svc: Clarify enum values for AddressSpaceBaseAddr and AddressSpaceSize in svcGetInfo()

So, one thing that's puzzled me is why the kernel seemed to *not* use
the direct code address ranges in some cases for some service functions.
For example, in svcMapMemory, the full address space width is compared
against for validity, but for svcMapSharedMemory, it compares against
0xFFE00000, 0xFF8000000, and 0x7FF8000000 as upper bounds, and uses
either 0x200000 or 0x8000000 as the lower-bounds as the beginning of the
compared range. Coincidentally, these exact same values are also used in
svcGetInfo, and also when initializing the user address space, so this
is actually retrieving the ASLR extents, not the extents of the address
space in general.

src/audio_core/audio_renderer.cpp

@@ -30,7 +30,7 @@ public:
         return info;
     }
 
-    VoiceInfo& Info() {
+    VoiceInfo& GetInfo() {
         return info;
     }
 
@@ -51,9 +51,30 @@ private:
     VoiceInfo info{};
 };
 
+class AudioRenderer::EffectState {
+public:
+    const EffectOutStatus& GetOutStatus() const {
+        return out_status;
+    }
+
+    const EffectInStatus& GetInfo() const {
+        return info;
+    }
+
+    EffectInStatus& GetInfo() {
+        return info;
+    }
+
+    void UpdateState();
+
+private:
+    EffectOutStatus out_status{};
+    EffectInStatus info{};
+};
 AudioRenderer::AudioRenderer(AudioRendererParameter params,
                              Kernel::SharedPtr<Kernel::Event> buffer_event)
-    : worker_params{params}, buffer_event{buffer_event}, voices(params.voice_count) {
+    : worker_params{params}, buffer_event{buffer_event}, voices(params.voice_count),
+      effects(params.effect_count) {
 
     audio_out = std::make_unique<AudioCore::AudioOut>();
     stream = audio_out->OpenStream(STREAM_SAMPLE_RATE, STREAM_NUM_CHANNELS, "AudioRenderer",
@@ -96,11 +117,29 @@ std::vector<u8> AudioRenderer::UpdateAudioRenderer(const std::vector<u8>& input_
                 memory_pool_count * sizeof(MemoryPoolInfo));
 
     // Copy VoiceInfo structs
-    std::size_t offset{sizeof(UpdateDataHeader) + config.behavior_size + config.memory_pools_size +
-                       config.voice_resource_size};
+    std::size_t voice_offset{sizeof(UpdateDataHeader) + config.behavior_size +
+                             config.memory_pools_size + config.voice_resource_size};
     for (auto& voice : voices) {
-        std::memcpy(&voice.Info(), input_params.data() + offset, sizeof(VoiceInfo));
-        offset += sizeof(VoiceInfo);
+        std::memcpy(&voice.GetInfo(), input_params.data() + voice_offset, sizeof(VoiceInfo));
+        voice_offset += sizeof(VoiceInfo);
+    }
+
+    std::size_t effect_offset{sizeof(UpdateDataHeader) + config.behavior_size +
+                              config.memory_pools_size + config.voice_resource_size +
+                              config.voices_size};
+    for (auto& effect : effects) {
+        std::memcpy(&effect.GetInfo(), input_params.data() + effect_offset, sizeof(EffectInStatus));
+        effect_offset += sizeof(EffectInStatus);
+    }
+
+    // Update memory pool state
+    std::vector<MemoryPoolEntry> memory_pool(memory_pool_count);
+    for (std::size_t index = 0; index < memory_pool.size(); ++index) {
+        if (mem_pool_info[index].pool_state == MemoryPoolStates::RequestAttach) {
+            memory_pool[index].state = MemoryPoolStates::Attached;
+        } else if (mem_pool_info[index].pool_state == MemoryPoolStates::RequestDetach) {
+            memory_pool[index].state = MemoryPoolStates::Detached;
+        }
     }
 
     // Update voices
@@ -114,14 +153,8 @@ std::vector<u8> AudioRenderer::UpdateAudioRenderer(const std::vector<u8>& input_
         }
     }
 
-    // Update memory pool state
-    std::vector<MemoryPoolEntry> memory_pool(memory_pool_count);
-    for (std::size_t index = 0; index < memory_pool.size(); ++index) {
-        if (mem_pool_info[index].pool_state == MemoryPoolStates::RequestAttach) {
-            memory_pool[index].state = MemoryPoolStates::Attached;
-        } else if (mem_pool_info[index].pool_state == MemoryPoolStates::RequestDetach) {
-            memory_pool[index].state = MemoryPoolStates::Detached;
-        }
+    for (auto& effect : effects) {
+        effect.UpdateState();
     }
 
     // Release previous buffers and queue next ones for playback
@@ -144,6 +177,14 @@ std::vector<u8> AudioRenderer::UpdateAudioRenderer(const std::vector<u8>& input_
         voice_out_status_offset += sizeof(VoiceOutStatus);
     }
 
+    std::size_t effect_out_status_offset{
+        sizeof(UpdateDataHeader) + response_data.memory_pools_size + response_data.voices_size +
+        response_data.voice_resource_size};
+    for (const auto& effect : effects) {
+        std::memcpy(output_params.data() + effect_out_status_offset, &effect.GetOutStatus(),
+                    sizeof(EffectOutStatus));
+        effect_out_status_offset += sizeof(EffectOutStatus);
+    }
     return output_params;
 }
 
@@ -244,11 +285,29 @@ void AudioRenderer::VoiceState::RefreshBuffer() {
         break;
     }
 
-    samples = Interpolate(interp_state, std::move(samples), Info().sample_rate, STREAM_SAMPLE_RATE);
+    samples =
+        Interpolate(interp_state, std::move(samples), GetInfo().sample_rate, STREAM_SAMPLE_RATE);
 
     is_refresh_pending = false;
 }
 
+void AudioRenderer::EffectState::UpdateState() {
+    if (info.is_new) {
+        out_status.state = EffectStatus::New;
+    } else {
+        if (info.type == Effect::Aux) {
+            ASSERT_MSG(Memory::Read32(info.aux_info.return_buffer_info) == 0,
+                       "Aux buffers tried to update");
+            ASSERT_MSG(Memory::Read32(info.aux_info.send_buffer_info) == 0,
+                       "Aux buffers tried to update");
+            ASSERT_MSG(Memory::Read32(info.aux_info.return_buffer_base) == 0,
+                       "Aux buffers tried to update");
+            ASSERT_MSG(Memory::Read32(info.aux_info.send_buffer_base) == 0,
+                       "Aux buffers tried to update");
+        }
+    }
+}
+
 static constexpr s16 ClampToS16(s32 value) {
     return static_cast<s16>(std::clamp(value, -32768, 32767));
 }

src/audio_core/audio_renderer.h

@@ -28,6 +28,16 @@ enum class PlayState : u8 {
     Paused = 2,
 };
 
+enum class Effect : u8 {
+    None = 0,
+    Aux = 2,
+};
+
+enum class EffectStatus : u8 {
+    None = 0,
+    New = 1,
+};
+
 struct AudioRendererParameter {
     u32_le sample_rate;
     u32_le sample_count;
@@ -128,6 +138,43 @@ struct VoiceOutStatus {
 };
 static_assert(sizeof(VoiceOutStatus) == 0x10, "VoiceOutStatus has wrong size");
 
+struct AuxInfo {
+    std::array<u8, 24> input_mix_buffers;
+    std::array<u8, 24> output_mix_buffers;
+    u32_le mix_buffer_count;
+    u32_le sample_rate; // Stored in the aux buffer currently
+    u32_le sampe_count;
+    u64_le send_buffer_info;
+    u64_le send_buffer_base;
+
+    u64_le return_buffer_info;
+    u64_le return_buffer_base;
+};
+static_assert(sizeof(AuxInfo) == 0x60, "AuxInfo is an invalid size");
+
+struct EffectInStatus {
+    Effect type;
+    u8 is_new;
+    u8 is_enabled;
+    INSERT_PADDING_BYTES(1);
+    u32_le mix_id;
+    u64_le buffer_base;
+    u64_le buffer_sz;
+    s32_le priority;
+    INSERT_PADDING_BYTES(4);
+    union {
+        std::array<u8, 0xa0> raw;
+        AuxInfo aux_info;
+    };
+};
+static_assert(sizeof(EffectInStatus) == 0xc0, "EffectInStatus is an invalid size");
+
+struct EffectOutStatus {
+    EffectStatus state;
+    INSERT_PADDING_BYTES(0xf);
+};
+static_assert(sizeof(EffectOutStatus) == 0x10, "EffectOutStatus is an invalid size");
+
 struct UpdateDataHeader {
     UpdateDataHeader() {}
 
@@ -173,11 +220,13 @@ public:
     Stream::State GetStreamState() const;
 
 private:
+    class EffectState;
     class VoiceState;
 
     AudioRendererParameter worker_params;
     Kernel::SharedPtr<Kernel::Event> buffer_event;
     std::vector<VoiceState> voices;
+    std::vector<EffectState> effects;
     std::unique_ptr<AudioOut> audio_out;
     AudioCore::StreamPtr stream;
 };

src/common/CMakeLists.txt

@@ -29,7 +29,7 @@ if ($ENV{CI})
       if (BUILD_VERSION)
         # This leaves a trailing space on the last word, but we actually want that
         # because of how it's styled in the title bar.
-        set(BUILD_FULLNAME "${REPO_NAME} #${BUILD_VERSION} ")
+        set(BUILD_FULLNAME "${REPO_NAME} ${BUILD_VERSION} ")
       else()
         set(BUILD_FULLNAME "")
       endif()

src/common/hex_util.cpp

@@ -18,6 +18,25 @@ u8 ToHexNibble(char c1) {
     return 0;
 }
 
+std::vector<u8> HexStringToVector(std::string_view str, bool little_endian) {
+    std::vector<u8> out(str.size() / 2);
+    if (little_endian) {
+        for (std::size_t i = str.size() - 2; i <= str.size(); i -= 2)
+            out[i / 2] = (ToHexNibble(str[i]) << 4) | ToHexNibble(str[i + 1]);
+    } else {
+        for (std::size_t i = 0; i < str.size(); i += 2)
+            out[i / 2] = (ToHexNibble(str[i]) << 4) | ToHexNibble(str[i + 1]);
+    }
+    return out;
+}
+
+std::string HexVectorToString(const std::vector<u8>& vector, bool upper) {
+    std::string out;
+    for (u8 c : vector)
+        out += fmt::format(upper ? "{:02X}" : "{:02x}", c);
+    return out;
+}
+
 std::array<u8, 16> operator""_array16(const char* str, std::size_t len) {
     if (len != 32) {
         LOG_ERROR(Common,

src/common/hex_util.h

@@ -7,6 +7,7 @@
 #include <array>
 #include <cstddef>
 #include <string>
+#include <vector>
 #include <fmt/format.h>
 #include "common/common_types.h"
 
@@ -14,6 +15,8 @@ namespace Common {
 
 u8 ToHexNibble(char c1);
 
+std::vector<u8> HexStringToVector(std::string_view str, bool little_endian);
+
 template <std::size_t Size, bool le = false>
 std::array<u8, Size> HexStringToArray(std::string_view str) {
     std::array<u8, Size> out{};
@@ -27,6 +30,8 @@ std::array<u8, Size> HexStringToArray(std::string_view str) {
     return out;
 }
 
+std::string HexVectorToString(const std::vector<u8>& vector, bool upper = true);
+
 template <std::size_t Size>
 std::string HexArrayToString(std::array<u8, Size> array, bool upper = true) {
     std::string out;

src/common/param_package.cpp

@@ -20,7 +20,15 @@ constexpr char KEY_VALUE_SEPARATOR_ESCAPE[] = "$0";
 constexpr char PARAM_SEPARATOR_ESCAPE[] = "$1";
 constexpr char ESCAPE_CHARACTER_ESCAPE[] = "$2";
 
+/// A placeholder for empty param packages to avoid empty strings
+/// (they may be recognized as "not set" by some frontend libraries like qt)
+constexpr char EMPTY_PLACEHOLDER[] = "[empty]";
+
 ParamPackage::ParamPackage(const std::string& serialized) {
+    if (serialized == EMPTY_PLACEHOLDER) {
+        return;
+    }
+
     std::vector<std::string> pairs;
     Common::SplitString(serialized, PARAM_SEPARATOR, pairs);
 
@@ -46,7 +54,7 @@ ParamPackage::ParamPackage(std::initializer_list<DataType::value_type> list) : d
 
 std::string ParamPackage::Serialize() const {
     if (data.empty())
-        return "";
+        return EMPTY_PLACEHOLDER;
 
     std::string result;
 
@@ -120,4 +128,12 @@ bool ParamPackage::Has(const std::string& key) const {
     return data.find(key) != data.end();
 }
 
+void ParamPackage::Erase(const std::string& key) {
+    data.erase(key);
+}
+
+void ParamPackage::Clear() {
+    data.clear();
+}
+
 } // namespace Common

src/common/param_package.h

@@ -32,6 +32,8 @@ public:
     void Set(const std::string& key, int value);
     void Set(const std::string& key, float value);
     bool Has(const std::string& key) const;
+    void Erase(const std::string& key);
+    void Clear();
 
 private:
     DataType data;

src/core/CMakeLists.txt

@@ -18,6 +18,8 @@ add_library(core STATIC
     crypto/encryption_layer.h
     crypto/key_manager.cpp
     crypto/key_manager.h
+    crypto/partition_data_manager.cpp
+    crypto/partition_data_manager.h
     crypto/ctr_encryption_layer.cpp
     crypto/ctr_encryption_layer.h
     crypto/xts_encryption_layer.cpp
@@ -70,6 +72,7 @@ add_library(core STATIC
     file_sys/vfs_real.cpp
     file_sys/vfs_real.h
     file_sys/vfs_static.h
+    file_sys/vfs_types.h
     file_sys/vfs_vector.cpp
     file_sys/vfs_vector.h
     file_sys/xts_archive.cpp

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -129,7 +129,7 @@ public:
 };
 
 std::unique_ptr<Dynarmic::A64::Jit> ARM_Dynarmic::MakeJit() const {
-    auto& current_process = Core::CurrentProcess();
+    auto* current_process = Core::CurrentProcess();
     auto** const page_table = current_process->VMManager().page_table.pointers.data();
 
     Dynarmic::A64::UserConfig config;

src/core/core.cpp

@@ -136,7 +136,8 @@ struct System::Impl {
         if (virtual_filesystem == nullptr)
             virtual_filesystem = std::make_shared<FileSys::RealVfsFilesystem>();
 
-        kernel.MakeCurrentProcess(Kernel::Process::Create(kernel, "main"));
+        auto main_process = Kernel::Process::Create(kernel, "main");
+        kernel.MakeCurrentProcess(main_process.get());
 
         cpu_barrier = std::make_shared<CpuBarrier>();
         cpu_exclusive_monitor = Cpu::MakeExclusiveMonitor(cpu_cores.size());
@@ -361,11 +362,11 @@ const std::shared_ptr<Kernel::Scheduler>& System::Scheduler(std::size_t core_ind
     return impl->cpu_cores[core_index]->Scheduler();
 }
 
-Kernel::SharedPtr<Kernel::Process>& System::CurrentProcess() {
+Kernel::Process* System::CurrentProcess() {
     return impl->kernel.CurrentProcess();
 }
 
-const Kernel::SharedPtr<Kernel::Process>& System::CurrentProcess() const {
+const Kernel::Process* System::CurrentProcess() const {
     return impl->kernel.CurrentProcess();
 }
 

src/core/core.h

@@ -174,11 +174,11 @@ public:
     /// Gets the scheduler for the CPU core with the specified index
     const std::shared_ptr<Kernel::Scheduler>& Scheduler(std::size_t core_index);
 
-    /// Provides a reference to the current process
-    Kernel::SharedPtr<Kernel::Process>& CurrentProcess();
+    /// Provides a pointer to the current process
+    Kernel::Process* CurrentProcess();
 
-    /// Provides a constant reference to the current process.
-    const Kernel::SharedPtr<Kernel::Process>& CurrentProcess() const;
+    /// Provides a constant pointer to the current process.
+    const Kernel::Process* CurrentProcess() const;
 
     /// Provides a reference to the kernel instance.
     Kernel::KernelCore& Kernel();
@@ -246,7 +246,7 @@ inline TelemetrySession& Telemetry() {
     return System::GetInstance().TelemetrySession();
 }
 
-inline Kernel::SharedPtr<Kernel::Process>& CurrentProcess() {
+inline Kernel::Process* CurrentProcess() {
     return System::GetInstance().CurrentProcess();
 }
 

src/core/crypto/key_manager.cpp

@@ -4,23 +4,56 @@
 
 #include <algorithm>
 #include <array>
+#include <bitset>
+#include <cctype>
 #include <fstream>
 #include <locale>
+#include <map>
 #include <sstream>
 #include <string_view>
 #include <tuple>
 #include <vector>
+#include <mbedtls/bignum.h>
+#include <mbedtls/cipher.h>
+#include <mbedtls/cmac.h>
+#include <mbedtls/sha256.h>
+#include "common/common_funcs.h"
 #include "common/common_paths.h"
 #include "common/file_util.h"
 #include "common/hex_util.h"
 #include "common/logging/log.h"
 #include "core/crypto/aes_util.h"
 #include "core/crypto/key_manager.h"
+#include "core/crypto/partition_data_manager.h"
+#include "core/file_sys/content_archive.h"
+#include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/partition_filesystem.h"
+#include "core/file_sys/registered_cache.h"
+#include "core/hle/service/filesystem/filesystem.h"
 #include "core/loader/loader.h"
 #include "core/settings.h"
 
 namespace Core::Crypto {
 
+constexpr u64 CURRENT_CRYPTO_REVISION = 0x5;
+
+using namespace Common;
+
+const std::array<SHA256Hash, 2> eticket_source_hashes{
+    "B71DB271DC338DF380AA2C4335EF8873B1AFD408E80B3582D8719FC81C5E511C"_array32, // eticket_rsa_kek_source
+    "E8965A187D30E57869F562D04383C996DE487BBA5761363D2D4D32391866A85C"_array32, // eticket_rsa_kekek_source
+};
+
+const std::map<std::pair<S128KeyType, u64>, std::string> KEYS_VARIABLE_LENGTH{
+    {{S128KeyType::Master, 0}, "master_key_"},
+    {{S128KeyType::Package1, 0}, "package1_key_"},
+    {{S128KeyType::Package2, 0}, "package2_key_"},
+    {{S128KeyType::Titlekek, 0}, "titlekek_"},
+    {{S128KeyType::Source, static_cast<u64>(SourceKeyType::Keyblob)}, "keyblob_key_source_"},
+    {{S128KeyType::Keyblob, 0}, "keyblob_key_"},
+    {{S128KeyType::KeyblobMAC, 0}, "keyblob_mac_key_"},
+};
+
 Key128 GenerateKeyEncryptionKey(Key128 source, Key128 master, Key128 kek_seed, Key128 key_seed) {
     Key128 out{};
 
@@ -37,6 +70,77 @@ Key128 GenerateKeyEncryptionKey(Key128 source, Key128 master, Key128 kek_seed, K
     return out;
 }
 
+Key128 DeriveKeyblobKey(const Key128& sbk, const Key128& tsec, Key128 source) {
+    AESCipher<Key128> sbk_cipher(sbk, Mode::ECB);
+    AESCipher<Key128> tsec_cipher(tsec, Mode::ECB);
+    tsec_cipher.Transcode(source.data(), source.size(), source.data(), Op::Decrypt);
+    sbk_cipher.Transcode(source.data(), source.size(), source.data(), Op::Decrypt);
+    return source;
+}
+
+Key128 DeriveMasterKey(const std::array<u8, 0x90>& keyblob, const Key128& master_source) {
+    Key128 master_root;
+    std::memcpy(master_root.data(), keyblob.data(), sizeof(Key128));
+
+    AESCipher<Key128> master_cipher(master_root, Mode::ECB);
+
+    Key128 master{};
+    master_cipher.Transcode(master_source.data(), master_source.size(), master.data(), Op::Decrypt);
+    return master;
+}
+
+std::array<u8, 144> DecryptKeyblob(const std::array<u8, 176>& encrypted_keyblob,
+                                   const Key128& key) {
+    std::array<u8, 0x90> keyblob;
+    AESCipher<Key128> cipher(key, Mode::CTR);
+    cipher.SetIV(std::vector<u8>(encrypted_keyblob.data() + 0x10, encrypted_keyblob.data() + 0x20));
+    cipher.Transcode(encrypted_keyblob.data() + 0x20, keyblob.size(), keyblob.data(), Op::Decrypt);
+    return keyblob;
+}
+
+void KeyManager::DeriveGeneralPurposeKeys(u8 crypto_revision) {
+    const auto kek_generation_source =
+        GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKekGeneration));
+    const auto key_generation_source =
+        GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKeyGeneration));
+
+    if (HasKey(S128KeyType::Master, crypto_revision)) {
+        for (auto kak_type :
+             {KeyAreaKeyType::Application, KeyAreaKeyType::Ocean, KeyAreaKeyType::System}) {
+            if (HasKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::KeyAreaKey),
+                       static_cast<u64>(kak_type))) {
+                const auto source =
+                    GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::KeyAreaKey),
+                           static_cast<u64>(kak_type));
+                const auto kek =
+                    GenerateKeyEncryptionKey(source, GetKey(S128KeyType::Master, crypto_revision),
+                                             kek_generation_source, key_generation_source);
+                SetKey(S128KeyType::KeyArea, kek, crypto_revision, static_cast<u64>(kak_type));
+            }
+        }
+
+        AESCipher<Key128> master_cipher(GetKey(S128KeyType::Master, crypto_revision), Mode::ECB);
+        for (auto key_type : {SourceKeyType::Titlekek, SourceKeyType::Package2}) {
+            if (HasKey(S128KeyType::Source, static_cast<u64>(key_type))) {
+                Key128 key{};
+                master_cipher.Transcode(
+                    GetKey(S128KeyType::Source, static_cast<u64>(key_type)).data(), key.size(),
+                    key.data(), Op::Decrypt);
+                SetKey(key_type == SourceKeyType::Titlekek ? S128KeyType::Titlekek
+                                                           : S128KeyType::Package2,
+                       key, crypto_revision);
+            }
+        }
+    }
+}
+
+Key128 DeriveKeyblobMACKey(const Key128& keyblob_key, const Key128& mac_source) {
+    AESCipher<Key128> mac_cipher(keyblob_key, Mode::ECB);
+    Key128 mac_key{};
+    mac_cipher.Transcode(mac_source.data(), mac_key.size(), mac_key.data(), Op::Decrypt);
+    return mac_key;
+}
+
 boost::optional<Key128> DeriveSDSeed() {
     const FileUtil::IOFile save_43(FileUtil::GetUserPath(FileUtil::UserPath::NANDDir) +
                                        "/system/save/8000000000000043",
@@ -71,23 +175,24 @@ boost::optional<Key128> DeriveSDSeed() {
     return seed;
 }
 
-Loader::ResultStatus DeriveSDKeys(std::array<Key256, 2>& sd_keys, const KeyManager& keys) {
-    if (!keys.HasKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::SDKEK)))
+Loader::ResultStatus DeriveSDKeys(std::array<Key256, 2>& sd_keys, KeyManager& keys) {
+    if (!keys.HasKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::SDKek)))
         return Loader::ResultStatus::ErrorMissingSDKEKSource;
-    if (!keys.HasKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKEKGeneration)))
+    if (!keys.HasKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKekGeneration)))
         return Loader::ResultStatus::ErrorMissingAESKEKGenerationSource;
     if (!keys.HasKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKeyGeneration)))
         return Loader::ResultStatus::ErrorMissingAESKeyGenerationSource;
 
     const auto sd_kek_source =
-        keys.GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::SDKEK));
+        keys.GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::SDKek));
     const auto aes_kek_gen =
-        keys.GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKEKGeneration));
+        keys.GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKekGeneration));
     const auto aes_key_gen =
         keys.GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKeyGeneration));
     const auto master_00 = keys.GetKey(S128KeyType::Master);
     const auto sd_kek =
         GenerateKeyEncryptionKey(sd_kek_source, master_00, aes_kek_gen, aes_key_gen);
+    keys.SetKey(S128KeyType::SDKek, sd_kek);
 
     if (!keys.HasKey(S128KeyType::SDSeed))
         return Loader::ResultStatus::ErrorMissingSDSeed;
@@ -118,9 +223,141 @@ Loader::ResultStatus DeriveSDKeys(std::array<Key256, 2>& sd_keys, const KeyManag
                        return source; ///< Return unaltered source to satisfy output requirement.
                    });
 
+    keys.SetKey(S256KeyType::SDKey, sd_keys[0], static_cast<u64>(SDKeyType::Save));
+    keys.SetKey(S256KeyType::SDKey, sd_keys[1], static_cast<u64>(SDKeyType::NCA));
+
     return Loader::ResultStatus::Success;
 }
 
+std::vector<TicketRaw> GetTicketblob(const FileUtil::IOFile& ticket_save) {
+    if (!ticket_save.IsOpen())
+        return {};
+
+    std::vector<u8> buffer(ticket_save.GetSize());
+    ticket_save.ReadBytes(buffer.data(), buffer.size());
+
+    std::vector<TicketRaw> out;
+    u32 magic{};
+    for (std::size_t offset = 0; offset + 0x4 < buffer.size(); ++offset) {
+        if (buffer[offset] == 0x4 && buffer[offset + 1] == 0x0 && buffer[offset + 2] == 0x1 &&
+            buffer[offset + 3] == 0x0) {
+            out.emplace_back();
+            auto& next = out.back();
+            std::memcpy(&next, buffer.data() + offset, sizeof(TicketRaw));
+            offset += next.size();
+        }
+    }
+
+    return out;
+}
+
+template <size_t size>
+static std::array<u8, size> operator^(const std::array<u8, size>& lhs,
+                                      const std::array<u8, size>& rhs) {
+    std::array<u8, size> out{};
+    std::transform(lhs.begin(), lhs.end(), rhs.begin(), out.begin(), std::bit_xor<>());
+    return out;
+}
+
+template <size_t target_size, size_t in_size>
+static std::array<u8, target_size> MGF1(const std::array<u8, in_size>& seed) {
+    std::array<u8, in_size + 4> seed_exp{};
+    std::memcpy(seed_exp.data(), seed.data(), in_size);
+
+    std::vector<u8> out;
+    size_t i = 0;
+    while (out.size() < target_size) {
+        out.resize(out.size() + 0x20);
+        seed_exp[in_size + 3] = i;
+        mbedtls_sha256(seed_exp.data(), seed_exp.size(), out.data() + out.size() - 0x20, 0);
+        ++i;
+    }
+
+    std::array<u8, target_size> target;
+    std::memcpy(target.data(), out.data(), target_size);
+    return target;
+}
+
+template <size_t size>
+static boost::optional<u64> FindTicketOffset(const std::array<u8, size>& data) {
+    u64 offset = 0;
+    for (size_t i = 0x20; i < data.size() - 0x10; ++i) {
+        if (data[i] == 0x1) {
+            offset = i + 1;
+            break;
+        } else if (data[i] != 0x0) {
+            return boost::none;
+        }
+    }
+
+    return offset;
+}
+
+boost::optional<std::pair<Key128, Key128>> ParseTicket(const TicketRaw& ticket,
+                                                       const RSAKeyPair<2048>& key) {
+    u32 cert_authority;
+    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'))
+        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));
+
+    if (rights_id == Key128{})
+        return boost::none;
+
+    Key128 key_temp{};
+
+    if (!std::any_of(ticket.begin() + 0x190, ticket.begin() + 0x280, [](u8 b) { return b != 0; })) {
+        std::memcpy(key_temp.data(), ticket.data() + 0x180, key_temp.size());
+        return std::make_pair(rights_id, key_temp);
+    }
+
+    mbedtls_mpi D; // RSA Private Exponent
+    mbedtls_mpi N; // RSA Modulus
+    mbedtls_mpi S; // Input
+    mbedtls_mpi M; // Output
+
+    mbedtls_mpi_init(&D);
+    mbedtls_mpi_init(&N);
+    mbedtls_mpi_init(&S);
+    mbedtls_mpi_init(&M);
+
+    mbedtls_mpi_read_binary(&D, key.decryption_key.data(), key.decryption_key.size());
+    mbedtls_mpi_read_binary(&N, key.modulus.data(), key.modulus.size());
+    mbedtls_mpi_read_binary(&S, ticket.data() + 0x180, 0x100);
+
+    mbedtls_mpi_exp_mod(&M, &S, &D, &N, nullptr);
+
+    std::array<u8, 0x100> rsa_step;
+    mbedtls_mpi_write_binary(&M, rsa_step.data(), rsa_step.size());
+
+    u8 m_0 = rsa_step[0];
+    std::array<u8, 0x20> m_1;
+    std::memcpy(m_1.data(), rsa_step.data() + 0x01, m_1.size());
+    std::array<u8, 0xDF> m_2;
+    std::memcpy(m_2.data(), rsa_step.data() + 0x21, m_2.size());
+
+    if (m_0 != 0)
+        return boost::none;
+
+    m_1 = m_1 ^ MGF1<0x20>(m_2);
+    m_2 = m_2 ^ MGF1<0xDF>(m_1);
+
+    const auto offset = FindTicketOffset(m_2);
+    if (offset == boost::none)
+        return boost::none;
+    ASSERT(offset.get() > 0);
+
+    std::memcpy(key_temp.data(), m_2.data() + offset.get(), key_temp.size());
+
+    return std::make_pair(rights_id, key_temp);
+}
+
 KeyManager::KeyManager() {
     // Initialize keys
     const std::string hactool_keys_dir = FileUtil::GetHactoolConfigurationPath();
@@ -137,6 +374,15 @@ KeyManager::KeyManager() {
 
     AttemptLoadKeyFile(yuzu_keys_dir, hactool_keys_dir, "title.keys", true);
     AttemptLoadKeyFile(yuzu_keys_dir, yuzu_keys_dir, "title.keys_autogenerated", true);
+    AttemptLoadKeyFile(yuzu_keys_dir, hactool_keys_dir, "console.keys", false);
+    AttemptLoadKeyFile(yuzu_keys_dir, yuzu_keys_dir, "console.keys_autogenerated", false);
+}
+
+static bool ValidCryptoRevisionString(std::string_view base, size_t begin, size_t length) {
+    if (base.size() < begin + length)
+        return false;
+    return std::all_of(base.begin() + begin, base.begin() + begin + length,
+                       [](u8 c) { return std::isdigit(c); });
 }
 
 void KeyManager::LoadFromFile(const std::string& filename, bool is_title_keys) {
@@ -158,6 +404,9 @@ void KeyManager::LoadFromFile(const std::string& filename, bool is_title_keys) {
         out[0].erase(std::remove(out[0].begin(), out[0].end(), ' '), out[0].end());
         out[1].erase(std::remove(out[1].begin(), out[1].end(), ' '), out[1].end());
 
+        if (out[0].compare(0, 1, "#") == 0)
+            continue;
+
         if (is_title_keys) {
             auto rights_id_raw = Common::HexStringToArray<16>(out[0]);
             u128 rights_id{};
@@ -174,6 +423,50 @@ void KeyManager::LoadFromFile(const std::string& filename, bool is_title_keys) {
                 const auto index = s256_file_id.at(out[0]);
                 Key256 key = Common::HexStringToArray<32>(out[1]);
                 s256_keys[{index.type, index.field1, index.field2}] = key;
+            } else if (out[0].compare(0, 8, "keyblob_") == 0 &&
+                       out[0].compare(0, 9, "keyblob_k") != 0) {
+                if (!ValidCryptoRevisionString(out[0], 8, 2))
+                    continue;
+
+                const auto index = std::stoul(out[0].substr(8, 2), nullptr, 16);
+                keyblobs[index] = Common::HexStringToArray<0x90>(out[1]);
+            } else if (out[0].compare(0, 18, "encrypted_keyblob_") == 0) {
+                if (!ValidCryptoRevisionString(out[0], 18, 2))
+                    continue;
+
+                const auto index = std::stoul(out[0].substr(18, 2), nullptr, 16);
+                encrypted_keyblobs[index] = Common::HexStringToArray<0xB0>(out[1]);
+            } else {
+                for (const auto& kv : KEYS_VARIABLE_LENGTH) {
+                    if (!ValidCryptoRevisionString(out[0], kv.second.size(), 2))
+                        continue;
+                    if (out[0].compare(0, kv.second.size(), kv.second) == 0) {
+                        const auto index =
+                            std::stoul(out[0].substr(kv.second.size(), 2), nullptr, 16);
+                        const auto sub = kv.first.second;
+                        if (sub == 0) {
+                            s128_keys[{kv.first.first, index, 0}] =
+                                Common::HexStringToArray<16>(out[1]);
+                        } else {
+                            s128_keys[{kv.first.first, kv.first.second, index}] =
+                                Common::HexStringToArray<16>(out[1]);
+                        }
+
+                        break;
+                    }
+                }
+
+                static constexpr std::array<const char*, 3> kak_names = {
+                    "key_area_key_application_", "key_area_key_ocean_", "key_area_key_system_"};
+                for (size_t j = 0; j < kak_names.size(); ++j) {
+                    const auto& match = kak_names[j];
+                    if (out[0].compare(0, std::strlen(match), match) == 0) {
+                        const auto index =
+                            std::stoul(out[0].substr(std::strlen(match), 2), nullptr, 16);
+                        s128_keys[{S128KeyType::KeyArea, index, j}] =
+                            Common::HexStringToArray<16>(out[1]);
+                    }
+                }
             }
         }
     }
@@ -187,6 +480,28 @@ void KeyManager::AttemptLoadKeyFile(const std::string& dir1, const std::string&
         LoadFromFile(dir2 + DIR_SEP + filename, title);
 }
 
+bool KeyManager::BaseDeriveNecessary() const {
+    const auto check_key_existence = [this](auto key_type, u64 index1 = 0, u64 index2 = 0) {
+        return !HasKey(key_type, index1, index2);
+    };
+
+    if (check_key_existence(S256KeyType::Header))
+        return true;
+
+    for (size_t i = 0; i < CURRENT_CRYPTO_REVISION; ++i) {
+        if (check_key_existence(S128KeyType::Master, i) ||
+            check_key_existence(S128KeyType::KeyArea, i,
+                                static_cast<u64>(KeyAreaKeyType::Application)) ||
+            check_key_existence(S128KeyType::KeyArea, i, static_cast<u64>(KeyAreaKeyType::Ocean)) ||
+            check_key_existence(S128KeyType::KeyArea, i,
+                                static_cast<u64>(KeyAreaKeyType::System)) ||
+            check_key_existence(S128KeyType::Titlekek, i))
+            return true;
+    }
+
+    return false;
+}
+
 bool KeyManager::HasKey(S128KeyType id, u64 field1, u64 field2) const {
     return s128_keys.find({id, field1, field2}) != s128_keys.end();
 }
@@ -207,13 +522,30 @@ Key256 KeyManager::GetKey(S256KeyType id, u64 field1, u64 field2) const {
     return s256_keys.at({id, field1, field2});
 }
 
-template <std::size_t Size>
-void KeyManager::WriteKeyToFile(bool title_key, std::string_view keyname,
+Key256 KeyManager::GetBISKey(u8 partition_id) const {
+    Key256 out{};
+
+    for (const auto& bis_type : {BISKeyType::Crypto, BISKeyType::Tweak}) {
+        if (HasKey(S128KeyType::BIS, partition_id, static_cast<u64>(bis_type))) {
+            std::memcpy(
+                out.data() + sizeof(Key128) * static_cast<u64>(bis_type),
+                s128_keys.at({S128KeyType::BIS, partition_id, static_cast<u64>(bis_type)}).data(),
+                sizeof(Key128));
+        }
+    }
+
+    return out;
+}
+
+template <size_t Size>
+void KeyManager::WriteKeyToFile(KeyCategory category, std::string_view keyname,
                                 const std::array<u8, Size>& key) {
     const std::string yuzu_keys_dir = FileUtil::GetUserPath(FileUtil::UserPath::KeysDir);
     std::string filename = "title.keys_autogenerated";
-    if (!title_key)
+    if (category == KeyCategory::Standard)
         filename = dev_mode ? "dev.keys_autogenerated" : "prod.keys_autogenerated";
+    else if (category == KeyCategory::Console)
+        filename = "console.keys_autogenerated";
     const auto add_info_text = !FileUtil::Exists(yuzu_keys_dir + DIR_SEP + filename);
     FileUtil::CreateFullPath(yuzu_keys_dir + DIR_SEP + filename);
     std::ofstream file(yuzu_keys_dir + DIR_SEP + filename, std::ios::app);
@@ -227,7 +559,7 @@ void KeyManager::WriteKeyToFile(bool title_key, std::string_view keyname,
     }
 
     file << fmt::format("\n{} = {}", keyname, Common::HexArrayToString(key));
-    AttemptLoadKeyFile(yuzu_keys_dir, yuzu_keys_dir, filename, title_key);
+    AttemptLoadKeyFile(yuzu_keys_dir, yuzu_keys_dir, filename, category == KeyCategory::Title);
 }
 
 void KeyManager::SetKey(S128KeyType id, Key128 key, u64 field1, u64 field2) {
@@ -237,8 +569,15 @@ void KeyManager::SetKey(S128KeyType id, Key128 key, u64 field1, u64 field2) {
         Key128 rights_id;
         std::memcpy(rights_id.data(), &field2, sizeof(u64));
         std::memcpy(rights_id.data() + sizeof(u64), &field1, sizeof(u64));
-        WriteKeyToFile(true, Common::HexArrayToString(rights_id), key);
+        WriteKeyToFile(KeyCategory::Title, Common::HexArrayToString(rights_id), key);
     }
+
+    auto category = KeyCategory::Standard;
+    if (id == S128KeyType::Keyblob || id == S128KeyType::KeyblobMAC || id == S128KeyType::TSEC ||
+        id == S128KeyType::SecureBoot || id == S128KeyType::SDSeed || id == S128KeyType::BIS) {
+        category = KeyCategory::Console;
+    }
+
     const auto iter2 = std::find_if(
         s128_file_id.begin(), s128_file_id.end(),
         [&id, &field1, &field2](const std::pair<std::string, KeyIndex<S128KeyType>> elem) {
@@ -246,7 +585,30 @@ void KeyManager::SetKey(S128KeyType id, Key128 key, u64 field1, u64 field2) {
                    std::tie(id, field1, field2);
         });
     if (iter2 != s128_file_id.end())
-        WriteKeyToFile(false, iter2->first, key);
+        WriteKeyToFile(category, iter2->first, key);
+
+    // Variable cases
+    if (id == S128KeyType::KeyArea) {
+        static constexpr std::array<const char*, 3> kak_names = {"key_area_key_application_{:02X}",
+                                                                 "key_area_key_ocean_{:02X}",
+                                                                 "key_area_key_system_{:02X}"};
+        WriteKeyToFile(category, fmt::format(kak_names.at(field2), field1), key);
+    } else if (id == S128KeyType::Master) {
+        WriteKeyToFile(category, fmt::format("master_key_{:02X}", field1), key);
+    } else if (id == S128KeyType::Package1) {
+        WriteKeyToFile(category, fmt::format("package1_key_{:02X}", field1), key);
+    } else if (id == S128KeyType::Package2) {
+        WriteKeyToFile(category, fmt::format("package2_key_{:02X}", field1), key);
+    } else if (id == S128KeyType::Titlekek) {
+        WriteKeyToFile(category, fmt::format("titlekek_{:02X}", field1), key);
+    } else if (id == S128KeyType::Keyblob) {
+        WriteKeyToFile(category, fmt::format("keyblob_key_{:02X}", field1), key);
+    } else if (id == S128KeyType::KeyblobMAC) {
+        WriteKeyToFile(category, fmt::format("keyblob_mac_key_{:02X}", field1), key);
+    } else if (id == S128KeyType::Source && field1 == static_cast<u64>(SourceKeyType::Keyblob)) {
+        WriteKeyToFile(category, fmt::format("keyblob_key_source_{:02X}", field2), key);
+    }
+
     s128_keys[{id, field1, field2}] = key;
 }
 
@@ -260,7 +622,7 @@ void KeyManager::SetKey(S256KeyType id, Key256 key, u64 field1, u64 field2) {
                    std::tie(id, field1, field2);
         });
     if (iter != s256_file_id.end())
-        WriteKeyToFile(false, iter->first, key);
+        WriteKeyToFile(KeyCategory::Standard, iter->first, key);
     s256_keys[{id, field1, field2}] = key;
 }
 
