --show-stats > ccache_before
-cmake .. -DYUZU_BUILD_UNICORN=ON -DCMAKE_BUILD_TYPE=Release -G Ninja
+cmake .. -DYUZU_BUILD_UNICORN=ON -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=/usr/lib/ccache/gcc -DCMAKE_CXX_COMPILER=/usr/lib/ccache/g++ -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -G Ninja
 ninja
-ccache --show-stats > ccache_after
-diff -U100 ccache_before ccache_after || true
 
 ctest -VV -C Release

.travis/macos/build.sh

@@ -5,14 +5,11 @@ set -o pipefail
 export MACOSX_DEPLOYMENT_TARGET=10.12
 export Qt5_DIR=$(brew --prefix)/opt/qt5
 export UNICORNDIR=$(pwd)/externals/unicorn
+export PATH="/usr/local/opt/ccache/libexec:$PATH"
 
 mkdir build && cd build
-export PATH=/usr/local/opt/ccache/libexec:$PATH
-ccache --show-stats > ccache_before
 cmake --version
-cmake .. -DYUZU_BUILD_UNICORN=ON -DCMAKE_BUILD_TYPE=Release
+cmake .. -DYUZU_BUILD_UNICORN=ON -DCMAKE_BUILD_TYPE=Release -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON
 make -j4
-ccache --show-stats > ccache_after
-diff -U100 ccache_before ccache_after || true
 
 ctest -VV -C Release

CMakeLists.txt

@@ -41,6 +41,19 @@ function(check_submodules_present)
 endfunction()
 check_submodules_present()
 
+configure_file(${CMAKE_SOURCE_DIR}/dist/compatibility_list/compatibility_list.qrc
+               ${CMAKE_BINARY_DIR}/dist/compatibility_list/compatibility_list.qrc
+               COPYONLY)
+if (ENABLE_COMPATIBILITY_LIST_DOWNLOAD AND NOT EXISTS ${CMAKE_BINARY_DIR}/dist/compatibility_list/compatibility_list.json)
+    message(STATUS "Downloading compatibility list for yuzu...")
+    file(DOWNLOAD
+        https://api.yuzu-emu.org/gamedb/
+        "${CMAKE_BINARY_DIR}/dist/compatibility_list/compatibility_list.json" SHOW_PROGRESS)
+endif()
+if (NOT EXISTS ${CMAKE_BINARY_DIR}/dist/compatibility_list/compatibility_list.json)
+    file(WRITE ${CMAKE_BINARY_DIR}/dist/compatibility_list/compatibility_list.json "")
+endif()
+
 # Detect current compilation architecture and create standard definitions
 # =======================================================================
 

appveyor.yml

@@ -41,9 +41,9 @@ before_build:
   - ps: |
         if ($env:BUILD_TYPE -eq 'msvc') {
           # redirect stderr and change the exit code to prevent powershell from cancelling the build if cmake prints a warning
-          cmd /C 'cmake -G "Visual Studio 15 2017 Win64" -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_BUNDLED_UNICORN=1 .. 2>&1 && exit 0'
+          cmd /C 'cmake -G "Visual Studio 15 2017 Win64" -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_BUNDLED_UNICORN=1 -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON .. 2>&1 && exit 0'
         } else {
-          C:\msys64\usr\bin\bash.exe -lc "cmake -G 'MSYS Makefiles' -DYUZU_BUILD_UNICORN=1 -DCMAKE_BUILD_TYPE=Release .. 2>&1"
+          C:\msys64\usr\bin\bash.exe -lc "cmake -G 'MSYS Makefiles' -DYUZU_BUILD_UNICORN=1 -DCMAKE_BUILD_TYPE=Release -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON .. 2>&1"
         }
   - cd ..
 

dist/compatibility_list/compatibility_list.qrc

@@ -0,0 +1,5 @@
+<RCC>
+  <qresource prefix="compatibility_list">
+      <file>compatibility_list.json</file>
+  </qresource>
+</RCC>

src/core/CMakeLists.txt

@@ -35,8 +35,12 @@ add_library(core STATIC
     file_sys/mode.h
     file_sys/nca_metadata.cpp
     file_sys/nca_metadata.h
+    file_sys/nca_patch.cpp
+    file_sys/nca_patch.h
     file_sys/partition_filesystem.cpp
     file_sys/partition_filesystem.h
+    file_sys/patch_manager.cpp
+    file_sys/patch_manager.h
     file_sys/program_metadata.cpp
     file_sys/program_metadata.h
     file_sys/registered_cache.cpp
@@ -49,6 +53,8 @@ add_library(core STATIC
     file_sys/savedata_factory.h
     file_sys/sdmc_factory.cpp
     file_sys/sdmc_factory.h
+    file_sys/submission_package.cpp
+    file_sys/submission_package.h
     file_sys/vfs.cpp
     file_sys/vfs.h
     file_sys/vfs_concat.cpp
@@ -359,6 +365,8 @@ add_library(core STATIC
     loader/nro.h
     loader/nso.cpp
     loader/nso.h
+    loader/nsp.cpp
+    loader/nsp.h
     loader/xci.cpp
     loader/xci.h
     memory.cpp

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -9,6 +9,7 @@
 #include "common/logging/log.h"
 #include "core/arm/dynarmic/arm_dynarmic.h"
 #include "core/core.h"
+#include "core/core_cpu.h"
 #include "core/core_timing.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/svc.h"

src/core/core.cpp

@@ -2,24 +2,35 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <array>
+#include <map>
 #include <memory>
+#include <thread>
 #include <utility>
+
 #include "common/logging/log.h"
 #include "common/string_util.h"
+#include "core/arm/exclusive_monitor.h"
 #include "core/core.h"
+#include "core/core_cpu.h"
 #include "core/core_timing.h"
 #include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
+#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/thread.h"
 #include "core/hle/service/service.h"
 #include "core/hle/service/sm/controller.h"
 #include "core/hle/service/sm/sm.h"
 #include "core/loader/loader.h"
+#include "core/perf_stats.h"
 #include "core/settings.h"
+#include "core/telemetry_session.h"
 #include "file_sys/vfs_concat.h"
 #include "file_sys/vfs_real.h"
+#include "video_core/debug_utils/debug_utils.h"
+#include "video_core/gpu.h"
 #include "video_core/renderer_base.h"
 #include "video_core/video_core.h"
 
@@ -27,71 +38,9 @@ namespace Core {
 
 /*static*/ System System::s_instance;
 
-System::System() = default;
-
-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;
-
-    // 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();
-
-        // If the loop is halted and we want to step, use a tiny (1) number of instructions to
-        // execute. Otherwise, get out of the loop function.
-        if (GDBStub::GetCpuHaltFlag()) {
-            if (GDBStub::GetCpuStepFlag()) {
-                tight_loop = false;
-            } else {
-                return ResultStatus::Success;
-            }
-        }
-    }
-
-    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;
-        }
-    }
-
-    if (GDBStub::IsServerEnabled()) {
-        GDBStub::SetCpuStepFlag(false);
-    }
-
-    return status;
-}
-
-System::ResultStatus System::SingleStep() {
-    return RunLoop(false);
-}
-
-static FileSys::VirtualFile GetGameFileFromPath(const FileSys::VirtualFilesystem& vfs,
-                                                const std::string& path) {
+namespace {
+FileSys::VirtualFile GetGameFileFromPath(const FileSys::VirtualFilesystem& vfs,
+                                         const std::string& path) {
     // To account for split 00+01+etc files.
     std::string dir_name;
     std::string filename;
@@ -121,165 +70,402 @@ static FileSys::VirtualFile GetGameFileFromPath(const FileSys::VirtualFilesystem
     return vfs->OpenFile(path, FileSys::Mode::Read);
 }
 
+/// Runs a CPU core while the system is powered on
+void RunCpuCore(std::shared_ptr<Cpu> cpu_state) {
+    while (Core::System::GetInstance().IsPoweredOn()) {
+        cpu_state->RunLoop(true);
+    }
+}
+} // Anonymous namespace
+
+struct System::Impl {
+    Cpu& CurrentCpuCore() {
+        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];
+    }
+
+    ResultStatus RunLoop(bool tight_loop) {
+        status = ResultStatus::Success;
+
+        // 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();
+
+            // If the loop is halted and we want to step, use a tiny (1) number of instructions to
+            // execute. Otherwise, get out of the loop function.
+            if (GDBStub::GetCpuHaltFlag()) {
+                if (GDBStub::GetCpuStepFlag()) {
+                    tight_loop = false;
+                } else {
+                    return ResultStatus::Success;
+                }
+            }
+        }
+
+        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;
+            }
+        }
+
+        if (GDBStub::IsServerEnabled()) {
+            GDBStub::SetCpuStepFlag(false);
+        }
+
+        return status;
+    }
+
+    ResultStatus Init(Frontend::EmuWindow& emu_window) {
+        LOG_DEBUG(HW_Memory, "initialized OK");
+
+        CoreTiming::Init();
+        kernel.Initialize();
+
+        // Create a default fs if one doesn't already exist.
+        if (virtual_filesystem == nullptr)
+            virtual_filesystem = std::make_shared<FileSys::RealVfsFilesystem>();
+
+        current_process = Kernel::Process::Create(kernel, "main");
+
+        cpu_barrier = std::make_shared<CpuBarrier>();
+        cpu_exclusive_monitor = Cpu::MakeExclusiveMonitor(cpu_cores.size());
+        for (size_t index = 0; index < cpu_cores.size(); ++index) {
+            cpu_cores[index] = std::make_shared<Cpu>(cpu_exclusive_monitor, cpu_barrier, index);
+        }
+
+        telemetry_session = std::make_unique<Core::TelemetrySession>();
+        service_manager = std::make_shared<Service::SM::ServiceManager>();
+
+        Service::Init(service_manager, virtual_filesystem);
+        GDBStub::Init();
+
+        renderer = VideoCore::CreateRenderer(emu_window);
+        if (!renderer->Init()) {
+            return ResultStatus::ErrorVideoCore;
+        }
+
+        gpu_core = std::make_unique<Tegra::GPU>(renderer->Rasterizer());
+
+        // 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];
+            }
+        }
+
+        LOG_DEBUG(Core, "Initialized OK");
+
+        // Reset counters and set time origin to current frame
+        GetAndResetPerfStats();
+        perf_stats.BeginSystemFrame();
+
+        return ResultStatus::Success;
+    }
+
+    ResultStatus Load(Frontend::EmuWindow& emu_window, const std::string& filepath) {
+        app_loader = Loader::GetLoader(GetGameFileFromPath(virtual_filesystem, filepath));
+
+        if (!app_loader) {
+            LOG_CRITICAL(Core, "Failed to obtain loader for {}!", filepath);
+            return ResultStatus::ErrorGetLoader;
+        }
+        std::pair<boost::optional<u32>, Loader::ResultStatus> system_mode =
+            app_loader->LoadKernelSystemMode();
+
+        if (system_mode.second != Loader::ResultStatus::Success) {
+            LOG_CRITICAL(Core, "Failed to determine system mode (Error {})!",
+                         static_cast<int>(system_mode.second));
+
+            return ResultStatus::ErrorSystemMode;
+        }
+
+        ResultStatus init_result{Init(emu_window)};
+        if (init_result != ResultStatus::Success) {
+            LOG_CRITICAL(Core, "Failed to initialize system (Error {})!",
+                         static_cast<int>(init_result));
+            Shutdown();
+            return init_result;
+        }
+
+        const Loader::ResultStatus load_result{app_loader->Load(current_process)};
+        if (load_result != Loader::ResultStatus::Success) {
+            LOG_CRITICAL(Core, "Failed to load ROM (Error {})!", static_cast<int>(load_result));
+            Shutdown();
+
+            return static_cast<ResultStatus>(static_cast<u32>(ResultStatus::ErrorLoader) +
+                                             static_cast<u32>(load_result));
+        }
+        status = ResultStatus::Success;
+        return status;
+    }
+
+    void Shutdown() {
+        // Log last frame performance stats
+        auto perf_results = GetAndResetPerfStats();
+        Telemetry().AddField(Telemetry::FieldType::Performance, "Shutdown_EmulationSpeed",
+                             perf_results.emulation_speed * 100.0);
+        Telemetry().AddField(Telemetry::FieldType::Performance, "Shutdown_Framerate",
+                             perf_results.game_fps);
+        Telemetry().AddField(Telemetry::FieldType::Performance, "Shutdown_Frametime",
+                             perf_results.frametime * 1000.0);
+
+        // Shutdown emulation session
+        renderer.reset();
+        GDBStub::Shutdown();
+        Service::Shutdown();
+        service_manager.reset();
+        telemetry_session.reset();
+        gpu_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();
+
+        // Shutdown kernel and core timing
+        kernel.Shutdown();
+        CoreTiming::Shutdown();
+
+        // Close app loader
+        app_loader.reset();
+
+        LOG_DEBUG(Core, "Shutdown OK");
+    }
+
+    Loader::ResultStatus GetGameName(std::string& out) const {
+        if (app_loader == nullptr)
+            return Loader::ResultStatus::ErrorNotInitialized;
+        return app_loader->ReadTitle(out);
+    }
+
+    void SetStatus(ResultStatus new_status, const char* details = nullptr) {
+        status = new_status;
+        if (details) {
+            status_details = details;
+        }
+    }
+
+    PerfStatsResults GetAndResetPerfStats() {
+        return perf_stats.GetAndResetStats(CoreTiming::GetGlobalTimeUs());
+    }
+
+    Kernel::KernelCore kernel;
+    /// RealVfsFilesystem instance
+    FileSys::VirtualFilesystem virtual_filesystem;
+    /// AppLoader used to load the current executing application
+    std::unique_ptr<Loader::AppLoader> app_loader;
+    std::unique_ptr<VideoCore::RendererBase> renderer;
+    std::unique_ptr<Tegra::GPU> gpu_core;
+    std::shared_ptr<Tegra::DebugContext> debug_context;
+    Kernel::SharedPtr<Kernel::Process> current_process;
+    std::shared_ptr<ExclusiveMonitor> cpu_exclusive_monitor;
+    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;
+
+    /// Telemetry session for this emulation session
+    std::unique_ptr<Core::TelemetrySession> telemetry_session;
+
+    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;
+
+    Core::PerfStats perf_stats;
+    Core::FrameLimiter frame_limiter;
+};
+
+System::System() : impl{std::make_unique<Impl>()} {}
+System::~System() = default;
+
+Cpu& System::CurrentCpuCore() {
+    return impl->CurrentCpuCore();
+}
+
+System::ResultStatus System::RunLoop(bool tight_loop) {
+    return impl->RunLoop(tight_loop);
+}
+
+System::ResultStatus System::SingleStep() {
+    return RunLoop(false);
+}
+
+void System::InvalidateCpuInstructionCaches() {
+    for (auto& cpu : impl->cpu_cores) {
+        cpu->ArmInterface().ClearInstructionCache();
+    }
+}
+
 System::ResultStatus System::Load(Frontend::EmuWindow& emu_window, const std::string& filepath) {
-    app_loader = Loader::GetLoader(GetGameFileFromPath(virtual_filesystem, filepath));
+    return impl->Load(emu_window, filepath);
+}
 
-    if (!app_loader) {
-        LOG_CRITICAL(Core, "Failed to obtain loader for {}!", filepath);
-        return ResultStatus::ErrorGetLoader;
-    }
-    std::pair<boost::optional<u32>, Loader::ResultStatus> system_mode =
-        app_loader->LoadKernelSystemMode();
-
-    if (system_mode.second != Loader::ResultStatus::Success) {
-        LOG_CRITICAL(Core, "Failed to determine system mode (Error {})!",
-                     static_cast<int>(system_mode.second));
-
-        return ResultStatus::ErrorSystemMode;
-    }
-
-    ResultStatus init_result{Init(emu_window)};
-    if (init_result != ResultStatus::Success) {
-        LOG_CRITICAL(Core, "Failed to initialize system (Error {})!",
-                     static_cast<int>(init_result));
-        System::Shutdown();
-        return init_result;
-    }
-
-    const Loader::ResultStatus load_result{app_loader->Load(current_process)};
-    if (load_result != Loader::ResultStatus::Success) {
-        LOG_CRITICAL(Core, "Failed to load ROM (Error {})!", static_cast<int>(load_result));
-        System::Shutdown();
-
-        return static_cast<ResultStatus>(static_cast<u32>(ResultStatus::ErrorLoader) +
-                                         static_cast<u32>(load_result));
-    }
-    status = ResultStatus::Success;
-    return status;
+bool System::IsPoweredOn() const {
+    return impl->cpu_barrier && impl->cpu_barrier->IsAlive();
 }
 
 void System::PrepareReschedule() {
     CurrentCpuCore().PrepareReschedule();
 }
 
-PerfStats::Results System::GetAndResetPerfStats() {
-    return perf_stats.GetAndResetStats(CoreTiming::GetGlobalTimeUs());
+PerfStatsResults System::GetAndResetPerfStats() {
+    return impl->GetAndResetPerfStats();
+}
+
+Core::TelemetrySession& System::TelemetrySession() const {
+    return *impl->telemetry_session;
+}
+
+ARM_Interface& System::CurrentArmInterface() {
+    return CurrentCpuCore().ArmInterface();
+}
+
+size_t System::CurrentCoreIndex() {
+    return CurrentCpuCore().CoreIndex();
+}
+
+Kernel::Scheduler& System::CurrentScheduler() {
+    return *CurrentCpuCore().Scheduler();
 }
 
 const std::shared_ptr<Kernel::Scheduler>& System::Scheduler(size_t core_index) {
     ASSERT(core_index < NUM_CPU_CORES);
-    return cpu_cores[core_index]->Scheduler();
+    return impl->cpu_cores[core_index]->Scheduler();
+}
+
+Kernel::SharedPtr<Kernel::Process>& System::CurrentProcess() {
+    return impl->current_process;
 }
 
 ARM_Interface& System::ArmInterface(size_t core_index) {
     ASSERT(core_index < NUM_CPU_CORES);
-    return cpu_cores[core_index]->ArmInterface();
+    return impl->cpu_cores[core_index]->ArmInterface();
 }
 
 Cpu& System::CpuCore(size_t core_index) {
     ASSERT(core_index < NUM_CPU_CORES);
-    return *cpu_cores[core_index];
+    return *impl->cpu_cores[core_index];
+}
+
+ExclusiveMonitor& System::Monitor() {
+    return *impl->cpu_exclusive_monitor;
+}
+
+Tegra::GPU& System::GPU() {
+    return *impl->gpu_core;
+}
+
+const Tegra::GPU& System::GPU() const {
+    return *impl->gpu_core;
+}
+
+VideoCore::RendererBase& System::Renderer() {
+    return *impl->renderer;
+}
+
+const VideoCore::RendererBase& System::Renderer() const {
+    return *impl->renderer;
+}
+
+Kernel::KernelCore& System::Kernel() {
+    return impl->kernel;
+}
+
+const Kernel::KernelCore& System::Kernel() const {
+    return impl->kernel;
+}
+
+Core::PerfStats& System::GetPerfStats() {
+    return impl->perf_stats;
+}
+
+const Core::PerfStats& System::GetPerfStats() const {
+    return impl->perf_stats;
+}
+
+Core::FrameLimiter& System::FrameLimiter() {
+    return impl->frame_limiter;
+}
+
+const Core::FrameLimiter& System::FrameLimiter() const {
+    return impl->frame_limiter;
+}
+
+Loader::ResultStatus System::GetGameName(std::string& out) const {
+    return impl->GetGameName(out);
+}
+
+void System::SetStatus(ResultStatus new_status, const char* details) {
+    impl->SetStatus(new_status, details);
+}
+
+const std::string& System::GetStatusDetails() const {
+    return impl->status_details;
+}
+
+Loader::AppLoader& System::GetAppLoader() const {
+    return *impl->app_loader;
+}
+
+void System::SetGPUDebugContext(std::shared_ptr<Tegra::DebugContext> context) {
+    impl->debug_context = std::move(context);
+}
+
+std::shared_ptr<Tegra::DebugContext> System::GetGPUDebugContext() const {
+    return impl->debug_context;
+}
+
+void System::SetFilesystem(std::shared_ptr<FileSys::VfsFilesystem> vfs) {
+    impl->virtual_filesystem = std::move(vfs);
+}
+
+std::shared_ptr<FileSys::VfsFilesystem> System::GetFilesystem() const {
+    return impl->virtual_filesystem;
 }
 
 System::ResultStatus System::Init(Frontend::EmuWindow& emu_window) {
-    LOG_DEBUG(HW_Memory, "initialized OK");
-
-    CoreTiming::Init();
-
-    // Create a default fs if one doesn't already exist.
-    if (virtual_filesystem == nullptr)
-        virtual_filesystem = std::make_shared<FileSys::RealVfsFilesystem>();
-
-    current_process = Kernel::Process::Create("main");
-
-    cpu_barrier = std::make_shared<CpuBarrier>();
-    cpu_exclusive_monitor = Cpu::MakeExclusiveMonitor(cpu_cores.size());
-    for (size_t index = 0; index < cpu_cores.size(); ++index) {
-        cpu_cores[index] = std::make_shared<Cpu>(cpu_exclusive_monitor, cpu_barrier, index);
-    }
-
-    telemetry_session = std::make_unique<Core::TelemetrySession>();
-    service_manager = std::make_shared<Service::SM::ServiceManager>();
-
-    Kernel::Init();
-    Service::Init(service_manager, virtual_filesystem);
-    GDBStub::Init();
-
-    renderer = VideoCore::CreateRenderer(emu_window);
-    if (!renderer->Init()) {
-        return ResultStatus::ErrorVideoCore;
-    }
-
-    gpu_core = std::make_unique<Tegra::GPU>(renderer->Rasterizer());
-
-    // 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];
-        }
-    }
-
-    LOG_DEBUG(Core, "Initialized OK");
-
-    // Reset counters and set time origin to current frame
-    GetAndResetPerfStats();
-    perf_stats.BeginSystemFrame();
-
-    return ResultStatus::Success;
+    return impl->Init(emu_window);
 }
 
 void System::Shutdown() {
-    // Log last frame performance stats
-    auto perf_results = GetAndResetPerfStats();
-    Telemetry().AddField(Telemetry::FieldType::Performance, "Shutdown_EmulationSpeed",
-                         perf_results.emulation_speed * 100.0);
-    Telemetry().AddField(Telemetry::FieldType::Performance, "Shutdown_Framerate",
-                         perf_results.game_fps);
-    Telemetry().AddField(Telemetry::FieldType::Performance, "Shutdown_Frametime",
-                         perf_results.frametime * 1000.0);
-
-    // Shutdown emulation session
-    renderer.reset();
-    GDBStub::Shutdown();
-    Service::Shutdown();
-    Kernel::Shutdown();
-    service_manager.reset();
-    telemetry_session.reset();
-    gpu_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();
-
-    LOG_DEBUG(Core, "Shutdown OK");
+    impl->Shutdown();
 }
 
 Service::SM::ServiceManager& System::ServiceManager() {
-    return *service_manager;
+    return *impl->service_manager;
 }
 
 const Service::SM::ServiceManager& System::ServiceManager() const {
-    return *service_manager;
+    return *impl->service_manager;
 }
 
 } // namespace Core

src/core/core.h

@@ -4,40 +4,55 @@
 
 #pragma once
 
-#include <array>
-#include <map>
+#include <cstddef>
 #include <memory>
 #include <string>
-#include <thread>
+
 #include "common/common_types.h"
-#include "core/arm/exclusive_monitor.h"
-#include "core/core_cpu.h"
 #include "core/hle/kernel/object.h"
-#include "core/hle/kernel/scheduler.h"
-#include "core/loader/loader.h"
-#include "core/memory.h"
-#include "core/perf_stats.h"
-#include "core/telemetry_session.h"
-#include "file_sys/vfs_real.h"
-#include "hle/service/filesystem/filesystem.h"
-#include "video_core/debug_utils/debug_utils.h"
-#include "video_core/gpu.h"
 
 namespace Core::Frontend {
 class EmuWindow;
-}
+} // namespace Core::Frontend
+
+namespace FileSys {
+class VfsFilesystem;
+} // namespace FileSys
+
+namespace Kernel {
+class KernelCore;
+class Process;
+class Scheduler;
+} // namespace Kernel
+
+namespace Loader {
+class AppLoader;
+enum class ResultStatus : u16;
+} // namespace Loader
 
 namespace Service::SM {
 class ServiceManager;
-}
+} // namespace Service::SM
+
+namespace Tegra {
+class DebugContext;
+class GPU;
+} // namespace Tegra
 
 namespace VideoCore {
 class RendererBase;
-}
+} // namespace VideoCore
 
 namespace Core {
 
 class ARM_Interface;
+class Cpu;
+class ExclusiveMonitor;
+class FrameLimiter;
+class PerfStats;
+class TelemetrySession;
+
+struct PerfStatsResults;
 
 class System {
 public:
@@ -93,11 +108,7 @@ public:
      * This function should only be used by GDB Stub to support breakpoints, memory updates and
      * step/continue commands.
      */
-    void InvalidateCpuInstructionCaches() {
-        for (auto& cpu : cpu_cores) {
-            cpu->ArmInterface().ClearInstructionCache();
-        }
-    }
+    void InvalidateCpuInstructionCaches();
 
     /// Shutdown the emulated system.
     void Shutdown();
@@ -116,33 +127,28 @@ public:
      * application).
      * @returns True if the emulated system is powered on, otherwise false.
      */
-    bool IsPoweredOn() const {
-        return cpu_barrier && cpu_barrier->IsAlive();
-    }
+    bool IsPoweredOn() const;
 
     /**
      * Returns a reference to the telemetry session for this emulation session.
      * @returns Reference to the telemetry session.
      */
-    Core::TelemetrySession& TelemetrySession() const {
-        return *telemetry_session;
-    }
+    Core::TelemetrySession& TelemetrySession() const;
 
     /// Prepare the core emulation for a reschedule
     void PrepareReschedule();
 
     /// Gets and resets core performance statistics
-    PerfStats::Results GetAndResetPerfStats();
+    PerfStatsResults GetAndResetPerfStats();
 
     /// Gets an ARM interface to the CPU core that is currently running
-    ARM_Interface& CurrentArmInterface() {
-        return CurrentCpuCore().ArmInterface();
-    }
+    ARM_Interface& CurrentArmInterface();
 
     /// Gets the index of the currently running CPU core
-    size_t CurrentCoreIndex() {
-        return CurrentCpuCore().CoreIndex();
-    }
+    size_t CurrentCoreIndex();
+
+    /// Gets the scheduler for the CPU core that is currently running
+    Kernel::Scheduler& CurrentScheduler();
 
     /// Gets an ARM interface to the CPU core with the specified index
     ARM_Interface& ArmInterface(size_t core_index);
@@ -150,87 +156,64 @@ public:
     /// Gets a CPU interface to the CPU core with the specified index
     Cpu& CpuCore(size_t core_index);
 
+    /// Gets the exclusive monitor
+    ExclusiveMonitor& Monitor();
+
     /// Gets a mutable reference to the GPU interface
-    Tegra::GPU& GPU() {
-        return *gpu_core;
-    }
+    Tegra::GPU& GPU();
 
     /// Gets an immutable reference to the GPU interface.
-    const Tegra::GPU& GPU() const {
-        return *gpu_core;
-    }
+    const Tegra::GPU& GPU() const;
 
     /// Gets a mutable reference to the renderer.
-    VideoCore::RendererBase& Renderer() {
-        return *renderer;
-    }
+    VideoCore::RendererBase& Renderer();
 
     /// Gets an immutable reference to the renderer.
-    const VideoCore::RendererBase& Renderer() const {
-        return *renderer;
-    }
-
-    /// Gets the scheduler for the CPU core that is currently running
-    Kernel::Scheduler& CurrentScheduler() {
-        return *CurrentCpuCore().Scheduler();
-    }
-
-    /// Gets the exclusive monitor
-    ExclusiveMonitor& Monitor() {
-        return *cpu_exclusive_monitor;
-    }
+    const VideoCore::RendererBase& Renderer() const;
 
