aoc_u: Stub GetAddOnContentListChangedEvent

This event signals the game when new DLC is purchased from the eShop while the game is running. Since, for the forseeable future, yuzu will not have this ability, it seems safe to stub with a dummy event that will never fire. This is needed to boot Sonic Mania Plus (update v1.04).

.travis/clang-format/script.sh

@@ -1,6 +1,6 @@
 #!/bin/bash -ex
 
-if grep -nr '\s$' src *.yml *.txt *.md Doxyfile .gitignore .gitmodules .travis* dist/*.desktop \
+if grep -nrI '\s$' src *.yml *.txt *.md Doxyfile .gitignore .gitmodules .travis* dist/*.desktop \
                  dist/*.svg dist/*.xml; then
     echo Trailing whitespace found, aborting
     exit 1

src/common/alignment.h

@@ -19,4 +19,16 @@ constexpr T AlignDown(T value, std::size_t size) {
     return static_cast<T>(value - value % size);
 }
 
+template <typename T>
+constexpr bool Is4KBAligned(T value) {
+    static_assert(std::is_unsigned_v<T>, "T must be an unsigned value.");
+    return (value & 0xFFF) == 0;
+}
+
+template <typename T>
+constexpr bool IsWordAligned(T value) {
+    static_assert(std::is_unsigned_v<T>, "T must be an unsigned value.");
+    return (value & 0b11) == 0;
+}
+
 } // namespace Common

src/common/web_result.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <string>
+#include "common/common_types.h"
 
 namespace Common {
 struct WebResult {

src/core/CMakeLists.txt

@@ -236,6 +236,24 @@ add_library(core STATIC
     hle/service/hid/irs.h
     hle/service/hid/xcd.cpp
     hle/service/hid/xcd.h
+    hle/service/hid/controllers/controller_base.cpp
+    hle/service/hid/controllers/controller_base.h
+    hle/service/hid/controllers/debug_pad.cpp
+    hle/service/hid/controllers/debug_pad.h
+    hle/service/hid/controllers/gesture.cpp
+    hle/service/hid/controllers/gesture.h
+    hle/service/hid/controllers/keyboard.cpp
+    hle/service/hid/controllers/keyboard.h
+    hle/service/hid/controllers/mouse.cpp
+    hle/service/hid/controllers/mouse.h
+    hle/service/hid/controllers/npad.cpp
+    hle/service/hid/controllers/npad.h
+    hle/service/hid/controllers/stubbed.cpp
+    hle/service/hid/controllers/stubbed.h
+    hle/service/hid/controllers/touchscreen.cpp
+    hle/service/hid/controllers/touchscreen.h
+    hle/service/hid/controllers/xpad.cpp
+    hle/service/hid/controllers/xpad.h
     hle/service/lbl/lbl.cpp
     hle/service/lbl/lbl.h
     hle/service/ldn/ldn.cpp
@@ -400,8 +418,8 @@ create_target_directory_groups(core)
 target_link_libraries(core PUBLIC common PRIVATE audio_core video_core)
 target_link_libraries(core PUBLIC Boost::boost PRIVATE fmt lz4_static mbedtls opus unicorn open_source_archives)
 if (ENABLE_WEB_SERVICE)
-    add_definitions(-DENABLE_WEB_SERVICE)
-    target_link_libraries(core PUBLIC json-headers web_service)
+    target_compile_definitions(core PRIVATE -DENABLE_WEB_SERVICE)
+    target_link_libraries(core PRIVATE web_service)
 endif()
 
 if (ARCHITECTURE_x86_64)

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -144,7 +144,7 @@ std::unique_ptr<Dynarmic::A64::Jit> ARM_Dynarmic::MakeJit() const {
 
     // Multi-process state
     config.processor_id = core_index;
-    config.global_monitor = &exclusive_monitor->monitor;
+    config.global_monitor = &exclusive_monitor.monitor;
 
     // System registers
     config.tpidrro_el0 = &cb->tpidrro_el0;
@@ -171,10 +171,9 @@ void ARM_Dynarmic::Step() {
     cb->InterpreterFallback(jit->GetPC(), 1);
 }
 
-ARM_Dynarmic::ARM_Dynarmic(std::shared_ptr<ExclusiveMonitor> exclusive_monitor,
-                           std::size_t core_index)
+ARM_Dynarmic::ARM_Dynarmic(ExclusiveMonitor& exclusive_monitor, std::size_t core_index)
     : cb(std::make_unique<ARM_Dynarmic_Callbacks>(*this)), core_index{core_index},
-      exclusive_monitor{std::dynamic_pointer_cast<DynarmicExclusiveMonitor>(exclusive_monitor)} {
+      exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {
     ThreadContext ctx{};
     inner_unicorn.SaveContext(ctx);
     PageTableChanged();

src/core/arm/dynarmic/arm_dynarmic.h

@@ -23,7 +23,7 @@ class DynarmicExclusiveMonitor;
 
 class ARM_Dynarmic final : public ARM_Interface {
 public:
-    ARM_Dynarmic(std::shared_ptr<ExclusiveMonitor> exclusive_monitor, std::size_t core_index);
+    ARM_Dynarmic(ExclusiveMonitor& exclusive_monitor, std::size_t core_index);
     ~ARM_Dynarmic();
 
     void MapBackingMemory(VAddr address, std::size_t size, u8* memory,
@@ -62,7 +62,7 @@ private:
     ARM_Unicorn inner_unicorn;
 
     std::size_t core_index;
-    std::shared_ptr<DynarmicExclusiveMonitor> exclusive_monitor;
+    DynarmicExclusiveMonitor& exclusive_monitor;
 
     Memory::PageTable* current_page_table = nullptr;
 };

src/core/core.cpp

@@ -71,9 +71,9 @@ FileSys::VirtualFile GetGameFileFromPath(const FileSys::VirtualFilesystem& vfs,
 }
 
 /// Runs a CPU core while the system is powered on
-void RunCpuCore(std::shared_ptr<Cpu> cpu_state) {
+void RunCpuCore(Cpu& cpu_state) {
     while (Core::System::GetInstance().IsPoweredOn()) {
-        cpu_state->RunLoop(true);
+        cpu_state.RunLoop(true);
     }
 }
 } // Anonymous namespace
@@ -95,7 +95,7 @@ struct System::Impl {
         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];
+        thread_to_cpu[std::this_thread::get_id()] = cpu_cores[0].get();
 
         if (GDBStub::IsServerEnabled()) {
             GDBStub::HandlePacket();
@@ -139,16 +139,16 @@ struct System::Impl {
         auto main_process = Kernel::Process::Create(kernel, "main");
         kernel.MakeCurrentProcess(main_process.get());
 
-        cpu_barrier = std::make_shared<CpuBarrier>();
+        cpu_barrier = std::make_unique<CpuBarrier>();
         cpu_exclusive_monitor = Cpu::MakeExclusiveMonitor(cpu_cores.size());
         for (std::size_t index = 0; index < cpu_cores.size(); ++index) {
-            cpu_cores[index] = std::make_shared<Cpu>(cpu_exclusive_monitor, cpu_barrier, index);
+            cpu_cores[index] = std::make_unique<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);
+        Service::Init(service_manager, *virtual_filesystem);
         GDBStub::Init();
 
         renderer = VideoCore::CreateRenderer(emu_window);
@@ -160,12 +160,12 @@ struct System::Impl {
 
         // 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];
+        thread_to_cpu[std::this_thread::get_id()] = cpu_cores[0].get();
         if (Settings::values.use_multi_core) {
             for (std::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];
+                    std::make_unique<std::thread>(RunCpuCore, std::ref(*cpu_cores[index + 1]));
+                thread_to_cpu[cpu_core_threads[index]->get_id()] = cpu_cores[index + 1].get();
             }
         }
 
@@ -245,6 +245,7 @@ struct System::Impl {
         for (auto& cpu_core : cpu_cores) {
             cpu_core.reset();
         }
+        cpu_exclusive_monitor.reset();
         cpu_barrier.reset();
 
         // Shutdown kernel and core timing
@@ -282,9 +283,9 @@ struct System::Impl {
     std::unique_ptr<VideoCore::RendererBase> renderer;
     std::unique_ptr<Tegra::GPU> gpu_core;
     std::shared_ptr<Tegra::DebugContext> debug_context;
-    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::unique_ptr<ExclusiveMonitor> cpu_exclusive_monitor;
+    std::unique_ptr<CpuBarrier> cpu_barrier;
+    std::array<std::unique_ptr<Cpu>, NUM_CPU_CORES> cpu_cores;
     std::array<std::unique_ptr<std::thread>, NUM_CPU_CORES - 1> cpu_core_threads;
     std::size_t active_core{}; ///< Active core, only used in single thread mode
 
@@ -298,7 +299,7 @@ struct System::Impl {
     std::string status_details = "";
 
     /// Map of guest threads to CPU cores
-    std::map<std::thread::id, std::shared_ptr<Cpu>> thread_to_cpu;
+    std::map<std::thread::id, Cpu*> thread_to_cpu;
 
     Core::PerfStats perf_stats;
     Core::FrameLimiter frame_limiter;
@@ -354,12 +355,15 @@ std::size_t System::CurrentCoreIndex() {
 }
 
 Kernel::Scheduler& System::CurrentScheduler() {
-    return *CurrentCpuCore().Scheduler();
+    return CurrentCpuCore().Scheduler();
 }
 
-const std::shared_ptr<Kernel::Scheduler>& System::Scheduler(std::size_t core_index) {
-    ASSERT(core_index < NUM_CPU_CORES);
-    return impl->cpu_cores[core_index]->Scheduler();
+Kernel::Scheduler& System::Scheduler(std::size_t core_index) {
+    return CpuCore(core_index).Scheduler();
+}
+
+const Kernel::Scheduler& System::Scheduler(std::size_t core_index) const {
+    return CpuCore(core_index).Scheduler();
 }
 
 Kernel::Process* System::CurrentProcess() {
@@ -371,8 +375,7 @@ const Kernel::Process* System::CurrentProcess() const {
 }
 
 ARM_Interface& System::ArmInterface(std::size_t core_index) {
-    ASSERT(core_index < NUM_CPU_CORES);
-    return impl->cpu_cores[core_index]->ArmInterface();
+    return CpuCore(core_index).ArmInterface();
 }
 
 Cpu& System::CpuCore(std::size_t core_index) {
@@ -380,6 +383,11 @@ Cpu& System::CpuCore(std::size_t core_index) {
     return *impl->cpu_cores[core_index];
 }
 
+const Cpu& System::CpuCore(std::size_t core_index) const {
+    ASSERT(core_index < NUM_CPU_CORES);
+    return *impl->cpu_cores[core_index];
+}
+
 ExclusiveMonitor& System::Monitor() {
     return *impl->cpu_exclusive_monitor;
 }

src/core/core.h

@@ -156,6 +156,9 @@ public:
     /// Gets a CPU interface to the CPU core with the specified index
     Cpu& CpuCore(std::size_t core_index);
 
+    /// Gets a CPU interface to the CPU core with the specified index
+    const Cpu& CpuCore(std::size_t core_index) const;
+
     /// Gets the exclusive monitor
     ExclusiveMonitor& Monitor();
 
@@ -172,7 +175,10 @@ public:
     const VideoCore::RendererBase& Renderer() const;
 
     /// Gets the scheduler for the CPU core with the specified index
-    const std::shared_ptr<Kernel::Scheduler>& Scheduler(std::size_t core_index);
+    Kernel::Scheduler& Scheduler(std::size_t core_index);
+
+    /// Gets the scheduler for the CPU core with the specified index
+    const Kernel::Scheduler& Scheduler(std::size_t core_index) const;
 
     /// Provides a pointer to the current process
     Kernel::Process* CurrentProcess();

src/core/core_cpu.cpp

@@ -49,10 +49,8 @@ bool CpuBarrier::Rendezvous() {
     return false;
 }
 
-Cpu::Cpu(std::shared_ptr<ExclusiveMonitor> exclusive_monitor,
-         std::shared_ptr<CpuBarrier> cpu_barrier, std::size_t core_index)
-    : cpu_barrier{std::move(cpu_barrier)}, core_index{core_index} {
-
+Cpu::Cpu(ExclusiveMonitor& exclusive_monitor, CpuBarrier& cpu_barrier, std::size_t core_index)
+    : cpu_barrier{cpu_barrier}, core_index{core_index} {
     if (Settings::values.use_cpu_jit) {
 #ifdef ARCHITECTURE_x86_64
         arm_interface = std::make_unique<ARM_Dynarmic>(exclusive_monitor, core_index);
@@ -64,15 +62,15 @@ Cpu::Cpu(std::shared_ptr<ExclusiveMonitor> exclusive_monitor,
         arm_interface = std::make_unique<ARM_Unicorn>();
     }
 
-    scheduler = std::make_shared<Kernel::Scheduler>(*arm_interface);
+    scheduler = std::make_unique<Kernel::Scheduler>(*arm_interface);
 }
 
 Cpu::~Cpu() = default;
 
-std::shared_ptr<ExclusiveMonitor> Cpu::MakeExclusiveMonitor(std::size_t num_cores) {
+std::unique_ptr<ExclusiveMonitor> Cpu::MakeExclusiveMonitor(std::size_t num_cores) {
     if (Settings::values.use_cpu_jit) {
 #ifdef ARCHITECTURE_x86_64
-        return std::make_shared<DynarmicExclusiveMonitor>(num_cores);
+        return std::make_unique<DynarmicExclusiveMonitor>(num_cores);
 #else
         return nullptr; // TODO(merry): Passthrough exclusive monitor
 #endif
@@ -83,7 +81,7 @@ std::shared_ptr<ExclusiveMonitor> Cpu::MakeExclusiveMonitor(std::size_t num_core
 
 void Cpu::RunLoop(bool tight_loop) {
     // Wait for all other CPU cores to complete the previous slice, such that they run in lock-step
-    if (!cpu_barrier->Rendezvous()) {
+    if (!cpu_barrier.Rendezvous()) {
         // If rendezvous failed, session has been killed
         return;
     }

src/core/core_cpu.h

@@ -41,8 +41,7 @@ private:
 
 class Cpu {
 public:
-    Cpu(std::shared_ptr<ExclusiveMonitor> exclusive_monitor,
-        std::shared_ptr<CpuBarrier> cpu_barrier, std::size_t core_index);
+    Cpu(ExclusiveMonitor& exclusive_monitor, CpuBarrier& cpu_barrier, std::size_t core_index);
     ~Cpu();
 
     void RunLoop(bool tight_loop = true);
@@ -59,8 +58,12 @@ public:
         return *arm_interface;
     }
 
-    const std::shared_ptr<Kernel::Scheduler>& Scheduler() const {
-        return scheduler;
+    Kernel::Scheduler& Scheduler() {
+        return *scheduler;
+    }
+
+    const Kernel::Scheduler& Scheduler() const {
+        return *scheduler;
     }
 
     bool IsMainCore() const {
@@ -71,14 +74,14 @@ public:
         return core_index;
     }
 
-    static std::shared_ptr<ExclusiveMonitor> MakeExclusiveMonitor(std::size_t num_cores);
+    static std::unique_ptr<ExclusiveMonitor> MakeExclusiveMonitor(std::size_t num_cores);
 
 private:
     void Reschedule();
 
     std::unique_ptr<ARM_Interface> arm_interface;
-    std::shared_ptr<CpuBarrier> cpu_barrier;
-    std::shared_ptr<Kernel::Scheduler> scheduler;
+    CpuBarrier& cpu_barrier;
+    std::unique_ptr<Kernel::Scheduler> scheduler;
 
     std::atomic<bool> reschedule_pending = false;
     std::size_t core_index;

src/core/crypto/key_manager.cpp

@@ -98,7 +98,7 @@ std::array<u8, 144> DecryptKeyblob(const std::array<u8, 176>& encrypted_keyblob,
     return keyblob;
 }
 
-void KeyManager::DeriveGeneralPurposeKeys(u8 crypto_revision) {
+void KeyManager::DeriveGeneralPurposeKeys(std::size_t crypto_revision) {
     const auto kek_generation_source =
         GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKekGeneration));
     const auto key_generation_source =
@@ -147,31 +147,38 @@ boost::optional<Key128> DeriveSDSeed() {
                                    "rb+");
     if (!save_43.IsOpen())
         return boost::none;
+
     const FileUtil::IOFile sd_private(
         FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir) + "/Nintendo/Contents/private", "rb+");
     if (!sd_private.IsOpen())
         return boost::none;
 
-    sd_private.Seek(0, SEEK_SET);
     std::array<u8, 0x10> private_seed{};
-    if (sd_private.ReadBytes(private_seed.data(), private_seed.size()) != 0x10)
+    if (sd_private.ReadBytes(private_seed.data(), private_seed.size()) != private_seed.size()) {
         return boost::none;
+    }
 
     std::array<u8, 0x10> buffer{};
     std::size_t offset = 0;
     for (; offset + 0x10 < save_43.GetSize(); ++offset) {
-        save_43.Seek(offset, SEEK_SET);
+        if (!save_43.Seek(offset, SEEK_SET)) {
+            return boost::none;
+        }
+
         save_43.ReadBytes(buffer.data(), buffer.size());
-        if (buffer == private_seed)
+        if (buffer == private_seed) {
             break;
+        }
     }
 
-    if (offset + 0x10 >= save_43.GetSize())
+    if (!save_43.Seek(offset + 0x10, SEEK_SET)) {
         return boost::none;
+    }
 
     Key128 seed{};
-    save_43.Seek(offset + 0x10, SEEK_SET);
-    save_43.ReadBytes(seed.data(), seed.size());
+    if (save_43.ReadBytes(seed.data(), seed.size()) != seed.size()) {
+        return boost::none;
+    }
     return seed;
 }
 
@@ -234,7 +241,9 @@ std::vector<TicketRaw> GetTicketblob(const FileUtil::IOFile& ticket_save) {
         return {};
 
     std::vector<u8> buffer(ticket_save.GetSize());
-    ticket_save.ReadBytes(buffer.data(), buffer.size());
+    if (ticket_save.ReadBytes(buffer.data(), buffer.size()) != buffer.size()) {
+        return {};
+    }
 
     std::vector<TicketRaw> out;
     u32 magic{};
@@ -261,6 +270,9 @@ static std::array<u8, size> operator^(const std::array<u8, size>& lhs,
 
 template <size_t target_size, size_t in_size>
 static std::array<u8, target_size> MGF1(const std::array<u8, in_size>& seed) {
+    // Avoids truncation overflow within the loop below.
+    static_assert(target_size <= 0xFF);
+
     std::array<u8, in_size + 4> seed_exp{};
     std::memcpy(seed_exp.data(), seed.data(), in_size);
 
@@ -268,7 +280,7 @@ static std::array<u8, target_size> MGF1(const std::array<u8, in_size>& seed) {
     size_t i = 0;
     while (out.size() < target_size) {
         out.resize(out.size() + 0x20);
-        seed_exp[in_size + 3] = i;
+        seed_exp[in_size + 3] = static_cast<u8>(i);
         mbedtls_sha256(seed_exp.data(), seed_exp.size(), out.data() + out.size() - 0x20, 0);
         ++i;
     }
@@ -299,10 +311,11 @@ boost::optional<std::pair<Key128, Key128>> ParseTicket(const TicketRaw& ticket,
     std::memcpy(&cert_authority, ticket.data() + 0x140, sizeof(cert_authority));
     if (cert_authority == 0)
         return boost::none;
-    if (cert_authority != Common::MakeMagic('R', 'o', 'o', 't'))
+    if (cert_authority != Common::MakeMagic('R', 'o', 'o', 't')) {
         LOG_INFO(Crypto,
                  "Attempting to parse ticket with non-standard certificate authority {:08X}.",
                  cert_authority);
+    }
 
     Key128 rights_id;
     std::memcpy(rights_id.data(), ticket.data() + 0x2A0, sizeof(Key128));
@@ -871,9 +884,9 @@ void KeyManager::DeriveETicket(PartitionDataManager& data) {
                                      "/system/save/80000000000000e2",
                                  "rb+");
 
+    const auto blob2 = GetTicketblob(save2);
     auto res = GetTicketblob(save1);
-    const auto res2 = GetTicketblob(save2);
-    std::copy(res2.begin(), res2.end(), std::back_inserter(res));
+    res.insert(res.end(), blob2.begin(), blob2.end());
 
     for (const auto& raw : res) {
         const auto pair = ParseTicket(raw, rsa_key);

src/core/crypto/key_manager.h

@@ -175,7 +175,7 @@ private:
     void WriteKeyToFile(KeyCategory category, std::string_view keyname,
                         const std::array<u8, Size>& key);
 
-    void DeriveGeneralPurposeKeys(u8 crypto_revision);
+    void DeriveGeneralPurposeKeys(std::size_t crypto_revision);
 
     void SetKeyWrapped(S128KeyType id, Key128 key, u64 field1 = 0, u64 field2 = 0);
     void SetKeyWrapped(S256KeyType id, Key256 key, u64 field1 = 0, u64 field2 = 0);

src/core/crypto/partition_data_manager.cpp

@@ -11,7 +11,6 @@
 #include <array>
 #include <cctype>
 #include <cstring>
-#include <boost/optional/optional.hpp>
 #include <mbedtls/sha256.h>
 #include "common/assert.h"
 #include "common/common_funcs.h"
@@ -19,7 +18,7 @@
 #include "common/hex_util.h"
 #include "common/logging/log.h"
 #include "common/string_util.h"
-#include "core/crypto/ctr_encryption_layer.h"
+#include "common/swap.h"
 #include "core/crypto/key_manager.h"
 #include "core/crypto/partition_data_manager.h"
 #include "core/crypto/xts_encryption_layer.h"
@@ -302,10 +301,10 @@ FileSys::VirtualFile FindFileInDirWithNames(const FileSys::VirtualDir& dir,
     return nullptr;
 }
 
-PartitionDataManager::PartitionDataManager(FileSys::VirtualDir sysdata_dir)
+PartitionDataManager::PartitionDataManager(const FileSys::VirtualDir& sysdata_dir)
     : boot0(FindFileInDirWithNames(sysdata_dir, "BOOT0")),
-      fuses(FindFileInDirWithNames(sysdata_dir, "fuse")),
-      kfuses(FindFileInDirWithNames(sysdata_dir, "kfuse")),
+      fuses(FindFileInDirWithNames(sysdata_dir, "fuses")),
+      kfuses(FindFileInDirWithNames(sysdata_dir, "kfuses")),
       package2({
           FindFileInDirWithNames(sysdata_dir, "BCPKG2-1-Normal-Main"),
           FindFileInDirWithNames(sysdata_dir, "BCPKG2-2-Normal-Sub"),
@@ -314,13 +313,14 @@ PartitionDataManager::PartitionDataManager(FileSys::VirtualDir sysdata_dir)
           FindFileInDirWithNames(sysdata_dir, "BCPKG2-5-Repair-Main"),
           FindFileInDirWithNames(sysdata_dir, "BCPKG2-6-Repair-Sub"),
       }),
+      prodinfo(FindFileInDirWithNames(sysdata_dir, "PRODINFO")),
       secure_monitor(FindFileInDirWithNames(sysdata_dir, "secmon")),
       package1_decrypted(FindFileInDirWithNames(sysdata_dir, "pkg1_decr")),
       secure_monitor_bytes(secure_monitor == nullptr ? std::vector<u8>{}
                                                      : secure_monitor->ReadAllBytes()),
       package1_decrypted_bytes(package1_decrypted == nullptr ? std::vector<u8>{}
-                                                             : package1_decrypted->ReadAllBytes()),
-      prodinfo(FindFileInDirWithNames(sysdata_dir, "PRODINFO")) {}
+                                                             : package1_decrypted->ReadAllBytes()) {
+}
 
 PartitionDataManager::~PartitionDataManager() = default;
 
@@ -332,18 +332,19 @@ FileSys::VirtualFile PartitionDataManager::GetBoot0Raw() const {
     return boot0;
 }
 
-std::array<u8, 176> PartitionDataManager::GetEncryptedKeyblob(u8 index) const {
-    if (HasBoot0() && index < 32)
+PartitionDataManager::EncryptedKeyBlob PartitionDataManager::GetEncryptedKeyblob(
+    std::size_t index) const {
+    if (HasBoot0() && index < NUM_ENCRYPTED_KEYBLOBS)
         return GetEncryptedKeyblobs()[index];
     return {};
 }
 
-std::array<std::array<u8, 176>, 32> PartitionDataManager::GetEncryptedKeyblobs() const {
+PartitionDataManager::EncryptedKeyBlobs PartitionDataManager::GetEncryptedKeyblobs() const {
     if (!HasBoot0())
         return {};
 
-    std::array<std::array<u8, 176>, 32> out{};
-    for (size_t i = 0; i < 0x20; ++i)
+    EncryptedKeyBlobs out{};
+    for (size_t i = 0; i < out.size(); ++i)
         boot0->Read(out[i].data(), out[i].size(), 0x180000 + i * 0x200);
     return out;
 }
@@ -389,7 +390,7 @@ std::array<u8, 16> PartitionDataManager::GetKeyblobMACKeySource() const {
     return FindKeyFromHex(package1_decrypted_bytes, source_hashes[0]);
 }
 
-std::array<u8, 16> PartitionDataManager::GetKeyblobKeySource(u8 revision) const {
+std::array<u8, 16> PartitionDataManager::GetKeyblobKeySource(std::size_t revision) const {
     if (keyblob_source_hashes[revision] == SHA256Hash{}) {
         LOG_WARNING(Crypto,
                     "No keyblob source hash for crypto revision {:02X}! Cannot derive keys...",
@@ -446,7 +447,7 @@ bool AttemptDecrypt(const std::array<u8, 16>& key, Package2Header& header) {
     return false;
 }
 
-void PartitionDataManager::DecryptPackage2(std::array<std::array<u8, 16>, 0x20> package2_keys,
+void PartitionDataManager::DecryptPackage2(const std::array<Key128, 0x20>& package2_keys,
                                            Package2Type type) {
     FileSys::VirtualFile file = std::make_shared<FileSys::OffsetVfsFile>(
         package2[static_cast<size_t>(type)],
@@ -456,43 +457,38 @@ void PartitionDataManager::DecryptPackage2(std::array<std::array<u8, 16>, 0x20>
     if (file->ReadObject(&header) != sizeof(Package2Header))
         return;
 
-    u8 revision = 0xFF;
+    std::size_t revision = 0xFF;
     if (header.magic != Common::MakeMagic('P', 'K', '2', '1')) {
-        for (size_t i = 0; i < package2_keys.size(); ++i) {
-            if (AttemptDecrypt(package2_keys[i], header))
+        for (std::size_t i = 0; i < package2_keys.size(); ++i) {
+            if (AttemptDecrypt(package2_keys[i], header)) {
                 revision = i;
+            }
         }
     }
 
     if (header.magic != Common::MakeMagic('P', 'K', '2', '1'))
         return;
 
-    const std::vector<u8> s1_iv(header.section_ctr[1].begin(), header.section_ctr[1].end());
-
     const auto a = std::make_shared<FileSys::OffsetVfsFile>(
         file, header.section_size[1], header.section_size[0] + sizeof(Package2Header));
 
     auto c = a->ReadAllBytes();
 
     AESCipher<Key128> cipher(package2_keys[revision], Mode::CTR);
-    cipher.SetIV(s1_iv);
+    cipher.SetIV({header.section_ctr[1].begin(), header.section_ctr[1].end()});
     cipher.Transcode(c.data(), c.size(), c.data(), Op::Decrypt);
 
-    // package2_decrypted[static_cast<size_t>(type)] = s1;
-
     INIHeader ini;
     std::memcpy(&ini, c.data(), sizeof(INIHeader));
     if (ini.magic != Common::MakeMagic('I', 'N', 'I', '1'))
         return;
 
-    std::map<u64, KIPHeader> kips{};
     u64 offset = sizeof(INIHeader);
     for (size_t i = 0; i < ini.process_count; ++i) {
         KIPHeader kip;
         std::memcpy(&kip, c.data() + offset, sizeof(KIPHeader));
         if (kip.magic != Common::MakeMagic('K', 'I', 'P', '1'))
             return;
-        kips.emplace(offset, kip);
 
         const auto name =
             Common::StringFromFixedZeroTerminatedBuffer(kip.name.data(), kip.name.size());
@@ -503,33 +499,29 @@ void PartitionDataManager::DecryptPackage2(std::array<std::array<u8, 16>, 0x20>
             continue;
         }
 
-        std::vector<u8> text(kip.sections[0].size_compressed);
-        std::vector<u8> rodata(kip.sections[1].size_compressed);
-        std::vector<u8> data(kip.sections[2].size_compressed);
+        const u64 initial_offset = sizeof(KIPHeader) + offset;
+        const auto text_begin = c.cbegin() + initial_offset;
+        const auto text_end = text_begin + kip.sections[0].size_compressed;
+        const std::vector<u8> text = DecompressBLZ({text_begin, text_end});
 
-        u64 offset_sec = sizeof(KIPHeader) + offset;
-        std::memcpy(text.data(), c.data() + offset_sec, text.size());
-        offset_sec += text.size();
-        std::memcpy(rodata.data(), c.data() + offset_sec, rodata.size());
-        offset_sec += rodata.size();
-        std::memcpy(data.data(), c.data() + offset_sec, data.size());
+        const auto rodata_end = text_end + kip.sections[1].size_compressed;
+        const std::vector<u8> rodata = DecompressBLZ({text_end, rodata_end});
 
-        offset += sizeof(KIPHeader) + kip.sections[0].size_compressed +
-                  kip.sections[1].size_compressed + kip.sections[2].size_compressed;
+        const auto data_end = rodata_end + kip.sections[2].size_compressed;
+        const std::vector<u8> data = DecompressBLZ({rodata_end, data_end});
 
-        text = DecompressBLZ(text);
-        rodata = DecompressBLZ(rodata);
-        data = DecompressBLZ(data);
+        std::vector<u8> out;
+        out.reserve(text.size() + rodata.size() + data.size());
+        out.insert(out.end(), text.begin(), text.end());
+        out.insert(out.end(), rodata.begin(), rodata.end());
+        out.insert(out.end(), data.begin(), data.end());
 
-        std::vector<u8> out(text.size() + rodata.size() + data.size());
-        std::memcpy(out.data(), text.data(), text.size());
-        std::memcpy(out.data() + text.size(), rodata.data(), rodata.size());
-        std::memcpy(out.data() + text.size() + rodata.size(), data.data(), data.size());
+        offset += sizeof(KIPHeader) + out.size();
 
         if (name == "FS")
-            package2_fs[static_cast<size_t>(type)] = out;
+            package2_fs[static_cast<size_t>(type)] = std::move(out);
         else if (name == "spl")
-            package2_spl[static_cast<size_t>(type)] = out;
+            package2_spl[static_cast<size_t>(type)] = std::move(out);
     }
 }
 

src/core/crypto/partition_data_manager.h

@@ -5,9 +5,7 @@
 #pragma once
 
 #include <vector>
-#include "common/common_funcs.h"
 #include "common/common_types.h"
-#include "common/swap.h"
 #include "core/file_sys/vfs_types.h"
 
 namespace Core::Crypto {
@@ -24,15 +22,20 @@ enum class Package2Type {
 class PartitionDataManager {
 public:
     static const u8 MAX_KEYBLOB_SOURCE_HASH;
+    static constexpr std::size_t NUM_ENCRYPTED_KEYBLOBS = 32;
+    static constexpr std::size_t ENCRYPTED_KEYBLOB_SIZE = 0xB0;
 
-    explicit PartitionDataManager(FileSys::VirtualDir sysdata_dir);
+    using EncryptedKeyBlob = std::array<u8, ENCRYPTED_KEYBLOB_SIZE>;
+    using EncryptedKeyBlobs = std::array<EncryptedKeyBlob, NUM_ENCRYPTED_KEYBLOBS>;
+
+    explicit PartitionDataManager(const FileSys::VirtualDir& sysdata_dir);
     ~PartitionDataManager();
 
     // BOOT0
     bool HasBoot0() const;
     FileSys::VirtualFile GetBoot0Raw() const;
-    std::array<u8, 0xB0> GetEncryptedKeyblob(u8 index) const;
-    std::array<std::array<u8, 0xB0>, 0x20> GetEncryptedKeyblobs() const;
+    EncryptedKeyBlob GetEncryptedKeyblob(std::size_t index) const;
+    EncryptedKeyBlobs GetEncryptedKeyblobs() const;
     std::vector<u8> GetSecureMonitor() const;
     std::array<u8, 0x10> GetPackage2KeySource() const;
     std::array<u8, 0x10> GetAESKekGenerationSource() const;
@@ -43,7 +46,7 @@ public:
     std::vector<u8> GetPackage1Decrypted() const;
     std::array<u8, 0x10> GetMasterKeySource() const;
     std::array<u8, 0x10> GetKeyblobMACKeySource() const;
-    std::array<u8, 0x10> GetKeyblobKeySource(u8 revision) const;
+    std::array<u8, 0x10> GetKeyblobKeySource(std::size_t revision) const;
 
     // Fuses
     bool HasFuses() const;
@@ -57,7 +60,8 @@ public:
     // Package2
     bool HasPackage2(Package2Type type = Package2Type::NormalMain) const;
     FileSys::VirtualFile GetPackage2Raw(Package2Type type = Package2Type::NormalMain) const;
-    void DecryptPackage2(std::array<std::array<u8, 16>, 0x20> package2, Package2Type type);
+    void DecryptPackage2(const std::array<std::array<u8, 16>, 0x20>& package2_keys,
+                         Package2Type type);
     const std::vector<u8>& GetPackage2FSDecompressed(
         Package2Type type = Package2Type::NormalMain) const;
     std::array<u8, 0x10> GetKeyAreaKeyApplicationSource(

src/core/file_sys/bis_factory.cpp

@@ -10,19 +10,19 @@ namespace FileSys {
 
 BISFactory::BISFactory(VirtualDir nand_root_, VirtualDir load_root_)
     : nand_root(std::move(nand_root_)), load_root(std::move(load_root_)),
-      sysnand_cache(std::make_shared<RegisteredCache>(
+      sysnand_cache(std::make_unique<RegisteredCache>(
           GetOrCreateDirectoryRelative(nand_root, "/system/Contents/registered"))),
-      usrnand_cache(std::make_shared<RegisteredCache>(
+      usrnand_cache(std::make_unique<RegisteredCache>(
           GetOrCreateDirectoryRelative(nand_root, "/user/Contents/registered"))) {}
 
 BISFactory::~BISFactory() = default;
 
-std::shared_ptr<RegisteredCache> BISFactory::GetSystemNANDContents() const {
-    return sysnand_cache;
+RegisteredCache* BISFactory::GetSystemNANDContents() const {
+    return sysnand_cache.get();
 }
 
-std::shared_ptr<RegisteredCache> BISFactory::GetUserNANDContents() const {
-    return usrnand_cache;
+RegisteredCache* BISFactory::GetUserNANDContents() const {
+    return usrnand_cache.get();
 }
 
 VirtualDir BISFactory::GetModificationLoadRoot(u64 title_id) const {

src/core/file_sys/bis_factory.h

@@ -20,8 +20,8 @@ public:
     explicit BISFactory(VirtualDir nand_root, VirtualDir load_root);
     ~BISFactory();
 
-    std::shared_ptr<RegisteredCache> GetSystemNANDContents() const;
-    std::shared_ptr<RegisteredCache> GetUserNANDContents() const;
+    RegisteredCache* GetSystemNANDContents() const;
+    RegisteredCache* GetUserNANDContents() const;
 
     VirtualDir GetModificationLoadRoot(u64 title_id) const;
 
@@ -29,8 +29,8 @@ private:
     VirtualDir nand_root;
     VirtualDir load_root;
 
-    std::shared_ptr<RegisteredCache> sysnand_cache;
-    std::shared_ptr<RegisteredCache> usrnand_cache;
+    std::unique_ptr<RegisteredCache> sysnand_cache;
+    std::unique_ptr<RegisteredCache> usrnand_cache;
 };
 
 } // namespace FileSys

src/core/file_sys/card_image.cpp

@@ -122,14 +122,16 @@ u64 XCI::GetProgramTitleID() const {
     return secure_partition->GetProgramTitleID();
 }
 
-std::shared_ptr<NCA> XCI::GetProgramNCA() const {
-    return program;
+bool XCI::HasProgramNCA() const {
+    return program != nullptr;
 }
 