@@ -290,59 +652,388 @@ void KeyManager::DeriveSDSeedLazy() {
         SetKey(S128KeyType::SDSeed, res.get());
 }
 
+static Key128 CalculateCMAC(const u8* source, size_t size, const Key128& key) {
+    Key128 out{};
+
+    mbedtls_cipher_cmac(mbedtls_cipher_info_from_type(MBEDTLS_CIPHER_AES_128_ECB), key.data(),
+                        key.size() * 8, source, size, out.data());
+    return out;
+}
+
+void KeyManager::DeriveBase() {
+    if (!BaseDeriveNecessary())
+        return;
+
+    if (!HasKey(S128KeyType::SecureBoot) || !HasKey(S128KeyType::TSEC))
+        return;
+
+    const auto has_bis = [this](u64 id) {
+        return HasKey(S128KeyType::BIS, id, static_cast<u64>(BISKeyType::Crypto)) &&
+               HasKey(S128KeyType::BIS, id, static_cast<u64>(BISKeyType::Tweak));
+    };
+
+    const auto copy_bis = [this](u64 id_from, u64 id_to) {
+        SetKey(S128KeyType::BIS,
+               GetKey(S128KeyType::BIS, id_from, static_cast<u64>(BISKeyType::Crypto)), id_to,
+               static_cast<u64>(BISKeyType::Crypto));
+
+        SetKey(S128KeyType::BIS,
+               GetKey(S128KeyType::BIS, id_from, static_cast<u64>(BISKeyType::Tweak)), id_to,
+               static_cast<u64>(BISKeyType::Tweak));
+    };
+
+    if (has_bis(2) && !has_bis(3))
+        copy_bis(2, 3);
+    else if (has_bis(3) && !has_bis(2))
+        copy_bis(3, 2);
+
+    std::bitset<32> revisions(0xFFFFFFFF);
+    for (size_t i = 0; i < revisions.size(); ++i) {
+        if (!HasKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::Keyblob), i) ||
+            encrypted_keyblobs[i] == std::array<u8, 0xB0>{}) {
+            revisions.reset(i);
+        }
+    }
+
+    if (!revisions.any())
+        return;
+
+    const auto sbk = GetKey(S128KeyType::SecureBoot);
+    const auto tsec = GetKey(S128KeyType::TSEC);
+    const auto master_source = GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::Master));
+
+    for (size_t i = 0; i < revisions.size(); ++i) {
+        if (!revisions[i])
+            continue;
+
+        // Derive keyblob key
+        const auto key = DeriveKeyblobKey(
+            sbk, tsec, GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::Keyblob), i));
+
+        SetKey(S128KeyType::Keyblob, key, i);
+
+        // Derive keyblob MAC key
+        if (!HasKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::KeyblobMAC)))
+            continue;
+
+        const auto mac_key = DeriveKeyblobMACKey(
+            key, GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::KeyblobMAC)));
+        SetKey(S128KeyType::KeyblobMAC, mac_key, i);
+
+        Key128 cmac = CalculateCMAC(encrypted_keyblobs[i].data() + 0x10, 0xA0, mac_key);
+        if (std::memcmp(cmac.data(), encrypted_keyblobs[i].data(), cmac.size()) != 0)
+            continue;
+
+        // Decrypt keyblob
+        if (keyblobs[i] == std::array<u8, 0x90>{}) {
+            keyblobs[i] = DecryptKeyblob(encrypted_keyblobs[i], key);
+            WriteKeyToFile<0x90>(KeyCategory::Console, fmt::format("keyblob_{:02X}", i),
+                                 keyblobs[i]);
+        }
+
+        Key128 package1;
+        std::memcpy(package1.data(), keyblobs[i].data() + 0x80, sizeof(Key128));
+        SetKey(S128KeyType::Package1, package1, i);
+
+        // Derive master key
+        if (HasKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::Master))) {
+            SetKey(S128KeyType::Master,
+                   DeriveMasterKey(keyblobs[i], GetKey(S128KeyType::Source,
+                                                       static_cast<u64>(SourceKeyType::Master))),
+                   i);
+        }
+    }
+
+    revisions.set();
+    for (size_t i = 0; i < revisions.size(); ++i) {
+        if (!HasKey(S128KeyType::Master, i))
+            revisions.reset(i);
+    }
+
+    if (!revisions.any())
+        return;
+
+    for (size_t i = 0; i < revisions.size(); ++i) {
+        if (!revisions[i])
+            continue;
+
+        // Derive general purpose keys
+        DeriveGeneralPurposeKeys(i);
+    }
+
+    if (HasKey(S128KeyType::Master, 0) &&
+        HasKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKeyGeneration)) &&
+        HasKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKekGeneration)) &&
+        HasKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::HeaderKek)) &&
+        HasKey(S256KeyType::HeaderSource)) {
+        const auto header_kek = GenerateKeyEncryptionKey(
+            GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::HeaderKek)),
+            GetKey(S128KeyType::Master, 0),
+            GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKekGeneration)),
+            GetKey(S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKeyGeneration)));
+        SetKey(S128KeyType::HeaderKek, header_kek);
+
+        AESCipher<Key128> header_cipher(header_kek, Mode::ECB);
+        Key256 out = GetKey(S256KeyType::HeaderSource);
+        header_cipher.Transcode(out.data(), out.size(), out.data(), Op::Decrypt);
+        SetKey(S256KeyType::Header, out);
+    }
+}
+
+void KeyManager::DeriveETicket(PartitionDataManager& data) {
+    // ETicket keys
+    const auto es = Service::FileSystem::GetUnionContents()->GetEntry(
+        0x0100000000000033, FileSys::ContentRecordType::Program);
+
+    if (es == nullptr)
+        return;
+
+    const auto exefs = es->GetExeFS();
+    if (exefs == nullptr)
+        return;
+
+    const auto main = exefs->GetFile("main");
+    if (main == nullptr)
+        return;
+
+    const auto bytes = main->ReadAllBytes();
+
+    const auto eticket_kek = FindKeyFromHex16(bytes, eticket_source_hashes[0]);
+    const auto eticket_kekek = FindKeyFromHex16(bytes, eticket_source_hashes[1]);
+
+    const auto seed3 = data.GetRSAKekSeed3();
+    const auto mask0 = data.GetRSAKekMask0();
+
+    if (eticket_kek != Key128{})
+        SetKey(S128KeyType::Source, eticket_kek, static_cast<size_t>(SourceKeyType::ETicketKek));
+    if (eticket_kekek != Key128{}) {
+        SetKey(S128KeyType::Source, eticket_kekek,
+               static_cast<size_t>(SourceKeyType::ETicketKekek));
+    }
+    if (seed3 != Key128{})
+        SetKey(S128KeyType::RSAKek, seed3, static_cast<size_t>(RSAKekType::Seed3));
+    if (mask0 != Key128{})
+        SetKey(S128KeyType::RSAKek, mask0, static_cast<size_t>(RSAKekType::Mask0));
+    if (eticket_kek == Key128{} || eticket_kekek == Key128{} || seed3 == Key128{} ||
+        mask0 == Key128{}) {
+        return;
+    }
+
+    Key128 rsa_oaep_kek{};
+    std::transform(seed3.begin(), seed3.end(), mask0.begin(), rsa_oaep_kek.begin(),
+                   std::bit_xor<>());
+
+    if (rsa_oaep_kek == Key128{})
+        return;
+
+    SetKey(S128KeyType::Source, rsa_oaep_kek,
+           static_cast<u64>(SourceKeyType::RSAOaepKekGeneration));
+
+    Key128 temp_kek{};
+    Key128 temp_kekek{};
+    Key128 eticket_final{};
+
+    // Derive ETicket RSA Kek
+    AESCipher<Key128> es_master(GetKey(S128KeyType::Master), Mode::ECB);
+    es_master.Transcode(rsa_oaep_kek.data(), rsa_oaep_kek.size(), temp_kek.data(), Op::Decrypt);
+    AESCipher<Key128> es_kekek(temp_kek, Mode::ECB);
+    es_kekek.Transcode(eticket_kekek.data(), eticket_kekek.size(), temp_kekek.data(), Op::Decrypt);
+    AESCipher<Key128> es_kek(temp_kekek, Mode::ECB);
+    es_kek.Transcode(eticket_kek.data(), eticket_kek.size(), eticket_final.data(), Op::Decrypt);
+
+    if (eticket_final == Key128{})
+        return;
+
+    SetKey(S128KeyType::ETicketRSAKek, eticket_final);
+
+    // Titlekeys
+    data.DecryptProdInfo(GetBISKey(0));
+
+    const auto eticket_extended_kek = data.GetETicketExtendedKek();
+
+    std::vector<u8> extended_iv(0x10);
+    std::memcpy(extended_iv.data(), eticket_extended_kek.data(), extended_iv.size());
+    std::array<u8, 0x230> extended_dec{};
+    AESCipher<Key128> rsa_1(eticket_final, Mode::CTR);
+    rsa_1.SetIV(extended_iv);
+    rsa_1.Transcode(eticket_extended_kek.data() + 0x10, eticket_extended_kek.size() - 0x10,
+                    extended_dec.data(), Op::Decrypt);
+
+    RSAKeyPair<2048> rsa_key{};
+    std::memcpy(rsa_key.decryption_key.data(), extended_dec.data(), rsa_key.decryption_key.size());
+    std::memcpy(rsa_key.modulus.data(), extended_dec.data() + 0x100, rsa_key.modulus.size());
+    std::memcpy(rsa_key.exponent.data(), extended_dec.data() + 0x200, rsa_key.exponent.size());
+
+    const FileUtil::IOFile save1(FileUtil::GetUserPath(FileUtil::UserPath::NANDDir) +
+                                     "/system/save/80000000000000e1",
+                                 "rb+");
+    const FileUtil::IOFile save2(FileUtil::GetUserPath(FileUtil::UserPath::NANDDir) +
+                                     "/system/save/80000000000000e2",
+                                 "rb+");
+
+    auto res = GetTicketblob(save1);
+    const auto res2 = GetTicketblob(save2);
+    std::copy(res2.begin(), res2.end(), std::back_inserter(res));
+
+    for (const auto& raw : res) {
+        const auto pair = ParseTicket(raw, rsa_key);
+        if (pair == boost::none)
+            continue;
+        const auto& [rid, key] = pair.value();
+        u128 rights_id;
+        std::memcpy(rights_id.data(), rid.data(), rid.size());
+        SetKey(S128KeyType::Titlekey, key, rights_id[1], rights_id[0]);
+    }
+}
+
+void KeyManager::SetKeyWrapped(S128KeyType id, Key128 key, u64 field1, u64 field2) {
+    if (key == Key128{})
+        return;
+    SetKey(id, key, field1, field2);
+}
+
+void KeyManager::SetKeyWrapped(S256KeyType id, Key256 key, u64 field1, u64 field2) {
+    if (key == Key256{})
+        return;
+    SetKey(id, key, field1, field2);
+}
+
+void KeyManager::PopulateFromPartitionData(PartitionDataManager& data) {
+    if (!BaseDeriveNecessary())
+        return;
+
+    if (!data.HasBoot0())
+        return;
+
+    for (size_t i = 0; i < encrypted_keyblobs.size(); ++i) {
+        if (encrypted_keyblobs[i] != std::array<u8, 0xB0>{})
+            continue;
+        encrypted_keyblobs[i] = data.GetEncryptedKeyblob(i);
+        WriteKeyToFile<0xB0>(KeyCategory::Console, fmt::format("encrypted_keyblob_{:02X}", i),
+                             encrypted_keyblobs[i]);
+    }
+
+    SetKeyWrapped(S128KeyType::Source, data.GetPackage2KeySource(),
+                  static_cast<u64>(SourceKeyType::Package2));
+    SetKeyWrapped(S128KeyType::Source, data.GetAESKekGenerationSource(),
+                  static_cast<u64>(SourceKeyType::AESKekGeneration));
+    SetKeyWrapped(S128KeyType::Source, data.GetTitlekekSource(),
+                  static_cast<u64>(SourceKeyType::Titlekek));
+    SetKeyWrapped(S128KeyType::Source, data.GetMasterKeySource(),
+                  static_cast<u64>(SourceKeyType::Master));
+    SetKeyWrapped(S128KeyType::Source, data.GetKeyblobMACKeySource(),
+                  static_cast<u64>(SourceKeyType::KeyblobMAC));
+
+    for (size_t i = 0; i < PartitionDataManager::MAX_KEYBLOB_SOURCE_HASH; ++i) {
+        SetKeyWrapped(S128KeyType::Source, data.GetKeyblobKeySource(i),
+                      static_cast<u64>(SourceKeyType::Keyblob), i);
+    }
+
+    if (data.HasFuses())
+        SetKeyWrapped(S128KeyType::SecureBoot, data.GetSecureBootKey());
+
+    DeriveBase();
+
+    Key128 latest_master{};
+    for (s8 i = 0x1F; i >= 0; --i) {
+        if (GetKey(S128KeyType::Master, static_cast<u8>(i)) != Key128{}) {
+            latest_master = GetKey(S128KeyType::Master, static_cast<u8>(i));
+            break;
+        }
+    }
+
+    const auto masters = data.GetTZMasterKeys(latest_master);
+    for (size_t i = 0; i < masters.size(); ++i) {
+        if (masters[i] != Key128{} && !HasKey(S128KeyType::Master, i))
+            SetKey(S128KeyType::Master, masters[i], i);
+    }
+
+    DeriveBase();
+
+    if (!data.HasPackage2())
+        return;
+
+    std::array<Key128, 0x20> package2_keys{};
+    for (size_t i = 0; i < package2_keys.size(); ++i) {
+        if (HasKey(S128KeyType::Package2, i))
+            package2_keys[i] = GetKey(S128KeyType::Package2, i);
+    }
+    data.DecryptPackage2(package2_keys, Package2Type::NormalMain);
+
+    SetKeyWrapped(S128KeyType::Source, data.GetKeyAreaKeyApplicationSource(),
+                  static_cast<u64>(SourceKeyType::KeyAreaKey),
+                  static_cast<u64>(KeyAreaKeyType::Application));
+    SetKeyWrapped(S128KeyType::Source, data.GetKeyAreaKeyOceanSource(),
+                  static_cast<u64>(SourceKeyType::KeyAreaKey),
+                  static_cast<u64>(KeyAreaKeyType::Ocean));
+    SetKeyWrapped(S128KeyType::Source, data.GetKeyAreaKeySystemSource(),
+                  static_cast<u64>(SourceKeyType::KeyAreaKey),
+                  static_cast<u64>(KeyAreaKeyType::System));
+    SetKeyWrapped(S128KeyType::Source, data.GetSDKekSource(),
+                  static_cast<u64>(SourceKeyType::SDKek));
+    SetKeyWrapped(S256KeyType::SDKeySource, data.GetSDSaveKeySource(),
+                  static_cast<u64>(SDKeyType::Save));
+    SetKeyWrapped(S256KeyType::SDKeySource, data.GetSDNCAKeySource(),
+                  static_cast<u64>(SDKeyType::NCA));
+    SetKeyWrapped(S128KeyType::Source, data.GetHeaderKekSource(),
+                  static_cast<u64>(SourceKeyType::HeaderKek));
+    SetKeyWrapped(S256KeyType::HeaderSource, data.GetHeaderKeySource());
+    SetKeyWrapped(S128KeyType::Source, data.GetAESKeyGenerationSource(),
+                  static_cast<u64>(SourceKeyType::AESKeyGeneration));
+
+    DeriveBase();
+}
+
 const boost::container::flat_map<std::string, KeyIndex<S128KeyType>> KeyManager::s128_file_id = {
-    {"master_key_00", {S128KeyType::Master, 0, 0}},
-    {"master_key_01", {S128KeyType::Master, 1, 0}},
-    {"master_key_02", {S128KeyType::Master, 2, 0}},
-    {"master_key_03", {S128KeyType::Master, 3, 0}},
-    {"master_key_04", {S128KeyType::Master, 4, 0}},
-    {"package1_key_00", {S128KeyType::Package1, 0, 0}},
-    {"package1_key_01", {S128KeyType::Package1, 1, 0}},
-    {"package1_key_02", {S128KeyType::Package1, 2, 0}},
-    {"package1_key_03", {S128KeyType::Package1, 3, 0}},
-    {"package1_key_04", {S128KeyType::Package1, 4, 0}},
-    {"package2_key_00", {S128KeyType::Package2, 0, 0}},
-    {"package2_key_01", {S128KeyType::Package2, 1, 0}},
-    {"package2_key_02", {S128KeyType::Package2, 2, 0}},
-    {"package2_key_03", {S128KeyType::Package2, 3, 0}},
-    {"package2_key_04", {S128KeyType::Package2, 4, 0}},
-    {"titlekek_00", {S128KeyType::Titlekek, 0, 0}},
-    {"titlekek_01", {S128KeyType::Titlekek, 1, 0}},
-    {"titlekek_02", {S128KeyType::Titlekek, 2, 0}},
-    {"titlekek_03", {S128KeyType::Titlekek, 3, 0}},
-    {"titlekek_04", {S128KeyType::Titlekek, 4, 0}},
     {"eticket_rsa_kek", {S128KeyType::ETicketRSAKek, 0, 0}},
-    {"key_area_key_application_00",
-     {S128KeyType::KeyArea, 0, static_cast<u64>(KeyAreaKeyType::Application)}},
-    {"key_area_key_application_01",
-     {S128KeyType::KeyArea, 1, static_cast<u64>(KeyAreaKeyType::Application)}},
-    {"key_area_key_application_02",
-     {S128KeyType::KeyArea, 2, static_cast<u64>(KeyAreaKeyType::Application)}},
-    {"key_area_key_application_03",
-     {S128KeyType::KeyArea, 3, static_cast<u64>(KeyAreaKeyType::Application)}},
-    {"key_area_key_application_04",
-     {S128KeyType::KeyArea, 4, static_cast<u64>(KeyAreaKeyType::Application)}},
-    {"key_area_key_ocean_00", {S128KeyType::KeyArea, 0, static_cast<u64>(KeyAreaKeyType::Ocean)}},
-    {"key_area_key_ocean_01", {S128KeyType::KeyArea, 1, static_cast<u64>(KeyAreaKeyType::Ocean)}},
-    {"key_area_key_ocean_02", {S128KeyType::KeyArea, 2, static_cast<u64>(KeyAreaKeyType::Ocean)}},
-    {"key_area_key_ocean_03", {S128KeyType::KeyArea, 3, static_cast<u64>(KeyAreaKeyType::Ocean)}},
-    {"key_area_key_ocean_04", {S128KeyType::KeyArea, 4, static_cast<u64>(KeyAreaKeyType::Ocean)}},
-    {"key_area_key_system_00", {S128KeyType::KeyArea, 0, static_cast<u64>(KeyAreaKeyType::System)}},
-    {"key_area_key_system_01", {S128KeyType::KeyArea, 1, static_cast<u64>(KeyAreaKeyType::System)}},
-    {"key_area_key_system_02", {S128KeyType::KeyArea, 2, static_cast<u64>(KeyAreaKeyType::System)}},
-    {"key_area_key_system_03", {S128KeyType::KeyArea, 3, static_cast<u64>(KeyAreaKeyType::System)}},
-    {"key_area_key_system_04", {S128KeyType::KeyArea, 4, static_cast<u64>(KeyAreaKeyType::System)}},
-    {"sd_card_kek_source", {S128KeyType::Source, static_cast<u64>(SourceKeyType::SDKEK), 0}},
+    {"eticket_rsa_kek_source",
+     {S128KeyType::Source, static_cast<u64>(SourceKeyType::ETicketKek), 0}},
+    {"eticket_rsa_kekek_source",
+     {S128KeyType::Source, static_cast<u64>(SourceKeyType::ETicketKekek), 0}},
+    {"rsa_kek_mask_0", {S128KeyType::RSAKek, static_cast<u64>(RSAKekType::Mask0), 0}},
+    {"rsa_kek_seed_3", {S128KeyType::RSAKek, static_cast<u64>(RSAKekType::Seed3), 0}},
+    {"rsa_oaep_kek_generation_source",
+     {S128KeyType::Source, static_cast<u64>(SourceKeyType::RSAOaepKekGeneration), 0}},
+    {"sd_card_kek_source", {S128KeyType::Source, static_cast<u64>(SourceKeyType::SDKek), 0}},
     {"aes_kek_generation_source",
-     {S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKEKGeneration), 0}},
+     {S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKekGeneration), 0}},
     {"aes_key_generation_source",
      {S128KeyType::Source, static_cast<u64>(SourceKeyType::AESKeyGeneration), 0}},
+    {"package2_key_source", {S128KeyType::Source, static_cast<u64>(SourceKeyType::Package2), 0}},
+    {"master_key_source", {S128KeyType::Source, static_cast<u64>(SourceKeyType::Master), 0}},
+    {"header_kek_source", {S128KeyType::Source, static_cast<u64>(SourceKeyType::HeaderKek), 0}},
+    {"key_area_key_application_source",
+     {S128KeyType::Source, static_cast<u64>(SourceKeyType::KeyAreaKey),
+      static_cast<u64>(KeyAreaKeyType::Application)}},
+    {"key_area_key_ocean_source",
+     {S128KeyType::Source, static_cast<u64>(SourceKeyType::KeyAreaKey),
+      static_cast<u64>(KeyAreaKeyType::Ocean)}},
+    {"key_area_key_system_source",
+     {S128KeyType::Source, static_cast<u64>(SourceKeyType::KeyAreaKey),
+      static_cast<u64>(KeyAreaKeyType::System)}},
+    {"titlekek_source", {S128KeyType::Source, static_cast<u64>(SourceKeyType::Titlekek), 0}},
+    {"keyblob_mac_key_source", {S128KeyType::Source, static_cast<u64>(SourceKeyType::KeyblobMAC)}},
+    {"tsec_key", {S128KeyType::TSEC, 0, 0}},
+    {"secure_boot_key", {S128KeyType::SecureBoot, 0, 0}},
     {"sd_seed", {S128KeyType::SDSeed, 0, 0}},
+    {"bis_key_0_crypt", {S128KeyType::BIS, 0, static_cast<u64>(BISKeyType::Crypto)}},
+    {"bis_key_0_tweak", {S128KeyType::BIS, 0, static_cast<u64>(BISKeyType::Tweak)}},
+    {"bis_key_1_crypt", {S128KeyType::BIS, 1, static_cast<u64>(BISKeyType::Crypto)}},
+    {"bis_key_1_tweak", {S128KeyType::BIS, 1, static_cast<u64>(BISKeyType::Tweak)}},
+    {"bis_key_2_crypt", {S128KeyType::BIS, 2, static_cast<u64>(BISKeyType::Crypto)}},
+    {"bis_key_2_tweak", {S128KeyType::BIS, 2, static_cast<u64>(BISKeyType::Tweak)}},
+    {"bis_key_3_crypt", {S128KeyType::BIS, 3, static_cast<u64>(BISKeyType::Crypto)}},
+    {"bis_key_3_tweak", {S128KeyType::BIS, 3, static_cast<u64>(BISKeyType::Tweak)}},
+    {"header_kek", {S128KeyType::HeaderKek, 0, 0}},
+    {"sd_card_kek", {S128KeyType::SDKek, 0, 0}},
 };
 
 const boost::container::flat_map<std::string, KeyIndex<S256KeyType>> KeyManager::s256_file_id = {
     {"header_key", {S256KeyType::Header, 0, 0}},
     {"sd_card_save_key_source", {S256KeyType::SDKeySource, static_cast<u64>(SDKeyType::Save), 0}},
     {"sd_card_nca_key_source", {S256KeyType::SDKeySource, static_cast<u64>(SDKeyType::NCA), 0}},
+    {"header_key_source", {S256KeyType::HeaderSource, 0, 0}},
+    {"sd_card_save_key", {S256KeyType::SDKey, static_cast<u64>(SDKeyType::Save), 0}},
+    {"sd_card_nca_key", {S256KeyType::SDKey, static_cast<u64>(SDKeyType::NCA), 0}},
 };
 } // namespace Core::Crypto

src/core/crypto/key_manager.h

@@ -5,11 +5,18 @@
 #pragma once
 
 #include <array>
+#include <map>
 #include <string>
 #include <boost/container/flat_map.hpp>
 #include <boost/optional.hpp>
 #include <fmt/format.h>
 #include "common/common_types.h"
+#include "core/crypto/partition_data_manager.h"
+#include "core/file_sys/vfs_types.h"
+
+namespace FileUtil {
+class IOFile;
+}
 
 namespace Loader {
 enum class ResultStatus : u16;
@@ -22,13 +29,30 @@ constexpr u64 TICKET_FILE_TITLEKEY_OFFSET = 0x180;
 using Key128 = std::array<u8, 0x10>;
 using Key256 = std::array<u8, 0x20>;
 using SHA256Hash = std::array<u8, 0x20>;
+using TicketRaw = std::array<u8, 0x400>;
 
 static_assert(sizeof(Key128) == 16, "Key128 must be 128 bytes big.");
-static_assert(sizeof(Key256) == 32, "Key128 must be 128 bytes big.");
+static_assert(sizeof(Key256) == 32, "Key256 must be 256 bytes big.");
+
+template <size_t bit_size, size_t byte_size = (bit_size >> 3)>
+struct RSAKeyPair {
+    std::array<u8, byte_size> encryption_key;
+    std::array<u8, byte_size> decryption_key;
+    std::array<u8, byte_size> modulus;
+    std::array<u8, 4> exponent;
+};
+
+enum class KeyCategory : u8 {
+    Standard,
+    Title,
+    Console,
+};
 
 enum class S256KeyType : u64 {
-    Header,      //
-    SDKeySource, // f1=SDKeyType
+    SDKey,        // f1=SDKeyType
+    Header,       //
+    SDKeySource,  // f1=SDKeyType
+    HeaderSource, //
 };
 
 enum class S128KeyType : u64 {
@@ -41,6 +65,14 @@ enum class S128KeyType : u64 {
     SDSeed,        //
     Titlekey,      // f1=rights id LSB f2=rights id MSB
     Source,        // f1=source type, f2= sub id
+    Keyblob,       // f1=crypto revision
+    KeyblobMAC,    // f1=crypto revision
+    TSEC,          //
+    SecureBoot,    //
+    BIS,           // f1=partition (0-3), f2=type {crypt, tweak}
+    HeaderKek,     //
+    SDKek,         //
+    RSAKek,        //
 };
 
 enum class KeyAreaKeyType : u8 {
@@ -50,9 +82,19 @@ enum class KeyAreaKeyType : u8 {
 };
 
 enum class SourceKeyType : u8 {
-    SDKEK,
-    AESKEKGeneration,
-    AESKeyGeneration,
+    SDKek,                //
+    AESKekGeneration,     //
+    AESKeyGeneration,     //
+    RSAOaepKekGeneration, //
+    Master,               //
+    Keyblob,              // f2=crypto revision
+    KeyAreaKey,           // f2=KeyAreaKeyType
+    Titlekek,             //
+    Package2,             //
+    HeaderKek,            //
+    KeyblobMAC,           //
+    ETicketKek,           //
+    ETicketKekek,         //
 };
 
 enum class SDKeyType : u8 {
@@ -60,6 +102,16 @@ enum class SDKeyType : u8 {
     NCA,
 };
 
+enum class BISKeyType : u8 {
+    Crypto,
+    Tweak,
+};
+
+enum class RSAKekType : u8 {
+    Mask0,
+    Seed3,
+};
+
 template <typename KeyType>
 struct KeyIndex {
     KeyType type;
@@ -91,6 +143,8 @@ public:
     Key128 GetKey(S128KeyType id, u64 field1 = 0, u64 field2 = 0) const;
     Key256 GetKey(S256KeyType id, u64 field1 = 0, u64 field2 = 0) const;
 
+    Key256 GetBISKey(u8 partition_id) const;
+
     void SetKey(S128KeyType id, Key128 key, u64 field1 = 0, u64 field2 = 0);
     void SetKey(S256KeyType id, Key256 key, u64 field1 = 0, u64 field2 = 0);
 
@@ -100,23 +154,51 @@ public:
     // 8*43 and the private file to exist.
     void DeriveSDSeedLazy();
 
+    bool BaseDeriveNecessary() const;
+    void DeriveBase();
+    void DeriveETicket(PartitionDataManager& data);
+
+    void PopulateFromPartitionData(PartitionDataManager& data);
+
 private:
-    boost::container::flat_map<KeyIndex<S128KeyType>, Key128> s128_keys;
-    boost::container::flat_map<KeyIndex<S256KeyType>, Key256> s256_keys;
+    std::map<KeyIndex<S128KeyType>, Key128> s128_keys;
+    std::map<KeyIndex<S256KeyType>, Key256> s256_keys;
+
+    std::array<std::array<u8, 0xB0>, 0x20> encrypted_keyblobs{};
+    std::array<std::array<u8, 0x90>, 0x20> keyblobs{};
 
     bool dev_mode;
     void LoadFromFile(const std::string& filename, bool is_title_keys);
     void AttemptLoadKeyFile(const std::string& dir1, const std::string& dir2,
                             const std::string& filename, bool title);
-    template <std::size_t Size>
-    void WriteKeyToFile(bool title_key, std::string_view keyname, const std::array<u8, Size>& key);
+    template <size_t Size>
+    void WriteKeyToFile(KeyCategory category, std::string_view keyname,
+                        const std::array<u8, Size>& key);
+
+    void DeriveGeneralPurposeKeys(u8 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);
 
     static const boost::container::flat_map<std::string, KeyIndex<S128KeyType>> s128_file_id;
     static const boost::container::flat_map<std::string, KeyIndex<S256KeyType>> s256_file_id;
 };
 
 Key128 GenerateKeyEncryptionKey(Key128 source, Key128 master, Key128 kek_seed, Key128 key_seed);
+Key128 DeriveKeyblobKey(const Key128& sbk, const Key128& tsec, Key128 source);
+Key128 DeriveKeyblobMACKey(const Key128& keyblob_key, const Key128& mac_source);
+Key128 DeriveMasterKey(const std::array<u8, 0x90>& keyblob, const Key128& master_source);
+std::array<u8, 0x90> DecryptKeyblob(const std::array<u8, 0xB0>& encrypted_keyblob,
+                                    const Key128& key);
+
 boost::optional<Key128> DeriveSDSeed();
-Loader::ResultStatus DeriveSDKeys(std::array<Key256, 2>& sd_keys, const KeyManager& keys);
+Loader::ResultStatus DeriveSDKeys(std::array<Key256, 2>& sd_keys, KeyManager& keys);
+
+std::vector<TicketRaw> GetTicketblob(const FileUtil::IOFile& ticket_save);
+
+// Returns a pair of {rights_id, titlekey}. Fails if the ticket has no certificate authority (offset
+// 0x140-0x144 is zero)
+boost::optional<std::pair<Key128, Key128>> ParseTicket(
+    const TicketRaw& ticket, const RSAKeyPair<2048>& eticket_extended_key);
 
