FileSys: Append the requested path to the filesystem base path in DeleteFile.

We were trying to delete things in the current directory instead of the actual filesystem directory. This may fix some savedata issues in some games.

src/common/string_util.cpp

@@ -134,7 +134,7 @@ bool SplitPath(const std::string& full_path, std::string* _pPath, std::string* _
     size_t dir_end = full_path.find_last_of("/"
 // windows needs the : included for something like just "C:" to be considered a directory
 #ifdef _WIN32
-                                            ":"
+                                            "\\:"
 #endif
     );
     if (std::string::npos == dir_end)

src/core/arm/unicorn/arm_unicorn.cpp

@@ -193,11 +193,11 @@ void ARM_Unicorn::ExecuteInstructions(int num_instructions) {
         }
         Kernel::Thread* thread = Kernel::GetCurrentThread();
         SaveContext(thread->context);
-        if (last_bkpt_hit) {
+        if (last_bkpt_hit || (num_instructions == 1)) {
             last_bkpt_hit = false;
             GDBStub::Break();
+            GDBStub::SendTrap(thread, 5);
         }
-        GDBStub::SendTrap(thread, 5);
     }
 }
 

src/core/file_sys/disk_filesystem.cpp

@@ -58,11 +58,13 @@ ResultVal<std::unique_ptr<StorageBackend>> Disk_FileSystem::OpenFile(const std::
 }
 
 ResultCode Disk_FileSystem::DeleteFile(const std::string& path) const {
-    if (!FileUtil::Exists(path)) {
+    std::string full_path = base_directory + path;
+
+    if (!FileUtil::Exists(full_path)) {
         return ERROR_PATH_NOT_FOUND;
     }
 
-    FileUtil::Delete(path);
+    FileUtil::Delete(full_path);
 
     return RESULT_SUCCESS;
 }

src/core/gdbstub/gdbstub.cpp

@@ -61,10 +61,16 @@ const u32 SIGTERM = 15;
 const u32 MSG_WAITALL = 8;
 #endif
 
-const u32 X30_REGISTER = 30;
+const u32 LR_REGISTER = 30;
 const u32 SP_REGISTER = 31;
 const u32 PC_REGISTER = 32;
 const u32 CPSR_REGISTER = 33;
+const u32 UC_ARM64_REG_Q0 = 34;
+const u32 FPSCR_REGISTER = 66;
+
+// TODO/WiP - Used while working on support for FPU
+const u32 TODO_DUMMY_REG_997 = 997;
+const u32 TODO_DUMMY_REG_998 = 998;
 
 // For sample XML files see the GDB source /gdb/features
 // GDB also wants the l character at the start
@@ -130,6 +136,8 @@ static const char* target_xml =
     </flags>
     <reg name="cpsr" bitsize="32" type="cpsr_flags"/>
   </feature>
+  <feature name="org.gnu.gdb.aarch64.fpu">
+  </feature>
 </target>
 )";
 
@@ -144,6 +152,7 @@ static u32 latest_signal = 0;
 static bool memory_break = false;
 
 static Kernel::Thread* current_thread = nullptr;
+static u32 current_core = 0;
 
 // Binding to a port within the reserved ports range (0-1023) requires root permissions,
 // so default to a port outside of that range.
@@ -171,13 +180,34 @@ static std::map<u64, Breakpoint> breakpoints_execute;
 static std::map<u64, Breakpoint> breakpoints_read;
 static std::map<u64, Breakpoint> breakpoints_write;
 
+struct Module {
+    std::string name;
+    PAddr beg;
+    PAddr end;
+};
+
+static std::vector<Module> modules;
+
+void RegisterModule(std::string name, PAddr beg, PAddr end, bool add_elf_ext) {
+    Module module;
+    if (add_elf_ext) {
+        Common::SplitPath(name, nullptr, &module.name, nullptr);
+        module.name += ".elf";
+    } else {
+        module.name = std::move(name);
+    }
+    module.beg = beg;
+    module.end = end;
+    modules.push_back(std::move(module));
+}
+
 static Kernel::Thread* FindThreadById(int id) {
-    for (int core = 0; core < Core::NUM_CPU_CORES; core++) {
-        auto threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
-        for (auto thread : threads) {
+    for (u32 core = 0; core < Core::NUM_CPU_CORES; core++) {
+        const auto& threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
+        for (auto& thread : threads) {
             if (thread->GetThreadId() == id) {
-                current_thread = thread.get();
-                return current_thread;
+                current_core = core;
+                return thread.get();
             }
         }
     }
@@ -197,6 +227,8 @@ static u64 RegRead(int id, Kernel::Thread* thread = nullptr) {
         return thread->context.pc;
     } else if (id == CPSR_REGISTER) {
         return thread->context.cpsr;
+    } else if (id > CPSR_REGISTER && id < FPSCR_REGISTER) {
+        return thread->context.fpu_registers[id - UC_ARM64_REG_Q0][0];
     } else {
         return 0;
     }
@@ -215,6 +247,8 @@ static void RegWrite(int id, u64 val, Kernel::Thread* thread = nullptr) {
         thread->context.pc = val;
     } else if (id == CPSR_REGISTER) {
         thread->context.cpsr = val;
+    } else if (id > CPSR_REGISTER && id < FPSCR_REGISTER) {
+        thread->context.fpu_registers[id - (CPSR_REGISTER + 1)][0] = val;
     }
 }
 
