Remove unused header files

GPU: Implemented predicated exit instructions in the shader programs.

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 -C mingw_build/ 2>&1'
+          C:\msys64\usr\bin\bash.exe -lc 'mingw32-make -j4 -C mingw_build/ 2>&1'
         }
 
 after_build:

src/common/logging/backend.cpp

@@ -41,6 +41,7 @@ namespace Log {
     SUB(Service, FS)                                                                               \
     SUB(Service, HID)                                                                              \
     SUB(Service, LM)                                                                               \
+    SUB(Service, MM)                                                                               \
     SUB(Service, NFP)                                                                              \
     SUB(Service, NIFM)                                                                             \
     SUB(Service, NS)                                                                               \

src/common/logging/log.h

@@ -61,6 +61,7 @@ enum class Class : ClassType {
     Service_FS,        ///< The FS (Filesystem) service
     Service_HID,       ///< The HID (Human interface device) service
     Service_LM,        ///< The LM (Logger) service
+    Service_MM,        ///< The MM (Multimedia) service
     Service_NFP,       ///< The NFP service
     Service_NIFM,      ///< The NIFM (Network interface) service
     Service_NS,        ///< The NS services

src/core/CMakeLists.txt

@@ -148,6 +148,8 @@ add_library(core STATIC
     hle/service/hid/hid.h
     hle/service/lm/lm.cpp
     hle/service/lm/lm.h
+    hle/service/mm/mm_u.cpp
+    hle/service/mm/mm_u.h
     hle/service/nifm/nifm.cpp
     hle/service/nifm/nifm.h
     hle/service/nifm/nifm_a.cpp

src/core/hle/ipc.h

@@ -29,7 +29,10 @@ enum class ControlCommand : u32 {
 };
 
 enum class CommandType : u32 {
+    Invalid = 0,
+    LegacyRequest = 1,
     Close = 2,
+    LegacyControl = 3,
     Request = 4,
     Control = 5,
     RequestWithContext = 6,

src/core/hle/kernel/errors.h

@@ -21,7 +21,9 @@ enum {
 
     // Confirmed Switch OS error codes
     MisalignedAddress = 102,
+    InvalidProcessorId = 113,
     InvalidHandle = 114,
+    InvalidCombination = 116,
     Timeout = 117,
     SynchronizationCanceled = 118,
     TooLarge = 119,

src/core/hle/kernel/svc.cpp

@@ -732,7 +732,7 @@ static ResultCode GetThreadCoreMask(Handle thread_handle, u32* core, u64* mask)
 }
 
 static ResultCode SetThreadCoreMask(Handle thread_handle, u32 core, u64 mask) {
-    NGLOG_TRACE(Kernel_SVC, "called, handle=0x{:08X}, mask=0x{:08X}, core=0x{:X}", thread_handle,
+    NGLOG_DEBUG(Kernel_SVC, "called, handle=0x{:08X}, mask=0x{:16X}, core=0x{:X}", thread_handle,
                 mask, core);
 
     const SharedPtr<Thread> thread = g_handle_table.Get<Thread>(thread_handle);
@@ -740,6 +740,31 @@ static ResultCode SetThreadCoreMask(Handle thread_handle, u32 core, u64 mask) {
         return ERR_INVALID_HANDLE;
     }
 
+    if (core == THREADPROCESSORID_DEFAULT) {
+        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;
+    }
+
+    if (mask == 0) {
+        return ResultCode(ErrorModule::Kernel, ErrCodes::InvalidCombination);
+    }
+
+    /// This value is used to only change the affinity mask without changing the current ideal core.
+    static constexpr u32 OnlyChangeMask = static_cast<u32>(-3);
+
+    if (core == OnlyChangeMask) {
+        core = thread->ideal_core;
+    } else if (core >= Core::NUM_CPU_CORES && core != -1) {
+        return ResultCode(ErrorModule::Kernel, ErrCodes::InvalidProcessorId);
+    }
+
+    // Error out if the input core isn't enabled in the input mask.
+    if (core < Core::NUM_CPU_CORES && (mask & (1 << core)) == 0) {
+        return ResultCode(ErrorModule::Kernel, ErrCodes::InvalidCombination);
+    }
+
     thread->ChangeCore(core, mask);
 
     return RESULT_SUCCESS;

src/core/hle/kernel/thread.cpp

@@ -133,8 +133,11 @@ static void ThreadWakeupCallback(u64 thread_handle, int cycles_late) {
 
         auto lock_owner = thread->lock_owner;
         // Threads waking up by timeout from WaitProcessWideKey do not perform priority inheritance
-        // and don't have a lock owner.
-        ASSERT(lock_owner == nullptr);
+        // and don't have a lock owner unless SignalProcessWideKey was called first and the thread
+        // wasn't awakened due to the mutex already being acquired.
+        if (lock_owner) {
+            lock_owner->RemoveMutexWaiter(thread);
+        }
     }
 
     if (resume)
@@ -460,13 +463,13 @@ void Thread::UpdatePriority() {
 
 void Thread::ChangeCore(u32 core, u64 mask) {
     ideal_core = core;
-    mask = mask;
+    affinity_mask = mask;
 
     if (status != THREADSTATUS_READY) {
         return;
     }
 
-    boost::optional<s32> new_processor_id{GetNextProcessorId(mask)};
+    boost::optional<s32> new_processor_id{GetNextProcessorId(affinity_mask)};
 
     if (!new_processor_id) {
         new_processor_id = processor_id;
@@ -476,7 +479,7 @@ void Thread::ChangeCore(u32 core, u64 mask) {
         new_processor_id = ideal_core;
     }
 
-    ASSERT(new_processor_id < 4);
+    ASSERT(*new_processor_id < 4);
 
     // Add thread to new core's scheduler
     auto& next_scheduler = Core::System().GetInstance().Scheduler(*new_processor_id);

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

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <cinttypes>
+#include <stack>
 #include "core/file_sys/filesystem.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/event.h"
@@ -154,7 +155,7 @@ ISelfController::ISelfController(std::shared_ptr<NVFlinger::NVFlinger> nvflinger
     RegisterHandlers(functions);
 
     launchable_event =
-        Kernel::Event::Create(Kernel::ResetType::OneShot, "ISelfController:LaunchableEvent");
+        Kernel::Event::Create(Kernel::ResetType::Sticky, "ISelfController:LaunchableEvent");
 }
 
 void ISelfController::SetFocusHandlingMode(Kernel::HLERequestContext& ctx) {
@@ -348,19 +349,100 @@ void ICommonStateGetter::GetPerformanceMode(Kernel::HLERequestContext& ctx) {
     NGLOG_WARNING(Service_AM, "(STUBBED) called");
 }
 
+class IStorageAccessor final : public ServiceFramework<IStorageAccessor> {
+public:
+    explicit IStorageAccessor(std::vector<u8> buffer)
+        : ServiceFramework("IStorageAccessor"), buffer(std::move(buffer)) {
+        static const FunctionInfo functions[] = {
+            {0, &IStorageAccessor::GetSize, "GetSize"},
+            {10, &IStorageAccessor::Write, "Write"},
+            {11, &IStorageAccessor::Read, "Read"},
+        };
+        RegisterHandlers(functions);
+    }
+
+private:
+    std::vector<u8> buffer;
+
+    void GetSize(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 4};
+
+        rb.Push(RESULT_SUCCESS);
+        rb.Push(static_cast<u64>(buffer.size()));
+
+        NGLOG_DEBUG(Service_AM, "called");
+    }
+
+    void Write(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+
+        const u64 offset{rp.Pop<u64>()};
+        const std::vector<u8> data{ctx.ReadBuffer()};
+
+        ASSERT(offset + data.size() <= buffer.size());
+
+        std::memcpy(&buffer[offset], data.data(), data.size());
+
+        IPC::ResponseBuilder rb{rp.MakeBuilder(2, 0, 0)};
+        rb.Push(RESULT_SUCCESS);
+
+        NGLOG_DEBUG(Service_AM, "called, offset={}", offset);
+    }
+
+    void Read(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+
+        const u64 offset{rp.Pop<u64>()};
+        const size_t size{ctx.GetWriteBufferSize()};
+
+        ASSERT(offset + size <= buffer.size());
+
+        ctx.WriteBuffer(buffer.data() + offset, size);
+
+        IPC::ResponseBuilder rb{rp.MakeBuilder(2, 0, 0)};
+        rb.Push(RESULT_SUCCESS);
+
+        NGLOG_DEBUG(Service_AM, "called, offset={}", offset);
+    }
+};
+
+class IStorage final : public ServiceFramework<IStorage> {
+public:
+    explicit IStorage(std::vector<u8> buffer)
+        : ServiceFramework("IStorage"), buffer(std::move(buffer)) {
+        static const FunctionInfo functions[] = {
+            {0, &IStorage::Open, "Open"},
+            {1, nullptr, "OpenTransferStorage"},
+        };
+        RegisterHandlers(functions);
+    }
+
+private:
+    std::vector<u8> buffer;
+
+    void Open(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<AM::IStorageAccessor>(buffer);
+
+        NGLOG_DEBUG(Service_AM, "called");
+    }
+};
+
 class ILibraryAppletAccessor final : public ServiceFramework<ILibraryAppletAccessor> {
 public:
     explicit ILibraryAppletAccessor() : ServiceFramework("ILibraryAppletAccessor") {
         static const FunctionInfo functions[] = {
             {0, &ILibraryAppletAccessor::GetAppletStateChangedEvent, "GetAppletStateChangedEvent"},
             {1, nullptr, "IsCompleted"},
-            {10, nullptr, "Start"},
+            {10, &ILibraryAppletAccessor::Start, "Start"},
             {20, nullptr, "RequestExit"},
             {25, nullptr, "Terminate"},
-            {30, nullptr, "GetResult"},
+            {30, &ILibraryAppletAccessor::GetResult, "GetResult"},
             {50, nullptr, "SetOutOfFocusApplicationSuspendingEnabled"},
-            {100, nullptr, "PushInData"},
-            {101, nullptr, "PopOutData"},
+            {100, &ILibraryAppletAccessor::PushInData, "PushInData"},
+            {101, &ILibraryAppletAccessor::PopOutData, "PopOutData"},
             {102, nullptr, "PushExtraStorage"},
             {103, nullptr, "PushInteractiveInData"},
             {104, nullptr, "PopInteractiveOutData"},
@@ -388,6 +470,41 @@ private:
         NGLOG_WARNING(Service_AM, "(STUBBED) called");
     }
 