     /// 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;
-    }
+    Kernel::SharedPtr<Kernel::Process>& CurrentProcess();
+
+    /// Provides a reference to the kernel instance.
+    Kernel::KernelCore& Kernel();
+
+    /// Provides a constant reference to the kernel instance.
+    const Kernel::KernelCore& Kernel() const;
+
+    /// Provides a reference to the internal PerfStats instance.
+    Core::PerfStats& GetPerfStats();
+
+    /// Provides a constant reference to the internal PerfStats instance.
+    const Core::PerfStats& GetPerfStats() const;
+
+    /// Provides a reference to the frame limiter;
+    Core::FrameLimiter& FrameLimiter();
+
+    /// Provides a constant referent to the frame limiter
+    const Core::FrameLimiter& FrameLimiter() const;
 
     /// Gets the name of the current game
-    Loader::ResultStatus GetGameName(std::string& out) const {
-        if (app_loader == nullptr)
-            return Loader::ResultStatus::ErrorNotInitialized;
-        return app_loader->ReadTitle(out);
-    }
+    Loader::ResultStatus GetGameName(std::string& out) const;
 
-    PerfStats perf_stats;
-    FrameLimiter frame_limiter;
+    void SetStatus(ResultStatus new_status, const char* details);
 
-    void SetStatus(ResultStatus new_status, const char* details = nullptr) {
-        status = new_status;
-        if (details) {
-            status_details = details;
-        }
-    }
+    const std::string& GetStatusDetails() const;
 
-    const std::string& GetStatusDetails() const {
-        return status_details;
-    }
-
-    Loader::AppLoader& GetAppLoader() const {
-        return *app_loader;
-    }
+    Loader::AppLoader& GetAppLoader() const;
 
     Service::SM::ServiceManager& ServiceManager();
     const Service::SM::ServiceManager& ServiceManager() const;
 
-    void SetGPUDebugContext(std::shared_ptr<Tegra::DebugContext> context) {
-        debug_context = std::move(context);
-    }
+    void SetGPUDebugContext(std::shared_ptr<Tegra::DebugContext> context);
 
-    std::shared_ptr<Tegra::DebugContext> GetGPUDebugContext() const {
-        return debug_context;
-    }
+    std::shared_ptr<Tegra::DebugContext> GetGPUDebugContext() const;
 
-    void SetFilesystem(FileSys::VirtualFilesystem vfs) {
-        virtual_filesystem = std::move(vfs);
-    }
+    void SetFilesystem(std::shared_ptr<FileSys::VfsFilesystem> vfs);
 
-    FileSys::VirtualFilesystem GetFilesystem() const {
-        return virtual_filesystem;
-    }
+    std::shared_ptr<FileSys::VfsFilesystem> GetFilesystem() const;
 
 private:
     System();
@@ -246,33 +229,10 @@ private:
      */
     ResultStatus Init(Frontend::EmuWindow& emu_window);
 
-    /// RealVfsFilesystem instance
-    FileSys::VirtualFilesystem virtual_filesystem;
-    /// AppLoader used to load the current executing application
-    std::unique_ptr<Loader::AppLoader> app_loader;
-    std::unique_ptr<VideoCore::RendererBase> renderer;
-    std::unique_ptr<Tegra::GPU> gpu_core;
-    std::shared_ptr<Tegra::DebugContext> debug_context;
-    Kernel::SharedPtr<Kernel::Process> current_process;
-    std::shared_ptr<ExclusiveMonitor> cpu_exclusive_monitor;
-    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;
-
-    /// Telemetry session for this emulation session
-    std::unique_ptr<Core::TelemetrySession> telemetry_session;
+    struct Impl;
+    std::unique_ptr<Impl> impl;
 
     static System s_instance;
-
-    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& CurrentArmInterface() {

src/core/crypto/aes_util.cpp

@@ -82,11 +82,25 @@ void AESCipher<Key, KeySize>::Transcode(const u8* src, size_t size, u8* dest, Op
         }
     } else {
         const auto block_size = mbedtls_cipher_get_block_size(context);
+        if (size < block_size) {
+            std::vector<u8> block(block_size);
+            std::memcpy(block.data(), src, size);
+            Transcode(block.data(), block.size(), block.data(), op);
+            std::memcpy(dest, block.data(), size);
+            return;
+        }
 
         for (size_t offset = 0; offset < size; offset += block_size) {
             auto length = std::min<size_t>(block_size, size - offset);
             mbedtls_cipher_update(context, src + offset, length, dest + offset, &written);
             if (written != length) {
+                if (length < block_size) {
+                    std::vector<u8> block(block_size);
+                    std::memcpy(block.data(), src + offset, length);
+                    Transcode(block.data(), block.size(), block.data(), op);
+                    std::memcpy(dest + offset, block.data(), length);
+                    return;
+                }
                 LOG_WARNING(Crypto, "Not all data was decrypted requested={:016X}, actual={:016X}.",
                             length, written);
             }

src/core/crypto/ctr_encryption_layer.cpp

@@ -21,7 +21,7 @@ size_t CTREncryptionLayer::Read(u8* data, size_t length, size_t offset) const {
         UpdateIV(base_offset + offset);
         std::vector<u8> raw = base->ReadBytes(length, offset);
         cipher.Transcode(raw.data(), raw.size(), data, Op::Decrypt);
-        return raw.size();
+        return length;
     }
 
     // offset does not fall on block boundary (0x10)

src/core/crypto/key_manager.cpp

@@ -228,18 +228,28 @@ void KeyManager::WriteKeyToFile(bool title_key, std::string_view keyname,
 }
 
 void KeyManager::SetKey(S128KeyType id, Key128 key, u64 field1, u64 field2) {
-    const auto iter = std::find_if(
+    if (s128_keys.find({id, field1, field2}) != s128_keys.end())
+        return;
+    if (id == S128KeyType::Titlekey) {
+        Key128 rights_id;
+        std::memcpy(rights_id.data(), &field2, sizeof(u64));
+        std::memcpy(rights_id.data() + sizeof(u64), &field1, sizeof(u64));
+        WriteKeyToFile(true, Common::HexArrayToString(rights_id), key);
+    }
+    const auto iter2 = std::find_if(
         s128_file_id.begin(), s128_file_id.end(),
         [&id, &field1, &field2](const std::pair<std::string, KeyIndex<S128KeyType>> elem) {
             return std::tie(elem.second.type, elem.second.field1, elem.second.field2) ==
                    std::tie(id, field1, field2);
         });
-    if (iter != s128_file_id.end())
-        WriteKeyToFile(id == S128KeyType::Titlekey, iter->first, key);
+    if (iter2 != s128_file_id.end())
+        WriteKeyToFile(false, iter2->first, key);
     s128_keys[{id, field1, field2}] = key;
 }
 
 void KeyManager::SetKey(S256KeyType id, Key256 key, u64 field1, u64 field2) {
+    if (s256_keys.find({id, field1, field2}) != s256_keys.end())
+        return;
     const auto iter = std::find_if(
         s256_file_id.begin(), s256_file_id.end(),
         [&id, &field1, &field2](const std::pair<std::string, KeyIndex<S256KeyType>> elem) {

src/core/crypto/key_manager.h

@@ -16,6 +16,8 @@
 
 namespace Core::Crypto {
 
+constexpr u64 TICKET_FILE_TITLEKEY_OFFSET = 0x180;
+
 using Key128 = std::array<u8, 0x10>;
 using Key256 = std::array<u8, 0x20>;
 using SHA256Hash = std::array<u8, 0x20>;

src/core/file_sys/card_image.cpp

@@ -9,7 +9,9 @@
 
 #include "common/logging/log.h"
 #include "core/file_sys/card_image.h"
+#include "core/file_sys/nca_metadata.h"
 #include "core/file_sys/partition_filesystem.h"
+#include "core/file_sys/submission_package.h"
 #include "core/file_sys/vfs_offset.h"
 #include "core/loader/loader.h"
 
@@ -43,15 +45,19 @@ XCI::XCI(VirtualFile file_) : file(std::move(file_)), partitions(0x4) {
             partitions[static_cast<size_t>(partition)] = std::make_shared<PartitionFilesystem>(raw);
     }
 
+    secure_partition = std::make_shared<NSP>(
+        main_hfs.GetFile(partition_names[static_cast<size_t>(XCIPartition::Secure)]));
+
+    const auto secure_ncas = secure_partition->GetNCAsCollapsed();
+    std::copy(secure_ncas.begin(), secure_ncas.end(), std::back_inserter(ncas));
+
     program_nca_status = Loader::ResultStatus::ErrorXCIMissingProgramNCA;
+    program =
+        secure_partition->GetNCA(secure_partition->GetProgramTitleID(), ContentRecordType::Program);
+    if (program != nullptr)
+        program_nca_status = program->GetStatus();
 
-    auto result = AddNCAFromPartition(XCIPartition::Secure);
-    if (result != Loader::ResultStatus::Success) {
-        status = result;
-        return;
-    }
-
-    result = AddNCAFromPartition(XCIPartition::Update);
+    auto result = AddNCAFromPartition(XCIPartition::Update);
     if (result != Loader::ResultStatus::Success) {
         status = result;
         return;
@@ -74,6 +80,8 @@ XCI::XCI(VirtualFile file_) : file(std::move(file_)), partitions(0x4) {
     status = Loader::ResultStatus::Success;
 }
 
+XCI::~XCI() = default;
+
 Loader::ResultStatus XCI::GetStatus() const {
     return status;
 }
@@ -86,6 +94,10 @@ VirtualDir XCI::GetPartition(XCIPartition partition) const {
     return partitions[static_cast<size_t>(partition)];
 }
 
+std::shared_ptr<NSP> XCI::GetSecurePartitionNSP() const {
+    return secure_partition;
+}
+
 VirtualDir XCI::GetSecurePartition() const {
     return GetPartition(XCIPartition::Secure);
 }
@@ -102,6 +114,16 @@ VirtualDir XCI::GetLogoPartition() const {
     return GetPartition(XCIPartition::Logo);
 }
 
+std::shared_ptr<NCA> XCI::GetProgramNCA() const {
+    return program;
+}
+
+VirtualFile XCI::GetProgramNCAFile() const {
+    if (GetProgramNCA() == nullptr)
+        return nullptr;
+    return GetProgramNCA()->GetBaseFile();
+}
+
 const std::vector<std::shared_ptr<NCA>>& XCI::GetNCAs() const {
     return ncas;
 }

src/core/file_sys/card_image.h

@@ -14,6 +14,8 @@
 
 namespace FileSys {
 
+class NSP;
+
 enum class GamecardSize : u8 {
     S_1GB = 0xFA,
     S_2GB = 0xF8,
@@ -57,6 +59,7 @@ enum class XCIPartition : u8 { Update, Normal, Secure, Logo };
 class XCI : public ReadOnlyVfsDirectory {
 public:
     explicit XCI(VirtualFile file);
+    ~XCI() override;
 
     Loader::ResultStatus GetStatus() const;
     Loader::ResultStatus GetProgramNCAStatus() const;
@@ -64,11 +67,14 @@ public:
     u8 GetFormatVersion() const;
 
     VirtualDir GetPartition(XCIPartition partition) const;
+    std::shared_ptr<NSP> GetSecurePartitionNSP() const;
     VirtualDir GetSecurePartition() const;
     VirtualDir GetNormalPartition() const;
     VirtualDir GetUpdatePartition() const;
     VirtualDir GetLogoPartition() const;
 
+    std::shared_ptr<NCA> GetProgramNCA() const;
+    VirtualFile GetProgramNCAFile() const;
     const std::vector<std::shared_ptr<NCA>>& GetNCAs() const;
     std::shared_ptr<NCA> GetNCAByType(NCAContentType type) const;
     VirtualFile GetNCAFileByType(NCAContentType type) const;
@@ -94,6 +100,8 @@ private:
     Loader::ResultStatus program_nca_status;
 
     std::vector<VirtualDir> partitions;
+    std::shared_ptr<NSP> secure_partition;
+    std::shared_ptr<NCA> program;
     std::vector<std::shared_ptr<NCA>> ncas;
 };
 } // namespace FileSys

src/core/file_sys/content_archive.cpp

@@ -9,7 +9,7 @@
 #include "core/crypto/aes_util.h"
 #include "core/crypto/ctr_encryption_layer.h"
 #include "core/file_sys/content_archive.h"
-#include "core/file_sys/romfs.h"
+#include "core/file_sys/nca_patch.h"
 #include "core/file_sys/vfs_offset.h"
 #include "core/loader/loader.h"
 
@@ -64,10 +64,31 @@ struct RomFSSuperblock {
 };
 static_assert(sizeof(RomFSSuperblock) == 0x200, "RomFSSuperblock has incorrect size.");
 
+struct BKTRHeader {
+    u64_le offset;
+    u64_le size;
+    u32_le magic;
+    INSERT_PADDING_BYTES(0x4);
+    u32_le number_entries;
+    INSERT_PADDING_BYTES(0x4);
+};
+static_assert(sizeof(BKTRHeader) == 0x20, "BKTRHeader has incorrect size.");
+
+struct BKTRSuperblock {
+    NCASectionHeaderBlock header_block;
+    IVFCHeader ivfc;
+    INSERT_PADDING_BYTES(0x18);
+    BKTRHeader relocation;
+    BKTRHeader subsection;
+    INSERT_PADDING_BYTES(0xC0);
+};
+static_assert(sizeof(BKTRSuperblock) == 0x200, "BKTRSuperblock has incorrect size.");
+
 union NCASectionHeader {
     NCASectionRaw raw;
     PFS0Superblock pfs0;
     RomFSSuperblock romfs;
+    BKTRSuperblock bktr;
 };
 static_assert(sizeof(NCASectionHeader) == 0x200, "NCASectionHeader has incorrect size.");
 
@@ -100,7 +121,7 @@ boost::optional<Core::Crypto::Key128> NCA::GetKeyAreaKey(NCASectionCryptoType ty
     Core::Crypto::Key128 out;
     if (type == NCASectionCryptoType::XTS)
         std::copy(key_area.begin(), key_area.begin() + 0x10, out.begin());
-    else if (type == NCASectionCryptoType::CTR)
+    else if (type == NCASectionCryptoType::CTR || type == NCASectionCryptoType::BKTR)
         std::copy(key_area.begin() + 0x20, key_area.begin() + 0x30, out.begin());
     else
         LOG_CRITICAL(Crypto, "Called GetKeyAreaKey on invalid NCASectionCryptoType type={:02X}",
@@ -150,6 +171,9 @@ VirtualFile NCA::Decrypt(NCASectionHeader s_header, VirtualFile in, u64 starting
         LOG_DEBUG(Crypto, "called with mode=NONE");
         return in;
     case NCASectionCryptoType::CTR:
+    // During normal BKTR decryption, this entire function is skipped. This is for the metadata,
+    // which uses the same CTR as usual.
+    case NCASectionCryptoType::BKTR:
         LOG_DEBUG(Crypto, "called with mode=CTR, starting_offset={:016X}", starting_offset);
         {
             boost::optional<Core::Crypto::Key128> key = boost::none;
@@ -186,7 +210,9 @@ VirtualFile NCA::Decrypt(NCASectionHeader s_header, VirtualFile in, u64 starting
     }
 }
 
-NCA::NCA(VirtualFile file_) : file(std::move(file_)) {
+NCA::NCA(VirtualFile file_, VirtualFile bktr_base_romfs_, u64 bktr_base_ivfc_offset)
+    : file(std::move(file_)),
+      bktr_base_romfs(bktr_base_romfs_ ? std::move(bktr_base_romfs_) : nullptr) {
     status = Loader::ResultStatus::Success;
 
     if (file == nullptr) {
@@ -261,22 +287,21 @@ NCA::NCA(VirtualFile file_) : file(std::move(file_)) {
     is_update = std::find_if(sections.begin(), sections.end(), [](const NCASectionHeader& header) {
                     return header.raw.header.crypto_type == NCASectionCryptoType::BKTR;
                 }) != sections.end();
+    ivfc_offset = 0;
 
     for (std::ptrdiff_t i = 0; i < number_sections; ++i) {
         auto section = sections[i];
 
         if (section.raw.header.filesystem_type == NCASectionFilesystemType::ROMFS) {
-            const size_t romfs_offset =
-                header.section_tables[i].media_offset * MEDIA_OFFSET_MULTIPLIER +
-                section.romfs.ivfc.levels[IVFC_MAX_LEVEL - 1].offset;
+            const size_t base_offset =
+                header.section_tables[i].media_offset * MEDIA_OFFSET_MULTIPLIER;
+            ivfc_offset = section.romfs.ivfc.levels[IVFC_MAX_LEVEL - 1].offset;
+            const size_t romfs_offset = base_offset + ivfc_offset;
             const size_t romfs_size = section.romfs.ivfc.levels[IVFC_MAX_LEVEL - 1].size;
-            auto dec =
-                Decrypt(section, std::make_shared<OffsetVfsFile>(file, romfs_size, romfs_offset),
-                        romfs_offset);
-            if (dec != nullptr) {
-                files.push_back(std::move(dec));
-                romfs = files.back();
-            } else {
+            auto raw = std::make_shared<OffsetVfsFile>(file, romfs_size, romfs_offset);
+            auto dec = Decrypt(section, raw, romfs_offset);
+
+            if (dec == nullptr) {
                 if (status != Loader::ResultStatus::Success)
                     return;
                 if (has_rights_id)
@@ -285,6 +310,120 @@ NCA::NCA(VirtualFile file_) : file(std::move(file_)) {
                     status = Loader::ResultStatus::ErrorIncorrectKeyAreaKey;
                 return;
             }
+
+            if (section.raw.header.crypto_type == NCASectionCryptoType::BKTR) {
+                if (section.bktr.relocation.magic != Common::MakeMagic('B', 'K', 'T', 'R') ||
+                    section.bktr.subsection.magic != Common::MakeMagic('B', 'K', 'T', 'R')) {
+                    status = Loader::ResultStatus::ErrorBadBKTRHeader;
+                    return;
+                }
+
+                if (section.bktr.relocation.offset + section.bktr.relocation.size !=
+                    section.bktr.subsection.offset) {
+                    status = Loader::ResultStatus::ErrorBKTRSubsectionNotAfterRelocation;
+                    return;
+                }
+
+                const u64 size =
+                    MEDIA_OFFSET_MULTIPLIER * (header.section_tables[i].media_end_offset -
+                                               header.section_tables[i].media_offset);
+                if (section.bktr.subsection.offset + section.bktr.subsection.size != size) {
+                    status = Loader::ResultStatus::ErrorBKTRSubsectionNotAtEnd;
+                    return;
+                }
+
+                const u64 offset = section.romfs.ivfc.levels[IVFC_MAX_LEVEL - 1].offset;
+                RelocationBlock relocation_block{};
+                if (dec->ReadObject(&relocation_block, section.bktr.relocation.offset - offset) !=
+                    sizeof(RelocationBlock)) {
+                    status = Loader::ResultStatus::ErrorBadRelocationBlock;
+                    return;
+                }
+                SubsectionBlock subsection_block{};
+                if (dec->ReadObject(&subsection_block, section.bktr.subsection.offset - offset) !=
+                    sizeof(RelocationBlock)) {
+                    status = Loader::ResultStatus::ErrorBadSubsectionBlock;
+                    return;
+                }
+
+                std::vector<RelocationBucketRaw> relocation_buckets_raw(
+                    (section.bktr.relocation.size - sizeof(RelocationBlock)) /
+                    sizeof(RelocationBucketRaw));
+                if (dec->ReadBytes(relocation_buckets_raw.data(),
+                                   section.bktr.relocation.size - sizeof(RelocationBlock),
+                                   section.bktr.relocation.offset + sizeof(RelocationBlock) -
+                                       offset) !=
+                    section.bktr.relocation.size - sizeof(RelocationBlock)) {
+                    status = Loader::ResultStatus::ErrorBadRelocationBuckets;
+                    return;
+                }
+
+                std::vector<SubsectionBucketRaw> subsection_buckets_raw(
+                    (section.bktr.subsection.size - sizeof(SubsectionBlock)) /
+                    sizeof(SubsectionBucketRaw));
+                if (dec->ReadBytes(subsection_buckets_raw.data(),
+                                   section.bktr.subsection.size - sizeof(SubsectionBlock),
+                                   section.bktr.subsection.offset + sizeof(SubsectionBlock) -
+                                       offset) !=
+                    section.bktr.subsection.size - sizeof(SubsectionBlock)) {
+                    status = Loader::ResultStatus::ErrorBadSubsectionBuckets;
+                    return;
+                }
+
+                std::vector<RelocationBucket> relocation_buckets(relocation_buckets_raw.size());
+                std::transform(relocation_buckets_raw.begin(), relocation_buckets_raw.end(),
+                               relocation_buckets.begin(), &ConvertRelocationBucketRaw);
+                std::vector<SubsectionBucket> subsection_buckets(subsection_buckets_raw.size());
+                std::transform(subsection_buckets_raw.begin(), subsection_buckets_raw.end(),
+                               subsection_buckets.begin(), &ConvertSubsectionBucketRaw);
+
+                u32 ctr_low;
+                std::memcpy(&ctr_low, section.raw.section_ctr.data(), sizeof(ctr_low));
+                subsection_buckets.back().entries.push_back(
+                    {section.bktr.relocation.offset, {0}, ctr_low});
+                subsection_buckets.back().entries.push_back({size, {0}, 0});
+
+                boost::optional<Core::Crypto::Key128> key = boost::none;
+                if (encrypted) {
+                    if (has_rights_id) {
+                        status = Loader::ResultStatus::Success;
+                        key = GetTitlekey();
+                        if (key == boost::none) {
+                            status = Loader::ResultStatus::ErrorMissingTitlekey;
+                            return;
+                        }
+                    } else {
+                        key = GetKeyAreaKey(NCASectionCryptoType::BKTR);
+                        if (key == boost::none) {
+                            status = Loader::ResultStatus::ErrorMissingKeyAreaKey;
+                            return;
+                        }
+                    }
+                }
+
+                if (bktr_base_romfs == nullptr) {
+                    status = Loader::ResultStatus::ErrorMissingBKTRBaseRomFS;
+                    return;
+                }
+
+                auto bktr = std::make_shared<BKTR>(
+                    bktr_base_romfs, std::make_shared<OffsetVfsFile>(file, romfs_size, base_offset),
+                    relocation_block, relocation_buckets, subsection_block, subsection_buckets,
+                    encrypted, encrypted ? key.get() : Core::Crypto::Key128{}, base_offset,
+                    bktr_base_ivfc_offset, section.raw.section_ctr);
+
+                // BKTR applies to entire IVFC, so make an offset version to level 6
+
+                files.push_back(std::make_shared<OffsetVfsFile>(
+                    bktr, romfs_size, section.romfs.ivfc.levels[IVFC_MAX_LEVEL - 1].offset));
+                romfs = files.back();
+            } else {
+                files.push_back(std::move(dec));
+                romfs = files.back();
+                const u64 raw_size =
+                    MEDIA_OFFSET_MULTIPLIER * (header.section_tables[i].media_end_offset -
+                                               header.section_tables[i].media_offset);
+            }
         } else if (section.raw.header.filesystem_type == NCASectionFilesystemType::PFS0) {
             u64 offset = (static_cast<u64>(header.section_tables[i].media_offset) *
                           MEDIA_OFFSET_MULTIPLIER) +
@@ -345,11 +484,17 @@ NCAContentType NCA::GetType() const {
 }
 
 u64 NCA::GetTitleId() const {
+    if (is_update || status == Loader::ResultStatus::ErrorMissingBKTRBaseRomFS)
+        return header.title_id | 0x800;
     if (status != Loader::ResultStatus::Success)
         return {};
     return header.title_id;
 }
 
+bool NCA::IsUpdate() const {
+    return is_update;
+}
+
 VirtualFile NCA::GetRomFS() const {
     return romfs;
 }
@@ -362,8 +507,8 @@ VirtualFile NCA::GetBaseFile() const {
     return file;
 }
 
-bool NCA::IsUpdate() const {
-    return is_update;
+u64 NCA::GetBaseIVFCOffset() const {
+    return ivfc_offset;
 }
 
 bool NCA::ReplaceFileWithSubdirectory(VirtualFile file, VirtualDir dir) {

src/core/file_sys/content_archive.h

@@ -15,6 +15,7 @@
 #include "control_metadata.h"
 #include "core/crypto/key_manager.h"
 #include "core/file_sys/partition_filesystem.h"
+#include "core/file_sys/romfs.h"
 #include "core/loader/loader.h"
 
 namespace FileSys {
@@ -77,7 +78,8 @@ bool IsValidNCA(const NCAHeader& header);
 // After construction, use GetStatus to determine if the file is valid and ready to be used.
 class NCA : public ReadOnlyVfsDirectory {
 public:
-    explicit NCA(VirtualFile file);
+    explicit NCA(VirtualFile file, VirtualFile bktr_base_romfs = nullptr,
+                 u64 bktr_base_ivfc_offset = 0);
     Loader::ResultStatus GetStatus() const;
 
     std::vector<std::shared_ptr<VfsFile>> GetFiles() const override;
@@ -87,13 +89,15 @@ public:
 
     NCAContentType GetType() const;
     u64 GetTitleId() const;
+    bool IsUpdate() const;
 
     VirtualFile GetRomFS() const;
     VirtualDir GetExeFS() const;
 
     VirtualFile GetBaseFile() const;
 
-    bool IsUpdate() const;
+    // Returns the base ivfc offset used in BKTR patching.
+    u64 GetBaseIVFCOffset() const;
 
 protected:
     bool ReplaceFileWithSubdirectory(VirtualFile file, VirtualDir dir) override;
@@ -110,14 +114,16 @@ private:
     VirtualFile romfs = nullptr;
     VirtualDir exefs = nullptr;
     VirtualFile file;
+    VirtualFile bktr_base_romfs;
+    u64 ivfc_offset;
 
     NCAHeader header{};
     bool has_rights_id{};
-    bool is_update{};
 
     Loader::ResultStatus status{};
 
     bool encrypted;
+    bool is_update;
 
