diff options
52 files changed, 490 insertions, 326 deletions
diff --git a/externals/dynarmic b/externals/dynarmic -Subproject 82417da7803e2cf18efc28a1cd3f3d0a4b6045a +Subproject 0e1112b7df77ae55a62a51622940d5c8f9e8c84 diff --git a/src/common/CMakeLists.txt b/src/common/CMakeLists.txt index 78c3bfb3b..5d54516eb 100644 --- a/src/common/CMakeLists.txt +++ b/src/common/CMakeLists.txt @@ -172,7 +172,6 @@ add_library(common STATIC virtual_buffer.h wall_clock.cpp wall_clock.h - web_result.h zstd_compression.cpp zstd_compression.h ) diff --git a/src/common/color.h b/src/common/color.h index 381d6332e..bbcac858e 100644 --- a/src/common/color.h +++ b/src/common/color.h @@ -10,7 +10,7 @@ #include "common/swap.h" #include "common/vector_math.h" -namespace Color { +namespace Common::Color { /// Convert a 1-bit color component to 8 bit [[nodiscard]] constexpr u8 Convert1To8(u8 value) { @@ -268,4 +268,4 @@ inline void EncodeX24S8(u8 stencil, u8* bytes) { bytes[3] = stencil; } -} // namespace Color +} // namespace Common::Color diff --git a/src/core/arm/dynarmic/arm_dynarmic_32.cpp b/src/core/arm/dynarmic/arm_dynarmic_32.cpp index 443ca72eb..b5f28a86e 100644 --- a/src/core/arm/dynarmic/arm_dynarmic_32.cpp +++ b/src/core/arm/dynarmic/arm_dynarmic_32.cpp @@ -143,7 +143,7 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable& config.wall_clock_cntpct = uses_wall_clock; // Safe optimizations - if (Settings::values.cpu_accuracy != Settings::CPUAccuracy::Accurate) { + if (Settings::values.cpu_accuracy == Settings::CPUAccuracy::DebugMode) { if (!Settings::values.cpuopt_page_tables) { config.page_table = nullptr; } @@ -170,6 +170,17 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable& } } + // Unsafe optimizations + if (Settings::values.cpu_accuracy == Settings::CPUAccuracy::Unsafe) { + config.unsafe_optimizations = true; + if (Settings::values.cpuopt_unsafe_unfuse_fma) { + config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_UnfuseFMA; + } + if (Settings::values.cpuopt_unsafe_reduce_fp_error) { + config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_ReducedErrorFP; + } + } + return std::make_unique<Dynarmic::A32::Jit>(config); } diff --git a/src/core/arm/dynarmic/arm_dynarmic_64.cpp b/src/core/arm/dynarmic/arm_dynarmic_64.cpp index a63a04a25..ce9968724 100644 --- a/src/core/arm/dynarmic/arm_dynarmic_64.cpp +++ b/src/core/arm/dynarmic/arm_dynarmic_64.cpp @@ -195,7 +195,7 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable& config.wall_clock_cntpct = uses_wall_clock; // Safe optimizations - if (Settings::values.cpu_accuracy != Settings::CPUAccuracy::Accurate) { + if (Settings::values.cpu_accuracy == Settings::CPUAccuracy::DebugMode) { if (!Settings::values.cpuopt_page_tables) { config.page_table = nullptr; } @@ -222,6 +222,17 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable& } } + // Unsafe optimizations + if (Settings::values.cpu_accuracy == Settings::CPUAccuracy::Unsafe) { + config.unsafe_optimizations = true; + if (Settings::values.cpuopt_unsafe_unfuse_fma) { + config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_UnfuseFMA; + } + if (Settings::values.cpuopt_unsafe_reduce_fp_error) { + config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_ReducedErrorFP; + } + } + return std::make_shared<Dynarmic::A64::Jit>(config); } diff --git a/src/core/core_timing.cpp b/src/core/core_timing.cpp index 71af26ec5..e6c8461a5 100644 --- a/src/core/core_timing.cpp +++ b/src/core/core_timing.cpp @@ -7,14 +7,14 @@ #include <string> #include <tuple> -#include "common/assert.h" #include "common/microprofile.h" #include "core/core_timing.h" #include "core/core_timing_util.h" +#include "core/hardware_properties.h" namespace Core::Timing { -constexpr u64 MAX_SLICE_LENGTH = 4000; +constexpr s64 MAX_SLICE_LENGTH = 4000; std::shared_ptr<EventType> CreateEvent(std::string name, TimedCallback&& callback) { return std::make_shared<EventType>(std::move(callback), std::move(name)); @@ -37,10 +37,8 @@ struct CoreTiming::Event { } }; -CoreTiming::CoreTiming() { - clock = - Common::CreateBestMatchingClock(Core::Hardware::BASE_CLOCK_RATE, Core::Hardware::CNTFREQ); -} +CoreTiming::CoreTiming() + : clock{Common::CreateBestMatchingClock(Hardware::BASE_CLOCK_RATE, Hardware::CNTFREQ)} {} CoreTiming::~CoreTiming() = default; @@ -136,7 +134,7 @@ void CoreTiming::UnscheduleEvent(const std::shared_ptr<EventType>& event_type, void CoreTiming::AddTicks(u64 ticks) { this->ticks += ticks; - downcount -= ticks; + downcount -= static_cast<s64>(ticks); } void CoreTiming::Idle() { diff --git a/src/core/core_timing_util.cpp b/src/core/core_timing_util.cpp index aefc63663..8ce8e602e 100644 --- a/src/core/core_timing_util.cpp +++ b/src/core/core_timing_util.cpp @@ -8,6 +8,7 @@ #include <limits> #include "common/logging/log.h" #include "common/uint128.h" +#include "core/hardware_properties.h" namespace Core::Timing { diff --git a/src/core/core_timing_util.h b/src/core/core_timing_util.h index 2ed979e14..e4a046bf9 100644 --- a/src/core/core_timing_util.h +++ b/src/core/core_timing_util.h @@ -6,7 +6,6 @@ #include <chrono> #include "common/common_types.h" -#include "core/hardware_properties.h" namespace Core::Timing { diff --git a/src/core/cpu_manager.cpp b/src/core/cpu_manager.cpp index 358943429..ef0bae556 100644 --- a/src/core/cpu_manager.cpp +++ b/src/core/cpu_manager.cpp @@ -41,9 +41,9 @@ void CpuManager::Shutdown() { running_mode = false; Pause(false); if (is_multicore) { - for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { - core_data[core].host_thread->join(); - core_data[core].host_thread.reset(); + for (auto& data : core_data) { + data.host_thread->join(); + data.host_thread.reset(); } } else { core_data[0].host_thread->join(); @@ -166,25 +166,23 @@ void CpuManager::MultiCorePause(bool paused) { bool all_not_barrier = false; while (!all_not_barrier) { all_not_barrier = true; - for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { - all_not_barrier &= - !core_data[core].is_running.load() && core_data[core].initialized.load(); + for (const auto& data : core_data) { + all_not_barrier &= !data.is_running.load() && data.initialized.load(); } } - for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { - core_data[core].enter_barrier->Set(); + for (auto& data : core_data) { + data.enter_barrier->Set(); } if (paused_state.load()) { bool all_barrier = false; while (!all_barrier) { all_barrier = true; - for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { - all_barrier &= - core_data[core].is_paused.load() && core_data[core].initialized.load(); + for (const auto& data : core_data) { + all_barrier &= data.is_paused.load() && data.initialized.load(); } } - for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { - core_data[core].exit_barrier->Set(); + for (auto& data : core_data) { + data.exit_barrier->Set(); } } } else { @@ -192,9 +190,8 @@ void CpuManager::MultiCorePause(bool paused) { bool all_barrier = false; while (!all_barrier) { all_barrier = true; - for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { - all_barrier &= - core_data[core].is_paused.load() && core_data[core].initialized.load(); + for (const auto& data : core_data) { + all_barrier &= data.is_paused.load() && data.initialized.load(); } } /// Don't release the barrier diff --git a/src/core/crypto/key_manager.cpp b/src/core/crypto/key_manager.cpp index 8783d1ac2..dc591c730 100644 --- a/src/core/crypto/key_manager.cpp +++ b/src/core/crypto/key_manager.cpp @@ -36,6 +36,7 @@ #include "core/settings.h" namespace Core::Crypto { +namespace { constexpr u64 CURRENT_CRYPTO_REVISION = 0x5; constexpr u64 FULL_TICKET_SIZE = 0x400; @@ -49,7 +50,72 @@ constexpr std::array eticket_source_hashes{ }; // clang-format on -const std::map<std::pair<S128KeyType, u64>, std::string> KEYS_VARIABLE_LENGTH{ +constexpr std::array<std::pair<std::string_view, KeyIndex<S128KeyType>>, 30> s128_file_id{{ + {"eticket_rsa_kek", {S128KeyType::ETicketRSAKek, 0, 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}}, + {"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), 0}}, + {"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}}, +}}; + +auto Find128ByName(std::string_view name) { + return std::find_if(s128_file_id.begin(), s128_file_id.end(), + [&name](const auto& pair) { return pair.first == name; }); +} + +constexpr std::array<std::pair<std::string_view, KeyIndex<S256KeyType>>, 6> 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}}, +}}; + +auto Find256ByName(std::string_view name) { + return std::find_if(s256_file_id.begin(), s256_file_id.end(), + [&name](const auto& pair) { return pair.first == name; }); +} + +using KeyArray = std::array<std::pair<std::pair<S128KeyType, u64>, std::string_view>, 7>; +constexpr KeyArray KEYS_VARIABLE_LENGTH{{ {{S128KeyType::Master, 0}, "master_key_"}, {{S128KeyType::Package1, 0}, "package1_key_"}, {{S128KeyType::Package2, 0}, "package2_key_"}, @@ -57,14 +123,13 @@ const