+    void GetResult(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+
+        NGLOG_WARNING(Service_AM, "(STUBBED) called");
+    }
+
+    void Start(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+
+        NGLOG_WARNING(Service_AM, "(STUBBED) called");
+    }
+
+    void PushInData(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        storage_stack.push(rp.PopIpcInterface<AM::IStorage>());
+
+        IPC::ResponseBuilder rb{rp.MakeBuilder(2, 0, 0)};
+        rb.Push(RESULT_SUCCESS);
+
+        NGLOG_DEBUG(Service_AM, "called");
+    }
+
+    void PopOutData(Kernel::HLERequestContext& ctx) {
+        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushIpcInterface<AM::IStorage>(std::move(storage_stack.top()));
+
+        storage_stack.pop();
+
+        NGLOG_DEBUG(Service_AM, "called");
+    }
+
+    std::stack<std::shared_ptr<AM::IStorage>> storage_stack;
     Kernel::SharedPtr<Kernel::Event> state_changed_event;
 };
 
@@ -396,7 +513,7 @@ ILibraryAppletCreator::ILibraryAppletCreator() : ServiceFramework("ILibraryApple
         {0, &ILibraryAppletCreator::CreateLibraryApplet, "CreateLibraryApplet"},
         {1, nullptr, "TerminateAllLibraryApplets"},
         {2, nullptr, "AreAnyLibraryAppletsLeft"},
-        {10, nullptr, "CreateStorage"},
+        {10, &ILibraryAppletCreator::CreateStorage, "CreateStorage"},
         {11, nullptr, "CreateTransferMemoryStorage"},
         {12, nullptr, "CreateHandleStorage"},
     };
@@ -412,72 +529,17 @@ void ILibraryAppletCreator::CreateLibraryApplet(Kernel::HLERequestContext& ctx)
     NGLOG_DEBUG(Service_AM, "called");
 }
 
-class IStorageAccessor final : public ServiceFramework<IStorageAccessor> {
-public:
-    explicit IStorageAccessor(std::vector<u8> buffer)
-        : ServiceFramework("IStorageAccessor"), buffer(std::move(buffer)) {
-        static const FunctionInfo functions[] = {
-            {0, &IStorageAccessor::GetSize, "GetSize"},
-            {10, nullptr, "Write"},
-            {11, &IStorageAccessor::Read, "Read"},
-        };
-        RegisterHandlers(functions);
-    }
+void ILibraryAppletCreator::CreateStorage(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    const u64 size{rp.Pop<u64>()};
+    std::vector<u8> buffer(size);
 
-private:
-    std::vector<u8> buffer;
+    IPC::ResponseBuilder rb{rp.MakeBuilder(2, 0, 1)};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushIpcInterface<AM::IStorage>(std::move(buffer));
 
-    void GetSize(Kernel::HLERequestContext& ctx) {
-        IPC::ResponseBuilder rb{ctx, 4};
-
-        rb.Push(RESULT_SUCCESS);
-        rb.Push(static_cast<u64>(buffer.size()));
-
-        NGLOG_DEBUG(Service_AM, "called");
-    }
-
-    void Read(Kernel::HLERequestContext& ctx) {
-        IPC::RequestParser rp{ctx};
-
-        u64 offset = rp.Pop<u64>();
-
-        const size_t size{ctx.GetWriteBufferSize()};
-
-        ASSERT(offset + size <= buffer.size());
-
-        ctx.WriteBuffer(buffer.data() + offset, size);
-
-        IPC::ResponseBuilder rb{ctx, 2};
-
-        rb.Push(RESULT_SUCCESS);
-
-        NGLOG_DEBUG(Service_AM, "called");
-    }
-};
-
-class IStorage final : public ServiceFramework<IStorage> {
-public:
-    explicit IStorage(std::vector<u8> buffer)
-        : ServiceFramework("IStorage"), buffer(std::move(buffer)) {
-        static const FunctionInfo functions[] = {
-            {0, &IStorage::Open, "Open"},
-            {1, nullptr, "OpenTransferStorage"},
-        };
-        RegisterHandlers(functions);
-    }
-
-private:
-    std::vector<u8> buffer;
-
-    void Open(Kernel::HLERequestContext& ctx) {
-        IPC::ResponseBuilder rb{ctx, 2, 0, 1};
-
-        rb.Push(RESULT_SUCCESS);
-        rb.PushIpcInterface<AM::IStorageAccessor>(buffer);
-
-        NGLOG_DEBUG(Service_AM, "called");
-    }
-};
+    NGLOG_DEBUG(Service_AM, "called, size={}", size);
+}
 
 IApplicationFunctions::IApplicationFunctions() : ServiceFramework("IApplicationFunctions") {
     static const FunctionInfo functions[] = {

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

@@ -121,6 +121,7 @@ public:
 
 private:
     void CreateLibraryApplet(Kernel::HLERequestContext& ctx);
+    void CreateStorage(Kernel::HLERequestContext& ctx);
 };
 
 class IApplicationFunctions final : public ServiceFramework<IApplicationFunctions> {

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

@@ -329,6 +329,7 @@ public:
             {130, nullptr, "SwapNpadAssignment"},
             {131, nullptr, "IsUnintendedHomeButtonInputProtectionEnabled"},
             {132, nullptr, "EnableUnintendedHomeButtonInputProtection"},
+            {133, nullptr, "SetNpadJoyAssignmentModeSingleWithDestination"},
             {200, &Hid::GetVibrationDeviceInfo, "GetVibrationDeviceInfo"},
             {201, &Hid::SendVibrationValue, "SendVibrationValue"},
             {202, &Hid::GetActualVibrationValue, "GetActualVibrationValue"},
@@ -336,12 +337,41 @@ public:
             {204, nullptr, "PermitVibration"},
             {205, nullptr, "IsVibrationPermitted"},
             {206, &Hid::SendVibrationValues, "SendVibrationValues"},
+            {207, nullptr, "SendVibrationGcErmCommand"},
+            {208, nullptr, "GetActualVibrationGcErmCommand"},
+            {209, nullptr, "BeginPermitVibrationSession"},
+            {210, nullptr, "EndPermitVibrationSession"},
             {300, nullptr, "ActivateConsoleSixAxisSensor"},
             {301, nullptr, "StartConsoleSixAxisSensor"},
             {302, nullptr, "StopConsoleSixAxisSensor"},
+            {303, nullptr, "ActivateSevenSixAxisSensor"},
+            {304, nullptr, "StartSevenSixAxisSensor"},
+            {305, nullptr, "StopSevenSixAxisSensor"},
+            {306, nullptr, "InitializeSevenSixAxisSensor"},
+            {307, nullptr, "FinalizeSevenSixAxisSensor"},
+            {308, nullptr, "SetSevenSixAxisSensorFusionStrength"},
+            {309, nullptr, "GetSevenSixAxisSensorFusionStrength"},
             {400, nullptr, "IsUsbFullKeyControllerEnabled"},
             {401, nullptr, "EnableUsbFullKeyController"},
             {402, nullptr, "IsUsbFullKeyControllerConnected"},
+            {403, nullptr, "HasBattery"},
+            {404, nullptr, "HasLeftRightBattery"},
+            {405, nullptr, "GetNpadInterfaceType"},
+            {406, nullptr, "GetNpadLeftRightInterfaceType"},
+            {500, nullptr, "GetPalmaConnectionHandle"},
+            {501, nullptr, "InitializePalma"},
+            {502, nullptr, "AcquirePalmaOperationCompleteEvent"},
+            {503, nullptr, "GetPalmaOperationInfo"},
+            {504, nullptr, "PlayPalmaActivity"},
+            {505, nullptr, "SetPalmaFrModeType"},
+            {506, nullptr, "ReadPalmaStep"},
+            {507, nullptr, "EnablePalmaStep"},
+            {508, nullptr, "SuspendPalmaStep"},
+            {509, nullptr, "ResetPalmaStep"},
+            {510, nullptr, "ReadPalmaApplicationSection"},
+            {511, nullptr, "WritePalmaApplicationSection"},
+            {512, nullptr, "ReadPalmaUniqueCode"},
+            {513, nullptr, "SetPalmaUniqueCodeInvalid"},
             {1000, nullptr, "SetNpadCommunicationMode"},
             {1001, nullptr, "GetNpadCommunicationMode"},
         };

