diff options
53 files changed, 2536 insertions, 434 deletions
diff --git a/externals/mbedtls b/externals/mbedtls -Subproject d409b75a4cf75a5b358b352c75826ddbca44db5 +Subproject a280e602f3a4ae001d3a83cbc3e6e04c99c2227 diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt index 194e5f3f7..78986deb5 100644 --- a/src/core/CMakeLists.txt +++ b/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 diff --git a/src/core/core.cpp b/src/core/core.cpp index e2fb9e038..32baa40dc 100644 --- a/src/core/core.cpp +++ b/src/core/core.cpp @@ -148,7 +148,7 @@ struct System::Impl { telemetry_session = std::make_unique<Core::TelemetrySession>(); service_manager = std::make_shared<Service::SM::ServiceManager>(); - Service::Init(service_manager, virtual_filesystem); + Service::Init(service_manager, *virtual_filesystem); GDBStub::Init(); renderer = VideoCore::CreateRenderer(emu_window); diff --git a/src/core/crypto/key_manager.cpp b/src/core/crypto/key_manager.cpp index bf3a70944..fd0786068 100644 --- a/src/core/crypto/key_manager.cpp +++ b/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{}; @@ -37,57 +70,136 @@ 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(std::size_t 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", "rb+"); if (!save_43.IsOpen()) return boost::none; + const FileUtil::IOFile sd_private( FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir) + "/Nintendo/Contents/private", "rb+"); if (!sd_private.IsOpen()) return boost::none; - sd_private.Seek(0, SEEK_SET); std::array<u8, 0x10> private_seed{}; - if (sd_private.ReadBytes(private_seed.data(), private_seed.size()) != 0x10) + if (sd_private.ReadBytes(private_seed.data(), private_seed.size()) != private_seed.size()) { return boost::none; + } std::array<u8, 0x10> buffer{}; std::size_t offset = 0; for (; offset + 0x10 < save_43.GetSize(); ++offset) { - save_43.Seek(offset, SEEK_SET); + if (!save_43.Seek(offset, SEEK_SET)) { + return boost::none; + } + save_43.ReadBytes(buffer.data(), buffer.size()); - if (buffer == private_seed) + if (buffer == private_seed) { break; + } } - if (offset + 0x10 >= save_43.GetSize()) + if (!save_43.Seek(offset + 0x10, SEEK_SET)) { return boost::none; + } Key128 seed{}; - save_43.Seek(offset + 0x10, SEEK_SET); - save_43.ReadBytes(seed.data(), seed.size()); + if (save_43.ReadBytes(seed.data(), seed.size()) != seed.size()) { + return boost::none; + } return seed; } -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 +230,147 @@ 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()); + if (ticket_save.ReadBytes(buffer.data(), buffer.size()) != buffer.size()) { + return {}; + } + + 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) { + // Avoids truncation overflow within the loop below. + static_assert(target_size <= 0xFF); + + std::array<u8, in_size + 4> seed_exp{}; + std::memcpy(seed_exp.data(), seed.data(), in_size); + + std::vector<u8> out; + size_t i = 0; + while (out.size() < target_size) { + out.resize(out.size() + 0x20); + seed_exp[in_size + 3] = static_cast<u8>(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 +387,15 @@ 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 +417,9 @@ 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 +436,50 @@ 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 +493,28 @@ 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 +535,30 @@ 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 +572,7 @@ 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 +582,15 @@ 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 +598,30 @@ 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 +635,7 @@ 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 +665,388 @@ 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+"); + + const auto blob2 = GetTicketblob(save2); + auto res = GetTicketblob(save1); + res.insert(res.end(), blob2.begin(), blob2.end()); + + 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 diff --git a/src/core/crypto/key_manager.h b/src/core/crypto/key_manager.h index 978eec8dc..cccb3c0ae 100644 --- a/src/core/crypto/key_manager.h +++ b/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(std::size_t crypto_revision); + + void SetKeyWrapped(S128KeyType id, Key128 key, u64 field1 = 0, u64 field2 = 0); + void SetKeyWrapped(S256KeyType id, Key256 key, u64 field1 = 0, u64 field2 = 0); 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 diff --git a/src/core/crypto/partition_data_manager.cpp b/src/core/crypto/partition_data_manager.cpp new file mode 100644 index 000000000..25cee1f3a --- /dev/null +++ b/src/core/crypto/partition_data_manager.cpp @@ -0,0 +1,593 @@ +// 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 <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 "common/swap.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(const FileSys::VirtualDir& sysdata_dir) + : boot0(FindFileInDirWithNames(sysdata_dir, "BOOT0")), + fuses(FindFileInDirWithNames(sysdata_dir, "fuses")), + kfuses(FindFileInDirWithNames(sysdata_dir, "kfuses")), + package2({ + FindFileInDirWithNames(sysdata_dir, "BCPKG2-1-Normal-Main"), + FindFileInDirWithNames(sysdata_dir, "BCPKG2-2-Normal-Sub"), + 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"), + }), + prodinfo(FindFileInDirWithNames(sysdata_dir, "PRODINFO")), + secure_monitor(FindFileInDirWithNames(sysdata_dir, "secmon")), + package1_decrypted(FindFileInDirWithNames(sysdata_dir, "pkg1_decr")), + secure_monitor_bytes(secure_monitor == nullptr ? std::vector<u8>{} + : secure_monitor->ReadAllBytes()), + package1_decrypted_bytes(package1_decrypted == nullptr ? std::vector<u8>{} + : package1_decrypted->ReadAllBytes()) { +} + +PartitionDataManager::~PartitionDataManager() = default; + +bool PartitionDataManager::HasBoot0() const { + return boot0 != nullptr; +} + +FileSys::VirtualFile PartitionDataManager::GetBoot0Raw() const { + return boot0; +} + +PartitionDataManager::EncryptedKeyBlob PartitionDataManager::GetEncryptedKeyblob( + std::size_t index) const { + if (HasBoot0() && index < NUM_ENCRYPTED_KEYBLOBS) + return GetEncryptedKeyblobs()[index]; + return {}; +} + +PartitionDataManager::EncryptedKeyBlobs PartitionDataManager::GetEncryptedKeyblobs() const { + if (!HasBoot0()) + return {}; + + EncryptedKeyBlobs out{}; + for (size_t i = 0; i < out.size(); ++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(std::size_t revision) const { + if (keyblob_source_hashes[revision] == SHA256Hash{}) { + LOG_WARNING(Crypto, + "No keyblob source hash for crypto revision {:02X}! Cannot derive keys...", + 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(const std::array<Key128, 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; + + std::size_t revision = 0xFF; + if (header.magic != Common::MakeMagic('P', 'K', '2', '1')) { + for (std::size_t i = 0; i < package2_keys.size(); ++i) { + if (AttemptDecrypt(package2_keys[i], header)) { + revision = i; + } + } + } + + if (header.magic != Common::MakeMagic('P', 'K', '2', '1')) + return; + + const 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({header.section_ctr[1].begin(), header.section_ctr[1].end()}); + cipher.Transcode(c.data(), c.size(), c.data(), Op::Decrypt); + + INIHeader ini; + std::memcpy(&ini, c.data(), sizeof(INIHeader)); + if (ini.magic != Common::MakeMagic('I', 'N', 'I', '1')) + return; + + 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; + + 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; + } + + const u64 initial_offset = sizeof(KIPHeader) + offset; + const auto text_begin = c.cbegin() + initial_offset; + const auto text_end = text_begin + kip.sections[0].size_compressed; + const std::vector<u8> text = DecompressBLZ({text_begin, text_end}); + + const auto rodata_end = text_end + kip.sections[1].size_compressed; + const std::vector<u8> rodata = DecompressBLZ({text_end, rodata_end}); + + const auto data_end = rodata_end + kip.sections[2].size_compressed; + const std::vector<u8> data = DecompressBLZ({rodata_end, data_end}); + + std::vector<u8> out; + out.reserve(text.size() + rodata.size() + data.size()); + out.insert(out.end(), text.begin(), text.end()); + out.insert(out.end(), rodata.begin(), rodata.end()); + out.insert(out.end(), data.begin(), data.end()); + + offset += sizeof(KIPHeader) + out.size(); + + if (name == "FS") + package2_fs[static_cast<size_t>(type)] = std::move(out); + else if (name == "spl") + package2_spl[static_cast<size_t>(type)] = std::move(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 diff --git a/src/core/crypto/partition_data_manager.h b/src/core/crypto/partition_data_manager.h new file mode 100644 index 000000000..0ad007c72 --- /dev/null +++ b/src/core/crypto/partition_data_manager.h @@ -0,0 +1,109 @@ +// 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_types.