std::map<std::pair<S128KeyType, u64>, std::string> KEYS_VARIABLE_LENGTH{ {{S128KeyType::Source, static_cast<u64>(SourceKeyType::Keyblob)}, "keyblob_key_source_"}, {{S128KeyType::Keyblob, 0}, "keyblob_key_"}, {{S128KeyType::KeyblobMAC, 0}, "keyblob_mac_key_"}, -}; +}}; -namespace { template <std::size_t Size> bool IsAllZeroArray(const std::array<u8, Size>& array) { return std::all_of(array.begin(), array.end(), [](const auto& elem) { return elem == 0; }); } -} // namespace +} // Anonymous namespace u64 GetSignatureTypeDataSize(SignatureType type) { switch (type) { @@ -564,13 +629,13 @@ void KeyManager::LoadFromFile(const std::string& filename, bool is_title_keys) { s128_keys[{S128KeyType::Titlekey, rights_id[1], rights_id[0]}] = key; } else { out[0] = Common::ToLower(out[0]); - if (s128_file_id.find(out[0]) != s128_file_id.end()) { - const auto index = s128_file_id.at(out[0]); - Key128 key = Common::HexStringToArray<16>(out[1]); + if (const auto iter128 = Find128ByName(out[0]); iter128 != s128_file_id.end()) { + const auto& index = iter128->second; + const Key128 key = Common::HexStringToArray<16>(out[1]); s128_keys[{index.type, index.field1, index.field2}] = key; - } else if (s256_file_id.find(out[0]) != s256_file_id.end()) { - const auto index = s256_file_id.at(out[0]); - Key256 key = Common::HexStringToArray<32>(out[1]); + } else if (const auto iter256 = Find256ByName(out[0]); iter256 != s256_file_id.end()) { + const auto& index = iter256->second; + const 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) { @@ -742,8 +807,7 @@ void KeyManager::SetKey(S128KeyType id, Key128 key, u64 field1, u64 field2) { } 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) { + s128_file_id.begin(), s128_file_id.end(), [&id, &field1, &field2](const auto& elem) { return std::tie(elem.second.type, elem.second.field1, elem.second.field2) == std::tie(id, field1, field2); }); @@ -753,9 +817,11 @@ void KeyManager::SetKey(S128KeyType id, Key128 key, u64 field1, u64 field2) { // 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}"}; + 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); @@ -781,8 +847,7 @@ void KeyManager::SetKey(S256KeyType id, Key256 key, u64 field1, u64 field2) { return; } const auto iter = std::find_if( - s256_file_id.begin(), s256_file_id.end(), - [&id, &field1, &field2](const std::pair<std::string, KeyIndex<S256KeyType>> elem) { + s256_file_id.begin(), s256_file_id.end(), [&id, &field1, &field2](const auto& elem) { return std::tie(elem.second.type, elem.second.field1, elem.second.field2) == std::tie(id, field1, field2); }); @@ -1245,58 +1310,4 @@ bool KeyManager::AddTicketPersonalized(Ticket raw) { SetKey(S128KeyType::Titlekey, key, rights_id[1], rights_id[0]); return true; } - -const boost::container::flat_map<std::string, KeyIndex<S128KeyType>> KeyManager::s128_file_id = { - {"eticket_rsa_kek", {S128KeyType::ETicketRSAKek, 0, 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}}, - {"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), 0}}, - {"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 bdca3770a..321b75323 100644 --- a/src/core/crypto/key_manager.h +++ b/src/core/crypto/key_manager.h @@ -10,7 +10,6 @@ #include <string> #include <variant> -#include <boost/container/flat_map.hpp> #include <fmt/format.h> #include "common/common_funcs.h" #include "common/common_types.h" @@ -293,9 +292,6 @@ private: 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); diff --git a/src/core/file_sys/vfs_real.cpp b/src/core/file_sys/vfs_real.cpp index 1dbf632c1..488687ba9 100644 --- a/src/core/file_sys/vfs_real.cpp +++ b/src/core/file_sys/vfs_real.cpp @@ -72,8 +72,10 @@ VfsEntryType RealVfsFilesystem::GetEntryType(std::string_view path_) const { VirtualFile RealVfsFilesystem::OpenFile(std::string_view path_, Mode perms) { const auto path = FS::SanitizePath(path_, FS::DirectorySeparator::PlatformDefault); - if (cache.find(path) != cache.end()) { - auto weak = cache[path]; + + if (const auto weak_iter = cache.find(path); weak_iter != cache.cend()) { + const auto& weak = weak_iter->second; + if (!weak.expired()) { return std::shared_ptr<RealVfsFile>(new RealVfsFile(*this, weak.lock(), path, perms)); } @@ -84,7 +86,7 @@ VirtualFile RealVfsFilesystem::OpenFile(std::string_view path_, Mode perms) { } auto backing = std::make_shared<FS::IOFile>(path, ModeFlagsToString(perms).c_str()); - cache[path] = backing; + cache.insert_or_assign(path, backing); // Cannot use make_shared as RealVfsFile constructor is private return std::shared_ptr<RealVfsFile>(new RealVfsFile(*this, backing, path, perms)); @@ -116,11 +118,12 @@ VirtualFile RealVfsFilesystem::CopyFile(std::string_view old_path_, std::string_ VirtualFile RealVfsFilesystem::MoveFile(std::string_view old_path_, std::string_view new_path_) { const auto old_path = FS::SanitizePath(old_path_, FS::DirectorySeparator::PlatformDefault); const auto new_path = FS::SanitizePath(new_path_, FS::DirectorySeparator::PlatformDefault); + const auto cached_file_iter = cache.find(old_path); - if (cache.find(old_path) != cache.end()) { - auto file = cache[old_path].lock(); + if (cached_file_iter != cache.cend()) { + auto file = cached_file_iter->second.lock(); - if (!cache[old_path].expired()) { + if (!cached_file_iter->second.expired()) { file->Close(); } @@ -131,7 +134,7 @@ VirtualFile RealVfsFilesystem::MoveFile(std::string_view old_path_, std::string_ cache.erase(old_path); file->Open(new_path, "r+b"); - cache[new_path] = file; + cache.insert_or_assign(new_path, std::move(file)); } else { UNREACHABLE(); return nullptr; @@ -142,12 +145,15 @@ VirtualFile RealVfsFilesystem::MoveFile(std::string_view old_path_, std::string_ bool RealVfsFilesystem::DeleteFile(std::string_view path_) { const auto path = FS::SanitizePath(path_, FS::DirectorySeparator::PlatformDefault); - if (cache.find(path) != cache.end()) { - if (!cache[path].expired()) { - cache[path].lock()->Close(); + const auto cached_iter = cache.find(path); + + if (cached_iter != cache.cend()) { + if (!cached_iter->second.expired()) { + cached_iter->second.lock()->Close(); } cache.erase(path); } + return FS::Delete(path); } @@ -192,21 +198,25 @@ VirtualDir RealVfsFilesystem::MoveDirectory(std::string_view old_path_, } for (auto& kv : cache) { - // Path in cache starts with old_path - if (kv.first.rfind(old_path, 0) == 0) { - const auto file_old_path = - FS::SanitizePath(kv.first, FS::DirectorySeparator::PlatformDefault); - const auto file_new_path = - FS::SanitizePath(new_path + DIR_SEP + kv.first.substr(old_path.size()), - FS::DirectorySeparator::PlatformDefault); - auto cached = cache[file_old_path]; - if (!cached.expired()) { - auto file = cached.lock(); - file->Open(file_new_path, "r+b"); - cache.erase(file_old_path); - cache[file_new_path] = file; - } + // If the path in the cache doesn't start with old_path, then bail on this file. + if (kv.first.rfind(old_path, 0) != 0) { + continue; + } + + const auto file_old_path = + FS::SanitizePath(kv.first, FS::DirectorySeparator::PlatformDefault); + auto file_new_path = FS::SanitizePath(new_path + DIR_SEP + kv.first.substr(old_path.size()), + FS::DirectorySeparator::PlatformDefault); + const auto& cached = cache[file_old_path]; + + if (cached.expired()) { + continue; } + + auto file = cached.lock(); + file->Open(file_new_path, "r+b"); + cache.erase(file_old_path); + cache.insert_or_assign(std::move(file_new_path), std::move(file)); } return OpenDirectory(new_path, Mode::ReadWrite); @@ -214,15 +224,21 @@ VirtualDir RealVfsFilesystem::MoveDirectory(std::string_view old_path_, bool RealVfsFilesystem::DeleteDirectory(std::string_view path_) { const auto path = FS::SanitizePath(path_, FS::DirectorySeparator::PlatformDefault); + for (auto& kv : cache) { - // Path in cache starts with old_path - if (kv.first.rfind(path, 0) == 0) { - if (!cache[kv.first].expired()) { - cache[kv.first].lock()->Close(); - } - cache.erase(kv.first); + // If the path in the cache doesn't start with path, then bail on this file. + if (kv.first.rfind(path, 0) != 0) { + continue; } + + const auto& entry = cache[kv.first]; + if (!entry.expired()) { + entry.lock()->Close(); + } + + cache.erase(kv.first); } + return FS::DeleteDirRecursively(path); } @@ -260,14 +276,14 @@ bool RealVfsFile::IsReadable() const { } std::size_t RealVfsFile::Read(u8* data, std::size_t length, std::size_t offset) const { - if (!backing->Seek(offset, SEEK_SET)) { + if (!backing->Seek(static_cast<s64>(offset), SEEK_SET)) { return 0; } return backing->ReadBytes(data, length); } std::size_t RealVfsFile::Write(const u8* data, std::size_t length, std::size_t offset) { - if (!backing->Seek(offset, SEEK_SET)) { + if (!backing->Seek(static_cast<s64>(offset), SEEK_SET)) { return 0; } return backing->WriteBytes(data, length); diff --git a/src/core/settings.h b/src/core/settings.h index bb145f193..3681b5e9d 100644 --- a/src/core/settings.h +++ b/src/core/settings.h @@ -359,7 +359,8 @@ enum class GPUAccuracy : u32 { enum class CPUAccuracy { Accurate = 0, - DebugMode = 1, + Unsafe = 1, + DebugMode = 2, }; extern bool configuring_global; @@ -419,6 +420,9 @@ struct Values { bool cpuopt_misc_ir; bool cpuopt_reduce_misalign_checks; + bool cpuopt_unsafe_unfuse_fma; + bool cpuopt_unsafe_reduce_fp_error; + // Renderer Setting<RendererBackend> renderer_backend; bool renderer_debug; diff --git a/src/video_core/engines/fermi_2d.cpp b/src/video_core/engines/fermi_2d.cpp index ff10ff40d..6e50661a3 100644 --- a/src/video_core/engines/fermi_2d.cpp +++ b/src/video_core/engines/fermi_2d.cpp @@ -10,7 +10,13 @@ namespace Tegra::Engines { -Fermi2D::Fermi2D(VideoCore::RasterizerInterface& rasterizer) : rasterizer{rasterizer} {} +Fermi2D::Fermi2D() = default; + +Fermi2D::~Fermi2D() = default; + +void Fermi2D::BindRasterizer(VideoCore::RasterizerInterface& rasterizer_) { + rasterizer = &rasterizer_; +} void Fermi2D::CallMethod(u32 method, u32 method_argument, bool is_last_call) { ASSERT_MSG(method < Regs::NUM_REGS, @@ -87,7 +93,7 @@ void Fermi2D::HandleSurfaceCopy() { copy_config.src_rect = src_rect; copy_config.dst_rect = dst_rect; - if (!rasterizer.AccelerateSurfaceCopy(regs.src, regs.dst, copy_config)) { + if (!rasterizer->AccelerateSurfaceCopy(regs.src, regs.dst, copy_config)) { UNIMPLEMENTED(); } } diff --git a/src/video_core/engines/fermi_2d.h b/src/video_core/engines/fermi_2d.h index 8f37d053f..213abfaae 100644 --- a/src/video_core/engines/fermi_2d.h +++ b/src/video_core/engines/fermi_2d.h @@ -34,8 +34,11 @@ namespace Tegra::Engines { class Fermi2D final : public EngineInterface { public: - explicit Fermi2D(VideoCore::RasterizerInterface& rasterizer); - ~Fermi2D() = default; + explicit Fermi2D(); + ~Fermi2D(); + + /// Binds a rasterizer to this engine. + void BindRasterizer(VideoCore::RasterizerInterface& rasterizer); /// Write the value to the register identified by method. void CallMethod(u32 method, u32 method_argument, bool is_last_call) override; @@ -149,7 +152,7 @@ public: }; private: - VideoCore::RasterizerInterface& rasterizer; + VideoCore::RasterizerInterface* rasterizer; /// Performs the copy from the source surface to the destination surface as configured in the /// registers. diff --git a/src/video_core/engines/kepler_compute.cpp b/src/video_core/engines/kepler_compute.cpp index a82b06a38..898370739 100644 --- a/src/video_core/engines/kepler_compute.cpp +++ b/src/video_core/engines/kepler_compute.cpp @@ -16,14 +16,15 @@ namespace Tegra::Engines { -KeplerCompute::KeplerCompute(Core::System& system, VideoCore::RasterizerInterface& rasterizer, - MemoryManager& memory_manager) - : system{system}, rasterizer{rasterizer}, memory_manager{memory_manager}, upload_state{ - memory_manager, - regs.upload} {} +KeplerCompute::KeplerCompute(Core::System& system_, MemoryManager& memory_manager_) + : system{system_}, memory_manager{memory_manager_}, upload_state{memory_manager, regs.upload} {} KeplerCompute::~KeplerCompute() = default; +void KeplerCompute::BindRasterizer(VideoCore::RasterizerInterface& rasterizer_) { + rasterizer = &rasterizer_; +} + void KeplerCompute::CallMethod(u32 method, u32 method_argument, bool is_last_call) { ASSERT_MSG(method < Regs::NUM_REGS, "Invalid KeplerCompute register, increase the size of the Regs structure"); @@ -104,11 +105,11 @@ SamplerDescriptor KeplerCompute::AccessSampler(u32 handle) const { } VideoCore::GuestDriverProfile& KeplerCompute::AccessGuestDriverProfile() { - return rasterizer.AccessGuestDriverProfile(); + return rasterizer->AccessGuestDriverProfile(); } const VideoCore::GuestDriverProfile& KeplerCompute::AccessGuestDriverProfile() const { - return rasterizer.AccessGuestDriverProfile(); + return rasterizer->AccessGuestDriverProfile(); } void KeplerCompute::ProcessLaunch() { @@ -119,7 +120,7 @@ void KeplerCompute::ProcessLaunch() { const GPUVAddr code_addr = regs.code_loc.Address() + launch_description.program_start; LOG_TRACE(HW_GPU, "Compute invocation launched at address 0x{:016x}", code_addr); - rasterizer.DispatchCompute(code_addr); + rasterizer->DispatchCompute(code_addr); } Texture::TICEntry KeplerCompute::GetTICEntry(u32 tic_index) const { diff --git a/src/video_core/engines/kepler_compute.h b/src/video_core/engines/kepler_compute.h index b7f668d88..7f2500aab 100644 --- a/src/video_core/engines/kepler_compute.h +++ b/src/video_core/engines/kepler_compute.h @@ -42,10 +42,12 @@ namespace Tegra::Engines { class KeplerCompute final : public ConstBufferEngineInterface, public EngineInterface { public: - explicit KeplerCompute(Core::System& system, VideoCore::RasterizerInterface& rasterizer, - MemoryManager& memory_manager); + explicit KeplerCompute(Core::System& system, MemoryManager& memory_manager); ~KeplerCompute(); + /// Binds a rasterizer to this engine. + void BindRasterizer(VideoCore::RasterizerInterface& rasterizer); + static constexpr std::size_t NumConstBuffers = 8; struct Regs { @@ -230,11 +232,6 @@ public: const VideoCore::GuestDriverProfile& AccessGuestDriverProfile() const override; private: - Core::System& system; - VideoCore::RasterizerInterface& rasterizer; - MemoryManager& memory_manager; - Upload::State upload_state; - void ProcessLaunch(); /// Retrieves information about a specific TIC entry from the TIC buffer. @@ -242,6 +239,11 @@ private: /// Retrieves information about a specific TSC entry from the TSC buffer. Texture::TSCEntry GetTSCEntry(u32 tsc_index) const; + + Core::System& system; + MemoryManager& memory_manager; + VideoCore::RasterizerInterface* rasterizer = nullptr; + Upload::State upload_state; }; #define ASSERT_REG_POSITION(field_name, position) \ diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp index c01436295..33854445f 100644 --- a/src/video_core/engines/maxwell_3d.cpp +++ b/src/video_core/engines/maxwell_3d.cpp @@ -22,14 +22,19 @@ using VideoCore::QueryType; /// First register id that is actually a Macro call. constexpr u32 MacroRegistersStart = 0xE00; -Maxwell3D::Maxwell3D(Core::System& system, VideoCore::RasterizerInterface& rasterizer, - MemoryManager& memory_manager) - : system{system}, rasterizer{rasterizer}, memory_manager{memory_manager}, - macro_engine{GetMacroEngine(*this)}, upload_state{memory_manager, regs.upload} { +Maxwell3D::Maxwell3D(Core::System& system_, MemoryManager& memory_manager_) + : system{system_}, memory_manager{memory_manager_}, macro_engine{GetMacroEngine(*this)}, + upload_state{memory_manager, regs.upload} { dirty.flags.flip(); InitializeRegisterDefaults(); } +Maxwell3D::~Maxwell3D() = default; + +void Maxwell3D::BindRasterizer(VideoCore::RasterizerInterface& rasterizer_) { + rasterizer = &rasterizer_; +} + void Maxwell3D::InitializeRegisterDefaults() { // Initializes registers to their default values - what games expect them to be at boot. This is // for certain registers that may not be explicitly set by games. @@ -192,7 +197,7 @@ void Maxwell3D::CallMethod(u32 method, u32 method_argument, bool is_last_call) { switch (method) { case MAXWELL3D_REG_INDEX(wait_for_idle): { - rasterizer.WaitForIdle(); + rasterizer->WaitForIdle(); break; } case MAXWELL3D_REG_INDEX(shadow_ram_control): { @@ -402,7 +407,7 @@ void Maxwell3D::FlushMMEInlineDraw() { const bool is_indexed = mme_draw.current_mode == MMEDrawMode::Indexed; if (ShouldExecute()) { - rasterizer.Draw(is_indexed, true); + rasterizer->Draw(is_indexed, true); } // TODO(bunnei): Below, we reset vertex count so that we can use these registers to determine if @@ -465,7 +470,7 @@ void Maxwell3D::ProcessQueryGet() { switch (regs.query.query_get.operation) { case Regs::QueryOperation::Release: if (regs.query.query_get.fence == 1) { - rasterizer.SignalSemaphore(regs.query.QueryAddress(), regs.query.query_sequence); + rasterizer->SignalSemaphore(regs.query.QueryAddress(), regs.query.query_sequence); } else { StampQueryResult(regs.query.query_sequence, regs.query.query_get.short_query == 0); } @@ -533,7 +538,7 @@ void Maxwell3D::ProcessQueryCondition() { void Maxwell3D::ProcessCounterReset() { switch (regs.counter_reset) { case Regs::CounterReset::SampleCnt: - rasterizer.ResetCounter(QueryType::SamplesPassed); + rasterizer->ResetCounter(QueryType::SamplesPassed); break; default: LOG_DEBUG(Render_OpenGL, "Unimplemented counter reset={}", @@ -547,7 +552,7 @@ void Maxwell3D::ProcessSyncPoint() { const u32 increment = regs.sync_info.increment.Value(); [[maybe_unused]] const u32 cache_flush = regs.sync_info.unknown.Value(); if (increment) { - rasterizer.SignalSyncPoint(sync_point); + rasterizer->SignalSyncPoint(sync_point); } } @@ -570,7 +575,7 @@ void Maxwell3D::DrawArrays() { const bool is_indexed{regs.index_array.count && !regs.vertex_buffer.count}; if (ShouldExecute()) { - rasterizer.Draw(is_indexed, false); + rasterizer->Draw(is_indexed, false); } // TODO(bunnei): Below, we reset vertex count so that we can use these registers to determine if @@ -590,8 +595,8 @@ std::optional<u64> Maxwell3D::GetQueryResult() { return 0; case