Service/Audio: update audren:u service

Build: Fixed some MSVC warnings in various parts of the code.

.travis.yml

@@ -42,3 +42,7 @@ notifications:
   webhooks:
     urls:
       - https://api.yuzu-emu.org/code/travis/notify
+
+cache:
+  directories:
+    - $HOME/.ccache

.travis/linux/build.sh

@@ -1,3 +1,3 @@
 #!/bin/bash -ex
 
-docker run -v $(pwd):/yuzu ubuntu:18.04 /bin/bash /yuzu/.travis/linux/docker.sh
+docker run -e CCACHE_DIR=/ccache -v $HOME/.ccache:/ccache -v $(pwd):/yuzu ubuntu:18.04 /bin/bash /yuzu/.travis/linux/docker.sh

.travis/linux/docker.sh

@@ -1,16 +1,18 @@
 #!/bin/bash -ex
 
 apt-get update
-apt-get install -y build-essential git libqt5opengl5-dev libsdl2-dev libssl-dev python qtbase5-dev wget
-
-# Get a recent version of CMake
-wget https://cmake.org/files/v3.10/cmake-3.10.1-Linux-x86_64.sh
-sh cmake-3.10.1-Linux-x86_64.sh --exclude-subdir --prefix=/ --skip-license
+apt-get install --no-install-recommends -y build-essential git libqt5opengl5-dev libsdl2-dev libssl-dev python qtbase5-dev wget cmake ninja-build ccache
 
 cd /yuzu
 
+export PATH=/usr/lib/ccache:$PATH
+ln -sf /usr/bin/ccache /usr/lib/ccache/cc
+ln -sf /usr/bin/ccache /usr/lib/ccache/c++
 mkdir build && cd build
-cmake .. -DYUZU_BUILD_UNICORN=ON -DCMAKE_BUILD_TYPE=Release
-make -j4
+ccache --show-stats > ccache_before
+cmake .. -DYUZU_BUILD_UNICORN=ON -DCMAKE_BUILD_TYPE=Release -G Ninja
+ninja
+ccache --show-stats > ccache_after
+diff -U100 ccache_before ccache_after || true
 
 ctest -VV -C Release

.travis/macos/build.sh

@@ -7,8 +7,12 @@ export Qt5_DIR=$(brew --prefix)/opt/qt5
 export UNICORNDIR=$(pwd)/externals/unicorn
 
 mkdir build && cd build
+export PATH=/usr/local/opt/ccache/libexec:$PATH
+ccache --show-stats > ccache_before
 cmake --version
 cmake .. -DYUZU_BUILD_UNICORN=ON -DCMAKE_BUILD_TYPE=Release
 make -j4
+ccache --show-stats > ccache_after
+diff -U100 ccache_before ccache_after || true
 
 ctest -VV -C Release

.travis/macos/deps.sh

@@ -1,5 +1,5 @@
 #!/bin/sh -ex
 
 brew update
-brew install dylibbundler p7zip qt5 sdl2
+brew install dylibbundler p7zip qt5 sdl2 ccache
 brew outdated cmake || brew upgrade cmake

appveyor.yml

@@ -53,7 +53,7 @@ build_script:
           # https://www.appveyor.com/docs/build-phase
           msbuild msvc_build/yuzu.sln /maxcpucount /logger:"C:\Program Files\AppVeyor\BuildAgent\Appveyor.MSBuildLogger.dll"
         } else {
-          C:\msys64\usr\bin\bash.exe -lc 'mingw32-make -j4 -C mingw_build/ 2>&1'
+          C:\msys64\usr\bin\bash.exe -lc 'mingw32-make -C mingw_build/ 2>&1'
         }
 
 after_build:

src/common/string_util.cpp

@@ -64,6 +64,10 @@ std::string ArrayToString(const u8* data, size_t size, int line_len, bool spaces
     return oss.str();
 }
 
+std::string StringFromBuffer(const std::vector<u8>& data) {
+    return std::string(data.begin(), std::find(data.begin(), data.end(), '\0'));
+}
+
 // Turns "  hej " into "hej". Also handles tabs.
 std::string StripSpaces(const std::string& str) {
     const size_t s = str.find_first_not_of(" \t\r\n");

src/common/string_util.h

@@ -21,6 +21,8 @@ std::string ToUpper(std::string str);
 
 std::string ArrayToString(const u8* data, size_t size, int line_len = 20, bool spaces = true);
 
+std::string StringFromBuffer(const std::vector<u8>& data);
+
 std::string StripSpaces(const std::string& s);
 std::string StripQuotes(const std::string& s);
 

src/core/arm/unicorn/arm_unicorn.cpp

@@ -35,6 +35,17 @@ LoadDll LoadDll::g_load_dll;
         }                                                                                          \
     } while (0)
 
+static void CodeHook(uc_engine* uc, uint64_t address, uint32_t size, void* user_data) {
+    GDBStub::BreakpointAddress bkpt =
+        GDBStub::GetNextBreakpointFromAddress(address, GDBStub::BreakpointType::Execute);
+    if (GDBStub::IsMemoryBreak() ||
+        (bkpt.type != GDBStub::BreakpointType::None && address == bkpt.address)) {
+        auto core = static_cast<ARM_Unicorn*>(user_data);
+        core->RecordBreak(bkpt);
+        uc_emu_stop(uc);
+    }
+}
+
 static void InterruptHook(uc_engine* uc, u32 intNo, void* user_data) {
     u32 esr{};
     CHECKED(uc_reg_read(uc, UC_ARM64_REG_ESR, &esr));
@@ -67,6 +78,10 @@ ARM_Unicorn::ARM_Unicorn() {
     uc_hook hook{};
     CHECKED(uc_hook_add(uc, &hook, UC_HOOK_INTR, (void*)InterruptHook, this, 0, -1));
     CHECKED(uc_hook_add(uc, &hook, UC_HOOK_MEM_INVALID, (void*)UnmappedMemoryHook, this, 0, -1));
+    if (GDBStub::IsServerEnabled()) {
+        CHECKED(uc_hook_add(uc, &hook, UC_HOOK_CODE, (void*)CodeHook, this, 0, -1));
+        last_bkpt_hit = false;
+    }
 }
 
 ARM_Unicorn::~ARM_Unicorn() {
@@ -155,7 +170,11 @@ void ARM_Unicorn::SetTlsAddress(VAddr base) {
 }
 
 void ARM_Unicorn::Run() {
-    ExecuteInstructions(std::max(CoreTiming::GetDowncount(), 0));
+    if (GDBStub::IsServerEnabled()) {
+        ExecuteInstructions(std::max(4000000, 0));
+    } else {
+        ExecuteInstructions(std::max(CoreTiming::GetDowncount(), 0));
+    }
 }
 
 void ARM_Unicorn::Step() {
@@ -168,6 +187,18 @@ void ARM_Unicorn::ExecuteInstructions(int num_instructions) {
     MICROPROFILE_SCOPE(ARM_Jit);
     CHECKED(uc_emu_start(uc, GetPC(), 1ULL << 63, 0, num_instructions));
     CoreTiming::AddTicks(num_instructions);
+    if (GDBStub::IsServerEnabled()) {
+        if (last_bkpt_hit) {
+            uc_reg_write(uc, UC_ARM64_REG_PC, &last_bkpt.address);
+        }
+        Kernel::Thread* thread = Kernel::GetCurrentThread();
+        SaveContext(thread->context);
+        if (last_bkpt_hit) {
+            last_bkpt_hit = false;
+            GDBStub::Break();
+        }
+        GDBStub::SendTrap(thread, 5);
+    }
 }
 
 void ARM_Unicorn::SaveContext(ARM_Interface::ThreadContext& ctx) {
@@ -233,3 +264,8 @@ void ARM_Unicorn::PrepareReschedule() {
 }
 
 void ARM_Unicorn::ClearInstructionCache() {}
+
+void ARM_Unicorn::RecordBreak(GDBStub::BreakpointAddress bkpt) {
+    last_bkpt = bkpt;
+    last_bkpt_hit = true;
+}

src/core/arm/unicorn/arm_unicorn.h

@@ -7,6 +7,7 @@
 #include <unicorn/unicorn.h>
 #include "common/common_types.h"
 #include "core/arm/arm_interface.h"
+#include "core/gdbstub/gdbstub.h"
 
 class ARM_Unicorn final : public ARM_Interface {
 public:
@@ -35,7 +36,10 @@ public:
     void Step() override;
     void ClearInstructionCache() override;
     void PageTableChanged() override{};
+    void RecordBreak(GDBStub::BreakpointAddress bkpt);
 
 private:
     uc_engine* uc{};
+    GDBStub::BreakpointAddress last_bkpt{};
+    bool last_bkpt_hit;
 };

src/core/file_sys/partition_filesystem.cpp

@@ -73,21 +73,21 @@ u32 PartitionFilesystem::GetNumEntries() const {
     return pfs_header.num_entries;
 }
 
-u64 PartitionFilesystem::GetEntryOffset(int index) const {
+u64 PartitionFilesystem::GetEntryOffset(u32 index) const {
     if (index > GetNumEntries())
         return 0;
 
     return content_offset + pfs_entries[index].fs_entry.offset;
 }
 
-u64 PartitionFilesystem::GetEntrySize(int index) const {
+u64 PartitionFilesystem::GetEntrySize(u32 index) const {
     if (index > GetNumEntries())
         return 0;
 
     return pfs_entries[index].fs_entry.size;
 }
 
-std::string PartitionFilesystem::GetEntryName(int index) const {
+std::string PartitionFilesystem::GetEntryName(u32 index) const {
     if (index > GetNumEntries())
         return "";
 

src/core/file_sys/partition_filesystem.h

@@ -27,9 +27,9 @@ public:
     Loader::ResultStatus Load(const std::vector<u8>& file_data, size_t offset = 0);
 
     u32 GetNumEntries() const;
-    u64 GetEntryOffset(int index) const;
-    u64 GetEntrySize(int index) const;
-    std::string GetEntryName(int index) const;
+    u64 GetEntryOffset(u32 index) const;
+    u64 GetEntrySize(u32 index) const;
+    std::string GetEntryName(u32 index) const;
     u64 GetFileOffset(const std::string& name) const;
     u64 GetFileSize(const std::string& name) const;
 

src/core/gdbstub/gdbstub.cpp

@@ -32,9 +32,13 @@
 
 #include "common/logging/log.h"
 #include "common/string_util.h"
+#include "common/swap.h"
 #include "core/arm/arm_interface.h"
 #include "core/core.h"
+#include "core/core_cpu.h"
 #include "core/gdbstub/gdbstub.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/scheduler.h"
 #include "core/loader/loader.h"
 #include "core/memory.h"
 
@@ -137,15 +141,17 @@ static u8 command_buffer[GDB_BUFFER_SIZE];
 static u32 command_length;
 
 static u32 latest_signal = 0;
-static bool step_break = false;
 static bool memory_break = false;
 
+static Kernel::Thread* current_thread = nullptr;
+
 // Binding to a port within the reserved ports range (0-1023) requires root permissions,
 // so default to a port outside of that range.
 static u16 gdbstub_port = 24689;
 
 static bool halt_loop = true;
 static bool step_loop = false;
+static bool send_trap = false;
 
 // If set to false, the server will never be started and no
 // gdbstub-related functions will be executed.
@@ -165,6 +171,53 @@ static std::map<u64, Breakpoint> breakpoints_execute;
 static std::map<u64, Breakpoint> breakpoints_read;
 static std::map<u64, Breakpoint> breakpoints_write;
 
+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) {
+            if (thread->GetThreadId() == id) {
+                current_thread = thread.get();
+                return current_thread;
+            }
+        }
+    }
+    return nullptr;
+}
+
+static u64 RegRead(int id, Kernel::Thread* thread = nullptr) {
+    if (!thread) {
+        return 0;
+    }
+
+    if (id < SP_REGISTER) {
+        return thread->context.cpu_registers[id];
+    } else if (id == SP_REGISTER) {
+        return thread->context.sp;
+    } else if (id == PC_REGISTER) {
+        return thread->context.pc;
+    } else if (id == CPSR_REGISTER) {
+        return thread->context.cpsr;
+    } else {
+        return 0;
+    }
+}
+
+static void RegWrite(int id, u64 val, Kernel::Thread* thread = nullptr) {
+    if (!thread) {
+        return;
+    }
+
+    if (id < SP_REGISTER) {
+        thread->context.cpu_registers[id] = val;
+    } else if (id == SP_REGISTER) {
+        thread->context.sp = val;
+    } else if (id == PC_REGISTER) {
+        thread->context.pc = val;
+    } else if (id == CPSR_REGISTER) {
+        thread->context.cpsr = val;
+    }
+}
+
 /**
  * Turns hex string character into the equivalent byte.
  *
@@ -193,7 +246,7 @@ static u8 NibbleToHex(u8 n) {
     if (n < 0xA) {
         return '0' + n;
     } else {
-        return 'A' + n - 0xA;
+        return 'a' + n - 0xA;
     }
 }
 
@@ -439,6 +492,8 @@ static void SendReply(const char* reply) {
         return;
     }
 
+    NGLOG_DEBUG(Debug_GDBStub, "Reply: {}", reply);
+
     memset(command_buffer, 0, sizeof(command_buffer));
 
     command_length = static_cast<u32>(strlen(reply));
@@ -483,6 +538,22 @@ static void HandleQuery() {
     } else if (strncmp(query, "Xfer:features:read:target.xml:",
                        strlen("Xfer:features:read:target.xml:")) == 0) {
         SendReply(target_xml);
+    } else if (strncmp(query, "Offsets", strlen("Offsets")) == 0) {
+        std::string buffer = fmt::format("TextSeg={:0x}", Memory::PROCESS_IMAGE_VADDR);
+        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) {
+                val += fmt::format("{:x}", thread->GetThreadId());
+                val += ",";
+            }
+        }
+        val.pop_back();
+        SendReply(val.c_str());
+    } else if (strncmp(query, "sThreadInfo", strlen("sThreadInfo")) == 0) {
+        SendReply("l");
     } else {
         SendReply("");
     }
@@ -490,11 +561,40 @@ static void HandleQuery() {
 
 /// Handle set thread command from gdb client.
 static void HandleSetThread() {
-    if (memcmp(command_buffer, "Hg0", 3) == 0 || memcmp(command_buffer, "Hc-1", 4) == 0 ||
-        memcmp(command_buffer, "Hc0", 4) == 0 || memcmp(command_buffer, "Hc1", 4) == 0) {
-        return SendReply("OK");
+    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;
+        }
     }
+    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)*/));
+    if (thread_id == 0) {
+        thread_id = 1;
+    }
+    if (FindThreadById(thread_id)) {
+        SendReply("OK");
+        return;
+    }
     SendReply("E01");
 }
 
@@ -503,15 +603,24 @@ static void HandleSetThread() {
  *
  * @param signal Signal to be sent to client.
  */
-static void SendSignal(u32 signal) {
+static void SendSignal(Kernel::Thread* thread, u32 signal, bool full = true) {
     if (gdbserver_socket == -1) {
         return;
     }
 
     latest_signal = signal;
 
-    std::string buffer = fmt::format("T{:02x}", latest_signal);
-    NGLOG_DEBUG(Debug_GDBStub, "Response: {}", buffer);
+    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);
+    }
+
+    buffer += fmt::format("thread:{:x};", thread->GetThreadId());
+
     SendReply(buffer.c_str());
 }
 
@@ -527,7 +636,7 @@ static void ReadCommand() {
     } else if (c == 0x03) {
         NGLOG_INFO(Debug_GDBStub, "gdb: found break command");
         halt_loop = true;
-        SendSignal(SIGTRAP);
+        SendSignal(current_thread, SIGTRAP);
         return;
     } else if (c != GDB_STUB_START) {
         NGLOG_DEBUG(Debug_GDBStub, "gdb: read invalid byte {:02X}", c);
@@ -598,11 +707,11 @@ static void ReadRegister() {
     }
 
     if (id <= SP_REGISTER) {
-        LongToGdbHex(reply, Core::CurrentArmInterface().GetReg(static_cast<int>(id)));
+        LongToGdbHex(reply, RegRead(id, current_thread));
     } else if (id == PC_REGISTER) {
-        LongToGdbHex(reply, Core::CurrentArmInterface().GetPC());
+        LongToGdbHex(reply, RegRead(id, current_thread));
     } else if (id == CPSR_REGISTER) {
-        IntToGdbHex(reply, Core::CurrentArmInterface().GetCPSR());
+        IntToGdbHex(reply, (u32)RegRead(id, current_thread));
     } else {
         return SendReply("E01");
     }
@@ -618,16 +727,16 @@ static void ReadRegisters() {
     u8* bufptr = buffer;
 
     for (int reg = 0; reg <= SP_REGISTER; reg++) {
-        LongToGdbHex(bufptr + reg * 16, Core::CurrentArmInterface().GetReg(reg));
+        LongToGdbHex(bufptr + reg * 16, RegRead(reg, current_thread));
     }
 
     bufptr += (32 * 16);
 
-    LongToGdbHex(bufptr, Core::CurrentArmInterface().GetPC());
+    LongToGdbHex(bufptr, RegRead(PC_REGISTER, current_thread));
 
     bufptr += 16;
 
-    IntToGdbHex(bufptr, Core::CurrentArmInterface().GetCPSR());
+    IntToGdbHex(bufptr, (u32)RegRead(CPSR_REGISTER, current_thread));
 
     bufptr += 8;
 
@@ -646,11 +755,11 @@ static void WriteRegister() {
     }
 
     if (id <= SP_REGISTER) {
-        Core::CurrentArmInterface().SetReg(id, GdbHexToLong(buffer_ptr));
+        RegWrite(id, GdbHexToLong(buffer_ptr), current_thread);
     } else if (id == PC_REGISTER) {
-        Core::CurrentArmInterface().SetPC(GdbHexToLong(buffer_ptr));
+        RegWrite(id, GdbHexToLong(buffer_ptr), current_thread);
     } else if (id == CPSR_REGISTER) {
-        Core::CurrentArmInterface().SetCPSR(GdbHexToInt(buffer_ptr));
+        RegWrite(id, GdbHexToInt(buffer_ptr), current_thread);
     } else {
         return SendReply("E01");
     }
@@ -667,11 +776,11 @@ static void WriteRegisters() {
 
     for (int i = 0, reg = 0; reg <= CPSR_REGISTER; i++, reg++) {
         if (reg <= SP_REGISTER) {
-            Core::CurrentArmInterface().SetReg(reg, GdbHexToLong(buffer_ptr + i * 16));
+            RegWrite(reg, GdbHexToLong(buffer_ptr + i * 16), current_thread);
         } else if (reg == PC_REGISTER) {
-            Core::CurrentArmInterface().SetPC(GdbHexToLong(buffer_ptr + i * 16));
+            RegWrite(PC_REGISTER, GdbHexToLong(buffer_ptr + i * 16), current_thread);
         } else if (reg == CPSR_REGISTER) {
-            Core::CurrentArmInterface().SetCPSR(GdbHexToInt(buffer_ptr + i * 16));
+            RegWrite(CPSR_REGISTER, GdbHexToInt(buffer_ptr + i * 16), current_thread);
         } else {
             UNIMPLEMENTED();
         }
@@ -734,7 +843,7 @@ static void WriteMemory() {
 void Break(bool is_memory_break) {
     if (!halt_loop) {
         halt_loop = true;
-        SendSignal(SIGTRAP);
+        send_trap = true;
     }
 
     memory_break = is_memory_break;
@@ -744,10 +853,10 @@ void Break(bool is_memory_break) {
 static void Step() {
     step_loop = true;
     halt_loop = true;
-    step_break = true;
-    SendSignal(SIGTRAP);
+    send_trap = true;
 }
 
+/// Tell the CPU if we hit a memory breakpoint.
 bool IsMemoryBreak() {
     if (IsConnected()) {
         return false;
@@ -759,7 +868,6 @@ bool IsMemoryBreak() {
 /// Tell the CPU to continue executing.
 static void Continue() {
     memory_break = false;
-    step_break = false;
     step_loop = false;
     halt_loop = false;
 }
@@ -898,7 +1006,7 @@ void HandlePacket() {
         HandleSetThread();
         break;
     case '?':
-        SendSignal(latest_signal);
+        SendSignal(current_thread, latest_signal);
         break;
     case 'k':
         Shutdown();
@@ -935,6 +1043,9 @@ void HandlePacket() {
     case 'Z':
         AddBreakpoint();
         break;
+    case 'T':
+        HandleThreadAlive();
+        break;
     default:
         SendReply("");
         break;
@@ -1079,4 +1190,11 @@ bool GetCpuStepFlag() {
 void SetCpuStepFlag(bool is_step) {
     step_loop = is_step;
 }
+
+void SendTrap(Kernel::Thread* thread, int trap) {
+    if (send_trap) {
+        send_trap = false;
+        SendSignal(thread, trap);
+    }
+}
 }; // namespace GDBStub

src/core/gdbstub/gdbstub.h

@@ -7,6 +7,7 @@
 #pragma once
 
 #include "common/common_types.h"
+#include "core/hle/kernel/thread.h"
 
 namespace GDBStub {
 
@@ -91,4 +92,12 @@ bool GetCpuStepFlag();
  * @param is_step
  */
 void SetCpuStepFlag(bool is_step);
+
+/**
+ * Send trap signal from thread back to the gdbstub server.
+ *
+ * @param thread Sending thread.
+ * @param trap Trap no.
+ */
+void SendTrap(Kernel::Thread* thread, int trap);
 } // namespace GDBStub

src/core/hle/kernel/svc.cpp

@@ -316,6 +316,11 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
                       "(STUBBED) Attempted to query privileged process id bounds, returned 0");
         *result = 0;
         break;
+    case GetInfoType::UserExceptionContextAddr:
+        NGLOG_WARNING(Kernel_SVC,
+                      "(STUBBED) Attempted to query user exception context address, returned 0");
+        *result = 0;
+        break;
     default:
         UNIMPLEMENTED();
     }
@@ -744,7 +749,7 @@ static ResultCode SetThreadCoreMask(Handle thread_handle, u32 core, u64 mask) {
         ASSERT(thread->owner_process->ideal_processor != THREADPROCESSORID_DEFAULT);
         // Set the target CPU to the one specified in the process' exheader.
         core = thread->owner_process->ideal_processor;
-        mask = 1 << core;
+        mask = 1ull << core;
     }
 
     if (mask == 0) {
@@ -761,7 +766,7 @@ static ResultCode SetThreadCoreMask(Handle thread_handle, u32 core, u64 mask) {
     }
 