 VirtualFile XCI::GetProgramNCAFile() const {
-    if (GetProgramNCA() == nullptr)
+    if (!HasProgramNCA()) {
         return nullptr;
-    return GetProgramNCA()->GetBaseFile();
+    }
+
+    return program->GetBaseFile();
 }
 
 const std::vector<std::shared_ptr<NCA>>& XCI::GetNCAs() const {

src/core/file_sys/card_image.h

@@ -80,7 +80,7 @@ public:
 
     u64 GetProgramTitleID() const;
 
-    std::shared_ptr<NCA> GetProgramNCA() const;
+    bool HasProgramNCA() const;
     VirtualFile GetProgramNCAFile() const;
     const std::vector<std::shared_ptr<NCA>>& GetNCAs() const;
     std::shared_ptr<NCA> GetNCAByType(NCAContentType type) const;

src/core/file_sys/content_archive.cpp

@@ -97,11 +97,288 @@ union NCASectionHeader {
 };
 static_assert(sizeof(NCASectionHeader) == 0x200, "NCASectionHeader has incorrect size.");
 
-bool IsValidNCA(const NCAHeader& header) {
+static bool IsValidNCA(const NCAHeader& header) {
     // TODO(DarkLordZach): Add NCA2/NCA0 support.
     return header.magic == Common::MakeMagic('N', 'C', 'A', '3');
 }
 
+NCA::NCA(VirtualFile file_, VirtualFile bktr_base_romfs_, u64 bktr_base_ivfc_offset)
+    : file(std::move(file_)), bktr_base_romfs(std::move(bktr_base_romfs_)) {
+    if (file == nullptr) {
+        status = Loader::ResultStatus::ErrorNullFile;
+        return;
+    }
+
+    if (sizeof(NCAHeader) != file->ReadObject(&header)) {
+        LOG_ERROR(Loader, "File reader errored out during header read.");
+        status = Loader::ResultStatus::ErrorBadNCAHeader;
+        return;
+    }
+
+    if (!HandlePotentialHeaderDecryption()) {
+        return;
+    }
+
+    has_rights_id = std::any_of(header.rights_id.begin(), header.rights_id.end(),
+                                [](char c) { return c != '\0'; });
+
+    const std::vector<NCASectionHeader> sections = ReadSectionHeaders();
+    is_update = std::any_of(sections.begin(), sections.end(), [](const NCASectionHeader& header) {
+        return header.raw.header.crypto_type == NCASectionCryptoType::BKTR;
+    });
+
+    if (!ReadSections(sections, bktr_base_ivfc_offset)) {
+        return;
+    }
+
+    status = Loader::ResultStatus::Success;
+}
+
+NCA::~NCA() = default;
+
+bool NCA::CheckSupportedNCA(const NCAHeader& nca_header) {
+    if (nca_header.magic == Common::MakeMagic('N', 'C', 'A', '2')) {
+        status = Loader::ResultStatus::ErrorNCA2;
+        return false;
+    }
+
+    if (nca_header.magic == Common::MakeMagic('N', 'C', 'A', '0')) {
+        status = Loader::ResultStatus::ErrorNCA0;
+        return false;
+    }
+
+    return true;
+}
+
+bool NCA::HandlePotentialHeaderDecryption() {
+    if (IsValidNCA(header)) {
+        return true;
+    }
+
+    if (!CheckSupportedNCA(header)) {
+        return false;
+    }
+
+    NCAHeader dec_header{};
+    Core::Crypto::AESCipher<Core::Crypto::Key256> cipher(
+        keys.GetKey(Core::Crypto::S256KeyType::Header), Core::Crypto::Mode::XTS);
+    cipher.XTSTranscode(&header, sizeof(NCAHeader), &dec_header, 0, 0x200,
+                        Core::Crypto::Op::Decrypt);
+    if (IsValidNCA(dec_header)) {
+        header = dec_header;
+        encrypted = true;
+    } else {
+        if (!CheckSupportedNCA(dec_header)) {
+            return false;
+        }
+
+        if (keys.HasKey(Core::Crypto::S256KeyType::Header)) {
+            status = Loader::ResultStatus::ErrorIncorrectHeaderKey;
+        } else {
+            status = Loader::ResultStatus::ErrorMissingHeaderKey;
+        }
+        return false;
+    }
+
+    return true;
+}
+
+std::vector<NCASectionHeader> NCA::ReadSectionHeaders() const {
+    const std::ptrdiff_t number_sections =
+        std::count_if(std::begin(header.section_tables), std::end(header.section_tables),
+                      [](NCASectionTableEntry entry) { return entry.media_offset > 0; });
+
+    std::vector<NCASectionHeader> sections(number_sections);
+    const auto length_sections = SECTION_HEADER_SIZE * number_sections;
+
+    if (encrypted) {
+        auto raw = file->ReadBytes(length_sections, SECTION_HEADER_OFFSET);
+        Core::Crypto::AESCipher<Core::Crypto::Key256> cipher(
+            keys.GetKey(Core::Crypto::S256KeyType::Header), Core::Crypto::Mode::XTS);
+        cipher.XTSTranscode(raw.data(), length_sections, sections.data(), 2, SECTION_HEADER_SIZE,
+                            Core::Crypto::Op::Decrypt);
+    } else {
+        file->ReadBytes(sections.data(), length_sections, SECTION_HEADER_OFFSET);
+    }
+
+    return sections;
+}
+
+bool NCA::ReadSections(const std::vector<NCASectionHeader>& sections, u64 bktr_base_ivfc_offset) {
+    for (std::size_t i = 0; i < sections.size(); ++i) {
+        const auto& section = sections[i];
+
+        if (section.raw.header.filesystem_type == NCASectionFilesystemType::ROMFS) {
+            if (!ReadRomFSSection(section, header.section_tables[i], bktr_base_ivfc_offset)) {
+                return false;
+            }
+        } else if (section.raw.header.filesystem_type == NCASectionFilesystemType::PFS0) {
+            if (!ReadPFS0Section(section, header.section_tables[i])) {
+                return false;
+            }
+        }
+    }
+
+    return true;
+}
+
+bool NCA::ReadRomFSSection(const NCASectionHeader& section, const NCASectionTableEntry& entry,
+                           u64 bktr_base_ivfc_offset) {
+    const std::size_t base_offset = entry.media_offset * MEDIA_OFFSET_MULTIPLIER;
+    ivfc_offset = section.romfs.ivfc.levels[IVFC_MAX_LEVEL - 1].offset;
+    const std::size_t romfs_offset = base_offset + ivfc_offset;
+    const std::size_t romfs_size = section.romfs.ivfc.levels[IVFC_MAX_LEVEL - 1].size;
+    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 false;
+        if (has_rights_id)
+            status = Loader::ResultStatus::ErrorIncorrectTitlekeyOrTitlekek;
+        else
+            status = Loader::ResultStatus::ErrorIncorrectKeyAreaKey;
+        return false;
+    }
+
+    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 false;
+        }
+
+        if (section.bktr.relocation.offset + section.bktr.relocation.size !=
+            section.bktr.subsection.offset) {
+            status = Loader::ResultStatus::ErrorBKTRSubsectionNotAfterRelocation;
+            return false;
+        }
+
+        const u64 size = MEDIA_OFFSET_MULTIPLIER * (entry.media_end_offset - entry.media_offset);
+        if (section.bktr.subsection.offset + section.bktr.subsection.size != size) {
+            status = Loader::ResultStatus::ErrorBKTRSubsectionNotAtEnd;
+            return false;
+        }
+
+        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 false;
+        }
+        SubsectionBlock subsection_block{};
+        if (dec->ReadObject(&subsection_block, section.bktr.subsection.offset - offset) !=
+            sizeof(RelocationBlock)) {
+            status = Loader::ResultStatus::ErrorBadSubsectionBlock;
+            return false;
+        }
+
+        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 false;
+        }
+
+        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 false;
+        }
+
+        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 false;
+                }
+            } else {
+                key = GetKeyAreaKey(NCASectionCryptoType::BKTR);
+                if (key == boost::none) {
+                    status = Loader::ResultStatus::ErrorMissingKeyAreaKey;
+                    return false;
+                }
+            }
+        }
+
+        if (bktr_base_romfs == nullptr) {
+            status = Loader::ResultStatus::ErrorMissingBKTRBaseRomFS;
+            return false;
+        }
+
+        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));
+    } else {
+        files.push_back(std::move(dec));
+    }
+
+    romfs = files.back();
+    return true;
+}
+
+bool NCA::ReadPFS0Section(const NCASectionHeader& section, const NCASectionTableEntry& entry) {
+    const u64 offset = (static_cast<u64>(entry.media_offset) * MEDIA_OFFSET_MULTIPLIER) +
+                       section.pfs0.pfs0_header_offset;
+    const u64 size = MEDIA_OFFSET_MULTIPLIER * (entry.media_end_offset - entry.media_offset);
+
+    auto dec = Decrypt(section, std::make_shared<OffsetVfsFile>(file, size, offset), offset);
+    if (dec != nullptr) {
+        auto npfs = std::make_shared<PartitionFilesystem>(std::move(dec));
+
+        if (npfs->GetStatus() == Loader::ResultStatus::Success) {
+            dirs.push_back(std::move(npfs));
+            if (IsDirectoryExeFS(dirs.back()))
+                exefs = dirs.back();
+        } else {
+            if (has_rights_id)
+                status = Loader::ResultStatus::ErrorIncorrectTitlekeyOrTitlekek;
+            else
+                status = Loader::ResultStatus::ErrorIncorrectKeyAreaKey;
+            return false;
+        }
+    } else {
+        if (status != Loader::ResultStatus::Success)
+            return false;
+        if (has_rights_id)
+            status = Loader::ResultStatus::ErrorIncorrectTitlekeyOrTitlekek;
+        else
+            status = Loader::ResultStatus::ErrorIncorrectKeyAreaKey;
+        return false;
+    }
+
+    return true;
+}
+
 u8 NCA::GetCryptoRevision() const {
     u8 master_key_id = header.crypto_type;
     if (header.crypto_type_2 > master_key_id)
@@ -133,7 +410,7 @@ boost::optional<Core::Crypto::Key128> NCA::GetKeyAreaKey(NCASectionCryptoType ty
                      static_cast<u8>(type));
     u128 out_128{};
     memcpy(out_128.data(), out.data(), 16);
-    LOG_DEBUG(Crypto, "called with crypto_rev={:02X}, kak_index={:02X}, key={:016X}{:016X}",
+    LOG_TRACE(Crypto, "called with crypto_rev={:02X}, kak_index={:02X}, key={:016X}{:016X}",
               master_key_id, header.key_index, out_128[1], out_128[0]);
 
     return out;
@@ -167,7 +444,7 @@ boost::optional<Core::Crypto::Key128> NCA::GetTitlekey() {
     return titlekey;
 }
 
-VirtualFile NCA::Decrypt(NCASectionHeader s_header, VirtualFile in, u64 starting_offset) {
+VirtualFile NCA::Decrypt(const NCASectionHeader& s_header, VirtualFile in, u64 starting_offset) {
     if (!encrypted)
         return in;
 
@@ -215,256 +492,6 @@ VirtualFile NCA::Decrypt(NCASectionHeader s_header, VirtualFile in, u64 starting
     }
 }
 
-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) {
-        status = Loader::ResultStatus::ErrorNullFile;
-        return;
-    }
-
-    if (sizeof(NCAHeader) != file->ReadObject(&header)) {
-        LOG_ERROR(Loader, "File reader errored out during header read.");
-        status = Loader::ResultStatus::ErrorBadNCAHeader;
-        return;
-    }
-
-    encrypted = false;
-
-    if (!IsValidNCA(header)) {
-        if (header.magic == Common::MakeMagic('N', 'C', 'A', '2')) {
-            status = Loader::ResultStatus::ErrorNCA2;
-            return;
-        }
-        if (header.magic == Common::MakeMagic('N', 'C', 'A', '0')) {
-            status = Loader::ResultStatus::ErrorNCA0;
-            return;
-        }
-
-        NCAHeader dec_header{};
-        Core::Crypto::AESCipher<Core::Crypto::Key256> cipher(
-            keys.GetKey(Core::Crypto::S256KeyType::Header), Core::Crypto::Mode::XTS);
-        cipher.XTSTranscode(&header, sizeof(NCAHeader), &dec_header, 0, 0x200,
-                            Core::Crypto::Op::Decrypt);
-        if (IsValidNCA(dec_header)) {
-            header = dec_header;
-            encrypted = true;
-        } else {
-            if (dec_header.magic == Common::MakeMagic('N', 'C', 'A', '2')) {
-                status = Loader::ResultStatus::ErrorNCA2;
-                return;
-            }
-            if (dec_header.magic == Common::MakeMagic('N', 'C', 'A', '0')) {
-                status = Loader::ResultStatus::ErrorNCA0;
-                return;
-            }
-
-            if (!keys.HasKey(Core::Crypto::S256KeyType::Header))
-                status = Loader::ResultStatus::ErrorMissingHeaderKey;
-            else
-                status = Loader::ResultStatus::ErrorIncorrectHeaderKey;
-            return;
-        }
-    }
-
-    has_rights_id = std::find_if_not(header.rights_id.begin(), header.rights_id.end(),
-                                     [](char c) { return c == '\0'; }) != header.rights_id.end();
-
-    const std::ptrdiff_t number_sections =
-        std::count_if(std::begin(header.section_tables), std::end(header.section_tables),
-                      [](NCASectionTableEntry entry) { return entry.media_offset > 0; });
-
-    std::vector<NCASectionHeader> sections(number_sections);
-    const auto length_sections = SECTION_HEADER_SIZE * number_sections;
-
-    if (encrypted) {
-        auto raw = file->ReadBytes(length_sections, SECTION_HEADER_OFFSET);
-        Core::Crypto::AESCipher<Core::Crypto::Key256> cipher(
-            keys.GetKey(Core::Crypto::S256KeyType::Header), Core::Crypto::Mode::XTS);
-        cipher.XTSTranscode(raw.data(), length_sections, sections.data(), 2, SECTION_HEADER_SIZE,
-                            Core::Crypto::Op::Decrypt);
-    } else {
-        file->ReadBytes(sections.data(), length_sections, SECTION_HEADER_OFFSET);
-    }
-
-    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 std::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 std::size_t romfs_offset = base_offset + ivfc_offset;
-            const std::size_t romfs_size = section.romfs.ivfc.levels[IVFC_MAX_LEVEL - 1].size;
-            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)
-                    status = Loader::ResultStatus::ErrorIncorrectTitlekeyOrTitlekek;
-                else
-                    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();
-            }
-        } else if (section.raw.header.filesystem_type == NCASectionFilesystemType::PFS0) {
-            u64 offset = (static_cast<u64>(header.section_tables[i].media_offset) *
-                          MEDIA_OFFSET_MULTIPLIER) +
-                         section.pfs0.pfs0_header_offset;
-            u64 size = MEDIA_OFFSET_MULTIPLIER * (header.section_tables[i].media_end_offset -
-                                                  header.section_tables[i].media_offset);
-            auto dec =
-                Decrypt(section, std::make_shared<OffsetVfsFile>(file, size, offset), offset);
-            if (dec != nullptr) {
-                auto npfs = std::make_shared<PartitionFilesystem>(std::move(dec));
-
-                if (npfs->GetStatus() == Loader::ResultStatus::Success) {
-                    dirs.push_back(std::move(npfs));
-                    if (IsDirectoryExeFS(dirs.back()))
-                        exefs = dirs.back();
-                } else {
-                    if (has_rights_id)
-                        status = Loader::ResultStatus::ErrorIncorrectTitlekeyOrTitlekek;
-                    else
-                        status = Loader::ResultStatus::ErrorIncorrectKeyAreaKey;
-                    return;
-                }
-            } else {
-                if (status != Loader::ResultStatus::Success)
-                    return;
-                if (has_rights_id)
-                    status = Loader::ResultStatus::ErrorIncorrectTitlekeyOrTitlekek;
-                else
-                    status = Loader::ResultStatus::ErrorIncorrectKeyAreaKey;
-                return;
-            }
-        }
-    }
-
-    status = Loader::ResultStatus::Success;
-}
-
-NCA::~NCA() = default;
-
 Loader::ResultStatus NCA::GetStatus() const {
     return status;
 }

src/core/file_sys/content_archive.h

@@ -73,8 +73,6 @@ inline bool IsDirectoryExeFS(const std::shared_ptr<VfsDirectory>& pfs) {
     return pfs->GetFile("main") != nullptr && pfs->GetFile("main.npdm") != nullptr;
 }
 
-bool IsValidNCA(const NCAHeader& header);
-
 // An implementation of VfsDirectory that represents a Nintendo Content Archive (NCA) conatiner.
 // After construction, use GetStatus to determine if the file is valid and ready to be used.
 class NCA : public ReadOnlyVfsDirectory {
@@ -106,10 +104,19 @@ protected:
     bool ReplaceFileWithSubdirectory(VirtualFile file, VirtualDir dir) override;
 
 private:
+    bool CheckSupportedNCA(const NCAHeader& header);
+    bool HandlePotentialHeaderDecryption();
+
+    std::vector<NCASectionHeader> ReadSectionHeaders() const;
+    bool ReadSections(const std::vector<NCASectionHeader>& sections, u64 bktr_base_ivfc_offset);
+    bool ReadRomFSSection(const NCASectionHeader& section, const NCASectionTableEntry& entry,
+                          u64 bktr_base_ivfc_offset);
+    bool ReadPFS0Section(const NCASectionHeader& section, const NCASectionTableEntry& entry);
+
     u8 GetCryptoRevision() const;
     boost::optional<Core::Crypto::Key128> GetKeyAreaKey(NCASectionCryptoType type) const;
     boost::optional<Core::Crypto::Key128> GetTitlekey();
-    VirtualFile Decrypt(NCASectionHeader header, VirtualFile in, u64 starting_offset);
+    VirtualFile Decrypt(const NCASectionHeader& header, VirtualFile in, u64 starting_offset);
 
     std::vector<VirtualDir> dirs;
     std::vector<VirtualFile> files;
@@ -118,15 +125,15 @@ private:
     VirtualDir exefs = nullptr;
     VirtualFile file;
     VirtualFile bktr_base_romfs;
-    u64 ivfc_offset;
+    u64 ivfc_offset = 0;
 
     NCAHeader header{};
     bool has_rights_id{};
 
     Loader::ResultStatus status{};
 
-    bool encrypted;
-    bool is_update;
+    bool encrypted = false;
+    bool is_update = false;
 
     Core::Crypto::KeyManager keys;
 };

src/core/file_sys/control_metadata.cpp

@@ -17,11 +17,13 @@ const std::array<const char*, 15> LANGUAGE_NAMES = {
 };
 
 std::string LanguageEntry::GetApplicationName() const {
-    return Common::StringFromFixedZeroTerminatedBuffer(application_name.data(), 0x200);
+    return Common::StringFromFixedZeroTerminatedBuffer(application_name.data(),
+                                                       application_name.size());
 }
 
 std::string LanguageEntry::GetDeveloperName() const {
-    return Common::StringFromFixedZeroTerminatedBuffer(developer_name.data(), 0x100);
+    return Common::StringFromFixedZeroTerminatedBuffer(developer_name.data(),
+                                                       developer_name.size());
 }
 
 NACP::NACP(VirtualFile file) : raw(std::make_unique<RawNACP>()) {
@@ -56,7 +58,12 @@ u64 NACP::GetTitleId() const {
     return raw->title_id;
 }
 
+u64 NACP::GetDLCBaseTitleId() const {
+    return raw->dlc_base_title_id;
+}
+
 std::string NACP::GetVersionString() const {
-    return Common::StringFromFixedZeroTerminatedBuffer(raw->version_string.data(), 0x10);
+    return Common::StringFromFixedZeroTerminatedBuffer(raw->version_string.data(),
+                                                       raw->version_string.size());
 }
 } // namespace FileSys

src/core/file_sys/control_metadata.h

@@ -79,6 +79,7 @@ public:
     std::string GetApplicationName(Language language = Language::Default) const;
     std::string GetDeveloperName(Language language = Language::Default) const;
     u64 GetTitleId() const;
+    u64 GetDLCBaseTitleId() const;
     std::string GetVersionString() const;
 
 private:

src/core/file_sys/patch_manager.cpp

@@ -214,8 +214,14 @@ static void ApplyLayeredFS(VirtualFile& romfs, u64 title_id, ContentRecordType t
 
 VirtualFile PatchManager::PatchRomFS(VirtualFile romfs, u64 ivfc_offset, ContentRecordType type,
                                      VirtualFile update_raw) const {
-    LOG_INFO(Loader, "Patching RomFS for title_id={:016X}, type={:02X}", title_id,
-             static_cast<u8>(type));
+    const auto log_string = fmt::format("Patching RomFS for title_id={:016X}, type={:02X}",
+                                        title_id, static_cast<u8>(type))
+                                .c_str();
+
+    if (type == ContentRecordType::Program)
+        LOG_INFO(Loader, log_string);
+    else
+        LOG_DEBUG(Loader, log_string);
 
     if (romfs == nullptr)
         return romfs;
@@ -346,7 +352,7 @@ std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNam
 }
 
 std::pair<std::unique_ptr<NACP>, VirtualFile> PatchManager::GetControlMetadata() const {
-    const auto& installed{Service::FileSystem::GetUnionContents()};
+    const auto installed{Service::FileSystem::GetUnionContents()};
 
     const auto base_control_nca = installed->GetEntry(title_id, ContentRecordType::Control);
     if (base_control_nca == nullptr)

src/core/file_sys/registered_cache.cpp

@@ -308,14 +308,14 @@ VirtualFile RegisteredCache::GetEntryRaw(RegisteredCacheEntry entry) const {
     return GetEntryRaw(entry.title_id, entry.type);
 }
 
-std::shared_ptr<NCA> RegisteredCache::GetEntry(u64 title_id, ContentRecordType type) const {
+std::unique_ptr<NCA> RegisteredCache::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);
+    return std::make_unique<NCA>(raw);
 }
 
-std::shared_ptr<NCA> RegisteredCache::GetEntry(RegisteredCacheEntry entry) const {
+std::unique_ptr<NCA> RegisteredCache::GetEntry(RegisteredCacheEntry entry) const {
     return GetEntry(entry.title_id, entry.type);
 }
 
@@ -516,7 +516,7 @@ bool RegisteredCache::RawInstallYuzuMeta(const CNMT& cnmt) {
                         }) != yuzu_meta.end();
 }
 
-RegisteredCacheUnion::RegisteredCacheUnion(std::vector<std::shared_ptr<RegisteredCache>> caches)
+RegisteredCacheUnion::RegisteredCacheUnion(std::vector<RegisteredCache*> caches)
     : caches(std::move(caches)) {}
 
 void RegisteredCacheUnion::Refresh() {
@@ -572,14 +572,14 @@ 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 {
+std::unique_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);
+    return std::make_unique<NCA>(raw);
 }
 
-std::shared_ptr<NCA> RegisteredCacheUnion::GetEntry(RegisteredCacheEntry entry) const {
+std::unique_ptr<NCA> RegisteredCacheUnion::GetEntry(RegisteredCacheEntry entry) const {
     return GetEntry(entry.title_id, entry.type);
 }
 

src/core/file_sys/registered_cache.h

@@ -88,8 +88,8 @@ public:
     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::unique_ptr<NCA> GetEntry(u64 title_id, ContentRecordType type) const;
+    std::unique_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.
@@ -142,7 +142,7 @@ private:
 // Combines multiple RegisteredCaches (i.e. SysNAND, UserNAND, SDMC) into one interface.
 class RegisteredCacheUnion {
 public:
-    explicit RegisteredCacheUnion(std::vector<std::shared_ptr<RegisteredCache>> caches);
+    explicit RegisteredCacheUnion(std::vector<RegisteredCache*> caches);
 
     void Refresh();
 
@@ -157,8 +157,8 @@ public:
     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::unique_ptr<NCA> GetEntry(u64 title_id, ContentRecordType type) const;
+    std::unique_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.
@@ -168,7 +168,7 @@ public:
         boost::optional<u64> title_id = boost::none) const;
 
 private:
-    std::vector<std::shared_ptr<RegisteredCache>> caches;
+    std::vector<RegisteredCache*> caches;
 };
 
 } // namespace FileSys

src/core/file_sys/savedata_factory.cpp

@@ -51,6 +51,13 @@ ResultVal<VirtualDir> SaveDataFactory::Open(SaveDataSpaceId space, SaveDataDescr
                     meta.title_id);
     }
 
+    if (meta.type == SaveDataType::DeviceSaveData && meta.user_id != u128{0, 0}) {
+        LOG_WARNING(Service_FS,
+                    "Possibly incorrect SaveDataDescriptor, type is DeviceSaveData but user_id is "
+                    "non-zero ({:016X}{:016X})",
+                    meta.user_id[1], meta.user_id[0]);
+    }
+
     std::string save_directory =
         GetFullPath(space, meta.type, meta.title_id, meta.user_id, meta.save_id);
 
@@ -92,6 +99,9 @@ std::string SaveDataFactory::GetFullPath(SaveDataSpaceId space, SaveDataType typ
     case SaveDataSpaceId::NandUser:
         out = "/user/";
         break;
+    case SaveDataSpaceId::TemporaryStorage:
+        out = "/temp/";
+        break;
     default:
         ASSERT_MSG(false, "Unrecognized SaveDataSpaceId: {:02X}", static_cast<u8>(space));
     }
@@ -100,10 +110,11 @@ std::string SaveDataFactory::GetFullPath(SaveDataSpaceId space, SaveDataType typ
     case SaveDataType::SystemSaveData:
         return fmt::format("{}save/{:016X}/{:016X}{:016X}", out, save_id, user_id[1], user_id[0]);
     case SaveDataType::SaveData:
+    case SaveDataType::DeviceSaveData:
         return fmt::format("{}save/{:016X}/{:016X}{:016X}/{:016X}", out, 0, user_id[1], user_id[0],
                            title_id);
     case SaveDataType::TemporaryStorage:
-        return fmt::format("{}temp/{:016X}/{:016X}{:016X}/{:016X}", out, 0, user_id[1], user_id[0],
+        return fmt::format("{}{:016X}/{:016X}{:016X}/{:016X}", out, 0, user_id[1], user_id[0],
                            title_id);
     default:
         ASSERT_MSG(false, "Unrecognized SaveDataType: {:02X}", static_cast<u8>(type));

src/core/file_sys/sdmc_factory.cpp

@@ -10,10 +10,10 @@
 namespace FileSys {
 
 SDMCFactory::SDMCFactory(VirtualDir dir_)
-    : dir(std::move(dir_)), contents(std::make_shared<RegisteredCache>(
+    : dir(std::move(dir_)), contents(std::make_unique<RegisteredCache>(
                                 GetOrCreateDirectoryRelative(dir, "/Nintendo/Contents/registered"),
                                 [](const VirtualFile& file, const NcaID& id) {
-                                    return std::make_shared<NAX>(file, id)->GetDecrypted();
+                                    return NAX{file, id}.GetDecrypted();
                                 })) {}
 
 SDMCFactory::~SDMCFactory() = default;
@@ -22,8 +22,8 @@ ResultVal<VirtualDir> SDMCFactory::Open() {
     return MakeResult<VirtualDir>(dir);
 }
 
-std::shared_ptr<RegisteredCache> SDMCFactory::GetSDMCContents() const {
-    return contents;
+RegisteredCache* SDMCFactory::GetSDMCContents() const {
+    return contents.get();
 }
 
 } // namespace FileSys

src/core/file_sys/sdmc_factory.h

@@ -19,12 +19,12 @@ public:
     ~SDMCFactory();
 
     ResultVal<VirtualDir> Open();
-    std::shared_ptr<RegisteredCache> GetSDMCContents() const;
+    RegisteredCache* GetSDMCContents() const;
 
 private:
     VirtualDir dir;
 
-    std::shared_ptr<RegisteredCache> contents;
+    std::unique_ptr<RegisteredCache> contents;
 };
 
 } // namespace FileSys

src/core/gdbstub/gdbstub.cpp

@@ -207,7 +207,7 @@ void RegisterModule(std::string name, VAddr beg, VAddr end, bool add_elf_ext) {
 
 static Kernel::Thread* FindThreadById(int id) {
     for (u32 core = 0; core < Core::NUM_CPU_CORES; core++) {
-        const auto& threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
+        const auto& threads = Core::System::GetInstance().Scheduler(core).GetThreadList();
         for (auto& thread : threads) {
             if (thread->GetThreadID() == static_cast<u32>(id)) {
                 current_core = core;
@@ -597,7 +597,7 @@ static void HandleQuery() {
     } else if (strncmp(query, "fThreadInfo", strlen("fThreadInfo")) == 0) {
         std::string val = "m";
         for (u32 core = 0; core < Core::NUM_CPU_CORES; core++) {
-            const auto& threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
+            const auto& threads = Core::System::GetInstance().Scheduler(core).GetThreadList();
             for (const auto& thread : threads) {
                 val += fmt::format("{:x}", thread->GetThreadID());
                 val += ",";
@@ -612,7 +612,7 @@ static void HandleQuery() {
         buffer += "l<?xml version=\"1.0\"?>";
         buffer += "<threads>";
         for (u32 core = 0; core < Core::NUM_CPU_CORES; core++) {
-            const auto& threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
+            const auto& threads = Core::System::GetInstance().Scheduler(core).GetThreadList();
             for (const auto& thread : threads) {
                 buffer +=
                     fmt::format(R"*(<thread id="{:x}" core="{:d}" name="Thread {:x}"></thread>)*",

src/core/hle/kernel/address_arbiter.cpp

@@ -39,7 +39,7 @@ static std::vector<SharedPtr<Thread>> GetThreadsWaitingOnAddress(VAddr address)
                                            std::vector<SharedPtr<Thread>>& waiting_threads,
                                            VAddr arb_addr) {
         const auto& scheduler = Core::System::GetInstance().Scheduler(core_index);
-        const auto& thread_list = scheduler->GetThreadList();
+        const auto& thread_list = scheduler.GetThreadList();
 
         for (const auto& thread : thread_list) {
             if (thread->GetArbiterWaitAddress() == arb_addr)

src/core/hle/kernel/process.cpp

@@ -153,11 +153,11 @@ void Process::PrepareForTermination() {
         }
     };
 
-    auto& system = Core::System::GetInstance();
-    stop_threads(system.Scheduler(0)->GetThreadList());
-    stop_threads(system.Scheduler(1)->GetThreadList());
-    stop_threads(system.Scheduler(2)->GetThreadList());
-    stop_threads(system.Scheduler(3)->GetThreadList());
+    const auto& system = Core::System::GetInstance();
+    stop_threads(system.Scheduler(0).GetThreadList());
+    stop_threads(system.Scheduler(1).GetThreadList());
+    stop_threads(system.Scheduler(2).GetThreadList());
+    stop_threads(system.Scheduler(3).GetThreadList());
 }
 
 /**

src/core/hle/kernel/process.h

@@ -24,6 +24,7 @@ class ProgramMetadata;
 namespace Kernel {
 
 class KernelCore;
+class ResourceLimit;
 
 struct AddressMapping {
     // Address and size must be page-aligned
@@ -57,9 +58,23 @@ union ProcessFlags {
     BitField<12, 1, u16> loaded_high; ///< Application loaded high (not at 0x00100000).
 };
 
-enum class ProcessStatus { Created, Running, Exited };
-
-class ResourceLimit;
+/**
+ * Indicates the status of a Process instance.
+ *
+ * @note These match the values as used by kernel,
+ *       so new entries should only be added if RE
+ *       shows that a new value has been introduced.
+ */
+enum class ProcessStatus {
+    Created,
+    CreatedWithDebuggerAttached,
+    Running,
+    WaitingForDebuggerToAttach,
+    DebuggerAttached,
+    Exiting,
+    Exited,
+    DebugBreak,
+};
 
 struct CodeSet final {
     struct Segment {

src/core/hle/kernel/svc.cpp

@@ -8,6 +8,7 @@
 #include <mutex>
 #include <vector>
 
+#include "common/alignment.h"
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "common/microprofile.h"
@@ -36,9 +37,6 @@
 
 namespace Kernel {
 namespace {
-constexpr bool Is4KBAligned(VAddr address) {
-    return (address & 0xFFF) == 0;
-}
 
 // Checks if address + size is greater than the given address
 // This can return false if the size causes an overflow of a 64-bit type
@@ -69,11 +67,11 @@ bool IsInsideNewMapRegion(const VMManager& vm, VAddr address, u64 size) {
 // in the same order.
 ResultCode MapUnmapMemorySanityChecks(const VMManager& vm_manager, VAddr dst_addr, VAddr src_addr,
                                       u64 size) {
-    if (!Is4KBAligned(dst_addr) || !Is4KBAligned(src_addr)) {
+    if (!Common::Is4KBAligned(dst_addr) || !Common::Is4KBAligned(src_addr)) {
         return ERR_INVALID_ADDRESS;
     }
 
-    if (size == 0 || !Is4KBAligned(size)) {
+    if (size == 0 || !Common::Is4KBAligned(size)) {
         return ERR_INVALID_SIZE;
     }
 
@@ -352,6 +350,10 @@ static ResultCode ArbitrateLock(Handle holding_thread_handle, VAddr mutex_addr,
         return ERR_INVALID_ADDRESS_STATE;
     }
 
+    if (!Common::IsWordAligned(mutex_addr)) {
+        return ERR_INVALID_ADDRESS;
+    }
+
     auto& handle_table = Core::System::GetInstance().Kernel().HandleTable();
     return Mutex::TryAcquire(handle_table, mutex_addr, holding_thread_handle,
                              requesting_thread_handle);
@@ -365,6 +367,10 @@ static ResultCode ArbitrateUnlock(VAddr mutex_addr) {
         return ERR_INVALID_ADDRESS_STATE;
     }
 
+    if (!Common::IsWordAligned(mutex_addr)) {
+        return ERR_INVALID_ADDRESS;
+    }
+
     return Mutex::Release(mutex_addr);
 }
 
@@ -448,25 +454,12 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
     case GetInfoType::RandomEntropy:
         *result = 0;
         break;
-    case GetInfoType::AddressSpaceBaseAddr:
-        *result = vm_manager.GetCodeRegionBaseAddress();
+    case GetInfoType::ASLRRegionBaseAddr:
+        *result = vm_manager.GetASLRRegionBaseAddress();
         break;
-    case GetInfoType::AddressSpaceSize: {
-        const u64 width = vm_manager.GetAddressSpaceWidth();
-
-        switch (width) {
-        case 32:
-            *result = 0xFFE00000;
-            break;
-        case 36:
-            *result = 0xFF8000000;
-            break;
-        case 39:
-            *result = 0x7FF8000000;
-            break;
-        }
+    case GetInfoType::ASLRRegionSize:
+        *result = vm_manager.GetASLRRegionSize();
         break;
-    }
     case GetInfoType::NewMapRegionBaseAddr:
         *result = vm_manager.GetNewMapRegionBaseAddress();
         break;
@@ -583,11 +576,11 @@ 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);
 
-    if (!Is4KBAligned(addr)) {
+    if (!Common::Is4KBAligned(addr)) {
         return ERR_INVALID_ADDRESS;
     }
 
-    if (size == 0 || !Is4KBAligned(size)) {
+    if (size == 0 || !Common::Is4KBAligned(size)) {
         return ERR_INVALID_SIZE;
     }
 
@@ -612,11 +605,11 @@ 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);
 
-    if (!Is4KBAligned(addr)) {
+    if (!Common::Is4KBAligned(addr)) {
         return ERR_INVALID_ADDRESS;
     }
 
-    if (size == 0 || !Is4KBAligned(size)) {
+    if (size == 0 || !Common::Is4KBAligned(size)) {
         return ERR_INVALID_SIZE;
     }
 
@@ -809,7 +802,7 @@ static ResultCode SignalProcessWideKey(VAddr condition_variable_addr, s32 target
                                            std::vector<SharedPtr<Thread>>& waiting_threads,
                                            VAddr condvar_addr) {
         const auto& scheduler = Core::System::GetInstance().Scheduler(core_index);
-        const auto& thread_list = scheduler->GetThreadList();
+        const auto& thread_list = scheduler.GetThreadList();
 
         for (const auto& thread : thread_list) {
             if (thread->GetCondVarWaitAddress() == condvar_addr)
@@ -1098,6 +1091,29 @@ static ResultCode ClearEvent(Handle handle) {
     return RESULT_SUCCESS;
 }
 