src/core/hle/service/mm/mm_u.cpp

@@ -0,0 +1,50 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/logging/log.h"
+#include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/client_session.h"
+#include "core/hle/service/mm/mm_u.h"
+
+namespace Service::MM {
+
+void InstallInterfaces(SM::ServiceManager& service_manager) {
+    std::make_shared<MM_U>()->InstallAsService(service_manager);
+}
+
+void MM_U::Initialize(Kernel::HLERequestContext& ctx) {
+    NGLOG_WARNING(Service_MM, "(STUBBED) called");
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+}
+
+void MM_U::SetAndWait(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    min = rp.Pop<u32>();
+    max = rp.Pop<u32>();
+    current = min;
+
+    NGLOG_WARNING(Service_MM, "(STUBBED) called, min=0x{:X}, max=0x{:X}", min, max);
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+}
+
+void MM_U::Get(Kernel::HLERequestContext& ctx) {
+    NGLOG_WARNING(Service_MM, "(STUBBED) called");
+    IPC::ResponseBuilder rb{ctx, 3};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push(current);
+}
+
+MM_U::MM_U() : ServiceFramework("mm:u") {
+    static const FunctionInfo functions[] = {
+        {0, nullptr, "InitializeOld"},        {1, nullptr, "FinalizeOld"},
+        {2, nullptr, "SetAndWaitOld"},        {3, nullptr, "GetOld"},
+        {4, &MM_U::Initialize, "Initialize"}, {5, nullptr, "Finalize"},
+        {6, &MM_U::SetAndWait, "SetAndWait"}, {7, &MM_U::Get, "Get"},
+    };
+    RegisterHandlers(functions);
+}
+
+} // namespace Service::MM

src/core/hle/service/mm/mm_u.h

@@ -0,0 +1,29 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/service.h"
+
+namespace Service::MM {
+
+class MM_U final : public ServiceFramework<MM_U> {
+public:
+    MM_U();
+    ~MM_U() = default;
+
+private:
+    void Initialize(Kernel::HLERequestContext& ctx);
+    void SetAndWait(Kernel::HLERequestContext& ctx);
+    void Get(Kernel::HLERequestContext& ctx);
+
+    u32 min{0};
+    u32 max{0};
+    u32 current{0};
+};
+
+/// Registers all MM services with the specified service manager.
+void InstallInterfaces(SM::ServiceManager& service_manager);
+
+} // namespace Service::MM

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

@@ -16,7 +16,11 @@ u32 nvhost_ctrl::ioctl(Ioctl command, const std::vector<u8>& input, std::vector<
     case IoctlCommand::IocGetConfigCommand:
         return NvOsGetConfigU32(input, output);
     case IoctlCommand::IocCtrlEventWaitCommand:
-        return IocCtrlEventWait(input, output);
+        return IocCtrlEventWait(input, output, false);
+    case IoctlCommand::IocCtrlEventWaitAsyncCommand:
+        return IocCtrlEventWait(input, output, true);
+    case IoctlCommand::IocCtrlEventRegisterCommand:
+        return IocCtrlEventRegister(input, output);
     }
     UNIMPLEMENTED_MSG("Unimplemented ioctl");
     return 0;
@@ -45,11 +49,13 @@ u32 nvhost_ctrl::NvOsGetConfigU32(const std::vector<u8>& input, std::vector<u8>&
     return 0;
 }
 
-u32 nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& output) {
+u32 nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& output,
+                                  bool is_async) {
     IocCtrlEventWaitParams params{};
     std::memcpy(&params, input.data(), sizeof(params));
-    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called, syncpt_id={} threshold={} timeout={}",
-                  params.syncpt_id, params.threshold, params.timeout);
+    NGLOG_WARNING(Service_NVDRV,
+                  "(STUBBED) called, syncpt_id={}, threshold={}, timeout={}, is_async={}",
+                  params.syncpt_id, params.threshold, params.timeout, is_async);
 
     // TODO(Subv): Implement actual syncpt waiting.
     params.value = 0;
@@ -57,4 +63,10 @@ u32 nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>&
     return 0;
 }
 
+u32 nvhost_ctrl::IocCtrlEventRegister(const std::vector<u8>& input, std::vector<u8>& output) {
+    NGLOG_WARNING(Service_NVDRV, "(STUBBED) called");
+    // TODO(bunnei): Implement this.
+    return 0;
+}
+
 } // namespace Service::Nvidia::Devices

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

@@ -134,7 +134,9 @@ private:
 
     u32 NvOsGetConfigU32(const std::vector<u8>& input, std::vector<u8>& output);
 
-    u32 IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& output, bool is_async);
+
+    u32 IocCtrlEventRegister(const std::vector<u8>& input, std::vector<u8>& output);
 };
 
 } // namespace Service::Nvidia::Devices

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

@@ -26,11 +26,19 @@ public:
 private:
     enum class IoctlCommand : u32_le {
         IocSetNVMAPfdCommand = 0x40044801,
+        IocAllocGPFIFOCommand = 0x40084805,
         IocSetClientDataCommand = 0x40084714,
         IocGetClientDataCommand = 0x80084715,
         IocZCullBind = 0xc010480b,
         IocSetErrorNotifierCommand = 0xC018480C,
         IocChannelSetPriorityCommand = 0x4004480D,
+        IocEnableCommand = 0x0000480E,
+        IocDisableCommand = 0x0000480F,
+        IocPreemptCommand = 0x00004810,
+        IocForceResetCommand = 0x00004811,
+        IocEventIdControlCommand = 0x40084812,
+        IocGetErrorNotificationCommand = 0xC0104817,
+        IocAllocGPFIFOExCommand = 0x40204818,
         IocAllocGPFIFOEx2Command = 0xC020481A,
         IocAllocObjCtxCommand = 0xC0104809,
         IocChannelGetWaitbaseCommand = 0xC0080003,
@@ -56,6 +64,12 @@ private:
     };
     static_assert(sizeof(IoctlChannelSetTimeout) == 4, "IoctlChannelSetTimeout is incorrect size");
 
+    struct IoctlAllocGPFIFO {
+        u32_le num_entries;
+        u32_le flags;
+    };
+    static_assert(sizeof(IoctlAllocGPFIFO) == 8, "IoctlAllocGPFIFO is incorrect size");
+
     struct IoctlClientData {
         u64_le data;
     };
@@ -76,12 +90,45 @@ private:
     };
     static_assert(sizeof(IoctlSetErrorNotifier) == 24, "IoctlSetErrorNotifier is incorrect size");
 
+    struct IoctlChannelSetPriority {
+        u32_le priority;
+    };
+    static_assert(sizeof(IoctlChannelSetPriority) == 4,
+                  "IoctlChannelSetPriority is incorrect size");
+
+    struct IoctlEventIdControl {
+        u32_le cmd; // 0=disable, 1=enable, 2=clear
+        u32_le id;
+    };
+    static_assert(sizeof(IoctlEventIdControl) == 8, "IoctlEventIdControl is incorrect size");
+
+    struct IoctlGetErrorNotification {
+        u64_le timestamp;
+        u32_le info32;
+        u16_le info16;
+        u16_le status; // always 0xFFFF
+    };
+    static_assert(sizeof(IoctlGetErrorNotification) == 16,
+                  "IoctlGetErrorNotification is incorrect size");
+
     struct IoctlFence {
         u32_le id;
         u32_le value;
     };
     static_assert(sizeof(IoctlFence) == 8, "IoctlFence is incorrect size");
 
+    struct IoctlAllocGpfifoEx {
+        u32_le num_entries;
+        u32_le flags;
+        u32_le unk0;
+        u32_le unk1;
+        u32_le unk2;
+        u32_le unk3;
+        u32_le unk4;
+        u32_le unk5;
+    };
+    static_assert(sizeof(IoctlAllocGpfifoEx) == 32, "IoctlAllocGpfifoEx is incorrect size");
+
     struct IoctlAllocGpfifoEx2 {
         u32_le num_entries;   // in
         u32_le flags;         // in

src/core/hle/service/service.cpp

@@ -26,6 +26,7 @@
 #include "core/hle/service/friend/friend.h"
 #include "core/hle/service/hid/hid.h"
 #include "core/hle/service/lm/lm.h"
+#include "core/hle/service/mm/mm_u.h"
 #include "core/hle/service/nfp/nfp.h"
 #include "core/hle/service/nifm/nifm.h"
 #include "core/hle/service/ns/ns.h"
@@ -191,6 +192,7 @@ void Init(std::shared_ptr<SM::ServiceManager>& sm) {
     Friend::InstallInterfaces(*sm);
     HID::InstallInterfaces(*sm);
     LM::InstallInterfaces(*sm);
+    MM::InstallInterfaces(*sm);
     NFP::InstallInterfaces(*sm);
     NIFM::InstallInterfaces(*sm);
     NS::InstallInterfaces(*sm);