Regs::QuerySelect::SamplesPassed: // Deferred. - rasterizer.Query(regs.query.QueryAddress(), VideoCore::QueryType::SamplesPassed, - system.GPU().GetTicks()); + rasterizer->Query(regs.query.QueryAddress(), VideoCore::QueryType::SamplesPassed, + system.GPU().GetTicks()); return {}; default: LOG_DEBUG(HW_GPU, "Unimplemented query select type {}", @@ -718,7 +723,7 @@ void Maxwell3D::ProcessClearBuffers() { regs.clear_buffers.R == regs.clear_buffers.B && regs.clear_buffers.R == regs.clear_buffers.A); - rasterizer.Clear(); + rasterizer->Clear(); } u32 Maxwell3D::AccessConstBuffer32(ShaderType stage, u64 const_buffer, u64 offset) const { @@ -752,11 +757,11 @@ SamplerDescriptor Maxwell3D::AccessSampler(u32 handle) const { } VideoCore::GuestDriverProfile& Maxwell3D::AccessGuestDriverProfile() { - return rasterizer.AccessGuestDriverProfile(); + return rasterizer->AccessGuestDriverProfile(); } const VideoCore::GuestDriverProfile& Maxwell3D::AccessGuestDriverProfile() const { - return rasterizer.AccessGuestDriverProfile(); + return rasterizer->AccessGuestDriverProfile(); } } // namespace Tegra::Engines diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h index c97eeb792..bc289c55d 100644 --- a/src/video_core/engines/maxwell_3d.h +++ b/src/video_core/engines/maxwell_3d.h @@ -51,9 +51,11 @@ namespace Tegra::Engines { class Maxwell3D final : public ConstBufferEngineInterface, public EngineInterface { public: - explicit Maxwell3D(Core::System& system, VideoCore::RasterizerInterface& rasterizer, - MemoryManager& memory_manager); - ~Maxwell3D() = default; + explicit Maxwell3D(Core::System& system, MemoryManager& memory_manager); + ~Maxwell3D(); + + /// Binds a rasterizer to this engine. + void BindRasterizer(VideoCore::RasterizerInterface& rasterizer); /// Register structure of the Maxwell3D engine. /// TODO(Subv): This structure will need to be made bigger as more registers are discovered. @@ -1418,12 +1420,12 @@ public: return execute_on; } - VideoCore::RasterizerInterface& GetRasterizer() { - return rasterizer; + VideoCore::RasterizerInterface& Rasterizer() { + return *rasterizer; } - const VideoCore::RasterizerInterface& GetRasterizer() const { - return rasterizer; + const VideoCore::RasterizerInterface& Rasterizer() const { + return *rasterizer; } /// Notify a memory write has happened. @@ -1460,11 +1462,10 @@ private: void InitializeRegisterDefaults(); Core::System& system; - - VideoCore::RasterizerInterface& rasterizer; - MemoryManager& memory_manager; + VideoCore::RasterizerInterface* rasterizer = nullptr; + /// Start offsets of each macro in macro_memory std::array<u32, 0x80> macro_positions = {}; diff --git a/src/video_core/fence_manager.h b/src/video_core/fence_manager.h index 8b2a6a42c..06cc12d5a 100644 --- a/src/video_core/fence_manager.h +++ b/src/video_core/fence_manager.h @@ -5,15 +5,10 @@ #pragma once #include <algorithm> -#include <array> -#include <memory> #include <queue> -#include "common/assert.h" #include "common/common_types.h" #include "core/core.h" -#include "core/memory.h" -#include "core/settings.h" #include "video_core/gpu.h" #include "video_core/memory_manager.h" #include "video_core/rasterizer_interface.h" diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp index 512578c8b..acb6e6d46 100644 --- a/src/video_core/gpu.cpp +++ b/src/video_core/gpu.cpp @@ -27,21 +27,28 @@ namespace Tegra { MICROPROFILE_DEFINE(GPU_wait, "GPU", "Wait for the GPU", MP_RGB(128, 128, 192)); -GPU::GPU(Core::System& system, std::unique_ptr<VideoCore::RendererBase>&& renderer_, bool is_async) - : system{system}, renderer{std::move(renderer_)}, is_async{is_async} { - auto& rasterizer{renderer->Rasterizer()}; - memory_manager = std::make_unique<Tegra::MemoryManager>(system, rasterizer); - dma_pusher = std::make_unique<Tegra::DmaPusher>(system, *this); - maxwell_3d = std::make_unique<Engines::Maxwell3D>(system, rasterizer, *memory_manager); - fermi_2d = std::make_unique<Engines::Fermi2D>(rasterizer); - kepler_compute = std::make_unique<Engines::KeplerCompute>(system, rasterizer, *memory_manager); - maxwell_dma = std::make_unique<Engines::MaxwellDMA>(system, *memory_manager); - kepler_memory = std::make_unique<Engines::KeplerMemory>(system, *memory_manager); - shader_notify = std::make_unique<VideoCore::ShaderNotify>(); -} +GPU::GPU(Core::System& system_, bool is_async_) + : system{system_}, dma_pusher{std::make_unique<Tegra::DmaPusher>(system, *this)}, + memory_manager{std::make_unique<Tegra::MemoryManager>(system)}, + maxwell_3d{std::make_unique<Engines::Maxwell3D>(system, *memory_manager)}, + fermi_2d{std::make_unique<Engines::Fermi2D>()}, + kepler_compute{std::make_unique<Engines::KeplerCompute>(system, *memory_manager)}, + maxwell_dma{std::make_unique<Engines::MaxwellDMA>(system, *memory_manager)}, + kepler_memory{std::make_unique<Engines::KeplerMemory>(system, *memory_manager)}, + shader_notify{std::make_unique<VideoCore::ShaderNotify>()}, is_async{is_async_} {} GPU::~GPU() = default; +void GPU::BindRenderer(std::unique_ptr<VideoCore::RendererBase> renderer_) { + renderer = std::move(renderer_); + + VideoCore::RasterizerInterface& rasterizer = renderer->Rasterizer(); + memory_manager->BindRasterizer(rasterizer); + maxwell_3d->BindRasterizer(rasterizer); + fermi_2d->BindRasterizer(rasterizer); + kepler_compute->BindRasterizer(rasterizer); +} + Engines::Maxwell3D& GPU::Maxwell3D() { return *maxwell_3d; } diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h index ebfc7b0c7..c7d11deb2 100644 --- a/src/video_core/gpu.h +++ b/src/video_core/gpu.h @@ -142,11 +142,6 @@ class MemoryManager; class GPU { public: - explicit GPU(Core::System& system, std::unique_ptr<VideoCore::RendererBase>&& renderer, - bool is_async); - - virtual ~GPU(); - struct MethodCall { u32 method{}; u32 argument{}; @@ -162,6 +157,12 @@ public: method_count(method_count) {} }; + explicit GPU(Core::System& system, bool is_async); + virtual ~GPU(); + + /// Binds a renderer to the GPU. + void BindRenderer(std::unique_ptr<VideoCore::RendererBase> renderer); + /// Calls a GPU method. void CallMethod(const MethodCall& method_call); @@ -345,8 +346,8 @@ private: bool ExecuteMethodOnEngine(u32 method); protected: - std::unique_ptr<Tegra::DmaPusher> dma_pusher; Core::System& system; + std::unique_ptr<Tegra::DmaPusher> dma_pusher; std::unique_ptr<VideoCore::RendererBase> renderer; private: diff --git a/src/video_core/gpu_asynch.cpp b/src/video_core/gpu_asynch.cpp index 7b855f63e..70a3d5738 100644 --- a/src/video_core/gpu_asynch.cpp +++ b/src/video_core/gpu_asynch.cpp @@ -10,16 +10,14 @@ namespace VideoCommon { -GPUAsynch::GPUAsynch(Core::System& system, std::unique_ptr<VideoCore::RendererBase>&& renderer_, - std::unique_ptr<Core::Frontend::GraphicsContext>&& context) - : GPU(system, std::move(renderer_), true), gpu_thread{system}, - cpu_context(renderer->GetRenderWindow().CreateSharedContext()), - gpu_context(std::move(context)) {} +GPUAsynch::GPUAsynch(Core::System& system) : GPU{system, true}, gpu_thread{system} {} GPUAsynch::~GPUAsynch() = default; void GPUAsynch::Start() { - gpu_thread.StartThread(*renderer, *gpu_context, *dma_pusher); + gpu_thread.StartThread(*renderer, renderer->Context(), *dma_pusher); + cpu_context = renderer->GetRenderWindow().CreateSharedContext(); + cpu_context->MakeCurrent(); } void GPUAsynch::ObtainContext() { diff --git a/src/video_core/gpu_asynch.h b/src/video_core/gpu_asynch.h index 15e9f1d38..f89c855a5 100644 --- a/src/video_core/gpu_asynch.h +++ b/src/video_core/gpu_asynch.h @@ -20,8 +20,7 @@ namespace VideoCommon { /// Implementation of GPU interface that runs the GPU asynchronously class GPUAsynch final : public Tegra::GPU { public: - explicit GPUAsynch(Core::System& system, std::unique_ptr<VideoCore::RendererBase>&& renderer, - std::unique_ptr<Core::Frontend::GraphicsContext>&& context); + explicit GPUAsynch(Core::System& system); ~GPUAsynch() override; void Start() override; @@ -42,7 +41,6 @@ protected: private: GPUThread::ThreadManager gpu_thread; std::unique_ptr<Core::Frontend::GraphicsContext> cpu_context; - std::unique_ptr<Core::Frontend::GraphicsContext> gpu_context; }; } // namespace VideoCommon diff --git a/src/video_core/gpu_synch.cpp b/src/video_core/gpu_synch.cpp index aaeb9811d..1ca47ddef 100644 --- a/src/video_core/gpu_synch.cpp +++ b/src/video_core/gpu_synch.cpp @@ -7,20 +7,18 @@ namespace VideoCommon { -GPUSynch::GPUSynch(Core::System& system, std::unique_ptr<VideoCore::RendererBase>&& renderer, - std::unique_ptr<Core::Frontend::GraphicsContext>&& context) - : GPU(system, std::move(renderer), false), context{std::move(context)} {} +GPUSynch::GPUSynch(Core::System& system) : GPU{system, false} {} GPUSynch::~GPUSynch() = default; void GPUSynch::Start() {} void GPUSynch::ObtainContext() { - context->MakeCurrent(); + renderer->Context().MakeCurrent(); } void GPUSynch::ReleaseContext() { - context->DoneCurrent(); + renderer->Context().DoneCurrent(); } void