     Core::Crypto::KeyManager keys;
 };

src/core/file_sys/nca_patch.cpp

@@ -0,0 +1,206 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/assert.h"
+#include "core/crypto/aes_util.h"
+#include "core/file_sys/nca_patch.h"
+
+namespace FileSys {
+
+BKTR::BKTR(VirtualFile base_romfs_, VirtualFile bktr_romfs_, RelocationBlock relocation_,
+           std::vector<RelocationBucket> relocation_buckets_, SubsectionBlock subsection_,
+           std::vector<SubsectionBucket> subsection_buckets_, bool is_encrypted_,
+           Core::Crypto::Key128 key_, u64 base_offset_, u64 ivfc_offset_,
+           std::array<u8, 8> section_ctr_)
+    : base_romfs(std::move(base_romfs_)), bktr_romfs(std::move(bktr_romfs_)),
+      relocation(relocation_), relocation_buckets(std::move(relocation_buckets_)),
+      subsection(subsection_), subsection_buckets(std::move(subsection_buckets_)),
+      encrypted(is_encrypted_), key(key_), base_offset(base_offset_), ivfc_offset(ivfc_offset_),
+      section_ctr(section_ctr_) {
+    for (size_t i = 0; i < relocation.number_buckets - 1; ++i) {
+        relocation_buckets[i].entries.push_back({relocation.base_offsets[i + 1], 0, 0});
+    }
+
+    for (size_t i = 0; i < subsection.number_buckets - 1; ++i) {
+        subsection_buckets[i].entries.push_back({subsection_buckets[i + 1].entries[0].address_patch,
+                                                 {0},
+                                                 subsection_buckets[i + 1].entries[0].ctr});
+    }
+
+    relocation_buckets.back().entries.push_back({relocation.size, 0, 0});
+}
+
+BKTR::~BKTR() = default;
+
+size_t BKTR::Read(u8* data, size_t length, size_t offset) const {
+    // Read out of bounds.
+    if (offset >= relocation.size)
+        return 0;
+    const auto relocation = GetRelocationEntry(offset);
+    const auto section_offset = offset - relocation.address_patch + relocation.address_source;
+    const auto bktr_read = relocation.from_patch;
+
+    const auto next_relocation = GetNextRelocationEntry(offset);
+
+    if (offset + length > next_relocation.address_patch) {
+        const u64 partition = next_relocation.address_patch - offset;
+        return Read(data, partition, offset) +
+               Read(data + partition, length - partition, offset + partition);
+    }
+
+    if (!bktr_read) {
+        ASSERT_MSG(section_offset > ivfc_offset, "Offset calculation negative.");
+        return base_romfs->Read(data, length, section_offset - ivfc_offset);
+    }
+
+    if (!encrypted) {
+        return bktr_romfs->Read(data, length, section_offset);
+    }
+
+    const auto subsection = GetSubsectionEntry(section_offset);
+    Core::Crypto::AESCipher<Core::Crypto::Key128> cipher(key, Core::Crypto::Mode::CTR);
+
+    // Calculate AES IV
+    std::vector<u8> iv(16);
+    auto subsection_ctr = subsection.ctr;
+    auto offset_iv = section_offset + base_offset;
+    for (size_t i = 0; i < section_ctr.size(); ++i)
+        iv[i] = section_ctr[0x8 - i - 1];
+    offset_iv >>= 4;
+    for (size_t i = 0; i < sizeof(u64); ++i) {
+        iv[0xF - i] = static_cast<u8>(offset_iv & 0xFF);
+        offset_iv >>= 8;
+    }
+    for (size_t i = 0; i < sizeof(u32); ++i) {
+        iv[0x7 - i] = static_cast<u8>(subsection_ctr & 0xFF);
+        subsection_ctr >>= 8;
+    }
+    cipher.SetIV(iv);
+
+    const auto next_subsection = GetNextSubsectionEntry(section_offset);
+
+    if (section_offset + length > next_subsection.address_patch) {
+        const u64 partition = next_subsection.address_patch - section_offset;
+        return Read(data, partition, offset) +
+               Read(data + partition, length - partition, offset + partition);
+    }
+
+    const auto block_offset = section_offset & 0xF;
+    if (block_offset != 0) {
+        auto block = bktr_romfs->ReadBytes(0x10, section_offset & ~0xF);
+        cipher.Transcode(block.data(), block.size(), block.data(), Core::Crypto::Op::Decrypt);
+        if (length + block_offset < 0x10) {
+            std::memcpy(data, block.data() + block_offset, std::min(length, block.size()));
+            return std::min(length, block.size());
+        }
+
+        const auto read = 0x10 - block_offset;
+        std::memcpy(data, block.data() + block_offset, read);
+        return read + Read(data + read, length - read, offset + read);
+    }
+
+    const auto raw_read = bktr_romfs->Read(data, length, section_offset);
+    cipher.Transcode(data, raw_read, data, Core::Crypto::Op::Decrypt);
+    return raw_read;
+}
+
+template <bool Subsection, typename BlockType, typename BucketType>
+std::pair<size_t, size_t> BKTR::SearchBucketEntry(u64 offset, BlockType block,
+                                                  BucketType buckets) const {
+    if constexpr (Subsection) {
+        const auto last_bucket = buckets[block.number_buckets - 1];
+        if (offset >= last_bucket.entries[last_bucket.number_entries].address_patch)
+            return {block.number_buckets - 1, last_bucket.number_entries};
+    } else {
+        ASSERT_MSG(offset <= block.size, "Offset is out of bounds in BKTR relocation block.");
+    }
+
+    size_t bucket_id = std::count_if(block.base_offsets.begin() + 1,
+                                     block.base_offsets.begin() + block.number_buckets,
+                                     [&offset](u64 base_offset) { return base_offset < offset; });
+
+    const auto bucket = buckets[bucket_id];
+
+    if (bucket.number_entries == 1)
+        return {bucket_id, 0};
+
+    size_t low = 0;
+    size_t mid = 0;
+    size_t high = bucket.number_entries - 1;
+    while (low <= high) {
+        mid = (low + high) / 2;
+        if (bucket.entries[mid].address_patch > offset) {
+            high = mid - 1;
+        } else {
+            if (mid == bucket.number_entries - 1 ||
+                bucket.entries[mid + 1].address_patch > offset) {
+                return {bucket_id, mid};
+            }
+
+            low = mid + 1;
+        }
+    }
+
+    UNREACHABLE_MSG("Offset could not be found in BKTR block.");
+}
+
+RelocationEntry BKTR::GetRelocationEntry(u64 offset) const {
+    const auto res = SearchBucketEntry<false>(offset, relocation, relocation_buckets);
+    return relocation_buckets[res.first].entries[res.second];
+}
+
+RelocationEntry BKTR::GetNextRelocationEntry(u64 offset) const {
+    const auto res = SearchBucketEntry<false>(offset, relocation, relocation_buckets);
+    const auto bucket = relocation_buckets[res.first];
+    if (res.second + 1 < bucket.entries.size())
+        return bucket.entries[res.second + 1];
+    return relocation_buckets[res.first + 1].entries[0];
+}
+
+SubsectionEntry BKTR::GetSubsectionEntry(u64 offset) const {
+    const auto res = SearchBucketEntry<true>(offset, subsection, subsection_buckets);
+    return subsection_buckets[res.first].entries[res.second];
+}
+
+SubsectionEntry BKTR::GetNextSubsectionEntry(u64 offset) const {
+    const auto res = SearchBucketEntry<true>(offset, subsection, subsection_buckets);
+    const auto bucket = subsection_buckets[res.first];
+    if (res.second + 1 < bucket.entries.size())
+        return bucket.entries[res.second + 1];
+    return subsection_buckets[res.first + 1].entries[0];
+}
+
+std::string BKTR::GetName() const {
+    return base_romfs->GetName();
+}
+
+size_t BKTR::GetSize() const {
+    return relocation.size;
+}
+
+bool BKTR::Resize(size_t new_size) {
+    return false;
+}
+
+std::shared_ptr<VfsDirectory> BKTR::GetContainingDirectory() const {
+    return base_romfs->GetContainingDirectory();
+}
+
+bool BKTR::IsWritable() const {
+    return false;
+}
+
+bool BKTR::IsReadable() const {
+    return true;
+}
+
+size_t BKTR::Write(const u8* data, size_t length, size_t offset) {
+    return 0;
+}
+
+bool BKTR::Rename(std::string_view name) {
+    return base_romfs->Rename(name);
+}
+
+} // namespace FileSys

src/core/file_sys/nca_patch.h

@@ -0,0 +1,147 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <vector>
+#include <common/common_funcs.h>
+#include "core/crypto/key_manager.h"
+#include "core/file_sys/romfs.h"
+
+namespace FileSys {
+
+#pragma pack(push, 1)
+struct RelocationEntry {
+    u64_le address_patch;
+    u64_le address_source;
+    u32 from_patch;
+};
+#pragma pack(pop)
+static_assert(sizeof(RelocationEntry) == 0x14, "RelocationEntry has incorrect size.");
+
+struct RelocationBucketRaw {
+    INSERT_PADDING_BYTES(4);
+    u32_le number_entries;
+    u64_le end_offset;
+    std::array<RelocationEntry, 0x332> relocation_entries;
+    INSERT_PADDING_BYTES(8);
+};
+static_assert(sizeof(RelocationBucketRaw) == 0x4000, "RelocationBucketRaw has incorrect size.");
+
+// Vector version of RelocationBucketRaw
+struct RelocationBucket {
+    u32 number_entries;
+    u64 end_offset;
+    std::vector<RelocationEntry> entries;
+};
+
+struct RelocationBlock {
+    INSERT_PADDING_BYTES(4);
+    u32_le number_buckets;
+    u64_le size;
+    std::array<u64, 0x7FE> base_offsets;
+};
+static_assert(sizeof(RelocationBlock) == 0x4000, "RelocationBlock has incorrect size.");
+
+struct SubsectionEntry {
+    u64_le address_patch;
+    INSERT_PADDING_BYTES(0x4);
+    u32_le ctr;
+};
+static_assert(sizeof(SubsectionEntry) == 0x10, "SubsectionEntry has incorrect size.");
+
+struct SubsectionBucketRaw {
+    INSERT_PADDING_BYTES(4);
+    u32_le number_entries;
+    u64_le end_offset;
+    std::array<SubsectionEntry, 0x3FF> subsection_entries;
+};
+static_assert(sizeof(SubsectionBucketRaw) == 0x4000, "SubsectionBucketRaw has incorrect size.");
+
+// Vector version of SubsectionBucketRaw
+struct SubsectionBucket {
+    u32 number_entries;
+    u64 end_offset;
+    std::vector<SubsectionEntry> entries;
+};
+
+struct SubsectionBlock {
+    INSERT_PADDING_BYTES(4);
+    u32_le number_buckets;
+    u64_le size;
+    std::array<u64, 0x7FE> base_offsets;
+};
+static_assert(sizeof(SubsectionBlock) == 0x4000, "SubsectionBlock has incorrect size.");
+
+inline RelocationBucket ConvertRelocationBucketRaw(RelocationBucketRaw raw) {
+    return {raw.number_entries,
+            raw.end_offset,
+            {raw.relocation_entries.begin(), raw.relocation_entries.begin() + raw.number_entries}};
+}
+
+inline SubsectionBucket ConvertSubsectionBucketRaw(SubsectionBucketRaw raw) {
+    return {raw.number_entries,
+            raw.end_offset,
+            {raw.subsection_entries.begin(), raw.subsection_entries.begin() + raw.number_entries}};
+}
+
+class BKTR : public VfsFile {
+public:
+    BKTR(VirtualFile base_romfs, VirtualFile bktr_romfs, RelocationBlock relocation,
+         std::vector<RelocationBucket> relocation_buckets, SubsectionBlock subsection,
+         std::vector<SubsectionBucket> subsection_buckets, bool is_encrypted,
+         Core::Crypto::Key128 key, u64 base_offset, u64 ivfc_offset, std::array<u8, 8> section_ctr);
+    ~BKTR() override;
+
+    size_t Read(u8* data, size_t length, size_t offset) const override;
+
+    std::string GetName() const override;
+
+    size_t GetSize() const override;
+
+    bool Resize(size_t new_size) override;
+
+    std::shared_ptr<VfsDirectory> GetContainingDirectory() const override;
+
+    bool IsWritable() const override;
+
+    bool IsReadable() const override;
+
+    size_t Write(const u8* data, size_t length, size_t offset) override;
+
+    bool Rename(std::string_view name) override;
+
+private:
+    template <bool Subsection, typename BlockType, typename BucketType>
+    std::pair<size_t, size_t> SearchBucketEntry(u64 offset, BlockType block,
+                                                BucketType buckets) const;
+
+    RelocationEntry GetRelocationEntry(u64 offset) const;
+    RelocationEntry GetNextRelocationEntry(u64 offset) const;
+
+    SubsectionEntry GetSubsectionEntry(u64 offset) const;
+    SubsectionEntry GetNextSubsectionEntry(u64 offset) const;
+
+    RelocationBlock relocation;
+    std::vector<RelocationBucket> relocation_buckets;
+    SubsectionBlock subsection;
+    std::vector<SubsectionBucket> subsection_buckets;
+
+    // Should be the raw base romfs, decrypted.
+    VirtualFile base_romfs;
+    // Should be the raw BKTR romfs, (located at media_offset with size media_size).
+    VirtualFile bktr_romfs;
+
+    bool encrypted;
+    Core::Crypto::Key128 key;
+
+    // Base offset into NCA, used for IV calculation.
+    u64 base_offset;
+    // Distance between IVFC start and RomFS start, used for base reads
+    u64 ivfc_offset;
+    std::array<u8, 8> section_ctr;
+};
+
+} // namespace FileSys

src/core/file_sys/patch_manager.cpp

@@ -0,0 +1,115 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/file_sys/patch_manager.h"
+#include "core/file_sys/registered_cache.h"
+#include "core/hle/service/filesystem/filesystem.h"
+
+namespace FileSys {
+
+constexpr u64 SINGLE_BYTE_MODULUS = 0x100;
+
+std::string FormatTitleVersion(u32 version, TitleVersionFormat format) {
+    std::array<u8, sizeof(u32)> bytes{};
+    bytes[0] = version % SINGLE_BYTE_MODULUS;
+    for (size_t i = 1; i < bytes.size(); ++i) {
+        version /= SINGLE_BYTE_MODULUS;
+        bytes[i] = version % SINGLE_BYTE_MODULUS;
+    }
+
+    if (format == TitleVersionFormat::FourElements)
+        return fmt::format("v{}.{}.{}.{}", bytes[3], bytes[2], bytes[1], bytes[0]);
+    return fmt::format("v{}.{}.{}", bytes[3], bytes[2], bytes[1]);
+}
+
+constexpr std::array<const char*, 1> PATCH_TYPE_NAMES{
+    "Update",
+};
+
+std::string FormatPatchTypeName(PatchType type) {
+    return PATCH_TYPE_NAMES.at(static_cast<size_t>(type));
+}
+
+PatchManager::PatchManager(u64 title_id) : title_id(title_id) {}
+
+VirtualDir PatchManager::PatchExeFS(VirtualDir exefs) const {
+    LOG_INFO(Loader, "Patching ExeFS for title_id={:016X}", title_id);
+
+    if (exefs == nullptr)
+        return exefs;
+
+    const auto installed = Service::FileSystem::GetUnionContents();
+
+    // Game Updates
+    const auto update_tid = GetUpdateTitleID(title_id);
+    const auto update = installed->GetEntry(update_tid, ContentRecordType::Program);
+    if (update != nullptr) {
+        if (update->GetStatus() == Loader::ResultStatus::ErrorMissingBKTRBaseRomFS &&
+            update->GetExeFS() != nullptr) {
+            LOG_INFO(Loader, "    ExeFS: Update ({}) applied successfully",
+                     FormatTitleVersion(installed->GetEntryVersion(update_tid).get_value_or(0)));
+            exefs = update->GetExeFS();
+        }
+    }
+
+    return exefs;
+}
+
+VirtualFile PatchManager::PatchRomFS(VirtualFile romfs, u64 ivfc_offset,
+                                     ContentRecordType type) const {
+    LOG_INFO(Loader, "Patching RomFS for title_id={:016X}, type={:02X}", title_id,
+             static_cast<u8>(type));
+
+    if (romfs == nullptr)
+        return romfs;
+
+    const auto installed = Service::FileSystem::GetUnionContents();
+
+    // Game Updates
+    const auto update_tid = GetUpdateTitleID(title_id);
+    const auto update = installed->GetEntryRaw(update_tid, type);
+    if (update != nullptr) {
+        const auto new_nca = std::make_shared<NCA>(update, romfs, ivfc_offset);
+        if (new_nca->GetStatus() == Loader::ResultStatus::Success &&
+            new_nca->GetRomFS() != nullptr) {
+            LOG_INFO(Loader, "    RomFS: Update ({}) applied successfully",
+                     FormatTitleVersion(installed->GetEntryVersion(update_tid).get_value_or(0)));
+            romfs = new_nca->GetRomFS();
+        }
+    }
+
+    return romfs;
+}
+
+std::map<PatchType, std::string> PatchManager::GetPatchVersionNames() const {
+    std::map<PatchType, std::string> out;
+    const auto installed = Service::FileSystem::GetUnionContents();
+
+    const auto update_tid = GetUpdateTitleID(title_id);
+    const auto update_control = installed->GetEntry(title_id, ContentRecordType::Control);
+    if (update_control != nullptr) {
+        do {
+            const auto romfs =
+                PatchRomFS(update_control->GetRomFS(), update_control->GetBaseIVFCOffset(),
+                           FileSys::ContentRecordType::Control);
+            if (romfs == nullptr)
+                break;
+
+            const auto control_dir = FileSys::ExtractRomFS(romfs);
+            if (control_dir == nullptr)
+                break;
+
+            const auto nacp_file = control_dir->GetFile("control.nacp");
+            if (nacp_file == nullptr)
+                break;
+
+            FileSys::NACP nacp(nacp_file);
+            out[PatchType::Update] = nacp.GetVersionString();
+        } while (false);
+    }
+
+    return out;
+}
+
+} // namespace FileSys

src/core/file_sys/patch_manager.h

@@ -0,0 +1,54 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <map>
+#include <string>
+#include "common/common_types.h"
+#include "core/file_sys/vfs.h"
+#include "nca_metadata.h"
+#include "romfs_factory.h"
+
+namespace FileSys {
+
+class NCA;
+
+enum class TitleVersionFormat : u8 {
+    ThreeElements, ///< vX.Y.Z
+    FourElements,  ///< vX.Y.Z.W
+};
+
+std::string FormatTitleVersion(u32 version,
+                               TitleVersionFormat format = TitleVersionFormat::ThreeElements);
+
+enum class PatchType {
+    Update,
+};
+
+std::string FormatPatchTypeName(PatchType type);
+
+// A centralized class to manage patches to games.
+class PatchManager {
+public:
+    explicit PatchManager(u64 title_id);
+
+    // Currently tracked ExeFS patches:
+    // - Game Updates
+    VirtualDir PatchExeFS(VirtualDir exefs) const;
+
+    // Currently tracked RomFS patches:
+    // - Game Updates
+    VirtualFile PatchRomFS(VirtualFile base, u64 ivfc_offset,
+                           ContentRecordType type = ContentRecordType::Program) const;
+
+    // Returns a vector of pairs between patch names and patch versions.
+    // i.e. Update v80 will return {Update, 80}
+    std::map<PatchType, std::string> GetPatchVersionNames() const;
+
+private:
+    u64 title_id;
+};
+
+} // namespace FileSys

src/core/file_sys/registered_cache.cpp

@@ -276,6 +276,18 @@ VirtualFile RegisteredCache::GetEntryUnparsed(RegisteredCacheEntry entry) const
     return GetEntryUnparsed(entry.title_id, entry.type);
 }
 
+boost::optional<u32> RegisteredCache::GetEntryVersion(u64 title_id) const {
+    const auto meta_iter = meta.find(title_id);
+    if (meta_iter != meta.end())
+        return meta_iter->second.GetTitleVersion();
+
+    const auto yuzu_meta_iter = yuzu_meta.find(title_id);
+    if (yuzu_meta_iter != yuzu_meta.end())
+        return yuzu_meta_iter->second.GetTitleVersion();
+
+    return boost::none;
+}
+
 VirtualFile RegisteredCache::GetEntryRaw(u64 title_id, ContentRecordType type) const {
     const auto id = GetNcaIDFromMetadata(title_id, type);
     if (id == boost::none)
@@ -355,17 +367,21 @@ std::vector<RegisteredCacheEntry> RegisteredCache::ListEntriesFilter(
     return out;
 }
 
-static std::shared_ptr<NCA> GetNCAFromXCIForID(std::shared_ptr<XCI> xci, const NcaID& id) {
-    const auto filename = fmt::format("{}.nca", Common::HexArrayToString(id, false));
-    const auto iter =
-        std::find_if(xci->GetNCAs().begin(), xci->GetNCAs().end(),
-                     [&filename](std::shared_ptr<NCA> nca) { return nca->GetName() == filename; });
-    return iter == xci->GetNCAs().end() ? nullptr : *iter;
+static std::shared_ptr<NCA> GetNCAFromNSPForID(std::shared_ptr<NSP> nsp, const NcaID& id) {
+    const auto file = nsp->GetFile(fmt::format("{}.nca", Common::HexArrayToString(id, false)));
+    if (file == nullptr)
+        return nullptr;
+    return std::make_shared<NCA>(file);
 }
 
 InstallResult RegisteredCache::InstallEntry(std::shared_ptr<XCI> xci, bool overwrite_if_exists,
                                             const VfsCopyFunction& copy) {
-    const auto& ncas = xci->GetNCAs();
+    return InstallEntry(xci->GetSecurePartitionNSP(), overwrite_if_exists, copy);
+}
+
+InstallResult RegisteredCache::InstallEntry(std::shared_ptr<NSP> nsp, bool overwrite_if_exists,
+                                            const VfsCopyFunction& copy) {
+    const auto& ncas = nsp->GetNCAsCollapsed();
     const auto& meta_iter = std::find_if(ncas.begin(), ncas.end(), [](std::shared_ptr<NCA> nca) {
         return nca->GetType() == NCAContentType::Meta;
     });
@@ -389,7 +405,7 @@ InstallResult RegisteredCache::InstallEntry(std::shared_ptr<XCI> xci, bool overw
     const auto cnmt_file = section0->GetFiles()[0];
     const CNMT cnmt(cnmt_file);
     for (const auto& record : cnmt.GetContentRecords()) {
-        const auto nca = GetNCAFromXCIForID(xci, record.nca_id);
+        const auto nca = GetNCAFromNSPForID(nsp, record.nca_id);
         if (nca == nullptr)
             return InstallResult::ErrorCopyFailed;
         const auto res2 = RawInstallNCA(nca, copy, overwrite_if_exists, record.nca_id);
@@ -490,4 +506,107 @@ bool RegisteredCache::RawInstallYuzuMeta(const CNMT& cnmt) {
                                    kv.second.GetTitleID() == cnmt.GetTitleID();
                         }) != yuzu_meta.end();
 }
+
+RegisteredCacheUnion::RegisteredCacheUnion(std::vector<std::shared_ptr<RegisteredCache>> caches)
+    : caches(std::move(caches)) {}
+
+void RegisteredCacheUnion::Refresh() {
+    for (const auto& c : caches)
+        c->Refresh();
+}
+
+bool RegisteredCacheUnion::HasEntry(u64 title_id, ContentRecordType type) const {
+    return std::any_of(caches.begin(), caches.end(), [title_id, type](const auto& cache) {
+        return cache->HasEntry(title_id, type);
+    });
+}
+
+bool RegisteredCacheUnion::HasEntry(RegisteredCacheEntry entry) const {
+    return HasEntry(entry.title_id, entry.type);
+}
+
+boost::optional<u32> RegisteredCacheUnion::GetEntryVersion(u64 title_id) const {
+    for (const auto& c : caches) {
+        const auto res = c->GetEntryVersion(title_id);
+        if (res != boost::none)
+            return res;
+    }
+
+    return boost::none;
+}
+
+VirtualFile RegisteredCacheUnion::GetEntryUnparsed(u64 title_id, ContentRecordType type) const {
+    for (const auto& c : caches) {
+        const auto res = c->GetEntryUnparsed(title_id, type);
+        if (res != nullptr)
+            return res;
+    }
+
+    return nullptr;
+}
+
+VirtualFile RegisteredCacheUnion::GetEntryUnparsed(RegisteredCacheEntry entry) const {
+    return GetEntryUnparsed(entry.title_id, entry.type);
+}
+
+VirtualFile RegisteredCacheUnion::GetEntryRaw(u64 title_id, ContentRecordType type) const {
+    for (const auto& c : caches) {
+        const auto res = c->GetEntryRaw(title_id, type);
+        if (res != nullptr)
+            return res;
+    }
+
+    return nullptr;
+}
+
+VirtualFile RegisteredCacheUnion::GetEntryRaw(RegisteredCacheEntry entry) const {
+    return GetEntryRaw(entry.title_id, entry.type);
+}
+
+std::shared_ptr<NCA> RegisteredCacheUnion::GetEntry(u64 title_id, ContentRecordType type) const {
+    const auto raw = GetEntryRaw(title_id, type);
+    if (raw == nullptr)
+        return nullptr;
+    return std::make_shared<NCA>(raw);
+}
+
+std::shared_ptr<NCA> RegisteredCacheUnion::GetEntry(RegisteredCacheEntry entry) const {
+    return GetEntry(entry.title_id, entry.type);
+}
+
+std::vector<RegisteredCacheEntry> RegisteredCacheUnion::ListEntries() const {
+    std::vector<RegisteredCacheEntry> out;
+    for (const auto& c : caches) {
+        c->IterateAllMetadata<RegisteredCacheEntry>(
+            out,
+            [](const CNMT& c, const ContentRecord& r) {
+                return RegisteredCacheEntry{c.GetTitleID(), r.type};
+            },
+            [](const CNMT& c, const ContentRecord& r) { return true; });
+    }
+    return out;
+}
+
+std::vector<RegisteredCacheEntry> RegisteredCacheUnion::ListEntriesFilter(
+    boost::optional<TitleType> title_type, boost::optional<ContentRecordType> record_type,
+    boost::optional<u64> title_id) const {
+    std::vector<RegisteredCacheEntry> out;
+    for (const auto& c : caches) {
+        c->IterateAllMetadata<RegisteredCacheEntry>(
+            out,
+            [](const CNMT& c, const ContentRecord& r) {
+                return RegisteredCacheEntry{c.GetTitleID(), r.type};
+            },
+            [&title_type, &record_type, &title_id](const CNMT& c, const ContentRecord& r) {
+                if (title_type != boost::none && title_type.get() != c.GetType())
+                    return false;
+                if (record_type != boost::none && record_type.get() != r.type)
+                    return false;
+                if (title_id != boost::none && title_id.get() != c.GetTitleID())
+                    return false;
+                return true;
+            });
+    }
+    return out;
+}
 } // namespace FileSys

src/core/file_sys/registered_cache.h

@@ -13,8 +13,9 @@
 #include <boost/container/flat_map.hpp>
 #include "common/common_funcs.h"
 #include "common/common_types.h"
-#include "content_archive.h"
+#include "core/file_sys/content_archive.h"
 #include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/submission_package.h"
 #include "core/file_sys/vfs.h"
 
 namespace FileSys {
@@ -39,6 +40,10 @@ struct RegisteredCacheEntry {
     std::string DebugInfo() const;
 };
 
+constexpr u64 GetUpdateTitleID(u64 base_title_id) {
+    return base_title_id | 0x800;
+}
+
 // boost flat_map requires operator< for O(log(n)) lookups.
 bool operator<(const RegisteredCacheEntry& lhs, const RegisteredCacheEntry& rhs);
 
@@ -56,6 +61,8 @@ bool operator<(const RegisteredCacheEntry& lhs, const RegisteredCacheEntry& rhs)
  * 4GB splitting can be ignored.)
  */
 class RegisteredCache {
+    friend class RegisteredCacheUnion;
+
 public:
     // Parsing function defines the conversion from raw file to NCA. If there are other steps
     // besides creating the NCA from the file (e.g. NAX0 on SD Card), that should go in a custom
@@ -70,6 +77,8 @@ public:
     bool HasEntry(u64 title_id, ContentRecordType type) const;
     bool HasEntry(RegisteredCacheEntry entry) const;
 
+    boost::optional<u32> GetEntryVersion(u64 title_id) const;
+
     VirtualFile GetEntryUnparsed(u64 title_id, ContentRecordType type) const;
     VirtualFile GetEntryUnparsed(RegisteredCacheEntry entry) const;
 
@@ -86,10 +95,12 @@ public:
         boost::optional<ContentRecordType> record_type = boost::none,
         boost::optional<u64> title_id = boost::none) const;
 
-    // Raw copies all the ncas from the xci to the csache. Does some quick checks to make sure there
-    // is a meta NCA and all of them are accessible.
+    // Raw copies all the ncas from the xci/nsp to the csache. Does some quick checks to make sure
+    // there is a meta NCA and all of them are accessible.
     InstallResult InstallEntry(std::shared_ptr<XCI> xci, bool overwrite_if_exists = false,
                                const VfsCopyFunction& copy = &VfsRawCopy);
+    InstallResult InstallEntry(std::shared_ptr<NSP> nsp, bool overwrite_if_exists = false,
+                               const VfsCopyFunction& copy = &VfsRawCopy);
 
     // Due to the fact that we must use Meta-type NCAs to determine the existance of files, this
     // poses quite a challenge. Instead of creating a new meta NCA for this file, yuzu will create a
@@ -125,4 +136,36 @@ private:
     boost::container::flat_map<u64, CNMT> yuzu_meta;
 };
 