@@ -534,7 +568,11 @@ static void HandleQuery() {
         SendReply("T0");
     } else if (strncmp(query, "Supported", strlen("Supported")) == 0) {
         // PacketSize needs to be large enough for target xml
-        SendReply("PacketSize=2000;qXfer:features:read+");
+        std::string buffer = "PacketSize=2000;qXfer:features:read+;qXfer:threads:read+";
+        if (!modules.empty()) {
+            buffer += ";qXfer:libraries:read+";
+        }
+        SendReply(buffer.c_str());
     } else if (strncmp(query, "Xfer:features:read:target.xml:",
                        strlen("Xfer:features:read:target.xml:")) == 0) {
         SendReply(target_xml);
@@ -543,9 +581,9 @@ static void HandleQuery() {
         SendReply(buffer.c_str());
     } else if (strncmp(query, "fThreadInfo", strlen("fThreadInfo")) == 0) {
         std::string val = "m";
-        for (int core = 0; core < Core::NUM_CPU_CORES; core++) {
-            auto threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
-            for (auto thread : threads) {
+        for (u32 core = 0; core < Core::NUM_CPU_CORES; core++) {
+            const auto& threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
+            for (const auto& thread : threads) {
                 val += fmt::format("{:x}", thread->GetThreadId());
                 val += ",";
             }
@@ -554,6 +592,31 @@ static void HandleQuery() {
         SendReply(val.c_str());
     } else if (strncmp(query, "sThreadInfo", strlen("sThreadInfo")) == 0) {
         SendReply("l");
+    } else if (strncmp(query, "Xfer:threads:read", strlen("Xfer:threads:read")) == 0) {
+        std::string buffer;
+        buffer += "l<?xml version=\"1.0\"?>";
+        buffer += "<threads>";
+        for (u32 core = 0; core < Core::NUM_CPU_CORES; core++) {
+            const auto& threads = Core::System::GetInstance().Scheduler(core)->GetThreadList();
+            for (const auto& thread : threads) {
+                buffer +=
+                    fmt::format(R"*(<thread id="{:x}" core="{:d}" name="Thread {:x}"></thread>)*",
+                                thread->GetThreadId(), core, thread->GetThreadId());
+            }
+        }
+        buffer += "</threads>";
+        SendReply(buffer.c_str());
+    } else if (strncmp(query, "Xfer:libraries:read", strlen("Xfer:libraries:read")) == 0) {
+        std::string buffer;
+        buffer += "l<?xml version=\"1.0\"?>";
+        buffer += "<library-list>";
+        for (const auto& module : modules) {
+            buffer +=
+                fmt::format(R"*("<library name = "{}"><segment address = "0x{:x}"/></library>)*",
+                            module.name, module.beg);
+        }
+        buffer += "</library-list>";
+        SendReply(buffer.c_str());
     } else {
         SendReply("");
     }
@@ -561,33 +624,27 @@ static void HandleQuery() {
 
 /// Handle set thread command from gdb client.
 static void HandleSetThread() {
-    if (memcmp(command_buffer, "Hc", 2) == 0 || memcmp(command_buffer, "Hg", 2) == 0) {
-        int thread_id = -1;
-        if (command_buffer[2] != '-') {
-            thread_id = static_cast<int>(HexToInt(
-                command_buffer + 2,
-                command_length - 2 /*strlen(reinterpret_cast<char*>(command_buffer) + 2)*/));
-        }
-        if (thread_id >= 1) {
-            current_thread = FindThreadById(thread_id);
-        }
-        if (!current_thread) {
-            thread_id = 1;
-            current_thread = FindThreadById(thread_id);
-        }
-        if (current_thread) {
-            SendReply("OK");
-            return;
-        }
+    int thread_id = -1;
+    if (command_buffer[2] != '-') {
+        thread_id = static_cast<int>(HexToInt(command_buffer + 2, command_length - 2));
+    }
+    if (thread_id >= 1) {
+        current_thread = FindThreadById(thread_id);
+    }
+    if (!current_thread) {
+        thread_id = 1;
+        current_thread = FindThreadById(thread_id);
+    }
+    if (current_thread) {
+        SendReply("OK");
+        return;
     }
     SendReply("E01");
 }
 
 /// Handle thread alive command from gdb client.
 static void HandleThreadAlive() {
-    int thread_id = static_cast<int>(
-        HexToInt(command_buffer + 1,
-                 command_length - 1 /*strlen(reinterpret_cast<char*>(command_buffer) + 1)*/));
+    int thread_id = static_cast<int>(HexToInt(command_buffer + 1, command_length - 1));
     if (thread_id == 0) {
         thread_id = 1;
     }
@@ -610,16 +667,23 @@ static void SendSignal(Kernel::Thread* thread, u32 signal, bool full = true) {
 
     latest_signal = signal;
 
-    std::string buffer;
-    if (full) {
-        buffer = fmt::format("T{:02x}{:02x}:{:016x};{:02x}:{:016x};", latest_signal, PC_REGISTER,
-                             Common::swap64(RegRead(PC_REGISTER, thread)), SP_REGISTER,
-                             Common::swap64(RegRead(SP_REGISTER, thread)));
-    } else {
-        buffer = fmt::format("T{:02x};", latest_signal);
+    if (!thread) {
+        full = false;
     }
 
-    buffer += fmt::format("thread:{:x};", thread->GetThreadId());
+    std::string buffer;
+    if (full) {
+        buffer = fmt::format("T{:02x}{:02x}:{:016x};{:02x}:{:016x};{:02x}:{:016x}", latest_signal,
+                             PC_REGISTER, Common::swap64(RegRead(PC_REGISTER, thread)), SP_REGISTER,
+                             Common::swap64(RegRead(SP_REGISTER, thread)), LR_REGISTER,
+                             Common::swap64(RegRead(LR_REGISTER, thread)));
+    } else {
+        buffer = fmt::format("T{:02x}", latest_signal);
+    }
+
+    if (thread) {
+        buffer += fmt::format(";thread:{:x};", thread->GetThreadId());
+    }
 
     SendReply(buffer.c_str());
 }
@@ -711,8 +775,12 @@ static void ReadRegister() {
         LongToGdbHex(reply, RegRead(id, current_thread));
     } else if (id == CPSR_REGISTER) {
         IntToGdbHex(reply, (u32)RegRead(id, current_thread));
+    } else if (id >= UC_ARM64_REG_Q0 && id < FPSCR_REGISTER) {
+        LongToGdbHex(reply, RegRead(id, current_thread));
+    } else if (id == FPSCR_REGISTER) {
+        LongToGdbHex(reply, RegRead(TODO_DUMMY_REG_998, current_thread));
     } else {
-        return SendReply("E01");
+        LongToGdbHex(reply, RegRead(TODO_DUMMY_REG_997, current_thread));
     }
 
     SendReply(reinterpret_cast<char*>(reply));
@@ -729,7 +797,7 @@ static void ReadRegisters() {
         LongToGdbHex(bufptr + reg * 16, RegRead(reg, current_thread));
     }
 
-    bufptr += (32 * 16);
+    bufptr += 32 * 16;
 
     LongToGdbHex(bufptr, RegRead(PC_REGISTER, current_thread));
 
