diff options
Diffstat (limited to 'src')
42 files changed, 1854 insertions, 349 deletions
diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt index f87d67db5..d1f173f42 100644 --- a/src/core/CMakeLists.txt +++ b/src/core/CMakeLists.txt @@ -614,7 +614,7 @@ endif() create_target_directory_groups(core) target_link_libraries(core PUBLIC common PRIVATE audio_core video_core) -target_link_libraries(core PUBLIC Boost::boost PRIVATE fmt::fmt nlohmann_json::nlohmann_json mbedtls Opus::Opus unicorn zip) +target_link_libraries(core PUBLIC Boost::boost PRIVATE fmt::fmt nlohmann_json::nlohmann_json mbedtls opus unicorn zip) if (YUZU_ENABLE_BOXCAT) target_compile_definitions(core PRIVATE -DYUZU_ENABLE_BOXCAT) diff --git a/src/core/arm/cpu_interrupt_handler.cpp b/src/core/arm/cpu_interrupt_handler.cpp index 2f1a1a269..df0350881 100644 --- a/src/core/arm/cpu_interrupt_handler.cpp +++ b/src/core/arm/cpu_interrupt_handler.cpp @@ -2,8 +2,6 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#pragma once - #include "common/thread.h" #include "core/arm/cpu_interrupt_handler.h" diff --git a/src/core/hle/service/acc/acc.cpp b/src/core/hle/service/acc/acc.cpp index 94d8c1fc6..8ac856ec3 100644 --- a/src/core/hle/service/acc/acc.cpp +++ b/src/core/hle/service/acc/acc.cpp @@ -776,6 +776,15 @@ void Module::Interface::ListQualifiedUsers(Kernel::HLERequestContext& ctx) { rb.Push(RESULT_SUCCESS); } +void Module::Interface::ListOpenContextStoredUsers(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_ACC, "(STUBBED) called"); + + // TODO(ogniK): Handle open contexts + ctx.WriteBuffer(profile_manager->GetOpenUsers()); + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + void Module::Interface::TrySelectUserWithoutInteraction(Kernel::HLERequestContext& ctx) { LOG_DEBUG(Service_ACC, "called"); // A u8 is passed into this function which we can safely ignore. It's to determine if we have diff --git a/src/core/hle/service/acc/acc.h b/src/core/hle/service/acc/acc.h index 74ca39d6e..d4c6395c6 100644 --- a/src/core/hle/service/acc/acc.h +++ b/src/core/hle/service/acc/acc.h @@ -34,6 +34,7 @@ public: void IsUserAccountSwitchLocked(Kernel::HLERequestContext& ctx); void GetProfileEditor(Kernel::HLERequestContext& ctx); void ListQualifiedUsers(Kernel::HLERequestContext& ctx); + void ListOpenContextStoredUsers(Kernel::HLERequestContext& ctx); private: ResultCode InitializeApplicationInfoBase(); diff --git a/src/core/hle/service/acc/acc_su.cpp b/src/core/hle/service/acc/acc_su.cpp index 85620bde3..d2bb8c2c8 100644 --- a/src/core/hle/service/acc/acc_su.cpp +++ b/src/core/hle/service/acc/acc_su.cpp @@ -20,7 +20,7 @@ ACC_SU::ACC_SU(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p {6, nullptr, "GetProfileDigest"}, // 3.0.0+ {50, &ACC_SU::IsUserRegistrationRequestPermitted, "IsUserRegistrationRequestPermitted"}, {51, &ACC_SU::TrySelectUserWithoutInteraction, "TrySelectUserWithoutInteraction"}, - {60, nullptr, "ListOpenContextStoredUsers"}, // 5.0.0 - 5.1.0 + {60, &ACC_SU::ListOpenContextStoredUsers, "ListOpenContextStoredUsers"}, // 5.0.0 - 5.1.0 {99, nullptr, "DebugActivateOpenContextRetention"}, // 6.0.0+ {100, nullptr, "GetUserRegistrationNotifier"}, {101, nullptr, "GetUserStateChangeNotifier"}, diff --git a/src/core/hle/service/acc/acc_u0.cpp b/src/core/hle/service/acc/acc_u0.cpp index 49f6e20f1..cb44e06b7 100644 --- a/src/core/hle/service/acc/acc_u0.cpp +++ b/src/core/hle/service/acc/acc_u0.cpp @@ -20,7 +20,7 @@ ACC_U0::ACC_U0(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p {6, nullptr, "GetProfileDigest"}, // 3.0.0+ {50, &ACC_U0::IsUserRegistrationRequestPermitted, "IsUserRegistrationRequestPermitted"}, {51, &ACC_U0::TrySelectUserWithoutInteraction, "TrySelectUserWithoutInteraction"}, - {60, nullptr, "ListOpenContextStoredUsers"}, // 5.0.0 - 5.1.0 + {60, &ACC_U0::ListOpenContextStoredUsers, "ListOpenContextStoredUsers"}, // 5.0.0 - 5.1.0 {99, nullptr, "DebugActivateOpenContextRetention"}, // 6.0.0+ {100, &ACC_U0::InitializeApplicationInfo, "InitializeApplicationInfo"}, {101, &ACC_U0::GetBaasAccountManagerForApplication, "GetBaasAccountManagerForApplication"}, @@ -30,7 +30,7 @@ ACC_U0::ACC_U0(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p {111, nullptr, "ClearSaveDataThumbnail"}, {120, nullptr, "CreateGuestLoginRequest"}, {130, nullptr, "LoadOpenContext"}, // 5.0.0+ - {131, nullptr, "ListOpenContextStoredUsers"}, // 6.0.0+ + {131, &ACC_U0::ListOpenContextStoredUsers, "ListOpenContextStoredUsers"}, // 6.0.0+ {140, &ACC_U0::InitializeApplicationInfoRestricted, "InitializeApplicationInfoRestricted"}, // 6.0.0+ {141, &ACC_U0::ListQualifiedUsers, "ListQualifiedUsers"}, // 6.0.0+ {150, &ACC_U0::IsUserAccountSwitchLocked, "IsUserAccountSwitchLocked"}, // 6.0.0+ diff --git a/src/core/hle/service/acc/acc_u1.cpp b/src/core/hle/service/acc/acc_u1.cpp index f47004f84..a4aa5316a 100644 --- a/src/core/hle/service/acc/acc_u1.cpp +++ b/src/core/hle/service/acc/acc_u1.cpp @@ -20,7 +20,7 @@ ACC_U1::ACC_U1(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p {6, nullptr, "GetProfileDigest"}, // 3.0.0+ {50, &ACC_U1::IsUserRegistrationRequestPermitted, "IsUserRegistrationRequestPermitted"}, {51, &ACC_U1::TrySelectUserWithoutInteraction, "TrySelectUserWithoutInteraction"}, - {60, nullptr, "ListOpenContextStoredUsers"}, // 5.0.0 - 5.1.0 + {60, &ACC_U1::ListOpenContextStoredUsers, "ListOpenContextStoredUsers"}, // 5.0.0 - 5.1.0 {99, nullptr, "DebugActivateOpenContextRetention"}, // 6.0.0+ {100, nullptr, "GetUserRegistrationNotifier"}, {101, nullptr, "GetUserStateChangeNotifier"}, diff --git a/src/core/hle/service/am/am.cpp b/src/core/hle/service/am/am.cpp index 20f366635..24cfb370b 100644 --- a/src/core/hle/service/am/am.cpp +++ b/src/core/hle/service/am/am.cpp @@ -10,6 +10,7 @@ #include "core/core.h" #include "core/file_sys/control_metadata.h" #include "core/file_sys/patch_manager.h" +#include "core/file_sys/registered_cache.h" #include "core/file_sys/savedata_factory.h" #include "core/hle/ipc_helpers.h" #include "core/hle/kernel/kernel.h" @@ -841,7 +842,7 @@ public: {110, nullptr, "NeedsToExitProcess"}, {120, nullptr, "GetLibraryAppletInfo"}, {150, nullptr, "RequestForAppletToGetForeground"}, - {160, nullptr, "GetIndirectLayerConsumerHandle"}, + {160, &ILibraryAppletAccessor::GetIndirectLayerConsumerHandle, "GetIndirectLayerConsumerHandle"}, }; // clang-format on @@ -960,6 +961,18 @@ private: rb.PushCopyObjects(applet->GetBroker().GetInteractiveDataEvent()); } + void GetIndirectLayerConsumerHandle(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_AM, "(STUBBED) called"); + + // We require a non-zero handle to be valid. Using 0xdeadbeef allows us to trace if this is + // actually used anywhere + constexpr u64 handle = 0xdeadbeef; + + IPC::ResponseBuilder rb{ctx, 4}; + rb.Push(RESULT_SUCCESS); + rb.Push(handle); + } + std::shared_ptr<Applets::Applet> applet; }; @@ -1341,14 +1354,25 @@ void IApplicationFunctions::GetDisplayVersion(Kernel::HLERequestContext& ctx) { std::array<u8, 0x10> version_string{}; - FileSys::PatchManager pm{system.CurrentProcess()->GetTitleID()}; - const auto res = pm.GetControlMetadata(); + const auto res = [this] { + const auto title_id = system.CurrentProcess()->GetTitleID(); + + FileSys::PatchManager pm{title_id}; + auto res = pm.GetControlMetadata(); + if (res.first != nullptr) { + return res; + } + + FileSys::PatchManager pm_update{FileSys::GetUpdateTitleID(title_id)}; + return pm_update.GetControlMetadata(); + }(); + if (res.first != nullptr) { const auto& version = res.first->GetVersionString(); std::copy(version.begin(), version.end(), version_string.begin()); } else { - constexpr u128 default_version = {1, 0}; - std::memcpy(version_string.data(), default_version.data(), sizeof(u128)); + constexpr char default_version[]{"1.0.0"}; + std::memcpy(version_string.data(), default_version, sizeof(default_version)); } IPC::ResponseBuilder rb{ctx, 6}; diff --git a/src/core/hle/service/set/set.cpp b/src/core/hle/service/set/set.cpp index f3b4b286c..e5cfd2101 100644 --- a/src/core/hle/service/set/set.cpp +++ b/src/core/hle/service/set/set.cpp @@ -3,6 +3,7 @@ // Refer to the license.txt file included. #include <algorithm> +#include <array> #include <chrono> #include "common/logging/log.h" #include "core/hle/ipc_helpers.h" @@ -31,6 +32,44 @@ constexpr std::array<LanguageCode, 17> available_language_codes = {{ LanguageCode::ZH_HANT, }}; +enum class KeyboardLayout : u64 { + Japanese = 0, + EnglishUs = 1, + EnglishUsInternational = 2, + EnglishUk = 3, + French = 4, + FrenchCa = 5, + Spanish = 6, + SpanishLatin = 7, + German = 8, + Italian = 9, + Portuguese = 10, + Russian = 11, + Korean = 12, + ChineseSimplified = 13, + ChineseTraditional = 14, +}; + +constexpr std::array<std::pair<LanguageCode, KeyboardLayout>, 17> language_to_layout{{ + {LanguageCode::JA, KeyboardLayout::Japanese}, + {LanguageCode::EN_US, KeyboardLayout::EnglishUs}, + {LanguageCode::FR, KeyboardLayout::French}, + {LanguageCode::DE, KeyboardLayout::German}, + {LanguageCode::IT, KeyboardLayout::Italian}, + {LanguageCode::ES, KeyboardLayout::Spanish}, + {LanguageCode::ZH_CN, KeyboardLayout::ChineseSimplified}, + {LanguageCode::KO, KeyboardLayout::Korean}, + {LanguageCode::NL, KeyboardLayout::EnglishUsInternational}, + {LanguageCode::PT, KeyboardLayout::Portuguese}, + {LanguageCode::RU, KeyboardLayout::Russian}, + {LanguageCode::ZH_TW, KeyboardLayout::ChineseTraditional}, + {LanguageCode::EN_GB, KeyboardLayout::EnglishUk}, + {LanguageCode::FR_CA, KeyboardLayout::FrenchCa}, + {LanguageCode::ES_419, KeyboardLayout::SpanishLatin}, + {LanguageCode::ZH_HANS, KeyboardLayout::ChineseSimplified}, + {LanguageCode::ZH_HANT, KeyboardLayout::ChineseTraditional}, +}}; + constexpr std::size_t pre4_0_0_max_entries = 15; constexpr std::size_t post4_0_0_max_entries = 17; @@ -50,6 +89,25 @@ void GetAvailableLanguageCodesImpl(Kernel::HLERequestContext& ctx, std::size_t m ctx.WriteBuffer(available_language_codes.data(), copy_size); PushResponseLanguageCode(ctx, copy_amount); } + +void GetKeyCodeMapImpl(Kernel::HLERequestContext& ctx) { + const auto language_code = available_language_codes[Settings::values.language_index]; + const auto key_code = + std::find_if(language_to_layout.cbegin(), language_to_layout.cend(), + [=](const auto& element) { return element.first == language_code; }); + KeyboardLayout layout = KeyboardLayout::EnglishUs; + if (key_code == language_to_layout.cend()) { + LOG_ERROR(Service_SET, + "Could not find keyboard layout for language index {}, defaulting to English us", + Settings::values.language_index); + } else { + layout = key_code->second; + } + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); + ctx.WriteBuffer(&layout, sizeof(KeyboardLayout)); +} } // Anonymous namespace LanguageCode GetLanguageCodeFromIndex(std::size_t index) { @@ -120,6 +178,16 @@ void SET::GetRegionCode(Kernel::HLERequestContext& ctx) { rb.Push(Settings::values.region_index); } +void SET::GetKeyCodeMap(Kernel::HLERequestContext& ctx) { + LOG_DEBUG(Service_SET, "Called {}", ctx.Description()); + GetKeyCodeMapImpl(ctx); +} + +void SET::GetKeyCodeMap2(Kernel::HLERequestContext& ctx) { + LOG_DEBUG(Service_SET, "Called {}", ctx.Description()); + GetKeyCodeMapImpl(ctx); +} + SET::SET() : ServiceFramework("set") { // clang-format off static const FunctionInfo functions[] = { @@ -130,9 +198,9 @@ SET::SET() : ServiceFramework("set") { {4, &SET::GetRegionCode, "GetRegionCode"}, {5, &SET::GetAvailableLanguageCodes2, "GetAvailableLanguageCodes2"}, {6, &SET::GetAvailableLanguageCodeCount2, "GetAvailableLanguageCodeCount2"}, - {7, nullptr, "GetKeyCodeMap"}, + {7, &SET::GetKeyCodeMap, "GetKeyCodeMap"}, {8, &SET::GetQuestFlag, "GetQuestFlag"}, - {9, nullptr, "GetKeyCodeMap2"}, + {9, &SET::GetKeyCodeMap2, "GetKeyCodeMap2"}, {10, nullptr, "GetFirmwareVersionForDebug"}, }; // clang-format on diff --git a/src/core/hle/service/set/set.h b/src/core/hle/service/set/set.h index 6084b345d..8ac9c169d 100644 --- a/src/core/hle/service/set/set.h +++ b/src/core/hle/service/set/set.h @@ -44,6 +44,8 @@ private: void GetAvailableLanguageCodeCount2(Kernel::HLERequestContext& ctx); void GetQuestFlag(Kernel::HLERequestContext& ctx); void GetRegionCode(Kernel::HLERequestContext& ctx); + void GetKeyCodeMap(Kernel::HLERequestContext& ctx); + void GetKeyCodeMap2(Kernel::HLERequestContext& ctx); }; } // namespace Service::Set diff --git a/src/core/memory.cpp b/src/core/memory.cpp index 7def00768..2c5588933 100644 --- a/src/core/memory.cpp +++ b/src/core/memory.cpp @@ -548,9 +548,9 @@ struct Memory::Impl { // longer exist, and we should just leave the pagetable entry blank. page_type = Common::PageType::Unmapped; } else { - page_type = Common::PageType::Memory; current_page_table->pointers[vaddr >> PAGE_BITS] = pointer - (vaddr & ~PAGE_MASK); + page_type = Common::PageType::Memory; } break; } @@ -591,9 +591,12 @@ struct Memory::Impl { base + page_table.pointers.size()); if (!target) { + ASSERT_MSG(type != Common::PageType::Memory, + "Mapping memory page without a pointer @ {:016x}", base * PAGE_SIZE); + while (base != end) { - page_table.pointers[base] = nullptr; page_table.attributes[base] = type; + page_table.pointers[base] = nullptr; page_table.backing_addr[base] = 0; base += 1; diff --git a/src/input_common/CMakeLists.txt b/src/input_common/CMakeLists.txt index a9c2392b1..3bd76dd23 100644 --- a/src/input_common/CMakeLists.txt +++ b/src/input_common/CMakeLists.txt @@ -7,6 +7,10 @@ add_library(input_common STATIC main.h motion_emu.cpp motion_emu.h + gcadapter/gc_adapter.cpp + gcadapter/gc_adapter.h + gcadapter/gc_poller.cpp + gcadapter/gc_poller.h sdl/sdl.cpp sdl/sdl.h udp/client.cpp @@ -26,5 +30,7 @@ if(SDL2_FOUND) target_compile_definitions(input_common PRIVATE HAVE_SDL2) endif() +target_link_libraries(input_common PUBLIC ${LIBUSB_LIBRARIES}) + create_target_directory_groups(input_common) target_link_libraries(input_common PUBLIC core PRIVATE common Boost::boost) diff --git a/src/input_common/gcadapter/gc_adapter.cpp b/src/input_common/gcadapter/gc_adapter.cpp new file mode 100644 index 000000000..6d9f4d9eb --- /dev/null +++ b/src/input_common/gcadapter/gc_adapter.cpp @@ -0,0 +1,398 @@ +// Copyright 2014 Dolphin Emulator Project +// Licensed under GPLv2+ +// Refer to the license.txt file included. + +#include <chrono> +#include <thread> +#include "common/logging/log.h" +#include "input_common/gcadapter/gc_adapter.h" + +namespace GCAdapter { + +/// Used to loop through and assign button in poller +constexpr std::array<PadButton, 12> PadButtonArray{ + PadButton::PAD_BUTTON_LEFT, PadButton::PAD_BUTTON_RIGHT, PadButton::PAD_BUTTON_DOWN, + PadButton::PAD_BUTTON_UP, PadButton::PAD_TRIGGER_Z, PadButton::PAD_TRIGGER_R, + PadButton::PAD_TRIGGER_L, PadButton::PAD_BUTTON_A, PadButton::PAD_BUTTON_B, + PadButton::PAD_BUTTON_X, PadButton::PAD_BUTTON_Y, PadButton::PAD_BUTTON_START, +}; + +Adapter::Adapter() { + if (usb_adapter_handle != nullptr) { + return; + } + LOG_INFO(Input, "GC Adapter Initialization started"); + + current_status = NO_ADAPTER_DETECTED; + + const int init_res = libusb_init(&libusb_ctx); + if (init_res == LIBUSB_SUCCESS) { + StartScanThread(); + } else { + LOG_ERROR(Input, "libusb could not be initialized. failed with error = {}", init_res); + } +} + +GCPadStatus Adapter::GetPadStatus(int port, const std::array<u8, 37>& adapter_payload) { + GCPadStatus pad = {}; + bool get_origin = false; + + ControllerTypes type = ControllerTypes(adapter_payload[1 + (9 * port)] >> 4); + if (type != ControllerTypes::None) { + get_origin = true; + } + + adapter_controllers_status[port] = type; + + static constexpr std::array<PadButton, 8> b1_buttons{ + PadButton::PAD_BUTTON_A, PadButton::PAD_BUTTON_B, PadButton::PAD_BUTTON_X, + PadButton::PAD_BUTTON_Y, PadButton::PAD_BUTTON_LEFT, PadButton::PAD_BUTTON_RIGHT, + PadButton::PAD_BUTTON_DOWN, PadButton::PAD_BUTTON_UP, + }; + + static constexpr std::array<PadButton, 4> b2_buttons{ + PadButton::PAD_BUTTON_START, + PadButton::PAD_TRIGGER_Z, + PadButton::PAD_TRIGGER_R, + PadButton::PAD_TRIGGER_L, + }; + + if (adapter_controllers_status[port] != ControllerTypes::None) { + const u8 b1 = adapter_payload[1 + (9 * port) + 1]; + const u8 b2 = adapter_payload[1 + (9 * port) + 2]; + + for (std::size_t i = 0; i < b1_buttons.size(); ++i) { + if ((b1 & (1U << i)) != 0) { + pad.button |= static_cast<u16>(b1_buttons[i]); + } + } + + for (std::size_t j = 0; j < b2_buttons.size(); ++j) { + if ((b2 & (1U << j)) != 0) { + pad.button |= static_cast<u16>(b2_buttons[j]); + } + } + + if (get_origin) { + pad.button |= PAD_GET_ORIGIN; + } + + pad.stick_x = adapter_payload[1 + (9 * port) + 3]; + pad.stick_y = adapter_payload[1 + (9 * port) + 4]; + pad.substick_x = adapter_payload[1 + (9 * port) + 5]; + pad.substick_y = adapter_payload[1 + (9 * port) + 6]; + pad.trigger_left = adapter_payload[1 + (9 * port) + 7]; + pad.trigger_right = adapter_payload[1 + (9 * port) + 8]; + } + return pad; +} + +void Adapter::PadToState(const GCPadStatus& pad, GCState& state) { + for (const auto& button : PadButtonArray) { + const u16 button_value = static_cast<u16>(button); + state.buttons.insert_or_assign(button_value, pad.button & button_value); + } + + state.axes.insert_or_assign(static_cast<u8>(PadAxes::StickX), pad.stick_x); + state.axes.insert_or_assign(static_cast<u8>(PadAxes::StickY), pad.stick_y); + state.axes.insert_or_assign(static_cast<u8>(PadAxes::SubstickX), pad.substick_x); + state.axes.insert_or_assign(static_cast<u8>(PadAxes::SubstickY), pad.substick_y); + state.axes.insert_or_assign(static_cast<u8>(PadAxes::TriggerLeft), pad.trigger_left); + state.axes.insert_or_assign(static_cast<u8>(PadAxes::TriggerRight), pad.trigger_right); +} + +void Adapter::Read() { + LOG_DEBUG(Input, "GC Adapter Read() thread started"); + + int payload_size_in, payload_size_copy; + std::array<u8, 37> adapter_payload; + std::array<u8, 37> adapter_payload_copy; + std::array<GCPadStatus, 4> pads; + + while (adapter_thread_running) { + libusb_interrupt_transfer(usb_adapter_handle, input_endpoint, adapter_payload.data(), + sizeof(adapter_payload), &payload_size_in, 16); + payload_size_copy = 0; + // this mutex might be redundant? + { + std::lock_guard<std::mutex> lk(s_mutex); + std::copy(std::begin(adapter_payload), std::end(adapter_payload), + std::begin(adapter_payload_copy)); + payload_size_copy = payload_size_in; + } + + if (payload_size_copy != sizeof(adapter_payload_copy) || + adapter_payload_copy[0] != LIBUSB_DT_HID) { + LOG_ERROR(Input, "error reading payload (size: {}, type: {:02x})", payload_size_copy, + adapter_payload_copy[0]); + adapter_thread_running = false; // error reading from adapter, stop reading. + break; + } + for (std::size_t port = 0; port < pads.size(); ++port) { + pads[port] = GetPadStatus(port, adapter_payload_copy); + if (DeviceConnected(port) && configuring) { + if (pads[port].button != PAD_GET_ORIGIN) { + pad_queue[port].Push(pads[port]); + } + + // Accounting for a threshold here because of some controller variance + if (pads[port].stick_x > pads[port].MAIN_STICK_CENTER_X + pads[port].THRESHOLD || + pads[port].stick_x < pads[port].MAIN_STICK_CENTER_X - pads[port].THRESHOLD) { + pads[port].axis = GCAdapter::PadAxes::StickX; + pads[port].axis_value = pads[port].stick_x; + pad_queue[port].Push(pads[port]); + } + if (pads[port].stick_y > pads[port].MAIN_STICK_CENTER_Y + pads[port].THRESHOLD || + pads[port].stick_y < pads[port].MAIN_STICK_CENTER_Y - pads[port].THRESHOLD) { + pads[port].axis = GCAdapter::PadAxes::StickY; + pads[port].axis_value = pads[port].stick_y; + pad_queue[port].Push(pads[port]); + } + if (pads[port].substick_x > pads[port].C_STICK_CENTER_X + pads[port].THRESHOLD || + pads[port].substick_x < pads[port].C_STICK_CENTER_X - pads[port].THRESHOLD) { + pads[port].axis = GCAdapter::PadAxes::SubstickX; + pads[port].axis_value = pads[port].substick_x; + pad_queue[port].Push(pads[port]); + } + if (pads[port].substick_y > pads[port].C_STICK_CENTER_Y + pads[port].THRESHOLD || + pads[port].substick_y < pads[port].C_STICK_CENTER_Y - pads[port].THRESHOLD) { + pads[port].axis = GCAdapter::PadAxes::SubstickY; + pads[port].axis_value = pads[port].substick_y; + pad_queue[port].Push(pads[port]); + } + if (pads[port].trigger_left > pads[port].TRIGGER_THRESHOLD) { + pads[port].axis = GCAdapter::PadAxes::TriggerLeft; + pads[port].axis_value = pads[port].trigger_left; + pad_queue[port].Push(pads[port]); + } + if (pads[port].trigger_right > pads[port].TRIGGER_THRESHOLD) { + pads[port].axis = GCAdapter::PadAxes::TriggerRight; + pads[port].axis_value = pads[port].trigger_right; + pad_queue[port].Push(pads[port]); + } + } + PadToState(pads[port], state[port]); + } + std::this_thread::yield(); + } +} + +void Adapter::ScanThreadFunc() { + LOG_INFO(Input, "GC Adapter scanning thread started"); + + while (detect_thread_running) { + if (usb_adapter_handle == nullptr) { + std::lock_guard<std::mutex> lk(initialization_mutex); + Setup(); + } + std::this_thread::sleep_for(std::chrono::milliseconds(500)); + } +} + +void Adapter::StartScanThread() { + if (detect_thread_running) { + return; + } + if (!libusb_ctx) { + return; + } + + detect_thread_running = true; + detect_thread = std::thread([=] { ScanThreadFunc(); }); +} + +void Adapter::StopScanThread() { + detect_thread_running = false; + detect_thread.join(); +} + +void Adapter::Setup() { + // Reset the error status in case the adapter gets unplugged + if (current_status < 0) { + current_status = NO_ADAPTER_DETECTED; + } + + adapter_controllers_status.fill(ControllerTypes::None); + + // pointer to list of connected usb devices + libusb_device** devices{}; + + // populate the list of devices, get the count + const ssize_t device_count = libusb_get_device_list(libusb_ctx, &devices); + if (device_count < 0) { + LOG_ERROR(Input, "libusb_get_device_list failed with error: {}", device_count); + detect_thread_running = false; // Stop the loop constantly checking for gc adapter + // TODO: For hotplug+gc adapter checkbox implementation, revert this. + return; + } + + if (devices != nullptr) { + for (std::size_t index = 0; index < device_count; ++index) { + if (CheckDeviceAccess(devices[index])) { + // GC Adapter found and accessible, registering it + GetGCEndpoint(devices[index]); + break; + } + } + libusb_free_device_list(devices, 1); + } +} + +bool Adapter::CheckDeviceAccess(libusb_device* device) { + libusb_device_descriptor desc; + const int get_descriptor_error = libusb_get_device_descriptor(device, &desc); + if (get_descriptor_error) { + // could not acquire the descriptor, no point in trying to use it. + LOG_ERROR(Input, "libusb_get_device_descriptor failed with error: {}", + get_descriptor_error); + return false; + } + + if (desc.idVendor != 0x057e || desc.idProduct != 0x0337) { + // This isn't the device we are looking for. + return false; + } + const int open_error = libusb_open(device, &usb_adapter_handle); + + if (open_error == LIBUSB_ERROR_ACCESS) { + LOG_ERROR(Input, "Yuzu can not gain access to this device: ID {:04X}:{:04X}.", + desc.idVendor, desc.idProduct); + return false; + } + if (open_error) { + LOG_ERROR(Input, "libusb_open failed to open device with error = {}", open_error); + return false; + } + + int kernel_driver_error = libusb_kernel_driver_active(usb_adapter_handle, 0); + if (kernel_driver_error == 1) { + kernel_driver_error = libusb_detach_kernel_driver(usb_adapter_handle, 0); + if (kernel_driver_error != 0 && kernel_driver_error != LIBUSB_ERROR_NOT_SUPPORTED) { + LOG_ERROR(Input, "libusb_detach_kernel_driver failed with error = {}", + kernel_driver_error); + } + } + + if (kernel_driver_error && kernel_driver_error != LIBUSB_ERROR_NOT_SUPPORTED) { + libusb_close(usb_adapter_handle); + usb_adapter_handle = nullptr; + return false; + } + + const int interface_claim_error = libusb_claim_interface(usb_adapter_handle, 0); + if (interface_claim_error) { + LOG_ERROR(Input, "libusb_claim_interface failed with error = {}", interface_claim_error); + libusb_close(usb_adapter_handle); + usb_adapter_handle = nullptr; + return false; + } + + return true; +} + +void Adapter::GetGCEndpoint(libusb_device* device) { + libusb_config_descriptor* config = nullptr; + const int config_descriptor_return = libusb_get_config_descriptor(device, 0, &config); + if (config_descriptor_return != LIBUSB_SUCCESS) { + LOG_ERROR(Input, "libusb_get_config_descriptor failed with error = {}", + config_descriptor_return); + return; + } + + for (u8 ic = 0; ic < config->bNumInterfaces; ic++) { + const libusb_interface* interfaceContainer = &config->interface[ic]; + for (int i = 0; i < interfaceContainer->num_altsetting; i++) { + const libusb_interface_descriptor* interface = &interfaceContainer->altsetting[i]; + for (u8 e = 0; e < interface->bNumEndpoints; e++) { + const libusb_endpoint_descriptor* endpoint = &interface->endpoint[e]; + if (endpoint->bEndpointAddress & LIBUSB_ENDPOINT_IN) { + input_endpoint = endpoint->bEndpointAddress; + } else { + output_endpoint = endpoint->bEndpointAddress; + } + } + } + } + // This transfer seems to be responsible for clearing the state of the adapter + // Used to clear the "busy" state of when the device is unexpectedly unplugged + unsigned char clear_payload = 0x13; + libusb_interrupt_transfer(usb_adapter_handle, output_endpoint, &clear_payload, + sizeof(clear_payload), nullptr, 16); + + adapter_thread_running = true; + current_status = ADAPTER_DETECTED; + adapter_input_thread = std::thread([=] { Read(); }); // Read input +} + +Adapter::~Adapter() { + StopScanThread(); + Reset(); +} + +void Adapter::Reset() { + std::unique_lock<std::mutex> lock(initialization_mutex, std::defer_lock); + if (!lock.try_lock()) { + return; + } + if (current_status != ADAPTER_DETECTED) { + return; + } + + if (adapter_thread_running) { + adapter_thread_running = false; + } + adapter_input_thread.join(); + + adapter_controllers_status.fill(ControllerTypes::None); + current_status = NO_ADAPTER_DETECTED; + + if (usb_adapter_handle) { + libusb_release_interface(usb_adapter_handle, 1); + libusb_close(usb_adapter_handle); + usb_adapter_handle = nullptr; + } + + if (libusb_ctx) { + libusb_exit(libusb_ctx); + } +} + +bool Adapter::DeviceConnected(int port) { + return adapter_controllers_status[port] != ControllerTypes::None; +} + +void Adapter::ResetDeviceType(int port) { + adapter_controllers_status[port] = ControllerTypes::None; +} + +void Adapter::BeginConfiguration() { + for (auto& pq : pad_queue) { + pq.Clear(); + } + configuring = true; +} + +void Adapter::EndConfiguration() { + for (auto& pq : pad_queue) { + pq.Clear(); + } + configuring = false; +} + +std::array<Common::SPSCQueue<GCPadStatus>, 4>& Adapter::GetPadQueue() { + return pad_queue; +} + +const std::array<Common::SPSCQueue<GCPadStatus>, 4>& Adapter::GetPadQueue() const { + return pad_queue; +} + +std::array<GCState, 4>& Adapter::GetPadState() { + return state; +} + +const std::array<GCState, 4>& Adapter::GetPadState() const { + return state; +} + +} // namespace GCAdapter diff --git a/src/input_common/gcadapter/gc_adapter.h b/src/input_common/gcadapter/gc_adapter.h new file mode 100644 index 000000000..b1c2a1958 --- /dev/null +++ b/src/input_common/gcadapter/gc_adapter.h @@ -0,0 +1,161 @@ +// Copyright 2014 Dolphin Emulator Project +// Licensed under GPLv2+ +// Refer to the license.txt file included. + +#pragma once +#include <algorithm> +#include <functional> +#include <mutex> +#include <thread> +#include <unordered_map> +#include <libusb.h> +#include "common/common_types.h" +#include "common/threadsafe_queue.h" + +namespace GCAdapter { + +enum { + PAD_USE_ORIGIN = 0x0080, + PAD_GET_ORIGIN = 0x2000, + PAD_ERR_STATUS = 0x8000, +}; + +enum class PadButton { + PAD_BUTTON_LEFT = 0x0001, + PAD_BUTTON_RIGHT = 0x0002, + PAD_BUTTON_DOWN = 0x0004, + PAD_BUTTON_UP = 0x0008, + PAD_TRIGGER_Z = 0x0010, + PAD_TRIGGER_R = 0x0020, + PAD_TRIGGER_L = 0x0040, + PAD_BUTTON_A = 0x0100, + PAD_BUTTON_B = 0x0200, + PAD_BUTTON_X = 0x0400, + PAD_BUTTON_Y = 0x0800, + PAD_BUTTON_START = 0x1000, + // Below is for compatibility with "AxisButton" type + PAD_STICK = 0x2000, +}; + +extern const std::array<PadButton, 12> PadButtonArray; + +enum class PadAxes : u8 { + StickX, + StickY, + SubstickX, + SubstickY, + TriggerLeft, + TriggerRight, + Undefined, +}; + +struct GCPadStatus { + u16 button{}; // Or-ed PAD_BUTTON_* and PAD_TRIGGER_* bits + u8 stick_x{}; // 0 <= stick_x <= 255 + u8 stick_y{}; // 0 <= stick_y <= 255 + u8 substick_x{}; // 0 <= substick_x <= 255 + u8 substick_y{}; // 0 <= substick_y <= 255 + u8 trigger_left{}; // 0 <= trigger_left <= 255 + u8 trigger_right{}; // 0 <= trigger_right <= 255 + + static constexpr u8 MAIN_STICK_CENTER_X = 0x80; + static constexpr u8 MAIN_STICK_CENTER_Y = 0x80; + static constexpr u8 MAIN_STICK_RADIUS = 0x7f; + static constexpr u8 C_STICK_CENTER_X = 0x80; + static constexpr u8 C_STICK_CENTER_Y = 0x80; + static constexpr u8 C_STICK_RADIUS = 0x7f; + static constexpr u8 THRESHOLD = 10; + + // 256/4, at least a quarter press to count as a press. For polling mostly + static constexpr u8 TRIGGER_THRESHOLD = 64; + + u8 port{}; + PadAxes axis{PadAxes::Undefined}; + u8 axis_value{255}; +}; + +struct GCState { + std::unordered_map<int, bool> buttons; + std::unordered_map<int, u16> axes; +}; + +enum class ControllerTypes { None, Wired, Wireless }; + +enum { + NO_ADAPTER_DETECTED = 0, + ADAPTER_DETECTED = 1, +}; + +class Adapter { +public: + /// Initialize the GC Adapter capture and read sequence + Adapter(); + + /// Close the adapter read thread and release the adapter + ~Adapter(); + /// Used for polling + void BeginConfiguration(); + void