diff options
Diffstat (limited to 'src')
188 files changed, 9504 insertions, 1477 deletions
diff --git a/src/common/CMakeLists.txt b/src/common/CMakeLists.txt index 4480c25f9..2ba1da195 100644 --- a/src/common/CMakeLists.txt +++ b/src/common/CMakeLists.txt @@ -58,7 +58,6 @@ add_library(common STATIC misc.cpp param_package.cpp param_package.h - platform.h quaternion.h scm_rev.cpp scm_rev.h diff --git a/src/common/file_util.h b/src/common/file_util.h index 630232a25..143f099eb 100644 --- a/src/common/file_util.h +++ b/src/common/file_util.h @@ -121,7 +121,7 @@ void CopyDir(const std::string& source_path, const std::string& dest_path); // Set the current directory to given directory bool SetCurrentDir(const std::string& directory); -// Returns a pointer to a string with a Citra data dir in the user's home +// Returns a pointer to a string with a yuzu data dir in the user's home // directory. To be used in "multi-user" mode (that is, installed). const std::string& GetUserPath(const unsigned int DirIDX, const std::string& newPath = ""); diff --git a/src/common/logging/backend.cpp b/src/common/logging/backend.cpp index 22afa1d66..3db654913 100644 --- a/src/common/logging/backend.cpp +++ b/src/common/logging/backend.cpp @@ -38,16 +38,20 @@ namespace Log { SUB(Service, AM) \ SUB(Service, AOC) \ SUB(Service, APM) \ + SUB(Service, Fatal) \ SUB(Service, Friend) \ SUB(Service, FS) \ SUB(Service, HID) \ SUB(Service, LM) \ + SUB(Service, NFP) \ SUB(Service, NIFM) \ SUB(Service, NS) \ SUB(Service, NVDRV) \ SUB(Service, PCTL) \ SUB(Service, SET) \ SUB(Service, SM) \ + SUB(Service, SPL) \ + SUB(Service, SSL) \ SUB(Service, Time) \ SUB(Service, VI) \ CLS(HW) \ diff --git a/src/common/logging/log.h b/src/common/logging/log.h index 7f6d2ade8..8432628ae 100644 --- a/src/common/logging/log.h +++ b/src/common/logging/log.h @@ -55,16 +55,20 @@ enum class Class : ClassType { Service_AOC, ///< The AOC (AddOn Content) service Service_APM, ///< The APM (Performance) service Service_Audio, ///< The Audio (Audio control) service + Service_Fatal, ///< The Fatal service Service_Friend, ///< The friend service Service_FS, ///< The FS (Filesystem) service Service_HID, ///< The HID (Human interface device) service Service_LM, ///< The LM (Logger) service + Service_NFP, ///< The NFP service Service_NIFM, ///< The NIFM (Network interface) service Service_NS, ///< The NS services Service_NVDRV, ///< The NVDRV (Nvidia driver) service Service_PCTL, ///< The PCTL (Parental control) service Service_SET, ///< The SET (Settings) service Service_SM, ///< The SM (Service manager) service + Service_SPL, ///< The SPL service + Service_SSL, ///< The SSL service Service_Time, ///< The time service Service_VI, ///< The VI (Video interface) service HW, ///< Low-level hardware emulation @@ -83,7 +87,7 @@ enum class Class : ClassType { Loader, ///< ROM loader Input, ///< Input emulation Network, ///< Network emulation - WebService, ///< Interface to Citra Web Services + WebService, ///< Interface to yuzu Web Services Count ///< Total number of logging classes }; diff --git a/src/common/platform.h b/src/common/platform.h deleted file mode 100644 index c62fb7c8f..000000000 --- a/src/common/platform.h +++ /dev/null @@ -1,34 +0,0 @@ -/** - * Copyright (C) 2005-2012 Gekko Emulator - * - * @file platform.h - * @author ShizZy <shizzy247@gmail.com> - * @date 2012-02-11 - * @brief Platform detection macros for portable compilation - * - * @section LICENSE - * This program is free software; you can redistribute it and/or - * modify it under the terms of the GNU General Public License as - * published by the Free Software Foundation; either version 2 of - * the License, or (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, but - * WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - * General Public License for more details at - * http://www.gnu.org/copyleft/gpl.html - * - * Official project repository can be found at: - * http://code.google.com/p/gekko-gc-emu/ - */ - -#pragma once - -//////////////////////////////////////////////////////////////////////////////////////////////////// -// Platform detection - -#if defined(ARCHITECTURE_x86_64) || defined(__aarch64__) -#define EMU_ARCH_BITS 64 -#elif defined(__i386) || defined(_M_IX86) || defined(__arm__) || defined(_M_ARM) -#define EMU_ARCH_BITS 32 -#endif diff --git a/src/common/swap.h b/src/common/swap.h index d94cbe6b2..4a4012d1a 100644 --- a/src/common/swap.h +++ b/src/common/swap.h @@ -103,7 +103,19 @@ inline __attribute__((always_inline)) u64 swap64(u64 _data) { return __builtin_bswap64(_data); } #elif defined(__Bitrig__) || defined(__OpenBSD__) -// swap16, swap32, swap64 are left as is +// redefine swap16, swap32, swap64 as inline functions +#undef swap16 +#undef swap32 +#undef swap64 +inline u16 swap16(u16 _data) { + return __swap16(_data); +} +inline u32 swap32(u32 _data) { + return __swap32(_data); +} +inline u64 swap64(u64 _data) { + return __swap64(_data); +} #elif defined(__DragonFly__) || defined(__FreeBSD__) || defined(__NetBSD__) inline u16 swap16(u16 _data) { return bswap16(_data); diff --git a/src/common/telemetry.h b/src/common/telemetry.h index 3694c76f2..7a09df0a7 100644 --- a/src/common/telemetry.h +++ b/src/common/telemetry.h @@ -15,7 +15,7 @@ namespace Telemetry { /// Field type, used for grouping fields together in the final submitted telemetry log enum class FieldType : u8 { None = 0, ///< No specified field group - App, ///< Citra application fields (e.g. version, branch, etc.) + App, ///< yuzu application fields (e.g. version, branch, etc.) Session, ///< Emulated session fields (e.g. title ID, log, etc.) Performance, ///< Emulated performance (e.g. fps, emulated CPU speed, etc.) UserFeedback, ///< User submitted feedback (e.g. star rating, user notes, etc.) diff --git a/src/common/x64/cpu_detect.cpp b/src/common/x64/cpu_detect.cpp index 62f17fbb5..2dfcd39c8 100644 --- a/src/common/x64/cpu_detect.cpp +++ b/src/common/x64/cpu_detect.cpp @@ -54,7 +54,7 @@ static CPUCaps Detect() { caps.num_cores = std::thread::hardware_concurrency(); // Assumes the CPU supports the CPUID instruction. Those that don't would likely not support - // Citra at all anyway + // yuzu at all anyway int cpu_id[4]; memset(caps.brand_string, 0, sizeof(caps.brand_string)); diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt index 1bc536075..6f8104516 100644 --- a/src/core/CMakeLists.txt +++ b/src/core/CMakeLists.txt @@ -7,6 +7,8 @@ add_library(core STATIC core_timing.cpp core_timing.h file_sys/directory.h + file_sys/disk_filesystem.cpp + file_sys/disk_filesystem.h file_sys/errors.h file_sys/filesystem.cpp file_sys/filesystem.h @@ -18,6 +20,10 @@ add_library(core STATIC file_sys/romfs_factory.h file_sys/romfs_filesystem.cpp file_sys/romfs_filesystem.h + file_sys/savedata_factory.cpp + file_sys/savedata_factory.h + file_sys/sdmc_factory.cpp + file_sys/sdmc_factory.h file_sys/storage.h frontend/emu_window.cpp frontend/emu_window.h @@ -110,6 +116,12 @@ add_library(core STATIC hle/service/audio/audren_u.h hle/service/audio/codecctl.cpp hle/service/audio/codecctl.h + hle/service/fatal/fatal.cpp + hle/service/fatal/fatal.h + hle/service/fatal/fatal_p.cpp + hle/service/fatal/fatal_p.h + hle/service/fatal/fatal_u.cpp + hle/service/fatal/fatal_u.h hle/service/filesystem/filesystem.cpp hle/service/filesystem/filesystem.h hle/service/filesystem/fsp_srv.cpp @@ -130,6 +142,10 @@ add_library(core STATIC hle/service/nifm/nifm_s.h hle/service/nifm/nifm_u.cpp hle/service/nifm/nifm_u.h + hle/service/nfp/nfp.cpp + hle/service/nfp/nfp.h + hle/service/nfp/nfp_user.cpp + hle/service/nfp/nfp_user.h hle/service/ns/ns.cpp hle/service/ns/ns.h hle/service/ns/pl_u.cpp @@ -165,16 +181,34 @@ add_library(core STATIC hle/service/service.h hle/service/set/set.cpp hle/service/set/set.h + hle/service/set/set_cal.cpp + hle/service/set/set_cal.h + hle/service/set/set_fd.cpp + hle/service/set/set_fd.h + hle/service/set/set_sys.cpp + hle/service/set/set_sys.h + hle/service/set/settings.cpp + hle/service/set/settings.h hle/service/sm/controller.cpp hle/service/sm/controller.h hle/service/sm/sm.cpp hle/service/sm/sm.h - hle/service/sockets/bsd_u.cpp - hle/service/sockets/bsd_u.h + hle/service/sockets/bsd.cpp + hle/service/sockets/bsd.h + hle/service/sockets/nsd.cpp + hle/service/sockets/nsd.h hle/service/sockets/sfdnsres.cpp hle/service/sockets/sfdnsres.h hle/service/sockets/sockets.cpp hle/service/sockets/sockets.h + hle/service/spl/csrng.cpp + hle/service/spl/csrng.h + hle/service/spl/module.cpp + hle/service/spl/module.h + hle/service/spl/spl.cpp + hle/service/spl/spl.h + hle/service/ssl/ssl.cpp + hle/service/ssl/ssl.h hle/service/time/time.cpp hle/service/time/time.h hle/service/time/time_s.cpp diff --git a/src/core/arm/arm_interface.h b/src/core/arm/arm_interface.h index 5ae60214e..32ff3c345 100644 --- a/src/core/arm/arm_interface.h +++ b/src/core/arm/arm_interface.h @@ -25,22 +25,18 @@ public: VAddr tls_address; }; - /** - * Runs the CPU for the given number of instructions - * @param num_instructions Number of instructions to run - */ - void Run(int num_instructions) { - ExecuteInstructions(num_instructions); - this->num_instructions += num_instructions; - } + /// Runs the CPU until an event happens + virtual void Run() = 0; /// Step CPU by one instruction - void Step() { - Run(1); - } + virtual void Step() = 0; + /// Maps a backing memory region for the CPU virtual void MapBackingMemory(VAddr address, size_t size, u8* memory, - Kernel::VMAPermission perms) {} + Kernel::VMAPermission perms) = 0; + + /// Unmaps a region of memory that was previously mapped using MapBackingMemory + virtual void UnmapMemory(VAddr address, size_t size) = 0; /// Clear all instruction cache virtual void ClearInstructionCache() = 0; @@ -122,19 +118,4 @@ public: /// Prepare core for thread reschedule (if needed to correctly handle state) virtual void PrepareReschedule() = 0; - - /// Getter for num_instructions - u64 GetNumInstructions() const { - return num_instructions; - } - -protected: - /** - * Executes the given number of instructions - * @param num_instructions Number of instructions to executes - */ - virtual void ExecuteInstructions(int num_instructions) = 0; - -private: - u64 num_instructions = 0; ///< Number of instructions executed }; diff --git a/src/core/arm/dynarmic/arm_dynarmic.cpp b/src/core/arm/dynarmic/arm_dynarmic.cpp index e7f6bf8c2..6afad0e0c 100644 --- a/src/core/arm/dynarmic/arm_dynarmic.cpp +++ b/src/core/arm/dynarmic/arm_dynarmic.cpp @@ -8,6 +8,7 @@ #include <dynarmic/A64/config.h> #include "common/logging/log.h" #include "core/arm/dynarmic/arm_dynarmic.h" +#include "core/core.h" #include "core/core_timing.h" #include "core/hle/kernel/memory.h" #include "core/hle/kernel/svc.h" @@ -85,28 +86,24 @@ public: } void AddTicks(u64 ticks) override { - if (ticks > ticks_remaining) { - ticks_remaining = 0; - return; - } - ticks -= ticks_remaining; + CoreTiming::AddTicks(ticks - num_interpreted_instructions); + num_interpreted_instructions = 0; } u64 GetTicksRemaining() override { - return ticks_remaining; + return std::max(CoreTiming::GetDowncount(), 0); } u64 GetCNTPCT() override { return CoreTiming::GetTicks(); } ARM_Dynarmic& parent; - size_t ticks_remaining = 0; size_t num_interpreted_instructions = 0; u64 tpidrro_el0 = 0; u64 tpidr_el0 = 0; }; std::unique_ptr<Dynarmic::A64::Jit> MakeJit(const std::unique_ptr<ARM_Dynarmic_Callbacks>& cb) { - const auto page_table = Kernel::g_current_process->vm_manager.page_table.pointers.data(); + const auto page_table = Core::CurrentProcess()->vm_manager.page_table.pointers.data(); Dynarmic::A64::UserConfig config; config.callbacks = cb.get(); @@ -121,11 +118,22 @@ std::unique_ptr<Dynarmic::A64::Jit> MakeJit(const std::unique_ptr<ARM_Dynarmic_C return std::make_unique<Dynarmic::A64::Jit>(config); } +void ARM_Dynarmic::Run() { + ASSERT(Memory::GetCurrentPageTable() == current_page_table); + + jit->Run(); +} + +void ARM_Dynarmic::Step() { + cb->InterpreterFallback(jit->GetPC(), 1); +} + ARM_Dynarmic::ARM_Dynarmic() : cb(std::make_unique<ARM_Dynarmic_Callbacks>(*this)), jit(MakeJit(cb)) { ARM_Interface::ThreadContext ctx; inner_unicorn.SaveContext(ctx); LoadContext(ctx); + PageTableChanged(); } ARM_Dynarmic::~ARM_Dynarmic() = default; @@ -135,6 +143,10 @@ void ARM_Dynarmic::MapBackingMemory(u64 address, size_t size, u8* memory, inner_unicorn.MapBackingMemory(address, size, memory, perms); } +void ARM_Dynarmic::UnmapMemory(u64 address, size_t size) { + inner_unicorn.UnmapMemory(address, size); +} + void ARM_Dynarmic::SetPC(u64 pc) { jit->SetPC(pc); } @@ -184,13 +196,6 @@ void ARM_Dynarmic::SetTlsAddress(u64 address) { cb->tpidrro_el0 = address; } -void ARM_Dynarmic::ExecuteInstructions(int num_instructions) { - cb->ticks_remaining = num_instructions; - jit->Run(); - CoreTiming::AddTicks(num_instructions - cb->num_interpreted_instructions); - cb->num_interpreted_instructions = 0; -} - void ARM_Dynarmic::SaveContext(ARM_Interface::ThreadContext& ctx) { ctx.cpu_registers = jit->GetRegisters(); ctx.sp = jit->GetSP(); @@ -223,4 +228,5 @@ void ARM_Dynarmic::ClearInstructionCache() { void ARM_Dynarmic::PageTableChanged() { jit = MakeJit(cb); + current_page_table = Memory::GetCurrentPageTable(); } diff --git a/src/core/arm/dynarmic/arm_dynarmic.h b/src/core/arm/dynarmic/arm_dynarmic.h index 1d9dcf5ff..128669d01 100644 --- a/src/core/arm/dynarmic/arm_dynarmic.h +++ b/src/core/arm/dynarmic/arm_dynarmic.h @@ -19,7 +19,7 @@ public: void MapBackingMemory(VAddr address, size_t size, u8* memory, Kernel::VMAPermission perms) override; - + void UnmapMemory(u64 address, size_t size) override; void SetPC(u64 pc) override; u64 GetPC() const override; u64 GetReg(int index) const override; @@ -29,6 +29,8 @@ public: u32 GetVFPReg(int index) const override; void SetVFPReg(int index, u32 value) override; u32 GetCPSR() const override; + void Run() override; + void Step() override; void SetCPSR(u32 cpsr) override; VAddr GetTlsAddress() const override; void SetTlsAddress(VAddr address) override; @@ -37,7 +39,6 @@ public: void LoadContext(const ThreadContext& ctx) override; void PrepareReschedule() override; - void ExecuteInstructions(int num_instructions) override; void ClearInstructionCache() override; void PageTableChanged() override; @@ -47,4 +48,6 @@ private: std::unique_ptr<ARM_Dynarmic_Callbacks> cb; std::unique_ptr<Dynarmic::A64::Jit> jit; ARM_Unicorn inner_unicorn; + + Memory::PageTable* current_page_table = nullptr; }; diff --git a/src/core/arm/unicorn/arm_unicorn.cpp b/src/core/arm/unicorn/arm_unicorn.cpp index 5d2956bfd..b0cdc2403 100644 --- a/src/core/arm/unicorn/arm_unicorn.cpp +++ b/src/core/arm/unicorn/arm_unicorn.cpp @@ -2,6 +2,7 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <algorithm> #include <unicorn/arm64.h> #include "common/assert.h" #include "common/microprofile.h" @@ -52,7 +53,7 @@ static bool UnmappedMemoryHook(uc_engine* uc, uc_mem_type type, u64 addr, int si void* user_data) { ARM_Interface::ThreadContext ctx{}; Core::CPU().SaveContext(ctx); - ASSERT_MSG(false, "Attempted to read from unmapped memory: 0x%llx, pc=0x%llx, lr=0x%llx", addr, + ASSERT_MSG(false, "Attempted to read from unmapped memory: 0x%lx, pc=0x%lx, lr=0x%lx", addr, ctx.pc, ctx.cpu_registers[30]); return {}; } @@ -77,6 +78,10 @@ void ARM_Unicorn::MapBackingMemory(VAddr address, size_t size, u8* memory, CHECKED(uc_mem_map_ptr(uc, address, size, static_cast<u32>(perms), memory)); } +void ARM_Unicorn::UnmapMemory(VAddr address, size_t size) { + CHECKED(uc_mem_unmap(uc, address, size)); +} + void ARM_Unicorn::SetPC(u64 pc) { CHECKED(uc_reg_write(uc, UC_ARM64_REG_PC, &pc)); } @@ -149,6 +154,14 @@ void ARM_Unicorn::SetTlsAddress(VAddr base) { CHECKED(uc_reg_write(uc, UC_ARM64_REG_TPIDRRO_EL0, &base)); } +void ARM_Unicorn::Run() { + ExecuteInstructions(std::max(CoreTiming::GetDowncount(), 0)); +} + +void ARM_Unicorn::Step() { + ExecuteInstructions(1); +} + MICROPROFILE_DEFINE(ARM_Jit, "ARM JIT", "ARM JIT", MP_RGB(255, 64, 64)); void ARM_Unicorn::ExecuteInstructions(int num_instructions) { diff --git a/src/core/arm/unicorn/arm_unicorn.h b/src/core/arm/unicorn/arm_unicorn.h index c9a561dec..b99b58e4c 100644 --- a/src/core/arm/unicorn/arm_unicorn.h +++ b/src/core/arm/unicorn/arm_unicorn.h @@ -14,6 +14,7 @@ public: ~ARM_Unicorn(); void MapBackingMemory(VAddr address, size_t size, u8* memory, Kernel::VMAPermission perms) override; + void UnmapMemory(VAddr address, size_t size) override; void SetPC(u64 pc) override; u64 GetPC() const override; u64 GetReg(int index) const override; @@ -29,7 +30,9 @@ public: void SaveContext(ThreadContext& ctx) override; void LoadContext(const ThreadContext& ctx) override; void PrepareReschedule() override; - void ExecuteInstructions(int num_instructions) override; + void ExecuteInstructions(int num_instructions); + void Run() override; + void Step() override; void ClearInstructionCache() override; void PageTableChanged() override{}; diff --git a/src/core/core.cpp b/src/core/core.cpp index 4fb035556..11654d4da 100644 --- a/src/core/core.cpp +++ b/src/core/core.cpp @@ -26,7 +26,7 @@ namespace Core { /*static*/ System System::s_instance; -System::ResultStatus System::RunLoop(int tight_loop) { +System::ResultStatus System::RunLoop(bool tight_loop) { status = ResultStatus::Success; if (!cpu_core) { return ResultStatus::ErrorNotInitialized; @@ -40,7 +40,7 @@ System::ResultStatus System::RunLoop(int tight_loop) { if (GDBStub::GetCpuHaltFlag()) { if (GDBStub::GetCpuStepFlag()) { GDBStub::SetCpuStepFlag(false); - tight_loop = 1; + tight_loop = false; } else { return ResultStatus::Success; } @@ -56,7 +56,11 @@ System::ResultStatus System::RunLoop(int tight_loop) { PrepareReschedule(); } else { CoreTiming::Advance(); - cpu_core->Run(tight_loop); + if (tight_loop) { + cpu_core->Run(); + } else { + cpu_core->Step(); + } } HW::Update(); @@ -66,7 +70,7 @@ System::ResultStatus System::RunLoop(int tight_loop) { } System::ResultStatus System::SingleStep() { - return RunLoop(1); + return RunLoop(false); } System::ResultStatus System::Load(EmuWindow* emu_window, const std::string& filepath) { @@ -95,14 +99,15 @@ System::ResultStatus System::Load(EmuWindow* emu_window, const std::string& file ResultStatus init_result{Init(emu_window, system_mode.first.get())}; if (init_result != ResultStatus::Success) { - LOG_CRITICAL(Core, "Failed to initialize system (Error %i)!", init_result); + LOG_CRITICAL(Core, "Failed to initialize system (Error %i)!", + static_cast<int>(init_result)); System::Shutdown(); return init_result; } - const Loader::ResultStatus load_result{app_loader->Load(Kernel::g_current_process)}; + const Loader::ResultStatus load_result{app_loader->Load(current_process)}; if (Loader::ResultStatus::Success != load_result) { - LOG_CRITICAL(Core, "Failed to load ROM (Error %i)!", load_result); + LOG_CRITICAL(Core, "Failed to load ROM (Error %i)!", static_cast<int>(load_result)); System::Shutdown(); switch (load_result) { @@ -141,19 +146,17 @@ System::ResultStatus System::Init(EmuWindow* emu_window, u32 system_mode) { CoreTiming::Init(); - switch (Settings::values.cpu_core) { - case Settings::CpuCore::Unicorn: - cpu_core = std::make_shared<ARM_Unicorn>(); - break; - case Settings::CpuCore::Dynarmic: - default: + current_process = Kernel::Process::Create("main"); + + if (Settings::values.use_cpu_jit) { #ifdef ARCHITECTURE_x86_64 cpu_core = std::make_shared<ARM_Dynarmic>(); #else cpu_core = std::make_shared<ARM_Unicorn>(); LOG_WARNING(Core, "CPU JIT requested, but Dynarmic not available"); #endif - break; + } else { + cpu_core = std::make_shared<ARM_Unicorn>(); } gpu_core = std::make_unique<Tegra::GPU>(); diff --git a/src/core/core.h b/src/core/core.h index ada23b347..ade456cfc 100644 --- a/src/core/core.h +++ b/src/core/core.h @@ -7,11 +7,13 @@ #include <memory> #include <string> #include "common/common_types.h" +#include "core/hle/kernel/kernel.h" #include "core/hle/kernel/scheduler.h" #include "core/loader/loader.h" #include "core/memory.h" #include "core/perf_stats.h" #include "core/telemetry_session.h" +#include "video_core/debug_utils/debug_utils.h" #include "video_core/gpu.h" class EmuWindow; @@ -52,10 +54,10 @@ public: * is not required to do a full dispatch with each instruction. NOTE: the number of instructions * requested is not guaranteed to run, as this will be interrupted preemptively if a hardware * update is requested (e.g. on a thread switch). - * @param tight_loop Number of instructions to execute. + * @param tight_loop If false, the CPU single-steps. * @return Result status, indicating whether or not the operation succeeded. */ - ResultStatus RunLoop(int tight_loop = 100000); + ResultStatus RunLoop(bool tight_loop = true); /** * Step the CPU one instruction @@ -112,6 +114,10 @@ public: return *scheduler; } + Kernel::SharedPtr<Kernel::Process>& CurrentProcess() { + return current_process; + } + PerfStats perf_stats; FrameLimiter frame_limiter; @@ -130,6 +136,14 @@ public: return *app_loader; } + void SetGPUDebugContext(std::shared_ptr<Tegra::DebugContext> context) { + debug_context = std::move(context); + } + + std::shared_ptr<Tegra::DebugContext> GetGPUDebugContext() const { + return debug_context; + } + private: /** * Initialize the emulated system. @@ -149,6 +163,10 @@ private: std::unique_ptr<Kernel::Scheduler> scheduler; std::unique_ptr<Tegra::GPU> gpu_core; + std::shared_ptr<Tegra::DebugContext> debug_context; + + Kernel::SharedPtr<Kernel::Process> current_process; + /// When true, signals that a reschedule should happen bool reschedule_pending{}; @@ -169,4 +187,8 @@ inline TelemetrySession& Telemetry() { return System::GetInstance().TelemetrySession(); } +inline Kernel::SharedPtr<Kernel::Process>& CurrentProcess() { + return System::GetInstance().CurrentProcess(); +} + } // namespace Core diff --git a/src/core/file_sys/directory.h b/src/core/file_sys/directory.h index 5a40bf472..c7639795e 100644 --- a/src/core/file_sys/directory.h +++ b/src/core/file_sys/directory.h @@ -6,34 +6,28 @@ #include <array> #include <cstddef> +#include "common/common_funcs.h" #include "common/common_types.h" +#include "core/file_sys/filesystem.h" //////////////////////////////////////////////////////////////////////////////////////////////////// // FileSys namespace namespace FileSys { -// Structure of a directory entry, from http://3dbrew.org/wiki/FSDir:Read#Entry_format -const size_t FILENAME_LENGTH = 0x20C / 2; +// Structure of a directory entry, from +// http://switchbrew.org/index.php?title=Filesystem_services#DirectoryEntry +const size_t FILENAME_LENGTH = 0x300; struct Entry { - char16_t filename[FILENAME_LENGTH]; // Entry name (UTF-16, null-terminated) - std::array<char, 9> short_name; // 8.3 file name ('longfilename' -> 'LONGFI~1', null-terminated) - char unknown1; // unknown (observed values: 0x0A, 0x70, 0xFD) - std::array<char, 4> - extension; // 8.3 file extension (set to spaces for directories, null-terminated) - char unknown2; // unknown (always 0x01) - char unknown3; // unknown (0x00 or 0x08) - char is_directory; // directory flag - char is_hidden; // hidden flag - char is_archive; // archive flag - char is_read_only; // read-only flag - u64 file_size; // file size (for files only) + char filename[FILENAME_LENGTH]; + INSERT_PADDING_BYTES(4); + EntryType type; + INSERT_PADDING_BYTES(3); + u64 file_size; }; -static_assert(sizeof(Entry) == 0x228, "Directory Entry struct isn't exactly 0x228 bytes long!"); -static_assert(offsetof(Entry, short_name) == 0x20C, "Wrong offset for short_name in Entry."); -static_assert(offsetof(Entry, extension) == 0x216, "Wrong offset for extension in Entry."); -static_assert(offsetof(Entry, is_archive) == 0x21E, "Wrong offset for is_archive in Entry."); -static_assert(offsetof(Entry, file_size) == 0x220, "Wrong offset for file_size in Entry."); +static_assert(sizeof(Entry) == 0x310, "Directory Entry struct isn't exactly 0x310 bytes long!"); +static_assert(offsetof(Entry, type) == 0x304, "Wrong offset for type in Entry."); +static_assert(offsetof(Entry, file_size) == 0x308, "Wrong offset for file_size in Entry."); class DirectoryBackend : NonCopyable { public: @@ -46,7 +40,10 @@ public: * @param entries Buffer to read data into * @return Number of entries listed */ - virtual u32 Read(const u32 count, Entry* entries) = 0; + virtual u64 Read(const u64 count, Entry* entries) = 0; + + /// Returns the number of entries still left to read. + virtual u64 GetEntryCount() const = 0; /** * Close the directory diff --git a/src/core/file_sys/disk_filesystem.cpp b/src/core/file_sys/disk_filesystem.cpp new file mode 100644 index 000000000..4235f3935 --- /dev/null +++ b/src/core/file_sys/disk_filesystem.cpp @@ -0,0 +1,228 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <cstring> +#include <memory> +#include "common/common_types.h" +#include "common/logging/log.h" +#include "core/file_sys/disk_filesystem.h" +#include "core/file_sys/errors.h" + +namespace FileSys { + +static std::string ModeFlagsToString(Mode mode) { + std::string mode_str; + u32 mode_flags = static_cast<u32>(mode); + + // Calculate the correct open mode for the file. + if ((mode_flags & static_cast<u32>(Mode::Read)) && + (mode_flags & static_cast<u32>(Mode::Write))) { + if (mode_flags & static_cast<u32>(Mode::Append)) + mode_str = "a+"; + else + mode_str = "r+"; + } else { + if (mode_flags & static_cast<u32>(Mode::Read)) + mode_str = "r"; + else if (mode_flags & static_cast<u32>(Mode::Append)) + mode_str = "a"; + else if (mode_flags & static_cast<u32>(Mode::Write)) + mode_str = "w"; + } + + mode_str += "b"; + + return mode_str; +} + +std::string Disk_FileSystem::GetName() const { + return "Disk"; +} + +ResultVal<std::unique_ptr<StorageBackend>> Disk_FileSystem::OpenFile(const std::string& path, + Mode mode) const { + + // Calculate the correct open mode for the file. + std::string mode_str = ModeFlagsToString(mode); + + std::string full_path = base_directory + path; + auto file = std::make_shared<FileUtil::IOFile>(full_path, mode_str.c_str()); + + if (!file->IsOpen()) { + return ERROR_PATH_NOT_FOUND; + } + + return MakeResult<std::unique_ptr<StorageBackend>>( + std::make_unique<Disk_Storage>(std::move(file))); +} + +ResultCode Disk_FileSystem::DeleteFile(const Path& path) const { + LOG_WARNING(Service_FS, "(STUBBED) called"); + // TODO(bunnei): Use correct error code + return ResultCode(-1); +} + +ResultCode Disk_FileSystem::RenameFile(const Path& src_path, const Path& dest_path) const { + LOG_WARNING(Service_FS, "(STUBBED) called"); + // TODO(wwylele): Use correct error code + return ResultCode(-1); +} + +ResultCode Disk_FileSystem::DeleteDirectory(const Path& path) const { + LOG_WARNING(Service_FS, "(STUBBED) called"); + // TODO(wwylele): Use correct error code + return ResultCode(-1); +} + +ResultCode Disk_FileSystem::DeleteDirectoryRecursively(const Path& path) const { + LOG_WARNING(Service_FS, "(STUBBED) called"); + // TODO(wwylele): Use correct error code + return ResultCode(-1); +} + +ResultCode Disk_FileSystem::CreateFile(const std::string& path, u64 size) const { + LOG_WARNING(Service_FS, "(STUBBED) called"); + + std::string full_path = base_directory + path; + if (size == 0) { + FileUtil::CreateEmptyFile(full_path); + return RESULT_SUCCESS; + } + + FileUtil::IOFile file(full_path, "wb"); + // Creates a sparse file (or a normal file on filesystems without the concept of sparse files) + // We do this by seeking to the right size, then writing a single null byte. + if (file.Seek(size - 1, SEEK_SET) && file.WriteBytes("", 1) == 1) { + return RESULT_SUCCESS; + } + + LOG_ERROR(Service_FS, "Too large file"); + // TODO(Subv): Find out the correct error code + return ResultCode(-1); +} + +ResultCode Disk_FileSystem::CreateDirectory(const std::string& path) const { + // TODO(Subv): Perform path validation to prevent escaping the emulator sandbox. + std::string full_path = base_directory + path; + + if (FileUtil::CreateDir(full_path)) { + return RESULT_SUCCESS; + } + + LOG_CRITICAL(Service_FS, "(unreachable) Unknown error creating %s", full_path.c_str()); + // TODO(wwylele): Use correct error code + return ResultCode(-1); +} + +ResultCode Disk_FileSystem::RenameDirectory(const Path& src_path, const Path& dest_path) const { + LOG_WARNING(Service_FS, "(STUBBED) called"); + // TODO(wwylele): Use correct error code + return ResultCode(-1); +} + +ResultVal<std::unique_ptr<DirectoryBackend>> Disk_FileSystem::OpenDirectory( + const std::string& path) const { + + std::string full_path = base_directory + path; + + if (!FileUtil::IsDirectory(full_path)) { + // TODO(Subv): Find the correct error code for this. + return ResultCode(-1); + } + + auto directory = std::make_unique<Disk_Directory>(full_path); + return MakeResult<std::unique_ptr<DirectoryBackend>>(std::move(directory)); +} + +u64 Disk_FileSystem::GetFreeSpaceSize() const { + LOG_WARNING(Service_FS, "(STUBBED) called"); + return 0; +} + +ResultVal<FileSys::EntryType> Disk_FileSystem::GetEntryType(const std::string& path) const { + std::string full_path = base_directory + path; + if (!FileUtil::Exists(full_path)) { + return ERROR_PATH_NOT_FOUND; + } + + if (FileUtil::IsDirectory(full_path)) + return MakeResult(EntryType::Directory); + + return MakeResult(EntryType::File); +} + +ResultVal<size_t> Disk_Storage::Read(const u64 offset, const size_t length, u8* buffer) const { + LOG_TRACE(Service_FS, "called offset=%llu, length=%zu", offset, length); + file->Seek(offset, SEEK_SET); + return MakeResult<size_t>(file->ReadBytes(buffer, length)); +} + +ResultVal<size_t> Disk_Storage::Write(const u64 offset, const size_t length, const bool flush, + const u8* buffer) const { + LOG_WARNING(Service_FS, "(STUBBED) called"); + file->Seek(offset, SEEK_SET); + size_t written = file->WriteBytes(buffer, length); + if (flush) { + file->Flush(); + } + return MakeResult<size_t>(written); +} + +u64 Disk_Storage::GetSize() const { + return file->GetSize(); +} + +bool Disk_Storage::SetSize(const u64 size) const { + file->Resize(size); + file->Flush(); + return true; +} + +Disk_Directory::Disk_Directory(const std::string& path) : directory() { + unsigned size = FileUtil::ScanDirectoryTree(path, directory); + directory.size = size; + directory.isDirectory = true; + children_iterator = directory.children.begin(); +} + +u64 Disk_Directory::Read(const u64 count, Entry* entries) { + u64 entries_read = 0; + + while (entries_read < count && children_iterator != directory.children.cend()) { + const FileUtil::FSTEntry& file = *children_iterator; + const std::string& filename = file.virtualName; + Entry& entry = entries[entries_read]; + + LOG_TRACE(Service_FS, "File %s: size=%llu dir=%d", filename.c_str(), file.size, + file.isDirectory); + + // TODO(Link Mauve): use a proper conversion to UTF-16. + for (size_t j = 0; j < FILENAME_LENGTH; ++j) { + entry.filename[j] = filename[j]; + if (!filename[j]) + break; + } + + if (file.isDirectory) { + entry.file_size = 0; + entry.type = EntryType::Directory; + } else { + entry.file_size = file.size; + entry.type = EntryType::File; + } + + ++entries_read; + ++children_iterator; + } + return entries_read; +} + +u64 Disk_Directory::GetEntryCount() const { + // We convert the children iterator into a const_iterator to allow template argument deduction + // in std::distance. + std::vector<FileUtil::FSTEntry>::const_iterator current = children_iterator; + return std::distance(current, directory.children.end()); +} + +} // namespace FileSys diff --git a/src/core/file_sys/disk_filesystem.h b/src/core/file_sys/disk_filesystem.h new file mode 100644 index 000000000..742d7db1a --- /dev/null +++ b/src/core/file_sys/disk_filesystem.h @@ -0,0 +1,85 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <cstddef> +#include <memory> +#include <string> +#include "common/common_types.h" +#include "common/file_util.h" +#include "core/file_sys/directory.h" +#include "core/file_sys/filesystem.h" +#include "core/file_sys/storage.h" +#include "core/hle/result.h" + +namespace FileSys { + +class Disk_FileSystem : public FileSystemBackend { +public: + explicit Disk_FileSystem(std::string base_directory) + : base_directory(std::move(base_directory)) {} + + std::string GetName() const override; + + ResultVal<std::unique_ptr<StorageBackend>> OpenFile(const std::string& path, + Mode mode) const override; + ResultCode DeleteFile(const Path& path) const override; + ResultCode RenameFile(const Path& src_path, const Path& dest_path) const override; + ResultCode DeleteDirectory(const Path& path) const override; + ResultCode DeleteDirectoryRecursively(const Path& path) const override; + ResultCode CreateFile(const std::string& path, u64 size) const override; + ResultCode CreateDirectory(const std::string& path) const override; + ResultCode RenameDirectory(const Path& src_path, const Path& dest_path) const override; + ResultVal<std::unique_ptr<DirectoryBackend>> OpenDirectory( + const std::string& path) const override; + u64 GetFreeSpaceSize() const override; + ResultVal<EntryType> GetEntryType(const std::string& path) const override; + +protected: + std::string base_directory; +}; + +class Disk_Storage : public StorageBackend { +public: + Disk_Storage(std::shared_ptr<FileUtil::IOFile> file) : file(std::move(file)) {} + + ResultVal<size_t> Read(u64 offset, size_t length, u8* buffer) const override; + ResultVal<size_t> Write(u64 offset, size_t length, bool flush, const u8* buffer) const override; + u64 GetSize() const override; + bool SetSize(u64 size) const override; + bool Close() const override { + return false; + } + void Flush() const override {} + +private: + std::shared_ptr<FileUtil::IOFile> file; +}; + +class Disk_Directory : public DirectoryBackend { +public: + Disk_Directory(const std::string& path); + + ~Disk_Directory() override { + Close(); + } + + u64 Read(const u64 count, Entry* entries) override; + u64 GetEntryCount() const override; + + bool Close() const override { + return true; + } + +protected: + u32 total_entries_in_directory; + FileUtil::FSTEntry directory; + + // We need to remember the last entry we returned, so a subsequent call to Read will continue + // from the next one. This iterator will always point to the next unread entry. + std::vector<FileUtil::FSTEntry>::iterator children_iterator; +}; + +} // namespace FileSys diff --git a/src/core/file_sys/errors.h b/src/core/file_sys/errors.h index be3224ef8..0ed7d2a0c 100644 --- a/src/core/file_sys/errors.h +++ b/src/core/file_sys/errors.h @@ -10,36 +10,17 @@ namespace FileSys { namespace ErrCodes { enum { - RomFSNotFound = 100, - ArchiveNotMounted = 101, - FileNotFound = 112, - PathNotFound = 113, - GameCardNotInserted = 141, - NotFound = 120, - FileAlreadyExists = 180, - DirectoryAlreadyExists = 185, - AlreadyExists = 190, - InvalidOpenFlags = 230, - DirectoryNotEmpty = 240, - NotAFile = 250, - NotFormatted = 340, ///< This is used by the FS service when creating a SaveData archive - ExeFSSectionNotFound = 567, - CommandNotAllowed = 630, - InvalidReadFlag = 700, - InvalidPath = 702, - WriteBeyondEnd = 705, - UnsupportedOpenFlags = 760, - IncorrectExeFSReadSize = 761, - UnexpectedFileOrDirectory = 770, + NotFound = 1, }; } +constexpr ResultCode ERROR_PATH_NOT_FOUND(ErrorModule::FS, ErrCodes::NotFound); + // TODO(bunnei): Replace these with correct errors for Switch OS constexpr ResultCode ERROR_INVALID_PATH(ResultCode(-1)); constexpr ResultCode ERROR_UNSUPPORTED_OPEN_FLAGS(ResultCode(-1)); constexpr ResultCode ERROR_INVALID_OPEN_FLAGS(ResultCode(-1)); constexpr ResultCode ERROR_FILE_NOT_FOUND(ResultCode(-1)); -constexpr ResultCode ERROR_PATH_NOT_FOUND(ResultCode(-1)); constexpr ResultCode ERROR_UNEXPECTED_FILE_OR_DIRECTORY(ResultCode(-1)); constexpr ResultCode ERROR_DIRECTORY_ALREADY_EXISTS(ResultCode(-1)); constexpr ResultCode ERROR_FILE_ALREADY_EXISTS(ResultCode(-1)); diff --git a/src/core/file_sys/filesystem.h b/src/core/file_sys/filesystem.h index 02705506b..399427ca2 100644 --- a/src/core/file_sys/filesystem.h +++ b/src/core/file_sys/filesystem.h @@ -27,11 +27,15 @@ enum LowPathType : u32 { Wchar = 4, }; -union Mode { - u32 hex; - BitField<0, 1, u32> read_flag; - BitField<1, 1, u32> write_flag; - BitField<2, 1, u32> create_flag; +enum EntryType : u8 { + Directory = 0, + File = 1, +}; + +enum class Mode : u32 { + Read = 1, + Write = 2, + Append = 4, }; class Path { @@ -86,7 +90,7 @@ public: * @param size The size of the new file, filled with zeroes * @return Result of the operation */ - virtual ResultCode CreateFile(const Path& path, u64 size) const = 0; + virtual ResultCode CreateFile(const std::string& path, u64 size) const = 0; /** * Delete a file specified by its path @@ -100,7 +104,7 @@ public: * @param path Path relative to the archive * @return Result of the operation */ - virtual ResultCode CreateDirectory(const Path& path) const = 0; + virtual ResultCode CreateDirectory(const std::string& path) const = 0; /** * Delete a directory specified by its path @@ -138,21 +142,28 @@ public: * @param mode Mode to open the file with * @return Opened file, or error code */ - virtual ResultVal<std::unique_ptr<StorageBackend>> OpenFile(const Path& path, - const Mode& mode) const = 0; + virtual ResultVal<std::unique_ptr<StorageBackend>> OpenFile(const std::string& path, + Mode mode) const = 0; /** * Open a directory specified by its path * @param path Path relative to the archive * @return Opened directory, or error code */ - virtual ResultVal<std::unique_ptr<DirectoryBackend>> OpenDirectory(const Path& path) const = 0; + virtual ResultVal<std::unique_ptr<DirectoryBackend>> OpenDirectory( + const std::string& path) const = 0; /** * Get the free space * @return The number of free bytes in the archive */ virtual u64 GetFreeSpaceSize() const = 0; + + /** + * Get the type of the specified path + * @return The type of the specified path or error code + */ + virtual ResultVal<EntryType> GetEntryType(const std::string& path) const = 0; }; class FileSystemFactory : NonCopyable { @@ -174,10 +185,9 @@ public: /** * Deletes the archive contents and then re-creates the base folder * @param path Path to the archive - * @param format_info Format information for the new archive * @return ResultCode of the operation, 0 on success */ - virtual ResultCode Format(const Path& path, const FileSys::ArchiveFormatInfo& format_info) = 0; + virtual ResultCode Format(const Path& path) = 0; /** * Retrieves the format info about the archive with the specified path diff --git a/src/core/file_sys/romfs_factory.cpp b/src/core/file_sys/romfs_factory.cpp index e0de49f05..b21427948 100644 --- a/src/core/file_sys/romfs_factory.cpp +++ b/src/core/file_sys/romfs_factory.cpp @@ -23,7 +23,7 @@ ResultVal<std::unique_ptr<FileSystemBackend>> RomFS_Factory::Open(const Path& pa return MakeResult<std::unique_ptr<FileSystemBackend>>(std::move(archive)); } -ResultCode RomFS_Factory::Format(const Path& path, const FileSys::ArchiveFormatInfo& format_info) { +ResultCode RomFS_Factory::Format(const Path& path) { LOG_ERROR(Service_FS, "Unimplemented Format archive %s", GetName().c_str()); // TODO(bunnei): Find the right error code for this return ResultCode(-1); diff --git a/src/core/file_sys/romfs_factory.h b/src/core/file_sys/romfs_factory.h index 10ea13966..e0698e642 100644 --- a/src/core/file_sys/romfs_factory.h +++ b/src/core/file_sys/romfs_factory.h @@ -23,7 +23,7 @@ public: return "ArchiveFactory_RomFS"; } ResultVal<std::unique_ptr<FileSystemBackend>> Open(const Path& path) override; - ResultCode Format(const Path& path, const FileSys::ArchiveFormatInfo& format_info) override; + ResultCode Format(const Path& path) override; ResultVal<ArchiveFormatInfo> GetFormatInfo(const Path& path) const override; private: diff --git a/src/core/file_sys/romfs_filesystem.cpp b/src/core/file_sys/romfs_filesystem.cpp index ca1463d7c..0c6cc3157 100644 --- a/src/core/file_sys/romfs_filesystem.cpp +++ b/src/core/file_sys/romfs_filesystem.cpp @@ -14,8 +14,8 @@ std::string RomFS_FileSystem::GetName() const { return "RomFS"; } -ResultVal<std::unique_ptr<StorageBackend>> RomFS_FileSystem::OpenFile(const Path& path, - const Mode& mode) const { +ResultVal<std::unique_ptr<StorageBackend>> RomFS_FileSystem::OpenFile(const std::string& path, + Mode mode) const { return MakeResult<std::unique_ptr<StorageBackend>>( std::make_unique<RomFS_Storage>(romfs_file, data_offset, data_size)); } @@ -48,14 +48,14 @@ ResultCode RomFS_FileSystem::DeleteDirectoryRecursively(const Path& path) const return ResultCode(-1); } -ResultCode RomFS_FileSystem::CreateFile(const Path& path, u64 size) const { +ResultCode RomFS_FileSystem::CreateFile(const std::string& path, u64 size) const { LOG_CRITICAL(Service_FS, "Attempted to create a file in an ROMFS archive (%s).", GetName().c_str()); // TODO(bunnei): Use correct error code return ResultCode(-1); } -ResultCode RomFS_FileSystem::CreateDirectory(const Path& path) const { +ResultCode RomFS_FileSystem::CreateDirectory(const std::string& path) const { LOG_CRITICAL(Service_FS, "Attempted to create a directory in an ROMFS archive (%s).", GetName().c_str()); // TODO(wwylele): Use correct error code @@ -70,7 +70,8 @@ ResultCode RomFS_FileSystem::RenameDirectory(const Path& src_path, const Path& d } ResultVal<std::unique_ptr<DirectoryBackend>> RomFS_FileSystem::OpenDirectory( - const Path& path) const { + const std::string& path) const { + LOG_WARNING(Service_FS, "Opening Directory in a ROMFS archive"); return MakeResult<std::unique_ptr<DirectoryBackend>>(std::make_unique<ROMFSDirectory>()); } @@ -79,6 +80,12 @@ u64 RomFS_FileSystem::GetFreeSpaceSize() const { return 0; } +ResultVal<FileSys::EntryType> RomFS_FileSystem::GetEntryType(const std::string& path) const { + LOG_CRITICAL(Service_FS, "Called within an ROMFS archive (path %s).", path.c_str()); + // TODO(wwylele): Use correct error code + return ResultCode(-1); +} + ResultVal<size_t> RomFS_Storage::Read(const u64 offset, const size_t length, u8* buffer) const { LOG_TRACE(Service_FS, "called offset=%llu, length=%zu", offset, length); romfs_file->Seek(data_offset + offset, SEEK_SET); diff --git a/src/core/file_sys/romfs_filesystem.h b/src/core/file_sys/romfs_filesystem.h index 900ea567a..3f94c04d0 100644 --- a/src/core/file_sys/romfs_filesystem.h +++ b/src/core/file_sys/romfs_filesystem.h @@ -29,17 +29,19 @@ public: std::string GetName() const override; - ResultVal<std::unique_ptr<StorageBackend>> OpenFile(const Path& path, - const Mode& mode) const override; + ResultVal<std::unique_ptr<StorageBackend>> OpenFile(const std::string& path, + Mode mode) const override; ResultCode DeleteFile(const Path& path) const override; ResultCode RenameFile(const Path& src_path, const Path& dest_path) const override; ResultCode DeleteDirectory(const Path& path) const override; ResultCode DeleteDirectoryRecursively(const Path& path) const override; - ResultCode CreateFile(const Path& path, u64 size) const override; - ResultCode CreateDirectory(const Path& path) const override; + ResultCode CreateFile(const std::string& path, u64 size) const override; + ResultCode CreateDirectory(const std::string& path) const override; ResultCode RenameDirectory(const Path& src_path, const Path& dest_path) const override; - ResultVal<std::unique_ptr<DirectoryBackend>> OpenDirectory(const Path& path) const override; + ResultVal<std::unique_ptr<DirectoryBackend>> OpenDirectory( + const std::string& path) const override; u64 GetFreeSpaceSize() const override; + ResultVal<EntryType> GetEntryType(const std::string& path) const override; protected: std::shared_ptr<FileUtil::IOFile> romfs_file; @@ -69,7 +71,10 @@ private: class ROMFSDirectory : public DirectoryBackend { public: - u32 Read(const u32 count, Entry* entries) override { + u64 Read(const u64 count, Entry* entries) override { + return 0; + } + u64 GetEntryCount() const override { return 0; } bool Close() const override { diff --git a/src/core/file_sys/savedata_factory.cpp b/src/core/file_sys/savedata_factory.cpp new file mode 100644 index 000000000..14868fed2 --- /dev/null +++ b/src/core/file_sys/savedata_factory.cpp @@ -0,0 +1,57 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <cinttypes> +#include <memory> +#include "common/common_types.h" +#include "common/logging/log.h" +#include "common/string_util.h" +#include "core/core.h" +#include "core/file_sys/disk_filesystem.h" +#include "core/file_sys/savedata_factory.h" +#include "core/hle/kernel/process.h" + +namespace FileSys { + +SaveData_Factory::SaveData_Factory(std::string nand_directory) + : nand_directory(std::move(nand_directory)) {} + +ResultVal<std::unique_ptr<FileSystemBackend>> SaveData_Factory::Open(const Path& path) { + std::string save_directory = GetFullPath(); + // Return an error if the save data doesn't actually exist. + if (!FileUtil::IsDirectory(save_directory)) { + // TODO(Subv): Find out correct error code. + return ResultCode(-1); + } + + auto archive = std::make_unique<Disk_FileSystem>(save_directory); + return MakeResult<std::unique_ptr<FileSystemBackend>>(std::move(archive)); +} + +ResultCode SaveData_Factory::Format(const Path& path) { + LOG_WARNING(Service_FS, "Format archive %s", GetName().c_str()); + // Create the save data directory. + if (!FileUtil::CreateFullPath(GetFullPath())) { + // TODO(Subv): Find the correct error code. + return ResultCode(-1); + } + + return RESULT_SUCCESS; +} + +ResultVal<ArchiveFormatInfo> SaveData_Factory::GetFormatInfo(const Path& path) const { + LOG_ERROR(Service_FS, "Unimplemented GetFormatInfo archive %s", GetName().c_str()); + // TODO(bunnei): Find the right error code for this + return ResultCode(-1); +} + +std::string SaveData_Factory::GetFullPath() const { + u64 title_id = Core::CurrentProcess()->program_id; + // TODO(Subv): Somehow obtain this value. + u32 user = 0; + return Common::StringFromFormat("%ssave/%016" PRIX64 "/%08X/", nand_directory.c_str(), title_id, + user); +} + +} // namespace FileSys diff --git a/src/core/file_sys/savedata_factory.h b/src/core/file_sys/savedata_factory.h new file mode 100644 index 000000000..73a42aab6 --- /dev/null +++ b/src/core/file_sys/savedata_factory.h @@ -0,0 +1,33 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <memory> +#include <string> +#include "common/common_types.h" +#include "core/file_sys/filesystem.h" +#include "core/hle/result.h" + +namespace FileSys { + +/// File system interface to the SaveData archive +class SaveData_Factory final : public FileSystemFactory { +public: + explicit SaveData_Factory(std::string nand_directory); + + std::string GetName() const override { + return "SaveData_Factory"; + } + ResultVal<std::unique_ptr<FileSystemBackend>> Open(const Path& path) override; + ResultCode Format(const Path& path) override; + ResultVal<ArchiveFormatInfo> GetFormatInfo(const Path& path) const override; + +private: + std::string nand_directory; + + std::string GetFullPath() const; +}; + +} // namespace FileSys diff --git a/src/core/file_sys/sdmc_factory.cpp b/src/core/file_sys/sdmc_factory.cpp new file mode 100644 index 000000000..00e80d2a7 --- /dev/null +++ b/src/core/file_sys/sdmc_factory.cpp @@ -0,0 +1,40 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <cinttypes> +#include <memory> +#include "common/common_types.h" +#include "common/logging/log.h" +#include "common/string_util.h" +#include "core/core.h" +#include "core/file_sys/disk_filesystem.h" +#include "core/file_sys/sdmc_factory.h" + +namespace FileSys { + +SDMC_Factory::SDMC_Factory(std::string sd_directory) : sd_directory(std::move(sd_directory)) {} + +ResultVal<std::unique_ptr<FileSystemBackend>> SDMC_Factory::Open(const Path& path) { + // Create the SD Card directory if it doesn't already exist. + if (!FileUtil::IsDirectory(sd_directory)) { + FileUtil::CreateFullPath(sd_directory); + } + + auto archive = std::make_unique<Disk_FileSystem>(sd_directory); + return MakeResult<std::unique_ptr<FileSystemBackend>>(std::move(archive)); +} + +ResultCode SDMC_Factory::Format(const Path& path) { + LOG_ERROR(Service_FS, "Unimplemented Format archive %s", GetName().c_str()); + // TODO(Subv): Find the right error code for this + return ResultCode(-1); +} + +ResultVal<ArchiveFormatInfo> SDMC_Factory::GetFormatInfo(const Path& path) const { + LOG_ERROR(Service_FS, "Unimplemented GetFormatInfo archive %s", GetName().c_str()); + // TODO(bunnei): Find the right error code for this + return ResultCode(-1); +} + +} // namespace FileSys diff --git a/src/core/file_sys/sdmc_factory.h b/src/core/file_sys/sdmc_factory.h new file mode 100644 index 000000000..93becda25 --- /dev/null +++ b/src/core/file_sys/sdmc_factory.h @@ -0,0 +1,31 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <memory> +#include <string> +#include "common/common_types.h" +#include "core/file_sys/filesystem.h" +#include "core/hle/result.h" + +namespace FileSys { + +/// File system interface to the SDCard archive +class SDMC_Factory final : public FileSystemFactory { +public: + explicit SDMC_Factory(std::string sd_directory); + + std::string GetName() const override { + return "SDMC_Factory"; + } + ResultVal<std::unique_ptr<FileSystemBackend>> Open(const Path& path) override; + ResultCode Format(const Path& path) override; + ResultVal<ArchiveFormatInfo> GetFormatInfo(const Path& path) const override; + +private: + std::string sd_directory; +}; + +} // namespace FileSys diff --git a/src/core/gdbstub/gdbstub.cpp b/src/core/gdbstub/gdbstub.cpp index 7a142dc21..e4f337a0a 100644 --- a/src/core/gdbstub/gdbstub.cpp +++ b/src/core/gdbstub/gdbstub.cpp @@ -693,7 +693,7 @@ static void ReadMemory() { u64 len = HexToLong(start_offset, static_cast<u64>((command_buffer + command_length) - start_offset)); - LOG_DEBUG(Debug_GDBStub, "gdb: addr: %016llx len: %016llx\n", addr, len); + LOG_DEBUG(Debug_GDBStub, "gdb: addr: %016lx len: %016lx\n", addr, len); if (len * 2 > sizeof(reply)) { SendReply("E01"); diff --git a/src/core/hle/ipc_helpers.h b/src/core/hle/ipc_helpers.h index 6066d8a18..3f87c4297 100644 --- a/src/core/hle/ipc_helpers.h +++ b/src/core/hle/ipc_helpers.h @@ -118,7 +118,8 @@ public: AlignWithPadding(); - if (context.Session()->IsDomain()) { + const bool request_has_domain_header{context.GetDomainMessageHeader() != nullptr}; + if (context.Session()->IsDomain() && request_has_domain_header) { IPC::DomainMessageHeader domain_header{}; domain_header.num_objects = num_domain_objects; PushRaw(domain_header); diff --git a/src/core/hle/kernel/handle_table.cpp b/src/core/hle/kernel/handle_table.cpp index 3beb55753..822449cd5 100644 --- a/src/core/hle/kernel/handle_table.cpp +++ b/src/core/hle/kernel/handle_table.cpp @@ -5,6 +5,7 @@ #include <utility> #include "common/assert.h" #include "common/logging/log.h" +#include "core/core.h" #include "core/hle/kernel/errors.h" #include "core/hle/kernel/handle_table.h" #include "core/hle/kernel/kernel.h" @@ -77,7 +78,7 @@ SharedPtr<Object> HandleTable::GetGeneric(Handle handle) const { if (handle == CurrentThread) { return GetCurrentThread(); } else if (handle == CurrentProcess) { - return g_current_process; + return Core::CurrentProcess(); } if (!IsValid(handle)) { diff --git a/src/core/hle/kernel/hle_ipc.cpp b/src/core/hle/kernel/hle_ipc.cpp index 25ba26f18..bef4f15f5 100644 --- a/src/core/hle/kernel/hle_ipc.cpp +++ b/src/core/hle/kernel/hle_ipc.cpp @@ -7,6 +7,7 @@ #include "common/common_funcs.h" #include "common/common_types.h" #include "core/hle/ipc_helpers.h" +#include "core/hle/kernel/event.h" #include "core/hle/kernel/handle_table.h" #include "core/hle/kernel/hle_ipc.h" #include "core/hle/kernel/kernel.h" @@ -26,6 +27,32 @@ void SessionRequestHandler::ClientDisconnected(SharedPtr<ServerSession> server_s boost::range::remove_erase(connected_sessions, server_session); } +SharedPtr<Event> HLERequestContext::SleepClientThread(SharedPtr<Thread> thread, + const std::string& reason, u64 timeout, + WakeupCallback&& callback) { + + // Put the client thread to sleep until the wait event is signaled or the timeout expires. + thread->wakeup_callback = + [context = *this, callback](ThreadWakeupReason reason, SharedPtr<Thread> thread, + SharedPtr<WaitObject> object, size_t index) mutable -> bool { + ASSERT(thread->status == THREADSTATUS_WAIT_HLE_EVENT); + callback(thread, context, reason); + context.WriteToOutgoingCommandBuffer(*thread); + return true; + }; + + auto event = Kernel::Event::Create(Kernel::ResetType::OneShot, "HLE Pause Event: " + reason); + thread->status = THREADSTATUS_WAIT_HLE_EVENT; + thread->wait_objects = {event}; + event->AddWaitingThread(thread); + + if (timeout > 0) { + thread->WakeAfterDelay(timeout); + } + + return event; +} + HLERequestContext::HLERequestContext(SharedPtr<Kernel::ServerSession> server_session) : server_session(std::move(server_session)) { cmd_buf[0] = 0; @@ -35,7 +62,7 @@ HLERequestContext::~HLERequestContext() = default; void HLERequestContext::ParseCommandBuffer(u32_le* src_cmdbuf, bool incoming) { IPC::RequestParser rp(src_cmdbuf); - command_header = std::make_unique<IPC::CommandHeader>(rp.PopRaw<IPC::CommandHeader>()); + command_header = std::make_shared<IPC::CommandHeader>(rp.PopRaw<IPC::CommandHeader>()); if (command_header->type == IPC::CommandType::Close) { // Close does not populate the rest of the IPC header @@ -45,7 +72,7 @@ void HLERequestContext::ParseCommandBuffer(u32_le* src_cmdbuf, bool incoming) { // If handle descriptor is present, add size of it if (command_header->enable_handle_descriptor) { handle_descriptor_header = - std::make_unique<IPC::HandleDescriptorHeader>(rp.PopRaw<IPC::HandleDescriptorHeader>()); + std::make_shared<IPC::HandleDescriptorHeader>(rp.PopRaw<IPC::HandleDescriptorHeader>()); if (handle_descriptor_header->send_current_pid) { rp.Skip(2, false); } @@ -85,13 +112,18 @@ void HLERequestContext::ParseCommandBuffer(u32_le* src_cmdbuf, bool incoming) { if (Session()->IsDomain() && (command_header->type == IPC::CommandType::Request || !incoming)) { // If this is an incoming message, only CommandType "Request" has a domain header - // All outgoing domain messages have the domain header - domain_message_header = - std::make_unique<IPC::DomainMessageHeader>(rp.PopRaw<IPC::DomainMessageHeader>()); + // All outgoing domain messages have the domain header, if only incoming has it + if (incoming || domain_message_header) { + domain_message_header = + std::make_shared<IPC::DomainMessageHeader>(rp.PopRaw<IPC::DomainMessageHeader>()); + } else { + if (Session()->IsDomain()) + LOG_WARNING(IPC, "Domain request has no DomainMessageHeader!"); + } } data_payload_header = - std::make_unique<IPC::DataPayloadHeader>(rp.PopRaw<IPC::DataPayloadHeader>()); + std::make_shared<IPC::DataPayloadHeader>(rp.PopRaw<IPC::DataPayloadHeader>()); data_payload_offset = rp.GetCurrentOffset(); @@ -154,8 +186,11 @@ ResultCode HLERequestContext::PopulateFromIncomingCommandBuffer(u32_le* src_cmdb return RESULT_SUCCESS; } -ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(u32_le* dst_cmdbuf, Process& dst_process, - HandleTable& dst_table) { +ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(Thread& thread) { + std::array<u32, IPC::COMMAND_BUFFER_LENGTH> dst_cmdbuf; + Memory::ReadBlock(*thread.owner_process, thread.GetTLSAddress(), dst_cmdbuf.data(), + dst_cmdbuf.size() * sizeof(u32)); + // The header was already built in the internal command buffer. Attempt to parse it to verify // the integrity and then copy it over to the target command buffer. ParseCommandBuffer(cmd_buf.data(), false); @@ -166,7 +201,7 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(u32_le* dst_cmdbuf, P if (domain_message_header) size -= sizeof(IPC::DomainMessageHeader) / sizeof(u32); - std::copy_n(cmd_buf.begin(), size, dst_cmdbuf); + std::copy_n(cmd_buf.begin(), size, dst_cmdbuf.data()); if (command_header->enable_handle_descriptor) { ASSERT_MSG(!move_objects.empty() || !copy_objects.empty(), @@ -196,7 +231,7 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(u32_le* dst_cmdbuf, P // TODO(Subv): Translate the X/A/B/W buffers. - if (Session()->IsDomain()) { + if (Session()->IsDomain() && domain_message_header) { ASSERT(domain_message_header->num_objects == domain_objects.size()); // Write the domain objects to the command buffer, these go after the raw untranslated data. // TODO(Subv): This completely ignores C buffers. @@ -208,6 +243,11 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(u32_le* dst_cmdbuf, P dst_cmdbuf[domain_offset++] = static_cast<u32_le>(request_handlers.size()); } } + + // Copy the translated command buffer back into the thread's command buffer area. + Memory::WriteBlock(*thread.owner_process, thread.GetTLSAddress(), dst_cmdbuf.data(), + dst_cmdbuf.size() * sizeof(u32)); + return RESULT_SUCCESS; } @@ -228,8 +268,11 @@ std::vector<u8> HLERequestContext::ReadBuffer() const { size_t HLERequestContext::WriteBuffer(const void* buffer, size_t size) const { const bool is_buffer_b{BufferDescriptorB().size() && BufferDescriptorB()[0].Size()}; - - ASSERT_MSG(size <= GetWriteBufferSize(), "Size %d is too big", size); + const size_t buffer_size{GetWriteBufferSize()}; + if (size > buffer_size) { + LOG_CRITICAL(Core, "size (%016zx) is greater than buffer_size (%016zx)", size, buffer_size); + size = buffer_size; // TODO(bunnei): This needs to be HW tested + } if (is_buffer_b) { Memory::WriteBlock(BufferDescriptorB()[0].Address(), buffer, size); diff --git a/src/core/hle/kernel/hle_ipc.h b/src/core/hle/kernel/hle_ipc.h index b5631b773..8b35da4c9 100644 --- a/src/core/hle/kernel/hle_ipc.h +++ b/src/core/hle/kernel/hle_ipc.h @@ -6,6 +6,7 @@ #include <array> #include <memory> +#include <string> #include <vector> #include <boost/container/small_vector.hpp> #include "common/common_types.h" @@ -13,6 +14,7 @@ #include "core/hle/ipc.h" #include "core/hle/kernel/kernel.h" #include "core/hle/kernel/server_session.h" +#include "core/hle/kernel/thread.h" namespace Service { class ServiceFrameworkBase; @@ -24,6 +26,7 @@ class Domain; class HandleTable; class HLERequestContext; class Process; +class Event; /** * Interface implemented by HLE Session handlers. @@ -102,14 +105,31 @@ public: return server_session; } + using WakeupCallback = std::function<void(SharedPtr<Thread> thread, HLERequestContext& context, + ThreadWakeupReason reason)>; + + /** + * Puts the specified guest thread to sleep until the returned event is signaled or until the + * specified timeout expires. + * @param thread Thread to be put to sleep. + * @param reason Reason for pausing the thread, to be used for debugging purposes. + * @param timeout Timeout in nanoseconds after which the thread will be awoken and the callback + * invoked with a Timeout reason. + * @param callback Callback to be invoked when the thread is resumed. This callback must write + * the entire command response once again, regardless of the state of it before this function + * was called. + * @returns Event that when signaled will resume the thread and call the callback function. + */ + SharedPtr<Event> SleepClientThread(SharedPtr<Thread> thread, const std::string& reason, + u64 timeout, WakeupCallback&& callback); + void ParseCommandBuffer(u32_le* src_cmdbuf, bool incoming); /// Populates this context with data from the requesting process/thread. ResultCode PopulateFromIncomingCommandBuffer(u32_le* src_cmdbuf, Process& src_process, HandleTable& src_table); /// Writes data from this context back to the requesting process/thread. - ResultCode WriteToOutgoingCommandBuffer(u32_le* dst_cmdbuf, Process& dst_process, - HandleTable& dst_table); + ResultCode WriteToOutgoingCommandBuffer(Thread& thread); u32_le GetCommand() const { return command; @@ -139,7 +159,7 @@ public: return buffer_c_desciptors; } - const std::unique_ptr<IPC::DomainMessageHeader>& GetDomainMessageHeader() const { + const std::shared_ptr<IPC::DomainMessageHeader>& GetDomainMessageHeader() const { return domain_message_header; } @@ -212,10 +232,10 @@ private: boost::container::small_vector<SharedPtr<Object>, 8> copy_objects; boost::container::small_vector<std::shared_ptr<SessionRequestHandler>, 8> domain_objects; - std::unique_ptr<IPC::CommandHeader> command_header; - std::unique_ptr<IPC::HandleDescriptorHeader> handle_descriptor_header; - std::unique_ptr<IPC::DataPayloadHeader> data_payload_header; - std::unique_ptr<IPC::DomainMessageHeader> domain_message_header; + std::shared_ptr<IPC::CommandHeader> command_header; + std::shared_ptr<IPC::HandleDescriptorHeader> handle_descriptor_header; + std::shared_ptr<IPC::DataPayloadHeader> data_payload_header; + std::shared_ptr<IPC::DomainMessageHeader> domain_message_header; std::vector<IPC::BufferDescriptorX> buffer_x_desciptors; std::vector<IPC::BufferDescriptorABW> buffer_a_desciptors; std::vector<IPC::BufferDescriptorABW> buffer_b_desciptors; diff --git a/src/core/hle/kernel/kernel.cpp b/src/core/hle/kernel/kernel.cpp index b0c3f4ae1..b325b879b 100644 --- a/src/core/hle/kernel/kernel.cpp +++ b/src/core/hle/kernel/kernel.cpp @@ -41,7 +41,6 @@ void Shutdown() { g_object_address_table.Clear(); Kernel::ThreadingShutdown(); - g_current_process = nullptr; Kernel::TimersShutdown(); Kernel::ResourceLimitsShutdown(); diff --git a/src/core/hle/kernel/kernel.h b/src/core/hle/kernel/kernel.h index c77e58f3c..053bf4e17 100644 --- a/src/core/hle/kernel/kernel.h +++ b/src/core/hle/kernel/kernel.h @@ -33,10 +33,6 @@ enum class HandleType : u32 { ServerSession, }; -enum { - DEFAULT_STACK_SIZE = 0x10000, -}; - enum class ResetType { OneShot, Sticky, diff --git a/src/core/hle/kernel/object_address_table.cpp b/src/core/hle/kernel/object_address_table.cpp index 434c16add..cd286f85d 100644 --- a/src/core/hle/kernel/object_address_table.cpp +++ b/src/core/hle/kernel/object_address_table.cpp @@ -10,12 +10,12 @@ namespace Kernel { ObjectAddressTable g_object_address_table; void ObjectAddressTable::Insert(VAddr addr, SharedPtr<Object> obj) { - ASSERT_MSG(objects.find(addr) == objects.end(), "Object already exists with addr=0x%llx", addr); + ASSERT_MSG(objects.find(addr) == objects.end(), "Object already exists with addr=0x%lx", addr); objects[addr] = obj; } void ObjectAddressTable::Close(VAddr addr) { - ASSERT_MSG(objects.find(addr) != objects.end(), "Object does not exist with addr=0x%llx", addr); + ASSERT_MSG(objects.find(addr) != objects.end(), "Object does not exist with addr=0x%lx", addr); objects.erase(addr); } diff --git a/src/core/hle/kernel/process.cpp b/src/core/hle/kernel/process.cpp index 8e74059ea..2cffec198 100644 --- a/src/core/hle/kernel/process.cpp +++ b/src/core/hle/kernel/process.cpp @@ -20,12 +20,9 @@ namespace Kernel { // Lists all processes that exist in the current session. static std::vector<SharedPtr<Process>> process_list; -SharedPtr<CodeSet> CodeSet::Create(std::string name, u64 program_id) { +SharedPtr<CodeSet> CodeSet::Create(std::string name) { SharedPtr<CodeSet> codeset(new