@@ -739,6 +807,16 @@ static void ReadRegisters() {
 
     bufptr += 8;
 
+    for (int reg = UC_ARM64_REG_Q0; reg <= UC_ARM64_REG_Q0 + 31; reg++) {
+        LongToGdbHex(bufptr + reg * 16, RegRead(reg, current_thread));
+    }
+
+    bufptr += 32 * 32;
+
+    LongToGdbHex(bufptr, RegRead(TODO_DUMMY_REG_998, current_thread));
+
+    bufptr += 8;
+
     SendReply(reinterpret_cast<char*>(buffer));
 }
 
@@ -759,10 +837,17 @@ static void WriteRegister() {
         RegWrite(id, GdbHexToLong(buffer_ptr), current_thread);
     } else if (id == CPSR_REGISTER) {
         RegWrite(id, GdbHexToInt(buffer_ptr), current_thread);
+    } else if (id >= UC_ARM64_REG_Q0 && id < FPSCR_REGISTER) {
+        RegWrite(id, GdbHexToLong(buffer_ptr), current_thread);
+    } else if (id == FPSCR_REGISTER) {
+        RegWrite(TODO_DUMMY_REG_998, GdbHexToLong(buffer_ptr), current_thread);
     } else {
-        return SendReply("E01");
+        RegWrite(TODO_DUMMY_REG_997, GdbHexToLong(buffer_ptr), current_thread);
     }
 
+    // Update Unicorn context skipping scheduler, no running threads at this point
+    Core::System::GetInstance().ArmInterface(current_core).LoadContext(current_thread->context);
+
     SendReply("OK");
 }
 
@@ -773,18 +858,25 @@ static void WriteRegisters() {
     if (command_buffer[0] != 'G')
         return SendReply("E01");
 
-    for (int i = 0, reg = 0; reg <= CPSR_REGISTER; i++, reg++) {
+    for (int i = 0, reg = 0; reg <= FPSCR_REGISTER; i++, reg++) {
         if (reg <= SP_REGISTER) {
             RegWrite(reg, GdbHexToLong(buffer_ptr + i * 16), current_thread);
         } else if (reg == PC_REGISTER) {
             RegWrite(PC_REGISTER, GdbHexToLong(buffer_ptr + i * 16), current_thread);
         } else if (reg == CPSR_REGISTER) {
             RegWrite(CPSR_REGISTER, GdbHexToInt(buffer_ptr + i * 16), current_thread);
+        } else if (reg >= UC_ARM64_REG_Q0 && reg < FPSCR_REGISTER) {
+            RegWrite(reg, GdbHexToLong(buffer_ptr + i * 16), current_thread);
+        } else if (reg == FPSCR_REGISTER) {
+            RegWrite(TODO_DUMMY_REG_998, GdbHexToLong(buffer_ptr + i * 16), current_thread);
         } else {
             UNIMPLEMENTED();
         }
     }
 
+    // Update Unicorn context skipping scheduler, no running threads at this point
+    Core::System::GetInstance().ArmInterface(current_core).LoadContext(current_thread->context);
+
     SendReply("OK");
 }
 
@@ -806,6 +898,10 @@ static void ReadMemory() {
         SendReply("E01");
     }
 
+    if (addr < Memory::PROCESS_IMAGE_VADDR || addr >= Memory::MAP_REGION_VADDR_END) {
+        return SendReply("E00");
+    }
+
     if (!Memory::IsValidVirtualAddress(addr)) {
         return SendReply("E00");
     }
@@ -840,16 +936,18 @@ static void WriteMemory() {
 }
 
 void Break(bool is_memory_break) {
-    if (!halt_loop) {
-        halt_loop = true;
-        send_trap = true;
-    }
+    send_trap = true;
 
     memory_break = is_memory_break;
 }
 
 /// Tell the CPU that it should perform a single step.
 static void Step() {
+    if (command_length > 1) {
+        RegWrite(PC_REGISTER, GdbHexToLong(command_buffer + 1), current_thread);
+        // Update Unicorn context skipping scheduler, no running threads at this point
+        Core::System::GetInstance().ArmInterface(current_core).LoadContext(current_thread->context);
+    }
     step_loop = true;
     halt_loop = true;
     send_trap = true;
@@ -1090,6 +1188,8 @@ static void Init(u16 port) {
     breakpoints_read.clear();
     breakpoints_write.clear();
 
+    modules.clear();
+
     // Start gdb server
     LOG_INFO(Debug_GDBStub, "Starting GDB server on port {}...", port);
 
@@ -1192,8 +1292,12 @@ void SetCpuStepFlag(bool is_step) {
 
 void SendTrap(Kernel::Thread* thread, int trap) {
     if (send_trap) {
+        if (!halt_loop || current_thread == thread) {
+            current_thread = thread;
+            SendSignal(thread, trap);
+        }
+        halt_loop = true;
         send_trap = false;
-        SendSignal(thread, trap);
     }
 }
 }; // namespace GDBStub

src/core/gdbstub/gdbstub.h

@@ -6,6 +6,7 @@
 
 #pragma once
 
+#include <string>
 #include "common/common_types.h"
 #include "core/hle/kernel/thread.h"
 
@@ -51,6 +52,9 @@ bool IsServerEnabled();
 /// Returns true if there is an active socket connection.
 bool IsConnected();
 
+/// Register module.
+void RegisterModule(std::string name, PAddr beg, PAddr end, bool add_elf_ext = true);
+
 /**
  * Signal to the gdbstub server that it should halt CPU execution.
  *
@@ -80,10 +84,10 @@ BreakpointAddress GetNextBreakpointFromAddress(PAddr addr, GDBStub::BreakpointTy
  */
 bool CheckBreakpoint(PAddr addr, GDBStub::BreakpointType type);
 
-// If set to true, the CPU will halt at the beginning of the next CPU loop.
+/// If set to true, the CPU will halt at the beginning of the next CPU loop.
 bool GetCpuHaltFlag();
 
-// If set to true and the CPU is halted, the CPU will step one instruction.
+/// If set to true and the CPU is halted, the CPU will step one instruction.
 bool GetCpuStepFlag();
 