GPUSynch::PushGPUEntries(Tegra::CommandList&& entries) { diff --git a/src/video_core/gpu_synch.h b/src/video_core/gpu_synch.h index 762c20aa5..297258cb1 100644 --- a/src/video_core/gpu_synch.h +++ b/src/video_core/gpu_synch.h @@ -19,8 +19,7 @@ namespace VideoCommon { /// Implementation of GPU interface that runs the GPU synchronously class GPUSynch final : public Tegra::GPU { public: - explicit GPUSynch(Core::System& system, std::unique_ptr<VideoCore::RendererBase>&& renderer, - std::unique_ptr<Core::Frontend::GraphicsContext>&& context); + explicit GPUSynch(Core::System& system); ~GPUSynch() override; void Start() override; @@ -36,9 +35,6 @@ public: protected: void TriggerCpuInterrupt([[maybe_unused]] u32 syncpoint_id, [[maybe_unused]] u32 value) const override {} - -private: - std::unique_ptr<Core::Frontend::GraphicsContext> context; }; } // namespace VideoCommon diff --git a/src/video_core/macro/macro_hle.cpp b/src/video_core/macro/macro_hle.cpp index 0c9ff59a4..df00b57df 100644 --- a/src/video_core/macro/macro_hle.cpp +++ b/src/video_core/macro/macro_hle.cpp @@ -24,7 +24,7 @@ void HLE_771BB18C62444DA0(Engines::Maxwell3D& maxwell3d, const std::vector<u32>& maxwell3d.regs.index_array.first = parameters[4]; if (maxwell3d.ShouldExecute()) { - maxwell3d.GetRasterizer().Draw(true, true); + maxwell3d.Rasterizer().Draw(true, true); } maxwell3d.regs.index_array.count = 0; maxwell3d.mme_draw.instance_count = 0; @@ -42,7 +42,7 @@ void HLE_0D61FC9FAAC9FCAD(Engines::Maxwell3D& maxwell3d, const std::vector<u32>& maxwell3d.mme_draw.instance_count = count; if (maxwell3d.ShouldExecute()) { - maxwell3d.GetRasterizer().Draw(false, true); + maxwell3d.Rasterizer().Draw(false, true); } maxwell3d.regs.vertex_buffer.count = 0; maxwell3d.mme_draw.instance_count = 0; @@ -65,7 +65,7 @@ void HLE_0217920100488FF7(Engines::Maxwell3D& maxwell3d, const std::vector<u32>& maxwell3d.regs.draw.topology.Assign( static_cast<Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology>(parameters[0])); if (maxwell3d.ShouldExecute()) { - maxwell3d.GetRasterizer().Draw(true, true); + maxwell3d.Rasterizer().Draw(true, true); } maxwell3d.regs.reg_array[0x446] = 0x0; // vertex id base? maxwell3d.regs.index_array.count = 0; diff --git a/src/video_core/macro/macro_interpreter.cpp b/src/video_core/macro/macro_interpreter.cpp index aa5256419..bd01fd1f2 100644 --- a/src/video_core/macro/macro_interpreter.cpp +++ b/src/video_core/macro/macro_interpreter.cpp @@ -34,7 +34,6 @@ void MacroInterpreterImpl::Execute(const std::vector<u32>& parameters, u32 metho this->parameters = std::make_unique<u32[]>(num_parameters); } std::memcpy(this->parameters.get(), parameters.data(), num_parameters * sizeof(u32)); - this->num_parameters = num_parameters; // Execute the code until we hit an exit condition. bool keep_executing = true; diff --git a/src/video_core/memory_manager.cpp b/src/video_core/memory_manager.cpp index 844164645..c217f5bb2 100644 --- a/src/video_core/memory_manager.cpp +++ b/src/video_core/memory_manager.cpp @@ -14,11 +14,15 @@ namespace Tegra { -MemoryManager::MemoryManager(Core::System& system, VideoCore::RasterizerInterface& rasterizer) - : system{system}, rasterizer{rasterizer}, page_table(page_table_size) {} +MemoryManager::MemoryManager(Core::System& system_) + : system{system_}, page_table(page_table_size) {} MemoryManager::~MemoryManager() = default; +void MemoryManager::BindRasterizer(VideoCore::RasterizerInterface& rasterizer_) { + rasterizer = &rasterizer_; +} + GPUVAddr MemoryManager::UpdateRange(GPUVAddr gpu_addr, PageEntry page_entry, std::size_t size) { u64 remaining_size{size}; for (u64 offset{}; offset < size; offset += page_size) { @@ -217,7 +221,7 @@ void MemoryManager::ReadBlock(GPUVAddr gpu_src_addr, void* dest_buffer, std::siz // Flush must happen on the rasterizer interface, such that memory is always synchronous // when it is read (even when in asynchronous GPU mode). Fixes Dead Cells title menu. - rasterizer.FlushRegion(src_addr, copy_amount); + rasterizer->FlushRegion(src_addr, copy_amount); system.Memory().ReadBlockUnsafe(src_addr, dest_buffer, copy_amount); } @@ -266,7 +270,7 @@ void MemoryManager::WriteBlock(GPUVAddr gpu_dest_addr, const void* src_buffer, s // Invalidate must happen on the rasterizer interface, such that memory is always // synchronous when it is written (even when in asynchronous GPU mode). - rasterizer.InvalidateRegion(dest_addr, copy_amount); + rasterizer->InvalidateRegion(dest_addr, copy_amount); system.Memory().WriteBlockUnsafe(dest_addr, src_buffer, copy_amount); } diff --git a/src/video_core/memory_manager.h b/src/video_core/memory_manager.h index 681bd9588..8953fcb53 100644 --- a/src/video_core/memory_manager.h +++ b/src/video_core/memory_manager.h @@ -68,9 +68,12 @@ static_assert(sizeof(PageEntry) == 4, "PageEntry is too large"); class MemoryManager final { public: - explicit MemoryManager(Core::System& system, VideoCore::RasterizerInterface& rasterizer); + explicit MemoryManager(Core::System& system); ~MemoryManager(); + /// Binds a renderer to the memory manager. + void BindRasterizer(VideoCore::RasterizerInterface& rasterizer); + std::optional<VAddr> GpuToCpuAddress(GPUVAddr addr) const; template <typename T> @@ -141,7 +144,7 @@ private: Core::System& system; - VideoCore::RasterizerInterface& rasterizer; + VideoCore::RasterizerInterface* rasterizer = nullptr; std::vector<PageEntry> page_table; }; diff --git a/src/video_core/renderer_base.cpp b/src/video_core/renderer_base.cpp index dfb06e87e..a93a1732c 100644 --- a/src/video_core/renderer_base.cpp +++ b/src/video_core/renderer_base.cpp @@ -9,7 +9,9 @@ namespace VideoCore { -RendererBase::RendererBase(Core::Frontend::EmuWindow& window) : render_window{window} { +RendererBase::RendererBase(Core::Frontend::EmuWindow& window_, + std::unique_ptr<Core::Frontend::GraphicsContext> context_) + : render_window{window_}, context{std::move(context_)} { RefreshBaseSettings(); } diff --git a/src/video_core/renderer_base.h b/src/video_core/renderer_base.h index 1d85219b6..649074acd 100644 --- a/src/video_core/renderer_base.h +++ b/src/video_core/renderer_base.h @@ -15,7 +15,8 @@ namespace Core::Frontend { class EmuWindow; -} +class GraphicsContext; +} // namespace Core::Frontend namespace VideoCore { @@ -25,14 +26,15 @@ struct RendererSettings { // Screenshot std::atomic<bool> screenshot_requested{false}; - void* screenshot_bits; + void* screenshot_bits{}; std::function<void()> screenshot_complete_callback; Layout::FramebufferLayout screenshot_framebuffer_layout; }; class RendererBase : NonCopyable { public: - explicit RendererBase(Core::Frontend::EmuWindow& window); + explicit RendererBase(Core::Frontend::EmuWindow& window, + std::unique_ptr<Core::Frontend::GraphicsContext> context); virtual ~RendererBase(); /// Initialize the renderer @@ -68,6 +70,14 @@ public: return *rasterizer; } + Core::Frontend::GraphicsContext& Context() { + return *context; + } + + const Core::Frontend::GraphicsContext& Context() const { + return *context; + } + Core::Frontend::EmuWindow& GetRenderWindow() { return render_window; } @@ -94,6 +104,7 @@ public: protected: Core::Frontend::EmuWindow& render_window; ///< Reference to the render window handle. std::unique_ptr<RasterizerInterface> rasterizer; + std::unique_ptr<Core::Frontend::GraphicsContext> context; f32 m_current_fps = 0.0f; ///< Current framerate, should be set by the renderer int m_current_frame = 0; ///< Current frame, should be set by the renderer diff --git a/src/video_core/renderer_opengl/gl_fence_manager.cpp b/src/video_core/renderer_opengl/gl_fence_manager.cpp index ec5421afa..3d2588dd2 100644 --- a/src/video_core/renderer_opengl/gl_fence_manager.cpp +++ b/src/video_core/renderer_opengl/gl_fence_manager.cpp @@ -4,16 +4,17 @@ #include "common/assert.h" +#include <glad/glad.h> + #include "video_core/renderer_opengl/gl_buffer_cache.h" #include "video_core/renderer_opengl/gl_fence_manager.h" namespace OpenGL { -GLInnerFence::GLInnerFence(u32 payload, bool is_stubbed) - : VideoCommon::FenceBase(payload, is_stubbed), sync_object{} {} +GLInnerFence::GLInnerFence(u32 payload, bool is_stubbed) : FenceBase(payload, is_stubbed) {} GLInnerFence::GLInnerFence(GPUVAddr address, u32 payload, bool is_stubbed) - : VideoCommon::FenceBase(address, payload, is_stubbed), sync_object{} {} + : FenceBase(address, payload, is_stubbed) {} GLInnerFence::~GLInnerFence() = default; diff --git a/src/video_core/renderer_opengl/gl_fence_manager.h b/src/video_core/renderer_opengl/gl_fence_manager.h index c917b3343..1686cf5c8 100644 --- a/src/video_core/renderer_opengl/gl_fence_manager.h +++ b/src/video_core/renderer_opengl/gl_fence_manager.h @@ -5,7 +5,6 @@ #pragma once #include <memory> -#include <glad/glad.h> #include "common/common_types.h" #include "video_core/fence_manager.h" diff --git a/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp b/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp index 