diff options
Diffstat (limited to 'src')
59 files changed, 1602 insertions, 1150 deletions
diff --git a/src/common/logging/backend.cpp b/src/common/logging/backend.cpp index 5753b871a..12f6d0114 100644 --- a/src/common/logging/backend.cpp +++ b/src/common/logging/backend.cpp @@ -13,7 +13,7 @@ #include <vector> #ifdef _WIN32 #include <share.h> // For _SH_DENYWR -#include <windows.h> // For OutputDebugStringA +#include <windows.h> // For OutputDebugStringW #else #define _SH_DENYWR 0 #endif @@ -148,7 +148,7 @@ void FileBackend::Write(const Entry& entry) { void DebuggerBackend::Write(const Entry& entry) { #ifdef _WIN32 - ::OutputDebugStringA(FormatLogMessage(entry).append(1, '\n').c_str()); + ::OutputDebugStringW(Common::UTF8ToUTF16W(FormatLogMessage(entry).append(1, '\n')).c_str()); #endif } diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt index 73aec8ab0..882c9ab59 100644 --- a/src/core/CMakeLists.txt +++ b/src/core/CMakeLists.txt @@ -63,6 +63,10 @@ add_library(core STATIC file_sys/sdmc_factory.h file_sys/submission_package.cpp file_sys/submission_package.h + file_sys/system_archive/ng_word.cpp + file_sys/system_archive/ng_word.h + file_sys/system_archive/system_archive.cpp + file_sys/system_archive/system_archive.h file_sys/vfs.cpp file_sys/vfs.h file_sys/vfs_concat.cpp diff --git a/src/core/crypto/key_manager.cpp b/src/core/crypto/key_manager.cpp index ed231927b..ca12fb4ab 100644 --- a/src/core/crypto/key_manager.cpp +++ b/src/core/crypto/key_manager.cpp @@ -246,7 +246,6 @@ std::vector<TicketRaw> GetTicketblob(const FileUtil::IOFile& ticket_save) { } std::vector<TicketRaw> out; - u32 magic{}; for (std::size_t offset = 0; offset + 0x4 < buffer.size(); ++offset) { if (buffer[offset] == 0x4 && buffer[offset + 1] == 0x0 && buffer[offset + 2] == 0x1 && buffer[offset + 3] == 0x0) { diff --git a/src/core/file_sys/registered_cache.cpp b/src/core/file_sys/registered_cache.cpp index 07c3af64a..128199063 100644 --- a/src/core/file_sys/registered_cache.cpp +++ b/src/core/file_sys/registered_cache.cpp @@ -107,42 +107,41 @@ static ContentRecordType GetCRTypeFromNCAType(NCAContentType type) { VirtualFile RegisteredCache::OpenFileOrDirectoryConcat(const VirtualDir& dir, std::string_view path) const { const auto file = dir->GetFileRelative(path); - if (file != nullptr) + if (file != nullptr) { return file; + } const auto nca_dir = dir->GetDirectoryRelative(path); - if (nca_dir != nullptr) { - const auto nca_dir = dir->GetDirectoryRelative(path); - VirtualFile file = nullptr; + if (nca_dir == nullptr) { + return nullptr; + } - const auto files = nca_dir->GetFiles(); - if (files.size() == 1 && files[0]->GetName() == "00") { - file = files[0]; + const auto files = nca_dir->GetFiles(); + if (files.size() == 1 && files[0]->GetName() == "00") { + return files[0]; + } + + std::vector<VirtualFile> concat; + // Since the files are a two-digit hex number, max is FF. + for (std::size_t i = 0; i < 0x100; ++i) { + auto next = nca_dir->GetFile(fmt::format("{:02X}", i)); + if (next != nullptr) { + concat.push_back(std::move(next)); } else { - std::vector<VirtualFile> concat; - // Since the files are a two-digit hex number, max is FF. - for (std::size_t i = 0; i < 0x100; ++i) { - auto next = nca_dir->GetFile(fmt::format("{:02X}", i)); - if (next != nullptr) { - concat.push_back(std::move(next)); - } else { - next = nca_dir->GetFile(fmt::format("{:02x}", i)); - if (next != nullptr) - concat.push_back(std::move(next)); - else - break; - } + next = nca_dir->GetFile(fmt::format("{:02x}", i)); + if (next != nullptr) { + concat.push_back(std::move(next)); + } else { + break; } - - if (concat.empty()) - return nullptr; - - file = ConcatenatedVfsFile::MakeConcatenatedFile(concat, concat.front()->GetName()); } + } - return file; + if (concat.empty()) { + return nullptr; } - return nullptr; + + return ConcatenatedVfsFile::MakeConcatenatedFile(concat, concat.front()->GetName()); } VirtualFile RegisteredCache::GetFileAtID(NcaID id) const { diff --git a/src/core/file_sys/savedata_factory.cpp b/src/core/file_sys/savedata_factory.cpp index 5434f2149..bd50fedc7 100644 --- a/src/core/file_sys/savedata_factory.cpp +++ b/src/core/file_sys/savedata_factory.cpp @@ -18,7 +18,11 @@ std::string SaveDataDescriptor::DebugInfo() const { static_cast<u8>(type), title_id, user_id[1], user_id[0], save_id); } -SaveDataFactory::SaveDataFactory(VirtualDir save_directory) : dir(std::move(save_directory)) {} +SaveDataFactory::SaveDataFactory(VirtualDir save_directory) : dir(std::move(save_directory)) { + // Delete all temporary storages + // On hardware, it is expected that temporary storage be empty at first use. + dir->DeleteSubdirectoryRecursive("temp"); +} SaveDataFactory::~SaveDataFactory() = default; @@ -120,6 +124,8 @@ std::string SaveDataFactory::GetFullPath(SaveDataSpaceId space, SaveDataType typ case SaveDataType::TemporaryStorage: return fmt::format("{}{:016X}/{:016X}{:016X}/{:016X}", out, 0, user_id[1], user_id[0], title_id); + case SaveDataType::CacheStorage: + return fmt::format("{}save/cache/{:016X}", out, title_id); default: ASSERT_MSG(false, "Unrecognized SaveDataType: {:02X}", static_cast<u8>(type)); } diff --git a/src/core/file_sys/savedata_factory.h b/src/core/file_sys/savedata_factory.h index 2a0088040..bd4919610 100644 --- a/src/core/file_sys/savedata_factory.h +++ b/src/core/file_sys/savedata_factory.h @@ -17,8 +17,10 @@ namespace FileSys { enum class SaveDataSpaceId : u8 { NandSystem = 0, NandUser = 1, - SdCard = 2, + SdCardSystem = 2, TemporaryStorage = 3, + SdCardUser = 4, + ProperSystem = 100, }; enum class SaveDataType : u8 { diff --git a/src/core/file_sys/system_archive/ng_word.cpp b/src/core/file_sys/system_archive/ng_word.cpp new file mode 100644 index 000000000..f4443784d --- /dev/null +++ b/src/core/file_sys/system_archive/ng_word.cpp @@ -0,0 +1,81 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <fmt/format.h> +#include "common/common_types.h" +#include "core/file_sys/system_archive/ng_word.h" +#include "core/file_sys/vfs_vector.h" + +namespace FileSys::SystemArchive { + +namespace NgWord1Data { + +constexpr std::size_t NUMBER_WORD_TXT_FILES = 0x10; + +// Should this archive replacement mysteriously not work on a future game, consider updating. +constexpr std::array<u8, 4> VERSION_DAT{0x0, 0x0, 0x0, 0x19}; // 5.1.0 System Version + +constexpr std::array<u8, 30> WORD_TXT{ + 0xFE, 0xFF, 0x00, 0x5E, 0x00, 0x76, 0x00, 0x65, 0x00, 0x72, 0x00, 0x79, 0x00, 0x62, 0x00, + 0x61, 0x00, 0x64, 0x00, 0x77, 0x00, 0x6F, 0x00, 0x72, 0x00, 0x64, 0x00, 0x24, 0x00, 0x0A, +}; // "^verybadword$" in UTF-16 + +} // namespace NgWord1Data + +VirtualDir NgWord1() { + std::vector<VirtualFile> files(NgWord1Data::NUMBER_WORD_TXT_FILES); + + for (std::size_t i = 0; i < files.size(); ++i) { + files[i] = std::make_shared<ArrayVfsFile<NgWord1Data::WORD_TXT.size()>>( + NgWord1Data::WORD_TXT, fmt::format("{}.txt", i)); + } + + files.push_back(std::make_shared<ArrayVfsFile<NgWord1Data::WORD_TXT.size()>>( + NgWord1Data::WORD_TXT, "common.txt")); + files.push_back(std::make_shared<ArrayVfsFile<NgWord1Data::VERSION_DAT.size()>>( + NgWord1Data::VERSION_DAT, "version.dat")); + + return std::make_shared<VectorVfsDirectory>(files, std::vector<VirtualDir>{}, "data"); +} + +namespace NgWord2Data { + +constexpr std::size_t NUMBER_AC_NX_FILES = 0x10; + +// Should this archive replacement mysteriously not work on a future game, consider updating. +constexpr std::array<u8, 4> VERSION_DAT{0x0, 0x0, 0x0, 0x15}; // 5.1.0 System Version + +constexpr std::array<u8, 0x2C> AC_NX_DATA{ + 0x1F, 0x8B, 0x08, 0x08, 0xD5, 0x2C, 0x09, 0x5C, 0x04, 0x00, 0x61, 0x63, 0x72, 0x61, 0x77, + 0x00, 0xED, 0xC1, 0x01, 0x0D, 0x00, 0x00, 0x00, 0xC2, 0x20, 0xFB, 0xA7, 0xB6, 0xC7, 0x07, + 0x0C, 0x00, 0x00, 0x00, 0xC8, 0x3B, 0x11, 0x00, 0x1C, 0xC7, 0x00, 0x10, 0x00, 0x00, +}; // Deserializes to no bad words + +} // namespace NgWord2Data + +VirtualDir NgWord2() { + std::vector<VirtualFile> files(NgWord2Data::NUMBER_AC_NX_FILES * 3); + + for (std::size_t i = 0; i < NgWord2Data::NUMBER_AC_NX_FILES; ++i) { + files[3 * i] = std::make_shared<ArrayVfsFile<NgWord2Data::AC_NX_DATA.size()>>( + NgWord2Data::AC_NX_DATA, fmt::format("ac_{}_b1_nx", i)); + files[3 * i + 1] = std::make_shared<ArrayVfsFile<NgWord2Data::AC_NX_DATA.size()>>( + NgWord2Data::AC_NX_DATA, fmt::format("ac_{}_b2_nx", i)); + files[3 * i + 2] = std::make_shared<ArrayVfsFile<NgWord2Data::AC_NX_DATA.size()>>( + NgWord2Data::AC_NX_DATA, fmt::format("ac_{}_not_b_nx", i)); + } + + files.push_back(std::make_shared<ArrayVfsFile<NgWord2Data::AC_NX_DATA.size()>>( + NgWord2Data::AC_NX_DATA, "ac_common_b1_nx")); + files.push_back(std::make_shared<ArrayVfsFile<NgWord2Data::AC_NX_DATA.size()>>( + NgWord2Data::AC_NX_DATA, "ac_common_b2_nx")); + files.push_back(std::make_shared<ArrayVfsFile<NgWord2Data::AC_NX_DATA.size()>>( + NgWord2Data::AC_NX_DATA, "ac_common_not_b_nx")); + files.push_back(std::make_shared<ArrayVfsFile<NgWord2Data::VERSION_DAT.size()>>( + NgWord2Data::VERSION_DAT, "version.dat")); + + return std::make_shared<VectorVfsDirectory>(files, std::vector<VirtualDir>{}, "data"); +} + +} // namespace FileSys::SystemArchive diff --git a/src/core/file_sys/system_archive/ng_word.h b/src/core/file_sys/system_archive/ng_word.h new file mode 100644 index 000000000..cd81e0abb --- /dev/null +++ b/src/core/file_sys/system_archive/ng_word.h @@ -0,0 +1,14 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/file_sys/vfs_types.h" + +namespace FileSys::SystemArchive { + +VirtualDir NgWord1(); +VirtualDir NgWord2(); + +} // namespace FileSys::SystemArchive diff --git a/src/core/file_sys/system_archive/system_archive.cpp b/src/core/file_sys/system_archive/system_archive.cpp new file mode 100644 index 000000000..e3e79f40a --- /dev/null +++ b/src/core/file_sys/system_archive/system_archive.cpp @@ -0,0 +1,90 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "common/logging/log.h" +#include "core/file_sys/romfs.h" +#include "core/file_sys/system_archive/ng_word.h" +#include "core/file_sys/system_archive/system_archive.h" + +namespace FileSys::SystemArchive { + +constexpr u64 SYSTEM_ARCHIVE_BASE_TITLE_ID = 0x0100000000000800; +constexpr std::size_t SYSTEM_ARCHIVE_COUNT = 0x28; + +using SystemArchiveSupplier = VirtualDir (*)(); + +struct SystemArchiveDescriptor { + u64 title_id; + const char* name; + SystemArchiveSupplier supplier; +}; + +constexpr std::array<SystemArchiveDescriptor, SYSTEM_ARCHIVE_COUNT> SYSTEM_ARCHIVES{{ + {0x0100000000000800, "CertStore", nullptr}, + {0x0100000000000801, "ErrorMessage", nullptr}, + {0x0100000000000802, "MiiModel", nullptr}, + {0x0100000000000803, "BrowserDll", nullptr}, + {0x0100000000000804, "Help", nullptr}, + {0x0100000000000805, "SharedFont", nullptr}, + {0x0100000000000806, "NgWord", &NgWord1}, + {0x0100000000000807, "SsidList", nullptr}, + {0x0100000000000808, "Dictionary", nullptr}, + {0x0100000000000809, "SystemVersion", nullptr}, + {0x010000000000080A, "AvatarImage", nullptr}, + {0x010000000000080B, "LocalNews", nullptr}, + {0x010000000000080C, "Eula", nullptr}, + {0x010000000000080D, "UrlBlackList", nullptr}, + {0x010000000000080E, "TimeZoneBinary", nullptr}, + {0x010000000000080F, "CertStoreCruiser", nullptr}, + {0x0100000000000810, "FontNintendoExtension", nullptr}, + {0x0100000000000811, "FontStandard", nullptr}, + {0x0100000000000812, "FontKorean", nullptr}, + {0x0100000000000813, "FontChineseTraditional", nullptr}, + {0x0100000000000814, "FontChineseSimple", nullptr}, + {0x0100000000000815, "FontBfcpx", nullptr}, + {0x0100000000000816, "SystemUpdate", nullptr}, + {0x0100000000000817, "0100000000000817", nullptr}, + {0x0100000000000818, "FirmwareDebugSettings", nullptr}, + {0x0100000000000819, "BootImagePackage", nullptr}, + {0x010000000000081A, "BootImagePackageSafe", nullptr}, + {0x010000000000081B, "BootImagePackageExFat", nullptr}, + {0x010000000000081C, "BootImagePackageExFatSafe", nullptr}, + {0x010000000000081D, "FatalMessage", nullptr}, + {0x010000000000081E, "ControllerIcon", nullptr}, + {0x010000000000081F, "PlatformConfigIcosa", nullptr}, + {0x0100000000000820, "PlatformConfigCopper", nullptr}, + {0x0100000000000821, "PlatformConfigHoag", nullptr}, + {0x0100000000000822, "ControllerFirmware", nullptr}, + {0x0100000000000823, "NgWord2", &NgWord2}, + {0x0100000000000824, "PlatformConfigIcosaMariko", nullptr}, + {0x0100000000000825, "ApplicationBlackList", nullptr}, + {0x0100000000000826, "RebootlessSystemUpdateVersion", nullptr}, + {0x0100000000000827, "ContentActionTable", nullptr}, +}}; + +VirtualFile SynthesizeSystemArchive(const u64 title_id) { + if (title_id < SYSTEM_ARCHIVES.front().title_id || title_id > SYSTEM_ARCHIVES.back().title_id) + return nullptr; + + const auto& desc = SYSTEM_ARCHIVES[title_id - SYSTEM_ARCHIVE_BASE_TITLE_ID]; + + LOG_INFO(Service_FS, "Synthesizing system archive '{}' (0x{:016X}).", desc.name, desc.title_id); + + if (desc.supplier == nullptr) + return nullptr; + + const auto dir = desc.supplier(); + + if (dir == nullptr) + return nullptr; + + const auto romfs = CreateRomFS(dir); + + if (romfs == nullptr) + return nullptr; + + LOG_INFO(Service_FS, " - System archive generation successful!"); + return romfs; +} +} // namespace FileSys::SystemArchive diff --git a/src/core/file_sys/system_archive/system_archive.h b/src/core/file_sys/system_archive/system_archive.h new file mode 100644 index 000000000..724a8eb17 --- /dev/null +++ b/src/core/file_sys/system_archive/system_archive.h @@ -0,0 +1,14 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "common/common_types.h" +#include "core/file_sys/vfs_types.h" + +namespace FileSys::SystemArchive { + +VirtualFile SynthesizeSystemArchive(u64 title_id); + +} // namespace FileSys::SystemArchive diff --git a/src/core/file_sys/vfs_vector.cpp b/src/core/file_sys/vfs_vector.cpp index 808f31e81..515626658 100644 --- a/src/core/file_sys/vfs_vector.cpp +++ b/src/core/file_sys/vfs_vector.cpp @@ -3,7 +3,6 @@ // Refer to the license.txt file included. #include <algorithm> -#include <cstring> #include <utility> #include "core/file_sys/vfs_vector.h" diff --git a/src/core/file_sys/vfs_vector.h b/src/core/file_sys/vfs_vector.h index 3e3f790c3..ac36cb2ee 100644 --- a/src/core/file_sys/vfs_vector.h +++ b/src/core/file_sys/vfs_vector.h @@ -4,10 +4,63 @@ #pragma once +#include <cstring> #include "core/file_sys/vfs.h" namespace FileSys { +// An implementation of VfsFile that is backed by a statically-sized array +template <std::size_t size> +class ArrayVfsFile : public VfsFile { +public: + ArrayVfsFile(std::array<u8, size> data, std::string name = "", VirtualDir parent = nullptr) + : data(data), name(std::move(name)), parent(std::move(parent)) {} + + std::string GetName() const override { + return name; + } + + std::size_t GetSize() const override { + return size; + } + + bool Resize(std::size_t new_size) override { + return false; + } + + std::shared_ptr<VfsDirectory> GetContainingDirectory() const override { + return parent; + } + + bool IsWritable() const override { + return false; + } + + bool IsReadable() const override { + return true; + } + + std::size_t Read(u8* data_, std::size_t length, std::size_t offset) const override { + const auto read = std::min(length, size - offset); + std::memcpy(data_, data.data() + offset, read); + return read; + } + + std::size_t Write(const u8* data, std::size_t length, std::size_t offset) override { + return 0; + } + + bool Rename(std::string_view name) override { + this->name = name; + return true; + } + +private: + std::array<u8, size> data; + std::string name; + VirtualDir parent; +}; + // An implementation of VfsFile that is backed by a vector optionally supplied upon construction class VectorVfsFile : public VfsFile { public: diff --git a/src/core/hle/kernel/handle_table.cpp b/src/core/hle/kernel/handle_table.cpp index 1bf79b692..c8acde5b1 100644 --- a/src/core/hle/kernel/handle_table.cpp +++ b/src/core/hle/kernel/handle_table.cpp @@ -42,9 +42,10 @@ ResultVal<Handle> HandleTable::Create(SharedPtr<Object> obj) { u16 generation = next_generation++; // Overflow count so it fits in the 15 bits dedicated to the generation in the handle. - // CTR-OS doesn't use generation 0, so skip straight to 1. - if (next_generation >= (1 << 15)) + // Horizon OS uses zero to represent an invalid handle, so skip to 1. + if (next_generation >= (1 << 15)) { next_generation = 1; + } generations[slot] = generation; objects[slot] = std::move(obj); diff --git a/src/core/hle/kernel/handle_table.h b/src/core/hle/kernel/handle_table.h index e3f3e3fb8..6b7927fd8 100644 --- a/src/core/hle/kernel/handle_table.h +++ b/src/core/hle/kernel/handle_table.h @@ -13,6 +13,7 @@ namespace Kernel { enum KernelHandle : Handle { + InvalidHandle = 0, CurrentThread = 0xFFFF8000, CurrentProcess = 0xFFFF8001, }; diff --git a/src/core/hle/kernel/object.cpp b/src/core/hle/kernel/object.cpp index d87a62bb9..0ea851a74 100644 --- a/src/core/hle/kernel/object.cpp +++ b/src/core/hle/kernel/object.cpp @@ -13,16 +13,17 @@ Object::~Object() = default; bool Object::IsWaitable() const { switch (GetHandleType()) { - case HandleType::Event: + case HandleType::ReadableEvent: case HandleType::Thread: + case HandleType::Process: case HandleType::Timer: case HandleType::ServerPort: case HandleType::ServerSession: return true; case HandleType::Unknown: + case HandleType::WritableEvent: case HandleType::SharedMemory: - case HandleType::Process: case HandleType::AddressArbiter: case HandleType::ResourceLimit: case HandleType::ClientPort: diff --git a/src/core/hle/kernel/object.h b/src/core/hle/kernel/object.h index 69082ce3e..f1606a204 100644 --- a/src/core/hle/kernel/object.h +++ b/src/core/hle/kernel/object.h @@ -19,7 +19,8 @@ using Handle = u32; enum class HandleType : u32 { Unknown, - Event, + WritableEvent, + ReadableEvent, SharedMemory, Thread, Process, diff --git a/src/core/hle/kernel/process.cpp b/src/core/hle/kernel/process.cpp index 7ca538401..c817fb449 100644 --- a/src/core/hle/kernel/process.cpp +++ b/src/core/hle/kernel/process.cpp @@ -9,6 +9,7 @@ #include "common/logging/log.h" #include "core/core.h" #include "core/file_sys/program_metadata.h" +#include "core/hle/kernel/errors.h" #include "core/hle/kernel/kernel.h" #include "core/hle/kernel/process.h" #include "core/hle/kernel/resource_limit.h" @@ -44,8 +45,28 @@ SharedPtr<Process> Process::Create(KernelCore& kernel, std::string&& name) { return process; } +SharedPtr<ResourceLimit> Process::GetResourceLimit() const { + return resource_limit; +} + +ResultCode Process::ClearSignalState() { + if (status == ProcessStatus::Exited) { + LOG_ERROR(Kernel, "called on a terminated process instance."); + return ERR_INVALID_STATE; + } + + if (!is_signaled) { + LOG_ERROR(Kernel, "called on a process instance that isn't signaled."); + return ERR_INVALID_STATE; + } + + is_signaled = false; + return RESULT_SUCCESS; +} + void Process::LoadFromMetadata(const FileSys::ProgramMetadata& metadata) { program_id = metadata.GetTitleID(); + ideal_processor = metadata.GetMainThreadCore(); is_64bit_process = metadata.Is64BitProgram(); vm_manager.Reset(metadata.GetAddressSpaceType()); } @@ -133,13 +154,13 @@ void Process::Run(VAddr entry_point, s32 main_thread_priority, u32 stack_size) { .Unwrap(); vm_manager.LogLayout(); - status = ProcessStatus::Running; + ChangeStatus(ProcessStatus::Running); Kernel::SetupMainThread(kernel, entry_point, main_thread_priority, *this); } void Process::PrepareForTermination() { - status = ProcessStatus::Exited; + ChangeStatus(ProcessStatus::Exiting); const auto stop_threads = [this](const std::vector<SharedPtr<Thread>>& thread_list) { for (auto& thread : thread_list) { @@ -163,6 +184,8 @@ void Process::PrepareForTermination() { stop_threads(system.Scheduler(1).GetThreadList()); stop_threads(system.Scheduler(2).GetThreadList()); stop_threads(system.Scheduler(3).GetThreadList()); + + ChangeStatus(ProcessStatus::Exited); } /** @@ -261,7 +284,25 @@ ResultCode Process::UnmapMemory(VAddr dst_addr, VAddr /*src_addr*/, u64 size) { return vm_manager.UnmapRange(dst_addr, size); } -Kernel::Process::Process(KernelCore& kernel) : Object{kernel} {} +Kernel::Process::Process(KernelCore& kernel) : WaitObject{kernel} {} Kernel::Process::~Process() {} +void Process::Acquire(Thread* thread) { + ASSERT_MSG(!ShouldWait(thread), "Object unavailable!"); +} + +bool Process::ShouldWait(Thread* thread) const { + return !is_signaled; +} + +void Process::ChangeStatus(ProcessStatus new_status) { + if (status == new_status) { + return; + } + + status = new_status; + is_signaled = true; + WakeupAllWaitingThreads(); +} + } // namespace Kernel diff --git a/src/core/hle/kernel/process.h b/src/core/hle/kernel/process.h index ada845c7f..bcb9ac4b8 100644 --- a/src/core/hle/kernel/process.h +++ b/src/core/hle/kernel/process.h @@ -14,9 +14,10 @@ #include "common/bit_field.h" #include "common/common_types.h" #include "core/hle/kernel/handle_table.h" -#include "core/hle/kernel/object.h" #include "core/hle/kernel/thread.h" #include "core/hle/kernel/vm_manager.h" +#include "core/hle/kernel/wait_object.h" +#include "core/hle/result.h" namespace FileSys { class ProgramMetadata; @@ -117,7 +118,7 @@ struct CodeSet final { VAddr entrypoint = 0; }; -class Process final : public Object { +class Process final : public WaitObject { public: static constexpr std::size_t RANDOM_ENTROPY_SIZE = 4; @@ -171,14 +172,7 @@ public: } /// Gets the resource limit descriptor for this process - ResourceLimit& GetResourceLimit() { - return *resource_limit; - } - - /// Gets the resource limit descriptor for this process - const ResourceLimit& GetResourceLimit() const { - return *resource_limit; - } + SharedPtr<ResourceLimit> GetResourceLimit() const; /// Gets the default CPU ID for this process u8 GetDefaultProcessorID() const { @@ -219,6 +213,16 @@ public: return random_entropy.at(index); } + /// Clears the signaled state of the process if and only if it's signaled. + /// + /// @pre The process must not be already terminated. If this is called on a + /// terminated process, then ERR_INVALID_STATE will be returned. + /// + /// @pre The process must be in a signaled state. If this is called on a + /// process instance that is not signaled, ERR_INVALID_STATE will be + /// returned. + ResultCode ClearSignalState(); + /** * Loads process-specifics configuration info with metadata provided * by an executable. @@ -267,6 +271,17 @@ private: explicit Process(KernelCore& kernel); ~Process() override; + /// Checks if the specified thread should wait until this process is available. + bool ShouldWait(Thread* thread) const override; + + /// Acquires/locks this process for the specified thread if it's available. + void Acquire(Thread* thread) override; + + /// Changes the process status. If the status is different + /// from the current process status, then this will trigger + /// a process signal. + void ChangeStatus(ProcessStatus new_status); + /// Memory manager for this process. Kernel::VMManager vm_manager; @@ -312,6 +327,10 @@ private: /// specified by metadata provided to the process during loading. bool is_64bit_process = true; + /// Whether or not this process is signaled. This occurs + /// upon the process changing to a different state. + bool is_signaled = false; + /// Total running time for the process in ticks. u64 total_process_running_time_ticks = 0; diff --git a/src/core/hle/kernel/readable_event.cpp b/src/core/hle/kernel/readable_event.cpp index 92e16b4e6..ba01f495c 100644 --- a/src/core/hle/kernel/readable_event.cpp +++ b/src/core/hle/kernel/readable_event.cpp @@ -4,10 +4,10 @@ #include <algorithm> #include "common/assert.h" +#include "core/hle/kernel/errors.h" #include "core/hle/kernel/object.h" #include "core/hle/kernel/readable_event.h" #include "core/hle/kernel/thread.h" -#include "core/hle/kernel/writable_event.h" namespace Kernel { @@ -34,6 +34,16 @@ void ReadableEvent::Clear() { signaled = false; } +ResultCode ReadableEvent::Reset() { + if (!signaled) { + return ERR_INVALID_STATE; + } + + Clear(); + + return RESULT_SUCCESS; +} + void ReadableEvent::WakeupAllWaitingThreads() { WaitObject::WakeupAllWaitingThreads(); diff --git a/src/core/hle/kernel/readable_event.h b/src/core/hle/kernel/readable_event.h index b1f1f4871..80b3b0aba 100644 --- a/src/core/hle/kernel/readable_event.h +++ b/src/core/hle/kernel/readable_event.h @@ -7,6 +7,8 @@ #include "core/hle/kernel/object.h" #include "core/hle/kernel/wait_object.h" +union ResultCode; + namespace Kernel { class KernelCore; @@ -29,7 +31,7 @@ public: return reset_type; } - static const HandleType HANDLE_TYPE = HandleType::Event; + static const HandleType HANDLE_TYPE = HandleType::ReadableEvent; HandleType GetHandleType() const override { return HANDLE_TYPE; } @@ -39,8 +41,17 @@ public: void WakeupAllWaitingThreads() override; + /// Unconditionally clears the readable event's state. void Clear(); + /// Clears the readable event's state if and only if it + /// has already been signaled. + /// + /// @pre The event must be in a signaled state. If this event + /// is in an unsignaled state and this function is called, + /// then ERR_INVALID_STATE will be returned. + ResultCode Reset(); + private: explicit ReadableEvent(KernelCore& kernel); diff --git a/src/core/hle/kernel/shared_memory.cpp b/src/core/hle/kernel/shared_memory.cpp index 0494581f5..d1ca60125 100644 --- a/src/core/hle/kernel/shared_memory.cpp +++ b/src/core/hle/kernel/shared_memory.cpp @@ -39,15 +39,15 @@ SharedPtr<SharedMemory> SharedMemory::Create(KernelCore& kernel, SharedPtr<Proce shared_memory->backing_block.get()); } } else { - auto& vm_manager = shared_memory->owner_process->VMManager(); + const auto& vm_manager = shared_memory->owner_process->VMManager(); // The memory is already available and mapped in the owner process. - auto vma = vm_manager.FindVMA(address); - ASSERT_MSG(vma != vm_manager.vma_map.end(), "Invalid memory address"); + const auto vma = vm_manager.FindVMA(address); + ASSERT_MSG(vm_manager.IsValidHandle(vma), "Invalid memory address"); ASSERT_MSG(vma->second.backing_block, "Backing block doesn't exist for address"); // The returned VMA might be a bigger one encompassing the desired address. - auto vma_offset = address - vma->first; + const auto vma_offset = address - vma->first; ASSERT_MSG(vma_offset + size <= vma->second.size, "Shared memory exceeds bounds of mapped block"); diff --git a/src/core/hle/kernel/svc.cpp b/src/core/hle/kernel/svc.cpp index 051b09d00..f43c7201c 100644 --- a/src/core/hle/kernel/svc.cpp +++ b/src/core/hle/kernel/svc.cpp @@ -35,7 +35,6 @@ #include "core/hle/lock.h" #include "core/hle/result.h" #include "core/hle/service/service.h" -#include "core/settings.h" namespace Kernel { namespace { @@ -240,7 +239,7 @@ static ResultCode SetMemoryPermission(VAddr addr, u64 size, u32 prot) { } const VMManager::VMAHandle iter = vm_manager.FindVMA(addr); - if (iter == vm_manager.vma_map.end()) { + if (!vm_manager.IsValidHandle(iter)) { LOG_ERROR(Kernel_SVC, "Unable to find VMA for address=0x{:016X}", addr); return ERR_INVALID_ADDRESS_STATE; } @@ -663,7 +662,7 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id) TotalMemoryUsage = 6, TotalHeapUsage = 7, IsCurrentProcessBeingDebugged = 8, - ResourceHandleLimit = 9, + RegisterResourceLimit = 9, IdleTickCount = 10, RandomEntropy = 11, PerformanceCounter = 0xF0000002, @@ -683,37 +682,137 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id) ThreadTickCount = 0xF0000002, }; - const auto* current_process = Core::CurrentProcess(); - const auto& vm_manager = current_process->VMManager(); + const auto info_id_type = static_cast<GetInfoType>(info_id); - switch (static_cast<GetInfoType>(info_id)) { + switch (info_id_type) { case GetInfoType::AllowedCpuIdBitmask: - *result = current_process->GetAllowedProcessorMask(); - break; case GetInfoType::AllowedThreadPrioBitmask: - *result = current_process->GetAllowedThreadPriorityMask(); - break; case GetInfoType::MapRegionBaseAddr: - *result = vm_manager.GetMapRegionBaseAddress(); - break; case GetInfoType::MapRegionSize: - *result = vm_manager.GetMapRegionSize(); - break; case GetInfoType::HeapRegionBaseAddr: - *result = vm_manager.GetHeapRegionBaseAddress(); - break; case GetInfoType::HeapRegionSize: - *result = vm_manager.GetHeapRegionSize(); - break; + case GetInfoType::ASLRRegionBaseAddr: + case GetInfoType::ASLRRegionSize: + case GetInfoType::NewMapRegionBaseAddr: + case GetInfoType::NewMapRegionSize: case GetInfoType::TotalMemoryUsage: - *result = vm_manager.GetTotalMemoryUsage(); - break; case GetInfoType::TotalHeapUsage: - *result = vm_manager.GetTotalHeapUsage(); - break; + case GetInfoType::IsVirtualAddressMemoryEnabled: + case GetInfoType::PersonalMmHeapUsage: + case GetInfoType::TitleId: + case GetInfoType::UserExceptionContextAddr: { + if (info_sub_id != 0) { + return ERR_INVALID_ENUM_VALUE; + } + + const auto& current_process_handle_table = Core::CurrentProcess()->GetHandleTable(); + const auto process = current_process_handle_table.Get<Process>(static_cast<Handle>(handle)); + if (!process) { + return ERR_INVALID_HANDLE; + } + + switch (info_id_type) { + case GetInfoType::AllowedCpuIdBitmask: + *result = process->GetAllowedProcessorMask(); + return RESULT_SUCCESS; + + case GetInfoType::AllowedThreadPrioBitmask: + *result = process->GetAllowedThreadPriorityMask(); + return RESULT_SUCCESS; + + case GetInfoType::MapRegionBaseAddr: + *result = process->VMManager().GetMapRegionBaseAddress(); + return RESULT_SUCCESS; + + case GetInfoType::MapRegionSize: + *result = process->VMManager().GetMapRegionSize(); + return RESULT_SUCCESS; + + case GetInfoType::HeapRegionBaseAddr: + *result = process->VMManager().GetHeapRegionBaseAddress(); + return RESULT_SUCCESS; + + case GetInfoType::HeapRegionSize: + *result = process->VMManager().GetHeapRegionSize(); + return RESULT_SUCCESS; + + case GetInfoType::ASLRRegionBaseAddr: + *result = process->VMManager().GetASLRRegionBaseAddress(); + return RESULT_SUCCESS; + + case GetInfoType::ASLRRegionSize: + *result = process->VMManager().GetASLRRegionSize(); + return RESULT_SUCCESS; + + case GetInfoType::NewMapRegionBaseAddr: + *result = process->VMManager().GetNewMapRegionBaseAddress(); + return RESULT_SUCCESS; + + case GetInfoType::NewMapRegionSize: + *result = process->VMManager().GetNewMapRegionSize(); + return RESULT_SUCCESS; + + case GetInfoType::TotalMemoryUsage: + *result = process->VMManager().GetTotalMemoryUsage(); + return RESULT_SUCCESS; + + case GetInfoType::TotalHeapUsage: + *result = process->VMManager().GetTotalHeapUsage(); + return RESULT_SUCCESS; + + case GetInfoType::IsVirtualAddressMemoryEnabled: + *result = process->IsVirtualMemoryEnabled(); + return RESULT_SUCCESS; + + case GetInfoType::TitleId: + *result = process->GetTitleID(); + return RESULT_SUCCESS; + + case GetInfoType::UserExceptionContextAddr: + LOG_WARNING(Kernel_SVC, + "(STUBBED) Attempted to query user exception context address, returned 0"); + *result = 0; + return RESULT_SUCCESS; + + default: + break; + } + + LOG_WARNING(Kernel_SVC, "(STUBBED) Unimplemented svcGetInfo id=0x{:016X}", info_id); + return ERR_INVALID_ENUM_VALUE; + } + case GetInfoType::IsCurrentProcessBeingDebugged: *result = 0; - break; + return RESULT_SUCCESS; + + case GetInfoType::RegisterResourceLimit: { + if (handle != 0) { + return ERR_INVALID_HANDLE; + } + + if (info_sub_id != 0) { + return ERR_INVALID_COMBINATION; + } + + Process* const current_process = Core::CurrentProcess(); + HandleTable& handle_table = current_process->GetHandleTable(); + const auto resource_limit = current_process->GetResourceLimit(); + if (!resource_limit) { + *result = KernelHandle::InvalidHandle; + // Yes, the kernel considers this a successful operation. + return RESULT_SUCCESS; + } + + const auto table_result = handle_table.Create(resource_limit); + if (table_result.Failed()) { + return table_result.Code(); + } + + *result = *table_result; + return RESULT_SUCCESS; + } + case GetInfoType::RandomEntropy: if (handle != 0) { LOG_ERROR(Kernel_SVC, "Process Handle is non zero, expected 0 result but got {:016X}", @@ -727,37 +826,15 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id) return ERR_INVALID_COMBINATION; } - *result = current_process->GetRandomEntropy(info_sub_id); + *result = Core::CurrentProcess()->GetRandomEntropy(info_sub_id); return RESULT_SUCCESS; - break; - case GetInfoType::ASLRRegionBaseAddr: - *result = vm_manager.GetASLRRegionBaseAddress(); - break; - case GetInfoType::ASLRRegionSize: - *result = vm_manager.GetASLRRegionSize(); - break; - case GetInfoType::NewMapRegionBaseAddr: - *result = vm_manager.GetNewMapRegionBaseAddress(); - break; - case GetInfoType::NewMapRegionSize: - *result = vm_manager.GetNewMapRegionSize(); - break; - case GetInfoType::IsVirtualAddressMemoryEnabled: - *result = current_process->IsVirtualMemoryEnabled(); - break; - case GetInfoType::TitleId: - *result = current_process->GetTitleID(); - break; + case GetInfoType::PrivilegedProcessId: LOG_WARNING(Kernel_SVC, "(STUBBED) Attempted to query privileged process id bounds, returned 0"); *result = 0; - break; - case GetInfoType::UserExceptionContextAddr: - LOG_WARNING(Kernel_SVC, - "(STUBBED) Attempted to query user exception context address, returned 0"); - *result = 0; - break; + return RESULT_SUCCESS; + case GetInfoType::ThreadTickCount: { constexpr u64 num_cpus = 4; if (info_sub_id != 0xFFFFFFFFFFFFFFFF && info_sub_id >= num_cpus) { @@ -767,7 +844,7 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id) } const auto thread = - current_process->GetHandleTable().Get<Thread>(static_cast<Handle>(handle)); + Core::CurrentProcess()->GetHandleTable().Get<Thread>(static_cast<Handle>(handle)); if (!thread) { LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle=0x{:08X}", static_cast<Handle>(handle)); @@ -790,14 +867,13 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id) } *result = out_ticks; - break; + return RESULT_SUCCESS; } + default: LOG_WARNING(Kernel_SVC, "(STUBBED) Unimplemented svcGetInfo id=0x{:016X}", info_id); return ERR_INVALID_ENUM_VALUE; } - - return RESULT_SUCCESS; } /// Sets the thread activity @@ -1001,19 +1077,23 @@ static ResultCode QueryProcessMemory(MemoryInfo* memory_info, PageInfo* /*page_i process_handle); return ERR_INVALID_HANDLE; } - auto vma = process->VMManager().FindVMA(addr); + + const auto& vm_manager = process->VMManager(); + const auto vma = vm_manager.FindVMA(addr); + memory_info->attributes = 0; - if (vma == process->VMManager().vma_map.end()) { - memory_info->base_address = 0; - memory_info->permission = static_cast<u32>(VMAPermission::None); - memory_info->size = 0; - memory_info->type = static_cast<u32>(MemoryState::Unmapped); - } else { + if (vm_manager.IsValidHandle(vma)) { memory_info->base_address = vma->second.base; memory_info->permission = static_cast<u32>(vma->second.permissions); memory_info->size = vma->second.size; memory_info->type = static_cast<u32>(vma->second.meminfo_state); + } else { + memory_info->base_address = 0; + memory_info->permission = static_cast<u32>(VMAPermission::None); + memory_info->size = 0; + memory_info->type = static_cast<u32>(MemoryState::Unmapped); } + return RESULT_SUCCESS; } @@ -1043,7 +1123,7 @@ static void ExitProcess() { static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, VAddr stack_top, u32 priority, s32 processor_id) { LOG_TRACE(Kernel_SVC, - "called entrypoint=0x{:08X} ({}), arg=0x{:08X}, stacktop=0x{:08X}, " + "called entrypoint=0x{:08X}, arg=0x{:08X}, stacktop=0x{:08X}, " "threadpriority=0x{:08X}, processorid=0x{:08X} : created handle=0x{:08X}", entry_point, arg, stack_top, priority, processor_id, *out_handle); @@ -1357,17 +1437,24 @@ static ResultCode CloseHandle(Handle handle) { return handle_table.Close(handle); } -/// Reset an event +/// Clears the signaled state of an event or process. static ResultCode ResetSignal(Handle handle) { LOG_DEBUG(Kernel_SVC, "called handle 0x{:08X}", handle); const auto& handle_table = Core::CurrentProcess()->GetHandleTable(); + auto event = handle_table.Get<ReadableEvent>(handle); + if (event) { + return event->Reset(); + } - ASSERT(event != nullptr); + auto process = handle_table.Get<Process>(handle); + if (process) { + return process->ClearSignalState(); + } - event->Clear(); - return RESULT_SUCCESS; + LOG_ERROR(Kernel_SVC, "Invalid handle (0x{:08X})", handle); + return ERR_INVALID_HANDLE; } /// Creates a TransferMemory object @@ -1521,17 +1608,67 @@ static ResultCode CreateSharedMemory(Handle* handle, u64 size, u32 local_permiss return RESULT_SUCCESS; } +static ResultCode CreateEvent(Handle* write_handle, Handle* read_handle) { + LOG_DEBUG(Kernel_SVC, "called"); + + auto& kernel = Core::System::GetInstance().Kernel(); + const auto [readable_event, writable_event] = + WritableEvent::CreateEventPair(kernel, ResetType::Sticky, "CreateEvent"); + + HandleTable& handle_table = kernel.CurrentProcess()->GetHandleTable(); + + const auto write_create_result = handle_table.Create(writable_event); + if (write_create_result.Failed()) { + return write_create_result.Code(); + } + *write_handle = *write_create_result; + + const auto read_create_result = handle_table.Create(readable_event); + if (read_create_result.Failed()) { + handle_table.Close(*write_create_result); + return read_create_result.Code(); + } + *read_handle = *read_create_result; + + LOG_DEBUG(Kernel_SVC, + "successful. Writable event handle=0x{:08X}, Readable event handle=0x{:08X}", + *write_create_result, *read_create_result); + return RESULT_SUCCESS; +} + static ResultCode ClearEvent(Handle handle) { LOG_TRACE(Kernel_SVC, "called, event=0x{:08X}", handle); const auto& handle_table = Core::CurrentProcess()->GetHandleTable(); - SharedPtr<ReadableEvent> evt = handle_table.Get<ReadableEvent>(handle); - if (evt == nullptr) { - LOG_ERROR(Kernel_SVC, "Event handle does not exist, handle=0x{:08X}", handle); + + auto writable_event = handle_table.Get<WritableEvent>(handle); + if (writable_event) { + writable_event->Clear(); + return RESULT_SUCCESS; + } + + auto readable_event = handle_table.Get<ReadableEvent>(handle); + if (readable_event) { + readable_event->Clear(); + return RESULT_SUCCESS; + } + + LOG_ERROR(Kernel_SVC, "Event handle does not exist, handle=0x{:08X}", handle); + return ERR_INVALID_HANDLE; +} + +static ResultCode SignalEvent(Handle handle) { + LOG_DEBUG(Kernel_SVC, "called. Handle=0x{:08X}", handle); + + HandleTable& handle_table = Core::CurrentProcess()->GetHandleTable(); + auto writable_event = handle_table.Get<WritableEvent>(handle); + + if (!writable_event) { + LOG_ERROR(Kernel_SVC, "Non-existent writable event handle used (0x{:08X})", handle); return ERR_INVALID_HANDLE; } - evt->Clear(); + writable_event->Signal(); return RESULT_SUCCESS; } @@ -1670,7 +1807,7 @@ static const FunctionDef SVC_Table[] = { {0x0E, SvcWrap<GetThreadCoreMask>, "GetThreadCoreMask"}, {0x0F, SvcWrap<SetThreadCoreMask>, "SetThreadCoreMask"}, {0x10, SvcWrap<GetCurrentProcessorNumber>, "GetCurrentProcessorNumber"}, - {0x11, nullptr, "SignalEvent"}, + {0x11, SvcWrap<SignalEvent>, "SignalEvent"}, {0x12, SvcWrap<ClearEvent>, "ClearEvent"}, {0x13, SvcWrap<MapSharedMemory>, "MapSharedMemory"}, {0x14, SvcWrap<UnmapSharedMemory>, "UnmapSharedMemory"}, @@ -1722,7 +1859,7 @@ static const FunctionDef SVC_Table[] = { {0x42, nullptr, "ReplyAndReceiveLight"}, {0x43, nullptr, "ReplyAndReceive"}, {0x44, nullptr, "ReplyAndReceiveWithUserBuffer"}, - {0x45, nullptr, "CreateEvent"}, + {0x45, SvcWrap<CreateEvent>, "CreateEvent"}, {0x46, nullptr, "Unknown"}, {0x47, nullptr, "Unknown"}, {0x48, nullptr, "MapPhysicalMemoryUnsafe"}, diff --git a/src/core/hle/kernel/svc_wrap.h b/src/core/hle/kernel/svc_wrap.h index fa1116624..24aef46c9 100644 --- a/src/core/hle/kernel/svc_wrap.h +++ b/src/core/hle/kernel/svc_wrap.h @@ -59,6 +59,19 @@ void SvcWrap() { FuncReturn(retval); } +template <ResultCode func(u32*, u32*)> +void SvcWrap() { + u32 param_1 = 0; + u32 param_2 = 0; + const u32 retval = func(¶m_1, ¶m_2).raw; + + auto& arm_interface = Core::CurrentArmInterface(); + arm_interface.SetReg(1, param_1); + arm_interface.SetReg(2, param_2); + + FuncReturn(retval); +} + template <ResultCode func(u32*, u64)> void SvcWrap() { u32 param_1 = 0; diff --git a/src/core/hle/kernel/vm_manager.cpp b/src/core/hle/kernel/vm_manager.cpp index 100f8f6bf..6187993ce 100644 --- a/src/core/hle/kernel/vm_manager.cpp +++ b/src/core/hle/kernel/vm_manager.cpp @@ -87,6 +87,10 @@ VMManager::VMAHandle VMManager::FindVMA(VAddr target) const { } } +bool VMManager::IsValidHandle(VMAHandle handle) const { + return handle != vma_map.cend(); +} + ResultVal<VMManager::VMAHandle> VMManager::MapMemoryBlock(VAddr target, std::shared_ptr<std::vector<u8>> block, std::size_t offset, u64 size, diff --git a/src/core/hle/kernel/vm_manager.h b/src/core/hle/kernel/vm_manager.h index d522404fe..a12419d1e 100644 --- a/src/core/hle/kernel/vm_manager.h +++ b/src/core/hle/kernel/vm_manager.h @@ -113,16 +113,10 @@ struct VirtualMemoryArea { * - http://duartes.org/gustavo/blog/post/page-cache-the-affair-between-memory-and-files/ */ class VMManager final { + using VMAMap = std::map<VAddr, VirtualMemoryArea>; + public: - /** - * A map covering the entirety of the managed address space, keyed by the `base` field of each - * VMA. It must always be modified by splitting or merging VMAs, so that the invariant - * `elem.base + elem.size == next.base` is preserved, and mergeable regions must always be - * merged when possible so that no two similar and adjacent regions exist that have not been - * merged. - */ - std::map<VAddr, VirtualMemoryArea> vma_map; - using VMAHandle = decltype(vma_map)::const_iterator; + using VMAHandle = VMAMap::const_iterator; VMManager(); ~VMManager(); @@ -133,6 +127,9 @@ public: /// Finds the VMA in which the given address is included in, or `vma_map.end()`. VMAHandle FindVMA(VAddr target) const; + /// Indicates whether or not the given handle is within the VMA map. + bool IsValidHandle(VMAHandle handle) const; + // TODO(yuriks): Should these functions actually return the handle? /** @@ -281,7 +278,7 @@ public: Memory::PageTable page_table; private: - using VMAIter = decltype(vma_map)::iterator; + using VMAIter = VMAMap::iterator; /// Converts a VMAHandle to a mutable VMAIter. VMAIter StripIterConstness(const VMAHandle& iter); @@ -328,6 +325,15 @@ private: /// Clears out the page table void ClearPageTable(); + /** + * A map covering the entirety of the managed address space, keyed by the `base` field of each + * VMA. It must always be modified by splitting or merging VMAs, so that the invariant + * `elem.base + elem.size == next.base` is preserved, and mergeable regions must always be + * merged when possible so that no two similar and adjacent regions exist that have not been + * merged. + */ + VMAMap vma_map; + u32 address_space_width = 0; VAddr address_space_base = 0; VAddr address_space_end = 0; diff --git a/src/core/hle/kernel/writable_event.h b/src/core/hle/kernel/writable_event.h index fc57d18d7..8fa8d68ee 100644 --- a/src/core/hle/kernel/writable_event.h +++ b/src/core/hle/kernel/writable_event.h @@ -39,7 +39,7 @@ public: return name; } - static const HandleType HANDLE_TYPE = HandleType::Event; + static const HandleType HANDLE_TYPE = HandleType::WritableEvent; HandleType GetHandleType() const override { return HANDLE_TYPE; } diff --git a/src/core/hle/service/filesystem/fsp_srv.cpp b/src/core/hle/service/filesystem/fsp_srv.cpp index 694ec40ec..63fa48133 100644 --- a/src/core/hle/service/filesystem/fsp_srv.cpp +++ b/src/core/hle/service/filesystem/fsp_srv.cpp @@ -20,6 +20,7 @@ #include "core/file_sys/nca_metadata.h" #include "core/file_sys/patch_manager.h" #include "core/file_sys/savedata_factory.h" +#include "core/file_sys/system_archive/system_archive.h" #include "core/file_sys/vfs.h" #include "core/hle/ipc_helpers.h" #include "core/hle/kernel/process.h" @@ -795,9 +796,18 @@ void FSP_SRV::OpenSaveDataInfoReaderBySaveDataSpaceId(Kernel::HLERequestContext& void FSP_SRV::GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service_FS, "(STUBBED) called"); + enum class LogMode : u32 { + Off, + Log, + RedirectToSdCard, + LogToSdCard = Log | RedirectToSdCard, + }; + + // Given we always want to receive logging information, + // we always specify logging as enabled. IPC::ResponseBuilder rb{ctx, 3}; rb.Push(RESULT_SUCCESS); - rb.Push<u32>(5); + rb.PushEnum(LogMode::Log); } void FSP_SRV::OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) { @@ -831,6 +841,15 @@ void FSP_SRV::OpenDataStorageByDataId(Kernel::HLERequestContext& ctx) { auto data = OpenRomFS(title_id, storage_id, FileSys::ContentRecordType::Data); if (data.Failed()) { + const auto archive = FileSys::SystemArchive::SynthesizeSystemArchive(title_id); + + if (archive != nullptr) { + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; + rb.Push(RESULT_SUCCESS); + rb.PushIpcInterface(std::make_shared<IStorage>(archive)); + return; + } + // TODO(DarkLordZach): Find the right error code to use here LOG_ERROR(Service_FS, "could not open data storage with title_id={:016X}, storage_id={:02X}", title_id, diff --git a/src/core/hle/service/ldr/ldr.cpp b/src/core/hle/service/ldr/ldr.cpp index ca119dd3a..13bcefe07 100644 --- a/src/core/hle/service/ldr/ldr.cpp +++ b/src/core/hle/service/ldr/ldr.cpp @@ -335,10 +335,7 @@ public: vm_manager.ReprotectRange(*map_address + header.rw_offset, header.rw_size, Kernel::VMAPermission::ReadWrite); - Core::System::GetInstance().ArmInterface(0).ClearInstructionCache(); - Core::System::GetInstance().ArmInterface(1).ClearInstructionCache(); - Core::System::GetInstance().ArmInterface(2).ClearInstructionCache(); - Core::System::GetInstance().ArmInterface(3).ClearInstructionCache(); + Core::System::GetInstance().InvalidateCpuInstructionCaches(); nro.insert_or_assign(*map_address, NROInfo{hash, nro_size + bss_size}); @@ -391,10 +388,7 @@ public: Kernel::MemoryState::ModuleCodeStatic) == RESULT_SUCCESS); ASSERT(process->UnmapMemory(mapped_addr, 0, nro_size) == RESULT_SUCCESS); - Core::System::GetInstance().ArmInterface(0).ClearInstructionCache(); - Core::System::GetInstance().ArmInterface(1).ClearInstructionCache(); - Core::System::GetInstance().ArmInterface(2).ClearInstructionCache(); - Core::System::GetInstance().ArmInterface(3).ClearInstructionCache(); + Core::System::GetInstance().InvalidateCpuInstructionCaches(); nro.erase(iter); IPC::ResponseBuilder rb{ctx, 2}; @@ -414,13 +408,13 @@ private: using SHA256Hash = std::array<u8, 0x20>; struct NROHeader { - u32_le entrypoint_insn; + INSERT_PADDING_WORDS(1); u32_le mod_offset; INSERT_PADDING_WORDS(2); u32_le magic; - INSERT_PADDING_WORDS(1); + u32_le version; u32_le nro_size; - INSERT_PADDING_WORDS(1); + u32_le flags; u32_le text_offset; u32_le text_size; u32_le ro_offset; @@ -436,9 +430,10 @@ private: struct NRRHeader { u32_le magic; - INSERT_PADDING_BYTES(0x1C); + INSERT_PADDING_BYTES(12); u64_le title_id_mask; u64_le title_id_pattern; + INSERT_PADDING_BYTES(16); std::array<u8, 0x100> modulus; std::array<u8, 0x100> signature_1; std::array<u8, 0x100> signature_2; diff --git a/src/core/hle/service/service.cpp b/src/core/hle/service/service.cpp index 1ec340466..d41df3732 100644 --- a/src/core/hle/service/service.cpp +++ b/src/core/hle/service/service.cpp @@ -70,10 +70,6 @@ #include "core/hle/service/vi/vi.h" #include "core/hle/service/wlan/wlan.h" -using Kernel::ClientPort; -using Kernel::ServerPort; -using Kernel::SharedPtr; - namespace Service { /** @@ -110,10 +106,8 @@ void ServiceFrameworkBase::InstallAsNamedPort() { ASSERT(port == nullptr); auto& kernel = Core::System::GetInstance().Kernel(); - SharedPtr<ServerPort> server_port; - SharedPtr<ClientPort> client_port; - std::tie(server_port, client_port) = - ServerPort::CreatePortPair(kernel, max_sessions, service_name); + auto [server_port, client_port] = + Kernel::ServerPort::CreatePortPair(kernel, max_sessions, service_name); server_port->SetHleHandler(shared_from_this()); kernel.AddNamedPort(service_name, std::move(client_port)); } @@ -122,11 +116,9 @@ Kernel::SharedPtr<Kernel::ClientPort> ServiceFrameworkBase::CreatePort() { ASSERT(port == nullptr); auto& kernel = Core::System::GetInstance().Kernel(); - Kernel::SharedPtr<Kernel::ServerPort> server_port; - Kernel::SharedPtr<Kernel::ClientPort> client_port; - std::tie(server_port, client_port) = + auto [server_port, client_port] = Kernel::ServerPort::CreatePortPair(kernel, max_sessions, service_name); - port = MakeResult<Kernel::SharedPtr<Kernel::ServerPort>>(std::move(server_port)).Unwrap(); + port = MakeResult(std::move(server_port)).Unwrap(); port->SetHleHandler(shared_from_this()); return client_port; } @@ -152,8 +144,7 @@ void ServiceFrameworkBase::ReportUnimplementedFunction(Kernel::HLERequestContext } buf.push_back('}'); - LOG_ERROR(Service, "unknown / unimplemented {}", fmt::to_string(buf)); - UNIMPLEMENTED(); + UNIMPLEMENTED_MSG("Unknown / unimplemented {}", fmt::to_string(buf)); } void ServiceFrameworkBase::InvokeRequest(Kernel::HLERequestContext& ctx) { diff --git a/src/core/hle/service/sm/sm.cpp b/src/core/hle/service/sm/sm.cpp index 0d0f63a78..d73530086 100644 --- a/src/core/hle/service/sm/sm.cpp +++ b/src/core/hle/service/sm/sm.cpp @@ -54,13 +54,11 @@ ResultVal<Kernel::SharedPtr<Kernel::ServerPort>> ServiceManager::RegisterService return ERR_ALREADY_REGISTERED; auto& kernel = Core::System::GetInstance().Kernel(); - Kernel::SharedPtr<Kernel::ServerPort> server_port; - Kernel::SharedPtr<Kernel::ClientPort> client_port; - std::tie(server_port, client_port) = + auto [server_port, client_port] = Kernel::ServerPort::CreatePortPair(kernel, max_sessions, name); registered_services.emplace(std::move(name), std::move(client_port)); - return MakeResult<Kernel::SharedPtr<Kernel::ServerPort>>(std::move(server_port)); + return MakeResult(std::move(server_port)); } ResultCode ServiceManager::UnregisterService(const std::string& name) { @@ -83,7 +81,7 @@ ResultVal<Kernel::SharedPtr<Kernel::ClientPort>> ServiceManager::GetServicePort( return ERR_SERVICE_NOT_REGISTERED; } - return MakeResult<Kernel::SharedPtr<Kernel::ClientPort>>(it->second); + return MakeResult(it->second); } ResultVal<Kernel::SharedPtr<Kernel::ClientSession>> ServiceManager::ConnectToService( diff --git a/src/core/loader/deconstructed_rom_directory.cpp b/src/core/loader/deconstructed_rom_directory.cpp index 8518dddcb..ac04d72d7 100644 --- a/src/core/loader/deconstructed_rom_directory.cpp +++ b/src/core/loader/deconstructed_rom_directory.cpp @@ -7,7 +7,6 @@ #include "common/common_funcs.h" #include "common/file_util.h" #include "common/logging/log.h" -#include "core/core.h" #include "core/file_sys/content_archive.h" #include "core/file_sys/control_metadata.h" #include "core/file_sys/patch_manager.h" @@ -146,7 +145,7 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(Kernel::Process& process) const