CodeSet); - codeset->name = std::move(name); - codeset->program_id = program_id; - return codeset; } @@ -41,6 +38,7 @@ SharedPtr<Process> Process::Create(std::string&& name) { process->flags.raw = 0; process->flags.memory_region.Assign(MemoryRegion::APPLICATION); process->status = ProcessStatus::Created; + process->program_id = 0; process_list.push_back(process); return process; @@ -119,11 +117,13 @@ void Process::ParseKernelCaps(const u32* kernel_caps, size_t len) { } void Process::Run(VAddr entry_point, s32 main_thread_priority, u32 stack_size) { - // Allocate and map stack + // Allocate and map the main thread stack + // TODO(bunnei): This is heap area that should be allocated by the kernel and not mapped as part + // of the user address space. vm_manager - .MapMemoryBlock(Memory::HEAP_VADDR_END - stack_size, + .MapMemoryBlock(Memory::STACK_AREA_VADDR_END - stack_size, std::make_shared<std::vector<u8>>(stack_size, 0), 0, stack_size, - MemoryState::Heap) + MemoryState::Mapped) .Unwrap(); misc_memory_used += stack_size; memory_region->used += stack_size; @@ -155,9 +155,9 @@ void Process::LoadModule(SharedPtr<CodeSet> module_, VAddr base_addr) { }; // Map CodeSet segments - MapSegment(module_->code, VMAPermission::ReadExecute, MemoryState::Code); - MapSegment(module_->rodata, VMAPermission::Read, MemoryState::Static); - MapSegment(module_->data, VMAPermission::ReadWrite, MemoryState::Static); + MapSegment(module_->code, VMAPermission::ReadExecute, MemoryState::CodeStatic); + MapSegment(module_->rodata, VMAPermission::Read, MemoryState::CodeMutable); + MapSegment(module_->data, VMAPermission::ReadWrite, MemoryState::CodeMutable); } VAddr Process::GetLinearHeapAreaAddress() const { @@ -184,6 +184,8 @@ ResultVal<VAddr> Process::HeapAllocate(VAddr target, u64 size, VMAPermission per // Initialize heap heap_memory = std::make_shared<std::vector<u8>>(); heap_start = heap_end = target; + } else { + vm_manager.UnmapRange(heap_start, heap_end - heap_start); } // If necessary, expand backing vector to cover new heap extents. @@ -203,7 +205,7 @@ ResultVal<VAddr> Process::HeapAllocate(VAddr target, u64 size, VMAPermission per size, MemoryState::Heap)); vm_manager.Reprotect(vma, perms); - heap_used += size; + heap_used = size; memory_region->used += size; return MakeResult<VAddr>(heap_end - size); @@ -290,7 +292,7 @@ ResultCode Process::MirrorMemory(VAddr dst_addr, VAddr src_addr, u64 size) { CASCADE_RESULT(auto new_vma, vm_manager.MapMemoryBlock(dst_addr, backing_block, backing_block_offset, size, - vma->second.meminfo_state)); + MemoryState::Mapped)); // Protect mirror with permissions from old region vm_manager.Reprotect(new_vma, vma->second.permissions); // Remove permissions from old region @@ -321,5 +323,4 @@ SharedPtr<Process> GetProcessById(u32 process_id) { return *itr; } -SharedPtr<Process> g_current_process; } // namespace Kernel diff --git a/src/core/hle/kernel/process.h b/src/core/hle/kernel/process.h index add98472f..68e77a4d1 100644 --- a/src/core/hle/kernel/process.h +++ b/src/core/hle/kernel/process.h @@ -56,7 +56,7 @@ class ResourceLimit; struct MemoryRegionInfo; struct CodeSet final : public Object { - static SharedPtr<CodeSet> Create(std::string name, u64 program_id); + static SharedPtr<CodeSet> Create(std::string name); std::string GetTypeName() const override { return "CodeSet"; @@ -72,8 +72,6 @@ struct CodeSet final : public Object { /// Name of the process std::string name; - /// Title ID corresponding to the process - u64 program_id; std::shared_ptr<std::vector<u8>> memory; @@ -113,6 +111,9 @@ public: static u32 next_process_id; + /// Title ID corresponding to the process + u64 program_id; + /// Resource limit descriptor for this process SharedPtr<ResourceLimit> resource_limit; @@ -202,5 +203,4 @@ void ClearProcessList(); /// Retrieves a process from the current list of processes. SharedPtr<Process> GetProcessById(u32 process_id); -extern SharedPtr<Process> g_current_process; } // namespace Kernel diff --git a/src/core/hle/kernel/scheduler.cpp b/src/core/hle/kernel/scheduler.cpp index 235068b22..921f27efb 100644 --- a/src/core/hle/kernel/scheduler.cpp +++ b/src/core/hle/kernel/scheduler.cpp @@ -2,6 +2,7 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include "core/core.h" #include "core/core_timing.h" #include "core/hle/kernel/process.h" #include "core/hle/kernel/scheduler.h" @@ -67,7 +68,7 @@ void Scheduler::SwitchContext(Thread* new_thread) { // Cancel any outstanding wakeup events for this thread new_thread->CancelWakeupTimer(); - auto previous_process = Kernel::g_current_process; + auto previous_process = Core::CurrentProcess(); current_thread = new_thread; @@ -75,8 +76,8 @@ void Scheduler::SwitchContext(Thread* new_thread) { new_thread->status = THREADSTATUS_RUNNING; if (previous_process != current_thread->owner_process) { - Kernel::g_current_process = current_thread->owner_process; - SetCurrentPageTable(&Kernel::g_current_process->vm_manager.page_table); + Core::CurrentProcess() = current_thread->owner_process; + SetCurrentPageTable(&Core::CurrentProcess()->vm_manager.page_table); } cpu_core->LoadContext(new_thread->context); diff --git a/src/core/hle/kernel/server_session.cpp b/src/core/hle/kernel/server_session.cpp index 5608418c3..33397d84f 100644 --- a/src/core/hle/kernel/server_session.cpp +++ b/src/core/hle/kernel/server_session.cpp @@ -4,6 +4,7 @@ #include <tuple> +#include "core/core.h" #include "core/hle/ipc_helpers.h" #include "core/hle/kernel/client_port.h" #include "core/hle/kernel/client_session.h" @@ -77,7 +78,8 @@ ResultCode ServerSession::HandleDomainSyncRequest(Kernel::HLERequestContext& con } } - LOG_CRITICAL(IPC, "Unknown domain command=%d", domain_message_header->command.Value()); + LOG_CRITICAL(IPC, "Unknown domain command=%d", + static_cast<int>(domain_message_header->command.Value())); ASSERT(false); } @@ -91,12 +93,12 @@ ResultCode ServerSession::HandleSyncRequest(SharedPtr<Thread> thread) { Kernel::HLERequestContext context(this); u32* cmd_buf = (u32*)Memory::GetPointer(thread->GetTLSAddress()); - context.PopulateFromIncomingCommandBuffer(cmd_buf, *Kernel::g_current_process, + context.PopulateFromIncomingCommandBuffer(cmd_buf, *Core::CurrentProcess(), Kernel::g_handle_table); ResultCode result = RESULT_SUCCESS; // If the session has been converted to a domain, handle the domain request - if (IsDomain()) { + if (IsDomain() && context.GetDomainMessageHeader()) { result = HandleDomainSyncRequest(context); // If there is no domain header, the regular session handler is used } else if (hle_handler != nullptr) { diff --git a/src/core/hle/kernel/shared_memory.cpp b/src/core/hle/kernel/shared_memory.cpp index d4505061e..88230bdd9 100644 --- a/src/core/hle/kernel/shared_memory.cpp +++ b/src/core/hle/kernel/shared_memory.cpp @@ -4,6 +4,7 @@ #include <cstring> #include "common/logging/log.h" +#include "core/core.h" #include "core/hle/kernel/errors.h" #include "core/hle/kernel/memory.h" #include "core/hle/kernel/shared_memory.h" @@ -51,8 +52,8 @@ SharedPtr<SharedMemory> SharedMemory::Create(SharedPtr<Process> owner_process, u } // Refresh the address mappings for the current process. - if (Kernel::g_current_process != nullptr) { - Kernel::g_current_process->vm_manager.RefreshMemoryBlockMappings(linheap_memory.get()); + if (Core::CurrentProcess() != nullptr) { + Core::CurrentProcess()->vm_manager.RefreshMemoryBlockMappings(linheap_memory.get()); } } else { auto& vm_manager = shared_memory->owner_process->vm_manager; @@ -106,7 +107,7 @@ ResultCode SharedMemory::Map(Process* target_process, VAddr address, MemoryPermi // Error out if the requested permissions don't match what the creator process allows. if (static_cast<u32>(permissions) & ~static_cast<u32>(own_other_permissions)) { - LOG_ERROR(Kernel, "cannot map id=%u, address=0x%llx name=%s, permissions don't match", + LOG_ERROR(Kernel, "cannot map id=%u, address=0x%lx name=%s, permissions don't match", GetObjectId(), address, name.c_str()); return ERR_INVALID_COMBINATION; } @@ -114,7 +115,7 @@ ResultCode SharedMemory::Map(Process* target_process, VAddr address, MemoryPermi // Error out if the provided permissions are not compatible with what the creator process needs. if (other_permissions != MemoryPermission::DontCare && static_cast<u32>(this->permissions) & ~static_cast<u32>(other_permissions)) { - LOG_ERROR(Kernel, "cannot map id=%u, address=0x%llx name=%s, permissions don't match", + LOG_ERROR(Kernel, "cannot map id=%u, address=0x%lx name=%s, permissions don't match", GetObjectId(), address, name.c_str()); return ERR_WRONG_PERMISSION; } @@ -125,7 +126,7 @@ ResultCode SharedMemory::Map(Process* target_process, VAddr address, MemoryPermi if (address != 0) { // TODO(shinyquagsire23): Check for virtual/mappable memory here too? if (address >= Memory::HEAP_VADDR && address < Memory::HEAP_VADDR_END) { - LOG_ERROR(Kernel, "cannot map id=%u, address=0x%llx name=%s, invalid address", + LOG_ERROR(Kernel, "cannot map id=%u, address=0x%lx name=%s, invalid address", GetObjectId(), address, name.c_str()); return ERR_INVALID_ADDRESS; } @@ -142,10 +143,9 @@ ResultCode SharedMemory::Map(Process* target_process, VAddr address, MemoryPermi auto result = target_process->vm_manager.MapMemoryBlock( target_address, backing_block, backing_block_offset, size, MemoryState::Shared); if (result.Failed()) { - LOG_ERROR( - Kernel, - "cannot map id=%u, target_address=0x%llx name=%s, error mapping to virtual memory", - GetObjectId(), target_address, name.c_str()); + LOG_ERROR(Kernel, + "cannot map id=%u, target_address=0x%lx name=%s, error mapping to virtual memory", + GetObjectId(), target_address, name.c_str()); return result.Code(); } diff --git a/src/core/hle/kernel/svc.cpp b/src/core/hle/kernel/svc.cpp index 1ab8cbd88..171bbd956 100644 --- a/src/core/hle/kernel/svc.cpp +++ b/src/core/hle/kernel/svc.cpp @@ -8,6 +8,7 @@ #include "common/logging/log.h" #include "common/microprofile.h" #include "common/string_util.h" +#include "core/core.h" #include "core/core_timing.h" #include "core/hle/kernel/client_port.h" #include "core/hle/kernel/client_session.h" @@ -31,14 +32,14 @@ namespace Kernel { /// Set the process heap to a given Size. It can both extend and shrink the heap. static ResultCode SetHeapSize(VAddr* heap_addr, u64 heap_size) { LOG_TRACE(Kernel_SVC, "called, heap_size=0x%llx", heap_size); - auto& process = *g_current_process; + auto& process = *Core::CurrentProcess(); CASCADE_RESULT(*heap_addr, process.HeapAllocate(Memory::HEAP_VADDR, heap_size, VMAPermission::ReadWrite)); return RESULT_SUCCESS; } static ResultCode SetMemoryAttribute(VAddr addr, u64 size, u32 state0, u32 state1) { - LOG_WARNING(Kernel_SVC, "(STUBBED) called, addr=0x%llx", addr); + LOG_WARNING(Kernel_SVC, "(STUBBED) called, addr=0x%lx", addr); return RESULT_SUCCESS; } @@ -46,14 +47,14 @@ static ResultCode SetMemoryAttribute(VAddr addr, u64 size, u32 state0, u32 state static ResultCode MapMemory(VAddr dst_addr, VAddr src_addr, u64 size) { LOG_TRACE(Kernel_SVC, "called, dst_addr=0x%llx, src_addr=0x%llx, size=0x%llx", dst_addr, src_addr, size); - return g_current_process->MirrorMemory(dst_addr, src_addr, size); + return Core::CurrentProcess()->MirrorMemory(dst_addr, src_addr, size); } /// Unmaps a region that was previously mapped with svcMapMemory static ResultCode UnmapMemory(VAddr dst_addr, VAddr src_addr, u64 size) { LOG_TRACE(Kernel_SVC, "called, dst_addr=0x%llx, src_addr=0x%llx, size=0x%llx", dst_addr, src_addr, size); - return g_current_process->UnmapMemory(dst_addr, src_addr, size); + return Core::CurrentProcess()->UnmapMemory(dst_addr, src_addr, size); } /// Connect to an OS service given the port name, returns the handle to the port to out @@ -306,23 +307,23 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id) LOG_TRACE(Kernel_SVC, "called info_id=0x%X, info_sub_id=0x%X, handle=0x%08X", info_id, info_sub_id, handle); - auto& vm_manager = g_current_process->vm_manager; + auto& vm_manager = Core::CurrentProcess()->vm_manager; switch (static_cast<GetInfoType>(info_id)) { case GetInfoType::AllowedCpuIdBitmask: - *result = g_current_process->allowed_processor_mask; + *result = Core::CurrentProcess()->allowed_processor_mask; break; case GetInfoType::AllowedThreadPrioBitmask: - *result = g_current_process->allowed_thread_priority_mask; + *result = Core::CurrentProcess()->allowed_thread_priority_mask; break; case GetInfoType::MapRegionBaseAddr: - *result = vm_manager.GetMapRegionBaseAddr(); + *result = Memory::MAP_REGION_VADDR; break; case GetInfoType::MapRegionSize: - *result = vm_manager.GetAddressSpaceSize(); + *result = Memory::MAP_REGION_SIZE; break; case GetInfoType::HeapRegionBaseAddr: - *result = vm_manager.GetNewMapRegionBaseAddr() + vm_manager.GetNewMapRegionSize(); + *result = Memory::HEAP_VADDR; break; case GetInfoType::HeapRegionSize: *result = Memory::HEAP_SIZE; @@ -346,13 +347,13 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id) *result = vm_manager.GetAddressSpaceSize(); break; case GetInfoType::NewMapRegionBaseAddr: - *result = vm_manager.GetNewMapRegionBaseAddr(); + *result = Memory::NEW_MAP_REGION_VADDR; break; case GetInfoType::NewMapRegionSize: - *result = vm_manager.GetNewMapRegionSize(); + *result = Memory::NEW_MAP_REGION_SIZE; break; case GetInfoType::IsVirtualAddressMemoryEnabled: - *result = g_current_process->is_virtual_address_memory_enabled; + *result = Core::CurrentProcess()->is_virtual_address_memory_enabled; break; case GetInfoType::TitleId: LOG_WARNING(Kernel_SVC, "(STUBBED) Attempted to query titleid, returned 0"); @@ -392,7 +393,7 @@ static ResultCode SetThreadPriority(Handle handle, u32 priority) { // Note: The kernel uses the current process's resource limit instead of // the one from the thread owner's resource limit. - SharedPtr<ResourceLimit>& resource_limit = g_current_process->resource_limit; + SharedPtr<ResourceLimit>& resource_limit = Core::CurrentProcess()->resource_limit; if (resource_limit->GetMaxResourceValue(ResourceTypes::PRIORITY) > priority) { return ERR_NOT_AUTHORIZED; } @@ -435,7 +436,7 @@ static ResultCode MapSharedMemory(Handle shared_memory_handle, VAddr addr, u64 s case MemoryPermission::WriteExecute: case MemoryPermission::ReadWriteExecute: case MemoryPermission::DontCare: - return shared_memory->Map(g_current_process.get(), addr, permissions_type, + return shared_memory->Map(Core::CurrentProcess().get(), addr, permissions_type, MemoryPermission::DontCare); default: LOG_ERROR(Kernel_SVC, "unknown permissions=0x%08X", permissions); @@ -451,7 +452,7 @@ static ResultCode UnmapSharedMemory(Handle shared_memory_handle, VAddr addr, u64 SharedPtr<SharedMemory> shared_memory = g_handle_table.Get<SharedMemory>(shared_memory_handle); - return shared_memory->Unmap(g_current_process.get(), addr); + return shared_memory->Unmap(Core::CurrentProcess().get(), addr); } /// Query process memory @@ -463,11 +464,11 @@ static ResultCode QueryProcessMemory(MemoryInfo* memory_info, PageInfo* /*page_i } auto vma = process->vm_manager.FindVMA(addr); memory_info->attributes = 0; - if (vma == g_current_process->vm_manager.vma_map.end()) { + if (vma == Core::CurrentProcess()->vm_manager.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::Free); + memory_info->type = static_cast<u32>(MemoryState::Unmapped); } else { memory_info->base_address = vma->second.base; memory_info->permission = static_cast<u32>(vma->second.permissions); @@ -487,16 +488,17 @@ static ResultCode QueryMemory(MemoryInfo* memory_info, PageInfo* page_info, VAdd /// Exits the current process static void ExitProcess() { - LOG_INFO(Kernel_SVC, "Process %u exiting", g_current_process->process_id); + LOG_INFO(Kernel_SVC, "Process %u exiting", Core::CurrentProcess()->process_id); - ASSERT_MSG(g_current_process->status == ProcessStatus::Running, "Process has already exited"); + ASSERT_MSG(Core::CurrentProcess()->status == ProcessStatus::Running, + "Process has already exited"); - g_current_process->status = ProcessStatus::Exited; + Core::CurrentProcess()->status = ProcessStatus::Exited; // Stop all the process threads that are currently waiting for objects. auto& thread_list = Core::System::GetInstance().Scheduler().GetThreadList(); for (auto& thread : thread_list) { - if (thread->owner_process != g_current_process) + if (thread->owner_process != Core::CurrentProcess()) continue; if (thread == GetCurrentThread()) @@ -525,14 +527,14 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V return ERR_OUT_OF_RANGE; } - SharedPtr<ResourceLimit>& resource_limit = g_current_process->resource_limit; + SharedPtr<ResourceLimit>& resource_limit = Core::CurrentProcess()->resource_limit; if (resource_limit->GetMaxResourceValue(ResourceTypes::PRIORITY) > priority) { return ERR_NOT_AUTHORIZED; } if (processor_id == THREADPROCESSORID_DEFAULT) { // Set the target CPU to the one specified in the process' exheader. - processor_id = g_current_process->ideal_processor; + processor_id = Core::CurrentProcess()->ideal_processor; ASSERT(processor_id != THREADPROCESSORID_DEFAULT); } @@ -554,7 +556,7 @@ static ResultCode CreateThread(Handle* out_handle, VAddr entry_point, u64 arg, V CASCADE_RESULT(SharedPtr<Thread> thread, Thread::Create(name, entry_point, priority, arg, processor_id, stack_top, - g_current_process)); + Core::CurrentProcess())); CASCADE_RESULT(thread->guest_handle, g_handle_table.Create(thread)); *out_handle = thread->guest_handle; @@ -748,14 +750,22 @@ static ResultCode ResetSignal(Handle handle) { /// Creates a TransferMemory object static ResultCode CreateTransferMemory(Handle* handle, VAddr addr, u64 size, u32 permissions) { - LOG_WARNING(Kernel_SVC, "(STUBBED) called addr=0x%llx, size=0x%llx, perms=%08X", addr, size, + LOG_WARNING(Kernel_SVC, "(STUBBED) called addr=0x%lx, size=0x%lx, perms=%08X", addr, size, permissions); *handle = 0; return RESULT_SUCCESS; } -static ResultCode SetThreadCoreMask(u64, u64, u64) { - LOG_WARNING(Kernel_SVC, "(STUBBED) called"); +static ResultCode GetThreadCoreMask(Handle handle, u32* mask, u64* unknown) { + LOG_WARNING(Kernel_SVC, "(STUBBED) called, handle=0x%08X", handle); + *mask = 0x0; + *unknown = 0xf; + return RESULT_SUCCESS; +} + +static ResultCode SetThreadCoreMask(Handle handle, u32 mask, u64 unknown) { + LOG_WARNING(Kernel_SVC, "(STUBBED) called, handle=0x%08X, mask=0x%08X, unknown=0x%lx", handle, + mask, unknown); return RESULT_SUCCESS; } @@ -807,7 +817,7 @@ static const FunctionDef SVC_Table[] = { {0x0B, SvcWrap<SleepThread>, "SleepThread"}, {0x0C, SvcWrap<GetThreadPriority>, "GetThreadPriority"}, {0x0D, SvcWrap<SetThreadPriority>, "SetThreadPriority"}, - {0x0E, nullptr, "GetThreadCoreMask"}, + {0x0E, SvcWrap<GetThreadCoreMask>, "GetThreadCoreMask"}, {0x0F, SvcWrap<SetThreadCoreMask>, "SetThreadCoreMask"}, {0x10, SvcWrap<GetCurrentProcessorNumber>, "GetCurrentProcessorNumber"}, {0x11, nullptr, "SignalEvent"}, diff --git a/src/core/hle/kernel/svc_wrap.h b/src/core/hle/kernel/svc_wrap.h index b224f5e67..5da4f5269 100644 --- a/src/core/hle/kernel/svc_wrap.h +++ b/src/core/hle/kernel/svc_wrap.h @@ -70,6 +70,21 @@ void SvcWrap() { FuncReturn(retval); } +template <ResultCode func(u32, u32, u64)> +void SvcWrap() { + FuncReturn(func((u32)(PARAM(0) & 0xFFFFFFFF), (u32)(PARAM(1) & 0xFFFFFFFF), PARAM(2)).raw); +} + +template <ResultCode func(u32, u32*, u64*)> +void SvcWrap() { + u32 param_1 = 0; + u64 param_2 = 0; + ResultCode retval = func((u32)(PARAM(2) & 0xFFFFFFFF), ¶m_1, ¶m_2); + Core::CPU().SetReg(1, param_1); + Core::CPU().SetReg(2, param_2); + FuncReturn(retval.raw); +} + template <ResultCode func(u64, u64, u32, u32)> void SvcWrap() { FuncReturn( diff --git a/src/core/hle/kernel/thread.cpp b/src/core/hle/kernel/thread.cpp index dd0a8ae48..f3a8aa4aa 100644 --- a/src/core/hle/kernel/thread.cpp +++ b/src/core/hle/kernel/thread.cpp @@ -55,16 +55,6 @@ inline static u32 const NewThreadId() { Thread::Thread() {} Thread::~Thread() {} -/** - * Check if the specified thread is waiting on the specified address to be arbitrated - * @param thread The thread to test - * @param wait_address The address to test against - * @return True if the thread is waiting, false otherwise - */ -static bool CheckWait_AddressArbiter(const Thread* thread, VAddr wait_address) { - return thread->status == THREADSTATUS_WAIT_ARB && wait_address == thread->wait_address; -} - void Thread::Stop() { // Cancel any outstanding wakeup events for this thread CoreTiming::UnscheduleEvent(ThreadWakeupEventType, callback_handle); @@ -94,7 +84,7 @@ void Thread::Stop() { u64 tls_page = (tls_address - Memory::TLS_AREA_VADDR) / Memory::PAGE_SIZE; u64 tls_slot = ((tls_address - Memory::TLS_AREA_VADDR) % Memory::PAGE_SIZE) / Memory::TLS_ENTRY_SIZE; - Kernel::g_current_process->tls_slots[tls_page].reset(tls_slot); + Core::CurrentProcess()->tls_slots[tls_page].reset(tls_slot); } void WaitCurrentThread_Sleep() { @@ -102,12 +92,6 @@ void WaitCurrentThread_Sleep() { thread->status = THREADSTATUS_WAIT_SLEEP; } -void WaitCurrentThread_ArbitrateAddress(VAddr wait_address) { - Thread* thread = GetCurrentThread(); - thread->wait_address = wait_address; - thread->status = THREADSTATUS_WAIT_ARB; -} - void ExitCurrentThread() { Thread* thread = GetCurrentThread(); thread->Stop(); @@ -129,7 +113,8 @@ static void ThreadWakeupCallback(u64 thread_handle, int cycles_late) { bool resume = true; if (thread->status == THREADSTATUS_WAIT_SYNCH_ANY || - thread->status == THREADSTATUS_WAIT_SYNCH_ALL || thread->status == THREADSTATUS_WAIT_ARB) { + thread->status == THREADSTATUS_WAIT_SYNCH_ALL || + thread->status == THREADSTATUS_WAIT_HLE_EVENT) { // Remove the thread from each of its waiting objects' waitlists for (auto& object : thread->wait_objects) @@ -163,7 +148,7 @@ void Thread::ResumeFromWait() { switch (status) { case THREADSTATUS_WAIT_SYNCH_ALL: case THREADSTATUS_WAIT_SYNCH_ANY: - case THREADSTATUS_WAIT_ARB: + case THREADSTATUS_WAIT_HLE_EVENT: case THREADSTATUS_WAIT_SLEEP: case THREADSTATUS_WAIT_IPC: break; @@ -314,7 +299,7 @@ ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point, // TODO(Subv): Find the correct MemoryState for this region. vm_manager.MapMemoryBlock(Memory::TLS_AREA_VADDR + available_page * Memory::PAGE_SIZE, linheap_memory, offset, Memory::PAGE_SIZE, - MemoryState::ThreadLocalStorage); + MemoryState::ThreadLocal); } // Mark the slot as used @@ -353,11 +338,11 @@ void Thread::BoostPriority(u32 priority) { SharedPtr<Thread> SetupMainThread(VAddr entry_point, u32 priority, SharedPtr<Process> owner_process) { // Setup page table so we can write to memory - SetCurrentPageTable(&Kernel::g_current_process->vm_manager.page_table); + SetCurrentPageTable(&Core::CurrentProcess()->vm_manager.page_table); // Initialize new "main" thread auto thread_res = Thread::Create("main", entry_point, priority, 0, THREADPROCESSORID_0, - Memory::HEAP_VADDR_END, owner_process); + Memory::STACK_AREA_VADDR_END, owner_process); SharedPtr<Thread> thread = std::move(thread_res).Unwrap(); @@ -406,6 +391,8 @@ void ThreadingInit() { next_thread_id = 1; } -void ThreadingShutdown() {} +void ThreadingShutdown() { + Kernel::ClearProcessList(); +} } // namespace Kernel diff --git a/src/core/hle/kernel/thread.h b/src/core/hle/kernel/thread.h index 4fd2fc2f8..dbf47e269 100644 --- a/src/core/hle/kernel/thread.h +++ b/src/core/hle/kernel/thread.h @@ -38,7 +38,7 @@ enum ThreadProcessorId : s32 { enum ThreadStatus { THREADSTATUS_RUNNING, ///< Currently running THREADSTATUS_READY, ///< Ready to run - THREADSTATUS_WAIT_ARB, ///< Waiting on an address arbiter + THREADSTATUS_WAIT_HLE_EVENT, ///< Waiting for hle event to finish THREADSTATUS_WAIT_SLEEP, ///< Waiting due to a SleepThread SVC THREADSTATUS_WAIT_IPC, ///< Waiting for the reply from an IPC request THREADSTATUS_WAIT_SYNCH_ANY, ///< Waiting due to WaitSynch1 or WaitSynchN with wait_all = false diff --git a/src/core/hle/kernel/vm_manager.cpp b/src/core/hle/kernel/vm_manager.cpp index d5b36d71a..1c2f873aa 100644 --- a/src/core/hle/kernel/vm_manager.cpp +++ b/src/core/hle/kernel/vm_manager.cpp @@ -18,8 +18,26 @@ namespace Kernel { static const char* GetMemoryStateName(MemoryState state) { static const char* names[] = { - "Free", "Reserved", "IO", "Static", "Code", "Private", - "Shared", "Continuous", "Aliased", "Alias", "AliasCode", "Locked", + "Unmapped", + "Io", + "Normal", + "CodeStatic", + "CodeMutable", + "Heap", + "Shared", + "Unknown1" + "ModuleCodeStatic", + "ModuleCodeMutable", + "IpcBuffer0", + "Mapped", + "ThreadLocal", + "TransferMemoryIsolated", + "TransferMemory", + "ProcessMemory", + "Unknown2" + "IpcBuffer1", + "IpcBuffer3", + "KernelStack", }; return names[(int)state]; @@ -142,7 +160,7 @@ VMManager::VMAIter VMManager::Unmap(VMAIter vma_handle) { VirtualMemoryArea& vma = vma_handle->second; vma.type = VMAType::Free; vma.permissions = VMAPermission::None; - vma.meminfo_state = MemoryState::Free; + vma.meminfo_state = MemoryState::Unmapped; vma.backing_block = nullptr; vma.offset = 0; @@ -166,6 +184,9 @@ ResultCode VMManager::UnmapRange(VAddr target, u64 size) { } ASSERT(FindVMA(target)->second.size >= size); + + Core::CPU().UnmapMemory(target, size); + return RESULT_SUCCESS; } @@ -377,19 +398,4 @@ u64 VMManager::GetAddressSpaceSize() { return MAX_ADDRESS; } -VAddr VMManager::GetMapRegionBaseAddr() { - LOG_WARNING(Kernel, "(STUBBED) called"); - return Memory::HEAP_VADDR; -} - -VAddr VMManager::GetNewMapRegionBaseAddr() { - LOG_WARNING(Kernel, "(STUBBED) called"); - return 0x8000000; -} - -u64 VMManager::GetNewMapRegionSize() { - LOG_WARNING(Kernel, "(STUBBED) called"); - return 0x8000000; -} - } // namespace Kernel diff --git a/src/core/hle/kernel/vm_manager.h b/src/core/hle/kernel/vm_manager.h index 8de704a60..4d66146f6 100644 --- a/src/core/hle/kernel/vm_manager.h +++ b/src/core/hle/kernel/vm_manager.h @@ -41,15 +41,24 @@ enum class VMAPermission : u8 { /// Set of values returned in MemoryInfo.state by svcQueryMemory. enum class MemoryState : u32 { - Free = 0, - IO = 1, - Normal = 2, - Code = 3, - Static = 4, - Heap = 5, - Shared = 6, - Mapped = 6, - ThreadLocalStorage = 12, + Unmapped = 0x0, + Io = 0x1, + Normal = 0x2, + CodeStatic = 0x3, + CodeMutable = 0x4, + Heap = 0x5, + Shared = 0x6, + ModuleCodeStatic = 0x8, + ModuleCodeMutable = 0x9, + IpcBuffer0 = 0xA, + Mapped = 0xB, + ThreadLocal = 0xC, + TransferMemoryIsolated = 0xD, + TransferMemory = 0xE, + ProcessMemory = 0xF, + IpcBuffer1 = 0x11, + IpcBuffer3 = 0x12, + KernelStack = 0x13, }; /** @@ -66,7 +75,7 @@ struct VirtualMemoryArea { VMAType type = VMAType::Free; VMAPermission permissions = VMAPermission::None; /// Tag returned by svcQueryMemory. Not otherwise used. - MemoryState meminfo_state = MemoryState::Free; + MemoryState meminfo_state = MemoryState::Unmapped; // Settings for type = AllocatedMemoryBlock /// Memory block backing this VMA. @@ -192,15 +201,6 @@ public: /// Gets the total address space address size, used by svcGetInfo u64 GetAddressSpaceSize(); - /// Gets the map region base address, used by svcGetInfo - VAddr GetMapRegionBaseAddr(); - - /// Gets the base address for a new memory region, used by svcGetInfo - VAddr GetNewMapRegionBaseAddr(); - - /// Gets the size for a new memory region, used by svcGetInfo - u64 GetNewMapRegionSize(); - /// Each VMManager has its own page table, which is set as the main one when the owning process /// is scheduled. Memory::PageTable page_table; diff --git a/src/core/hle/kernel/wait_object.cpp b/src/core/hle/kernel/wait_object.cpp index ec147b84c..b08ac72c1 100644 --- a/src/core/hle/kernel/wait_object.cpp +++ b/src/core/hle/kernel/wait_object.cpp @@ -39,7 +39,8 @@ SharedPtr<Thread> WaitObject::GetHighestPriorityReadyThread() { for (const auto& thread : waiting_threads) { // The list of waiting threads must not contain threads that are not waiting to be awakened. ASSERT_MSG(thread->status == THREADSTATUS_WAIT_SYNCH_ANY || - thread->status == THREADSTATUS_WAIT_SYNCH_ALL, + thread->status == THREADSTATUS_WAIT_SYNCH_ALL || + thread->status == THREADSTATUS_WAIT_HLE_EVENT, "Inconsistent thread statuses in waiting_threads"); if (thread->current_priority >= candidate_priority) diff --git a/src/core/hle/result.h b/src/core/hle/result.h index 656e1b4a7..052f49979 100644 --- a/src/core/hle/result.h +++ b/src/core/hle/result.h @@ -108,11 +108,11 @@ union ResultCode { } constexpr bool IsSuccess() const { - return is_error.ExtractValue(raw) == 0; + return raw == 0; } constexpr bool IsError() const { - return is_error.ExtractValue(raw) == 1; + return raw != 0; } }; @@ -200,6 +200,9 @@ public: } ResultVal& operator=(const ResultVal& o) { + if (this == &o) { + return *this; + } if (!empty()) { if (!o.empty()) { object = o.object; diff --git a/src/core/hle/service/am/am.cpp b/src/core/hle/service/am/am.cpp index d3a674cf6..bab338205 100644 --- a/src/core/hle/service/am/am.cpp +++ b/src/core/hle/service/am/am.cpp @@ -2,13 +2,17 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <cinttypes> +#include "core/file_sys/filesystem.h" #include "core/hle/ipc_helpers.h" #include "core/hle/kernel/event.h" #include "core/hle/service/am/am.h" #include "core/hle/service/am/applet_ae.h" #include "core/hle/service/am/applet_oe.h" #include "core/hle/service/apm/apm.h" +#include "core/hle/service/filesystem/filesystem.h" #include "core/hle/service/nvflinger/nvflinger.h" +#include "core/settings.h" namespace Service { namespace AM { @@ -238,17 +242,20 @@ void ICommonStateGetter::GetCurrentFocusState(Kernel::HLERequestContext& ctx) { } void ICommonStateGetter::GetOperationMode(Kernel::HLERequestContext& ctx) { + const bool use_docked_mode{Settings::values.use_docked_mode}; IPC::ResponseBuilder rb{ctx, 3}; rb.Push(RESULT_SUCCESS); - rb.Push(static_cast<u8>(OperationMode::Handheld)); + rb.Push(static_cast<u8>(use_docked_mode ? OperationMode::Docked : OperationMode::Handheld)); LOG_WARNING(Service_AM, "(STUBBED) called"); } void ICommonStateGetter::GetPerformanceMode(Kernel::HLERequestContext& ctx) { + const bool use_docked_mode{Settings::values.use_docked_mode}; IPC::ResponseBuilder rb{ctx, 3}; rb.Push(RESULT_SUCCESS); - rb.Push(static_cast<u32>(APM::PerformanceMode::Handheld)); + rb.Push(static_cast<u32>(use_docked_mode ? APM::PerformanceMode::Docked + : APM::PerformanceMode::Handheld)); LOG_WARNING(Service_AM, "(STUBBED) called"); } @@ -416,9 +423,24 @@ void IApplicationFunctions::PopLaunchParameter(Kernel::HLERequestContext& ctx) { } void IApplicationFunctions::EnsureSaveData(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service, "(STUBBED) called"); + IPC::RequestParser rp{ctx}; + u128 uid = rp.PopRaw<u128>(); + + LOG_WARNING(Service, "(STUBBED) called uid = %016" PRIX64 "%016" PRIX64, uid[1], uid[0]); + IPC::ResponseBuilder rb{ctx, 4}; - rb.Push(RESULT_SUCCESS); + + FileSys::Path unused; + auto savedata = FileSystem::OpenFileSystem(FileSystem::Type::SaveData, unused); + if (savedata.Failed()) { + // Create the save data and return an error indicating that the operation was performed. + FileSystem::FormatFileSystem(FileSystem::Type::SaveData); + // TODO(Subv): Find out the correct error code for this. + rb.Push(ResultCode(ErrorModule::FS, 40)); + } else { + rb.Push(RESULT_SUCCESS); + } + rb.Push<u64>(0); } diff --git a/src/core/hle/service/audio/audout_u.cpp b/src/core/hle/service/audio/audout_u.cpp index 780a4e6e5..e873d768f 100644 --- a/src/core/hle/service/audio/audout_u.cpp +++ b/src/core/hle/service/audio/audout_u.cpp @@ -52,7 +52,9 @@ public: CoreTiming::ScheduleEvent(audio_ticks, audio_event); } - ~IAudioOut() = default; + ~IAudioOut() { + CoreTiming::UnscheduleEvent(audio_event, 0); + } private: void StartAudioOut(Kernel::HLERequestContext& ctx) { diff --git a/src/core/hle/service/audio/audren_u.cpp b/src/core/hle/service/audio/audren_u.cpp index dda135d18..6d0461bbc 100644 --- a/src/core/hle/service/audio/audren_u.cpp +++ b/src/core/hle/service/audio/audren_u.cpp @@ -45,7 +45,9 @@ public: // Start the audio event CoreTiming::ScheduleEvent(audio_ticks, audio_event); } - ~IAudioRenderer() = default; + ~IAudioRenderer() { + CoreTiming::UnscheduleEvent(audio_event, 0); + } private: void UpdateAudioCallback() { @@ -57,12 +59,12 @@ private: AudioRendererResponseData response_data{}; response_data.section_0_size = - response_data.state_entries.size() * sizeof(AudioRendererStateEntry); - response_data.section_1_size = response_data.section_1.size(); - response_data.section_2_size = response_data.section_2.size(); - response_data.section_3_size = response_data.section_3.size(); - response_data.section_4_size = response_data.section_4.size(); - response_data.section_5_size = response_data.section_5.size(); + static_cast<u32>(response_data.state_entries.size() * sizeof(AudioRendererStateEntry)); + response_data.section_1_size = static_cast<u32>(response_data.section_1.size()); + response_data.section_2_size = static_cast<u32>(response_data.section_2.size()); + response_data.section_3_size = static_cast<u32>(response_data.section_3.size()); + response_data.section_4_size = static_cast<u32>(response_data.section_4.size()); + response_data.section_5_size = static_cast<u32>(response_data.section_5.size()); response_data.total_size = sizeof(AudioRendererResponseData); for (unsigned i = 0; i < response_data.state_entries.size(); i++) { @@ -149,12 +151,80 @@ private: Kernel::SharedPtr<Kernel::Event> system_event; }; +class IAudioDevice final : public ServiceFramework<IAudioDevice> { +public: + IAudioDevice() : ServiceFramework("IAudioDevice") { + static const FunctionInfo functions[] = { + {0x0, &IAudioDevice::ListAudioDeviceName, "ListAudioDeviceName"}, + {0x1, &IAudioDevice::SetAudioDeviceOutputVolume, "SetAudioDeviceOutputVolume"}, + {0x2, nullptr, "GetAudioDeviceOutputVolume"}, + {0x3, nullptr, "GetActiveAudioDeviceName"}, + {0x4, &IAudioDevice::QueryAudioDeviceSystemEvent, "QueryAudioDeviceSystemEvent"}, + {0x5, &IAudioDevice::GetActiveChannelCount, "GetActiveChannelCount"}, + {0x6, nullptr, "ListAudioDeviceNameAuto"}, + {0x7, nullptr, "SetAudioDeviceOutputVolumeAuto"}, + {0x8, nullptr, "GetAudioDeviceOutputVolumeAuto"}, + {0x10, nullptr, "GetActiveAudioDeviceNameAuto"}, + {0x11, nullptr, "QueryAudioDeviceInputEvent"}, + {0x12, nullptr, "QueryAudioDeviceOutputEvent"}}; + RegisterHandlers(functions); + + buffer_event = + Kernel::Event::Create(Kernel::ResetType::OneShot, "IAudioOutBufferReleasedEvent"); + } + +private: + void ListAudioDeviceName(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_Audio, "(STUBBED) called"); + IPC::RequestParser rp{ctx}; + + const std::string audio_interface = "AudioInterface"; + ctx.WriteBuffer(audio_interface.c_str(), audio_interface.size()); + + IPC::ResponseBuilder rb = rp.MakeBuilder(3, 0, 0); + rb.Push(RESULT_SUCCESS); + rb.Push<u32>(1); + } + + void SetAudioDeviceOutputVolume(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_Audio, "(STUBBED) called"); + + IPC::RequestParser rp{ctx}; + f32 volume = static_cast<f32>(rp.Pop<u32>()); + + auto file_buffer = ctx.ReadBuffer(); + auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0'); + + IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0); + rb.Push(RESULT_SUCCESS); + } + + void QueryAudioDeviceSystemEvent(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_Audio, "(STUBBED) called"); + + buffer_event->Signal(); + + IPC::ResponseBuilder rb{ctx, 2, 1}; + rb.Push(RESULT_SUCCESS); + rb.PushCopyObjects(buffer_event); + } + + void GetActiveChannelCount(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_Audio, "(STUBBED) called"); + IPC::ResponseBuilder rb{ctx, 3}; + rb.Push(RESULT_SUCCESS); + rb.Push<u32>(1); + } + + Kernel::SharedPtr<Kernel::Event> buffer_event; + +}; // namespace Audio + AudRenU::AudRenU() : ServiceFramework("audren:u") { static const FunctionInfo functions[] = { {0, &AudRenU::OpenAudioRenderer, "OpenAudioRenderer"}, {1, &AudRenU::GetAudioRendererWorkBufferSize, "GetAudioRendererWorkBufferSize"}, - {2, &AudRenU::GetAudioRenderersProcessMasterVolume, "GetAudioRenderersProcessMasterVolume"}, - {3, nullptr, "SetAudioRenderersProcessMasterVolume"}, + {2, &AudRenU::GetAudioDevice, "GetAudioDevice"}, }; RegisterHandlers(functions); } @@ -177,12 +247,13 @@ void AudRenU::GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service_Audio, "(STUBBED) called"); } -void AudRenU::GetAudioRenderersProcessMasterVolume(Kernel::HLERequestContext& ctx) { - IPC::ResponseBuilder rb{ctx, 3}; +void AudRenU::GetAudioDevice(Kernel::HLERequestContext& ctx) { + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; rb.Push(RESULT_SUCCESS); - rb.Push<u32>(100); - LOG_WARNING(Service_Audio, "(STUBBED) called"); + rb.PushIpcInterface<Audio::IAudioDevice>(); + + LOG_DEBUG(Service_Audio, "called"); } } // namespace Audio diff --git a/src/core/hle/service/audio/audren_u.h b/src/core/hle/service/audio/audren_u.h index 939d353a9..f59d1627d 100644 --- a/src/core/hle/service/audio/audren_u.h +++ b/src/core/hle/service/audio/audren_u.h @@ -21,7 +21,7 @@ public: private: void OpenAudioRenderer(Kernel::HLERequestContext& ctx); void GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx); - void GetAudioRenderersProcessMasterVolume(Kernel::HLERequestContext& ctx); + void GetAudioDevice(Kernel::HLERequestContext& ctx); }; } // namespace Audio diff --git a/src/core/hle/service/fatal/fatal.cpp b/src/core/hle/service/fatal/fatal.cpp new file mode 100644 index 000000000..1a18e0051 --- /dev/null +++ b/src/core/hle/service/fatal/fatal.cpp @@ -0,0 +1,38 @@ +// 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/hle/ipc_helpers.h" +#include "core/hle/service/fatal/fatal.h" +#include "core/hle/service/fatal/fatal_p.h" +#include "core/hle/service/fatal/fatal_u.h" + +namespace Service { +namespace Fatal { + +Module::Interface::Interface(std::shared_ptr<Module> module, const char* name) + : ServiceFramework(name), module(std::move(module)) {} + +void Module::Interface::FatalSimple(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp(ctx); + u32 error_code = rp.Pop<u32>(); + LOG_WARNING(Service_Fatal, "(STUBBED) called, error_code=0x%X", error_code); + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + +void Module::Interface::TransitionToFatalError(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_Fatal, "(STUBBED) called"); + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + +void InstallInterfaces(SM::ServiceManager& service_manager) { + auto module = std::make_shared<Module>(); + std::make_shared<Fatal_P>(module)->InstallAsService(service_manager); + std::make_shared<Fatal_U>(module)->InstallAsService(service_manager); +} + +} // namespace Fatal +} // namespace Service diff --git a/src/core/hle/service/fatal/fatal.h b/src/core/hle/service/fatal/fatal.h new file mode 100644 index 000000000..85272b4be --- /dev/null +++ b/src/core/hle/service/fatal/fatal.h @@ -0,0 +1,29 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/service/service.h" + +namespace Service { +namespace Fatal { + +class Module final { +public: + class Interface : public ServiceFramework<Interface> { + public: + Interface(std::shared_ptr<Module> module, const char* name); + + void FatalSimple(Kernel::HLERequestContext& ctx); + void TransitionToFatalError(Kernel::HLERequestContext& ctx); + + protected: + std::shared_ptr<Module> module; + }; +}; + +void InstallInterfaces(SM::ServiceManager& service_manager); + +} // namespace Fatal +} // namespace Service diff --git a/src/core/hle/service/fatal/fatal_p.cpp b/src/core/hle/service/fatal/fatal_p.cpp new file mode 100644 index 000000000..ba194e340 --- /dev/null +++ b/src/core/hle/service/fatal/fatal_p.cpp @@ -0,0 +1,14 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "core/hle/service/fatal/fatal_p.h" + +namespace Service { +namespace Fatal { + +Fatal_P::Fatal_P(std::shared_ptr<Module> module) + : Module::Interface(std::move(module), "fatal:p") {} + +} // namespace Fatal +} // namespace Service diff --git a/src/core/hle/service/fatal/fatal_p.h b/src/core/hle/service/fatal/fatal_p.h new file mode 100644 index 000000000..d77b24bc4 --- /dev/null +++ b/src/core/hle/service/fatal/fatal_p.h @@ -0,0 +1,18 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/service/fatal/fatal.h" + +namespace Service { +namespace Fatal { + +class Fatal_P final : public Module::Interface { +public: + explicit Fatal_P(std::shared_ptr<Module> module); +}; + +} // namespace Fatal +} // namespace Service diff --git a/src/core/hle/service/fatal/fatal_u.cpp b/src/core/hle/service/fatal/fatal_u.cpp new file mode 100644 index 000000000..065cc868d --- /dev/null +++ b/src/core/hle/service/fatal/fatal_u.cpp @@ -0,0 +1,19 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "core/hle/service/fatal/fatal_u.h" + +namespace Service { +namespace Fatal { + +Fatal_U::Fatal_U(std::shared_ptr<Module> module) : Module::Interface(std::move(module), "fatal:u") { + static const FunctionInfo functions[] = { + {1, &Fatal_U::FatalSimple, "FatalSimple"}, + {2, &Fatal_U::TransitionToFatalError, "TransitionToFatalError"}, + }; + RegisterHandlers(functions); +} + +} // namespace Fatal +} // namespace Service diff --git a/src/core/hle/service/fatal/fatal_u.h b/src/core/hle/service/fatal/fatal_u.h new file mode 100644 index 000000000..22374755e --- /dev/null +++ b/src/core/hle/service/fatal/fatal_u.h @@ -0,0 +1,18 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/service/fatal/fatal.h" + +namespace Service { +namespace Fatal { + +class Fatal_U final : public Module::Interface { +public: + explicit Fatal_U(std::shared_ptr<Module> module); +}; + +} // namespace Fatal +} // namespace Service diff --git a/src/core/hle/service/filesystem/filesystem.cpp b/src/core/hle/service/filesystem/filesystem.cpp index 4b47548fd..945832e98 100644 --- a/src/core/hle/service/filesystem/filesystem.cpp +++ b/src/core/hle/service/filesystem/filesystem.cpp @@ -3,7 +3,10 @@ // Refer to the license.txt file included. #include <boost/container/flat_map.hpp> +#include "common/file_util.h" #include "core/file_sys/filesystem.h" +#include "core/file_sys/savedata_factory.h" +#include "core/file_sys/sdmc_factory.h" #include "core/hle/service/filesystem/filesystem.h" #include "core/hle/service/filesystem/fsp_srv.h" @@ -41,12 +44,34 @@ ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenFileSystem(Type type, return itr->second->Open(path); } -void UnregisterFileSystems() { +ResultCode FormatFileSystem(Type type) { + LOG_TRACE(Service_FS, "Formatting FileSystem with type=%d", type); + + auto itr = filesystem_map.find(type); + if (itr == filesystem_map.end()) { + // TODO(bunnei): Find a better error code for this + return ResultCode(-1); + } + + FileSys::Path unused; + return itr->second->Format(unused); +} + +void RegisterFileSystems() { filesystem_map.clear(); + + std::string nand_directory = FileUtil::GetUserPath(D_NAND_IDX); + std::string sd_directory = FileUtil::GetUserPath(D_SDMC_IDX); + + auto savedata = std::make_unique<FileSys::SaveData_Factory>(std::move(nand_directory)); + RegisterFileSystem(std::move(savedata), Type::SaveData); + + auto sdcard = std::make_unique<FileSys::SDMC_Factory>(std::move(sd_directory)); + RegisterFileSystem(std::move(sdcard), Type::SDMC); } void InstallInterfaces(SM::ServiceManager& service_manager) { - UnregisterFileSystems(); + RegisterFileSystems(); std::make_shared<FSP_SRV>()->InstallAsService(service_manager); } diff --git a/src/core/hle/service/filesystem/filesystem.h b/src/core/hle/service/filesystem/filesystem.h index a674c9493..56d26146e 100644 --- a/src/core/hle/service/filesystem/filesystem.h +++ b/src/core/hle/service/filesystem/filesystem.h @@ -25,6 +25,8 @@ namespace FileSystem { /// Supported FileSystem types enum class Type { RomFS = 1, + SaveData = 2, + SDMC = 3, }; /** @@ -43,6 +45,13 @@ ResultCode RegisterFileSystem(std::unique_ptr<FileSys::FileSystemFactory>&& fact ResultVal<std::unique_ptr<FileSys::FileSystemBackend>> OpenFileSystem(Type type, FileSys::Path& path); +/** + * Formats a file system + * @param type Type of the file system to format + * @return ResultCode of the operation + */ +ResultCode FormatFileSystem(Type type); + /// Registers all Filesystem services with the specified service manager. void InstallInterfaces(SM::ServiceManager& service_manager); diff --git a/src/core/hle/service/filesystem/fsp_srv.cpp b/src/core/hle/service/filesystem/fsp_srv.cpp index 87a07e457..89fa70ae6 100644 --- a/src/core/hle/service/filesystem/fsp_srv.cpp +++ b/src/core/hle/service/filesystem/fsp_srv.cpp @@ -2,8 +2,10 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <cinttypes> #include "common/logging/log.h" #include "core/core.h" +#include "core/file_sys/directory.h" #include "core/file_sys/filesystem.h" #include "core/file_sys/storage.h" #include "core/hle/ipc_helpers.h" @@ -34,7 +36,7 @@ private: const s64 offset = rp.Pop<s64>(); const s64 length = rp.Pop<s64>(); - LOG_DEBUG(Service_FS, "called, offset=0x%llx, length=0x%llx", offset, length); + LOG_DEBUG(Service_FS, "called, offset=0x%ld, length=0x%ld", offset, length); // Error checking if (length < 0) { @@ -65,10 +67,299 @@ private: } }; +class IFile final : public ServiceFramework<IFile> { +public: + explicit IFile(std::unique_ptr<FileSys::StorageBackend>&& backend) + : ServiceFramework("IFile"), backend(std::move(backend)) { + static const FunctionInfo functions[] = { + {0, &IFile::Read, "Read"}, {1, &IFile::Write, "Write"}, {2, nullptr, "Flush"}, + {3, &IFile::SetSize, "SetSize"}, {4, &IFile::GetSize, "GetSize"}, + }; + RegisterHandlers(functions); + } + +private: + std::unique_ptr<FileSys::StorageBackend> backend; + + void Read(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const u64 unk = rp.Pop<u64>(); + const s64 offset = rp.Pop<s64>(); + const s64 length = rp.Pop<s64>(); + + LOG_DEBUG(Service_FS, "called, offset=0x%ld, length=0x%ld", offset, length); + + // Error checking + if (length < 0) { + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(ResultCode(ErrorModule::FS, ErrorDescription::InvalidLength)); + return; + } + if (offset < 0) { + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(ResultCode(ErrorModule::FS, ErrorDescription::InvalidOffset)); + return; + } + + // Read the data from the Storage backend + std::vector<u8> output(length); + ResultVal<size_t> res = backend->Read(offset, length, output.data()); + if (res.Failed()) { + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(res.Code()); + return; + } + + // Write the data to memory + ctx.WriteBuffer(output); + + IPC::ResponseBuilder rb{ctx, 4}; + rb.Push(RESULT_SUCCESS); + rb.Push(static_cast<u64>(*res)); + } + + void Write(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const u64 unk = rp.Pop<u64>(); + const s64 offset = rp.Pop<s64>(); + const s64 length = rp.Pop<s64>(); + + LOG_DEBUG(Service_FS, "called, offset=0x%ld, length=0x%ld", offset, length); + + // Error checking + if (length < 0) { + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(ResultCode(ErrorModule::FS, ErrorDescription::InvalidLength)); + return; + } + if (offset < 0) { + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(ResultCode(ErrorModule::FS, ErrorDescription::InvalidOffset)); + return; + } + + // Write the data to the Storage backend + std::vector<u8> data = ctx.ReadBuffer(); + ResultVal<size_t> res = backend->Write(offset, length, true, data.data()); + if (res.Failed()) { + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(res.Code()); + return; + } + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); + } + + void SetSize(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const u64 size = rp.Pop<u64>(); + backend->SetSize(size); + LOG_DEBUG(Service_FS, "called, size=%" PRIu64, size); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); + } + + void GetSize(Kernel::HLERequestContext& ctx) { + const u64 size = backend->GetSize(); + LOG_DEBUG(Service_FS, "called, size=%" PRIu64, size); + + IPC::ResponseBuilder rb{ctx, 4}; + rb.Push(RESULT_SUCCESS); + rb.Push<u64>(size); + } +}; + +class IDirectory final : public ServiceFramework<IDirectory> { +public: + explicit IDirectory(std::unique_ptr<FileSys::DirectoryBackend>&& backend) + : ServiceFramework("IDirectory"), backend(std::move(backend)) { + static const FunctionInfo functions[] = { + {0, &IDirectory::Read, "Read"}, + {1, &IDirectory::GetEntryCount, "GetEntryCount"}, + }; + RegisterHandlers(functions); + } + +private: + std::unique_ptr<FileSys::DirectoryBackend> backend; + + void Read(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const u64 unk = rp.Pop<u64>(); + + LOG_DEBUG(Service_FS, "called, unk=0x%llx", unk); + + // Calculate how many entries we can fit in the output buffer + u64 count_entries = ctx.GetWriteBufferSize() / sizeof(FileSys::Entry); + + // Read the data from the Directory backend + std::vector<FileSys::Entry> entries(count_entries); + u64 read_entries = backend->Read(count_entries, entries.data()); + + // Convert the data into a byte array + std::vector<u8> output(entries.size() * sizeof(FileSys::Entry)); + std::memcpy(output.data(), entries.data(), output.size()); + + // Write the data to memory + ctx.WriteBuffer(output); + + IPC::ResponseBuilder rb{ctx, 4}; + rb.Push(RESULT_SUCCESS); + rb.Push(read_entries); + } + + void GetEntryCount(Kernel::HLERequestContext& ctx) { + LOG_DEBUG(Service_FS, "called"); + + u64 count = backend->GetEntryCount(); + + IPC::ResponseBuilder rb{ctx, 4}; + rb.Push(RESULT_SUCCESS); + rb.Push(count); + } +}; + +class IFileSystem final : public ServiceFramework<IFileSystem> { +public: + explicit IFileSystem(std::unique_ptr<FileSys::FileSystemBackend>&& backend) + : ServiceFramework("IFileSystem"), backend(std::move(backend)) { + static const FunctionInfo functions[] = { + {0, &IFileSystem::CreateFile, "CreateFile"}, + {2, &IFileSystem::CreateDirectory, "CreateDirectory"}, + {7, &IFileSystem::GetEntryType, "GetEntryType"}, + {8, &IFileSystem::OpenFile, "OpenFile"}, + {9, &IFileSystem::OpenDirectory, "OpenDirectory"}, + {10, &IFileSystem::Commit, "Commit"}, + }; + RegisterHandlers(functions); + } + + void CreateFile(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + + auto file_buffer = ctx.ReadBuffer(); + auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0'); + + std::string name(file_buffer.begin(), end); + + u64 mode = rp.Pop<u64>(); + u32 size = rp.Pop<u32>(); + + LOG_DEBUG(Service_FS, "called file %s mode 0x%" PRIX64 " size 0x%08X", name.c_str(), mode, + size); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(backend->CreateFile(name, size)); + } + + void CreateDirectory(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + + auto file_buffer = ctx.ReadBuffer(); + auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0'); + + std::string name(file_buffer.begin(), end); + + LOG_DEBUG(Service_FS, "called directory %s", name.c_str()); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(backend->CreateDirectory(name)); + } + + void OpenFile(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + + auto file_buffer = ctx.ReadBuffer(); + auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0'); + + std::string name(file_buffer.begin(), end); + + auto mode = static_cast<FileSys::Mode>(rp.Pop<u32>()); + + LOG_DEBUG(Service_FS, "called file %s mode %u", name.c_str(), static_cast<u32>(mode)); + + auto result = backend->OpenFile(name, mode); + if (result.Failed()) { + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(result.Code()); + return; + } + + auto file = std::move(result.Unwrap()); + + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; + rb.Push(RESULT_SUCCESS); + rb.PushIpcInterface<IFile>(std::move(file)); + } + + void OpenDirectory(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + + auto file_buffer = ctx.ReadBuffer(); + auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0'); + + std::string name(file_buffer.begin(), end); + + // TODO(Subv): Implement this filter. + u32 filter_flags = rp.Pop<u32>(); + + LOG_DEBUG(Service_FS, "called directory %s filter %u", name.c_str(), filter_flags); + + auto result = backend->OpenDirectory(name); + if (result.Failed()) { + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(result.Code()); + return; + } + + auto directory = std::move(result.Unwrap()); + + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; + rb.Push(RESULT_SUCCESS); + rb.PushIpcInterface<IDirectory>(std::move(directory)); + } + + void GetEntryType(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + + auto file_buffer = ctx.ReadBuffer(); + auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0'); + + std::string name(file_buffer.begin(), end); + + LOG_DEBUG(Service_FS, "called file %s", name.c_str()); + + auto result = backend->GetEntryType(name); + if (result.Failed()) { + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(result.Code()); + return; + } + + IPC::ResponseBuilder rb{ctx, 3}; + rb.Push(RESULT_SUCCESS); + rb.Push<u32>(static_cast<u32>(*result)); + } + + void Commit(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_FS, "(STUBBED) called"); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); + } + +private: + std::unique_ptr<FileSys::FileSystemBackend> backend; +}; + FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") { static const FunctionInfo functions[] = { {1, &FSP_SRV::Initalize, "Initalize"}, {18, &FSP_SRV::MountSdCard, "MountSdCard"}, + {22, &FSP_SRV::CreateSaveData, "CreateSaveData"}, + {51, &FSP_SRV::MountSaveData, "MountSaveData"}, {200, &FSP_SRV::OpenDataStorageByCurrentProcess, "OpenDataStorageByCurrentProcess"}, {202, nullptr, "OpenDataStorageByDataId"}, {203, &FSP_SRV::OpenRomStorage, "OpenRomStorage"}, @@ -96,12 +387,40 @@ void FSP_SRV::Initalize(Kernel::HLERequestContext& ctx) { } void FSP_SRV::MountSdCard(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_FS, "(STUBBED) called"); + LOG_DEBUG(Service_FS, "called"); + + FileSys::Path unused; + auto filesystem = OpenFileSystem(Type::SDMC, unused).Unwrap(); + + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; + rb.Push(RESULT_SUCCESS); + rb.PushIpcInterface<IFileSystem>(std::move(filesystem)); +} + +void FSP_SRV::CreateSaveData(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + + auto save_struct = rp.PopRaw<std::array<u8, 0x40>>(); + auto save_create_struct = rp.PopRaw<std::array<u8, 0x40>>(); + u128 uid = rp.PopRaw<u128>(); + + LOG_WARNING(Service_FS, "(STUBBED) called uid = %016" PRIX64 "%016" PRIX64, uid[1], uid[0]); IPC::ResponseBuilder rb{ctx, 2}; rb.Push(RESULT_SUCCESS); } +void FSP_SRV::MountSaveData(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_FS, "(STUBBED) called"); + + FileSys::Path unused; + auto filesystem = OpenFileSystem(Type::SaveData, unused).Unwrap(); + + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; + rb.Push(RESULT_SUCCESS); + rb.PushIpcInterface<IFileSystem>(std::move(filesystem)); +} + void FSP_SRV::GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service_FS, "(STUBBED) called"); diff --git a/src/core/hle/service/filesystem/fsp_srv.h b/src/core/hle/service/filesystem/fsp_srv.h index 56afc4b90..e15ba4375 100644 --- a/src/core/hle/service/filesystem/fsp_srv.h +++ b/src/core/hle/service/filesystem/fsp_srv.h @@ -24,6 +24,8 @@ private: void Initalize(Kernel::HLERequestContext& ctx); void MountSdCard(Kernel::HLERequestContext& ctx); + void CreateSaveData(Kernel::HLERequestContext& ctx); + void MountSaveData(Kernel::HLERequestContext& ctx); void GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx); void OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx); void OpenRomStorage(Kernel::HLERequestContext& ctx); diff --git a/src/core/hle/service/hid/hid.cpp b/src/core/hle/service/hid/hid.cpp index dacd1862d..019a09444 100644 --- a/src/core/hle/service/hid/hid.cpp +++ b/src/core/hle/service/hid/hid.cpp @@ -45,6 +45,10 @@ public: CoreTiming::ScheduleEvent(pad_update_ticks, pad_update_event); } + ~IAppletResource() { + CoreTiming::UnscheduleEvent(pad_update_event, 0); + } + private: void GetSharedMemoryHandle(Kernel::HLERequestContext& ctx) { IPC::ResponseBuilder rb{ctx, 2, 1}; @@ -61,13 +65,14 @@ private: } void UpdatePadCallback(u64 userdata, int cycles_late) { - SharedMemory* mem = reinterpret_cast<SharedMemory*>(shared_mem->GetPointer()); + SharedMemory mem{}; + std::memcpy(&mem, shared_mem->GetPointer(), sizeof(SharedMemory)); if (is_device_reload_pending.exchange(false)) LoadInputDevices(); // Set up controllers as neon red+blue Joy-Con attached to console - ControllerHeader& controller_header = mem->controllers[Controller_Handheld].header; + ControllerHeader& controller_header = mem.controllers[Controller_Handheld].header; controller_header.type = ControllerType_Handheld | ControllerType_JoyconPair; controller_header.single_colors_descriptor = ColorDesc_ColorsNonexistent; controller_header.right_color_body = JOYCON_BODY_NEON_RED; @@ -75,8 +80,8 @@ private: controller_header.left_color_body = JOYCON_BODY_NEON_BLUE; controller_header.left_color_buttons = JOYCON_BUTTONS_NEON_BLUE; - for (int layoutIdx = 0; layoutIdx < HID_NUM_LAYOUTS; layoutIdx++) { - ControllerLayout& layout = mem->controllers[Controller_Handheld].layouts[layoutIdx]; + for (int index = 0; index < HID_NUM_LAYOUTS; index++) { + ControllerLayout& layout = mem.controllers[Controller_Handheld].layouts[index]; layout.header.num_entries = HID_NUM_ENTRIES; layout.header.max_entry_index = HID_NUM_ENTRIES - 1; @@ -132,10 +137,25 @@ private: // layouts) } - // TODO(shinyquagsire23): Update touch info + // TODO(bunnei): Properly implement the touch screen, the below will just write empty data + + TouchScreen& touchscreen = mem.touchscreen; + const u64 last_entry = touchscreen.header.latest_entry; + const u64 curr_entry = (last_entry + 1) % touchscreen.entries.size(); + const u64 timestamp = CoreTiming::GetTicks(); + const u64 sample_counter = touchscreen.entries[last_entry].header.timestamp + 1; + touchscreen.header.timestamp_ticks = timestamp; + touchscreen.header.num_entries = touchscreen.entries.size(); + touchscreen.header.latest_entry = curr_entry; + touchscreen.header.max_entry_index = touchscreen.entries.size(); + touchscreen.header.timestamp = timestamp; + touchscreen.entries[curr_entry].header.timestamp = sample_counter; + touchscreen.entries[curr_entry].header.num_touches = 0; // TODO(shinyquagsire23): Signal events + std::memcpy(shared_mem->GetPointer(), &mem, sizeof(SharedMemory)); + // Reschedule recurrent event CoreTiming::ScheduleEvent(pad_update_ticks - cycles_late, pad_update_event); } @@ -181,6 +201,7 @@ public: {66, &Hid::StartSixAxisSensor, "StartSixAxisSensor"}, {79, &Hid::SetGyroscopeZeroDriftMode, "SetGyroscopeZeroDriftMode"}, {100, &Hid::SetSupportedNpadStyleSet, "SetSupportedNpadStyleSet"}, + {101, &Hid::GetSupportedNpadStyleSet, "GetSupportedNpadStyleSet"}, {102, &Hid::SetSupportedNpadIdType, "SetSupportedNpadIdType"}, {103, &Hid::ActivateNpad, "ActivateNpad"}, {106, &Hid::AcquireNpadStyleSetUpdateEventHandle, @@ -189,7 +210,7 @@ public: {121, &Hid::GetNpadJoyHoldType, "GetNpadJoyHoldType"}, {122, &Hid::SetNpadJoyAssignmentModeSingleByDefault, "SetNpadJoyAssignmentModeSingleByDefault"}, - {124, nullptr, "SetNpadJoyAssignmentModeDual"}, + {124, &Hid::SetNpadJoyAssignmentModeDual, "SetNpadJoyAssignmentModeDual"}, {128, &Hid::SetNpadHandheldActivationMode, "SetNpadHandheldActivationMode"}, {200, &Hid::GetVibrationDeviceInfo, "GetVibrationDeviceInfo"}, {201, &Hid::SendVibrationValue, "SendVibrationValue"}, @@ -261,6 +282,13 @@ private: LOG_WARNING(Service_HID, "(STUBBED) called"); } + void GetSupportedNpadStyleSet(Kernel::HLERequestContext& ctx) { + IPC::ResponseBuilder rb{ctx, 3}; + rb.Push(RESULT_SUCCESS); + rb.Push<u32>(0); + LOG_WARNING(Service_HID, "(STUBBED) called"); + } + void SetSupportedNpadIdType(Kernel::HLERequestContext& ctx) { IPC::ResponseBuilder rb{ctx, 2}; rb.Push(RESULT_SUCCESS); @@ -311,6 +339,12 @@ private: LOG_WARNING(Service_HID, "(STUBBED) called"); } + void SetNpadJoyAssignmentModeDual(Kernel::HLERequestContext& ctx) { + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); + LOG_WARNING(Service_HID, "(STUBBED) called"); + } + void SetNpadHandheldActivationMode(Kernel::HLERequestContext& ctx) { IPC::ResponseBuilder rb{ctx, 2}; rb.Push(RESULT_SUCCESS); diff --git a/src/core/hle/service/nfp/nfp.cpp b/src/core/hle/service/nfp/nfp.cpp new file mode 100644 index 000000000..49870841c --- /dev/null +++ b/src/core/hle/service/nfp/nfp.cpp @@ -0,0 +1,28 @@ +// 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/hle/ipc_helpers.h" +#include "core/hle/service/nfp/nfp.h" +#include "core/hle/service/nfp/nfp_user.h" + +namespace Service { +namespace NFP { + +Module::Interface::Interface(std::shared_ptr<Module> module, const char* name) + : ServiceFramework(name), module(std::move(module)) {} + +void Module::Interface::Unknown(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_NFP, "(STUBBED) called"); + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + +void InstallInterfaces(SM::ServiceManager& service_manager) { + auto module = std::make_shared<Module>(); + std::make_shared<NFP_User>(module)->InstallAsService(service_manager); +} + +} // namespace NFP +} // namespace Service diff --git a/src/core/hle/service/nfp/nfp.h b/src/core/hle/service/nfp/nfp.h new file mode 100644 index 000000000..1163e9954 --- /dev/null +++ b/src/core/hle/service/nfp/nfp.h @@ -0,0 +1,28 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/service/service.h" + +namespace Service { +namespace NFP { + +class Module final { +public: + class Interface : public ServiceFramework<Interface> { + public: + Interface(std::shared_ptr<Module> module, const char* name); + + void Unknown(Kernel::HLERequestContext& ctx); + + protected: + std::shared_ptr<Module> module; + }; +}; + +void InstallInterfaces(SM::ServiceManager& service_manager); + +} // namespace NFP +} // namespace Service diff --git a/src/core/hle/service/nfp/nfp_user.cpp b/src/core/hle/service/nfp/nfp_user.cpp new file mode 100644 index 000000000..14e5647c4 --- /dev/null +++ b/src/core/hle/service/nfp/nfp_user.cpp @@ -0,0 +1,19 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "core/hle/service/nfp/nfp_user.h" + +namespace Service { +namespace NFP { + +NFP_User::NFP_User(std::shared_ptr<Module> module) + : Module::Interface(std::move(module), "nfp:user") { + static const FunctionInfo functions[] = { + {0, &NFP_User::Unknown, "Unknown"}, + }; + RegisterHandlers(functions); +} + +} // namespace NFP +} // namespace Service diff --git a/src/core/hle/service/nfp/nfp_user.h b/src/core/hle/service/nfp/nfp_user.h new file mode 100644 index 000000000..1606444ca --- /dev/null +++ b/src/core/hle/service/nfp/nfp_user.h @@ -0,0 +1,18 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/service/nfp/nfp.h" + +namespace Service { +namespace NFP { + +class NFP_User final : public Module::Interface { +public: + explicit NFP_User(std::shared_ptr<Module> module); +}; + +} // namespace