+static ResultCode GetProcessInfo(u64* out, Handle process_handle, u32 type) {
+    LOG_DEBUG(Kernel_SVC, "called, handle=0x{:08X}, type=0x{:X}", process_handle, type);
+
+    // This function currently only allows retrieving a process' status.
+    enum class InfoType {
+        Status,
+    };
+
+    const auto& kernel = Core::System::GetInstance().Kernel();
+    const auto process = kernel.HandleTable().Get<Process>(process_handle);
+    if (!process) {
+        return ERR_INVALID_HANDLE;
+    }
+
+    const auto info_type = static_cast<InfoType>(type);
+    if (info_type != InfoType::Status) {
+        return ERR_INVALID_ENUM_VALUE;
+    }
+
+    *out = static_cast<u64>(process->GetStatus());
+    return RESULT_SUCCESS;
+}
+
 namespace {
 struct FunctionDef {
     using Func = void();
@@ -1233,7 +1249,7 @@ static const FunctionDef SVC_Table[] = {
     {0x79, nullptr, "CreateProcess"},
     {0x7A, nullptr, "StartProcess"},
     {0x7B, nullptr, "TerminateProcess"},
-    {0x7C, nullptr, "GetProcessInfo"},
+    {0x7C, SvcWrap<GetProcessInfo>, "GetProcessInfo"},
     {0x7D, nullptr, "CreateResourceLimit"},
     {0x7E, nullptr, "SetResourceLimitLimitValue"},
     {0x7F, nullptr, "CallSecureMonitor"},

src/core/hle/kernel/svc.h

@@ -41,8 +41,8 @@ enum class GetInfoType : u64 {
     RandomEntropy = 11,
     PerformanceCounter = 0xF0000002,
     // 2.0.0+
-    AddressSpaceBaseAddr = 12,
-    AddressSpaceSize = 13,
+    ASLRRegionBaseAddr = 12,
+    ASLRRegionSize = 13,
     NewMapRegionBaseAddr = 14,
     NewMapRegionSize = 15,
     // 3.0.0+

src/core/hle/kernel/svc_wrap.h

@@ -77,6 +77,14 @@ void SvcWrap() {
     FuncReturn(retval);
 }
 
+template <ResultCode func(u64*, u32, u32)>
+void SvcWrap() {
+    u64 param_1 = 0;
+    u32 retval = func(&param_1, static_cast<u32>(Param(1)), static_cast<u32>(Param(2))).raw;
+    Core::CurrentArmInterface().SetReg(1, param_1);
+    FuncReturn(retval);
+}
+
 template <ResultCode func(u32, u64)>
 void SvcWrap() {
     FuncReturn(func(static_cast<u32>(Param(0)), Param(1)).raw);

src/core/hle/kernel/thread.cpp

@@ -97,7 +97,7 @@ void Thread::CancelWakeupTimer() {
 static boost::optional<s32> GetNextProcessorId(u64 mask) {
     for (s32 index = 0; index < Core::NUM_CPU_CORES; ++index) {
         if (mask & (1ULL << index)) {
-            if (!Core::System::GetInstance().Scheduler(index)->GetCurrentThread()) {
+            if (!Core::System::GetInstance().Scheduler(index).GetCurrentThread()) {
                 // Core is enabled and not running any threads, use this one
                 return index;
             }
@@ -147,14 +147,14 @@ void Thread::ResumeFromWait() {
         new_processor_id = processor_id;
     }
     if (ideal_core != -1 &&
-        Core::System::GetInstance().Scheduler(ideal_core)->GetCurrentThread() == nullptr) {
+        Core::System::GetInstance().Scheduler(ideal_core).GetCurrentThread() == nullptr) {
         new_processor_id = ideal_core;
     }
 
     ASSERT(*new_processor_id < 4);
 
     // Add thread to new core's scheduler
-    auto& next_scheduler = Core::System::GetInstance().Scheduler(*new_processor_id);
+    auto* next_scheduler = &Core::System::GetInstance().Scheduler(*new_processor_id);
 
     if (*new_processor_id != processor_id) {
         // Remove thread from previous core's scheduler
@@ -169,7 +169,7 @@ void Thread::ResumeFromWait() {
     next_scheduler->ScheduleThread(this, current_priority);
 
     // Change thread's scheduler
-    scheduler = next_scheduler.get();
+    scheduler = next_scheduler;
 
     Core::System::GetInstance().CpuCore(processor_id).PrepareReschedule();
 }
@@ -230,7 +230,7 @@ ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name
     thread->name = std::move(name);
     thread->callback_handle = kernel.ThreadWakeupCallbackHandleTable().Create(thread).Unwrap();
     thread->owner_process = &owner_process;
-    thread->scheduler = Core::System::GetInstance().Scheduler(processor_id).get();
+    thread->scheduler = &Core::System::GetInstance().Scheduler(processor_id);
     thread->scheduler->AddThread(thread, priority);
     thread->tls_address = thread->owner_process->MarkNextAvailableTLSSlotAsUsed(*thread);
 
@@ -375,14 +375,14 @@ void Thread::ChangeCore(u32 core, u64 mask) {
         new_processor_id = processor_id;
     }
     if (ideal_core != -1 &&
-        Core::System::GetInstance().Scheduler(ideal_core)->GetCurrentThread() == nullptr) {
+        Core::System::GetInstance().Scheduler(ideal_core).GetCurrentThread() == nullptr) {
         new_processor_id = ideal_core;
     }
 
     ASSERT(*new_processor_id < 4);
 
     // Add thread to new core's scheduler
-    auto& next_scheduler = Core::System::GetInstance().Scheduler(*new_processor_id);
+    auto* next_scheduler = &Core::System::GetInstance().Scheduler(*new_processor_id);
 
     if (*new_processor_id != processor_id) {
         // Remove thread from previous core's scheduler
@@ -397,7 +397,7 @@ void Thread::ChangeCore(u32 core, u64 mask) {
     next_scheduler->ScheduleThread(this, current_priority);
 
     // Change thread's scheduler
-    scheduler = next_scheduler.get();
+    scheduler = next_scheduler;
 
     Core::System::GetInstance().CpuCore(processor_id).PrepareReschedule();
 }

src/core/hle/kernel/vm_manager.cpp

@@ -393,30 +393,35 @@ void VMManager::InitializeMemoryRegionRanges(FileSys::ProgramAddressSpaceType ty
 
     switch (type) {
     case FileSys::ProgramAddressSpaceType::Is32Bit:
+    case FileSys::ProgramAddressSpaceType::Is32BitNoMap:
         address_space_width = 32;
         code_region_base = 0x200000;
         code_region_end = code_region_base + 0x3FE00000;
-        map_region_size = 0x40000000;
-        heap_region_size = 0x40000000;
+        aslr_region_base = 0x200000;
+        aslr_region_end = aslr_region_base + 0xFFE00000;
+        if (type == FileSys::ProgramAddressSpaceType::Is32Bit) {
+            map_region_size = 0x40000000;
+            heap_region_size = 0x40000000;
+        } else {
+            map_region_size = 0;
+            heap_region_size = 0x80000000;
+        }
         break;
     case FileSys::ProgramAddressSpaceType::Is36Bit:
         address_space_width = 36;
         code_region_base = 0x8000000;
         code_region_end = code_region_base + 0x78000000;
+        aslr_region_base = 0x8000000;
+        aslr_region_end = aslr_region_base + 0xFF8000000;
         map_region_size = 0x180000000;
         heap_region_size = 0x180000000;
         break;
-    case FileSys::ProgramAddressSpaceType::Is32BitNoMap:
-        address_space_width = 32;
-        code_region_base = 0x200000;
-        code_region_end = code_region_base + 0x3FE00000;
-        map_region_size = 0;
-        heap_region_size = 0x80000000;
-        break;
     case FileSys::ProgramAddressSpaceType::Is39Bit:
         address_space_width = 39;
         code_region_base = 0x8000000;
         code_region_end = code_region_base + 0x80000000;
+        aslr_region_base = 0x8000000;
+        aslr_region_end = aslr_region_base + 0x7FF8000000;
         map_region_size = 0x1000000000;
         heap_region_size = 0x180000000;
         new_map_region_size = 0x80000000;
@@ -490,6 +495,18 @@ u64 VMManager::GetAddressSpaceWidth() const {
     return address_space_width;
 }
 
+VAddr VMManager::GetASLRRegionBaseAddress() const {
+    return aslr_region_base;
+}
+
+VAddr VMManager::GetASLRRegionEndAddress() const {
+    return aslr_region_end;
+}
+
+u64 VMManager::GetASLRRegionSize() const {
+    return aslr_region_end - aslr_region_base;
+}
+
 VAddr VMManager::GetCodeRegionBaseAddress() const {
     return code_region_base;
 }

src/core/hle/kernel/vm_manager.h

@@ -205,6 +205,15 @@ public:
     /// Gets the address space width in bits.
     u64 GetAddressSpaceWidth() const;
 
+    /// Gets the base address of the ASLR region.
+    VAddr GetASLRRegionBaseAddress() const;
+
+    /// Gets the end address of the ASLR region.
+    VAddr GetASLRRegionEndAddress() const;
+
+    /// Gets the size of the ASLR region
+    u64 GetASLRRegionSize() const;
+
     /// Gets the base address of the code region.
     VAddr GetCodeRegionBaseAddress() const;
 
@@ -306,6 +315,9 @@ private:
     VAddr address_space_base = 0;
     VAddr address_space_end = 0;
 
+    VAddr aslr_region_base = 0;
+    VAddr aslr_region_end = 0;
+
     VAddr code_region_base = 0;
     VAddr code_region_end = 0;
 

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

@@ -638,10 +638,12 @@ IApplicationFunctions::IApplicationFunctions() : ServiceFramework("IApplicationF
         {24, nullptr, "GetLaunchStorageInfoForDebug"},
         {25, nullptr, "ExtendSaveData"},
         {26, nullptr, "GetSaveDataSize"},
-        {30, nullptr, "BeginBlockingHomeButtonShortAndLongPressed"},
-        {31, nullptr, "EndBlockingHomeButtonShortAndLongPressed"},
-        {32, nullptr, "BeginBlockingHomeButton"},
-        {33, nullptr, "EndBlockingHomeButton"},
+        {30, &IApplicationFunctions::BeginBlockingHomeButtonShortAndLongPressed,
+         "BeginBlockingHomeButtonShortAndLongPressed"},
+        {31, &IApplicationFunctions::EndBlockingHomeButtonShortAndLongPressed,
+         "EndBlockingHomeButtonShortAndLongPressed"},
+        {32, &IApplicationFunctions::BeginBlockingHomeButton, "BeginBlockingHomeButton"},
+        {33, &IApplicationFunctions::EndBlockingHomeButton, "EndBlockingHomeButton"},
         {40, &IApplicationFunctions::NotifyRunning, "NotifyRunning"},
         {50, &IApplicationFunctions::GetPseudoDeviceId, "GetPseudoDeviceId"},
         {60, nullptr, "SetMediaPlaybackStateForApplication"},
@@ -669,6 +671,32 @@ IApplicationFunctions::IApplicationFunctions() : ServiceFramework("IApplicationF
 
 IApplicationFunctions::~IApplicationFunctions() = default;
 
+void IApplicationFunctions::BeginBlockingHomeButtonShortAndLongPressed(
+    Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+}
+
+void IApplicationFunctions::EndBlockingHomeButtonShortAndLongPressed(
+    Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+}
+
+void IApplicationFunctions::BeginBlockingHomeButton(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+}
+
+void IApplicationFunctions::EndBlockingHomeButton(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+}
+
 void IApplicationFunctions::PopLaunchParameter(Kernel::HLERequestContext& ctx) {
     constexpr std::array<u8, 0x88> data{{
         0xca, 0x97, 0x94, 0xc7, // Magic

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

@@ -154,6 +154,10 @@ private:
     void SetGamePlayRecordingState(Kernel::HLERequestContext& ctx);
     void NotifyRunning(Kernel::HLERequestContext& ctx);
     void GetPseudoDeviceId(Kernel::HLERequestContext& ctx);
+    void BeginBlockingHomeButtonShortAndLongPressed(Kernel::HLERequestContext& ctx);
+    void EndBlockingHomeButtonShortAndLongPressed(Kernel::HLERequestContext& ctx);
+    void BeginBlockingHomeButton(Kernel::HLERequestContext& ctx);
+    void EndBlockingHomeButton(Kernel::HLERequestContext& ctx);
 };
 
 class IHomeMenuFunctions final : public ServiceFramework<IHomeMenuFunctions> {

src/core/hle/service/aoc/aoc_u.cpp

@@ -7,10 +7,13 @@
 #include <vector>
 #include "common/logging/log.h"
 #include "core/file_sys/content_archive.h"
+#include "core/file_sys/control_metadata.h"
 #include "core/file_sys/nca_metadata.h"
 #include "core/file_sys/partition_filesystem.h"
+#include "core/file_sys/patch_manager.h"
 #include "core/file_sys/registered_cache.h"
 #include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/event.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/service/aoc/aoc_u.h"
 #include "core/hle/service/filesystem/filesystem.h"
@@ -19,7 +22,7 @@
 namespace Service::AOC {
 
 constexpr u64 DLC_BASE_TITLE_ID_MASK = 0xFFFFFFFFFFFFE000;
-constexpr u64 DLC_BASE_TO_AOC_ID_MASK = 0x1000;
+constexpr u64 DLC_BASE_TO_AOC_ID = 0x1000;
 
 static bool CheckAOCTitleIDMatchesBase(u64 base, u64 aoc) {
     return (aoc & DLC_BASE_TITLE_ID_MASK) == base;
@@ -53,9 +56,13 @@ AOC_U::AOC_U() : ServiceFramework("aoc:u"), add_on_content(AccumulateAOCTitleIDs
         {5, &AOC_U::GetAddOnContentBaseId, "GetAddOnContentBaseId"},
         {6, nullptr, "PrepareAddOnContentByApplicationId"},
         {7, &AOC_U::PrepareAddOnContent, "PrepareAddOnContent"},
-        {8, nullptr, "GetAddOnContentListChangedEvent"},
+        {8, &AOC_U::GetAddOnContentListChangedEvent, "GetAddOnContentListChangedEvent"},
     };
     RegisterHandlers(functions);
+
+    auto& kernel = Core::System::GetInstance().Kernel();
+    aoc_change_event = Kernel::Event::Create(kernel, Kernel::ResetType::Sticky,
+                                             "GetAddOnContentListChanged:Event");
 }
 
 AOC_U::~AOC_U() = default;
@@ -97,14 +104,24 @@ void AOC_U::ListAddOnContent(Kernel::HLERequestContext& ctx) {
 
     ctx.WriteBuffer(out);
 
-    IPC::ResponseBuilder rb{ctx, 2};
+    IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
+    rb.Push(count);
 }
 
 void AOC_U::GetAddOnContentBaseId(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 4};
     rb.Push(RESULT_SUCCESS);
-    rb.Push(Core::System::GetInstance().CurrentProcess()->GetTitleID() | DLC_BASE_TO_AOC_ID_MASK);
+    const auto title_id = Core::System::GetInstance().CurrentProcess()->GetTitleID();
+    FileSys::PatchManager pm{title_id};
+
+    const auto res = pm.GetControlMetadata();
+    if (res.first == nullptr) {
+        rb.Push(title_id + DLC_BASE_TO_AOC_ID);
+        return;
+    }
+
+    rb.Push(res.first->GetDLCBaseTitleId());
 }
 
 void AOC_U::PrepareAddOnContent(Kernel::HLERequestContext& ctx) {
@@ -118,6 +135,14 @@ void AOC_U::PrepareAddOnContent(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
 }
 
+void AOC_U::GetAddOnContentListChangedEvent(Kernel::HLERequestContext& ctx) {
+    LOG_WARNING(Service_AOC, "(STUBBED) called");
+
+    IPC::ResponseBuilder rb{ctx, 2, 1};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushCopyObjects(aoc_change_event);
+}
+
 void InstallInterfaces(SM::ServiceManager& service_manager) {
     std::make_shared<AOC_U>()->InstallAsService(service_manager);
 }

src/core/hle/service/aoc/aoc_u.h

@@ -18,8 +18,10 @@ private:
     void ListAddOnContent(Kernel::HLERequestContext& ctx);
     void GetAddOnContentBaseId(Kernel::HLERequestContext& ctx);
     void PrepareAddOnContent(Kernel::HLERequestContext& ctx);
+    void GetAddOnContentListChangedEvent(Kernel::HLERequestContext& ctx);
 
     std::vector<u64> add_on_content;
+    Kernel::SharedPtr<Kernel::Event> aoc_change_event;
 };
 
 /// Registers all AOC services with the specified service manager.

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

@@ -319,13 +319,12 @@ 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::unique_ptr<FileSys::RegisteredCacheUnion> GetUnionContents() {
+    return std::make_unique<FileSys::RegisteredCacheUnion>(std::vector<FileSys::RegisteredCache*>{
+        GetSystemNANDContents(), GetUserNANDContents(), GetSDMCContents()});
 }
 
-std::shared_ptr<FileSys::RegisteredCache> GetSystemNANDContents() {
+FileSys::RegisteredCache* GetSystemNANDContents() {
     LOG_TRACE(Service_FS, "Opening System NAND Contents");
 
     if (bis_factory == nullptr)
@@ -334,7 +333,7 @@ std::shared_ptr<FileSys::RegisteredCache> GetSystemNANDContents() {
     return bis_factory->GetSystemNANDContents();
 }
 
-std::shared_ptr<FileSys::RegisteredCache> GetUserNANDContents() {
+FileSys::RegisteredCache* GetUserNANDContents() {
     LOG_TRACE(Service_FS, "Opening User NAND Contents");
 
     if (bis_factory == nullptr)
@@ -343,7 +342,7 @@ std::shared_ptr<FileSys::RegisteredCache> GetUserNANDContents() {
     return bis_factory->GetUserNANDContents();
 }
 
-std::shared_ptr<FileSys::RegisteredCache> GetSDMCContents() {
+FileSys::RegisteredCache* GetSDMCContents() {
     LOG_TRACE(Service_FS, "Opening SDMC Contents");
 
     if (sdmc_factory == nullptr)
@@ -361,19 +360,19 @@ FileSys::VirtualDir GetModificationLoadRoot(u64 title_id) {
     return bis_factory->GetModificationLoadRoot(title_id);
 }
 
-void CreateFactories(const FileSys::VirtualFilesystem& vfs, bool overwrite) {
+void CreateFactories(FileSys::VfsFilesystem& vfs, bool overwrite) {
     if (overwrite) {
         bis_factory = nullptr;
         save_data_factory = nullptr;
         sdmc_factory = nullptr;
     }
 
-    auto nand_directory = vfs->OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::NANDDir),
-                                             FileSys::Mode::ReadWrite);
-    auto sd_directory = vfs->OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir),
-                                           FileSys::Mode::ReadWrite);
-    auto load_directory = vfs->OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::LoadDir),
-                                             FileSys::Mode::ReadWrite);
+    auto nand_directory = vfs.OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::NANDDir),
+                                            FileSys::Mode::ReadWrite);
+    auto sd_directory = vfs.OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir),
+                                          FileSys::Mode::ReadWrite);
+    auto load_directory = vfs.OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::LoadDir),
+                                            FileSys::Mode::ReadWrite);
 
     if (bis_factory == nullptr)
         bis_factory = std::make_unique<FileSys::BISFactory>(nand_directory, load_directory);
@@ -383,7 +382,7 @@ void CreateFactories(const FileSys::VirtualFilesystem& vfs, bool overwrite) {
         sdmc_factory = std::make_unique<FileSys::SDMCFactory>(std::move(sd_directory));
 }
 
-void InstallInterfaces(SM::ServiceManager& service_manager, const FileSys::VirtualFilesystem& vfs) {
+void InstallInterfaces(SM::ServiceManager& service_manager, FileSys::VfsFilesystem& vfs) {
     romfs_factory = nullptr;
     CreateFactories(vfs, false);
     std::make_shared<FSP_LDR>()->InstallAsService(service_manager);

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

@@ -47,19 +47,19 @@ ResultVal<FileSys::VirtualDir> OpenSaveData(FileSys::SaveDataSpaceId space,
                                             FileSys::SaveDataDescriptor save_struct);
 ResultVal<FileSys::VirtualDir> OpenSDMC();
 
-std::shared_ptr<FileSys::RegisteredCacheUnion> GetUnionContents();
+std::unique_ptr<FileSys::RegisteredCacheUnion> GetUnionContents();
 
-std::shared_ptr<FileSys::RegisteredCache> GetSystemNANDContents();
-std::shared_ptr<FileSys::RegisteredCache> GetUserNANDContents();
-std::shared_ptr<FileSys::RegisteredCache> GetSDMCContents();
+FileSys::RegisteredCache* GetSystemNANDContents();
+FileSys::RegisteredCache* GetUserNANDContents();
+FileSys::RegisteredCache* GetSDMCContents();
 
 FileSys::VirtualDir GetModificationLoadRoot(u64 title_id);
 
 // Creates the SaveData, SDMC, and BIS Factories. Should be called once and before any function
 // above is called.
-void CreateFactories(const FileSys::VirtualFilesystem& vfs, bool overwrite = true);
+void CreateFactories(FileSys::VfsFilesystem& vfs, bool overwrite = true);
 
-void InstallInterfaces(SM::ServiceManager& service_manager, const FileSys::VirtualFilesystem& vfs);
+void InstallInterfaces(SM::ServiceManager& service_manager, FileSys::VfsFilesystem& vfs);
 
 // A class that wraps a VfsDirectory with methods that return ResultVal and ResultCode instead of
 // pointers and booleans. This makes using a VfsDirectory with switch services much easier and

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

@@ -0,0 +1,28 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/service/hid/controllers/controller_base.h"
+
+namespace Service::HID {
+ControllerBase::~ControllerBase() = default;
+
+void ControllerBase::ActivateController() {
+    if (is_activated) {
+        OnRelease();
+    }
+    is_activated = true;
+    OnInit();
+}
+
+void ControllerBase::DeactivateController() {
+    if (is_activated) {
+        OnRelease();
+    }
+    is_activated = false;
+}
+
+bool ControllerBase::IsControllerActivated() const {
+    return is_activated;
+}
+} // namespace Service::HID

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

@@ -0,0 +1,45 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_types.h"
+#include "common/swap.h"
+
+namespace Service::HID {
+class ControllerBase {
+public:
+    ControllerBase() = default;
+    virtual ~ControllerBase() = 0;
+
+    // Called when the controller is initialized
+    virtual void OnInit() = 0;
+
+    // When the controller is released
+    virtual void OnRelease() = 0;
+
+    // When the controller is requesting an update for the shared memory
+    virtual void OnUpdate(u8* data, std::size_t size) = 0;
+
+    // Called when input devices should be loaded
+    virtual void OnLoadInputDevices() = 0;
+
+    void ActivateController();
+
+    void DeactivateController();
+
+    bool IsControllerActivated() const;
+
+protected:
+    bool is_activated{false};
+
+    struct CommonHeader {
+        s64_le timestamp;
+        s64_le total_entry_count;
+        s64_le last_entry_index;
+        s64_le entry_count;
+    };
+    static_assert(sizeof(CommonHeader) == 0x20, "CommonHeader is an invalid size");
+};
+} // namespace Service::HID

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

@@ -0,0 +1,42 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/core_timing.h"
+#include "core/hle/service/hid/controllers/debug_pad.h"
+
+namespace Service::HID {
+
+Controller_DebugPad::Controller_DebugPad() = default;
+
+void Controller_DebugPad::OnInit() {}
+
+void Controller_DebugPad::OnRelease() {}
+
+void Controller_DebugPad::OnUpdate(u8* data, std::size_t size) {
+    shared_memory.header.timestamp = CoreTiming::GetTicks();
+    shared_memory.header.total_entry_count = 17;
+
+    if (!IsControllerActivated()) {
+        shared_memory.header.entry_count = 0;
+        shared_memory.header.last_entry_index = 0;
+        return;
+    }
+    shared_memory.header.entry_count = 16;
+
+    const auto& last_entry = shared_memory.pad_states[shared_memory.header.last_entry_index];
+    shared_memory.header.last_entry_index = (shared_memory.header.last_entry_index + 1) % 17;
+    auto& cur_entry = shared_memory.pad_states[shared_memory.header.last_entry_index];
+
+    cur_entry.sampling_number = last_entry.sampling_number + 1;
+    cur_entry.sampling_number2 = cur_entry.sampling_number;
+    // TODO(ogniK): Update debug pad states
+
+    std::memcpy(data, &shared_memory, sizeof(SharedMemory));
+}
+
+void Controller_DebugPad::OnLoadInputDevices() {}
+} // namespace Service::HID

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

@@ -0,0 +1,55 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/hle/service/hid/controllers/controller_base.h"
+
+namespace Service::HID {
+class Controller_DebugPad final : public ControllerBase {
+public:
+    Controller_DebugPad();
+
+    // Called when the controller is initialized
+    void OnInit() override;
+
+    // When the controller is released
+    void OnRelease() override;
+
+    // When the controller is requesting an update for the shared memory
+    void OnUpdate(u8* data, std::size_t size) override;
+
+    // Called when input devices should be loaded
+    void OnLoadInputDevices() override;
+
+private:
+    struct AnalogStick {
+        s32_le x;
+        s32_le y;
+    };
+    static_assert(sizeof(AnalogStick) == 0x8);
+
+    struct PadStates {
+        s64_le sampling_number;
+        s64_le sampling_number2;
+        u32_le attribute;
+        u32_le button_state;
+        AnalogStick r_stick;
+        AnalogStick l_stick;
+    };
+    static_assert(sizeof(PadStates) == 0x28, "PadStates is an invalid state");
+
+    struct SharedMemory {
+        CommonHeader header;
+        std::array<PadStates, 17> pad_states;
+        INSERT_PADDING_BYTES(0x138);
+    };
+    static_assert(sizeof(SharedMemory) == 0x400, "SharedMemory is an invalid size");
+    SharedMemory shared_memory{};
+};
+} // namespace Service::HID

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

@@ -0,0 +1,43 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/core_timing.h"
+#include "core/hle/service/hid/controllers/gesture.h"
+
+namespace Service::HID {
+constexpr std::size_t SHARED_MEMORY_OFFSET = 0x3BA00;
+
+Controller_Gesture::Controller_Gesture() = default;
+
+void Controller_Gesture::OnInit() {}
+
+void Controller_Gesture::OnRelease() {}
+
+void Controller_Gesture::OnUpdate(u8* data, std::size_t size) {
+    shared_memory.header.timestamp = CoreTiming::GetTicks();
+    shared_memory.header.total_entry_count = 17;
+
+    if (!IsControllerActivated()) {
+        shared_memory.header.entry_count = 0;
+        shared_memory.header.last_entry_index = 0;
+        return;
+    }
+    shared_memory.header.entry_count = 16;
+
+    const auto& last_entry = shared_memory.gesture_states[shared_memory.header.last_entry_index];
+    shared_memory.header.last_entry_index = (shared_memory.header.last_entry_index + 1) % 17;
+    auto& cur_entry = shared_memory.gesture_states[shared_memory.header.last_entry_index];
+
+    cur_entry.sampling_number = last_entry.sampling_number + 1;
+    cur_entry.sampling_number2 = cur_entry.sampling_number;
+    // TODO(ogniK): Update gesture states
+
+    std::memcpy(data + SHARED_MEMORY_OFFSET, &shared_memory, sizeof(SharedMemory));
+}
+
+void Controller_Gesture::OnLoadInputDevices() {}
+} // namespace Service::HID

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

@@ -0,0 +1,62 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/hle/service/hid/controllers/controller_base.h"
+
+namespace Service::HID {
+class Controller_Gesture final : public ControllerBase {
+public:
+    Controller_Gesture();
+
+    // Called when the controller is initialized
+    void OnInit() override;
+
+    // When the controller is released
+    void OnRelease() override;
+
+    // When the controller is requesting an update for the shared memory
+    void OnUpdate(u8* data, size_t size) override;
+
+    // Called when input devices should be loaded
+    void OnLoadInputDevices() override;
+
+private:
+    struct Locations {
+        s32_le x;
+        s32_le y;
+    };
+
+    struct GestureState {
+        s64_le sampling_number;
+        s64_le sampling_number2;
+
+        s64_le detection_count;
+        s32_le type;
+        s32_le dir;
+        s32_le x;
+        s32_le y;
+        s32_le delta_x;
+        s32_le delta_y;
+        f32 vel_x;
+        f32 vel_y;
+        s32_le attributes;
+        f32 scale;
+        f32 rotation;
+        s32_le location_count;
+        std::array<Locations, 4> locations;
+    };
+    static_assert(sizeof(GestureState) == 0x68, "GestureState is an invalid size");
+
+    struct SharedMemory {
+        CommonHeader header;
+        std::array<GestureState, 17> gesture_states;
+    };
+    SharedMemory shared_memory{};
+};
+} // namespace Service::HID

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

@@ -0,0 +1,43 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/core_timing.h"
+#include "core/hle/service/hid/controllers/keyboard.h"
+
+namespace Service::HID {
+constexpr std::size_t SHARED_MEMORY_OFFSET = 0x3800;
+
+Controller_Keyboard::Controller_Keyboard() = default;
+
+void Controller_Keyboard::OnInit() {}
+
+void Controller_Keyboard::OnRelease() {}
+
+void Controller_Keyboard::OnUpdate(u8* data, std::size_t size) {
+    shared_memory.header.timestamp = CoreTiming::GetTicks();
+    shared_memory.header.total_entry_count = 17;
+
+    if (!IsControllerActivated()) {
+        shared_memory.header.entry_count = 0;
+        shared_memory.header.last_entry_index = 0;
+        return;
+    }
+    shared_memory.header.entry_count = 16;
+
+    const auto& last_entry = shared_memory.pad_states[shared_memory.header.last_entry_index];
+    shared_memory.header.last_entry_index = (shared_memory.header.last_entry_index + 1) % 17;
+    auto& cur_entry = shared_memory.pad_states[shared_memory.header.last_entry_index];
+
+    cur_entry.sampling_number = last_entry.sampling_number + 1;
+    cur_entry.sampling_number2 = cur_entry.sampling_number;
+    // TODO(ogniK): Update keyboard states
+
+    std::memcpy(data + SHARED_MEMORY_OFFSET, &shared_memory, sizeof(SharedMemory));
+}
+
+void Controller_Keyboard::OnLoadInputDevices() {}
+} // namespace Service::HID

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

@@ -0,0 +1,49 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/hle/service/hid/controllers/controller_base.h"
+
+namespace Service::HID {
+class Controller_Keyboard final : public ControllerBase {
+public:
+    Controller_Keyboard();
+
+    // Called when the controller is initialized
+    void OnInit() override;
+
+    // When the controller is released
+    void OnRelease() override;
+
+    // When the controller is requesting an update for the shared memory
+    void OnUpdate(u8* data, std::size_t size) override;
+
+    // Called when input devices should be loaded
+    void OnLoadInputDevices() override;
+
+private:
+    struct KeyboardState {
+        s64_le sampling_number;
+        s64_le sampling_number2;
+
+        s32_le modifier;
+        s32_le attribute;
+        std::array<u8, 32> key;
+    };
+    static_assert(sizeof(KeyboardState) == 0x38, "KeyboardState is an invalid size");
+
+    struct SharedMemory {
+        CommonHeader header;
+        std::array<KeyboardState, 17> pad_states;
+        INSERT_PADDING_BYTES(0x28);
+    };
+    static_assert(sizeof(SharedMemory) == 0x400, "SharedMemory is an invalid size");
+    SharedMemory shared_memory{};
+};
+} // namespace Service::HID

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

@@ -0,0 +1,43 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/core_timing.h"
+#include "core/hle/service/hid/controllers/mouse.h"
+
+namespace Service::HID {
+constexpr std::size_t SHARED_MEMORY_OFFSET = 0x3400;
+
+Controller_Mouse::Controller_Mouse() = default;
+
+void Controller_Mouse::OnInit() {}
+
+void Controller_Mouse::OnRelease() {}
+
+void Controller_Mouse::OnUpdate(u8* data, std::size_t size) {
+    shared_memory.header.timestamp = CoreTiming::GetTicks();
+    shared_memory.header.total_entry_count = 17;
+
+    if (!IsControllerActivated()) {
+        shared_memory.header.entry_count = 0;
+        shared_memory.header.last_entry_index = 0;
+        return;
+    }
+    shared_memory.header.entry_count = 16;
+
+    auto& last_entry = shared_memory.mouse_states[shared_memory.header.last_entry_index];
+    shared_memory.header.last_entry_index = (shared_memory.header.last_entry_index + 1) % 17;
+    auto& cur_entry = shared_memory.mouse_states[shared_memory.header.last_entry_index];
+
+    cur_entry.sampling_number = last_entry.sampling_number + 1;
+    cur_entry.sampling_number2 = cur_entry.sampling_number;
+    // TODO(ogniK): Update mouse states
+
+    std::memcpy(data + SHARED_MEMORY_OFFSET, &shared_memory, sizeof(SharedMemory));
+}
+
+void Controller_Mouse::OnLoadInputDevices() {}
+} // namespace Service::HID

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

@@ -0,0 +1,49 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/hle/service/hid/controllers/controller_base.h"
+
+namespace Service::HID {
+class Controller_Mouse final : public ControllerBase {
+public:
+    Controller_Mouse();
+
+    // Called when the controller is initialized
+    void OnInit() override;
+
+    // When the controller is released
+    void OnRelease() override;
+
+    // When the controller is requesting an update for the shared memory
+    void OnUpdate(u8* data, std::size_t size) override;
+
+    // Called when input devices should be loaded
+    void OnLoadInputDevices() override;
+
+private:
+    struct MouseState {
+        s64_le sampling_number;
+        s64_le sampling_number2;
+        s32_le x;
+        s32_le y;
+        s32_le delta_x;
+        s32_le delta_y;
+        s32_le mouse_wheel;
+        s32_le button;
+        s32_le attribute;
+    };
+    static_assert(sizeof(MouseState) == 0x30, "MouseState is an invalid size");
+
+    struct SharedMemory {
+        CommonHeader header;
+        std::array<MouseState, 17> mouse_states;
+    };
+    SharedMemory shared_memory{};
+};
+} // namespace Service::HID