src/core/hle/service/time/time.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <chrono>
+#include <ctime>
 #include "common/logging/log.h"
 #include "core/core_timing.h"
 #include "core/hle/ipc_helpers.h"
@@ -77,7 +78,7 @@ public:
             {3, nullptr, "LoadLocationNameList"},
             {4, &ITimeZoneService::LoadTimeZoneRule, "LoadTimeZoneRule"},
             {5, nullptr, "GetTimeZoneRuleVersion"},
-            {100, nullptr, "ToCalendarTime"},
+            {100, &ITimeZoneService::ToCalendarTime, "ToCalendarTime"},
             {101, &ITimeZoneService::ToCalendarTimeWithMyRule, "ToCalendarTimeWithMyRule"},
             {200, nullptr, "ToPosixTime"},
             {201, nullptr, "ToPosixTimeWithMyRule"},
@@ -86,9 +87,11 @@ public:
     }
 
 private:
+    LocationName location_name{"UTC"};
+    TimeZoneRule my_time_zone_rule{};
+
     void GetDeviceLocationName(Kernel::HLERequestContext& ctx) {
-        NGLOG_WARNING(Service_Time, "(STUBBED) called");
-        LocationName location_name{};
+        NGLOG_DEBUG(Service_Time, "called");
         IPC::ResponseBuilder rb{ctx, (sizeof(LocationName) / 4) + 2};
         rb.Push(RESULT_SUCCESS);
         rb.PushRaw(location_name);
@@ -103,23 +106,70 @@ private:
 
     void LoadTimeZoneRule(Kernel::HLERequestContext& ctx) {
         NGLOG_WARNING(Service_Time, "(STUBBED) called");
+
+        ctx.WriteBuffer(&my_time_zone_rule, sizeof(TimeZoneRule));
+
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
 
+    void ToCalendarTime(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        const u64 posix_time = rp.Pop<u64>();
+
+        NGLOG_WARNING(Service_Time, "(STUBBED) called, posix_time=0x{:016X}", posix_time);
+
+        TimeZoneRule time_zone_rule{};
+        auto buffer = ctx.ReadBuffer();
+        std::memcpy(&time_zone_rule, buffer.data(), buffer.size());
+
+        CalendarTime calendar_time{2018, 1, 1, 0, 0, 0};
+        CalendarAdditionalInfo additional_info{};
+
+        PosixToCalendar(posix_time, calendar_time, additional_info, time_zone_rule);
+
+        IPC::ResponseBuilder rb{ctx, 10};
+        rb.Push(RESULT_SUCCESS);
+        rb.PushRaw(calendar_time);
+        rb.PushRaw(additional_info);
+    }
+
     void ToCalendarTimeWithMyRule(Kernel::HLERequestContext& ctx) {
         IPC::RequestParser rp{ctx};
-        u64 posix_time = rp.Pop<u64>();
+        const u64 posix_time = rp.Pop<u64>();
 
         NGLOG_WARNING(Service_Time, "(STUBBED) called, posix_time=0x{:016X}", posix_time);
 
         CalendarTime calendar_time{2018, 1, 1, 0, 0, 0};
         CalendarAdditionalInfo additional_info{};
+
+        PosixToCalendar(posix_time, calendar_time, additional_info, my_time_zone_rule);
+
         IPC::ResponseBuilder rb{ctx, 10};
         rb.Push(RESULT_SUCCESS);
         rb.PushRaw(calendar_time);
         rb.PushRaw(additional_info);
     }
+
+    void PosixToCalendar(u64 posix_time, CalendarTime& calendar_time,
+                         CalendarAdditionalInfo& additional_info, const TimeZoneRule& /*rule*/) {
+        std::time_t t(posix_time);
+        std::tm* tm = std::localtime(&t);
+        if (!tm) {
+            return;
+        }
+        calendar_time.year = tm->tm_year + 1900;
+        calendar_time.month = tm->tm_mon + 1;
+        calendar_time.day = tm->tm_mday;
+        calendar_time.hour = tm->tm_hour;
+        calendar_time.minute = tm->tm_min;
+        calendar_time.second = tm->tm_sec;
+
+        additional_info.day_of_week = tm->tm_wday;
+        additional_info.day_of_year = tm->tm_yday;
+        std::memcpy(additional_info.name.data(), "UTC", sizeof("UTC"));
+        additional_info.utc_offset = 0;
+    }
 };
 
 void Module::Interface::GetStandardUserSystemClock(Kernel::HLERequestContext& ctx) {

src/core/hle/service/time/time.h

@@ -4,13 +4,13 @@
 
 #pragma once
 
+#include <array>
 #include "core/hle/service/service.h"
 
 namespace Service::Time {
 
-// TODO(Rozelette) RE this structure
 struct LocationName {
-    INSERT_PADDING_BYTES(0x24);
+    std::array<u8, 0x24> name;
 };
 static_assert(sizeof(LocationName) == 0x24, "LocationName is incorrect size");
 
@@ -25,26 +25,34 @@ struct CalendarTime {
 };
 static_assert(sizeof(CalendarTime) == 0x8, "CalendarTime structure has incorrect size");
 
-// TODO(Rozelette) RE this structure
 struct CalendarAdditionalInfo {
-    INSERT_PADDING_BYTES(0x18);
+    u32_le day_of_week;
+    u32_le day_of_year;
+    std::array<u8, 8> name;
+    INSERT_PADDING_BYTES(1);
+    s32_le utc_offset;
 };
 static_assert(sizeof(CalendarAdditionalInfo) == 0x18,
               "CalendarAdditionalInfo structure has incorrect size");
 
-// TODO(bunnei) RE this structure
-struct SystemClockContext {
-    INSERT_PADDING_BYTES(0x20);
+// TODO(mailwl) RE this structure
+struct TimeZoneRule {
+    INSERT_PADDING_BYTES(0x4000);
 };
-static_assert(sizeof(SystemClockContext) == 0x20,
-              "SystemClockContext structure has incorrect size");
 
 struct SteadyClockTimePoint {
-    u64 value;
+    u64_le value;
     INSERT_PADDING_WORDS(4);
 };
 static_assert(sizeof(SteadyClockTimePoint) == 0x18, "SteadyClockTimePoint is incorrect size");
 
+struct SystemClockContext {
+    u64_le offset;
+    SteadyClockTimePoint time_point;
+};
+static_assert(sizeof(SystemClockContext) == 0x20,
+              "SystemClockContext structure has incorrect size");
+
 class Module final {
 public:
     class Interface : public ServiceFramework<Interface> {

src/video_core/engines/maxwell_3d.cpp

@@ -156,16 +156,15 @@ void Maxwell3D::ProcessQueryGet() {
     // TODO(Subv): Support the other query units.
     ASSERT_MSG(regs.query.query_get.unit == Regs::QueryUnit::Crop,
                "Units other than CROP are unimplemented");
-    ASSERT_MSG(regs.query.query_get.short_query,
-               "Writing the entire query result structure is unimplemented");
 
     u32 value = Memory::Read32(*address);
-    u32 result = 0;
+    u64 result = 0;
 
     // TODO(Subv): Support the other query variables
     switch (regs.query.query_get.select) {
     case Regs::QuerySelect::Zero:
-        result = 0;
+        // This seems to actually write the query sequence to the query address.
+        result = regs.query.query_sequence;
         break;
     default:
         UNIMPLEMENTED_MSG("Unimplemented query select type {}",
@@ -174,15 +173,31 @@ void Maxwell3D::ProcessQueryGet() {
 
     // TODO(Subv): Research and implement how query sync conditions work.
 
+    struct LongQueryResult {
+        u64_le value;
+        u64_le timestamp;
+    };
+    static_assert(sizeof(LongQueryResult) == 16, "LongQueryResult has wrong size");
+
     switch (regs.query.query_get.mode) {
     case Regs::QueryMode::Write:
     case Regs::QueryMode::Write2: {
-        // Write the current query sequence to the sequence address.
         u32 sequence = regs.query.query_sequence;
-        Memory::Write32(*address, sequence);
-
-        // TODO(Subv): Write the proper query response structure to the address when not using short
-        // mode.
+        if (regs.query.query_get.short_query) {
+            // Write the current query sequence to the sequence address.
+            // TODO(Subv): Find out what happens if you use a long query type but mark it as a short
+            // query.
+            Memory::Write32(*address, sequence);
+        } else {
+            // Write the 128-bit result structure in long mode. Note: We emulate an infinitely fast
+            // GPU, this command may actually take a while to complete in real hardware due to GPU
+            // wait queues.
+            LongQueryResult query_result{};
+            query_result.value = result;
+            // TODO(Subv): Generate a real GPU timestamp and write it here instead of 0
+            query_result.timestamp = 0;
+            Memory::WriteBlock(*address, &query_result, sizeof(query_result));
+        }
         break;
     }
     default:

src/video_core/engines/maxwell_3d.h

@@ -354,10 +354,35 @@ public:
                     f32 scale_x;
                     f32 scale_y;
                     f32 scale_z;
-                    u32 translate_x;
-                    u32 translate_y;
-                    u32 translate_z;
+                    f32 translate_x;
+                    f32 translate_y;
+                    f32 translate_z;
                     INSERT_PADDING_WORDS(2);
+
+                    MathUtil::Rectangle<s32> GetRect() const {
+                        return {
+                            GetX(),               // left
+                            GetY() + GetHeight(), // top
+                            GetX() + GetWidth(),  // right
+                            GetY()                // bottom
+                        };
+                    };
+
+                    s32 GetX() const {
+                        return static_cast<s32>(std::max(0.0f, translate_x - std::fabs(scale_x)));
+                    }
+
+                    s32 GetY() const {
+                        return static_cast<s32>(std::max(0.0f, translate_y - std::fabs(scale_y)));
+                    }
+
+                    s32 GetWidth() const {
+                        return static_cast<s32>(translate_x + std::fabs(scale_x)) - GetX();
+                    }
+
+                    s32 GetHeight() const {
+                        return static_cast<s32>(translate_y + std::fabs(scale_y)) - GetY();
+                    }
                 } viewport_transform[NumViewports];
 
                 struct {
@@ -371,15 +396,6 @@ public:
                     };
                     float depth_range_near;
                     float depth_range_far;
-
-                    MathUtil::Rectangle<s32> GetRect() const {
-                        return {
-                            static_cast<s32>(x),          // left
-                            static_cast<s32>(y + height), // top
-                            static_cast<s32>(x + width),  // right
-                            static_cast<s32>(y)           // bottom
-                        };
-                    };
                 } viewport[NumViewports];
 