+// Combines multiple RegisteredCaches (i.e. SysNAND, UserNAND, SDMC) into one interface.
+class RegisteredCacheUnion {
+public:
+    explicit RegisteredCacheUnion(std::vector<std::shared_ptr<RegisteredCache>> caches);
+
+    void Refresh();
+
+    bool HasEntry(u64 title_id, ContentRecordType type) const;
+    bool HasEntry(RegisteredCacheEntry entry) const;
+
+    boost::optional<u32> GetEntryVersion(u64 title_id) const;
+
+    VirtualFile GetEntryUnparsed(u64 title_id, ContentRecordType type) const;
+    VirtualFile GetEntryUnparsed(RegisteredCacheEntry entry) const;
+
+    VirtualFile GetEntryRaw(u64 title_id, ContentRecordType type) const;
+    VirtualFile GetEntryRaw(RegisteredCacheEntry entry) const;
+
+    std::shared_ptr<NCA> GetEntry(u64 title_id, ContentRecordType type) const;
+    std::shared_ptr<NCA> GetEntry(RegisteredCacheEntry entry) const;
+
+    std::vector<RegisteredCacheEntry> ListEntries() const;
+    // If a parameter is not boost::none, it will be filtered for from all entries.
+    std::vector<RegisteredCacheEntry> ListEntriesFilter(
+        boost::optional<TitleType> title_type = boost::none,
+        boost::optional<ContentRecordType> record_type = boost::none,
+        boost::optional<u64> title_id = boost::none) const;
+
+private:
+    std::vector<std::shared_ptr<RegisteredCache>> caches;
+};
+
 } // namespace FileSys

src/core/file_sys/romfs_factory.cpp

@@ -6,7 +6,9 @@
 #include <memory>
 #include "common/common_types.h"
 #include "common/logging/log.h"
+#include "core/core.h"
 #include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/patch_manager.h"
 #include "core/file_sys/registered_cache.h"
 #include "core/file_sys/romfs_factory.h"
 #include "core/hle/kernel/process.h"
@@ -20,10 +22,17 @@ RomFSFactory::RomFSFactory(Loader::AppLoader& app_loader) {
     if (app_loader.ReadRomFS(file) != Loader::ResultStatus::Success) {
         LOG_ERROR(Service_FS, "Unable to read RomFS!");
     }
+
+    updatable = app_loader.IsRomFSUpdatable();
+    ivfc_offset = app_loader.ReadRomFSIVFCOffset();
 }
 
 ResultVal<VirtualFile> RomFSFactory::OpenCurrentProcess() {
-    return MakeResult<VirtualFile>(file);
+    if (!updatable)
+        return MakeResult<VirtualFile>(file);
+
+    const PatchManager patch_manager(Core::CurrentProcess()->program_id);
+    return MakeResult<VirtualFile>(patch_manager.PatchRomFS(file, ivfc_offset));
 }
 
 ResultVal<VirtualFile> RomFSFactory::Open(u64 title_id, StorageId storage, ContentRecordType type) {

src/core/file_sys/romfs_factory.h

@@ -36,6 +36,8 @@ public:
 
 private:
     VirtualFile file;
+    bool updatable;
+    u64 ivfc_offset;
 };
 
 } // namespace FileSys

src/core/file_sys/savedata_factory.cpp

@@ -7,6 +7,7 @@
 #include "common/logging/log.h"
 #include "core/core.h"
 #include "core/file_sys/savedata_factory.h"
+#include "core/file_sys/vfs.h"
 #include "core/hle/kernel/process.h"
 
 namespace FileSys {

src/core/file_sys/savedata_factory.h

@@ -8,6 +8,7 @@
 #include <string>
 #include "common/common_types.h"
 #include "common/swap.h"
+#include "core/file_sys/vfs.h"
 #include "core/hle/result.h"
 
 namespace FileSys {

src/core/file_sys/submission_package.cpp

@@ -0,0 +1,235 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <fmt/ostream.h>
+#include "common/assert.h"
+#include "common/hex_util.h"
+#include "core/file_sys/content_archive.h"
+#include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/submission_package.h"
+#include "core/loader/loader.h"
+
+namespace FileSys {
+NSP::NSP(VirtualFile file_)
+    : file(std::move(file_)),
+      pfs(std::make_shared<PartitionFilesystem>(file)), status{Loader::ResultStatus::Success} {
+    if (pfs->GetStatus() != Loader::ResultStatus::Success) {
+        status = pfs->GetStatus();
+        return;
+    }
+
+    if (IsDirectoryExeFS(pfs)) {
+        extracted = true;
+        exefs = pfs;
+
+        const auto& files = pfs->GetFiles();
+        const auto romfs_iter =
+            std::find_if(files.begin(), files.end(), [](const FileSys::VirtualFile& file) {
+                return file->GetName().find(".romfs") != std::string::npos;
+            });
+        if (romfs_iter != files.end())
+            romfs = *romfs_iter;
+        return;
+    }
+
+    extracted = false;
+    const auto files = pfs->GetFiles();
+
+    Core::Crypto::KeyManager keys;
+    for (const auto& ticket_file : files) {
+        if (ticket_file->GetExtension() == "tik") {
+            if (ticket_file == nullptr ||
+                ticket_file->GetSize() <
+                    Core::Crypto::TICKET_FILE_TITLEKEY_OFFSET + sizeof(Core::Crypto::Key128)) {
+                continue;
+            }
+
+            Core::Crypto::Key128 key{};
+            ticket_file->Read(key.data(), key.size(), Core::Crypto::TICKET_FILE_TITLEKEY_OFFSET);
+            std::string_view name_only(ticket_file->GetName());
+            name_only.remove_suffix(4);
+            const auto rights_id_raw = Common::HexStringToArray<16>(name_only);
+            u128 rights_id;
+            std::memcpy(rights_id.data(), rights_id_raw.data(), sizeof(u128));
+            keys.SetKey(Core::Crypto::S128KeyType::Titlekey, key, rights_id[1], rights_id[0]);
+        }
+    }
+
+    for (const auto& outer_file : files) {
+        if (outer_file->GetName().substr(outer_file->GetName().size() - 9) == ".cnmt.nca") {
+            const auto nca = std::make_shared<NCA>(outer_file);
+            if (nca->GetStatus() != Loader::ResultStatus::Success)
+                continue;
+            const auto section0 = nca->GetSubdirectories()[0];
+
+            for (const auto& inner_file : section0->GetFiles()) {
+                if (inner_file->GetExtension() != "cnmt")
+                    continue;
+
+                const CNMT cnmt(inner_file);
+                auto& ncas_title = ncas[cnmt.GetTitleID()];
+
+                ncas_title[ContentRecordType::Meta] = nca;
+                for (const auto& rec : cnmt.GetContentRecords()) {
+                    const auto id_string = Common::HexArrayToString(rec.nca_id, false);
+                    const auto next_file = pfs->GetFile(fmt::format("{}.nca", id_string));
+                    if (next_file == nullptr) {
+                        LOG_WARNING(Service_FS,
+                                    "NCA with ID {}.nca is listed in content metadata, but cannot "
+                                    "be found in PFS. NSP appears to be corrupted.",
+                                    id_string);
+                        continue;
+                    }
+
+                    auto next_nca = std::make_shared<NCA>(next_file);
+                    if (next_nca->GetType() == NCAContentType::Program)
+                        program_status[cnmt.GetTitleID()] = next_nca->GetStatus();
+                    if (next_nca->GetStatus() == Loader::ResultStatus::Success)
+                        ncas_title[rec.type] = std::move(next_nca);
+                }
+
+                break;
+            }
+        }
+    }
+}
+
+NSP::~NSP() = default;
+
+Loader::ResultStatus NSP::GetStatus() const {
+    return status;
+}
+
+Loader::ResultStatus NSP::GetProgramStatus(u64 title_id) const {
+    const auto iter = program_status.find(title_id);
+    if (iter == program_status.end())
+        return Loader::ResultStatus::ErrorNSPMissingProgramNCA;
+    return iter->second;
+}
+
+u64 NSP::GetFirstTitleID() const {
+    if (program_status.empty())
+        return 0;
+    return program_status.begin()->first;
+}
+
+u64 NSP::GetProgramTitleID() const {
+    const auto out = GetFirstTitleID();
+    if ((out & 0x800) == 0)
+        return out;
+
+    const auto ids = GetTitleIDs();
+    const auto iter =
+        std::find_if(ids.begin(), ids.end(), [](u64 tid) { return (tid & 0x800) == 0; });
+    return iter == ids.end() ? out : *iter;
+}
+
+std::vector<u64> NSP::GetTitleIDs() const {
+    std::vector<u64> out;
+    out.reserve(ncas.size());
+    for (const auto& kv : ncas)
+        out.push_back(kv.first);
+    return out;
+}
+
+bool NSP::IsExtractedType() const {
+    return extracted;
+}
+
+VirtualFile NSP::GetRomFS() const {
+    return romfs;
+}
+
+VirtualDir NSP::GetExeFS() const {
+    return exefs;
+}
+
+std::vector<std::shared_ptr<NCA>> NSP::GetNCAsCollapsed() const {
+    if (extracted)
+        LOG_WARNING(Service_FS, "called on an NSP that is of type extracted.");
+    std::vector<std::shared_ptr<NCA>> out;
+    for (const auto& map : ncas) {
+        for (const auto& inner_map : map.second)
+            out.push_back(inner_map.second);
+    }
+    return out;
+}
+
+std::multimap<u64, std::shared_ptr<NCA>> NSP::GetNCAsByTitleID() const {
+    if (extracted)
+        LOG_WARNING(Service_FS, "called on an NSP that is of type extracted.");
+    std::multimap<u64, std::shared_ptr<NCA>> out;
+    for (const auto& map : ncas) {
+        for (const auto& inner_map : map.second)
+            out.emplace(map.first, inner_map.second);
+    }
+    return out;
+}
+
+std::map<u64, std::map<ContentRecordType, std::shared_ptr<NCA>>> NSP::GetNCAs() const {
+    return ncas;
+}
+
+std::shared_ptr<NCA> NSP::GetNCA(u64 title_id, ContentRecordType type) const {
+    if (extracted)
+        LOG_WARNING(Service_FS, "called on an NSP that is of type extracted.");
+
+    const auto title_id_iter = ncas.find(title_id);
+    if (title_id_iter == ncas.end())
+        return nullptr;
+
+    const auto type_iter = title_id_iter->second.find(type);
+    if (type_iter == title_id_iter->second.end())
+        return nullptr;
+
+    return type_iter->second;
+}
+
+VirtualFile NSP::GetNCAFile(u64 title_id, ContentRecordType type) const {
+    if (extracted)
+        LOG_WARNING(Service_FS, "called on an NSP that is of type extracted.");
+    const auto nca = GetNCA(title_id, type);
+    if (nca != nullptr)
+        return nca->GetBaseFile();
+    return nullptr;
+}
+
+std::vector<Core::Crypto::Key128> NSP::GetTitlekey() const {
+    if (extracted)
+        LOG_WARNING(Service_FS, "called on an NSP that is of type extracted.");
+    std::vector<Core::Crypto::Key128> out;
+    for (const auto& ticket_file : ticket_files) {
+        if (ticket_file == nullptr ||
+            ticket_file->GetSize() <
+                Core::Crypto::TICKET_FILE_TITLEKEY_OFFSET + sizeof(Core::Crypto::Key128)) {
+            continue;
+        }
+
+        out.emplace_back();
+        ticket_file->Read(out.back().data(), out.back().size(),
+                          Core::Crypto::TICKET_FILE_TITLEKEY_OFFSET);
+    }
+    return out;
+}
+
+std::vector<VirtualFile> NSP::GetFiles() const {
+    return pfs->GetFiles();
+}
+
+std::vector<VirtualDir> NSP::GetSubdirectories() const {
+    return pfs->GetSubdirectories();
+}
+
+std::string NSP::GetName() const {
+    return file->GetName();
+}
+
+VirtualDir NSP::GetParentDirectory() const {
+    return file->GetContainingDirectory();
+}
+
+bool NSP::ReplaceFileWithSubdirectory(VirtualFile file, VirtualDir dir) {
+    return false;
+}
+} // namespace FileSys

src/core/file_sys/submission_package.h

@@ -0,0 +1,72 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <map>
+#include <vector>
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/file_sys/content_archive.h"
+#include "core/file_sys/romfs_factory.h"
+#include "core/file_sys/vfs.h"
+#include "core/loader/loader.h"
+#include "romfs_factory.h"
+
+namespace FileSys {
+
+class NSP : public ReadOnlyVfsDirectory {
+public:
+    explicit NSP(VirtualFile file);
+    ~NSP();
+
+    Loader::ResultStatus GetStatus() const;
+    Loader::ResultStatus GetProgramStatus(u64 title_id) const;
+    // Should only be used when one title id can be assured.
+    u64 GetFirstTitleID() const;
+    u64 GetProgramTitleID() const;
+    std::vector<u64> GetTitleIDs() const;
+
+    bool IsExtractedType() const;
+
+    // Common (Can be safely called on both types)
+    VirtualFile GetRomFS() const;
+    VirtualDir GetExeFS() const;
+
+    // Type 0 Only (Collection of NCAs + Certificate + Ticket + Meta XML)
+    std::vector<std::shared_ptr<NCA>> GetNCAsCollapsed() const;
+    std::multimap<u64, std::shared_ptr<NCA>> GetNCAsByTitleID() const;
+    std::map<u64, std::map<ContentRecordType, std::shared_ptr<NCA>>> GetNCAs() const;
+    std::shared_ptr<NCA> GetNCA(u64 title_id, ContentRecordType type) const;
+    VirtualFile GetNCAFile(u64 title_id, ContentRecordType type) const;
+    std::vector<Core::Crypto::Key128> GetTitlekey() const;
+
+    std::vector<VirtualFile> GetFiles() const override;
+
+    std::vector<VirtualDir> GetSubdirectories() const override;
+
+    std::string GetName() const override;
+
+    VirtualDir GetParentDirectory() const override;
+
+protected:
+    bool ReplaceFileWithSubdirectory(VirtualFile file, VirtualDir dir) override;
+
+private:
+    VirtualFile file;
+
+    bool extracted;
+    Loader::ResultStatus status;
+    std::map<u64, Loader::ResultStatus> program_status;
+
+    std::shared_ptr<PartitionFilesystem> pfs;
+    // Map title id -> {map type -> NCA}
+    std::map<u64, std::map<ContentRecordType, std::shared_ptr<NCA>>> ncas;
+    std::vector<VirtualFile> ticket_files;
+
+    VirtualFile romfs;
+    VirtualDir exefs;
+};
+} // namespace FileSys

src/core/hle/ipc_helpers.h

@@ -12,6 +12,7 @@
 #include <utility>
 #include "common/assert.h"
 #include "common/common_types.h"
+#include "core/core.h"
 #include "core/hle/ipc.h"
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/client_session.h"
@@ -135,7 +136,9 @@ public:
         if (context->Session()->IsDomain()) {
             context->AddDomainObject(std::move(iface));
         } else {
-            auto sessions = Kernel::ServerSession::CreateSessionPair(iface->GetServiceName());
+            auto& kernel = Core::System::GetInstance().Kernel();
+            auto sessions =
+                Kernel::ServerSession::CreateSessionPair(kernel, iface->GetServiceName());
             auto server = std::get<Kernel::SharedPtr<Kernel::ServerSession>>(sessions);
             auto client = std::get<Kernel::SharedPtr<Kernel::ClientSession>>(sessions);
             iface->ClientConnected(server);

src/core/hle/kernel/address_arbiter.cpp

@@ -8,9 +8,11 @@
 #include "common/assert.h"
 #include "common/common_types.h"
 #include "core/core.h"
+#include "core/core_cpu.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/process.h"
+#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/thread.h"
 #include "core/hle/result.h"
 #include "core/memory.h"

src/core/hle/kernel/client_port.cpp

@@ -14,7 +14,7 @@
 
 namespace Kernel {
 
-ClientPort::ClientPort() = default;
+ClientPort::ClientPort(KernelCore& kernel) : Object{kernel} {}
 ClientPort::~ClientPort() = default;
 
 ResultVal<SharedPtr<ClientSession>> ClientPort::Connect() {
@@ -27,7 +27,7 @@ ResultVal<SharedPtr<ClientSession>> ClientPort::Connect() {
     active_sessions++;
 
     // Create a new session pair, let the created sessions inherit the parent port's HLE handler.
-    auto sessions = ServerSession::CreateSessionPair(server_port->GetName(), this);
+    auto sessions = ServerSession::CreateSessionPair(kernel, server_port->GetName(), this);
 
     if (server_port->hle_handler)
         server_port->hle_handler->ClientConnected(std::get<SharedPtr<ServerSession>>(sessions));

src/core/hle/kernel/client_port.h

@@ -11,8 +11,9 @@
 
 namespace Kernel {
 
-class ServerPort;
 class ClientSession;
+class KernelCore;
+class ServerPort;
 
 class ClientPort final : public Object {
 public:
@@ -44,7 +45,7 @@ public:
     void ConnectionClosed();
 
 private:
-    ClientPort();
+    explicit ClientPort(KernelCore& kernel);
     ~ClientPort() override;
 
     SharedPtr<ServerPort> server_port; ///< ServerPort associated with this client port.

src/core/hle/kernel/client_session.cpp

@@ -11,7 +11,7 @@
 
 namespace Kernel {
 
-ClientSession::ClientSession() = default;
+ClientSession::ClientSession(KernelCore& kernel) : Object{kernel} {}
 ClientSession::~ClientSession() {
     // This destructor will be called automatically when the last ClientSession handle is closed by
     // the emulated application.

src/core/hle/kernel/client_session.h

@@ -12,8 +12,9 @@
 
 namespace Kernel {
 
-class ServerSession;
+class KernelCore;
 class Session;
+class ServerSession;
 class Thread;
 
 class ClientSession final : public Object {
@@ -41,7 +42,7 @@ public:
     std::shared_ptr<Session> parent;
 
 private:
-    ClientSession();
+    explicit ClientSession(KernelCore& kernel);
     ~ClientSession() override;
 };
 

src/core/hle/kernel/event.cpp

@@ -10,11 +10,11 @@
 
 namespace Kernel {
 
-Event::Event() {}
-Event::~Event() {}
+Event::Event(KernelCore& kernel) : WaitObject{kernel} {}
+Event::~Event() = default;
 
-SharedPtr<Event> Event::Create(ResetType reset_type, std::string name) {
-    SharedPtr<Event> evt(new Event);
+SharedPtr<Event> Event::Create(KernelCore& kernel, ResetType reset_type, std::string name) {
+    SharedPtr<Event> evt(new Event(kernel));
 
     evt->signaled = false;
     evt->reset_type = reset_type;

src/core/hle/kernel/event.h

@@ -10,14 +10,18 @@
 
 namespace Kernel {
 
+class KernelCore;
+
 class Event final : public WaitObject {
 public:
     /**
      * Creates an event
+     * @param kernel The kernel instance to create this event under.
      * @param reset_type ResetType describing how to create event
      * @param name Optional name of event
      */
-    static SharedPtr<Event> Create(ResetType reset_type, std::string name = "Unknown");
+    static SharedPtr<Event> Create(KernelCore& kernel, ResetType reset_type,
+                                   std::string name = "Unknown");
 
     std::string GetTypeName() const override {
         return "Event";
@@ -44,7 +48,7 @@ public:
     void Clear();
 
 private:
-    Event();
+    explicit Event(KernelCore& kernel);
     ~Event() override;
 
     ResetType reset_type; ///< Current ResetType

src/core/hle/kernel/handle_table.cpp

@@ -13,8 +13,6 @@
 
 namespace Kernel {
 
-HandleTable g_handle_table;
-
 HandleTable::HandleTable() {
     next_generation = 1;
     Clear();

src/core/hle/kernel/handle_table.h

@@ -121,6 +121,4 @@ private:
     u16 next_free_slot;
 };
 
-extern HandleTable g_handle_table;
-
 } // namespace Kernel

src/core/hle/kernel/hle_ipc.cpp

@@ -13,10 +13,12 @@
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "common/logging/log.h"
+#include "core/core.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/event.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/hle_ipc.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/server_session.h"
@@ -51,7 +53,9 @@ SharedPtr<Event> HLERequestContext::SleepClientThread(SharedPtr<Thread> thread,
 
     if (!event) {
         // Create event if not provided
-        event = Kernel::Event::Create(Kernel::ResetType::OneShot, "HLE Pause Event: " + reason);
+        auto& kernel = Core::System::GetInstance().Kernel();
+        event =
+            Kernel::Event::Create(kernel, Kernel::ResetType::OneShot, "HLE Pause Event: " + reason);
     }
 
     event->Clear();
@@ -90,12 +94,14 @@ void HLERequestContext::ParseCommandBuffer(u32_le* src_cmdbuf, bool incoming) {
             rp.Skip(2, false);
         }
         if (incoming) {
+            auto& handle_table = Core::System::GetInstance().Kernel().HandleTable();
+
             // Populate the object lists with the data in the IPC request.
             for (u32 handle = 0; handle < handle_descriptor_header->num_handles_to_copy; ++handle) {
-                copy_objects.push_back(Kernel::g_handle_table.GetGeneric(rp.Pop<Handle>()));
+                copy_objects.push_back(handle_table.GetGeneric(rp.Pop<Handle>()));
             }
             for (u32 handle = 0; handle < handle_descriptor_header->num_handles_to_move; ++handle) {
-                move_objects.push_back(Kernel::g_handle_table.GetGeneric(rp.Pop<Handle>()));
+                move_objects.push_back(handle_table.GetGeneric(rp.Pop<Handle>()));
             }
         } else {
             // For responses we just ignore the handles, they're empty and will be populated when
@@ -230,17 +236,19 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(const Thread& thread)
         ASSERT(copy_objects.size() == handle_descriptor_header->num_handles_to_copy);
         ASSERT(move_objects.size() == handle_descriptor_header->num_handles_to_move);
 
+        auto& handle_table = Core::System::GetInstance().Kernel().HandleTable();
+
         // We don't make a distinction between copy and move handles when translating since HLE
         // services don't deal with handles directly. However, the guest applications might check
         // for specific values in each of these descriptors.
         for (auto& object : copy_objects) {
             ASSERT(object != nullptr);
-            dst_cmdbuf[current_offset++] = Kernel::g_handle_table.Create(object).Unwrap();
+            dst_cmdbuf[current_offset++] = handle_table.Create(object).Unwrap();
         }
 
         for (auto& object : move_objects) {
             ASSERT(object != nullptr);
-            dst_cmdbuf[current_offset++] = Kernel::g_handle_table.Create(object).Unwrap();
+            dst_cmdbuf[current_offset++] = handle_table.Create(object).Unwrap();
         }
     }
 

src/core/hle/kernel/kernel.cpp

@@ -2,38 +2,291 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <array>
+#include <atomic>
+#include <memory>
+#include <mutex>
+#include <utility>
+
+#include "common/assert.h"
+#include "common/logging/log.h"
+
+#include "core/core.h"
+#include "core/core_timing.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/hle/kernel/thread.h"
 #include "core/hle/kernel/timer.h"
+#include "core/hle/lock.h"
+#include "core/hle/result.h"
 
 namespace Kernel {
 
-std::atomic<u32> Object::next_object_id{0};
+/**
+ * Callback that will wake up the thread it was scheduled for
+ * @param thread_handle The handle of the thread that's been awoken
+ * @param cycles_late The number of CPU cycles that have passed since the desired wakeup time
+ */
+static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] int cycles_late) {
+    const auto proper_handle = static_cast<Handle>(thread_handle);
+    auto& system = Core::System::GetInstance();
 
-/// Initialize the kernel
-void Init() {
-    Kernel::ResourceLimitsInit();
-    Kernel::ThreadingInit();
-    Kernel::TimersInit();
+    // Lock the global kernel mutex when we enter the kernel HLE.
+    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
 
-    Object::next_object_id = 0;
-    // TODO(Subv): Start the process ids from 10 for now, as lower PIDs are
-    // reserved for low-level services
-    Process::next_process_id = 10;
+    SharedPtr<Thread> thread =
+        system.Kernel().RetrieveThreadFromWakeupCallbackHandleTable(proper_handle);
+    if (thread == nullptr) {
+        LOG_CRITICAL(Kernel, "Callback fired for invalid thread {:08X}", proper_handle);
+        return;
+    }
+
+    bool resume = true;
+
+    if (thread->status == ThreadStatus::WaitSynchAny ||
+        thread->status == ThreadStatus::WaitSynchAll ||
+        thread->status == ThreadStatus::WaitHLEEvent) {
+        // Remove the thread from each of its waiting objects' waitlists
+        for (auto& object : thread->wait_objects) {
+            object->RemoveWaitingThread(thread.get());
+        }
+        thread->wait_objects.clear();
+
+        // Invoke the wakeup callback before clearing the wait objects
+        if (thread->wakeup_callback) {
+            resume = thread->wakeup_callback(ThreadWakeupReason::Timeout, thread, nullptr, 0);
+        }
+    }
+
+    if (thread->mutex_wait_address != 0 || thread->condvar_wait_address != 0 ||
+        thread->wait_handle) {
+        ASSERT(thread->status == ThreadStatus::WaitMutex);
+        thread->mutex_wait_address = 0;
+        thread->condvar_wait_address = 0;
+        thread->wait_handle = 0;
+
+        auto lock_owner = thread->lock_owner;
+        // Threads waking up by timeout from WaitProcessWideKey do not perform priority inheritance
+        // and don't have a lock owner unless SignalProcessWideKey was called first and the thread
+        // wasn't awakened due to the mutex already being acquired.
+        if (lock_owner) {
+            lock_owner->RemoveMutexWaiter(thread);
+        }
+    }
+
+    if (thread->arb_wait_address != 0) {
+        ASSERT(thread->status == ThreadStatus::WaitArb);
+        thread->arb_wait_address = 0;
+    }
+
+    if (resume) {
+        thread->ResumeFromWait();
+    }
 }
 
-/// Shutdown the kernel
-void Shutdown() {
-    // Free all kernel objects
-    g_handle_table.Clear();
+/// The timer callback event, called when a timer is fired
+static void TimerCallback(u64 timer_handle, int cycles_late) {
+    const auto proper_handle = static_cast<Handle>(timer_handle);
+    auto& system = Core::System::GetInstance();
+    SharedPtr<Timer> timer = system.Kernel().RetrieveTimerFromCallbackHandleTable(proper_handle);
 
-    Kernel::ThreadingShutdown();
+    if (timer == nullptr) {
+        LOG_CRITICAL(Kernel, "Callback fired for invalid timer {:016X}", timer_handle);
+        return;
+    }
 