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

@@ -0,0 +1,429 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <algorithm>
+#include <array>
+#include <cstring>
+#include "common/assert.h"
+#include "common/bit_field.h"
+#include "common/common_types.h"
+#include "common/logging/log.h"
+#include "common/swap.h"
+#include "core/core.h"
+#include "core/core_timing.h"
+#include "core/frontend/input.h"
+#include "core/hle/kernel/event.h"
+#include "core/hle/service/hid/controllers/npad.h"
+#include "core/settings.h"
+
+namespace Service::HID {
+constexpr u32 JOYCON_BODY_NEON_RED = 0xFF3C28;
+constexpr u32 JOYCON_BUTTONS_NEON_RED = 0x1E0A0A;
+constexpr u32 JOYCON_BODY_NEON_BLUE = 0x0AB9E6;
+constexpr u32 JOYCON_BUTTONS_NEON_BLUE = 0x001E1E;
+constexpr s32 HID_JOYSTICK_MAX = 0x7fff;
+constexpr s32 HID_JOYSTICK_MIN = -0x7fff;
+constexpr std::size_t NPAD_OFFSET = 0x9A00;
+constexpr u32 BATTERY_FULL = 2;
+enum class JoystickId : std::size_t { Joystick_Left, Joystick_Right };
+
+Controller_NPad::Controller_NPad() = default;
+
+void Controller_NPad::InitNewlyAddedControler(std::size_t controller_idx) {
+    const auto controller_type = connected_controllers[controller_idx].type;
+    auto& controller = shared_memory_entries[controller_idx];
+    if (controller_type == NPadControllerType::None) {
+        return;
+    }
+    controller.joy_styles.raw = 0; // Zero out
+    controller.device_type.raw = 0;
+    switch (controller_type) {
+    case NPadControllerType::Handheld:
+        controller.joy_styles.handheld.Assign(1);
+        controller.device_type.handheld.Assign(1);
+        controller.pad_assignment = NPadAssignments::Dual;
+        break;
+    case NPadControllerType::JoyDual:
+        controller.joy_styles.joycon_dual.Assign(1);
+        controller.device_type.joycon_left.Assign(1);
+        controller.device_type.joycon_right.Assign(1);
+        controller.pad_assignment = NPadAssignments::Dual;
+        break;
+    case NPadControllerType::JoyLeft:
+        controller.joy_styles.joycon_left.Assign(1);
+        controller.device_type.joycon_left.Assign(1);
+        controller.pad_assignment = NPadAssignments::Dual;
+        break;
+    case NPadControllerType::JoyRight:
+        controller.joy_styles.joycon_right.Assign(1);
+        controller.device_type.joycon_right.Assign(1);
+        controller.pad_assignment = NPadAssignments::Dual;
+        break;
+    case NPadControllerType::Pokeball:
+        controller.joy_styles.pokeball.Assign(1);
+        controller.device_type.pokeball.Assign(1);
+        controller.pad_assignment = NPadAssignments::Single;
+        break;
+    case NPadControllerType::ProController:
+        controller.joy_styles.pro_controller.Assign(1);
+        controller.device_type.pro_controller.Assign(1);
+        controller.pad_assignment = NPadAssignments::Single;
+        break;
+    }
+
+    controller.single_color_error = ColorReadError::ReadOk;
+    controller.single_color.body_color = 0;
+    controller.single_color.button_color = 0;
+
+    controller.dual_color_error = ColorReadError::ReadOk;
+    controller.left_color.body_color = JOYCON_BODY_NEON_BLUE;
+    controller.left_color.button_color = JOYCON_BUTTONS_NEON_BLUE;
+    controller.right_color.body_color = JOYCON_BODY_NEON_RED;
+    controller.right_color.button_color = JOYCON_BUTTONS_NEON_RED;
+
+    controller.properties.is_vertical.Assign(1); // TODO(ogniK): Swap joycons orientations
+    controller.properties.use_plus.Assign(1);
+    controller.properties.use_minus.Assign(1);
+    controller.battery_level[0] = BATTERY_FULL;
+    controller.battery_level[1] = BATTERY_FULL;
+    controller.battery_level[2] = BATTERY_FULL;
+}
+
+void Controller_NPad::OnInit() {
+    auto& kernel = Core::System::GetInstance().Kernel();
+    styleset_changed_event =
+        Kernel::Event::Create(kernel, Kernel::ResetType::OneShot, "npad:NpadStyleSetChanged");
+
+    if (!IsControllerActivated())
+        return;
+    std::size_t controller{};
+    if (style.raw == 0) {
+        // We want to support all controllers
+        style.handheld.Assign(1);
+        style.joycon_left.Assign(1);
+        style.joycon_right.Assign(1);
+        style.joycon_dual.Assign(1);
+        style.pro_controller.Assign(1);
+        style.pokeball.Assign(1);
+    }
+    if (std::none_of(connected_controllers.begin(), connected_controllers.end(),
+                     [](const ControllerHolder& controller) { return controller.is_connected; })) {
+        supported_npad_id_types.resize(npad_id_list.size());
+        std::memcpy(supported_npad_id_types.data(), npad_id_list.data(),
+                    npad_id_list.size() * sizeof(u32));
+        AddNewController(NPadControllerType::JoyDual);
+    }
+}
+
+void Controller_NPad::OnLoadInputDevices() {
+    std::transform(Settings::values.buttons.begin() + Settings::NativeButton::BUTTON_HID_BEGIN,
+                   Settings::values.buttons.begin() + Settings::NativeButton::BUTTON_HID_END,
+                   buttons.begin(), Input::CreateDevice<Input::ButtonDevice>);
+    std::transform(Settings::values.analogs.begin() + Settings::NativeAnalog::STICK_HID_BEGIN,
+                   Settings::values.analogs.begin() + Settings::NativeAnalog::STICK_HID_END,
+                   sticks.begin(), Input::CreateDevice<Input::AnalogDevice>);
+}
+
+void Controller_NPad::OnRelease() {}
+
+void Controller_NPad::OnUpdate(u8* data, std::size_t data_len) {
+    if (!IsControllerActivated())
+        return;
+    for (std::size_t i = 0; i < shared_memory_entries.size(); i++) {
+        auto& npad = shared_memory_entries[i];
+        const std::array<NPadGeneric*, 7> controller_npads{&npad.main_controller_states,
+                                                           &npad.handheld_states,
+                                                           &npad.dual_states,
+                                                           &npad.left_joy_states,
+                                                           &npad.right_joy_states,
+                                                           &npad.pokeball_states,
+                                                           &npad.libnx};
+
+        for (auto* main_controller : controller_npads) {
+            main_controller->common.entry_count = 16;
+            main_controller->common.total_entry_count = 17;
+
+            const auto& last_entry =
+                main_controller->npad[main_controller->common.last_entry_index];
+
+            main_controller->common.timestamp = CoreTiming::GetTicks();
+            main_controller->common.last_entry_index =
+                (main_controller->common.last_entry_index + 1) % 17;
+
+            auto& cur_entry = main_controller->npad[main_controller->common.last_entry_index];
+
+            cur_entry.timestamp = last_entry.timestamp + 1;
+            cur_entry.timestamp2 = cur_entry.timestamp;
+        }
+
+        const auto& controller_type = connected_controllers[i].type;
+
+        if (controller_type == NPadControllerType::None || !connected_controllers[i].is_connected) {
+            continue;
+        }
+
+        // Pad states
+        ControllerPadState pad_state{};
+        using namespace Settings::NativeButton;
+        pad_state.a.Assign(buttons[A - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.b.Assign(buttons[B - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.x.Assign(buttons[X - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.y.Assign(buttons[Y - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.l_stick.Assign(buttons[LStick - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.r_stick.Assign(buttons[RStick - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.l.Assign(buttons[L - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.r.Assign(buttons[R - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.zl.Assign(buttons[ZL - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.zr.Assign(buttons[ZR - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.plus.Assign(buttons[Plus - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.minus.Assign(buttons[Minus - BUTTON_HID_BEGIN]->GetStatus());
+
+        pad_state.d_left.Assign(buttons[DLeft - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.d_up.Assign(buttons[DUp - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.d_right.Assign(buttons[DRight - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.d_down.Assign(buttons[DDown - BUTTON_HID_BEGIN]->GetStatus());
+
+        pad_state.l_stick_left.Assign(buttons[LStick_Left - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.l_stick_up.Assign(buttons[LStick_Up - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.l_stick_right.Assign(buttons[LStick_Right - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.l_stick_down.Assign(buttons[LStick_Down - BUTTON_HID_BEGIN]->GetStatus());
+
+        pad_state.r_stick_left.Assign(buttons[RStick_Left - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.r_stick_up.Assign(buttons[RStick_Up - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.r_stick_right.Assign(buttons[RStick_Right - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.r_stick_down.Assign(buttons[RStick_Down - BUTTON_HID_BEGIN]->GetStatus());
+
+        pad_state.sl.Assign(buttons[SL - BUTTON_HID_BEGIN]->GetStatus());
+        pad_state.sr.Assign(buttons[SR - BUTTON_HID_BEGIN]->GetStatus());
+
+        AnalogPosition lstick_entry{};
+        AnalogPosition rstick_entry{};
+
+        const auto [stick_l_x_f, stick_l_y_f] =
+            sticks[static_cast<std::size_t>(JoystickId::Joystick_Left)]->GetStatus();
+        const auto [stick_r_x_f, stick_r_y_f] =
+            sticks[static_cast<std::size_t>(JoystickId::Joystick_Right)]->GetStatus();
+        lstick_entry.x = static_cast<s32>(stick_l_x_f * HID_JOYSTICK_MAX);
+        lstick_entry.y = static_cast<s32>(stick_l_y_f * HID_JOYSTICK_MAX);
+        rstick_entry.x = static_cast<s32>(stick_r_x_f * HID_JOYSTICK_MAX);
+        rstick_entry.y = static_cast<s32>(stick_r_y_f * HID_JOYSTICK_MAX);
+
+        auto& main_controller =
+            npad.main_controller_states.npad[npad.main_controller_states.common.last_entry_index];
+        auto& handheld_entry =
+            npad.handheld_states.npad[npad.handheld_states.common.last_entry_index];
+        auto& dual_entry = npad.dual_states.npad[npad.dual_states.common.last_entry_index];
+        auto& left_entry = npad.left_joy_states.npad[npad.left_joy_states.common.last_entry_index];
+        auto& right_entry =
+            npad.right_joy_states.npad[npad.right_joy_states.common.last_entry_index];
+        auto& pokeball_entry =
+            npad.pokeball_states.npad[npad.pokeball_states.common.last_entry_index];
+        auto& libnx_entry = npad.libnx.npad[npad.libnx.common.last_entry_index];
+
+        if (hold_type == NpadHoldType::Horizontal) {
+            // TODO(ogniK): Remap buttons for different orientations
+        }
+        libnx_entry.connection_status.raw = 0;
+
+        switch (controller_type) {
+        case NPadControllerType::Handheld:
+            handheld_entry.connection_status.raw = 0;
+            handheld_entry.connection_status.IsConnected.Assign(1);
+            if (!Settings::values.use_docked_mode) {
+                handheld_entry.connection_status.IsWired.Assign(1);
+            }
+            handheld_entry.pad_states.raw = pad_state.raw;
+            handheld_entry.l_stick = lstick_entry;
+            handheld_entry.r_stick = rstick_entry;
+            break;
+        case NPadControllerType::JoyDual:
+            dual_entry.connection_status.raw = 0;
+
+            dual_entry.connection_status.IsLeftJoyConnected.Assign(1);
+            dual_entry.connection_status.IsRightJoyConnected.Assign(1);
+            dual_entry.connection_status.IsConnected.Assign(1);
+
+            libnx_entry.connection_status.IsLeftJoyConnected.Assign(1);
+            libnx_entry.connection_status.IsRightJoyConnected.Assign(1);
+            libnx_entry.connection_status.IsConnected.Assign(1);
+
+            dual_entry.pad_states.raw = pad_state.raw;
+            dual_entry.l_stick = lstick_entry;
+            dual_entry.r_stick = rstick_entry;
+        case NPadControllerType::JoyLeft:
+            left_entry.connection_status.raw = 0;
+
+            left_entry.connection_status.IsConnected.Assign(1);
+            left_entry.pad_states.raw = pad_state.raw;
+            left_entry.l_stick = lstick_entry;
+            left_entry.r_stick = rstick_entry;
+            break;
+        case NPadControllerType::JoyRight:
+            right_entry.connection_status.raw = 0;
+
+            right_entry.connection_status.IsConnected.Assign(1);
+            right_entry.pad_states.raw = pad_state.raw;
+            right_entry.l_stick = lstick_entry;
+            right_entry.r_stick = rstick_entry;
+            break;
+        case NPadControllerType::Pokeball:
+            pokeball_entry.connection_status.raw = 0;
+
+            pokeball_entry.connection_status.IsConnected.Assign(1);
+            pokeball_entry.connection_status.IsWired.Assign(1);
+
+            pokeball_entry.pad_states.raw = pad_state.raw;
+            pokeball_entry.l_stick = lstick_entry;
+            pokeball_entry.r_stick = rstick_entry;
+            break;
+        case NPadControllerType::ProController:
+            main_controller.connection_status.raw = 0;
+
+            main_controller.connection_status.IsConnected.Assign(1);
+            main_controller.connection_status.IsWired.Assign(1);
+            main_controller.pad_states.raw = pad_state.raw;
+            main_controller.l_stick = lstick_entry;
+            main_controller.r_stick = rstick_entry;
+            break;
+        }
+
+        // LibNX exclusively uses this section, so we always update it since LibNX doesn't activate
+        // any controllers.
+        libnx_entry.pad_states.raw = pad_state.raw;
+        libnx_entry.l_stick = lstick_entry;
+        libnx_entry.r_stick = rstick_entry;
+    }
+    std::memcpy(data + NPAD_OFFSET, shared_memory_entries.data(),
+                shared_memory_entries.size() * sizeof(NPadEntry));
+} // namespace Service::HID
+
+void Controller_NPad::SetSupportedStyleSet(NPadType style_set) {
+    style.raw = style_set.raw;
+}
+
+Controller_NPad::NPadType Controller_NPad::GetSupportedStyleSet() const {
+    return style;
+}
+
+void Controller_NPad::SetSupportedNPadIdTypes(u8* data, std::size_t length) {
+    ASSERT(length > 0 && (length % sizeof(u32)) == 0);
+    supported_npad_id_types.clear();
+    supported_npad_id_types.resize(length / sizeof(u32));
+    std::memcpy(supported_npad_id_types.data(), data, length);
+}
+
+const void Controller_NPad::GetSupportedNpadIdTypes(u32* data, std::size_t max_length) {
+    ASSERT(max_length < supported_npad_id_types.size());
+    std::memcpy(data, supported_npad_id_types.data(), supported_npad_id_types.size());
+}
+
+std::size_t Controller_NPad::GetSupportedNPadIdTypesSize() const {
+    return supported_npad_id_types.size();
+}
+
+void Controller_NPad::SetHoldType(NpadHoldType joy_hold_type) {
+    hold_type = joy_hold_type;
+}
+Controller_NPad::NpadHoldType Controller_NPad::GetHoldType() const {
+    return hold_type;
+}
+
+void Controller_NPad::SetNpadMode(u32 npad_id, NPadAssignments assignment_mode) {
+    ASSERT(npad_id < shared_memory_entries.size());
+    shared_memory_entries[npad_id].pad_assignment = assignment_mode;
+}
+
+void Controller_NPad::VibrateController(const std::vector<u32>& controller_ids,
+                                        const std::vector<Vibration>& vibrations) {
+    if (!can_controllers_vibrate) {
+        return;
+    }
+    for (std::size_t i = 0; i < controller_ids.size(); i++) {
+        std::size_t controller_pos = i;
+        // Handheld controller conversion
+        if (controller_pos == 32) {
+            controller_pos = 8;
+        }
+        // Unknown controller conversion
+        if (controller_pos == 16) {
+            controller_pos = 9;
+        }
+        if (connected_controllers[controller_pos].is_connected) {
+            // TODO(ogniK): Vibrate the physical controller
+        }
+    }
+    LOG_WARNING(Service_HID, "(STUBBED) called");
+    last_processed_vibration = vibrations.back();
+}
+
+Kernel::SharedPtr<Kernel::Event> Controller_NPad::GetStyleSetChangedEvent() const {
+    return styleset_changed_event;
+}
+
+Controller_NPad::Vibration Controller_NPad::GetLastVibration() const {
+    return last_processed_vibration;
+}
+void Controller_NPad::AddNewController(NPadControllerType controller) {
+    if (controller == NPadControllerType::Handheld) {
+        connected_controllers[8] = {controller, true};
+        InitNewlyAddedControler(8);
+        return;
+    }
+    const auto pos =
+        std::find_if(connected_controllers.begin(), connected_controllers.end() - 2,
+                     [](const ControllerHolder& holder) { return !holder.is_connected; });
+    if (pos == connected_controllers.end() - 2) {
+        LOG_ERROR(Service_HID, "Cannot connect any more controllers!");
+        return;
+    }
+    const auto controller_id = std::distance(connected_controllers.begin(), pos);
+    connected_controllers[controller_id] = {controller, true};
+    InitNewlyAddedControler(controller_id);
+}
+
+void Controller_NPad::ConnectNPad(u32 npad_id) {
+    if (npad_id >= connected_controllers.size())
+        return;
+    connected_controllers[npad_id].is_connected = true;
+}
+
+void Controller_NPad::DisconnectNPad(u32 npad_id) {
+    if (npad_id >= connected_controllers.size())
+        return;
+    connected_controllers[npad_id].is_connected = false;
+}
+
+Controller_NPad::LedPattern Controller_NPad::GetLedPattern(u32 npad_id) {
+    if (npad_id == npad_id_list.back() || npad_id == npad_id_list[npad_id_list.size() - 2]) {
+        // These are controllers without led patterns
+        return LedPattern{0, 0, 0, 0};
+    }
+    switch (npad_id) {
+    case 0:
+        return LedPattern{1, 0, 0, 0};
+    case 1:
+        return LedPattern{0, 1, 0, 0};
+    case 2:
+        return LedPattern{0, 0, 1, 0};
+    case 3:
+        return LedPattern{0, 0, 0, 1};
+    case 4:
+        return LedPattern{1, 0, 0, 1};
+    case 5:
+        return LedPattern{1, 0, 1, 0};
+    case 6:
+        return LedPattern{1, 0, 1, 1};
+    case 7:
+        return LedPattern{0, 1, 1, 0};
+    default:
+        UNIMPLEMENTED_MSG("Unhandled npad_id {}", npad_id);
+        return LedPattern{0, 0, 0, 0};
+    };
+}
+void Controller_NPad::SetVibrationEnabled(bool can_vibrate) {
+    can_controllers_vibrate = can_vibrate;
+}
+} // namespace Service::HID

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

@@ -0,0 +1,288 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include "common/common_types.h"
+#include "core/frontend/input.h"
+#include "core/hle/service/hid/controllers/controller_base.h"
+#include "core/settings.h"
+
+namespace Service::HID {
+
+class Controller_NPad final : public ControllerBase {
+public:
+    Controller_NPad();
+
+    // Called when the controller is initialized
+    void OnInit() override;
+
+    // When the controller is released
+    void OnRelease() override;
+
+    // When the controller is requesting an update for the shared memory
+    void OnUpdate(u8* data, std::size_t size) override;
+
+    // Called when input devices should be loaded
+    void OnLoadInputDevices() override;
+
+    struct NPadType {
+        union {
+            u32_le raw{};
+
+            BitField<0, 1, u32_le> pro_controller;
+            BitField<1, 1, u32_le> handheld;
+            BitField<2, 1, u32_le> joycon_dual;
+            BitField<3, 1, u32_le> joycon_left;
+            BitField<4, 1, u32_le> joycon_right;
+
+            BitField<6, 1, u32_le> pokeball; // TODO(ogniK): Confirm when possible
+        };
+    };
+    static_assert(sizeof(NPadType) == 4, "NPadType is an invalid size");
+
+    struct Vibration {
+        f32 amp_low;
+        f32 freq_low;
+        f32 amp_high;
+        f32 freq_high;
+    };
+    static_assert(sizeof(Vibration) == 0x10, "Vibration is an invalid size");
+
+    enum class NpadHoldType : u64 {
+        Vertical = 0,
+        Horizontal = 1,
+    };
+
+    enum class NPadAssignments : u32_le {
+        Dual = 0,
+        Single = 1,
+    };
+
+    enum class NPadControllerType {
+        None,
+        ProController,
+        Handheld,
+        JoyDual,
+        JoyLeft,
+        JoyRight,
+        Pokeball,
+    };
+
+    struct LedPattern {
+        explicit LedPattern(u64 light1, u64 light2, u64 light3, u64 light4) {
+            position1.Assign(light1);
+            position1.Assign(light2);
+            position1.Assign(light3);
+            position1.Assign(light4);
+        };
+        union {
+            u64 raw{};
+            BitField<0, 1, u64> position1;
+            BitField<1, 1, u64> position2;
+            BitField<2, 1, u64> position3;
+            BitField<3, 1, u64> position4;
+        };
+    };
+
+    void SetSupportedStyleSet(NPadType style_set);
+    NPadType GetSupportedStyleSet() const;
+
+    void SetSupportedNPadIdTypes(u8* data, std::size_t length);
+    const void GetSupportedNpadIdTypes(u32* data, std::size_t max_length);
+    std::size_t GetSupportedNPadIdTypesSize() const;
+
+    void SetHoldType(NpadHoldType joy_hold_type);
+    NpadHoldType GetHoldType() const;
+
+    void SetNpadMode(u32 npad_id, NPadAssignments assignment_mode);
+
+    void VibrateController(const std::vector<u32>& controller_ids,
+                           const std::vector<Vibration>& vibrations);
+
+    Kernel::SharedPtr<Kernel::Event> GetStyleSetChangedEvent() const;
+    Vibration GetLastVibration() const;
+
+    void AddNewController(NPadControllerType controller);
+
+    void ConnectNPad(u32 npad_id);
+    void DisconnectNPad(u32 npad_id);
+    LedPattern GetLedPattern(u32 npad_id);
+    void SetVibrationEnabled(bool can_vibrate);
+
+private:
+    struct CommonHeader {
+        s64_le timestamp;
+        s64_le total_entry_count;
+        s64_le last_entry_index;
+        s64_le entry_count;
+    };
+    static_assert(sizeof(CommonHeader) == 0x20, "CommonHeader is an invalid size");
+
+    struct ControllerColor {
+        u32_le body_color;
+        u32_le button_color;
+    };
+    static_assert(sizeof(ControllerColor) == 8, "ControllerColor is an invalid size");
+
+    struct ControllerPadState {
+        union {
+            u64_le raw{};
+            // Button states
+            BitField<0, 1, u64_le> a;
+            BitField<1, 1, u64_le> b;
+            BitField<2, 1, u64_le> x;
+            BitField<3, 1, u64_le> y;
+            BitField<4, 1, u64_le> l_stick;
+            BitField<5, 1, u64_le> r_stick;
+            BitField<6, 1, u64_le> l;
+            BitField<7, 1, u64_le> r;
+            BitField<8, 1, u64_le> zl;
+            BitField<9, 1, u64_le> zr;
+            BitField<10, 1, u64_le> plus;
+            BitField<11, 1, u64_le> minus;
+
+            // D-Pad
+            BitField<12, 1, u64_le> d_left;
+            BitField<13, 1, u64_le> d_up;
+            BitField<14, 1, u64_le> d_right;
+            BitField<15, 1, u64_le> d_down;
+
+            // Left JoyStick
+            BitField<16, 1, u64_le> l_stick_left;
+            BitField<17, 1, u64_le> l_stick_up;
+            BitField<18, 1, u64_le> l_stick_right;
+            BitField<19, 1, u64_le> l_stick_down;
+
+            // Right JoyStick
+            BitField<20, 1, u64_le> r_stick_left;
+            BitField<21, 1, u64_le> r_stick_up;
+            BitField<22, 1, u64_le> r_stick_right;
+            BitField<23, 1, u64_le> r_stick_down;
+
+            // Not always active?
+            BitField<24, 1, u64_le> sl;
+            BitField<25, 1, u64_le> sr;
+        };
+    };
+    static_assert(sizeof(ControllerPadState) == 8, "ControllerPadState is an invalid size");
+
+    struct AnalogPosition {
+        s32_le x;
+        s32_le y;
+    };
+    static_assert(sizeof(AnalogPosition) == 8, "AnalogPosition is an invalid size");
+
+    struct ConnectionState {
+        union {
+            u32_le raw{};
+            BitField<0, 1, u32_le> IsConnected;
+            BitField<1, 1, u32_le> IsWired;
+            BitField<2, 1, u32_le> IsLeftJoyConnected;
+            BitField<3, 1, u32_le> IsLeftJoyWired;
+            BitField<4, 1, u32_le> IsRightJoyConnected;
+            BitField<5, 1, u32_le> IsRightJoyWired;
+        };
+    };
+    static_assert(sizeof(ConnectionState) == 4, "ConnectionState is an invalid size");
+
+    struct GenericStates {
+        s64_le timestamp;
+        s64_le timestamp2;
+        ControllerPadState pad_states;
+        AnalogPosition l_stick;
+        AnalogPosition r_stick;
+        ConnectionState connection_status;
+    };
+    static_assert(sizeof(GenericStates) == 0x30, "NPadGenericStates is an invalid size");
+
+    struct NPadGeneric {
+        CommonHeader common;
+        std::array<GenericStates, 17> npad;
+    };
+    static_assert(sizeof(NPadGeneric) == 0x350, "NPadGeneric is an invalid size");
+
+    enum class ColorReadError : u32_le {
+        ReadOk = 0,
+        ColorDoesntExist = 1,
+        NoController = 2,
+    };
+
+    struct NPadProperties {
+        union {
+            s64_le raw{};
+            BitField<11, 1, s64_le> is_vertical;
+            BitField<12, 1, s64_le> is_horizontal;
+            BitField<13, 1, s64_le> use_plus;
+            BitField<14, 1, s64_le> use_minus;
+        };
+    };
+
+    struct NPadDevice {
+        union {
+            u32_le raw{};
+            BitField<0, 1, s32_le> pro_controller;
+            BitField<1, 1, s32_le> handheld;
+            BitField<2, 1, s32_le> handheld_left;
+            BitField<3, 1, s32_le> handheld_right;
+            BitField<4, 1, s32_le> joycon_left;
+            BitField<5, 1, s32_le> joycon_right;
+            BitField<6, 1, s32_le> pokeball;
+        };
+    };
+
+    struct NPadEntry {
+        NPadType joy_styles;
+        NPadAssignments pad_assignment;
+
+        ColorReadError single_color_error;
+        ControllerColor single_color;
+
+        ColorReadError dual_color_error;
+        ControllerColor left_color;
+        ControllerColor right_color;
+
+        NPadGeneric main_controller_states;
+        NPadGeneric handheld_states;
+        NPadGeneric dual_states;
+        NPadGeneric left_joy_states;
+        NPadGeneric right_joy_states;
+        NPadGeneric pokeball_states;
+        NPadGeneric libnx; // TODO(ogniK): Find out what this actually is, libnx seems to only be
+                           // relying on this for the time being
+        INSERT_PADDING_BYTES(
+            0x708 *
+            6); // TODO(ogniK): SixAxis states, require more information before implementation
+        NPadDevice device_type;
+        NPadProperties properties;
+        INSERT_PADDING_WORDS(1);
+        std::array<u32, 3> battery_level;
+        INSERT_PADDING_BYTES(0x5c);
+        INSERT_PADDING_BYTES(0xdf8);
+    };
+    static_assert(sizeof(NPadEntry) == 0x5000, "NPadEntry is an invalid size");
+
+    struct ControllerHolder {
+        Controller_NPad::NPadControllerType type;
+        bool is_connected;
+    };
+
+    NPadType style{};
+    std::array<NPadEntry, 10> shared_memory_entries{};
+    std::array<std::unique_ptr<Input::ButtonDevice>, Settings::NativeButton::NUM_BUTTONS_HID>
+        buttons;
+    std::array<std::unique_ptr<Input::AnalogDevice>, Settings::NativeAnalog::NUM_STICKS_HID> sticks;
+    std::vector<u32> supported_npad_id_types{};
+    NpadHoldType hold_type{NpadHoldType::Vertical};
+    Kernel::SharedPtr<Kernel::Event> styleset_changed_event;
+    std::size_t dump_idx{};
+    Vibration last_processed_vibration{};
+    static constexpr std::array<u32, 10> npad_id_list{0, 1, 2, 3, 4, 5, 6, 7, 32, 16};
+    std::array<ControllerHolder, 10> connected_controllers{};
+    bool can_controllers_vibrate{true};
+
+    void InitNewlyAddedControler(std::size_t controller_idx);
+};
+} // namespace Service::HID

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

@@ -0,0 +1,39 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/core_timing.h"
+#include "core/hle/service/hid/controllers/stubbed.h"
+
+namespace Service::HID {
+
+Controller_Stubbed::Controller_Stubbed() = default;
+
+void Controller_Stubbed::OnInit() {}
+
+void Controller_Stubbed::OnRelease() {}
+
+void Controller_Stubbed::OnUpdate(u8* data, std::size_t size) {
+    if (!smart_update) {
+        return;
+    }
+
+    CommonHeader header{};
+    header.timestamp = CoreTiming::GetTicks();
+    header.total_entry_count = 17;
+    header.entry_count = 0;
+    header.last_entry_index = 0;
+
+    std::memcpy(data + common_offset, &header, sizeof(CommonHeader));
+}
+
+void Controller_Stubbed::OnLoadInputDevices() {}
+
+void Controller_Stubbed::SetCommonHeaderOffset(std::size_t off) {
+    common_offset = off;
+    smart_update = true;
+}
+} // namespace Service::HID

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

@@ -0,0 +1,33 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_types.h"
+#include "core/hle/service/hid/controllers/controller_base.h"
+
+namespace Service::HID {
+class Controller_Stubbed final : public ControllerBase {
+public:
+    Controller_Stubbed();
+
+    // Called when the controller is initialized
+    void OnInit() override;
+
+    // When the controller is released
+    void OnRelease() override;
+
+    // When the controller is requesting an update for the shared memory
+    void OnUpdate(u8* data, std::size_t size) override;
+
+    // Called when input devices should be loaded
+    void OnLoadInputDevices() override;
+
+    void SetCommonHeaderOffset(std::size_t off);
+
+private:
+    bool smart_update{};
+    std::size_t common_offset{};
+};
+} // namespace Service::HID

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

@@ -0,0 +1,65 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/core_timing.h"
+#include "core/frontend/emu_window.h"
+#include "core/frontend/input.h"
+#include "core/hle/service/hid/controllers/touchscreen.h"
+#include "core/settings.h"
+
+namespace Service::HID {
+constexpr std::size_t SHARED_MEMORY_OFFSET = 0x400;
+
+Controller_Touchscreen::Controller_Touchscreen() = default;
+
+void Controller_Touchscreen::OnInit() {}
+
+void Controller_Touchscreen::OnRelease() {}
+
+void Controller_Touchscreen::OnUpdate(u8* data, std::size_t size) {
+    shared_memory.header.timestamp = CoreTiming::GetTicks();
+    shared_memory.header.total_entry_count = 17;
+
+    if (!IsControllerActivated()) {
+        shared_memory.header.entry_count = 0;
+        shared_memory.header.last_entry_index = 0;
+        return;
+    }
+    shared_memory.header.entry_count = 16;
+
+    const auto& last_entry =
+        shared_memory.shared_memory_entries[shared_memory.header.last_entry_index];
+    shared_memory.header.last_entry_index = (shared_memory.header.last_entry_index + 1) % 17;
+    auto& cur_entry = shared_memory.shared_memory_entries[shared_memory.header.last_entry_index];
+
+    cur_entry.sampling_number = last_entry.sampling_number + 1;
+    cur_entry.sampling_number2 = cur_entry.sampling_number;
+
+    const auto [x, y, pressed] = touch_device->GetStatus();
+    auto& touch_entry = cur_entry.states[0];
+    if (pressed) {
+        touch_entry.x = static_cast<u16>(x * Layout::ScreenUndocked::Width);
+        touch_entry.y = static_cast<u16>(y * Layout::ScreenUndocked::Height);
+        touch_entry.diameter_x = 15;
+        touch_entry.diameter_y = 15;
+        touch_entry.rotation_angle = 0;
+        const u64 tick = CoreTiming::GetTicks();
+        touch_entry.delta_time = tick - last_touch;
+        last_touch = tick;
+        touch_entry.finger = 0;
+        cur_entry.entry_count = 1;
+    } else {
+        cur_entry.entry_count = 0;
+    }
+
+    std::memcpy(data + SHARED_MEMORY_OFFSET, &shared_memory, sizeof(TouchScreenSharedMemory));
+}
+
+void Controller_Touchscreen::OnLoadInputDevices() {
+    touch_device = Input::CreateDevice<Input::TouchDevice>(Settings::values.touch_device);
+}
+} // namespace Service::HID

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

@@ -0,0 +1,62 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/frontend/input.h"
+#include "core/hle/service/hid/controllers/controller_base.h"
+
+namespace Service::HID {
+class Controller_Touchscreen final : public ControllerBase {
+public:
+    Controller_Touchscreen();
+
+    // Called when the controller is initialized
+    void OnInit() override;
+
+    // When the controller is released
+    void OnRelease() override;
+
+    // When the controller is requesting an update for the shared memory
+    void OnUpdate(u8* data, std::size_t size) override;
+
+    // Called when input devices should be loaded
+    void OnLoadInputDevices() override;
+
+private:
+    struct TouchState {
+        u64_le delta_time;
+        u32_le attribute;
+        u32_le finger;
+        u32_le x;
+        u32_le y;
+        u32_le diameter_x;
+        u32_le diameter_y;
+        u32_le rotation_angle;
+    };
+    static_assert(sizeof(TouchState) == 0x28, "Touchstate is an invalid size");
+
+    struct TouchScreenEntry {
+        s64_le sampling_number;
+        s64_le sampling_number2;
+        s32_le entry_count;
+        std::array<TouchState, 16> states;
+    };
+    static_assert(sizeof(TouchScreenEntry) == 0x298, "TouchScreenEntry is an invalid size");
+
+    struct TouchScreenSharedMemory {
+        CommonHeader header;
+        std::array<TouchScreenEntry, 17> shared_memory_entries{};
+        INSERT_PADDING_BYTES(0x3c8);
+    };
+    static_assert(sizeof(TouchScreenSharedMemory) == 0x3000,
+                  "TouchScreenSharedMemory is an invalid size");
+    TouchScreenSharedMemory shared_memory{};
+    std::unique_ptr<Input::TouchDevice> touch_device;
+    s64_le last_touch{};
+};
+} // namespace Service::HID

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

@@ -0,0 +1,45 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/core_timing.h"
+#include "core/hle/service/hid/controllers/xpad.h"
+
+namespace Service::HID {
+constexpr std::size_t SHARED_MEMORY_OFFSET = 0x3C00;
+
+Controller_XPad::Controller_XPad() = default;
+
+void Controller_XPad::OnInit() {}
+
+void Controller_XPad::OnRelease() {}
+
+void Controller_XPad::OnUpdate(u8* data, std::size_t size) {
+    for (auto& xpad_entry : shared_memory.shared_memory_entries) {
+        xpad_entry.header.timestamp = CoreTiming::GetTicks();
+        xpad_entry.header.total_entry_count = 17;
+
+        if (!IsControllerActivated()) {
+            xpad_entry.header.entry_count = 0;
+            xpad_entry.header.last_entry_index = 0;
+            return;
+        }
+        xpad_entry.header.entry_count = 16;
+
+        const auto& last_entry = xpad_entry.pad_states[xpad_entry.header.last_entry_index];
+        xpad_entry.header.last_entry_index = (xpad_entry.header.last_entry_index + 1) % 17;
+        auto& cur_entry = xpad_entry.pad_states[xpad_entry.header.last_entry_index];
+
+        cur_entry.sampling_number = last_entry.sampling_number + 1;
+        cur_entry.sampling_number2 = cur_entry.sampling_number;
+    }
+    // TODO(ogniK): Update xpad states
+
+    std::memcpy(data + SHARED_MEMORY_OFFSET, &shared_memory, sizeof(SharedMemory));
+}
+
+void Controller_XPad::OnLoadInputDevices() {}
+} // namespace Service::HID

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

@@ -0,0 +1,60 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/frontend/input.h"
+#include "core/hle/service/hid/controllers/controller_base.h"
+
+namespace Service::HID {
+class Controller_XPad final : public ControllerBase {
+public:
+    Controller_XPad();
+
+    // Called when the controller is initialized
+    void OnInit() override;
+
+    // When the controller is released
+    void OnRelease() override;
+
+    // When the controller is requesting an update for the shared memory
+    void OnUpdate(u8* data, std::size_t size) override;
+
+    // Called when input devices should be loaded
+    void OnLoadInputDevices() override;
+
+private:
+    struct AnalogStick {
+        s32_le x;
+        s32_le y;
+    };
+    static_assert(sizeof(AnalogStick) == 0x8, "AnalogStick is an invalid size");
+
+    struct XPadState {
+        s64_le sampling_number;
+        s64_le sampling_number2;
+        s32_le attributes;
+        u32_le pad_states;
+        AnalogStick x_stick;
+        AnalogStick y_stick;
+    };
+    static_assert(sizeof(XPadState) == 0x28, "XPadState is an invalid size");
+
+    struct XPadEntry {
+        CommonHeader header;
+        std::array<XPadState, 17> pad_states{};
+        INSERT_PADDING_BYTES(0x138);
+    };
+    static_assert(sizeof(XPadEntry) == 0x400, "XPadEntry is an invalid size");
+
+    struct SharedMemory {
+        std::array<XPadEntry, 4> shared_memory_entries{};
+    };
+    static_assert(sizeof(SharedMemory) == 0x1000, "SharedMemory is an invalid size");
+    SharedMemory shared_memory{};
+};
+} // namespace Service::HID

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

@@ -2,6 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <array>
+#include "common/common_types.h"
 #include "common/logging/log.h"
 #include "core/core.h"
 #include "core/core_timing.h"
@@ -19,6 +21,16 @@
 #include "core/hle/service/service.h"
 #include "core/settings.h"
 
+#include "core/hle/service/hid/controllers/controller_base.h"
+#include "core/hle/service/hid/controllers/debug_pad.h"
+#include "core/hle/service/hid/controllers/gesture.h"
+#include "core/hle/service/hid/controllers/keyboard.h"
+#include "core/hle/service/hid/controllers/mouse.h"
+#include "core/hle/service/hid/controllers/npad.h"
+#include "core/hle/service/hid/controllers/stubbed.h"
+#include "core/hle/service/hid/controllers/touchscreen.h"
+#include "core/hle/service/hid/controllers/xpad.h"
+
 namespace Service::HID {
 
 // Updating period for each HID device.
@@ -26,6 +38,22 @@ namespace Service::HID {
 constexpr u64 pad_update_ticks = CoreTiming::BASE_CLOCK_RATE / 100;
 constexpr u64 accelerometer_update_ticks = CoreTiming::BASE_CLOCK_RATE / 100;
 constexpr u64 gyroscope_update_ticks = CoreTiming::BASE_CLOCK_RATE / 100;
+constexpr std::size_t SHARED_MEMORY_SIZE = 0x40000;
+enum class HidController : std::size_t {
+    DebugPad,
+    Touchscreen,
+    Mouse,
+    Keyboard,
+    XPad,
+    Unknown1,
+    Unknown2,
+    Unknown3,
+    SixAxisSensor,
+    NPad,
+    Gesture,
+
+    MaxControllers,
+};
 
 class IAppletResource final : public ServiceFramework<IAppletResource> {
 public:
@@ -37,19 +65,57 @@ public:
 
         auto& kernel = Core::System::GetInstance().Kernel();
         shared_mem = Kernel::SharedMemory::Create(
-            kernel, nullptr, 0x40000, Kernel::MemoryPermission::ReadWrite,
+            kernel, nullptr, SHARED_MEMORY_SIZE, Kernel::MemoryPermission::ReadWrite,
             Kernel::MemoryPermission::Read, 0, Kernel::MemoryRegion::BASE, "HID:SharedMemory");
 
+        MakeController<Controller_DebugPad>(HidController::DebugPad);
+        MakeController<Controller_Touchscreen>(HidController::Touchscreen);
+        MakeController<Controller_Mouse>(HidController::Mouse);
+        MakeController<Controller_Keyboard>(HidController::Keyboard);
+        MakeController<Controller_XPad>(HidController::XPad);
+        MakeController<Controller_Stubbed>(HidController::Unknown1);
+        MakeController<Controller_Stubbed>(HidController::Unknown2);
+        MakeController<Controller_Stubbed>(HidController::Unknown3);
+        MakeController<Controller_Stubbed>(HidController::SixAxisSensor);
+        MakeController<Controller_NPad>(HidController::NPad);
+        MakeController<Controller_Gesture>(HidController::Gesture);
+
+        // Homebrew doesn't try to activate some controllers, so we activate them by default
+        GetController<Controller_NPad>(HidController::NPad).ActivateController();
+        GetController<Controller_Touchscreen>(HidController::Touchscreen).ActivateController();
+
+        GetController<Controller_Stubbed>(HidController::Unknown1).SetCommonHeaderOffset(0x4c00);
+        GetController<Controller_Stubbed>(HidController::Unknown2).SetCommonHeaderOffset(0x4e00);
+        GetController<Controller_Stubbed>(HidController::Unknown3).SetCommonHeaderOffset(0x5000);
+
         // Register update callbacks
         pad_update_event = CoreTiming::RegisterEvent(
             "HID::UpdatePadCallback",
-            [this](u64 userdata, int cycles_late) { UpdatePadCallback(userdata, cycles_late); });
+            [this](u64 userdata, int cycles_late) { UpdateControllers(userdata, cycles_late); });
 
         // TODO(shinyquagsire23): Other update callbacks? (accel, gyro?)
 