NFP +} // namespace Service diff --git a/src/core/hle/service/nifm/nifm.cpp b/src/core/hle/service/nifm/nifm.cpp index e6f05eae5..dd2d5fe63 100644 --- a/src/core/hle/service/nifm/nifm.cpp +++ b/src/core/hle/service/nifm/nifm.cpp @@ -32,7 +32,7 @@ public: {0, &IRequest::GetRequestState, "GetRequestState"}, {1, &IRequest::GetResult, "GetResult"}, {2, &IRequest::GetSystemEventReadableHandles, "GetSystemEventReadableHandles"}, - {3, nullptr, "Cancel"}, + {3, &IRequest::Cancel, "Cancel"}, {4, nullptr, "Submit"}, {5, nullptr, "SetRequirement"}, {6, nullptr, "SetRequirementPreset"}, @@ -80,6 +80,11 @@ private: rb.Push(RESULT_SUCCESS); rb.PushCopyObjects(event1, event2); } + void Cancel(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_NIFM, "(STUBBED) called"); + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); + } Kernel::SharedPtr<Kernel::Event> event1, event2; }; @@ -96,13 +101,56 @@ public: } }; +class IGeneralService final : public ServiceFramework<IGeneralService> { +public: + IGeneralService(); + +private: + void GetClientId(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_NIFM, "(STUBBED) called"); + IPC::ResponseBuilder rb{ctx, 4}; + rb.Push(RESULT_SUCCESS); + rb.Push<u64>(0); + } + void CreateScanRequest(Kernel::HLERequestContext& ctx) { + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; + + rb.Push(RESULT_SUCCESS); + rb.PushIpcInterface<IScanRequest>(); + + LOG_DEBUG(Service_NIFM, "called"); + } + void CreateRequest(Kernel::HLERequestContext& ctx) { + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; + + rb.Push(RESULT_SUCCESS); + rb.PushIpcInterface<IRequest>(); + + LOG_DEBUG(Service_NIFM, "called"); + } + void RemoveNetworkProfile(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_NIFM, "(STUBBED) called"); + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); + } + void CreateTemporaryNetworkProfile(Kernel::HLERequestContext& ctx) { + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; + + rb.Push(RESULT_SUCCESS); + rb.PushIpcInterface<INetworkProfile>(); + + LOG_DEBUG(Service_NIFM, "called"); + } +}; + IGeneralService::IGeneralService() : ServiceFramework("IGeneralService") { static const FunctionInfo functions[] = { {1, &IGeneralService::GetClientId, "GetClientId"}, {2, &IGeneralService::CreateScanRequest, "CreateScanRequest"}, {4, &IGeneralService::CreateRequest, "CreateRequest"}, - {6, nullptr, "GetCurrentNetworkProfile"}, - {7, nullptr, "EnumerateNetworkInterfaces"}, + {5, nullptr, "GetCurrentNetworkProfile"}, + {6, nullptr, "EnumerateNetworkInterfaces"}, + {7, nullptr, "EnumerateNetworkProfiles"}, {8, nullptr, "GetNetworkProfile"}, {9, nullptr, "SetNetworkProfile"}, {10, &IGeneralService::RemoveNetworkProfile, "RemoveNetworkProfile"}, @@ -137,50 +185,28 @@ IGeneralService::IGeneralService() : ServiceFramework("IGeneralService") { RegisterHandlers(functions); } -void IGeneralService::GetClientId(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_NIFM, "(STUBBED) called"); - IPC::ResponseBuilder rb{ctx, 4}; - rb.Push(RESULT_SUCCESS); - rb.Push<u64>(0); -} - -void IGeneralService::CreateScanRequest(Kernel::HLERequestContext& ctx) { +void Module::Interface::CreateGeneralServiceOld(Kernel::HLERequestContext& ctx) { IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<IScanRequest>(); - + rb.PushIpcInterface<IGeneralService>(); LOG_DEBUG(Service_NIFM, "called"); } -void IGeneralService::CreateRequest(Kernel::HLERequestContext& ctx) { +void Module::Interface::CreateGeneralService(Kernel::HLERequestContext& ctx) { IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<IRequest>(); - + rb.PushIpcInterface<IGeneralService>(); LOG_DEBUG(Service_NIFM, "called"); } -void IGeneralService::RemoveNetworkProfile(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_NIFM, "(STUBBED) called"); - IPC::ResponseBuilder rb{ctx, 2}; - rb.Push(RESULT_SUCCESS); -} - -void IGeneralService::CreateTemporaryNetworkProfile(Kernel::HLERequestContext& ctx) { - IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<INetworkProfile>(); - - LOG_DEBUG(Service_NIFM, "called"); -} +Module::Interface::Interface(std::shared_ptr<Module> module, const char* name) + : ServiceFramework(name), module(std::move(module)) {} void InstallInterfaces(SM::ServiceManager& service_manager) { - std::make_shared<NIFM_A>()->InstallAsService(service_manager); - std::make_shared<NIFM_S>()->InstallAsService(service_manager); - std::make_shared<NIFM_U>()->InstallAsService(service_manager); + auto module = std::make_shared<Module>(); + std::make_shared<NIFM_A>(module)->InstallAsService(service_manager); + std::make_shared<NIFM_S>(module)->InstallAsService(service_manager); + std::make_shared<NIFM_U>(module)->InstallAsService(service_manager); } } // namespace NIFM diff --git a/src/core/hle/service/nifm/nifm.h b/src/core/hle/service/nifm/nifm.h index 6edbfe4a4..11d263b12 100644 --- a/src/core/hle/service/nifm/nifm.h +++ b/src/core/hle/service/nifm/nifm.h @@ -9,16 +9,18 @@ namespace Service { namespace NIFM { -class IGeneralService final : public ServiceFramework<IGeneralService> { +class Module final { public: - IGeneralService(); - -private: - void GetClientId(Kernel::HLERequestContext& ctx); - void CreateScanRequest(Kernel::HLERequestContext& ctx); - void CreateRequest(Kernel::HLERequestContext& ctx); - void RemoveNetworkProfile(Kernel::HLERequestContext& ctx); - void CreateTemporaryNetworkProfile(Kernel::HLERequestContext& ctx); + class Interface : public ServiceFramework<Interface> { + public: + Interface(std::shared_ptr<Module> module, const char* name); + + void CreateGeneralServiceOld(Kernel::HLERequestContext& ctx); + void CreateGeneralService(Kernel::HLERequestContext& ctx); + + protected: + std::shared_ptr<Module> module; + }; }; void InstallInterfaces(SM::ServiceManager& service_manager); diff --git a/src/core/hle/service/nifm/nifm_a.cpp b/src/core/hle/service/nifm/nifm_a.cpp index ee61d8ff4..f75df8c04 100644 --- a/src/core/hle/service/nifm/nifm_a.cpp +++ b/src/core/hle/service/nifm/nifm_a.cpp @@ -2,29 +2,12 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include "common/logging/log.h" -#include "core/hle/ipc_helpers.h" -#include "core/hle/service/nifm/nifm.h" #include "core/hle/service/nifm/nifm_a.h" namespace Service { namespace NIFM { -void NIFM_A::CreateGeneralServiceOld(Kernel::HLERequestContext& ctx) { - IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<IGeneralService>(); - LOG_DEBUG(Service_NIFM, "called"); -} - -void NIFM_A::CreateGeneralService(Kernel::HLERequestContext& ctx) { - IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<IGeneralService>(); - LOG_DEBUG(Service_NIFM, "called"); -} - -NIFM_A::NIFM_A() : ServiceFramework("nifm:a") { +NIFM_A::NIFM_A(std::shared_ptr<Module> module) : Module::Interface(std::move(module), "nifm:a") { static const FunctionInfo functions[] = { {4, &NIFM_A::CreateGeneralServiceOld, "CreateGeneralServiceOld"}, {5, &NIFM_A::CreateGeneralService, "CreateGeneralService"}, diff --git a/src/core/hle/service/nifm/nifm_a.h b/src/core/hle/service/nifm/nifm_a.h index 06a92a93c..eaea14e29 100644 --- a/src/core/hle/service/nifm/nifm_a.h +++ b/src/core/hle/service/nifm/nifm_a.h @@ -4,20 +4,14 @@ #pragma once -#include "core/hle/kernel/hle_ipc.h" -#include "core/hle/service/service.h" +#include "core/hle/service/nifm/nifm.h" namespace Service { namespace NIFM { -class NIFM_A final : public ServiceFramework<NIFM_A> { +class NIFM_A final : public Module::Interface { public: - NIFM_A(); - ~NIFM_A() = default; - -private: - void CreateGeneralServiceOld(Kernel::HLERequestContext& ctx); - void CreateGeneralService(Kernel::HLERequestContext& ctx); + explicit NIFM_A(std::shared_ptr<Module> module); }; } // namespace NIFM diff --git a/src/core/hle/service/nifm/nifm_s.cpp b/src/core/hle/service/nifm/nifm_s.cpp index c38b2a4c7..9c0b300e4 100644 --- a/src/core/hle/service/nifm/nifm_s.cpp +++ b/src/core/hle/service/nifm/nifm_s.cpp @@ -2,29 +2,12 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include "common/logging/log.h" -#include "core/hle/ipc_helpers.h" -#include "core/hle/service/nifm/nifm.h" #include "core/hle/service/nifm/nifm_s.h" namespace Service { namespace NIFM { -void NIFM_S::CreateGeneralServiceOld(Kernel::HLERequestContext& ctx) { - IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<IGeneralService>(); - LOG_DEBUG(Service_NIFM, "called"); -} - -void NIFM_S::CreateGeneralService(Kernel::HLERequestContext& ctx) { - IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<IGeneralService>(); - LOG_DEBUG(Service_NIFM, "called"); -} - -NIFM_S::NIFM_S() : ServiceFramework("nifm:s") { +NIFM_S::NIFM_S(std::shared_ptr<Module> module) : Module::Interface(std::move(module), "nifm:s") { static const FunctionInfo functions[] = { {4, &NIFM_S::CreateGeneralServiceOld, "CreateGeneralServiceOld"}, {5, &NIFM_S::CreateGeneralService, "CreateGeneralService"}, diff --git a/src/core/hle/service/nifm/nifm_s.h b/src/core/hle/service/nifm/nifm_s.h index d11a1ec29..f9e2d8039 100644 --- a/src/core/hle/service/nifm/nifm_s.h +++ b/src/core/hle/service/nifm/nifm_s.h @@ -4,20 +4,14 @@ #pragma once -#include "core/hle/kernel/hle_ipc.h" -#include "core/hle/service/service.h" +#include "core/hle/service/nifm/nifm.h" namespace Service { namespace NIFM { -class NIFM_S final : public ServiceFramework<NIFM_S> { +class NIFM_S final : public Module::Interface { public: - NIFM_S(); - ~NIFM_S() = default; - -private: - void CreateGeneralServiceOld(Kernel::HLERequestContext& ctx); - void CreateGeneralService(Kernel::HLERequestContext& ctx); + explicit NIFM_S(std::shared_ptr<Module> module); }; } // namespace NIFM diff --git a/src/core/hle/service/nifm/nifm_u.cpp b/src/core/hle/service/nifm/nifm_u.cpp index a5895c13c..44e6f483d 100644 --- a/src/core/hle/service/nifm/nifm_u.cpp +++ b/src/core/hle/service/nifm/nifm_u.cpp @@ -2,29 +2,12 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include "common/logging/log.h" -#include "core/hle/ipc_helpers.h" -#include "core/hle/service/nifm/nifm.h" #include "core/hle/service/nifm/nifm_u.h" namespace Service { namespace NIFM { -void NIFM_U::CreateGeneralServiceOld(Kernel::HLERequestContext& ctx) { - IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<IGeneralService>(); - LOG_DEBUG(Service_NIFM, "called"); -} - -void NIFM_U::CreateGeneralService(Kernel::HLERequestContext& ctx) { - IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<IGeneralService>(); - LOG_DEBUG(Service_NIFM, "called"); -} - -NIFM_U::NIFM_U() : ServiceFramework("nifm:u") { +NIFM_U::NIFM_U(std::shared_ptr<Module> module) : Module::Interface(std::move(module), "nifm:u") { static const FunctionInfo functions[] = { {4, &NIFM_U::CreateGeneralServiceOld, "CreateGeneralServiceOld"}, {5, &NIFM_U::CreateGeneralService, "CreateGeneralService"}, diff --git a/src/core/hle/service/nifm/nifm_u.h b/src/core/hle/service/nifm/nifm_u.h index da40b604f..912006775 100644 --- a/src/core/hle/service/nifm/nifm_u.h +++ b/src/core/hle/service/nifm/nifm_u.h @@ -4,20 +4,14 @@ #pragma once -#include "core/hle/kernel/hle_ipc.h" -#include "core/hle/service/service.h" +#include "core/hle/service/nifm/nifm.h" namespace Service { namespace NIFM { -class NIFM_U final : public ServiceFramework<NIFM_U> { +class NIFM_U final : public Module::Interface { public: - NIFM_U(); - ~NIFM_U() = default; - -private: - void CreateGeneralServiceOld(Kernel::HLERequestContext& ctx); - void CreateGeneralService(Kernel::HLERequestContext& ctx); + explicit NIFM_U(std::shared_ptr<Module> module); }; } // namespace NIFM diff --git a/src/core/hle/service/ns/pl_u.cpp b/src/core/hle/service/ns/pl_u.cpp index cc9d03a7c..ef3c7799a 100644 --- a/src/core/hle/service/ns/pl_u.cpp +++ b/src/core/hle/service/ns/pl_u.cpp @@ -4,6 +4,7 @@ #include "common/common_paths.h" #include "common/file_util.h" +#include "core/core.h" #include "core/hle/ipc_helpers.h" #include "core/hle/service/ns/pl_u.h" @@ -32,6 +33,7 @@ enum class LoadState : u32 { PL_U::PL_U() : ServiceFramework("pl:u") { static const FunctionInfo functions[] = { + {0, &PL_U::RequestLoad, "RequestLoad"}, {1, &PL_U::GetLoadState, "GetLoadState"}, {2, &PL_U::GetSize, "GetSize"}, {3, &PL_U::GetSharedMemoryAddressOffset, "GetSharedMemoryAddressOffset"}, @@ -53,6 +55,15 @@ PL_U::PL_U() : ServiceFramework("pl:u") { } } +void PL_U::RequestLoad(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const u32 shared_font_type{rp.Pop<u32>()}; + + LOG_DEBUG(Service_NS, "called, shared_font_type=%d", shared_font_type); + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + void PL_U::GetLoadState(Kernel::HLERequestContext& ctx) { IPC::RequestParser rp{ctx}; const u32 font_id{rp.Pop<u32>()}; @@ -90,13 +101,13 @@ void PL_U::GetSharedMemoryNativeHandle(Kernel::HLERequestContext& ctx) { // dump. In the future, we need to replace this with a more robust solution. // Map backing memory for the font data - Kernel::g_current_process->vm_manager.MapMemoryBlock(SHARED_FONT_MEM_VADDR, shared_font, 0, - SHARED_FONT_MEM_SIZE, - Kernel::MemoryState::Shared); + Core::CurrentProcess()->vm_manager.MapMemoryBlock(SHARED_FONT_MEM_VADDR, shared_font, 0, + SHARED_FONT_MEM_SIZE, + Kernel::MemoryState::Shared); // Create shared font memory object shared_font_mem = Kernel::SharedMemory::Create( - Kernel::g_current_process, SHARED_FONT_MEM_SIZE, Kernel::MemoryPermission::ReadWrite, + Core::CurrentProcess(), SHARED_FONT_MEM_SIZE, Kernel::MemoryPermission::ReadWrite, Kernel::MemoryPermission::Read, SHARED_FONT_MEM_VADDR, Kernel::MemoryRegion::BASE, "PL_U:shared_font_mem"); } diff --git a/src/core/hle/service/ns/pl_u.h b/src/core/hle/service/ns/pl_u.h index 7a4766338..360482d13 100644 --- a/src/core/hle/service/ns/pl_u.h +++ b/src/core/hle/service/ns/pl_u.h @@ -17,6 +17,7 @@ public: ~PL_U() = default; private: + void RequestLoad(Kernel::HLERequestContext& ctx); void GetLoadState(Kernel::HLERequestContext& ctx); void GetSize(Kernel::HLERequestContext& ctx); void GetSharedMemoryAddressOffset(Kernel::HLERequestContext& ctx); diff --git a/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp b/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp index 7674d332d..87b3a2d74 100644 --- a/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp +++ b/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp @@ -23,17 +23,16 @@ void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u3 u32 stride, NVFlinger::BufferQueue::BufferTransformFlags transform) { VAddr addr = nvmap_dev->GetObjectAddress(buffer_handle); LOG_WARNING(Service, - "Drawing from address %llx offset %08X Width %u Height %u Stride %u Format %u", - addr, offset, width, height, stride, format); + "Drawing from address %lx offset %08X Width %u Height %u Stride %u Format %u", addr, + offset, width, height, stride, format); - using PixelFormat = RendererBase::FramebufferInfo::PixelFormat; - using Flags = NVFlinger::BufferQueue::BufferTransformFlags; - const bool flip_vertical = static_cast<u32>(transform) & static_cast<u32>(Flags::FlipV); - const RendererBase::FramebufferInfo framebuffer_info{ - addr, offset, width, height, stride, static_cast<PixelFormat>(format), flip_vertical}; + using PixelFormat = Tegra::FramebufferConfig::PixelFormat; + const Tegra::FramebufferConfig framebuffer{ + addr, offset, width, height, stride, static_cast<PixelFormat>(format), transform}; Core::System::GetInstance().perf_stats.EndGameFrame(); - VideoCore::g_renderer->SwapBuffers(framebuffer_info); + + VideoCore::g_renderer->SwapBuffers(framebuffer); } } // namespace Devices diff --git a/src/core/hle/service/nvflinger/buffer_queue.cpp b/src/core/hle/service/nvflinger/buffer_queue.cpp index ff7b6b039..e4ff2e267 100644 --- a/src/core/hle/service/nvflinger/buffer_queue.cpp +++ b/src/core/hle/service/nvflinger/buffer_queue.cpp @@ -26,26 +26,30 @@ void BufferQueue::SetPreallocatedBuffer(u32 slot, IGBPBuffer& igbp_buffer) { LOG_WARNING(Service, "Adding graphics buffer %u", slot); queue.emplace_back(buffer); + + if (buffer_wait_event) { + buffer_wait_event->Signal(); + } } -u32 BufferQueue::DequeueBuffer(u32 pixel_format, u32 width, u32 height) { +boost::optional<u32> BufferQueue::DequeueBuffer(u32 width, u32 height) { auto itr = std::find_if(queue.begin(), queue.end(), [&](const Buffer& buffer) { // Only consider free buffers. Buffers become free once again after they've been Acquired // and Released by the compositor, see the NVFlinger::Compose method. - if (buffer.status != Buffer::Status::Free) + if (buffer.status != Buffer::Status::Free) { return false; + } // Make sure that the parameters match. - auto& igbp_buffer = buffer.igbp_buffer; - return igbp_buffer.format == pixel_format && igbp_buffer.width == width && - igbp_buffer.height == height; + return buffer.igbp_buffer.width == width && buffer.igbp_buffer.height == height; }); + if (itr == queue.end()) { - LOG_CRITICAL(Service_NVDRV, "no free buffers for pixel_format=%d, width=%d, height=%d", - pixel_format, width, height); - itr = queue.begin(); + return boost::none; } + buffer_wait_event = nullptr; + itr->status = Buffer::Status::Dequeued; return itr->slot; } @@ -83,6 +87,10 @@ void BufferQueue::ReleaseBuffer(u32 slot) { ASSERT(itr != queue.end()); ASSERT(itr->status == Buffer::Status::Acquired); itr->status = Buffer::Status::Free; + + if (buffer_wait_event) { + buffer_wait_event->Signal(); + } } u32 BufferQueue::Query(QueryType type) { @@ -98,5 +106,10 @@ u32 BufferQueue::Query(QueryType type) { return 0; } +void BufferQueue::SetBufferWaitEvent(Kernel::SharedPtr<Kernel::Event>&& wait_event) { + ASSERT_MSG(!buffer_wait_event, "buffer_wait_event only supports a single waiting thread!"); + buffer_wait_event = std::move(wait_event); +} + } // namespace NVFlinger } // namespace Service diff --git a/src/core/hle/service/nvflinger/buffer_queue.h b/src/core/hle/service/nvflinger/buffer_queue.h index ef9732769..1de5767cb 100644 --- a/src/core/hle/service/nvflinger/buffer_queue.h +++ b/src/core/hle/service/nvflinger/buffer_queue.h @@ -47,6 +47,8 @@ public: ~BufferQueue() = default; enum class BufferTransformFlags : u32 { + /// No transform flags are set + Unset = 0x00, /// Flip source image horizontally (around the vertical axis) FlipH = 0x01, /// Flip source image vertically (around the horizontal axis) @@ -69,12 +71,13 @@ public: }; void SetPreallocatedBuffer(u32 slot, IGBPBuffer& buffer); - u32 DequeueBuffer(u32 pixel_format, u32 width, u32 height); + boost::optional<u32> DequeueBuffer(u32 width, u32 height); const IGBPBuffer& RequestBuffer(u32 slot) const; void QueueBuffer(u32 slot, BufferTransformFlags transform); boost::optional<const Buffer&> AcquireBuffer(); void ReleaseBuffer(u32 slot); u32 Query(QueryType type); + void SetBufferWaitEvent(Kernel::SharedPtr<Kernel::Event>&& wait_event); u32 GetId() const { return id; @@ -90,6 +93,9 @@ private: std::vector<Buffer> queue; Kernel::SharedPtr<Kernel::Event> native_handle; + + /// Used to signal waiting thread when no buffers are available + Kernel::SharedPtr<Kernel::Event> buffer_wait_event; }; } // namespace NVFlinger diff --git a/src/core/hle/service/nvflinger/nvflinger.cpp b/src/core/hle/service/nvflinger/nvflinger.cpp index a54239b0f..0d30f54dc 100644 --- a/src/core/hle/service/nvflinger/nvflinger.cpp +++ b/src/core/hle/service/nvflinger/nvflinger.cpp @@ -150,6 +150,9 @@ void NVFlinger::Compose() { igbp_buffer.width, igbp_buffer.height, igbp_buffer.stride, buffer->transform); buffer_queue->ReleaseBuffer(buffer->slot); + + // TODO(Subv): Figure out when we should actually signal this event. + buffer_queue->GetNativeHandle()->Signal(); } } diff --git a/src/core/hle/service/service.cpp b/src/core/hle/service/service.cpp index 6a2d6a4ef..c5490c1ae 100644 --- a/src/core/hle/service/service.cpp +++ b/src/core/hle/service/service.cpp @@ -7,6 +7,7 @@ #include "common/assert.h" #include "common/logging/log.h" #include "common/string_util.h" +#include "core/core.h" #include "core/hle/ipc.h" #include "core/hle/ipc_helpers.h" #include "core/hle/kernel/client_port.h" @@ -19,19 +20,23 @@ #include "core/hle/service/aoc/aoc_u.h" #include "core/hle/service/apm/apm.h" #include "core/hle/service/audio/audio.h" +#include "core/hle/service/fatal/fatal.h" #include "core/hle/service/filesystem/filesystem.h" #include "core/hle/service/friend/friend.h" #include "core/hle/service/hid/hid.h" #include "core/hle/service/lm/lm.h" +#include "core/hle/service/nfp/nfp.h" #include "core/hle/service/nifm/nifm.h" #include "core/hle/service/ns/ns.h" #include "core/hle/service/nvdrv/nvdrv.h" #include "core/hle/service/pctl/pctl.h" #include "core/hle/service/service.h" -#include "core/hle/service/set/set.h" +#include "core/hle/service/set/settings.h" #include "core/hle/service/sm/controller.h" #include "core/hle/service/sm/sm.h" #include "core/hle/service/sockets/sockets.h" +#include "core/hle/service/spl/module.h" +#include "core/hle/service/ssl/ssl.h" #include "core/hle/service/time/time.h" #include "core/hle/service/vi/vi.h" @@ -107,15 +112,15 @@ void ServiceFrameworkBase::ReportUnimplementedFunction(Kernel::HLERequestContext auto cmd_buf = ctx.CommandBuffer(); std::string function_name = info == nullptr ? fmt::format("{}", ctx.GetCommand()) : info->name; - fmt::MemoryWriter w; - w.write("function '{}': port='{}' cmd_buf={{[0]={:#x}", function_name, service_name, - cmd_buf[0]); + fmt::memory_buffer buf; + fmt::format_to(buf, "function '{}': port='{}' cmd_buf={{[0]={:#x}", function_name, service_name, + cmd_buf[0]); for (int i = 1; i <= 8; ++i) { - w.write(", [{}]={:#x}", i, cmd_buf[i]); + fmt::format_to(buf, ", [{}]={:#x}", i, cmd_buf[i]); } - w << '}'; + buf.push_back('}'); - LOG_ERROR(Service, "unknown / unimplemented %s", w.c_str()); + LOG_ERROR(Service, "unknown / unimplemented %s", fmt::to_string(buf).c_str()); UNIMPLEMENTED(); } @@ -148,12 +153,10 @@ ResultCode ServiceFrameworkBase::HandleSyncRequest(Kernel::HLERequestContext& co break; } default: - UNIMPLEMENTED_MSG("command_type=%d", context.GetCommandType()); + UNIMPLEMENTED_MSG("command_type=%d", static_cast<int>(context.GetCommandType())); } - u32* cmd_buf = (u32*)Memory::GetPointer(Kernel::GetCurrentThread()->GetTLSAddress()); - context.WriteToOutgoingCommandBuffer(cmd_buf, *Kernel::g_current_process, - Kernel::g_handle_table); + context.WriteToOutgoingCommandBuffer(*Kernel::GetCurrentThread()); return RESULT_SUCCESS; } @@ -180,15 +183,19 @@ void Init() { AOC::InstallInterfaces(*SM::g_service_manager); APM::InstallInterfaces(*SM::g_service_manager); Audio::InstallInterfaces(*SM::g_service_manager); + Fatal::InstallInterfaces(*SM::g_service_manager); FileSystem::InstallInterfaces(*SM::g_service_manager); Friend::InstallInterfaces(*SM::g_service_manager); HID::InstallInterfaces(*SM::g_service_manager); LM::InstallInterfaces(*SM::g_service_manager); + NFP::InstallInterfaces(*SM::g_service_manager); NIFM::InstallInterfaces(*SM::g_service_manager); NS::InstallInterfaces(*SM::g_service_manager); Nvidia::InstallInterfaces(*SM::g_service_manager); PCTL::InstallInterfaces(*SM::g_service_manager); Sockets::InstallInterfaces(*SM::g_service_manager); + SPL::InstallInterfaces(*SM::g_service_manager); + SSL::InstallInterfaces(*SM::g_service_manager); Time::InstallInterfaces(*SM::g_service_manager); VI::InstallInterfaces(*SM::g_service_manager, nv_flinger); Set::InstallInterfaces(*SM::g_service_manager); diff --git a/src/core/hle/service/set/set.cpp b/src/core/hle/service/set/set.cpp index 3001ee411..aa7c924e7 100644 --- a/src/core/hle/service/set/set.cpp +++ b/src/core/hle/service/set/set.cpp @@ -26,16 +26,19 @@ void SET::GetAvailableLanguageCodes(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service_SET, "(STUBBED) called"); } -SET::SET(const char* name) : ServiceFramework(name) { +SET::SET() : ServiceFramework("set") { static const FunctionInfo functions[] = { + {0, nullptr, "GetLanguageCode"}, {1, &SET::GetAvailableLanguageCodes, "GetAvailableLanguageCodes"}, + {2, nullptr, "MakeLanguageCode"}, + {3, nullptr, "GetAvailableLanguageCodeCount"}, + {4, nullptr, "GetRegionCode"}, + {5, nullptr, "GetAvailableLanguageCodes2"}, + {6, nullptr, "GetAvailableLanguageCodeCount2"}, + {7, nullptr, "GetKeyCodeMap"}, }; RegisterHandlers(functions); } -void InstallInterfaces(SM::ServiceManager& service_manager) { - std::make_shared<SET>("set")->InstallAsService(service_manager); -} - } // namespace Set } // namespace Service diff --git a/src/core/hle/service/set/set.h b/src/core/hle/service/set/set.h index 61e957946..7b7814ed1 100644 --- a/src/core/hle/service/set/set.h +++ b/src/core/hle/service/set/set.h @@ -11,15 +11,12 @@ namespace Set { class SET final : public ServiceFramework<SET> { public: - explicit SET(const char* name); + explicit SET(); ~SET() = default; private: void GetAvailableLanguageCodes(Kernel::HLERequestContext& ctx); }; -/// Registers all Set services with the specified service manager. -void InstallInterfaces(SM::ServiceManager& service_manager); - } // namespace Set } // namespace Service diff --git a/src/core/hle/service/set/set_cal.cpp b/src/core/hle/service/set/set_cal.cpp new file mode 100644 index 000000000..6231acd96 --- /dev/null +++ b/src/core/hle/service/set/set_cal.cpp @@ -0,0 +1,40 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "core/hle/service/set/set_cal.h" + +namespace Service { +namespace Set { + +SET_CAL::SET_CAL() : ServiceFramework("set:cal") { + static const FunctionInfo functions[] = { + {0, nullptr, "GetBluetoothBdAddress"}, + {1, nullptr, "GetConfigurationId1"}, + {2, nullptr, "GetAccelerometerOffset"}, + {3, nullptr, "GetAccelerometerScale"}, + {4, nullptr, "GetGyroscopeOffset"}, + {5, nullptr, "GetGyroscopeScale"}, + {6, nullptr, "GetWirelessLanMacAddress"}, + {7, nullptr, "GetWirelessLanCountryCodeCount"}, + {8, nullptr, "GetWirelessLanCountryCodes"}, + {9, nullptr, "GetSerialNumber"}, + {10, nullptr, "SetInitialSystemAppletProgramId"}, + {11, nullptr, "SetOverlayDispProgramId"}, + {12, nullptr, "GetBatteryLot"}, + {14, nullptr, "GetEciDeviceCertificate"}, + {15, nullptr, "GetEticketDeviceCertificate"}, + {16, nullptr, "GetSslKey"}, + {17, nullptr, "GetSslCertificate"}, + {18, nullptr, "GetGameCardKey"}, + {19, nullptr, "GetGameCardCertificate"}, + {20, nullptr, "GetEciDeviceKey"}, + {21, nullptr, "GetEticketDeviceKey"}, + {22, nullptr, "GetSpeakerParameter"}, + {23, nullptr, "GetLcdVendorId"}, + }; + RegisterHandlers(functions); +} + +} // namespace Set +} // namespace Service diff --git a/src/core/hle/service/set/set_cal.h b/src/core/hle/service/set/set_cal.h new file mode 100644 index 000000000..9c0b851d0 --- /dev/null +++ b/src/core/hle/service/set/set_cal.h @@ -0,0 +1,19 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/service/service.h" + +namespace Service { +namespace Set { + +class SET_CAL final : public ServiceFramework<SET_CAL> { +public: + explicit SET_CAL(); + ~SET_CAL() = default; +}; + +} // namespace Set +} // namespace Service diff --git a/src/core/hle/service/set/set_fd.cpp b/src/core/hle/service/set/set_fd.cpp new file mode 100644 index 000000000..8320d4250 --- /dev/null +++ b/src/core/hle/service/set/set_fd.cpp @@ -0,0 +1,25 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "core/hle/service/set/set_fd.h" + +namespace Service { +namespace Set { + +SET_FD::SET_FD() : ServiceFramework("set:fd") { + static const FunctionInfo functions[] = { + {2, nullptr, "SetSettingsItemValue"}, + {3, nullptr, "ResetSettingsItemValue"}, + {4, nullptr, "CreateSettingsItemKeyIterator"}, + {10, nullptr, "ReadSettings"}, + {11, nullptr, "ResetSettings"}, + {20, nullptr, "SetWebInspectorFlag"}, + {21, nullptr, "SetAllowedSslHosts"}, + {22, nullptr, "SetHostFsMountPoint"}, + }; + RegisterHandlers(functions); +} + +} // namespace Set +} // namespace Service diff --git a/src/core/hle/service/set/set_fd.h b/src/core/hle/service/set/set_fd.h new file mode 100644 index 000000000..65b36bcb3 --- /dev/null +++ b/src/core/hle/service/set/set_fd.h @@ -0,0 +1,19 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/service/service.h" + +namespace Service { +namespace Set { + +class SET_FD final : public ServiceFramework<SET_FD> { +public: + explicit SET_FD(); + ~SET_FD() = default; +}; + +} // namespace Set +} // namespace Service diff --git a/src/core/hle/service/set/set_sys.cpp b/src/core/hle/service/set/set_sys.cpp new file mode 100644 index 000000000..363abd10a --- /dev/null +++ b/src/core/hle/service/set/set_sys.cpp @@ -0,0 +1,167 @@ +// 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/hle/ipc_helpers.h" +#include "core/hle/kernel/client_port.h" +#include "core/hle/service/set/set_sys.h" + +namespace Service { +namespace Set { + +void SET_SYS::GetColorSetId(Kernel::HLERequestContext& ctx) { + + IPC::ResponseBuilder rb{ctx, 3}; + + rb.Push(RESULT_SUCCESS); + rb.Push<u32>(0); + + LOG_WARNING(Service_SET, "(STUBBED) called"); +} + +SET_SYS::SET_SYS() : ServiceFramework("set:sys") { + static const FunctionInfo functions[] = { + {0, nullptr, "SetLanguageCode"}, + {1, nullptr, "SetNetworkSettings"}, + {2, nullptr, "GetNetworkSettings"}, + {3, nullptr, "GetFirmwareVersion"}, + {4, nullptr, "GetFirmwareVersion2"}, + {7, nullptr, "GetLockScreenFlag"}, + {8, nullptr, "SetLockScreenFlag"}, + {9, nullptr, "GetBacklightSettings"}, + {10, nullptr, "SetBacklightSettings"}, + {11, nullptr, "SetBluetoothDevicesSettings"}, + {12, nullptr, "GetBluetoothDevicesSettings"}, + {13, nullptr, "GetExternalSteadyClockSourceId"}, + {14, nullptr, "SetExternalSteadyClockSourceId"}, + {15, nullptr, "GetUserSystemClockContext"}, + {16, nullptr, "SetUserSystemClockContext"}, + {17, nullptr, "GetAccountSettings"}, + {18, nullptr, "SetAccountSettings"}, + {19, nullptr, "GetAudioVolume"}, + {20, nullptr, "SetAudioVolume"}, + {21, nullptr, "GetEulaVersions"}, + {22, nullptr, "SetEulaVersions"}, + {23, &SET_SYS::GetColorSetId, "GetColorSetId"}, + {24, nullptr, "SetColorSetId"}, + {25, nullptr, "GetConsoleInformationUploadFlag"}, + {26, nullptr, "SetConsoleInformationUploadFlag"}, + {27, nullptr, "GetAutomaticApplicationDownloadFlag"}, + {28, nullptr, "SetAutomaticApplicationDownloadFlag"}, + {29, nullptr, "GetNotificationSettings"}, + {30, nullptr, "SetNotificationSettings"}, + {31, nullptr, "GetAccountNotificationSettings"}, + {32, nullptr, "SetAccountNotificationSettings"}, + {35, nullptr, "GetVibrationMasterVolume"}, + {36, nullptr, "SetVibrationMasterVolume"}, + {37, nullptr, "GetSettingsItemValueSize"}, + {38, nullptr, "GetSettingsItemValue"}, + {39, nullptr, "GetTvSettings"}, + {40, nullptr, "SetTvSettings"}, + {41, nullptr, "GetEdid"}, + {42, nullptr, "SetEdid"}, + {43, nullptr, "GetAudioOutputMode"}, + {44, nullptr, "SetAudioOutputMode"}, + {45, nullptr, "IsForceMuteOnHeadphoneRemoved"}, + {46, nullptr, "SetForceMuteOnHeadphoneRemoved"}, + {47, nullptr, "GetQuestFlag"}, + {48, nullptr, "SetQuestFlag"}, + {49, nullptr, "GetDataDeletionSettings"}, + {50, nullptr, "SetDataDeletionSettings"}, + {51, nullptr, "GetInitialSystemAppletProgramId"}, + {52, nullptr, "GetOverlayDispProgramId"}, + {53, nullptr, "GetDeviceTimeZoneLocationName"}, + {54, nullptr, "SetDeviceTimeZoneLocationName"}, + {55, nullptr, "GetWirelessCertificationFileSize"}, + {56, nullptr, "GetWirelessCertificationFile"}, + {57, nullptr, "SetRegionCode"}, + {58, nullptr, "GetNetworkSystemClockContext"}, + {59, nullptr, "SetNetworkSystemClockContext"}, + {60, nullptr, "IsUserSystemClockAutomaticCorrectionEnabled"}, + {61, nullptr, "SetUserSystemClockAutomaticCorrectionEnabled"}, + {62, nullptr, "GetDebugModeFlag"}, + {63, nullptr, "GetPrimaryAlbumStorage"}, + {64, nullptr, "SetPrimaryAlbumStorage"}, + {65, nullptr, "GetUsb30EnableFlag"}, + {66, nullptr, "SetUsb30EnableFlag"}, + {67, nullptr, "GetBatteryLot"}, + {68, nullptr, "GetSerialNumber"}, + {69, nullptr, "GetNfcEnableFlag"}, + {70, nullptr, "SetNfcEnableFlag"}, + {71, nullptr, "GetSleepSettings"}, + {72, nullptr, "SetSleepSettings"}, + {73, nullptr, "GetWirelessLanEnableFlag"}, + {74, nullptr, "SetWirelessLanEnableFlag"}, + {75, nullptr, "GetInitialLaunchSettings"}, + {76, nullptr, "SetInitialLaunchSettings"}, + {77, nullptr, "GetDeviceNickName"}, + {78, nullptr, "SetDeviceNickName"}, + {79, nullptr, "GetProductModel"}, + {80, nullptr, "GetLdnChannel"}, + {81, nullptr, "SetLdnChannel"}, + {82, nullptr, "AcquireTelemetryDirtyFlagEventHandle"}, + {83, nullptr, "GetTelemetryDirtyFlags"}, + {84, nullptr, "GetPtmBatteryLot"}, + {85, nullptr, "SetPtmBatteryLot"}, + {86, nullptr, "GetPtmFuelGaugeParameter"}, + {87, nullptr, "SetPtmFuelGaugeParameter"}, + {88, nullptr, "GetBluetoothEnableFlag"}, + {89, nullptr, "SetBluetoothEnableFlag"}, + {90, nullptr, "GetMiiAuthorId"}, + {91, nullptr, "SetShutdownRtcValue"}, + {92, nullptr, "GetShutdownRtcValue"}, + {93, nullptr, "AcquireFatalDirtyFlagEventHandle"}, + {94, nullptr, "GetFatalDirtyFlags"}, + {95, nullptr, "GetAutoUpdateEnableFlag"}, + {96, nullptr, "SetAutoUpdateEnableFlag"}, + {97, nullptr, "GetNxControllerSettings"}, + {98, nullptr, "SetNxControllerSettings"}, + {99, nullptr, "GetBatteryPercentageFlag"}, + {100, nullptr, "SetBatteryPercentageFlag"}, + {101, nullptr, "GetExternalRtcResetFlag"}, + {102, nullptr, "SetExternalRtcResetFlag"}, + {103, nullptr, "GetUsbFullKeyEnableFlag"}, + {104, nullptr, "SetUsbFullKeyEnableFlag"}, + {105, nullptr, "SetExternalSteadyClockInternalOffset"}, + {106, nullptr, "GetExternalSteadyClockInternalOffset"}, + {107, nullptr, "GetBacklightSettingsEx"}, + {108, nullptr, "SetBacklightSettingsEx"}, + {109, nullptr, "GetHeadphoneVolumeWarningCount"}, + {110, nullptr, "SetHeadphoneVolumeWarningCount"}, + {111, nullptr, "GetBluetoothAfhEnableFlag"}, + {112, nullptr, "SetBluetoothAfhEnableFlag"}, + {113, nullptr, "GetBluetoothBoostEnableFlag"}, + {114, nullptr, "SetBluetoothBoostEnableFlag"}, + {115, nullptr, "GetInRepairProcessEnableFlag"}, + {116, nullptr, "SetInRepairProcessEnableFlag"}, + {117, nullptr, "GetHeadphoneVolumeUpdateFlag"}, + {118, nullptr, "SetHeadphoneVolumeUpdateFlag"}, + {119, nullptr, "NeedsToUpdateHeadphoneVolume"}, + {120, nullptr, "GetPushNotificationActivityModeOnSleep"}, + {121, nullptr, "SetPushNotificationActivityModeOnSleep"}, + {122, nullptr, "GetServiceDiscoveryControlSettings"}, + {123, nullptr, "SetServiceDiscoveryControlSettings"}, + {124, nullptr, "GetErrorReportSharePermission"}, + {125, nullptr, "SetErrorReportSharePermission"}, + {126, nullptr, "GetAppletLaunchFlags"}, + {127, nullptr, "SetAppletLaunchFlags"}, + {128, nullptr, "GetConsoleSixAxisSensorAccelerationBias"}, + {129, nullptr, "SetConsoleSixAxisSensorAccelerationBias"}, + {130, nullptr, "GetConsoleSixAxisSensorAngularVelocityBias"}, + {131, nullptr, "SetConsoleSixAxisSensorAngularVelocityBias"}, + {132, nullptr, "GetConsoleSixAxisSensorAccelerationGain"}, + {133, nullptr, "SetConsoleSixAxisSensorAccelerationGain"}, + {134, nullptr, "GetConsoleSixAxisSensorAngularVelocityGain"}, + {135, nullptr, "SetConsoleSixAxisSensorAngularVelocityGain"}, + {136, nullptr, "GetKeyboardLayout"}, + {137, nullptr, "SetKeyboardLayout"}, + {138, nullptr, "GetWebInspectorFlag"}, + {139, nullptr, "GetAllowedSslHosts"}, + {140, nullptr, "GetHostFsMountPoint"}, + }; + RegisterHandlers(functions); +} + +} // namespace Set +} // namespace Service diff --git a/src/core/hle/service/set/set_sys.h b/src/core/hle/service/set/set_sys.h new file mode 100644 index 000000000..105f1a3c7 --- /dev/null +++ b/src/core/hle/service/set/set_sys.h @@ -0,0 +1,22 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/service/service.h" + +namespace Service { +namespace Set { + +class SET_SYS final : public ServiceFramework<SET_SYS> { +public: + explicit SET_SYS(); + ~SET_SYS() = default; + +private: + void GetColorSetId(Kernel::HLERequestContext& ctx); +}; + +} // namespace Set +} // namespace Service diff --git a/src/core/hle/service/set/settings.cpp b/src/core/hle/service/set/settings.cpp new file mode 100644 index 000000000..c6bc9e240 --- /dev/null +++ b/src/core/hle/service/set/settings.cpp @@ -0,0 +1,22 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "core/hle/service/set/set.h" +#include "core/hle/service/set/set_cal.h" +#include "core/hle/service/set/set_fd.h" +#include "core/hle/service/set/set_sys.h" +#include "core/hle/service/set/settings.h" + +namespace Service { +namespace Set { + +void InstallInterfaces(SM::ServiceManager& service_manager) { + std::make_shared<SET>()->InstallAsService(service_manager); + std::make_shared<SET_CAL>()->InstallAsService(service_manager); + std::make_shared<SET_FD>()->InstallAsService(service_manager); + std::make_shared<SET_SYS>()->InstallAsService(service_manager); +} + +} // namespace Set +} // namespace Service diff --git a/src/core/hle/service/set/settings.h b/src/core/hle/service/set/settings.h new file mode 100644 index 000000000..6c8d5a58c --- /dev/null +++ b/src/core/hle/service/set/settings.h @@ -0,0 +1,16 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/service/service.h" + +namespace Service { +namespace Set { + +/// Registers all Settings services with the specified service manager. +void InstallInterfaces(SM::ServiceManager& service_manager); + +} // namespace Set +} // namespace Service diff --git a/src/core/hle/service/sockets/bsd_u.cpp b/src/core/hle/service/sockets/bsd.cpp index 2ca1000ca..790ff82b3 100644 --- a/src/core/hle/service/sockets/bsd_u.cpp +++ b/src/core/hle/service/sockets/bsd.cpp @@ -3,12 +3,12 @@ // Refer to the license.txt file included. #include "core/hle/ipc_helpers.h" -#include "core/hle/service/sockets/bsd_u.h" +#include "core/hle/service/sockets/bsd.h" namespace Service { namespace Sockets { -void BSD_U::RegisterClient(Kernel::HLERequestContext& ctx) { +void BSD::RegisterClient(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service, "(STUBBED) called"); IPC::ResponseBuilder rb{ctx, 3}; @@ -17,7 +17,7 @@ void BSD_U::RegisterClient(Kernel::HLERequestContext& ctx) { rb.Push<u32>(0); // bsd errno } -void BSD_U::StartMonitoring(Kernel::HLERequestContext& ctx) { +void BSD::StartMonitoring(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service, "(STUBBED) called"); IPC::ResponseBuilder rb{ctx, 3}; @@ -26,7 +26,7 @@ void BSD_U::StartMonitoring(Kernel::HLERequestContext& ctx) { rb.Push<u32>(0); // bsd errno } -void BSD_U::Socket(Kernel::HLERequestContext& ctx) { +void BSD::Socket(Kernel::HLERequestContext& ctx) { IPC::RequestParser rp{ctx}; u32 domain = rp.Pop<u32>(); @@ -44,7 +44,7 @@ void BSD_U::Socket(Kernel::HLERequestContext& ctx) { rb.Push<u32>(0); // bsd errno } -void BSD_U::Connect(Kernel::HLERequestContext& ctx) { +void BSD::Connect(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service, "(STUBBED) called"); IPC::ResponseBuilder rb{ctx, 4}; @@ -54,7 +54,7 @@ void BSD_U::Connect(Kernel::HLERequestContext& ctx) { rb.Push<u32>(0); // bsd errno } -void BSD_U::SendTo(Kernel::HLERequestContext& ctx) { +void BSD::SendTo(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service, "(STUBBED) called"); IPC::ResponseBuilder rb{ctx, 4}; @@ -64,7 +64,7 @@ void BSD_U::SendTo(Kernel::HLERequestContext& ctx) { rb.Push<u32>(0); // bsd errno } -void BSD_U::Close(Kernel::HLERequestContext& ctx) { +void BSD::Close(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service, "(STUBBED) called"); IPC::ResponseBuilder rb{ctx, 4}; @@ -74,13 +74,15 @@ void BSD_U::Close(Kernel::HLERequestContext& ctx) { rb.Push<u32>(0); // bsd errno } -BSD_U::BSD_U() : ServiceFramework("bsd:u") { - static const FunctionInfo functions[] = {{0, &BSD_U::RegisterClient, "RegisterClient"}, - {1, &BSD_U::StartMonitoring, "StartMonitoring"}, - {2, &BSD_U::Socket, "Socket"}, - {11, &BSD_U::SendTo, "SendTo"}, - {14, &BSD_U::Connect, "Connect"}, - {26, &BSD_U::Close, "Close"}}; +BSD::BSD(const char* name) : ServiceFramework(name) { + static const FunctionInfo functions[] = { + {0, &BSD::RegisterClient, "RegisterClient"}, + {1, &BSD::StartMonitoring, "StartMonitoring"}, + {2, &BSD::Socket, "Socket"}, + {11, &BSD::SendTo, "SendTo"}, + {14, &BSD::Connect, "Connect"}, + {26, &BSD::Close, "Close"}, + }; RegisterHandlers(functions); } diff --git a/src/core/hle/service/sockets/bsd_u.h b/src/core/hle/service/sockets/bsd.h index 4e1252e9d..32d949e95 100644 --- a/src/core/hle/service/sockets/bsd_u.h +++ b/src/core/hle/service/sockets/bsd.h @@ -10,10 +10,10 @@ namespace Service { namespace Sockets { -class BSD_U final : public ServiceFramework<BSD_U> { +class BSD final : public ServiceFramework<BSD> { public: - BSD_U(); - ~BSD_U() = default; + explicit BSD(const char* name); + ~BSD() = default; private: void RegisterClient(Kernel::HLERequestContext& ctx); diff --git a/src/core/hle/service/sockets/nsd.cpp b/src/core/hle/service/sockets/nsd.cpp new file mode 100644 index 000000000..e3542d325 --- /dev/null +++ b/src/core/hle/service/sockets/nsd.cpp @@ -0,0 +1,34 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "core/hle/service/sockets/nsd.h" + +namespace Service { +namespace Sockets { + +NSD::NSD(const char* name) : ServiceFramework(name) { + static const FunctionInfo functions[] = { + {10, nullptr, "GetSettingName"}, + {11, nullptr, "GetEnvironmentIdentifier"}, + {12, nullptr, "GetDeviceId"}, + {13, nullptr, "DeleteSettings"}, + {14, nullptr, "ImportSettings"}, + {20, nullptr, "Resolve"}, + {21, nullptr, "ResolveEx"}, + {30, nullptr, "GetNasServiceSetting"}, + {31, nullptr, "GetNasServiceSettingEx"}, + {40, nullptr, "GetNasRequestFqdn"}, + {41, nullptr, "GetNasRequestFqdnEx"}, + {42, nullptr, "GetNasApiFqdn"}, + {43, nullptr, "GetNasApiFqdnEx"}, + {50, nullptr, "GetCurrentSetting"}, + {60, nullptr, "ReadSaveDataFromFsForTest"}, + {61, nullptr, "WriteSaveDataToFsForTest"}, + {62, nullptr, "DeleteSaveDataOfFsForTest"}, + }; + RegisterHandlers(functions); +} + +} // namespace Sockets +} // namespace Service diff --git a/src/core/hle/service/sockets/nsd.h b/src/core/hle/service/sockets/nsd.h new file mode 100644 index 000000000..a7c15a860 --- /dev/null +++ b/src/core/hle/service/sockets/nsd.h @@ -0,0 +1,20 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/kernel/hle_ipc.h" +#include "core/hle/service/service.h" + +namespace Service { +namespace Sockets { + +class NSD final : public ServiceFramework<NSD> { +public: + explicit NSD(const char* name); + ~NSD() = default; +}; + +} // namespace Sockets +} // namespace Service diff --git a/src/core/hle/service/sockets/sfdnsres.cpp b/src/core/hle/service/sockets/sfdnsres.cpp index 4d7bc7c3e..eb4b5fa57 100644 --- a/src/core/hle/service/sockets/sfdnsres.cpp +++ b/src/core/hle/service/sockets/sfdnsres.cpp @@ -19,16 +19,18 @@ void SFDNSRES::GetAddrInfo(Kernel::HLERequestContext& ctx) { } SFDNSRES::SFDNSRES() : ServiceFramework("sfdnsres") { - static const FunctionInfo functions[] = {{0, nullptr, "SetDnsAddressesPrivate"}, - {1, nullptr, "GetDnsAddressPrivate"}, - {2, nullptr, "GetHostByName"}, - {3, nullptr, "GetHostByAddr"}, - {4, nullptr, "GetHostStringError"}, - {5, nullptr, "GetGaiStringError"}, - {6, &SFDNSRES::GetAddrInfo, "GetAddrInfo"}, - {7, nullptr, "GetNameInfo"}, - {8, nullptr, "RequestCancelHandle"}, - {9, nullptr, "CancelSocketCall"}}; + static const FunctionInfo functions[] = { + {0, nullptr, "SetDnsAddressesPrivate"}, + {1, nullptr, "GetDnsAddressPrivate"}, + {2, nullptr, "GetHostByName"}, + {3, nullptr, "GetHostByAddr"}, + {4, nullptr, "GetHostStringError"}, + {5, nullptr, "GetGaiStringError"}, + {6, &SFDNSRES::GetAddrInfo, "GetAddrInfo"}, + {7, nullptr, "GetNameInfo"}, + {8, nullptr, "RequestCancelHandle"}, + {9, nullptr, "CancelSocketCall"}, + }; RegisterHandlers(functions); } diff --git a/src/core/hle/service/sockets/sfdnsres.h b/src/core/hle/service/sockets/sfdnsres.h index b726a30fd..c07cc1594 100644 --- a/src/core/hle/service/sockets/sfdnsres.h +++ b/src/core/hle/service/sockets/sfdnsres.h @@ -12,7 +12,7 @@ namespace Sockets { class SFDNSRES final : public ServiceFramework<SFDNSRES> { public: - SFDNSRES(); + explicit SFDNSRES(); ~SFDNSRES() = default; private: diff --git a/src/core/hle/service/sockets/sockets.cpp b/src/core/hle/service/sockets/sockets.cpp index f1396eaa1..cedc276d9 100644 --- a/src/core/hle/service/sockets/sockets.cpp +++ b/src/core/hle/service/sockets/sockets.cpp @@ -2,7 +2,8 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include "core/hle/service/sockets/bsd_u.h" +#include "core/hle/service/sockets/bsd.h" +#include "core/hle/service/sockets/nsd.h" #include "core/hle/service/sockets/sfdnsres.h" #include "core/hle/service/sockets/sockets.h" @@ -10,7 +11,10 @@ namespace Service { namespace Sockets { void InstallInterfaces(SM::ServiceManager& service_manager) { - std::make_shared<BSD_U>()->InstallAsService(service_manager); + std::make_shared<BSD>("bsd:s")->InstallAsService(service_manager); + std::make_shared<BSD>("bsd:u")->InstallAsService(service_manager); + std::make_shared<NSD>("nsd:a")->InstallAsService(service_manager); + std::make_shared<NSD>("nsd:u")->InstallAsService(service_manager); std::make_shared<SFDNSRES>()->InstallAsService(service_manager); } diff --git a/src/core/hle/service/spl/csrng.cpp b/src/core/hle/service/spl/csrng.cpp new file mode 100644 index 000000000..cde05717a --- /dev/null +++ b/src/core/hle/service/spl/csrng.cpp @@ -0,0 +1,18 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "core/hle/service/spl/csrng.h" + +namespace Service { +namespace SPL { + +CSRNG::CSRNG(std::shared_ptr<Module> module) : Module::Interface(std::move(module), "csrng") { + static const FunctionInfo functions[] = { + {0, &CSRNG::GetRandomBytes, "GetRandomBytes"}, + }; + RegisterHandlers(functions); +} + +} // namespace SPL +} // namespace Service diff --git a/src/core/hle/service/spl/csrng.h b/src/core/hle/service/spl/csrng.h new file mode 100644 index 000000000..59ca794dd --- /dev/null +++ b/src/core/hle/service/spl/csrng.h @@ -0,0 +1,18 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/service/spl/module.h" + +namespace Service { +namespace SPL { + +class CSRNG final : public Module::Interface { +public: + explicit CSRNG(std::shared_ptr<Module> module); +}; + +} // namespace SPL +} // namespace Service diff --git a/src/core/hle/service/spl/module.cpp b/src/core/hle/service/spl/module.cpp new file mode 100644 index 000000000..fc1bcd94c --- /dev/null +++ b/src/core/hle/service/spl/module.cpp @@ -0,0 +1,42 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <algorithm> +#include <cstdlib> +#include <vector> +#include "common/logging/log.h" +#include "core/hle/ipc_helpers.h" +#include "core/hle/service/spl/csrng.h" +#include "core/hle/service/spl/module.h" +#include "core/hle/service/spl/spl.h" + +namespace Service { +namespace SPL { + +Module::Interface::Interface(std::shared_ptr<Module> module, const char* name) + : ServiceFramework(name), module(std::move(module)) {} + +void Module::Interface::GetRandomBytes(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + + size_t size = ctx.GetWriteBufferSize(); + + std::vector<u8> data(size); + std::generate(data.begin(), data.end(), std::rand); + + ctx.WriteBuffer(data); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); + LOG_DEBUG(Service_SPL, "called"); +} + +void InstallInterfaces(SM::ServiceManager& service_manager) { + auto module = std::make_shared<Module>(); + std::make_shared<CSRNG>(module)->InstallAsService(service_manager); + std::make_shared<SPL>(module)->InstallAsService(service_manager); +} + +} // namespace SPL +} // namespace Service diff --git a/src/core/hle/service/spl/module.h b/src/core/hle/service/spl/module.h new file mode 100644 index 000000000..12cdb2980 --- /dev/null +++ b/src/core/hle/service/spl/module.h @@ -0,0 +1,29 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/service/service.h" + +namespace Service { +namespace SPL { + +class Module final { +public: + class Interface : public ServiceFramework<Interface> { + public: + Interface(std::shared_ptr<Module> module, const char* name); + + void GetRandomBytes(Kernel::HLERequestContext& ctx); + + protected: + std::shared_ptr<Module> module; + }; +}; + +/// Registers all SPL services with the specified service manager. +void InstallInterfaces(SM::ServiceManager& service_manager); + +} // namespace SPL +} // namespace Service diff --git a/src/core/hle/service/spl/spl.cpp b/src/core/hle/service/spl/spl.cpp new file mode 100644 index 000000000..deab29b91 --- /dev/null +++ b/src/core/hle/service/spl/spl.cpp @@ -0,0 +1,41 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "core/hle/service/spl/spl.h" + +namespace Service { +namespace SPL { + +SPL::SPL(std::shared_ptr<Module> module) : Module::Interface(std::move(module), "spl:") { + static const FunctionInfo functions[] = { + {0, nullptr, "GetConfig"}, + {1, nullptr, "UserExpMod"}, + {2, nullptr, "GenerateAesKek"}, + {3, nullptr, "LoadAesKey"}, + {4, nullptr, "GenerateAesKey"}, + {5, nullptr, "SetConfig"}, + {7, &SPL::GetRandomBytes, "GetRandomBytes"}, + {9, nullptr, "LoadSecureExpModKey"}, + {10, nullptr, "SecureExpMod"}, + {11, nullptr, "IsDevelopment"}, + {12, nullptr, "GenerateSpecificAesKey"}, + {13, nullptr, "DecryptPrivk"}, + {14, nullptr, "DecryptAesKey"}, + {15, nullptr, "DecryptAesCtr"}, + {16, nullptr, "ComputeCmac"}, + {17, nullptr, "LoadRsaOaepKey"}, + {18, nullptr, "UnwrapRsaOaepWrappedTitleKey"}, + {19, nullptr, "LoadTitleKey"}, + {20, nullptr, "UnwrapAesWrappedTitleKey"}, + {21, nullptr, "LockAesEngine"}, + {22, nullptr, "UnlockAesEngine"}, + {23, nullptr, "GetSplWaitEvent"}, + {24, nullptr, "SetSharedData"}, + {25, nullptr, "GetSharedData"}, + }; + RegisterHandlers(functions); +} + +} // namespace SPL +} // namespace Service diff --git a/src/core/hle/service/spl/spl.h b/src/core/hle/service/spl/spl.h new file mode 100644 index 000000000..9fd6059af --- /dev/null +++ b/src/core/hle/service/spl/spl.h @@ -0,0 +1,18 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/service/spl/module.h" + +namespace Service { +namespace SPL { + +class SPL final : public Module::Interface { +public: + explicit SPL(std::shared_ptr<Module> module); +}; + +} // namespace SPL +} // namespace Service diff --git a/src/core/hle/service/ssl/ssl.cpp b/src/core/hle/service/ssl/ssl.cpp new file mode 100644 index 000000000..afa8d5d79 --- /dev/null +++ b/src/core/hle/service/ssl/ssl.cpp @@ -0,0 +1,17 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "core/hle/service/ssl/ssl.h" + +namespace Service { +namespace SSL { + +SSL::SSL() : ServiceFramework("ssl") {} + +void InstallInterfaces(SM::ServiceManager& service_manager) { + std::make_shared<SSL>()->InstallAsService(service_manager); +} + +} // namespace SSL +} // namespace Service diff --git a/src/core/hle/service/ssl/ssl.h b/src/core/hle/service/ssl/ssl.h new file mode 100644 index 000000000..645dad003 --- /dev/null +++ b/src/core/hle/service/ssl/ssl.h @@ -0,0 +1,22 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/service/service.h" + +namespace Service { +namespace SSL { + +class SSL final : public ServiceFramework<SSL> { +public: + explicit SSL(); + ~SSL() = default; +}; + +/// Registers all SSL services with the specified service manager. +void InstallInterfaces(SM::ServiceManager& service_manager); + +} // namespace SSL +} // namespace Service diff --git a/src/core/hle/service/time/time.cpp b/src/core/hle/service/time/time.cpp index ad49f4265..c3e46f866 100644 --- a/src/core/hle/service/time/time.cpp +++ b/src/core/hle/service/time/time.cpp @@ -107,7 +107,7 @@ private: IPC::RequestParser rp{ctx}; u64 posix_time = rp.Pop<u64>(); - LOG_WARNING(Service_Time, "(STUBBED) called, posix_time=0x%016llX", posix_time); + LOG_WARNING(Service_Time, "(STUBBED) called, posix_time=0x%016lX", posix_time); CalendarTime calendar_time{2018, 1, 1, 0, 0, 0}; CalendarAdditionalInfo additional_info{}; diff --git a/src/core/hle/service/vi/vi.cpp b/src/core/hle/service/vi/vi.cpp index 0aa621dfe..06c34e979 100644 --- a/src/core/hle/service/vi/vi.cpp +++ b/src/core/hle/service/vi/vi.cpp @@ -4,10 +4,13 @@ #include <algorithm> #include <array> +#include <memory> +#include <boost/optional.hpp> #include "common/alignment.h" #include "common/scope_exit.h" #include "core/core_timing.h" #include "core/hle/ipc_helpers.h" +#include "core/hle/kernel/event.h" #include "core/hle/service/nvdrv/nvdrv.h" #include "core/hle/service/nvflinger/buffer_queue.h" #include "core/hle/service/vi/vi.h" @@ -471,7 +474,7 @@ private: u32 flags = rp.Pop<u32>(); auto buffer_queue = nv_flinger->GetBufferQueue(id); - LOG_DEBUG(Service_VI, "called, transaction=%x", transaction); + LOG_DEBUG(Service_VI, "called, transaction=%x", static_cast<u32>(transaction)); if (transaction == TransactionId::Connect) { IGBPConnectRequestParcel request{ctx.ReadBuffer()}; @@ -486,12 +489,30 @@ private: ctx.WriteBuffer(response.Serialize()); } else if (transaction == TransactionId::DequeueBuffer) { IGBPDequeueBufferRequestParcel request{ctx.ReadBuffer()}; - - u32 slot = buffer_queue->DequeueBuffer(request.data.pixel_format, request.data.width, - request.data.height); - - IGBPDequeueBufferResponseParcel response{slot}; - ctx.WriteBuffer(response.Serialize()); + const u32 width{request.data.width}; + const u32 height{request.data.height}; + boost::optional<u32> slot = buffer_queue->DequeueBuffer(width, height); + + if (slot != boost::none) { + // Buffer is available + IGBPDequeueBufferResponseParcel response{*slot}; + ctx.WriteBuffer(response.Serialize()); + } else { + // Wait the current thread until a buffer becomes available + auto wait_event = ctx.SleepClientThread( + Kernel::GetCurrentThread(), "IHOSBinderDriver::DequeueBuffer", -1, + [=](Kernel::SharedPtr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx, + ThreadWakeupReason reason) { + // Repeat TransactParcel DequeueBuffer when a buffer is available + auto buffer_queue = nv_flinger->GetBufferQueue(id); + boost::optional<u32> slot = buffer_queue->DequeueBuffer(width, height); + IGBPDequeueBufferResponseParcel response{*slot}; + ctx.WriteBuffer(response.Serialize()); + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); + }); + buffer_queue->SetBufferWaitEvent(std::move(wait_event)); + } } else if (transaction == TransactionId::RequestBuffer) { IGBPRequestBufferRequestParcel request{ctx.ReadBuffer()}; @@ -628,144 +649,152 @@ private: std::shared_ptr<NVFlinger::NVFlinger> nv_flinger; }; -void IApplicationDisplayService::GetRelayService(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_VI, "(STUBBED) called"); +class IApplicationDisplayService final : public ServiceFramework<IApplicationDisplayService> { +public: + IApplicationDisplayService(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger); + ~IApplicationDisplayService() = default; - IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<IHOSBinderDriver>(nv_flinger); -} +private: + void GetRelayService(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_VI, "(STUBBED) called"); -void IApplicationDisplayService::GetSystemDisplayService(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_VI, "(STUBBED) called"); + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; + rb.Push(RESULT_SUCCESS); + rb.PushIpcInterface<IHOSBinderDriver>(nv_flinger); + } - IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<ISystemDisplayService>(); -} + void GetSystemDisplayService(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_VI, "(STUBBED) called"); -void IApplicationDisplayService::GetManagerDisplayService(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_VI, "(STUBBED) called"); + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; + rb.Push(RESULT_SUCCESS); + rb.PushIpcInterface<ISystemDisplayService>(); + } - IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<IManagerDisplayService>(nv_flinger); -} + void GetManagerDisplayService(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_VI, "(STUBBED) called"); -void IApplicationDisplayService::GetIndirectDisplayTransactionService( - Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_VI, "(STUBBED) called"); + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; + rb.Push(RESULT_SUCCESS); + rb.PushIpcInterface<IManagerDisplayService>(nv_flinger); + } - IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<IHOSBinderDriver>(nv_flinger); -} + void GetIndirectDisplayTransactionService(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_VI, "(STUBBED) called"); -void IApplicationDisplayService::OpenDisplay(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_VI, "(STUBBED) called"); - IPC::RequestParser rp{ctx}; - auto name_buf = rp.PopRaw<std::array<u8, 0x40>>(); - auto end = std::find(name_buf.begin(), name_buf.end(), '\0'); + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; + rb.Push(RESULT_SUCCESS); + rb.PushIpcInterface<IHOSBinderDriver>(nv_flinger); + } - std::string name(name_buf.begin(), end); + void OpenDisplay(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_VI, "(STUBBED) called"); + IPC::RequestParser rp{ctx}; + auto name_buf = rp.PopRaw<std::array<u8, 0x40>>(); + auto end = std::find(name_buf.begin(), name_buf.end(), '\0'); - ASSERT_MSG(name == "Default", "Non-default displays aren't supported yet"); + std::string name(name_buf.begin(), end); - IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0); - rb.Push(RESULT_SUCCESS); - rb.Push<u64>(nv_flinger->OpenDisplay(name)); -} + ASSERT_MSG(name == "Default", "Non-default displays aren't supported yet"); -void IApplicationDisplayService::CloseDisplay(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_VI, "(STUBBED) called"); - IPC::RequestParser rp{ctx}; - u64 display_id = rp.Pop<u64>(); + IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0); + rb.Push(RESULT_SUCCESS); + rb.Push<u64>(nv_flinger->OpenDisplay(name)); + } - IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0); - rb.Push(RESULT_SUCCESS); -} + void CloseDisplay(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_VI, "(STUBBED) called"); + IPC::RequestParser rp{ctx}; + u64 display_id = rp.Pop<u64>(); + + IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0); + rb.Push(RESULT_SUCCESS); + } + + void SetLayerScalingMode(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_VI, "(STUBBED) called"); + IPC::RequestParser rp{ctx}; + u32 scaling_mode = rp.Pop<u32>(); + u64 unknown = rp.Pop<u64>(); -void IApplicationDisplayService::OpenLayer(Kernel::HLERequestContext& ctx) { - LOG_DEBUG(Service_VI, "called"); - IPC::RequestParser rp{ctx}; - auto name_buf = rp.PopRaw<std::array<u8, 0x40>>(); - auto end = std::find(name_buf.begin(), name_buf.end(), '\0'); + IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0); + rb.Push(RESULT_SUCCESS); + } - std::string display_name(name_buf.begin(), end); + void ListDisplays(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + DisplayInfo display_info; + ctx.WriteBuffer(&display_info, sizeof(DisplayInfo)); + IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0); + rb.Push(RESULT_SUCCESS); + rb.Push<u64>(1); + LOG_WARNING(Service_VI, "(STUBBED) called"); + } - u64 layer_id = rp.Pop<u64>(); - u64 aruid = rp.Pop<u64>(); + void OpenLayer(Kernel::HLERequestContext& ctx) { + LOG_DEBUG(Service_VI, "called"); + IPC::RequestParser rp{ctx}; + auto name_buf = rp.PopRaw<std::array<u8, 0x40>>(); + auto end = std::find(name_buf.begin(), name_buf.end(), '\0'); - u64 display_id = nv_flinger->OpenDisplay(display_name); - u32 buffer_queue_id = nv_flinger->GetBufferQueueId(display_id, layer_id); + std::string display_name(name_buf.begin(), end); - NativeWindow native_window{buffer_queue_id}; - IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0); - rb.Push(RESULT_SUCCESS); - rb.Push<u64>(ctx.WriteBuffer(native_window.Serialize())); -} + u64 layer_id = rp.Pop<u64>(); + u64 aruid = rp.Pop<u64>(); -void IApplicationDisplayService::CreateStrayLayer(Kernel::HLERequestContext& ctx) { - LOG_DEBUG(Service_VI, "called"); + u64 display_id = nv_flinger->OpenDisplay(display_name); + u32 buffer_queue_id = nv_flinger->GetBufferQueueId(display_id, layer_id); - IPC::RequestParser rp{ctx}; - u32 flags = rp.Pop<u32>(); - rp.Pop<u32>(); // padding - u64 display_id = rp.Pop<u64>(); + NativeWindow native_window{buffer_queue_id}; + IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0); + rb.Push(RESULT_SUCCESS); + rb.Push<u64>(ctx.WriteBuffer(native_window.Serialize())); + } - // TODO(Subv): What's the difference between a Stray and a Managed layer? + void CreateStrayLayer(Kernel::HLERequestContext& ctx) { + LOG_DEBUG(Service_VI, "called"); - u64 layer_id = nv_flinger->CreateLayer(display_id); - u32 buffer_queue_id = nv_flinger->GetBufferQueueId(display_id, layer_id); + IPC::RequestParser rp{ctx}; + u32 flags = rp.Pop<u32>(); + rp.Pop<u32>(); // padding + u64 display_id = rp.Pop<u64>(); - NativeWindow native_window{buffer_queue_id}; - IPC::ResponseBuilder rb = rp.MakeBuilder(6, 0, 0); - rb.Push(RESULT_SUCCESS); - rb.Push(layer_id); - rb.Push<u64>(ctx.WriteBuffer(native_window.Serialize())); -} + // TODO(Subv): What's the difference between a Stray and a Managed layer? -void IApplicationDisplayService::DestroyStrayLayer(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_VI, "(STUBBED) called"); + u64 layer_id = nv_flinger->CreateLayer(display_id); + u32 buffer_queue_id = nv_flinger->GetBufferQueueId(display_id, layer_id); - IPC::RequestParser rp{ctx}; - u64 layer_id = rp.Pop<u64>(); + NativeWindow native_window{buffer_queue_id}; + IPC::ResponseBuilder rb = rp.MakeBuilder(6, 0, 0); + rb.Push(RESULT_SUCCESS); + rb.Push(layer_id); + rb.Push<u64>(ctx.WriteBuffer(native_window.Serialize())); + } - IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0); - rb.Push(RESULT_SUCCESS); -} + void DestroyStrayLayer(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_VI, "(STUBBED) called"); -void IApplicationDisplayService::SetLayerScalingMode(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_VI, "(STUBBED) called"); - IPC::RequestParser rp{ctx}; - u32 scaling_mode = rp.Pop<u32>(); - u64 unknown = rp.Pop<u64>(); + IPC::RequestParser rp{ctx}; + u64 layer_id = rp.Pop<u64>(); - IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0); - rb.Push(RESULT_SUCCESS); -} + IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0); + rb.Push(RESULT_SUCCESS); + } -void IApplicationDisplayService::ListDisplays(Kernel::HLERequestContext& ctx) { - IPC::RequestParser rp{ctx}; - DisplayInfo display_info; - ctx.WriteBuffer(&display_info, sizeof(DisplayInfo)); - IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0); - rb.Push(RESULT_SUCCESS); - rb.Push<u64>(1); - LOG_WARNING(Service_VI, "(STUBBED) called"); -} + void GetDisplayVsyncEvent(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_VI, "(STUBBED) called"); + IPC::RequestParser rp{ctx}; + u64 display_id = rp.Pop<u64>(); -void IApplicationDisplayService::GetDisplayVsyncEvent(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_VI, "(STUBBED) called"); - IPC::RequestParser rp{ctx}; - u64 display_id = rp.Pop<u64>(); + auto vsync_event = nv_flinger->GetVsyncEvent(display_id); - auto