     // Error out if the input core isn't enabled in the input mask.
-    if (core < Core::NUM_CPU_CORES && (mask & (1 << core)) == 0) {
+    if (core < Core::NUM_CPU_CORES && (mask & (1ull << core)) == 0) {
         return ResultCode(ErrorModule::Kernel, ErrCodes::InvalidCombination);
     }
 

src/core/hle/service/am/am.cpp

@@ -561,7 +561,7 @@ IApplicationFunctions::IApplicationFunctions() : ServiceFramework("IApplicationF
         {32, nullptr, "BeginBlockingHomeButton"},
         {33, nullptr, "EndBlockingHomeButton"},
         {40, &IApplicationFunctions::NotifyRunning, "NotifyRunning"},
-        {50, nullptr, "GetPseudoDeviceId"},
+        {50, &IApplicationFunctions::GetPseudoDeviceId, "GetPseudoDeviceId"},
         {60, nullptr, "SetMediaPlaybackStateForApplication"},
         {65, nullptr, "IsGamePlayRecordingSupported"},
         {66, &IApplicationFunctions::InitializeGamePlayRecording, "InitializeGamePlayRecording"},
@@ -684,6 +684,17 @@ void IApplicationFunctions::NotifyRunning(Kernel::HLERequestContext& ctx) {
     NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
+void IApplicationFunctions::GetPseudoDeviceId(Kernel::HLERequestContext& ctx) {
+    IPC::ResponseBuilder rb{ctx, 6};
+    rb.Push(RESULT_SUCCESS);
+
+    // Returns a 128-bit UUID
+    rb.Push<u64>(0);
+    rb.Push<u64>(0);
+
+    NGLOG_WARNING(Service_AM, "(STUBBED) called");
+}
+
 void InstallInterfaces(SM::ServiceManager& service_manager,
                        std::shared_ptr<NVFlinger::NVFlinger> nvflinger) {
     std::make_shared<AppletAE>(nvflinger)->InstallAsService(service_manager);

src/core/hle/service/am/am.h

@@ -138,6 +138,7 @@ private:
     void InitializeGamePlayRecording(Kernel::HLERequestContext& ctx);
     void SetGamePlayRecordingState(Kernel::HLERequestContext& ctx);
     void NotifyRunning(Kernel::HLERequestContext& ctx);
+    void GetPseudoDeviceId(Kernel::HLERequestContext& ctx);
 };
 
 class IHomeMenuFunctions final : public ServiceFramework<IHomeMenuFunctions> {

src/core/hle/service/audio/audren_u.cpp

@@ -57,27 +57,26 @@ private:
     }
 
     void RequestUpdateAudioRenderer(Kernel::HLERequestContext& ctx) {
-        NGLOG_DEBUG(Service_Audio, "{}", ctx.Description());
-        AudioRendererResponseData response_data{};
+        AudioRendererConfig config;
+        auto buf = ctx.ReadBuffer();
+        std::memcpy(&config, buf.data(), sizeof(AudioRendererConfig));
 
-        response_data.section_0_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);
+        AudioRendererResponse response_data{config};
 
-        for (unsigned i = 0; i < response_data.state_entries.size(); i++) {
-            // 4 = Busy and 5 = Ready?
-            response_data.state_entries[i].state = 5;
+        ASSERT(ctx.GetWriteBufferSize() == response_data.total_size);
+
+        std::vector<u8> output(response_data.total_size);
+        std::memcpy(output.data(), &response_data, sizeof(AudioRendererResponse));
+        std::vector<MemoryPoolEntry> memory_pool(config.memory_pools_size / 0x20);
+        for (auto& entry : memory_pool) {
+            entry.state = 5;
         }
+        std::memcpy(output.data() + sizeof(AudioRendererResponse), memory_pool.data(),
+                    response_data.memory_pools_size);
 
-        ctx.WriteBuffer(&response_data, response_data.total_size);
+        ctx.WriteBuffer(output);
 
         IPC::ResponseBuilder rb{ctx, 2};
-
         rb.Push(RESULT_SUCCESS);
 
         NGLOG_WARNING(Service_Audio, "(STUBBED) called");
@@ -109,43 +108,55 @@ private:
         NGLOG_WARNING(Service_Audio, "(STUBBED) called");
     }
 
-    struct AudioRendererStateEntry {
+    struct MemoryPoolEntry {
         u32_le state;
         u32_le unknown_4;
         u32_le unknown_8;
         u32_le unknown_c;
     };
-    static_assert(sizeof(AudioRendererStateEntry) == 0x10,
-                  "AudioRendererStateEntry has wrong size");
+    static_assert(sizeof(MemoryPoolEntry) == 0x10, "MemoryPoolEntry has wrong size");
 
-    struct AudioRendererResponseData {
-        u32_le unknown_0;
-        u32_le section_5_size;
-        u32_le section_0_size;
-        u32_le section_1_size;
-        u32_le unknown_10;
-        u32_le section_2_size;
-        u32_le unknown_18;
-        u32_le section_3_size;
-        u32_le section_4_size;
-        u32_le unknown_24;
-        u32_le unknown_28;
-        u32_le unknown_2c;
-        u32_le unknown_30;
-        u32_le unknown_34;
-        u32_le unknown_38;
-        u32_le total_size;
-
-        std::array<AudioRendererStateEntry, 0x18e> state_entries;
-
-        std::array<u8, 0x600> section_1;
-        std::array<u8, 0xe0> section_2;
-        std::array<u8, 0x20> section_3;
-        std::array<u8, 0x10> section_4;
-        std::array<u8, 0xb0> section_5;
+    struct AudioRendererConfig {
+        u32 revision;
+        u32 behavior_size;
+        u32 memory_pools_size;
+        u32 voices_size;
+        u32 voice_resource_size;
+        u32 effects_size;
+        u32 mixes_size;
+        u32 sinks_size;
+        u32 performance_buffer_size;
+        INSERT_PADDING_WORDS(6);
+        u32 total_size;
     };
-    static_assert(sizeof(AudioRendererResponseData) == 0x20e0,
-                  "AudioRendererResponseData has wrong size");
+    static_assert(sizeof(AudioRendererConfig) == 0x40, "AudioRendererConfig has wrong size");
+
+    struct AudioRendererResponse {
+        AudioRendererResponse(const AudioRendererConfig& config) {
+            revision = config.revision;
+            error_info_size = 0xb0;
+            memory_pools_size = (config.memory_pools_size / 0x20) * 0x10;
+            voices_size = (config.voices_size / 0x170) * 0x10;
+            effects_size = (config.effects_size / 0xC0) * 0x10;
+            sinks_size = (config.sinks_size / 0x140) * 0x20;
+            performance_manager_size = 0x10;
+            total_size = sizeof(AudioRendererResponse) + error_info_size + memory_pools_size +
+                         voices_size + effects_size + sinks_size + performance_manager_size;
+        }
+
+        u32_le revision;
+        u32_le error_info_size;
+        u32_le memory_pools_size;
+        u32_le voices_size;
+        u32_le unknown_10;
+        u32_le effects_size;
+        u32_le unknown_18;
+        u32_le sinks_size;
+        u32_le performance_manager_size;
+        INSERT_PADDING_WORDS(6);
+        u32_le total_size;
+    };
+    static_assert(sizeof(AudioRendererResponse) == 0x40, "AudioRendererResponse has wrong size");
 
     /// This is used to trigger the audio event callback.
     CoreTiming::EventType* audio_event;
@@ -258,7 +269,7 @@ void AudRenU::OpenAudioRenderer(Kernel::HLERequestContext& ctx) {
 
 void AudRenU::GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
-    auto params = rp.PopRaw<WorkerBufferParameters>();
+    auto params = rp.PopRaw<AudioRendererParameters>();
 
     u64 buffer_sz = Common::AlignUp(4 * params.unknown8, 0x40);
     buffer_sz += params.unknownC * 1024;
@@ -328,7 +339,7 @@ bool AudRenU::IsFeatureSupported(AudioFeatures feature, u32_le revision) const {
     u32_be version_num = (revision - Common::MakeMagic('R', 'E', 'V', '0')); // Byte swap
     switch (feature) {
     case AudioFeatures::Splitter:
-        return version_num >= 2;
+        return version_num >= 2u;
     default:
         return false;
     }

src/core/hle/service/audio/audren_u.h

@@ -22,7 +22,7 @@ private:
     void GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx);
     void GetAudioDevice(Kernel::HLERequestContext& ctx);
 
-    struct WorkerBufferParameters {
+    struct AudioRendererParameters {
         u32_le sample_rate;
         u32_le sample_count;
         u32_le unknown8;
@@ -38,8 +38,8 @@ private:
         u8 padding2[4];
         u32_le magic;
     };
-    static_assert(sizeof(WorkerBufferParameters) == 52,
-                  "WorkerBufferParameters is an invalid size");
+    static_assert(sizeof(AudioRendererParameters) == 52,
+                  "AudioRendererParameters is an invalid size");
 
     enum class AudioFeatures : u32 {
         Splitter,

src/core/hle/service/filesystem/fsp_srv.cpp

@@ -4,6 +4,7 @@
 
 #include <cinttypes>
 #include "common/logging/log.h"
+#include "common/string_util.h"
 #include "core/core.h"
 #include "core/file_sys/directory.h"
 #include "core/file_sys/filesystem.h"
@@ -258,9 +259,7 @@ public:
         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);
+        std::string name = Common::StringFromBuffer(file_buffer);
 
         u64 mode = rp.Pop<u64>();
         u32 size = rp.Pop<u32>();
@@ -275,9 +274,7 @@ public:
         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);
+        std::string name = Common::StringFromBuffer(file_buffer);
 
         NGLOG_DEBUG(Service_FS, "called file {}", name);
 
@@ -289,9 +286,7 @@ public:
         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);
+        std::string name = Common::StringFromBuffer(file_buffer);
 
         NGLOG_DEBUG(Service_FS, "called directory {}", name);
 
@@ -305,13 +300,11 @@ public:
         std::vector<u8> buffer;
         buffer.resize(ctx.BufferDescriptorX()[0].Size());
         Memory::ReadBlock(ctx.BufferDescriptorX()[0].Address(), buffer.data(), buffer.size());
-        auto end = std::find(buffer.begin(), buffer.end(), '\0');
-        std::string src_name(buffer.begin(), end);
+        std::string src_name = Common::StringFromBuffer(buffer);
 
         buffer.resize(ctx.BufferDescriptorX()[1].Size());
         Memory::ReadBlock(ctx.BufferDescriptorX()[1].Address(), buffer.data(), buffer.size());
-        end = std::find(buffer.begin(), buffer.end(), '\0');
-        std::string dst_name(buffer.begin(), end);
+        std::string dst_name = Common::StringFromBuffer(buffer);
 
         NGLOG_DEBUG(Service_FS, "called file '{}' to file '{}'", src_name, dst_name);
 
@@ -323,9 +316,7 @@ public:
         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);
+        std::string name = Common::StringFromBuffer(file_buffer);
 
         auto mode = static_cast<FileSys::Mode>(rp.Pop<u32>());
 
@@ -349,9 +340,7 @@ public:
         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);
+        std::string name = Common::StringFromBuffer(file_buffer);
 
         // TODO(Subv): Implement this filter.
         u32 filter_flags = rp.Pop<u32>();
@@ -376,9 +365,7 @@ public:
         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);
+        std::string name = Common::StringFromBuffer(file_buffer);
 
         NGLOG_DEBUG(Service_FS, "called file {}", name);
 

src/core/hle/service/hid/hid.cpp

@@ -84,6 +84,10 @@ private:
 
         for (size_t controller = 0; controller < mem.controllers.size(); controller++) {
             for (int index = 0; index < HID_NUM_LAYOUTS; index++) {
+                // TODO(DarkLordZach): Is this layout/controller config actually invalid?
+                if (controller == Controller_Handheld && index == Layout_Single)
+                    continue;
+
                 ControllerLayout& layout = mem.controllers[controller].layouts[index];
                 layout.header.num_entries = HID_NUM_ENTRIES;
                 layout.header.max_entry_index = HID_NUM_ENTRIES - 1;
@@ -94,7 +98,6 @@ private:
                 layout.header.latest_entry = (layout.header.latest_entry + 1) % HID_NUM_ENTRIES;
 
                 ControllerInputEntry& entry = layout.entries[layout.header.latest_entry];
-                entry.connection_state = ConnectionState_Connected | ConnectionState_Wired;
                 entry.timestamp++;
                 // TODO(shinyquagsire23): Is this always identical to timestamp?
                 entry.timestamp_2++;
@@ -103,6 +106,8 @@ private:
                 if (controller != Controller_Handheld)
                     continue;
 
+                entry.connection_state = ConnectionState_Connected | ConnectionState_Wired;
+
                 // TODO(shinyquagsire23): Set up some LUTs for each layout mapping in the future?
                 // For now everything is just the default handheld layout, but split Joy-Con will
                 // rotate the face buttons and directions for certain layouts.

src/core/hle/service/hid/hid.h

@@ -12,7 +12,7 @@ namespace Service::HID {
 // Begin enums and output structs
 
 constexpr u32 HID_NUM_ENTRIES = 17;
-constexpr u32 HID_NUM_LAYOUTS = 2;
+constexpr u32 HID_NUM_LAYOUTS = 7;
 constexpr s32 HID_JOYSTICK_MAX = 0x8000;
 constexpr s32 HID_JOYSTICK_MIN = -0x8000;
 

src/core/hle/service/nfp/nfp.cpp

@@ -4,6 +4,8 @@
 
 #include "common/logging/log.h"
 #include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/event.h"
+#include "core/hle/service/hid/hid.h"
 #include "core/hle/service/nfp/nfp.h"
 #include "core/hle/service/nfp/nfp_user.h"
 
@@ -18,7 +20,7 @@ public:
         static const FunctionInfo functions[] = {
             {0, &IUser::Initialize, "Initialize"},
             {1, nullptr, "Finalize"},
-            {2, nullptr, "ListDevices"},
+            {2, &IUser::ListDevices, "ListDevices"},
             {3, nullptr, "StartDetection"},
             {4, nullptr, "StopDetection"},
             {5, nullptr, "Mount"},
@@ -33,24 +35,116 @@ public:
             {14, nullptr, "GetRegisterInfo"},
             {15, nullptr, "GetCommonInfo"},
             {16, nullptr, "GetModelInfo"},
-            {17, nullptr, "AttachActivateEvent"},
-            {18, nullptr, "AttachDeactivateEvent"},
-            {19, nullptr, "GetState"},
-            {20, nullptr, "GetDeviceState"},
-            {21, nullptr, "GetNpadId"},
+            {17, &IUser::AttachActivateEvent, "AttachActivateEvent"},
+            {18, &IUser::AttachDeactivateEvent, "AttachDeactivateEvent"},
+            {19, &IUser::GetState, "GetState"},
+            {20, &IUser::GetDeviceState, "GetDeviceState"},
+            {21, &IUser::GetNpadId, "GetNpadId"},
             {22, nullptr, "GetApplicationArea2"},
-            {23, nullptr, "AttachAvailabilityChangeEvent"},
+            {23, &IUser::AttachAvailabilityChangeEvent, "AttachAvailabilityChangeEvent"},
             {24, nullptr, "RecreateApplicationArea"},
         };
         RegisterHandlers(functions);
+
+        activate_event = Kernel::Event::Create(Kernel::ResetType::OneShot, "IUser:ActivateEvent");
+        deactivate_event =
+            Kernel::Event::Create(Kernel::ResetType::OneShot, "IUser:DeactivateEvent");
+        availability_change_event =
+            Kernel::Event::Create(Kernel::ResetType::OneShot, "IUser:AvailabilityChangeEvent");
     }
 
 private:
+    enum class State : u32 {
+        NonInitialized = 0,
+        Initialized = 1,
+    };
+
+    enum class DeviceState : u32 {
+        Initialized = 0,
+    };
+
     void Initialize(Kernel::HLERequestContext& ctx) {
         NGLOG_WARNING(Service_NFP, "(STUBBED) called");
+
+        state = State::Initialized;
+
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
+
+    void ListDevices(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const u32 array_size = rp.Pop<u32>();
+
+        ctx.WriteBuffer(&device_handle, sizeof(device_handle));
+
+        NGLOG_WARNING(Service_NFP, "(STUBBED) called, array_size={}", array_size);
+
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(0);
+    }
+
+    void AttachActivateEvent(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const u64 dev_handle = rp.Pop<u64>();
+        NGLOG_WARNING(Service_NFP, "(STUBBED) called, dev_handle=0x{:X}", dev_handle);
+
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(activate_event);
+    }
+
+    void AttachDeactivateEvent(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const u64 dev_handle = rp.Pop<u64>();
+        NGLOG_WARNING(Service_NFP, "(STUBBED) called, dev_handle=0x{:X}", dev_handle);
+
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(deactivate_event);
+    }
+
+    void GetState(Kernel::HLERequestContext& ctx) {
+        NGLOG_WARNING(Service_NFP, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(static_cast<u32>(state));
+    }
+
+    void GetDeviceState(Kernel::HLERequestContext& ctx) {
+        NGLOG_WARNING(Service_NFP, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(static_cast<u32>(device_state));
+    }
+
+    void GetNpadId(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const u64 dev_handle = rp.Pop<u64>();
+        NGLOG_WARNING(Service_NFP, "(STUBBED) called, dev_handle=0x{:X}", dev_handle);
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push<u32>(npad_id);
+    }
+
+    void AttachAvailabilityChangeEvent(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const u64 dev_handle = rp.Pop<u64>();
+        NGLOG_WARNING(Service_NFP, "(STUBBED) called, dev_handle=0x{:X}", dev_handle);
+
+        IPC::ResponseBuilder rb{ctx, 2, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushCopyObjects(availability_change_event);
+    }
+
+    const u64 device_handle{0xDEAD};
+    const HID::ControllerID npad_id{HID::Controller_Player1};
+    State state{State::NonInitialized};
+    DeviceState device_state{DeviceState::Initialized};
+    Kernel::SharedPtr<Kernel::Event> activate_event;
+    Kernel::SharedPtr<Kernel::Event> deactivate_event;
+    Kernel::SharedPtr<Kernel::Event> availability_change_event;
 };
 
 void Module::Interface::CreateUserInterface(Kernel::HLERequestContext& ctx) {

src/core/hle/service/nifm/nifm.cpp

@@ -38,7 +38,7 @@ public:
             {8, nullptr, "SetPriority"},
             {9, nullptr, "SetNetworkProfileId"},
             {10, nullptr, "SetRejectable"},
-            {11, nullptr, "SetConnectionConfirmationOption"},
+            {11, &IRequest::SetConnectionConfirmationOption, "SetConnectionConfirmationOption"},
             {12, nullptr, "SetPersistent"},
             {13, nullptr, "SetInstant"},
             {14, nullptr, "SetSustainable"},
@@ -67,23 +67,32 @@ private:
         rb.Push(RESULT_SUCCESS);
         rb.Push<u32>(0);
     }
+
     void GetResult(Kernel::HLERequestContext& ctx) {
         NGLOG_WARNING(Service_NIFM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
+
     void GetSystemEventReadableHandles(Kernel::HLERequestContext& ctx) {
         NGLOG_WARNING(Service_NIFM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2, 2};
         rb.Push(RESULT_SUCCESS);
         rb.PushCopyObjects(event1, event2);
     }
+
     void Cancel(Kernel::HLERequestContext& ctx) {
         NGLOG_WARNING(Service_NIFM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
 
+    void SetConnectionConfirmationOption(Kernel::HLERequestContext& ctx) {
+        NGLOG_WARNING(Service_NIFM, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+
     Kernel::SharedPtr<Kernel::Event> event1, event2;
 };
 

src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp

@@ -26,6 +26,10 @@ u32 nvhost_ctrl_gpu::ioctl(Ioctl command, const std::vector<u8>& input, std::vec
         return ZCullGetInfo(input, output);
     case IoctlCommand::IocZbcSetTable:
         return ZBCSetTable(input, output);
+    case IoctlCommand::IocZbcQueryTable:
+        return ZBCQueryTable(input, output);
+    case IoctlCommand::IocFlushL2:
+        return FlushL2(input, output);
     }
     UNIMPLEMENTED_MSG("Unimplemented ioctl");
     return 0;
@@ -136,4 +140,22 @@ u32 nvhost_ctrl_gpu::ZBCSetTable(const std::vector<u8>& input, std::vector<u8>&
     return 0;
 }
 
+u32 nvhost_ctrl_gpu::ZBCQueryTable(const std::vector<u8>& input, std::vector<u8>& output) {
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called");
+    IoctlZbcQueryTable params{};
+    std::memcpy(&params, input.data(), input.size());
+    // TODO : To implement properly
+    std::memcpy(output.data(), &params, output.size());
+    return 0;
+}
+
+u32 nvhost_ctrl_gpu::FlushL2(const std::vector<u8>& input, std::vector<u8>& output) {
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called");
+    IoctlFlushL2 params{};
+    std::memcpy(&params, input.data(), input.size());
+    // TODO : To implement properly
+    std::memcpy(output.data(), &params, output.size());
+    return 0;
+}
+
 } // namespace Service::Nvidia::Devices

src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h

@@ -26,6 +26,18 @@ private:
         IocZcullGetCtxSizeCommand = 0x80044701,
         IocZcullGetInfo = 0x80284702,
         IocZbcSetTable = 0x402C4703,
+        IocZbcQueryTable = 0xC0344704,
+        IocFlushL2 = 0x40084707,
+        IocInvalICache = 0x4008470D,
+        IocSetMmudebugMode = 0x4008470E,
+        IocSetSmDebugMode = 0x4010470F,
+        IocWaitForPause = 0xC0084710,
+        IocGetTcpExceptionEnStatus = 0x80084711,
+        IocNumVsms = 0x80084712,
+        IocVsmsMapping = 0xC0044713,
+        IocGetErrorChannelUserData = 0xC008471B,
+        IocGetGpuTime = 0xC010471C,
+        IocGetCpuTimeCorrelationInfo = 0xC108471D,
     };
 
     struct IoctlGpuCharacteristics {
@@ -127,12 +139,31 @@ private:
     };
     static_assert(sizeof(IoctlZbcSetTable) == 44, "IoctlZbcSetTable is incorrect size");
 