52fbab3c1..40c0877c1 100644 --- a/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp +++ b/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp @@ -214,20 +214,20 @@ std::optional<std::vector<ShaderDiskCacheEntry>> ShaderDiskCacheOpenGL::LoadTran // Skip games without title id const bool has_title_id = system.CurrentProcess()->GetTitleID() != 0; if (!Settings::values.use_disk_shader_cache.GetValue() || !has_title_id) { - return {}; + return std::nullopt; } Common::FS::IOFile file(GetTransferablePath(), "rb"); if (!file.IsOpen()) { LOG_INFO(Render_OpenGL, "No transferable shader cache found"); is_usable = true; - return {}; + return std::nullopt; } u32 version{}; if (file.ReadBytes(&version, sizeof(version)) != sizeof(version)) { LOG_ERROR(Render_OpenGL, "Failed to get transferable cache version, skipping it"); - return {}; + return std::nullopt; } if (version < NativeVersion) { @@ -235,12 +235,12 @@ std::optional<std::vector<ShaderDiskCacheEntry>> ShaderDiskCacheOpenGL::LoadTran file.Close(); InvalidateTransferable(); is_usable = true; - return {}; + return std::nullopt; } if (version > NativeVersion) { LOG_WARNING(Render_OpenGL, "Transferable shader cache was generated with a newer version " "of the emulator, skipping"); - return {}; + return std::nullopt; } // Version is valid, load the shaders @@ -249,7 +249,7 @@ std::optional<std::vector<ShaderDiskCacheEntry>> ShaderDiskCacheOpenGL::LoadTran ShaderDiskCacheEntry& entry = entries.emplace_back(); if (!entry.Load(file)) { LOG_ERROR(Render_OpenGL, "Failed to load transferable raw entry, skipping"); - return {}; + return std::nullopt; } } @@ -290,12 +290,12 @@ std::optional<std::vector<ShaderDiskCachePrecompiled>> ShaderDiskCacheOpenGL::Lo ShaderCacheVersionHash file_hash{}; if (!LoadArrayFromPrecompiled(file_hash.data(), file_hash.size())) { precompiled_cache_virtual_file_offset = 0; - return {}; + return std::nullopt; } if (GetShaderCacheVersionHash() != file_hash) { LOG_INFO(Render_OpenGL, "Precompiled cache is from another version of the emulator"); precompiled_cache_virtual_file_offset = 0; - return {}; + return std::nullopt; } std::vector<ShaderDiskCachePrecompiled> entries; @@ -305,15 +305,16 @@ std::optional<std::vector<ShaderDiskCachePrecompiled>> ShaderDiskCacheOpenGL::Lo if (!LoadObjectFromPrecompiled(entry.unique_identifier) || !LoadObjectFromPrecompiled(entry.binary_format) || !LoadObjectFromPrecompiled(binary_size)) { - return {}; + return std::nullopt; } entry.binary.resize(binary_size); if (!LoadArrayFromPrecompiled(entry.binary.data(), entry.binary.size())) { - return {}; + return std::nullopt; } } - return entries; + + return std::move(entries); } void ShaderDiskCacheOpenGL::InvalidateTransferable() { diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp index 0a7bc9e2b..f403f388a 100644 --- a/src/video_core/renderer_opengl/gl_texture_cache.cpp +++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp @@ -403,7 +403,7 @@ void CachedSurface::DecorateSurfaceName() { LabelGLObject(GL_TEXTURE, texture.handle, GetGpuAddr(), params.TargetName()); } -void CachedSurfaceView::DecorateViewName(GPUVAddr gpu_addr, std::string prefix) { +void CachedSurfaceView::DecorateViewName(GPUVAddr gpu_addr, const std::string& prefix) { LabelGLObject(GL_TEXTURE, main_view.handle, gpu_addr, prefix); } diff --git a/src/video_core/renderer_opengl/gl_texture_cache.h b/src/video_core/renderer_opengl/gl_texture_cache.h index bfc4ddf5d..de8f18489 100644 --- a/src/video_core/renderer_opengl/gl_texture_cache.h +++ b/src/video_core/renderer_opengl/gl_texture_cache.h @@ -90,7 +90,7 @@ public: Tegra::Texture::SwizzleSource z_source, Tegra::Texture::SwizzleSource w_source); - void DecorateViewName(GPUVAddr gpu_addr, std::string prefix); + void DecorateViewName(GPUVAddr gpu_addr, const std::string& prefix); void MarkAsModified(u64 tick) { surface.MarkAsModified(true, tick); diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp index 14bbc3a1c..c39663db7 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.cpp +++ b/src/video_core/renderer_opengl/renderer_opengl.cpp @@ -313,10 +313,11 @@ public: } }; -RendererOpenGL::RendererOpenGL(Core::Frontend::EmuWindow& emu_window, Core::System& system, - Core::Frontend::GraphicsContext& context) - : RendererBase{emu_window}, emu_window{emu_window}, system{system}, context{context}, - program_manager{device}, has_debug_tool{HasDebugTool()} {} +RendererOpenGL::RendererOpenGL(Core::System& system_, Core::Frontend::EmuWindow& emu_window_, + Tegra::GPU& gpu_, + std::unique_ptr<Core::Frontend::GraphicsContext> context_) + : RendererBase{emu_window_, std::move(context_)}, system{system_}, + emu_window{emu_window_}, gpu{gpu_}, program_manager{device}, has_debug_tool{HasDebugTool()} {} RendererOpenGL::~RendererOpenGL() = default; @@ -384,7 +385,7 @@ void RendererOpenGL::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) { if (has_debug_tool) { glBindFramebuffer(GL_DRAW_FRAMEBUFFER, 0); Present(0); - context.SwapBuffers(); + context->SwapBuffers(); } } diff --git a/src/video_core/renderer_opengl/renderer_opengl.h b/src/video_core/renderer_opengl/renderer_opengl.h index 8b18d32e6..52ea76b7d 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.h +++ b/src/video_core/renderer_opengl/renderer_opengl.h @@ -56,8 +56,9 @@ class FrameMailbox; class RendererOpenGL final : public VideoCore::RendererBase { public: - explicit RendererOpenGL(Core::Frontend::EmuWindow& emu_window, Core::System& system, - Core::Frontend::GraphicsContext& context); + explicit RendererOpenGL(Core::System& system, Core::Frontend::EmuWindow& emu_window, + Tegra::GPU& gpu, + std::unique_ptr<Core::Frontend::GraphicsContext> context); ~RendererOpenGL() override; bool Init() override; @@ -93,9 +94,9 @@ private: bool Present(int timeout_ms); - Core::Frontend::EmuWindow& emu_window; Core::System& system; - Core::Frontend::GraphicsContext& context; + Core::Frontend::EmuWindow& emu_window; + Tegra::GPU& gpu; const Device device; StateTracker state_tracker{system}; @@ -120,7 +121,7 @@ private: std::vector<u8> gl_framebuffer_data; /// Used for transforming the framebuffer orientation - Tegra::FramebufferConfig::TransformFlags framebuffer_transform_flags; + Tegra::FramebufferConfig::TransformFlags framebuffer_transform_flags{}; Common::Rectangle<int> framebuffer_crop_rect; /// Frame presentation mailbox diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.cpp b/src/video_core/renderer_vulkan/renderer_vulkan.cpp index 6e49699d0..ae46e0444 100644 --- a/src/video_core/renderer_vulkan/renderer_vulkan.cpp +++ b/src/video_core/renderer_vulkan/renderer_vulkan.cpp @@ -237,8 +237,10 @@ std::string BuildCommaSeparatedExtensions(std::vector<std::string> available_ext } // Anonymous namespace -RendererVulkan::RendererVulkan(Core::Frontend::EmuWindow& window, Core::System& system) - : RendererBase(window), system{system} {} +RendererVulkan::RendererVulkan(Core::System& system_, Core::Frontend::EmuWindow& emu_window, + Tegra::GPU& gpu_, + std::unique_ptr<Core::Frontend::GraphicsContext> context) + : RendererBase{emu_window, std::move(context)}, system{system_}, gpu{gpu_} {} RendererVulkan::~RendererVulkan() { ShutDown(); diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.h b/src/video_core/renderer_vulkan/renderer_vulkan.h index 522b5bff8..13debbbc0 100644 --- a/src/video_core/renderer_vulkan/renderer_vulkan.h +++ b/src/video_core/renderer_vulkan/renderer_vulkan.h @@ -38,7 +38,9 @@ struct VKScreenInfo { class RendererVulkan final : public VideoCore::RendererBase { public: - explicit RendererVulkan(Core::Frontend::EmuWindow& window, Core::System& system); + explicit RendererVulkan(Core::System& system, Core::Frontend::EmuWindow& emu_window, + Tegra::GPU& gpu, + std::unique_ptr<Core::Frontend::GraphicsContext> context); ~RendererVulkan() override; bool Init() override; @@ -58,6 +60,7 @@ private: void Report() const; Core::System& system; + Tegra::GPU& gpu; Common::DynamicLibrary library; vk::InstanceDispatch dld; diff --git a/src/video_core/video_core.cpp b/src/video_core/video_core.cpp index 45f360bdd..4e3a092c7 100644 --- a/src/video_core/video_core.cpp +++ b/src/video_core/video_core.cpp @@ -3,6 +3,7 @@ // Refer to the license.txt file included. #include <memory> + #include "common/logging/log.h" #include "core/core.h" #include "core/settings.h" @@ -16,37 +17,46 @@ #include "video_core/video_core.h" namespace { -std::unique_ptr<VideoCore::RendererBase> CreateRenderer(Core::Frontend::EmuWindow& emu_window, - Core::System& system, - Core::Frontend::GraphicsContext& context) { + +std::unique_ptr<VideoCore::RendererBase> CreateRenderer( + Core::System& system, Core::Frontend::EmuWindow& emu_window, Tegra::GPU& gpu, + std::unique_ptr<Core::Frontend::GraphicsContext> context) { switch (Settings::values.renderer_backend.GetValue()) { case Settings::RendererBackend::OpenGL: - return std::make_unique<OpenGL::RendererOpenGL>(emu_window, system, context); + return