EndConfiguration(); + + std::array<Common::SPSCQueue<GCPadStatus>, 4>& GetPadQueue(); + const std::array<Common::SPSCQueue<GCPadStatus>, 4>& GetPadQueue() const; + + std::array<GCState, 4>& GetPadState(); + const std::array<GCState, 4>& GetPadState() const; + +private: + GCPadStatus GetPadStatus(int port, const std::array<u8, 37>& adapter_payload); + + void PadToState(const GCPadStatus& pad, GCState& state); + + void Read(); + void ScanThreadFunc(); + /// Begin scanning for the GC Adapter. + void StartScanThread(); + + /// Stop scanning for the adapter + void StopScanThread(); + + /// Returns true if there is a device connected to port + bool DeviceConnected(int port); + + /// Resets status of device connected to port + void ResetDeviceType(int port); + + /// Returns true if we successfully gain access to GC Adapter + bool CheckDeviceAccess(libusb_device* device); + + /// Captures GC Adapter endpoint address, + void GetGCEndpoint(libusb_device* device); + + /// For shutting down, clear all data, join all threads, release usb + void Reset(); + + /// For use in initialization, querying devices to find the adapter + void Setup(); + + int current_status = NO_ADAPTER_DETECTED; + libusb_device_handle* usb_adapter_handle = nullptr; + std::array<ControllerTypes, 4> adapter_controllers_status{}; + + std::mutex s_mutex; + + std::thread adapter_input_thread; + bool adapter_thread_running; + + std::mutex initialization_mutex; + std::thread detect_thread; + bool detect_thread_running = false; + + libusb_context* libusb_ctx; + + u8 input_endpoint = 0; + u8 output_endpoint = 0; + + bool configuring = false; + + std::array<Common::SPSCQueue<GCPadStatus>, 4> pad_queue; + std::array<GCState, 4> state; +}; + +} // namespace GCAdapter diff --git a/src/input_common/gcadapter/gc_poller.cpp b/src/input_common/gcadapter/gc_poller.cpp new file mode 100644 index 000000000..385ce8430 --- /dev/null +++ b/src/input_common/gcadapter/gc_poller.cpp @@ -0,0 +1,272 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <atomic> +#include <list> +#include <mutex> +#include <utility> +#include "common/threadsafe_queue.h" +#include "input_common/gcadapter/gc_adapter.h" +#include "input_common/gcadapter/gc_poller.h" + +namespace InputCommon { + +class GCButton final : public Input::ButtonDevice { +public: + explicit GCButton(int port_, int button_, GCAdapter::Adapter* adapter) + : port(port_), button(button_), gcadapter(adapter) {} + + ~GCButton() override; + + bool GetStatus() const override { + return gcadapter->GetPadState()[port].buttons.at(button); + } + +private: + const int port; + const int button; + GCAdapter::Adapter* gcadapter; +}; + +class GCAxisButton final : public Input::ButtonDevice { +public: + explicit GCAxisButton(int port_, int axis_, float threshold_, bool trigger_if_greater_, + GCAdapter::Adapter* adapter) + : port(port_), axis(axis_), threshold(threshold_), trigger_if_greater(trigger_if_greater_), + gcadapter(adapter) { + // L/R triggers range is only in positive direction beginning near 0 + // 0.0 threshold equates to near half trigger press, but threshold accounts for variability. + if (axis > 3) { + threshold *= -0.5; + } + } + + bool GetStatus() const override { + const float axis_value = (gcadapter->GetPadState()[port].axes.at(axis) - 128.0f) / 128.0f; + if (trigger_if_greater) { + // TODO: Might be worthwile to set a slider for the trigger threshold. It is currently + // always set to 0.5 in configure_input_player.cpp ZL/ZR HandleClick + return axis_value > threshold; + } + return axis_value < -threshold; + } + +private: + const int port; + const int axis; + float threshold; + bool trigger_if_greater; + GCAdapter::Adapter* gcadapter; +}; + +GCButtonFactory::GCButtonFactory(std::shared_ptr<GCAdapter::Adapter> adapter_) + : adapter(std::move(adapter_)) {} + +GCButton::~GCButton() = default; + +std::unique_ptr<Input::ButtonDevice> GCButtonFactory::Create(const Common::ParamPackage& params) { + const int button_id = params.Get("button", 0); + const int port = params.Get("port", 0); + + constexpr int PAD_STICK_ID = static_cast<u16>(GCAdapter::PadButton::PAD_STICK); + + // button is not an axis/stick button + if (button_id != PAD_STICK_ID) { + auto button = std::make_unique<GCButton>(port, button_id, adapter.get()); + return std::move(button); + } + + // For Axis buttons, used by the binary sticks. + if (button_id == PAD_STICK_ID) { + const int axis = params.Get("axis", 0); + const float threshold = params.Get("threshold", 0.25f); + const std::string direction_name = params.Get("direction", ""); + bool trigger_if_greater; + if (direction_name == "+") { + trigger_if_greater = true; + } else if (direction_name == "-") { + trigger_if_greater = false; + } else { + trigger_if_greater = true; + LOG_ERROR(Input, "Unknown direction {}", direction_name); + } + return std::make_unique<GCAxisButton>(port, axis, threshold, trigger_if_greater, + adapter.get()); + } +} + +Common::ParamPackage GCButtonFactory::GetNextInput() { + Common::ParamPackage params; + GCAdapter::GCPadStatus pad; + auto& queue = adapter->GetPadQueue(); + for (std::size_t port = 0; port < queue.size(); ++port) { + while (queue[port].Pop(pad)) { + // This while loop will break on the earliest detected button + params.Set("engine", "gcpad"); + params.Set("port", static_cast<int>(port)); + for (const auto& button : GCAdapter::PadButtonArray) { + const u16 button_value = static_cast<u16>(button); + if (pad.button & button_value) { + params.Set("button", button_value); + break; + } + } + + // For Axis button implementation + if (pad.axis != GCAdapter::PadAxes::Undefined) { + params.Set("axis", static_cast<u8>(pad.axis)); + params.Set("button", static_cast<u16>(GCAdapter::PadButton::PAD_STICK)); + if (pad.axis_value > 128) { + params.Set("direction", "+"); + params.Set("threshold", "0.25"); + } else { + params.Set("direction", "-"); + params.Set("threshold", "-0.25"); + } + break; + } + } + } + return params; +} + +void GCButtonFactory::BeginConfiguration() { + polling = true; + adapter->BeginConfiguration(); +} + +void GCButtonFactory::EndConfiguration() { + polling = false; + adapter->EndConfiguration(); +} + +class GCAnalog final : public Input::AnalogDevice { +public: + GCAnalog(int port_, int axis_x_, int axis_y_, float deadzone_, GCAdapter::Adapter* adapter) + : port(port_), axis_x(axis_x_), axis_y(axis_y_), deadzone(deadzone_), gcadapter(adapter) {} + + float GetAxis(int axis) const { + std::lock_guard lock{mutex}; + // division is not by a perfect 128 to account for some variance in center location + // e.g. my device idled at 131 in X, 120 in Y, and full range of motion was in range + // [20-230] + return (gcadapter->GetPadState()[port].axes.at(axis) - 128.0f) / 95.0f; + } + + std::pair<float, float> GetAnalog(int axis_x, int axis_y) const { + float x = GetAxis(axis_x); + float y = GetAxis(axis_y); + + // Make sure the coordinates are in the unit circle, + // otherwise normalize it. + float r = x * x + y * y; + if (r > 1.0f) { + r = std::sqrt(r); + x /= r; + y /= r; + } + + return {x, y}; + } + + std::tuple<float, float> GetStatus() const override { + const auto [x, y] = GetAnalog(axis_x, axis_y); + const float r = std::sqrt((x * x) + (y * y)); + if (r > deadzone) { + return {x / r * (r - deadzone) / (1 - deadzone), + y / r * (r - deadzone) / (1 - deadzone)}; + } + return {0.0f, 0.0f}; + } + + bool GetAnalogDirectionStatus(Input::AnalogDirection direction) const override { + const auto [x, y] = GetStatus(); + const float directional_deadzone = 0.4f; + switch (direction) { + case Input::AnalogDirection::RIGHT: + return x > directional_deadzone; + case Input::AnalogDirection::LEFT: + return x < -directional_deadzone; + case Input::AnalogDirection::UP: + return y > directional_deadzone; + case Input::AnalogDirection::DOWN: + return y < -directional_deadzone; + } + return false; + } + +private: + const int port; + const int axis_x; + const int axis_y; + const float deadzone; + mutable std::mutex mutex; + GCAdapter::Adapter* gcadapter; +}; + +/// An analog device factory that creates analog devices from GC Adapter +GCAnalogFactory::GCAnalogFactory(std::shared_ptr<GCAdapter::Adapter> adapter_) + : adapter(std::move(adapter_)) {} + +/** + * Creates analog device from joystick axes + * @param params contains parameters for creating the device: + * - "port": the nth gcpad on the adapter + * - "axis_x": the index of the axis to be bind as x-axis + * - "axis_y": the index of the axis to be bind as y-axis + */ +std::unique_ptr<Input::AnalogDevice> GCAnalogFactory::Create(const Common::ParamPackage& params) { + const int port = params.Get("port", 0); + const int axis_x = params.Get("axis_x", 0); + const int axis_y = params.Get("axis_y", 1); + const float deadzone = std::clamp(params.Get("deadzone", 0.0f), 0.0f, .99f); + + return std::make_unique<GCAnalog>(port, axis_x, axis_y, deadzone, adapter.get()); +} + +void GCAnalogFactory::BeginConfiguration() { + polling = true; + adapter->BeginConfiguration(); +} + +void GCAnalogFactory::EndConfiguration() { + polling = false; + adapter->EndConfiguration(); +} + +Common::ParamPackage GCAnalogFactory::GetNextInput() { + GCAdapter::GCPadStatus pad; + auto& queue = adapter->GetPadQueue(); + for (std::size_t port = 0; port < queue.size(); ++port) { + while (queue[port].Pop(pad)) { + if (pad.axis == GCAdapter::PadAxes::Undefined || + std::abs((pad.axis_value - 128.0f) / 128.0f) < 0.1) { + continue; + } + // An analog device needs two axes, so we need to store the axis for later and wait for + // a second input event. The axes also must be from the same joystick. + const u8 axis = static_cast<u8>(pad.axis); + if (analog_x_axis == -1) { + analog_x_axis = axis; + controller_number = port; + } else if (analog_y_axis == -1 && analog_x_axis != axis && controller_number == port) { + analog_y_axis = axis; + } + } + } + Common::ParamPackage params; + if (analog_x_axis != -1 && analog_y_axis != -1) { + params.Set("engine", "gcpad"); + params.Set("port", controller_number); + params.Set("axis_x", analog_x_axis); + params.Set("axis_y", analog_y_axis); + analog_x_axis = -1; + analog_y_axis = -1; + controller_number = -1; + return params; + } + return params; +} + +} // namespace InputCommon diff --git a/src/input_common/gcadapter/gc_poller.h b/src/input_common/gcadapter/gc_poller.h new file mode 100644 index 000000000..e96af7d51 --- /dev/null +++ b/src/input_common/gcadapter/gc_poller.h @@ -0,0 +1,67 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <memory> +#include "core/frontend/input.h" +#include "input_common/gcadapter/gc_adapter.h" + +namespace InputCommon { + +/** + * A button device factory representing a gcpad. It receives gcpad events and forward them + * to all button devices it created. + */ +class GCButtonFactory final : public Input::Factory<Input::ButtonDevice> { +public: + explicit GCButtonFactory(std::shared_ptr<GCAdapter::Adapter> adapter_); + + /** + * Creates a button device from a button press + * @param params contains parameters for creating the device: + * - "code": the code of the key to bind with the button + */ + std::unique_ptr<Input::ButtonDevice> Create(const Common::ParamPackage& params) override; + + Common::ParamPackage GetNextInput(); + + /// For device input configuration/polling + void BeginConfiguration(); + void EndConfiguration(); + + bool IsPolling() const { + return polling; + } + +private: + std::shared_ptr<GCAdapter::Adapter> adapter; + bool polling = false; +}; + +/// An analog device factory that creates analog devices from GC Adapter +class GCAnalogFactory final : public Input::Factory<Input::AnalogDevice> { +public: + explicit GCAnalogFactory(std::shared_ptr<GCAdapter::Adapter> adapter_); + + std::unique_ptr<Input::AnalogDevice> Create(const Common::ParamPackage& params) override; + Common::ParamPackage GetNextInput(); + + /// For device input configuration/polling + void BeginConfiguration(); + void EndConfiguration(); + + bool IsPolling() const { + return polling; + } + +private: + std::shared_ptr<GCAdapter::Adapter> adapter; + int analog_x_axis = -1; + int analog_y_axis = -1; + int controller_number = -1; + bool polling = false; +}; + +} // namespace InputCommon diff --git a/src/input_common/main.cpp b/src/input_common/main.cpp index 95e351e24..fd0af1019 100644 --- a/src/input_common/main.cpp +++ b/src/input_common/main.cpp @@ -4,8 +4,11 @@ #include <memory> #include <thread> +#include <libusb.h> #include "common/param_package.h" #include "input_common/analog_from_button.h" +#include "input_common/gcadapter/gc_adapter.h" +#include "input_common/gcadapter/gc_poller.h" #include "input_common/keyboard.h" #include "input_common/main.h" #include "input_common/motion_emu.h" @@ -22,8 +25,16 @@ static std::shared_ptr<MotionEmu> motion_emu; static std::unique_ptr<SDL::State> sdl; #endif static std::unique_ptr<CemuhookUDP::State> udp; +static std::shared_ptr<GCButtonFactory> gcbuttons; +static std::shared_ptr<GCAnalogFactory> gcanalog; void Init() { + auto gcadapter = std::make_shared<GCAdapter::Adapter>(); + gcbuttons = std::make_shared<GCButtonFactory>(gcadapter); + Input::RegisterFactory<Input::ButtonDevice>("gcpad", gcbuttons); + gcanalog = std::make_shared<GCAnalogFactory>(gcadapter); + Input::RegisterFactory<Input::AnalogDevice>("gcpad", gcanalog); + keyboard = std::make_shared<Keyboard>(); Input::RegisterFactory<Input::ButtonDevice>("keyboard", keyboard); Input::RegisterFactory<Input::AnalogDevice>("analog_from_button", @@ -48,6 +59,11 @@ void Shutdown() { sdl.reset(); #endif udp.reset(); + Input::UnregisterFactory<Input::ButtonDevice>("gcpad"); + Input::UnregisterFactory<Input::AnalogDevice>("gcpad"); + + gcbuttons.reset(); + gcanalog.reset(); } Keyboard* GetKeyboard() { @@ -58,6 +74,14 @@ MotionEmu* GetMotionEmu() { return motion_emu.get(); } +GCButtonFactory* GetGCButtons() { + return gcbuttons.get(); +} + +GCAnalogFactory* GetGCAnalogs() { + return gcanalog.get(); +} + std::string GenerateKeyboardParam(int key_code) { Common::ParamPackage param{ {"engine", "keyboard"}, diff --git a/src/input_common/main.h b/src/input_common/main.h index 77a0ce90b..0e32856f6 100644 --- a/src/input_common/main.h +++ b/src/input_common/main.h @@ -7,6 +7,7 @@ #include <memory> #include <string> #include <vector> +#include "input_common/gcadapter/gc_poller.h" namespace Common { class ParamPackage; @@ -30,6 +31,10 @@ class MotionEmu; /// Gets the motion emulation factory. MotionEmu* GetMotionEmu(); +GCButtonFactory* GetGCButtons(); + +GCAnalogFactory* GetGCAnalogs(); + /// Generates a serialized param package for creating a keyboard button device std::string GenerateKeyboardParam(int key_code); diff --git a/src/video_core/buffer_cache/buffer_cache.h b/src/video_core/buffer_cache/buffer_cache.h index cf8bdd021..c6479af9f 100644 --- a/src/video_core/buffer_cache/buffer_cache.h +++ b/src/video_core/buffer_cache/buffer_cache.h @@ -322,8 +322,7 @@ protected: } private: - MapInterval* MapAddress(const Buffer* block, GPUVAddr gpu_addr, VAddr cpu_addr, - std::size_t size) { + MapInterval* MapAddress(Buffer* block, GPUVAddr gpu_addr, VAddr cpu_addr, std::size_t size) { const VectorMapInterval overlaps = GetMapsInRange(cpu_addr, size); if (overlaps.empty()) { auto& memory_manager = system.GPU().MemoryManager(); @@ -377,8 +376,7 @@ private: return map; } - void UpdateBlock(const Buffer* block, VAddr start, VAddr end, - const VectorMapInterval& overlaps) { + void UpdateBlock(Buffer* block, VAddr start, VAddr end, const VectorMapInterval& overlaps) { const IntervalType base_interval{start, end}; IntervalSet interval_set{}; interval_set.add(base_interval); diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.cpp b/src/video_core/renderer_opengl/gl_buffer_cache.cpp index d9f7b4cc6..e461e4c70 100644 --- a/src/video_core/renderer_opengl/gl_buffer_cache.cpp +++ b/src/video_core/renderer_opengl/gl_buffer_cache.cpp @@ -34,20 +34,27 @@ Buffer::Buffer(const Device& device, VAddr cpu_addr, std::size_t size) Buffer::~Buffer() = default; -void Buffer::Upload(std::size_t offset, std::size_t size, const u8* data) const { +void Buffer::Upload(std::size_t offset, std::size_t size, const u8* data) { glNamedBufferSubData(Handle(), static_cast<GLintptr>(offset), static_cast<GLsizeiptr>(size), data); } -void Buffer::Download(std::size_t offset, std::size_t size, u8* data) const { +void Buffer::Download(std::size_t offset, std::size_t size, u8* data) { MICROPROFILE_SCOPE(OpenGL_Buffer_Download); + const GLsizeiptr gl_size = static_cast<GLsizeiptr>(size); + const GLintptr gl_offset = static_cast<GLintptr>(offset); + if (read_buffer.handle == 0) { + read_buffer.Create(); + glNamedBufferData(read_buffer.handle, static_cast<GLsizeiptr>(Size()), nullptr, + GL_STREAM_READ); + } glMemoryBarrier(GL_BUFFER_UPDATE_BARRIER_BIT); - glGetNamedBufferSubData(Handle(), static_cast<GLintptr>(offset), static_cast<GLsizeiptr>(size), - data); + glCopyNamedBufferSubData(gl_buffer.handle, read_buffer.handle, gl_offset, gl_offset, gl_size); + glGetNamedBufferSubData(read_buffer.handle, gl_offset, gl_size, data); } void Buffer::CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst_offset, - std::size_t size) const { + std::size_t size) { glCopyNamedBufferSubData(src.Handle(), Handle(), static_cast<GLintptr>(src_offset), static_cast<GLintptr>(dst_offset), static_cast<GLsizeiptr>(size)); } diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.h b/src/video_core/renderer_opengl/gl_buffer_cache.h index 59d95adbc..88fdc0536 100644 --- a/src/video_core/renderer_opengl/gl_buffer_cache.h +++ b/src/video_core/renderer_opengl/gl_buffer_cache.h @@ -28,12 +28,12 @@ public: explicit Buffer(const Device& device, VAddr cpu_addr, std::size_t size); ~Buffer(); - void Upload(std::size_t offset, std::size_t size, const u8* data) const; + void Upload(std::size_t offset, std::size_t size, const u8* data); - void Download(std::size_t offset, std::size_t size, u8* data) const; + void Download(std::size_t offset, std::size_t size, u8* data); void CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst_offset, - std::size_t size) const; + std::size_t size); GLuint Handle() const noexcept { return gl_buffer.handle; @@ -45,6 +45,7 @@ public: private: OGLBuffer gl_buffer; + OGLBuffer read_buffer; u64 gpu_address = 0; }; diff --git a/src/video_core/renderer_opengl/maxwell_to_gl.h b/src/video_core/renderer_opengl/maxwell_to_gl.h index 774e70a5b..fe9bd4b5a 100644 --- a/src/video_core/renderer_opengl/maxwell_to_gl.h +++ b/src/video_core/renderer_opengl/maxwell_to_gl.h @@ -191,6 +191,12 @@ inline GLenum WrapMode(Tegra::Texture::WrapMode wrap_mode) { } else { return GL_MIRROR_CLAMP_TO_EDGE; } + case Tegra::Texture::WrapMode::MirrorOnceClampOGL: + if (GL_EXT_texture_mirror_clamp) { + return GL_MIRROR_CLAMP_EXT; + } else { + return GL_MIRROR_CLAMP_TO_EDGE; + } } UNIMPLEMENTED_MSG("Unimplemented texture wrap mode={}", static_cast<u32>(wrap_mode)); return GL_REPEAT; diff --git a/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp b/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp index 424278816..d1f0ea932 100644 --- a/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp +++ b/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp @@ -39,52 +39,18 @@ constexpr std::array POLYGON_OFFSET_ENABLE_LUT = { } // Anonymous namespace -void FixedPipelineState::DepthStencil::Fill(const Maxwell& regs) noexcept { - raw = 0; - front.action_stencil_fail.Assign(PackStencilOp(regs.stencil_front_op_fail)); - front.action_depth_fail.Assign(PackStencilOp(regs.stencil_front_op_zfail)); - front.action_depth_pass.Assign(PackStencilOp(regs.stencil_front_op_zpass)); - front.test_func.Assign(PackComparisonOp(regs.stencil_front_func_func)); - if (regs.stencil_two_side_enable) { - back.action_stencil_fail.Assign(PackStencilOp(regs.stencil_back_op_fail)); - back.action_depth_fail.Assign(PackStencilOp(regs.stencil_back_op_zfail)); - back.action_depth_pass.Assign(PackStencilOp(regs.stencil_back_op_zpass)); - back.test_func.Assign(PackComparisonOp(regs.stencil_back_func_func)); - } else { - back.action_stencil_fail.Assign(front.action_stencil_fail); - back.action_depth_fail.Assign(front.action_depth_fail); - back.action_depth_pass.Assign(front.action_depth_pass); - back.test_func.Assign(front.test_func); - } - depth_test_enable.Assign(regs.depth_test_enable); - depth_write_enable.Assign(regs.depth_write_enabled); - depth_bounds_enable.Assign(regs.depth_bounds_enable); - stencil_enable.Assign(regs.stencil_enable); - depth_test_func.Assign(PackComparisonOp(regs.depth_test_func)); -} - -void FixedPipelineState::Rasterizer::Fill(const Maxwell& regs) noexcept { +void FixedPipelineState::Fill(const Maxwell& regs, bool has_extended_dynamic_state) { const auto& clip = regs.view_volume_clip_control; const std::array enabled_lut = {regs.polygon_offset_point_enable, regs.polygon_offset_line_enable, regs.polygon_offset_fill_enable}; const u32 topology_index = static_cast<u32>(regs.draw.topology.Value()); - u32 packed_front_face = PackFrontFace(regs.front_face); - if (regs.screen_y_control.triangle_rast_flip != 0) { - // Flip front face - packed_front_face = 1 - packed_front_face; - } - raw = 0; - topology.Assign(topology_index); primitive_restart_enable.Assign(regs.primitive_restart.enabled != 0 ? 1 : 0); - cull_enable.Assign(regs.cull_test_enabled != 0 ? 1 : 0); depth_bias_enable.Assign(enabled_lut[POLYGON_OFFSET_ENABLE_LUT[topology_index]] != 0 ? 1 : 0); depth_clamp_disabled.Assign(regs.view_volume_clip_control.depth_clamp_disabled.Value()); ndc_minus_one_to_one.Assign(regs.depth_mode == Maxwell::DepthMode::MinusOneToOne ? 1 : 0); - cull_face.Assign(PackCullFace(regs.cull_face)); - front_face.Assign(packed_front_face); polygon_mode.Assign(PackPolygonMode(regs.polygon_mode_front)); patch_control_points_minus_one.Assign(regs.patch_vertices - 1); tessellation_primitive.Assign(static_cast<u32>(regs.tess_mode.prim.Value())); @@ -93,19 +59,37 @@ void FixedPipelineState::Rasterizer::Fill(const Maxwell& regs) noexcept { logic_op_enable.Assign(regs.logic_op.enable != 0 ? 1 : 0); logic_op.Assign(PackLogicOp(regs.logic_op.operation)); rasterize_enable.Assign(regs.rasterize_enable != 0 ? 1 : 0); + std::memcpy(&point_size, ®s.point_size, sizeof(point_size)); // TODO: C++20 std::bit_cast -} -void FixedPipelineState::ColorBlending::Fill(const Maxwell& regs) noexcept { + for (std::size_t index = 0; index < Maxwell::NumVertexArrays; ++index) { + binding_divisors[index] = + regs.instanced_arrays.IsInstancingEnabled(index) ? regs.vertex_array[index].divisor : 0; + } + + for (std::size_t index = 0; index < Maxwell::NumVertexAttributes; ++index) { + const auto& input = regs.vertex_attrib_format[index]; + auto& attribute = attributes[index]; + attribute.raw = 0; + attribute.enabled.Assign(input.IsConstant() ? 0 : 1); + attribute.buffer.Assign(input.buffer); + attribute.offset.Assign(input.offset); + attribute.type.Assign(static_cast<u32>(input.type.Value())); + attribute.size.Assign(static_cast<u32>(input.size.Value())); + } + for (std::size_t index = 0; index < std::size(attachments); ++index) { attachments[index].Fill(regs, index); } -} -void FixedPipelineState::ViewportSwizzles::Fill(const Maxwell& regs) noexcept { const auto& transform = regs.viewport_transform; - std::transform(transform.begin(), transform.end(), swizzles.begin(), + std::transform(transform.begin(), transform.end(), viewport_swizzles.begin(), [](const auto& viewport) { return static_cast<u16>(viewport.swizzle.raw); }); + + if (!has_extended_dynamic_state) { + no_extended_dynamic_state.Assign(1); + dynamic_state.Fill(regs); + } } void FixedPipelineState::BlendingAttachment::Fill(const Maxwell& regs, std::size_t index) { @@ -147,20 +131,57 @@ void FixedPipelineState::BlendingAttachment::Fill(const Maxwell& regs, std::size enable.Assign(1); } -void FixedPipelineState::Fill(const Maxwell& regs) { - rasterizer.Fill(regs); - depth_stencil.Fill(regs); - color_blending.Fill(regs); - viewport_swizzles.Fill(regs); +void FixedPipelineState::DynamicState::Fill(const Maxwell& regs) { + const u32 topology_index = static_cast<u32>(regs.draw.topology.Value()); + u32 packed_front_face = PackFrontFace(regs.front_face); + if (regs.screen_y_control.triangle_rast_flip != 0) { + // Flip front face + packed_front_face = 1 - packed_front_face; + } + + raw1 = 0; + raw2 = 0; + front.action_stencil_fail.Assign(PackStencilOp(regs.stencil_front_op_fail)); + front.action_depth_fail.Assign(PackStencilOp(regs.stencil_front_op_zfail)); + front.action_depth_pass.Assign(PackStencilOp(regs.stencil_front_op_zpass)); + front.test_func.Assign(PackComparisonOp(regs.stencil_front_func_func)); + if (regs.stencil_two_side_enable) { + back.action_stencil_fail.Assign(PackStencilOp(regs.stencil_back_op_fail)); + back.action_depth_fail.Assign(PackStencilOp(regs.stencil_back_op_zfail)); + back.action_depth_pass.Assign(PackStencilOp(regs.stencil_back_op_zpass)); + back.test_func.Assign(PackComparisonOp(regs.stencil_back_func_func)); + } else { + back.action_stencil_fail.Assign(front.action_stencil_fail); + back.action_depth_fail.Assign(front.action_depth_fail); + back.action_depth_pass.Assign(front.action_depth_pass); + back.test_func.Assign(front.test_func); + } + stencil_enable.Assign(regs.stencil_enable); + depth_write_enable.Assign(regs.depth_write_enabled); + depth_bounds_enable.Assign(regs.depth_bounds_enable); + depth_test_enable.Assign(regs.depth_test_enable); + front_face.Assign(packed_front_face); + depth_test_func.Assign(PackComparisonOp(regs.depth_test_func)); + topology.Assign(topology_index); + cull_face.Assign(PackCullFace(regs.cull_face)); + cull_enable.Assign(regs.cull_test_enabled != 0 ? 1 : 0); + + for (std::size_t index = 0; index < Maxwell::NumVertexArrays; ++index) { + const auto& input = regs.vertex_array[index]; + VertexBinding& binding = vertex_bindings[index]; + binding.raw = 0; + binding.enabled.Assign(input.IsEnabled() ? 1 : 0); + binding.stride.Assign(static_cast<u16>(input.stride.Value())); + } } std::size_t FixedPipelineState::Hash() const noexcept { - const u64 hash = Common::CityHash64(reinterpret_cast<const char*>(this), sizeof *this); + const u64 hash = Common::CityHash64(reinterpret_cast<const char*>(this), Size()); return static_cast<std::size_t>(hash); } bool FixedPipelineState::operator==(const FixedPipelineState& rhs) const noexcept { - return std::memcmp(this, &rhs, sizeof *this) == 0; + return std::memcmp(this, &rhs, Size()) == 0; } u32 FixedPipelineState::PackComparisonOp(Maxwell::ComparisonOp op) noexcept { diff --git a/src/video_core/renderer_vulkan/fixed_pipeline_state.h b/src/video_core/renderer_vulkan/fixed_pipeline_state.h index 31a6398f2..cdcbb65f5 100644 --- a/src/video_core/renderer_vulkan/fixed_pipeline_state.h +++ b/src/video_core/renderer_vulkan/fixed_pipeline_state.h @@ -60,14 +60,6 @@ struct FixedPipelineState { void Fill(const Maxwell& regs, std::size_t index); - std::size_t Hash() const noexcept; - - bool operator==(const BlendingAttachment& rhs) const noexcept; - - bool operator!=(const BlendingAttachment& rhs) const noexcept { - return !operator==(rhs); - } - constexpr std::array<bool, 4> Mask() const noexcept { return {mask_r != 0, mask_g != 0, mask_b != 0, mask_a != 0}; } @@ -97,156 +89,116 @@ struct FixedPipelineState { } }; - struct VertexInput { - union Binding { - u16 raw; - BitField<0, 1, u16> enabled; - BitField<1, 12, u16> stride; - }; + union VertexAttribute { + u32 raw; + BitField<0, 1, u32> enabled; + BitField<1, 5, u32> buffer; + BitField<6, 14, u32> offset; + BitField<20, 3, u32> type; + BitField<23, 6, u32> size; - union Attribute { - u32 raw; - BitField<0, 1, u32> enabled; - BitField<1, 5, u32> buffer; - BitField<6, 14, u32> offset; - BitField<20, 3, u32> type; - BitField<23, 6, u32> size; - - constexpr Maxwell::VertexAttribute::Type Type() const noexcept { - return static_cast<Maxwell::VertexAttribute::Type>(type.Value()); - } - - constexpr Maxwell::VertexAttribute::Size Size() const noexcept { - return static_cast<Maxwell::VertexAttribute::Size>(size.Value()); - } - }; - - std::array<Binding, Maxwell::NumVertexArrays> bindings; - std::array<u32, Maxwell::NumVertexArrays> binding_divisors; - std::array<Attribute, Maxwell::NumVertexAttributes> attributes; - - void SetBinding(std::size_t index, bool enabled, u32 stride, u32 divisor) noexcept { - auto& binding = bindings[index]; - binding.raw = 0; - binding.enabled.Assign(enabled ? 1 : 0); - binding.stride.Assign(static_cast<u16>(stride)); - binding_divisors[index] = divisor; + constexpr Maxwell::VertexAttribute::Type Type() const noexcept { + return static_cast<Maxwell::VertexAttribute::Type>(type.Value()); } - void SetAttribute(std::size_t index, bool enabled, u32 buffer, u32 offset, - Maxwell::VertexAttribute::Type type, - Maxwell::VertexAttribute::Size size) noexcept { - auto& attribute = attributes[index]; - attribute.raw = 0; - attribute.enabled.Assign(enabled ? 