-    Kernel::TimersShutdown();
-    Kernel::ResourceLimitsShutdown();
+    timer->Signal(cycles_late);
+}
+
+struct KernelCore::Impl {
+    void Initialize(KernelCore& kernel) {
+        Shutdown();
+
+        InitializeResourceLimits(kernel);
+        InitializeThreads();
+        InitializeTimers();
+    }
+
+    void Shutdown() {
+        next_object_id = 0;
+        next_process_id = 10;
+        next_thread_id = 1;
+
+        process_list.clear();
+
+        handle_table.Clear();
+        resource_limits.fill(nullptr);
+
+        thread_wakeup_callback_handle_table.Clear();
+        thread_wakeup_event_type = nullptr;
+
+        timer_callback_handle_table.Clear();
+        timer_callback_event_type = nullptr;
+    }
+
+    void InitializeResourceLimits(KernelCore& kernel) {
+        // Create the four resource limits that the system uses
+        // Create the APPLICATION resource limit
+        SharedPtr<ResourceLimit> resource_limit = ResourceLimit::Create(kernel, "Applications");
+        resource_limit->max_priority = 0x18;
+        resource_limit->max_commit = 0x4000000;
+        resource_limit->max_threads = 0x20;
+        resource_limit->max_events = 0x20;
+        resource_limit->max_mutexes = 0x20;
+        resource_limit->max_semaphores = 0x8;
+        resource_limit->max_timers = 0x8;
+        resource_limit->max_shared_mems = 0x10;
+        resource_limit->max_address_arbiters = 0x2;
+        resource_limit->max_cpu_time = 0x1E;
+        resource_limits[static_cast<u8>(ResourceLimitCategory::APPLICATION)] = resource_limit;
+
+        // Create the SYS_APPLET resource limit
+        resource_limit = ResourceLimit::Create(kernel, "System Applets");
+        resource_limit->max_priority = 0x4;
+        resource_limit->max_commit = 0x5E00000;
+        resource_limit->max_threads = 0x1D;
+        resource_limit->max_events = 0xB;
+        resource_limit->max_mutexes = 0x8;
+        resource_limit->max_semaphores = 0x4;
+        resource_limit->max_timers = 0x4;
+        resource_limit->max_shared_mems = 0x8;
+        resource_limit->max_address_arbiters = 0x3;
+        resource_limit->max_cpu_time = 0x2710;
+        resource_limits[static_cast<u8>(ResourceLimitCategory::SYS_APPLET)] = resource_limit;
+
+        // Create the LIB_APPLET resource limit
+        resource_limit = ResourceLimit::Create(kernel, "Library Applets");
+        resource_limit->max_priority = 0x4;
+        resource_limit->max_commit = 0x600000;
+        resource_limit->max_threads = 0xE;
+        resource_limit->max_events = 0x8;
+        resource_limit->max_mutexes = 0x8;
+        resource_limit->max_semaphores = 0x4;
+        resource_limit->max_timers = 0x4;
+        resource_limit->max_shared_mems = 0x8;
+        resource_limit->max_address_arbiters = 0x1;
+        resource_limit->max_cpu_time = 0x2710;
+        resource_limits[static_cast<u8>(ResourceLimitCategory::LIB_APPLET)] = resource_limit;
+
+        // Create the OTHER resource limit
+        resource_limit = ResourceLimit::Create(kernel, "Others");
+        resource_limit->max_priority = 0x4;
+        resource_limit->max_commit = 0x2180000;
+        resource_limit->max_threads = 0xE1;
+        resource_limit->max_events = 0x108;
+        resource_limit->max_mutexes = 0x25;
+        resource_limit->max_semaphores = 0x43;
+        resource_limit->max_timers = 0x2C;
+        resource_limit->max_shared_mems = 0x1F;
+        resource_limit->max_address_arbiters = 0x2D;
+        resource_limit->max_cpu_time = 0x3E8;
+        resource_limits[static_cast<u8>(ResourceLimitCategory::OTHER)] = resource_limit;
+    }
+
+    void InitializeThreads() {
+        thread_wakeup_event_type =
+            CoreTiming::RegisterEvent("ThreadWakeupCallback", ThreadWakeupCallback);
+    }
+
+    void InitializeTimers() {
+        timer_callback_handle_table.Clear();
+        timer_callback_event_type = CoreTiming::RegisterEvent("TimerCallback", TimerCallback);
+    }
+
+    std::atomic<u32> next_object_id{0};
+    // TODO(Subv): Start the process ids from 10 for now, as lower PIDs are
+    // reserved for low-level services
+    std::atomic<u32> next_process_id{10};
+    std::atomic<u32> next_thread_id{1};
+
+    // Lists all processes that exist in the current session.
+    std::vector<SharedPtr<Process>> process_list;
+
+    Kernel::HandleTable handle_table;
+    std::array<SharedPtr<ResourceLimit>, 4> resource_limits;
+
+    /// The event type of the generic timer callback event
+    CoreTiming::EventType* timer_callback_event_type = nullptr;
+    // TODO(yuriks): This can be removed if Timer objects are explicitly pooled in the future,
+    // allowing us to simply use a pool index or similar.
+    Kernel::HandleTable timer_callback_handle_table;
+
+    CoreTiming::EventType* thread_wakeup_event_type = nullptr;
+    // TODO(yuriks): This can be removed if Thread objects are explicitly pooled in the future,
+    // allowing us to simply use a pool index or similar.
+    Kernel::HandleTable thread_wakeup_callback_handle_table;
+};
+
+KernelCore::KernelCore() : impl{std::make_unique<Impl>()} {}
+KernelCore::~KernelCore() {
+    Shutdown();
+}
+
+void KernelCore::Initialize() {
+    impl->Initialize(*this);
+}
+
+void KernelCore::Shutdown() {
+    impl->Shutdown();
+}
+
+Kernel::HandleTable& KernelCore::HandleTable() {
+    return impl->handle_table;
+}
+
+const Kernel::HandleTable& KernelCore::HandleTable() const {
+    return impl->handle_table;
+}
+
+SharedPtr<ResourceLimit> KernelCore::ResourceLimitForCategory(
+    ResourceLimitCategory category) const {
+    return impl->resource_limits.at(static_cast<std::size_t>(category));
+}
+
+SharedPtr<Thread> KernelCore::RetrieveThreadFromWakeupCallbackHandleTable(Handle handle) const {
+    return impl->thread_wakeup_callback_handle_table.Get<Thread>(handle);
+}
+
+SharedPtr<Timer> KernelCore::RetrieveTimerFromCallbackHandleTable(Handle handle) const {
+    return impl->timer_callback_handle_table.Get<Timer>(handle);
+}
+
+void KernelCore::AppendNewProcess(SharedPtr<Process> process) {
+    impl->process_list.push_back(std::move(process));
+}
+
+u32 KernelCore::CreateNewObjectID() {
+    return impl->next_object_id++;
+}
+
+u32 KernelCore::CreateNewThreadID() {
+    return impl->next_thread_id++;
+}
+
+u32 KernelCore::CreateNewProcessID() {
+    return impl->next_process_id++;
+}
+
+ResultVal<Handle> KernelCore::CreateTimerCallbackHandle(const SharedPtr<Timer>& timer) {
+    return impl->timer_callback_handle_table.Create(timer);
+}
+
+CoreTiming::EventType* KernelCore::ThreadWakeupCallbackEventType() const {
+    return impl->thread_wakeup_event_type;
+}
+
+CoreTiming::EventType* KernelCore::TimerCallbackEventType() const {
+    return impl->timer_callback_event_type;
+}
+
+Kernel::HandleTable& KernelCore::ThreadWakeupCallbackHandleTable() {
+    return impl->thread_wakeup_callback_handle_table;
+}
+
+const Kernel::HandleTable& KernelCore::ThreadWakeupCallbackHandleTable() const {
+    return impl->thread_wakeup_callback_handle_table;
 }
 
 } // namespace Kernel

src/core/hle/kernel/kernel.h

@@ -4,14 +4,93 @@
 
 #pragma once
 
-#include "common/common_types.h"
+#include "core/hle/kernel/object.h"
+
+template <typename T>
+class ResultVal;
+
+namespace CoreTiming {
+struct EventType;
+}
 
 namespace Kernel {
 
-/// Initialize the kernel with the specified system mode.
-void Init();
+class HandleTable;
+class Process;
+class ResourceLimit;
+class Thread;
+class Timer;
 
-/// Shutdown the kernel
-void Shutdown();
+enum class ResourceLimitCategory : u8;
+
+/// Represents a single instance of the kernel.
+class KernelCore {
+public:
+    KernelCore();
+    ~KernelCore();
+
+    KernelCore(const KernelCore&) = delete;
+    KernelCore& operator=(const KernelCore&) = delete;
+
+    KernelCore(KernelCore&&) = delete;
+    KernelCore& operator=(KernelCore&&) = delete;
+
+    /// Resets the kernel to a clean slate for use.
+    void Initialize();
+
+    /// Clears all resources in use by the kernel instance.
+    void Shutdown();
+
+    /// Provides a reference to the handle table.
+    Kernel::HandleTable& HandleTable();
+
+    /// Provides a const reference to the handle table.
+    const Kernel::HandleTable& HandleTable() const;
+
+    /// Retrieves a shared pointer to a ResourceLimit identified by the given category.
+    SharedPtr<ResourceLimit> ResourceLimitForCategory(ResourceLimitCategory category) const;
+
+    /// Retrieves a shared pointer to a Thread instance within the thread wakeup handle table.
+    SharedPtr<Thread> RetrieveThreadFromWakeupCallbackHandleTable(Handle handle) const;
+
+    /// Retrieves a shared pointer to a Timer instance within the timer callback handle table.
+    SharedPtr<Timer> RetrieveTimerFromCallbackHandleTable(Handle handle) const;
+
+    /// Adds the given shared pointer to an internal list of active processes.
+    void AppendNewProcess(SharedPtr<Process> process);
+
+private:
+    friend class Object;
+    friend class Process;
+    friend class Thread;
+    friend class Timer;
+
+    /// Creates a new object ID, incrementing the internal object ID counter.
+    u32 CreateNewObjectID();
+
+    /// Creates a new process ID, incrementing the internal process ID counter;
+    u32 CreateNewProcessID();
+
+    /// Creates a new thread ID, incrementing the internal thread ID counter.
+    u32 CreateNewThreadID();
+
+    /// Creates a timer callback handle for the given timer.
+    ResultVal<Handle> CreateTimerCallbackHandle(const SharedPtr<Timer>& timer);
+
+    /// Retrieves the event type used for thread wakeup callbacks.
+    CoreTiming::EventType* ThreadWakeupCallbackEventType() const;
+
+    /// Retrieves the event type used for timer callbacks.
+    CoreTiming::EventType* TimerCallbackEventType() const;
+
+    /// Provides a reference to the thread wakeup callback handle table.
+    Kernel::HandleTable& ThreadWakeupCallbackHandleTable();
+
+    /// Provides a const reference to the thread wakeup callback handle table.
+    const Kernel::HandleTable& ThreadWakeupCallbackHandleTable() const;
+
+    struct Impl;
+    std::unique_ptr<Impl> impl;
+};
 
 } // namespace Kernel

src/core/hle/kernel/mutex.cpp

@@ -58,15 +58,15 @@ static void TransferMutexOwnership(VAddr mutex_addr, SharedPtr<Thread> current_t
     }
 }
 
-ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle,
+ResultCode Mutex::TryAcquire(HandleTable& handle_table, VAddr address, Handle holding_thread_handle,
                              Handle requesting_thread_handle) {
     // The mutex address must be 4-byte aligned
     if ((address % sizeof(u32)) != 0) {
         return ResultCode(ErrorModule::Kernel, ErrCodes::InvalidAddress);
     }
 
-    SharedPtr<Thread> holding_thread = g_handle_table.Get<Thread>(holding_thread_handle);
-    SharedPtr<Thread> requesting_thread = g_handle_table.Get<Thread>(requesting_thread_handle);
+    SharedPtr<Thread> holding_thread = handle_table.Get<Thread>(holding_thread_handle);
+    SharedPtr<Thread> requesting_thread = handle_table.Get<Thread>(requesting_thread_handle);
 
     // TODO(Subv): It is currently unknown if it is possible to lock a mutex in behalf of another
     // thread.

src/core/hle/kernel/mutex.h

@@ -11,6 +11,7 @@ union ResultCode;
 
 namespace Kernel {
 
+class HandleTable;
 class Thread;
 
 class Mutex final {
@@ -21,8 +22,8 @@ public:
     static constexpr u32 MutexOwnerMask = 0xBFFFFFFF;
 
     /// Attempts to acquire a mutex at the specified address.
-    static ResultCode TryAcquire(VAddr address, Handle holding_thread_handle,
-                                 Handle requesting_thread_handle);
+    static ResultCode TryAcquire(HandleTable& handle_table, VAddr address,
+                                 Handle holding_thread_handle, Handle requesting_thread_handle);
 
     /// Releases the mutex at the specified address.
     static ResultCode Release(VAddr address);

src/core/hle/kernel/object.cpp

@@ -3,10 +3,12 @@
 // Refer to the license.txt file included.
 
 #include "common/assert.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/object.h"
 
 namespace Kernel {
 
+Object::Object(KernelCore& kernel) : kernel{kernel}, object_id{kernel.CreateNewObjectID()} {}
 Object::~Object() = default;
 
 bool Object::IsWaitable() const {

src/core/hle/kernel/object.h

@@ -14,6 +14,8 @@
 
 namespace Kernel {
 
+class KernelCore;
+
 using Handle = u32;
 
 enum class HandleType : u32 {
@@ -40,6 +42,7 @@ enum class ResetType {
 
 class Object : NonCopyable {
 public:
+    explicit Object(KernelCore& kernel);
     virtual ~Object();
 
     /// Returns a unique identifier for the object. For debugging purposes only.
@@ -61,15 +64,16 @@ public:
      */
     bool IsWaitable() const;
 
-public:
-    static std::atomic<u32> next_object_id;
+protected:
+    /// The kernel instance this object was created under.
+    KernelCore& kernel;
 
 private:
     friend void intrusive_ptr_add_ref(Object*);
     friend void intrusive_ptr_release(Object*);
 
     std::atomic<u32> ref_count{0};
-    std::atomic<u32> object_id{next_object_id++};
+    std::atomic<u32> object_id{0};
 };
 
 // Special functions used by boost::instrusive_ptr to do automatic ref-counting

src/core/hle/kernel/process.cpp

@@ -8,6 +8,7 @@
 #include "common/common_funcs.h"
 #include "common/logging/log.h"
 #include "core/hle/kernel/errors.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/hle/kernel/thread.h"
@@ -16,30 +17,26 @@
 
 namespace Kernel {
 
-// Lists all processes that exist in the current session.
-static std::vector<SharedPtr<Process>> process_list;
-
-SharedPtr<CodeSet> CodeSet::Create(std::string name) {
-    SharedPtr<CodeSet> codeset(new CodeSet);
+SharedPtr<CodeSet> CodeSet::Create(KernelCore& kernel, std::string name) {
+    SharedPtr<CodeSet> codeset(new CodeSet(kernel));
     codeset->name = std::move(name);
     return codeset;
 }
 
-CodeSet::CodeSet() {}
-CodeSet::~CodeSet() {}
+CodeSet::CodeSet(KernelCore& kernel) : Object{kernel} {}
+CodeSet::~CodeSet() = default;
 
-u32 Process::next_process_id;
-
-SharedPtr<Process> Process::Create(std::string&& name) {
-    SharedPtr<Process> process(new Process);
+SharedPtr<Process> Process::Create(KernelCore& kernel, std::string&& name) {
+    SharedPtr<Process> process(new Process(kernel));
 
     process->name = std::move(name);
     process->flags.raw = 0;
     process->flags.memory_region.Assign(MemoryRegion::APPLICATION);
     process->status = ProcessStatus::Created;
     process->program_id = 0;
+    process->process_id = kernel.CreateNewProcessID();
 
-    process_list.push_back(process);
+    kernel.AppendNewProcess(process);
     return process;
 }
 
@@ -128,7 +125,7 @@ void Process::Run(VAddr entry_point, s32 main_thread_priority, u32 stack_size) {
     vm_manager.LogLayout();
     status = ProcessStatus::Running;
 
-    Kernel::SetupMainThread(entry_point, main_thread_priority, this);
+    Kernel::SetupMainThread(kernel, entry_point, main_thread_priority, this);
 }
 
 void Process::LoadModule(SharedPtr<CodeSet> module_, VAddr base_addr) {
@@ -231,22 +228,7 @@ ResultCode Process::UnmapMemory(VAddr dst_addr, VAddr /*src_addr*/, u64 size) {
     return vm_manager.UnmapRange(dst_addr, size);
 }
 
-Kernel::Process::Process() {}
+Kernel::Process::Process(KernelCore& kernel) : Object{kernel} {}
 Kernel::Process::~Process() {}
 
-void ClearProcessList() {
-    process_list.clear();
-}
-
-SharedPtr<Process> GetProcessById(u32 process_id) {
-    auto itr = std::find_if(
-        process_list.begin(), process_list.end(),
-        [&](const SharedPtr<Process>& process) { return process->process_id == process_id; });
-
-    if (itr == process_list.end())
-        return nullptr;
-
-    return *itr;
-}
-
 } // namespace Kernel

src/core/hle/kernel/process.h

@@ -19,6 +19,8 @@
 
 namespace Kernel {
 
+class KernelCore;
+
 struct AddressMapping {
     // Address and size must be page-aligned
     VAddr address;
@@ -62,7 +64,7 @@ struct CodeSet final : public Object {
         u32 size = 0;
     };
 
-    static SharedPtr<CodeSet> Create(std::string name);
+    static SharedPtr<CodeSet> Create(KernelCore& kernel, std::string name);
 
     std::string GetTypeName() const override {
         return "CodeSet";
@@ -109,13 +111,13 @@ struct CodeSet final : public Object {
     std::string name;
 
 private:
-    CodeSet();
+    explicit CodeSet(KernelCore& kernel);
     ~CodeSet() override;
 };
 
 class Process final : public Object {
 public:
-    static SharedPtr<Process> Create(std::string&& name);
+    static SharedPtr<Process> Create(KernelCore& kernel, std::string&& name);
 
     std::string GetTypeName() const override {
         return "Process";
@@ -129,8 +131,6 @@ public:
         return HANDLE_TYPE;
     }
 
-    static u32 next_process_id;
-
     /// Title ID corresponding to the process
     u64 program_id;
 
@@ -157,8 +157,8 @@ public:
     /// Current status of the process
     ProcessStatus status;
 
-    /// The id of this process
-    u32 process_id = next_process_id++;
+    /// The ID of this process
+    u32 process_id = 0;
 
     /**
      * Parses a list of kernel capability descriptors (as found in the ExHeader) and applies them
@@ -206,13 +206,8 @@ public:
     ResultCode UnmapMemory(VAddr dst_addr, VAddr src_addr, u64 size);
 
 private:
-    Process();
+    explicit Process(KernelCore& kernel);
     ~Process() override;
 };
 
-void ClearProcessList();
-
-/// Retrieves a process from the current list of processes.
-SharedPtr<Process> GetProcessById(u32 process_id);
-
 } // namespace Kernel

src/core/hle/kernel/resource_limit.cpp

@@ -9,31 +9,16 @@
 
 namespace Kernel {
 
-static SharedPtr<ResourceLimit> resource_limits[4];
+ResourceLimit::ResourceLimit(KernelCore& kernel) : Object{kernel} {}
+ResourceLimit::~ResourceLimit() = default;
 
-ResourceLimit::ResourceLimit() {}
-ResourceLimit::~ResourceLimit() {}
-
-SharedPtr<ResourceLimit> ResourceLimit::Create(std::string name) {
-    SharedPtr<ResourceLimit> resource_limit(new ResourceLimit);
+SharedPtr<ResourceLimit> ResourceLimit::Create(KernelCore& kernel, std::string name) {
+    SharedPtr<ResourceLimit> resource_limit(new ResourceLimit(kernel));
 
     resource_limit->name = std::move(name);
     return resource_limit;
 }
 
-SharedPtr<ResourceLimit> ResourceLimit::GetForCategory(ResourceLimitCategory category) {
-    switch (category) {
-    case ResourceLimitCategory::APPLICATION:
-    case ResourceLimitCategory::SYS_APPLET:
-    case ResourceLimitCategory::LIB_APPLET:
-    case ResourceLimitCategory::OTHER:
-        return resource_limits[static_cast<u8>(category)];
-    default:
-        LOG_CRITICAL(Kernel, "Unknown resource limit category");
-        UNREACHABLE();
-    }
-}
-
 s32 ResourceLimit::GetCurrentResourceValue(ResourceType resource) const {
     switch (resource) {
     case ResourceType::Commit:
@@ -89,66 +74,4 @@ u32 ResourceLimit::GetMaxResourceValue(ResourceType resource) const {
         return 0;
     }
 }
-
-void ResourceLimitsInit() {
-    // Create the four resource limits that the system uses
-    // Create the APPLICATION resource limit
-    SharedPtr<ResourceLimit> resource_limit = ResourceLimit::Create("Applications");
-    resource_limit->max_priority = 0x18;
-    resource_limit->max_commit = 0x4000000;
-    resource_limit->max_threads = 0x20;
-    resource_limit->max_events = 0x20;
-    resource_limit->max_mutexes = 0x20;
-    resource_limit->max_semaphores = 0x8;
-    resource_limit->max_timers = 0x8;
-    resource_limit->max_shared_mems = 0x10;
-    resource_limit->max_address_arbiters = 0x2;
-    resource_limit->max_cpu_time = 0x1E;
-    resource_limits[static_cast<u8>(ResourceLimitCategory::APPLICATION)] = resource_limit;
-
-    // Create the SYS_APPLET resource limit
-    resource_limit = ResourceLimit::Create("System Applets");
-    resource_limit->max_priority = 0x4;
-    resource_limit->max_commit = 0x5E00000;
-    resource_limit->max_threads = 0x1D;
-    resource_limit->max_events = 0xB;
-    resource_limit->max_mutexes = 0x8;
-    resource_limit->max_semaphores = 0x4;
-    resource_limit->max_timers = 0x4;
-    resource_limit->max_shared_mems = 0x8;
-    resource_limit->max_address_arbiters = 0x3;
-    resource_limit->max_cpu_time = 0x2710;
-    resource_limits[static_cast<u8>(ResourceLimitCategory::SYS_APPLET)] = resource_limit;
-
-    // Create the LIB_APPLET resource limit
-    resource_limit = ResourceLimit::Create("Library Applets");
-    resource_limit->max_priority = 0x4;
-    resource_limit->max_commit = 0x600000;
-    resource_limit->max_threads = 0xE;
-    resource_limit->max_events = 0x8;
-    resource_limit->max_mutexes = 0x8;
-    resource_limit->max_semaphores = 0x4;
-    resource_limit->max_timers = 0x4;
-    resource_limit->max_shared_mems = 0x8;
-    resource_limit->max_address_arbiters = 0x1;
-    resource_limit->max_cpu_time = 0x2710;
-    resource_limits[static_cast<u8>(ResourceLimitCategory::LIB_APPLET)] = resource_limit;
-
-    // Create the OTHER resource limit
-    resource_limit = ResourceLimit::Create("Others");
-    resource_limit->max_priority = 0x4;
-    resource_limit->max_commit = 0x2180000;
-    resource_limit->max_threads = 0xE1;
-    resource_limit->max_events = 0x108;
-    resource_limit->max_mutexes = 0x25;
-    resource_limit->max_semaphores = 0x43;
-    resource_limit->max_timers = 0x2C;
-    resource_limit->max_shared_mems = 0x1F;
-    resource_limit->max_address_arbiters = 0x2D;
-    resource_limit->max_cpu_time = 0x3E8;
-    resource_limits[static_cast<u8>(ResourceLimitCategory::OTHER)] = resource_limit;
-}
-
-void ResourceLimitsShutdown() {}
-
 } // namespace Kernel

src/core/hle/kernel/resource_limit.h

@@ -9,6 +9,8 @@
 
 namespace Kernel {
 
+class KernelCore;
+
 enum class ResourceLimitCategory : u8 {
     APPLICATION = 0,
     SYS_APPLET = 1,
@@ -34,14 +36,7 @@ public:
     /**
      * Creates a resource limit object.
      */
-    static SharedPtr<ResourceLimit> Create(std::string name = "Unknown");
-
-    /**
-     * Retrieves the resource limit associated with the specified resource limit category.
-     * @param category The resource limit category
-     * @returns The resource limit associated with the category
-     */
-    static SharedPtr<ResourceLimit> GetForCategory(ResourceLimitCategory category);
+    static SharedPtr<ResourceLimit> Create(KernelCore& kernel, std::string name = "Unknown");
 
     std::string GetTypeName() const override {
         return "ResourceLimit";
@@ -113,14 +108,8 @@ public:
     s32 current_cpu_time = 0;
 
 private:
-    ResourceLimit();
+    explicit ResourceLimit(KernelCore& kernel);
     ~ResourceLimit() override;
 };
 
-/// Initializes the resource limits
-void ResourceLimitsInit();
-
-// Destroys the resource limits
-void ResourceLimitsShutdown();
-
 } // namespace Kernel

src/core/hle/kernel/server_port.cpp

@@ -13,8 +13,8 @@
 
 namespace Kernel {
 
-ServerPort::ServerPort() {}
-ServerPort::~ServerPort() {}
+ServerPort::ServerPort(KernelCore& kernel) : WaitObject{kernel} {}
+ServerPort::~ServerPort() = default;
 
 ResultVal<SharedPtr<ServerSession>> ServerPort::Accept() {
     if (pending_sessions.empty()) {
@@ -36,10 +36,10 @@ void ServerPort::Acquire(Thread* thread) {
 }
 
 std::tuple<SharedPtr<ServerPort>, SharedPtr<ClientPort>> ServerPort::CreatePortPair(
-    u32 max_sessions, std::string name) {
+    KernelCore& kernel, u32 max_sessions, std::string name) {
 
-    SharedPtr<ServerPort> server_port(new ServerPort);
-    SharedPtr<ClientPort> client_port(new ClientPort);
+    SharedPtr<ServerPort> server_port(new ServerPort(kernel));
+    SharedPtr<ClientPort> client_port(new ClientPort(kernel));
 
     server_port->name = name + "_Server";
     client_port->name = name + "_Client";

src/core/hle/kernel/server_port.h

@@ -15,6 +15,7 @@
 namespace Kernel {
 
 class ClientPort;
+class KernelCore;
 class ServerSession;
 class SessionRequestHandler;
 
@@ -23,12 +24,13 @@ public:
     /**
      * Creates a pair of ServerPort and an associated ClientPort.
      *
+     * @param kernel The kernel instance to create the port pair under.
      * @param max_sessions Maximum number of sessions to the port
      * @param name Optional name of the ports
      * @return The created port tuple
      */
     static std::tuple<SharedPtr<ServerPort>, SharedPtr<ClientPort>> CreatePortPair(
-        u32 max_sessions, std::string name = "UnknownPort");
+        KernelCore& kernel, u32 max_sessions, std::string name = "UnknownPort");
 
     std::string GetTypeName() const override {
         return "ServerPort";
@@ -69,7 +71,7 @@ public:
     void Acquire(Thread* thread) override;
 
 private:
-    ServerPort();
+    explicit ServerPort(KernelCore& kernel);
     ~ServerPort() override;
 };
 

src/core/hle/kernel/server_session.cpp

@@ -13,6 +13,7 @@
 #include "core/hle/kernel/client_session.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/hle_ipc.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/server_session.h"
 #include "core/hle/kernel/session.h"
@@ -20,7 +21,7 @@
 
 namespace Kernel {
 
-ServerSession::ServerSession() = default;
+ServerSession::ServerSession(KernelCore& kernel) : WaitObject{kernel} {}
 ServerSession::~ServerSession() {
     // This destructor will be called automatically when the last ServerSession handle is closed by
     // the emulated application.
@@ -35,8 +36,8 @@ ServerSession::~ServerSession() {
     parent->server = nullptr;
 }
 
-ResultVal<SharedPtr<ServerSession>> ServerSession::Create(std::string name) {
-    SharedPtr<ServerSession> server_session(new ServerSession);
+ResultVal<SharedPtr<ServerSession>> ServerSession::Create(KernelCore& kernel, std::string name) {
+    SharedPtr<ServerSession> server_session(new ServerSession(kernel));
 
     server_session->name = std::move(name);
     server_session->parent = nullptr;
@@ -104,11 +105,10 @@ ResultCode ServerSession::HandleSyncRequest(SharedPtr<Thread> thread) {
     // The ServerSession received a sync request, this means that there's new data available
     // from its ClientSession, so wake up any threads that may be waiting on a svcReplyAndReceive or
     // similar.
-
     Kernel::HLERequestContext context(this);
     u32* cmd_buf = (u32*)Memory::GetPointer(thread->GetTLSAddress());
     context.PopulateFromIncomingCommandBuffer(cmd_buf, *Core::CurrentProcess(),
-                                              Kernel::g_handle_table);
+                                              kernel.HandleTable());
 
     ResultCode result = RESULT_SUCCESS;
     // If the session has been converted to a domain, handle the domain request
@@ -160,10 +160,11 @@ ResultCode ServerSession::HandleSyncRequest(SharedPtr<Thread> thread) {
     return result;
 }
 