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; + static constexpr std::size_t NUM_ENCRYPTED_KEYBLOBS = 32; + static constexpr std::size_t ENCRYPTED_KEYBLOB_SIZE = 0xB0; + + using EncryptedKeyBlob = std::array<u8, ENCRYPTED_KEYBLOB_SIZE>; + using EncryptedKeyBlobs = std::array<EncryptedKeyBlob, NUM_ENCRYPTED_KEYBLOBS>; + + explicit PartitionDataManager(const FileSys::VirtualDir& sysdata_dir); + ~PartitionDataManager(); + + // BOOT0 + bool HasBoot0() const; + FileSys::VirtualFile GetBoot0Raw() const; + EncryptedKeyBlob GetEncryptedKeyblob(std::size_t index) const; + EncryptedKeyBlobs GetEncryptedKeyblobs() const; + std::vector<u8> GetSecureMonitor() const; + std::array<u8, 0x10> GetPackage2KeySource() const; + std::array<u8, 0x10> GetAESKekGenerationSource() const; + 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(std::size_t 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(const std::array<std::array<u8, 16>, 0x20>& package2_keys, + Package2Type type); + const std::vector<u8>& GetPackage2FSDecompressed( + Package2Type type = Package2Type::NormalMain) const; + std::array<u8, 0x10> GetKeyAreaKeyApplicationSource( + 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 diff --git a/src/core/file_sys/control_metadata.cpp b/src/core/file_sys/control_metadata.cpp index 5b1177a03..a012c2be9 100644 --- a/src/core/file_sys/control_metadata.cpp +++ b/src/core/file_sys/control_metadata.cpp @@ -17,11 +17,13 @@ const std::array<const char*, 15> LANGUAGE_NAMES = { }; std::string LanguageEntry::GetApplicationName() const { - return Common::StringFromFixedZeroTerminatedBuffer(application_name.data(), 0x200); + return Common::StringFromFixedZeroTerminatedBuffer(application_name.data(), + application_name.size()); } std::string LanguageEntry::GetDeveloperName() const { - return Common::StringFromFixedZeroTerminatedBuffer(developer_name.data(), 0x100); + return Common::StringFromFixedZeroTerminatedBuffer(developer_name.data(), + developer_name.size()); } NACP::NACP(VirtualFile file) : raw(std::make_unique<RawNACP>()) { @@ -56,7 +58,12 @@ u64 NACP::GetTitleId() const { return raw->title_id; } +u64 NACP::GetDLCBaseTitleId() const { + return raw->dlc_base_title_id; +} + std::string NACP::GetVersionString() const { - return Common::StringFromFixedZeroTerminatedBuffer(raw->version_string.data(), 0x10); + return Common::StringFromFixedZeroTerminatedBuffer(raw->version_string.data(), + raw->version_string.size()); } } // namespace FileSys diff --git a/src/core/file_sys/control_metadata.h b/src/core/file_sys/control_metadata.h index 43d6f0719..141f7e056 100644 --- a/src/core/file_sys/control_metadata.h +++ b/src/core/file_sys/control_metadata.h @@ -79,6 +79,7 @@ public: std::string GetApplicationName(Language language = Language::Default) const; std::string GetDeveloperName(Language language = Language::Default) const; u64 GetTitleId() const; + u64 GetDLCBaseTitleId() const; std::string GetVersionString() const; private: diff --git a/src/core/file_sys/vfs.h b/src/core/file_sys/vfs.h index 270291631..7f0d520ca 100644 --- a/src/core/file_sys/vfs.h +++ b/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, diff --git a/src/core/file_sys/vfs_types.h b/src/core/file_sys/vfs_types.h new file mode 100644 index 000000000..6215ed7af --- /dev/null +++ b/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 diff --git a/src/core/hle/kernel/errors.h b/src/core/hle/kernel/errors.h index e5fa67ae8..885259618 100644 --- a/src/core/hle/kernel/errors.h +++ b/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); diff --git a/src/core/hle/kernel/object.cpp b/src/core/hle/kernel/object.cpp index d51562d92..d87a62bb9 100644 --- a/src/core/hle/kernel/object.cpp +++ b/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; diff --git a/src/core/hle/kernel/object.h b/src/core/hle/kernel/object.h index 9eb72315c..c9f4d0bb3 100644 --- a/src/core/hle/kernel/object.h +++ b/src/core/hle/kernel/object.h @@ -26,7 +26,6 @@ enum class HandleType : u32 { AddressArbiter, Timer, ResourceLimit, - CodeSet, ClientPort, ServerPort, ClientSession, diff --git a/src/core/hle/kernel/process.cpp b/src/core/hle/kernel/process.cpp index fb0027a71..c80b2c507 100644 --- a/src/core/hle/kernel/process.cpp +++ b/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) { diff --git a/src/core/hle/kernel/process.h b/src/core/hle/kernel/process.h index 590e0c73d..f2816943a 100644 --- a/src/core/hle/kernel/process.h +++ b/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 }; - -class ResourceLimit; +/** + * Indicates the status of a Process instance. + * + * @note These match the values as used by kernel, + * so new entries should only be added if RE + * shows that a new value has been introduced. + */ +enum class ProcessStatus { + Created, + CreatedWithDebuggerAttached, + Running, + WaitingForDebuggerToAttach, + DebuggerAttached, + Exiting, + Exited, + DebugBreak, +}; -struct CodeSet final : 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 diff --git a/src/core/hle/kernel/svc.cpp b/src/core/hle/kernel/svc.cpp index 3afcce3fe..7a053da1e 100644 --- a/src/core/hle/kernel/svc.cpp +++ b/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(); - if (size == 0 || !Is4KBAligned(size)) { - return ERR_INVALID_SIZE; + const auto result = MapUnmapMemorySanityChecks(vm_manager, dst_addr, src_addr, size); + if (result != RESULT_SUCCESS) { + return result; } - 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(); - if (size == 0 || !Is4KBAligned(size)) { - return ERR_INVALID_SIZE; + const auto result = MapUnmapMemorySanityChecks(vm_manager, dst_addr, src_addr, size); + if (result != RESULT_SUCCESS) { + return result; } - 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 @@ -303,15 +370,15 @@ static ResultCode ArbitrateUnlock(VAddr mutex_addr) { struct BreakReason { union { - u64 raw; - BitField<31, 1, u64> dont_kill_application; + u32 raw; + BitField<31, 1, u32> signal_debugger; }; }; /// Break program execution -static void Break(u64 reason, u64 info1, u64 info2) { +static void Break(u32 reason, u64 info1, u64 info2) { BreakReason break_reason{reason}; - if (break_reason.dont_kill_application) { + if (break_reason.signal_debugger) { LOG_ERROR( Debug_Emulated, "Emulated program broke execution! reason=0x{:016X}, info1=0x{:016X}, info2=0x{:016X}", @@ -322,6 +389,12 @@ static void Break(u64 reason, u64 info1, u64 info2) { "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(); } } @@ -1025,6 +1098,29 @@ 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(); @@ -1160,7 +1256,7 @@ 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"}, diff --git a/src/core/hle/kernel/svc_wrap.h b/src/core/hle/kernel/svc_wrap.h index 22712e64f..b09753c80 100644 --- a/src/core/hle/kernel/svc_wrap.h +++ b/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(¶m_1, (u32)Param(1)).raw; + u32 retval = func(¶m_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(¶m_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), ¶m_1, ¶m_2); + ResultCode retval = func(static_cast<u32>(Param(2)), ¶m_1, ¶m_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(¶m_1, Param(1), (u32)(Param(2) & 0xFFFFFFFF), (s64)Param(3)); + ResultCode retval = + func(¶m_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(¶m_1, Param(1), Param(2), Param(3), (u32)Param(4), (s32)(Param(5) & 0xFFFFFFFF)) - .raw; + u32 retval = func(¶m_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(¶m_1, Param(1), Param(2), (u32)(Param(3) & 