 } // namespace Core::Crypto

src/core/crypto/partition_data_manager.cpp

@@ -0,0 +1,601 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// NOTE TO FUTURE MAINTAINERS:
+// When a new version of switch cryptography is released,
+// hash the new keyblob source and master key and add the hashes to
+// the arrays below.
+
+#include <algorithm>
+#include <array>
+#include <cctype>
+#include <cstring>
+#include <boost/optional/optional.hpp>
+#include <mbedtls/sha256.h>
+#include "common/assert.h"
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "common/hex_util.h"
+#include "common/logging/log.h"
+#include "common/string_util.h"
+#include "core/crypto/ctr_encryption_layer.h"
+#include "core/crypto/key_manager.h"
+#include "core/crypto/partition_data_manager.h"
+#include "core/crypto/xts_encryption_layer.h"
+#include "core/file_sys/vfs.h"
+#include "core/file_sys/vfs_offset.h"
+
+using namespace Common;
+
+namespace Core::Crypto {
+
+struct Package2Header {
+    std::array<u8, 0x100> signature;
+    Key128 header_ctr;
+    std::array<Key128, 4> section_ctr;
+    u32_le magic;
+    u32_le base_offset;
+    INSERT_PADDING_BYTES(4);
+    u8 version_max;
+    u8 version_min;
+    INSERT_PADDING_BYTES(2);
+    std::array<u32_le, 4> section_size;
+    std::array<u32_le, 4> section_offset;
+    std::array<SHA256Hash, 4> section_hash;
+};
+static_assert(sizeof(Package2Header) == 0x200, "Package2Header has incorrect size.");
+
+struct INIHeader {
+    u32_le magic;
+    u32_le size;
+    u32_le process_count;
+    INSERT_PADDING_BYTES(4);
+};
+static_assert(sizeof(INIHeader) == 0x10, "INIHeader has incorrect size.");
+
+struct SectionHeader {
+    u32_le offset;
+    u32_le size_decompressed;
+    u32_le size_compressed;
+    u32_le attribute;
+};
+static_assert(sizeof(SectionHeader) == 0x10, "SectionHeader has incorrect size.");
+
+struct KIPHeader {
+    u32_le magic;
+    std::array<char, 12> name;
+    u64_le title_id;
+    u32_le category;
+    u8 priority;
+    u8 core;
+    INSERT_PADDING_BYTES(1);
+    u8 flags;
+    std::array<SectionHeader, 6> sections;
+    std::array<u32, 0x20> capabilities;
+};
+static_assert(sizeof(KIPHeader) == 0x100, "KIPHeader has incorrect size.");
+
+const std::array<SHA256Hash, 0x10> source_hashes{
+    "B24BD293259DBC7AC5D63F88E60C59792498E6FC5443402C7FFE87EE8B61A3F0"_array32, // keyblob_mac_key_source
+    "7944862A3A5C31C6720595EFD302245ABD1B54CCDCF33000557681E65C5664A4"_array32, // master_key_source
+    "21E2DF100FC9E094DB51B47B9B1D6E94ED379DB8B547955BEF8FE08D8DD35603"_array32, // package2_key_source
+    "FC02B9D37B42D7A1452E71444F1F700311D1132E301A83B16062E72A78175085"_array32, // aes_kek_generation_source
+    "FBD10056999EDC7ACDB96098E47E2C3606230270D23281E671F0F389FC5BC585"_array32, // aes_key_generation_source
+    "C48B619827986C7F4E3081D59DB2B460C84312650E9A8E6B458E53E8CBCA4E87"_array32, // titlekek_source
+    "04AD66143C726B2A139FB6B21128B46F56C553B2B3887110304298D8D0092D9E"_array32, // key_area_key_application_source
+    "FD434000C8FF2B26F8E9A9D2D2C12F6BE5773CBB9DC86300E1BD99F8EA33A417"_array32, // key_area_key_ocean_source
+    "1F17B1FD51AD1C2379B58F152CA4912EC2106441E51722F38700D5937A1162F7"_array32, // key_area_key_system_source
+    "6B2ED877C2C52334AC51E59ABFA7EC457F4A7D01E46291E9F2EAA45F011D24B7"_array32, // sd_card_kek_source
+    "D482743563D3EA5DCDC3B74E97C9AC8A342164FA041A1DC80F17F6D31E4BC01C"_array32, // sd_card_save_key_source
+    "2E751CECF7D93A2B957BD5FFCB082FD038CC2853219DD3092C6DAB9838F5A7CC"_array32, // sd_card_nca_key_source
+    "1888CAED5551B3EDE01499E87CE0D86827F80820EFB275921055AA4E2ABDFFC2"_array32, // header_kek_source
+    "8F783E46852DF6BE0BA4E19273C4ADBAEE16380043E1B8C418C4089A8BD64AA6"_array32, // header_key_source
+    "D1757E52F1AE55FA882EC690BC6F954AC46A83DC22F277F8806BD55577C6EED7"_array32, // rsa_kek_seed3
+    "FC02B9D37B42D7A1452E71444F1F700311D1132E301A83B16062E72A78175085"_array32, // rsa_kek_mask0
+};
+
+const std::array<SHA256Hash, 0x20> keyblob_source_hashes{
+    "8A06FE274AC491436791FDB388BCDD3AB9943BD4DEF8094418CDAC150FD73786"_array32, // keyblob_key_source_00
+    "2D5CAEB2521FEF70B47E17D6D0F11F8CE2C1E442A979AD8035832C4E9FBCCC4B"_array32, // keyblob_key_source_01
+    "61C5005E713BAE780641683AF43E5F5C0E03671117F702F401282847D2FC6064"_array32, // keyblob_key_source_02
+    "8E9795928E1C4428E1B78F0BE724D7294D6934689C11B190943923B9D5B85903"_array32, // keyblob_key_source_03
+    "95FA33AF95AFF9D9B61D164655B32710ED8D615D46C7D6CC3CC70481B686B402"_array32, // keyblob_key_source_04
+    "3F5BE7B3C8B1ABD8C10B4B703D44766BA08730562C172A4FE0D6B866B3E2DB3E"_array32, // keyblob_key_source_05
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_06
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_07
+
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_08
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_09
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_0A
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_0B
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_0C
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_0D
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_0E
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_0F
+
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_10
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_11
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_12
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_13
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_14
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_15
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_16
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_17
+
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_18
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_19
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_1A
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_1B
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_1C
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_1D
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_1E
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // keyblob_key_source_1F
+};
+
+const std::array<SHA256Hash, 0x20> master_key_hashes{
+    "0EE359BE3C864BB0782E1D70A718A0342C551EED28C369754F9C4F691BECF7CA"_array32, // master_key_00
+    "4FE707B7E4ABDAF727C894AAF13B1351BFE2AC90D875F73B2E20FA94B9CC661E"_array32, // master_key_01
+    "79277C0237A2252EC3DFAC1F7C359C2B3D121E9DB15BB9AB4C2B4408D2F3AE09"_array32, // master_key_02
+    "4F36C565D13325F65EE134073C6A578FFCB0008E02D69400836844EAB7432754"_array32, // master_key_03
+    "75FF1D95D26113550EE6FCC20ACB58E97EDEB3A2FF52543ED5AEC63BDCC3DA50"_array32, // master_key_04
+    "EBE2BCD6704673EC0F88A187BB2AD9F1CC82B718C389425941BDC194DC46B0DD"_array32, // master_key_05
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_06
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_07
+
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_08
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_09
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_0A
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_0B
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_0C
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_0D
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_0E
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_0F
+
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_10
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_11
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_12
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_13
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_14
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_15
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_16
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_17
+
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_18
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_19
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_1A
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_1B
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_1C
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_1D
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_1E
+    "0000000000000000000000000000000000000000000000000000000000000000"_array32, // master_key_1F
+};
+
+static std::vector<u8> DecompressBLZ(const std::vector<u8>& in) {
+    const auto data_size = in.size() - 0xC;
+
+    u32 compressed_size{};
+    u32 init_index{};
+    u32 additional_size{};
+    std::memcpy(&compressed_size, in.data() + data_size, sizeof(u32));
+    std::memcpy(&init_index, in.data() + data_size + 0x4, sizeof(u32));
+    std::memcpy(&additional_size, in.data() + data_size + 0x8, sizeof(u32));
+
+    std::vector<u8> out(in.size() + additional_size);
+
+    if (compressed_size == in.size())
+        std::memcpy(out.data(), in.data() + in.size() - compressed_size, compressed_size);
+    else
+        std::memcpy(out.data(), in.data(), compressed_size);
+
+    auto index = in.size() - init_index;
+    auto out_index = out.size();
+
+    while (out_index > 0) {
+        --index;
+        auto control = in[index];
+        for (size_t i = 0; i < 8; ++i) {
+            if ((control & 0x80) > 0) {
+                ASSERT(index >= 2);
+                index -= 2;
+                u64 segment_offset = in[index] | in[index + 1] << 8;
+                u64 segment_size = ((segment_offset >> 12) & 0xF) + 3;
+                segment_offset &= 0xFFF;
+                segment_offset += 3;
+
+                if (out_index < segment_size)
+                    segment_size = out_index;
+
+                ASSERT(out_index >= segment_size);
+
+                out_index -= segment_size;
+
+                for (size_t j = 0; j < segment_size; ++j) {
+                    ASSERT(out_index + j + segment_offset < out.size());
+                    out[out_index + j] = out[out_index + j + segment_offset];
+                }
+            } else {
+                ASSERT(out_index >= 1);
+                --out_index;
+                --index;
+                out[out_index] = in[index];
+            }
+
+            control <<= 1;
+            if (out_index == 0)
+                return out;
+        }
+    }
+
+    return out;
+}
+
+static u8 CalculateMaxKeyblobSourceHash() {
+    for (s8 i = 0x1F; i >= 0; --i) {
+        if (keyblob_source_hashes[i] != SHA256Hash{})
+            return static_cast<u8>(i + 1);
+    }
+
+    return 0;
+}
+
+const u8 PartitionDataManager::MAX_KEYBLOB_SOURCE_HASH = CalculateMaxKeyblobSourceHash();
+
+template <size_t key_size = 0x10>
+std::array<u8, key_size> FindKeyFromHex(const std::vector<u8>& binary,
+                                        const std::array<u8, 0x20>& hash) {
+    if (binary.size() < key_size)
+        return {};
+
+    std::array<u8, 0x20> temp{};
+    for (size_t i = 0; i < binary.size() - key_size; ++i) {
+        mbedtls_sha256(binary.data() + i, key_size, temp.data(), 0);
+
+        if (temp != hash)
+            continue;
+
+        std::array<u8, key_size> out{};
+        std::memcpy(out.data(), binary.data() + i, key_size);
+        return out;
+    }
+
+    return {};
+}
+
+std::array<u8, 16> FindKeyFromHex16(const std::vector<u8>& binary, std::array<u8, 32> hash) {
+    return FindKeyFromHex<0x10>(binary, hash);
+}
+
+static std::array<Key128, 0x20> FindEncryptedMasterKeyFromHex(const std::vector<u8>& binary,
+                                                              const Key128& key) {
+    if (binary.size() < 0x10)
+        return {};
+
+    SHA256Hash temp{};
+    Key128 dec_temp{};
+    std::array<Key128, 0x20> out{};
+    AESCipher<Key128> cipher(key, Mode::ECB);
+    for (size_t i = 0; i < binary.size() - 0x10; ++i) {
+        cipher.Transcode(binary.data() + i, dec_temp.size(), dec_temp.data(), Op::Decrypt);
+        mbedtls_sha256(dec_temp.data(), dec_temp.size(), temp.data(), 0);
+
+        for (size_t k = 0; k < out.size(); ++k) {
+            if (temp == master_key_hashes[k]) {
+                out[k] = dec_temp;
+                break;
+            }
+        }
+    }
+
+    return out;
+}
+
+FileSys::VirtualFile FindFileInDirWithNames(const FileSys::VirtualDir& dir,
+                                            const std::string& name) {
+    auto upper = name;
+    std::transform(upper.begin(), upper.end(), upper.begin(), [](u8 c) { return std::toupper(c); });
+    for (const auto& fname : {name, name + ".bin", upper, upper + ".BIN"}) {
+        if (dir->GetFile(fname) != nullptr)
+            return dir->GetFile(fname);
+    }
+
+    return nullptr;
+}
+
+PartitionDataManager::PartitionDataManager(FileSys::VirtualDir sysdata_dir)
+    : boot0(FindFileInDirWithNames(sysdata_dir, "BOOT0")),
+      fuses(FindFileInDirWithNames(sysdata_dir, "fuse")),
+      kfuses(FindFileInDirWithNames(sysdata_dir, "kfuse")),
+      package2({
+          FindFileInDirWithNames(sysdata_dir, "BCPKG2-1-Normal-Main"),
+          FindFileInDirWithNames(sysdata_dir, "BCPKG2-2-Normal-Sub"),
+          FindFileInDirWithNames(sysdata_dir, "BCPKG2-3-SafeMode-Main"),
+          FindFileInDirWithNames(sysdata_dir, "BCPKG2-4-SafeMode-Sub"),
+          FindFileInDirWithNames(sysdata_dir, "BCPKG2-5-Repair-Main"),
+          FindFileInDirWithNames(sysdata_dir, "BCPKG2-6-Repair-Sub"),
+      }),
+      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")) {}
+
+PartitionDataManager::~PartitionDataManager() = default;
+
+bool PartitionDataManager::HasBoot0() const {
+    return boot0 != nullptr;
+}
+
+FileSys::VirtualFile PartitionDataManager::GetBoot0Raw() const {
+    return boot0;
+}
+
+std::array<u8, 176> PartitionDataManager::GetEncryptedKeyblob(u8 index) const {
+    if (HasBoot0() && index < 32)
+        return GetEncryptedKeyblobs()[index];
+    return {};
+}
+
+std::array<std::array<u8, 176>, 32> PartitionDataManager::GetEncryptedKeyblobs() const {
+    if (!HasBoot0())
+        return {};
+
+    std::array<std::array<u8, 176>, 32> out{};
+    for (size_t i = 0; i < 0x20; ++i)
+        boot0->Read(out[i].data(), out[i].size(), 0x180000 + i * 0x200);
+    return out;
+}
+
+std::vector<u8> PartitionDataManager::GetSecureMonitor() const {
+    return secure_monitor_bytes;
+}
+
+std::array<u8, 16> PartitionDataManager::GetPackage2KeySource() const {
+    return FindKeyFromHex(secure_monitor_bytes, source_hashes[2]);
+}
+
+std::array<u8, 16> PartitionDataManager::GetAESKekGenerationSource() const {
+    return FindKeyFromHex(secure_monitor_bytes, source_hashes[3]);
+}
+
+std::array<u8, 16> PartitionDataManager::GetTitlekekSource() const {
+    return FindKeyFromHex(secure_monitor_bytes, source_hashes[5]);
+}
+
+std::array<std::array<u8, 16>, 32> PartitionDataManager::GetTZMasterKeys(
+    std::array<u8, 0x10> master_key) const {
+    return FindEncryptedMasterKeyFromHex(secure_monitor_bytes, master_key);
+}
+
+std::array<u8, 16> PartitionDataManager::GetRSAKekSeed3() const {
+    return FindKeyFromHex(secure_monitor_bytes, source_hashes[14]);
+}
+
+std::array<u8, 16> PartitionDataManager::GetRSAKekMask0() const {
+    return FindKeyFromHex(secure_monitor_bytes, source_hashes[15]);
+}
+
+std::vector<u8> PartitionDataManager::GetPackage1Decrypted() const {
+    return package1_decrypted_bytes;
+}
+
+std::array<u8, 16> PartitionDataManager::GetMasterKeySource() const {
+    return FindKeyFromHex(package1_decrypted_bytes, source_hashes[1]);
+}
+
+std::array<u8, 16> PartitionDataManager::GetKeyblobMACKeySource() const {
+    return FindKeyFromHex(package1_decrypted_bytes, source_hashes[0]);
+}
+
+std::array<u8, 16> PartitionDataManager::GetKeyblobKeySource(u8 revision) const {
+    if (keyblob_source_hashes[revision] == SHA256Hash{}) {
+        LOG_WARNING(Crypto,
+                    "No keyblob source hash for crypto revision {:02X}! Cannot derive keys...",
+                    revision);
+    }
+    return FindKeyFromHex(package1_decrypted_bytes, keyblob_source_hashes[revision]);
+}
+
+bool PartitionDataManager::HasFuses() const {
+    return fuses != nullptr;
+}
+
+FileSys::VirtualFile PartitionDataManager::GetFusesRaw() const {
+    return fuses;
+}
+
+std::array<u8, 16> PartitionDataManager::GetSecureBootKey() const {
+    if (!HasFuses())
+        return {};
+    Key128 out{};
+    fuses->Read(out.data(), out.size(), 0xA4);
+    return out;
+}
+
+bool PartitionDataManager::HasKFuses() const {
+    return kfuses != nullptr;
+}
+
+FileSys::VirtualFile PartitionDataManager::GetKFusesRaw() const {
+    return kfuses;
+}
+
+bool PartitionDataManager::HasPackage2(Package2Type type) const {
+    return package2.at(static_cast<size_t>(type)) != nullptr;
+}
+
+FileSys::VirtualFile PartitionDataManager::GetPackage2Raw(Package2Type type) const {
+    return package2.at(static_cast<size_t>(type));
+}
+
+bool AttemptDecrypt(const std::array<u8, 16>& key, Package2Header& header) {
+
+    const std::vector<u8> iv(header.header_ctr.begin(), header.header_ctr.end());
+    Package2Header temp = header;
+    AESCipher<Key128> cipher(key, Mode::CTR);
+    cipher.SetIV(iv);
+    cipher.Transcode(&temp.header_ctr, sizeof(Package2Header) - 0x100, &temp.header_ctr,
+                     Op::Decrypt);
+    if (temp.magic == Common::MakeMagic('P', 'K', '2', '1')) {
+        header = temp;
+        return true;
+    }
+
+    return false;
+}
+
+void PartitionDataManager::DecryptPackage2(std::array<std::array<u8, 16>, 0x20> package2_keys,
+                                           Package2Type type) {
+    FileSys::VirtualFile file = std::make_shared<FileSys::OffsetVfsFile>(
+        package2[static_cast<size_t>(type)],
+        package2[static_cast<size_t>(type)]->GetSize() - 0x4000, 0x4000);
+
+    Package2Header header{};
+    if (file->ReadObject(&header) != sizeof(Package2Header))
+        return;
+
+    u8 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))
+                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.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());
+
+        if (name != "FS" && name != "spl") {
+            offset += sizeof(KIPHeader) + kip.sections[0].size_compressed +
+                      kip.sections[1].size_compressed + kip.sections[2].size_compressed;
+            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);
+
+        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());
+
+        offset += sizeof(KIPHeader) + kip.sections[0].size_compressed +
+                  kip.sections[1].size_compressed + kip.sections[2].size_compressed;
+
+        text = DecompressBLZ(text);
+        rodata = DecompressBLZ(rodata);
+        data = DecompressBLZ(data);
+
+        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());
+
+        if (name == "FS")
+            package2_fs[static_cast<size_t>(type)] = out;
+        else if (name == "spl")
+            package2_spl[static_cast<size_t>(type)] = out;
+    }
+}
+
+const std::vector<u8>& PartitionDataManager::GetPackage2FSDecompressed(Package2Type type) const {
+    return package2_fs.at(static_cast<size_t>(type));
+}
+
+std::array<u8, 16> PartitionDataManager::GetKeyAreaKeyApplicationSource(Package2Type type) const {
+    return FindKeyFromHex(package2_fs.at(static_cast<size_t>(type)), source_hashes[6]);
+}
+
+std::array<u8, 16> PartitionDataManager::GetKeyAreaKeyOceanSource(Package2Type type) const {
+    return FindKeyFromHex(package2_fs.at(static_cast<size_t>(type)), source_hashes[7]);
+}
+
+std::array<u8, 16> PartitionDataManager::GetKeyAreaKeySystemSource(Package2Type type) const {
+    return FindKeyFromHex(package2_fs.at(static_cast<size_t>(type)), source_hashes[8]);
+}
+
+std::array<u8, 16> PartitionDataManager::GetSDKekSource(Package2Type type) const {
+    return FindKeyFromHex(package2_fs.at(static_cast<size_t>(type)), source_hashes[9]);
+}
+
+std::array<u8, 32> PartitionDataManager::GetSDSaveKeySource(Package2Type type) const {
+    return FindKeyFromHex<0x20>(package2_fs.at(static_cast<size_t>(type)), source_hashes[10]);
+}
+
+std::array<u8, 32> PartitionDataManager::GetSDNCAKeySource(Package2Type type) const {
+    return FindKeyFromHex<0x20>(package2_fs.at(static_cast<size_t>(type)), source_hashes[11]);
+}
+
+std::array<u8, 16> PartitionDataManager::GetHeaderKekSource(Package2Type type) const {
+    return FindKeyFromHex(package2_fs.at(static_cast<size_t>(type)), source_hashes[12]);
+}
+
+std::array<u8, 32> PartitionDataManager::GetHeaderKeySource(Package2Type type) const {
+    return FindKeyFromHex<0x20>(package2_fs.at(static_cast<size_t>(type)), source_hashes[13]);
+}
+
+const std::vector<u8>& PartitionDataManager::GetPackage2SPLDecompressed(Package2Type type) const {
+    return package2_spl.at(static_cast<size_t>(type));
+}
+
+std::array<u8, 16> PartitionDataManager::GetAESKeyGenerationSource(Package2Type type) const {
+    return FindKeyFromHex(package2_spl.at(static_cast<size_t>(type)), source_hashes[4]);
+}
+
+bool PartitionDataManager::HasProdInfo() const {
+    return prodinfo != nullptr;
+}
+
+FileSys::VirtualFile PartitionDataManager::GetProdInfoRaw() const {
+    return prodinfo;
+}
+
+void PartitionDataManager::DecryptProdInfo(std::array<u8, 0x20> bis_key) {
+    if (prodinfo == nullptr)
+        return;
+
+    prodinfo_decrypted = std::make_shared<XTSEncryptionLayer>(prodinfo, bis_key);
+}
+
+std::array<u8, 576> PartitionDataManager::GetETicketExtendedKek() const {
+    std::array<u8, 0x240> out{};
+    if (prodinfo_decrypted != nullptr)
+        prodinfo_decrypted->Read(out.data(), out.size(), 0x3890);
+    return out;
+}
+} // namespace Core::Crypto

src/core/crypto/partition_data_manager.h

@@ -0,0 +1,105 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#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 {
+
+enum class Package2Type {
+    NormalMain,
+    NormalSub,
+    SafeModeMain,
+    SafeModeSub,
+    RepairMain,
+    RepairSub,
+};
+
+class PartitionDataManager {
+public:
+    static const u8 MAX_KEYBLOB_SOURCE_HASH;
+
+    explicit PartitionDataManager(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;
+    std::vector<u8> GetSecureMonitor() const;
+    std::array<u8, 0x10> GetPackage2KeySource() const;
+    std::array<u8, 0x10> GetAESKekGenerationSource() const;
+    std::array<u8, 0x10> GetTitlekekSource() const;
+    std::array<std::array<u8, 0x10>, 0x20> GetTZMasterKeys(std::array<u8, 0x10> master_key) const;
+    std::array<u8, 0x10> GetRSAKekSeed3() const;
+    std::array<u8, 0x10> GetRSAKekMask0() const;
+    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;
+
+    // Fuses
+    bool HasFuses() const;
+    FileSys::VirtualFile GetFusesRaw() const;
+    std::array<u8, 0x10> GetSecureBootKey() const;
+
+    // K-Fuses
+    bool HasKFuses() const;
+    FileSys::VirtualFile GetKFusesRaw() const;
+
+    // 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);
+    const std::vector<u8>& GetPackage2FSDecompressed(
+        Package2Type type = Package2Type::NormalMain) const;
+    std::array<u8, 0x10> GetKeyAreaKeyApplicationSource(
+        Package2Type type = Package2Type::NormalMain) const;
+    std::array<u8, 0x10> GetKeyAreaKeyOceanSource(
+        Package2Type type = Package2Type::NormalMain) const;
+    std::array<u8, 0x10> GetKeyAreaKeySystemSource(
+        Package2Type type = Package2Type::NormalMain) const;
+    std::array<u8, 0x10> GetSDKekSource(Package2Type type = Package2Type::NormalMain) const;
+    std::array<u8, 0x20> GetSDSaveKeySource(Package2Type type = Package2Type::NormalMain) const;
+    std::array<u8, 0x20> GetSDNCAKeySource(Package2Type type = Package2Type::NormalMain) const;
+    std::array<u8, 0x10> GetHeaderKekSource(Package2Type type = Package2Type::NormalMain) const;
+    std::array<u8, 0x20> GetHeaderKeySource(Package2Type type = Package2Type::NormalMain) const;
+    const std::vector<u8>& GetPackage2SPLDecompressed(
+        Package2Type type = Package2Type::NormalMain) const;
+    std::array<u8, 0x10> GetAESKeyGenerationSource(
+        Package2Type type = Package2Type::NormalMain) const;
+
+    // PRODINFO
+    bool HasProdInfo() const;
+    FileSys::VirtualFile GetProdInfoRaw() const;
+    void DecryptProdInfo(std::array<u8, 0x20> bis_key);
+    std::array<u8, 0x240> GetETicketExtendedKek() const;
+
+private:
+    FileSys::VirtualFile boot0;
+    FileSys::VirtualFile fuses;
+    FileSys::VirtualFile kfuses;
+    std::array<FileSys::VirtualFile, 6> package2;
+    FileSys::VirtualFile prodinfo;
+    FileSys::VirtualFile secure_monitor;
+    FileSys::VirtualFile package1_decrypted;
+
+    // Processed
+    std::array<FileSys::VirtualFile, 6> package2_decrypted;
+    FileSys::VirtualFile prodinfo_decrypted;
+    std::vector<u8> secure_monitor_bytes;
+    std::vector<u8> package1_decrypted_bytes;
+    std::array<std::vector<u8>, 6> package2_fs;
+    std::array<std::vector<u8>, 6> package2_spl;
+};
+
+std::array<u8, 0x10> FindKeyFromHex16(const std::vector<u8>& binary, std::array<u8, 0x20> hash);
+
+} // namespace Core::Crypto

src/core/file_sys/fsmitm_romfsbuild.cpp

@@ -26,6 +26,7 @@
 #include "common/alignment.h"
 #include "common/assert.h"
 #include "core/file_sys/fsmitm_romfsbuild.h"
+#include "core/file_sys/ips_layer.h"
 #include "core/file_sys/vfs.h"
 #include "core/file_sys/vfs_vector.h"
 
@@ -123,7 +124,7 @@ static u64 romfs_get_hash_table_count(u64 num_entries) {
     return count;
 }
 
-void RomFSBuildContext::VisitDirectory(VirtualDir root_romfs,
+void RomFSBuildContext::VisitDirectory(VirtualDir root_romfs, VirtualDir ext,
                                        std::shared_ptr<RomFSBuildDirectoryContext> parent) {
     std::vector<std::shared_ptr<RomFSBuildDirectoryContext>> child_dirs;
 
@@ -144,6 +145,9 @@ void RomFSBuildContext::VisitDirectory(VirtualDir root_romfs,
             child->path_len = child->cur_path_ofs + static_cast<u32>(kv.first.size());
             child->path = parent->path + "/" + kv.first;
 
+            if (ext != nullptr && ext->GetFileRelative(child->path + ".stub") != nullptr)
+                continue;
+
             // Sanity check on path_len
             ASSERT(child->path_len < FS_MAX_PATH);
 
@@ -157,11 +161,24 @@ void RomFSBuildContext::VisitDirectory(VirtualDir root_romfs,
             child->path_len = child->cur_path_ofs + static_cast<u32>(kv.first.size());
             child->path = parent->path + "/" + kv.first;
 
+            if (ext != nullptr && ext->GetFileRelative(child->path + ".stub") != nullptr)
+                continue;
+
             // Sanity check on path_len
             ASSERT(child->path_len < FS_MAX_PATH);
 
             child->source = root_romfs->GetFileRelative(child->path);
 
+            if (ext != nullptr) {
+                const auto ips = ext->GetFileRelative(child->path + ".ips");
+
+                if (ips != nullptr) {
+                    auto patched = PatchIPS(child->source, ips);
+                    if (patched != nullptr)
+                        child->source = std::move(patched);
+                }
+            }
+
             child->size = child->source->GetSize();
 
             AddFile(parent, child);
@@ -169,7 +186,7 @@ void RomFSBuildContext::VisitDirectory(VirtualDir root_romfs,
     }
 
     for (auto& child : child_dirs) {
-        this->VisitDirectory(root_romfs, child);
+        this->VisitDirectory(root_romfs, ext, child);
     }
 }
 
@@ -208,14 +225,15 @@ bool RomFSBuildContext::AddFile(std::shared_ptr<RomFSBuildDirectoryContext> pare
     return true;
 }
 
-RomFSBuildContext::RomFSBuildContext(VirtualDir base_) : base(std::move(base_)) {
+RomFSBuildContext::RomFSBuildContext(VirtualDir base_, VirtualDir ext_)
+    : base(std::move(base_)), ext(std::move(ext_)) {
     root = std::make_shared<RomFSBuildDirectoryContext>();
     root->path = "\0";
     directories.emplace(root->path, root);
     num_dirs = 1;
     dir_table_size = 0x18;
 
-    VisitDirectory(base, root);
+    VisitDirectory(base, ext, root);
 }
 
 RomFSBuildContext::~RomFSBuildContext() = default;

src/core/file_sys/fsmitm_romfsbuild.h

@@ -40,7 +40,7 @@ struct RomFSFileEntry;
 
 class RomFSBuildContext {
 public:
-    explicit RomFSBuildContext(VirtualDir base);
+    explicit RomFSBuildContext(VirtualDir base, VirtualDir ext = nullptr);
     ~RomFSBuildContext();
 
     // This finalizes the context.
@@ -48,6 +48,7 @@ public:
 
 private:
     VirtualDir base;
+    VirtualDir ext;
     std::shared_ptr<RomFSBuildDirectoryContext> root;
     std::map<std::string, std::shared_ptr<RomFSBuildDirectoryContext>, std::less<>> directories;
     std::map<std::string, std::shared_ptr<RomFSBuildFileContext>, std::less<>> files;
@@ -59,7 +60,8 @@ private:
     u64 file_hash_table_size = 0;
     u64 file_partition_size = 0;
 
-    void VisitDirectory(VirtualDir filesys, std::shared_ptr<RomFSBuildDirectoryContext> parent);
+    void VisitDirectory(VirtualDir filesys, VirtualDir ext,
+                        std::shared_ptr<RomFSBuildDirectoryContext> parent);
 
     bool AddDirectory(std::shared_ptr<RomFSBuildDirectoryContext> parent_dir_ctx,
                       std::shared_ptr<RomFSBuildDirectoryContext> dir_ctx);

src/core/file_sys/ips_layer.cpp

@@ -2,7 +2,15 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "common/assert.h"
+#include <algorithm>
+#include <cstring>
+#include <map>
+#include <sstream>
+#include <string>
+#include <utility>
+
+#include "common/hex_util.h"
+#include "common/logging/log.h"
 #include "common/swap.h"
 #include "core/file_sys/ips_layer.h"
 #include "core/file_sys/vfs_vector.h"
@@ -15,16 +23,48 @@ enum class IPSFileType {
     Error,
 };
 
+constexpr std::array<std::pair<const char*, const char*>, 11> ESCAPE_CHARACTER_MAP{{
+    {"\\a", "\a"},
+    {"\\b", "\b"},
+    {"\\f", "\f"},
+    {"\\n", "\n"},
+    {"\\r", "\r"},
+    {"\\t", "\t"},
+    {"\\v", "\v"},
+    {"\\\\", "\\"},
+    {"\\\'", "\'"},
+    {"\\\"", "\""},
+    {"\\\?", "\?"},
+}};
+
 static IPSFileType IdentifyMagic(const std::vector<u8>& magic) {
-    if (magic.size() != 5)
+    if (magic.size() != 5) {
         return IPSFileType::Error;
-    if (magic == std::vector<u8>{'P', 'A', 'T', 'C', 'H'})
+    }
+
+    constexpr std::array<u8, 5> patch_magic{{'P', 'A', 'T', 'C', 'H'}};
+    if (std::equal(magic.begin(), magic.end(), patch_magic.begin())) {
         return IPSFileType::IPS;
-    if (magic == std::vector<u8>{'I', 'P', 'S', '3', '2'})
+    }
+
+    constexpr std::array<u8, 5> ips32_magic{{'I', 'P', 'S', '3', '2'}};
+    if (std::equal(magic.begin(), magic.end(), ips32_magic.begin())) {
         return IPSFileType::IPS32;
+    }
+
     return IPSFileType::Error;
 }
 
+static bool IsEOF(IPSFileType type, const std::vector<u8>& data) {
+    constexpr std::array<u8, 3> eof{{'E', 'O', 'F'}};
+    if (type == IPSFileType::IPS && std::equal(data.begin(), data.end(), eof.begin())) {
+        return true;
+    }
+
+    constexpr std::array<u8, 4> eeof{{'E', 'E', 'O', 'F'}};
+    return type == IPSFileType::IPS32 && std::equal(data.begin(), data.end(), eeof.begin());
+}
+
 VirtualFile PatchIPS(const VirtualFile& in, const VirtualFile& ips) {
     if (in == nullptr || ips == nullptr)
         return nullptr;
@@ -39,8 +79,7 @@ VirtualFile PatchIPS(const VirtualFile& in, const VirtualFile& ips) {
     u64 offset = 5; // After header
     while (ips->Read(temp.data(), temp.size(), offset) == temp.size()) {
         offset += temp.size();
-        if (type == IPSFileType::IPS32 && temp == std::vector<u8>{'E', 'E', 'O', 'F'} ||
-            type == IPSFileType::IPS && temp == std::vector<u8>{'E', 'O', 'F'}) {
+        if (IsEOF(type, temp)) {
             break;
         }
 
@@ -68,21 +107,232 @@ VirtualFile PatchIPS(const VirtualFile& in, const VirtualFile& ips) {
                 return nullptr;
 
             if (real_offset + rle_size > in_data.size())
-                rle_size = in_data.size() - real_offset;
+                rle_size = static_cast<u16>(in_data.size() - real_offset);
             std::memset(in_data.data() + real_offset, data.get(), rle_size);
         } else { // Standard Patch
             auto read = data_size;
             if (real_offset + read > in_data.size())
-                read = in_data.size() - real_offset;
+                read = static_cast<u16>(in_data.size() - real_offset);
             if (ips->Read(in_data.data() + real_offset, read, offset) != data_size)
                 return nullptr;
             offset += data_size;
         }
     }
 
-    if (temp != std::vector<u8>{'E', 'E', 'O', 'F'} && temp != std::vector<u8>{'E', 'O', 'F'})
+    if (!IsEOF(type, temp)) {
         return nullptr;
-    return std::make_shared<VectorVfsFile>(in_data, in->GetName(), in->GetContainingDirectory());
+    }
+
+    return std::make_shared<VectorVfsFile>(std::move(in_data), in->GetName(),
+                                           in->GetContainingDirectory());
+}
+
+struct IPSwitchCompiler::IPSwitchPatch {
+    std::string name;
+    bool enabled;
+    std::map<u32, std::vector<u8>> records;
+};
+
+IPSwitchCompiler::IPSwitchCompiler(VirtualFile patch_text_) : patch_text(std::move(patch_text_)) {
+    Parse();
+}
+
+IPSwitchCompiler::~IPSwitchCompiler() = default;
+
+std::array<u8, 32> IPSwitchCompiler::GetBuildID() const {
+    return nso_build_id;
+}
+
+bool IPSwitchCompiler::IsValid() const {
+    return valid;
+}
+
+static bool StartsWith(std::string_view base, std::string_view check) {
+    return base.size() >= check.size() && base.substr(0, check.size()) == check;
+}
+
+static std::string EscapeStringSequences(std::string in) {
+    for (const auto& seq : ESCAPE_CHARACTER_MAP) {
+        for (auto index = in.find(seq.first); index != std::string::npos;
+             index = in.find(seq.first, index)) {
+            in.replace(index, std::strlen(seq.first), seq.second);
+            index += std::strlen(seq.second);
+        }
+    }
+
+    return in;
+}
+
+void IPSwitchCompiler::ParseFlag(const std::string& line) {
+    if (StartsWith(line, "@flag offset_shift ")) {
+        // Offset Shift Flag
+        offset_shift = std::stoll(line.substr(19), nullptr, 0);
+    } else if (StartsWith(line, "@little-endian")) {
+        // Set values to read as little endian
+        is_little_endian = true;
+    } else if (StartsWith(line, "@big-endian")) {
+        // Set values to read as big endian
+        is_little_endian = false;
+    } else if (StartsWith(line, "@flag print_values")) {
+        // Force printing of applied values
+        print_values = true;
+    }
+}
+
+void IPSwitchCompiler::Parse() {
+    const auto bytes = patch_text->ReadAllBytes();
+    std::stringstream s;
+    s.write(reinterpret_cast<const char*>(bytes.data()), bytes.size());
+
+    std::vector<std::string> lines;
+    std::string stream_line;
+    while (std::getline(s, stream_line)) {
+        // Remove a trailing \r
+        if (!stream_line.empty() && stream_line.back() == '\r')
+            stream_line.pop_back();
+        lines.push_back(std::move(stream_line));
+    }
+
+    for (std::size_t i = 0; i < lines.size(); ++i) {
+        auto line = lines[i];
+
+        // Remove midline comments
+        std::size_t comment_index = std::string::npos;
+        bool within_string = false;
+        for (std::size_t k = 0; k < line.size(); ++k) {
+            if (line[k] == '\"' && (k > 0 && line[k - 1] != '\\')) {
+                within_string = !within_string;
+            } else if (line[k] == '\\' && (k < line.size() - 1 && line[k + 1] == '\\')) {
+                comment_index = k;
+                break;
+            }
+        }
+
+        if (!StartsWith(line, "//") && comment_index != std::string::npos) {
+            last_comment = line.substr(comment_index + 2);
+            line = line.substr(0, comment_index);
+        }
+
+        if (StartsWith(line, "@stop")) {
+            // Force stop
+            break;
+        } else if (StartsWith(line, "@nsobid-")) {
+            // NSO Build ID Specifier
+            auto raw_build_id = line.substr(8);
+            if (raw_build_id.size() != 0x40)
+                raw_build_id.resize(0x40, '0');
+            nso_build_id = Common::HexStringToArray<0x20>(raw_build_id);
+        } else if (StartsWith(line, "#")) {
+            // Mandatory Comment
+            LOG_INFO(Loader, "[IPSwitchCompiler ('{}')] Forced output comment: {}",
+                     patch_text->GetName(), line.substr(1));
+        } else if (StartsWith(line, "//")) {
+            // Normal Comment
+            last_comment = line.substr(2);
+            if (last_comment.find_first_not_of(' ') == std::string::npos)
+                continue;
+            if (last_comment.find_first_not_of(' ') != 0)
+                last_comment = last_comment.substr(last_comment.find_first_not_of(' '));
+        } else if (StartsWith(line, "@enabled") || StartsWith(line, "@disabled")) {
+            // Start of patch
+            const auto enabled = StartsWith(line, "@enabled");
+            if (i == 0)
+                return;
+            LOG_INFO(Loader, "[IPSwitchCompiler ('{}')] Parsing patch '{}' ({})",
+                     patch_text->GetName(), last_comment, line.substr(1));
+
+            IPSwitchPatch patch{last_comment, enabled, {}};
+
+            // Read rest of patch
+            while (true) {
+                if (i + 1 >= lines.size())
+                    break;
+                const auto patch_line = lines[++i];
+
+                // Start of new patch
+                if (StartsWith(patch_line, "@enabled") || StartsWith(patch_line, "@disabled")) {
+                    --i;
+                    break;
+                }
+
+                // Check for a flag
+                if (StartsWith(patch_line, "@")) {
+                    ParseFlag(patch_line);
+                    continue;
+                }
+
+                // 11 - 8 hex digit offset + space + minimum two digit overwrite val
+                if (patch_line.length() < 11)
+                    break;
+                auto offset = std::stoul(patch_line.substr(0, 8), nullptr, 16);
+                offset += static_cast<unsigned long>(offset_shift);
+
+                std::vector<u8> replace;
+                // 9 - first char of replacement val
+                if (patch_line[9] == '\"') {
+                    // string replacement
+                    auto end_index = patch_line.find('\"', 10);
+                    if (end_index == std::string::npos || end_index < 10)
+                        return;
+                    while (patch_line[end_index - 1] == '\\') {
+                        end_index = patch_line.find('\"', end_index + 1);
+                        if (end_index == std::string::npos || end_index < 10)
+                            return;
+                    }
+
+                    auto value = patch_line.substr(10, end_index - 10);
+                    value = EscapeStringSequences(value);
+                    replace.reserve(value.size());
+                    std::copy(value.begin(), value.end(), std::back_inserter(replace));
+                } else {
+                    // hex replacement
+                    const auto value = patch_line.substr(9);
+                    replace.reserve(value.size() / 2);
+                    replace = Common::HexStringToVector(value, is_little_endian);
+                }
+
+                if (print_values) {
+                    LOG_INFO(Loader,
+                             "[IPSwitchCompiler ('{}')]     - Patching value at offset 0x{:08X} "
+                             "with byte string '{}'",
+                             patch_text->GetName(), offset, Common::HexVectorToString(replace));
+                }
+
+                patch.records.insert_or_assign(offset, std::move(replace));
+            }
+
+            patches.push_back(std::move(patch));
+        } else if (StartsWith(line, "@")) {
+            ParseFlag(line);
+        }
+    }
+
+    valid = true;
+}
+
+VirtualFile IPSwitchCompiler::Apply(const VirtualFile& in) const {
+    if (in == nullptr || !valid)
+        return nullptr;
+
+    auto in_data = in->ReadAllBytes();
+
+    for (const auto& patch : patches) {
+        if (!patch.enabled)
+            continue;
+
+        for (const auto& record : patch.records) {
+            if (record.first >= in_data.size())
+                continue;
+            auto replace_size = record.second.size();
+            if (record.first + replace_size > in_data.size())
+                replace_size = in_data.size() - record.first;
+            for (std::size_t i = 0; i < replace_size; ++i)
+                in_data[i + record.first] = record.second[i];
+        }
+    }
+
+    return std::make_shared<VectorVfsFile>(std::move(in_data), in->GetName(),
+                                           in->GetContainingDirectory());
 }
 
 } // namespace FileSys

src/core/file_sys/ips_layer.h

@@ -4,12 +4,41 @@
 
 #pragma once
 
+#include <array>
 #include <memory>
+#include <vector>
 
+#include "common/common_types.h"
 #include "core/file_sys/vfs.h"
 
 namespace FileSys {
 
 VirtualFile PatchIPS(const VirtualFile& in, const VirtualFile& ips);
 
+class IPSwitchCompiler {
+public:
+    explicit IPSwitchCompiler(VirtualFile patch_text);
+    ~IPSwitchCompiler();
+
+    std::array<u8, 0x20> GetBuildID() const;
+    bool IsValid() const;
+    VirtualFile Apply(const VirtualFile& in) const;
+
+private:
+    struct IPSwitchPatch;
+
+    void ParseFlag(const std::string& flag);
+    void Parse();
+
+    bool valid = false;
+
+    VirtualFile patch_text;
+    std::vector<IPSwitchPatch> patches;
+    std::array<u8, 0x20> nso_build_id{};
+    bool is_little_endian = false;
+    s64 offset_shift = 0;
+    bool print_values = false;
+    std::string last_comment = "";
+};
+
 } // namespace FileSys

src/core/file_sys/patch_manager.cpp

@@ -73,27 +73,38 @@ VirtualDir PatchManager::PatchExeFS(VirtualDir exefs) const {
     return exefs;
 }
 
-static std::vector<VirtualFile> CollectIPSPatches(const std::vector<VirtualDir>& patch_dirs,
-                                                  const std::string& build_id) {
-    std::vector<VirtualFile> ips;
-    ips.reserve(patch_dirs.size());
+static std::vector<VirtualFile> CollectPatches(const std::vector<VirtualDir>& patch_dirs,
+                                               const std::string& build_id) {
+    std::vector<VirtualFile> out;
+    out.reserve(patch_dirs.size());
     for (const auto& subdir : patch_dirs) {
         auto exefs_dir = subdir->GetSubdirectory("exefs");
         if (exefs_dir != nullptr) {
             for (const auto& file : exefs_dir->GetFiles()) {
-                if (file->GetExtension() != "ips")
-                    continue;
-                auto name = file->GetName();
-                const auto p1 = name.substr(0, name.find('.'));
-                const auto this_build_id = p1.substr(0, p1.find_last_not_of('0') + 1);
+                if (file->GetExtension() == "ips") {
+                    auto name = file->GetName();
+                    const auto p1 = name.substr(0, name.find('.'));
+                    const auto this_build_id = p1.substr(0, p1.find_last_not_of('0') + 1);
 
-                if (build_id == this_build_id)
-                    ips.push_back(file);
+                    if (build_id == this_build_id)
+                        out.push_back(file);
+                } else if (file->GetExtension() == "pchtxt") {
+                    IPSwitchCompiler compiler{file};
+                    if (!compiler.IsValid())
+                        continue;
+
+                    auto this_build_id = Common::HexArrayToString(compiler.GetBuildID());
+                    this_build_id =
+                        this_build_id.substr(0, this_build_id.find_last_not_of('0') + 1);
+
+                    if (build_id == this_build_id)
+                        out.push_back(file);
+                }
             }
         }
     }
 
-    return ips;
+    return out;
 }
 
 std::vector<u8> PatchManager::PatchNSO(const std::vector<u8>& nso) const {
@@ -115,15 +126,24 @@ std::vector<u8> PatchManager::PatchNSO(const std::vector<u8>& nso) const {
     auto patch_dirs = load_dir->GetSubdirectories();
     std::sort(patch_dirs.begin(), patch_dirs.end(),
               [](const VirtualDir& l, const VirtualDir& r) { return l->GetName() < r->GetName(); });
-    const auto ips = CollectIPSPatches(patch_dirs, build_id);
+    const auto patches = CollectPatches(patch_dirs, build_id);
 
     auto out = nso;
-    for (const auto& ips_file : ips) {
-        LOG_INFO(Loader, "    - Appling IPS patch from mod \"{}\"",
-                 ips_file->GetContainingDirectory()->GetParentDirectory()->GetName());
-        const auto patched = PatchIPS(std::make_shared<VectorVfsFile>(out), ips_file);
-        if (patched != nullptr)
-            out = patched->ReadAllBytes();
+    for (const auto& patch_file : patches) {
+        if (patch_file->GetExtension() == "ips") {
+            LOG_INFO(Loader, "    - Applying IPS patch from mod \"{}\"",
+                     patch_file->GetContainingDirectory()->GetParentDirectory()->GetName());
+            const auto patched = PatchIPS(std::make_shared<VectorVfsFile>(out), patch_file);
+            if (patched != nullptr)
+                out = patched->ReadAllBytes();
+        } else if (patch_file->GetExtension() == "pchtxt") {
+            LOG_INFO(Loader, "    - Applying IPSwitch patch from mod \"{}\"",
+                     patch_file->GetContainingDirectory()->GetParentDirectory()->GetName());
+            const IPSwitchCompiler compiler{patch_file};
+            const auto patched = compiler.Apply(std::make_shared<VectorVfsFile>(out));
+            if (patched != nullptr)
+                out = patched->ReadAllBytes();
+        }
     }
 
     if (out.size() < 0x100)
@@ -143,7 +163,7 @@ bool PatchManager::HasNSOPatch(const std::array<u8, 32>& build_id_) const {
     std::sort(patch_dirs.begin(), patch_dirs.end(),
               [](const VirtualDir& l, const VirtualDir& r) { return l->GetName() < r->GetName(); });
 
-    return !CollectIPSPatches(patch_dirs, build_id).empty();
+    return !CollectPatches(patch_dirs, build_id).empty();
 }
 
 static void ApplyLayeredFS(VirtualFile& romfs, u64 title_id, ContentRecordType type) {
@@ -162,11 +182,17 @@ static void ApplyLayeredFS(VirtualFile& romfs, u64 title_id, ContentRecordType t
               [](const VirtualDir& l, const VirtualDir& r) { return l->GetName() < r->GetName(); });
 
     std::vector<VirtualDir> layers;
+    std::vector<VirtualDir> layers_ext;
     layers.reserve(patch_dirs.size() + 1);
+    layers_ext.reserve(patch_dirs.size() + 1);
     for (const auto& subdir : patch_dirs) {
         auto romfs_dir = subdir->GetSubdirectory("romfs");
         if (romfs_dir != nullptr)
             layers.push_back(std::move(romfs_dir));
+
+        auto ext_dir = subdir->GetSubdirectory("romfs_ext");
+        if (ext_dir != nullptr)
+            layers_ext.push_back(std::move(ext_dir));
     }
     layers.push_back(std::move(extracted));
 
@@ -175,7 +201,9 @@ static void ApplyLayeredFS(VirtualFile& romfs, u64 title_id, ContentRecordType t
         return;
     }
 
-    auto packed = CreateRomFS(std::move(layered));
+    auto layered_ext = LayeredVfsDirectory::MakeLayeredDirectory(std::move(layers_ext));
+
+    auto packed = CreateRomFS(std::move(layered), std::move(layered_ext));
     if (packed == nullptr) {
         return;
     }
@@ -184,8 +212,8 @@ static void ApplyLayeredFS(VirtualFile& romfs, u64 title_id, ContentRecordType t
     romfs = std::move(packed);
 }
 
-VirtualFile PatchManager::PatchRomFS(VirtualFile romfs, u64 ivfc_offset,
-                                     ContentRecordType type) const {
+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));
 
@@ -205,6 +233,13 @@ VirtualFile PatchManager::PatchRomFS(VirtualFile romfs, u64 ivfc_offset,
                      FormatTitleVersion(installed->GetEntryVersion(update_tid).get_value_or(0)));
             romfs = new_nca->GetRomFS();
         }
+    } else if (update_raw != nullptr) {
+        const auto new_nca = std::make_shared<NCA>(update_raw, romfs, ivfc_offset);
+        if (new_nca->GetStatus() == Loader::ResultStatus::Success &&
+            new_nca->GetRomFS() != nullptr) {
+            LOG_INFO(Loader, "    RomFS: Update (PACKED) applied successfully");
+            romfs = new_nca->GetRomFS();
+        }
     }
 