VAddr load_addr = next_load_addr; const bool should_pass_arguments = std::strcmp(module, "rtld") == 0; const auto tentative_next_load_addr = - AppLoader_NSO::LoadModule(*module_file, load_addr, should_pass_arguments, pm); + AppLoader_NSO::LoadModule(process, *module_file, load_addr, should_pass_arguments, pm); if (!tentative_next_load_addr) { return ResultStatus::ErrorLoadingNSO; } diff --git a/src/core/loader/deconstructed_rom_directory.h b/src/core/loader/deconstructed_rom_directory.h index d109ed2b5..1615cb5a8 100644 --- a/src/core/loader/deconstructed_rom_directory.h +++ b/src/core/loader/deconstructed_rom_directory.h @@ -33,7 +33,7 @@ public: */ static FileType IdentifyType(const FileSys::VirtualFile& file); - FileType GetFileType() override { + FileType GetFileType() const override { return IdentifyType(file); } diff --git a/src/core/loader/elf.h b/src/core/loader/elf.h index 6af76441c..a2d33021c 100644 --- a/src/core/loader/elf.h +++ b/src/core/loader/elf.h @@ -22,7 +22,7 @@ public: */ static FileType IdentifyType(const FileSys::VirtualFile& file); - FileType GetFileType() override { + FileType GetFileType() const override { return IdentifyType(file); } diff --git a/src/core/loader/loader.h b/src/core/loader/loader.h index c589b3bd1..0838e303b 100644 --- a/src/core/loader/loader.h +++ b/src/core/loader/loader.h @@ -132,7 +132,7 @@ public: * Returns the type of this file * @return FileType corresponding to the loaded file */ - virtual FileType GetFileType() = 0; + virtual FileType GetFileType() const = 0; /** * Load the application and return the created Process instance diff --git a/src/core/loader/nax.cpp b/src/core/loader/nax.cpp index 42f4a777b..a093e3d36 100644 --- a/src/core/loader/nax.cpp +++ b/src/core/loader/nax.cpp @@ -37,7 +37,7 @@ FileType AppLoader_NAX::IdentifyType(const FileSys::VirtualFile& file) { return IdentifyTypeImpl(nax); } -FileType AppLoader_NAX::GetFileType() { +FileType AppLoader_NAX::GetFileType() const { return IdentifyTypeImpl(*nax); } diff --git a/src/core/loader/nax.h b/src/core/loader/nax.h index b4d93bd01..0a97511b8 100644 --- a/src/core/loader/nax.h +++ b/src/core/loader/nax.h @@ -31,7 +31,7 @@ public: */ static FileType IdentifyType(const FileSys::VirtualFile& file); - FileType GetFileType() override; + FileType GetFileType() const override; ResultStatus Load(Kernel::Process& process) override; diff --git a/src/core/loader/nca.h b/src/core/loader/nca.h index 95d9b73a1..cbbe701d2 100644 --- a/src/core/loader/nca.h +++ b/src/core/loader/nca.h @@ -29,7 +29,7 @@ public: */ static FileType IdentifyType(const FileSys::VirtualFile& file); - FileType GetFileType() override { + FileType GetFileType() const override { return IdentifyType(file); } diff --git a/src/core/loader/nro.cpp b/src/core/loader/nro.cpp index fbbd6b0de..4fad0c0dd 100644 --- a/src/core/loader/nro.cpp +++ b/src/core/loader/nro.cpp @@ -10,7 +10,6 @@ #include "common/file_util.h" #include "common/logging/log.h" #include "common/swap.h" -#include "core/core.h" #include "core/file_sys/control_metadata.h" #include "core/file_sys/romfs_factory.h" #include "core/file_sys/vfs_offset.h" @@ -129,9 +128,8 @@ static constexpr u32 PageAlignSize(u32 size) { return (size + Memory::PAGE_MASK) & ~Memory::PAGE_MASK; } -/*static*/ bool AppLoader_NRO::LoadNro(const std::vector<u8>& data, const std::string& name, - VAddr load_base) { - +static bool LoadNroImpl(Kernel::Process& process, const std::vector<u8>& data, + const std::string& name, VAddr load_base) { if (data.size() < sizeof(NroHeader)) { return {}; } @@ -189,7 +187,7 @@ static constexpr u32 PageAlignSize(u32 size) { // Load codeset for current process codeset.memory = std::make_shared<std::vector<u8>>(std::move(program_image)); - Core::CurrentProcess()->LoadModule(std::move(codeset), load_base); + process.LoadModule(std::move(codeset), load_base); // Register module with GDBStub GDBStub::RegisterModule(name, load_base, load_base); @@ -197,8 +195,9 @@ static constexpr u32 PageAlignSize(u32 size) { return true; } -bool AppLoader_NRO::LoadNro(const FileSys::VfsFile& file, VAddr load_base) { - return AppLoader_NRO::LoadNro(file.ReadAllBytes(), file.GetName(), load_base); +bool AppLoader_NRO::LoadNro(Kernel::Process& process, const FileSys::VfsFile& file, + VAddr load_base) { + return LoadNroImpl(process, file.ReadAllBytes(), file.GetName(), load_base); } ResultStatus AppLoader_NRO::Load(Kernel::Process& process) { @@ -209,7 +208,7 @@ ResultStatus AppLoader_NRO::Load(Kernel::Process& process) { // Load NRO const VAddr base_address = process.VMManager().GetCodeRegionBaseAddress(); - if (!LoadNro(*file, base_address)) { + if (!LoadNro(process, *file, base_address)) { return ResultStatus::ErrorLoadingNRO; } diff --git a/src/core/loader/nro.h b/src/core/loader/nro.h index 3e6959302..013d629c0 100644 --- a/src/core/loader/nro.h +++ b/src/core/loader/nro.h @@ -14,6 +14,10 @@ namespace FileSys { class NACP; } +namespace Kernel { +class Process; +} + namespace Loader { /// Loads an NRO file @@ -29,7 +33,7 @@ public: */ static FileType IdentifyType(const FileSys::VirtualFile& file); - FileType GetFileType() override { + FileType GetFileType() const override { return IdentifyType(file); } @@ -41,10 +45,8 @@ public: ResultStatus ReadTitle(std::string& title) override; bool IsRomFSUpdatable() const override; - static bool LoadNro(const std::vector<u8>& data, const std::string& name, VAddr load_base); - private: - bool LoadNro(const FileSys::VfsFile& file, VAddr load_base); + bool LoadNro(Kernel::Process& process, const FileSys::VfsFile& file, VAddr load_base); std::vector<u8> icon_data; std::unique_ptr<FileSys::NACP> nacp; diff --git a/src/core/loader/nso.cpp b/src/core/loader/nso.cpp index aaf006309..6ded0b707 100644 --- a/src/core/loader/nso.cpp +++ b/src/core/loader/nso.cpp @@ -9,7 +9,6 @@ #include "common/file_util.h" #include "common/logging/log.h" #include "common/swap.h" -#include "core/core.h" #include "core/file_sys/patch_manager.h" #include "core/gdbstub/gdbstub.h" #include "core/hle/kernel/process.h" @@ -93,7 +92,8 @@ static constexpr u32 PageAlignSize(u32 size) { return (size + Memory::PAGE_MASK) & ~Memory::PAGE_MASK; } -std::optional<VAddr> AppLoader_NSO::LoadModule(const FileSys::VfsFile& file, VAddr load_base, +std::optional<VAddr> AppLoader_NSO::LoadModule(Kernel::Process& process, + const FileSys::VfsFile& file, VAddr load_base, bool should_pass_arguments, std::optional<FileSys::PatchManager> pm) { if (file.GetSize() < sizeof(NsoHeader)) @@ -166,7 +166,7 @@ std::optional<VAddr> AppLoader_NSO::LoadModule(const FileSys::VfsFile& file, VAd // Load codeset for current process codeset.memory = std::make_shared<std::vector<u8>>(std::move(program_image)); - Core::CurrentProcess()->LoadModule(std::move(codeset), load_base); + process.LoadModule(std::move(codeset), load_base); // Register module with GDBStub GDBStub::RegisterModule(file.GetName(), load_base, load_base); @@ -181,7 +181,7 @@ ResultStatus AppLoader_NSO::Load(Kernel::Process& process) { // Load module const VAddr base_address = process.VMManager().GetCodeRegionBaseAddress(); - if (!LoadModule(*file, base_address, true)) { + if (!LoadModule(process, *file, base_address, true)) { return ResultStatus::ErrorLoadingNSO; } LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", file->GetName(), base_address); diff --git a/src/core/loader/nso.h b/src/core/loader/nso.h index 433306139..135b6ea5a 100644 --- a/src/core/loader/nso.h +++ b/src/core/loader/nso.h @@ -10,6 +10,10 @@ #include "core/loader/linker.h" #include "core/loader/loader.h" +namespace Kernel { +class Process; +} + namespace Loader { constexpr u64 NSO_ARGUMENT_DATA_ALLOCATION_SIZE = 0x9000; @@ -33,12 +37,12 @@ public: */ static FileType IdentifyType(const FileSys::VirtualFile& file); - FileType GetFileType() override { + FileType GetFileType() const override { return IdentifyType(file); } - static std::optional<VAddr> LoadModule(const FileSys::VfsFile& file, VAddr load_base, - bool should_pass_arguments, + static std::optional<VAddr> LoadModule(Kernel::Process& process, const FileSys::VfsFile& file, + VAddr load_base, bool should_pass_arguments, std::optional<FileSys::PatchManager> pm = {}); ResultStatus Load(Kernel::Process& process) override; diff --git a/src/core/loader/nsp.h b/src/core/loader/nsp.h index 842327a49..2b1e0719b 100644 --- a/src/core/loader/nsp.h +++ b/src/core/loader/nsp.h @@ -31,7 +31,7 @@ public: */ static FileType IdentifyType(const FileSys::VirtualFile& file); - FileType GetFileType() override { + FileType GetFileType() const override { return IdentifyType(file); } diff --git a/src/core/loader/xci.h b/src/core/loader/xci.h index f3b5a3936..15d1b1a23 100644 --- a/src/core/loader/xci.h +++ b/src/core/loader/xci.h @@ -31,7 +31,7 @@ public: */ static FileType IdentifyType(const FileSys::VirtualFile& file); - FileType GetFileType() override { + FileType GetFileType() const override { return IdentifyType(file); } diff --git a/src/core/memory.cpp b/src/core/memory.cpp index 70abd856a..643afdee8 100644 --- a/src/core/memory.cpp +++ b/src/core/memory.cpp @@ -53,6 +53,14 @@ void PageTable::Resize(std::size_t address_space_width_in_bits) { pointers.resize(num_page_table_entries); attributes.resize(num_page_table_entries); + + // The default is a 39-bit address space, which causes an initial 1GB allocation size. If the + // vector size is subsequently decreased (via resize), the vector might not automatically + // actually reallocate/resize its underlying allocation, which wastes up to ~800 MB for + // 36-bit titles. Call shrink_to_fit to reduce capacity to what's actually in use. + + pointers.shrink_to_fit(); + attributes.shrink_to_fit(); } static void MapPages(PageTable& page_table, VAddr base, u64 size, u8* memory, PageType type) { @@ -117,14 +125,13 @@ void RemoveDebugHook(PageTable& page_table, VAddr base, u64 size, MemoryHookPoin * using a VMA from the current process */ static u8* GetPointerFromVMA(const Kernel::Process& process, VAddr vaddr) { - u8* direct_pointer = nullptr; + const auto& vm_manager = process.VMManager(); - auto& vm_manager = process.VMManager(); + const auto it = vm_manager.FindVMA(vaddr); + DEBUG_ASSERT(vm_manager.IsValidHandle(it)); - auto it = vm_manager.FindVMA(vaddr); - ASSERT(it != vm_manager.vma_map.end()); - - auto& vma = it->second; + u8* direct_pointer = nullptr; + const auto& vma = it->second; switch (vma.type) { case Kernel::VMAType::AllocatedMemoryBlock: direct_pointer = vma.backing_block->data() + vma.offset; diff --git a/src/core/telemetry_session.cpp b/src/core/telemetry_session.cpp index a3b08c740..09ed74d78 100644 --- a/src/core/telemetry_session.cpp +++ b/src/core/telemetry_session.cpp @@ -103,13 +103,8 @@ bool VerifyLogin(const std::string& username, const std::string& token) { TelemetrySession::TelemetrySession() { #ifdef ENABLE_WEB_SERVICE - if (Settings::values.enable_telemetry) { - backend = std::make_unique<WebService::TelemetryJson>(Settings::values.web_api_url, - Settings::values.yuzu_username, - Settings::values.yuzu_token); - } else { - backend = std::make_unique<Telemetry::NullVisitor>(); - } + backend = std::make_unique<WebService::TelemetryJson>( + Settings::values.web_api_url, Settings::values.yuzu_username, Settings::values.yuzu_token); #else backend = std::make_unique<Telemetry::NullVisitor>(); #endif @@ -180,7 +175,8 @@ TelemetrySession::~TelemetrySession() { // This is just a placeholder to wrap up the session once the core completes and this is // destroyed. This will be moved elsewhere once we are actually doing real I/O with the service. field_collection.Accept(*backend); - backend->Complete(); + if (Settings::values.enable_telemetry) + backend->Complete(); backend = nullptr; } diff --git a/src/video_core/command_processor.cpp b/src/video_core/command_processor.cpp deleted file mode 100644 index 8b9c548cc..000000000 --- a/src/video_core/command_processor.cpp +++ /dev/null @@ -1,142 +0,0 @@ -// Copyright 2018 yuzu Emulator Project -// Licensed under GPLv2 or any later version -// Refer to the license.txt file included. - -#include <array> -#include <cstddef> -#include <memory> -#include <utility> -#include "common/assert.h" -#include "common/logging/log.h" -#include "common/microprofile.h" -#include "common/vector_math.h" -#include "core/memory.h" -#include "core/tracer/recorder.h" -#include "video_core/command_processor.h" -#include "video_core/engines/fermi_2d.h" -#include "video_core/engines/kepler_memory.h" -#include "video_core/engines/maxwell_3d.h" -#include "video_core/engines/maxwell_compute.h" -#include "video_core/engines/maxwell_dma.h" -#include "video_core/gpu.h" -#include "video_core/renderer_base.h" -#include "video_core/video_core.h" - -namespace Tegra { - -enum class BufferMethods { - BindObject = 0, - CountBufferMethods = 0x40, -}; - -MICROPROFILE_DEFINE(ProcessCommandLists, "GPU", "Execute command buffer", MP_RGB(128, 128, 192)); - -void GPU::ProcessCommandLists(const std::vector<CommandListHeader>& commands) { - MICROPROFILE_SCOPE(ProcessCommandLists); - - // On entering GPU code, assume all memory may be touched by the ARM core. - maxwell_3d->dirty_flags.OnMemoryWrite(); - - auto WriteReg = [this](u32 method, u32 subchannel, u32 value, u32 remaining_params) { - LOG_TRACE(HW_GPU, - "Processing method {:08X} on subchannel {} value " - "{:08X} remaining params {}", - method, subchannel, value, remaining_params); - - ASSERT(subchannel < bound_engines.size()); - - if (method == static_cast<u32>(BufferMethods::BindObject)) { - // Bind the current subchannel to the desired engine id. - LOG_DEBUG(HW_GPU, "Binding subchannel {} to engine {}", subchannel, value); - bound_engines[subchannel] = static_cast<EngineID>(value); - return; - } - - if (method < static_cast<u32>(BufferMethods::CountBufferMethods)) { - // TODO(Subv): Research and implement these methods. - LOG_ERROR(HW_GPU, "Special buffer methods other than Bind are not implemented"); - return; - } - - const EngineID engine = bound_engines[subchannel]; - - switch (engine) { - case EngineID::FERMI_TWOD_A: - fermi_2d->WriteReg(method, value); - break; - case EngineID::MAXWELL_B: - maxwell_3d->WriteReg(method, value, remaining_params); - break; - case EngineID::MAXWELL_COMPUTE_B: - maxwell_compute->WriteReg(method, value); - break; - case EngineID::MAXWELL_DMA_COPY_A: - maxwell_dma->WriteReg(method, value); - break; - case EngineID::KEPLER_INLINE_TO_MEMORY_B: - kepler_memory->WriteReg(method, value); - break; - default: - UNIMPLEMENTED_MSG("Unimplemented engine"); - } - }; - - for (auto entry : commands) { - Tegra::GPUVAddr address = entry.Address(); - u32 size = entry.sz; - const std::optional<VAddr> head_address = memory_manager->GpuToCpuAddress(address); - VAddr current_addr = *head_address; - while (current_addr < *head_address + size * sizeof(CommandHeader)) { - const CommandHeader header = {Memory::Read32(current_addr)}; - current_addr += sizeof(u32); - - switch (header.mode.Value()) { - case SubmissionMode::IncreasingOld: - case SubmissionMode::Increasing: { - // Increase the method value with each argument. - for (unsigned i = 0; i < header.arg_count; ++i) { - WriteReg(header.method + i, header.subchannel, Memory::Read32(current_addr), - header.arg_count - i - 1); - current_addr += sizeof(u32); - } - break; - } - case SubmissionMode::NonIncreasingOld: - case SubmissionMode::NonIncreasing: { - // Use the same method value for all arguments. - for (unsigned i = 0; i < header.arg_count; ++i) { - WriteReg(header.method, header.subchannel, Memory::Read32(current_addr), - header.arg_count - i - 1); - current_addr += sizeof(u32); - } - break; - } - case SubmissionMode::IncreaseOnce: { - ASSERT(header.arg_count.Value() >= 1); - - // Use the original method for the first argument and then the next method for all - // other arguments. - WriteReg(header.method, header.subchannel, Memory::Read32(current_addr), - header.arg_count - 1); - current_addr += sizeof(u32); - - for (unsigned i = 1; i < header.arg_count; ++i) { - WriteReg(header.method + 1, header.subchannel, Memory::Read32(current_addr), - header.arg_count - i - 1); - current_addr += sizeof(u32); - } - break; - } - case SubmissionMode::Inline: { - // The register value is stored in the bits 16-28 as an immediate - WriteReg(header.method, header.subchannel, header.inline_data, 0); - break; - } - default: - UNIMPLEMENTED(); - } - } - } -} - -} // namespace Tegra diff --git a/src/video_core/engines/shader_bytecode.h b/src/video_core/engines/shader_bytecode.h index b9faaf8e0..5ea094e64 100644 --- a/src/video_core/engines/shader_bytecode.h +++ b/src/video_core/engines/shader_bytecode.h @@ -1049,6 +1049,7 @@ union Instruction { BitField<49, 1, u64> nodep_flag; BitField<50, 3, u64> component_mask_selector; BitField<53, 4, u64> texture_info; + BitField<60, 1, u64> fp32_flag; TextureType GetTextureType() const { // The TEXS