         CoreTiming::ScheduleEvent(pad_update_ticks, pad_update_event);
     }
 
+    void ActivateController(HidController controller) {
+        controllers[static_cast<size_t>(controller)]->ActivateController();
+    }
+
+    void DeactivateController(HidController controller) {
+        controllers[static_cast<size_t>(controller)]->DeactivateController();
+    }
+
+    template <typename T>
+    void MakeController(HidController controller) {
+        controllers[static_cast<std::size_t>(controller)] = std::make_unique<T>();
+    }
+
+    template <typename T>
+    T& GetController(HidController controller) {
+        return static_cast<T&>(*controllers[static_cast<size_t>(controller)]);
+    }
+
     ~IAppletResource() {
         CoreTiming::UnscheduleEvent(pad_update_event, 0);
     }
@@ -62,200 +128,15 @@ private:
         LOG_DEBUG(Service_HID, "called");
     }
 
-    void LoadInputDevices() {
-        std::transform(Settings::values.buttons.begin() + Settings::NativeButton::BUTTON_HID_BEGIN,
-                       Settings::values.buttons.begin() + Settings::NativeButton::BUTTON_HID_END,
-                       buttons.begin(), Input::CreateDevice<Input::ButtonDevice>);
-        std::transform(Settings::values.analogs.begin() + Settings::NativeAnalog::STICK_HID_BEGIN,
-                       Settings::values.analogs.begin() + Settings::NativeAnalog::STICK_HID_END,
-                       sticks.begin(), Input::CreateDevice<Input::AnalogDevice>);
-        touch_device = Input::CreateDevice<Input::TouchDevice>(Settings::values.touch_device);
-        // TODO(shinyquagsire23): gyro, mouse, keyboard
-    }
-
-    void UpdatePadCallback(u64 userdata, int cycles_late) {
-        SharedMemory mem{};
-        std::memcpy(&mem, shared_mem->GetPointer(), sizeof(SharedMemory));
-
-        if (Settings::values.is_device_reload_pending.exchange(false))
-            LoadInputDevices();
-
-        // Set up controllers as neon red+blue Joy-Con attached to console
-        ControllerHeader& controller_header = mem.controllers[Controller_Handheld].header;
-        controller_header.type = ControllerType_Handheld;
-        controller_header.single_colors_descriptor = ColorDesc_ColorsNonexistent;
-        controller_header.right_color_body = JOYCON_BODY_NEON_RED;
-        controller_header.right_color_buttons = JOYCON_BUTTONS_NEON_RED;
-        controller_header.left_color_body = JOYCON_BODY_NEON_BLUE;
-        controller_header.left_color_buttons = JOYCON_BUTTONS_NEON_BLUE;
-
-        for (std::size_t controller = 0; controller < mem.controllers.size(); controller++) {
-            for (auto& layout : mem.controllers[controller].layouts) {
-                layout.header.num_entries = HID_NUM_ENTRIES;
-                layout.header.max_entry_index = HID_NUM_ENTRIES - 1;
-
-                // HID shared memory stores the state of the past 17 samples in a circlular buffer,
-                // each with a timestamp in number of samples since boot.
-                const ControllerInputEntry& last_entry = layout.entries[layout.header.latest_entry];
-
-                layout.header.timestamp_ticks = CoreTiming::GetTicks();
-                layout.header.latest_entry = (layout.header.latest_entry + 1) % HID_NUM_ENTRIES;
-
-                ControllerInputEntry& entry = layout.entries[layout.header.latest_entry];
-                entry.timestamp = last_entry.timestamp + 1;
-                // TODO(shinyquagsire23): Is this always identical to timestamp?
-                entry.timestamp_2 = entry.timestamp;
-
-                // TODO(shinyquagsire23): More than just handheld input
-                if (controller != Controller_Handheld)
-                    continue;
-
-                entry.connection_state = ConnectionState_Connected | ConnectionState_Wired;
-
-                // TODO(shinyquagsire23): Set up some LUTs for each layout mapping in the future?
-                // For now everything is just the default handheld layout, but split Joy-Con will
-                // rotate the face buttons and directions for certain layouts.
-                ControllerPadState& state = entry.buttons;
-                using namespace Settings::NativeButton;
-                state.a.Assign(buttons[A - BUTTON_HID_BEGIN]->GetStatus());
-                state.b.Assign(buttons[B - BUTTON_HID_BEGIN]->GetStatus());
-                state.x.Assign(buttons[X - BUTTON_HID_BEGIN]->GetStatus());
-                state.y.Assign(buttons[Y - BUTTON_HID_BEGIN]->GetStatus());
-                state.lstick.Assign(buttons[LStick - BUTTON_HID_BEGIN]->GetStatus());
-                state.rstick.Assign(buttons[RStick - BUTTON_HID_BEGIN]->GetStatus());
-                state.l.Assign(buttons[L - BUTTON_HID_BEGIN]->GetStatus());
-                state.r.Assign(buttons[R - BUTTON_HID_BEGIN]->GetStatus());
-                state.zl.Assign(buttons[ZL - BUTTON_HID_BEGIN]->GetStatus());
-                state.zr.Assign(buttons[ZR - BUTTON_HID_BEGIN]->GetStatus());
-                state.plus.Assign(buttons[Plus - BUTTON_HID_BEGIN]->GetStatus());
-                state.minus.Assign(buttons[Minus - BUTTON_HID_BEGIN]->GetStatus());
-
-                state.dleft.Assign(buttons[DLeft - BUTTON_HID_BEGIN]->GetStatus());
-                state.dup.Assign(buttons[DUp - BUTTON_HID_BEGIN]->GetStatus());
-                state.dright.Assign(buttons[DRight - BUTTON_HID_BEGIN]->GetStatus());
-                state.ddown.Assign(buttons[DDown - BUTTON_HID_BEGIN]->GetStatus());
-
-                state.lstick_left.Assign(buttons[LStick_Left - BUTTON_HID_BEGIN]->GetStatus());
-                state.lstick_up.Assign(buttons[LStick_Up - BUTTON_HID_BEGIN]->GetStatus());
-                state.lstick_right.Assign(buttons[LStick_Right - BUTTON_HID_BEGIN]->GetStatus());
-                state.lstick_down.Assign(buttons[LStick_Down - BUTTON_HID_BEGIN]->GetStatus());
-
-                state.rstick_left.Assign(buttons[RStick_Left - BUTTON_HID_BEGIN]->GetStatus());
-                state.rstick_up.Assign(buttons[RStick_Up - BUTTON_HID_BEGIN]->GetStatus());
-                state.rstick_right.Assign(buttons[RStick_Right - BUTTON_HID_BEGIN]->GetStatus());
-                state.rstick_down.Assign(buttons[RStick_Down - BUTTON_HID_BEGIN]->GetStatus());
-
-                state.sl.Assign(buttons[SL - BUTTON_HID_BEGIN]->GetStatus());
-                state.sr.Assign(buttons[SR - BUTTON_HID_BEGIN]->GetStatus());
-
-                const auto [stick_l_x_f, stick_l_y_f] = sticks[Joystick_Left]->GetStatus();
-                const auto [stick_r_x_f, stick_r_y_f] = sticks[Joystick_Right]->GetStatus();
-                entry.joystick_left_x = static_cast<s32>(stick_l_x_f * HID_JOYSTICK_MAX);
-                entry.joystick_left_y = static_cast<s32>(stick_l_y_f * HID_JOYSTICK_MAX);
-                entry.joystick_right_x = static_cast<s32>(stick_r_x_f * HID_JOYSTICK_MAX);
-                entry.joystick_right_y = static_cast<s32>(stick_r_y_f * HID_JOYSTICK_MAX);
+    void UpdateControllers(u64 userdata, int cycles_late) {
+        const bool should_reload = Settings::values.is_device_reload_pending.exchange(false);
+        for (const auto& controller : controllers) {
+            if (should_reload) {
+                controller->OnLoadInputDevices();
             }
+            controller->OnUpdate(shared_mem->GetPointer(), SHARED_MEMORY_SIZE);
         }
 
-        TouchScreen& touchscreen = mem.touchscreen;
-        const u64 last_entry = touchscreen.header.latest_entry;
-        const u64 curr_entry = (last_entry + 1) % touchscreen.entries.size();
-        const u64 timestamp = CoreTiming::GetTicks();
-        const u64 sample_counter = touchscreen.entries[last_entry].header.timestamp + 1;
-        touchscreen.header.timestamp_ticks = timestamp;
-        touchscreen.header.num_entries = touchscreen.entries.size();
-        touchscreen.header.latest_entry = curr_entry;
-        touchscreen.header.max_entry_index = touchscreen.entries.size();
-        touchscreen.header.timestamp = timestamp;
-        touchscreen.entries[curr_entry].header.timestamp = sample_counter;
-
-        TouchScreenEntryTouch touch_entry{};
-        auto [x, y, pressed] = touch_device->GetStatus();
-        touch_entry.timestamp = timestamp;
-        touch_entry.x = static_cast<u16>(x * Layout::ScreenUndocked::Width);
-        touch_entry.y = static_cast<u16>(y * Layout::ScreenUndocked::Height);
-        touch_entry.touch_index = 0;
-
-        // TODO(DarkLordZach): Maybe try to derive these from EmuWindow?
-        touch_entry.diameter_x = 15;
-        touch_entry.diameter_y = 15;
-        touch_entry.angle = 0;
-
-        // TODO(DarkLordZach): Implement multi-touch support
-        if (pressed) {
-            touchscreen.entries[curr_entry].header.num_touches = 1;
-            touchscreen.entries[curr_entry].touches[0] = touch_entry;
-        } else {
-            touchscreen.entries[curr_entry].header.num_touches = 0;
-        }
-
-        // TODO(shinyquagsire23): Properly implement mouse
-        Mouse& mouse = mem.mouse;
-        const u64 last_mouse_entry = mouse.header.latest_entry;
-        const u64 curr_mouse_entry = (mouse.header.latest_entry + 1) % mouse.entries.size();
-        const u64 mouse_sample_counter = mouse.entries[last_mouse_entry].timestamp + 1;
-        mouse.header.timestamp_ticks = timestamp;
-        mouse.header.num_entries = mouse.entries.size();
-        mouse.header.max_entry_index = mouse.entries.size();
-        mouse.header.latest_entry = curr_mouse_entry;
-
-        mouse.entries[curr_mouse_entry].timestamp = mouse_sample_counter;
-        mouse.entries[curr_mouse_entry].timestamp_2 = mouse_sample_counter;
-
-        // TODO(shinyquagsire23): Properly implement keyboard
-        Keyboard& keyboard = mem.keyboard;
-        const u64 last_keyboard_entry = keyboard.header.latest_entry;
-        const u64 curr_keyboard_entry =
-            (keyboard.header.latest_entry + 1) % keyboard.entries.size();
-        const u64 keyboard_sample_counter = keyboard.entries[last_keyboard_entry].timestamp + 1;
-        keyboard.header.timestamp_ticks = timestamp;
-        keyboard.header.num_entries = keyboard.entries.size();
-        keyboard.header.latest_entry = last_keyboard_entry;
-        keyboard.header.max_entry_index = keyboard.entries.size();
-
-        keyboard.entries[curr_keyboard_entry].timestamp = keyboard_sample_counter;
-        keyboard.entries[curr_keyboard_entry].timestamp_2 = keyboard_sample_counter;
-
-        // TODO(shinyquagsire23): Figure out what any of these are
-        for (auto& input : mem.unk_input_1) {
-            const u64 last_input_entry = input.header.latest_entry;
-            const u64 curr_input_entry = (input.header.latest_entry + 1) % input.entries.size();
-            const u64 input_sample_counter = input.entries[last_input_entry].timestamp + 1;
-
-            input.header.timestamp_ticks = timestamp;
-            input.header.num_entries = input.entries.size();
-            input.header.latest_entry = last_input_entry;
-            input.header.max_entry_index = input.entries.size();
-
-            input.entries[curr_input_entry].timestamp = input_sample_counter;
-            input.entries[curr_input_entry].timestamp_2 = input_sample_counter;
-        }
-
-        for (auto& input : mem.unk_input_2) {
-            input.header.timestamp_ticks = timestamp;
-            input.header.num_entries = 17;
-            input.header.latest_entry = 0;
-            input.header.max_entry_index = 0;
-        }
-
-        UnkInput3& input = mem.unk_input_3;
-        const u64 last_input_entry = input.header.latest_entry;
-        const u64 curr_input_entry = (input.header.latest_entry + 1) % input.entries.size();
-        const u64 input_sample_counter = input.entries[last_input_entry].timestamp + 1;
-
-        input.header.timestamp_ticks = timestamp;
-        input.header.num_entries = input.entries.size();
-        input.header.latest_entry = last_input_entry;
-        input.header.max_entry_index = input.entries.size();
-
-        input.entries[curr_input_entry].timestamp = input_sample_counter;
-        input.entries[curr_input_entry].timestamp_2 = input_sample_counter;
-
-        // TODO(shinyquagsire23): Signal events
-
-        std::memcpy(shared_mem->GetPointer(), &mem, sizeof(SharedMemory));
-
-        // Reschedule recurrent event
         CoreTiming::ScheduleEvent(pad_update_ticks - cycles_late, pad_update_event);
     }
 
@@ -265,11 +146,8 @@ private:
     // CoreTiming update events
     CoreTiming::EventType* pad_update_event;
 
-    // Stored input state info
-    std::array<std::unique_ptr<Input::ButtonDevice>, Settings::NativeButton::NUM_BUTTONS_HID>
-        buttons;
-    std::array<std::unique_ptr<Input::AnalogDevice>, Settings::NativeAnalog::NUM_STICKS_HID> sticks;
-    std::unique_ptr<Input::TouchDevice> touch_device;
+    std::array<std::unique_ptr<ControllerBase>, static_cast<size_t>(HidController::MaxControllers)>
+        controllers{};
 };
 
 class IActiveVibrationDeviceList final : public ServiceFramework<IActiveVibrationDeviceList> {
@@ -301,7 +179,7 @@ public:
             {31, &Hid::ActivateKeyboard, "ActivateKeyboard"},
             {40, nullptr, "AcquireXpadIdEventHandle"},
             {41, nullptr, "ReleaseXpadIdEventHandle"},
-            {51, nullptr, "ActivateXpad"},
+            {51, &Hid::ActivateXpad, "ActivateXpad"},
             {55, nullptr, "GetXpadIds"},
             {56, nullptr, "ActivateJoyXpad"},
             {58, nullptr, "GetJoyXpadLifoHandle"},
@@ -362,8 +240,8 @@ public:
             {206, &Hid::SendVibrationValues, "SendVibrationValues"},
             {207, nullptr, "SendVibrationGcErmCommand"},
             {208, nullptr, "GetActualVibrationGcErmCommand"},
-            {209, nullptr, "BeginPermitVibrationSession"},
-            {210, nullptr, "EndPermitVibrationSession"},
+            {209, &Hid::BeginPermitVibrationSession, "BeginPermitVibrationSession"},
+            {210, &Hid::EndPermitVibrationSession, "EndPermitVibrationSession"},
             {300, &Hid::ActivateConsoleSixAxisSensor, "ActivateConsoleSixAxisSensor"},
             {301, &Hid::StartConsoleSixAxisSensor, "StartConsoleSixAxisSensor"},
             {302, nullptr, "StopConsoleSixAxisSensor"},
@@ -401,16 +279,11 @@ public:
         // clang-format on
 
         RegisterHandlers(functions);
-
-        auto& kernel = Core::System::GetInstance().Kernel();
-        event = Kernel::Event::Create(kernel, Kernel::ResetType::OneShot, "hid:EventHandle");
     }
     ~Hid() = default;
 
 private:
     std::shared_ptr<IAppletResource> applet_resource;
-    u32 joy_hold_type{0};
-    Kernel::SharedPtr<Kernel::Event> event;
 
     void CreateAppletResource(Kernel::HLERequestContext& ctx) {
         if (applet_resource == nullptr) {
@@ -423,31 +296,59 @@ private:
         LOG_DEBUG(Service_HID, "called");
     }
 
-    void ActivateDebugPad(Kernel::HLERequestContext& ctx) {
+    void ActivateXpad(Kernel::HLERequestContext& ctx) {
+        applet_resource->ActivateController(HidController::XPad);
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        LOG_DEBUG(Service_HID, "called");
+    }
+
+    void ActivateDebugPad(Kernel::HLERequestContext& ctx) {
+        applet_resource->ActivateController(HidController::DebugPad);
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void ActivateTouchScreen(Kernel::HLERequestContext& ctx) {
+        applet_resource->ActivateController(HidController::Touchscreen);
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void ActivateMouse(Kernel::HLERequestContext& ctx) {
+        applet_resource->ActivateController(HidController::Mouse);
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void ActivateKeyboard(Kernel::HLERequestContext& ctx) {
+        applet_resource->ActivateController(HidController::Keyboard);
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        LOG_DEBUG(Service_HID, "called");
+    }
+
+    void ActivateGesture(Kernel::HLERequestContext& ctx) {
+        applet_resource->ActivateController(HidController::Gesture);
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_DEBUG(Service_HID, "called");
+    }
+
+    void ActivateNpadWithRevision(Kernel::HLERequestContext& ctx) {
+        // Should have no effect with how our npad sets up the data
+        applet_resource->ActivateController(HidController::NPad);
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void StartSixAxisSensor(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        auto handle = rp.PopRaw<u32>();
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
         LOG_WARNING(Service_HID, "(STUBBED) called");
@@ -468,84 +369,168 @@ private:
     }
 
     void SetSupportedNpadStyleSet(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        auto supported_styleset = rp.PopRaw<u32>();
+        applet_resource->GetController<Controller_NPad>(HidController::NPad)
+            .SetSupportedStyleSet({supported_styleset});
+
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void GetSupportedNpadStyleSet(Kernel::HLERequestContext& ctx) {
+        auto& controller = applet_resource->GetController<Controller_NPad>(HidController::NPad);
+
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
-        rb.Push<u32>(0);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        rb.Push<u32>(controller.GetSupportedStyleSet().raw);
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void SetSupportedNpadIdType(Kernel::HLERequestContext& ctx) {
+        applet_resource->GetController<Controller_NPad>(HidController::NPad)
+            .SetSupportedNPadIdTypes(ctx.ReadBuffer().data(), ctx.GetReadBufferSize());
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void ActivateNpad(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        applet_resource->ActivateController(HidController::NPad);
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void AcquireNpadStyleSetUpdateEventHandle(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        auto npad_id = rp.PopRaw<u32>();
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(RESULT_SUCCESS);
-        rb.PushCopyObjects(event);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        rb.PushCopyObjects(applet_resource->GetController<Controller_NPad>(HidController::NPad)
+                               .GetStyleSetChangedEvent());
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void DisconnectNpad(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        auto npad_id = rp.PopRaw<u32>();
+        applet_resource->GetController<Controller_NPad>(HidController::NPad)
+            .DisconnectNPad(npad_id);
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void GetPlayerLedPattern(Kernel::HLERequestContext& ctx) {
-        IPC::ResponseBuilder rb{ctx, 2};
+        IPC::RequestParser rp{ctx};
+        auto npad_id = rp.PopRaw<u32>();
+        IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        rb.PushRaw<u64>(applet_resource->GetController<Controller_NPad>(HidController::NPad)
+                            .GetLedPattern(npad_id)
+                            .raw);
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void SetNpadJoyHoldType(Kernel::HLERequestContext& ctx) {
+        auto& controller = applet_resource->GetController<Controller_NPad>(HidController::NPad);
+        IPC::RequestParser rp{ctx};
+        const auto hold_type = rp.PopRaw<u64>();
+        controller.SetHoldType(Controller_NPad::NpadHoldType{hold_type});
+
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void GetNpadJoyHoldType(Kernel::HLERequestContext& ctx) {
-        IPC::ResponseBuilder rb{ctx, 3};
+        const auto& controller =
+            applet_resource->GetController<Controller_NPad>(HidController::NPad);
+        IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
-        rb.Push(joy_hold_type);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        rb.Push<u64>(static_cast<u64>(controller.GetHoldType()));
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void SetNpadJoyAssignmentModeSingleByDefault(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        auto npad_id = rp.PopRaw<u32>();
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
         LOG_WARNING(Service_HID, "(STUBBED) called");
     }
 
+    void BeginPermitVibrationSession(Kernel::HLERequestContext& ctx) {
+        applet_resource->GetController<Controller_NPad>(HidController::NPad)
+            .SetVibrationEnabled(true);
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_DEBUG(Service_HID, "called");
+    }
+
+    void EndPermitVibrationSession(Kernel::HLERequestContext& ctx) {
+        applet_resource->GetController<Controller_NPad>(HidController::NPad)
+            .SetVibrationEnabled(false);
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_DEBUG(Service_HID, "called");
+    }
+
     void SendVibrationValue(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const auto controller_id = rp.PopRaw<u32>();
+        const auto vibration_values = rp.PopRaw<Controller_NPad::Vibration>();
+
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+
+        applet_resource->GetController<Controller_NPad>(HidController::NPad)
+            .VibrateController({controller_id}, {vibration_values});
+        LOG_DEBUG(Service_HID, "called");
+    }
+
+    void SendVibrationValues(Kernel::HLERequestContext& ctx) {
+        const auto controllers = ctx.ReadBuffer(0);
+        const auto vibrations = ctx.ReadBuffer(1);
+
+        std::vector<u32> controller_list(controllers.size() / sizeof(u32));
+        std::vector<Controller_NPad::Vibration> vibration_list(vibrations.size() /
+                                                               sizeof(Controller_NPad::Vibration));
+
+        std::memcpy(controller_list.data(), controllers.data(), controllers.size());
+        std::memcpy(vibration_list.data(), vibrations.data(), vibrations.size());
+        std::transform(controller_list.begin(), controller_list.end(), controller_list.begin(),
+                       [](u32 controller_id) { return controller_id - 3; });
+
+        applet_resource->GetController<Controller_NPad>(HidController::NPad)
+            .VibrateController(controller_list, vibration_list);
+
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void GetActualVibrationValue(Kernel::HLERequestContext& ctx) {
-        IPC::ResponseBuilder rb{ctx, 2};
+        IPC::ResponseBuilder rb{ctx, 6};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        rb.PushRaw<Controller_NPad::Vibration>(
+            applet_resource->GetController<Controller_NPad>(HidController::NPad)
+                .GetLastVibration());
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void SetNpadJoyAssignmentModeDual(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const auto npad_id = rp.PopRaw<u32>();
+        auto& controller = applet_resource->GetController<Controller_NPad>(HidController::NPad);
+        controller.SetNpadMode(npad_id, Controller_NPad::NPadAssignments::Dual);
+
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void MergeSingleJoyAsDualJoy(Kernel::HLERequestContext& ctx) {
@@ -555,6 +540,8 @@ private:
     }
 
     void SetNpadHandheldActivationMode(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        auto mode = rp.PopRaw<u32>();
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
         LOG_WARNING(Service_HID, "(STUBBED) called");
@@ -563,8 +550,9 @@ private:
     void GetVibrationDeviceInfo(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
-        rb.Push<u64>(0);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
+        rb.Push<u32>(1);
+        rb.Push<u32>(0);
+        LOG_DEBUG(Service_HID, "called");
     }
 
     void CreateActiveVibrationDeviceList(Kernel::HLERequestContext& ctx) {
@@ -574,12 +562,6 @@ private:
         LOG_DEBUG(Service_HID, "called");
     }
 
-    void SendVibrationValues(Kernel::HLERequestContext& ctx) {
-        IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
-    }
-
     void ActivateConsoleSixAxisSensor(Kernel::HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
@@ -597,18 +579,6 @@ private:
         rb.Push(RESULT_SUCCESS);
         LOG_WARNING(Service_HID, "(STUBBED) called");
     }
-
-    void ActivateGesture(Kernel::HLERequestContext& ctx) {
-        IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
-    }
-
-    void ActivateNpadWithRevision(Kernel::HLERequestContext& ctx) {
-        IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_HID, "(STUBBED) called");
-    }
 };
 
 class HidDbg final : public ServiceFramework<HidDbg> {

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

@@ -4,408 +4,12 @@
 
 #pragma once
 
-#include <array>
-#include "common/bit_field.h"
-#include "common/common_types.h"
-#include "core/hle/service/service.h"
+namespace SM {
+class ServiceManager;
+}
 
 namespace Service::HID {
 
-// Begin enums and output structs
-
-constexpr u32 HID_NUM_ENTRIES = 17;
-constexpr u32 HID_NUM_LAYOUTS = 7;
-constexpr s32 HID_JOYSTICK_MAX = 0x8000;
-constexpr s32 HID_JOYSTICK_MIN = -0x8000;
-
-constexpr u32 JOYCON_BODY_NEON_RED = 0xFF3C28;
-constexpr u32 JOYCON_BUTTONS_NEON_RED = 0x1E0A0A;
-constexpr u32 JOYCON_BODY_NEON_BLUE = 0x0AB9E6;
-constexpr u32 JOYCON_BUTTONS_NEON_BLUE = 0x001E1E;
-
-enum ControllerType : u32 {
-    ControllerType_ProController = 1 << 0,
-    ControllerType_Handheld = 1 << 1,
-    ControllerType_JoyconPair = 1 << 2,
-    ControllerType_JoyconLeft = 1 << 3,
-    ControllerType_JoyconRight = 1 << 4,
-};
-
-enum ControllerLayoutType : u32 {
-    Layout_ProController = 0, // Pro Controller or HID gamepad
-    Layout_Handheld = 1,      // Two Joy-Con docked to rails
-    Layout_Single = 2, // Horizontal single Joy-Con or pair of Joy-Con, adjusted for orientation
-    Layout_Left = 3,   // Only raw left Joy-Con state, no orientation adjustment
-    Layout_Right = 4,  // Only raw right Joy-Con state, no orientation adjustment
-    Layout_DefaultDigital = 5, // Same as next, but sticks have 8-direction values only
-    Layout_Default = 6, // Safe default, single Joy-Con have buttons/sticks rotated for orientation
-};
-
-enum ControllerColorDescription {
-    ColorDesc_ColorsNonexistent = 1 << 1,
-};
-
-enum ControllerConnectionState {
-    ConnectionState_Connected = 1 << 0,
-    ConnectionState_Wired = 1 << 1,
-};
-
-enum ControllerJoystick {
-    Joystick_Left = 0,
-    Joystick_Right = 1,
-};
-
-enum ControllerID {
-    Controller_Player1 = 0,
-    Controller_Player2 = 1,
-    Controller_Player3 = 2,
-    Controller_Player4 = 3,
-    Controller_Player5 = 4,
-    Controller_Player6 = 5,
-    Controller_Player7 = 6,
-    Controller_Player8 = 7,
-    Controller_Handheld = 8,
-    Controller_Unknown = 9,
-};
-
-// End enums and output structs
-
-// Begin UnkInput3
-
-struct UnkInput3Header {
-    u64 timestamp_ticks;
-    u64 num_entries;
-    u64 latest_entry;
-    u64 max_entry_index;
-};
-static_assert(sizeof(UnkInput3Header) == 0x20, "HID UnkInput3 header structure has incorrect size");
-
-struct UnkInput3Entry {
-    u64 timestamp;
-    u64 timestamp_2;
-    u64 unk_8;
-    u64 unk_10;
-    u64 unk_18;
-};
-static_assert(sizeof(UnkInput3Entry) == 0x28, "HID UnkInput3 entry structure has incorrect size");
-
-struct UnkInput3 {
-    UnkInput3Header header;
-    std::array<UnkInput3Entry, 17> entries;
-    std::array<u8, 0x138> padding;
-};
-static_assert(sizeof(UnkInput3) == 0x400, "HID UnkInput3 structure has incorrect size");
-
-// End UnkInput3
-
-// Begin TouchScreen
-
-struct TouchScreenHeader {
-    u64 timestamp_ticks;
-    u64 num_entries;
-    u64 latest_entry;
-    u64 max_entry_index;
-    u64 timestamp;
-};
-static_assert(sizeof(TouchScreenHeader) == 0x28,
-              "HID touch screen header structure has incorrect size");
-
-struct TouchScreenEntryHeader {
-    u64 timestamp;
-    u64 num_touches;
-};
-static_assert(sizeof(TouchScreenEntryHeader) == 0x10,
-              "HID touch screen entry header structure has incorrect size");
-
-struct TouchScreenEntryTouch {
-    u64 timestamp;
-    u32 padding;
-    u32 touch_index;
-    u32 x;
-    u32 y;
-    u32 diameter_x;
-    u32 diameter_y;
-    u32 angle;
-    u32 padding_2;
-};
-static_assert(sizeof(TouchScreenEntryTouch) == 0x28,
-              "HID touch screen touch structure has incorrect size");
-
-struct TouchScreenEntry {
-    TouchScreenEntryHeader header;
-    std::array<TouchScreenEntryTouch, 16> touches;
-    u64 unk;
-};
-static_assert(sizeof(TouchScreenEntry) == 0x298,
-              "HID touch screen entry structure has incorrect size");
-
-struct TouchScreen {
-    TouchScreenHeader header;
-    std::array<TouchScreenEntry, 17> entries;
-    std::array<u8, 0x3c0> padding;
-};
-static_assert(sizeof(TouchScreen) == 0x3000, "HID touch screen structure has incorrect size");
-
-// End TouchScreen
-
-// Begin Mouse
-
-struct MouseHeader {
-    u64 timestamp_ticks;
-    u64 num_entries;
-    u64 latest_entry;
-    u64 max_entry_index;
-};
-static_assert(sizeof(MouseHeader) == 0x20, "HID mouse header structure has incorrect size");
-
-struct MouseButtonState {
-    union {
-        u64 hex{};
-
-        // Buttons
-        BitField<0, 1, u64> left;
-        BitField<1, 1, u64> right;
-        BitField<2, 1, u64> middle;
-        BitField<3, 1, u64> forward;
-        BitField<4, 1, u64> back;
-    };
-};
-
-struct MouseEntry {
-    u64 timestamp;
-    u64 timestamp_2;
-    u32 x;
-    u32 y;
-    u32 velocity_x;
-    u32 velocity_y;
-    u32 scroll_velocity_x;
-    u32 scroll_velocity_y;
-    MouseButtonState buttons;
-};
-static_assert(sizeof(MouseEntry) == 0x30, "HID mouse entry structure has incorrect size");
-
-struct Mouse {
-    MouseHeader header;
-    std::array<MouseEntry, 17> entries;
-    std::array<u8, 0xB0> padding;
-};
-static_assert(sizeof(Mouse) == 0x400, "HID mouse structure has incorrect size");
-
-// End Mouse
-
-// Begin Keyboard
-
-struct KeyboardHeader {
-    u64 timestamp_ticks;
-    u64 num_entries;
-    u64 latest_entry;
-    u64 max_entry_index;
-};
-static_assert(sizeof(KeyboardHeader) == 0x20, "HID keyboard header structure has incorrect size");
-
-struct KeyboardModifierKeyState {
-    union {
-        u64 hex{};
-
-        // Buttons
-        BitField<0, 1, u64> lctrl;
-        BitField<1, 1, u64> lshift;
-        BitField<2, 1, u64> lalt;
-        BitField<3, 1, u64> lmeta;
-        BitField<4, 1, u64> rctrl;
-        BitField<5, 1, u64> rshift;
-        BitField<6, 1, u64> ralt;
-        BitField<7, 1, u64> rmeta;
-        BitField<8, 1, u64> capslock;
-        BitField<9, 1, u64> scrolllock;
-        BitField<10, 1, u64> numlock;
-    };
-};
-
-struct KeyboardEntry {
-    u64 timestamp;
-    u64 timestamp_2;
-    KeyboardModifierKeyState modifier;
-    u32 keys[8];
-};
-static_assert(sizeof(KeyboardEntry) == 0x38, "HID keyboard entry structure has incorrect size");
-
-struct Keyboard {
-    KeyboardHeader header;
-    std::array<KeyboardEntry, 17> entries;
-    std::array<u8, 0x28> padding;
-};
-static_assert(sizeof(Keyboard) == 0x400, "HID keyboard structure has incorrect size");
-
-// End Keyboard
-
-// Begin UnkInput1
-
-struct UnkInput1Header {
-    u64 timestamp_ticks;
-    u64 num_entries;
-    u64 latest_entry;
-    u64 max_entry_index;
-};
-static_assert(sizeof(UnkInput1Header) == 0x20, "HID UnkInput1 header structure has incorrect size");
-
-struct UnkInput1Entry {
-    u64 timestamp;
-    u64 timestamp_2;
-    u64 unk_8;
-    u64 unk_10;
-    u64 unk_18;
-};
-static_assert(sizeof(UnkInput1Entry) == 0x28, "HID UnkInput1 entry structure has incorrect size");
-
-struct UnkInput1 {
-    UnkInput1Header header;
-    std::array<UnkInput1Entry, 17> entries;
-    std::array<u8, 0x138> padding;
-};
-static_assert(sizeof(UnkInput1) == 0x400, "HID UnkInput1 structure has incorrect size");
-
-// End UnkInput1
-
-// Begin UnkInput2
-
-struct UnkInput2Header {
-    u64 timestamp_ticks;
-    u64 num_entries;
-    u64 latest_entry;
-    u64 max_entry_index;
-};
-static_assert(sizeof(UnkInput2Header) == 0x20, "HID UnkInput2 header structure has incorrect size");
-
-struct UnkInput2 {
-    UnkInput2Header header;
-    std::array<u8, 0x1E0> padding;
-};
-static_assert(sizeof(UnkInput2) == 0x200, "HID UnkInput2 structure has incorrect size");
-
-// End UnkInput2
-
-// Begin Controller
-
-struct ControllerMAC {
-    u64 timestamp;
-    std::array<u8, 0x8> mac;
-    u64 unk;
-    u64 timestamp_2;
-};
-static_assert(sizeof(ControllerMAC) == 0x20, "HID controller MAC structure has incorrect size");
-
-struct ControllerHeader {
-    u32 type;
-    u32 is_half;
-    u32 single_colors_descriptor;
-    u32 single_color_body;
-    u32 single_color_buttons;
-    u32 split_colors_descriptor;
-    u32 left_color_body;
-    u32 left_color_buttons;
-    u32 right_color_body;
-    u32 right_color_buttons;
-};
-static_assert(sizeof(ControllerHeader) == 0x28,
-              "HID controller header structure has incorrect size");
-
-struct ControllerLayoutHeader {
-    u64 timestamp_ticks;
-    u64 num_entries;
-    u64 latest_entry;
-    u64 max_entry_index;
-};
-static_assert(sizeof(ControllerLayoutHeader) == 0x20,
-              "HID controller layout header structure has incorrect size");
-
-struct ControllerPadState {
-    union {
-        u64 hex{};
-
-        // Buttons
-        BitField<0, 1, u64> a;
-        BitField<1, 1, u64> b;
-        BitField<2, 1, u64> x;
-        BitField<3, 1, u64> y;
-        BitField<4, 1, u64> lstick;
-        BitField<5, 1, u64> rstick;
-        BitField<6, 1, u64> l;
-        BitField<7, 1, u64> r;
-        BitField<8, 1, u64> zl;
-        BitField<9, 1, u64> zr;
-        BitField<10, 1, u64> plus;
-        BitField<11, 1, u64> minus;
-
-        // D-pad buttons
-        BitField<12, 1, u64> dleft;
-        BitField<13, 1, u64> dup;
-        BitField<14, 1, u64> dright;
-        BitField<15, 1, u64> ddown;
-
-        // Left stick directions
-        BitField<16, 1, u64> lstick_left;
-        BitField<17, 1, u64> lstick_up;
-        BitField<18, 1, u64> lstick_right;
-        BitField<19, 1, u64> lstick_down;
-
-        // Right stick directions
-        BitField<20, 1, u64> rstick_left;
-        BitField<21, 1, u64> rstick_up;
-        BitField<22, 1, u64> rstick_right;
-        BitField<23, 1, u64> rstick_down;
-
-        BitField<24, 1, u64> sl;
-        BitField<25, 1, u64> sr;
-    };
-};
-
-struct ControllerInputEntry {
-    u64 timestamp;
-    u64 timestamp_2;
-    ControllerPadState buttons;
-    s32 joystick_left_x;
-    s32 joystick_left_y;
-    s32 joystick_right_x;
-    s32 joystick_right_y;
-    u64 connection_state;
-};
-static_assert(sizeof(ControllerInputEntry) == 0x30,
-              "HID controller input entry structure has incorrect size");
-
-struct ControllerLayout {
-    ControllerLayoutHeader header;
-    std::array<ControllerInputEntry, 17> entries;
-};
-static_assert(sizeof(ControllerLayout) == 0x350,
-              "HID controller layout structure has incorrect size");
-
-struct Controller {
-    ControllerHeader header;
-    std::array<ControllerLayout, HID_NUM_LAYOUTS> layouts;
-    std::array<u8, 0x2a70> unk_1;
-    ControllerMAC mac_left;
-    ControllerMAC mac_right;
-    std::array<u8, 0xdf8> unk_2;
-};
-static_assert(sizeof(Controller) == 0x5000, "HID controller structure has incorrect size");
-
-// End Controller
-
-struct SharedMemory {
-    UnkInput3 unk_input_3;
-    TouchScreen touchscreen;
-    Mouse mouse;
-    Keyboard keyboard;
-    std::array<UnkInput1, 4> unk_input_1;
-    std::array<UnkInput2, 3> unk_input_2;
-    std::array<u8, 0x800> unk_section_8;
-    std::array<u8, 0x4000> controller_serials;
-    std::array<Controller, 10> controllers;
-    std::array<u8, 0x4600> unk_section_9;
-};
-static_assert(sizeof(SharedMemory) == 0x40000, "HID Shared Memory structure has incorrect size");
-
 /// Reload input devices. Used when input configuration changed
 void ReloadInputDevices();
 

src/core/hle/service/mm/mm_u.cpp

@@ -14,14 +14,14 @@ public:
     explicit MM_U() : ServiceFramework{"mm:u"} {
         // clang-format off
         static const FunctionInfo functions[] = {
-            {0, &MM_U::Initialize, "InitializeOld"},
-            {1, &MM_U::Finalize, "FinalizeOld"},
-            {2, &MM_U::SetAndWait, "SetAndWaitOld"},
-            {3, &MM_U::Get, "GetOld"},
-            {4, &MM_U::Initialize, "Initialize"},
-            {5, &MM_U::Finalize, "Finalize"},
-            {6, &MM_U::SetAndWait, "SetAndWait"},
-            {7, &MM_U::Get, "Get"},
+            {0, &MM_U::Initialize, "Initialize"},
+            {1, &MM_U::Finalize, "Finalize"},
+            {2, &MM_U::SetAndWait, "SetAndWait"},
+            {3, &MM_U::Get, "Get"},
+            {4, &MM_U::InitializeWithId, "InitializeWithId"},
+            {5, &MM_U::FinalizeWithId, "FinalizeWithId"},
+            {6, &MM_U::SetAndWaitWithId, "SetAndWaitWithId"},
+            {7, &MM_U::GetWithId, "GetWithId"},
         };
         // clang-format on
 