                 INSERT_PADDING_WORDS(0x1D);

src/video_core/engines/shader_bytecode.h

@@ -75,6 +75,10 @@ union Attribute {
     enum class Index : u64 {
         Position = 7,
         Attribute_0 = 8,
+        // This attribute contains a tuple of (~, ~, InstanceId, VertexId) when inside a vertex
+        // shader, and a tuple of (TessCoord.x, TessCoord.y, TessCoord.z, ~) when inside a Tess Eval
+        // shader.
+        TessCoordInstanceIDVertexID = 47,
     };
 
     union {
@@ -152,6 +156,13 @@ enum class PredOperation : u64 {
     Xor = 2,
 };
 
+enum class LogicOperation : u64 {
+    And = 0,
+    Or = 1,
+    Xor = 2,
+    PassB = 3,
+};
+
 enum class SubOp : u64 {
     Cos = 0x0,
     Sin = 0x1,
@@ -198,6 +209,12 @@ union Instruction {
             BitField<42, 1, u64> negate_pred;
         } fmnmx;
 
+        union {
+            BitField<53, 2, LogicOperation> operation;
+            BitField<55, 1, u64> invert_a;
+            BitField<56, 1, u64> invert_b;
+        } lop;
+
         float GetImm20_19() const {
             float result{};
             u32 imm{static_cast<u32>(imm20_19)};
@@ -215,6 +232,22 @@ union Instruction {
         }
     } 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));
+            // Sign extend the 20-bit value.
+            u32 mask = 1U << (20 - 1);
+            return static_cast<s32>((immediate ^ mask) - mask);
+        }
+
+    } iscadd;
+
     union {
         BitField<48, 1, u64> negate_b;
         BitField<49, 1, u64> negate_c;
@@ -234,6 +267,16 @@ union Instruction {
         BitField<56, 1, u64> neg_b;
     } fsetp;
 
+    union {
+        BitField<0, 3, u64> pred0;
+        BitField<3, 3, u64> pred3;
+        BitField<39, 3, u64> pred39;
+        BitField<42, 1, u64> neg_pred;
+        BitField<45, 2, PredOperation> op;
+        BitField<48, 1, u64> is_signed;
+        BitField<49, 3, PredCondition> cond;
+    } isetp;
+
     union {
         BitField<39, 3, u64> pred39;
         BitField<42, 1, u64> neg_pred;
@@ -241,9 +284,9 @@ union Instruction {
         BitField<44, 1, u64> abs_b;
         BitField<45, 2, PredOperation> op;
         BitField<48, 4, PredCondition> cond;
+        BitField<52, 1, u64> bf;
         BitField<53, 1, u64> neg_b;
         BitField<54, 1, u64> abs_a;
-        BitField<52, 1, u64> bf;
         BitField<55, 1, u64> ftz;
         BitField<56, 1, u64> neg_imm;
     } fset;
@@ -257,6 +300,46 @@ union Instruction {
         BitField<50, 1, u64> saturate_a;
     } conversion;
 
+    union {
+        BitField<31, 4, u64> component_mask;
+
+        bool IsComponentEnabled(size_t component) const {
+            return ((1 << component) & component_mask) != 0;
+        }
+    } tex;
+
+    union {
+        BitField<50, 3, u64> component_mask_selector;
+        BitField<28, 8, Register> gpr28;
+
+        bool HasTwoDestinations() const {
+            return gpr28.Value() != Register::ZeroIndex;
+        }
+
+        bool IsComponentEnabled(size_t component) const {
+            static constexpr std::array<size_t, 5> one_dest_mask{0x1, 0x2, 0x4, 0x8, 0x3};
+            static constexpr std::array<size_t, 5> two_dest_mask{0x7, 0xb, 0xd, 0xe, 0xf};
+            const auto& mask{HasTwoDestinations() ? two_dest_mask : one_dest_mask};
+
+            ASSERT(component_mask_selector < mask.size());
+
+            return ((1 << component) & mask[component_mask_selector]) != 0;
+        }
+    } texs;
+
+    union {
+        BitField<20, 5, u64> target;
+        BitField<5, 1, u64> constant_buffer;
+
+        s32 GetBranchTarget() const {
+            // Sign extend the branch target offset
+            u32 mask = 1U << (5 - 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;
+        }
+    } bra;
+
     BitField<61, 1, u64> is_b_imm;
     BitField<60, 1, u64> is_b_gpr;
     BitField<59, 1, u64> is_c_gpr;
@@ -275,8 +358,10 @@ class OpCode {
 public:
     enum class Id {
         KIL,
+        BRA,
         LD_A,
         ST_A,
+        TEX,
         TEXQ, // Texture Query
         TEXS, // Texture Fetch with scalar/non-vec4 source/destinations
         TLDS, // Texture Load with scalar/non-vec4 source/destinations
@@ -293,8 +378,13 @@ public:
         FMUL_R,
         FMUL_IMM,
         FMUL32_IMM,
-        MUFU, // Multi-Function Operator
-        RRO,  // Range Reduction Operator
+        ISCADD_C, // Scale and Add
+        ISCADD_R,
+        ISCADD_IMM,
+        MUFU,  // Multi-Function Operator
+        RRO_C, // Range Reduction Operator
+        RRO_R,
+        RRO_IMM,
         F2F_C,
         F2F_R,
         F2F_IMM,
@@ -333,6 +423,8 @@ public:
     enum class Type {
         Trivial,
         Arithmetic,
+        Logic,
+        ScaledAdd,
         Ffma,
         Flow,
         Memory,
@@ -436,8 +528,10 @@ 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("111000100100----", Id::BRA, Type::Flow, "BRA"),
             INST("1110111111011---", Id::LD_A, Type::Memory, "LD_A"),
             INST("1110111111110---", Id::ST_A, Type::Memory, "ST_A"),
+            INST("1100000000111---", Id::TEX, Type::Memory, "TEX"),
             INST("1101111101001---", Id::TEXQ, Type::Memory, "TEXQ"),
             INST("1101100---------", Id::TEXS, Type::Memory, "TEXS"),
             INST("1101101---------", Id::TLDS, Type::Memory, "TLDS"),
@@ -454,15 +548,19 @@ private:
             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("0101000010000---", Id::MUFU, Type::Arithmetic, "MUFU"),
-            INST("0101110010010---", Id::RRO, Type::Arithmetic, "RRO"),
-            INST("0100110010101---", Id::F2F_C, Type::Arithmetic, "F2F_C"),
-            INST("0101110010101---", Id::F2F_R, Type::Arithmetic, "F2F_R"),
-            INST("0011100-10101---", Id::F2F_IMM, Type::Arithmetic, "F2F_IMM"),
+            INST("0100110010010---", Id::RRO_C, Type::Arithmetic, "RRO_C"),
+            INST("0101110010010---", Id::RRO_R, Type::Arithmetic, "RRO_R"),
+            INST("0011100-10010---", Id::RRO_IMM, Type::Arithmetic, "RRO_IMM"),
+            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("000001----------", Id::LOP32I, Type::Arithmetic, "LOP32I"),
             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"),
@@ -473,6 +571,7 @@ private:
             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("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"),

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -298,7 +298,7 @@ void RasterizerOpenGL::DrawArrays() {
     const bool has_stencil = false;
     const bool using_color_fb = true;
     const bool using_depth_fb = false;
-    const MathUtil::Rectangle<s32> viewport_rect{regs.viewport[0].GetRect()};
+    const MathUtil::Rectangle<s32> viewport_rect{regs.viewport_transform[0].GetRect()};
 
     const bool write_color_fb =
         state.color_mask.red_enabled == GL_TRUE || state.color_mask.green_enabled == GL_TRUE ||
@@ -702,7 +702,7 @@ void RasterizerOpenGL::BindFramebufferSurfaces(const Surface& color_surface,
 
 void RasterizerOpenGL::SyncViewport(const MathUtil::Rectangle<u32>& surfaces_rect, u16 res_scale) {
     const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
-    const MathUtil::Rectangle<s32> viewport_rect{regs.viewport[0].GetRect()};
+    const MathUtil::Rectangle<s32> viewport_rect{regs.viewport_transform[0].GetRect()};
 
     state.viewport.x = static_cast<GLint>(surfaces_rect.left) + viewport_rect.left * res_scale;
     state.viewport.y = static_cast<GLint>(surfaces_rect.bottom) + viewport_rect.bottom * res_scale;