0xFFFFFFFF)).raw; + u32 retval = func(¶m_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(¶m_1, Param(1), (u32)(Param(2) & 0xFFFFFFFF), (u32)(Param(3) & 0xFFFFFFFF)).raw; + func(¶m_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 diff --git a/src/core/hle/kernel/thread.cpp b/src/core/hle/kernel/thread.cpp index 33aed8c23..352ce1725 100644 --- a/src/core/hle/kernel/thread.cpp +++ b/src/core/hle/kernel/thread.cpp @@ -183,13 +183,10 @@ 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, diff --git a/src/core/hle/service/aoc/aoc_u.cpp b/src/core/hle/service/aoc/aoc_u.cpp index 0ecfb5af1..518161bf7 100644 --- a/src/core/hle/service/aoc/aoc_u.cpp +++ b/src/core/hle/service/aoc/aoc_u.cpp @@ -7,8 +7,10 @@ #include <vector> #include "common/logging/log.h" #include "core/file_sys/content_archive.h" +#include "core/file_sys/control_metadata.h" #include "core/file_sys/nca_metadata.h" #include "core/file_sys/partition_filesystem.h" +#include "core/file_sys/patch_manager.h" #include "core/file_sys/registered_cache.h" #include "core/hle/ipc_helpers.h" #include "core/hle/kernel/process.h" @@ -19,7 +21,7 @@ namespace Service::AOC { constexpr u64 DLC_BASE_TITLE_ID_MASK = 0xFFFFFFFFFFFFE000; -constexpr u64 DLC_BASE_TO_AOC_ID_MASK = 0x1000; +constexpr u64 DLC_BASE_TO_AOC_ID = 0x1000; static bool CheckAOCTitleIDMatchesBase(u64 base, u64 aoc) { return (aoc & DLC_BASE_TITLE_ID_MASK) == base; @@ -97,14 +99,24 @@ void AOC_U::ListAddOnContent(Kernel::HLERequestContext& ctx) { ctx.WriteBuffer(out); - IPC::ResponseBuilder rb{ctx, 2}; + IPC::ResponseBuilder rb{ctx, 3}; rb.Push(RESULT_SUCCESS); + rb.Push(count); } void AOC_U::GetAddOnContentBaseId(Kernel::HLERequestContext& ctx) { IPC::ResponseBuilder rb{ctx, 4}; rb.Push(RESULT_SUCCESS); - rb.Push(Core::System::GetInstance().CurrentProcess()->GetTitleID() | DLC_BASE_TO_AOC_ID_MASK); + const auto title_id = Core::System::GetInstance().CurrentProcess()->GetTitleID(); + FileSys::PatchManager pm{title_id}; + + const auto res = pm.GetControlMetadata(); + if (res.first == nullptr) { + rb.Push(title_id + DLC_BASE_TO_AOC_ID); + return; + } + + rb.Push(res.first->GetDLCBaseTitleId()); } void AOC_U::PrepareAddOnContent(Kernel::HLERequestContext& ctx) { diff --git a/src/core/hle/service/audio/hwopus.cpp b/src/core/hle/service/audio/hwopus.cpp index fc6067e59..7168c6a10 100644 --- a/src/core/hle/service/audio/hwopus.cpp +++ b/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; } diff --git a/src/core/hle/service/filesystem/filesystem.cpp b/src/core/hle/service/filesystem/filesystem.cpp index 439e62d27..e06712603 100644 --- a/src/core/hle/service/filesystem/filesystem.cpp +++ b/src/core/hle/service/filesystem/filesystem.cpp @@ -361,19 +361,19 @@ FileSys::VirtualDir GetModificationLoadRoot(u64 title_id) { return bis_factory->GetModificationLoadRoot(title_id); } -void CreateFactories(const FileSys::VirtualFilesystem& vfs, bool overwrite) { +void CreateFactories(FileSys::VfsFilesystem& vfs, bool overwrite) { if (overwrite) { bis_factory = nullptr; save_data_factory = nullptr; sdmc_factory = nullptr; } - auto nand_directory = vfs->OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::NANDDir), - FileSys::Mode::ReadWrite); - auto sd_directory = vfs->OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir), - FileSys::Mode::ReadWrite); - auto load_directory = vfs->OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::LoadDir), - FileSys::Mode::ReadWrite); + auto nand_directory = vfs.OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::NANDDir), + FileSys::Mode::ReadWrite); + auto sd_directory = vfs.OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir), + FileSys::Mode::ReadWrite); + auto load_directory = vfs.OpenDirectory(FileUtil::GetUserPath(FileUtil::UserPath::LoadDir), + FileSys::Mode::ReadWrite); if (bis_factory == nullptr) bis_factory = std::make_unique<FileSys::BISFactory>(nand_directory, load_directory); @@ -383,7 +383,7 @@ void CreateFactories(const FileSys::VirtualFilesystem& vfs, bool overwrite) { sdmc_factory = std::make_unique<FileSys::SDMCFactory>(std::move(sd_directory)); } -void InstallInterfaces(SM::ServiceManager& service_manager, const FileSys::VirtualFilesystem& vfs) { +void InstallInterfaces(SM::ServiceManager& service_manager, FileSys::VfsFilesystem& vfs) { romfs_factory = nullptr; CreateFactories(vfs, false); std::make_shared<FSP_LDR>()->InstallAsService(service_manager); diff --git a/src/core/hle/service/filesystem/filesystem.h b/src/core/hle/service/filesystem/filesystem.h index 53b01bb01..2df1faeb0 100644 --- a/src/core/hle/service/filesystem/filesystem.h +++ b/src/core/hle/service/filesystem/filesystem.h @@ -57,9 +57,9 @@ FileSys::VirtualDir GetModificationLoadRoot(u64 title_id); // Creates the SaveData, SDMC, and BIS Factories. Should be called once and before any function // above is called. -void CreateFactories(const FileSys::VirtualFilesystem& vfs, bool overwrite = true); +void CreateFactories(FileSys::VfsFilesystem& vfs, bool overwrite = true); -void InstallInterfaces(SM::ServiceManager& service_manager, const FileSys::VirtualFilesystem& vfs); +void InstallInterfaces(SM::ServiceManager& service_manager, FileSys::VfsFilesystem& vfs); // A class that wraps a VfsDirectory with methods that return ResultVal and ResultCode instead of // pointers and booleans. This makes using a VfsDirectory with switch services much easier and diff --git a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp b/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp index 7555bbe7d..c41ef7058 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp +++ b/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); @@ -167,10 +174,11 @@ u32 nvhost_as_gpu::UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& ou 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); diff --git a/src/core/hle/service/nvdrv/devices/nvmap.cpp b/src/core/hle/service/nvdrv/devices/nvmap.cpp index a2287cc1b..43651d8a6 100644 --- a/src/core/hle/service/nvdrv/devices/nvmap.cpp +++ b/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(¶ms, 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(), ¶ms, 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(¶ms, 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(), ¶ms, 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--; diff --git a/src/core/hle/service/service.cpp b/src/core/hle/service/service.cpp index 62f049660..a225cb4cb 100644 --- a/src/core/hle/service/service.cpp +++ b/src/core/hle/service/service.cpp @@ -197,7 +197,7 @@ ResultCode ServiceFrameworkBase::HandleSyncRequest(Kernel::HLERequestContext& co // Module interface /// Initialize ServiceManager -void Init(std::shared_ptr<SM::ServiceManager>& sm, const FileSys::VirtualFilesystem& rfs) { +void Init(std::shared_ptr<SM::ServiceManager>& sm, FileSys::VfsFilesystem& vfs) { // NVFlinger needs to be accessed by several services like Vi and AppletOE so we instantiate it // here and pass it into the respective InstallInterfaces functions. auto nv_flinger = std::make_shared<NVFlinger::NVFlinger>(); @@ -220,7 +220,7 @@ void Init(std::shared_ptr<SM::ServiceManager>& sm, const FileSys::VirtualFilesys EUPLD::InstallInterfaces(*sm); Fatal::InstallInterfaces(*sm); FGM::InstallInterfaces(*sm); - FileSystem::InstallInterfaces(*sm, rfs); + FileSystem::InstallInterfaces(*sm, vfs); Friend::InstallInterfaces(*sm); GRC::InstallInterfaces(*sm); HID::InstallInterfaces(*sm); diff --git a/src/core/hle/service/service.h b/src/core/hle/service/service.h index 2fc57a82e..98483ecf1 100644 --- a/src/core/hle/service/service.h +++ b/src/core/hle/service/service.h @@ -180,8 +180,7 @@ private: }; /// Initialize ServiceManager -void Init(std::shared_ptr<SM::ServiceManager>& sm, - const std::shared_ptr<FileSys::VfsFilesystem>& vfs); +void Init(std::shared_ptr<SM::ServiceManager>& sm, FileSys::VfsFilesystem& vfs); /// Shutdown ServiceManager void Shutdown(); diff --git a/src/core/loader/deconstructed_rom_directory.cpp b/src/core/loader/deconstructed_rom_directory.cpp index 951fd8257..8518dddcb 