+    struct IoctlZbcQueryTable {
+        u32_le color_ds[4];
+        u32_le color_l2[4];
+        u32_le depth;
+        u32_le ref_cnt;
+        u32_le format;
+        u32_le type;
+        u32_le index_size;
+    };
+    static_assert(sizeof(IoctlZbcQueryTable) == 52, "IoctlZbcQueryTable is incorrect size");
+
+    struct IoctlFlushL2 {
+        u32_le flush; // l2_flush | l2_invalidate << 1 | fb_flush << 2
+        u32_le reserved;
+    };
+    static_assert(sizeof(IoctlFlushL2) == 8, "IoctlFlushL2 is incorrect size");
+
     u32 GetCharacteristics(const std::vector<u8>& input, std::vector<u8>& output);
     u32 GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output);
     u32 GetActiveSlotMask(const std::vector<u8>& input, std::vector<u8>& output);
     u32 ZCullGetCtxSize(const std::vector<u8>& input, std::vector<u8>& output);
     u32 ZCullGetInfo(const std::vector<u8>& input, std::vector<u8>& output);
     u32 ZBCSetTable(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 ZBCQueryTable(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 FlushL2(const std::vector<u8>& input, std::vector<u8>& output);
 };
 
 } // namespace Service::Nvidia::Devices

src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp

@@ -121,8 +121,9 @@ u32 nvhost_gpu::AllocateObjectContext(const std::vector<u8>& input, std::vector<
 }
 
 u32 nvhost_gpu::SubmitGPFIFO(const std::vector<u8>& input, std::vector<u8>& output) {
-    if (input.size() < sizeof(IoctlSubmitGpfifo))
+    if (input.size() < sizeof(IoctlSubmitGpfifo)) {
         UNIMPLEMENTED();
+    }
     IoctlSubmitGpfifo params{};
     std::memcpy(&params, input.data(), sizeof(IoctlSubmitGpfifo));
     NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, gpfifo={:X}, num_entries={:X}, flags={:X}",

src/core/hle/service/set/set.cpp

@@ -12,9 +12,6 @@
 namespace Service::Set {
 
 void SET::GetAvailableLanguageCodes(Kernel::HLERequestContext& ctx) {
-    IPC::RequestParser rp{ctx};
-    u32 id = rp.Pop<u32>();
-
     static constexpr std::array<LanguageCode, 17> available_language_codes = {{
         LanguageCode::JA,
         LanguageCode::EN_US,
@@ -50,7 +47,7 @@ SET::SET() : ServiceFramework("set") {
         {2, nullptr, "MakeLanguageCode"},
         {3, nullptr, "GetAvailableLanguageCodeCount"},
         {4, nullptr, "GetRegionCode"},
-        {5, nullptr, "GetAvailableLanguageCodes2"},
+        {5, &SET::GetAvailableLanguageCodes, "GetAvailableLanguageCodes2"},
         {6, nullptr, "GetAvailableLanguageCodeCount2"},
         {7, nullptr, "GetKeyCodeMap"},
         {8, nullptr, "GetQuestFlag"},

src/video_core/CMakeLists.txt

@@ -9,6 +9,8 @@ add_library(video_core STATIC
     engines/maxwell_3d.h
     engines/maxwell_compute.cpp
     engines/maxwell_compute.h
+    engines/maxwell_dma.cpp
+    engines/maxwell_dma.h
     engines/shader_bytecode.h
     gpu.cpp
     gpu.h
@@ -39,6 +41,8 @@ add_library(video_core STATIC
     renderer_opengl/maxwell_to_gl.h
     renderer_opengl/renderer_opengl.cpp
     renderer_opengl/renderer_opengl.h
+    textures/astc.cpp
+    textures/astc.h
     textures/decoders.cpp
     textures/decoders.h
     textures/texture.h

src/video_core/command_processor.cpp

@@ -16,6 +16,7 @@
 #include "video_core/engines/fermi_2d.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/engines/maxwell_compute.h"
+#include "video_core/engines/maxwell_dma.h"
 #include "video_core/gpu.h"
 #include "video_core/renderer_base.h"
 #include "video_core/video_core.h"
@@ -60,8 +61,11 @@ void GPU::WriteReg(u32 method, u32 subchannel, u32 value, u32 remaining_params)
     case EngineID::MAXWELL_COMPUTE_B:
         maxwell_compute->WriteReg(method, value);
         break;
+    case EngineID::MAXWELL_DMA_COPY_A:
+        maxwell_dma->WriteReg(method, value);
+        break;
     default:
-        UNIMPLEMENTED();
+        UNIMPLEMENTED_MSG("Unimplemented engine");
     }
 }
 

src/video_core/engines/fermi_2d.cpp

@@ -47,6 +47,7 @@ void Fermi2D::HandleSurfaceCopy() {
 
     if (regs.src.linear == regs.dst.linear) {
         // If the input layout and the output layout are the same, just perform a raw copy.
+        ASSERT(regs.src.BlockHeight() == regs.dst.BlockHeight());
         Memory::CopyBlock(dest_cpu, source_cpu,
                           src_bytes_per_pixel * regs.dst.width * regs.dst.height);
         return;

src/video_core/engines/maxwell_3d.cpp

@@ -328,8 +328,9 @@ std::vector<Texture::FullTextureInfo> Maxwell3D::GetStageTextures(Regs::ShaderSt
 
         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);
+        tex_info.index =
+            static_cast<u32>(current_texture - tex_info_buffer.address - TextureInfoOffset) /
+            sizeof(Texture::TextureHandle);
 
         // Load the TIC data.
         if (tex_handle.tic_id != 0) {
@@ -354,6 +355,40 @@ std::vector<Texture::FullTextureInfo> Maxwell3D::GetStageTextures(Regs::ShaderSt
     return textures;
 }
 
+Texture::FullTextureInfo Maxwell3D::GetStageTexture(Regs::ShaderStage stage, size_t offset) const {
+    auto& shader = state.shader_stages[static_cast<size_t>(stage)];
+    auto& tex_info_buffer = shader.const_buffers[regs.tex_cb_index];
+    ASSERT(tex_info_buffer.enabled && tex_info_buffer.address != 0);
+
+    GPUVAddr tex_info_address = tex_info_buffer.address + offset * sizeof(Texture::TextureHandle);
+
+    ASSERT(tex_info_address < tex_info_buffer.address + tex_info_buffer.size);
+
+    boost::optional<VAddr> tex_address_cpu = memory_manager.GpuToCpuAddress(tex_info_address);
+    Texture::TextureHandle tex_handle{Memory::Read32(*tex_address_cpu)};
+
+    Texture::FullTextureInfo tex_info{};
+    tex_info.index = static_cast<u32>(offset);
+
+    // 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));
+    }
+
+    return tex_info;
+}
+
 u32 Maxwell3D::GetRegisterValue(u32 method) const {
     ASSERT_MSG(method < Regs::NUM_REGS, "Invalid Maxwell3D register");
     return regs.reg_array[method];

src/video_core/engines/maxwell_3d.h

@@ -318,6 +318,7 @@ public:
             Equation equation_a;
             Factor factor_source_a;
             Factor factor_dest_a;
+            INSERT_PADDING_WORDS(1);
         };
 
         union {
@@ -432,7 +433,27 @@ public:
                     };
                 } rt_control;
 
-                INSERT_PADDING_WORDS(0xCF);
+                INSERT_PADDING_WORDS(0x31);
+
+                u32 independent_blend_enable;
+
+                INSERT_PADDING_WORDS(0x15);
+
+                struct {
+                    u32 separate_alpha;
+                    Blend::Equation equation_rgb;
+                    Blend::Factor factor_source_rgb;
+                    Blend::Factor factor_dest_rgb;
+                    Blend::Equation equation_a;
+                    Blend::Factor factor_source_a;
+                    INSERT_PADDING_WORDS(1);
+                    Blend::Factor factor_dest_a;
+
+                    u32 enable_common;
+                    u32 enable[NumRenderTargets];
+                } blend;
+
+                INSERT_PADDING_WORDS(0x77);
 
                 struct {
                     u32 tsc_address_high;
@@ -557,9 +578,7 @@ public:
 
                 } vertex_array[NumVertexArrays];
 
-                Blend blend;
-
-                INSERT_PADDING_WORDS(0x39);
+                Blend independent_blend[NumRenderTargets];
 
                 struct {
                     u32 limit_high;
@@ -664,6 +683,9 @@ public:
     /// Returns a list of enabled textures for the specified shader stage.
     std::vector<Texture::FullTextureInfo> GetStageTextures(Regs::ShaderStage stage) const;
 
+    /// 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;
 
@@ -719,6 +741,8 @@ 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(independent_blend_enable, 0x4B9);
+ASSERT_REG_POSITION(blend, 0x4CF);
 ASSERT_REG_POSITION(tsc, 0x557);
 ASSERT_REG_POSITION(tic, 0x55D);
 ASSERT_REG_POSITION(code_address, 0x582);
@@ -726,7 +750,7 @@ ASSERT_REG_POSITION(draw, 0x585);
 ASSERT_REG_POSITION(index_array, 0x5F2);
 ASSERT_REG_POSITION(query, 0x6C0);
 ASSERT_REG_POSITION(vertex_array[0], 0x700);
-ASSERT_REG_POSITION(blend, 0x780);
+ASSERT_REG_POSITION(independent_blend, 0x780);
 ASSERT_REG_POSITION(vertex_array_limit[0], 0x7C0);
 ASSERT_REG_POSITION(shader_config[0], 0x800);
 ASSERT_REG_POSITION(const_buffer, 0x8E0);

src/video_core/engines/maxwell_dma.cpp

@@ -0,0 +1,69 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/memory.h"
+#include "video_core/engines/maxwell_dma.h"
+#include "video_core/textures/decoders.h"
+
+namespace Tegra {
+namespace Engines {
+
+MaxwellDMA::MaxwellDMA(MemoryManager& memory_manager) : memory_manager(memory_manager) {}
+
+void MaxwellDMA::WriteReg(u32 method, u32 value) {
+    ASSERT_MSG(method < Regs::NUM_REGS,
+               "Invalid MaxwellDMA register, increase the size of the Regs structure");
+
+    regs.reg_array[method] = value;
+
+#define MAXWELLDMA_REG_INDEX(field_name)                                                           \
+    (offsetof(Tegra::Engines::MaxwellDMA::Regs, field_name) / sizeof(u32))
+
+    switch (method) {
+    case MAXWELLDMA_REG_INDEX(exec): {
+        HandleCopy();
+        break;
+    }
+    }
+
+#undef MAXWELLDMA_REG_INDEX
+}
+
+void MaxwellDMA::HandleCopy() {
+    NGLOG_WARNING(HW_GPU, "Requested a DMA copy");
+
+    const GPUVAddr source = regs.src_address.Address();
+    const GPUVAddr dest = regs.dst_address.Address();
+
+    const VAddr source_cpu = *memory_manager.GpuToCpuAddress(source);
+    const VAddr dest_cpu = *memory_manager.GpuToCpuAddress(dest);
+
+    // TODO(Subv): Perform more research and implement all features of this engine.
+    ASSERT(regs.exec.enable_swizzle == 0);
+    ASSERT(regs.exec.enable_2d == 1);
+    ASSERT(regs.exec.query_mode == Regs::QueryMode::None);
+    ASSERT(regs.exec.query_intr == Regs::QueryIntr::None);
+    ASSERT(regs.exec.copy_mode == Regs::CopyMode::Unk2);
+    ASSERT(regs.src_params.pos_x == 0);
+    ASSERT(regs.src_params.pos_y == 0);
+    ASSERT(regs.dst_params.pos_x == 0);
+    ASSERT(regs.dst_params.pos_y == 0);
+    ASSERT(regs.exec.is_dst_linear != regs.exec.is_src_linear);
+
+    u8* src_buffer = Memory::GetPointer(source_cpu);
+    u8* dst_buffer = Memory::GetPointer(dest_cpu);
+
+    if (regs.exec.is_dst_linear && !regs.exec.is_src_linear) {
+        // If the input is tiled and the output is linear, deswizzle the input and copy it over.
+        Texture::CopySwizzledData(regs.src_params.size_x, regs.src_params.size_y, 1, 1, src_buffer,
+                                  dst_buffer, true, regs.src_params.BlockHeight());
+    } else {
+        // If the input is linear and the output is tiled, swizzle the input and copy it over.
+        Texture::CopySwizzledData(regs.dst_params.size_x, regs.dst_params.size_y, 1, 1, dst_buffer,
+                                  src_buffer, false, regs.dst_params.BlockHeight());
+    }
+}
+
+} // namespace Engines
+} // namespace Tegra

src/video_core/engines/maxwell_dma.h

@@ -0,0 +1,155 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include "common/assert.h"
+#include "common/bit_field.h"
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "video_core/gpu.h"
+#include "video_core/memory_manager.h"
+
+namespace Tegra {
+namespace Engines {
+
+class MaxwellDMA final {
+public:
+    explicit MaxwellDMA(MemoryManager& memory_manager);
+    ~MaxwellDMA() = default;
+
+    /// Write the value to the register identified by method.
+    void WriteReg(u32 method, u32 value);
+
+    struct Regs {
+        static constexpr size_t NUM_REGS = 0x1D6;
+
+        struct Parameters {
+            union {
+                BitField<0, 4, u32> block_depth;
+                BitField<4, 4, u32> block_height;
+                BitField<8, 4, u32> block_width;
+            };
+            u32 size_x;
+            u32 size_y;
+            u32 size_z;
+            u32 pos_z;
+            union {
+                BitField<0, 16, u32> pos_x;
+                BitField<16, 16, u32> pos_y;
+            };
+
+            u32 BlockHeight() const {
+                return 1 << block_height;
+            }
+        };
+
+        static_assert(sizeof(Parameters) == 24, "Parameters has wrong size");
+
+        enum class CopyMode : u32 {
+            None = 0,
+            Unk1 = 1,
+            Unk2 = 2,
+        };
+
+        enum class QueryMode : u32 {
+            None = 0,
+            Short = 1,
+            Long = 2,
+        };
+
+        enum class QueryIntr : u32 {
+            None = 0,
+            Block = 1,
+            NonBlock = 2,
+        };
+
+        union {
+            struct {
+                INSERT_PADDING_WORDS(0xC0);
+
+                struct {
+                    union {
+                        BitField<0, 2, CopyMode> copy_mode;
+                        BitField<2, 1, u32> flush;
+
+                        BitField<3, 2, QueryMode> query_mode;
+                        BitField<5, 2, QueryIntr> query_intr;
+
+                        BitField<7, 1, u32> is_src_linear;
+                        BitField<8, 1, u32> is_dst_linear;
+
+                        BitField<9, 1, u32> enable_2d;
+                        BitField<10, 1, u32> enable_swizzle;
+                    };
+                } exec;
+
+                INSERT_PADDING_WORDS(0x3F);
+
+                struct {
+                    u32 address_high;
+                    u32 address_low;
+
+                    GPUVAddr Address() const {
+                        return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) |
+                                                     address_low);
+                    }
+                } src_address;
+
+                struct {
+                    u32 address_high;
+                    u32 address_low;
+
+                    GPUVAddr Address() const {
+                        return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) |
+                                                     address_low);
+                    }
+                } dst_address;
+
+                u32 src_pitch;
+                u32 dst_pitch;
+                u32 x_count;
+                u32 y_count;
+
+                INSERT_PADDING_WORDS(0xBB);
+
+                Parameters dst_params;
+
+                INSERT_PADDING_WORDS(1);
+
+                Parameters src_params;
+
+                INSERT_PADDING_WORDS(0x13);
+            };
+            std::array<u32, NUM_REGS> reg_array;
+        };
+    } regs{};
+
+    MemoryManager& memory_manager;
+
+private:
+    /// Performs the copy from the source buffer to the destination buffer as configured in the
+    /// registers.
+    void HandleCopy();
+};
+
+#define ASSERT_REG_POSITION(field_name, position)                                                  \
+    static_assert(offsetof(MaxwellDMA::Regs, field_name) == position * 4,                          \
+                  "Field " #field_name " has invalid position")
+
+ASSERT_REG_POSITION(exec, 0xC0);
+ASSERT_REG_POSITION(src_address, 0x100);
+ASSERT_REG_POSITION(dst_address, 0x102);
+ASSERT_REG_POSITION(src_pitch, 0x104);
+ASSERT_REG_POSITION(dst_pitch, 0x105);
+ASSERT_REG_POSITION(x_count, 0x106);
+ASSERT_REG_POSITION(y_count, 0x107);
+ASSERT_REG_POSITION(dst_params, 0x1C3);
+ASSERT_REG_POSITION(src_params, 0x1CA);
+
+#undef ASSERT_REG_POSITION
+
+} // namespace Engines
+} // namespace Tegra

src/video_core/engines/shader_bytecode.h

@@ -109,11 +109,6 @@ union Sampler {
     u64 value{};
 };
 
-union Uniform {
-    BitField<20, 14, u64> offset;
-    BitField<34, 5, u64> index;
-};
-
 } // namespace Shader
 } // namespace Tegra
 
@@ -173,6 +168,31 @@ enum class SubOp : u64 {
     Min = 0x8,
 };
 
+enum class F2iRoundingOp : u64 {
+    None = 0,
+    Floor = 1,
+    Ceil = 2,
+    Trunc = 3,
+};
+
+enum class F2fRoundingOp : u64 {
+    None = 0,
+    Pass = 3,
+    Round = 8,
+    Floor = 9,
+    Ceil = 10,
+    Trunc = 11,
+};
+
+enum class UniformType : u64 {
+    UnsignedByte = 0,
+    SignedByte = 1,
+    UnsignedShort = 2,
+    SignedShort = 3,
+    Single = 4,
+    Double = 5,
+};
+
 union Instruction {
     Instruction& operator=(const Instruction& instr) {
         value = instr.value;
@@ -196,12 +216,12 @@ union Instruction {
 
     union {
         BitField<20, 19, u64> imm20_19;
-        BitField<20, 32, u64> imm20_32;
+        BitField<20, 32, s64> imm20_32;
         BitField<45, 1, u64> negate_b;
         BitField<46, 1, u64> abs_a;
         BitField<48, 1, u64> negate_a;
         BitField<49, 1, u64> abs_b;
-        BitField<50, 1, u64> abs_d;
+        BitField<50, 1, u64> saturate_d;
         BitField<56, 1, u64> negate_imm;
 
         union {
@@ -209,11 +229,19 @@ union Instruction {
             BitField<42, 1, u64> negate_pred;
         } fmnmx;
 
+        union {
+            BitField<39, 1, u64> invert_a;
+            BitField<40, 1, u64> invert_b;
+            BitField<41, 2, LogicOperation> operation;
+            BitField<44, 2, u64> unk44;
+            BitField<48, 3, Pred> pred48;
+        } lop;
+
         union {
             BitField<53, 2, LogicOperation> operation;
             BitField<55, 1, u64> invert_a;
             BitField<56, 1, u64> invert_b;
-        } lop;
+        } lop32i;
 
         float GetImm20_19() const {
             float result{};
@@ -226,33 +254,55 @@ union Instruction {
 
         float GetImm20_32() const {
             float result{};
-            u32 imm{static_cast<u32>(imm20_32)};
+            s32 imm{static_cast<s32>(imm20_32)};
             std::memcpy(&result, &imm, sizeof(imm));
             return result;
         }
-    } alu;
 
-    union {
-        BitField<39, 5, u64> shift_amount;
-        BitField<20, 19, u64> immediate_low;
-        BitField<56, 1, u64> immediate_high;
-        BitField<48, 1, u64> negate_b;
-        BitField<49, 1, u64> negate_a;
-
-        s32 GetImmediate() const {
-            u32 immediate = static_cast<u32>(immediate_low | (immediate_high << 19));
+        s32 GetSignedImm20_20() const {
+            u32 immediate = static_cast<u32>(imm20_19 | (negate_imm << 19));
             // Sign extend the 20-bit value.
             u32 mask = 1U << (20 - 1);
             return static_cast<s32>((immediate ^ mask) - mask);
         }
+    } alu;
 
-    } iscadd;
+    union {
+        BitField<48, 1, u64> is_signed;
+    } shift;
+
+    union {
+        BitField<39, 5, u64> shift_amount;
+        BitField<48, 1, u64> negate_b;
+        BitField<49, 1, u64> negate_a;
+    } alu_integer;
+
+    union {
+        BitField<54, 1, u64> saturate;
+        BitField<56, 1, u64> negate_a;
+    } iadd32i;
+
+    union {
+        BitField<20, 8, u64> shift_position;
+        BitField<28, 8, u64> shift_length;
+        BitField<48, 1, u64> negate_b;
+        BitField<49, 1, u64> negate_a;
+
+        u64 GetLeftShiftValue() const {
+            return 32 - (shift_position + shift_length);
+        }
+    } bfe;
 
     union {
         BitField<48, 1, u64> negate_b;
         BitField<49, 1, u64> negate_c;
     } ffma;
 
+    union {
+        BitField<48, 3, UniformType> type;
+        BitField<44, 2, u64> unknown;
+    } ld_c;
+
     union {
         BitField<0, 3, u64> pred0;
         BitField<3, 3, u64> pred3;
@@ -292,19 +342,37 @@ union Instruction {
     } fset;
 
     union {
-        BitField<10, 2, Register::Size> size;
-        BitField<13, 1, u64> is_signed;
+        BitField<39, 3, u64> pred39;
+        BitField<42, 1, u64> neg_pred;
+        BitField<44, 1, u64> bf;
+        BitField<45, 2, PredOperation> op;
+        BitField<48, 1, u64> is_signed;
+        BitField<49, 3, PredCondition> cond;
+    } iset;
+
+    union {
+        BitField<8, 2, Register::Size> dest_size;
+        BitField<10, 2, Register::Size> src_size;
+        BitField<12, 1, u64> is_output_signed;
+        BitField<13, 1, u64> is_input_signed;
         BitField<41, 2, u64> selector;
         BitField<45, 1, u64> negate_a;
         BitField<49, 1, u64> abs_a;
-        BitField<50, 1, u64> saturate_a;
+
+        union {
+            BitField<39, 2, F2iRoundingOp> rounding;
+        } f2i;
+
+        union {
+            BitField<39, 4, F2fRoundingOp> rounding;
+        } f2f;
     } conversion;
 
     union {
         BitField<31, 4, u64> component_mask;
 
         bool IsComponentEnabled(size_t component) const {
-            return ((1 << component) & component_mask) != 0;
+            return ((1ull << component) & component_mask) != 0;
         }
     } tex;
 
@@ -323,29 +391,39 @@ union Instruction {
 
             ASSERT(component_mask_selector < mask.size());
 