-ServerSession::SessionPair ServerSession::CreateSessionPair(const std::string& name,
+ServerSession::SessionPair ServerSession::CreateSessionPair(KernelCore& kernel,
+                                                            const std::string& name,
                                                             SharedPtr<ClientPort> port) {
-    auto server_session = ServerSession::Create(name + "_Server").Unwrap();
-    SharedPtr<ClientSession> client_session(new ClientSession);
+    auto server_session = ServerSession::Create(kernel, name + "_Server").Unwrap();
+    SharedPtr<ClientSession> client_session(new ClientSession(kernel));
     client_session->name = name + "_Client";
 
     std::shared_ptr<Session> parent(new Session);

src/core/hle/kernel/server_session.h

@@ -15,13 +15,14 @@
 
 namespace Kernel {
 
-class ClientSession;
 class ClientPort;
+class ClientSession;
+class HLERequestContext;
+class KernelCore;
 class ServerSession;
 class Session;
 class SessionRequestHandler;
 class Thread;
-class HLERequestContext;
 
 /**
  * Kernel object representing the server endpoint of an IPC session. Sessions are the basic CTR-OS
@@ -50,11 +51,12 @@ public:
 
     /**
      * Creates a pair of ServerSession and an associated ClientSession.
+     * @param kernel      The kernal instance to create the session pair under.
      * @param name        Optional name of the ports.
      * @param client_port Optional The ClientPort that spawned this session.
      * @return The created session tuple
      */
-    static SessionPair CreateSessionPair(const std::string& name = "Unknown",
+    static SessionPair CreateSessionPair(KernelCore& kernel, const std::string& name = "Unknown",
                                          SharedPtr<ClientPort> client_port = nullptr);
 
     /**
@@ -111,16 +113,18 @@ public:
     }
 
 private:
-    ServerSession();
+    explicit ServerSession(KernelCore& kernel);
     ~ServerSession() override;
 
     /**
      * Creates a server session. The server session can have an optional HLE handler,
      * which will be invoked to handle the IPC requests that this session receives.
+     * @param kernel The kernel instance to create this server session under.
      * @param name Optional name of the server session.
      * @return The created server session
      */
-    static ResultVal<SharedPtr<ServerSession>> Create(std::string name = "Unknown");
+    static ResultVal<SharedPtr<ServerSession>> Create(KernelCore& kernel,
+                                                      std::string name = "Unknown");
 
     /// Handles a SyncRequest to a domain, forwarding the request to the proper object or closing an
     /// object handle.

src/core/hle/kernel/shared_memory.cpp

@@ -13,14 +13,14 @@
 
 namespace Kernel {
 
-SharedMemory::SharedMemory() {}
-SharedMemory::~SharedMemory() {}
+SharedMemory::SharedMemory(KernelCore& kernel) : Object{kernel} {}
+SharedMemory::~SharedMemory() = default;
 
-SharedPtr<SharedMemory> SharedMemory::Create(SharedPtr<Process> owner_process, u64 size,
-                                             MemoryPermission permissions,
+SharedPtr<SharedMemory> SharedMemory::Create(KernelCore& kernel, SharedPtr<Process> owner_process,
+                                             u64 size, MemoryPermission permissions,
                                              MemoryPermission other_permissions, VAddr address,
                                              MemoryRegion region, std::string name) {
-    SharedPtr<SharedMemory> shared_memory(new SharedMemory);
+    SharedPtr<SharedMemory> shared_memory(new SharedMemory(kernel));
 
     shared_memory->owner_process = std::move(owner_process);
     shared_memory->name = std::move(name);
@@ -59,12 +59,10 @@ SharedPtr<SharedMemory> SharedMemory::Create(SharedPtr<Process> owner_process, u
     return shared_memory;
 }
 
-SharedPtr<SharedMemory> SharedMemory::CreateForApplet(std::shared_ptr<std::vector<u8>> heap_block,
-                                                      u32 offset, u32 size,
-                                                      MemoryPermission permissions,
-                                                      MemoryPermission other_permissions,
-                                                      std::string name) {
-    SharedPtr<SharedMemory> shared_memory(new SharedMemory);
+SharedPtr<SharedMemory> SharedMemory::CreateForApplet(
+    KernelCore& kernel, std::shared_ptr<std::vector<u8>> heap_block, u32 offset, u32 size,
+    MemoryPermission permissions, MemoryPermission other_permissions, std::string name) {
+    SharedPtr<SharedMemory> shared_memory(new SharedMemory(kernel));
 
     shared_memory->owner_process = nullptr;
     shared_memory->name = std::move(name);

src/core/hle/kernel/shared_memory.h

@@ -15,6 +15,8 @@
 
 namespace Kernel {
 
+class KernelCore;
+
 /// Permissions for mapped shared memory blocks
 enum class MemoryPermission : u32 {
     None = 0,
@@ -32,6 +34,7 @@ class SharedMemory final : public Object {
 public:
     /**
      * Creates a shared memory object.
+     * @param kernel The kernel instance to create a shared memory instance under.
      * @param owner_process Process that created this shared memory object.
      * @param size Size of the memory block. Must be page-aligned.
      * @param permissions Permission restrictions applied to the process which created the block.
@@ -42,14 +45,15 @@ public:
      * linear heap.
      * @param name Optional object name, used for debugging purposes.
      */
-    static SharedPtr<SharedMemory> Create(SharedPtr<Process> owner_process, u64 size,
-                                          MemoryPermission permissions,
+    static SharedPtr<SharedMemory> Create(KernelCore& kernel, SharedPtr<Process> owner_process,
+                                          u64 size, MemoryPermission permissions,
                                           MemoryPermission other_permissions, VAddr address = 0,
                                           MemoryRegion region = MemoryRegion::BASE,
                                           std::string name = "Unknown");
 
     /**
      * Creates a shared memory object from a block of memory managed by an HLE applet.
+     * @param kernel The kernel instance to create a shared memory instance under.
      * @param heap_block Heap block of the HLE applet.
      * @param offset The offset into the heap block that the SharedMemory will map.
      * @param size Size of the memory block. Must be page-aligned.
@@ -58,7 +62,8 @@ public:
      * block.
      * @param name Optional object name, used for debugging purposes.
      */
-    static SharedPtr<SharedMemory> CreateForApplet(std::shared_ptr<std::vector<u8>> heap_block,
+    static SharedPtr<SharedMemory> CreateForApplet(KernelCore& kernel,
+                                                   std::shared_ptr<std::vector<u8>> heap_block,
                                                    u32 offset, u32 size,
                                                    MemoryPermission permissions,
                                                    MemoryPermission other_permissions,
@@ -125,7 +130,7 @@ public:
     std::string name;
 
 private:
-    SharedMemory();
+    explicit SharedMemory(KernelCore& kernel);
     ~SharedMemory() override;
 };
 

src/core/hle/kernel/svc.cpp

@@ -12,16 +12,20 @@
 #include "common/logging/log.h"
 #include "common/microprofile.h"
 #include "common/string_util.h"
+#include "core/arm/exclusive_monitor.h"
 #include "core/core.h"
+#include "core/core_cpu.h"
 #include "core/core_timing.h"
 #include "core/hle/kernel/address_arbiter.h"
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/client_session.h"
 #include "core/hle/kernel/event.h"
 #include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/mutex.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
+#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/shared_memory.h"
 #include "core/hle/kernel/svc.h"
 #include "core/hle/kernel/svc_wrap.h"
@@ -87,13 +91,15 @@ static ResultCode ConnectToNamedPort(Handle* out_handle, VAddr port_name_address
     CASCADE_RESULT(client_session, client_port->Connect());
 
     // Return the client session
-    CASCADE_RESULT(*out_handle, g_handle_table.Create(client_session));
+    auto& kernel = Core::System::GetInstance().Kernel();
+    CASCADE_RESULT(*out_handle, kernel.HandleTable().Create(client_session));
     return RESULT_SUCCESS;
 }
 
 /// Makes a blocking IPC call to an OS service.
 static ResultCode SendSyncRequest(Handle handle) {
-    SharedPtr<ClientSession> session = g_handle_table.Get<ClientSession>(handle);
+    auto& kernel = Core::System::GetInstance().Kernel();
+    SharedPtr<ClientSession> session = kernel.HandleTable().Get<ClientSession>(handle);
     if (!session) {
         LOG_ERROR(Kernel_SVC, "called with invalid handle=0x{:08X}", handle);
         return ERR_INVALID_HANDLE;
@@ -112,7 +118,8 @@ static ResultCode SendSyncRequest(Handle handle) {
 static ResultCode GetThreadId(u32* thread_id, Handle thread_handle) {
     LOG_TRACE(Kernel_SVC, "called thread=0x{:08X}", thread_handle);
 
-    const SharedPtr<Thread> thread = g_handle_table.Get<Thread>(thread_handle);
+    auto& kernel = Core::System::GetInstance().Kernel();
+    const SharedPtr<Thread> thread = kernel.HandleTable().Get<Thread>(thread_handle);
     if (!thread) {
         return ERR_INVALID_HANDLE;
     }
@@ -125,7 +132,8 @@ static ResultCode GetThreadId(u32* thread_id, Handle thread_handle) {
 static ResultCode GetProcessId(u32* process_id, Handle process_handle) {
     LOG_TRACE(Kernel_SVC, "called process=0x{:08X}", process_handle);
 
-    const SharedPtr<Process> process = g_handle_table.Get<Process>(process_handle);
+    auto& kernel = Core::System::GetInstance().Kernel();
+    const SharedPtr<Process> process = kernel.HandleTable().Get<Process>(process_handle);
     if (!process) {
         return ERR_INVALID_HANDLE;
     }
@@ -168,10 +176,11 @@ static ResultCode WaitSynchronization(Handle* index, VAddr handles_address, u64
 
     using ObjectPtr = SharedPtr<WaitObject>;
     std::vector<ObjectPtr> objects(handle_count);
+    auto& kernel = Core::System::GetInstance().Kernel();
 
     for (u64 i = 0; i < handle_count; ++i) {
         const Handle handle = Memory::Read32(handles_address + i * sizeof(Handle));
-        const auto object = g_handle_table.Get<WaitObject>(handle);
+        const auto object = kernel.HandleTable().Get<WaitObject>(handle);
 
         if (object == nullptr) {
             return ERR_INVALID_HANDLE;
@@ -219,7 +228,8 @@ static ResultCode WaitSynchronization(Handle* index, VAddr handles_address, u64
 static ResultCode CancelSynchronization(Handle thread_handle) {
     LOG_TRACE(Kernel_SVC, "called thread=0x{:X}", thread_handle);
 
-    const SharedPtr<Thread> thread = g_handle_table.Get<Thread>(thread_handle);
+    auto& kernel = Core::System::GetInstance().Kernel();
+    const SharedPtr<Thread> thread = kernel.HandleTable().Get<Thread>(thread_handle);
     if (!thread) {
         return ERR_INVALID_HANDLE;
     }
@@ -239,7 +249,9 @@ static ResultCode ArbitrateLock(Handle holding_thread_handle, VAddr mutex_addr,
               "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);
+    auto& handle_table = Core::System::GetInstance().Kernel().HandleTable();
+    return Mutex::TryAcquire(handle_table, mutex_addr, holding_thread_handle,
+                             requesting_thread_handle);
 }
 
 /// Unlock a mutex
@@ -352,7 +364,8 @@ static ResultCode GetThreadContext(Handle handle, VAddr addr) {
 
 /// Gets the priority for the specified thread
 static ResultCode GetThreadPriority(u32* priority, Handle handle) {
-    const SharedPtr<Thread> thread = g_handle_table.Get<Thread>(handle);
+    auto& kernel = Core::System::GetInstance().Kernel();
+    const SharedPtr<Thread> thread = kernel.HandleTable().Get<Thread>(handle);
     if (!thread)
         return ERR_INVALID_HANDLE;
 
@@ -366,7 +379,8 @@ static ResultCode SetThreadPriority(Handle handle, u32 priority) {
         return ERR_OUT_OF_RANGE;
     }
 
-    SharedPtr<Thread> thread = g_handle_table.Get<Thread>(handle);
+    auto& kernel = Core::System::GetInstance().Kernel();
+    SharedPtr<Thread> thread = kernel.HandleTable().Get<Thread>(handle);
     if (!thread)
         return ERR_INVALID_HANDLE;
 
@@ -395,7 +409,8 @@ static ResultCode MapSharedMemory(Handle shared_memory_handle, VAddr addr, u64 s
               "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);
+    auto& kernel = Core::System::GetInstance().Kernel();
+    auto shared_memory = kernel.HandleTable().Get<SharedMemory>(shared_memory_handle);
     if (!shared_memory) {
         return ERR_INVALID_HANDLE;
     }
@@ -423,7 +438,8 @@ static ResultCode UnmapSharedMemory(Handle shared_memory_handle, VAddr addr, u64
     LOG_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);
+    auto& kernel = Core::System::GetInstance().Kernel();
+    auto shared_memory = kernel.HandleTable().Get<SharedMemory>(shared_memory_handle);
 
     return shared_memory->Unmap(Core::CurrentProcess().get(), addr);
 }
@@ -431,7 +447,9 @@ static ResultCode UnmapSharedMemory(Handle shared_memory_handle, VAddr addr, u64
 /// Query process memory
 static ResultCode QueryProcessMemory(MemoryInfo* memory_info, PageInfo* /*page_info*/,
                                      Handle process_handle, u64 addr) {
-    SharedPtr<Process> process = g_handle_table.Get<Process>(process_handle);
+
+    auto& kernel = Core::System::GetInstance().Kernel();
+    SharedPtr<Process> process = kernel.HandleTable().Get<Process>(process_handle);
     if (!process) {
         return ERR_INVALID_HANDLE;
     }
@@ -528,10 +546,11 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V
         break;
     }
 
+    auto& kernel = Core::System::GetInstance().Kernel();
     CASCADE_RESULT(SharedPtr<Thread> thread,
-                   Thread::Create(name, entry_point, priority, arg, processor_id, stack_top,
+                   Thread::Create(kernel, name, entry_point, priority, arg, processor_id, stack_top,
                                   Core::CurrentProcess()));
-    CASCADE_RESULT(thread->guest_handle, g_handle_table.Create(thread));
+    CASCADE_RESULT(thread->guest_handle, kernel.HandleTable().Create(thread));
     *out_handle = thread->guest_handle;
 
     Core::System::GetInstance().CpuCore(thread->processor_id).PrepareReschedule();
@@ -548,7 +567,8 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V
 static ResultCode StartThread(Handle thread_handle) {
     LOG_TRACE(Kernel_SVC, "called thread=0x{:08X}", thread_handle);
 
-    const SharedPtr<Thread> thread = g_handle_table.Get<Thread>(thread_handle);
+    auto& kernel = Core::System::GetInstance().Kernel();
+    const SharedPtr<Thread> thread = kernel.HandleTable().Get<Thread>(thread_handle);
     if (!thread) {
         return ERR_INVALID_HANDLE;
     }
@@ -595,7 +615,8 @@ static ResultCode WaitProcessWideKeyAtomic(VAddr mutex_addr, VAddr condition_var
         "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);
+    auto& kernel = Core::System::GetInstance().Kernel();
+    SharedPtr<Thread> thread = kernel.HandleTable().Get<Thread>(thread_handle);
     ASSERT(thread);
 
     CASCADE_CODE(Mutex::Release(mutex_addr));
@@ -704,8 +725,9 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
                                                mutex_val | Mutex::MutexHasWaitersFlag));
 
             // The mutex is already owned by some other thread, make this thread wait on it.
+            auto& kernel = Core::System::GetInstance().Kernel();
             Handle owner_handle = static_cast<Handle>(mutex_val & Mutex::MutexOwnerMask);
-            auto owner = g_handle_table.Get<Thread>(owner_handle);
+            auto owner = kernel.HandleTable().Get<Thread>(owner_handle);
             ASSERT(owner);
             ASSERT(thread->status == ThreadStatus::WaitMutex);
             thread->wakeup_callback = nullptr;
@@ -783,14 +805,20 @@ static u64 GetSystemTick() {
 /// Close a handle
 static ResultCode CloseHandle(Handle handle) {
     LOG_TRACE(Kernel_SVC, "Closing handle 0x{:08X}", handle);
-    return g_handle_table.Close(handle);
+
+    auto& kernel = Core::System::GetInstance().Kernel();
+    return kernel.HandleTable().Close(handle);
 }
 
 /// Reset an event
 static ResultCode ResetSignal(Handle handle) {
     LOG_WARNING(Kernel_SVC, "(STUBBED) called handle 0x{:08X}", handle);
-    auto event = g_handle_table.Get<Event>(handle);
+
+    auto& kernel = Core::System::GetInstance().Kernel();
+    auto event = kernel.HandleTable().Get<Event>(handle);
+
     ASSERT(event != nullptr);
+
     event->Clear();
     return RESULT_SUCCESS;
 }
@@ -806,7 +834,8 @@ static ResultCode CreateTransferMemory(Handle* handle, VAddr addr, u64 size, u32
 static ResultCode GetThreadCoreMask(Handle thread_handle, u32* core, u64* mask) {
     LOG_TRACE(Kernel_SVC, "called, handle=0x{:08X}", thread_handle);
 
-    const SharedPtr<Thread> thread = g_handle_table.Get<Thread>(thread_handle);
+    auto& kernel = Core::System::GetInstance().Kernel();
+    const SharedPtr<Thread> thread = kernel.HandleTable().Get<Thread>(thread_handle);
     if (!thread) {
         return ERR_INVALID_HANDLE;
     }
@@ -821,7 +850,8 @@ static ResultCode SetThreadCoreMask(Handle thread_handle, u32 core, u64 mask) {
     LOG_DEBUG(Kernel_SVC, "called, handle=0x{:08X}, mask=0x{:16X}, core=0x{:X}", thread_handle,
               mask, core);
 
-    const SharedPtr<Thread> thread = g_handle_table.Get<Thread>(thread_handle);
+    auto& kernel = Core::System::GetInstance().Kernel();
+    const SharedPtr<Thread> thread = kernel.HandleTable().Get<Thread>(thread_handle);
     if (!thread) {
         return ERR_INVALID_HANDLE;
     }
@@ -861,19 +891,23 @@ static ResultCode CreateSharedMemory(Handle* handle, u64 size, u32 local_permiss
                                      u32 remote_permissions) {
     LOG_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,
+
+    auto& kernel = Core::System::GetInstance().Kernel();
+    auto& handle_table = kernel.HandleTable();
+    auto shared_mem_handle =
+        SharedMemory::Create(kernel, handle_table.Get<Process>(KernelHandle::CurrentProcess), size,
                              static_cast<MemoryPermission>(local_permissions),
                              static_cast<MemoryPermission>(remote_permissions));
 
-    CASCADE_RESULT(*handle, g_handle_table.Create(sharedMemHandle));
+    CASCADE_RESULT(*handle, handle_table.Create(shared_mem_handle));
     return RESULT_SUCCESS;
 }
 
 static ResultCode ClearEvent(Handle handle) {
     LOG_TRACE(Kernel_SVC, "called, event=0x{:08X}", handle);
 
-    SharedPtr<Event> evt = g_handle_table.Get<Event>(handle);
+    auto& kernel = Core::System::GetInstance().Kernel();
+    SharedPtr<Event> evt = kernel.HandleTable().Get<Event>(handle);
     if (evt == nullptr)
         return ERR_INVALID_HANDLE;
     evt->Clear();

src/core/hle/kernel/thread.cpp

@@ -16,22 +16,21 @@
 #include "common/thread_queue_list.h"
 #include "core/arm/arm_interface.h"
 #include "core/core.h"
+#include "core/core_cpu.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/process.h"
+#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/thread.h"
-#include "core/hle/lock.h"
 #include "core/hle/result.h"
 #include "core/memory.h"
 
 namespace Kernel {
 
-/// Event type for the thread wake up event
-static CoreTiming::EventType* ThreadWakeupEventType = nullptr;
-
 bool Thread::ShouldWait(Thread* thread) const {
     return status != ThreadStatus::Dead;
 }
@@ -40,32 +39,17 @@ void Thread::Acquire(Thread* thread) {
     ASSERT_MSG(!ShouldWait(thread), "object unavailable!");
 }
 
-// TODO(yuriks): This can be removed if Thread objects are explicitly pooled in the future, allowing
-//               us to simply use a pool index or similar.
-static Kernel::HandleTable wakeup_callback_handle_table;
-
-// The first available thread id at startup
-static u32 next_thread_id;
-
-/**
- * Creates a new thread ID
- * @return The new thread ID
- */
-inline static u32 const NewThreadId() {
-    return next_thread_id++;
-}
-
-Thread::Thread() {}
-Thread::~Thread() {}
+Thread::Thread(KernelCore& kernel) : WaitObject{kernel} {}
+Thread::~Thread() = default;
 
 void Thread::Stop() {
     // Cancel any outstanding wakeup events for this thread
-    CoreTiming::UnscheduleEvent(ThreadWakeupEventType, callback_handle);
-    wakeup_callback_handle_table.Close(callback_handle);
+    CoreTiming::UnscheduleEvent(kernel.ThreadWakeupCallbackEventType(), callback_handle);
+    kernel.ThreadWakeupCallbackHandleTable().Close(callback_handle);
     callback_handle = 0;
 
     // Clean up thread from ready queue
-    // This is only needed when the thread is termintated forcefully (SVC TerminateProcess)
+    // This is only needed when the thread is terminated forcefully (SVC TerminateProcess)
     if (status == ThreadStatus::Ready) {
         scheduler->UnscheduleThread(this, current_priority);
     }
@@ -98,63 +82,6 @@ void ExitCurrentThread() {
     Core::System::GetInstance().CurrentScheduler().RemoveThread(thread);
 }
 
-/**
- * Callback that will wake up the thread it was scheduled for
- * @param thread_handle The handle of the thread that's been awoken
- * @param cycles_late The number of CPU cycles that have passed since the desired wakeup time
- */
-static void ThreadWakeupCallback(u64 thread_handle, int cycles_late) {
-    const auto proper_handle = static_cast<Handle>(thread_handle);
-
-    // Lock the global kernel mutex when we enter the kernel HLE.
-    std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock);
-
-    SharedPtr<Thread> thread = wakeup_callback_handle_table.Get<Thread>(proper_handle);
-    if (thread == nullptr) {
-        LOG_CRITICAL(Kernel, "Callback fired for invalid thread {:08X}", proper_handle);
-        return;
-    }
-
-    bool resume = true;
-
-    if (thread->status == ThreadStatus::WaitSynchAny ||
-        thread->status == ThreadStatus::WaitSynchAll ||
-        thread->status == ThreadStatus::WaitHLEEvent) {
-        // Remove the thread from each of its waiting objects' waitlists
-        for (auto& object : thread->wait_objects)
-            object->RemoveWaitingThread(thread.get());
-        thread->wait_objects.clear();
-
-        // Invoke the wakeup callback before clearing the wait objects
-        if (thread->wakeup_callback)
-            resume = thread->wakeup_callback(ThreadWakeupReason::Timeout, thread, nullptr, 0);
-    }
-
-    if (thread->mutex_wait_address != 0 || thread->condvar_wait_address != 0 ||
-        thread->wait_handle) {
-        ASSERT(thread->status == ThreadStatus::WaitMutex);
-        thread->mutex_wait_address = 0;
-        thread->condvar_wait_address = 0;
-        thread->wait_handle = 0;
-
-        auto lock_owner = thread->lock_owner;
-        // Threads waking up by timeout from WaitProcessWideKey do not perform priority inheritance
-        // and don't have a lock owner unless SignalProcessWideKey was called first and the thread
-        // wasn't awakened due to the mutex already being acquired.
-        if (lock_owner) {
-            lock_owner->RemoveMutexWaiter(thread);
-        }
-    }
-
-    if (thread->arb_wait_address != 0) {
-        ASSERT(thread->status == ThreadStatus::WaitArb);
-        thread->arb_wait_address = 0;
-    }
-
-    if (resume)
-        thread->ResumeFromWait();
-}
-
 void Thread::WakeAfterDelay(s64 nanoseconds) {
     // Don't schedule a wakeup if the thread wants to wait forever
     if (nanoseconds == -1)
@@ -162,12 +89,12 @@ void Thread::WakeAfterDelay(s64 nanoseconds) {
 
     // This function might be called from any thread so we have to be cautious and use the
     // thread-safe version of ScheduleEvent.
-    CoreTiming::ScheduleEventThreadsafe(CoreTiming::nsToCycles(nanoseconds), ThreadWakeupEventType,
-                                        callback_handle);
+    CoreTiming::ScheduleEventThreadsafe(CoreTiming::nsToCycles(nanoseconds),
+                                        kernel.ThreadWakeupCallbackEventType(), callback_handle);
 }
 
 void Thread::CancelWakeupTimer() {
-    CoreTiming::UnscheduleEventThreadsafe(ThreadWakeupEventType, callback_handle);
+    CoreTiming::UnscheduleEventThreadsafe(kernel.ThreadWakeupCallbackEventType(), callback_handle);
 }
 
 static boost::optional<s32> GetNextProcessorId(u64 mask) {
@@ -294,9 +221,9 @@ static void ResetThreadContext(Core::ARM_Interface::ThreadContext& context, VAdd
     context.fpscr = 0;
 }
 
-ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point, u32 priority,
-                                            u64 arg, s32 processor_id, VAddr stack_top,
-                                            SharedPtr<Process> owner_process) {
+ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name, VAddr entry_point,
+                                            u32 priority, u64 arg, s32 processor_id,
+                                            VAddr stack_top, SharedPtr<Process> owner_process) {
     // Check if priority is in ranged. Lowest priority -> highest priority id.
     if (priority > THREADPRIO_LOWEST) {
         LOG_ERROR(Kernel_SVC, "Invalid thread priority: {}", priority);
@@ -316,9 +243,9 @@ ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point,
         return ResultCode(-1);
     }
 
-    SharedPtr<Thread> thread(new Thread);
+    SharedPtr<Thread> thread(new Thread(kernel));
 
-    thread->thread_id = NewThreadId();
+    thread->thread_id = kernel.CreateNewThreadID();
     thread->status = ThreadStatus::Dormant;
     thread->entry_point = entry_point;
     thread->stack_top = stack_top;
@@ -333,7 +260,7 @@ ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point,
     thread->condvar_wait_address = 0;
     thread->wait_handle = 0;
     thread->name = std::move(name);
-    thread->callback_handle = wakeup_callback_handle_table.Create(thread).Unwrap();
+    thread->callback_handle = kernel.ThreadWakeupCallbackHandleTable().Create(thread).Unwrap();
     thread->owner_process = owner_process;
     thread->scheduler = Core::System::GetInstance().Scheduler(processor_id);
     thread->scheduler->AddThread(thread, priority);
@@ -383,19 +310,19 @@ void Thread::BoostPriority(u32 priority) {
     current_priority = priority;
 }
 
-SharedPtr<Thread> SetupMainThread(VAddr entry_point, u32 priority,
+SharedPtr<Thread> SetupMainThread(KernelCore& kernel, VAddr entry_point, u32 priority,
                                   SharedPtr<Process> owner_process) {
     // Setup page table so we can write to memory
     SetCurrentPageTable(&Core::CurrentProcess()->vm_manager.page_table);
 
     // Initialize new "main" thread
-    auto thread_res = Thread::Create("main", entry_point, priority, 0, THREADPROCESSORID_0,
+    auto thread_res = Thread::Create(kernel, "main", entry_point, priority, 0, THREADPROCESSORID_0,
                                      Memory::STACK_AREA_VADDR_END, std::move(owner_process));
 
     SharedPtr<Thread> thread = std::move(thread_res).Unwrap();
 
     // Register 1 must be a handle to the main thread
-    thread->guest_handle = Kernel::g_handle_table.Create(thread).Unwrap();
+    thread->guest_handle = kernel.HandleTable().Create(thread).Unwrap();
 
     thread->context.cpu_registers[1] = thread->guest_handle;
 
@@ -528,13 +455,4 @@ Thread* GetCurrentThread() {
     return Core::System::GetInstance().CurrentScheduler().GetCurrentThread();
 }
 