     // LayeredFS
@@ -224,7 +259,8 @@ static bool IsDirValidAndNonEmpty(const VirtualDir& dir) {
     return dir != nullptr && (!dir->GetFiles().empty() || !dir->GetSubdirectories().empty());
 }
 
-std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNames() const {
+std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNames(
+    VirtualFile update_raw) const {
     std::map<std::string, std::string, std::less<>> out;
     const auto installed = Service::FileSystem::GetUnionContents();
 
@@ -245,6 +281,8 @@ std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNam
                     "Update",
                     FormatTitleVersion(meta_ver.get(), TitleVersionFormat::ThreeElements));
             }
+        } else if (update_raw != nullptr) {
+            out.insert_or_assign("Update", "PACKED");
         }
     }
 
@@ -253,8 +291,24 @@ std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNam
     if (mod_dir != nullptr && mod_dir->GetSize() > 0) {
         for (const auto& mod : mod_dir->GetSubdirectories()) {
             std::string types;
-            if (IsDirValidAndNonEmpty(mod->GetSubdirectory("exefs")))
-                AppendCommaIfNotEmpty(types, "IPS");
+
+            const auto exefs_dir = mod->GetSubdirectory("exefs");
+            if (IsDirValidAndNonEmpty(exefs_dir)) {
+                bool ips = false;
+                bool ipswitch = false;
+
+                for (const auto& file : exefs_dir->GetFiles()) {
+                    if (file->GetExtension() == "ips")
+                        ips = true;
+                    else if (file->GetExtension() == "pchtxt")
+                        ipswitch = true;
+                }
+
+                if (ips)
+                    AppendCommaIfNotEmpty(types, "IPS");
+                if (ipswitch)
+                    AppendCommaIfNotEmpty(types, "IPSwitch");
+            }
             if (IsDirValidAndNonEmpty(mod->GetSubdirectory("romfs")))
                 AppendCommaIfNotEmpty(types, "LayeredFS");
 
@@ -291,23 +345,22 @@ std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNam
     return out;
 }
 
-std::pair<std::shared_ptr<NACP>, VirtualFile> PatchManager::GetControlMetadata() const {
+std::pair<std::unique_ptr<NACP>, VirtualFile> PatchManager::GetControlMetadata() const {
     const auto& installed{Service::FileSystem::GetUnionContents()};
 
     const auto base_control_nca = installed->GetEntry(title_id, ContentRecordType::Control);
     if (base_control_nca == nullptr)
         return {};
 
-    return ParseControlNCA(base_control_nca);
+    return ParseControlNCA(*base_control_nca);
 }
 
-std::pair<std::shared_ptr<NACP>, VirtualFile> PatchManager::ParseControlNCA(
-    const std::shared_ptr<NCA>& nca) const {
-    const auto base_romfs = nca->GetRomFS();
+std::pair<std::unique_ptr<NACP>, VirtualFile> PatchManager::ParseControlNCA(const NCA& nca) const {
+    const auto base_romfs = nca.GetRomFS();
     if (base_romfs == nullptr)
         return {};
 
-    const auto romfs = PatchRomFS(base_romfs, nca->GetBaseIVFCOffset(), ContentRecordType::Control);
+    const auto romfs = PatchRomFS(base_romfs, nca.GetBaseIVFCOffset(), ContentRecordType::Control);
     if (romfs == nullptr)
         return {};
 
@@ -319,7 +372,7 @@ std::pair<std::shared_ptr<NACP>, VirtualFile> PatchManager::ParseControlNCA(
     if (nacp_file == nullptr)
         nacp_file = extracted->GetFile("Control.nacp");
 
-    const auto nacp = nacp_file == nullptr ? nullptr : std::make_shared<NACP>(nacp_file);
+    auto nacp = nacp_file == nullptr ? nullptr : std::make_unique<NACP>(nacp_file);
 
     VirtualFile icon_file;
     for (const auto& language : FileSys::LANGUAGE_NAMES) {
@@ -328,6 +381,6 @@ std::pair<std::shared_ptr<NACP>, VirtualFile> PatchManager::ParseControlNCA(
             break;
     }
 
-    return {nacp, icon_file};
+    return {std::move(nacp), icon_file};
 }
 } // namespace FileSys

src/core/file_sys/patch_manager.h

@@ -36,6 +36,7 @@ public:
 
     // Currently tracked NSO patches:
     // - IPS
+    // - IPSwitch
     std::vector<u8> PatchNSO(const std::vector<u8>& nso) const;
 
     // Checks to see if PatchNSO() will have any effect given the NSO's build ID.
@@ -46,19 +47,20 @@ public:
     // - Game Updates
     // - LayeredFS
     VirtualFile PatchRomFS(VirtualFile base, u64 ivfc_offset,
-                           ContentRecordType type = ContentRecordType::Program) const;
+                           ContentRecordType type = ContentRecordType::Program,
+                           VirtualFile update_raw = nullptr) const;
 
     // Returns a vector of pairs between patch names and patch versions.
     // i.e. Update 3.2.2 will return {"Update", "3.2.2"}
-    std::map<std::string, std::string, std::less<>> GetPatchVersionNames() const;
+    std::map<std::string, std::string, std::less<>> GetPatchVersionNames(
+        VirtualFile update_raw = nullptr) const;
 
     // Given title_id of the program, attempts to get the control data of the update and parse it,
     // falling back to the base control data.
-    std::pair<std::shared_ptr<NACP>, VirtualFile> GetControlMetadata() const;
+    std::pair<std::unique_ptr<NACP>, VirtualFile> GetControlMetadata() const;
 
     // Version of GetControlMetadata that takes an arbitrary NCA
-    std::pair<std::shared_ptr<NACP>, VirtualFile> ParseControlNCA(
-        const std::shared_ptr<NCA>& nca) const;
+    std::pair<std::unique_ptr<NACP>, VirtualFile> ParseControlNCA(const NCA& nca) const;
 
 private:
     u64 title_id;

src/core/file_sys/romfs.cpp

@@ -129,11 +129,11 @@ VirtualDir ExtractRomFS(VirtualFile file, RomFSExtractionType type) {
     return out;
 }
 
-VirtualFile CreateRomFS(VirtualDir dir) {
+VirtualFile CreateRomFS(VirtualDir dir, VirtualDir ext) {
     if (dir == nullptr)
         return nullptr;
 
-    RomFSBuildContext ctx{dir};
+    RomFSBuildContext ctx{dir, ext};
     return ConcatenatedVfsFile::MakeConcatenatedFile(0, ctx.Build(), dir->GetName());
 }
 

src/core/file_sys/romfs.h

@@ -44,6 +44,6 @@ VirtualDir ExtractRomFS(VirtualFile file,
 
 // Converts a VFS filesystem into a RomFS binary
 // Returns nullptr on failure
-VirtualFile CreateRomFS(VirtualDir dir);
+VirtualFile CreateRomFS(VirtualDir dir, VirtualDir ext = nullptr);
 
 } // namespace FileSys

src/core/file_sys/romfs_factory.cpp

@@ -30,12 +30,17 @@ RomFSFactory::RomFSFactory(Loader::AppLoader& app_loader) {
 
 RomFSFactory::~RomFSFactory() = default;
 
+void RomFSFactory::SetPackedUpdate(VirtualFile update_raw) {
+    this->update_raw = std::move(update_raw);
+}
+
 ResultVal<VirtualFile> RomFSFactory::OpenCurrentProcess() {
     if (!updatable)
         return MakeResult<VirtualFile>(file);
 
     const PatchManager patch_manager(Core::CurrentProcess()->GetTitleID());
-    return MakeResult<VirtualFile>(patch_manager.PatchRomFS(file, ivfc_offset));
+    return MakeResult<VirtualFile>(
+        patch_manager.PatchRomFS(file, ivfc_offset, ContentRecordType::Program, update_raw));
 }
 
 ResultVal<VirtualFile> RomFSFactory::Open(u64 title_id, StorageId storage, ContentRecordType type) {

src/core/file_sys/romfs_factory.h

@@ -32,11 +32,13 @@ public:
     explicit RomFSFactory(Loader::AppLoader& app_loader);
     ~RomFSFactory();
 
+    void SetPackedUpdate(VirtualFile update_raw);
     ResultVal<VirtualFile> OpenCurrentProcess();
     ResultVal<VirtualFile> Open(u64 title_id, StorageId storage, ContentRecordType type);
 
 private:
     VirtualFile file;
+    VirtualFile update_raw;
     bool updatable;
     u64 ivfc_offset;
 };

src/core/file_sys/submission_package.cpp

@@ -259,8 +259,11 @@ void NSP::ReadNCAs(const std::vector<VirtualFile>& files) {
                 auto next_nca = std::make_shared<NCA>(next_file);
                 if (next_nca->GetType() == NCAContentType::Program)
                     program_status[cnmt.GetTitleID()] = next_nca->GetStatus();
-                if (next_nca->GetStatus() == Loader::ResultStatus::Success)
+                if (next_nca->GetStatus() == Loader::ResultStatus::Success ||
+                    (next_nca->GetStatus() == Loader::ResultStatus::ErrorMissingBKTRBaseRomFS &&
+                     (cnmt.GetTitleID() & 0x800) != 0)) {
                     ncas_title[rec.type] = std::move(next_nca);
+                }
             }
 
             break;

src/core/file_sys/vfs.h

@@ -12,20 +12,12 @@
 #include <vector>
 #include <boost/optional.hpp>
 #include "common/common_types.h"
+#include "core/file_sys/vfs_types.h"
 
 namespace FileSys {
 
-class VfsDirectory;
-class VfsFile;
-class VfsFilesystem;
-
 enum class Mode : u32;
 
-// Convenience typedefs to use Vfs* interfaces
-using VirtualFilesystem = std::shared_ptr<VfsFilesystem>;
-using VirtualDir = std::shared_ptr<VfsDirectory>;
-using VirtualFile = std::shared_ptr<VfsFile>;
-
 // An enumeration representing what can be at the end of a path in a VfsFilesystem
 enum class VfsEntryType {
     None,

src/core/file_sys/vfs_types.h

@@ -0,0 +1,21 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+
+namespace FileSys {
+
+class VfsDirectory;
+class VfsFile;
+class VfsFilesystem;
+
+// Declarations for Vfs* pointer types
+
+using VirtualDir = std::shared_ptr<VfsDirectory>;
+using VirtualFile = std::shared_ptr<VfsFile>;
+using VirtualFilesystem = std::shared_ptr<VfsFilesystem>;
+
+} // namespace FileSys

src/core/hle/kernel/errors.h

@@ -22,6 +22,7 @@ enum {
     HandleTableFull = 105,
     InvalidMemoryState = 106,
     InvalidMemoryPermissions = 108,
+    InvalidMemoryRange = 110,
     InvalidThreadPriority = 112,
     InvalidProcessorId = 113,
     InvalidHandle = 114,
@@ -56,6 +57,7 @@ constexpr ResultCode ERR_INVALID_ADDRESS(ErrorModule::Kernel, ErrCodes::InvalidA
 constexpr ResultCode ERR_INVALID_ADDRESS_STATE(ErrorModule::Kernel, ErrCodes::InvalidMemoryState);
 constexpr ResultCode ERR_INVALID_MEMORY_PERMISSIONS(ErrorModule::Kernel,
                                                     ErrCodes::InvalidMemoryPermissions);
+constexpr ResultCode ERR_INVALID_MEMORY_RANGE(ErrorModule::Kernel, ErrCodes::InvalidMemoryRange);
 constexpr ResultCode ERR_INVALID_HANDLE(ErrorModule::Kernel, ErrCodes::InvalidHandle);
 constexpr ResultCode ERR_INVALID_PROCESSOR_ID(ErrorModule::Kernel, ErrCodes::InvalidProcessorId);
 constexpr ResultCode ERR_INVALID_SIZE(ErrorModule::Kernel, ErrCodes::InvalidSize);

src/core/hle/kernel/kernel.cpp

@@ -116,7 +116,7 @@ struct KernelCore::Impl {
         next_thread_id = 1;
 
         process_list.clear();
-        current_process.reset();
+        current_process = nullptr;
 
         handle_table.Clear();
         resource_limits.fill(nullptr);
@@ -207,7 +207,7 @@ struct KernelCore::Impl {
 
     // Lists all processes that exist in the current session.
     std::vector<SharedPtr<Process>> process_list;
-    SharedPtr<Process> current_process;
+    Process* current_process = nullptr;
 
     Kernel::HandleTable handle_table;
     std::array<SharedPtr<ResourceLimit>, 4> resource_limits;
@@ -266,15 +266,15 @@ void KernelCore::AppendNewProcess(SharedPtr<Process> process) {
     impl->process_list.push_back(std::move(process));
 }
 
-void KernelCore::MakeCurrentProcess(SharedPtr<Process> process) {
-    impl->current_process = std::move(process);
+void KernelCore::MakeCurrentProcess(Process* process) {
+    impl->current_process = process;
 }
 
-SharedPtr<Process>& KernelCore::CurrentProcess() {
+Process* KernelCore::CurrentProcess() {
     return impl->current_process;
 }
 
-const SharedPtr<Process>& KernelCore::CurrentProcess() const {
+const Process* KernelCore::CurrentProcess() const {
     return impl->current_process;
 }
 

src/core/hle/kernel/kernel.h

@@ -66,13 +66,13 @@ public:
     void AppendNewProcess(SharedPtr<Process> process);
 
     /// Makes the given process the new current process.
-    void MakeCurrentProcess(SharedPtr<Process> process);
+    void MakeCurrentProcess(Process* process);
 
-    /// Retrieves a reference to the current process.
-    SharedPtr<Process>& CurrentProcess();
+    /// Retrieves a pointer to the current process.
+    Process* CurrentProcess();
 
-    /// Retrieves a const reference to the current process.
-    const SharedPtr<Process>& CurrentProcess() const;
+    /// Retrieves a const pointer to the current process.
+    const Process* CurrentProcess() const;
 
     /// Adds a port to the named port table
     void AddNamedPort(std::string name, SharedPtr<ClientPort> port);

src/core/hle/kernel/object.cpp

@@ -25,7 +25,6 @@ bool Object::IsWaitable() const {
     case HandleType::Process:
     case HandleType::AddressArbiter:
     case HandleType::ResourceLimit:
-    case HandleType::CodeSet:
     case HandleType::ClientPort:
     case HandleType::ClientSession:
         return false;

src/core/hle/kernel/object.h

@@ -26,7 +26,6 @@ enum class HandleType : u32 {
     AddressArbiter,
     Timer,
     ResourceLimit,
-    CodeSet,
     ClientPort,
     ServerPort,
     ClientSession,

src/core/hle/kernel/process.cpp

@@ -20,13 +20,7 @@
 
 namespace Kernel {
 
-SharedPtr<CodeSet> CodeSet::Create(KernelCore& kernel, std::string name) {
-    SharedPtr<CodeSet> codeset(new CodeSet(kernel));
-    codeset->name = std::move(name);
-    return codeset;
-}
-
-CodeSet::CodeSet(KernelCore& kernel) : Object{kernel} {}
+CodeSet::CodeSet() = default;
 CodeSet::~CodeSet() = default;
 
 SharedPtr<Process> Process::Create(KernelCore& kernel, std::string&& name) {
@@ -224,20 +218,20 @@ void Process::FreeTLSSlot(VAddr tls_address) {
     tls_slots[tls_page].reset(tls_slot);
 }
 
-void Process::LoadModule(SharedPtr<CodeSet> module_, VAddr base_addr) {
+void Process::LoadModule(CodeSet module_, VAddr base_addr) {
     const auto MapSegment = [&](CodeSet::Segment& segment, VMAPermission permissions,
                                 MemoryState memory_state) {
-        auto vma = vm_manager
-                       .MapMemoryBlock(segment.addr + base_addr, module_->memory, segment.offset,
-                                       segment.size, memory_state)
-                       .Unwrap();
+        const auto vma = vm_manager
+                             .MapMemoryBlock(segment.addr + base_addr, module_.memory,
+                                             segment.offset, segment.size, memory_state)
+                             .Unwrap();
         vm_manager.Reprotect(vma, permissions);
     };
 
     // Map CodeSet segments
-    MapSegment(module_->CodeSegment(), VMAPermission::ReadExecute, MemoryState::CodeStatic);
-    MapSegment(module_->RODataSegment(), VMAPermission::Read, MemoryState::CodeMutable);
-    MapSegment(module_->DataSegment(), VMAPermission::ReadWrite, MemoryState::CodeMutable);
+    MapSegment(module_.CodeSegment(), VMAPermission::ReadExecute, MemoryState::CodeStatic);
+    MapSegment(module_.RODataSegment(), VMAPermission::Read, MemoryState::CodeMutable);
+    MapSegment(module_.DataSegment(), VMAPermission::ReadWrite, MemoryState::CodeMutable);
 }
 
 ResultVal<VAddr> Process::HeapAllocate(VAddr target, u64 size, VMAPermission perms) {

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
@@ -57,30 +58,33 @@ union ProcessFlags {
     BitField<12, 1, u16> loaded_high; ///< Application loaded high (not at 0x00100000).
 };
 
-enum class ProcessStatus { Created, Running, Exited };
+/**
+ * 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,
+};
 
-class ResourceLimit;
-
-struct CodeSet final : public Object {
+struct CodeSet final {
     struct Segment {
         std::size_t offset = 0;
         VAddr addr = 0;
         u32 size = 0;
     };
 
-    static SharedPtr<CodeSet> Create(KernelCore& kernel, std::string name);
-
-    std::string GetTypeName() const override {
-        return "CodeSet";
-    }
-    std::string GetName() const override {
-        return name;
-    }
-
-    static const HandleType HANDLE_TYPE = HandleType::CodeSet;
-    HandleType GetHandleType() const override {
-        return HANDLE_TYPE;
-    }
+    explicit CodeSet();
+    ~CodeSet();
 
     Segment& CodeSegment() {
         return segments[0];
@@ -109,14 +113,7 @@ struct CodeSet final : public Object {
     std::shared_ptr<std::vector<u8>> memory;
 
     std::array<Segment, 3> segments;
-    VAddr entrypoint;
-
-    /// Name of the process
-    std::string name;
-
-private:
-    explicit CodeSet(KernelCore& kernel);
-    ~CodeSet() override;
+    VAddr entrypoint = 0;
 };
 
 class Process final : public Object {
@@ -219,7 +216,7 @@ public:
      */
     void PrepareForTermination();
 
-    void LoadModule(SharedPtr<CodeSet> module_, VAddr base_addr);
+    void LoadModule(CodeSet module_, VAddr base_addr);
 
     ///////////////////////////////////////////////////////////////////////////////////////////////
     // Memory Management

src/core/hle/kernel/scheduler.cpp

@@ -9,7 +9,7 @@
 #include "common/logging/log.h"
 #include "core/arm/arm_interface.h"
 #include "core/core.h"
-#include "core/core_timing.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/scheduler.h"
 
@@ -78,16 +78,16 @@ void Scheduler::SwitchContext(Thread* new_thread) {
         // Cancel any outstanding wakeup events for this thread
         new_thread->CancelWakeupTimer();
 
-        auto previous_process = Core::CurrentProcess();
+        auto* const previous_process = Core::CurrentProcess();
 
         current_thread = new_thread;
 
         ready_queue.remove(new_thread->GetPriority(), new_thread);
         new_thread->SetStatus(ThreadStatus::Running);
 
-        const auto thread_owner_process = current_thread->GetOwnerProcess();
+        auto* const thread_owner_process = current_thread->GetOwnerProcess();
         if (previous_process != thread_owner_process) {
-            Core::CurrentProcess() = thread_owner_process;
+            Core::System::GetInstance().Kernel().MakeCurrentProcess(thread_owner_process);
             SetCurrentPageTable(&Core::CurrentProcess()->VMManager().page_table);
         }
 

src/core/hle/kernel/svc.cpp

@@ -39,6 +39,73 @@ 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
+// or if the given size is zero.
+constexpr bool IsValidAddressRange(VAddr address, u64 size) {
+    return address + size > address;
+}
+
+// Checks if a given address range lies within a larger address range.
+constexpr bool IsInsideAddressRange(VAddr address, u64 size, VAddr address_range_begin,
+                                    VAddr address_range_end) {
+    const VAddr end_address = address + size - 1;
+    return address_range_begin <= address && end_address <= address_range_end - 1;
+}
+
+bool IsInsideAddressSpace(const VMManager& vm, VAddr address, u64 size) {
+    return IsInsideAddressRange(address, size, vm.GetAddressSpaceBaseAddress(),
+                                vm.GetAddressSpaceEndAddress());
+}
+
+bool IsInsideNewMapRegion(const VMManager& vm, VAddr address, u64 size) {
+    return IsInsideAddressRange(address, size, vm.GetNewMapRegionBaseAddress(),
+                                vm.GetNewMapRegionEndAddress());
+}
+
+// Helper function that performs the common sanity checks for svcMapMemory
+// and svcUnmapMemory. This is doable, as both functions perform their sanitizing
+// 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)) {
+        return ERR_INVALID_ADDRESS;
+    }
+
+    if (size == 0 || !Is4KBAligned(size)) {
+        return ERR_INVALID_SIZE;
+    }
+
+    if (!IsValidAddressRange(dst_addr, size)) {
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+
+    if (!IsValidAddressRange(src_addr, size)) {
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+
+    if (!IsInsideAddressSpace(vm_manager, src_addr, size)) {
+        return ERR_INVALID_ADDRESS_STATE;
+    }
+
+    if (!IsInsideNewMapRegion(vm_manager, dst_addr, size)) {
+        return ERR_INVALID_MEMORY_RANGE;
+    }
+
+    const VAddr dst_end_address = dst_addr + size;
+    if (dst_end_address > vm_manager.GetHeapRegionBaseAddress() &&
+        vm_manager.GetHeapRegionEndAddress() > dst_addr) {
+        return ERR_INVALID_MEMORY_RANGE;
+    }
+
+    if (dst_end_address > vm_manager.GetMapRegionBaseAddress() &&
+        vm_manager.GetMapRegionEndAddress() > dst_addr) {
+        return ERR_INVALID_MEMORY_RANGE;
+    }
+
+    return RESULT_SUCCESS;
+}
 } // Anonymous namespace
 
 /// Set the process heap to a given Size. It can both extend and shrink the heap.
@@ -69,15 +136,15 @@ static ResultCode MapMemory(VAddr dst_addr, VAddr src_addr, u64 size) {
     LOG_TRACE(Kernel_SVC, "called, dst_addr=0x{:X}, src_addr=0x{:X}, size=0x{:X}", dst_addr,
               src_addr, size);
 
-    if (!Is4KBAligned(dst_addr) || !Is4KBAligned(src_addr)) {
-        return ERR_INVALID_ADDRESS;
+    auto* const current_process = Core::CurrentProcess();
+    const auto& vm_manager = current_process->VMManager();
+
+    const auto result = MapUnmapMemorySanityChecks(vm_manager, dst_addr, src_addr, size);
+    if (result != RESULT_SUCCESS) {
+        return result;
     }
 
-    if (size == 0 || !Is4KBAligned(size)) {
-        return ERR_INVALID_SIZE;
-    }
-
-    return Core::CurrentProcess()->MirrorMemory(dst_addr, src_addr, size);
+    return current_process->MirrorMemory(dst_addr, src_addr, size);
 }
 
 /// Unmaps a region that was previously mapped with svcMapMemory
@@ -85,15 +152,15 @@ static ResultCode UnmapMemory(VAddr dst_addr, VAddr src_addr, u64 size) {
     LOG_TRACE(Kernel_SVC, "called, dst_addr=0x{:X}, src_addr=0x{:X}, size=0x{:X}", dst_addr,
               src_addr, size);
 
-    if (!Is4KBAligned(dst_addr) || !Is4KBAligned(src_addr)) {
-        return ERR_INVALID_ADDRESS;
+    auto* const current_process = Core::CurrentProcess();
+    const auto& vm_manager = current_process->VMManager();
+
+    const auto result = MapUnmapMemorySanityChecks(vm_manager, dst_addr, src_addr, size);
+    if (result != RESULT_SUCCESS) {
+        return result;
     }
 
-    if (size == 0 || !Is4KBAligned(size)) {
-        return ERR_INVALID_SIZE;
-    }
-
-    return Core::CurrentProcess()->UnmapMemory(dst_addr, src_addr, size);
+    return current_process->UnmapMemory(dst_addr, src_addr, size);
 }
 
 /// Connect to an OS service given the port name, returns the handle to the port to out
@@ -301,13 +368,34 @@ static ResultCode ArbitrateUnlock(VAddr mutex_addr) {
     return Mutex::Release(mutex_addr);
 }
 
+struct BreakReason {
+    union {
+        u32 raw;
+        BitField<31, 1, u32> signal_debugger;
+    };
+};
+
 /// Break program execution
-static void Break(u64 reason, u64 info1, u64 info2) {
-    LOG_CRITICAL(
-        Debug_Emulated,
-        "Emulated program broke execution! reason=0x{:016X}, info1=0x{:016X}, info2=0x{:016X}",
-        reason, info1, info2);
-    ASSERT(false);
+static void Break(u32 reason, u64 info1, u64 info2) {
+    BreakReason break_reason{reason};
+    if (break_reason.signal_debugger) {
+        LOG_ERROR(
+            Debug_Emulated,
+            "Emulated program broke execution! reason=0x{:016X}, info1=0x{:016X}, info2=0x{:016X}",
+            reason, info1, info2);
+    } else {
+        LOG_CRITICAL(
+            Debug_Emulated,
+            "Emulated program broke execution! reason=0x{:016X}, info1=0x{:016X}, info2=0x{:016X}",
+            reason, info1, info2);
+        ASSERT(false);
+
+        Core::CurrentProcess()->PrepareForTermination();
+
+        // Kill the current thread
+        GetCurrentThread()->Stop();
+        Core::System::GetInstance().PrepareReschedule();
+    }
 }
 
 /// Used to output a message on a debug hardware unit - does nothing on a retail unit
@@ -326,7 +414,7 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
     LOG_TRACE(Kernel_SVC, "called info_id=0x{:X}, info_sub_id=0x{:X}, handle=0x{:08X}", info_id,
               info_sub_id, handle);
 
-    const auto& current_process = Core::CurrentProcess();
+    const auto* current_process = Core::CurrentProcess();
     const auto& vm_manager = current_process->VMManager();
 
     switch (static_cast<GetInfoType>(info_id)) {
@@ -360,25 +448,12 @@ 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;
@@ -424,7 +499,7 @@ static ResultCode GetThreadContext(VAddr thread_context, Handle handle) {
         return ERR_INVALID_HANDLE;
     }
 
-    const auto current_process = Core::CurrentProcess();
+    const auto* current_process = Core::CurrentProcess();
     if (thread->GetOwnerProcess() != current_process) {
         return ERR_INVALID_HANDLE;
     }
@@ -516,7 +591,7 @@ static ResultCode MapSharedMemory(Handle shared_memory_handle, VAddr addr, u64 s
         return ERR_INVALID_HANDLE;
     }
 
-    return shared_memory->Map(Core::CurrentProcess().get(), addr, permissions_type,
+    return shared_memory->Map(Core::CurrentProcess(), addr, permissions_type,
                               MemoryPermission::DontCare);
 }
 
@@ -535,7 +610,7 @@ static ResultCode UnmapSharedMemory(Handle shared_memory_handle, VAddr addr, u64
     auto& kernel = Core::System::GetInstance().Kernel();
     auto shared_memory = kernel.HandleTable().Get<SharedMemory>(shared_memory_handle);
 
-    return shared_memory->Unmap(Core::CurrentProcess().get(), addr);
+    return shared_memory->Unmap(Core::CurrentProcess(), addr);
 }
 
 /// Query process memory
@@ -573,7 +648,7 @@ static ResultCode QueryMemory(MemoryInfo* memory_info, PageInfo* page_info, VAdd
 
 /// Exits the current process
 static void ExitProcess() {
-    auto& current_process = Core::CurrentProcess();
+    auto* current_process = Core::CurrentProcess();
 
     LOG_INFO(Kernel_SVC, "Process {} exiting", current_process->GetProcessID());
     ASSERT_MSG(current_process->GetStatus() == ProcessStatus::Running,
@@ -621,7 +696,7 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V
     auto& kernel = Core::System::GetInstance().Kernel();
     CASCADE_RESULT(SharedPtr<Thread> thread,
                    Thread::Create(kernel, name, entry_point, priority, arg, processor_id, stack_top,
-                                  Core::CurrentProcess()));
+                                  *Core::CurrentProcess()));
     const auto new_guest_handle = kernel.HandleTable().Create(thread);
     if (new_guest_handle.Failed()) {
         return new_guest_handle.Code();
@@ -1010,6 +1085,29 @@ 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();
@@ -1145,7 +1243,7 @@ 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

@@ -35,18 +35,18 @@ void SvcWrap() {
 
 template <ResultCode func(u32)>
 void SvcWrap() {
-    FuncReturn(func((u32)Param(0)).raw);
+    FuncReturn(func(static_cast<u32>(Param(0))).raw);
 }
 
 template <ResultCode func(u32, u32)>
 void SvcWrap() {
-    FuncReturn(func((u32)Param(0), (u32)Param(1)).raw);
+    FuncReturn(func(static_cast<u32>(Param(0)), static_cast<u32>(Param(1))).raw);
 }
 
 template <ResultCode func(u32*, u32)>
 void SvcWrap() {
     u32 param_1 = 0;
-    u32 retval = func(&param_1, (u32)Param(1)).raw;
+    u32 retval = func(&param_1, static_cast<u32>(Param(1))).raw;
     Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
@@ -61,7 +61,7 @@ void SvcWrap() {
 
 template <ResultCode func(u64, s32)>
 void SvcWrap() {
-    FuncReturn(func(Param(0), (s32)Param(1)).raw);
+    FuncReturn(func(Param(0), static_cast<s32>(Param(1))).raw);
 }
 
 template <ResultCode func(u64, u32)>
@@ -77,21 +77,29 @@ 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((u32)(Param(0) & 0xFFFFFFFF), Param(1)).raw);
+    FuncReturn(func(static_cast<u32>(Param(0)), Param(1)).raw);
 }
 
 template <ResultCode func(u32, u32, u64)>
 void SvcWrap() {
-    FuncReturn(func((u32)(Param(0) & 0xFFFFFFFF), (u32)(Param(1) & 0xFFFFFFFF), Param(2)).raw);
+    FuncReturn(func(static_cast<u32>(Param(0)), static_cast<u32>(Param(1)), Param(2)).raw);
 }
 
 template <ResultCode func(u32, u32*, u64*)>
 void SvcWrap() {
     u32 param_1 = 0;
     u64 param_2 = 0;
-    ResultCode retval = func((u32)(Param(2) & 0xFFFFFFFF), &param_1, &param_2);
+    ResultCode retval = func(static_cast<u32>(Param(2)), &param_1, &param_2);
     Core::CurrentArmInterface().SetReg(1, param_1);
     Core::CurrentArmInterface().SetReg(2, param_2);
     FuncReturn(retval.raw);
@@ -100,12 +108,12 @@ void SvcWrap() {
 template <ResultCode func(u64, u64, u32, u32)>
 void SvcWrap() {
     FuncReturn(
-        func(Param(0), Param(1), (u32)(Param(3) & 0xFFFFFFFF), (u32)(Param(3) & 0xFFFFFFFF)).raw);
+        func(Param(0), Param(1), static_cast<u32>(Param(3)), static_cast<u32>(Param(3))).raw);
 }
 
 template <ResultCode func(u32, u64, u32)>
 void SvcWrap() {
-    FuncReturn(func((u32)Param(0), Param(1), (u32)Param(2)).raw);
+    FuncReturn(func(static_cast<u32>(Param(0)), Param(1), static_cast<u32>(Param(2))).raw);
 }
 
 template <ResultCode func(u64, u64, u64)>
@@ -115,25 +123,28 @@ void SvcWrap() {
 
 template <ResultCode func(u32, u64, u64, u32)>
 void SvcWrap() {
-    FuncReturn(func((u32)Param(0), Param(1), Param(2), (u32)Param(3)).raw);
+    FuncReturn(
+        func(static_cast<u32>(Param(0)), Param(1), Param(2), static_cast<u32>(Param(3))).raw);
 }
 
 template <ResultCode func(u32, u64, u64)>
 void SvcWrap() {
-    FuncReturn(func((u32)Param(0), Param(1), Param(2)).raw);
+    FuncReturn(func(static_cast<u32>(Param(0)), Param(1), Param(2)).raw);
 }
 
 template <ResultCode func(u32*, u64, u64, s64)>
 void SvcWrap() {
     u32 param_1 = 0;
-    ResultCode retval = func(&param_1, Param(1), (u32)(Param(2) & 0xFFFFFFFF), (s64)Param(3));
+    ResultCode retval =
+        func(&param_1, Param(1), static_cast<u32>(Param(2)), static_cast<s64>(Param(3)));
     Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval.raw);
 }
 
 template <ResultCode func(u64, u64, u32, s64)>
 void SvcWrap() {
-    FuncReturn(func(Param(0), Param(1), (u32)Param(2), (s64)Param(3)).raw);
+    FuncReturn(
+        func(Param(0), Param(1), static_cast<u32>(Param(2)), static_cast<s64>(Param(3))).raw);
 }
 
 template <ResultCode func(u64*, u64, u64, u64)>
@@ -147,9 +158,9 @@ void SvcWrap() {
 template <ResultCode func(u32*, u64, u64, u64, u32, s32)>
 void SvcWrap() {
     u32 param_1 = 0;
-    u32 retval =
-        func(&param_1, Param(1), Param(2), Param(3), (u32)Param(4), (s32)(Param(5) & 0xFFFFFFFF))
-            .raw;
+    u32 retval = func(&param_1, Param(1), Param(2), Param(3), static_cast<u32>(Param(4)),
+                      static_cast<s32>(Param(5)))
+                     .raw;
     Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
@@ -172,7 +183,7 @@ void SvcWrap() {
 template <ResultCode func(u32*, u64, u64, u32)>
 void SvcWrap() {
     u32 param_1 = 0;
-    u32 retval = func(&param_1, Param(1), Param(2), (u32)(Param(3) & 0xFFFFFFFF)).raw;
+    u32 retval = func(&param_1, Param(1), Param(2), static_cast<u32>(Param(3))).raw;
     Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
@@ -181,22 +192,22 @@ template <ResultCode func(Handle*, u64, u32, u32)>
 void SvcWrap() {
     u32 param_1 = 0;
     u32 retval =
-        func(&param_1, Param(1), (u32)(Param(2) & 0xFFFFFFFF), (u32)(Param(3) & 0xFFFFFFFF)).raw;
+        func(&param_1, Param(1), static_cast<u32>(Param(2)), static_cast<u32>(Param(3))).raw;
     Core::CurrentArmInterface().SetReg(1, param_1);
     FuncReturn(retval);
 }
 
 template <ResultCode func(u64, u32, s32, s64)>
 void SvcWrap() {
-    FuncReturn(
-        func(Param(0), (u32)(Param(1) & 0xFFFFFFFF), (s32)(Param(2) & 0xFFFFFFFF), (s64)Param(3))
-            .raw);
+    FuncReturn(func(Param(0), static_cast<u32>(Param(1)), static_cast<s32>(Param(2)),
+                    static_cast<s64>(Param(3)))
+                   .raw);
 }
 
 template <ResultCode func(u64, u32, s32, s32)>
 void SvcWrap() {
-    FuncReturn(func(Param(0), (u32)(Param(1) & 0xFFFFFFFF), (s32)(Param(2) & 0xFFFFFFFF),
-                    (s32)(Param(3) & 0xFFFFFFFF))
+    FuncReturn(func(Param(0), static_cast<u32>(Param(1)), static_cast<s32>(Param(2)),
+                    static_cast<s32>(Param(3)))
                    .raw);
 }
 
@@ -226,7 +237,7 @@ void SvcWrap() {
 
 template <void func(s64)>
 void SvcWrap() {
-    func((s64)Param(0));
+    func(static_cast<s64>(Param(0)));
 }
 
 template <void func(u64, u64 len)>
@@ -239,4 +250,9 @@ void SvcWrap() {
     func(Param(0), Param(1), Param(2));
 }
 
+template <void func(u32, u64, u64)>
+void SvcWrap() {
+    func(static_cast<u32>(Param(0)), Param(1), Param(2));
+}
+
 } // namespace Kernel

src/core/hle/kernel/thread.cpp

@@ -183,18 +183,15 @@ void Thread::ResumeFromWait() {
  */
 static void ResetThreadContext(Core::ARM_Interface::ThreadContext& context, VAddr stack_top,
                                VAddr entry_point, u64 arg) {
-    memset(&context, 0, sizeof(Core::ARM_Interface::ThreadContext));
-
+    context = {};
     context.cpu_registers[0] = arg;
     context.pc = entry_point;
     context.sp = stack_top;
-    context.pstate = 0;
-    context.fpcr = 0;
 }
 
 ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name, VAddr entry_point,
                                             u32 priority, u64 arg, s32 processor_id,
-                                            VAddr stack_top, SharedPtr<Process> owner_process) {
+                                            VAddr stack_top, Process& owner_process) {
     // Check if priority is in ranged. Lowest priority -> highest priority id.
     if (priority > THREADPRIO_LOWEST) {
         LOG_ERROR(Kernel_SVC, "Invalid thread priority: {}", priority);
@@ -208,7 +205,7 @@ ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name
 
     // TODO(yuriks): Other checks, returning 0xD9001BEA
 
-    if (!Memory::IsValidVirtualAddress(*owner_process, entry_point)) {
+    if (!Memory::IsValidVirtualAddress(owner_process, entry_point)) {
         LOG_ERROR(Kernel_SVC, "(name={}): invalid entry {:016X}", name, entry_point);
         // TODO (bunnei): Find the correct error code to use here
         return ResultCode(-1);
@@ -232,7 +229,7 @@ ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name
     thread->wait_handle = 0;
     thread->name = std::move(name);
     thread->callback_handle = kernel.ThreadWakeupCallbackHandleTable().Create(thread).Unwrap();
-    thread->owner_process = owner_process;
+    thread->owner_process = &owner_process;
     thread->scheduler = Core::System::GetInstance().Scheduler(processor_id).get();
     thread->scheduler->AddThread(thread, priority);
     thread->tls_address = thread->owner_process->MarkNextAvailableTLSSlotAsUsed(*thread);
@@ -264,7 +261,7 @@ SharedPtr<Thread> SetupMainThread(KernelCore& kernel, VAddr entry_point, u32 pri
     // Initialize new "main" thread
     const VAddr stack_top = owner_process.VMManager().GetTLSIORegionEndAddress();
     auto thread_res = Thread::Create(kernel, "main", entry_point, priority, 0, THREADPROCESSORID_0,
-                                     stack_top, &owner_process);
+                                     stack_top, owner_process);
 
     SharedPtr<Thread> thread = std::move(thread_res).Unwrap();
 

src/core/hle/kernel/thread.h

@@ -89,7 +89,7 @@ public:
     static ResultVal<SharedPtr<Thread>> Create(KernelCore& kernel, std::string name,
                                                VAddr entry_point, u32 priority, u64 arg,
                                                s32 processor_id, VAddr stack_top,
-                                               SharedPtr<Process> owner_process);
+                                               Process& owner_process);
 
     std::string GetName() const override {
         return name;
@@ -262,11 +262,11 @@ public:
         return processor_id;
     }
 
-    SharedPtr<Process>& GetOwnerProcess() {
+    Process* GetOwnerProcess() {
         return owner_process;
     }
 
-    const SharedPtr<Process>& GetOwnerProcess() const {
+    const Process* GetOwnerProcess() const {
         return owner_process;
     }
 
@@ -386,7 +386,7 @@ private:
     u64 tpidr_el0 = 0;     ///< TPIDR_EL0 read/write system register.
 
     /// Process that owns this thread
-    SharedPtr<Process> owner_process;
+    Process* owner_process;
 
     /// Objects that the thread is waiting on, in the same order as they were
     /// passed to WaitSynchronization1/N.