-            return ((1 << component) & mask[component_mask_selector]) != 0;
+            return ((1ull << component) & mask[component_mask_selector]) != 0;
         }
     } texs;
 
     union {
-        BitField<20, 5, u64> target;
+        BitField<20, 24, u64> target;
         BitField<5, 1, u64> constant_buffer;
 
         s32 GetBranchTarget() const {
             // Sign extend the branch target offset
-            u32 mask = 1U << (5 - 1);
+            u32 mask = 1U << (24 - 1);
             u32 value = static_cast<u32>(target);
-            // The branch offset is relative to the next instruction, so add 1 to it.
-            return static_cast<s32>((value ^ mask) - mask) + 1;
+            // The branch offset is relative to the next instruction and is stored in bytes, so
+            // divide it by the size of an instruction and add 1 to it.
+            return static_cast<s32>((value ^ mask) - mask) / sizeof(Instruction) + 1;
         }
     } bra;
 
+    union {
+        BitField<20, 14, u64> offset;
+        BitField<34, 5, u64> index;
+    } cbuf34;
+
+    union {
+        BitField<20, 16, s64> offset;
+        BitField<36, 5, u64> index;
+    } cbuf36;
+
     BitField<61, 1, u64> is_b_imm;
     BitField<60, 1, u64> is_b_gpr;
     BitField<59, 1, u64> is_c_gpr;
 
     Attribute attribute;
-    Uniform uniform;
     Sampler sampler;
 
     u64 value;
@@ -358,8 +436,13 @@ class OpCode {
 public:
     enum class Id {
         KIL,
+        SSY,
+        BFE_C,
+        BFE_R,
+        BFE_IMM,
         BRA,
         LD_A,
+        LD_C,
         ST_A,
         TEX,
         TEXQ, // Texture Query
@@ -378,6 +461,10 @@ public:
         FMUL_R,
         FMUL_IMM,
         FMUL32_IMM,
+        IADD_C,
+        IADD_R,
+        IADD_IMM,
+        IADD32I,
         ISCADD_C, // Scale and Add
         ISCADD_R,
         ISCADD_IMM,
@@ -397,17 +484,26 @@ public:
         I2I_C,
         I2I_R,
         I2I_IMM,
+        LOP_C,
+        LOP_R,
+        LOP_IMM,
         LOP32I,
         MOV_C,
         MOV_R,
         MOV_IMM,
         MOV32_IMM,
+        SHL_C,
+        SHL_R,
+        SHL_IMM,
         SHR_C,
         SHR_R,
         SHR_IMM,
         FMNMX_C,
         FMNMX_R,
         FMNMX_IMM,
+        IMNMX_C,
+        IMNMX_R,
+        IMNMX_IMM,
         FSETP_C, // Set Predicate
         FSETP_R,
         FSETP_IMM,
@@ -417,19 +513,30 @@ public:
         ISETP_C,
         ISETP_IMM,
         ISETP_R,
+        ISET_R,
+        ISET_C,
+        ISET_IMM,
         PSETP,
+        XMAD_IMM,
+        XMAD_CR,
+        XMAD_RC,
+        XMAD_RR,
     };
 
     enum class Type {
         Trivial,
         Arithmetic,
-        Logic,
-        ScaledAdd,
+        ArithmeticImmediate,
+        ArithmeticInteger,
+        ArithmeticIntegerImmediate,
+        Bfe,
+        Shift,
         Ffma,
         Flow,
         Memory,
         FloatSet,
         FloatSetPredicate,
+        IntegerSet,
         IntegerSetPredicate,
         PredicateSetPredicate,
         Conversion,
@@ -528,8 +635,10 @@ private:
         std::vector<Matcher> table = {
 #define INST(bitstring, op, type, name) Detail::GetMatcher(bitstring, op, type, name)
             INST("111000110011----", Id::KIL, Type::Flow, "KIL"),
+            INST("111000101001----", Id::SSY, Type::Flow, "SSY"),
             INST("111000100100----", Id::BRA, Type::Flow, "BRA"),
             INST("1110111111011---", Id::LD_A, Type::Memory, "LD_A"),
+            INST("1110111110010---", Id::LD_C, Type::Memory, "LD_C"),
             INST("1110111111110---", Id::ST_A, Type::Memory, "ST_A"),
             INST("1100000000111---", Id::TEX, Type::Memory, "TEX"),
             INST("1101111101001---", Id::TEXQ, Type::Memory, "TEXQ"),
@@ -547,10 +656,14 @@ private:
             INST("0100110001101---", Id::FMUL_C, Type::Arithmetic, "FMUL_C"),
             INST("0101110001101---", Id::FMUL_R, Type::Arithmetic, "FMUL_R"),
             INST("0011100-01101---", Id::FMUL_IMM, Type::Arithmetic, "FMUL_IMM"),
-            INST("00011110--------", Id::FMUL32_IMM, Type::Arithmetic, "FMUL32_IMM"),
-            INST("0100110000011---", Id::ISCADD_C, Type::ScaledAdd, "ISCADD_C"),
-            INST("0101110000011---", Id::ISCADD_R, Type::ScaledAdd, "ISCADD_R"),
-            INST("0011100-00011---", Id::ISCADD_IMM, Type::ScaledAdd, "ISCADD_IMM"),
+            INST("00011110--------", Id::FMUL32_IMM, Type::ArithmeticImmediate, "FMUL32_IMM"),
+            INST("0100110000010---", Id::IADD_C, Type::ArithmeticInteger, "IADD_C"),
+            INST("0101110000010---", Id::IADD_R, Type::ArithmeticInteger, "IADD_R"),
+            INST("0011100-00010---", Id::IADD_IMM, Type::ArithmeticInteger, "IADD_IMM"),
+            INST("0001110---------", Id::IADD32I, Type::ArithmeticIntegerImmediate, "IADD32I"),
+            INST("0100110000011---", Id::ISCADD_C, Type::ArithmeticInteger, "ISCADD_C"),
+            INST("0101110000011---", Id::ISCADD_R, Type::ArithmeticInteger, "ISCADD_R"),
+            INST("0011100-00011---", Id::ISCADD_IMM, Type::ArithmeticInteger, "ISCADD_IMM"),
             INST("0101000010000---", Id::MUFU, Type::Arithmetic, "MUFU"),
             INST("0100110010010---", Id::RRO_C, Type::Arithmetic, "RRO_C"),
             INST("0101110010010---", Id::RRO_R, Type::Arithmetic, "RRO_R"),
@@ -558,20 +671,32 @@ private:
             INST("0100110010101---", Id::F2F_C, Type::Conversion, "F2F_C"),
             INST("0101110010101---", Id::F2F_R, Type::Conversion, "F2F_R"),
             INST("0011100-10101---", Id::F2F_IMM, Type::Conversion, "F2F_IMM"),
-            INST("0100110010110---", Id::F2I_C, Type::Arithmetic, "F2I_C"),
-            INST("0101110010110---", Id::F2I_R, Type::Arithmetic, "F2I_R"),
-            INST("0011100-10110---", Id::F2I_IMM, Type::Arithmetic, "F2I_IMM"),
+            INST("0100110010110---", Id::F2I_C, Type::Conversion, "F2I_C"),
+            INST("0101110010110---", Id::F2I_R, Type::Conversion, "F2I_R"),
+            INST("0011100-10110---", Id::F2I_IMM, Type::Conversion, "F2I_IMM"),
             INST("0100110010011---", Id::MOV_C, Type::Arithmetic, "MOV_C"),
             INST("0101110010011---", Id::MOV_R, Type::Arithmetic, "MOV_R"),
             INST("0011100-10011---", Id::MOV_IMM, Type::Arithmetic, "MOV_IMM"),
-            INST("000000010000----", Id::MOV32_IMM, Type::Arithmetic, "MOV32_IMM"),
-            INST("0100110000101---", Id::SHR_C, Type::Arithmetic, "SHR_C"),
-            INST("0101110000101---", Id::SHR_R, Type::Arithmetic, "SHR_R"),
-            INST("0011100-00101---", Id::SHR_IMM, Type::Arithmetic, "SHR_IMM"),
+            INST("000000010000----", Id::MOV32_IMM, Type::ArithmeticImmediate, "MOV32_IMM"),
             INST("0100110001100---", Id::FMNMX_C, Type::Arithmetic, "FMNMX_C"),
             INST("0101110001100---", Id::FMNMX_R, Type::Arithmetic, "FMNMX_R"),
             INST("0011100-01100---", Id::FMNMX_IMM, Type::Arithmetic, "FMNMX_IMM"),
-            INST("000001----------", Id::LOP32I, Type::Logic, "LOP32I"),
+            INST("0100110000100---", Id::IMNMX_C, Type::Arithmetic, "FMNMX_IMM"),
+            INST("0101110000100---", Id::IMNMX_R, Type::Arithmetic, "FMNMX_IMM"),
+            INST("0011100-00100---", Id::IMNMX_IMM, Type::Arithmetic, "FMNMX_IMM"),
+            INST("0100110000000---", Id::BFE_C, Type::Bfe, "BFE_C"),
+            INST("0101110000000---", Id::BFE_R, Type::Bfe, "BFE_R"),
+            INST("0011100-00000---", Id::BFE_IMM, Type::Bfe, "BFE_IMM"),
+            INST("0100110001000---", Id::LOP_C, Type::ArithmeticInteger, "LOP_C"),
+            INST("0101110001000---", Id::LOP_R, Type::ArithmeticInteger, "LOP_R"),
+            INST("0011100001000---", Id::LOP_IMM, Type::ArithmeticInteger, "LOP_IMM"),
+            INST("000001----------", Id::LOP32I, Type::ArithmeticIntegerImmediate, "LOP32I"),
+            INST("0100110001001---", Id::SHL_C, Type::Shift, "SHL_C"),
+            INST("0101110001001---", Id::SHL_R, Type::Shift, "SHL_R"),
+            INST("0011100-01001---", Id::SHL_IMM, Type::Shift, "SHL_IMM"),
+            INST("0100110000101---", Id::SHR_C, Type::Shift, "SHR_C"),
+            INST("0101110000101---", Id::SHR_R, Type::Shift, "SHR_R"),
+            INST("0011100-00101---", Id::SHR_IMM, Type::Shift, "SHR_IMM"),
             INST("0100110011100---", Id::I2I_C, Type::Conversion, "I2I_C"),
             INST("0101110011100---", Id::I2I_R, Type::Conversion, "I2I_R"),
             INST("01110001-1000---", Id::I2I_IMM, Type::Conversion, "I2I_IMM"),
@@ -587,7 +712,14 @@ private:
             INST("010010110110----", Id::ISETP_C, Type::IntegerSetPredicate, "ISETP_C"),
             INST("010110110110----", Id::ISETP_R, Type::IntegerSetPredicate, "ISETP_R"),
             INST("0011011-0110----", Id::ISETP_IMM, Type::IntegerSetPredicate, "ISETP_IMM"),
+            INST("010110110101----", Id::ISET_R, Type::IntegerSet, "ISET_R"),
+            INST("010010110101----", Id::ISET_C, Type::IntegerSet, "ISET_C"),
+            INST("0011011-0101----", Id::ISET_IMM, Type::IntegerSet, "ISET_IMM"),
             INST("0101000010010---", Id::PSETP, Type::PredicateSetPredicate, "PSETP"),
+            INST("0011011-00------", Id::XMAD_IMM, Type::Arithmetic, "XMAD_IMM"),
+            INST("0100111---------", Id::XMAD_CR, Type::Arithmetic, "XMAD_CR"),
+            INST("010100010-------", Id::XMAD_RC, Type::Arithmetic, "XMAD_RC"),
+            INST("0101101100------", Id::XMAD_RR, Type::Arithmetic, "XMAD_RR"),
         };
 #undef INST
         std::stable_sort(table.begin(), table.end(), [](const auto& a, const auto& b) {

src/video_core/gpu.cpp

@@ -5,6 +5,7 @@
 #include "video_core/engines/fermi_2d.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/engines/maxwell_compute.h"
+#include "video_core/engines/maxwell_dma.h"
 #include "video_core/gpu.h"
 
 namespace Tegra {
@@ -14,6 +15,7 @@ GPU::GPU() {
     maxwell_3d = std::make_unique<Engines::Maxwell3D>(*memory_manager);
     fermi_2d = std::make_unique<Engines::Fermi2D>(*memory_manager);
     maxwell_compute = std::make_unique<Engines::MaxwellCompute>();
+    maxwell_dma = std::make_unique<Engines::MaxwellDMA>(*memory_manager);
 }
 
 GPU::~GPU() = default;
@@ -26,6 +28,10 @@ u32 RenderTargetBytesPerPixel(RenderTargetFormat format) {
     ASSERT(format != RenderTargetFormat::NONE);
 
     switch (format) {
+    case RenderTargetFormat::RGBA32_FLOAT:
+        return 16;
+    case RenderTargetFormat::RGBA16_FLOAT:
+        return 8;
     case RenderTargetFormat::RGBA8_UNORM:
     case RenderTargetFormat::RGB10_A2_UNORM:
         return 4;

src/video_core/gpu.h

@@ -15,10 +15,12 @@ namespace Tegra {
 
 enum class RenderTargetFormat : u32 {
     NONE = 0x0,
+    RGBA32_FLOAT = 0xC0,
     RGBA16_FLOAT = 0xCA,
     RGB10_A2_UNORM = 0xD1,
     RGBA8_UNORM = 0xD5,
     RGBA8_SRGB = 0xD6,
+    R11G11B10_FLOAT = 0xE0,
 };
 
 /// Returns the number of bytes per pixel of each rendertarget format.
@@ -61,6 +63,7 @@ namespace Engines {
 class Fermi2D;
 class Maxwell3D;
 class MaxwellCompute;
+class MaxwellDMA;
 } // namespace Engines
 
 enum class EngineID {
@@ -101,6 +104,8 @@ private:
     std::unique_ptr<Engines::Fermi2D> fermi_2d;
     /// Compute engine
     std::unique_ptr<Engines::MaxwellCompute> maxwell_compute;
+    /// DMA engine
+    std::unique_ptr<Engines::MaxwellDMA> maxwell_dma;
 };
 
 } // namespace Tegra

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -196,8 +196,10 @@ 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 bindpoint to use when uploading the const buffers to the GLSL shaders.
+    // Next available bindpoints to use when uploading the const buffers and textures to the GLSL
+    // shaders.
     u32 current_constbuffer_bindpoint = 0;
+    u32 current_texture_bindpoint = 0;
 
     for (unsigned index = 1; index < Maxwell::MaxShaderProgram; ++index) {
         auto& shader_config = gpu.regs.shader_config[index];
@@ -212,13 +214,17 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
             continue;
         }
 
+        GLShader::MaxwellUniformData ubo{};
+        ubo.SetFromRegs(gpu.state.shader_stages[stage]);
+        std::memcpy(buffer_ptr, &ubo, sizeof(ubo));
+
+        // Flush the buffer so that the GPU can see the data we just wrote.
+        glFlushMappedBufferRange(GL_ARRAY_BUFFER, buffer_offset, sizeof(ubo));
+
         // Upload uniform data as one UBO per stage
         const GLintptr ubo_offset = buffer_offset;
         copy_buffer(uniform_buffers[stage].handle, ubo_offset,
                     sizeof(GLShader::MaxwellUniformData));
-        GLShader::MaxwellUniformData* ub_ptr =
-            reinterpret_cast<GLShader::MaxwellUniformData*>(buffer_ptr);
-        ub_ptr->SetFromRegs(gpu.state.shader_stages[stage]);
 
         buffer_ptr += sizeof(GLShader::MaxwellUniformData);
         buffer_offset += sizeof(GLShader::MaxwellUniformData);
@@ -258,6 +264,11 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset) {
         current_constbuffer_bindpoint =
             SetupConstBuffers(static_cast<Maxwell::ShaderStage>(stage), gl_stage_program,
                               current_constbuffer_bindpoint, shader_resources.const_buffer_entries);
+
+        // Configure the textures for this shader stage.
+        current_texture_bindpoint =
+            SetupTextures(static_cast<Maxwell::ShaderStage>(stage), gl_stage_program,
+                          current_texture_bindpoint, shader_resources.texture_samplers);
     }
 
     shader_program_manager->UseTrivialGeometryShader();
@@ -338,12 +349,12 @@ void RasterizerOpenGL::DrawArrays() {
     // Sync the viewport
     SyncViewport(surfaces_rect, res_scale);
 
+    // Sync the blend state registers
+    SyncBlendState();
+
     // TODO(bunnei): Sync framebuffer_scale uniform here
     // TODO(bunnei): Sync scissorbox uniform(s) here
 
-    // Sync and bind the texture surfaces
-    BindTextures();
-
     // 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;
@@ -447,65 +458,7 @@ void RasterizerOpenGL::DrawArrays() {
     }
 }
 
-void RasterizerOpenGL::BindTextures() {
-    using Regs = Tegra::Engines::Maxwell3D::Regs;
-    auto& maxwell3d = Core::System::GetInstance().GPU().Get3DEngine();
-
-    // Each Maxwell shader stage can have an arbitrary number of textures, but we're limited to a
-    // certain number in OpenGL. We try to only use the minimum amount of host textures by not
-    // keeping a 1:1 relation between guest texture ids and host texture ids, ie, guest texture id 8
-    // can be host texture id 0 if it's the only texture used in the guest shader program.
-    u32 host_texture_index = 0;
-    for (u32 stage = 0; stage < Regs::MaxShaderStage; ++stage) {
-        ASSERT(host_texture_index < texture_samplers.size());
-        const auto textures = maxwell3d.GetStageTextures(static_cast<Regs::ShaderStage>(stage));
-        for (unsigned texture_index = 0; texture_index < textures.size(); ++texture_index) {
-            const auto& texture = textures[texture_index];
-
-            if (texture.enabled) {
-                texture_samplers[host_texture_index].SyncWithConfig(texture.tsc);
-                Surface surface = res_cache.GetTextureSurface(texture);
-                if (surface != nullptr) {
-                    state.texture_units[host_texture_index].texture_2d = surface->texture.handle;
-                } else {
-                    // Can occur when texture addr is null or its memory is unmapped/invalid
-                    state.texture_units[texture_index].texture_2d = 0;
-                }
-
-                ++host_texture_index;
-            } else {
-                state.texture_units[texture_index].texture_2d = 0;
-            }
-        }
-    }
-}
-
-void RasterizerOpenGL::NotifyMaxwellRegisterChanged(u32 method) {
-    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
-    switch (method) {
-    case MAXWELL3D_REG_INDEX(blend.separate_alpha):
-        ASSERT_MSG(false, "unimplemented");
-        break;
-    case MAXWELL3D_REG_INDEX(blend.equation_rgb):
-        state.blend.rgb_equation = MaxwellToGL::BlendEquation(regs.blend.equation_rgb);
-        break;
-    case MAXWELL3D_REG_INDEX(blend.factor_source_rgb):
-        state.blend.src_rgb_func = MaxwellToGL::BlendFunc(regs.blend.factor_source_rgb);
-        break;
-    case MAXWELL3D_REG_INDEX(blend.factor_dest_rgb):
-        state.blend.dst_rgb_func = MaxwellToGL::BlendFunc(regs.blend.factor_dest_rgb);
-        break;
-    case MAXWELL3D_REG_INDEX(blend.equation_a):
-        state.blend.a_equation = MaxwellToGL::BlendEquation(regs.blend.equation_a);
-        break;
-    case MAXWELL3D_REG_INDEX(blend.factor_source_a):
-        state.blend.src_a_func = MaxwellToGL::BlendFunc(regs.blend.factor_source_a);
-        break;
-    case MAXWELL3D_REG_INDEX(blend.factor_dest_a):
-        state.blend.dst_a_func = MaxwellToGL::BlendFunc(regs.blend.factor_dest_a);
-        break;
-    }
-}
+void RasterizerOpenGL::NotifyMaxwellRegisterChanged(u32 method) {}
 
 void RasterizerOpenGL::FlushAll() {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
@@ -654,7 +607,16 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, GLuint progr
         buffer_draw_state.bindpoint = current_bindpoint + bindpoint;
 
         boost::optional<VAddr> addr = gpu.memory_manager->GpuToCpuAddress(buffer.address);
-        std::vector<u8> data(used_buffer.GetSize() * sizeof(float));
+
+        std::vector<u8> data;
+        if (used_buffer.IsIndirect()) {
+            // Buffer is accessed indirectly, so upload the entire thing
+            data.resize(buffer.size * sizeof(float));
+        } else {
+            // Buffer is accessed directly, upload just what we use
+            data.resize(used_buffer.GetSize() * sizeof(float));
+        }
+
         Memory::ReadBlock(*addr, data.data(), data.size());
 
         glBindBuffer(GL_SHADER_STORAGE_BUFFER, buffer_draw_state.ssbo);
@@ -671,7 +633,53 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, GLuint progr
 
     state.Apply();
 
-    return current_bindpoint + entries.size();
+    return current_bindpoint + static_cast<u32>(entries.size());
+}
+
+u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, GLuint program, u32 current_unit,
+                                    const std::vector<GLShader::SamplerEntry>& entries) {
+    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.");
+
+    for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
+        const auto& entry = entries[bindpoint];
+        u32 current_bindpoint = current_unit + bindpoint;
+
+        // Bind the uniform to the sampler.
+        GLint uniform = glGetUniformLocation(program, entry.GetName().c_str());
+        ASSERT(uniform != -1);
+        glProgramUniform1i(program, uniform, current_bindpoint);
+
+        const auto texture = maxwell3d.GetStageTexture(entry.GetStage(), entry.GetOffset());
+        ASSERT(texture.enabled);
+
+        texture_samplers[current_bindpoint].SyncWithConfig(texture.tsc);
+        Surface surface = res_cache.GetTextureSurface(texture);
+        if (surface != nullptr) {
+            state.texture_units[current_bindpoint].texture_2d = surface->texture.handle;
+            state.texture_units[current_bindpoint].swizzle.r =
+                MaxwellToGL::SwizzleSource(texture.tic.x_source);
+            state.texture_units[current_bindpoint].swizzle.g =
+                MaxwellToGL::SwizzleSource(texture.tic.y_source);
+            state.texture_units[current_bindpoint].swizzle.b =
+                MaxwellToGL::SwizzleSource(texture.tic.z_source);
+            state.texture_units[current_bindpoint].swizzle.a =
+                MaxwellToGL::SwizzleSource(texture.tic.w_source);
+        } else {
+            // Can occur when texture addr is null or its memory is unmapped/invalid
+            state.texture_units[current_bindpoint].texture_2d = 0;
+        }
+    }
+
+    state.Apply();
+
+    return current_unit + static_cast<u32>(entries.size());
 }
 
 void RasterizerOpenGL::BindFramebufferSurfaces(const Surface& color_surface,
@@ -730,14 +738,21 @@ void RasterizerOpenGL::SyncDepthOffset() {
     UNREACHABLE();
 }
 
-void RasterizerOpenGL::SyncBlendEnabled() {
-    UNREACHABLE();
-}
+void RasterizerOpenGL::SyncBlendState() {
+    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
 
