diff options
Diffstat (limited to 'src')
38 files changed, 595 insertions, 290 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/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/system_archive/ng_word.cpp b/src/core/file_sys/system_archive/ng_word.cpp index a24f1e496..f4443784d 100644 --- a/src/core/file_sys/system_archive/ng_word.cpp +++ b/src/core/file_sys/system_archive/ng_word.cpp @@ -39,4 +39,43 @@ VirtualDir NgWord1() { 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 index f4bc67344..cd81e0abb 100644 --- a/src/core/file_sys/system_archive/ng_word.h +++ b/src/core/file_sys/system_archive/ng_word.h @@ -9,5 +9,6 @@ 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 index d3883267c..e3e79f40a 100644 --- a/src/core/file_sys/system_archive/system_archive.cpp +++ b/src/core/file_sys/system_archive/system_archive.cpp @@ -56,7 +56,7 @@ constexpr std::array<SystemArchiveDescriptor, SYSTEM_ARCHIVE_COUNT> SYSTEM_ARCHI {0x0100000000000820, "PlatformConfigCopper", nullptr}, {0x0100000000000821, "PlatformConfigHoag", nullptr}, {0x0100000000000822, "ControllerFirmware", nullptr}, - {0x0100000000000823, "NgWord2", nullptr}, + {0x0100000000000823, "NgWord2", &NgWord2}, {0x0100000000000824, "PlatformConfigIcosaMariko", nullptr}, {0x0100000000000825, "ApplicationBlackList", nullptr}, {0x0100000000000826, "RebootlessSystemUpdateVersion", nullptr}, diff --git a/src/core/hle/kernel/object.cpp b/src/core/hle/kernel/object.cpp index bb1b68778..0ea851a74 100644 --- a/src/core/hle/kernel/object.cpp +++ b/src/core/hle/kernel/object.cpp @@ -15,6 +15,7 @@ bool Object::IsWaitable() const { switch (GetHandleType()) { case HandleType::ReadableEvent: case HandleType::Thread: + case HandleType::Process: case HandleType::Timer: case HandleType::ServerPort: case HandleType::ServerSession: @@ -23,7 +24,6 @@ bool Object::IsWaitable() const { 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/process.cpp b/src/core/hle/kernel/process.cpp index 4ecb8c926..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" @@ -48,8 +49,24 @@ 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()); } @@ -137,13 +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) { @@ -167,6 +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); } /** @@ -265,7 +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 49345aa66..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; @@ -212,6 +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. @@ -260,6 +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; @@ -305,6 +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 867ff3051..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; @@ -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 e6c77f9db..f43c7201c 100644 --- a/src/core/hle/kernel/svc.cpp +++ b/src/core/hle/kernel/svc.cpp @@ -239,7 +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; } @@ -1077,19 +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; } @@ -1433,17 +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 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/service/ldr/ldr.cpp b/src/core/hle/service/ldr/ldr.cpp index 453d90a22..13bcefe07 100644 --- a/src/core/hle/service/ldr/ldr.cpp +++ b/src/core/hle/service/ldr/ldr.cpp @@ -408,13 +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; @@ -430,9 +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.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 7686634bf..5390ab9ee 100644 --- a/src/core/loader/loader.h +++ b/src/core/loader/loader.h @@ -131,7 +131,7 @@ 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.h b/src/core/loader/nro.h index 6deff3a51..013d629c0 100644 --- a/src/core/loader/nro.h +++ b/src/core/loader/nro.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/nso.h b/src/core/loader/nso.h index 0c1defbb6..135b6ea5a 100644 --- a/src/core/loader/nso.h +++ b/src/core/loader/nso.h @@ -37,7 +37,7 @@ public: */ static FileType IdentifyType(const FileSys::VirtualFile& file); - FileType GetFileType() override { + FileType GetFileType() const override { return IdentifyType(file); } diff --git a/src/core/loader/nsp.h b/src/core/loader/nsp.h index db91cd01e..9ed8a59cf 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 46f8dfc9e..ded5bb88a 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 41fd2a6a0..643afdee8 100644 --- a/src/core/memory.cpp +++ b/src/core/memory.cpp @@ -125,14 +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; - - auto& vm_manager = process.VMManager(); + const auto& vm_manager = process.VMManager(); - auto it = vm_manager.FindVMA(vaddr); - ASSERT(it != vm_manager.vma_map.end()); + const auto it = vm_manager.FindVMA(vaddr); + DEBUG_ASSERT(vm_manager.IsValidHandle(it)); - 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/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/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 8d68156bf..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: @@ -1649,7 +1681,7 @@ private: 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 (!last) { + if (i < coord_count - 1) { coord += ','; } } @@ -1670,6 +1702,99 @@ private: 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. @@ -2766,33 +2891,33 @@ private: 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 auto [coord, texture] = + auto [coord, texture] = GetTEXSCode(instr, texture_type, process_mode, depth_compare, is_array); shader.AddLine(coord); - if (!depth_compare) { - shader.AddLine("vec4 texture_tmp = " + texture + ';'); + if (depth_compare) { + texture = "vec4(" + texture + ')'; + } + shader.AddLine("vec4 texture_tmp = " + texture + ';'); + if (instr.texs.fp32_flag) { + WriteTexsInstructionFloat(instr, "texture_tmp"); } else { - shader.AddLine("vec4 texture_tmp = vec4(" + texture + ");"); + WriteTexsInstructionHalfFloat(instr, "texture_tmp"); } - - WriteTexsInstruction(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), @@ -2800,54 +2925,16 @@ private: UNIMPLEMENTED_IF_MSG(instr.tlds.UsesMiscMode(Tegra::Shader::TextureMiscMode::MZ), "MZ is not implemented"); - u32 extra_op_offset = 0; - - 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 auto [coord, texture] = GetTLDSCode(instr, texture_type, is_array); - 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"); @@ -2857,56 +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 bool is_array = instr.tld4.array != 0; - const auto scope = shader.Scope(); - - 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; } @@ -2920,27 +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 + ')'; - } - WriteTexsInstruction(instr, 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 + ')'; + } + shader.AddLine(coords); + shader.AddLine("vec4 texture_tmp = " + texture + ';'); + WriteTexsInstructionFloat(instr, "texture_tmp"); break; } case OpCode::Id::TXQ: { 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/yuzu/configuration/config.cpp b/src/yuzu/configuration/config.cpp index 83ebbd1fe..c26161169 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; } } @@ -511,30 +508,28 @@ 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/game_list_worker.cpp b/src/yuzu/game_list_worker.cpp index 9fd074223..20f5e8798 100644 --- a/src/yuzu/game_list_worker.cpp +++ b/src/yuzu/game_list_worker.cpp @@ -99,12 +99,14 @@ QList<QStandardItem*> MakeGameListEntry(const std::string& path, const std::stri 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), - QString::fromStdString(Loader::GetFileTypeString(loader.GetFileType())), program_id), + new GameListItemPath(FormatGameName(path), icon, QString::fromStdString(name), + file_type_string, program_id), new GameListItemCompat(compatibility), - new GameListItem(QString::fromStdString(Loader::GetFileTypeString(loader.GetFileType()))), + new GameListItem(file_type_string), new GameListItemSize(FileUtil::GetSize(path)), }; @@ -196,12 +198,16 @@ void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsign 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); |