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/aoc/aoc_u.cpp

@@ -61,13 +61,13 @@ AOC_U::AOC_U() : ServiceFramework("aoc:u"), add_on_content(AccumulateAOCTitleIDs
 AOC_U::~AOC_U() = default;
 
 void AOC_U::CountAddOnContent(Kernel::HLERequestContext& ctx) {
-    IPC::ResponseBuilder rb{ctx, 4};
+    IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
 
     const auto current = Core::System::GetInstance().CurrentProcess()->GetTitleID();
-    rb.Push<u32>(std::count_if(add_on_content.begin(), add_on_content.end(), [&current](u64 tid) {
-        return (tid & DLC_BASE_TITLE_ID_MASK) == current;
-    }));
+    rb.Push<u32>(static_cast<u32>(
+        std::count_if(add_on_content.begin(), add_on_content.end(),
+                      [&current](u64 tid) { return (tid & DLC_BASE_TITLE_ID_MASK) == current; })));
 }
 
 void AOC_U::ListAddOnContent(Kernel::HLERequestContext& ctx) {
@@ -91,7 +91,7 @@ void AOC_U::ListAddOnContent(Kernel::HLERequestContext& ctx) {
         return;
     }
 
-    count = std::min<size_t>(out.size() - offset, count);
+    count = static_cast<u32>(std::min<size_t>(out.size() - offset, count));
     std::rotate(out.begin(), out.begin() + offset, out.end());
     out.resize(count);
 

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

@@ -2,8 +2,10 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <chrono>
 #include <cstring>
 #include <memory>
+#include <optional>
 #include <vector>
 
 #include <opus.h>
@@ -33,7 +35,8 @@ public:
             {1, nullptr, "SetContext"},
             {2, nullptr, "DecodeInterleavedForMultiStream"},
             {3, nullptr, "SetContextForMultiStream"},
-            {4, nullptr, "Unknown4"},
+            {4, &IHardwareOpusDecoderManager::DecodeInterleavedWithPerformance,
+             "DecodeInterleavedWithPerformance"},
             {5, nullptr, "Unknown5"},
             {6, nullptr, "Unknown6"},
             {7, nullptr, "Unknown7"},
@@ -59,8 +62,31 @@ private:
         ctx.WriteBuffer(samples.data(), samples.size() * sizeof(s16));
     }
 
-    bool Decoder_DecodeInterleaved(u32& consumed, u32& sample_count, const std::vector<u8>& input,
-                                   std::vector<opus_int16>& output) {
+    void DecodeInterleavedWithPerformance(Kernel::HLERequestContext& ctx) {
+        u32 consumed = 0;
+        u32 sample_count = 0;
+        u64 performance = 0;
+        std::vector<opus_int16> samples(ctx.GetWriteBufferSize() / sizeof(opus_int16));
+        if (!Decoder_DecodeInterleaved(consumed, sample_count, ctx.ReadBuffer(), samples,
+                                       performance)) {
+            IPC::ResponseBuilder rb{ctx, 2};
+            // TODO(ogniK): Use correct error code
+            rb.Push(ResultCode(-1));
+            return;
+        }
+        IPC::ResponseBuilder rb{ctx, 6};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(consumed);
+        rb.Push<u64>(performance);
+        rb.Push<u32>(sample_count);
+        ctx.WriteBuffer(samples.data(), samples.size() * sizeof(s16));
+    }
+
+    bool Decoder_DecodeInterleaved(
+        u32& consumed, u32& sample_count, const std::vector<u8>& input,
+        std::vector<opus_int16>& output,
+        std::optional<std::reference_wrapper<u64>> performance_time = std::nullopt) {
+        const auto start_time = std::chrono::high_resolution_clock::now();
         std::size_t raw_output_sz = output.size() * sizeof(opus_int16);
         if (sizeof(OpusHeader) > input.size())
             return false;
@@ -80,8 +106,13 @@ private:
                         (static_cast<int>(raw_output_sz / sizeof(s16) / channel_count)), 0);
         if (out_sample_count < 0)
             return false;
+        const auto end_time = std::chrono::high_resolution_clock::now() - start_time;
         sample_count = out_sample_count;
         consumed = static_cast<u32>(sizeof(OpusHeader) + hdr.sz);
+        if (performance_time.has_value()) {
+            performance_time->get() =
+                std::chrono::duration_cast<std::chrono::milliseconds>(end_time).count();
+        }
         return true;
     }
 

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

@@ -264,6 +264,15 @@ ResultCode RegisterBIS(std::unique_ptr<FileSys::BISFactory>&& factory) {
     return RESULT_SUCCESS;
 }
 
+void SetPackedUpdate(FileSys::VirtualFile update_raw) {
+    LOG_TRACE(Service_FS, "Setting packed update for romfs");
+
+    if (romfs_factory == nullptr)
+        return;
+
+    romfs_factory->SetPackedUpdate(std::move(update_raw));
+}
+
 ResultVal<FileSys::VirtualFile> OpenRomFSCurrentProcess() {
     LOG_TRACE(Service_FS, "Opening RomFS for current process");
 

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

@@ -39,6 +39,7 @@ ResultCode RegisterSaveData(std::unique_ptr<FileSys::SaveDataFactory>&& factory)
 ResultCode RegisterSDMC(std::unique_ptr<FileSys::SDMCFactory>&& factory);
 ResultCode RegisterBIS(std::unique_ptr<FileSys::BISFactory>&& factory);
 
+void SetPackedUpdate(FileSys::VirtualFile update_raw);
 ResultVal<FileSys::VirtualFile> OpenRomFSCurrentProcess();
 ResultVal<FileSys::VirtualFile> OpenRomFS(u64 title_id, FileSys::StorageId storage_id,
                                           FileSys::ContentRecordType type);

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

@@ -15,6 +15,11 @@
 #include "video_core/renderer_base.h"
 
 namespace Service::Nvidia::Devices {
+namespace NvErrCodes {
+enum {
+    InvalidNmapHandle = -22,
+};
+}
 
 nvhost_as_gpu::nvhost_as_gpu(std::shared_ptr<nvmap> nvmap_dev) : nvmap_dev(std::move(nvmap_dev)) {}
 nvhost_as_gpu::~nvhost_as_gpu() = default;
@@ -79,14 +84,16 @@ u32 nvhost_as_gpu::Remap(const std::vector<u8>& input, std::vector<u8>& output)
     std::memcpy(entries.data(), input.data(), input.size());
 
     auto& gpu = Core::System::GetInstance().GPU();
-
     for (const auto& entry : entries) {
         LOG_WARNING(Service_NVDRV, "remap entry, offset=0x{:X} handle=0x{:X} pages=0x{:X}",
                     entry.offset, entry.nvmap_handle, entry.pages);
         Tegra::GPUVAddr offset = static_cast<Tegra::GPUVAddr>(entry.offset) << 0x10;
-
         auto object = nvmap_dev->GetObject(entry.nvmap_handle);
-        ASSERT(object);
+        if (!object) {
+            LOG_CRITICAL(Service_NVDRV, "nvmap {} is an invalid handle!", entry.nvmap_handle);
+            std::memcpy(output.data(), entries.data(), output.size());
+            return static_cast<u32>(NvErrCodes::InvalidNmapHandle);
+        }
 
         ASSERT(object->status == nvmap::Object::Status::Allocated);
 
@@ -157,15 +164,21 @@ u32 nvhost_as_gpu::UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& ou
     LOG_DEBUG(Service_NVDRV, "called, offset=0x{:X}", params.offset);
 
     const auto itr = buffer_mappings.find(params.offset);
-    ASSERT_MSG(itr != buffer_mappings.end(), "Tried to unmap invalid mapping");
+    if (itr == buffer_mappings.end()) {
+        LOG_WARNING(Service_NVDRV, "Tried to unmap an invalid offset 0x{:X}", params.offset);
+        // Hardware tests shows that unmapping an already unmapped buffer always returns successful
+        // and doesn't fail.
+        return 0;
+    }
 
     auto& system_instance = Core::System::GetInstance();
 
     // Remove this memory region from the rasterizer cache.
-    system_instance.Renderer().Rasterizer().FlushAndInvalidateRegion(params.offset,
-                                                                     itr->second.size);
-
     auto& gpu = system_instance.GPU();
+    auto cpu_addr = gpu.MemoryManager().GpuToCpuAddress(params.offset);
+    ASSERT(cpu_addr);
+    system_instance.Renderer().Rasterizer().FlushAndInvalidateRegion(*cpu_addr, itr->second.size);
+
     params.offset = gpu.MemoryManager().UnmapBuffer(params.offset, itr->second.size);
 
     buffer_mappings.erase(itr->second.offset);

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

@@ -11,6 +11,13 @@
 
 namespace Service::Nvidia::Devices {
 
+namespace NvErrCodes {
+enum {
+    OperationNotPermitted = -1,
+    InvalidValue = -22,
+};
+}
+
 nvmap::nvmap() = default;
 nvmap::~nvmap() = default;
 
@@ -44,7 +51,11 @@ u32 nvmap::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<u8>& o
 u32 nvmap::IocCreate(const std::vector<u8>& input, std::vector<u8>& output) {
     IocCreateParams params;
     std::memcpy(&params, input.data(), sizeof(params));
+    LOG_DEBUG(Service_NVDRV, "size=0x{:08X}", params.size);
 
+    if (!params.size) {
+        return static_cast<u32>(NvErrCodes::InvalidValue);
+    }
     // Create a new nvmap object and obtain a handle to it.
     auto object = std::make_shared<Object>();
     object->id = next_id++;
@@ -55,8 +66,6 @@ u32 nvmap::IocCreate(const std::vector<u8>& input, std::vector<u8>& output) {
     u32 handle = next_handle++;
     handles[handle] = std::move(object);
 
-    LOG_DEBUG(Service_NVDRV, "size=0x{:08X}", params.size);
-
     params.handle = handle;
 
     std::memcpy(output.data(), &params, sizeof(params));
@@ -66,9 +75,29 @@ u32 nvmap::IocCreate(const std::vector<u8>& input, std::vector<u8>& output) {
 u32 nvmap::IocAlloc(const std::vector<u8>& input, std::vector<u8>& output) {
     IocAllocParams params;
     std::memcpy(&params, input.data(), sizeof(params));
+    LOG_DEBUG(Service_NVDRV, "called, addr={:X}", params.addr);
+
+    if (!params.handle) {
+        return static_cast<u32>(NvErrCodes::InvalidValue);
+    }
+
+    if ((params.align - 1) & params.align) {
+        return static_cast<u32>(NvErrCodes::InvalidValue);
+    }
+
+    const u32 min_alignment = 0x1000;
+    if (params.align < min_alignment) {
+        params.align = min_alignment;
+    }
 
     auto object = GetObject(params.handle);
-    ASSERT(object);
+    if (!object) {
+        return static_cast<u32>(NvErrCodes::InvalidValue);
+    }
+
+    if (object->status == Object::Status::Allocated) {
+        return static_cast<u32>(NvErrCodes::OperationNotPermitted);
+    }
 
     object->flags = params.flags;
     object->align = params.align;
@@ -76,8 +105,6 @@ u32 nvmap::IocAlloc(const std::vector<u8>& input, std::vector<u8>& output) {
     object->addr = params.addr;
     object->status = Object::Status::Allocated;
 
-    LOG_DEBUG(Service_NVDRV, "called, addr={:X}", params.addr);
-
     std::memcpy(output.data(), &params, sizeof(params));
     return 0;
 }
@@ -88,8 +115,14 @@ u32 nvmap::IocGetId(const std::vector<u8>& input, std::vector<u8>& output) {
 
     LOG_WARNING(Service_NVDRV, "called");
 
+    if (!params.handle) {
+        return static_cast<u32>(NvErrCodes::InvalidValue);
+    }
+
     auto object = GetObject(params.handle);
-    ASSERT(object);
+    if (!object) {
+        return static_cast<u32>(NvErrCodes::OperationNotPermitted);
+    }
 
     params.id = object->id;
 
@@ -105,7 +138,14 @@ u32 nvmap::IocFromId(const std::vector<u8>& input, std::vector<u8>& output) {
 
     auto itr = std::find_if(handles.begin(), handles.end(),
                             [&](const auto& entry) { return entry.second->id == params.id; });
-    ASSERT(itr != handles.end());
+    if (itr == handles.end()) {
+        return static_cast<u32>(NvErrCodes::InvalidValue);
+    }
+
+    auto& object = itr->second;
+    if (object->status != Object::Status::Allocated) {
+        return static_cast<u32>(NvErrCodes::InvalidValue);
+    }
 
     itr->second->refcount++;
 
@@ -125,8 +165,13 @@ u32 nvmap::IocParam(const std::vector<u8>& input, std::vector<u8>& output) {
     LOG_WARNING(Service_NVDRV, "(STUBBED) called type={}", params.param);
 
     auto object = GetObject(params.handle);
-    ASSERT(object);
-    ASSERT(object->status == Object::Status::Allocated);
+    if (!object) {
+        return static_cast<u32>(NvErrCodes::InvalidValue);
+    }
+
+    if (object->status != Object::Status::Allocated) {
+        return static_cast<u32>(NvErrCodes::OperationNotPermitted);
+    }
 
     switch (static_cast<ParamTypes>(params.param)) {
     case ParamTypes::Size:
@@ -163,9 +208,12 @@ u32 nvmap::IocFree(const std::vector<u8>& input, std::vector<u8>& output) {
     LOG_WARNING(Service_NVDRV, "(STUBBED) called");
 
     auto itr = handles.find(params.handle);
-    ASSERT(itr != handles.end());
-
-    ASSERT(itr->second->refcount > 0);
+    if (itr == handles.end()) {
+        return static_cast<u32>(NvErrCodes::InvalidValue);
+    }
+    if (!itr->second->refcount) {
+        return static_cast<u32>(NvErrCodes::InvalidValue);
+    }
 
     itr->second->refcount--;
 

src/core/loader/deconstructed_rom_directory.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <cinttypes>
+#include <cstring>
 #include "common/common_funcs.h"
 #include "common/file_util.h"
 #include "common/logging/log.h"
@@ -140,7 +141,8 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(Kernel::Process& process)
         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, pm);
+            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);

src/core/loader/elf.cpp

@@ -9,16 +9,11 @@
 #include "common/common_types.h"
 #include "common/file_util.h"
 #include "common/logging/log.h"
-#include "core/core.h"
-#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/vm_manager.h"
 #include "core/loader/elf.h"
 #include "core/memory.h"
 
-using Kernel::CodeSet;
-using Kernel::SharedPtr;
-
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 // ELF Header Constants
 
@@ -211,7 +206,7 @@ public:
     u32 GetFlags() const {
         return (u32)(header->e_flags);
     }
-    SharedPtr<CodeSet> LoadInto(VAddr vaddr);
+    Kernel::CodeSet LoadInto(VAddr vaddr);
 
     int GetNumSegments() const {
         return (int)(header->e_phnum);
@@ -274,7 +269,7 @@ const char* ElfReader::GetSectionName(int section) const {
     return nullptr;
 }
 
-SharedPtr<CodeSet> ElfReader::LoadInto(VAddr vaddr) {
+Kernel::CodeSet ElfReader::LoadInto(VAddr vaddr) {
     LOG_DEBUG(Loader, "String section: {}", header->e_shstrndx);
 
     // Should we relocate?
@@ -302,8 +297,7 @@ SharedPtr<CodeSet> ElfReader::LoadInto(VAddr vaddr) {
     std::vector<u8> program_image(total_image_size);
     std::size_t current_image_position = 0;
 
-    auto& kernel = Core::System::GetInstance().Kernel();
-    SharedPtr<CodeSet> codeset = CodeSet::Create(kernel, "");
+    Kernel::CodeSet codeset;
 
     for (unsigned int i = 0; i < header->e_phnum; ++i) {
         const Elf32_Phdr* p = &segments[i];
@@ -311,14 +305,14 @@ SharedPtr<CodeSet> ElfReader::LoadInto(VAddr vaddr) {
                   p->p_vaddr, p->p_filesz, p->p_memsz);
 
         if (p->p_type == PT_LOAD) {
-            CodeSet::Segment* codeset_segment;
+            Kernel::CodeSet::Segment* codeset_segment;
             u32 permission_flags = p->p_flags & (PF_R | PF_W | PF_X);
             if (permission_flags == (PF_R | PF_X)) {
-                codeset_segment = &codeset->CodeSegment();
+                codeset_segment = &codeset.CodeSegment();
             } else if (permission_flags == (PF_R)) {
-                codeset_segment = &codeset->RODataSegment();
+                codeset_segment = &codeset.RODataSegment();
             } else if (permission_flags == (PF_R | PF_W)) {
-                codeset_segment = &codeset->DataSegment();
+                codeset_segment = &codeset.DataSegment();
             } else {
                 LOG_ERROR(Loader, "Unexpected ELF PT_LOAD segment id {} with flags {:X}", i,
                           p->p_flags);
@@ -345,8 +339,8 @@ SharedPtr<CodeSet> ElfReader::LoadInto(VAddr vaddr) {
         }
     }
 
-    codeset->entrypoint = base_addr + header->e_entry;
-    codeset->memory = std::make_shared<std::vector<u8>>(std::move(program_image));
+    codeset.entrypoint = base_addr + header->e_entry;
+    codeset.memory = std::make_shared<std::vector<u8>>(std::move(program_image));
 
     LOG_DEBUG(Loader, "Done loading.");
 
@@ -397,11 +391,11 @@ ResultStatus AppLoader_ELF::Load(Kernel::Process& process) {
 
     const VAddr base_address = process.VMManager().GetCodeRegionBaseAddress();
     ElfReader elf_reader(&buffer[0]);
-    SharedPtr<CodeSet> codeset = elf_reader.LoadInto(base_address);
-    codeset->name = file->GetName();
+    Kernel::CodeSet codeset = elf_reader.LoadInto(base_address);
+    const VAddr entry_point = codeset.entrypoint;
 
-    process.LoadModule(codeset, codeset->entrypoint);
-    process.Run(codeset->entrypoint, 48, Memory::DEFAULT_STACK_SIZE);
+    process.LoadModule(std::move(codeset), entry_point);
+    process.Run(entry_point, 48, Memory::DEFAULT_STACK_SIZE);
 
     is_loaded = true;
     return ResultStatus::Success;

src/core/loader/loader.cpp

@@ -93,7 +93,7 @@ std::string GetFileTypeString(FileType type) {
     return "unknown";
 }
 
-constexpr std::array<const char*, 58> RESULT_MESSAGES{
+constexpr std::array<const char*, 59> RESULT_MESSAGES{
     "The operation completed successfully.",
     "The loader requested to load is already loaded.",
     "The operation is not implemented.",
@@ -152,6 +152,7 @@ constexpr std::array<const char*, 58> RESULT_MESSAGES{
     "The BKTR-type NCA has a bad Relocation bucket.",
     "The BKTR-type NCA has a bad Subsection bucket.",
     "The BKTR-type NCA is missing the base RomFS.",
+    "The NSP or XCI does not contain an update in addition to the base game.",
 };
 
 std::ostream& operator<<(std::ostream& os, ResultStatus status) {

src/core/loader/loader.h

@@ -114,6 +114,7 @@ enum class ResultStatus : u16 {
     ErrorBadRelocationBuckets,
     ErrorBadSubsectionBuckets,
     ErrorMissingBKTRBaseRomFS,
+    ErrorNoPackedUpdate,
 };
 
 std::ostream& operator<<(std::ostream& os, ResultStatus status);
@@ -196,10 +197,19 @@ public:
     /**
      * Get the RomFS of the application
      * Since the RomFS can be huge, we return a file reference instead of copying to a buffer
-     * @param dir The directory containing the RomFS
+     * @param file The directory containing the RomFS
      * @return ResultStatus result of function
      */
-    virtual ResultStatus ReadRomFS(FileSys::VirtualFile& dir) {
+    virtual ResultStatus ReadRomFS(FileSys::VirtualFile& file) {
+        return ResultStatus::ErrorNotImplemented;
+    }
+
+    /**
+     * Get the raw update of the application, should it come packed with one
+     * @param file The raw update NCA file (Program-type
+     * @return ResultStatus result of function
+     */
+    virtual ResultStatus ReadUpdateRaw(FileSys::VirtualFile& file) {
         return ResultStatus::ErrorNotImplemented;
     }
 

src/core/loader/nax.cpp

@@ -72,6 +72,10 @@ ResultStatus AppLoader_NAX::ReadRomFS(FileSys::VirtualFile& dir) {
     return nca_loader->ReadRomFS(dir);
 }
 
+u64 AppLoader_NAX::ReadRomFSIVFCOffset() const {
+    return nca_loader->ReadRomFSIVFCOffset();
+}
+
 ResultStatus AppLoader_NAX::ReadProgramId(u64& out_program_id) {
     return nca_loader->ReadProgramId(out_program_id);
 }

src/core/loader/nax.h

@@ -36,6 +36,7 @@ public:
     ResultStatus Load(Kernel::Process& process) override;
 
     ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
+    u64 ReadRomFSIVFCOffset() const override;
     ResultStatus ReadProgramId(u64& out_program_id) override;
 
 private:

src/core/loader/nro.cpp

@@ -14,11 +14,12 @@
 #include "core/file_sys/control_metadata.h"
 #include "core/file_sys/vfs_offset.h"
 #include "core/gdbstub/gdbstub.h"
-#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/vm_manager.h"
 #include "core/loader/nro.h"
+#include "core/loader/nso.h"
 #include "core/memory.h"
+#include "core/settings.h"
 
 namespace Loader {
 
@@ -137,17 +138,29 @@ bool AppLoader_NRO::LoadNro(FileSys::VirtualFile file, VAddr load_base) {
     }
 
     // Build program image
-    auto& kernel = Core::System::GetInstance().Kernel();
-    Kernel::SharedPtr<Kernel::CodeSet> codeset = Kernel::CodeSet::Create(kernel, "");
     std::vector<u8> program_image = file->ReadBytes(PageAlignSize(nro_header.file_size));
     if (program_image.size() != PageAlignSize(nro_header.file_size)) {
         return {};
     }
 
+    Kernel::CodeSet codeset;
     for (std::size_t i = 0; i < nro_header.segments.size(); ++i) {
-        codeset->segments[i].addr = nro_header.segments[i].offset;
-        codeset->segments[i].offset = nro_header.segments[i].offset;
-        codeset->segments[i].size = PageAlignSize(nro_header.segments[i].size);
+        codeset.segments[i].addr = nro_header.segments[i].offset;
+        codeset.segments[i].offset = nro_header.segments[i].offset;
+        codeset.segments[i].size = PageAlignSize(nro_header.segments[i].size);
+    }
+
+    if (!Settings::values.program_args.empty()) {
+        const auto arg_data = Settings::values.program_args;
+        codeset.DataSegment().size += NSO_ARGUMENT_DATA_ALLOCATION_SIZE;
+        NSOArgumentHeader args_header{
+            NSO_ARGUMENT_DATA_ALLOCATION_SIZE, static_cast<u32_le>(arg_data.size()), {}};
+        const auto end_offset = program_image.size();
+        program_image.resize(static_cast<u32>(program_image.size()) +
+                             NSO_ARGUMENT_DATA_ALLOCATION_SIZE);
+        std::memcpy(program_image.data() + end_offset, &args_header, sizeof(NSOArgumentHeader));
+        std::memcpy(program_image.data() + end_offset + sizeof(NSOArgumentHeader), arg_data.data(),
+                    arg_data.size());
     }
 
     // Read MOD header
@@ -161,16 +174,15 @@ bool AppLoader_NRO::LoadNro(FileSys::VirtualFile file, VAddr load_base) {
         // Resize program image to include .bss section and page align each section
         bss_size = PageAlignSize(mod_header.bss_end_offset - mod_header.bss_start_offset);
     }
-    codeset->DataSegment().size += bss_size;
+    codeset.DataSegment().size += bss_size;
     program_image.resize(static_cast<u32>(program_image.size()) + bss_size);
 
     // Load codeset for current process
-    codeset->name = file->GetName();
-    codeset->memory = std::make_shared<std::vector<u8>>(std::move(program_image));
-    Core::CurrentProcess()->LoadModule(codeset, load_base);
+    codeset.memory = std::make_shared<std::vector<u8>>(std::move(program_image));
+    Core::CurrentProcess()->LoadModule(std::move(codeset), load_base);
 
     // Register module with GDBStub
-    GDBStub::RegisterModule(codeset->name, load_base, load_base);
+    GDBStub::RegisterModule(file->GetName(), load_base, load_base);
 
     return true;
 }

src/core/loader/nso.cpp

@@ -12,11 +12,11 @@
 #include "core/core.h"
 #include "core/file_sys/patch_manager.h"
 #include "core/gdbstub/gdbstub.h"
-#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/vm_manager.h"
 #include "core/loader/nso.h"
 #include "core/memory.h"
+#include "core/settings.h"
 
 namespace Loader {
 
@@ -94,6 +94,7 @@ static constexpr u32 PageAlignSize(u32 size) {
 }
 
 VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base,
+                                bool should_pass_arguments,
                                 boost::optional<FileSys::PatchManager> pm) {
     if (file == nullptr)
         return {};
@@ -109,8 +110,7 @@ VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base,
         return {};
 
     // Build program image
-    auto& kernel = Core::System::GetInstance().Kernel();
-    Kernel::SharedPtr<Kernel::CodeSet> codeset = Kernel::CodeSet::Create(kernel, "");
+    Kernel::CodeSet codeset;
     std::vector<u8> program_image;
     for (std::size_t i = 0; i < nso_header.segments.size(); ++i) {
         std::vector<u8> data =
@@ -120,9 +120,22 @@ VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base,
         }
         program_image.resize(nso_header.segments[i].location);
         program_image.insert(program_image.end(), data.begin(), data.end());
-        codeset->segments[i].addr = nso_header.segments[i].location;
-        codeset->segments[i].offset = nso_header.segments[i].location;
-        codeset->segments[i].size = PageAlignSize(static_cast<u32>(data.size()));
+        codeset.segments[i].addr = nso_header.segments[i].location;
+        codeset.segments[i].offset = nso_header.segments[i].location;
+        codeset.segments[i].size = PageAlignSize(static_cast<u32>(data.size()));
+    }
+
+    if (should_pass_arguments && !Settings::values.program_args.empty()) {
+        const auto arg_data = Settings::values.program_args;
+        codeset.DataSegment().size += NSO_ARGUMENT_DATA_ALLOCATION_SIZE;
+        NSOArgumentHeader args_header{
+            NSO_ARGUMENT_DATA_ALLOCATION_SIZE, static_cast<u32_le>(arg_data.size()), {}};
+        const auto end_offset = program_image.size();
+        program_image.resize(static_cast<u32>(program_image.size()) +
+                             NSO_ARGUMENT_DATA_ALLOCATION_SIZE);
+        std::memcpy(program_image.data() + end_offset, &args_header, sizeof(NSOArgumentHeader));
+        std::memcpy(program_image.data() + end_offset + sizeof(NSOArgumentHeader), arg_data.data(),
+                    arg_data.size());
     }
 
     // MOD header pointer is at .text offset + 4
@@ -139,7 +152,7 @@ VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base,
         // Resize program image to include .bss section and page align each section
         bss_size = PageAlignSize(mod_header.bss_end_offset - mod_header.bss_start_offset);
     }
-    codeset->DataSegment().size += bss_size;
+    codeset.DataSegment().size += bss_size;
     const u32 image_size{PageAlignSize(static_cast<u32>(program_image.size()) + bss_size)};
     program_image.resize(image_size);
 
@@ -155,12 +168,11 @@ VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base,
     }
 
     // Load codeset for current process
-    codeset->name = file->GetName();
-    codeset->memory = std::make_shared<std::vector<u8>>(std::move(program_image));
-    Core::CurrentProcess()->LoadModule(codeset, load_base);
+    codeset.memory = std::make_shared<std::vector<u8>>(std::move(program_image));
+    Core::CurrentProcess()->LoadModule(std::move(codeset), load_base);
 
     // Register module with GDBStub
-    GDBStub::RegisterModule(codeset->name, load_base, load_base);
+    GDBStub::RegisterModule(file->GetName(), load_base, load_base);
 
     return load_base + image_size;
 }
@@ -172,7 +184,7 @@ ResultStatus AppLoader_NSO::Load(Kernel::Process& process) {
 
     // Load module
     const VAddr base_address = process.VMManager().GetCodeRegionBaseAddress();
-    LoadModule(file, base_address);
+    LoadModule(file, base_address, true);
     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

@@ -11,6 +11,15 @@
 
 namespace Loader {
 
+constexpr u64 NSO_ARGUMENT_DATA_ALLOCATION_SIZE = 0x9000;
+
+struct NSOArgumentHeader {
+    u32_le allocated_size;
+    u32_le actual_size;
+    INSERT_PADDING_BYTES(0x18);
+};
+static_assert(sizeof(NSOArgumentHeader) == 0x20, "NSOArgumentHeader has incorrect size.");
+
 /// Loads an NSO file
 class AppLoader_NSO final : public AppLoader, Linker {
 public:
@@ -27,7 +36,7 @@ public:
         return IdentifyType(file);
     }
 
-    static VAddr LoadModule(FileSys::VirtualFile file, VAddr load_base,
+    static VAddr LoadModule(FileSys::VirtualFile file, VAddr load_base, bool should_pass_arguments,
                             boost::optional<FileSys::PatchManager> pm = boost::none);
 
     ResultStatus Load(Kernel::Process& process) override;

src/core/loader/nsp.cpp

@@ -10,8 +10,10 @@
 #include "core/file_sys/control_metadata.h"
 #include "core/file_sys/nca_metadata.h"
 #include "core/file_sys/patch_manager.h"
+#include "core/file_sys/registered_cache.h"
 #include "core/file_sys/submission_package.h"
 #include "core/hle/kernel/process.h"
+#include "core/hle/service/filesystem/filesystem.h"
 #include "core/loader/deconstructed_rom_directory.h"
 #include "core/loader/nca.h"
 #include "core/loader/nsp.h"
@@ -33,7 +35,7 @@ AppLoader_NSP::AppLoader_NSP(FileSys::VirtualFile file)
         return;
 
     std::tie(nacp_file, icon_file) =
-        FileSys::PatchManager(nsp->GetProgramTitleID()).ParseControlNCA(control_nca);
+        FileSys::PatchManager(nsp->GetProgramTitleID()).ParseControlNCA(*control_nca);
 }
 
 AppLoader_NSP::~AppLoader_NSP() = default;
@@ -91,13 +93,39 @@ ResultStatus AppLoader_NSP::Load(Kernel::Process& process) {
     if (result != ResultStatus::Success)
         return result;
 
+    FileSys::VirtualFile update_raw;
+    if (ReadUpdateRaw(update_raw) == ResultStatus::Success && update_raw != nullptr)
+        Service::FileSystem::SetPackedUpdate(std::move(update_raw));
+
     is_loaded = true;
 
     return ResultStatus::Success;
 }
 
-ResultStatus AppLoader_NSP::ReadRomFS(FileSys::VirtualFile& dir) {
-    return secondary_loader->ReadRomFS(dir);
+ResultStatus AppLoader_NSP::ReadRomFS(FileSys::VirtualFile& file) {
+    return secondary_loader->ReadRomFS(file);
+}
+
+u64 AppLoader_NSP::ReadRomFSIVFCOffset() const {
+    return secondary_loader->ReadRomFSIVFCOffset();
+}
+
+ResultStatus AppLoader_NSP::ReadUpdateRaw(FileSys::VirtualFile& file) {
+    if (nsp->IsExtractedType())
+        return ResultStatus::ErrorNoPackedUpdate;
+
+    const auto read =
+        nsp->GetNCAFile(FileSys::GetUpdateTitleID(title_id), FileSys::ContentRecordType::Program);
+
+    if (read == nullptr)
+        return ResultStatus::ErrorNoPackedUpdate;
+    const auto nca_test = std::make_shared<FileSys::NCA>(read);
+
+    if (nca_test->GetStatus() != ResultStatus::ErrorMissingBKTRBaseRomFS)
+        return nca_test->GetStatus();
+
+    file = read;
+    return ResultStatus::Success;
 }
 
 ResultStatus AppLoader_NSP::ReadProgramId(u64& out_program_id) {

src/core/loader/nsp.h

@@ -37,7 +37,9 @@ public:
 
     ResultStatus Load(Kernel::Process& process) override;
 
-    ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
+    ResultStatus ReadRomFS(FileSys::VirtualFile& file) override;
+    u64 ReadRomFSIVFCOffset() const override;
+    ResultStatus ReadUpdateRaw(FileSys::VirtualFile& file) override;
     ResultStatus ReadProgramId(u64& out_program_id) override;
     ResultStatus ReadIcon(std::vector<u8>& buffer) override;
     ResultStatus ReadTitle(std::string& title) override;
@@ -47,7 +49,7 @@ private:
     std::unique_ptr<AppLoader> secondary_loader;
 
     FileSys::VirtualFile icon_file;
-    std::shared_ptr<FileSys::NACP> nacp_file;
+    std::unique_ptr<FileSys::NACP> nacp_file;
     u64 title_id;
 };
 

src/core/loader/xci.cpp

@@ -9,7 +9,11 @@
 #include "core/file_sys/content_archive.h"
 #include "core/file_sys/control_metadata.h"
 #include "core/file_sys/patch_manager.h"
+#include "core/file_sys/registered_cache.h"
+#include "core/file_sys/romfs.h"
+#include "core/file_sys/submission_package.h"
 #include "core/hle/kernel/process.h"
+#include "core/hle/service/filesystem/filesystem.h"
 #include "core/loader/nca.h"
 #include "core/loader/xci.h"
 
@@ -26,7 +30,7 @@ AppLoader_XCI::AppLoader_XCI(FileSys::VirtualFile file)
         return;
 
     std::tie(nacp_file, icon_file) =
-        FileSys::PatchManager(xci->GetProgramTitleID()).ParseControlNCA(control_nca);
+        FileSys::PatchManager(xci->GetProgramTitleID()).ParseControlNCA(*control_nca);
 }
 
 AppLoader_XCI::~AppLoader_XCI() = default;
@@ -63,13 +67,41 @@ ResultStatus AppLoader_XCI::Load(Kernel::Process& process) {
     if (result != ResultStatus::Success)
         return result;
 
+    FileSys::VirtualFile update_raw;
+    if (ReadUpdateRaw(update_raw) == ResultStatus::Success && update_raw != nullptr)
+        Service::FileSystem::SetPackedUpdate(std::move(update_raw));
+
     is_loaded = true;
 
     return ResultStatus::Success;
 }
 
-ResultStatus AppLoader_XCI::ReadRomFS(FileSys::VirtualFile& dir) {
-    return nca_loader->ReadRomFS(dir);
+ResultStatus AppLoader_XCI::ReadRomFS(FileSys::VirtualFile& file) {
+    return nca_loader->ReadRomFS(file);
+}
+
+u64 AppLoader_XCI::ReadRomFSIVFCOffset() const {
+    return nca_loader->ReadRomFSIVFCOffset();
+}
+
+ResultStatus AppLoader_XCI::ReadUpdateRaw(FileSys::VirtualFile& file) {
+    u64 program_id{};
+    nca_loader->ReadProgramId(program_id);
+    if (program_id == 0)
+        return ResultStatus::ErrorXCIMissingProgramNCA;
+
+    const auto read = xci->GetSecurePartitionNSP()->GetNCAFile(
+        FileSys::GetUpdateTitleID(program_id), FileSys::ContentRecordType::Program);
+
+    if (read == nullptr)
+        return ResultStatus::ErrorNoPackedUpdate;
+    const auto nca_test = std::make_shared<FileSys::NCA>(read);
+
+    if (nca_test->GetStatus() != ResultStatus::ErrorMissingBKTRBaseRomFS)
+        return nca_test->GetStatus();
+
+    file = read;
+    return ResultStatus::Success;
 }
 
 ResultStatus AppLoader_XCI::ReadProgramId(u64& out_program_id) {

src/core/loader/xci.h

@@ -37,7 +37,9 @@ public:
 
     ResultStatus Load(Kernel::Process& process) override;
 
-    ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
+    ResultStatus ReadRomFS(FileSys::VirtualFile& file) override;
+    u64 ReadRomFSIVFCOffset() const override;
+    ResultStatus ReadUpdateRaw(FileSys::VirtualFile& file) override;
     ResultStatus ReadProgramId(u64& out_program_id) override;
     ResultStatus ReadIcon(std::vector<u8>& buffer) override;
     ResultStatus ReadTitle(std::string& title) override;
@@ -47,7 +49,7 @@ private:
     std::unique_ptr<AppLoader_NCA> nca_loader;
 
     FileSys::VirtualFile icon_file;
-    std::shared_ptr<FileSys::NACP> nacp_file;
+    std::unique_ptr<FileSys::NACP> nacp_file;
 };
 
 } // namespace Loader

src/core/settings.h

@@ -155,6 +155,7 @@ struct Values {
     // Debugging
     bool use_gdbstub;
     u16 gdbstub_port;
+    std::string program_args;
 
     // WebService
     bool enable_telemetry;

src/core/telemetry_session.cpp

@@ -2,12 +2,16 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "common/assert.h"
-#include "common/common_types.h"
-#include "common/file_util.h"
+#include <array>
 
 #include <mbedtls/ctr_drbg.h>
 #include <mbedtls/entropy.h>
+
+#include "common/assert.h"
+#include "common/common_types.h"
+#include "common/file_util.h"
+#include "common/logging/log.h"
+
 #include "core/core.h"
 #include "core/file_sys/control_metadata.h"
 #include "core/file_sys/patch_manager.h"
@@ -28,11 +32,11 @@ static u64 GenerateTelemetryId() {
     mbedtls_entropy_context entropy;
     mbedtls_entropy_init(&entropy);
     mbedtls_ctr_drbg_context ctr_drbg;
-    std::string personalization = "yuzu Telemetry ID";
+    constexpr std::array<char, 18> personalization{{"yuzu Telemetry ID"}};
 
     mbedtls_ctr_drbg_init(&ctr_drbg);
     ASSERT(mbedtls_ctr_drbg_seed(&ctr_drbg, mbedtls_entropy_func, &entropy,
-                                 reinterpret_cast<const unsigned char*>(personalization.c_str()),
+                                 reinterpret_cast<const unsigned char*>(personalization.data()),
                                  personalization.size()) == 0);
     ASSERT(mbedtls_ctr_drbg_random(&ctr_drbg, reinterpret_cast<unsigned char*>(&telemetry_id),
                                    sizeof(u64)) == 0);

src/core/telemetry_session.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <memory>
+#include <string>
 #include "common/telemetry.h"
 
 namespace Core {
@@ -30,8 +31,6 @@ public:
         field_collection.AddField(type, name, std::move(value));
     }
 
-    static void FinalizeAsyncJob();
-
 private:
     Telemetry::FieldCollection field_collection; ///< Tracks all added fields for the session
     std::unique_ptr<Telemetry::VisitorInterface> backend; ///< Backend interface that logs fields
@@ -53,7 +52,6 @@ u64 RegenerateTelemetryId();
  * Verifies the username and token.
  * @param username yuzu username to use for authentication.
  * @param token yuzu token to use for authentication.
- * @param func A function that gets exectued when the verification is finished
  * @returns Future with bool indicating whether the verification succeeded
  */
 bool VerifyLogin(const std::string& username, const std::string& token);

src/tests/core/arm/arm_test_common.cpp

@@ -15,7 +15,8 @@ namespace ArmTests {
 TestEnvironment::TestEnvironment(bool mutable_memory_)
     : mutable_memory(mutable_memory_), test_memory(std::make_shared<TestMemory>(this)) {
 
-    Core::CurrentProcess() = Kernel::Process::Create(kernel, "");
+    auto process = Kernel::Process::Create(kernel, "");
+    kernel.MakeCurrentProcess(process.get());
     page_table = &Core::CurrentProcess()->VMManager().page_table;
 
     std::fill(page_table->pointers.begin(), page_table->pointers.end(), nullptr);

src/video_core/engines/fermi_2d.cpp

@@ -4,11 +4,13 @@
 
 #include "core/memory.h"
 #include "video_core/engines/fermi_2d.h"
+#include "video_core/rasterizer_interface.h"
 #include "video_core/textures/decoders.h"
 
 namespace Tegra::Engines {
 
-Fermi2D::Fermi2D(MemoryManager& memory_manager) : memory_manager(memory_manager) {}
+Fermi2D::Fermi2D(VideoCore::RasterizerInterface& rasterizer, MemoryManager& memory_manager)
+    : memory_manager(memory_manager), rasterizer{rasterizer} {}
 
 void Fermi2D::WriteReg(u32 method, u32 value) {
     ASSERT_MSG(method < Regs::NUM_REGS,
@@ -44,27 +46,31 @@ void Fermi2D::HandleSurfaceCopy() {
     u32 src_bytes_per_pixel = RenderTargetBytesPerPixel(regs.src.format);
     u32 dst_bytes_per_pixel = RenderTargetBytesPerPixel(regs.dst.format);
 
-    if (regs.src.linear == regs.dst.linear) {
-        // If the input layout and the output layout are the same, just perform a raw copy.
-        ASSERT(regs.src.BlockHeight() == regs.dst.BlockHeight());
-        Memory::CopyBlock(dest_cpu, source_cpu,
-                          src_bytes_per_pixel * regs.dst.width * regs.dst.height);
-        return;
-    }
+    if (!rasterizer.AccelerateSurfaceCopy(regs.src, regs.dst)) {
+        // TODO(bunnei): The below implementation currently will not get hit, as
+        // AccelerateSurfaceCopy tries to always copy and will always return success. This should be
+        // changed once we properly support flushing.
 