1 : 0); - attribute.buffer.Assign(buffer); - attribute.offset.Assign(offset); - attribute.type.Assign(static_cast<u32>(type)); - attribute.size.Assign(static_cast<u32>(size)); + constexpr Maxwell::VertexAttribute::Size Size() const noexcept { + return static_cast<Maxwell::VertexAttribute::Size>(size.Value()); } }; - struct Rasterizer { - union { - u32 raw; - BitField<0, 4, u32> topology; - BitField<4, 1, u32> primitive_restart_enable; - BitField<5, 1, u32> cull_enable; - BitField<6, 1, u32> depth_bias_enable; - BitField<7, 1, u32> depth_clamp_disabled; - BitField<8, 1, u32> ndc_minus_one_to_one; - BitField<9, 2, u32> cull_face; - BitField<11, 1, u32> front_face; - BitField<12, 2, u32> polygon_mode; - BitField<14, 5, u32> patch_control_points_minus_one; - BitField<19, 2, u32> tessellation_primitive; - BitField<21, 2, u32> tessellation_spacing; - BitField<23, 1, u32> tessellation_clockwise; - BitField<24, 1, u32> logic_op_enable; - BitField<25, 4, u32> logic_op; - BitField<29, 1, u32> rasterize_enable; - }; - - // TODO(Rodrigo): Move this to push constants - u32 point_size; + template <std::size_t Position> + union StencilFace { + BitField<Position + 0, 3, u32> action_stencil_fail; + BitField<Position + 3, 3, u32> action_depth_fail; + BitField<Position + 6, 3, u32> action_depth_pass; + BitField<Position + 9, 3, u32> test_func; - void Fill(const Maxwell& regs) noexcept; + Maxwell::StencilOp ActionStencilFail() const noexcept { + return UnpackStencilOp(action_stencil_fail); + } - constexpr Maxwell::PrimitiveTopology Topology() const noexcept { - return static_cast<Maxwell::PrimitiveTopology>(topology.Value()); + Maxwell::StencilOp ActionDepthFail() const noexcept { + return UnpackStencilOp(action_depth_fail); } - Maxwell::CullFace CullFace() const noexcept { - return UnpackCullFace(cull_face.Value()); + Maxwell::StencilOp ActionDepthPass() const noexcept { + return UnpackStencilOp(action_depth_pass); } - Maxwell::FrontFace FrontFace() const noexcept { - return UnpackFrontFace(front_face.Value()); + Maxwell::ComparisonOp TestFunc() const noexcept { + return UnpackComparisonOp(test_func); } }; - struct DepthStencil { - template <std::size_t Position> - union StencilFace { - BitField<Position + 0, 3, u32> action_stencil_fail; - BitField<Position + 3, 3, u32> action_depth_fail; - BitField<Position + 6, 3, u32> action_depth_pass; - BitField<Position + 9, 3, u32> test_func; - - Maxwell::StencilOp ActionStencilFail() const noexcept { - return UnpackStencilOp(action_stencil_fail); - } - - Maxwell::StencilOp ActionDepthFail() const noexcept { - return UnpackStencilOp(action_depth_fail); - } - - Maxwell::StencilOp ActionDepthPass() const noexcept { - return UnpackStencilOp(action_depth_pass); - } - - Maxwell::ComparisonOp TestFunc() const noexcept { - return UnpackComparisonOp(test_func); - } - }; + union VertexBinding { + u16 raw; + BitField<0, 12, u16> stride; + BitField<12, 1, u16> enabled; + }; + struct DynamicState { union { - u32 raw; + u32 raw1; StencilFace<0> front; StencilFace<12> back; - BitField<24, 1, u32> depth_test_enable; + BitField<24, 1, u32> stencil_enable; BitField<25, 1, u32> depth_write_enable; BitField<26, 1, u32> depth_bounds_enable; - BitField<27, 1, u32> stencil_enable; - BitField<28, 3, u32> depth_test_func; + BitField<27, 1, u32> depth_test_enable; + BitField<28, 1, u32> front_face; + BitField<29, 3, u32> depth_test_func; + }; + union { + u32 raw2; + BitField<0, 4, u32> topology; + BitField<4, 2, u32> cull_face; + BitField<6, 1, u32> cull_enable; }; + std::array<VertexBinding, Maxwell::NumVertexArrays> vertex_bindings; - void Fill(const Maxwell& regs) noexcept; + void Fill(const Maxwell& regs); Maxwell::ComparisonOp DepthTestFunc() const noexcept { return UnpackComparisonOp(depth_test_func); } - }; - - struct ColorBlending { - std::array<BlendingAttachment, Maxwell::NumRenderTargets> attachments; - void Fill(const Maxwell& regs) noexcept; - }; + Maxwell::CullFace CullFace() const noexcept { + return UnpackCullFace(cull_face.Value()); + } - struct ViewportSwizzles { - std::array<u16, Maxwell::NumViewports> swizzles; + Maxwell::FrontFace FrontFace() const noexcept { + return UnpackFrontFace(front_face.Value()); + } - void Fill(const Maxwell& regs) noexcept; + constexpr Maxwell::PrimitiveTopology Topology() const noexcept { + return static_cast<Maxwell::PrimitiveTopology>(topology.Value()); + } }; - VertexInput vertex_input; - Rasterizer rasterizer; - DepthStencil depth_stencil; - ColorBlending color_blending; - ViewportSwizzles viewport_swizzles; + union { + u32 raw; + BitField<0, 1, u32> no_extended_dynamic_state; + BitField<2, 1, u32> primitive_restart_enable; + BitField<3, 1, u32> depth_bias_enable; + BitField<4, 1, u32> depth_clamp_disabled; + BitField<5, 1, u32> ndc_minus_one_to_one; + BitField<6, 2, u32> polygon_mode; + BitField<8, 5, u32> patch_control_points_minus_one; + BitField<13, 2, u32> tessellation_primitive; + BitField<15, 2, u32> tessellation_spacing; + BitField<17, 1, u32> tessellation_clockwise; + BitField<18, 1, u32> logic_op_enable; + BitField<19, 4, u32> logic_op; + BitField<23, 1, u32> rasterize_enable; + }; + u32 point_size; + std::array<u32, Maxwell::NumVertexArrays> binding_divisors; + std::array<VertexAttribute, Maxwell::NumVertexAttributes> attributes; + std::array<BlendingAttachment, Maxwell::NumRenderTargets> attachments; + std::array<u16, Maxwell::NumViewports> viewport_swizzles; + DynamicState dynamic_state; - void Fill(const Maxwell& regs); + void Fill(const Maxwell& regs, bool has_extended_dynamic_state); std::size_t Hash() const noexcept; @@ -255,6 +207,11 @@ struct FixedPipelineState { bool operator!=(const FixedPipelineState& rhs) const noexcept { return !operator==(rhs); } + + std::size_t Size() const noexcept { + const std::size_t total_size = sizeof *this; + return total_size - (no_extended_dynamic_state != 0 ? 0 : sizeof(DynamicState)); + } }; static_assert(std::has_unique_object_representations_v<FixedPipelineState>); static_assert(std::is_trivially_copyable_v<FixedPipelineState>); diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp index f10f96cd8..2be38d419 100644 --- a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp @@ -56,7 +56,7 @@ Buffer::Buffer(const VKDevice& device, VKMemoryManager& memory_manager, VKSchedu Buffer::~Buffer() = default; -void Buffer::Upload(std::size_t offset, std::size_t size, const u8* data) const { +void Buffer::Upload(std::size_t offset, std::size_t size, const u8* data) { const auto& staging = staging_pool.GetUnusedBuffer(size, true); std::memcpy(staging.commit->Map(size), data, size); @@ -81,7 +81,7 @@ void Buffer::Upload(std::size_t offset, std::size_t size, const u8* data) const }); } -void Buffer::Download(std::size_t offset, std::size_t size, u8* data) const { +void Buffer::Download(std::size_t offset, std::size_t size, u8* data) { const auto& staging = staging_pool.GetUnusedBuffer(size, true); scheduler.RequestOutsideRenderPassOperationContext(); @@ -110,7 +110,7 @@ void Buffer::Download(std::size_t offset, std::size_t size, u8* data) const { } void Buffer::CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst_offset, - std::size_t size) const { + std::size_t size) { scheduler.RequestOutsideRenderPassOperationContext(); const VkBuffer dst_buffer = Handle(); diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.h b/src/video_core/renderer_vulkan/vk_buffer_cache.h index 3630aca77..991ee451c 100644 --- a/src/video_core/renderer_vulkan/vk_buffer_cache.h +++ b/src/video_core/renderer_vulkan/vk_buffer_cache.h @@ -29,12 +29,12 @@ public: VKStagingBufferPool& staging_pool, VAddr cpu_addr, std::size_t size); ~Buffer(); - void Upload(std::size_t offset, std::size_t size, const u8* data) const; + void Upload(std::size_t offset, std::size_t size, const u8* data); - void Download(std::size_t offset, std::size_t size, u8* data) const; + void Download(std::size_t offset, std::size_t size, u8* data); void CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst_offset, - std::size_t size) const; + std::size_t size); VkBuffer Handle() const { return *buffer.handle; diff --git a/src/video_core/renderer_vulkan/vk_device.cpp b/src/video_core/renderer_vulkan/vk_device.cpp index 9fd8ac3f6..fdaea4210 100644 --- a/src/video_core/renderer_vulkan/vk_device.cpp +++ b/src/video_core/renderer_vulkan/vk_device.cpp @@ -313,6 +313,16 @@ bool VKDevice::Create() { LOG_INFO(Render_Vulkan, "Device doesn't support custom border colors"); } + VkPhysicalDeviceExtendedDynamicStateFeaturesEXT dynamic_state; + if (ext_extended_dynamic_state) { + dynamic_state.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_EXTENDED_DYNAMIC_STATE_FEATURES_EXT; + dynamic_state.pNext = nullptr; + dynamic_state.extendedDynamicState = VK_TRUE; + SetNext(next, dynamic_state); + } else { + LOG_INFO(Render_Vulkan, "Device doesn't support extended dynamic state"); + } + if (!ext_depth_range_unrestricted) { LOG_INFO(Render_Vulkan, "Device doesn't support depth range unrestricted"); } @@ -541,6 +551,7 @@ std::vector<const char*> VKDevice::LoadExtensions() { bool has_ext_subgroup_size_control{}; bool has_ext_transform_feedback{}; bool has_ext_custom_border_color{}; + bool has_ext_extended_dynamic_state{}; for (const auto& extension : physical.EnumerateDeviceExtensionProperties()) { Test(extension, nv_viewport_swizzle, VK_NV_VIEWPORT_SWIZZLE_EXTENSION_NAME, true); Test(extension, khr_uniform_buffer_standard_layout, @@ -558,6 +569,8 @@ std::vector<const char*> VKDevice::LoadExtensions() { false); Test(extension, has_ext_custom_border_color, VK_EXT_CUSTOM_BORDER_COLOR_EXTENSION_NAME, false); + Test(extension, has_ext_extended_dynamic_state, + VK_EXT_EXTENDED_DYNAMIC_STATE_EXTENSION_NAME, false); if (Settings::values.renderer_debug) { Test(extension, nv_device_diagnostics_config, VK_NV_DEVICE_DIAGNOSTICS_CONFIG_EXTENSION_NAME, true); @@ -643,6 +656,19 @@ std::vector<const char*> VKDevice::LoadExtensions() { } } + if (has_ext_extended_dynamic_state) { + VkPhysicalDeviceExtendedDynamicStateFeaturesEXT dynamic_state; + dynamic_state.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_EXTENDED_DYNAMIC_STATE_FEATURES_EXT; + dynamic_state.pNext = nullptr; + features.pNext = &dynamic_state; + physical.GetFeatures2KHR(features); + + if (dynamic_state.extendedDynamicState) { + extensions.push_back(VK_EXT_EXTENDED_DYNAMIC_STATE_EXTENSION_NAME); + ext_extended_dynamic_state = true; + } + } + return extensions; } diff --git a/src/video_core/renderer_vulkan/vk_device.h b/src/video_core/renderer_vulkan/vk_device.h index 6b9227b09..ae5c21baa 100644 --- a/src/video_core/renderer_vulkan/vk_device.h +++ b/src/video_core/renderer_vulkan/vk_device.h @@ -182,6 +182,11 @@ public: return ext_custom_border_color; } + /// Returns true if the device supports VK_EXT_extended_dynamic_state. + bool IsExtExtendedDynamicStateSupported() const { + return ext_extended_dynamic_state; + } + /// Returns the vendor name reported from Vulkan. std::string_view GetVendorName() const { return vendor_name; @@ -239,6 +244,7 @@ private: bool ext_shader_viewport_index_layer{}; ///< Support for VK_EXT_shader_viewport_index_layer. bool ext_transform_feedback{}; ///< Support for VK_EXT_transform_feedback. bool ext_custom_border_color{}; ///< Support for VK_EXT_custom_border_color. + bool ext_extended_dynamic_state{}; ///< Support for VK_EXT_extended_dynamic_state. bool nv_device_diagnostics_config{}; ///< Support for VK_NV_device_diagnostics_config. // Telemetry parameters diff --git a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp index 69b6bba00..844445105 100644 --- a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp +++ b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp @@ -176,20 +176,32 @@ std::vector<vk::ShaderModule> VKGraphicsPipeline::CreateShaderModules( vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpass_params, const SPIRVProgram& program) const { - const auto& vi = fixed_state.vertex_input; - const auto& ds = fixed_state.depth_stencil; - const auto& cd = fixed_state.color_blending; - const auto& rs = fixed_state.rasterizer; - const auto& viewport_swizzles = fixed_state.viewport_swizzles.swizzles; + const auto& state = fixed_state; + const auto& viewport_swizzles = state.viewport_swizzles; + + FixedPipelineState::DynamicState dynamic; + if (device.IsExtExtendedDynamicStateSupported()) { + // Insert dummy values, as long as they are valid they don't matter as extended dynamic + // state is ignored + dynamic.raw1 = 0; + dynamic.raw2 = 0; + for (FixedPipelineState::VertexBinding& binding : dynamic.vertex_bindings) { + // Enable all vertex bindings + binding.raw = 0; + binding.enabled.Assign(1); + } + } else { + dynamic = state.dynamic_state; + } std::vector<VkVertexInputBindingDescription> vertex_bindings; std::vector<VkVertexInputBindingDivisorDescriptionEXT> vertex_binding_divisors; - for (std::size_t index = 0; index < std::size(vi.bindings); ++index) { - const auto& binding = vi.bindings[index]; + for (std::size_t index = 0; index < Maxwell::NumVertexArrays; ++index) { + const auto& binding = dynamic.vertex_bindings[index]; if (!binding.enabled) { continue; } - const bool instanced = vi.binding_divisors[index] != 0; + const bool instanced = state.binding_divisors[index] != 0; const auto rate = instanced ? VK_VERTEX_INPUT_RATE_INSTANCE : VK_VERTEX_INPUT_RATE_VERTEX; auto& vertex_binding = vertex_bindings.emplace_back(); @@ -200,14 +212,14 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa if (instanced) { auto& binding_divisor = vertex_binding_divisors.emplace_back(); binding_divisor.binding = static_cast<u32>(index); - binding_divisor.divisor = vi.binding_divisors[index]; + binding_divisor.divisor = state.binding_divisors[index]; } } std::vector<VkVertexInputAttributeDescription> vertex_attributes; const auto& input_attributes = program[0]->entries.attributes; - for (std::size_t index = 0; index < std::size(vi.attributes); ++index) { - const auto& attribute = vi.attributes[index]; + for (std::size_t index = 0; index < state.attributes.size(); ++index) { + const auto& attribute = state.attributes[index]; if (!attribute.enabled) { continue; } @@ -244,15 +256,15 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa input_assembly_ci.sType = VK_STRUCTURE_TYPE_PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO; input_assembly_ci.pNext = nullptr; input_assembly_ci.flags = 0; - input_assembly_ci.topology = MaxwellToVK::PrimitiveTopology(device, rs.Topology()); + input_assembly_ci.topology = MaxwellToVK::PrimitiveTopology(device, dynamic.Topology()); input_assembly_ci.primitiveRestartEnable = - rs.primitive_restart_enable != 0 && SupportsPrimitiveRestart(input_assembly_ci.topology); + state.primitive_restart_enable != 0 && SupportsPrimitiveRestart(input_assembly_ci.topology); VkPipelineTessellationStateCreateInfo tessellation_ci; tessellation_ci.sType = VK_STRUCTURE_TYPE_PIPELINE_TESSELLATION_STATE_CREATE_INFO; tessellation_ci.pNext = nullptr; tessellation_ci.flags = 0; - tessellation_ci.patchControlPoints = rs.patch_control_points_minus_one.Value() + 1; + tessellation_ci.patchControlPoints = state.patch_control_points_minus_one.Value() + 1; VkPipelineViewportStateCreateInfo viewport_ci; viewport_ci.sType = VK_STRUCTURE_TYPE_PIPELINE_VIEWPORT_STATE_CREATE_INFO; @@ -280,13 +292,13 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa rasterization_ci.sType = VK_STRUCTURE_TYPE_PIPELINE_RASTERIZATION_STATE_CREATE_INFO; rasterization_ci.pNext = nullptr; rasterization_ci.flags = 0; - rasterization_ci.depthClampEnable = rs.depth_clamp_disabled == 0 ? VK_TRUE : VK_FALSE; - rasterization_ci.rasterizerDiscardEnable = rs.rasterize_enable == 0 ? VK_TRUE : VK_FALSE; + rasterization_ci.depthClampEnable = state.depth_clamp_disabled == 0 ? VK_TRUE : VK_FALSE; + rasterization_ci.rasterizerDiscardEnable = state.rasterize_enable == 0 ? VK_TRUE : VK_FALSE; rasterization_ci.polygonMode = VK_POLYGON_MODE_FILL; rasterization_ci.cullMode = - rs.cull_enable ? MaxwellToVK::CullFace(rs.CullFace()) : VK_CULL_MODE_NONE; - rasterization_ci.frontFace = MaxwellToVK::FrontFace(rs.FrontFace()); - rasterization_ci.depthBiasEnable = rs.depth_bias_enable; + dynamic.cull_enable ? MaxwellToVK::CullFace(dynamic.CullFace()) : VK_CULL_MODE_NONE; + rasterization_ci.frontFace = MaxwellToVK::FrontFace(dynamic.FrontFace()); + rasterization_ci.depthBiasEnable = state.depth_bias_enable; rasterization_ci.depthBiasConstantFactor = 0.0f; rasterization_ci.depthBiasClamp = 0.0f; rasterization_ci.depthBiasSlopeFactor = 0.0f; @@ -307,14 +319,15 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa depth_stencil_ci.sType = VK_STRUCTURE_TYPE_PIPELINE_DEPTH_STENCIL_STATE_CREATE_INFO; depth_stencil_ci.pNext = nullptr; depth_stencil_ci.flags = 0; - depth_stencil_ci.depthTestEnable = ds.depth_test_enable; - depth_stencil_ci.depthWriteEnable = ds.depth_write_enable; - depth_stencil_ci.depthCompareOp = - ds.depth_test_enable ? MaxwellToVK::ComparisonOp(ds.DepthTestFunc()) : VK_COMPARE_OP_ALWAYS; - depth_stencil_ci.depthBoundsTestEnable = ds.depth_bounds_enable; - depth_stencil_ci.stencilTestEnable = ds.stencil_enable; - depth_stencil_ci.front = GetStencilFaceState(ds.front); - depth_stencil_ci.back = GetStencilFaceState(ds.back); + depth_stencil_ci.depthTestEnable = dynamic.depth_test_enable; + depth_stencil_ci.depthWriteEnable = dynamic.depth_write_enable; + depth_stencil_ci.depthCompareOp = dynamic.depth_test_enable + ? MaxwellToVK::ComparisonOp(dynamic.DepthTestFunc()) + : VK_COMPARE_OP_ALWAYS; + depth_stencil_ci.depthBoundsTestEnable = dynamic.depth_bounds_enable; + depth_stencil_ci.stencilTestEnable = dynamic.stencil_enable; + depth_stencil_ci.front = GetStencilFaceState(dynamic.front); + depth_stencil_ci.back = GetStencilFaceState(dynamic.back); depth_stencil_ci.minDepthBounds = 0.0f; depth_stencil_ci.maxDepthBounds = 0.0f; @@ -324,7 +337,7 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa static constexpr std::array COMPONENT_TABLE = { VK_COLOR_COMPONENT_R_BIT, VK_COLOR_COMPONENT_G_BIT, VK_COLOR_COMPONENT_B_BIT, VK_COLOR_COMPONENT_A_BIT}; - const auto& blend = cd.attachments[index]; + const auto& blend = state.attachments[index]; VkColorComponentFlags color_components = 0; for (std::size_t i = 0; i < COMPONENT_TABLE.size(); ++i) { @@ -354,11 +367,27 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa color_blend_ci.pAttachments = cb_attachments.data(); std::memset(color_blend_ci.blendConstants, 0, sizeof(color_blend_ci.blendConstants)); - static constexpr std::array dynamic_states = { + std::vector dynamic_states = { VK_DYNAMIC_STATE_VIEWPORT, VK_DYNAMIC_STATE_SCISSOR, VK_DYNAMIC_STATE_DEPTH_BIAS, VK_DYNAMIC_STATE_BLEND_CONSTANTS, VK_DYNAMIC_STATE_DEPTH_BOUNDS, VK_DYNAMIC_STATE_STENCIL_COMPARE_MASK, - VK_DYNAMIC_STATE_STENCIL_WRITE_MASK, VK_DYNAMIC_STATE_STENCIL_REFERENCE}; + VK_DYNAMIC_STATE_STENCIL_WRITE_MASK, VK_DYNAMIC_STATE_STENCIL_REFERENCE, + }; + if (device.IsExtExtendedDynamicStateSupported()) { + static constexpr std::array extended = { + VK_DYNAMIC_STATE_CULL_MODE_EXT, + VK_DYNAMIC_STATE_FRONT_FACE_EXT, + VK_DYNAMIC_STATE_PRIMITIVE_TOPOLOGY_EXT, + VK_DYNAMIC_STATE_VERTEX_INPUT_BINDING_STRIDE_EXT, + VK_DYNAMIC_STATE_DEPTH_TEST_ENABLE_EXT, + VK_DYNAMIC_STATE_DEPTH_WRITE_ENABLE_EXT, + VK_DYNAMIC_STATE_DEPTH_COMPARE_OP_EXT, + VK_DYNAMIC_STATE_DEPTH_BOUNDS_TEST_ENABLE_EXT, + VK_DYNAMIC_STATE_STENCIL_TEST_ENABLE_EXT, + VK_DYNAMIC_STATE_STENCIL_OP_EXT, + }; + dynamic_states.insert(dynamic_states.end(), extended.begin(), extended.end()); + } VkPipelineDynamicStateCreateInfo dynamic_state_ci; dynamic_state_ci.sType = VK_STRUCTURE_TYPE_PIPELINE_DYNAMIC_STATE_CREATE_INFO; diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp index ea66e621e..3da835324 100644 --- a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp @@ -116,12 +116,12 @@ u32 FillDescriptorLayout(const ShaderEntries& entries, } // Anonymous namespace std::size_t GraphicsPipelineCacheKey::Hash() const noexcept { - const u64 hash = Common::CityHash64(reinterpret_cast<const char*>(this), sizeof *this); + const u64 hash = Common::CityHash64(reinterpret_cast<const char*>(this), Size()); return static_cast<std::size_t>(hash); } bool GraphicsPipelineCacheKey::operator==(const GraphicsPipelineCacheKey& rhs) const noexcept { - return std::memcmp(&rhs, this, sizeof *this) == 0; + return std::memcmp(&rhs, this, Size()) == 0; } std::size_t ComputePipelineCacheKey::Hash() const noexcept { @@ -312,18 +312,19 @@ VKPipelineCache::DecompileShaders(const GraphicsPipelineCacheKey& key) { const auto& gpu = system.GPU().Maxwell3D(); Specialization specialization; - if (fixed_state.rasterizer.Topology() == Maxwell::PrimitiveTopology::Points) { + if (fixed_state.dynamic_state.Topology() == Maxwell::PrimitiveTopology::Points || + device.IsExtExtendedDynamicStateSupported()) { float point_size; - std::memcpy(&point_size, &fixed_state.rasterizer.point_size, sizeof(float)); + std::memcpy(&point_size, &fixed_state.point_size, sizeof(float)); specialization.point_size = point_size; ASSERT(point_size != 0.0f); } for (std::size_t i = 0; i < Maxwell::NumVertexAttributes; ++i) { - const auto& attribute = fixed_state.vertex_input.attributes[i]; + const auto& attribute = fixed_state.attributes[i]; specialization.enabled_attributes[i] = attribute.enabled.Value() != 0; specialization.attribute_types[i] = attribute.Type(); } - specialization.ndc_minus_one_to_one = fixed_state.rasterizer.ndc_minus_one_to_one; + specialization.ndc_minus_one_to_one = fixed_state.ndc_minus_one_to_one; SPIRVProgram program; std::vector<VkDescriptorSetLayoutBinding> bindings; diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.h b/src/video_core/renderer_vulkan/vk_pipeline_cache.h index 0a36e5112..0a3fe65fb 100644 --- a/src/video_core/renderer_vulkan/vk_pipeline_cache.h +++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.h @@ -44,10 +44,10 @@ class VKUpdateDescriptorQueue; using Maxwell = Tegra::Engines::Maxwell3D::Regs; struct GraphicsPipelineCacheKey { - FixedPipelineState fixed_state; RenderPassParams renderpass_params; + u32 padding; std::array<GPUVAddr, Maxwell::MaxShaderProgram> shaders; - u64 padding; // This is necessary for unique object representations + FixedPipelineState fixed_state; std::size_t Hash() const noexcept; @@ -56,6 +56,10 @@ struct GraphicsPipelineCacheKey { bool operator!=(const GraphicsPipelineCacheKey& rhs) const noexcept { return !operator==(rhs); } + + std::size_t Size() const noexcept { + return sizeof(renderpass_params) + sizeof(padding) + sizeof(shaders) + fixed_state.Size(); + } }; static_assert(std::has_unique_object_representations_v<GraphicsPipelineCacheKey>); static_assert(std::is_trivially_copyable_v<GraphicsPipelineCacheKey>); diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp index a8d94eac3..380ed532b 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp +++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp @@ -186,13 +186,22 @@ bool HasToPreserveDepthContents(bool is_clear, const Maxwell& regs) { scissor.max_y < regs.zeta_height; } +template <std::size_t N> +std::array<VkDeviceSize, N> ExpandStrides(const std::array<u16, N>& strides) { + std::array<VkDeviceSize, N> expanded; + std::copy(strides.begin(), strides.end(), expanded.begin()); + return expanded; +} + } // Anonymous namespace class BufferBindings final { public: - void AddVertexBinding(VkBuffer buffer, VkDeviceSize offset) { + void AddVertexBinding(VkBuffer buffer, VkDeviceSize offset, VkDeviceSize size, u32 stride) { vertex.buffers[vertex.num_buffers] = buffer; vertex.offsets[vertex.num_buffers] = offset; + vertex.sizes[vertex.num_buffers] = size; + vertex.strides[vertex.num_buffers] = static_cast<u16>(stride); ++vertex.num_buffers; } @@ -202,76 +211,76 @@ public: index.type = type; } - void Bind(VKScheduler& scheduler) const { + void Bind(const VKDevice& device, VKScheduler& scheduler) const { // Use this large switch case to avoid dispatching more memory in the record lambda than // what we need. It looks horrible, but it's the best we can do on standard C++. switch (vertex.num_buffers) { case 0: - return BindStatic<0>(scheduler); + return BindStatic<0>(device, scheduler); case 1: - return BindStatic<1>(scheduler); + return BindStatic<1>(device, scheduler); case 2: - return BindStatic<2>(scheduler); + return BindStatic<2>(device, scheduler); case 3: - return BindStatic<3>(scheduler); + return BindStatic<3>(device, scheduler); case 4: - return BindStatic<4>(scheduler); + return BindStatic<4>(device, scheduler); case 5: - return BindStatic<5>(scheduler); + return BindStatic<5>(device, scheduler); case 6: - return BindStatic<6>(scheduler); + return BindStatic<6>(device, scheduler); case 7: - return BindStatic<7>(scheduler); + return BindStatic<7>(device, scheduler); case 8: - return BindStatic<8>(scheduler); + return BindStatic<8>(device, scheduler); case 9: - return BindStatic<9>(scheduler); + return BindStatic<9>(device, scheduler); case 10: - return BindStatic<10>(scheduler); + return BindStatic<10>(device, scheduler); case 11: - return BindStatic<11>(scheduler); + return BindStatic<11>(device, scheduler); case 12: - return BindStatic<12>(scheduler); + return BindStatic<12>(device, scheduler); case 13: - return BindStatic<13>(scheduler); + return BindStatic<13>(device, scheduler); case 14: - return BindStatic<14>(scheduler); + return BindStatic<14>(device, scheduler); case 15: - return BindStatic<15>(scheduler); + return BindStatic<15>(device, scheduler); case 16: - return BindStatic<16>(scheduler); + return BindStatic<16>(device, scheduler); case 17: - return BindStatic<17>(scheduler); + return BindStatic<17>(device, scheduler); case 18: - return BindStatic<18>(scheduler); + return BindStatic<18>(device, scheduler); case 19: - return BindStatic<19>(scheduler); + return BindStatic<19>(device, scheduler); case 20: - return BindStatic<20>(scheduler); + return BindStatic<20>(device, scheduler); case 21: - return BindStatic<21>(scheduler); + return BindStatic<21>(device, scheduler); case 22: - return BindStatic<22>(scheduler); + return BindStatic<22>(device, scheduler); case 23: - return BindStatic<23>(scheduler); + return BindStatic<23>(device, scheduler); case 24: - return BindStatic<24>(scheduler); + return BindStatic<24>(device, scheduler); case 25: - return BindStatic<25>(scheduler); + return BindStatic<25>(device, scheduler); case 26: - return BindStatic<26>(scheduler); + return BindStatic<26>(device, scheduler); case 27: - return BindStatic<27>(scheduler); + return BindStatic<27>(device, scheduler); case 28: - return BindStatic<28>(scheduler); + return BindStatic<28>(device, scheduler); case 29: - return BindStatic<29>(scheduler); + return BindStatic<29>(device, scheduler); case 30: - return BindStatic<30>(scheduler); + return BindStatic<30>(device, scheduler); case 31: - return BindStatic<31>(scheduler); + return BindStatic<31>(device, scheduler); case 32: - return BindStatic<32>(scheduler); + return BindStatic<32>(device, scheduler); } UNREACHABLE(); } @@ -282,6 +291,8 @@ private: std::size_t num_buffers = 0; std::array<VkBuffer, Maxwell::NumVertexArrays> buffers; std::array<VkDeviceSize, Maxwell::NumVertexArrays> offsets; + std::array<VkDeviceSize, Maxwell::NumVertexArrays> sizes; + std::array<u16, Maxwell::NumVertexArrays> strides; } vertex; struct { @@ -291,15 +302,23 @@ private: } index; template <std::size_t N> - void BindStatic(VKScheduler& scheduler) const { - if (index.buffer) { - BindStatic<N, true>(scheduler); + void BindStatic(const VKDevice& device, VKScheduler& scheduler) const { + if (device.IsExtExtendedDynamicStateSupported()) { + if (index.buffer) { + BindStatic<N, true, true>(scheduler); + } else { + BindStatic<N, false, true>(scheduler); + } } else { - BindStatic<N, false>(scheduler); + if (index.buffer) { + BindStatic<N, true, false>(scheduler); + } else { + BindStatic<N, false, false>(scheduler); + } } } - template <std::size_t N, bool is_indexed> + template <std::size_t N, bool is_indexed, bool has_extended_dynamic_state> void BindStatic(VKScheduler& scheduler) const { static_assert(N <= Maxwell::NumVertexArrays); if constexpr (N == 0) { @@ -311,6 +330,31 @@ private: std::copy(vertex.buffers.begin(), vertex.buffers.begin() + N, buffers.begin()); std::copy(vertex.offsets.begin(), vertex.offsets.begin() + N, offsets.begin()); + if constexpr (has_extended_dynamic_state) { + // With extended dynamic states we can specify the length and stride of a vertex buffer + // std::array<VkDeviceSize, N> sizes; + std::array<u16, N> strides; + // std::copy(vertex.sizes.begin(), vertex.sizes.begin() + N, sizes.begin()); + std::copy(vertex.strides.begin(), vertex.strides.begin() + N, strides.begin()); + + if constexpr (is_indexed) { + scheduler.Record( + [buffers, offsets, strides, index = index](vk::CommandBuffer cmdbuf) { + cmdbuf.BindIndexBuffer(index.buffer, index.offset, index.type); + cmdbuf.BindVertexBuffers2EXT(0, static_cast<u32>(N), buffers.data(), + offsets.data(), nullptr, + ExpandStrides(strides).data()); + }); + } else { + scheduler.Record([buffers, offsets, strides](vk::CommandBuffer cmdbuf) { + cmdbuf.BindVertexBuffers2EXT(0, static_cast<u32>(N), buffers.data(), + offsets.data(), nullptr, + ExpandStrides(strides).data()); + }); + } + return; + } + if constexpr (is_indexed) { // Indexed draw scheduler.Record([buffers, offsets, index = index](vk::CommandBuffer cmdbuf) { @@ -369,7 +413,7 @@ void RasterizerVulkan::Draw(bool is_indexed, bool is_instanced) { const auto& gpu = system.GPU().Maxwell3D(); GraphicsPipelineCacheKey key; - key.fixed_state.Fill(gpu.regs); + key.fixed_state.Fill(gpu.regs, device.IsExtExtendedDynamicStateSupported()); buffer_cache.Map(CalculateGraphicsStreamBufferSize(is_indexed)); @@ -402,7 +446,7 @@ void RasterizerVulkan::Draw(bool is_indexed, bool is_instanced) { UpdateDynamicStates(); - buffer_bindings.Bind(scheduler); + buffer_bindings.Bind(device, scheduler); BeginTransformFeedback(); @@ -822,7 +866,7 @@ RasterizerVulkan::DrawParameters RasterizerVulkan::SetupGeometry(FixedPipelineSt const auto& gpu = system.GPU().Maxwell3D(); const auto& regs = gpu.regs; - SetupVertexArrays(fixed_state.vertex_input, buffer_bindings); + SetupVertexArrays(buffer_bindings); const u32 base_instance = regs.vb_base_instance; const u32 num_instances = is_instanced ? gpu.mme_draw.instance_count : 1; @@ -893,6 +937,17 @@ void