std::make_unique<OpenGL::RendererOpenGL>(system, emu_window, gpu, + std::move(context)); #ifdef HAS_VULKAN case Settings::RendererBackend::Vulkan: - return std::make_unique<Vulkan::RendererVulkan>(emu_window, system); + return std::make_unique<Vulkan::RendererVulkan>(system, emu_window, gpu, + std::move(context)); #endif default: return nullptr; } } + } // Anonymous namespace namespace VideoCore { std::unique_ptr<Tegra::GPU> CreateGPU(Core::Frontend::EmuWindow& emu_window, Core::System& system) { + std::unique_ptr<Tegra::GPU> gpu; + if (Settings::values.use_asynchronous_gpu_emulation.GetValue()) { + gpu = std::make_unique<VideoCommon::GPUAsynch>(system); + } else { + gpu = std::make_unique<VideoCommon::GPUSynch>(system); + } + auto context = emu_window.CreateSharedContext(); const auto scope = context->Acquire(); - auto renderer = CreateRenderer(emu_window, system, *context); + + auto renderer = CreateRenderer(system, emu_window, *gpu, std::move(context)); if (!renderer->Init()) { return nullptr; } - if (Settings::values.use_asynchronous_gpu_emulation.GetValue()) { - return std::make_unique<VideoCommon::GPUAsynch>(system, std::move(renderer), - std::move(context)); - } - return std::make_unique<VideoCommon::GPUSynch>(system, std::move(renderer), std::move(context)); + gpu->BindRenderer(std::move(renderer)); + return gpu; } u16 GetResolutionScaleFactor(const RendererBase& renderer) { diff --git a/src/web_service/CMakeLists.txt b/src/web_service/CMakeLists.txt index 06ab7c59d..7e484b906 100644 --- a/src/web_service/CMakeLists.txt +++ b/src/web_service/CMakeLists.txt @@ -5,6 +5,7 @@ add_library(web_service STATIC verify_login.h web_backend.cpp web_backend.h + web_result.h ) create_target_directory_groups(web_service) diff --git a/src/web_service/telemetry_json.cpp b/src/web_service/telemetry_json.cpp index c89a3a0db..6215c914f 100644 --- a/src/web_service/telemetry_json.cpp +++ b/src/web_service/telemetry_json.cpp @@ -4,9 +4,9 @@ #include <nlohmann/json.hpp> #include "common/detached_tasks.h" -#include "common/web_result.h" #include "web_service/telemetry_json.h" #include "web_service/web_backend.h" +#include "web_service/web_result.h" namespace WebService { @@ -125,7 +125,7 @@ bool TelemetryJson::SubmitTestcase() { Client client(impl->host, impl->username, impl->token); auto value = client.PostJson("/gamedb/testcase", content, false); - return value.result_code == Common::WebResult::Code::Success; + return value.result_code == WebResult::Code::Success; } } // namespace WebService diff --git a/src/web_service/verify_login.cpp b/src/web_service/verify_login.cpp index bfaa5b70a..ceb55ca6b 100644 --- a/src/web_service/verify_login.cpp +++ b/src/web_service/verify_login.cpp @@ -3,9 +3,9 @@ // Refer to the license.txt file included. #include <nlohmann/json.hpp> -#include "common/web_result.h" #include "web_service/verify_login.h" #include "web_service/web_backend.h" +#include "web_service/web_result.h" namespace WebService { diff --git a/src/web_service/web_backend.cpp b/src/web_service/web_backend.cpp index 09d1651ac..74e287045 100644 --- a/src/web_service/web_backend.cpp +++ b/src/web_service/web_backend.cpp @@ -6,13 +6,14 @@ #include <cstdlib> #include <mutex> #include <string> + #include <LUrlParser.h> #include <fmt/format.h> #include <httplib.h> -#include "common/common_types.h" + #include "common/logging/log.h" -#include "common/web_result.h" #include "web_service/web_backend.h" +#include "web_service/web_result.h" namespace WebService { @@ -33,17 +34,16 @@ struct Client::Impl { } /// A generic function handles POST, GET and DELETE request together - Common::WebResult GenericRequest(const std::string& method, const std::string& path, - const std::string& data, bool allow_anonymous, - const std::string& accept) { + WebResult GenericRequest(const std::string& method, const std::string& path, + const std::string& data, bool allow_anonymous, + const std::string& accept) { if (jwt.empty()) { UpdateJWT(); } if (jwt.empty() && !allow_anonymous) { LOG_ERROR(WebService, "Credentials must be provided for authenticated requests"); - return Common::WebResult{Common::WebResult::Code::CredentialsMissing, - "Credentials needed", ""}; + return WebResult{WebResult::Code::CredentialsMissing, "Credentials needed", ""}; } auto result = GenericRequest(method, path, data, accept, jwt); @@ -62,10 +62,10 @@ struct Client::Impl { * username + token is used if jwt is empty but username and token are * not empty anonymous if all of jwt, username and token are empty */ - Common::WebResult GenericRequest(const std::string& method, const std::string& path, - const std::string& data, const std::string& accept, - const std::string& jwt = "", const std::string& username = "", - const std::string& token = "") { + WebResult GenericRequest(const std::string& method, const std::string& path, + const std::string& data, const std::string& accept, + const std::string& jwt = "", const std::string& username = "", + const std::string& token = "") { if (cli == nullptr) { auto parsedUrl = LUrlParser::clParseURL::ParseURL(host); int port; @@ -81,12 +81,12 @@ struct Client::Impl { cli = std::make_unique<httplib::SSLClient>(parsedUrl.m_Host.c_str(), port); } else { LOG_ERROR(WebService, "Bad URL scheme {}", parsedUrl.m_Scheme); - return Common::WebResult{Common::WebResult::Code::InvalidURL, "Bad URL scheme", ""}; + return WebResult{WebResult::Code::InvalidURL, "Bad URL scheme", ""}; } } if (cli == nullptr) { LOG_ERROR(WebService, "Invalid URL {}", host + path); - return Common::WebResult{Common::WebResult::Code::InvalidURL, "Invalid URL", ""}; + return WebResult{WebResult::Code::InvalidURL, "Invalid URL", ""}; } cli->set_timeout_sec(TIMEOUT_SECONDS); @@ -106,7 +106,7 @@ struct Client::Impl { std::string(API_VERSION.begin(), API_VERSION.end())); if (method != "GET") { params.emplace(std::string("Content-Type"), std::string("application/json")); - }; + } httplib::Request request; request.method = method; @@ -118,29 +118,28 @@ struct Client::Impl { if (!cli->send(request, response)) { LOG_ERROR(WebService, "{} to {} returned null", method, host + path); - return Common::WebResult{Common::WebResult::Code::LibError, "Null response", ""}; + return WebResult{WebResult::Code::LibError, "Null response", ""}; } if (response.status >= 400) { LOG_ERROR(WebService, "{} to {} returned error status code: {}", method, host + path, response.status); - return Common::WebResult{Common::WebResult::Code::HttpError, - std::to_string(response.status), ""}; + return WebResult{WebResult::Code::HttpError, std::to_string(response.status), ""}; } auto content_type = response.headers.find("content-type"); if (content_type == response.headers.end()) { LOG_ERROR(WebService, "{} to {} returned no content", method, host + path); - return Common::WebResult{Common::WebResult::Code::WrongContent, "", ""}; + return WebResult{WebResult::Code::WrongContent, "", ""}; } if (content_type->second.find(accept) == std::string::npos) { LOG_ERROR(WebService, "{} to {} returned wrong content: {}", method, host + path, content_type->second); - return Common::WebResult{Common::WebResult::Code::WrongContent, "Wrong content", ""}; + return WebResult{WebResult::Code::WrongContent, "Wrong content", ""}; } - return Common::WebResult{Common::WebResult::Code::Success, "", response.body}; + return WebResult{WebResult::Code::Success, "", response.body}; } // Retrieve a new JWT from given username and token @@ -150,7 +149,7 @@ struct Client::Impl { } auto result = GenericRequest("POST", "/jwt/internal", "", "text/html", "", username, token); - if (result.result_code != Common::WebResult::Code::Success) { + if (result.result_code != WebResult::Code::Success) { LOG_ERROR(WebService, "UpdateJWT failed"); } else { std::lock_guard lock{jwt_cache.mutex}; @@ -180,29 +179,28 @@ Client::Client(std::string host, std::string username, std::string token) Client::~Client() = default; -Common::WebResult Client::PostJson(const std::string& path, const std::string& data, - bool allow_anonymous) { +WebResult Client::PostJson(const std::string& path, const std::string& data, bool allow_anonymous) { return impl->GenericRequest("POST", path, data, allow_anonymous, "application/json"); } -Common::WebResult Client::GetJson(const std::string& path, bool allow_anonymous) { +WebResult Client::GetJson(const std::string& path, bool allow_anonymous) { return impl->GenericRequest("GET", path, "", allow_anonymous, "application/json"); } -Common::WebResult Client::DeleteJson(const std::string& path, const std::string& data, - bool allow_anonymous) { +WebResult Client::DeleteJson(const std::string& path, const std::string& data, + bool allow_anonymous) { return impl->GenericRequest("DELETE", path, data, allow_anonymous, "application/json"); } -Common::WebResult Client::GetPlain(const std::string& path, bool allow_anonymous) { +WebResult Client::GetPlain(const std::string& path, bool