instruction has a weird encoding for the texture type. @@ -1549,7 +1550,7 @@ private: INST("1110111011011---", Id::STG, Type::Memory, "STG"), INST("110000----111---", Id::TEX, Type::Memory, "TEX"), INST("1101111101001---", Id::TXQ, Type::Memory, "TXQ"), - INST("1101100---------", Id::TEXS, Type::Memory, "TEXS"), + INST("1101-00---------", Id::TEXS, Type::Memory, "TEXS"), INST("1101101---------", Id::TLDS, Type::Memory, "TLDS"), INST("110010----111---", Id::TLD4, Type::Memory, "TLD4"), INST("1101111100------", Id::TLD4S, Type::Memory, "TLD4S"), diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp index fd1242333..88c45a423 100644 --- a/src/video_core/gpu.cpp +++ b/src/video_core/gpu.cpp @@ -141,6 +141,12 @@ void GPU::CallMethod(const MethodCall& method_call) { return; } + if (method_call.method < static_cast<u32>(BufferMethods::CountBufferMethods)) { + // TODO(Subv): Research and implement these methods. + LOG_ERROR(HW_GPU, "Special buffer methods other than Bind are not implemented"); + return; + } + const EngineID engine = bound_engines[method_call.subchannel]; switch (engine) { diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index 9e93bd609..2b29fc45f 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -79,6 +79,26 @@ struct DrawParameters { } }; +struct FramebufferCacheKey { + bool is_single_buffer = false; + bool stencil_enable = false; + + std::array<GLenum, Maxwell::NumRenderTargets> color_attachments{}; + std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::NumRenderTargets> colors{}; + u32 colors_count = 0; + + GLuint zeta = 0; + + auto Tie() const { + return std::tie(is_single_buffer, stencil_enable, color_attachments, colors, colors_count, + zeta); + } + + bool operator<(const FramebufferCacheKey& rhs) const { + return Tie() < rhs.Tie(); + } +}; + RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo& info) : res_cache{*this}, shader_cache{*this}, emu_window{window}, screen_info{info}, buffer_cache(*this, STREAM_BUFFER_SIZE) { @@ -90,9 +110,6 @@ RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo OpenGLState::ApplyDefaultState(); - // Create render framebuffer - framebuffer.Create(); - shader_program_manager = std::make_unique<GLShader::ProgramManager>(); state.draw.shader_program = 0; state.Apply(); @@ -361,6 +378,44 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) { SyncClipEnabled(clip_distances); } +void RasterizerOpenGL::SetupCachedFramebuffer(const FramebufferCacheKey& fbkey, + OpenGLState& current_state) { + const auto [entry, is_cache_miss] = framebuffer_cache.try_emplace(fbkey); + auto& framebuffer = entry->second; + + if (is_cache_miss) + framebuffer.Create(); + + current_state.draw.draw_framebuffer = framebuffer.handle; + current_state.ApplyFramebufferState(); + + if (!is_cache_miss) + return; + + if (fbkey.is_single_buffer) { + if (fbkey.color_attachments[0] != GL_NONE) { + glFramebufferTexture(GL_DRAW_FRAMEBUFFER, fbkey.color_attachments[0], fbkey.colors[0], + 0); + } + glDrawBuffer(fbkey.color_attachments[0]); + } else { + for (std::size_t index = 0; index < Maxwell::NumRenderTargets; ++index) { + if (fbkey.colors[index]) { + glFramebufferTexture(GL_DRAW_FRAMEBUFFER, + GL_COLOR_ATTACHMENT0 + static_cast<GLenum>(index), + fbkey.colors[index], 0); + } + } + glDrawBuffers(fbkey.colors_count, fbkey.color_attachments.data()); + } + + if (fbkey.zeta) { + GLenum zeta_attachment = + fbkey.stencil_enable ? GL_DEPTH_STENCIL_ATTACHMENT : GL_DEPTH_ATTACHMENT; + glFramebufferTexture(GL_DRAW_FRAMEBUFFER, zeta_attachment, fbkey.zeta, 0); + } +} + std::size_t RasterizerOpenGL::CalculateVertexArraysSize() const { const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs; @@ -444,10 +499,10 @@ void RasterizerOpenGL::ConfigureFramebuffers(OpenGLState& current_state, bool us UNIMPLEMENTED_IF(regs.rt_separate_frag_data != 0); // Bind the framebuffer surfaces - current_state.draw.draw_framebuffer = framebuffer.handle; - current_state.ApplyFramebufferState(); current_state.framebuffer_srgb.enabled = regs.framebuffer_srgb != 0; + FramebufferCacheKey fbkey; + if (using_color_fb) { if (single_color_target) { // Used when just a single color attachment is enabled, e.g. for clearing a color buffer @@ -463,14 +518,12 @@ void RasterizerOpenGL::ConfigureFramebuffers(OpenGLState& current_state, bool us state.framebuffer_srgb.enabled |= color_surface->GetSurfaceParams().srgb_conversion; } - glFramebufferTexture2D( - GL_DRAW_FRAMEBUFFER, - GL_COLOR_ATTACHMENT0 + static_cast<GLenum>(*single_color_target), GL_TEXTURE_2D, - color_surface != nullptr ? color_surface->Texture().handle : 0, 0); - glDrawBuffer(GL_COLOR_ATTACHMENT0 + static_cast<GLenum>(*single_color_target)); + fbkey.is_single_buffer = true; + fbkey.color_attachments[0] = + GL_COLOR_ATTACHMENT0 + static_cast<GLenum>(*single_color_target); + fbkey.colors[0] = color_surface != nullptr ? color_surface->Texture().handle : 0; } else { // Multiple color attachments are enabled - std::array<GLenum, Maxwell::NumRenderTargets> buffers; for (std::size_t index = 0; index < Maxwell::NumRenderTargets; ++index) { Surface color_surface = res_cache.GetColorBufferSurface(index, preserve_contents); @@ -485,22 +538,17 @@ void RasterizerOpenGL::ConfigureFramebuffers(OpenGLState& current_state, bool us color_surface->GetSurfaceParams().srgb_conversion; } - buffers[index] = GL_COLOR_ATTACHMENT0 + regs.rt_control.GetMap(index); - glFramebufferTexture2D( - GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0 + static_cast<GLenum>(index), - GL_TEXTURE_2D, color_surface != nullptr ? color_surface->Texture().handle : 0, - 0); + fbkey.color_attachments[index] = + GL_COLOR_ATTACHMENT0 + regs.rt_control.GetMap(index); + fbkey.colors[index] = + color_surface != nullptr ? color_surface->Texture().handle : 0; } - glDrawBuffers(regs.rt_control.count, buffers.data()); + fbkey.is_single_buffer = false; + fbkey.colors_count = regs.rt_control.count; } } else { - // No color attachments are enabled - zero out all of them - for (std::size_t index = 0; index < Maxwell::NumRenderTargets; ++index) { - glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, - GL_COLOR_ATTACHMENT0 + static_cast<GLenum>(index), GL_TEXTURE_2D, - 0, 0); - } - glDrawBuffer(GL_NONE); + // No color attachments are enabled - leave them as zero + fbkey.is_single_buffer = true; } if (depth_surface) { @@ -508,22 +556,12 @@ void RasterizerOpenGL::ConfigureFramebuffers(OpenGLState& current_state, bool us // the shader doesn't actually write to it. depth_surface->MarkAsModified(true, res_cache); - if (regs.stencil_enable) { - // Attach both depth and stencil - glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, - depth_surface->Texture().handle, 0); - } else { - // Attach depth - glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D, - depth_surface->Texture().handle, 0); - // Clear stencil attachment - glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0); - } - } else { - // Clear both depth and stencil attachment - glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, - 0); + fbkey.zeta = depth_surface->Texture().handle; + fbkey.stencil_enable = regs.stencil_enable; } + + SetupCachedFramebuffer(fbkey, current_state); + SyncViewport(current_state); } diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h index 988fa3e27..8a891ffc7 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.h +++ b/src/video_core/renderer_opengl/gl_rasterizer.h @@ -40,6 +40,7 @@ namespace OpenGL { struct ScreenInfo; struct DrawParameters; +struct FramebufferCacheKey; class RasterizerOpenGL : public VideoCore::RasterizerInterface { public: @@ -195,11 +196,12 @@ private: OGLVertexArray> vertex_array_cache; + std::map<FramebufferCacheKey, OGLFramebuffer> framebuffer_cache; + std::array<SamplerInfo, Tegra::Engines::Maxwell3D::Regs::NumTextureSamplers> texture_samplers; static constexpr std::size_t STREAM_BUFFER_SIZE = 128 * 1024 * 1024; OGLBufferCache buffer_cache; - OGLFramebuffer framebuffer; PrimitiveAssembler primitive_assembler{buffer_cache}; GLint uniform_buffer_alignment; @@ -214,6 +216,8 @@ private: void SetupShaders(GLenum primitive_mode); + void SetupCachedFramebuffer(const FramebufferCacheKey& fbkey, OpenGLState& current_state); + enum class AccelDraw { Disabled, Arrays, Indexed }; AccelDraw accelerate_draw = AccelDraw::Disabled; diff --git a/src/video_core/renderer_opengl/gl_shader_cache.cpp b/src/video_core/renderer_opengl/gl_shader_cache.cpp index 038b25c75..a4265f498 100644 --- a/src/video_core/renderer_opengl/gl_shader_cache.cpp +++ b/src/video_core/renderer_opengl/gl_shader_cache.cpp @@ -2,7 +2,9 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <boost/functional/hash.hpp> #include "common/assert.h" +#include "common/hash.h" #include "core/core.h" #include "core/memory.h" #include "video_core/engines/maxwell_3d.h" @@ -66,14 +68,17 @@ CachedShader::CachedShader(VAddr addr, Maxwell::ShaderProgram program_type) // stage here. setup.SetProgramB(GetShaderCode(GetShaderAddress(Maxwell::ShaderProgram::VertexB))); case Maxwell::ShaderProgram::VertexB: + CalculateProperties(); program_result = GLShader::GenerateVertexShader(setup); gl_type = GL_VERTEX_SHADER; break; case Maxwell::ShaderProgram::Geometry: + CalculateProperties(); program_result = GLShader::GenerateGeometryShader(setup); gl_type = GL_GEOMETRY_SHADER; break; case Maxwell::ShaderProgram::Fragment: + CalculateProperties(); program_result = GLShader::GenerateFragmentShader(setup); gl_type = GL_FRAGMENT_SHADER; break; @@ -140,6 +145,46 @@ GLuint CachedShader::LazyGeometryProgram(OGLProgram& target_program, return target_program.handle; }; +static bool IsSchedInstruction(u32 offset, u32 main_offset) { + // sched instructions appear once every 4 instructions. + static constexpr std::size_t SchedPeriod = 4; + const std::size_t absolute_offset = offset - main_offset; + return (absolute_offset % SchedPeriod) == 0; +} + +static std::size_t CalculateProgramSize(const GLShader::ProgramCode& program) { + const std::size_t start_offset = 10; + std::size_t offset = start_offset; + std::size_t size = start_offset * sizeof(u64); + while (offset < program.size()) { + const u64 inst = program[offset]; + if (!IsSchedInstruction(offset, start_offset)) { + if (inst == 0 || (inst >> 52) == 0x50b) { + break; + } + } + size += 8; + offset++; + } + return size; +} + +void CachedShader::CalculateProperties() { + setup.program.real_size = CalculateProgramSize(setup.program.code); + setup.program.real_size_b = 0; + setup.program.unique_identifier = Common::CityHash64( + reinterpret_cast<const char*>(setup.program.code.data()), setup.program.real_size); + if (program_type == Maxwell::ShaderProgram::VertexA) { + std::size_t seed = 0; + boost::hash_combine(seed, setup.program.unique_identifier); + setup.program.real_size_b = CalculateProgramSize(setup.program.code_b); + const u64 identifier_b = Common::CityHash64( + reinterpret_cast<const char*>(setup.program.code_b.data()), setup.program.real_size_b); + boost::hash_combine(seed, identifier_b); + setup.program.unique_identifier = static_cast<u64>(seed); + } +} + ShaderCacheOpenGL::ShaderCacheOpenGL(RasterizerOpenGL& rasterizer) : RasterizerCache{rasterizer} {} Shader ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program) { diff --git a/src/video_core/renderer_opengl/gl_shader_cache.h b/src/video_core/renderer_opengl/gl_shader_cache.h index 08f470de3..b4ef6030d 100644 --- a/src/video_core/renderer_opengl/gl_shader_cache.h +++ b/src/video_core/renderer_opengl/gl_shader_cache.h @@ -81,6 +81,8 @@ private: GLuint LazyGeometryProgram(OGLProgram& target_program, const std::string& glsl_topology, u32 max_vertices, const std::string& debug_name); + void CalculateProperties(); + VAddr addr; std::size_t shader_length; Maxwell::ShaderProgram program_type; diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp index d235bfcd4..e7057a9cb 100644 --- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp +++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp @@ -50,6 +50,14 @@ public: using std::runtime_error::runtime_error; }; +/// Generates code to use for a swizzle operation. +static std::string GetSwizzle(u64 elem) { + ASSERT(elem <= 3); + std::string swizzle = "."; + swizzle += "xyzw"[elem]; + return swizzle; +} + /// Translate topology static std::string GetTopologyName(Tegra::Shader::OutputTopology topology) { switch (topology) { @@ -1004,14 +1012,6 @@ private: } } - /// Generates code to use for a swizzle operation. - static std::string GetSwizzle(u64 elem) { - ASSERT(elem <= 3); - std::string swizzle = "."; - swizzle += "xyzw"[elem]; - return swizzle; - } - ShaderWriter& shader; ShaderWriter& declarations; std::vector<GLSLRegister> regs; @@ -1343,7 +1343,7 @@ private: regs.SetRegisterToInteger(dest, true, 0, result, 1, 1); } - void WriteTexsInstruction(const Instruction& instr, const std::string& texture) { + void WriteTexsInstructionFloat(const Instruction& instr, const std::string& texture) { // TEXS has two destination registers and a swizzle. The first two elements in the swizzle // go into gpr0+0 and gpr0+1, and the rest goes into gpr28+0 and gpr28+1 @@ -1368,6 +1368,38 @@ private: } } + void WriteTexsInstructionHalfFloat(const Instruction& instr, const std::string& texture) { + // TEXS.F16 destionation registers are packed in two registers in pairs (just like any half + // float instruction). + + std::array<std::string, 4> components; + u32 written_components = 0; + + for (u32 component = 0; component < 4; ++component) { + if (!instr.texs.IsComponentEnabled(component)) + continue; + components[written_components++] = texture + GetSwizzle(component); + } + if (written_components == 0) + return; + + const auto BuildComponent = [&](std::string low, std::string high, bool high_enabled) { + return "vec2(" + low + ", " + (high_enabled ? high : "0") + ')'; + }; + + regs.SetRegisterToHalfFloat( + instr.gpr0, 0, BuildComponent(components[0], components[1], written_components > 1), + Tegra::Shader::HalfMerge::H0_H1, 1, 1); + + if (written_components > 2) { + ASSERT(instr.texs.HasTwoDestinations()); + regs.SetRegisterToHalfFloat( + instr.gpr28, 0, + BuildComponent(components[2], components[3], written_components > 3), + Tegra::Shader::HalfMerge::H0_H1, 1, 1); + } + } + static u32 TextureCoordinates(Tegra::Shader::TextureType texture_type) { switch (texture_type) { case Tegra::Shader::TextureType::Texture1D: @@ -1515,6 +1547,254 @@ private: } } + std::pair<size_t, std::string> ValidateAndGetCoordinateElement( + const Tegra::Shader::TextureType texture_type, const bool depth_compare, + const bool is_array, const bool lod_bias_enabled, size_t max_coords, size_t max_inputs) { + const size_t coord_count = TextureCoordinates(texture_type); + + size_t total_coord_count = coord_count + (is_array ? 1 : 0) + (depth_compare ? 1 : 0); + const size_t total_reg_count = total_coord_count + (lod_bias_enabled ? 1 : 0); + if (total_coord_count > max_coords || total_reg_count > max_inputs) { + UNIMPLEMENTED_MSG("Unsupported Texture operation"); + total_coord_count = std::min(total_coord_count, max_coords); + } + // 1D.DC opengl is using a vec3 but 2nd component is ignored later. + total_coord_count += + (depth_compare && !is_array && texture_type == Tegra::Shader::TextureType::Texture1D) + ? 1 + : 0; + + constexpr std::array<const char*, 5> coord_container{ + {"", "float coord = (", "vec2 coord = vec2(", "vec3 coord = vec3(", + "vec4 coord = vec4("}}; + + return std::pair<size_t, std::string>(coord_count, coord_container[total_coord_count]); + } + + std::string GetTextureCode(const Tegra::Shader::Instruction& instr, + const Tegra::Shader::TextureType texture_type, + const Tegra::Shader::TextureProcessMode process_mode, + const bool depth_compare, const bool is_array, + const size_t bias_offset) { + + if ((texture_type == Tegra::Shader::TextureType::Texture3D && + (is_array || depth_compare)) || + (texture_type == Tegra::Shader::TextureType::TextureCube && is_array && + depth_compare)) { + UNIMPLEMENTED_MSG("This method is not supported."); + } + + const std::string sampler = + GetSampler(instr.sampler, texture_type, is_array, depth_compare); + + const bool lod_needed = process_mode == Tegra::Shader::TextureProcessMode::LZ || + process_mode == Tegra::Shader::TextureProcessMode::LL || + process_mode == Tegra::Shader::TextureProcessMode::LLA; + + const bool gl_lod_supported = !