RasterizerVulkan::UpdateDynamicStates() { UpdateBlendConstants(regs); UpdateDepthBounds(regs); UpdateStencilFaces(regs); + if (device.IsExtExtendedDynamicStateSupported()) { + UpdateCullMode(regs); + UpdateDepthBoundsTestEnable(regs); + UpdateDepthTestEnable(regs); + UpdateDepthWriteEnable(regs); + UpdateDepthCompareOp(regs); + UpdateFrontFace(regs); + UpdatePrimitiveTopology(regs); + UpdateStencilOp(regs); + UpdateStencilTestEnable(regs); + } } void RasterizerVulkan::BeginTransformFeedback() { @@ -940,41 +995,25 @@ void RasterizerVulkan::EndTransformFeedback() { [](vk::CommandBuffer cmdbuf) { cmdbuf.EndTransformFeedbackEXT(0, 0, nullptr, nullptr); }); } -void RasterizerVulkan::SetupVertexArrays(FixedPipelineState::VertexInput& vertex_input, - BufferBindings& buffer_bindings) { +void RasterizerVulkan::SetupVertexArrays(BufferBindings& buffer_bindings) { const auto& regs = system.GPU().Maxwell3D().regs; - for (std::size_t index = 0; index < Maxwell::NumVertexAttributes; ++index) { - const auto& attrib = regs.vertex_attrib_format[index]; - if (attrib.IsConstant()) { - vertex_input.SetAttribute(index, false, 0, 0, {}, {}); - continue; - } - vertex_input.SetAttribute(index, true, attrib.buffer, attrib.offset, attrib.type.Value(), - attrib.size.Value()); - } - for (std::size_t index = 0; index < Maxwell::NumVertexArrays; ++index) { const auto& vertex_array = regs.vertex_array[index]; if (!vertex_array.IsEnabled()) { - vertex_input.SetBinding(index, false, 0, 0); continue; } - vertex_input.SetBinding( - index, true, vertex_array.stride, - regs.instanced_arrays.IsInstancingEnabled(index) ? vertex_array.divisor : 0); - const GPUVAddr start{vertex_array.StartAddress()}; const GPUVAddr end{regs.vertex_array_limit[index].LimitAddress()}; ASSERT(end >= start); - const std::size_t size{end - start}; + const std::size_t size = end - start; if (size == 0) { - buffer_bindings.AddVertexBinding(DefaultBuffer(), 0); + buffer_bindings.AddVertexBinding(DefaultBuffer(), 0, DEFAULT_BUFFER_SIZE, 0); continue; } const auto info = buffer_cache.UploadMemory(start, size); - buffer_bindings.AddVertexBinding(info.handle, info.offset); + buffer_bindings.AddVertexBinding(info.handle, info.offset, size, vertex_array.stride); } } @@ -1326,6 +1365,117 @@ void RasterizerVulkan::UpdateStencilFaces(Tegra::Engines::Maxwell3D::Regs& regs) } } +void RasterizerVulkan::UpdateCullMode(Tegra::Engines::Maxwell3D::Regs& regs) { + if (!state_tracker.TouchCullMode()) { + return; + } + scheduler.Record( + [enabled = regs.cull_test_enabled, cull_face = regs.cull_face](vk::CommandBuffer cmdbuf) { + cmdbuf.SetCullModeEXT(enabled ? MaxwellToVK::CullFace(cull_face) : VK_CULL_MODE_NONE); + }); +} + +void RasterizerVulkan::UpdateDepthBoundsTestEnable(Tegra::Engines::Maxwell3D::Regs& regs) { + if (!state_tracker.TouchDepthBoundsTestEnable()) { + return; + } + scheduler.Record([enable = regs.depth_bounds_enable](vk::CommandBuffer cmdbuf) { + cmdbuf.SetDepthBoundsTestEnableEXT(enable); + }); +} + +void RasterizerVulkan::UpdateDepthTestEnable(Tegra::Engines::Maxwell3D::Regs& regs) { + if (!state_tracker.TouchDepthTestEnable()) { + return; + } + scheduler.Record([enable = regs.depth_test_enable](vk::CommandBuffer cmdbuf) { + cmdbuf.SetDepthTestEnableEXT(enable); + }); +} + +void RasterizerVulkan::UpdateDepthWriteEnable(Tegra::Engines::Maxwell3D::Regs& regs) { + if (!state_tracker.TouchDepthWriteEnable()) { + return; + } + scheduler.Record([enable = regs.depth_write_enabled](vk::CommandBuffer cmdbuf) { + cmdbuf.SetDepthWriteEnableEXT(enable); + }); +} + +void RasterizerVulkan::UpdateDepthCompareOp(Tegra::Engines::Maxwell3D::Regs& regs) { + if (!state_tracker.TouchDepthCompareOp()) { + return; + } + scheduler.Record([func = regs.depth_test_func](vk::CommandBuffer cmdbuf) { + cmdbuf.SetDepthCompareOpEXT(MaxwellToVK::ComparisonOp(func)); + }); +} + +void RasterizerVulkan::UpdateFrontFace(Tegra::Engines::Maxwell3D::Regs& regs) { + if (!state_tracker.TouchFrontFace()) { + return; + } + + VkFrontFace front_face = MaxwellToVK::FrontFace(regs.front_face); + if (regs.screen_y_control.triangle_rast_flip != 0) { + front_face = front_face == VK_FRONT_FACE_CLOCKWISE ? VK_FRONT_FACE_COUNTER_CLOCKWISE + : VK_FRONT_FACE_CLOCKWISE; + } + scheduler.Record( + [front_face](vk::CommandBuffer cmdbuf) { cmdbuf.SetFrontFaceEXT(front_face); }); +} + +void RasterizerVulkan::UpdatePrimitiveTopology(Tegra::Engines::Maxwell3D::Regs& regs) { + if (!state_tracker.TouchPrimitiveTopology()) { + return; + } + const Maxwell::PrimitiveTopology primitive_topology = regs.draw.topology.Value(); + scheduler.Record([this, primitive_topology](vk::CommandBuffer cmdbuf) { + cmdbuf.SetPrimitiveTopologyEXT(MaxwellToVK::PrimitiveTopology(device, primitive_topology)); + }); +} + +void RasterizerVulkan::UpdateStencilOp(Tegra::Engines::Maxwell3D::Regs& regs) { + if (!state_tracker.TouchStencilOp()) { + return; + } + const Maxwell::StencilOp fail = regs.stencil_front_op_fail; + const Maxwell::StencilOp zfail = regs.stencil_front_op_zfail; + const Maxwell::StencilOp zpass = regs.stencil_front_op_zpass; + const Maxwell::ComparisonOp compare = regs.stencil_front_func_func; + if (regs.stencil_two_side_enable) { + scheduler.Record([fail, zfail, zpass, compare](vk::CommandBuffer cmdbuf) { + cmdbuf.SetStencilOpEXT(VK_STENCIL_FACE_FRONT_AND_BACK, MaxwellToVK::StencilOp(fail), + MaxwellToVK::StencilOp(zpass), MaxwellToVK::StencilOp(zfail), + MaxwellToVK::ComparisonOp(compare)); + }); + } else { + const Maxwell::StencilOp back_fail = regs.stencil_back_op_fail; + const Maxwell::StencilOp back_zfail = regs.stencil_back_op_zfail; + const Maxwell::StencilOp back_zpass = regs.stencil_back_op_zpass; + const Maxwell::ComparisonOp back_compare = regs.stencil_back_func_func; + scheduler.Record([fail, zfail, zpass, compare, back_fail, back_zfail, back_zpass, + back_compare](vk::CommandBuffer cmdbuf) { + cmdbuf.SetStencilOpEXT(VK_STENCIL_FACE_FRONT_BIT, MaxwellToVK::StencilOp(fail), + MaxwellToVK::StencilOp(zpass), MaxwellToVK::StencilOp(zfail), + MaxwellToVK::ComparisonOp(compare)); + cmdbuf.SetStencilOpEXT(VK_STENCIL_FACE_BACK_BIT, MaxwellToVK::StencilOp(back_fail), + MaxwellToVK::StencilOp(back_zpass), + MaxwellToVK::StencilOp(back_zfail), + MaxwellToVK::ComparisonOp(back_compare)); + }); + } +} + +void RasterizerVulkan::UpdateStencilTestEnable(Tegra::Engines::Maxwell3D::Regs& regs) { + if (!state_tracker.TouchStencilTestEnable()) { + return; + } + scheduler.Record([enable = regs.stencil_enable](vk::CommandBuffer cmdbuf) { + cmdbuf.SetStencilTestEnableEXT(enable); + }); +} + std::size_t RasterizerVulkan::CalculateGraphicsStreamBufferSize(bool is_indexed) const { std::size_t size = CalculateVertexArraysSize(); if (is_indexed) { diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.h b/src/video_core/renderer_vulkan/vk_rasterizer.h index 83e00e7e9..923178b0b 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.h +++ b/src/video_core/renderer_vulkan/vk_rasterizer.h @@ -185,8 +185,7 @@ private: bool WalkAttachmentOverlaps(const CachedSurfaceView& attachment); - void SetupVertexArrays(FixedPipelineState::VertexInput& vertex_input, - BufferBindings& buffer_bindings); + void SetupVertexArrays(BufferBindings& buffer_bindings); void SetupIndexBuffer(BufferBindings& buffer_bindings, DrawParameters& params, bool is_indexed); @@ -246,6 +245,16 @@ private: void UpdateDepthBounds(Tegra::Engines::Maxwell3D::Regs& regs); void UpdateStencilFaces(Tegra::Engines::Maxwell3D::Regs& regs); + void UpdateCullMode(Tegra::Engines::Maxwell3D::Regs& regs); + void UpdateDepthBoundsTestEnable(Tegra::Engines::Maxwell3D::Regs& regs); + void UpdateDepthTestEnable(Tegra::Engines::Maxwell3D::Regs& regs); + void UpdateDepthWriteEnable(Tegra::Engines::Maxwell3D::Regs& regs); + void UpdateDepthCompareOp(Tegra::Engines::Maxwell3D::Regs& regs); + void UpdateFrontFace(Tegra::Engines::Maxwell3D::Regs& regs); + void UpdatePrimitiveTopology(Tegra::Engines::Maxwell3D::Regs& regs); + void UpdateStencilOp(Tegra::Engines::Maxwell3D::Regs& regs); + void UpdateStencilTestEnable(Tegra::Engines::Maxwell3D::Regs& regs); + std::size_t CalculateGraphicsStreamBufferSize(bool is_indexed) const; std::size_t CalculateComputeStreamBufferSize() const; diff --git a/src/video_core/renderer_vulkan/vk_state_tracker.cpp b/src/video_core/renderer_vulkan/vk_state_tracker.cpp index 94a89e388..e5a583dd5 100644 --- a/src/video_core/renderer_vulkan/vk_state_tracker.cpp +++ b/src/video_core/renderer_vulkan/vk_state_tracker.cpp @@ -36,6 +36,15 @@ Flags MakeInvalidationFlags() { flags[BlendConstants] = true; flags[DepthBounds] = true; flags[StencilProperties] = true; + flags[CullMode] = true; + flags[DepthBoundsEnable] = true; + flags[DepthTestEnable] = true; + flags[DepthWriteEnable] = true; + flags[DepthCompareOp] = true; + flags[FrontFace] = true; + flags[PrimitiveTopology] = true; + flags[StencilOp] = true; + flags[StencilTestEnable] = true; return flags; } @@ -75,6 +84,57 @@ void SetupDirtyStencilProperties(Tables& tables) { table[OFF(stencil_back_func_mask)] = StencilProperties; } +void SetupDirtyCullMode(Tables& tables) { + auto& table = tables[0]; + table[OFF(cull_face)] = CullMode; + table[OFF(cull_test_enabled)] = CullMode; +} + +void SetupDirtyDepthBoundsEnable(Tables& tables) { + tables[0][OFF(depth_bounds_enable)] = DepthBoundsEnable; +} + +void SetupDirtyDepthTestEnable(Tables& tables) { + tables[0][OFF(depth_test_enable)] = DepthTestEnable; +} + +void SetupDirtyDepthWriteEnable(Tables& tables) { + tables[0][OFF(depth_write_enabled)] = DepthWriteEnable; +} + +void SetupDirtyDepthCompareOp(Tables& tables) { + tables[0][OFF(depth_test_func)] = DepthCompareOp; +} + +void SetupDirtyFrontFace(Tables& tables) { + auto& table = tables[0]; + table[OFF(front_face)] = FrontFace; + table[OFF(screen_y_control)] = FrontFace; +} + +void SetupDirtyPrimitiveTopology(Tables& tables) { + tables[0][OFF(draw.topology)] = PrimitiveTopology; +} + +void SetupDirtyStencilOp(Tables& tables) { + auto& table = tables[0]; + table[OFF(stencil_front_op_fail)] = StencilOp; + table[OFF(stencil_front_op_zfail)] = StencilOp; + table[OFF(stencil_front_op_zpass)] = StencilOp; + table[OFF(stencil_front_func_func)] = StencilOp; + table[OFF(stencil_back_op_fail)] = StencilOp; + table[OFF(stencil_back_op_zfail)] = StencilOp; + table[OFF(stencil_back_op_zpass)] = StencilOp; + table[OFF(stencil_back_func_func)] = StencilOp; + + // Table 0 is used by StencilProperties + tables[1][OFF(stencil_two_side_enable)] = StencilOp; +} + +void SetupDirtyStencilTestEnable(Tables& tables) { + tables[0][OFF(stencil_enable)] = StencilTestEnable; +} + } // Anonymous namespace StateTracker::StateTracker(Core::System& system) @@ -90,6 +150,14 @@ void StateTracker::Initialize() { SetupDirtyBlendConstants(tables); SetupDirtyDepthBounds(tables); SetupDirtyStencilProperties(tables); + SetupDirtyCullMode(tables); + SetupDirtyDepthBoundsEnable(tables); + SetupDirtyDepthTestEnable(tables); + SetupDirtyDepthWriteEnable(tables); + SetupDirtyDepthCompareOp(tables); + SetupDirtyFrontFace(tables); + SetupDirtyPrimitiveTopology(tables); + SetupDirtyStencilOp(tables); } void StateTracker::InvalidateCommandBufferState() { diff --git a/src/video_core/renderer_vulkan/vk_state_tracker.h b/src/video_core/renderer_vulkan/vk_state_tracker.h index 03bc415b2..54ca0d6c6 100644 --- a/src/video_core/renderer_vulkan/vk_state_tracker.h +++ b/src/video_core/renderer_vulkan/vk_state_tracker.h @@ -26,6 +26,16 @@ enum : u8 { DepthBounds, StencilProperties, + CullMode, + DepthBoundsEnable, + DepthTestEnable, + DepthWriteEnable, + DepthCompareOp, + FrontFace, + PrimitiveTopology, + StencilOp, + StencilTestEnable, + Last }; static_assert(Last <= std::numeric_limits<u8>::max()); @@ -64,6 +74,46 @@ public: return Exchange(Dirty::StencilProperties, false); } + bool TouchCullMode() { + return Exchange(Dirty::CullMode, false); + } + + bool TouchDepthBoundsTestEnable() { + return Exchange(Dirty::DepthBoundsEnable, false); + } + + bool TouchDepthTestEnable() { + return Exchange(Dirty::DepthTestEnable, false); + } + + bool TouchDepthBoundsEnable() { + return Exchange(Dirty::DepthBoundsEnable, false); + } + + bool TouchDepthWriteEnable() { + return Exchange(Dirty::DepthWriteEnable, false); + } + + bool TouchDepthCompareOp() { + return Exchange(Dirty::DepthCompareOp, false); + } + + bool TouchFrontFace() { + return Exchange(Dirty::FrontFace, false); + } + + bool TouchPrimitiveTopology() { + return Exchange(Dirty::PrimitiveTopology, false); + } + + bool TouchStencilOp() { + return Exchange(Dirty::StencilOp, false); + } + + bool TouchStencilTestEnable() { + return Exchange(Dirty::StencilTestEnable, false); + } + private: bool Exchange(std::size_t id, bool new_value) const noexcept { auto& flags = system.GPU().Maxwell3D().dirty.flags; diff --git a/src/video_core/renderer_vulkan/wrapper.cpp b/src/video_core/renderer_vulkan/wrapper.cpp index 0d485a662..051298cc8 100644 --- a/src/video_core/renderer_vulkan/wrapper.cpp +++ b/src/video_core/renderer_vulkan/wrapper.cpp @@ -88,6 +88,16 @@ void Load(VkDevice device, DeviceDispatch& dld) noexcept { X(vkCmdSetStencilWriteMask); X(vkCmdSetViewport); X(vkCmdWaitEvents); + X(vkCmdBindVertexBuffers2EXT); + X(vkCmdSetCullModeEXT); + X(vkCmdSetDepthBoundsTestEnableEXT); + X(vkCmdSetDepthCompareOpEXT); + X(vkCmdSetDepthTestEnableEXT); + X(vkCmdSetDepthWriteEnableEXT); + X(vkCmdSetFrontFaceEXT); + X(vkCmdSetPrimitiveTopologyEXT); + X(vkCmdSetStencilOpEXT); + X(vkCmdSetStencilTestEnableEXT); X(vkCreateBuffer); X(vkCreateBufferView); X(vkCreateCommandPool); diff --git a/src/video_core/renderer_vulkan/wrapper.h b/src/video_core/renderer_vulkan/wrapper.h index d56fdb3f9..71daac9d7 100644 --- a/src/video_core/renderer_vulkan/wrapper.h +++ b/src/video_core/renderer_vulkan/wrapper.h @@ -207,6 +207,16 @@ struct DeviceDispatch : public InstanceDispatch { PFN_vkCmdSetStencilWriteMask vkCmdSetStencilWriteMask; PFN_vkCmdSetViewport vkCmdSetViewport; PFN_vkCmdWaitEvents vkCmdWaitEvents; + PFN_vkCmdBindVertexBuffers2EXT vkCmdBindVertexBuffers2EXT; + PFN_vkCmdSetCullModeEXT vkCmdSetCullModeEXT; + PFN_vkCmdSetDepthBoundsTestEnableEXT vkCmdSetDepthBoundsTestEnableEXT; + PFN_vkCmdSetDepthCompareOpEXT vkCmdSetDepthCompareOpEXT; + PFN_vkCmdSetDepthTestEnableEXT vkCmdSetDepthTestEnableEXT; + PFN_vkCmdSetDepthWriteEnableEXT vkCmdSetDepthWriteEnableEXT; + PFN_vkCmdSetFrontFaceEXT vkCmdSetFrontFaceEXT; + PFN_vkCmdSetPrimitiveTopologyEXT vkCmdSetPrimitiveTopologyEXT; + PFN_vkCmdSetStencilOpEXT vkCmdSetStencilOpEXT; + PFN_vkCmdSetStencilTestEnableEXT vkCmdSetStencilTestEnableEXT; PFN_vkCreateBuffer vkCreateBuffer; PFN_vkCreateBufferView vkCreateBufferView; PFN_vkCreateCommandPool vkCreateCommandPool; @@ -969,6 +979,50 @@ public: buffer_barriers.data(), image_barriers.size(), image_barriers.data()); } + void