 /**

src/core/loader/deconstructed_rom_directory.cpp

@@ -9,6 +9,7 @@
 #include "common/logging/log.h"
 #include "common/string_util.h"
 #include "core/file_sys/romfs_factory.h"
+#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/hle/service/filesystem/filesystem.h"
@@ -133,6 +134,8 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(
         next_load_addr = AppLoader_NSO::LoadModule(path, load_addr);
         if (next_load_addr) {
             LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", module, load_addr);
+            // Register module with GDBStub
+            GDBStub::RegisterModule(module, load_addr, next_load_addr - 1, false);
         } else {
             next_load_addr = load_addr;
         }

src/core/loader/nca.cpp

@@ -7,10 +7,12 @@
 #include "common/common_funcs.h"
 #include "common/file_util.h"
 #include "common/logging/log.h"
+#include "common/string_util.h"
 #include "common/swap.h"
 #include "core/core.h"
 #include "core/file_sys/program_metadata.h"
 #include "core/file_sys/romfs_factory.h"
+#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/hle/service/filesystem/filesystem.h"
@@ -259,6 +261,8 @@ ResultStatus AppLoader_NCA::Load(Kernel::SharedPtr<Kernel::Process>& process) {
         next_load_addr = AppLoader_NSO::LoadModule(module, nca->GetExeFsFile(module), load_addr);
         if (next_load_addr) {
             LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", module, load_addr);
+            // Register module with GDBStub
+            GDBStub::RegisterModule(module, load_addr, next_load_addr - 1, false);
         } else {
             next_load_addr = load_addr;
         }

src/core/loader/nro.cpp

@@ -9,6 +9,7 @@
 #include "common/logging/log.h"
 #include "common/swap.h"
 #include "core/core.h"
+#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/loader/nro.h"
@@ -115,6 +116,9 @@ bool AppLoader_NRO::LoadNro(const std::string& path, VAddr load_base) {
     codeset->memory = std::make_shared<std::vector<u8>>(std::move(program_image));
     Core::CurrentProcess()->LoadModule(codeset, load_base);
 
+    // Register module with GDBStub
+    GDBStub::RegisterModule(codeset->name, load_base, load_base);
+
     return true;
 }
 

src/core/loader/nso.cpp

@@ -10,6 +10,7 @@
 #include "common/logging/log.h"
 #include "common/swap.h"
 #include "core/core.h"
+#include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/loader/nso.h"
@@ -147,6 +148,9 @@ VAddr AppLoader_NSO::LoadModule(const std::string& name, const std::vector<u8>&
     codeset->memory = std::make_shared<std::vector<u8>>(std::move(program_image));
     Core::CurrentProcess()->LoadModule(codeset, load_base);
 
+    // Register module with GDBStub
+    GDBStub::RegisterModule(codeset->name, load_base, load_base);
+
     return load_base + image_size;
 }
 

src/video_core/engines/maxwell_3d.cpp

@@ -398,27 +398,6 @@ u32 Maxwell3D::GetRegisterValue(u32 method) const {
     return regs.reg_array[method];
 }
 
-bool Maxwell3D::IsShaderStageEnabled(Regs::ShaderStage stage) const {
-    // The Vertex stage is always enabled.
-    if (stage == Regs::ShaderStage::Vertex)
-        return true;
-
-    switch (stage) {
-    case Regs::ShaderStage::TesselationControl:
-        return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::TesselationControl)]
-                   .enable != 0;
-    case Regs::ShaderStage::TesselationEval:
-        return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::TesselationEval)]
-                   .enable != 0;
-    case Regs::ShaderStage::Geometry:
-        return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::Geometry)].enable != 0;
-    case Regs::ShaderStage::Fragment:
-        return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::Fragment)].enable != 0;
-    }
-
-    UNREACHABLE();
-}
-
 void Maxwell3D::ProcessClearBuffers() {
     ASSERT(regs.clear_buffers.R == regs.clear_buffers.G &&
            regs.clear_buffers.R == regs.clear_buffers.B &&

src/video_core/engines/maxwell_3d.h

@@ -379,6 +379,14 @@ public:
             }
         };
 
+        bool IsShaderConfigEnabled(size_t index) const {
+            // The VertexB is always enabled.
+            if (index == static_cast<size_t>(Regs::ShaderProgram::VertexB)) {
+                return true;
+            }
+            return shader_config[index].enable != 0;
+        }
+
         union {
             struct {
                 INSERT_PADDING_WORDS(0x45);
@@ -780,9 +788,6 @@ public:
     /// Returns the texture information for a specific texture in a specific shader stage.
     Texture::FullTextureInfo GetStageTexture(Regs::ShaderStage stage, size_t offset) const;
 
-    /// Returns whether the specified shader stage is enabled or not.
-    bool IsShaderStageEnabled(Regs::ShaderStage stage) const;
-
 private:
     std::unordered_map<u32, std::vector<u32>> uploaded_macros;
 

src/video_core/engines/shader_bytecode.h

@@ -142,6 +142,7 @@ enum class PredCondition : u64 {
     GreaterThan = 4,
     NotEqual = 5,
     GreaterEqual = 6,
+    LessThanWithNan = 9,
     NotEqualWithNan = 13,
     // TODO(Subv): Other condition types
 };
@@ -201,6 +202,11 @@ enum class IMinMaxExchange : u64 {
     XHi = 3,
 };
 
+enum class FlowCondition : u64 {
+    Always = 0xF,
+    Fcsm_Tr = 0x1C, // TODO(bunnei): What is this used for?
+};
+
 union Instruction {
     Instruction& operator=(const Instruction& instr) {
         value = instr.value;
@@ -315,6 +321,10 @@ union Instruction {
         }
     } bfe;
 
+    union {
+        BitField<0, 5, FlowCondition> cond;
+    } flow;
+
     union {
         BitField<48, 1, u64> negate_b;
         BitField<49, 1, u64> negate_c;

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -181,6 +181,19 @@ std::pair<u8*, GLintptr> RasterizerOpenGL::SetupVertexArrays(u8* array_ptr,
     return {array_ptr, buffer_offset};
 }
 