( + (texture_type == Tegra::Shader::TextureType::Texture2D && is_array && depth_compare) || + (texture_type == Tegra::Shader::TextureType::TextureCube && !is_array && + depth_compare)); + + const std::string read_method = lod_needed && gl_lod_supported ? "textureLod(" : "texture("; + std::string texture = read_method + sampler + ", coord"; + + if (process_mode != Tegra::Shader::TextureProcessMode::None) { + if (process_mode == Tegra::Shader::TextureProcessMode::LZ) { + if (gl_lod_supported) { + texture += ", 0"; + } else { + // Lod 0 is emulated by a big negative bias + // in scenarios that are not supported by glsl + texture += ", -1000"; + } + } else { + // If present, lod or bias are always stored in the register indexed by the + // gpr20 + // field with an offset depending on the usage of the other registers + texture += ',' + regs.GetRegisterAsFloat(instr.gpr20.Value() + bias_offset); + } + } + texture += ")"; + return texture; + } + + std::pair<std::string, std::string> GetTEXCode( + const Instruction& instr, const Tegra::Shader::TextureType texture_type, + const Tegra::Shader::TextureProcessMode process_mode, const bool depth_compare, + const bool is_array) { + const bool lod_bias_enabled = (process_mode != Tegra::Shader::TextureProcessMode::None && + process_mode != Tegra::Shader::TextureProcessMode::LZ); + + const auto [coord_count, coord_dcl] = ValidateAndGetCoordinateElement( + texture_type, depth_compare, is_array, lod_bias_enabled, 4, 5); + // If enabled arrays index is always stored in the gpr8 field + const u64 array_register = instr.gpr8.Value(); + // First coordinate index is the gpr8 or gpr8 + 1 when arrays are used + const u64 coord_register = array_register + (is_array ? 1 : 0); + + std::string coord = coord_dcl; + for (size_t i = 0; i < coord_count;) { + coord += regs.GetRegisterAsFloat(coord_register + i); + ++i; + if (i != coord_count) { + coord += ','; + } + } + // 1D.DC in opengl the 2nd component is ignored. + if (depth_compare && !is_array && texture_type == Tegra::Shader::TextureType::Texture1D) { + coord += ",0.0"; + } + if (depth_compare) { + // Depth is always stored in the register signaled by gpr20 + // or in the next register if lod or bias are used + const u64 depth_register = instr.gpr20.Value() + (lod_bias_enabled ? 1 : 0); + coord += ',' + regs.GetRegisterAsFloat(depth_register); + } + if (is_array) { + coord += ',' + regs.GetRegisterAsInteger(array_register); + } + coord += ");"; + return std::make_pair( + coord, GetTextureCode(instr, texture_type, process_mode, depth_compare, is_array, 0)); + } + + std::pair<std::string, std::string> GetTEXSCode( + const Instruction& instr, const Tegra::Shader::TextureType texture_type, + const Tegra::Shader::TextureProcessMode process_mode, const bool depth_compare, + const bool is_array) { + const bool lod_bias_enabled = (process_mode != Tegra::Shader::TextureProcessMode::None && + process_mode != Tegra::Shader::TextureProcessMode::LZ); + + const auto [coord_count, coord_dcl] = ValidateAndGetCoordinateElement( + texture_type, depth_compare, is_array, lod_bias_enabled, 4, 4); + // If enabled arrays index is always stored in the gpr8 field + const u64 array_register = instr.gpr8.Value(); + // First coordinate index is stored in gpr8 field or (gpr8 + 1) when arrays are used + const u64 coord_register = array_register + (is_array ? 1 : 0); + const u64 last_coord_register = + (is_array || !(lod_bias_enabled || depth_compare) || (coord_count > 2)) + ? static_cast<u64>(instr.gpr20.Value()) + : coord_register + 1; + + std::string coord = coord_dcl; + for (size_t i = 0; i < coord_count; ++i) { + const bool last = (i == (coord_count - 1)) && (coord_count > 1); + coord += regs.GetRegisterAsFloat(last ? last_coord_register : coord_register + i); + if (i < coord_count - 1) { + coord += ','; + } + } + + if (depth_compare) { + // Depth is always stored in the register signaled by gpr20 + // or in the next register if lod or bias are used + const u64 depth_register = instr.gpr20.Value() + (lod_bias_enabled ? 1 : 0); + coord += ',' + regs.GetRegisterAsFloat(depth_register); + } + if (is_array) { + coord += ',' + regs.GetRegisterAsInteger(array_register); + } + coord += ");"; + + return std::make_pair(coord, + GetTextureCode(instr, texture_type, process_mode, depth_compare, + is_array, (coord_count > 2 ? 1 : 0))); + } + + std::pair<std::string, std::string> GetTLD4Code(const Instruction& instr, + const Tegra::Shader::TextureType texture_type, + const bool depth_compare, const bool is_array) { + + const size_t coord_count = TextureCoordinates(texture_type); + const size_t total_coord_count = coord_count + (is_array ? 1 : 0); + const size_t total_reg_count = total_coord_count + (depth_compare ? 1 : 0); + + constexpr std::array<const char*, 5> coord_container{ + {"", "", "vec2 coord = vec2(", "vec3 coord = vec3(", "vec4 coord = vec4("}}; + + // If enabled arrays index is always stored in the gpr8 field + const u64 array_register = instr.gpr8.Value(); + // First coordinate index is the gpr8 or gpr8 + 1 when arrays are used + const u64 coord_register = array_register + (is_array ? 1 : 0); + + std::string coord = coord_container[total_coord_count]; + for (size_t i = 0; i < coord_count;) { + coord += regs.GetRegisterAsFloat(coord_register + i); + ++i; + if (i != coord_count) { + coord += ','; + } + } + + if (is_array) { + coord += ',' + regs.GetRegisterAsInteger(array_register); + } + coord += ");"; + + const std::string sampler = + GetSampler(instr.sampler, texture_type, is_array, depth_compare); + + std::string texture = "textureGather(" + sampler + ", coord, "; + if (depth_compare) { + // Depth is always stored in the register signaled by gpr20 + texture += regs.GetRegisterAsFloat(instr.gpr20.Value()) + ')'; + } else { + texture += std::to_string(instr.tld4.component) + ')'; + } + return std::make_pair(coord, texture); + } + + std::pair<std::string, std::string> GetTLDSCode(const Instruction& instr, + const Tegra::Shader::TextureType texture_type, + const bool is_array) { + + const size_t coord_count = TextureCoordinates(texture_type); + const size_t total_coord_count = coord_count + (is_array ? 1 : 0); + const bool lod_enabled = + instr.tlds.GetTextureProcessMode() == Tegra::Shader::TextureProcessMode::LL; + + constexpr std::array<const char*, 4> coord_container{ + {"", "int coord = (", "ivec2 coord = ivec2(", "ivec3 coord = ivec3("}}; + + std::string coord = coord_container[total_coord_count]; + + // If enabled arrays index is always stored in the gpr8 field + const u64 array_register = instr.gpr8.Value(); + + // if is array gpr20 is used + const u64 coord_register = is_array ? instr.gpr20.Value() : instr.gpr8.Value(); + + const u64 last_coord_register = + ((coord_count > 2) || (coord_count == 2 && !lod_enabled)) && !is_array + ? static_cast<u64>(instr.gpr20.Value()) + : coord_register + 1; + + for (size_t i = 0; i < coord_count; ++i) { + const bool last = (i == (coord_count - 1)) && (coord_count > 1); + coord += regs.GetRegisterAsInteger(last ? last_coord_register : coord_register + i); + if (i < coord_count - 1) { + coord += ','; + } + } + if (is_array) { + coord += ',' + regs.GetRegisterAsInteger(array_register); + } + coord += ");"; + + const std::string sampler = GetSampler(instr.sampler, texture_type, is_array, false); + + std::string texture = "texelFetch(" + sampler + ", coords"; + + if (lod_enabled) { + // When lod is used always is in grp20 + texture += ", " + regs.GetRegisterAsInteger(instr.gpr20) + ')'; + } else { + texture += ", 0)"; + } + return std::make_pair(coord, texture); + } + /** * Compiles a single instruction from Tegra to GLSL. * @param offset the offset of the Tegra shader instruction. @@ -2574,168 +2854,32 @@ private: case OpCode::Id::TEX: { Tegra::Shader::TextureType texture_type{instr.tex.texture_type}; const bool is_array = instr.tex.array != 0; - + const bool depth_compare = + instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC); + const auto process_mode = instr.tex.GetTextureProcessMode(); UNIMPLEMENTED_IF_MSG(instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP), "NODEP is not implemented"); UNIMPLEMENTED_IF_MSG(instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::AOFFI), "AOFFI is not implemented"); - const bool depth_compare = - instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC); - u32 num_coordinates = TextureCoordinates(texture_type); - u32 start_index = 0; - std::string array_elem; - if (is_array) { - array_elem = regs.GetRegisterAsInteger(instr.gpr8); - start_index = 1; - } - const auto process_mode = instr.tex.GetTextureProcessMode(); - u32 start_index_b = 0; - std::string lod_value; - if (process_mode != Tegra::Shader::TextureProcessMode::LZ && - process_mode != Tegra::Shader::TextureProcessMode::None) { - start_index_b = 1; - lod_value = regs.GetRegisterAsFloat(instr.gpr20); - } - - std::string depth_value; - if (depth_compare) { - depth_value = regs.GetRegisterAsFloat(instr.gpr20.Value() + start_index_b); - } - - bool depth_compare_extra = false; + const auto [coord, texture] = + GetTEXCode(instr, texture_type, process_mode, depth_compare, is_array); const auto scope = shader.Scope(); - - switch (num_coordinates) { - case 1: { - const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index); - if (is_array) { - if (depth_compare) { - shader.AddLine("vec3 coords = vec3(" + x + ", " + depth_value + ", " + - array_elem + ");"); - } else { - shader.AddLine("vec2 coords = vec2(" + x + ", " + array_elem + ");"); - } - } else { - if (depth_compare) { - shader.AddLine("vec2 coords = vec2(" + x + ", " + depth_value + ");"); - } else { - shader.AddLine("float coords = " + x + ';'); - } - } - break; - } - case 2: { - const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index); - const std::string y = - regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index + 1); - if (is_array) { - if (depth_compare) { - shader.AddLine("vec4 coords = vec4(" + x + ", " + y + ", " + - depth_value + ", " + array_elem + ");"); - } else { - shader.AddLine("vec3 coords = vec3(" + x + ", " + y + ", " + - array_elem + ");"); - } - } else { - if (depth_compare) { - shader.AddLine("vec3 coords = vec3(" + x + ", " + y + ", " + - depth_value + ");"); - } else { - shader.AddLine("vec2 coords = vec2(" + x + ", " + y + ");"); - } - } - break; - } - case 3: { - const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index); - const std::string y = - regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index + 1); - const std::string z = - regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index + 2); - if (is_array) { - depth_compare_extra = depth_compare; - shader.AddLine("vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " + - array_elem + ");"); - } else { - if (depth_compare) { - shader.AddLine("vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " + - depth_value + ");"); - } else { - shader.AddLine("vec3 coords = vec3(" + x + ", " + y + ", " + z + ");"); - } - } - break; - } - default: - UNIMPLEMENTED_MSG("Unhandled coordinates number {}", - static_cast<u32>(num_coordinates)); - - // Fallback to interpreting as a 2D texture for now - const std::string x = regs.GetRegisterAsFloat(instr.gpr8); - const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1); - shader.AddLine("vec2 coords = vec2(" + x + ", " + y + ");"); - texture_type = Tegra::Shader::TextureType::Texture2D; - } - - const std::string sampler = - GetSampler(instr.sampler, texture_type, is_array, depth_compare); - // Add an extra scope and declare the texture coords inside to prevent - // overwriting them in case they are used as outputs of the texs instruction. - - const std::string texture = [&]() { - switch (instr.tex.GetTextureProcessMode()) { - case Tegra::Shader::TextureProcessMode::None: - if (depth_compare_extra) { - return "texture(" + sampler + ", coords, " + depth_value + ')'; - } - return "texture(" + sampler + ", coords)"; - case Tegra::Shader::TextureProcessMode::LZ: - if (depth_compare_extra) { - return "texture(" + sampler + ", coords, " + depth_value + ')'; - } - return "textureLod(" + sampler + ", coords, 0.0)"; - case Tegra::Shader::TextureProcessMode::LB: - case Tegra::Shader::TextureProcessMode::LBA: - // TODO: Figure if A suffix changes the equation at all. - if (depth_compare_extra) { - LOG_WARNING( - HW_GPU, - "OpenGL Limitation: can't set bias value along depth compare"); - return "texture(" + sampler + ", coords, " + depth_value + ')'; - } - return "texture(" + sampler + ", coords, " + lod_value + ')'; - case Tegra::Shader::TextureProcessMode::LL: - case Tegra::Shader::TextureProcessMode::LLA: - // TODO: Figure if A suffix changes the equation at all. - if (depth_compare_extra) { - LOG_WARNING( - HW_GPU, - "OpenGL Limitation: can't set lod value along depth compare"); - return "texture(" + sampler + ", coords, " + depth_value + ')'; - } - return "textureLod(" + sampler + ", coords, " + lod_value + ')'; - default: - UNIMPLEMENTED_MSG("Unhandled texture process mode {}", - static_cast<u32>(instr.tex.GetTextureProcessMode())); - if (depth_compare_extra) { - return "texture(" + sampler + ", coords, " + depth_value + ')'; - } - return "texture(" + sampler + ", coords)"; - } - }(); + shader.AddLine(coord); if (depth_compare) { regs.SetRegisterToFloat(instr.gpr0, 0, texture, 1, 1, false); } else { + shader.AddLine("vec4 texture_tmp = " + texture + ';'); std::size_t dest_elem{}; for (std::size_t elem = 0; elem < 4; ++elem) { if (!instr.tex.IsComponentEnabled(elem)) { // Skip disabled components continue; } - regs.SetRegisterToFloat(instr.gpr0, elem, texture, 1, 4, false, dest_elem); + regs.SetRegisterToFloat(instr.gpr0, elem, "texture_tmp", 1, 4, false, + dest_elem); ++dest_elem; } } @@ -2743,138 +2887,37 @@ private: } case OpCode::Id::TEXS: { Tegra::Shader::TextureType texture_type{instr.texs.GetTextureType()}; - bool is_array{instr.texs.IsArrayTexture()}; + const bool is_array{instr.texs.IsArrayTexture()}; + const bool depth_compare = + instr.texs.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC); + const auto process_mode = instr.texs.GetTextureProcessMode(); UNIMPLEMENTED_IF_MSG(instr.texs.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP), "NODEP is not implemented"); const auto scope = shader.Scope(); - const bool depth_compare = - instr.texs.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC); - u32 num_coordinates = TextureCoordinates(texture_type); - const auto process_mode = instr.texs.GetTextureProcessMode(); - u32 lod_offset = 0; - if (process_mode == Tegra::Shader::TextureProcessMode::LL) { - if (num_coordinates > 2) { - shader.AddLine("float lod_value = " + - regs.GetRegisterAsFloat(instr.gpr20.Value() + 1) + ';'); - lod_offset = 2; - } else { - shader.AddLine("float lod_value = " + regs.GetRegisterAsFloat(instr.gpr20) + - ';'); - lod_offset = 1; - } - } + auto [coord, texture] = + GetTEXSCode(instr, texture_type, process_mode, depth_compare, is_array); - switch (num_coordinates) { - case 1: { - shader.AddLine("float coords = " + regs.GetRegisterAsFloat(instr.gpr8) + ';'); - break; - } - case 2: { - if (is_array) { - if (depth_compare) { - const std::string index = regs.GetRegisterAsInteger(instr.gpr8); - const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1); - const std::string y = regs.GetRegisterAsFloat(instr.gpr20); - const std::string z = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1); - shader.AddLine("vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " + - index + ");"); - } else { - const std::string index = regs.GetRegisterAsInteger(instr.gpr8); - const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1); - const std::string y = regs.GetRegisterAsFloat(instr.gpr20); - shader.AddLine("vec3 coords = vec3(" + x + ", " + y + ", " + index + - ");"); - } - } else { - if (lod_offset != 0) { - if (depth_compare) { - const std::string x = regs.GetRegisterAsFloat(instr.gpr8); - const std::string y = - regs.GetRegisterAsFloat(instr.gpr8.Value() + 1); - const std::string z = - regs.GetRegisterAsFloat(instr.gpr20.Value() + lod_offset); - shader.AddLine("vec3 coords = vec3(" + x + ", " + y + ", " + z + - ");"); - } else { - const std::string x = regs.GetRegisterAsFloat(instr.gpr8); - const std::string y = - regs.GetRegisterAsFloat(instr.gpr8.Value() + 1); - shader.AddLine("vec2 coords = vec2(" + x + ", " + y + ");"); - } - } else { - if (depth_compare) { - const std::string x = regs.GetRegisterAsFloat(instr.gpr8); - const std::string y = - regs.GetRegisterAsFloat(instr.gpr8.Value() + 1); - const std::string z = regs.GetRegisterAsFloat(instr.gpr20); - shader.AddLine("vec3 coords = vec3(" + x + ", " + y + ", " + z + - ");"); - } else { - const std::string x = regs.GetRegisterAsFloat(instr.gpr8); - const std::string y = regs.GetRegisterAsFloat(instr.gpr20); - shader.AddLine("vec2 coords = vec2(" + x + ", " + y + ");"); - } - } - } - break; - } - case 3: { - const std::string x = regs.GetRegisterAsFloat(instr.gpr8); - const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1); - const std::string z = regs.GetRegisterAsFloat(instr.gpr20); - shader.AddLine("vec3 coords = vec3(" + x + ", " + y + ", " + z + ");"); - break; - } - default: - UNIMPLEMENTED_MSG("Unhandled coordinates number {}", - static_cast<u32>(num_coordinates)); + shader.AddLine(coord); - // Fallback to interpreting as a 2D texture for now - const std::string x = regs.GetRegisterAsFloat(instr.gpr8); - const std::string y = regs.GetRegisterAsFloat(instr.gpr20); - shader.AddLine("vec2 coords = vec2(" + x + ", " + y + ");"); - texture_type = Tegra::Shader::TextureType::Texture2D; - is_array = false; - } - const std::string sampler = - GetSampler(instr.sampler, texture_type, is_array, depth_compare); - - std::string texture = [&]() { - switch (process_mode) { - case Tegra::Shader::TextureProcessMode::None: - return "texture(" + sampler + ", coords)"; - case Tegra::Shader::TextureProcessMode::LZ: - if (depth_compare && is_array) { - return "texture(" + sampler + ", coords)"; - } else { - return "textureLod(" + sampler + ", coords, 0.0)"; - } - break; - case Tegra::Shader::TextureProcessMode::LL: - return "textureLod(" + sampler + ", coords, lod_value)"; - default: - UNIMPLEMENTED_MSG("Unhandled texture process mode {}", - static_cast<u32>(instr.texs.GetTextureProcessMode())); - return "texture(" + sampler + ", coords)"; - } - }(); if (depth_compare) { texture = "vec4(" + texture + ')'; } + shader.AddLine("vec4 texture_tmp = " + texture + ';'); - WriteTexsInstruction(instr, texture); + if (instr.texs.fp32_flag) { + WriteTexsInstructionFloat(instr, "texture_tmp"); + } else { + WriteTexsInstructionHalfFloat(instr, "texture_tmp"); + } break; } case OpCode::Id::TLDS: { const Tegra::Shader::TextureType texture_type{instr.tlds.GetTextureType()}; const bool is_array{instr.tlds.IsArrayTexture()}; - ASSERT(texture_type == Tegra::Shader::TextureType::Texture2D); - ASSERT(is_array == false); - UNIMPLEMENTED_IF_MSG(instr.tlds.