-void ThreadingInit() {
-    ThreadWakeupEventType = CoreTiming::RegisterEvent("ThreadWakeupCallback", ThreadWakeupCallback);
-    next_thread_id = 1;
-}
-
-void ThreadingShutdown() {
-    Kernel::ClearProcessList();
-}
-
 } // namespace Kernel

src/core/hle/kernel/thread.h

@@ -56,6 +56,7 @@ enum class ThreadWakeupReason {
 
 namespace Kernel {
 
+class KernelCore;
 class Process;
 class Scheduler;
 
@@ -63,6 +64,7 @@ class Thread final : public WaitObject {
 public:
     /**
      * Creates and returns a new thread. The new thread is immediately scheduled
+     * @param kernel The kernel instance this thread will be created under.
      * @param name The friendly name desired for the thread
      * @param entry_point The address at which the thread should start execution
      * @param priority The thread's priority
@@ -72,8 +74,9 @@ public:
      * @param owner_process The parent process for the thread
      * @return A shared pointer to the newly created thread
      */
-    static ResultVal<SharedPtr<Thread>> Create(std::string name, VAddr entry_point, u32 priority,
-                                               u64 arg, s32 processor_id, VAddr stack_top,
+    static ResultVal<SharedPtr<Thread>> Create(KernelCore& kernel, std::string name,
+                                               VAddr entry_point, u32 priority, u64 arg,
+                                               s32 processor_id, VAddr stack_top,
                                                SharedPtr<Process> owner_process);
 
     std::string GetName() const override {
@@ -263,7 +266,7 @@ public:
     u64 affinity_mask{0x1};
 
 private:
-    Thread();
+    explicit Thread(KernelCore& kernel);
     ~Thread() override;
 
     std::shared_ptr<std::vector<u8>> tls_memory = std::make_shared<std::vector<u8>>();
@@ -271,12 +274,13 @@ private:
 
 /**
  * Sets up the primary application thread
+ * @param kernel The kernel instance to create the main thread under.
  * @param entry_point The address at which the thread should start execution
  * @param priority The priority to give the main thread
  * @param owner_process The parent process for the main thread
  * @return A shared pointer to the main thread
  */
-SharedPtr<Thread> SetupMainThread(VAddr entry_point, u32 priority,
+SharedPtr<Thread> SetupMainThread(KernelCore& kernel, VAddr entry_point, u32 priority,
                                   SharedPtr<Process> owner_process);
 
 /**
@@ -294,14 +298,4 @@ void WaitCurrentThread_Sleep();
  */
 void ExitCurrentThread();
 
-/**
- * Initialize threading
- */
-void ThreadingInit();
-
-/**
- * Shutdown threading
- */
-void ThreadingShutdown();
-
 } // namespace Kernel

src/core/hle/kernel/timer.cpp

@@ -2,36 +2,31 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include <cinttypes>
 #include "common/assert.h"
 #include "common/logging/log.h"
+#include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
 #include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/thread.h"
 #include "core/hle/kernel/timer.h"
 
 namespace Kernel {
 
-/// The event type of the generic timer callback event
-static CoreTiming::EventType* timer_callback_event_type = nullptr;
-// TODO(yuriks): This can be removed if Timer objects are explicitly pooled in the future, allowing
-//               us to simply use a pool index or similar.
-static Kernel::HandleTable timer_callback_handle_table;
+Timer::Timer(KernelCore& kernel) : WaitObject{kernel} {}
+Timer::~Timer() = default;
 
-Timer::Timer() {}
-Timer::~Timer() {}
-
-SharedPtr<Timer> Timer::Create(ResetType reset_type, std::string name) {
-    SharedPtr<Timer> timer(new Timer);
+SharedPtr<Timer> Timer::Create(KernelCore& kernel, ResetType reset_type, std::string name) {
+    SharedPtr<Timer> timer(new Timer(kernel));
 
     timer->reset_type = reset_type;
     timer->signaled = false;
     timer->name = std::move(name);
     timer->initial_delay = 0;
     timer->interval_delay = 0;
-    timer->callback_handle = timer_callback_handle_table.Create(timer).Unwrap();
+    timer->callback_handle = kernel.CreateTimerCallbackHandle(timer).Unwrap();
 
     return timer;
 }
@@ -58,13 +53,13 @@ void Timer::Set(s64 initial, s64 interval) {
         // Immediately invoke the callback
         Signal(0);
     } else {
-        CoreTiming::ScheduleEvent(CoreTiming::nsToCycles(initial), timer_callback_event_type,
+        CoreTiming::ScheduleEvent(CoreTiming::nsToCycles(initial), kernel.TimerCallbackEventType(),
                                   callback_handle);
     }
 }
 
 void Timer::Cancel() {
-    CoreTiming::UnscheduleEvent(timer_callback_event_type, callback_handle);
+    CoreTiming::UnscheduleEvent(kernel.TimerCallbackEventType(), callback_handle);
 }
 
 void Timer::Clear() {
@@ -89,28 +84,8 @@ void Timer::Signal(int cycles_late) {
     if (interval_delay != 0) {
         // Reschedule the timer with the interval delay
         CoreTiming::ScheduleEvent(CoreTiming::nsToCycles(interval_delay) - cycles_late,
-                                  timer_callback_event_type, callback_handle);
+                                  kernel.TimerCallbackEventType(), callback_handle);
     }
 }
 
-/// The timer callback event, called when a timer is fired
-static void TimerCallback(u64 timer_handle, int cycles_late) {
-    SharedPtr<Timer> timer =
-        timer_callback_handle_table.Get<Timer>(static_cast<Handle>(timer_handle));
-
-    if (timer == nullptr) {
-        LOG_CRITICAL(Kernel, "Callback fired for invalid timer {:016X}", timer_handle);
-        return;
-    }
-
-    timer->Signal(cycles_late);
-}
-
-void TimersInit() {
-    timer_callback_handle_table.Clear();
-    timer_callback_event_type = CoreTiming::RegisterEvent("TimerCallback", TimerCallback);
-}
-
-void TimersShutdown() {}
-
 } // namespace Kernel

src/core/hle/kernel/timer.h

@@ -10,15 +10,19 @@
 
 namespace Kernel {
 
+class KernelCore;
+
 class Timer final : public WaitObject {
 public:
     /**
      * Creates a timer
+     * @param kernel The kernel instance to create the timer callback handle for.
      * @param reset_type ResetType describing how to create the timer
      * @param name Optional name of timer
      * @return The created Timer
      */
-    static SharedPtr<Timer> Create(ResetType reset_type, std::string name = "Unknown");
+    static SharedPtr<Timer> Create(KernelCore& kernel, ResetType reset_type,
+                                   std::string name = "Unknown");
 
     std::string GetTypeName() const override {
         return "Timer";
@@ -68,7 +72,7 @@ public:
     void Signal(int cycles_late);
 
 private:
-    Timer();
+    explicit Timer(KernelCore& kernel);
     ~Timer() override;
 
     ResetType reset_type; ///< The ResetType of this timer
@@ -83,9 +87,4 @@ private:
     Handle callback_handle;
 };
 
-/// Initializes the required variables for timers
-void TimersInit();
-/// Tears down the timer variables
-void TimersShutdown();
-
 } // namespace Kernel

src/core/hle/kernel/wait_object.cpp

@@ -12,6 +12,9 @@
 
 namespace Kernel {
 
+WaitObject::WaitObject(KernelCore& kernel) : Object{kernel} {}
+WaitObject::~WaitObject() = default;
+
 void WaitObject::AddWaitingThread(SharedPtr<Thread> thread) {
     auto itr = std::find(waiting_threads.begin(), waiting_threads.end(), thread);
     if (itr == waiting_threads.end())

src/core/hle/kernel/wait_object.h

@@ -11,11 +11,15 @@
 
 namespace Kernel {
 
+class KernelCore;
 class Thread;
 
 /// Class that represents a Kernel object that a thread can be waiting on
 class WaitObject : public Object {
 public:
+    explicit WaitObject(KernelCore& kernel);
+    ~WaitObject() override;
+
     /**
      * Check if the specified thread should wait until the object is available
      * @param thread The thread about which we're deciding.

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

@@ -160,8 +160,9 @@ ISelfController::ISelfController(std::shared_ptr<NVFlinger::NVFlinger> nvflinger
     };
     RegisterHandlers(functions);
 
+    auto& kernel = Core::System::GetInstance().Kernel();
     launchable_event =
-        Kernel::Event::Create(Kernel::ResetType::Sticky, "ISelfController:LaunchableEvent");
+        Kernel::Event::Create(kernel, Kernel::ResetType::Sticky, "ISelfController:LaunchableEvent");
 }
 
 void ISelfController::SetFocusHandlingMode(Kernel::HLERequestContext& ctx) {
@@ -332,7 +333,8 @@ ICommonStateGetter::ICommonStateGetter() : ServiceFramework("ICommonStateGetter"
     };
     RegisterHandlers(functions);
 
-    event = Kernel::Event::Create(Kernel::ResetType::OneShot, "ICommonStateGetter:Event");
+    auto& kernel = Core::System::GetInstance().Kernel();
+    event = Kernel::Event::Create(kernel, Kernel::ResetType::OneShot, "ICommonStateGetter:Event");
 }
 
 void ICommonStateGetter::GetBootMode(Kernel::HLERequestContext& ctx) {
@@ -505,7 +507,8 @@ public:
         };
         RegisterHandlers(functions);
 
-        state_changed_event = Kernel::Event::Create(Kernel::ResetType::OneShot,
+        auto& kernel = Core::System::GetInstance().Kernel();
+        state_changed_event = Kernel::Event::Create(kernel, Kernel::ResetType::OneShot,
                                                     "ILibraryAppletAccessor:StateChangedEvent");
     }
 

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

@@ -47,7 +47,9 @@ public:
         RegisterHandlers(functions);
 
         // This is the event handle used to check if the audio buffer was released
-        buffer_event = Kernel::Event::Create(Kernel::ResetType::Sticky, "IAudioOutBufferReleased");
+        auto& kernel = Core::System::GetInstance().Kernel();
+        buffer_event =
+            Kernel::Event::Create(kernel, Kernel::ResetType::Sticky, "IAudioOutBufferReleased");
 
         stream = audio_core.OpenStream(audio_params.sample_rate, audio_params.channel_count,
                                        "IAudioOut", [=]() { buffer_event->Signal(); });

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

@@ -35,8 +35,9 @@ public:
         };
         RegisterHandlers(functions);
 
+        auto& kernel = Core::System::GetInstance().Kernel();
         system_event =
-            Kernel::Event::Create(Kernel::ResetType::Sticky, "IAudioRenderer:SystemEvent");
+            Kernel::Event::Create(kernel, Kernel::ResetType::Sticky, "IAudioRenderer:SystemEvent");
         renderer = std::make_unique<AudioCore::AudioRenderer>(audren_params, system_event);
     }
 
@@ -121,8 +122,9 @@ public:
         };
         RegisterHandlers(functions);
 
-        buffer_event =
-            Kernel::Event::Create(Kernel::ResetType::OneShot, "IAudioOutBufferReleasedEvent");
+        auto& kernel = Core::System::GetInstance().Kernel();
+        buffer_event = Kernel::Event::Create(kernel, Kernel::ResetType::OneShot,
+                                             "IAudioOutBufferReleasedEvent");
     }
 
 private:

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

@@ -19,6 +19,7 @@
 #include "core/hle/service/filesystem/fsp_ldr.h"
 #include "core/hle/service/filesystem/fsp_pr.h"
 #include "core/hle/service/filesystem/fsp_srv.h"
+#include "filesystem.h"
 
 namespace Service::FileSystem {
 
@@ -304,6 +305,12 @@ ResultVal<FileSys::VirtualDir> OpenSDMC() {
     return sdmc_factory->Open();
 }
 
+std::shared_ptr<FileSys::RegisteredCacheUnion> GetUnionContents() {
+    return std::make_shared<FileSys::RegisteredCacheUnion>(
+        std::vector<std::shared_ptr<FileSys::RegisteredCache>>{
+            GetSystemNANDContents(), GetUserNANDContents(), GetSDMCContents()});
+}
+
 std::shared_ptr<FileSys::RegisteredCache> GetSystemNANDContents() {
     LOG_TRACE(Service_FS, "Opening System NAND Contents");
 

src/core/hle/service/filesystem/filesystem.h

@@ -7,11 +7,13 @@
 #include <memory>
 #include "common/common_types.h"
 #include "core/file_sys/directory.h"
+#include "core/file_sys/vfs.h"
 #include "core/hle/result.h"
 
 namespace FileSys {
 class BISFactory;
 class RegisteredCache;
+class RegisteredCacheUnion;
 class RomFSFactory;
 class SaveDataFactory;
 class SDMCFactory;
@@ -44,6 +46,8 @@ ResultVal<FileSys::VirtualDir> OpenSaveData(FileSys::SaveDataSpaceId space,
                                             FileSys::SaveDataDescriptor save_struct);
 ResultVal<FileSys::VirtualDir> OpenSDMC();
 
+std::shared_ptr<FileSys::RegisteredCacheUnion> GetUnionContents();
+
 std::shared_ptr<FileSys::RegisteredCache> GetSystemNANDContents();
 std::shared_ptr<FileSys::RegisteredCache> GetUserNANDContents();
 std::shared_ptr<FileSys::RegisteredCache> GetSDMCContents();

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

@@ -4,6 +4,7 @@
 
 #include <atomic>
 #include "common/logging/log.h"
+#include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
 #include "core/frontend/emu_window.h"
@@ -35,9 +36,10 @@ public:
         };
         RegisterHandlers(functions);
 
+        auto& kernel = Core::System::GetInstance().Kernel();
         shared_mem = Kernel::SharedMemory::Create(
-            nullptr, 0x40000, Kernel::MemoryPermission::ReadWrite, Kernel::MemoryPermission::Read,
-            0, Kernel::MemoryRegion::BASE, "HID:SharedMemory");
+            kernel, nullptr, 0x40000, Kernel::MemoryPermission::ReadWrite,
+            Kernel::MemoryPermission::Read, 0, Kernel::MemoryRegion::BASE, "HID:SharedMemory");
 
         // Register update callbacks
         pad_update_event = CoreTiming::RegisterEvent(
@@ -402,7 +404,8 @@ public:
 
         RegisterHandlers(functions);
 
-        event = Kernel::Event::Create(Kernel::ResetType::OneShot, "hid:EventHandle");
+        auto& kernel = Core::System::GetInstance().Kernel();
+        event = Kernel::Event::Create(kernel, Kernel::ResetType::OneShot, "hid:EventHandle");
     }
     ~Hid() = default;
 

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

@@ -46,11 +46,13 @@ public:
         };
         RegisterHandlers(functions);
 
-        activate_event = Kernel::Event::Create(Kernel::ResetType::OneShot, "IUser:ActivateEvent");
+        auto& kernel = Core::System::GetInstance().Kernel();
+        activate_event =
+            Kernel::Event::Create(kernel, Kernel::ResetType::OneShot, "IUser:ActivateEvent");
         deactivate_event =
-            Kernel::Event::Create(Kernel::ResetType::OneShot, "IUser:DeactivateEvent");
-        availability_change_event =
-            Kernel::Event::Create(Kernel::ResetType::OneShot, "IUser:AvailabilityChangeEvent");
+            Kernel::Event::Create(kernel, Kernel::ResetType::OneShot, "IUser:DeactivateEvent");
+        availability_change_event = Kernel::Event::Create(kernel, Kernel::ResetType::OneShot,
+                                                          "IUser:AvailabilityChangeEvent");
     }
 
 private:

src/core/hle/service/nifm/nifm.cpp

@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include "core/core.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/event.h"
 #include "core/hle/service/nifm/nifm.h"
@@ -54,8 +55,9 @@ public:
         };
         RegisterHandlers(functions);
 
-        event1 = Kernel::Event::Create(Kernel::ResetType::OneShot, "IRequest:Event1");
-        event2 = Kernel::Event::Create(Kernel::ResetType::OneShot, "IRequest:Event2");
+        auto& kernel = Core::System::GetInstance().Kernel();
+        event1 = Kernel::Event::Create(kernel, Kernel::ResetType::OneShot, "IRequest:Event1");
+        event2 = Kernel::Event::Create(kernel, Kernel::ResetType::OneShot, "IRequest:Event2");
     }
 
 private:

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

@@ -266,8 +266,9 @@ void PL_U::GetSharedMemoryNativeHandle(Kernel::HLERequestContext& ctx) {
         SHARED_FONT_MEM_VADDR, shared_font, 0, SHARED_FONT_MEM_SIZE, Kernel::MemoryState::Shared);
 
     // Create shared font memory object
+    auto& kernel = Core::System::GetInstance().Kernel();
     shared_font_mem = Kernel::SharedMemory::Create(
-        Core::CurrentProcess(), SHARED_FONT_MEM_SIZE, Kernel::MemoryPermission::ReadWrite,
+        kernel, Core::CurrentProcess(), SHARED_FONT_MEM_SIZE, Kernel::MemoryPermission::ReadWrite,
         Kernel::MemoryPermission::Read, SHARED_FONT_MEM_VADDR, Kernel::MemoryRegion::BASE,
         "PL_U:shared_font_mem");
 

src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp

@@ -7,6 +7,7 @@
 #include "core/core.h"
 #include "core/hle/service/nvdrv/devices/nvdisp_disp0.h"
 #include "core/hle/service/nvdrv/devices/nvmap.h"
+#include "core/perf_stats.h"
 #include "video_core/gpu.h"
 #include "video_core/renderer_base.h"
 
@@ -31,7 +32,7 @@ void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u3
         transform, crop_rect};
 
     auto& instance = Core::System::GetInstance();
-    instance.perf_stats.EndGameFrame();
+    instance.GetPerfStats().EndGameFrame();
     instance.Renderer().SwapBuffers(framebuffer);
 }
 

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

@@ -10,6 +10,7 @@
 #include "common/common_types.h"
 #include "common/swap.h"
 #include "core/hle/service/nvdrv/devices/nvdevice.h"
+#include "video_core/memory_manager.h"
 
 namespace Service::Nvidia::Devices {
 

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

@@ -4,6 +4,7 @@
 
 #include <cinttypes>
 #include "common/logging/log.h"
+#include "core/core.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/event.h"
 #include "core/hle/service/nvdrv/interface.h"
@@ -107,7 +108,8 @@ NVDRV::NVDRV(std::shared_ptr<Module> nvdrv, const char* name)
     };
     RegisterHandlers(functions);
 
-    query_event = Kernel::Event::Create(Kernel::ResetType::OneShot, "NVDRV::query_event");
+    auto& kernel = Core::System::GetInstance().Kernel();
+    query_event = Kernel::Event::Create(kernel, Kernel::ResetType::OneShot, "NVDRV::query_event");
 }
 
 } // namespace Service::Nvidia

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

@@ -6,14 +6,16 @@
 
 #include "common/assert.h"
 #include "common/logging/log.h"
+#include "core/core.h"
 #include "core/hle/service/nvflinger/buffer_queue.h"
 
 namespace Service {
 namespace NVFlinger {
 
 BufferQueue::BufferQueue(u32 id, u64 layer_id) : id(id), layer_id(layer_id) {
+    auto& kernel = Core::System::GetInstance().Kernel();
     buffer_wait_event =
-        Kernel::Event::Create(Kernel::ResetType::Sticky, "BufferQueue NativeHandle");
+        Kernel::Event::Create(kernel, Kernel::ResetType::Sticky, "BufferQueue NativeHandle");
 }
 
 void BufferQueue::SetPreallocatedBuffer(u32 slot, const IGBPBuffer& igbp_buffer) {

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

@@ -17,6 +17,7 @@
 #include "core/hle/service/nvdrv/nvdrv.h"
 #include "core/hle/service/nvflinger/buffer_queue.h"
 #include "core/hle/service/nvflinger/nvflinger.h"
+#include "core/perf_stats.h"
 #include "video_core/renderer_base.h"
 #include "video_core/video_core.h"
 
@@ -137,7 +138,7 @@ void NVFlinger::Compose() {
             auto& system_instance = Core::System::GetInstance();
 
             // There was no queued buffer to draw, render previous frame
-            system_instance.perf_stats.EndGameFrame();
+            system_instance.GetPerfStats().EndGameFrame();
             system_instance.Renderer().SwapBuffers({});
             continue;
         }
@@ -161,7 +162,8 @@ void NVFlinger::Compose() {
 Layer::Layer(u64 id, std::shared_ptr<BufferQueue> queue) : id(id), buffer_queue(std::move(queue)) {}
 
 Display::Display(u64 id, std::string name) : id(id), name(std::move(name)) {
-    vsync_event = Kernel::Event::Create(Kernel::ResetType::Pulse, "Display VSync Event");
+    auto& kernel = Core::System::GetInstance().Kernel();
+    vsync_event = Kernel::Event::Create(kernel, Kernel::ResetType::Pulse, "Display VSync Event");
 }
 
 } // namespace Service::NVFlinger

src/core/hle/service/service.cpp

@@ -107,19 +107,24 @@ void ServiceFrameworkBase::InstallAsService(SM::ServiceManager& service_manager)
 
 void ServiceFrameworkBase::InstallAsNamedPort() {
     ASSERT(port == nullptr);
+
+    auto& kernel = Core::System::GetInstance().Kernel();
     SharedPtr<ServerPort> server_port;
     SharedPtr<ClientPort> client_port;
-    std::tie(server_port, client_port) = ServerPort::CreatePortPair(max_sessions, service_name);
+    std::tie(server_port, client_port) =
+        ServerPort::CreatePortPair(kernel, max_sessions, service_name);
     server_port->SetHleHandler(shared_from_this());
     AddNamedPort(service_name, std::move(client_port));
 }
 
 Kernel::SharedPtr<Kernel::ClientPort> ServiceFrameworkBase::CreatePort() {
     ASSERT(port == nullptr);
+
+    auto& kernel = Core::System::GetInstance().Kernel();
     Kernel::SharedPtr<Kernel::ServerPort> server_port;
     Kernel::SharedPtr<Kernel::ClientPort> client_port;
     std::tie(server_port, client_port) =
-        Kernel::ServerPort::CreatePortPair(max_sessions, service_name);
+        Kernel::ServerPort::CreatePortPair(kernel, max_sessions, service_name);
     port = MakeResult<Kernel::SharedPtr<Kernel::ServerPort>>(std::move(server_port)).Unwrap();
     port->SetHleHandler(shared_from_this());
     return client_port;

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

@@ -4,6 +4,7 @@
 
 #include <tuple>
 #include "common/assert.h"
+#include "core/core.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/client_session.h"
@@ -47,9 +48,11 @@ ResultVal<Kernel::SharedPtr<Kernel::ServerPort>> ServiceManager::RegisterService
     if (registered_services.find(name) != registered_services.end())
         return ERR_ALREADY_REGISTERED;
 
+    auto& kernel = Core::System::GetInstance().Kernel();
     Kernel::SharedPtr<Kernel::ServerPort> server_port;
     Kernel::SharedPtr<Kernel::ClientPort> client_port;
-    std::tie(server_port, client_port) = Kernel::ServerPort::CreatePortPair(max_sessions, name);
+    std::tie(server_port, client_port) =
+        Kernel::ServerPort::CreatePortPair(kernel, max_sessions, name);
 
     registered_services.emplace(std::move(name), std::move(client_port));
     return MakeResult<Kernel::SharedPtr<Kernel::ServerPort>>(std::move(server_port));

src/core/loader/deconstructed_rom_directory.cpp

@@ -6,10 +6,13 @@
 #include "common/common_funcs.h"
 #include "common/file_util.h"
 #include "common/logging/log.h"
+#include "core/core.h"
 #include "core/file_sys/content_archive.h"
 #include "core/file_sys/control_metadata.h"
+#include "core/file_sys/patch_manager.h"
 #include "core/file_sys/romfs_factory.h"
 #include "core/gdbstub/gdbstub.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/hle/service/filesystem/filesystem.h"
@@ -19,8 +22,9 @@
 
 namespace Loader {
 
-AppLoader_DeconstructedRomDirectory::AppLoader_DeconstructedRomDirectory(FileSys::VirtualFile file_)
-    : AppLoader(std::move(file_)) {
+AppLoader_DeconstructedRomDirectory::AppLoader_DeconstructedRomDirectory(FileSys::VirtualFile file_,
+                                                                         bool override_update)
+    : AppLoader(std::move(file_)), override_update(override_update) {
     const auto dir = file->GetContainingDirectory();
 
     // Icon
@@ -59,14 +63,14 @@ AppLoader_DeconstructedRomDirectory::AppLoader_DeconstructedRomDirectory(FileSys
 
     if (nacp_file != nullptr) {
         FileSys::NACP nacp(nacp_file);
-        title_id = nacp.GetTitleId();
         name = nacp.GetApplicationName();
     }
 }
 
 AppLoader_DeconstructedRomDirectory::AppLoader_DeconstructedRomDirectory(
-    FileSys::VirtualDir directory)
-    : AppLoader(directory->GetFile("main")), dir(std::move(directory)) {}
+    FileSys::VirtualDir directory, bool override_update)
+    : AppLoader(directory->GetFile("main")), dir(std::move(directory)),
+      override_update(override_update) {}
 
 FileType AppLoader_DeconstructedRomDirectory::IdentifyType(const FileSys::VirtualFile& file) {
     if (FileSys::IsDirectoryExeFS(file->GetContainingDirectory())) {
@@ -88,7 +92,8 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(
         dir = file->GetContainingDirectory();
     }
 
-    const FileSys::VirtualFile npdm = dir->GetFile("main.npdm");
+    // Read meta to determine title ID
+    FileSys::VirtualFile npdm = dir->GetFile("main.npdm");
     if (npdm == nullptr)
         return ResultStatus::ErrorMissingNPDM;
 
@@ -96,6 +101,21 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(
     if (result != ResultStatus::Success) {
         return result;
     }
+
+    if (override_update) {
+        const FileSys::PatchManager patch_manager(metadata.GetTitleID());
+        dir = patch_manager.PatchExeFS(dir);
+    }
+
+    // Reread in case PatchExeFS affected the main.npdm
+    npdm = dir->GetFile("main.npdm");
+    if (npdm == nullptr)
+        return ResultStatus::ErrorMissingNPDM;
+
+    ResultStatus result2 = metadata.Load(npdm);
+    if (result2 != ResultStatus::Success) {
+        return result2;
+    }
     metadata.Print();
 
     const FileSys::ProgramAddressSpaceType arch_bits{metadata.GetAddressSpaceType()};
@@ -117,10 +137,12 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(
         }
     }
 
+    auto& kernel = Core::System::GetInstance().Kernel();
+    title_id = metadata.GetTitleID();
     process->program_id = metadata.GetTitleID();
     process->svc_access_mask.set();
     process->resource_limit =
-        Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION);
+        kernel.ResourceLimitForCategory(Kernel::ResourceLimitCategory::APPLICATION);
     process->Run(Memory::PROCESS_IMAGE_VADDR, metadata.GetMainThreadPriority(),
                  metadata.GetMainThreadStackSize());
 
@@ -156,8 +178,6 @@ ResultStatus AppLoader_DeconstructedRomDirectory::ReadIcon(std::vector<u8>& buff
 }
 
 ResultStatus AppLoader_DeconstructedRomDirectory::ReadProgramId(u64& out_program_id) {
-    if (name.empty())
-        return ResultStatus::ErrorNoControl;
     out_program_id = title_id;
     return ResultStatus::Success;
 }
@@ -169,4 +189,8 @@ ResultStatus AppLoader_DeconstructedRomDirectory::ReadTitle(std::string& title)
     return ResultStatus::Success;
 }
 
+bool AppLoader_DeconstructedRomDirectory::IsRomFSUpdatable() const {
+    return false;
+}
+
 } // namespace Loader

src/core/loader/deconstructed_rom_directory.h

@@ -20,10 +20,12 @@ namespace Loader {
  */
 class AppLoader_DeconstructedRomDirectory final : public AppLoader {
 public:
-    explicit AppLoader_DeconstructedRomDirectory(FileSys::VirtualFile main_file);
+    explicit AppLoader_DeconstructedRomDirectory(FileSys::VirtualFile main_file,
+                                                 bool override_update = false);
 