-void RasterizerOpenGL::SyncBlendFuncs() {
-    UNREACHABLE();
-}
+    // TODO(Subv): Support more than just render target 0.
+    state.blend.enabled = regs.blend.enable[0] != 0;
 
-void RasterizerOpenGL::SyncBlendColor() {
-    UNREACHABLE();
+    if (!state.blend.enabled)
+        return;
+
+    ASSERT_MSG(regs.independent_blend_enable == 1, "Only independent blending is implemented");
+    ASSERT_MSG(!regs.independent_blend[0].separate_alpha, "Unimplemented");
+    state.blend.rgb_equation = MaxwellToGL::BlendEquation(regs.independent_blend[0].equation_rgb);
+    state.blend.src_rgb_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_source_rgb);
+    state.blend.dst_rgb_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_dest_rgb);
+    state.blend.a_equation = MaxwellToGL::BlendEquation(regs.independent_blend[0].equation_a);
+    state.blend.src_a_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_source_a);
+    state.blend.dst_a_func = MaxwellToGL::BlendFunc(regs.independent_blend[0].factor_dest_a);
 }

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -80,9 +80,6 @@ private:
     void BindFramebufferSurfaces(const Surface& color_surface, const Surface& depth_surface,
                                  bool has_stencil);
 
-    /// Binds the required textures to OpenGL before drawing a batch.
-    void BindTextures();
-
     /*
      * Configures the current constbuffers to use for the draw command.
      * @param stage The shader stage to configure buffers for.
@@ -95,6 +92,17 @@ private:
                           u32 current_bindpoint,
                           const std::vector<GLShader::ConstBufferEntry>& entries);
 
+    /*
+     * Configures the current textures to use for the draw command.
+     * @param stage The shader stage to configure textures for.
+     * @param program The OpenGL program object that contains the specified stage.
+     * @param current_unit The offset at which to start counting unused texture units.
+     * @param entries Vector describing the textures that are actually used in the guest shader.
+     * @returns The next available bindpoint for use in the next shader stage.
+     */
+    u32 SetupTextures(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, GLuint program,
+                      u32 current_unit, const std::vector<GLShader::SamplerEntry>& entries);
+
     /// Syncs the viewport to match the guest state
     void SyncViewport(const MathUtil::Rectangle<u32>& surfaces_rect, u16 res_scale);
 
@@ -113,14 +121,8 @@ private:
     /// 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();
+    /// Syncs the blend state to match the guest state
+    void SyncBlendState();
 
     bool has_ARB_buffer_storage;
     bool has_ARB_direct_state_access;

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -28,6 +28,7 @@
 #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/textures/astc.h"
 #include "video_core/textures/decoders.h"
 #include "video_core/utils.h"
 #include "video_core/video_core.h"
@@ -50,18 +51,22 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_1_5_5_5_REV, false},                // A1B5G5R5
     {GL_R8, GL_RED, GL_UNSIGNED_BYTE, false},                                   // R8
     {GL_RGBA16F, GL_RGBA, GL_HALF_FLOAT, false},                                // RGBA16F
+    {GL_R11F_G11F_B10F, GL_RGB, GL_UNSIGNED_INT_10F_11F_11F_REV, false},        // R11FG11FB10F
     {GL_COMPRESSED_RGB_S3TC_DXT1_EXT, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, true},   // DXT1
     {GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT23
     {GL_COMPRESSED_RGBA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT45
     {GL_COMPRESSED_RED_RGTC1, GL_RED, GL_UNSIGNED_INT_8_8_8_8, true},           // DXN1
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, false},                               // ASTC_2D_4X4
 }};
 
 static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType component_type) {
     const SurfaceType type = SurfaceParams::GetFormatType(pixel_format);
     if (type == SurfaceType::ColorTexture) {
         ASSERT(static_cast<size_t>(pixel_format) < tex_format_tuples.size());
-        // For now only UNORM components are supported, or RGBA16F which is type FLOAT
-        ASSERT(component_type == ComponentType::UNorm || pixel_format == PixelFormat::RGBA16F);
+        // For now only UNORM components are supported, or either R11FG11FB10F or RGBA16F which are
+        // type FLOAT
+        ASSERT(component_type == ComponentType::UNorm || pixel_format == PixelFormat::RGBA16F ||
+               pixel_format == PixelFormat::R11FG11FB10F);
         return tex_format_tuples[static_cast<unsigned int>(pixel_format)];
     } else if (type == SurfaceType::Depth || type == SurfaceType::DepthStencil) {
         // TODO(Subv): Implement depth formats
@@ -83,6 +88,23 @@ static u16 GetResolutionScaleFactor() {
                                 : Settings::values.resolution_factor);
 }
 
+static void ConvertASTCToRGBA8(std::vector<u8>& data, PixelFormat format, u32 width, u32 height) {
+    u32 block_width{};
+    u32 block_height{};
+
+    switch (format) {
+    case PixelFormat::ASTC_2D_4X4:
+        block_width = 4;
+        block_height = 4;
+        break;
+    default:
+        NGLOG_CRITICAL(HW_GPU, "Unhandled format: {}", static_cast<u32>(format));
+        UNREACHABLE();
+    }
+
+    data = Tegra::Texture::ASTC::Decompress(data, width, height, block_width, block_height);
+}
+
 template <bool morton_to_gl, PixelFormat format>
 void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, Tegra::GPUVAddr base,
                 Tegra::GPUVAddr start, Tegra::GPUVAddr end) {
@@ -94,6 +116,12 @@ void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, Tegra::
         auto data = Tegra::Texture::UnswizzleTexture(
             *gpu.memory_manager->GpuToCpuAddress(base),
             SurfaceParams::TextureFormatFromPixelFormat(format), stride, height, block_height);
+
+        if (SurfaceParams::IsFormatASTC(format)) {
+            // ASTC formats are converted to RGBA8 in software, as most PC GPUs do not support this
+            ConvertASTCToRGBA8(data, format, stride, height);
+        }
+
         std::memcpy(gl_buffer, data.data(), data.size());
     } else {
         // TODO(bunnei): Assumes the default rendering GOB size of 16 (128 lines). We should check
@@ -110,11 +138,12 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr, Tegra:
                                      Tegra::GPUVAddr),
                             SurfaceParams::MaxPixelFormat>
     morton_to_gl_fns = {
-        MortonCopy<true, PixelFormat::ABGR8>,       MortonCopy<true, PixelFormat::B5G6R5>,
-        MortonCopy<true, PixelFormat::A2B10G10R10>, MortonCopy<true, PixelFormat::A1B5G5R5>,
-        MortonCopy<true, PixelFormat::R8>,          MortonCopy<true, PixelFormat::RGBA16F>,
-        MortonCopy<true, PixelFormat::DXT1>,        MortonCopy<true, PixelFormat::DXT23>,
-        MortonCopy<true, PixelFormat::DXT45>,       MortonCopy<true, PixelFormat::DXN1>,
+        MortonCopy<true, PixelFormat::ABGR8>,        MortonCopy<true, PixelFormat::B5G6R5>,
+        MortonCopy<true, PixelFormat::A2B10G10R10>,  MortonCopy<true, PixelFormat::A1B5G5R5>,
+        MortonCopy<true, PixelFormat::R8>,           MortonCopy<true, PixelFormat::RGBA16F>,
+        MortonCopy<true, PixelFormat::R11FG11FB10F>, MortonCopy<true, PixelFormat::DXT1>,
+        MortonCopy<true, PixelFormat::DXT23>,        MortonCopy<true, PixelFormat::DXT45>,
+        MortonCopy<true, PixelFormat::DXN1>,         MortonCopy<true, PixelFormat::ASTC_2D_4X4>,
 };
 
 static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr, Tegra::GPUVAddr,
@@ -127,11 +156,13 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr, Tegra:
         MortonCopy<false, PixelFormat::A1B5G5R5>,
         MortonCopy<false, PixelFormat::R8>,
         MortonCopy<false, PixelFormat::RGBA16F>,
+        MortonCopy<false, PixelFormat::R11FG11FB10F>,
         // TODO(Subv): Swizzling the DXT1/DXT23/DXT45/DXN1 formats is not yet supported
         nullptr,
         nullptr,
         nullptr,
         nullptr,
+        MortonCopy<false, PixelFormat::ABGR8>,
 };
 
 // Allocate an uninitialized texture of appropriate size and format for the surface
@@ -164,60 +195,10 @@ static void AllocateSurfaceTexture(GLuint texture, const FormatTuple& format_tup
 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::ColorTexture) {
-        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);
 
+    glCopyImageSubData(src_tex, GL_TEXTURE_2D, 0, src_rect.left, src_rect.bottom, 0, dst_tex,
+                       GL_TEXTURE_2D, 0, dst_rect.left, dst_rect.bottom, 0, src_rect.GetWidth(),
+                       src_rect.GetHeight(), 0);
     return true;
 }
 
@@ -594,7 +575,7 @@ void CachedSurface::UploadGLTexture(const MathUtil::Rectangle<u32>& rect, GLuint
         glCompressedTexImage2D(GL_TEXTURE_2D, 0, tuple.internal_format,
                                static_cast<GLsizei>(rect.GetWidth() * GetCompresssionFactor()),
                                static_cast<GLsizei>(rect.GetHeight() * GetCompresssionFactor()), 0,
-                               size, &gl_buffer[buffer_offset]);
+                               static_cast<GLsizei>(size), &gl_buffer[buffer_offset]);
     } else {
         glTexSubImage2D(GL_TEXTURE_2D, 0, x0, y0, static_cast<GLsizei>(rect.GetWidth()),
                         static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
@@ -933,9 +914,6 @@ Surface RasterizerCacheOpenGL::GetSurface(const SurfaceParams& params, ScaleMatc
     // Use GetSurfaceSubRect instead
     ASSERT(params.width == params.stride);
 
-    ASSERT(!params.is_tiled ||
-           (params.GetActualWidth() % 8 == 0 && params.GetActualHeight() % 8 == 0));
-
     // Check for an exact match in existing surfaces
     Surface surface =
         FindMatch<MatchFlags::Exact | MatchFlags::Invalid>(surface_cache, params, match_res_scale);
@@ -1078,8 +1056,11 @@ Surface RasterizerCacheOpenGL::GetTextureSurface(const Tegra::Texture::FullTextu
     params.addr = config.tic.Address();
     params.is_tiled = config.tic.IsTiled();
     params.pixel_format = SurfaceParams::PixelFormatFromTextureFormat(config.tic.format);
-    params.width = config.tic.Width() / params.GetCompresssionFactor();
-    params.height = config.tic.Height() / params.GetCompresssionFactor();
+
+    params.width = Common::AlignUp(config.tic.Width(), params.GetCompresssionFactor()) /
+                   params.GetCompresssionFactor();
+    params.height = Common::AlignUp(config.tic.Height(), params.GetCompresssionFactor()) /
+                    params.GetCompresssionFactor();
 
     // TODO(Subv): Different types per component are not supported.
     ASSERT(config.tic.r_type.Value() == config.tic.g_type.Value() &&
@@ -1090,6 +1071,13 @@ Surface RasterizerCacheOpenGL::GetTextureSurface(const Tegra::Texture::FullTextu
 
     if (config.tic.IsTiled()) {
         params.block_height = config.tic.BlockHeight();
+
+        // TODO(bunnei): The below align up is a hack. This is here because some compressed textures
+        // are not a multiple of their own compression factor, and so this accounts for that. This
+        // could potentially result in an extra row of 4px being decoded if a texture is not a
+        // multiple of 4.
+        params.width = Common::AlignUp(params.width, 4);
+        params.height = Common::AlignUp(params.height, 4);
     } else {
         // Use the texture-provided stride value if the texture isn't tiled.
         params.stride = static_cast<u32>(params.PixelsInBytes(config.tic.Pitch()));
@@ -1097,23 +1085,6 @@ Surface RasterizerCacheOpenGL::GetTextureSurface(const Tegra::Texture::FullTextu
 
     params.UpdateParams();
 
-    if (config.tic.Width() % 8 != 0 || config.tic.Height() % 8 != 0 ||
-        params.stride != params.width) {
-        Surface src_surface;
-        MathUtil::Rectangle<u32> rect;
-        std::tie(src_surface, rect) = GetSurfaceSubRect(params, ScaleMatch::Ignore, true);
-
-        params.res_scale = src_surface->res_scale;
-        Surface tmp_surface = CreateSurface(params);
-        BlitTextures(src_surface->texture.handle, rect, tmp_surface->texture.handle,
-                     tmp_surface->GetScaledRect(),
-                     SurfaceParams::GetFormatType(params.pixel_format), read_framebuffer.handle,
-                     draw_framebuffer.handle);
-
-        remove_surfaces.emplace(tmp_surface);
-        return tmp_surface;
-    }
-
     return GetSurface(params, ScaleMatch::Ignore, true);
 }
 
@@ -1288,7 +1259,7 @@ void RasterizerCacheOpenGL::ValidateSurface(const Surface& surface, Tegra::GPUVA
 
         const auto interval = *it & validate_interval;
         // Look for a valid surface to copy from
-        SurfaceParams params = surface->FromInterval(interval);
+        SurfaceParams params = *surface;
 
         Surface copy_surface =
             FindMatch<MatchFlags::Copy>(surface_cache, params, ScaleMatch::Ignore, interval);

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -60,10 +60,12 @@ struct SurfaceParams {
         A1B5G5R5 = 3,
         R8 = 4,
         RGBA16F = 5,
-        DXT1 = 6,
-        DXT23 = 7,
-        DXT45 = 8,
-        DXN1 = 9, // This is also known as BC4
+        R11FG11FB10F = 6,
+        DXT1 = 7,
+        DXT23 = 8,
+        DXT45 = 9,
+        DXN1 = 10, // This is also known as BC4
+        ASTC_2D_4X4 = 11,
 
         Max,
         Invalid = 255,
@@ -104,11 +106,13 @@ struct SurfaceParams {
             1, // A2B10G10R10
             1, // A1B5G5R5
             1, // R8
-            2, // RGBA16F
+            1, // RGBA16F
+            1, // R11FG11FB10F
             4, // DXT1
             4, // DXT23
             4, // DXT45
             4, // DXN1
+            1, // ASTC_2D_4X4
         }};
 
         ASSERT(static_cast<size_t>(format) < compression_factor_table.size());
@@ -129,10 +133,12 @@ struct SurfaceParams {
             16,  // A1B5G5R5
             8,   // R8
             64,  // RGBA16F
+            32,  // R11FG11FB10F
             64,  // DXT1
             128, // DXT23
             128, // DXT45
             64,  // DXN1
+            32,  // ASTC_2D_4X4
         }};
 
         ASSERT(static_cast<size_t>(format) < bpp_table.size());
@@ -151,12 +157,23 @@ struct SurfaceParams {
             return PixelFormat::A2B10G10R10;
         case Tegra::RenderTargetFormat::RGBA16_FLOAT:
             return PixelFormat::RGBA16F;
+        case Tegra::RenderTargetFormat::R11G11B10_FLOAT:
+            return PixelFormat::R11FG11FB10F;
         default:
             NGLOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
         }
     }
 
+    static bool IsFormatASTC(PixelFormat format) {
+        switch (format) {
+        case PixelFormat::ASTC_2D_4X4:
+            return true;
+        default:
+            return false;
+        }
+    }
+
     static PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat format) {
         switch (format) {
         case Tegra::FramebufferConfig::PixelFormat::ABGR8:
@@ -182,6 +199,8 @@ struct SurfaceParams {
             return PixelFormat::R8;
         case Tegra::Texture::TextureFormat::R16_G16_B16_A16:
             return PixelFormat::RGBA16F;
+        case Tegra::Texture::TextureFormat::BF10GF11RF11:
+            return PixelFormat::R11FG11FB10F;
         case Tegra::Texture::TextureFormat::DXT1:
             return PixelFormat::DXT1;
         case Tegra::Texture::TextureFormat::DXT23:
@@ -190,6 +209,8 @@ struct SurfaceParams {
             return PixelFormat::DXT45;
         case Tegra::Texture::TextureFormat::DXN1:
             return PixelFormat::DXN1;
+        case Tegra::Texture::TextureFormat::ASTC_2D_4X4:
+            return PixelFormat::ASTC_2D_4X4;
         default:
             NGLOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
@@ -211,6 +232,8 @@ struct SurfaceParams {
             return Tegra::Texture::TextureFormat::R8;
         case PixelFormat::RGBA16F:
             return Tegra::Texture::TextureFormat::R16_G16_B16_A16;
+        case PixelFormat::R11FG11FB10F:
+            return Tegra::Texture::TextureFormat::BF10GF11RF11;
         case PixelFormat::DXT1:
             return Tegra::Texture::TextureFormat::DXT1;
         case PixelFormat::DXT23:
@@ -219,6 +242,8 @@ struct SurfaceParams {
             return Tegra::Texture::TextureFormat::DXT45;
         case PixelFormat::DXN1:
             return Tegra::Texture::TextureFormat::DXN1;
+        case PixelFormat::ASTC_2D_4X4:
+            return Tegra::Texture::TextureFormat::ASTC_2D_4X4;
         default:
             UNREACHABLE();
         }
@@ -243,6 +268,7 @@ struct SurfaceParams {
         case Tegra::RenderTargetFormat::RGB10_A2_UNORM:
             return ComponentType::UNorm;
         case Tegra::RenderTargetFormat::RGBA16_FLOAT:
+        case Tegra::RenderTargetFormat::R11G11B10_FLOAT:
             return ComponentType::Float;
         default:
             NGLOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -16,11 +16,11 @@ namespace Decompiler {
 
 using Tegra::Shader::Attribute;
 using Tegra::Shader::Instruction;
+using Tegra::Shader::LogicOperation;
 using Tegra::Shader::OpCode;
 using Tegra::Shader::Register;
 using Tegra::Shader::Sampler;
 using Tegra::Shader::SubOp;
-using Tegra::Shader::Uniform;
 
 constexpr u32 PROGRAM_END = MAX_PROGRAM_CODE_LENGTH;
 
@@ -266,6 +266,27 @@ public:
         BuildRegisterList();
     }
 
+    /**
+     * Returns code that does an integer size conversion for the specified size.
+     * @param value Value to perform integer size conversion on.
+     * @param size Register size to use for conversion instructions.
+     * @returns GLSL string corresponding to the value converted to the specified size.
+     */
+    static std::string ConvertIntegerSize(const std::string& value, Register::Size size) {
+        switch (size) {
+        case Register::Size::Byte:
+            return "((" + value + " << 24) >> 24)";
+        case Register::Size::Short:
+            return "((" + value + " << 16) >> 16)";
+        case Register::Size::Word:
+            // Default - do nothing
+            return value;
+        default:
+            NGLOG_CRITICAL(HW_GPU, "Unimplemented conversion size {}", static_cast<u32>(size));
+            UNREACHABLE();
+        }
+    }
+
     /**
      * Gets a register as an float.
      * @param reg The register to get.
@@ -282,15 +303,18 @@ public:
      * @param reg The register to get.
      * @param elem The element to use for the operation.
      * @param is_signed Whether to get the register as a signed (or unsigned) integer.
+     * @param size Register size to use for conversion instructions.
      * @returns GLSL string corresponding to the register as an integer.
      */
-    std::string GetRegisterAsInteger(const Register& reg, unsigned elem = 0,
-                                     bool is_signed = true) {
+    std::string GetRegisterAsInteger(const Register& reg, unsigned elem = 0, bool is_signed = true,
+                                     Register::Size size = Register::Size::Word) {
         const std::string func = GetGLSLConversionFunc(
             GLSLRegister::Type::Float,
             is_signed ? GLSLRegister::Type::Integer : GLSLRegister::Type::UnsignedInteger);
 
-        return func + '(' + GetRegister(reg, elem) + ')';
+        std::string value = func + '(' + GetRegister(reg, elem) + ')';
+
+        return ConvertIntegerSize(value, size);
     }
 
     /**
@@ -300,13 +324,15 @@ public:
      * @param value The code representing the value to assign.
      * @param dest_num_components Number of components in the destination.
      * @param value_num_components Number of components in the value.
-     * @param is_abs Optional, when True, applies absolute value to output.
+     * @param is_saturated Optional, when True, saturates the provided value.
      * @param dest_elem Optional, the destination element to use for the operation.
      */
     void SetRegisterToFloat(const Register& reg, u64 elem, const std::string& value,
-                            u64 dest_num_components, u64 value_num_components, bool is_abs = false,
-                            u64 dest_elem = 0) {
-        SetRegister(reg, elem, value, dest_num_components, value_num_components, is_abs, dest_elem);
+                            u64 dest_num_components, u64 value_num_components,
+                            bool is_saturated = false, u64 dest_elem = 0) {
+
+        SetRegister(reg, elem, is_saturated ? "clamp(" + value + ", 0.0, 1.0)" : value,
+                    dest_num_components, value_num_components, dest_elem);
     }
 
     /**
@@ -316,18 +342,22 @@ public:
      * @param value The code representing the value to assign.
      * @param dest_num_components Number of components in the destination.
      * @param value_num_components Number of components in the value.
-     * @param is_abs Optional, when True, applies absolute value to output.
+     * @param is_saturated Optional, when True, saturates the provided value.
      * @param dest_elem Optional, the destination element to use for the operation.
+     * @param size Register size to use for conversion instructions.
      */
     void SetRegisterToInteger(const Register& reg, bool is_signed, u64 elem,
                               const std::string& value, u64 dest_num_components,
-                              u64 value_num_components, bool is_abs = false, u64 dest_elem = 0) {
+                              u64 value_num_components, bool is_saturated = false,
+                              u64 dest_elem = 0, Register::Size size = Register::Size::Word) {
+        ASSERT_MSG(!is_saturated, "Unimplemented");
+
         const std::string func = GetGLSLConversionFunc(
             is_signed ? GLSLRegister::Type::Integer : GLSLRegister::Type::UnsignedInteger,
             GLSLRegister::Type::Float);
 
-        SetRegister(reg, elem, func + '(' + value + ')', dest_num_components, value_num_components,
-                    is_abs, dest_elem);
+        SetRegister(reg, elem, func + '(' + ConvertIntegerSize(value, size) + ')',
+                    dest_num_components, value_num_components, dest_elem);
     }
 