+static GLShader::ProgramCode GetShaderProgramCode(Maxwell::ShaderProgram program) {
+    auto& gpu = Core::System().GetInstance().GPU().Maxwell3D();
+
+    // Fetch program code from memory
+    GLShader::ProgramCode program_code;
+    auto& shader_config = gpu.regs.shader_config[static_cast<size_t>(program)];
+    const u64 gpu_address{gpu.regs.code_address.CodeAddress() + shader_config.offset};
+    const boost::optional<VAddr> cpu_address{gpu.memory_manager.GpuToCpuAddress(gpu_address)};
+    Memory::ReadBlock(*cpu_address, program_code.data(), program_code.size() * sizeof(u64));
+
+    return program_code;
+}
+
 void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
     // Helper function for uploading uniform data
     const auto copy_buffer = [&](GLuint handle, GLintptr offset, GLsizeiptr size) {
@@ -193,26 +206,23 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
     };
 
     auto& gpu = Core::System().GetInstance().GPU().Maxwell3D();
-    ASSERT_MSG(!gpu.regs.shader_config[0].enable, "VertexA is unsupported!");
 
     // Next available bindpoints to use when uploading the const buffers and textures to the GLSL
     // shaders. The constbuffer bindpoint starts after the shader stage configuration bind points.
     u32 current_constbuffer_bindpoint = uniform_buffers.size();
     u32 current_texture_bindpoint = 0;
 
-    for (unsigned index = 1; index < Maxwell::MaxShaderProgram; ++index) {
+    for (size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) {
         auto& shader_config = gpu.regs.shader_config[index];
         const Maxwell::ShaderProgram program{static_cast<Maxwell::ShaderProgram>(index)};
 
-        const auto& stage = index - 1; // Stage indices are 0 - 5
-
-        const bool is_enabled = gpu.IsShaderStageEnabled(static_cast<Maxwell::ShaderStage>(stage));
-
         // Skip stages that are not enabled
-        if (!is_enabled) {
+        if (!gpu.regs.IsShaderConfigEnabled(index)) {
             continue;
         }
 
+        const size_t stage{index == 0 ? 0 : index - 1}; // Stage indices are 0 - 5
+
         GLShader::MaxwellUniformData ubo{};
         ubo.SetFromRegs(gpu.state.shader_stages[stage]);
         std::memcpy(buffer_ptr, &ubo, sizeof(ubo));
@@ -228,16 +238,21 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
         buffer_ptr += sizeof(GLShader::MaxwellUniformData);
         buffer_offset += sizeof(GLShader::MaxwellUniformData);
 
-        // Fetch program code from memory
-        GLShader::ProgramCode program_code;
-        const u64 gpu_address{gpu.regs.code_address.CodeAddress() + shader_config.offset};
-        const boost::optional<VAddr> cpu_address{gpu.memory_manager.GpuToCpuAddress(gpu_address)};
-        Memory::ReadBlock(*cpu_address, program_code.data(), program_code.size() * sizeof(u64));
-        GLShader::ShaderSetup setup{std::move(program_code)};
-
+        GLShader::ShaderSetup setup{GetShaderProgramCode(program)};
         GLShader::ShaderEntries shader_resources;
 
         switch (program) {
+        case Maxwell::ShaderProgram::VertexA: {
+            // VertexB is always enabled, so when VertexA is enabled, we have two vertex shaders.
+            // Conventional HW does not support this, so we combine VertexA and VertexB into one
+            // stage here.
+            setup.SetProgramB(GetShaderProgramCode(Maxwell::ShaderProgram::VertexB));
+            GLShader::MaxwellVSConfig vs_config{setup};
+            shader_resources =
+                shader_program_manager->UseProgrammableVertexShader(vs_config, setup);
+            break;
+        }
+
         case Maxwell::ShaderProgram::VertexB: {
             GLShader::MaxwellVSConfig vs_config{setup};
             shader_resources =
@@ -268,6 +283,12 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
         current_texture_bindpoint =
             SetupTextures(static_cast<Maxwell::ShaderStage>(stage), gl_stage_program,
                           current_texture_bindpoint, shader_resources.texture_samplers);
+
+        // When VertexA is enabled, we have dual vertex shaders
+        if (program == Maxwell::ShaderProgram::VertexA) {
+            // VertexB was combined with VertexA, so we skip the VertexB iteration
+            index++;
+        }
     }
 
     shader_program_manager->UseTrivialGeometryShader();
@@ -301,9 +322,6 @@ std::pair<Surface, Surface> RasterizerOpenGL::ConfigureFramebuffers(bool using_c
                                                                     bool using_depth_fb) {
     const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
 
-    // Sync the depth test state before configuring the framebuffer surfaces.
-    SyncDepthTestState();
-
     // TODO(bunnei): Implement this
     const bool has_stencil = false;
 
@@ -368,6 +386,13 @@ void RasterizerOpenGL::Clear() {
     if (regs.clear_buffers.Z) {
         clear_mask |= GL_DEPTH_BUFFER_BIT;
         use_depth_fb = true;
+
+        // Always enable the depth write when clearing the depth buffer. The depth write mask is
+        // ignored when clearing the buffer in the Switch, but OpenGL obeys it so we set it to true.
+        state.depth.test_enabled = true;
+        state.depth.write_mask = GL_TRUE;
+        state.depth.test_func = GL_ALWAYS;
+        state.Apply();
     }
 
     if (clear_mask == 0)
@@ -402,6 +427,7 @@ void RasterizerOpenGL::DrawArrays() {
     auto [dirty_color_surface, dirty_depth_surface] =
         ConfigureFramebuffers(true, regs.zeta.Address() != 0);
 
+    SyncDepthTestState();
     SyncBlendState();
     SyncCullMode();
 
@@ -605,9 +631,6 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, GLuint progr
     auto& gpu = Core::System::GetInstance().GPU();
     auto& maxwell3d = gpu.Get3DEngine();
 
-    ASSERT_MSG(maxwell3d.IsShaderStageEnabled(stage),
-               "Attempted to upload constbuffer of disabled shader stage");
-
     // Reset all buffer draw state for this stage.
     for (auto& buffer : state.draw.const_buffers[static_cast<size_t>(stage)]) {
         buffer.bindpoint = 0;
@@ -674,9 +697,6 @@ u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, GLuint program,
     auto& gpu = Core::System::GetInstance().GPU();
     auto& maxwell3d = gpu.Get3DEngine();
 
-    ASSERT_MSG(maxwell3d.IsShaderStageEnabled(stage),
-               "Attempted to upload textures of disabled shader stage");
-
     ASSERT_MSG(current_unit + entries.size() <= std::size(state.texture_units),
                "Exceeded the number of active textures.");
 