vsync_event = nv_flinger->GetVsyncEvent(display_id); + IPC::ResponseBuilder rb = rp.MakeBuilder(2, 1, 0); + rb.Push(RESULT_SUCCESS); + rb.PushCopyObjects(vsync_event); + } - IPC::ResponseBuilder rb = rp.MakeBuilder(2, 1, 0); - rb.Push(RESULT_SUCCESS); - rb.PushCopyObjects(vsync_event); -} + std::shared_ptr<NVFlinger::NVFlinger> nv_flinger; +}; IApplicationDisplayService::IApplicationDisplayService( std::shared_ptr<NVFlinger::NVFlinger> nv_flinger) @@ -788,11 +817,24 @@ IApplicationDisplayService::IApplicationDisplayService( RegisterHandlers(functions); } +Module::Interface::Interface(std::shared_ptr<Module> module, const char* name, + std::shared_ptr<NVFlinger::NVFlinger> nv_flinger) + : ServiceFramework(name), module(std::move(module)), nv_flinger(std::move(nv_flinger)) {} + +void Module::Interface::GetDisplayService(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_VI, "(STUBBED) called"); + + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; + rb.Push(RESULT_SUCCESS); + rb.PushIpcInterface<IApplicationDisplayService>(nv_flinger); +} + void InstallInterfaces(SM::ServiceManager& service_manager, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger) { - std::make_shared<VI_M>(nv_flinger)->InstallAsService(service_manager); - std::make_shared<VI_S>(nv_flinger)->InstallAsService(service_manager); - std::make_shared<VI_U>(nv_flinger)->InstallAsService(service_manager); + auto module = std::make_shared<Module>(); + std::make_shared<VI_M>(module, nv_flinger)->InstallAsService(service_manager); + std::make_shared<VI_S>(module, nv_flinger)->InstallAsService(service_manager); + std::make_shared<VI_U>(module, nv_flinger)->InstallAsService(service_manager); } } // namespace VI diff --git a/src/core/hle/service/vi/vi.h b/src/core/hle/service/vi/vi.h index f6be7d1e6..985c9d27c 100644 --- a/src/core/hle/service/vi/vi.h +++ b/src/core/hle/service/vi/vi.h @@ -4,9 +4,6 @@ #pragma once -#include <memory> -#include <boost/optional.hpp> -#include "core/hle/kernel/event.h" #include "core/hle/service/nvflinger/nvflinger.h" #include "core/hle/service/service.h" @@ -17,26 +14,19 @@ struct EventType; namespace Service { namespace VI { -class IApplicationDisplayService final : public ServiceFramework<IApplicationDisplayService> { +class Module final { public: - IApplicationDisplayService(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger); - ~IApplicationDisplayService() = default; - -private: - void GetRelayService(Kernel::HLERequestContext& ctx); - void GetSystemDisplayService(Kernel::HLERequestContext& ctx); - void GetManagerDisplayService(Kernel::HLERequestContext& ctx); - void GetIndirectDisplayTransactionService(Kernel::HLERequestContext& ctx); - void OpenDisplay(Kernel::HLERequestContext& ctx); - void CloseDisplay(Kernel::HLERequestContext& ctx); - void SetLayerScalingMode(Kernel::HLERequestContext& ctx); - void ListDisplays(Kernel::HLERequestContext& ctx); - void OpenLayer(Kernel::HLERequestContext& ctx); - void CreateStrayLayer(Kernel::HLERequestContext& ctx); - void DestroyStrayLayer(Kernel::HLERequestContext& ctx); - void GetDisplayVsyncEvent(Kernel::HLERequestContext& ctx); - - std::shared_ptr<NVFlinger::NVFlinger> nv_flinger; + class Interface : public ServiceFramework<Interface> { + public: + Interface(std::shared_ptr<Module> module, const char* name, + std::shared_ptr<NVFlinger::NVFlinger> nv_flinger); + + void GetDisplayService(Kernel::HLERequestContext& ctx); + + protected: + std::shared_ptr<Module> module; + std::shared_ptr<NVFlinger::NVFlinger> nv_flinger; + }; }; /// Registers all VI services with the specified service manager. diff --git a/src/core/hle/service/vi/vi_m.cpp b/src/core/hle/service/vi/vi_m.cpp index 5d99647dc..5781fa9ec 100644 --- a/src/core/hle/service/vi/vi_m.cpp +++ b/src/core/hle/service/vi/vi_m.cpp @@ -2,24 +2,13 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include "common/logging/log.h" -#include "core/hle/ipc_helpers.h" -#include "core/hle/service/vi/vi.h" #include "core/hle/service/vi/vi_m.h" namespace Service { namespace VI { -void VI_M::GetDisplayService(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_VI, "(STUBBED) called"); - - IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<IApplicationDisplayService>(nv_flinger); -} - -VI_M::VI_M(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger) - : ServiceFramework("vi:m"), nv_flinger(std::move(nv_flinger)) { +VI_M::VI_M(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger) + : Module::Interface(std::move(module), "vi:m", std::move(nv_flinger)) { static const FunctionInfo functions[] = { {2, &VI_M::GetDisplayService, "GetDisplayService"}, {3, nullptr, "GetDisplayServiceWithProxyNameExchange"}, diff --git a/src/core/hle/service/vi/vi_m.h b/src/core/hle/service/vi/vi_m.h index e5319b1e7..0f7b799d6 100644 --- a/src/core/hle/service/vi/vi_m.h +++ b/src/core/hle/service/vi/vi_m.h @@ -4,25 +4,14 @@ #pragma once -#include <memory> -#include "core/hle/service/service.h" +#include "core/hle/service/vi/vi.h" namespace Service { -namespace NVFlinger { -class NVFlinger; -} - namespace VI { -class VI_M final : public ServiceFramework<VI_M> { +class VI_M final : public Module::Interface { public: - VI_M(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger); - ~VI_M() = default; - -private: - void GetDisplayService(Kernel::HLERequestContext& ctx); - - std::shared_ptr<NVFlinger::NVFlinger> nv_flinger; + explicit VI_M(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger); }; } // namespace VI diff --git a/src/core/hle/service/vi/vi_s.cpp b/src/core/hle/service/vi/vi_s.cpp index 411757981..1f937b2a8 100644 --- a/src/core/hle/service/vi/vi_s.cpp +++ b/src/core/hle/service/vi/vi_s.cpp @@ -2,24 +2,13 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include "common/logging/log.h" -#include "core/hle/ipc_helpers.h" -#include "core/hle/service/vi/vi.h" #include "core/hle/service/vi/vi_s.h" namespace Service { namespace VI { -void VI_S::GetDisplayService(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_VI, "(STUBBED) called"); - - IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<IApplicationDisplayService>(nv_flinger); -} - -VI_S::VI_S(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger) - : ServiceFramework("vi:s"), nv_flinger(std::move(nv_flinger)) { +VI_S::VI_S(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger) + : Module::Interface(std::move(module), "vi:s", std::move(nv_flinger)) { static const FunctionInfo functions[] = { {1, &VI_S::GetDisplayService, "GetDisplayService"}, {3, nullptr, "GetDisplayServiceWithProxyNameExchange"}, diff --git a/src/core/hle/service/vi/vi_s.h b/src/core/hle/service/vi/vi_s.h index 6978fd700..7b32fdddc 100644 --- a/src/core/hle/service/vi/vi_s.h +++ b/src/core/hle/service/vi/vi_s.h @@ -4,25 +4,14 @@ #pragma once -#include <memory> -#include "core/hle/service/service.h" +#include "core/hle/service/vi/vi.h" namespace Service { -namespace NVFlinger { -class NVFlinger; -} - namespace VI { -class VI_S final : public ServiceFramework<VI_S> { +class VI_S final : public Module::Interface { public: - VI_S(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger); - ~VI_S() = default; - -private: - void GetDisplayService(Kernel::HLERequestContext& ctx); - - std::shared_ptr<NVFlinger::NVFlinger> nv_flinger; + explicit VI_S(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger); }; } // namespace VI diff --git a/src/core/hle/service/vi/vi_u.cpp b/src/core/hle/service/vi/vi_u.cpp index f5568383b..14e375b86 100644 --- a/src/core/hle/service/vi/vi_u.cpp +++ b/src/core/hle/service/vi/vi_u.cpp @@ -2,24 +2,13 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include "common/logging/log.h" -#include "core/hle/ipc_helpers.h" -#include "core/hle/service/vi/vi.h" #include "core/hle/service/vi/vi_u.h" namespace Service { namespace VI { -void VI_U::GetDisplayService(Kernel::HLERequestContext& ctx) { - LOG_WARNING(Service_VI, "(STUBBED) called"); - - IPC::ResponseBuilder rb{ctx, 2, 0, 1}; - rb.Push(RESULT_SUCCESS); - rb.PushIpcInterface<IApplicationDisplayService>(nv_flinger); -} - -VI_U::VI_U(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger) - : ServiceFramework("vi:u"), nv_flinger(std::move(nv_flinger)) { +VI_U::VI_U(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger) + : Module::Interface(std::move(module), "vi:u", std::move(nv_flinger)) { static const FunctionInfo functions[] = { {0, &VI_U::GetDisplayService, "GetDisplayService"}, {3, nullptr, "GetDisplayServiceWithProxyNameExchange"}, diff --git a/src/core/hle/service/vi/vi_u.h b/src/core/hle/service/vi/vi_u.h index b3e9c094d..c557a2235 100644 --- a/src/core/hle/service/vi/vi_u.h +++ b/src/core/hle/service/vi/vi_u.h @@ -4,25 +4,14 @@ #pragma once -#include <memory> -#include "core/hle/service/service.h" +#include "core/hle/service/vi/vi.h" namespace Service { -namespace NVFlinger { -class NVFlinger; -} - namespace VI { -class VI_U final : public ServiceFramework<VI_U> { +class VI_U final : public Module::Interface { public: - VI_U(std::shared_ptr<NVFlinger::NVFlinger> nv_flinger); - ~VI_U() = default; - -private: - void GetDisplayService(Kernel::HLERequestContext& ctx); - - std::shared_ptr<NVFlinger::NVFlinger> nv_flinger; + explicit VI_U(std::shared_ptr<Module> module, std::shared_ptr<NVFlinger::NVFlinger> nv_flinger); }; } // namespace VI diff --git a/src/core/loader/deconstructed_rom_directory.cpp b/src/core/loader/deconstructed_rom_directory.cpp index 864cf25cd..8b4ee970f 100644 --- a/src/core/loader/deconstructed_rom_directory.cpp +++ b/src/core/loader/deconstructed_rom_directory.cpp @@ -76,7 +76,7 @@ FileType AppLoader_DeconstructedRomDirectory::IdentifyType(FileUtil::IOFile& fil } else if (Common::ToLower(virtual_name) == "sdk") { is_sdk_found = true; } else { - // Contrinue searching + // Continue searching return true; } @@ -110,8 +110,6 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load( return ResultStatus::Error; } - process = Kernel::Process::Create("main"); - const std::string directory = filepath.substr(0, filepath.find_last_of("/\\")) + DIR_SEP; const std::string npdm_path = directory + DIR_SEP + "main.npdm"; @@ -127,7 +125,7 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load( "subsdk4", "subsdk5", "subsdk6", "subsdk7", "sdk"}) { const std::string path = directory + DIR_SEP + module; const VAddr load_addr = next_load_addr; - next_load_addr = AppLoader_NSO::LoadModule(path, load_addr, metadata.GetTitleID()); + next_load_addr = AppLoader_NSO::LoadModule(path, load_addr); if (next_load_addr) { LOG_DEBUG(Loader, "loaded module %s @ 0x%" PRIx64, module, load_addr); } else { @@ -135,6 +133,7 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load( } } + process->program_id = metadata.GetTitleID(); process->svc_access_mask.set(); process->address_mappings = default_address_mappings; process->resource_limit = diff --git a/src/core/loader/elf.cpp b/src/core/loader/elf.cpp index b87320656..e9f462196 100644 --- a/src/core/loader/elf.cpp +++ b/src/core/loader/elf.cpp @@ -300,7 +300,7 @@ SharedPtr<CodeSet> ElfReader::LoadInto(u32 vaddr) { std::vector<u8> program_image(total_image_size); size_t current_image_position = 0; - SharedPtr<CodeSet> codeset = CodeSet::Create("", 0); + SharedPtr<CodeSet> codeset = CodeSet::Create(""); for (unsigned int i = 0; i < header->e_phnum; ++i) { Elf32_Phdr* p = &segments[i]; @@ -406,7 +406,6 @@ ResultStatus AppLoader_ELF::Load(Kernel::SharedPtr<Kernel::Process>& process) { SharedPtr<CodeSet> codeset = elf_reader.LoadInto(Memory::PROCESS_IMAGE_VADDR); codeset->name = filename; - process = Kernel::Process::Create("main"); process->LoadModule(codeset, codeset->entrypoint); process->svc_access_mask.set(); process->address_mappings = default_address_mappings; @@ -415,7 +414,7 @@ ResultStatus AppLoader_ELF::Load(Kernel::SharedPtr<Kernel::Process>& process) { process->resource_limit = Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION); - process->Run(codeset->entrypoint, 48, Kernel::DEFAULT_STACK_SIZE); + process->Run(codeset->entrypoint, 48, Memory::DEFAULT_STACK_SIZE); is_loaded = true; return ResultStatus::Success; diff --git a/src/core/loader/linker.cpp b/src/core/loader/linker.cpp index 87cc65e91..69198e3e3 100644 --- a/src/core/loader/linker.cpp +++ b/src/core/loader/linker.cpp @@ -84,7 +84,7 @@ void Linker::WriteRelocations(std::vector<u8>& program_image, const std::vector< } break; default: - LOG_CRITICAL(Loader, "Unknown relocation type: %d", rela.type); + LOG_CRITICAL(Loader, "Unknown relocation type: %d", static_cast<int>(rela.type)); break; } } diff --git a/src/core/loader/nro.cpp b/src/core/loader/nro.cpp index 6f8a2f21e..b5133e4d6 100644 --- a/src/core/loader/nro.cpp +++ b/src/core/loader/nro.cpp @@ -8,6 +8,7 @@ #include "common/file_util.h" #include "common/logging/log.h" #include "common/swap.h" +#include "core/core.h" #include "core/hle/kernel/process.h" #include "core/hle/kernel/resource_limit.h" #include "core/loader/nro.h" @@ -83,7 +84,7 @@ bool AppLoader_NRO::LoadNro(const std::string& path, VAddr load_base) { } // Build program image - Kernel::SharedPtr<Kernel::CodeSet> codeset = Kernel::CodeSet::Create("", 0); + Kernel::SharedPtr<Kernel::CodeSet> codeset = Kernel::CodeSet::Create(""); std::vector<u8> program_image; program_image.resize(PageAlignSize(nro_header.file_size)); file.Seek(0, SEEK_SET); @@ -112,7 +113,7 @@ bool AppLoader_NRO::LoadNro(const std::string& path, VAddr load_base) { // Load codeset for current process codeset->name = path; codeset->memory = std::make_shared<std::vector<u8>>(std::move(program_image)); - Kernel::g_current_process->LoadModule(codeset, load_base); + Core::CurrentProcess()->LoadModule(codeset, load_base); return true; } @@ -125,8 +126,6 @@ ResultStatus AppLoader_NRO::Load(Kernel::SharedPtr<Kernel::Process>& process) { return ResultStatus::Error; } - process = Kernel::Process::Create("main"); - // Load NRO static constexpr VAddr base_addr{Memory::PROCESS_IMAGE_VADDR}; @@ -138,7 +137,7 @@ ResultStatus AppLoader_NRO::Load(Kernel::SharedPtr<Kernel::Process>& process) { process->address_mappings = default_address_mappings; process->resource_limit = Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION); - process->Run(base_addr, 48, Kernel::DEFAULT_STACK_SIZE); + process->Run(base_addr, 48, Memory::DEFAULT_STACK_SIZE); is_loaded = true; return ResultStatus::Success; diff --git a/src/core/loader/nso.cpp b/src/core/loader/nso.cpp index 7f8d24dd6..3bc10ed0d 100644 --- a/src/core/loader/nso.cpp +++ b/src/core/loader/nso.cpp @@ -9,6 +9,7 @@ #include "common/file_util.h" #include "common/logging/log.h" #include "common/swap.h" +#include "core/core.h" #include "core/hle/kernel/process.h" #include "core/hle/kernel/resource_limit.h" #include "core/loader/nso.h" @@ -92,7 +93,7 @@ static constexpr u32 PageAlignSize(u32 size) { return (size + Memory::PAGE_MASK) & ~Memory::PAGE_MASK; } -VAddr AppLoader_NSO::LoadModule(const std::string& path, VAddr load_base, u64 tid) { +VAddr AppLoader_NSO::LoadModule(const std::string& path, VAddr load_base) { FileUtil::IOFile file(path, "rb"); if (!file.IsOpen()) { return {}; @@ -109,7 +110,7 @@ VAddr AppLoader_NSO::LoadModule(const std::string& path, VAddr load_base, u64 ti } // Build program image - Kernel::SharedPtr<Kernel::CodeSet> codeset = Kernel::CodeSet::Create("", tid); + Kernel::SharedPtr<Kernel::CodeSet> codeset = Kernel::CodeSet::Create(""); std::vector<u8> program_image; for (int i = 0; i < nso_header.segments.size(); ++i) { std::vector<u8> data = @@ -142,7 +143,7 @@ VAddr AppLoader_NSO::LoadModule(const std::string& path, VAddr load_base, u64 ti // Load codeset for current process codeset->name = path; codeset->memory = std::make_shared<std::vector<u8>>(std::move(program_image)); - Kernel::g_current_process->LoadModule(codeset, load_base); + Core::CurrentProcess()->LoadModule(codeset, load_base); return load_base + image_size; } @@ -155,10 +156,8 @@ ResultStatus AppLoader_NSO::Load(Kernel::SharedPtr<Kernel::Process>& process) { return ResultStatus::Error; } - process = Kernel::Process::Create("main"); - // Load module - LoadModule(filepath, Memory::PROCESS_IMAGE_VADDR, 0); + LoadModule(filepath, Memory::PROCESS_IMAGE_VADDR); LOG_DEBUG(Loader, "loaded module %s @ 0x%" PRIx64, filepath.c_str(), Memory::PROCESS_IMAGE_VADDR); @@ -166,7 +165,7 @@ ResultStatus AppLoader_NSO::Load(Kernel::SharedPtr<Kernel::Process>& process) { process->address_mappings = default_address_mappings; process->resource_limit = Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION); - process->Run(Memory::PROCESS_IMAGE_VADDR, 48, Kernel::DEFAULT_STACK_SIZE); + process->Run(Memory::PROCESS_IMAGE_VADDR, 48, Memory::DEFAULT_STACK_SIZE); is_loaded = true; return ResultStatus::Success; diff --git a/src/core/loader/nso.h b/src/core/loader/nso.h index 14eb1d87e..1ae30a824 100644 --- a/src/core/loader/nso.h +++ b/src/core/loader/nso.h @@ -29,7 +29,7 @@ public: return IdentifyType(file, filepath); } - static VAddr LoadModule(const std::string& path, VAddr load_base, u64 tid); + static VAddr LoadModule(const std::string& path, VAddr load_base); ResultStatus Load(Kernel::SharedPtr<Kernel::Process>& process) override; diff --git a/src/core/memory.cpp b/src/core/memory.cpp index ce62666d7..291bf066f 100644 --- a/src/core/memory.cpp +++ b/src/core/memory.cpp @@ -15,6 +15,7 @@ #include "core/core.h" #include "core/hle/kernel/memory.h" #include "core/hle/kernel/process.h" +#include "core/hle/lock.h" #include "core/memory.h" #include "core/memory_setup.h" #include "video_core/renderer_base.h" @@ -23,7 +24,6 @@ namespace Memory { static std::array<u8, Memory::VRAM_SIZE> vram; -static std::array<u8, Memory::N3DS_EXTRA_RAM_SIZE> n3ds_extra_ram; static PageTable* current_page_table = nullptr; @@ -42,6 +42,9 @@ static void MapPages(PageTable& page_table, VAddr base, u64 size, u8* memory, Pa LOG_DEBUG(HW_Memory, "Mapping %p onto %016" PRIX64 "-%016" PRIX64, memory, base * PAGE_SIZE, (base + size) * PAGE_SIZE); + RasterizerFlushVirtualRegion(base << PAGE_BITS, size * PAGE_SIZE, + FlushMode::FlushAndInvalidate); + VAddr end = base + size; while (base != end) { ASSERT_MSG(base < PAGE_TABLE_NUM_ENTRIES, "out of range mapping at %016" PRIX64, base); @@ -109,100 +112,129 @@ static std::set<MemoryHookPointer> GetSpecialHandlers(const PageTable& page_tabl } static std::set<MemoryHookPointer> GetSpecialHandlers(VAddr vaddr, u64 size) { - const PageTable& page_table = Kernel::g_current_process->vm_manager.page_table; + const PageTable& page_table = Core::CurrentProcess()->vm_manager.page_table; return GetSpecialHandlers(page_table, vaddr, size); } -template <typename T> -boost::optional<T> ReadSpecial(VAddr addr); +/** + * Gets a pointer to the exact memory at the virtual address (i.e. not page aligned) + * using a VMA from the current process + */ +static u8* GetPointerFromVMA(const Kernel::Process& process, VAddr vaddr) { + u8* direct_pointer = nullptr; + + auto& vm_manager = process.vm_manager; + + auto it = vm_manager.FindVMA(vaddr); + ASSERT(it != vm_manager.vma_map.end()); + + auto& vma = it->second; + switch (vma.type) { + case Kernel::VMAType::AllocatedMemoryBlock: + direct_pointer = vma.backing_block->data() + vma.offset; + break; + case Kernel::VMAType::BackingMemory: + direct_pointer = vma.backing_memory; + break; + case Kernel::VMAType::Free: + return nullptr; + default: + UNREACHABLE(); + } + + return direct_pointer + (vaddr - vma.base); +} + +/** + * Gets a pointer to the exact memory at the virtual address (i.e. not page aligned) + * using a VMA from the current process. + */ +static u8* GetPointerFromVMA(VAddr vaddr) { + return GetPointerFromVMA(*Core::CurrentProcess(), vaddr); +} template <typename T> T Read(const VAddr vaddr) { - if ((vaddr >> PAGE_BITS) >= PAGE_TABLE_NUM_ENTRIES) { - LOG_ERROR(HW_Memory, "Read%lu after page table @ 0x%016" PRIX64, sizeof(T) * 8, vaddr); - return 0; + const u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS]; + if (page_pointer) { + // NOTE: Avoid adding any extra logic to this fast-path block + T value; + std::memcpy(&value, &page_pointer[vaddr & PAGE_MASK], sizeof(T)); + return value; } - const PageType type = current_page_table->attributes[vaddr >> PAGE_BITS]; + // The memory access might do an MMIO or cached access, so we have to lock the HLE kernel state + std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock); + + PageType type = current_page_table->attributes[vaddr >> PAGE_BITS]; switch (type) { case PageType::Unmapped: - LOG_ERROR(HW_Memory, "unmapped Read%zu @ 0x%016" PRIX64, sizeof(T) * 8, vaddr); + LOG_ERROR(HW_Memory, "unmapped Read%lu @ 0x%08X", sizeof(T) * 8, vaddr); return 0; - case PageType::Special: { - if (auto result = ReadSpecial<T>(vaddr)) - return *result; - [[fallthrough]]; - } - case PageType::Memory: { - const u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS]; - ASSERT_MSG(page_pointer, "Mapped memory page without a pointer @ %016" PRIX64, vaddr); + case PageType::Memory: + ASSERT_MSG(false, "Mapped memory page without a pointer @ %08X", vaddr); + break; + case PageType::RasterizerCachedMemory: { + RasterizerFlushVirtualRegion(vaddr, sizeof(T), FlushMode::Flush); T value; - std::memcpy(&value, &page_pointer[vaddr & PAGE_MASK], sizeof(T)); + std::memcpy(&value, GetPointerFromVMA(vaddr), sizeof(T)); return value; } + default: + UNREACHABLE(); } - UNREACHABLE(); - return 0; } template <typename T> -bool WriteSpecial(VAddr addr, const T data); - -template <typename T> void Write(const VAddr vaddr, const T data) { - if ((vaddr >> PAGE_BITS) >= PAGE_TABLE_NUM_ENTRIES) { - LOG_ERROR(HW_Memory, "Write%lu after page table 0x%08X @ 0x%016" PRIX64, sizeof(data) * 8, - (u32)data, vaddr); + u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS]; + if (page_pointer) { + // NOTE: Avoid adding any extra logic to this fast-path block + std::memcpy(&page_pointer[vaddr & PAGE_MASK], &data, sizeof(T)); return; } - const PageType type = current_page_table->attributes[vaddr >> PAGE_BITS]; + // The memory access might do an MMIO or cached access, so we have to lock the HLE kernel state + std::lock_guard<std::recursive_mutex> lock(HLE::g_hle_lock); + + PageType type = current_page_table->attributes[vaddr >> PAGE_BITS]; switch (type) { case PageType::Unmapped: - LOG_ERROR(HW_Memory, "unmapped Write%zu 0x%08X @ 0x%016" PRIX64, sizeof(data) * 8, - static_cast<u32>(data), vaddr); - return; - case PageType::Special: { - if (WriteSpecial<T>(vaddr, data)) - return; - [[fallthrough]]; - } - case PageType::Memory: { - u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS]; - ASSERT_MSG(page_pointer, "Mapped memory page without a pointer @ %016" PRIX64, vaddr); - std::memcpy(&page_pointer[vaddr & PAGE_MASK], &data, sizeof(T)); + LOG_ERROR(HW_Memory, "unmapped Write%lu 0x%08X @ 0x%08X", sizeof(data) * 8, (u32)data, + vaddr); return; + case PageType::Memory: + ASSERT_MSG(false, "Mapped memory page without a pointer @ %08X", vaddr); + break; + case PageType::RasterizerCachedMemory: { + RasterizerFlushVirtualRegion(vaddr, sizeof(T), FlushMode::Invalidate); + std::memcpy(GetPointerFromVMA(vaddr), &data, sizeof(T)); + break; } + default: + UNREACHABLE(); } - UNREACHABLE(); } bool IsValidVirtualAddress(const Kernel::Process& process, const VAddr vaddr) { auto& page_table = process.vm_manager.page_table; - if ((vaddr >> PAGE_BITS) >= PAGE_TABLE_NUM_ENTRIES) - return false; + const u8* page_pointer = page_table.pointers[vaddr >> PAGE_BITS]; + if (page_pointer) + return true; - const PageType type = current_page_table->attributes[vaddr >> PAGE_BITS]; - switch (type) { - case PageType::Unmapped: - return false; - case PageType::Memory: + if (page_table.attributes[vaddr >> PAGE_BITS] == PageType::RasterizerCachedMemory) return true; - case PageType::Special: { - for (auto handler : GetSpecialHandlers(page_table, vaddr, 1)) - if (auto result = handler->IsValidAddress(vaddr)) - return *result; - return current_page_table->pointers[vaddr >> PAGE_BITS] != nullptr; - } - } - UNREACHABLE(); + + if (page_table.attributes[vaddr >> PAGE_BITS] != PageType::Special) + return false; + return false; } bool IsValidVirtualAddress(const VAddr vaddr) { - return IsValidVirtualAddress(*Kernel::g_current_process, vaddr); + return IsValidVirtualAddress(*Core::CurrentProcess(), vaddr); } bool IsValidPhysicalAddress(const PAddr paddr) { @@ -215,7 +247,11 @@ u8* GetPointer(const VAddr vaddr) { return page_pointer + (vaddr & PAGE_MASK); } - LOG_ERROR(HW_Memory, "unknown GetPointer @ 0x%016" PRIx64, vaddr); + if (current_page_table->attributes[vaddr >> PAGE_BITS] == PageType::RasterizerCachedMemory) { + return GetPointerFromVMA(vaddr); + } + + LOG_ERROR(HW_Memory, "unknown GetPointer @ 0x%08x", vaddr); return nullptr; } @@ -244,7 +280,6 @@ u8* GetPhysicalPointer(PAddr address) { {IO_AREA_PADDR, IO_AREA_SIZE}, {DSP_RAM_PADDR, DSP_RAM_SIZE}, {FCRAM_PADDR, FCRAM_N3DS_SIZE}, - {N3DS_EXTRA_RAM_PADDR, N3DS_EXTRA_RAM_SIZE}, }; const auto area = @@ -283,9 +318,6 @@ u8* GetPhysicalPointer(PAddr address) { } ASSERT_MSG(target_pointer != nullptr, "Invalid FCRAM address"); break; - case N3DS_EXTRA_RAM_PADDR: - target_pointer = n3ds_extra_ram.data() + offset_into_region; - break; default: UNREACHABLE(); } @@ -293,6 +325,95 @@ u8* GetPhysicalPointer(PAddr address) { return target_pointer; } +void RasterizerMarkRegionCached(VAddr start, u64 size, bool cached) { + if (start == 0) { + return; + } + + u64 num_pages = ((start + size - 1) >> PAGE_BITS) - (start >> PAGE_BITS) + 1; + VAddr vaddr = start; + + for (unsigned i = 0; i < num_pages; ++i, vaddr += PAGE_SIZE) { + PageType& page_type = current_page_table->attributes[vaddr >> PAGE_BITS]; + + if (cached) { + // Switch page type to cached if now cached + switch (page_type) { + case PageType::Unmapped: + // It is not necessary for a process to have this region mapped into its address + // space, for example, a system module need not have a VRAM mapping. + break; + case PageType::Memory: + page_type = PageType::RasterizerCachedMemory; + current_page_table->pointers[vaddr >> PAGE_BITS] = nullptr; + break; + default: + UNREACHABLE(); + } + } else { + // Switch page type to uncached if now uncached + switch (page_type) { + case PageType::Unmapped: + // It is not necessary for a process to have this region mapped into its address + // space, for example, a system module need not have a VRAM mapping. + break; + case PageType::RasterizerCachedMemory: { + u8* pointer = GetPointerFromVMA(vaddr & ~PAGE_MASK); + if (pointer == nullptr) { + // It's possible that this function has been called while updating the pagetable + // after unmapping a VMA. In that case the underlying VMA will no longer exist, + // and we should just leave the pagetable entry blank. + page_type = PageType::Unmapped; + } else { + page_type = PageType::Memory; + current_page_table->pointers[vaddr >> PAGE_BITS] = pointer; + } + break; + } + default: + UNREACHABLE(); + } + } + } +} + +void RasterizerFlushVirtualRegion(VAddr start, u64 size, FlushMode mode) { + // Since pages are unmapped on shutdown after video core is shutdown, the renderer may be + // null here + if (VideoCore::g_renderer == nullptr) { + return; + } + + VAddr end = start + size; + + auto CheckRegion = [&](VAddr region_start, VAddr region_end) { + if (start >= region_end || end <= region_start) { + // No overlap with region + return; + } + + VAddr overlap_start = std::max(start, region_start); + VAddr overlap_end = std::min(end, region_end); + u64 overlap_size = overlap_end - overlap_start; + + auto* rasterizer = VideoCore::g_renderer->Rasterizer(); + switch (mode) { + case FlushMode::Flush: + rasterizer->FlushRegion(overlap_start, overlap_size); + break; + case FlushMode::Invalidate: + rasterizer->InvalidateRegion(overlap_start, overlap_size); + break; + case FlushMode::FlushAndInvalidate: + rasterizer->FlushAndInvalidateRegion(overlap_start, overlap_size); + break; + } + }; + + CheckRegion(PROCESS_IMAGE_VADDR, PROCESS_IMAGE_VADDR_END); + CheckRegion(HEAP_VADDR, HEAP_VADDR_END); +} + u8 Read8(const VAddr addr) { return Read<u8>(addr); } @@ -309,17 +430,6 @@ u64 Read64(const VAddr addr) { return Read<u64_le>(addr); } -static bool ReadSpecialBlock(const Kernel::Process& process, const VAddr src_addr, - void* dest_buffer, const size_t size) { - auto& page_table = process.vm_manager.page_table; - for (const auto& handler : GetSpecialHandlers(page_table, src_addr, size)) { - if (handler->ReadBlock(src_addr, dest_buffer, size)) { - return true; - } - } - return false; -} - void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_buffer, const size_t size) { auto& page_table = process.vm_manager.page_table; @@ -329,21 +439,16 @@ void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_ size_t page_offset = src_addr & PAGE_MASK; while (remaining_size > 0) { - const size_t copy_amount = std::min<size_t>(PAGE_SIZE - page_offset, remaining_size); + const size_t copy_amount = + std::min(static_cast<size_t>(PAGE_SIZE) - page_offset, remaining_size); const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset); switch (page_table.attributes[page_index]) { - case PageType::Unmapped: - LOG_ERROR(HW_Memory, - "unmapped ReadBlock @ 0x%016" PRIX64 " (start address = 0x%" PRIx64 - ", size = %zu)", + case PageType::Unmapped: { + LOG_ERROR(HW_Memory, "unmapped ReadBlock @ 0x%08X (start address = 0x%08X, size = %zu)", current_vaddr, src_addr, size); std::memset(dest_buffer, 0, copy_amount); break; - case PageType::Special: { - if (ReadSpecialBlock(process, current_vaddr, dest_buffer, copy_amount)) - break; - [[fallthrough]]; } case PageType::Memory: { DEBUG_ASSERT(page_table.pointers[page_index]); @@ -352,6 +457,12 @@ void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_ std::memcpy(dest_buffer, src_ptr, copy_amount); break; } + case PageType::RasterizerCachedMemory: { + RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount), + FlushMode::Flush); + std::memcpy(dest_buffer, GetPointerFromVMA(process, current_vaddr), copy_amount); + break; + } default: UNREACHABLE(); } @@ -364,7 +475,7 @@ void ReadBlock(const Kernel::Process& process, const VAddr src_addr, void* dest_ } void ReadBlock(const VAddr src_addr, void* dest_buffer, const size_t size) { - ReadBlock(*Kernel::g_current_process, src_addr, dest_buffer, size); + ReadBlock(*Core::CurrentProcess(), src_addr, dest_buffer, size); } void Write8(const VAddr addr, const u8 data) { @@ -383,17 +494,6 @@ void Write64(const VAddr addr, const u64 data) { Write<u64_le>(addr, data); } -static bool WriteSpecialBlock(const Kernel::Process& process, const VAddr dest_addr, - const void* src_buffer, const size_t size) { - auto& page_table = process.vm_manager.page_table; - for (const auto& handler : GetSpecialHandlers(page_table, dest_addr, size)) { - if (handler->WriteBlock(dest_addr, src_buffer, size)) { - return true; - } - } - return false; -} - void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const void* src_buffer, const size_t size) { auto& page_table = process.vm_manager.page_table; @@ -402,20 +502,17 @@ void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const voi size_t page_offset = dest_addr & PAGE_MASK; while (remaining_size > 0) { - const size_t copy_amount = std::min<size_t>(PAGE_SIZE - page_offset, remaining_size); + const size_t copy_amount = + std::min(static_cast<size_t>(PAGE_SIZE) - page_offset, remaining_size); const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset); switch (page_table.attributes[page_index]) { - case PageType::Unmapped: + case PageType::Unmapped: { LOG_ERROR(HW_Memory, - "unmapped WriteBlock @ 0x%016" PRIX64 " (start address = 0x%016" PRIX64 - ", size = %zu)", + "unmapped WriteBlock @ 0x%08X (start address = 0x%08X, size = %zu)", current_vaddr, dest_addr, size); break; - case PageType::Special: - if (WriteSpecialBlock(process, current_vaddr, src_buffer, copy_amount)) - break; - [[fallthrough]]; + } case PageType::Memory: { DEBUG_ASSERT(page_table.pointers[page_index]); @@ -423,6 +520,12 @@ void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const voi std::memcpy(dest_ptr, src_buffer, copy_amount); break; } + case PageType::RasterizerCachedMemory: { + RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount), + FlushMode::Invalidate); + std::memcpy(GetPointerFromVMA(process, current_vaddr), src_buffer, copy_amount); + break; + } default: UNREACHABLE(); } @@ -435,12 +538,11 @@ void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const voi } void WriteBlock(const VAddr dest_addr, const void* src_buffer, const size_t size) { - WriteBlock(*Kernel::g_current_process, dest_addr, src_buffer, size); + WriteBlock(*Core::CurrentProcess(), dest_addr, src_buffer, size); } -void ZeroBlock(const VAddr dest_addr, const size_t size) { - const auto& process = *Kernel::g_current_process; - +void ZeroBlock(const Kernel::Process& process, const VAddr dest_addr, const size_t size) { + auto& page_table = process.vm_manager.page_table; size_t remaining_size = size; size_t page_index = dest_addr >> PAGE_BITS; size_t page_offset = dest_addr & PAGE_MASK; @@ -448,27 +550,29 @@ void ZeroBlock(const VAddr dest_addr, const size_t size) { static const std::array<u8, PAGE_SIZE> zeros = {}; while (remaining_size > 0) { - const size_t copy_amount = std::min<size_t>(PAGE_SIZE - page_offset, remaining_size); + const size_t copy_amount = + std::min(static_cast<size_t>(PAGE_SIZE) - page_offset, remaining_size); const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset); - switch (current_page_table->attributes[page_index]) { - case PageType::Unmapped: - LOG_ERROR(HW_Memory, - "unmapped ZeroBlock @ 0x%016" PRIX64 " (start address = 0x%016" PRIX64 - ", size = %zu)", + switch (page_table.attributes[page_index]) { + case PageType::Unmapped: { + LOG_ERROR(HW_Memory, "unmapped ZeroBlock @ 0x%08X (start address = 0x%08X, size = %zu)", current_vaddr, dest_addr, size); break; - case PageType::Special: - if (WriteSpecialBlock(process, current_vaddr, zeros.data(), copy_amount)) - break; - [[fallthrough]]; + } case PageType::Memory: { - DEBUG_ASSERT(current_page_table->pointers[page_index]); + DEBUG_ASSERT(page_table.pointers[page_index]); - u8* dest_ptr = current_page_table->pointers[page_index] + page_offset; + u8* dest_ptr = page_table.pointers[page_index] + page_offset; std::memset(dest_ptr, 0, copy_amount); break; } + case PageType::RasterizerCachedMemory: { + RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount), + FlushMode::Invalidate); + std::memset(GetPointerFromVMA(process, current_vaddr), 0, copy_amount); + break; + } default: UNREACHABLE(); } @@ -479,37 +583,34 @@ void ZeroBlock(const VAddr dest_addr, const size_t size) { } } -void CopyBlock(VAddr dest_addr, VAddr src_addr, const size_t size) { - const auto& process = *Kernel::g_current_process; - +void CopyBlock(const Kernel::Process& process, VAddr dest_addr, VAddr src_addr, const size_t size) { + auto& page_table = process.vm_manager.page_table; size_t remaining_size = size; size_t page_index = src_addr >> PAGE_BITS; size_t page_offset = src_addr & PAGE_MASK; while (remaining_size > 0) { - const size_t copy_amount = std::min<size_t>(PAGE_SIZE - page_offset, remaining_size); + const size_t copy_amount = + std::min(static_cast<size_t>(PAGE_SIZE) - page_offset, remaining_size); const VAddr current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset); - switch (current_page_table->attributes[page_index]) { - case PageType::Unmapped: - LOG_ERROR(HW_Memory, - "unmapped CopyBlock @ 0x%016" PRIX64 " (start address = 0x%016" PRIX64 - ", size = %zu)", + switch (page_table.attributes[page_index]) { + case PageType::Unmapped: { + LOG_ERROR(HW_Memory, "unmapped CopyBlock @ 0x%08X (start address = 0x%08X, size = %zu)", current_vaddr, src_addr, size); - ZeroBlock(dest_addr, copy_amount); + ZeroBlock(process, dest_addr, copy_amount); break; - case PageType::Special: { - std::vector<u8> buffer(copy_amount); - if (ReadSpecialBlock(process, current_vaddr, buffer.data(), buffer.size())) { - WriteBlock(dest_addr, buffer.data(), buffer.size()); - break; - } - [[fallthrough]]; } case PageType::Memory: { - DEBUG_ASSERT(current_page_table->pointers[page_index]); - const u8* src_ptr = current_page_table->pointers[page_index] + page_offset; - WriteBlock(dest_addr, src_ptr, copy_amount); + DEBUG_ASSERT(page_table.pointers[page_index]); + const u8* src_ptr = page_table.pointers[page_index] + page_offset; + WriteBlock(process, dest_addr, src_ptr, copy_amount); + break; + } + case PageType::RasterizerCachedMemory: { + RasterizerFlushVirtualRegion(current_vaddr, static_cast<u32>(copy_amount), + FlushMode::Flush); + WriteBlock(process, dest_addr, GetPointerFromVMA(process, current_vaddr), copy_amount); break; } default: @@ -524,78 +625,6 @@ void CopyBlock(VAddr dest_addr, VAddr src_addr, const size_t size) { } } -template <> -boost::optional<u8> ReadSpecial<u8>(VAddr addr) { - const PageTable& page_table = Kernel::g_current_process->vm_manager.page_table; - for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u8))) - if (auto result = handler->Read8(addr)) - return *result; - return {}; -} - -template <> -boost::optional<u16> ReadSpecial<u16>(VAddr addr) { - const PageTable& page_table = Kernel::g_current_process->vm_manager.page_table; - for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u16))) - if (auto result = handler->Read16(addr)) - return *result; - return {}; -} - -template <> -boost::optional<u32> ReadSpecial<u32>(VAddr addr) { - const PageTable& page_table = Kernel::g_current_process->vm_manager.page_table; - for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u32))) - if (auto result = handler->Read32(addr)) - return *result; - return {}; -} - -template <> -boost::optional<u64> ReadSpecial<u64>(VAddr addr) { - const PageTable& page_table = Kernel::g_current_process->vm_manager.page_table; - for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u64))) - if (auto result = handler->Read64(addr)) - return *result; - return {}; -} - -template <> -bool WriteSpecial<u8>(VAddr addr, const u8 data) { - const PageTable& page_table = Kernel::g_current_process->vm_manager.page_table; - for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u8))) - if (handler->Write8(addr, data)) - return true; - return false; -} - -template <> -bool WriteSpecial<u16>(VAddr addr, const u16 data) { - const PageTable& page_table = Kernel::g_current_process->vm_manager.page_table; - for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u16))) - if (handler->Write16(addr, data)) - return true; - return false; -} - -template <> -bool WriteSpecial<u32>(VAddr addr, const u32 data) { - const PageTable& page_table = Kernel::g_current_process->vm_manager.page_table; - for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u32))) - if (handler->Write32(addr, data)) - return true; - return false; -} - -template <> -bool WriteSpecial<u64>(VAddr addr, const u64 data) { - const PageTable& page_table = Kernel::g_current_process->vm_manager.page_table; - for (const auto& handler : GetSpecialHandlers(page_table, addr, sizeof(u64))) - if (handler->Write64(addr, data)) - return true; - return false; -} - boost::optional<PAddr> TryVirtualToPhysicalAddress(const VAddr addr) { if (addr == 0) { return 0; @@ -609,8 +638,6 @@ boost::optional<PAddr> TryVirtualToPhysicalAddress(const VAddr addr) { return addr - DSP_RAM_VADDR + DSP_RAM_PADDR; } else if (addr >= IO_AREA_VADDR && addr < IO_AREA_VADDR_END) { return addr - IO_AREA_VADDR + IO_AREA_PADDR; - } else if (addr >= N3DS_EXTRA_RAM_VADDR && addr < N3DS_EXTRA_RAM_VADDR_END) { - return addr - N3DS_EXTRA_RAM_VADDR + N3DS_EXTRA_RAM_PADDR; } return boost::none; @@ -632,13 +659,11 @@ boost::optional<VAddr> PhysicalToVirtualAddress(const PAddr addr) { } else if (addr >= VRAM_PADDR && addr < VRAM_PADDR_END) { return addr - VRAM_PADDR + VRAM_VADDR; } else if (addr >= FCRAM_PADDR && addr < FCRAM_PADDR_END) { - return addr - FCRAM_PADDR + Kernel::g_current_process->GetLinearHeapAreaAddress(); + return addr - FCRAM_PADDR + Core::CurrentProcess()->GetLinearHeapAreaAddress(); } else if (addr >= DSP_RAM_PADDR && addr < DSP_RAM_PADDR_END) { return addr - DSP_RAM_PADDR + DSP_RAM_VADDR; } else if (addr >= IO_AREA_PADDR && addr < IO_AREA_PADDR_END) { return addr - IO_AREA_PADDR + IO_AREA_VADDR; - } else if (addr >= N3DS_EXTRA_RAM_PADDR && addr < N3DS_EXTRA_RAM_PADDR_END) { - return addr - N3DS_EXTRA_RAM_PADDR + N3DS_EXTRA_RAM_VADDR; } return boost::none; diff --git a/src/core/memory.h b/src/core/memory.h index f3ace7a98..e9b8ca873 100644 --- a/src/core/memory.h +++ b/src/core/memory.h @@ -36,7 +36,10 @@ enum class PageType : u8 { Unmapped, /// Page is mapped to regular memory. This is the only type you can get pointers to. Memory, - /// Page is mapped to a memory hook, which intercepts read and write requests. + /// Page is mapped to regular memory, but also needs to check for rasterizer cache flushing and + /// invalidation + RasterizerCachedMemory, + /// Page is mapped to a I/O region. Writing and reading to this page is handled by functions. Special, }; @@ -98,12 +101,6 @@ enum : PAddr { VRAM_SIZE = 0x00600000, ///< VRAM size (6MB) VRAM_PADDR_END = VRAM_PADDR + VRAM_SIZE, - /// New 3DS additional memory. Supposedly faster than regular FCRAM. Part of it can be used by - /// applications and system modules if mapped via the ExHeader. - N3DS_EXTRA_RAM_PADDR = 0x1F000000, - N3DS_EXTRA_RAM_SIZE = 0x00400000, ///< New 3DS additional memory size (4MB) - N3DS_EXTRA_RAM_PADDR_END = N3DS_EXTRA_RAM_PADDR + N3DS_EXTRA_RAM_SIZE, - /// DSP memory DSP_RAM_PADDR = 0x1FF00000, DSP_RAM_SIZE = 0x00080000, ///< DSP memory size (512KB) @@ -119,7 +116,6 @@ enum : PAddr { FCRAM_SIZE = 0x08000000, ///< FCRAM size on the Old 3DS (128MB) FCRAM_N3DS_SIZE = 0x10000000, ///< FCRAM size on the New 3DS (256MB) FCRAM_PADDR_END = FCRAM_PADDR + FCRAM_SIZE, - FCRAM_N3DS_PADDR_END = FCRAM_PADDR + FCRAM_N3DS_SIZE, }; /// Virtual user-space memory regions @@ -129,31 +125,12 @@ enum : VAddr { PROCESS_IMAGE_MAX_SIZE = 0x08000000, PROCESS_IMAGE_VADDR_END = PROCESS_IMAGE_VADDR + PROCESS_IMAGE_MAX_SIZE, - /// Area where IPC buffers are mapped onto. - IPC_MAPPING_VADDR = 0x04000000, - IPC_MAPPING_SIZE = 0x04000000, - IPC_MAPPING_VADDR_END = IPC_MAPPING_VADDR + IPC_MAPPING_SIZE, - - /// Application heap (includes stack). - HEAP_VADDR = 0x108000000, - HEAP_SIZE = 0xF0000000, - HEAP_VADDR_END = HEAP_VADDR + HEAP_SIZE, - - /// Area where shared memory buffers are mapped onto. - SHARED_MEMORY_VADDR = 0x10000000, - SHARED_MEMORY_SIZE = 0x04000000, - SHARED_MEMORY_VADDR_END = SHARED_MEMORY_VADDR + SHARED_MEMORY_SIZE, - /// Maps 1:1 to an offset in FCRAM. Used for HW allocations that need to be linear in physical /// memory. LINEAR_HEAP_VADDR = 0x14000000, LINEAR_HEAP_SIZE = 0x08000000, LINEAR_HEAP_VADDR_END = LINEAR_HEAP_VADDR + LINEAR_HEAP_SIZE, - /// Maps 1:1 to New 3DS additional memory - N3DS_EXTRA_RAM_VADDR = 0x1E800000, - N3DS_EXTRA_RAM_VADDR_END = N3DS_EXTRA_RAM_VADDR + N3DS_EXTRA_RAM_SIZE, - /// Maps 1:1 to the IO register area. IO_AREA_VADDR = 0x1EC00000, IO_AREA_VADDR_END = IO_AREA_VADDR + IO_AREA_SIZE, @@ -176,14 +153,40 @@ enum : VAddr { SHARED_PAGE_SIZE = 0x00001000, SHARED_PAGE_VADDR_END = SHARED_PAGE_VADDR + SHARED_PAGE_SIZE, - /// Area where TLS (Thread-Local Storage) buffers are allocated. - TLS_AREA_VADDR = 0x228000000, - TLS_ENTRY_SIZE = 0x200, - /// Equivalent to LINEAR_HEAP_VADDR, but expanded to cover the extra memory in the New 3DS. NEW_LINEAR_HEAP_VADDR = 0x30000000, NEW_LINEAR_HEAP_SIZE = 0x10000000, NEW_LINEAR_HEAP_VADDR_END = NEW_LINEAR_HEAP_VADDR + NEW_LINEAR_HEAP_SIZE, + + /// Area where TLS (Thread-Local Storage) buffers are allocated. + TLS_AREA_VADDR = NEW_LINEAR_HEAP_VADDR_END, + TLS_ENTRY_SIZE = 0x200, + TLS_AREA_SIZE = 0x10000000, + TLS_AREA_VADDR_END = TLS_AREA_VADDR + TLS_AREA_SIZE, + + /// Application stack + STACK_AREA_VADDR = TLS_AREA_VADDR_END, + STACK_AREA_SIZE = 0x10000000, + STACK_AREA_VADDR_END = STACK_AREA_VADDR + STACK_AREA_SIZE, + DEFAULT_STACK_SIZE = 0x100000, + + /// Application heap + /// Size is confirmed to be a static value on fw 3.0.0 + HEAP_VADDR = 0x108000000, + HEAP_SIZE = 0x180000000, + HEAP_VADDR_END = HEAP_VADDR + HEAP_SIZE, + + /// New map region + /// Size is confirmed to be a static value on fw 3.0.0 + NEW_MAP_REGION_VADDR = HEAP_VADDR_END, + NEW_MAP_REGION_SIZE = 0x80000000, + NEW_MAP_REGION_VADDR_END = NEW_MAP_REGION_VADDR + NEW_MAP_REGION_SIZE, + + /// Map region + /// Size is confirmed to be a static value on fw 3.0.0 + MAP_REGION_VADDR = NEW_MAP_REGION_VADDR_END, + MAP_REGION_SIZE = 0x1000000000, + MAP_REGION_VADDR_END = MAP_REGION_VADDR + MAP_REGION_SIZE, }; /// Currently active page table @@ -243,4 +246,24 @@ boost::optional<VAddr> PhysicalToVirtualAddress(PAddr addr); */ u8* GetPhysicalPointer(PAddr address); +enum class FlushMode { + /// Write back modified surfaces to RAM + Flush, + /// Remove region from the cache + Invalidate, + /// Write back modified surfaces to RAM, and also remove them from the cache + FlushAndInvalidate, +}; + +/** + * Mark each page touching the region as cached. + */ +void RasterizerMarkRegionCached(VAddr start, u64 size, bool cached); + +/** + * Flushes and invalidates any externally cached rasterizer resources touching the given virtual + * address region. + */ +void RasterizerFlushVirtualRegion(VAddr start, u64 size, FlushMode mode); + } // namespace Memory diff --git a/src/core/settings.h b/src/core/settings.h index 6f8cd0f03..2c94caab7 100644 --- a/src/core/settings.h +++ b/src/core/settings.h @@ -105,12 +105,10 @@ static const std::array<const char*, NumAnalogs> mapping = {{ }}; } // namespace NativeAnalog -enum class CpuCore { - Unicorn, - Dynarmic, -}; - struct Values { + // System + bool use_docked_mode; + // Controls std::array<std::string, NativeButton::NumButtons> buttons; std::array<std::string, NativeAnalog::NumAnalogs> analogs; @@ -118,7 +116,7 @@ struct Values { std::string touch_device; // Core - CpuCore cpu_core; + bool use_cpu_jit; // Data Storage bool use_virtual_sd; diff --git a/src/core/telemetry_session.cpp b/src/core/telemetry_session.cpp index bea05a09b..cecf0a5cb 100644 --- a/src/core/telemetry_session.cpp +++ b/src/core/telemetry_session.cpp @@ -87,8 +87,8 @@ TelemetrySession::TelemetrySession() { #ifdef ENABLE_WEB_SERVICE if (Settings::values.enable_telemetry) { backend = std::make_unique<WebService::TelemetryJson>( - Settings::values.telemetry_endpoint_url, Settings::values.citra_username, - Settings::values.citra_token); + Settings::values.telemetry_endpoint_url, Settings::values.yuzu_username, + Settings::values.yuzu_token); } else { backend = std::make_unique<Telemetry::NullVisitor>(); } @@ -154,12 +154,13 @@ TelemetrySession::TelemetrySession() { #endif // Log user configuration information - AddField(Telemetry::FieldType::UserConfig, "Core_CpuCore", - static_cast<int>(Settings::values.cpu_core)); + AddField(Telemetry::FieldType::UserConfig, "Core_UseCpuJit", Settings::values.use_cpu_jit); AddField(Telemetry::FieldType::UserConfig, "Renderer_ResolutionFactor", Settings::values.resolution_factor); AddField(Telemetry::FieldType::UserConfig, "Renderer_ToggleFramelimit", Settings::values.toggle_framelimit); + AddField(Telemetry::FieldType::UserConfig, "System_UseDockedMode", + Settings::values.use_docked_mode); } TelemetrySession::~TelemetrySession() { diff --git a/src/core/telemetry_session.h b/src/core/telemetry_session.h index 550c6ea2d..dbc4f8bd4 100644 --- a/src/core/telemetry_session.h +++ b/src/core/telemetry_session.h @@ -50,8 +50,8 @@ u64 RegenerateTelemetryId(); /** * Verifies the username and token. - * @param username Citra username to use for authentication. - * @param token Citra token to use for authentication. + * @param username yuzu username to use for authentication. + * @param token yuzu token to use for authentication. * @param func A function that gets exectued when the verification is finished * @returns Future with bool indicating whether the verification succeeded */ diff --git a/src/tests/core/arm/arm_test_common.cpp b/src/tests/core/arm/arm_test_common.cpp index 88bbbc95c..7f9f27e19 100644 --- a/src/tests/core/arm/arm_test_common.cpp +++ b/src/tests/core/arm/arm_test_common.cpp @@ -15,8 +15,8 @@ static Memory::PageTable* page_table = nullptr; TestEnvironment::TestEnvironment(bool mutable_memory_) : mutable_memory(mutable_memory_), test_memory(std::make_shared<TestMemory>(this)) { - Kernel::g_current_process = Kernel::Process::Create(""); - page_table = &Kernel::g_current_process->vm_manager.page_table; + Core::CurrentProcess() = Kernel::Process::Create(""); + page_table = &Core::CurrentProcess()->vm_manager.page_table; page_table->pointers.fill(nullptr); page_table->special_regions.clear(); diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt index ed87f8ff1..a710c4bc5 100644 --- a/src/video_core/CMakeLists.txt +++ b/src/video_core/CMakeLists.txt @@ -1,24 +1,44 @@ add_library(video_core STATIC command_processor.cpp command_processor.h + debug_utils/debug_utils.cpp + debug_utils/debug_utils.h engines/fermi_2d.cpp engines/fermi_2d.h engines/maxwell_3d.cpp engines/maxwell_3d.h engines/maxwell_compute.cpp engines/maxwell_compute.h + gpu.cpp gpu.h + macro_interpreter.cpp + macro_interpreter.h memory_manager.cpp memory_manager.h + rasterizer_interface.h renderer_base.cpp renderer_base.h + renderer_opengl/gl_rasterizer.cpp + renderer_opengl/gl_rasterizer.h + renderer_opengl/gl_rasterizer_cache.cpp + renderer_opengl/gl_rasterizer_cache.h renderer_opengl/gl_resource_manager.h + renderer_opengl/gl_shader_decompiler.cpp + renderer_opengl/gl_shader_decompiler.h + renderer_opengl/gl_shader_gen.cpp + renderer_opengl/gl_shader_gen.h renderer_opengl/gl_shader_util.cpp renderer_opengl/gl_shader_util.h renderer_opengl/gl_state.cpp renderer_opengl/gl_state.h + renderer_opengl/gl_stream_buffer.cpp + renderer_opengl/gl_stream_buffer.h + renderer_opengl/maxwell_to_gl.h renderer_opengl/renderer_opengl.cpp renderer_opengl/renderer_opengl.h + textures/decoders.cpp + textures/decoders.h + textures/texture.h utils.h video_core.cpp video_core.h diff --git a/src/video_core/command_processor.cpp b/src/video_core/command_processor.cpp index 21d672085..d4cdb4ab2 100644 --- a/src/video_core/command_processor.cpp +++ b/src/video_core/command_processor.cpp @@ -24,12 +24,37 @@ namespace Tegra { enum class BufferMethods { BindObject = 0, + SetGraphMacroCode = 0x45, + SetGraphMacroCodeArg = 0x46, + SetGraphMacroEntry = 0x47, CountBufferMethods = 0x100, }; -void GPU::WriteReg(u32 method, u32 subchannel, u32 value) { - LOG_WARNING(HW_GPU, "Processing method %08X on subchannel %u value %08X", method, subchannel, - value); +void GPU::WriteReg(u32 method, u32 subchannel, u32 value, u32 remaining_params) { + LOG_WARNING(HW_GPU, "Processing method %08X on subchannel %u value %08X remaining params %u", + method, subchannel, value, remaining_params); + + if (method == static_cast<u32>(BufferMethods::SetGraphMacroEntry)) { + // Prepare to upload a new macro, reset the upload counter. + LOG_DEBUG(HW_GPU, "Uploading GPU macro %08X", value); + current_macro_entry = value; + current_macro_code.clear(); + return; + } + + if (method == static_cast<u32>(BufferMethods::SetGraphMacroCodeArg)) { + // Append a new code word to the current macro. + current_macro_code.push_back(value); + + // There are no more params remaining, submit the code to the 3D engine. + if (remaining_params == 0) { + maxwell_3d->SubmitMacroCode(current_macro_entry, std::move(current_macro_code)); + current_macro_entry = InvalidGraphMacroEntry; + current_macro_code.clear(); + } + + return; + } if (method == static_cast<u32>(BufferMethods::BindObject)) { // Bind the current subchannel to the desired engine id. @@ -54,7 +79,7 @@ void GPU::WriteReg(u32 method, u32 subchannel, u32 value) { fermi_2d->WriteReg(method, value); break; case EngineID::MAXWELL_B: - maxwell_3d->WriteReg(method, value); + maxwell_3d->WriteReg(method, value, remaining_params); break; case EngineID::MAXWELL_COMPUTE_B: maxwell_compute->WriteReg(method, value); @@ -78,7 +103,8 @@ void GPU::ProcessCommandList(GPUVAddr address, u32 size) { 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)); + WriteReg(header.method + i, header.subchannel, Memory::Read32(current_addr), + header.arg_count - i - 1); current_addr += sizeof(u32); } break; @@ -87,27 +113,31 @@ void GPU::ProcessCommandList(GPUVAddr address, u32 size) { 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)); + 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)); + WriteReg(header.method, header.subchannel, Memory::Read32(current_addr), + header.arg_count - 1); current_addr += sizeof(u32); - // Use the same method value for all arguments. + for (unsigned i = 1; i < header.arg_count; ++i) { - WriteReg(header.method + 1, header.subchannel, Memory::Read32(current_addr)); + 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); + WriteReg(header.method, header.subchannel, header.inline_data, 0); break; } default: diff --git a/src/video_core/command_processor.h b/src/video_core/command_processor.h index b511bfcf7..f7214ffec 100644 --- a/src/video_core/command_processor.h +++ b/src/video_core/command_processor.h @@ -34,6 +34,4 @@ static_assert(std::is_standard_layout<CommandHeader>::value == true, "CommandHeader does not use standard layout"); static_assert(sizeof(CommandHeader) == sizeof(u32), "CommandHeader has incorrect size!"); -void ProcessCommandList(VAddr address, u32 size); - } // namespace Tegra diff --git a/src/video_core/debug_utils/debug_utils.cpp b/src/video_core/debug_utils/debug_utils.cpp new file mode 100644 index 000000000..22d44aab2 --- /dev/null +++ b/src/video_core/debug_utils/debug_utils.cpp @@ -0,0 +1,64 @@ +// Copyright 2014 Citra Emulator Project +// Licensed under GPLv2 +// Refer to the license.txt file included. + +#include <algorithm> +#include <condition_variable> +#include <cstdint> +#include <cstring> +#include <fstream> +#include <map> +#include <mutex> +#include <string> + +#include "common/assert.h" +#include "common/bit_field.h" +#include "common/color.h" +#include "common/common_types.h" +#include "common/file_util.h" +#include "common/logging/log.h" +#include "common/math_util.h" +#include "common/vector_math.h" +#include "video_core/debug_utils/debug_utils.h" + +namespace Tegra { + +void DebugContext::DoOnEvent(Event event, void* data) { + { + std::unique_lock<std::mutex> lock(breakpoint_mutex); + + // TODO(Subv): Commit the rasterizer's caches so framebuffers, render targets, etc. will + // show on debug widgets + + // TODO: Should stop the CPU thread here once we multithread emulation. + + active_breakpoint = event; + at_breakpoint = true; + + // Tell all observers that we hit a breakpoint + for (auto& breakpoint_observer : breakpoint_observers) { + breakpoint_observer->OnMaxwellBreakPointHit(event, data); + } + + // Wait until another thread tells us to Resume() + resume_from_breakpoint.wait(lock, [&] { return !at_breakpoint; }); + } +} + +void DebugContext::Resume() { + { + std::lock_guard<std::mutex> lock(breakpoint_mutex); + + // Tell all observers that we are about to resume + for (auto& breakpoint_observer : breakpoint_observers) { + breakpoint_observer->OnMaxwellResume(); + } + + // Resume the waiting thread (i.e. OnEvent()) + at_breakpoint = false; + } + + resume_from_breakpoint.notify_one(); +} + +} // namespace Tegra diff --git a/src/video_core/debug_utils/debug_utils.h b/src/video_core/debug_utils/debug_utils.h new file mode 100644 index 000000000..bbba8e380 --- /dev/null +++ b/src/video_core/debug_utils/debug_utils.h @@ -0,0 +1,163 @@ +// Copyright 2014 Citra Emulator Project +// Licensed under GPLv2 +// Refer to the license.txt file included. + +#pragma once + +#include <algorithm> +#include <array> +#include <condition_variable> +#include <iterator> +#include <list> +#include <map> +#include <memory> +#include <mutex> +#include <string> +#include <utility> +#include <vector> +#include "common/common_types.h" +#include "common/vector_math.h" + +namespace Tegra { + +class DebugContext { +public: + enum class Event { + FirstEvent = 0, + + MaxwellCommandLoaded = FirstEvent, + MaxwellCommandProcessed, + IncomingPrimitiveBatch, + FinishedPrimitiveBatch, + + NumEvents + }; + + /** + * Inherit from this class to be notified of events registered to some debug context. + * Most importantly this is used for our debugger GUI. + * + * To implement event handling, override the OnMaxwellBreakPointHit and OnMaxwellResume methods. + * @warning All BreakPointObservers need to be on the same thread to guarantee thread-safe state + * access + * @todo Evaluate an alternative interface, in which there is only one managing observer and + * multiple child observers running (by design) on the same thread. + */ + class BreakPointObserver { + public: + /// Constructs the object such that it observes events of the given DebugContext. + BreakPointObserver(std::shared_ptr<DebugContext> debug_context) + : context_weak(debug_context) { + std::unique_lock<std::mutex> lock(debug_context->breakpoint_mutex); + debug_context->breakpoint_observers.push_back(this); + } + + virtual ~BreakPointObserver() { + auto context = context_weak.lock(); + if (context) { + std::unique_lock<std::mutex> lock(context->breakpoint_mutex); + context->breakpoint_observers.remove(this); + + // If we are the last observer to be destroyed, tell the debugger context that + // it is free to continue. In particular, this is required for a proper yuzu + // shutdown, when the emulation thread is waiting at a breakpoint. + if (context->breakpoint_observers.empty()) + context->Resume(); + } + } + + /** + * Action to perform when a breakpoint was reached. + * @param event Type of event which triggered the breakpoint + * @param data Optional data pointer (if unused, this is a nullptr) + * @note This function will perform nothing unless it is overridden in the child class. + */ + virtual void OnMaxwellBreakPointHit(Event event, void* data) {} + + /** + * Action to perform when emulation is resumed from a breakpoint. + * @note This function will perform nothing unless it is overridden in the child class. + */ + virtual void OnMaxwellResume() {} + + protected: + /** + * Weak context pointer. This need not be valid, so when requesting a shared_ptr via + * context_weak.lock(), always compare the result against nullptr. + */ + std::weak_ptr<DebugContext> context_weak; + }; + + /** + * Simple structure defining a breakpoint state + */ + struct BreakPoint { + bool enabled = false; + }; + + /** + * Static constructor used to create a shared_ptr of a DebugContext. + */ + static std::shared_ptr<DebugContext> Construct() { + return std::shared_ptr<DebugContext>(new DebugContext); + } + + /** + * Used by the emulation core when a given event has happened. If a breakpoint has been set + * for this event, OnEvent calls the event handlers of the registered breakpoint observers. + * The current thread then is halted until Resume() is called from another thread (or until + * emulation is stopped). + * @param event Event which has happened + * @param data Optional data pointer (pass nullptr if unused). Needs to remain valid until + * Resume() is called. + */ + void OnEvent(Event event, void* data) { + // This check is left in the header to allow the compiler to inline it. + if (!breakpoints[(int)event].enabled) + return; + // For the rest of event handling, call a separate function. + DoOnEvent(event, data); + } + + void DoOnEvent(Event event, void* data); + + /** + * Resume from the current breakpoint. + * @warning Calling this from the same thread that OnEvent was called in will cause a deadlock. + * Calling from any other thread is safe. + */ + void Resume(); + + /** + * Delete all set breakpoints and resume emulation. + */ + void ClearBreakpoints() { + for (auto& bp : breakpoints) { + bp.enabled = false; + } + Resume(); + } + + // TODO: Evaluate if access to these members should be hidden behind a public interface. + std::array<BreakPoint, (int)Event::NumEvents> breakpoints; + Event active_breakpoint; + bool at_breakpoint = false; + +private: + /** + * Private default constructor to make sure people always construct this through Construct() + * instead. + */ + DebugContext() = default; + + /// Mutex protecting current breakpoint state and the observer list. + std::mutex breakpoint_mutex; + + /// Used by OnEvent to wait for resumption. + std::condition_variable resume_from_breakpoint; + + /// List of registered observers + std::list<BreakPointObserver*> breakpoint_observers; +}; + +} // namespace Tegra diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp index 9f699399f..124753032 100644 --- a/src/video_core/engines/maxwell_3d.cpp +++ b/src/video_core/engines/maxwell_3d.cpp @@ -2,23 +2,132 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <cinttypes> #include "common/assert.h" +#include "core/core.h" +#include "video_core/debug_utils/debug_utils.h" #include "video_core/engines/maxwell_3d.h" +#include "video_core/rasterizer_interface.h" +#include "video_core/renderer_base.h" +#include "video_core/textures/decoders.h" +#include "video_core/textures/texture.h" +#include "video_core/video_core.h" namespace Tegra { namespace Engines { -Maxwell3D::Maxwell3D(MemoryManager& memory_manager) : memory_manager(memory_manager) {} +/// First register id that is actually a Macro call. +constexpr u32 MacroRegistersStart = 0xE00; -void Maxwell3D::WriteReg(u32 method, u32 value) { +Maxwell3D::Maxwell3D(MemoryManager& memory_manager) + : memory_manager(memory_manager), macro_interpreter(*this) {} + +void Maxwell3D::SubmitMacroCode(u32 entry, std::vector<u32> code) { + uploaded_macros[entry * 2 + MacroRegistersStart] = std::move(code); +} + +void Maxwell3D::CallMacroMethod(u32 method, std::vector<u32> parameters) { + auto macro_code = uploaded_macros.find(method); + // The requested macro must have been uploaded already. + ASSERT_MSG(macro_code != uploaded_macros.end(), "Macro %08X was not uploaded", method); + + // Reset the current macro and execute it. + executing_macro = 0; + macro_interpreter.Execute(macro_code->second, std::move(parameters)); +} + +void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) { ASSERT_MSG(method < Regs::NUM_REGS, "Invalid Maxwell3D register, increase the size of the Regs structure"); + auto debug_context = Core::System::GetInstance().GetGPUDebugContext(); + + // It is an error to write to a register other than the current macro's ARG register before it + // has finished execution. + if (executing_macro != 0) { + ASSERT(method == executing_macro + 1); + } + + // Methods after 0xE00 are special, they're actually triggers for some microcode that was + // uploaded to the GPU during initialization. + if (method >= MacroRegistersStart) { + // We're trying to execute a macro + if (executing_macro == 0) { + // A macro call must begin by writing the macro method's register, not its argument. + ASSERT_MSG((method % 2) == 0, + "Can't start macro execution by writing to the ARGS register"); + executing_macro = method; + } + + macro_params.push_back(value); + + // Call the macro when there are no more parameters in the command buffer + if (remaining_params == 0) { + CallMacroMethod(executing_macro, std::move(macro_params)); + } + return; + } + + if (debug_context) { + debug_context->OnEvent(Tegra::DebugContext::Event::MaxwellCommandLoaded, nullptr); + } + regs.reg_array[method] = value; #define MAXWELL3D_REG_INDEX(field_name) (offsetof(Regs, field_name) / sizeof(u32)) switch (method) { + case MAXWELL3D_REG_INDEX(code_address.code_address_high): + case MAXWELL3D_REG_INDEX(code_address.code_address_low): { + // Note: For some reason games (like Puyo Puyo Tetris) seem to write 0 to the CODE_ADDRESS + // register, we do not currently know if that's intended or a bug, so we assert it lest + // stuff breaks in other places (like the shader address calculation). + ASSERT_MSG(regs.code_address.CodeAddress() == 0, "Unexpected CODE_ADDRESS register value."); + break; + } + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[0]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[1]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[2]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[3]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[4]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[5]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[6]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[7]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[8]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[9]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[10]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[11]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[12]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[13]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[14]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[15]): { + ProcessCBData(value); + break; + } + case MAXWELL3D_REG_INDEX(cb_bind[0].raw_config): { + ProcessCBBind(Regs::ShaderStage::Vertex); + break; + } + case MAXWELL3D_REG_INDEX(cb_bind[1].raw_config): { + ProcessCBBind(Regs::ShaderStage::TesselationControl); + break; + } + case MAXWELL3D_REG_INDEX(cb_bind[2].raw_config): { + ProcessCBBind(Regs::ShaderStage::TesselationEval); + break; + } + case MAXWELL3D_REG_INDEX(cb_bind[3].raw_config): { + ProcessCBBind(Regs::ShaderStage::Geometry); + break; + } + case MAXWELL3D_REG_INDEX(cb_bind[4].raw_config): { + ProcessCBBind(Regs::ShaderStage::Fragment); + break; + } + case MAXWELL3D_REG_INDEX(draw.vertex_end_gl): { + DrawArrays(); + break; + } case MAXWELL3D_REG_INDEX(query.query_get): { ProcessQueryGet(); break; @@ -28,6 +137,10 @@ void Maxwell3D::WriteReg(u32 method, u32 value) { } #undef MAXWELL3D_REG_INDEX + + if (debug_context) { + debug_context->OnEvent(Tegra::DebugContext::Event::MaxwellCommandProcessed, nullptr); + } } void Maxwell3D::ProcessQueryGet() { @@ -44,8 +157,147 @@ void Maxwell3D::ProcessQueryGet() { break; } default: - UNIMPLEMENTED_MSG("Query mode %u not implemented", regs.query.query_get.mode.Value()); + UNIMPLEMENTED_MSG("Query mode %u not implemented", + static_cast<u32>(regs.query.query_get.mode.Value())); + } +} + +void Maxwell3D::DrawArrays() { + LOG_DEBUG(HW_GPU, "called, topology=%d, count=%d", regs.draw.topology.Value(), + regs.vertex_buffer.count); + + auto debug_context = Core::System::GetInstance().GetGPUDebugContext(); + + if (debug_context) { + debug_context->OnEvent(Tegra::DebugContext::Event::IncomingPrimitiveBatch, nullptr); + } + + if (debug_context) { + debug_context->OnEvent(Tegra::DebugContext::Event::FinishedPrimitiveBatch, nullptr); + } + + VideoCore::g_renderer->Rasterizer()->AccelerateDrawBatch(false /*is_indexed*/); +} + +void Maxwell3D::ProcessCBBind(Regs::ShaderStage stage) { + // Bind the buffer currently in CB_ADDRESS to the specified index in the desired shader stage. + auto& shader = state.shader_stages[static_cast<size_t>(stage)]; + auto& bind_data = regs.cb_bind[static_cast<size_t>(stage)]; + + auto& buffer = shader.const_buffers[bind_data.index]; + + buffer.enabled = bind_data.valid.Value() != 0; + buffer.index = bind_data.index; + buffer.address = regs.const_buffer.BufferAddress(); + buffer.size = regs.const_buffer.cb_size; +} + +void Maxwell3D::ProcessCBData(u32 value) { + // Write the input value to the current const buffer at the current position. + GPUVAddr buffer_address = regs.const_buffer.BufferAddress(); + ASSERT(buffer_address != 0); + + // Don't allow writing past the end of the buffer. + ASSERT(regs.const_buffer.cb_pos + sizeof(u32) <= regs.const_buffer.cb_size); + + VAddr address = + memory_manager.PhysicalToVirtualAddress(buffer_address + regs.const_buffer.cb_pos); + + Memory::Write32(address, value); + + // Increment the current buffer position. + regs.const_buffer.cb_pos = regs.const_buffer.cb_pos + 4; +} + +Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const { + GPUVAddr tic_base_address = regs.tic.TICAddress(); + + GPUVAddr tic_address_gpu = tic_base_address + tic_index * sizeof(Texture::TICEntry); + VAddr tic_address_cpu = memory_manager.PhysicalToVirtualAddress(tic_address_gpu); + + Texture::TICEntry tic_entry; + Memory::ReadBlock(tic_address_cpu, &tic_entry, sizeof(Texture::TICEntry)); + + ASSERT_MSG(tic_entry.header_version == Texture::TICHeaderVersion::BlockLinear, + "TIC versions other than BlockLinear are unimplemented"); + + ASSERT_MSG(tic_entry.texture_type == Texture::TextureType::Texture2D, + "Texture types other than Texture2D are unimplemented"); + + auto r_type = tic_entry.r_type.Value(); + auto g_type = tic_entry.g_type.Value(); + auto b_type = tic_entry.b_type.Value(); + auto a_type = tic_entry.a_type.Value(); + + // TODO(Subv): Different data types for separate components are not supported + ASSERT(r_type == g_type && r_type == b_type && r_type == a_type); + + return tic_entry; +} + +Texture::TSCEntry Maxwell3D::GetTSCEntry(u32 tsc_index) const { + GPUVAddr tsc_base_address = regs.tsc.TSCAddress(); + + GPUVAddr tsc_address_gpu = tsc_base_address + tsc_index * sizeof(Texture::TSCEntry); + VAddr tsc_address_cpu = memory_manager.PhysicalToVirtualAddress(tsc_address_gpu); + + Texture::TSCEntry tsc_entry; + Memory::ReadBlock(tsc_address_cpu, &tsc_entry, sizeof(Texture::TSCEntry)); + return tsc_entry; +} + +std::vector<Texture::FullTextureInfo> Maxwell3D::GetStageTextures(Regs::ShaderStage stage) const { + std::vector<Texture::FullTextureInfo> textures; + + auto& fragment_shader = state.shader_stages[static_cast<size_t>(stage)]; + auto& tex_info_buffer = fragment_shader.const_buffers[regs.tex_cb_index]; + ASSERT(tex_info_buffer.enabled && tex_info_buffer.address != 0); + + GPUVAddr tic_base_address = regs.tic.TICAddress(); + + GPUVAddr tex_info_buffer_end = tex_info_buffer.address + tex_info_buffer.size; + + // Offset into the texture constbuffer where the texture info begins. + static constexpr size_t TextureInfoOffset = 0x20; + + for (GPUVAddr current_texture = tex_info_buffer.address + TextureInfoOffset; + current_texture < tex_info_buffer_end; current_texture += sizeof(Texture::TextureHandle)) { + + Texture::TextureHandle tex_handle{ + Memory::Read32(memory_manager.PhysicalToVirtualAddress(current_texture))}; + + Texture::FullTextureInfo tex_info{}; + // TODO(Subv): Use the shader to determine which textures are actually accessed. + tex_info.index = (current_texture - tex_info_buffer.address - TextureInfoOffset) / + sizeof(Texture::TextureHandle); + + // Load the TIC data. + if (tex_handle.tic_id != 0) { + tex_info.enabled = true; + + auto tic_entry = GetTICEntry(tex_handle.tic_id); + // TODO(Subv): Workaround for BitField's move constructor being deleted. + std::memcpy(&tex_info.tic, &tic_entry, sizeof(tic_entry)); + } + + // Load the TSC data + if (tex_handle.tsc_id != 0) { + auto tsc_entry = GetTSCEntry(tex_handle.tsc_id); + // TODO(Subv): Workaround for BitField's move constructor being deleted. + std::memcpy(&tex_info.tsc, &tsc_entry, sizeof(tsc_entry)); + } + + if (tex_info.enabled) + textures.push_back(tex_info); } + + return textures; } + +u32 Maxwell3D::GetRegisterValue(u32 method) const { + ASSERT_MSG(method < Regs::NUM_REGS, "Invalid Maxwell3D register"); + return regs.reg_array[method]; +} + } // namespace Engines } // namespace Tegra diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h index 1eeef6857..98b39b2ff 100644 --- a/src/video_core/engines/maxwell_3d.h +++ b/src/video_core/engines/maxwell_3d.h @@ -4,10 +4,18 @@ #pragma once +#include <array> +#include <unordered_map> +#include <vector> +#include "common/assert.h" #include "common/bit_field.h" #include "common/common_funcs.h" #include "common/common_types.h" +#include "common/math_util.h" +#include "video_core/gpu.h" +#include "video_core/macro_interpreter.h" #include "video_core/memory_manager.h" +#include "video_core/textures/texture.h" namespace Tegra { namespace Engines { @@ -17,22 +25,357 @@ public: explicit Maxwell3D(MemoryManager& memory_manager); ~Maxwell3D() = default; - /// Write the value to the register identified by method. - void WriteReg(u32 method, u32 value); - /// Register structure of the Maxwell3D engine. /// TODO(Subv): This structure will need to be made bigger as more registers are discovered. struct Regs { static constexpr size_t NUM_REGS = 0xE36; + static constexpr size_t NumRenderTargets = 8; + static constexpr size_t NumViewports = 16; + static constexpr size_t NumCBData = 16; + static constexpr size_t NumVertexArrays = 32; + static constexpr size_t NumVertexAttributes = 32; + static constexpr size_t MaxShaderProgram = 6; + static constexpr size_t MaxShaderStage = 5; + // Maximum number of const buffers per shader stage. + static constexpr size_t MaxConstBuffers = 16; + enum class QueryMode : u32 { Write = 0, Sync = 1, }; + enum class ShaderProgram : u32 { + VertexA = 0, + VertexB = 1, + TesselationControl = 2, + TesselationEval = 3, + Geometry = 4, + Fragment = 5, + }; + + enum class ShaderStage : u32 { + Vertex = 0, + TesselationControl = 1, + TesselationEval = 2, + Geometry = 3, + Fragment = 4, + }; + + struct VertexAttribute { + enum class Size : u32 { + Size_32_32_32_32 = 0x01, + Size_32_32_32 = 0x02, + Size_16_16_16_16 = 0x03, + Size_32_32 = 0x04, + Size_16_16_16 = 0x05, + Size_8_8_8_8 = 0x0a, + Size_16_16 = 0x0f, + Size_32 = 0x12, + Size_8_8_8 = 0x13, + Size_8_8 = 0x18, + Size_16 = 0x1b, + Size_8 = 0x1d, + Size_10_10_10_2 = 0x30, + Size_11_11_10 = 0x31, + }; + + enum class Type : u32 { + SignedNorm = 1, + UnsignedNorm = 2, + SignedInt = 3, + UnsignedInt = 4, + UnsignedScaled = 5, + SignedScaled = 6, + Float = 7, + }; + + union { + BitField<0, 5, u32> buffer; + BitField<6, 1, u32> constant; + BitField<7, 14, u32> offset; + BitField<21, 6, Size> size; + BitField<27, 3, Type> type; + BitField<31, 1, u32> bgra; + }; + + u32 ComponentCount() const { + switch (size) { + case Size::Size_32_32_32_32: + return 4; + case Size::Size_32_32_32: + return 3; + case Size::Size_16_16_16_16: + return 4; + case Size::Size_32_32: + return 2; + case Size::Size_16_16_16: + return 3; + case Size::Size_8_8_8_8: + return 4; + case Size::Size_16_16: + return 2; + case Size::Size_32: + return 1; + case Size::Size_8_8_8: + return 3; + case Size::Size_8_8: + return 2; + case Size::Size_16: + return 1; + case Size::Size_8: + return 1; + case Size::Size_10_10_10_2: + return 4; + case Size::Size_11_11_10: + return 3; + default: + UNREACHABLE(); + } + } + + u32 SizeInBytes() const { + switch (size) { + case Size::Size_32_32_32_32: + return 16; + case Size::Size_32_32_32: + return 12; + case Size::Size_16_16_16_16: + return 8; + case Size::Size_32_32: + return 8; + case Size::Size_16_16_16: + return 6; + case Size::Size_8_8_8_8: + return 4; + case Size::Size_16_16: + return 4; + case Size::Size_32: + return 4; + case Size::Size_8_8_8: + return 3; + case Size::Size_8_8: + return 2; + case Size::Size_16: + return 2; + case Size::Size_8: + return 1; + case Size::Size_10_10_10_2: + return 4; + case Size::Size_11_11_10: + return 4; + default: + UNREACHABLE(); + } + } + + std::string SizeString() const { + switch (size) { + case Size::Size_32_32_32_32: + return "32_32_32_32"; + case Size::Size_32_32_32: + return "32_32_32"; + case Size::Size_16_16_16_16: + return "16_16_16_16"; + case Size::Size_32_32: + return "32_32"; + case Size::Size_16_16_16: + return "16_16_16"; + case Size::Size_8_8_8_8: + return "8_8_8_8"; + case Size::Size_16_16: + return "16_16"; + case Size::Size_32: + return "32"; + case Size::Size_8_8_8: + return "8_8_8"; + case Size::Size_8_8: + return "8_8"; + case Size::Size_16: + return "16"; + case Size::Size_8: + return "8"; + case Size::Size_10_10_10_2: + return "10_10_10_2"; + case Size::Size_11_11_10: + return "11_11_10"; + } + UNREACHABLE(); + return {}; + } + + std::string TypeString() const { + switch (type) { + case Type::SignedNorm: + return "SNORM"; + case Type::UnsignedNorm: + return "UNORM"; + case Type::SignedInt: + return "SINT"; + case Type::UnsignedInt: + return "UINT"; + case Type::UnsignedScaled: + return "USCALED"; + case Type::SignedScaled: + return "SSCALED"; + case Type::Float: + return "FLOAT"; + } + UNREACHABLE(); + return {}; + } + + bool IsNormalized() const { + return (type == Type::SignedNorm) || (type == Type::UnsignedNorm); + } + }; + + enum class PrimitiveTopology : u32 { + Points = 0x0, + Lines = 0x1, + LineLoop = 0x2, + LineStrip = 0x3, + Triangles = 0x4, + TriangleStrip = 0x5, + TriangleFan = 0x6, + Quads = 0x7, + QuadStrip = 0x8, + Polygon = 0x9, + LinesAdjacency = 0xa, + LineStripAdjacency = 0xb, + TrianglesAdjacency = 0xc, + TriangleStripAdjacency = 0xd, + Patches = 0xe, + }; + union { struct { - INSERT_PADDING_WORDS(0x6C0); + INSERT_PADDING_WORDS(0x200); + + struct { + u32 address_high; + u32 address_low; + u32 width; + u32 height; + Tegra::RenderTargetFormat format; + u32 block_dimensions; + u32 array_mode; + u32 layer_stride; + u32 base_layer; + INSERT_PADDING_WORDS(7); + + GPUVAddr Address() const { + return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) | + address_low); + } + } rt[NumRenderTargets]; + + INSERT_PADDING_WORDS(0x80); + + struct { + union { + BitField<0, 16, u32> x; + BitField<16, 16, u32> width; + }; + union { + BitField<0, 16, u32> y; + BitField<16, 16, u32> height; + }; + float depth_range_near; + float depth_range_far; + + MathUtil::Rectangle<s32> GetRect() const { + return { + static_cast<s32>(x), // left + static_cast<s32>(y + height), // top + static_cast<s32>(x + width), // right + static_cast<s32>(y) // bottom + }; + }; + } viewport[NumViewports]; + + INSERT_PADDING_WORDS(0x1D); + + struct { + u32 first; + u32 count; + } vertex_buffer; + + INSERT_PADDING_WORDS(0x99); + + struct { + u32 address_high; + u32 address_low; + u32 format; + u32 block_dimensions; + u32 layer_stride; + + GPUVAddr Address() const { + return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) | + address_low); + } + } zeta; + + INSERT_PADDING_WORDS(0x5B); + + VertexAttribute vertex_attrib_format[NumVertexAttributes]; + + INSERT_PADDING_WORDS(0xF); + + struct { + union { + BitField<0, 4, u32> count; + }; + } rt_control; + + INSERT_PADDING_WORDS(0xCF); + + struct { + u32 tsc_address_high; + u32 tsc_address_low; + u32 tsc_limit; + + GPUVAddr TSCAddress() const { + return static_cast<GPUVAddr>( + (static_cast<GPUVAddr>(tsc_address_high) << 32) | tsc_address_low); + } + } tsc; + + INSERT_PADDING_WORDS(0x3); + + struct { + u32 tic_address_high; + u32 tic_address_low; + u32 tic_limit; + + GPUVAddr TICAddress() const { + return static_cast<GPUVAddr>( + (static_cast<GPUVAddr>(tic_address_high) << 32) | tic_address_low); + } + } tic; + + INSERT_PADDING_WORDS(0x22); + + struct { + u32 code_address_high; + u32 code_address_low; + + GPUVAddr CodeAddress() const { + return static_cast<GPUVAddr>( + (static_cast<GPUVAddr>(code_address_high) << 32) | code_address_low); + } + } code_address; + INSERT_PADDING_WORDS(1); + + struct { + u32 vertex_end_gl; + union { + u32 vertex_begin_gl; + BitField<0, 16, PrimitiveTopology> topology; + }; + } draw; + + INSERT_PADDING_WORDS(0x139); struct { u32 query_address_high; u32 query_address_low; @@ -49,7 +392,98 @@ public: (static_cast<GPUVAddr>(query_address_high) << 32) | query_address_low); } } query; - INSERT_PADDING_WORDS(0x772); + + INSERT_PADDING_WORDS(0x3C); + + struct { + union { + BitField<0, 12, u32> stride; + BitField<12, 1, u32> enable; + }; + u32 start_high; + u32 start_low; + u32 divisor; + + GPUVAddr StartAddress() const { + return static_cast<GPUVAddr>((static_cast<GPUVAddr>(start_high) << 32) | + start_low); + } + } vertex_array[NumVertexArrays]; + + INSERT_PADDING_WORDS(0x40); + + struct { + u32 limit_high; + u32 limit_low; + + GPUVAddr LimitAddress() const { + return static_cast<GPUVAddr>((static_cast<GPUVAddr>(limit_high) << 32) | + limit_low); + } + } vertex_array_limit[NumVertexArrays]; + + struct { + union { + BitField<0, 1, u32> enable; + BitField<4, 4, ShaderProgram> program; + }; + u32 start_id; + INSERT_PADDING_WORDS(1); + u32 gpr_alloc; + ShaderStage type; + INSERT_PADDING_WORDS(9); + } shader_config[MaxShaderProgram]; + + INSERT_PADDING_WORDS(0x8C); + + struct { + u32 cb_size; + u32 cb_address_high; + u32 cb_address_low; + u32 cb_pos; + u32 cb_data[NumCBData]; + + GPUVAddr BufferAddress() const { + return static_cast<GPUVAddr>( + (static_cast<GPUVAddr>(cb_address_high) << 32) | cb_address_low); + } + } const_buffer; + + INSERT_PADDING_WORDS(0x10); + + struct { + union { + u32 raw_config; + BitField<0, 1, u32> valid; + BitField<4, 5, u32> index; + }; + INSERT_PADDING_WORDS(7); + } cb_bind[MaxShaderStage]; + + INSERT_PADDING_WORDS(0x56); + + u32 tex_cb_index; + + INSERT_PADDING_WORDS(0x395); + + struct { + /// Compressed address of a buffer that holds information about bound SSBOs. + /// This address is usually bound to c0 in the shaders. + u32 buffer_address; + + GPUVAddr BufferAddress() const { + return static_cast<GPUVAddr>(buffer_address) << 8; + } + } ssbo_info; + + INSERT_PADDING_WORDS(0x11); + + struct { + u32 address[MaxShaderStage]; + u32 size[MaxShaderStage]; + } tex_info_buffers; + + INSERT_PADDING_WORDS(0x102); }; std::array<u32, NUM_REGS> reg_array; }; @@ -57,18 +491,98 @@ public: static_assert(sizeof(Regs) == Regs::NUM_REGS * sizeof(u32), "Maxwell3D Regs has wrong size"); + struct State { + struct ConstBufferInfo { + GPUVAddr address; + u32 index; + u32 size; + bool enabled; + }; + + struct ShaderStageInfo { + std::array<ConstBufferInfo, Regs::MaxConstBuffers> const_buffers; + }; + + std::array<ShaderStageInfo, Regs::MaxShaderStage> shader_stages; + }; + + State state{}; + + /// Reads a register value located at the input method address + u32 GetRegisterValue(u32 method) const; + + /// Write the value to the register identified by method. + void WriteReg(u32 method, u32 value, u32 remaining_params); + + /// Uploads the code for a GPU macro program associated with the specified entry. + void SubmitMacroCode(u32 entry, std::vector<u32> code); + + /// Returns a list of enabled textures for the specified shader stage. + std::vector<Texture::FullTextureInfo> GetStageTextures(Regs::ShaderStage stage) const; + private: + MemoryManager& memory_manager; + + std::unordered_map<u32, std::vector<u32>> uploaded_macros; + + /// Macro method that is currently being executed / being fed parameters. + u32 executing_macro = 0; + /// Parameters that have been submitted to the macro call so far. + std::vector<u32> macro_params; + + /// Interpreter for the macro codes uploaded to the GPU. + MacroInterpreter macro_interpreter; + + /// Retrieves information about a specific TIC entry from the TIC buffer. + Texture::TICEntry GetTICEntry(u32 tic_index) const; + + /// Retrieves information about a specific TSC entry from the TSC buffer. + Texture::TSCEntry GetTSCEntry(u32 tsc_index) const; + + /** + * Call a macro on this engine. + * @param method Method to call + * @param parameters Arguments to the method call + */ + void CallMacroMethod(u32 method, std::vector<u32> parameters); + /// Handles a write to the QUERY_GET register. void ProcessQueryGet(); - MemoryManager& memory_manager; + /// Handles a write to the CB_DATA[i] register. + void ProcessCBData(u32 value); + + /// Handles a write to the CB_BIND register. + void ProcessCBBind(Regs::ShaderStage stage); + + /// Handles a write to the VERTEX_END_GL register, triggering a draw. + void DrawArrays(); }; #define ASSERT_REG_POSITION(field_name, position) \ static_assert(offsetof(Maxwell3D::Regs, field_name) == position * 4, \ "Field " #field_name " has invalid position") +ASSERT_REG_POSITION(rt, 0x200); +ASSERT_REG_POSITION(viewport, 0x300); +ASSERT_REG_POSITION(vertex_buffer, 0x35D); +ASSERT_REG_POSITION(zeta, 0x3F8); +ASSERT_REG_POSITION(vertex_attrib_format[0], 0x458); +ASSERT_REG_POSITION(rt_control, 0x487); +ASSERT_REG_POSITION(tsc, 0x557); +ASSERT_REG_POSITION(tic, 0x55D); +ASSERT_REG_POSITION(code_address, 0x582); +ASSERT_REG_POSITION(draw, 0x585); ASSERT_REG_POSITION(query, 0x6C0); +ASSERT_REG_POSITION(vertex_array[0], 0x700); +ASSERT_REG_POSITION(vertex_array_limit[0], 0x7C0); +ASSERT_REG_POSITION(shader_config[0], 0x800); +ASSERT_REG_POSITION(const_buffer, 0x8E0); +ASSERT_REG_POSITION(cb_bind[0], 0x904); +ASSERT_REG_POSITION(tex_cb_index, 0x982); +ASSERT_REG_POSITION(ssbo_info, 0xD18); +ASSERT_REG_POSITION(tex_info_buffers.address[0], 0xD2A); +ASSERT_REG_POSITION(tex_info_buffers.size[0], 0xD2F); #undef ASSERT_REG_POSITION diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp new file mode 100644 index 000000000..9463cd5d6 --- /dev/null +++ b/src/video_core/gpu.cpp @@ -0,0 +1,25 @@ +// Copyright 2018 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "video_core/engines/fermi_2d.h" +#include "video_core/engines/maxwell_3d.h" +#include "video_core/engines/maxwell_compute.h" +#include "video_core/gpu.h" + +namespace Tegra { + +GPU::GPU() { + memory_manager = std::make_unique<MemoryManager>(); + maxwell_3d = std::make_unique<Engines::Maxwell3D>(*memory_manager); + fermi_2d = std::make_unique<Engines::Fermi2D>(); + maxwell_compute = std::make_unique<Engines::MaxwellCompute>(); +} + +GPU::~GPU() = default; + +const Tegra::Engines::Maxwell3D& GPU::Get3DEngine() const { + return *maxwell_3d; +} + +} // namespace Tegra diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h index ba7781756..71a8661b4 100644 --- a/src/video_core/gpu.h +++ b/src/video_core/gpu.h @@ -6,14 +6,57 @@ #include <memory> #include <unordered_map> +#include <vector> #include "common/common_types.h" -#include "video_core/engines/fermi_2d.h" -#include "video_core/engines/maxwell_3d.h" -#include "video_core/engines/maxwell_compute.h" +#include "core/hle/service/nvflinger/buffer_queue.h" #include "video_core/memory_manager.h" namespace Tegra { +enum class RenderTargetFormat : u32 { + NONE = 0x0, + RGBA8_UNORM = 0xD5, +}; + +class DebugContext; + +/** + * Struct describing framebuffer configuration + */ +struct FramebufferConfig { + enum class PixelFormat : u32 { + ABGR8 = 1, + }; + + /** + * Returns the number of bytes per pixel. + */ + static u32 BytesPerPixel(PixelFormat format) { + switch (format) { + case PixelFormat::ABGR8: + return 4; + } + + UNREACHABLE(); + } + + VAddr address; + u32 offset; + u32 width; + u32 height; + u32 stride; + PixelFormat pixel_format; + + using TransformFlags = Service::NVFlinger::BufferQueue::BufferTransformFlags; + TransformFlags transform_flags; +}; + +namespace Engines { +class Fermi2D; +class Maxwell3D; +class MaxwellCompute; +} // namespace Engines + enum class EngineID { FERMI_TWOD_A = 0x902D, // 2D Engine MAXWELL_B = 0xB197, // 3D Engine @@ -24,22 +67,26 @@ enum class EngineID { class GPU final { public: - GPU() { - memory_manager = std::make_unique<MemoryManager>(); - maxwell_3d = std::make_unique<Engines::Maxwell3D>(*memory_manager); - fermi_2d = std::make_unique<Engines::Fermi2D>(); - maxwell_compute = std::make_unique<Engines::MaxwellCompute>(); - } - ~GPU() = default; + GPU(); + ~GPU(); /// Processes a command list stored at the specified address in GPU memory. void ProcessCommandList(GPUVAddr address, u32 size); + /// Returns a reference to the Maxwell3D GPU engine. + const Engines::Maxwell3D& Get3DEngine() const; + std::unique_ptr<MemoryManager> memory_manager; + Engines::Maxwell3D& Maxwell3D() { + return *maxwell_3d; + } + private: + static constexpr u32 InvalidGraphMacroEntry = 0xFFFFFFFF; + /// Writes a single register in the engine bound to the specified subchannel - void WriteReg(u32 method, u32 subchannel, u32 value); + void WriteReg(u32 method, u32 subchannel, u32 value, u32 remaining_params); /// Mapping of command subchannels to their bound engine ids. std::unordered_map<u32, EngineID> bound_engines; @@ -50,6 +97,11 @@ private: std::unique_ptr<Engines::Fermi2D> fermi_2d; /// Compute engine std::unique_ptr<Engines::MaxwellCompute> maxwell_compute; + + /// Entry of the macro that is currently being uploaded + u32 current_macro_entry = InvalidGraphMacroEntry; + /// Code being uploaded for the current macro + std::vector<u32> current_macro_code; }; } // namespace Tegra diff --git a/src/video_core/macro_interpreter.cpp b/src/video_core/macro_interpreter.cpp new file mode 100644 index 000000000..993a67746 --- /dev/null +++ b/src/video_core/macro_interpreter.cpp @@ -0,0 +1,257 @@ +// Copyright 2018 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "common/assert.h" +#include "common/logging/log.h" +#include "video_core/engines/maxwell_3d.h" +#include "video_core/macro_interpreter.h" + +namespace Tegra { + +MacroInterpreter::MacroInterpreter(Engines::Maxwell3D& maxwell3d) : maxwell3d(maxwell3d) {} + +void MacroInterpreter::Execute(const std::vector<u32>& code, std::vector<u32> parameters) { + Reset(); + registers[1] = parameters[0]; + this->parameters = std::move(parameters); + + // Execute the code until we hit an exit condition. + bool keep_executing = true; + while (keep_executing) { + keep_executing = Step(code, false); + } + + // Assert the the macro used all the input parameters + ASSERT(next_parameter_index == this->parameters.size()); +} + +void MacroInterpreter::Reset() { + registers = {}; + pc = 0; + delayed_pc = boost::none; + method_address.raw = 0; + parameters.clear(); + // The next parameter index starts at 1, because $r1 already has the value of the first + // parameter. + next_parameter_index = 1; +} + +bool MacroInterpreter::Step(const std::vector<u32>& code, bool is_delay_slot) { + u32 base_address = pc; + + Opcode opcode = GetOpcode(code); + pc += 4; + + // Update the program counter if we were delayed + if (delayed_pc != boost::none) { + ASSERT(is_delay_slot); + pc = *delayed_pc; + delayed_pc = boost::none; + } + + switch (opcode.operation) { + case Operation::ALU: { + u32 result = GetALUResult(opcode.alu_operation, GetRegister(opcode.src_a), + GetRegister(opcode.src_b)); + ProcessResult(opcode.result_operation, opcode.dst, result); + break; + } + case Operation::AddImmediate: { + ProcessResult(opcode.result_operation, opcode.dst, + GetRegister(opcode.src_a) + opcode.immediate); + break; + } + case Operation::ExtractInsert: { + u32 dst = GetRegister(opcode.src_a); + u32 src = GetRegister(opcode.src_b); + + src = (src >> opcode.bf_src_bit) & opcode.GetBitfieldMask(); + dst &= ~(opcode.GetBitfieldMask() << opcode.bf_dst_bit); + dst |= src << opcode.bf_dst_bit; + ProcessResult(opcode.result_operation, opcode.dst, dst); + break; + } + case Operation::ExtractShiftLeftImmediate: { + u32 dst = GetRegister(opcode.src_a); + u32 src = GetRegister(opcode.src_b); + + u32 result = ((src >> dst) & opcode.GetBitfieldMask()) << opcode.bf_dst_bit; + + ProcessResult(opcode.result_operation, opcode.dst, result); + break; + } + case Operation::ExtractShiftLeftRegister: { + u32 dst = GetRegister(opcode.src_a); + u32 src = GetRegister(opcode.src_b); + + u32 result = ((src >> opcode.bf_src_bit) & opcode.GetBitfieldMask()) << dst; + + ProcessResult(opcode.result_operation, opcode.dst, result); + break; + } + case Operation::Read: { + u32 result = Read(GetRegister(opcode.src_a) + opcode.immediate); + ProcessResult(opcode.result_operation, opcode.dst, result); + break; + } + case Operation::Branch: { + ASSERT_MSG(!is_delay_slot, "Executing a branch in a delay slot is not valid"); + u32 value = GetRegister(opcode.src_a); + bool taken = EvaluateBranchCondition(opcode.branch_condition, value); + if (taken) { + // Ignore the delay slot if the branch has the annul bit. + if (opcode.branch_annul) { + pc = base_address + (opcode.immediate << 2); + return true; + } + + delayed_pc = base_address + (opcode.immediate << 2); + // Execute one more instruction due to the delay slot. + return Step(code, true); + } + break; + } + default: + UNIMPLEMENTED_MSG("Unimplemented macro operation %u", + static_cast<u32>(opcode.operation.Value())); + } + + if (opcode.is_exit) { + // Exit has a delay slot, execute the next instruction + // Note: Executing an exit during a branch delay slot will cause the instruction at the + // branch target to be executed before exiting. + Step(code, true); + return false; + } + + return true; +} + +MacroInterpreter::Opcode MacroInterpreter::GetOpcode(const std::vector<u32>& code) const { + ASSERT((pc % sizeof(u32)) == 0); + ASSERT(pc < code.size() * sizeof(u32)); + return {code[pc / sizeof(u32)]}; +} + +u32 MacroInterpreter::GetALUResult(ALUOperation operation, u32 src_a, u32 src_b) const { + switch (operation) { + case ALUOperation::Add: + return src_a + src_b; + // TODO(Subv): Implement AddWithCarry + case ALUOperation::Subtract: + return src_a - src_b; + // TODO(Subv): Implement SubtractWithBorrow + case ALUOperation::Xor: + return src_a ^ src_b; + case ALUOperation::Or: + return src_a | src_b; + case ALUOperation::And: + return src_a & src_b; + case ALUOperation::AndNot: + return src_a & ~src_b; + case ALUOperation::Nand: + return ~(src_a & src_b); + + default: + UNIMPLEMENTED_MSG("Unimplemented ALU operation %u", static_cast<u32>(operation)); + } +} + +void MacroInterpreter::ProcessResult(ResultOperation operation, u32 reg, u32 result) { + switch (operation) { + case ResultOperation::IgnoreAndFetch: + // Fetch parameter and ignore result. + SetRegister(reg, FetchParameter()); + break; + case ResultOperation::Move: + // Move result. + SetRegister(reg, result); + break; + case ResultOperation::MoveAndSetMethod: + // Move result and use as Method Address. + SetRegister(reg, result); + SetMethodAddress(result); + break; + case ResultOperation::FetchAndSend: + // Fetch parameter and send result. + SetRegister(reg, FetchParameter()); + Send(result); + break; + case ResultOperation::MoveAndSend: + // Move and send result. + SetRegister(reg, result); + Send(result); + break; + case ResultOperation::FetchAndSetMethod: + // Fetch parameter and use result as Method Address. + SetRegister(reg, FetchParameter()); + SetMethodAddress(result); + break; + case ResultOperation::MoveAndSetMethodFetchAndSend: + // Move result and use as Method Address, then fetch and send parameter. + SetRegister(reg, result); + SetMethodAddress(result); + Send(FetchParameter()); + break; + case ResultOperation::MoveAndSetMethodSend: + // Move result and use as Method Address, then send bits 12:17 of result. + SetRegister(reg, result); + SetMethodAddress(result); + Send((result >> 12) & 0b111111); + break; + default: + UNIMPLEMENTED_MSG("Unimplemented result operation %u", static_cast<u32>(operation)); + } +} + +u32 MacroInterpreter::FetchParameter() { + ASSERT(next_parameter_index < parameters.size()); + return parameters[next_parameter_index++]; +} + +u32 MacroInterpreter::GetRegister(u32 register_id) const { + // Register 0 is supposed to always return 0. + if (register_id == 0) + return 0; + + ASSERT(register_id < registers.size()); + return registers[register_id]; +} + +void MacroInterpreter::SetRegister(u32 register_id, u32 value) { + // Register 0 is supposed to always return 0. NOP is implemented as a store to the zero + // register. + if (register_id == 0) + return; + + ASSERT(register_id < registers.size()); + registers[register_id] = value; +} + +void MacroInterpreter::SetMethodAddress(u32 address) { + method_address.raw = address; +} + +void MacroInterpreter::Send(u32 value) { + maxwell3d.WriteReg(method_address.address, value, 0); + // Increment the method address by the method increment. + method_address.address.Assign(method_address.address.Value() + + method_address.increment.Value()); +} + +u32 MacroInterpreter::Read(u32 method) const { + return maxwell3d.GetRegisterValue(method); +} + +bool MacroInterpreter::EvaluateBranchCondition(BranchCondition cond, u32 value) const { + switch (cond) { + case BranchCondition::Zero: + return value == 0; + case BranchCondition::NotZero: + return value != 0; + } + UNREACHABLE(); +} + +} // namespace Tegra diff --git a/src/video_core/macro_interpreter.h b/src/video_core/macro_interpreter.h new file mode 100644 index 000000000..a71e359d8 --- /dev/null +++ b/src/video_core/macro_interpreter.h @@ -0,0 +1,164 @@ +// Copyright 2018 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <array> +#include <vector> +#include <boost/optional.hpp> +#include "common/bit_field.h" +#include "common/common_types.h" + +namespace Tegra { +namespace Engines { +class Maxwell3D; +} + +class MacroInterpreter final { +public: + explicit MacroInterpreter(Engines::Maxwell3D& maxwell3d); + + /** + * Executes the macro code with the specified input parameters. + * @param code The macro byte code to execute + * @param parameters The parameters of the macro + */ + void Execute(const std::vector<u32>& code, std::vector<u32> parameters); + +private: + enum class Operation : u32 { + ALU = 0, + AddImmediate = 1, + ExtractInsert = 2, + ExtractShiftLeftImmediate = 3, + ExtractShiftLeftRegister = 4, + Read = 5, + Unused = 6, // This operation doesn't seem to be a valid encoding. + Branch = 7, + }; + + enum class ALUOperation : u32 { + Add = 0, + AddWithCarry = 1, + Subtract = 2, + SubtractWithBorrow = 3, + // Operations 4-7 don't seem to be valid encodings. + Xor = 8, + Or = 9, + And = 10, + AndNot = 11, + Nand = 12 + }; + + enum class ResultOperation : u32 { + IgnoreAndFetch = 0, + Move = 1, + MoveAndSetMethod = 2, + FetchAndSend = 3, + MoveAndSend = 4, + FetchAndSetMethod = 5, + MoveAndSetMethodFetchAndSend = 6, + MoveAndSetMethodSend = 7 + }; + + enum class BranchCondition : u32 { + Zero = 0, + NotZero = 1, + }; + + union Opcode { + u32 raw; + BitField<0, 3, Operation> operation; + BitField<4, 3, ResultOperation> result_operation; + BitField<4, 1, BranchCondition> branch_condition; + BitField<5, 1, u32> + branch_annul; // If set on a branch, then the branch doesn't have a delay slot. + BitField<7, 1, u32> is_exit; + BitField<8, 3, u32> dst; + BitField<11, 3, u32> src_a; + BitField<14, 3, u32> src_b; + // The signed immediate overlaps the second source operand and the alu operation. + BitField<14, 18, s32> immediate; + + BitField<17, 5, ALUOperation> alu_operation; + + // Bitfield instructions data + BitField<17, 5, u32> bf_src_bit; + BitField<22, 5, u32> bf_size; + BitField<27, 5, u32> bf_dst_bit; + + u32 GetBitfieldMask() const { + return (1 << bf_size) - 1; + } + }; + + union MethodAddress { + u32 raw; + BitField<0, 12, u32> address; + BitField<12, 6, u32> increment; + }; + + /// Resets the execution engine state, zeroing registers, etc. + void Reset(); + + /** + * Executes a single macro instruction located at the current program counter. Returns whether + * the interpreter should keep running. + * @param code The macro code to execute. + * @param is_delay_slot Whether the current step is being executed due to a delay slot in a + * previous instruction. + */ + bool Step(const std::vector<u32>& code, bool is_delay_slot); + + /// Calculates the result of an ALU operation. src_a OP src_b; + u32 GetALUResult(ALUOperation operation, u32 src_a, u32 src_b) const; + + /// Performs the result operation on the input result and stores it in the specified register + /// (if necessary). + void ProcessResult(ResultOperation operation, u32 reg, u32 result); + + /// Evaluates the branch condition and returns whether the branch should be taken or not. + bool EvaluateBranchCondition(BranchCondition cond, u32 value) const; + + /// Reads an opcode at the current program counter location. + Opcode GetOpcode(const std::vector<u32>& code) const; + + /// Returns the specified register's value. Register 0 is hardcoded to always return 0. + u32 GetRegister(u32 register_id) const; + + /// Sets the register to the input value. + void SetRegister(u32 register_id, u32 value); + + /// Sets the method address to use for the next Send instruction. + void SetMethodAddress(u32 address); + + /// Calls a GPU Engine method with the input parameter. + void Send(u32 value); + + /// Reads a GPU register located at the method address. + u32 Read(u32 method) const; + + /// Returns the next parameter in the parameter queue. + u32 FetchParameter(); + + Engines::Maxwell3D& maxwell3d; + + u32 pc; ///< Current program counter + boost::optional<u32> + delayed_pc; ///< Program counter to execute at after the delay slot is executed. + + static constexpr size_t NumMacroRegisters = 8; + + /// General purpose macro registers. + std::array<u32, NumMacroRegisters> registers = {}; + + /// Method address to use for the next Send instruction. + MethodAddress method_address = {}; + + /// Input parameters of the current macro. + std::vector<u32> parameters; + /// Index of the next parameter that will be fetched by the 'parm' instruction. + u32 next_parameter_index = 0; +}; +} // namespace Tegra diff --git a/src/video_core/rasterizer_interface.h b/src/video_core/rasterizer_interface.h new file mode 100644 index 000000000..8239f9aad --- /dev/null +++ b/src/video_core/rasterizer_interface.h @@ -0,0 +1,63 @@ +// Copyright 2018 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "common/common_types.h" +#include "video_core/gpu.h" + +struct ScreenInfo; + +namespace VideoCore { + +class RasterizerInterface { +public: + virtual ~RasterizerInterface() {} + + /// Draw the current batch of vertex arrays + virtual void DrawArrays() = 0; + + /// Notify rasterizer that the specified Maxwell register has been changed + virtual void NotifyMaxwellRegisterChanged(u32 id) = 0; + + /// Notify rasterizer that all caches should be flushed to 3DS memory + virtual void FlushAll() = 0; + + /// Notify rasterizer that any caches of the specified region should be flushed to 3DS memory + virtual void FlushRegion(VAddr addr, u64 size) = 0; + + /// Notify rasterizer that any caches of the specified region should be invalidated + virtual void InvalidateRegion(VAddr addr, u64 size) = 0; + + /// Notify rasterizer that any caches of the specified region should be flushed to 3DS memory + /// and invalidated + virtual void FlushAndInvalidateRegion(VAddr addr, u64 size) = 0; + + /// Attempt to use a faster method to perform a display transfer with is_texture_copy = 0 + virtual bool AccelerateDisplayTransfer(const void* config) { + return false; + } + + /// Attempt to use a faster method to perform a display transfer with is_texture_copy = 1 + virtual bool AccelerateTextureCopy(const void* config) { + return false; + } + + /// Attempt to use a faster method to fill a region + virtual bool AccelerateFill(const void* config) { + return false; + } + + /// Attempt to use a faster method to display the framebuffer to screen + virtual bool AccelerateDisplay(const Tegra::FramebufferConfig& framebuffer, + VAddr framebuffer_addr, u32 pixel_stride, + ScreenInfo& screen_info) { + return false; + } + + virtual bool AccelerateDrawBatch(bool is_indexed) { + return false; + } +}; +} // namespace VideoCore diff --git a/src/video_core/renderer_base.cpp b/src/video_core/renderer_base.cpp index 51e1d45f9..30075b23c 100644 --- a/src/video_core/renderer_base.cpp +++ b/src/video_core/renderer_base.cpp @@ -5,6 +5,11 @@ #include <atomic> #include <memory> #include "video_core/renderer_base.h" +#include "video_core/renderer_opengl/gl_rasterizer.h" #include "video_core/video_core.h" -void RendererBase::RefreshRasterizerSetting() {} +void RendererBase::RefreshRasterizerSetting() { + if (rasterizer == nullptr) { + rasterizer = std::make_unique<RasterizerOpenGL>(); + } +} diff --git a/src/video_core/renderer_base.h b/src/video_core/renderer_base.h index 2aba50eda..89a960eaf 100644 --- a/src/video_core/renderer_base.h +++ b/src/video_core/renderer_base.h @@ -8,6 +8,8 @@ #include <boost/optional.hpp> #include "common/assert.h" #include "common/common_types.h" +#include "video_core/gpu.h" +#include "video_core/rasterizer_interface.h" class EmuWindow; @@ -16,40 +18,10 @@ public: /// Used to reference a framebuffer enum kFramebuffer { kFramebuffer_VirtualXFB = 0, kFramebuffer_EFB, kFramebuffer_Texture }; - /** - * Struct describing framebuffer metadata - * TODO(bunnei): This struct belongs in the GPU code, but we don't have a good place for it yet. - */ - struct FramebufferInfo { - enum class PixelFormat : u32 { - ABGR8 = 1, - }; - - /** - * Returns the number of bytes per pixel. - */ - static u32 BytesPerPixel(PixelFormat format) { - switch (format) { - case PixelFormat::ABGR8: - return 4; - } - - UNREACHABLE(); - } - - VAddr address; - u32 offset; - u32 width; - u32 height; - u32 stride; - PixelFormat pixel_format; - bool flip_vertical; - }; - virtual ~RendererBase() {} /// Swap buffers (render frame) - virtual void SwapBuffers(boost::optional<const FramebufferInfo&> framebuffer_info) = 0; + virtual void SwapBuffers(boost::optional<const Tegra::FramebufferConfig&> framebuffer) = 0; /** * Set the emulator window to use for renderer @@ -74,12 +46,16 @@ public: return m_current_frame; } + VideoCore::RasterizerInterface* Rasterizer() const { + return rasterizer.get(); + } + void RefreshRasterizerSetting(); protected: + std::unique_ptr<VideoCore::RasterizerInterface> rasterizer; f32 m_current_fps = 0.0f; ///< Current framerate, should be set by the renderer int m_current_frame = 0; ///< Current frame, should be set by the renderer private: - bool opengl_rasterizer_active = false; }; diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp new file mode 100644 index 000000000..911890f16 --- /dev/null +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -0,0 +1,578 @@ +// Copyright 2015 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <memory> +#include <string> +#include <tuple> +#include <utility> +#include <glad/glad.h> +#include "common/alignment.h" +#include "common/assert.h" +#include "common/logging/log.h" +#include "common/math_util.h" +#include "common/microprofile.h" +#include "common/scope_exit.h" +#include "common/vector_math.h" +#include "core/core.h" +#include "core/hle/kernel/process.h" +#include "core/settings.h" +#include "video_core/engines/maxwell_3d.h" +#include "video_core/renderer_opengl/gl_rasterizer.h" +#include "video_core/renderer_opengl/gl_shader_gen.h" +#include "video_core/renderer_opengl/maxwell_to_gl.h" +#include "video_core/renderer_opengl/renderer_opengl.h" + +using Maxwell = Tegra::Engines::Maxwell3D::Regs; +using PixelFormat = SurfaceParams::PixelFormat; +using SurfaceType = SurfaceParams::SurfaceType; + +MICROPROFILE_DEFINE(OpenGL_VAO, "OpenGL", "Vertex Array Setup", MP_RGB(128, 128, 192)); +MICROPROFILE_DEFINE(OpenGL_VS, "OpenGL", "Vertex Shader Setup", MP_RGB(128, 128, 192)); +MICROPROFILE_DEFINE(OpenGL_FS, "OpenGL", "Fragment Shader Setup", MP_RGB(128, 128, 192)); +MICROPROFILE_DEFINE(OpenGL_Drawing, "OpenGL", "Drawing", MP_RGB(128, 128, 192)); +MICROPROFILE_DEFINE(OpenGL_Blits, "OpenGL", "Blits", MP_RGB(100, 100, 255)); +MICROPROFILE_DEFINE(OpenGL_CacheManagement, "OpenGL", "Cache Mgmt", MP_RGB(100, 255, 100)); + +enum class UniformBindings : GLuint { Common, VS, FS }; + +static void SetShaderUniformBlockBinding(GLuint shader, const char* name, UniformBindings binding, + size_t expected_size) { + GLuint ub_index = glGetUniformBlockIndex(shader, name); + if (ub_index != GL_INVALID_INDEX) { + GLint ub_size = 0; + glGetActiveUniformBlockiv(shader, ub_index, GL_UNIFORM_BLOCK_DATA_SIZE, &ub_size); + ASSERT_MSG(ub_size == expected_size, + "Uniform block size did not match! Got %d, expected %zu", + static_cast<int>(ub_size), expected_size); + glUniformBlockBinding(shader, ub_index, static_cast<GLuint>(binding)); + } +} + +static void SetShaderUniformBlockBindings(GLuint shader) { + SetShaderUniformBlockBinding(shader, "shader_data", UniformBindings::Common, + sizeof(RasterizerOpenGL::UniformData)); + SetShaderUniformBlockBinding(shader, "vs_config", UniformBindings::VS, + sizeof(RasterizerOpenGL::VSUniformData)); + SetShaderUniformBlockBinding(shader, "fs_config", UniformBindings::FS, + sizeof(RasterizerOpenGL::FSUniformData)); +} + +RasterizerOpenGL::RasterizerOpenGL() { + shader_dirty = true; + + has_ARB_buffer_storage = false; + has_ARB_direct_state_access = false; + has_ARB_separate_shader_objects = false; + has_ARB_vertex_attrib_binding = false; + + GLint ext_num; + glGetIntegerv(GL_NUM_EXTENSIONS, &ext_num); + for (GLint i = 0; i < ext_num; i++) { + std::string extension{reinterpret_cast<const char*>(glGetStringi(GL_EXTENSIONS, i))}; + + if (extension == "GL_ARB_buffer_storage") { + has_ARB_buffer_storage = true; + } else if (extension == "GL_ARB_direct_state_access") { + has_ARB_direct_state_access = true; + } else if (extension == "GL_ARB_separate_shader_objects") { + has_ARB_separate_shader_objects = true; + } else if (extension == "GL_ARB_vertex_attrib_binding") { + has_ARB_vertex_attrib_binding = true; + } + } + + // Clipping plane 0 is always enabled for PICA fixed clip plane z <= 0 + state.clip_distance[0] = true; + + // Generate VBO, VAO and UBO + vertex_buffer = OGLStreamBuffer::MakeBuffer(GLAD_GL_ARB_buffer_storage, GL_ARRAY_BUFFER); + vertex_buffer->Create(VERTEX_BUFFER_SIZE, VERTEX_BUFFER_SIZE / 2); + sw_vao.Create(); + uniform_buffer.Create(); + + state.draw.vertex_array = sw_vao.handle; + state.draw.vertex_buffer = vertex_buffer->GetHandle(); + state.draw.uniform_buffer = uniform_buffer.handle; + state.Apply(); + + glBufferData(GL_UNIFORM_BUFFER, sizeof(UniformData), nullptr, GL_STATIC_DRAW); + glBindBufferBase(GL_UNIFORM_BUFFER, 0, uniform_buffer.handle); + + uniform_block_data.dirty = true; + + // Create render framebuffer + framebuffer.Create(); + + if (has_ARB_separate_shader_objects) { + hw_vao.Create(); + hw_vao_enabled_attributes.fill(false); + + stream_buffer = OGLStreamBuffer::MakeBuffer(has_ARB_buffer_storage, GL_ARRAY_BUFFER); + stream_buffer->Create(STREAM_BUFFER_SIZE, STREAM_BUFFER_SIZE / 2); + state.draw.vertex_buffer = stream_buffer->GetHandle(); + + pipeline.Create(); + state.draw.program_pipeline = pipeline.handle; + state.draw.shader_program = 0; + state.draw.vertex_array = hw_vao.handle; + state.Apply(); + + glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, stream_buffer->GetHandle()); + + vs_uniform_buffer.Create(); + glBindBuffer(GL_UNIFORM_BUFFER, vs_uniform_buffer.handle); + glBufferData(GL_UNIFORM_BUFFER, sizeof(VSUniformData), nullptr, GL_STREAM_COPY); + glBindBufferBase(GL_UNIFORM_BUFFER, 1, vs_uniform_buffer.handle); + } else { + UNREACHABLE(); + } + + accelerate_draw = AccelDraw::Disabled; + + glEnable(GL_BLEND); + + LOG_CRITICAL(Render_OpenGL, "Sync fixed function OpenGL state here!"); +} + +RasterizerOpenGL::~RasterizerOpenGL() { + if (stream_buffer != nullptr) { + state.draw.vertex_buffer = stream_buffer->GetHandle(); + state.Apply(); + stream_buffer->Release(); + } +} + +void RasterizerOpenGL::AnalyzeVertexArray(bool is_indexed) { + const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs; + + if (is_indexed) { + UNREACHABLE(); + } + + // TODO(bunnei): Add support for 1+ vertex arrays + vs_input_size = regs.vertex_buffer.count * regs.vertex_array[0].stride; +} + +void RasterizerOpenGL::SetupVertexArray(u8* array_ptr, GLintptr buffer_offset) { + MICROPROFILE_SCOPE(OpenGL_VAO); + const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs; + const auto& memory_manager = Core::System().GetInstance().GPU().memory_manager; + + state.draw.vertex_array = hw_vao.handle; + state.draw.vertex_buffer = stream_buffer->GetHandle(); + state.Apply(); + + // TODO(bunnei): Add support for 1+ vertex arrays + const auto& vertex_array{regs.vertex_array[0]}; + ASSERT_MSG(vertex_array.enable, "vertex array 0 is disabled?"); + ASSERT_MSG(!vertex_array.divisor, "vertex array 0 divisor is unimplemented!"); + for (unsigned index = 1; index < Maxwell::NumVertexArrays; ++index) { + ASSERT_MSG(!regs.vertex_array[index].enable, "vertex array %d is unimplemented!", index); + } + + // Use the vertex array as-is, assumes that the data is formatted correctly for OpenGL. + // Enables the first 16 vertex attributes always, as we don't know which ones are actually used + // until shader time. Note, Tegra technically supports 32, but we're cappinig this to 16 for now + // to avoid OpenGL errors. + for (unsigned index = 0; index < 16; ++index) { + auto& attrib = regs.vertex_attrib_format[index]; + glVertexAttribPointer(index, attrib.ComponentCount(), MaxwellToGL::VertexType(attrib), + attrib.IsNormalized() ? GL_TRUE : GL_FALSE, vertex_array.stride, + reinterpret_cast<GLvoid*>(buffer_offset + attrib.offset)); + glEnableVertexAttribArray(index); + hw_vao_enabled_attributes[index] = true; + } + + // Copy vertex array data + const u32 data_size{vertex_array.stride * regs.vertex_buffer.count}; + const VAddr data_addr{memory_manager->PhysicalToVirtualAddress(vertex_array.StartAddress())}; + res_cache.FlushRegion(data_addr, data_size, nullptr); + Memory::ReadBlock(data_addr, array_ptr, data_size); + + array_ptr += data_size; + buffer_offset += data_size; +} + +void RasterizerOpenGL::SetupVertexShader(VSUniformData* ub_ptr, GLintptr buffer_offset) { + MICROPROFILE_SCOPE(OpenGL_VS); + LOG_CRITICAL(Render_OpenGL, "Emulated shaders are not supported! Using a passthrough shader."); + glUseProgramStages(pipeline.handle, GL_VERTEX_SHADER_BIT, current_shader->shader.handle); +} + +void RasterizerOpenGL::SetupFragmentShader(FSUniformData* ub_ptr, GLintptr buffer_offset) { + MICROPROFILE_SCOPE(OpenGL_FS); + UNREACHABLE(); +} + +bool RasterizerOpenGL::AccelerateDrawBatch(bool is_indexed) { + if (!has_ARB_separate_shader_objects) { + UNREACHABLE(); + return false; + } + + accelerate_draw = is_indexed ? AccelDraw::Indexed : AccelDraw::Arrays; + DrawArrays(); + + return true; +} + +void RasterizerOpenGL::DrawArrays() { + if (accelerate_draw == AccelDraw::Disabled) + return; + + MICROPROFILE_SCOPE(OpenGL_Drawing); + const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs; + + // TODO(bunnei): Implement these + const bool has_stencil = false; + const bool using_color_fb = true; + const bool using_depth_fb = false; + const MathUtil::Rectangle<s32> viewport_rect{regs.viewport[0].GetRect()}; + + const bool write_color_fb = + state.color_mask.red_enabled == GL_TRUE || state.color_mask.green_enabled == GL_TRUE || + state.color_mask.blue_enabled == GL_TRUE || state.color_mask.alpha_enabled == GL_TRUE; + + const bool write_depth_fb = + (state.depth.test_enabled && state.depth.write_mask == GL_TRUE) || + (has_stencil && state.stencil.test_enabled && state.stencil.write_mask != 0); + + Surface color_surface; + Surface depth_surface; + MathUtil::Rectangle<u32> surfaces_rect; + std::tie(color_surface, depth_surface, surfaces_rect) = + res_cache.GetFramebufferSurfaces(using_color_fb, using_depth_fb, viewport_rect); + + const u16 res_scale = color_surface != nullptr + ? color_surface->res_scale + : (depth_surface == nullptr ? 1u : depth_surface->res_scale); + + MathUtil::Rectangle<u32> draw_rect{ + static_cast<u32>(MathUtil::Clamp<s32>(static_cast<s32>(surfaces_rect.left) + + viewport_rect.left * res_scale, + surfaces_rect.left, surfaces_rect.right)), // Left + static_cast<u32>(MathUtil::Clamp<s32>(static_cast<s32>(surfaces_rect.bottom) + + viewport_rect.top * res_scale, + surfaces_rect.bottom, surfaces_rect.top)), // Top + static_cast<u32>(MathUtil::Clamp<s32>(static_cast<s32>(surfaces_rect.left) + + viewport_rect.right * res_scale, + surfaces_rect.left, surfaces_rect.right)), // Right + static_cast<u32>(MathUtil::Clamp<s32>(static_cast<s32>(surfaces_rect.bottom) + + viewport_rect.bottom * res_scale, + surfaces_rect.bottom, surfaces_rect.top))}; // Bottom + + // Bind the framebuffer surfaces + BindFramebufferSurfaces(color_surface, depth_surface, has_stencil); + + // Sync the viewport + SyncViewport(surfaces_rect, res_scale); + + // TODO(bunnei): Sync framebuffer_scale uniform here + // TODO(bunnei): Sync scissorbox uniform(s) here + // TODO(bunnei): Sync and bind the texture surfaces + + // Sync and bind the shader + if (shader_dirty) { + SetShader(); + shader_dirty = false; + } + + // Sync the uniform data + if (uniform_block_data.dirty) { + glBufferSubData(GL_UNIFORM_BUFFER, 0, sizeof(UniformData), &uniform_block_data.data); + uniform_block_data.dirty = false; + } + + // Viewport can have negative offsets or larger dimensions than our framebuffer sub-rect. Enable + // scissor test to prevent drawing outside of the framebuffer region + state.scissor.enabled = true; + state.scissor.x = draw_rect.left; + state.scissor.y = draw_rect.bottom; + state.scissor.width = draw_rect.GetWidth(); + state.scissor.height = draw_rect.GetHeight(); + state.Apply(); + + // Draw the vertex batch + const bool is_indexed = accelerate_draw == AccelDraw::Indexed; + AnalyzeVertexArray(is_indexed); + state.draw.vertex_buffer = stream_buffer->GetHandle(); + state.Apply(); + + size_t buffer_size = static_cast<size_t>(vs_input_size); + if (is_indexed) { + UNREACHABLE(); + } + buffer_size += sizeof(VSUniformData); + + size_t ptr_pos = 0; + u8* buffer_ptr; + GLintptr buffer_offset; + std::tie(buffer_ptr, buffer_offset) = + stream_buffer->Map(static_cast<GLsizeiptr>(buffer_size), 4); + + SetupVertexArray(buffer_ptr, buffer_offset); + ptr_pos += vs_input_size; + + GLintptr index_buffer_offset = 0; + if (is_indexed) { + UNREACHABLE(); + } + + SetupVertexShader(reinterpret_cast<VSUniformData*>(&buffer_ptr[ptr_pos]), + buffer_offset + static_cast<GLintptr>(ptr_pos)); + const GLintptr vs_ubo_offset = buffer_offset + static_cast<GLintptr>(ptr_pos); + ptr_pos += sizeof(VSUniformData); + + stream_buffer->Unmap(); + + const auto copy_buffer = [&](GLuint handle, GLintptr offset, GLsizeiptr size) { + if (has_ARB_direct_state_access) { + glCopyNamedBufferSubData(stream_buffer->GetHandle(), handle, offset, 0, size); + } else { + glBindBuffer(GL_COPY_WRITE_BUFFER, handle); + glCopyBufferSubData(GL_ARRAY_BUFFER, GL_COPY_WRITE_BUFFER, offset, 0, size); + } + }; + + copy_buffer(vs_uniform_buffer.handle, vs_ubo_offset, sizeof(VSUniformData)); + + glUseProgramStages(pipeline.handle, GL_FRAGMENT_SHADER_BIT, current_shader->shader.handle); + + if (is_indexed) { + UNREACHABLE(); + } else { + glDrawArrays(MaxwellToGL::PrimitiveTopology(regs.draw.topology), 0, + regs.vertex_buffer.count); + } + + // Disable scissor test + state.scissor.enabled = false; + + accelerate_draw = AccelDraw::Disabled; + + // Unbind textures for potential future use as framebuffer attachments + for (auto& texture_unit : state.texture_units) { + texture_unit.texture_2d = 0; + } + state.Apply(); + + // Mark framebuffer surfaces as dirty + MathUtil::Rectangle<u32> draw_rect_unscaled{ + draw_rect.left / res_scale, draw_rect.top / res_scale, draw_rect.right / res_scale, + draw_rect.bottom / res_scale}; + + if (color_surface != nullptr && write_color_fb) { + auto interval = color_surface->GetSubRectInterval(draw_rect_unscaled); + res_cache.InvalidateRegion(boost::icl::first(interval), boost::icl::length(interval), + color_surface); + } + if (depth_surface != nullptr && write_depth_fb) { + auto interval = depth_surface->GetSubRectInterval(draw_rect_unscaled); + res_cache.InvalidateRegion(boost::icl::first(interval), boost::icl::length(interval), + depth_surface); + } +} + +void RasterizerOpenGL::NotifyMaxwellRegisterChanged(u32 id) {} + +void RasterizerOpenGL::FlushAll() { + MICROPROFILE_SCOPE(OpenGL_CacheManagement); + res_cache.FlushAll(); +} + +void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) { + MICROPROFILE_SCOPE(OpenGL_CacheManagement); + res_cache.FlushRegion(addr, size); +} + +void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) { + MICROPROFILE_SCOPE(OpenGL_CacheManagement); + res_cache.InvalidateRegion(addr, size, nullptr); +} + +void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) { + MICROPROFILE_SCOPE(OpenGL_CacheManagement); + res_cache.FlushRegion(addr, size); + res_cache.InvalidateRegion(addr, size, nullptr); +} + +bool RasterizerOpenGL::AccelerateDisplayTransfer(const void* config) { + MICROPROFILE_SCOPE(OpenGL_Blits); + UNREACHABLE(); + return true; +} + +bool RasterizerOpenGL::AccelerateTextureCopy(const void* config) { + UNREACHABLE(); + return true; +} + +bool RasterizerOpenGL::AccelerateFill(const void* config) { + UNREACHABLE(); + return true; +} + +bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& framebuffer, + VAddr framebuffer_addr, u32 pixel_stride, + ScreenInfo& screen_info) { + if (framebuffer_addr == 0) { + return false; + } + MICROPROFILE_SCOPE(OpenGL_CacheManagement); + + SurfaceParams src_params; + src_params.addr = framebuffer_addr; + src_params.width = std::min(framebuffer.width, pixel_stride); + src_params.height = framebuffer.height; + src_params.stride = pixel_stride; + src_params.is_tiled = false; + src_params.pixel_format = + SurfaceParams::PixelFormatFromGPUPixelFormat(framebuffer.pixel_format); + src_params.UpdateParams(); + + MathUtil::Rectangle<u32> src_rect; + Surface src_surface; + std::tie(src_surface, src_rect) = + res_cache.GetSurfaceSubRect(src_params, ScaleMatch::Ignore, true); + + if (src_surface == nullptr) { + return false; + } + + u32 scaled_width = src_surface->GetScaledWidth(); + u32 scaled_height = src_surface->GetScaledHeight(); + + screen_info.display_texcoords = MathUtil::Rectangle<float>( + (float)src_rect.bottom / (float)scaled_height, (float)src_rect.left / (float)scaled_width, + (float)src_rect.top / (float)scaled_height, (float)src_rect.right / (float)scaled_width); + + screen_info.display_texture = src_surface->texture.handle; + + return true; +} + +void RasterizerOpenGL::SetShader() { + // TODO(bunnei): The below sets up a static test shader for passing untransformed vertices to + // OpenGL for rendering. This should be removed/replaced when we start emulating Maxwell + // shaders. + + static constexpr char vertex_shader[] = R"( +#version 150 core + +in vec2 vert_position; +in vec2 vert_tex_coord; +out vec2 frag_tex_coord; + +void main() { + // Multiply input position by the rotscale part of the matrix and then manually translate by + // the last column. This is equivalent to using a full 3x3 matrix and expanding the vector + // to `vec3(vert_position.xy, 1.0)` + gl_Position = vec4(mat2(mat3x2(0.0015625f, 0.0, 0.0, -0.0027778, -1.0, 1.0)) * vert_position + mat3x2(0.0015625f, 0.0, 0.0, -0.0027778, -1.0, 1.0)[2], 0.0, 1.0); + frag_tex_coord = vert_tex_coord; +} +)"; + + static constexpr char fragment_shader[] = R"( +#version 150 core + +in vec2 frag_tex_coord; +out vec4 color; + +uniform sampler2D color_texture; + +void main() { + color = vec4(1.0, 0.0, 1.0, 0.0); +} +)"; + + if (current_shader) { + return; + } + + LOG_CRITICAL(Render_OpenGL, "Emulated shaders are not supported! Using a passthrough shader."); + + current_shader = &test_shader; + if (has_ARB_separate_shader_objects) { + test_shader.shader.Create(vertex_shader, nullptr, fragment_shader, {}, true); + glActiveShaderProgram(pipeline.handle, test_shader.shader.handle); + } else { + UNREACHABLE(); + } + + state.draw.shader_program = test_shader.shader.handle; + state.Apply(); + + if (has_ARB_separate_shader_objects) { + state.draw.shader_program = 0; + state.Apply(); + } +} + +void RasterizerOpenGL::BindFramebufferSurfaces(const Surface& color_surface, + const Surface& depth_surface, bool has_stencil) { + state.draw.draw_framebuffer = framebuffer.handle; + state.Apply(); + + glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, + color_surface != nullptr ? color_surface->texture.handle : 0, 0); + if (depth_surface != nullptr) { + if (has_stencil) { + // 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); + } +} + +void RasterizerOpenGL::SyncViewport(const MathUtil::Rectangle<u32>& surfaces_rect, u16 res_scale) { + const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs; + const MathUtil::Rectangle<s32> viewport_rect{regs.viewport[0].GetRect()}; + + state.viewport.x = static_cast<GLint>(surfaces_rect.left) + viewport_rect.left * res_scale; + state.viewport.y = static_cast<GLint>(surfaces_rect.bottom) + viewport_rect.bottom * res_scale; + state.viewport.width = static_cast<GLsizei>(viewport_rect.GetWidth() * res_scale); + state.viewport.height = static_cast<GLsizei>(viewport_rect.GetHeight() * res_scale); +} + +void RasterizerOpenGL::SyncClipEnabled() { + UNREACHABLE(); +} + +void RasterizerOpenGL::SyncClipCoef() { + UNREACHABLE(); +} + +void RasterizerOpenGL::SyncCullMode() { + UNREACHABLE(); +} + +void RasterizerOpenGL::SyncDepthScale() { + UNREACHABLE(); +} + +void RasterizerOpenGL::SyncDepthOffset() { + UNREACHABLE(); +} + +void RasterizerOpenGL::SyncBlendEnabled() { + UNREACHABLE(); +} + +void RasterizerOpenGL::SyncBlendFuncs() { + UNREACHABLE(); +} + +void RasterizerOpenGL::SyncBlendColor() { + UNREACHABLE(); +} diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h new file mode 100644 index 000000000..fd53e94cd --- /dev/null +++ b/src/video_core/renderer_opengl/gl_rasterizer.h @@ -0,0 +1,179 @@ +// Copyright 2015 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <array> +#include <cstddef> +#include <cstring> +#include <memory> +#include <unordered_map> +#include <vector> +#include <glad/glad.h> +#include "common/bit_field.h" +#include "common/common_types.h" +#include "common/hash.h" +#include "common/vector_math.h" +#include "video_core/rasterizer_interface.h" +#include "video_core/renderer_opengl/gl_rasterizer_cache.h" +#include "video_core/renderer_opengl/gl_resource_manager.h" +#include "video_core/renderer_opengl/gl_shader_gen.h" +#include "video_core/renderer_opengl/gl_state.h" +#include "video_core/renderer_opengl/gl_stream_buffer.h" + +struct ScreenInfo; + +class RasterizerOpenGL : public VideoCore::RasterizerInterface { +public: + RasterizerOpenGL(); + ~RasterizerOpenGL() override; + + void