BindVertexBuffers2EXT(u32 first_binding, u32 binding_count, const VkBuffer* buffers, + const VkDeviceSize* offsets, const VkDeviceSize* sizes, + const VkDeviceSize* strides) const noexcept { + dld->vkCmdBindVertexBuffers2EXT(handle, first_binding, binding_count, buffers, offsets, + sizes, strides); + } + + void SetCullModeEXT(VkCullModeFlags cull_mode) const noexcept { + dld->vkCmdSetCullModeEXT(handle, cull_mode); + } + + void SetDepthBoundsTestEnableEXT(bool enable) const noexcept { + dld->vkCmdSetDepthBoundsTestEnableEXT(handle, enable ? VK_TRUE : VK_FALSE); + } + + void SetDepthCompareOpEXT(VkCompareOp compare_op) const noexcept { + dld->vkCmdSetDepthCompareOpEXT(handle, compare_op); + } + + void SetDepthTestEnableEXT(bool enable) const noexcept { + dld->vkCmdSetDepthTestEnableEXT(handle, enable ? VK_TRUE : VK_FALSE); + } + + void SetDepthWriteEnableEXT(bool enable) const noexcept { + dld->vkCmdSetDepthWriteEnableEXT(handle, enable ? VK_TRUE : VK_FALSE); + } + + void SetFrontFaceEXT(VkFrontFace front_face) const noexcept { + dld->vkCmdSetFrontFaceEXT(handle, front_face); + } + + void SetPrimitiveTopologyEXT(VkPrimitiveTopology primitive_topology) const noexcept { + dld->vkCmdSetPrimitiveTopologyEXT(handle, primitive_topology); + } + + void SetStencilOpEXT(VkStencilFaceFlags face_mask, VkStencilOp fail_op, VkStencilOp pass_op, + VkStencilOp depth_fail_op, VkCompareOp compare_op) const noexcept { + dld->vkCmdSetStencilOpEXT(handle, face_mask, fail_op, pass_op, depth_fail_op, compare_op); + } + + void SetStencilTestEnableEXT(bool enable) const noexcept { + dld->vkCmdSetStencilTestEnableEXT(handle, enable ? VK_TRUE : VK_FALSE); + } + void BindTransformFeedbackBuffersEXT(u32 first, u32 count, const VkBuffer* buffers, const VkDeviceSize* offsets, const VkDeviceSize* sizes) const noexcept { diff --git a/src/video_core/shader_cache.h b/src/video_core/shader_cache.h index 2dd270e99..b7608fc7b 100644 --- a/src/video_core/shader_cache.h +++ b/src/video_core/shader_cache.h @@ -20,6 +20,7 @@ namespace VideoCommon { template <class T> class ShaderCache { static constexpr u64 PAGE_BITS = 14; + static constexpr u64 PAGE_SIZE = u64(1) << PAGE_BITS; struct Entry { VAddr addr_start; @@ -87,8 +88,8 @@ protected: const VAddr addr_end = addr + size; Entry* const entry = NewEntry(addr, addr_end, data.get()); - const u64 page_end = addr_end >> PAGE_BITS; - for (u64 page = addr >> PAGE_BITS; page <= page_end; ++page) { + const u64 page_end = (addr_end + PAGE_SIZE - 1) >> PAGE_BITS; + for (u64 page = addr >> PAGE_BITS; page < page_end; ++page) { invalidation_cache[page].push_back(entry); } @@ -108,20 +109,13 @@ private: /// @pre invalidation_mutex is locked void InvalidatePagesInRegion(VAddr addr, std::size_t size) { const VAddr addr_end = addr + size; - const u64 page_end = addr_end >> PAGE_BITS; - for (u64 page = addr >> PAGE_BITS; page <= page_end; ++page) { - const auto it = invalidation_cache.find(page); + const u64 page_end = (addr_end + PAGE_SIZE - 1) >> PAGE_BITS; + for (u64 page = addr >> PAGE_BITS; page < page_end; ++page) { + auto it = invalidation_cache.find(page); if (it == invalidation_cache.end()) { continue; } - - std::vector<Entry*>& entries = it->second; - InvalidatePageEntries(entries, addr, addr_end); - - // If there's nothing else in this page, remove it to avoid overpopulating the hash map. - if (entries.empty()) { - invalidation_cache.erase(it); - } + InvalidatePageEntries(it->second, addr, addr_end); } } @@ -131,15 +125,22 @@ private: if (marked_for_removal.empty()) { return; } - std::scoped_lock lock{lookup_mutex}; + // Remove duplicates + std::sort(marked_for_removal.begin(), marked_for_removal.end()); + marked_for_removal.erase(std::unique(marked_for_removal.begin(), marked_for_removal.end()), + marked_for_removal.end()); std::vector<T*> removed_shaders; removed_shaders.reserve(marked_for_removal.size()); + std::scoped_lock lock{lookup_mutex}; + for (Entry* const entry : marked_for_removal) { - if (lookup_cache.erase(entry->addr_start) > 0) { - removed_shaders.push_back(entry->data); - } + removed_shaders.push_back(entry->data); + + const auto it = lookup_cache.find(entry->addr_start); + ASSERT(it != lookup_cache.end()); + lookup_cache.erase(it); } marked_for_removal.clear(); @@ -154,17 +155,33 @@ private: /// @param addr_end Non-inclusive end address of the invalidation /// @pre invalidation_mutex is locked void InvalidatePageEntries(std::vector<Entry*>& entries, VAddr addr, VAddr addr_end) { - auto it = entries.begin(); - while (it != entries.end()) { - Entry* const entry = *it; + std::size_t index = 0; + while (index < entries.size()) { + Entry* const entry = entries[index]; if (!entry->Overlaps(addr, addr_end)) { - ++it; + ++index; continue; } + UnmarkMemory(entry); + RemoveEntryFromInvalidationCache(entry); marked_for_removal.push_back(entry); + } + } - it = entries.erase(it); + /// @brief Removes all references to an entry in the invalidation cache + /// @param entry Entry to remove from the invalidation cache + /// @pre invalidation_mutex is locked + void RemoveEntryFromInvalidationCache(const Entry* entry) { + const u64 page_end = (entry->addr_end + PAGE_SIZE - 1) >> PAGE_BITS; + for (u64 page = entry->addr_start >> PAGE_BITS; page < page_end; ++page) { + const auto entries_it = invalidation_cache.find(page); + ASSERT(entries_it != invalidation_cache.end()); + std::vector<Entry*>& entries = entries_it->second; + + const auto entry_it = std::find(entries.begin(), entries.end(), entry); + ASSERT(entry_it != entries.end()); + entries.erase(entry_it); } } @@ -182,16 +199,11 @@ private: } /// @brief Removes a vector of shaders from a list - /// @param removed_shaders Shaders to be removed from the storage, it can contain duplicates + /// @param removed_shaders Shaders to be removed from the storage /// @pre invalidation_mutex is locked /// @pre lookup_mutex is locked void RemoveShadersFromStorage(std::vector<T*> removed_shaders) { - // Remove duplicates - std::sort(removed_shaders.begin(), removed_shaders.end()); - removed_shaders.erase(std::unique(removed_shaders.begin(), removed_shaders.end()), - removed_shaders.end()); - - // Now that there are no duplicates, we can notify removals + // Notify removals for (T* const shader : removed_shaders) { OnShaderRemoval(shader); } diff --git a/src/yuzu/configuration/config.cpp b/src/yuzu/configuration/config.cpp index bbbd96113..5e0d0e7af 100644 --- a/src/yuzu/configuration/config.cpp +++ b/src/yuzu/configuration/config.cpp @@ -212,7 +212,7 @@ const std::array<int, Settings::NativeKeyboard::NumKeyboardMods> Config::default // UISetting::values.shortcuts, which is alphabetically ordered. // clang-format off const std::array<UISettings::Shortcut, 16> Config::default_hotkeys{{ - {QStringLiteral("Capture Screenshot"), QStringLiteral("Main Window"), {QStringLiteral("Ctrl+P"), Qt::ApplicationShortcut}}, + {QStringLiteral("Capture Screenshot"), QStringLiteral("Main Window"), {QStringLiteral("Ctrl+P"), Qt::WidgetWithChildrenShortcut}}, {QStringLiteral("Change Docked Mode"), QStringLiteral("Main Window"), {QStringLiteral("F10"), Qt::ApplicationShortcut}}, {QStringLiteral("Continue/Pause Emulation"), QStringLiteral("Main Window"), {QStringLiteral("F4"), Qt::WindowShortcut}}, {QStringLiteral("Decrease Speed Limit"), QStringLiteral("Main Window"), {QStringLiteral("-"), Qt::ApplicationShortcut}}, @@ -220,8 +220,8 @@ const std::array<UISettings::Shortcut, 16> Config::default_hotkeys{{ {QStringLiteral("Exit yuzu"), QStringLiteral("Main Window"), {QStringLiteral("Ctrl+Q"), Qt::WindowShortcut}}, {QStringLiteral("Fullscreen"), QStringLiteral("Main Window"), {QStringLiteral("F11"), Qt::WindowShortcut}}, {QStringLiteral("Increase Speed Limit"), QStringLiteral("Main Window"), {QStringLiteral("+"), Qt::ApplicationShortcut}}, - {QStringLiteral("Load Amiibo"), QStringLiteral("Main Window"), {QStringLiteral("F2"), Qt::ApplicationShortcut}}, - {QStringLiteral("Load File"), QStringLiteral("Main Window"), {QStringLiteral("Ctrl+O"), Qt::WindowShortcut}}, + {QStringLiteral("Load Amiibo"), QStringLiteral("Main Window"), {QStringLiteral("F2"), Qt::WidgetWithChildrenShortcut}}, + {QStringLiteral("Load File"), QStringLiteral("Main Window"), {QStringLiteral("Ctrl+O"), Qt::WidgetWithChildrenShortcut}}, {QStringLiteral("Mute Audio"), QStringLiteral("Main Window"), {QStringLiteral("Ctrl+M"), Qt::WindowShortcut}}, {QStringLiteral("Restart Emulation"), QStringLiteral("Main Window"), {QStringLiteral("F6"), Qt::WindowShortcut}}, {QStringLiteral("Stop Emulation"), QStringLiteral("Main Window"), {QStringLiteral("F5"), Qt::WindowShortcut}}, @@ -665,11 +665,13 @@ void Config::ReadShortcutValues() { const auto& [keyseq, context] = shortcut; qt_config->beginGroup(group); qt_config->beginGroup(name); + // No longer using ReadSetting for shortcut.second as it innacurately returns a value of 1 + // for WidgetWithChildrenShortcut which is a value of 3. Needed to fix shortcuts the open + // a file dialog in windowed mode UISettings::values.shortcuts.push_back( {name, group, - {ReadSetting(QStringLiteral("KeySeq"), keyseq).toString(), - ReadSetting(QStringLiteral("Context"), context).toInt()}}); + {ReadSetting(QStringLiteral("KeySeq"), keyseq).toString(), shortcut.second}}); qt_config->endGroup(); qt_config->endGroup(); } diff --git a/src/yuzu/configuration/configure_graphics.cpp b/src/yuzu/configuration/configure_graphics.cpp index 304625cd7..431f51d73 100644 --- a/src/yuzu/configuration/configure_graphics.cpp +++ b/src/yuzu/configuration/configure_graphics.cpp @@ -28,9 +28,9 @@ ConfigureGraphics::ConfigureGraphics(QWidget* parent) SetConfiguration(); - connect(ui->api, static_cast<void (QComboBox::*)(int)>(&QComboBox::currentIndexChanged), this, + connect(ui->api, qOverload<int>(&QComboBox::currentIndexChanged), this, [this] { UpdateDeviceComboBox(); }); - connect(ui->device, static_cast<void (QComboBox::*)(int)>(&QComboBox::activated), this, + connect(ui->device, qOverload<int>(&QComboBox::activated), this, [this](int device) { UpdateDeviceSelection(device); }); connect(ui->bg_button, &QPushButton::clicked, this, [this] { @@ -112,7 +112,7 @@ void ConfigureGraphics::UpdateDeviceComboBox() { enabled = false; break; case Settings::RendererBackend::Vulkan: - for (const auto device : vulkan_devices) { + for (const auto& device : vulkan_devices) { ui->device->addItem(device); } ui->device->setCurrentIndex(vulkan_device); diff --git a/src/yuzu/configuration/configure_input_player.cpp b/src/yuzu/configuration/configure_input_player.cpp index a05fa64ba..00433926d 100644 --- a/src/yuzu/configuration/configure_input_player.cpp +++ b/src/yuzu/configuration/configure_input_player.cpp @@ -70,6 +70,20 @@ static QString ButtonToText(const Common::ParamPackage& param) { return GetKeyName(param.Get("code", 0)); } + if (param.Get("engine", "") == "gcpad") { + if (param.Has("axis")) { + const QString axis_str = QString::fromStdString(param.Get("axis", "")); + const QString direction_str = QString::fromStdString(param.Get("direction", "")); + + return QObject::tr("GC Axis %1%2").arg(axis_str, direction_str); + } + if (param.Has("button")) { + const QString button_str = QString::number(int(std::log2(param.Get("button", 0)))); + return QObject::tr("GC Button %1").arg(button_str); + } + return GetKeyName(param.Get("code", 0)); + } + if (param.Get("engine", "") == "sdl") { if (param.Has("hat")) { const QString hat_str = QString::fromStdString(param.Get("hat", "")); @@ -126,6 +140,25 @@ static QString AnalogToText(const Common::ParamPackage& param, const std::string return {}; } + if (param.Get("engine", "") == "gcpad") { + if (dir == "modifier") { + return QObject::tr("[unused]"); + } + + if (dir == "left" || dir == "right") { + const QString axis_x_str = QString::fromStdString(param.Get("axis_x", "")); + + return QObject::tr("GC Axis %1").arg(axis_x_str); + } + + if (dir == "up" || dir == "down") { + const QString axis_y_str = QString::fromStdString(param.Get("axis_y", "")); + + return QObject::tr("GC Axis %1").arg(axis_y_str); + } + + return {}; + } return QObject::tr("[unknown]"); } @@ -332,7 +365,8 @@ ConfigureInputPlayer::ConfigureInputPlayer(QWidget* parent, std::size_t player_i connect(analog_map_deadzone_and_modifier_slider[analog_id], &QSlider::valueChanged, [=] { const float slider_value = analog_map_deadzone_and_modifier_slider[analog_id]->value(); - if (analogs_param[analog_id].Get("engine", "") == "sdl") { + if (analogs_param[analog_id].Get("engine", "") == "sdl" || + analogs_param[analog_id].Get("engine", "") == "gcpad") { analog_map_deadzone_and_modifier_slider_label[analog_id]->setText( tr("Deadzone: %1%").arg(slider_value)); analogs_param[analog_id].Set("deadzone", slider_value / 100.0f); @@ -352,6 +386,20 @@ ConfigureInputPlayer::ConfigureInputPlayer(QWidget* parent, std::size_t player_i connect(poll_timer.get(), &QTimer::timeout, [this] { Common::ParamPackage params; + if (InputCommon::GetGCButtons()->IsPolling()) { + params = InputCommon::GetGCButtons()->GetNextInput(); + if (params.Has("engine")) { + SetPollingResult(params, false); + return; + } + } + if (InputCommon::GetGCAnalogs()->IsPolling()) { + params = InputCommon::GetGCAnalogs()->GetNextInput(); + if (params.Has("engine")) { + SetPollingResult(params, false); + return; + } + } for (auto& poller : device_pollers) { params = poller->GetNextInput(); if (params.Has("engine")) { @@ -534,7 +582,7 @@ void ConfigureInputPlayer::UpdateButtonLabels() { analog_map_deadzone_and_modifier_slider_label[analog_id]; if (param.Has("engine")) { - if (param.Get("engine", "") == "sdl") { + if (param.Get("engine", "") == "sdl" || param.Get("engine", "") == "gcpad") { if (!param.Has("deadzone")) { param.Set("deadzone", 0.1f); } @@ -583,6 +631,11 @@ void ConfigureInputPlayer::HandleClick( grabKeyboard(); grabMouse(); + if (type == InputCommon::Polling::DeviceType::Button) { + InputCommon::GetGCButtons()->BeginConfiguration(); + } else { + InputCommon::GetGCAnalogs()->BeginConfiguration(); + } timeout_timer->start(5000); // Cancel after 5 seconds poll_timer->start(200); // Check for new inputs every 200ms } @@ -596,6 +649,9 @@ void ConfigureInputPlayer::SetPollingResult(const Common::ParamPackage& params, poller->Stop(); } + InputCommon::GetGCButtons()->EndConfiguration(); + InputCommon::GetGCAnalogs()->EndConfiguration(); + if (!abort) { (*input_setter)(params); } diff --git a/src/yuzu/main.cpp b/src/yuzu/main.cpp index fb299a39b..9844e4764 100644 --- a/src/yuzu/main.cpp +++ b/src/yuzu/main.cpp @@ -753,7 +753,7 @@ void GMainWindow::InitializeHotkeys() { }); connect(hotkey_registry.GetHotkey(main_window, QStringLiteral("Capture Screenshot"), this), &QShortcut::activated, this, [&] { - if (emu_thread->IsRunning()) { + if (emu_thread != nullptr && emu_thread->IsRunning()) { OnCaptureScreenshot(); } }); |