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -42,13 +42,14 @@ enum class ExitMethod {
 struct Subroutine {
     /// Generates a name suitable for GLSL source code.
     std::string GetName() const {
-        return "sub_" + std::to_string(begin) + '_' + std::to_string(end);
+        return "sub_" + std::to_string(begin) + '_' + std::to_string(end) + '_' + suffix;
     }
 
-    u32 begin;              ///< Entry point of the subroutine.
-    u32 end;                ///< Return point of the subroutine.
-    ExitMethod exit_method; ///< Exit method of the subroutine.
-    std::set<u32> labels;   ///< Addresses refereced by JMP instructions.
+    u32 begin;                 ///< Entry point of the subroutine.
+    u32 end;                   ///< Return point of the subroutine.
+    const std::string& suffix; ///< Suffix of the shader, used to make a unique subroutine name
+    ExitMethod exit_method;    ///< Exit method of the subroutine.
+    std::set<u32> labels;      ///< Addresses refereced by JMP instructions.
 
     bool operator<(const Subroutine& rhs) const {
         return std::tie(begin, end) < std::tie(rhs.begin, rhs.end);
@@ -58,11 +59,11 @@ struct Subroutine {
 /// Analyzes shader code and produces a set of subroutines.
 class ControlFlowAnalyzer {
 public:
-    ControlFlowAnalyzer(const ProgramCode& program_code, u32 main_offset)
+    ControlFlowAnalyzer(const ProgramCode& program_code, u32 main_offset, const std::string& suffix)
         : program_code(program_code) {
 
         // Recursively finds all subroutines.
-        const Subroutine& program_main = AddSubroutine(main_offset, PROGRAM_END);
+        const Subroutine& program_main = AddSubroutine(main_offset, PROGRAM_END, suffix);
         if (program_main.exit_method != ExitMethod::AlwaysEnd)
             throw DecompileFail("Program does not always end");
     }
@@ -77,12 +78,12 @@ private:
     std::map<std::pair<u32, u32>, ExitMethod> exit_method_map;
 
     /// Adds and analyzes a new subroutine if it is not added yet.
-    const Subroutine& AddSubroutine(u32 begin, u32 end) {
-        auto iter = subroutines.find(Subroutine{begin, end});
+    const Subroutine& AddSubroutine(u32 begin, u32 end, const std::string& suffix) {
+        auto iter = subroutines.find(Subroutine{begin, end, suffix});
         if (iter != subroutines.end())
             return *iter;
 
-        Subroutine subroutine{begin, end};
+        Subroutine subroutine{begin, end, suffix};
         subroutine.exit_method = Scan(begin, end, subroutine.labels);
         if (subroutine.exit_method == ExitMethod::Undetermined)
             throw DecompileFail("Recursive function detected");
@@ -191,7 +192,8 @@ public:
         UnsignedInteger,
     };
 
-    GLSLRegister(size_t index, ShaderWriter& shader) : index{index}, shader{shader} {}
+    GLSLRegister(size_t index, ShaderWriter& shader, const std::string& suffix)
+        : index{index}, shader{shader}, suffix{suffix} {}
 
     /// Gets the GLSL type string for a register
     static std::string GetTypeString(Type type) {
@@ -216,7 +218,7 @@ public:
     /// Returns a GLSL string representing the current state of the register
     const std::string GetActiveString() {
         declr_type.insert(active_type);
-        return GetPrefixString(active_type) + std::to_string(index);
+        return GetPrefixString(active_type) + std::to_string(index) + '_' + suffix;
     }
 
     /// Returns true if the active type is a float
@@ -251,6 +253,7 @@ private:
     ShaderWriter& shader;
     Type active_type{Type::Float};
     std::set<Type> declr_type;
+    const std::string& suffix;
 };
 
 /**
@@ -262,8 +265,8 @@ private:
 class GLSLRegisterManager {
 public:
     GLSLRegisterManager(ShaderWriter& shader, ShaderWriter& declarations,
-                        const Maxwell3D::Regs::ShaderStage& stage)
-        : shader{shader}, declarations{declarations}, stage{stage} {
+                        const Maxwell3D::Regs::ShaderStage& stage, const std::string& suffix)
+        : shader{shader}, declarations{declarations}, stage{stage}, suffix{suffix} {
         BuildRegisterList();
     }
 
@@ -430,12 +433,12 @@ public:
     }
 
     /// Add declarations for registers
-    void GenerateDeclarations() {
+    void GenerateDeclarations(const std::string& suffix) {
         for (const auto& reg : regs) {
             for (const auto& type : reg.DeclaredTypes()) {
                 declarations.AddLine(GLSLRegister::GetTypeString(type) + ' ' +
-                                     GLSLRegister::GetPrefixString(type) +
-                                     std::to_string(reg.GetIndex()) + " = 0;");
+                                     reg.GetPrefixString(type) + std::to_string(reg.GetIndex()) +
+                                     '_' + suffix + " = 0;");
             }
         }
         declarations.AddNewLine();
@@ -558,7 +561,7 @@ private:
     /// Build the GLSL register list.
     void BuildRegisterList() {
         for (size_t index = 0; index < Register::NumRegisters; ++index) {
-            regs.emplace_back(index, shader);
+            regs.emplace_back(index, shader, suffix);
         }
     }
 
@@ -620,16 +623,17 @@ private:
     std::array<ConstBufferEntry, Maxwell3D::Regs::MaxConstBuffers> declr_const_buffers;
     std::vector<SamplerEntry> used_samplers;
     const Maxwell3D::Regs::ShaderStage& stage;
+    const std::string& suffix;
 };
 
 class GLSLGenerator {
 public:
     GLSLGenerator(const std::set<Subroutine>& subroutines, const ProgramCode& program_code,
-                  u32 main_offset, Maxwell3D::Regs::ShaderStage stage)
+                  u32 main_offset, Maxwell3D::Regs::ShaderStage stage, const std::string& suffix)
         : subroutines(subroutines), program_code(program_code), main_offset(main_offset),
-          stage(stage) {
+          stage(stage), suffix(suffix) {
 