-    u8* src_buffer = Memory::GetPointer(source_cpu);
-    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());
-    } 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());
+        if (regs.src.linear == regs.dst.linear) {
+            // If the input layout and the output layout are the same, just perform a raw copy.
+            ASSERT(regs.src.BlockHeight() == regs.dst.BlockHeight());
+            Memory::CopyBlock(dest_cpu, source_cpu,
+                              src_bytes_per_pixel * regs.dst.width * regs.dst.height);
+            return;
+        }
+        u8* src_buffer = Memory::GetPointer(source_cpu);
+        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());
+        } 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());
+        }
     }
 }
 

src/video_core/engines/fermi_2d.h

@@ -12,6 +12,10 @@
 #include "video_core/gpu.h"
 #include "video_core/memory_manager.h"
 
+namespace VideoCore {
+class RasterizerInterface;
+}
+
 namespace Tegra::Engines {
 
 #define FERMI2D_REG_INDEX(field_name)                                                              \
@@ -19,7 +23,7 @@ namespace Tegra::Engines {
 
 class Fermi2D final {
 public:
-    explicit Fermi2D(MemoryManager& memory_manager);
+    explicit Fermi2D(VideoCore::RasterizerInterface& rasterizer, MemoryManager& memory_manager);
     ~Fermi2D() = default;
 
     /// Write the value to the register identified by method.
@@ -32,9 +36,9 @@ public:
             RenderTargetFormat format;
             BitField<0, 1, u32> linear;
             union {
-                BitField<0, 4, u32> block_depth;
+                BitField<0, 4, u32> block_width;
                 BitField<4, 4, u32> block_height;
-                BitField<8, 4, u32> block_width;
+                BitField<8, 4, u32> block_depth;
             };
             u32 depth;
             u32 layer;
@@ -49,10 +53,20 @@ public:
                                              address_low);
             }
 
+            u32 BlockWidth() const {
+                // The block width is stored in log2 format.
+                return 1 << block_width;
+            }
+
             u32 BlockHeight() const {
                 // The block height is stored in log2 format.
                 return 1 << block_height;
             }
+
+            u32 BlockDepth() const {
+                // The block depth is stored in log2 format.
+                return 1 << block_depth;
+            }
         };
         static_assert(sizeof(Surface) == 0x28, "Surface has incorrect size");
 
@@ -94,6 +108,8 @@ public:
     MemoryManager& memory_manager;
 
 private:
+    VideoCore::RasterizerInterface& rasterizer;
+
     /// Performs the copy from the source surface to the destination surface as configured in the
     /// registers.
     void HandleSurfaceCopy();

src/video_core/engines/maxwell_3d.h

@@ -347,6 +347,16 @@ public:
             DecrWrap = 8,
         };
 
+        enum class MemoryLayout : u32 {
+            Linear = 0,
+            BlockLinear = 1,
+        };
+
+        enum class InvMemoryLayout : u32 {
+            BlockLinear = 0,
+            Linear = 1,
+        };
+
         struct Cull {
             enum class FrontFace : u32 {
                 ClockWise = 0x0900,
@@ -432,7 +442,12 @@ public:
             u32 width;
             u32 height;
             Tegra::RenderTargetFormat format;
-            u32 block_dimensions;
+            union {
+                BitField<0, 3, u32> block_width;
+                BitField<4, 3, u32> block_height;
+                BitField<8, 3, u32> block_depth;
+                BitField<12, 1, InvMemoryLayout> type;
+            } memory_layout;
             u32 array_mode;
             u32 layer_stride;
             u32 base_layer;
@@ -532,7 +547,21 @@ public:
                 INSERT_PADDING_WORDS(0x3);
                 s32 clear_stencil;
 
-                INSERT_PADDING_WORDS(0x6C);
+                INSERT_PADDING_WORDS(0x17);
+
+                struct {
+                    u32 enable;
+                    union {
+                        BitField<0, 16, u32> min_x;
+                        BitField<16, 16, u32> max_x;
+                    };
+                    union {
+                        BitField<0, 16, u32> min_y;
+                        BitField<16, 16, u32> max_y;
+                    };
+                } scissor_test;
+
+                INSERT_PADDING_WORDS(0x52);
 
                 s32 stencil_back_func_ref;
                 u32 stencil_back_mask;
@@ -548,7 +577,12 @@ public:
                     u32 address_high;
                     u32 address_low;
                     Tegra::DepthFormat format;
-                    u32 block_dimensions;
+                    union {
+                        BitField<0, 4, u32> block_width;
+                        BitField<4, 4, u32> block_height;
+                        BitField<8, 4, u32> block_depth;
+                        BitField<20, 1, InvMemoryLayout> type;
+                    } memory_layout;
                     u32 layer_stride;
 
                     GPUVAddr Address() const {
@@ -1002,6 +1036,7 @@ ASSERT_REG_POSITION(vertex_buffer, 0x35D);
 ASSERT_REG_POSITION(clear_color[0], 0x360);
 ASSERT_REG_POSITION(clear_depth, 0x364);
 ASSERT_REG_POSITION(clear_stencil, 0x368);
+ASSERT_REG_POSITION(scissor_test, 0x380);
 ASSERT_REG_POSITION(stencil_back_func_ref, 0x3D5);
 ASSERT_REG_POSITION(stencil_back_mask, 0x3D6);
 ASSERT_REG_POSITION(stencil_back_func_mask, 0x3D7);

src/video_core/engines/shader_bytecode.h

@@ -214,6 +214,18 @@ enum class IMinMaxExchange : u64 {
     XHi = 3,
 };
 
+enum class VmadType : u64 {
+    Size16_Low = 0,
+    Size16_High = 1,
+    Size32 = 2,
+    Invalid = 3,
+};
+
+enum class VmadShr : u64 {
+    Shr7 = 1,
+    Shr15 = 2,
+};
+
 enum class XmadMode : u64 {
     None = 0,
     CLo = 1,
@@ -314,6 +326,15 @@ enum class TextureMiscMode : u64 {
     PTP,
 };
 
+enum class IsberdMode : u64 {
+    None = 0,
+    Patch = 1,
+    Prim = 2,
+    Attr = 3,
+};
+
+enum class IsberdShift : u64 { None = 0, U16 = 1, B32 = 2 };
+
 enum class IpaInterpMode : u64 {
     Linear = 0,
     Perspective = 1,
@@ -340,6 +361,87 @@ struct IpaMode {
     }
 };
 
+enum class SystemVariable : u64 {
+    LaneId = 0x00,
+    VirtCfg = 0x02,
+    VirtId = 0x03,
+    Pm0 = 0x04,
+    Pm1 = 0x05,
+    Pm2 = 0x06,
+    Pm3 = 0x07,
+    Pm4 = 0x08,
+    Pm5 = 0x09,
+    Pm6 = 0x0a,
+    Pm7 = 0x0b,
+    OrderingTicket = 0x0f,
+    PrimType = 0x10,
+    InvocationId = 0x11,
+    Ydirection = 0x12,
+    ThreadKill = 0x13,
+    ShaderType = 0x14,
+    DirectBeWriteAddressLow = 0x15,
+    DirectBeWriteAddressHigh = 0x16,
+    DirectBeWriteEnabled = 0x17,
+    MachineId0 = 0x18,
+    MachineId1 = 0x19,
+    MachineId2 = 0x1a,
+    MachineId3 = 0x1b,
+    Affinity = 0x1c,
+    InvocationInfo = 0x1d,
+    WscaleFactorXY = 0x1e,
+    WscaleFactorZ = 0x1f,
+    Tid = 0x20,
+    TidX = 0x21,
+    TidY = 0x22,
+    TidZ = 0x23,
+    CtaParam = 0x24,
+    CtaIdX = 0x25,
+    CtaIdY = 0x26,
+    CtaIdZ = 0x27,
+    NtId = 0x28,
+    CirQueueIncrMinusOne = 0x29,
+    Nlatc = 0x2a,
+    SmSpaVersion = 0x2c,
+    MultiPassShaderInfo = 0x2d,
+    LwinHi = 0x2e,
+    SwinHi = 0x2f,
+    SwinLo = 0x30,
+    SwinSz = 0x31,
+    SmemSz = 0x32,
+    SmemBanks = 0x33,
+    LwinLo = 0x34,
+    LwinSz = 0x35,
+    LmemLosz = 0x36,
+    LmemHioff = 0x37,
+    EqMask = 0x38,
+    LtMask = 0x39,
+    LeMask = 0x3a,
+    GtMask = 0x3b,
+    GeMask = 0x3c,
+    RegAlloc = 0x3d,
+    CtxAddr = 0x3e,      // .fmask = F_SM50
+    BarrierAlloc = 0x3e, // .fmask = F_SM60
+    GlobalErrorStatus = 0x40,
+    WarpErrorStatus = 0x42,
+    WarpErrorStatusClear = 0x43,
+    PmHi0 = 0x48,
+    PmHi1 = 0x49,
+    PmHi2 = 0x4a,
+    PmHi3 = 0x4b,
+    PmHi4 = 0x4c,
+    PmHi5 = 0x4d,
+    PmHi6 = 0x4e,
+    PmHi7 = 0x4f,
+    ClockLo = 0x50,
+    ClockHi = 0x51,
+    GlobalTimerLo = 0x52,
+    GlobalTimerHi = 0x53,
+    HwTaskId = 0x60,
+    CircularQueueEntryIndex = 0x61,
+    CircularQueueEntryAddressLow = 0x62,
+    CircularQueueEntryAddressHigh = 0x63,
+};
+
 union Instruction {
     Instruction& operator=(const Instruction& instr) {
         value = instr.value;
@@ -362,6 +464,7 @@ union Instruction {
     BitField<48, 16, u64> opcode;
 
     union {
+        BitField<20, 16, u64> imm20_16;
         BitField<20, 19, u64> imm20_19;
         BitField<20, 32, s64> imm20_32;
         BitField<45, 1, u64> negate_b;
@@ -403,6 +506,10 @@ union Instruction {
             }
         } lop3;
 
+        u16 GetImm20_16() const {
+            return static_cast<u16>(imm20_16);
+        }
+
         u32 GetImm20_19() const {
             u32 imm{static_cast<u32>(imm20_19)};
             imm <<= 12;
@@ -914,6 +1021,35 @@ union Instruction {
         }
     } bra;
 
+    union {
+        BitField<39, 1, u64> emit; // EmitVertex
+        BitField<40, 1, u64> cut;  // EndPrimitive
+    } out;
+
+    union {
+        BitField<31, 1, u64> skew;
+        BitField<32, 1, u64> o;
+        BitField<33, 2, IsberdMode> mode;
+        BitField<47, 2, IsberdShift> shift;
+    } isberd;
+
+    union {
+        BitField<48, 1, u64> signed_a;
+        BitField<38, 1, u64> is_byte_chunk_a;
+        BitField<36, 2, VmadType> type_a;
+        BitField<36, 2, u64> byte_height_a;
+
+        BitField<49, 1, u64> signed_b;
+        BitField<50, 1, u64> use_register_b;
+        BitField<30, 1, u64> is_byte_chunk_b;
+        BitField<28, 2, VmadType> type_b;
+        BitField<28, 2, u64> byte_height_b;
+
+        BitField<51, 2, VmadShr> shr;
+        BitField<55, 1, u64> saturate; // Saturates the result (a * b + c)
+        BitField<47, 1, u64> cc;
+    } vmad;
+
     union {
         BitField<20, 16, u64> imm20_16;
         BitField<36, 1, u64> product_shift_left;
@@ -936,6 +1072,10 @@ union Instruction {
         BitField<36, 5, u64> index;
     } cbuf36;
 
+    // Unsure about the size of this one.
+    // It's always used with a gpr0, so any size should be fine.
+    BitField<20, 8, SystemVariable> sys20;
+
     BitField<47, 1, u64> generates_cc;
     BitField<61, 1, u64> is_b_imm;
     BitField<60, 1, u64> is_b_gpr;
@@ -975,6 +1115,9 @@ public:
         TMML,   // Texture Mip Map Level
         EXIT,
         IPA,
+        OUT_R, // Emit vertex/primitive
+        ISBERD,
+        VMAD,
         FFMA_IMM, // Fused Multiply and Add
         FFMA_CR,
         FFMA_RC,
@@ -1034,6 +1177,7 @@ public:
         MOV_C,
         MOV_R,
         MOV_IMM,
+        MOV_SYS,
         MOV32_IMM,
         SHL_C,
         SHL_R,
@@ -1209,6 +1353,9 @@ private:
             INST("1101111101011---", Id::TMML, Type::Memory, "TMML"),
             INST("111000110000----", Id::EXIT, Type::Trivial, "EXIT"),
             INST("11100000--------", Id::IPA, Type::Trivial, "IPA"),
+            INST("1111101111100---", Id::OUT_R, Type::Trivial, "OUT_R"),
+            INST("1110111111010---", Id::ISBERD, Type::Trivial, "ISBERD"),
+            INST("01011111--------", Id::VMAD, Type::Trivial, "VMAD"),
             INST("0011001-1-------", Id::FFMA_IMM, Type::Ffma, "FFMA_IMM"),
             INST("010010011-------", Id::FFMA_CR, Type::Ffma, "FFMA_CR"),
             INST("010100011-------", Id::FFMA_RC, Type::Ffma, "FFMA_RC"),
@@ -1255,6 +1402,7 @@ private:
             INST("0100110010011---", Id::MOV_C, Type::Arithmetic, "MOV_C"),
             INST("0101110010011---", Id::MOV_R, Type::Arithmetic, "MOV_R"),
             INST("0011100-10011---", Id::MOV_IMM, Type::Arithmetic, "MOV_IMM"),
+            INST("1111000011001---", Id::MOV_SYS, Type::Trivial, "MOV_SYS"),
             INST("000000010000----", Id::MOV32_IMM, Type::ArithmeticImmediate, "MOV32_IMM"),
             INST("0100110001100---", Id::FMNMX_C, Type::Arithmetic, "FMNMX_C"),
             INST("0101110001100---", Id::FMNMX_R, Type::Arithmetic, "FMNMX_R"),

src/video_core/gpu.cpp

@@ -25,7 +25,7 @@ u32 FramebufferConfig::BytesPerPixel(PixelFormat format) {
 GPU::GPU(VideoCore::RasterizerInterface& rasterizer) {
     memory_manager = std::make_unique<Tegra::MemoryManager>();
     maxwell_3d = std::make_unique<Engines::Maxwell3D>(rasterizer, *memory_manager);
-    fermi_2d = std::make_unique<Engines::Fermi2D>(*memory_manager);
+    fermi_2d = std::make_unique<Engines::Fermi2D>(rasterizer, *memory_manager);
     maxwell_compute = std::make_unique<Engines::MaxwellCompute>();
     maxwell_dma = std::make_unique<Engines::MaxwellDMA>(*memory_manager);
     kepler_memory = std::make_unique<Engines::KeplerMemory>(*memory_manager);

src/video_core/rasterizer_interface.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include "common/common_types.h"
+#include "video_core/engines/fermi_2d.h"
 #include "video_core/gpu.h"
 #include "video_core/memory_manager.h"
 
@@ -33,13 +34,9 @@ public:
     /// and invalidated
     virtual void FlushAndInvalidateRegion(VAddr addr, u64 size) = 0;
 
-    /// Attempt to use a faster method to perform a display transfer with is_texture_copy = 0
-    virtual bool AccelerateDisplayTransfer(const void* config) {
-        return false;
-    }
-
-    /// Attempt to use a faster method to perform a display transfer with is_texture_copy = 1
-    virtual bool AccelerateTextureCopy(const void* config) {
+    /// Attempt to use a faster method to perform a surface copy
+    virtual bool AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
+                                       const Tegra::Engines::Fermi2D::Regs::Surface& dst) {
         return false;
     }
 

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -255,7 +255,7 @@ DrawParameters RasterizerOpenGL::SetupDraw() {
     return params;
 }
 
-void RasterizerOpenGL::SetupShaders() {
+void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
     MICROPROFILE_SCOPE(OpenGL_Shader);
     const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
 
@@ -270,6 +270,11 @@ void RasterizerOpenGL::SetupShaders() {
 
         // Skip stages that are not enabled
         if (!gpu.regs.IsShaderConfigEnabled(index)) {
+            switch (program) {
+            case Maxwell::ShaderProgram::Geometry:
+                shader_program_manager->UseTrivialGeometryShader();
+                break;
+            }
             continue;
         }
 
@@ -288,11 +293,18 @@ void RasterizerOpenGL::SetupShaders() {
         switch (program) {
         case Maxwell::ShaderProgram::VertexA:
         case Maxwell::ShaderProgram::VertexB: {
-            shader_program_manager->UseProgrammableVertexShader(shader->GetProgramHandle());
+            shader_program_manager->UseProgrammableVertexShader(
+                shader->GetProgramHandle(primitive_mode));
+            break;
+        }
+        case Maxwell::ShaderProgram::Geometry: {
+            shader_program_manager->UseProgrammableGeometryShader(
+                shader->GetProgramHandle(primitive_mode));
             break;
         }
         case Maxwell::ShaderProgram::Fragment: {
-            shader_program_manager->UseProgrammableFragmentShader(shader->GetProgramHandle());
+            shader_program_manager->UseProgrammableFragmentShader(
+                shader->GetProgramHandle(primitive_mode));
             break;
         }
         default:
@@ -302,12 +314,13 @@ void RasterizerOpenGL::SetupShaders() {
         }
 
         // Configure the const buffers for this shader stage.
-        current_constbuffer_bindpoint = SetupConstBuffers(static_cast<Maxwell::ShaderStage>(stage),
-                                                          shader, current_constbuffer_bindpoint);
+        current_constbuffer_bindpoint =
+            SetupConstBuffers(static_cast<Maxwell::ShaderStage>(stage), shader, primitive_mode,
+                              current_constbuffer_bindpoint);
 
         // Configure the textures for this shader stage.
         current_texture_bindpoint = SetupTextures(static_cast<Maxwell::ShaderStage>(stage), shader,
-                                                  current_texture_bindpoint);
+                                                  primitive_mode, current_texture_bindpoint);
 
         // When VertexA is enabled, we have dual vertex shaders
         if (program == Maxwell::ShaderProgram::VertexA) {
@@ -317,8 +330,6 @@ void RasterizerOpenGL::SetupShaders() {
     }
 
     state.Apply();
-
-    shader_program_manager->UseTrivialGeometryShader();
 }
 
 std::size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
@@ -541,6 +552,7 @@ void RasterizerOpenGL::DrawArrays() {
     SyncLogicOpState();
     SyncCullMode();
     SyncAlphaTest();
+    SyncScissorTest();
     SyncTransformFeedback();
     SyncPointState();
 
@@ -580,7 +592,7 @@ void RasterizerOpenGL::DrawArrays() {
 
     SetupVertexArrays();
     DrawParameters params = SetupDraw();
-    SetupShaders();
+    SetupShaders(params.primitive_mode);
 
     buffer_cache.Unmap();
 
@@ -617,14 +629,10 @@ void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) {
     InvalidateRegion(addr, size);
 }
 
-bool RasterizerOpenGL::AccelerateDisplayTransfer(const void* config) {
+bool RasterizerOpenGL::AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
+                                             const Tegra::Engines::Fermi2D::Regs::Surface& dst) {
     MICROPROFILE_SCOPE(OpenGL_Blits);
-    UNREACHABLE();
-    return true;
-}
-
-bool RasterizerOpenGL::AccelerateTextureCopy(const void* config) {
-    UNREACHABLE();
+    res_cache.FermiCopySurface(src, dst);
     return true;
 }
 
@@ -662,10 +670,13 @@ void RasterizerOpenGL::SamplerInfo::Create() {
     sampler.Create();
     mag_filter = min_filter = Tegra::Texture::TextureFilter::Linear;
     wrap_u = wrap_v = wrap_p = Tegra::Texture::WrapMode::Wrap;
+    uses_depth_compare = false;
+    depth_compare_func = Tegra::Texture::DepthCompareFunc::Never;
 
     // default is GL_LINEAR_MIPMAP_LINEAR
     glSamplerParameteri(sampler.handle, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
     // Other attributes have correct defaults
+    glSamplerParameteri(sampler.handle, GL_TEXTURE_COMPARE_FUNC, GL_NEVER);
 }
 
 void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntry& config) {
@@ -693,6 +704,21 @@ void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntr
         glSamplerParameteri(s, GL_TEXTURE_WRAP_R, MaxwellToGL::WrapMode(wrap_p));
     }
 
+    if (uses_depth_compare != (config.depth_compare_enabled == 1)) {
+        uses_depth_compare = (config.depth_compare_enabled == 1);
+        if (uses_depth_compare) {
+            glSamplerParameteri(s, GL_TEXTURE_COMPARE_MODE, GL_COMPARE_REF_TO_TEXTURE);
+        } else {
+            glSamplerParameteri(s, GL_TEXTURE_COMPARE_MODE, GL_NONE);
+        }
+    }
+
+    if (depth_compare_func != config.depth_compare_func) {
+        depth_compare_func = config.depth_compare_func;
+        glSamplerParameteri(s, GL_TEXTURE_COMPARE_FUNC,
+                            MaxwellToGL::DepthCompareFunc(depth_compare_func));
+    }
+
     if (wrap_u == Tegra::Texture::WrapMode::Border || wrap_v == Tegra::Texture::WrapMode::Border ||
         wrap_p == Tegra::Texture::WrapMode::Border) {
         const GLvec4 new_border_color = {{config.border_color_r, config.border_color_g,
@@ -705,7 +731,7 @@ void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntr
 }
 
 u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shader,
-                                        u32 current_bindpoint) {
+                                        GLenum primitive_mode, u32 current_bindpoint) {
     MICROPROFILE_SCOPE(OpenGL_UBO);
     const auto& gpu = Core::System::GetInstance().GPU();
     const auto& maxwell3d = gpu.Maxwell3D();
@@ -757,7 +783,7 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shad
             buffer.address, size, static_cast<std::size_t>(uniform_buffer_alignment));
 
         // Now configure the bindpoint of the buffer inside the shader
-        glUniformBlockBinding(shader->GetProgramHandle(),
+        glUniformBlockBinding(shader->GetProgramHandle(primitive_mode),
                               shader->GetProgramResourceIndex(used_buffer),
                               current_bindpoint + bindpoint);
 
@@ -773,7 +799,8 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shad
     return current_bindpoint + static_cast<u32>(entries.size());
 }
 
-u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader, u32 current_unit) {
+u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader,
+                                    GLenum primitive_mode, u32 current_unit) {
     MICROPROFILE_SCOPE(OpenGL_Texture);
     const auto& gpu = Core::System::GetInstance().GPU();
     const auto& maxwell3d = gpu.Maxwell3D();
@@ -788,8 +815,8 @@ u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader,
 
         // Bind the uniform to the sampler.
 
-        glProgramUniform1i(shader->GetProgramHandle(), shader->GetUniformLocation(entry),
-                           current_bindpoint);
+        glProgramUniform1i(shader->GetProgramHandle(primitive_mode),
+                           shader->GetUniformLocation(entry), current_bindpoint);
 
         const auto texture = maxwell3d.GetStageTexture(entry.GetStage(), entry.GetOffset());
 
@@ -958,6 +985,22 @@ void RasterizerOpenGL::SyncAlphaTest() {
     }
 }
 
+void RasterizerOpenGL::SyncScissorTest() {
+    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+
+    state.scissor.enabled = (regs.scissor_test.enable != 0);
+    // TODO(Blinkhawk): Figure if the hardware supports scissor testing per viewport and how it's
+    // implemented.
+    if (regs.scissor_test.enable != 0) {
+        const u32 width = regs.scissor_test.max_x - regs.scissor_test.min_x;
+        const u32 height = regs.scissor_test.max_y - regs.scissor_test.min_y;
+        state.scissor.x = regs.scissor_test.min_x;
+        state.scissor.y = regs.scissor_test.min_y;
+        state.scissor.width = width;
+        state.scissor.height = height;
+    }
+}
+
 void RasterizerOpenGL::SyncTransformFeedback() {
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -52,8 +52,8 @@ public:
     void FlushRegion(VAddr addr, u64 size) override;
     void InvalidateRegion(VAddr addr, u64 size) override;
     void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
-    bool AccelerateDisplayTransfer(const void* config) override;
-    bool AccelerateTextureCopy(const void* config) override;
+    bool AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
+                               const Tegra::Engines::Fermi2D::Regs::Surface& dst) override;
     bool AccelerateFill(const void* config) override;
     bool AccelerateDisplay(const Tegra::FramebufferConfig& config, VAddr framebuffer_addr,
                            u32 pixel_stride) override;
@@ -96,6 +96,8 @@ private:
         Tegra::Texture::WrapMode wrap_u;
         Tegra::Texture::WrapMode wrap_v;
         Tegra::Texture::WrapMode wrap_p;
+        bool uses_depth_compare;
+        Tegra::Texture::DepthCompareFunc depth_compare_func;
         GLvec4 border_color;
     };
 
@@ -118,7 +120,7 @@ private:
      * @returns The next available bindpoint for use in the next shader stage.
      */
     u32 SetupConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, Shader& shader,
-                          u32 current_bindpoint);
+                          GLenum primitive_mode, u32 current_bindpoint);
 
     /*
      * Configures the current textures to use for the draw command.
@@ -128,7 +130,7 @@ private:
      * @returns The next available bindpoint for use in the next shader stage.
      */
     u32 SetupTextures(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, Shader& shader,
-                      u32 current_unit);
+                      GLenum primitive_mode, u32 current_unit);
 
     /// Syncs the viewport to match the guest state
     void SyncViewport();
@@ -163,6 +165,9 @@ private:
     /// Syncs the alpha test state to match the guest state
     void SyncAlphaTest();
 
+    /// Syncs the scissor test state to match the guest state
+    void SyncScissorTest();
+
     /// Syncs the transform feedback state to match the guest state
     void SyncTransformFeedback();
 
@@ -205,7 +210,7 @@ private:
 
     DrawParameters SetupDraw();
 
-    void SetupShaders();
+    void SetupShaders(GLenum primitive_mode);
 
     enum class AccelDraw { Disabled, Arrays, Indexed };
     AccelDraw accelerate_draw = AccelDraw::Disabled;

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -45,7 +45,9 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     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,
+    params.block_depth = params.is_tiled ? config.tic.BlockDepth() : 0,
     params.pixel_format =
         PixelFormatFromTextureFormat(config.tic.format, config.tic.r_type.Value());
     params.component_type = ComponentTypeFromTexture(config.tic.r_type.Value());
@@ -97,8 +99,11 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     const auto& config{Core::System::GetInstance().GPU().Maxwell3D().regs.rt[index]};
     SurfaceParams params{};
     params.addr = TryGetCpuAddr(config.Address());
-    params.is_tiled = true;
-    params.block_height = Tegra::Texture::TICEntry::DefaultBlockHeight;
+    params.is_tiled =
+        config.memory_layout.type == Tegra::Engines::Maxwell3D::Regs::InvMemoryLayout::BlockLinear;
+    params.block_width = 1 << config.memory_layout.block_width;
+    params.block_height = 1 << config.memory_layout.block_height;
+    params.block_depth = 1 << config.memory_layout.block_depth;
     params.pixel_format = PixelFormatFromRenderTargetFormat(config.format);
     params.component_type = ComponentTypeFromRenderTarget(config.format);
     params.type = GetFormatType(params.pixel_format);
@@ -120,13 +125,16 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     return params;
 }
 
-/*static*/ SurfaceParams SurfaceParams::CreateForDepthBuffer(u32 zeta_width, u32 zeta_height,
-                                                             Tegra::GPUVAddr zeta_address,
-                                                             Tegra::DepthFormat format) {
+/*static*/ SurfaceParams SurfaceParams::CreateForDepthBuffer(
+    u32 zeta_width, u32 zeta_height, Tegra::GPUVAddr zeta_address, Tegra::DepthFormat format,
+    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 = true;
-    params.block_height = Tegra::Texture::TICEntry::DefaultBlockHeight;
+    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);
+    params.block_depth = 1 << std::min(block_depth, 5U);
     params.pixel_format = PixelFormatFromDepthFormat(format);
     params.component_type = ComponentTypeFromDepthFormat(format);
     params.type = GetFormatType(params.pixel_format);
@@ -143,6 +151,30 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     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,
+    params.block_depth = params.is_tiled ? std::min(config.BlockDepth(), 32U) : 0,
+    params.pixel_format = PixelFormatFromRenderTargetFormat(config.format);
+    params.component_type = ComponentTypeFromRenderTarget(config.format);
+    params.type = GetFormatType(params.pixel_format);
+    params.width = config.width;
+    params.height = config.height;
+    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 = {};
+
+    return params;
+}
+
 static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_format_tuples = {{
     {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, ComponentType::UNorm, false}, // ABGR8U
     {GL_RGBA8, GL_RGBA, GL_BYTE, ComponentType::SNorm, false},                     // ABGR8S
@@ -559,6 +591,18 @@ static bool BlitSurface(const Surface& src_surface, const Surface& dst_surface,
     return true;
 }
 
+static void FastCopySurface(const Surface& src_surface, const Surface& dst_surface) {
+    const auto& src_params{src_surface->GetSurfaceParams()};
+    const auto& dst_params{dst_surface->GetSurfaceParams()};
+
+    const u32 width{std::min(src_params.width, dst_params.width)};
+    const u32 height{std::min(src_params.height, dst_params.height)};
+
+    glCopyImageSubData(src_surface->Texture().handle, SurfaceTargetToGL(src_params.target), 0, 0, 0,
+                       0, dst_surface->Texture().handle, SurfaceTargetToGL(dst_params.target), 0, 0,
+                       0, 0, width, height, 1);
+}
+
 static void CopySurface(const Surface& src_surface, const Surface& dst_surface,
                         GLuint copy_pbo_handle, GLenum src_attachment = 0,
                         GLenum dst_attachment = 0, std::size_t cubemap_face = 0) {
@@ -784,6 +828,11 @@ void CachedSurface::LoadGLBuffer() {
     if (params.is_tiled) {
         gl_buffer.resize(total_size);
 
+        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) {
@@ -955,7 +1004,9 @@ Surface RasterizerCacheOpenGL::GetDepthBufferSurface(bool preserve_contents) {
     }
 
     SurfaceParams depth_params{SurfaceParams::CreateForDepthBuffer(
-        regs.zeta_width, regs.zeta_height, regs.zeta.Address(), regs.zeta.format)};
+        regs.zeta_width, regs.zeta_height, regs.zeta.Address(), regs.zeta.format,
+        regs.zeta.memory_layout.block_width, regs.zeta.memory_layout.block_height,
+        regs.zeta.memory_layout.block_depth, regs.zeta.memory_layout.type)};
 
     return GetSurface(depth_params, preserve_contents);
 }
@@ -1033,6 +1084,26 @@ Surface RasterizerCacheOpenGL::GetUncachedSurface(const SurfaceParams& params) {
     return surface;
 }
 
+void RasterizerCacheOpenGL::FermiCopySurface(
+    const Tegra::Engines::Fermi2D::Regs::Surface& src_config,
+    const Tegra::Engines::Fermi2D::Regs::Surface& dst_config) {
+
+    const auto& src_params = SurfaceParams::CreateForFermiCopySurface(src_config);
+    const auto& dst_params = SurfaceParams::CreateForFermiCopySurface(dst_config);
+
+    ASSERT(src_params.width == dst_params.width);
+    ASSERT(src_params.height == dst_params.height);
+    ASSERT(src_params.pixel_format == dst_params.pixel_format);
+    ASSERT(src_params.block_height == dst_params.block_height);
+    ASSERT(src_params.is_tiled == dst_params.is_tiled);
+    ASSERT(src_params.depth == dst_params.depth);
+    ASSERT(src_params.depth == 1); // Currently, FastCopySurface only works with 2D surfaces
+    ASSERT(src_params.target == dst_params.target);
+    ASSERT(src_params.rt.index == dst_params.rt.index);
+
+    FastCopySurface(GetSurface(src_params, true), GetSurface(dst_params, false));
+}
+
 Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& old_surface,
                                                const SurfaceParams& new_params) {
     // Verify surface is compatible for blitting
@@ -1041,6 +1112,15 @@ 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)};
 
+    // 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 &&
+        SurfaceParams::GetFormatBpp(old_params.pixel_format) ==
+            SurfaceParams::GetFormatBpp(new_params.pixel_format)) {
+        FastCopySurface(old_surface, new_surface);
+        return new_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

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -13,6 +13,7 @@
 #include "common/common_types.h"
 #include "common/hash.h"
 #include "common/math_util.h"
+#include "video_core/engines/fermi_2d.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
@@ -715,9 +716,14 @@ struct SurfaceParams {
     static SurfaceParams CreateForFramebuffer(std::size_t index);
 
     /// Creates SurfaceParams for a depth buffer configuration
-    static SurfaceParams CreateForDepthBuffer(u32 zeta_width, u32 zeta_height,
-                                              Tegra::GPUVAddr zeta_address,
-                                              Tegra::DepthFormat format);
+    static SurfaceParams CreateForDepthBuffer(
+        u32 zeta_width, u32 zeta_height, Tegra::GPUVAddr zeta_address, Tegra::DepthFormat format,
+        u32 block_width, u32 block_height, u32 block_depth,
+        Tegra::Engines::Maxwell3D::Regs::InvMemoryLayout type);
+
+    /// Creates SurfaceParams for a Fermi2D surface copy
+    static SurfaceParams CreateForFermiCopySurface(
+        const Tegra::Engines::Fermi2D::Regs::Surface& config);
 
     /// Checks if surfaces are compatible for caching
     bool IsCompatibleSurface(const SurfaceParams& other) const {
@@ -728,7 +734,9 @@ struct SurfaceParams {
 
     VAddr addr;
     bool is_tiled;
+    u32 block_width;
     u32 block_height;
+    u32 block_depth;
     PixelFormat pixel_format;
     ComponentType component_type;
     SurfaceType type;
@@ -837,6 +845,10 @@ public:
     /// Tries to find a framebuffer using on the provided CPU address
     Surface TryFindFramebufferSurface(VAddr addr) const;
 
+    /// Copies the contents of one surface to another
+    void FermiCopySurface(const Tegra::Engines::Fermi2D::Regs::Surface& src_config,
+                          const Tegra::Engines::Fermi2D::Regs::Surface& dst_config);
+
 private:
     void LoadSurface(const Surface& surface);
     Surface GetSurface(const SurfaceParams& params, bool preserve_contents = true);

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -68,6 +68,10 @@ CachedShader::CachedShader(VAddr addr, Maxwell::ShaderProgram program_type)
         program_result = GLShader::GenerateVertexShader(setup);
         gl_type = GL_VERTEX_SHADER;
         break;
+    case Maxwell::ShaderProgram::Geometry:
+        program_result = GLShader::GenerateGeometryShader(setup);
+        gl_type = GL_GEOMETRY_SHADER;
+        break;
     case Maxwell::ShaderProgram::Fragment:
         program_result = GLShader::GenerateFragmentShader(setup);
         gl_type = GL_FRAGMENT_SHADER;
@@ -80,11 +84,16 @@ CachedShader::CachedShader(VAddr addr, Maxwell::ShaderProgram program_type)
 
     entries = program_result.second;
 
-    OGLShader shader;
-    shader.Create(program_result.first.c_str(), gl_type);
-    program.Create(true, shader.handle);
-    SetShaderUniformBlockBindings(program.handle);
-    VideoCore::LabelGLObject(GL_PROGRAM, program.handle, addr);
+    if (program_type != Maxwell::ShaderProgram::Geometry) {
+        OGLShader shader;
+        shader.Create(program_result.first.c_str(), gl_type);
+        program.Create(true, shader.handle);
+        SetShaderUniformBlockBindings(program.handle);
+        VideoCore::LabelGLObject(GL_PROGRAM, program.handle, addr);
+    } else {
+        // Store shader's code to lazily build it on draw
+        geometry_programs.code = program_result.first;
+    }
 }
 
 GLuint CachedShader::GetProgramResourceIndex(const GLShader::ConstBufferEntry& buffer) {
@@ -110,6 +119,21 @@ GLint CachedShader::GetUniformLocation(const GLShader::SamplerEntry& sampler) {
     return search->second;
 }
 
+GLuint CachedShader::LazyGeometryProgram(OGLProgram& target_program,
+                                         const std::string& glsl_topology,
+                                         const std::string& debug_name) {
+    if (target_program.handle != 0) {
+        return target_program.handle;
+    }
+    const std::string source{geometry_programs.code + "layout (" + glsl_topology + ") in;\n"};
+    OGLShader shader;
+    shader.Create(source.c_str(), GL_GEOMETRY_SHADER);
+    target_program.Create(true, shader.handle);
+    SetShaderUniformBlockBindings(target_program.handle);
+    VideoCore::LabelGLObject(GL_PROGRAM, target_program.handle, addr, debug_name);
+    return target_program.handle;
+};
+
 Shader ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program) {
     const VAddr program_addr{GetShaderAddress(program)};
 

src/video_core/renderer_opengl/gl_shader_cache.h

@@ -7,6 +7,7 @@
 #include <map>
 #include <memory>
 
+#include "common/assert.h"
 #include "common/common_types.h"
 #include "video_core/rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
@@ -38,8 +39,31 @@ public:
     }
 
     /// Gets the GL program handle for the shader
-    GLuint GetProgramHandle() const {
-        return program.handle;
+    GLuint GetProgramHandle(GLenum primitive_mode) {
+        if (program_type != Maxwell::ShaderProgram::Geometry) {
+            return program.handle;
+        }
+        switch (primitive_mode) {
+        case GL_POINTS:
+            return LazyGeometryProgram(geometry_programs.points, "points", "ShaderPoints");
+        case GL_LINES:
+        case GL_LINE_STRIP:
+            return LazyGeometryProgram(geometry_programs.lines, "lines", "ShaderLines");
+        case GL_LINES_ADJACENCY:
+        case GL_LINE_STRIP_ADJACENCY:
+            return LazyGeometryProgram(geometry_programs.lines_adjacency, "lines_adjacency",
+                                       "ShaderLinesAdjacency");
+        case GL_TRIANGLES:
+        case GL_TRIANGLE_STRIP:
+        case GL_TRIANGLE_FAN:
+            return LazyGeometryProgram(geometry_programs.triangles, "triangles", "ShaderTriangles");
+        case GL_TRIANGLES_ADJACENCY:
+        case GL_TRIANGLE_STRIP_ADJACENCY:
+            return LazyGeometryProgram(geometry_programs.triangles_adjacency, "triangles_adjacency",
+                                       "ShaderLines");
+        default:
+            UNREACHABLE_MSG("Unknown primitive mode.");
+        }
     }
 
     /// Gets the GL program resource location for the specified resource, caching as needed
@@ -49,12 +73,30 @@ public:
     GLint GetUniformLocation(const GLShader::SamplerEntry& sampler);
 
 private:
+    /// Generates a geometry shader or returns one that already exists.
+    GLuint LazyGeometryProgram(OGLProgram& target_program, const std::string& glsl_topology,
+                               const std::string& debug_name);
+
     VAddr addr;
     Maxwell::ShaderProgram program_type;
     GLShader::ShaderSetup setup;
     GLShader::ShaderEntries entries;
+
+    // Non-geometry program.
     OGLProgram program;
 
+    // Geometry programs. These are needed because GLSL needs an input topology but it's not
+    // declared by the hardware. Workaround this issue by generating a different shader per input
+    // topology class.
+    struct {
+        std::string code;
+        OGLProgram points;
+        OGLProgram lines;
+        OGLProgram lines_adjacency;
+        OGLProgram triangles;
+        OGLProgram triangles_adjacency;
+    } geometry_programs;
+
     std::map<u32, GLuint> resource_cache;
     std::map<u32, GLint> uniform_cache;
 };