@@ -59,9 +59,43 @@ private:
         rb.Push(current);
     }
 
+    void InitializeWithId(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_MM, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(id); // Any non zero value
+    }
+
+    void FinalizeWithId(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_MM, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+
+    void SetAndWaitWithId(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        u32 input_id = rp.Pop<u32>();
+        min = rp.Pop<u32>();
+        max = rp.Pop<u32>();
+        current = min;
+
+        LOG_WARNING(Service_MM, "(STUBBED) called, input_id=0x{:X}, min=0x{:X}, max=0x{:X}",
+                    input_id, min, max);
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+
+    void GetWithId(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_MM, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push(current);
+    }
+
     u32 min{0};
     u32 max{0};
     u32 current{0};
+    u32 id{1};
 };
 
 void InstallInterfaces(SM::ServiceManager& service_manager) {

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

@@ -144,7 +144,7 @@ private:
     }
 
     const u64 device_handle{0xDEAD};
-    const HID::ControllerID npad_id{HID::Controller_Player1};
+    const u32 npad_id{0}; // This is the first player controller id
     State state{State::NonInitialized};
     DeviceState device_state{DeviceState::Initialized};
     Kernel::SharedPtr<Kernel::Event> activate_event;

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

@@ -161,7 +161,7 @@ PL_U::PL_U() : ServiceFramework("pl:u"), impl{std::make_unique<Impl>()} {
     };
     RegisterHandlers(functions);
     // Attempt to load shared font data from disk
-    const auto nand = FileSystem::GetSystemNANDContents();
+    const auto* nand = FileSystem::GetSystemNANDContents();
     std::size_t offset = 0;
     // Rebuild shared fonts from data ncas
     if (nand->HasEntry(static_cast<u64>(FontArchives::Standard),

src/core/hle/service/service.cpp

@@ -197,7 +197,7 @@ ResultCode ServiceFrameworkBase::HandleSyncRequest(Kernel::HLERequestContext& co
 // Module interface
 
 /// Initialize ServiceManager
-void Init(std::shared_ptr<SM::ServiceManager>& sm, const FileSys::VirtualFilesystem& rfs) {
+void Init(std::shared_ptr<SM::ServiceManager>& sm, FileSys::VfsFilesystem& vfs) {
     // NVFlinger needs to be accessed by several services like Vi and AppletOE so we instantiate it
     // here and pass it into the respective InstallInterfaces functions.
     auto nv_flinger = std::make_shared<NVFlinger::NVFlinger>();
@@ -220,7 +220,7 @@ void Init(std::shared_ptr<SM::ServiceManager>& sm, const FileSys::VirtualFilesys
     EUPLD::InstallInterfaces(*sm);
     Fatal::InstallInterfaces(*sm);
     FGM::InstallInterfaces(*sm);
-    FileSystem::InstallInterfaces(*sm, rfs);
+    FileSystem::InstallInterfaces(*sm, vfs);
     Friend::InstallInterfaces(*sm);
     GRC::InstallInterfaces(*sm);
     HID::InstallInterfaces(*sm);

src/core/hle/service/service.h

@@ -180,8 +180,7 @@ private:
 };
 
 /// Initialize ServiceManager
-void Init(std::shared_ptr<SM::ServiceManager>& sm,
-          const std::shared_ptr<FileSys::VfsFilesystem>& vfs);
+void Init(std::shared_ptr<SM::ServiceManager>& sm, FileSys::VfsFilesystem& vfs);
 
 /// Shutdown ServiceManager
 void Shutdown();

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

@@ -968,6 +968,54 @@ private:
         rb.PushCopyObjects(vsync_event);
     }
 
+    enum class ConvertedScaleMode : u64 {
+        None = 0, // VI seems to name this as "Unknown" but lots of games pass it, assume it's no
+                  // scaling/default
+        Freeze = 1,
+        ScaleToWindow = 2,
+        Crop = 3,
+        NoCrop = 4,
+    };
+
+    // This struct is different, currently it's 1:1 but this might change in the future.
+    enum class NintendoScaleMode : u32 {
+        None = 0,
+        Freeze = 1,
+        ScaleToWindow = 2,
+        Crop = 3,
+        NoCrop = 4,
+    };
+
+    void ConvertScalingMode(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        auto mode = rp.PopEnum<NintendoScaleMode>();
+        LOG_DEBUG(Service_VI, "called mode={}", static_cast<u32>(mode));
+
+        IPC::ResponseBuilder rb{ctx, 4};
+        rb.Push(RESULT_SUCCESS);
+        switch (mode) {
+        case NintendoScaleMode::None:
+            rb.PushEnum(ConvertedScaleMode::None);
+            break;
+        case NintendoScaleMode::Freeze:
+            rb.PushEnum(ConvertedScaleMode::Freeze);
+            break;
+        case NintendoScaleMode::ScaleToWindow:
+            rb.PushEnum(ConvertedScaleMode::ScaleToWindow);
+            break;
+        case NintendoScaleMode::Crop:
+            rb.PushEnum(ConvertedScaleMode::Crop);
+            break;
+        case NintendoScaleMode::NoCrop:
+            rb.PushEnum(ConvertedScaleMode::NoCrop);
+            break;
+        default:
+            UNIMPLEMENTED_MSG("Unknown scaling mode {}", static_cast<u32>(mode));
+            rb.PushEnum(ConvertedScaleMode::None);
+            break;
+        }
+    }
+
     std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
 };
 
@@ -991,7 +1039,7 @@ IApplicationDisplayService::IApplicationDisplayService(
         {2030, &IApplicationDisplayService::CreateStrayLayer, "CreateStrayLayer"},
         {2031, &IApplicationDisplayService::DestroyStrayLayer, "DestroyStrayLayer"},
         {2101, &IApplicationDisplayService::SetLayerScalingMode, "SetLayerScalingMode"},
-        {2102, nullptr, "ConvertScalingMode"},
+        {2102, &IApplicationDisplayService::ConvertScalingMode, "ConvertScalingMode"},
         {2450, nullptr, "GetIndirectLayerImageMap"},
         {2451, nullptr, "GetIndirectLayerImageCropMap"},
         {2460, nullptr, "GetIndirectLayerImageRequiredMemoryInfo"},

src/core/loader/deconstructed_rom_directory.cpp

@@ -139,14 +139,22 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(Kernel::Process& process)
     for (const auto& module : {"rtld", "main", "subsdk0", "subsdk1", "subsdk2", "subsdk3",
                                "subsdk4", "subsdk5", "subsdk6", "subsdk7", "sdk"}) {
         const FileSys::VirtualFile module_file = dir->GetFile(module);
-        if (module_file != nullptr) {
-            const VAddr load_addr = next_load_addr;
-            next_load_addr = AppLoader_NSO::LoadModule(module_file, load_addr,
-                                                       std::strcmp(module, "rtld") == 0, pm);
-            LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", module, load_addr);
-            // Register module with GDBStub
-            GDBStub::RegisterModule(module, load_addr, next_load_addr - 1, false);
+        if (module_file == nullptr) {
+            continue;
         }
+
+        const VAddr load_addr = next_load_addr;
+        const bool should_pass_arguments = std::strcmp(module, "rtld") == 0;
+        const auto tentative_next_load_addr =
+            AppLoader_NSO::LoadModule(*module_file, load_addr, should_pass_arguments, pm);
+        if (!tentative_next_load_addr) {
+            return ResultStatus::ErrorLoadingNSO;
+        }
+
+        next_load_addr = *tentative_next_load_addr;
+        LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", module, load_addr);
+        // Register module with GDBStub
+        GDBStub::RegisterModule(module, load_addr, next_load_addr - 1, false);
     }
 
     process.Run(base_address, metadata.GetMainThreadPriority(), metadata.GetMainThreadStackSize());

src/core/loader/loader.cpp

@@ -93,7 +93,7 @@ std::string GetFileTypeString(FileType type) {
     return "unknown";
 }
 
-constexpr std::array<const char*, 59> RESULT_MESSAGES{
+constexpr std::array<const char*, 60> RESULT_MESSAGES{
     "The operation completed successfully.",
     "The loader requested to load is already loaded.",
     "The operation is not implemented.",
@@ -128,6 +128,7 @@ constexpr std::array<const char*, 59> RESULT_MESSAGES{
     "The RomFS could not be found.",
     "The ELF file has incorrect size as determined by the header.",
     "There was a general error loading the NRO into emulated memory.",
+    "There was a general error loading the NSO into emulated memory.",
     "There is no icon available.",
     "There is no control data available.",
     "The NAX file has a bad header.",

src/core/loader/loader.h

@@ -90,6 +90,7 @@ enum class ResultStatus : u16 {
     ErrorNoRomFS,
     ErrorIncorrectELFFileSize,
     ErrorLoadingNRO,
+    ErrorLoadingNSO,
     ErrorNoIcon,
     ErrorNoControl,
     ErrorBadNAXHeader,

src/core/loader/nro.cpp

@@ -127,10 +127,10 @@ static constexpr u32 PageAlignSize(u32 size) {
     return (size + Memory::PAGE_MASK) & ~Memory::PAGE_MASK;
 }
 
-bool AppLoader_NRO::LoadNro(FileSys::VirtualFile file, VAddr load_base) {
+bool AppLoader_NRO::LoadNro(const FileSys::VfsFile& file, VAddr load_base) {
     // Read NSO header
     NroHeader nro_header{};
-    if (sizeof(NroHeader) != file->ReadObject(&nro_header)) {
+    if (sizeof(NroHeader) != file.ReadObject(&nro_header)) {
         return {};
     }
     if (nro_header.magic != Common::MakeMagic('N', 'R', 'O', '0')) {
@@ -138,7 +138,7 @@ bool AppLoader_NRO::LoadNro(FileSys::VirtualFile file, VAddr load_base) {
     }
 
     // Build program image
-    std::vector<u8> program_image = file->ReadBytes(PageAlignSize(nro_header.file_size));
+    std::vector<u8> program_image = file.ReadBytes(PageAlignSize(nro_header.file_size));
     if (program_image.size() != PageAlignSize(nro_header.file_size)) {
         return {};
     }
@@ -182,7 +182,7 @@ bool AppLoader_NRO::LoadNro(FileSys::VirtualFile file, VAddr load_base) {
     Core::CurrentProcess()->LoadModule(std::move(codeset), load_base);
 
     // Register module with GDBStub
-    GDBStub::RegisterModule(file->GetName(), load_base, load_base);
+    GDBStub::RegisterModule(file.GetName(), load_base, load_base);
 
     return true;
 }
@@ -195,7 +195,7 @@ ResultStatus AppLoader_NRO::Load(Kernel::Process& process) {
     // Load NRO
     const VAddr base_address = process.VMManager().GetCodeRegionBaseAddress();
 
-    if (!LoadNro(file, base_address)) {
+    if (!LoadNro(*file, base_address)) {
         return ResultStatus::ErrorLoadingNRO;
     }
 

src/core/loader/nro.h

@@ -41,7 +41,7 @@ public:
     bool IsRomFSUpdatable() const override;
 
 private:
-    bool LoadNro(FileSys::VirtualFile file, VAddr load_base);
+    bool LoadNro(const FileSys::VfsFile& file, VAddr load_base);
 
     std::vector<u8> icon_data;
     std::unique_ptr<FileSys::NACP> nacp;

src/core/loader/nso.cpp

@@ -93,17 +93,14 @@ static constexpr u32 PageAlignSize(u32 size) {
     return (size + Memory::PAGE_MASK) & ~Memory::PAGE_MASK;
 }
 
-VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base,
-                                bool should_pass_arguments,
-                                boost::optional<FileSys::PatchManager> pm) {
-    if (file == nullptr)
-        return {};
-
-    if (file->GetSize() < sizeof(NsoHeader))
+std::optional<VAddr> AppLoader_NSO::LoadModule(const FileSys::VfsFile& file, VAddr load_base,
+                                               bool should_pass_arguments,
+                                               std::optional<FileSys::PatchManager> pm) {
+    if (file.GetSize() < sizeof(NsoHeader))
         return {};
 
     NsoHeader nso_header{};
-    if (sizeof(NsoHeader) != file->ReadObject(&nso_header))
+    if (sizeof(NsoHeader) != file.ReadObject(&nso_header))
         return {};
 
     if (nso_header.magic != Common::MakeMagic('N', 'S', 'O', '0'))
@@ -114,7 +111,7 @@ VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base,
     std::vector<u8> program_image;
     for (std::size_t i = 0; i < nso_header.segments.size(); ++i) {
         std::vector<u8> data =
-            file->ReadBytes(nso_header.segments_compressed_size[i], nso_header.segments[i].offset);
+            file.ReadBytes(nso_header.segments_compressed_size[i], nso_header.segments[i].offset);
         if (nso_header.IsSegmentCompressed(i)) {
             data = DecompressSegment(data, nso_header.segments[i]);
         }
@@ -157,7 +154,7 @@ VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base,
     program_image.resize(image_size);
 
     // Apply patches if necessary
-    if (pm != boost::none && pm->HasNSOPatch(nso_header.build_id)) {
+    if (pm && pm->HasNSOPatch(nso_header.build_id)) {
         std::vector<u8> pi_header(program_image.size() + 0x100);
         std::memcpy(pi_header.data(), &nso_header, sizeof(NsoHeader));
         std::memcpy(pi_header.data() + 0x100, program_image.data(), program_image.size());
@@ -172,7 +169,7 @@ VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base,
     Core::CurrentProcess()->LoadModule(std::move(codeset), load_base);
 
     // Register module with GDBStub
-    GDBStub::RegisterModule(file->GetName(), load_base, load_base);
+    GDBStub::RegisterModule(file.GetName(), load_base, load_base);
 
     return load_base + image_size;
 }
@@ -184,7 +181,9 @@ ResultStatus AppLoader_NSO::Load(Kernel::Process& process) {
 
     // Load module
     const VAddr base_address = process.VMManager().GetCodeRegionBaseAddress();
-    LoadModule(file, base_address, true);
+    if (!LoadModule(*file, base_address, true)) {
+        return ResultStatus::ErrorLoadingNSO;
+    }
     LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", file->GetName(), base_address);
 
     process.Run(base_address, Kernel::THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);

src/core/loader/nso.h

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <optional>
 #include "common/common_types.h"
 #include "core/file_sys/patch_manager.h"
 #include "core/loader/linker.h"
@@ -36,8 +37,9 @@ public:
         return IdentifyType(file);
     }
 
-    static VAddr LoadModule(FileSys::VirtualFile file, VAddr load_base, bool should_pass_arguments,
-                            boost::optional<FileSys::PatchManager> pm = boost::none);
+    static std::optional<VAddr> LoadModule(const FileSys::VfsFile& file, VAddr load_base,
+                                           bool should_pass_arguments,
+                                           std::optional<FileSys::PatchManager> pm = {});
 
     ResultStatus Load(Kernel::Process& process) override;
 };

src/core/loader/xci.cpp

@@ -59,8 +59,7 @@ ResultStatus AppLoader_XCI::Load(Kernel::Process& process) {
     if (xci->GetProgramNCAStatus() != ResultStatus::Success)
         return xci->GetProgramNCAStatus();
 
-    const auto nca = xci->GetProgramNCA();
-    if (nca == nullptr && !Core::Crypto::KeyManager::KeyFileExists(false))
+    if (!xci->HasProgramNCA() && !Core::Crypto::KeyManager::KeyFileExists(false))
         return ResultStatus::ErrorMissingProductionKeyFile;
 
     const auto result = nca_loader->Load(process);

src/core/settings.h

@@ -136,7 +136,7 @@ struct Values {
     float resolution_factor;
     bool use_frame_limit;
     u16 frame_limit;
-    bool use_accurate_framebuffers;
+    bool use_accurate_gpu_emulation;
 
     float bg_red;
     float bg_green;

src/core/telemetry_session.cpp

@@ -163,8 +163,8 @@ TelemetrySession::TelemetrySession() {
     AddField(Telemetry::FieldType::UserConfig, "Renderer_UseFrameLimit",
              Settings::values.use_frame_limit);
     AddField(Telemetry::FieldType::UserConfig, "Renderer_FrameLimit", Settings::values.frame_limit);
-    AddField(Telemetry::FieldType::UserConfig, "Renderer_UseAccurateFramebuffers",
-             Settings::values.use_accurate_framebuffers);
+    AddField(Telemetry::FieldType::UserConfig, "Renderer_UseAccurateGpuEmulation",
+             Settings::values.use_accurate_gpu_emulation);
     AddField(Telemetry::FieldType::UserConfig, "System_UseDockedMode",
              Settings::values.use_docked_mode);
 }

src/video_core/engines/fermi_2d.cpp

@@ -62,14 +62,16 @@ void Fermi2D::HandleSurfaceCopy() {
         u8* dst_buffer = Memory::GetPointer(dest_cpu);
         if (!regs.src.linear && regs.dst.linear) {
             // If the input is tiled and the output is linear, deswizzle the input and copy it over.
-            Texture::CopySwizzledData(regs.src.width, regs.src.height, src_bytes_per_pixel,
-                                      dst_bytes_per_pixel, src_buffer, dst_buffer, true,
-                                      regs.src.BlockHeight());
+            Texture::CopySwizzledData(regs.src.width, regs.src.height, regs.src.depth,
+                                      src_bytes_per_pixel, dst_bytes_per_pixel, src_buffer,
+                                      dst_buffer, true, regs.src.BlockHeight(),
+                                      regs.src.BlockDepth());
         } else {
             // If the input is linear and the output is tiled, swizzle the input and copy it over.
-            Texture::CopySwizzledData(regs.src.width, regs.src.height, src_bytes_per_pixel,
-                                      dst_bytes_per_pixel, dst_buffer, src_buffer, false,
-                                      regs.dst.BlockHeight());
+            Texture::CopySwizzledData(regs.src.width, regs.src.height, regs.src.depth,
+                                      src_bytes_per_pixel, dst_bytes_per_pixel, dst_buffer,
+                                      src_buffer, false, regs.dst.BlockHeight(),
+                                      regs.dst.BlockDepth());
         }
     }
 }

src/video_core/engines/maxwell_3d.h

@@ -448,7 +448,10 @@ public:
                 BitField<8, 3, u32> block_depth;
                 BitField<12, 1, InvMemoryLayout> type;
             } memory_layout;
-            u32 array_mode;
+            union {
+                BitField<0, 16, u32> array_mode;
+                BitField<16, 1, u32> volume;
+            };
             u32 layer_stride;
             u32 base_layer;
             INSERT_PADDING_WORDS(7);

src/video_core/engines/maxwell_dma.cpp

@@ -68,12 +68,14 @@ void MaxwellDMA::HandleCopy() {
 
     if (regs.exec.is_dst_linear && !regs.exec.is_src_linear) {
         // If the input is tiled and the output is linear, deswizzle the input and copy it over.
-        Texture::CopySwizzledData(regs.src_params.size_x, regs.src_params.size_y, 1, 1, src_buffer,
-                                  dst_buffer, true, regs.src_params.BlockHeight());
+        Texture::CopySwizzledData(regs.src_params.size_x, regs.src_params.size_y,
+                                  regs.src_params.size_z, 1, 1, src_buffer, dst_buffer, true,
+                                  regs.src_params.BlockHeight(), regs.src_params.BlockDepth());
     } else {
         // If the input is linear and the output is tiled, swizzle the input and copy it over.
-        Texture::CopySwizzledData(regs.dst_params.size_x, regs.dst_params.size_y, 1, 1, dst_buffer,
-                                  src_buffer, false, regs.dst_params.BlockHeight());
+        Texture::CopySwizzledData(regs.dst_params.size_x, regs.dst_params.size_y,
+                                  regs.dst_params.size_z, 1, 1, dst_buffer, src_buffer, false,
+                                  regs.dst_params.BlockHeight(), regs.dst_params.BlockDepth());
     }
 }
 

src/video_core/engines/maxwell_dma.h

@@ -43,6 +43,10 @@ public:
             u32 BlockHeight() const {
                 return 1 << block_height;
             }
+
+            u32 BlockDepth() const {
+                return 1 << block_depth;
+            }
         };
 
         static_assert(sizeof(Parameters) == 24, "Parameters has wrong size");

src/video_core/engines/shader_bytecode.h

@@ -267,7 +267,7 @@ enum class ControlCode : u64 {
     GTU = 12,
     NEU = 13,
     GEU = 14,
-    //
+    T = 15,
     OFF = 16,
     LO = 17,
     SFF = 18,

src/video_core/memory_manager.cpp

@@ -87,6 +87,16 @@ GPUVAddr MemoryManager::UnmapBuffer(GPUVAddr gpu_addr, u64 size) {
     return gpu_addr;
 }
 
+GPUVAddr MemoryManager::GetRegionEnd(GPUVAddr region_start) const {
+    for (const auto& region : mapped_regions) {
+        const GPUVAddr region_end{region.gpu_addr + region.size};
+        if (region_start >= region.gpu_addr && region_start < region_end) {
+            return region_end;
+        }
+    }
+    return {};
+}
+
 boost::optional<GPUVAddr> MemoryManager::FindFreeBlock(u64 size, u64 align) {
     GPUVAddr gpu_addr = 0;
     u64 free_space = 0;

src/video_core/memory_manager.h

@@ -26,6 +26,7 @@ public:
     GPUVAddr MapBufferEx(VAddr cpu_addr, u64 size);
     GPUVAddr MapBufferEx(VAddr cpu_addr, GPUVAddr gpu_addr, u64 size);
     GPUVAddr UnmapBuffer(GPUVAddr gpu_addr, u64 size);
+    GPUVAddr GetRegionEnd(GPUVAddr region_start) const;
     boost::optional<VAddr> GpuToCpuAddress(GPUVAddr gpu_addr);
     std::vector<GPUVAddr> CpuToGpuAddress(VAddr cpu_addr) const;
 

src/video_core/rasterizer_cache.h

@@ -11,32 +11,77 @@
 
 #include "common/common_types.h"
 #include "core/core.h"
+#include "core/settings.h"
 #include "video_core/rasterizer_interface.h"
 #include "video_core/renderer_base.h"
 
+class RasterizerCacheObject {
+public:
+    /// Gets the address of the shader in guest memory, required for cache management
+    virtual VAddr GetAddr() const = 0;
+
+    /// Gets the size of the shader in guest memory, required for cache management
+    virtual std::size_t GetSizeInBytes() const = 0;
+
+    /// Wriets any cached resources back to memory
+    virtual void Flush() = 0;
+
+    /// Sets whether the cached object should be considered registered
+    void SetIsRegistered(bool registered) {
+        is_registered = registered;
+    }
+
+    /// Returns true if the cached object is registered
+    bool IsRegistered() const {
+        return is_registered;
+    }
+
+    /// Returns true if the cached object is dirty
+    bool IsDirty() const {
+        return is_dirty;
+    }
+
+    /// Returns ticks from when this cached object was last modified
+    u64 GetLastModifiedTicks() const {
+        return last_modified_ticks;
+    }
+
+    /// Marks an object as recently modified, used to specify whether it is clean or dirty
+    template <class T>
+    void MarkAsModified(bool dirty, T& cache) {
+        is_dirty = dirty;
+        last_modified_ticks = cache.GetModifiedTicks();
+    }
+
+private:
+    bool is_registered{};      ///< Whether the object is currently registered with the cache
+    bool is_dirty{};           ///< Whether the object is dirty (out of sync with guest memory)
+    u64 last_modified_ticks{}; ///< When the object was last modified, used for in-order flushing
+};
+
 template <class T>
 class RasterizerCache : NonCopyable {
+    friend class RasterizerCacheObject;
+
 public:
+    /// Write any cached resources overlapping the specified region back to memory
+    void FlushRegion(Tegra::GPUVAddr addr, size_t size) {
+        const auto& objects{GetSortedObjectsFromRegion(addr, size)};
+        for (auto& object : objects) {
+            FlushObject(object);
+        }
+    }
+
     /// Mark the specified region as being invalidated
     void InvalidateRegion(VAddr addr, u64 size) {
-        if (size == 0)
-            return;
-
-        const ObjectInterval interval{addr, addr + size};
-        for (auto& pair : boost::make_iterator_range(object_cache.equal_range(interval))) {
-            for (auto& cached_object : pair.second) {
-                if (!cached_object)
-                    continue;
-
-                remove_objects.emplace(cached_object);
+        const auto& objects{GetSortedObjectsFromRegion(addr, size)};
+        for (auto& object : objects) {
+            if (!object->IsRegistered()) {
+                // Skip duplicates
+                continue;
             }
+            Unregister(object);
         }
-
-        for (auto& remove_object : remove_objects) {
-            Unregister(remove_object);
-        }
-
-        remove_objects.clear();
     }
 
     /// Invalidates everything in the cache
@@ -62,6 +107,7 @@ protected:
 
     /// Register an object into the cache
     void Register(const T& object) {
+        object->SetIsRegistered(true);
         object_cache.add({GetInterval(object), ObjectSet{object}});
         auto& rasterizer = Core::System::GetInstance().Renderer().Rasterizer();
         rasterizer.UpdatePagesCachedCount(object->GetAddr(), object->GetSizeInBytes(), 1);
@@ -69,12 +115,57 @@ protected:
 
     /// Unregisters an object from the cache
     void Unregister(const T& object) {
+        object->SetIsRegistered(false);
         auto& rasterizer = Core::System::GetInstance().Renderer().Rasterizer();
         rasterizer.UpdatePagesCachedCount(object->GetAddr(), object->GetSizeInBytes(), -1);
+
+        // Only flush if use_accurate_gpu_emulation is enabled, as it incurs a performance hit
+        if (Settings::values.use_accurate_gpu_emulation) {
+            FlushObject(object);
+        }
+
         object_cache.subtract({GetInterval(object), ObjectSet{object}});
     }
 
+    /// Returns a ticks counter used for tracking when cached objects were last modified
+    u64 GetModifiedTicks() {
+        return ++modified_ticks;
+    }
+
 private:
+    /// Returns a list of cached objects from the specified memory region, ordered by access time
+    std::vector<T> GetSortedObjectsFromRegion(VAddr addr, u64 size) {
+        if (size == 0) {
+            return {};
+        }
+
+        std::vector<T> objects;
+        const ObjectInterval interval{addr, addr + size};
+        for (auto& pair : boost::make_iterator_range(object_cache.equal_range(interval))) {
+            for (auto& cached_object : pair.second) {
+                if (!cached_object) {
+                    continue;
+                }
+                objects.push_back(cached_object);
+            }
+        }
+
+        std::sort(objects.begin(), objects.end(), [](const T& a, const T& b) -> bool {
+            return a->GetLastModifiedTicks() < b->GetLastModifiedTicks();
+        });
+
+        return objects;
+    }
+
+    /// Flushes the specified object, updating appropriate cache state as needed
+    void FlushObject(const T& object) {
+        if (!object->IsDirty()) {
+            return;
+        }
+        object->Flush();
+        object->MarkAsModified(false, *this);
+    }
+
     using ObjectSet = std::set<T>;
     using ObjectCache = boost::icl::interval_map<VAddr, ObjectSet>;
     using ObjectInterval = typename ObjectCache::interval_type;
@@ -84,6 +175,6 @@ private:
                                           object->GetAddr() + object->GetSizeInBytes());
     }
 
-    ObjectCache object_cache;
-    ObjectSet remove_objects;
+    ObjectCache object_cache; ///< Cache of objects
+    u64 modified_ticks{};     ///< Counter of cache state ticks, used for in-order flushing
 };

src/video_core/renderer_opengl/gl_buffer_cache.h

@@ -15,15 +15,18 @@
 
 namespace OpenGL {
 
-struct CachedBufferEntry final {
-    VAddr GetAddr() const {
+struct CachedBufferEntry final : public RasterizerCacheObject {
+    VAddr GetAddr() const override {
         return addr;
     }
 
-    std::size_t GetSizeInBytes() const {
+    std::size_t GetSizeInBytes() const override {
         return size;
     }
 
+    // We do not have to flush this cache as things in it are never modified by us.
+    void Flush() override {}
+
     VAddr addr;
     std::size_t size;
     GLintptr offset;

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -286,7 +286,8 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
             &ubo, sizeof(ubo), static_cast<std::size_t>(uniform_buffer_alignment));
 
         // Bind the buffer
-        glBindBufferRange(GL_UNIFORM_BUFFER, stage, buffer_cache.GetHandle(), offset, sizeof(ubo));
+        glBindBufferRange(GL_UNIFORM_BUFFER, static_cast<GLuint>(stage), buffer_cache.GetHandle(),
+                          offset, static_cast<GLsizeiptr>(sizeof(ubo)));
 
         Shader shader{shader_cache.GetStageProgram(program)};
 
@@ -423,6 +424,13 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
             // Used when just a single color attachment is enabled, e.g. for clearing a color buffer
             Surface color_surface =
                 res_cache.GetColorBufferSurface(*single_color_target, preserve_contents);
+
+            if (color_surface) {
+                // Assume that a surface will be written to if it is used as a framebuffer, even if
+                // the shader doesn't actually write to it.
+                color_surface->MarkAsModified(true, res_cache);
+            }
+
             glFramebufferTexture2D(
                 GL_DRAW_FRAMEBUFFER,
                 GL_COLOR_ATTACHMENT0 + static_cast<GLenum>(*single_color_target), GL_TEXTURE_2D,
@@ -433,6 +441,13 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
             std::array<GLenum, Maxwell::NumRenderTargets> buffers;
             for (std::size_t index = 0; index < Maxwell::NumRenderTargets; ++index) {
                 Surface color_surface = res_cache.GetColorBufferSurface(index, preserve_contents);
+
+                if (color_surface) {
+                    // Assume that a surface will be written to if it is used as a framebuffer, even
+                    // if the shader doesn't actually write to it.
+                    color_surface->MarkAsModified(true, res_cache);
+                }
+
                 buffers[index] = GL_COLOR_ATTACHMENT0 + regs.rt_control.GetMap(index);
                 glFramebufferTexture2D(
                     GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + static_cast<GLenum>(index),
@@ -452,6 +467,10 @@ void RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb, bool using_dep
     }
 
     if (depth_surface) {
+        // Assume that a surface will be written to if it is used as a framebuffer, even if
+        // the shader doesn't actually write to it.
+        depth_surface->MarkAsModified(true, res_cache);
+
         if (regs.stencil_enable) {
             // Attach both depth and stencil
             glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
@@ -616,7 +635,14 @@ void RasterizerOpenGL::DrawArrays() {
 
 void RasterizerOpenGL::FlushAll() {}
 
-void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {}
+void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {
+    MICROPROFILE_SCOPE(OpenGL_CacheManagement);
+
+    if (Settings::values.use_accurate_gpu_emulation) {
+        // Only flush if use_accurate_gpu_emulation is enabled, as it incurs a performance hit
+        res_cache.FlushRegion(addr, size);
+    }
+}
 
 void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
@@ -626,6 +652,7 @@ void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
 }
 
 void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) {
+    FlushRegion(addr, size);
     InvalidateRegion(addr, size);
 }
 

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -34,16 +34,53 @@ struct FormatTuple {
     bool compressed;
 };
 