DrawArrays() override; + void NotifyMaxwellRegisterChanged(u32 id) override; + void FlushAll() override; + void FlushRegion(VAddr addr, u64 size) override; + void InvalidateRegion(VAddr addr, u64 size) override; + void FlushAndInvalidateRegion(VAddr addr, u64 size) override; + bool AccelerateDisplayTransfer(const void* config) override; + bool AccelerateTextureCopy(const void* config) override; + bool AccelerateFill(const void* config) override; + bool AccelerateDisplay(const Tegra::FramebufferConfig& framebuffer, VAddr framebuffer_addr, + u32 pixel_stride, ScreenInfo& screen_info) override; + bool AccelerateDrawBatch(bool is_indexed) override; + + /// OpenGL shader generated for a given Maxwell register state + struct MaxwellShader { + /// OpenGL shader resource + OGLShader shader; + }; + + struct VertexShader { + OGLShader shader; + }; + + struct FragmentShader { + OGLShader shader; + }; + + /// Uniform structure for the Uniform Buffer Object, all vectors must be 16-byte aligned + // NOTE: Always keep a vec4 at the end. The GL spec is not clear wether the alignment at + // the end of a uniform block is included in UNIFORM_BLOCK_DATA_SIZE or not. + // Not following that rule will cause problems on some AMD drivers. + struct UniformData {}; + + // static_assert( + // sizeof(UniformData) == 0x460, + // "The size of the UniformData structure has changed, update the structure in the shader"); + static_assert(sizeof(UniformData) < 16384, + "UniformData structure must be less than 16kb as per the OpenGL spec"); + + struct VSUniformData {}; + // static_assert( + // sizeof(VSUniformData) == 1856, + // "The size of the VSUniformData structure has changed, update the structure in the + // shader"); + static_assert(sizeof(VSUniformData) < 16384, + "VSUniformData structure must be less than 16kb as per the OpenGL spec"); + + struct FSUniformData {}; + // static_assert( + // sizeof(FSUniformData) == 1856, + // "The size of the FSUniformData structure has changed, update the structure in the + // shader"); + static_assert(sizeof(FSUniformData) < 16384, + "FSUniformData structure must be less than 16kb as per the OpenGL spec"); + +private: + struct SamplerInfo {}; + + /// Binds the framebuffer color and depth surface + void BindFramebufferSurfaces(const Surface& color_surface, const Surface& depth_surface, + bool has_stencil); + + /// Syncs the viewport to match the guest state + void SyncViewport(const MathUtil::Rectangle<u32>& surfaces_rect, u16 res_scale); + + /// Syncs the clip enabled status to match the guest state + void SyncClipEnabled(); + + /// Syncs the clip coefficients to match the guest state + void SyncClipCoef(); + + /// Sets the OpenGL shader in accordance with the current guest state + void SetShader(); + + /// Syncs the cull mode to match the guest state + void SyncCullMode(); + + /// Syncs the depth scale to match the guest state + void SyncDepthScale(); + + /// Syncs the depth offset to match the guest state + void SyncDepthOffset(); + + /// Syncs the blend enabled status to match the guest state + void SyncBlendEnabled(); + + /// Syncs the blend functions to match the guest state + void SyncBlendFuncs(); + + /// Syncs the blend color to match the guest state + void SyncBlendColor(); + + bool has_ARB_buffer_storage; + bool has_ARB_direct_state_access; + bool has_ARB_separate_shader_objects; + bool has_ARB_vertex_attrib_binding; + + OpenGLState state; + + RasterizerCacheOpenGL res_cache; + + /// Shader used for test renderering - to be removed once we have emulated shaders + MaxwellShader test_shader{}; + + const MaxwellShader* current_shader{}; + bool shader_dirty{}; + + struct { + UniformData data; + bool dirty; + } uniform_block_data = {}; + + OGLPipeline pipeline; + OGLVertexArray sw_vao; + OGLVertexArray hw_vao; + std::array<bool, 16> hw_vao_enabled_attributes; + + std::array<SamplerInfo, 32> texture_samplers; + static constexpr size_t VERTEX_BUFFER_SIZE = 128 * 1024 * 1024; + std::unique_ptr<OGLStreamBuffer> vertex_buffer; + OGLBuffer uniform_buffer; + OGLFramebuffer framebuffer; + + static constexpr size_t STREAM_BUFFER_SIZE = 4 * 1024 * 1024; + std::unique_ptr<OGLStreamBuffer> stream_buffer; + + GLsizeiptr vs_input_size; + + void AnalyzeVertexArray(bool is_indexed); + void SetupVertexArray(u8* array_ptr, GLintptr buffer_offset); + + OGLBuffer vs_uniform_buffer; + std::unordered_map<GLShader::MaxwellVSConfig, VertexShader*> vs_shader_map; + std::unordered_map<std::string, VertexShader> vs_shader_cache; + OGLShader vs_default_shader; + + void SetupVertexShader(VSUniformData* ub_ptr, GLintptr buffer_offset); + + OGLBuffer fs_uniform_buffer; + std::unordered_map<GLShader::MaxwellFSConfig, FragmentShader*> fs_shader_map; + std::unordered_map<std::string, FragmentShader> fs_shader_cache; + OGLShader fs_default_shader; + + void SetupFragmentShader(FSUniformData* ub_ptr, GLintptr buffer_offset); + + enum class AccelDraw { Disabled, Arrays, Indexed }; + AccelDraw accelerate_draw; +}; diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp new file mode 100644 index 000000000..2ffbd3bab --- /dev/null +++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp @@ -0,0 +1,1432 @@ +// Copyright 2015 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <algorithm> +#include <atomic> +#include <cstring> +#include <iterator> +#include <memory> +#include <unordered_set> +#include <utility> +#include <vector> +#include <boost/optional.hpp> +#include <boost/range/iterator_range.hpp> +#include <glad/glad.h> +#include "common/alignment.h" +#include "common/bit_field.h" +#include "common/color.h" +#include "common/logging/log.h" +#include "common/math_util.h" +#include "common/microprofile.h" +#include "common/scope_exit.h" +#include "common/vector_math.h" +#include "core/core.h" +#include "core/frontend/emu_window.h" +#include "core/hle/kernel/process.h" +#include "core/hle/kernel/vm_manager.h" +#include "core/memory.h" +#include "core/settings.h" +#include "video_core/engines/maxwell_3d.h" +#include "video_core/renderer_opengl/gl_rasterizer_cache.h" +#include "video_core/renderer_opengl/gl_state.h" +#include "video_core/utils.h" +#include "video_core/video_core.h" + +using SurfaceType = SurfaceParams::SurfaceType; +using PixelFormat = SurfaceParams::PixelFormat; + +struct FormatTuple { + GLint internal_format; + GLenum format; + GLenum type; +}; + +static constexpr std::array<FormatTuple, 5> fb_format_tuples = {{ + {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8}, // RGBA8 + {GL_RGB8, GL_BGR, GL_UNSIGNED_BYTE}, // RGB8 + {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_5_5_5_1}, // RGB5A1 + {GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5}, // RGB565 + {GL_RGBA4, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4}, // RGBA4 +}}; + +static constexpr std::array<FormatTuple, 4> depth_format_tuples = {{ + {GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT}, // D16 + {}, + {GL_DEPTH_COMPONENT24, GL_DEPTH_COMPONENT, GL_UNSIGNED_INT}, // D24 + {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8}, // D24S8 +}}; + +static constexpr FormatTuple tex_tuple = {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE}; + +static const FormatTuple& GetFormatTuple(PixelFormat pixel_format) { + const SurfaceType type = SurfaceParams::GetFormatType(pixel_format); + if (type == SurfaceType::Color) { + ASSERT(static_cast<size_t>(pixel_format) < fb_format_tuples.size()); + return fb_format_tuples[static_cast<unsigned int>(pixel_format)]; + } else if (type == SurfaceType::Depth || type == SurfaceType::DepthStencil) { + size_t tuple_idx = static_cast<size_t>(pixel_format) - 14; + ASSERT(tuple_idx < depth_format_tuples.size()); + return depth_format_tuples[tuple_idx]; + } + return tex_tuple; +} + +template <typename Map, typename Interval> +constexpr auto RangeFromInterval(Map& map, const Interval& interval) { + return boost::make_iterator_range(map.equal_range(interval)); +} + +static u16 GetResolutionScaleFactor() { + return static_cast<u16>(!Settings::values.resolution_factor + ? VideoCore::g_emu_window->GetFramebufferLayout().GetScalingRatio() + : Settings::values.resolution_factor); +} + +template <bool morton_to_gl, PixelFormat format> +static void MortonCopyTile(u32 stride, u8* tile_buffer, u8* gl_buffer) { + constexpr u32 bytes_per_pixel = SurfaceParams::GetFormatBpp(format) / 8; + constexpr u32 gl_bytes_per_pixel = CachedSurface::GetGLBytesPerPixel(format); + for (u32 y = 0; y < 8; ++y) { + for (u32 x = 0; x < 8; ++x) { + u8* tile_ptr = tile_buffer + VideoCore::MortonInterleave(x, y) * bytes_per_pixel; + u8* gl_ptr = gl_buffer + ((7 - y) * stride + x) * gl_bytes_per_pixel; + if (morton_to_gl) { + if (format == PixelFormat::D24S8) { + gl_ptr[0] = tile_ptr[3]; + std::memcpy(gl_ptr + 1, tile_ptr, 3); + } else { + std::memcpy(gl_ptr, tile_ptr, bytes_per_pixel); + } + } else { + if (format == PixelFormat::D24S8) { + std::memcpy(tile_ptr, gl_ptr + 1, 3); + tile_ptr[3] = gl_ptr[0]; + } else { + std::memcpy(tile_ptr, gl_ptr, bytes_per_pixel); + } + } + } + } +} + +template <bool morton_to_gl, PixelFormat format> +static void MortonCopy(u32 stride, u32 height, u8* gl_buffer, VAddr base, VAddr start, VAddr end) { + constexpr u32 bytes_per_pixel = SurfaceParams::GetFormatBpp(format) / 8; + constexpr u32 gl_bytes_per_pixel = CachedSurface::GetGLBytesPerPixel(format); + + // TODO(bunnei): Assumes the default rendering GOB size of 16 (128 lines). We should check the + // configuration for this and perform more generic un/swizzle + LOG_WARNING(Render_OpenGL, "need to use correct swizzle/GOB parameters!"); + VideoCore::MortonCopyPixels128(stride, height, bytes_per_pixel, gl_bytes_per_pixel, + Memory::GetPointer(base), gl_buffer, morton_to_gl); +} + +static constexpr std::array<void (*)(u32, u32, u8*, VAddr, VAddr, VAddr), 18> morton_to_gl_fns = { + MortonCopy<true, PixelFormat::RGBA8>, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, +}; + +static constexpr std::array<void (*)(u32, u32, u8*, VAddr, VAddr, VAddr), 18> gl_to_morton_fns = { + MortonCopy<false, PixelFormat::RGBA8>, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, + nullptr, +}; + +// Allocate an uninitialized texture of appropriate size and format for the surface +static void AllocateSurfaceTexture(GLuint texture, const FormatTuple& format_tuple, u32 width, + u32 height) { + OpenGLState cur_state = OpenGLState::GetCurState(); + + // Keep track of previous texture bindings + GLuint old_tex = cur_state.texture_units[0].texture_2d; + cur_state.texture_units[0].texture_2d = texture; + cur_state.Apply(); + glActiveTexture(GL_TEXTURE0); + + glTexImage2D(GL_TEXTURE_2D, 0, format_tuple.internal_format, width, height, 0, + format_tuple.format, format_tuple.type, nullptr); + + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 0); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); + glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); + + // Restore previous texture bindings + cur_state.texture_units[0].texture_2d = old_tex; + cur_state.Apply(); +} + +static bool BlitTextures(GLuint src_tex, const MathUtil::Rectangle<u32>& src_rect, GLuint dst_tex, + const MathUtil::Rectangle<u32>& dst_rect, SurfaceType type, + GLuint read_fb_handle, GLuint draw_fb_handle) { + OpenGLState state = OpenGLState::GetCurState(); + + OpenGLState prev_state = state; + SCOPE_EXIT({ prev_state.Apply(); }); + + // Make sure textures aren't bound to texture units, since going to bind them to framebuffer + // components + state.ResetTexture(src_tex); + state.ResetTexture(dst_tex); + + state.draw.read_framebuffer = read_fb_handle; + state.draw.draw_framebuffer = draw_fb_handle; + state.Apply(); + + u32 buffers = 0; + + if (type == SurfaceType::Color || type == SurfaceType::Texture) { + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, src_tex, + 0); + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, + 0); + + glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, dst_tex, + 0); + glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, + 0); + + buffers = GL_COLOR_BUFFER_BIT; + } else if (type == SurfaceType::Depth) { + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0); + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D, src_tex, 0); + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0); + + glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0); + glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D, dst_tex, 0); + glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0); + + buffers = GL_DEPTH_BUFFER_BIT; + } else if (type == SurfaceType::DepthStencil) { + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0); + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, + src_tex, 0); + + glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0); + glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, + dst_tex, 0); + + buffers = GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT; + } + + glBlitFramebuffer(src_rect.left, src_rect.bottom, src_rect.right, src_rect.top, dst_rect.left, + dst_rect.bottom, dst_rect.right, dst_rect.top, buffers, + buffers == GL_COLOR_BUFFER_BIT ? GL_LINEAR : GL_NEAREST); + + return true; +} + +static bool FillSurface(const Surface& surface, const u8* fill_data, + const MathUtil::Rectangle<u32>& fill_rect, GLuint draw_fb_handle) { + UNREACHABLE(); + return {}; +} + +SurfaceParams SurfaceParams::FromInterval(SurfaceInterval interval) const { + SurfaceParams params = *this; + const u32 tiled_size = is_tiled ? 8 : 1; + const u64 stride_tiled_bytes = BytesInPixels(stride * tiled_size); + VAddr aligned_start = + addr + Common::AlignDown(boost::icl::first(interval) - addr, stride_tiled_bytes); + VAddr aligned_end = + addr + Common::AlignUp(boost::icl::last_next(interval) - addr, stride_tiled_bytes); + + if (aligned_end - aligned_start > stride_tiled_bytes) { + params.addr = aligned_start; + params.height = static_cast<u32>((aligned_end - aligned_start) / BytesInPixels(stride)); + } else { + // 1 row + ASSERT(aligned_end - aligned_start == stride_tiled_bytes); + const u64 tiled_alignment = BytesInPixels(is_tiled ? 8 * 8 : 1); + aligned_start = + addr + Common::AlignDown(boost::icl::first(interval) - addr, tiled_alignment); + aligned_end = + addr + Common::AlignUp(boost::icl::last_next(interval) - addr, tiled_alignment); + params.addr = aligned_start; + params.width = static_cast<u32>(PixelsInBytes(aligned_end - aligned_start) / tiled_size); + params.stride = params.width; + params.height = tiled_size; + } + params.UpdateParams(); + + return params; +} + +SurfaceInterval SurfaceParams::GetSubRectInterval(MathUtil::Rectangle<u32> unscaled_rect) const { + if (unscaled_rect.GetHeight() == 0 || unscaled_rect.GetWidth() == 0) { + return {}; + } + + if (is_tiled) { + unscaled_rect.left = Common::AlignDown(unscaled_rect.left, 8) * 8; + unscaled_rect.bottom = Common::AlignDown(unscaled_rect.bottom, 8) / 8; + unscaled_rect.right = Common::AlignUp(unscaled_rect.right, 8) * 8; + unscaled_rect.top = Common::AlignUp(unscaled_rect.top, 8) / 8; + } + + const u32 stride_tiled = !is_tiled ? stride : stride * 8; + + const u32 pixel_offset = + stride_tiled * (!is_tiled ? unscaled_rect.bottom : (height / 8) - unscaled_rect.top) + + unscaled_rect.left; + + const u32 pixels = (unscaled_rect.GetHeight() - 1) * stride_tiled + unscaled_rect.GetWidth(); + + return {addr + BytesInPixels(pixel_offset), addr + BytesInPixels(pixel_offset + pixels)}; +} + +MathUtil::Rectangle<u32> SurfaceParams::GetSubRect(const SurfaceParams& sub_surface) const { + const u32 begin_pixel_index = static_cast<u32>(PixelsInBytes(sub_surface.addr - addr)); + + if (is_tiled) { + const int x0 = (begin_pixel_index % (stride * 8)) / 8; + const int y0 = (begin_pixel_index / (stride * 8)) * 8; + // Top to bottom + return MathUtil::Rectangle<u32>(x0, height - y0, x0 + sub_surface.width, + height - (y0 + sub_surface.height)); + } + + const int x0 = begin_pixel_index % stride; + const int y0 = begin_pixel_index / stride; + // Bottom to top + return MathUtil::Rectangle<u32>(x0, y0 + sub_surface.height, x0 + sub_surface.width, y0); +} + +MathUtil::Rectangle<u32> SurfaceParams::GetScaledSubRect(const SurfaceParams& sub_surface) const { + auto rect = GetSubRect(sub_surface); + rect.left = rect.left * res_scale; + rect.right = rect.right * res_scale; + rect.top = rect.top * res_scale; + rect.bottom = rect.bottom * res_scale; + return rect; +} + +bool SurfaceParams::ExactMatch(const SurfaceParams& other_surface) const { + return std::tie(other_surface.addr, other_surface.width, other_surface.height, + other_surface.stride, other_surface.pixel_format, other_surface.is_tiled) == + std::tie(addr, width, height, stride, pixel_format, is_tiled) && + pixel_format != PixelFormat::Invalid; +} + +bool SurfaceParams::CanSubRect(const SurfaceParams& sub_surface) const { + return sub_surface.addr >= addr && sub_surface.end <= end && + sub_surface.pixel_format == pixel_format && pixel_format != PixelFormat::Invalid && + sub_surface.is_tiled == is_tiled && + (sub_surface.addr - addr) % BytesInPixels(is_tiled ? 64 : 1) == 0 && + (sub_surface.stride == stride || sub_surface.height <= (is_tiled ? 8u : 1u)) && + GetSubRect(sub_surface).left + sub_surface.width <= stride; +} + +bool SurfaceParams::CanExpand(const SurfaceParams& expanded_surface) const { + return pixel_format != PixelFormat::Invalid && pixel_format == expanded_surface.pixel_format && + addr <= expanded_surface.end && expanded_surface.addr <= end && + is_tiled == expanded_surface.is_tiled && stride == expanded_surface.stride && + (std::max(expanded_surface.addr, addr) - std::min(expanded_surface.addr, addr)) % + BytesInPixels(stride * (is_tiled ? 8 : 1)) == + 0; +} + +bool SurfaceParams::CanTexCopy(const SurfaceParams& texcopy_params) const { + if (pixel_format == PixelFormat::Invalid || addr > texcopy_params.addr || + end < texcopy_params.end) { + return false; + } + if (texcopy_params.width != texcopy_params.stride) { + const u32 tile_stride = static_cast<u32>(BytesInPixels(stride * (is_tiled ? 8 : 1))); + return (texcopy_params.addr - addr) % BytesInPixels(is_tiled ? 64 : 1) == 0 && + texcopy_params.width % BytesInPixels(is_tiled ? 64 : 1) == 0 && + (texcopy_params.height == 1 || texcopy_params.stride == tile_stride) && + ((texcopy_params.addr - addr) % tile_stride) + texcopy_params.width <= tile_stride; + } + return FromInterval(texcopy_params.GetInterval()).GetInterval() == texcopy_params.GetInterval(); +} + +bool CachedSurface::CanFill(const SurfaceParams& dest_surface, + SurfaceInterval fill_interval) const { + if (type == SurfaceType::Fill && IsRegionValid(fill_interval) && + boost::icl::first(fill_interval) >= addr && + boost::icl::last_next(fill_interval) <= end && // dest_surface is within our fill range + dest_surface.FromInterval(fill_interval).GetInterval() == + fill_interval) { // make sure interval is a rectangle in dest surface + if (fill_size * 8 != dest_surface.GetFormatBpp()) { + // Check if bits repeat for our fill_size + const u32 dest_bytes_per_pixel = std::max(dest_surface.GetFormatBpp() / 8, 1u); + std::vector<u8> fill_test(fill_size * dest_bytes_per_pixel); + + for (u32 i = 0; i < dest_bytes_per_pixel; ++i) + std::memcpy(&fill_test[i * fill_size], &fill_data[0], fill_size); + + for (u32 i = 0; i < fill_size; ++i) + if (std::memcmp(&fill_test[dest_bytes_per_pixel * i], &fill_test[0], + dest_bytes_per_pixel) != 0) + return false; + + if (dest_surface.GetFormatBpp() == 4 && (fill_test[0] & 0xF) != (fill_test[0] >> 4)) + return false; + } + return true; + } + return false; +} + +bool CachedSurface::CanCopy(const SurfaceParams& dest_surface, + SurfaceInterval copy_interval) const { + SurfaceParams subrect_params = dest_surface.FromInterval(copy_interval); + ASSERT(subrect_params.GetInterval() == copy_interval); + if (CanSubRect(subrect_params)) + return true; + + if (CanFill(dest_surface, copy_interval)) + return true; + + return false; +} + +SurfaceInterval SurfaceParams::GetCopyableInterval(const Surface& src_surface) const { + SurfaceInterval result{}; + const auto valid_regions = + SurfaceRegions(GetInterval() & src_surface->GetInterval()) - src_surface->invalid_regions; + for (auto& valid_interval : valid_regions) { + const SurfaceInterval aligned_interval{ + addr + Common::AlignUp(boost::icl::first(valid_interval) - addr, + BytesInPixels(is_tiled ? 8 * 8 : 1)), + addr + Common::AlignDown(boost::icl::last_next(valid_interval) - addr, + BytesInPixels(is_tiled ? 8 * 8 : 1))}; + + if (BytesInPixels(is_tiled ? 8 * 8 : 1) > boost::icl::length(valid_interval) || + boost::icl::length(aligned_interval) == 0) { + continue; + } + + // Get the rectangle within aligned_interval + const u32 stride_bytes = static_cast<u32>(BytesInPixels(stride)) * (is_tiled ? 8 : 1); + SurfaceInterval rect_interval{ + addr + Common::AlignUp(boost::icl::first(aligned_interval) - addr, stride_bytes), + addr + Common::AlignDown(boost::icl::last_next(aligned_interval) - addr, stride_bytes), + }; + if (boost::icl::first(rect_interval) > boost::icl::last_next(rect_interval)) { + // 1 row + rect_interval = aligned_interval; + } else if (boost::icl::length(rect_interval) == 0) { + // 2 rows that do not make a rectangle, return the larger one + const SurfaceInterval row1{boost::icl::first(aligned_interval), + boost::icl::first(rect_interval)}; + const SurfaceInterval row2{boost::icl::first(rect_interval), + boost::icl::last_next(aligned_interval)}; + rect_interval = (boost::icl::length(row1) > boost::icl::length(row2)) ? row1 : row2; + } + + if (boost::icl::length(rect_interval) > boost::icl::length(result)) { + result = rect_interval; + } + } + return result; +} + +void RasterizerCacheOpenGL::CopySurface(const Surface& src_surface, const Surface& dst_surface, + SurfaceInterval copy_interval) { + SurfaceParams subrect_params = dst_surface->FromInterval(copy_interval); + ASSERT(subrect_params.GetInterval() == copy_interval); + + ASSERT(src_surface != dst_surface); + + // This is only called when CanCopy is true, no need to run checks here + if (src_surface->type == SurfaceType::Fill) { + // FillSurface needs a 4 bytes buffer + const u64 fill_offset = + (boost::icl::first(copy_interval) - src_surface->addr) % src_surface->fill_size; + std::array<u8, 4> fill_buffer; + + u64 fill_buff_pos = fill_offset; + for (int i : {0, 1, 2, 3}) + fill_buffer[i] = src_surface->fill_data[fill_buff_pos++ % src_surface->fill_size]; + + FillSurface(dst_surface, &fill_buffer[0], dst_surface->GetScaledSubRect(subrect_params), + draw_framebuffer.handle); + return; + } + if (src_surface->CanSubRect(subrect_params)) { + BlitTextures(src_surface->texture.handle, src_surface->GetScaledSubRect(subrect_params), + dst_surface->texture.handle, dst_surface->GetScaledSubRect(subrect_params), + src_surface->type, read_framebuffer.handle, draw_framebuffer.handle); + return; + } + UNREACHABLE(); +} + +MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64, 192)); +void CachedSurface::LoadGLBuffer(VAddr load_start, VAddr load_end) { + ASSERT(type != SurfaceType::Fill); + + u8* const texture_src_data = Memory::GetPointer(addr); + if (texture_src_data == nullptr) + return; + + if (gl_buffer == nullptr) { + gl_buffer_size = width * height * GetGLBytesPerPixel(pixel_format); + gl_buffer.reset(new u8[gl_buffer_size]); + } + + MICROPROFILE_SCOPE(OpenGL_SurfaceLoad); + + ASSERT(load_start >= addr && load_end <= end); + const u64 start_offset = load_start - addr; + + if (!is_tiled) { + ASSERT(type == SurfaceType::Color); + const u32 bytes_per_pixel{GetFormatBpp() >> 3}; + + // TODO(bunnei): Assumes the default rendering GOB size of 16 (128 lines). We should check + // the configuration for this and perform more generic un/swizzle + LOG_WARNING(Render_OpenGL, "need to use correct swizzle/GOB parameters!"); + VideoCore::MortonCopyPixels128(width, height, bytes_per_pixel, 4, + texture_src_data + start_offset, &gl_buffer[start_offset], + true); + } else { + morton_to_gl_fns[static_cast<size_t>(pixel_format)](stride, height, &gl_buffer[0], addr, + load_start, load_end); + } +} + +MICROPROFILE_DEFINE(OpenGL_SurfaceFlush, "OpenGL", "Surface Flush", MP_RGB(128, 192, 64)); +void CachedSurface::FlushGLBuffer(VAddr flush_start, VAddr flush_end) { + u8* const dst_buffer = Memory::GetPointer(addr); + if (dst_buffer == nullptr) + return; + + ASSERT(gl_buffer_size == width * height * GetGLBytesPerPixel(pixel_format)); + + // TODO: Should probably be done in ::Memory:: and check for other regions too + // same as loadglbuffer() + if (flush_start < Memory::VRAM_VADDR_END && flush_end > Memory::VRAM_VADDR_END) + flush_end = Memory::VRAM_VADDR_END; + + if (flush_start < Memory::VRAM_VADDR && flush_end > Memory::VRAM_VADDR) + flush_start = Memory::VRAM_VADDR; + + MICROPROFILE_SCOPE(OpenGL_SurfaceFlush); + + ASSERT(flush_start >= addr && flush_end <= end); + const u64 start_offset = flush_start - addr; + const u64 end_offset = flush_end - addr; + + if (type == SurfaceType::Fill) { + const u64 coarse_start_offset = start_offset - (start_offset % fill_size); + const u64 backup_bytes = start_offset % fill_size; + std::array<u8, 4> backup_data; + if (backup_bytes) + std::memcpy(&backup_data[0], &dst_buffer[coarse_start_offset], backup_bytes); + + for (u64 offset = coarse_start_offset; offset < end_offset; offset += fill_size) { + std::memcpy(&dst_buffer[offset], &fill_data[0], + std::min(fill_size, end_offset - offset)); + } + + if (backup_bytes) + std::memcpy(&dst_buffer[coarse_start_offset], &backup_data[0], backup_bytes); + } else if (!is_tiled) { + ASSERT(type == SurfaceType::Color); + std::memcpy(dst_buffer + start_offset, &gl_buffer[start_offset], flush_end - flush_start); + } else { + gl_to_morton_fns[static_cast<size_t>(pixel_format)](stride, height, &gl_buffer[0], addr, + flush_start, flush_end); + } +} + +MICROPROFILE_DEFINE(OpenGL_TextureUL, "OpenGL", "Texture Upload", MP_RGB(128, 64, 192)); +void CachedSurface::UploadGLTexture(const MathUtil::Rectangle<u32>& rect, GLuint read_fb_handle, + GLuint draw_fb_handle) { + if (type == SurfaceType::Fill) + return; + + MICROPROFILE_SCOPE(OpenGL_TextureUL); + + ASSERT(gl_buffer_size == width * height * GetGLBytesPerPixel(pixel_format)); + + // Load data from memory to the surface + GLint x0 = static_cast<GLint>(rect.left); + GLint y0 = static_cast<GLint>(rect.bottom); + size_t buffer_offset = (y0 * stride + x0) * GetGLBytesPerPixel(pixel_format); + + const FormatTuple& tuple = GetFormatTuple(pixel_format); + GLuint target_tex = texture.handle; + + // If not 1x scale, create 1x texture that we will blit from to replace texture subrect in + // surface + OGLTexture unscaled_tex; + if (res_scale != 1) { + x0 = 0; + y0 = 0; + + unscaled_tex.Create(); + AllocateSurfaceTexture(unscaled_tex.handle, tuple, rect.GetWidth(), rect.GetHeight()); + target_tex = unscaled_tex.handle; + } + + OpenGLState cur_state = OpenGLState::GetCurState(); + + GLuint old_tex = cur_state.texture_units[0].texture_2d; + cur_state.texture_units[0].texture_2d = target_tex; + cur_state.Apply(); + + // Ensure no bad interactions with GL_UNPACK_ALIGNMENT + ASSERT(stride * GetGLBytesPerPixel(pixel_format) % 4 == 0); + glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(stride)); + + glActiveTexture(GL_TEXTURE0); + glTexSubImage2D(GL_TEXTURE_2D, 0, x0, y0, static_cast<GLsizei>(rect.GetWidth()), + static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type, + &gl_buffer[buffer_offset]); + + glPixelStorei(GL_UNPACK_ROW_LENGTH, 0); + + cur_state.texture_units[0].texture_2d = old_tex; + cur_state.Apply(); + + if (res_scale != 1) { + auto scaled_rect = rect; + scaled_rect.left *= res_scale; + scaled_rect.top *= res_scale; + scaled_rect.right *= res_scale; + scaled_rect.bottom *= res_scale; + + BlitTextures(unscaled_tex.handle, {0, rect.GetHeight(), rect.GetWidth(), 0}, texture.handle, + scaled_rect, type, read_fb_handle, draw_fb_handle); + } +} + +MICROPROFILE_DEFINE(OpenGL_TextureDL, "OpenGL", "Texture Download", MP_RGB(128, 192, 64)); +void CachedSurface::DownloadGLTexture(const MathUtil::Rectangle<u32>& rect, GLuint read_fb_handle, + GLuint draw_fb_handle) { + if (type == SurfaceType::Fill) + return; + + MICROPROFILE_SCOPE(OpenGL_TextureDL); + + if (gl_buffer == nullptr) { + gl_buffer_size = width * height * GetGLBytesPerPixel(pixel_format); + gl_buffer.reset(new u8[gl_buffer_size]); + } + + OpenGLState state = OpenGLState::GetCurState(); + OpenGLState prev_state = state; + SCOPE_EXIT({ prev_state.Apply(); }); + + const FormatTuple& tuple = GetFormatTuple(pixel_format); + + // Ensure no bad interactions with GL_PACK_ALIGNMENT + ASSERT(stride * GetGLBytesPerPixel(pixel_format) % 4 == 0); + glPixelStorei(GL_PACK_ROW_LENGTH, static_cast<GLint>(stride)); + size_t buffer_offset = (rect.bottom * stride + rect.left) * GetGLBytesPerPixel(pixel_format); + + // If not 1x scale, blit scaled texture to a new 1x texture and use that to flush + if (res_scale != 1) { + auto scaled_rect = rect; + scaled_rect.left *= res_scale; + scaled_rect.top *= res_scale; + scaled_rect.right *= res_scale; + scaled_rect.bottom *= res_scale; + + OGLTexture unscaled_tex; + unscaled_tex.Create(); + + MathUtil::Rectangle<u32> unscaled_tex_rect{0, rect.GetHeight(), rect.GetWidth(), 0}; + AllocateSurfaceTexture(unscaled_tex.handle, tuple, rect.GetWidth(), rect.GetHeight()); + BlitTextures(texture.handle, scaled_rect, unscaled_tex.handle, unscaled_tex_rect, type, + read_fb_handle, draw_fb_handle); + + state.texture_units[0].texture_2d = unscaled_tex.handle; + state.Apply(); + + glActiveTexture(GL_TEXTURE0); + glGetTexImage(GL_TEXTURE_2D, 0, tuple.format, tuple.type, &gl_buffer[buffer_offset]); + } else { + state.ResetTexture(texture.handle); + state.draw.read_framebuffer = read_fb_handle; + state.Apply(); + + if (type == SurfaceType::Color || type == SurfaceType::Texture) { + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, + texture.handle, 0); + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, + 0, 0); + } else if (type == SurfaceType::Depth) { + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0); + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D, + texture.handle, 0); + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0); + } else { + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0); + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, + texture.handle, 0); + } + glReadPixels(static_cast<GLint>(rect.left), static_cast<GLint>(rect.bottom), + static_cast<GLsizei>(rect.GetWidth()), static_cast<GLsizei>(rect.GetHeight()), + tuple.format, tuple.type, &gl_buffer[buffer_offset]); + } + + glPixelStorei(GL_PACK_ROW_LENGTH, 0); +} + +enum MatchFlags { + Invalid = 1, // Flag that can be applied to other match types, invalid matches require + // validation before they can be used + Exact = 1 << 1, // Surfaces perfectly match + SubRect = 1 << 2, // Surface encompasses params + Copy = 1 << 3, // Surface we can copy from + Expand = 1 << 4, // Surface that can expand params + TexCopy = 1 << 5 // Surface that will match a display transfer "texture copy" parameters +}; + +constexpr MatchFlags operator|(MatchFlags lhs, MatchFlags rhs) { + return static_cast<MatchFlags>(static_cast<int>(lhs) | static_cast<int>(rhs)); +} + +/// Get the best surface match (and its match type) for the given flags +template <MatchFlags find_flags> +Surface FindMatch(const SurfaceCache& surface_cache, const SurfaceParams& params, + ScaleMatch match_scale_type, + boost::optional<SurfaceInterval> validate_interval = boost::none) { + Surface match_surface = nullptr; + bool match_valid = false; + u32 match_scale = 0; + SurfaceInterval match_interval{}; + + for (auto& pair : RangeFromInterval(surface_cache, params.GetInterval())) { + for (auto& surface : pair.second) { + bool res_scale_matched = match_scale_type == ScaleMatch::Exact + ? (params.res_scale == surface->res_scale) + : (params.res_scale <= surface->res_scale); + // validity will be checked in GetCopyableInterval + bool is_valid = + find_flags & MatchFlags::Copy + ? true + : surface->IsRegionValid(validate_interval.value_or(params.GetInterval())); + + if (!(find_flags & MatchFlags::Invalid) && !is_valid) + continue; + + auto IsMatch_Helper = [&](auto check_type, auto match_fn) { + if (!(find_flags & check_type)) + return; + + bool matched; + SurfaceInterval surface_interval; + std::tie(matched, surface_interval) = match_fn(); + if (!matched) + return; + + if (!res_scale_matched && match_scale_type != ScaleMatch::Ignore && + surface->type != SurfaceType::Fill) + return; + + // Found a match, update only if this is better than the previous one + auto UpdateMatch = [&] { + match_surface = surface; + match_valid = is_valid; + match_scale = surface->res_scale; + match_interval = surface_interval; + }; + + if (surface->res_scale > match_scale) { + UpdateMatch(); + return; + } else if (surface->res_scale < match_scale) { + return; + } + + if (is_valid && !match_valid) { + UpdateMatch(); + return; + } else if (is_valid != match_valid) { + return; + } + + if (boost::icl::length(surface_interval) > boost::icl::length(match_interval)) { + UpdateMatch(); + } + }; + IsMatch_Helper(std::integral_constant<MatchFlags, MatchFlags::Exact>{}, [&] { + return std::make_pair(surface->ExactMatch(params), surface->GetInterval()); + }); + IsMatch_Helper(std::integral_constant<MatchFlags, MatchFlags::SubRect>{}, [&] { + return std::make_pair(surface->CanSubRect(params), surface->GetInterval()); + }); + IsMatch_Helper(std::integral_constant<MatchFlags, MatchFlags::Copy>{}, [&] { + auto copy_interval = + params.FromInterval(*validate_interval).GetCopyableInterval(surface); + bool matched = boost::icl::length(copy_interval & *validate_interval) != 0 && + surface->CanCopy(params, copy_interval); + return std::make_pair(matched, copy_interval); + }); + IsMatch_Helper(std::integral_constant<MatchFlags, MatchFlags::Expand>{}, [&] { + return std::make_pair(surface->CanExpand(params), surface->GetInterval()); + }); + IsMatch_Helper(std::integral_constant<MatchFlags, MatchFlags::TexCopy>{}, [&] { + return std::make_pair(surface->CanTexCopy(params), surface->GetInterval()); + }); + } + } + return match_surface; +} + +RasterizerCacheOpenGL::RasterizerCacheOpenGL() { + read_framebuffer.Create(); + draw_framebuffer.Create(); + + attributeless_vao.Create(); + + d24s8_abgr_buffer.Create(); + d24s8_abgr_buffer_size = 0; + + const char* vs_source = R"( +#version 330 core +const vec2 vertices[4] = vec2[4](vec2(-1.0, -1.0), vec2(1.0, -1.0), vec2(-1.0, 1.0), vec2(1.0, 1.0)); +void main() { + gl_Position = vec4(vertices[gl_VertexID], 0.0, 1.0); +} +)"; + const char* fs_source = R"( +#version 330 core + +uniform samplerBuffer tbo; +uniform vec2 tbo_size; +uniform vec4 viewport; + +out vec4 color; + +void main() { + vec2 tbo_coord = (gl_FragCoord.xy - viewport.xy) * tbo_size / viewport.zw; + int tbo_offset = int(tbo_coord.y) * int(tbo_size.x) + int(tbo_coord.x); + color = texelFetch(tbo, tbo_offset).rabg; +} +)"; + d24s8_abgr_shader.Create(vs_source, nullptr, fs_source); + + OpenGLState state = OpenGLState::GetCurState(); + GLuint old_program = state.draw.shader_program; + state.draw.shader_program = d24s8_abgr_shader.handle; + state.Apply(); + + GLint tbo_u_id = glGetUniformLocation(d24s8_abgr_shader.handle, "tbo"); + ASSERT(tbo_u_id != -1); + glUniform1i(tbo_u_id, 0); + + state.draw.shader_program = old_program; + state.Apply(); + + d24s8_abgr_tbo_size_u_id = glGetUniformLocation(d24s8_abgr_shader.handle, "tbo_size"); + ASSERT(d24s8_abgr_tbo_size_u_id != -1); + d24s8_abgr_viewport_u_id = glGetUniformLocation(d24s8_abgr_shader.handle, "viewport"); + ASSERT(d24s8_abgr_viewport_u_id != -1); +} + +RasterizerCacheOpenGL::~RasterizerCacheOpenGL() { + FlushAll(); + while (!surface_cache.empty()) + UnregisterSurface(*surface_cache.begin()->second.begin()); +} + +bool RasterizerCacheOpenGL::BlitSurfaces(const Surface& src_surface, + const MathUtil::Rectangle<u32>& src_rect, + const Surface& dst_surface, + const MathUtil::Rectangle<u32>& dst_rect) { + if (!SurfaceParams::CheckFormatsBlittable(src_surface->pixel_format, dst_surface->pixel_format)) + return false; + + return BlitTextures(src_surface->texture.handle, src_rect, dst_surface->texture.handle, + dst_rect, src_surface->type, read_framebuffer.handle, + draw_framebuffer.handle); +} + +void RasterizerCacheOpenGL::ConvertD24S8toABGR(GLuint src_tex, + const MathUtil::Rectangle<u32>& src_rect, + GLuint dst_tex, + const MathUtil::Rectangle<u32>& dst_rect) { + OpenGLState prev_state = OpenGLState::GetCurState(); + SCOPE_EXIT({ prev_state.Apply(); }); + + OpenGLState state; + state.draw.read_framebuffer = read_framebuffer.handle; + state.draw.draw_framebuffer = draw_framebuffer.handle; + state.Apply(); + + glBindBuffer(GL_PIXEL_PACK_BUFFER, d24s8_abgr_buffer.handle); + + GLsizeiptr target_pbo_size = src_rect.GetWidth() * src_rect.GetHeight() * 4; + if (target_pbo_size > d24s8_abgr_buffer_size) { + d24s8_abgr_buffer_size = target_pbo_size * 2; + glBufferData(GL_PIXEL_PACK_BUFFER, d24s8_abgr_buffer_size, nullptr, GL_STREAM_COPY); + } + + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0); + glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, src_tex, + 0); + glReadPixels(static_cast<GLint>(src_rect.left), static_cast<GLint>(src_rect.bottom), + static_cast<GLsizei>(src_rect.GetWidth()), + static_cast<GLsizei>(src_rect.GetHeight()), GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, + 0); + + glBindBuffer(GL_PIXEL_PACK_BUFFER, 0); + + // PBO now contains src_tex in RABG format + state.draw.shader_program = d24s8_abgr_shader.handle; + state.draw.vertex_array = attributeless_vao.handle; + state.viewport.x = static_cast<GLint>(dst_rect.left); + state.viewport.y = static_cast<GLint>(dst_rect.bottom); + state.viewport.width = static_cast<GLsizei>(dst_rect.GetWidth()); + state.viewport.height = static_cast<GLsizei>(dst_rect.GetHeight()); + state.Apply(); + + OGLTexture tbo; + tbo.Create(); + glActiveTexture(GL_TEXTURE0); + glBindTexture(GL_TEXTURE_BUFFER, tbo.handle); + glTexBuffer(GL_TEXTURE_BUFFER, GL_RGBA8, d24s8_abgr_buffer.handle); + + glUniform2f(d24s8_abgr_tbo_size_u_id, static_cast<GLfloat>(src_rect.GetWidth()), + static_cast<GLfloat>(src_rect.GetHeight())); + glUniform4f(d24s8_abgr_viewport_u_id, static_cast<GLfloat>(state.viewport.x), + static_cast<GLfloat>(state.viewport.y), static_cast<GLfloat>(state.viewport.width), + static_cast<GLfloat>(state.viewport.height)); + + glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, dst_tex, 0); + glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0); + glDrawArrays(GL_TRIANGLE_STRIP, 0, 4); + + glBindTexture(GL_TEXTURE_BUFFER, 0); +} + +Surface RasterizerCacheOpenGL::GetSurface(const SurfaceParams& params, ScaleMatch match_res_scale, + bool load_if_create) { + if (params.addr == 0 || params.height * params.width == 0) { + return nullptr; + } + // Use GetSurfaceSubRect instead + ASSERT(params.width == params.stride); + + ASSERT(!params.is_tiled || (params.width % 8 == 0 && params.height % 8 == 0)); + + // Check for an exact match in existing surfaces + Surface surface = + FindMatch<MatchFlags::Exact | MatchFlags::Invalid>(surface_cache, params, match_res_scale); + + if (surface == nullptr) { + u16 target_res_scale = params.res_scale; + if (match_res_scale != ScaleMatch::Exact) { + // This surface may have a subrect of another surface with a higher res_scale, find it + // to adjust our params + SurfaceParams find_params = params; + Surface expandable = FindMatch<MatchFlags::Expand | MatchFlags::Invalid>( + surface_cache, find_params, match_res_scale); + if (expandable != nullptr && expandable->res_scale > target_res_scale) { + target_res_scale = expandable->res_scale; + } + // Keep res_scale when reinterpreting d24s8 -> rgba8 + if (params.pixel_format == PixelFormat::RGBA8) { + find_params.pixel_format = PixelFormat::D24S8; + expandable = FindMatch<MatchFlags::Expand | MatchFlags::Invalid>( + surface_cache, find_params, match_res_scale); + if (expandable != nullptr && expandable->res_scale > target_res_scale) { + target_res_scale = expandable->res_scale; + } + } + } + SurfaceParams new_params = params; + new_params.res_scale = target_res_scale; + surface = CreateSurface(new_params); + RegisterSurface(surface); + } + + if (load_if_create) { + ValidateSurface(surface, params.addr, params.size); + } + + return surface; +} + +SurfaceRect_Tuple RasterizerCacheOpenGL::GetSurfaceSubRect(const SurfaceParams& params, + ScaleMatch match_res_scale, + bool load_if_create) { + if (params.addr == 0 || params.height * params.width == 0) { + return std::make_tuple(nullptr, MathUtil::Rectangle<u32>{}); + } + + // Attempt to find encompassing surface + Surface surface = FindMatch<MatchFlags::SubRect | MatchFlags::Invalid>(surface_cache, params, + match_res_scale); + + // Check if FindMatch failed because of res scaling + // If that's the case create a new surface with + // the dimensions of the lower res_scale surface + // to suggest it should not be used again + if (surface == nullptr && match_res_scale != ScaleMatch::Ignore) { + surface = FindMatch<MatchFlags::SubRect | MatchFlags::Invalid>(surface_cache, params, + ScaleMatch::Ignore); + if (surface != nullptr) { + ASSERT(surface->res_scale < params.res_scale); + SurfaceParams new_params = *surface; + new_params.res_scale = params.res_scale; + + surface = CreateSurface(new_params); + RegisterSurface(surface); + } + } + + SurfaceParams aligned_params = params; + if (params.is_tiled) { + aligned_params.height = Common::AlignUp(params.height, 8); + aligned_params.width = Common::AlignUp(params.width, 8); + aligned_params.stride = Common::AlignUp(params.stride, 8); + aligned_params.UpdateParams(); + } + + // Check for a surface we can expand before creating a new one + if (surface == nullptr) { + surface = FindMatch<MatchFlags::Expand | MatchFlags::Invalid>(surface_cache, aligned_params, + match_res_scale); + if (surface != nullptr) { + aligned_params.width = aligned_params.stride; + aligned_params.UpdateParams(); + + SurfaceParams new_params = *surface; + new_params.addr = std::min(aligned_params.addr, surface->addr); + new_params.end = std::max(aligned_params.end, surface->end); + new_params.size = new_params.end - new_params.addr; + new_params.height = static_cast<u32>( + new_params.size / aligned_params.BytesInPixels(aligned_params.stride)); + ASSERT(new_params.size % aligned_params.BytesInPixels(aligned_params.stride) == 0); + + Surface new_surface = CreateSurface(new_params); + DuplicateSurface(surface, new_surface); + + // Delete the expanded surface, this can't be done safely yet + // because it may still be in use + remove_surfaces.emplace(surface); + + surface = new_surface; + RegisterSurface(new_surface); + } + } + + // No subrect found - create and return a new surface + if (surface == nullptr) { + SurfaceParams new_params = aligned_params; + // Can't have gaps in a surface + new_params.width = aligned_params.stride; + new_params.UpdateParams(); + // GetSurface will create the new surface and possibly adjust res_scale if necessary + surface = GetSurface(new_params, match_res_scale, load_if_create); + } else if (load_if_create) { + ValidateSurface(surface, aligned_params.addr, aligned_params.size); + } + + return std::make_tuple(surface, surface->GetScaledSubRect(params)); +} + +Surface RasterizerCacheOpenGL::GetTextureSurface(const void* config) { + UNREACHABLE(); + return {}; +} + +SurfaceSurfaceRect_Tuple RasterizerCacheOpenGL::GetFramebufferSurfaces( + bool using_color_fb, bool using_depth_fb, const MathUtil::Rectangle<s32>& viewport) { + const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs; + const auto& memory_manager = Core::System().GetInstance().GPU().memory_manager; + const auto& config = regs.rt[0]; + + // TODO(bunnei): This is hard corded to use just the first render buffer + LOG_WARNING(Render_OpenGL, "hard-coded for render target 0!"); + + // update resolution_scale_factor and reset cache if changed + // TODO (bunnei): This code was ported as-is from Citra, and is technically not thread-safe. We + // need to fix this before making the renderer multi-threaded. + static u16 resolution_scale_factor = GetResolutionScaleFactor(); + if (resolution_scale_factor != GetResolutionScaleFactor()) { + resolution_scale_factor = GetResolutionScaleFactor(); + FlushAll(); + while (!surface_cache.empty()) + UnregisterSurface(*surface_cache.begin()->second.begin()); + } + + MathUtil::Rectangle<u32> viewport_clamped{ + static_cast<u32>(MathUtil::Clamp(viewport.left, 0, static_cast<s32>(config.width))), + static_cast<u32>(MathUtil::Clamp(viewport.top, 0, static_cast<s32>(config.height))), + static_cast<u32>(MathUtil::Clamp(viewport.right, 0, static_cast<s32>(config.width))), + static_cast<u32>(MathUtil::Clamp(viewport.bottom, 0, static_cast<s32>(config.height)))}; + + // get color and depth surfaces + SurfaceParams color_params; + color_params.is_tiled = true; + color_params.res_scale = resolution_scale_factor; + color_params.width = config.width; + color_params.height = config.height; + SurfaceParams depth_params = color_params; + + color_params.addr = memory_manager->PhysicalToVirtualAddress(config.Address()); + color_params.pixel_format = SurfaceParams::PixelFormatFromRenderTargetFormat(config.format); + color_params.UpdateParams(); + + ASSERT_MSG(!using_depth_fb, "depth buffer is unimplemented"); + // depth_params.addr = config.GetDepthBufferPhysicalAddress(); + // depth_params.pixel_format = SurfaceParams::PixelFormatFromDepthFormat(config.depth_format); + // depth_params.UpdateParams(); + + auto color_vp_interval = color_params.GetSubRectInterval(viewport_clamped); + auto depth_vp_interval = depth_params.GetSubRectInterval(viewport_clamped); + + // Make sure that framebuffers don't overlap if both color and depth are being used + if (using_color_fb && using_depth_fb && + boost::icl::length(color_vp_interval & depth_vp_interval)) { + LOG_CRITICAL(Render_OpenGL, "Color and depth framebuffer memory regions overlap; " + "overlapping framebuffers not supported!"); + using_depth_fb = false; + } + + MathUtil::Rectangle<u32> color_rect{}; + Surface color_surface = nullptr; + if (using_color_fb) + std::tie(color_surface, color_rect) = + GetSurfaceSubRect(color_params, ScaleMatch::Exact, false); + + MathUtil::Rectangle<u32> depth_rect{}; + Surface depth_surface = nullptr; + if (using_depth_fb) + std::tie(depth_surface, depth_rect) = + GetSurfaceSubRect(depth_params, ScaleMatch::Exact, false); + + MathUtil::Rectangle<u32> fb_rect{}; + if (color_surface != nullptr && depth_surface != nullptr) { + fb_rect = color_rect; + // Color and Depth surfaces must have the same dimensions and offsets + if (color_rect.bottom != depth_rect.bottom || color_rect.top != depth_rect.top || + color_rect.left != depth_rect.left || color_rect.right != depth_rect.right) { + color_surface = GetSurface(color_params, ScaleMatch::Exact, false); + depth_surface = GetSurface(depth_params, ScaleMatch::Exact, false); + fb_rect = color_surface->GetScaledRect(); + } + } else if (color_surface != nullptr) { + fb_rect = color_rect; + } else if (depth_surface != nullptr) { + fb_rect = depth_rect; + } + + if (color_surface != nullptr) { + ValidateSurface(color_surface, boost::icl::first(color_vp_interval), + boost::icl::length(color_vp_interval)); + } + if (depth_surface != nullptr) { + ValidateSurface(depth_surface, boost::icl::first(depth_vp_interval), + boost::icl::length(depth_vp_interval)); + } + + return std::make_tuple(color_surface, depth_surface, fb_rect); +} + +Surface RasterizerCacheOpenGL::GetFillSurface(const void* config) { + UNREACHABLE(); + return {}; +} + +SurfaceRect_Tuple RasterizerCacheOpenGL::GetTexCopySurface(const SurfaceParams& params) { + MathUtil::Rectangle<u32> rect{}; + + Surface match_surface = FindMatch<MatchFlags::TexCopy | MatchFlags::Invalid>( + surface_cache, params, ScaleMatch::Ignore); + + if (match_surface != nullptr) { + ValidateSurface(match_surface, params.addr, params.size); + + SurfaceParams match_subrect; + if (params.width != params.stride) { + const u32 tiled_size = match_surface->is_tiled ? 8 : 1; + match_subrect = params; + match_subrect.width = + static_cast<u32>(match_surface->PixelsInBytes(params.width) / tiled_size); + match_subrect.stride = + static_cast<u32>(match_surface->PixelsInBytes(params.stride) / tiled_size); + match_subrect.height *= tiled_size; + } else { + match_subrect = match_surface->FromInterval(params.GetInterval()); + ASSERT(match_subrect.GetInterval() == params.GetInterval()); + } + + rect = match_surface->GetScaledSubRect(match_subrect); + } + + return std::make_tuple(match_surface, rect); +} + +void RasterizerCacheOpenGL::DuplicateSurface(const Surface& src_surface, + const Surface& dest_surface) { + ASSERT(dest_surface->addr <= src_surface->addr && dest_surface->end >= src_surface->end); + + BlitSurfaces(src_surface, src_surface->GetScaledRect(), dest_surface, + dest_surface->GetScaledSubRect(*src_surface)); + + dest_surface->invalid_regions -= src_surface->GetInterval(); + dest_surface->invalid_regions += src_surface->invalid_regions; + + SurfaceRegions regions; + for (auto& pair : RangeFromInterval(dirty_regions, src_surface->GetInterval())) { + if (pair.second == src_surface) { + regions += pair.first; + } + } + for (auto& interval : regions) { + dirty_regions.set({interval, dest_surface}); + } +} + +void RasterizerCacheOpenGL::ValidateSurface(const Surface& surface, VAddr addr, u64 size) { + if (size == 0) + return; + + const SurfaceInterval validate_interval(addr, addr + size); + + if (surface->type == SurfaceType::Fill) { + // Sanity check, fill surfaces will always be valid when used + ASSERT(surface->IsRegionValid(validate_interval)); + return; + } + + while (true) { + const auto it = surface->invalid_regions.find(validate_interval); + if (it == surface->invalid_regions.end()) + break; + + const auto interval = *it & validate_interval; + // Look for a valid surface to copy from + SurfaceParams params = surface->FromInterval(interval); + + Surface copy_surface = + FindMatch<MatchFlags::Copy>(surface_cache, params, ScaleMatch::Ignore, interval); + if (copy_surface != nullptr) { + SurfaceInterval copy_interval = params.GetCopyableInterval(copy_surface); + CopySurface(copy_surface, surface, copy_interval); + surface->invalid_regions.erase(copy_interval); + continue; + } + + // D24S8 to RGBA8 + if (surface->pixel_format == PixelFormat::RGBA8) { + params.pixel_format = PixelFormat::D24S8; + Surface reinterpret_surface = + FindMatch<MatchFlags::Copy>(surface_cache, params, ScaleMatch::Ignore, interval); + if (reinterpret_surface != nullptr) { + ASSERT(reinterpret_surface->pixel_format == PixelFormat::D24S8); + + SurfaceInterval convert_interval = params.GetCopyableInterval(reinterpret_surface); + SurfaceParams convert_params = surface->FromInterval(convert_interval); + auto src_rect = reinterpret_surface->GetScaledSubRect(convert_params); + auto dest_rect = surface->GetScaledSubRect(convert_params); + + ConvertD24S8toABGR(reinterpret_surface->texture.handle, src_rect, + surface->texture.handle, dest_rect); + + surface->invalid_regions.erase(convert_interval); + continue; + } + } + + // Load data from 3DS memory + FlushRegion(params.addr, params.size); + surface->LoadGLBuffer(params.addr, params.end); + surface->UploadGLTexture(surface->GetSubRect(params), read_framebuffer.handle, + draw_framebuffer.handle); + surface->invalid_regions.erase(params.GetInterval()); + } +} + +void RasterizerCacheOpenGL::FlushRegion(VAddr addr, u64 size, Surface flush_surface) { + if (size == 0) + return; + + const SurfaceInterval flush_interval(addr, addr + size); + SurfaceRegions flushed_intervals; + + for (auto& pair : RangeFromInterval(dirty_regions, flush_interval)) { + // small sizes imply that this most likely comes from the cpu, flush the entire region + // the point is to avoid thousands of small writes every frame if the cpu decides to access + // that region, anything higher than 8 you're guaranteed it comes from a service + const auto interval = size <= 8 ? pair.first : pair.first & flush_interval; + auto& surface = pair.second; + + if (flush_surface != nullptr && surface != flush_surface) + continue; + + // Sanity check, this surface is the last one that marked this region dirty + ASSERT(surface->IsRegionValid(interval)); + + if (surface->type != SurfaceType::Fill) { + SurfaceParams params = surface->FromInterval(interval); + surface->DownloadGLTexture(surface->GetSubRect(params), read_framebuffer.handle, + draw_framebuffer.handle); + } + surface->FlushGLBuffer(boost::icl::first(interval), boost::icl::last_next(interval)); + flushed_intervals += interval; + } + // Reset dirty regions + dirty_regions -= flushed_intervals; +} + +void RasterizerCacheOpenGL::FlushAll() { + FlushRegion(0, Kernel::VMManager::MAX_ADDRESS); +} + +void RasterizerCacheOpenGL::InvalidateRegion(VAddr addr, u64 size, const Surface& region_owner) { + if (size == 0) + return; + + const SurfaceInterval invalid_interval(addr, addr + size); + + if (region_owner != nullptr) { + ASSERT(region_owner->type != SurfaceType::Texture); + ASSERT(addr >= region_owner->addr && addr + size <= region_owner->end); + // Surfaces can't have a gap + ASSERT(region_owner->width == region_owner->stride); + region_owner->invalid_regions.erase(invalid_interval); + } + + for (auto& pair : RangeFromInterval(surface_cache, invalid_interval)) { + for (auto& cached_surface : pair.second) { + if (cached_surface == region_owner) + continue; + + // If cpu is invalidating this region we want to remove it + // to (likely) mark the memory pages as uncached + if (region_owner == nullptr && size <= 8) { + FlushRegion(cached_surface->addr, cached_surface->size, cached_surface); + remove_surfaces.emplace(cached_surface); + continue; + } + + const auto interval = cached_surface->GetInterval() & invalid_interval; + cached_surface->invalid_regions.insert(interval); + + // Remove only "empty" fill surfaces to avoid destroying and recreating OGL textures + if (cached_surface->type == SurfaceType::Fill && + cached_surface->IsSurfaceFullyInvalid()) { + remove_surfaces.emplace(cached_surface); + } + } + } + + if (region_owner != nullptr) + dirty_regions.set({invalid_interval, region_owner}); + else + dirty_regions.erase(invalid_interval); + + for (auto& remove_surface : remove_surfaces) { + if (remove_surface == region_owner) { + Surface expanded_surface = FindMatch<MatchFlags::SubRect | MatchFlags::Invalid>( + surface_cache, *region_owner, ScaleMatch::Ignore); + ASSERT(expanded_surface); + + if ((region_owner->invalid_regions - expanded_surface->invalid_regions).empty()) { + DuplicateSurface(region_owner, expanded_surface); + } else { + continue; + } + } + UnregisterSurface(remove_surface); + } + + remove_surfaces.clear(); +} + +Surface RasterizerCacheOpenGL::CreateSurface(const SurfaceParams& params) { + Surface surface = std::make_shared<CachedSurface>(); + static_cast<SurfaceParams&>(*surface) = params; + + surface->texture.Create(); + + surface->gl_buffer_size = 0; + surface->invalid_regions.insert(surface->GetInterval()); + AllocateSurfaceTexture(surface->texture.handle, GetFormatTuple(surface->pixel_format), + surface->GetScaledWidth(), surface->GetScaledHeight()); + + return surface; +} + +void RasterizerCacheOpenGL::RegisterSurface(const Surface& surface) { + if (surface->registered) { + return; + } + surface->registered = true; + surface_cache.add({surface->GetInterval(), SurfaceSet{surface}}); + UpdatePagesCachedCount(surface->addr, surface->size, 1); +} + +void RasterizerCacheOpenGL::UnregisterSurface(const Surface& surface) { + if (!surface->registered) { + return; + } + surface->registered = false; + UpdatePagesCachedCount(surface->addr, surface->size, -1); + surface_cache.subtract({surface->GetInterval(), SurfaceSet{surface}}); +} + +void RasterizerCacheOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) { + const u64 num_pages = + ((addr + size - 1) >> Memory::PAGE_BITS) - (addr >> Memory::PAGE_BITS) + 1; + const u64 page_start = addr >> Memory::PAGE_BITS; + const u64 page_end = page_start + num_pages; + + // Interval maps will erase segments if count reaches 0, so if delta is negative we have to + // subtract after iterating + const auto pages_interval = PageMap::interval_type::right_open(page_start, page_end); + if (delta > 0) + cached_pages.add({pages_interval, delta}); + + for (const auto& pair : RangeFromInterval(cached_pages, pages_interval)) { + const auto interval = pair.first & pages_interval; + const int count = pair.second; + + const VAddr interval_start_addr = boost::icl::first(interval) << Memory::PAGE_BITS; + const VAddr interval_end_addr = boost::icl::last_next(interval) << Memory::PAGE_BITS; + const u64 interval_size = interval_end_addr - interval_start_addr; + + if (delta > 0 && count == delta) + Memory::RasterizerMarkRegionCached(interval_start_addr, interval_size, true); + else if (delta < 0 && count == -delta) + Memory::RasterizerMarkRegionCached(interval_start_addr, interval_size, false); + else + ASSERT(count >= 0); + } + + if (delta < 0) + cached_pages.add({pages_interval, delta}); +} diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.h b/src/video_core/renderer_opengl/gl_rasterizer_cache.h new file mode 100644 index 000000000..1f660d30c --- /dev/null +++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.h @@ -0,0 +1,369 @@ +// Copyright 2015 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <array> +#include <memory> +#include <set> +#include <tuple> +#ifdef __GNUC__ +#pragma GCC diagnostic push +#pragma GCC diagnostic ignored "-Wunused-local-typedefs" +#endif +#include <boost/icl/interval_map.hpp> +#include <boost/icl/interval_set.hpp> +#ifdef __GNUC__ +#pragma GCC diagnostic pop +#endif +#include <glad/glad.h> +#include "common/assert.h" +#include "common/common_funcs.h" +#include "common/common_types.h" +#include "common/math_util.h" +#include "video_core/gpu.h" +#include "video_core/renderer_opengl/gl_resource_manager.h" + +struct CachedSurface; +using Surface = std::shared_ptr<CachedSurface>; +using SurfaceSet = std::set<Surface>; + +using SurfaceRegions = boost::icl::interval_set<VAddr>; +using SurfaceMap = boost::icl::interval_map<VAddr, Surface>; +using SurfaceCache = boost::icl::interval_map<VAddr, SurfaceSet>; + +using SurfaceInterval = SurfaceCache::interval_type; +static_assert(std::is_same<SurfaceRegions::interval_type, SurfaceCache::interval_type>() && + std::is_same<SurfaceMap::interval_type, SurfaceCache::interval_type>(), + "incorrect interval types"); + +using SurfaceRect_Tuple = std::tuple<Surface, MathUtil::Rectangle<u32>>; +using SurfaceSurfaceRect_Tuple = std::tuple<Surface, Surface, MathUtil::Rectangle<u32>>; + +using PageMap = boost::icl::interval_map<u64, int>; + +enum class ScaleMatch { + Exact, // only accept same res scale + Upscale, // only allow higher scale than params + Ignore // accept every scaled res +}; + +struct SurfaceParams { + enum class PixelFormat { + // First 5 formats are shared between textures and color buffers + RGBA8 = 0, + RGB8 = 1, + RGB5A1 = 2, + RGB565 = 3, + RGBA4 = 4, + + // Texture-only formats + IA8 = 5, + RG8 = 6, + I8 = 7, + A8 = 8, + IA4 = 9, + I4 = 10, + A4 = 11, + ETC1 = 12, + ETC1A4 = 13, + + // Depth buffer-only formats + D16 = 14, + // gap + D24 = 16, + D24S8 = 17, + + Invalid = 255, + }; + + enum class SurfaceType { + Color = 0, + Texture = 1, + Depth = 2, + DepthStencil = 3, + Fill = 4, + Invalid = 5 + }; + + static constexpr unsigned int GetFormatBpp(PixelFormat format) { + constexpr std::array<unsigned int, 18> bpp_table = { + 32, // RGBA8 + 24, // RGB8 + 16, // RGB5A1 + 16, // RGB565 + 16, // RGBA4 + 16, // IA8 + 16, // RG8 + 8, // I8 + 8, // A8 + 8, // IA4 + 4, // I4 + 4, // A4 + 4, // ETC1 + 8, // ETC1A4 + 16, // D16 + 0, + 24, // D24 + 32, // D24S8 + }; + + assert(static_cast<size_t>(format) < bpp_table.size()); + return bpp_table[static_cast<size_t>(format)]; + } + unsigned int GetFormatBpp() const { + return GetFormatBpp(pixel_format); + } + + static PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format) { + switch (format) { + case Tegra::RenderTargetFormat::RGBA8_UNORM: + return PixelFormat::RGBA8; + default: + UNREACHABLE(); + } + } + + static PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat format) { + switch (format) { + case Tegra::FramebufferConfig::PixelFormat::ABGR8: + return PixelFormat::RGBA8; + default: + UNREACHABLE(); + } + } + + static bool CheckFormatsBlittable(PixelFormat pixel_format_a, PixelFormat pixel_format_b) { + SurfaceType a_type = GetFormatType(pixel_format_a); + SurfaceType b_type = GetFormatType(pixel_format_b); + + if ((a_type == SurfaceType::Color || a_type == SurfaceType::Texture) && + (b_type == SurfaceType::Color || b_type == SurfaceType::Texture)) { + return true; + } + + if (a_type == SurfaceType::Depth && b_type == SurfaceType::Depth) { + return true; + } + + if (a_type == SurfaceType::DepthStencil && b_type == SurfaceType::DepthStencil) { + return true; + } + + return false; + } + + static constexpr SurfaceType GetFormatType(PixelFormat pixel_format) { + if ((unsigned int)pixel_format < 5) { + return SurfaceType::Color; + } + + if ((unsigned int)pixel_format < 14) { + return SurfaceType::Texture; + } + + if (pixel_format == PixelFormat::D16 || pixel_format == PixelFormat::D24) { + return SurfaceType::Depth; + } + + if (pixel_format == PixelFormat::D24S8) { + return SurfaceType::DepthStencil; + } + + return SurfaceType::Invalid; + } + + /// Update the params "size", "end" and "type" from the already set "addr", "width", "height" + /// and "pixel_format" + void UpdateParams() { + if (stride == 0) { + stride = width; + } + type = GetFormatType(pixel_format); + size = !is_tiled ? BytesInPixels(stride * (height - 1) + width) + : BytesInPixels(stride * 8 * (height / 8 - 1) + width * 8); + end = addr + size; + } + + SurfaceInterval GetInterval() const { + return SurfaceInterval::right_open(addr, end); + } + + // Returns the outer rectangle containing "interval" + SurfaceParams FromInterval(SurfaceInterval interval) const; + + SurfaceInterval GetSubRectInterval(MathUtil::Rectangle<u32> unscaled_rect) const; + + // Returns the region of the biggest valid rectange within interval + SurfaceInterval GetCopyableInterval(const Surface& src_surface) const; + + u32 GetScaledWidth() const { + return width * res_scale; + } + + u32 GetScaledHeight() const { + return height * res_scale; + } + + MathUtil::Rectangle<u32> GetRect() const { + return {0, height, width, 0}; + } + + MathUtil::Rectangle<u32> GetScaledRect() const { + return {0, GetScaledHeight(), GetScaledWidth(), 0}; + } + + u64 PixelsInBytes(u64 size) const { + return size * CHAR_BIT / GetFormatBpp(pixel_format); + } + + u64 BytesInPixels(u64 pixels) const { + return pixels * GetFormatBpp(pixel_format) / CHAR_BIT; + } + + bool ExactMatch(const SurfaceParams& other_surface) const; + bool CanSubRect(const SurfaceParams& sub_surface) const; + bool CanExpand(const SurfaceParams& expanded_surface) const; + bool CanTexCopy(const SurfaceParams& texcopy_params) const; + + MathUtil::Rectangle<u32> GetSubRect(const SurfaceParams& sub_surface) const; + MathUtil::Rectangle<u32> GetScaledSubRect(const SurfaceParams& sub_surface) const; + + VAddr addr = 0; + VAddr end = 0; + u64 size = 0; + + u32 width = 0; + u32 height = 0; + u32 stride = 0; + u16 res_scale = 1; + + bool is_tiled = false; + PixelFormat pixel_format = PixelFormat::Invalid; + SurfaceType type = SurfaceType::Invalid; +}; + +struct CachedSurface : SurfaceParams { + bool CanFill(const SurfaceParams& dest_surface, SurfaceInterval fill_interval) const; + bool CanCopy(const SurfaceParams& dest_surface, SurfaceInterval copy_interval) const; + + bool IsRegionValid(SurfaceInterval interval) const { + return (invalid_regions.find(interval) == invalid_regions.end()); + } + + bool IsSurfaceFullyInvalid() const { + return (invalid_regions & GetInterval()) == SurfaceRegions(GetInterval()); + } + + bool registered = false; + SurfaceRegions invalid_regions; + + u64 fill_size = 0; /// Number of bytes to read from fill_data + std::array<u8, 4> fill_data; + + OGLTexture texture; + + static constexpr unsigned int GetGLBytesPerPixel(PixelFormat format) { + // OpenGL needs 4 bpp alignment for D24 since using GL_UNSIGNED_INT as type + return format == PixelFormat::Invalid + ? 