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -7,6 +7,7 @@
 #include <string>
 #include <string_view>
 
+#include <boost/optional.hpp>
 #include <fmt/format.h>
 
 #include "common/assert.h"
@@ -29,11 +30,32 @@ using Tegra::Shader::SubOp;
 constexpr u32 PROGRAM_END = MAX_PROGRAM_CODE_LENGTH;
 constexpr u32 PROGRAM_HEADER_SIZE = sizeof(Tegra::Shader::Header);
 
+enum : u32 { POSITION_VARYING_LOCATION = 0, GENERIC_VARYING_START_LOCATION = 1 };
+
+constexpr u32 MAX_GEOMETRY_BUFFERS = 6;
+constexpr u32 MAX_ATTRIBUTES = 0x100; // Size in vec4s, this value is untested
+
 class DecompileFail : public std::runtime_error {
 public:
     using std::runtime_error::runtime_error;
 };
 
+/// Translate topology
+static std::string GetTopologyName(Tegra::Shader::OutputTopology topology) {
+    switch (topology) {
+    case Tegra::Shader::OutputTopology::PointList:
+        return "points";
+    case Tegra::Shader::OutputTopology::LineStrip:
+        return "line_strip";
+    case Tegra::Shader::OutputTopology::TriangleStrip:
+        return "triangle_strip";
+    default:
+        LOG_CRITICAL(Render_OpenGL, "Unknown output topology {}", static_cast<u32>(topology));
+        UNREACHABLE();
+        return "points";
+    }
+}
+
 /// Describes the behaviour of code path of a given entry point and a return point.
 enum class ExitMethod {
     Undetermined, ///< Internal value. Only occur when analyzing JMP loop.
@@ -253,8 +275,9 @@ enum class InternalFlag : u64 {
 class GLSLRegisterManager {
 public:
     GLSLRegisterManager(ShaderWriter& shader, ShaderWriter& declarations,
-                        const Maxwell3D::Regs::ShaderStage& stage, const std::string& suffix)
-        : shader{shader}, declarations{declarations}, stage{stage}, suffix{suffix} {
+                        const Maxwell3D::Regs::ShaderStage& stage, const std::string& suffix,
+                        const Tegra::Shader::Header& header)
+        : shader{shader}, declarations{declarations}, stage{stage}, suffix{suffix}, header{header} {
         BuildRegisterList();
         BuildInputList();
     }
@@ -358,11 +381,13 @@ public:
      * @param reg The destination register to use.
      * @param elem The element to use for the operation.
      * @param attribute The input attribute to use as the source value.
+     * @param vertex The register that decides which vertex to read from (used in GS).
      */
     void SetRegisterToInputAttibute(const Register& reg, u64 elem, Attribute::Index attribute,
-                                    const Tegra::Shader::IpaMode& input_mode) {
+                                    const Tegra::Shader::IpaMode& input_mode,
+                                    boost::optional<Register> vertex = {}) {
         const std::string dest = GetRegisterAsFloat(reg);
-        const std::string src = GetInputAttribute(attribute, input_mode) + GetSwizzle(elem);
+        const std::string src = GetInputAttribute(attribute, input_mode, vertex) + GetSwizzle(elem);
         shader.AddLine(dest + " = " + src + ';');
     }
 
@@ -391,16 +416,29 @@ public:
      * are stored as floats, so this may require conversion.
      * @param attribute The destination output attribute.
      * @param elem The element to use for the operation.
-     * @param reg The register to use as the source value.
+     * @param val_reg The register to use as the source value.
+     * @param buf_reg The register that tells which buffer to write to (used in geometry shaders).
      */
-    void SetOutputAttributeToRegister(Attribute::Index attribute, u64 elem, const Register& reg) {
+    void SetOutputAttributeToRegister(Attribute::Index attribute, u64 elem, const Register& val_reg,
+                                      const Register& buf_reg) {
         const std::string dest = GetOutputAttribute(attribute);
-        const std::string src = GetRegisterAsFloat(reg);
+        const std::string src = GetRegisterAsFloat(val_reg);
 
         if (!dest.empty()) {
             // Can happen with unknown/unimplemented output attributes, in which case we ignore the
             // instruction for now.
-            shader.AddLine(dest + GetSwizzle(elem) + " = " + src + ';');
+            if (stage == Maxwell3D::Regs::ShaderStage::Geometry) {
+                // TODO(Rodrigo): nouveau sets some attributes after setting emitting a geometry
+                // shader. These instructions use a dirty register as buffer index. To avoid some
+                // drivers from complaining for the out of boundary writes, guard them.
+                const std::string buf_index{"min(" + GetRegisterAsInteger(buf_reg) + ", " +
+                                            std::to_string(MAX_GEOMETRY_BUFFERS - 1) + ')'};
+                shader.AddLine("amem[" + buf_index + "][" +
+                               std::to_string(static_cast<u32>(attribute)) + ']' +
+                               GetSwizzle(elem) + " = " + src + ';');
+            } else {
+                shader.AddLine(dest + GetSwizzle(elem) + " = " + src + ';');
+            }
         }
     }
 
@@ -441,41 +479,123 @@ public:
         }
     }
 
-    /// Add declarations for registers
+    /// Add declarations.
     void GenerateDeclarations(const std::string& suffix) {
+        GenerateRegisters(suffix);
+        GenerateInternalFlags();
+        GenerateInputAttrs();
+        GenerateOutputAttrs();
+        GenerateConstBuffers();
+        GenerateSamplers();
+        GenerateGeometry();
+    }
+
+    /// Returns a list of constant buffer declarations.
+    std::vector<ConstBufferEntry> GetConstBuffersDeclarations() const {
+        std::vector<ConstBufferEntry> result;
+        std::copy_if(declr_const_buffers.begin(), declr_const_buffers.end(),
+                     std::back_inserter(result), [](const auto& entry) { return entry.IsUsed(); });
+        return result;
+    }
+
+    /// Returns a list of samplers used in the shader.
+    const std::vector<SamplerEntry>& GetSamplers() const {
+        return used_samplers;
+    }
+
+    /// Returns the GLSL sampler used for the input shader sampler, and creates a new one if
+    /// necessary.
+    std::string AccessSampler(const Sampler& sampler, Tegra::Shader::TextureType type,
+                              bool is_array, bool is_shadow) {
+        const auto offset = static_cast<std::size_t>(sampler.index.Value());
+
+        // If this sampler has already been used, return the existing mapping.
+        const auto itr =
+            std::find_if(used_samplers.begin(), used_samplers.end(),
+                         [&](const SamplerEntry& entry) { return entry.GetOffset() == offset; });
+
+        if (itr != used_samplers.end()) {
+            ASSERT(itr->GetType() == type && itr->IsArray() == is_array &&
+                   itr->IsShadow() == is_shadow);
+            return itr->GetName();
+        }
+
+        // Otherwise create a new mapping for this sampler
+        const std::size_t next_index = used_samplers.size();
+        const SamplerEntry entry{stage, offset, next_index, type, is_array, is_shadow};
+        used_samplers.emplace_back(entry);
+        return entry.GetName();
+    }
+
+private:
+    /// Generates declarations for registers.
+    void GenerateRegisters(const std::string& suffix) {
         for (const auto& reg : regs) {
             declarations.AddLine(GLSLRegister::GetTypeString() + ' ' + reg.GetPrefixString() +
                                  std::to_string(reg.GetIndex()) + '_' + suffix + " = 0;");
         }
         declarations.AddNewLine();
+    }
 
+    /// Generates declarations for internal flags.
+    void GenerateInternalFlags() {
         for (u32 ii = 0; ii < static_cast<u64>(InternalFlag::Amount); ii++) {
             const InternalFlag code = static_cast<InternalFlag>(ii);
             declarations.AddLine("bool " + GetInternalFlag(code) + " = false;");
         }
         declarations.AddNewLine();
+    }
+
+    /// Generates declarations for input attributes.
+    void GenerateInputAttrs() {
+        if (stage != Maxwell3D::Regs::ShaderStage::Vertex) {
+            const std::string attr =
+                stage == Maxwell3D::Regs::ShaderStage::Geometry ? "gs_position[]" : "position";
+            declarations.AddLine("layout (location = " + std::to_string(POSITION_VARYING_LOCATION) +
+                                 ") in vec4 " + attr + ';');
+        }
 
         for (const auto element : declr_input_attribute) {
             // TODO(bunnei): Use proper number of elements for these
             u32 idx =
                 static_cast<u32>(element.first) - static_cast<u32>(Attribute::Index::Attribute_0);
-            declarations.AddLine("layout(location = " + std::to_string(idx) + ")" +
-                                 GetInputFlags(element.first) + "in vec4 " +
-                                 GetInputAttribute(element.first, element.second) + ';');
-        }
-        declarations.AddNewLine();
+            if (stage != Maxwell3D::Regs::ShaderStage::Vertex) {
+                // If inputs are varyings, add an offset
+                idx += GENERIC_VARYING_START_LOCATION;
+            }
 
+            std::string attr{GetInputAttribute(element.first, element.second)};
+            if (stage == Maxwell3D::Regs::ShaderStage::Geometry) {
+                attr = "gs_" + attr + "[]";
+            }
+            declarations.AddLine("layout (location = " + std::to_string(idx) + ") " +
+                                 GetInputFlags(element.first) + "in vec4 " + attr + ';');
+        }
+
+        declarations.AddNewLine();
+    }
+
+    /// Generates declarations for output attributes.
+    void GenerateOutputAttrs() {
+        if (stage != Maxwell3D::Regs::ShaderStage::Fragment) {
+            declarations.AddLine("layout (location = " + std::to_string(POSITION_VARYING_LOCATION) +
+                                 ") out vec4 position;");
+        }
         for (const auto& index : declr_output_attribute) {
             // TODO(bunnei): Use proper number of elements for these
-            declarations.AddLine("layout(location = " +
-                                 std::to_string(static_cast<u32>(index) -
-                                                static_cast<u32>(Attribute::Index::Attribute_0)) +
-                                 ") out vec4 " + GetOutputAttribute(index) + ';');
+            const u32 idx = static_cast<u32>(index) -
+                            static_cast<u32>(Attribute::Index::Attribute_0) +
+                            GENERIC_VARYING_START_LOCATION;
+            declarations.AddLine("layout (location = " + std::to_string(idx) + ") out vec4 " +
+                                 GetOutputAttribute(index) + ';');
         }
         declarations.AddNewLine();
+    }
 
+    /// Generates declarations for constant buffers.
+    void GenerateConstBuffers() {
         for (const auto& entry : GetConstBuffersDeclarations()) {
-            declarations.AddLine("layout(std140) uniform " + entry.GetName());
+            declarations.AddLine("layout (std140) uniform " + entry.GetName());
             declarations.AddLine('{');
             declarations.AddLine("    vec4 c" + std::to_string(entry.GetIndex()) +
                                  "[MAX_CONSTBUFFER_ELEMENTS];");
@@ -483,7 +603,10 @@ public:
             declarations.AddNewLine();
         }
         declarations.AddNewLine();
+    }
 
+    /// Generates declarations for samplers.
+    void GenerateSamplers() {
         const auto& samplers = GetSamplers();
         for (const auto& sampler : samplers) {
             declarations.AddLine("uniform " + sampler.GetTypeString() + ' ' + sampler.GetName() +
@@ -492,43 +615,42 @@ public:
         declarations.AddNewLine();
     }
 
-    /// Returns a list of constant buffer declarations
-    std::vector<ConstBufferEntry> GetConstBuffersDeclarations() const {
-        std::vector<ConstBufferEntry> result;
-        std::copy_if(declr_const_buffers.begin(), declr_const_buffers.end(),
-                     std::back_inserter(result), [](const auto& entry) { return entry.IsUsed(); });
-        return result;
-    }
+    /// Generates declarations used for geometry shaders.
+    void GenerateGeometry() {
+        if (stage != Maxwell3D::Regs::ShaderStage::Geometry)
+            return;
 
-    /// Returns a list of samplers used in the shader
-    const std::vector<SamplerEntry>& GetSamplers() const {
-        return used_samplers;
-    }
+        declarations.AddLine(
+            "layout (" + GetTopologyName(header.common3.output_topology) +
+            ", max_vertices = " + std::to_string(header.common4.max_output_vertices) + ") out;");
+        declarations.AddNewLine();
 
-    /// Returns the GLSL sampler used for the input shader sampler, and creates a new one if
-    /// necessary.
-    std::string AccessSampler(const Sampler& sampler, Tegra::Shader::TextureType type,
-                              bool is_array) {
-        const std::size_t offset = static_cast<std::size_t>(sampler.index.Value());
+        declarations.AddLine("vec4 amem[" + std::to_string(MAX_GEOMETRY_BUFFERS) + "][" +
+                             std::to_string(MAX_ATTRIBUTES) + "];");
+        declarations.AddNewLine();
 
-        // If this sampler has already been used, return the existing mapping.
-        const auto itr =
-            std::find_if(used_samplers.begin(), used_samplers.end(),
-                         [&](const SamplerEntry& entry) { return entry.GetOffset() == offset; });
-
-        if (itr != used_samplers.end()) {
-            ASSERT(itr->GetType() == type && itr->IsArray() == is_array);
-            return itr->GetName();
+        constexpr char buffer[] = "amem[output_buffer]";
+        declarations.AddLine("void emit_vertex(uint output_buffer) {");
+        ++declarations.scope;
+        for (const auto element : declr_output_attribute) {
+            declarations.AddLine(GetOutputAttribute(element) + " = " + buffer + '[' +
+                                 std::to_string(static_cast<u32>(element)) + "];");
         }
 
-        // Otherwise create a new mapping for this sampler
-        const std::size_t next_index = used_samplers.size();
-        const SamplerEntry entry{stage, offset, next_index, type, is_array};
-        used_samplers.emplace_back(entry);
-        return entry.GetName();
+        declarations.AddLine("position = " + std::string(buffer) + '[' +
+                             std::to_string(static_cast<u32>(Attribute::Index::Position)) + "];");
+
+        // If a geometry shader is attached, it will always flip (it's the last stage before
+        // fragment). For more info about flipping, refer to gl_shader_gen.cpp.
+        declarations.AddLine("position.xy *= viewport_flip.xy;");
+        declarations.AddLine("gl_Position = position;");
+        declarations.AddLine("position.w = 1.0;");
+        declarations.AddLine("EmitVertex();");
+        --declarations.scope;
+        declarations.AddLine('}');
+        declarations.AddNewLine();
     }
 
-private:
     /// Generates code representing a temporary (GPR) register.
     std::string GetRegister(const Register& reg, unsigned elem) {
         if (reg == Register::ZeroIndex) {
@@ -585,11 +707,19 @@ private:
 
     /// Generates code representing an input attribute register.
     std::string GetInputAttribute(Attribute::Index attribute,
-                                  const Tegra::Shader::IpaMode& input_mode) {
+                                  const Tegra::Shader::IpaMode& input_mode,
+                                  boost::optional<Register> vertex = {}) {
+        auto GeometryPass = [&](const std::string& name) {
+            if (stage == Maxwell3D::Regs::ShaderStage::Geometry && vertex) {
+                return "gs_" + name + '[' + GetRegisterAsInteger(vertex.value(), 0, false) + ']';
+            }
+            return name;
+        };
+
         switch (attribute) {
         case Attribute::Index::Position:
             if (stage != Maxwell3D::Regs::ShaderStage::Fragment) {
-                return "position";
+                return GeometryPass("position");
             } else {
                 return "vec4(gl_FragCoord.x, gl_FragCoord.y, gl_FragCoord.z, 1.0)";
             }
@@ -618,7 +748,7 @@ private:
                         UNREACHABLE();
                     }
                 }
-                return "input_attribute_" + std::to_string(index);
+                return GeometryPass("input_attribute_" + std::to_string(index));
             }
 
             LOG_CRITICAL(HW_GPU, "Unhandled input attribute: {}", static_cast<u32>(attribute));
@@ -671,7 +801,7 @@ private:
         return out;
     }
 
-    /// Generates code representing an output attribute register.
+    /// Generates code representing the declaration name of an output attribute register.
     std::string GetOutputAttribute(Attribute::Index attribute) {
         switch (attribute) {
         case Attribute::Index::Position:
@@ -707,6 +837,7 @@ private:
     std::vector<SamplerEntry> used_samplers;
     const Maxwell3D::Regs::ShaderStage& stage;
     const std::string& suffix;
+    const Tegra::Shader::Header& header;
 };
 
 class GLSLGenerator {
@@ -747,8 +878,9 @@ private:
     }
 
     /// Generates code representing a texture sampler.
-    std::string GetSampler(const Sampler& sampler, Tegra::Shader::TextureType type, bool is_array) {
-        return regs.AccessSampler(sampler, type, is_array);
+    std::string GetSampler(const Sampler& sampler, Tegra::Shader::TextureType type, bool is_array,
+                           bool is_shadow) {
+        return regs.AccessSampler(sampler, type, is_array, is_shadow);
     }
 
     /**
@@ -1002,6 +1134,24 @@ private:
         shader.AddLine('}');
     }
 
+    static u32 TextureCoordinates(Tegra::Shader::TextureType texture_type) {
+        switch (texture_type) {
+        case Tegra::Shader::TextureType::Texture1D: {
+            return 1;
+        }
+        case Tegra::Shader::TextureType::Texture2D: {
+            return 2;
+        }
+        case Tegra::Shader::TextureType::TextureCube: {
+            return 3;
+        }
+        default:
+            LOG_CRITICAL(HW_GPU, "Unhandled texture type {}", static_cast<u32>(texture_type));
+            UNREACHABLE();
+            return 0;
+        }
+    }
+
     /*
      * Emits code to push the input target address to the SSY address stack, incrementing the stack
      * top.
@@ -1083,8 +1233,8 @@ private:
             return offset + 1;
         }
 
-        shader.AddLine("// " + std::to_string(offset) + ": " + opcode->GetName() + " (" +
-                       std::to_string(instr.value) + ')');
+        shader.AddLine(
+            fmt::format("// {}: {} (0x{:016x})", offset, opcode->GetName(), instr.value));
 
         using Tegra::Shader::Pred;
         ASSERT_MSG(instr.pred.full_pred != Pred::NeverExecute,
@@ -1806,7 +1956,7 @@ private:
                 const auto LoadNextElement = [&](u32 reg_offset) {
                     regs.SetRegisterToInputAttibute(instr.gpr0.Value() + reg_offset, next_element,
                                                     static_cast<Attribute::Index>(next_index),
-                                                    input_mode);
+                                                    input_mode, instr.gpr39.Value());
 
                     // Load the next attribute element into the following register. If the element
                     // to load goes beyond the vec4 size, load the first element of the next
@@ -1870,8 +2020,8 @@ private:
 
                 const auto StoreNextElement = [&](u32 reg_offset) {
                     regs.SetOutputAttributeToRegister(static_cast<Attribute::Index>(next_index),
-                                                      next_element,
-                                                      instr.gpr0.Value() + reg_offset);
+                                                      next_element, instr.gpr0.Value() + reg_offset,
+                                                      instr.gpr39.Value());
 
                     // Load the next attribute element into the following register. If the element
                     // to load goes beyond the vec4 size, load the first element of the next
@@ -1896,24 +2046,35 @@ private:
                            "NODEP is not implemented");
                 ASSERT_MSG(!instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::AOFFI),
                            "AOFFI is not implemented");
-                ASSERT_MSG(!instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC),
-                           "DC is not implemented");
 
-                switch (texture_type) {
-                case Tegra::Shader::TextureType::Texture1D: {
+                const bool depth_compare =
+                    instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC);
+                u32 num_coordinates = TextureCoordinates(texture_type);
+                if (depth_compare)
+                    num_coordinates += 1;
+
+                switch (num_coordinates) {
+                case 1: {
                     const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
                     coord = "float coords = " + x + ';';
                     break;
                 }
-                case Tegra::Shader::TextureType::Texture2D: {
+                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 + ");";
                     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 + ");";
+                    break;
+                }
                 default:
-                    LOG_CRITICAL(HW_GPU, "Unhandled texture type {}",
-                                 static_cast<u32>(texture_type));
+                    LOG_CRITICAL(HW_GPU, "Unhandled coordinates number {}",
+                                 static_cast<u32>(num_coordinates));
                     UNREACHABLE();
 
                     // Fallback to interpreting as a 2D texture for now
@@ -1924,9 +2085,10 @@ private:
                 }
                 // TODO: make sure coordinates are always indexed to gpr8 and gpr20 is always bias
                 // or lod.
-                const std::string op_c = regs.GetRegisterAsFloat(instr.gpr20);
+                std::string op_c;
 
-                const std::string sampler = GetSampler(instr.sampler, texture_type, false);
+                const std::string sampler =
+                    GetSampler(instr.sampler, texture_type, false, 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.
 
@@ -1935,7 +2097,7 @@ private:
                 shader.AddLine(coord);
                 std::string texture;
 
-                switch (instr.tex.process_mode) {
+                switch (instr.tex.GetTextureProcessMode()) {
                 case Tegra::Shader::TextureProcessMode::None: {
                     texture = "texture(" + sampler + ", coords)";
                     break;
@@ -1946,12 +2108,22 @@ private:
                 }
                 case Tegra::Shader::TextureProcessMode::LB:
                 case Tegra::Shader::TextureProcessMode::LBA: {
+                    if (num_coordinates <= 2) {
+                        op_c = regs.GetRegisterAsFloat(instr.gpr20);
+                    } else {
+                        op_c = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
+                    }
                     // TODO: Figure if A suffix changes the equation at all.
                     texture = "texture(" + sampler + ", coords, " + op_c + ')';
                     break;
                 }
                 case Tegra::Shader::TextureProcessMode::LL:
                 case Tegra::Shader::TextureProcessMode::LLA: {
+                    if (num_coordinates <= 2) {
+                        op_c = regs.GetRegisterAsFloat(instr.gpr20);
+                    } else {
+                        op_c = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
+                    }
                     // TODO: Figure if A suffix changes the equation at all.
                     texture = "textureLod(" + sampler + ", coords, " + op_c + ')';
                     break;
@@ -1959,18 +2131,22 @@ private:
                 default: {
                     texture = "texture(" + sampler + ", coords)";
                     LOG_CRITICAL(HW_GPU, "Unhandled texture process mode {}",
-                                 static_cast<u32>(instr.tex.process_mode.Value()));
+                                 static_cast<u32>(instr.tex.GetTextureProcessMode()));
                     UNREACHABLE();
                 }
                 }
-                std::size_t dest_elem{};
-                for (std::size_t elem = 0; elem < 4; ++elem) {
-                    if (!instr.tex.IsComponentEnabled(elem)) {
-                        // Skip disabled components
-                        continue;
+                if (!depth_compare) {
+                    std::size_t dest_elem{};
+                    for (std::size_t elem = 0; elem < 4; ++elem) {
+                        if (!instr.tex.IsComponentEnabled(elem)) {
+                            // Skip disabled components
+                            continue;
+                        }
+                        regs.SetRegisterToFloat(instr.gpr0, elem, texture, 1, 4, false, dest_elem);
+                        ++dest_elem;
                     }
-                    regs.SetRegisterToFloat(instr.gpr0, elem, texture, 1, 4, false, dest_elem);
-                    ++dest_elem;
+                } else {
+                    regs.SetRegisterToFloat(instr.gpr0, 0, texture, 1, 1, false);
                 }
                 --shader.scope;
                 shader.AddLine("}");
@@ -1983,11 +2159,15 @@ private:
 
                 ASSERT_MSG(!instr.texs.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
                            "NODEP is not implemented");
-                ASSERT_MSG(!instr.texs.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC),
-                           "DC is not implemented");
 
-                switch (texture_type) {
-                case Tegra::Shader::TextureType::Texture2D: {
+                const bool depth_compare =
+                    instr.texs.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC);
+                u32 num_coordinates = TextureCoordinates(texture_type);
+                if (depth_compare)
+                    num_coordinates += 1;
+
+                switch (num_coordinates) {
+                case 2: {
                     if (is_array) {
                         const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
                         const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
@@ -2000,17 +2180,25 @@ private:
                     }
                     break;
                 }
-                case Tegra::Shader::TextureType::TextureCube: {
-                    ASSERT_MSG(!is_array, "Unimplemented");
-                    std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                    std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
-                    std::string z = regs.GetRegisterAsFloat(instr.gpr20);
-                    coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
+                case 3: {
+                    if (is_array) {
+                        UNIMPLEMENTED_MSG("3-coordinate arrays not fully implemented");
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                        const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
+                        coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+                        texture_type = Tegra::Shader::TextureType::Texture2D;
+                        is_array = false;
+                    } 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 + ");";
+                    }
                     break;
                 }
                 default:
-                    LOG_CRITICAL(HW_GPU, "Unhandled texture type {}",
-                                 static_cast<u32>(texture_type));
+                    LOG_CRITICAL(HW_GPU, "Unhandled coordinates number {}",
+                                 static_cast<u32>(num_coordinates));
                     UNREACHABLE();
 
                     // Fallback to interpreting as a 2D texture for now
@@ -2020,9 +2208,35 @@ private:
                     texture_type = Tegra::Shader::TextureType::Texture2D;
                     is_array = false;
                 }
-                const std::string sampler = GetSampler(instr.sampler, texture_type, is_array);
-                const std::string texture = "texture(" + sampler + ", coords)";
-                WriteTexsInstruction(instr, coord, texture);
+                const std::string sampler =
+                    GetSampler(instr.sampler, texture_type, is_array, depth_compare);
+                std::string texture;
+                switch (instr.texs.GetTextureProcessMode()) {
+                case Tegra::Shader::TextureProcessMode::None: {
+                    texture = "texture(" + sampler + ", coords)";
+                    break;
+                }
+                case Tegra::Shader::TextureProcessMode::LZ: {
+                    texture = "textureLod(" + sampler + ", coords, 0.0)";
+                    break;
+                }
+                case Tegra::Shader::TextureProcessMode::LL: {
+                    const std::string op_c = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
+                    texture = "textureLod(" + sampler + ", coords, " + op_c + ')';
+                    break;
+                }
+                default: {
+                    texture = "texture(" + sampler + ", coords)";
+                    LOG_CRITICAL(HW_GPU, "Unhandled texture process mode {}",
+                                 static_cast<u32>(instr.texs.GetTextureProcessMode()));
+                    UNREACHABLE();
+                }
+                }
+                if (!depth_compare) {
+                    WriteTexsInstruction(instr, coord, texture);
+                } else {
+                    WriteTexsInstruction(instr, coord, "vec4(" + texture + ')');
+                }
                 break;
             }
             case OpCode::Id::TLDS: {
@@ -2062,9 +2276,26 @@ private:
                                  static_cast<u32>(texture_type));
                     UNREACHABLE();
                 }
-
-                const std::string sampler = GetSampler(instr.sampler, texture_type, is_array);
-                const std::string texture = "texelFetch(" + sampler + ", coords, 0)";
+                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: {
+                    texture = "texelFetch(" + sampler + ", coords, " + op_c + ')';
+                    break;
+                }
+                default: {
+                    texture = "texelFetch(" + sampler + ", coords, 0)";
+                    LOG_CRITICAL(HW_GPU, "Unhandled texture process mode {}",
+                                 static_cast<u32>(instr.tlds.GetTextureProcessMode()));
+                    UNREACHABLE();
+                }
+                }
                 WriteTexsInstruction(instr, coord, texture);
                 break;
             }
@@ -2077,28 +2308,43 @@ private:
                            "NODEP is not implemented");
                 ASSERT_MSG(!instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::AOFFI),
                            "AOFFI is not implemented");
-                ASSERT_MSG(!instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC),
-                           "DC is not implemented");
                 ASSERT_MSG(!instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::NDV),
                            "NDV is not implemented");
                 ASSERT_MSG(!instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::PTP),
                            "PTP is not implemented");
+                const bool depth_compare =
+                    instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC);
+                auto texture_type = instr.tld4.texture_type.Value();
+                u32 num_coordinates = TextureCoordinates(texture_type);
+                if (depth_compare)
+                    num_coordinates += 1;
 
-                switch (instr.tld4.texture_type) {
-                case Tegra::Shader::TextureType::Texture2D: {
+                switch (num_coordinates) {
+                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 + ");";
                     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.gpr8.Value() + 2);
+                    coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
+                    break;
+                }
                 default:
-                    LOG_CRITICAL(HW_GPU, "Unhandled texture type {}",
-                                 static_cast<u32>(instr.tld4.texture_type.Value()));
+                    LOG_CRITICAL(HW_GPU, "Unhandled coordinates number {}",
+                                 static_cast<u32>(num_coordinates));
                     UNREACHABLE();
+                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                    const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                    coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+                    texture_type = Tegra::Shader::TextureType::Texture2D;
                 }
 
                 const std::string sampler =
-                    GetSampler(instr.sampler, instr.tld4.texture_type, false);
+                    GetSampler(instr.sampler, texture_type, false, 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.
                 shader.AddLine("{");
@@ -2106,15 +2352,18 @@ private:
                 shader.AddLine(coord);
                 const std::string texture = "textureGather(" + sampler + ", coords, " +
                                             std::to_string(instr.tld4.component) + ')';
-
-                std::size_t dest_elem{};
-                for (std::size_t elem = 0; elem < 4; ++elem) {
-                    if (!instr.tex.IsComponentEnabled(elem)) {
-                        // Skip disabled components
-                        continue;
+                if (!depth_compare) {
+                    std::size_t dest_elem{};
+                    for (std::size_t elem = 0; elem < 4; ++elem) {
+                        if (!instr.tex.IsComponentEnabled(elem)) {
+                            // Skip disabled components
+                            continue;
+                        }
+                        regs.SetRegisterToFloat(instr.gpr0, elem, texture, 1, 4, false, dest_elem);
+                        ++dest_elem;
                     }
-                    regs.SetRegisterToFloat(instr.gpr0, elem, texture, 1, 4, false, dest_elem);
-                    ++dest_elem;
+                } else {
+                    regs.SetRegisterToFloat(instr.gpr0, 0, texture, 1, 1, false);
                 }
                 --shader.scope;
                 shader.AddLine("}");
@@ -2125,18 +2374,30 @@ private:
                            "NODEP is not implemented");
                 ASSERT_MSG(!instr.tld4s.UsesMiscMode(Tegra::Shader::TextureMiscMode::AOFFI),
                            "AOFFI is not implemented");
-                ASSERT_MSG(!instr.tld4s.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC),
-                           "DC is not implemented");
 
+                const bool depth_compare =
+                    instr.tld4s.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC);
                 const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
                 const std::string op_b = regs.GetRegisterAsFloat(instr.gpr20);
                 // TODO(Subv): Figure out how the sampler type is encoded in the TLD4S instruction.
-                const std::string sampler =
-                    GetSampler(instr.sampler, Tegra::Shader::TextureType::Texture2D, false);
-                const std::string coord = "vec2 coords = vec2(" + op_a + ", " + op_b + ");";
+                const std::string sampler = GetSampler(
+                    instr.sampler, Tegra::Shader::TextureType::Texture2D, false, depth_compare);
+                std::string coord;
+                if (!depth_compare) {
+                    coord = "vec2 coords = vec2(" + op_a + ", " + op_b + ");";
+                } else {
+                    // Note: TLD4S coordinate encoding works just like TEXS's
+                    const std::string op_c = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                    coord = "vec3 coords = vec3(" + op_a + ", " + op_c + ", " + op_b + ");";
+                }
                 const std::string texture = "textureGather(" + sampler + ", coords, " +
                                             std::to_string(instr.tld4s.component) + ')';
-                WriteTexsInstruction(instr, coord, texture);
+
+                if (!depth_compare) {
+                    WriteTexsInstruction(instr, coord, texture);
+                } else {
+                    WriteTexsInstruction(instr, coord, "vec4(" + texture + ')');
+                }
                 break;
             }
             case OpCode::Id::TXQ: {
@@ -2147,7 +2408,7 @@ private:
                 // Sadly, not all texture instructions specify the type of texture their sampler
                 // uses. This must be fixed at a later instance.
                 const std::string sampler =
-                    GetSampler(instr.sampler, Tegra::Shader::TextureType::Texture2D, false);
+                    GetSampler(instr.sampler, Tegra::Shader::TextureType::Texture2D, false, false);
                 switch (instr.txq.query_type) {
                 case Tegra::Shader::TextureQueryType::Dimension: {
                     const std::string texture = "textureQueryLevels(" + sampler + ')';
@@ -2168,24 +2429,22 @@ private:
                 ASSERT_MSG(!instr.tmml.UsesMiscMode(Tegra::Shader::TextureMiscMode::NDV),
                            "NDV is not implemented");
 
-                const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
-                const std::string op_b = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
                 const bool is_array = instr.tmml.array != 0;
                 auto texture_type = instr.tmml.texture_type.Value();
-                const std::string sampler = GetSampler(instr.sampler, texture_type, is_array);
+                const std::string sampler =
+                    GetSampler(instr.sampler, texture_type, is_array, false);
 
                 // TODO: add coordinates for different samplers once other texture types are
                 // implemented.
                 std::string coord;
                 switch (texture_type) {
                 case Tegra::Shader::TextureType::Texture1D: {
-                    std::string x = regs.GetRegisterAsFloat(instr.gpr8);
                     coord = "float coords = " + x + ';';
                     break;
                 }
                 case Tegra::Shader::TextureType::Texture2D: {
-                    std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                    std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                    const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
                     coord = "vec2 coords = vec2(" + x + ", " + y + ");";
                     break;
                 }
@@ -2195,8 +2454,7 @@ private:
                     UNREACHABLE();
 
                     // Fallback to interpreting as a 2D texture for now
-                    std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                    std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                    const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
                     coord = "vec2 coords = vec2(" + x + ", " + y + ");";
                     texture_type = Tegra::Shader::TextureType::Texture2D;
                 }
@@ -2606,6 +2864,52 @@ private:
 
                 break;
             }
+            case OpCode::Id::OUT_R: {
+                ASSERT(instr.gpr20.Value() == Register::ZeroIndex);
+                ASSERT_MSG(stage == Maxwell3D::Regs::ShaderStage::Geometry,
+                           "OUT is expected to be used in a geometry shader.");
+
+                if (instr.out.emit) {
+                    // gpr0 is used to store the next address. Hardware returns a pointer but
+                    // we just return the next index with a cyclic cap.
+                    const std::string current{regs.GetRegisterAsInteger(instr.gpr8, 0, false)};
+                    const std::string next = "((" + current + " + 1" + ") % " +
+                                             std::to_string(MAX_GEOMETRY_BUFFERS) + ')';
+                    shader.AddLine("emit_vertex(" + current + ");");
+                    regs.SetRegisterToInteger(instr.gpr0, false, 0, next, 1, 1);
+                }
+                if (instr.out.cut) {
+                    shader.AddLine("EndPrimitive();");
+                }
+
+                break;
+            }
+            case OpCode::Id::MOV_SYS: {
+                switch (instr.sys20) {
+                case Tegra::Shader::SystemVariable::InvocationInfo: {
+                    LOG_WARNING(HW_GPU, "MOV_SYS instruction with InvocationInfo is incomplete");
+                    regs.SetRegisterToInteger(instr.gpr0, false, 0, "0u", 1, 1);
+                    break;
+                }
+                default: {
+                    LOG_CRITICAL(HW_GPU, "Unhandled system move: {}",
+                                 static_cast<u32>(instr.sys20.Value()));
+                    UNREACHABLE();
+                }
+                }
+                break;
+            }
+            case OpCode::Id::ISBERD: {
+                ASSERT(instr.isberd.o == 0);
+                ASSERT(instr.isberd.skew == 0);
+                ASSERT(instr.isberd.shift == Tegra::Shader::IsberdShift::None);
+                ASSERT(instr.isberd.mode == Tegra::Shader::IsberdMode::None);
+                ASSERT_MSG(stage == Maxwell3D::Regs::ShaderStage::Geometry,
+                           "ISBERD is expected to be used in a geometry shader.");
+                LOG_WARNING(HW_GPU, "ISBERD instruction is incomplete");
+                regs.SetRegisterToFloat(instr.gpr0, 0, regs.GetRegisterAsFloat(instr.gpr8), 1, 1);
+                break;
+            }
             case OpCode::Id::BRA: {
                 ASSERT_MSG(instr.bra.constant_buffer == 0,
                            "BRA with constant buffers are not implemented");
@@ -2649,6 +2953,88 @@ private:
                 LOG_WARNING(HW_GPU, "DEPBAR instruction is stubbed");
                 break;
             }
+            case OpCode::Id::VMAD: {
+                const bool signed_a = instr.vmad.signed_a == 1;
+                const bool signed_b = instr.vmad.signed_b == 1;
+                const bool result_signed = signed_a || signed_b;
+                boost::optional<std::string> forced_result;
+
+                auto Unpack = [&](const std::string& op, bool is_chunk, bool is_signed,
+                                  Tegra::Shader::VmadType type, u64 byte_height) {
+                    const std::string value = [&]() {
+                        if (!is_chunk) {
+                            const auto offset = static_cast<u32>(byte_height * 8);
+                            return "((" + op + " >> " + std::to_string(offset) + ") & 0xff)";
+                        }
+                        const std::string zero = "0";
+
+                        switch (type) {
+                        case Tegra::Shader::VmadType::Size16_Low:
+                            return '(' + op + " & 0xffff)";
+                        case Tegra::Shader::VmadType::Size16_High:
+                            return '(' + op + " >> 16)";
+                        case Tegra::Shader::VmadType::Size32:
+                            // TODO(Rodrigo): From my hardware tests it becomes a bit "mad" when
+                            // this type is used (1 * 1 + 0 == 0x5b800000). Until a better
+                            // explanation is found: assert.
+                            UNREACHABLE_MSG("Unimplemented");
+                            return zero;
+                        case Tegra::Shader::VmadType::Invalid:
+                            // Note(Rodrigo): This flag is invalid according to nvdisasm. From my
+                            // testing (even though it's invalid) this makes the whole instruction
+                            // assign zero to target register.
+                            forced_result = boost::make_optional(zero);
+                            return zero;
+                        default:
+                            UNREACHABLE();
+                            return zero;
+                        }
+                    }();
+
+                    if (is_signed) {
+                        return "int(" + value + ')';
+                    }
+                    return value;
+                };
+
+                const std::string op_a = Unpack(regs.GetRegisterAsInteger(instr.gpr8, 0, false),
+                                                instr.vmad.is_byte_chunk_a != 0, signed_a,
+                                                instr.vmad.type_a, instr.vmad.byte_height_a);
+
+                std::string op_b;
+                if (instr.vmad.use_register_b) {
+                    op_b = Unpack(regs.GetRegisterAsInteger(instr.gpr20, 0, false),
+                                  instr.vmad.is_byte_chunk_b != 0, signed_b, instr.vmad.type_b,
+                                  instr.vmad.byte_height_b);
+                } else {
+                    op_b = '(' +
+                           std::to_string(signed_b ? static_cast<s16>(instr.alu.GetImm20_16())
+                                                   : instr.alu.GetImm20_16()) +
+                           ')';
+                }
+
+                const std::string op_c = regs.GetRegisterAsInteger(instr.gpr39, 0, result_signed);
+
+                std::string result;
+                if (forced_result) {
+                    result = *forced_result;
+                } else {
+                    result = '(' + op_a + " * " + op_b + " + " + op_c + ')';
+
+                    switch (instr.vmad.shr) {
+                    case Tegra::Shader::VmadShr::Shr7:
+                        result = '(' + result + " >> 7)";
+                        break;
+                    case Tegra::Shader::VmadShr::Shr15:
+                        result = '(' + result + " >> 15)";
+                        break;
+                    }
+                }
+                regs.SetRegisterToInteger(instr.gpr0, result_signed, 1, result, 1, 1,
+                                          instr.vmad.saturate == 1, 0, Register::Size::Word,
+                                          instr.vmad.cc);
+                break;
+            }
             default: {
                 LOG_CRITICAL(HW_GPU, "Unhandled instruction: {}", opcode->GetName());
                 UNREACHABLE();
@@ -2779,7 +3165,7 @@ private:
 
     ShaderWriter shader;
     ShaderWriter declarations;
-    GLSLRegisterManager regs{shader, declarations, stage, suffix};
+    GLSLRegisterManager regs{shader, declarations, stage, suffix, header};
 
     // Declarations
     std::set<std::string> declr_predicates;