-        Generate();
+        Generate(suffix);
     }
 
     std::string GetShaderCode() {
@@ -644,7 +648,7 @@ public:
 private:
     /// Gets the Subroutine object corresponding to the specified address.
     const Subroutine& GetSubroutine(u32 begin, u32 end) const {
-        auto iter = subroutines.find(Subroutine{begin, end});
+        auto iter = subroutines.find(Subroutine{begin, end, suffix});
         ASSERT(iter != subroutines.end());
         return *iter;
     }
@@ -689,7 +693,7 @@ private:
         // Can't assign to the constant predicate.
         ASSERT(pred != static_cast<u64>(Pred::UnusedIndex));
 
-        std::string variable = 'p' + std::to_string(pred);
+        std::string variable = 'p' + std::to_string(pred) + '_' + suffix;
         shader.AddLine(variable + " = " + value + ';');
         declr_predicates.insert(std::move(variable));
     }
@@ -707,7 +711,7 @@ private:
         if (index == static_cast<u64>(Pred::UnusedIndex))
             variable = "true";
         else
-            variable = 'p' + std::to_string(index);
+            variable = 'p' + std::to_string(index) + '_' + suffix;
 
         if (negate) {
             return "!(" + variable + ')';
@@ -728,10 +732,10 @@ private:
                                        const std::string& op_a, const std::string& op_b) const {
         using Tegra::Shader::PredCondition;
         static const std::unordered_map<PredCondition, const char*> PredicateComparisonStrings = {
-            {PredCondition::LessThan, "<"},         {PredCondition::Equal, "=="},
-            {PredCondition::LessEqual, "<="},       {PredCondition::GreaterThan, ">"},
-            {PredCondition::NotEqual, "!="},        {PredCondition::GreaterEqual, ">="},
-            {PredCondition::NotEqualWithNan, "!="},
+            {PredCondition::LessThan, "<"},        {PredCondition::Equal, "=="},
+            {PredCondition::LessEqual, "<="},      {PredCondition::GreaterThan, ">"},
+            {PredCondition::NotEqual, "!="},       {PredCondition::GreaterEqual, ">="},
+            {PredCondition::LessThanWithNan, "<"}, {PredCondition::NotEqualWithNan, "!="},
         };
 
         const auto& comparison{PredicateComparisonStrings.find(condition)};
@@ -739,7 +743,8 @@ private:
                    "Unknown predicate comparison operation");
 
         std::string predicate{'(' + op_a + ") " + comparison->second + " (" + op_b + ')'};
-        if (condition == PredCondition::NotEqualWithNan) {
+        if (condition == PredCondition::LessThanWithNan ||
+            condition == PredCondition::NotEqualWithNan) {
             predicate += " || isnan(" + op_a + ") || isnan(" + op_b + ')';
         }
 
@@ -1639,16 +1644,32 @@ private:
                     shader.AddLine("color.a = " + regs.GetRegisterAsFloat(3) + ';');
                 }
 
-                shader.AddLine("return true;");
-                if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
-                    // If this is an unconditional exit then just end processing here, otherwise
-                    // we have to account for the possibility of the condition not being met, so
-                    // continue processing the next instruction.
-                    offset = PROGRAM_END - 1;
+                switch (instr.flow.cond) {
+                case Tegra::Shader::FlowCondition::Always:
+                    shader.AddLine("return true;");
+                    if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
+                        // If this is an unconditional exit then just end processing here,
+                        // otherwise we have to account for the possibility of the condition
+                        // not being met, so continue processing the next instruction.
+                        offset = PROGRAM_END - 1;
+                    }
+                    break;
+
+                case Tegra::Shader::FlowCondition::Fcsm_Tr:
+                    // TODO(bunnei): What is this used for? If we assume this conditon is not
+                    // satisifed, dual vertex shaders in Farming Simulator make more sense
+                    LOG_CRITICAL(HW_GPU, "Skipping unknown FlowCondition::Fcsm_Tr");
+                    break;
+
+                default:
+                    LOG_CRITICAL(HW_GPU, "Unhandled flow condition: {}",
+                                 static_cast<u32>(instr.flow.cond.Value()));
+                    UNREACHABLE();
                 }
                 break;
             }
             case OpCode::Id::KIL: {
+                ASSERT(instr.flow.cond == Tegra::Shader::FlowCondition::Always);
                 shader.AddLine("discard;");
                 break;
             }
@@ -1669,8 +1690,9 @@ private:
                 // can ignore this when generating GLSL code.
                 break;
             }
-            case OpCode::Id::DEPBAR:
-            case OpCode::Id::SYNC: {
+            case OpCode::Id::SYNC:
+                ASSERT(instr.flow.cond == Tegra::Shader::FlowCondition::Always);
+            case OpCode::Id::DEPBAR: {
                 // TODO(Subv): Find out if we actually have to care about these instructions or if
                 // the GLSL compiler takes care of that for us.
                 LOG_WARNING(HW_GPU, "DEPBAR/SYNC instruction is stubbed");
@@ -1710,7 +1732,7 @@ private:
         return program_counter;
     }
 
-    void Generate() {
+    void Generate(const std::string& suffix) {
         // Add declarations for all subroutines
         for (const auto& subroutine : subroutines) {
             shader.AddLine("bool " + subroutine.GetName() + "();");
@@ -1718,7 +1740,7 @@ private:
         shader.AddNewLine();
 
         // Add the main entry point
-        shader.AddLine("bool exec_shader() {");
+        shader.AddLine("bool exec_" + suffix + "() {");
         ++shader.scope;
         CallSubroutine(GetSubroutine(main_offset, PROGRAM_END));
         --shader.scope;
@@ -1781,7 +1803,7 @@ private:
 
     /// Add declarations for registers
     void GenerateDeclarations() {
-        regs.GenerateDeclarations();
+        regs.GenerateDeclarations(suffix);
 
         for (const auto& pred : declr_predicates) {
             declarations.AddLine("bool " + pred + " = false;");
@@ -1794,27 +1816,30 @@ private:
     const ProgramCode& program_code;
     const u32 main_offset;
     Maxwell3D::Regs::ShaderStage stage;
+    const std::string& suffix;
 
     ShaderWriter shader;
     ShaderWriter declarations;
-    GLSLRegisterManager regs{shader, declarations, stage};
+    GLSLRegisterManager regs{shader, declarations, stage, suffix};
 