0 + : (format == PixelFormat::D24 || GetFormatType(format) == SurfaceType::Texture) + ? 4 + : SurfaceParams::GetFormatBpp(format) / 8; + } + + std::unique_ptr<u8[]> gl_buffer; + size_t gl_buffer_size = 0; + + // Read/Write data in Switch memory to/from gl_buffer + void LoadGLBuffer(VAddr load_start, VAddr load_end); + void FlushGLBuffer(VAddr flush_start, VAddr flush_end); + + // Upload/Download data in gl_buffer in/to this surface's texture + void UploadGLTexture(const MathUtil::Rectangle<u32>& rect, GLuint read_fb_handle, + GLuint draw_fb_handle); + void DownloadGLTexture(const MathUtil::Rectangle<u32>& rect, GLuint read_fb_handle, + GLuint draw_fb_handle); +}; + +class RasterizerCacheOpenGL : NonCopyable { +public: + RasterizerCacheOpenGL(); + ~RasterizerCacheOpenGL(); + + /// Blit one surface's texture to another + bool BlitSurfaces(const Surface& src_surface, const MathUtil::Rectangle<u32>& src_rect, + const Surface& dst_surface, const MathUtil::Rectangle<u32>& dst_rect); + + void ConvertD24S8toABGR(GLuint src_tex, const MathUtil::Rectangle<u32>& src_rect, + GLuint dst_tex, const MathUtil::Rectangle<u32>& dst_rect); + + /// Copy one surface's region to another + void CopySurface(const Surface& src_surface, const Surface& dst_surface, + SurfaceInterval copy_interval); + + /// Load a texture from 3DS memory to OpenGL and cache it (if not already cached) + Surface GetSurface(const SurfaceParams& params, ScaleMatch match_res_scale, + bool load_if_create); + + /// Attempt to find a subrect (resolution scaled) of a surface, otherwise loads a texture from + /// 3DS memory to OpenGL and caches it (if not already cached) + SurfaceRect_Tuple GetSurfaceSubRect(const SurfaceParams& params, ScaleMatch match_res_scale, + bool load_if_create); + + /// Get a surface based on the texture configuration + Surface GetTextureSurface(const void* config); + + /// Get the color and depth surfaces based on the framebuffer configuration + SurfaceSurfaceRect_Tuple GetFramebufferSurfaces(bool using_color_fb, bool using_depth_fb, + const MathUtil::Rectangle<s32>& viewport); + + /// Get a surface that matches the fill config + Surface GetFillSurface(const void* config); + + /// Get a surface that matches a "texture copy" display transfer config + SurfaceRect_Tuple GetTexCopySurface(const SurfaceParams& params); + + /// Write any cached resources overlapping the region back to memory (if dirty) + void FlushRegion(VAddr addr, u64 size, Surface flush_surface = nullptr); + + /// Mark region as being invalidated by region_owner (nullptr if 3DS memory) + void InvalidateRegion(VAddr addr, u64 size, const Surface& region_owner); + + /// Flush all cached resources tracked by this cache manager + void FlushAll(); + +private: + void DuplicateSurface(const Surface& src_surface, const Surface& dest_surface); + + /// Update surface's texture for given region when necessary + void ValidateSurface(const Surface& surface, VAddr addr, u64 size); + + /// Create a new surface + Surface CreateSurface(const SurfaceParams& params); + + /// Register surface into the cache + void RegisterSurface(const Surface& surface); + + /// Remove surface from the cache + void UnregisterSurface(const Surface& surface); + + /// Increase/decrease the number of surface in pages touching the specified region + void UpdatePagesCachedCount(VAddr addr, u64 size, int delta); + + SurfaceCache surface_cache; + PageMap cached_pages; + SurfaceMap dirty_regions; + SurfaceSet remove_surfaces; + + OGLFramebuffer read_framebuffer; + OGLFramebuffer draw_framebuffer; + + OGLVertexArray attributeless_vao; + OGLBuffer d24s8_abgr_buffer; + GLsizeiptr d24s8_abgr_buffer_size; + OGLShader d24s8_abgr_shader; + GLint d24s8_abgr_tbo_size_u_id; + GLint d24s8_abgr_viewport_u_id; +}; diff --git a/src/video_core/renderer_opengl/gl_resource_manager.h b/src/video_core/renderer_opengl/gl_resource_manager.h index 13301ec9f..7da5e74d1 100644 --- a/src/video_core/renderer_opengl/gl_resource_manager.h +++ b/src/video_core/renderer_opengl/gl_resource_manager.h @@ -36,7 +36,7 @@ public: if (handle == 0) return; glDeleteTextures(1, &handle); - OpenGLState::ResetTexture(handle); + OpenGLState::GetCurState().ResetTexture(handle).Apply(); handle = 0; } @@ -69,7 +69,7 @@ public: if (handle == 0) return; glDeleteSamplers(1, &handle); - OpenGLState::ResetSampler(handle); + OpenGLState::GetCurState().ResetSampler(handle).Apply(); handle = 0; } @@ -91,10 +91,13 @@ public: } /// Creates a new internal OpenGL resource and stores the handle - void Create(const char* vert_shader, const char* frag_shader) { + void Create(const char* vert_shader, const char* geo_shader, const char* frag_shader, + const std::vector<const char*>& feedback_vars = {}, + bool separable_program = false) { if (handle != 0) return; - handle = GLShader::LoadProgram(vert_shader, frag_shader); + handle = GLShader::LoadProgram(vert_shader, geo_shader, frag_shader, feedback_vars, + separable_program); } /// Deletes the internal OpenGL resource @@ -102,7 +105,40 @@ public: if (handle == 0) return; glDeleteProgram(handle); - OpenGLState::ResetProgram(handle); + OpenGLState::GetCurState().ResetProgram(handle).Apply(); + handle = 0; + } + + GLuint handle = 0; +}; + +class OGLPipeline : private NonCopyable { +public: + OGLPipeline() = default; + OGLPipeline(OGLPipeline&& o) { + handle = std::exchange<GLuint>(o.handle, 0); + } + ~OGLPipeline() { + Release(); + } + OGLPipeline& operator=(OGLPipeline&& o) { + handle = std::exchange<GLuint>(o.handle, 0); + return *this; + } + + /// Creates a new internal OpenGL resource and stores the handle + void Create() { + if (handle != 0) + return; + glGenProgramPipelines(1, &handle); + } + + /// Deletes the internal OpenGL resource + void Release() { + if (handle == 0) + return; + glDeleteProgramPipelines(1, &handle); + OpenGLState::GetCurState().ResetPipeline(handle).Apply(); handle = 0; } @@ -135,13 +171,46 @@ public: if (handle == 0) return; glDeleteBuffers(1, &handle); - OpenGLState::ResetBuffer(handle); + OpenGLState::GetCurState().ResetBuffer(handle).Apply(); handle = 0; } GLuint handle = 0; }; +class OGLSync : private NonCopyable { +public: + OGLSync() = default; + + OGLSync(OGLSync&& o) : handle(std::exchange(o.handle, nullptr)) {} + + ~OGLSync() { + Release(); + } + OGLSync& operator=(OGLSync&& o) { + Release(); + handle = std::exchange(o.handle, nullptr); + return *this; + } + + /// Creates a new internal OpenGL resource and stores the handle + void Create() { + if (handle != 0) + return; + handle = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0); + } + + /// Deletes the internal OpenGL resource + void Release() { + if (handle == 0) + return; + glDeleteSync(handle); + handle = 0; + } + + GLsync handle = 0; +}; + class OGLVertexArray : private NonCopyable { public: OGLVertexArray() = default; @@ -168,7 +237,7 @@ public: if (handle == 0) return; glDeleteVertexArrays(1, &handle); - OpenGLState::ResetVertexArray(handle); + OpenGLState::GetCurState().ResetVertexArray(handle).Apply(); handle = 0; } @@ -201,7 +270,7 @@ public: if (handle == 0) return; glDeleteFramebuffers(1, &handle); - OpenGLState::ResetFramebuffer(handle); + OpenGLState::GetCurState().ResetFramebuffer(handle).Apply(); handle = 0; } diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp new file mode 100644 index 000000000..564ea8f9e --- /dev/null +++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp @@ -0,0 +1,58 @@ +// Copyright 2018 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <string> +#include <queue> +#include "common/assert.h" +#include "common/common_types.h" +#include "video_core/renderer_opengl/gl_shader_decompiler.h" + +namespace Maxwell3D { +namespace Shader { +namespace Decompiler { + +constexpr u32 PROGRAM_END = MAX_PROGRAM_CODE_LENGTH; + +class Impl { +public: + Impl(const std::array<u32, MAX_PROGRAM_CODE_LENGTH>& program_code, + const std::array<u32, MAX_SWIZZLE_DATA_LENGTH>& swizzle_data, u32 main_offset, + const std::function<std::string(u32)>& inputreg_getter, + const std::function<std::string(u32)>& outputreg_getter, bool sanitize_mul, + const std::string& emit_cb, const std::string& setemit_cb) + : program_code(program_code), swizzle_data(swizzle_data), main_offset(main_offset), + inputreg_getter(inputreg_getter), outputreg_getter(outputreg_getter), + sanitize_mul(sanitize_mul), emit_cb(emit_cb), setemit_cb(setemit_cb) {} + + std::string Decompile() { + UNREACHABLE(); + return {}; + } + +private: + const std::array<u32, MAX_PROGRAM_CODE_LENGTH>& program_code; + const std::array<u32, MAX_SWIZZLE_DATA_LENGTH>& swizzle_data; + u32 main_offset; + const std::function<std::string(u32)>& inputreg_getter; + const std::function<std::string(u32)>& outputreg_getter; + bool sanitize_mul; + const std::string& emit_cb; + const std::string& setemit_cb; +}; + +std::string DecompileProgram(const std::array<u32, MAX_PROGRAM_CODE_LENGTH>& program_code, + const std::array<u32, MAX_SWIZZLE_DATA_LENGTH>& swizzle_data, + u32 main_offset, + const std::function<std::string(u32)>& inputreg_getter, + const std::function<std::string(u32)>& outputreg_getter, + bool sanitize_mul, const std::string& emit_cb, + const std::string& setemit_cb) { + Impl impl(program_code, swizzle_data, main_offset, inputreg_getter, outputreg_getter, + sanitize_mul, emit_cb, setemit_cb); + return impl.Decompile(); +} + +} // namespace Decompiler +} // namespace Shader +} // namespace Maxwell3D diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.h b/src/video_core/renderer_opengl/gl_shader_decompiler.h new file mode 100644 index 000000000..02ebfcbe8 --- /dev/null +++ b/src/video_core/renderer_opengl/gl_shader_decompiler.h @@ -0,0 +1,27 @@ +// Copyright 2018 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <array> +#include <functional> +#include <string> +#include "common/common_types.h" + +namespace Maxwell3D { +namespace Shader { +namespace Decompiler { + +constexpr size_t MAX_PROGRAM_CODE_LENGTH{0x100000}; +constexpr size_t MAX_SWIZZLE_DATA_LENGTH{0x100000}; + +std::string DecompileProgram(const std::array<u32, MAX_PROGRAM_CODE_LENGTH>& program_code, + const std::array<u32, MAX_SWIZZLE_DATA_LENGTH>& swizzle_data, + u32 main_offset, + const std::function<std::string(u32)>& inputreg_getter, + const std::function<std::string(u32)>& outputreg_getter, + bool sanitize_mul, const std::string& emit_cb = "", + const std::string& setemit_cb = ""); + +} // namespace Decompiler +} // namespace Shader +} // namespace Maxwell3D diff --git a/src/video_core/renderer_opengl/gl_shader_gen.cpp b/src/video_core/renderer_opengl/gl_shader_gen.cpp new file mode 100644 index 000000000..8f3c98800 --- /dev/null +++ b/src/video_core/renderer_opengl/gl_shader_gen.cpp @@ -0,0 +1,20 @@ +// Copyright 2018 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "common/assert.h" +#include "video_core/renderer_opengl/gl_shader_gen.h" + +namespace GLShader { + +std::string GenerateVertexShader(const MaxwellVSConfig& config) { + UNREACHABLE(); + return {}; +} + +std::string GenerateFragmentShader(const MaxwellFSConfig& config) { + UNREACHABLE(); + return {}; +} + +} // namespace GLShader diff --git a/src/video_core/renderer_opengl/gl_shader_gen.h b/src/video_core/renderer_opengl/gl_shader_gen.h new file mode 100644 index 000000000..5101e7d30 --- /dev/null +++ b/src/video_core/renderer_opengl/gl_shader_gen.h @@ -0,0 +1,66 @@ +// Copyright 2018 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <cstring> +#include <string> +#include <type_traits> +#include "common/hash.h" + +namespace GLShader { + +enum Attributes { + ATTRIBUTE_POSITION, + ATTRIBUTE_COLOR, + ATTRIBUTE_TEXCOORD0, + ATTRIBUTE_TEXCOORD1, + ATTRIBUTE_TEXCOORD2, + ATTRIBUTE_TEXCOORD0_W, + ATTRIBUTE_NORMQUAT, + ATTRIBUTE_VIEW, +}; + +struct MaxwellShaderConfigCommon { + explicit MaxwellShaderConfigCommon(){}; +}; + +struct MaxwellVSConfig : MaxwellShaderConfigCommon { + explicit MaxwellVSConfig() : MaxwellShaderConfigCommon() {} + + bool operator==(const MaxwellVSConfig& o) const { + return std::memcmp(this, &o, sizeof(MaxwellVSConfig)) == 0; + }; +}; + +struct MaxwellFSConfig : MaxwellShaderConfigCommon { + explicit MaxwellFSConfig() : MaxwellShaderConfigCommon() {} + + bool operator==(const MaxwellFSConfig& o) const { + return std::memcmp(this, &o, sizeof(MaxwellFSConfig)) == 0; + }; +}; + +std::string GenerateVertexShader(const MaxwellVSConfig& config); +std::string GenerateFragmentShader(const MaxwellFSConfig& config); + +} // namespace GLShader + +namespace std { + +template <> +struct hash<GLShader::MaxwellVSConfig> { + size_t operator()(const GLShader::MaxwellVSConfig& k) const { + return Common::ComputeHash64(&k, sizeof(GLShader::MaxwellVSConfig)); + } +}; + +template <> +struct hash<GLShader::MaxwellFSConfig> { + size_t operator()(const GLShader::MaxwellFSConfig& k) const { + return Common::ComputeHash64(&k, sizeof(GLShader::MaxwellFSConfig)); + } +}; + +} // namespace std diff --git a/src/video_core/renderer_opengl/gl_shader_util.cpp b/src/video_core/renderer_opengl/gl_shader_util.cpp index 4da241d83..a6c6204d5 100644 --- a/src/video_core/renderer_opengl/gl_shader_util.cpp +++ b/src/video_core/renderer_opengl/gl_shader_util.cpp @@ -10,53 +10,85 @@ namespace GLShader { -GLuint LoadProgram(const char* vertex_shader, const char* fragment_shader) { - +GLuint LoadProgram(const char* vertex_shader, const char* geometry_shader, + const char* fragment_shader, const std::vector<const char*>& feedback_vars, + bool separable_program) { // Create the shaders - GLuint vertex_shader_id = glCreateShader(GL_VERTEX_SHADER); - GLuint fragment_shader_id = glCreateShader(GL_FRAGMENT_SHADER); + GLuint vertex_shader_id = vertex_shader ? glCreateShader(GL_VERTEX_SHADER) : 0; + GLuint geometry_shader_id = geometry_shader ? glCreateShader(GL_GEOMETRY_SHADER) : 0; + GLuint fragment_shader_id = fragment_shader ? glCreateShader(GL_FRAGMENT_SHADER) : 0; GLint result = GL_FALSE; int info_log_length; - // Compile Vertex Shader - LOG_DEBUG(Render_OpenGL, "Compiling vertex shader..."); - - glShaderSource(vertex_shader_id, 1, &vertex_shader, nullptr); - glCompileShader(vertex_shader_id); - - // Check Vertex Shader - glGetShaderiv(vertex_shader_id, GL_COMPILE_STATUS, &result); - glGetShaderiv(vertex_shader_id, GL_INFO_LOG_LENGTH, &info_log_length); - - if (info_log_length > 1) { - std::vector<char> vertex_shader_error(info_log_length); - glGetShaderInfoLog(vertex_shader_id, info_log_length, nullptr, &vertex_shader_error[0]); - if (result == GL_TRUE) { - LOG_DEBUG(Render_OpenGL, "%s", &vertex_shader_error[0]); - } else { - LOG_ERROR(Render_OpenGL, "Error compiling vertex shader:\n%s", &vertex_shader_error[0]); + if (vertex_shader) { + // Compile Vertex Shader + LOG_DEBUG(Render_OpenGL, "Compiling vertex shader..."); + + glShaderSource(vertex_shader_id, 1, &vertex_shader, nullptr); + glCompileShader(vertex_shader_id); + + // Check Vertex Shader + glGetShaderiv(vertex_shader_id, GL_COMPILE_STATUS, &result); + glGetShaderiv(vertex_shader_id, GL_INFO_LOG_LENGTH, &info_log_length); + + if (info_log_length > 1) { + std::vector<char> vertex_shader_error(info_log_length); + glGetShaderInfoLog(vertex_shader_id, info_log_length, nullptr, &vertex_shader_error[0]); + if (result == GL_TRUE) { + LOG_DEBUG(Render_OpenGL, "%s", &vertex_shader_error[0]); + } else { + LOG_CRITICAL(Render_OpenGL, "Error compiling vertex shader:\n%s", + &vertex_shader_error[0]); + } } } - // Compile Fragment Shader - LOG_DEBUG(Render_OpenGL, "Compiling fragment shader..."); - - glShaderSource(fragment_shader_id, 1, &fragment_shader, nullptr); - glCompileShader(fragment_shader_id); - - // Check Fragment Shader - glGetShaderiv(fragment_shader_id, GL_COMPILE_STATUS, &result); - glGetShaderiv(fragment_shader_id, GL_INFO_LOG_LENGTH, &info_log_length); + if (geometry_shader) { + // Compile Geometry Shader + LOG_DEBUG(Render_OpenGL, "Compiling geometry shader..."); + + glShaderSource(geometry_shader_id, 1, &geometry_shader, nullptr); + glCompileShader(geometry_shader_id); + + // Check Geometry Shader + glGetShaderiv(geometry_shader_id, GL_COMPILE_STATUS, &result); + glGetShaderiv(geometry_shader_id, GL_INFO_LOG_LENGTH, &info_log_length); + + if (info_log_length > 1) { + std::vector<char> geometry_shader_error(info_log_length); + glGetShaderInfoLog(geometry_shader_id, info_log_length, nullptr, + &geometry_shader_error[0]); + if (result == GL_TRUE) { + LOG_DEBUG(Render_OpenGL, "%s", &geometry_shader_error[0]); + } else { + LOG_CRITICAL(Render_OpenGL, "Error compiling geometry shader:\n%s", + &geometry_shader_error[0]); + } + } + } - if (info_log_length > 1) { - std::vector<char> fragment_shader_error(info_log_length); - glGetShaderInfoLog(fragment_shader_id, info_log_length, nullptr, &fragment_shader_error[0]); - if (result == GL_TRUE) { - LOG_DEBUG(Render_OpenGL, "%s", &fragment_shader_error[0]); - } else { - LOG_ERROR(Render_OpenGL, "Error compiling fragment shader:\n%s", - &fragment_shader_error[0]); + if (fragment_shader) { + // Compile Fragment Shader + LOG_DEBUG(Render_OpenGL, "Compiling fragment shader..."); + + glShaderSource(fragment_shader_id, 1, &fragment_shader, nullptr); + glCompileShader(fragment_shader_id); + + // Check Fragment Shader + glGetShaderiv(fragment_shader_id, GL_COMPILE_STATUS, &result); + glGetShaderiv(fragment_shader_id, GL_INFO_LOG_LENGTH, &info_log_length); + + if (info_log_length > 1) { + std::vector<char> fragment_shader_error(info_log_length); + glGetShaderInfoLog(fragment_shader_id, info_log_length, nullptr, + &fragment_shader_error[0]); + if (result == GL_TRUE) { + LOG_DEBUG(Render_OpenGL, "%s", &fragment_shader_error[0]); + } else { + LOG_CRITICAL(Render_OpenGL, "Error compiling fragment shader:\n%s", + &fragment_shader_error[0]); + } } } @@ -64,8 +96,25 @@ GLuint LoadProgram(const char* vertex_shader, const char* fragment_shader) { LOG_DEBUG(Render_OpenGL, "Linking program..."); GLuint program_id = glCreateProgram(); - glAttachShader(program_id, vertex_shader_id); - glAttachShader(program_id, fragment_shader_id); + if (vertex_shader) { + glAttachShader(program_id, vertex_shader_id); + } + if (geometry_shader) { + glAttachShader(program_id, geometry_shader_id); + } + if (fragment_shader) { + glAttachShader(program_id, fragment_shader_id); + } + + if (!feedback_vars.empty()) { + auto varyings = feedback_vars; + glTransformFeedbackVaryings(program_id, static_cast<GLsizei>(feedback_vars.size()), + &varyings[0], GL_INTERLEAVED_ATTRIBS); + } + + if (separable_program) { + glProgramParameteri(program_id, GL_PROGRAM_SEPARABLE, GL_TRUE); + } glLinkProgram(program_id); @@ -79,19 +128,36 @@ GLuint LoadProgram(const char* vertex_shader, const char* fragment_shader) { if (result == GL_TRUE) { LOG_DEBUG(Render_OpenGL, "%s", &program_error[0]); } else { - LOG_ERROR(Render_OpenGL, "Error linking shader:\n%s", &program_error[0]); + LOG_CRITICAL(Render_OpenGL, "Error linking shader:\n%s", &program_error[0]); } } // If the program linking failed at least one of the shaders was probably bad if (result == GL_FALSE) { - LOG_ERROR(Render_OpenGL, "Vertex shader:\n%s", vertex_shader); - LOG_ERROR(Render_OpenGL, "Fragment shader:\n%s", fragment_shader); + if (vertex_shader) { + LOG_CRITICAL(Render_OpenGL, "Vertex shader:\n%s", vertex_shader); + } + if (geometry_shader) { + LOG_CRITICAL(Render_OpenGL, "Geometry shader:\n%s", geometry_shader); + } + if (fragment_shader) { + LOG_CRITICAL(Render_OpenGL, "Fragment shader:\n%s", fragment_shader); + } } ASSERT_MSG(result == GL_TRUE, "Shader not linked"); - glDeleteShader(vertex_shader_id); - glDeleteShader(fragment_shader_id); + if (vertex_shader) { + glDetachShader(program_id, vertex_shader_id); + glDeleteShader(vertex_shader_id); + } + if (geometry_shader) { + glDetachShader(program_id, geometry_shader_id); + glDeleteShader(geometry_shader_id); + } + if (fragment_shader) { + glDetachShader(program_id, fragment_shader_id); + glDeleteShader(fragment_shader_id); + } return program_id; } diff --git a/src/video_core/renderer_opengl/gl_shader_util.h b/src/video_core/renderer_opengl/gl_shader_util.h index a4bcffdfa..fc7b5e080 100644 --- a/src/video_core/renderer_opengl/gl_shader_util.h +++ b/src/video_core/renderer_opengl/gl_shader_util.h @@ -4,6 +4,7 @@ #pragma once +#include <vector> #include <glad/glad.h> namespace GLShader { @@ -11,9 +12,12 @@ namespace GLShader { /** * Utility function to create and compile an OpenGL GLSL shader program (vertex + fragment shader) * @param vertex_shader String of the GLSL vertex shader program + * @param geometry_shader String of the GLSL geometry shader program * @param fragment_shader String of the GLSL fragment shader program * @returns Handle of the newly created OpenGL shader object */ -GLuint LoadProgram(const char* vertex_shader, const char* fragment_shader); +GLuint LoadProgram(const char* vertex_shader, const char* geometry_shader, + const char* fragment_shader, const std::vector<const char*>& feedback_vars = {}, + bool separable_program = false); } // namespace GLShader diff --git a/src/video_core/renderer_opengl/gl_state.cpp b/src/video_core/renderer_opengl/gl_state.cpp index 5770ae08f..1d396728b 100644 --- a/src/video_core/renderer_opengl/gl_state.cpp +++ b/src/video_core/renderer_opengl/gl_state.cpp @@ -33,7 +33,7 @@ OpenGLState::OpenGLState() { stencil.action_depth_pass = GL_KEEP; stencil.action_stencil_fail = GL_KEEP; - blend.enabled = false; + blend.enabled = true; blend.rgb_equation = GL_FUNC_ADD; blend.a_equation = GL_FUNC_ADD; blend.src_rgb_func = GL_ONE; @@ -68,6 +68,18 @@ OpenGLState::OpenGLState() { draw.vertex_buffer = 0; draw.uniform_buffer = 0; draw.shader_program = 0; + draw.program_pipeline = 0; + + scissor.enabled = false; + scissor.x = 0; + scissor.y = 0; + scissor.width = 0; + scissor.height = 0; + + viewport.x = 0; + viewport.y = 0; + viewport.width = 0; + viewport.height = 0; clip_distance = {}; } @@ -148,9 +160,6 @@ void OpenGLState::Apply() const { if (blend.enabled != cur_state.blend.enabled) { if (blend.enabled) { glEnable(GL_BLEND); - - cur_state.logic_op = GL_COPY; - glLogicOp(cur_state.logic_op); glDisable(GL_COLOR_LOGIC_OP); } else { glDisable(GL_BLEND); @@ -196,7 +205,7 @@ void OpenGLState::Apply() const { // Lighting LUTs if (lighting_lut.texture_buffer != cur_state.lighting_lut.texture_buffer) { glActiveTexture(TextureUnits::LightingLUT.Enum()); - glBindTexture(GL_TEXTURE_BUFFER, cur_state.lighting_lut.texture_buffer); + glBindTexture(GL_TEXTURE_BUFFER, lighting_lut.texture_buffer); } // Fog LUT @@ -263,6 +272,31 @@ void OpenGLState::Apply() const { glUseProgram(draw.shader_program); } + // Program pipeline + if (draw.program_pipeline != cur_state.draw.program_pipeline) { + glBindProgramPipeline(draw.program_pipeline); + } + + // Scissor test + if (scissor.enabled != cur_state.scissor.enabled) { + if (scissor.enabled) { + glEnable(GL_SCISSOR_TEST); + } else { + glDisable(GL_SCISSOR_TEST); + } + } + + if (scissor.x != cur_state.scissor.x || scissor.y != cur_state.scissor.y || + scissor.width != cur_state.scissor.width || scissor.height != cur_state.scissor.height) { + glScissor(scissor.x, scissor.y, scissor.width, scissor.height); + } + + if (viewport.x != cur_state.viewport.x || viewport.y != cur_state.viewport.y || + viewport.width != cur_state.viewport.width || + viewport.height != cur_state.viewport.height) { + glViewport(viewport.x, viewport.y, viewport.width, viewport.height); + } + // Clip distance for (size_t i = 0; i < clip_distance.size(); ++i) { if (clip_distance[i] != cur_state.clip_distance[i]) { @@ -277,62 +311,75 @@ void OpenGLState::Apply() const { cur_state = *this; } -void OpenGLState::ResetTexture(GLuint handle) { - for (auto& unit : cur_state.texture_units) { +OpenGLState& OpenGLState::ResetTexture(GLuint handle) { + for (auto& unit : texture_units) { if (unit.texture_2d == handle) { unit.texture_2d = 0; } } - if (cur_state.lighting_lut.texture_buffer == handle) - cur_state.lighting_lut.texture_buffer = 0; - if (cur_state.fog_lut.texture_buffer == handle) - cur_state.fog_lut.texture_buffer = 0; - if (cur_state.proctex_noise_lut.texture_buffer == handle) - cur_state.proctex_noise_lut.texture_buffer = 0; - if (cur_state.proctex_color_map.texture_buffer == handle) - cur_state.proctex_color_map.texture_buffer = 0; - if (cur_state.proctex_alpha_map.texture_buffer == handle) - cur_state.proctex_alpha_map.texture_buffer = 0; - if (cur_state.proctex_lut.texture_buffer == handle) - cur_state.proctex_lut.texture_buffer = 0; - if (cur_state.proctex_diff_lut.texture_buffer == handle) - cur_state.proctex_diff_lut.texture_buffer = 0; + if (lighting_lut.texture_buffer == handle) + lighting_lut.texture_buffer = 0; + if (fog_lut.texture_buffer == handle) + fog_lut.texture_buffer = 0; + if (proctex_noise_lut.texture_buffer == handle) + proctex_noise_lut.texture_buffer = 0; + if (proctex_color_map.texture_buffer == handle) + proctex_color_map.texture_buffer = 0; + if (proctex_alpha_map.texture_buffer == handle) + proctex_alpha_map.texture_buffer = 0; + if (proctex_lut.texture_buffer == handle) + proctex_lut.texture_buffer = 0; + if (proctex_diff_lut.texture_buffer == handle) + proctex_diff_lut.texture_buffer = 0; + return *this; } -void OpenGLState::ResetSampler(GLuint handle) { - for (auto& unit : cur_state.texture_units) { +OpenGLState& OpenGLState::ResetSampler(GLuint handle) { + for (auto& unit : texture_units) { if (unit.sampler == handle) { unit.sampler = 0; } } + return *this; +} + +OpenGLState& OpenGLState::ResetProgram(GLuint handle) { + if (draw.shader_program == handle) { + draw.shader_program = 0; + } + return *this; } -void OpenGLState::ResetProgram(GLuint handle) { - if (cur_state.draw.shader_program == handle) { - cur_state.draw.shader_program = 0; +OpenGLState& OpenGLState::ResetPipeline(GLuint handle) { + if (draw.program_pipeline == handle) { + draw.program_pipeline = 0; } + return *this; } -void OpenGLState::ResetBuffer(GLuint handle) { - if (cur_state.draw.vertex_buffer == handle) { - cur_state.draw.vertex_buffer = 0; +OpenGLState& OpenGLState::ResetBuffer(GLuint handle) { + if (draw.vertex_buffer == handle) { + draw.vertex_buffer = 0; } - if (cur_state.draw.uniform_buffer == handle) { - cur_state.draw.uniform_buffer = 0; + if (draw.uniform_buffer == handle) { + draw.uniform_buffer = 0; } + return *this; } -void OpenGLState::ResetVertexArray(GLuint handle) { - if (cur_state.draw.vertex_array == handle) { - cur_state.draw.vertex_array = 0; +OpenGLState& OpenGLState::ResetVertexArray(GLuint handle) { + if (draw.vertex_array == handle) { + draw.vertex_array = 0; } + return *this; } -void OpenGLState::ResetFramebuffer(GLuint handle) { - if (cur_state.draw.read_framebuffer == handle) { - cur_state.draw.read_framebuffer = 0; +OpenGLState& OpenGLState::ResetFramebuffer(GLuint handle) { + if (draw.read_framebuffer == handle) { + draw.read_framebuffer = 0; } - if (cur_state.draw.draw_framebuffer == handle) { - cur_state.draw.draw_framebuffer = 0; + if (draw.draw_framebuffer == handle) { + draw.draw_framebuffer = 0; } + return *this; } diff --git a/src/video_core/renderer_opengl/gl_state.h b/src/video_core/renderer_opengl/gl_state.h index 437fe34c4..c1f4efc8c 100644 --- a/src/video_core/renderer_opengl/gl_state.h +++ b/src/video_core/renderer_opengl/gl_state.h @@ -85,7 +85,7 @@ public: struct { GLuint texture_2d; // GL_TEXTURE_BINDING_2D GLuint sampler; // GL_SAMPLER_BINDING - } texture_units[3]; + } texture_units[32]; struct { GLuint texture_buffer; // GL_TEXTURE_BINDING_BUFFER @@ -122,27 +122,44 @@ public: GLuint vertex_buffer; // GL_ARRAY_BUFFER_BINDING GLuint uniform_buffer; // GL_UNIFORM_BUFFER_BINDING GLuint shader_program; // GL_CURRENT_PROGRAM + GLuint program_pipeline; // GL_PROGRAM_PIPELINE_BINDING } draw; + struct { + bool enabled; // GL_SCISSOR_TEST + GLint x; + GLint y; + GLsizei width; + GLsizei height; + } scissor; + + struct { + GLint x; + GLint y; + GLsizei width; + GLsizei height; + } viewport; + std::array<bool, 2> clip_distance; // GL_CLIP_DISTANCE OpenGLState(); /// Get the currently active OpenGL state - static const OpenGLState& GetCurState() { + static OpenGLState GetCurState() { return cur_state; } /// Apply this state as the current OpenGL state void Apply() const; - /// Resets and unbinds any references to the given resource in the current OpenGL state - static void ResetTexture(GLuint handle); - static void ResetSampler(GLuint handle); - static void ResetProgram(GLuint handle); - static void ResetBuffer(GLuint handle); - static void ResetVertexArray(GLuint handle); - static void ResetFramebuffer(GLuint handle); + /// Resets any references to the given resource + OpenGLState& ResetTexture(GLuint handle); + OpenGLState& ResetSampler(GLuint handle); + OpenGLState& ResetProgram(GLuint handle); + OpenGLState& ResetPipeline(GLuint handle); + OpenGLState& ResetBuffer(GLuint handle); + OpenGLState& ResetVertexArray(GLuint handle); + OpenGLState& ResetFramebuffer(GLuint handle); private: static OpenGLState cur_state; diff --git a/src/video_core/renderer_opengl/gl_stream_buffer.cpp b/src/video_core/renderer_opengl/gl_stream_buffer.cpp new file mode 100644 index 000000000..a2713e9f0 --- /dev/null +++ b/src/video_core/renderer_opengl/gl_stream_buffer.cpp @@ -0,0 +1,182 @@ +// Copyright 2018 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <deque> +#include <vector> +#include "common/alignment.h" +#include "common/assert.h" +#include "video_core/renderer_opengl/gl_state.h" +#include "video_core/renderer_opengl/gl_stream_buffer.h" + +class OrphanBuffer : public OGLStreamBuffer { +public: + explicit OrphanBuffer(GLenum target) : OGLStreamBuffer(target) {} + ~OrphanBuffer() override; + +private: + void Create(size_t size, size_t sync_subdivide) override; + void Release() override; + + std::pair<u8*, GLintptr> Map(size_t size, size_t alignment) override; + void Unmap() override; + + std::vector<u8> data; +}; + +class StorageBuffer : public OGLStreamBuffer { +public: + explicit StorageBuffer(GLenum target) : OGLStreamBuffer(target) {} + ~StorageBuffer() override; + +private: + void Create(size_t size, size_t sync_subdivide) override; + void Release() override; + + std::pair<u8*, GLintptr> Map(size_t size, size_t alignment) override; + void Unmap() override; + + struct Fence { + OGLSync sync; + size_t offset; + }; + std::deque<Fence> head; + std::deque<Fence> tail; + + u8* mapped_ptr; +}; + +OGLStreamBuffer::OGLStreamBuffer(GLenum target) { + gl_target = target; +} + +GLuint OGLStreamBuffer::GetHandle() const { + return gl_buffer.handle; +} + +std::unique_ptr<OGLStreamBuffer> OGLStreamBuffer::MakeBuffer(bool storage_buffer, GLenum target) { + if (storage_buffer) { + return std::make_unique<StorageBuffer>(target); + } + return std::make_unique<OrphanBuffer>(target); +} + +OrphanBuffer::~OrphanBuffer() { + Release(); +} + +void OrphanBuffer::Create(size_t size, size_t /*sync_subdivide*/) { + buffer_pos = 0; + buffer_size = size; + data.resize(buffer_size); + + if (gl_buffer.handle == 0) { + gl_buffer.Create(); + glBindBuffer(gl_target, gl_buffer.handle); + } + + glBufferData(gl_target, static_cast<GLsizeiptr>(buffer_size), nullptr, GL_STREAM_DRAW); +} + +void OrphanBuffer::Release() { + gl_buffer.Release(); +} + +std::pair<u8*, GLintptr> OrphanBuffer::Map(size_t size, size_t alignment) { + buffer_pos = Common::AlignUp(buffer_pos, alignment); + + if (buffer_pos + size > buffer_size) { + Create(std::max(buffer_size, size), 0); + } + + mapped_size = size; + return std::make_pair(&data[buffer_pos], static_cast<GLintptr>(buffer_pos)); +} + +void OrphanBuffer::Unmap() { + glBufferSubData(gl_target, static_cast<GLintptr>(buffer_pos), + static_cast<GLsizeiptr>(mapped_size), &data[buffer_pos]); + buffer_pos += mapped_size; +} + +StorageBuffer::~StorageBuffer() { + Release(); +} + +void StorageBuffer::Create(size_t size, size_t sync_subdivide) { + if (gl_buffer.handle != 0) + return; + + buffer_pos = 0; + buffer_size = size; + buffer_sync_subdivide = std::max<size_t>(sync_subdivide, 1); + + gl_buffer.Create(); + glBindBuffer(gl_target, gl_buffer.handle); + + glBufferStorage(gl_target, static_cast<GLsizeiptr>(buffer_size), nullptr, + GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT); + mapped_ptr = reinterpret_cast<u8*>( + glMapBufferRange(gl_target, 0, static_cast<GLsizeiptr>(buffer_size), + GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | GL_MAP_FLUSH_EXPLICIT_BIT)); +} + +void StorageBuffer::Release() { + if (gl_buffer.handle == 0) + return; + + glUnmapBuffer(gl_target); + + gl_buffer.Release(); + head.clear(); + tail.clear(); +} + +std::pair<u8*, GLintptr> StorageBuffer::Map(size_t size, size_t alignment) { + ASSERT(size <= buffer_size); + + OGLSync sync; + + buffer_pos = Common::AlignUp(buffer_pos, alignment); + size_t effective_offset = Common::AlignDown(buffer_pos, buffer_sync_subdivide); + + if (!head.empty() && + (effective_offset > head.back().offset || buffer_pos + size > buffer_size)) { + ASSERT(head.back().sync.handle == 0); + head.back().sync.Create(); + } + + if (buffer_pos + size > buffer_size) { + if (!tail.empty()) { + std::swap(sync, tail.back().sync); + tail.clear(); + } + std::swap(tail, head); + buffer_pos = 0; + effective_offset = 0; + } + + while (!tail.empty() && buffer_pos + size > tail.front().offset) { + std::swap(sync, tail.front().sync); + tail.pop_front(); + } + + if (sync.handle != 0) { + glClientWaitSync(sync.handle, GL_SYNC_FLUSH_COMMANDS_BIT, GL_TIMEOUT_IGNORED); + sync.Release(); + } + + if (head.empty() || effective_offset > head.back().offset) { + head.emplace_back(); + head.back().offset = effective_offset; + } + + mapped_size = size; + return std::make_pair(&mapped_ptr[buffer_pos], static_cast<GLintptr>(buffer_pos)); +} + +void StorageBuffer::Unmap() { + glFlushMappedBufferRange(gl_target, static_cast<GLintptr>(buffer_pos), + static_cast<GLsizeiptr>(mapped_size)); + buffer_pos += mapped_size; +} diff --git a/src/video_core/renderer_opengl/gl_stream_buffer.h b/src/video_core/renderer_opengl/gl_stream_buffer.h new file mode 100644 index 000000000..4bc2f52e0 --- /dev/null +++ b/src/video_core/renderer_opengl/gl_stream_buffer.h @@ -0,0 +1,34 @@ +// Copyright 2018 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <memory> +#include <glad/glad.h> +#include "common/common_types.h" +#include "video_core/renderer_opengl/gl_resource_manager.h" + +class OGLStreamBuffer : private NonCopyable { +public: + explicit OGLStreamBuffer(GLenum target); + virtual ~OGLStreamBuffer() = default; + +public: + static std::unique_ptr<OGLStreamBuffer> MakeBuffer(bool storage_buffer, GLenum target); + + virtual void Create(size_t size, size_t sync_subdivide) = 0; + virtual void Release() {} + + GLuint GetHandle() const; + + virtual std::pair<u8*, GLintptr> Map(size_t size, size_t alignment) = 0; + virtual void Unmap() = 0; + +protected: + OGLBuffer gl_buffer; + GLenum gl_target; + + size_t buffer_pos = 0; + size_t buffer_size = 0; + size_t buffer_sync_subdivide = 0; + size_t mapped_size = 0; +}; diff --git a/src/video_core/renderer_opengl/maxwell_to_gl.h b/src/video_core/renderer_opengl/maxwell_to_gl.h new file mode 100644 index 000000000..d847317ac --- /dev/null +++ b/src/video_core/renderer_opengl/maxwell_to_gl.h @@ -0,0 +1,50 @@ +// Copyright 2018 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <array> +#include <glad/glad.h> +#include "common/common_types.h" +#include "common/logging/log.h" +#include "video_core/engines/maxwell_3d.h" + +namespace MaxwellToGL { + +using Maxwell = Tegra::Engines::Maxwell3D::Regs; + +inline GLenum VertexType(Maxwell::VertexAttribute attrib) { + switch (attrib.type) { + case Maxwell::VertexAttribute::Type::UnsignedNorm: { + + switch (attrib.size) { + case Maxwell::VertexAttribute::Size::Size_8_8_8_8: + return GL_UNSIGNED_BYTE; + } + + LOG_CRITICAL(Render_OpenGL, "Unimplemented vertex size=%s", attrib.SizeString().c_str()); + UNREACHABLE(); + return {}; + } + + case Maxwell::VertexAttribute::Type::Float: + return GL_FLOAT; + } + + LOG_CRITICAL(Render_OpenGL, "Unimplemented vertex type=%s", attrib.TypeString().c_str()); + UNREACHABLE(); + return {}; +} + +inline GLenum PrimitiveTopology(Maxwell::PrimitiveTopology topology) { + switch (topology) { + case Maxwell::PrimitiveTopology::TriangleStrip: + return GL_TRIANGLE_STRIP; + } + LOG_CRITICAL(Render_OpenGL, "Unimplemented primitive topology=%d", topology); + UNREACHABLE(); + return {}; +} + +} // namespace MaxwellToGL diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp index 7f921fa32..78b50b227 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.cpp +++ b/src/video_core/renderer_opengl/renderer_opengl.cpp @@ -20,6 +20,7 @@ #include "core/settings.h" #include "core/tracer/recorder.h" #include "video_core/renderer_opengl/renderer_opengl.h" +#include "video_core/utils.h" #include "video_core/video_core.h" static const char vertex_shader[] = R"( @@ -98,197 +99,88 @@ RendererOpenGL::RendererOpenGL() = default; RendererOpenGL::~RendererOpenGL() = default; /// Swap buffers (render frame) -void RendererOpenGL::SwapBuffers(boost::optional<const FramebufferInfo&> framebuffer_info) { +void RendererOpenGL::SwapBuffers(boost::optional<const Tegra::FramebufferConfig&> framebuffer) { + Core::System::GetInstance().perf_stats.EndSystemFrame(); + // Maintain the rasterizer's state as a priority OpenGLState prev_state = OpenGLState::GetCurState(); state.Apply(); - if (framebuffer_info != boost::none) { - // If framebuffer_info is provided, reload it from memory to a texture - if (screen_info.texture.width != (GLsizei)framebuffer_info->width || - screen_info.texture.height != (GLsizei)framebuffer_info->height || - screen_info.texture.pixel_format != framebuffer_info->pixel_format) { + if (framebuffer != boost::none) { + // If framebuffer is provided, reload it from memory to a texture + if (screen_info.texture.width != (GLsizei)framebuffer->width || + screen_info.texture.height != (GLsizei)framebuffer->height || + screen_info.texture.pixel_format != framebuffer->pixel_format) { // Reallocate texture if the framebuffer size has changed. // This is expected to not happen very often and hence should not be a // performance problem. - ConfigureFramebufferTexture(screen_info.texture, *framebuffer_info); + ConfigureFramebufferTexture(screen_info.texture, *framebuffer); } - LoadFBToScreenInfo(*framebuffer_info, screen_info); - } - - DrawScreens(); - Core::System::GetInstance().perf_stats.EndSystemFrame(); + // Load the framebuffer from memory, draw it to the screen, and swap buffers + LoadFBToScreenInfo(*framebuffer, screen_info); + DrawScreen(); + render_window->SwapBuffers(); + } - // Swap buffers render_window->PollEvents(); - render_window->SwapBuffers(); Core::System::GetInstance().frame_limiter.DoFrameLimiting(CoreTiming::GetGlobalTimeUs()); Core::System::GetInstance().perf_stats.BeginSystemFrame(); + // Restore the rasterizer state prev_state.Apply(); RefreshRasterizerSetting(); } -static inline u32 MortonInterleave128(u32 x, u32 y) { - // 128x128 Z-Order coordinate from 2D coordinates - static constexpr u32 xlut[] = { - 0x0000, 0x0001, 0x0002, 0x0003, 0x0008, 0x0009, 0x000a, 0x000b, 0x0040, 0x0041, 0x0042, - 0x0043, 0x0048, 0x0049, 0x004a, 0x004b, 0x0800, 0x0801, 0x0802, 0x0803, 0x0808, 0x0809, - 0x080a, 0x080b, 0x0840, 0x0841, 0x0842, 0x0843, 0x0848, 0x0849, 0x084a, 0x084b, 0x1000, - 0x1001, 0x1002, 0x1003, 0x1008, 0x1009, 0x100a, 0x100b, 0x1040, 0x1041, 0x1042, 0x1043, - 0x1048, 0x1049, 0x104a, 0x104b, 0x1800, 0x1801, 0x1802, 0x1803, 0x1808, 0x1809, 0x180a, - 0x180b, 0x1840, 0x1841, 0x1842, 0x1843, 0x1848, 0x1849, 0x184a, 0x184b, 0x2000, 0x2001, - 0x2002, 0x2003, 0x2008, 0x2009, 0x200a, 0x200b, 0x2040, 0x2041, 0x2042, 0x2043, 0x2048, - 0x2049, 0x204a, 0x204b, 0x2800, 0x2801, 0x2802, 0x2803, 0x2808, 0x2809, 0x280a, 0x280b, - 0x2840, 0x2841, 0x2842, 0x2843, 0x2848, 0x2849, 0x284a, 0x284b, 0x3000, 0x3001, 0x3002, - 0x3003, 0x3008, 0x3009, 0x300a, 0x300b, 0x3040, 0x3041, 0x3042, 0x3043, 0x3048, 0x3049, - 0x304a, 0x304b, 0x3800, 0x3801, 0x3802, 0x3803, 0x3808, 0x3809, 0x380a, 0x380b, 0x3840, - 0x3841, 0x3842, 0x3843, 0x3848, 0x3849, 0x384a, 0x384b, 0x0000, 0x0001, 0x0002, 0x0003, - 0x0008, 0x0009, 0x000a, 0x000b, 0x0040, 0x0041, 0x0042, 0x0043, 0x0048, 0x0049, 0x004a, - 0x004b, 0x0800, 0x0801, 0x0802, 0x0803, 0x0808, 0x0809, 0x080a, 0x080b, 0x0840, 0x0841, - 0x0842, 0x0843, 0x0848, 0x0849, 0x084a, 0x084b, 0x1000, 0x1001, 0x1002, 0x1003, 0x1008, - 0x1009, 0x100a, 0x100b, 0x1040, 0x1041, 0x1042, 0x1043, 0x1048, 0x1049, 0x104a, 0x104b, - 0x1800, 0x1801, 0x1802, 0x1803, 0x1808, 0x1809, 0x180a, 0x180b, 0x1840, 0x1841, 0x1842, - 0x1843, 0x1848, 0x1849, 0x184a, 0x184b, 0x2000, 0x2001, 0x2002, 0x2003, 0x2008, 0x2009, - 0x200a, 0x200b, 0x2040, 0x2041, 0x2042, 0x2043, 0x2048, 0x2049, 0x204a, 0x204b, 0x2800, - 0x2801, 0x2802, 0x2803, 0x2808, 0x2809, 0x280a, 0x280b, 0x2840, 0x2841, 0x2842, 0x2843, - 0x2848, 0x2849, 0x284a, 0x284b, 0x3000, 0x3001, 0x3002, 0x3003, 0x3008, 0x3009, 0x300a, - 0x300b, 0x3040, 0x3041, 0x3042, 0x3043, 0x3048, 0x3049, 0x304a, 0x304b, 0x3800, 0x3801, - 0x3802, 0x3803, 0x3808, 0x3809, 0x380a, 0x380b, 0x3840, 0x3841, 0x3842, 0x3843, 0x3848, - 0x3849, 0x384a, 0x384b, 0x0000, 0x0001, 0x0002, 0x0003, 0x0008, 0x0009, 0x000a, 0x000b, - 0x0040, 0x0041, 0x0042, 0x0043, 0x0048, 0x0049, 0x004a, 0x004b, 0x0800, 0x0801, 0x0802, - 0x0803, 0x0808, 0x0809, 0x080a, 0x080b, 0x0840, 0x0841, 0x0842, 0x0843, 0x0848, 0x0849, - 0x084a, 0x084b, 0x1000, 0x1001, 0x1002, 0x1003, 0x1008, 0x1009, 0x100a, 0x100b, 0x1040, - 0x1041, 0x1042, 0x1043, 0x1048, 0x1049, 0x104a, 0x104b, 0x1800, 0x1801, 0x1802, 0x1803, - 0x1808, 0x1809, 0x180a, 0x180b, 0x1840, 0x1841, 0x1842, 0x1843, 0x1848, 0x1849, 0x184a, - 0x184b, 0x2000, 0x2001, 0x2002, 0x2003, 0x2008, 0x2009, 0x200a, 0x200b, 0x2040, 0x2041, - 0x2042, 0x2043, 0x2048, 0x2049, 0x204a, 0x204b, 0x2800, 0x2801, 0x2802, 0x2803, 0x2808, - 0x2809, 0x280a, 0x280b, 0x2840, 0x2841, 0x2842, 0x2843, 0x2848, 0x2849, 0x284a, 0x284b, - 0x3000, 0x3001, 0x3002, 0x3003, 0x3008, 0x3009, 0x300a, 0x300b, 0x3040, 0x3041, 0x3042, - 0x3043, 0x3048, 0x3049, 0x304a, 0x304b, 0x3800, 0x3801, 0x3802, 0x3803, 0x3808, 0x3809, - 0x380a, 0x380b, 0x3840, 0x3841, 0x3842, 0x3843, 0x3848, 0x3849, 0x384a, 0x384b, - }; - static constexpr u32 ylut[] = { - 0x0000, 0x0004, 0x0010, 0x0014, 0x0020, 0x0024, 0x0030, 0x0034, 0x0080, 0x0084, 0x0090, - 0x0094, 0x00a0, 0x00a4, 0x00b0, 0x00b4, 0x0100, 0x0104, 0x0110, 0x0114, 0x0120, 0x0124, - 0x0130, 0x0134, 0x0180, 0x0184, 0x0190, 0x0194, 0x01a0, 0x01a4, 0x01b0, 0x01b4, 0x0200, - 0x0204, 0x0210, 0x0214, 0x0220, 0x0224, 0x0230, 0x0234, 0x0280, 0x0284, 0x0290, 0x0294, - 0x02a0, 0x02a4, 0x02b0, 0x02b4, 0x0300, 0x0304, 0x0310, 0x0314, 0x0320, 0x0324, 0x0330, - 0x0334, 0x0380, 0x0384, 0x0390, 0x0394, 0x03a0, 0x03a4, 0x03b0, 0x03b4, 0x0400, 0x0404, - 0x0410, 0x0414, 0x0420, 0x0424, 0x0430, 0x0434, 0x0480, 0x0484, 0x0490, 0x0494, 0x04a0, - 0x04a4, 0x04b0, 0x04b4, 0x0500, 0x0504, 0x0510, 0x0514, 0x0520, 0x0524, 0x0530, 0x0534, - 0x0580, 0x0584, 0x0590, 0x0594, 0x05a0, 0x05a4, 0x05b0, 0x05b4, 0x0600, 0x0604, 0x0610, - 0x0614, 0x0620, 0x0624, 0x0630, 0x0634, 0x0680, 0x0684, 0x0690, 0x0694, 0x06a0, 0x06a4, - 0x06b0, 0x06b4, 0x0700, 0x0704, 0x0710, 0x0714, 0x0720, 0x0724, 0x0730, 0x0734, 0x0780, - 0x0784, 0x0790, 0x0794, 0x07a0, 0x07a4, 0x07b0, 0x07b4, 0x0000, 0x0004, 0x0010, 0x0014, - 0x0020, 0x0024, 0x0030, 0x0034, 0x0080, 0x0084, 0x0090, 0x0094, 0x00a0, 0x00a4, 0x00b0, - 0x00b4, 0x0100, 0x0104, 0x0110, 0x0114, 0x0120, 0x0124, 0x0130, 0x0134, 0x0180, 0x0184, - 0x0190, 0x0194, 0x01a0, 0x01a4, 0x01b0, 0x01b4, 0x0200, 0x0204, 0x0210, 0x0214, 0x0220, - 0x0224, 0x0230, 0x0234, 0x0280, 0x0284, 0x0290, 0x0294, 0x02a0, 0x02a4, 0x02b0, 0x02b4, - 0x0300, 0x0304, 0x0310, 0x0314, 0x0320, 0x0324, 0x0330, 0x0334, 0x0380, 0x0384, 0x0390, - 0x0394, 0x03a0, 0x03a4, 0x03b0, 0x03b4, 0x0400, 0x0404, 0x0410, 0x0414, 0x0420, 0x0424, - 0x0430, 0x0434, 0x0480, 0x0484, 0x0490, 0x0494, 0x04a0, 0x04a4, 0x04b0, 0x04b4, 0x0500, - 0x0504, 0x0510, 0x0514, 0x0520, 0x0524, 0x0530, 0x0534, 0x0580, 0x0584, 0x0590, 0x0594, - 0x05a0, 0x05a4, 0x05b0, 0x05b4, 0x0600, 0x0604, 0x0610, 0x0614, 0x0620, 0x0624, 0x0630, - 0x0634, 0x0680, 0x0684, 0x0690, 0x0694, 0x06a0, 0x06a4, 0x06b0, 0x06b4, 0x0700, 0x0704, - 0x0710, 0x0714, 0x0720, 0x0724, 0x0730, 0x0734, 0x0780, 0x0784, 0x0790, 0x0794, 0x07a0, - 0x07a4, 0x07b0, 0x07b4, 0x0000, 0x0004, 0x0010, 0x0014, 0x0020, 0x0024, 0x0030, 0x0034, - 0x0080, 0x0084, 0x0090, 0x0094, 0x00a0, 0x00a4, 0x00b0, 0x00b4, 0x0100, 0x0104, 0x0110, - 0x0114, 0x0120, 0x0124, 0x0130, 0x0134, 0x0180, 0x0184, 0x0190, 0x0194, 0x01a0, 0x01a4, - 0x01b0, 0x01b4, 0x0200, 0x0204, 0x0210, 0x0214, 0x0220, 0x0224, 0x0230, 0x0234, 0x0280, - 0x0284, 0x0290, 0x0294, 0x02a0, 0x02a4, 0x02b0, 0x02b4, 0x0300, 0x0304, 0x0310, 0x0314, - 0x0320, 0x0324, 0x0330, 0x0334, 0x0380, 0x0384, 0x0390, 0x0394, 0x03a0, 0x03a4, 0x03b0, - 0x03b4, 0x0400, 0x0404, 0x0410, 0x0414, 0x0420, 0x0424, 0x0430, 0x0434, 0x0480, 0x0484, - 0x0490, 0x0494, 0x04a0, 0x04a4, 0x04b0, 0x04b4, 0x0500, 0x0504, 0x0510, 0x0514, 0x0520, - 0x0524, 0x0530, 0x0534, 0x0580, 0x0584, 0x0590, 0x0594, 0x05a0, 0x05a4, 0x05b0, 0x05b4, - 0x0600, 0x0604, 0x0610, 0x0614, 0x0620, 0x0624, 0x0630, 0x0634, 0x0680, 0x0684, 0x0690, - 0x0694, 0x06a0, 0x06a4, 0x06b0, 0x06b4, 0x0700, 0x0704, 0x0710, 0x0714, 0x0720, 0x0724, - 0x0730, 0x0734, 0x0780, 0x0784, 0x0790, 0x0794, 0x07a0, 0x07a4, 0x07b0, 0x07b4, - }; - return xlut[x % 128] + ylut[y % 128]; -} - -static inline u32 GetMortonOffset128(u32 x, u32 y, u32 bytes_per_pixel) { - // Calculates the offset of the position of the pixel in Morton order - // Framebuffer images are split into 128x128 tiles. - - const unsigned int block_height = 128; - const unsigned int coarse_x = x & ~127; - - u32 i = MortonInterleave128(x, y); - - const unsigned int offset = coarse_x * block_height; - - return (i + offset) * bytes_per_pixel; -} - -static void MortonCopyPixels128(u32 width, u32 height, u32 bytes_per_pixel, u32 gl_bytes_per_pixel, - u8* morton_data, u8* gl_data, bool morton_to_gl) { - u8* data_ptrs[2]; - for (unsigned y = 0; y < height; ++y) { - for (unsigned x = 0; x < width; ++x) { - const u32 coarse_y = y & ~127; - u32 morton_offset = - GetMortonOffset128(x, y, bytes_per_pixel) + coarse_y * width * bytes_per_pixel; - u32 gl_pixel_index = (x + (height - 1 - y) * width) * gl_bytes_per_pixel; - - data_ptrs[morton_to_gl] = morton_data + morton_offset; - data_ptrs[!morton_to_gl] = &gl_data[gl_pixel_index]; - - memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel); - } - } -} - /** * Loads framebuffer from emulated memory into the active OpenGL texture. */ -void RendererOpenGL::LoadFBToScreenInfo(const FramebufferInfo& framebuffer_info, +void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuffer, ScreenInfo& screen_info) { - const u32 bpp{FramebufferInfo::BytesPerPixel(framebuffer_info.pixel_format)}; - const u32 size_in_bytes{framebuffer_info.stride * framebuffer_info.height * bpp}; + const u32 bytes_per_pixel{Tegra::FramebufferConfig::BytesPerPixel(framebuffer.pixel_format)}; + const u64 size_in_bytes{framebuffer.stride * framebuffer.height * bytes_per_pixel}; + const VAddr framebuffer_addr{framebuffer.address + framebuffer.offset}; - MortonCopyPixels128(framebuffer_info.width, framebuffer_info.height, bpp, 4, - Memory::GetPointer(framebuffer_info.address), gl_framebuffer_data.data(), - true); - - LOG_TRACE(Render_OpenGL, "0x%08x bytes from 0x%llx(%dx%d), fmt %x", size_in_bytes, - framebuffer_info.address, framebuffer_info.width, framebuffer_info.height, - (int)framebuffer_info.pixel_format); + // Framebuffer orientation handling + framebuffer_transform_flags = framebuffer.transform_flags; // Ensure no bad interactions with GL_UNPACK_ALIGNMENT, which by default // only allows rows to have a memory alignement of 4. - ASSERT(framebuffer_info.stride % 4 == 0); + ASSERT(framebuffer.stride % 4 == 0); - framebuffer_flip_vertical = framebuffer_info.flip_vertical; + if (!Rasterizer()->AccelerateDisplay(framebuffer, framebuffer_addr, framebuffer.stride, + screen_info)) { + // Reset the screen info's display texture to its own permanent texture + screen_info.display_texture = screen_info.texture.resource.handle; + screen_info.display_texcoords = MathUtil::Rectangle<float>(0.f, 0.f, 1.f, 1.f); - // Reset the screen info's display texture to its own permanent texture - screen_info.display_texture = screen_info.texture.resource.handle; - screen_info.display_texcoords = MathUtil::Rectangle<float>(0.f, 0.f, 1.f, 1.f); + Rasterizer()->FlushRegion(framebuffer_addr, size_in_bytes); - // Memory::RasterizerFlushRegion(framebuffer_info.address, size_in_bytes); + VideoCore::MortonCopyPixels128(framebuffer.width, framebuffer.height, bytes_per_pixel, 4, + Memory::GetPointer(framebuffer_addr), + gl_framebuffer_data.data(), true); - state.texture_units[0].texture_2d = screen_info.texture.resource.handle; - state.Apply(); + state.texture_units[0].texture_2d = screen_info.texture.resource.handle; + state.Apply(); - glActiveTexture(GL_TEXTURE0); - glPixelStorei(GL_UNPACK_ROW_LENGTH, (GLint)framebuffer_info.stride); + glActiveTexture(GL_TEXTURE0); + glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(framebuffer.stride)); - // Update existing texture - // TODO: Test what happens on hardware when you change the framebuffer dimensions so that - // they differ from the LCD resolution. - // TODO: Applications could theoretically crash Citra here by specifying too large - // framebuffer sizes. We should make sure that this cannot happen. - glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, framebuffer_info.width, framebuffer_info.height, - screen_info.texture.gl_format, screen_info.texture.gl_type, - gl_framebuffer_data.data()); + // Update existing texture + // TODO: Test what happens on hardware when you change the framebuffer dimensions so that + // they differ from the LCD resolution. + // TODO: Applications could theoretically crash yuzu here by specifying too large + // framebuffer sizes. We should make sure that this cannot happen. + glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, framebuffer.width, framebuffer.height, + screen_info.texture.gl_format, screen_info.texture.gl_type, + gl_framebuffer_data.data()); - glPixelStorei(GL_UNPACK_ROW_LENGTH, 0); + glPixelStorei(GL_UNPACK_ROW_LENGTH, 0); - state.texture_units[0].texture_2d = 0; - state.Apply(); + state.texture_units[0].texture_2d = 0; + state.Apply(); + } } /** @@ -318,7 +210,7 @@ void RendererOpenGL::InitOpenGLObjects() { 0.0f); // Link shaders and get variable locations - shader.Create(vertex_shader, fragment_shader); + shader.Create(vertex_shader, nullptr, fragment_shader); state.draw.shader_program = shader.handle; state.Apply(); uniform_modelview_matrix = glGetUniformLocation(shader.handle, "modelview_matrix"); @@ -372,14 +264,14 @@ void RendererOpenGL::InitOpenGLObjects() { } void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture, - const FramebufferInfo& framebuffer_info) { + const Tegra::FramebufferConfig& framebuffer) { - texture.width = framebuffer_info.width; - texture.height = framebuffer_info.height; + texture.width = framebuffer.width; + texture.height = framebuffer.height; GLint internal_format; - switch (framebuffer_info.pixel_format) { - case FramebufferInfo::PixelFormat::ABGR8: + switch (framebuffer.pixel_format) { + case Tegra::FramebufferConfig::PixelFormat::ABGR8: // Use RGBA8 and swap in the fragment shader internal_format = GL_RGBA; texture.gl_format = GL_RGBA; @@ -387,7 +279,7 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture, gl_framebuffer_data.resize(texture.width * texture.height * 4); break; default: - UNIMPLEMENTED(); + UNREACHABLE(); } state.texture_units[0].texture_2d = texture.resource.handle; @@ -401,11 +293,22 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture, state.Apply(); } -void RendererOpenGL::DrawSingleScreen(const ScreenInfo& screen_info, float x, float y, float w, - float h) { +void RendererOpenGL::DrawScreenTriangles(const ScreenInfo& screen_info, float x, float y, float w, + float h) { const auto& texcoords = screen_info.display_texcoords; - const auto& left = framebuffer_flip_vertical ? texcoords.right : texcoords.left; - const auto& right = framebuffer_flip_vertical ? texcoords.left : texcoords.right; + auto left = texcoords.left; + auto right = texcoords.right; + if (framebuffer_transform_flags != Tegra::FramebufferConfig::TransformFlags::Unset) + if (framebuffer_transform_flags == Tegra::FramebufferConfig::TransformFlags::FlipV) { + // Flip the framebuffer vertically + left = texcoords.right; + right = texcoords.left; + } else { + // Other transformations are unsupported + LOG_CRITICAL(Render_OpenGL, "Unsupported framebuffer_transform_flags=%d", + framebuffer_transform_flags); + UNIMPLEMENTED(); + } std::array<ScreenRectVertex, 4> vertices = {{ ScreenRectVertex(x, y, texcoords.top, right), @@ -427,7 +330,7 @@ void RendererOpenGL::DrawSingleScreen(const ScreenInfo& screen_info, float x, fl /** * Draws the emulated screens to the emulator window. */ -void RendererOpenGL::DrawScreens() { +void RendererOpenGL::DrawScreen() { const auto& layout = render_window->GetFramebufferLayout(); const auto& screen = layout.screen; @@ -443,8 +346,8 @@ void RendererOpenGL::DrawScreens() { glActiveTexture(GL_TEXTURE0); glUniform1i(uniform_color_texture, 0); - DrawSingleScreen(screen_info, (float)screen.left, (float)screen.top, (float)screen.GetWidth(), - (float)screen.GetHeight()); + DrawScreenTriangles(screen_info, (float)screen.left, (float)screen.top, + (float)screen.GetWidth(), (float)screen.GetHeight()); m_current_frame++; } diff --git a/src/video_core/renderer_opengl/renderer_opengl.h b/src/video_core/renderer_opengl/renderer_opengl.h index 05bb3c5cf..fffd0f9f4 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.h +++ b/src/video_core/renderer_opengl/renderer_opengl.h @@ -21,7 +21,7 @@ struct TextureInfo { GLsizei height; GLenum gl_format; GLenum gl_type; - RendererBase::FramebufferInfo::PixelFormat pixel_format; + Tegra::FramebufferConfig::PixelFormat pixel_format; }; /// Structure used for storing information about the display target for each 3DS screen @@ -37,7 +37,7 @@ public: ~RendererOpenGL() override; /// Swap buffers (render frame) - void SwapBuffers(boost::optional<const FramebufferInfo&> framebuffer_info) override; + void SwapBuffers(boost::optional<const Tegra::FramebufferConfig&> framebuffer) override; /** * Set the emulator window to use for renderer @@ -53,13 +53,14 @@ public: private: void InitOpenGLObjects(); - void ConfigureFramebufferTexture(TextureInfo& texture, const FramebufferInfo& framebuffer_info); - void DrawScreens(); - void DrawSingleScreen(const ScreenInfo& screen_info, float x, float y, float w, float h); + void ConfigureFramebufferTexture(TextureInfo& texture, + const Tegra::FramebufferConfig& framebuffer); + void DrawScreen(); + void DrawScreenTriangles(const ScreenInfo& screen_info, float x, float y, float w, float h); void UpdateFramerate(); // Loads framebuffer from emulated memory into the display information structure - void LoadFBToScreenInfo(const FramebufferInfo& framebuffer_info, ScreenInfo& screen_info); + void LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuffer, ScreenInfo& screen_info); // Fills active OpenGL texture with the given RGBA color. void LoadColorToActiveGLTexture(u8 color_r, u8 color_g, u8 color_b, u8 color_a, const TextureInfo& texture); @@ -87,6 +88,6 @@ private: GLuint attrib_position; GLuint attrib_tex_coord; - /// Flips the framebuffer vertically when true - bool framebuffer_flip_vertical; + /// Used for transforming the framebuffer orientation + Tegra::FramebufferConfig::TransformFlags framebuffer_transform_flags; }; diff --git a/src/video_core/textures/decoders.cpp b/src/video_core/textures/decoders.cpp new file mode 100644 index 000000000..2e87281eb --- /dev/null +++ b/src/video_core/textures/decoders.cpp @@ -0,0 +1,105 @@ +// Copyright 2018 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <cstring> +#include "common/assert.h" +#include "video_core/textures/decoders.h" +#include "video_core/textures/texture.h" + +namespace Tegra { +namespace Texture { + +/** + * Calculates the offset of an (x, y) position within a swizzled texture. + * Taken from the Tegra X1 TRM. + */ +static u32 GetSwizzleOffset(u32 x, u32 y, u32 image_width, u32 bytes_per_pixel, u32 block_height) { + u32 image_width_in_gobs = image_width * bytes_per_pixel / 64; + u32 GOB_address = 0 + (y / (8 * block_height)) * 512 * block_height * image_width_in_gobs + + (x * bytes_per_pixel / 64) * 512 * block_height + + (y % (8 * block_height) / 8) * 512; + x *= bytes_per_pixel; + u32 address = GOB_address + ((x % 64) / 32) * 256 + ((y % 8) / 2) * 64 + ((x % 32) / 16) * 32 + + (y % 2) * 16 + (x % 16); + + return address; +} + +static void CopySwizzledData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel, + u8* swizzled_data, u8* unswizzled_data, bool unswizzle, + u32 block_height) { + u8* data_ptrs[2]; + for (unsigned y = 0; y < height; ++y) { + for (unsigned x = 0; x < width; ++x) { + u32 swizzle_offset = GetSwizzleOffset(x, y, width, bytes_per_pixel, block_height); + u32 pixel_index = (x + y * width) * out_bytes_per_pixel; + + data_ptrs[unswizzle] = swizzled_data + swizzle_offset; + data_ptrs[!unswizzle] = &unswizzled_data[pixel_index]; + + std::memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel); + } + } +} + +u32 BytesPerPixel(TextureFormat format) { + switch (format) { + case TextureFormat::DXT1: + // In this case a 'pixel' actually refers to a 4x4 tile. + return 8; + case TextureFormat::A8R8G8B8: + return 4; + default: + UNIMPLEMENTED_MSG("Format not implemented"); + break; + } +} + +std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width, u32 height) { + u8* data = Memory::GetPointer(address); + u32 bytes_per_pixel = BytesPerPixel(format); + + static constexpr u32 DefaultBlockHeight = 16; + + std::vector<u8> unswizzled_data(width * height * bytes_per_pixel); + + switch (format) { + case TextureFormat::DXT1: + // In the DXT1 format, each 4x4 tile is swizzled instead of just individual pixel values. + CopySwizzledData(width / 4, height / 4, bytes_per_pixel, bytes_per_pixel, data, + unswizzled_data.data(), true, DefaultBlockHeight); + break; + case TextureFormat::A8R8G8B8: + CopySwizzledData(width, height, bytes_per_pixel, bytes_per_pixel, data, + unswizzled_data.data(), true, DefaultBlockHeight); + break; + default: + UNIMPLEMENTED_MSG("Format not implemented"); + break; + } + + return unswizzled_data; +} + +std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat format, u32 width, + u32 height) { + std::vector<u8> rgba_data; + + // TODO(Subv): Implement. + switch (format) { + case TextureFormat::DXT1: + case TextureFormat::A8R8G8B8: + // TODO(Subv): For the time being just forward the same data without any decoding. + rgba_data = texture_data; + break; + default: + UNIMPLEMENTED_MSG("Format not implemented"); + break; + } + + return rgba_data; +} + +} // namespace Texture +} // namespace Tegra diff --git a/src/video_core/textures/decoders.h b/src/video_core/textures/decoders.h new file mode 100644 index 000000000..0c21694ff --- /dev/null +++ b/src/video_core/textures/decoders.h @@ -0,0 +1,26 @@ +// Copyright 2018 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <vector> +#include "common/common_types.h" +#include "video_core/textures/texture.h" + +namespace Tegra { +namespace Texture { + +/** + * Unswizzles a swizzled texture without changing its format. + */ +std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width, u32 height); + +/** + * Decodes an unswizzled texture into a A8R8G8B8 texture. + */ +std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat format, u32 width, + u32 height); + +} // namespace Texture +} // namespace Tegra diff --git a/src/video_core/textures/texture.h b/src/video_core/textures/texture.h new file mode 100644 index 000000000..07936f8a3 --- /dev/null +++ b/src/video_core/textures/texture.h @@ -0,0 +1,137 @@ +// Copyright 2018 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "common/bit_field.h" +#include "common/common_funcs.h" +#include "common/common_types.h" +#include "video_core/memory_manager.h" + +namespace Tegra { +namespace Texture { + +enum class TextureFormat : u32 { + A8R8G8B8 = 8, + DXT1 = 0x24, +}; + +enum class TextureType : u32 { + Texture1D = 0, + Texture2D = 1, + Texture3D = 2, + TextureCubemap = 3, + Texture1DArray = 4, + Texture2DArray = 5, + Texture1DBuffer = 6, + Texture2DNoMipmap = 7, + TextureCubeArray = 8, +}; + +enum class TICHeaderVersion : u32 { + OneDBuffer = 0, + PitchColorKey = 1, + Pitch = 2, + BlockLinear = 3, + BlockLinearColorKey = 4, +}; + +union TextureHandle { + u32 raw; + BitField<0, 20, u32> tic_id; + BitField<20, 12, u32> tsc_id; +}; +static_assert(sizeof(TextureHandle) == 4, "TextureHandle has wrong size"); + +struct TICEntry { + union { + u32 raw; + BitField<0, 7, TextureFormat> format; + BitField<7, 3, u32> r_type; + BitField<10, 3, u32> g_type; + BitField<13, 