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -45,21 +45,23 @@ struct FormatTuple {
 
 static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_format_tuples = {{
     {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, false},                    // ABGR8
-    {GL_RGB, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV, false},                       // B5G6R5
+    {GL_RGB, GL_RGB, GL_UNSIGNED_SHORT_5_6_5, false},                           // B5G6R5
     {GL_RGB10_A2, GL_RGBA, GL_UNSIGNED_INT_2_10_10_10_REV, false},              // A2B10G10R10
     {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_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
 }};
 
 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
-        ASSERT(component_type == ComponentType::UNorm);
+        // For now only UNORM components are supported, or RGBA16F which is type FLOAT
+        ASSERT(component_type == ComponentType::UNorm || pixel_format == PixelFormat::RGBA16F);
         return tex_format_tuples[static_cast<unsigned int>(pixel_format)];
     } else if (type == SurfaceType::Depth || type == SurfaceType::DepthStencil) {
         // TODO(Subv): Implement depth formats
@@ -110,8 +112,9 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr, Tegra:
     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::DXT1>,
-        MortonCopy<true, PixelFormat::DXT23>,       MortonCopy<true, PixelFormat::DXT45>,
+        MortonCopy<true, PixelFormat::R8>,          MortonCopy<true, PixelFormat::RGBA16F>,
+        MortonCopy<true, PixelFormat::DXT1>,        MortonCopy<true, PixelFormat::DXT23>,
+        MortonCopy<true, PixelFormat::DXT45>,       MortonCopy<true, PixelFormat::DXN1>,
 };
 
 static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr, Tegra::GPUVAddr,
@@ -123,7 +126,9 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr, Tegra:
         MortonCopy<false, PixelFormat::A2B10G10R10>,
         MortonCopy<false, PixelFormat::A1B5G5R5>,
         MortonCopy<false, PixelFormat::R8>,
-        // TODO(Subv): Swizzling the DXT1/DXT23/DXT45 formats is not yet supported
+        MortonCopy<false, PixelFormat::RGBA16F>,
+        // TODO(Subv): Swizzling the DXT1/DXT23/DXT45/DXN1 formats is not yet supported
+        nullptr,
         nullptr,
         nullptr,
         nullptr,
@@ -928,7 +933,8 @@ Surface RasterizerCacheOpenGL::GetSurface(const SurfaceParams& params, ScaleMatc
     // Use GetSurfaceSubRect instead
     ASSERT(params.width == params.stride);
 
-    ASSERT(!params.is_tiled || (params.width % 8 == 0 && params.height % 8 == 0));
+    ASSERT(!params.is_tiled ||
+           (params.GetActualWidth() % 8 == 0 && params.GetActualHeight() % 8 == 0));
 
     // Check for an exact match in existing surfaces
     Surface surface =

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -59,9 +59,11 @@ struct SurfaceParams {
         A2B10G10R10 = 2,
         A1B5G5R5 = 3,
         R8 = 4,
-        DXT1 = 5,
-        DXT23 = 6,
-        DXT45 = 7,
+        RGBA16F = 5,
+        DXT1 = 6,
+        DXT23 = 7,
+        DXT45 = 8,
+        DXN1 = 9, // This is also known as BC4
 
         Max,
         Invalid = 255,
@@ -102,9 +104,11 @@ struct SurfaceParams {
             1, // A2B10G10R10
             1, // A1B5G5R5
             1, // R8
+            2, // RGBA16F
             4, // DXT1
             4, // DXT23
             4, // DXT45
+            4, // DXN1
         }};
 
         ASSERT(static_cast<size_t>(format) < compression_factor_table.size());
@@ -124,9 +128,11 @@ struct SurfaceParams {
             32,  // A2B10G10R10
             16,  // A1B5G5R5
             8,   // R8
+            64,  // RGBA16F
             64,  // DXT1
             128, // DXT23
             128, // DXT45
+            64,  // DXN1
         }};
 
         ASSERT(static_cast<size_t>(format) < bpp_table.size());
@@ -143,6 +149,8 @@ struct SurfaceParams {
             return PixelFormat::ABGR8;
         case Tegra::RenderTargetFormat::RGB10_A2_UNORM:
             return PixelFormat::A2B10G10R10;
+        case Tegra::RenderTargetFormat::RGBA16_FLOAT:
+            return PixelFormat::RGBA16F;
         default:
             NGLOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
@@ -172,12 +180,16 @@ struct SurfaceParams {
             return PixelFormat::A1B5G5R5;
         case Tegra::Texture::TextureFormat::R8:
             return PixelFormat::R8;
+        case Tegra::Texture::TextureFormat::R16_G16_B16_A16:
+            return PixelFormat::RGBA16F;
         case Tegra::Texture::TextureFormat::DXT1:
             return PixelFormat::DXT1;
         case Tegra::Texture::TextureFormat::DXT23:
             return PixelFormat::DXT23;
         case Tegra::Texture::TextureFormat::DXT45:
             return PixelFormat::DXT45;
+        case Tegra::Texture::TextureFormat::DXN1:
+            return PixelFormat::DXN1;
         default:
             NGLOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
@@ -197,12 +209,16 @@ struct SurfaceParams {
             return Tegra::Texture::TextureFormat::A1B5G5R5;
         case PixelFormat::R8:
             return Tegra::Texture::TextureFormat::R8;
+        case PixelFormat::RGBA16F:
+            return Tegra::Texture::TextureFormat::R16_G16_B16_A16;
         case PixelFormat::DXT1:
             return Tegra::Texture::TextureFormat::DXT1;
         case PixelFormat::DXT23:
             return Tegra::Texture::TextureFormat::DXT23;
         case PixelFormat::DXT45:
             return Tegra::Texture::TextureFormat::DXT45;
+        case PixelFormat::DXN1:
+            return Tegra::Texture::TextureFormat::DXN1;
         default:
             UNREACHABLE();
         }
@@ -226,6 +242,8 @@ struct SurfaceParams {
         case Tegra::RenderTargetFormat::RGBA8_SRGB:
         case Tegra::RenderTargetFormat::RGB10_A2_UNORM:
             return ComponentType::UNorm;
+        case Tegra::RenderTargetFormat::RGBA16_FLOAT:
+            return ComponentType::Float;
         default:
             NGLOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -88,6 +88,20 @@ private:
         return *subroutines.insert(std::move(subroutine)).first;
     }
 
+    /// Merges exit method of two parallel branches.
+    static ExitMethod ParallelExit(ExitMethod a, ExitMethod b) {
+        if (a == ExitMethod::Undetermined) {
+            return b;
+        }
+        if (b == ExitMethod::Undetermined) {
+            return a;
+        }
+        if (a == b) {
+            return a;
+        }
+        return ExitMethod::Conditional;
+    }
+
     /// Scans a range of code for labels and determines the exit method.
     ExitMethod Scan(u32 begin, u32 end, std::set<u32>& labels) {
         auto [iter, inserted] =
@@ -97,10 +111,27 @@ private:
             return exit_method;
 
         for (u32 offset = begin; offset != end && offset != PROGRAM_END; ++offset) {
-            if (const auto opcode = OpCode::Decode({program_code[offset]})) {
+            const Instruction instr = {program_code[offset]};
+            if (const auto opcode = OpCode::Decode(instr)) {
                 switch (opcode->GetId()) {
                 case OpCode::Id::EXIT: {
-                    return exit_method = ExitMethod::AlwaysEnd;
+                    // The EXIT instruction can be predicated, which means that the shader can
+                    // conditionally end on this instruction. We have to consider the case where the
+                    // condition is not met and check the exit method of that other basic block.
+                    using Tegra::Shader::Pred;
+                    if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
+                        return exit_method = ExitMethod::AlwaysEnd;
+                    } else {
+                        ExitMethod not_met = Scan(offset + 1, end, labels);
+                        return exit_method = ParallelExit(ExitMethod::AlwaysEnd, not_met);
+                    }
+                }
+                case OpCode::Id::BRA: {
+                    u32 target = offset + instr.bra.GetBranchTarget();
+                    labels.insert(target);
+                    ExitMethod no_jmp = Scan(offset + 1, end, labels);
+                    ExitMethod jmp = Scan(target, end, labels);
+                    return exit_method = ParallelExit(no_jmp, jmp);
                 }
                 }
             }
@@ -197,6 +228,11 @@ public:
         return active_type == Type::Integer;
     }
 
+    /// Returns the current active type of the register
+    Type GetActiveType() const {
+        return active_type;
+    }
+
     /// Returns the index of the register
     size_t GetIndex() const {
         return index;
@@ -299,7 +335,7 @@ public:
      * are stored as floats, so this may require conversion.
      * @param reg The destination register to use.
      * @param elem The element to use for the operation.
-     * @param attribute The input attibute to use as the source value.
+     * @param attribute The input attribute to use as the source value.
      */
     void SetRegisterToInputAttibute(const Register& reg, u64 elem, Attribute::Index attribute) {
         std::string dest = GetRegisterAsFloat(reg);
@@ -328,22 +364,28 @@ public:
         shader.AddLine(dest + " = " + src + ';');
     }
 