     // Overload to accept exefs directory. Must contain 'main' and 'main.npdm'
-    explicit AppLoader_DeconstructedRomDirectory(FileSys::VirtualDir directory);
+    explicit AppLoader_DeconstructedRomDirectory(FileSys::VirtualDir directory,
+                                                 bool override_update = false);
 
     /**
      * Returns the type of the file
@@ -42,6 +44,7 @@ public:
     ResultStatus ReadIcon(std::vector<u8>& buffer) override;
     ResultStatus ReadProgramId(u64& out_program_id) override;
     ResultStatus ReadTitle(std::string& title) override;
+    bool IsRomFSUpdatable() const override;
 
 private:
     FileSys::ProgramMetadata metadata;
@@ -51,6 +54,7 @@ private:
     std::vector<u8> icon_data;
     std::string name;
     u64 title_id{};
+    bool override_update;
 };
 
 } // namespace Loader

src/core/loader/elf.cpp

@@ -9,6 +9,8 @@
 #include "common/common_types.h"
 #include "common/file_util.h"
 #include "common/logging/log.h"
+#include "core/core.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/loader/elf.h"
@@ -300,7 +302,8 @@ SharedPtr<CodeSet> ElfReader::LoadInto(u32 vaddr) {
     std::vector<u8> program_image(total_image_size);
     size_t current_image_position = 0;
 
-    SharedPtr<CodeSet> codeset = CodeSet::Create("");
+    auto& kernel = Core::System::GetInstance().Kernel();
+    SharedPtr<CodeSet> codeset = CodeSet::Create(kernel, "");
 
     for (unsigned int i = 0; i < header->e_phnum; ++i) {
         Elf32_Phdr* p = &segments[i];
@@ -400,8 +403,9 @@ ResultStatus AppLoader_ELF::Load(Kernel::SharedPtr<Kernel::Process>& process) {
     process->svc_access_mask.set();
 
     // Attach the default resource limit (APPLICATION) to the process
+    auto& kernel = Core::System::GetInstance().Kernel();
     process->resource_limit =
-        Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION);
+        kernel.ResourceLimitForCategory(Kernel::ResourceLimitCategory::APPLICATION);
 
     process->Run(codeset->entrypoint, 48, Memory::DEFAULT_STACK_SIZE);
 

src/core/loader/loader.cpp

@@ -15,6 +15,7 @@
 #include "core/loader/nca.h"
 #include "core/loader/nro.h"
 #include "core/loader/nso.h"
+#include "core/loader/nsp.h"
 #include "core/loader/xci.h"
 
 namespace Loader {
@@ -34,6 +35,7 @@ FileType IdentifyFile(FileSys::VirtualFile file) {
     CHECK_TYPE(NCA)
     CHECK_TYPE(XCI)
     CHECK_TYPE(NAX)
+    CHECK_TYPE(NSP)
 
 #undef CHECK_TYPE
 
@@ -59,6 +61,8 @@ FileType GuessFromFilename(const std::string& name) {
         return FileType::NCA;
     if (extension == "xci")
         return FileType::XCI;
+    if (extension == "nsp")
+        return FileType::NSP;
 
     return FileType::Unknown;
 }
@@ -77,6 +81,8 @@ std::string GetFileTypeString(FileType type) {
         return "XCI";
     case FileType::NAX:
         return "NAX";
+    case FileType::NSP:
+        return "NSP";
     case FileType::DeconstructedRomDirectory:
         return "Directory";
     case FileType::Error:
@@ -87,7 +93,7 @@ std::string GetFileTypeString(FileType type) {
     return "unknown";
 }
 
-constexpr std::array<const char*, 49> RESULT_MESSAGES{
+constexpr std::array<const char*, 58> RESULT_MESSAGES{
     "The operation completed successfully.",
     "The loader requested to load is already loaded.",
     "The operation is not implemented.",
@@ -137,6 +143,15 @@ constexpr std::array<const char*, 49> RESULT_MESSAGES{
     "The AES Key Generation Source could not be found.",
     "The SD Save Key Source could not be found.",
     "The SD NCA Key Source could not be found.",
+    "The NSP file is missing a Program-type NCA.",
+    "The BKTR-type NCA has a bad BKTR header.",
+    "The BKTR Subsection entry is not located immediately after the Relocation entry.",
+    "The BKTR Subsection entry is not at the end of the media block.",
+    "The BKTR-type NCA has a bad Relocation block.",
+    "The BKTR-type NCA has a bad Subsection block.",
+    "The BKTR-type NCA has a bad Relocation bucket.",
+    "The BKTR-type NCA has a bad Subsection bucket.",
+    "The BKTR-type NCA is missing the base RomFS.",
 };
 
 std::ostream& operator<<(std::ostream& os, ResultStatus status) {
@@ -179,6 +194,10 @@ static std::unique_ptr<AppLoader> GetFileLoader(FileSys::VirtualFile file, FileT
     case FileType::NAX:
         return std::make_unique<AppLoader_NAX>(std::move(file));
 
+    // NX NSP (Nintendo Submission Package) file format
+    case FileType::NSP:
+        return std::make_unique<AppLoader_NSP>(std::move(file));
+
     // NX deconstructed ROM directory.
     case FileType::DeconstructedRomDirectory:
         return std::make_unique<AppLoader_DeconstructedRomDirectory>(std::move(file));

src/core/loader/loader.h

@@ -31,6 +31,7 @@ enum class FileType {
     NSO,
     NRO,
     NCA,
+    NSP,
     XCI,
     NAX,
     DeconstructedRomDirectory,
@@ -107,6 +108,15 @@ enum class ResultStatus : u16 {
     ErrorMissingAESKeyGenerationSource,
     ErrorMissingSDSaveKeySource,
     ErrorMissingSDNCAKeySource,
+    ErrorNSPMissingProgramNCA,
+    ErrorBadBKTRHeader,
+    ErrorBKTRSubsectionNotAfterRelocation,
+    ErrorBKTRSubsectionNotAtEnd,
+    ErrorBadRelocationBlock,
+    ErrorBadSubsectionBlock,
+    ErrorBadRelocationBuckets,
+    ErrorBadSubsectionBuckets,
+    ErrorMissingBKTRBaseRomFS,
 };
 
 std::ostream& operator<<(std::ostream& os, ResultStatus status);
@@ -197,13 +207,22 @@ public:
     }
 
     /**
-     * Get the update RomFS of the application
-     * Since the RomFS can be huge, we return a file reference instead of copying to a buffer
-     * @param file The file containing the RomFS
-     * @return ResultStatus result of function
+     * Get whether or not updates can be applied to the RomFS.
+     * By default, this is true, however for formats where it cannot be guaranteed that the RomFS is
+     * the base game it should be set to false.
+     * @return bool whether or not updatable.
      */
-    virtual ResultStatus ReadUpdateRomFS(FileSys::VirtualFile& file) {
-        return ResultStatus::ErrorNotImplemented;
+    virtual bool IsRomFSUpdatable() const {
+        return true;
+    }
+
+    /**
+     * Gets the difference between the start of the IVFC header and the start of level 6 (RomFS)
+     * data. Needed for bktr patching.
+     * @return IVFC offset for romfs.
+     */
+    virtual u64 ReadRomFSIVFCOffset() const {
+        return 0;
     }
 
     /**

src/core/loader/nca.cpp

@@ -48,7 +48,7 @@ ResultStatus AppLoader_NCA::Load(Kernel::SharedPtr<Kernel::Process>& process) {
     if (exefs == nullptr)
         return ResultStatus::ErrorNoExeFS;
 
-    directory_loader = std::make_unique<AppLoader_DeconstructedRomDirectory>(exefs);
+    directory_loader = std::make_unique<AppLoader_DeconstructedRomDirectory>(exefs, true);
 
     const auto load_result = directory_loader->Load(process);
     if (load_result != ResultStatus::Success)
@@ -71,6 +71,12 @@ ResultStatus AppLoader_NCA::ReadRomFS(FileSys::VirtualFile& dir) {
     return ResultStatus::Success;
 }
 
+u64 AppLoader_NCA::ReadRomFSIVFCOffset() const {
+    if (nca == nullptr)
+        return 0;
+    return nca->GetBaseIVFCOffset();
+}
+
 ResultStatus AppLoader_NCA::ReadProgramId(u64& out_program_id) {
     if (nca == nullptr || nca->GetStatus() != ResultStatus::Success)
         return ResultStatus::ErrorNotInitialized;

src/core/loader/nca.h

@@ -37,6 +37,7 @@ public:
     ResultStatus Load(Kernel::SharedPtr<Kernel::Process>& process) override;
 
     ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
+    u64 ReadRomFSIVFCOffset() const override;
     ResultStatus ReadProgramId(u64& out_program_id) override;
 
 private:

src/core/loader/nro.cpp

@@ -14,6 +14,7 @@
 #include "core/file_sys/control_metadata.h"
 #include "core/file_sys/vfs_offset.h"
 #include "core/gdbstub/gdbstub.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/loader/nro.h"
@@ -136,7 +137,8 @@ bool AppLoader_NRO::LoadNro(FileSys::VirtualFile file, VAddr load_base) {
     }
 
     // Build program image
-    Kernel::SharedPtr<Kernel::CodeSet> codeset = Kernel::CodeSet::Create("");
+    auto& kernel = Core::System::GetInstance().Kernel();
+    Kernel::SharedPtr<Kernel::CodeSet> codeset = Kernel::CodeSet::Create(kernel, "");
     std::vector<u8> program_image = file->ReadBytes(PageAlignSize(nro_header.file_size));
     if (program_image.size() != PageAlignSize(nro_header.file_size)) {
         return {};
@@ -185,9 +187,10 @@ ResultStatus AppLoader_NRO::Load(Kernel::SharedPtr<Kernel::Process>& process) {
         return ResultStatus::ErrorLoadingNRO;
     }
 
+    auto& kernel = Core::System::GetInstance().Kernel();
     process->svc_access_mask.set();
     process->resource_limit =
-        Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION);
+        kernel.ResourceLimitForCategory(Kernel::ResourceLimitCategory::APPLICATION);
     process->Run(base_addr, THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);
 
     is_loaded = true;
@@ -229,4 +232,9 @@ ResultStatus AppLoader_NRO::ReadTitle(std::string& title) {
     title = nacp->GetApplicationName();
     return ResultStatus::Success;
 }
+
+bool AppLoader_NRO::IsRomFSUpdatable() const {
+    return false;
+}
+
 } // namespace Loader

src/core/loader/nro.h

@@ -39,6 +39,7 @@ public:
     ResultStatus ReadProgramId(u64& out_program_id) override;
     ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
     ResultStatus ReadTitle(std::string& title) override;
+    bool IsRomFSUpdatable() const override;
 
 private:
     bool LoadNro(FileSys::VirtualFile file, VAddr load_base);

src/core/loader/nso.cpp

@@ -11,6 +11,7 @@
 #include "common/swap.h"
 #include "core/core.h"
 #include "core/gdbstub/gdbstub.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/loader/nso.h"
@@ -100,7 +101,8 @@ VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base) {
         return {};
 
     // Build program image
-    Kernel::SharedPtr<Kernel::CodeSet> codeset = Kernel::CodeSet::Create("");
+    auto& kernel = Core::System::GetInstance().Kernel();
+    Kernel::SharedPtr<Kernel::CodeSet> codeset = Kernel::CodeSet::Create(kernel, "");
     std::vector<u8> program_image;
     for (std::size_t i = 0; i < nso_header.segments.size(); ++i) {
         const std::vector<u8> compressed_data =
@@ -151,9 +153,10 @@ ResultStatus AppLoader_NSO::Load(Kernel::SharedPtr<Kernel::Process>& process) {
     LoadModule(file, Memory::PROCESS_IMAGE_VADDR);
     LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", file->GetName(), Memory::PROCESS_IMAGE_VADDR);
 
+    auto& kernel = Core::System::GetInstance().Kernel();
     process->svc_access_mask.set();
     process->resource_limit =
-        Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION);
+        kernel.ResourceLimitForCategory(Kernel::ResourceLimitCategory::APPLICATION);
     process->Run(Memory::PROCESS_IMAGE_VADDR, THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);
 
     is_loaded = true;

src/core/loader/nsp.cpp

@@ -0,0 +1,135 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <vector>
+
+#include "common/common_types.h"
+#include "core/file_sys/card_image.h"
+#include "core/file_sys/content_archive.h"
+#include "core/file_sys/control_metadata.h"
+#include "core/file_sys/registered_cache.h"
+#include "core/file_sys/romfs.h"
+#include "core/file_sys/submission_package.h"
+#include "core/hle/kernel/process.h"
+#include "core/loader/deconstructed_rom_directory.h"
+#include "core/loader/nca.h"
+#include "core/loader/nsp.h"
+
+namespace Loader {
+
+AppLoader_NSP::AppLoader_NSP(FileSys::VirtualFile file)
+    : AppLoader(file), nsp(std::make_unique<FileSys::NSP>(file)),
+      title_id(nsp->GetProgramTitleID()) {
+
+    if (nsp->GetStatus() != ResultStatus::Success)
+        return;
+    if (nsp->IsExtractedType())
+        return;
+
+    const auto control_nca =
+        nsp->GetNCA(nsp->GetFirstTitleID(), FileSys::ContentRecordType::Control);
+    if (control_nca == nullptr || control_nca->GetStatus() != ResultStatus::Success)
+        return;
+
+    const auto romfs = FileSys::ExtractRomFS(control_nca->GetRomFS());
+    if (romfs == nullptr)
+        return;
+
+    for (const auto& language : FileSys::LANGUAGE_NAMES) {
+        icon_file = romfs->GetFile("icon_" + std::string(language) + ".dat");
+        if (icon_file != nullptr)
+            break;
+    }
+
+    const auto nacp_raw = romfs->GetFile("control.nacp");
+    if (nacp_raw == nullptr)
+        return;
+    nacp_file = std::make_shared<FileSys::NACP>(nacp_raw);
+}
+
+AppLoader_NSP::~AppLoader_NSP() = default;
+
+FileType AppLoader_NSP::IdentifyType(const FileSys::VirtualFile& file) {
+    FileSys::NSP nsp(file);
+
+    if (nsp.GetStatus() == ResultStatus::Success) {
+        // Extracted Type case
+        if (nsp.IsExtractedType() && nsp.GetExeFS() != nullptr &&
+            FileSys::IsDirectoryExeFS(nsp.GetExeFS()) && nsp.GetRomFS() != nullptr) {
+            return FileType::NSP;
+        }
+
+        // Non-Ectracted Type case
+        if (!nsp.IsExtractedType() &&
+            nsp.GetNCA(nsp.GetFirstTitleID(), FileSys::ContentRecordType::Program) != nullptr &&
+            AppLoader_NCA::IdentifyType(nsp.GetNCAFile(
+                nsp.GetFirstTitleID(), FileSys::ContentRecordType::Program)) == FileType::NCA) {
+            return FileType::NSP;
+        }
+    }
+
+    return FileType::Error;
+}
+
+ResultStatus AppLoader_NSP::Load(Kernel::SharedPtr<Kernel::Process>& process) {
+    if (is_loaded) {
+        return ResultStatus::ErrorAlreadyLoaded;
+    }
+
+    if (nsp->IsExtractedType()) {
+        secondary_loader = std::make_unique<AppLoader_DeconstructedRomDirectory>(nsp->GetExeFS());
+    } else {
+        if (title_id == 0)
+            return ResultStatus::ErrorNSPMissingProgramNCA;
+
+        secondary_loader = std::make_unique<AppLoader_NCA>(
+            nsp->GetNCAFile(title_id, FileSys::ContentRecordType::Program));
+
+        if (nsp->GetStatus() != ResultStatus::Success)
+            return nsp->GetStatus();
+
+        if (nsp->GetProgramStatus(title_id) != ResultStatus::Success)
+            return nsp->GetProgramStatus(title_id);
+
+        if (nsp->GetNCA(title_id, FileSys::ContentRecordType::Program) == nullptr) {
+            if (!Core::Crypto::KeyManager::KeyFileExists(false))
+                return ResultStatus::ErrorMissingProductionKeyFile;
+            return ResultStatus::ErrorNSPMissingProgramNCA;
+        }
+    }
+
+    const auto result = secondary_loader->Load(process);
+    if (result != ResultStatus::Success)
+        return result;
+
+    is_loaded = true;
+
+    return ResultStatus::Success;
+}
+
+ResultStatus AppLoader_NSP::ReadRomFS(FileSys::VirtualFile& dir) {
+    return secondary_loader->ReadRomFS(dir);
+}
+
+ResultStatus AppLoader_NSP::ReadProgramId(u64& out_program_id) {
+    if (title_id == 0)
+        return ResultStatus::ErrorNotInitialized;
+    out_program_id = title_id;
+    return ResultStatus::Success;
+}
+
+ResultStatus AppLoader_NSP::ReadIcon(std::vector<u8>& buffer) {
+    if (icon_file == nullptr)
+        return ResultStatus::ErrorNoControl;
+    buffer = icon_file->ReadAllBytes();
+    return ResultStatus::Success;
+}
+
+ResultStatus AppLoader_NSP::ReadTitle(std::string& title) {
+    if (nacp_file == nullptr)
+        return ResultStatus::ErrorNoControl;
+    title = nacp_file->GetApplicationName();
+    return ResultStatus::Success;
+}
+} // namespace Loader

src/core/loader/nsp.h

@@ -0,0 +1,54 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include "common/common_types.h"
+#include "core/file_sys/vfs.h"
+#include "core/loader/loader.h"
+
+namespace FileSys {
+class NACP;
+class NSP;
+} // namespace FileSys
+
+namespace Loader {
+
+class AppLoader_NCA;
+
+/// Loads an XCI file
+class AppLoader_NSP final : public AppLoader {
+public:
+    explicit AppLoader_NSP(FileSys::VirtualFile file);
+    ~AppLoader_NSP() override;
+
+    /**
+     * Returns the type of the file
+     * @param file std::shared_ptr<VfsFile> open file
+     * @return FileType found, or FileType::Error if this loader doesn't know it
+     */
+    static FileType IdentifyType(const FileSys::VirtualFile& file);
+
+    FileType GetFileType() override {
+        return IdentifyType(file);
+    }
+
+    ResultStatus Load(Kernel::SharedPtr<Kernel::Process>& process) override;
+
+    ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
+    ResultStatus ReadProgramId(u64& out_program_id) override;
+    ResultStatus ReadIcon(std::vector<u8>& buffer) override;
+    ResultStatus ReadTitle(std::string& title) override;
+
+private:
+    std::unique_ptr<FileSys::NSP> nsp;
+    std::unique_ptr<AppLoader> secondary_loader;
+
+    FileSys::VirtualFile icon_file;
+    std::shared_ptr<FileSys::NACP> nacp_file;
+    u64 title_id;
+};
+
+} // namespace Loader

src/core/loader/xci.cpp

@@ -8,6 +8,7 @@
 #include "core/file_sys/card_image.h"
 #include "core/file_sys/content_archive.h"
 #include "core/file_sys/control_metadata.h"
+#include "core/file_sys/patch_manager.h"
 #include "core/file_sys/romfs.h"
 #include "core/hle/kernel/process.h"
 #include "core/loader/nca.h"
@@ -17,14 +18,21 @@ namespace Loader {
 
 AppLoader_XCI::AppLoader_XCI(FileSys::VirtualFile file)
     : AppLoader(file), xci(std::make_unique<FileSys::XCI>(file)),
-      nca_loader(std::make_unique<AppLoader_NCA>(
-          xci->GetNCAFileByType(FileSys::NCAContentType::Program))) {
+      nca_loader(std::make_unique<AppLoader_NCA>(xci->GetProgramNCAFile())) {
     if (xci->GetStatus() != ResultStatus::Success)
         return;
+
     const auto control_nca = xci->GetNCAByType(FileSys::NCAContentType::Control);
+
     if (control_nca == nullptr || control_nca->GetStatus() != ResultStatus::Success)
         return;
-    const auto romfs = FileSys::ExtractRomFS(control_nca->GetRomFS());
+
+    auto romfs_raw = control_nca->GetRomFS();
+    FileSys::PatchManager patch{xci->GetNCAByType(FileSys::NCAContentType::Program)->GetTitleId()};
+    romfs_raw = patch.PatchRomFS(romfs_raw, control_nca->GetBaseIVFCOffset(),
+                                 FileSys::ContentRecordType::Control);
+
+    const auto romfs = FileSys::ExtractRomFS(romfs_raw);
     if (romfs == nullptr)
         return;
     for (const auto& language : FileSys::LANGUAGE_NAMES) {
@@ -64,11 +72,11 @@ ResultStatus AppLoader_XCI::Load(Kernel::SharedPtr<Kernel::Process>& process) {
     if (xci->GetProgramNCAStatus() != ResultStatus::Success)
         return xci->GetProgramNCAStatus();
 
-    const auto nca = xci->GetNCAFileByType(FileSys::NCAContentType::Program);
+    const auto nca = xci->GetProgramNCA();
     if (nca == nullptr && !Core::Crypto::KeyManager::KeyFileExists(false))
         return ResultStatus::ErrorMissingProductionKeyFile;
 
-    auto result = nca_loader->Load(process);
+    const auto result = nca_loader->Load(process);
     if (result != ResultStatus::Success)
         return result;
 

src/core/memory.cpp

@@ -251,8 +251,8 @@ std::string ReadCString(VAddr vaddr, std::size_t max_length) {
     return string;
 }
 
-void RasterizerMarkRegionCached(Tegra::GPUVAddr gpu_addr, u64 size, bool cached) {
-    if (gpu_addr == 0) {
+void RasterizerMarkRegionCached(VAddr vaddr, u64 size, bool cached) {
+    if (vaddr == 0) {
         return;
     }
 
@@ -261,19 +261,8 @@ void RasterizerMarkRegionCached(Tegra::GPUVAddr gpu_addr, u64 size, bool cached)
     // CPU pages, hence why we iterate on a CPU page basis (note: GPU page size is different). This
     // assumes the specified GPU address region is contiguous as well.
 
-    u64 num_pages = ((gpu_addr + size - 1) >> PAGE_BITS) - (gpu_addr >> PAGE_BITS) + 1;
-    for (unsigned i = 0; i < num_pages; ++i, gpu_addr += PAGE_SIZE) {
-        boost::optional<VAddr> maybe_vaddr =
-            Core::System::GetInstance().GPU().MemoryManager().GpuToCpuAddress(gpu_addr);
-        // The GPU <-> CPU virtual memory mapping is not 1:1
-        if (!maybe_vaddr) {
-            LOG_ERROR(HW_Memory,
-                      "Trying to flush a cached region to an invalid physical address {:016X}",
-                      gpu_addr);
-            continue;
-        }
-        VAddr vaddr = *maybe_vaddr;
-
+    u64 num_pages = ((vaddr + size - 1) >> PAGE_BITS) - (vaddr >> PAGE_BITS) + 1;
+    for (unsigned i = 0; i < num_pages; ++i, vaddr += PAGE_SIZE) {
         PageType& page_type = current_page_table->attributes[vaddr >> PAGE_BITS];
 
         if (cached) {
@@ -344,29 +333,19 @@ void RasterizerFlushVirtualRegion(VAddr start, u64 size, FlushMode mode) {
 
         const VAddr overlap_start = std::max(start, region_start);
         const VAddr overlap_end = std::min(end, region_end);
-
-        const std::vector<Tegra::GPUVAddr> gpu_addresses =
-            system_instance.GPU().MemoryManager().CpuToGpuAddress(overlap_start);
-
-        if (gpu_addresses.empty()) {
-            return;
-        }
-
         const u64 overlap_size = overlap_end - overlap_start;
 
-        for (const auto& gpu_address : gpu_addresses) {
-            auto& rasterizer = system_instance.Renderer().Rasterizer();
-            switch (mode) {
-            case FlushMode::Flush:
-                rasterizer.FlushRegion(gpu_address, overlap_size);
-                break;
-            case FlushMode::Invalidate:
-                rasterizer.InvalidateRegion(gpu_address, overlap_size);
-                break;
-            case FlushMode::FlushAndInvalidate:
-                rasterizer.FlushAndInvalidateRegion(gpu_address, overlap_size);
-                break;
-            }
+        auto& rasterizer = system_instance.Renderer().Rasterizer();
+        switch (mode) {
+        case FlushMode::Flush:
+            rasterizer.FlushRegion(overlap_start, overlap_size);
+            break;
+        case FlushMode::Invalidate:
+            rasterizer.InvalidateRegion(overlap_start, overlap_size);
+            break;
+        case FlushMode::FlushAndInvalidate:
+            rasterizer.FlushAndInvalidateRegion(overlap_start, overlap_size);
+            break;
         }
     };
 

src/core/memory.h

@@ -11,7 +11,6 @@
 #include <boost/icl/interval_map.hpp>
 #include "common/common_types.h"
 #include "core/memory_hook.h"
-#include "video_core/memory_manager.h"
 
 namespace Kernel {
 class Process;
@@ -179,7 +178,7 @@ enum class FlushMode {
 /**
  * Mark each page touching the region as cached.
  */
-void RasterizerMarkRegionCached(Tegra::GPUVAddr gpu_addr, u64 size, bool cached);
+void RasterizerMarkRegionCached(VAddr vaddr, u64 size, bool cached);
 
 /**
  * Flushes and invalidates any externally cached rasterizer resources touching the given virtual

src/core/perf_stats.cpp

@@ -40,7 +40,7 @@ void PerfStats::EndGameFrame() {
     game_frames += 1;
 }
 
-PerfStats::Results PerfStats::GetAndResetStats(microseconds current_system_time_us) {
+PerfStatsResults PerfStats::GetAndResetStats(microseconds current_system_time_us) {
     std::lock_guard<std::mutex> lock(object_mutex);
 
     const auto now = Clock::now();
@@ -49,7 +49,7 @@ PerfStats::Results PerfStats::GetAndResetStats(microseconds current_system_time_
 
     const auto system_us_per_second = (current_system_time_us - reset_point_system_us) / interval;
 
-    Results results{};
+    PerfStatsResults results{};
     results.system_fps = static_cast<double>(system_frames) / interval;
     results.game_fps = static_cast<double>(game_frames) / interval;
     results.frametime = duration_cast<DoubleSecs>(accumulated_frametime).count() /

src/core/perf_stats.h

@@ -10,6 +10,17 @@
 
 namespace Core {
 
+struct PerfStatsResults {
+    /// System FPS (LCD VBlanks) in Hz
+    double system_fps;
+    /// Game FPS (GSP frame submissions) in Hz
+    double game_fps;
+    /// Walltime per system frame, in seconds, excluding any waits
+    double frametime;
+    /// Ratio of walltime / emulated time elapsed
+    double emulation_speed;
+};
+
 /**
  * Class to manage and query performance/timing statistics. All public functions of this class are
  * thread-safe unless stated otherwise.
@@ -18,22 +29,11 @@ class PerfStats {
 public:
     using Clock = std::chrono::high_resolution_clock;
 
-    struct Results {
-        /// System FPS (LCD VBlanks) in Hz
-        double system_fps;
-        /// Game FPS (GSP frame submissions) in Hz
-        double game_fps;
-        /// Walltime per system frame, in seconds, excluding any waits
-        double frametime;
-        /// Ratio of walltime / emulated time elapsed
-        double emulation_speed;
-    };
-
     void BeginSystemFrame();
     void EndSystemFrame();
     void EndGameFrame();
 
-    Results GetAndResetStats(std::chrono::microseconds current_system_time_us);
+    PerfStatsResults GetAndResetStats(std::chrono::microseconds current_system_time_us);
 
     /**
      * Gets the ratio between walltime and the emulated time of the previous system frame. This is

src/core/telemetry_session.cpp

@@ -7,6 +7,7 @@
 #include "common/file_util.h"
 
 #include "core/core.h"
+#include "core/loader/loader.h"
 #include "core/settings.h"
 #include "core/telemetry_session.h"