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -17,7 +17,18 @@ ProgramResult GenerateVertexShader(const ShaderSetup& setup) {
     std::string out = "#version 430 core\n";
     out += "#extension GL_ARB_separate_shader_objects : enable\n\n";
     out += Decompiler::GetCommonDeclarations();
-    out += "bool exec_vertex();\n";
+
+    out += R"(
+out gl_PerVertex {
+    vec4 gl_Position;
+};
+
+layout(std140) uniform vs_config {
+    vec4 viewport_flip;
+    uvec4 instance_id;
+    uvec4 flip_stage;
+};
+)";
 
     if (setup.IsDualProgram()) {
         out += "bool exec_vertex_b();\n";
@@ -28,19 +39,18 @@ ProgramResult GenerateVertexShader(const ShaderSetup& setup) {
                                      Maxwell3D::Regs::ShaderStage::Vertex, "vertex")
             .get_value_or({});
 
+    out += program.first;
+
+    if (setup.IsDualProgram()) {
+        ProgramResult program_b =
+            Decompiler::DecompileProgram(setup.program.code_b, PROGRAM_OFFSET,
+                                         Maxwell3D::Regs::ShaderStage::Vertex, "vertex_b")
+                .get_value_or({});
+        out += program_b.first;
+    }
+
     out += R"(
 
-out gl_PerVertex {
-    vec4 gl_Position;
-};
-
-out vec4 position;
-
-layout (std140) uniform vs_config {
-    vec4 viewport_flip;
-    uvec4 instance_id;
-};
-
 void main() {
     position = vec4(0.0, 0.0, 0.0, 0.0);
     exec_vertex();
@@ -52,27 +62,52 @@ void main() {
 
     out += R"(
 
-    // Viewport can be flipped, which is unsupported by glViewport
-    position.xy *= viewport_flip.xy;
+    // Check if the flip stage is VertexB
+    if (flip_stage[0] == 1) {
+        // Viewport can be flipped, which is unsupported by glViewport
+        position.xy *= viewport_flip.xy;
+    }
     gl_Position = position;
 
     // TODO(bunnei): This is likely a hack, position.w should be interpolated as 1.0
     // For now, this is here to bring order in lieu of proper emulation
-    position.w = 1.0;
+    if (flip_stage[0] == 1) {
+        position.w = 1.0;
+    }
 }
 
 )";
 
+    return {out, program.second};
+}
+
+ProgramResult GenerateGeometryShader(const ShaderSetup& setup) {
+    std::string out = "#version 430 core\n";
+    out += "#extension GL_ARB_separate_shader_objects : enable\n\n";
+    out += Decompiler::GetCommonDeclarations();
+    out += "bool exec_geometry();\n";
+
+    ProgramResult program =
+        Decompiler::DecompileProgram(setup.program.code, PROGRAM_OFFSET,
+                                     Maxwell3D::Regs::ShaderStage::Geometry, "geometry")
+            .get_value_or({});
+    out += R"(
+out gl_PerVertex {
+    vec4 gl_Position;
+};
+
+layout (std140) uniform gs_config {
+    vec4 viewport_flip;
+    uvec4 instance_id;
+    uvec4 flip_stage;
+};
+
+void main() {
+    exec_geometry();
+}
+
+)";
     out += program.first;
-
-    if (setup.IsDualProgram()) {
-        ProgramResult program_b =
-            Decompiler::DecompileProgram(setup.program.code_b, PROGRAM_OFFSET,
-                                         Maxwell3D::Regs::ShaderStage::Vertex, "vertex_b")
-                .get_value_or({});
-        out += program_b.first;
-    }
-
     return {out, program.second};
 }
 
@@ -87,7 +122,6 @@ ProgramResult GenerateFragmentShader(const ShaderSetup& setup) {
                                      Maxwell3D::Regs::ShaderStage::Fragment, "fragment")
             .get_value_or({});
     out += R"(
-in vec4 position;
 layout(location = 0) out vec4 FragColor0;
 layout(location = 1) out vec4 FragColor1;
 layout(location = 2) out vec4 FragColor2;
@@ -100,6 +134,7 @@ layout(location = 7) out vec4 FragColor7;
 layout (std140) uniform fs_config {
     vec4 viewport_flip;
     uvec4 instance_id;
+    uvec4 flip_stage;
 };
 
 void main() {
@@ -110,5 +145,4 @@ void main() {
     out += program.first;
     return {out, program.second};
 }
-
-} // namespace OpenGL::GLShader
+} // namespace OpenGL::GLShader

src/video_core/renderer_opengl/gl_shader_gen.h

@@ -75,8 +75,9 @@ class SamplerEntry {
 
 public:
     SamplerEntry(Maxwell::ShaderStage stage, std::size_t offset, std::size_t index,
-                 Tegra::Shader::TextureType type, bool is_array)
-        : offset(offset), stage(stage), sampler_index(index), type(type), is_array(is_array) {}
+                 Tegra::Shader::TextureType type, bool is_array, bool is_shadow)
+        : offset(offset), stage(stage), sampler_index(index), type(type), is_array(is_array),
+          is_shadow(is_shadow) {}
 
     std::size_t GetOffset() const {
         return offset;
@@ -117,6 +118,8 @@ public:
         }
         if (is_array)
             glsl_type += "Array";
+        if (is_shadow)
+            glsl_type += "Shadow";
         return glsl_type;
     }
 
@@ -128,6 +131,10 @@ public:
         return is_array;
     }
 
+    bool IsShadow() const {
+        return is_shadow;
+    }
+
     u32 GetHash() const {
         return (static_cast<u32>(stage) << 16) | static_cast<u32>(sampler_index);
     }
@@ -147,7 +154,8 @@ private:
     Maxwell::ShaderStage stage;      ///< Shader stage where this sampler was used.
     std::size_t sampler_index;       ///< Value used to index into the generated GLSL sampler array.
     Tegra::Shader::TextureType type; ///< The type used to sample this texture (Texture2D, etc)
-    bool is_array; ///< Whether the texture is being sampled as an array texture or not.
+    bool is_array;  ///< Whether the texture is being sampled as an array texture or not.
+    bool is_shadow; ///< Whether the texture is being sampled as a depth texture or not.
 };
 
 struct ShaderEntries {
@@ -187,6 +195,12 @@ private:
  */
 ProgramResult GenerateVertexShader(const ShaderSetup& setup);
 
+/**
+ * Generates the GLSL geometry shader program source code for the given GS program
+ * @returns String of the shader source code
+ */
+ProgramResult GenerateGeometryShader(const ShaderSetup& setup);
+
 /**
  * Generates the GLSL fragment shader program source code for the given FS program
  * @returns String of the shader source code

src/video_core/renderer_opengl/gl_shader_manager.cpp

@@ -18,6 +18,14 @@ void MaxwellUniformData::SetFromRegs(const Maxwell3D::State::ShaderStageInfo& sh
 
     // We only assign the instance to the first component of the vector, the rest is just padding.
     instance_id[0] = state.current_instance;
+
+    // Assign in which stage the position has to be flipped
+    // (the last stage before the fragment shader).
+    if (gpu.regs.shader_config[static_cast<u32>(Maxwell3D::Regs::ShaderProgram::Geometry)].enable) {
+        flip_stage[0] = static_cast<u32>(Maxwell3D::Regs::ShaderProgram::Geometry);
+    } else {
+        flip_stage[0] = static_cast<u32>(Maxwell3D::Regs::ShaderProgram::VertexB);
+    }
 }
 
 } // namespace OpenGL::GLShader

src/video_core/renderer_opengl/gl_shader_manager.h

@@ -21,8 +21,9 @@ struct MaxwellUniformData {
     void SetFromRegs(const Maxwell3D::State::ShaderStageInfo& shader_stage);
     alignas(16) GLvec4 viewport_flip;
     alignas(16) GLuvec4 instance_id;
+    alignas(16) GLuvec4 flip_stage;
 };
-static_assert(sizeof(MaxwellUniformData) == 32, "MaxwellUniformData structure size is incorrect");
+static_assert(sizeof(MaxwellUniformData) == 48, "MaxwellUniformData structure size is incorrect");
 static_assert(sizeof(MaxwellUniformData) < 16384,
               "MaxwellUniformData structure must be less than 16kb as per the OpenGL spec");
 
@@ -36,6 +37,10 @@ public:
         vs = program;
     }
 
+    void UseProgrammableGeometryShader(GLuint program) {
+        gs = program;
+    }
+
     void UseProgrammableFragmentShader(GLuint program) {
         fs = program;
     }

src/video_core/renderer_opengl/maxwell_to_gl.h

@@ -159,6 +159,31 @@ inline GLenum WrapMode(Tegra::Texture::WrapMode wrap_mode) {
     return {};
 }
 
+inline GLenum DepthCompareFunc(Tegra::Texture::DepthCompareFunc func) {
+    switch (func) {
+    case Tegra::Texture::DepthCompareFunc::Never:
+        return GL_NEVER;
+    case Tegra::Texture::DepthCompareFunc::Less:
+        return GL_LESS;
+    case Tegra::Texture::DepthCompareFunc::LessEqual:
+        return GL_LEQUAL;
+    case Tegra::Texture::DepthCompareFunc::Equal:
+        return GL_EQUAL;
+    case Tegra::Texture::DepthCompareFunc::NotEqual:
+        return GL_NOTEQUAL;
+    case Tegra::Texture::DepthCompareFunc::Greater:
+        return GL_GREATER;
+    case Tegra::Texture::DepthCompareFunc::GreaterEqual:
+        return GL_GEQUAL;
+    case Tegra::Texture::DepthCompareFunc::Always:
+        return GL_ALWAYS;
+    }
+    LOG_CRITICAL(Render_OpenGL, "Unimplemented texture depth compare function ={}",
+                 static_cast<u32>(func));
+    UNREACHABLE();
+    return {};
+}
+
 inline GLenum BlendEquation(Maxwell::Blend::Equation equation) {
     switch (equation) {
     case Maxwell::Blend::Equation::Add:

src/video_core/textures/decoders.cpp

@@ -4,6 +4,7 @@
 
 #include <cmath>
 #include <cstring>
+#include "common/alignment.h"
 #include "common/assert.h"
 #include "core/memory.h"
 #include "video_core/gpu.h"
@@ -199,4 +200,19 @@ std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat
     return rgba_data;
 }
 
+std::size_t CalculateSize(bool tiled, u32 bytes_per_pixel, u32 width, u32 height, u32 depth,
+                          u32 block_height, u32 block_depth) {
+    if (tiled) {
+        const u32 gobs_in_x = 64 / bytes_per_pixel;
+        const u32 gobs_in_y = 8;
+        const u32 gobs_in_z = 1;
+        const u32 aligned_width = Common::AlignUp(width, gobs_in_x);
+        const u32 aligned_height = Common::AlignUp(height, gobs_in_y * block_height);
+        const u32 aligned_depth = Common::AlignUp(depth, gobs_in_z * block_depth);
+        return aligned_width * aligned_height * aligned_depth * bytes_per_pixel;
+    } else {
+        return width * height * depth * bytes_per_pixel;
+    }
+}
+
 } // namespace Tegra::Texture

src/video_core/textures/decoders.h

@@ -32,4 +32,10 @@ void CopySwizzledData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_
 std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat format, u32 width,
                               u32 height);
 
+/**
+ * This function calculates the correct size of a texture depending if it's tiled or not.
+ */
+std::size_t CalculateSize(bool tiled, u32 bytes_per_pixel, u32 width, u32 height, u32 depth,
+                          u32 block_height, u32 block_depth);
+
 } // namespace Tegra::Texture

src/video_core/textures/texture.h

@@ -161,7 +161,9 @@ struct TICEntry {
         BitField<21, 3, TICHeaderVersion> header_version;
     };
     union {
+        BitField<0, 3, u32> block_width;
         BitField<3, 3, u32> block_height;
+        BitField<6, 3, u32> block_depth;
 
         // High 16 bits of the pitch value
         BitField<0, 16, u32> pitch_high;
@@ -202,13 +204,24 @@ struct TICEntry {
         return depth_minus_1 + 1;
     }
 
+    u32 BlockWidth() const {
+        ASSERT(IsTiled());
+        // The block height is stored in log2 format.
+        return 1 << block_width;
+    }
+
     u32 BlockHeight() const {
-        ASSERT(header_version == TICHeaderVersion::BlockLinear ||
-               header_version == TICHeaderVersion::BlockLinearColorKey);
+        ASSERT(IsTiled());
         // The block height is stored in log2 format.
         return 1 << block_height;
     }
 
+    u32 BlockDepth() const {
+        ASSERT(IsTiled());
+        // The block height is stored in log2 format.
+        return 1 << block_depth;
+    }
+
     bool IsTiled() const {
         return header_version == TICHeaderVersion::BlockLinear ||
                header_version == TICHeaderVersion::BlockLinearColorKey;
@@ -227,6 +240,17 @@ enum class WrapMode : u32 {
     MirrorOnceClampOGL = 7,
 };
 
+enum class DepthCompareFunc : u32 {
+    Never = 0,
+    Less = 1,
+    Equal = 2,
+    LessEqual = 3,
+    Greater = 4,
+    NotEqual = 5,
+    GreaterEqual = 6,
+    Always = 7,
+};
+
 enum class TextureFilter : u32 {
     Nearest = 1,
     Linear = 2,
@@ -244,7 +268,7 @@ struct TSCEntry {
         BitField<3, 3, WrapMode> wrap_v;
         BitField<6, 3, WrapMode> wrap_p;
         BitField<9, 1, u32> depth_compare_enabled;
-        BitField<10, 3, u32> depth_compare_func;
+        BitField<10, 3, DepthCompareFunc> depth_compare_func;
     };
     union {
         BitField<0, 2, TextureFilter> mag_filter;

src/video_core/utils.h

@@ -169,16 +169,20 @@ static void LabelGLObject(GLenum identifier, GLuint handle, VAddr addr,
     const std::string nice_addr = fmt::format("0x{:016x}", addr);
     std::string object_label;
 
-    switch (identifier) {
-    case GL_TEXTURE:
-        object_label = extra_info + "@" + nice_addr;
-        break;
-    case GL_PROGRAM:
-        object_label = "ShaderProgram@" + nice_addr;
-        break;
-    default:
-        object_label = fmt::format("Object(0x{:x})@{}", identifier, nice_addr);
-        break;
+    if (extra_info.empty()) {
+        switch (identifier) {
+        case GL_TEXTURE:
+            object_label = "Texture@" + nice_addr;
+            break;
+        case GL_PROGRAM:
+            object_label = "Shader@" + nice_addr;
+            break;
+        default:
+            object_label = fmt::format("Object(0x{:x})@{}", identifier, nice_addr);
+            break;
+        }
+    } else {
+        object_label = extra_info + '@' + nice_addr;
     }
     glObjectLabel(identifier, handle, -1, static_cast<const GLchar*>(object_label.c_str()));
 }

src/yuzu/bootmanager.cpp

@@ -110,6 +110,7 @@ GRenderWindow::GRenderWindow(QWidget* parent, EmuThread* emu_thread)
     std::string window_title = fmt::format("yuzu {} | {}-{}", Common::g_build_name,
                                            Common::g_scm_branch, Common::g_scm_desc);
     setWindowTitle(QString::fromStdString(window_title));
+    setAttribute(Qt::WA_AcceptTouchEvents);
 
     InputCommon::Init();
     InputCommon::StartJoystickEventHandler();
@@ -190,11 +191,17 @@ QByteArray GRenderWindow::saveGeometry() {
         return geometry;
 }
 
-qreal GRenderWindow::windowPixelRatio() {
+qreal GRenderWindow::windowPixelRatio() const {
     // windowHandle() might not be accessible until the window is displayed to screen.
     return windowHandle() ? windowHandle()->screen()->devicePixelRatio() : 1.0f;
 }
 
+std::pair<unsigned, unsigned> GRenderWindow::ScaleTouch(const QPointF pos) const {
+    const qreal pixel_ratio = windowPixelRatio();
+    return {static_cast<unsigned>(std::max(std::round(pos.x() * pixel_ratio), qreal{0.0})),
+            static_cast<unsigned>(std::max(std::round(pos.y() * pixel_ratio), qreal{0.0}))};
+}
+
 void GRenderWindow::closeEvent(QCloseEvent* event) {
     emit Closed();
     QWidget::closeEvent(event);
@@ -209,31 +216,81 @@ void GRenderWindow::keyReleaseEvent(QKeyEvent* event) {
 }
 
 void GRenderWindow::mousePressEvent(QMouseEvent* event) {
+    if (event->source() == Qt::MouseEventSynthesizedBySystem)
+        return; // touch input is handled in TouchBeginEvent
+
     auto pos = event->pos();
     if (event->button() == Qt::LeftButton) {
-        qreal pixelRatio = windowPixelRatio();
-        this->TouchPressed(static_cast<unsigned>(pos.x() * pixelRatio),
-                           static_cast<unsigned>(pos.y() * pixelRatio));
+        const auto [x, y] = ScaleTouch(pos);
+        this->TouchPressed(x, y);
     } else if (event->button() == Qt::RightButton) {
         InputCommon::GetMotionEmu()->BeginTilt(pos.x(), pos.y());
     }
 }
 
 void GRenderWindow::mouseMoveEvent(QMouseEvent* event) {
+    if (event->source() == Qt::MouseEventSynthesizedBySystem)
+        return; // touch input is handled in TouchUpdateEvent
+
     auto pos = event->pos();
-    qreal pixelRatio = windowPixelRatio();
-    this->TouchMoved(std::max(static_cast<unsigned>(pos.x() * pixelRatio), 0u),
-                     std::max(static_cast<unsigned>(pos.y() * pixelRatio), 0u));
+    const auto [x, y] = ScaleTouch(pos);
+    this->TouchMoved(x, y);
     InputCommon::GetMotionEmu()->Tilt(pos.x(), pos.y());
 }
 
 void GRenderWindow::mouseReleaseEvent(QMouseEvent* event) {
+    if (event->source() == Qt::MouseEventSynthesizedBySystem)
+        return; // touch input is handled in TouchEndEvent
+
     if (event->button() == Qt::LeftButton)
         this->TouchReleased();
     else if (event->button() == Qt::RightButton)
         InputCommon::GetMotionEmu()->EndTilt();
 }
 
+void GRenderWindow::TouchBeginEvent(const QTouchEvent* event) {
+    // TouchBegin always has exactly one touch point, so take the .first()
+    const auto [x, y] = ScaleTouch(event->touchPoints().first().pos());
+    this->TouchPressed(x, y);
+}
+
+void GRenderWindow::TouchUpdateEvent(const QTouchEvent* event) {
+    QPointF pos;
+    int active_points = 0;
+
+    // average all active touch points
+    for (const auto tp : event->touchPoints()) {
+        if (tp.state() & (Qt::TouchPointPressed | Qt::TouchPointMoved | Qt::TouchPointStationary)) {
+            active_points++;
+            pos += tp.pos();
+        }
+    }
+
+    pos /= active_points;
+
+    const auto [x, y] = ScaleTouch(pos);
+    this->TouchMoved(x, y);
+}
+
+void GRenderWindow::TouchEndEvent() {
+    this->TouchReleased();
+}
+
+bool GRenderWindow::event(QEvent* event) {
+    if (event->type() == QEvent::TouchBegin) {
+        TouchBeginEvent(static_cast<QTouchEvent*>(event));
+        return true;
+    } else if (event->type() == QEvent::TouchUpdate) {
+        TouchUpdateEvent(static_cast<QTouchEvent*>(event));
+        return true;
+    } else if (event->type() == QEvent::TouchEnd || event->type() == QEvent::TouchCancel) {
+        TouchEndEvent();
+        return true;
+    }
+
+    return QWidget::event(event);
+}
+
 void GRenderWindow::focusOutEvent(QFocusEvent* event) {
     QWidget::focusOutEvent(event);
     InputCommon::GetKeyboard()->ReleaseAllKeys();

src/yuzu/bootmanager.h

@@ -15,6 +15,7 @@
 
 class QKeyEvent;
 class QScreen;
+class QTouchEvent;
 
 class GGLWidgetInternal;
 class GMainWindow;
@@ -119,7 +120,7 @@ public:
     void restoreGeometry(const QByteArray& geometry); // overridden
     QByteArray saveGeometry();                        // overridden
 
-    qreal windowPixelRatio();
+    qreal windowPixelRatio() const;
 
     void closeEvent(QCloseEvent* event) override;
 
@@ -130,6 +131,8 @@ public:
     void mouseMoveEvent(QMouseEvent* event) override;
     void mouseReleaseEvent(QMouseEvent* event) override;
 
+    bool event(QEvent* event) override;
+
     void focusOutEvent(QFocusEvent* event) override;
 
     void OnClientAreaResized(unsigned width, unsigned height);
@@ -148,6 +151,11 @@ signals:
     void Closed();
 
 private:
+    std::pair<unsigned, unsigned> ScaleTouch(const QPointF pos) const;
+    void TouchBeginEvent(const QTouchEvent* event);
+    void TouchUpdateEvent(const QTouchEvent* event);
+    void TouchEndEvent();
+
     void OnMinimalClientAreaChangeRequest(
         const std::pair<unsigned, unsigned>& minimal_size) override;
 

src/yuzu/configuration/config.cpp

@@ -134,6 +134,7 @@ void Config::ReadValues() {
     qt_config->beginGroup("Debugging");
     Settings::values.use_gdbstub = qt_config->value("use_gdbstub", false).toBool();
     Settings::values.gdbstub_port = qt_config->value("gdbstub_port", 24689).toInt();
+    Settings::values.program_args = qt_config->value("program_args", "").toString().toStdString();
     qt_config->endGroup();
 
     qt_config->beginGroup("WebService");
@@ -269,6 +270,7 @@ void Config::SaveValues() {
     qt_config->beginGroup("Debugging");
     qt_config->setValue("use_gdbstub", Settings::values.use_gdbstub);
     qt_config->setValue("gdbstub_port", Settings::values.gdbstub_port);
+    qt_config->setValue("program_args", QString::fromStdString(Settings::values.program_args));
     qt_config->endGroup();
 
     qt_config->beginGroup("WebService");

src/yuzu/configuration/configure_debug.cpp

@@ -33,6 +33,7 @@ void ConfigureDebug::setConfiguration() {
     ui->toggle_console->setEnabled(!Core::System::GetInstance().IsPoweredOn());
     ui->toggle_console->setChecked(UISettings::values.show_console);
     ui->log_filter_edit->setText(QString::fromStdString(Settings::values.log_filter));
+    ui->homebrew_args_edit->setText(QString::fromStdString(Settings::values.program_args));
 }
 
 void ConfigureDebug::applyConfiguration() {
@@ -40,6 +41,7 @@ void ConfigureDebug::applyConfiguration() {
     Settings::values.gdbstub_port = ui->gdbport_spinbox->value();
     UISettings::values.show_console = ui->toggle_console->isChecked();
     Settings::values.log_filter = ui->log_filter_edit->text().toStdString();
+    Settings::values.program_args = ui->homebrew_args_edit->text().toStdString();
     Debugger::ToggleConsole();
     Log::Filter filter;
     filter.ParseFilterString(Settings::values.log_filter);

src/yuzu/configuration/configure_debug.ui

@@ -106,6 +106,29 @@
      </layout>
     </widget>
    </item>
+   <item>
+    <widget class="QGroupBox" name="groupBox_3">
+     <property name="title">
+      <string>Homebrew</string>
+     </property>
+     <layout class="QVBoxLayout" name="verticalLayout">
+      <item>
+       <layout class="QHBoxLayout" name="horizontalLayout">
+        <item>
+         <widget class="QLabel" name="label">
+          <property name="text">
+           <string>Arguments String</string>
+          </property>
+         </widget>
+        </item>
+        <item>
+         <widget class="QLineEdit" name="homebrew_args_edit"/>
+        </item>
+       </layout>
+      </item>
+     </layout>
+    </widget>
+   </item>
    <item>
     <spacer name="verticalSpacer">
      <property name="orientation">

src/yuzu/configuration/configure_input.cpp

@@ -5,6 +5,7 @@
 #include <algorithm>
 #include <memory>
 #include <utility>
+#include <QMenu>
 #include <QMessageBox>
 #include <QTimer>
 #include "common/param_package.h"
@@ -128,28 +129,63 @@ ConfigureInput::ConfigureInput(QWidget* parent)
     analog_map_stick = {ui->buttonLStickAnalog, ui->buttonRStickAnalog};
 
     for (int button_id = 0; button_id < Settings::NativeButton::NumButtons; button_id++) {
-        if (button_map[button_id])
-            connect(button_map[button_id], &QPushButton::released, [=]() {
-                handleClick(
-                    button_map[button_id],
-                    [=](const Common::ParamPackage& params) { buttons_param[button_id] = params; },
-                    InputCommon::Polling::DeviceType::Button);
-            });
+        if (!button_map[button_id])
+            continue;
+        button_map[button_id]->setContextMenuPolicy(Qt::CustomContextMenu);
+        connect(button_map[button_id], &QPushButton::released, [=]() {
+            handleClick(
+                button_map[button_id],
+                [=](const Common::ParamPackage& params) { buttons_param[button_id] = params; },
+                InputCommon::Polling::DeviceType::Button);
+        });
+        connect(button_map[button_id], &QPushButton::customContextMenuRequested,
+                [=](const QPoint& menu_location) {
+                    QMenu context_menu;
+                    context_menu.addAction(tr("Clear"), [&] {
+                        buttons_param[button_id].Clear();
+                        button_map[button_id]->setText(tr("[not set]"));
+                    });
+                    context_menu.addAction(tr("Restore Default"), [&] {
+                        buttons_param[button_id] = Common::ParamPackage{
+                            InputCommon::GenerateKeyboardParam(Config::default_buttons[button_id])};
+                        button_map[button_id]->setText(ButtonToText(buttons_param[button_id]));
+                    });
+                    context_menu.exec(button_map[button_id]->mapToGlobal(menu_location));
+                });
     }
 
     for (int analog_id = 0; analog_id < Settings::NativeAnalog::NumAnalogs; analog_id++) {
         for (int sub_button_id = 0; sub_button_id < ANALOG_SUB_BUTTONS_NUM; sub_button_id++) {
-            if (analog_map_buttons[analog_id][sub_button_id] != nullptr) {
-                connect(analog_map_buttons[analog_id][sub_button_id], &QPushButton::released,
-                        [=]() {
-                            handleClick(analog_map_buttons[analog_id][sub_button_id],
-                                        [=](const Common::ParamPackage& params) {
-                                            SetAnalogButton(params, analogs_param[analog_id],
-                                                            analog_sub_buttons[sub_button_id]);
-                                        },
-                                        InputCommon::Polling::DeviceType::Button);
+            if (!analog_map_buttons[analog_id][sub_button_id])
+                continue;
+            analog_map_buttons[analog_id][sub_button_id]->setContextMenuPolicy(
+                Qt::CustomContextMenu);
+            connect(analog_map_buttons[analog_id][sub_button_id], &QPushButton::released, [=]() {
+                handleClick(analog_map_buttons[analog_id][sub_button_id],
+                            [=](const Common::ParamPackage& params) {
+                                SetAnalogButton(params, analogs_param[analog_id],
+                                                analog_sub_buttons[sub_button_id]);
+                            },
+                            InputCommon::Polling::DeviceType::Button);
+            });
+            connect(analog_map_buttons[analog_id][sub_button_id],
+                    &QPushButton::customContextMenuRequested, [=](const QPoint& menu_location) {
+                        QMenu context_menu;
+                        context_menu.addAction(tr("Clear"), [&] {
+                            analogs_param[analog_id].Erase(analog_sub_buttons[sub_button_id]);
+                            analog_map_buttons[analog_id][sub_button_id]->setText(tr("[not set]"));
                         });
-            }
+                        context_menu.addAction(tr("Restore Default"), [&] {
+                            Common::ParamPackage params{InputCommon::GenerateKeyboardParam(
+                                Config::default_analogs[analog_id][sub_button_id])};
+                            SetAnalogButton(params, analogs_param[analog_id],
+                                            analog_sub_buttons[sub_button_id]);
+                            analog_map_buttons[analog_id][sub_button_id]->setText(AnalogToText(
+                                analogs_param[analog_id], analog_sub_buttons[sub_button_id]));
+                        });
+                        context_menu.exec(analog_map_buttons[analog_id][sub_button_id]->mapToGlobal(
+                            menu_location));
+                    });
         }
         connect(analog_map_stick[analog_id], &QPushButton::released, [=]() {
             QMessageBox::information(this, tr("Information"),
@@ -162,6 +198,7 @@ ConfigureInput::ConfigureInput(QWidget* parent)
         });
     }
 
+    connect(ui->buttonClearAll, &QPushButton::released, [this] { ClearAll(); });
     connect(ui->buttonRestoreDefaults, &QPushButton::released, [this]() { restoreDefaults(); });
 
     timeout_timer->setSingleShot(true);
@@ -215,7 +252,21 @@ void ConfigureInput::restoreDefaults() {
         }
     }
     updateButtonLabels();
-    applyConfiguration();
+}
+
+void ConfigureInput::ClearAll() {
+    for (int button_id = 0; button_id < Settings::NativeButton::NumButtons; button_id++) {
+        if (button_map[button_id] && button_map[button_id]->isEnabled())
+            buttons_param[button_id].Clear();
+    }
+    for (int analog_id = 0; analog_id < Settings::NativeAnalog::NumAnalogs; analog_id++) {
+        for (int sub_button_id = 0; sub_button_id < ANALOG_SUB_BUTTONS_NUM; sub_button_id++) {
+            if (analog_map_buttons[analog_id][sub_button_id] &&
+                analog_map_buttons[analog_id][sub_button_id]->isEnabled())
+                analogs_param[analog_id].Erase(analog_sub_buttons[sub_button_id]);
+        }
+    }
+    updateButtonLabels();
 }
 
 void ConfigureInput::updateButtonLabels() {

src/yuzu/configuration/configure_input.h

@@ -72,6 +72,9 @@ private:
     void loadConfiguration();
     /// Restore all buttons to their default values.
     void restoreDefaults();
+    /// Clear all input configuration
+    void ClearAll();
+
     /// Update UI to reflect current configuration.
     void updateButtonLabels();
 

src/yuzu/configuration/configure_input.ui

@@ -694,6 +694,34 @@ Capture:</string>
        </property>
       </spacer>
      </item>
+     <item>
+      <widget class="QPushButton" name="buttonClearAll">
+       <property name="sizePolicy">
+        <sizepolicy hsizetype="Minimum" vsizetype="Fixed">
+         <horstretch>0</horstretch>
+         <verstretch>0</verstretch>
+        </sizepolicy>
+       </property>
+       <property name="sizeIncrement">
+        <size>
+         <width>0</width>
+         <height>0</height>
+        </size>
+       </property>
+       <property name="baseSize">
+        <size>
+         <width>0</width>
+         <height>0</height>
+        </size>
+       </property>
+       <property name="layoutDirection">
+        <enum>Qt::LeftToRight</enum>
+       </property>
+       <property name="text">
+        <string>Clear All</string>
+       </property>
+      </widget>
+     </item>
      <item>
       <widget class="QPushButton" name="buttonRestoreDefaults">
        <property name="sizePolicy">

src/yuzu/game_list_worker.cpp

@@ -27,9 +27,8 @@
 #include "yuzu/ui_settings.h"
 
 namespace {
-void GetMetadataFromControlNCA(const FileSys::PatchManager& patch_manager,
-                               const std::shared_ptr<FileSys::NCA>& nca, std::vector<u8>& icon,
-                               std::string& name) {
+void GetMetadataFromControlNCA(const FileSys::PatchManager& patch_manager, const FileSys::NCA& nca,
+                               std::vector<u8>& icon, std::string& name) {
     auto [nacp, icon_file] = patch_manager.ParseControlNCA(nca);
     if (icon_file != nullptr)
         icon = icon_file->ReadAllBytes();
@@ -57,16 +56,25 @@ QString FormatGameName(const std::string& physical_name) {
     return physical_name_as_qstring;
 }
 
-QString FormatPatchNameVersions(const FileSys::PatchManager& patch_manager, bool updatable = true) {
+QString FormatPatchNameVersions(const FileSys::PatchManager& patch_manager,
+                                Loader::AppLoader& loader, bool updatable = true) {
     QString out;
-    for (const auto& kv : patch_manager.GetPatchVersionNames()) {
+    FileSys::VirtualFile update_raw;
+    loader.ReadUpdateRaw(update_raw);
+    for (const auto& kv : patch_manager.GetPatchVersionNames(update_raw)) {
         if (!updatable && kv.first == "Update")
             continue;
 
         if (kv.second.empty()) {
             out.append(fmt::format("{}\n", kv.first).c_str());
         } else {
-            out.append(fmt::format("{} ({})\n", kv.first, kv.second).c_str());
+            auto ver = kv.second;
+
+            // Display container name for packed updates
+            if (ver == "PACKED" && kv.first == "Update")
+                ver = Loader::GetFileTypeString(loader.GetFileType());
+
+            out.append(fmt::format("{} ({})\n", kv.first, ver).c_str());
         }
     }
 
@@ -101,7 +109,7 @@ void GameListWorker::AddInstalledTitlesToGameList() {
         const FileSys::PatchManager patch{program_id};
         const auto& control = cache->GetEntry(game.title_id, FileSys::ContentRecordType::Control);
         if (control != nullptr)
-            GetMetadataFromControlNCA(patch, control, icon, name);
+            GetMetadataFromControlNCA(patch, *control, icon, name);
 
         auto it = FindMatchingCompatibilityEntry(compatibility_list, program_id);
 
@@ -116,7 +124,7 @@ void GameListWorker::AddInstalledTitlesToGameList() {
                 QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType())),
                 program_id),
             new GameListItemCompat(compatibility),
-            new GameListItem(FormatPatchNameVersions(patch)),
+            new GameListItem(FormatPatchNameVersions(patch, *loader)),
             new GameListItem(
                 QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))),
             new GameListItemSize(file->GetSize()),
@@ -188,8 +196,8 @@ void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsign
                 res2 == Loader::ResultStatus::Success) {
                 // Use from metadata pool.
                 if (nca_control_map.find(program_id) != nca_control_map.end()) {
-                    const auto nca = nca_control_map[program_id];
-                    GetMetadataFromControlNCA(patch, nca, icon, name);
+                    const auto& nca = nca_control_map[program_id];
+                    GetMetadataFromControlNCA(patch, *nca, icon, name);
                 }
             }
 
@@ -206,7 +214,8 @@ void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsign
                     QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType())),
                     program_id),
                 new GameListItemCompat(compatibility),
-                new GameListItem(FormatPatchNameVersions(patch, loader->IsRomFSUpdatable())),
+                new GameListItem(
+                    FormatPatchNameVersions(patch, *loader, loader->IsRomFSUpdatable())),
                 new GameListItem(
                     QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))),
                 new GameListItemSize(FileUtil::GetSize(physical_name)),

src/yuzu/main.cpp

@@ -31,6 +31,7 @@ static FileSys::VirtualFile VfsDirectoryCreateFileWrapper(const FileSys::Virtual
 #include <QDialogButtonBox>
 #include <QFileDialog>
 #include <QMessageBox>
+#include <QtConcurrent/QtConcurrent>
 #include <QtGui>
 #include <QtWidgets>
 #include <fmt/format.h>
@@ -171,6 +172,9 @@ GMainWindow::GMainWindow()
                        .arg(Common::g_build_fullname, Common::g_scm_branch, Common::g_scm_desc));
     show();
 
+    // Gen keys if necessary
+    OnReinitializeKeys(ReinitializeKeyBehavior::NoWarning);
+
     // Necessary to load titles from nand in gamelist.
     Service::FileSystem::CreateFactories(vfs);
     game_list->LoadCompatibilityList();
@@ -443,6 +447,8 @@ void GMainWindow::ConnectMenuEvents() {
     connect(ui.action_Fullscreen, &QAction::triggered, this, &GMainWindow::ToggleFullscreen);
 
     // Help
+    connect(ui.action_Rederive, &QAction::triggered, this,
+            std::bind(&GMainWindow::OnReinitializeKeys, this, ReinitializeKeyBehavior::Warning));
     connect(ui.action_About, &QAction::triggered, this, &GMainWindow::OnAbout);
 }
 
@@ -485,6 +491,8 @@ QStringList GMainWindow::GetUnsupportedGLExtensions() {
         unsupported_ext.append("ARB_texture_storage");
     if (!GLAD_GL_ARB_multi_bind)
         unsupported_ext.append("ARB_multi_bind");
+    if (!GLAD_GL_ARB_copy_image)
+        unsupported_ext.append("ARB_copy_image");
 
     // Extensions required to support some texture formats.
     if (!GLAD_GL_EXT_texture_compression_s3tc)
@@ -1373,6 +1381,82 @@ void GMainWindow::OnCoreError(Core::System::ResultStatus result, std::string det
     }
 }
 
+void GMainWindow::OnReinitializeKeys(ReinitializeKeyBehavior behavior) {
+    if (behavior == ReinitializeKeyBehavior::Warning) {
+        const auto res = QMessageBox::information(
+            this, tr("Confirm Key Rederivation"),
+            tr("You are about to force rederive all of your keys. \nIf you do not know what this "
+               "means or what you are doing, \nthis is a potentially destructive action. \nPlease "
+               "make "
+               "sure this is what you want \nand optionally make backups.\n\nThis will delete your "
+               "autogenerated key files and re-run the key derivation module."),
+            QMessageBox::StandardButtons{QMessageBox::Ok, QMessageBox::Cancel});
+
+        if (res == QMessageBox::Cancel)
+            return;
+
+        FileUtil::Delete(FileUtil::GetUserPath(FileUtil::UserPath::KeysDir) +
+                         "prod.keys_autogenerated");
+        FileUtil::Delete(FileUtil::GetUserPath(FileUtil::UserPath::KeysDir) +
+                         "console.keys_autogenerated");
+        FileUtil::Delete(FileUtil::GetUserPath(FileUtil::UserPath::KeysDir) +
+                         "title.keys_autogenerated");
+    }
+
+    Core::Crypto::KeyManager keys{};
+    if (keys.BaseDeriveNecessary()) {
+        Core::Crypto::PartitionDataManager pdm{vfs->OpenDirectory(
+            FileUtil::GetUserPath(FileUtil::UserPath::SysDataDir), FileSys::Mode::Read)};
+
+        const auto function = [this, &keys, &pdm] {
+            keys.PopulateFromPartitionData(pdm);
+            Service::FileSystem::CreateFactories(vfs);
+            keys.DeriveETicket(pdm);
+        };
+
+        QString errors;
+
+        if (!pdm.HasFuses())
+            errors += tr("- Missing fuses - Cannot derive SBK\n");
+        if (!pdm.HasBoot0())
+            errors += tr("- Missing BOOT0 - Cannot derive master keys\n");
+        if (!pdm.HasPackage2())
+            errors += tr("- Missing BCPKG2-1-Normal-Main - Cannot derive general keys\n");
+        if (!pdm.HasProdInfo())
+            errors += tr("- Missing PRODINFO - Cannot derive title keys\n");
+
+        if (!errors.isEmpty()) {
+
+            QMessageBox::warning(
+                this, tr("Warning Missing Derivation Components"),
+                tr("The following are missing from your configuration that may hinder key "
+                   "derivation. It will be attempted but may not complete.\n\n") +
+                    errors);
+        }
+
+        QProgressDialog prog;
+        prog.setRange(0, 0);
+        prog.setLabelText(tr("Deriving keys...\nThis may take up to a minute depending \non your "
+                             "system's performance."));
+        prog.setWindowTitle(tr("Deriving Keys"));
+
+        prog.show();
+
+        auto future = QtConcurrent::run(function);
+        while (!future.isFinished()) {
+            QCoreApplication::processEvents();
+        }
+
+        prog.close();
+    }
+
+    Service::FileSystem::CreateFactories(vfs);
+
+    if (behavior == ReinitializeKeyBehavior::Warning) {
+        game_list->PopulateAsync(UISettings::values.gamedir, UISettings::values.gamedir_deepscan);
+    }
+}
+
 bool GMainWindow::ConfirmClose() {
     if (emu_thread == nullptr || !UISettings::values.confirm_before_closing)
         return true;

src/yuzu/main.h

@@ -41,6 +41,11 @@ enum class EmulatedDirectoryTarget {
     SDMC,
 };
 
+enum class ReinitializeKeyBehavior {
+    NoWarning,
+    Warning,
+};
+
 namespace DiscordRPC {
 class DiscordInterface;
 }
@@ -167,6 +172,7 @@ private slots:
     void HideFullscreen();
     void ToggleWindowMode();
     void OnCoreError(Core::System::ResultStatus, std::string);
+    void OnReinitializeKeys(ReinitializeKeyBehavior behavior);
 
 private:
     void UpdateStatusBar();