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP), "NODEP is not implemented"); UNIMPLEMENTED_IF_MSG(instr.tlds.UsesMiscMode(Tegra::Shader::TextureMiscMode::AOFFI), @@ -2882,54 +2925,16 @@ private: UNIMPLEMENTED_IF_MSG(instr.tlds.UsesMiscMode(Tegra::Shader::TextureMiscMode::MZ), "MZ is not implemented"); - u32 extra_op_offset = 0; + const auto [coord, texture] = GetTLDSCode(instr, texture_type, is_array); - ShaderScopedScope scope = shader.Scope(); - - switch (texture_type) { - case Tegra::Shader::TextureType::Texture1D: { - const std::string x = regs.GetRegisterAsInteger(instr.gpr8); - shader.AddLine("float coords = " + x + ';'); - break; - } - case Tegra::Shader::TextureType::Texture2D: { - UNIMPLEMENTED_IF_MSG(is_array, "Unhandled 2d array texture"); - - const std::string x = regs.GetRegisterAsInteger(instr.gpr8); - const std::string y = regs.GetRegisterAsInteger(instr.gpr20); - // shader.AddLine("ivec2 coords = ivec2(" + x + ", " + y + ");"); - shader.AddLine("ivec2 coords = ivec2(" + x + ", " + y + ");"); - extra_op_offset = 1; - break; - } - default: - UNIMPLEMENTED_MSG("Unhandled texture type {}", static_cast<u32>(texture_type)); - } - const std::string sampler = - GetSampler(instr.sampler, texture_type, is_array, false); - - const std::string texture = [&]() { - switch (instr.tlds.GetTextureProcessMode()) { - case Tegra::Shader::TextureProcessMode::LZ: - return "texelFetch(" + sampler + ", coords, 0)"; - case Tegra::Shader::TextureProcessMode::LL: - shader.AddLine( - "float lod = " + - regs.GetRegisterAsInteger(instr.gpr20.Value() + extra_op_offset) + ';'); - return "texelFetch(" + sampler + ", coords, lod)"; - default: - UNIMPLEMENTED_MSG("Unhandled texture process mode {}", - static_cast<u32>(instr.tlds.GetTextureProcessMode())); - return "texelFetch(" + sampler + ", coords, 0)"; - } - }(); + const auto scope = shader.Scope(); - WriteTexsInstruction(instr, texture); + shader.AddLine(coord); + shader.AddLine("vec4 texture_tmp = " + texture + ';'); + WriteTexsInstructionFloat(instr, "texture_tmp"); break; } case OpCode::Id::TLD4: { - ASSERT(instr.tld4.texture_type == Tegra::Shader::TextureType::Texture2D); - ASSERT(instr.tld4.array == 0); UNIMPLEMENTED_IF_MSG(instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP), "NODEP is not implemented"); @@ -2939,56 +2944,29 @@ private: "NDV is not implemented"); UNIMPLEMENTED_IF_MSG(instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::PTP), "PTP is not implemented"); + + auto texture_type = instr.tld4.texture_type.Value(); const bool depth_compare = instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC); - auto texture_type = instr.tld4.texture_type.Value(); - u32 num_coordinates = TextureCoordinates(texture_type); - if (depth_compare) - num_coordinates += 1; - - const auto scope = shader.Scope(); + const bool is_array = instr.tld4.array != 0; - switch (num_coordinates) { - case 2: { - const std::string x = regs.GetRegisterAsFloat(instr.gpr8); - const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1); - shader.AddLine("vec2 coords = vec2(" + x + ", " + y + ");"); - break; - } - case 3: { - const std::string x = regs.GetRegisterAsFloat(instr.gpr8); - const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1); - const std::string z = regs.GetRegisterAsFloat(instr.gpr8.Value() + 2); - shader.AddLine("vec3 coords = vec3(" + x + ", " + y + ", " + z + ");"); - break; - } - default: - UNIMPLEMENTED_MSG("Unhandled coordinates number {}", - static_cast<u32>(num_coordinates)); - const std::string x = regs.GetRegisterAsFloat(instr.gpr8); - const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1); - shader.AddLine("vec2 coords = vec2(" + x + ", " + y + ");"); - texture_type = Tegra::Shader::TextureType::Texture2D; - } + const auto [coord, texture] = + GetTLD4Code(instr, texture_type, depth_compare, is_array); - const std::string sampler = - GetSampler(instr.sampler, texture_type, false, depth_compare); + const auto scope = shader.Scope(); - const std::string texture = "textureGather(" + sampler + ", coords, " + - std::to_string(instr.tld4.component) + ')'; + shader.AddLine(coord); + std::size_t dest_elem{}; - if (depth_compare) { - regs.SetRegisterToFloat(instr.gpr0, 0, texture, 1, 1, false); - } else { - std::size_t dest_elem{}; - for (std::size_t elem = 0; elem < 4; ++elem) { - if (!instr.tex.IsComponentEnabled(elem)) { - // Skip disabled components - continue; - } - regs.SetRegisterToFloat(instr.gpr0, elem, texture, 1, 4, false, dest_elem); - ++dest_elem; + shader.AddLine("vec4 texture_tmp = " + texture + ';'); + for (std::size_t elem = 0; elem < 4; ++elem) { + if (!instr.tex.IsComponentEnabled(elem)) { + // Skip disabled components + continue; } + regs.SetRegisterToFloat(instr.gpr0, elem, "texture_tmp", 1, 4, false, + dest_elem); + ++dest_elem; } break; } @@ -3002,27 +2980,31 @@ private: const auto scope = shader.Scope(); + std::string coords; + const bool depth_compare = instr.tld4s.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC); - const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8); - const std::string op_b = regs.GetRegisterAsFloat(instr.gpr20); - // TODO(Subv): Figure out how the sampler type is encoded in the TLD4S instruction. + const std::string sampler = GetSampler( instr.sampler, Tegra::Shader::TextureType::Texture2D, false, depth_compare); - if (depth_compare) { - // Note: TLD4S coordinate encoding works just like TEXS's - const std::string op_y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1); - shader.AddLine("vec3 coords = vec3(" + op_a + ", " + op_y + ", " + op_b + ");"); - } else { - shader.AddLine("vec2 coords = vec2(" + op_a + ", " + op_b + ");"); - } - std::string texture = "textureGather(" + sampler + ", coords, " + - std::to_string(instr.tld4s.component) + ')'; - if (depth_compare) { - texture = "vec4(" + texture + ')'; + const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8); + coords = "vec2 coords = vec2(" + op_a + ", "; + std::string texture = "textureGather(" + sampler + ", coords, "; + + if (!depth_compare) { + const std::string op_b = regs.GetRegisterAsFloat(instr.gpr20); + coords += op_b + ");"; + texture += std::to_string(instr.tld4s.component) + ')'; + } else { + const std::string op_b = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1); + const std::string op_c = regs.GetRegisterAsFloat(instr.gpr20); + coords += op_b + ");"; + texture += op_c + ')'; } - WriteTexsInstruction(instr, texture); + shader.AddLine(coords); + shader.AddLine("vec4 texture_tmp = " + texture + ';'); + WriteTexsInstructionFloat(instr, "texture_tmp"); break; } case OpCode::Id::TXQ: { @@ -3934,4 +3916,4 @@ std::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u return {}; } -} // namespace OpenGL::GLShader::Decompiler +} // namespace OpenGL::GLShader::Decompiler
\ No newline at end of file diff --git a/src/video_core/renderer_opengl/gl_shader_gen.cpp b/src/video_core/renderer_opengl/gl_shader_gen.cpp index 23ed91e27..5d0819dc5 100644 --- a/src/video_core/renderer_opengl/gl_shader_gen.cpp +++ b/src/video_core/renderer_opengl/gl_shader_gen.cpp @@ -2,6 +2,7 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <fmt/format.h> #include "common/assert.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/renderer_opengl/gl_shader_decompiler.h" @@ -16,6 +17,8 @@ static constexpr u32 PROGRAM_OFFSET{10}; ProgramResult GenerateVertexShader(const ShaderSetup& setup) { std::string out = "#version 430 core\n"; out += "#extension GL_ARB_separate_shader_objects : enable\n\n"; + const std::string id = fmt::format("{:016x}", setup.program.unique_identifier); + out += "// Shader Unique Id: VS" + id + "\n\n"; out += Decompiler::GetCommonDeclarations(); out += R"( @@ -84,6 +87,8 @@ void main() { ProgramResult GenerateGeometryShader(const ShaderSetup& setup) { // Version is intentionally skipped in shader generation, it's added by the lazy compilation. std::string out = "#extension GL_ARB_separate_shader_objects : enable\n\n"; + const std::string id = fmt::format("{:016x}", setup.program.unique_identifier); + out += "// Shader Unique Id: GS" + id + "\n\n"; out += Decompiler::GetCommonDeclarations(); out += "bool exec_geometry();\n"; @@ -117,6 +122,8 @@ void main() { ProgramResult GenerateFragmentShader(const ShaderSetup& setup) { std::string out = "#version 430 core\n"; out += "#extension GL_ARB_separate_shader_objects : enable\n\n"; + const std::string id = fmt::format("{:016x}", setup.program.unique_identifier); + out += "// Shader Unique Id: FS" + id + "\n\n"; out += Decompiler::GetCommonDeclarations(); out += "bool exec_fragment();\n"; diff --git a/src/video_core/renderer_opengl/gl_shader_gen.h b/src/video_core/renderer_opengl/gl_shader_gen.h index 4fa6d7612..fcc20d3b4 100644 --- a/src/video_core/renderer_opengl/gl_shader_gen.h +++ b/src/video_core/renderer_opengl/gl_shader_gen.h @@ -177,6 +177,9 @@ struct ShaderSetup { struct { ProgramCode code; ProgramCode code_b; // Used for dual vertex shaders + u64 unique_identifier; + std::size_t real_size; + std::size_t real_size_b; } program; /// Used in scenarios where we have a dual vertex shaders diff --git a/src/video_core/surface.h b/src/video_core/surface.h index 0dd3eb2e4..e23cfecbc 100644 --- a/src/video_core/surface.h +++ b/src/video_core/surface.h @@ -125,6 +125,75 @@ enum class SurfaceTarget { TextureCubeArray, }; +constexpr std::array<u32, MaxPixelFormat> compression_factor_table = {{ + 1, // ABGR8U + 1, // ABGR8S + 1, // ABGR8UI + 1, // B5G6R5U + 1, // A2B10G10R10U + 1, // A1B5G5R5U + 1, // R8U + 1, // R8UI + 1, // RGBA16F + 1, // RGBA16U + 1, // RGBA16UI + 1, // R11FG11FB10F + 1, // RGBA32UI + 4, // DXT1 + 4, // DXT23 + 4, // DXT45 + 4, // DXN1 + 4, // DXN2UNORM + 4, // DXN2SNORM + 4, // BC7U + 4, // BC6H_UF16 + 4, // BC6H_SF16 + 4, // ASTC_2D_4X4 + 1, // G8R8U + 1, // G8R8S + 1, // BGRA8 + 1, // RGBA32F + 1, // RG32F + 1, // R32F + 1, // R16F + 1, // R16U + 1, // R16S + 1, // R16UI + 1, // R16I + 1, // RG16 + 1, // RG16F + 1, // RG16UI + 1, // RG16I + 1, // RG16S + 1, // RGB32F + 1, // RGBA8_SRGB + 1, // RG8U + 1, // RG8S + 1, // RG32UI + 1, // R32UI + 4, // ASTC_2D_8X8 + 4, // ASTC_2D_8X5 + 4, // ASTC_2D_5X4 + 1, // BGRA8_SRGB + 4, // DXT1_SRGB + 4, // DXT23_SRGB + 4, // DXT45_SRGB + 4, // BC7U_SRGB + 4, // ASTC_2D_4X4_SRGB + 4, // ASTC_2D_8X8_SRGB + 4, // ASTC_2D_8X5_SRGB + 4, // ASTC_2D_5X4_SRGB + 4, // ASTC_2D_5X5 + 4, // ASTC_2D_5X5_SRGB + 4, // ASTC_2D_10X8 + 4, // ASTC_2D_10X8_SRGB + 1, // Z32F + 1, // Z16 + 1, // Z24S8 + 1, // S8Z24 + 1, // Z32FS8 +}}; + /** * Gets the compression factor for the specified PixelFormat. This applies to just the * "compressed width" and "compressed height", not the overall compression factor of a @@ -135,304 +204,237 @@ static constexpr u32 GetCompressionFactor(PixelFormat format) { if (format == PixelFormat::Invalid) return 0; - constexpr std::array<u32, MaxPixelFormat> compression_factor_table = {{ - 1, // ABGR8U - 1, // ABGR8S - 1, // ABGR8UI - 1, // B5G6R5U - 1, // A2B10G10R10U - 1, // A1B5G5R5U - 1, // R8U - 1, // R8UI - 1, // RGBA16F - 1, // RGBA16U - 1, // RGBA16UI - 1, // R11FG11FB10F - 1, // RGBA32UI - 4, // DXT1 - 4, // DXT23 - 4, // DXT45 - 4, // DXN1 - 4, // DXN2UNORM - 4, // DXN2SNORM - 4, // BC7U - 4, // BC6H_UF16 - 4, // BC6H_SF16 - 4, // ASTC_2D_4X4 - 1, // G8R8U - 1, // G8R8S - 1, // BGRA8 - 1, // RGBA32F - 1, // RG32F - 1, // R32F - 1, // R16F - 1, // R16U - 1, // R16S - 1, // R16UI - 1, // R16I - 1, // RG16 - 1, // RG16F - 1, // RG16UI - 1, // RG16I - 1, // RG16S - 1, // RGB32F - 1, // RGBA8_SRGB - 1, // RG8U - 1, // RG8S - 1, // RG32UI - 1, // R32UI - 4, // ASTC_2D_8X8 - 4, // ASTC_2D_8X5 - 4, // ASTC_2D_5X4 - 1, // BGRA8_SRGB - 4, // DXT1_SRGB - 4, // DXT23_SRGB - 4, // DXT45_SRGB - 4, // BC7U_SRGB - 4, // ASTC_2D_4X4_SRGB - 4, // ASTC_2D_8X8_SRGB - 4, // ASTC_2D_8X5_SRGB - 4, // ASTC_2D_5X4_SRGB - 4, // ASTC_2D_5X5 - 4, // ASTC_2D_5X5_SRGB - 4, // ASTC_2D_10X8 - 4, // ASTC_2D_10X8_SRGB - 1, // Z32F - 1, // Z16 - 1, // Z24S8 - 1, // S8Z24 - 1, // Z32FS8 - }}; - ASSERT(static_cast<std::size_t>(format) < compression_factor_table.size()); return compression_factor_table[static_cast<std::size_t>(format)]; } +constexpr std::array<u32, MaxPixelFormat> block_width_table = {{ + 1, // ABGR8U + 1, // ABGR8S + 1, // ABGR8UI + 1, // B5G6R5U + 1, // A2B10G10R10U + 1, // A1B5G5R5U + 1, // R8U + 1, // R8UI + 1, // RGBA16F + 1, // RGBA16U + 1, // RGBA16UI + 1, // R11FG11FB10F + 1, // RGBA32UI + 4, // DXT1 + 4, // DXT23 + 4, // DXT45 + 4, // DXN1 + 4, // DXN2UNORM + 4, // DXN2SNORM + 4, // BC7U + 4, // BC6H_UF16 + 4, // BC6H_SF16 + 4, // ASTC_2D_4X4 + 1, // G8R8U + 1, // G8R8S + 1, // BGRA8 + 1, // RGBA32F + 1, // RG32F + 1, // R32F + 1, // R16F + 1, // R16U + 1, // R16S + 1, // R16UI + 1, // R16I + 1, // RG16 + 1, // RG16F + 1, // RG16UI + 1, // RG16I + 1, // RG16S + 1, // RGB32F + 1, // RGBA8_SRGB + 1, // RG8U + 1, // RG8S + 1, // RG32UI + 1, // R32UI + 8, // ASTC_2D_8X8 + 8, // ASTC_2D_8X5 + 5, // ASTC_2D_5X4 + 1, // BGRA8_SRGB + 4, // DXT1_SRGB + 4, // DXT23_SRGB + 4, // DXT45_SRGB + 4, // BC7U_SRGB + 4, // ASTC_2D_4X4_SRGB + 8, // ASTC_2D_8X8_SRGB + 8, // ASTC_2D_8X5_SRGB + 5, // ASTC_2D_5X4_SRGB + 5, // ASTC_2D_5X5 + 5, // ASTC_2D_5X5_SRGB + 10, // ASTC_2D_10X8 + 10, // ASTC_2D_10X8_SRGB + 1, // Z32F + 1, // Z16 + 1, // Z24S8 + 1, // S8Z24 + 1, // Z32FS8 +}}; + static constexpr u32 GetDefaultBlockWidth(PixelFormat format) { if (format == PixelFormat::Invalid) return 0; - constexpr std::array<u32, MaxPixelFormat> block_width_table = {{ - 1, // ABGR8U - 1, // ABGR8S - 1, // ABGR8UI - 1, // B5G6R5U - 1, // A2B10G10R10U - 1, // A1B5G5R5U - 1, // R8U - 1, // R8UI - 1, // RGBA16F - 1, // RGBA16U - 1, // RGBA16UI - 1, // R11FG11FB10F - 1, // RGBA32UI - 4, // DXT1 - 4, // DXT23 - 4, // DXT45 - 4, // DXN1 - 4, // DXN2UNORM - 4, // DXN2SNORM - 4, // BC7U - 4, // BC6H_UF16 - 4, // BC6H_SF16 - 4, // ASTC_2D_4X4 - 1, // G8R8U - 1, // G8R8S - 1, // BGRA8 - 1, // RGBA32F - 1, // RG32F - 1, // R32F - 1, // R16F - 1, // R16U - 1, // R16S - 1, // R16UI - 1, // R16I - 1, // RG16 - 1, // RG16F - 1, // RG16UI - 1, // RG16I - 1, // RG16S - 1, // RGB32F - 1, // RGBA8_SRGB - 1, // RG8U - 1, // RG8S - 1, // RG32UI - 1, // R32UI - 8, // ASTC_2D_8X8 - 8, // ASTC_2D_8X5 - 5, // ASTC_2D_5X4 - 1, // BGRA8_SRGB - 4, // DXT1_SRGB - 4, // DXT23_SRGB - 4, // DXT45_SRGB - 4, // BC7U_SRGB - 4, // ASTC_2D_4X4_SRGB - 8, // ASTC_2D_8X8_SRGB - 8, // ASTC_2D_8X5_SRGB - 5, // ASTC_2D_5X4_SRGB - 5, // ASTC_2D_5X5 - 5, // ASTC_2D_5X5_SRGB - 10, // ASTC_2D_10X8 - 10, // ASTC_2D_10X8_SRGB - 1, // Z32F - 1, // Z16 - 1, // Z24S8 - 1, // S8Z24 - 1, // Z32FS8 - }}; + ASSERT(static_cast<std::size_t>(format) < block_width_table.size()); return block_width_table[static_cast<std::size_t>(format)]; } +constexpr std::array<u32, MaxPixelFormat> block_height_table = {{ + 1, // ABGR8U + 1, // ABGR8S + 1, // ABGR8UI + 1, // B5G6R5U + 1, // A2B10G10R10U + 1, // A1B5G5R5U + 1, // R8U + 1, // R8UI + 1, // RGBA16F + 1, // RGBA16U + 1, // RGBA16UI + 1, // R11FG11FB10F + 1, // RGBA32UI + 4, // DXT1 + 4, // DXT23 + 4, // DXT45 + 4, // DXN1 + 4, // DXN2UNORM + 4, // DXN2SNORM + 4, // BC7U + 4, // BC6H_UF16 + 4, // BC6H_SF16 + 4, // ASTC_2D_4X4 + 1, // G8R8U + 1, // G8R8S + 1, // BGRA8 + 1, // RGBA32F + 1, // RG32F + 1, // R32F + 1, // R16F + 1, // R16U + 1, // R16S + 1, // R16UI + 1, // R16I + 1, // RG16 + 1, // RG16F + 1, // RG16UI + 1, // RG16I + 1, // RG16S + 1, // RGB32F + 1, // RGBA8_SRGB + 1, // RG8U + 