100644 --- a/src/core/loader/deconstructed_rom_directory.cpp +++ b/src/core/loader/deconstructed_rom_directory.cpp @@ -139,14 +139,22 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(Kernel::Process& process) for (const auto& module : {"rtld", "main", "subsdk0", "subsdk1", "subsdk2", "subsdk3", "subsdk4", "subsdk5", "subsdk6", "subsdk7", "sdk"}) { const FileSys::VirtualFile module_file = dir->GetFile(module); - if (module_file != nullptr) { - const VAddr load_addr = next_load_addr; - next_load_addr = AppLoader_NSO::LoadModule(module_file, load_addr, - std::strcmp(module, "rtld") == 0, pm); - LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", module, load_addr); - // Register module with GDBStub - GDBStub::RegisterModule(module, load_addr, next_load_addr - 1, false); + if (module_file == nullptr) { + continue; } + + const VAddr load_addr = next_load_addr; + const bool should_pass_arguments = std::strcmp(module, "rtld") == 0; + const auto tentative_next_load_addr = + AppLoader_NSO::LoadModule(*module_file, load_addr, should_pass_arguments, pm); + if (!tentative_next_load_addr) { + return ResultStatus::ErrorLoadingNSO; + } + + next_load_addr = *tentative_next_load_addr; + LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", module, load_addr); + // Register module with GDBStub + GDBStub::RegisterModule(module, load_addr, next_load_addr - 1, false); } process.Run(base_address, metadata.GetMainThreadPriority(), metadata.GetMainThreadStackSize()); diff --git a/src/core/loader/elf.cpp b/src/core/loader/elf.cpp index e67b49fc9..6057c7f26 100644 --- a/src/core/loader/elf.cpp +++ b/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; diff --git a/src/core/loader/loader.cpp b/src/core/loader/loader.cpp index 91659ec17..9cd0b0ccd 100644 --- a/src/core/loader/loader.cpp +++ b/src/core/loader/loader.cpp @@ -93,7 +93,7 @@ std::string GetFileTypeString(FileType type) { return "unknown"; } -constexpr std::array<const char*, 59> RESULT_MESSAGES{ +constexpr std::array<const char*, 60> RESULT_MESSAGES{ "The operation completed successfully.", "The loader requested to load is already loaded.", "The operation is not implemented.", @@ -128,6 +128,7 @@ constexpr std::array<const char*, 59> RESULT_MESSAGES{ "The RomFS could not be found.", "The ELF file has incorrect size as determined by the header.", "There was a general error loading the NRO into emulated memory.", + "There was a general error loading the NSO into emulated memory.", "There is no icon available.", "There is no control data available.", "The NAX file has a bad header.", diff --git a/src/core/loader/loader.h b/src/core/loader/loader.h index 0e0333db5..e562b3a04 100644 --- a/src/core/loader/loader.h +++ b/src/core/loader/loader.h @@ -90,6 +90,7 @@ enum class ResultStatus : u16 { ErrorNoRomFS, ErrorIncorrectELFFileSize, ErrorLoadingNRO, + ErrorLoadingNSO, ErrorNoIcon, ErrorNoControl, ErrorBadNAXHeader, diff --git a/src/core/loader/nro.cpp b/src/core/loader/nro.cpp index 25dd3f04e..243b499f2 100644 --- a/src/core/loader/nro.cpp +++ b/src/core/loader/nro.cpp @@ -14,7 +14,6 @@ #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" @@ -128,10 +127,10 @@ static constexpr u32 PageAlignSize(u32 size) { return (size + Memory::PAGE_MASK) & ~Memory::PAGE_MASK; } -bool AppLoader_NRO::LoadNro(FileSys::VirtualFile file, VAddr load_base) { +bool AppLoader_NRO::LoadNro(const FileSys::VfsFile& file, VAddr load_base) { // Read NSO header NroHeader nro_header{}; - if (sizeof(NroHeader) != file->ReadObject(&nro_header)) { + if (sizeof(NroHeader) != file.ReadObject(&nro_header)) { return {}; } if (nro_header.magic != Common::MakeMagic('N', 'R', 'O', '0')) { @@ -139,22 +138,21 @@ 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)); + 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; + 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(); @@ -176,16 +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; } @@ -198,7 +195,7 @@ ResultStatus AppLoader_NRO::Load(Kernel::Process& process) { // Load NRO const VAddr base_address = process.VMManager().GetCodeRegionBaseAddress(); - if (!LoadNro(file, base_address)) { + if (!LoadNro(*file, base_address)) { return ResultStatus::ErrorLoadingNRO; } diff --git a/src/core/loader/nro.h b/src/core/loader/nro.h index 04b46119a..50ee5a78a 100644 --- a/src/core/loader/nro.h +++ b/src/core/loader/nro.h @@ -41,7 +41,7 @@ public: bool IsRomFSUpdatable() const override; private: - bool LoadNro(FileSys::VirtualFile file, VAddr load_base); + bool LoadNro(const FileSys::VfsFile& file, VAddr load_base); std::vector<u8> icon_data; std::unique_ptr<FileSys::NACP> nacp; diff --git a/src/core/loader/nso.cpp b/src/core/loader/nso.cpp index 28c6dd9b7..68efca5c0 100644 --- a/src/core/loader/nso.cpp +++ b/src/core/loader/nso.cpp @@ -12,7 +12,6 @@ #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" @@ -94,42 +93,38 @@ static constexpr u32 PageAlignSize(u32 size) { return (size + Memory::PAGE_MASK) & ~Memory::PAGE_MASK; } -VAddr AppLoader_NSO::LoadModule(FileSys::VirtualFile file, VAddr load_base, - bool should_pass_arguments, - boost::optional<FileSys::PatchManager> pm) { - if (file == nullptr) - return {}; - - if (file->GetSize() < sizeof(NsoHeader)) +std::optional<VAddr> AppLoader_NSO::LoadModule(const FileSys::VfsFile& file, VAddr load_base, + bool should_pass_arguments, + std::optional<FileSys::PatchManager> pm) { + if (file.GetSize() < sizeof(NsoHeader)) return {}; NsoHeader nso_header{}; - if (sizeof(NsoHeader) != file->ReadObject(&nso_header)) + if (sizeof(NsoHeader) != file.ReadObject(&nso_header)) return {}; if (nso_header.magic != Common::MakeMagic('N', 'S', 'O', '0')) 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 = - file->ReadBytes(nso_header.segments_compressed_size[i], nso_header.segments[i].offset); + file.ReadBytes(nso_header.segments_compressed_size[i], nso_header.segments[i].offset); if (nso_header.IsSegmentCompressed(i)) { data = DecompressSegment(data, nso_header.segments[i]); } 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; + 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(); @@ -154,12 +149,12 @@ 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); // Apply patches if necessary - if (pm != boost::none && pm->HasNSOPatch(nso_header.build_id)) { + if (pm && pm->HasNSOPatch(nso_header.build_id)) { std::vector<u8> pi_header(program_image.size() + 0x100); std::memcpy(pi_header.data(), &nso_header, sizeof(NsoHeader)); std::memcpy(pi_header.data() + 0x100, program_image.data(), program_image.size()); @@ -170,12 +165,11 @@ 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; } @@ -187,7 +181,9 @@ ResultStatus AppLoader_NSO::Load(Kernel::Process& process) { // Load module const VAddr base_address = process.VMManager().GetCodeRegionBaseAddress(); - LoadModule(file, base_address, true); + if (!LoadModule(*file, base_address, true)) { + return ResultStatus::ErrorLoadingNSO; + } LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", file->GetName(), base_address); process.Run(base_address, Kernel::THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE); diff --git a/src/core/loader/nso.h b/src/core/loader/nso.h index 70ab3b718..433306139 100644 --- a/src/core/loader/nso.h +++ b/src/core/loader/nso.h @@ -4,6 +4,7 @@ #pragma once +#include <optional> #include "common/common_types.h" #include "core/file_sys/patch_manager.h" #include "core/loader/linker.h" @@ -36,8 +37,9 @@ public: return IdentifyType(file); } - static VAddr LoadModule(FileSys::VirtualFile file, VAddr load_base, bool should_pass_arguments, - boost::optional<FileSys::PatchManager> pm = boost::none); + static std::optional<VAddr> LoadModule(const FileSys::VfsFile& file, VAddr load_base, + bool should_pass_arguments, + std::optional<FileSys::PatchManager> pm = {}); ResultStatus Load(Kernel::Process& process) override; }; diff --git a/src/video_core/engines/fermi_2d.cpp b/src/video_core/engines/fermi_2d.cpp index 912e785b9..597b279b9 100644 --- a/src/video_core/engines/fermi_2d.cpp +++ b/src/video_core/engines/fermi_2d.cpp @@ -62,14 +62,16 @@ void