-    /// Generates code representing a uniform (C buffer) register.
-    std::string GetUniform(const Uniform& uniform, const Register& dest_reg) {
+    /// 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) + ']';
 
-        if (regs[dest_reg].IsFloat()) {
+        if (type == GLSLRegister::Type::Float) {
             return value;
-        } else if (regs[dest_reg].IsInteger()) {
+        } else if (type == GLSLRegister::Type::Integer) {
             return "floatBitsToInt(" + value + ')';
         } else {
             UNREACHABLE();
         }
     }
 
+    /// 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());
+    }
+
     /// Add declarations for registers
     void GenerateDeclarations() {
         for (const auto& reg : regs) {
@@ -451,6 +493,12 @@ private:
         switch (attribute) {
         case Attribute::Index::Position:
             return "position";
+        case Attribute::Index::TessCoordInstanceIDVertexID:
+            // TODO(Subv): Find out what the values are for the first two elements when inside a
+            // 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)";
         default:
             const u32 index{static_cast<u32>(attribute) -
                             static_cast<u32>(Attribute::Index::Attribute_0)};
@@ -606,9 +654,9 @@ private:
     std::string GetPredicateComparison(Tegra::Shader::PredCondition condition) const {
         using Tegra::Shader::PredCondition;
         static const std::unordered_map<PredCondition, const char*> PredicateComparisonStrings = {
-            {PredCondition::LessThan, "<"},      {PredCondition::Equal, "=="},
-            {PredCondition::LessEqual, "<="},    {PredCondition::GreaterThan, ">"},
-            {PredCondition::GreaterEqual, ">="},
+            {PredCondition::LessThan, "<"},   {PredCondition::Equal, "=="},
+            {PredCondition::LessEqual, "<="}, {PredCondition::GreaterThan, ">"},
+            {PredCondition::NotEqual, "!="},  {PredCondition::GreaterEqual, ">="},
         };
 
         auto comparison = PredicateComparisonStrings.find(condition);
@@ -786,8 +834,13 @@ private:
                                         1, 1);
                 break;
             }
-            case OpCode::Id::RRO: {
-                NGLOG_DEBUG(HW_GPU, "Skipping RRO instruction");
+            case OpCode::Id::RRO_C:
+            case OpCode::Id::RRO_R:
+            case OpCode::Id::RRO_IMM: {
+                // Currently RRO is only implemented as a register move.
+                // Usage of `abs_b` and `negate_b` here should also be correct.
+                regs.SetRegisterToFloat(instr.gpr0, 0, op_b, 1, 1);
+                NGLOG_WARNING(HW_GPU, "RRO instruction is incomplete");
                 break;
             }
             default: {
@@ -797,6 +850,73 @@ private:
             }
             break;
         }
+        case OpCode::Type::Logic: {
+            std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, false);
+
+            if (instr.alu.lop.invert_a)
+                op_a = "~(" + op_a + ')';
+
+            switch (opcode->GetId()) {
+            case OpCode::Id::LOP32I: {
+                u32 imm = static_cast<u32>(instr.alu.imm20_32.Value());
+
+                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();
+                }
+                break;
+            }
+            default: {
+                NGLOG_CRITICAL(HW_GPU, "Unhandled logic instruction: {}", opcode->GetName());
+                UNREACHABLE();
+            }
+            }
+            break;
+        }
+        case OpCode::Type::ScaledAdd: {
+            std::string op_a = regs.GetRegisterAsInteger(instr.gpr8);
+
+            if (instr.iscadd.negate_a)
+                op_a = '-' + op_a;
+
+            std::string op_b = instr.iscadd.negate_b ? "-" : "";
+
+            if (instr.is_b_imm) {
+                op_b += '(' + std::to_string(instr.iscadd.GetImmediate()) + ')';
+            } else {
+                if (instr.is_b_gpr) {
+                    op_b += regs.GetRegisterAsInteger(instr.gpr20);
+                } else {
+                    op_b += regs.GetUniform(instr.uniform, instr.gpr0);
+                }
+            }
+
+            std::string shift = std::to_string(instr.iscadd.shift_amount.Value());
+
+            regs.SetRegisterToInteger(instr.gpr0, true, 0,
+                                      "((" + op_a + " << " + shift + ") + " + op_b + ')', 1, 1);
+            break;
+        }
         case OpCode::Type::Ffma: {
             std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
             std::string op_b = instr.ffma.negate_b ? "-" : "";
@@ -834,13 +954,13 @@ private:
         }
         case OpCode::Type::Conversion: {
             ASSERT_MSG(instr.conversion.size == Register::Size::Word, "Unimplemented");
-            ASSERT_MSG(!instr.conversion.selector, "Unimplemented");
             ASSERT_MSG(!instr.conversion.negate_a, "Unimplemented");
             ASSERT_MSG(!instr.conversion.saturate_a, "Unimplemented");
 
             switch (opcode->GetId()) {
-            case OpCode::Id::I2I_R:
-            case OpCode::Id::I2F_R: {
+            case OpCode::Id::I2I_R: {
+                ASSERT_MSG(!instr.conversion.selector, "Unimplemented");
+
                 std::string op_a =
                     regs.GetRegisterAsInteger(instr.gpr20, 0, instr.conversion.is_signed);
 
@@ -851,6 +971,27 @@ private:
                 regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_signed, 0, op_a, 1, 1);
                 break;
             }
+            case OpCode::Id::I2F_R: {
+                std::string op_a =
+                    regs.GetRegisterAsInteger(instr.gpr20, 0, instr.conversion.is_signed);
+
+                if (instr.conversion.abs_a) {
+                    op_a = "abs(" + op_a + ')';
+                }
+
+                regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
+                break;
+            }
+            case OpCode::Id::F2F_R: {
+                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);
+                break;
+            }
             default: {
                 NGLOG_CRITICAL(HW_GPU, "Unhandled conversion instruction: {}", opcode->GetName());
                 UNREACHABLE();
@@ -874,10 +1015,10 @@ private:
                                                   instr.gpr0);
                 break;
             }
-            case OpCode::Id::TEXS: {
+            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.gpr20);
+                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
@@ -886,8 +1027,52 @@ private:
                 ++shader.scope;
                 shader.AddLine(coord);
                 const std::string texture = "texture(" + sampler + ", coords)";
-                for (unsigned elem = 0; elem < instr.attribute.fmt20.size; ++elem) {
-                    regs.SetRegisterToFloat(instr.gpr0, elem, texture, 1, 4, false, elem);
+
+                size_t dest_elem{};
+                for (size_t elem = 0; elem < instr.attribute.fmt20.size; ++elem) {
+                    if (!instr.tex.IsComponentEnabled(elem)) {
+                        // Skip disabled components
+                        continue;
+                    }
+                    regs.SetRegisterToFloat(instr.gpr0, elem, texture, 1, 4, false, dest_elem);
+                    ++dest_elem;
+                }
+                --shader.scope;
+                shader.AddLine("}");
+                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);
+                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.
+                shader.AddLine("{");
+                ++shader.scope;
+                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
+                size_t offset{};
+
+                for (const auto& dest : {instr.gpr0.Value(), instr.gpr28.Value()}) {
+                    for (unsigned elem = 0; elem < 2; ++elem) {
+                        if (!instr.texs.IsComponentEnabled(elem)) {
+                            // Skip disabled components
+                            continue;
+                        }
+                        regs.SetRegisterToFloat(dest, elem + offset, texture, 1, 4, false, elem);
+                    }
+
+                    if (!instr.texs.HasTwoDestinations()) {
+                        // Skip the second destination
+                        break;
+                    }
+
+                    offset += 2;
                 }
                 --shader.scope;
                 shader.AddLine("}");
@@ -920,7 +1105,7 @@ private:
                 if (instr.is_b_gpr) {
                     op_b += regs.GetRegisterAsFloat(instr.gpr20);
                 } else {
-                    op_b += regs.GetUniform(instr.uniform, instr.gpr0);
+                    op_b += regs.GetUniform(instr.uniform, GLSLRegister::Type::Float);
                 }
             }
 