     /**
@@ -365,11 +395,9 @@ public:
     }
 
     /// Generates code representing a uniform (C buffer) register, interpreted as the input type.
-    std::string GetUniform(const Uniform& uniform, GLSLRegister::Type type) {
-        declr_const_buffers[uniform.index].MarkAsUsed(static_cast<unsigned>(uniform.index),
-                                                      static_cast<unsigned>(uniform.offset), stage);
-        std::string value =
-            'c' + std::to_string(uniform.index) + '[' + std::to_string(uniform.offset) + ']';
+    std::string GetUniform(u64 index, u64 offset, GLSLRegister::Type type) {
+        declr_const_buffers[index].MarkAsUsed(index, offset, stage);
+        std::string value = 'c' + std::to_string(index) + '[' + std::to_string(offset) + ']';
 
         if (type == GLSLRegister::Type::Float) {
             return value;
@@ -380,10 +408,19 @@ public:
         }
     }
 
-    /// Generates code representing a uniform (C buffer) register, interpreted as the type of the
-    /// destination register.
-    std::string GetUniform(const Uniform& uniform, const Register& dest_reg) {
-        return GetUniform(uniform, regs[dest_reg].GetActiveType());
+    std::string GetUniformIndirect(u64 index, s64 offset, const Register& index_reg,
+                                   GLSLRegister::Type type) {
+        declr_const_buffers[index].MarkAsUsedIndirect(index, stage);
+        std::string value = 'c' + std::to_string(index) + "[(floatBitsToInt(" +
+                            GetRegister(index_reg, 0) + ") + " + std::to_string(offset) + ") / 4]";
+
+        if (type == GLSLRegister::Type::Float) {
+            return value;
+        } else if (type == GLSLRegister::Type::Integer) {
+            return "floatBitsToInt(" + value + ')';
+        } else {
+            UNREACHABLE();
+        }
     }
 
     /// Add declarations for registers
@@ -425,6 +462,14 @@ public:
             ++const_buffer_layout;
         }
         declarations.AddNewLine();
+
+        // Append the sampler2D array for the used textures.
+        size_t num_samplers = GetSamplers().size();
+        if (num_samplers > 0) {
+            declarations.AddLine("uniform sampler2D " + SamplerEntry::GetArrayName(stage) + '[' +
+                                 std::to_string(num_samplers) + "];");
+            declarations.AddNewLine();
+        }
     }
 
     /// Returns a list of constant buffer declarations
@@ -435,6 +480,32 @@ public:
         return result;
     }
 
+    /// Returns a list of samplers used in the shader
+    std::vector<SamplerEntry> GetSamplers() const {
+        return used_samplers;
+    }
+
+    /// Returns the GLSL sampler used for the input shader sampler, and creates a new one if
+    /// necessary.
+    std::string AccessSampler(const Sampler& sampler) {
+        size_t offset = static_cast<size_t>(sampler.index.Value());
+
+        // If this sampler has already been used, return the existing mapping.
+        auto itr =
+            std::find_if(used_samplers.begin(), used_samplers.end(),
+                         [&](const SamplerEntry& entry) { return entry.GetOffset() == offset; });
+
+        if (itr != used_samplers.end()) {
+            return itr->GetName();
+        }
+
+        // Otherwise create a new mapping for this sampler
+        size_t next_index = used_samplers.size();
+        SamplerEntry entry{stage, offset, next_index};
+        used_samplers.emplace_back(entry);
+        return entry.GetName();
+    }
+
 private:
     /// Build GLSL conversion function, e.g. floatBitsToInt, intBitsToFloat, etc.
     const std::string GetGLSLConversionFunc(GLSLRegister::Type src, GLSLRegister::Type dest) const {
@@ -460,13 +531,11 @@ private:
      * @param value The code representing the value to assign.
      * @param dest_num_components Number of components in the destination.
      * @param value_num_components Number of components in the value.
-     * @param is_abs Optional, when True, applies absolute value to output.
      * @param dest_elem Optional, the destination element to use for the operation.
      */
     void SetRegister(const Register& reg, u64 elem, const std::string& value,
-                     u64 dest_num_components, u64 value_num_components, bool is_abs,
-                     u64 dest_elem) {
-        std::string dest = GetRegister(reg, dest_elem);
+                     u64 dest_num_components, u64 value_num_components, u64 dest_elem) {
+        std::string dest = GetRegister(reg, static_cast<u32>(dest_elem));
         if (dest_num_components > 1) {
             dest += GetSwizzle(elem);
         }
@@ -476,8 +545,6 @@ private:
             src += GetSwizzle(elem);
         }
 
-        src = is_abs ? "abs(" + src + ')' : src;
-
         shader.AddLine(dest + " = " + src + ';');
     }
 
@@ -498,7 +565,7 @@ private:
             // vertex shader, and what's the value of the fourth element when inside a Tess Eval
             // shader.
             ASSERT(stage == Maxwell3D::Regs::ShaderStage::Vertex);
-            return "vec4(0, 0, gl_InstanceID, gl_VertexID)";
+            return "vec4(0, 0, uintBitsToFloat(gl_InstanceID), uintBitsToFloat(gl_VertexID))";
         default:
             const u32 index{static_cast<u32>(attribute) -
                             static_cast<u32>(Attribute::Index::Attribute_0)};
@@ -544,6 +611,7 @@ private:
     std::set<Attribute::Index> declr_input_attribute;
     std::set<Attribute::Index> declr_output_attribute;
     std::array<ConstBufferEntry, Maxwell3D::Regs::MaxConstBuffers> declr_const_buffers;
+    std::vector<SamplerEntry> used_samplers;
     const Maxwell3D::Regs::ShaderStage& stage;
 };
 
@@ -563,7 +631,7 @@ public:
 
     /// Returns entries in the shader that are useful for external functions
     ShaderEntries GetEntries() const {
-        return {regs.GetConstBuffersDeclarations()};
+        return {regs.GetConstBuffersDeclarations(), regs.GetSamplers()};
     }
 
 private:
@@ -585,12 +653,8 @@ private:
     }
 
     /// Generates code representing a texture sampler.
-    std::string GetSampler(const Sampler& sampler) const {
-        // TODO(Subv): Support more than just texture sampler 0
-        ASSERT_MSG(sampler.index == Sampler::Index::Sampler_0, "unsupported");
-        const unsigned index{static_cast<unsigned>(sampler.index.Value()) -
-                             static_cast<unsigned>(Sampler::Index::Sampler_0)};
-        return "tex[" + std::to_string(index) + ']';
+    std::string GetSampler(const Sampler& sampler) {
+        return regs.AccessSampler(sampler);
     }
 
     /**
@@ -696,6 +760,31 @@ private:
         return (absolute_offset % SchedPeriod) == 0;
     }
 
+    void WriteLogicOperation(Register dest, LogicOperation logic_op, const std::string& op_a,
+                             const std::string& op_b) {
+        switch (logic_op) {
+        case LogicOperation::And: {
+            regs.SetRegisterToInteger(dest, true, 0, '(' + op_a + " & " + op_b + ')', 1, 1);
+            break;
+        }
+        case LogicOperation::Or: {
+            regs.SetRegisterToInteger(dest, true, 0, '(' + op_a + " | " + op_b + ')', 1, 1);
+            break;
+        }
+        case LogicOperation::Xor: {
+            regs.SetRegisterToInteger(dest, true, 0, '(' + op_a + " ^ " + op_b + ')', 1, 1);
+            break;
+        }
+        case LogicOperation::PassB: {
+            regs.SetRegisterToInteger(dest, true, 0, op_b, 1, 1);
+            break;
+        }
+        default:
+            NGLOG_CRITICAL(HW_GPU, "Unimplemented logic operation: {}", static_cast<u32>(logic_op));
+            UNREACHABLE();
+        }
+    }
+
     /**
      * Compiles a single instruction from Tegra to GLSL.
      * @param offset the offset of the Tegra shader instruction.
@@ -733,21 +822,25 @@ private:
 
         switch (opcode->GetType()) {
         case OpCode::Type::Arithmetic: {
-            std::string op_a = instr.alu.negate_a ? "-" : "";
-            op_a += regs.GetRegisterAsFloat(instr.gpr8);
+            std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
             if (instr.alu.abs_a) {
                 op_a = "abs(" + op_a + ')';
             }
 
-            std::string op_b = instr.alu.negate_b ? "-" : "";
+            if (instr.alu.negate_a) {
+                op_a = "-(" + op_a + ')';
+            }
+
+            std::string op_b;
 
             if (instr.is_b_imm) {
-                op_b += GetImmediate19(instr);
+                op_b = GetImmediate19(instr);
             } else {
                 if (instr.is_b_gpr) {
-                    op_b += regs.GetRegisterAsFloat(instr.gpr20);
+                    op_b = regs.GetRegisterAsFloat(instr.gpr20);
                 } else {
-                    op_b += regs.GetUniform(instr.uniform, instr.gpr0);
+                    op_b = regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
+                                           GLSLRegister::Type::Float);
                 }
             }
 
@@ -755,6 +848,10 @@ private:
                 op_b = "abs(" + op_b + ')';
             }
 
+            if (instr.alu.negate_b) {
+                op_b = "-(" + op_b + ')';
+            }
+
             switch (opcode->GetId()) {
             case OpCode::Id::MOV_C:
             case OpCode::Id::MOV_R: {
@@ -762,58 +859,49 @@ private:
                 break;
             }
 
-            case OpCode::Id::MOV32_IMM: {
-                // mov32i doesn't have abs or neg bits.
-                regs.SetRegisterToFloat(instr.gpr0, 0, GetImmediate32(instr), 1, 1);
-                break;
-            }
             case OpCode::Id::FMUL_C:
             case OpCode::Id::FMUL_R:
             case OpCode::Id::FMUL_IMM: {
-                regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " * " + op_b, 1, 1, instr.alu.abs_d);
-                break;
-            }
-            case OpCode::Id::FMUL32_IMM: {
-                // fmul32i doesn't have abs or neg bits.
-                regs.SetRegisterToFloat(
-                    instr.gpr0, 0,
-                    regs.GetRegisterAsFloat(instr.gpr8) + " * " + GetImmediate32(instr), 1, 1);
+                regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " * " + op_b, 1, 1,
+                                        instr.alu.saturate_d);
                 break;
             }
             case OpCode::Id::FADD_C:
             case OpCode::Id::FADD_R:
             case OpCode::Id::FADD_IMM: {
-                regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " + " + op_b, 1, 1, instr.alu.abs_d);
+                regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " + " + op_b, 1, 1,
+                                        instr.alu.saturate_d);
                 break;
             }
             case OpCode::Id::MUFU: {
                 switch (instr.sub_op) {
                 case SubOp::Cos:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "cos(" + op_a + ')', 1, 1,
-                                            instr.alu.abs_d);
+                                            instr.alu.saturate_d);
                     break;
                 case SubOp::Sin:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "sin(" + op_a + ')', 1, 1,
-                                            instr.alu.abs_d);
+                                            instr.alu.saturate_d);
                     break;
                 case SubOp::Ex2:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "exp2(" + op_a + ')', 1, 1,
-                                            instr.alu.abs_d);
+                                            instr.alu.saturate_d);
                     break;
                 case SubOp::Lg2:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "log2(" + op_a + ')', 1, 1,
-                                            instr.alu.abs_d);
+                                            instr.alu.saturate_d);
                     break;
                 case SubOp::Rcp:
-                    regs.SetRegisterToFloat(instr.gpr0, 0, "1.0 / " + op_a, 1, 1, instr.alu.abs_d);
+                    regs.SetRegisterToFloat(instr.gpr0, 0, "1.0 / " + op_a, 1, 1,
+                                            instr.alu.saturate_d);
                     break;
                 case SubOp::Rsq:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "inversesqrt(" + op_a + ')', 1, 1,
-                                            instr.alu.abs_d);
+                                            instr.alu.saturate_d);
                     break;
                 case SubOp::Min:
                     regs.SetRegisterToFloat(instr.gpr0, 0, "min(" + op_a + "," + op_b + ')', 1, 1,
-                                            instr.alu.abs_d);
+                                            instr.alu.saturate_d);
                     break;
                 default:
                     NGLOG_CRITICAL(HW_GPU, "Unhandled MUFU sub op: {0:x}",
@@ -850,71 +938,184 @@ private:
             }
             break;
         }
-        case OpCode::Type::Logic: {
-            std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, false);
+        case OpCode::Type::ArithmeticImmediate: {
+            switch (opcode->GetId()) {
+            case OpCode::Id::MOV32_IMM: {
+                regs.SetRegisterToFloat(instr.gpr0, 0, GetImmediate32(instr), 1, 1);
+                break;
+            }
+            case OpCode::Id::FMUL32_IMM: {
+                regs.SetRegisterToFloat(
+                    instr.gpr0, 0,
+                    regs.GetRegisterAsFloat(instr.gpr8) + " * " + GetImmediate32(instr), 1, 1);
+                break;
+            }
+            }
+            break;
+        }
+        case OpCode::Type::Bfe: {
+            ASSERT_MSG(!instr.bfe.negate_b, "Unimplemented");
 
-            if (instr.alu.lop.invert_a)
-                op_a = "~(" + op_a + ')';
+            std::string op_a = instr.bfe.negate_a ? "-" : "";
+            op_a += regs.GetRegisterAsInteger(instr.gpr8);
 
             switch (opcode->GetId()) {
-            case OpCode::Id::LOP32I: {
-                u32 imm = static_cast<u32>(instr.alu.imm20_32.Value());
+            case OpCode::Id::BFE_IMM: {
+                std::string inner_shift =
+                    '(' + op_a + " << " + std::to_string(instr.bfe.GetLeftShiftValue()) + ')';
+                std::string outer_shift =
+                    '(' + inner_shift + " >> " +
+                    std::to_string(instr.bfe.GetLeftShiftValue() + instr.bfe.shift_position) + ')';
 
-                if (instr.alu.lop.invert_b)
-                    imm = ~imm;
-
-                switch (instr.alu.lop.operation) {
-                case Tegra::Shader::LogicOperation::And: {
-                    regs.SetRegisterToInteger(instr.gpr0, false, 0,
-                                              '(' + op_a + " & " + std::to_string(imm) + ')', 1, 1);
-                    break;
-                }
-                case Tegra::Shader::LogicOperation::Or: {
-                    regs.SetRegisterToInteger(instr.gpr0, false, 0,
-                                              '(' + op_a + " | " + std::to_string(imm) + ')', 1, 1);
-                    break;
-                }
-                case Tegra::Shader::LogicOperation::Xor: {
-                    regs.SetRegisterToInteger(instr.gpr0, false, 0,
-                                              '(' + op_a + " ^ " + std::to_string(imm) + ')', 1, 1);
-                    break;
-                }
-                default:
-                    NGLOG_CRITICAL(HW_GPU, "Unimplemented lop32i operation: {}",
-                                   static_cast<u32>(instr.alu.lop.operation.Value()));
-                    UNREACHABLE();
-                }
+                regs.SetRegisterToInteger(instr.gpr0, true, 0, outer_shift, 1, 1);
                 break;
             }
             default: {
-                NGLOG_CRITICAL(HW_GPU, "Unhandled logic instruction: {}", opcode->GetName());
+                NGLOG_CRITICAL(HW_GPU, "Unhandled BFE instruction: {}", opcode->GetName());
+                UNREACHABLE();
+            }
+            }
+
+            break;
+        }
+
+        case OpCode::Type::Shift: {
+            std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, true);
+            std::string op_b;
+
+            if (instr.is_b_imm) {
+                op_b += '(' + std::to_string(instr.alu.GetSignedImm20_20()) + ')';
+            } else {
+                if (instr.is_b_gpr) {
+                    op_b += regs.GetRegisterAsInteger(instr.gpr20);
+                } else {
+                    op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
+                                            GLSLRegister::Type::Integer);
+                }
+            }
+
+            switch (opcode->GetId()) {
+            case OpCode::Id::SHR_C:
+            case OpCode::Id::SHR_R:
+            case OpCode::Id::SHR_IMM: {
+                if (!instr.shift.is_signed) {
+                    // Logical shift right
+                    op_a = "uint(" + op_a + ')';
+                }
+
+                // Cast to int is superfluous for arithmetic shift, it's only for a logical shift
+                regs.SetRegisterToInteger(instr.gpr0, true, 0, "int(" + op_a + " >> " + op_b + ')',
+                                          1, 1);
+                break;
+            }
+            case OpCode::Id::SHL_C:
+            case OpCode::Id::SHL_R:
+            case OpCode::Id::SHL_IMM:
+                regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " << " + op_b, 1, 1);
+                break;
+            default: {
+                NGLOG_CRITICAL(HW_GPU, "Unhandled shift instruction: {}", opcode->GetName());
                 UNREACHABLE();
             }
             }
             break;
         }
-        case OpCode::Type::ScaledAdd: {
+
+        case OpCode::Type::ArithmeticIntegerImmediate: {
             std::string op_a = regs.GetRegisterAsInteger(instr.gpr8);
+            std::string op_b = std::to_string(instr.alu.imm20_32.Value());
 
-            if (instr.iscadd.negate_a)
-                op_a = '-' + op_a;
+            switch (opcode->GetId()) {
+            case OpCode::Id::IADD32I:
+                if (instr.iadd32i.negate_a)
+                    op_a = "-(" + op_a + ')';
 
-            std::string op_b = instr.iscadd.negate_b ? "-" : "";
+                regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " + " + op_b, 1, 1,
+                                          instr.iadd32i.saturate != 0);
+                break;
+            case OpCode::Id::LOP32I: {
+                if (instr.alu.lop32i.invert_a)
+                    op_a = "~(" + op_a + ')';
 
+                if (instr.alu.lop32i.invert_b)
+                    op_b = "~(" + op_b + ')';
+
+                WriteLogicOperation(instr.gpr0, instr.alu.lop32i.operation, op_a, op_b);
+                break;
+            }
+            default: {
+                NGLOG_CRITICAL(HW_GPU, "Unhandled ArithmeticIntegerImmediate instruction: {}",
+                               opcode->GetName());
+                UNREACHABLE();
+            }
+            }
+            break;
+        }
+        case OpCode::Type::ArithmeticInteger: {
+            std::string op_a = regs.GetRegisterAsInteger(instr.gpr8);
+            std::string op_b;
             if (instr.is_b_imm) {
-                op_b += '(' + std::to_string(instr.iscadd.GetImmediate()) + ')';
+                op_b += '(' + std::to_string(instr.alu.GetSignedImm20_20()) + ')';
             } else {
                 if (instr.is_b_gpr) {
                     op_b += regs.GetRegisterAsInteger(instr.gpr20);
                 } else {
-                    op_b += regs.GetUniform(instr.uniform, instr.gpr0);
+                    op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
+                                            GLSLRegister::Type::Integer);
                 }
             }
 
-            std::string shift = std::to_string(instr.iscadd.shift_amount.Value());
+            switch (opcode->GetId()) {
+            case OpCode::Id::IADD_C:
+            case OpCode::Id::IADD_R:
+            case OpCode::Id::IADD_IMM: {
+                if (instr.alu_integer.negate_a)
+                    op_a = "-(" + op_a + ')';
+
+                if (instr.alu_integer.negate_b)
+                    op_b = "-(" + op_b + ')';
+
+                regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " + " + op_b, 1, 1,
+                                          instr.alu.saturate_d);
+                break;
+            }
+            case OpCode::Id::ISCADD_C:
+            case OpCode::Id::ISCADD_R:
+            case OpCode::Id::ISCADD_IMM: {
+                if (instr.alu_integer.negate_a)
+                    op_a = "-(" + op_a + ')';
+
+                if (instr.alu_integer.negate_b)
+                    op_b = "-(" + op_b + ')';
+
+                std::string shift = std::to_string(instr.alu_integer.shift_amount.Value());
+
+                regs.SetRegisterToInteger(instr.gpr0, true, 0,
+                                          "((" + op_a + " << " + shift + ") + " + op_b + ')', 1, 1);
+                break;
+            }
+            case OpCode::Id::LOP_C:
+            case OpCode::Id::LOP_R:
+            case OpCode::Id::LOP_IMM: {
+                ASSERT_MSG(!instr.alu.lop.unk44, "Unimplemented");
+                ASSERT_MSG(instr.alu.lop.pred48 == Pred::UnusedIndex, "Unimplemented");
+
+                if (instr.alu.lop.invert_a)
+                    op_a = "~(" + op_a + ')';
+
+                if (instr.alu.lop.invert_b)
+                    op_b = "~(" + op_b + ')';
+
+                WriteLogicOperation(instr.gpr0, instr.alu.lop.operation, op_a, op_b);
+                break;
+            }
+            default: {
+                NGLOG_CRITICAL(HW_GPU, "Unhandled ArithmeticInteger instruction: {}",
+                               opcode->GetName());
+                UNREACHABLE();
+            }
+            }
 
-            regs.SetRegisterToInteger(instr.gpr0, true, 0,
-                                      "((" + op_a + " << " + shift + ") + " + op_b + ')', 1, 1);
             break;
         }
         case OpCode::Type::Ffma: {
@@ -924,7 +1125,8 @@ private:
 
             switch (opcode->GetId()) {
             case OpCode::Id::FFMA_CR: {
-                op_b += regs.GetUniform(instr.uniform, instr.gpr0);
+                op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
+                                        GLSLRegister::Type::Float);
                 op_c += regs.GetRegisterAsFloat(instr.gpr39);
                 break;
             }
@@ -935,7 +1137,8 @@ private:
             }
             case OpCode::Id::FFMA_RC: {
                 op_b += regs.GetRegisterAsFloat(instr.gpr39);
-                op_c += regs.GetUniform(instr.uniform, instr.gpr0);
+                op_c += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
+                                        GLSLRegister::Type::Float);
                 break;
             }
             case OpCode::Id::FFMA_IMM: {
@@ -949,31 +1152,33 @@ private:
             }
             }
 
-            regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " * " + op_b + " + " + op_c, 1, 1);
+            regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " * " + op_b + " + " + op_c, 1, 1,
+                                    instr.alu.saturate_d);
             break;
         }
         case OpCode::Type::Conversion: {
-            ASSERT_MSG(instr.conversion.size == Register::Size::Word, "Unimplemented");
             ASSERT_MSG(!instr.conversion.negate_a, "Unimplemented");
-            ASSERT_MSG(!instr.conversion.saturate_a, "Unimplemented");
 
             switch (opcode->GetId()) {
             case OpCode::Id::I2I_R: {
                 ASSERT_MSG(!instr.conversion.selector, "Unimplemented");
 
-                std::string op_a =
-                    regs.GetRegisterAsInteger(instr.gpr20, 0, instr.conversion.is_signed);
+                std::string op_a = regs.GetRegisterAsInteger(
+                    instr.gpr20, 0, instr.conversion.is_input_signed, instr.conversion.src_size);
 
                 if (instr.conversion.abs_a) {
                     op_a = "abs(" + op_a + ')';
                 }
 