Fermi2D::HandleSurfaceCopy() { u8* dst_buffer = Memory::GetPointer(dest_cpu); if (!regs.src.linear && regs.dst.linear) { // If the input is tiled and the output is linear, deswizzle the input and copy it over. - Texture::CopySwizzledData(regs.src.width, regs.src.height, src_bytes_per_pixel, - dst_bytes_per_pixel, src_buffer, dst_buffer, true, - regs.src.BlockHeight()); + Texture::CopySwizzledData(regs.src.width, regs.src.height, regs.src.depth, + src_bytes_per_pixel, dst_bytes_per_pixel, src_buffer, + dst_buffer, true, regs.src.BlockHeight(), + regs.src.BlockDepth()); } else { // If the input is linear and the output is tiled, swizzle the input and copy it over. - Texture::CopySwizzledData(regs.src.width, regs.src.height, src_bytes_per_pixel, - dst_bytes_per_pixel, dst_buffer, src_buffer, false, - regs.dst.BlockHeight()); + Texture::CopySwizzledData(regs.src.width, regs.src.height, regs.src.depth, + src_bytes_per_pixel, dst_bytes_per_pixel, dst_buffer, + src_buffer, false, regs.dst.BlockHeight(), + regs.dst.BlockDepth()); } } } diff --git a/src/video_core/engines/fermi_2d.h b/src/video_core/engines/fermi_2d.h index 81d15c62a..2a6e8bbbb 100644 --- a/src/video_core/engines/fermi_2d.h +++ b/src/video_core/engines/fermi_2d.h @@ -36,9 +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; @@ -53,10 +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"); diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h index 20e1884da..c8d1b6478 100644 --- a/src/video_core/engines/maxwell_3d.h +++ b/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; @@ -562,7 +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 { diff --git a/src/video_core/engines/maxwell_dma.cpp b/src/video_core/engines/maxwell_dma.cpp index aa7481b8c..bf2a21bb6 100644 --- a/src/video_core/engines/maxwell_dma.cpp +++ b/src/video_core/engines/maxwell_dma.cpp @@ -68,12 +68,14 @@ void MaxwellDMA::HandleCopy() { if (regs.exec.is_dst_linear && !regs.exec.is_src_linear) { // If the input is tiled and the output is linear, deswizzle the input and copy it over. - Texture::CopySwizzledData(regs.src_params.size_x, regs.src_params.size_y, 1, 1, src_buffer, - dst_buffer, true, regs.src_params.BlockHeight()); + Texture::CopySwizzledData(regs.src_params.size_x, regs.src_params.size_y, + regs.src_params.size_z, 1, 1, src_buffer, dst_buffer, true, + regs.src_params.BlockHeight(), regs.src_params.BlockDepth()); } else { // If the input is linear and the output is tiled, swizzle the input and copy it over. - Texture::CopySwizzledData(regs.dst_params.size_x, regs.dst_params.size_y, 1, 1, dst_buffer, - src_buffer, false, regs.dst_params.BlockHeight()); + Texture::CopySwizzledData(regs.dst_params.size_x, regs.dst_params.size_y, + regs.dst_params.size_z, 1, 1, dst_buffer, src_buffer, false, + regs.dst_params.BlockHeight(), regs.dst_params.BlockDepth()); } } diff --git a/src/video_core/engines/maxwell_dma.h b/src/video_core/engines/maxwell_dma.h index 311ccb616..df19e02e2 100644 --- a/src/video_core/engines/maxwell_dma.h +++ b/src/video_core/engines/maxwell_dma.h @@ -43,6 +43,10 @@ public: u32 BlockHeight() const { return 1 << block_height; } + + u32 BlockDepth() const { + return 1 << block_depth; + } }; static_assert(sizeof(Parameters) == 24, "Parameters has wrong size"); diff --git a/src/video_core/engines/shader_bytecode.h b/src/video_core/engines/shader_bytecode.h index 550ab1148..f356f9a03 100644 --- a/src/video_core/engines/shader_bytecode.h +++ b/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, @@ -255,7 +267,7 @@ enum class ControlCode : u64 { GTU = 12, NEU = 13, GEU = 14, - // + T = 15, OFF = 16, LO = 17, SFF = 18, @@ -452,6 +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; @@ -493,6 +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; @@ -1017,6 +1034,23 @@ union Instruction { } 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; BitField<37, 1, u64> merge_37; @@ -1083,6 +1117,7 @@ public: IPA, OUT_R, // Emit vertex/primitive ISBERD, + VMAD, FFMA_IMM, // Fused Multiply and Add FFMA_CR, FFMA_RC, @@ -1320,6 +1355,7 @@ private: 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"), diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index 84582c777..8d5f277e2 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -286,7 +286,8 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) { &ubo, sizeof(ubo), static_cast<std::size_t>(uniform_buffer_alignment)); // Bind the buffer - glBindBufferRange(GL_UNIFORM_BUFFER, stage, buffer_cache.GetHandle(), offset, sizeof(ubo)); + glBindBufferRange(GL_UNIFORM_BUFFER, static_cast<GLuint>(stage), buffer_cache.GetHandle(), + offset, static_cast<GLsizeiptr>(sizeof(ubo))); Shader shader{shader_cache.GetStageProgram(program)}; diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp index 56ff83eff..801d45144 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp +++ b/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); @@ -148,7 +156,9 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) { SurfaceParams params{}; params.addr = TryGetCpuAddr(config.Address()); params.is_tiled = !config.linear; - params.block_height = params.is_tiled ? config.BlockHeight() : 0, + 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); @@ -221,6 +231,8 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form {GL_RG32UI, GL_RG_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false}, // RG32UI {GL_R32UI, GL_RED_INTEGER, GL_UNSIGNED_INT, ComponentType::UInt, false}, // R32UI {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X8 + {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X5 + {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_5X4 // Depth formats {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, ComponentType::Float, false}, // Z32F @@ -267,7 +279,9 @@ static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType static bool IsPixelFormatASTC(PixelFormat format) { switch (format) { case PixelFormat::ASTC_2D_4X4: + case PixelFormat::ASTC_2D_5X4: case PixelFormat::ASTC_2D_8X8: + case PixelFormat::ASTC_2D_8X5: return true; default: return false; @@ -278,8 +292,12 @@ static std::pair<u32, u32> GetASTCBlockSize(PixelFormat format) { switch (format) { case PixelFormat::ASTC_2D_4X4: return {4, 4}; + case PixelFormat::ASTC_2D_5X4: + return {5, 4}; case PixelFormat::ASTC_2D_8X8: return {8, 8}; + case PixelFormat::ASTC_2D_8X5: + return {8, 5}; default: LOG_CRITICAL(HW_GPU, "Unhandled format: {}", static_cast<u32>(format)); UNREACHABLE(); @@ -313,8 +331,8 @@ static bool IsFormatBCn(PixelFormat format) { } template <bool morton_to_gl, PixelFormat format> -void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, std::size_t gl_buffer_size, - VAddr addr) { +void MortonCopy(u32 stride, u32 block_height, u32 height, u32 block_depth, u32 depth, u8* gl_buffer, + std::size_t gl_buffer_size, VAddr addr) { constexpr u32 bytes_per_pixel = SurfaceParams::GetFormatBpp(format) / CHAR_BIT; constexpr u32 gl_bytes_per_pixel = CachedSurface::GetGLBytesPerPixel(format); @@ -323,7 +341,7 @@ void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, std::si // pixel values. const u32 tile_size{IsFormatBCn(format) ? 