allow_anonymous) { return impl->GenericRequest("GET", path, "", allow_anonymous, "text/plain"); } -Common::WebResult Client::GetImage(const std::string& path, bool allow_anonymous) { +WebResult Client::GetImage(const std::string& path, bool allow_anonymous) { return impl->GenericRequest("GET", path, "", allow_anonymous, "image/png"); } -Common::WebResult Client::GetExternalJWT(const std::string& audience) { +WebResult Client::GetExternalJWT(const std::string& audience) { return impl->GenericRequest("POST", fmt::format("/jwt/external/{}", audience), "", false, "text/html"); } diff --git a/src/web_service/web_backend.h b/src/web_service/web_backend.h index 04121f17e..81f58583c 100644 --- a/src/web_service/web_backend.h +++ b/src/web_service/web_backend.h @@ -7,12 +7,10 @@ #include <memory> #include <string> -namespace Common { -struct WebResult; -} - namespace WebService { +struct WebResult; + class Client { public: Client(std::string host, std::string username, std::string token); @@ -25,8 +23,7 @@ public: * @param allow_anonymous If true, allow anonymous unauthenticated requests. * @return the result of the request. */ - Common::WebResult PostJson(const std::string& path, const std::string& data, - bool allow_anonymous); + WebResult PostJson(const std::string& path, const std::string& data, bool allow_anonymous); /** * Gets JSON from the specified path. @@ -34,7 +31,7 @@ public: * @param allow_anonymous If true, allow anonymous unauthenticated requests. * @return the result of the request. */ - Common::WebResult GetJson(const std::string& path, bool allow_anonymous); + WebResult GetJson(const std::string& path, bool allow_anonymous); /** * Deletes JSON to the specified path. @@ -43,8 +40,7 @@ public: * @param allow_anonymous If true, allow anonymous unauthenticated requests. * @return the result of the request. */ - Common::WebResult DeleteJson(const std::string& path, const std::string& data, - bool allow_anonymous); + WebResult DeleteJson(const std::string& path, const std::string& data, bool allow_anonymous); /** * Gets a plain string from the specified path. @@ -52,7 +48,7 @@ public: * @param allow_anonymous If true, allow anonymous unauthenticated requests. * @return the result of the request. */ - Common::WebResult GetPlain(const std::string& path, bool allow_anonymous); + WebResult GetPlain(const std::string& path, bool allow_anonymous); /** * Gets an PNG image from the specified path. @@ -60,14 +56,14 @@ public: * @param allow_anonymous If true, allow anonymous unauthenticated requests. * @return the result of the request. */ - Common::WebResult GetImage(const std::string& path, bool allow_anonymous); + WebResult GetImage(const std::string& path, bool allow_anonymous); /** * Requests an external JWT for the specific audience provided. * @param audience the audience of the JWT requested. * @return the result of the request. */ - Common::WebResult GetExternalJWT(const std::string& audience); + WebResult GetExternalJWT(const std::string& audience); private: struct Impl; diff --git a/src/common/web_result.h b/src/web_service/web_result.h index 8bfa2141d..3aeeb5288 100644 --- a/src/common/web_result.h +++ b/src/web_service/web_result.h @@ -7,7 +7,7 @@ #include <string> #include "common/common_types.h" -namespace Common { +namespace WebService { struct WebResult { enum class Code : u32 { Success, @@ -22,4 +22,4 @@ struct WebResult { std::string result_string; std::string returned_data; }; -} // namespace Common +} // namespace WebService diff --git a/src/yuzu/configuration/config.cpp b/src/yuzu/configuration/config.cpp index a372190cc..7af974d8d 100644 --- a/src/yuzu/configuration/config.cpp +++ b/src/yuzu/configuration/config.cpp @@ -635,6 +635,11 @@ void Config::ReadCpuValues() { ReadSetting(QStringLiteral("cpuopt_misc_ir"), true).toBool(); Settings::values.cpuopt_reduce_misalign_checks = ReadSetting(QStringLiteral("cpuopt_reduce_misalign_checks"), true).toBool(); + + Settings::values.cpuopt_unsafe_unfuse_fma = + ReadSetting(QStringLiteral("cpuopt_unsafe_unfuse_fma"), true).toBool(); + Settings::values.cpuopt_unsafe_reduce_fp_error = + ReadSetting(QStringLiteral("cpuopt_unsafe_reduce_fp_error"), true).toBool(); } qt_config->endGroup(); @@ -1132,6 +1137,11 @@ void Config::SaveCpuValues() { WriteSetting(QStringLiteral("cpuopt_misc_ir"), Settings::values.cpuopt_misc_ir, true); WriteSetting(QStringLiteral("cpuopt_reduce_misalign_checks"), Settings::values.cpuopt_reduce_misalign_checks, true); + + WriteSetting(QStringLiteral("cpuopt_unsafe_unfuse_fma"), + Settings::values.cpuopt_unsafe_unfuse_fma, true); + WriteSetting(QStringLiteral("cpuopt_unsafe_reduce_fp_error"), + Settings::values.cpuopt_unsafe_reduce_fp_error, true); } qt_config->endGroup(); diff --git a/src/yuzu/configuration/configure_cpu.cpp b/src/yuzu/configuration/configure_cpu.cpp index 7493e5ffb..37fcd6adc 100644 --- a/src/yuzu/configuration/configure_cpu.cpp +++ b/src/yuzu/configuration/configure_cpu.cpp @@ -19,6 +19,8 @@ ConfigureCpu::ConfigureCpu(QWidget* parent) : QWidget(parent), ui(new Ui::Config connect(ui->accuracy, qOverload<int>(&QComboBox::activated), this, &ConfigureCpu::AccuracyUpdated); + connect(ui->accuracy, qOverload<int>(&QComboBox::currentIndexChanged), this, + &ConfigureCpu::UpdateGroup); } ConfigureCpu::~ConfigureCpu() = default; @@ -28,6 +30,12 @@ void ConfigureCpu::SetConfiguration() { ui->accuracy->setEnabled(runtime_lock); ui->accuracy->setCurrentIndex(static_cast<int>(Settings::values.cpu_accuracy)); + UpdateGroup(static_cast<int>(Settings::values.cpu_accuracy)); + + ui->cpuopt_unsafe_unfuse_fma->setEnabled(runtime_lock); + ui->cpuopt_unsafe_unfuse_fma->setChecked(Settings::values.cpuopt_unsafe_unfuse_fma); + ui->cpuopt_unsafe_reduce_fp_error->setEnabled(runtime_lock); + ui->cpuopt_unsafe_reduce_fp_error->setChecked(Settings::values.cpuopt_unsafe_reduce_fp_error); } void ConfigureCpu::AccuracyUpdated(int index) { @@ -38,14 +46,21 @@ void ConfigureCpu::AccuracyUpdated(int index) { QMessageBox::Yes | QMessageBox::No); if (result == QMessageBox::No) { ui->accuracy->setCurrentIndex(static_cast<int>(Settings::CPUAccuracy::Accurate)); - return; + UpdateGroup(static_cast<int>(Settings::CPUAccuracy::Accurate)); } } } +void ConfigureCpu::UpdateGroup(int index) { + ui->unsafe_group->setVisible(static_cast<Settings::CPUAccuracy>(index) == + Settings::CPUAccuracy::Unsafe); +} + void ConfigureCpu::ApplyConfiguration() { Settings::values.cpu_accuracy = static_cast<Settings::CPUAccuracy>(ui->accuracy->currentIndex()); + Settings::values.cpuopt_unsafe_unfuse_fma = ui->cpuopt_unsafe_unfuse_fma->isChecked(); + Settings::values.cpuopt_unsafe_reduce_fp_error = ui->cpuopt_unsafe_reduce_fp_error->isChecked(); } void ConfigureCpu::changeEvent(QEvent* event) { diff --git a/src/yuzu/configuration/configure_cpu.h b/src/yuzu/configuration/configure_cpu.h index e4741d3a4..3c5683d81 100644 --- a/src/yuzu/configuration/configure_cpu.h +++ b/src/yuzu/configuration/configure_cpu.h @@ -26,6 +26,7 @@ private: void RetranslateUI(); void AccuracyUpdated(int index); + void UpdateGroup(int index); void SetConfiguration(); diff --git a/src/yuzu/configuration/configure_cpu.ui b/src/yuzu/configuration/configure_cpu.ui index bf6ea79bb..ebdd2e6e9 100644 --- a/src/yuzu/configuration/configure_cpu.ui +++ b/src/yuzu/configuration/configure_cpu.ui @@ -40,6 +40,11 @@ </item> <item> <property name="text"> + <string>Unsafe</string> + </property> + </item> + <item> + <property name="text"> <string>Enable Debug Mode</string> </property> </item> @@ -63,6 +68,53 @@ </layout> </item> <item> + <layout class="QVBoxLayout"> + <item> + <widget class="QGroupBox" name="unsafe_group"> + <property name="title"> + <string>Unsafe CPU Optimization Settings</string> + </property> + <layout class="QVBoxLayout"> + <item> + <widget class="QLabel"> + <property name="wordWrap"> + <bool>1</bool> + </property> + <property name="text"> + <string>These settings reduce accuracy for speed.</string> + </property> + </widget> + </item> + <item> + <widget class="QCheckBox" name="cpuopt_unsafe_unfuse_fma"> + <property name="text"> + <string>Unfuse FMA (improve performance on CPUs without FMA)</string> + </property> + <property name="toolTip"> + <string> + <div>This option improves speed by reducing accuracy of fused-multiply-add instructions on CPUs without native FMA support.</div> + </string> + </property> + </widget> + </item> + <item> + <widget class="QCheckBox" name="cpuopt_unsafe_reduce_fp_error"> + <property name="text"> + <string>Faster FRSQRTE and FRECPE</string> + </property> + <property name="toolTip"> + <string> + <div>This option improves the speed of some approximate floating-point functions by using less accurate native approximations.</div> + </string> + </property> + </widget> + </item> + </layout> + </widget> + </item> + </layout> + </item> + <item> <spacer name="verticalSpacer"> <property name="orientation"> <enum>Qt::Vertical</enum> |