-                regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_signed, 0, op_a, 1, 1);
+                regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_output_signed, 0, op_a, 1,
+                                          1, instr.alu.saturate_d, 0, instr.conversion.dest_size);
                 break;
             }
             case OpCode::Id::I2F_R: {
-                std::string op_a =
-                    regs.GetRegisterAsInteger(instr.gpr20, 0, instr.conversion.is_signed);
+                ASSERT_MSG(instr.conversion.dest_size == Register::Size::Word, "Unimplemented");
+                ASSERT_MSG(!instr.conversion.selector, "Unimplemented");
+                std::string op_a = regs.GetRegisterAsInteger(
+                    instr.gpr20, 0, instr.conversion.is_input_signed, instr.conversion.src_size);
 
                 if (instr.conversion.abs_a) {
                     op_a = "abs(" + op_a + ')';
@@ -983,13 +1188,71 @@ private:
                 break;
             }
             case OpCode::Id::F2F_R: {
+                ASSERT_MSG(instr.conversion.dest_size == Register::Size::Word, "Unimplemented");
+                ASSERT_MSG(instr.conversion.src_size == Register::Size::Word, "Unimplemented");
+                std::string op_a = regs.GetRegisterAsFloat(instr.gpr20);
+
+                switch (instr.conversion.f2f.rounding) {
+                case Tegra::Shader::F2fRoundingOp::None:
+                    break;
+                case Tegra::Shader::F2fRoundingOp::Floor:
+                    op_a = "floor(" + op_a + ')';
+                    break;
+                case Tegra::Shader::F2fRoundingOp::Ceil:
+                    op_a = "ceil(" + op_a + ')';
+                    break;
+                case Tegra::Shader::F2fRoundingOp::Trunc:
+                    op_a = "trunc(" + op_a + ')';
+                    break;
+                default:
+                    NGLOG_CRITICAL(HW_GPU, "Unimplemented f2f rounding mode {}",
+                                   static_cast<u32>(instr.conversion.f2f.rounding.Value()));
+                    UNREACHABLE();
+                    break;
+                }
+
+                if (instr.conversion.abs_a) {
+                    op_a = "abs(" + op_a + ')';
+                }
+
+                regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1, instr.alu.saturate_d);
+                break;
+            }
+            case OpCode::Id::F2I_R: {
+                ASSERT_MSG(instr.conversion.src_size == Register::Size::Word, "Unimplemented");
                 std::string op_a = regs.GetRegisterAsFloat(instr.gpr20);
 
                 if (instr.conversion.abs_a) {
                     op_a = "abs(" + op_a + ')';
                 }
 
-                regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
+                switch (instr.conversion.f2i.rounding) {
+                case Tegra::Shader::F2iRoundingOp::None:
+                    break;
+                case Tegra::Shader::F2iRoundingOp::Floor:
+                    op_a = "floor(" + op_a + ')';
+                    break;
+                case Tegra::Shader::F2iRoundingOp::Ceil:
+                    op_a = "ceil(" + op_a + ')';
+                    break;
+                case Tegra::Shader::F2iRoundingOp::Trunc:
+                    op_a = "trunc(" + op_a + ')';
+                    break;
+                default:
+                    NGLOG_CRITICAL(HW_GPU, "Unimplemented f2i rounding mode {}",
+                                   static_cast<u32>(instr.conversion.f2i.rounding.Value()));
+                    UNREACHABLE();
+                    break;
+                }
+
+                if (instr.conversion.is_output_signed) {
+                    op_a = "int(" + op_a + ')';
+                } else {
+                    op_a = "uint(" + op_a + ')';
+                }
+
+                regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_output_signed, 0, op_a, 1,
+                                          1, false, 0, instr.conversion.dest_size);
                 break;
             }
             default: {
@@ -1000,36 +1263,60 @@ private:
             break;
         }
         case OpCode::Type::Memory: {
-            const Attribute::Index attribute = instr.attribute.fmt20.index;
-
             switch (opcode->GetId()) {
             case OpCode::Id::LD_A: {
                 ASSERT_MSG(instr.attribute.fmt20.size == 0, "untested");
                 regs.SetRegisterToInputAttibute(instr.gpr0, instr.attribute.fmt20.element,
-                                                attribute);
+                                                instr.attribute.fmt20.index);
+                break;
+            }
+            case OpCode::Id::LD_C: {
+                ASSERT_MSG(instr.ld_c.unknown == 0, "Unimplemented");
+
+                std::string op_a =
+                    regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 0, instr.gpr8,
+                                            GLSLRegister::Type::Float);
+                std::string op_b =
+                    regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 4, instr.gpr8,
+                                            GLSLRegister::Type::Float);
+
+                switch (instr.ld_c.type.Value()) {
+                case Tegra::Shader::UniformType::Single:
+                    regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
+                    break;
+
+                case Tegra::Shader::UniformType::Double:
+                    regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
+                    regs.SetRegisterToFloat(instr.gpr0.Value() + 1, 0, op_b, 1, 1);
+                    break;
+
+                default:
+                    NGLOG_CRITICAL(HW_GPU, "Unhandled type: {}",
+                                   static_cast<unsigned>(instr.ld_c.type.Value()));
+                    UNREACHABLE();
+                }
                 break;
             }
             case OpCode::Id::ST_A: {
                 ASSERT_MSG(instr.attribute.fmt20.size == 0, "untested");
-                regs.SetOutputAttributeToRegister(attribute, instr.attribute.fmt20.element,
-                                                  instr.gpr0);
+                regs.SetOutputAttributeToRegister(instr.attribute.fmt20.index,
+                                                  instr.attribute.fmt20.element, instr.gpr0);
                 break;
             }
             case OpCode::Id::TEX: {
-                ASSERT_MSG(instr.attribute.fmt20.size == 4, "untested");
                 const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
                 const std::string op_b = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
                 const std::string sampler = GetSampler(instr.sampler);
                 const std::string coord = "vec2 coords = vec2(" + op_a + ", " + op_b + ");";
-                // Add an extra scope and declare the texture coords inside to prevent overwriting
-                // them in case they are used as outputs of the texs instruction.
+                // Add an extra scope and declare the texture coords inside to prevent
+                // overwriting them in case they are used as outputs of the texs instruction.
                 shader.AddLine("{");
                 ++shader.scope;
                 shader.AddLine(coord);
                 const std::string texture = "texture(" + sampler + ", coords)";
 
                 size_t dest_elem{};
-                for (size_t elem = 0; elem < instr.attribute.fmt20.size; ++elem) {
+                for (size_t elem = 0; elem < 4; ++elem) {
                     if (!instr.tex.IsComponentEnabled(elem)) {
                         // Skip disabled components
                         continue;
@@ -1042,7 +1329,6 @@ private:
                 break;
             }
             case OpCode::Id::TEXS: {
-                ASSERT_MSG(instr.attribute.fmt20.size == 4, "untested");
                 const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
                 const std::string op_b = regs.GetRegisterAsFloat(instr.gpr20);
                 const std::string sampler = GetSampler(instr.sampler);
@@ -1054,8 +1340,8 @@ private:
                 shader.AddLine(coord);
                 const std::string texture = "texture(" + sampler + ", coords)";
 
-                // TEXS has two destination registers. RG goes into gpr0+0 and gpr0+1, and BA goes
-                // into gpr28+0 and gpr28+1
+                // TEXS has two destination registers. RG goes into gpr0+0 and gpr0+1, and BA
+                // goes into gpr28+0 and gpr28+1
                 size_t offset{};
 
                 for (const auto& dest : {instr.gpr0.Value(), instr.gpr28.Value()}) {
@@ -1105,7 +1391,8 @@ private:
                 if (instr.is_b_gpr) {
                     op_b += regs.GetRegisterAsFloat(instr.gpr20);
                 } else {
-                    op_b += regs.GetUniform(instr.uniform, GLSLRegister::Type::Float);
+                    op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
+                                            GLSLRegister::Type::Float);
                 }
             }
 
@@ -1138,15 +1425,17 @@ private:
         }
         case OpCode::Type::IntegerSetPredicate: {
             std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, instr.isetp.is_signed);
+            std::string op_b;
 
-            std::string op_b{};
-
-            ASSERT_MSG(!instr.is_b_imm, "ISETP_IMM not implemented");
-
-            if (instr.is_b_gpr) {
-                op_b += regs.GetRegisterAsInteger(instr.gpr20, 0, instr.isetp.is_signed);
+            if (instr.is_b_imm) {
+                op_b += '(' + std::to_string(instr.alu.GetSignedImm20_20()) + ')';
             } else {
-                op_b += regs.GetUniform(instr.uniform, GLSLRegister::Type::Integer);
+                if (instr.is_b_gpr) {
+                    op_b += regs.GetRegisterAsInteger(instr.gpr20, 0, instr.isetp.is_signed);
+                } else {
+                    op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
+                                            GLSLRegister::Type::Integer);
+                }
             }
 
             using Tegra::Shader::Pred;
@@ -1192,7 +1481,8 @@ private:
                 if (instr.is_b_gpr) {
                     op_b += regs.GetRegisterAsFloat(instr.gpr20);
                 } else {
-                    op_b += regs.GetUniform(instr.uniform, GLSLRegister::Type::Float);
+                    op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
+                                            GLSLRegister::Type::Float);
                 }
             }
 
@@ -1200,8 +1490,8 @@ private:
                 op_b = "abs(" + op_b + ')';
             }
 
-            // The fset instruction sets a register to 1.0 if the condition is true, and to 0
-            // otherwise.
+            // The fset instruction sets a register to 1.0 or -1 (depending on the bf bit) if the
+            // condition is true, and to 0 otherwise.
             std::string second_pred =
                 GetPredicateCondition(instr.fset.pred39, instr.fset.neg_pred != 0);
 
@@ -1219,6 +1509,41 @@ private:
             }
             break;
         }
+        case OpCode::Type::IntegerSet: {
+            std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, instr.iset.is_signed);
+
+            std::string op_b;
+
+            if (instr.is_b_imm) {
+                op_b = std::to_string(instr.alu.GetSignedImm20_20());
+            } else {
+                if (instr.is_b_gpr) {
+                    op_b = regs.GetRegisterAsInteger(instr.gpr20, 0, instr.iset.is_signed);
+                } else {
+                    op_b = regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
+                                           GLSLRegister::Type::Integer);
+                }
+            }
+
+            // The iset instruction sets a register to 1.0 or -1 (depending on the bf bit) if the
+            // condition is true, and to 0 otherwise.
+            std::string second_pred =
+                GetPredicateCondition(instr.iset.pred39, instr.iset.neg_pred != 0);
+
+            std::string comparator = GetPredicateComparison(instr.iset.cond);
+            std::string combiner = GetPredicateCombiner(instr.iset.op);
+
+            std::string predicate = "(((" + op_a + ") " + comparator + " (" + op_b + ")) " +
+                                    combiner + " (" + second_pred + "))";
+
+            if (instr.iset.bf) {
+                regs.SetRegisterToFloat(instr.gpr0, 0, predicate + " ? 1.0 : 0.0", 1, 1);
+            } else {
+                regs.SetRegisterToInteger(instr.gpr0, false, 0, predicate + " ? 0xFFFFFFFF : 0", 1,
+                                          1);
+            }
+            break;
+        }
         default: {
             switch (opcode->GetId()) {
             case OpCode::Id::EXIT: {
@@ -1232,8 +1557,8 @@ private:
 
                 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
+                    // 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;
                 }
@@ -1255,6 +1580,11 @@ private:
                 regs.SetRegisterToInputAttibute(instr.gpr0, attribute.element, attribute.index);
                 break;
             }
+            case OpCode::Id::SSY: {
+                // The SSY opcode tells the GPU where to re-converge divergent execution paths, we
+                // can ignore this when generating GLSL code.
+                break;
+            }
             default: {
                 NGLOG_CRITICAL(HW_GPU, "Unhandled instruction: {}", opcode->GetName());
                 UNREACHABLE();

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -39,6 +39,10 @@ void main() {
     // Viewport can be flipped, which is unsupported by glViewport
     position.xy *= viewport_flip.xy;
     gl_Position = position;
+
+    // TODO(bunnei): This is likely a hack, position.w should be interpolated as 1.0
+    // For now, this is here to bring order in lieu of proper emulation
+    position.w = 1.0;
 }
 )";
     out += program.first;
@@ -62,8 +66,6 @@ layout (std140) uniform fs_config {
     vec4 viewport_flip;
 };
 
-uniform sampler2D tex[32];
-
 void main() {
     exec_shader();
 }

src/video_core/renderer_opengl/gl_shader_gen.h

@@ -22,17 +22,28 @@ class ConstBufferEntry {
     using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 
 public:
-    void MarkAsUsed(unsigned index, unsigned offset, Maxwell::ShaderStage stage) {
+    void MarkAsUsed(u64 index, u64 offset, Maxwell::ShaderStage stage) {
         is_used = true;
-        this->index = index;
+        this->index = static_cast<unsigned>(index);
+        this->stage = stage;
+        max_offset = std::max(max_offset, static_cast<unsigned>(offset));
+    }
+
+    void MarkAsUsedIndirect(u64 index, Maxwell::ShaderStage stage) {
+        is_used = true;
+        is_indirect = true;
+        this->index = static_cast<unsigned>(index);
         this->stage = stage;
-        max_offset = std::max(max_offset, offset);
     }
 
     bool IsUsed() const {
         return is_used;
     }
 
+    bool IsIndirect() const {
+        return is_indirect;
+    }
+
     unsigned GetIndex() const {
         return index;
     }
@@ -51,13 +62,54 @@ private:
     };
 
     bool is_used{};
+    bool is_indirect{};
     unsigned index{};
     unsigned max_offset{};
     Maxwell::ShaderStage stage;
 };
 
+class SamplerEntry {
+    using Maxwell = Tegra::Engines::Maxwell3D::Regs;
+
+public:
+    SamplerEntry(Maxwell::ShaderStage stage, size_t offset, size_t index)
+        : offset(offset), stage(stage), sampler_index(index) {}
+
+    size_t GetOffset() const {
+        return offset;
+    }
+
+    size_t GetIndex() const {
+        return sampler_index;
+    }
+
+    Maxwell::ShaderStage GetStage() const {
+        return stage;
+    }
+
+    std::string GetName() const {
+        return std::string(TextureSamplerNames[static_cast<size_t>(stage)]) + '[' +
+               std::to_string(sampler_index) + ']';
+    }
+
+    static std::string GetArrayName(Maxwell::ShaderStage stage) {
+        return TextureSamplerNames[static_cast<size_t>(stage)];
+    }
+
+private:
+    static constexpr std::array<const char*, Maxwell::MaxShaderStage> TextureSamplerNames = {
+        "tex_vs", "tex_tessc", "tex_tesse", "tex_gs", "tex_fs",
+    };
+    /// Offset in TSC memory from which to read the sampler object, as specified by the sampling
+    /// instruction.
+    size_t offset;
+    Maxwell::ShaderStage stage; ///< Shader stage where this sampler was used.
+    size_t sampler_index;       ///< Value used to index into the generated GLSL sampler array.
+};
+
 struct ShaderEntries {
     std::vector<ConstBufferEntry> const_buffer_entries;
+    std::vector<SamplerEntry> texture_samplers;
 };
 
 using ProgramResult = std::pair<std::string, ShaderEntries>;

src/video_core/renderer_opengl/gl_shader_manager.cpp

@@ -32,33 +32,14 @@ void SetShaderUniformBlockBindings(GLuint shader) {
                                  sizeof(MaxwellUniformData));
 }
 
-void SetShaderSamplerBindings(GLuint shader) {
-    OpenGLState cur_state = OpenGLState::GetCurState();
-    GLuint old_program = std::exchange(cur_state.draw.shader_program, shader);
-    cur_state.Apply();
-
-    // Set the texture samplers to correspond to different texture units
-    for (u32 texture = 0; texture < NumTextureSamplers; ++texture) {
-        // Set the texture samplers to correspond to different texture units
-        std::string uniform_name = "tex[" + std::to_string(texture) + "]";
-        GLint uniform_tex = glGetUniformLocation(shader, uniform_name.c_str());
-        if (uniform_tex != -1) {
-            glUniform1i(uniform_tex, TextureUnits::MaxwellTexture(texture).id);
-        }
-    }
-
-    cur_state.draw.shader_program = old_program;
-    cur_state.Apply();
-}
-
 } // namespace Impl
 
 void MaxwellUniformData::SetFromRegs(const Maxwell3D::State::ShaderStageInfo& shader_stage) {
     const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
 
     // TODO(bunnei): Support more than one viewport
-    viewport_flip[0] = regs.viewport_transform[0].scale_x < 0.0 ? -1.0 : 1.0;
-    viewport_flip[1] = regs.viewport_transform[0].scale_y < 0.0 ? -1.0 : 1.0;
+    viewport_flip[0] = regs.viewport_transform[0].scale_x < 0.0 ? -1.0f : 1.0f;
+    viewport_flip[1] = regs.viewport_transform[0].scale_y < 0.0 ? -1.0f : 1.0f;
 }
 
 } // namespace GLShader

src/video_core/renderer_opengl/gl_shader_manager.h

@@ -45,7 +45,6 @@ public:
         shader.Create(program_result.first.c_str(), type);
         program.Create(true, shader.handle);
         Impl::SetShaderUniformBlockBindings(program.handle);
-        Impl::SetShaderSamplerBindings(program.handle);
         entries = program_result.second;
     }
     GLuint GetHandle() const {

src/video_core/renderer_opengl/gl_state.cpp

@@ -50,6 +50,10 @@ OpenGLState::OpenGLState() {
     for (auto& texture_unit : texture_units) {
         texture_unit.texture_2d = 0;
         texture_unit.sampler = 0;
+        texture_unit.swizzle.r = GL_RED;
+        texture_unit.swizzle.g = GL_GREEN;
+        texture_unit.swizzle.b = GL_BLUE;
+        texture_unit.swizzle.a = GL_ALPHA;
     }
 
     lighting_lut.texture_buffer = 0;
@@ -192,13 +196,22 @@ void OpenGLState::Apply() const {
     }
 
     // Textures
-    for (size_t i = 0; i < std::size(texture_units); ++i) {
+    for (int i = 0; i < std::size(texture_units); ++i) {
         if (texture_units[i].texture_2d != cur_state.texture_units[i].texture_2d) {
             glActiveTexture(TextureUnits::MaxwellTexture(i).Enum());
             glBindTexture(GL_TEXTURE_2D, texture_units[i].texture_2d);
         }
         if (texture_units[i].sampler != cur_state.texture_units[i].sampler) {
-            glBindSampler(i, texture_units[i].sampler);
+            glBindSampler(static_cast<GLuint>(i), texture_units[i].sampler);
+        }
+        // Update the texture swizzle
+        if (texture_units[i].swizzle.r != cur_state.texture_units[i].swizzle.r ||
+            texture_units[i].swizzle.g != cur_state.texture_units[i].swizzle.g ||
+            texture_units[i].swizzle.b != cur_state.texture_units[i].swizzle.b ||
+            texture_units[i].swizzle.a != cur_state.texture_units[i].swizzle.a) {
+            std::array<GLint, 4> mask = {texture_units[i].swizzle.r, texture_units[i].swizzle.g,
+                                         texture_units[i].swizzle.b, texture_units[i].swizzle.a};
+            glTexParameteriv(GL_TEXTURE_2D, GL_TEXTURE_SWIZZLE_RGBA, mask.data());
         }
     }
 

src/video_core/renderer_opengl/gl_state.h

@@ -85,6 +85,12 @@ public:
     struct {
         GLuint texture_2d; // GL_TEXTURE_BINDING_2D
         GLuint sampler;    // GL_SAMPLER_BINDING
+        struct {
+            GLint r; // GL_TEXTURE_SWIZZLE_R
+            GLint g; // GL_TEXTURE_SWIZZLE_G
+            GLint b; // GL_TEXTURE_SWIZZLE_B
+            GLint a; // GL_TEXTURE_SWIZZLE_A
+        } swizzle;
     } texture_units[32];
 
     struct {

src/video_core/renderer_opengl/maxwell_to_gl.h

@@ -100,6 +100,8 @@ inline GLenum WrapMode(Tegra::Texture::WrapMode wrap_mode) {
     switch (wrap_mode) {
     case Tegra::Texture::WrapMode::Wrap:
         return GL_REPEAT;
+    case Tegra::Texture::WrapMode::Mirror:
+        return GL_MIRRORED_REPEAT;
     case Tegra::Texture::WrapMode::ClampToEdge:
         return GL_CLAMP_TO_EDGE;
     case Tegra::Texture::WrapMode::ClampOGL:
@@ -178,4 +180,25 @@ inline GLenum BlendFunc(Maxwell::Blend::Factor factor) {
     return {};
 }
 
+inline GLenum SwizzleSource(Tegra::Texture::SwizzleSource source) {
+    switch (source) {
+    case Tegra::Texture::SwizzleSource::Zero:
+        return GL_ZERO;
+    case Tegra::Texture::SwizzleSource::R:
+        return GL_RED;
+    case Tegra::Texture::SwizzleSource::G:
+        return GL_GREEN;
+    case Tegra::Texture::SwizzleSource::B:
+        return GL_BLUE;
+    case Tegra::Texture::SwizzleSource::A:
+        return GL_ALPHA;
+    case Tegra::Texture::SwizzleSource::OneInt:
+    case Tegra::Texture::SwizzleSource::OneFloat:
+        return GL_ONE;
+    }
+    NGLOG_CRITICAL(Render_OpenGL, "Unimplemented swizzle source={}", static_cast<u32>(source));
+    UNREACHABLE();
+    return {};
+}
+
 } // namespace MaxwellToGL