1, // RG8S + 1, // RG32UI + 1, // R32UI + 8, // ASTC_2D_8X8 + 5, // ASTC_2D_8X5 + 4, // ASTC_2D_5X4 + 1, // BGRA8_SRGB + 4, // DXT1_SRGB + 4, // DXT23_SRGB + 4, // DXT45_SRGB + 4, // BC7U_SRGB + 4, // ASTC_2D_4X4_SRGB + 8, // ASTC_2D_8X8_SRGB + 5, // ASTC_2D_8X5_SRGB + 4, // ASTC_2D_5X4_SRGB + 5, // ASTC_2D_5X5 + 5, // ASTC_2D_5X5_SRGB + 8, // ASTC_2D_10X8 + 8, // ASTC_2D_10X8_SRGB + 1, // Z32F + 1, // Z16 + 1, // Z24S8 + 1, // S8Z24 + 1, // Z32FS8 +}}; + static constexpr u32 GetDefaultBlockHeight(PixelFormat format) { if (format == PixelFormat::Invalid) return 0; - constexpr std::array<u32, MaxPixelFormat> block_height_table = {{ - 1, // ABGR8U - 1, // ABGR8S - 1, // ABGR8UI - 1, // B5G6R5U - 1, // A2B10G10R10U - 1, // A1B5G5R5U - 1, // R8U - 1, // R8UI - 1, // RGBA16F - 1, // RGBA16U - 1, // RGBA16UI - 1, // R11FG11FB10F - 1, // RGBA32UI - 4, // DXT1 - 4, // DXT23 - 4, // DXT45 - 4, // DXN1 - 4, // DXN2UNORM - 4, // DXN2SNORM - 4, // BC7U - 4, // BC6H_UF16 - 4, // BC6H_SF16 - 4, // ASTC_2D_4X4 - 1, // G8R8U - 1, // G8R8S - 1, // BGRA8 - 1, // RGBA32F - 1, // RG32F - 1, // R32F - 1, // R16F - 1, // R16U - 1, // R16S - 1, // R16UI - 1, // R16I - 1, // RG16 - 1, // RG16F - 1, // RG16UI - 1, // RG16I - 1, // RG16S - 1, // RGB32F - 1, // RGBA8_SRGB - 1, // RG8U - 1, // RG8S - 1, // RG32UI - 1, // R32UI - 8, // ASTC_2D_8X8 - 5, // ASTC_2D_8X5 - 4, // ASTC_2D_5X4 - 1, // BGRA8_SRGB - 4, // DXT1_SRGB - 4, // DXT23_SRGB - 4, // DXT45_SRGB - 4, // BC7U_SRGB - 4, // ASTC_2D_4X4_SRGB - 8, // ASTC_2D_8X8_SRGB - 5, // ASTC_2D_8X5_SRGB - 4, // ASTC_2D_5X4_SRGB - 5, // ASTC_2D_5X5 - 5, // ASTC_2D_5X5_SRGB - 8, // ASTC_2D_10X8 - 8, // ASTC_2D_10X8_SRGB - 1, // Z32F - 1, // Z16 - 1, // Z24S8 - 1, // S8Z24 - 1, // Z32FS8 - }}; - ASSERT(static_cast<std::size_t>(format) < block_height_table.size()); return block_height_table[static_cast<std::size_t>(format)]; } +constexpr std::array<u32, MaxPixelFormat> bpp_table = {{ + 32, // ABGR8U + 32, // ABGR8S + 32, // ABGR8UI + 16, // B5G6R5U + 32, // A2B10G10R10U + 16, // A1B5G5R5U + 8, // R8U + 8, // R8UI + 64, // RGBA16F + 64, // RGBA16U + 64, // RGBA16UI + 32, // R11FG11FB10F + 128, // RGBA32UI + 64, // DXT1 + 128, // DXT23 + 128, // DXT45 + 64, // DXN1 + 128, // DXN2UNORM + 128, // DXN2SNORM + 128, // BC7U + 128, // BC6H_UF16 + 128, // BC6H_SF16 + 128, // ASTC_2D_4X4 + 16, // G8R8U + 16, // G8R8S + 32, // BGRA8 + 128, // RGBA32F + 64, // RG32F + 32, // R32F + 16, // R16F + 16, // R16U + 16, // R16S + 16, // R16UI + 16, // R16I + 32, // RG16 + 32, // RG16F + 32, // RG16UI + 32, // RG16I + 32, // RG16S + 96, // RGB32F + 32, // RGBA8_SRGB + 16, // RG8U + 16, // RG8S + 64, // RG32UI + 32, // R32UI + 128, // ASTC_2D_8X8 + 128, // ASTC_2D_8X5 + 128, // ASTC_2D_5X4 + 32, // BGRA8_SRGB + 64, // DXT1_SRGB + 128, // DXT23_SRGB + 128, // DXT45_SRGB + 128, // BC7U + 128, // ASTC_2D_4X4_SRGB + 128, // ASTC_2D_8X8_SRGB + 128, // ASTC_2D_8X5_SRGB + 128, // ASTC_2D_5X4_SRGB + 128, // ASTC_2D_5X5 + 128, // ASTC_2D_5X5_SRGB + 128, // ASTC_2D_10X8 + 128, // ASTC_2D_10X8_SRGB + 32, // Z32F + 16, // Z16 + 32, // Z24S8 + 32, // S8Z24 + 64, // Z32FS8 +}}; + static constexpr u32 GetFormatBpp(PixelFormat format) { if (format == PixelFormat::Invalid) return 0; - constexpr std::array<u32, MaxPixelFormat> bpp_table = {{ - 32, // ABGR8U - 32, // ABGR8S - 32, // ABGR8UI - 16, // B5G6R5U - 32, // A2B10G10R10U - 16, // A1B5G5R5U - 8, // R8U - 8, // R8UI - 64, // RGBA16F - 64, // RGBA16U - 64, // RGBA16UI - 32, // R11FG11FB10F - 128, // RGBA32UI - 64, // DXT1 - 128, // DXT23 - 128, // DXT45 - 64, // DXN1 - 128, // DXN2UNORM - 128, // DXN2SNORM - 128, // BC7U - 128, // BC6H_UF16 - 128, // BC6H_SF16 - 128, // ASTC_2D_4X4 - 16, // G8R8U - 16, // G8R8S - 32, // BGRA8 - 128, // RGBA32F - 64, // RG32F - 32, // R32F - 16, // R16F - 16, // R16U - 16, // R16S - 16, // R16UI - 16, // R16I - 32, // RG16 - 32, // RG16F - 32, // RG16UI - 32, // RG16I - 32, // RG16S - 96, // RGB32F - 32, // RGBA8_SRGB - 16, // RG8U - 16, // RG8S - 64, // RG32UI - 32, // R32UI - 128, // ASTC_2D_8X8 - 128, // ASTC_2D_8X5 - 128, // ASTC_2D_5X4 - 32, // BGRA8_SRGB - 64, // DXT1_SRGB - 128, // DXT23_SRGB - 128, // DXT45_SRGB - 128, // BC7U - 128, // ASTC_2D_4X4_SRGB - 128, // ASTC_2D_8X8_SRGB - 128, // ASTC_2D_8X5_SRGB - 128, // ASTC_2D_5X4_SRGB - 128, // ASTC_2D_5X5 - 128, // ASTC_2D_5X5_SRGB - 128, // ASTC_2D_10X8 - 128, // ASTC_2D_10X8_SRGB - 32, // Z32F - 16, // Z16 - 32, // Z24S8 - 32, // S8Z24 - 64, // Z32FS8 - }}; - ASSERT(static_cast<std::size_t>(format) < bpp_table.size()); return bpp_table[static_cast<std::size_t>(format)]; } diff --git a/src/yuzu/configuration/config.cpp b/src/yuzu/configuration/config.cpp index 97a1633b0..eb2077b0f 100644 --- a/src/yuzu/configuration/config.cpp +++ b/src/yuzu/configuration/config.cpp @@ -206,60 +206,57 @@ const std::array<int, Settings::NativeKeyboard::NumKeyboardMods> Config::default void Config::ReadPlayerValues() { for (std::size_t p = 0; p < Settings::values.players.size(); ++p) { - Settings::values.players[p].connected = - qt_config->value(QString("player_%1_connected").arg(p), false).toBool(); + auto& player = Settings::values.players[p]; - Settings::values.players[p].type = static_cast<Settings::ControllerType>( + player.connected = qt_config->value(QString("player_%1_connected").arg(p), false).toBool(); + + player.type = static_cast<Settings::ControllerType>( qt_config ->value(QString("player_%1_type").arg(p), static_cast<u8>(Settings::ControllerType::DualJoycon)) .toUInt()); - Settings::values.players[p].body_color_left = - qt_config - ->value(QString("player_%1_body_color_left").arg(p), - Settings::JOYCON_BODY_NEON_BLUE) - .toUInt(); - Settings::values.players[p].body_color_right = - qt_config - ->value(QString("player_%1_body_color_right").arg(p), - Settings::JOYCON_BODY_NEON_RED) - .toUInt(); - Settings::values.players[p].button_color_left = - qt_config - ->value(QString("player_%1_button_color_left").arg(p), - Settings::JOYCON_BUTTONS_NEON_BLUE) - .toUInt(); - Settings::values.players[p].button_color_right = - qt_config - ->value(QString("player_%1_button_color_right").arg(p), - Settings::JOYCON_BUTTONS_NEON_RED) - .toUInt(); + player.body_color_left = qt_config + ->value(QString("player_%1_body_color_left").arg(p), + Settings::JOYCON_BODY_NEON_BLUE) + .toUInt(); + player.body_color_right = qt_config + ->value(QString("player_%1_body_color_right").arg(p), + Settings::JOYCON_BODY_NEON_RED) + .toUInt(); + player.button_color_left = qt_config + ->value(QString("player_%1_button_color_left").arg(p), + Settings::JOYCON_BUTTONS_NEON_BLUE) + .toUInt(); + player.button_color_right = qt_config + ->value(QString("player_%1_button_color_right").arg(p), + Settings::JOYCON_BUTTONS_NEON_RED) + .toUInt(); for (int i = 0; i < Settings::NativeButton::NumButtons; ++i) { std::string default_param = InputCommon::GenerateKeyboardParam(default_buttons[i]); - Settings::values.players[p].buttons[i] = + player.buttons[i] = qt_config ->value(QString("player_%1_").arg(p) + Settings::NativeButton::mapping[i], QString::fromStdString(default_param)) .toString() .toStdString(); - if (Settings::values.players[p].buttons[i].empty()) - Settings::values.players[p].buttons[i] = default_param; + if (player.buttons[i].empty()) + player.buttons[i] = default_param; } for (int i = 0; i < Settings::NativeAnalog::NumAnalogs; ++i) { std::string default_param = InputCommon::GenerateAnalogParamFromKeys( default_analogs[i][0], default_analogs[i][1], default_analogs[i][2], default_analogs[i][3], default_analogs[i][4], 0.5f); - Settings::values.players[p].analogs[i] = + player.analogs[i] = qt_config ->value(QString("player_%1_").arg(p) + Settings::NativeAnalog::mapping[i], QString::fromStdString(default_param)) .toString() .toStdString(); - if (Settings::values.players[p].analogs[i].empty()) - Settings::values.players[p].analogs[i] = default_param; + if (player.analogs[i].empty()) + player.analogs[i] = default_param; } } @@ -526,30 +523,28 @@ void Config::ReadValues() { } void Config::SavePlayerValues() { - for (int p = 0; p < Settings::values.players.size(); ++p) { - qt_config->setValue(QString("player_%1_connected").arg(p), - Settings::values.players[p].connected); - qt_config->setValue(QString("player_%1_type").arg(p), - static_cast<u8>(Settings::values.players[p].type)); - - qt_config->setValue(QString("player_%1_body_color_left").arg(p), - Settings::values.players[p].body_color_left); - qt_config->setValue(QString("player_%1_body_color_right").arg(p), - Settings::values.players[p].body_color_right); + for (std::size_t p = 0; p < Settings::values.players.size(); ++p) { + const auto& player = Settings::values.players[p]; + + qt_config->setValue(QString("player_%1_connected").arg(p), player.connected); + qt_config->setValue(QString("player_%1_type").arg(p), static_cast<u8>(player.type)); + + qt_config->setValue(QString("player_%1_body_color_left").arg(p), player.body_color_left); + qt_config->setValue(QString("player_%1_body_color_right").arg(p), player.body_color_right); qt_config->setValue(QString("player_%1_button_color_left").arg(p), - Settings::values.players[p].button_color_left); + player.button_color_left); qt_config->setValue(QString("player_%1_button_color_right").arg(p), - Settings::values.players[p].button_color_right); + player.button_color_right); for (int i = 0; i < Settings::NativeButton::NumButtons; ++i) { qt_config->setValue(QString("player_%1_").arg(p) + QString::fromStdString(Settings::NativeButton::mapping[i]), - QString::fromStdString(Settings::values.players[p].buttons[i])); + QString::fromStdString(player.buttons[i])); } for (int i = 0; i < Settings::NativeAnalog::NumAnalogs; ++i) { qt_config->setValue(QString("player_%1_").arg(p) + QString::fromStdString(Settings::NativeAnalog::mapping[i]), - QString::fromStdString(Settings::values.players[p].analogs[i])); + QString::fromStdString(player.analogs[i])); } } } diff --git a/src/yuzu/debugger/wait_tree.cpp b/src/yuzu/debugger/wait_tree.cpp index 96c57fe97..f9c18ede4 100644 --- a/src/yuzu/debugger/wait_tree.cpp +++ b/src/yuzu/debugger/wait_tree.cpp @@ -153,7 +153,7 @@ QString WaitTreeWaitObject::GetText() const { std::unique_ptr<WaitTreeWaitObject> WaitTreeWaitObject::make(const Kernel::WaitObject& object) { switch (object.GetHandleType()) { - case Kernel::HandleType::Event: + case Kernel::HandleType::ReadableEvent: return std::make_unique<WaitTreeEvent>(static_cast<const Kernel::ReadableEvent&>(object)); case Kernel::HandleType::Timer: return std::make_unique<WaitTreeTimer>(static_cast<const Kernel::Timer&>(object)); diff --git a/src/yuzu/game_list_worker.cpp b/src/yuzu/game_list_worker.cpp index 553269269..b37710f59 100644 --- a/src/yuzu/game_list_worker.cpp +++ b/src/yuzu/game_list_worker.cpp @@ -86,6 +86,37 @@ QString FormatPatchNameVersions(const FileSys::PatchManager& patch_manager, out.chop(1); return out; } + +QList<QStandardItem*> MakeGameListEntry(const std::string& path, const std::string& name, + const std::vector<u8>& icon, Loader::AppLoader& loader, + u64 program_id, const CompatibilityList& compatibility_list, + const FileSys::PatchManager& patch) { + const auto it = FindMatchingCompatibilityEntry(compatibility_list, program_id); + + // The game list uses this as compatibility number for untested games + QString compatibility{"99"}; + if (it != compatibility_list.end()) { + compatibility = it->second.first; + } + + const auto file_type = loader.GetFileType(); + const auto file_type_string = QString::fromStdString(Loader::GetFileTypeString(file_type)); + + QList<QStandardItem*> list{ + new GameListItemPath(FormatGameName(path), icon, QString::fromStdString(name), + file_type_string, program_id), + new GameListItemCompat(compatibility), + new GameListItem(file_type_string), + new GameListItemSize(FileUtil::GetSize(path)), + }; + + if (UISettings::values.show_add_ons) { + list.insert( + 2, new GameListItem(FormatPatchNameVersions(patch, loader, loader.IsRomFSUpdatable()))); + } + + return list; +} } // Anonymous namespace GameListWorker::GameListWorker(FileSys::VirtualFilesystem vfs, QString dir_path, bool deep_scan, @@ -116,29 +147,8 @@ void GameListWorker::AddInstalledTitlesToGameList() { if (control != nullptr) GetMetadataFromControlNCA(patch, *control, icon, name); - auto it = FindMatchingCompatibilityEntry(compatibility_list, program_id); - - // The game list uses this as compatibility number for untested games - QString compatibility("99"); - if (it != compatibility_list.end()) - compatibility = it->second.first; - - QList<QStandardItem*> list{ - new GameListItemPath( - FormatGameName(file->GetFullPath()), icon, QString::fromStdString(name), - QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType())), - program_id), - new GameListItemCompat(compatibility), - new GameListItem( - QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))), - new GameListItemSize(file->GetSize()), - }; - - if (UISettings::values.show_add_ons) { - list.insert(2, new GameListItem(FormatPatchNameVersions(patch, *loader))); - } - - emit EntryReady(list); + emit EntryReady(MakeGameListEntry(file->GetFullPath(), name, icon, *loader, program_id, + compatibility_list, patch)); } const auto control_data = cache.ListEntriesFilter(FileSys::TitleType::Application, @@ -155,14 +165,14 @@ void GameListWorker::AddInstalledTitlesToGameList() { void GameListWorker::FillControlMap(const std::string& dir_path) { const auto nca_control_callback = [this](u64* num_entries_out, const std::string& directory, const std::string& virtual_name) -> bool { - std::string physical_name = directory + DIR_SEP + virtual_name; - - if (stop_processing) - return false; // Breaks the callback loop. + if (stop_processing) { + // Breaks the callback loop + return false; + } - bool is_dir = FileUtil::IsDirectory(physical_name); - QFileInfo file_info(physical_name.c_str()); - if (!is_dir && file_info.suffix().toStdString() == "nca") { + const std::string physical_name = directory + DIR_SEP + virtual_name; + const QFileInfo file_info(QString::fromStdString(physical_name)); + if (!file_info.isDir() && file_info.suffix() == QStringLiteral("nca")) { auto nca = std::make_unique<FileSys::NCA>(vfs->OpenFile(physical_name, FileSys::Mode::Read)); if (nca->GetType() == FileSys::NCAContentType::Control) { @@ -179,20 +189,25 @@ void GameListWorker::FillControlMap(const std::string& dir_path) { void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsigned int recursion) { const auto callback = [this, recursion](u64* num_entries_out, const std::string& directory, const std::string& virtual_name) -> bool { - std::string physical_name = directory + DIR_SEP + virtual_name; - - if (stop_processing) - return false; // Breaks the callback loop. + if (stop_processing) { + // Breaks the callback loop. + return false; + } - bool is_dir = FileUtil::IsDirectory(physical_name); + const std::string physical_name = directory + DIR_SEP + virtual_name; + const bool is_dir = FileUtil::IsDirectory(physical_name); if (!is_dir && (HasSupportedFileExtension(physical_name) || IsExtractedNCAMain(physical_name))) { - std::unique_ptr<Loader::AppLoader> loader = - Loader::GetLoader(vfs->OpenFile(physical_name, FileSys::Mode::Read)); - if (!loader || ((loader->GetFileType() == Loader::FileType::Unknown || - loader->GetFileType() == Loader::FileType::Error) && - !UISettings::values.show_unknown)) + auto loader = Loader::GetLoader(vfs->OpenFile(physical_name, FileSys::Mode::Read)); + if (!loader) { return true; + } + + const auto file_type = loader->GetFileType(); + if ((file_type == Loader::FileType::Unknown || file_type == Loader::FileType::Error) && + !UISettings::values.show_unknown) { + return true; + } std::vector<u8> icon; const auto res1 = loader->ReadIcon(icon); @@ -214,30 +229,8 @@ void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsign } } - auto it = FindMatchingCompatibilityEntry(compatibility_list, program_id); - - // The game list uses this as compatibility number for untested games - QString compatibility("99"); - if (it != compatibility_list.end()) - compatibility = it->second.first; - - QList<QStandardItem*> list{ - new GameListItemPath( - FormatGameName(physical_name), icon, QString::fromStdString(name), - QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType())), - program_id), - new GameListItemCompat(compatibility), - new GameListItem( - QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))), - new GameListItemSize(FileUtil::GetSize(physical_name)), - }; - - if (UISettings::values.show_add_ons) { - list.insert(2, new GameListItem(FormatPatchNameVersions( - patch, *loader, loader->IsRomFSUpdatable()))); - } - - emit EntryReady(std::move(list)); + emit EntryReady(MakeGameListEntry(physical_name, name, icon, *loader, program_id, + compatibility_list, patch)); } else if (is_dir && recursion > 0) { watch_list.append(QString::fromStdString(physical_name)); AddFstEntriesToGameList(physical_name, recursion - 1); |