-static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
-    auto& gpu{Core::System::GetInstance().GPU()};
-    const auto cpu_addr{gpu.MemoryManager().GpuToCpuAddress(gpu_addr)};
-    return cpu_addr ? *cpu_addr : 0;
+static bool IsPixelFormatASTC(PixelFormat format) {
+    switch (format) {
+    case PixelFormat::ASTC_2D_4X4:
+    case PixelFormat::ASTC_2D_5X4:
+    case PixelFormat::ASTC_2D_8X8:
+    case PixelFormat::ASTC_2D_8X5:
+        return true;
+    default:
+        return false;
+    }
+}
+
+static std::pair<u32, u32> GetASTCBlockSize(PixelFormat format) {
+    switch (format) {
+    case PixelFormat::ASTC_2D_4X4:
+        return {4, 4};
+    case PixelFormat::ASTC_2D_5X4:
+        return {5, 4};
+    case PixelFormat::ASTC_2D_8X8:
+        return {8, 8};
+    case PixelFormat::ASTC_2D_8X5:
+        return {8, 5};
+    default:
+        LOG_CRITICAL(HW_GPU, "Unhandled format: {}", static_cast<u32>(format));
+        UNREACHABLE();
+    }
+}
+
+void SurfaceParams::InitCacheParameters(Tegra::GPUVAddr gpu_addr_) {
+    auto& memory_manager{Core::System::GetInstance().GPU().MemoryManager()};
+    const auto cpu_addr{memory_manager.GpuToCpuAddress(gpu_addr_)};
+
+    addr = cpu_addr ? *cpu_addr : 0;
+    gpu_addr = gpu_addr_;
+    size_in_bytes = SizeInBytesRaw();
+
+    if (IsPixelFormatASTC(pixel_format)) {
+        // ASTC is uncompressed in software, in emulated as RGBA8
+        size_in_bytes_gl = width * height * depth * 4;
+    } else {
+        size_in_bytes_gl = SizeInBytesGL();
+    }
 }
 
 /*static*/ SurfaceParams SurfaceParams::CreateForTexture(
     const Tegra::Texture::FullTextureInfo& config, const GLShader::SamplerEntry& entry) {
     SurfaceParams params{};
-    params.addr = TryGetCpuAddr(config.tic.Address());
     params.is_tiled = config.tic.IsTiled();
     params.block_width = params.is_tiled ? config.tic.BlockWidth() : 0,
     params.block_height = params.is_tiled ? config.tic.BlockHeight() : 0,
@@ -87,18 +124,18 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
         break;
     }
 
-    params.size_in_bytes_total = params.SizeInBytesTotal();
-    params.size_in_bytes_2d = params.SizeInBytes2D();
     params.max_mip_level = config.tic.max_mip_level + 1;
     params.rt = {};
 
+    params.InitCacheParameters(config.tic.Address());
+
     return params;
 }
 
 /*static*/ SurfaceParams SurfaceParams::CreateForFramebuffer(std::size_t index) {
     const auto& config{Core::System::GetInstance().GPU().Maxwell3D().regs.rt[index]};
     SurfaceParams params{};
-    params.addr = TryGetCpuAddr(config.Address());
+
     params.is_tiled =
         config.memory_layout.type == Tegra::Engines::Maxwell3D::Regs::InvMemoryLayout::BlockLinear;
     params.block_width = 1 << config.memory_layout.block_width;
@@ -112,16 +149,17 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     params.unaligned_height = config.height;
     params.target = SurfaceTarget::Texture2D;
     params.depth = 1;
-    params.size_in_bytes_total = params.SizeInBytesTotal();
-    params.size_in_bytes_2d = params.SizeInBytes2D();
     params.max_mip_level = 0;
 
     // Render target specific parameters, not used for caching
     params.rt.index = static_cast<u32>(index);
     params.rt.array_mode = config.array_mode;
     params.rt.layer_stride = config.layer_stride;
+    params.rt.volume = config.volume;
     params.rt.base_layer = config.base_layer;
 
+    params.InitCacheParameters(config.Address());
+
     return params;
 }
 
@@ -130,7 +168,7 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     u32 block_width, u32 block_height, u32 block_depth,
     Tegra::Engines::Maxwell3D::Regs::InvMemoryLayout type) {
     SurfaceParams params{};
-    params.addr = TryGetCpuAddr(zeta_address);
+
     params.is_tiled = type == Tegra::Engines::Maxwell3D::Regs::InvMemoryLayout::BlockLinear;
     params.block_width = 1 << std::min(block_width, 5U);
     params.block_height = 1 << std::min(block_height, 5U);
@@ -143,18 +181,18 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     params.unaligned_height = zeta_height;
     params.target = SurfaceTarget::Texture2D;
     params.depth = 1;
-    params.size_in_bytes_total = params.SizeInBytesTotal();
-    params.size_in_bytes_2d = params.SizeInBytes2D();
     params.max_mip_level = 0;
     params.rt = {};
 
+    params.InitCacheParameters(zeta_address);
+
     return params;
 }
 
 /*static*/ SurfaceParams SurfaceParams::CreateForFermiCopySurface(
     const Tegra::Engines::Fermi2D::Regs::Surface& config) {
     SurfaceParams params{};
-    params.addr = TryGetCpuAddr(config.Address());
+
     params.is_tiled = !config.linear;
     params.block_width = params.is_tiled ? std::min(config.BlockWidth(), 32U) : 0,
     params.block_height = params.is_tiled ? std::min(config.BlockHeight(), 32U) : 0,
@@ -167,11 +205,11 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     params.unaligned_height = config.height;
     params.target = SurfaceTarget::Texture2D;
     params.depth = 1;
-    params.size_in_bytes_total = params.SizeInBytesTotal();
-    params.size_in_bytes_2d = params.SizeInBytes2D();
     params.max_mip_level = 0;
     params.rt = {};
 
+    params.InitCacheParameters(config.Address());
+
     return params;
 }
 
@@ -231,6 +269,8 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {GL_RG32UI, GL_RG_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false},              // RG32UI
     {GL_R32UI, GL_RED_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false},              // R32UI
     {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X8
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X5
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_5X4
 
     // Depth formats
     {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, ComponentType::Float, false}, // Z32F
@@ -274,28 +314,6 @@ static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType
     return format;
 }
 
-static bool IsPixelFormatASTC(PixelFormat format) {
-    switch (format) {
-    case PixelFormat::ASTC_2D_4X4:
-    case PixelFormat::ASTC_2D_8X8:
-        return true;
-    default:
-        return false;
-    }
-}
-
-static std::pair<u32, u32> GetASTCBlockSize(PixelFormat format) {
-    switch (format) {
-    case PixelFormat::ASTC_2D_4X4:
-        return {4, 4};
-    case PixelFormat::ASTC_2D_8X8:
-        return {8, 8};
-    default:
-        LOG_CRITICAL(HW_GPU, "Unhandled format: {}", static_cast<u32>(format));
-        UNREACHABLE();
-    }
-}
-
 MathUtil::Rectangle<u32> SurfaceParams::GetRect() const {
     u32 actual_height{unaligned_height};
     if (IsPixelFormatASTC(pixel_format)) {
@@ -323,29 +341,27 @@ static bool IsFormatBCn(PixelFormat format) {
 }
 
 template <bool morton_to_gl, PixelFormat format>
-void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, std::size_t gl_buffer_size,
-                VAddr addr) {
-    constexpr u32 bytes_per_pixel = SurfaceParams::GetFormatBpp(format) / CHAR_BIT;
-    constexpr u32 gl_bytes_per_pixel = CachedSurface::GetGLBytesPerPixel(format);
+void MortonCopy(u32 stride, u32 block_height, u32 height, u32 block_depth, u32 depth, u8* gl_buffer,
+                std::size_t gl_buffer_size, VAddr addr) {
+    constexpr u32 bytes_per_pixel = SurfaceParams::GetBytesPerPixel(format);
+
+    // With the BCn formats (DXT and DXN), each 4x4 tile is swizzled instead of just individual
+    // pixel values.
+    const u32 tile_size{IsFormatBCn(format) ? 4U : 1U};
 
     if (morton_to_gl) {
-        // With the BCn formats (DXT and DXN), each 4x4 tile is swizzled instead of just individual
-        // pixel values.
-        const u32 tile_size{IsFormatBCn(format) ? 4U : 1U};
         const std::vector<u8> data = Tegra::Texture::UnswizzleTexture(
-            addr, tile_size, bytes_per_pixel, stride, height, block_height);
+            addr, tile_size, bytes_per_pixel, stride, height, depth, block_height, block_depth);
         const std::size_t size_to_copy{std::min(gl_buffer_size, data.size())};
         memcpy(gl_buffer, data.data(), size_to_copy);
     } else {
-        // TODO(bunnei): Assumes the default rendering GOB size of 16 (128 lines). We should
-        // check the configuration for this and perform more generic un/swizzle
-        LOG_WARNING(Render_OpenGL, "need to use correct swizzle/GOB parameters!");
-        VideoCore::MortonCopyPixels128(stride, height, bytes_per_pixel, gl_bytes_per_pixel,
-                                       Memory::GetPointer(addr), gl_buffer, morton_to_gl);
+        Tegra::Texture::CopySwizzledData(stride / tile_size, height / tile_size, depth,
+                                         bytes_per_pixel, bytes_per_pixel, Memory::GetPointer(addr),
+                                         gl_buffer, false, block_height, block_depth);
     }
 }
 
-static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr),
+static constexpr std::array<void (*)(u32, u32, u32, u32, u32, u8*, std::size_t, VAddr),
                             SurfaceParams::MaxPixelFormat>
     morton_to_gl_fns = {
         // clang-format off
@@ -395,6 +411,8 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr),
         MortonCopy<true, PixelFormat::RG32UI>,
         MortonCopy<true, PixelFormat::R32UI>,
         MortonCopy<true, PixelFormat::ASTC_2D_8X8>,
+        MortonCopy<true, PixelFormat::ASTC_2D_8X5>,
+        MortonCopy<true, PixelFormat::ASTC_2D_5X4>,
         MortonCopy<true, PixelFormat::Z32F>,
         MortonCopy<true, PixelFormat::Z16>,
         MortonCopy<true, PixelFormat::Z24S8>,
@@ -403,7 +421,7 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr),
         // clang-format on
 };
 
-static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr),
+static constexpr std::array<void (*)(u32, u32, u32, u32, u32, u8*, std::size_t, VAddr),
                             SurfaceParams::MaxPixelFormat>
     gl_to_morton_fns = {
         // clang-format off
@@ -420,17 +438,16 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr),
         MortonCopy<false, PixelFormat::RGBA16UI>,
         MortonCopy<false, PixelFormat::R11FG11FB10F>,
         MortonCopy<false, PixelFormat::RGBA32UI>,
-        // TODO(Subv): Swizzling DXT1/DXT23/DXT45/DXN1/DXN2/BC7U/BC6H_UF16/BC6H_SF16/ASTC_2D_4X4
-        // formats are not supported
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
+        MortonCopy<false, PixelFormat::DXT1>,
+        MortonCopy<false, PixelFormat::DXT23>,
+        MortonCopy<false, PixelFormat::DXT45>,
+        MortonCopy<false, PixelFormat::DXN1>,
+        MortonCopy<false, PixelFormat::DXN2UNORM>,
+        MortonCopy<false, PixelFormat::DXN2SNORM>,
+        MortonCopy<false, PixelFormat::BC7U>,
+        MortonCopy<false, PixelFormat::BC6H_UF16>,
+        MortonCopy<false, PixelFormat::BC6H_SF16>,
+        // TODO(Subv): Swizzling ASTC formats are not supported
         nullptr,
         MortonCopy<false, PixelFormat::G8R8U>,
         MortonCopy<false, PixelFormat::G8R8S>,
@@ -455,6 +472,8 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr),
         MortonCopy<false, PixelFormat::RG32UI>,
         MortonCopy<false, PixelFormat::R32UI>,
         nullptr,
+        nullptr,
+        nullptr,
         MortonCopy<false, PixelFormat::Z32F>,
         MortonCopy<false, PixelFormat::Z16>,
         MortonCopy<false, PixelFormat::Z24S8>,
@@ -614,22 +633,21 @@ static void CopySurface(const Surface& src_surface, const Surface& dst_surface,
     auto source_format = GetFormatTuple(src_params.pixel_format, src_params.component_type);
     auto dest_format = GetFormatTuple(dst_params.pixel_format, dst_params.component_type);
 
-    std::size_t buffer_size =
-        std::max(src_params.size_in_bytes_total, dst_params.size_in_bytes_total);
+    std::size_t buffer_size = std::max(src_params.size_in_bytes, dst_params.size_in_bytes);
 
     glBindBuffer(GL_PIXEL_PACK_BUFFER, copy_pbo_handle);
     glBufferData(GL_PIXEL_PACK_BUFFER, buffer_size, nullptr, GL_STREAM_DRAW_ARB);
     if (source_format.compressed) {
         glGetCompressedTextureImage(src_surface->Texture().handle, src_attachment,
-                                    static_cast<GLsizei>(src_params.size_in_bytes_total), nullptr);
+                                    static_cast<GLsizei>(src_params.size_in_bytes), nullptr);
     } else {
         glGetTextureImage(src_surface->Texture().handle, src_attachment, source_format.format,
-                          source_format.type, static_cast<GLsizei>(src_params.size_in_bytes_total),
+                          source_format.type, static_cast<GLsizei>(src_params.size_in_bytes),
                           nullptr);
     }
     // If the new texture is bigger than the previous one, we need to fill in the rest with data
     // from the CPU.
-    if (src_params.size_in_bytes_total < dst_params.size_in_bytes_total) {
+    if (src_params.size_in_bytes < dst_params.size_in_bytes) {
         // Upload the rest of the memory.
         if (dst_params.is_tiled) {
             // TODO(Subv): We might have to de-tile the subtexture and re-tile it with the rest
@@ -639,12 +657,12 @@ static void CopySurface(const Surface& src_surface, const Surface& dst_surface,
             LOG_DEBUG(HW_GPU, "Trying to upload extra texture data from the CPU during "
                               "reinterpretation but the texture is tiled.");
         }
-        std::size_t remaining_size =
-            dst_params.size_in_bytes_total - src_params.size_in_bytes_total;
+        std::size_t remaining_size = dst_params.size_in_bytes - src_params.size_in_bytes;
         std::vector<u8> data(remaining_size);
-        Memory::ReadBlock(dst_params.addr + src_params.size_in_bytes_total, data.data(),
-                          data.size());
-        glBufferSubData(GL_PIXEL_PACK_BUFFER, src_params.size_in_bytes_total, remaining_size,
+        std::memcpy(data.data(), Memory::GetPointer(dst_params.addr + src_params.size_in_bytes),
+                    data.size());
+
+        glBufferSubData(GL_PIXEL_PACK_BUFFER, src_params.size_in_bytes, remaining_size,
                         data.data());
     }
 
@@ -690,7 +708,8 @@ static void CopySurface(const Surface& src_surface, const Surface& dst_surface,
 }
 
 CachedSurface::CachedSurface(const SurfaceParams& params)
-    : params(params), gl_target(SurfaceTargetToGL(params.target)) {
+    : params(params), gl_target(SurfaceTargetToGL(params.target)),
+      cached_size_in_bytes(params.size_in_bytes) {
     texture.Create();
     const auto& rect{params.GetRect()};
 
@@ -740,9 +759,21 @@ CachedSurface::CachedSurface(const SurfaceParams& params)
 
     VideoCore::LabelGLObject(GL_TEXTURE, texture.handle, params.addr,
                              SurfaceParams::SurfaceTargetName(params.target));
+
+    // Clamp size to mapped GPU memory region
+    // TODO(bunnei): Super Mario Odyssey maps a 0x40000 byte region and then uses it for a 0x80000
+    // R32F render buffer. We do not yet know if this is a game bug or something else, but this
+    // check is necessary to prevent flushing from overwriting unmapped memory.
+
+    auto& memory_manager{Core::System::GetInstance().GPU().MemoryManager()};
+    const u64 max_size{memory_manager.GetRegionEnd(params.gpu_addr) - params.gpu_addr};
+    if (cached_size_in_bytes > max_size) {
+        LOG_ERROR(HW_GPU, "Surface size {} exceeds region size {}", params.size_in_bytes, max_size);
+        cached_size_in_bytes = max_size;
+    }
 }
 
-static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
+static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height, bool reverse) {
     union S8Z24 {
         BitField<0, 24, u32> z24;
         BitField<24, 8, u32> s8;
@@ -755,22 +786,29 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
     };
     static_assert(sizeof(Z24S8) == 4, "Z24S8 is incorrect size");
 
-    S8Z24 input_pixel{};
-    Z24S8 output_pixel{};
-    constexpr auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::S8Z24)};
+    S8Z24 s8z24_pixel{};
+    Z24S8 z24s8_pixel{};
+    constexpr auto bpp{SurfaceParams::GetBytesPerPixel(PixelFormat::S8Z24)};
     for (std::size_t y = 0; y < height; ++y) {
         for (std::size_t x = 0; x < width; ++x) {
             const std::size_t offset{bpp * (y * width + x)};
-            std::memcpy(&input_pixel, &data[offset], sizeof(S8Z24));
-            output_pixel.s8.Assign(input_pixel.s8);
-            output_pixel.z24.Assign(input_pixel.z24);
-            std::memcpy(&data[offset], &output_pixel, sizeof(Z24S8));
+            if (reverse) {
+                std::memcpy(&z24s8_pixel, &data[offset], sizeof(Z24S8));
+                s8z24_pixel.s8.Assign(z24s8_pixel.s8);
+                s8z24_pixel.z24.Assign(z24s8_pixel.z24);
+                std::memcpy(&data[offset], &s8z24_pixel, sizeof(S8Z24));
+            } else {
+                std::memcpy(&s8z24_pixel, &data[offset], sizeof(S8Z24));
+                z24s8_pixel.s8.Assign(s8z24_pixel.s8);
+                z24s8_pixel.z24.Assign(s8z24_pixel.z24);
+                std::memcpy(&data[offset], &z24s8_pixel, sizeof(Z24S8));
+            }
         }
     }
 }
 
 static void ConvertG8R8ToR8G8(std::vector<u8>& data, u32 width, u32 height) {
-    constexpr auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::G8R8U)};
+    constexpr auto bpp{SurfaceParams::GetBytesPerPixel(PixelFormat::G8R8U)};
     for (std::size_t y = 0; y < height; ++y) {
         for (std::size_t x = 0; x < width; ++x) {
             const std::size_t offset{bpp * (y * width + x)};
@@ -790,7 +828,9 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
                                                u32 width, u32 height) {
     switch (pixel_format) {
     case PixelFormat::ASTC_2D_4X4:
-    case PixelFormat::ASTC_2D_8X8: {
+    case PixelFormat::ASTC_2D_8X8:
+    case PixelFormat::ASTC_2D_8X5:
+    case PixelFormat::ASTC_2D_5X4: {
         // Convert ASTC pixel formats to RGBA8, as most desktop GPUs do not support ASTC.
         u32 block_width{};
         u32 block_height{};
@@ -800,7 +840,7 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
     }
     case PixelFormat::S8Z24:
         // Convert the S8Z24 depth format to Z24S8, as OpenGL does not support S8Z24.
-        ConvertS8Z24ToZ24S8(data, width, height);
+        ConvertS8Z24ToZ24S8(data, width, height, false);
         break;
 
     case PixelFormat::G8R8U:
@@ -811,54 +851,54 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
     }
 }
 
+/**
+ * Helper function to perform software conversion (as needed) when flushing a buffer from OpenGL to
+ * Switch memory. This is for Maxwell pixel formats that cannot be represented as-is in OpenGL or
+ * with typical desktop GPUs.
+ */
+static void ConvertFormatAsNeeded_FlushGLBuffer(std::vector<u8>& data, PixelFormat pixel_format,
+                                                u32 width, u32 height) {
+    switch (pixel_format) {
+    case PixelFormat::G8R8U:
+    case PixelFormat::G8R8S:
+    case PixelFormat::ASTC_2D_4X4:
+    case PixelFormat::ASTC_2D_8X8: {
+        LOG_CRITICAL(HW_GPU, "Conversion of format {} after texture flushing is not implemented",
+                     static_cast<u32>(pixel_format));
+        UNREACHABLE();
+        break;
+    }
+    case PixelFormat::S8Z24:
+        // Convert the Z24S8 depth format to S8Z24, as OpenGL does not support S8Z24.
+        ConvertS8Z24ToZ24S8(data, width, height, true);
+        break;
+    }
+}
+
 MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64, 192));
 void CachedSurface::LoadGLBuffer() {
-    ASSERT(params.type != SurfaceType::Fill);
-
-    const u8* const texture_src_data = Memory::GetPointer(params.addr);
-
-    ASSERT(texture_src_data);
-
-    const u32 bytes_per_pixel = GetGLBytesPerPixel(params.pixel_format);
-    const u32 copy_size = params.width * params.height * bytes_per_pixel;
-    const std::size_t total_size = copy_size * params.depth;
-
     MICROPROFILE_SCOPE(OpenGL_SurfaceLoad);
 
+    gl_buffer.resize(params.size_in_bytes_gl);
     if (params.is_tiled) {
-        gl_buffer.resize(total_size);
+        u32 depth = params.depth;
+        u32 block_depth = params.block_depth;
 
         ASSERT_MSG(params.block_width == 1, "Block width is defined as {} on texture type {}",
                    params.block_width, static_cast<u32>(params.target));
-        ASSERT_MSG(params.block_depth == 1, "Block depth is defined as {} on texture type {}",
-                   params.block_depth, static_cast<u32>(params.target));
 
-        // TODO(bunnei): This only unswizzles and copies a 2D texture - we do not yet know how to do
-        // this for 3D textures, etc.
-        switch (params.target) {
-        case SurfaceParams::SurfaceTarget::Texture2D:
-            // Pass impl. to the fallback code below
-            break;
-        case SurfaceParams::SurfaceTarget::Texture2DArray:
-        case SurfaceParams::SurfaceTarget::TextureCubemap:
-            for (std::size_t index = 0; index < params.depth; ++index) {
-                const std::size_t offset{index * copy_size};
-                morton_to_gl_fns[static_cast<std::size_t>(params.pixel_format)](
-                    params.width, params.block_height, params.height, gl_buffer.data() + offset,
-                    copy_size, params.addr + offset);
-            }
-            break;
-        default:
-            LOG_CRITICAL(HW_GPU, "Unimplemented tiled load for target={}",
-                         static_cast<u32>(params.target));
-            UNREACHABLE();
+        if (params.target == SurfaceParams::SurfaceTarget::Texture2D) {
+            // TODO(Blinkhawk): Eliminate this condition once all texture types are implemented.
+            depth = 1U;
+            block_depth = 1U;
         }
 
         morton_to_gl_fns[static_cast<std::size_t>(params.pixel_format)](
-            params.width, params.block_height, params.height, gl_buffer.data(), copy_size,
-            params.addr);
+            params.width, params.block_height, params.height, block_depth, depth, gl_buffer.data(),
+            gl_buffer.size(), params.addr);
     } else {
-        const u8* const texture_src_data_end{texture_src_data + total_size};
+        const auto texture_src_data{Memory::GetPointer(params.addr)};
+        const auto texture_src_data_end{texture_src_data + params.size_in_bytes_gl};
         gl_buffer.assign(texture_src_data, texture_src_data_end);
     }
 
@@ -867,7 +907,44 @@ void CachedSurface::LoadGLBuffer() {
 
 MICROPROFILE_DEFINE(OpenGL_SurfaceFlush, "OpenGL", "Surface Flush", MP_RGB(128, 192, 64));
 void CachedSurface::FlushGLBuffer() {
-    ASSERT_MSG(false, "Unimplemented");
+    MICROPROFILE_SCOPE(OpenGL_SurfaceFlush);
+
+    ASSERT_MSG(!IsPixelFormatASTC(params.pixel_format), "Unimplemented");
+
+    // OpenGL temporary buffer needs to be big enough to store raw texture size
+    gl_buffer.resize(GetSizeInBytes());
+
+    const FormatTuple& tuple = GetFormatTuple(params.pixel_format, params.component_type);
+    // Ensure no bad interactions with GL_UNPACK_ALIGNMENT
+    ASSERT(params.width * SurfaceParams::GetBytesPerPixel(params.pixel_format) % 4 == 0);
+    glPixelStorei(GL_PACK_ROW_LENGTH, static_cast<GLint>(params.width));
+    ASSERT(!tuple.compressed);
+    glBindBuffer(GL_PIXEL_PACK_BUFFER, 0);
+    glGetTextureImage(texture.handle, 0, tuple.format, tuple.type, gl_buffer.size(),
+                      gl_buffer.data());
+    glPixelStorei(GL_PACK_ROW_LENGTH, 0);
+    ConvertFormatAsNeeded_FlushGLBuffer(gl_buffer, params.pixel_format, params.width,
+                                        params.height);
+    ASSERT(params.type != SurfaceType::Fill);
+    const u8* const texture_src_data = Memory::GetPointer(params.addr);
+    ASSERT(texture_src_data);
+    if (params.is_tiled) {
+        u32 depth = params.depth;
+        u32 block_depth = params.block_depth;
+
+        ASSERT_MSG(params.block_width == 1, "Block width is defined as {} on texture type {}",
+                   params.block_width, static_cast<u32>(params.target));
+
+        if (params.target == SurfaceParams::SurfaceTarget::Texture2D) {
+            // TODO(Blinkhawk): Eliminate this condition once all texture types are implemented.
+            depth = 1U;
+        }
+        gl_to_morton_fns[static_cast<size_t>(params.pixel_format)](
+            params.width, params.block_height, params.height, block_depth, depth, gl_buffer.data(),
+            gl_buffer.size(), GetAddr());
+    } else {
+        std::memcpy(Memory::GetPointer(GetAddr()), gl_buffer.data(), GetSizeInBytes());
+    }
 }
 
 MICROPROFILE_DEFINE(OpenGL_TextureUL, "OpenGL", "Texture Upload", MP_RGB(128, 64, 192));
@@ -877,9 +954,6 @@ void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle
 
     MICROPROFILE_SCOPE(OpenGL_TextureUL);
 
-    ASSERT(gl_buffer.size() == static_cast<std::size_t>(params.width) * params.height *
-                                   GetGLBytesPerPixel(params.pixel_format) * params.depth);
-
     const auto& rect{params.GetRect()};
 
     // Load data from memory to the surface
@@ -888,7 +962,7 @@ void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle
     std::size_t buffer_offset =
         static_cast<std::size_t>(static_cast<std::size_t>(y0) * params.width +
                                  static_cast<std::size_t>(x0)) *
-        GetGLBytesPerPixel(params.pixel_format);
+        SurfaceParams::GetBytesPerPixel(params.pixel_format);
 
     const FormatTuple& tuple = GetFormatTuple(params.pixel_format, params.component_type);
     const GLuint target_tex = texture.handle;
@@ -904,7 +978,7 @@ void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle
     cur_state.Apply();
 
     // Ensure no bad interactions with GL_UNPACK_ALIGNMENT
-    ASSERT(params.width * GetGLBytesPerPixel(params.pixel_format) % 4 == 0);
+    ASSERT(params.width * SurfaceParams::GetBytesPerPixel(params.pixel_format) % 4 == 0);
     glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(params.width));
 
     glActiveTexture(GL_TEXTURE0);
@@ -914,7 +988,7 @@ void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle
             glCompressedTexImage2D(
                 SurfaceTargetToGL(params.target), 0, tuple.internal_format,
                 static_cast<GLsizei>(params.width), static_cast<GLsizei>(params.height), 0,
-                static_cast<GLsizei>(params.size_in_bytes_2d), &gl_buffer[buffer_offset]);
+                static_cast<GLsizei>(params.size_in_bytes_gl), &gl_buffer[buffer_offset]);
             break;
         case SurfaceParams::SurfaceTarget::Texture3D:
         case SurfaceParams::SurfaceTarget::Texture2DArray:
@@ -922,16 +996,16 @@ void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle
                 SurfaceTargetToGL(params.target), 0, tuple.internal_format,
                 static_cast<GLsizei>(params.width), static_cast<GLsizei>(params.height),
                 static_cast<GLsizei>(params.depth), 0,
-                static_cast<GLsizei>(params.size_in_bytes_total), &gl_buffer[buffer_offset]);
+                static_cast<GLsizei>(params.size_in_bytes_gl), &gl_buffer[buffer_offset]);
             break;
         case SurfaceParams::SurfaceTarget::TextureCubemap:
             for (std::size_t face = 0; face < params.depth; ++face) {
                 glCompressedTexImage2D(static_cast<GLenum>(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face),
                                        0, tuple.internal_format, static_cast<GLsizei>(params.width),
                                        static_cast<GLsizei>(params.height), 0,
-                                       static_cast<GLsizei>(params.size_in_bytes_2d),
+                                       static_cast<GLsizei>(params.SizeInBytesCubeFaceGL()),
                                        &gl_buffer[buffer_offset]);
-                buffer_offset += params.size_in_bytes_2d;
+                buffer_offset += params.SizeInBytesCubeFace();
             }
             break;
         default:
@@ -941,7 +1015,7 @@ void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle
             glCompressedTexImage2D(
                 GL_TEXTURE_2D, 0, tuple.internal_format, static_cast<GLsizei>(params.width),
                 static_cast<GLsizei>(params.height), 0,
-                static_cast<GLsizei>(params.size_in_bytes_2d), &gl_buffer[buffer_offset]);
+                static_cast<GLsizei>(params.size_in_bytes_gl), &gl_buffer[buffer_offset]);
         }
     } else {
 
@@ -970,7 +1044,7 @@ void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle
                                 y0, static_cast<GLsizei>(rect.GetWidth()),
                                 static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
                                 &gl_buffer[buffer_offset]);
-                buffer_offset += params.size_in_bytes_2d;
+                buffer_offset += params.SizeInBytesCubeFace();
             }
             break;
         default:
@@ -1032,10 +1106,7 @@ Surface RasterizerCacheOpenGL::GetColorBufferSurface(std::size_t index, bool pre
 void RasterizerCacheOpenGL::LoadSurface(const Surface& surface) {
     surface->LoadGLBuffer();
     surface->UploadGLTexture(read_framebuffer.handle, draw_framebuffer.handle);
-}
-
-void RasterizerCacheOpenGL::FlushSurface(const Surface& surface) {
-    surface->FlushGLBuffer();
+    surface->MarkAsModified(false, *this);
 }
 
 Surface RasterizerCacheOpenGL::GetSurface(const SurfaceParams& params, bool preserve_contents) {
@@ -1052,8 +1123,8 @@ Surface RasterizerCacheOpenGL::GetSurface(const SurfaceParams& params, bool pres
         } else if (preserve_contents) {
             // If surface parameters changed and we care about keeping the previous data, recreate
             // the surface from the old one
-            Unregister(surface);
             Surface new_surface{RecreateSurface(surface, params)};
+            Unregister(surface);
             Register(new_surface);
             return new_surface;
         } else {
@@ -1104,6 +1175,14 @@ void RasterizerCacheOpenGL::FermiCopySurface(
     FastCopySurface(GetSurface(src_params, true), GetSurface(dst_params, false));
 }
 
+void RasterizerCacheOpenGL::AccurateCopySurface(const Surface& src_surface,
+                                                const Surface& dst_surface) {
+    const auto& src_params{src_surface->GetSurfaceParams()};
+    const auto& dst_params{dst_surface->GetSurfaceParams()};
+    FlushRegion(src_params.addr, dst_params.size_in_bytes);
+    LoadSurface(dst_surface);
+}
+
 Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
                                                const SurfaceParams& new_params) {
     // Verify surface is compatible for blitting
@@ -1112,6 +1191,12 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
     // Get a new surface with the new parameters, and blit the previous surface to it
     Surface new_surface{GetUncachedSurface(new_params)};
 
+    // With use_accurate_gpu_emulation enabled, do an accurate surface copy
+    if (Settings::values.use_accurate_gpu_emulation) {
+        AccurateCopySurface(old_surface, new_surface);
+        return new_surface;
+    }
+
     // For compatible surfaces, we can just do fast glCopyImageSubData based copy
     if (old_params.target == new_params.target && old_params.type == new_params.type &&
         old_params.depth == new_params.depth && old_params.depth == 1 &&
@@ -1123,11 +1208,10 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
 
     // If the format is the same, just do a framebuffer blit. This is significantly faster than
     // using PBOs. The is also likely less accurate, as textures will be converted rather than
-    // reinterpreted. When use_accurate_framebuffers setting is enabled, perform a more accurate
+    // reinterpreted. When use_accurate_gpu_emulation setting is enabled, perform a more accurate
     // surface copy, where pixels are reinterpreted as a new format (without conversion). This
     // code path uses OpenGL PBOs and is quite slow.
-    const bool is_blit{old_params.pixel_format == new_params.pixel_format ||
-                       !Settings::values.use_accurate_framebuffers};
+    const bool is_blit{old_params.pixel_format == new_params.pixel_format};
 
     switch (new_params.target) {
     case SurfaceParams::SurfaceTarget::Texture2D:
@@ -1137,6 +1221,9 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
             CopySurface(old_surface, new_surface, copy_pbo.handle);
         }
         break;
+    case SurfaceParams::SurfaceTarget::Texture3D:
+        AccurateCopySurface(old_surface, new_surface);
+        break;
     case SurfaceParams::SurfaceTarget::TextureCubemap: {
         if (old_params.rt.array_mode != 1) {
             // TODO(bunnei): This is used by Breath of the Wild, I'm not sure how to implement this

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -18,6 +18,7 @@
 #include "video_core/rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/renderer_opengl/gl_shader_gen.h"
+#include "video_core/textures/decoders.h"
 #include "video_core/textures/texture.h"
 
 namespace OpenGL {
@@ -74,19 +75,21 @@ struct SurfaceParams {
         RG32UI = 43,
         R32UI = 44,
         ASTC_2D_8X8 = 45,
+        ASTC_2D_8X5 = 46,
+        ASTC_2D_5X4 = 47,
 
         MaxColorFormat,
 
         // Depth formats
-        Z32F = 46,
-        Z16 = 47,
+        Z32F = 48,
+        Z16 = 49,
 
         MaxDepthFormat,
 
         // DepthStencil formats
-        Z24S8 = 48,
-        S8Z24 = 49,
-        Z32FS8 = 50,
+        Z24S8 = 50,
+        S8Z24 = 51,
+        Z32FS8 = 52,
 
         MaxDepthStencilFormat,
 
@@ -129,6 +132,8 @@ struct SurfaceParams {
         case Tegra::Texture::TextureType::Texture2D:
         case Tegra::Texture::TextureType::Texture2DNoMipmap:
             return SurfaceTarget::Texture2D;
+        case Tegra::Texture::TextureType::Texture3D:
+            return SurfaceTarget::Texture3D;
         case Tegra::Texture::TextureType::TextureCubemap:
             return SurfaceTarget::TextureCubemap;
         case Tegra::Texture::TextureType::Texture1DArray:
@@ -220,6 +225,8 @@ struct SurfaceParams {
             1, // RG32UI
             1, // R32UI
             4, // ASTC_2D_8X8
+            4, // ASTC_2D_8X5
+            4, // ASTC_2D_5X4
             1, // Z32F
             1, // Z16
             1, // Z24S8
@@ -282,6 +289,8 @@ struct SurfaceParams {
             64,  // RG32UI
             32,  // R32UI
             16,  // ASTC_2D_8X8
+            32,  // ASTC_2D_8X5
+            32,  // ASTC_2D_5X4
             32,  // Z32F
             16,  // Z16
             32,  // Z24S8
@@ -553,8 +562,12 @@ struct SurfaceParams {
             return PixelFormat::BC6H_SF16;
         case Tegra::Texture::TextureFormat::ASTC_2D_4X4:
             return PixelFormat::ASTC_2D_4X4;
+        case Tegra::Texture::TextureFormat::ASTC_2D_5X4:
+            return PixelFormat::ASTC_2D_5X4;
         case Tegra::Texture::TextureFormat::ASTC_2D_8X8:
             return PixelFormat::ASTC_2D_8X8;
+        case Tegra::Texture::TextureFormat::ASTC_2D_8X5:
+            return PixelFormat::ASTC_2D_8X5;
         case Tegra::Texture::TextureFormat::R16_G16:
             switch (component_type) {
             case Tegra::Texture::ComponentType::FLOAT:
@@ -691,21 +704,42 @@ struct SurfaceParams {
         return SurfaceType::Invalid;
     }
 