3, u32> b_type; + BitField<16, 3, u32> a_type; + }; + u32 address_low; + union { + BitField<0, 16, u32> address_high; + BitField<21, 3, TICHeaderVersion> header_version; + }; + INSERT_PADDING_BYTES(4); + union { + BitField<0, 16, u32> width_minus_1; + BitField<23, 4, TextureType> texture_type; + }; + u16 height_minus_1; + INSERT_PADDING_BYTES(10); + + GPUVAddr Address() const { + return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) | address_low); + } + + u32 Width() const { + return width_minus_1 + 1; + } + + u32 Height() const { + return height_minus_1 + 1; + } +}; +static_assert(sizeof(TICEntry) == 0x20, "TICEntry has wrong size"); + +enum class WrapMode : u32 { + Wrap = 0, + Mirror = 1, + ClampToEdge = 2, + Border = 3, + ClampOGL = 4, + MirrorOnceClampToEdge = 5, + MirrorOnceBorder = 6, + MirrorOnceClampOGL = 7, +}; + +enum class TextureFilter : u32 { + Nearest = 1, + Linear = 2, +}; + +enum class TextureMipmapFilter : u32 { + None = 1, + Nearest = 2, + Linear = 3, +}; + +struct TSCEntry { + union { + BitField<0, 3, WrapMode> wrap_u; + BitField<3, 3, WrapMode> wrap_v; + BitField<6, 3, WrapMode> wrap_p; + BitField<9, 1, u32> depth_compare_enabled; + BitField<10, 3, u32> depth_compare_func; + }; + union { + BitField<0, 2, TextureFilter> mag_filter; + BitField<4, 2, TextureFilter> min_filter; + BitField<6, 2, TextureMipmapFilter> mip_filter; + }; + INSERT_PADDING_BYTES(8); + u32 border_color_r; + u32 border_color_g; + u32 border_color_b; + u32 border_color_a; +}; +static_assert(sizeof(TSCEntry) == 0x20, "TSCEntry has wrong size"); + +struct FullTextureInfo { + u32 index; + TICEntry tic; + TSCEntry tsc; + bool enabled; +}; + +/// Returns the number of bytes per pixel of the input texture format. +u32 BytesPerPixel(TextureFormat format); + +} // namespace Texture +} // namespace Tegra diff --git a/src/video_core/utils.h b/src/video_core/utils.h index d94a10417..be0f7e22b 100644 --- a/src/video_core/utils.h +++ b/src/video_core/utils.h @@ -49,4 +49,116 @@ static inline u32 GetMortonOffset(u32 x, u32 y, u32 bytes_per_pixel) { return (i + offset) * bytes_per_pixel; } +static inline u32 MortonInterleave128(u32 x, u32 y) { + // 128x128 Z-Order coordinate from 2D coordinates + static constexpr u32 xlut[] = { + 0x0000, 0x0001, 0x0002, 0x0003, 0x0008, 0x0009, 0x000a, 0x000b, 0x0040, 0x0041, 0x0042, + 0x0043, 0x0048, 0x0049, 0x004a, 0x004b, 0x0800, 0x0801, 0x0802, 0x0803, 0x0808, 0x0809, + 0x080a, 0x080b, 0x0840, 0x0841, 0x0842, 0x0843, 0x0848, 0x0849, 0x084a, 0x084b, 0x1000, + 0x1001, 0x1002, 0x1003, 0x1008, 0x1009, 0x100a, 0x100b, 0x1040, 0x1041, 0x1042, 0x1043, + 0x1048, 0x1049, 0x104a, 0x104b, 0x1800, 0x1801, 0x1802, 0x1803, 0x1808, 0x1809, 0x180a, + 0x180b, 0x1840, 0x1841, 0x1842, 0x1843, 0x1848, 0x1849, 0x184a, 0x184b, 0x2000, 0x2001, + 0x2002, 0x2003, 0x2008, 0x2009, 0x200a, 0x200b, 0x2040, 0x2041, 0x2042, 0x2043, 0x2048, + 0x2049, 0x204a, 0x204b, 0x2800, 0x2801, 0x2802, 0x2803, 0x2808, 0x2809, 0x280a, 0x280b, + 0x2840, 0x2841, 0x2842, 0x2843, 0x2848, 0x2849, 0x284a, 0x284b, 0x3000, 0x3001, 0x3002, + 0x3003, 0x3008, 0x3009, 0x300a, 0x300b, 0x3040, 0x3041, 0x3042, 0x3043, 0x3048, 0x3049, + 0x304a, 0x304b, 0x3800, 0x3801, 0x3802, 0x3803, 0x3808, 0x3809, 0x380a, 0x380b, 0x3840, + 0x3841, 0x3842, 0x3843, 0x3848, 0x3849, 0x384a, 0x384b, 0x0000, 0x0001, 0x0002, 0x0003, + 0x0008, 0x0009, 0x000a, 0x000b, 0x0040, 0x0041, 0x0042, 0x0043, 0x0048, 0x0049, 0x004a, + 0x004b, 0x0800, 0x0801, 0x0802, 0x0803, 0x0808, 0x0809, 0x080a, 0x080b, 0x0840, 0x0841, + 0x0842, 0x0843, 0x0848, 0x0849, 0x084a, 0x084b, 0x1000, 0x1001, 0x1002, 0x1003, 0x1008, + 0x1009, 0x100a, 0x100b, 0x1040, 0x1041, 0x1042, 0x1043, 0x1048, 0x1049, 0x104a, 0x104b, + 0x1800, 0x1801, 0x1802, 0x1803, 0x1808, 0x1809, 0x180a, 0x180b, 0x1840, 0x1841, 0x1842, + 0x1843, 0x1848, 0x1849, 0x184a, 0x184b, 0x2000, 0x2001, 0x2002, 0x2003, 0x2008, 0x2009, + 0x200a, 0x200b, 0x2040, 0x2041, 0x2042, 0x2043, 0x2048, 0x2049, 0x204a, 0x204b, 0x2800, + 0x2801, 0x2802, 0x2803, 0x2808, 0x2809, 0x280a, 0x280b, 0x2840, 0x2841, 0x2842, 0x2843, + 0x2848, 0x2849, 0x284a, 0x284b, 0x3000, 0x3001, 0x3002, 0x3003, 0x3008, 0x3009, 0x300a, + 0x300b, 0x3040, 0x3041, 0x3042, 0x3043, 0x3048, 0x3049, 0x304a, 0x304b, 0x3800, 0x3801, + 0x3802, 0x3803, 0x3808, 0x3809, 0x380a, 0x380b, 0x3840, 0x3841, 0x3842, 0x3843, 0x3848, + 0x3849, 0x384a, 0x384b, 0x0000, 0x0001, 0x0002, 0x0003, 0x0008, 0x0009, 0x000a, 0x000b, + 0x0040, 0x0041, 0x0042, 0x0043, 0x0048, 0x0049, 0x004a, 0x004b, 0x0800, 0x0801, 0x0802, + 0x0803, 0x0808, 0x0809, 0x080a, 0x080b, 0x0840, 0x0841, 0x0842, 0x0843, 0x0848, 0x0849, + 0x084a, 0x084b, 0x1000, 0x1001, 0x1002, 0x1003, 0x1008, 0x1009, 0x100a, 0x100b, 0x1040, + 0x1041, 0x1042, 0x1043, 0x1048, 0x1049, 0x104a, 0x104b, 0x1800, 0x1801, 0x1802, 0x1803, + 0x1808, 0x1809, 0x180a, 0x180b, 0x1840, 0x1841, 0x1842, 0x1843, 0x1848, 0x1849, 0x184a, + 0x184b, 0x2000, 0x2001, 0x2002, 0x2003, 0x2008, 0x2009, 0x200a, 0x200b, 0x2040, 0x2041, + 0x2042, 0x2043, 0x2048, 0x2049, 0x204a, 0x204b, 0x2800, 0x2801, 0x2802, 0x2803, 0x2808, + 0x2809, 0x280a, 0x280b, 0x2840, 0x2841, 0x2842, 0x2843, 0x2848, 0x2849, 0x284a, 0x284b, + 0x3000, 0x3001, 0x3002, 0x3003, 0x3008, 0x3009, 0x300a, 0x300b, 0x3040, 0x3041, 0x3042, + 0x3043, 0x3048, 0x3049, 0x304a, 0x304b, 0x3800, 0x3801, 0x3802, 0x3803, 0x3808, 0x3809, + 0x380a, 0x380b, 0x3840, 0x3841, 0x3842, 0x3843, 0x3848, 0x3849, 0x384a, 0x384b, + }; + static constexpr u32 ylut[] = { + 0x0000, 0x0004, 0x0010, 0x0014, 0x0020, 0x0024, 0x0030, 0x0034, 0x0080, 0x0084, 0x0090, + 0x0094, 0x00a0, 0x00a4, 0x00b0, 0x00b4, 0x0100, 0x0104, 0x0110, 0x0114, 0x0120, 0x0124, + 0x0130, 0x0134, 0x0180, 0x0184, 0x0190, 0x0194, 0x01a0, 0x01a4, 0x01b0, 0x01b4, 0x0200, + 0x0204, 0x0210, 0x0214, 0x0220, 0x0224, 0x0230, 0x0234, 0x0280, 0x0284, 0x0290, 0x0294, + 0x02a0, 0x02a4, 0x02b0, 0x02b4, 0x0300, 0x0304, 0x0310, 0x0314, 0x0320, 0x0324, 0x0330, + 0x0334, 0x0380, 0x0384, 0x0390, 0x0394, 0x03a0, 0x03a4, 0x03b0, 0x03b4, 0x0400, 0x0404, + 0x0410, 0x0414, 0x0420, 0x0424, 0x0430, 0x0434, 0x0480, 0x0484, 0x0490, 0x0494, 0x04a0, + 0x04a4, 0x04b0, 0x04b4, 0x0500, 0x0504, 0x0510, 0x0514, 0x0520, 0x0524, 0x0530, 0x0534, + 0x0580, 0x0584, 0x0590, 0x0594, 0x05a0, 0x05a4, 0x05b0, 0x05b4, 0x0600, 0x0604, 0x0610, + 0x0614, 0x0620, 0x0624, 0x0630, 0x0634, 0x0680, 0x0684, 0x0690, 0x0694, 0x06a0, 0x06a4, + 0x06b0, 0x06b4, 0x0700, 0x0704, 0x0710, 0x0714, 0x0720, 0x0724, 0x0730, 0x0734, 0x0780, + 0x0784, 0x0790, 0x0794, 0x07a0, 0x07a4, 0x07b0, 0x07b4, 0x0000, 0x0004, 0x0010, 0x0014, + 0x0020, 0x0024, 0x0030, 0x0034, 0x0080, 0x0084, 0x0090, 0x0094, 0x00a0, 0x00a4, 0x00b0, + 0x00b4, 0x0100, 0x0104, 0x0110, 0x0114, 0x0120, 0x0124, 0x0130, 0x0134, 0x0180, 0x0184, + 0x0190, 0x0194, 0x01a0, 0x01a4, 0x01b0, 0x01b4, 0x0200, 0x0204, 0x0210, 0x0214, 0x0220, + 0x0224, 0x0230, 0x0234, 0x0280, 0x0284, 0x0290, 0x0294, 0x02a0, 0x02a4, 0x02b0, 0x02b4, + 0x0300, 0x0304, 0x0310, 0x0314, 0x0320, 0x0324, 0x0330, 0x0334, 0x0380, 0x0384, 0x0390, + 0x0394, 0x03a0, 0x03a4, 0x03b0, 0x03b4, 0x0400, 0x0404, 0x0410, 0x0414, 0x0420, 0x0424, + 0x0430, 0x0434, 0x0480, 0x0484, 0x0490, 0x0494, 0x04a0, 0x04a4, 0x04b0, 0x04b4, 0x0500, + 0x0504, 0x0510, 0x0514, 0x0520, 0x0524, 0x0530, 0x0534, 0x0580, 0x0584, 0x0590, 0x0594, + 0x05a0, 0x05a4, 0x05b0, 0x05b4, 0x0600, 0x0604, 0x0610, 0x0614, 0x0620, 0x0624, 0x0630, + 0x0634, 0x0680, 0x0684, 0x0690, 0x0694, 0x06a0, 0x06a4, 0x06b0, 0x06b4, 0x0700, 0x0704, + 0x0710, 0x0714, 0x0720, 0x0724, 0x0730, 0x0734, 0x0780, 0x0784, 0x0790, 0x0794, 0x07a0, + 0x07a4, 0x07b0, 0x07b4, 0x0000, 0x0004, 0x0010, 0x0014, 0x0020, 0x0024, 0x0030, 0x0034, + 0x0080, 0x0084, 0x0090, 0x0094, 0x00a0, 0x00a4, 0x00b0, 0x00b4, 0x0100, 0x0104, 0x0110, + 0x0114, 0x0120, 0x0124, 0x0130, 0x0134, 0x0180, 0x0184, 0x0190, 0x0194, 0x01a0, 0x01a4, + 0x01b0, 0x01b4, 0x0200, 0x0204, 0x0210, 0x0214, 0x0220, 0x0224, 0x0230, 0x0234, 0x0280, + 0x0284, 0x0290, 0x0294, 0x02a0, 0x02a4, 0x02b0, 0x02b4, 0x0300, 0x0304, 0x0310, 0x0314, + 0x0320, 0x0324, 0x0330, 0x0334, 0x0380, 0x0384, 0x0390, 0x0394, 0x03a0, 0x03a4, 0x03b0, + 0x03b4, 0x0400, 0x0404, 0x0410, 0x0414, 0x0420, 0x0424, 0x0430, 0x0434, 0x0480, 0x0484, + 0x0490, 0x0494, 0x04a0, 0x04a4, 0x04b0, 0x04b4, 0x0500, 0x0504, 0x0510, 0x0514, 0x0520, + 0x0524, 0x0530, 0x0534, 0x0580, 0x0584, 0x0590, 0x0594, 0x05a0, 0x05a4, 0x05b0, 0x05b4, + 0x0600, 0x0604, 0x0610, 0x0614, 0x0620, 0x0624, 0x0630, 0x0634, 0x0680, 0x0684, 0x0690, + 0x0694, 0x06a0, 0x06a4, 0x06b0, 0x06b4, 0x0700, 0x0704, 0x0710, 0x0714, 0x0720, 0x0724, + 0x0730, 0x0734, 0x0780, 0x0784, 0x0790, 0x0794, 0x07a0, 0x07a4, 0x07b0, 0x07b4, + }; + return xlut[x % 128] + ylut[y % 128]; +} + +static inline u32 GetMortonOffset128(u32 x, u32 y, u32 bytes_per_pixel) { + // Calculates the offset of the position of the pixel in Morton order + // Framebuffer images are split into 128x128 tiles. + + const unsigned int block_height = 128; + const unsigned int coarse_x = x & ~127; + + u32 i = MortonInterleave128(x, y); + + const unsigned int offset = coarse_x * block_height; + + return (i + offset) * bytes_per_pixel; +} + +static inline void MortonCopyPixels128(u32 width, u32 height, u32 bytes_per_pixel, + u32 gl_bytes_per_pixel, u8* morton_data, u8* gl_data, + bool morton_to_gl) { + u8* data_ptrs[2]; + for (unsigned y = 0; y < height; ++y) { + for (unsigned x = 0; x < width; ++x) { + const u32 coarse_y = y & ~127; + u32 morton_offset = + GetMortonOffset128(x, y, bytes_per_pixel) + coarse_y * width * bytes_per_pixel; + u32 gl_pixel_index = (x + (height - 1 - y) * width) * gl_bytes_per_pixel; + + data_ptrs[morton_to_gl] = morton_data + morton_offset; + data_ptrs[!morton_to_gl] = &gl_data[gl_pixel_index]; + + memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel); + } + } +} + } // namespace VideoCore diff --git a/src/video_core/video_core.cpp b/src/video_core/video_core.cpp index 864691baa..289140f31 100644 --- a/src/video_core/video_core.cpp +++ b/src/video_core/video_core.cpp @@ -26,7 +26,7 @@ bool Init(EmuWindow* emu_window) { if (g_renderer->Init()) { LOG_DEBUG(Render, "initialized OK"); } else { - LOG_ERROR(Render, "initialization failed !"); + LOG_CRITICAL(Render, "initialization failed !"); return false; } return true; diff --git a/src/video_core/video_core.h b/src/video_core/video_core.h index 1fd90b9d0..37da62436 100644 --- a/src/video_core/video_core.h +++ b/src/video_core/video_core.h @@ -15,6 +15,8 @@ class RendererBase; namespace VideoCore { +enum class Renderer { Software, OpenGL }; + extern std::unique_ptr<RendererBase> g_renderer; ///< Renderer plugin extern EmuWindow* g_emu_window; ///< Emu window diff --git a/src/yuzu/CMakeLists.txt b/src/yuzu/CMakeLists.txt index 0c4056c49..5af3154d7 100644 --- a/src/yuzu/CMakeLists.txt +++ b/src/yuzu/CMakeLists.txt @@ -23,6 +23,13 @@ add_executable(yuzu configuration/configure_input.h configuration/configure_system.cpp configuration/configure_system.h + debugger/graphics/graphics_breakpoint_observer.cpp + debugger/graphics/graphics_breakpoint_observer.h + debugger/graphics/graphics_breakpoints.cpp + debugger/graphics/graphics_breakpoints.h + debugger/graphics/graphics_breakpoints_p.h + debugger/graphics/graphics_surface.cpp + debugger/graphics/graphics_surface.h debugger/profiler.cpp debugger/profiler.h debugger/registers.cpp diff --git a/src/yuzu/about_dialog.cpp b/src/yuzu/about_dialog.cpp index da3429822..d6647eeea 100644 --- a/src/yuzu/about_dialog.cpp +++ b/src/yuzu/about_dialog.cpp @@ -2,12 +2,14 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <QIcon> #include "common/scm_rev.h" #include "ui_aboutdialog.h" #include "yuzu/about_dialog.h" AboutDialog::AboutDialog(QWidget* parent) : QDialog(parent), ui(new Ui::AboutDialog) { ui->setupUi(this); + ui->labelLogo->setPixmap(QIcon::fromTheme("yuzu").pixmap(200)); ui->labelBuildInfo->setText(ui->labelBuildInfo->text().arg( Common::g_build_name, Common::g_scm_branch, Common::g_scm_desc)); } diff --git a/src/yuzu/configuration/config.cpp b/src/yuzu/configuration/config.cpp index f9ddb9edc..8843f2078 100644 --- a/src/yuzu/configuration/config.cpp +++ b/src/yuzu/configuration/config.cpp @@ -77,8 +77,7 @@ void Config::ReadValues() { qt_config->endGroup(); qt_config->beginGroup("Core"); - Settings::values.cpu_core = - static_cast<Settings::CpuCore>(qt_config->value("cpu_core", 0).toInt()); + Settings::values.use_cpu_jit = qt_config->value("use_cpu_jit", true).toBool(); qt_config->endGroup(); qt_config->beginGroup("Renderer"); @@ -94,6 +93,10 @@ void Config::ReadValues() { Settings::values.use_virtual_sd = qt_config->value("use_virtual_sd", true).toBool(); qt_config->endGroup(); + qt_config->beginGroup("System"); + Settings::values.use_docked_mode = qt_config->value("use_docked_mode", true).toBool(); + qt_config->endGroup(); + qt_config->beginGroup("Miscellaneous"); Settings::values.log_filter = qt_config->value("log_filter", "*:Info").toString().toStdString(); qt_config->endGroup(); @@ -104,6 +107,8 @@ void Config::ReadValues() { qt_config->endGroup(); qt_config->beginGroup("UI"); + UISettings::values.theme = qt_config->value("theme", UISettings::themes[0].second).toString(); + qt_config->beginGroup("UILayout"); UISettings::values.geometry = qt_config->value("geometry").toByteArray(); UISettings::values.state = qt_config->value("state").toByteArray(); @@ -171,7 +176,7 @@ void Config::SaveValues() { qt_config->endGroup(); qt_config->beginGroup("Core"); - qt_config->setValue("cpu_core", static_cast<int>(Settings::values.cpu_core)); + qt_config->setValue("use_cpu_jit", Settings::values.use_cpu_jit); qt_config->endGroup(); qt_config->beginGroup("Renderer"); @@ -188,6 +193,10 @@ void Config::SaveValues() { qt_config->setValue("use_virtual_sd", Settings::values.use_virtual_sd); qt_config->endGroup(); + qt_config->beginGroup("System"); + qt_config->setValue("use_docked_mode", Settings::values.use_docked_mode); + qt_config->endGroup(); + qt_config->beginGroup("Miscellaneous"); qt_config->setValue("log_filter", QString::fromStdString(Settings::values.log_filter)); qt_config->endGroup(); diff --git a/src/yuzu/configuration/configure_general.cpp b/src/yuzu/configuration/configure_general.cpp index 92fd6ab02..2d73fc5aa 100644 --- a/src/yuzu/configuration/configure_general.cpp +++ b/src/yuzu/configuration/configure_general.cpp @@ -13,9 +13,14 @@ ConfigureGeneral::ConfigureGeneral(QWidget* parent) ui->setupUi(this); + for (auto theme : UISettings::themes) { + ui->theme_combobox->addItem(theme.first, theme.second); + } + this->setConfiguration(); - ui->cpu_core_combobox->setEnabled(!Core::System::GetInstance().IsPoweredOn()); + ui->use_cpu_jit->setEnabled(!Core::System::GetInstance().IsPoweredOn()); + ui->use_docked_mode->setEnabled(!Core::System::GetInstance().IsPoweredOn()); } ConfigureGeneral::~ConfigureGeneral() {} @@ -23,13 +28,18 @@ ConfigureGeneral::~ConfigureGeneral() {} void ConfigureGeneral::setConfiguration() { ui->toggle_deepscan->setChecked(UISettings::values.gamedir_deepscan); ui->toggle_check_exit->setChecked(UISettings::values.confirm_before_closing); - ui->cpu_core_combobox->setCurrentIndex(static_cast<int>(Settings::values.cpu_core)); + ui->theme_combobox->setCurrentIndex(ui->theme_combobox->findData(UISettings::values.theme)); + ui->use_cpu_jit->setChecked(Settings::values.use_cpu_jit); + ui->use_docked_mode->setChecked(Settings::values.use_docked_mode); } void ConfigureGeneral::applyConfiguration() { UISettings::values.gamedir_deepscan = ui->toggle_deepscan->isChecked(); UISettings::values.confirm_before_closing = ui->toggle_check_exit->isChecked(); - Settings::values.cpu_core = - static_cast<Settings::CpuCore>(ui->cpu_core_combobox->currentIndex()); + UISettings::values.theme = + ui->theme_combobox->itemData(ui->theme_combobox->currentIndex()).toString(); + + Settings::values.use_cpu_jit = ui->use_cpu_jit->isChecked(); + Settings::values.use_docked_mode = ui->use_docked_mode->isChecked(); Settings::Apply(); } diff --git a/src/yuzu/configuration/configure_general.ui b/src/yuzu/configuration/configure_general.ui index 573c4cb0e..1775c4d40 100644 --- a/src/yuzu/configuration/configure_general.ui +++ b/src/yuzu/configuration/configure_general.ui @@ -13,17 +13,17 @@ <property name="windowTitle"> <string>Form</string> </property> - <layout class="QHBoxLayout" name="horizontalLayout"> + <layout class="QHBoxLayout" name="HorizontalLayout"> <item> - <layout class="QVBoxLayout" name="verticalLayout"> + <layout class="QVBoxLayout" name="VerticalLayout"> <item> - <widget class="QGroupBox" name="groupBox"> + <widget class="QGroupBox" name="GeneralGroupBox"> <property name="title"> <string>General</string> </property> - <layout class="QHBoxLayout" name="horizontalLayout_3"> + <layout class="QHBoxLayout" name="GeneralHorizontalLayout"> <item> - <layout class="QVBoxLayout" name="verticalLayout_2"> + <layout class="QVBoxLayout" name="GeneralVerticalLayout"> <item> <widget class="QCheckBox" name="toggle_deepscan"> <property name="text"> @@ -44,40 +44,80 @@ </widget> </item> <item> - <widget class="QGroupBox" name="groupBox_2"> - <property name="title"> - <string>CPU Core</string> - </property> - <layout class="QHBoxLayout" name="horizontalLayout_7"> - <item> - <layout class="QVBoxLayout" name="verticalLayout_5"> - <item> - <widget class="QComboBox" name="cpu_core_combobox"> - <item> - <property name="text"> - <string>Unicorn</string> - </property> - </item> - <item> - <property name="text"> - <string>Dynarmic</string> - </property> - </item> - </widget> - </item> - </layout> - </item> + <widget class="QGroupBox" name="PerformanceGroupBox"> + <property name="title"> + <string>Performance</string> + </property> + <layout class="QHBoxLayout" name="PerformanceHorizontalLayout"> + <item> + <layout class="QVBoxLayout" name="PerformanceVerticalLayout"> + <item> + <widget class="QCheckBox" name="use_cpu_jit"> + <property name="text"> + <string>Enable CPU JIT</string> + </property> + </widget> + </item> </layout> - </widget> + </item> + </layout> + </widget> + </item> + <item> + <widget class="QGroupBox" name="EmulationGroupBox"> + <property name="title"> + <string>Emulation</string> + </property> + <layout class="QHBoxLayout" name="EmulationHorizontalLayout"> + <item> + <layout class="QVBoxLayout" name="EmulationVerticalLayout"> + <item> + <widget class="QCheckBox" name="use_docked_mode"> + <property name="text"> + <string>Enable docked mode</string> + </property> + </widget> + </item> + </layout> + </item> + </layout> + </widget> + </item> + <item> + <widget class="QGroupBox" name="theme_group_box"> + <property name="title"> + <string>Theme</string> + </property> + <layout class="QHBoxLayout" name="theme_qhbox_layout"> + <item> + <layout class="QVBoxLayout" name="theme_qvbox_layout"> + <item> + <layout class="QHBoxLayout" name="theme_qhbox_layout_2"> + <item> + <widget class="QLabel" name="theme_label"> + <property name="text"> + <string>Theme:</string> + </property> + </widget> + </item> + <item> + <widget class="QComboBox" name="theme_combobox"/> + </item> + </layout> + </item> + </layout> + </item> + </layout> + </widget> </item> <item> - <widget class="QGroupBox" name="groupBox_3"> + <widget class="QGroupBox" name="HotKeysGroupBox"> <property name="title"> <string>Hotkeys</string> </property> - <layout class="QHBoxLayout" name="horizontalLayout_4"> + <layout class="QHBoxLayout" name="HotKeysHorizontalLayout"> <item> - <layout class="QVBoxLayout" name="verticalLayout_4"> + <layout class="QVBoxLayout" name="HotKeysVerticalLayout"> <item> <widget class="GHotkeysDialog" name="widget" native="true"/> </item> diff --git a/src/yuzu/debugger/graphics/graphics_breakpoint_observer.cpp b/src/yuzu/debugger/graphics/graphics_breakpoint_observer.cpp new file mode 100644 index 000000000..d6d61a739 --- /dev/null +++ b/src/yuzu/debugger/graphics/graphics_breakpoint_observer.cpp @@ -0,0 +1,27 @@ +// Copyright 2014 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <QMetaType> +#include "yuzu/debugger/graphics/graphics_breakpoint_observer.h" + +BreakPointObserverDock::BreakPointObserverDock(std::shared_ptr<Tegra::DebugContext> debug_context, + const QString& title, QWidget* parent) + : QDockWidget(title, parent), BreakPointObserver(debug_context) { + qRegisterMetaType<Tegra::DebugContext::Event>("Tegra::DebugContext::Event"); + + connect(this, SIGNAL(Resumed()), this, SLOT(OnResumed())); + + // NOTE: This signal is emitted from a non-GUI thread, but connect() takes + // care of delaying its handling to the GUI thread. + connect(this, SIGNAL(BreakPointHit(Tegra::DebugContext::Event, void*)), this, + SLOT(OnBreakPointHit(Tegra::DebugContext::Event, void*)), Qt::BlockingQueuedConnection); +} + +void BreakPointObserverDock::OnMaxwellBreakPointHit(Tegra::DebugContext::Event event, void* data) { + emit BreakPointHit(event, data); +} + +void BreakPointObserverDock::OnMaxwellResume() { + emit Resumed(); +} diff --git a/src/yuzu/debugger/graphics/graphics_breakpoint_observer.h b/src/yuzu/debugger/graphics/graphics_breakpoint_observer.h new file mode 100644 index 000000000..9d05493cf --- /dev/null +++ b/src/yuzu/debugger/graphics/graphics_breakpoint_observer.h @@ -0,0 +1,33 @@ +// Copyright 2014 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <QDockWidget> +#include "video_core/debug_utils/debug_utils.h" + +/** + * Utility class which forwards calls to OnMaxwellBreakPointHit and OnMaxwellResume to public slots. + * This is because the Maxwell breakpoint callbacks are called from a non-GUI thread, while + * the widget usually wants to perform reactions in the GUI thread. + */ +class BreakPointObserverDock : public QDockWidget, + protected Tegra::DebugContext::BreakPointObserver { + Q_OBJECT + +public: + BreakPointObserverDock(std::shared_ptr<Tegra::DebugContext> debug_context, const QString& title, + QWidget* parent = nullptr); + + void OnMaxwellBreakPointHit(Tegra::DebugContext::Event event, void* data) override; + void OnMaxwellResume() override; + +private slots: + virtual void OnBreakPointHit(Tegra::DebugContext::Event event, void* data) = 0; + virtual void OnResumed() = 0; + +signals: + void Resumed(); + void BreakPointHit(Tegra::DebugContext::Event event, void* data); +}; diff --git a/src/yuzu/debugger/graphics/graphics_breakpoints.cpp b/src/yuzu/debugger/graphics/graphics_breakpoints.cpp new file mode 100644 index 000000000..f98cc8152 --- /dev/null +++ b/src/yuzu/debugger/graphics/graphics_breakpoints.cpp @@ -0,0 +1,212 @@ +// Copyright 2014 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <QLabel> +#include <QMetaType> +#include <QPushButton> +#include <QTreeView> +#include <QVBoxLayout> +#include "common/assert.h" +#include "yuzu/debugger/graphics/graphics_breakpoints.h" +#include "yuzu/debugger/graphics/graphics_breakpoints_p.h" + +BreakPointModel::BreakPointModel(std::shared_ptr<Tegra::DebugContext> debug_context, + QObject* parent) + : QAbstractListModel(parent), context_weak(debug_context), + at_breakpoint(debug_context->at_breakpoint), + active_breakpoint(debug_context->active_breakpoint) {} + +int BreakPointModel::columnCount(const QModelIndex& parent) const { + return 1; +} + +int BreakPointModel::rowCount(const QModelIndex& parent) const { + return static_cast<int>(Tegra::DebugContext::Event::NumEvents); +} + +QVariant BreakPointModel::data(const QModelIndex& index, int role) const { + const auto event = static_cast<Tegra::DebugContext::Event>(index.row()); + + switch (role) { + case Qt::DisplayRole: { + if (index.column() == 0) { + static const std::map<Tegra::DebugContext::Event, QString> map = { + {Tegra::DebugContext::Event::MaxwellCommandLoaded, tr("Maxwell command loaded")}, + {Tegra::DebugContext::Event::MaxwellCommandProcessed, + tr("Maxwell command processed")}, + {Tegra::DebugContext::Event::IncomingPrimitiveBatch, + tr("Incoming primitive batch")}, + {Tegra::DebugContext::Event::FinishedPrimitiveBatch, + tr("Finished primitive batch")}, + }; + + DEBUG_ASSERT(map.size() == static_cast<size_t>(Tegra::DebugContext::Event::NumEvents)); + return (map.find(event) != map.end()) ? map.at(event) : QString(); + } + + break; + } + + case Qt::CheckStateRole: { + if (index.column() == 0) + return data(index, Role_IsEnabled).toBool() ? Qt::Checked : Qt::Unchecked; + break; + } + + case Qt::BackgroundRole: { + if (at_breakpoint && index.row() == static_cast<int>(active_breakpoint)) { + return QBrush(QColor(0xE0, 0xE0, 0x10)); + } + break; + } + + case Role_IsEnabled: { + auto context = context_weak.lock(); + return context && context->breakpoints[(int)event].enabled; + } + + default: + break; + } + return QVariant(); +} + +Qt::ItemFlags BreakPointModel::flags(const QModelIndex& index) const { + if (!index.isValid()) + return 0; + + Qt::ItemFlags flags = Qt::ItemIsEnabled; + if (index.column() == 0) + flags |= Qt::ItemIsUserCheckable; + return flags; +} + +bool BreakPointModel::setData(const QModelIndex& index, const QVariant& value, int role) { + const auto event = static_cast<Tegra::DebugContext::Event>(index.row()); + + switch (role) { + case Qt::CheckStateRole: { + if (index.column() != 0) + return false; + + auto context = context_weak.lock(); + if (!context) + return false; + + context->breakpoints[(int)event].enabled = value == Qt::Checked; + QModelIndex changed_index = createIndex(index.row(), 0); + emit dataChanged(changed_index, changed_index); + return true; + } + } + + return false; +} + +void BreakPointModel::OnBreakPointHit(Tegra::DebugContext::Event event) { + auto context = context_weak.lock(); + if (!context) + return; + + active_breakpoint = context->active_breakpoint; + at_breakpoint = context->at_breakpoint; + emit dataChanged(createIndex(static_cast<int>(event), 0), + createIndex(static_cast<int>(event), 0)); +} + +void BreakPointModel::OnResumed() { + auto context = context_weak.lock(); + if (!context) + return; + + at_breakpoint = context->at_breakpoint; + emit dataChanged(createIndex(static_cast<int>(active_breakpoint), 0), + createIndex(static_cast<int>(active_breakpoint), 0)); + active_breakpoint = context->active_breakpoint; +} + +GraphicsBreakPointsWidget::GraphicsBreakPointsWidget( + std::shared_ptr<Tegra::DebugContext> debug_context, QWidget* parent) + : QDockWidget(tr("Maxwell Breakpoints"), parent), Tegra::DebugContext::BreakPointObserver( + debug_context) { + setObjectName("TegraBreakPointsWidget"); + + status_text = new QLabel(tr("Emulation running")); + resume_button = new QPushButton(tr("Resume")); + resume_button->setEnabled(false); + + breakpoint_model = new BreakPointModel(debug_context, this); + breakpoint_list = new QTreeView; + breakpoint_list->setRootIsDecorated(false); + breakpoint_list->setHeaderHidden(true); + breakpoint_list->setModel(breakpoint_model); + + qRegisterMetaType<Tegra::DebugContext::Event>("Tegra::DebugContext::Event"); + + connect(breakpoint_list, SIGNAL(doubleClicked(const QModelIndex&)), this, + SLOT(OnItemDoubleClicked(const QModelIndex&))); + + connect(resume_button, SIGNAL(clicked()), this, SLOT(OnResumeRequested())); + + connect(this, SIGNAL(BreakPointHit(Tegra::DebugContext::Event, void*)), this, + SLOT(OnBreakPointHit(Tegra::DebugContext::Event, void*)), Qt::BlockingQueuedConnection); + connect(this, SIGNAL(Resumed()), this, SLOT(OnResumed())); + + connect(this, SIGNAL(BreakPointHit(Tegra::DebugContext::Event, void*)), breakpoint_model, + SLOT(OnBreakPointHit(Tegra::DebugContext::Event)), Qt::BlockingQueuedConnection); + connect(this, SIGNAL(Resumed()), breakpoint_model, SLOT(OnResumed())); + + connect(this, SIGNAL(BreakPointsChanged(const QModelIndex&, const QModelIndex&)), + breakpoint_model, SIGNAL(dataChanged(const QModelIndex&, const QModelIndex&))); + + QWidget* main_widget = new QWidget; + auto main_layout = new QVBoxLayout; + { + auto sub_layout = new QHBoxLayout; + sub_layout->addWidget(status_text); + sub_layout->addWidget(resume_button); + main_layout->addLayout(sub_layout); + } + main_layout->addWidget(breakpoint_list); + main_widget->setLayout(main_layout); + + setWidget(main_widget); +} + +void GraphicsBreakPointsWidget::OnMaxwellBreakPointHit(Event event, void* data) { + // Process in GUI thread + emit BreakPointHit(event, data); +} + +void GraphicsBreakPointsWidget::OnBreakPointHit(Tegra::DebugContext::Event event, void* data) { + status_text->setText(tr("Emulation halted at breakpoint")); + resume_button->setEnabled(true); +} + +void GraphicsBreakPointsWidget::OnMaxwellResume() { + // Process in GUI thread + emit Resumed(); +} + +void GraphicsBreakPointsWidget::OnResumed() { + status_text->setText(tr("Emulation running")); + resume_button->setEnabled(false); +} + +void GraphicsBreakPointsWidget::OnResumeRequested() { + if (auto context = context_weak.lock()) + context->Resume(); +} + +void GraphicsBreakPointsWidget::OnItemDoubleClicked(const QModelIndex& index) { + if (!index.isValid()) + return; + + QModelIndex check_index = breakpoint_list->model()->index(index.row(), 0); + QVariant enabled = breakpoint_list->model()->data(check_index, Qt::CheckStateRole); + QVariant new_state = Qt::Unchecked; + if (enabled == Qt::Unchecked) + new_state = Qt::Checked; + breakpoint_list->model()->setData(check_index, new_state, Qt::CheckStateRole); +} diff --git a/src/yuzu/debugger/graphics/graphics_breakpoints.h b/src/yuzu/debugger/graphics/graphics_breakpoints.h new file mode 100644 index 000000000..ae0ede2e8 --- /dev/null +++ b/src/yuzu/debugger/graphics/graphics_breakpoints.h @@ -0,0 +1,46 @@ +// Copyright 2014 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <memory> +#include <QDockWidget> +#include "video_core/debug_utils/debug_utils.h" + +class QLabel; +class QPushButton; +class QTreeView; + +class BreakPointModel; + +class GraphicsBreakPointsWidget : public QDockWidget, Tegra::DebugContext::BreakPointObserver { + Q_OBJECT + + using Event = Tegra::DebugContext::Event; + +public: + explicit GraphicsBreakPointsWidget(std::shared_ptr<Tegra::DebugContext> debug_context, + QWidget* parent = nullptr); + + void OnMaxwellBreakPointHit(Tegra::DebugContext::Event event, void* data) override; + void OnMaxwellResume() override; + +public slots: + void OnBreakPointHit(Tegra::DebugContext::Event event, void* data); + void OnItemDoubleClicked(const QModelIndex&); + void OnResumeRequested(); + void OnResumed(); + +signals: + void Resumed(); + void BreakPointHit(Tegra::DebugContext::Event event, void* data); + void BreakPointsChanged(const QModelIndex& topLeft, const QModelIndex& bottomRight); + +private: + QLabel* status_text; + QPushButton* resume_button; + + BreakPointModel* breakpoint_model; + QTreeView* breakpoint_list; +}; diff --git a/src/yuzu/debugger/graphics/graphics_breakpoints_p.h b/src/yuzu/debugger/graphics/graphics_breakpoints_p.h new file mode 100644 index 000000000..35a6876ae --- /dev/null +++ b/src/yuzu/debugger/graphics/graphics_breakpoints_p.h @@ -0,0 +1,36 @@ +// Copyright 2014 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <memory> +#include <QAbstractListModel> +#include "video_core/debug_utils/debug_utils.h" + +class BreakPointModel : public QAbstractListModel { + Q_OBJECT + +public: + enum { + Role_IsEnabled = Qt::UserRole, + }; + + BreakPointModel(std::shared_ptr<Tegra::DebugContext> context, QObject* parent); + + int columnCount(const QModelIndex& parent = QModelIndex()) const override; + int rowCount(const QModelIndex& parent = QModelIndex()) const override; + QVariant data(const QModelIndex& index, int role = Qt::DisplayRole) const override; + Qt::ItemFlags flags(const QModelIndex& index) const override; + + bool setData(const QModelIndex& index, const QVariant& value, int role = Qt::EditRole) override; + +public slots: + void OnBreakPointHit(Tegra::DebugContext::Event event); + void OnResumed(); + +private: + std::weak_ptr<Tegra::DebugContext> context_weak; + bool at_breakpoint; + Tegra::DebugContext::Event active_breakpoint; +}; diff --git a/src/yuzu/debugger/graphics/graphics_surface.cpp b/src/yuzu/debugger/graphics/graphics_surface.cpp new file mode 100644 index 000000000..1e4844b57 --- /dev/null +++ b/src/yuzu/debugger/graphics/graphics_surface.cpp @@ -0,0 +1,451 @@ +// Copyright 2014 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <QBoxLayout> +#include <QComboBox> +#include <QDebug> +#include <QFileDialog> +#include <QLabel> +#include <QMouseEvent> +#include <QPushButton> +#include <QScrollArea> +#include <QSpinBox> +#include "core/core.h" +#include "video_core/engines/maxwell_3d.h" +#include "video_core/gpu.h" +#include "video_core/textures/decoders.h" +#include "video_core/textures/texture.h" +#include "video_core/utils.h" +#include "yuzu/debugger/graphics/graphics_surface.h" +#include "yuzu/util/spinbox.h" + +static Tegra::Texture::TextureFormat ConvertToTextureFormat( + Tegra::RenderTargetFormat render_target_format) { + switch (render_target_format) { + case Tegra::RenderTargetFormat::RGBA8_UNORM: + return Tegra::Texture::TextureFormat::A8R8G8B8; + default: + UNIMPLEMENTED_MSG("Unimplemented RT format"); + } +} + +SurfacePicture::SurfacePicture(QWidget* parent, GraphicsSurfaceWidget* surface_widget_) + : QLabel(parent), surface_widget(surface_widget_) {} +SurfacePicture::~SurfacePicture() {} + +void SurfacePicture::mousePressEvent(QMouseEvent* event) { + // Only do something while the left mouse button is held down + if (!(event->buttons() & Qt::LeftButton)) + return; + + if (pixmap() == nullptr) + return; + + if (surface_widget) + surface_widget->Pick(event->x() * pixmap()->width() / width(), + event->y() * pixmap()->height() / height()); +} + +void SurfacePicture::mouseMoveEvent(QMouseEvent* event) { + // We also want to handle the event if the user moves the mouse while holding down the LMB + mousePressEvent(event); +} + +GraphicsSurfaceWidget::GraphicsSurfaceWidget(std::shared_ptr<Tegra::DebugContext> debug_context, + QWidget* parent) + : BreakPointObserverDock(debug_context, tr("Maxwell Surface Viewer"), parent), + surface_source(Source::RenderTarget0) { + setObjectName("MaxwellSurface"); + + surface_source_list = new QComboBox; + surface_source_list->addItem(tr("Render Target 0")); + surface_source_list->addItem(tr("Render Target 1")); + surface_source_list->addItem(tr("Render Target 2")); + surface_source_list->addItem(tr("Render Target 3")); + surface_source_list->addItem(tr("Render Target 4")); + surface_source_list->addItem(tr("Render Target 5")); + surface_source_list->addItem(tr("Render Target 6")); + surface_source_list->addItem(tr("Render Target 7")); + surface_source_list->addItem(tr("Z Buffer")); + surface_source_list->addItem(tr("Custom")); + surface_source_list->setCurrentIndex(static_cast<int>(surface_source)); + + surface_address_control = new CSpinBox; + surface_address_control->SetBase(16); + surface_address_control->SetRange(0, 0x7FFFFFFFFFFFFFFF); + surface_address_control->SetPrefix("0x"); + + unsigned max_dimension = 16384; // TODO: Find actual maximum + + surface_width_control = new QSpinBox; + surface_width_control->setRange(0, max_dimension); + + surface_height_control = new QSpinBox; + surface_height_control->setRange(0, max_dimension); + + surface_picker_x_control = new QSpinBox; + surface_picker_x_control->setRange(0, max_dimension - 1); + + surface_picker_y_control = new QSpinBox; + surface_picker_y_control->setRange(0, max_dimension - 1); + + surface_format_control = new QComboBox; + + // Color formats sorted by Maxwell texture format index + surface_format_control->addItem(tr("None")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("A8R8G8B8")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("Unknown")); + surface_format_control->addItem(tr("DXT1")); + surface_format_control->addItem(tr("DXT23")); + surface_format_control->addItem(tr("DXT45")); + surface_format_control->addItem(tr("DXN1")); + surface_format_control->addItem(tr("DXN2")); + + surface_info_label = new QLabel(); + surface_info_label->setWordWrap(true); + + surface_picture_label = new SurfacePicture(0, this); + surface_picture_label->setSizePolicy(QSizePolicy::Fixed, QSizePolicy::Fixed); + surface_picture_label->setAlignment(Qt::AlignLeft | Qt::AlignTop); + surface_picture_label->setScaledContents(false); + + auto scroll_area = new QScrollArea(); + scroll_area->setBackgroundRole(QPalette::Dark); + scroll_area->setWidgetResizable(false); + scroll_area->setWidget(surface_picture_label); + + save_surface = new QPushButton(QIcon::fromTheme("document-save"), tr("Save")); + + // Connections + connect(this, SIGNAL(Update()), this, SLOT(OnUpdate())); + connect(surface_source_list, SIGNAL(currentIndexChanged(int)), this, + SLOT(OnSurfaceSourceChanged(int))); + connect(surface_address_control, SIGNAL(ValueChanged(qint64)), this, + SLOT(OnSurfaceAddressChanged(qint64))); + connect(surface_width_control, SIGNAL(valueChanged(int)), this, + SLOT(OnSurfaceWidthChanged(int))); + connect(surface_height_control, SIGNAL(valueChanged(int)), this, + SLOT(OnSurfaceHeightChanged(int))); + connect(surface_format_control, SIGNAL(currentIndexChanged(int)), this, + SLOT(OnSurfaceFormatChanged(int))); + connect(surface_picker_x_control, SIGNAL(valueChanged(int)), this, + SLOT(OnSurfacePickerXChanged(int))); + connect(surface_picker_y_control, SIGNAL(valueChanged(int)), this, + SLOT(OnSurfacePickerYChanged(int))); + connect(save_surface, SIGNAL(clicked()), this, SLOT(SaveSurface())); + + auto main_widget = new QWidget; + auto main_layout = new QVBoxLayout; + { + auto sub_layout = new QHBoxLayout; + sub_layout->addWidget(new QLabel(tr("Source:"))); + sub_layout->addWidget(surface_source_list); + main_layout->addLayout(sub_layout); + } + { + auto sub_layout = new QHBoxLayout; + sub_layout->addWidget(new QLabel(tr("GPU Address:"))); + sub_layout->addWidget(surface_address_control); + main_layout->addLayout(sub_layout); + } + { + auto sub_layout = new QHBoxLayout; + sub_layout->addWidget(new QLabel(tr("Width:"))); + sub_layout->addWidget(surface_width_control); + main_layout->addLayout(sub_layout); + } + { + auto sub_layout = new QHBoxLayout; + sub_layout->addWidget(new QLabel(tr("Height:"))); + sub_layout->addWidget(surface_height_control); + main_layout->addLayout(sub_layout); + } + { + auto sub_layout = new QHBoxLayout; + sub_layout->addWidget(new QLabel(tr("Format:"))); + sub_layout->addWidget(surface_format_control); + main_layout->addLayout(sub_layout); + } + main_layout->addWidget(scroll_area); + + auto info_layout = new QHBoxLayout; + { + auto xy_layout = new QVBoxLayout; + { + { + auto sub_layout = new QHBoxLayout; + sub_layout->addWidget(new QLabel(tr("X:"))); + sub_layout->addWidget(surface_picker_x_control); + xy_layout->addLayout(sub_layout); + } + { + auto sub_layout = new QHBoxLayout; + sub_layout->addWidget(new QLabel(tr("Y:"))); + sub_layout->addWidget(surface_picker_y_control); + xy_layout->addLayout(sub_layout); + } + } + info_layout->addLayout(xy_layout); + surface_info_label->setSizePolicy(QSizePolicy::Expanding, QSizePolicy::Minimum); + info_layout->addWidget(surface_info_label); + } + main_layout->addLayout(info_layout); + + main_layout->addWidget(save_surface); + main_widget->setLayout(main_layout); + setWidget(main_widget); + + // Load current data - TODO: Make sure this works when emulation is not running + if (debug_context && debug_context->at_breakpoint) { + emit Update(); + widget()->setEnabled(debug_context->at_breakpoint); + } else { + widget()->setEnabled(false); + } +} + +void GraphicsSurfaceWidget::OnBreakPointHit(Tegra::DebugContext::Event event, void* data) { + emit Update(); + widget()->setEnabled(true); +} + +void GraphicsSurfaceWidget::OnResumed() { + widget()->setEnabled(false); +} + +void GraphicsSurfaceWidget::OnSurfaceSourceChanged(int new_value) { + surface_source = static_cast<Source>(new_value); + emit Update(); +} + +void GraphicsSurfaceWidget::OnSurfaceAddressChanged(qint64 new_value) { + if (surface_address != new_value) { + surface_address = static_cast<Tegra::GPUVAddr>(new_value); + + surface_source_list->setCurrentIndex(static_cast<int>(Source::Custom)); + emit Update(); + } +} + +void GraphicsSurfaceWidget::OnSurfaceWidthChanged(int new_value) { + if (surface_width != static_cast<unsigned>(new_value)) { + surface_width = static_cast<unsigned>(new_value); + + surface_source_list->setCurrentIndex(static_cast<int>(Source::Custom)); + emit Update(); + } +} + +void GraphicsSurfaceWidget::OnSurfaceHeightChanged(int new_value) { + if (surface_height != static_cast<unsigned>(new_value)) { + surface_height = static_cast<unsigned>(new_value); + + surface_source_list->setCurrentIndex(static_cast<int>(Source::Custom)); + emit Update(); + } +} + +void GraphicsSurfaceWidget::OnSurfaceFormatChanged(int new_value) { + if (surface_format != static_cast<Tegra::Texture::TextureFormat>(new_value)) { + surface_format = static_cast<Tegra::Texture::TextureFormat>(new_value); + + surface_source_list->setCurrentIndex(static_cast<int>(Source::Custom)); + emit Update(); + } +} + +void GraphicsSurfaceWidget::OnSurfacePickerXChanged(int new_value) { + if (surface_picker_x != new_value) { + surface_picker_x = new_value; + Pick(surface_picker_x, surface_picker_y); + } +} + +void GraphicsSurfaceWidget::OnSurfacePickerYChanged(int new_value) { + if (surface_picker_y != new_value) { + surface_picker_y = new_value; + Pick(surface_picker_x, surface_picker_y); + } +} + +void GraphicsSurfaceWidget::Pick(int x, int y) { + surface_picker_x_control->setValue(x); + surface_picker_y_control->setValue(y); + + if (x < 0 || x >= static_cast<int>(surface_width) || y < 0 || + y >= static_cast<int>(surface_height)) { + surface_info_label->setText(tr("Pixel out of bounds")); + surface_info_label->setAlignment(Qt::AlignLeft | Qt::AlignVCenter); + return; + } + + surface_info_label->setText(QString("Raw: <Unimplemented>\n(%1)").arg("<Unimplemented>")); + surface_info_label->setAlignment(Qt::AlignLeft | Qt::AlignVCenter); +} + +void GraphicsSurfaceWidget::OnUpdate() { + auto& gpu = Core::System::GetInstance().GPU(); + + QPixmap pixmap; + + switch (surface_source) { + case Source::RenderTarget0: + case Source::RenderTarget1: + case Source::RenderTarget2: + case Source::RenderTarget3: + case Source::RenderTarget4: + case Source::RenderTarget5: + case Source::RenderTarget6: + case Source::RenderTarget7: { + // TODO: Store a reference to the registers in the debug context instead of accessing them + // directly... + + auto& registers = gpu.Get3DEngine().regs; + auto& rt = registers.rt[static_cast<size_t>(surface_source) - + static_cast<size_t>(Source::RenderTarget0)]; + + surface_address = rt.Address(); + surface_width = rt.width; + surface_height = rt.height; + if (rt.format != Tegra::RenderTargetFormat::NONE) { + surface_format = ConvertToTextureFormat(rt.format); + } + + break; + } + + case Source::Custom: { + // Keep user-specified values + break; + } + + default: + qDebug() << "Unknown surface source " << static_cast<int>(surface_source); + break; + } + + surface_address_control->SetValue(surface_address); + surface_width_control->setValue(surface_width); + surface_height_control->setValue(surface_height); + surface_format_control->setCurrentIndex(static_cast<int>(surface_format)); + + if (surface_address == 0) { + surface_picture_label->hide(); + surface_info_label->setText(tr("(invalid surface address)")); + surface_info_label->setAlignment(Qt::AlignCenter); + surface_picker_x_control->setEnabled(false); + surface_picker_y_control->setEnabled(false); + save_surface->setEnabled(false); + return; + } + + // TODO: Implement a good way to visualize alpha components! + + QImage decoded_image(surface_width, surface_height, QImage::Format_ARGB32); + VAddr address = gpu.memory_manager->PhysicalToVirtualAddress(surface_address); + + auto unswizzled_data = + Tegra::Texture::UnswizzleTexture(address, surface_format, surface_width, surface_height); + + auto texture_data = Tegra::Texture::DecodeTexture(unswizzled_data, surface_format, + surface_width, surface_height); + + surface_picture_label->show(); + + for (unsigned int y = 0; y < surface_height; ++y) { + for (unsigned int x = 0; x < surface_width; ++x) { + Math::Vec4<u8> color; + color[0] = texture_data[x + y * surface_width + 0]; + color[1] = texture_data[x + y * surface_width + 1]; + color[2] = texture_data[x + y * surface_width + 2]; + color[3] = texture_data[x + y * surface_width + 3]; + decoded_image.setPixel(x, y, qRgba(color.r(), color.g(), color.b(), color.a())); + } + } + + pixmap = QPixmap::fromImage(decoded_image); + surface_picture_label->setPixmap(pixmap); + surface_picture_label->resize(pixmap.size()); + + // Update the info with pixel data + surface_picker_x_control->setEnabled(true); + surface_picker_y_control->setEnabled(true); + Pick(surface_picker_x, surface_picker_y); + + // Enable saving the converted pixmap to file + save_surface->setEnabled(true); +} + +void GraphicsSurfaceWidget::SaveSurface() { + QString png_filter = tr("Portable Network Graphic (*.png)"); + QString bin_filter = tr("Binary data (*.bin)"); + + QString selectedFilter; + QString filename = QFileDialog::getSaveFileName( + this, tr("Save Surface"), + QString("texture-0x%1.png").arg(QString::number(surface_address, 16)), + QString("%1;;%2").arg(png_filter, bin_filter), &selectedFilter); + + if (filename.isEmpty()) { + // If the user canceled the dialog, don't save anything. + return; + } + + if (selectedFilter == png_filter) { + const QPixmap* pixmap = surface_picture_label->pixmap(); + ASSERT_MSG(pixmap != nullptr, "No pixmap set"); + + QFile file(filename); + file.open(QIODevice::WriteOnly); + if (pixmap) + pixmap->save(&file, "PNG"); + } else if (selectedFilter == bin_filter) { + auto& gpu = Core::System::GetInstance().GPU(); + VAddr address = gpu.memory_manager->PhysicalToVirtualAddress(surface_address); + + const u8* buffer = Memory::GetPointer(address); + ASSERT_MSG(buffer != nullptr, "Memory not accessible"); + + QFile file(filename); + file.open(QIODevice::WriteOnly); + int size = surface_width * surface_height * Tegra::Texture::BytesPerPixel(surface_format); + QByteArray data(reinterpret_cast<const char*>(buffer), size); + file.write(data); + } else { + UNREACHABLE_MSG("Unhandled filter selected"); + } +} diff --git a/src/yuzu/debugger/graphics/graphics_surface.h b/src/yuzu/debugger/graphics/graphics_surface.h new file mode 100644 index 000000000..6a344bdfc --- /dev/null +++ b/src/yuzu/debugger/graphics/graphics_surface.h @@ -0,0 +1,97 @@ +// Copyright 2014 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <QLabel> +#include <QPushButton> +#include "video_core/memory_manager.h" +#include "video_core/textures/texture.h" +#include "yuzu/debugger/graphics/graphics_breakpoint_observer.h" + +class QComboBox; +class QSpinBox; +class CSpinBox; + +class GraphicsSurfaceWidget; + +class SurfacePicture : public QLabel { + Q_OBJECT + +public: + explicit SurfacePicture(QWidget* parent = nullptr, + GraphicsSurfaceWidget* surface_widget = nullptr); + ~SurfacePicture(); + +protected slots: + virtual void mouseMoveEvent(QMouseEvent* event); + virtual void mousePressEvent(QMouseEvent* event); + +private: + GraphicsSurfaceWidget* surface_widget; +}; + +class GraphicsSurfaceWidget : public BreakPointObserverDock { + Q_OBJECT + + using Event = Tegra::DebugContext::Event; + + enum class Source { + RenderTarget0 = 0, + RenderTarget1 = 1, + RenderTarget2 = 2, + RenderTarget3 = 3, + RenderTarget4 = 4, + RenderTarget5 = 5, + RenderTarget6 = 6, + RenderTarget7 = 7, + ZBuffer = 8, + Custom = 9, + }; + +public: + explicit GraphicsSurfaceWidget(std::shared_ptr<Tegra::DebugContext> debug_context, + QWidget* parent = nullptr); + void Pick(int x, int y); + +public slots: + void OnSurfaceSourceChanged(int new_value); + void OnSurfaceAddressChanged(qint64 new_value); + void OnSurfaceWidthChanged(int new_value); + void OnSurfaceHeightChanged(int new_value); + void OnSurfaceFormatChanged(int new_value); + void OnSurfacePickerXChanged(int new_value); + void OnSurfacePickerYChanged(int new_value); + void OnUpdate(); + +private slots: + void OnBreakPointHit(Tegra::DebugContext::Event event, void* data) override; + void OnResumed() override; + + void SaveSurface(); + +signals: + void Update(); + +private: + QComboBox* surface_source_list; + CSpinBox* surface_address_control; + QSpinBox* surface_width_control; + QSpinBox* surface_height_control; + QComboBox* surface_format_control; + + SurfacePicture* surface_picture_label; + QSpinBox* surface_picker_x_control; + QSpinBox* surface_picker_y_control; + QLabel* surface_info_label; + QPushButton* save_surface; + + Source surface_source; + Tegra::GPUVAddr surface_address; + unsigned surface_width; + unsigned surface_height; + Tegra::Texture::TextureFormat surface_format; + int surface_picker_x = 0; + int surface_picker_y = 0; +}; diff --git a/src/yuzu/debugger/wait_tree.cpp b/src/yuzu/debugger/wait_tree.cpp index 7a62f57b5..cae2864e5 100644 --- a/src/yuzu/debugger/wait_tree.cpp +++ b/src/yuzu/debugger/wait_tree.cpp @@ -150,8 +150,8 @@ QString WaitTreeThread::GetText() const { case THREADSTATUS_READY: status = tr("ready"); break; - case THREADSTATUS_WAIT_ARB: - status = tr("waiting for address 0x%1").arg(thread.wait_address, 8, 16, QLatin1Char('0')); + case THREADSTATUS_WAIT_HLE_EVENT: + status = tr("waiting for HLE return"); break; case THREADSTATUS_WAIT_SLEEP: status = tr("sleeping"); @@ -180,7 +180,7 @@ QColor WaitTreeThread::GetColor() const { return QColor(Qt::GlobalColor::darkGreen); case THREADSTATUS_READY: return QColor(Qt::GlobalColor::darkBlue); - case THREADSTATUS_WAIT_ARB: + case THREADSTATUS_WAIT_HLE_EVENT: return QColor(Qt::GlobalColor::darkRed); case THREADSTATUS_WAIT_SLEEP: return QColor(Qt::GlobalColor::darkYellow); diff --git a/src/yuzu/main.cpp b/src/yuzu/main.cpp index 9b2047e04..793d9d739 100644 --- a/src/yuzu/main.cpp +++ b/src/yuzu/main.cpp @@ -18,7 +18,6 @@ #include "common/logging/log.h" #include "common/logging/text_formatter.h" #include "common/microprofile.h" -#include "common/platform.h" #include "common/scm_rev.h" #include "common/scope_exit.h" #include "common/string_util.h" @@ -26,10 +25,13 @@ #include "core/gdbstub/gdbstub.h" #include "core/loader/loader.h" #include "core/settings.h" +#include "video_core/debug_utils/debug_utils.h" #include "yuzu/about_dialog.h" #include "yuzu/bootmanager.h" #include "yuzu/configuration/config.h" #include "yuzu/configuration/configure_dialog.h" +#include "yuzu/debugger/graphics/graphics_breakpoints.h" +#include "yuzu/debugger/graphics/graphics_surface.h" #include "yuzu/debugger/profiler.h" #include "yuzu/debugger/registers.h" #include "yuzu/debugger/wait_tree.h" @@ -69,10 +71,16 @@ static void ShowCalloutMessage(const QString& message, CalloutFlag flag) { void GMainWindow::ShowCallouts() {} GMainWindow::GMainWindow() : config(new Config()), emu_thread(nullptr) { + + debug_context = Tegra::DebugContext::Construct(); + setAcceptDrops(true); ui.setupUi(this); statusBar()->hide(); + default_theme_paths = QIcon::themeSearchPaths(); + UpdateUITheme(); + InitializeWidgets(); InitializeDebugWidgets(); InitializeRecentFileMenuActions(); @@ -161,6 +169,16 @@ void GMainWindow::InitializeDebugWidgets() { connect(this, &GMainWindow::EmulationStopping, registersWidget, &RegistersWidget::OnEmulationStopping); + graphicsBreakpointsWidget = new GraphicsBreakPointsWidget(debug_context, this); + addDockWidget(Qt::RightDockWidgetArea, graphicsBreakpointsWidget); + graphicsBreakpointsWidget->hide(); + debug_menu->addAction(graphicsBreakpointsWidget->toggleViewAction()); + + graphicsSurfaceWidget = new GraphicsSurfaceWidget(debug_context, this); + addDockWidget(Qt::RightDockWidgetArea, graphicsSurfaceWidget); + graphicsSurfaceWidget->hide(); + debug_menu->addAction(graphicsSurfaceWidget->toggleViewAction()); + waitTreeWidget = new WaitTreeWidget(this); addDockWidget(Qt::LeftDockWidgetArea, waitTreeWidget); waitTreeWidget->hide(); @@ -187,7 +205,8 @@ void GMainWindow::InitializeHotkeys() { RegisterHotkey("Main Window", "Load File", QKeySequence::Open); RegisterHotkey("Main Window", "Start Emulation"); RegisterHotkey("Main Window", "Fullscreen", QKeySequence::FullScreen); - RegisterHotkey("Main Window", "Exit Fullscreen", QKeySequence::Cancel, Qt::ApplicationShortcut); + RegisterHotkey("Main Window", "Exit Fullscreen", QKeySequence(Qt::Key_Escape), + Qt::ApplicationShortcut); LoadHotkeys(); connect(GetHotkey("Main Window", "Load File", this), &QShortcut::activated, this, @@ -324,6 +343,8 @@ bool GMainWindow::LoadROM(const QString& filename) { Core::System& system{Core::System::GetInstance()}; + system.SetGPUDebugContext(debug_context); + const Core::System::ResultStatus result{system.Load(render_window, filename.toStdString())}; Core::Telemetry().AddField(Telemetry::FieldType::App, "Frontend", "Qt"); @@ -350,9 +371,9 @@ bool GMainWindow::LoadROM(const QString& filename) { "yuzu. A real Switch is required.<br/><br/>" "For more information on dumping and decrypting games, please see the following " "wiki pages: <ul>" - "<li><a href='https://citra-emu.org/wiki/dumping-game-cartridges/'>Dumping Game " + "<li><a href='https://yuzu-emu.org/wiki/dumping-game-cartridges/'>Dumping Game " "Cartridges</a></li>" - "<li><a href='https://citra-emu.org/wiki/dumping-installed-titles/'>Dumping " + "<li><a href='https://yuzu-emu.org/wiki/dumping-installed-titles/'>Dumping " "Installed Titles</a></li>" "</ul>")); break; @@ -635,6 +656,7 @@ void GMainWindow::OnConfigure() { auto result = configureDialog.exec(); if (result == QDialog::Accepted) { configureDialog.applyConfiguration(); + UpdateUITheme(); config->Save(); } } @@ -673,18 +695,18 @@ void GMainWindow::UpdateStatusBar() { void GMainWindow::OnCoreError(Core::System::ResultStatus result, std::string details) { QMessageBox::StandardButton answer; QString status_message; - const QString common_message = - tr("The game you are trying to load requires additional files from your 3DS to be dumped " - "before playing.<br/><br/>For more information on dumping these files, please see the " - "following wiki page: <a " - "href='https://citra-emu.org/wiki/" - "dumping-system-archives-and-the-shared-fonts-from-a-3ds-console/'>Dumping System " - "Archives and the Shared Fonts from a 3DS Console</a>.<br/><br/>Would you like to quit " - "back to the game list? Continuing emulation may result in crashes, corrupted save " - "data, or other bugs."); + const QString common_message = tr( + "The game you are trying to load requires additional files from your Switch to be dumped " + "before playing.<br/><br/>For more information on dumping these files, please see the " + "following wiki page: <a " + "href='https://yuzu-emu.org/wiki/" + "dumping-system-archives-and-the-shared-fonts-from-a-switch-console/'>Dumping System " + "Archives and the Shared Fonts from a Switch Console</a>.<br/><br/>Would you like to quit " + "back to the game list? Continuing emulation may result in crashes, corrupted save " + "data, or other bugs."); switch (result) { case Core::System::ResultStatus::ErrorSystemFiles: { - QString message = "Citra was unable to locate a 3DS system archive"; + QString message = "yuzu was unable to locate a Switch system archive"; if (!details.empty()) { message.append(tr(": %1. ").arg(details.c_str())); } else { @@ -699,7 +721,7 @@ void GMainWindow::OnCoreError(Core::System::ResultStatus result, std::string det } case Core::System::ResultStatus::ErrorSharedFont: { - QString message = tr("Citra was unable to locate the 3DS shared fonts. "); + QString message = tr("yuzu was unable to locate the Switch shared fonts. "); message.append(common_message); answer = QMessageBox::question(this, tr("Shared Fonts Not Found"), message, QMessageBox::Yes | QMessageBox::No, QMessageBox::No); @@ -815,6 +837,32 @@ void GMainWindow::filterBarSetChecked(bool state) { emit(OnToggleFilterBar()); } +void GMainWindow::UpdateUITheme() { + QStringList theme_paths(default_theme_paths); + if (UISettings::values.theme != UISettings::themes[0].second && + !UISettings::values.theme.isEmpty()) { + QString theme_uri(":" + UISettings::values.theme + "/style.qss"); + QFile f(theme_uri); + if (!f.exists()) { + LOG_ERROR(Frontend, "Unable to set style, stylesheet file not found"); + } else { + f.open(QFile::ReadOnly | QFile::Text); + QTextStream ts(&f); + qApp->setStyleSheet(ts.readAll()); + GMainWindow::setStyleSheet(ts.readAll()); + } + theme_paths.append(QStringList{":/icons/default", ":/icons/" + UISettings::values.theme}); + QIcon::setThemeName(":/icons/" + UISettings::values.theme); + } else { + qApp->setStyleSheet(""); + GMainWindow::setStyleSheet(""); + theme_paths.append(QStringList{":/icons/default"}); + QIcon::setThemeName(":/icons/default"); + } + QIcon::setThemeSearchPaths(theme_paths); + emit UpdateThemedIcons(); +} + #ifdef main #undef main #endif diff --git a/src/yuzu/main.h b/src/yuzu/main.h index 4a0d912bb..20ff65314 100644 --- a/src/yuzu/main.h +++ b/src/yuzu/main.h @@ -2,8 +2,7 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#ifndef _CITRA_QT_MAIN_HXX_ -#define _CITRA_QT_MAIN_HXX_ +#pragma once #include <memory> #include <QMainWindow> @@ -15,17 +14,18 @@ class Config; class EmuThread; class GameList; class GImageInfo; -class GPUCommandStreamWidget; -class GPUCommandListWidget; class GraphicsBreakPointsWidget; -class GraphicsTracingWidget; -class GraphicsVertexShaderWidget; +class GraphicsSurfaceWidget; class GRenderWindow; class MicroProfileDialog; class ProfilerWidget; class RegistersWidget; class WaitTreeWidget; +namespace Tegra { +class DebugContext; +} + class GMainWindow : public QMainWindow { Q_OBJECT @@ -64,6 +64,9 @@ signals: */ void EmulationStopping(); + // Signal that tells widgets to update icons to use the current theme + void UpdateThemedIcons(); + private: void InitializeWidgets(); void InitializeDebugWidgets(); @@ -138,6 +141,8 @@ private: Ui::MainWindow ui; + std::shared_ptr<Tegra::DebugContext> debug_context; + GRenderWindow* render_window; GameList* game_list; @@ -150,7 +155,7 @@ private: std::unique_ptr<Config> config; - // Whether emulation is currently running in Citra. + // Whether emulation is currently running in yuzu. bool emulation_running = false; std::unique_ptr<EmuThread> emu_thread; @@ -158,14 +163,17 @@ private: ProfilerWidget* profilerWidget; MicroProfileDialog* microProfileDialog; RegistersWidget* registersWidget; + GraphicsBreakPointsWidget* graphicsBreakpointsWidget; + GraphicsSurfaceWidget* graphicsSurfaceWidget; WaitTreeWidget* waitTreeWidget; QAction* actions_recent_files[max_recent_files_item]; + // stores default icon theme search paths for the platform + QStringList default_theme_paths; + protected: void dropEvent(QDropEvent* event) override; void dragEnterEvent(QDragEnterEvent* event) override; void dragMoveEvent(QDragMoveEvent* event) override; }; - -#endif // _CITRA_QT_MAIN_HXX_ diff --git a/src/yuzu/ui_settings.h b/src/yuzu/ui_settings.h index 9036ce2c1..8e215a002 100644 --- a/src/yuzu/ui_settings.h +++ b/src/yuzu/ui_settings.h @@ -15,6 +15,10 @@ namespace UISettings { using ContextualShortcut = std::pair<QString, int>; using Shortcut = std::pair<QString, ContextualShortcut>; +static const std::array<std::pair<QString, QString>, 2> themes = { + {std::make_pair(QString("Default"), QString("default")), + std::make_pair(QString("Dark"), QString("qdarkstyle"))}}; + struct Values { QByteArray geometry; QByteArray state; diff --git a/src/yuzu_cmd/config.cpp b/src/yuzu_cmd/config.cpp index bf79d2e81..8b479bc6d 100644 --- a/src/yuzu_cmd/config.cpp +++ b/src/yuzu_cmd/config.cpp @@ -90,8 +90,7 @@ void Config::ReadValues() { sdl2_config->Get("Controls", "touch_device", "engine:emu_window"); // Core - Settings::values.cpu_core = - static_cast<Settings::CpuCore>(sdl2_config->GetInteger("Core", "cpu_core", 0)); + Settings::values.use_cpu_jit = sdl2_config->GetBoolean("Core", "use_cpu_jit", true); // Renderer Settings::values.resolution_factor = @@ -107,6 +106,9 @@ void Config::ReadValues() { Settings::values.use_virtual_sd = sdl2_config->GetBoolean("Data Storage", "use_virtual_sd", true); + // System + Settings::values.use_docked_mode = sdl2_config->GetBoolean("System", "use_docked_mode", true); + // Miscellaneous Settings::values.log_filter = sdl2_config->Get("Miscellaneous", "log_filter", "*:Trace"); diff --git a/src/yuzu_cmd/default_ini.h b/src/yuzu_cmd/default_ini.h index 156f7a1a4..895a42c39 100644 --- a/src/yuzu_cmd/default_ini.h +++ b/src/yuzu_cmd/default_ini.h @@ -76,9 +76,9 @@ motion_device= touch_device= [Core] -# Which CPU core to use for CPU emulation -# 0 (default): Unicorn (slow), 1: Dynarmic (faster) -cpu_core = +# Whether to use the Just-In-Time (JIT) compiler for CPU emulation +# 0: Interpreter (slow), 1 (default): JIT (fast) +use_cpu_jit = [Renderer] # Whether to use software or hardware rendering. @@ -154,9 +154,9 @@ output_device = use_virtual_sd = [System] -# The system model that Citra will try to emulate -# 0: Old 3DS (default), 1: New 3DS -is_new_3ds = +# Whether the system is docked +# 1 (default): Yes, 0: No +use_docked_mode = # The system region that Citra will use during emulation # -1: Auto-select (default), 0: Japan, 1: USA, 2: Europe, 3: Australia, 4: China, 5: Korea, 6: Taiwan @@ -177,12 +177,12 @@ gdbstub_port=24689 # 0: No, 1 (default): Yes enable_telemetry = # Endpoint URL for submitting telemetry data -telemetry_endpoint_url = https://services.citra-emu.org/api/telemetry +telemetry_endpoint_url = # Endpoint URL to verify the username and token -verify_endpoint_url = https://services.citra-emu.org/api/profile -# Username and token for Citra Web Service +verify_endpoint_url = +# Username and token for yuzu Web Service # See https://services.citra-emu.org/ for more info -citra_username = -citra_token = +yuzu_username = +yuzu_token = )"; } diff --git a/src/yuzu_cmd/yuzu.cpp b/src/yuzu_cmd/yuzu.cpp index 8f7c75796..261312f62 100644 --- a/src/yuzu_cmd/yuzu.cpp +++ b/src/yuzu_cmd/yuzu.cpp @@ -147,7 +147,7 @@ int main(int argc, char** argv) { LOG_CRITICAL(Frontend, "The game that you are trying to load must be decrypted before " "being used with yuzu. \n\n For more information on dumping and " "decrypting games, please refer to: " - "https://citra-emu.org/wiki/dumping-game-cartridges/"); + "https://yuzu-emu.org/wiki/dumping-game-cartridges/"); return -1; case Core::System::ResultStatus::ErrorLoader_ErrorInvalidFormat: LOG_CRITICAL(Frontend, "Error while loading ROM: The ROM format is not supported."); diff --git a/src/yuzu_cmd/yuzu.rc b/src/yuzu_cmd/yuzu.rc index 7cb8a14e1..7de8ef3d9 100644 --- a/src/yuzu_cmd/yuzu.rc +++ b/src/yuzu_cmd/yuzu.rc @@ -6,7 +6,7 @@ // Icon with lowest ID value placed first to ensure application icon // remains consistent on all systems. -CITRA_ICON ICON "../../dist/yuzu.ico" +YUZU_ICON ICON "../../dist/yuzu.ico" ///////////////////////////////////////////////////////////////////////////// |