4U : 1U}; const std::vector<u8> data = Tegra::Texture::UnswizzleTexture( - addr, tile_size, bytes_per_pixel, stride, height, block_height); + addr, tile_size, bytes_per_pixel, stride, height, depth, block_height, block_depth); const std::size_t size_to_copy{std::min(gl_buffer_size, data.size())}; memcpy(gl_buffer, data.data(), size_to_copy); } else { @@ -335,7 +353,7 @@ void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, std::si } } -static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr), +static constexpr std::array<void (*)(u32, u32, u32, u32, u32, u8*, std::size_t, VAddr), SurfaceParams::MaxPixelFormat> morton_to_gl_fns = { // clang-format off @@ -385,6 +403,8 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr), MortonCopy<true, PixelFormat::RG32UI>, MortonCopy<true, PixelFormat::R32UI>, MortonCopy<true, PixelFormat::ASTC_2D_8X8>, + MortonCopy<true, PixelFormat::ASTC_2D_8X5>, + MortonCopy<true, PixelFormat::ASTC_2D_5X4>, MortonCopy<true, PixelFormat::Z32F>, MortonCopy<true, PixelFormat::Z16>, MortonCopy<true, PixelFormat::Z24S8>, @@ -393,7 +413,7 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr), // clang-format on }; -static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr), +static constexpr std::array<void (*)(u32, u32, u32, u32, u32, u8*, std::size_t, VAddr), SurfaceParams::MaxPixelFormat> gl_to_morton_fns = { // clang-format off @@ -445,6 +465,8 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, std::size_t, VAddr), MortonCopy<false, PixelFormat::RG32UI>, MortonCopy<false, PixelFormat::R32UI>, nullptr, + nullptr, + nullptr, MortonCopy<false, PixelFormat::Z32F>, MortonCopy<false, PixelFormat::Z16>, MortonCopy<false, PixelFormat::Z24S8>, @@ -780,7 +802,9 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma u32 width, u32 height) { switch (pixel_format) { case PixelFormat::ASTC_2D_4X4: - case PixelFormat::ASTC_2D_8X8: { + case PixelFormat::ASTC_2D_8X8: + case PixelFormat::ASTC_2D_8X5: + case PixelFormat::ASTC_2D_5X4: { // Convert ASTC pixel formats to RGBA8, as most desktop GPUs do not support ASTC. u32 block_width{}; u32 block_height{}; @@ -817,31 +841,23 @@ void CachedSurface::LoadGLBuffer() { if (params.is_tiled) { gl_buffer.resize(total_size); + u32 depth = params.depth; + u32 block_depth = params.block_depth; - // TODO(bunnei): This only unswizzles and copies a 2D texture - we do not yet know how to do - // this for 3D textures, etc. - switch (params.target) { - case SurfaceParams::SurfaceTarget::Texture2D: - // Pass impl. to the fallback code below - break; - case SurfaceParams::SurfaceTarget::Texture2DArray: - case SurfaceParams::SurfaceTarget::TextureCubemap: - for (std::size_t index = 0; index < params.depth; ++index) { - const std::size_t offset{index * copy_size}; - morton_to_gl_fns[static_cast<std::size_t>(params.pixel_format)]( - params.width, params.block_height, params.height, gl_buffer.data() + offset, - copy_size, params.addr + offset); - } - break; - default: - LOG_CRITICAL(HW_GPU, "Unimplemented tiled load for target={}", - static_cast<u32>(params.target)); - UNREACHABLE(); + ASSERT_MSG(params.block_width == 1, "Block width is defined as {} on texture type {}", + params.block_width, static_cast<u32>(params.target)); + + if (params.target == SurfaceParams::SurfaceTarget::Texture2D) { + // TODO(Blinkhawk): Eliminate this condition once all texture types are implemented. + depth = 1U; + block_depth = 1U; } + const std::size_t size = copy_size * depth; + morton_to_gl_fns[static_cast<std::size_t>(params.pixel_format)]( - params.width, params.block_height, params.height, gl_buffer.data(), copy_size, - params.addr); + params.width, params.block_height, params.height, block_depth, depth, gl_buffer.data(), + size, params.addr); } else { const u8* const texture_src_data_end{texture_src_data + total_size}; gl_buffer.assign(texture_src_data, texture_src_data_end); @@ -989,7 +1005,9 @@ 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); } diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.h b/src/video_core/renderer_opengl/gl_rasterizer_cache.h index 0b4940b3c..0b8ae3eb4 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer_cache.h +++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.h @@ -74,19 +74,21 @@ struct SurfaceParams { RG32UI = 43, R32UI = 44, ASTC_2D_8X8 = 45, + ASTC_2D_8X5 = 46, + ASTC_2D_5X4 = 47, MaxColorFormat, // Depth formats - Z32F = 46, - Z16 = 47, + Z32F = 48, + Z16 = 49, MaxDepthFormat, // DepthStencil formats - Z24S8 = 48, - S8Z24 = 49, - Z32FS8 = 50, + Z24S8 = 50, + S8Z24 = 51, + Z32FS8 = 52, MaxDepthStencilFormat, @@ -220,6 +222,8 @@ struct SurfaceParams { 1, // RG32UI 1, // R32UI 4, // ASTC_2D_8X8 + 4, // ASTC_2D_8X5 + 4, // ASTC_2D_5X4 1, // Z32F 1, // Z16 1, // Z24S8 @@ -282,6 +286,8 @@ struct SurfaceParams { 64, // RG32UI 32, // R32UI 16, // ASTC_2D_8X8 + 32, // ASTC_2D_8X5 + 32, // ASTC_2D_5X4 32, // Z32F 16, // Z16 32, // Z24S8 @@ -553,8 +559,12 @@ struct SurfaceParams { return PixelFormat::BC6H_SF16; case Tegra::Texture::TextureFormat::ASTC_2D_4X4: return PixelFormat::ASTC_2D_4X4; + case Tegra::Texture::TextureFormat::ASTC_2D_5X4: + return PixelFormat::ASTC_2D_5X4; case Tegra::Texture::TextureFormat::ASTC_2D_8X8: return PixelFormat::ASTC_2D_8X8; + case Tegra::Texture::TextureFormat::ASTC_2D_8X5: + return PixelFormat::ASTC_2D_8X5; case Tegra::Texture::TextureFormat::R16_G16: switch (component_type) { case Tegra::Texture::ComponentType::FLOAT: @@ -716,9 +726,10 @@ 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( @@ -733,7 +744,9 @@ struct SurfaceParams { VAddr addr; bool is_tiled; + u32 block_width; u32 block_height; + u32 block_depth; PixelFormat pixel_format; ComponentType component_type; SurfaceType type; diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp index c82a0dcfa..ca063d90d 100644 --- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp +++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp @@ -1436,7 +1436,6 @@ private: break; } - case OpCode::Type::Shift: { std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, true); std::string op_b; @@ -1478,7 +1477,6 @@ private: } break; } - case OpCode::Type::ArithmeticIntegerImmediate: { std::string op_a = regs.GetRegisterAsInteger(instr.gpr8); std::string op_b = std::to_string(instr.alu.imm20_32.Value()); @@ -2626,14 +2624,14 @@ private: const std::string pred = GetPredicateCondition(instr.csetp.pred39, instr.csetp.neg_pred39 != 0); const std::string combiner = GetPredicateCombiner(instr.csetp.op); - const std::string controlCode = regs.GetControlCode(instr.csetp.cc); + const std::string control_code = regs.GetControlCode(instr.csetp.cc); if (instr.csetp.pred3 != static_cast<u64>(Pred::UnusedIndex)) { SetPredicate(instr.csetp.pred3, - '(' + controlCode + ") " + combiner + " (" + pred + ')'); + '(' + control_code + ") " + combiner + " (" + pred + ')'); } if (instr.csetp.pred0 != static_cast<u64>(Pred::UnusedIndex)) { SetPredicate(instr.csetp.pred0, - "!(" + controlCode + ") " + combiner + " (" + pred + ')'); + "!(" + control_code + ") " + combiner + " (" + pred + ')'); } break; } @@ -2953,6 +2951,88 @@ 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(); diff --git a/src/video_core/textures/decoders.cpp b/src/video_core/textures/decoders.cpp index 3d5476e5d..18ab723f7 100644 --- a/src/video_core/textures/decoders.cpp +++ b/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" @@ -39,72 +40,146 @@ struct alignas(64) SwizzleTable { constexpr auto legacy_swizzle_table = SwizzleTable<8, 64, 1>(); constexpr auto fast_swizzle_table = SwizzleTable<8, 4, 16>(); -static void LegacySwizzleData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel, - u8* swizzled_data, u8* unswizzled_data, bool unswizzle, - u32 block_height) { +/** + * This function manages ALL the GOBs(Group of Bytes) Inside a single block. + * Instead of going gob by gob, we map the coordinates inside a block and manage from + * those. Block_Width is assumed to be 1. + */ +void PreciseProcessBlock(u8* swizzled_data, u8* unswizzled_data, const bool unswizzle, + const u32 x_start, const u32 y_start, const u32 z_start, const u32 x_end, + const u32 y_end, const u32 z_end, const u32 tile_offset, + const u32 xy_block_size, const u32 layer_z, const u32 stride_x, + const u32 bytes_per_pixel, const u32 out_bytes_per_pixel) { std::array<u8*, 2> data_ptrs; - const std::size_t stride = width * bytes_per_pixel; - const std::size_t gobs_in_x = 64; - const std::size_t gobs_in_y = 8; - const std::size_t gobs_size = gobs_in_x * gobs_in_y; - const std::size_t image_width_in_gobs{(stride + gobs_in_x - 1) / gobs_in_x}; - for (std::size_t y = 0; y < height; ++y) { - const std::size_t gob_y_address = - (y / (gobs_in_y * block_height)) * gobs_size * block_height * image_width_in_gobs + - (y % (gobs_in_y * block_height) / gobs_in_y) * gobs_size; - const auto& table = legacy_swizzle_table[y % gobs_in_y]; - for (std::size_t x = 0; x < width; ++x) { - const std::size_t gob_address = - gob_y_address + (x * bytes_per_pixel / gobs_in_x) * gobs_size * block_height; - const