@@ -944,13 +1129,49 @@ private:
                          '(' + predicate + ") " + combiner + " (" + second_pred + ')');
 
             if (instr.fsetp.pred0 != static_cast<u64>(Pred::UnusedIndex)) {
-                // Set the secondary predicate to the result of !Predicate OP SecondPredicate, if
-                // enabled
+                // Set the secondary predicate to the result of !Predicate OP SecondPredicate,
+                // if enabled
                 SetPredicate(instr.fsetp.pred0,
                              "!(" + predicate + ") " + combiner + " (" + second_pred + ')');
             }
             break;
         }
+        case OpCode::Type::IntegerSetPredicate: {
+            std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, instr.isetp.is_signed);
+
+            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);
+            } else {
+                op_b += regs.GetUniform(instr.uniform, GLSLRegister::Type::Integer);
+            }
+
+            using Tegra::Shader::Pred;
+            // We can't use the constant predicate as destination.
+            ASSERT(instr.isetp.pred3 != static_cast<u64>(Pred::UnusedIndex));
+
+            std::string second_pred =
+                GetPredicateCondition(instr.isetp.pred39, instr.isetp.neg_pred != 0);
+
+            std::string comparator = GetPredicateComparison(instr.isetp.cond);
+            std::string combiner = GetPredicateCombiner(instr.isetp.op);
+
+            std::string predicate = '(' + op_a + ") " + comparator + " (" + op_b + ')';
+            // Set the primary predicate to the result of Predicate OP SecondPredicate
+            SetPredicate(instr.isetp.pred3,
+                         '(' + predicate + ") " + combiner + " (" + second_pred + ')');
+
+            if (instr.isetp.pred0 != static_cast<u64>(Pred::UnusedIndex)) {
+                // Set the secondary predicate to the result of !Predicate OP SecondPredicate,
+                // if enabled
+                SetPredicate(instr.isetp.pred0,
+                             "!(" + predicate + ") " + combiner + " (" + second_pred + ')');
+            }
+            break;
+        }
         case OpCode::Type::FloatSet: {
             std::string op_a = instr.fset.neg_a ? "-" : "";
             op_a += regs.GetRegisterAsFloat(instr.gpr8);
@@ -971,7 +1192,7 @@ private:
                 if (instr.is_b_gpr) {
                     op_b += regs.GetRegisterAsFloat(instr.gpr20);
                 } else {
-                    op_b += regs.GetUniform(instr.uniform, instr.gpr0);
+                    op_b += regs.GetUniform(instr.uniform, GLSLRegister::Type::Float);
                 }
             }
 
@@ -990,15 +1211,17 @@ private:
             std::string predicate = "(((" + op_a + ") " + comparator + " (" + op_b + ")) " +
                                     combiner + " (" + second_pred + "))";
 
-            regs.SetRegisterToFloat(instr.gpr0, 0, predicate + " ? 1.0 : 0.0", 1, 1);
+            if (instr.fset.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: {
-                ASSERT_MSG(instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex),
-                           "Predicated exits not implemented");
-
                 // Final color output is currently hardcoded to GPR0-3 for fragment shaders
                 if (stage == Maxwell3D::Regs::ShaderStage::Fragment) {
                     shader.AddLine("color.r = " + regs.GetRegisterAsFloat(0) + ';');
@@ -1008,13 +1231,25 @@ private:
                 }
 
                 shader.AddLine("return true;");
-                offset = PROGRAM_END - 1;
+                if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
+                    // If this is an unconditional exit then just end processing here, otherwise we
+                    // have to account for the possibility of the condition not being met, so
+                    // continue processing the next instruction.
+                    offset = PROGRAM_END - 1;
+                }
                 break;
             }
             case OpCode::Id::KIL: {
                 shader.AddLine("discard;");
                 break;
             }
+            case OpCode::Id::BRA: {
+                ASSERT_MSG(instr.bra.constant_buffer == 0,
+                           "BRA with constant buffers are not implemented");
+                u32 target = offset + instr.bra.GetBranchTarget();
+                shader.AddLine("{ jmp_to = " + std::to_string(target) + "u; break; }");
+                break;
+            }
             case OpCode::Id::IPA: {
                 const auto& attribute = instr.attribute.fmt28;
                 regs.SetRegisterToInputAttibute(instr.gpr0, attribute.element, attribute.index);

src/video_core/textures/decoders.cpp

@@ -46,6 +46,7 @@ void CopySwizzledData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_
 u32 BytesPerPixel(TextureFormat format) {
     switch (format) {
     case TextureFormat::DXT1:
+    case TextureFormat::DXN1:
         // In this case a 'pixel' actually refers to a 4x4 tile.
         return 8;
     case TextureFormat::DXT23:
@@ -60,6 +61,8 @@ u32 BytesPerPixel(TextureFormat format) {
         return 2;
     case TextureFormat::R8:
         return 1;
+    case TextureFormat::R16_G16_B16_A16:
+        return 8;
     default:
         UNIMPLEMENTED_MSG("Format not implemented");
         break;
@@ -77,7 +80,9 @@ std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width,
     case TextureFormat::DXT1:
     case TextureFormat::DXT23:
     case TextureFormat::DXT45:
-        // In the DXT formats, each 4x4 tile is swizzled instead of just individual pixel values.
+    case TextureFormat::DXN1:
+        // In the DXT and DXN formats, each 4x4 tile is swizzled instead of just individual pixel
+        // values.
         CopySwizzledData(width / 4, height / 4, bytes_per_pixel, bytes_per_pixel, data,
                          unswizzled_data.data(), true, block_height);
         break;
@@ -86,6 +91,7 @@ std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width,
     case TextureFormat::A1B5G5R5:
     case TextureFormat::B5G6R5:
     case TextureFormat::R8:
+    case TextureFormat::R16_G16_B16_A16:
         CopySwizzledData(width, height, bytes_per_pixel, bytes_per_pixel, data,
                          unswizzled_data.data(), true, block_height);
         break;
@@ -106,6 +112,7 @@ std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat
     case TextureFormat::DXT1:
     case TextureFormat::DXT23:
     case TextureFormat::DXT45:
+    case TextureFormat::DXN1:
     case TextureFormat::A8R8G8B8:
     case TextureFormat::A2B10G10R10:
     case TextureFormat::A1B5G5R5:

src/yuzu/main.cpp

@@ -335,6 +335,24 @@ void GMainWindow::OnDisplayTitleBars(bool show) {
     }
 }
 
+bool GMainWindow::SupportsRequiredGLExtensions() {
+    QStringList unsupported_ext;
+
+    if (!GLAD_GL_ARB_program_interface_query)
+        unsupported_ext.append("ARB_program_interface_query");
+    if (!GLAD_GL_ARB_separate_shader_objects)
+        unsupported_ext.append("ARB_separate_shader_objects");
+    if (!GLAD_GL_ARB_shader_storage_buffer_object)
+        unsupported_ext.append("ARB_shader_storage_buffer_object");
+    if (!GLAD_GL_ARB_vertex_attrib_binding)
+        unsupported_ext.append("ARB_vertex_attrib_binding");
+
+    for (const QString& ext : unsupported_ext)
+        NGLOG_CRITICAL(Frontend, "Unsupported GL extension: {}", ext.toStdString());
+
+    return unsupported_ext.empty();
+}
+
 bool GMainWindow::LoadROM(const QString& filename) {
     // Shutdown previous session if the emu thread is still active...
     if (emu_thread != nullptr)
@@ -350,6 +368,14 @@ bool GMainWindow::LoadROM(const QString& filename) {
         return false;
     }
 
+    if (!SupportsRequiredGLExtensions()) {
+        QMessageBox::critical(
+            this, tr("Error while initializing OpenGL Core!"),
+            tr("Your GPU may not support one or more required OpenGL extensions. Please "
+               "ensure you have the latest graphics driver. See the log for more details."));
+        return false;
+    }
+
     Core::System& system{Core::System::GetInstance()};
 
     system.SetGPUDebugContext(debug_context);

src/yuzu/main.h

@@ -79,6 +79,7 @@ private:
     void ConnectWidgetEvents();
     void ConnectMenuEvents();
 
+    bool SupportsRequiredGLExtensions();
     bool LoadROM(const QString& filename);
     void BootGame(const QString& filename);
     void ShutdownGame();

src/yuzu_cmd/emu_window/emu_window_sdl2.cpp

@@ -78,6 +78,24 @@ void EmuWindow_SDL2::Fullscreen() {
     SDL_MaximizeWindow(render_window);
 }
 
+bool EmuWindow_SDL2::SupportsRequiredGLExtensions() {
+    std::vector<std::string> unsupported_ext;
+
+    if (!GLAD_GL_ARB_program_interface_query)
+        unsupported_ext.push_back("ARB_program_interface_query");
+    if (!GLAD_GL_ARB_separate_shader_objects)
+        unsupported_ext.push_back("ARB_separate_shader_objects");
+    if (!GLAD_GL_ARB_shader_storage_buffer_object)
+        unsupported_ext.push_back("ARB_shader_storage_buffer_object");
+    if (!GLAD_GL_ARB_vertex_attrib_binding)
+        unsupported_ext.push_back("ARB_vertex_attrib_binding");
+
+    for (const std::string& ext : unsupported_ext)
+        NGLOG_CRITICAL(Frontend, "Unsupported GL extension: {}", ext);
+
+    return unsupported_ext.empty();
+}
+
 EmuWindow_SDL2::EmuWindow_SDL2(bool fullscreen) {
     InputCommon::Init();
 
@@ -128,6 +146,11 @@ EmuWindow_SDL2::EmuWindow_SDL2(bool fullscreen) {
         exit(1);
     }
 
+    if (!SupportsRequiredGLExtensions()) {
+        NGLOG_CRITICAL(Frontend, "GPU does not support all required OpenGL extensions! Exiting...");
+        exit(1);
+    }
+
     OnResize();
     OnMinimalClientAreaChangeRequest(GetActiveConfig().min_client_area_size);
     SDL_PumpEvents();

src/yuzu_cmd/emu_window/emu_window_sdl2.h

@@ -46,6 +46,9 @@ private:
     /// Called when user passes the fullscreen parameter flag
     void Fullscreen();
 
+    /// Whether the GPU and driver supports the OpenGL extension required
+    bool SupportsRequiredGLExtensions();
+
     /// Called when a configuration change affects the minimal size of the window
     void OnMinimalClientAreaChangeRequest(
         const std::pair<unsigned, unsigned>& minimal_size) override;