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -316,6 +316,7 @@ void RendererOpenGL::DrawScreenTriangles(const ScreenInfo& screen_info, float x,
     }};
 
     state.texture_units[0].texture_2d = screen_info.display_texture;
+    state.texture_units[0].swizzle = {GL_RED, GL_GREEN, GL_BLUE, GL_ALPHA};
     state.Apply();
 
     glBufferSubData(GL_ARRAY_BUFFER, 0, sizeof(vertices), vertices.data());

src/video_core/textures/astc.h

@@ -0,0 +1,15 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <cstdint>
+#include <vector>
+
+namespace Tegra::Texture::ASTC {
+
+std::vector<uint8_t> Decompress(std::vector<uint8_t>& data, uint32_t width, uint32_t height,
+                                uint32_t block_width, uint32_t block_height);
+
+} // namespace Tegra::Texture::ASTC

src/video_core/textures/decoders.cpp

@@ -53,8 +53,10 @@ u32 BytesPerPixel(TextureFormat format) {
     case TextureFormat::DXT45:
         // In this case a 'pixel' actually refers to a 4x4 tile.
         return 16;
+    case TextureFormat::ASTC_2D_4X4:
     case TextureFormat::A8R8G8B8:
     case TextureFormat::A2B10G10R10:
+    case TextureFormat::BF10GF11RF11:
         return 4;
     case TextureFormat::A1B5G5R5:
     case TextureFormat::B5G6R5:
@@ -92,6 +94,8 @@ std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width,
     case TextureFormat::B5G6R5:
     case TextureFormat::R8:
     case TextureFormat::R16_G16_B16_A16:
+    case TextureFormat::BF10GF11RF11:
+    case TextureFormat::ASTC_2D_4X4:
         CopySwizzledData(width, height, bytes_per_pixel, bytes_per_pixel, data,
                          unswizzled_data.data(), true, block_height);
         break;
@@ -113,11 +117,13 @@ std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat
     case TextureFormat::DXT23:
     case TextureFormat::DXT45:
     case TextureFormat::DXN1:
+    case TextureFormat::ASTC_2D_4X4:
     case TextureFormat::A8R8G8B8:
     case TextureFormat::A2B10G10R10:
     case TextureFormat::A1B5G5R5:
     case TextureFormat::B5G6R5:
     case TextureFormat::R8:
+    case TextureFormat::BF10GF11RF11:
         // TODO(Subv): For the time being just forward the same data without any decoding.
         rgba_data = texture_data;
         break;

src/video_core/textures/texture.h

@@ -122,6 +122,17 @@ enum class ComponentType : u32 {
     FLOAT = 7
 };
 
+enum class SwizzleSource : u32 {
+    Zero = 0,
+
+    R = 2,
+    G = 3,
+    B = 4,
+    A = 5,
+    OneInt = 6,
+    OneFloat = 7,
+};
+
 union TextureHandle {
     u32 raw;
     BitField<0, 20, u32> tic_id;
@@ -139,6 +150,11 @@ struct TICEntry {
         BitField<10, 3, ComponentType> g_type;
         BitField<13, 3, ComponentType> b_type;
         BitField<16, 3, ComponentType> a_type;
+
+        BitField<19, 3, SwizzleSource> x_source;
+        BitField<22, 3, SwizzleSource> y_source;
+        BitField<25, 3, SwizzleSource> z_source;
+        BitField<28, 3, SwizzleSource> w_source;
     };
     u32 address_low;
     union {

src/yuzu/CMakeLists.txt

@@ -32,8 +32,6 @@ add_executable(yuzu
     debugger/graphics/graphics_surface.h
     debugger/profiler.cpp
     debugger/profiler.h
-    debugger/registers.cpp
-    debugger/registers.h
     debugger/wait_tree.cpp
     debugger/wait_tree.h
     game_list.cpp
@@ -60,7 +58,6 @@ set(UIS
     configuration/configure_graphics.ui
     configuration/configure_input.ui
     configuration/configure_system.ui
-    debugger/registers.ui
     hotkeys.ui
     main.ui
 )

src/yuzu/debugger/registers.cpp

@@ -1,190 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <QTreeWidgetItem>
-#include "core/arm/arm_interface.h"
-#include "core/core.h"
-#include "yuzu/debugger/registers.h"
-#include "yuzu/util/util.h"
-
-RegistersWidget::RegistersWidget(QWidget* parent) : QDockWidget(parent) {
-    cpu_regs_ui.setupUi(this);
-
-    tree = cpu_regs_ui.treeWidget;
-    tree->addTopLevelItem(core_registers = new QTreeWidgetItem(QStringList(tr("Registers"))));
-    tree->addTopLevelItem(vfp_registers = new QTreeWidgetItem(QStringList(tr("VFP Registers"))));
-    tree->addTopLevelItem(vfp_system_registers =
-                              new QTreeWidgetItem(QStringList(tr("VFP System Registers"))));
-    tree->addTopLevelItem(cpsr = new QTreeWidgetItem(QStringList("CPSR")));
-
-    for (int i = 0; i < 16; ++i) {
-        QTreeWidgetItem* child = new QTreeWidgetItem(QStringList(QString("R[%1]").arg(i)));
-        core_registers->addChild(child);
-    }
-
-    for (int i = 0; i < 32; ++i) {
-        QTreeWidgetItem* child = new QTreeWidgetItem(QStringList(QString("S[%1]").arg(i)));
-        vfp_registers->addChild(child);
-    }
-
-    QFont font = GetMonospaceFont();
-
-    CreateCPSRChildren();
-    CreateVFPSystemRegisterChildren();
-
-    // Set Registers to display in monospace font
-    for (int i = 0; i < core_registers->childCount(); ++i)
-        core_registers->child(i)->setFont(1, font);
-
-    for (int i = 0; i < vfp_registers->childCount(); ++i)
-        vfp_registers->child(i)->setFont(1, font);
-
-    for (int i = 0; i < vfp_system_registers->childCount(); ++i) {
-        vfp_system_registers->child(i)->setFont(1, font);
-        for (int x = 0; x < vfp_system_registers->child(i)->childCount(); ++x) {
-            vfp_system_registers->child(i)->child(x)->setFont(1, font);
-        }
-    }
-    // Set CSPR to display in monospace font
-    cpsr->setFont(1, font);
-    for (int i = 0; i < cpsr->childCount(); ++i) {
-        cpsr->child(i)->setFont(1, font);
-        for (int x = 0; x < cpsr->child(i)->childCount(); ++x) {
-            cpsr->child(i)->child(x)->setFont(1, font);
-        }
-    }
-    setEnabled(false);
-}
-
-void RegistersWidget::OnDebugModeEntered() {
-    if (!Core::System::GetInstance().IsPoweredOn())
-        return;
-
-    for (int i = 0; i < core_registers->childCount(); ++i)
-        core_registers->child(i)->setText(
-            1, QString("0x%1").arg(Core::CurrentArmInterface().GetReg(i), 8, 16, QLatin1Char('0')));
-
-    UpdateCPSRValues();
-}
-
-void RegistersWidget::OnDebugModeLeft() {}
-
-void RegistersWidget::OnEmulationStarting(EmuThread* emu_thread) {
-    setEnabled(true);
-}
-
-void RegistersWidget::OnEmulationStopping() {
-    // Reset widget text
-    for (int i = 0; i < core_registers->childCount(); ++i)
-        core_registers->child(i)->setText(1, QString(""));
-
-    for (int i = 0; i < vfp_registers->childCount(); ++i)
-        vfp_registers->child(i)->setText(1, QString(""));
-
-    for (int i = 0; i < cpsr->childCount(); ++i)
-        cpsr->child(i)->setText(1, QString(""));
-
-    cpsr->setText(1, QString(""));
-
-    // FPSCR
-    for (int i = 0; i < vfp_system_registers->child(0)->childCount(); ++i)
-        vfp_system_registers->child(0)->child(i)->setText(1, QString(""));
-
-    // FPEXC
-    for (int i = 0; i < vfp_system_registers->child(1)->childCount(); ++i)
-        vfp_system_registers->child(1)->child(i)->setText(1, QString(""));
-
-    vfp_system_registers->child(0)->setText(1, QString(""));
-    vfp_system_registers->child(1)->setText(1, QString(""));
-    vfp_system_registers->child(2)->setText(1, QString(""));
-    vfp_system_registers->child(3)->setText(1, QString(""));
-
-    setEnabled(false);
-}
-
-void RegistersWidget::CreateCPSRChildren() {
-    cpsr->addChild(new QTreeWidgetItem(QStringList("M")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("T")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("F")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("I")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("A")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("E")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("IT")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("GE")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("DNM")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("J")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("Q")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("V")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("C")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("Z")));
-    cpsr->addChild(new QTreeWidgetItem(QStringList("N")));
-}
-
-void RegistersWidget::UpdateCPSRValues() {
-    const u32 cpsr_val = Core::CurrentArmInterface().GetCPSR();
-
-    cpsr->setText(1, QString("0x%1").arg(cpsr_val, 8, 16, QLatin1Char('0')));
-    cpsr->child(0)->setText(
-        1, QString("b%1").arg(cpsr_val & 0x1F, 5, 2, QLatin1Char('0'))); // M - Mode
-    cpsr->child(1)->setText(1, QString::number((cpsr_val >> 5) & 1));    // T - State
-    cpsr->child(2)->setText(1, QString::number((cpsr_val >> 6) & 1));    // F - FIQ disable
-    cpsr->child(3)->setText(1, QString::number((cpsr_val >> 7) & 1));    // I - IRQ disable
-    cpsr->child(4)->setText(1, QString::number((cpsr_val >> 8) & 1));    // A - Imprecise abort
-    cpsr->child(5)->setText(1, QString::number((cpsr_val >> 9) & 1));    // E - Data endianness
-    cpsr->child(6)->setText(1,
-                            QString::number((cpsr_val >> 10) & 0x3F)); // IT - If-Then state (DNM)
-    cpsr->child(7)->setText(1,
-                            QString::number((cpsr_val >> 16) & 0xF)); // GE - Greater-than-or-Equal
-    cpsr->child(8)->setText(1, QString::number((cpsr_val >> 20) & 0xF)); // DNM - Do not modify
-    cpsr->child(9)->setText(1, QString::number((cpsr_val >> 24) & 1));   // J - Jazelle
-    cpsr->child(10)->setText(1, QString::number((cpsr_val >> 27) & 1));  // Q - Saturation
-    cpsr->child(11)->setText(1, QString::number((cpsr_val >> 28) & 1));  // V - Overflow
-    cpsr->child(12)->setText(1, QString::number((cpsr_val >> 29) & 1));  // C - Carry/Borrow/Extend
-    cpsr->child(13)->setText(1, QString::number((cpsr_val >> 30) & 1));  // Z - Zero
-    cpsr->child(14)->setText(1, QString::number((cpsr_val >> 31) & 1));  // N - Negative/Less than
-}
-
-void RegistersWidget::CreateVFPSystemRegisterChildren() {
-    QTreeWidgetItem* const fpscr = new QTreeWidgetItem(QStringList("FPSCR"));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("IOC")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("DZC")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("OFC")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("UFC")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("IXC")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("IDC")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("IOE")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("DZE")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("OFE")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("UFE")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("IXE")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("IDE")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList(tr("Vector Length"))));
-    fpscr->addChild(new QTreeWidgetItem(QStringList(tr("Vector Stride"))));
-    fpscr->addChild(new QTreeWidgetItem(QStringList(tr("Rounding Mode"))));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("FZ")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("DN")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("V")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("C")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("Z")));
-    fpscr->addChild(new QTreeWidgetItem(QStringList("N")));
-
-    QTreeWidgetItem* const fpexc = new QTreeWidgetItem(QStringList("FPEXC"));
-    fpexc->addChild(new QTreeWidgetItem(QStringList("IOC")));
-    fpexc->addChild(new QTreeWidgetItem(QStringList("OFC")));
-    fpexc->addChild(new QTreeWidgetItem(QStringList("UFC")));
-    fpexc->addChild(new QTreeWidgetItem(QStringList("INV")));
-    fpexc->addChild(new QTreeWidgetItem(QStringList(tr("Vector Iteration Count"))));
-    fpexc->addChild(new QTreeWidgetItem(QStringList("FP2V")));
-    fpexc->addChild(new QTreeWidgetItem(QStringList("EN")));
-    fpexc->addChild(new QTreeWidgetItem(QStringList("EX")));
-
-    vfp_system_registers->addChild(fpscr);
-    vfp_system_registers->addChild(fpexc);
-    vfp_system_registers->addChild(new QTreeWidgetItem(QStringList("FPINST")));
-    vfp_system_registers->addChild(new QTreeWidgetItem(QStringList("FPINST2")));
-}
-
-void RegistersWidget::UpdateVFPSystemRegisterValues() {
-    UNIMPLEMENTED();
-}

src/yuzu/debugger/registers.h

@@ -1,42 +0,0 @@
-// Copyright 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <QDockWidget>
-#include "ui_registers.h"
-
-class QTreeWidget;
-class QTreeWidgetItem;
-class EmuThread;
-
-class RegistersWidget : public QDockWidget {
-    Q_OBJECT
-
-public:
-    explicit RegistersWidget(QWidget* parent = nullptr);
-
-public slots:
-    void OnDebugModeEntered();
-    void OnDebugModeLeft();
-
-    void OnEmulationStarting(EmuThread* emu_thread);
-    void OnEmulationStopping();
-
-private:
-    void CreateCPSRChildren();
-    void UpdateCPSRValues();
-
-    void CreateVFPSystemRegisterChildren();
-    void UpdateVFPSystemRegisterValues();
-
-    Ui::ARMRegisters cpu_regs_ui;
-
-    QTreeWidget* tree;
-
-    QTreeWidgetItem* core_registers;
-    QTreeWidgetItem* vfp_registers;
-    QTreeWidgetItem* vfp_system_registers;
-    QTreeWidgetItem* cpsr;
-};

src/yuzu/debugger/registers.ui

@@ -1,40 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<ui version="4.0">
- <class>ARMRegisters</class>
- <widget class="QDockWidget" name="ARMRegisters">
-  <property name="geometry">
-   <rect>
-    <x>0</x>
-    <y>0</y>
-    <width>400</width>
-    <height>300</height>
-   </rect>
-  </property>
-  <property name="windowTitle">
-   <string>ARM Registers</string>
-  </property>
-  <widget class="QWidget" name="dockWidgetContents">
-   <layout class="QVBoxLayout" name="verticalLayout">
-    <item>
-     <widget class="QTreeWidget" name="treeWidget">
-      <property name="alternatingRowColors">
-       <bool>true</bool>
-      </property>
-      <column>
-       <property name="text">
-        <string>Register</string>
-       </property>
-      </column>
-      <column>
-       <property name="text">
-        <string>Value</string>
-       </property>
-      </column>
-     </widget>
-    </item>
-   </layout>
-  </widget>
- </widget>
- <resources/>
- <connections/>
-</ui>

src/yuzu/game_list.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <QApplication>
+#include <QDir>
 #include <QFileInfo>
 #include <QHeaderView>
 #include <QKeyEvent>
@@ -264,8 +265,17 @@ void GameList::ValidateEntry(const QModelIndex& item) {
     if (file_path.isEmpty())
         return;
     std::string std_file_path(file_path.toStdString());
-    if (!FileUtil::Exists(std_file_path) || FileUtil::IsDirectory(std_file_path))
+    if (!FileUtil::Exists(std_file_path))
         return;
+    if (FileUtil::IsDirectory(std_file_path)) {
+        QDir dir(std_file_path.c_str());
+        QStringList matching_main = dir.entryList(QStringList("main"), QDir::Files);
+        if (matching_main.size() == 1) {
+            emit GameChosen(dir.path() + DIR_SEP + matching_main[0]);
+        }
+        return;
+    }
+
     // Users usually want to run a diffrent game after closing one
     search_field->clear();
     emit GameChosen(file_path);
@@ -363,6 +373,19 @@ static bool HasSupportedFileExtension(const std::string& file_name) {
     return GameList::supported_file_extensions.contains(file.suffix(), Qt::CaseInsensitive);
 }
 
+static bool IsExtractedNCAMain(const std::string& file_name) {
+    return QFileInfo(file_name.c_str()).fileName() == "main";
+}
+
+static QString FormatGameName(const std::string& physical_name) {
+    QFileInfo file_info(physical_name.c_str());
+    if (IsExtractedNCAMain(physical_name)) {
+        return file_info.dir().path();
+    } else {
+        return QString::fromStdString(physical_name);
+    }
+}
+
 void GameList::RefreshGameDirectory() {
     if (!UISettings::values.gamedir.isEmpty() && current_worker != nullptr) {
         NGLOG_INFO(Frontend, "Change detected in the games directory. Reloading game list.");
@@ -380,7 +403,8 @@ void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsign
             return false; // Breaks the callback loop.
 
         bool is_dir = FileUtil::IsDirectory(physical_name);
-        if (!is_dir && HasSupportedFileExtension(physical_name)) {
+        if (!is_dir &&
+            (HasSupportedFileExtension(physical_name) || IsExtractedNCAMain(physical_name))) {
             std::unique_ptr<Loader::AppLoader> loader = Loader::GetLoader(physical_name);
             if (!loader)
                 return true;
@@ -392,7 +416,7 @@ void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsign
             loader->ReadProgramId(program_id);
 
             emit EntryReady({
-                new GameListItemPath(QString::fromStdString(physical_name), smdh, program_id),
+                new GameListItemPath(FormatGameName(physical_name), smdh, program_id),
                 new GameListItem(
                     QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))),
                 new GameListItemSize(FileUtil::GetSize(physical_name)),

src/yuzu/main.cpp

@@ -13,6 +13,7 @@
 #include <QMessageBox>
 #include <QtGui>
 #include <QtWidgets>
+#include "common/common_paths.h"
 #include "common/logging/backend.h"
 #include "common/logging/filter.h"
 #include "common/logging/log.h"
@@ -33,7 +34,6 @@
 #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"
 #include "yuzu/game_list.h"
 #include "yuzu/hotkeys.h"
@@ -169,15 +169,6 @@ void GMainWindow::InitializeDebugWidgets() {
     debug_menu->addAction(microProfileDialog->toggleViewAction());
 #endif
 
-    registersWidget = new RegistersWidget(this);
-    addDockWidget(Qt::RightDockWidgetArea, registersWidget);
-    registersWidget->hide();
-    debug_menu->addAction(registersWidget->toggleViewAction());
-    connect(this, &GMainWindow::EmulationStarting, registersWidget,
-            &RegistersWidget::OnEmulationStarting);
-    connect(this, &GMainWindow::EmulationStopping, registersWidget,
-            &RegistersWidget::OnEmulationStopping);
-
     graphicsBreakpointsWidget = new GraphicsBreakPointsWidget(debug_context, this);
     addDockWidget(Qt::RightDockWidgetArea, graphicsBreakpointsWidget);
     graphicsBreakpointsWidget->hide();
@@ -288,6 +279,7 @@ void GMainWindow::ConnectWidgetEvents() {
 void GMainWindow::ConnectMenuEvents() {
     // File
     connect(ui.action_Load_File, &QAction::triggered, this, &GMainWindow::OnMenuLoadFile);
+    connect(ui.action_Load_Folder, &QAction::triggered, this, &GMainWindow::OnMenuLoadFolder);
     connect(ui.action_Select_Game_List_Root, &QAction::triggered, this,
             &GMainWindow::OnMenuSelectGameListRoot);
     connect(ui.action_Exit, &QAction::triggered, this, &QMainWindow::close);
@@ -460,17 +452,12 @@ void GMainWindow::BootGame(const QString& filename) {
     connect(render_window, &GRenderWindow::Closed, this, &GMainWindow::OnStopGame);
     // BlockingQueuedConnection is important here, it makes sure we've finished refreshing our views
     // before the CPU continues
-    connect(emu_thread.get(), &EmuThread::DebugModeEntered, registersWidget,
-            &RegistersWidget::OnDebugModeEntered, Qt::BlockingQueuedConnection);
     connect(emu_thread.get(), &EmuThread::DebugModeEntered, waitTreeWidget,
             &WaitTreeWidget::OnDebugModeEntered, Qt::BlockingQueuedConnection);
-    connect(emu_thread.get(), &EmuThread::DebugModeLeft, registersWidget,
-            &RegistersWidget::OnDebugModeLeft, Qt::BlockingQueuedConnection);
     connect(emu_thread.get(), &EmuThread::DebugModeLeft, waitTreeWidget,
             &WaitTreeWidget::OnDebugModeLeft, Qt::BlockingQueuedConnection);
 
     // Update the GUI
-    registersWidget->OnDebugModeEntered();
     if (ui.action_Single_Window_Mode->isChecked()) {
         game_list->hide();
     }
@@ -565,6 +552,8 @@ void GMainWindow::OnMenuLoadFile() {
     for (const auto& piece : game_list->supported_file_extensions)
         extensions += "*." + piece + " ";
 
+    extensions += "main ";
+
     QString file_filter = tr("Switch Executable") + " (" + extensions + ")";
     file_filter += ";;" + tr("All Files (*.*)");
 
@@ -577,6 +566,18 @@ void GMainWindow::OnMenuLoadFile() {
     }
 }
 
+void GMainWindow::OnMenuLoadFolder() {
+    QDir dir = QFileDialog::getExistingDirectory(this, tr("Open Extracted ROM Directory"));
+
+    QStringList matching_main = dir.entryList(QStringList("main"), QDir::Files);
+    if (matching_main.size() == 1) {
+        BootGame(dir.path() + DIR_SEP + matching_main[0]);
+    } else {
+        QMessageBox::warning(this, tr("Invalid Directory Selected"),
+                             tr("The directory you have selected does not contain a 'main' file."));
+    }
+}
+
 void GMainWindow::OnMenuSelectGameListRoot() {
     QString dir_path = QFileDialog::getExistingDirectory(this, tr("Select Directory"));
     if (!dir_path.isEmpty()) {

src/yuzu/main.h

@@ -19,7 +19,6 @@ class GraphicsSurfaceWidget;
 class GRenderWindow;
 class MicroProfileDialog;
 class ProfilerWidget;
-class RegistersWidget;
 class WaitTreeWidget;
 
 namespace Tegra {
@@ -124,6 +123,7 @@ private slots:
     void OnGameListLoadFile(QString game_path);
     void OnGameListOpenSaveFolder(u64 program_id);
     void OnMenuLoadFile();
+    void OnMenuLoadFolder();
     /// Called whenever a user selects the "File->Select Game List Root" menu item
     void OnMenuSelectGameListRoot();
     void OnMenuRecentFile();
@@ -163,7 +163,6 @@ private:
     // Debugger panes
     ProfilerWidget* profilerWidget;
     MicroProfileDialog* microProfileDialog;
-    RegistersWidget* registersWidget;
     GraphicsBreakPointsWidget* graphicsBreakpointsWidget;
     GraphicsSurfaceWidget* graphicsSurfaceWidget;
     WaitTreeWidget* waitTreeWidget;

src/yuzu/main.ui

@@ -58,6 +58,7 @@
      </property>
     </widget>
     <addaction name="action_Load_File"/>
+    <addaction name="action_Load_Folder"/>
     <addaction name="separator"/>
     <addaction name="action_Select_Game_List_Root"/>
     <addaction name="menu_recent_files"/>
@@ -106,6 +107,11 @@
     <string>Load File...</string>
    </property>
   </action>
+  <action name="action_Load_Folder">
+   <property name="text">
+    <string>Load Folder...</string>
+   </property>
+  </action>
   <action name="action_Load_Symbol_Map">
    <property name="text">
     <string>Load Symbol Map...</string>