std::size_t x2 = x * bytes_per_pixel; - const std::size_t swizzle_offset = gob_address + table[x2 % gobs_in_x]; - const std::size_t pixel_index = (x + y * width) * out_bytes_per_pixel; - - data_ptrs[unswizzle] = swizzled_data + swizzle_offset; - data_ptrs[!unswizzle] = unswizzled_data + pixel_index; - - std::memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel); + u32 z_address = tile_offset; + const u32 gob_size_x = 64; + const u32 gob_size_y = 8; + const u32 gob_size_z = 1; + const u32 gob_size = gob_size_x * gob_size_y * gob_size_z; + for (u32 z = z_start; z < z_end; z++) { + u32 y_address = z_address; + u32 pixel_base = layer_z * z + y_start * stride_x; + for (u32 y = y_start; y < y_end; y++) { + const auto& table = legacy_swizzle_table[y % gob_size_y]; + for (u32 x = x_start; x < x_end; x++) { + const u32 swizzle_offset{y_address + table[x * bytes_per_pixel % gob_size_x]}; + const u32 pixel_index{x * out_bytes_per_pixel + pixel_base}; + data_ptrs[unswizzle] = swizzled_data + swizzle_offset; + data_ptrs[!unswizzle] = unswizzled_data + pixel_index; + std::memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel); + } + pixel_base += stride_x; + if ((y + 1) % gob_size_y == 0) + y_address += gob_size; } + z_address += xy_block_size; } } -static void FastSwizzleData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel, - u8* swizzled_data, u8* unswizzled_data, bool unswizzle, - u32 block_height) { +/** + * This function manages ALL the GOBs(Group of Bytes) Inside a single block. + * Instead of going gob by gob, we map the coordinates inside a block and manage from + * those. Block_Width is assumed to be 1. + */ +void FastProcessBlock(u8* swizzled_data, u8* unswizzled_data, const bool unswizzle, + const u32 x_start, const u32 y_start, const u32 z_start, const u32 x_end, + const u32 y_end, const u32 z_end, const u32 tile_offset, + const u32 xy_block_size, const u32 layer_z, const u32 stride_x, + const u32 bytes_per_pixel, const u32 out_bytes_per_pixel) { std::array<u8*, 2> data_ptrs; - const std::size_t stride{width * bytes_per_pixel}; - const std::size_t gobs_in_x = 64; - const std::size_t gobs_in_y = 8; - const std::size_t gobs_size = gobs_in_x * gobs_in_y; - const std::size_t image_width_in_gobs{(stride + gobs_in_x - 1) / gobs_in_x}; - const std::size_t copy_size{16}; - for (std::size_t y = 0; y < height; ++y) { - const std::size_t initial_gob = - (y / (gobs_in_y * block_height)) * gobs_size * block_height * image_width_in_gobs + - (y % (gobs_in_y * block_height) / gobs_in_y) * gobs_size; - const std::size_t pixel_base{y * width * out_bytes_per_pixel}; - const auto& table = fast_swizzle_table[y % gobs_in_y]; - for (std::size_t xb = 0; xb < stride; xb += copy_size) { - const std::size_t gob_address{initial_gob + - (xb / gobs_in_x) * gobs_size * block_height}; - const std::size_t swizzle_offset{gob_address + table[(xb / 16) % 4]}; - const std::size_t out_x = xb * out_bytes_per_pixel / bytes_per_pixel; - const std::size_t pixel_index{out_x + pixel_base}; - data_ptrs[unswizzle] = swizzled_data + swizzle_offset; - data_ptrs[!unswizzle] = unswizzled_data + pixel_index; - std::memcpy(data_ptrs[0], data_ptrs[1], copy_size); + u32 z_address = tile_offset; + const u32 x_startb = x_start * bytes_per_pixel; + const u32 x_endb = x_end * bytes_per_pixel; + const u32 copy_size = 16; + const u32 gob_size_x = 64; + const u32 gob_size_y = 8; + const u32 gob_size_z = 1; + const u32 gob_size = gob_size_x * gob_size_y * gob_size_z; + for (u32 z = z_start; z < z_end; z++) { + u32 y_address = z_address; + u32 pixel_base = layer_z * z + y_start * stride_x; + for (u32 y = y_start; y < y_end; y++) { + const auto& table = fast_swizzle_table[y % gob_size_y]; + for (u32 xb = x_startb; xb < x_endb; xb += copy_size) { + const u32 swizzle_offset{y_address + table[(xb / copy_size) % 4]}; + const u32 out_x = xb * out_bytes_per_pixel / bytes_per_pixel; + const u32 pixel_index{out_x + pixel_base}; + data_ptrs[unswizzle] = swizzled_data + swizzle_offset; + data_ptrs[!unswizzle] = unswizzled_data + pixel_index; + std::memcpy(data_ptrs[0], data_ptrs[1], copy_size); + } + pixel_base += stride_x; + if ((y + 1) % gob_size_y == 0) + y_address += gob_size; } + z_address += xy_block_size; } } -void CopySwizzledData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel, - u8* swizzled_data, u8* unswizzled_data, bool unswizzle, u32 block_height) { +/** + * This function unswizzles or swizzles a texture by mapping Linear to BlockLinear Textue. + * The body of this function takes care of splitting the swizzled texture into blocks, + * and managing the extents of it. Once all the parameters of a single block are obtained, + * the function calls 'ProcessBlock' to process that particular Block. + * + * Documentation for the memory layout and decoding can be found at: + * https://envytools.readthedocs.io/en/latest/hw/memory/g80-surface.html#blocklinear-surfaces + */ +template <bool fast> +void SwizzledData(u8* swizzled_data, u8* unswizzled_data, const bool unswizzle, const u32 width, + const u32 height, const u32 depth, const u32 bytes_per_pixel, + const u32 out_bytes_per_pixel, const u32 block_height, const u32 block_depth) { + auto div_ceil = [](const u32 x, const u32 y) { return ((x + y - 1) / y); }; + const u32 stride_x = width * out_bytes_per_pixel; + const u32 layer_z = height * stride_x; + const u32 gob_x_bytes = 64; + const u32 gob_elements_x = gob_x_bytes / bytes_per_pixel; + const u32 gob_elements_y = 8; + const u32 gob_elements_z = 1; + const u32 block_x_elements = gob_elements_x; + const u32 block_y_elements = gob_elements_y * block_height; + const u32 block_z_elements = gob_elements_z * block_depth; + const u32 blocks_on_x = div_ceil(width, block_x_elements); + const u32 blocks_on_y = div_ceil(height, block_y_elements); + const u32 blocks_on_z = div_ceil(depth, block_z_elements); + const u32 blocks = blocks_on_x * blocks_on_y * blocks_on_z; + const u32 gob_size = gob_x_bytes * gob_elements_y * gob_elements_z; + const u32 xy_block_size = gob_size * block_height; + const u32 block_size = xy_block_size * block_depth; + u32 tile_offset = 0; + for (u32 zb = 0; zb < blocks_on_z; zb++) { + const u32 z_start = zb * block_z_elements; + const u32 z_end = std::min(depth, z_start + block_z_elements); + for (u32 yb = 0; yb < blocks_on_y; yb++) { + const u32 y_start = yb * block_y_elements; + const u32 y_end = std::min(height, y_start + block_y_elements); + for (u32 xb = 0; xb < blocks_on_x; xb++) { + const u32 x_start = xb * block_x_elements; + const u32 x_end = std::min(width, x_start + block_x_elements); + if (fast) { + FastProcessBlock(swizzled_data, unswizzled_data, unswizzle, x_start, y_start, + z_start, x_end, y_end, z_end, tile_offset, xy_block_size, + layer_z, stride_x, bytes_per_pixel, out_bytes_per_pixel); + } else { + PreciseProcessBlock(swizzled_data, unswizzled_data, unswizzle, x_start, y_start, + z_start, x_end, y_end, z_end, tile_offset, xy_block_size, + layer_z, stride_x, bytes_per_pixel, out_bytes_per_pixel); + } + tile_offset += block_size; + } + } + } +} + +void CopySwizzledData(u32 width, u32 height, u32 depth, u32 bytes_per_pixel, + u32 out_bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data, + bool unswizzle, u32 block_height, u32 block_depth) { if (bytes_per_pixel % 3 != 0 && (width * bytes_per_pixel) % 16 == 0) { - FastSwizzleData(width, height, bytes_per_pixel, out_bytes_per_pixel, swizzled_data, - unswizzled_data, unswizzle, block_height); + SwizzledData<true>(swizzled_data, unswizzled_data, unswizzle, width, height, depth, + bytes_per_pixel, out_bytes_per_pixel, block_height, block_depth); } else { - LegacySwizzleData(width, height, bytes_per_pixel, out_bytes_per_pixel, swizzled_data, - unswizzled_data, unswizzle, block_height); + SwizzledData<false>(swizzled_data, unswizzled_data, unswizzle, width, height, depth, + bytes_per_pixel, out_bytes_per_pixel, block_height, block_depth); } } @@ -125,7 +200,9 @@ u32 BytesPerPixel(TextureFormat format) { case TextureFormat::R32_G32_B32: return 12; case TextureFormat::ASTC_2D_4X4: + case TextureFormat::ASTC_2D_5X4: case TextureFormat::ASTC_2D_8X8: + case TextureFormat::ASTC_2D_8X5: case TextureFormat::A8R8G8B8: case TextureFormat::A2B10G10R10: case TextureFormat::BF10GF11RF11: @@ -152,10 +229,11 @@ u32 BytesPerPixel(TextureFormat format) { } std::vector<u8> UnswizzleTexture(VAddr address, u32 tile_size, u32 bytes_per_pixel, u32 width, - u32 height, u32 block_height) { - std::vector<u8> unswizzled_data(width * height * bytes_per_pixel); - CopySwizzledData(width / tile_size, height / tile_size, bytes_per_pixel, bytes_per_pixel, - Memory::GetPointer(address), unswizzled_data.data(), true, block_height); + u32 height, u32 depth, u32 block_height, u32 block_depth) { + std::vector<u8> unswizzled_data(width * height * depth * bytes_per_pixel); + CopySwizzledData(width / tile_size, height / tile_size, depth, bytes_per_pixel, bytes_per_pixel, + Memory::GetPointer(address), unswizzled_data.data(), true, block_height, + block_depth); return unswizzled_data; } @@ -199,4 +277,19 @@ 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 diff --git a/src/video_core/textures/decoders.h b/src/video_core/textures/decoders.h index 1f7b731be..aaf316947 100644 --- a/src/video_core/textures/decoders.h +++ b/src/video_core/textures/decoders.h @@ -14,17 +14,14 @@ namespace Tegra::Texture { * Unswizzles a swizzled texture without changing its format. */ std::vector<u8> UnswizzleTexture(VAddr address, u32 tile_size, u32 bytes_per_pixel, u32 width, - u32 height, u32 block_height = TICEntry::DefaultBlockHeight); - -/** - * Unswizzles a swizzled depth texture without changing its format. - */ -std::vector<u8> UnswizzleDepthTexture(VAddr address, DepthFormat format, u32 width, u32 height, - u32 block_height = TICEntry::DefaultBlockHeight); + u32 height, u32 depth, + u32 block_height = TICEntry::DefaultBlockHeight, + u32 block_depth = TICEntry::DefaultBlockHeight); /// Copies texture data from a buffer and performs swizzling/unswizzling as necessary. -void CopySwizzledData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel, - u8* swizzled_data, u8* unswizzled_data, bool unswizzle, u32 block_height); +void CopySwizzledData(u32 width, u32 height, u32 depth, u32 bytes_per_pixel, + u32 out_bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data, + bool unswizzle, u32 block_height, u32 block_depth); /** * Decodes an unswizzled texture into a A8R8G8B8 texture. @@ -32,4 +29,10 @@ 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 diff --git a/src/video_core/textures/texture.h b/src/video_core/textures/texture.h index 8f31d825a..5947bd2b9 100644 --- a/src/video_core/textures/texture.h +++ b/src/video_core/textures/texture.h @@ -141,6 +141,7 @@ static_assert(sizeof(TextureHandle) == 4, "TextureHandle has wrong size"); struct TICEntry { static constexpr u32 DefaultBlockHeight = 16; + static constexpr u32 DefaultBlockDepth = 1; union { u32 raw; @@ -161,7 +162,9 @@ 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 +205,24 @@ 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; diff --git a/src/yuzu/debugger/graphics/graphics_surface.cpp b/src/yuzu/debugger/graphics/graphics_surface.cpp index cbcd5dd5f..44d423da2 100644 --- a/src/yuzu/debugger/graphics/graphics_surface.cpp +++ b/src/yuzu/debugger/graphics/graphics_surface.cpp @@ -386,8 +386,9 @@ void GraphicsSurfaceWidget::OnUpdate() { // TODO(bunnei): Will not work with BCn formats that swizzle 4x4 tiles. // Needs to be fixed if we plan to use this feature more, otherwise we may remove it. - auto unswizzled_data = Tegra::Texture::UnswizzleTexture( - *address, 1, Tegra::Texture::BytesPerPixel(surface_format), surface_width, surface_height); + auto unswizzled_data = + Tegra::Texture::UnswizzleTexture(*address, 1, Tegra::Texture::BytesPerPixel(surface_format), + surface_width, surface_height, 1U); auto texture_data = Tegra::Texture::DecodeTexture(unswizzled_data, surface_format, surface_width, surface_height); diff --git a/src/yuzu/main.cpp b/src/yuzu/main.cpp index e11833c5a..bef9df00d 100644 --- a/src/yuzu/main.cpp +++ b/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> @@ -171,8 +172,11 @@ 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); + Service::FileSystem::CreateFactories(*vfs); game_list->LoadCompatibilityList(); game_list->PopulateAsync(UISettings::values.gamedir, UISettings::values.gamedir_deepscan); @@ -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); } @@ -902,22 +908,20 @@ void GMainWindow::OnGameListNavigateToGamedbEntry(u64 program_id, } void GMainWindow::OnMenuLoadFile() { - QString extensions; - for (const auto& piece : game_list->supported_file_extensions) - extensions += "*." + piece + " "; - - extensions += "main "; + const QString extensions = + QString("*.").append(GameList::supported_file_extensions.join(" *.")).append(" main"); + const QString file_filter = tr("Switch Executable (%1);;All Files (*.*)", + "%1 is an identifier for the Switch executable file extensions.") + .arg(extensions); + const QString filename = QFileDialog::getOpenFileName( + this, tr("Load File"), UISettings::values.roms_path, file_filter); - QString file_filter = tr("Switch Executable") + " (" + extensions + ")"; - file_filter += ";;" + tr("All Files (*.*)"); - - QString filename = QFileDialog::getOpenFileName(this, tr("Load File"), - UISettings::values.roms_path, file_filter); - if (!filename.isEmpty()) { - UISettings::values.roms_path = QFileInfo(filename).path(); - - BootGame(filename); + if (filename.isEmpty()) { + return; } + + UISettings::values.roms_path = QFileInfo(filename).path(); + BootGame(filename); } void GMainWindow::OnMenuLoadFolder() { @@ -1133,7 +1137,7 @@ void GMainWindow::OnMenuSelectEmulatedDirectory(EmulatedDirectoryTarget target) FileUtil::GetUserPath(target == EmulatedDirectoryTarget::SDMC ? FileUtil::UserPath::SDMCDir : FileUtil::UserPath::NANDDir, dir_path.toStdString()); - Service::FileSystem::CreateFactories(vfs); + Service::FileSystem::CreateFactories(*vfs); game_list->PopulateAsync(UISettings::values.gamedir, UISettings::values.gamedir_deepscan); } } @@ -1375,6 +1379,86 @@ 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.<br><br>") + + errors + + tr("<br><br>You can get all of these and dump all of your games easily by " + "following <a href='https://yuzu-emu.org/help/quickstart/quickstart/'>the " + "quickstart guide</a>. Alternatively, you can use another method of dumping " + "to obtain all of your keys.")); + } + + 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; @@ -1483,7 +1567,7 @@ void GMainWindow::UpdateUITheme() { emit UpdateThemedIcons(); } -void GMainWindow::SetDiscordEnabled(bool state) { +void GMainWindow::SetDiscordEnabled([[maybe_unused]] bool state) { #ifdef USE_DISCORD_PRESENCE if (state) { discord_rpc = std::make_unique<DiscordRPC::DiscordImpl>(); diff --git a/src/yuzu/main.h b/src/yuzu/main.h index fe0e9a50a..3663d6aed 100644 --- a/src/yuzu/main.h +++ b/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(); diff --git a/src/yuzu/main.ui b/src/yuzu/main.ui index cb1664b21..9851f507d 100644 --- a/src/yuzu/main.ui +++ b/src/yuzu/main.ui @@ -103,6 +103,7 @@ </property> <addaction name="action_Report_Compatibility"/> <addaction name="separator"/> + <addaction name="action_Rederive"/> <addaction name="action_About"/> </widget> <addaction name="menu_File"/> @@ -159,6 +160,11 @@ <string>&Stop</string> </property> </action> + <action name="action_Rederive"> + <property name="text"> + <string>Reinitialize keys...</string> + </property> + </action> <action name="action_About"> <property name="text"> <string>About yuzu</string> diff --git a/src/yuzu_cmd/yuzu.cpp b/src/yuzu_cmd/yuzu.cpp index 27aba95f6..c8b93b85b 100644 --- a/src/yuzu_cmd/yuzu.cpp +++ b/src/yuzu_cmd/yuzu.cpp @@ -175,7 +175,7 @@ int main(int argc, char** argv) { Core::System& system{Core::System::GetInstance()}; system.SetFilesystem(std::make_shared<FileSys::RealVfsFilesystem>()); - Service::FileSystem::CreateFactories(system.GetFilesystem()); + Service::FileSystem::CreateFactories(*system.GetFilesystem()); SCOPE_EXIT({ system.Shutdown(); }); |