+    /// Returns the sizer in bytes of the specified pixel format
+    static constexpr u32 GetBytesPerPixel(PixelFormat pixel_format) {
+        if (pixel_format == SurfaceParams::PixelFormat::Invalid) {
+            return 0;
+        }
+        return GetFormatBpp(pixel_format) / CHAR_BIT;
+    }
+
     /// Returns the rectangle corresponding to this surface
     MathUtil::Rectangle<u32> GetRect() const;
 
-    /// Returns the size of this surface as a 2D texture in bytes, adjusted for compression
-    std::size_t SizeInBytes2D() const {
+    /// Returns the total size of this surface in bytes, adjusted for compression
+    std::size_t SizeInBytesRaw(bool ignore_tiled = false) const {
         const u32 compression_factor{GetCompressionFactor(pixel_format)};
-        ASSERT(width % compression_factor == 0);
-        ASSERT(height % compression_factor == 0);
-        return (width / compression_factor) * (height / compression_factor) *
-               GetFormatBpp(pixel_format) / CHAR_BIT;
+        const u32 bytes_per_pixel{GetBytesPerPixel(pixel_format)};
+        const size_t uncompressed_size{
+            Tegra::Texture::CalculateSize((ignore_tiled ? false : is_tiled), bytes_per_pixel, width,
+                                          height, depth, block_height, block_depth)};
+
+        // Divide by compression_factor^2, as height and width are factored by this
+        return uncompressed_size / (compression_factor * compression_factor);
     }
 
-    /// Returns the total size of this surface in bytes, adjusted for compression
-    std::size_t SizeInBytesTotal() const {
-        return SizeInBytes2D() * depth;
+    /// Returns the size of this surface as an OpenGL texture in bytes
+    std::size_t SizeInBytesGL() const {
+        return SizeInBytesRaw(true);
+    }
+
+    /// Returns the size of this surface as a cube face in bytes
+    std::size_t SizeInBytesCubeFace() const {
+        return size_in_bytes / 6;
+    }
+
+    /// Returns the size of this surface as an OpenGL cube face in bytes
+    std::size_t SizeInBytesCubeFaceGL() const {
+        return size_in_bytes_gl / 6;
     }
 
     /// Creates SurfaceParams from a texture configuration
@@ -732,7 +766,9 @@ struct SurfaceParams {
                         other.depth);
     }
 
-    VAddr addr;
+    /// Initializes parameters for caching, should be called after everything has been initialized
+    void InitCacheParameters(Tegra::GPUVAddr gpu_addr);
+
     bool is_tiled;
     u32 block_width;
     u32 block_height;
@@ -744,15 +780,20 @@ struct SurfaceParams {
     u32 height;
     u32 depth;
     u32 unaligned_height;
-    std::size_t size_in_bytes_total;
-    std::size_t size_in_bytes_2d;
     SurfaceTarget target;
     u32 max_mip_level;
 
+    // Parameters used for caching
+    VAddr addr;
+    Tegra::GPUVAddr gpu_addr;
+    std::size_t size_in_bytes;
+    std::size_t size_in_bytes_gl;
+
     // Render target specific parameters, not used in caching
     struct {
         u32 index;
         u32 array_mode;
+        u32 volume;
         u32 layer_stride;
         u32 base_layer;
     } rt;
@@ -765,7 +806,8 @@ struct SurfaceReserveKey : Common::HashableStruct<OpenGL::SurfaceParams> {
     static SurfaceReserveKey Create(const OpenGL::SurfaceParams& params) {
         SurfaceReserveKey res;
         res.state = params;
-        res.state.rt = {}; // Ignore rt config in caching
+        res.state.gpu_addr = {}; // Ignore GPU vaddr in caching
+        res.state.rt = {};       // Ignore rt config in caching
         return res;
     }
 };
@@ -780,16 +822,20 @@ struct hash<SurfaceReserveKey> {
 
 namespace OpenGL {
 
-class CachedSurface final {
+class CachedSurface final : public RasterizerCacheObject {
 public:
     CachedSurface(const SurfaceParams& params);
 
-    VAddr GetAddr() const {
+    VAddr GetAddr() const override {
         return params.addr;
     }
 
-    std::size_t GetSizeInBytes() const {
-        return params.size_in_bytes_total;
+    std::size_t GetSizeInBytes() const override {
+        return cached_size_in_bytes;
+    }
+
+    void Flush() override {
+        FlushGLBuffer();
     }
 
     const OGLTexture& Texture() const {
@@ -800,13 +846,6 @@ public:
         return gl_target;
     }
 
-    static constexpr unsigned int GetGLBytesPerPixel(SurfaceParams::PixelFormat format) {
-        if (format == SurfaceParams::PixelFormat::Invalid)
-            return 0;
-
-        return SurfaceParams::GetFormatBpp(format) / CHAR_BIT;
-    }
-
     const SurfaceParams& GetSurfaceParams() const {
         return params;
     }
@@ -823,6 +862,7 @@ private:
     std::vector<u8> gl_buffer;
     SurfaceParams params;
     GLenum gl_target;
+    std::size_t cached_size_in_bytes;
 };
 
 class RasterizerCacheOpenGL final : public RasterizerCache<Surface> {
@@ -839,9 +879,6 @@ public:
     /// Get the color surface based on the framebuffer configuration and the specified render target
     Surface GetColorBufferSurface(std::size_t index, bool preserve_contents);
 
-    /// Flushes the surface to Switch memory
-    void FlushSurface(const Surface& surface);
-
     /// Tries to find a framebuffer using on the provided CPU address
     Surface TryFindFramebufferSurface(VAddr addr) const;
 
@@ -865,6 +902,9 @@ private:
     /// Tries to get a reserved surface for the specified parameters
     Surface TryGetReservedSurface(const SurfaceParams& params);
 
+    /// Performs a slow but accurate surface copy, flushing to RAM and reinterpreting the data
+    void AccurateCopySurface(const Surface& src_surface, const Surface& dst_surface);
+
     /// The surface reserve is a "backup" cache, this is where we put unique surfaces that have
     /// previously been used. This is to prevent surfaces from being constantly created and
     /// destroyed when used with different surface parameters.

src/video_core/renderer_opengl/gl_shader_cache.h

@@ -19,20 +19,21 @@ class CachedShader;
 using Shader = std::shared_ptr<CachedShader>;
 using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 
-class CachedShader final {
+class CachedShader final : public RasterizerCacheObject {
 public:
     CachedShader(VAddr addr, Maxwell::ShaderProgram program_type);
 
-    /// Gets the address of the shader in guest memory, required for cache management
-    VAddr GetAddr() const {
+    VAddr GetAddr() const override {
         return addr;
     }
 
-    /// Gets the size of the shader in guest memory, required for cache management
-    std::size_t GetSizeInBytes() const {
+    std::size_t GetSizeInBytes() const override {
         return GLShader::MAX_PROGRAM_CODE_LENGTH * sizeof(u64);
     }
 
+    // We do not have to flush this cache as things in it are never modified by us.
+    void Flush() override {}
+
     /// Gets the shader entries for the shader
     const GLShader::ShaderEntries& GetShaderEntries() const {
         return entries;

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -1142,6 +1142,7 @@ private:
         case Tegra::Shader::TextureType::Texture2D: {
             return 2;
         }
+        case Tegra::Shader::TextureType::Texture3D:
         case Tegra::Shader::TextureType::TextureCube: {
             return 3;
         }
@@ -1436,7 +1437,6 @@ private:
 
             break;
         }
-
         case OpCode::Type::Shift: {
             std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, true);
             std::string op_b;
@@ -1478,7 +1478,6 @@ private:
             }
             break;
         }
-
         case OpCode::Type::ArithmeticIntegerImmediate: {
             std::string op_a = regs.GetRegisterAsInteger(instr.gpr8);
             std::string op_b = std::to_string(instr.alu.imm20_32.Value());
@@ -2038,9 +2037,9 @@ private:
                 break;
             }
             case OpCode::Id::TEX: {
-                ASSERT_MSG(instr.tex.array == 0, "TEX arrays unimplemented");
                 Tegra::Shader::TextureType texture_type{instr.tex.texture_type};
                 std::string coord;
+                const bool is_array = instr.tex.array != 0;
 
                 ASSERT_MSG(!instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
                            "NODEP is not implemented");
@@ -2055,21 +2054,59 @@ private:
 
                 switch (num_coordinates) {
                 case 1: {
-                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                    coord = "float coords = " + x + ';';
+                    if (is_array) {
+                        const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                        coord = "vec2 coords = vec2(" + x + ", " + index + ");";
+                    } else {
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                        coord = "float coords = " + x + ';';
+                    }
                     break;
                 }
                 case 2: {
-                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                    const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
-                    coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+                    if (is_array) {
+                        const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                        const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 2);
+                        coord = "vec3 coords = vec3(" + x + ", " + y + ", " + index + ");";
+                    } else {
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                        const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                        coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+                    }
                     break;
                 }
                 case 3: {
-                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                    const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
-                    const std::string z = regs.GetRegisterAsFloat(instr.gpr20);
-                    coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
+                    if (depth_compare) {
+                        if (is_array) {
+                            const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+                            const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                            const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
+                            const std::string z = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
+                            coord = "vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " + index +
+                                    ");";
+                        } else {
+                            const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                            const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                            const std::string z = regs.GetRegisterAsFloat(instr.gpr20);
+                            coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
+                        }
+                    } else {
+                        if (is_array) {
+                            const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+                            const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                            const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 2);
+                            const std::string z = regs.GetRegisterAsFloat(instr.gpr8.Value() + 3);
+                            coord = "vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " + index +
+                                    ");";
+                        } else {
+                            const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                            const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                            const std::string z = regs.GetRegisterAsFloat(instr.gpr8.Value() + 2);
+                            coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
+                        }
+                    }
                     break;
                 }
                 default:
@@ -2088,7 +2125,7 @@ private:
                 std::string op_c;
 
                 const std::string sampler =
-                    GetSampler(instr.sampler, texture_type, false, depth_compare);
+                    GetSampler(instr.sampler, texture_type, is_array, depth_compare);
                 // Add an extra scope and declare the texture coords inside to prevent
                 // overwriting them in case they are used as outputs of the texs instruction.
 
@@ -2108,10 +2145,13 @@ private:
                 }
                 case Tegra::Shader::TextureProcessMode::LB:
                 case Tegra::Shader::TextureProcessMode::LBA: {
-                    if (num_coordinates <= 2) {
-                        op_c = regs.GetRegisterAsFloat(instr.gpr20);
+                    if (depth_compare) {
+                        if (is_array)
+                            op_c = regs.GetRegisterAsFloat(instr.gpr20.Value() + 2);
+                        else
+                            op_c = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
                     } else {
-                        op_c = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
+                        op_c = regs.GetRegisterAsFloat(instr.gpr20);
                     }
                     // TODO: Figure if A suffix changes the equation at all.
                     texture = "texture(" + sampler + ", coords, " + op_c + ')';
@@ -2254,6 +2294,8 @@ private:
                 ASSERT_MSG(!instr.tlds.UsesMiscMode(Tegra::Shader::TextureMiscMode::MZ),
                            "MZ is not implemented");
 
+                u32 op_c_offset = 0;
+
                 switch (texture_type) {
                 case Tegra::Shader::TextureType::Texture1D: {
                     const std::string x = regs.GetRegisterAsInteger(instr.gpr8);
@@ -2268,6 +2310,7 @@ private:
                         const std::string x = regs.GetRegisterAsInteger(instr.gpr8);
                         const std::string y = regs.GetRegisterAsInteger(instr.gpr20);
                         coord = "ivec2 coords = ivec2(" + x + ", " + y + ");";
+                        op_c_offset = 1;
                     }
                     break;
                 }
@@ -2279,13 +2322,14 @@ private:
                 const std::string sampler =
                     GetSampler(instr.sampler, texture_type, is_array, false);
                 std::string texture = "texelFetch(" + sampler + ", coords, 0)";
-                const std::string op_c = regs.GetRegisterAsInteger(instr.gpr20.Value() + 1);
                 switch (instr.tlds.GetTextureProcessMode()) {
                 case Tegra::Shader::TextureProcessMode::LZ: {
                     texture = "texelFetch(" + sampler + ", coords, 0)";
                     break;
                 }
                 case Tegra::Shader::TextureProcessMode::LL: {
+                    const std::string op_c =
+                        regs.GetRegisterAsInteger(instr.gpr20.Value() + op_c_offset);
                     texture = "texelFetch(" + sampler + ", coords, " + op_c + ')';
                     break;
                 }
@@ -2626,14 +2670,14 @@ private:
                 const std::string pred =
                     GetPredicateCondition(instr.csetp.pred39, instr.csetp.neg_pred39 != 0);
                 const std::string combiner = GetPredicateCombiner(instr.csetp.op);
-                const std::string controlCode = regs.GetControlCode(instr.csetp.cc);
+                const std::string control_code = regs.GetControlCode(instr.csetp.cc);
                 if (instr.csetp.pred3 != static_cast<u64>(Pred::UnusedIndex)) {
                     SetPredicate(instr.csetp.pred3,
-                                 '(' + controlCode + ") " + combiner + " (" + pred + ')');
+                                 '(' + control_code + ") " + combiner + " (" + pred + ')');
                 }
                 if (instr.csetp.pred0 != static_cast<u64>(Pred::UnusedIndex)) {
                     SetPredicate(instr.csetp.pred0,
-                                 "!(" + controlCode + ") " + combiner + " (" + pred + ')');
+                                 "!(" + control_code + ") " + combiner + " (" + pred + ')');
                 }
                 break;
             }

src/video_core/textures/decoders.cpp

@@ -40,72 +40,146 @@ struct alignas(64) SwizzleTable {
 constexpr auto legacy_swizzle_table = SwizzleTable<8, 64, 1>();
 constexpr auto fast_swizzle_table = SwizzleTable<8, 4, 16>();
 
-static void LegacySwizzleData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel,
-                              u8* swizzled_data, u8* unswizzled_data, bool unswizzle,
-                              u32 block_height) {
+/**
+ * This function manages ALL the GOBs(Group of Bytes) Inside a single block.
+ * Instead of going gob by gob, we map the coordinates inside a block and manage from
+ * those. Block_Width is assumed to be 1.
+ */
+void PreciseProcessBlock(u8* swizzled_data, u8* unswizzled_data, const bool unswizzle,
+                         const u32 x_start, const u32 y_start, const u32 z_start, const u32 x_end,
+                         const u32 y_end, const u32 z_end, const u32 tile_offset,
+                         const u32 xy_block_size, const u32 layer_z, const u32 stride_x,
+                         const u32 bytes_per_pixel, const u32 out_bytes_per_pixel) {
     std::array<u8*, 2> data_ptrs;
-    const std::size_t stride = width * bytes_per_pixel;
-    const std::size_t gobs_in_x = 64;
-    const std::size_t gobs_in_y = 8;
-    const std::size_t gobs_size = gobs_in_x * gobs_in_y;
-    const std::size_t image_width_in_gobs{(stride + gobs_in_x - 1) / gobs_in_x};
-    for (std::size_t y = 0; y < height; ++y) {
-        const std::size_t gob_y_address =
-            (y / (gobs_in_y * block_height)) * gobs_size * block_height * image_width_in_gobs +
-            (y % (gobs_in_y * block_height) / gobs_in_y) * gobs_size;
-        const auto& table = legacy_swizzle_table[y % gobs_in_y];
-        for (std::size_t x = 0; x < width; ++x) {
-            const std::size_t gob_address =
-                gob_y_address + (x * bytes_per_pixel / gobs_in_x) * gobs_size * block_height;
-            const std::size_t x2 = x * bytes_per_pixel;
-            const std::size_t swizzle_offset = gob_address + table[x2 % gobs_in_x];
-            const std::size_t pixel_index = (x + y * width) * out_bytes_per_pixel;
+    u32 z_address = tile_offset;
+    const u32 gob_size_x = 64;
+    const u32 gob_size_y = 8;
+    const u32 gob_size_z = 1;
+    const u32 gob_size = gob_size_x * gob_size_y * gob_size_z;
+    for (u32 z = z_start; z < z_end; z++) {
+        u32 y_address = z_address;
+        u32 pixel_base = layer_z * z + y_start * stride_x;
+        for (u32 y = y_start; y < y_end; y++) {
+            const auto& table = legacy_swizzle_table[y % gob_size_y];
+            for (u32 x = x_start; x < x_end; x++) {
+                const u32 swizzle_offset{y_address + table[x * bytes_per_pixel % gob_size_x]};
+                const u32 pixel_index{x * out_bytes_per_pixel + pixel_base};
+                data_ptrs[unswizzle] = swizzled_data + swizzle_offset;
+                data_ptrs[!unswizzle] = unswizzled_data + pixel_index;
+                std::memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel);
+            }
+            pixel_base += stride_x;
+            if ((y + 1) % gob_size_y == 0)
+                y_address += gob_size;
+        }
+        z_address += xy_block_size;
+    }
+}
 
-            data_ptrs[unswizzle] = swizzled_data + swizzle_offset;
-            data_ptrs[!unswizzle] = unswizzled_data + pixel_index;
+/**
+ * This function manages ALL the GOBs(Group of Bytes) Inside a single block.
+ * Instead of going gob by gob, we map the coordinates inside a block and manage from
+ * those. Block_Width is assumed to be 1.
+ */
+void FastProcessBlock(u8* swizzled_data, u8* unswizzled_data, const bool unswizzle,
+                      const u32 x_start, const u32 y_start, const u32 z_start, const u32 x_end,
+                      const u32 y_end, const u32 z_end, const u32 tile_offset,
+                      const u32 xy_block_size, const u32 layer_z, const u32 stride_x,
+                      const u32 bytes_per_pixel, const u32 out_bytes_per_pixel) {
+    std::array<u8*, 2> data_ptrs;
+    u32 z_address = tile_offset;
+    const u32 x_startb = x_start * bytes_per_pixel;
+    const u32 x_endb = x_end * bytes_per_pixel;
+    const u32 copy_size = 16;
+    const u32 gob_size_x = 64;
+    const u32 gob_size_y = 8;
+    const u32 gob_size_z = 1;
+    const u32 gob_size = gob_size_x * gob_size_y * gob_size_z;
+    for (u32 z = z_start; z < z_end; z++) {
+        u32 y_address = z_address;
+        u32 pixel_base = layer_z * z + y_start * stride_x;
+        for (u32 y = y_start; y < y_end; y++) {
+            const auto& table = fast_swizzle_table[y % gob_size_y];
+            for (u32 xb = x_startb; xb < x_endb; xb += copy_size) {
+                const u32 swizzle_offset{y_address + table[(xb / copy_size) % 4]};
+                const u32 out_x = xb * out_bytes_per_pixel / bytes_per_pixel;
+                const u32 pixel_index{out_x + pixel_base};
+                data_ptrs[unswizzle] = swizzled_data + swizzle_offset;
+                data_ptrs[!unswizzle] = unswizzled_data + pixel_index;
+                std::memcpy(data_ptrs[0], data_ptrs[1], copy_size);
+            }
+            pixel_base += stride_x;
+            if ((y + 1) % gob_size_y == 0)
+                y_address += gob_size;
+        }
+        z_address += xy_block_size;
+    }
+}
 
-            std::memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel);
+/**
+ * This function unswizzles or swizzles a texture by mapping Linear to BlockLinear Textue.
+ * The body of this function takes care of splitting the swizzled texture into blocks,
+ * and managing the extents of it. Once all the parameters of a single block are obtained,
+ * the function calls 'ProcessBlock' to process that particular Block.
+ *
+ * Documentation for the memory layout and decoding can be found at:
+ *  https://envytools.readthedocs.io/en/latest/hw/memory/g80-surface.html#blocklinear-surfaces
+ */
+template <bool fast>
+void SwizzledData(u8* swizzled_data, u8* unswizzled_data, const bool unswizzle, const u32 width,
+                  const u32 height, const u32 depth, const u32 bytes_per_pixel,
+                  const u32 out_bytes_per_pixel, const u32 block_height, const u32 block_depth) {
+    auto div_ceil = [](const u32 x, const u32 y) { return ((x + y - 1) / y); };
+    const u32 stride_x = width * out_bytes_per_pixel;
+    const u32 layer_z = height * stride_x;
+    const u32 gob_x_bytes = 64;
+    const u32 gob_elements_x = gob_x_bytes / bytes_per_pixel;
+    const u32 gob_elements_y = 8;
+    const u32 gob_elements_z = 1;
+    const u32 block_x_elements = gob_elements_x;
+    const u32 block_y_elements = gob_elements_y * block_height;
+    const u32 block_z_elements = gob_elements_z * block_depth;
+    const u32 blocks_on_x = div_ceil(width, block_x_elements);
+    const u32 blocks_on_y = div_ceil(height, block_y_elements);
+    const u32 blocks_on_z = div_ceil(depth, block_z_elements);
+    const u32 blocks = blocks_on_x * blocks_on_y * blocks_on_z;
+    const u32 gob_size = gob_x_bytes * gob_elements_y * gob_elements_z;
+    const u32 xy_block_size = gob_size * block_height;
+    const u32 block_size = xy_block_size * block_depth;
+    u32 tile_offset = 0;
+    for (u32 zb = 0; zb < blocks_on_z; zb++) {
+        const u32 z_start = zb * block_z_elements;
+        const u32 z_end = std::min(depth, z_start + block_z_elements);
+        for (u32 yb = 0; yb < blocks_on_y; yb++) {
+            const u32 y_start = yb * block_y_elements;
+            const u32 y_end = std::min(height, y_start + block_y_elements);
+            for (u32 xb = 0; xb < blocks_on_x; xb++) {
+                const u32 x_start = xb * block_x_elements;
+                const u32 x_end = std::min(width, x_start + block_x_elements);
+                if (fast) {
+                    FastProcessBlock(swizzled_data, unswizzled_data, unswizzle, x_start, y_start,
+                                     z_start, x_end, y_end, z_end, tile_offset, xy_block_size,
+                                     layer_z, stride_x, bytes_per_pixel, out_bytes_per_pixel);
+                } else {
+                    PreciseProcessBlock(swizzled_data, unswizzled_data, unswizzle, x_start, y_start,
+                                        z_start, x_end, y_end, z_end, tile_offset, xy_block_size,
+                                        layer_z, stride_x, bytes_per_pixel, out_bytes_per_pixel);
+                }
+                tile_offset += block_size;
+            }
         }
     }
 }
 
-static void FastSwizzleData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel,
-                            u8* swizzled_data, u8* unswizzled_data, bool unswizzle,
-                            u32 block_height) {
-    std::array<u8*, 2> data_ptrs;
-    const std::size_t stride{width * bytes_per_pixel};
-    const std::size_t gobs_in_x = 64;
-    const std::size_t gobs_in_y = 8;
-    const std::size_t gobs_size = gobs_in_x * gobs_in_y;
-    const std::size_t image_width_in_gobs{(stride + gobs_in_x - 1) / gobs_in_x};
-    const std::size_t copy_size{16};
-    for (std::size_t y = 0; y < height; ++y) {
-        const std::size_t initial_gob =
-            (y / (gobs_in_y * block_height)) * gobs_size * block_height * image_width_in_gobs +
-            (y % (gobs_in_y * block_height) / gobs_in_y) * gobs_size;
-        const std::size_t pixel_base{y * width * out_bytes_per_pixel};
-        const auto& table = fast_swizzle_table[y % gobs_in_y];
-        for (std::size_t xb = 0; xb < stride; xb += copy_size) {
-            const std::size_t gob_address{initial_gob +
-                                          (xb / gobs_in_x) * gobs_size * block_height};
-            const std::size_t swizzle_offset{gob_address + table[(xb / 16) % 4]};
-            const std::size_t out_x = xb * out_bytes_per_pixel / bytes_per_pixel;
-            const std::size_t pixel_index{out_x + pixel_base};
-            data_ptrs[unswizzle] = swizzled_data + swizzle_offset;
-            data_ptrs[!unswizzle] = unswizzled_data + pixel_index;
-            std::memcpy(data_ptrs[0], data_ptrs[1], copy_size);
-        }
-    }
-}
-
-void CopySwizzledData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel,
-                      u8* swizzled_data, u8* unswizzled_data, bool unswizzle, u32 block_height) {
+void CopySwizzledData(u32 width, u32 height, u32 depth, u32 bytes_per_pixel,
+                      u32 out_bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data,
+                      bool unswizzle, u32 block_height, u32 block_depth) {
     if (bytes_per_pixel % 3 != 0 && (width * bytes_per_pixel) % 16 == 0) {
-        FastSwizzleData(width, height, bytes_per_pixel, out_bytes_per_pixel, swizzled_data,
-                        unswizzled_data, unswizzle, block_height);
+        SwizzledData<true>(swizzled_data, unswizzled_data, unswizzle, width, height, depth,
+                           bytes_per_pixel, out_bytes_per_pixel, block_height, block_depth);
     } else {
-        LegacySwizzleData(width, height, bytes_per_pixel, out_bytes_per_pixel, swizzled_data,
-                          unswizzled_data, unswizzle, block_height);
+        SwizzledData<false>(swizzled_data, unswizzled_data, unswizzle, width, height, depth,
+                            bytes_per_pixel, out_bytes_per_pixel, block_height, block_depth);
     }
 }
 
@@ -126,7 +200,9 @@ u32 BytesPerPixel(TextureFormat format) {
     case TextureFormat::R32_G32_B32:
         return 12;
     case TextureFormat::ASTC_2D_4X4:
+    case TextureFormat::ASTC_2D_5X4:
     case TextureFormat::ASTC_2D_8X8:
+    case TextureFormat::ASTC_2D_8X5:
     case TextureFormat::A8R8G8B8:
     case TextureFormat::A2B10G10R10:
     case TextureFormat::BF10GF11RF11:
@@ -153,10 +229,11 @@ u32 BytesPerPixel(TextureFormat format) {
 }
 
 std::vector<u8> UnswizzleTexture(VAddr address, u32 tile_size, u32 bytes_per_pixel, u32 width,
-                                 u32 height, u32 block_height) {
-    std::vector<u8> unswizzled_data(width * height * bytes_per_pixel);
-    CopySwizzledData(width / tile_size, height / tile_size, bytes_per_pixel, bytes_per_pixel,
-                     Memory::GetPointer(address), unswizzled_data.data(), true, block_height);
+                                 u32 height, u32 depth, u32 block_height, u32 block_depth) {
+    std::vector<u8> unswizzled_data(width * height * depth * bytes_per_pixel);
+    CopySwizzledData(width / tile_size, height / tile_size, depth, bytes_per_pixel, bytes_per_pixel,
+                     Memory::GetPointer(address), unswizzled_data.data(), true, block_height,
+                     block_depth);
     return unswizzled_data;
 }
 

src/video_core/textures/decoders.h

@@ -14,17 +14,14 @@ namespace Tegra::Texture {
  * Unswizzles a swizzled texture without changing its format.
  */
 std::vector<u8> UnswizzleTexture(VAddr address, u32 tile_size, u32 bytes_per_pixel, u32 width,
-                                 u32 height, u32 block_height = TICEntry::DefaultBlockHeight);
-
-/**
- * Unswizzles a swizzled depth texture without changing its format.
- */
-std::vector<u8> UnswizzleDepthTexture(VAddr address, DepthFormat format, u32 width, u32 height,
-                                      u32 block_height = TICEntry::DefaultBlockHeight);
+                                 u32 height, u32 depth,
+                                 u32 block_height = TICEntry::DefaultBlockHeight,
+                                 u32 block_depth = TICEntry::DefaultBlockHeight);
 
 /// Copies texture data from a buffer and performs swizzling/unswizzling as necessary.
-void CopySwizzledData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel,
-                      u8* swizzled_data, u8* unswizzled_data, bool unswizzle, u32 block_height);
+void CopySwizzledData(u32 width, u32 height, u32 depth, u32 bytes_per_pixel,
+                      u32 out_bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data,
+                      bool unswizzle, u32 block_height, u32 block_depth);
 
 /**
  * Decodes an unswizzled texture into a A8R8G8B8 texture.

src/video_core/textures/texture.h

@@ -141,6 +141,7 @@ static_assert(sizeof(TextureHandle) == 4, "TextureHandle has wrong size");
 
 struct TICEntry {
     static constexpr u32 DefaultBlockHeight = 16;
+    static constexpr u32 DefaultBlockDepth = 1;
 
     union {
         u32 raw;

src/web_service/telemetry_json.cpp

@@ -2,96 +2,114 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include <thread>
-#include "common/assert.h"
+#include <json.hpp>
 #include "common/detached_tasks.h"
+#include "common/web_result.h"
 #include "web_service/telemetry_json.h"
 #include "web_service/web_backend.h"
 
 namespace WebService {
 
-TelemetryJson::TelemetryJson(const std::string& host, const std::string& username,
-                             const std::string& token)
-    : host(std::move(host)), username(std::move(username)), token(std::move(token)) {}
+struct TelemetryJson::Impl {
+    Impl(std::string host, std::string username, std::string token)
+        : host{std::move(host)}, username{std::move(username)}, token{std::move(token)} {}
+
+    nlohmann::json& TopSection() {
+        return sections[static_cast<u8>(Telemetry::FieldType::None)];
+    }
+
+    const nlohmann::json& TopSection() const {
+        return sections[static_cast<u8>(Telemetry::FieldType::None)];
+    }
+
+    template <class T>
+    void Serialize(Telemetry::FieldType type, const std::string& name, T value) {
+        sections[static_cast<u8>(type)][name] = value;
+    }
+
+    void SerializeSection(Telemetry::FieldType type, const std::string& name) {
+        TopSection()[name] = sections[static_cast<unsigned>(type)];
+    }
+
+    nlohmann::json output;
+    std::array<nlohmann::json, 7> sections;
+    std::string host;
+    std::string username;
+    std::string token;
+};
+
+TelemetryJson::TelemetryJson(std::string host, std::string username, std::string token)
+    : impl{std::make_unique<Impl>(std::move(host), std::move(username), std::move(token))} {}
 TelemetryJson::~TelemetryJson() = default;
 
-template <class T>
-void TelemetryJson::Serialize(Telemetry::FieldType type, const std::string& name, T value) {
-    sections[static_cast<u8>(type)][name] = value;
-}
-
-void TelemetryJson::SerializeSection(Telemetry::FieldType type, const std::string& name) {
-    TopSection()[name] = sections[static_cast<unsigned>(type)];
-}
-
 void TelemetryJson::Visit(const Telemetry::Field<bool>& field) {
-    Serialize(field.GetType(), field.GetName(), field.GetValue());
+    impl->Serialize(field.GetType(), field.GetName(), field.GetValue());
 }
 
 void TelemetryJson::Visit(const Telemetry::Field<double>& field) {
-    Serialize(field.GetType(), field.GetName(), field.GetValue());
+    impl->Serialize(field.GetType(), field.GetName(), field.GetValue());
 }
 
 void TelemetryJson::Visit(const Telemetry::Field<float>& field) {
-    Serialize(field.GetType(), field.GetName(), field.GetValue());
+    impl->Serialize(field.GetType(), field.GetName(), field.GetValue());
 }
 
 void TelemetryJson::Visit(const Telemetry::Field<u8>& field) {
-    Serialize(field.GetType(), field.GetName(), field.GetValue());
+    impl->Serialize(field.GetType(), field.GetName(), field.GetValue());
 }
 
 void TelemetryJson::Visit(const Telemetry::Field<u16>& field) {
-    Serialize(field.GetType(), field.GetName(), field.GetValue());
+    impl->Serialize(field.GetType(), field.GetName(), field.GetValue());
 }
 
 void TelemetryJson::Visit(const Telemetry::Field<u32>& field) {
-    Serialize(field.GetType(), field.GetName(), field.GetValue());
+    impl->Serialize(field.GetType(), field.GetName(), field.GetValue());
 }
 
 void TelemetryJson::Visit(const Telemetry::Field<u64>& field) {
-    Serialize(field.GetType(), field.GetName(), field.GetValue());
+    impl->Serialize(field.GetType(), field.GetName(), field.GetValue());
 }
 
 void TelemetryJson::Visit(const Telemetry::Field<s8>& field) {
-    Serialize(field.GetType(), field.GetName(), field.GetValue());
+    impl->Serialize(field.GetType(), field.GetName(), field.GetValue());
 }
 
 void TelemetryJson::Visit(const Telemetry::Field<s16>& field) {
-    Serialize(field.GetType(), field.GetName(), field.GetValue());
+    impl->Serialize(field.GetType(), field.GetName(), field.GetValue());
 }
 
 void TelemetryJson::Visit(const Telemetry::Field<s32>& field) {
-    Serialize(field.GetType(), field.GetName(), field.GetValue());
+    impl->Serialize(field.GetType(), field.GetName(), field.GetValue());
 }
 
 void TelemetryJson::Visit(const Telemetry::Field<s64>& field) {
-    Serialize(field.GetType(), field.GetName(), field.GetValue());
+    impl->Serialize(field.GetType(), field.GetName(), field.GetValue());
 }
 
 void TelemetryJson::Visit(const Telemetry::Field<std::string>& field) {
-    Serialize(field.GetType(), field.GetName(), field.GetValue());
+    impl->Serialize(field.GetType(), field.GetName(), field.GetValue());
 }
 
 void TelemetryJson::Visit(const Telemetry::Field<const char*>& field) {
-    Serialize(field.GetType(), field.GetName(), std::string(field.GetValue()));
+    impl->Serialize(field.GetType(), field.GetName(), std::string(field.GetValue()));
 }
 
 void TelemetryJson::Visit(const Telemetry::Field<std::chrono::microseconds>& field) {
-    Serialize(field.GetType(), field.GetName(), field.GetValue().count());
+    impl->Serialize(field.GetType(), field.GetName(), field.GetValue().count());
 }
 
 void TelemetryJson::Complete() {
-    SerializeSection(Telemetry::FieldType::App, "App");
-    SerializeSection(Telemetry::FieldType::Session, "Session");
-    SerializeSection(Telemetry::FieldType::Performance, "Performance");
-    SerializeSection(Telemetry::FieldType::UserFeedback, "UserFeedback");
-    SerializeSection(Telemetry::FieldType::UserConfig, "UserConfig");
-    SerializeSection(Telemetry::FieldType::UserSystem, "UserSystem");
+    impl->SerializeSection(Telemetry::FieldType::App, "App");
+    impl->SerializeSection(Telemetry::FieldType::Session, "Session");
+    impl->SerializeSection(Telemetry::FieldType::Performance, "Performance");
+    impl->SerializeSection(Telemetry::FieldType::UserFeedback, "UserFeedback");
+    impl->SerializeSection(Telemetry::FieldType::UserConfig, "UserConfig");
+    impl->SerializeSection(Telemetry::FieldType::UserSystem, "UserSystem");
 
-    auto content = TopSection().dump();
+    auto content = impl->TopSection().dump();
     // Send the telemetry async but don't handle the errors since they were written to the log
     Common::DetachedTasks::AddTask(
-        [host{this->host}, username{this->username}, token{this->token}, content]() {
+        [host{impl->host}, username{impl->username}, token{impl->token}, content]() {
             Client{host, username, token}.PostJson("/telemetry", content, true);
         });
 }

src/web_service/telemetry_json.h

@@ -4,11 +4,9 @@
 
 #pragma once
 
-#include <array>
+#include <chrono>
 #include <string>
-#include <json.hpp>
 #include "common/telemetry.h"
-#include "common/web_result.h"
 
 namespace WebService {
 
@@ -18,8 +16,8 @@ namespace WebService {
  */
 class TelemetryJson : public Telemetry::VisitorInterface {
 public:
-    TelemetryJson(const std::string& host, const std::string& username, const std::string& token);
-    ~TelemetryJson();
+    TelemetryJson(std::string host, std::string username, std::string token);
+    ~TelemetryJson() override;
 
     void Visit(const Telemetry::Field<bool>& field) override;
     void Visit(const Telemetry::Field<double>& field) override;
@@ -39,20 +37,8 @@ public:
     void Complete() override;
 
 private:
-    nlohmann::json& TopSection() {
-        return sections[static_cast<u8>(Telemetry::FieldType::None)];
-    }
-
-    template <class T>
-    void Serialize(Telemetry::FieldType type, const std::string& name, T value);
-
-    void SerializeSection(Telemetry::FieldType type, const std::string& name);
-
-    nlohmann::json output;
-    std::array<nlohmann::json, 7> sections;
-    std::string host;
-    std::string username;
-    std::string token;
+    struct Impl;
+    std::unique_ptr<Impl> impl;
 };
 
 } // namespace WebService

src/web_service/verify_login.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <json.hpp>
+#include "common/web_result.h"
 #include "web_service/verify_login.h"
 #include "web_service/web_backend.h"