     // Declarations
     std::set<std::string> declr_predicates;
 }; // namespace Decompiler
 
 std::string GetCommonDeclarations() {
-    std::string declarations = "bool exec_shader();\n";
+    std::string declarations;
     declarations += "#define MAX_CONSTBUFFER_ELEMENTS " +
                     std::to_string(RasterizerOpenGL::MaxConstbufferSize / (sizeof(GLvec4)));
+    declarations += '\n';
     return declarations;
 }
 
 boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
-                                                Maxwell3D::Regs::ShaderStage stage) {
+                                                Maxwell3D::Regs::ShaderStage stage,
+                                                const std::string& suffix) {
     try {
-        auto subroutines = ControlFlowAnalyzer(program_code, main_offset).GetSubroutines();
-        GLSLGenerator generator(subroutines, program_code, main_offset, stage);
+        auto subroutines = ControlFlowAnalyzer(program_code, main_offset, suffix).GetSubroutines();
+        GLSLGenerator generator(subroutines, program_code, main_offset, stage, suffix);
         return ProgramResult{generator.GetShaderCode(), generator.GetEntries()};
     } catch (const DecompileFail& exception) {
         LOG_ERROR(HW_GPU, "Shader decompilation failed: {}", exception.what());

src/video_core/renderer_opengl/gl_shader_decompiler.h

@@ -20,7 +20,8 @@ using Tegra::Engines::Maxwell3D;
 std::string GetCommonDeclarations();
 
 boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
-                                                Maxwell3D::Regs::ShaderStage stage);
+                                                Maxwell3D::Regs::ShaderStage stage,
+                                                const std::string& suffix);
 
 } // namespace Decompiler
 } // namespace GLShader

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -17,10 +17,17 @@ ProgramResult GenerateVertexShader(const ShaderSetup& setup, const MaxwellVSConf
     std::string out = "#version 430 core\n";
     out += "#extension GL_ARB_separate_shader_objects : enable\n\n";
     out += Decompiler::GetCommonDeclarations();
+    out += "bool exec_vertex();\n";
+
+    if (setup.IsDualProgram()) {
+        out += "bool exec_vertex_b();\n";
+    }
+
+    ProgramResult program =
+        Decompiler::DecompileProgram(setup.program.code, PROGRAM_OFFSET,
+                                     Maxwell3D::Regs::ShaderStage::Vertex, "vertex")
+            .get_value_or({});
 
-    ProgramResult program = Decompiler::DecompileProgram(setup.program_code, PROGRAM_OFFSET,
-                                                         Maxwell3D::Regs::ShaderStage::Vertex)
-                                .get_value_or({});
     out += R"(
 
 out gl_PerVertex {
@@ -34,7 +41,14 @@ layout (std140) uniform vs_config {
 };
 
 void main() {
-    exec_shader();
+    exec_vertex();
+)";
+
+    if (setup.IsDualProgram()) {
+        out += "    exec_vertex_b();";
+    }
+
+    out += R"(
 
     // Viewport can be flipped, which is unsupported by glViewport
     position.xy *= viewport_flip.xy;
@@ -44,8 +58,19 @@ void main() {
     // For now, this is here to bring order in lieu of proper emulation
     position.w = 1.0;
 }
+
 )";
+
     out += program.first;
+
+    if (setup.IsDualProgram()) {
+        ProgramResult program_b =
+            Decompiler::DecompileProgram(setup.program.code_b, PROGRAM_OFFSET,
+                                         Maxwell3D::Regs::ShaderStage::Vertex, "vertex_b")
+                .get_value_or({});
+        out += program_b.first;
+    }
+
     return {out, program.second};
 }
 
@@ -53,12 +78,13 @@ ProgramResult GenerateFragmentShader(const ShaderSetup& setup, const MaxwellFSCo
     std::string out = "#version 430 core\n";
     out += "#extension GL_ARB_separate_shader_objects : enable\n\n";
     out += Decompiler::GetCommonDeclarations();
+    out += "bool exec_fragment();\n";
 
-    ProgramResult program = Decompiler::DecompileProgram(setup.program_code, PROGRAM_OFFSET,
-                                                         Maxwell3D::Regs::ShaderStage::Fragment)
-                                .get_value_or({});
+    ProgramResult program =
+        Decompiler::DecompileProgram(setup.program.code, PROGRAM_OFFSET,
+                                     Maxwell3D::Regs::ShaderStage::Fragment, "fragment")
+            .get_value_or({});
     out += R"(
-
 in vec4 position;
 out vec4 color;
 
@@ -67,7 +93,7 @@ layout (std140) uniform fs_config {
 };
 
 void main() {
-    exec_shader();
+    exec_fragment();
 }
 
 )";

src/video_core/renderer_opengl/gl_shader_gen.h

@@ -115,21 +115,48 @@ struct ShaderEntries {
 using ProgramResult = std::pair<std::string, ShaderEntries>;
 
 struct ShaderSetup {
-    ShaderSetup(ProgramCode&& program_code) : program_code(std::move(program_code)) {}
+    ShaderSetup(const ProgramCode& program_code) {
+        program.code = program_code;
+    }
+
+    struct {
+        ProgramCode code;
+        ProgramCode code_b; // Used for dual vertex shaders
+    } program;
 
-    ProgramCode program_code;
     bool program_code_hash_dirty = true;
 
     u64 GetProgramCodeHash() {
         if (program_code_hash_dirty) {
-            program_code_hash = Common::ComputeHash64(&program_code, sizeof(program_code));
+            program_code_hash = GetNewHash();
             program_code_hash_dirty = false;
         }
         return program_code_hash;
     }
 
+    /// Used in scenarios where we have a dual vertex shaders
+    void SetProgramB(const ProgramCode& program_b) {
+        program.code_b = program_b;
+        has_program_b = true;
+    }
+
+    bool IsDualProgram() const {
+        return has_program_b;
+    }
+
 private:
+    u64 GetNewHash() const {
+        if (has_program_b) {
+            // Compute hash over dual shader programs
+            return Common::ComputeHash64(&program, sizeof(program));
+        } else {
+            // Compute hash over a single shader program
+            return Common::ComputeHash64(&program.code, program.code.size());
+        }
+    }
+
     u64 program_code_hash{};
+    bool has_program_b{};
 };
 
 struct MaxwellShaderConfigCommon {